{"afmoe/modeling_afmoe.py:AfmoeRotaryEmbedding.__init__": ["Callable", "False", "ModelConfig", "ModelRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "afmoe/modeling_afmoe.py:AfmoeRotaryEmbedding.compute_default_rope_parameters": ["ModelConfig", "ModelRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "afmoe/modeling_afmoe.py:AfmoeRotaryEmbedding.forward": ["False", "ModelRotaryEmbedding", "Module", "None", "and", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "afmoe/modeling_afmoe.py:AfmoeRMSNorm.__init__": ["ModelRMSNorm", "Module", "Parameter", "__init__", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "afmoe/modeling_afmoe.py:AfmoeRMSNorm.forward": ["ModelRMSNorm", "Module", "True", "class", "def", "dtype", "float32", "forward", "hidden_states", "input_dtype", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "afmoe/modeling_afmoe.py:AfmoeRMSNorm.extra_repr": ["ModelRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "variance_epsilon", "weight"], "afmoe/modeling_afmoe.py:AfmoeMLP.__init__": ["ACT2FN", "False", "Linear", "ModelMLP", "Module", "None", "__init__", "act_fn", "bias", "class", "config", "def", "down_proj", "else", "gate_proj", "hidden_act", "hidden_size", "if", "intermediate_size", "is", "nn", "self", "super", "up_proj"], "afmoe/modeling_afmoe.py:AfmoeMLP.forward": ["ModelMLP", "Module", "act_fn", "class", "def", "down_proj", "forward", "gate_proj", "nn", "return", "self", "up_proj", "x"], "afmoe/modeling_afmoe.py:AfmoeTokenChoiceRouter.__init__": ["False", "Linear", "ModelTokenChoiceRouter", "Module", "__init__", "bias", "class", "config", "def", "gate", "hidden_size", "nn", "num_experts", "num_experts_per_tok", "route_scale", "self", "super", "top_k"], "afmoe/modeling_afmoe.py:AfmoeTokenChoiceRouter.forward": ["ModelTokenChoiceRouter", "Module", "Tensor", "True", "_", "class", "def", "denominator", "dim", "expert_bias", "float32", "forward", "gate", "gather", "hidden_dim", "hidden_states", "index", "k", "keepdim", "nn", "return", "route_scale", "scores", "selected_experts", "self", "shape", "sigmoid", "sum", "to", "top_k", "top_scores", "topk", "torch", "view"], "afmoe/modeling_afmoe.py:AfmoeExperts.__init__": ["ModelConfig", "ModelExperts", "ModelMLP", "ModuleList", "_", "__init__", "append", "class", "config", "def", "for", "in", "intermediate_size", "moe_intermediate_size", "nn", "num_experts", "num_experts_per_tok", "range", "self", "super", "top_k"], "afmoe/modeling_afmoe.py:AfmoeExperts.forward": ["ModelExperts", "ModuleList", "Tensor", "True", "aggregated", "arange", "argsort", "batch_size", "class", "continue", "count", "counts", "def", "device", "dispatched_tokens", "dtype", "end", "expand_as", "expert_id", "expert_indices", "expert_input", "expert_output", "expert_outputs", "float32", "for", "forward", "hidden_dim", "hidden_states", "hidden_states_flat", "if", "in", "index_select", "long", "new_zeros", "nn", "repeat_interleave", "reshape", "return", "return_counts", "routing_weights", "scatter_add_", "scatter_indices", "selected_experts", "self", "seq_len", "shape", "sorting", "stable", "start", "to", "token_indices", "tolist", "top_k", "torch", "unique_consecutive", "unique_experts", "unsqueeze", "view", "weighted_outputs", "zeros_like", "zip"], "afmoe/modeling_afmoe.py:AfmoeMoE.__init__": ["False", "ModelExperts", "ModelMLP", "ModelMoE", "ModelTokenChoiceRouter", "Module", "Parameter", "__init__", "class", "config", "def", "dtype", "expert_bias", "experts", "float32", "moe_intermediate_size", "nn", "num_experts", "num_shared_experts", "requires_grad", "router", "self", "shared_experts", "super", "torch", "zeros"], "afmoe/modeling_afmoe.py:AfmoeMoE.forward": ["ModelMoE", "Module", "batch_size", "class", "config", "def", "expert_bias", "experts", "forward", "hidden_dim", "hidden_states", "hidden_states_flat", "nn", "num_experts_per_tok", "return", "routed_output", "router", "selected_experts", "self", "seq_len", "shape", "shared_experts", "shared_output", "top_scores", "view"], "afmoe/modeling_afmoe.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "afmoe/modeling_afmoe.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cos", "def", "k", "k_embed", "q", "q_embed", "return", "rotate_half", "sin", "unsqueeze", "unsqueeze_dim"], "afmoe/modeling_afmoe.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "afmoe/modeling_afmoe.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value", "value_states"], "afmoe/modeling_afmoe.py:AfmoeAttention.__init__": ["False", "Linear", "ModelAttention", "ModelConfig", "ModelRMSNorm", "Module", "None", "True", "__init__", "attention_bias", "attention_dropout", "bias", "class", "config", "def", "else", "eps", "gate_proj", "getattr", "head_dim", "hidden_size", "if", "int", "is_causal", "is_local_attention", "k_norm", "k_proj", "layer_idx", "layer_types", "nn", "num_attention_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "q_norm", "q_proj", "rms_norm_eps", "scaling", "self", "sliding_attention", "sliding_window", "super", "v_proj"], "afmoe/modeling_afmoe.py:AfmoeAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "LongTensor", "ModelAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_attn_implementation", "apply_rotary_pos_emb", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "cache_kwargs", "cache_position", "class", "config", "contiguous", "cos", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "gate_proj", "gate_states", "head_dim", "hidden_shape", "hidden_states", "if", "input_shape", "is", "is_local_attention", "k_norm", "k_proj", "key_states", "kwargs", "layer_idx", "nn", "not", "o_proj", "output", "past_key_value", "position_embeddings", "q_norm", "q_proj", "query_states", "return", "scaling", "self", "shape", "sigmoid", "sin", "sliding_window", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "view"], "afmoe/modeling_afmoe.py:AfmoeDecoderLayer.__init__": ["GradientCheckpointingLayer", "ModelAttention", "ModelConfig", "ModelDecoderLayer", "ModelMLP", "ModelMoE", "ModelRMSNorm", "__init__", "attention_type", "class", "config", "def", "else", "eps", "hidden_size", "if", "input_layernorm", "int", "layer_idx", "layer_types", "mlp", "moe_enabled", "num_dense_layers", "post_attention_layernorm", "post_mlp_layernorm", "pre_mlp_layernorm", "rms_norm_eps", "self", "self_attn", "super"], "afmoe/modeling_afmoe.py:AfmoeDecoderLayer.forward": ["Cache", "FloatTensor", "GradientCheckpointingLayer", "LongTensor", "ModelDecoderLayer", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "attention_mask", "bool", "cache_position", "class", "def", "forward", "hidden_states", "input_layernorm", "kwargs", "mlp", "past_key_value", "position_embeddings", "position_ids", "post_attention_layernorm", "post_mlp_layernorm", "pre_mlp_layernorm", "residual", "return", "self", "self_attn", "torch", "tuple", "use_cache"], "afmoe/modeling_afmoe.py:AfmoePreTrainedModel._init_weights": ["ModelMoE", "ModelPreTrainedModel", "ModelTokenChoiceRouter", "PreTrainedModel", "_init_weights", "class", "def", "elif", "expert_bias", "gate", "if", "init", "isinstance", "module", "self", "super", "weight", "zeros_"], "afmoe/modeling_afmoe.py:AfmoeModel.__init__": ["Embedding", "False", "ModelConfig", "ModelDecoderLayer", "ModelModel", "ModelPreTrainedModel", "ModelRMSNorm", "ModelRotaryEmbedding", "ModuleList", "__init__", "class", "config", "def", "embed_tokens", "eps", "for", "gradient_checkpointing", "hidden_size", "in", "layer_idx", "layers", "nn", "norm", "num_hidden_layers", "pad_token_id", "padding_idx", "post_init", "range", "rms_norm_eps", "rotary_emb", "self", "super", "vocab_size"], "afmoe/modeling_afmoe.py:AfmoeModel.forward": ["Cache", "DynamicCache", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "MoeModelOutputWithPast", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "and", "arange", "attention_mask", "attention_type", "bool", "cache_position", "causal_mask_mapping", "class", "config", "create_causal_mask", "create_sliding_window_causal_mask", "decoder_layer", "def", "device", "dict", "else", "embed_tokens", "exactly", "for", "forward", "full_attention", "get_seq_length", "hidden_size", "hidden_states", "if", "in", "input_embeds", "input_ids", "inputs_embeds", "is", "isinstance", "kwargs", "last_hidden_state", "layers", "mask_kwargs", "mup_enabled", "must", "norm", "not", "of", "one", "or", "past_key_value", "past_key_values", "past_seen_tokens", "position_embeddings", "position_ids", "raise", "return", "rotary_emb", "self", "shape", "sliding_attention", "specify", "torch", "unsqueeze", "use_cache"], "afmoe/modeling_afmoe.py:AfmoeForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "ModelForCausalLM", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "vocab_size"], "afmoe/modeling_afmoe.py:AfmoeForCausalLM.forward": ["BaseModelOutputWithPast", "Cache", "CausalLMOutputWithPast", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForCausalLM", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "outputs", "past_key_values", "position_ids", "r", "return", "self", "slice", "slice_indices", "torch", "use_cache", "vocab_size"], "aimv2/modeling_aimv2.py:Aimv2Output.to_tuple": ["Any", "ModelOutput", "class", "def", "else", "for", "getattr", "if", "in", "k", "keys", "not", "return", "self", "text_model_output", "to_tuple", "tuple", "vision_model_output"], "aimv2/modeling_aimv2.py:Aimv2RMSNorm.__init__": ["ModelRMSNorm", "Module", "Parameter", "__init__", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "aimv2/modeling_aimv2.py:Aimv2RMSNorm.forward": ["ModelRMSNorm", "Module", "True", "class", "def", "dtype", "float32", "forward", "hidden_states", "input_dtype", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "aimv2/modeling_aimv2.py:Aimv2RMSNorm.extra_repr": ["ModelRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "variance_epsilon", "weight"], "aimv2/modeling_aimv2.py:Aimv2MLP.__init__": ["ACT2FN", "Linear", "ModelMLP", "Module", "__init__", "act_fn", "bias", "class", "config", "def", "down_proj", "gate_proj", "hidden_act", "hidden_size", "intermediate_size", "mlp_bias", "nn", "self", "super", "up_proj"], "aimv2/modeling_aimv2.py:Aimv2MLP.forward": ["ModelMLP", "Module", "act_fn", "class", "def", "down_proj", "forward", "gate_proj", "nn", "return", "self", "up_proj", "x"], "aimv2/modeling_aimv2.py:Aimv2VisionEmbeddings.__init__": ["Conv2d", "Embedding", "False", "ModelRMSNorm", "ModelVisionConfig", "ModelVisionEmbeddings", "Module", "__init__", "arange", "class", "config", "def", "expand", "hidden_size", "if", "image_size", "is_native", "kernel_size", "nn", "not", "num_channels", "num_patches", "patch_embed", "patch_size", "persistent", "position_embedding", "position_ids", "register_buffer", "rms_norm", "rms_norm_eps", "self", "stride", "super", "torch"], "aimv2/modeling_aimv2.py:Aimv2VisionEmbeddings.build_2d_sincos_position_embedding": ["ModelVisionEmbeddings", "Module", "None", "Tensor", "arange", "build_2d_sincos_position_embedding", "class", "concat", "cos", "cpu", "def", "device", "dim", "dtype", "embed_dim", "flatten", "float32", "grid_h", "grid_w", "height", "indexing", "int", "meshgrid", "nn", "omega", "out_h", "out_w", "pos_dim", "return", "sin", "temperature", "torch", "width", "xy"], "aimv2/modeling_aimv2.py:Aimv2VisionEmbeddings.forward": ["ModelVisionEmbeddings", "Module", "Tensor", "_", "build_2d_sincos_position_embedding", "class", "config", "def", "device", "dtype", "else", "embed_dim", "flatten", "forward", "height", "hidden_size", "hidden_states", "if", "is_native", "nn", "patch_embed", "patch_size", "pixel_values", "pos_embed", "position_embedding", "position_ids", "return", "rms_norm", "self", "size", "torch", "transpose", "width"], "aimv2/modeling_aimv2.py:Aimv2TextEmbeddings.__init__": ["Embedding", "False", "ModelTextConfig", "ModelTextEmbeddings", "Module", "__init__", "arange", "class", "config", "def", "embed_dim", "expand", "hidden_size", "max_position_embeddings", "nn", "persistent", "position_embedding", "position_ids", "register_buffer", "self", "super", "token_embedding", "torch", "vocab_size"], "aimv2/modeling_aimv2.py:Aimv2TextEmbeddings.forward": ["FloatTensor", "LongTensor", "ModelTextEmbeddings", "Module", "None", "Sequence", "Tensor", "ValueError", "and", "be", "class", "def", "else", "embeddings", "f", "forward", "got", "if", "input_ids", "inputs_embeds", "is", "length", "less", "max_position_embedding", "max_position_embeddings", "must", "nn", "not", "position_embedding", "position_embeddings", "position_ids", "raise", "return", "self", "seq_length", "sequence", "shape", "than", "token_embedding", "torch", "weight"], "aimv2/modeling_aimv2.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "attention_mask", "attn_output", "attn_weights", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "softmax", "to", "torch", "training", "transpose", "value"], "aimv2/modeling_aimv2.py:Aimv2Attention.__init__": ["False", "Linear", "ModelAttention", "Module", "ValueError", "__init__", "and", "attention_dropout", "be", "bias", "by", "class", "config", "def", "divisible", "dropout", "embed_dim", "f", "got", "head_dim", "hidden_size", "if", "is_causal", "k_proj", "must", "nn", "num_attention_heads", "num_heads", "out_proj", "q_proj", "qkv_bias", "raise", "scale", "self", "super", "v_proj"], "aimv2/modeling_aimv2.py:Aimv2Attention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "ModelAttention", "Module", "None", "Tensor", "_attn_implementation", "attention_interface", "attention_mask", "attn_output", "attn_weights", "batch_size", "class", "config", "contiguous", "def", "dropout", "eager", "eager_attention_forward", "else", "embed_dim", "forward", "head_dim", "hidden_states", "if", "is_causal", "k_proj", "keys", "kwargs", "nn", "not", "num_heads", "out_proj", "q_proj", "queries", "reshape", "return", "scale", "scaling", "self", "seq_length", "shape", "torch", "training", "transpose", "tuple", "v_proj", "values", "view"], "aimv2/modeling_aimv2.py:Aimv2EncoderLayer.__init__": ["GradientCheckpointingLayer", "ModelAttention", "ModelEncoderLayer", "ModelMLP", "ModelRMSNorm", "ModelVisionConfig", "__init__", "attention", "class", "config", "def", "ffn", "hidden_size", "rms_norm1", "rms_norm2", "rms_norm_eps", "self", "super"], "aimv2/modeling_aimv2.py:Aimv2EncoderLayer.forward": ["GradientCheckpointingLayer", "ModelEncoderLayer", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "attention", "attention_mask", "attn_output", "class", "def", "ffn", "forward", "hidden_states", "kwargs", "mlp_output", "norm_hidden_states", "return", "rms_norm1", "rms_norm2", "self", "torch"], "aimv2/modeling_aimv2.py:Aimv2Encoder.__init__": ["False", "ModelConfig", "ModelEncoder", "ModelEncoderLayer", "Module", "ModuleList", "_", "__init__", "class", "config", "def", "for", "gradient_checkpointing", "in", "layers", "nn", "num_hidden_layers", "range", "self", "super"], "aimv2/modeling_aimv2.py:Aimv2Encoder.forward": ["BaseModelOutput", "ModelEncoder", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "class", "def", "encoder_layer", "for", "forward", "hidden_states", "in", "inputs_embeds", "kwargs", "last_hidden_state", "layers", "nn", "return", "self", "torch"], "aimv2/modeling_aimv2.py:Aimv2AttentionPoolingHead.__init__": ["Linear", "ModelAttentionPoolingHead", "ModelVisionConfig", "Module", "Parameter", "True", "__init__", "bias", "class", "cls_token", "config", "def", "hidden_size", "k_proj", "nn", "num_attention_heads", "num_heads", "output_proj", "qkv_bias", "self", "super", "torch", "v_proj", "zeros"], "aimv2/modeling_aimv2.py:Aimv2AttentionPoolingHead.forward": ["F", "ModelAttentionPoolingHead", "Module", "Tensor", "attn_output", "batch_size", "class", "cls_token", "def", "dim", "expand", "forward", "hidden_dim", "hidden_states", "k_proj", "key", "mean", "nn", "num_heads", "output", "output_proj", "permute", "query", "reshape", "return", "scaled_dot_product_attention", "self", "seq_len", "shape", "torch", "transpose", "v_proj", "value"], "aimv2/modeling_aimv2.py:Aimv2PreTrainedModel._init_weights": ["ModelAttentionPoolingHead", "ModelPreTrainedModel", "ModelTextEmbeddings", "ModelVisionEmbeddings", "Parameter", "PreTrainedModel", "_init_weights", "arange", "class", "cls_token", "config", "constant_", "copy_", "def", "elif", "expand", "hasattr", "if", "init", "initializer_range", "isinstance", "log", "logit_scale", "math", "mean", "module", "nn", "normal_", "position_ids", "self", "shape", "std", "super", "torch"], "aimv2/modeling_aimv2.py:Aimv2VisionModel.__init__": ["ModelAttentionPoolingHead", "ModelEncoder", "ModelPreTrainedModel", "ModelRMSNorm", "ModelVisionConfig", "ModelVisionEmbeddings", "ModelVisionModel", "__init__", "class", "config", "def", "embeddings", "encoder", "head", "hidden_size", "if", "post_init", "rms_norm", "rms_norm_eps", "self", "super", "use_head"], "aimv2/modeling_aimv2.py:Aimv2VisionModel.get_input_embeddings": ["ModelPreTrainedModel", "ModelVisionModel", "Module", "class", "def", "embeddings", "get_input_embeddings", "nn", "patch_embed", "return", "self"], "aimv2/modeling_aimv2.py:Aimv2VisionModel.forward": ["BaseModelOutput", "BaseModelOutputWithPooling", "ModelPreTrainedModel", "ModelVisionModel", "None", "TransformersKwargs", "Unpack", "class", "def", "else", "embeddings", "encoder", "encoder_outputs", "forward", "head", "hidden_states", "if", "inputs_embeds", "kwargs", "last_hidden_state", "pixel_values", "pooler_output", "r", "return", "rms_norm", "self", "use_head"], "aimv2/modeling_aimv2.py:Aimv2TextModel.__init__": ["ModelEncoder", "ModelPreTrainedModel", "ModelRMSNorm", "ModelTextConfig", "ModelTextEmbeddings", "ModelTextModel", "__init__", "class", "config", "def", "embeddings", "encoder", "eos_token_id", "hidden_size", "post_init", "rms_norm", "rms_norm_eps", "self", "super"], "aimv2/modeling_aimv2.py:Aimv2TextModel.get_input_embeddings": ["ModelPreTrainedModel", "ModelTextModel", "Module", "class", "def", "embeddings", "get_input_embeddings", "nn", "return", "self", "token_embedding"], "aimv2/modeling_aimv2.py:Aimv2TextModel.set_input_embeddings": ["ModelPreTrainedModel", "ModelTextModel", "class", "def", "embeddings", "self", "set_input_embeddings", "token_embedding", "value"], "aimv2/modeling_aimv2.py:Aimv2TextModel.forward": ["BaseModelOutputWithPooling", "ModelPreTrainedModel", "ModelTextModel", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "arange", "argmax", "attention_mask", "batch_size", "cache_position", "class", "config", "create_causal_mask", "def", "device", "dim", "dtype", "embeddings", "encoder", "encoder_outputs", "eos_token_id", "expand", "forward", "hidden_states", "if", "input_embeds", "input_ids", "inputs_embeds", "int", "is", "kwargs", "last_hidden_state", "long", "not", "past_key_values", "pooled_output", "pooler_output", "position_ids", "return", "rms_norm", "self", "seq_len", "shape", "to", "torch", "unsqueeze"], "aimv2/modeling_aimv2.py:_get_vector_norm": ["Tensor", "True", "_get_vector_norm", "def", "dim", "keepdim", "normed_tensor", "pow", "return", "square_tensor", "sum", "sum_tensor", "tensor", "torch"], "aimv2/modeling_aimv2.py:Aimv2Model.__init__": ["False", "Linear", "ModelConfig", "ModelModel", "ModelPreTrainedModel", "ModelTextModel", "ModelVisionModel", "Parameter", "__init__", "_from_config", "bias", "class", "config", "def", "hidden_size", "log", "logit_scale", "logit_scale_init_value", "math", "max_log_logit_scale", "max_logit_scale", "nn", "post_init", "projection_dim", "self", "super", "tensor", "text_config", "text_embed_dim", "text_model", "text_projection", "torch", "vision_config", "vision_embed_dim", "vision_model", "visual_projection"], "aimv2/modeling_aimv2.py:Aimv2Model.get_text_features": ["BaseModelOutputWithPooling", "FloatTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "attention_mask", "class", "def", "get_text_features", "input_ids", "pooled_output", "pooler_output", "position_ids", "r", "return", "self", "text_features", "text_model", "text_outputs", "text_projection", "torch"], "aimv2/modeling_aimv2.py:Aimv2Model.get_image_features": ["BaseModelOutputWithPooling", "False", "FloatTensor", "ModelModel", "ModelPreTrainedModel", "bool", "class", "def", "get_image_features", "image_features", "interpolate_pos_encoding", "pixel_values", "pooled_output", "pooler_output", "r", "return", "self", "torch", "vision_model", "vision_outputs", "visual_projection"], "aimv2/modeling_aimv2.py:Aimv2Model.forward": ["BaseModelOutputWithPooling", "FloatTensor", "LongTensor", "ModelModel", "ModelOutput", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "_get_vector_norm", "attention_mask", "clamp", "class", "def", "device", "exp", "forward", "image_embeds", "input_ids", "kwargs", "logit_scale", "logits_per_image", "logits_per_text", "max_log_logit_scale", "pixel_values", "pooler_output", "r", "return", "self", "t", "text_embeds", "text_model", "text_model_output", "text_outputs", "text_projection", "to", "torch", "vision_model", "vision_model_output", "vision_outputs", "visual_projection"], "albert/modeling_albert.py:AlbertEmbeddings.__init__": ["Dropout", "Embedding", "False", "LayerNorm", "ModelConfig", "ModelEmbeddings", "Module", "__init__", "arange", "class", "config", "def", "dropout", "dtype", "embedding_size", "eps", "expand", "hidden_dropout_prob", "layer_norm_eps", "long", "max_position_embeddings", "nn", "pad_token_id", "padding_idx", "persistent", "position_embeddings", "position_ids", "register_buffer", "self", "size", "super", "token_type_embeddings", "token_type_ids", "torch", "type_vocab_size", "vocab_size", "word_embeddings", "zeros"], "albert/modeling_albert.py:AlbertEmbeddings.forward": ["FloatTensor", "LayerNorm", "LongTensor", "ModelEmbeddings", "Module", "None", "Tensor", "batch_size", "buffered_token_type_ids", "class", "def", "device", "dim", "dropout", "dtype", "else", "embeddings", "expand", "forward", "gather", "hasattr", "if", "index", "input_ids", "input_shape", "inputs_embeds", "is", "long", "nn", "not", "position_embeddings", "position_ids", "return", "self", "seq_length", "shape", "size", "token_type_embeddings", "token_type_ids", "torch", "word_embeddings", "zeros"], "albert/modeling_albert.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "contiguous", "def", "dim", "dropout", "float", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "shape", "size", "softmax", "torch", "training", "transpose", "value"], "albert/modeling_albert.py:AlbertAttention.__init__": ["Dropout", "False", "LayerNorm", "Linear", "ModelAttention", "ModelConfig", "Module", "The", "ValueError", "__init__", "a", "all_head_size", "and", "attention", "attention_dropout", "attention_head_size", "attention_probs_dropout_prob", "class", "config", "def", "dense", "embedding_size", "eps", "f", "hasattr", "heads", "hidden", "hidden_dropout_prob", "hidden_size", "if", "is", "is_causal", "key", "layer_norm_eps", "multiple", "nn", "not", "num_attention_heads", "number", "of", "output_dropout", "query", "raise", "scaling", "self", "size", "super", "the", "value"], "albert/modeling_albert.py:AlbertAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "FloatTensor", "LayerNorm", "ModelAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_attn_implementation", "attention_dropout", "attention_head_size", "attention_interface", "attention_mask", "attn_output", "attn_weights", "class", "config", "contiguous", "def", "dense", "dropout", "eager", "eager_attention_forward", "else", "forward", "hidden_shape", "hidden_states", "if", "input_shape", "key", "key_layer", "kwargs", "nn", "not", "output_dropout", "p", "query", "query_layer", "reshape", "return", "scaling", "self", "shape", "torch", "training", "transpose", "tuple", "value", "value_layer", "view"], "albert/modeling_albert.py:AlbertLayer.__init__": ["ACT2FN", "Dropout", "LayerNorm", "Linear", "ModelAttention", "ModelConfig", "ModelLayer", "Module", "__init__", "activation", "attention", "chunk_size_feed_forward", "class", "config", "def", "dropout", "eps", "ffn", "ffn_output", "full_layer_layer_norm", "hidden_act", "hidden_dropout_prob", "hidden_size", "intermediate_size", "layer_norm_eps", "nn", "self", "seq_len_dim", "super"], "albert/modeling_albert.py:AlbertLayer.forward": ["FloatTensor", "ModelLayer", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "apply_chunking_to_forward", "attention", "attention_mask", "attention_output", "chunk_size_feed_forward", "class", "def", "ff_chunk", "ffn_output", "forward", "full_layer_layer_norm", "hidden_states", "kwargs", "nn", "return", "self", "seq_len_dim", "torch", "tuple"], "albert/modeling_albert.py:AlbertLayer.ff_chunk": ["ModelLayer", "Module", "Tensor", "activation", "attention_output", "class", "def", "ff_chunk", "ffn", "ffn_output", "nn", "return", "self", "torch"], "albert/modeling_albert.py:AlbertLayerGroup.__init__": ["ModelConfig", "ModelLayer", "ModelLayerGroup", "Model_layers", "Module", "ModuleList", "_", "__init__", "class", "config", "def", "for", "in", "inner_group_num", "nn", "range", "self", "super"], "albert/modeling_albert.py:AlbertLayerGroup.forward": ["FloatTensor", "ModelLayerGroup", "Model_layer", "Model_layers", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "class", "def", "enumerate", "for", "forward", "hidden_states", "in", "kwargs", "layer_index", "nn", "return", "self", "torch", "tuple"], "albert/modeling_albert.py:AlbertTransformer.__init__": ["Linear", "ModelConfig", "ModelLayerGroup", "ModelTransformer", "Model_layer_groups", "Module", "ModuleList", "_", "__init__", "class", "config", "def", "embedding_hidden_mapping_in", "embedding_size", "for", "hidden_size", "in", "nn", "num_hidden_groups", "range", "self", "super"], "albert/modeling_albert.py:AlbertTransformer.forward": ["BaseModelOutput", "FloatTensor", "ModelTransformer", "Model_layer_groups", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "class", "config", "def", "embedding_hidden_mapping_in", "for", "forward", "group_idx", "hidden_states", "i", "in", "int", "kwargs", "last_hidden_state", "nn", "num_hidden_groups", "num_hidden_layers", "range", "return", "self", "torch", "tuple"], "albert/modeling_albert.py:AlbertPreTrainedModel._init_weights": ["Embedding", "False", "LayerNorm", "Linear", "ModelEmbeddings", "ModelMLMHead", "ModelPreTrainedModel", "None", "PreTrainedModel", "_init_weights", "_is_hf_initialized", "and", "arange", "bias", "class", "config", "copy_", "def", "elif", "expand", "getattr", "if", "init", "initializer_range", "is", "isinstance", "mean", "module", "nn", "normal_", "not", "ones_", "padding_idx", "position_ids", "self", "shape", "std", "token_type_ids", "torch", "weight", "zeros_"], "albert/modeling_albert.py:AlbertModel.__init__": ["Linear", "ModelConfig", "ModelEmbeddings", "ModelModel", "ModelPreTrainedModel", "ModelTransformer", "None", "Tanh", "True", "__init__", "_attn_implementation", "add_pooling_layer", "attn_implementation", "bool", "class", "config", "def", "else", "embeddings", "encoder", "hidden_size", "if", "nn", "pooler", "pooler_activation", "post_init", "r", "self", "super"], "albert/modeling_albert.py:AlbertModel.get_input_embeddings": ["Embedding", "ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "nn", "return", "self", "word_embeddings"], "albert/modeling_albert.py:AlbertModel.set_input_embeddings": ["Embedding", "ModelModel", "ModelPreTrainedModel", "None", "class", "def", "embeddings", "nn", "self", "set_input_embeddings", "value", "word_embeddings"], "albert/modeling_albert.py:AlbertModel.forward": ["BaseModelOutputWithPooling", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "TransformersKwargs", "Unpack", "ValueError", "You", "attention_mask", "class", "config", "create_bidirectional_mask", "def", "else", "embedding_output", "embeddings", "encoder", "encoder_outputs", "exactly", "forward", "if", "input_embeds", "input_ids", "inputs_embeds", "is", "kwargs", "last_hidden_state", "must", "not", "of", "one", "or", "pooled_output", "pooler", "pooler_activation", "pooler_output", "position_ids", "raise", "return", "self", "sequence_output", "specify", "token_type_ids", "torch", "tuple"], "albert/modeling_albert.py:AlbertForPreTraining.__init__": ["Model", "ModelConfig", "ModelForPreTraining", "ModelMLMHead", "ModelModel", "ModelPreTrainedModel", "ModelSOPHead", "__init__", "class", "config", "def", "post_init", "predictions", "self", "sop_classifier", "super"], "albert/modeling_albert.py:AlbertForPreTraining.get_output_embeddings": ["Linear", "ModelForPreTraining", "ModelPreTrainedModel", "class", "decoder", "def", "get_output_embeddings", "nn", "predictions", "return", "self"], "albert/modeling_albert.py:AlbertForPreTraining.set_output_embeddings": ["Linear", "ModelForPreTraining", "ModelPreTrainedModel", "None", "class", "decoder", "def", "new_embeddings", "nn", "predictions", "self", "set_output_embeddings"], "albert/modeling_albert.py:AlbertForPreTraining.get_input_embeddings": ["Embedding", "Model", "ModelForPreTraining", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "nn", "return", "self", "word_embeddings"], "albert/modeling_albert.py:AlbertForPreTraining.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "Model", "ModelForPreTraining", "ModelForPreTrainingOutput", "ModelPreTrainedModel", "None", "TransformersKwargs", "True", "Unpack", "and", "attention_mask", "attentions", "class", "config", "def", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "loss", "loss_fct", "masked_lm_loss", "not", "outputs", "pooled_output", "position_ids", "prediction_logits", "prediction_scores", "predictions", "r", "return", "return_dict", "self", "sentence_order_label", "sentence_order_loss", "sequence_output", "sop_classifier", "sop_logits", "sop_scores", "token_type_ids", "torch", "total_loss", "tuple", "view", "vocab_size"], "albert/modeling_albert.py:AlbertMLMHead.__init__": ["ACT2FN", "LayerNorm", "Linear", "ModelConfig", "ModelMLMHead", "Module", "Parameter", "__init__", "activation", "bias", "class", "config", "decoder", "def", "dense", "embedding_size", "eps", "hidden_act", "hidden_size", "layer_norm_eps", "nn", "self", "super", "torch", "vocab_size", "zeros"], "albert/modeling_albert.py:AlbertMLMHead.forward": ["LayerNorm", "ModelMLMHead", "Module", "Tensor", "activation", "class", "decoder", "def", "dense", "forward", "hidden_states", "nn", "prediction_scores", "return", "self", "torch"], "albert/modeling_albert.py:AlbertSOPHead.__init__": ["Dropout", "Linear", "ModelConfig", "ModelSOPHead", "Module", "__init__", "class", "classifier", "classifier_dropout_prob", "config", "def", "dropout", "hidden_size", "nn", "num_labels", "self", "super"], "albert/modeling_albert.py:AlbertSOPHead.forward": ["ModelSOPHead", "Module", "Tensor", "class", "classifier", "def", "dropout", "dropout_pooled_output", "forward", "logits", "nn", "pooled_output", "return", "self", "torch"], "albert/modeling_albert.py:AlbertForMaskedLM.__init__": ["False", "Model", "ModelForMaskedLM", "ModelMLMHead", "ModelModel", "ModelPreTrainedModel", "__init__", "add_pooling_layer", "class", "config", "def", "post_init", "predictions", "self", "super"], "albert/modeling_albert.py:AlbertForMaskedLM.get_output_embeddings": ["Linear", "ModelForMaskedLM", "ModelPreTrainedModel", "class", "decoder", "def", "get_output_embeddings", "nn", "predictions", "return", "self"], "albert/modeling_albert.py:AlbertForMaskedLM.set_output_embeddings": ["Linear", "ModelForMaskedLM", "ModelPreTrainedModel", "None", "bias", "class", "decoder", "def", "new_embeddings", "nn", "predictions", "self", "set_output_embeddings"], "albert/modeling_albert.py:AlbertForMaskedLM.get_input_embeddings": ["Embedding", "Model", "ModelForMaskedLM", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "nn", "return", "self", "word_embeddings"], "albert/modeling_albert.py:AlbertForMaskedLM.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "MaskedLMOutput", "Model", "ModelForMaskedLM", "ModelPreTrainedModel", "None", "TransformersKwargs", "True", "Unpack", "attention_mask", "attentions", "class", "config", "def", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "masked_lm_loss", "not", "outputs", "position_ids", "prediction_scores", "predictions", "r", "return", "return_dict", "self", "sequence_outputs", "token_type_ids", "torch", "tuple", "view", "vocab_size"], "albert/modeling_albert.py:AlbertForSequenceClassification.__init__": ["Dropout", "Linear", "Model", "ModelConfig", "ModelForSequenceClassification", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "classifier", "classifier_dropout_prob", "config", "def", "dropout", "hidden_size", "nn", "num_labels", "post_init", "self", "super"], "albert/modeling_albert.py:AlbertForSequenceClassification.forward": ["BCEWithLogitsLoss", "CrossEntropyLoss", "FloatTensor", "LongTensor", "MSELoss", "Model", "ModelForSequenceClassification", "ModelPreTrainedModel", "None", "SequenceClassifierOutput", "TransformersKwargs", "True", "Unpack", "and", "attention_mask", "attentions", "class", "classifier", "config", "def", "dropout", "dtype", "elif", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "kwargs", "labels", "logits", "long", "loss", "loss_fct", "multi_label_classification", "not", "num_labels", "or", "outputs", "pooled_output", "position_ids", "problem_type", "r", "regression", "return", "return_dict", "self", "single_label_classification", "squeeze", "token_type_ids", "torch", "tuple", "view"], "albert/modeling_albert.py:AlbertForTokenClassification.__init__": ["Dropout", "False", "Linear", "Model", "ModelConfig", "ModelForTokenClassification", "ModelModel", "ModelPreTrainedModel", "None", "__init__", "add_pooling_layer", "class", "classifier", "classifier_dropout_prob", "config", "def", "dropout", "else", "hidden_dropout_prob", "hidden_size", "if", "is", "nn", "not", "num_labels", "post_init", "self", "super"], "albert/modeling_albert.py:AlbertForTokenClassification.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "Model", "ModelForTokenClassification", "ModelPreTrainedModel", "None", "TokenClassifierOutput", "TransformersKwargs", "True", "Unpack", "attention_mask", "attentions", "class", "classifier", "def", "dropout", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "not", "num_labels", "outputs", "position_ids", "r", "return", "return_dict", "self", "sequence_output", "token_type_ids", "torch", "tuple", "view"], "albert/modeling_albert.py:AlbertForQuestionAnswering.__init__": ["False", "Linear", "Model", "ModelConfig", "ModelForQuestionAnswering", "ModelModel", "ModelPreTrainedModel", "__init__", "add_pooling_layer", "class", "config", "def", "hidden_size", "nn", "num_labels", "post_init", "qa_outputs", "self", "super"], "albert/modeling_albert.py:AlbertForQuestionAnswering.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "Model", "ModelForPreTrainingOutput", "ModelForQuestionAnswering", "ModelPreTrainedModel", "None", "QuestionAnsweringModelOutput", "Tensor", "TransformersKwargs", "True", "Unpack", "and", "attention_mask", "attentions", "clamp", "class", "contiguous", "def", "dim", "end_logits", "end_loss", "end_positions", "forward", "hidden_states", "if", "ignore_index", "ignored_index", "input_ids", "inputs_embeds", "is", "kwargs", "len", "logits", "loss", "loss_fct", "not", "outputs", "position_ids", "qa_outputs", "return", "return_dict", "self", "sequence_output", "size", "split", "squeeze", "start_logits", "start_loss", "start_positions", "token_type_ids", "torch", "total_loss", "tuple"], "albert/modeling_albert.py:AlbertForMultipleChoice.__init__": ["Dropout", "Linear", "Model", "ModelConfig", "ModelForMultipleChoice", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "classifier", "classifier_dropout_prob", "config", "def", "dropout", "hidden_size", "nn", "post_init", "self", "super"], "albert/modeling_albert.py:AlbertForMultipleChoice.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "Model", "ModelForMultipleChoice", "ModelForPreTrainingOutput", "ModelPreTrainedModel", "MultipleChoiceModelOutput", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "attention_mask", "attentions", "class", "classifier", "def", "dropout", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "not", "num_choices", "outputs", "pooled_output", "position_ids", "r", "reshaped_logits", "return", "return_dict", "self", "shape", "size", "token_type_ids", "torch", "tuple", "view"], "align/modeling_align.py:AlignOutput.to_tuple": ["Any", "ModelOutput", "class", "def", "else", "for", "getattr", "if", "in", "k", "keys", "not", "return", "self", "text_model_output", "to_tuple", "tuple", "vision_model_output"], "align/modeling_align.py:contrastive_loss": ["Model_loss", "Tensor", "arange", "cross_entropy", "def", "device", "functional", "label_smoothing", "len", "logits", "nn", "return", "torch"], "align/modeling_align.py:align_loss": ["Model_loss", "Tensor", "caption_loss", "contrastive_loss", "def", "image_loss", "return", "similarity", "t", "torch"], "align/modeling_align.py:round_filters": ["ModelVisionConfig", "Model_filters", "config", "def", "depth_divisor", "divisor", "if", "int", "max", "new_dim", "num_channels", "r", "return", "width_coefficient"], "align/modeling_align.py:correct_pad": ["Model", "Model_pad", "True", "adjust", "bool", "def", "else", "if", "int", "isinstance", "kernel_size", "r", "return", "tuple"], "align/modeling_align.py:AlignVisionEmbeddings.__init__": ["ACT2FN", "BatchNorm2d", "Conv2d", "False", "ModelVisionConfig", "ModelVisionEmbeddings", "Module", "ZeroPad2d", "__init__", "activation", "batch_norm_eps", "batch_norm_momentum", "batchnorm", "bias", "class", "config", "convolution", "def", "eps", "hidden_act", "kernel_size", "momentum", "nn", "num_channels", "out_dim", "padding", "round_filters", "self", "stride", "super", "valid"], "align/modeling_align.py:AlignVisionEmbeddings.forward": ["ModelVisionEmbeddings", "Module", "Tensor", "activation", "batchnorm", "class", "convolution", "def", "features", "forward", "nn", "padding", "pixel_values", "return", "self", "torch"], "align/modeling_align.py:AlignVisionDepthwiseConv2d.__init__": ["Conv2d", "ModelVisionDepthwiseConv2d", "True", "__init__", "bias", "class", "def", "depth_multiplier", "dilation", "groups", "in_channels", "kernel_size", "nn", "out_channels", "padding", "padding_mode", "self", "stride", "super", "zeros"], "align/modeling_align.py:AlignVisionExpansionLayer.__init__": ["ACT2FN", "BatchNorm2d", "Conv2d", "False", "ModelVisionConfig", "ModelVisionExpansionLayer", "Module", "__init__", "batch_norm_eps", "bias", "class", "config", "def", "eps", "expand_act", "expand_bn", "expand_conv", "hidden_act", "in_channels", "in_dim", "int", "kernel_size", "nn", "num_features", "out_channels", "out_dim", "padding", "same", "self", "stride", "super"], "align/modeling_align.py:AlignVisionExpansionLayer.forward": ["FloatTensor", "ModelVisionExpansionLayer", "Module", "Tensor", "class", "def", "expand_act", "expand_bn", "expand_conv", "forward", "hidden_states", "nn", "return", "self", "torch"], "align/modeling_align.py:AlignVisionDepthwiseLayer.__init__": ["ACT2FN", "BatchNorm2d", "False", "ModelVisionConfig", "ModelVisionDepthwiseConv2d", "ModelVisionDepthwiseLayer", "Module", "ZeroPad2d", "__init__", "adjust", "adjust_padding", "batch_norm_eps", "batch_norm_momentum", "bias", "bool", "class", "config", "conv_pad", "correct_pad", "def", "depthwise_act", "depthwise_conv", "depthwise_conv_pad", "depthwise_norm", "else", "eps", "hidden_act", "if", "in_dim", "int", "kernel_size", "momentum", "nn", "num_features", "padding", "same", "self", "stride", "super", "valid"], "align/modeling_align.py:AlignVisionDepthwiseLayer.forward": ["FloatTensor", "ModelVisionDepthwiseLayer", "Module", "Tensor", "class", "def", "depthwise_act", "depthwise_conv", "depthwise_conv_pad", "depthwise_norm", "forward", "hidden_states", "if", "nn", "return", "self", "stride", "torch"], "align/modeling_align.py:AlignVisionSqueezeExciteLayer.__init__": ["ACT2FN", "AdaptiveAvgPool2d", "Conv2d", "False", "ModelVisionConfig", "ModelVisionSqueezeExciteLayer", "Module", "Sigmoid", "__init__", "act_expand", "act_reduce", "bool", "class", "config", "def", "dim", "dim_se", "else", "expand", "expand_dim", "hidden_act", "if", "in_channels", "in_dim", "int", "kernel_size", "max", "nn", "out_channels", "output_size", "padding", "reduce", "same", "self", "squeeze", "squeeze_expansion_ratio", "super"], "align/modeling_align.py:AlignVisionSqueezeExciteLayer.forward": ["FloatTensor", "ModelVisionSqueezeExciteLayer", "Module", "Tensor", "act_expand", "act_reduce", "class", "def", "expand", "forward", "hidden_states", "inputs", "mul", "nn", "reduce", "return", "self", "squeeze", "torch"], "align/modeling_align.py:AlignVisionFinalBlockLayer.__init__": ["BatchNorm2d", "Conv2d", "Dropout", "False", "ModelVisionConfig", "ModelVisionFinalBlockLayer", "Module", "__init__", "and", "apply_dropout", "batch_norm_eps", "batch_norm_momentum", "bias", "bool", "class", "config", "def", "drop_rate", "dropout", "eps", "float", "id_skip", "in_channels", "in_dim", "int", "kernel_size", "momentum", "nn", "not", "num_features", "out_channels", "out_dim", "p", "padding", "project_bn", "project_conv", "same", "self", "stride", "super"], "align/modeling_align.py:AlignVisionFinalBlockLayer.forward": ["FloatTensor", "ModelVisionFinalBlockLayer", "Module", "Tensor", "apply_dropout", "class", "def", "dropout", "embeddings", "forward", "hidden_states", "if", "nn", "project_bn", "project_conv", "return", "self", "torch"], "align/modeling_align.py:AlignVisionBlock.__init__": ["ModelVisionBlock", "ModelVisionConfig", "ModelVisionDepthwiseLayer", "ModelVisionExpansionLayer", "ModelVisionFinalBlockLayer", "ModelVisionSqueezeExciteLayer", "Module", "__init__", "adjust_padding", "bool", "class", "config", "def", "depthwise_conv", "drop_rate", "else", "expand", "expand_dim", "expand_in_dim", "expand_ratio", "expansion", "float", "id_skip", "if", "in_dim", "int", "kernel_size", "nn", "out_dim", "projection", "self", "squeeze_excite", "stride", "super"], "align/modeling_align.py:AlignVisionBlock.forward": ["FloatTensor", "ModelVisionBlock", "Module", "Tensor", "class", "def", "depthwise_conv", "embeddings", "expand_ratio", "expansion", "forward", "hidden_states", "if", "nn", "projection", "return", "self", "squeeze_excite", "torch"], "align/modeling_align.py:AlignVisionEncoder.__init__": ["ModelVisionBlock", "ModelVisionConfig", "ModelVisionEncoder", "Module", "ModuleList", "__init__", "adjust_padding", "append", "block", "blocks", "ceil", "class", "config", "curr_block_num", "def", "depth_coefficient", "depthwise_padding", "drop_connect_rate", "drop_rate", "else", "expand_ratio", "expand_ratios", "for", "i", "id_skip", "if", "in", "in_channels", "in_dim", "int", "j", "kernel_size", "kernel_sizes", "len", "math", "n", "nn", "not", "num_base_blocks", "num_block_repeats", "num_blocks", "out_channels", "out_dim", "range", "repeats", "return", "round_filters", "round_repeats", "self", "stride", "strides", "sum", "super"], "align/modeling_align.py:AlignVisionEncoder.forward": ["BaseModelOutputWithNoAttention", "BaseModelOutputWithPoolingAndNoAttention", "False", "FloatTensor", "ModelVisionEncoder", "Module", "None", "True", "all_hidden_states", "block", "blocks", "bool", "class", "def", "else", "for", "forward", "hidden_states", "if", "in", "is", "last_hidden_state", "nn", "not", "output_hidden_states", "return", "return_dict", "self", "torch", "tuple", "v"], "align/modeling_align.py:AlignTextEmbeddings.__init__": ["Dropout", "Embedding", "False", "LayerNorm", "ModelTextEmbeddings", "Module", "__init__", "arange", "class", "config", "def", "dropout", "dtype", "eps", "expand", "hidden_dropout_prob", "hidden_size", "layer_norm_eps", "long", "max_position_embeddings", "nn", "pad_token_id", "padding_idx", "persistent", "position_embeddings", "position_ids", "register_buffer", "self", "size", "super", "token_type_embeddings", "token_type_ids", "torch", "type_vocab_size", "vocab_size", "word_embeddings", "zeros"], "align/modeling_align.py:AlignTextEmbeddings.forward": ["FloatTensor", "LayerNorm", "LongTensor", "ModelTextEmbeddings", "Module", "None", "Tensor", "buffered_token_type_ids", "buffered_token_type_ids_expanded", "class", "def", "device", "dropout", "dtype", "else", "embeddings", "expand", "forward", "hasattr", "if", "input_ids", "input_shape", "inputs_embeds", "is", "long", "nn", "not", "position_embeddings", "position_ids", "return", "self", "seq_length", "size", "token_type_embeddings", "token_type_ids", "torch", "word_embeddings", "zeros"], "align/modeling_align.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value"], "align/modeling_align.py:AlignTextSelfAttention.__init__": ["Dropout", "Linear", "ModelTextSelfAttention", "Module", "The", "ValueError", "__init__", "a", "all_head_size", "and", "attention", "attention_dropout", "attention_head_size", "attention_probs_dropout_prob", "class", "config", "def", "dropout", "embedding_size", "f", "hasattr", "heads", "hidden", "hidden_size", "if", "int", "is", "key", "multiple", "nn", "not", "num_attention_heads", "number", "of", "query", "raise", "scaling", "self", "size", "super", "the", "value"], "align/modeling_align.py:AlignTextSelfAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "False", "FloatTensor", "ModelTextSelfAttention", "Module", "None", "Tensor", "_attn_implementation", "attention_dropout", "attention_head_size", "attention_interface", "attention_mask", "attn_output", "attn_weights", "bool", "class", "config", "contiguous", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "hidden_shape", "hidden_states", "if", "input_shape", "key", "key_states", "kwargs", "nn", "not", "output_attentions", "outputs", "query", "query_states", "reshape", "return", "scaling", "self", "shape", "torch", "training", "transpose", "tuple", "value", "value_states", "view"], "align/modeling_align.py:AlignTextSelfOutput.__init__": ["Dropout", "LayerNorm", "Linear", "ModelTextSelfOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "eps", "hidden_dropout_prob", "hidden_size", "layer_norm_eps", "nn", "self", "super"], "align/modeling_align.py:AlignTextSelfOutput.forward": ["LayerNorm", "ModelTextSelfOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "align/modeling_align.py:AlignTextAttention.__init__": ["ModelTextAttention", "ModelTextSelfAttention", "ModelTextSelfOutput", "Module", "__init__", "class", "config", "def", "nn", "output", "self", "super"], "align/modeling_align.py:AlignTextAttention.forward": ["False", "FloatTensor", "ModelTextAttention", "Module", "None", "Tensor", "attention_mask", "attention_output", "bool", "class", "def", "forward", "hidden_states", "kwargs", "nn", "output", "output_attentions", "outputs", "return", "self", "self_outputs", "torch", "tuple"], "align/modeling_align.py:AlignTextIntermediate.__init__": ["ACT2FN", "Linear", "ModelTextIntermediate", "Module", "__init__", "class", "config", "def", "dense", "else", "hidden_act", "hidden_size", "if", "intermediate_act_fn", "intermediate_size", "isinstance", "nn", "self", "str", "super"], "align/modeling_align.py:AlignTextIntermediate.forward": ["ModelTextIntermediate", "Module", "Tensor", "class", "def", "dense", "forward", "hidden_states", "intermediate_act_fn", "nn", "return", "self", "torch"], "align/modeling_align.py:AlignTextOutput.__init__": ["Dropout", "LayerNorm", "Linear", "ModelTextOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "eps", "hidden_dropout_prob", "hidden_size", "intermediate_size", "layer_norm_eps", "nn", "self", "super"], "align/modeling_align.py:AlignTextOutput.forward": ["LayerNorm", "ModelTextOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "align/modeling_align.py:AlignTextLayer.__init__": ["GradientCheckpointingLayer", "ModelTextAttention", "ModelTextIntermediate", "ModelTextLayer", "ModelTextOutput", "__init__", "attention", "chunk_size_feed_forward", "class", "config", "def", "intermediate", "output", "self", "seq_len_dim", "super"], "align/modeling_align.py:AlignTextLayer.forward": ["False", "FloatTensor", "GradientCheckpointingLayer", "ModelTextLayer", "None", "Tensor", "apply_chunking_to_forward", "attention", "attention_mask", "attention_output", "bool", "chunk_size_feed_forward", "class", "def", "feed_forward_chunk", "forward", "hidden_states", "kwargs", "layer_output", "output_attentions", "outputs", "return", "self", "self_attention_outputs", "seq_len_dim", "torch", "tuple"], "align/modeling_align.py:AlignTextLayer.feed_forward_chunk": ["GradientCheckpointingLayer", "ModelTextLayer", "attention_output", "class", "def", "feed_forward_chunk", "intermediate", "intermediate_output", "layer_output", "output", "return", "self"], "align/modeling_align.py:AlignTextEncoder.__init__": ["False", "ModelTextEncoder", "ModelTextLayer", "Module", "ModuleList", "__init__", "class", "config", "def", "for", "gradient_checkpointing", "i", "in", "layer", "nn", "num_hidden_layers", "range", "self", "super"], "align/modeling_align.py:AlignTextEncoder.forward": ["BaseModelOutput", "False", "FloatTensor", "ModelTextEncoder", "Module", "None", "Tensor", "True", "all_hidden_states", "all_self_attentions", "attention_mask", "attentions", "bool", "class", "def", "else", "enumerate", "for", "forward", "hidden_states", "i", "if", "in", "kwargs", "last_hidden_state", "layer", "layer_module", "layer_outputs", "nn", "output_attentions", "output_hidden_states", "return", "return_dict", "self", "torch", "tuple"], "align/modeling_align.py:AlignTextPooler.__init__": ["Linear", "ModelTextPooler", "Module", "Tanh", "__init__", "activation", "class", "config", "def", "dense", "hidden_size", "nn", "self", "super"], "align/modeling_align.py:AlignTextPooler.forward": ["ModelTextPooler", "Module", "Tensor", "activation", "class", "def", "dense", "first_token_tensor", "forward", "hidden_states", "nn", "pooled_output", "return", "self", "torch"], "align/modeling_align.py:AlignPreTrainedModel._init_weights": ["BatchNorm2d", "Conv2d", "Embedding", "False", "LayerNorm", "Linear", "ModelModel", "ModelPreTrainedModel", "ModelTextEmbeddings", "Module", "None", "PreTrainedModel", "_init_weights", "_is_hf_initialized", "and", "arange", "bias", "class", "config", "constant_", "copy_", "def", "elif", "expand", "getattr", "if", "init", "initializer_range", "is", "isinstance", "mean", "module", "nn", "normal_", "not", "num_batches_tracked", "ones_", "padding_idx", "position_ids", "running_mean", "running_var", "self", "shape", "std", "temperature", "temperature_init_value", "text_projection", "token_type_ids", "torch", "weight", "xavier_uniform_", "zeros_"], "align/modeling_align.py:AlignTextModel.__init__": ["ModelPreTrainedModel", "ModelTextConfig", "ModelTextEmbeddings", "ModelTextEncoder", "ModelTextModel", "ModelTextPooler", "None", "True", "__init__", "add_pooling_layer", "bool", "class", "config", "def", "else", "embeddings", "encoder", "if", "pooler", "post_init", "r", "self", "super"], "align/modeling_align.py:AlignTextModel.get_input_embeddings": ["ModelPreTrainedModel", "ModelTextModel", "class", "def", "embeddings", "get_input_embeddings", "return", "self", "word_embeddings"], "align/modeling_align.py:AlignTextModel.set_input_embeddings": ["ModelPreTrainedModel", "ModelTextModel", "class", "def", "embeddings", "self", "set_input_embeddings", "value", "word_embeddings"], "align/modeling_align.py:AlignTextModel.forward": ["BaseModelOutputWithPooling", "ModelPreTrainedModel", "ModelTextModel", "None", "Tensor", "True", "ValueError", "You", "and", "at", "attention_mask", "attentions", "batch_size", "bool", "both", "buffered_token_type_ids", "buffered_token_type_ids_expanded", "cannot", "class", "config", "def", "device", "dtype", "either", "elif", "else", "embedding_output", "embeddings", "encoder", "encoder_outputs", "expand", "extended_attention_mask", "forward", "get_extended_attention_mask", "hasattr", "have", "hidden_states", "if", "input_ids", "input_shape", "inputs_embeds", "is", "kwargs", "last_hidden_state", "long", "not", "ones", "or", "output_attentions", "output_hidden_states", "pooled_output", "pooler", "pooler_output", "position_ids", "r", "raise", "return", "return_dict", "same", "self", "seq_length", "sequence_output", "size", "specify", "the", "time", "to", "token_type_ids", "torch", "tuple", "use_return_dict", "warn_if_padding_and_no_attention_mask", "zeros"], "align/modeling_align.py:AlignVisionModel.__init__": ["AvgPool2d", "MaxPool2d", "ModelPreTrainedModel", "ModelVisionConfig", "ModelVisionEmbeddings", "ModelVisionEncoder", "ModelVisionModel", "True", "ValueError", "__init__", "be", "ceil_mode", "class", "config", "def", "elif", "else", "embeddings", "encoder", "f", "got", "hidden_dim", "if", "max", "mean", "must", "nn", "of", "one", "pooler", "pooling", "pooling_type", "post_init", "raise", "self", "super"], "align/modeling_align.py:AlignVisionModel.forward": ["BaseModelOutputWithPoolingAndNoAttention", "FloatTensor", "ModelPreTrainedModel", "ModelVisionModel", "None", "True", "ValueError", "You", "bool", "class", "config", "def", "else", "embedding_output", "embeddings", "encoder", "encoder_outputs", "forward", "have", "hidden_states", "if", "is", "kwargs", "last_hidden_state", "not", "output_hidden_states", "pixel_values", "pooled_output", "pooler", "pooler_output", "r", "raise", "reshape", "return", "return_dict", "self", "shape", "specify", "to", "torch", "tuple", "use_return_dict"], "align/modeling_align.py:AlignModel.__init__": ["Linear", "ModelConfig", "ModelModel", "ModelPreTrainedModel", "ModelTextConfig", "ModelTextModel", "ModelVisionConfig", "ModelVisionModel", "Parameter", "TypeError", "__init__", "be", "but", "class", "config", "def", "expected", "f", "hidden_size", "if", "is", "isinstance", "nn", "not", "of", "post_init", "projection_dim", "raise", "self", "super", "temperature", "temperature_init_value", "tensor", "text_config", "text_embed_dim", "text_model", "text_projection", "to", "torch", "type", "vision_config", "vision_model"], "align/modeling_align.py:AlignModel.get_text_features": ["FloatTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "attention_mask", "class", "def", "get_text_features", "input_ids", "inputs_embeds", "last_hidden_state", "position_ids", "r", "return", "self", "text_features", "text_model", "text_outputs", "text_projection", "token_type_ids", "torch"], "align/modeling_align.py:AlignModel.get_image_features": ["FloatTensor", "ModelModel", "ModelPreTrainedModel", "class", "def", "get_image_features", "image_features", "pixel_values", "pooler_output", "r", "return", "self", "torch", "vision_model", "vision_outputs"], "align/modeling_align.py:AlignModel.forward": ["FloatTensor", "LongTensor", "ModelModel", "ModelOutput", "ModelPreTrainedModel", "Model_loss", "None", "Tensor", "True", "attention_mask", "bool", "class", "config", "def", "dim", "else", "forward", "if", "image_embeds", "input_ids", "inputs_embeds", "is", "keepdim", "kwargs", "logits_per_image", "logits_per_text", "loss", "matmul", "norm", "not", "output_attentions", "output_hidden_states", "p", "pixel_values", "position_ids", "r", "return", "return_dict", "return_loss", "self", "t", "temperature", "text_embeds", "text_model", "text_model_output", "text_outputs", "text_projection", "token_type_ids", "torch", "tuple", "use_return_dict", "vision_model", "vision_model_output", "vision_outputs"], "altclip/modeling_altclip.py:contrastive_loss": ["Model_loss", "Tensor", "arange", "cross_entropy", "def", "device", "functional", "len", "logits", "nn", "return", "torch"], "altclip/modeling_altclip.py:clip_loss": ["Model_loss", "Tensor", "caption_loss", "contrastive_loss", "def", "image_loss", "return", "similarity", "t", "torch"], "altclip/modeling_altclip.py:AltCLIPOutput.to_tuple": ["Any", "ModelOutput", "class", "def", "else", "for", "getattr", "if", "in", "k", "keys", "not", "return", "self", "text_model_output", "to_tuple", "tuple", "vision_model_output"], "altclip/modeling_altclip.py:AltRobertaEmbeddings.__init__": ["Dropout", "Embedding", "False", "LayerNorm", "ModelRobertaEmbeddings", "Module", "__init__", "arange", "class", "config", "def", "dropout", "dtype", "eps", "expand", "hidden_dropout_prob", "hidden_size", "layer_norm_eps", "long", "max_position_embeddings", "nn", "pad_token_id", "padding_idx", "persistent", "position_embeddings", "position_ids", "register_buffer", "self", "size", "super", "token_type_embeddings", "token_type_ids", "torch", "type_vocab_size", "vocab_size", "word_embeddings", "zeros"], "altclip/modeling_altclip.py:AltRobertaEmbeddings.forward": ["FloatTensor", "LayerNorm", "LongTensor", "ModelRobertaEmbeddings", "Module", "None", "Tensor", "batch_size", "buffered_token_type_ids", "class", "create_position_ids_from_input_ids", "create_position_ids_from_inputs_embeds", "def", "device", "dim", "dropout", "dtype", "else", "embeddings", "expand", "forward", "gather", "hasattr", "if", "index", "input_ids", "input_shape", "inputs_embeds", "int", "is", "long", "nn", "not", "padding_idx", "past_key_values_length", "position_embeddings", "position_ids", "return", "self", "seq_length", "shape", "size", "token_type_embeddings", "token_type_ids", "torch", "word_embeddings", "zeros"], "altclip/modeling_altclip.py:AltRobertaEmbeddings.create_position_ids_from_inputs_embeds": ["ModelRobertaEmbeddings", "Module", "arange", "class", "create_position_ids_from_inputs_embeds", "def", "device", "dtype", "expand", "input_shape", "inputs_embeds", "long", "nn", "padding_idx", "position_ids", "return", "sequence_length", "size", "torch", "unsqueeze"], "altclip/modeling_altclip.py:AltRobertaEmbeddings.create_position_ids_from_input_ids": ["ModelRobertaEmbeddings", "Module", "class", "create_position_ids_from_input_ids", "cumsum", "def", "dim", "incremental_indices", "input_ids", "int", "long", "mask", "ne", "nn", "padding_idx", "past_key_values_length", "return", "torch", "type_as"], "altclip/modeling_altclip.py:AltRobertaSelfAttention.__init__": ["Dropout", "Linear", "ModelRobertaSelfAttention", "Module", "The", "ValueError", "__init__", "a", "all_head_size", "and", "attention", "attention_head_size", "attention_probs_dropout_prob", "class", "config", "def", "dropout", "embedding_size", "f", "hasattr", "heads", "hidden", "hidden_size", "if", "int", "is", "key", "multiple", "nn", "not", "num_attention_heads", "number", "of", "query", "raise", "self", "size", "super", "the", "value"], "altclip/modeling_altclip.py:AltRobertaSelfAttention.forward": ["False", "FloatTensor", "ModelRobertaSelfAttention", "Module", "None", "Tensor", "all_head_size", "attention_head_size", "attention_mask", "attention_probs", "attention_scores", "bool", "class", "context_layer", "contiguous", "def", "dim", "dropout", "else", "forward", "functional", "hidden_shape", "hidden_states", "if", "input_shape", "is", "key", "key_layer", "math", "matmul", "new_context_layer_shape", "nn", "not", "output_attentions", "outputs", "permute", "query", "query_layer", "return", "self", "shape", "size", "softmax", "sqrt", "torch", "transpose", "tuple", "value", "value_layer", "view"], "altclip/modeling_altclip.py:AltRobertaSelfOutput.__init__": ["Dropout", "LayerNorm", "Linear", "ModelRobertaSelfOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "eps", "hidden_dropout_prob", "hidden_size", "layer_norm_eps", "nn", "self", "super"], "altclip/modeling_altclip.py:AltRobertaSelfOutput.forward": ["LayerNorm", "ModelRobertaSelfOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "altclip/modeling_altclip.py:AltRobertaAttention.__init__": ["ModelRobertaAttention", "ModelRobertaSelfOutput", "Model_ROBERTA_SELF_ATTENTION_CLASSES", "Module", "__init__", "_attn_implementation", "class", "config", "def", "nn", "output", "self", "super"], "altclip/modeling_altclip.py:AltRobertaAttention.forward": ["False", "FloatTensor", "ModelRobertaAttention", "Module", "None", "Tensor", "attention_mask", "attention_output", "bool", "class", "def", "forward", "hidden_states", "nn", "output", "output_attentions", "outputs", "return", "self", "self_outputs", "torch", "tuple"], "altclip/modeling_altclip.py:AltRobertaIntermediate.__init__": ["ACT2FN", "Linear", "ModelRobertaIntermediate", "Module", "__init__", "class", "config", "def", "dense", "else", "hidden_act", "hidden_size", "if", "intermediate_act_fn", "intermediate_size", "isinstance", "nn", "self", "str", "super"], "altclip/modeling_altclip.py:AltRobertaIntermediate.forward": ["ModelRobertaIntermediate", "Module", "Tensor", "class", "def", "dense", "forward", "hidden_states", "intermediate_act_fn", "nn", "return", "self", "torch"], "altclip/modeling_altclip.py:AltRobertaOutput.__init__": ["Dropout", "LayerNorm", "Linear", "ModelRobertaOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "eps", "hidden_dropout_prob", "hidden_size", "intermediate_size", "layer_norm_eps", "nn", "self", "super"], "altclip/modeling_altclip.py:AltRobertaOutput.forward": ["LayerNorm", "ModelRobertaOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "altclip/modeling_altclip.py:AltRobertaLayer.__init__": ["GradientCheckpointingLayer", "ModelRobertaAttention", "ModelRobertaIntermediate", "ModelRobertaLayer", "ModelRobertaOutput", "__init__", "attention", "chunk_size_feed_forward", "class", "config", "def", "intermediate", "output", "self", "seq_len_dim", "super"], "altclip/modeling_altclip.py:AltRobertaLayer.forward": ["False", "FloatTensor", "GradientCheckpointingLayer", "ModelRobertaLayer", "None", "Tensor", "apply_chunking_to_forward", "attention", "attention_mask", "attention_output", "bool", "chunk_size_feed_forward", "class", "def", "feed_forward_chunk", "forward", "hidden_states", "kwargs", "layer_output", "output_attentions", "outputs", "return", "self", "self_attention_outputs", "seq_len_dim", "torch", "tuple"], "altclip/modeling_altclip.py:AltRobertaLayer.feed_forward_chunk": ["GradientCheckpointingLayer", "ModelRobertaLayer", "attention_output", "class", "def", "feed_forward_chunk", "intermediate", "intermediate_output", "layer_output", "output", "return", "self"], "altclip/modeling_altclip.py:AltRobertaEncoder.__init__": ["False", "ModelRobertaEncoder", "ModelRobertaLayer", "Module", "ModuleList", "__init__", "class", "config", "def", "for", "gradient_checkpointing", "i", "in", "layer", "nn", "num_hidden_layers", "range", "self", "super"], "altclip/modeling_altclip.py:AltRobertaEncoder.forward": ["BaseModelOutput", "False", "FloatTensor", "ModelRobertaEncoder", "Module", "None", "Tensor", "True", "all_hidden_states", "all_self_attentions", "attention_mask", "attentions", "bool", "class", "def", "else", "enumerate", "for", "forward", "hidden_states", "i", "if", "in", "kwargs", "last_hidden_state", "layer", "layer_module", "layer_outputs", "nn", "output_attentions", "output_hidden_states", "return", "return_dict", "self", "torch", "tuple"], "altclip/modeling_altclip.py:AltRobertaPooler.__init__": ["Linear", "ModelRobertaPooler", "Module", "Tanh", "__init__", "activation", "class", "config", "def", "dense", "hidden_size", "nn", "self", "super"], "altclip/modeling_altclip.py:AltRobertaPooler.forward": ["ModelRobertaPooler", "Module", "Tensor", "activation", "class", "def", "dense", "first_token_tensor", "forward", "hidden_states", "nn", "pooled_output", "return", "self", "torch"], "altclip/modeling_altclip.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "attention_mask", "attn_output", "attn_weights", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "softmax", "to", "torch", "training", "transpose", "value"], "altclip/modeling_altclip.py:AltCLIPAttention.__init__": ["False", "Linear", "ModelAttention", "Module", "ValueError", "__init__", "and", "attention_dropout", "be", "by", "class", "config", "def", "divisible", "dropout", "embed_dim", "f", "got", "head_dim", "hidden_size", "if", "is_causal", "k_proj", "must", "nn", "num_attention_heads", "num_heads", "out_proj", "q_proj", "raise", "scale", "self", "super", "v_proj"], "altclip/modeling_altclip.py:AltCLIPAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "False", "ModelAttention", "Module", "None", "Tensor", "_attn_implementation", "and", "attention_interface", "attention_mask", "attn_output", "attn_weights", "batch_size", "bool", "causal_attention_mask", "class", "config", "contiguous", "def", "dropout", "eager", "eager_attention_forward", "elif", "else", "embed_dim", "flash_attention_2", "forward", "head_dim", "hidden_states", "if", "is", "is_causal", "k_proj", "keys", "nn", "not", "num_heads", "out_proj", "output_attentions", "q_proj", "queries", "reshape", "return", "scale", "scaling", "self", "seq_length", "shape", "torch", "training", "transpose", "tuple", "v_proj", "values", "view"], "altclip/modeling_altclip.py:AltCLIPMLP.__init__": ["ACT2FN", "Linear", "ModelMLP", "Module", "__init__", "activation_fn", "class", "config", "def", "fc1", "fc2", "hidden_act", "hidden_size", "intermediate_size", "nn", "self", "super"], "altclip/modeling_altclip.py:AltCLIPMLP.forward": ["ModelMLP", "Module", "Tensor", "activation_fn", "class", "def", "fc1", "fc2", "forward", "hidden_states", "nn", "return", "self", "torch"], "altclip/modeling_altclip.py:AltCLIPEncoderLayer.__init__": ["GradientCheckpointingLayer", "LayerNorm", "ModelAttention", "ModelConfig", "ModelEncoderLayer", "ModelMLP", "__init__", "class", "config", "def", "embed_dim", "eps", "hidden_size", "layer_norm1", "layer_norm2", "layer_norm_eps", "mlp", "nn", "self", "self_attn", "super"], "altclip/modeling_altclip.py:AltCLIPEncoderLayer.forward": ["False", "FloatTensor", "GradientCheckpointingLayer", "ModelEncoderLayer", "None", "Tensor", "attention_mask", "attn_weights", "bool", "causal_attention_mask", "class", "def", "forward", "hidden_states", "if", "layer_norm1", "layer_norm2", "mlp", "output_attentions", "outputs", "residual", "return", "self", "self_attn", "torch", "tuple"], "altclip/modeling_altclip.py:AltCLIPEncoder.__init__": ["False", "ModelConfig", "ModelEncoder", "ModelEncoderLayer", "Module", "ModuleList", "_", "__init__", "class", "config", "def", "for", "gradient_checkpointing", "in", "layers", "nn", "num_hidden_layers", "range", "self", "super"], "altclip/modeling_altclip.py:AltCLIPEncoder.forward": ["BaseModelOutput", "ModelEncoder", "Module", "None", "Tensor", "all_attentions", "attention_mask", "attentions", "bool", "causal_attention_mask", "class", "config", "def", "else", "encoder_layer", "encoder_states", "enumerate", "for", "forward", "hidden_states", "idx", "if", "in", "inputs_embeds", "is", "last_hidden_state", "layer_outputs", "layers", "nn", "not", "output_attentions", "output_hidden_states", "r", "return", "return_dict", "self", "torch", "tuple", "use_return_dict"], "altclip/modeling_altclip.py:AltCLIPVisionEmbeddings.__init__": ["Conv2d", "Embedding", "False", "ModelVisionConfig", "ModelVisionEmbeddings", "Module", "Parameter", "__init__", "arange", "bias", "class", "class_embedding", "config", "def", "embed_dim", "expand", "hidden_size", "image_size", "in_channels", "kernel_size", "nn", "num_channels", "num_patches", "num_positions", "out_channels", "patch_embedding", "patch_size", "persistent", "position_embedding", "position_ids", "randn", "register_buffer", "self", "stride", "super", "torch"], "altclip/modeling_altclip.py:AltCLIPVisionEmbeddings.interpolate_pos_encoding": ["False", "ModelVisionEmbeddings", "Module", "Tensor", "align_corners", "and", "bicubic", "cat", "class", "class_pos_embed", "def", "dim", "embeddings", "functional", "height", "if", "int", "interpolate", "interpolate_pos_encoding", "is_tracing", "jit", "mode", "new_height", "new_width", "nn", "not", "num_patches", "num_positions", "patch_pos_embed", "patch_size", "permute", "position_embedding", "position_ids", "reshape", "return", "self", "shape", "size", "sqrt_num_positions", "torch", "torch_int", "unsqueeze", "view", "weight", "width"], "altclip/modeling_altclip.py:AltCLIPVisionEmbeddings.forward": ["False", "FloatTensor", "Input", "ModelVisionEmbeddings", "Module", "Tensor", "ValueError", "_", "and", "batch_size", "cat", "class", "class_embedding", "class_embeds", "def", "dim", "doesn", "dtype", "else", "embeddings", "expand", "f", "flatten", "forward", "height", "if", "image", "image_size", "interpolate_pos_encoding", "match", "model", "nn", "not", "or", "patch_embedding", "patch_embeds", "pixel_values", "position_embedding", "position_ids", "raise", "return", "self", "shape", "size", "t", "target_dtype", "to", "torch", "transpose", "weight", "width"], "altclip/modeling_altclip.py:AltCLIPPreTrainedModel._init_weights": ["Embedding", "False", "LayerNorm", "Linear", "ModelAttention", "ModelMLP", "ModelModel", "ModelPreTrainedModel", "ModelRobertaEmbeddings", "ModelVisionEmbeddings", "None", "PreTrainedModel", "_init_weights", "_is_hf_initialized", "and", "arange", "bias", "class", "class_embedding", "config", "copy_", "def", "elif", "embed_dim", "expand", "factor", "fc1", "fc2", "fc_std", "getattr", "hidden_size", "if", "in_proj_std", "init", "initializer_factor", "initializer_range", "is", "isinstance", "k_proj", "mean", "module", "nn", "normal_", "not", "num_hidden_layers", "num_positions", "ones_", "out_proj", "out_proj_std", "padding_idx", "patch_embedding", "position_embedding", "position_ids", "q_proj", "self", "shape", "std", "text_embed_dim", "text_projection", "token_type_ids", "torch", "v_proj", "vision_embed_dim", "visual_projection", "weight", "zeros_"], "altclip/modeling_altclip.py:AltCLIPVisionTransformer.__init__": ["LayerNorm", "ModelEncoder", "ModelVisionConfig", "ModelVisionEmbeddings", "ModelVisionTransformer", "Module", "__init__", "class", "config", "def", "embed_dim", "embeddings", "encoder", "eps", "hidden_size", "layer_norm_eps", "nn", "post_layernorm", "pre_layrnorm", "self", "super"], "altclip/modeling_altclip.py:AltCLIPVisionTransformer.forward": ["BaseModelOutputWithPooling", "False", "FloatTensor", "ModelVisionTransformer", "Module", "None", "True", "ValueError", "You", "attentions", "bool", "class", "config", "def", "else", "embeddings", "encoder", "encoder_outputs", "forward", "have", "hidden_states", "if", "inputs_embeds", "interpolate_pos_encoding", "is", "last_hidden_state", "nn", "not", "output_attentions", "output_hidden_states", "pixel_values", "pooled_output", "pooler_output", "post_layernorm", "pre_layrnorm", "raise", "return", "return_dict", "self", "specify", "to", "torch", "tuple", "use_return_dict"], "altclip/modeling_altclip.py:AltCLIPVisionModel.__init__": ["ModelPreTrainedModel", "ModelVisionConfig", "ModelVisionModel", "ModelVisionTransformer", "__init__", "class", "config", "def", "post_init", "self", "super", "vision_model"], "altclip/modeling_altclip.py:AltCLIPVisionModel.get_input_embeddings": ["ModelPreTrainedModel", "ModelVisionModel", "Module", "class", "def", "embeddings", "get_input_embeddings", "nn", "patch_embedding", "return", "self", "vision_model"], "altclip/modeling_altclip.py:AltCLIPVisionModel.forward": ["BaseModelOutputWithPooling", "False", "FloatTensor", "ModelPreTrainedModel", "ModelVisionModel", "None", "bool", "class", "config", "def", "else", "forward", "if", "interpolate_pos_encoding", "is", "kwargs", "not", "output_attentions", "output_hidden_states", "pixel_values", "r", "return", "return_dict", "self", "torch", "tuple", "use_return_dict", "vision_model"], "altclip/modeling_altclip.py:AltRobertaModel.__init__": ["ModelPreTrainedModel", "ModelRobertaEmbeddings", "ModelRobertaEncoder", "ModelRobertaModel", "ModelRobertaPooler", "None", "True", "__init__", "add_pooling_layer", "class", "config", "def", "else", "embeddings", "encoder", "if", "pooler", "post_init", "r", "self", "super"], "altclip/modeling_altclip.py:AltRobertaModel.get_input_embeddings": ["ModelPreTrainedModel", "ModelRobertaModel", "class", "def", "embeddings", "get_input_embeddings", "return", "self", "word_embeddings"], "altclip/modeling_altclip.py:AltRobertaModel.set_input_embeddings": ["ModelPreTrainedModel", "ModelRobertaModel", "class", "def", "embeddings", "self", "set_input_embeddings", "value", "word_embeddings"], "altclip/modeling_altclip.py:AltRobertaModel.forward": ["BaseModelOutputWithPooling", "BaseModelOutputWithPoolingAndCrossAttentions", "ModelPreTrainedModel", "ModelRobertaModel", "None", "Tensor", "True", "ValueError", "You", "and", "at", "attention_mask", "attentions", "batch_size", "bool", "both", "buffered_token_type_ids", "buffered_token_type_ids_expanded", "cannot", "class", "config", "def", "device", "dtype", "either", "elif", "else", "embedding_output", "embeddings", "encoder", "encoder_outputs", "expand", "extended_attention_mask", "forward", "get_extended_attention_mask", "hasattr", "have", "hidden_states", "if", "input_ids", "input_shape", "inputs_embeds", "is", "kwargs", "last_hidden_state", "long", "not", "ones", "or", "output_attentions", "output_hidden_states", "pooled_output", "pooler", "pooler_output", "position_ids", "raise", "return", "return_dict", "same", "self", "seq_length", "sequence_output", "size", "specify", "the", "time", "to", "token_type_ids", "torch", "tuple", "use_return_dict", "warn_if_padding_and_no_attention_mask", "zeros"], "altclip/modeling_altclip.py:AltCLIPTextModel.__init__": ["False", "LayerNorm", "Linear", "ModelPreTrainedModel", "ModelRobertaModel", "ModelTextModel", "__init__", "add_pooling_layer", "class", "config", "def", "eps", "hidden_size", "layer_norm_eps", "nn", "post_init", "pre_LN", "project_dim", "roberta", "self", "super", "transformation"], "altclip/modeling_altclip.py:AltCLIPTextModel.get_input_embeddings": ["ModelPreTrainedModel", "ModelTextModel", "Module", "class", "def", "embeddings", "get_input_embeddings", "nn", "return", "roberta", "self", "word_embeddings"], "altclip/modeling_altclip.py:AltCLIPTextModel.set_input_embeddings": ["Embedding", "ModelPreTrainedModel", "ModelTextModel", "None", "class", "def", "embeddings", "nn", "roberta", "self", "set_input_embeddings", "value", "word_embeddings"], "altclip/modeling_altclip.py:AltCLIPTextModel.resize_token_embeddings": ["Embedding", "ModelPreTrainedModel", "ModelTextModel", "None", "class", "def", "int", "new_num_tokens", "nn", "resize_token_embeddings", "return", "self", "super"], "altclip/modeling_altclip.py:AltCLIPTextModel.forward": ["BaseModelOutputWithPoolingAndProjection", "ModelPreTrainedModel", "ModelTextModel", "None", "Tensor", "True", "attention_mask", "attentions", "bool", "class", "config", "def", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "last_hidden_state", "not", "output_attentions", "output_hidden_states", "outputs", "pooler_output", "position_ids", "pre_LN", "projection_state", "r", "return", "return_dict", "roberta", "self", "sequence_output", "token_type_ids", "torch", "transformation", "tuple", "use_return_dict"], "altclip/modeling_altclip.py:AltCLIPModel.__init__": ["False", "Linear", "ModelConfig", "ModelModel", "ModelPreTrainedModel", "ModelTextConfig", "ModelTextModel", "ModelVisionConfig", "ModelVisionTransformer", "Parameter", "TypeError", "__init__", "_attn_implementation", "be", "bias", "but", "class", "config", "def", "expected", "f", "hidden_size", "if", "is", "isinstance", "logit_scale", "logit_scale_init_value", "nn", "not", "of", "post_init", "project_dim", "projection_dim", "raise", "self", "super", "tensor", "text_config", "text_embed_dim", "text_model", "text_projection", "to", "torch", "type", "vision_config", "vision_embed_dim", "vision_model", "visual_projection"], "altclip/modeling_altclip.py:AltCLIPModel.get_text_features": ["FloatTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "attention_mask", "class", "def", "get_text_features", "input_ids", "pooled_output", "pooler_output", "position_ids", "r", "return", "self", "text_features", "text_model", "text_outputs", "text_projection", "token_type_ids", "torch"], "altclip/modeling_altclip.py:AltCLIPModel.get_image_features": ["False", "FloatTensor", "ModelModel", "ModelPreTrainedModel", "bool", "class", "def", "get_image_features", "image_features", "interpolate_pos_encoding", "pixel_values", "pooled_output", "pooler_output", "r", "return", "self", "torch", "vision_model", "vision_outputs", "visual_projection"], "altclip/modeling_altclip.py:AltCLIPModel.forward": ["False", "FloatTensor", "LongTensor", "ModelModel", "ModelOutput", "ModelPreTrainedModel", "None", "T", "Tensor", "True", "attention_mask", "bool", "class", "clip_loss", "config", "def", "dim", "else", "exp", "forward", "if", "image_embeds", "input_ids", "interpolate_pos_encoding", "is", "keepdim", "kwargs", "logit_scale", "logits_per_image", "logits_per_text", "loss", "matmul", "norm", "not", "output", "output_attentions", "output_hidden_states", "p", "pixel_values", "position_ids", "r", "return", "return_dict", "return_loss", "self", "t", "text_embeds", "text_model", "text_model_output", "text_outputs", "text_projection", "token_type_ids", "torch", "tuple", "use_return_dict", "vision_model", "vision_model_output", "vision_outputs", "visual_projection"], "apertus/modeling_apertus.py:ApertusMLP.__init__": ["ACT2CLS", "ACT2FN", "False", "Linear", "ModelMLP", "Module", "__init__", "act_fn", "bias", "class", "config", "def", "down_proj", "dtype", "hidden_act", "hidden_size", "if", "intermediate_size", "nn", "self", "super", "up_proj", "xielu"], "apertus/modeling_apertus.py:ApertusMLP.forward": ["ModelMLP", "Module", "act_fn", "class", "def", "down_proj", "forward", "nn", "return", "self", "up_proj", "x"], "apertus/modeling_apertus.py:ApertusRMSNorm.__init__": ["ModelRMSNorm", "Module", "Parameter", "__init__", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "apertus/modeling_apertus.py:ApertusRMSNorm.forward": ["ModelRMSNorm", "Module", "True", "class", "def", "dtype", "float32", "forward", "hidden_states", "input_dtype", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "apertus/modeling_apertus.py:ApertusRMSNorm.extra_repr": ["ModelRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "variance_epsilon", "weight"], "apertus/modeling_apertus.py:ApertusRotaryEmbedding.__init__": ["Callable", "False", "ModelConfig", "ModelRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "apertus/modeling_apertus.py:ApertusRotaryEmbedding.compute_default_rope_parameters": ["ModelConfig", "ModelRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "apertus/modeling_apertus.py:ApertusRotaryEmbedding.forward": ["False", "ModelRotaryEmbedding", "Module", "None", "and", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "apertus/modeling_apertus.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "apertus/modeling_apertus.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cos", "def", "k", "k_embed", "q", "q_embed", "return", "rotate_half", "sin", "unsqueeze", "unsqueeze_dim"], "apertus/modeling_apertus.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "apertus/modeling_apertus.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value", "value_states"], "apertus/modeling_apertus.py:ApertusAttention.__init__": ["Linear", "ModelAttention", "ModelConfig", "ModelRMSNorm", "Module", "None", "True", "__init__", "attention_bias", "attention_dropout", "bias", "class", "config", "def", "getattr", "head_dim", "hidden_size", "int", "is_causal", "k_norm", "k_proj", "layer_idx", "nn", "num_attention_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "q_norm", "q_proj", "rms_norm_eps", "scaling", "self", "super", "v_proj"], "apertus/modeling_apertus.py:ApertusAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "LongTensor", "ModelAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_attn_implementation", "apply_rotary_pos_emb", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "cache_kwargs", "cache_position", "class", "config", "contiguous", "cos", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "head_dim", "hidden_shape", "hidden_states", "if", "input_shape", "is", "k_norm", "k_proj", "key_states", "kwargs", "layer_idx", "nn", "not", "o_proj", "past_key_values", "position_embeddings", "q_norm", "q_proj", "query_states", "reshape", "return", "scaling", "self", "shape", "sin", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "view"], "apertus/modeling_apertus.py:ApertusDecoderLayer.__init__": ["GradientCheckpointingLayer", "ModelAttention", "ModelConfig", "ModelDecoderLayer", "ModelMLP", "ModelRMSNorm", "__init__", "attention_layernorm", "class", "config", "def", "eps", "feedforward_layernorm", "hidden_size", "int", "layer_idx", "mlp", "rms_norm_eps", "self", "self_attn", "super"], "apertus/modeling_apertus.py:ApertusDecoderLayer.forward": ["Cache", "False", "GradientCheckpointingLayer", "LongTensor", "ModelDecoderLayer", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "attention_layernorm", "attention_mask", "bool", "cache_position", "class", "def", "feedforward_layernorm", "forward", "hidden_states", "kwargs", "mlp", "past_key_values", "position_embeddings", "position_ids", "residual", "return", "self", "self_attn", "torch", "tuple", "use_cache"], "apertus/modeling_apertus.py:ApertusModel.__init__": ["Embedding", "False", "ModelConfig", "ModelDecoderLayer", "ModelModel", "ModelPreTrainedModel", "ModelRMSNorm", "ModelRotaryEmbedding", "ModuleList", "__init__", "class", "config", "def", "embed_tokens", "eps", "for", "gradient_checkpointing", "hidden_size", "in", "layer_idx", "layers", "nn", "norm", "num_hidden_layers", "pad_token_id", "padding_idx", "post_init", "range", "rms_norm_eps", "rotary_emb", "self", "super", "vocab_size"], "apertus/modeling_apertus.py:ApertusModel.forward": ["BaseModelOutputWithPast", "Cache", "DynamicCache", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "and", "arange", "attention_mask", "bool", "cache_position", "causal_mask", "class", "config", "create_causal_mask", "decoder_layer", "def", "device", "else", "embed_tokens", "exactly", "for", "forward", "get_seq_length", "hidden_states", "if", "in", "input_embeds", "input_ids", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layers", "must", "norm", "not", "num_hidden_layers", "of", "one", "or", "past_key_values", "past_seen_tokens", "position_embeddings", "position_ids", "raise", "return", "rotary_emb", "self", "shape", "specify", "torch", "unsqueeze", "use_cache"], "apertus/modeling_apertus.py:ApertusForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "ModelForCausalLM", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "vocab_size"], "apertus/modeling_apertus.py:ApertusForCausalLM.forward": ["BaseModelOutputWithPast", "Cache", "CausalLMOutputWithPast", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForCausalLM", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "outputs", "past_key_values", "position_ids", "r", "return", "self", "slice", "slice_indices", "torch", "use_cache", "vocab_size"], "arcee/modeling_arcee.py:ArceeMLP.__init__": ["ACT2FN", "Linear", "ModelMLP", "Module", "__init__", "act_fn", "bias", "class", "config", "def", "down_proj", "hidden_act", "hidden_size", "intermediate_size", "mlp_bias", "nn", "self", "super", "up_proj"], "arcee/modeling_arcee.py:ArceeMLP.forward": ["ModelMLP", "Module", "act_fn", "class", "def", "down_proj", "forward", "nn", "return", "self", "up_proj", "x"], "arcee/modeling_arcee.py:ArceeRMSNorm.__init__": ["ModelRMSNorm", "Module", "Parameter", "__init__", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "arcee/modeling_arcee.py:ArceeRMSNorm.forward": ["ModelRMSNorm", "Module", "True", "class", "def", "dtype", "float32", "forward", "hidden_states", "input_dtype", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "arcee/modeling_arcee.py:ArceeRMSNorm.extra_repr": ["ModelRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "variance_epsilon", "weight"], "arcee/modeling_arcee.py:ArceeRotaryEmbedding.__init__": ["Callable", "False", "ModelConfig", "ModelRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "arcee/modeling_arcee.py:ArceeRotaryEmbedding.compute_default_rope_parameters": ["ModelConfig", "ModelRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "arcee/modeling_arcee.py:ArceeRotaryEmbedding.forward": ["False", "ModelRotaryEmbedding", "Module", "None", "and", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "arcee/modeling_arcee.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "arcee/modeling_arcee.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cos", "def", "k", "k_embed", "q", "q_embed", "return", "rotate_half", "sin", "unsqueeze", "unsqueeze_dim"], "arcee/modeling_arcee.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "arcee/modeling_arcee.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value", "value_states"], "arcee/modeling_arcee.py:ArceeAttention.__init__": ["Linear", "ModelAttention", "ModelConfig", "Module", "True", "__init__", "attention_bias", "attention_dropout", "bias", "class", "config", "def", "getattr", "head_dim", "hidden_size", "int", "is_causal", "k_proj", "layer_idx", "nn", "num_attention_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "q_proj", "scaling", "self", "super", "v_proj"], "arcee/modeling_arcee.py:ArceeAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "LongTensor", "ModelAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_attn_implementation", "apply_rotary_pos_emb", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "cache_kwargs", "cache_position", "class", "config", "contiguous", "cos", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "head_dim", "hidden_shape", "hidden_states", "if", "input_shape", "is", "k_proj", "key_states", "kwargs", "layer_idx", "nn", "not", "o_proj", "past_key_values", "position_embeddings", "q_proj", "query_states", "reshape", "return", "scaling", "self", "shape", "sin", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "view"], "arcee/modeling_arcee.py:ArceeDecoderLayer.__init__": ["GradientCheckpointingLayer", "ModelAttention", "ModelConfig", "ModelDecoderLayer", "ModelMLP", "ModelRMSNorm", "__init__", "class", "config", "def", "eps", "hidden_size", "input_layernorm", "int", "layer_idx", "mlp", "post_attention_layernorm", "rms_norm_eps", "self", "self_attn", "super"], "arcee/modeling_arcee.py:ArceeDecoderLayer.forward": ["Cache", "False", "GradientCheckpointingLayer", "LongTensor", "ModelDecoderLayer", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "attention_mask", "bool", "cache_position", "class", "def", "forward", "hidden_states", "input_layernorm", "kwargs", "mlp", "past_key_values", "position_embeddings", "position_ids", "post_attention_layernorm", "residual", "return", "self", "self_attn", "torch", "tuple", "use_cache"], "arcee/modeling_arcee.py:ArceeModel.__init__": ["Embedding", "False", "ModelConfig", "ModelDecoderLayer", "ModelModel", "ModelPreTrainedModel", "ModelRMSNorm", "ModelRotaryEmbedding", "ModuleList", "__init__", "class", "config", "def", "embed_tokens", "eps", "for", "gradient_checkpointing", "hidden_size", "in", "layer_idx", "layers", "nn", "norm", "num_hidden_layers", "pad_token_id", "padding_idx", "post_init", "range", "rms_norm_eps", "rotary_emb", "self", "super", "vocab_size"], "arcee/modeling_arcee.py:ArceeModel.forward": ["BaseModelOutputWithPast", "Cache", "DynamicCache", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "and", "arange", "attention_mask", "bool", "cache_position", "causal_mask", "class", "config", "create_causal_mask", "decoder_layer", "def", "device", "else", "embed_tokens", "exactly", "for", "forward", "get_seq_length", "hidden_states", "if", "in", "input_embeds", "input_ids", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layers", "must", "norm", "not", "num_hidden_layers", "of", "one", "or", "past_key_values", "past_seen_tokens", "position_embeddings", "position_ids", "raise", "return", "rotary_emb", "self", "shape", "specify", "torch", "unsqueeze", "use_cache"], "arcee/modeling_arcee.py:ArceeForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "ModelForCausalLM", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "vocab_size"], "arcee/modeling_arcee.py:ArceeForCausalLM.forward": ["BaseModelOutputWithPast", "Cache", "CausalLMOutputWithPast", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForCausalLM", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "outputs", "past_key_values", "position_ids", "r", "return", "self", "slice", "slice_indices", "torch", "use_cache", "vocab_size"], "aria/modeling_aria.py:AriaTextRMSNorm.__init__": ["ModelTextRMSNorm", "Module", "Parameter", "__init__", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "vModelnce_epsilon", "weight"], "aria/modeling_aria.py:AriaTextRMSNorm.forward": ["ModelTextRMSNorm", "Module", "True", "class", "def", "dtype", "float32", "forward", "hidden_states", "input_dtype", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "vModelnce", "vModelnce_epsilon", "weight"], "aria/modeling_aria.py:AriaTextRMSNorm.extra_repr": ["ModelTextRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "vModelnce_epsilon", "weight"], "aria/modeling_aria.py:AriaProjectorMLP.__init__": ["ACT2FN", "False", "Linear", "ModelProjectorMLP", "Module", "__init__", "act", "bias", "class", "def", "gelu_new", "hidden_features", "in_features", "linear_in", "linear_out", "nn", "output_dim", "self", "super"], "aria/modeling_aria.py:AriaProjectorMLP.forward": ["ModelProjectorMLP", "Module", "act", "class", "def", "forward", "hidden_states", "linear_in", "linear_out", "nn", "return", "self"], "aria/modeling_aria.py:AriaCrossAttention.__init__": ["Dropout", "False", "LayerNorm", "Linear", "ModelConfig", "ModelCrossAttention", "Module", "MultiheadAttention", "True", "__init__", "batch_first", "bias", "class", "config", "def", "dropout", "dropout_rate", "float", "hidden_size", "k_proj", "layer_norm", "layer_norm_kv", "linear", "multihead_attn", "nn", "num_attention_heads", "num_heads", "q_proj", "self", "super", "v_proj", "vision_config"], "aria/modeling_aria.py:AriaCrossAttention.forward": ["ModelCrossAttention", "Module", "None", "_", "attn_mask", "attn_output", "class", "def", "dropout", "forward", "hidden_states", "k_proj", "key", "key_value_states", "layer_norm", "layer_norm_kv", "linear", "multihead_attn", "nn", "q_proj", "query", "return", "self", "v_proj", "value"], "aria/modeling_aria.py:AriaProjector.__init__": ["LayerNorm", "ModelConfig", "ModelCrossAttention", "ModelProjector", "ModelProjectorMLP", "Module", "Parameter", "__init__", "class", "config", "cross_attn", "def", "feed_forward", "hidden_features", "hidden_size", "in_features", "kv_dim", "layer_norm", "max_value_projector_patch_to_query_dict", "nn", "num_attention_heads", "num_heads", "output_dim", "patch_to_query_dict", "projector_patch_to_query_dict", "query", "self", "super", "text_config", "torch", "vision_config", "zeros"], "aria/modeling_aria.py:AriaProjector.forward": ["KeyError", "ModelProjector", "Module", "None", "Number", "Tensor", "amongst", "attention_out", "attn_mask", "batch_size", "class", "cross_attn", "def", "expand", "f", "feed_forward", "forward", "found", "if", "in", "is", "key_value_states", "keys", "layer_norm", "nn", "not", "num_heads", "num_patches", "of", "out", "patch_to_query_dict", "patches", "possible", "queries", "query", "query_num", "raise", "repeat", "repeat_interleave", "return", "self", "shape", "size", "torch", "unsqueeze", "values"], "aria/modeling_aria.py:AriaSharedExpertsMLP.__init__": ["ACT2FN", "Linear", "ModelSharedExpertsMLP", "ModelTextConfig", "Module", "__init__", "act_fn", "bias", "class", "config", "def", "down_proj", "gate_proj", "hidden_act", "hidden_size", "intermediate_size", "mlp_bias", "moe_num_shared_experts", "nn", "self", "super", "up_proj"], "aria/modeling_aria.py:AriaSharedExpertsMLP.forward": ["ModelSharedExpertsMLP", "Module", "act_fn", "class", "def", "down_proj", "forward", "gate_proj", "nn", "return", "self", "up_proj", "x"], "aria/modeling_aria.py:sequential_experts_gemm": ["Model_experts_gemm", "cat", "cumsum", "cumsum_num_tokens", "def", "device", "dim", "dtype", "end", "expert_num", "expert_weights", "for", "in", "long", "matmul", "num_tokens", "out", "out_features", "output", "range", "return", "shape", "start", "token_states", "tokens", "tokens_per_expert", "torch", "zero_tensor", "zeros"], "aria/modeling_aria.py:AriaGroupedExpertsGemm.__init__": ["ModelGroupedExpertsGemm", "Module", "Parameter", "__init__", "class", "def", "empty", "groups", "in_features", "nn", "out_features", "self", "super", "torch", "weight"], "aria/modeling_aria.py:AriaGroupedExpertsGemm.forward": ["ModelGroupedExpertsGemm", "Module", "class", "cpu", "def", "forward", "input", "nn", "return", "self", "sequential_experts_gemm", "tokens_per_expert", "weight"], "aria/modeling_aria.py:AriaExperts.__init__": ["ModelExperts", "ModelGroupedExpertsGemm", "ModelTextConfig", "Module", "None", "__init__", "class", "config", "def", "fc1", "fc2", "hidden_size", "intermediate_size", "moe_num_experts", "nn", "self", "super"], "aria/modeling_aria.py:AriaExperts.route_tokens_to_experts": ["ModelExperts", "Module", "class", "config", "def", "dim", "functional", "k", "moe_topk", "nn", "return", "route_tokens_to_experts", "router_logits", "scores", "self", "softmax", "top_indices", "top_logits", "topk", "torch"], "aria/modeling_aria.py:AriaExperts.forward": ["ModelExperts", "Module", "Tensor", "argsort", "bins", "chunk", "class", "config", "def", "device", "dim", "dtype", "expert_output", "fc1", "fc1_output", "fc2", "flatten", "flatten_indices", "float32", "forward", "functional", "gate", "hidden_states", "histc", "index_copy_", "index_select", "indices", "max", "min", "moe_num_experts", "moe_topk", "nn", "original_dtype", "output", "permuted_tokens", "projection", "return", "route_tokens_to_experts", "router_logits", "self", "shape", "silu", "size", "sorted_indices", "sum", "to", "tokens_per_expert", "top_k_index", "top_k_weights", "torch", "unpermuted_tokens", "unsqueeze", "view", "zeros"], "aria/modeling_aria.py:AriaTextMoELayer.__init__": ["False", "Linear", "ModelExperts", "ModelSharedExpertsMLP", "ModelTextConfig", "ModelTextMoELayer", "Module", "__init__", "bias", "class", "config", "def", "experts", "hidden_size", "moe_num_experts", "nn", "router", "self", "shared_experts", "super"], "aria/modeling_aria.py:AriaTextMoELayer.forward": ["ModelTextMoELayer", "Module", "Tensor", "class", "def", "expert_output", "experts", "forward", "hidden_states", "nn", "original_shape", "return", "router", "router_logits", "self", "shape", "shared_expert_output", "shared_experts", "size", "torch", "view"], "aria/modeling_aria.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "aria/modeling_aria.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cos", "def", "k", "k_embed", "q", "q_embed", "return", "rotate_half", "sin", "unsqueeze", "unsqueeze_dim"], "aria/modeling_aria.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "aria/modeling_aria.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value", "value_states"], "aria/modeling_aria.py:AriaTextAttention.__init__": ["Linear", "ModelTextAttention", "ModelTextConfig", "Module", "True", "__init__", "attention_bias", "attention_dropout", "bias", "class", "config", "def", "getattr", "head_dim", "hidden_size", "int", "is_causal", "k_proj", "layer_idx", "nn", "num_attention_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "q_proj", "scaling", "self", "super", "v_proj"], "aria/modeling_aria.py:AriaTextAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "LongTensor", "ModelTextAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_attn_implementation", "apply_rotary_pos_emb", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "cache_kwargs", "cache_position", "class", "config", "contiguous", "cos", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "head_dim", "hidden_shape", "hidden_states", "if", "input_shape", "is", "k_proj", "key_states", "kwargs", "layer_idx", "nn", "not", "o_proj", "past_key_values", "position_embeddings", "q_proj", "query_states", "reshape", "return", "scaling", "self", "shape", "sin", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "view"], "aria/modeling_aria.py:AriaTextDecoderLayer.__init__": ["GradientCheckpointingLayer", "ModelTextAttention", "ModelTextConfig", "ModelTextDecoderLayer", "ModelTextMoELayer", "ModelTextRMSNorm", "__init__", "class", "config", "def", "eps", "hidden_size", "input_layernorm", "int", "layer_idx", "mlp", "post_attention_layernorm", "rms_norm_eps", "self", "self_attn", "super"], "aria/modeling_aria.py:AriaTextDecoderLayer.forward": ["Cache", "False", "GradientCheckpointingLayer", "LongTensor", "ModelTextDecoderLayer", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "attention_mask", "bool", "cache_position", "class", "def", "forward", "hidden_states", "input_layernorm", "kwargs", "mlp", "past_key_values", "position_embeddings", "position_ids", "post_attention_layernorm", "residual", "return", "self", "self_attn", "torch", "tuple", "use_cache"], "aria/modeling_aria.py:AriaTextPreTrainedModel._init_weights": ["ModelGroupedExpertsGemm", "ModelTextPreTrainedModel", "PreTrainedModel", "_init_weights", "class", "config", "def", "if", "init", "initializer_range", "isinstance", "mean", "module", "normal_", "self", "std", "super", "weight"], "aria/modeling_aria.py:AriaPreTrainedModel._init_weights": ["ModelPreTrainedModel", "ModelProjector", "PreTrainedModel", "_init_weights", "class", "config", "def", "if", "init", "initializer_range", "isinstance", "module", "query", "self", "std", "super", "trunc_normal_"], "aria/modeling_aria.py:AriaTextRotaryEmbedding.__init__": ["Callable", "False", "ModelTextConfig", "ModelTextRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "aria/modeling_aria.py:AriaTextRotaryEmbedding.compute_default_rope_parameters": ["ModelTextConfig", "ModelTextRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "aria/modeling_aria.py:AriaTextRotaryEmbedding.forward": ["False", "ModelTextRotaryEmbedding", "Module", "None", "and", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "aria/modeling_aria.py:AriaTextModel.__init__": ["Embedding", "False", "ModelTextConfig", "ModelTextDecoderLayer", "ModelTextModel", "ModelTextPreTrainedModel", "ModelTextRMSNorm", "ModelTextRotaryEmbedding", "ModuleList", "__init__", "class", "config", "def", "embed_tokens", "eps", "for", "gradient_checkpointing", "hidden_size", "in", "layer_idx", "layers", "nn", "norm", "num_hidden_layers", "pad_token_id", "padding_idx", "post_init", "range", "rms_norm_eps", "rotary_emb", "self", "super", "vocab_size"], "aria/modeling_aria.py:AriaTextModel.forward": ["BaseModelOutputWithPast", "Cache", "DynamicCache", "FloatTensor", "LongTensor", "ModelTextModel", "ModelTextPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "and", "arange", "attention_mask", "bool", "cache_position", "causal_mask", "class", "config", "create_causal_mask", "decoder_layer", "def", "device", "else", "embed_tokens", "exactly", "for", "forward", "get_seq_length", "hidden_states", "if", "in", "input_embeds", "input_ids", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layers", "must", "norm", "not", "num_hidden_layers", "of", "one", "or", "past_key_values", "past_seen_tokens", "position_embeddings", "position_ids", "raise", "return", "rotary_emb", "self", "shape", "specify", "torch", "unsqueeze", "use_cache"], "aria/modeling_aria.py:AriaTextForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "ModelTextConfig", "ModelTextForCausalLM", "ModelTextModel", "ModelTextPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "vocab_size"], "aria/modeling_aria.py:AriaTextForCausalLM.forward": ["BaseModelOutputWithPast", "Cache", "CausalLMOutputWithPast", "FloatTensor", "GenerationMixin", "LongTensor", "ModelTextForCausalLM", "ModelTextPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "outputs", "past_key_values", "position_ids", "r", "return", "self", "slice", "slice_indices", "torch", "use_cache", "vocab_size"], "aria/modeling_aria.py:AriaModel.__init__": ["AutoModel", "ModelConfig", "ModelModel", "ModelPreTrainedModel", "ModelProjector", "__init__", "class", "config", "def", "from_config", "language_model", "multi_modal_projector", "post_init", "self", "super", "text_config", "vision_config", "vision_tower"], "aria/modeling_aria.py:AriaModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "language_model", "return", "self"], "aria/modeling_aria.py:AriaModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "language_model", "self", "set_input_embeddings", "value"], "aria/modeling_aria.py:AriaModel.get_image_features": ["FloatTensor", "ModelModel", "ModelPreTrainedModel", "None", "True", "_create_patch_attention_mask", "attn_mask", "class", "config", "def", "else", "flatten", "flattened_mask", "get_image_features", "hidden_states", "if", "image_attn_mask", "image_features", "image_outputs", "int", "is", "logical_not", "multi_modal_projector", "not", "output_hidden_states", "patch_attention_mask", "pixel_mask", "pixel_values", "return", "selected_image_feature", "self", "torch", "vision_feature_layer", "vision_tower"], "aria/modeling_aria.py:AriaModel.get_placeholder_mask": ["FloatTensor", "Image", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "ValueError", "all", "and", "class", "config", "def", "device", "do", "dtype", "else", "expand_as", "f", "features", "get_input_embeddings", "get_placeholder_mask", "if", "image", "image_features", "image_token_id", "input_ids", "inputs_embeds", "is", "long", "match", "n_image_features", "n_image_tokens", "not", "numel", "raise", "return", "self", "shape", "special_image_mask", "sum", "tensor", "to", "tokens", "torch", "unsqueeze"], "aria/modeling_aria.py:AriaModel.forward": ["Cache", "FlashAttentionKwargs", "FloatTensor", "LongTensor", "ModelModel", "ModelModelOutputWithPast", "ModelPreTrainedModel", "None", "Tensor", "Unpack", "and", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "device", "dtype", "else", "forward", "get_image_features", "get_input_embeddings", "get_placeholder_mask", "hidden_states", "if", "image_features", "image_hidden_states", "input_ids", "inputs_embeds", "is", "kwargs", "language_model", "last_hidden_state", "masked_scatter", "not", "outputs", "past_key_values", "pixel_mask", "pixel_values", "position_ids", "return", "self", "shape", "special_image_mask", "to", "torch", "tuple", "use_cache", "vision_feature_layer"], "aria/modeling_aria.py:AriaModel._create_patch_attention_mask": ["ModelModel", "ModelPreTrainedModel", "None", "_create_patch_attention_mask", "bool", "class", "config", "def", "dim", "dimension", "if", "is", "patch_size", "patches_subgrid", "pixel_mask", "return", "self", "size", "step", "sum", "unfold", "vision_tower"], "aria/modeling_aria.py:AriaForConditionalGeneration.__init__": ["False", "GenerationMixin", "Linear", "ModelConfig", "ModelForConditionalGeneration", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "text_config", "vocab_size"], "aria/modeling_aria.py:AriaForConditionalGeneration.get_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "model", "return", "self"], "aria/modeling_aria.py:AriaForConditionalGeneration.set_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "model", "self", "set_input_embeddings", "value"], "aria/modeling_aria.py:AriaForConditionalGeneration.get_output_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "Module", "class", "def", "get_output_embeddings", "lm_head", "nn", "return", "self"], "aria/modeling_aria.py:AriaForConditionalGeneration.get_image_features": ["FloatTensor", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "class", "def", "get_image_features", "int", "model", "pixel_mask", "pixel_values", "return", "self", "torch", "vision_feature_layer"], "aria/modeling_aria.py:AriaForConditionalGeneration.forward": ["Cache", "FloatTensor", "GenerationMixin", "LongTensor", "ModelCausalLMOutputWithPast", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "outputs", "past_key_values", "pixel_mask", "pixel_values", "position_ids", "r", "return", "self", "slice", "slice_indices", "text_config", "torch", "tuple", "use_cache", "vocab_size"], "aria/modeling_aria.py:AriaForConditionalGeneration.prepare_inputs_for_generation": ["False", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "True", "attention_mask", "cache_position", "class", "def", "get", "if", "input_ids", "inputs_embeds", "is_first_iteration", "kwargs", "logits_to_keep", "model_inputs", "not", "or", "past_key_values", "pixel_mask", "pixel_values", "prepare_inputs_for_generation", "return", "self", "super", "use_cache"], "audio_spectrogram_transformer/modeling_audio_spectrogram_transformer.py:ASTEmbeddings.__init__": ["ASTConfig", "ASTPatchEmbeddings", "Dropout", "Model", "Module", "None", "Parameter", "__init__", "class", "cls_token", "config", "def", "distillation_token", "dropout", "frequency_out_dimension", "get_shape", "hidden_dropout_prob", "hidden_size", "nn", "num_patches", "patch_embeddings", "position_embeddings", "self", "super", "time_out_dimension", "torch", "zeros"], "audio_spectrogram_transformer/modeling_audio_spectrogram_transformer.py:ASTEmbeddings.get_shape": ["Model", "Module", "class", "config", "def", "frequency_out_dimension", "frequency_stride", "get_shape", "max_length", "nn", "num_mel_bins", "patch_size", "return", "self", "time_out_dimension", "time_stride"], "audio_spectrogram_transformer/modeling_audio_spectrogram_transformer.py:ASTEmbeddings.forward": ["Model", "Module", "Tensor", "batch_size", "cat", "class", "cls_token", "cls_tokens", "def", "dim", "distillation_token", "distillation_tokens", "dropout", "embeddings", "expand", "forward", "input_values", "nn", "patch_embeddings", "position_embeddings", "return", "self", "shape", "torch"], "audio_spectrogram_transformer/modeling_audio_spectrogram_transformer.py:ASTPatchEmbeddings.__init__": ["ASTConfig", "Conv2d", "Model", "Module", "__init__", "class", "config", "def", "frequency_stride", "hidden_size", "kernel_size", "nn", "patch_size", "projection", "self", "stride", "super", "time_stride"], "audio_spectrogram_transformer/modeling_audio_spectrogram_transformer.py:ASTPatchEmbeddings.forward": ["Model", "Module", "Tensor", "class", "def", "embeddings", "flatten", "forward", "input_values", "nn", "projection", "return", "self", "torch", "transpose", "unsqueeze"], "audio_spectrogram_transformer/modeling_audio_spectrogram_transformer.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "contiguous", "def", "dim", "dropout", "float", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "shape", "size", "softmax", "torch", "training", "transpose", "value"], "audio_spectrogram_transformer/modeling_audio_spectrogram_transformer.py:ASTSelfAttention.__init__": ["ASTConfig", "False", "Linear", "Model", "Module", "The", "ValueError", "__init__", "a", "all_head_size", "and", "attention", "attention_head_size", "attention_probs_dropout_prob", "bias", "class", "config", "def", "dropout_prob", "embedding_size", "f", "hasattr", "heads", "hidden", "hidden_size", "if", "int", "is", "is_causal", "key", "multiple", "nn", "not", "num_attention_heads", "number", "of", "qkv_bias", "query", "raise", "scaling", "self", "size", "super", "the", "value"], "audio_spectrogram_transformer/modeling_audio_spectrogram_transformer.py:ASTSelfAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "Model", "Module", "None", "Tensor", "_attn_implementation", "all_head_size", "attention_head_size", "attention_interface", "attention_probs", "batch_size", "class", "config", "context_layer", "def", "dropout", "dropout_prob", "eager", "eager_attention_forward", "else", "forward", "hidden_states", "if", "is_causal", "key", "key_layer", "new_context_layer_shape", "new_shape", "nn", "not", "num_attention_heads", "query", "query_layer", "reshape", "return", "scaling", "self", "shape", "size", "torch", "training", "transpose", "tuple", "value", "value_layer", "view"], "audio_spectrogram_transformer/modeling_audio_spectrogram_transformer.py:ASTSelfOutput.__init__": ["ASTConfig", "Dropout", "Linear", "Model", "Module", "__init__", "class", "config", "def", "dense", "dropout", "hidden_dropout_prob", "hidden_size", "nn", "self", "super"], "audio_spectrogram_transformer/modeling_audio_spectrogram_transformer.py:ASTSelfOutput.forward": ["Model", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "audio_spectrogram_transformer/modeling_audio_spectrogram_transformer.py:ASTAttention.__init__": ["ASTConfig", "ASTSelfAttention", "ASTSelfOutput", "Model", "Module", "__init__", "attention", "class", "config", "def", "nn", "output", "self", "super"], "audio_spectrogram_transformer/modeling_audio_spectrogram_transformer.py:ASTAttention.forward": ["Model", "Module", "Tensor", "_", "attention", "class", "def", "forward", "hidden_states", "nn", "output", "return", "self", "self_attn_output", "torch"], "audio_spectrogram_transformer/modeling_audio_spectrogram_transformer.py:ASTIntermediate.__init__": ["ACT2FN", "ASTConfig", "Linear", "Model", "Module", "__init__", "class", "config", "def", "dense", "else", "hidden_act", "hidden_size", "if", "intermediate_act_fn", "intermediate_size", "isinstance", "nn", "self", "str", "super"], "audio_spectrogram_transformer/modeling_audio_spectrogram_transformer.py:ASTIntermediate.forward": ["Model", "Module", "Tensor", "class", "def", "dense", "forward", "hidden_states", "intermediate_act_fn", "nn", "return", "self", "torch"], "audio_spectrogram_transformer/modeling_audio_spectrogram_transformer.py:ASTOutput.__init__": ["ASTConfig", "Dropout", "Linear", "Model", "Module", "__init__", "class", "config", "def", "dense", "dropout", "hidden_dropout_prob", "hidden_size", "intermediate_size", "nn", "self", "super"], "audio_spectrogram_transformer/modeling_audio_spectrogram_transformer.py:ASTOutput.forward": ["Model", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "audio_spectrogram_transformer/modeling_audio_spectrogram_transformer.py:ASTLayer.__init__": ["ASTAttention", "ASTConfig", "ASTIntermediate", "ASTOutput", "GradientCheckpointingLayer", "LayerNorm", "Model", "__init__", "attention", "chunk_size_feed_forward", "class", "config", "def", "eps", "hidden_size", "intermediate", "layer_norm_eps", "layernorm_after", "layernorm_before", "nn", "output", "self", "seq_len_dim", "super"], "audio_spectrogram_transformer/modeling_audio_spectrogram_transformer.py:ASTLayer.forward": ["GradientCheckpointingLayer", "Model", "Tensor", "attention", "attention_output", "class", "def", "forward", "hidden_states", "hidden_states_norm", "intermediate", "layer_output", "layernorm_after", "layernorm_before", "output", "return", "self", "torch"], "audio_spectrogram_transformer/modeling_audio_spectrogram_transformer.py:ASTEncoder.__init__": ["ASTConfig", "ASTLayer", "False", "Model", "Module", "ModuleList", "_", "__init__", "class", "config", "def", "for", "gradient_checkpointing", "in", "layer", "nn", "num_hidden_layers", "range", "self", "super"], "audio_spectrogram_transformer/modeling_audio_spectrogram_transformer.py:ASTEncoder.forward": ["BaseModelOutput", "Model", "Module", "Tensor", "class", "def", "enumerate", "for", "forward", "hidden_states", "i", "in", "last_hidden_state", "layer", "layer_module", "nn", "return", "self", "torch"], "audio_spectrogram_transformer/modeling_audio_spectrogram_transformer.py:ASTPreTrainedModel._init_weights": ["ASTEmbeddings", "Conv2d", "LayerNorm", "Linear", "Model", "None", "PreTrainedModel", "_init_weights", "bias", "class", "cls_token", "config", "def", "distillation_token", "elif", "if", "init", "initializer_range", "is", "isinstance", "mean", "module", "nn", "not", "ones_", "position_embeddings", "self", "std", "trunc_normal_", "weight", "zeros_"], "audio_spectrogram_transformer/modeling_audio_spectrogram_transformer.py:ASTModel.__init__": ["ASTConfig", "ASTEmbeddings", "ASTEncoder", "ASTPreTrainedModel", "LayerNorm", "Model", "None", "__init__", "class", "config", "def", "embeddings", "encoder", "eps", "hidden_size", "layer_norm_eps", "layernorm", "nn", "post_init", "self", "super"], "audio_spectrogram_transformer/modeling_audio_spectrogram_transformer.py:ASTModel.get_input_embeddings": ["ASTPatchEmbeddings", "ASTPreTrainedModel", "Model", "class", "def", "embeddings", "get_input_embeddings", "patch_embeddings", "return", "self"], "audio_spectrogram_transformer/modeling_audio_spectrogram_transformer.py:ASTModel.forward": ["ASTPreTrainedModel", "BaseModelOutput", "BaseModelOutputWithPooling", "Model", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "class", "def", "embedding_output", "embeddings", "encoder", "encoder_outputs", "forward", "have", "if", "input_values", "is", "kwargs", "last_hidden_state", "layernorm", "pooled_output", "pooler_output", "r", "raise", "return", "self", "sequence_output", "specify", "to", "torch"], "audio_spectrogram_transformer/modeling_audio_spectrogram_transformer.py:ASTMLPHead.__init__": ["ASTConfig", "Identity", "LayerNorm", "Linear", "Model", "Module", "__init__", "class", "config", "def", "dense", "else", "eps", "hidden_size", "if", "layer_norm_eps", "layernorm", "nn", "num_labels", "self", "super"], "audio_spectrogram_transformer/modeling_audio_spectrogram_transformer.py:ASTMLPHead.forward": ["Model", "Module", "class", "def", "dense", "forward", "hidden_state", "layernorm", "nn", "return", "self"], "audio_spectrogram_transformer/modeling_audio_spectrogram_transformer.py:ASTForAudioClassification.__init__": ["ASTConfig", "ASTMLPHead", "ASTModel", "ASTPreTrainedModel", "Model", "None", "__init__", "class", "classifier", "config", "def", "num_labels", "post_init", "self", "super"], "audio_spectrogram_transformer/modeling_audio_spectrogram_transformer.py:ASTForAudioClassification.forward": ["ASTPreTrainedModel", "BaseModelOutputWithPooling", "Model", "None", "SequenceClassifierOutput", "Tensor", "TransformersKwargs", "Unpack", "attentions", "class", "classifier", "config", "def", "forward", "hidden_states", "if", "input_values", "is", "kwargs", "labels", "logits", "loss", "loss_function", "not", "outputs", "pooled_output", "pooler_output", "r", "return", "self", "torch"], "audioflamingo3/modeling_audioflamingo3.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "and", "attention_mask", "attn_output", "attn_weights", "contiguous", "def", "dim", "dropout", "float", "functional", "if", "is", "key", "kwargs", "matmul", "module", "ndim", "nn", "not", "p", "query", "return", "scaling", "shape", "size", "softmax", "torch", "training", "transpose", "value"], "audioflamingo3/modeling_audioflamingo3.py:AudioFlamingo3Attention.__init__": ["False", "Instantiating", "Linear", "ModelAttention", "ModelConfig", "Module", "None", "Please", "True", "ValueError", "__class__", "__init__", "__name__", "a", "and", "be", "bias", "bool", "by", "caching", "call", "class", "config", "creating", "decoder", "def", "divisible", "dropout", "during", "embed_dim", "errors", "f", "float", "forward", "got", "head_dim", "if", "int", "is", "is_causal", "is_decoder", "k_proj", "layer_idx", "logger", "make", "must", "nn", "not", "num_heads", "out_proj", "passing", "provide", "q_proj", "raise", "recommended", "scaling", "self", "super", "sure", "the", "this", "to", "used", "v_proj", "warning_once", "when", "will", "without"], "audioflamingo3/modeling_audioflamingo3.py:AudioFlamingo3Attention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "EncoderDecoderCache", "False", "FlashAttentionKwargs", "ModelAttention", "Module", "None", "Tensor", "True", "Unpack", "_attn_implementation", "and", "attention_interface", "attention_mask", "attn_output", "attn_weights", "bool", "bsz", "cache_position", "class", "config", "contiguous", "cross_attention_cache", "current_states", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "get", "head_dim", "hidden_states", "if", "is", "is_cross_attention", "is_updated", "isinstance", "k_proj", "key_states", "key_value_states", "keys", "kwargs", "layer_idx", "layers", "nn", "not", "num_heads", "out_proj", "output_attentions", "past_key_values", "q_input_shape", "q_proj", "query_states", "reshape", "return", "scaling", "self", "self_attention_cache", "shape", "tgt_len", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "values", "view"], "audioflamingo3/modeling_audioflamingo3.py:AudioFlamingo3EncoderLayer.__init__": ["ACT2FN", "GradientCheckpointingLayer", "LayerNorm", "Linear", "ModelAttention", "ModelConfig", "ModelEncoderLayer", "__init__", "activation_dropout", "activation_fn", "activation_function", "attention_dropout", "class", "config", "d_model", "def", "dropout", "embed_dim", "encoder_attention_heads", "encoder_ffn_dim", "fc1", "fc2", "final_layer_norm", "nn", "num_heads", "self", "self_attn", "self_attn_layer_norm", "super"], "audioflamingo3/modeling_audioflamingo3.py:AudioFlamingo3EncoderLayer.forward": ["False", "GradientCheckpointingLayer", "ModelEncoderLayer", "Tensor", "activation_dropout", "activation_fn", "attention_mask", "attn_weights", "bool", "clamp", "clamp_value", "class", "def", "dropout", "dtype", "fc1", "fc2", "final_layer_norm", "finfo", "float16", "forward", "functional", "hidden_states", "if", "max", "min", "nn", "output_attentions", "p", "residual", "return", "self", "self_attn", "self_attn_layer_norm", "torch", "training"], "audioflamingo3/modeling_audioflamingo3.py:AudioFlamingo3Encoder.__init__": ["AvgPool1d", "Conv1d", "Embedding", "False", "LayerNorm", "ModelEncoder", "ModelEncoderConfig", "ModelEncoderLayer", "ModelPreTrainedModel", "ModuleList", "_", "__init__", "avg_pooler", "class", "config", "conv1", "conv2", "d_model", "def", "dropout", "else", "embed_dim", "embed_positions", "embed_scale", "encoder_layerdrop", "encoder_layers", "for", "gradient_checkpointing", "if", "in", "kernel_size", "layer_norm", "layerdrop", "layers", "math", "max_source_positions", "nn", "num_mel_bins", "pad_token_id", "padding", "padding_idx", "post_init", "range", "requires_grad_", "scale_embedding", "self", "sqrt", "stride", "super"], "audioflamingo3/modeling_audioflamingo3.py:AudioFlamingo3Encoder._freeze_parameters": ["False", "ModelEncoder", "ModelPreTrainedModel", "_freeze_parameters", "_requires_grad", "class", "def", "for", "in", "param", "parameters", "requires_grad", "self"], "audioflamingo3/modeling_audioflamingo3.py:AudioFlamingo3Encoder.get_input_embeddings": ["ModelEncoder", "ModelPreTrainedModel", "Module", "class", "conv1", "def", "get_input_embeddings", "nn", "return", "self"], "audioflamingo3/modeling_audioflamingo3.py:AudioFlamingo3Encoder.set_input_embeddings": ["ModelEncoder", "ModelPreTrainedModel", "Module", "class", "conv1", "def", "nn", "self", "set_input_embeddings", "value"], "audioflamingo3/modeling_audioflamingo3.py:AudioFlamingo3Encoder.forward": ["BaseModelOutput", "ModelEncoder", "ModelPreTrainedModel", "None", "Tensor", "and", "arange", "attention_mask", "avg_pooler", "class", "config", "conv1", "conv2", "create_bidirectional_mask", "def", "device", "drop", "dropout", "embed_positions", "for", "forward", "functional", "gelu", "hidden_states", "if", "in", "input_embeds", "input_features", "input_features_lengths", "input_features_mask", "inputs_embeds", "kwargs", "last_hidden_state", "layer", "layer_norm", "layerdrop", "layers", "nn", "not", "p", "permute", "r", "rand", "return", "self", "seq_len", "shape", "sum", "torch", "training", "weight"], "audioflamingo3/modeling_audioflamingo3.py:AudioFlamingo3Encoder._get_feat_extract_output_lengths": ["LongTensor", "ModelEncoder", "ModelPreTrainedModel", "_get_feat_extract_output_lengths", "class", "def", "input_lengths", "output_lengths", "return", "self", "torch"], "audioflamingo3/modeling_audioflamingo3.py:AudioFlamingo3MultiModalProjector.__init__": ["ACT2FN", "Linear", "ModelConfig", "ModelMultiModalProjector", "Model_config", "Module", "__init__", "act", "bias", "class", "config", "def", "hidden_size", "linear_1", "linear_2", "nn", "projector_bias", "projector_hidden_act", "self", "super", "text_config"], "audioflamingo3/modeling_audioflamingo3.py:AudioFlamingo3MultiModalProjector.forward": ["ModelMultiModalProjector", "Model_features", "Module", "act", "class", "def", "forward", "hidden_states", "linear_1", "linear_2", "nn", "return", "self"], "audioflamingo3/modeling_audioflamingo3.py:AudioFlamingo3ForConditionalGeneration.__init__": ["AutoModel", "AutoModelForCausalLM", "GenerationMixin", "ModelForConditionalGeneration", "ModelMultiModalProjector", "ModelPreTrainedModel", "Model_config", "Model_tower", "__init__", "class", "config", "def", "from_config", "language_model", "multi_modal_projector", "post_init", "self", "super", "text_config", "vocab_size"], "audioflamingo3/modeling_audioflamingo3.py:AudioFlamingo3ForConditionalGeneration.get_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "language_model", "return", "self"], "audioflamingo3/modeling_audioflamingo3.py:AudioFlamingo3ForConditionalGeneration.set_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "language_model", "self", "set_input_embeddings", "value"], "audioflamingo3/modeling_audioflamingo3.py:AudioFlamingo3ForConditionalGeneration.get_output_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "get_output_embeddings", "language_model", "return", "self"], "audioflamingo3/modeling_audioflamingo3.py:AudioFlamingo3ForConditionalGeneration.set_output_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "language_model", "new_embeddings", "self", "set_output_embeddings"], "audioflamingo3/modeling_audioflamingo3.py:AudioFlamingo3ForConditionalGeneration.set_decoder": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "decoder", "def", "language_model", "self", "set_decoder"], "audioflamingo3/modeling_audioflamingo3.py:AudioFlamingo3ForConditionalGeneration.get_decoder": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "get_decoder", "language_model", "return", "self"], "audioflamingo3/modeling_audioflamingo3.py:AudioFlamingo3ForConditionalGeneration.get_audio_features": ["FloatTensor", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "Model_embeds", "Model_tower", "None", "Tensor", "arange", "class", "def", "device", "encoder_output", "get_Model_features", "input_features", "input_features_mask", "last_hidden_state", "multi_modal_projector", "post_lengths", "return", "self", "shape", "sum", "to", "torch", "valid_mask"], "audioflamingo3/modeling_audioflamingo3.py:AudioFlamingo3ForConditionalGeneration.forward": ["Cache", "CausalLMOutputWithPast", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForConditionalGeneration", "ModelPreTrainedModel", "Model_embeds", "Model_token_id", "Model_token_mask", "None", "Tensor", "TransformersKwargs", "Unpack", "and", "attention_mask", "bool", "cache_position", "class", "config", "def", "device", "forward", "get_Model_features", "get_input_embeddings", "if", "input_features", "input_features_mask", "input_ids", "inputs_embeds", "int", "is", "kwargs", "labels", "language_model", "logits_to_keep", "masked_scatter", "not", "outputs", "past_key_values", "position_ids", "r", "return", "self", "to", "torch", "unsqueeze", "use_cache"], "audioflamingo3/modeling_audioflamingo3.py:AudioFlamingo3ForConditionalGeneration.prepare_inputs_for_generation": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "and", "args", "cache_position", "class", "def", "get", "if", "input_features", "input_features_mask", "is", "kwargs", "model_inputs", "not", "pop", "prepare_inputs_for_generation", "return", "self", "super"], "auto/modeling_auto.py:AutoModelForCausalLM.from_pretrained": ["ModelModelForCausalLM", "PathLike", "_BaseModelModelClass", "_BaseModelWithGenerate", "class", "cls", "def", "from_pretrained", "kwargs", "model_args", "os", "pretrained_model_name_or_path", "return", "str", "super", "type"], "auto/modeling_auto.py:AutoModelForImageTextToText.from_pretrained": ["ModelModelForImageTextToText", "PathLike", "_BaseModelModelClass", "_BaseModelWithGenerate", "class", "cls", "def", "from_pretrained", "kwargs", "model_args", "os", "pretrained_model_name_or_path", "return", "str", "super", "type"], "autoformer/modeling_autoformer.py:AutoformerFeatureEmbedder.__init__": ["Embedding", "ModelFeatureEmbedder", "Module", "ModuleList", "None", "__init__", "c", "cardinalities", "class", "d", "def", "embedders", "embedding_dims", "for", "in", "int", "len", "list", "nn", "num_features", "self", "super", "zip"], "autoformer/modeling_autoformer.py:AutoformerFeatureEmbedder.forward": ["ModelFeatureEmbedder", "Module", "Tensor", "cat", "cat_feature_slice", "cat_feature_slices", "chunk", "class", "def", "dim", "else", "embed", "embedders", "features", "for", "forward", "if", "in", "nn", "num_features", "return", "self", "squeeze", "torch", "zip"], "autoformer/modeling_autoformer.py:AutoformerStdScaler.__init__": ["ModelConfig", "ModelStdScaler", "Module", "True", "__init__", "class", "config", "def", "dim", "else", "hasattr", "if", "keepdim", "minimum_scale", "nn", "scaling_dim", "self", "super"], "autoformer/modeling_autoformer.py:AutoformerStdScaler.forward": ["ModelStdScaler", "Module", "Tensor", "clamp_min", "class", "data", "def", "denominator", "dim", "forward", "keepdim", "loc", "minimum_scale", "nn", "observed_indicator", "return", "scale", "self", "sqrt", "sum", "torch", "tuple", "variance"], "autoformer/modeling_autoformer.py:AutoformerMeanScaler.__init__": ["ModelConfig", "ModelMeanScaler", "Module", "None", "True", "__init__", "class", "config", "def", "default_scale", "dim", "else", "hasattr", "if", "keepdim", "minimum_scale", "nn", "scaling_dim", "self", "super"], "autoformer/modeling_autoformer.py:AutoformerMeanScaler.forward": ["ModelMeanScaler", "Module", "None", "Tensor", "True", "abs", "batch_observations", "batch_sum", "clamp", "class", "data", "def", "default_scale", "dim", "else", "forward", "if", "is", "keepdim", "min", "minimum_scale", "nn", "not", "num_observed", "observed_indicator", "ones_like", "return", "scale", "scaled_data", "self", "squeeze", "sum", "torch", "ts_sum", "tuple", "where", "zeros_like"], "autoformer/modeling_autoformer.py:AutoformerNOPScaler.__init__": ["ModelConfig", "ModelNOPScaler", "Module", "True", "__init__", "class", "config", "def", "dim", "else", "hasattr", "if", "keepdim", "nn", "scaling_dim", "self", "super"], "autoformer/modeling_autoformer.py:AutoformerNOPScaler.forward": ["False", "ModelNOPScaler", "Module", "None", "Tensor", "class", "data", "def", "dim", "forward", "keepdim", "loc", "mean", "nn", "observed_indicator", "ones_like", "requires_grad", "return", "scale", "self", "torch", "tuple", "zeros_like"], "autoformer/modeling_autoformer.py:weighted_average": ["Model_average", "Model_tensor", "None", "Tensor", "clamp", "def", "dim", "else", "if", "input_tensor", "is", "mean", "min", "not", "return", "sum", "sum_weights", "torch", "weights", "where", "zeros_like"], "autoformer/modeling_autoformer.py:nll": ["Distribution", "Model", "Tensor", "def", "distributions", "input", "log_prob", "return", "target", "torch"], "autoformer/modeling_autoformer.py:AutoformerSinusoidalPositionalEmbedding.__init__": ["Embedding", "ModelSinusoidalPositionalEmbedding", "None", "True", "__init__", "_freeze", "class", "def", "embedding_dim", "int", "nn", "num_positions", "padding_idx", "self", "super"], "autoformer/modeling_autoformer.py:AutoformerSinusoidalPositionalEmbedding.create_weight": ["Embedding", "False", "FloatTensor", "ModelSinusoidalPositionalEmbedding", "array", "class", "cos", "create_weight", "def", "dim", "dtype", "else", "empty", "for", "if", "in", "j", "n_pos", "nn", "np", "out", "pos", "position_enc", "power", "range", "requires_grad", "return", "self", "sentinel", "shape", "sin", "torch", "weight"], "autoformer/modeling_autoformer.py:AutoformerSinusoidalPositionalEmbedding.forward": ["Embedding", "ModelSinusoidalPositionalEmbedding", "None", "Size", "Tensor", "arange", "bsz", "class", "def", "device", "dtype", "forward", "if", "input_ids_shape", "int", "is", "long", "nn", "past_key_values_length", "position_ids", "return", "self", "seq_len", "super", "torch", "weight"], "autoformer/modeling_autoformer.py:AutoformerValueEmbedding.__init__": ["False", "Linear", "ModelValueEmbedding", "Module", "__init__", "bias", "class", "d_model", "def", "feature_size", "in_features", "nn", "out_features", "self", "super", "value_projection"], "autoformer/modeling_autoformer.py:AutoformerValueEmbedding.forward": ["ModelValueEmbedding", "Module", "class", "def", "forward", "nn", "return", "self", "value_projection", "x"], "autoformer/modeling_autoformer.py:AutoformerSeriesDecompositionLayer.__init__": ["AvgPool1d", "ModelConfig", "ModelSeriesDecompositionLayer", "Module", "__init__", "avg", "class", "config", "def", "kernel_size", "moving_average", "nn", "padding", "self", "stride", "super"], "autoformer/modeling_autoformer.py:AutoformerSeriesDecompositionLayer.forward": ["ModelSeriesDecompositionLayer", "Module", "avg", "cat", "class", "def", "dim", "end", "forward", "front", "kernel_size", "nn", "num_of_pads", "permute", "repeat", "return", "self", "torch", "x", "x_padded", "x_seasonal", "x_trend"], "autoformer/modeling_autoformer.py:AutoformerLayernorm.__init__": ["LayerNorm", "ModelConfig", "ModelLayernorm", "Module", "__init__", "class", "config", "d_model", "def", "layernorm", "nn", "self", "super"], "autoformer/modeling_autoformer.py:AutoformerLayernorm.forward": ["ModelLayernorm", "Module", "bias", "class", "def", "dim", "forward", "layernorm", "mean", "nn", "repeat", "return", "self", "shape", "torch", "unsqueeze", "x", "x_hat"], "autoformer/modeling_autoformer.py:AutoformerAttention.__init__": ["False", "Linear", "ModelAttention", "Module", "None", "True", "ValueError", "__init__", "and", "autocorrelation_factor", "be", "bias", "bool", "by", "class", "def", "divisible", "dropout", "embed_dim", "f", "float", "got", "head_dim", "if", "int", "is_decoder", "k_proj", "layer_idx", "must", "nn", "num_heads", "out_proj", "q_proj", "raise", "scaling", "self", "super", "v_proj"], "autoformer/modeling_autoformer.py:AutoformerAttention.forward": ["Attention", "Cache", "EncoderDecoderCache", "False", "ModelAttention", "Module", "None", "Tensor", "True", "ValueError", "_", "and", "arange", "attention_mask", "attn_output", "attn_weights", "attn_weights_reshaped", "autocorrelation_factor", "autocorrelations", "autocorrelations_mean_on_bsz", "autocorrelations_mean_on_head_channel", "be", "bool", "bsz", "but", "cache_position", "cat", "channel", "class", "conj", "contiguous", "cross_attention_cache", "curr_past_key_values", "current_states", "def", "delays_agg", "device", "dim", "dims", "else", "embed_dim", "f", "fft", "float", "for", "forward", "gather", "get", "head_dim", "hidden_states", "i", "if", "in", "index", "init_index", "int", "irfft", "is", "is_cross_attention", "is_updated", "isinstance", "k_proj", "key_states", "key_states_fft", "key_value_states", "keys", "layer_idx", "layers", "log", "mask", "math", "mean", "n", "nn", "not", "num_heads", "of", "out_proj", "output_attentions", "past_key_values", "proj_shape", "q_proj", "queries_time_length", "query_states", "query_states_fft", "raise", "range", "repeat", "reshape", "return", "rfft", "roll", "self", "self_attention_cache", "shifts", "should", "size", "softmax", "src_len", "stack", "tgt_len", "time_length", "tmp_delay", "tmp_values", "to", "top_k", "top_k_autocorrelations", "top_k_autocorrelations_at_delay", "top_k_delays_index", "topk", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "value_states_roll_delay", "values", "values_time_length", "view", "weights", "zeros", "zeros_like"], "autoformer/modeling_autoformer.py:AutoformerEncoderLayer.__init__": ["ACT2FN", "GradientCheckpointingLayer", "LayerNorm", "Linear", "ModelAttention", "ModelConfig", "ModelEncoderLayer", "ModelLayernorm", "ModelSeriesDecompositionLayer", "__init__", "activation_dropout", "activation_fn", "activation_function", "attention_dropout", "autocorrelation_factor", "class", "config", "d_model", "decomp1", "decomp2", "def", "dropout", "embed_dim", "encoder_attention_heads", "encoder_ffn_dim", "fc1", "fc2", "final_layer_norm", "nn", "num_heads", "self", "self_attn", "self_attn_layer_norm", "super"], "autoformer/modeling_autoformer.py:AutoformerEncoderLayer.forward": ["False", "FloatTensor", "GradientCheckpointingLayer", "ModelEncoderLayer", "None", "_", "activation_dropout", "activation_fn", "and", "any", "attention_mask", "attn_weights", "bool", "clamp", "clamp_value", "class", "decomp1", "decomp2", "def", "dropout", "dtype", "fc1", "fc2", "final_layer_norm", "finfo", "float16", "forward", "functional", "hidden_states", "if", "isinf", "isnan", "max", "min", "nn", "or", "output_attentions", "outputs", "p", "residual", "return", "self", "self_attn", "self_attn_layer_norm", "torch", "training", "tuple"], "autoformer/modeling_autoformer.py:AutoformerDecoderLayer.__init__": ["ACT2FN", "Conv1d", "False", "GradientCheckpointingLayer", "LayerNorm", "Linear", "ModelAttention", "ModelConfig", "ModelDecoderLayer", "ModelLayernorm", "ModelSeriesDecompositionLayer", "None", "True", "__init__", "activation_dropout", "activation_fn", "activation_function", "attention_dropout", "autocorrelation_factor", "bias", "circular", "class", "config", "d_model", "decoder_attention_heads", "decoder_ffn_dim", "decomp1", "decomp2", "decomp3", "def", "dropout", "embed_dim", "encoder_attn", "encoder_attn_layer_norm", "fc1", "fc2", "feature_size", "final_layer_norm", "in_channels", "is_decoder", "kernel_size", "layer_idx", "nn", "num_heads", "out_channels", "padding", "padding_mode", "self", "self_attn", "self_attn_layer_norm", "stride", "super", "trend_projection"], "autoformer/modeling_autoformer.py:AutoformerDecoderLayer.forward": ["Cache", "False", "FloatTensor", "GradientCheckpointingLayer", "ModelDecoderLayer", "None", "Tensor", "True", "activation_dropout", "activation_fn", "attention_mask", "bool", "cache_position", "class", "cross_attn_weights", "decomp1", "decomp2", "decomp3", "def", "dropout", "else", "encoder_attention_mask", "encoder_attn", "encoder_attn_layer_norm", "encoder_hidden_states", "fc1", "fc2", "final_layer_norm", "forward", "functional", "hidden_states", "if", "is", "key_value_states", "nn", "not", "output_attentions", "outputs", "p", "past_key_values", "permute", "residual", "residual_trend", "return", "self", "self_attn", "self_attn_layer_norm", "self_attn_weights", "torch", "training", "transpose", "trend1", "trend2", "trend3", "trend_projection", "tuple", "use_cache"], "autoformer/modeling_autoformer.py:AutoformerPreTrainedModel._init_weights": ["Conv1d", "Embedding", "False", "LayerNorm", "Linear", "ModelPreTrainedModel", "ModelSinusoidalPositionalEmbedding", "Module", "None", "PreTrainedModel", "_init_weights", "_is_hf_initialized", "and", "bias", "class", "config", "copy_", "create_weight", "def", "elif", "getattr", "if", "init", "init_std", "is", "isinstance", "mean", "module", "nn", "normal_", "not", "ones_", "padding_idx", "self", "std", "weight", "zeros_"], "autoformer/modeling_autoformer.py:AutoformerPreTrainedModel._update_full_mask": ["False", "ModelPreTrainedModel", "None", "PreTrainedModel", "Tensor", "_attn_implementation", "_prepare_4d_attention_mask", "_prepare_4d_attention_mask_for_sdpa", "_update_full_mask", "attention_mask", "class", "config", "def", "dtype", "elif", "else", "flash", "flex_attention", "if", "in", "inputs_embeds", "is", "is_causal", "isinstance", "make_flex_block_causal_mask", "not", "return", "sdpa", "self", "torch"], "autoformer/modeling_autoformer.py:AutoformerEncoder.__init__": ["False", "LayerNorm", "ModelConfig", "ModelEncoder", "ModelEncoderLayer", "ModelPreTrainedModel", "ModelSinusoidalPositionalEmbedding", "ModelValueEmbedding", "ModuleList", "None", "The", "ValueError", "_", "__init__", "be", "class", "config", "context_length", "d_model", "def", "dropout", "embed_positions", "encoder_layerdrop", "encoder_layers", "feature_size", "for", "gradient_checkpointing", "if", "in", "is", "layerdrop", "layernorm_embedding", "layers", "needs", "nn", "post_init", "prediction_length", "raise", "range", "self", "specified", "super", "to", "value_embedding"], "autoformer/modeling_autoformer.py:AutoformerEncoder.forward": ["BaseModelOutput", "False", "FloatTensor", "ModelEncoder", "ModelPreTrainedModel", "None", "Tensor", "True", "_update_full_mask", "all_attentions", "attention_mask", "attentions", "bool", "class", "config", "def", "dropout", "dropout_probability", "else", "embed_pos", "embed_positions", "encoder_layer", "encoder_states", "enumerate", "for", "forward", "functional", "hidden_states", "idx", "if", "in", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layer_outputs", "layerdrop", "layernorm_embedding", "layers", "nn", "not", "output_attentions", "output_hidden_states", "p", "r", "rand", "return", "return_dict", "self", "size", "to_drop", "torch", "training", "tuple", "use_return_dict", "v", "value_embedding"], "autoformer/modeling_autoformer.py:AutoformerDecoder.__init__": ["False", "LayerNorm", "Linear", "ModelConfig", "ModelDecoder", "ModelDecoderLayer", "ModelPreTrainedModel", "ModelSinusoidalPositionalEmbedding", "ModelValueEmbedding", "ModuleList", "None", "The", "ValueError", "__init__", "be", "class", "config", "context_length", "d_model", "decoder_layerdrop", "decoder_layers", "def", "dropout", "embed_positions", "feature_size", "for", "gradient_checkpointing", "i", "if", "in", "is", "layer_idx", "layerdrop", "layernorm_embedding", "layers", "needs", "nn", "post_init", "prediction_length", "raise", "range", "seasonality_projection", "self", "specified", "super", "to", "value_embedding"], "autoformer/modeling_autoformer.py:AutoformerDecoder.forward": ["Cache", "DynamicCache", "EncoderDecoderCache", "False", "FloatTensor", "LongTensor", "ModelDecoder", "ModelDecoderOutput", "ModelPreTrainedModel", "None", "Setting", "Tensor", "True", "_prepare_4d_attention_mask", "all_cross_attentions", "all_hidden_states", "all_self_attns", "and", "attention_mask", "attentions", "bool", "cache_position", "checkpointing", "class", "config", "context_length", "continue", "cross_attentions", "decoder_layer", "def", "dropout", "dropout_probability", "dtype", "else", "embed_pos", "embed_positions", "encoder_attention_mask", "encoder_hidden_states", "enumerate", "for", "forward", "functional", "gradient", "gradient_checkpointing", "hidden_states", "idx", "if", "in", "incompatible", "input_shape", "inputs_embeds", "is", "kwargs", "label_length", "last_hidden_state", "layer_outputs", "layerdrop", "layernorm_embedding", "layers", "logger", "nn", "not", "output_attentions", "output_hidden_states", "p", "past_key_values", "past_key_values_length", "r", "rand", "residual_trend", "return", "return_dict", "seasonality_projection", "self", "size", "tgt_len", "torch", "training", "trend", "tuple", "use_cache", "use_return_dict", "v", "value_embedding", "warning", "with"], "autoformer/modeling_autoformer.py:AutoformerModel.__init__": ["ModelConfig", "ModelDecoder", "ModelEncoder", "ModelFeatureEmbedder", "ModelMeanScaler", "ModelModel", "ModelNOPScaler", "ModelPreTrainedModel", "ModelSeriesDecompositionLayer", "ModelStdScaler", "True", "__init__", "cardinalities", "cardinality", "class", "config", "decoder", "decomposition_layer", "def", "elif", "else", "embedder", "embedding_dimension", "embedding_dims", "encoder", "if", "is", "mean", "num_static_categorical_features", "or", "post_init", "scaler", "scaling", "self", "std", "super"], "autoformer/modeling_autoformer.py:AutoformerModel._past_length": ["ModelModel", "ModelPreTrainedModel", "_past_length", "class", "config", "context_length", "def", "int", "lags_sequence", "max", "return", "self"], "autoformer/modeling_autoformer.py:AutoformerModel.get_lagged_subsequences": ["ModelModel", "ModelPreTrainedModel", "None", "Tensor", "ValueError", "append", "begin_index", "cannot", "class", "config", "def", "dim", "else", "end_index", "f", "for", "found", "further", "get_lagged_subsequences", "go", "history", "if", "in", "indices", "int", "is", "lag", "lag_index", "lagged_values", "lags", "lags_sequence", "length", "max", "only", "raise", "return", "self", "sequence", "sequence_length", "shape", "shift", "stack", "subsequences_length", "than", "torch", "while"], "autoformer/modeling_autoformer.py:AutoformerModel.create_network_inputs": ["ModelModel", "ModelPreTrainedModel", "None", "Tensor", "ValueError", "_", "_past_length", "abs", "and", "cat", "class", "config", "context", "context_length", "create_network_inputs", "def", "dim", "does", "else", "embedded_cat", "embedder", "expand", "expanded_static_feat", "f", "feature", "features", "future_time_features", "future_values", "get_lagged_subsequences", "if", "input", "input_size", "inputs", "is", "lagged_sequence", "lags_shape", "length", "lengths", "loc", "log", "log1p", "log_abs_loc", "log_scale", "match", "not", "observed_context", "ones_like", "past_observed_mask", "past_time_features", "past_values", "prediction_length", "raise", "reshape", "reshaped_lagged_sequence", "return", "scale", "scaler", "self", "sequence", "shape", "squeeze", "static_categorical_features", "static_feat", "static_real_features", "subsequences_length", "time", "time_feat", "torch", "tuple", "unsqueeze"], "autoformer/modeling_autoformer.py:AutoformerModel.forward": ["BaseModelOutput", "Cache", "FloatTensor", "LongTensor", "ModelDecoderOutput", "ModelModel", "ModelModelOutput", "ModelPreTrainedModel", "None", "Tensor", "and", "attention_mask", "attentions", "bool", "cache_position", "cat", "class", "config", "context_length", "create_network_inputs", "cross_attentions", "decoder", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_input", "decoder_outputs", "decomposition_layer", "def", "device", "dim", "elif", "else", "enc_input", "encoder", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "forward", "future_time_features", "future_values", "hidden_states", "if", "inputs_embeds", "is", "isinstance", "kwargs", "label_length", "last_hidden_state", "len", "list", "loc", "mean", "not", "output_attentions", "output_hidden_states", "past_key_values", "past_observed_mask", "past_time_features", "past_values", "prediction_length", "r", "repeat", "return", "return_dict", "scale", "seasonal_input", "self", "shape", "static_categorical_features", "static_feat", "static_features", "static_real_features", "temporal_features", "torch", "transformer_inputs", "trend", "trend_init", "trend_input", "tuple", "unsqueeze", "use_cache", "use_return_dict", "zeros"], "autoformer/modeling_autoformer.py:AutoformerForPrediction.__init__": ["ModelConfig", "ModelForPrediction", "ModelModel", "ModelPreTrainedModel", "NegativeBinomialOutput", "NormalOutput", "StudentTOutput", "Unknown", "ValueError", "__init__", "class", "config", "def", "dim", "distribution", "distribution_output", "elif", "else", "event_shape", "f", "feature_size", "function", "get_parameter_projection", "if", "input_size", "loss", "model", "negative_binomial", "nll", "normal", "output", "parameter_projection", "post_init", "raise", "self", "student_t", "super", "target_shape"], "autoformer/modeling_autoformer.py:AutoformerForPrediction.output_params": ["ModelForPrediction", "ModelPreTrainedModel", "class", "config", "decoder_output", "def", "output_params", "parameter_projection", "prediction_length", "return", "self"], "autoformer/modeling_autoformer.py:AutoformerForPrediction.output_distribution": ["Distribution", "ModelForPrediction", "ModelPreTrainedModel", "None", "class", "def", "distribution", "distribution_output", "distributions", "for", "if", "in", "is", "loc", "not", "output_distribution", "p", "params", "return", "scale", "self", "sliced_params", "torch", "trailing_n"], "autoformer/modeling_autoformer.py:AutoformerForPrediction.forward": ["Cache", "False", "FloatTensor", "LongTensor", "ModelForPrediction", "ModelPreTrainedModel", "None", "Seq2SeqTSPredictionOutput", "Tensor", "_", "bool", "class", "config", "cross_attentions", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "def", "dim", "distribution", "else", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "forward", "future_observed_mask", "future_time_features", "future_values", "if", "is", "keepdim", "kwargs", "len", "list", "loc", "loss", "loss_weights", "min", "model", "not", "ones_like", "output_attentions", "output_distribution", "output_hidden_states", "output_params", "outputs", "params", "past_key_values", "past_observed_mask", "past_time_features", "past_values", "prediction_loss", "r", "return", "return_dict", "scale", "self", "static_categorical_features", "static_features", "static_real_features", "target_shape", "torch", "tuple", "use_cache", "use_return_dict", "weighted_average", "weights"], "autoformer/modeling_autoformer.py:AutoformerForPrediction.generate": ["False", "ModelForPrediction", "ModelPreTrainedModel", "None", "SampleTSPredictionOutput", "Tensor", "True", "bool", "cat", "class", "config", "context_length", "decoder", "decoder_input", "decoder_last_hidden", "decoder_outputs", "decomposition_layer", "def", "device", "dim", "distr", "enc_last_hidden", "encoder_hidden_states", "encoder_last_hidden_state", "expand", "expanded_static_feat", "features", "future_samples", "future_time_features", "future_values", "generate", "get_decoder", "get_lagged_subsequences", "inputs_embeds", "label_length", "lagged_sequence", "lags_shape", "last_hidden_state", "loc", "mean", "model", "num_parallel_samples", "output_attentions", "output_distribution", "output_hidden_states", "output_params", "outputs", "params", "past_observed_mask", "past_time_features", "past_values", "prediction_length", "r", "repeat", "repeat_interleave", "repeated_enc_last_hidden", "repeated_features", "repeated_loc", "repeated_past_values", "repeated_scale", "repeats", "reshape", "reshaped_lagged_sequence", "return", "return_dict", "sample", "scale", "seasonal_input", "self", "sequence", "sequences", "shape", "static_categorical_features", "static_feat", "static_features", "static_real_features", "subsequences_length", "target_shape", "time_features", "torch", "trend", "trend_init", "trend_input", "unsqueeze", "use_cache", "zeros"], "aya_vision/modeling_aya_vision.py:AyaVisionMultiModalProjector.__init__": ["ACT2FN", "LayerNorm", "Linear", "ModelConfig", "ModelMultiModalProjector", "Module", "True", "__init__", "act", "adapter_layer_norm_eps", "alignment_intermediate_size", "bias", "class", "config", "def", "downsample_factor", "eps", "getattr", "hidden_size", "layernorm", "linear_1", "linear_2", "nn", "self", "silu", "super", "text_config", "vision_config"], "aya_vision/modeling_aya_vision.py:AyaVisionMultiModalProjector.forward": ["ModelMultiModalProjector", "Module", "act", "chunk", "class", "def", "dim", "forward", "gate", "hidden_states", "image_features", "layernorm", "linear_1", "linear_2", "nn", "pixel_shuffle", "return", "self", "x"], "aya_vision/modeling_aya_vision.py:AyaVisionMultiModalProjector.pixel_shuffle": ["ModelMultiModalProjector", "Module", "batch_size", "channels", "class", "def", "downsample_factor", "feature_dim", "height", "image_features", "int", "nn", "permute", "pixel_shuffle", "reshape", "return", "self", "seq_length", "shape", "width"], "aya_vision/modeling_aya_vision.py:AyaVisionModel.__init__": ["AutoModel", "ModelConfig", "ModelModel", "ModelMultiModalProjector", "ModelPreTrainedModel", "__init__", "class", "config", "def", "from_config", "language_model", "multi_modal_projector", "post_init", "self", "super", "text_config", "vision_config", "vision_tower"], "aya_vision/modeling_aya_vision.py:AyaVisionModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "language_model", "return", "self"], "aya_vision/modeling_aya_vision.py:AyaVisionModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "language_model", "self", "set_input_embeddings", "value"], "aya_vision/modeling_aya_vision.py:AyaVisionModel.get_image_features": ["FloatTensor", "ModelModel", "ModelPreTrainedModel", "None", "True", "Unexpected", "ValueError", "cat", "class", "config", "def", "default", "dim", "else", "f", "feature", "for", "full", "get_image_features", "hidden_states", "hs", "hs_pool", "if", "image_features", "image_outputs", "in", "int", "is", "isinstance", "items", "k", "kwargs", "layer_idx", "list", "multi_modal_projector", "not", "output_hidden_states", "pixel_values", "raise", "return", "select", "selected_image_feature", "self", "str", "strategy", "torch", "v", "vision_feature_layer", "vision_feature_select_strategy", "vision_tower"], "aya_vision/modeling_aya_vision.py:AyaVisionModel.get_placeholder_mask": ["FloatTensor", "Image", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "ValueError", "all", "and", "class", "config", "def", "device", "do", "dtype", "else", "expand_as", "f", "features", "get_input_embeddings", "get_placeholder_mask", "if", "image", "image_features", "image_token_id", "input_ids", "inputs_embeds", "is", "long", "match", "n_image_features", "n_image_tokens", "not", "numel", "raise", "return", "self", "shape", "special_image_mask", "sum", "tensor", "to", "tokens", "torch", "unsqueeze"], "aya_vision/modeling_aya_vision.py:AyaVisionModel.forward": ["Cache", "FloatTensor", "LongTensor", "ModelModel", "ModelModelOutputWithPast", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "device", "dtype", "else", "exactly", "forward", "get_image_features", "get_input_embeddings", "get_placeholder_mask", "hidden_states", "if", "image_features", "image_hidden_states", "input_ids", "inputs_embeds", "int", "is", "kwargs", "language_model", "last_hidden_state", "list", "masked_scatter", "must", "not", "of", "one", "or", "outputs", "past_key_values", "pixel_values", "position_ids", "raise", "return", "self", "special_image_mask", "specify", "str", "to", "torch", "tuple", "use_cache", "vision_feature_layer", "vision_feature_select_strategy"], "aya_vision/modeling_aya_vision.py:AyaVisionForConditionalGeneration.__init__": ["False", "GenerationMixin", "Linear", "ModelConfig", "ModelForConditionalGeneration", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "text_config", "vocab_size"], "aya_vision/modeling_aya_vision.py:AyaVisionForConditionalGeneration.get_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "model", "return", "self"], "aya_vision/modeling_aya_vision.py:AyaVisionForConditionalGeneration.set_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "model", "self", "set_input_embeddings", "value"], "aya_vision/modeling_aya_vision.py:AyaVisionForConditionalGeneration.get_output_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "Module", "class", "def", "get_output_embeddings", "lm_head", "nn", "return", "self"], "aya_vision/modeling_aya_vision.py:AyaVisionForConditionalGeneration.get_image_features": ["FloatTensor", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "class", "def", "get_image_features", "int", "kwargs", "list", "model", "pixel_values", "return", "self", "str", "torch", "vision_feature_layer", "vision_feature_select_strategy"], "aya_vision/modeling_aya_vision.py:AyaVisionForConditionalGeneration.forward": ["Cache", "FloatTensor", "GenerationMixin", "LongTensor", "ModelCausalLMOutputWithPast", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "cache_position", "class", "config", "def", "else", "forward", "hidden_states", "if", "image_hidden_states", "image_sizes", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "list", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "outputs", "past_key_values", "pixel_values", "position_ids", "r", "return", "self", "slice", "slice_indices", "str", "text_config", "torch", "tuple", "vision_feature_layer", "vision_feature_select_strategy", "vocab_size"], "aya_vision/modeling_aya_vision.py:AyaVisionForConditionalGeneration.prepare_inputs_for_generation": ["False", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "True", "attention_mask", "cache_position", "class", "def", "get", "if", "input_ids", "inputs_embeds", "is_first_iteration", "kwargs", "logits_to_keep", "model_inputs", "not", "or", "past_key_values", "pixel_values", "prepare_inputs_for_generation", "return", "self", "super", "use_cache"], "bamba/modeling_bamba.py:HybridMambaAttentionDynamicCache.__init__": ["False", "ModelConfig", "ModelMambaAttentionDynamicCache", "None", "_", "__init__", "append", "batch_size", "class", "config", "conv_kernel_size", "conv_states", "def", "device", "dtype", "else", "float16", "for", "has_previous_state", "hidden_size", "i", "if", "in", "key_cache", "layers_block_type", "mamba", "mamba_d_conv", "mamba_d_head", "mamba_d_state", "mamba_expand", "mamba_n_groups", "mamba_n_heads", "num_hidden_layers", "range", "self", "ssm_state_size", "ssm_states", "tensor", "torch", "transformer_layers", "value_cache", "zeros"], "bamba/modeling_bamba.py:HybridMambaAttentionDynamicCache.__len__": ["ModelMambaAttentionDynamicCache", "__len__", "class", "def", "key_cache", "len", "return", "self"], "bamba/modeling_bamba.py:HybridMambaAttentionDynamicCache.__getitem__": ["ModelMambaAttentionDynamicCache", "__getitem__", "class", "def", "key_cache", "layer_idx", "return", "self", "value_cache"], "bamba/modeling_bamba.py:HybridMambaAttentionDynamicCache.update": ["Any", "ModelMambaAttentionDynamicCache", "None", "Tensor", "cache_kwargs", "cat", "class", "def", "dict", "dim", "else", "if", "int", "key_cache", "key_states", "layer_idx", "return", "self", "shape", "str", "torch", "tuple", "update", "value_cache", "value_states"], "bamba/modeling_bamba.py:HybridMambaAttentionDynamicCache.reorder_cache": ["LongTensor", "ModelMambaAttentionDynamicCache", "beam_idx", "class", "conv_states", "def", "device", "for", "get_seq_length", "if", "in", "index_select", "key_cache", "layer_idx", "len", "range", "reorder_cache", "self", "ssm_states", "to", "torch", "value_cache"], "bamba/modeling_bamba.py:HybridMambaAttentionDynamicCache.get_mask_sizes": ["ModelMambaAttentionDynamicCache", "Tensor", "cache_position", "class", "def", "get_mask_sizes", "get_seq_length", "int", "kv_length", "kv_offset", "layer_idx", "query_length", "return", "self", "shape", "torch", "tuple"], "bamba/modeling_bamba.py:HybridMambaAttentionDynamicCache.get_seq_length": ["ModelMambaAttentionDynamicCache", "None", "class", "def", "else", "get_seq_length", "if", "in", "int", "key_cache", "layer_idx", "len", "not", "or", "return", "self", "shape", "transformer_layers"], "bamba/modeling_bamba.py:BambaRotaryEmbedding.__init__": ["Callable", "False", "ModelConfig", "ModelRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "bamba/modeling_bamba.py:BambaRotaryEmbedding.compute_default_rope_parameters": ["ModelConfig", "ModelRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "bamba/modeling_bamba.py:BambaRotaryEmbedding.forward": ["False", "ModelRotaryEmbedding", "Module", "None", "and", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "bamba/modeling_bamba.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "bamba/modeling_bamba.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "bamba/modeling_bamba.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value", "value_states"], "bamba/modeling_bamba.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cat", "cos", "def", "dim", "k", "k_embed", "k_pass", "k_rot", "q", "q_embed", "q_pass", "q_rot", "return", "rotary_dim", "rotate_half", "shape", "sin", "torch", "unsqueeze", "unsqueeze_dim"], "bamba/modeling_bamba.py:BambaAttention.__init__": ["Linear", "ModelAttention", "ModelConfig", "Module", "True", "__init__", "attention_bias", "attention_dropout", "bias", "class", "config", "def", "getattr", "head_dim", "hidden_size", "int", "is_causal", "k_proj", "layer_idx", "nn", "num_attention_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "q_proj", "scaling", "self", "super", "v_proj"], "bamba/modeling_bamba.py:BambaAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "LongTensor", "ModelAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_attn_implementation", "apply_rotary_pos_emb", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "cache_kwargs", "cache_position", "class", "config", "contiguous", "cos", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "head_dim", "hidden_shape", "hidden_states", "if", "input_shape", "is", "k_proj", "key_states", "kwargs", "layer_idx", "nn", "not", "o_proj", "past_key_values", "position_embeddings", "q_proj", "query_states", "reshape", "return", "scaling", "self", "shape", "sin", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "view"], "bamba/modeling_bamba.py:BambaRMSNormGated.__init__": ["ModelRMSNormGated", "Module", "Parameter", "__init__", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "bamba/modeling_bamba.py:BambaRMSNormGated.forward": ["ModelRMSNormGated", "Module", "None", "True", "class", "def", "dtype", "float32", "forward", "functional", "gate", "hidden_states", "if", "input_dtype", "is", "keepdim", "mean", "nn", "not", "pow", "return", "rsqrt", "self", "silu", "to", "torch", "variance", "variance_epsilon", "weight"], "bamba/modeling_bamba.py:pad_tensor_by_size": ["Model", "Model_shape", "Model_size", "Model_tensor_by_size", "Tensor", "constant", "def", "else", "functional", "if", "input_tensor", "int", "len", "mode", "nn", "return", "shape", "torch", "value"], "bamba/modeling_bamba.py:reshape_into_chunks": ["Model", "Model_into_chunks", "chunk_size", "def", "else", "if", "input_tensor", "len", "pad_size", "pad_tensor_by_size", "return", "shape"], "bamba/modeling_bamba.py:segment_sum": ["Model_sum", "None", "bool", "chunk_size", "cumsum", "def", "device", "diagonal", "dim", "dtype", "expand", "inf", "input_tensor", "mask", "masked_fill", "ones", "return", "size", "tensor_segsum", "torch", "tril"], "bamba/modeling_bamba.py:apply_mask_to_padding_states": ["Model_mask_to_padding_states", "None", "and", "attention_mask", "def", "dtype", "hidden_states", "if", "is", "not", "return", "shape", "to"], "bamba/modeling_bamba.py:BambaMixer.__init__": ["A", "ACT2FN", "AILab", "A_log", "Conv1d", "D", "Dao", "Falling", "GPU", "Linear", "Model", "ModelConfig", "ModelMixer", "ModelRMSNormGated", "Module", "None", "Parameter", "The", "To", "__init__", "a", "act", "activation", "all", "arange", "available", "back", "be", "because", "bias", "causal", "causal_conv1d", "causal_conv1d_fn", "causal_conv1d_update", "chunk_size", "class", "com", "config", "conv1d", "conv_dim", "conv_kernel_size", "def", "dt_bias", "else", "eps", "fast", "float", "follow", "for", "getattr", "github", "global", "groups", "head_dim", "hidden_act", "hidden_size", "https", "if", "implementation", "in_channels", "in_proj", "inf", "install", "int", "intermediate_size", "is", "is_fast_path_available", "kernel_size", "layer_idx", "layer_norm_epsilon", "lazy_load_kernel", "log", "logger", "mamba", "mamba_chunk_scan_combined", "mamba_chunk_size", "mamba_conv_bias", "mamba_d_conv", "mamba_d_head", "mamba_d_state", "mamba_expand", "mamba_n_groups", "mamba_n_heads", "mamba_proj_bias", "mamba_split_conv1d_scan_combined", "mamba_ssm", "model", "n_groups", "naive", "nn", "norm", "not", "num_heads", "of", "on", "one", "ones", "out_channels", "out_proj", "padding", "path", "projection_size", "rms_norm_eps", "running", "selective_state_update", "self", "spaces", "ssm", "ssm_state_size", "state", "super", "the", "time_step_limit", "time_step_max", "time_step_min", "to", "torch", "use_bias", "use_conv_bias", "used", "warning_once", "when", "will"], "bamba/modeling_bamba.py:BambaMixer.cuda_kernels_forward": ["A", "A_log", "B", "C", "D", "False", "HybridMambaAttentionDynamicCache", "IntTensor", "LongTensor", "ModelMixer", "Module", "None", "Tensor", "True", "_", "act", "activation", "and", "apply_mask_to_padding_states", "attention_mask", "batch_size", "bias", "cache_params", "cache_position", "causal_conv1d_fn", "causal_conv1d_update", "chunk_size", "class", "conv1d", "conv_dim", "conv_kernel_size", "conv_states", "copy_", "cuda_kernels_forward", "def", "dim", "dt", "dt_bias", "dt_limit", "dt_limit_kwargs", "dt_softplus", "dtype", "else", "exp", "expand", "float", "float32", "functional", "gate", "groups_time_state_size", "has_previous_state", "head_dim", "headdim", "hidden_states", "hidden_states_B_C", "hidden_states_B_C_transposed", "hidden_states_reshaped", "if", "in", "in_proj", "inf", "intermediate_size", "is", "layer_idx", "mamba_chunk_scan_combined", "mamba_split_conv1d_scan_combined", "n_groups", "ngroups", "nn", "norm", "norm_before_gate", "not", "num_heads", "out", "out_proj", "outproj_bias", "outproj_weight", "pad", "projected_states", "return", "return_final_states", "rmsnorm_eps", "rmsnorm_weight", "scan_output", "selective_state_update", "self", "seq_idx", "seq_len", "shape", "silu", "split", "squeeze", "ssm_state", "ssm_state_size", "ssm_states", "swish", "time_step_limit", "to", "torch", "training", "transpose", "use_precomputed_states", "variance_epsilon", "view", "weight", "x", "z"], "bamba/modeling_bamba.py:BambaMixer.torch_forward": ["A", "A_cumsum", "A_log", "B", "B_decay", "C", "C_reshaped", "C_times_states", "D", "D_residual", "G", "G_intermediate", "HybridMambaAttentionDynamicCache", "L", "LongTensor", "M", "M_intermediate", "ModelMixer", "Module", "None", "Tensor", "Y_diag", "Y_off", "_", "act", "and", "apply_mask_to_padding_states", "attention_mask", "batch_size", "bias", "bmm", "cache_device", "cache_params", "cache_position", "cat", "chunk_size", "clamp", "class", "contextualized_states", "contiguous", "conv1d", "conv_dim", "conv_kernel_size", "conv_states", "copy_", "cumsum", "dA", "dB", "dBx", "decay_chunk", "decay_states", "def", "device", "dim", "dims", "dt", "dt_bias", "dtype", "else", "exp", "expand", "float", "float32", "for", "functional", "gate", "has_previous_state", "head_dim", "hidden_states", "hidden_states_B_C", "hidden_states_B_C_transposed", "if", "in", "in_proj", "input_states", "intermediate_size", "is", "layer_idx", "n_groups", "new_states", "nn", "norm", "not", "num_heads", "out_proj", "output_size", "pad", "pad_size", "pad_tensor_by_size", "permute", "previous_states", "projected_states", "repeat_interleave", "reshape", "reshape_into_chunks", "return", "roll", "scan_output", "segment_sum", "self", "seq_len", "shape", "shifts", "softplus", "split", "squeeze", "ssm_state", "ssm_state_size", "ssm_states", "ssm_states_reshaped", "state_decay_out", "state_decay_out_permuted", "states", "sum", "t", "time_step_limit", "to", "torch", "torch_forward", "transpose", "use_conv_bias", "use_precomputed_states", "view", "weight", "y", "zeros_like"], "bamba/modeling_bamba.py:BambaMixer.forward": ["HybridMambaAttentionDynamicCache", "IntTensor", "LongTensor", "ModelMixer", "Module", "None", "NotImplementedError", "Please", "Tensor", "and", "attention_mask", "cache_params", "cache_position", "causal_conv1d", "class", "cuda", "cuda_kernels_forward", "def", "device", "dtype", "fast", "forward", "hidden_states", "if", "in", "in_proj", "install", "is", "is_fast_path_available", "kwargs", "mamba_ssm", "nn", "not", "path", "raise", "requires", "return", "self", "seq_idx", "shape", "support", "to", "torch", "torch_forward", "type", "weight"], "bamba/modeling_bamba.py:BambaMLP.__init__": ["ACT2FN", "Linear", "ModelMLP", "Module", "__init__", "act_fn", "bias", "class", "config", "def", "down_proj", "gate_proj", "hidden_act", "hidden_size", "intermediate_size", "mlp_bias", "nn", "self", "super", "up_proj"], "bamba/modeling_bamba.py:BambaMLP.forward": ["ModelMLP", "Module", "act_fn", "class", "def", "down_proj", "forward", "gate_proj", "nn", "return", "self", "up_proj", "x"], "bamba/modeling_bamba.py:BambaRMSNorm.__init__": ["ModelRMSNorm", "Module", "Parameter", "__init__", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "bamba/modeling_bamba.py:BambaRMSNorm.forward": ["ModelRMSNorm", "Module", "True", "class", "def", "dtype", "float32", "forward", "hidden_states", "input_dtype", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "bamba/modeling_bamba.py:BambaRMSNorm.extra_repr": ["ModelRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "variance_epsilon", "weight"], "bamba/modeling_bamba.py:BambaDecoderLayer.__init__": ["GradientCheckpointingLayer", "Invalid", "ModelAttention", "ModelConfig", "ModelDecoderLayer", "ModelMLP", "ModelMixer", "ModelRMSNorm", "None", "ValueError", "__init__", "attention", "class", "config", "def", "elif", "else", "eps", "feed_forward", "ffn_layer_class", "hidden_size", "if", "input_layernorm", "int", "layer_idx", "layer_type", "mamba", "num_experts", "pre_ff_layernorm", "raise", "rms_norm_eps", "self", "self_attn", "str", "super"], "bamba/modeling_bamba.py:BambaDecoderLayer.forward": ["False", "FloatTensor", "GradientCheckpointingLayer", "HybridMambaAttentionDynamicCache", "LongTensor", "ModelDecoderLayer", "ModelFlashAttentionKwargs", "None", "Tensor", "Unpack", "attention", "attention_mask", "bool", "cache_params", "cache_position", "class", "def", "elif", "feed_forward", "forward", "hidden_states", "if", "input_layernorm", "kwargs", "layer_type", "mamba", "output_attentions", "outputs", "past_key_values", "position_embeddings", "position_ids", "pre_ff_layernorm", "residual", "return", "self", "self_attn", "self_attn_weights", "torch", "tuple", "use_cache"], "bamba/modeling_bamba.py:BambaPreTrainedModel._init_weights": ["A_log", "D", "ModelMixer", "ModelPreTrainedModel", "PreTrainedModel", "_init_weights", "arange", "class", "copy_", "def", "dt_bias", "if", "init", "isinstance", "log", "module", "num_heads", "ones_", "self", "super", "torch"], "bamba/modeling_bamba.py:BambaModel.__init__": ["Embedding", "False", "ModelConfig", "ModelDecoderLayer", "ModelModel", "ModelPreTrainedModel", "ModelRMSNorm", "ModelRotaryEmbedding", "ModuleList", "__init__", "_attn_implementation", "append", "class", "config", "decoder_layers", "def", "embed_tokens", "eps", "final_layernorm", "for", "gradient_checkpointing", "hidden_size", "i", "in", "layer_idx", "layer_type", "layers", "layers_block_type", "nn", "num_hidden_layers", "pad_token_id", "padding_idx", "post_init", "range", "rms_norm_eps", "rotary_emb", "self", "super", "vocab_size"], "bamba/modeling_bamba.py:BambaModel.forward": ["BaseModelOutputWithPast", "False", "FloatTensor", "HybridMambaAttentionDynamicCache", "LongTensor", "Model", "ModelFlashAttentionKwargs", "ModelModel", "ModelPreTrainedModel", "None", "Setting", "Tensor", "True", "Unpack", "ValueError", "You", "_update_causal_mask", "_update_mamba_mask", "a", "all_hidden_states", "all_self_attns", "an", "and", "arange", "attention_mask", "attentions", "be", "bool", "cache", "cache_position", "causal_mask", "checkpointing", "class", "config", "decoder_layer", "def", "device", "else", "embed_tokens", "exactly", "final_layernorm", "for", "forward", "gradient", "gradient_checkpointing", "has_previous_state", "hidden_states", "if", "in", "incompatible", "initialized", "input_ids", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layer_mask", "layer_outputs", "layer_type", "layers", "logger", "mamba", "mamba_mask", "must", "next_cache", "no", "not", "of", "one", "or", "output_attentions", "output_hidden_states", "past_key_values", "position_embeddings", "position_ids", "provided", "raise", "requires", "return", "returned", "rotary_emb", "self", "shape", "so", "specify", "to", "torch", "training", "unsqueeze", "use_cache", "warning_once", "was", "will", "with"], "bamba/modeling_bamba.py:BambaModel._update_causal_mask": ["AttentionMaskConverter", "HybridMambaAttentionDynamicCache", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "_attn_implementation", "_ignore_causal_mask_sdpa", "_prepare_4d_causal_attention_mask_with_cache_position", "_unmask_unattended", "_update_causal_mask", "and", "attention_mask", "batch_size", "bool", "cache_position", "causal_mask", "class", "config", "cuda", "def", "device", "dtype", "else", "finfo", "flash_attention_2", "get_seq_length", "if", "in", "input_tensor", "inputs_embeds", "is", "is_training", "isinstance", "min", "min_dtype", "not", "npu", "output_attentions", "past_key_values", "past_key_values_length", "past_seen_tokens", "return", "sdpa", "self", "sequence_length", "shape", "target_length", "torch", "training", "type", "xpu"], "bamba/modeling_bamba.py:BambaModel._prepare_4d_causal_attention_mask_with_cache_position": ["ModelModel", "ModelPreTrainedModel", "None", "Tensor", "_prepare_4d_causal_attention_mask_with_cache_position", "and", "arange", "attention_mask", "batch_size", "cache_position", "causal_mask", "class", "clone", "def", "device", "diagonal", "dim", "dtype", "else", "expand", "fill_value", "finfo", "full", "if", "int", "is", "kwargs", "mask_length", "masked_fill", "min", "min_dtype", "not", "padding_attention_mask", "padding_mask", "reshape", "return", "sequence_length", "shape", "target_length", "to", "torch", "triu"], "bamba/modeling_bamba.py:BambaModel._update_mamba_mask": ["ModelModel", "ModelPreTrainedModel", "None", "_update_mamba_mask", "all", "and", "attention_mask", "cache_position", "class", "def", "if", "is", "mamba_mask", "not", "or", "return", "self", "torch"], "bamba/modeling_bamba.py:BambaForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "ModelForCausalLM", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "vocab_size", "z_loss_coefficient"], "bamba/modeling_bamba.py:BambaForCausalLM.forward": ["BaseModelOutputWithPast", "CausalLMOutputWithPast", "FloatTensor", "GenerationMixin", "HybridMambaAttentionDynamicCache", "LongTensor", "ModelForCausalLM", "ModelPreTrainedModel", "None", "Tensor", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "dim", "dtype", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "logits", "logits_to_keep", "logsumexp", "loss", "loss_function", "mean", "model", "not", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "position_ids", "pow", "r", "return", "self", "slice", "slice_indices", "to", "torch", "use_cache", "vocab_size", "z_loss", "z_loss_coefficient"], "bamba/modeling_bamba.py:BambaForCausalLM.prepare_inputs_for_generation": ["False", "GenerationMixin", "HybridMambaAttentionDynamicCache", "ModelForCausalLM", "ModelPreTrainedModel", "None", "True", "and", "attention_mask", "cache_position", "class", "config", "contiguous", "cumsum", "def", "device", "dtype", "elif", "else", "empty_past_kv", "for", "if", "in", "input_ids", "inputs_embeds", "is", "is_first_iteration", "items", "key", "kwargs", "logits_to_keep", "long", "masked_fill_", "model_inputs", "not", "num_logits_to_keep", "or", "past_key_values", "position_ids", "prepare_inputs_for_generation", "return", "self", "shape", "update", "use_cache", "value"], "bark/modeling_bark.py:BarkSelfAttention.__init__": ["Dropout", "False", "Linear", "ModelSelfAttention", "Module", "None", "ValueError", "__init__", "and", "att_proj", "attn_dropout", "be", "bias", "block_size", "bool", "by", "class", "config", "def", "divisible", "dropout", "dtype", "embed_dim", "f", "got", "head_dim", "hidden_size", "if", "is_causal", "layer_idx", "must", "nn", "num_heads", "ones", "out_proj", "raise", "register_buffer", "resid_dropout", "self", "super", "torch", "tril", "view"], "bark/modeling_bark.py:BarkSelfAttention._split_heads": ["ModelSelfAttention", "Module", "_split_heads", "attn_head_size", "class", "def", "new_shape", "nn", "num_heads", "permute", "return", "self", "size", "tensor", "view"], "bark/modeling_bark.py:BarkSelfAttention._merge_heads": ["ModelSelfAttention", "Module", "_merge_heads", "attn_head_size", "class", "contiguous", "def", "nn", "num_heads", "return", "self", "size", "tensor", "transpose", "view"], "bark/modeling_bark.py:BarkSelfAttention._attn": ["ModelSelfAttention", "Module", "None", "_attn", "attention_mask", "attn_dropout", "attn_output", "attn_weights", "bias", "class", "def", "dim", "dtype", "finfo", "functional", "head_dim", "if", "is", "is_causal", "key", "key_length", "masked_fill", "math", "matmul", "min", "nn", "not", "query", "query_length", "return", "self", "size", "softmax", "sqrt", "to", "torch", "transpose", "value"], "bark/modeling_bark.py:BarkSelfAttention.forward": ["False", "ModelSelfAttention", "Module", "None", "_attn", "_merge_heads", "_split_heads", "att_proj", "attention_mask", "attn_output", "attn_weights", "cache_position", "class", "def", "dim", "embed_dim", "forward", "head_dim", "hidden_states", "if", "is", "key", "layer_idx", "nn", "not", "num_heads", "out_proj", "output_attentions", "past_key_values", "query", "resid_dropout", "return", "self", "split", "update", "use_cache", "value"], "bark/modeling_bark.py:BarkSelfFlashAttention2.__init__": ["ModelSelfAttention", "ModelSelfFlashAttention2", "__init__", "_flash_attn_uses_top_left_mask", "args", "class", "def", "flash_attn_supports_top_left_mask", "kwargs", "self", "super"], "bark/modeling_bark.py:BarkSelfFlashAttention2._split_heads": ["ModelSelfAttention", "ModelSelfFlashAttention2", "_split_heads", "attn_head_size", "class", "def", "new_shape", "num_heads", "return", "self", "size", "tensor", "view"], "bark/modeling_bark.py:BarkSelfFlashAttention2._merge_heads": ["ModelSelfAttention", "ModelSelfFlashAttention2", "_merge_heads", "attn_head_size", "class", "def", "num_heads", "return", "self", "size", "tensor", "view"], "bark/modeling_bark.py:BarkSelfFlashAttention2.forward": ["False", "ModelSelfAttention", "ModelSelfFlashAttention2", "None", "_", "_flash_attention_forward", "_flash_attn_uses_top_left_mask", "_merge_heads", "_split_heads", "att_proj", "attention_mask", "attn_output", "batch_size", "cache_position", "class", "def", "dim", "dropout", "else", "embed_dim", "forward", "get_target_dtype", "head_dim", "hidden_states", "if", "is", "is_causal", "key", "layer_idx", "not", "num_heads", "out_proj", "output_attentions", "past_key_values", "query", "query_len", "resid_dropout", "return", "self", "size", "split", "target_dtype", "training", "update", "use_cache", "use_top_left_mask", "value"], "bark/modeling_bark.py:BarkMLP.__init__": ["Dropout", "GELU", "Linear", "ModelMLP", "Module", "__init__", "bias", "class", "config", "def", "dropout", "gelu", "hidden_size", "in_proj", "nn", "out_proj", "self", "super"], "bark/modeling_bark.py:BarkMLP.forward": ["ModelMLP", "Module", "class", "def", "dropout", "forward", "gelu", "hidden_states", "in_proj", "nn", "out_proj", "return", "self"], "bark/modeling_bark.py:BarkBlock.__init__": ["False", "GradientCheckpointingLayer", "LayerNorm", "ModelBlock", "ModelMLP", "Model_ATTENTION_CLASSES", "None", "__init__", "_attn_implementation", "attn", "bias", "class", "config", "def", "else", "hidden_size", "if", "is_causal", "layer_idx", "layernorm_1", "layernorm_2", "mlp", "nn", "self", "super"], "bark/modeling_bark.py:BarkBlock.forward": ["False", "GradientCheckpointingLayer", "ModelBlock", "None", "attention_mask", "attn", "attn_output", "attn_outputs", "cache_position", "class", "def", "forward", "hidden_states", "intermediary_hidden_states", "layernorm_1", "layernorm_2", "mlp", "output_attentions", "outputs", "past_key_values", "return", "self", "use_cache"], "bark/modeling_bark.py:BarkPreTrainedModel.device": ["ModelPreTrainedModel", "None", "PreTrainedModel", "_hf_hook", "and", "class", "def", "device", "execution_device", "for", "hasattr", "if", "in", "is", "module", "modules", "not", "return", "self", "super", "torch"], "bark/modeling_bark.py:BarkPreTrainedModel._init_weights": ["ModelPreTrainedModel", "ModelSelfAttention", "PreTrainedModel", "_init_weights", "bias", "block_size", "bool", "class", "config", "copy_", "def", "dtype", "if", "init", "is_causal", "isinstance", "module", "ones", "self", "super", "torch", "tril", "view"], "bark/modeling_bark.py:BarkCausalModel.__init__": ["Dropout", "Embedding", "False", "GenerationMixin", "LayerNorm", "Linear", "ModelBlock", "ModelCausalModel", "ModelPreTrainedModel", "ModuleList", "True", "__init__", "bias", "block_size", "class", "config", "def", "drop", "dropout", "for", "gradient_checkpointing", "hidden_size", "i", "in", "input_embeds_layer", "input_vocab_size", "is_causal", "layer_idx", "layernorm_final", "layers", "lm_head", "nn", "num_layers", "output_vocab_size", "position_embeds_layer", "post_init", "range", "self", "super"], "bark/modeling_bark.py:BarkCausalModel.get_output_embeddings": ["GenerationMixin", "ModelCausalModel", "ModelPreTrainedModel", "None", "class", "def", "get_output_embeddings", "return", "self"], "bark/modeling_bark.py:BarkCausalModel.get_input_embeddings": ["GenerationMixin", "ModelCausalModel", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "input_embeds_layer", "return", "self"], "bark/modeling_bark.py:BarkCausalModel.set_input_embeddings": ["GenerationMixin", "ModelCausalModel", "ModelPreTrainedModel", "class", "def", "input_embeds_layer", "new_embeddings", "self", "set_input_embeddings"], "bark/modeling_bark.py:BarkCausalModel.prepare_inputs_for_generation": ["GenerationMixin", "ModelCausalModel", "ModelPreTrainedModel", "None", "attention_mask", "cache_position", "class", "def", "input_embeds", "input_ids", "inputs_embeds", "kwargs", "model_inputs", "past_key_values", "pop", "position_ids", "prepare_inputs_for_generation", "return", "self", "super", "use_cache"], "bark/modeling_bark.py:BarkCausalModel.forward": ["Cache", "CausalLMOutputWithPast", "DynamicCache", "False", "GenerationMixin", "LongTensor", "Model", "ModelCausalModel", "ModelPreTrainedModel", "None", "NotImplementedError", "Setting", "Tensor", "Training", "True", "ValueError", "You", "_attn_implementation", "_prepare_4d_attention_mask", "all_hidden_states", "all_self_attentions", "and", "arange", "at", "attention_mask", "attentions", "batch_size", "be", "block", "bool", "both", "cache_position", "cannot", "checkpointing", "class", "config", "def", "defined", "device", "do", "drop", "dtype", "either", "elif", "else", "ensure", "enumerate", "flash_attention_2", "for", "forward", "get_seq_length", "gradient", "gradient_checkpointing", "has", "have", "hidden_states", "i", "if", "implemented", "in", "incompatible", "input_embeds", "input_embeds_layer", "input_ids", "input_shape", "is", "kwargs", "labels", "layernorm_final", "layers", "lm_head", "logger", "logits", "long", "loss", "model", "not", "or", "output_attentions", "output_hidden_states", "output_shape", "outputs", "pass", "past_key_values", "past_length", "position_embeds", "position_embeds_layer", "position_ids", "r", "raise", "return", "return_dict", "same", "self", "seq_length", "shape", "size", "specify", "tgt_len", "the", "time", "to", "torch", "training", "tuple", "unsqueeze", "use_cache", "use_return_dict", "v", "view", "warning_once", "with", "yet", "you"], "bark/modeling_bark.py:BarkSemanticModel.generate": ["False", "LongTensor", "ModelCausalModel", "ModelEosPrioritizerLogitsProcessor", "ModelSemanticGenerationConfig", "ModelSemanticModel", "None", "SuppressTokensLogitsProcessor", "Tensor", "ValueError", "attention_mask", "batch_size", "be", "bool", "cat", "class", "config", "constant", "def", "device", "dict", "dim", "dtype", "early_stopping_logits_processor", "else", "eos_token_id", "extend", "functional", "generate", "generation_config", "get", "has", "history_prompt", "if", "infer_array", "input_embeds", "input_embeds_layer", "input_ids", "int", "is", "kwargs", "len", "list", "logits_processor", "masked_fill", "max_input_semantic_length", "min_eos_p", "mode", "nn", "not", "ones", "output_vocab_size", "pad", "provided", "raise", "range", "repeat_interleave", "return", "return_dict_in_generate", "self", "semantic_generation_config", "semantic_history", "semantic_infer_token", "semantic_output", "semantic_pad_token", "semantic_prompt", "semantic_vocab_size", "sequences", "shape", "str", "super", "suppress_tokens_logits_processor", "tensor", "text_encoding_offset", "text_pad_token", "to", "tokens_to_suppress", "torch", "value"], "bark/modeling_bark.py:BarkCoarseModel.preprocess_histories": ["ModelCausalModel", "ModelCoarseModel", "None", "Tensor", "batch_size", "class", "clone", "coarse_prompt", "codebook_size", "def", "device", "dict", "dim", "dtype", "else", "floor", "for", "history_prompt", "if", "in", "int", "is", "max_coarse_history", "max_semantic_history", "min", "n", "n_coarse_hist_provided", "n_semantic_hist_provided", "not", "np", "preprocess_histories", "range", "repeat_interleave", "reshape", "return", "round", "self", "semantic_generation_config", "semantic_prompt", "semantic_to_coarse_ratio", "semantic_vocab_size", "shape", "str", "tensor", "torch", "transpose", "x_coarse_history", "x_semantic_history"], "bark/modeling_bark.py:BarkCoarseModel.generate": ["AlternatingCodebooksLogitsProcessor", "F", "False", "LongTensor", "ModelCausalModel", "ModelCoarseGenerationConfig", "ModelCoarseModel", "ModelSemanticGenerationConfig", "None", "Tensor", "ValueError", "_", "alternatingLogitsProcessor", "base_semantic_idx", "batch_size", "be", "bool", "ceil", "class", "coarse_generation_config", "coarse_infer_token", "coarse_output", "coarse_rate_hz", "coarse_semantic_pad_token", "codebook_size", "constant", "def", "del", "device", "dict", "else", "floor", "for", "generate", "generation_config", "get", "has", "history_prompt", "hstack", "if", "in", "input_coarse", "input_coarse_len", "int", "is", "item", "kwargs", "len_coarse_history", "logits_processor", "masked_fill_", "max", "max_coarse_history", "max_coarse_input_length", "max_generated_len", "max_new_tokens", "max_semantic_history", "min", "n_coarse_codebooks", "n_window_steps", "np", "output_coarse", "output_lengths", "pad", "preprocess_histories", "provided", "raise", "range", "return", "return_dict_in_generate", "return_output_lengths", "round", "self", "semantic_generation_config", "semantic_idx", "semantic_output", "semantic_pad_token", "semantic_rate_hz", "semantic_to_coarse_ratio", "semantic_vocab_size", "sequences", "shape", "sliding_window_len", "str", "sum", "super", "tensor", "to", "torch", "total_generated_len", "tuple", "x_coarse", "x_semantic_history"], "bark/modeling_bark.py:BarkFineModel.__init__": ["Dropout", "Embedding", "False", "LayerNorm", "Linear", "ModelBlock", "ModelFineModel", "ModelPreTrainedModel", "ModuleList", "_", "__init__", "_tied_weights_keys", "bias", "block_size", "class", "config", "def", "drop", "dropout", "f", "for", "gradient_checkpointing", "hidden_size", "i", "in", "input_embeds_layers", "input_vocab_size", "is_causal", "layer_idx", "layernorm_final", "layers", "lm_heads", "n_codes_given", "n_codes_total", "nn", "num_layers", "output_vocab_size", "position_embeds_layer", "post_init", "range", "self", "super", "weight"], "bark/modeling_bark.py:BarkFineModel.get_input_embeddings": ["ModelFineModel", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "input_embeds_layers", "return", "self"], "bark/modeling_bark.py:BarkFineModel.set_input_embeddings": ["ModelFineModel", "ModelPreTrainedModel", "class", "def", "input_embeds_layers", "new_embeddings", "self", "set_input_embeddings"], "bark/modeling_bark.py:BarkFineModel.get_output_embeddings": ["ModelFineModel", "ModelPreTrainedModel", "class", "def", "get_output_embeddings", "lm_heads", "return", "self"], "bark/modeling_bark.py:BarkFineModel.set_output_embeddings": ["ModelFineModel", "ModelPreTrainedModel", "class", "def", "lm_heads", "new_output_embeddings", "self", "set_output_embeddings"], "bark/modeling_bark.py:BarkFineModel._resize_token_embeddings": ["ModelFineModel", "ModelPreTrainedModel", "ModuleList", "None", "True", "_get_resized_embeddings", "_get_resized_lm_head", "_resize_token_embeddings", "and", "class", "config", "def", "for", "get_input_embeddings", "get_output_embeddings", "if", "in", "is", "mean_resizing", "new_embeddings_list", "new_lm_head_list", "new_num_tokens", "nn", "not", "old_embeddings", "old_embeddings_list", "old_lm_head", "old_lm_head_list", "pad_to_multiple_of", "return", "self", "set_input_embeddings", "set_output_embeddings", "shape", "tie_word_embeddings", "weight"], "bark/modeling_bark.py:BarkFineModel.resize_token_embeddings": ["Embedding", "ModelFineModel", "ModelPreTrainedModel", "None", "True", "_resize_token_embeddings", "and", "bool", "class", "config", "def", "if", "int", "is", "mean_resizing", "model_embeds", "new_num_tokens", "nn", "output_vocab_size", "pad_to_multiple_of", "resize_token_embeddings", "return", "self", "shape", "tie_weights", "vocab_size", "weight"], "bark/modeling_bark.py:BarkFineModel.forward": ["Cannot", "LongTensor", "MaskedLMOutput", "ModelFineModel", "ModelPreTrainedModel", "None", "NotImplementedError", "Tensor", "Training", "ValueError", "You", "_attn_implementation", "_prepare_4d_attention_mask", "all_hidden_states", "all_self_attentions", "and", "arange", "at", "attention_mask", "attentions", "batch_size", "be", "block", "bool", "both", "by", "cannot", "cat", "class", "coarse", "codebook", "codebook_idx", "config", "def", "defined", "device", "dim", "drop", "dtype", "either", "else", "enumerate", "flash_attention_2", "for", "forward", "has", "have", "hidden_states", "i", "if", "implemented", "in", "input_embeds", "input_embeds_layer", "input_embeds_layers", "input_ids", "input_shape", "int", "is", "kwargs", "labels", "layernorm_final", "layers", "lm_heads", "logits", "long", "loss", "model", "n_codes_given", "not", "or", "output_attentions", "output_hidden_states", "output_shape", "outputs", "position_embeds", "position_embeds_layer", "position_ids", "predict", "predicted", "r", "raise", "return", "return_dict", "same", "self", "seq_length", "shape", "should", "size", "specify", "sum", "tgt_len", "the", "time", "to", "torch", "tuple", "unsqueeze", "use_return_dict", "v", "view", "yet"], "bark/modeling_bark.py:BarkFineModel.generate": ["F", "LongTensor", "ModelCoarseGenerationConfig", "ModelFineGenerationConfig", "ModelFineModel", "ModelPreTrainedModel", "ModelSemanticGenerationConfig", "None", "T", "Tensor", "ValueError", "and", "argmax", "batch_size", "be", "cat", "ceil", "class", "coarse_generation_config", "coarse_output", "codebook_preds", "codebook_size", "constant", "def", "del", "dict", "dim", "else", "fine_generation_config", "fine_input", "fine_prompt", "for", "forward", "generate", "get", "has", "have", "history_prompt", "if", "in", "input", "input_buffer", "int", "int32", "is", "kwargs", "logits", "max", "max_fine_history_length", "max_fine_input_length", "min", "mode", "multinomial", "n_coarse", "n_coarse_codebooks", "n_fine_codebooks", "n_history", "n_inner", "n_loops", "n_outer", "n_remove_from_end", "not", "np", "num_samples", "or", "output", "pad", "probs", "provided", "raise", "range", "rel_start_fill_idx", "relevant_logits", "remainder", "repeat_interleave", "reshape", "return", "same", "self", "semantic_generation_config", "semantic_vocab_size", "seq_len", "shape", "should", "softmax", "start_fill_idx", "start_idx", "str", "temperature", "the", "to", "torch", "transpose", "value", "view", "x_fine_history"], "bark/modeling_bark.py:BarkModel.__init__": ["AutoModel", "GenerationMixin", "ModelCoarseModel", "ModelFineModel", "ModelModel", "ModelPreTrainedModel", "ModelSemanticModel", "__init__", "class", "coarse_acoustics", "coarse_acoustics_config", "codec_config", "codec_model", "config", "def", "fine_acoustics", "fine_acoustics_config", "from_config", "post_init", "self", "semantic", "semantic_config", "super"], "bark/modeling_bark.py:BarkModel.can_generate": ["GenerationMixin", "ModelModel", "ModelPreTrainedModel", "True", "bool", "can_generate", "class", "cls", "def", "return"], "bark/modeling_bark.py:BarkModel.device": ["GenerationMixin", "ModelModel", "ModelPreTrainedModel", "None", "_hf_hook", "and", "class", "def", "device", "execution_device", "for", "hasattr", "if", "in", "is", "module", "modules", "not", "return", "self", "semantic", "super", "torch"], "bark/modeling_bark.py:BarkModel.enable_cpu_offload": ["GenerationMixin", "ImportError", "ModelModel", "ModelPreTrainedModel", "None", "_", "accelerate", "accelerator", "accelerator_id", "class", "coarse_acoustics", "codec_model", "codec_model_hook", "cpu", "cpu_offload_with_hook", "cpu_offloaded_model", "cuda", "current_accelerator", "def", "device", "device_type", "else", "empty_cache", "enable_cpu_offload", "enable_model_cpu_offload", "f", "fine_acoustics", "fine_acoustics_hook", "for", "getattr", "hook", "if", "in", "input_embeds_layer", "int", "is_accelerate_available", "is_torch_accelerator_available", "kwargs", "prev_module_hook", "r", "raise", "requires", "self", "semantic", "to", "torch", "torch_accelerator_module", "type"], "bark/modeling_bark.py:BarkModel.codec_decode": ["GenerationMixin", "ModelModel", "ModelPreTrainedModel", "None", "audio_arr", "class", "codec_decode", "codec_model", "decode", "decoder", "def", "else", "emb", "fine_output", "for", "if", "in", "is", "l", "not", "out", "output_lengths", "quantizer", "return", "sample", "self", "squeeze", "transpose", "unsqueeze", "zip"], "bark/modeling_bark.py:BarkModel.generate": ["GenerationMixin", "LongTensor", "ModelCoarseGenerationConfig", "ModelFineGenerationConfig", "ModelModel", "ModelPreTrainedModel", "ModelSemanticGenerationConfig", "None", "Tensor", "True", "attention_mask", "audio", "batch_first", "bool", "class", "coarse_", "coarse_acoustics", "coarse_acoustics_config", "coarse_generation_config", "coarse_output", "codebook_size", "codec_decode", "codec_model", "codec_model_hook", "def", "device", "dict", "elif", "else", "fine_", "fine_acoustics", "fine_acoustics_config", "fine_acoustics_hook", "fine_generation_config", "for", "generate", "generation_config", "getattr", "history_prompt", "if", "in", "input_ids", "is", "items", "key", "kwargs", "kwargs_coarse", "kwargs_fine", "kwargs_semantic", "len", "min_eos_p", "n_coarse_codebooks", "nn", "not", "offload", "output", "output_lengths", "pad_sequence", "padding_value", "pop", "return", "return_output_lengths", "rnn", "sample", "self", "semantic", "semantic_", "semantic_config", "semantic_generation_config", "semantic_output", "startswith", "str", "to", "torch", "utils", "value"], "bart/modeling_bart.py:shift_tokens_right": ["Model_tokens_right", "Modeled_input_ids", "None", "Tensor", "ValueError", "be", "clone", "config", "decoder_start_token_id", "def", "defined", "has", "if", "input_ids", "int", "is", "masked_fill_", "model", "new_zeros", "pad_token_id", "raise", "return", "self", "shape", "to", "torch"], "bart/modeling_bart.py:BartLearnedPositionalEmbedding.__init__": ["Embedding", "ModelLearnedPositionalEmbedding", "__init__", "class", "def", "embedding_dim", "int", "nn", "num_embeddings", "offset", "self", "super"], "bart/modeling_bart.py:BartLearnedPositionalEmbedding.forward": ["Embedding", "ModelLearnedPositionalEmbedding", "None", "Tensor", "arange", "bsz", "class", "def", "device", "dtype", "else", "expand", "forward", "if", "input_ids", "int", "is", "long", "nn", "offset", "past_key_values_length", "position_ids", "return", "self", "seq_len", "shape", "super", "torch", "unsqueeze", "weight"], "bart/modeling_bart.py:BartScaledWordEmbedding.__init__": ["Embedding", "ModelScaledWordEmbedding", "None", "__init__", "class", "def", "embed_scale", "embedding_dim", "float", "int", "nn", "num_embeddings", "padding_idx", "self", "super"], "bart/modeling_bart.py:BartScaledWordEmbedding.forward": ["Embedding", "ModelScaledWordEmbedding", "Tensor", "class", "def", "embed_scale", "forward", "input_ids", "nn", "return", "self", "super", "torch"], "bart/modeling_bart.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "contiguous", "def", "dim", "dropout", "float", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "shape", "size", "softmax", "torch", "training", "transpose", "value"], "bart/modeling_bart.py:BartAttention.__init__": ["False", "Instantiating", "Linear", "ModelAttention", "ModelConfig", "Module", "None", "Please", "True", "ValueError", "__class__", "__init__", "__name__", "a", "and", "be", "bias", "bool", "by", "caching", "call", "class", "config", "creating", "decoder", "def", "divisible", "dropout", "during", "embed_dim", "errors", "f", "float", "forward", "got", "head_dim", "if", "int", "is", "is_causal", "is_decoder", "k_proj", "layer_idx", "lead", "logger", "make", "must", "nn", "not", "num_heads", "out_proj", "passing", "provide", "q_proj", "raise", "recommended", "scaling", "self", "super", "sure", "the", "this", "to", "used", "v_proj", "warning_once", "when", "will", "without"], "bart/modeling_bart.py:BartAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "EncoderDecoderCache", "False", "FlashAttentionKwargs", "ModelAttention", "Module", "None", "Tensor", "True", "Unpack", "_attn_implementation", "and", "attention_interface", "attention_mask", "attn_output", "attn_weights", "bool", "bsz", "cache_position", "class", "config", "contiguous", "cross_attention_cache", "curr_past_key_values", "current_states", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "get", "head_dim", "hidden_states", "if", "is", "is_cross_attention", "is_updated", "isinstance", "k_proj", "key_states", "key_value_states", "keys", "kv_input_shape", "kwargs", "layer_idx", "layers", "nn", "not", "out_proj", "output_attentions", "past_key_values", "q_input_shape", "q_proj", "query_states", "reshape", "return", "scaling", "self", "self_attention_cache", "shape", "src_len", "tgt_len", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "values", "view"], "bart/modeling_bart.py:BartEncoderLayer.__init__": ["ACT2FN", "GradientCheckpointingLayer", "LayerNorm", "Linear", "ModelAttention", "ModelConfig", "ModelEncoderLayer", "None", "__init__", "activation_dropout", "activation_fn", "activation_function", "attention_dropout", "class", "config", "d_model", "def", "dropout", "embed_dim", "encoder_attention_heads", "encoder_ffn_dim", "fc1", "fc2", "final_layer_norm", "int", "layer_idx", "nn", "num_heads", "self", "self_attn", "self_attn_layer_norm", "super"], "bart/modeling_bart.py:BartEncoderLayer.forward": ["False", "FloatTensor", "GradientCheckpointingLayer", "ModelEncoderLayer", "None", "activation_dropout", "activation_fn", "and", "any", "attention_mask", "attn_weights", "bool", "clamp", "clamp_value", "class", "def", "dropout", "dtype", "fc1", "fc2", "final_layer_norm", "finfo", "float16", "forward", "functional", "hidden_states", "if", "isinf", "isnan", "max", "min", "nn", "or", "output_attentions", "outputs", "p", "residual", "return", "self", "self_attn", "self_attn_layer_norm", "torch", "training", "tuple"], "bart/modeling_bart.py:BartDecoderLayer.__init__": ["ACT2FN", "GradientCheckpointingLayer", "LayerNorm", "Linear", "ModelAttention", "ModelConfig", "ModelDecoderLayer", "None", "True", "__init__", "activation_dropout", "activation_fn", "activation_function", "attention_dropout", "class", "config", "d_model", "decoder_attention_heads", "decoder_ffn_dim", "def", "dropout", "embed_dim", "encoder_attn", "encoder_attn_layer_norm", "fc1", "fc2", "final_layer_norm", "int", "is_causal", "is_decoder", "layer_idx", "nn", "num_heads", "self", "self_attn", "self_attn_layer_norm", "super"], "bart/modeling_bart.py:BartDecoderLayer.forward": ["Cache", "False", "FloatTensor", "GradientCheckpointingLayer", "ModelDecoderLayer", "None", "Tensor", "True", "activation_dropout", "activation_fn", "attention_mask", "bool", "cache_position", "class", "cross_attn_weights", "def", "dropout", "encoder_attention_mask", "encoder_attn", "encoder_attn_layer_norm", "encoder_hidden_states", "fc1", "fc2", "final_layer_norm", "forward", "functional", "hidden_states", "if", "is", "key_value_states", "nn", "not", "output_attentions", "outputs", "p", "past_key_values", "residual", "return", "self", "self_attn", "self_attn_layer_norm", "self_attn_weights", "torch", "training", "tuple", "use_cache"], "bart/modeling_bart.py:BartClassificationHead.__init__": ["Dropout", "Linear", "ModelClassificationHead", "Module", "__init__", "class", "def", "dense", "dropout", "float", "inner_dim", "input_dim", "int", "nn", "num_classes", "out_proj", "p", "pooler_dropout", "self", "super"], "bart/modeling_bart.py:BartClassificationHead.forward": ["ModelClassificationHead", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "nn", "out_proj", "return", "self", "tanh", "torch"], "bart/modeling_bart.py:BartPreTrainedModel._init_weights": ["ModelForConditionalGeneration", "ModelPreTrainedModel", "PreTrainedModel", "_init_weights", "class", "def", "final_logits_bias", "if", "init", "isinstance", "module", "self", "super", "zeros_"], "bart/modeling_bart.py:BartPreTrainedModel.dummy_inputs": ["ModelPreTrainedModel", "PreTrainedModel", "attention_mask", "class", "config", "def", "device", "dummy_inputs", "input_ids", "ne", "pad_token", "pad_token_id", "return", "self", "tensor", "torch"], "bart/modeling_bart.py:PretrainedBartModel.__init_subclass__": ["FutureWarning", "ModelModelModel", "The", "__init_subclass__", "been", "class", "def", "depreciated", "has", "instead", "please", "self", "use", "warn", "warnings"], "bart/modeling_bart.py:BartPretrainedModel.__init_subclass__": ["FutureWarning", "ModelPreTrainedModel", "ModelPretrainedModel", "PretrainedModelModel", "The", "__init_subclass__", "been", "class", "def", "depreciated", "has", "instead", "please", "self", "use", "warn", "warnings"], "bart/modeling_bart.py:BartEncoder.__init__": ["False", "LayerNorm", "ModelConfig", "ModelEncoder", "ModelEncoderLayer", "ModelLearnedPositionalEmbedding", "ModelPreTrainedModel", "ModelScaledWordEmbedding", "ModuleList", "__init__", "class", "config", "d_model", "def", "dropout", "else", "embed_dim", "embed_positions", "embed_scale", "embed_tokens", "encoder_layerdrop", "encoder_layers", "for", "gradient_checkpointing", "i", "if", "in", "layer_idx", "layerdrop", "layernorm_embedding", "layers", "math", "max_position_embeddings", "max_source_positions", "nn", "pad_token_id", "padding_idx", "post_init", "range", "scale_embedding", "self", "sqrt", "super", "vocab_size"], "bart/modeling_bart.py:BartEncoder.forward": ["BaseModelOutput", "False", "FloatTensor", "LongTensor", "ModelEncoder", "ModelPreTrainedModel", "None", "Tensor", "True", "ValueError", "You", "all_attentions", "and", "at", "attention_mask", "attentions", "bool", "both", "cannot", "class", "config", "create_bidirectional_mask", "def", "device", "dropout", "dropout_probability", "either", "elif", "else", "embed_pos", "embed_positions", "embed_tokens", "encoder_layer", "encoder_states", "enumerate", "for", "forward", "functional", "have", "hidden_states", "idx", "if", "in", "input", "input_embeds", "input_ids", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layer_outputs", "layerdrop", "layernorm_embedding", "layers", "nn", "not", "or", "output_attentions", "output_hidden_states", "p", "r", "raise", "rand", "return", "return_dict", "same", "self", "shape", "specify", "the", "time", "to", "to_drop", "torch", "training", "tuple", "use_return_dict", "v", "view"], "bart/modeling_bart.py:BartDecoder.__init__": ["False", "LayerNorm", "ModelConfig", "ModelDecoder", "ModelDecoderLayer", "ModelLearnedPositionalEmbedding", "ModelPreTrainedModel", "ModelScaledWordEmbedding", "ModuleList", "__init__", "class", "config", "d_model", "decoder_layerdrop", "decoder_layers", "def", "dropout", "else", "embed_positions", "embed_scale", "embed_tokens", "for", "gradient_checkpointing", "i", "if", "in", "layer_idx", "layerdrop", "layernorm_embedding", "layers", "math", "max_position_embeddings", "max_target_positions", "nn", "pad_token_id", "padding_idx", "post_init", "range", "scale_embedding", "self", "sqrt", "super", "vocab_size"], "bart/modeling_bart.py:BartDecoder.forward": ["BaseModelOutputWithPastAndCrossAttentions", "Cache", "DynamicCache", "EncoderDecoderCache", "False", "FloatTensor", "LongTensor", "ModelDecoder", "ModelPreTrainedModel", "None", "Setting", "Tensor", "True", "ValueError", "You", "all_cross_attentions", "all_hidden_states", "all_self_attns", "and", "arange", "at", "attention_mask", "attentions", "batch_size", "bool", "both", "cache_position", "cannot", "checkpointing", "class", "config", "continue", "create_bidirectional_mask", "create_causal_mask", "cross_attentions", "decoder_input_ids", "decoder_inputs_embeds", "decoder_layer", "def", "device", "dropout", "dropout_probability", "either", "elif", "else", "embed_positions", "embed_tokens", "encoder_attention_mask", "encoder_hidden_states", "enumerate", "for", "forward", "functional", "get_seq_length", "gradient", "gradient_checkpointing", "have", "hidden_states", "idx", "if", "in", "incompatible", "input", "input_embeds", "input_ids", "input_shape", "inputs_embeds", "is", "is_encoder_decoder", "is_torchdynamo_compiling", "isinstance", "kwargs", "last_hidden_state", "layer_outputs", "layerdrop", "layernorm_embedding", "layers", "logger", "mask_seq_length", "nn", "not", "ones", "or", "output_attentions", "output_hidden_states", "p", "past_key_values", "past_key_values_length", "position_ids", "positions", "r", "raise", "rand", "return", "return_dict", "same", "self", "self_attention_cache", "self_attn_cache", "seq_length", "shape", "size", "specify", "the", "time", "to", "torch", "training", "tuple", "use_cache", "use_return_dict", "v", "view", "warning_once", "with"], "bart/modeling_bart.py:BartModel.__init__": ["ModelConfig", "ModelDecoder", "ModelEncoder", "ModelModel", "ModelPreTrainedModel", "ModelScaledWordEmbedding", "__init__", "class", "config", "d_model", "decoder", "def", "else", "embed_scale", "encoder", "if", "math", "pad_token_id", "padding_idx", "post_init", "scale_embedding", "self", "shared", "sqrt", "super", "vocab_size"], "bart/modeling_bart.py:BartModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "return", "self", "shared"], "bart/modeling_bart.py:BartModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "decoder", "def", "embed_tokens", "encoder", "self", "set_input_embeddings", "shared", "value"], "bart/modeling_bart.py:BartModel.forward": ["BaseModelOutput", "Cache", "FloatTensor", "If", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Please", "Seq2SeqModelOutput", "Tensor", "ValueError", "and", "are", "attention_mask", "attentions", "be", "bool", "cache_position", "cannot", "class", "config", "cross_attentions", "decoder", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_input_ids", "decoder_inputs_embeds", "decoder_outputs", "decoder_start_token_id", "def", "either", "elif", "else", "encoder", "encoder_attention_mask", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "isinstance", "kwargs", "last_hidden_state", "len", "list", "no", "not", "or", "output_attentions", "output_hidden_states", "pad_token_id", "pass", "passed", "past_key_values", "r", "raise", "return", "return_dict", "self", "shift_tokens_right", "torch", "tuple", "use_cache", "use_return_dict"], "bart/modeling_bart.py:BartForConditionalGeneration.__init__": ["False", "GenerationMixin", "Linear", "ModelConfig", "ModelForConditionalGeneration", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "d_model", "def", "final_logits_bias", "lm_head", "model", "nn", "num_embeddings", "post_init", "register_buffer", "self", "shared", "super", "torch", "zeros"], "bart/modeling_bart.py:BartForConditionalGeneration.resize_token_embeddings": ["Embedding", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "True", "_resize_final_logits_bias", "bool", "class", "def", "int", "mean_resizing", "new_embeddings", "new_num_tokens", "nn", "pad_to_multiple_of", "resize_token_embeddings", "return", "self", "shape", "super", "weight"], "bart/modeling_bart.py:BartForConditionalGeneration._resize_final_logits_bias": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "_resize_final_logits_bias", "cat", "class", "def", "device", "dim", "else", "extra_bias", "final_logits_bias", "if", "int", "new_bias", "new_num_tokens", "old_num_tokens", "register_buffer", "self", "shape", "torch", "zeros"], "bart/modeling_bart.py:BartForConditionalGeneration.forward": ["Cache", "CrossEntropyLoss", "False", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Seq2SeqLMOutput", "Tensor", "The", "and", "argument", "attention_mask", "bool", "cache_position", "changed", "class", "config", "cross_attentions", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_input_ids", "decoder_inputs_embeds", "decoder_start_token_id", "def", "device", "else", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "final_logits_bias", "forward", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "list", "lm_head", "lm_logits", "logger", "logits", "loss", "loss_fct", "masked_lm_loss", "model", "not", "output", "output_attentions", "output_hidden_states", "outputs", "pad_token_id", "past_key_values", "provided", "r", "return", "return_dict", "self", "shift_tokens_right", "since", "to", "torch", "tuple", "use_cache", "use_return_dict", "view", "vocab_size", "warning"], "bart/modeling_bart.py:BartForConditionalGeneration.prepare_decoder_input_ids_from_labels": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "Tensor", "class", "config", "decoder_start_token_id", "def", "labels", "pad_token_id", "prepare_decoder_input_ids_from_labels", "return", "self", "shift_tokens_right", "torch"], "bart/modeling_bart.py:BartForSequenceClassification.__init__": ["ModelClassificationHead", "ModelConfig", "ModelForSequenceClassification", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "classification_head", "classifier_dropout", "config", "d_model", "def", "kwargs", "model", "num_labels", "post_init", "self", "super"], "bart/modeling_bart.py:BartForSequenceClassification.forward": ["All", "BCEWithLogitsLoss", "CrossEntropyLoss", "False", "FloatTensor", "LongTensor", "MSELoss", "ModelForSequenceClassification", "ModelPreTrainedModel", "None", "NotImplementedError", "Passing", "Seq2SeqSequenceClassifierOutput", "Tensor", "ValueError", "__class__", "__name__", "and", "attention_mask", "bool", "cache_position", "class", "classification_head", "config", "cross_attentions", "currently", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_input_ids", "decoder_inputs_embeds", "def", "device", "dtype", "elif", "else", "embeddings", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "eos", "eos_mask", "eos_token_id", "eq", "examples", "f", "for", "forward", "have", "hidden_states", "if", "input", "input_ids", "inputs_embeds", "int", "is", "kwargs", "labels", "len", "list", "logits", "long", "loss", "loss_fct", "model", "multi_label_classification", "must", "not", "num_labels", "number", "of", "or", "output", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "problem_type", "r", "raise", "regression", "return", "return_dict", "same", "self", "sentence_representation", "single_label_classification", "size", "squeeze", "sum", "supported", "the", "to", "tokens", "torch", "tuple", "unique_consecutive", "use_cache", "use_return_dict", "view"], "bart/modeling_bart.py:BartForQuestionAnswering.__init__": ["Linear", "ModelForQuestionAnswering", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "config", "def", "hidden_size", "model", "nn", "num_labels", "post_init", "qa_outputs", "self", "super"], "bart/modeling_bart.py:BartForQuestionAnswering.forward": ["CrossEntropyLoss", "False", "FloatTensor", "LongTensor", "ModelForQuestionAnswering", "ModelPreTrainedModel", "None", "Seq2SeqQuestionAnsweringModelOutput", "Tensor", "and", "attention_mask", "bool", "cache_position", "clamp", "class", "config", "contiguous", "cross_attentions", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_input_ids", "decoder_inputs_embeds", "def", "dim", "else", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "end_logits", "end_loss", "end_positions", "forward", "if", "ignore_index", "ignored_index", "input_ids", "inputs_embeds", "is", "kwargs", "len", "list", "logits", "loss", "loss_fct", "model", "not", "output", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "qa_outputs", "r", "return", "return_dict", "self", "sequence_output", "size", "split", "squeeze", "start_logits", "start_loss", "start_positions", "torch", "total_loss", "tuple", "use_cache", "use_return_dict"], "bart/modeling_bart.py:BartDecoderWrapper.__init__": ["ModelDecoder", "ModelDecoderWrapper", "ModelPreTrainedModel", "__init__", "class", "config", "decoder", "def", "post_init", "self", "super"], "bart/modeling_bart.py:BartDecoderWrapper.forward": ["ModelDecoderWrapper", "ModelPreTrainedModel", "args", "class", "decoder", "def", "forward", "kwargs", "return", "self"], "bart/modeling_bart.py:BartForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "ModelDecoderWrapper", "ModelForCausalLM", "ModelPreTrainedModel", "True", "__init__", "bias", "class", "config", "def", "hidden_size", "is_decoder", "is_encoder_decoder", "lm_head", "model", "nn", "post_init", "self", "super", "vocab_size"], "bart/modeling_bart.py:BartForCausalLM.get_input_embeddings": ["GenerationMixin", "ModelForCausalLM", "ModelPreTrainedModel", "class", "decoder", "def", "embed_tokens", "get_input_embeddings", "model", "return", "self"], "bart/modeling_bart.py:BartForCausalLM.set_input_embeddings": ["GenerationMixin", "ModelForCausalLM", "ModelPreTrainedModel", "class", "decoder", "def", "embed_tokens", "model", "self", "set_input_embeddings", "value"], "bart/modeling_bart.py:BartForCausalLM.forward": ["Cache", "CausalLMOutputWithCrossAttentions", "CrossEntropyLoss", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForCausalLM", "ModelPreTrainedModel", "None", "Tensor", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "cross_attentions", "decoder", "def", "device", "else", "encoder_attention_mask", "encoder_hidden_states", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "lm_head", "logits", "logits_to_keep", "loss", "loss_fct", "model", "not", "output", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "r", "return", "return_dict", "self", "slice", "slice_indices", "to", "torch", "tuple", "use_cache", "use_return_dict", "view", "vocab_size"], "beit/modeling_beit.py:drop_path": ["False", "Model_path", "Model_prob", "Tensor", "bool", "def", "device", "div", "dtype", "float", "floor_", "if", "input", "keep_prob", "ndim", "not", "or", "output", "rand", "random_tensor", "return", "shape", "torch", "training"], "beit/modeling_beit.py:BeitDropPath.__init__": ["ModelDropPath", "Module", "None", "__init__", "class", "def", "drop_prob", "float", "nn", "self", "super"], "beit/modeling_beit.py:BeitDropPath.forward": ["ModelDropPath", "Module", "Tensor", "class", "def", "drop_path", "drop_prob", "forward", "hidden_states", "nn", "return", "self", "torch", "training"], "beit/modeling_beit.py:BeitDropPath.extra_repr": ["ModelDropPath", "Module", "class", "def", "drop_prob", "extra_repr", "f", "nn", "p", "return", "self", "str"], "beit/modeling_beit.py:BeitEmbeddings.__init__": ["Dropout", "Iterable", "ModelConfig", "ModelEmbeddings", "ModelPatchEmbeddings", "Module", "None", "Parameter", "__init__", "abc", "class", "cls_token", "collections", "config", "def", "dropout", "else", "hidden_dropout_prob", "hidden_size", "if", "image_size", "isinstance", "mask_token", "nn", "num_patches", "patch_embeddings", "patch_size", "position_embeddings", "self", "super", "torch", "use_absolute_position_embeddings", "use_mask_token", "zeros"], "beit/modeling_beit.py:BeitEmbeddings.interpolate_pos_encoding": ["False", "ModelEmbeddings", "Module", "Tensor", "align_corners", "and", "bicubic", "cat", "class", "class_pos_embed", "def", "dim", "embeddings", "functional", "height", "if", "int", "interpolate", "interpolate_pos_encoding", "is_tracing", "jit", "mode", "new_height", "new_width", "nn", "not", "num_patches", "num_positions", "patch_pos_embed", "patch_size", "permute", "position_embeddings", "reshape", "return", "self", "shape", "size", "sqrt_num_positions", "torch", "torch_int", "view", "width"], "beit/modeling_beit.py:BeitEmbeddings.forward": ["BoolTensor", "ModelEmbeddings", "Module", "None", "Tensor", "_", "batch_size", "bool_masked_pos", "cat", "class", "cls_token", "cls_tokens", "def", "dim", "dropout", "embeddings", "expand", "forward", "height", "if", "interpolate_pos_encoding", "is", "mask_token", "mask_tokens", "nn", "not", "patch_embeddings", "patch_height", "patch_width", "pixel_values", "position_embeddings", "return", "self", "seq_len", "shape", "size", "torch", "type_as", "unsqueeze", "w", "width"], "beit/modeling_beit.py:BeitPatchEmbeddings.__init__": ["Conv2d", "Iterable", "ModelPatchEmbeddings", "Module", "__init__", "abc", "class", "collections", "config", "def", "else", "hidden_size", "if", "image_size", "isinstance", "kernel_size", "nn", "num_channels", "num_patches", "patch_shape", "patch_size", "projection", "self", "stride", "super"], "beit/modeling_beit.py:BeitPatchEmbeddings.forward": ["Make", "ModelPatchEmbeddings", "Module", "Tensor", "ValueError", "batch_size", "channel", "class", "configuration", "def", "dimension", "dtype", "embeddings", "flatten", "forward", "height", "if", "in", "match", "nn", "num_channels", "of", "one", "patch_height", "patch_width", "pixel", "pixel_values", "projection", "raise", "return", "self", "set", "shape", "sure", "that", "the", "to", "torch", "transpose", "values", "weight", "width", "with"], "beit/modeling_beit.py:BeitSelfAttention.__init__": ["Dropout", "False", "Linear", "ModelConfig", "ModelRelativePositionBias", "ModelSelfAttention", "Module", "None", "The", "ValueError", "__init__", "a", "all_head_size", "and", "attention", "attention_head_size", "attention_probs_dropout_prob", "bias", "bool", "class", "config", "def", "dropout", "embedding_size", "f", "has_relative_position_bias", "hasattr", "heads", "hidden", "hidden_size", "if", "int", "is", "key", "multiple", "nn", "not", "num_attention_heads", "number", "of", "query", "raise", "relative_position_bias", "self", "size", "super", "the", "tuple", "value", "window_size"], "beit/modeling_beit.py:BeitSelfAttention.forward": ["False", "ModelSelfAttention", "Module", "None", "Tensor", "_", "all_head_size", "attention_head_size", "attention_probs", "attention_scores", "batch_size", "bool", "class", "config", "context_layer", "contiguous", "def", "dim", "dim_size", "dropout", "else", "forward", "functional", "has_relative_position_bias", "height", "hidden_states", "if", "int", "interpolate_pos_encoding", "is", "key", "key_layer", "math", "matmul", "new_context_layer_shape", "nn", "not", "num_attention_heads", "output_attentions", "outputs", "patch_size", "permute", "query", "query_layer", "relative_position_bias", "resolution", "return", "self", "seq_length", "shape", "size", "softmax", "sqrt", "torch", "transpose", "tuple", "value", "value_layer", "view", "width", "window_size"], "beit/modeling_beit.py:BeitSdpaSelfAttention.forward": ["False", "If", "ModelSdpaSelfAttention", "ModelSelfAttention", "None", "Tensor", "The", "True", "_", "__class__", "__name__", "all_head_size", "attention", "attention_head_size", "attention_probs_dropout_prob", "attn_bias", "attn_implementation", "attn_mask", "batch_size", "be", "bool", "class", "config", "context_layer", "contiguous", "def", "dim_size", "does", "dropout_p", "eager", "else", "f", "forward", "functional", "get", "has_relative_position_bias", "height", "hidden_states", "if", "int", "interpolate_pos_encoding", "is", "is_causal", "key", "key_layer", "loading", "logger", "math", "model", "new_context_layer_shape", "nn", "not", "num_attention_heads", "output_attentions", "patch_size", "permute", "please", "query", "query_layer", "relative_position_bias", "resolution", "return", "returned", "scale", "scaled_dot_product_attention", "scaling", "self", "seq_length", "set", "shape", "size", "sqrt", "support", "the", "to", "torch", "training", "transpose", "tuple", "value", "value_layer", "view", "want", "warning_once", "weights", "when", "width", "will", "window_size", "you"], "beit/modeling_beit.py:BeitSelfOutput.__init__": ["Dropout", "Linear", "ModelConfig", "ModelSelfOutput", "Module", "None", "__init__", "class", "config", "def", "dense", "dropout", "hidden_dropout_prob", "hidden_size", "nn", "self", "super"], "beit/modeling_beit.py:BeitSelfOutput.forward": ["ModelSelfOutput", "Module", "None", "Tensor", "class", "def", "dense", "dropout", "forward", "gamma", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "beit/modeling_beit.py:BeitAttention.__init__": ["ModelAttention", "ModelConfig", "ModelSelfOutput", "Model_SELF_ATTENTION_CLASSES", "Module", "None", "__init__", "_attn_implementation", "attention", "class", "config", "def", "nn", "output", "self", "super", "tuple", "window_size"], "beit/modeling_beit.py:BeitAttention.forward": ["False", "ModelAttention", "Module", "None", "Tensor", "attention", "attention_output", "bool", "class", "def", "forward", "hidden_states", "int", "interpolate_pos_encoding", "nn", "output", "output_attentions", "outputs", "relative_position_bias", "resolution", "return", "self", "self_outputs", "torch", "tuple"], "beit/modeling_beit.py:BeitIntermediate.__init__": ["ACT2FN", "Linear", "ModelConfig", "ModelIntermediate", "Module", "None", "__init__", "class", "config", "def", "dense", "else", "hidden_act", "hidden_size", "if", "intermediate_act_fn", "intermediate_size", "isinstance", "nn", "self", "str", "super"], "beit/modeling_beit.py:BeitIntermediate.forward": ["ModelIntermediate", "Module", "Tensor", "class", "def", "dense", "forward", "hidden_states", "intermediate_act_fn", "nn", "return", "self", "torch"], "beit/modeling_beit.py:BeitOutput.__init__": ["Dropout", "Linear", "ModelConfig", "ModelOutput", "Module", "None", "__init__", "class", "config", "def", "dense", "dropout", "hidden_dropout_prob", "hidden_size", "intermediate_size", "nn", "self", "super"], "beit/modeling_beit.py:BeitOutput.forward": ["ModelOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "nn", "return", "self", "torch"], "beit/modeling_beit.py:BeitLayer.__init__": ["GradientCheckpointingLayer", "Identity", "LayerNorm", "ModelAttention", "ModelConfig", "ModelDropPath", "ModelIntermediate", "ModelLayer", "ModelOutput", "None", "Parameter", "True", "__init__", "attention", "chunk_size_feed_forward", "class", "config", "def", "drop_path", "drop_path_rate", "else", "eps", "float", "hidden_size", "if", "init_values", "intermediate", "lambda_1", "lambda_2", "layer_norm_eps", "layer_scale_init_value", "layernorm_after", "layernorm_before", "nn", "ones", "output", "requires_grad", "self", "seq_len_dim", "super", "torch", "tuple", "window_size"], "beit/modeling_beit.py:BeitLayer.forward": ["False", "GradientCheckpointingLayer", "ModelLayer", "None", "Tensor", "attention", "attention_output", "bool", "class", "def", "drop_path", "forward", "hidden_states", "if", "int", "intermediate", "interpolate_pos_encoding", "is", "lambda_1", "lambda_2", "layer_output", "layernorm_after", "layernorm_before", "not", "output", "output_attentions", "outputs", "relative_position_bias", "resolution", "return", "self", "self_attention_outputs", "torch", "tuple"], "beit/modeling_beit.py:BeitRelativePositionBias.__init__": ["ModelConfig", "ModelRelativePositionBias", "Module", "None", "Parameter", "__init__", "class", "config", "def", "nn", "num_attention_heads", "num_relative_distance", "relative_position_bias_table", "self", "super", "torch", "tuple", "window_size", "zeros"], "beit/modeling_beit.py:BeitRelativePositionBias.generate_relative_position_index": ["ModelRelativePositionBias", "Module", "None", "Tensor", "arange", "class", "contiguous", "coords", "coords_flatten", "def", "dtype", "flatten", "generate_relative_position_index", "grid", "ij", "indexing", "int", "meshgrid", "nn", "num_relative_distance", "permute", "relative_coords", "relative_position_index", "return", "self", "size", "stack", "sum", "torch", "tuple", "window_area", "window_size", "zeros"], "beit/modeling_beit.py:BeitRelativePositionBias.forward": ["False", "ModelRelativePositionBias", "Module", "None", "Tensor", "align_corners", "bilinear", "bool", "cat", "class", "contiguous", "def", "dim_size", "forward", "functional", "generate_relative_position_index", "if", "interpolate", "interpolate_pos_encoding", "mode", "new_height", "new_num_relative_distance", "new_relative_position_bias_table", "new_sub_table", "new_width", "nn", "num_relative_distance", "old_height", "old_num_relative_distance", "old_relative_position_bias_table", "old_sub_table", "old_width", "permute", "relative_position_bias", "relative_position_bias_table", "relative_position_index", "reshape", "return", "self", "size", "squeeze", "torch", "torch_int", "unsqueeze", "view", "window_size"], "beit/modeling_beit.py:BeitEncoder.__init__": ["False", "ModelConfig", "ModelEncoder", "ModelLayer", "ModelRelativePositionBias", "Module", "ModuleList", "None", "__init__", "class", "config", "cpu", "def", "device", "dpr", "drop_path_rate", "else", "for", "gradient_checkpointing", "has_relative_position_bias", "i", "if", "in", "item", "layer", "linspace", "nn", "num_hidden_layers", "range", "relative_position_bias", "self", "super", "torch", "tuple", "use_relative_position_bias", "use_shared_relative_position_bias", "window_size", "x"], "beit/modeling_beit.py:BeitEncoder.forward": ["BaseModelOutput", "False", "ModelEncoder", "Module", "None", "Tensor", "True", "all_hidden_states", "all_self_attentions", "attentions", "bool", "class", "config", "def", "dim_size", "else", "enumerate", "for", "forward", "has_relative_position_bias", "height", "hidden_states", "i", "if", "in", "int", "interpolate_pos_encoding", "is", "last_hidden_state", "layer", "layer_module", "layer_outputs", "nn", "not", "output_attentions", "output_hidden_states", "patch_size", "relative_position_bias", "resolution", "return", "return_dict", "self", "shape", "torch", "tuple", "v", "width", "window_size"], "beit/modeling_beit.py:BeitPreTrainedModel._init_weights": ["ModelEmbeddings", "ModelLayer", "ModelPreTrainedModel", "ModelRelativePositionBias", "None", "PreTrainedModel", "_init_weights", "class", "cls_token", "config", "constant_", "def", "elif", "if", "init", "is", "isinstance", "lambda_1", "lambda_2", "layer_scale_init_value", "mask_token", "module", "not", "position_embeddings", "relative_position_bias_table", "self", "super", "zeros_"], "beit/modeling_beit.py:BeitModel.__init__": ["Identity", "LayerNorm", "ModelConfig", "ModelEmbeddings", "ModelEncoder", "ModelModel", "ModelPooler", "ModelPreTrainedModel", "None", "True", "__init__", "add_pooling_layer", "bool", "class", "config", "def", "else", "embeddings", "encoder", "eps", "hidden_size", "if", "layer_norm_eps", "layernorm", "nn", "patch_embeddings", "patch_shape", "pooler", "post_init", "r", "self", "super", "use_mean_pooling", "window_size"], "beit/modeling_beit.py:BeitModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "patch_embeddings", "return", "self"], "beit/modeling_beit.py:BeitModel.forward": ["BoolTensor", "False", "ModelModel", "ModelModelOutputWithPooling", "ModelPreTrainedModel", "None", "Tensor", "_", "attentions", "bool", "bool_masked_pos", "class", "config", "def", "else", "embedding_output", "embeddings", "encoder", "encoder_outputs", "forward", "head_outputs", "hidden_states", "if", "interpolate_pos_encoding", "is", "kwargs", "last_hidden_state", "layernorm", "not", "output_attentions", "output_hidden_states", "pixel_values", "pooled_output", "pooler", "pooler_output", "r", "resolution", "return", "return_dict", "self", "sequence_output", "shape", "torch", "tuple", "use_return_dict"], "beit/modeling_beit.py:BeitPooler.__init__": ["LayerNorm", "ModelConfig", "ModelPooler", "Module", "None", "__init__", "class", "config", "def", "else", "eps", "hidden_size", "if", "layer_norm_eps", "layernorm", "nn", "self", "super", "use_mean_pooling"], "beit/modeling_beit.py:BeitPooler.forward": ["ModelPooler", "Module", "None", "Tensor", "class", "def", "else", "forward", "hidden_states", "if", "is", "layernorm", "mean", "nn", "not", "patch_tokens", "pooled_output", "return", "self", "torch"], "beit/modeling_beit.py:BeitForMaskedImageModeling.__init__": ["False", "LayerNorm", "Linear", "Model", "ModelConfig", "ModelForMaskedImageModeling", "ModelModel", "ModelPreTrainedModel", "None", "__init__", "add_pooling_layer", "class", "config", "def", "eps", "hidden_size", "layer_norm_eps", "layernorm", "lm_head", "nn", "num_labels", "post_init", "self", "super", "vocab_size"], "beit/modeling_beit.py:BeitForMaskedImageModeling.get_output_embeddings": ["ModelForMaskedImageModeling", "ModelPreTrainedModel", "None", "class", "def", "get_output_embeddings", "return", "self"], "beit/modeling_beit.py:BeitForMaskedImageModeling.forward": ["BoolTensor", "CrossEntropyLoss", "False", "MaskedLMOutput", "Model", "ModelForMaskedImageModeling", "ModelPreTrainedModel", "None", "Tensor", "attentions", "bool", "bool_masked_pos", "class", "config", "def", "else", "forward", "hidden_states", "if", "interpolate_pos_encoding", "is", "kwargs", "labels", "layernorm", "lm_head", "logits", "loss", "loss_fct", "masked_lm_loss", "not", "output", "output_attentions", "output_hidden_states", "outputs", "pixel_values", "prediction_scores", "r", "return", "return_dict", "self", "sequence_output", "torch", "tuple", "use_return_dict"], "beit/modeling_beit.py:BeitForImageClassification.__init__": ["Identity", "Linear", "Model", "ModelConfig", "ModelForImageClassification", "ModelModel", "ModelPreTrainedModel", "None", "True", "__init__", "add_pooling_layer", "class", "classifier", "config", "def", "else", "hidden_size", "if", "nn", "num_labels", "post_init", "self", "super"], "beit/modeling_beit.py:BeitForImageClassification.forward": ["False", "ImageClassifierOutput", "Model", "ModelForImageClassification", "ModelPreTrainedModel", "None", "Tensor", "attentions", "bool", "class", "classifier", "config", "def", "else", "forward", "hidden_states", "if", "interpolate_pos_encoding", "is", "kwargs", "labels", "logits", "loss", "loss_function", "not", "output", "output_attentions", "output_hidden_states", "outputs", "pixel_values", "pooled_output", "pooler_output", "r", "return", "return_dict", "self", "torch", "tuple", "use_return_dict"], "beit/modeling_beit.py:BeitConvModule.__init__": ["BatchNorm2d", "Conv2d", "False", "ModelConvModule", "Module", "None", "ReLU", "__init__", "activation", "bias", "bn", "bool", "class", "conv", "def", "dilation", "in_channels", "int", "kernel_size", "nn", "out_channels", "padding", "self", "str", "super", "tuple"], "beit/modeling_beit.py:BeitConvModule.forward": ["ModelConvModule", "Module", "Tensor", "activation", "bn", "class", "conv", "def", "forward", "input", "nn", "output", "return", "self", "torch"], "beit/modeling_beit.py:BeitPyramidPoolingBlock.__init__": ["AdaptiveAvgPool2d", "ModelConvModule", "ModelPyramidPoolingBlock", "Module", "None", "__init__", "add_module", "channels", "class", "def", "enumerate", "for", "i", "in", "in_channels", "int", "kernel_size", "layer", "layers", "nn", "pool_scale", "self", "str", "super"], "beit/modeling_beit.py:BeitPyramidPoolingBlock.forward": ["ModelPyramidPoolingBlock", "Module", "Tensor", "class", "def", "for", "forward", "hidden_state", "in", "input", "layer", "layers", "nn", "return", "self", "torch"], "beit/modeling_beit.py:BeitPyramidPoolingModule.__init__": ["ModelPyramidPoolingBlock", "ModelPyramidPoolingModule", "Module", "None", "__init__", "add_module", "align_corners", "append", "block", "blocks", "bool", "channels", "class", "def", "enumerate", "for", "i", "in", "in_channels", "int", "nn", "pool_scale", "pool_scales", "self", "str", "super", "tuple"], "beit/modeling_beit.py:BeitPyramidPoolingModule.forward": ["ModelPyramidPoolingModule", "Module", "Tensor", "align_corners", "append", "bilinear", "blocks", "class", "def", "for", "forward", "functional", "in", "interpolate", "list", "mode", "nn", "ppm", "ppm_out", "ppm_outs", "return", "self", "size", "torch", "upsampled_ppm_out", "x"], "beit/modeling_beit.py:BeitUperHead.__init__": ["Conv2d", "False", "ModelConfig", "ModelConvModule", "ModelPyramidPoolingModule", "ModelUperHead", "Module", "ModuleList", "None", "__init__", "align_corners", "append", "bottleneck", "channels", "class", "classifier", "config", "def", "for", "fpn_bottleneck", "fpn_conv", "fpn_convs", "hidden_size", "in", "in_channels", "kernel_size", "l_conv", "lateral_convs", "len", "nn", "num_labels", "padding", "pool_scales", "psp_modules", "self", "super"], "beit/modeling_beit.py:BeitUperHead.psp_forward": ["ModelUperHead", "Module", "bottleneck", "cat", "class", "def", "dim", "extend", "inputs", "nn", "output", "psp_forward", "psp_modules", "psp_outs", "return", "self", "torch", "x"], "beit/modeling_beit.py:BeitUperHead.forward": ["ModelUperHead", "Module", "Tensor", "align_corners", "append", "bilinear", "cat", "class", "classifier", "def", "dim", "encoder_hidden_states", "enumerate", "for", "forward", "fpn_bottleneck", "fpn_convs", "fpn_outs", "functional", "i", "in", "interpolate", "lateral_conv", "lateral_convs", "laterals", "len", "mode", "nn", "output", "prev_shape", "psp_forward", "range", "return", "self", "shape", "size", "torch", "used_backbone_levels"], "beit/modeling_beit.py:BeitFCNHead.__init__": ["Conv2d", "Identity", "ModelConfig", "ModelConvModule", "ModelFCNHead", "Module", "None", "Sequential", "__init__", "append", "auxiliary_channels", "auxiliary_concat_input", "auxiliary_num_convs", "channels", "class", "classifier", "concat_input", "config", "conv_cat", "conv_padding", "convs", "def", "dilation", "else", "for", "hidden_size", "i", "if", "in", "in_channels", "in_index", "int", "kernel_size", "nn", "num_convs", "num_labels", "padding", "range", "self", "super", "tuple"], "beit/modeling_beit.py:BeitFCNHead.forward": ["ModelFCNHead", "Module", "Tensor", "cat", "class", "classifier", "concat_input", "conv_cat", "convs", "def", "dim", "encoder_hidden_states", "forward", "hidden_states", "if", "in_index", "nn", "output", "return", "self", "torch"], "beit/modeling_beit.py:BeitForSemanticSegmentation.__init__": ["BatchNorm2d", "ConvTranspose2d", "False", "GELU", "Identity", "MaxPool2d", "Model", "ModelConfig", "ModelFCNHead", "ModelForSemanticSegmentation", "ModelModel", "ModelPreTrainedModel", "ModelUperHead", "None", "One", "Sequential", "ValueError", "__init__", "a", "add_pooling_layer", "architecture", "auxiliary_head", "backbone", "base", "be", "can", "case", "class", "config", "decode_head", "def", "else", "features", "fpn1", "fpn2", "fpn3", "fpn4", "from", "hidden_size", "if", "in", "integers", "kernel_size", "len", "list", "nn", "num_labels", "of", "out_indices", "post_init", "raise", "requires", "self", "sized", "specifying", "stride", "super", "the", "to", "use", "use_auxiliary_head", "which"], "beit/modeling_beit.py:BeitForSemanticSegmentation.compute_loss": ["CrossEntropyLoss", "False", "ModelForSemanticSegmentation", "ModelPreTrainedModel", "None", "align_corners", "auxiliary_logits", "auxiliary_loss", "auxiliary_loss_weight", "bilinear", "class", "compute_loss", "config", "def", "functional", "if", "ignore_index", "interpolate", "is", "labels", "logits", "loss", "loss_fct", "main_loss", "mode", "nn", "not", "return", "self", "semantic_loss_ignore_index", "shape", "size", "upsampled_auxiliary_logits", "upsampled_logits"], "beit/modeling_beit.py:BeitForSemanticSegmentation.forward": ["False", "Model", "ModelForSemanticSegmentation", "ModelPreTrainedModel", "None", "SemanticSegmenterOutput", "Tensor", "The", "True", "ValueError", "and", "attentions", "auxiliary_head", "auxiliary_logits", "batch_size", "be", "bool", "class", "compute_loss", "config", "decode_head", "def", "else", "encoder_hidden_states", "enumerate", "feature", "features", "for", "forward", "fpn1", "fpn2", "fpn3", "fpn4", "greater", "hidden_states", "i", "idx", "if", "image_size", "in", "interpolate_pos_encoding", "is", "kwargs", "labels", "len", "logits", "loss", "not", "num_labels", "number", "of", "one", "ops", "out_indices", "output", "output_attentions", "output_hidden_states", "outputs", "patch_resolution", "patch_size", "permute", "pixel_values", "r", "raise", "range", "reshape", "return", "return_dict", "self", "shape", "should", "than", "torch", "tuple", "use_return_dict", "x"], "beit/modeling_beit.py:BeitBackbone.__init__": ["BackboneMixin", "BatchNorm2d", "ConvTranspose2d", "GELU", "Identity", "MaxPool2d", "ModelBackbone", "ModelEmbeddings", "ModelEncoder", "ModelPreTrainedModel", "One", "Sequential", "ValueError", "_", "__init__", "_init_backbone", "a", "add_fpn", "architecture", "backbone", "base", "batch_norm_eps", "be", "can", "case", "class", "config", "def", "embeddings", "encoder", "eps", "features", "for", "fpn1", "fpn2", "fpn3", "fpn4", "from", "hidden_size", "if", "in", "integers", "kernel_size", "len", "list", "nn", "num_features", "num_hidden_layers", "of", "out_indices", "patch_embeddings", "patch_shape", "post_init", "raise", "range", "requires", "self", "sized", "specifying", "stride", "super", "the", "to", "use", "which", "window_size"], "beit/modeling_beit.py:BeitBackbone.get_input_embeddings": ["BackboneMixin", "ModelBackbone", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "patch_embeddings", "return", "self"], "beit/modeling_beit.py:BeitBackbone.forward": ["BackboneMixin", "BackboneOutput", "ModelBackbone", "ModelPreTrainedModel", "None", "Tensor", "True", "add_fpn", "attentions", "batch_size", "bool", "class", "config", "def", "else", "embedding_output", "embeddings", "encoder", "feature_maps", "for", "forward", "fpn1", "fpn2", "fpn3", "fpn4", "hidden_state", "hidden_states", "if", "in", "is", "kwargs", "not", "out_features", "output", "output_attentions", "output_hidden_states", "outputs", "patch_height", "patch_width", "permute", "pixel_values", "r", "reshape", "reshape_hidden_states", "resolution", "return", "return_dict", "self", "shape", "stage", "stage_names", "tuple", "use_return_dict", "zip"], "bert/modeling_bert.py:BertEmbeddings.__init__": ["Dropout", "Embedding", "False", "LayerNorm", "ModelEmbeddings", "Module", "__init__", "arange", "class", "config", "def", "dropout", "dtype", "eps", "expand", "hidden_dropout_prob", "hidden_size", "layer_norm_eps", "long", "max_position_embeddings", "nn", "pad_token_id", "padding_idx", "persistent", "position_embeddings", "position_ids", "register_buffer", "self", "size", "super", "token_type_embeddings", "token_type_ids", "torch", "type_vocab_size", "vocab_size", "word_embeddings", "zeros"], "bert/modeling_bert.py:BertEmbeddings.forward": ["FloatTensor", "LayerNorm", "LongTensor", "ModelEmbeddings", "Module", "None", "Tensor", "batch_size", "buffered_token_type_ids", "class", "def", "device", "dim", "dropout", "dtype", "else", "embeddings", "expand", "forward", "gather", "hasattr", "if", "index", "input_ids", "input_shape", "inputs_embeds", "int", "is", "long", "nn", "not", "past_key_values_length", "position_embeddings", "position_ids", "return", "self", "seq_length", "shape", "size", "token_type_embeddings", "token_type_ids", "torch", "word_embeddings", "zeros"], "bert/modeling_bert.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "contiguous", "def", "dim", "dropout", "float", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "shape", "size", "softmax", "torch", "training", "transpose", "value"], "bert/modeling_bert.py:BertSelfAttention.__init__": ["Dropout", "False", "Linear", "ModelSelfAttention", "Module", "None", "The", "ValueError", "__init__", "a", "all_head_size", "and", "attention", "attention_head_size", "attention_probs_dropout_prob", "class", "config", "def", "dropout", "embedding_size", "f", "hasattr", "heads", "hidden", "hidden_size", "if", "int", "is", "is_causal", "is_decoder", "key", "layer_idx", "multiple", "nn", "not", "num_attention_heads", "number", "of", "query", "raise", "scaling", "self", "size", "super", "the", "value"], "bert/modeling_bert.py:BertSelfAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "EncoderDecoderCache", "FloatTensor", "ModelSelfAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_attn_implementation", "attention_head_size", "attention_interface", "attention_mask", "attn_output", "attn_weights", "cache_position", "class", "config", "contiguous", "current_past_key_values", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "hidden_shape", "hidden_states", "if", "input_shape", "is", "isinstance", "key", "key_layer", "kwargs", "layer_idx", "nn", "not", "p", "past_key_values", "query", "query_layer", "reshape", "return", "scaling", "self", "self_attention_cache", "shape", "torch", "training", "transpose", "tuple", "update", "value", "value_layer", "view"], "bert/modeling_bert.py:BertCrossAttention.__init__": ["Dropout", "False", "Linear", "ModelCrossAttention", "Module", "None", "The", "ValueError", "__init__", "a", "all_head_size", "and", "attention", "attention_head_size", "attention_probs_dropout_prob", "class", "config", "def", "dropout", "embedding_size", "f", "hasattr", "heads", "hidden", "hidden_size", "if", "int", "is", "is_causal", "key", "layer_idx", "multiple", "nn", "not", "num_attention_heads", "number", "of", "query", "raise", "scaling", "self", "size", "super", "the", "value"], "bert/modeling_bert.py:BertCrossAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "EncoderDecoderCache", "False", "FloatTensor", "ModelCrossAttention", "Module", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "_attn_implementation", "and", "attention_head_size", "attention_interface", "attention_mask", "attn_output", "attn_weights", "bsz", "class", "config", "contiguous", "cross_attention_cache", "def", "dropout", "eager", "eager_attention_forward", "else", "encoder_hidden_states", "forward", "get", "hidden_states", "if", "is", "is_updated", "key", "key_layer", "keys", "kv_input_shape", "kwargs", "layer_idx", "layers", "nn", "not", "p", "past_key_values", "q_input_shape", "query", "query_layer", "reshape", "return", "scaling", "self", "shape", "src_len", "tgt_len", "torch", "training", "transpose", "tuple", "update", "value", "value_layer", "values", "view"], "bert/modeling_bert.py:BertSelfOutput.__init__": ["Dropout", "LayerNorm", "Linear", "ModelSelfOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "eps", "hidden_dropout_prob", "hidden_size", "layer_norm_eps", "nn", "self", "super"], "bert/modeling_bert.py:BertSelfOutput.forward": ["LayerNorm", "ModelSelfOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "bert/modeling_bert.py:BertAttention.__init__": ["False", "ModelAttention", "ModelCrossAttention", "ModelSelfAttention", "ModelSelfOutput", "Module", "None", "__init__", "attention_class", "class", "config", "def", "else", "if", "is_causal", "is_cross_attention", "layer_idx", "nn", "output", "self", "super"], "bert/modeling_bert.py:BertAttention.forward": ["Cache", "FloatTensor", "ModelAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attention_output", "attn_weights", "cache_position", "class", "def", "else", "encoder_attention_mask", "encoder_hidden_states", "forward", "hidden_states", "if", "is_cross_attention", "kwargs", "nn", "not", "output", "past_key_values", "return", "self", "torch", "tuple"], "bert/modeling_bert.py:BertIntermediate.__init__": ["ACT2FN", "Linear", "ModelIntermediate", "Module", "__init__", "class", "config", "def", "dense", "else", "hidden_act", "hidden_size", "if", "intermediate_act_fn", "intermediate_size", "isinstance", "nn", "self", "str", "super"], "bert/modeling_bert.py:BertIntermediate.forward": ["ModelIntermediate", "Module", "Tensor", "class", "def", "dense", "forward", "hidden_states", "intermediate_act_fn", "nn", "return", "self", "torch"], "bert/modeling_bert.py:BertOutput.__init__": ["Dropout", "LayerNorm", "Linear", "ModelOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "eps", "hidden_dropout_prob", "hidden_size", "intermediate_size", "layer_norm_eps", "nn", "self", "super"], "bert/modeling_bert.py:BertOutput.forward": ["LayerNorm", "ModelOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "bert/modeling_bert.py:BertLayer.__init__": ["False", "GradientCheckpointingLayer", "ModelAttention", "ModelIntermediate", "ModelLayer", "ModelOutput", "None", "True", "ValueError", "__init__", "a", "add_cross_attention", "added", "as", "attention", "be", "chunk_size_feed_forward", "class", "config", "cross", "crossattention", "decoder", "def", "f", "if", "intermediate", "is", "is_causal", "is_cross_attention", "is_decoder", "layer_idx", "model", "not", "output", "raise", "self", "seq_len_dim", "should", "super", "used"], "bert/modeling_bert.py:BertLayer.forward": ["Cache", "FloatTensor", "GradientCheckpointingLayer", "If", "ModelLayer", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "ValueError", "_", "add_cross_attention", "and", "apply_chunking_to_forward", "are", "attention", "attention_mask", "attention_output", "be", "by", "cache_position", "chunk_size_feed_forward", "class", "config", "cross", "cross_attention_output", "crossattention", "def", "encoder_attention_mask", "encoder_hidden_states", "f", "feed_forward_chunk", "forward", "has", "hasattr", "hidden_states", "if", "instantiated", "is", "is_decoder", "kwargs", "layer_output", "layers", "not", "passed", "past_key_values", "raise", "return", "self", "self_attention_output", "seq_len_dim", "setting", "to", "torch", "tuple", "with"], "bert/modeling_bert.py:BertLayer.feed_forward_chunk": ["GradientCheckpointingLayer", "ModelLayer", "attention_output", "class", "def", "feed_forward_chunk", "intermediate", "intermediate_output", "layer_output", "output", "return", "self"], "bert/modeling_bert.py:BertEncoder.__init__": ["ModelEncoder", "ModelLayer", "Module", "ModuleList", "__init__", "class", "config", "def", "for", "i", "in", "layer", "layer_idx", "nn", "num_hidden_layers", "range", "self", "super"], "bert/modeling_bert.py:BertEncoder.forward": ["BaseModelOutputWithPastAndCrossAttentions", "Cache", "FloatTensor", "ModelEncoder", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "bool", "cache_position", "class", "def", "else", "encoder_attention_mask", "encoder_hidden_states", "enumerate", "for", "forward", "hidden_states", "i", "if", "in", "kwargs", "last_hidden_state", "layer", "layer_module", "nn", "past_key_values", "return", "self", "torch", "tuple", "use_cache"], "bert/modeling_bert.py:BertPooler.__init__": ["Linear", "ModelPooler", "Module", "Tanh", "__init__", "activation", "class", "config", "def", "dense", "hidden_size", "nn", "self", "super"], "bert/modeling_bert.py:BertPooler.forward": ["ModelPooler", "Module", "Tensor", "activation", "class", "def", "dense", "first_token_tensor", "forward", "hidden_states", "nn", "pooled_output", "return", "self", "torch"], "bert/modeling_bert.py:BertPredictionHeadTransform.__init__": ["ACT2FN", "LayerNorm", "Linear", "ModelPredictionHeadTransform", "Module", "__init__", "class", "config", "def", "dense", "else", "eps", "hidden_act", "hidden_size", "if", "isinstance", "layer_norm_eps", "nn", "self", "str", "super", "transform_act_fn"], "bert/modeling_bert.py:BertPredictionHeadTransform.forward": ["LayerNorm", "ModelPredictionHeadTransform", "Module", "Tensor", "class", "def", "dense", "forward", "hidden_states", "nn", "return", "self", "torch", "transform_act_fn"], "bert/modeling_bert.py:BertLMPredictionHead.__init__": ["Linear", "ModelLMPredictionHead", "ModelPredictionHeadTransform", "Module", "Parameter", "True", "__init__", "bias", "class", "config", "decoder", "def", "hidden_size", "nn", "self", "super", "torch", "transform", "vocab_size", "zeros"], "bert/modeling_bert.py:BertLMPredictionHead.forward": ["ModelLMPredictionHead", "Module", "class", "decoder", "def", "forward", "hidden_states", "nn", "return", "self", "transform"], "bert/modeling_bert.py:BertOnlyMLMHead.__init__": ["ModelLMPredictionHead", "ModelOnlyMLMHead", "Module", "__init__", "class", "config", "def", "nn", "predictions", "self", "super"], "bert/modeling_bert.py:BertOnlyMLMHead.forward": ["ModelOnlyMLMHead", "Module", "Tensor", "class", "def", "forward", "nn", "prediction_scores", "predictions", "return", "self", "sequence_output", "torch"], "bert/modeling_bert.py:BertOnlyNSPHead.__init__": ["Linear", "ModelOnlyNSPHead", "Module", "__init__", "class", "config", "def", "hidden_size", "nn", "self", "seq_relationship", "super"], "bert/modeling_bert.py:BertOnlyNSPHead.forward": ["ModelOnlyNSPHead", "Module", "class", "def", "forward", "nn", "pooled_output", "return", "self", "seq_relationship", "seq_relationship_score"], "bert/modeling_bert.py:BertPreTrainingHeads.__init__": ["Linear", "ModelLMPredictionHead", "ModelPreTrainingHeads", "Module", "__init__", "class", "config", "def", "hidden_size", "nn", "predictions", "self", "seq_relationship", "super"], "bert/modeling_bert.py:BertPreTrainingHeads.forward": ["ModelPreTrainingHeads", "Module", "class", "def", "forward", "nn", "pooled_output", "prediction_scores", "predictions", "return", "self", "seq_relationship", "seq_relationship_score", "sequence_output"], "bert/modeling_bert.py:BertPreTrainedModel._init_weights": ["ModelEmbeddings", "ModelLMPredictionHead", "ModelPreTrainedModel", "PreTrainedModel", "_init_weights", "arange", "bias", "class", "copy_", "def", "elif", "expand", "if", "init", "isinstance", "module", "position_ids", "self", "shape", "super", "token_type_ids", "torch", "zeros_"], "bert/modeling_bert.py:BertModel.__init__": ["False", "ModelEmbeddings", "ModelEncoder", "ModelModel", "ModelPooler", "ModelPreTrainedModel", "None", "True", "__init__", "add_pooling_layer", "class", "config", "def", "else", "embeddings", "encoder", "gradient_checkpointing", "if", "pooler", "post_init", "r", "self", "super"], "bert/modeling_bert.py:BertModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "return", "self", "word_embeddings"], "bert/modeling_bert.py:BertModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "self", "set_input_embeddings", "value", "word_embeddings"], "bert/modeling_bert.py:BertModel.forward": ["BaseModelOutputWithPoolingAndCrossAttentions", "Cache", "DynamicCache", "EncoderDecoderCache", "False", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "_create_attention_masks", "and", "arange", "attention_mask", "bool", "cache_position", "class", "config", "def", "device", "else", "embedding_output", "embeddings", "encoder", "encoder_attention_mask", "encoder_hidden_states", "encoder_outputs", "exactly", "forward", "get_seq_length", "if", "input_ids", "inputs_embeds", "is", "is_decoder", "is_encoder_decoder", "kwargs", "last_hidden_state", "must", "not", "of", "one", "or", "past_key_values", "past_key_values_length", "pooled_output", "pooler", "pooler_output", "position_ids", "raise", "return", "self", "seq_length", "sequence_output", "shape", "specify", "token_type_ids", "torch", "tuple", "use_cache"], "bert/modeling_bert.py:BertModel._create_attention_masks": ["ModelModel", "ModelPreTrainedModel", "None", "_create_attention_masks", "attention_mask", "cache_position", "class", "config", "create_bidirectional_mask", "create_causal_mask", "def", "else", "embedding_output", "encoder_attention_mask", "encoder_hidden_states", "if", "input_embeds", "is", "is_decoder", "not", "past_key_values", "return", "self"], "bert/modeling_bert.py:BertForPreTraining.__init__": ["Model", "ModelForPreTraining", "ModelModel", "ModelPreTrainedModel", "ModelPreTrainingHeads", "__init__", "class", "cls", "config", "def", "post_init", "self", "super"], "bert/modeling_bert.py:BertForPreTraining.get_output_embeddings": ["ModelForPreTraining", "ModelPreTrainedModel", "class", "cls", "decoder", "def", "get_output_embeddings", "predictions", "return", "self"], "bert/modeling_bert.py:BertForPreTraining.set_output_embeddings": ["ModelForPreTraining", "ModelPreTrainedModel", "bias", "class", "cls", "decoder", "def", "new_embeddings", "predictions", "self", "set_output_embeddings"], "bert/modeling_bert.py:BertForPreTraining.forward": ["CrossEntropyLoss", "Model", "ModelForPreTraining", "ModelForPreTrainingOutput", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "and", "attention_mask", "attentions", "class", "cls", "config", "def", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "loss", "loss_fct", "masked_lm_loss", "next_sentence_label", "next_sentence_loss", "not", "outputs", "pooled_output", "position_ids", "prediction_logits", "prediction_scores", "r", "return", "return_dict", "self", "seq_relationship_logits", "seq_relationship_score", "sequence_output", "token_type_ids", "torch", "total_loss", "tuple", "view", "vocab_size"], "bert/modeling_bert.py:BertLMHeadModel.__init__": ["False", "GenerationMixin", "If", "Model", "ModelLMHeadModel", "ModelModel", "ModelOnlyMLMHead", "ModelPreTrainedModel", "True", "__init__", "a", "add", "add_pooling_layer", "as", "class", "cls", "config", "def", "if", "is_decoder", "logger", "not", "post_init", "self", "standalone", "super", "to", "use", "want", "warning", "you"], "bert/modeling_bert.py:BertLMHeadModel.get_output_embeddings": ["GenerationMixin", "ModelLMHeadModel", "ModelPreTrainedModel", "class", "cls", "decoder", "def", "get_output_embeddings", "predictions", "return", "self"], "bert/modeling_bert.py:BertLMHeadModel.set_output_embeddings": ["GenerationMixin", "ModelLMHeadModel", "ModelPreTrainedModel", "bias", "class", "cls", "decoder", "def", "new_embeddings", "predictions", "self", "set_output_embeddings"], "bert/modeling_bert.py:BertLMHeadModel.forward": ["BaseModelOutputWithPoolingAndCrossAttentions", "Cache", "CausalLMOutputWithCrossAttentions", "False", "GenerationMixin", "Model", "ModelLMHeadModel", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "attention_mask", "attentions", "bool", "cache_position", "class", "cls", "config", "cross_attentions", "def", "else", "encoder_attention_mask", "encoder_hidden_states", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "logits", "logits_to_keep", "loss", "loss_function", "not", "outputs", "past_key_values", "position_ids", "r", "return", "return_dict", "self", "slice", "slice_indices", "token_type_ids", "torch", "tuple", "use_cache", "vocab_size"], "bert/modeling_bert.py:BertForMaskedLM.__init__": ["False", "If", "Model", "ModelForMaskedLM", "ModelModel", "ModelOnlyMLMHead", "ModelPreTrainedModel", "__init__", "add_pooling_layer", "attention", "bi", "class", "cls", "config", "def", "directional", "for", "if", "is_decoder", "logger", "make", "post_init", "self", "super", "sure", "to", "use", "want", "warning", "you"], "bert/modeling_bert.py:BertForMaskedLM.get_output_embeddings": ["ModelForMaskedLM", "ModelPreTrainedModel", "class", "cls", "decoder", "def", "get_output_embeddings", "predictions", "return", "self"], "bert/modeling_bert.py:BertForMaskedLM.set_output_embeddings": ["ModelForMaskedLM", "ModelPreTrainedModel", "bias", "class", "cls", "decoder", "def", "new_embeddings", "predictions", "self", "set_output_embeddings"], "bert/modeling_bert.py:BertForMaskedLM.forward": ["CrossEntropyLoss", "MaskedLMOutput", "Model", "ModelForMaskedLM", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "attention_mask", "attentions", "class", "cls", "config", "def", "encoder_attention_mask", "encoder_hidden_states", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "masked_lm_loss", "not", "outputs", "position_ids", "prediction_scores", "r", "return", "return_dict", "self", "sequence_output", "token_type_ids", "torch", "tuple", "view", "vocab_size"], "bert/modeling_bert.py:BertForMaskedLM.prepare_inputs_for_generation": ["ModelForMaskedLM", "ModelPreTrainedModel", "None", "PAD", "The", "ValueError", "attention_mask", "be", "cat", "class", "config", "def", "defined", "device", "dim", "dtype", "dummy_token", "effective_batch_size", "for", "full", "generation", "if", "input_ids", "input_shape", "is", "long", "model_kwargs", "new_zeros", "pad_token_id", "prepare_inputs_for_generation", "raise", "return", "self", "shape", "should", "token", "torch"], "bert/modeling_bert.py:BertForMaskedLM.can_generate": ["False", "ModelForMaskedLM", "ModelPreTrainedModel", "bool", "can_generate", "class", "cls", "def", "return"], "bert/modeling_bert.py:BertForNextSentencePrediction.__init__": ["Model", "ModelForNextSentencePrediction", "ModelModel", "ModelOnlyNSPHead", "ModelPreTrainedModel", "__init__", "class", "cls", "config", "def", "post_init", "self", "super"], "bert/modeling_bert.py:BertForNextSentencePrediction.forward": ["CrossEntropyLoss", "Model", "ModelForNextSentencePrediction", "ModelPreTrainedModel", "NextSentencePredictorOutput", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "attention_mask", "attentions", "class", "cls", "def", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "next_sentence_loss", "not", "outputs", "pooled_output", "position_ids", "r", "return", "return_dict", "self", "seq_relationship_scores", "token_type_ids", "torch", "tuple", "view"], "bert/modeling_bert.py:BertForSequenceClassification.__init__": ["Dropout", "Linear", "Model", "ModelForSequenceClassification", "ModelModel", "ModelPreTrainedModel", "None", "__init__", "class", "classifier", "classifier_dropout", "config", "def", "dropout", "else", "hidden_dropout_prob", "hidden_size", "if", "is", "nn", "not", "num_labels", "post_init", "self", "super"], "bert/modeling_bert.py:BertForSequenceClassification.forward": ["BCEWithLogitsLoss", "CrossEntropyLoss", "MSELoss", "Model", "ModelForSequenceClassification", "ModelPreTrainedModel", "None", "SequenceClassifierOutput", "Tensor", "TransformersKwargs", "True", "Unpack", "and", "attention_mask", "attentions", "class", "classifier", "config", "def", "dropout", "dtype", "elif", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "kwargs", "labels", "logits", "long", "loss", "loss_fct", "multi_label_classification", "not", "num_labels", "or", "outputs", "pooled_output", "position_ids", "problem_type", "r", "regression", "return", "return_dict", "self", "single_label_classification", "squeeze", "token_type_ids", "torch", "tuple", "view"], "bert/modeling_bert.py:BertForMultipleChoice.__init__": ["Dropout", "Linear", "Model", "ModelForMultipleChoice", "ModelModel", "ModelPreTrainedModel", "None", "__init__", "class", "classifier", "classifier_dropout", "config", "def", "dropout", "else", "hidden_dropout_prob", "hidden_size", "if", "is", "nn", "not", "post_init", "self", "super"], "bert/modeling_bert.py:BertForMultipleChoice.forward": ["CrossEntropyLoss", "Model", "ModelForMultipleChoice", "ModelPreTrainedModel", "MultipleChoiceModelOutput", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "attention_mask", "attentions", "class", "classifier", "def", "dropout", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "not", "num_choices", "outputs", "pooled_output", "position_ids", "r", "reshaped_logits", "return", "return_dict", "self", "shape", "size", "token_type_ids", "torch", "tuple", "view"], "bert/modeling_bert.py:BertForTokenClassification.__init__": ["Dropout", "False", "Linear", "Model", "ModelForTokenClassification", "ModelModel", "ModelPreTrainedModel", "None", "__init__", "add_pooling_layer", "class", "classifier", "classifier_dropout", "config", "def", "dropout", "else", "hidden_dropout_prob", "hidden_size", "if", "is", "nn", "not", "num_labels", "post_init", "self", "super"], "bert/modeling_bert.py:BertForTokenClassification.forward": ["CrossEntropyLoss", "Model", "ModelForTokenClassification", "ModelPreTrainedModel", "None", "Tensor", "TokenClassifierOutput", "TransformersKwargs", "True", "Unpack", "attention_mask", "attentions", "class", "classifier", "def", "dropout", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "not", "num_labels", "outputs", "position_ids", "r", "return", "return_dict", "self", "sequence_output", "token_type_ids", "torch", "tuple", "view"], "bert/modeling_bert.py:BertForQuestionAnswering.__init__": ["False", "Linear", "Model", "ModelForQuestionAnswering", "ModelModel", "ModelPreTrainedModel", "__init__", "add_pooling_layer", "class", "config", "def", "hidden_size", "nn", "num_labels", "post_init", "qa_outputs", "self", "super"], "bert/modeling_bert.py:BertForQuestionAnswering.forward": ["CrossEntropyLoss", "Model", "ModelForQuestionAnswering", "ModelPreTrainedModel", "None", "QuestionAnsweringModelOutput", "Tensor", "TransformersKwargs", "True", "Unpack", "and", "attention_mask", "attentions", "clamp", "class", "contiguous", "def", "dim", "end_logits", "end_loss", "end_positions", "forward", "hidden_states", "if", "ignore_index", "ignored_index", "input_ids", "inputs_embeds", "is", "kwargs", "len", "logits", "loss", "loss_fct", "not", "outputs", "position_ids", "qa_outputs", "return", "return_dict", "self", "sequence_output", "size", "split", "squeeze", "start_logits", "start_loss", "start_positions", "token_type_ids", "torch", "total_loss", "tuple"], "bert_generation/modeling_bert_generation.py:BertGenerationSelfOutput.__init__": ["Dropout", "LayerNorm", "Linear", "ModelSelfOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "eps", "hidden_dropout_prob", "hidden_size", "layer_norm_eps", "nn", "self", "super"], "bert_generation/modeling_bert_generation.py:BertGenerationSelfOutput.forward": ["LayerNorm", "ModelSelfOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "bert_generation/modeling_bert_generation.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "contiguous", "def", "dim", "dropout", "float", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "shape", "size", "softmax", "torch", "training", "transpose", "value"], "bert_generation/modeling_bert_generation.py:BertGenerationSelfAttention.__init__": ["Dropout", "False", "Linear", "ModelSelfAttention", "Module", "None", "The", "ValueError", "__init__", "a", "all_head_size", "and", "attention", "attention_head_size", "attention_probs_dropout_prob", "class", "config", "def", "dropout", "embedding_size", "f", "hasattr", "heads", "hidden", "hidden_size", "if", "int", "is", "is_causal", "is_decoder", "key", "layer_idx", "multiple", "nn", "not", "num_attention_heads", "number", "of", "query", "raise", "scaling", "self", "size", "super", "the", "value"], "bert_generation/modeling_bert_generation.py:BertGenerationSelfAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "EncoderDecoderCache", "FloatTensor", "ModelSelfAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_attn_implementation", "attention_head_size", "attention_interface", "attention_mask", "attn_output", "attn_weights", "cache_position", "class", "config", "contiguous", "current_past_key_values", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "hidden_shape", "hidden_states", "if", "input_shape", "is", "isinstance", "key", "key_layer", "kwargs", "layer_idx", "nn", "not", "p", "past_key_values", "query", "query_layer", "reshape", "return", "scaling", "self", "self_attention_cache", "shape", "torch", "training", "transpose", "tuple", "update", "value", "value_layer", "view"], "bert_generation/modeling_bert_generation.py:BertGenerationCrossAttention.__init__": ["Dropout", "False", "Linear", "ModelCrossAttention", "Module", "None", "The", "ValueError", "__init__", "a", "all_head_size", "and", "attention", "attention_head_size", "attention_probs_dropout_prob", "class", "config", "def", "dropout", "embedding_size", "f", "hasattr", "heads", "hidden", "hidden_size", "if", "int", "is", "is_causal", "key", "layer_idx", "multiple", "nn", "not", "num_attention_heads", "number", "of", "query", "raise", "scaling", "self", "size", "super", "the", "value"], "bert_generation/modeling_bert_generation.py:BertGenerationCrossAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "EncoderDecoderCache", "False", "FloatTensor", "ModelCrossAttention", "Module", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "_attn_implementation", "and", "attention_head_size", "attention_interface", "attention_mask", "attn_output", "attn_weights", "bsz", "class", "config", "contiguous", "cross_attention_cache", "def", "dropout", "eager", "eager_attention_forward", "else", "encoder_hidden_states", "forward", "get", "hidden_states", "if", "is", "is_updated", "key", "key_layer", "keys", "kv_input_shape", "kwargs", "layer_idx", "layers", "nn", "not", "p", "past_key_values", "q_input_shape", "query", "query_layer", "reshape", "return", "scaling", "self", "shape", "src_len", "tgt_len", "torch", "training", "transpose", "tuple", "update", "value", "value_layer", "values", "view"], "bert_generation/modeling_bert_generation.py:BertGenerationAttention.__init__": ["False", "ModelAttention", "ModelCrossAttention", "ModelSelfAttention", "ModelSelfOutput", "Module", "None", "__init__", "attention_class", "class", "config", "def", "else", "if", "is_causal", "is_cross_attention", "layer_idx", "nn", "output", "self", "super"], "bert_generation/modeling_bert_generation.py:BertGenerationAttention.forward": ["Cache", "FloatTensor", "ModelAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attention_output", "attn_weights", "cache_position", "class", "def", "else", "encoder_attention_mask", "encoder_hidden_states", "forward", "hidden_states", "if", "is_cross_attention", "kwargs", "nn", "not", "output", "past_key_values", "return", "self", "torch", "tuple"], "bert_generation/modeling_bert_generation.py:BertGenerationIntermediate.__init__": ["ACT2FN", "Linear", "ModelIntermediate", "Module", "__init__", "class", "config", "def", "dense", "else", "hidden_act", "hidden_size", "if", "intermediate_act_fn", "intermediate_size", "isinstance", "nn", "self", "str", "super"], "bert_generation/modeling_bert_generation.py:BertGenerationIntermediate.forward": ["ModelIntermediate", "Module", "Tensor", "class", "def", "dense", "forward", "hidden_states", "intermediate_act_fn", "nn", "return", "self", "torch"], "bert_generation/modeling_bert_generation.py:BertGenerationOutput.__init__": ["Dropout", "LayerNorm", "Linear", "ModelOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "eps", "hidden_dropout_prob", "hidden_size", "intermediate_size", "layer_norm_eps", "nn", "self", "super"], "bert_generation/modeling_bert_generation.py:BertGenerationOutput.forward": ["LayerNorm", "ModelOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "bert_generation/modeling_bert_generation.py:BertGenerationLayer.__init__": ["False", "GradientCheckpointingLayer", "ModelAttention", "ModelIntermediate", "ModelLayer", "ModelOutput", "None", "True", "ValueError", "__init__", "a", "add_cross_attention", "added", "as", "attention", "be", "chunk_size_feed_forward", "class", "config", "cross", "crossattention", "decoder", "def", "f", "if", "intermediate", "is", "is_causal", "is_cross_attention", "is_decoder", "layer_idx", "model", "not", "output", "raise", "self", "seq_len_dim", "should", "super", "used"], "bert_generation/modeling_bert_generation.py:BertGenerationLayer.forward": ["Cache", "FloatTensor", "GradientCheckpointingLayer", "If", "ModelLayer", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "ValueError", "_", "add_cross_attention", "and", "apply_chunking_to_forward", "are", "attention", "attention_mask", "attention_output", "be", "by", "cache_position", "chunk_size_feed_forward", "class", "config", "cross", "cross_attention_output", "crossattention", "def", "encoder_attention_mask", "encoder_hidden_states", "f", "feed_forward_chunk", "forward", "has", "hasattr", "hidden_states", "if", "instantiated", "is", "is_decoder", "kwargs", "layer_output", "layers", "not", "passed", "past_key_values", "raise", "return", "self", "self_attention_output", "seq_len_dim", "setting", "to", "torch", "tuple", "with"], "bert_generation/modeling_bert_generation.py:BertGenerationLayer.feed_forward_chunk": ["GradientCheckpointingLayer", "ModelLayer", "attention_output", "class", "def", "feed_forward_chunk", "intermediate", "intermediate_output", "layer_output", "output", "return", "self"], "bert_generation/modeling_bert_generation.py:BertEncoder.__init__": ["ModelEncoder", "ModelLayer", "Module", "ModuleList", "__init__", "class", "config", "def", "for", "i", "in", "layer", "layer_idx", "nn", "num_hidden_layers", "range", "self", "super"], "bert_generation/modeling_bert_generation.py:BertEncoder.forward": ["BaseModelOutputWithPastAndCrossAttentions", "Cache", "FloatTensor", "ModelEncoder", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "bool", "cache_position", "class", "def", "else", "encoder_attention_mask", "encoder_hidden_states", "enumerate", "for", "forward", "hidden_states", "i", "if", "in", "kwargs", "last_hidden_state", "layer", "layer_module", "nn", "past_key_values", "return", "self", "torch", "tuple", "use_cache"], "bert_generation/modeling_bert_generation.py:BertGenerationEmbeddings.__init__": ["Dropout", "Embedding", "False", "LayerNorm", "ModelEmbeddings", "Module", "__init__", "arange", "class", "config", "def", "dropout", "eps", "expand", "hidden_dropout_prob", "hidden_size", "layer_norm_eps", "max_position_embeddings", "nn", "pad_token_id", "padding_idx", "persistent", "position_embeddings", "position_ids", "register_buffer", "self", "super", "torch", "vocab_size", "word_embeddings"], "bert_generation/modeling_bert_generation.py:BertGenerationEmbeddings.forward": ["LayerNorm", "ModelEmbeddings", "Module", "None", "class", "def", "dropout", "else", "embeddings", "forward", "if", "input_ids", "input_shape", "inputs_embeds", "is", "nn", "not", "past_key_values_length", "position_embeddings", "position_ids", "return", "self", "seq_length", "size", "word_embeddings"], "bert_generation/modeling_bert_generation.py:BertGenerationPreTrainedModel._init_weights": ["ModelEmbeddings", "ModelOnlyLMHead", "ModelPreTrainedModel", "PreTrainedModel", "_init_weights", "arange", "bias", "class", "copy_", "def", "elif", "expand", "if", "init", "isinstance", "module", "position_ids", "self", "shape", "super", "torch", "zeros_"], "bert_generation/modeling_bert_generation.py:BertGenerationEncoder.__init__": ["False", "ModelEmbeddings", "ModelEncoder", "ModelPreTrainedModel", "__init__", "class", "config", "def", "embeddings", "encoder", "gradient_checkpointing", "post_init", "self", "super"], "bert_generation/modeling_bert_generation.py:BertGenerationEncoder.get_input_embeddings": ["ModelEncoder", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "return", "self", "word_embeddings"], "bert_generation/modeling_bert_generation.py:BertGenerationEncoder.set_input_embeddings": ["ModelEncoder", "ModelPreTrainedModel", "class", "def", "embeddings", "self", "set_input_embeddings", "value", "word_embeddings"], "bert_generation/modeling_bert_generation.py:BertGenerationEncoder.forward": ["BaseModelOutputWithPastAndCrossAttentions", "Cache", "DynamicCache", "EncoderDecoderCache", "False", "ModelEncoder", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "_create_attention_masks", "and", "arange", "attention_mask", "bool", "cache_position", "class", "config", "def", "device", "else", "embedding_output", "embeddings", "encoder", "encoder_attention_mask", "encoder_hidden_states", "encoder_outputs", "exactly", "forward", "get_seq_length", "if", "input_ids", "input_shape", "inputs_embeds", "is", "is_decoder", "is_encoder_decoder", "kwargs", "last_hidden_state", "must", "not", "of", "one", "or", "past_key_values", "past_key_values_length", "position_ids", "raise", "return", "self", "seq_length", "sequence_output", "shape", "specify", "torch", "tuple", "use_cache"], "bert_generation/modeling_bert_generation.py:BertGenerationEncoder._create_attention_masks": ["ModelEncoder", "ModelPreTrainedModel", "None", "_create_attention_masks", "attention_mask", "cache_position", "class", "config", "create_bidirectional_mask", "create_causal_mask", "def", "else", "embedding_output", "encoder_attention_mask", "encoder_hidden_states", "if", "input_embeds", "is", "is_decoder", "not", "past_key_values", "return", "self"], "bert_generation/modeling_bert_generation.py:BertGenerationOnlyLMHead.__init__": ["Linear", "ModelOnlyLMHead", "Module", "Parameter", "__init__", "bias", "class", "config", "decoder", "def", "hidden_size", "nn", "self", "super", "torch", "vocab_size", "zeros"], "bert_generation/modeling_bert_generation.py:BertGenerationOnlyLMHead.forward": ["ModelOnlyLMHead", "Module", "class", "decoder", "def", "forward", "hidden_states", "logits", "nn", "return", "self"], "bert_generation/modeling_bert_generation.py:BertGenerationDecoder.__init__": ["GenerationMixin", "If", "Model", "ModelDecoder", "ModelEncoder", "ModelOnlyLMHead", "ModelPreTrainedModel", "True", "__init__", "a", "add", "as", "class", "config", "def", "if", "is_decoder", "lm_head", "logger", "not", "post_init", "self", "standalone", "super", "to", "use", "want", "warning", "you"], "bert_generation/modeling_bert_generation.py:BertGenerationDecoder.get_output_embeddings": ["GenerationMixin", "ModelDecoder", "ModelPreTrainedModel", "class", "decoder", "def", "get_output_embeddings", "lm_head", "return", "self"], "bert_generation/modeling_bert_generation.py:BertGenerationDecoder.set_output_embeddings": ["GenerationMixin", "ModelDecoder", "ModelPreTrainedModel", "bias", "class", "decoder", "def", "lm_head", "new_embeddings", "self", "set_output_embeddings"], "bert_generation/modeling_bert_generation.py:BertGenerationDecoder.forward": ["BaseModelOutputWithPastAndCrossAttentions", "CausalLMOutputWithCrossAttentions", "False", "FloatTensor", "GenerationMixin", "Model", "ModelDecoder", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "cross_attentions", "def", "else", "encoder_attention_mask", "encoder_hidden_states", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "not", "outputs", "past_key_values", "position_ids", "r", "return", "return_dict", "self", "slice", "slice_indices", "torch", "tuple", "use_cache", "vocab_size"], "big_bird/modeling_big_bird.py:BigBirdEmbeddings.__init__": ["Dropout", "Embedding", "False", "LayerNorm", "ModelEmbeddings", "Module", "__init__", "arange", "class", "config", "def", "dropout", "dtype", "eps", "expand", "hidden_dropout_prob", "hidden_size", "layer_norm_eps", "long", "max_position_embeddings", "nn", "pad_token_id", "padding_idx", "persistent", "position_embeddings", "position_ids", "register_buffer", "rescale_embeddings", "self", "size", "super", "token_type_embeddings", "token_type_ids", "torch", "type_vocab_size", "vocab_size", "word_embeddings", "zeros"], "big_bird/modeling_big_bird.py:BigBirdEmbeddings.forward": ["LayerNorm", "ModelEmbeddings", "Module", "None", "buffered_token_type_ids", "buffered_token_type_ids_expanded", "class", "def", "device", "dropout", "dtype", "else", "embeddings", "expand", "forward", "hasattr", "hidden_size", "if", "input_ids", "input_shape", "inputs_embeds", "is", "long", "nn", "not", "past_key_values_length", "position_embeddings", "position_ids", "rescale_embeddings", "return", "self", "seq_length", "size", "token_type_embeddings", "token_type_ids", "torch", "word_embeddings", "zeros"], "big_bird/modeling_big_bird.py:BigBirdSelfAttention.__init__": ["Dropout", "Linear", "ModelSelfAttention", "Module", "None", "The", "ValueError", "__init__", "a", "all_head_size", "and", "attention", "attention_head_size", "attention_probs_dropout_prob", "bias", "class", "config", "def", "dropout", "embedding_size", "f", "hasattr", "heads", "hidden", "hidden_size", "if", "int", "is", "is_decoder", "key", "layer_idx", "multiple", "nn", "not", "num_attention_heads", "number", "of", "query", "raise", "self", "size", "super", "the", "use_bias", "value"], "big_bird/modeling_big_bird.py:BigBirdSelfAttention.forward": ["False", "ModelSelfAttention", "Module", "None", "_", "all_head_size", "and", "attention_head_size", "attention_mask", "attention_probs", "attention_scores", "batch_size", "cache_position", "class", "context_layer", "contiguous", "current_states", "def", "dim", "dropout", "else", "encoder_attention_mask", "encoder_hidden_states", "forward", "functional", "get_seq_length", "hidden_states", "if", "is", "is_cross_attention", "key", "key_layer", "keys", "layer_idx", "layers", "math", "matmul", "new_context_layer_shape", "nn", "not", "num_attention_heads", "output_attentions", "past_key_values", "permute", "query", "query_layer", "return", "self", "seq_length", "shape", "size", "softmax", "sqrt", "torch", "transpose", "update", "value", "value_layer", "values", "view"], "big_bird/modeling_big_bird.py:BigBirdBlockSparseAttention.__init__": ["Linear", "ModelBlockSparseAttention", "Module", "None", "The", "ValueError", "__init__", "a", "all_head_size", "attention", "attention_head_size", "bias", "block_size", "class", "config", "def", "f", "heads", "hidden", "hidden_size", "if", "int", "is", "key", "max_position_embeddings", "max_seqlen", "multiple", "nn", "not", "num_attention_heads", "num_random_blocks", "number", "of", "query", "raise", "seed", "self", "size", "super", "the", "use_bias", "value"], "big_bird/modeling_big_bird.py:BigBirdBlockSparseAttention.forward": ["Key", "ModelBlockSparseAttention", "Model_block_sparse_attention", "Module", "None", "Query", "Value", "ValueError", "_", "attention_head_size", "attention_probs", "band_mask", "batch_size", "be", "block", "block_size", "class", "context_layer", "contiguous", "def", "forward", "from_block_size", "from_blocked_mask", "from_mask", "from_seq_length", "hidden_states", "if", "key", "key_layer", "length", "multiple", "must", "nn", "num_attention_heads", "num_random_blocks", "of", "output_attentions", "plan_from_length", "plan_num_rand_blocks", "query", "query_layer", "raise", "return", "seed", "self", "seqlen", "sequence", "sided", "size", "to_block_size", "to_blocked_mask", "to_mask", "to_seq_length", "transpose", "value", "value_layer", "view"], "big_bird/modeling_big_bird.py:BigBirdBlockSparseAttention.torch_bmm_nd": ["ModelBlockSparseAttention", "Module", "None", "bmm", "class", "def", "inp_1", "inp_2", "ndim", "nn", "reshape", "return", "shape", "torch", "torch_bmm_nd", "view"], "big_bird/modeling_big_bird.py:BigBirdBlockSparseAttention.torch_bmm_nd_transpose": ["ModelBlockSparseAttention", "Module", "None", "bmm", "class", "def", "inp_1", "inp_2", "ndim", "nn", "reshape", "return", "shape", "torch", "torch_bmm_nd_transpose", "transpose", "view"], "big_bird/modeling_big_bird.py:BigBirdBlockSparseAttention.bigbird_block_sparse_attention": ["Error", "ModelBlockSparseAttention", "Model_block_sparse_attention", "Module", "None", "ValueError", "_", "_Model_block_rand_mask", "_Model_block_rand_mask_with_head", "_create_rand_mask_from_inputs", "_get_rand_attn_plan", "attention_head_size", "attention_probs", "attn_mask_penalty", "attn_probs_view", "attn_weights", "axis", "band_mask", "band_product", "batch_size", "be", "bhkd", "bhlqd", "bhlqk", "blocked_key_matrix", "blocked_query_matrix", "blocked_value_matrix", "blocks", "bsz", "cat", "class", "context_layer", "def", "device", "dim", "dtype", "einsum", "else", "exp_blocked_key_matrix", "exp_blocked_value_matrix", "first_attn_weights", "first_band_product", "first_context_layer", "first_product", "float", "for", "from_block_size", "from_blocked_mask", "from_mask", "from_seq_len", "from_seq_length", "functional", "gathered_key", "gathered_value", "i1", "i2", "if", "in", "inner_band_product", "is", "key_layer", "last_attn_weights", "last_band_product", "last_context_layer", "last_idx", "last_product", "len", "long", "math", "max_seqlen", "middle_query_matrix", "minimum", "n_heads", "n_rand_blocks", "ndim", "needs", "new_ones", "nn", "np", "num_heads", "number", "of", "output_attentions", "p1", "p2", "plan_from_length", "plan_num_rand_blocks", "q_idx", "query_layer", "raise", "rand_attn", "rand_band_product", "rand_mask", "random", "range", "return", "right_slice", "rsqrt_d", "same", "second_attn_weights", "second_context_layer", "second_key_mat", "second_last_attn_weights", "second_last_context_layer", "second_last_key_mat", "second_last_product", "second_last_rand_pad", "second_last_seq_pad", "second_last_value_mat", "second_product", "second_rand_pad", "second_seq_pad", "second_value_mat", "seed", "self", "softmax", "sqrt", "stack", "tensor", "the", "to", "to_block_size", "to_blocked_mask", "to_mask", "to_seq_len", "to_seq_length", "torch", "torch_bmm_nd", "torch_bmm_nd_transpose", "torch_gather_b2", "transpose", "unsqueeze", "unsqueeze_", "value_layer", "view", "w1", "w2", "zeros", "zip"], "big_bird/modeling_big_bird.py:BigBirdBlockSparseAttention.torch_gather_b2": ["Make", "ModelBlockSparseAttention", "Module", "ValueError", "and", "arange", "are", "but", "class", "def", "device", "dimensions", "div", "f", "first", "flattened_indices", "flattened_params", "floor", "identical", "if", "index_select", "indices", "indices_shift", "nn", "num_indices_to_gather", "num_indices_to_pick_from", "of", "out", "out_flattened", "params", "raise", "reshape", "return", "rounding_mode", "shape", "shift", "sure", "that", "the", "they", "torch", "torch_gather_b2", "two", "view", "vs"], "big_bird/modeling_big_bird.py:BigBirdBlockSparseAttention._create_rand_mask_from_inputs": ["ModelBlockSparseAttention", "Module", "_create_rand_mask_from_inputs", "batch_size", "bhlk", "bhlqk", "blq", "class", "def", "einsum", "flatten", "for", "from_block_size", "from_blocked_mask", "from_seq_length", "i1", "in", "nn", "num_attention_heads", "num_rand_blocks", "num_windows", "p1", "rand_attn", "rand_mask", "return", "stack", "to_blocked_mask", "torch", "view", "zip"], "big_bird/modeling_big_bird.py:BigBirdBlockSparseAttention._get_rand_attn_plan": ["ModelBlockSparseAttention", "Module", "_get_rand_attn_plan", "append", "class", "def", "elif", "else", "from_block_size", "from_seq_length", "if", "int", "nn", "num_rand_blocks", "plan_from_length", "plan_num_rand_blocks", "return"], "big_bird/modeling_big_bird.py:BigBirdBlockSparseAttention._bigbird_block_rand_mask": ["Error", "ModelBlockSparseAttention", "Module", "ValueError", "_Model_block_rand_mask", "arange", "be", "blocks", "class", "concatenate", "def", "dtype", "elif", "else", "end", "for", "from_block_size", "from_seq_length", "i", "if", "in", "int32", "last", "last_idx", "middle_seq", "needs", "nn", "not", "np", "num_rand_blocks", "number", "of", "permutation", "r", "raise", "rand_attn", "random", "range", "return", "same", "self", "start", "the", "to", "to_block_size", "to_seq_length", "training", "zeros"], "big_bird/modeling_big_bird.py:BigBirdBlockSparseAttention._bigbird_block_rand_mask_with_head": ["Error", "ModelBlockSparseAttention", "Module", "ValueError", "_Model_block_rand_mask_with_head", "_get_single_block_row_attention", "array", "be", "blk_rw_idx", "block_id", "blocks", "class", "continue", "curr_r_cnt", "def", "dtype", "for", "from", "from_block_size", "from_seq_length", "from_start_block_id", "global_block_bottom", "global_block_left", "global_block_right", "global_block_top", "h", "i", "if", "in", "index", "int", "int32", "length", "max_plan_idx", "needs", "nh", "nn", "not", "np", "num_blocks", "num_heads", "num_rand_blocks", "number", "of", "pl_id", "plan", "plan_block_length", "plan_from_length", "plan_idx", "plan_num_rand_blocks", "raise", "rand_attn", "range", "return", "rnd_r_cnt", "same", "self", "sequence", "sum", "the", "to", "to_block_size", "to_end_block_id", "to_seq_length", "to_start_block_id", "training", "window_block_left", "window_block_right", "zeros"], "big_bird/modeling_big_bird.py:BigBirdBlockSparseAttention._get_single_block_row_attention": ["ModelBlockSparseAttention", "Module", "_get_single_block_row_attention", "append", "arange", "array", "block_id", "break", "class", "def", "dtype", "extend", "for", "global_block_left", "global_block_right", "i", "if", "illegal_blocks", "in", "int32", "len", "list", "nn", "not", "np", "num_rand_blocks", "perm_block", "permutation", "random", "range", "return", "selected_random_blocks", "to_block_list", "to_end_block_id", "to_start_block_id", "window_block_left", "window_block_right"], "big_bird/modeling_big_bird.py:BigBirdSelfOutput.__init__": ["Dropout", "LayerNorm", "Linear", "ModelSelfOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "eps", "hidden_dropout_prob", "hidden_size", "layer_norm_eps", "nn", "self", "super"], "big_bird/modeling_big_bird.py:BigBirdSelfOutput.forward": ["LayerNorm", "ModelSelfOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "big_bird/modeling_big_bird.py:BigBirdAttention.__init__": ["ModelAttention", "ModelBlockSparseAttention", "ModelSelfAttention", "ModelSelfOutput", "Module", "None", "ValueError", "__init__", "attention_type", "be", "block_sparse", "but", "can", "class", "config", "def", "either", "elif", "else", "f", "if", "is", "layer_idx", "nn", "or", "original_full", "output", "raise", "seed", "self", "super"], "big_bird/modeling_big_bird.py:BigBirdAttention.set_attention_type": ["ModelAttention", "ModelBlockSparseAttention", "ModelSelfAttention", "Module", "None", "ValueError", "attention_type", "attn_weights", "be", "block_sparse", "but", "can", "class", "config", "def", "either", "else", "eval", "f", "if", "in", "is", "key", "layer_idx", "nn", "not", "only", "or", "original_full", "query", "raise", "return", "seed", "self", "set", "set_attention_type", "str", "to", "training", "value"], "big_bird/modeling_big_bird.py:BigBirdAttention.forward": ["False", "Model", "ModelAttention", "Module", "None", "ValueError", "a", "as", "attention_mask", "attention_output", "attention_type", "band_mask", "be", "cache_position", "cannot", "class", "config", "decoder", "def", "dtype", "else", "encoder_attention_mask", "encoder_hidden_states", "forward", "from_blocked_mask", "from_mask", "hidden_states", "if", "is", "nn", "not", "original_full", "output", "output_attentions", "outputs", "past_key_values", "raise", "return", "self", "self_outputs", "to", "to_blocked_mask", "to_mask", "used", "when"], "big_bird/modeling_big_bird.py:BigBirdIntermediate.__init__": ["ACT2FN", "Linear", "ModelIntermediate", "Module", "__init__", "class", "config", "def", "dense", "else", "hidden_act", "hidden_size", "if", "intermediate_act_fn", "intermediate_size", "isinstance", "nn", "self", "str", "super"], "big_bird/modeling_big_bird.py:BigBirdIntermediate.forward": ["ModelIntermediate", "Module", "Tensor", "class", "def", "dense", "forward", "hidden_states", "intermediate_act_fn", "nn", "return", "self", "torch"], "big_bird/modeling_big_bird.py:BigBirdOutput.__init__": ["Dropout", "LayerNorm", "Linear", "ModelOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "eps", "hidden_dropout_prob", "hidden_size", "intermediate_size", "layer_norm_eps", "nn", "self", "super"], "big_bird/modeling_big_bird.py:BigBirdOutput.forward": ["LayerNorm", "ModelOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "big_bird/modeling_big_bird.py:BigBirdLayer.__init__": ["GradientCheckpointingLayer", "ModelAttention", "ModelIntermediate", "ModelLayer", "ModelOutput", "None", "TypeError", "__init__", "a", "add_cross_attention", "added", "as", "attention", "attention_type", "be", "chunk_size_feed_forward", "class", "config", "cross", "crossattention", "decoder", "def", "f", "if", "intermediate", "is", "is_decoder", "model", "not", "output", "raise", "seed", "self", "seq_len_dim", "should", "super", "used"], "big_bird/modeling_big_bird.py:BigBirdLayer.set_attention_type": ["GradientCheckpointingLayer", "ModelLayer", "None", "ValueError", "add_cross_attention", "attention", "attention_type", "be", "block_sparse", "but", "can", "class", "crossattention", "def", "either", "f", "if", "in", "is", "layer_idx", "not", "only", "or", "original_full", "raise", "return", "self", "set", "set_attention_type", "str", "to", "value"], "big_bird/modeling_big_bird.py:BigBirdLayer.forward": ["False", "GradientCheckpointingLayer", "If", "ModelLayer", "None", "True", "ValueError", "add_cross_attention", "and", "apply_chunking_to_forward", "are", "attention", "attention_mask", "attention_output", "band_mask", "be", "blocked_encoder_mask", "by", "cache_position", "chunk_size_feed_forward", "class", "config", "cross", "cross_attention_outputs", "crossattention", "def", "encoder_attention_mask", "encoder_hidden_states", "f", "feed_forward_chunk", "forward", "from_blocked_mask", "from_mask", "has", "hasattr", "hidden_states", "if", "instantiated", "is", "is_decoder", "layer_output", "layers", "not", "output_attentions", "outputs", "passed", "past_key_values", "raise", "return", "self", "self_attention_outputs", "seq_len_dim", "setting", "to", "to_blocked_mask", "to_mask", "with"], "big_bird/modeling_big_bird.py:BigBirdLayer.feed_forward_chunk": ["GradientCheckpointingLayer", "ModelLayer", "attention_output", "class", "def", "feed_forward_chunk", "intermediate", "intermediate_output", "layer_output", "output", "return", "self"], "big_bird/modeling_big_bird.py:BigBirdEncoder.__init__": ["False", "ModelEncoder", "ModelLayer", "Module", "ModuleList", "__init__", "attention_type", "class", "config", "def", "for", "gradient_checkpointing", "in", "layer", "layer_idx", "nn", "num_hidden_layers", "range", "seed", "self", "super"], "big_bird/modeling_big_bird.py:BigBirdEncoder.set_attention_type": ["ModelEncoder", "Module", "ValueError", "attention_type", "be", "block_sparse", "but", "can", "class", "def", "either", "enumerate", "f", "for", "i", "if", "in", "is", "layer", "layer_idx", "nn", "not", "only", "or", "original_full", "raise", "return", "self", "set", "set_attention_type", "str", "to", "value"], "big_bird/modeling_big_bird.py:BigBirdEncoder.forward": ["BaseModelOutputWithPastAndCrossAttentions", "DynamicCache", "False", "ModelEncoder", "Module", "None", "Setting", "True", "add_cross_attention", "all_cross_attentions", "all_hidden_states", "all_self_attentions", "and", "attention_mask", "attentions", "band_mask", "blocked_encoder_mask", "cache_position", "checkpointing", "class", "config", "cross_attentions", "def", "else", "encoder_attention_mask", "encoder_hidden_states", "enumerate", "for", "forward", "from_mask", "gradient", "gradient_checkpointing", "hidden_states", "i", "if", "in", "incompatible", "is", "last_hidden_state", "layer", "layer_module", "layer_outputs", "logger", "nn", "not", "output_attentions", "output_hidden_states", "past_key_values", "return", "return_dict", "self", "to_mask", "training", "tuple", "use_cache", "v", "warning_once", "with"], "big_bird/modeling_big_bird.py:BigBirdPredictionHeadTransform.__init__": ["ACT2FN", "LayerNorm", "Linear", "ModelPredictionHeadTransform", "Module", "__init__", "class", "config", "def", "dense", "else", "eps", "hidden_act", "hidden_size", "if", "isinstance", "layer_norm_eps", "nn", "self", "str", "super", "transform_act_fn"], "big_bird/modeling_big_bird.py:BigBirdPredictionHeadTransform.forward": ["LayerNorm", "ModelPredictionHeadTransform", "Module", "Tensor", "class", "def", "dense", "forward", "hidden_states", "nn", "return", "self", "torch", "transform_act_fn"], "big_bird/modeling_big_bird.py:BigBirdLMPredictionHead.__init__": ["Linear", "ModelLMPredictionHead", "ModelPredictionHeadTransform", "Module", "Parameter", "True", "__init__", "bias", "class", "config", "decoder", "def", "hidden_size", "nn", "self", "super", "torch", "transform", "vocab_size", "zeros"], "big_bird/modeling_big_bird.py:BigBirdLMPredictionHead.forward": ["ModelLMPredictionHead", "Module", "class", "decoder", "def", "forward", "hidden_states", "nn", "return", "self", "transform"], "big_bird/modeling_big_bird.py:BigBirdOnlyMLMHead.__init__": ["ModelLMPredictionHead", "ModelOnlyMLMHead", "Module", "__init__", "class", "config", "def", "nn", "predictions", "self", "super"], "big_bird/modeling_big_bird.py:BigBirdOnlyMLMHead.forward": ["ModelOnlyMLMHead", "Module", "Tensor", "class", "def", "forward", "nn", "prediction_scores", "predictions", "return", "self", "sequence_output", "torch"], "big_bird/modeling_big_bird.py:BigBirdOnlyNSPHead.__init__": ["Linear", "ModelOnlyNSPHead", "Module", "__init__", "class", "config", "def", "hidden_size", "nn", "self", "seq_relationship", "super"], "big_bird/modeling_big_bird.py:BigBirdOnlyNSPHead.forward": ["ModelOnlyNSPHead", "Module", "class", "def", "forward", "nn", "pooled_output", "return", "self", "seq_relationship", "seq_relationship_score"], "big_bird/modeling_big_bird.py:BigBirdPreTrainingHeads.__init__": ["Linear", "ModelLMPredictionHead", "ModelPreTrainingHeads", "Module", "__init__", "class", "config", "def", "hidden_size", "nn", "predictions", "self", "seq_relationship", "super"], "big_bird/modeling_big_bird.py:BigBirdPreTrainingHeads.forward": ["ModelPreTrainingHeads", "Module", "class", "def", "forward", "nn", "pooled_output", "prediction_scores", "predictions", "return", "self", "seq_relationship", "seq_relationship_score", "sequence_output"], "big_bird/modeling_big_bird.py:BigBirdPreTrainedModel._init_weights": ["ModelEmbeddings", "ModelLMPredictionHead", "ModelPreTrainedModel", "PreTrainedModel", "_init_weights", "arange", "bias", "class", "copy_", "def", "elif", "expand", "if", "init", "isinstance", "module", "position_ids", "self", "shape", "super", "token_type_ids", "torch", "zeros_"], "big_bird/modeling_big_bird.py:BigBirdModel.__init__": ["Linear", "ModelEmbeddings", "ModelEncoder", "ModelForCausalLM", "ModelModel", "ModelPreTrainedModel", "None", "Setting", "Tanh", "True", "When", "__init__", "activation", "add_cross_attention", "add_pooling_layer", "and", "as", "attention_type", "be", "block_size", "class", "config", "decoder", "def", "else", "embeddings", "encoder", "hidden_size", "if", "logger", "must", "nn", "original_full", "pooler", "post_init", "r", "self", "set_attention_type", "super", "then", "using", "warning"], "big_bird/modeling_big_bird.py:BigBirdModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "return", "self", "word_embeddings"], "big_bird/modeling_big_bird.py:BigBirdModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "self", "set_input_embeddings", "value", "word_embeddings"], "big_bird/modeling_big_bird.py:BigBirdModel.set_attention_type": ["ModelModel", "ModelPreTrainedModel", "ValueError", "attention_type", "be", "block_sparse", "but", "can", "class", "def", "either", "encoder", "f", "if", "in", "is", "not", "only", "or", "original_full", "raise", "return", "self", "set", "set_attention_type", "str", "to", "value"], "big_bird/modeling_big_bird.py:BigBirdModel.forward": ["Attention", "BaseModelOutputWithPoolingAndCrossAttentions", "Cache", "Changing", "False", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "ValueError", "You", "_", "_pad_to_block_size", "activation", "additional", "and", "at", "attention", "attention_mask", "attention_type", "attentions", "band_mask", "batch_size", "be", "block_size", "block_sparse", "blocked_encoder_mask", "bool", "both", "buffer", "buffered_token_type_ids", "buffered_token_type_ids_expanded", "but", "cache_position", "can", "cannot", "class", "config", "create_masks_for_block_sparse_attn", "cross_attentions", "def", "device", "dtype", "either", "elif", "else", "embedding_output", "embeddings", "encoder", "encoder_attention_mask", "encoder_batch_size", "encoder_extended_attention_mask", "encoder_hidden_shape", "encoder_hidden_states", "encoder_outputs", "encoder_sequence_length", "expand", "extended_attention_mask", "f", "forward", "from_mask", "get_extended_attention_mask", "get_seq_length", "global", "hasattr", "have", "hidden_states", "if", "input_ids", "input_shape", "inputs_embeds", "invert_attention_mask", "is", "is_decoder", "kwargs", "last_hidden_state", "logger", "long", "max_tokens_to_attend", "min", "not", "num", "num_random_blocks", "ones", "or", "original_full", "output_attentions", "output_hidden_states", "pad_token_id", "padding_len", "past_key_values", "past_key_values_length", "pooler", "pooler_output", "position_ids", "possible", "raise", "return", "return_dict", "same", "self", "seq_length", "sequence_length", "sequence_output", "set_attention_type", "size", "sliding", "specify", "the", "time", "to", "to_mask", "token_type_ids", "tokens", "torch", "tuple", "type", "use_cache", "use_return_dict", "warn_if_padding_and_no_attention_mask", "warning", "with", "zeros"], "big_bird/modeling_big_bird.py:BigBirdModel.create_masks_for_block_sparse_attn": ["ModelModel", "ModelPreTrainedModel", "Sequence", "Tensor", "ValueError", "attention_mask", "band_mask", "batch_size", "be", "blk", "block", "block_size", "blocked_encoder_mask", "blq", "blqk", "but", "cat", "class", "create_band_mask_from_inputs", "create_masks_for_block_sparse_attn", "def", "dim", "einsum", "exp_blocked_to_pad", "f", "from_blocked_mask", "from_mask", "if", "int", "is", "length", "multiple", "must", "of", "raise", "return", "seq_length", "sequence", "size", "to_blocked_mask", "to_mask", "torch", "unsqueeze_", "view", "while"], "big_bird/modeling_big_bird.py:BigBirdModel._pad_to_block_size": ["False", "Input", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "_pad_to_block_size", "a", "are", "attention_mask", "automatically", "batch_size", "be", "block_size", "cat", "class", "config", "def", "dim", "dtype", "else", "embeddings", "f", "from", "functional", "ids", "if", "input_ids", "input_ids_padding", "input_shape", "inputs_embeds", "inputs_embeds_padding", "int", "is", "logger", "long", "multiple", "new_full", "nn", "not", "of", "pad", "pad_token_id", "padded", "padding_len", "position_ids", "return", "self", "seq_len", "shape", "to", "token_type_ids", "torch", "value", "warning_once"], "big_bird/modeling_big_bird.py:BigBirdForPreTraining.__init__": ["ModelForPreTraining", "ModelModel", "ModelPreTrainedModel", "ModelPreTrainingHeads", "True", "__init__", "add_pooling_layer", "bert", "class", "cls", "config", "def", "post_init", "self", "super"], "big_bird/modeling_big_bird.py:BigBirdForPreTraining.get_output_embeddings": ["ModelForPreTraining", "ModelPreTrainedModel", "class", "cls", "decoder", "def", "get_output_embeddings", "predictions", "return", "self"], "big_bird/modeling_big_bird.py:BigBirdForPreTraining.set_output_embeddings": ["ModelForPreTraining", "ModelPreTrainedModel", "bias", "class", "cls", "decoder", "def", "new_embeddings", "predictions", "self", "set_output_embeddings"], "big_bird/modeling_big_bird.py:BigBirdForPreTraining.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "ModelForPreTraining", "ModelForPreTrainingOutput", "ModelPreTrainedModel", "None", "and", "attention_mask", "attentions", "bert", "bool", "class", "cls", "config", "def", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "loss", "loss_fct", "next_sentence_label", "next_sentence_loss", "not", "output", "output_attentions", "output_hidden_states", "outputs", "pooled_output", "position_ids", "prediction_logits", "prediction_scores", "r", "return", "return_dict", "self", "seq_relationship_logits", "seq_relationship_score", "sequence_output", "token_type_ids", "torch", "total_loss", "tuple", "use_return_dict", "view", "vocab_size"], "big_bird/modeling_big_bird.py:BigBirdForMaskedLM.__init__": ["False", "If", "ModelForMaskedLM", "ModelModel", "ModelOnlyMLMHead", "ModelPreTrainedModel", "__init__", "attention", "bert", "bi", "class", "cls", "config", "def", "directional", "for", "if", "is_decoder", "logger", "make", "post_init", "self", "super", "sure", "to", "use", "want", "warning", "you"], "big_bird/modeling_big_bird.py:BigBirdForMaskedLM.get_output_embeddings": ["ModelForMaskedLM", "ModelPreTrainedModel", "class", "cls", "decoder", "def", "get_output_embeddings", "predictions", "return", "self"], "big_bird/modeling_big_bird.py:BigBirdForMaskedLM.set_output_embeddings": ["ModelForMaskedLM", "ModelPreTrainedModel", "bias", "class", "cls", "decoder", "def", "new_embeddings", "predictions", "self", "set_output_embeddings"], "big_bird/modeling_big_bird.py:BigBirdForMaskedLM.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "MaskedLMOutput", "ModelForMaskedLM", "ModelPreTrainedModel", "None", "attention_mask", "attentions", "bert", "bool", "class", "cls", "config", "def", "else", "encoder_attention_mask", "encoder_hidden_states", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "masked_lm_loss", "not", "output", "output_attentions", "output_hidden_states", "outputs", "position_ids", "prediction_scores", "r", "return", "return_dict", "self", "sequence_output", "token_type_ids", "torch", "tuple", "use_return_dict", "view", "vocab_size"], "big_bird/modeling_big_bird.py:BigBirdForMaskedLM.prepare_inputs_for_generation": ["ModelForMaskedLM", "ModelPreTrainedModel", "None", "PAD", "The", "ValueError", "attention_mask", "be", "cat", "class", "config", "def", "defined", "device", "dim", "dtype", "dummy_token", "effective_batch_size", "for", "full", "generation", "if", "input_ids", "input_shape", "is", "long", "model_kwargs", "new_zeros", "pad_token_id", "prepare_inputs_for_generation", "raise", "return", "self", "shape", "should", "token", "torch"], "big_bird/modeling_big_bird.py:BigBirdForCausalLM.__init__": ["GenerationMixin", "If", "ModelForCausalLM", "ModelModel", "ModelOnlyMLMHead", "ModelPreTrainedModel", "True", "__init__", "a", "add", "as", "bert", "class", "cls", "config", "def", "if", "is_decoder", "logger", "not", "post_init", "self", "standalone", "super", "to", "use", "want", "warning", "you"], "big_bird/modeling_big_bird.py:BigBirdForCausalLM.get_output_embeddings": ["GenerationMixin", "ModelForCausalLM", "ModelPreTrainedModel", "class", "cls", "decoder", "def", "get_output_embeddings", "predictions", "return", "self"], "big_bird/modeling_big_bird.py:BigBirdForCausalLM.set_output_embeddings": ["GenerationMixin", "ModelForCausalLM", "ModelPreTrainedModel", "bias", "class", "cls", "decoder", "def", "new_embeddings", "predictions", "self", "set_output_embeddings"], "big_bird/modeling_big_bird.py:BigBirdForCausalLM.forward": ["Cache", "CausalLMOutputWithCrossAttentions", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForCausalLM", "ModelPreTrainedModel", "None", "Tensor", "attention_mask", "attentions", "bert", "bool", "cache_position", "class", "cls", "config", "cross_attentions", "def", "else", "encoder_attention_mask", "encoder_hidden_states", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "logits", "logits_to_keep", "loss", "loss_function", "not", "output", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "position_ids", "r", "return", "return_dict", "self", "slice", "slice_indices", "token_type_ids", "torch", "tuple", "use_cache", "use_return_dict", "vocab_size"], "big_bird/modeling_big_bird.py:BigBirdClassificationHead.__init__": ["Dropout", "Linear", "ModelClassificationHead", "Module", "None", "__init__", "class", "classifier_dropout", "config", "def", "dense", "dropout", "else", "hidden_dropout_prob", "hidden_size", "if", "is", "nn", "not", "num_labels", "out_proj", "self", "super"], "big_bird/modeling_big_bird.py:BigBirdClassificationHead.forward": ["ACT2FN", "ModelClassificationHead", "Module", "class", "config", "def", "dense", "dropout", "features", "forward", "hidden_act", "kwargs", "nn", "out_proj", "return", "self", "x"], "big_bird/modeling_big_bird.py:BigBirdForSequenceClassification.__init__": ["ModelClassificationHead", "ModelForSequenceClassification", "ModelModel", "ModelPreTrainedModel", "__init__", "bert", "class", "classifier", "config", "def", "num_labels", "post_init", "self", "super"], "big_bird/modeling_big_bird.py:BigBirdForSequenceClassification.forward": ["BCEWithLogitsLoss", "CrossEntropyLoss", "FloatTensor", "LongTensor", "MSELoss", "ModelForSequenceClassification", "ModelPreTrainedModel", "None", "SequenceClassifierOutput", "and", "attention_mask", "attentions", "bert", "bool", "class", "classifier", "config", "def", "dtype", "elif", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "kwargs", "labels", "logits", "long", "loss", "loss_fct", "multi_label_classification", "not", "num_labels", "or", "output", "output_attentions", "output_hidden_states", "outputs", "position_ids", "problem_type", "r", "regression", "return", "return_dict", "self", "sequence_output", "single_label_classification", "squeeze", "token_type_ids", "torch", "tuple", "use_return_dict", "view"], "big_bird/modeling_big_bird.py:BigBirdForMultipleChoice.__init__": ["Dropout", "Linear", "ModelForMultipleChoice", "ModelModel", "ModelPreTrainedModel", "__init__", "bert", "class", "classifier", "config", "def", "dropout", "hidden_dropout_prob", "hidden_size", "nn", "post_init", "self", "super"], "big_bird/modeling_big_bird.py:BigBirdForMultipleChoice.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "ModelForMultipleChoice", "ModelPreTrainedModel", "MultipleChoiceModelOutput", "None", "attention_mask", "attentions", "bert", "bool", "class", "classifier", "config", "def", "dropout", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "not", "num_choices", "output", "output_attentions", "output_hidden_states", "outputs", "pooled_output", "position_ids", "r", "reshaped_logits", "return", "return_dict", "self", "shape", "size", "token_type_ids", "torch", "tuple", "use_return_dict", "view"], "big_bird/modeling_big_bird.py:BigBirdForTokenClassification.__init__": ["Dropout", "Linear", "ModelForTokenClassification", "ModelModel", "ModelPreTrainedModel", "None", "__init__", "bert", "class", "classifier", "classifier_dropout", "config", "def", "dropout", "else", "hidden_dropout_prob", "hidden_size", "if", "is", "nn", "not", "num_labels", "post_init", "self", "super"], "big_bird/modeling_big_bird.py:BigBirdForTokenClassification.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "ModelForTokenClassification", "ModelPreTrainedModel", "None", "TokenClassifierOutput", "attention_mask", "attentions", "bert", "bool", "class", "classifier", "config", "def", "dropout", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "not", "num_labels", "output", "output_attentions", "output_hidden_states", "outputs", "position_ids", "r", "return", "return_dict", "self", "sequence_output", "token_type_ids", "torch", "tuple", "use_return_dict", "view"], "big_bird/modeling_big_bird.py:BigBirdForQuestionAnsweringHead.__init__": ["Dropout", "Linear", "ModelForQuestionAnsweringHead", "ModelIntermediate", "ModelOutput", "Module", "__init__", "class", "config", "def", "dropout", "hidden_dropout_prob", "hidden_size", "intermediate", "nn", "num_labels", "output", "qa_outputs", "self", "super"], "big_bird/modeling_big_bird.py:BigBirdForQuestionAnsweringHead.forward": ["ModelForQuestionAnsweringHead", "Module", "class", "def", "dropout", "encoder_output", "forward", "hidden_states", "intermediate", "nn", "output", "qa_outputs", "return", "self"], "big_bird/modeling_big_bird.py:BigBirdForQuestionAnswering.__init__": ["False", "ModelForQuestionAnswering", "ModelForQuestionAnsweringHead", "ModelModel", "ModelPreTrainedModel", "__init__", "add_pooling_layer", "bert", "class", "config", "def", "num_labels", "post_init", "qa_classifier", "r", "self", "sep_token_id", "super"], "big_bird/modeling_big_bird.py:BigBirdForQuestionAnswering.forward": ["CrossEntropyLoss", "False", "FloatTensor", "LongTensor", "ModelForQuestionAnswering", "ModelForQuestionAnsweringModelOutput", "ModelPreTrainedModel", "None", "and", "argmax", "attention_mask", "attentions", "bert", "bool", "clamp", "class", "config", "contiguous", "def", "device", "dim", "dtype", "else", "end_logits", "end_loss", "end_positions", "eq", "forward", "hidden_states", "if", "ignore_index", "ignored_index", "input_ids", "inputs_embeds", "int", "is", "kwargs", "len", "logits", "logits_mask", "loss", "loss_fct", "not", "ones", "output", "output_attentions", "output_hidden_states", "outputs", "pooler_output", "position_ids", "prepare_question_mask", "qa_classifier", "question_lengths", "r", "return", "return_dict", "self", "sep_token_id", "seqlen", "sequence_output", "size", "split", "squeeze", "start_logits", "start_loss", "start_positions", "token_type_ids", "torch", "total_loss", "tuple", "unsqueeze_", "use_return_dict"], "big_bird/modeling_big_bird.py:BigBirdForQuestionAnswering.prepare_question_mask": ["ModelForQuestionAnswering", "ModelPreTrainedModel", "Tensor", "arange", "class", "def", "device", "int", "mask", "maxlen", "prepare_question_mask", "q_lengths", "return", "to", "torch", "unsqueeze_", "where"], "bigbird_pegasus/modeling_bigbird_pegasus.py:shift_tokens_right": ["Model_tokens_right", "Modeled_input_ids", "None", "Tensor", "ValueError", "be", "clone", "config", "decoder_start_token_id", "def", "defined", "has", "if", "input_ids", "int", "is", "masked_fill_", "model", "new_zeros", "pad_token_id", "raise", "return", "self", "shape", "to", "torch"], "bigbird_pegasus/modeling_bigbird_pegasus.py:BigBirdPegasusLearnedPositionalEmbedding.__init__": ["Embedding", "ModelLearnedPositionalEmbedding", "__init__", "class", "def", "embedding_dim", "int", "nn", "num_embeddings", "self", "super"], "bigbird_pegasus/modeling_bigbird_pegasus.py:BigBirdPegasusLearnedPositionalEmbedding.forward": ["Embedding", "ModelLearnedPositionalEmbedding", "None", "Size", "Tensor", "arange", "bsz", "class", "def", "device", "dtype", "forward", "if", "input_ids_shape", "int", "is", "long", "nn", "past_key_values_length", "position_ids", "return", "self", "seq_len", "super", "torch", "weight"], "bigbird_pegasus/modeling_bigbird_pegasus.py:BigBirdPegasusScaledWordEmbedding.__init__": ["Embedding", "ModelScaledWordEmbedding", "None", "__init__", "class", "def", "embed_scale", "embedding_dim", "float", "int", "nn", "num_embeddings", "padding_idx", "self", "super"], "bigbird_pegasus/modeling_bigbird_pegasus.py:BigBirdPegasusScaledWordEmbedding.forward": ["Embedding", "ModelScaledWordEmbedding", "Tensor", "class", "def", "embed_scale", "forward", "input_ids", "nn", "return", "self", "super", "torch"], "bigbird_pegasus/modeling_bigbird_pegasus.py:BigBirdPegasusSelfAttention.__init__": ["Dropout", "Linear", "ModelSelfAttention", "Module", "None", "The", "ValueError", "__init__", "a", "all_head_size", "and", "attention", "attention_head_size", "attention_probs_dropout_prob", "bias", "class", "config", "def", "dropout", "embedding_size", "f", "hasattr", "heads", "hidden", "hidden_size", "if", "int", "is", "is_decoder", "key", "layer_idx", "multiple", "nn", "not", "num_attention_heads", "number", "of", "query", "raise", "self", "size", "super", "the", "use_bias", "value"], "bigbird_pegasus/modeling_bigbird_pegasus.py:BigBirdPegasusSelfAttention.forward": ["False", "ModelSelfAttention", "Module", "None", "_", "all_head_size", "and", "attention_head_size", "attention_mask", "attention_probs", "attention_scores", "batch_size", "cache_position", "class", "context_layer", "contiguous", "current_states", "def", "dim", "dropout", "else", "encoder_attention_mask", "encoder_hidden_states", "forward", "functional", "get_seq_length", "hidden_states", "if", "is", "is_cross_attention", "key", "key_layer", "keys", "layer_idx", "layers", "math", "matmul", "new_context_layer_shape", "nn", "not", "num_attention_heads", "output_attentions", "past_key_values", "permute", "query", "query_layer", "return", "self", "seq_length", "shape", "size", "softmax", "sqrt", "torch", "transpose", "update", "value", "value_layer", "values", "view"], "bigbird_pegasus/modeling_bigbird_pegasus.py:BigBirdPegasusBlockSparseAttention.__init__": ["Linear", "ModelBlockSparseAttention", "Module", "None", "The", "ValueError", "__init__", "a", "all_head_size", "attention", "attention_head_size", "bias", "block_size", "class", "config", "def", "f", "heads", "hidden", "hidden_size", "if", "int", "is", "key", "max_position_embeddings", "max_seqlen", "multiple", "nn", "not", "num_attention_heads", "num_random_blocks", "number", "of", "query", "raise", "seed", "self", "size", "super", "the", "use_bias", "value"], "bigbird_pegasus/modeling_bigbird_pegasus.py:BigBirdPegasusBlockSparseAttention.forward": ["Key", "ModelBlockSparseAttention", "Modelbird_block_sparse_attention", "Module", "None", "Query", "Value", "ValueError", "_", "attention_head_size", "attention_probs", "band_mask", "batch_size", "be", "block", "block_size", "class", "context_layer", "contiguous", "def", "forward", "from_block_size", "from_blocked_mask", "from_mask", "from_seq_length", "hidden_states", "if", "key", "key_layer", "length", "multiple", "must", "nn", "num_attention_heads", "num_random_blocks", "of", "output_attentions", "plan_from_length", "plan_num_rand_blocks", "query", "query_layer", "raise", "return", "seed", "self", "seqlen", "sequence", "sided", "size", "to_block_size", "to_blocked_mask", "to_mask", "to_seq_length", "transpose", "value", "value_layer", "view"], "bigbird_pegasus/modeling_bigbird_pegasus.py:BigBirdPegasusBlockSparseAttention.torch_bmm_nd": ["ModelBlockSparseAttention", "Module", "None", "bmm", "class", "def", "inp_1", "inp_2", "ndim", "nn", "reshape", "return", "shape", "torch", "torch_bmm_nd", "view"], "bigbird_pegasus/modeling_bigbird_pegasus.py:BigBirdPegasusBlockSparseAttention.torch_bmm_nd_transpose": ["ModelBlockSparseAttention", "Module", "None", "bmm", "class", "def", "inp_1", "inp_2", "ndim", "nn", "reshape", "return", "shape", "torch", "torch_bmm_nd_transpose", "transpose", "view"], "bigbird_pegasus/modeling_bigbird_pegasus.py:BigBirdPegasusBlockSparseAttention.bigbird_block_sparse_attention": ["Error", "ModelBlockSparseAttention", "Modelbird_block_sparse_attention", "Module", "None", "ValueError", "_", "_Modelbird_block_rand_mask", "_Modelbird_block_rand_mask_with_head", "_create_rand_mask_from_inputs", "_get_rand_attn_plan", "attention_head_size", "attention_probs", "attn_mask_penalty", "attn_probs_view", "attn_weights", "axis", "band_mask", "band_product", "batch_size", "be", "bhkd", "bhlqd", "bhlqk", "blocked_key_matrix", "blocked_query_matrix", "blocked_value_matrix", "blocks", "bsz", "cat", "class", "context_layer", "def", "device", "dim", "dtype", "einsum", "else", "exp_blocked_key_matrix", "exp_blocked_value_matrix", "first_attn_weights", "first_band_product", "first_context_layer", "first_product", "float", "for", "from_block_size", "from_blocked_mask", "from_mask", "from_seq_len", "from_seq_length", "functional", "gathered_key", "gathered_value", "i1", "i2", "if", "in", "inner_band_product", "is", "key_layer", "last_attn_weights", "last_band_product", "last_context_layer", "last_idx", "last_product", "len", "long", "math", "max_seqlen", "middle_query_matrix", "minimum", "n_heads", "n_rand_blocks", "ndim", "needs", "new_ones", "nn", "np", "num_heads", "number", "of", "output_attentions", "p1", "p2", "plan_from_length", "plan_num_rand_blocks", "q_idx", "query_layer", "raise", "rand_attn", "rand_band_product", "rand_mask", "random", "range", "return", "right_slice", "rsqrt_d", "same", "second_attn_weights", "second_context_layer", "second_key_mat", "second_last_attn_weights", "second_last_context_layer", "second_last_key_mat", "second_last_product", "second_last_rand_pad", "second_last_seq_pad", "second_last_value_mat", "second_product", "second_rand_pad", "second_seq_pad", "second_value_mat", "seed", "self", "softmax", "sqrt", "stack", "tensor", "the", "to", "to_block_size", "to_blocked_mask", "to_mask", "to_seq_len", "to_seq_length", "torch", "torch_bmm_nd", "torch_bmm_nd_transpose", "torch_gather_b2", "transpose", "unsqueeze", "unsqueeze_", "value_layer", "view", "w1", "w2", "zeros", "zip"], "bigbird_pegasus/modeling_bigbird_pegasus.py:BigBirdPegasusBlockSparseAttention.torch_gather_b2": ["Make", "ModelBlockSparseAttention", "Module", "ValueError", "and", "arange", "are", "but", "class", "def", "device", "dimensions", "div", "f", "first", "flattened_indices", "flattened_params", "floor", "identical", "if", "index_select", "indices", "indices_shift", "nn", "num_indices_to_gather", "num_indices_to_pick_from", "of", "out", "out_flattened", "params", "raise", "reshape", "return", "rounding_mode", "shape", "shift", "sure", "that", "the", "they", "torch", "torch_gather_b2", "two", "view", "vs"], "bigbird_pegasus/modeling_bigbird_pegasus.py:BigBirdPegasusBlockSparseAttention._create_rand_mask_from_inputs": ["ModelBlockSparseAttention", "Module", "_create_rand_mask_from_inputs", "batch_size", "bhlk", "bhlqk", "blq", "class", "def", "einsum", "flatten", "for", "from_block_size", "from_blocked_mask", "from_seq_length", "i1", "in", "nn", "num_attention_heads", "num_rand_blocks", "num_windows", "p1", "rand_attn", "rand_mask", "return", "stack", "to_blocked_mask", "torch", "view", "zip"], "bigbird_pegasus/modeling_bigbird_pegasus.py:BigBirdPegasusBlockSparseAttention._get_rand_attn_plan": ["ModelBlockSparseAttention", "Module", "_get_rand_attn_plan", "append", "class", "def", "elif", "else", "from_block_size", "from_seq_length", "if", "int", "nn", "num_rand_blocks", "plan_from_length", "plan_num_rand_blocks", "return"], "bigbird_pegasus/modeling_bigbird_pegasus.py:BigBirdPegasusBlockSparseAttention._bigbird_block_rand_mask": ["Error", "ModelBlockSparseAttention", "Module", "ValueError", "_Modelbird_block_rand_mask", "arange", "be", "blocks", "class", "concatenate", "def", "dtype", "elif", "else", "end", "for", "from_block_size", "from_seq_length", "i", "if", "in", "int32", "last", "last_idx", "middle_seq", "needs", "nn", "not", "np", "num_rand_blocks", "number", "of", "permutation", "r", "raise", "rand_attn", "random", "range", "return", "same", "self", "start", "the", "to", "to_block_size", "to_seq_length", "training", "zeros"], "bigbird_pegasus/modeling_bigbird_pegasus.py:BigBirdPegasusBlockSparseAttention._bigbird_block_rand_mask_with_head": ["Error", "ModelBlockSparseAttention", "Module", "ValueError", "_Modelbird_block_rand_mask_with_head", "_get_single_block_row_attention", "array", "be", "blk_rw_idx", "block_id", "blocks", "class", "continue", "curr_r_cnt", "def", "dtype", "for", "from", "from_block_size", "from_seq_length", "from_start_block_id", "global_block_bottom", "global_block_left", "global_block_right", "global_block_top", "h", "i", "if", "in", "index", "int", "int32", "length", "max_plan_idx", "needs", "nh", "nn", "not", "np", "num_blocks", "num_heads", "num_rand_blocks", "number", "of", "pl_id", "plan", "plan_block_length", "plan_from_length", "plan_idx", "plan_num_rand_blocks", "raise", "rand_attn", "range", "return", "rnd_r_cnt", "same", "self", "sequence", "sum", "the", "to", "to_block_size", "to_end_block_id", "to_seq_length", "to_start_block_id", "training", "window_block_left", "window_block_right", "zeros"], "bigbird_pegasus/modeling_bigbird_pegasus.py:BigBirdPegasusBlockSparseAttention._get_single_block_row_attention": ["ModelBlockSparseAttention", "Module", "_get_single_block_row_attention", "append", "arange", "array", "block_id", "break", "class", "def", "dtype", "extend", "for", "global_block_left", "global_block_right", "i", "if", "illegal_blocks", "in", "int32", "len", "list", "nn", "not", "np", "num_rand_blocks", "perm_block", "permutation", "random", "range", "return", "selected_random_blocks", "to_block_list", "to_end_block_id", "to_start_block_id", "window_block_left", "window_block_right"], "bigbird_pegasus/modeling_bigbird_pegasus.py:BigBirdPegasusEncoderAttention.__init__": ["Linear", "ModelBlockSparseAttention", "ModelEncoderAttention", "ModelSelfAttention", "Module", "None", "ValueError", "__init__", "attention_type", "be", "bias", "block_sparse", "but", "can", "class", "config", "def", "either", "elif", "else", "f", "hidden_size", "if", "is", "nn", "or", "original_full", "output", "raise", "seed", "self", "super", "use_bias"], "bigbird_pegasus/modeling_bigbird_pegasus.py:BigBirdPegasusEncoderAttention.set_attention_type": ["ModelBlockSparseAttention", "ModelEncoderAttention", "ModelSelfAttention", "Module", "ValueError", "attention_type", "attn_weights", "be", "block_sparse", "but", "can", "class", "config", "def", "either", "else", "eval", "f", "if", "in", "is", "key", "nn", "not", "only", "or", "original_full", "query", "raise", "return", "seed", "self", "set", "set_attention_type", "str", "to", "training", "value"], "bigbird_pegasus/modeling_bigbird_pegasus.py:BigBirdPegasusEncoderAttention.forward": ["False", "ModelEncoderAttention", "Module", "None", "attention_mask", "attention_output", "attention_type", "band_mask", "class", "def", "else", "forward", "from_blocked_mask", "from_mask", "hidden_states", "if", "nn", "original_full", "output", "output_attentions", "outputs", "return", "self", "self_outputs", "to_blocked_mask", "to_mask"], "bigbird_pegasus/modeling_bigbird_pegasus.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "contiguous", "def", "dim", "dropout", "dtype", "float", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "shape", "size", "softmax", "to", "torch", "training", "transpose", "value"], "bigbird_pegasus/modeling_bigbird_pegasus.py:BigBirdPegasusDecoderAttention.__init__": ["False", "Instantiating", "Linear", "ModelConfig", "ModelDecoderAttention", "Module", "None", "Please", "True", "ValueError", "__class__", "__init__", "__name__", "a", "and", "be", "bias", "bool", "by", "caching", "call", "class", "config", "creating", "decoder", "def", "divisible", "dropout", "during", "embed_dim", "errors", "f", "float", "forward", "got", "head_dim", "if", "int", "is", "is_causal", "is_decoder", "k_proj", "layer_idx", "lead", "logger", "make", "must", "nn", "not", "num_heads", "out_proj", "passing", "provide", "q_proj", "raise", "recommended", "scaling", "self", "super", "sure", "the", "this", "to", "used", "v_proj", "warning_once", "when", "will", "without"], "bigbird_pegasus/modeling_bigbird_pegasus.py:BigBirdPegasusDecoderAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "EncoderDecoderCache", "False", "FlashAttentionKwargs", "ModelDecoderAttention", "Module", "None", "Tensor", "True", "Unpack", "_attn_implementation", "and", "attention_interface", "attention_mask", "attn_output", "attn_weights", "bool", "bsz", "cache_position", "class", "config", "contiguous", "cross_attention_cache", "curr_past_key_values", "current_states", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "get", "head_dim", "hidden_states", "if", "is", "is_cross_attention", "is_updated", "isinstance", "k_proj", "key_states", "key_value_states", "keys", "kv_input_shape", "kwargs", "layer_idx", "layers", "nn", "not", "out_proj", "output_attentions", "past_key_values", "q_input_shape", "q_proj", "query_states", "reshape", "return", "scaling", "self", "self_attention_cache", "shape", "src_len", "tgt_len", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "values", "view"], "bigbird_pegasus/modeling_bigbird_pegasus.py:BigBirdPegasusEncoderLayer.__init__": ["ACT2FN", "GradientCheckpointingLayer", "LayerNorm", "Linear", "ModelConfig", "ModelEncoderAttention", "ModelEncoderLayer", "None", "__init__", "activation_dropout", "activation_fn", "activation_function", "attention_type", "class", "config", "d_model", "def", "dropout", "embed_dim", "encoder_ffn_dim", "fc1", "fc2", "final_layer_norm", "nn", "seed", "self", "self_attn", "self_attn_layer_norm", "super"], "bigbird_pegasus/modeling_bigbird_pegasus.py:BigBirdPegasusEncoderLayer.forward": ["False", "GradientCheckpointingLayer", "ModelEncoderLayer", "None", "Tensor", "activation_fn", "and", "any", "attention_mask", "band_mask", "bool", "clamp", "clamp_value", "class", "def", "dropout", "dtype", "fc1", "fc2", "final_layer_norm", "finfo", "float16", "forward", "from_blocked_mask", "from_mask", "functional", "hidden_states", "if", "isinf", "isnan", "max", "min", "nn", "or", "output_attentions", "outputs", "p", "residual", "return", "self", "self_attention_outputs", "self_attn", "self_attn_layer_norm", "to_blocked_mask", "to_mask", "torch", "training"], "bigbird_pegasus/modeling_bigbird_pegasus.py:BigBirdPegasusEncoderLayer.set_attention_type": ["GradientCheckpointingLayer", "ModelEncoderLayer", "ValueError", "attention_type", "be", "block_sparse", "but", "can", "class", "def", "either", "f", "if", "in", "is", "not", "only", "or", "original_full", "raise", "return", "self", "self_attn", "set", "set_attention_type", "str", "to", "value"], "bigbird_pegasus/modeling_bigbird_pegasus.py:BigBirdPegasusDecoderLayer.__init__": ["ACT2FN", "GradientCheckpointingLayer", "LayerNorm", "Linear", "ModelConfig", "ModelDecoderAttention", "ModelDecoderLayer", "None", "True", "__init__", "activation_dropout", "activation_fn", "activation_function", "attention_dropout", "bias", "class", "config", "d_model", "decoder_attention_heads", "decoder_ffn_dim", "def", "dropout", "embed_dim", "encoder_attn", "encoder_attn_layer_norm", "fc1", "fc2", "final_layer_norm", "int", "is_decoder", "layer_idx", "nn", "num_heads", "self", "self_attn", "self_attn_layer_norm", "super", "use_bias"], "bigbird_pegasus/modeling_bigbird_pegasus.py:BigBirdPegasusDecoderLayer.forward": ["Cache", "False", "GradientCheckpointingLayer", "ModelDecoderLayer", "None", "Tensor", "True", "activation_dropout", "activation_fn", "attention_mask", "bool", "cache_position", "class", "cross_attn_weights", "def", "dropout", "encoder_attention_mask", "encoder_attn", "encoder_attn_layer_norm", "encoder_hidden_states", "fc1", "fc2", "final_layer_norm", "forward", "functional", "hidden_states", "if", "is", "key_value_states", "nn", "not", "output_attentions", "outputs", "p", "past_key_values", "residual", "return", "self", "self_attn", "self_attn_layer_norm", "self_attn_weights", "torch", "training", "use_cache"], "bigbird_pegasus/modeling_bigbird_pegasus.py:BigBirdPegasusClassificationHead.__init__": ["Dropout", "Linear", "ModelClassificationHead", "Module", "__init__", "class", "def", "dense", "dropout", "float", "inner_dim", "input_dim", "int", "nn", "num_classes", "out_proj", "p", "pooler_dropout", "self", "super"], "bigbird_pegasus/modeling_bigbird_pegasus.py:BigBirdPegasusClassificationHead.forward": ["ModelClassificationHead", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "nn", "out_proj", "return", "self", "tanh", "torch"], "bigbird_pegasus/modeling_bigbird_pegasus.py:BigBirdPegasusPreTrainedModel._init_weights": ["ModelForConditionalGeneration", "ModelPreTrainedModel", "PreTrainedModel", "_init_weights", "class", "def", "final_logits_bias", "if", "init", "isinstance", "module", "self", "super", "zeros_"], "bigbird_pegasus/modeling_bigbird_pegasus.py:BigBirdPegasusPreTrainedModel.dummy_inputs": ["ModelPreTrainedModel", "PreTrainedModel", "attention_mask", "class", "config", "def", "device", "dummy_inputs", "input_ids", "ne", "pad_token", "pad_token_id", "return", "self", "tensor", "torch"], "bigbird_pegasus/modeling_bigbird_pegasus.py:BigBirdPegasusEncoder.__init__": ["False", "LayerNorm", "ModelConfig", "ModelEncoder", "ModelEncoderLayer", "ModelLearnedPositionalEmbedding", "ModelPreTrainedModel", "ModelScaledWordEmbedding", "ModuleList", "__init__", "attention_type", "block_size", "class", "config", "d_model", "def", "dropout", "else", "embed_dim", "embed_positions", "embed_scale", "embed_tokens", "encoder_layerdrop", "encoder_layers", "for", "gradient_checkpointing", "i", "if", "in", "layerdrop", "layernorm_embedding", "layers", "math", "max_position_embeddings", "max_source_positions", "nn", "pad_token_id", "padding_idx", "post_init", "range", "scale_embedding", "seed", "self", "sqrt", "super", "vocab_size"], "bigbird_pegasus/modeling_bigbird_pegasus.py:BigBirdPegasusEncoder.forward": ["Attention", "BaseModelOutput", "Changing", "False", "ModelEncoder", "ModelPreTrainedModel", "None", "Tensor", "True", "ValueError", "You", "_pad_to_block_size", "_prepare_4d_attention_mask", "additional", "all_attentions", "and", "at", "attention", "attention_mask", "attention_type", "attentions", "band_mask", "be", "block_size", "block_sparse", "blocked_encoder_mask", "bool", "both", "buffer", "but", "can", "cannot", "class", "config", "create_masks_for_block_sparse_attn", "def", "device", "dropout", "dropout_probability", "dtype", "either", "elif", "else", "embed_pos", "embed_positions", "embed_tokens", "encoder_layer", "encoder_states", "enumerate", "f", "for", "forward", "from_blocked_mask", "from_mask", "functional", "global", "have", "hidden_states", "idx", "if", "in", "input_ids", "input_shape", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layer_outputs", "layerdrop", "layernorm_embedding", "layers", "logger", "long", "max_tokens_to_attend", "min", "nn", "not", "num", "num_random_blocks", "ones", "or", "original_full", "output_attentions", "output_hidden_states", "p", "padding_len", "possible", "r", "raise", "rand", "return", "return_dict", "same", "self", "sequence_length", "set_attention_type", "size", "sliding", "specify", "the", "time", "to", "to_blocked_mask", "to_drop", "to_mask", "tokens", "torch", "training", "tuple", "type", "use_return_dict", "v", "view", "warn_if_padding_and_no_attention_mask", "warning", "with"], "bigbird_pegasus/modeling_bigbird_pegasus.py:BigBirdPegasusEncoder.set_attention_type": ["ModelEncoder", "ModelPreTrainedModel", "ValueError", "attention_type", "be", "block_sparse", "but", "can", "class", "def", "either", "f", "for", "if", "in", "is", "layer", "layers", "not", "only", "or", "original_full", "raise", "return", "self", "set", "set_attention_type", "str", "to", "value"], "bigbird_pegasus/modeling_bigbird_pegasus.py:BigBirdPegasusEncoder.create_masks_for_block_sparse_attn": ["ModelEncoder", "ModelPreTrainedModel", "Sequence", "Tensor", "ValueError", "attention_mask", "band_mask", "batch_size", "be", "blk", "block", "block_size", "blocked_encoder_mask", "blq", "blqk", "but", "cat", "class", "create_band_mask_from_inputs", "create_masks_for_block_sparse_attn", "def", "dim", "einsum", "exp_blocked_to_pad", "f", "from_blocked_mask", "from_mask", "if", "int", "is", "length", "multiple", "must", "of", "raise", "return", "seq_length", "sequence", "size", "to_blocked_mask", "to_mask", "torch", "unsqueeze_", "view", "while"], "bigbird_pegasus/modeling_bigbird_pegasus.py:BigBirdPegasusEncoder._pad_to_block_size": ["Input", "ModelEncoder", "ModelPreTrainedModel", "Tensor", "_pad_to_block_size", "a", "are", "attention_mask", "automatically", "batch_size", "be", "block_size", "cat", "class", "config", "def", "device", "dim", "dtype", "embed_tokens", "f", "from", "functional", "hidden_states", "ids", "if", "input_ids_padding", "inputs_embeds_padding", "logger", "long", "multiple", "nn", "of", "ones", "pad", "pad_id", "pad_token_id", "padded", "padding_len", "return", "self", "seq_len", "shape", "to", "torch", "value", "warning_once"], "bigbird_pegasus/modeling_bigbird_pegasus.py:BigBirdPegasusDecoder.__init__": ["False", "LayerNorm", "ModelConfig", "ModelDecoder", "ModelDecoderLayer", "ModelLearnedPositionalEmbedding", "ModelPreTrainedModel", "ModelScaledWordEmbedding", "ModuleList", "__init__", "class", "config", "d_model", "decoder_layerdrop", "decoder_layers", "def", "dropout", "else", "embed_positions", "embed_scale", "embed_tokens", "for", "gradient_checkpointing", "i", "if", "in", "layer_idx", "layerdrop", "layernorm_embedding", "layers", "math", "max_position_embeddings", "max_target_positions", "nn", "pad_token_id", "padding_idx", "post_init", "range", "scale_embedding", "self", "sqrt", "super", "vocab_size"], "bigbird_pegasus/modeling_bigbird_pegasus.py:BigBirdPegasusDecoder.forward": ["BaseModelOutputWithPastAndCrossAttentions", "Cache", "DynamicCache", "EncoderDecoderCache", "False", "ModelDecoder", "ModelPreTrainedModel", "None", "Setting", "Tensor", "True", "ValueError", "You", "all_cross_attentions", "all_hidden_states", "all_self_attns", "and", "arange", "at", "attention_mask", "attentions", "batch_size", "bool", "both", "cache_position", "cannot", "checkpointing", "class", "config", "continue", "create_bidirectional_mask", "create_causal_mask", "cross_attentions", "decoder_input_ids", "decoder_inputs_embeds", "decoder_layer", "def", "device", "dropout", "dropout_probability", "either", "elif", "else", "embed_positions", "embed_tokens", "encoder_attention_mask", "encoder_hidden_states", "enumerate", "for", "forward", "functional", "get_seq_length", "gradient", "gradient_checkpointing", "have", "hidden_states", "idx", "if", "in", "incompatible", "input", "input_embeds", "input_ids", "input_shape", "inputs_embeds", "is", "is_encoder_decoder", "is_torchdynamo_compiling", "isinstance", "kwargs", "last_hidden_state", "layer_outputs", "layerdrop", "layernorm_embedding", "layers", "logger", "mask_seq_length", "nn", "not", "ones", "or", "output_attentions", "output_hidden_states", "p", "past_key_values", "past_key_values_length", "position_ids", "positions", "r", "raise", "rand", "return", "return_dict", "same", "self", "self_attention_cache", "self_attn_cache", "seq_length", "size", "specify", "the", "time", "to", "torch", "training", "tuple", "use_cache", "use_return_dict", "v", "view", "warning_once", "with"], "bigbird_pegasus/modeling_bigbird_pegasus.py:BigBirdPegasusModel.__init__": ["ModelConfig", "ModelDecoder", "ModelEncoder", "ModelModel", "ModelPreTrainedModel", "ModelScaledWordEmbedding", "__init__", "class", "config", "d_model", "decoder", "def", "else", "embed_scale", "encoder", "if", "math", "pad_token_id", "padding_idx", "post_init", "scale_embedding", "self", "shared", "sqrt", "super", "vocab_size"], "bigbird_pegasus/modeling_bigbird_pegasus.py:BigBirdPegasusModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "return", "self", "shared"], "bigbird_pegasus/modeling_bigbird_pegasus.py:BigBirdPegasusModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "decoder", "def", "embed_tokens", "encoder", "self", "set_input_embeddings", "shared", "value"], "bigbird_pegasus/modeling_bigbird_pegasus.py:BigBirdPegasusModel.forward": ["BaseModelOutput", "Cache", "FloatTensor", "If", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Please", "Seq2SeqModelOutput", "Tensor", "ValueError", "and", "are", "attention_mask", "attentions", "be", "bool", "cache_position", "cannot", "class", "config", "cross_attentions", "decoder", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_input_ids", "decoder_inputs_embeds", "decoder_outputs", "decoder_start_token_id", "def", "either", "elif", "else", "encoder", "encoder_attention_mask", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "isinstance", "kwargs", "last_hidden_state", "len", "list", "no", "not", "or", "output_attentions", "output_hidden_states", "pad_token_id", "pass", "passed", "past_key_values", "r", "raise", "return", "return_dict", "self", "shift_tokens_right", "torch", "tuple", "use_cache", "use_return_dict"], "bigbird_pegasus/modeling_bigbird_pegasus.py:BigBirdPegasusForConditionalGeneration.__init__": ["False", "GenerationMixin", "Linear", "ModelConfig", "ModelForConditionalGeneration", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "d_model", "def", "final_logits_bias", "lm_head", "model", "nn", "num_embeddings", "post_init", "register_buffer", "self", "shared", "super", "torch", "zeros"], "bigbird_pegasus/modeling_bigbird_pegasus.py:BigBirdPegasusForConditionalGeneration.resize_token_embeddings": ["Embedding", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "True", "_resize_final_logits_bias", "bool", "class", "def", "int", "mean_resizing", "new_embeddings", "new_num_tokens", "nn", "pad_to_multiple_of", "resize_token_embeddings", "return", "self", "shape", "super", "weight"], "bigbird_pegasus/modeling_bigbird_pegasus.py:BigBirdPegasusForConditionalGeneration._resize_final_logits_bias": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "_resize_final_logits_bias", "cat", "class", "def", "device", "dim", "else", "extra_bias", "final_logits_bias", "if", "int", "new_bias", "new_num_tokens", "old_num_tokens", "register_buffer", "self", "shape", "torch", "zeros"], "bigbird_pegasus/modeling_bigbird_pegasus.py:BigBirdPegasusForConditionalGeneration.forward": ["Cache", "CrossEntropyLoss", "False", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Seq2SeqLMOutput", "Tensor", "The", "and", "argument", "attention_mask", "bool", "cache_position", "changed", "class", "config", "cross_attentions", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_input_ids", "decoder_inputs_embeds", "decoder_start_token_id", "def", "device", "else", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "final_logits_bias", "forward", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "list", "lm_head", "lm_logits", "logger", "logits", "loss", "loss_fct", "masked_lm_loss", "model", "not", "output", "output_attentions", "output_hidden_states", "outputs", "pad_token_id", "past_key_values", "provided", "r", "return", "return_dict", "self", "shift_tokens_right", "since", "to", "torch", "tuple", "use_cache", "use_return_dict", "view", "vocab_size", "warning"], "bigbird_pegasus/modeling_bigbird_pegasus.py:BigBirdPegasusForConditionalGeneration.prepare_decoder_input_ids_from_labels": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "Tensor", "class", "config", "decoder_start_token_id", "def", "labels", "pad_token_id", "prepare_decoder_input_ids_from_labels", "return", "self", "shift_tokens_right", "torch"], "bigbird_pegasus/modeling_bigbird_pegasus.py:BigBirdPegasusForSequenceClassification.__init__": ["ModelClassificationHead", "ModelConfig", "ModelForSequenceClassification", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "classification_head", "classifier_dropout", "config", "d_model", "def", "kwargs", "model", "num_labels", "post_init", "self", "super"], "bigbird_pegasus/modeling_bigbird_pegasus.py:BigBirdPegasusForSequenceClassification.forward": ["All", "BCEWithLogitsLoss", "CrossEntropyLoss", "False", "FloatTensor", "LongTensor", "MSELoss", "ModelForSequenceClassification", "ModelPreTrainedModel", "None", "NotImplementedError", "Passing", "Seq2SeqSequenceClassifierOutput", "Tensor", "ValueError", "__class__", "__name__", "and", "attention_mask", "bool", "cache_position", "class", "classification_head", "config", "cross_attentions", "currently", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_input_ids", "decoder_inputs_embeds", "def", "device", "dtype", "elif", "else", "embeddings", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "eos", "eos_mask", "eos_token_id", "eq", "examples", "f", "for", "forward", "have", "hidden_states", "if", "input", "input_ids", "inputs_embeds", "int", "is", "kwargs", "labels", "len", "list", "logits", "long", "loss", "loss_fct", "model", "multi_label_classification", "must", "not", "num_labels", "number", "of", "or", "output", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "problem_type", "r", "raise", "regression", "return", "return_dict", "same", "self", "sentence_representation", "single_label_classification", "size", "squeeze", "sum", "supported", "the", "to", "tokens", "torch", "tuple", "unique_consecutive", "use_cache", "use_return_dict", "view"], "bigbird_pegasus/modeling_bigbird_pegasus.py:BigBirdPegasusForQuestionAnswering.__init__": ["Linear", "ModelForQuestionAnswering", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "config", "def", "hidden_size", "model", "nn", "num_labels", "post_init", "qa_outputs", "self", "super"], "bigbird_pegasus/modeling_bigbird_pegasus.py:BigBirdPegasusForQuestionAnswering.forward": ["CrossEntropyLoss", "False", "FloatTensor", "LongTensor", "ModelForQuestionAnswering", "ModelPreTrainedModel", "None", "Seq2SeqQuestionAnsweringModelOutput", "Tensor", "and", "attention_mask", "bool", "cache_position", "clamp", "class", "config", "contiguous", "cross_attentions", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_input_ids", "decoder_inputs_embeds", "def", "dim", "else", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "end_logits", "end_loss", "end_positions", "forward", "if", "ignore_index", "ignored_index", "input_ids", "inputs_embeds", "is", "kwargs", "len", "list", "logits", "loss", "loss_fct", "model", "not", "output", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "qa_outputs", "r", "return", "return_dict", "self", "sequence_output", "size", "split", "squeeze", "start_logits", "start_loss", "start_positions", "torch", "total_loss", "tuple", "use_cache", "use_return_dict"], "bigbird_pegasus/modeling_bigbird_pegasus.py:BigBirdPegasusDecoderWrapper.__init__": ["ModelDecoder", "ModelDecoderWrapper", "ModelPreTrainedModel", "__init__", "class", "config", "decoder", "def", "post_init", "self", "super"], "bigbird_pegasus/modeling_bigbird_pegasus.py:BigBirdPegasusDecoderWrapper.forward": ["ModelDecoderWrapper", "ModelPreTrainedModel", "args", "class", "decoder", "def", "forward", "kwargs", "return", "self"], "bigbird_pegasus/modeling_bigbird_pegasus.py:BigBirdPegasusForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "ModelDecoderWrapper", "ModelForCausalLM", "ModelPreTrainedModel", "True", "__init__", "bias", "class", "config", "def", "hidden_size", "is_decoder", "is_encoder_decoder", "lm_head", "model", "nn", "post_init", "self", "super", "vocab_size"], "bigbird_pegasus/modeling_bigbird_pegasus.py:BigBirdPegasusForCausalLM.get_input_embeddings": ["GenerationMixin", "ModelForCausalLM", "ModelPreTrainedModel", "class", "decoder", "def", "embed_tokens", "get_input_embeddings", "model", "return", "self"], "bigbird_pegasus/modeling_bigbird_pegasus.py:BigBirdPegasusForCausalLM.set_input_embeddings": ["GenerationMixin", "ModelForCausalLM", "ModelPreTrainedModel", "class", "decoder", "def", "embed_tokens", "model", "self", "set_input_embeddings", "value"], "bigbird_pegasus/modeling_bigbird_pegasus.py:BigBirdPegasusForCausalLM.forward": ["Cache", "CausalLMOutputWithCrossAttentions", "CrossEntropyLoss", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForCausalLM", "ModelPreTrainedModel", "None", "Tensor", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "cross_attentions", "decoder", "def", "else", "encoder_attention_mask", "encoder_hidden_states", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "lm_head", "logits", "logits_to_keep", "loss", "loss_fct", "model", "not", "output", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "r", "return", "return_dict", "self", "slice", "slice_indices", "torch", "tuple", "use_cache", "use_return_dict", "view", "vocab_size"], "biogpt/modeling_biogpt.py:BioGptLearnedPositionalEmbedding.__init__": ["Embedding", "ModelLearnedPositionalEmbedding", "__init__", "class", "def", "embedding_dim", "int", "nn", "num_embeddings", "offset", "self", "super"], "biogpt/modeling_biogpt.py:BioGptLearnedPositionalEmbedding.forward": ["Embedding", "LongTensor", "ModelLearnedPositionalEmbedding", "None", "attention_mask", "class", "cumsum", "def", "dim", "forward", "if", "int", "is", "long", "nn", "offset", "past_key_values_length", "position_ids", "return", "self", "super", "torch"], "biogpt/modeling_biogpt.py:BioGptScaledWordEmbedding.__init__": ["Embedding", "ModelScaledWordEmbedding", "None", "__init__", "class", "def", "embed_scale", "embedding_dim", "float", "int", "nn", "num_embeddings", "padding_idx", "self", "super"], "biogpt/modeling_biogpt.py:BioGptScaledWordEmbedding.forward": ["Embedding", "ModelScaledWordEmbedding", "Tensor", "class", "def", "embed_scale", "forward", "input_ids", "nn", "return", "self", "super", "torch"], "biogpt/modeling_biogpt.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "contiguous", "def", "dim", "dropout", "float", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "shape", "size", "softmax", "torch", "training", "transpose", "value"], "biogpt/modeling_biogpt.py:BioGptAttention.__init__": ["False", "Instantiating", "Linear", "ModelAttention", "ModelConfig", "Module", "None", "Please", "True", "ValueError", "__class__", "__init__", "__name__", "a", "and", "be", "bias", "bool", "by", "caching", "call", "class", "config", "creating", "decoder", "def", "divisible", "dropout", "during", "embed_dim", "errors", "f", "float", "forward", "got", "head_dim", "if", "int", "is", "is_causal", "is_decoder", "k_proj", "layer_idx", "lead", "logger", "make", "must", "nn", "not", "num_heads", "out_proj", "passing", "provide", "q_proj", "raise", "recommended", "scaling", "self", "super", "sure", "the", "this", "to", "used", "v_proj", "warning_once", "when", "will", "without"], "biogpt/modeling_biogpt.py:BioGptAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "EncoderDecoderCache", "False", "FlashAttentionKwargs", "ModelAttention", "Module", "None", "Tensor", "True", "Unpack", "_attn_implementation", "and", "attention_interface", "attention_mask", "attn_output", "attn_weights", "bool", "bsz", "cache_position", "class", "config", "contiguous", "cross_attention_cache", "curr_past_key_values", "current_states", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "get", "head_dim", "hidden_states", "if", "is", "is_cross_attention", "is_updated", "isinstance", "k_proj", "key_states", "key_value_states", "keys", "kv_input_shape", "kwargs", "layer_idx", "layers", "nn", "not", "out_proj", "output_attentions", "past_key_values", "q_input_shape", "q_proj", "query_states", "reshape", "return", "scaling", "self", "self_attention_cache", "shape", "src_len", "tgt_len", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "values", "view"], "biogpt/modeling_biogpt.py:BioGptDecoderLayer.__init__": ["ACT2FN", "GradientCheckpointingLayer", "LayerNorm", "Linear", "ModelAttention", "ModelConfig", "ModelDecoderLayer", "None", "True", "__init__", "activation_dropout", "activation_fn", "attention_probs_dropout_prob", "class", "config", "def", "dropout", "embed_dim", "fc1", "fc2", "final_layer_norm", "hidden_act", "hidden_dropout_prob", "hidden_size", "int", "intermediate_size", "is_causal", "is_decoder", "layer_idx", "nn", "num_attention_heads", "num_heads", "self", "self_attn", "self_attn_layer_norm", "super"], "biogpt/modeling_biogpt.py:BioGptDecoderLayer.forward": ["Cache", "False", "FloatTensor", "GradientCheckpointingLayer", "LongTensor", "ModelDecoderLayer", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "activation_dropout", "activation_fn", "attention_mask", "bool", "cache_position", "class", "def", "dropout", "fc1", "fc2", "final_layer_norm", "forward", "functional", "hidden_states", "if", "kwargs", "nn", "output_attentions", "outputs", "p", "past_key_values", "position_ids", "residual", "return", "self", "self_attn", "self_attn_layer_norm", "self_attn_weights", "torch", "training", "tuple", "use_cache"], "biogpt/modeling_biogpt.py:BioGptModel.__init__": ["False", "LayerNorm", "ModelConfig", "ModelDecoderLayer", "ModelLearnedPositionalEmbedding", "ModelModel", "ModelPreTrainedModel", "ModelScaledWordEmbedding", "ModuleList", "__init__", "class", "config", "def", "dropout", "else", "embed_dim", "embed_positions", "embed_scale", "embed_tokens", "for", "gradient_checkpointing", "hidden_dropout_prob", "hidden_size", "i", "if", "in", "layer_idx", "layer_norm", "layerdrop", "layers", "math", "max_position_embeddings", "nn", "num_hidden_layers", "pad_token_id", "padding_idx", "post_init", "range", "scale_embedding", "self", "sqrt", "super", "vocab_size"], "biogpt/modeling_biogpt.py:BioGptModel.forward": ["BaseModelOutputWithPastAndCrossAttentions", "Cache", "DynamicCache", "False", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Setting", "Tensor", "TransformersKwargs", "True", "Unpack", "ValueError", "You", "all_cross_attentions", "all_hidden_states", "all_self_attns", "and", "arange", "at", "attention_mask", "attentions", "batch_size", "bool", "both", "cache_position", "cannot", "causal_mask", "checkpointing", "class", "config", "continue", "create_causal_mask", "cross_attentions", "cumsum", "decoder_input_ids", "decoder_inputs_embeds", "decoder_layer", "def", "device", "dim", "dropout", "dropout_probability", "either", "elif", "else", "embed_positions", "embed_tokens", "enumerate", "for", "forward", "functional", "get_seq_length", "gradient", "gradient_checkpointing", "have", "hidden_states", "idx", "if", "in", "incompatible", "input", "input_embeds", "input_ids", "input_shape", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layer_norm", "layer_outputs", "layerdrop", "layers", "logger", "long", "mask_seq_length", "nn", "not", "ones", "or", "output_attentions", "output_hidden_states", "p", "past_key_values", "past_key_values_length", "position_ids", "positions", "raise", "rand", "return", "return_dict", "same", "self", "self_attn_cache", "seq_length", "shape", "size", "specify", "the", "time", "to", "torch", "training", "tuple", "use_cache", "use_return_dict", "v", "view", "warning_once", "with"], "biogpt/modeling_biogpt.py:BioGptForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "Model", "ModelForCausalLM", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "nn", "output_projection", "post_init", "self", "super", "vocab_size"], "biogpt/modeling_biogpt.py:BioGptForCausalLM.get_output_embeddings": ["GenerationMixin", "ModelForCausalLM", "ModelPreTrainedModel", "class", "def", "get_output_embeddings", "output_projection", "return", "self"], "biogpt/modeling_biogpt.py:BioGptForCausalLM.set_output_embeddings": ["GenerationMixin", "ModelForCausalLM", "ModelPreTrainedModel", "class", "def", "new_embeddings", "output_projection", "self", "set_output_embeddings"], "biogpt/modeling_biogpt.py:BioGptForCausalLM.forward": ["Cache", "CausalLMOutputWithCrossAttentions", "FloatTensor", "GenerationMixin", "LongTensor", "Model", "ModelForCausalLM", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "cross_attentions", "def", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "logits", "logits_to_keep", "loss", "loss_function", "not", "output", "output_attentions", "output_hidden_states", "output_projection", "outputs", "past_key_values", "position_ids", "r", "return", "return_dict", "self", "slice", "slice_indices", "torch", "tuple", "use_cache", "use_return_dict", "vocab_size"], "biogpt/modeling_biogpt.py:BioGptForTokenClassification.__init__": ["Dropout", "Linear", "Model", "ModelForTokenClassification", "ModelModel", "ModelPreTrainedModel", "None", "__init__", "and", "class", "classifier", "classifier_dropout", "config", "def", "dropout", "else", "hasattr", "hidden_dropout_prob", "hidden_size", "if", "is", "nn", "not", "num_labels", "post_init", "self", "super"], "biogpt/modeling_biogpt.py:BioGptForTokenClassification.forward": ["Cache", "CrossEntropyLoss", "FloatTensor", "LongTensor", "Model", "ModelForTokenClassification", "ModelPreTrainedModel", "None", "Tensor", "TokenClassifierOutput", "active_labels", "active_logits", "active_loss", "attention_mask", "attentions", "bool", "cache_position", "class", "classifier", "config", "def", "dropout", "else", "forward", "hidden_states", "if", "ignore_index", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "not", "num_labels", "output", "output_attentions", "output_hidden_states", "past_key_values", "position_ids", "r", "return", "return_dict", "self", "tensor", "token_type_ids", "torch", "transformer_outputs", "tuple", "type_as", "use_cache", "use_return_dict", "view", "where"], "biogpt/modeling_biogpt.py:BioGptForSequenceClassification.__init__": ["False", "Linear", "Model", "ModelConfig", "ModelForSequenceClassification", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "nn", "num_labels", "post_init", "score", "self", "super"], "biogpt/modeling_biogpt.py:BioGptForSequenceClassification.forward": ["BCEWithLogitsLoss", "Cache", "CrossEntropyLoss", "FloatTensor", "LongTensor", "MSELoss", "Model", "ModelForSequenceClassification", "ModelPreTrainedModel", "None", "Results", "SequenceClassifierOutputWithPast", "Tensor", "__class__", "__name__", "and", "arange", "attention_mask", "attentions", "batch_size", "be", "bool", "cache_position", "class", "config", "conjunction", "def", "detect", "device", "dtype", "elif", "else", "f", "forward", "hidden_states", "if", "in", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "logger", "logits", "logits_to_keep", "long", "loss", "loss_fct", "may", "multi_label_classification", "ne", "not", "num_labels", "or", "output", "output_attentions", "output_hidden_states", "pad_token_id", "padding", "past_key_values", "pooled_logits", "position_ids", "problem_type", "r", "regression", "return", "return_dict", "score", "self", "sequence_length", "shape", "single_label_classification", "slice", "slice_indices", "squeeze", "sum", "to", "tokens", "torch", "transformer_outputs", "tuple", "unexpected", "use_cache", "use_return_dict", "using", "view", "warning_once", "will", "with"], "biogpt/modeling_biogpt.py:BioGptForSequenceClassification.get_input_embeddings": ["Model", "ModelForSequenceClassification", "ModelPreTrainedModel", "class", "def", "embed_tokens", "get_input_embeddings", "return", "self"], "biogpt/modeling_biogpt.py:BioGptForSequenceClassification.set_input_embeddings": ["Model", "ModelForSequenceClassification", "ModelPreTrainedModel", "class", "def", "embed_tokens", "self", "set_input_embeddings", "value"], "bit/modeling_bit.py:get_padding_value": ["False", "Model_padding_value", "None", "True", "and", "bool", "def", "dilation", "dynamic", "elif", "else", "if", "is", "isinstance", "kernel_size", "lower", "padding", "r", "return", "same", "str", "stride", "tuple", "valid"], "bit/modeling_bit.py:WeightStandardizedConv2d.__init__": ["Conv2d", "DynamicPad2d", "False", "ModelStandardizedConv2d", "None", "SAME", "__init__", "bias", "class", "def", "dilation", "else", "eps", "get_padding_value", "groups", "if", "in_channel", "is_dynamic", "kernel_size", "nn", "out_channels", "pad", "padding", "self", "stride", "super"], "bit/modeling_bit.py:WeightStandardizedConv2d.forward": ["Conv2d", "Model", "ModelStandardizedConv2d", "None", "True", "batch_norm", "bias", "class", "conv2d", "def", "dilation", "eps", "forward", "functional", "groups", "hidden_state", "if", "is", "momentum", "nn", "not", "out_channels", "pad", "padding", "reshape", "reshape_as", "return", "self", "stride", "training"], "bit/modeling_bit.py:BitGroupNormActivation.__init__": ["ACT2FN", "GroupNorm", "Identity", "ModelGroupNormActivation", "True", "__init__", "activation", "affine", "apply_activation", "class", "config", "def", "else", "eps", "hidden_act", "if", "nn", "num_channels", "num_groups", "self", "super"], "bit/modeling_bit.py:BitGroupNormActivation.forward": ["GroupNorm", "ModelGroupNormActivation", "activation", "bias", "class", "def", "eps", "forward", "functional", "group_norm", "hidden_state", "nn", "num_groups", "return", "self", "weight"], "bit/modeling_bit.py:DynamicPad2d.__init__": ["ModelPad2d", "Module", "__init__", "ceil", "class", "compute_padding", "def", "dilation", "if", "int", "isinstance", "kernel_size", "math", "max", "nn", "return", "self", "stride", "super", "value", "x"], "bit/modeling_bit.py:DynamicPad2d.forward": ["ModelPad2d", "Module", "class", "compute_padding", "def", "dilation", "forward", "functional", "if", "input", "input_height", "input_width", "kernel_size", "nn", "or", "pad", "padding_height", "padding_width", "return", "self", "size", "stride", "value"], "bit/modeling_bit.py:BitMaxPool2d.__init__": ["DynamicPad2d", "False", "Identity", "Iterable", "MaxPool2d", "ModelMaxPool2d", "None", "True", "__init__", "abc", "ceil_mode", "class", "collections", "def", "dilation", "else", "if", "int", "isinstance", "kernel_size", "nn", "pad", "padding", "padding_value", "self", "stride", "super", "use_dynamic_padding"], "bit/modeling_bit.py:BitMaxPool2d.forward": ["MaxPool2d", "ModelMaxPool2d", "ceil_mode", "class", "def", "dilation", "forward", "functional", "hidden_states", "kernel_size", "max_pool2d", "nn", "pad", "padding", "return", "self", "stride"], "bit/modeling_bit.py:BitEmbeddings.__init__": ["ConstantPad2d", "Identity", "ModelConfig", "ModelEmbeddings", "ModelGroupNormActivation", "ModelMaxPool2d", "Module", "None", "SAME", "WeightStandardizedConv2d", "__init__", "and", "class", "config", "convolution", "def", "else", "embedding_dynamic_padding", "embedding_size", "eps", "global_padding", "if", "is", "kernel_size", "layer_type", "nn", "norm", "not", "num_channels", "pad", "padding", "pooler", "preactivation", "self", "stride", "super", "upper", "use_dynamic_padding", "value"], "bit/modeling_bit.py:BitEmbeddings.forward": ["Make", "ModelEmbeddings", "Module", "Tensor", "ValueError", "channel", "class", "configuration", "convolution", "def", "dimension", "embedding", "forward", "if", "in", "match", "nn", "norm", "num_channels", "of", "one", "pad", "pixel", "pixel_values", "pooler", "raise", "return", "self", "set", "shape", "sure", "that", "the", "values", "with"], "bit/modeling_bit.py:drop_path": ["False", "Model_path", "Model_prob", "Tensor", "bool", "def", "device", "div", "dtype", "float", "floor_", "if", "input", "keep_prob", "ndim", "not", "or", "output", "rand", "random_tensor", "return", "shape", "torch", "training"], "bit/modeling_bit.py:BitDropPath.__init__": ["ModelDropPath", "Module", "None", "__init__", "class", "def", "drop_prob", "float", "nn", "self", "super"], "bit/modeling_bit.py:BitDropPath.forward": ["ModelDropPath", "Module", "Tensor", "class", "def", "drop_path", "drop_prob", "forward", "hidden_states", "nn", "return", "self", "torch", "training"], "bit/modeling_bit.py:BitDropPath.extra_repr": ["ModelDropPath", "Module", "class", "def", "drop_prob", "extra_repr", "f", "nn", "p", "return", "self", "str"], "bit/modeling_bit.py:make_div": ["Model_div", "def", "divisor", "if", "int", "max", "min_value", "new_value", "return", "value"], "bit/modeling_bit.py:BitPreActivationBottleneckLayer.__init__": ["False", "Identity", "ModelDownsampleConv", "ModelDropPath", "ModelGroupNormActivation", "ModelPreActivationBottleneckLayer", "Module", "None", "True", "WeightStandardizedConv2d", "__init__", "bottle_ratio", "class", "config", "conv1", "conv2", "conv3", "def", "dilation", "downsample", "drop_path", "drop_path_rate", "else", "eps", "first_dilation", "global_padding", "groups", "if", "in_channels", "is_first_layer", "make_div", "mid_channels", "nn", "norm1", "norm2", "norm3", "num_channels", "or", "out_channels", "padding", "preact", "self", "stride", "super"], "bit/modeling_bit.py:BitPreActivationBottleneckLayer.forward": ["ModelPreActivationBottleneckLayer", "Module", "None", "class", "conv1", "conv2", "conv3", "def", "downsample", "drop_path", "forward", "hidden_states", "hidden_states_preact", "if", "is", "nn", "norm1", "norm2", "norm3", "not", "return", "self", "shortcut"], "bit/modeling_bit.py:BitBottleneckLayer.__init__": ["ACT2FN", "False", "Identity", "ModelBottleneckLayer", "ModelDownsampleConv", "ModelDropPath", "ModelGroupNormActivation", "Module", "None", "WeightStandardizedConv2d", "__init__", "activation", "apply_activation", "bottle_ratio", "class", "config", "conv1", "conv2", "conv3", "def", "dilation", "downsample", "drop_path", "drop_path_rate", "else", "eps", "first_dilation", "global_padding", "groups", "hidden_act", "if", "in_channels", "is_first_layer", "make_div", "mid_chs", "nn", "norm1", "norm2", "norm3", "num_channels", "or", "out_channels", "padding", "preact", "self", "stride", "super"], "bit/modeling_bit.py:BitBottleneckLayer.forward": ["ModelBottleneckLayer", "Module", "None", "activation", "class", "conv1", "conv2", "conv3", "def", "downsample", "drop_path", "forward", "hidden_states", "if", "is", "nn", "norm1", "norm2", "norm3", "not", "return", "self", "shortcut"], "bit/modeling_bit.py:BitDownsampleConv.__init__": ["False", "Identity", "ModelDownsampleConv", "ModelGroupNormActivation", "Module", "True", "WeightStandardizedConv2d", "__init__", "apply_activation", "class", "config", "conv", "def", "else", "eps", "global_padding", "if", "in_channels", "nn", "norm", "num_channels", "out_channels", "padding", "preact", "self", "stride", "super"], "bit/modeling_bit.py:BitDownsampleConv.forward": ["ModelDownsampleConv", "Module", "class", "conv", "def", "forward", "nn", "norm", "return", "self", "x"], "bit/modeling_bit.py:BitStage.__init__": ["ModelBottleneckLayer", "ModelPreActivationBottleneckLayer", "ModelStage", "Module", "None", "Sequential", "__init__", "_get_updated_hyperparameters", "add_module", "bottle_ratio", "bottleneck", "class", "config", "def", "depth", "dilation", "drop_path_rate", "else", "first_dilation", "for", "if", "in", "in_channels", "is_first_layer", "layer_cls", "layer_dropout", "layer_idx", "layer_type", "layers", "nn", "out_channels", "prev_chs", "range", "self", "str", "stride", "super"], "bit/modeling_bit.py:BitStage._get_updated_hyperparameters": ["ModelStage", "Module", "_get_updated_hyperparameters", "class", "def", "drop_path_rate", "else", "if", "is_first_layer", "layer_dropout", "layer_idx", "nn", "r", "return", "self", "stride"], "bit/modeling_bit.py:BitStage.forward": ["ModelStage", "Module", "Tensor", "_", "class", "def", "enumerate", "for", "forward", "hidden_state", "in", "input", "layer", "layers", "nn", "return", "self"], "bit/modeling_bit.py:BitEncoder.__init__": ["ModelConfig", "ModelEncoder", "ModelStage", "Module", "ModuleList", "Tensor", "__init__", "_get_updated_hyperparameters", "add_module", "class", "config", "current_depth", "current_hidden_size", "current_stride", "def", "depth", "depths", "dilation", "drop_path_rate", "embedding_size", "enumerate", "for", "hidden_sizes", "in", "layer_dropout", "layer_dropouts", "linspace", "nn", "np", "out_channels", "prev_chs", "self", "split", "stage", "stage_idx", "stages", "str", "stride", "sum", "super", "tolist", "torch", "x", "zip"], "bit/modeling_bit.py:BitEncoder._get_updated_hyperparameters": ["ModelEncoder", "Module", "_get_updated_hyperparameters", "class", "config", "current_hidden_size", "current_stride", "def", "dilation", "else", "if", "make_div", "nn", "out_channels", "output_stride", "return", "self", "stage_idx", "stride", "width_factor"], "bit/modeling_bit.py:BitEncoder.forward": ["BaseModelOutputWithNoAttention", "False", "ModelEncoder", "Module", "None", "Tensor", "True", "bool", "class", "def", "else", "for", "forward", "hidden_state", "hidden_states", "if", "in", "is", "last_hidden_state", "nn", "not", "output_hidden_states", "return", "return_dict", "self", "stage_module", "stages", "tuple", "v"], "bit/modeling_bit.py:BitPreTrainedModel._init_weights": ["BatchNorm2d", "Conv2d", "GroupNorm", "Linear", "ModelPreTrainedModel", "None", "PreTrainedModel", "_", "_calculate_fan_in_and_fan_out", "_init_weights", "a", "bias", "bound", "class", "constant_", "def", "elif", "else", "fan_in", "fan_out", "getattr", "if", "init", "is", "isinstance", "kaiming_normal_", "kaiming_uniform_", "math", "mode", "module", "nn", "nonlinearity", "not", "num_batches_tracked", "ones_", "relu", "running_mean", "running_var", "self", "sqrt", "torch", "uniform_", "weight", "zeros_"], "bit/modeling_bit.py:BitModel.__init__": ["AdaptiveAvgPool2d", "Identity", "ModelEmbeddings", "ModelEncoder", "ModelGroupNormActivation", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "config", "def", "else", "embedder", "encoder", "hidden_sizes", "if", "layer_type", "nn", "norm", "num_channels", "pooler", "post_init", "preactivation", "self", "super"], "bit/modeling_bit.py:BitModel.forward": ["BaseModelOutputWithPoolingAndNoAttention", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "bool", "class", "config", "def", "else", "embedder", "embedding_output", "encoder", "encoder_outputs", "forward", "hidden_states", "if", "is", "kwargs", "last_hidden_state", "norm", "not", "output_hidden_states", "pixel_values", "pooled_output", "pooler", "pooler_output", "return", "return_dict", "self", "use_return_dict"], "bit/modeling_bit.py:BitForImageClassification.__init__": ["Flatten", "Identity", "Linear", "Model", "ModelForImageClassification", "ModelModel", "ModelPreTrainedModel", "Sequential", "__init__", "class", "classifier", "config", "def", "else", "hidden_sizes", "if", "nn", "num_labels", "post_init", "self", "super"], "bit/modeling_bit.py:BitForImageClassification.forward": ["FloatTensor", "ImageClassifierOutputWithNoAttention", "LongTensor", "Model", "ModelForImageClassification", "ModelPreTrainedModel", "None", "bool", "class", "classifier", "config", "def", "else", "forward", "hidden_states", "if", "is", "kwargs", "labels", "logits", "loss", "loss_function", "not", "output", "output_hidden_states", "outputs", "pixel_values", "pooled_output", "pooler_output", "r", "return", "return_dict", "self", "torch", "use_return_dict"], "bit/modeling_bit.py:BitBackbone.__init__": ["BackboneMixin", "Model", "ModelBackbone", "ModelModel", "ModelPreTrainedModel", "__init__", "_init_backbone", "class", "config", "def", "embedding_size", "hidden_sizes", "num_features", "post_init", "self", "super"], "bit/modeling_bit.py:BitBackbone.forward": ["BackboneMixin", "BackboneOutput", "Model", "ModelBackbone", "ModelPreTrainedModel", "None", "Tensor", "True", "attentions", "bool", "class", "config", "def", "else", "enumerate", "feature_maps", "for", "forward", "hidden_states", "idx", "if", "in", "is", "kwargs", "not", "out_features", "output", "output_hidden_states", "outputs", "pixel_values", "r", "return", "return_dict", "self", "stage", "stage_names", "use_return_dict"], "bitnet/modeling_bitnet.py:BitNetRMSNorm.__init__": ["ModelRMSNorm", "Module", "Parameter", "__init__", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "bitnet/modeling_bitnet.py:BitNetRMSNorm.forward": ["ModelRMSNorm", "Module", "True", "class", "def", "dtype", "float32", "forward", "hidden_states", "input_dtype", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "bitnet/modeling_bitnet.py:BitNetRMSNorm.extra_repr": ["ModelRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "variance_epsilon", "weight"], "bitnet/modeling_bitnet.py:BitNetMLP.__init__": ["ACT2FN", "False", "Linear", "ModelConfig", "ModelMLP", "ModelRMSNorm", "Module", "__init__", "act_fn", "bias", "class", "config", "def", "down_proj", "eps", "ffn_sub_norm", "gate_proj", "hidden_act", "hidden_size", "intermediate_size", "nn", "rms_norm_eps", "self", "super", "up_proj"], "bitnet/modeling_bitnet.py:BitNetMLP.forward": ["ModelMLP", "Module", "act_fn", "class", "def", "down_proj", "ffn_sub_norm", "forward", "gate_proj", "nn", "return", "self", "up_proj", "x"], "bitnet/modeling_bitnet.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "bitnet/modeling_bitnet.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cos", "def", "k", "k_embed", "q", "q_embed", "return", "rotate_half", "sin", "unsqueeze", "unsqueeze_dim"], "bitnet/modeling_bitnet.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "bitnet/modeling_bitnet.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value", "value_states"], "bitnet/modeling_bitnet.py:BitNetAttention.__init__": ["Linear", "ModelAttention", "ModelConfig", "ModelRMSNorm", "Module", "True", "__init__", "attention_bias", "attention_dropout", "attn_sub_norm", "bias", "class", "config", "def", "eps", "getattr", "head_dim", "hidden_size", "int", "is_causal", "k_proj", "layer_idx", "nn", "num_attention_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "q_proj", "rms_norm_eps", "scaling", "self", "super", "v_proj"], "bitnet/modeling_bitnet.py:BitNetAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "FlashAttentionKwargs", "LongTensor", "ModelAttention", "Module", "None", "Tensor", "Unpack", "_attn_implementation", "apply_rotary_pos_emb", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_sub_norm", "attn_weights", "cache_kwargs", "cache_position", "class", "config", "contiguous", "cos", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "head_dim", "hidden_shape", "hidden_states", "if", "input_shape", "is", "k_proj", "key_states", "kwargs", "layer_idx", "nn", "not", "o_proj", "past_key_values", "position_embeddings", "q_proj", "query_states", "reshape", "return", "scaling", "self", "shape", "sin", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "view"], "bitnet/modeling_bitnet.py:BitNetDecoderLayer.__init__": ["GradientCheckpointingLayer", "ModelAttention", "ModelConfig", "ModelDecoderLayer", "ModelMLP", "ModelRMSNorm", "__init__", "class", "config", "def", "eps", "hidden_size", "input_layernorm", "int", "layer_idx", "mlp", "post_attention_layernorm", "rms_norm_eps", "self", "self_attn", "super"], "bitnet/modeling_bitnet.py:BitNetDecoderLayer.forward": ["Cache", "False", "GradientCheckpointingLayer", "LongTensor", "ModelDecoderLayer", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "attention_mask", "bool", "cache_position", "class", "def", "forward", "hidden_states", "input_layernorm", "kwargs", "mlp", "past_key_values", "position_embeddings", "position_ids", "post_attention_layernorm", "residual", "return", "self", "self_attn", "torch", "tuple", "use_cache"], "bitnet/modeling_bitnet.py:BitNetRotaryEmbedding.__init__": ["Callable", "False", "ModelConfig", "ModelRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "bitnet/modeling_bitnet.py:BitNetRotaryEmbedding.compute_default_rope_parameters": ["ModelConfig", "ModelRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "bitnet/modeling_bitnet.py:BitNetRotaryEmbedding.forward": ["False", "ModelRotaryEmbedding", "Module", "None", "and", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "bitnet/modeling_bitnet.py:BitNetModel.__init__": ["Embedding", "False", "ModelConfig", "ModelDecoderLayer", "ModelModel", "ModelPreTrainedModel", "ModelRMSNorm", "ModelRotaryEmbedding", "ModuleList", "__init__", "class", "config", "def", "embed_tokens", "eps", "for", "gradient_checkpointing", "hidden_size", "in", "layer_idx", "layers", "nn", "norm", "num_hidden_layers", "pad_token_id", "padding_idx", "post_init", "range", "rms_norm_eps", "rotary_emb", "self", "super", "vocab_size"], "bitnet/modeling_bitnet.py:BitNetModel.forward": ["BaseModelOutputWithPast", "Cache", "DynamicCache", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "and", "arange", "attention_mask", "bool", "cache_position", "causal_mask", "class", "config", "create_causal_mask", "decoder_layer", "def", "device", "else", "embed_tokens", "exactly", "for", "forward", "get_seq_length", "hidden_states", "if", "in", "input_embeds", "input_ids", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layers", "must", "norm", "not", "num_hidden_layers", "of", "one", "or", "past_key_values", "past_seen_tokens", "position_embeddings", "position_ids", "raise", "return", "rotary_emb", "self", "shape", "specify", "torch", "unsqueeze", "use_cache"], "bitnet/modeling_bitnet.py:BitNetForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "ModelForCausalLM", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "vocab_size"], "bitnet/modeling_bitnet.py:BitNetForCausalLM.forward": ["BaseModelOutputWithPast", "Cache", "CausalLMOutputWithPast", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForCausalLM", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "outputs", "past_key_values", "position_ids", "r", "return", "self", "slice", "slice_indices", "torch", "use_cache", "vocab_size"], "blenderbot/modeling_blenderbot.py:shift_tokens_right": ["Model_tokens_right", "Modeled_input_ids", "None", "Tensor", "ValueError", "be", "clone", "config", "decoder_start_token_id", "def", "defined", "has", "if", "input_ids", "int", "is", "masked_fill_", "model", "new_zeros", "pad_token_id", "raise", "return", "self", "shape", "to", "torch"], "blenderbot/modeling_blenderbot.py:BlenderbotLearnedPositionalEmbedding.__init__": ["Embedding", "ModelLearnedPositionalEmbedding", "__init__", "class", "def", "embedding_dim", "int", "nn", "num_embeddings", "self", "super"], "blenderbot/modeling_blenderbot.py:BlenderbotLearnedPositionalEmbedding.forward": ["Embedding", "ModelLearnedPositionalEmbedding", "None", "Size", "Tensor", "arange", "bsz", "class", "def", "device", "dtype", "forward", "if", "input_ids_shape", "int", "is", "long", "nn", "past_key_values_length", "position_ids", "return", "self", "seq_len", "super", "torch", "weight"], "blenderbot/modeling_blenderbot.py:BlenderbotScaledWordEmbedding.__init__": ["Embedding", "ModelScaledWordEmbedding", "None", "__init__", "class", "def", "embed_scale", "embedding_dim", "float", "int", "nn", "num_embeddings", "padding_idx", "self", "super"], "blenderbot/modeling_blenderbot.py:BlenderbotScaledWordEmbedding.forward": ["Embedding", "ModelScaledWordEmbedding", "Tensor", "class", "def", "embed_scale", "forward", "input_ids", "nn", "return", "self", "super", "torch"], "blenderbot/modeling_blenderbot.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "contiguous", "def", "dim", "dropout", "float", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "shape", "size", "softmax", "torch", "training", "transpose", "value"], "blenderbot/modeling_blenderbot.py:BlenderbotAttention.__init__": ["False", "Instantiating", "Linear", "ModelAttention", "ModelConfig", "Module", "None", "Please", "True", "ValueError", "__class__", "__init__", "__name__", "a", "and", "be", "bias", "bool", "by", "caching", "call", "class", "config", "creating", "decoder", "def", "divisible", "dropout", "during", "embed_dim", "errors", "f", "float", "forward", "got", "head_dim", "if", "int", "is", "is_causal", "is_decoder", "k_proj", "layer_idx", "lead", "logger", "make", "must", "nn", "not", "num_heads", "out_proj", "passing", "provide", "q_proj", "raise", "recommended", "scaling", "self", "super", "sure", "the", "this", "to", "used", "v_proj", "warning_once", "when", "will", "without"], "blenderbot/modeling_blenderbot.py:BlenderbotAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "EncoderDecoderCache", "False", "FlashAttentionKwargs", "ModelAttention", "Module", "None", "Tensor", "True", "Unpack", "_attn_implementation", "and", "attention_interface", "attention_mask", "attn_output", "attn_weights", "bool", "bsz", "cache_position", "class", "config", "contiguous", "cross_attention_cache", "curr_past_key_values", "current_states", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "get", "head_dim", "hidden_states", "if", "is", "is_cross_attention", "is_updated", "isinstance", "k_proj", "key_states", "key_value_states", "keys", "kv_input_shape", "kwargs", "layer_idx", "layers", "nn", "not", "out_proj", "output_attentions", "past_key_values", "q_input_shape", "q_proj", "query_states", "reshape", "return", "scaling", "self", "self_attention_cache", "shape", "src_len", "tgt_len", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "values", "view"], "blenderbot/modeling_blenderbot.py:BlenderbotEncoderLayer.__init__": ["ACT2FN", "GradientCheckpointingLayer", "LayerNorm", "Linear", "ModelAttention", "ModelConfig", "ModelEncoderLayer", "__init__", "activation_dropout", "activation_fn", "activation_function", "attention_dropout", "class", "config", "d_model", "def", "dropout", "embed_dim", "encoder_attention_heads", "encoder_ffn_dim", "fc1", "fc2", "final_layer_norm", "nn", "num_heads", "self", "self_attn", "self_attn_layer_norm", "super"], "blenderbot/modeling_blenderbot.py:BlenderbotEncoderLayer.forward": ["False", "GradientCheckpointingLayer", "ModelEncoderLayer", "Tensor", "activation_dropout", "activation_fn", "attention_mask", "attn_weights", "bool", "clamp", "clamp_value", "class", "def", "dropout", "dtype", "fc1", "fc2", "final_layer_norm", "finfo", "float16", "forward", "functional", "hidden_states", "if", "max", "min", "nn", "output_attentions", "p", "residual", "return", "self", "self_attn", "self_attn_layer_norm", "torch", "training"], "blenderbot/modeling_blenderbot.py:BlenderbotDecoderLayer.__init__": ["ACT2FN", "GradientCheckpointingLayer", "LayerNorm", "Linear", "ModelAttention", "ModelConfig", "ModelDecoderLayer", "None", "True", "__init__", "activation_dropout", "activation_fn", "activation_function", "attention_dropout", "class", "config", "d_model", "decoder_attention_heads", "decoder_ffn_dim", "def", "dropout", "embed_dim", "encoder_attn", "encoder_attn_layer_norm", "fc1", "fc2", "final_layer_norm", "int", "is_causal", "is_decoder", "layer_idx", "nn", "num_heads", "self", "self_attn", "self_attn_layer_norm", "super"], "blenderbot/modeling_blenderbot.py:BlenderbotDecoderLayer.forward": ["Cache", "False", "GradientCheckpointingLayer", "ModelDecoderLayer", "None", "Tensor", "True", "activation_dropout", "activation_fn", "attention_mask", "bool", "cache_position", "class", "cross_attn_weights", "def", "dropout", "encoder_attention_mask", "encoder_attn", "encoder_attn_layer_norm", "encoder_hidden_states", "fc1", "fc2", "final_layer_norm", "forward", "functional", "hidden_states", "if", "is", "key_value_states", "nn", "not", "output_attentions", "outputs", "p", "past_key_values", "residual", "return", "self", "self_attn", "self_attn_layer_norm", "self_attn_weights", "torch", "training", "use_cache"], "blenderbot/modeling_blenderbot.py:BlenderbotPreTrainedModel._init_weights": ["ModelForConditionalGeneration", "ModelPreTrainedModel", "PreTrainedModel", "_init_weights", "class", "def", "final_logits_bias", "if", "init", "isinstance", "module", "self", "super", "zeros_"], "blenderbot/modeling_blenderbot.py:BlenderbotPreTrainedModel.dummy_inputs": ["ModelPreTrainedModel", "PreTrainedModel", "attention_mask", "class", "config", "decoder_input_ids", "def", "device", "dummy_inputs", "input_ids", "ne", "pad_token", "pad_token_id", "return", "self", "tensor", "torch"], "blenderbot/modeling_blenderbot.py:BlenderbotEncoder.__init__": ["False", "LayerNorm", "ModelConfig", "ModelEncoder", "ModelEncoderLayer", "ModelLearnedPositionalEmbedding", "ModelPreTrainedModel", "ModelScaledWordEmbedding", "ModuleList", "_", "__init__", "class", "config", "d_model", "def", "dropout", "else", "embed_dim", "embed_positions", "embed_scale", "embed_tokens", "encoder_layerdrop", "encoder_layers", "for", "gradient_checkpointing", "if", "in", "layer_norm", "layerdrop", "layers", "math", "max_position_embeddings", "max_source_positions", "nn", "pad_token_id", "padding_idx", "post_init", "range", "scale_embedding", "self", "sqrt", "super", "vocab_size"], "blenderbot/modeling_blenderbot.py:BlenderbotEncoder.forward": ["BaseModelOutput", "False", "ModelEncoder", "ModelPreTrainedModel", "None", "True", "ValueError", "You", "all_attentions", "and", "at", "attention_mask", "attentions", "both", "cannot", "class", "config", "create_bidirectional_mask", "def", "dropout", "dropout_probability", "either", "elif", "else", "embed_pos", "embed_positions", "embed_tokens", "encoder_layer", "encoder_states", "enumerate", "for", "forward", "functional", "have", "hidden_states", "idx", "if", "in", "input_embeds", "input_ids", "input_shape", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layer_norm", "layer_outputs", "layerdrop", "layers", "nn", "not", "or", "output_attentions", "output_hidden_states", "p", "r", "raise", "rand", "return", "return_dict", "same", "self", "size", "specify", "the", "time", "to", "to_drop", "torch", "training", "tuple", "use_return_dict", "v", "view", "warn_if_padding_and_no_attention_mask"], "blenderbot/modeling_blenderbot.py:BlenderbotDecoder.__init__": ["False", "LayerNorm", "ModelConfig", "ModelDecoder", "ModelDecoderLayer", "ModelLearnedPositionalEmbedding", "ModelPreTrainedModel", "ModelScaledWordEmbedding", "ModuleList", "__init__", "class", "config", "d_model", "decoder_layerdrop", "decoder_layers", "def", "dropout", "else", "embed_positions", "embed_scale", "embed_tokens", "for", "gradient_checkpointing", "i", "if", "in", "layer_idx", "layer_norm", "layerdrop", "layers", "math", "max_position_embeddings", "max_target_positions", "nn", "pad_token_id", "padding_idx", "post_init", "range", "scale_embedding", "self", "sqrt", "super", "vocab_size"], "blenderbot/modeling_blenderbot.py:BlenderbotDecoder.forward": ["BaseModelOutputWithPastAndCrossAttentions", "DynamicCache", "EncoderDecoderCache", "False", "ModelDecoder", "ModelPreTrainedModel", "None", "Setting", "Tensor", "True", "ValueError", "You", "all_cross_attentions", "all_hidden_states", "all_self_attns", "and", "arange", "at", "attention_mask", "attentions", "batch_size", "both", "cache_position", "cannot", "causal_mask", "checkpointing", "class", "config", "continue", "create_bidirectional_mask", "create_causal_mask", "cross_attentions", "decoder_input_ids", "decoder_inputs_embeds", "decoder_layer", "def", "device", "dropout", "dropout_probability", "either", "elif", "else", "embed_positions", "embed_tokens", "encoder_attention_mask", "encoder_hidden_states", "enumerate", "for", "forward", "functional", "get_seq_length", "gradient", "gradient_checkpointing", "have", "hidden_states", "idx", "if", "in", "incompatible", "input", "input_embeds", "input_ids", "input_shape", "inputs_embeds", "is", "is_encoder_decoder", "is_torchdynamo_compiling", "isinstance", "kwargs", "last_hidden_state", "layer_norm", "layer_outputs", "layerdrop", "layers", "logger", "mask_seq_length", "nn", "not", "ones", "or", "output_attentions", "output_hidden_states", "p", "past_key_values", "past_key_values_length", "position_ids", "r", "raise", "rand", "return", "return_dict", "same", "self", "self_attention_cache", "self_attn_cache", "seq_length", "shape", "size", "specify", "the", "time", "to", "torch", "training", "tuple", "use_cache", "use_return_dict", "v", "view", "warning_once", "with"], "blenderbot/modeling_blenderbot.py:BlenderbotModel.__init__": ["ModelConfig", "ModelDecoder", "ModelEncoder", "ModelModel", "ModelPreTrainedModel", "ModelScaledWordEmbedding", "__init__", "class", "config", "d_model", "decoder", "def", "else", "embed_scale", "encoder", "if", "math", "pad_token_id", "padding_idx", "post_init", "scale_embedding", "self", "shared", "sqrt", "super", "vocab_size"], "blenderbot/modeling_blenderbot.py:BlenderbotModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "return", "self", "shared"], "blenderbot/modeling_blenderbot.py:BlenderbotModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "decoder", "def", "embed_tokens", "encoder", "self", "set_input_embeddings", "shared", "value"], "blenderbot/modeling_blenderbot.py:BlenderbotModel.forward": ["BaseModelOutput", "Cache", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Seq2SeqModelOutput", "Tensor", "and", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "cross_attentions", "decoder", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_input_ids", "decoder_inputs_embeds", "decoder_outputs", "def", "elif", "else", "encoder", "encoder_attention_mask", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "isinstance", "kwargs", "last_hidden_state", "len", "not", "output_attentions", "output_hidden_states", "past_key_values", "r", "return", "return_dict", "self", "torch", "tuple", "use_cache", "use_return_dict"], "blenderbot/modeling_blenderbot.py:BlenderbotForConditionalGeneration.__init__": ["False", "GenerationMixin", "Linear", "ModelConfig", "ModelForConditionalGeneration", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "d_model", "def", "final_logits_bias", "lm_head", "model", "nn", "num_embeddings", "post_init", "register_buffer", "self", "shared", "super", "torch", "zeros"], "blenderbot/modeling_blenderbot.py:BlenderbotForConditionalGeneration.resize_token_embeddings": ["Embedding", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "True", "_resize_final_logits_bias", "bool", "class", "def", "int", "mean_resizing", "new_embeddings", "new_num_tokens", "nn", "pad_to_multiple_of", "resize_token_embeddings", "return", "self", "shape", "super", "weight"], "blenderbot/modeling_blenderbot.py:BlenderbotForConditionalGeneration._resize_final_logits_bias": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "_resize_final_logits_bias", "cat", "class", "def", "device", "dim", "else", "extra_bias", "final_logits_bias", "if", "int", "new_bias", "new_num_tokens", "old_num_tokens", "register_buffer", "self", "shape", "torch", "zeros"], "blenderbot/modeling_blenderbot.py:BlenderbotForConditionalGeneration.forward": ["BaseModelOutput", "Cache", "CrossEntropyLoss", "False", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Seq2SeqLMOutput", "Tensor", "The", "and", "argument", "attention_mask", "bool", "cache_position", "changed", "class", "config", "cross_attentions", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_input_ids", "decoder_inputs_embeds", "decoder_start_token_id", "def", "else", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "final_logits_bias", "forward", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "lm_head", "lm_logits", "logger", "logits", "loss", "loss_fct", "masked_lm_loss", "model", "not", "output", "output_attentions", "output_hidden_states", "outputs", "pad_token_id", "past_key_values", "provided", "r", "return", "return_dict", "self", "shift_tokens_right", "since", "to", "torch", "tuple", "use_cache", "use_return_dict", "view", "vocab_size", "warning"], "blenderbot/modeling_blenderbot.py:BlenderbotDecoderWrapper.__init__": ["ModelDecoder", "ModelDecoderWrapper", "ModelPreTrainedModel", "__init__", "class", "config", "decoder", "def", "post_init", "self", "super"], "blenderbot/modeling_blenderbot.py:BlenderbotDecoderWrapper.forward": ["ModelDecoderWrapper", "ModelPreTrainedModel", "args", "class", "decoder", "def", "forward", "kwargs", "return", "self"], "blenderbot/modeling_blenderbot.py:BlenderbotForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "ModelDecoderWrapper", "ModelForCausalLM", "ModelPreTrainedModel", "True", "__init__", "bias", "class", "config", "def", "hidden_size", "is_decoder", "is_encoder_decoder", "lm_head", "model", "nn", "post_init", "self", "super", "vocab_size"], "blenderbot/modeling_blenderbot.py:BlenderbotForCausalLM.get_input_embeddings": ["GenerationMixin", "ModelForCausalLM", "ModelPreTrainedModel", "class", "decoder", "def", "embed_tokens", "get_input_embeddings", "model", "return", "self"], "blenderbot/modeling_blenderbot.py:BlenderbotForCausalLM.set_input_embeddings": ["GenerationMixin", "ModelForCausalLM", "ModelPreTrainedModel", "class", "decoder", "def", "embed_tokens", "model", "self", "set_input_embeddings", "value"], "blenderbot/modeling_blenderbot.py:BlenderbotForCausalLM.forward": ["Cache", "CausalLMOutputWithCrossAttentions", "CrossEntropyLoss", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForCausalLM", "ModelPreTrainedModel", "None", "Tensor", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "cross_attentions", "decoder", "def", "device", "else", "encoder_attention_mask", "encoder_hidden_states", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "lm_head", "logits", "logits_to_keep", "loss", "loss_fct", "model", "not", "output", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "r", "return", "return_dict", "self", "slice", "slice_indices", "to", "torch", "tuple", "use_cache", "use_return_dict", "view", "vocab_size"], "blenderbot_small/modeling_blenderbot_small.py:shift_tokens_right": ["Model_tokens_right", "Modeled_input_ids", "None", "Tensor", "ValueError", "be", "clone", "config", "decoder_start_token_id", "def", "defined", "has", "if", "input_ids", "int", "is", "masked_fill_", "model", "new_zeros", "pad_token_id", "raise", "return", "self", "shape", "to", "torch"], "blenderbot_small/modeling_blenderbot_small.py:BlenderbotSmallLearnedPositionalEmbedding.__init__": ["Embedding", "ModelLearnedPositionalEmbedding", "__init__", "class", "def", "embedding_dim", "int", "nn", "num_embeddings", "self", "super"], "blenderbot_small/modeling_blenderbot_small.py:BlenderbotSmallLearnedPositionalEmbedding.forward": ["Embedding", "ModelLearnedPositionalEmbedding", "None", "Size", "Tensor", "arange", "bsz", "class", "def", "device", "dtype", "forward", "if", "input_ids_shape", "int", "is", "long", "nn", "past_key_values_length", "position_ids", "return", "self", "seq_len", "super", "torch", "weight"], "blenderbot_small/modeling_blenderbot_small.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "contiguous", "def", "dim", "dropout", "float", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "shape", "size", "softmax", "torch", "training", "transpose", "value"], "blenderbot_small/modeling_blenderbot_small.py:BlenderbotSmallAttention.__init__": ["False", "Instantiating", "Linear", "ModelAttention", "ModelConfig", "Module", "None", "Please", "True", "ValueError", "__class__", "__init__", "__name__", "a", "and", "be", "bias", "bool", "by", "caching", "call", "class", "config", "creating", "decoder", "def", "divisible", "dropout", "during", "embed_dim", "errors", "f", "float", "forward", "got", "head_dim", "if", "int", "is", "is_causal", "is_decoder", "k_proj", "layer_idx", "lead", "logger", "make", "must", "nn", "not", "num_heads", "out_proj", "passing", "provide", "q_proj", "raise", "recommended", "scaling", "self", "super", "sure", "the", "this", "to", "used", "v_proj", "warning_once", "when", "will", "without"], "blenderbot_small/modeling_blenderbot_small.py:BlenderbotSmallAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "EncoderDecoderCache", "False", "FlashAttentionKwargs", "ModelAttention", "Module", "None", "Tensor", "True", "Unpack", "_attn_implementation", "and", "attention_interface", "attention_mask", "attn_output", "attn_weights", "bool", "bsz", "cache_position", "class", "config", "contiguous", "cross_attention_cache", "curr_past_key_values", "current_states", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "get", "head_dim", "hidden_states", "if", "is", "is_cross_attention", "is_updated", "isinstance", "k_proj", "key_states", "key_value_states", "keys", "kv_input_shape", "kwargs", "layer_idx", "layers", "nn", "not", "out_proj", "output_attentions", "past_key_values", "q_input_shape", "q_proj", "query_states", "reshape", "return", "scaling", "self", "self_attention_cache", "shape", "src_len", "tgt_len", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "values", "view"], "blenderbot_small/modeling_blenderbot_small.py:BlenderbotSmallEncoderLayer.__init__": ["ACT2FN", "GradientCheckpointingLayer", "LayerNorm", "Linear", "ModelAttention", "ModelConfig", "ModelEncoderLayer", "None", "__init__", "activation_dropout", "activation_fn", "activation_function", "attention_dropout", "class", "config", "d_model", "def", "dropout", "embed_dim", "encoder_attention_heads", "encoder_ffn_dim", "fc1", "fc2", "final_layer_norm", "int", "layer_idx", "nn", "num_heads", "self", "self_attn", "self_attn_layer_norm", "super"], "blenderbot_small/modeling_blenderbot_small.py:BlenderbotSmallEncoderLayer.forward": ["False", "FloatTensor", "GradientCheckpointingLayer", "ModelEncoderLayer", "None", "activation_dropout", "activation_fn", "and", "any", "attention_mask", "attn_weights", "bool", "clamp", "clamp_value", "class", "def", "dropout", "dtype", "fc1", "fc2", "final_layer_norm", "finfo", "float16", "forward", "functional", "hidden_states", "if", "isinf", "isnan", "max", "min", "nn", "or", "output_attentions", "outputs", "p", "residual", "return", "self", "self_attn", "self_attn_layer_norm", "torch", "training", "tuple"], "blenderbot_small/modeling_blenderbot_small.py:BlenderbotSmallDecoderLayer.__init__": ["ACT2FN", "GradientCheckpointingLayer", "LayerNorm", "Linear", "ModelAttention", "ModelConfig", "ModelDecoderLayer", "None", "True", "__init__", "activation_dropout", "activation_fn", "activation_function", "attention_dropout", "class", "config", "d_model", "decoder_attention_heads", "decoder_ffn_dim", "def", "dropout", "embed_dim", "encoder_attn", "encoder_attn_layer_norm", "fc1", "fc2", "final_layer_norm", "int", "is_causal", "is_decoder", "layer_idx", "nn", "num_heads", "self", "self_attn", "self_attn_layer_norm", "super"], "blenderbot_small/modeling_blenderbot_small.py:BlenderbotSmallDecoderLayer.forward": ["Cache", "False", "FloatTensor", "GradientCheckpointingLayer", "ModelDecoderLayer", "None", "Tensor", "True", "activation_dropout", "activation_fn", "attention_mask", "bool", "cache_position", "class", "cross_attn_weights", "def", "dropout", "encoder_attention_mask", "encoder_attn", "encoder_attn_layer_norm", "encoder_hidden_states", "fc1", "fc2", "final_layer_norm", "forward", "functional", "hidden_states", "if", "is", "key_value_states", "nn", "not", "output_attentions", "outputs", "p", "past_key_values", "residual", "return", "self", "self_attn", "self_attn_layer_norm", "self_attn_weights", "torch", "training", "tuple", "use_cache"], "blenderbot_small/modeling_blenderbot_small.py:BlenderbotSmallPreTrainedModel._init_weights": ["ModelForConditionalGeneration", "ModelPreTrainedModel", "PreTrainedModel", "_init_weights", "class", "def", "final_logits_bias", "if", "init", "isinstance", "module", "self", "super", "zeros_"], "blenderbot_small/modeling_blenderbot_small.py:BlenderbotSmallPreTrainedModel.dummy_inputs": ["ModelPreTrainedModel", "PreTrainedModel", "attention_mask", "class", "config", "decoder_input_ids", "def", "device", "dummy_inputs", "input_ids", "ne", "pad_token", "pad_token_id", "return", "self", "tensor", "torch"], "blenderbot_small/modeling_blenderbot_small.py:BlenderbotSmallEncoder.__init__": ["Embedding", "False", "LayerNorm", "ModelConfig", "ModelEncoder", "ModelEncoderLayer", "ModelLearnedPositionalEmbedding", "ModelPreTrainedModel", "ModuleList", "_", "__init__", "class", "config", "d_model", "def", "dropout", "else", "embed_dim", "embed_positions", "embed_scale", "embed_tokens", "encoder_layerdrop", "encoder_layers", "for", "gradient_checkpointing", "if", "in", "layerdrop", "layernorm_embedding", "layers", "math", "max_position_embeddings", "max_source_positions", "nn", "pad_token_id", "padding_idx", "post_init", "range", "scale_embedding", "self", "sqrt", "super", "vocab_size"], "blenderbot_small/modeling_blenderbot_small.py:BlenderbotSmallEncoder.forward": ["BaseModelOutput", "False", "ModelEncoder", "ModelPreTrainedModel", "None", "True", "ValueError", "You", "all_attentions", "and", "at", "attention_mask", "attentions", "both", "cannot", "class", "config", "create_bidirectional_mask", "def", "dropout", "dropout_probability", "either", "elif", "else", "embed_pos", "embed_positions", "embed_scale", "embed_tokens", "encoder_layer", "encoder_states", "enumerate", "for", "forward", "functional", "have", "hidden_states", "idx", "if", "in", "input_embeds", "input_ids", "input_shape", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layer_outputs", "layerdrop", "layernorm_embedding", "layers", "nn", "not", "or", "output_attentions", "output_hidden_states", "p", "r", "raise", "rand", "return", "return_dict", "same", "self", "size", "specify", "the", "time", "to", "to_drop", "torch", "training", "tuple", "use_return_dict", "v", "view", "warn_if_padding_and_no_attention_mask"], "blenderbot_small/modeling_blenderbot_small.py:BlenderbotSmallDecoder.__init__": ["Embedding", "False", "LayerNorm", "ModelConfig", "ModelDecoder", "ModelDecoderLayer", "ModelLearnedPositionalEmbedding", "ModelPreTrainedModel", "ModuleList", "__init__", "class", "config", "d_model", "decoder_layerdrop", "decoder_layers", "def", "dropout", "else", "embed_positions", "embed_scale", "embed_tokens", "for", "gradient_checkpointing", "i", "if", "in", "layer_idx", "layerdrop", "layernorm_embedding", "layers", "math", "max_position_embeddings", "max_target_positions", "nn", "pad_token_id", "padding_idx", "post_init", "range", "scale_embedding", "self", "sqrt", "super", "vocab_size"], "blenderbot_small/modeling_blenderbot_small.py:BlenderbotSmallDecoder.forward": ["BaseModelOutputWithPastAndCrossAttentions", "DynamicCache", "EncoderDecoderCache", "False", "ModelDecoder", "ModelPreTrainedModel", "None", "Setting", "True", "ValueError", "You", "all_cross_attentions", "all_hidden_states", "all_self_attns", "and", "arange", "at", "attention_mask", "attentions", "batch_size", "both", "cache_position", "cannot", "causal_mask", "checkpointing", "class", "config", "continue", "create_bidirectional_mask", "create_causal_mask", "cross_attentions", "decoder_input_ids", "decoder_inputs_embeds", "decoder_layer", "def", "device", "dropout", "dropout_probability", "either", "elif", "else", "embed_positions", "embed_tokens", "encoder_attention_mask", "encoder_hidden_states", "enumerate", "for", "forward", "functional", "get_seq_length", "gradient", "gradient_checkpointing", "have", "hidden_states", "idx", "if", "in", "incompatible", "input", "input_embeds", "input_ids", "input_shape", "inputs_embeds", "is", "is_encoder_decoder", "is_torchdynamo_compiling", "isinstance", "kwargs", "last_hidden_state", "layer_outputs", "layerdrop", "layernorm_embedding", "layers", "logger", "mask_seq_length", "nn", "not", "ones", "or", "output_attentions", "output_hidden_states", "p", "past_key_values", "past_key_values_length", "position_ids", "r", "raise", "rand", "return", "return_dict", "same", "self", "self_attention_cache", "self_attn_cache", "seq_length", "shape", "size", "specify", "the", "time", "to", "torch", "training", "tuple", "use_cache", "use_return_dict", "v", "view", "warning_once", "with"], "blenderbot_small/modeling_blenderbot_small.py:BlenderbotSmallModel.__init__": ["Embedding", "ModelConfig", "ModelDecoder", "ModelEncoder", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "config", "d_model", "decoder", "def", "encoder", "nn", "pad_token_id", "padding_idx", "post_init", "self", "shared", "super", "vocab_size"], "blenderbot_small/modeling_blenderbot_small.py:BlenderbotSmallModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "return", "self", "shared"], "blenderbot_small/modeling_blenderbot_small.py:BlenderbotSmallModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "decoder", "def", "embed_tokens", "encoder", "self", "set_input_embeddings", "shared", "value"], "blenderbot_small/modeling_blenderbot_small.py:BlenderbotSmallModel.forward": ["BaseModelOutput", "Cache", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Seq2SeqModelOutput", "Tensor", "and", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "cross_attentions", "decoder", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_input_ids", "decoder_inputs_embeds", "decoder_outputs", "def", "elif", "else", "encoder", "encoder_attention_mask", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "isinstance", "kwargs", "last_hidden_state", "len", "not", "output_attentions", "output_hidden_states", "past_key_values", "r", "return", "return_dict", "self", "torch", "tuple", "use_cache", "use_return_dict"], "blenderbot_small/modeling_blenderbot_small.py:BlenderbotSmallForConditionalGeneration.__init__": ["False", "GenerationMixin", "Linear", "ModelConfig", "ModelForConditionalGeneration", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "d_model", "def", "final_logits_bias", "lm_head", "model", "nn", "num_embeddings", "post_init", "register_buffer", "self", "shared", "super", "torch", "zeros"], "blenderbot_small/modeling_blenderbot_small.py:BlenderbotSmallForConditionalGeneration.resize_token_embeddings": ["Embedding", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "True", "_resize_final_logits_bias", "bool", "class", "def", "int", "mean_resizing", "new_embeddings", "new_num_tokens", "nn", "pad_to_multiple_of", "resize_token_embeddings", "return", "self", "shape", "super", "weight"], "blenderbot_small/modeling_blenderbot_small.py:BlenderbotSmallForConditionalGeneration._resize_final_logits_bias": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "_resize_final_logits_bias", "cat", "class", "def", "device", "dim", "else", "extra_bias", "final_logits_bias", "if", "int", "new_bias", "new_num_tokens", "old_num_tokens", "register_buffer", "self", "shape", "torch", "zeros"], "blenderbot_small/modeling_blenderbot_small.py:BlenderbotSmallForConditionalGeneration.forward": ["BaseModelOutput", "Cache", "CrossEntropyLoss", "False", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Seq2SeqLMOutput", "Tensor", "The", "and", "argument", "attention_mask", "bool", "cache_position", "changed", "class", "config", "cross_attentions", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_input_ids", "decoder_inputs_embeds", "decoder_start_token_id", "def", "else", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "final_logits_bias", "forward", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "lm_head", "lm_logits", "logger", "logits", "loss", "loss_fct", "masked_lm_loss", "model", "not", "output", "output_attentions", "output_hidden_states", "outputs", "pad_token_id", "past_key_values", "provided", "r", "return", "return_dict", "self", "shift_tokens_right", "since", "to", "torch", "tuple", "use_cache", "use_return_dict", "view", "vocab_size", "warning"], "blenderbot_small/modeling_blenderbot_small.py:BlenderbotSmallDecoderWrapper.__init__": ["ModelDecoder", "ModelDecoderWrapper", "ModelPreTrainedModel", "__init__", "class", "config", "decoder", "def", "post_init", "self", "super"], "blenderbot_small/modeling_blenderbot_small.py:BlenderbotSmallDecoderWrapper.forward": ["ModelDecoderWrapper", "ModelPreTrainedModel", "args", "class", "decoder", "def", "forward", "kwargs", "return", "self"], "blenderbot_small/modeling_blenderbot_small.py:BlenderbotSmallForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "ModelDecoderWrapper", "ModelForCausalLM", "ModelPreTrainedModel", "True", "__init__", "bias", "class", "config", "def", "hidden_size", "is_decoder", "is_encoder_decoder", "lm_head", "model", "nn", "post_init", "self", "super", "vocab_size"], "blenderbot_small/modeling_blenderbot_small.py:BlenderbotSmallForCausalLM.get_input_embeddings": ["GenerationMixin", "ModelForCausalLM", "ModelPreTrainedModel", "class", "decoder", "def", "embed_tokens", "get_input_embeddings", "model", "return", "self"], "blenderbot_small/modeling_blenderbot_small.py:BlenderbotSmallForCausalLM.set_input_embeddings": ["GenerationMixin", "ModelForCausalLM", "ModelPreTrainedModel", "class", "decoder", "def", "embed_tokens", "model", "self", "set_input_embeddings", "value"], "blenderbot_small/modeling_blenderbot_small.py:BlenderbotSmallForCausalLM.forward": ["Cache", "CausalLMOutputWithCrossAttentions", "CrossEntropyLoss", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForCausalLM", "ModelPreTrainedModel", "None", "Tensor", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "cross_attentions", "decoder", "def", "device", "else", "encoder_attention_mask", "encoder_hidden_states", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "lm_head", "logits", "logits_to_keep", "loss", "loss_fct", "model", "not", "output", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "r", "return", "return_dict", "self", "slice", "slice_indices", "to", "torch", "tuple", "use_cache", "use_return_dict", "view", "vocab_size"], "blip/modeling_blip.py:contrastive_loss": ["Model_loss", "Tensor", "arange", "cross_entropy", "def", "device", "functional", "len", "logits", "nn", "return", "torch"], "blip/modeling_blip.py:blip_loss": ["Model_loss", "Tensor", "caption_loss", "contrastive_loss", "def", "image_loss", "return", "similarity", "t", "torch"], "blip/modeling_blip.py:BlipOutput.to_tuple": ["Any", "ModelOutput", "class", "def", "else", "for", "getattr", "if", "in", "k", "keys", "not", "return", "self", "text_model_output", "to_tuple", "tuple", "vision_model_output"], "blip/modeling_blip.py:BlipVisionEmbeddings.__init__": ["Conv2d", "ModelVisionConfig", "ModelVisionEmbeddings", "Module", "Parameter", "__init__", "class", "class_embedding", "config", "def", "embed_dim", "hidden_size", "image_size", "in_channels", "kernel_size", "nn", "num_patches", "num_positions", "out_channels", "patch_embedding", "patch_size", "position_embedding", "randn", "self", "stride", "super", "torch"], "blip/modeling_blip.py:BlipVisionEmbeddings.interpolate_pos_encoding": ["False", "ModelVisionEmbeddings", "Module", "Tensor", "align_corners", "and", "bicubic", "cat", "class", "class_pos_embed", "def", "dim", "embeddings", "functional", "height", "if", "int", "interpolate", "interpolate_pos_encoding", "is_tracing", "jit", "mode", "new_height", "new_width", "nn", "not", "num_patches", "num_positions", "patch_pos_embed", "patch_size", "permute", "position_embedding", "reshape", "return", "self", "shape", "size", "sqrt_num_positions", "torch", "torch_int", "view", "width"], "blip/modeling_blip.py:BlipVisionEmbeddings.forward": ["False", "FloatTensor", "ModelVisionEmbeddings", "Module", "Tensor", "_", "batch_size", "bool", "cat", "class", "class_embedding", "class_embeds", "def", "dim", "dtype", "else", "embeddings", "expand", "flatten", "forward", "height", "if", "interpolate_pos_encoding", "nn", "patch_embedding", "patch_embeds", "pixel_values", "position_embedding", "return", "self", "shape", "size", "target_dtype", "to", "torch", "transpose", "weight", "width"], "blip/modeling_blip.py:BlipTextEmbeddings.__init__": ["Embedding", "False", "ModelTextConfig", "ModelTextEmbeddings", "Module", "__init__", "arange", "class", "config", "def", "embed_dim", "expand", "hidden_size", "max_position_embeddings", "nn", "persistent", "position_embedding", "position_ids", "register_buffer", "self", "super", "token_embedding", "torch", "vocab_size"], "blip/modeling_blip.py:BlipTextEmbeddings.forward": ["FloatTensor", "LongTensor", "ModelTextEmbeddings", "Module", "None", "Sequence", "Tensor", "ValueError", "and", "be", "class", "def", "else", "embeddings", "f", "forward", "got", "if", "input_ids", "inputs_embeds", "is", "length", "less", "max_position_embedding", "max_position_embeddings", "must", "nn", "not", "position_embedding", "position_embeddings", "position_ids", "raise", "return", "self", "seq_length", "sequence", "shape", "than", "token_embedding", "torch", "weight"], "blip/modeling_blip.py:BlipAttention.__init__": ["Dropout", "Linear", "ModelAttention", "Module", "ValueError", "__init__", "and", "attention_dropout", "be", "by", "class", "config", "def", "divisible", "dropout", "embed_dim", "f", "got", "head_dim", "hidden_size", "if", "must", "nn", "num_attention_heads", "num_heads", "projection", "qkv", "raise", "scale", "self", "super"], "blip/modeling_blip.py:BlipAttention._shape": ["ModelAttention", "Module", "Tensor", "_shape", "bsz", "class", "contiguous", "def", "head_dim", "int", "nn", "num_heads", "return", "self", "seq_len", "tensor", "torch", "transpose", "view"], "blip/modeling_blip.py:BlipAttention.forward": ["ModelAttention", "Module", "Tensor", "TransformersKwargs", "Unpack", "attention_probs", "attention_scores", "bsz", "class", "context_layer", "def", "dim", "dropout", "embed_dim", "forward", "functional", "hidden_states", "key_states", "kwargs", "matmul", "mixed_qkv", "new_context_layer_shape", "nn", "num_heads", "output", "permute", "projection", "qkv", "query_states", "reshape", "return", "scale", "self", "size", "softmax", "tgt_len", "torch", "transpose", "tuple", "value_states"], "blip/modeling_blip.py:BlipMLP.__init__": ["ACT2FN", "Linear", "ModelMLP", "Module", "__init__", "activation_fn", "class", "config", "def", "fc1", "fc2", "hidden_act", "hidden_size", "intermediate_size", "nn", "self", "super"], "blip/modeling_blip.py:BlipMLP.forward": ["ModelMLP", "Module", "Tensor", "activation_fn", "class", "def", "fc1", "fc2", "forward", "hidden_states", "nn", "return", "self", "torch"], "blip/modeling_blip.py:BlipEncoderLayer.__init__": ["GradientCheckpointingLayer", "LayerNorm", "ModelAttention", "ModelConfig", "ModelEncoderLayer", "ModelMLP", "__init__", "class", "config", "def", "embed_dim", "eps", "hidden_size", "layer_norm1", "layer_norm2", "layer_norm_eps", "mlp", "nn", "self", "self_attn", "super"], "blip/modeling_blip.py:BlipEncoderLayer.forward": ["FloatTensor", "GradientCheckpointingLayer", "ModelEncoderLayer", "Tensor", "TransformersKwargs", "Unpack", "_", "class", "def", "forward", "hidden_states", "kwargs", "layer_norm1", "layer_norm2", "mlp", "residual", "return", "self", "self_attn", "torch"], "blip/modeling_blip.py:BlipPreTrainedModel._init_weights": ["ModelPreTrainedModel", "ModelTextEmbeddings", "ModelVisionEmbeddings", "PreTrainedModel", "_init_weights", "arange", "class", "class_embedding", "config", "copy_", "def", "elif", "expand", "hasattr", "if", "init", "initializer_range", "isinstance", "mean", "module", "position_embedding", "position_ids", "self", "shape", "std", "super", "torch", "trunc_normal_", "vision_config"], "blip/modeling_blip.py:BlipEncoder.__init__": ["False", "ModelConfig", "ModelEncoder", "ModelEncoderLayer", "Module", "ModuleList", "_", "__init__", "class", "config", "def", "for", "gradient_checkpointing", "in", "layers", "nn", "num_hidden_layers", "range", "self", "super"], "blip/modeling_blip.py:BlipEncoder.forward": ["BaseModelOutput", "ModelEncoder", "Module", "TransformersKwargs", "Unpack", "class", "def", "encoder_layer", "for", "forward", "hidden_states", "in", "inputs_embeds", "kwargs", "last_hidden_state", "layers", "nn", "return", "self", "tuple"], "blip/modeling_blip.py:BlipVisionModel.__init__": ["LayerNorm", "ModelEncoder", "ModelPreTrainedModel", "ModelVisionConfig", "ModelVisionEmbeddings", "ModelVisionModel", "__init__", "class", "config", "def", "embed_dim", "embeddings", "encoder", "eps", "hidden_size", "layer_norm_eps", "nn", "post_init", "post_layernorm", "self", "super"], "blip/modeling_blip.py:BlipVisionModel.forward": ["BaseModelOutput", "BaseModelOutputWithPooling", "False", "FloatTensor", "ModelPreTrainedModel", "ModelVisionModel", "None", "TransformersKwargs", "Unpack", "ValueError", "You", "bool", "class", "def", "embeddings", "encoder", "encoder_outputs", "forward", "have", "hidden_states", "if", "inputs_embeds", "interpolate_pos_encoding", "is", "kwargs", "last_hidden_state", "pixel_values", "pooled_output", "pooler_output", "post_layernorm", "raise", "return", "self", "specify", "to", "torch", "tuple"], "blip/modeling_blip.py:BlipVisionModel.get_input_embeddings": ["ModelPreTrainedModel", "ModelVisionModel", "class", "def", "embeddings", "get_input_embeddings", "return", "self"], "blip/modeling_blip.py:BlipModel.__init__": ["False", "Linear", "ModelConfig", "ModelForConditionalGeneration", "ModelForImageTextRetrieval", "ModelForQuestionAnswering", "ModelModel", "ModelPreTrainedModel", "ModelTextConfig", "ModelTextModel", "ModelVisionConfig", "ModelVisionModel", "Parameter", "TypeError", "__init__", "be", "bias", "but", "class", "config", "def", "depending", "deprecated", "expected", "f", "future", "going", "hidden_size", "if", "in", "is", "isinstance", "logger", "logit_scale", "logit_scale_init_value", "nn", "not", "of", "on", "or", "please", "post_init", "projection_dim", "raise", "release", "self", "super", "tensor", "text_config", "text_embed_dim", "text_model", "text_projection", "to", "torch", "type", "use", "usecase", "vision_config", "vision_embed_dim", "vision_model", "visual_projection", "warning", "your"], "blip/modeling_blip.py:BlipModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "return", "self", "text_model"], "blip/modeling_blip.py:BlipModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "self", "set_input_embeddings", "text_model", "value"], "blip/modeling_blip.py:BlipModel.get_text_features": ["FloatTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "attention_mask", "class", "def", "get_text_features", "input_ids", "pooled_output", "position_ids", "r", "return", "self", "text_features", "text_model", "text_outputs", "text_projection", "torch"], "blip/modeling_blip.py:BlipModel.get_image_features": ["False", "FloatTensor", "ModelModel", "ModelPreTrainedModel", "None", "bool", "class", "def", "get_image_features", "image_features", "interpolate_pos_encoding", "pixel_values", "pooled_output", "r", "return", "self", "torch", "vision_model", "vision_outputs", "visual_projection"], "blip/modeling_blip.py:BlipModel.get_multimodal_features": ["False", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "attention_mask", "bool", "class", "def", "dtype", "encoder_attention_mask", "encoder_hidden_states", "get_multimodal_features", "image_atts", "image_embeds", "input_ids", "interpolate_pos_encoding", "long", "multimodal_features", "ones", "pixel_values", "pooled_output", "r", "return", "self", "size", "text_model", "text_outputs", "text_projection", "torch", "vision_model", "vision_outputs"], "blip/modeling_blip.py:BlipModel.forward": ["False", "FloatTensor", "LongTensor", "ModelModel", "ModelOutput", "ModelPreTrainedModel", "Model_loss", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "attention_mask", "bool", "class", "def", "device", "dim", "dtype", "exp", "forward", "if", "image_embeds", "input_ids", "interpolate_pos_encoding", "keepdim", "kwargs", "logit_scale", "logits_per_image", "logits_per_text", "loss", "matmul", "norm", "p", "pixel_values", "pooler_output", "position_ids", "r", "return", "return_loss", "self", "t", "text_embeds", "text_model", "text_model_output", "text_outputs", "text_projection", "to", "torch", "tuple", "vision_model", "vision_model_output", "vision_outputs", "visual_projection"], "blip/modeling_blip.py:BlipForConditionalGeneration.__init__": ["GenerationMixin", "ModelConfig", "ModelForConditionalGeneration", "ModelPreTrainedModel", "ModelTextLMHeadModel", "ModelVisionModel", "__init__", "bos_token_id", "class", "config", "decoder_input_ids", "decoder_pad_token_id", "def", "pad_token_id", "post_init", "self", "super", "text_config", "text_decoder", "vision_config", "vision_model"], "blip/modeling_blip.py:BlipForConditionalGeneration.get_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "return", "self", "text_decoder"], "blip/modeling_blip.py:BlipForConditionalGeneration.set_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "self", "set_input_embeddings", "text_decoder", "value"], "blip/modeling_blip.py:BlipForConditionalGeneration.forward": ["False", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForConditionalGeneration", "ModelForConditionalGenerationModelOutput", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "bool", "class", "def", "encoder_hidden_states", "forward", "hidden_states", "image_embeds", "input_ids", "int", "interpolate_pos_encoding", "kwargs", "labels", "last_hidden_state", "logits", "logits_to_keep", "loss", "mean", "outputs", "pixel_values", "r", "reduction", "return", "self", "text_decoder", "torch", "tuple", "vision_model", "vision_outputs"], "blip/modeling_blip.py:BlipForConditionalGeneration.generate": ["False", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "attention_mask", "batch_size", "bool", "bos_token_id", "class", "config", "decoder_input_ids", "def", "device", "dtype", "elif", "else", "encoder_attention_mask", "encoder_hidden_states", "eos_token_id", "generate", "generate_kwargs", "if", "image_attention_mask", "image_embeds", "input_ids", "interpolate_pos_encoding", "is", "isinstance", "list", "long", "not", "ones", "outputs", "pad_token_id", "pixel_values", "r", "repeat", "return", "self", "sep_token_id", "shape", "size", "text_config", "text_decoder", "to", "torch", "vision_model", "vision_outputs"], "blip/modeling_blip.py:BlipForQuestionAnswering.__init__": ["False", "GenerationMixin", "ModelConfig", "ModelForQuestionAnswering", "ModelPreTrainedModel", "ModelTextLMHeadModel", "ModelTextModel", "ModelVisionModel", "__init__", "add_pooling_layer", "bos_token_id", "class", "config", "decoder_pad_token_id", "decoder_start_token_id", "def", "pad_token_id", "post_init", "self", "super", "text_config", "text_decoder", "text_encoder", "vision_config", "vision_model"], "blip/modeling_blip.py:BlipForQuestionAnswering.set_input_embeddings": ["GenerationMixin", "ModelForQuestionAnswering", "ModelPreTrainedModel", "class", "def", "self", "set_input_embeddings", "text_encoder", "value"], "blip/modeling_blip.py:BlipForQuestionAnswering.get_input_embeddings": ["GenerationMixin", "ModelForQuestionAnswering", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "return", "self", "text_encoder"], "blip/modeling_blip.py:BlipForQuestionAnswering.forward": ["Either", "False", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForQuestionAnswering", "ModelPreTrainedModel", "ModelTextVisionModelOutput", "None", "TransformersKwargs", "Unpack", "ValueError", "and", "answer_output", "are", "attention_mask", "attentions", "be", "bool", "call", "calling", "class", "decoder_attention_mask", "decoder_input_ids", "decoder_loss", "def", "dtype", "else", "encoder_attention_mask", "encoder_hidden_states", "for", "forward", "generate", "hidden_states", "if", "image_attention_mask", "image_embeds", "inference", "input_ids", "interpolate_pos_encoding", "is", "kwargs", "labels", "last_hidden_state", "long", "loss", "make", "mean", "model", "not", "ones", "or", "passed", "pixel_values", "question_embeds", "r", "raise", "reduction", "return", "self", "should", "size", "sure", "text_decoder", "text_encoder", "that", "the", "torch", "training", "tuple", "using", "vision_model", "vision_outputs", "when", "with", "you"], "blip/modeling_blip.py:BlipForQuestionAnswering.generate": ["False", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForQuestionAnswering", "ModelPreTrainedModel", "None", "attention_mask", "bool", "bos_ids", "class", "config", "decoder_start_token_id", "def", "device", "dtype", "encoder_attention_mask", "encoder_hidden_states", "eos_token_id", "fill_value", "full", "generate", "generate_kwargs", "if", "image_attention_mask", "image_embeds", "input_ids", "interpolate_pos_encoding", "isinstance", "list", "long", "ones", "outputs", "pad_token_id", "pixel_values", "question_attention_mask", "question_embeds", "question_outputs", "r", "return", "return_dict", "self", "sep_token_id", "size", "text_config", "text_decoder", "text_encoder", "torch", "vision_model", "vision_outputs"], "blip/modeling_blip.py:BlipForImageTextRetrieval.__init__": ["False", "Linear", "ModelConfig", "ModelForImageTextRetrieval", "ModelPreTrainedModel", "ModelTextModel", "ModelVisionModel", "__init__", "add_pooling_layer", "bos_token_id", "class", "config", "decoder_pad_token_id", "decoder_start_token_id", "def", "else", "hasattr", "hidden_size", "if", "image_text_hidden_size", "itm_head", "nn", "not", "pad_token_id", "post_init", "self", "super", "text_config", "text_encoder", "text_proj", "vision_config", "vision_model", "vision_proj"], "blip/modeling_blip.py:BlipForImageTextRetrieval.get_input_embeddings": ["ModelForImageTextRetrieval", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "return", "self", "text_encoder"], "blip/modeling_blip.py:BlipForImageTextRetrieval.set_input_embeddings": ["ModelForImageTextRetrieval", "ModelPreTrainedModel", "class", "def", "self", "set_input_embeddings", "text_encoder", "value"], "blip/modeling_blip.py:BlipForImageTextRetrieval.forward": ["False", "FloatTensor", "LongTensor", "ModelForImageTextRetrieval", "ModelImageTextMatchingModelOutput", "ModelPreTrainedModel", "ModelTextVisionModelOutput", "None", "TransformersKwargs", "True", "Unpack", "attention_mask", "attentions", "bool", "class", "def", "dim", "dtype", "else", "encoder_attention_mask", "encoder_hidden_states", "forward", "hidden_states", "if", "image_atts", "image_embeds", "image_feat", "input_ids", "interpolate_pos_encoding", "itm_head", "itm_score", "kwargs", "last_hidden_state", "long", "normalize", "ones", "output", "pixel_values", "question_embeds", "r", "return", "self", "size", "t", "text_encoder", "text_feat", "text_proj", "torch", "tuple", "use_itm_head", "vision_model", "vision_outputs", "vision_proj"], "blip/modeling_blip_text.py:BlipTextEmbeddings.__init__": ["Dropout", "Embedding", "False", "LayerNorm", "ModelTextEmbeddings", "Module", "__init__", "arange", "class", "config", "def", "dropout", "eps", "expand", "hidden_dropout_prob", "hidden_size", "layer_norm_eps", "max_position_embeddings", "nn", "pad_token_id", "padding_idx", "persistent", "position_embeddings", "position_ids", "register_buffer", "self", "super", "torch", "vocab_size", "word_embeddings"], "blip/modeling_blip_text.py:BlipTextEmbeddings.forward": ["FloatTensor", "LayerNorm", "LongTensor", "ModelTextEmbeddings", "Module", "None", "Tensor", "class", "def", "dropout", "else", "embeddings", "forward", "if", "input_ids", "input_shape", "inputs_embeds", "int", "is", "nn", "not", "past_key_values_length", "position_embeddings", "position_ids", "return", "self", "seq_length", "size", "torch", "word_embeddings"], "blip/modeling_blip_text.py:BlipTextSelfAttention.__init__": ["Dropout", "Linear", "ModelTextSelfAttention", "Module", "None", "The", "ValueError", "__init__", "a", "all_head_size", "and", "attention", "attention_head_size", "attention_probs_dropout_prob", "class", "config", "d", "def", "dropout", "else", "embedding_size", "encoder_hidden_size", "hasattr", "heads", "hidden", "hidden_size", "if", "int", "is", "is_cross_attention", "key", "layer_idx", "multiple", "nn", "not", "num_attention_heads", "number", "of", "query", "raise", "self", "size", "super", "the", "value"], "blip/modeling_blip_text.py:BlipTextSelfAttention.save_attn_gradients": ["ModelTextSelfAttention", "Module", "attn_gradients", "class", "def", "nn", "save_attn_gradients", "self"], "blip/modeling_blip_text.py:BlipTextSelfAttention.get_attn_gradients": ["ModelTextSelfAttention", "Module", "attn_gradients", "class", "def", "get_attn_gradients", "nn", "return", "self"], "blip/modeling_blip_text.py:BlipTextSelfAttention.save_attention_map": ["ModelTextSelfAttention", "Module", "attention_map", "class", "def", "nn", "save_attention_map", "self"], "blip/modeling_blip_text.py:BlipTextSelfAttention.get_attention_map": ["ModelTextSelfAttention", "Module", "attention_map", "class", "def", "get_attention_map", "nn", "return", "self"], "blip/modeling_blip_text.py:BlipTextSelfAttention.forward": ["Cache", "EncoderDecoderCache", "False", "FloatTensor", "ModelTextSelfAttention", "Module", "None", "Softmax", "Tensor", "True", "_", "all_head_size", "and", "attention_head_size", "attention_mask", "attention_probs", "attention_probs_dropped", "attention_scores", "batch_size", "bool", "cache_position", "class", "context_layer", "contiguous", "cross_attention_cache", "curr_past_key_values", "current_states", "def", "device", "dim", "dropout", "else", "encoder_attention_mask", "encoder_hidden_states", "forward", "get", "hidden_states", "if", "is", "is_cross_attention", "is_updated", "isinstance", "key", "key_layer", "keys", "layer_idx", "layers", "math", "matmul", "new_context_layer_shape", "nn", "not", "num_attention_heads", "output_attentions", "past_key_values", "permute", "query", "query_layer", "return", "self", "self_attention_cache", "seq_length", "shape", "size", "sqrt", "to", "torch", "transpose", "tuple", "update", "value", "value_layer", "values", "view"], "blip/modeling_blip_text.py:BlipTextSelfOutput.__init__": ["Dropout", "LayerNorm", "Linear", "ModelTextSelfOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "eps", "hidden_dropout_prob", "hidden_size", "layer_norm_eps", "nn", "self", "super"], "blip/modeling_blip_text.py:BlipTextSelfOutput.forward": ["LayerNorm", "ModelTextSelfOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "blip/modeling_blip_text.py:BlipTextAttention.__init__": ["False", "ModelTextAttention", "ModelTextSelfAttention", "ModelTextSelfOutput", "Module", "None", "__init__", "class", "config", "def", "is_cross_attention", "layer_idx", "nn", "output", "self", "super"], "blip/modeling_blip_text.py:BlipTextAttention.forward": ["Cache", "False", "FloatTensor", "ModelTextAttention", "Module", "None", "Tensor", "attention_mask", "attention_output", "bool", "cache_position", "class", "def", "encoder_hidden_states", "forward", "hidden_states", "nn", "output", "output_attentions", "outputs", "past_key_values", "return", "self", "self_outputs", "torch", "tuple"], "blip/modeling_blip_text.py:BlipTextIntermediate.__init__": ["ACT2FN", "Linear", "ModelTextIntermediate", "Module", "__init__", "class", "config", "def", "dense", "else", "hidden_act", "hidden_size", "if", "intermediate_act_fn", "intermediate_size", "isinstance", "nn", "self", "str", "super"], "blip/modeling_blip_text.py:BlipTextIntermediate.forward": ["ModelTextIntermediate", "Module", "Tensor", "class", "def", "dense", "forward", "hidden_states", "intermediate_act_fn", "nn", "return", "self", "torch"], "blip/modeling_blip_text.py:BlipTextOutput.__init__": ["Dropout", "LayerNorm", "Linear", "ModelTextOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "eps", "hidden_dropout_prob", "hidden_size", "intermediate_size", "layer_norm_eps", "nn", "self", "super"], "blip/modeling_blip_text.py:BlipTextOutput.forward": ["LayerNorm", "ModelTextOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "blip/modeling_blip_text.py:BlipTextLayer.__init__": ["GradientCheckpointingLayer", "ModelTextAttention", "ModelTextIntermediate", "ModelTextLayer", "ModelTextOutput", "__init__", "attention", "chunk_size_feed_forward", "class", "config", "crossattention", "def", "if", "intermediate", "is_cross_attention", "is_decoder", "layer_idx", "layer_num", "output", "self", "seq_len_dim", "super"], "blip/modeling_blip_text.py:BlipTextLayer.forward": ["Cache", "False", "FloatTensor", "GradientCheckpointingLayer", "ModelTextLayer", "None", "Tensor", "apply_chunking_to_forward", "attention", "attention_mask", "attention_output", "bool", "cache_position", "chunk_size_feed_forward", "class", "cross_attention_outputs", "crossattention", "def", "encoder_attention_mask", "encoder_hidden_states", "feed_forward_chunk", "forward", "hidden_states", "if", "is", "layer_output", "not", "output_attentions", "outputs", "past_key_values", "return", "self", "self_attention_outputs", "seq_len_dim", "torch", "tuple"], "blip/modeling_blip_text.py:BlipTextLayer.feed_forward_chunk": ["GradientCheckpointingLayer", "ModelTextLayer", "attention_output", "class", "def", "feed_forward_chunk", "intermediate", "intermediate_output", "layer_output", "output", "return", "self"], "blip/modeling_blip_text.py:BlipTextEncoder.__init__": ["False", "ModelTextEncoder", "ModelTextLayer", "Module", "ModuleList", "__init__", "class", "config", "def", "for", "gradient_checkpointing", "i", "in", "layer", "nn", "num_hidden_layers", "range", "self", "super"], "blip/modeling_blip_text.py:BlipTextEncoder.forward": ["BaseModelOutputWithPastAndCrossAttentions", "Cache", "DynamicCache", "EncoderDecoderCache", "False", "FloatTensor", "ModelTextEncoder", "Module", "None", "Setting", "Tensor", "True", "all_cross_attentions", "all_hidden_states", "all_self_attentions", "and", "attention_mask", "attentions", "bool", "cache_position", "checkpointing", "class", "config", "cross_attentions", "def", "elif", "else", "encoder_attention_mask", "encoder_hidden_states", "for", "forward", "gradient", "gradient_checkpointing", "hidden_states", "i", "if", "in", "incompatible", "is", "isinstance", "last_hidden_state", "layer", "layer_module", "layer_outputs", "logger", "nn", "not", "num_hidden_layers", "output_attentions", "output_hidden_states", "past_key_values", "range", "return", "return_dict", "self", "torch", "training", "tuple", "use_cache", "v", "warning", "with"], "blip/modeling_blip_text.py:BlipTextPooler.__init__": ["Linear", "ModelTextPooler", "Module", "Tanh", "__init__", "activation", "class", "config", "def", "dense", "hidden_size", "nn", "self", "super"], "blip/modeling_blip_text.py:BlipTextPooler.forward": ["ModelTextPooler", "Module", "Tensor", "activation", "class", "def", "dense", "first_token_tensor", "forward", "hidden_states", "nn", "pooled_output", "return", "self", "torch"], "blip/modeling_blip_text.py:BlipTextPredictionHeadTransform.__init__": ["ACT2FN", "LayerNorm", "Linear", "ModelTextPredictionHeadTransform", "Module", "__init__", "class", "config", "def", "dense", "else", "eps", "hidden_act", "hidden_size", "if", "isinstance", "layer_norm_eps", "nn", "self", "str", "super", "transform_act_fn"], "blip/modeling_blip_text.py:BlipTextPredictionHeadTransform.forward": ["LayerNorm", "ModelTextPredictionHeadTransform", "Module", "Tensor", "class", "def", "dense", "forward", "hidden_states", "nn", "return", "self", "torch", "transform_act_fn"], "blip/modeling_blip_text.py:BlipTextLMPredictionHead.__init__": ["Linear", "ModelTextLMPredictionHead", "ModelTextPredictionHeadTransform", "Module", "Parameter", "True", "__init__", "bias", "class", "config", "decoder", "def", "hidden_size", "nn", "self", "super", "torch", "transform", "vocab_size", "zeros"], "blip/modeling_blip_text.py:BlipTextLMPredictionHead.forward": ["ModelTextLMPredictionHead", "Module", "class", "decoder", "def", "forward", "hidden_states", "nn", "return", "self", "transform"], "blip/modeling_blip_text.py:BlipTextOnlyMLMHead.__init__": ["ModelTextLMPredictionHead", "ModelTextOnlyMLMHead", "Module", "__init__", "class", "config", "def", "nn", "predictions", "self", "super"], "blip/modeling_blip_text.py:BlipTextOnlyMLMHead.forward": ["ModelTextOnlyMLMHead", "Module", "Tensor", "class", "def", "forward", "nn", "prediction_scores", "predictions", "return", "self", "sequence_output", "torch"], "blip/modeling_blip_text.py:BlipTextPreTrainedModel._init_weights": ["ModelTextEmbeddings", "ModelTextPreTrainedModel", "PreTrainedModel", "_init_weights", "arange", "class", "copy_", "def", "expand", "if", "init", "isinstance", "module", "position_ids", "self", "shape", "super", "torch"], "blip/modeling_blip_text.py:BlipTextModel.__init__": ["ModelTextEmbeddings", "ModelTextEncoder", "ModelTextModel", "ModelTextPooler", "ModelTextPreTrainedModel", "None", "True", "__init__", "add_pooling_layer", "class", "config", "def", "else", "embeddings", "encoder", "if", "pooler", "post_init", "self", "super"], "blip/modeling_blip_text.py:BlipTextModel.get_input_embeddings": ["ModelTextModel", "ModelTextPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "return", "self", "word_embeddings"], "blip/modeling_blip_text.py:BlipTextModel.set_input_embeddings": ["ModelTextModel", "ModelTextPreTrainedModel", "class", "def", "embeddings", "self", "set_input_embeddings", "value", "word_embeddings"], "blip/modeling_blip_text.py:BlipTextModel.get_extended_attention_mask": ["ModelTextModel", "ModelTextPreTrainedModel", "None", "Tensor", "ValueError", "Wrong", "arange", "attention_mask", "axis", "batch_size", "bool", "cat", "causal_mask", "class", "def", "device", "dim", "dtype", "elif", "else", "extended_attention_mask", "f", "for", "get_extended_attention_mask", "if", "input_ids", "input_shape", "int", "is_decoder", "ones", "or", "prefix_seq_len", "raise", "repeat", "return", "self", "seq_ids", "seq_length", "shape", "to", "torch", "tuple"], "blip/modeling_blip_text.py:BlipTextModel.forward": ["BaseModelOutputWithPoolingAndCrossAttentions", "Cache", "False", "ModelTextModel", "ModelTextPreTrainedModel", "None", "Tensor", "ValueError", "You", "_", "and", "at", "attention_mask", "attentions", "batch_size", "bool", "both", "cache_position", "cannot", "class", "config", "cross_attentions", "def", "device", "either", "elif", "else", "embedding_output", "embeddings", "encoder", "encoder_attention_mask", "encoder_batch_size", "encoder_embeds", "encoder_extended_attention_mask", "encoder_hidden_shape", "encoder_hidden_states", "encoder_outputs", "encoder_sequence_length", "extended_attention_mask", "for", "forward", "get_extended_attention_mask", "get_seq_length", "have", "hidden_states", "if", "in", "input_ids", "input_shape", "inputs_embeds", "invert_attention_mask", "is", "is_decoder", "isinstance", "kwargs", "last_hidden_state", "list", "mask", "not", "ones", "or", "output_attentions", "output_hidden_states", "past_key_values", "past_key_values_length", "pooled_output", "pooler", "pooler_output", "position_ids", "r", "raise", "return", "return_dict", "same", "self", "seq_length", "sequence_output", "size", "specify", "the", "time", "to", "torch", "tuple", "use_cache", "use_return_dict", "warn_if_padding_and_no_attention_mask"], "blip/modeling_blip_text.py:BlipTextLMHeadModel.__init__": ["False", "GenerationMixin", "ModelTextLMHeadModel", "ModelTextModel", "ModelTextOnlyMLMHead", "ModelTextPreTrainedModel", "__init__", "add_pooling_layer", "bert", "class", "cls", "config", "def", "label_smoothing", "post_init", "self", "super"], "blip/modeling_blip_text.py:BlipTextLMHeadModel.get_input_embeddings": ["GenerationMixin", "ModelTextLMHeadModel", "ModelTextPreTrainedModel", "bert", "class", "def", "get_input_embeddings", "return", "self"], "blip/modeling_blip_text.py:BlipTextLMHeadModel.set_input_embeddings": ["GenerationMixin", "ModelTextLMHeadModel", "ModelTextPreTrainedModel", "bert", "class", "def", "new_embeddings", "self", "set_input_embeddings"], "blip/modeling_blip_text.py:BlipTextLMHeadModel.get_output_embeddings": ["GenerationMixin", "ModelTextLMHeadModel", "ModelTextPreTrainedModel", "class", "cls", "decoder", "def", "get_output_embeddings", "predictions", "return", "self"], "blip/modeling_blip_text.py:BlipTextLMHeadModel.set_output_embeddings": ["GenerationMixin", "ModelTextLMHeadModel", "ModelTextPreTrainedModel", "bias", "class", "cls", "decoder", "def", "new_embeddings", "predictions", "self", "set_output_embeddings"], "blip/modeling_blip_text.py:BlipTextLMHeadModel.forward": ["Cache", "CausalLMOutputWithCrossAttentions", "CrossEntropyLoss", "False", "GenerationMixin", "ModelTextLMHeadModel", "ModelTextPreTrainedModel", "None", "Tensor", "True", "attention_mask", "attentions", "bert", "bool", "cache_position", "class", "cls", "config", "contiguous", "cross_attentions", "def", "device", "else", "encoder_attention_mask", "encoder_hidden_states", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "is_decoder", "isinstance", "kwargs", "label_smoothing", "labels", "lm_loss", "logits", "logits_to_keep", "loss", "loss_fct", "mean", "none", "not", "output", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "position_ids", "prediction_scores", "r", "reduction", "return", "return_dict", "return_logits", "self", "shifted_prediction_scores", "size", "slice", "slice_indices", "str", "sum", "to", "torch", "tuple", "use_cache", "use_return_dict", "view", "vocab_size"], "blip/modeling_blip_text.py:BlipTextLMHeadModel.prepare_inputs_for_generation": ["GenerationMixin", "ModelTextLMHeadModel", "ModelTextPreTrainedModel", "None", "True", "attention_mask", "class", "def", "input_ids", "is_decoder", "model_inputs", "model_kwargs", "past_key_values", "prepare_inputs_for_generation", "return", "self", "super"], "blip_2/modeling_blip_2.py:Blip2ForConditionalGenerationModelOutput.to_tuple": ["Any", "ModelForConditionalGenerationModelOutput", "ModelOutput", "class", "def", "else", "for", "getattr", "if", "in", "k", "keys", "language_model_outputs", "not", "qformer_outputs", "return", "self", "to_tuple", "tuple", "vision_outputs"], "blip_2/modeling_blip_2.py:Blip2ImageTextMatchingModelOutput.to_tuple": ["Any", "ModelImageTextMatchingModelOutput", "ModelOutput", "class", "def", "else", "for", "getattr", "if", "in", "k", "keys", "not", "return", "self", "text_model_output", "to_tuple", "tuple", "vision_model_output"], "blip_2/modeling_blip_2.py:Blip2VisionEmbeddings.__init__": ["Conv2d", "ModelVisionConfig", "ModelVisionEmbeddings", "Module", "Parameter", "__init__", "class", "class_embedding", "config", "def", "embed_dim", "hidden_size", "image_size", "in_channels", "kernel_size", "nn", "num_patches", "num_positions", "out_channels", "patch_embedding", "patch_size", "position_embedding", "randn", "self", "stride", "super", "torch"], "blip_2/modeling_blip_2.py:Blip2VisionEmbeddings.interpolate_pos_encoding": ["False", "ModelVisionEmbeddings", "Module", "Tensor", "align_corners", "and", "bicubic", "cat", "class", "class_pos_embed", "def", "dim", "embeddings", "functional", "height", "if", "int", "interpolate", "interpolate_pos_encoding", "is_tracing", "jit", "mode", "new_height", "new_width", "nn", "not", "num_patches", "num_positions", "patch_pos_embed", "patch_size", "permute", "position_embedding", "reshape", "return", "self", "shape", "size", "sqrt_num_positions", "torch", "torch_int", "view", "width"], "blip_2/modeling_blip_2.py:Blip2VisionEmbeddings.forward": ["False", "FloatTensor", "ModelVisionEmbeddings", "Module", "Tensor", "_", "batch_size", "bool", "cat", "class", "class_embedding", "class_embeds", "def", "dim", "dtype", "else", "embeddings", "expand", "flatten", "forward", "height", "if", "interpolate_pos_encoding", "nn", "patch_embedding", "patch_embeds", "pixel_values", "position_embedding", "return", "self", "shape", "size", "target_dtype", "to", "torch", "transpose", "weight", "width"], "blip_2/modeling_blip_2.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "attention_mask", "attn_output", "attn_weights", "contiguous", "def", "dim", "dropout", "float", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "softmax", "torch", "training", "transpose", "value"], "blip_2/modeling_blip_2.py:Blip2Attention.__init__": ["False", "Linear", "ModelAttention", "Module", "None", "Parameter", "ValueError", "__init__", "and", "attention_dropout", "be", "bias", "by", "cat", "class", "config", "def", "divisible", "else", "embed_dim", "f", "got", "head_dim", "hidden_size", "if", "is", "is_causal", "must", "nn", "not", "num_attention_heads", "num_heads", "projection", "q_bias", "qkv", "qkv_bias", "raise", "requires_grad", "scale", "self", "super", "torch", "v_bias", "zeros", "zeros_like"], "blip_2/modeling_blip_2.py:Blip2Attention._shape": ["ModelAttention", "Module", "Tensor", "_shape", "bsz", "class", "contiguous", "def", "head_dim", "int", "nn", "num_heads", "return", "self", "seq_len", "tensor", "torch", "transpose", "view"], "blip_2/modeling_blip_2.py:Blip2Attention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "ModelAttention", "Module", "None", "Tensor", "_attn_implementation", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "bsz", "class", "config", "contiguous", "def", "dropout", "eager", "eager_attention_forward", "else", "embed_dim", "forward", "hidden_states", "if", "key_states", "kwargs", "mixed_qkv", "nn", "not", "num_heads", "permute", "projection", "qkv", "query_states", "reshape", "return", "scale", "scaling", "self", "size", "tgt_len", "torch", "training", "tuple", "value_states"], "blip_2/modeling_blip_2.py:Blip2MLP.__init__": ["ACT2FN", "Linear", "ModelMLP", "Module", "__init__", "activation_fn", "class", "config", "def", "fc1", "fc2", "hidden_act", "hidden_size", "intermediate_size", "nn", "self", "super"], "blip_2/modeling_blip_2.py:Blip2MLP.forward": ["ModelMLP", "Module", "Tensor", "activation_fn", "class", "def", "fc1", "fc2", "forward", "hidden_states", "nn", "return", "self", "torch"], "blip_2/modeling_blip_2.py:Blip2EncoderLayer.__init__": ["GradientCheckpointingLayer", "LayerNorm", "ModelAttention", "ModelConfig", "ModelEncoderLayer", "ModelMLP", "__init__", "class", "config", "def", "embed_dim", "eps", "hidden_size", "layer_norm1", "layer_norm2", "layer_norm_eps", "mlp", "nn", "self", "self_attn", "super"], "blip_2/modeling_blip_2.py:Blip2EncoderLayer.forward": ["FloatTensor", "GradientCheckpointingLayer", "ModelEncoderLayer", "Tensor", "TransformersKwargs", "Unpack", "_", "class", "def", "forward", "hidden_states", "kwargs", "layer_norm1", "layer_norm2", "mlp", "residual", "return", "self", "self_attn", "torch"], "blip_2/modeling_blip_2.py:Blip2PreTrainedModel._init_weights": ["ModelForConditionalGeneration", "ModelForImageTextRetrieval", "ModelModel", "ModelPreTrainedModel", "ModelTextEmbeddings", "ModelTextModelWithProjection", "ModelVisionEmbeddings", "ModelVisionModelWithProjection", "PreTrainedModel", "_init_weights", "arange", "class", "class_embedding", "config", "copy_", "def", "elif", "expand", "if", "init", "initializer_range", "isinstance", "mean", "module", "position_embedding", "position_ids", "query_tokens", "self", "shape", "std", "super", "torch", "trunc_normal_", "zeros_"], "blip_2/modeling_blip_2.py:Blip2Encoder.__init__": ["False", "ModelConfig", "ModelEncoder", "ModelEncoderLayer", "Module", "ModuleList", "_", "__init__", "class", "config", "def", "for", "gradient_checkpointing", "in", "layers", "nn", "num_hidden_layers", "range", "self", "super"], "blip_2/modeling_blip_2.py:Blip2Encoder.forward": ["BaseModelOutput", "ModelEncoder", "Module", "TransformersKwargs", "Unpack", "class", "def", "encoder_layer", "for", "forward", "hidden_states", "in", "inputs_embeds", "kwargs", "last_hidden_state", "layers", "nn", "return", "self", "tuple"], "blip_2/modeling_blip_2.py:Blip2VisionModel.__init__": ["LayerNorm", "ModelEncoder", "ModelPreTrainedModel", "ModelVisionConfig", "ModelVisionEmbeddings", "ModelVisionModel", "__init__", "class", "config", "def", "embed_dim", "embeddings", "encoder", "eps", "hidden_size", "layer_norm_eps", "nn", "post_init", "post_layernorm", "self", "super"], "blip_2/modeling_blip_2.py:Blip2VisionModel.forward": ["BaseModelOutput", "BaseModelOutputWithPooling", "False", "FloatTensor", "ModelPreTrainedModel", "ModelVisionModel", "None", "TransformersKwargs", "Unpack", "ValueError", "You", "bool", "class", "def", "embeddings", "encoder", "encoder_outputs", "forward", "have", "hidden_states", "if", "inputs_embeds", "interpolate_pos_encoding", "is", "kwargs", "last_hidden_state", "pixel_values", "pooled_output", "pooler_output", "post_layernorm", "raise", "return", "self", "specify", "to", "torch", "tuple"], "blip_2/modeling_blip_2.py:Blip2VisionModel.get_input_embeddings": ["ModelPreTrainedModel", "ModelVisionModel", "class", "def", "embeddings", "get_input_embeddings", "return", "self"], "blip_2/modeling_blip_2.py:Blip2QFormerMultiHeadAttention.__init__": ["Dropout", "False", "Linear", "ModelQFormerMultiHeadAttention", "Module", "The", "ValueError", "__init__", "a", "all_head_size", "and", "attention", "attention_head_size", "attention_probs_dropout_prob", "class", "config", "d", "def", "dropout", "else", "embedding_size", "encoder_hidden_size", "hasattr", "heads", "hidden", "hidden_size", "if", "int", "is", "is_cross_attention", "key", "multiple", "nn", "not", "num_attention_heads", "number", "of", "query", "raise", "save_attention", "self", "size", "super", "the", "value"], "blip_2/modeling_blip_2.py:Blip2QFormerMultiHeadAttention.save_attn_gradients": ["ModelQFormerMultiHeadAttention", "Module", "attn_gradients", "class", "def", "nn", "save_attn_gradients", "self"], "blip_2/modeling_blip_2.py:Blip2QFormerMultiHeadAttention.get_attn_gradients": ["ModelQFormerMultiHeadAttention", "Module", "attn_gradients", "class", "def", "get_attn_gradients", "nn", "return", "self"], "blip_2/modeling_blip_2.py:Blip2QFormerMultiHeadAttention.save_attention_map": ["ModelQFormerMultiHeadAttention", "Module", "attention_map", "class", "def", "nn", "save_attention_map", "self"], "blip_2/modeling_blip_2.py:Blip2QFormerMultiHeadAttention.get_attention_map": ["ModelQFormerMultiHeadAttention", "Module", "attention_map", "class", "def", "get_attention_map", "nn", "return", "self"], "blip_2/modeling_blip_2.py:Blip2QFormerMultiHeadAttention.transpose_for_scores": ["ModelQFormerMultiHeadAttention", "Module", "attention_head_size", "class", "def", "new_x_shape", "nn", "num_attention_heads", "permute", "return", "self", "size", "transpose_for_scores", "view", "x"], "blip_2/modeling_blip_2.py:Blip2QFormerMultiHeadAttention.forward": ["ModelQFormerMultiHeadAttention", "Module", "None", "Softmax", "TransformersKwargs", "Unpack", "all_head_size", "and", "attention_head_size", "attention_mask", "attention_probs", "attention_probs_dropped", "attention_scores", "class", "context_layer", "contiguous", "def", "dim", "dropout", "dtype", "else", "encoder_attention_mask", "encoder_hidden_states", "forward", "hidden_states", "if", "is", "is_cross_attention", "key", "key_layer", "kwargs", "math", "matmul", "mixed_query_layer", "new_context_layer_shape", "nn", "not", "permute", "query", "query_layer", "register_hook", "return", "save_attention", "save_attention_map", "save_attn_gradients", "self", "size", "sqrt", "to", "torch", "transpose", "transpose_for_scores", "value", "value_layer", "view"], "blip_2/modeling_blip_2.py:Blip2QFormerSelfOutput.__init__": ["Dropout", "LayerNorm", "Linear", "ModelQFormerSelfOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "eps", "hidden_dropout_prob", "hidden_size", "layer_norm_eps", "nn", "self", "super"], "blip_2/modeling_blip_2.py:Blip2QFormerSelfOutput.forward": ["LayerNorm", "ModelQFormerSelfOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "blip_2/modeling_blip_2.py:Blip2QFormerAttention.__init__": ["False", "ModelQFormerAttention", "ModelQFormerMultiHeadAttention", "ModelQFormerSelfOutput", "Module", "__init__", "attention", "class", "config", "def", "is_cross_attention", "nn", "output", "self", "super"], "blip_2/modeling_blip_2.py:Blip2QFormerAttention.forward": ["FloatTensor", "ModelQFormerAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "attention", "attention_mask", "attention_output", "attn_output", "class", "def", "encoder_attention_mask", "encoder_hidden_states", "forward", "hidden_states", "kwargs", "nn", "output", "return", "self", "torch"], "blip_2/modeling_blip_2.py:Blip2QFormerIntermediate.__init__": ["ACT2FN", "Linear", "ModelQFormerIntermediate", "Module", "__init__", "class", "config", "def", "dense", "else", "hidden_act", "hidden_size", "if", "intermediate_act_fn", "intermediate_size", "isinstance", "nn", "self", "str", "super"], "blip_2/modeling_blip_2.py:Blip2QFormerIntermediate.forward": ["ModelQFormerIntermediate", "Module", "Tensor", "class", "def", "dense", "forward", "hidden_states", "intermediate_act_fn", "nn", "return", "self", "torch"], "blip_2/modeling_blip_2.py:Blip2QFormerOutput.__init__": ["Dropout", "LayerNorm", "Linear", "ModelQFormerOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "eps", "hidden_dropout_prob", "hidden_size", "intermediate_size", "layer_norm_eps", "nn", "self", "super"], "blip_2/modeling_blip_2.py:Blip2QFormerOutput.forward": ["LayerNorm", "ModelQFormerOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "blip_2/modeling_blip_2.py:Blip2QFormerLayer.__init__": ["False", "GradientCheckpointingLayer", "ModelQFormerAttention", "ModelQFormerIntermediate", "ModelQFormerLayer", "ModelQFormerOutput", "True", "__init__", "attention", "chunk_size_feed_forward", "class", "config", "cross_attention_frequency", "crossattention", "def", "else", "has_cross_attention", "if", "intermediate", "intermediate_query", "is_cross_attention", "layer_idx", "output", "output_query", "self", "seq_len_dim", "super", "use_qformer_text_input"], "blip_2/modeling_blip_2.py:Blip2QFormerLayer.forward": ["GradientCheckpointingLayer", "ModelQFormerLayer", "None", "TransformersKwargs", "Unpack", "ValueError", "apply_chunking_to_forward", "attention", "attention_mask", "attention_output", "be", "cat", "chunk_size_feed_forward", "class", "cross", "crossattention", "def", "dim", "else", "encoder_attention_mask", "encoder_hidden_states", "feed_forward_chunk", "feed_forward_chunk_query", "for", "forward", "given", "has_cross_attention", "hidden_states", "if", "is", "kwargs", "layer_output", "layer_output_text", "layers", "must", "query_attention_output", "query_length", "raise", "return", "self", "seq_len_dim", "shape", "torch"], "blip_2/modeling_blip_2.py:Blip2QFormerLayer.feed_forward_chunk": ["GradientCheckpointingLayer", "ModelQFormerLayer", "attention_output", "class", "def", "feed_forward_chunk", "intermediate", "intermediate_output", "layer_output", "output", "return", "self"], "blip_2/modeling_blip_2.py:Blip2QFormerLayer.feed_forward_chunk_query": ["GradientCheckpointingLayer", "ModelQFormerLayer", "attention_output", "class", "def", "feed_forward_chunk_query", "intermediate_output", "intermediate_query", "layer_output", "output_query", "return", "self"], "blip_2/modeling_blip_2.py:Blip2QFormerEncoder.__init__": ["False", "ModelQFormerEncoder", "ModelQFormerLayer", "Module", "ModuleList", "__init__", "class", "config", "def", "for", "gradient_checkpointing", "in", "layer", "layer_idx", "nn", "num_hidden_layers", "range", "self", "super"], "blip_2/modeling_blip_2.py:Blip2QFormerEncoder.forward": ["BaseModelOutputWithPastAndCrossAttentions", "ModelQFormerEncoder", "Module", "None", "TransformersKwargs", "Unpack", "attention_mask", "class", "config", "def", "encoder_attention_mask", "encoder_hidden_states", "for", "forward", "hidden_states", "i", "in", "kwargs", "last_hidden_state", "layer", "layer_module", "nn", "num_hidden_layers", "query_length", "range", "return", "self"], "blip_2/modeling_blip_2.py:Blip2TextEmbeddings.__init__": ["Embedding", "False", "ModelTextEmbeddings", "Module", "__init__", "arange", "class", "config", "def", "expand", "hidden_size", "max_position_embeddings", "nn", "pad_token_id", "padding_idx", "persistent", "position_embeddings", "position_ids", "register_buffer", "self", "super", "torch", "vocab_size", "word_embeddings"], "blip_2/modeling_blip_2.py:Blip2TextEmbeddings.forward": ["FloatTensor", "LongTensor", "ModelTextEmbeddings", "Module", "None", "Tensor", "cat", "class", "def", "device", "dim", "dtype", "else", "embeddings", "forward", "if", "input_ids", "is", "nn", "not", "position_embeddings", "position_ids", "query_embeds", "return", "self", "seq_length", "size", "to", "torch", "weight", "word_embeddings"], "blip_2/modeling_blip_2.py:Blip2QFormerModel.__init__": ["Dropout", "LayerNorm", "ModelPreTrainedModel", "ModelQFormerConfig", "ModelQFormerEncoder", "ModelQFormerModel", "__init__", "class", "config", "def", "dropout", "encoder", "eps", "hidden_dropout_prob", "hidden_size", "layer_norm_eps", "layernorm", "nn", "post_init", "self", "super"], "blip_2/modeling_blip_2.py:Blip2QFormerModel.get_input_embeddings": ["ModelPreTrainedModel", "ModelQFormerModel", "None", "class", "def", "get_input_embeddings", "return", "self"], "blip_2/modeling_blip_2.py:Blip2QFormerModel.set_input_embeddings": ["ModelPreTrainedModel", "ModelQFormerModel", "NotImplementedError", "and", "cannot", "class", "def", "does", "embeddings", "input", "not", "own", "raise", "self", "set", "set_input_embeddings", "them", "value"], "blip_2/modeling_blip_2.py:Blip2QFormerModel.get_extended_attention_mask": ["False", "ModelPreTrainedModel", "ModelQFormerModel", "None", "Tensor", "ValueError", "Wrong", "attention_mask", "bool", "class", "def", "device", "dim", "dtype", "elif", "else", "extended_attention_mask", "f", "for", "get_extended_attention_mask", "has_query", "if", "input_ids", "input_shape", "int", "or", "raise", "return", "self", "shape", "to", "torch", "tuple"], "blip_2/modeling_blip_2.py:Blip2QFormerModel.forward": ["BaseModelOutput", "BaseModelOutputWithPoolingAndCrossAttentions", "FloatTensor", "ModelPreTrainedModel", "ModelQFormerModel", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "attention_mask", "batch_size", "class", "def", "device", "dropout", "dtype", "elif", "else", "embedding_output", "encoder", "encoder_attention_mask", "encoder_batch_size", "encoder_extended_attention_mask", "encoder_hidden_shape", "encoder_hidden_states", "encoder_outputs", "encoder_sequence_length", "extended_attention_mask", "for", "forward", "get_extended_attention_mask", "if", "in", "input_shape", "int", "invert_attention_mask", "is", "isinstance", "kwargs", "last_hidden_state", "layernorm", "list", "mask", "not", "ones", "pooled_output", "pooler_output", "query_embeds", "query_length", "r", "return", "self", "seq_length", "sequence_output", "shape", "size", "to", "torch", "tuple", "weight"], "blip_2/modeling_blip_2.py:Blip2Model.__init__": ["AutoModelForCausalLM", "AutoModelForSeq2SeqLM", "Linear", "ModelConfig", "ModelModel", "ModelPreTrainedModel", "ModelQFormerModel", "ModelVisionModel", "Parameter", "__init__", "_from_config", "class", "config", "def", "else", "from_config", "hidden_size", "if", "language_model", "language_projection", "nn", "num_query_tokens", "post_init", "qformer", "qformer_config", "query_tokens", "self", "super", "text_config", "torch", "use_decoder_only_language_model", "vision_config", "vision_model", "zeros"], "blip_2/modeling_blip_2.py:Blip2Model.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "language_model", "return", "self"], "blip_2/modeling_blip_2.py:Blip2Model.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "language_model", "self", "set_input_embeddings", "value"], "blip_2/modeling_blip_2.py:Blip2Model.set_output_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "language_model", "new_embeddings", "self", "set_output_embeddings"], "blip_2/modeling_blip_2.py:Blip2Model.get_output_embeddings": ["ModelModel", "ModelPreTrainedModel", "Module", "class", "def", "get_output_embeddings", "language_model", "nn", "return", "self"], "blip_2/modeling_blip_2.py:Blip2Model.get_encoder": ["ModelModel", "ModelPreTrainedModel", "None", "class", "def", "else", "get_encoder", "if", "is", "language_model", "modality", "return", "self", "super"], "blip_2/modeling_blip_2.py:Blip2Model.get_text_features": ["CausalLMOutputWithPast", "FloatTensor", "ModelModel", "ModelPreTrainedModel", "None", "Seq2SeqLMOutput", "Tensor", "True", "attention_mask", "class", "config", "decoder_attention_mask", "decoder_input_ids", "def", "else", "get_input_embeddings", "get_text_features", "if", "input_ids", "inputs_embeds", "labels", "language_model", "logits", "r", "return", "return_dict", "self", "text_outputs", "torch", "use_decoder_only_language_model"], "blip_2/modeling_blip_2.py:Blip2Model.get_image_features": ["CausalLMOutputWithPast", "False", "FloatTensor", "ModelModel", "ModelPreTrainedModel", "True", "bool", "class", "def", "get_image_features", "interpolate_pos_encoding", "pixel_values", "pooler_output", "r", "return", "return_dict", "self", "torch", "vision_model", "vision_outputs"], "blip_2/modeling_blip_2.py:Blip2Model.get_qformer_features": ["BaseModelOutputWithPooling", "BaseModelOutputWithPoolingAndCrossAttentions", "False", "FloatTensor", "ModelModel", "ModelPreTrainedModel", "True", "bool", "class", "def", "device", "dtype", "encoder_attention_mask", "encoder_hidden_states", "expand", "get_qformer_features", "image_attention_mask", "image_embeds", "interpolate_pos_encoding", "last_hidden_state", "long", "ones", "pixel_values", "qformer", "query_embeds", "query_outputs", "query_tokens", "r", "return", "return_dict", "self", "shape", "size", "torch", "vision_model", "vision_outputs"], "blip_2/modeling_blip_2.py:Blip2Model.get_placeholder_mask": ["FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "all", "class", "config", "def", "device", "dtype", "else", "expand_as", "get_input_embeddings", "get_placeholder_mask", "if", "image_token_id", "input_ids", "inputs_embeds", "is", "long", "return", "self", "special_image_mask", "tensor", "to", "torch", "unsqueeze"], "blip_2/modeling_blip_2.py:Blip2Model.forward": ["CrossEntropyLoss", "False", "FloatTensor", "LongTensor", "ModelForConditionalGenerationModelOutput", "ModelModel", "ModelPreTrainedModel", "None", "TransformersKwargs", "True", "Unpack", "attention_mask", "bool", "class", "config", "contiguous", "decoder_attention_mask", "decoder_input_ids", "def", "device", "dtype", "else", "encoder_attention_mask", "encoder_hidden_states", "expand", "forward", "get_input_embeddings", "get_placeholder_mask", "if", "image_attention_mask", "image_embeds", "input_ids", "inputs_embeds", "interpolate_pos_encoding", "is", "kwargs", "labels", "language_model", "language_model_inputs", "language_model_outputs", "language_projection", "logits", "long", "loss", "loss_fct", "masked_scatter", "mean", "not", "ones", "ones_like", "outputs", "pixel_values", "qformer", "qformer_outputs", "query_embeds", "query_output", "query_outputs", "query_tokens", "r", "reduction", "return", "return_dict", "self", "shape", "shift_labels", "shift_logits", "size", "special_image_mask", "text_config", "to", "torch", "tuple", "use_decoder_only_language_model", "view", "vision_model", "vision_outputs", "vocab_size"], "blip_2/modeling_blip_2.py:Blip2TextModelWithProjection.__init__": ["Linear", "ModelConfig", "ModelPreTrainedModel", "ModelQFormerModel", "ModelTextEmbeddings", "ModelTextModelWithProjection", "Parameter", "__init__", "class", "config", "def", "embeddings", "hidden_size", "image_text_hidden_size", "nn", "num_query_tokens", "post_init", "qformer", "qformer_config", "query_tokens", "self", "super", "text_projection", "torch", "zeros"], "blip_2/modeling_blip_2.py:Blip2TextModelWithProjection.get_input_embeddings": ["ModelPreTrainedModel", "ModelTextModelWithProjection", "class", "def", "embeddings", "get_input_embeddings", "return", "self", "word_embeddings"], "blip_2/modeling_blip_2.py:Blip2TextModelWithProjection.set_input_embeddings": ["ModelPreTrainedModel", "ModelTextModelWithProjection", "class", "def", "embeddings", "self", "set_input_embeddings", "value", "word_embeddings"], "blip_2/modeling_blip_2.py:Blip2TextModelWithProjection.forward": ["ModelPreTrainedModel", "ModelTextModelOutput", "ModelTextModelWithProjection", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "class", "def", "dim", "dtype", "embeddings", "forward", "functional", "hidden_states", "input_ids", "kwargs", "last_hidden_state", "nn", "normalize", "pooled_output", "position_ids", "qformer", "query_embeds", "query_length", "r", "return", "self", "text_embeds", "text_outputs", "text_projection", "to", "torch", "tuple", "weight"], "blip_2/modeling_blip_2.py:Blip2VisionModelWithProjection.__init__": ["Linear", "ModelConfig", "ModelPreTrainedModel", "ModelQFormerModel", "ModelVisionModel", "ModelVisionModelWithProjection", "Parameter", "__init__", "_from_config", "class", "config", "def", "hidden_size", "image_text_hidden_size", "nn", "num_query_tokens", "post_init", "qformer", "qformer_config", "query_tokens", "self", "super", "torch", "vision_config", "vision_model", "vision_projection", "zeros"], "blip_2/modeling_blip_2.py:Blip2VisionModelWithProjection.get_input_embeddings": ["ModelPreTrainedModel", "ModelVisionModelWithProjection", "Module", "class", "def", "embeddings", "get_input_embeddings", "nn", "patch_embedding", "return", "self", "vision_model"], "blip_2/modeling_blip_2.py:Blip2VisionModelWithProjection.forward": ["FloatTensor", "ModelPreTrainedModel", "ModelVisionModelOutput", "ModelVisionModelWithProjection", "None", "TransformersKwargs", "Unpack", "attentions", "class", "def", "device", "dim", "dtype", "embeds", "encoder_attention_mask", "encoder_hidden_states", "expand", "forward", "functional", "hidden_states", "image_attention_mask", "image_embeds", "kwargs", "last_hidden_state", "long", "nn", "normalize", "ones", "pixel_values", "pooled_output", "qformer", "query_embeds", "query_outputs", "query_tokens", "r", "return", "self", "shape", "size", "to", "torch", "tuple", "vision_model", "vision_outputs", "vision_projection", "weight"], "blip_2/modeling_blip_2.py:Blip2ForConditionalGeneration.__init__": ["AutoModelForCausalLM", "AutoModelForSeq2SeqLM", "GenerationMixin", "Linear", "ModelConfig", "ModelForConditionalGeneration", "ModelPreTrainedModel", "ModelQFormerModel", "ModelVisionModel", "Parameter", "__init__", "_from_config", "class", "config", "def", "else", "from_config", "hidden_size", "if", "language_model", "language_projection", "nn", "num_query_tokens", "post_init", "qformer", "qformer_config", "query_tokens", "self", "super", "text_config", "torch", "use_decoder_only_language_model", "vision_config", "vision_model", "zeros"], "blip_2/modeling_blip_2.py:Blip2ForConditionalGeneration.get_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "language_model", "return", "self"], "blip_2/modeling_blip_2.py:Blip2ForConditionalGeneration.set_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "language_model", "self", "set_input_embeddings", "value"], "blip_2/modeling_blip_2.py:Blip2ForConditionalGeneration.set_output_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "language_model", "new_embeddings", "self", "set_output_embeddings"], "blip_2/modeling_blip_2.py:Blip2ForConditionalGeneration.get_output_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "Module", "class", "def", "get_output_embeddings", "language_model", "nn", "return", "self"], "blip_2/modeling_blip_2.py:Blip2ForConditionalGeneration.get_encoder": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "class", "def", "else", "get_encoder", "if", "is", "language_model", "modality", "return", "self", "super"], "blip_2/modeling_blip_2.py:Blip2ForConditionalGeneration._preprocess_accelerate": ["GPU", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "Please", "The", "True", "_hf_hook", "_preprocess_accelerate", "a", "accelerate", "and", "are", "behavior", "blob", "blog", "class", "com", "contains", "creating", "cuda", "def", "details", "device_count", "device_map", "dictionary", "environment", "for", "github", "hasattr", "hf_device_map", "https", "huggingface", "if", "in", "io_same_device", "is", "language_model", "large", "lead", "len", "logger", "main", "may", "md", "models", "more", "multi", "not", "on", "pass", "r", "refer", "remove", "running", "script", "self", "that", "the", "this", "to", "torch", "unexpected", "using", "warning", "when", "you", "your"], "blip_2/modeling_blip_2.py:Blip2ForConditionalGeneration.get_image_features": ["False", "FloatTensor", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "True", "bool", "class", "def", "device", "dtype", "encoder_attention_mask", "encoder_hidden_states", "expand", "get_image_features", "if", "image_attention_mask", "image_embeds", "interpolate_pos_encoding", "language_model_inputs", "language_projection", "long", "ones", "pixel_values", "qformer", "query_embeds", "query_output", "query_outputs", "query_tokens", "return", "return_dict", "self", "shape", "size", "to", "torch", "vision_model", "vision_outputs"], "blip_2/modeling_blip_2.py:Blip2ForConditionalGeneration.get_placeholder_mask": ["FloatTensor", "GenerationMixin", "LongTensor", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "all", "class", "config", "def", "device", "dtype", "else", "expand_as", "get_input_embeddings", "get_placeholder_mask", "if", "image_token_id", "input_ids", "inputs_embeds", "is", "long", "return", "self", "special_image_mask", "tensor", "to", "torch", "unsqueeze"], "blip_2/modeling_blip_2.py:Blip2ForConditionalGeneration.forward": ["CrossEntropyLoss", "False", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForConditionalGeneration", "ModelForConditionalGenerationModelOutput", "ModelPreTrainedModel", "None", "TransformersKwargs", "True", "Unpack", "attention_mask", "bool", "class", "config", "contiguous", "decoder_attention_mask", "decoder_input_ids", "def", "device", "dtype", "else", "forward", "get_image_features", "get_input_embeddings", "get_placeholder_mask", "if", "input_ids", "inputs_embeds", "interpolate_pos_encoding", "is", "kwargs", "labels", "language_model", "language_model_inputs", "language_model_outputs", "logits", "loss", "loss_fct", "masked_scatter", "mean", "not", "ones_like", "outputs", "pixel_values", "qformer_outputs", "query_outputs", "r", "reduction", "return", "return_dict", "self", "shift_labels", "shift_logits", "size", "special_image_mask", "text_config", "to", "torch", "tuple", "use_decoder_only_language_model", "view", "vision_outputs", "vocab_size"], "blip_2/modeling_blip_2.py:Blip2ForConditionalGeneration.generate": ["False", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "True", "_preprocess_accelerate", "all", "attention_mask", "batch_size", "bool", "bos_token_id", "class", "config", "def", "device", "dtype", "else", "encoder_attention_mask", "encoder_hidden_states", "expand", "expand_as", "generate", "generate_kwargs", "get_input_embeddings", "hasattr", "hf_device_map", "if", "image_attention_mask", "image_embeds", "image_token_id", "image_token_index", "image_tokens", "input_ids", "inputs", "inputs_embeds", "interpolate_pos_encoding", "is", "is_encoder_decoder", "language_model", "language_model_inputs", "language_projection", "last_hidden_state", "long", "masked_scatter", "not", "num_query_tokens", "ones", "ones_like", "outputs", "pixel_values", "qformer", "query_embeds", "query_output", "query_outputs", "query_tokens", "repeat", "return", "return_dict", "self", "shape", "size", "special_image_mask", "start_tokens", "tensor", "text_config", "to", "torch", "unsqueeze", "vision_model"], "blip_2/modeling_blip_2.py:Blip2ForImageTextRetrieval.__init__": ["Linear", "ModelConfig", "ModelForImageTextRetrieval", "ModelPreTrainedModel", "ModelQFormerModel", "ModelTextEmbeddings", "ModelVisionModel", "Parameter", "__init__", "_from_config", "class", "config", "def", "embeddings", "hidden_size", "image_text_hidden_size", "itm_head", "nn", "num_query_tokens", "post_init", "qformer", "qformer_config", "query_tokens", "self", "super", "text_projection", "torch", "vision_config", "vision_model", "vision_projection", "zeros"], "blip_2/modeling_blip_2.py:Blip2ForImageTextRetrieval.get_input_embeddings": ["ModelForImageTextRetrieval", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "return", "self", "word_embeddings"], "blip_2/modeling_blip_2.py:Blip2ForImageTextRetrieval.set_input_embeddings": ["ModelForImageTextRetrieval", "ModelPreTrainedModel", "class", "def", "embeddings", "self", "set_input_embeddings", "value", "word_embeddings"], "blip_2/modeling_blip_2.py:Blip2ForImageTextRetrieval.forward": ["False", "FloatTensor", "LongTensor", "ModelForImageTextRetrieval", "ModelImageTextMatchingModelOutput", "ModelPreTrainedModel", "None", "_", "attention_mask", "bool", "cat", "class", "config", "def", "device", "dim", "dtype", "else", "embeddings", "encoder_attention_mask", "encoder_hidden_states", "expand", "forward", "functional", "if", "image_attention_mask", "image_embeds", "image_token_index", "input_ids", "is", "itm_head", "kwargs", "last_hidden_state", "logits_per_image", "logits_per_text", "long", "matmul", "max", "mean", "nn", "normalize", "not", "num_query_tokens", "ones", "output", "output_attentions", "output_hidden_states", "pixel_values", "qformer", "query_attention_mask", "query_embeds", "query_length", "query_outputs", "query_tokens", "question_embeds", "r", "return", "return_dict", "self", "shape", "size", "t", "text_embeds", "text_model_output", "text_outputs", "text_projection", "to", "torch", "tuple", "use_image_text_matching_head", "use_return_dict", "vision_model", "vision_model_output", "vision_outputs", "vision_projection", "weight"], "bloom/modeling_bloom.py:build_alibi_tensor": ["Model_alibi_tensor", "None", "Tensor", "alibi", "arange", "arange_tensor", "attention_mask", "base", "batch_size", "cat", "closest_power_of_2", "cumsum", "def", "device", "dim", "dtype", "extra_base", "extra_powers", "float32", "floor", "if", "int", "int32", "log2", "math", "min", "num_heads", "num_remaining_heads", "pow", "powers", "reshape", "return", "seq_length", "shape", "slopes", "tensor", "to", "torch"], "bloom/modeling_bloom.py:dropout_add": ["F", "Model", "Model_add", "Tensor", "bool", "def", "float", "out", "p", "prob", "residual", "return", "torch", "training", "x"], "bloom/modeling_bloom.py:bloom_gelu_forward": ["Model_gelu_forward", "Tensor", "def", "return", "tanh", "torch", "x"], "bloom/modeling_bloom.py:bloom_gelu_back": ["Model_gelu_back", "Tensor", "def", "ff", "g", "return", "tanh", "tanh_out", "torch", "x"], "bloom/modeling_bloom.py:GeLUFunction.forward": ["Function", "GeLUFunction", "Model_gelu_forward", "Tensor", "autograd", "class", "ctx", "def", "forward", "input", "return", "save_for_backward", "torch"], "bloom/modeling_bloom.py:GeLUFunction.backward": ["Function", "GeLUFunction", "Model_gelu_back", "Tensor", "autograd", "backward", "class", "ctx", "def", "grad_output", "input", "return", "saved_tensors", "tmp", "torch"], "bloom/modeling_bloom.py:BloomGelu.__init__": ["ModelGelu", "Module", "__init__", "class", "def", "nn", "self", "super"], "bloom/modeling_bloom.py:BloomGelu.forward": ["GeLUFunction", "ModelGelu", "Module", "Tensor", "apply", "class", "def", "forward", "nn", "return", "self", "torch", "x"], "bloom/modeling_bloom.py:BloomAttention.__init__": ["Dropout", "Instantiating", "Linear", "ModelAttention", "ModelConfig", "Module", "None", "Please", "True", "ValueError", "__class__", "__init__", "__name__", "a", "and", "attention_dropout", "be", "beta", "bias", "by", "caching", "call", "class", "config", "creating", "def", "dense", "divisible", "during", "errors", "f", "forward", "got", "head_dim", "hidden_dropout", "hidden_size", "if", "int", "inv_norm_factor", "is", "layer_idx", "lead", "logger", "make", "math", "must", "n_head", "nn", "not", "num_heads", "passing", "pretraining_tp", "provide", "query_key_value", "raise", "recommended", "self", "slow_but_exact", "split_size", "sqrt", "super", "sure", "the", "this", "to", "used", "warning_once", "when", "will", "without"], "bloom/modeling_bloom.py:BloomAttention._reshape": ["ModelAttention", "Module", "Tensor", "_reshape", "batch_size", "class", "def", "fused_qkv", "head_dim", "key_layer", "nn", "num_heads", "query_layer", "return", "self", "seq_length", "shape", "three_times_hidden_size", "torch", "transpose", "tuple", "value_layer", "view"], "bloom/modeling_bloom.py:BloomAttention._merge_heads": ["ModelAttention", "Module", "Tensor", "_", "_merge_heads", "batch_size", "batch_size_and_num_heads", "class", "def", "head_dim", "nn", "num_heads", "permute", "reshape", "return", "self", "seq_length", "shape", "torch", "view", "x"], "bloom/modeling_bloom.py:BloomAttention.forward": ["Cache", "F", "False", "LongTensor", "ModelAttention", "Module", "None", "Tensor", "_", "_merge_heads", "_reshape", "alibi", "alpha", "and", "attention_dropout", "attention_mask", "attention_probs", "attention_probs_reshaped", "attention_scores", "attn_weights", "baddbmm", "batch1", "batch2", "batch_size", "beta", "bmm", "bool", "cache_kwargs", "cache_position", "causal_mask", "class", "context_layer", "def", "dense", "dim", "dropout_add", "dtype", "else", "float32", "for", "forward", "fused_qkv", "head_dim", "hidden_dropout", "hidden_size", "hidden_states", "i", "if", "in", "int", "inv_norm_factor", "is", "key_layer", "layer_idx", "layer_past", "linear", "nn", "not", "num_heads", "output_attentions", "output_tensor", "pretraining_tp", "q_length", "query_key_value", "query_layer", "range", "reshape", "residual", "return", "self", "shape", "slices", "slow_but_exact", "softmax", "to", "torch", "training", "transpose", "update", "use_cache", "value_layer", "view", "weight", "zeros_like"], "bloom/modeling_bloom.py:BloomMLP.__init__": ["Linear", "ModelConfig", "ModelGelu", "ModelMLP", "Module", "__init__", "class", "config", "def", "dense_4h_to_h", "dense_h_to_4h", "gelu_impl", "hidden_dropout", "hidden_size", "nn", "pretraining_tp", "self", "slow_but_exact", "super"], "bloom/modeling_bloom.py:BloomMLP.forward": ["F", "ModelMLP", "Module", "Tensor", "and", "class", "def", "dense_4h_to_h", "dense_h_to_4h", "dropout_add", "else", "for", "forward", "gelu_impl", "hidden_dropout", "hidden_states", "i", "if", "in", "int", "intermediate_output", "linear", "nn", "output", "pretraining_tp", "range", "residual", "return", "self", "shape", "slices", "slow_but_exact", "torch", "training", "weight", "zeros_like"], "bloom/modeling_bloom.py:BloomBlock.__init__": ["GradientCheckpointingLayer", "LayerNorm", "ModelAttention", "ModelBlock", "ModelConfig", "ModelMLP", "None", "__init__", "apply_residual_connection_post_layernorm", "class", "config", "def", "eps", "hidden_dropout", "hidden_size", "input_layernorm", "int", "layer_idx", "layer_norm_epsilon", "mlp", "n_head", "num_heads", "post_attention_layernorm", "self", "self_attention", "super"], "bloom/modeling_bloom.py:BloomBlock.forward": ["Cache", "False", "GradientCheckpointingLayer", "LongTensor", "ModelBlock", "None", "Tensor", "alibi", "apply_residual_connection_post_layernorm", "attention_mask", "attention_output", "attn_weights", "bool", "cache_position", "class", "def", "else", "forward", "hidden_states", "if", "input_layernorm", "layer_past", "layernorm_output", "mlp", "output", "output_attentions", "post_attention_layernorm", "residual", "return", "self", "self_attention", "torch", "use_cache"], "bloom/modeling_bloom.py:BloomModel.__init__": ["Embedding", "False", "LayerNorm", "ModelBlock", "ModelConfig", "ModelModel", "ModelPreTrainedModel", "ModuleList", "__init__", "class", "config", "def", "embed_dim", "eps", "for", "gradient_checkpointing", "h", "hidden_size", "i", "in", "layer_idx", "layer_norm_epsilon", "ln_f", "n_head", "nn", "num_heads", "num_hidden_layers", "post_init", "range", "self", "super", "vocab_size", "word_embeddings", "word_embeddings_layernorm"], "bloom/modeling_bloom.py:BloomModel.build_alibi_tensor": ["ModelModel", "ModelPreTrainedModel", "Tensor", "attention_mask", "build_alibi_tensor", "class", "def", "dtype", "int", "num_heads", "return", "self", "torch"], "bloom/modeling_bloom.py:BloomModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "return", "self", "word_embeddings"], "bloom/modeling_bloom.py:BloomModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "Tensor", "class", "def", "new_embeddings", "self", "set_input_embeddings", "torch", "word_embeddings"], "bloom/modeling_bloom.py:BloomModel.forward": ["BaseModelOutputWithPastAndCrossAttentions", "Cache", "DynamicCache", "False", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Setting", "Tensor", "True", "ValueError", "You", "_", "_update_causal_mask", "alibi", "all_hidden_states", "all_self_attentions", "and", "arange", "attention_mask", "attentions", "batch_size", "block", "bool", "build_alibi_tensor", "cache_position", "causal_mask", "checkpointing", "class", "config", "def", "device", "dtype", "else", "enumerate", "exactly", "for", "forward", "get_seq_length", "gradient", "gradient_checkpointing", "h", "hidden_states", "i", "if", "in", "incompatible", "input_ids", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layer_past", "ln_f", "logger", "must", "not", "num_heads", "of", "one", "ones", "or", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "past_length", "r", "raise", "return", "return_dict", "self", "seq_length", "seq_length_with_past", "shape", "specify", "to", "torch", "training", "tuple", "use_cache", "use_return_dict", "v", "warning_once", "with", "word_embeddings", "word_embeddings_layernorm"], "bloom/modeling_bloom.py:BloomModel._update_causal_mask": ["AttentionMaskConverter", "BlockMask", "Cache", "False", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "Union", "_attn_implementation", "_ignore_causal_mask_sdpa", "_prepare_4d_causal_attention_mask_with_cache_position", "_unmask_unattended", "_update_causal_mask", "and", "any", "attention_mask", "batch_size", "bool", "cache_position", "causal_mask", "class", "config", "cuda", "def", "device", "dtype", "else", "finfo", "flash_attention_2", "flex_attention", "get_max_cache_shape", "get_seq_length", "if", "in", "input_tensor", "inputs_embeds", "is", "is_compileable", "is_training", "isinstance", "make_flex_block_causal_mask", "min", "min_dtype", "not", "npu", "output_attentions", "past_key_values", "past_key_values_length", "past_seen_tokens", "return", "sdpa", "self", "sequence_length", "shape", "target_length", "torch", "training", "type", "using_compilable_cache", "xpu"], "bloom/modeling_bloom.py:BloomModel._prepare_4d_causal_attention_mask_with_cache_position": ["ModelModel", "ModelPreTrainedModel", "None", "Tensor", "_prepare_4d_causal_attention_mask_with_cache_position", "and", "arange", "attention_mask", "batch_size", "cache_position", "causal_mask", "class", "clone", "def", "device", "diagonal", "dim", "dtype", "else", "expand", "fill_value", "finfo", "full", "if", "int", "is", "kwargs", "mask_length", "masked_fill", "min", "min_dtype", "not", "padding_mask", "reshape", "return", "sequence_length", "shape", "target_length", "to", "torch", "triu"], "bloom/modeling_bloom.py:BloomForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "ModelConfig", "ModelForCausalLM", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "nn", "post_init", "self", "super", "transformer", "vocab_size"], "bloom/modeling_bloom.py:BloomForCausalLM.set_output_embeddings": ["GenerationMixin", "ModelForCausalLM", "ModelPreTrainedModel", "Tensor", "class", "def", "lm_head", "new_embeddings", "self", "set_output_embeddings", "torch"], "bloom/modeling_bloom.py:BloomForCausalLM.prepare_inputs_for_generation": ["False", "GenerationMixin", "ModelForCausalLM", "ModelPreTrainedModel", "None", "StaticCache", "True", "and", "attention_mask", "batch_size", "cache_position", "cat", "class", "def", "device", "diff", "dim", "dtype", "get_max_cache_shape", "if", "input_ids", "inputs_embeds", "is", "is_first_iteration", "isinstance", "kwargs", "model_inputs", "new_attn_mask", "not", "past_key_values", "prepare_inputs_for_generation", "return", "self", "seq_length", "shape", "super", "target_length", "torch", "use_cache", "zeros"], "bloom/modeling_bloom.py:BloomForCausalLM.forward": ["Cache", "CausalLMOutputWithCrossAttentions", "GenerationMixin", "LongTensor", "ModelForCausalLM", "ModelPreTrainedModel", "None", "Tensor", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "else", "forward", "get", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "not", "num_items_in_batch", "output", "output_attentions", "output_hidden_states", "past_key_values", "r", "return", "return_dict", "self", "slice", "slice_indices", "torch", "transformer", "transformer_outputs", "tuple", "use_cache", "use_return_dict", "vocab_size"], "bloom/modeling_bloom.py:BloomForSequenceClassification.__init__": ["False", "Linear", "ModelConfig", "ModelForSequenceClassification", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "nn", "num_labels", "post_init", "score", "self", "super", "transformer"], "bloom/modeling_bloom.py:BloomForSequenceClassification.forward": ["BCEWithLogitsLoss", "Cache", "Cannot", "CrossEntropyLoss", "LongTensor", "MSELoss", "ModelForSequenceClassification", "ModelPreTrainedModel", "None", "Results", "SequenceClassifierOutputWithPast", "Tensor", "ValueError", "__class__", "__name__", "and", "arange", "argmax", "attention_mask", "attentions", "batch", "batch_size", "be", "bool", "class", "config", "conjunction", "def", "defined", "detect", "device", "dtype", "elif", "else", "f", "forward", "handle", "hidden_states", "if", "in", "input_ids", "inputs_embeds", "int", "int32", "is", "kwargs", "labels", "last_non_pad_token", "logger", "logits", "long", "loss", "loss_fct", "may", "multi_label_classification", "no", "non_pad_mask", "not", "num_labels", "or", "output", "output_attentions", "output_hidden_states", "pad_token_id", "padding", "past_key_values", "pooled_logits", "problem_type", "r", "raise", "regression", "return", "return_dict", "score", "self", "shape", "single_label_classification", "sizes", "squeeze", "to", "token", "token_indices", "tokens", "torch", "transformer", "transformer_outputs", "tuple", "unexpected", "use_cache", "use_return_dict", "using", "warning_once", "will", "with"], "bloom/modeling_bloom.py:BloomForTokenClassification.__init__": ["Dropout", "Linear", "ModelConfig", "ModelForTokenClassification", "ModelModel", "ModelPreTrainedModel", "None", "__init__", "and", "class", "classifier", "classifier_dropout", "config", "def", "dropout", "elif", "else", "hasattr", "hidden_dropout", "hidden_size", "if", "is", "nn", "not", "num_labels", "post_init", "self", "super", "transformer"], "bloom/modeling_bloom.py:BloomForTokenClassification.forward": ["Cache", "CrossEntropyLoss", "LongTensor", "ModelForTokenClassification", "ModelPreTrainedModel", "None", "Tensor", "TokenClassifierOutput", "attention_mask", "attentions", "batch_size", "bool", "class", "classifier", "config", "def", "device", "dropout", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "not", "num_labels", "output", "output_attentions", "output_hidden_states", "past_key_values", "r", "return", "return_dict", "self", "seq_length", "shape", "to", "torch", "transformer", "transformer_outputs", "tuple", "use_cache", "use_return_dict", "view"], "bloom/modeling_bloom.py:BloomForQuestionAnswering.__init__": ["Linear", "ModelForQuestionAnswering", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "config", "def", "hidden_size", "nn", "post_init", "qa_outputs", "self", "super", "transformer"], "bloom/modeling_bloom.py:BloomForQuestionAnswering.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "ModelForQuestionAnswering", "ModelPreTrainedModel", "None", "QuestionAnsweringModelOutput", "and", "attention_mask", "attentions", "bool", "clamp", "class", "config", "contiguous", "def", "dim", "else", "end_logits", "end_loss", "end_positions", "forward", "hidden_states", "if", "ignore_index", "ignored_index", "input_ids", "inputs_embeds", "is", "kwargs", "len", "logits", "loss", "loss_fct", "not", "output", "output_attentions", "output_hidden_states", "outputs", "qa_outputs", "r", "return", "return_dict", "self", "sequence_output", "size", "split", "squeeze", "start_logits", "start_loss", "start_positions", "torch", "total_loss", "transformer", "tuple", "use_return_dict"], "blt/modeling_blt.py:BltMLP.__init__": ["ACT2FN", "False", "Linear", "ModelMLP", "Module", "__init__", "act_fn", "bias", "class", "config", "def", "down_proj", "gate_proj", "hidden_act", "hidden_size", "intermediate_size", "nn", "self", "super", "up_proj"], "blt/modeling_blt.py:BltMLP.forward": ["ModelMLP", "Module", "act_fn", "class", "def", "down_proj", "forward", "gate_proj", "nn", "return", "self", "up_proj", "x"], "blt/modeling_blt.py:BltRMSNorm.__init__": ["ModelRMSNorm", "Module", "Parameter", "__init__", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "blt/modeling_blt.py:BltRMSNorm.forward": ["ModelRMSNorm", "Module", "True", "class", "def", "dtype", "float32", "forward", "hidden_states", "input_dtype", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "blt/modeling_blt.py:BltRMSNorm.extra_repr": ["ModelRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "variance_epsilon", "weight"], "blt/modeling_blt.py:BltRotaryEmbedding.__init__": ["Callable", "False", "ModelConfig", "ModelRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "blt/modeling_blt.py:BltRotaryEmbedding.compute_default_rope_parameters": ["ModelConfig", "ModelRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "blt/modeling_blt.py:BltRotaryEmbedding.forward": ["False", "ModelRotaryEmbedding", "Module", "None", "and", "attention_scaling", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "repeat_interleave", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "blt/modeling_blt.py:BltTransformerLayer.__init__": ["GradientCheckpointingLayer", "ModelMLP", "ModelRMSNorm", "ModelSelfAttention", "ModelTransformerLayer", "__init__", "class", "config", "def", "eps", "hidden_size", "input_layernorm", "int", "layer_idx", "mlp", "post_attention_layernorm", "rms_norm_eps", "self", "self_attn", "super"], "blt/modeling_blt.py:BltTransformerLayer.forward": ["Cache", "False", "FlashAttentionKwargs", "FloatTensor", "GradientCheckpointingLayer", "LongTensor", "ModelTransformerLayer", "None", "Tensor", "Unpack", "attention_mask", "bool", "cache_position", "class", "cross_attention_mask", "cross_attention_states", "def", "forward", "full_text_row_masked_out_mask", "hidden_states", "input_layernorm", "kwargs", "mlp", "past_key_values", "position_embeddings", "position_ids", "post_attention_layernorm", "residual", "return", "self", "self_attn", "self_attn_weights", "torch", "tuple", "use_cache"], "blt/modeling_blt.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "blt/modeling_blt.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value", "value_states"], "blt/modeling_blt.py:rotate_half": ["Model_half", "def", "dim", "flatten", "return", "rot_x", "stack", "torch", "x", "x1", "x2"], "blt/modeling_blt.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cos", "def", "k", "k_embed", "q", "q_embed", "return", "rotate_half", "sin", "unsqueeze", "unsqueeze_dim"], "blt/modeling_blt.py:BltSelfAttention.__init__": ["False", "Linear", "ModelConfig", "ModelSelfAttention", "Module", "True", "__init__", "bias", "class", "config", "def", "dropout", "head_dim", "hidden_size", "int", "is_causal", "k_proj", "layer_idx", "nn", "num_attention_heads", "num_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "q_proj", "scaling", "self", "super", "v_proj"], "blt/modeling_blt.py:BltSelfAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "ModelSelfAttention", "Module", "None", "Tensor", "_", "_attn_implementation", "apply_rotary_pos_emb", "attention_interface", "attention_mask", "attn_output", "attn_weights", "bsz", "cache_kwargs", "cache_position", "class", "config", "contiguous", "cos", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "head_dim", "hidden_states", "if", "is", "k_proj", "key_states", "kwargs", "layer_idx", "nn", "not", "num_heads", "num_key_value_heads", "o_proj", "past_key_values", "position_embeddings", "q_len", "q_proj", "query_states", "reshape", "return", "scaling", "self", "sin", "size", "torch", "training", "transpose", "update", "v_proj", "value_states", "view"], "blt/modeling_blt.py:BltCrossAttention.__init__": ["False", "Linear", "ModelConfig", "ModelCrossAttention", "ModelRMSNorm", "Module", "None", "__init__", "bias", "class", "config", "def", "dropout", "eps", "head_dim", "hidden_size", "int", "is_causal", "k_norm", "k_proj", "layer_idx", "nn", "num_attention_heads", "num_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "q_norm", "q_proj", "rms_norm_eps", "scaling", "self", "super", "v_proj"], "blt/modeling_blt.py:BltCrossAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "ModelCrossAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "_attn_implementation", "attention_interface", "attention_mask", "attn_output", "attn_weights", "bsz", "class", "config", "contiguous", "cross_attention_states", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "head_dim", "hidden_states", "if", "k_norm", "k_proj", "key_states", "kwargs", "nn", "not", "num_heads", "num_key_value_heads", "o_proj", "q_len", "q_norm", "q_proj", "query_states", "reshape", "return", "scaling", "self", "size", "torch", "training", "transpose", "tuple", "v_proj", "value_states", "view"], "blt/modeling_blt.py:BltPreTrainedModel._init_weights": ["Embedding", "LayerNorm", "Linear", "MllamaTextCrossAttention", "MllamaTextMLP", "MllamaTextSelfAttention", "ModelCrossAttention", "ModelMLP", "ModelPreTrainedModel", "ModelRMSNorm", "ModelRotaryEmbedding", "ModelSelfAttention", "None", "PreTrainedModel", "RMSNorm", "ROPE_INIT_FUNCTIONS", "_", "__class__", "__name__", "_init_weights", "a", "and", "b", "bias", "break", "buffer_value", "class", "class_name", "compute_default_rope_parameters", "config", "copy_", "decoder_config", "def", "default", "dense", "dim", "down_proj", "else", "embedding_dim", "encoder_config", "fan_in", "fc1", "fc2", "for", "gate_proj", "getattr", "hasattr", "hidden_dim", "hidden_size", "if", "in", "in_features", "in_std", "init", "inv_freq", "is", "isinstance", "k_proj", "mean", "module", "name", "nn", "not", "o_proj", "ones_", "or", "original_inv_freq", "out_std", "padding_idx", "proj", "proj_name", "q_proj", "return", "rope_fn", "rope_type", "self", "shape", "std", "trunc_normal_", "up_proj", "v_proj", "weight", "zeros_"], "blt/modeling_blt.py:BltLocalEncoder.__init__": ["Embedding", "False", "Linear", "ModelCrossAttention", "ModelLocalEncoder", "ModelLocalEncoderConfig", "ModelPreTrainedModel", "ModelRotaryEmbedding", "ModelTransformerLayer", "ModuleList", "__init__", "append", "bias", "class", "config", "cross_attn_all_layers", "cross_attn_k", "cross_attn_layers", "def", "else", "embed_tokens", "for", "gradient_checkpointing", "hidden_size", "if", "in", "in_features", "layer_idx", "layers", "layers_to_add", "nn", "num_hidden_layers", "out_features", "patch_embedding_projection", "post_init", "range", "rotary_emb", "self", "super", "vocab_size"], "blt/modeling_blt.py:BltLocalEncoder.forward": ["Cache", "F", "LongTensor", "ModelLocalEncoder", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "arange", "attention_mask", "batch_size", "cache_position", "class", "config", "cross_attention_output", "cross_attention_states", "cross_attn_all_layers", "cross_attn_k", "cross_attn_layers", "def", "device", "dropout", "else", "embed_tokens", "encoder_attention_mask", "encoder_cross_states", "enumerate", "expand", "for", "forward", "hidden_size", "hidden_states", "idx", "if", "in", "input_ids", "inputs_embeds", "int", "is", "kwargs", "layer", "layer_idx", "layers", "len", "num_patches", "or", "p", "past_key_values", "patch_embedding_projection", "patch_embeds", "patch_ids", "patch_reduce", "position_embeddings", "position_ids", "reshape", "return", "rotary_emb", "self", "shape", "torch", "training", "unsqueeze"], "blt/modeling_blt.py:BltLocalEncoder.patch_reduce": ["False", "ModelLocalEncoder", "ModelPreTrainedModel", "amax", "batch_size", "class", "def", "device", "dim", "dtype", "embedding_dim", "expand", "hidden_states", "include_self", "index", "max_num_patches", "patch_ids", "patch_reduce", "reduce", "reduced_embeddings", "return", "scatter_reduce", "self", "shape", "src", "torch", "unsqueeze", "zeros"], "blt/modeling_blt.py:BltLocalDecoder.__init__": ["False", "Linear", "ModelCrossAttention", "ModelLocalDecoder", "ModelLocalDecoderConfig", "ModelPreTrainedModel", "ModelRMSNorm", "ModelRotaryEmbedding", "ModelTransformerLayer", "ModuleList", "True", "__init__", "append", "bias", "class", "config", "cross_attn_all_layers", "cross_attn_decoder", "cross_attn_k", "cross_attn_layers", "def", "else", "eps", "for", "gradient_checkpointing", "hidden_size", "hidden_size_global", "if", "in", "in_features", "layer_idx", "layers", "layers_to_add", "nn", "norm", "num_hidden_layers", "out_features", "patch_embedding_projection", "post_init", "range", "rms_norm_eps", "rotary_emb", "self", "super"], "blt/modeling_blt.py:BltLocalDecoder.forward": ["Cache", "F", "LongTensor", "ModelLocalDecoder", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "and", "arange", "attention_mask", "batch_size", "cache_position", "class", "config", "cross_attention_output", "cross_attention_states", "cross_attn_all_layers", "cross_attn_decoder", "cross_attn_k", "cross_attn_layers", "def", "device", "dropout", "encoder_attention_mask", "enumerate", "expand", "for", "forward", "hidden_size", "hidden_states", "i", "if", "in", "input_ids", "inputs_embeds", "is", "kwargs", "layer", "layers", "logits", "norm", "not", "or", "p", "past_key_values", "patch_embedding_projection", "patch_embeds", "position_embeddings", "position_ids", "reshape", "return", "rotary_emb", "self", "shape", "torch", "training", "unsqueeze"], "blt/modeling_blt.py:BltGlobalTransformer.__init__": ["False", "Identity", "Linear", "ModelGlobalTransformer", "ModelGlobalTransformerConfig", "ModelPreTrainedModel", "ModelRotaryEmbedding", "ModelTransformerLayer", "ModuleList", "None", "__init__", "append", "bias", "class", "config", "def", "else", "encoder_cross_output_size", "for", "getattr", "hidden_size", "if", "in", "is", "layer_idx", "layers", "nn", "not", "num_hidden_layers", "post_init", "range", "rotary_emb", "self", "super", "token_embedding_projection"], "blt/modeling_blt.py:BltGlobalTransformer.forward": ["Cache", "F", "LongTensor", "ModelGlobalTransformer", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "arange", "attention_mask", "batch_size", "cache_position", "class", "config", "def", "device", "dropout", "enumerate", "expand", "for", "forward", "hidden_states", "i", "if", "in", "input_embeds", "is", "kwargs", "layer", "layers", "p", "past_key_values", "position_embeddings", "position_ids", "return", "rotary_emb", "self", "seq_len", "shape", "token_embedding_projection", "torch", "training", "unsqueeze"], "blt/modeling_blt.py:process_patch_lengths": ["Model_patch_lengths", "Modeled", "None", "Tensor", "any", "append", "batch_size", "def", "device", "dim", "divmod", "dtype", "enumerate", "extend", "for", "full_chunks", "i", "if", "in", "int", "is", "item", "last_nonzero", "len", "length", "max", "max_len", "max_patch_length", "nonzero", "padded", "patch_lengths", "remainder", "return", "seq", "shape", "size", "splits", "sum", "tensor", "torch", "zeros"], "blt/modeling_blt.py:BltPatcher.__init__": ["Embedding", "False", "Linear", "ModelPatcher", "ModelPatcherConfig", "ModelPreTrainedModel", "ModelRMSNorm", "ModelRotaryEmbedding", "ModelTransformerLayer", "ModuleList", "__init__", "append", "bias", "class", "config", "def", "embed_tokens", "eps", "for", "hidden_size", "in", "layer_idx", "layers", "lm_head", "nn", "norm", "num_hidden_layers", "post_init", "range", "rms_norm_eps", "rotary_emb", "self", "super", "vocab_size"], "blt/modeling_blt.py:BltPatcher.forward": ["Cache", "Categorical", "DynamicCache", "FloatTensor", "LongTensor", "ModelPatcher", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "and", "arange", "attention_mask", "batch_size", "bool", "cache_position", "causal_mask", "class", "config", "create_causal_mask", "def", "device", "distributions", "dtype", "else", "embed_tokens", "entropies", "entropy", "exactly", "float", "for", "forward", "get_seq_length", "hidden_states", "if", "in", "input_embeds", "input_ids", "inputs_embeds", "int", "is", "kwargs", "layer", "layers", "lm_head", "logits", "max_patch_length", "must", "norm", "not", "of", "one", "ones", "or", "past_key_values", "past_seen_tokens", "patch_lengths", "patch_lengths_from_entropies", "patch_size", "position_embeddings", "position_ids", "prediction_entropies", "process_patch_lengths", "raise", "return", "rotary_emb", "self", "sequence_length", "shape", "specify", "threshold", "torch", "unsqueeze", "use_cache"], "blt/modeling_blt.py:BltPatcher.patch_lengths_from_entropies": ["ModelPatcher", "ModelPreTrainedModel", "None", "arange", "batch_size", "cat", "class", "def", "device", "dim", "dtype", "entropies", "expand", "full_like", "init_tokens", "last_token", "long", "max", "max_valid_patches", "offset", "padded_indices", "padded_mask", "patch_ends", "patch_lengths", "patch_lengths_from_entropies", "patch_mask", "patch_size", "patch_start_ids", "patch_starts", "repeat", "reshape", "return", "sentinel", "seq_len", "sequence_length", "shape", "sum", "tensor", "threshold", "token_indices", "torch", "unsqueeze"], "blt/modeling_blt.py:rolling_polynomial_hash": ["Model_polynomial_hash", "arange", "def", "device", "dim", "dtype", "int", "int64", "powers", "prime", "prime_powers", "prime_tensor", "return", "shape", "sum", "tensor", "token_tensor", "torch"], "blt/modeling_blt.py:byte_group_hash_function": ["Model_group_hash_function", "Tensor", "batch_size", "cat", "def", "device", "dim", "dtype", "group_size", "hash_values", "hashes", "int", "int64", "max_hash", "no_grad", "padded_tokens", "padding", "prime", "return", "rolling_polynomial_hash", "seq_len", "shape", "token_ids", "torch", "unfold", "windows", "with", "zeros"], "blt/modeling_blt.py:compute_hash_embeddings": ["Embedding", "Model_hash_embeddings", "Tensor", "byte_group_hash_function", "def", "device", "embed_tokens", "embedding_idx", "embeddings", "encoder_hash_byte_group_nb_functions", "encoder_hash_byte_group_size", "encoder_hash_byte_group_vocab", "encoder_hash_tok_embedding", "for", "func_nb", "group_size", "hash_ids", "in", "int", "len", "list", "local_encoder", "local_encoder_tokens", "nn", "offset_hash_ids", "prime", "primes", "range", "return", "to", "torch"], "blt/modeling_blt.py:_prepare_patch_cross_attention_mask": ["Cross", "False", "Tensor", "ValueError", "_prepare_patch_cross_attention_mask", "arange", "attention", "batch_size", "bool", "cross_attention_mask", "cross_attn_k", "def", "device", "dim", "doesn", "dtype", "else", "expand", "expected", "expected_shape", "f", "finfo", "float32", "if", "int", "inverted_cross_attn_mask", "kv_len", "kv_patch_ids", "mask", "masked_fill", "match", "min", "num_patches", "patch_ids", "patches_as_queries", "q_len", "q_patch_ids", "raise", "repeat_dim", "repeat_interleave", "return", "seq_len", "sequence_length", "shape", "t", "to", "torch", "tuple", "unsqueeze"], "blt/modeling_blt.py:BltModel.__init__": ["Embedding", "False", "ModelConfig", "ModelGlobalTransformer", "ModelLocalDecoder", "ModelLocalEncoder", "ModelModel", "ModelPatcher", "ModelPreTrainedModel", "None", "__init__", "class", "config", "decoder_config", "def", "else", "encoder_config", "encoder_hash_byte_group_nb_functions", "encoder_hash_byte_group_size", "encoder_hash_byte_group_vocab", "encoder_hash_tok_embedding", "eval", "for", "global_config", "global_transformer", "gradient_checkpointing", "hidden_size", "if", "in", "len", "local_decoder", "local_encoder", "nn", "num_embeddings", "param", "parameters", "patch_in_forward", "patcher", "patcher_config", "post_init", "requires_grad", "self", "super", "total_vocab_size"], "blt/modeling_blt.py:BltModel.forward": ["BaseModelOutputWithPast", "Cache", "DynamicCache", "EncoderDecoderCache", "False", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "ValueError", "You", "_", "_patch_ids_from_lengths", "_prepare_patch_cross_attention_mask", "and", "arange", "attention_mask", "based", "batch_size", "bool", "cache_position", "causal_mask", "class", "compute_hash_embeddings", "config", "create_causal_mask", "cross_attention_cache", "cross_attn_k", "cross_attn_mask_dec", "cross_attn_mask_enc", "decoder_patch_ids", "def", "device", "dtype", "else", "encoder_attention_mask", "encoder_cross_states", "encoder_embeds", "encoder_hash_byte_group_nb_functions", "encoder_hash_byte_group_size", "encoder_hash_byte_group_vocab", "encoder_hash_tok_embedding", "encoder_hidden_states", "entropy", "exactly", "for", "forward", "get_seq_length", "global_cache_position", "global_causal_mask", "global_hidden_states", "global_position_ids", "global_transformer", "if", "input_embeds", "input_ids", "inputs_embeds", "is", "kwargs", "last_hidden_state", "local_decoder", "local_encoder", "max_patch_length", "must", "not", "num_patches", "of", "one", "ones", "or", "output", "past_key_values", "past_seen_tokens", "patch_embeds", "patch_ids", "patch_lengths", "patch_size", "patcher", "patches_as_queries", "patching", "patching_batch_size", "patching_mode", "patching_threshold", "position_ids", "process_patch_lengths", "raise", "required", "return", "self", "self_attention_cache", "sequence_length", "shape", "specify", "threshold", "torch", "unsqueeze", "use_cache", "view"], "blt/modeling_blt.py:BltModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embed_tokens", "get_input_embeddings", "local_encoder", "return", "self"], "blt/modeling_blt.py:BltModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embed_tokens", "local_encoder", "self", "set_input_embeddings", "value"], "blt/modeling_blt.py:BltModel._patch_ids_from_lengths": ["ModelModel", "ModelPreTrainedModel", "Tensor", "_patch_ids_from_lengths", "arange", "batch_size", "cat", "class", "cumsum", "def", "device", "dim", "dtype", "int", "patch_lengths", "patch_starts", "return", "self", "seq_len", "shape", "sum", "token_positions", "torch", "unsqueeze", "zeros"], "blt/modeling_blt.py:BltForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "ModelConfig", "ModelForCausalLM", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "decoder_config", "def", "get_text_config", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "text_config", "vocab_size"], "blt/modeling_blt.py:BltForCausalLM.forward": ["Cache", "CausalLMOutputWithPast", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForCausalLM", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "bool", "cache_position", "class", "cross_attention_mask", "cross_attention_states", "def", "else", "float", "forward", "full_text_row_masked_out_mask", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "outputs", "past_key_values", "position_ids", "r", "return", "self", "slice", "slice_indices", "torch", "tuple", "use_cache", "vocab_size"], "bridgetower/modeling_bridgetower.py:BridgeTowerResidualAttention.__init__": ["LayerNorm", "Linear", "ModelResidualAttention", "Module", "ModuleDict", "MultiheadAttention", "None", "OrderedDict", "QuickGELUActivation", "__init__", "attn", "attn_mask", "c_fc", "c_proj", "class", "config", "def", "eps", "gelu", "hidden_size", "layer_norm_eps", "ln_1", "ln_2", "mlp", "nn", "self", "super"], "bridgetower/modeling_bridgetower.py:BridgeTowerResidualAttention.attention": ["False", "ModelResidualAttention", "Module", "None", "Tensor", "attention", "attention_mask", "attn", "attn_mask", "bool", "class", "def", "device", "dtype", "else", "hidden_state", "if", "is", "key_padding_mask", "need_weights", "nn", "not", "return", "self", "to", "torch"], "bridgetower/modeling_bridgetower.py:BridgeTowerResidualAttention.forward": ["ModelResidualAttention", "Module", "None", "Tensor", "attention", "attention_mask", "class", "def", "for", "forward", "hidden_state", "in", "layer", "ln_1", "ln_2", "mlp", "nn", "residual_state", "return", "self", "torch", "values"], "bridgetower/modeling_bridgetower.py:BridgeTowerTransformer.__init__": ["ModelResidualAttention", "ModelTransformer", "Module", "ModuleList", "_", "__init__", "class", "config", "def", "else", "for", "hidden_size", "if", "in", "nn", "num_hidden_layers", "range", "remove_last_layer", "resblocks", "self", "stop_gradient", "super"], "bridgetower/modeling_bridgetower.py:BridgeTowerTransformer.forward": ["ModelTransformer", "Module", "None", "Tensor", "append", "attention_mask", "block", "class", "def", "detach", "else", "for", "forward", "hidden_state", "hidden_states", "if", "in", "nn", "resblocks", "return", "self", "stop_gradient", "torch"], "bridgetower/modeling_bridgetower.py:BridgeTowerVisionEmbeddings.__init__": ["Conv2d", "Embedding", "False", "ModelVisionConfig", "ModelVisionEmbeddings", "Module", "Parameter", "__init__", "arange", "bias", "class", "class_embedding", "config", "def", "embed_dim", "expand", "hidden_size", "image_size", "in_channels", "kernel_size", "nn", "num_channels", "num_patches", "num_positions", "out_channels", "patch_embedding", "patch_size", "persistent", "position_embedding", "position_ids", "randn", "register_buffer", "self", "stride", "super", "torch"], "bridgetower/modeling_bridgetower.py:BridgeTowerVisionEmbeddings.interpolate_pos_encoding": ["False", "ModelVisionEmbeddings", "Module", "Tensor", "align_corners", "and", "bicubic", "cat", "class", "class_pos_embed", "def", "dim", "embeddings", "functional", "height", "if", "int", "interpolate", "interpolate_pos_encoding", "is_tracing", "jit", "mode", "new_height", "new_width", "nn", "not", "num_patches", "num_positions", "patch_pos_embed", "patch_size", "permute", "position_embedding", "position_ids", "reshape", "return", "self", "shape", "size", "sqrt_num_positions", "torch", "torch_int", "unsqueeze", "view", "weight", "width"], "bridgetower/modeling_bridgetower.py:BridgeTowerVisionEmbeddings.forward": ["False", "FloatTensor", "Input", "ModelVisionEmbeddings", "Module", "Tensor", "ValueError", "_", "and", "batch_size", "cat", "class", "class_embedding", "class_embeds", "def", "dim", "doesn", "dtype", "else", "embeddings", "expand", "f", "flatten", "forward", "height", "if", "image", "image_size", "interpolate_pos_encoding", "match", "model", "nn", "not", "or", "patch_embedding", "patch_embeds", "pixel_values", "position_embedding", "position_ids", "raise", "return", "self", "shape", "size", "t", "target_dtype", "to", "torch", "transpose", "weight", "width"], "bridgetower/modeling_bridgetower.py:BridgeTowerVisionTransformer.__init__": ["LayerNorm", "ModelTransformer", "ModelVisionEmbeddings", "ModelVisionTransformer", "Module", "ModuleList", "_", "__init__", "class", "config", "def", "embeddings", "eps", "for", "hidden_size", "if", "in", "layer_norm_eps", "ln_post", "ln_pre", "ln_separate", "nn", "not", "num_hidden_layers", "range", "self", "share_layernorm", "super", "transformer"], "bridgetower/modeling_bridgetower.py:BridgeTowerVisionTransformer.forward": ["False", "ModelVisionTransformer", "Module", "Tensor", "append", "attention_mask", "bool", "class", "def", "dim", "else", "embeddings", "for", "forward", "hidden_states", "hidden_states_stack", "if", "in", "interpolate_pos_encoding", "ln", "ln_post", "ln_pre", "ln_separate", "nn", "permute", "pixel_values", "return", "self", "share_layernorm", "stack", "torch", "transformer", "zip"], "bridgetower/modeling_bridgetower.py:BridgeTowerVisionTransformer.forward_pre": ["False", "ModelVisionTransformer", "Module", "Tensor", "bool", "class", "def", "embeddings", "forward_pre", "hidden_states", "interpolate_pos_encoding", "ln_pre", "nn", "permute", "pixel_values", "return", "self", "torch"], "bridgetower/modeling_bridgetower.py:BridgeTowerVisionTransformer.forward_post": ["ModelVisionTransformer", "Module", "Tensor", "class", "def", "forward_post", "hidden_state", "ln_post", "nn", "permute", "return", "self", "torch", "visual_output_post"], "bridgetower/modeling_bridgetower.py:BridgeTowerLinkTower.__init__": ["LayerNorm", "ModelLinkTower", "Module", "NotImplementedError", "Parameter", "__init__", "add", "beta", "class", "config", "def", "elif", "else", "eps", "f", "hidden_size", "if", "implemented", "in", "interpolate", "is", "layer_norm_eps", "link_tower_type", "nn", "not", "raise", "scaled_add", "scaled_factor", "self", "super", "tensor", "torch"], "bridgetower/modeling_bridgetower.py:BridgeTowerLinkTower.forward": ["LayerNorm", "ModelLinkTower", "Module", "NotImplementedError", "add", "attention_mask", "beta", "class", "cross_modal_hidden_states", "def", "elif", "else", "f", "forward", "hidden_states", "if", "implemented", "interpolate", "is", "link_tower_type", "nn", "not", "raise", "return", "scaled_add", "scaled_factor", "self"], "bridgetower/modeling_bridgetower.py:BridgeTowerSelfOutput.__init__": ["Dropout", "LayerNorm", "Linear", "ModelSelfOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "eps", "hidden_dropout_prob", "hidden_size", "layer_norm_eps", "nn", "self", "super"], "bridgetower/modeling_bridgetower.py:BridgeTowerSelfOutput.forward": ["LayerNorm", "ModelSelfOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "bridgetower/modeling_bridgetower.py:BridgeTowerIntermediate.__init__": ["ACT2FN", "Linear", "ModelIntermediate", "Module", "__init__", "class", "config", "def", "dense", "else", "hidden_act", "hidden_size", "if", "intermediate_act_fn", "intermediate_size", "isinstance", "nn", "self", "str", "super"], "bridgetower/modeling_bridgetower.py:BridgeTowerIntermediate.forward": ["ModelIntermediate", "Module", "Tensor", "class", "def", "dense", "forward", "hidden_states", "intermediate_act_fn", "nn", "return", "self", "torch"], "bridgetower/modeling_bridgetower.py:BridgeTowerOutput.__init__": ["Dropout", "LayerNorm", "Linear", "ModelOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "eps", "hidden_dropout_prob", "hidden_size", "intermediate_size", "layer_norm_eps", "nn", "self", "super"], "bridgetower/modeling_bridgetower.py:BridgeTowerOutput.forward": ["LayerNorm", "ModelOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "bridgetower/modeling_bridgetower.py:BridgeTowerPooler.__init__": ["Linear", "ModelPooler", "Module", "Tanh", "__init__", "activation", "class", "config", "def", "dense", "hidden_size", "nn", "self", "super"], "bridgetower/modeling_bridgetower.py:BridgeTowerPooler.forward": ["ModelPooler", "Module", "Tensor", "activation", "class", "def", "dense", "first_token_tensor", "forward", "hidden_states", "nn", "pooled_output", "return", "self", "torch"], "bridgetower/modeling_bridgetower.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "contiguous", "def", "dim", "dropout", "float", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "shape", "size", "softmax", "torch", "training", "transpose", "value"], "bridgetower/modeling_bridgetower.py:BridgeTowerSelfAttention.__init__": ["Dropout", "False", "Linear", "ModelSelfAttention", "Module", "None", "The", "ValueError", "__init__", "a", "all_head_size", "and", "attention", "attention_head_size", "attention_probs_dropout_prob", "class", "config", "def", "dropout", "embedding_size", "f", "hasattr", "heads", "hidden", "hidden_size", "if", "int", "is", "is_causal", "is_decoder", "key", "layer_idx", "multiple", "nn", "not", "num_attention_heads", "number", "of", "query", "raise", "scaling", "self", "size", "super", "the", "value"], "bridgetower/modeling_bridgetower.py:BridgeTowerSelfAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "EncoderDecoderCache", "FloatTensor", "ModelSelfAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_attn_implementation", "attention_head_size", "attention_interface", "attention_mask", "attn_output", "attn_weights", "cache_position", "class", "config", "contiguous", "current_past_key_values", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "hidden_shape", "hidden_states", "if", "input_shape", "is", "isinstance", "key", "key_layer", "kwargs", "layer_idx", "nn", "not", "p", "past_key_values", "query", "query_layer", "reshape", "return", "scaling", "self", "self_attention_cache", "shape", "torch", "training", "transpose", "tuple", "update", "value", "value_layer", "view"], "bridgetower/modeling_bridgetower.py:BridgeTowerCrossAttention.__init__": ["Dropout", "False", "Linear", "ModelCrossAttention", "Module", "None", "The", "ValueError", "__init__", "a", "all_head_size", "and", "attention", "attention_head_size", "attention_probs_dropout_prob", "class", "config", "def", "dropout", "embedding_size", "f", "hasattr", "heads", "hidden", "hidden_size", "if", "int", "is", "is_causal", "key", "layer_idx", "multiple", "nn", "not", "num_attention_heads", "number", "of", "query", "raise", "scaling", "self", "size", "super", "the", "value"], "bridgetower/modeling_bridgetower.py:BridgeTowerCrossAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "EncoderDecoderCache", "False", "FloatTensor", "ModelCrossAttention", "Module", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "_attn_implementation", "and", "attention_head_size", "attention_interface", "attention_mask", "attn_output", "attn_weights", "bsz", "class", "config", "contiguous", "cross_attention_cache", "def", "dropout", "eager", "eager_attention_forward", "else", "encoder_hidden_states", "forward", "get", "hidden_states", "if", "is", "is_updated", "key", "key_layer", "keys", "kv_input_shape", "kwargs", "layer_idx", "layers", "nn", "not", "p", "past_key_values", "q_input_shape", "query", "query_layer", "reshape", "return", "scaling", "self", "shape", "src_len", "tgt_len", "torch", "training", "transpose", "tuple", "update", "value", "value_layer", "values", "view"], "bridgetower/modeling_bridgetower.py:BridgeTowerAttention.__init__": ["False", "ModelAttention", "ModelCrossAttention", "ModelSelfAttention", "ModelSelfOutput", "Module", "None", "__init__", "attention_class", "class", "config", "def", "else", "if", "is_causal", "is_cross_attention", "layer_idx", "nn", "output", "self", "super"], "bridgetower/modeling_bridgetower.py:BridgeTowerAttention.forward": ["Cache", "FloatTensor", "ModelAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attention_output", "attn_weights", "cache_position", "class", "def", "else", "encoder_attention_mask", "encoder_hidden_states", "forward", "hidden_states", "if", "is_cross_attention", "kwargs", "nn", "not", "output", "past_key_values", "return", "self", "torch", "tuple"], "bridgetower/modeling_bridgetower.py:BridgeTowerBertCrossLayer.__init__": ["False", "ModelAttention", "ModelBertCrossLayer", "ModelIntermediate", "ModelOutput", "Module", "None", "True", "__init__", "add_cross_attention", "attention", "chunk_size_feed_forward", "class", "config", "crossattention", "def", "intermediate", "is_causal", "is_cross_attention", "is_decoder", "layer_idx", "nn", "output", "self", "seq_len_dim", "super"], "bridgetower/modeling_bridgetower.py:BridgeTowerBertCrossLayer.forward": ["ModelBertCrossLayer", "Module", "None", "TransformersKwargs", "Unpack", "apply_chunking_to_forward", "attention", "attention_mask", "attention_output", "chunk_size_feed_forward", "class", "cross_attention_output", "cross_attn_weights", "crossattention", "def", "encoder_attention_mask", "encoder_hidden_states", "feed_forward_chunk", "forward", "hidden_states", "kwargs", "layer_output", "nn", "past_key_values", "return", "self", "self_attention_output", "self_attn_weights", "seq_len_dim"], "bridgetower/modeling_bridgetower.py:BridgeTowerBertCrossLayer.feed_forward_chunk": ["ModelBertCrossLayer", "Module", "attention_output", "class", "def", "feed_forward_chunk", "intermediate", "intermediate_output", "layer_output", "nn", "output", "return", "self"], "bridgetower/modeling_bridgetower.py:BridgeTowerTextLayer.__init__": ["False", "GradientCheckpointingLayer", "ModelAttention", "ModelIntermediate", "ModelOutput", "ModelTextLayer", "None", "True", "ValueError", "__init__", "a", "add_cross_attention", "added", "as", "attention", "be", "chunk_size_feed_forward", "class", "config", "cross", "crossattention", "decoder", "def", "f", "if", "intermediate", "is", "is_causal", "is_cross_attention", "is_decoder", "layer_idx", "model", "not", "output", "raise", "self", "seq_len_dim", "should", "super", "used"], "bridgetower/modeling_bridgetower.py:BridgeTowerTextLayer.forward": ["Cache", "FloatTensor", "GradientCheckpointingLayer", "If", "ModelTextLayer", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "ValueError", "add_cross_attention", "and", "apply_chunking_to_forward", "are", "attention", "attention_mask", "attention_output", "be", "by", "cache_position", "chunk_size_feed_forward", "class", "config", "cross", "cross_attention_output", "cross_attn_weights", "crossattention", "def", "encoder_attention_mask", "encoder_hidden_states", "f", "feed_forward_chunk", "forward", "has", "hasattr", "hidden_states", "if", "instantiated", "is", "is_decoder", "kwargs", "layer_output", "layers", "not", "outputs", "passed", "past_key_values", "raise", "return", "self", "self_attention_output", "self_attn_weights", "seq_len_dim", "setting", "to", "torch", "tuple", "with"], "bridgetower/modeling_bridgetower.py:BridgeTowerTextLayer.feed_forward_chunk": ["GradientCheckpointingLayer", "ModelTextLayer", "attention_output", "class", "def", "feed_forward_chunk", "intermediate", "intermediate_output", "layer_output", "output", "return", "self"], "bridgetower/modeling_bridgetower.py:BridgeTowerTextEncoder.__init__": ["ModelTextEncoder", "ModelTextLayer", "Module", "ModuleList", "__init__", "class", "config", "def", "for", "i", "in", "layer", "layer_idx", "nn", "num_hidden_layers", "range", "self", "super"], "bridgetower/modeling_bridgetower.py:BridgeTowerTextEncoder.forward": ["BaseModelOutputWithPastAndCrossAttentions", "Cache", "False", "FloatTensor", "ModelTextEncoder", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "add_cross_attention", "all_cross_attentions", "all_hidden_states", "all_self_attentions", "and", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "cross_attentions", "def", "else", "encoder_attention_mask", "encoder_hidden_states", "enumerate", "for", "forward", "hidden_states", "i", "if", "in", "kwargs", "last_hidden_state", "layer", "layer_module", "layer_outputs", "nn", "output_attentions", "output_hidden_states", "past_key_values", "return", "self", "torch", "tuple", "use_cache"], "bridgetower/modeling_bridgetower.py:BridgeTowerTextEmbeddings.__init__": ["Dropout", "Embedding", "False", "LayerNorm", "ModelTextEmbeddings", "Module", "__init__", "arange", "class", "config", "def", "dropout", "dtype", "eps", "expand", "hidden_dropout_prob", "hidden_size", "layer_norm_eps", "long", "max_position_embeddings", "nn", "pad_token_id", "padding_idx", "persistent", "position_embeddings", "position_ids", "register_buffer", "self", "size", "super", "token_type_embeddings", "token_type_ids", "torch", "type_vocab_size", "vocab_size", "word_embeddings", "zeros"], "bridgetower/modeling_bridgetower.py:BridgeTowerTextEmbeddings.forward": ["FloatTensor", "LayerNorm", "LongTensor", "ModelTextEmbeddings", "Module", "None", "Tensor", "batch_size", "buffered_token_type_ids", "class", "create_position_ids_from_input_ids", "create_position_ids_from_inputs_embeds", "def", "device", "dim", "dropout", "dtype", "else", "embeddings", "expand", "forward", "gather", "hasattr", "if", "index", "input_ids", "input_shape", "inputs_embeds", "int", "is", "long", "nn", "not", "padding_idx", "past_key_values_length", "position_embeddings", "position_ids", "return", "self", "seq_length", "shape", "size", "token_type_embeddings", "token_type_ids", "torch", "word_embeddings", "zeros"], "bridgetower/modeling_bridgetower.py:BridgeTowerTextEmbeddings.create_position_ids_from_inputs_embeds": ["ModelTextEmbeddings", "Module", "arange", "class", "create_position_ids_from_inputs_embeds", "def", "device", "dtype", "expand", "input_shape", "inputs_embeds", "long", "nn", "padding_idx", "position_ids", "return", "sequence_length", "size", "torch", "unsqueeze"], "bridgetower/modeling_bridgetower.py:BridgeTowerTextEmbeddings.create_position_ids_from_input_ids": ["ModelTextEmbeddings", "Module", "class", "create_position_ids_from_input_ids", "cumsum", "def", "dim", "incremental_indices", "input_ids", "int", "long", "mask", "ne", "nn", "padding_idx", "past_key_values_length", "return", "torch", "type_as"], "bridgetower/modeling_bridgetower.py:BridgeTowerPreTrainedModel._init_weights": ["Conv2d", "Embedding", "LayerNorm", "Linear", "ModelForContrastiveLearning", "ModelMLMHead", "ModelPreTrainedModel", "ModelTextEmbeddings", "ModelVisionEmbeddings", "ModelVisionTransformer", "Module", "None", "PreTrainedModel", "_init_weights", "and", "arange", "attn", "attn_std", "bias", "block", "c_fc", "c_proj", "class", "class_embedding", "config", "constant_", "copy_", "def", "elif", "embeddings", "expand", "fc_std", "for", "hidden_size", "if", "in", "in_proj_bias", "in_proj_weight", "init", "initializer_factor", "is", "isinstance", "logit_scale", "logit_scale_init_value", "mean", "mlp", "module", "nn", "normal_", "not", "num_hidden_layers", "num_positions", "ones_", "out_proj", "position_embedding", "position_ids", "proj_std", "resblocks", "self", "shape", "std", "token_type_ids", "torch", "transformer", "weight", "zeros_"], "bridgetower/modeling_bridgetower.py:BridgeTowerVisionModel.__init__": ["ModelPreTrainedModel", "ModelVisionModel", "ModelVisionTransformer", "__init__", "class", "config", "def", "post_init", "self", "super", "visual"], "bridgetower/modeling_bridgetower.py:BridgeTowerVisionModel.dtype": ["ModelPreTrainedModel", "ModelVisionModel", "class", "def", "dtype", "embeddings", "patch_embedding", "return", "self", "visual", "weight"], "bridgetower/modeling_bridgetower.py:BridgeTowerVisionModel.forward": ["False", "ModelPreTrainedModel", "ModelVisionModel", "None", "class", "def", "dtype", "forward", "image", "image_mask", "interpolate_pos_encoding", "kwargs", "return", "self", "type", "visual"], "bridgetower/modeling_bridgetower.py:BridgeTowerTextModel.__init__": ["False", "ModelPooler", "ModelPreTrainedModel", "ModelTextEmbeddings", "ModelTextEncoder", "ModelTextModel", "None", "True", "__init__", "add_pooling_layer", "class", "config", "def", "else", "embeddings", "encoder", "gradient_checkpointing", "if", "pooler", "post_init", "r", "self", "super"], "bridgetower/modeling_bridgetower.py:BridgeTowerTextModel.get_input_embeddings": ["ModelPreTrainedModel", "ModelTextModel", "class", "def", "embeddings", "get_input_embeddings", "return", "self", "word_embeddings"], "bridgetower/modeling_bridgetower.py:BridgeTowerTextModel.set_input_embeddings": ["ModelPreTrainedModel", "ModelTextModel", "class", "def", "embeddings", "self", "set_input_embeddings", "value", "word_embeddings"], "bridgetower/modeling_bridgetower.py:BridgeTowerTextModel.forward": ["BaseModelOutputWithPoolingAndCrossAttentions", "Cache", "DynamicCache", "EncoderDecoderCache", "False", "ModelPreTrainedModel", "ModelTextModel", "None", "Setting", "Tensor", "TransformersKwargs", "True", "Unpack", "ValueError", "You", "_create_attention_masks", "and", "arange", "attention_mask", "attentions", "bool", "cache_position", "checkpointing", "class", "config", "cross_attentions", "def", "device", "else", "embedding_output", "embeddings", "encoder", "encoder_attention_mask", "encoder_hidden_states", "encoder_outputs", "exactly", "forward", "get_seq_length", "gradient", "gradient_checkpointing", "hidden_states", "if", "incompatible", "input_ids", "input_shape", "inputs_embeds", "is", "is_decoder", "kwargs", "last_hidden_state", "logger", "must", "not", "of", "one", "or", "output_attentions", "output_hidden_states", "past_key_values", "past_key_values_length", "pooled_output", "pooler", "pooler_output", "position_ids", "raise", "return", "self", "seq_length", "sequence_output", "shape", "specify", "token_type_ids", "torch", "training", "tuple", "use_cache", "warning_once", "with"], "bridgetower/modeling_bridgetower.py:BridgeTowerTextModel._create_attention_masks": ["ModelPreTrainedModel", "ModelTextModel", "None", "_create_attention_masks", "attention_mask", "cache_position", "class", "config", "create_bidirectional_mask", "create_causal_mask", "def", "else", "embedding_output", "encoder_attention_mask", "encoder_hidden_states", "if", "input_embeds", "is", "is_decoder", "not", "past_key_values", "return", "self"], "bridgetower/modeling_bridgetower.py:BridgeTowerModel.__init__": ["Embedding", "LayerNorm", "Linear", "ModelBertCrossLayer", "ModelLinkTower", "ModelModel", "ModelPooler", "ModelPreTrainedModel", "ModelTextModel", "ModelVisionModel", "ModuleList", "_", "__init__", "and", "bias", "class", "config", "cross_modal_image_layernorm", "cross_modal_image_layers", "cross_modal_image_link_tower", "cross_modal_image_pooler", "cross_modal_image_transform", "cross_modal_ln_separate", "cross_modal_text_layernorm", "cross_modal_text_layers", "cross_modal_text_link_tower", "cross_modal_text_pooler", "cross_modal_text_transform", "data", "def", "else", "eps", "for", "hidden_size", "if", "in", "init_layernorm_from_vision_encoder", "layer_norm_eps", "ln", "ln_post", "nn", "not", "num_hidden_layers", "post_init", "range", "self", "share_cross_modal_transformer_layers", "share_layernorm", "share_link_tower_layers", "super", "text_config", "text_model", "token_type_embeddings", "vision_config", "vision_model", "visual", "weight"], "bridgetower/modeling_bridgetower.py:BridgeTowerModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "return", "self", "text_model"], "bridgetower/modeling_bridgetower.py:BridgeTowerModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "self", "set_input_embeddings", "text_model", "value"], "bridgetower/modeling_bridgetower.py:BridgeTowerModel.forward": ["False", "FloatTensor", "LongTensor", "Make", "ModelModel", "ModelModelOutput", "ModelPreTrainedModel", "None", "NotImplementedError", "Tensor", "all_hidden_states", "all_hidden_states_cross", "all_hidden_states_image", "all_hidden_states_text", "all_self_attentions", "and", "attention_mask", "attentions", "block", "bool", "class", "cls_features", "config", "cross_image_features", "cross_image_features_", "cross_modal_image", "cross_modal_image_layernorm", "cross_modal_image_layers", "cross_modal_image_link_tower", "cross_modal_image_transform", "cross_modal_text", "cross_modal_text_layernorm", "cross_modal_text_layers", "cross_modal_text_link_tower", "cross_modal_text_transform", "cross_text_features", "cross_text_features_", "def", "device", "does", "dtype", "else", "embeddings", "encoder", "encoder_attention_mask", "expand_as", "extend_image_masks", "extend_text_masks", "for", "forward", "forward_post", "forward_pre", "full", "get_cls_features", "get_extended_attention_mask", "hidden_states", "i", "if", "image_embeds", "image_embeds_with_ln", "image_features", "image_link_tower", "image_token_type_embeddings", "image_token_type_idx", "in", "input_ids", "input_shape", "inputs_embeds", "instead", "int", "interpolate_pos_encoding", "is", "kwargs", "labels", "layer", "layer_outputs_image", "layer_outputs_text", "len", "link_layer_index", "long", "not", "num_hidden_layers", "ones", "output_attentions", "output_hidden_states", "pass", "permute", "pixel_mask", "pixel_values", "pooler_output", "r", "raise", "range", "resblocks", "return", "return_dict", "self", "size", "split_index", "sure", "text_embeds", "text_features", "text_link_tower", "text_model", "text_token_type_embeddings", "to", "token_type_embeddings", "token_type_ids", "torch", "transformer", "tuple", "type", "use", "use_return_dict", "v", "vision_model", "visual", "zeros"], "bridgetower/modeling_bridgetower.py:BridgeTowerModel.get_cls_features": ["ModelModel", "ModelPreTrainedModel", "cat", "class", "cls_features_image", "cls_features_text", "cross_modal_image_pooler", "cross_modal_text_pooler", "def", "dim", "get_cls_features", "image_features", "return", "self", "text_features", "torch"], "bridgetower/modeling_bridgetower.py:BridgeTowerPredictionHeadTransform.__init__": ["ACT2FN", "LayerNorm", "Linear", "ModelPredictionHeadTransform", "Module", "__init__", "class", "config", "def", "dense", "else", "eps", "hidden_act", "hidden_size", "if", "isinstance", "layer_norm_eps", "nn", "self", "str", "super", "transform_act_fn"], "bridgetower/modeling_bridgetower.py:BridgeTowerPredictionHeadTransform.forward": ["LayerNorm", "ModelPredictionHeadTransform", "Module", "class", "def", "dense", "forward", "hidden_states", "nn", "return", "self", "transform_act_fn"], "bridgetower/modeling_bridgetower.py:BridgeTowerMLMHead.__init__": ["False", "Linear", "ModelMLMHead", "ModelPredictionHeadTransform", "Module", "None", "Parameter", "__init__", "bias", "class", "config", "decoder", "def", "hidden_size", "if", "is", "nn", "not", "self", "super", "text_config", "torch", "transform", "vocab_size", "weight", "zeros"], "bridgetower/modeling_bridgetower.py:BridgeTowerMLMHead.forward": ["ModelMLMHead", "Module", "bias", "class", "decoder", "def", "forward", "mlm_score", "nn", "return", "self", "transform", "x"], "bridgetower/modeling_bridgetower.py:BridgeTowerITMHead.__init__": ["Linear", "ModelITMHead", "Module", "__init__", "class", "def", "fc", "hidden_size", "nn", "self", "super"], "bridgetower/modeling_bridgetower.py:BridgeTowerITMHead.forward": ["ModelITMHead", "Module", "class", "def", "fc", "forward", "itm_score", "nn", "return", "self", "x"], "bridgetower/modeling_bridgetower.py:BridgeTowerForMaskedLM.__init__": ["Model", "ModelForMaskedLM", "ModelMLMHead", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "config", "def", "mlm_score", "post_init", "self", "super"], "bridgetower/modeling_bridgetower.py:BridgeTowerForMaskedLM.get_output_embeddings": ["ModelForMaskedLM", "ModelPreTrainedModel", "class", "decoder", "def", "get_output_embeddings", "mlm_score", "return", "self"], "bridgetower/modeling_bridgetower.py:BridgeTowerForMaskedLM.set_output_embeddings": ["ModelForMaskedLM", "ModelPreTrainedModel", "class", "decoder", "def", "mlm_score", "new_embeddings", "self", "set_output_embeddings"], "bridgetower/modeling_bridgetower.py:BridgeTowerForMaskedLM.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "MaskedLMOutput", "Model", "ModelForMaskedLM", "ModelPreTrainedModel", "None", "attention_mask", "attentions", "bool", "class", "config", "def", "device", "else", "forward", "hidden_states", "if", "image_embeds", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "masked_lm_loss", "mlm_logits", "mlm_score", "not", "output", "output_attentions", "output_hidden_states", "outputs", "pixel_mask", "pixel_values", "r", "return", "return_dict", "self", "text_config", "text_features", "to", "token_type_ids", "torch", "tuple", "use_return_dict", "view", "vocab_size"], "bridgetower/modeling_bridgetower.py:BridgeTowerForImageAndTextRetrieval.__init__": ["Model", "ModelForImageAndTextRetrieval", "ModelITMHead", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "config", "def", "hidden_size", "itm_score", "post_init", "self", "super"], "bridgetower/modeling_bridgetower.py:BridgeTowerForImageAndTextRetrieval.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "Model", "ModelForImageAndTextRetrieval", "ModelPreTrainedModel", "None", "SequenceClassifierOutput", "attention_mask", "attentions", "bool", "class", "config", "def", "device", "else", "forward", "hidden_states", "if", "image_embeds", "input_ids", "inputs_embeds", "is", "itm_loss", "itm_score", "kwargs", "labels", "logits", "loss", "loss_fct", "not", "output", "output_attentions", "output_hidden_states", "outputs", "pixel_mask", "pixel_values", "pooler_output", "r", "return", "return_dict", "self", "to", "token_type_ids", "torch", "tuple", "use_return_dict"], "bridgetower/modeling_bridgetower.py:BridgeTowerContrastiveHead.__init__": ["Linear", "ModelContrastiveHead", "Module", "__init__", "class", "def", "embed_size", "fc", "hidden_size", "nn", "self", "super"], "bridgetower/modeling_bridgetower.py:BridgeTowerContrastiveHead.forward": ["ModelContrastiveHead", "Module", "class", "def", "fc", "forward", "nn", "return", "self", "x"], "bridgetower/modeling_bridgetower.py:BridgeTowerForContrastiveLearning.__init__": ["Model", "ModelContrastiveHead", "ModelForContrastiveLearning", "ModelModel", "ModelPreTrainedModel", "Parameter", "__init__", "class", "config", "contrastive_hidden_size", "def", "hidden_size", "itc_cross_modal_head", "itc_image_head", "itc_text_head", "logit_scale", "logit_scale_init_value", "nn", "post_init", "self", "super", "tensor", "torch"], "bridgetower/modeling_bridgetower.py:BridgeTowerForContrastiveLearning.forward": ["FloatTensor", "LongTensor", "Model", "ModelContrastiveOutput", "ModelForContrastiveLearning", "ModelPreTrainedModel", "None", "True", "arange", "attention_mask", "attentions", "bool", "class", "config", "cross_embeds", "cross_entropy", "cross_modal_image_transform", "def", "device", "dim", "dtype", "else", "exp", "expand_as", "forward", "forward_post", "full", "functional", "hidden_states", "hidden_states_cross_modal", "hidden_states_img", "hidden_states_txt", "if", "image_embeds", "image_embeds_with_ln", "image_to_cross_loss", "image_token_type_embeddings", "input_ids", "inputs_embeds", "is", "itc_cross_modal_head", "itc_image_head", "itc_loss", "itc_text_head", "kwargs", "labels", "len", "logit_scale", "logits", "logits_image_to_cross", "logits_text_to_cross", "logits_text_to_image", "long", "loss", "matmul", "nn", "normalize", "not", "output", "output_attentions", "output_hidden_states", "outputs", "p", "pixel_mask", "pixel_values", "pooler_output", "r", "return", "return_dict", "return_loss", "self", "stack", "t", "text_embeds", "text_to_cross_loss", "text_to_image_loss", "to", "token_type_embeddings", "token_type_ids", "torch", "tuple", "use_return_dict", "vision_model", "visual", "weight"], "bros/modeling_bros.py:BrosPositionalEmbedding1D.__init__": ["ModelPositionalEmbedding1D", "Module", "__init__", "arange", "class", "config", "def", "dim_bbox_sinusoid_emb_1d", "inv_freq", "nn", "register_buffer", "self", "super", "torch"], "bros/modeling_bros.py:BrosPositionalEmbedding1D.forward": ["ModelPositionalEmbedding1D", "Module", "Tensor", "b1", "b2", "b3", "cat", "class", "cos", "def", "dim", "dim_bbox_sinusoid_emb_1d", "forward", "inv_freq", "nn", "pos_emb", "pos_seq", "return", "self", "seq_size", "sin", "sinusoid_inp", "size", "torch", "view"], "bros/modeling_bros.py:BrosPositionalEmbedding2D.__init__": ["ModelPositionalEmbedding1D", "ModelPositionalEmbedding2D", "Module", "__init__", "class", "config", "def", "dim_bbox", "nn", "self", "super", "x_pos_emb", "y_pos_emb"], "bros/modeling_bros.py:BrosPositionalEmbedding2D.forward": ["ModelPositionalEmbedding2D", "Module", "Tensor", "append", "bbox", "bbox_pos_emb", "cat", "class", "def", "dim", "dim_bbox", "else", "for", "forward", "i", "if", "in", "nn", "range", "return", "self", "stack", "torch", "x_pos_emb", "y_pos_emb"], "bros/modeling_bros.py:BrosBboxEmbeddings.__init__": ["False", "Linear", "ModelBboxEmbeddings", "ModelPositionalEmbedding2D", "Module", "__init__", "bbox_projection", "bbox_sinusoid_emb", "bias", "class", "config", "def", "dim_bbox_projection", "dim_bbox_sinusoid_emb_2d", "nn", "self", "super"], "bros/modeling_bros.py:BrosBboxEmbeddings.forward": ["ModelBboxEmbeddings", "Module", "None", "Tensor", "bbox", "bbox_pos", "bbox_pos_emb", "bbox_projection", "bbox_sinusoid_emb", "bbox_t", "class", "def", "forward", "nn", "return", "self", "torch", "transpose"], "bros/modeling_bros.py:BrosTextEmbeddings.__init__": ["Dropout", "Embedding", "False", "LayerNorm", "ModelTextEmbeddings", "Module", "__init__", "arange", "class", "config", "def", "device", "dropout", "dtype", "eps", "expand", "hidden_dropout_prob", "hidden_size", "layer_norm_eps", "long", "max_position_embeddings", "nn", "pad_token_id", "padding_idx", "persistent", "position_embeddings", "position_ids", "register_buffer", "self", "size", "super", "token_type_embeddings", "token_type_ids", "torch", "type_vocab_size", "vocab_size", "word_embeddings", "zeros"], "bros/modeling_bros.py:BrosTextEmbeddings.forward": ["LayerNorm", "ModelTextEmbeddings", "Module", "None", "Tensor", "buffered_token_type_ids", "buffered_token_type_ids_expanded", "class", "def", "device", "dropout", "dtype", "else", "embeddings", "expand", "forward", "hasattr", "if", "input_ids", "input_shape", "inputs_embeds", "is", "long", "nn", "not", "position_embeddings", "position_ids", "return", "self", "seq_length", "size", "token_type_embeddings", "token_type_ids", "torch", "word_embeddings", "zeros"], "bros/modeling_bros.py:BrosSelfAttention.__init__": ["Dropout", "Linear", "ModelSelfAttention", "Module", "The", "ValueError", "__init__", "a", "all_head_size", "and", "attention", "attention_head_size", "attention_probs_dropout_prob", "class", "config", "def", "dropout", "embedding_size", "f", "hasattr", "heads", "hidden", "hidden_size", "if", "int", "is", "is_decoder", "key", "multiple", "nn", "not", "num_attention_heads", "number", "of", "query", "raise", "self", "size", "super", "the", "value"], "bros/modeling_bros.py:BrosSelfAttention.forward": ["False", "ModelSelfAttention", "Module", "None", "Softmax", "Tensor", "all_head_size", "attention_head_size", "attention_mask", "attention_probs", "attention_scores", "batch_size", "bbox_pos_emb", "bbox_pos_scores", "bijd", "bnid", "bnij", "class", "context_layer", "contiguous", "d_head", "def", "dim", "dropout", "einsum", "else", "encoder_attention_mask", "encoder_hidden_states", "forward", "hidden_shape", "hidden_states", "if", "is", "is_cross_attention", "is_decoder", "key", "key_layer", "math", "matmul", "n_head", "new_context_layer_shape", "nn", "not", "num_attention_heads", "output_attentions", "outputs", "permute", "query", "query_layer", "return", "self", "seq_length", "shape", "size", "sqrt", "torch", "transpose", "tuple", "value", "value_layer", "view"], "bros/modeling_bros.py:BrosSelfOutput.__init__": ["Dropout", "LayerNorm", "Linear", "ModelSelfOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "eps", "hidden_dropout_prob", "hidden_size", "layer_norm_eps", "nn", "self", "super"], "bros/modeling_bros.py:BrosSelfOutput.forward": ["LayerNorm", "ModelSelfOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "bros/modeling_bros.py:BrosAttention.__init__": ["ModelAttention", "ModelSelfAttention", "ModelSelfOutput", "Module", "__init__", "class", "config", "def", "nn", "output", "self", "super"], "bros/modeling_bros.py:BrosAttention.forward": ["False", "ModelAttention", "Module", "None", "Tensor", "attention_mask", "attention_output", "bbox_pos_emb", "bool", "class", "def", "encoder_attention_mask", "encoder_hidden_states", "forward", "hidden_states", "nn", "output", "output_attentions", "outputs", "return", "self", "self_outputs", "torch", "tuple"], "bros/modeling_bros.py:BrosIntermediate.__init__": ["ACT2FN", "Linear", "ModelIntermediate", "Module", "__init__", "class", "config", "def", "dense", "else", "hidden_act", "hidden_size", "if", "intermediate_act_fn", "intermediate_size", "isinstance", "nn", "self", "str", "super"], "bros/modeling_bros.py:BrosIntermediate.forward": ["ModelIntermediate", "Module", "Tensor", "class", "def", "dense", "forward", "hidden_states", "intermediate_act_fn", "nn", "return", "self", "torch"], "bros/modeling_bros.py:BrosOutput.__init__": ["Dropout", "LayerNorm", "Linear", "ModelOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "eps", "hidden_dropout_prob", "hidden_size", "intermediate_size", "layer_norm_eps", "nn", "self", "super"], "bros/modeling_bros.py:BrosOutput.forward": ["LayerNorm", "ModelOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "bros/modeling_bros.py:BrosLayer.__init__": ["Exception", "GradientCheckpointingLayer", "ModelAttention", "ModelIntermediate", "ModelLayer", "ModelOutput", "__init__", "a", "add_cross_attention", "added", "as", "attention", "be", "chunk_size_feed_forward", "class", "config", "cross", "crossattention", "decoder", "def", "f", "if", "intermediate", "is", "is_decoder", "model", "not", "output", "raise", "self", "seq_len_dim", "should", "super", "used"], "bros/modeling_bros.py:BrosLayer.forward": ["Exception", "False", "FloatTensor", "GradientCheckpointingLayer", "If", "ModelLayer", "None", "Tensor", "True", "add_cross_attention", "and", "apply_chunking_to_forward", "are", "attention", "attention_mask", "attention_output", "bbox_pos_emb", "be", "bool", "by", "chunk_size_feed_forward", "class", "config", "cross", "cross_attention_outputs", "crossattention", "def", "else", "encoder_attention_mask", "encoder_hidden_states", "f", "feed_forward_chunk", "forward", "has", "hasattr", "hidden_states", "if", "instantiated", "is", "is_decoder", "layer_output", "layers", "not", "output_attentions", "outputs", "passed", "raise", "return", "self", "self_attention_outputs", "seq_len_dim", "setting", "to", "torch", "tuple", "with"], "bros/modeling_bros.py:BrosLayer.feed_forward_chunk": ["GradientCheckpointingLayer", "ModelLayer", "attention_output", "class", "def", "feed_forward_chunk", "intermediate", "intermediate_output", "layer_output", "output", "return", "self"], "bros/modeling_bros.py:BrosEncoder.__init__": ["ModelEncoder", "ModelLayer", "Module", "ModuleList", "_", "__init__", "class", "config", "def", "for", "in", "layer", "nn", "num_hidden_layers", "range", "self", "super"], "bros/modeling_bros.py:BrosEncoder.forward": ["BaseModelOutputWithCrossAttentions", "False", "FloatTensor", "ModelEncoder", "Module", "None", "Tensor", "True", "add_cross_attention", "all_cross_attentions", "all_hidden_states", "all_self_attentions", "and", "attention_mask", "attentions", "bbox_pos_emb", "bool", "class", "config", "cross_attentions", "def", "else", "encoder_attention_mask", "encoder_hidden_states", "enumerate", "for", "forward", "hidden_states", "i", "if", "in", "last_hidden_state", "layer", "layer_module", "layer_outputs", "nn", "output_attentions", "output_hidden_states", "return", "return_dict", "self", "torch", "tuple"], "bros/modeling_bros.py:BrosPooler.__init__": ["Linear", "ModelPooler", "Module", "Tanh", "__init__", "activation", "class", "config", "def", "dense", "hidden_size", "nn", "self", "super"], "bros/modeling_bros.py:BrosPooler.forward": ["ModelPooler", "Module", "Tensor", "activation", "class", "def", "dense", "first_token_tensor", "forward", "hidden_states", "nn", "pooled_output", "return", "self", "torch"], "bros/modeling_bros.py:BrosRelationExtractor.__init__": ["Dropout", "Linear", "ModelRelationExtractor", "Module", "Parameter", "__init__", "backbone_hidden_size", "class", "classifier_dropout_prob", "config", "def", "drop", "dummy_node", "head_hidden_size", "hidden_size", "key", "n_relations", "nn", "query", "self", "super", "torch", "zeros"], "bros/modeling_bros.py:BrosRelationExtractor.forward": ["ModelRelationExtractor", "Module", "Tensor", "axis", "cat", "class", "def", "drop", "dummy_node", "dummy_vec", "forward", "head_hidden_size", "key", "key_layer", "matmul", "n_relations", "nn", "permute", "query", "query_layer", "relation_score", "repeat", "return", "self", "size", "torch", "unsqueeze", "view"], "bros/modeling_bros.py:BrosPreTrainedModel._init_weights": ["ModelPositionalEmbedding1D", "ModelPreTrainedModel", "ModelRelationExtractor", "ModelTextEmbeddings", "Module", "PreTrainedModel", "_init_weights", "arange", "class", "config", "copy_", "def", "dim_bbox_sinusoid_emb_1d", "dummy_node", "elif", "expand", "if", "init", "initializer_range", "inv_freq", "isinstance", "module", "nn", "normal_", "position_ids", "self", "shape", "std", "super", "token_type_ids", "torch", "zeros_"], "bros/modeling_bros.py:BrosModel.__init__": ["ModelBboxEmbeddings", "ModelEncoder", "ModelModel", "ModelPooler", "ModelPreTrainedModel", "ModelTextEmbeddings", "None", "True", "__init__", "add_pooling_layer", "bbox_embeddings", "class", "config", "def", "else", "embeddings", "encoder", "if", "pooler", "post_init", "r", "self", "super"], "bros/modeling_bros.py:BrosModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "return", "self", "word_embeddings"], "bros/modeling_bros.py:BrosModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "self", "set_input_embeddings", "value", "word_embeddings"], "bros/modeling_bros.py:BrosModel.forward": ["BaseModelOutputWithPoolingAndCrossAttentions", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "True", "ValueError", "You", "_", "and", "at", "attention_mask", "attentions", "batch_size", "bbox", "bbox_embeddings", "bbox_pos_emb", "bbox_position_embeddings", "bbox_scale", "bool", "both", "buffered_token_type_ids", "buffered_token_type_ids_expanded", "cannot", "class", "config", "cross_attentions", "def", "device", "dtype", "either", "elif", "else", "embedding_output", "embeddings", "encoder", "encoder_attention_mask", "encoder_batch_size", "encoder_extended_attention_mask", "encoder_hidden_shape", "encoder_hidden_states", "encoder_outputs", "encoder_sequence_length", "expand", "extended_attention_mask", "forward", "get_extended_attention_mask", "hasattr", "have", "hidden_states", "if", "input_ids", "input_shape", "inputs_embeds", "invert_attention_mask", "is", "is_decoder", "kwargs", "last_hidden_state", "long", "not", "ones", "or", "output_attentions", "output_hidden_states", "pooled_output", "pooler", "pooler_output", "position_ids", "r", "raise", "return", "return_dict", "same", "scaled_bbox", "self", "seq_length", "sequence_output", "shape", "size", "specify", "the", "time", "to", "token_type_ids", "torch", "tuple", "use_return_dict", "zeros"], "bros/modeling_bros.py:BrosForTokenClassification.__init__": ["Dropout", "Linear", "Model", "ModelForTokenClassification", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "classifier", "classifier_dropout", "config", "def", "dropout", "else", "hasattr", "hidden_dropout_prob", "hidden_size", "if", "nn", "num_labels", "post_init", "self", "super"], "bros/modeling_bros.py:BrosForTokenClassification.forward": ["CrossEntropyLoss", "Model", "ModelForTokenClassification", "ModelPreTrainedModel", "None", "Tensor", "TokenClassifierOutput", "True", "attention_mask", "attentions", "bbox", "bbox_first_token_mask", "bool", "class", "classifier", "config", "def", "dropout", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "not", "num_labels", "output_attentions", "output_hidden_states", "outputs", "position_ids", "r", "return", "return_dict", "self", "sequence_output", "token_type_ids", "torch", "tuple", "use_return_dict", "view"], "bros/modeling_bros.py:BrosSpadeEEForTokenClassification.__init__": ["Dropout", "Linear", "Model", "ModelModel", "ModelPreTrainedModel", "ModelRelationExtractor", "ModelSpadeEEForTokenClassification", "Sequential", "__init__", "backbone_hidden_size", "class", "classifier_dropout", "config", "def", "else", "hasattr", "hidden_dropout_prob", "hidden_size", "if", "initial_token_classifier", "n_relations", "nn", "num_labels", "post_init", "self", "subsequent_token_classifier", "super"], "bros/modeling_bros.py:BrosSpadeEEForTokenClassification.forward": ["CrossEntropyLoss", "Model", "ModelPreTrainedModel", "ModelSpadeEEForTokenClassification", "ModelSpadeOutput", "None", "Tensor", "True", "and", "attention_mask", "attentions", "axis", "batch_size", "bbox", "bbox_first_token_mask", "bool", "cat", "class", "config", "contiguous", "def", "device", "dtype", "else", "eye", "finfo", "forward", "hidden_states", "if", "initial_token_classifier", "initial_token_labels", "initial_token_logits", "initial_token_loss", "input_ids", "inputs_embeds", "inv_attention_mask", "invalid_token_mask", "is", "kwargs", "last_hidden_states", "loss", "loss_fct", "masked_fill", "max_seq_length", "min", "not", "num_labels", "output_attentions", "output_hidden_states", "outputs", "position_ids", "r", "return", "return_dict", "self", "self_token_mask", "shape", "squeeze", "subsequent_token_classifier", "subsequent_token_labels", "subsequent_token_logits", "subsequent_token_loss", "subsequent_token_mask", "to", "token_type_ids", "torch", "transpose", "tuple", "use_return_dict", "view", "zeros"], "bros/modeling_bros.py:BrosSpadeELForTokenClassification.__init__": ["Model", "ModelModel", "ModelPreTrainedModel", "ModelRelationExtractor", "ModelSpadeELForTokenClassification", "__init__", "backbone_hidden_size", "class", "classifier_dropout", "config", "def", "else", "entity_linker", "hasattr", "hidden_dropout_prob", "hidden_size", "if", "n_relations", "num_labels", "post_init", "self", "super"], "bros/modeling_bros.py:BrosSpadeELForTokenClassification.forward": ["CrossEntropyLoss", "Model", "ModelPreTrainedModel", "ModelSpadeELForTokenClassification", "None", "Tensor", "TokenClassifierOutput", "True", "attention_mask", "attentions", "axis", "batch_size", "bbox", "bbox_first_token_mask", "bool", "cat", "class", "config", "contiguous", "def", "device", "dtype", "else", "entity_linker", "eye", "finfo", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "last_hidden_states", "logits", "loss", "loss_fct", "mask", "masked_fill", "max_seq_length", "min", "not", "output_attentions", "output_hidden_states", "outputs", "position_ids", "r", "return", "return_dict", "self", "self_token_mask", "shape", "squeeze", "to", "token_type_ids", "torch", "transpose", "tuple", "use_return_dict", "view", "zeros"], "camembert/modeling_camembert.py:CamembertEmbeddings.__init__": ["Dropout", "Embedding", "False", "LayerNorm", "ModelEmbeddings", "Module", "__init__", "arange", "class", "config", "def", "dropout", "dtype", "eps", "expand", "hidden_dropout_prob", "hidden_size", "layer_norm_eps", "long", "max_position_embeddings", "nn", "pad_token_id", "padding_idx", "persistent", "position_embeddings", "position_ids", "register_buffer", "self", "size", "super", "token_type_embeddings", "token_type_ids", "torch", "type_vocab_size", "vocab_size", "word_embeddings", "zeros"], "camembert/modeling_camembert.py:CamembertEmbeddings.forward": ["FloatTensor", "LayerNorm", "LongTensor", "ModelEmbeddings", "Module", "None", "Tensor", "batch_size", "buffered_token_type_ids", "class", "create_position_ids_from_input_ids", "create_position_ids_from_inputs_embeds", "def", "device", "dim", "dropout", "dtype", "else", "embeddings", "expand", "forward", "gather", "hasattr", "if", "index", "input_ids", "input_shape", "inputs_embeds", "int", "is", "long", "nn", "not", "padding_idx", "past_key_values_length", "position_embeddings", "position_ids", "return", "self", "seq_length", "shape", "size", "token_type_embeddings", "token_type_ids", "torch", "word_embeddings", "zeros"], "camembert/modeling_camembert.py:CamembertEmbeddings.create_position_ids_from_inputs_embeds": ["ModelEmbeddings", "Module", "arange", "class", "create_position_ids_from_inputs_embeds", "def", "device", "dtype", "expand", "input_shape", "inputs_embeds", "long", "nn", "padding_idx", "position_ids", "return", "sequence_length", "size", "torch", "unsqueeze"], "camembert/modeling_camembert.py:CamembertEmbeddings.create_position_ids_from_input_ids": ["ModelEmbeddings", "Module", "class", "create_position_ids_from_input_ids", "cumsum", "def", "dim", "incremental_indices", "input_ids", "int", "long", "mask", "ne", "nn", "padding_idx", "past_key_values_length", "return", "torch", "type_as"], "camembert/modeling_camembert.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "contiguous", "def", "dim", "dropout", "float", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "shape", "size", "softmax", "torch", "training", "transpose", "value"], "camembert/modeling_camembert.py:CamembertSelfAttention.__init__": ["Dropout", "False", "Linear", "ModelSelfAttention", "Module", "None", "The", "ValueError", "__init__", "a", "all_head_size", "and", "attention", "attention_head_size", "attention_probs_dropout_prob", "class", "config", "def", "dropout", "embedding_size", "f", "hasattr", "heads", "hidden", "hidden_size", "if", "int", "is", "is_causal", "is_decoder", "key", "layer_idx", "multiple", "nn", "not", "num_attention_heads", "number", "of", "query", "raise", "scaling", "self", "size", "super", "the", "value"], "camembert/modeling_camembert.py:CamembertSelfAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "EncoderDecoderCache", "FloatTensor", "ModelSelfAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_attn_implementation", "attention_head_size", "attention_interface", "attention_mask", "attn_output", "attn_weights", "cache_position", "class", "config", "contiguous", "current_past_key_values", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "hidden_shape", "hidden_states", "if", "input_shape", "is", "isinstance", "key", "key_layer", "kwargs", "layer_idx", "nn", "not", "p", "past_key_values", "query", "query_layer", "reshape", "return", "scaling", "self", "self_attention_cache", "shape", "torch", "training", "transpose", "tuple", "update", "value", "value_layer", "view"], "camembert/modeling_camembert.py:CamembertCrossAttention.__init__": ["Dropout", "False", "Linear", "ModelCrossAttention", "Module", "None", "The", "ValueError", "__init__", "a", "all_head_size", "and", "attention", "attention_head_size", "attention_probs_dropout_prob", "class", "config", "def", "dropout", "embedding_size", "f", "hasattr", "heads", "hidden", "hidden_size", "if", "int", "is", "is_causal", "key", "layer_idx", "multiple", "nn", "not", "num_attention_heads", "number", "of", "query", "raise", "scaling", "self", "size", "super", "the", "value"], "camembert/modeling_camembert.py:CamembertCrossAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "EncoderDecoderCache", "False", "FloatTensor", "ModelCrossAttention", "Module", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "_attn_implementation", "and", "attention_head_size", "attention_interface", "attention_mask", "attn_output", "attn_weights", "bsz", "class", "config", "contiguous", "cross_attention_cache", "def", "dropout", "eager", "eager_attention_forward", "else", "encoder_hidden_states", "forward", "get", "hidden_states", "if", "is", "is_updated", "key", "key_layer", "keys", "kv_input_shape", "kwargs", "layer_idx", "layers", "nn", "not", "p", "past_key_values", "q_input_shape", "query", "query_layer", "reshape", "return", "scaling", "self", "shape", "src_len", "tgt_len", "torch", "training", "transpose", "tuple", "update", "value", "value_layer", "values", "view"], "camembert/modeling_camembert.py:CamembertSelfOutput.__init__": ["Dropout", "LayerNorm", "Linear", "ModelSelfOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "eps", "hidden_dropout_prob", "hidden_size", "layer_norm_eps", "nn", "self", "super"], "camembert/modeling_camembert.py:CamembertSelfOutput.forward": ["LayerNorm", "ModelSelfOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "camembert/modeling_camembert.py:CamembertAttention.__init__": ["False", "ModelAttention", "ModelCrossAttention", "ModelSelfAttention", "ModelSelfOutput", "Module", "None", "__init__", "attention_class", "class", "config", "def", "else", "if", "is_causal", "is_cross_attention", "layer_idx", "nn", "output", "self", "super"], "camembert/modeling_camembert.py:CamembertAttention.forward": ["Cache", "FloatTensor", "ModelAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attention_output", "attn_weights", "cache_position", "class", "def", "else", "encoder_attention_mask", "encoder_hidden_states", "forward", "hidden_states", "if", "is_cross_attention", "kwargs", "nn", "not", "output", "past_key_values", "return", "self", "torch", "tuple"], "camembert/modeling_camembert.py:CamembertIntermediate.__init__": ["ACT2FN", "Linear", "ModelIntermediate", "Module", "__init__", "class", "config", "def", "dense", "else", "hidden_act", "hidden_size", "if", "intermediate_act_fn", "intermediate_size", "isinstance", "nn", "self", "str", "super"], "camembert/modeling_camembert.py:CamembertIntermediate.forward": ["ModelIntermediate", "Module", "Tensor", "class", "def", "dense", "forward", "hidden_states", "intermediate_act_fn", "nn", "return", "self", "torch"], "camembert/modeling_camembert.py:CamembertOutput.__init__": ["Dropout", "LayerNorm", "Linear", "ModelOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "eps", "hidden_dropout_prob", "hidden_size", "intermediate_size", "layer_norm_eps", "nn", "self", "super"], "camembert/modeling_camembert.py:CamembertOutput.forward": ["LayerNorm", "ModelOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "camembert/modeling_camembert.py:CamembertLayer.__init__": ["False", "GradientCheckpointingLayer", "ModelAttention", "ModelIntermediate", "ModelLayer", "ModelOutput", "None", "True", "ValueError", "__init__", "a", "add_cross_attention", "added", "as", "attention", "be", "chunk_size_feed_forward", "class", "config", "cross", "crossattention", "decoder", "def", "f", "if", "intermediate", "is", "is_causal", "is_cross_attention", "is_decoder", "layer_idx", "model", "not", "output", "raise", "self", "seq_len_dim", "should", "super", "used"], "camembert/modeling_camembert.py:CamembertLayer.forward": ["Cache", "FloatTensor", "GradientCheckpointingLayer", "If", "ModelLayer", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "ValueError", "_", "add_cross_attention", "and", "apply_chunking_to_forward", "are", "attention", "attention_mask", "attention_output", "be", "by", "cache_position", "chunk_size_feed_forward", "class", "config", "cross", "cross_attention_output", "crossattention", "def", "encoder_attention_mask", "encoder_hidden_states", "f", "feed_forward_chunk", "forward", "has", "hasattr", "hidden_states", "if", "instantiated", "is", "is_decoder", "kwargs", "layer_output", "layers", "not", "passed", "past_key_values", "raise", "return", "self", "self_attention_output", "seq_len_dim", "setting", "to", "torch", "tuple", "with"], "camembert/modeling_camembert.py:CamembertLayer.feed_forward_chunk": ["GradientCheckpointingLayer", "ModelLayer", "attention_output", "class", "def", "feed_forward_chunk", "intermediate", "intermediate_output", "layer_output", "output", "return", "self"], "camembert/modeling_camembert.py:CamembertLMHead.__init__": ["LayerNorm", "Linear", "ModelLMHead", "Module", "Parameter", "__init__", "bias", "class", "config", "decoder", "def", "dense", "eps", "hidden_size", "layer_norm", "layer_norm_eps", "nn", "self", "super", "torch", "vocab_size", "zeros"], "camembert/modeling_camembert.py:CamembertLMHead.forward": ["ModelLMHead", "Module", "class", "decoder", "def", "dense", "features", "forward", "gelu", "kwargs", "layer_norm", "nn", "return", "self", "x"], "camembert/modeling_camembert.py:CamembertPreTrainedModel._init_weights": ["ModelEmbeddings", "ModelLMHead", "ModelPreTrainedModel", "PreTrainedModel", "_init_weights", "arange", "bias", "class", "copy_", "def", "elif", "expand", "if", "init", "isinstance", "module", "position_ids", "self", "shape", "super", "token_type_ids", "torch", "zeros_"], "camembert/modeling_camembert.py:CamembertEncoder.__init__": ["ModelEncoder", "ModelLayer", "Module", "ModuleList", "__init__", "class", "config", "def", "for", "i", "in", "layer", "layer_idx", "nn", "num_hidden_layers", "range", "self", "super"], "camembert/modeling_camembert.py:CamembertEncoder.forward": ["BaseModelOutputWithPastAndCrossAttentions", "Cache", "FloatTensor", "ModelEncoder", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "bool", "cache_position", "class", "def", "else", "encoder_attention_mask", "encoder_hidden_states", "enumerate", "for", "forward", "hidden_states", "i", "if", "in", "kwargs", "last_hidden_state", "layer", "layer_module", "nn", "past_key_values", "return", "self", "torch", "tuple", "use_cache"], "camembert/modeling_camembert.py:CamembertPooler.__init__": ["Linear", "ModelPooler", "Module", "Tanh", "__init__", "activation", "class", "config", "def", "dense", "hidden_size", "nn", "self", "super"], "camembert/modeling_camembert.py:CamembertPooler.forward": ["ModelPooler", "Module", "Tensor", "activation", "class", "def", "dense", "first_token_tensor", "forward", "hidden_states", "nn", "pooled_output", "return", "self", "torch"], "camembert/modeling_camembert.py:CamembertModel.__init__": ["False", "ModelEmbeddings", "ModelEncoder", "ModelModel", "ModelPooler", "ModelPreTrainedModel", "None", "True", "__init__", "add_pooling_layer", "class", "config", "def", "else", "embeddings", "encoder", "gradient_checkpointing", "if", "pooler", "post_init", "r", "self", "super"], "camembert/modeling_camembert.py:CamembertModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "return", "self", "word_embeddings"], "camembert/modeling_camembert.py:CamembertModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "self", "set_input_embeddings", "value", "word_embeddings"], "camembert/modeling_camembert.py:CamembertModel.forward": ["BaseModelOutputWithPoolingAndCrossAttentions", "Cache", "DynamicCache", "EncoderDecoderCache", "False", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "_create_attention_masks", "and", "arange", "attention_mask", "bool", "cache_position", "class", "config", "def", "device", "else", "embedding_output", "embeddings", "encoder", "encoder_attention_mask", "encoder_hidden_states", "encoder_outputs", "exactly", "forward", "get_seq_length", "if", "input_ids", "inputs_embeds", "is", "is_decoder", "is_encoder_decoder", "kwargs", "last_hidden_state", "must", "not", "of", "one", "or", "past_key_values", "past_key_values_length", "pooled_output", "pooler", "pooler_output", "position_ids", "raise", "return", "self", "seq_length", "sequence_output", "shape", "specify", "token_type_ids", "torch", "tuple", "use_cache"], "camembert/modeling_camembert.py:CamembertModel._create_attention_masks": ["ModelModel", "ModelPreTrainedModel", "None", "_create_attention_masks", "attention_mask", "cache_position", "class", "config", "create_bidirectional_mask", "create_causal_mask", "def", "else", "embedding_output", "encoder_attention_mask", "encoder_hidden_states", "if", "input_embeds", "is", "is_decoder", "not", "past_key_values", "return", "self"], "camembert/modeling_camembert.py:CamembertForMaskedLM.__init__": ["False", "If", "ModelForMaskedLM", "ModelLMHead", "ModelModel", "ModelPreTrainedModel", "__init__", "add_pooling_layer", "attention", "bi", "class", "config", "def", "directional", "for", "if", "is_decoder", "lm_head", "logger", "make", "post_init", "roberta", "self", "super", "sure", "to", "use", "want", "warning", "you"], "camembert/modeling_camembert.py:CamembertForMaskedLM.get_output_embeddings": ["ModelForMaskedLM", "ModelPreTrainedModel", "class", "decoder", "def", "get_output_embeddings", "lm_head", "return", "self"], "camembert/modeling_camembert.py:CamembertForMaskedLM.set_output_embeddings": ["ModelForMaskedLM", "ModelPreTrainedModel", "class", "decoder", "def", "lm_head", "new_embeddings", "self", "set_output_embeddings"], "camembert/modeling_camembert.py:CamembertForMaskedLM.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "MaskedLMOutput", "ModelForMaskedLM", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "attention_mask", "attentions", "class", "config", "def", "device", "encoder_attention_mask", "encoder_hidden_states", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "lm_head", "logits", "loss", "loss_fct", "masked_lm_loss", "not", "outputs", "position_ids", "prediction_scores", "r", "return", "return_dict", "roberta", "self", "sequence_output", "to", "token_type_ids", "torch", "tuple", "view", "vocab_size"], "camembert/modeling_camembert.py:CamembertClassificationHead.__init__": ["Dropout", "Linear", "ModelClassificationHead", "Module", "None", "__init__", "class", "classifier_dropout", "config", "def", "dense", "dropout", "else", "hidden_dropout_prob", "hidden_size", "if", "is", "nn", "not", "num_labels", "out_proj", "self", "super"], "camembert/modeling_camembert.py:CamembertClassificationHead.forward": ["ModelClassificationHead", "Module", "class", "def", "dense", "dropout", "features", "forward", "kwargs", "nn", "out_proj", "return", "self", "tanh", "torch", "x"], "camembert/modeling_camembert.py:CamembertForSequenceClassification.__init__": ["False", "ModelClassificationHead", "ModelForSequenceClassification", "ModelModel", "ModelPreTrainedModel", "__init__", "add_pooling_layer", "class", "classifier", "config", "def", "num_labels", "post_init", "roberta", "self", "super"], "camembert/modeling_camembert.py:CamembertForSequenceClassification.forward": ["BCEWithLogitsLoss", "CrossEntropyLoss", "FloatTensor", "LongTensor", "MSELoss", "ModelForSequenceClassification", "ModelPreTrainedModel", "None", "SequenceClassifierOutput", "Tensor", "TransformersKwargs", "True", "Unpack", "and", "attention_mask", "attentions", "class", "classifier", "config", "def", "device", "dtype", "elif", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "kwargs", "labels", "logits", "long", "loss", "loss_fct", "multi_label_classification", "not", "num_labels", "or", "outputs", "position_ids", "problem_type", "r", "regression", "return", "return_dict", "roberta", "self", "sequence_output", "single_label_classification", "squeeze", "to", "token_type_ids", "torch", "tuple", "view"], "camembert/modeling_camembert.py:CamembertForMultipleChoice.__init__": ["Dropout", "False", "Linear", "ModelForMultipleChoice", "ModelModel", "ModelPreTrainedModel", "__init__", "add_pooling_layer", "class", "classifier", "config", "def", "dropout", "hidden_dropout_prob", "hidden_size", "nn", "post_init", "roberta", "self", "super"], "camembert/modeling_camembert.py:CamembertForMultipleChoice.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "ModelForMultipleChoice", "ModelPreTrainedModel", "MultipleChoiceModelOutput", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "attention_mask", "attentions", "class", "classifier", "def", "device", "dropout", "else", "flat_attention_mask", "flat_input_ids", "flat_inputs_embeds", "flat_position_ids", "flat_token_type_ids", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "not", "num_choices", "outputs", "pooled_output", "position_ids", "r", "reshaped_logits", "return", "return_dict", "roberta", "self", "shape", "size", "to", "token_type_ids", "torch", "tuple", "view"], "camembert/modeling_camembert.py:CamembertForTokenClassification.__init__": ["Dropout", "False", "Linear", "ModelForTokenClassification", "ModelModel", "ModelPreTrainedModel", "None", "__init__", "add_pooling_layer", "class", "classifier", "classifier_dropout", "config", "def", "dropout", "else", "hidden_dropout_prob", "hidden_size", "if", "is", "nn", "not", "num_labels", "post_init", "roberta", "self", "super"], "camembert/modeling_camembert.py:CamembertForTokenClassification.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "ModelForTokenClassification", "ModelPreTrainedModel", "None", "Tensor", "TokenClassifierOutput", "TransformersKwargs", "True", "Unpack", "attention_mask", "attentions", "class", "classifier", "def", "device", "dropout", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "not", "num_labels", "outputs", "position_ids", "r", "return", "return_dict", "roberta", "self", "sequence_output", "to", "token_type_ids", "torch", "tuple", "view"], "camembert/modeling_camembert.py:CamembertForQuestionAnswering.__init__": ["False", "Linear", "ModelForQuestionAnswering", "ModelModel", "ModelPreTrainedModel", "__init__", "add_pooling_layer", "class", "config", "def", "hidden_size", "nn", "num_labels", "post_init", "qa_outputs", "roberta", "self", "super"], "camembert/modeling_camembert.py:CamembertForQuestionAnswering.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "ModelForQuestionAnswering", "ModelPreTrainedModel", "None", "QuestionAnsweringModelOutput", "Tensor", "TransformersKwargs", "True", "Unpack", "and", "attention_mask", "attentions", "clamp", "class", "contiguous", "def", "dim", "end_logits", "end_loss", "end_positions", "forward", "hidden_states", "if", "ignore_index", "ignored_index", "input_ids", "inputs_embeds", "is", "kwargs", "len", "logits", "loss", "loss_fct", "not", "outputs", "position_ids", "qa_outputs", "r", "return", "return_dict", "roberta", "self", "sequence_output", "size", "split", "squeeze", "start_logits", "start_loss", "start_positions", "token_type_ids", "torch", "total_loss", "tuple"], "camembert/modeling_camembert.py:CamembertForCausalLM.__init__": ["False", "GenerationMixin", "If", "ModelForCausalLM", "ModelLMHead", "ModelLMHeadModel", "ModelModel", "ModelPreTrainedModel", "True", "__init__", "a", "add", "add_pooling_layer", "as", "class", "config", "def", "if", "is_decoder", "lm_head", "logger", "not", "post_init", "roberta", "self", "standalone", "super", "to", "use", "want", "warning", "you"], "camembert/modeling_camembert.py:CamembertForCausalLM.get_output_embeddings": ["GenerationMixin", "ModelForCausalLM", "ModelPreTrainedModel", "class", "decoder", "def", "get_output_embeddings", "lm_head", "return", "self"], "camembert/modeling_camembert.py:CamembertForCausalLM.set_output_embeddings": ["GenerationMixin", "ModelForCausalLM", "ModelPreTrainedModel", "class", "decoder", "def", "lm_head", "new_embeddings", "self", "set_output_embeddings"], "camembert/modeling_camembert.py:CamembertForCausalLM.forward": ["BaseModelOutputWithPoolingAndCrossAttentions", "CausalLMOutputWithCrossAttentions", "False", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForCausalLM", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "cross_attentions", "def", "else", "encoder_attention_mask", "encoder_hidden_states", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "not", "outputs", "past_key_values", "position_ids", "r", "return", "return_dict", "roberta", "self", "slice", "slice_indices", "token_type_ids", "torch", "tuple", "use_cache", "vocab_size"], "canine/modeling_canine.py:CanineEmbeddings.__init__": ["Dropout", "Embedding", "False", "HashBucketCodepointEmbedder_", "LayerNorm", "ModelEmbeddings", "Module", "__init__", "arange", "char_position_embeddings", "class", "config", "def", "dropout", "eps", "expand", "f", "for", "hidden_dropout_prob", "hidden_size", "i", "in", "layer_norm_eps", "max_position_embeddings", "name", "nn", "num_hash_buckets", "num_hash_functions", "persistent", "position_ids", "range", "register_buffer", "self", "setattr", "shard_embedding_size", "super", "token_type_embeddings", "torch", "type_vocab_size"], "canine/modeling_canine.py:CanineEmbeddings._hash_bucket_tensors": ["ModelEmbeddings", "Module", "ValueError", "_PRIMES", "_hash_bucket_tensors", "append", "be", "class", "def", "f", "for", "hashed", "if", "in", "input_ids", "int", "len", "must", "nn", "num_buckets", "num_hashes", "prime", "primes", "raise", "result_tensors", "return", "self"], "canine/modeling_canine.py:CanineEmbeddings._embed_hash_buckets": ["Expected", "HashBucketCodepointEmbedder_", "ModelEmbeddings", "Module", "ValueError", "_embed_hash_buckets", "_hash_bucket_tensors", "append", "cat", "class", "def", "dim", "embedding_shards", "embedding_size", "enumerate", "f", "for", "getattr", "hash_bucket_ids", "hash_bucket_tensors", "i", "if", "in", "input_ids", "int", "name", "nn", "num_buckets", "num_hashes", "raise", "return", "self", "shard_embeddings", "torch"], "canine/modeling_canine.py:CanineEmbeddings.forward": ["FloatTensor", "LayerNorm", "LongTensor", "ModelEmbeddings", "Module", "None", "_embed_hash_buckets", "char_position_embeddings", "class", "config", "def", "device", "dropout", "dtype", "else", "embeddings", "forward", "hidden_size", "if", "input_ids", "input_shape", "inputs_embeds", "is", "long", "nn", "not", "num_hash_buckets", "num_hash_functions", "position_embeddings", "position_ids", "return", "self", "seq_length", "size", "token_type_embeddings", "token_type_ids", "torch", "zeros"], "canine/modeling_canine.py:CharactersToMolecules.__init__": ["ACT2FN", "Conv1d", "LayerNorm", "ModelToMolecules", "Module", "__init__", "activation", "class", "config", "conv", "def", "downsampling_rate", "eps", "hidden_act", "hidden_size", "in_channels", "kernel_size", "layer_norm_eps", "nn", "out_channels", "self", "stride", "super"], "canine/modeling_canine.py:CharactersToMolecules.forward": ["LayerNorm", "ModelToMolecules", "Module", "Tensor", "activation", "cat", "char_encoding", "class", "cls_encoding", "conv", "def", "dim", "downsampled", "downsampled_truncated", "forward", "nn", "result", "return", "self", "torch", "transpose"], "canine/modeling_canine.py:ConvProjection.__init__": ["ACT2FN", "Dropout", "LayerNorm", "Model", "Model1d", "ModelProjection", "Module", "__init__", "activation", "class", "config", "def", "dropout", "eps", "hidden_act", "hidden_dropout_prob", "hidden_size", "in_channels", "kernel_size", "layer_norm_eps", "nn", "out_channels", "self", "stride", "super", "upsampling_kernel_size"], "canine/modeling_canine.py:ConvProjection.forward": ["ConstantPad1d", "LayerNorm", "Model", "ModelForMaskedLM", "ModelProjection", "Module", "None", "NotImplementedError", "Tensor", "activation", "class", "config", "currently", "def", "dropout", "else", "final_char_seq", "final_seq_char_positions", "forward", "if", "inputs", "is", "nn", "not", "pad", "pad_beg", "pad_end", "pad_total", "query_seq", "raise", "result", "return", "self", "supported", "torch", "transpose", "upsampling_kernel_size"], "canine/modeling_canine.py:CanineSelfAttention.__init__": ["Dropout", "Linear", "ModelSelfAttention", "Module", "The", "ValueError", "__init__", "a", "all_head_size", "and", "attention", "attention_head_size", "attention_probs_dropout_prob", "class", "config", "def", "dropout", "embedding_size", "f", "hasattr", "heads", "hidden", "hidden_size", "if", "int", "is", "key", "multiple", "nn", "not", "num_attention_heads", "number", "of", "query", "raise", "self", "size", "super", "the", "value"], "canine/modeling_canine.py:CanineSelfAttention.forward": ["False", "FloatTensor", "ModelSelfAttention", "Module", "None", "Tensor", "_", "all_head_size", "attention_head_size", "attention_mask", "attention_probs", "attention_scores", "batch_size", "bool", "class", "context_layer", "contiguous", "def", "dim", "dropout", "dtype", "else", "finfo", "float", "forward", "from_tensor", "functional", "if", "is", "key", "key_layer", "math", "matmul", "min", "ndim", "new_context_layer_shape", "nn", "not", "num_attention_heads", "output_attentions", "outputs", "permute", "query", "query_layer", "return", "self", "seq_length", "shape", "size", "softmax", "sqrt", "to_tensor", "torch", "transpose", "tuple", "unsqueeze", "value", "value_layer", "view"], "canine/modeling_canine.py:CanineSelfOutput.__init__": ["Dropout", "LayerNorm", "Linear", "ModelSelfOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "eps", "hidden_dropout_prob", "hidden_size", "layer_norm_eps", "nn", "self", "super"], "canine/modeling_canine.py:CanineSelfOutput.forward": ["FloatTensor", "LayerNorm", "ModelSelfOutput", "Module", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch", "tuple"], "canine/modeling_canine.py:CanineAttention.__init__": ["False", "ModelAttention", "ModelSelfAttention", "ModelSelfOutput", "Module", "ValueError", "__init__", "always_attend_to_first_position", "attend_from_chunk_stride", "attend_from_chunk_width", "attend_to_chunk_stride", "attend_to_chunk_width", "bool", "cause", "class", "config", "def", "first_position_attends_to_all", "get", "if", "int", "local", "nn", "output", "positions", "raise", "self", "sequence", "skipped", "super", "to", "would"], "canine/modeling_canine.py:CanineAttention.forward": ["Check", "Expected", "False", "FloatTensor", "ModelAttention", "Module", "None", "ValueError", "always_attend_to_first_position", "and", "append", "attend_from_chunk_stride", "attend_from_chunk_width", "attend_to_chunk_stride", "attend_to_chunk_width", "attention_mask", "attention_mask_chunk", "attention_output", "attention_output_chunks", "attention_outputs_chunk", "attention_probs_chunks", "bool", "cat", "chunk_end", "chunk_start", "class", "cls_attention_mask", "cls_position", "def", "dim", "else", "f", "first_position_attends_to_all", "for", "forward", "from_chunks", "from_end", "from_seq_length", "from_start", "from_tensor", "from_tensor_chunk", "have", "hidden_states", "if", "in", "len", "local", "min", "nn", "not", "number", "of", "output", "output_attentions", "outputs", "raise", "range", "return", "same", "self", "self_outputs", "shape", "strides", "to", "to_chunks", "to_end", "to_seq_length", "to_start", "to_tensor", "to_tensor_chunk", "torch", "tuple", "zip"], "canine/modeling_canine.py:CanineIntermediate.__init__": ["ACT2FN", "Linear", "ModelIntermediate", "Module", "__init__", "class", "config", "def", "dense", "else", "hidden_act", "hidden_size", "if", "intermediate_act_fn", "intermediate_size", "isinstance", "nn", "self", "str", "super"], "canine/modeling_canine.py:CanineIntermediate.forward": ["FloatTensor", "ModelIntermediate", "Module", "class", "def", "dense", "forward", "hidden_states", "intermediate_act_fn", "nn", "return", "self", "torch"], "canine/modeling_canine.py:CanineOutput.__init__": ["Dropout", "LayerNorm", "Linear", "ModelOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "eps", "hidden_dropout_prob", "hidden_size", "intermediate_size", "layer_norm_eps", "nn", "self", "super"], "canine/modeling_canine.py:CanineOutput.forward": ["FloatTensor", "LayerNorm", "ModelOutput", "Module", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch", "tuple"], "canine/modeling_canine.py:CanineLayer.__init__": ["GradientCheckpointingLayer", "ModelAttention", "ModelIntermediate", "ModelLayer", "ModelOutput", "__init__", "always_attend_to_first_position", "attend_from_chunk_stride", "attend_from_chunk_width", "attend_to_chunk_stride", "attend_to_chunk_width", "attention", "chunk_size_feed_forward", "class", "config", "def", "first_position_attends_to_all", "intermediate", "local", "output", "self", "seq_len_dim", "super"], "canine/modeling_canine.py:CanineLayer.forward": ["False", "FloatTensor", "GradientCheckpointingLayer", "ModelLayer", "None", "apply_chunking_to_forward", "attention", "attention_mask", "attention_output", "bool", "chunk_size_feed_forward", "class", "def", "feed_forward_chunk", "forward", "hidden_states", "layer_output", "output_attentions", "outputs", "return", "self", "self_attention_outputs", "seq_len_dim", "torch", "tuple"], "canine/modeling_canine.py:CanineLayer.feed_forward_chunk": ["GradientCheckpointingLayer", "ModelLayer", "attention_output", "class", "def", "feed_forward_chunk", "intermediate", "intermediate_output", "layer_output", "output", "return", "self"], "canine/modeling_canine.py:CanineEncoder.__init__": ["False", "ModelEncoder", "ModelLayer", "Module", "ModuleList", "_", "__init__", "always_attend_to_first_position", "attend_from_chunk_stride", "attend_from_chunk_width", "attend_to_chunk_stride", "attend_to_chunk_width", "class", "config", "def", "first_position_attends_to_all", "for", "gradient_checkpointing", "in", "layer", "local", "nn", "num_hidden_layers", "range", "self", "super"], "canine/modeling_canine.py:CanineEncoder.forward": ["BaseModelOutput", "False", "FloatTensor", "ModelEncoder", "Module", "None", "True", "all_hidden_states", "all_self_attentions", "attention_mask", "attentions", "bool", "class", "def", "else", "enumerate", "for", "forward", "hidden_states", "i", "if", "in", "is", "last_hidden_state", "layer", "layer_module", "layer_outputs", "nn", "not", "output_attentions", "output_hidden_states", "return", "return_dict", "self", "torch", "tuple", "v"], "canine/modeling_canine.py:CaninePooler.__init__": ["Linear", "ModelPooler", "Module", "Tanh", "__init__", "activation", "class", "config", "def", "dense", "hidden_size", "nn", "self", "super"], "canine/modeling_canine.py:CaninePooler.forward": ["FloatTensor", "ModelPooler", "Module", "activation", "class", "def", "dense", "first_token_tensor", "forward", "hidden_states", "nn", "pooled_output", "return", "self", "torch", "tuple"], "canine/modeling_canine.py:CaninePredictionHeadTransform.__init__": ["ACT2FN", "LayerNorm", "Linear", "ModelPredictionHeadTransform", "Module", "__init__", "class", "config", "def", "dense", "else", "eps", "hidden_act", "hidden_size", "if", "isinstance", "layer_norm_eps", "nn", "self", "str", "super", "transform_act_fn"], "canine/modeling_canine.py:CaninePredictionHeadTransform.forward": ["FloatTensor", "LayerNorm", "ModelPredictionHeadTransform", "Module", "class", "def", "dense", "forward", "hidden_states", "nn", "return", "self", "torch", "transform_act_fn", "tuple"], "canine/modeling_canine.py:CanineLMPredictionHead.__init__": ["Linear", "ModelLMPredictionHead", "ModelPredictionHeadTransform", "Module", "Parameter", "True", "__init__", "bias", "class", "config", "decoder", "def", "hidden_size", "nn", "self", "super", "torch", "transform", "vocab_size", "zeros"], "canine/modeling_canine.py:CanineLMPredictionHead.forward": ["FloatTensor", "ModelLMPredictionHead", "Module", "class", "decoder", "def", "forward", "hidden_states", "nn", "return", "self", "torch", "transform", "tuple"], "canine/modeling_canine.py:CanineOnlyMLMHead.__init__": ["ModelLMPredictionHead", "ModelOnlyMLMHead", "Module", "__init__", "class", "config", "def", "nn", "predictions", "self", "super"], "canine/modeling_canine.py:CanineOnlyMLMHead.forward": ["ModelOnlyMLMHead", "Module", "Tensor", "class", "def", "forward", "nn", "prediction_scores", "predictions", "return", "self", "sequence_output", "torch", "tuple"], "canine/modeling_canine.py:CaninePreTrainedModel._init_weights": ["ModelEmbeddings", "ModelPreTrainedModel", "PreTrainedModel", "_init_weights", "arange", "class", "copy_", "def", "expand", "if", "init", "isinstance", "module", "position_ids", "self", "shape", "super", "torch"], "canine/modeling_canine.py:CanineModel.__init__": ["CharactersToMolecules", "ConvProjection", "False", "ModelEmbeddings", "ModelEncoder", "ModelModel", "ModelPooler", "ModelPreTrainedModel", "None", "True", "__init__", "add_pooling_layer", "always_attend_to_first_position", "attend_from_chunk_stride", "attend_from_chunk_width", "attend_to_chunk_stride", "attend_to_chunk_width", "char_embeddings", "chars_to_molecules", "class", "config", "copy", "deepcopy", "def", "else", "encoder", "final_char_encoder", "first_position_attends_to_all", "if", "initial_char_encoder", "local", "local_transformer_stride", "num_hidden_layers", "pooler", "post_init", "projection", "r", "self", "shallow_config", "super"], "canine/modeling_canine.py:CanineModel._create_3d_attention_mask_from_input_mask": ["ModelModel", "ModelPreTrainedModel", "_create_3d_attention_mask_from_input_mask", "batch_size", "broadcast_ones", "class", "def", "device", "dtype", "float", "float32", "from_seq_length", "from_tensor", "mask", "ones", "reshape", "return", "self", "shape", "size", "to_mask", "to_seq_length", "torch"], "canine/modeling_canine.py:CanineModel._downsample_attention_mask": ["MaxPool1d", "ModelModel", "ModelPreTrainedModel", "Tensor", "_downsample_attention_mask", "batch_size", "char_attention_mask", "char_seq_len", "class", "def", "dim", "downsampling_rate", "float", "int", "kernel_size", "molecule_attention_mask", "nn", "poolable_char_mask", "pooled_molecule_mask", "reshape", "return", "self", "shape", "squeeze", "stride", "torch"], "canine/modeling_canine.py:CanineModel._repeat_molecules": ["ModelModel", "ModelPreTrainedModel", "Tensor", "_repeat_molecules", "cat", "char_seq_length", "class", "config", "def", "dim", "downsampling_rate", "int", "last_molecule", "molecules", "molecules_without_extra_cls", "rate", "remainder_length", "remainder_repeated", "repeat_interleave", "repeated", "repeats", "return", "self", "torch"], "canine/modeling_canine.py:CanineModel.forward": ["FloatTensor", "LongTensor", "ModelModel", "ModelModelOutputWithPooling", "ModelPreTrainedModel", "None", "Tensor", "ValueError", "You", "_create_3d_attention_mask_from_input_mask", "_downsample_attention_mask", "_repeat_molecules", "all_hidden_states", "all_self_attentions", "and", "at", "attention_mask", "attentions", "batch_size", "bool", "both", "cannot", "cat", "char_attention_mask", "char_embeddings", "char_seq_length", "chars_to_molecules", "class", "concat", "config", "deep_encoder_hidden_states", "deep_encoder_self_attentions", "def", "device", "dim", "downsampling_rate", "dtype", "either", "elif", "else", "encoder", "encoder_outputs", "extended_attention_mask", "extended_molecule_attention_mask", "final_char_encoder", "final_chars_encoder_outputs", "for", "forward", "get_extended_attention_mask", "have", "hidden_states", "if", "in", "init_chars_encoder_outputs", "init_molecule_encoding", "initial_char_encoder", "input_char_embeddings", "input_char_encoding", "input_ids", "input_shape", "inputs_embeds", "is", "kwargs", "last_hidden_state", "long", "molecule_attention_mask", "molecule_sequence_output", "not", "ones", "or", "output", "output_attentions", "output_hidden_states", "pooled_output", "pooler", "pooler_output", "position_ids", "projection", "raise", "repeated_molecules", "return", "return_dict", "same", "self", "seq_length", "sequence_output", "shape", "size", "specify", "the", "time", "to", "token_type_ids", "torch", "tuple", "use_return_dict", "v", "warn_if_padding_and_no_attention_mask", "zeros"], "canine/modeling_canine.py:CanineForSequenceClassification.__init__": ["Dropout", "Linear", "Model", "ModelForSequenceClassification", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "classifier", "config", "def", "dropout", "hidden_dropout_prob", "hidden_size", "nn", "num_labels", "post_init", "self", "super"], "canine/modeling_canine.py:CanineForSequenceClassification.forward": ["BCEWithLogitsLoss", "CrossEntropyLoss", "FloatTensor", "LongTensor", "MSELoss", "Model", "ModelForSequenceClassification", "ModelPreTrainedModel", "None", "SequenceClassifierOutput", "and", "attention_mask", "attentions", "bool", "class", "classifier", "config", "def", "dropout", "dtype", "elif", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "kwargs", "labels", "logits", "long", "loss", "loss_fct", "multi_label_classification", "not", "num_labels", "or", "output", "output_attentions", "output_hidden_states", "outputs", "pooled_output", "position_ids", "problem_type", "r", "regression", "return", "return_dict", "self", "single_label_classification", "squeeze", "token_type_ids", "torch", "tuple", "use_return_dict", "view"], "canine/modeling_canine.py:CanineForMultipleChoice.__init__": ["Dropout", "Linear", "Model", "ModelForMultipleChoice", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "classifier", "config", "def", "dropout", "hidden_dropout_prob", "hidden_size", "nn", "post_init", "self", "super"], "canine/modeling_canine.py:CanineForMultipleChoice.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "Model", "ModelForMultipleChoice", "ModelPreTrainedModel", "MultipleChoiceModelOutput", "None", "attention_mask", "attentions", "bool", "class", "classifier", "config", "def", "dropout", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "not", "num_choices", "output", "output_attentions", "output_hidden_states", "outputs", "pooled_output", "position_ids", "r", "reshaped_logits", "return", "return_dict", "self", "shape", "size", "token_type_ids", "torch", "tuple", "use_return_dict", "view"], "canine/modeling_canine.py:CanineForTokenClassification.__init__": ["Dropout", "Linear", "Model", "ModelForTokenClassification", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "classifier", "config", "def", "dropout", "hidden_dropout_prob", "hidden_size", "nn", "num_labels", "post_init", "self", "super"], "canine/modeling_canine.py:CanineForTokenClassification.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "Model", "ModelForTokenClassification", "ModelPreTrainedModel", "None", "TokenClassifierOutput", "attention_mask", "attentions", "bool", "class", "classifier", "config", "def", "dropout", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "not", "num_labels", "output", "output_attentions", "output_hidden_states", "outputs", "position_ids", "r", "return", "return_dict", "self", "sequence_output", "token_type_ids", "torch", "tuple", "use_return_dict", "view"], "canine/modeling_canine.py:CanineForQuestionAnswering.__init__": ["Linear", "Model", "ModelForQuestionAnswering", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "config", "def", "hidden_size", "nn", "num_labels", "post_init", "qa_outputs", "self", "super"], "canine/modeling_canine.py:CanineForQuestionAnswering.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "Model", "ModelForQuestionAnswering", "ModelPreTrainedModel", "None", "QuestionAnsweringModelOutput", "and", "attention_mask", "attentions", "bool", "clamp_", "class", "config", "def", "dim", "else", "end_logits", "end_loss", "end_positions", "forward", "hidden_states", "if", "ignore_index", "ignored_index", "input_ids", "inputs_embeds", "is", "kwargs", "len", "logits", "loss", "loss_fct", "not", "output", "output_attentions", "output_hidden_states", "outputs", "position_ids", "qa_outputs", "return", "return_dict", "self", "sequence_output", "size", "split", "squeeze", "start_logits", "start_loss", "start_positions", "token_type_ids", "torch", "total_loss", "tuple", "use_return_dict"], "chameleon/modeling_chameleon.py:ChameleonRMSNorm.__init__": ["ModelRMSNorm", "Module", "Parameter", "__init__", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "chameleon/modeling_chameleon.py:ChameleonRMSNorm.forward": ["ModelRMSNorm", "Module", "True", "class", "def", "dtype", "float32", "forward", "hidden_states", "input_dtype", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "chameleon/modeling_chameleon.py:ChameleonRMSNorm.extra_repr": ["ModelRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "variance_epsilon", "weight"], "chameleon/modeling_chameleon.py:ChameleonRotaryEmbedding.__init__": ["Callable", "False", "ModelConfig", "ModelRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "chameleon/modeling_chameleon.py:ChameleonRotaryEmbedding.compute_default_rope_parameters": ["ModelConfig", "ModelRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "chameleon/modeling_chameleon.py:ChameleonRotaryEmbedding.forward": ["False", "ModelRotaryEmbedding", "Module", "None", "and", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "chameleon/modeling_chameleon.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "chameleon/modeling_chameleon.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cos", "def", "k", "k_embed", "q", "q_embed", "return", "rotate_half", "sin", "unsqueeze", "unsqueeze_dim"], "chameleon/modeling_chameleon.py:ChameleonMLP.__init__": ["ACT2FN", "Linear", "ModelMLP", "Module", "__init__", "act_fn", "bias", "class", "config", "def", "down_proj", "gate_proj", "hidden_act", "hidden_size", "intermediate_size", "mlp_bias", "nn", "self", "super", "up_proj"], "chameleon/modeling_chameleon.py:ChameleonMLP.forward": ["ModelMLP", "Module", "act_fn", "class", "def", "down_proj", "forward", "gate_proj", "nn", "return", "self", "up_proj", "x"], "chameleon/modeling_chameleon.py:ChameleonLayerNorm.__init__": ["LayerNorm", "ModelLayerNorm", "__init__", "args", "class", "def", "hidden_size", "kwargs", "nn", "normalized_shape", "self", "super"], "chameleon/modeling_chameleon.py:ChameleonLayerNorm.forward": ["F", "LayerNorm", "ModelLayerNorm", "None", "bias", "class", "def", "eps", "forward", "hidden_states", "layer_norm", "nn", "normalized_shape", "return", "self", "weight"], "chameleon/modeling_chameleon.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "chameleon/modeling_chameleon.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value", "value_states"], "chameleon/modeling_chameleon.py:ChameleonAttention.__init__": ["Instantiating", "Linear", "ModelAttention", "ModelConfig", "ModelLayerNorm", "Module", "None", "Please", "True", "ValueError", "__class__", "__init__", "__name__", "a", "and", "attention_bias", "attention_dropout", "be", "bias", "by", "caching", "call", "class", "config", "creating", "def", "divisible", "during", "errors", "f", "forward", "got", "head_dim", "hidden_size", "if", "int", "is", "is_causal", "k_norm", "k_proj", "layer_idx", "lead", "logger", "make", "max_position_embeddings", "model_parallel_size", "must", "nn", "not", "num_attention_heads", "num_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "passing", "provide", "q_norm", "q_proj", "raise", "recommended", "scaling", "self", "super", "sure", "the", "this", "to", "used", "v_proj", "warning_once", "when", "will", "without"], "chameleon/modeling_chameleon.py:ChameleonAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "False", "LongTensor", "ModelAttention", "Module", "None", "Tensor", "_", "_attn_implementation", "apply_rotary_pos_emb", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "bool", "bsz", "cache_kwargs", "cache_position", "class", "config", "contiguous", "cos", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "head_dim", "hidden_states", "if", "is", "k_norm", "k_proj", "key_states", "kwargs", "layer_idx", "nn", "not", "num_heads", "num_key_value_heads", "o_proj", "output_attentions", "past_key_values", "position_embeddings", "position_ids", "q_len", "q_norm", "q_proj", "query_states", "reshape", "return", "scaling", "self", "sin", "size", "torch", "training", "transpose", "tuple", "update", "use_cache", "v_proj", "value_states", "view"], "chameleon/modeling_chameleon.py:ChameleonDecoderLayer.__init__": ["GradientCheckpointingLayer", "ModelAttention", "ModelConfig", "ModelDecoderLayer", "ModelMLP", "ModelRMSNorm", "__init__", "class", "config", "def", "eps", "hidden_size", "input_layernorm", "int", "layer_idx", "mlp", "post_attention_layernorm", "rms_norm_eps", "self", "self_attn", "super"], "chameleon/modeling_chameleon.py:ChameleonDecoderLayer.forward": ["Cache", "False", "FloatTensor", "GradientCheckpointingLayer", "LongTensor", "ModelDecoderLayer", "None", "Tensor", "attention_mask", "bool", "cache_position", "class", "def", "forward", "hidden_states", "if", "input_layernorm", "kwargs", "mlp", "output_attentions", "outputs", "past_key_values", "position_embeddings", "position_ids", "post_attention_layernorm", "residual", "return", "self", "self_attn", "self_attn_weights", "torch", "tuple", "use_cache"], "chameleon/modeling_chameleon.py:ChameleonSwinDecoderLayer.__init__": ["GradientCheckpointingLayer", "ModelAttention", "ModelConfig", "ModelMLP", "ModelRMSNorm", "ModelSwinDecoderLayer", "__init__", "class", "config", "def", "eps", "hidden_size", "input_layernorm", "int", "layer_idx", "mlp", "post_attention_layernorm", "rms_norm_eps", "self", "self_attn", "super"], "chameleon/modeling_chameleon.py:ChameleonSwinDecoderLayer.forward": ["Cache", "False", "FloatTensor", "GradientCheckpointingLayer", "LongTensor", "ModelSwinDecoderLayer", "None", "Tensor", "attention_mask", "bool", "cache_position", "class", "def", "forward", "hidden_states", "if", "input_layernorm", "kwargs", "mlp", "output_attentions", "outputs", "past_key_values", "position_embeddings", "position_ids", "post_attention_layernorm", "residual", "return", "self", "self_attn", "self_attn_weights", "torch", "tuple", "use_cache"], "chameleon/modeling_chameleon.py:ChameleonVQVAEVectorQuantizer.__init__": ["Embedding", "ModelVQVAEVectorQuantizer", "Module", "__init__", "beta", "class", "config", "def", "embed_dim", "embedding", "embedding_dim", "getattr", "nn", "num_embeddings", "self", "super"], "chameleon/modeling_chameleon.py:ChameleonVQVAEVectorQuantizer.forward": ["ModelVQVAEVectorQuantizer", "Module", "Tensor", "True", "argmin", "bd", "beta", "bn", "class", "contiguous", "def", "detach", "dim", "distances", "dn", "einsum", "embedding", "embedding_dim", "forward", "hidden_state", "hidden_state_flattened", "hidden_state_quant", "keepdim", "loss", "mean", "min_encoding_indices", "nn", "permute", "return", "self", "shape", "sum", "torch", "transpose", "view", "weight"], "chameleon/modeling_chameleon.py:ChameleonVQVAEEncoderConvDownsample.__init__": ["Conv2d", "ModelVQVAEEncoderConvDownsample", "Module", "__init__", "class", "conv", "def", "in_channels", "kernel_size", "nn", "padding", "self", "stride", "super"], "chameleon/modeling_chameleon.py:ChameleonVQVAEEncoderConvDownsample.forward": ["F", "ModelVQVAEEncoderConvDownsample", "Module", "class", "constant", "conv", "def", "forward", "hidden_states", "mode", "nn", "pad", "return", "self", "value"], "chameleon/modeling_chameleon.py:ChameleonVQVAEEncoderResnetBlock.__init__": ["Conv2d", "Dropout", "False", "GroupNorm", "ModelVQVAEEncoderResnetBlock", "Module", "None", "True", "__init__", "affine", "class", "config", "conv1", "conv2", "conv_shortcut", "def", "dropout", "else", "eps", "if", "in_channels", "is", "kernel_size", "nin_shortcut", "nn", "norm1", "norm2", "num_channels", "num_groups", "out_channels", "padding", "self", "stride", "super", "torch", "use_conv_shortcut"], "chameleon/modeling_chameleon.py:ChameleonVQVAEEncoderResnetBlock.forward": ["ModelVQVAEEncoderResnetBlock", "Module", "class", "conv1", "conv2", "conv_shortcut", "def", "dropout", "else", "forward", "hidden_states", "if", "in_channels", "nin_shortcut", "nn", "norm1", "norm2", "out_channels", "residual", "return", "self", "sigmoid", "torch", "use_conv_shortcut"], "chameleon/modeling_chameleon.py:ChameleonVQVAEEncoderAttnBlock.__init__": ["Conv2d", "GroupNorm", "ModelVQVAEEncoderAttnBlock", "Module", "True", "__init__", "affine", "class", "def", "eps", "in_channels", "k", "kernel_size", "nn", "norm", "num_channels", "num_groups", "padding", "proj_out", "q", "self", "stride", "super", "torch", "v"], "chameleon/modeling_chameleon.py:ChameleonVQVAEEncoderAttnBlock.forward": ["F", "ModelVQVAEEncoderAttnBlock", "Module", "attn_output", "attn_weights", "batch_size", "bmm", "channels", "class", "def", "dim", "forward", "height", "hidden_states", "int", "k", "key_states", "nn", "norm", "permute", "proj_out", "q", "query_states", "reshape", "residual", "return", "self", "shape", "softmax", "torch", "v", "value_states", "width"], "chameleon/modeling_chameleon.py:ChameleonVQVAEEncoder.__init__": ["Conv2d", "GroupNorm", "Identity", "ModelVQVAEEncoder", "ModelVQVAEEncoderAttnBlock", "ModelVQVAEEncoderConvDownsample", "ModelVQVAEEncoderResnetBlock", "Module", "ModuleList", "None", "True", "__init__", "affine", "and", "append", "attn", "attn_1", "attn_resolutions", "attn_type", "base_channels", "block", "block_1", "block_2", "block_in", "block_out", "channel_multiplier", "class", "config", "conv_in", "conv_out", "curr_res", "def", "double_latent", "down", "downsample", "else", "eps", "for", "i_block", "i_level", "if", "in", "in_channel_multiplier", "in_channels", "is", "kernel_size", "latent_channels", "len", "mid", "nn", "norm_out", "not", "num_channels", "num_groups", "num_res_blocks", "num_resolutions", "out_channels", "padding", "range", "resolution", "self", "stride", "super", "torch", "tuple", "vanilla"], "chameleon/modeling_chameleon.py:ChameleonVQVAEEncoder.forward": ["LongTensor", "ModelVQVAEEncoder", "Module", "append", "attn", "attn_1", "block", "block_1", "block_2", "class", "conv_in", "conv_out", "def", "down", "downsample", "for", "forward", "hidden_state", "hidden_states", "i_block", "i_level", "if", "in", "last_hidden_state", "len", "mid", "nn", "norm_out", "num_res_blocks", "num_resolutions", "pixel_values", "range", "return", "self", "sigmoid", "torch"], "chameleon/modeling_chameleon.py:ChameleonImageVocabularyMapping.__init__": ["ModelImageVocabularyMapping", "__init__", "class", "def", "get", "image", "image_token_id", "self", "vocab_map"], "chameleon/modeling_chameleon.py:ChameleonImageVocabularyMapping.val2name": ["ModelImageVocabularyMapping", "class", "def", "for", "in", "items", "k", "return", "self", "v", "val2name", "vocab_map"], "chameleon/modeling_chameleon.py:ChameleonImageVocabularyMapping.image_tokens": ["IMGIMG", "ModelImageVocabularyMapping", "class", "def", "for", "if", "image_tokens", "in", "items", "name", "return", "self", "sorted", "startswith", "val", "vocab_map"], "chameleon/modeling_chameleon.py:ChameleonImageVocabularyMapping.bpe2img": ["A", "IMGIMG", "ModelImageVocabularyMapping", "bpe2img", "c", "chr", "class", "def", "for", "get", "i", "image_tokens", "img_tkn_chr_mapping", "in", "int", "join", "len", "old_name", "ord", "range", "remap", "return", "self", "str", "tok", "val2name"], "chameleon/modeling_chameleon.py:ChameleonImageVocabularyMapping.img2bpe": ["ModelImageVocabularyMapping", "bpe2img", "class", "def", "for", "img2bpe", "in", "items", "k", "return", "self", "v"], "chameleon/modeling_chameleon.py:ChameleonImageVocabularyMapping.bpe2img_search_tensors": ["ModelImageVocabularyMapping", "bpe2img", "bpe2img_search_tensors", "class", "def", "keys", "return", "self", "sorted", "tensor", "torch", "values"], "chameleon/modeling_chameleon.py:ChameleonImageVocabularyMapping.img2bpe_mapping_tensor": ["ModelImageVocabularyMapping", "class", "def", "dtype", "for", "img2bpe", "img2bpe_mapping_tensor", "in", "int", "items", "k", "keys", "mapping", "max", "return", "self", "torch", "v", "zeros"], "chameleon/modeling_chameleon.py:ChameleonImageVocabularyMapping.convert_img2bpe": ["ModelImageVocabularyMapping", "Tensor", "class", "convert_img2bpe", "cpu", "def", "device", "img2bpe_mapping_tensor", "img_batch", "img_tokens", "return", "self", "to", "torch"], "chameleon/modeling_chameleon.py:ChameleonVQVAE.__init__": ["Conv2d", "ModelPreTrainedModel", "ModelVQVAE", "ModelVQVAEConfig", "ModelVQVAEEncoder", "ModelVQVAEVectorQuantizer", "__init__", "class", "config", "def", "embed_dim", "encoder", "eval", "latent_channels", "nn", "post_init", "post_quant_conv", "quant_conv", "quantize", "self", "super", "torch"], "chameleon/modeling_chameleon.py:ChameleonVQVAE.encode": ["LongTensor", "ModelPreTrainedModel", "ModelVQVAE", "class", "def", "emb_loss", "encode", "encoder", "hidden_states", "indices", "pixel_values", "quant", "quant_conv", "quantize", "return", "self", "torch"], "chameleon/modeling_chameleon.py:ChameleonModel.__init__": ["Embedding", "False", "ModelConfig", "ModelDecoderLayer", "ModelImageVocabularyMapping", "ModelModel", "ModelPreTrainedModel", "ModelRMSNorm", "ModelRotaryEmbedding", "ModelSwinDecoderLayer", "ModelVQVAE", "ModuleList", "__init__", "_from_config", "class", "config", "decoder_layer", "def", "else", "embed_tokens", "eps", "for", "gradient_checkpointing", "hidden_size", "if", "in", "layer_idx", "layers", "nn", "norm", "not", "num_hidden_layers", "pad_token_id", "padding_idx", "post_init", "range", "rms_norm_eps", "rotary_emb", "self", "super", "swin_norm", "vocab_size", "vocabulary_map", "vocabulary_mapping", "vq_config", "vqmodel"], "chameleon/modeling_chameleon.py:ChameleonModel.get_image_tokens": ["FloatTensor", "ModelModel", "ModelPreTrainedModel", "_", "batch_size", "bpe_toks", "class", "convert_img2bpe", "def", "encode", "get_image_tokens", "image_toks", "pixel_values", "return", "self", "shape", "torch", "view", "vocabulary_mapping", "vqmodel"], "chameleon/modeling_chameleon.py:ChameleonModel.get_image_features": ["FloatTensor", "ModelModel", "ModelPreTrainedModel", "class", "def", "get_image_features", "get_image_tokens", "get_input_embeddings", "image_tokens", "pixel_values", "return", "self", "torch", "vision_embeddings"], "chameleon/modeling_chameleon.py:ChameleonModel.get_placeholder_mask": ["FloatTensor", "Image", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "ValueError", "all", "and", "class", "def", "device", "do", "dtype", "else", "expand_as", "f", "features", "get_input_embeddings", "get_placeholder_mask", "if", "image", "image_features", "image_token_id", "input_ids", "inputs_embeds", "is", "long", "match", "n_image_features", "n_image_tokens", "not", "numel", "raise", "return", "self", "shape", "special_image_mask", "sum", "tensor", "to", "tokens", "torch", "unsqueeze", "vocabulary_mapping"], "chameleon/modeling_chameleon.py:ChameleonModel.forward": ["BaseModelOutputWithPast", "Cache", "DynamicCache", "False", "FlashAttentionKwargs", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Setting", "Tensor", "True", "Unpack", "ValueError", "You", "all_hidden_states", "all_self_attns", "and", "arange", "attention_mask", "attentions", "bool", "cache_position", "causal_mask", "checkpointing", "class", "config", "create_causal_mask", "decoder_layer", "def", "device", "else", "embed_tokens", "exactly", "for", "forward", "get_image_features", "get_placeholder_mask", "get_seq_length", "gradient", "gradient_checkpointing", "hidden_states", "if", "image_embeds", "image_features", "in", "incompatible", "input_embeds", "input_ids", "inputs_embeds", "is", "is_tracing", "jit", "kwargs", "last_hidden_state", "layer_outputs", "layers", "logger", "masked_scatter", "must", "norm", "not", "of", "one", "or", "output_attentions", "output_hidden_states", "past_key_values", "past_seen_tokens", "pixel_values", "position_embeddings", "position_ids", "raise", "return", "return_dict", "rotary_emb", "self", "shape", "special_image_mask", "specify", "torch", "training", "tuple", "unsqueeze", "use_cache", "use_return_dict", "v", "warning_once", "with"], "chameleon/modeling_chameleon.py:ChameleonForConditionalGeneration.__init__": ["False", "GenerationMixin", "Linear", "ModelForConditionalGeneration", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "vocab_size"], "chameleon/modeling_chameleon.py:ChameleonForConditionalGeneration.get_image_tokens": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "get_image_tokens", "model", "pixel_values", "return", "self"], "chameleon/modeling_chameleon.py:ChameleonForConditionalGeneration.get_image_features": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "get_image_features", "model", "pixel_values", "return", "self"], "chameleon/modeling_chameleon.py:ChameleonForConditionalGeneration.forward": ["Cache", "CausalLMOutputWithPast", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "dtype", "else", "finfo", "forward", "hidden_states", "if", "image_tokens", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "min", "model", "not", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "pixel_values", "position_ids", "r", "return", "return_dict", "self", "slice", "slice_indices", "torch", "tuple", "use_cache", "vocab_size", "vocabulary_mapping"], "chameleon/modeling_chameleon.py:ChameleonForConditionalGeneration.prepare_inputs_for_generation": ["False", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "True", "and", "attention_mask", "cache_position", "class", "def", "if", "input_ids", "inputs_embeds", "is_first_iteration", "kwargs", "model_inputs", "not", "past_key_values", "pixel_values", "position_ids", "prepare_inputs_for_generation", "return", "self", "super", "use_cache"], "chinese_clip/modeling_chinese_clip.py:contrastive_loss": ["Model_loss", "Tensor", "arange", "cross_entropy", "def", "device", "functional", "len", "logits", "nn", "return", "torch"], "chinese_clip/modeling_chinese_clip.py:chinese_clip_loss": ["Model_loss", "Tensor", "caption_loss", "contrastive_loss", "def", "image_loss", "return", "similarity", "t", "torch"], "chinese_clip/modeling_chinese_clip.py:ChineseCLIPOutput.to_tuple": ["Any", "ModelOutput", "class", "def", "else", "for", "getattr", "if", "in", "k", "keys", "not", "return", "self", "text_model_output", "to_tuple", "tuple", "vision_model_output"], "chinese_clip/modeling_chinese_clip.py:ChineseCLIPTextEmbeddings.__init__": ["Dropout", "Embedding", "False", "LayerNorm", "ModelTextEmbeddings", "Module", "__init__", "arange", "class", "config", "def", "dropout", "dtype", "eps", "expand", "hidden_dropout_prob", "hidden_size", "layer_norm_eps", "long", "max_position_embeddings", "nn", "pad_token_id", "padding_idx", "persistent", "position_embeddings", "position_ids", "register_buffer", "self", "size", "super", "token_type_embeddings", "token_type_ids", "torch", "type_vocab_size", "vocab_size", "word_embeddings", "zeros"], "chinese_clip/modeling_chinese_clip.py:ChineseCLIPTextEmbeddings.forward": ["FloatTensor", "LayerNorm", "LongTensor", "ModelTextEmbeddings", "Module", "None", "Tensor", "buffered_token_type_ids", "buffered_token_type_ids_expanded", "class", "def", "device", "dropout", "dtype", "else", "embeddings", "expand", "forward", "hasattr", "if", "input_ids", "input_shape", "inputs_embeds", "is", "long", "nn", "not", "position_embeddings", "position_ids", "return", "self", "seq_length", "size", "token_type_embeddings", "token_type_ids", "torch", "word_embeddings", "zeros"], "chinese_clip/modeling_chinese_clip.py:ChineseCLIPVisionEmbeddings.__init__": ["Conv2d", "Embedding", "False", "ModelVisionConfig", "ModelVisionEmbeddings", "Module", "Parameter", "__init__", "arange", "bias", "class", "class_embedding", "config", "def", "embed_dim", "expand", "hidden_size", "image_size", "in_channels", "kernel_size", "nn", "num_channels", "num_patches", "num_positions", "out_channels", "patch_embedding", "patch_size", "persistent", "position_embedding", "position_ids", "randn", "register_buffer", "self", "stride", "super", "torch"], "chinese_clip/modeling_chinese_clip.py:ChineseCLIPVisionEmbeddings.interpolate_pos_encoding": ["False", "ModelVisionEmbeddings", "Module", "Tensor", "align_corners", "and", "bicubic", "cat", "class", "class_pos_embed", "def", "dim", "embeddings", "functional", "height", "if", "int", "interpolate", "interpolate_pos_encoding", "is_tracing", "jit", "mode", "new_height", "new_width", "nn", "not", "num_patches", "num_positions", "patch_pos_embed", "patch_size", "permute", "position_embedding", "position_ids", "reshape", "return", "self", "shape", "size", "sqrt_num_positions", "torch", "torch_int", "unsqueeze", "view", "weight", "width"], "chinese_clip/modeling_chinese_clip.py:ChineseCLIPVisionEmbeddings.forward": ["False", "FloatTensor", "Input", "ModelVisionEmbeddings", "Module", "Tensor", "ValueError", "_", "and", "batch_size", "cat", "class", "class_embedding", "class_embeds", "def", "dim", "doesn", "dtype", "else", "embeddings", "expand", "f", "flatten", "forward", "height", "if", "image", "image_size", "interpolate_pos_encoding", "match", "model", "nn", "not", "or", "patch_embedding", "patch_embeds", "pixel_values", "position_embedding", "position_ids", "raise", "return", "self", "shape", "size", "t", "target_dtype", "to", "torch", "transpose", "weight", "width"], "chinese_clip/modeling_chinese_clip.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value"], "chinese_clip/modeling_chinese_clip.py:ChineseCLIPTextSelfAttention.__init__": ["Dropout", "Linear", "ModelTextSelfAttention", "Module", "The", "ValueError", "__init__", "a", "all_head_size", "and", "attention", "attention_dropout", "attention_head_size", "attention_probs_dropout_prob", "class", "config", "def", "dropout", "embedding_size", "f", "hasattr", "heads", "hidden", "hidden_size", "if", "int", "is", "key", "multiple", "nn", "not", "num_attention_heads", "number", "of", "query", "raise", "scaling", "self", "size", "super", "the", "value"], "chinese_clip/modeling_chinese_clip.py:ChineseCLIPTextSelfAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "False", "FloatTensor", "ModelTextSelfAttention", "Module", "None", "Tensor", "_attn_implementation", "attention_dropout", "attention_head_size", "attention_interface", "attention_mask", "attn_output", "attn_weights", "bool", "class", "config", "contiguous", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "hidden_shape", "hidden_states", "if", "input_shape", "key", "key_states", "kwargs", "nn", "not", "output_attentions", "outputs", "query", "query_states", "reshape", "return", "scaling", "self", "shape", "torch", "training", "transpose", "tuple", "value", "value_states", "view"], "chinese_clip/modeling_chinese_clip.py:ChineseCLIPTextSelfOutput.__init__": ["Dropout", "LayerNorm", "Linear", "ModelTextSelfOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "eps", "hidden_dropout_prob", "hidden_size", "layer_norm_eps", "nn", "self", "super"], "chinese_clip/modeling_chinese_clip.py:ChineseCLIPTextSelfOutput.forward": ["LayerNorm", "ModelTextSelfOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "chinese_clip/modeling_chinese_clip.py:ChineseCLIPTextAttention.__init__": ["ModelTextAttention", "ModelTextSelfAttention", "ModelTextSelfOutput", "Module", "__init__", "class", "config", "def", "nn", "output", "self", "super"], "chinese_clip/modeling_chinese_clip.py:ChineseCLIPTextAttention.forward": ["False", "FloatTensor", "ModelTextAttention", "Module", "None", "Tensor", "attention_mask", "attention_output", "bool", "class", "def", "forward", "hidden_states", "kwargs", "nn", "output", "output_attentions", "outputs", "return", "self", "self_outputs", "torch", "tuple"], "chinese_clip/modeling_chinese_clip.py:ChineseCLIPVisionAttention.__init__": ["Linear", "ModelVisionAttention", "Module", "ValueError", "__init__", "and", "attention_dropout", "be", "by", "class", "config", "def", "divisible", "dropout", "embed_dim", "f", "got", "head_dim", "hidden_size", "if", "k_proj", "must", "nn", "num_attention_heads", "num_heads", "out_proj", "q_proj", "raise", "scale", "self", "super", "v_proj"], "chinese_clip/modeling_chinese_clip.py:ChineseCLIPVisionAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "False", "ModelVisionAttention", "Module", "None", "Tensor", "_attn_implementation", "attention_interface", "attn_output", "attn_weights", "bool", "class", "config", "contiguous", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "head_dim", "hidden_shape", "hidden_states", "if", "input_shape", "k_proj", "key_states", "kwargs", "nn", "not", "out_proj", "output_attentions", "q_proj", "query_states", "reshape", "return", "scale", "scaling", "self", "shape", "torch", "training", "transpose", "tuple", "v_proj", "value_states", "view"], "chinese_clip/modeling_chinese_clip.py:ChineseCLIPTextIntermediate.__init__": ["ACT2FN", "Linear", "ModelTextIntermediate", "Module", "__init__", "class", "config", "def", "dense", "else", "hidden_act", "hidden_size", "if", "intermediate_act_fn", "intermediate_size", "isinstance", "nn", "self", "str", "super"], "chinese_clip/modeling_chinese_clip.py:ChineseCLIPTextIntermediate.forward": ["ModelTextIntermediate", "Module", "Tensor", "class", "def", "dense", "forward", "hidden_states", "intermediate_act_fn", "nn", "return", "self", "torch"], "chinese_clip/modeling_chinese_clip.py:ChineseCLIPTextOutput.__init__": ["Dropout", "LayerNorm", "Linear", "ModelTextOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "eps", "hidden_dropout_prob", "hidden_size", "intermediate_size", "layer_norm_eps", "nn", "self", "super"], "chinese_clip/modeling_chinese_clip.py:ChineseCLIPTextOutput.forward": ["LayerNorm", "ModelTextOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "chinese_clip/modeling_chinese_clip.py:ChineseCLIPVisionMLP.__init__": ["ACT2FN", "Linear", "ModelVisionMLP", "Module", "__init__", "activation_fn", "class", "config", "def", "fc1", "fc2", "hidden_act", "hidden_size", "intermediate_size", "nn", "self", "super"], "chinese_clip/modeling_chinese_clip.py:ChineseCLIPVisionMLP.forward": ["ModelVisionMLP", "Module", "Tensor", "activation_fn", "class", "def", "fc1", "fc2", "forward", "hidden_states", "nn", "return", "self", "torch"], "chinese_clip/modeling_chinese_clip.py:ChineseCLIPTextLayer.__init__": ["GradientCheckpointingLayer", "ModelTextAttention", "ModelTextIntermediate", "ModelTextLayer", "ModelTextOutput", "__init__", "attention", "chunk_size_feed_forward", "class", "config", "def", "intermediate", "output", "self", "seq_len_dim", "super"], "chinese_clip/modeling_chinese_clip.py:ChineseCLIPTextLayer.forward": ["False", "FloatTensor", "GradientCheckpointingLayer", "ModelTextLayer", "None", "Tensor", "apply_chunking_to_forward", "attention", "attention_mask", "attention_output", "bool", "chunk_size_feed_forward", "class", "def", "feed_forward_chunk", "forward", "hidden_states", "kwargs", "layer_output", "output_attentions", "outputs", "return", "self", "self_attention_outputs", "seq_len_dim", "torch", "tuple"], "chinese_clip/modeling_chinese_clip.py:ChineseCLIPTextLayer.feed_forward_chunk": ["GradientCheckpointingLayer", "ModelTextLayer", "attention_output", "class", "def", "feed_forward_chunk", "intermediate", "intermediate_output", "layer_output", "output", "return", "self"], "chinese_clip/modeling_chinese_clip.py:ChineseCLIPVisionLayer.__init__": ["GradientCheckpointingLayer", "LayerNorm", "ModelConfig", "ModelVisionAttention", "ModelVisionLayer", "ModelVisionMLP", "__init__", "class", "config", "def", "embed_dim", "eps", "hidden_size", "layer_norm1", "layer_norm2", "layer_norm_eps", "mlp", "nn", "self", "self_attn", "super"], "chinese_clip/modeling_chinese_clip.py:ChineseCLIPVisionLayer.forward": ["False", "FloatTensor", "GradientCheckpointingLayer", "ModelVisionLayer", "None", "Tensor", "attn_weights", "bool", "class", "def", "forward", "hidden_states", "if", "layer_norm1", "layer_norm2", "mlp", "output_attentions", "outputs", "residual", "return", "self", "self_attn", "torch", "tuple"], "chinese_clip/modeling_chinese_clip.py:ChineseCLIPTextPooler.__init__": ["Linear", "ModelTextPooler", "Module", "Tanh", "__init__", "activation", "class", "config", "def", "dense", "hidden_size", "nn", "self", "super"], "chinese_clip/modeling_chinese_clip.py:ChineseCLIPTextPooler.forward": ["ModelTextPooler", "Module", "Tensor", "activation", "class", "def", "dense", "first_token_tensor", "forward", "hidden_states", "nn", "pooled_output", "return", "self", "torch"], "chinese_clip/modeling_chinese_clip.py:ChineseCLIPPreTrainedModel._init_weights": ["LayerNorm", "Linear", "ModelModel", "ModelPreTrainedModel", "ModelTextEmbeddings", "ModelVisionAttention", "ModelVisionEmbeddings", "ModelVisionMLP", "None", "PreTrainedModel", "_init_weights", "arange", "bias", "class", "class_embedding", "config", "copy_", "def", "elif", "embed_dim", "embedding", "expand", "factor", "fc1", "fc2", "fc_std", "for", "hidden_size", "if", "in", "in_proj_std", "init", "initializer_factor", "initializer_range", "is", "isinstance", "k_proj", "mean", "module", "nn", "normal_", "not", "num_hidden_layers", "num_positions", "ones_", "out_proj", "out_proj_std", "padding_idx", "patch_embedding", "position_embedding", "position_embeddings", "position_ids", "q_proj", "self", "shape", "std", "text_embed_dim", "text_projection", "token_type_embeddings", "token_type_ids", "torch", "v_proj", "vision_embed_dim", "visual_projection", "weight", "word_embeddings", "zeros_"], "chinese_clip/modeling_chinese_clip.py:ChineseCLIPTextEncoder.__init__": ["False", "ModelTextEncoder", "ModelTextLayer", "Module", "ModuleList", "__init__", "class", "config", "def", "for", "gradient_checkpointing", "i", "in", "layer", "nn", "num_hidden_layers", "range", "self", "super"], "chinese_clip/modeling_chinese_clip.py:ChineseCLIPTextEncoder.forward": ["BaseModelOutput", "False", "FloatTensor", "ModelTextEncoder", "Module", "None", "Tensor", "True", "all_hidden_states", "all_self_attentions", "attention_mask", "attentions", "bool", "class", "def", "else", "enumerate", "for", "forward", "hidden_states", "i", "if", "in", "kwargs", "last_hidden_state", "layer", "layer_module", "layer_outputs", "nn", "output_attentions", "output_hidden_states", "return", "return_dict", "self", "torch", "tuple"], "chinese_clip/modeling_chinese_clip.py:ChineseCLIPVisionEncoder.__init__": ["False", "ModelConfig", "ModelVisionEncoder", "ModelVisionLayer", "Module", "ModuleList", "_", "__init__", "class", "config", "def", "for", "gradient_checkpointing", "in", "layers", "nn", "num_hidden_layers", "range", "self", "super"], "chinese_clip/modeling_chinese_clip.py:ChineseCLIPVisionEncoder.forward": ["BaseModelOutput", "ModelVisionEncoder", "Module", "None", "all_attentions", "attentions", "bool", "class", "config", "def", "else", "encoder_layer", "encoder_states", "enumerate", "for", "forward", "hidden_states", "idx", "if", "in", "inputs_embeds", "is", "last_hidden_state", "layer_outputs", "layers", "nn", "not", "output_attentions", "output_hidden_states", "r", "return", "return_dict", "self", "tuple", "use_return_dict"], "chinese_clip/modeling_chinese_clip.py:ChineseCLIPVisionTransformer.__init__": ["LayerNorm", "ModelVisionConfig", "ModelVisionEmbeddings", "ModelVisionEncoder", "ModelVisionTransformer", "Module", "__init__", "class", "config", "def", "embed_dim", "embeddings", "encoder", "eps", "hidden_size", "layer_norm_eps", "nn", "post_layernorm", "pre_layrnorm", "self", "super"], "chinese_clip/modeling_chinese_clip.py:ChineseCLIPVisionTransformer.forward": ["BaseModelOutputWithPooling", "False", "FloatTensor", "ModelVisionTransformer", "Module", "None", "True", "ValueError", "You", "attentions", "bool", "class", "config", "def", "else", "embeddings", "encoder", "encoder_outputs", "forward", "have", "hidden_states", "if", "inputs_embeds", "interpolate_pos_encoding", "is", "last_hidden_state", "nn", "not", "output_attentions", "output_hidden_states", "pixel_values", "pooled_output", "pooler_output", "post_layernorm", "pre_layrnorm", "raise", "return", "return_dict", "self", "specify", "to", "torch", "tuple", "use_return_dict"], "chinese_clip/modeling_chinese_clip.py:ChineseCLIPTextModel.__init__": ["ModelPreTrainedModel", "ModelTextEmbeddings", "ModelTextEncoder", "ModelTextModel", "ModelTextPooler", "None", "True", "__init__", "add_pooling_layer", "class", "config", "def", "else", "embeddings", "encoder", "if", "pooler", "post_init", "r", "self", "super"], "chinese_clip/modeling_chinese_clip.py:ChineseCLIPTextModel.get_input_embeddings": ["ModelPreTrainedModel", "ModelTextModel", "class", "def", "embeddings", "get_input_embeddings", "return", "self", "word_embeddings"], "chinese_clip/modeling_chinese_clip.py:ChineseCLIPTextModel.set_input_embeddings": ["ModelPreTrainedModel", "ModelTextModel", "class", "def", "embeddings", "self", "set_input_embeddings", "value", "word_embeddings"], "chinese_clip/modeling_chinese_clip.py:ChineseCLIPTextModel.forward": ["BaseModelOutputWithPooling", "Cache", "ModelPreTrainedModel", "ModelTextModel", "None", "Tensor", "True", "ValueError", "You", "and", "at", "attention_mask", "attentions", "batch_size", "bool", "both", "buffered_token_type_ids", "buffered_token_type_ids_expanded", "cannot", "class", "config", "def", "device", "dtype", "either", "elif", "else", "embedding_output", "embeddings", "encoder", "encoder_attention_mask", "encoder_hidden_states", "encoder_outputs", "expand", "extended_attention_mask", "forward", "get_extended_attention_mask", "hasattr", "have", "hidden_states", "if", "input_ids", "input_shape", "inputs_embeds", "is", "kwargs", "last_hidden_state", "long", "not", "ones", "or", "output_attentions", "output_hidden_states", "past_key_values", "pooled_output", "pooler", "pooler_output", "position_ids", "raise", "return", "return_dict", "same", "self", "seq_length", "sequence_output", "size", "specify", "the", "time", "to", "token_type_ids", "torch", "tuple", "use_cache", "use_return_dict", "warn_if_padding_and_no_attention_mask", "zeros"], "chinese_clip/modeling_chinese_clip.py:ChineseCLIPVisionModel.__init__": ["ModelPreTrainedModel", "ModelVisionConfig", "ModelVisionModel", "ModelVisionTransformer", "__init__", "class", "config", "def", "post_init", "self", "super", "vision_model"], "chinese_clip/modeling_chinese_clip.py:ChineseCLIPVisionModel.get_input_embeddings": ["ModelPreTrainedModel", "ModelVisionModel", "Module", "class", "def", "embeddings", "get_input_embeddings", "nn", "patch_embedding", "return", "self", "vision_model"], "chinese_clip/modeling_chinese_clip.py:ChineseCLIPVisionModel.forward": ["BaseModelOutputWithPooling", "False", "FloatTensor", "ModelPreTrainedModel", "ModelVisionModel", "None", "bool", "class", "config", "def", "else", "forward", "if", "interpolate_pos_encoding", "is", "kwargs", "not", "output_attentions", "output_hidden_states", "pixel_values", "r", "return", "return_dict", "self", "torch", "tuple", "use_return_dict", "vision_model"], "chinese_clip/modeling_chinese_clip.py:ChineseCLIPModel.__init__": ["False", "Linear", "ModelConfig", "ModelModel", "ModelPreTrainedModel", "ModelTextConfig", "ModelTextModel", "ModelVisionConfig", "ModelVisionTransformer", "Parameter", "TypeError", "__init__", "_attn_implementation", "add_pooling_layer", "be", "bias", "but", "class", "config", "def", "expected", "f", "hidden_size", "if", "is", "isinstance", "logit_scale", "logit_scale_init_value", "nn", "not", "of", "post_init", "projection_dim", "raise", "self", "super", "tensor", "text_config", "text_embed_dim", "text_model", "text_projection", "to", "torch", "type", "vision_config", "vision_embed_dim", "vision_model", "visual_projection"], "chinese_clip/modeling_chinese_clip.py:ChineseCLIPModel.get_text_features": ["FloatTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "attention_mask", "class", "def", "get_text_features", "input_ids", "pooled_output", "position_ids", "r", "return", "self", "text_features", "text_model", "text_outputs", "text_projection", "token_type_ids", "torch"], "chinese_clip/modeling_chinese_clip.py:ChineseCLIPModel.get_image_features": ["BaseModelOutputWithPooling", "False", "FloatTensor", "ModelModel", "ModelPreTrainedModel", "bool", "class", "def", "get_image_features", "image_features", "interpolate_pos_encoding", "pixel_values", "pooled_output", "pooler_output", "r", "return", "self", "torch", "vision_model", "vision_outputs", "visual_projection"], "chinese_clip/modeling_chinese_clip.py:ChineseCLIPModel.forward": ["False", "FloatTensor", "LongTensor", "ModelModel", "ModelOutput", "ModelPreTrainedModel", "Model_loss", "None", "Tensor", "True", "attention_mask", "bool", "class", "config", "def", "dim", "else", "exp", "forward", "if", "image_embeds", "input_ids", "interpolate_pos_encoding", "is", "keepdim", "kwargs", "logit_scale", "logits_per_image", "logits_per_text", "loss", "matmul", "norm", "not", "output_attentions", "output_hidden_states", "p", "pixel_values", "position_ids", "r", "return", "return_dict", "return_loss", "self", "t", "text_embeds", "text_model", "text_model_output", "text_outputs", "text_projection", "token_type_ids", "torch", "tuple", "vision_model", "vision_model_output", "vision_outputs", "visual_projection"], "clap/modeling_clap.py:interpolate": ["Model", "None", "batch_size", "classes_num", "def", "hidden_states", "ratio", "repeat", "reshape", "return", "shape", "time_length", "upsampled"], "clap/modeling_clap.py:window_partition": ["Model_partition", "Model_size", "Models", "batch_size", "contiguous", "def", "height", "hidden_states", "num_channels", "permute", "return", "shape", "view", "width"], "clap/modeling_clap.py:window_reverse": ["Model_reverse", "Model_size", "Models", "contiguous", "def", "height", "num_channels", "permute", "return", "shape", "view", "width"], "clap/modeling_clap.py:contrastive_loss": ["Model_loss", "Tensor", "arange", "cross_entropy", "def", "device", "functional", "labels", "len", "logits", "nn", "return", "torch"], "clap/modeling_clap.py:ClapOutput.to_tuple": ["Any", "ModelOutput", "audio_model_output", "class", "def", "else", "for", "getattr", "if", "in", "k", "keys", "not", "return", "self", "text_model_output", "to_tuple", "tuple"], "clap/modeling_clap.py:ClapDropPath.__init__": ["ModelDropPath", "Module", "None", "__init__", "class", "def", "drop_prob", "nn", "self", "super"], "clap/modeling_clap.py:ClapDropPath.forward": ["ModelDropPath", "Module", "class", "def", "device", "div", "drop_prob", "dtype", "floor_", "forward", "hidden_states", "if", "keep_prob", "ndim", "nn", "not", "or", "output", "rand", "random_tensor", "return", "self", "shape", "torch", "training"], "clap/modeling_clap.py:ClapAudioAFFBlock.__init__": ["AdaptiveAvgPool2d", "BatchNorm2d", "Conv2d", "ModelAudioAFFBlock", "ModelAudioConfig", "Module", "ReLU", "Sequential", "Sigmoid", "True", "__init__", "aff_block_r", "channels", "class", "config", "def", "downsize_ratio", "global_att", "inplace", "int", "inter_channels", "kernel_size", "local_att", "nn", "padding", "patch_embeds_hidden_size", "self", "sigmoid", "stride", "super"], "clap/modeling_clap.py:ClapAudioAFFBlock.forward": ["ModelAudioAFFBlock", "Module", "attention_input", "class", "def", "forward", "fused_layer_output", "global_att", "hidden_states", "local_att", "nn", "output", "residual", "return", "self", "sigmoid"], "clap/modeling_clap.py:ClapAudioPatchEmbed.__init__": ["Conv2d", "Identity", "LayerNorm", "ModelAudioAFFBlock", "ModelAudioConfig", "ModelAudioPatchEmbed", "Module", "__init__", "and", "channel_map", "class", "config", "def", "else", "enable_fusion", "enable_patch_layer_norm", "flatten", "flatten_patch_embeds", "fusion_model", "fusion_type", "grid_size", "if", "img_size", "int", "isinstance", "kernel_size", "mel_conv2d", "nn", "norm", "num_patches", "padding", "patch_embed_input_channels", "patch_embeds_hidden_size", "patch_size", "patch_stride", "proj", "scale_factor", "self", "spec_size", "stride", "super"], "clap/modeling_clap.py:ClapAudioPatchEmbed.forward": ["Input", "ModelAudioPatchEmbed", "Module", "None", "ValueError", "_", "audio", "batch_size", "class", "constant", "contiguous", "def", "doesn", "else", "enable_fusion", "f", "features", "flatten", "forward", "functional", "fusion_model", "global_hidden_states", "height", "hidden_states", "if", "img_size", "is_longer_idx", "len", "local_hidden_states", "local_width", "match", "mel_conv2d", "model", "nn", "norm", "num_channels", "or", "output_width", "pad", "permute", "proj", "raise", "return", "self", "shape", "size", "t", "torch", "transpose", "view", "width"], "clap/modeling_clap.py:ClapAudioSelfAttention.__init__": ["Dropout", "Iterable", "Linear", "ModelAudioSelfAttention", "Module", "Parameter", "The", "ValueError", "__init__", "a", "abc", "all_head_size", "attention", "attention_head_size", "attention_probs_dropout_prob", "bias", "class", "collections", "config", "create_relative_position_index", "def", "dim", "dropout", "else", "f", "heads", "hidden", "if", "int", "is", "isinstance", "key", "multiple", "nn", "not", "num_attention_heads", "num_heads", "number", "of", "qkv_bias", "query", "raise", "register_buffer", "relative_position_bias_table", "relative_position_index", "self", "size", "super", "the", "torch", "value", "window_size", "zeros"], "clap/modeling_clap.py:ClapAudioSelfAttention.forward": ["False", "FloatTensor", "ModelAudioSelfAttention", "Module", "None", "Tensor", "all_head_size", "attention_head_size", "attention_mask", "attention_probs", "attention_scores", "batch_size", "bool", "class", "context_layer", "contiguous", "def", "dim", "dropout", "else", "forward", "functional", "hidden_shape", "hidden_states", "if", "is", "key", "key_layer", "mask_shape", "math", "matmul", "new_context_layer_shape", "nn", "not", "num_attention_heads", "num_channels", "output_attentions", "outputs", "permute", "query", "query_layer", "relative_position_bias", "relative_position_bias_table", "relative_position_index", "return", "self", "shape", "size", "softmax", "sqrt", "torch", "transpose", "tuple", "unsqueeze", "value", "value_layer", "view", "window_size"], "clap/modeling_clap.py:ClapAudioSelfAttention.create_relative_position_index": ["ModelAudioSelfAttention", "Module", "None", "arange", "class", "contiguous", "coords", "coords_flatten", "coords_h", "coords_w", "create_relative_position_index", "def", "flatten", "ij", "indexing", "meshgrid", "nn", "permute", "relative_coords", "relative_position_index", "return", "self", "stack", "sum", "torch", "window_size"], "clap/modeling_clap.py:ClapAudioSelfOutput.__init__": ["Dropout", "Linear", "ModelAudioSelfOutput", "Module", "__init__", "attention_probs_dropout_prob", "class", "config", "def", "dense", "dim", "dropout", "nn", "self", "super"], "clap/modeling_clap.py:ClapAudioSelfOutput.forward": ["ModelAudioSelfOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "clap/modeling_clap.py:ClapAudioAttention.__init__": ["ModelAudioAttention", "ModelAudioSelfAttention", "ModelAudioSelfOutput", "Module", "__init__", "class", "config", "def", "dim", "nn", "num_heads", "output", "self", "super", "window_size"], "clap/modeling_clap.py:ClapAudioAttention.forward": ["False", "FloatTensor", "ModelAudioAttention", "Module", "None", "Tensor", "attention_mask", "attention_output", "bool", "class", "def", "forward", "hidden_states", "nn", "output", "output_attentions", "outputs", "return", "self", "self_outputs", "torch", "tuple"], "clap/modeling_clap.py:ClapAudioIntermediate.__init__": ["ACT2FN", "Linear", "ModelAudioIntermediate", "Module", "__init__", "class", "config", "def", "dense", "dim", "else", "hidden_act", "if", "int", "intermediate_act_fn", "isinstance", "mlp_ratio", "nn", "self", "str", "super"], "clap/modeling_clap.py:ClapAudioIntermediate.forward": ["ModelAudioIntermediate", "Module", "Tensor", "class", "def", "dense", "forward", "hidden_states", "intermediate_act_fn", "nn", "return", "self", "torch"], "clap/modeling_clap.py:ClapAudioOutput.__init__": ["Dropout", "Linear", "ModelAudioOutput", "Module", "__init__", "class", "config", "def", "dense", "dim", "dropout", "hidden_dropout_prob", "int", "mlp_ratio", "nn", "self", "super"], "clap/modeling_clap.py:ClapAudioOutput.forward": ["ModelAudioOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "nn", "return", "self", "torch"], "clap/modeling_clap.py:ClapAudioLayer.__init__": ["Identity", "LayerNorm", "ModelAudioAttention", "ModelAudioIntermediate", "ModelAudioLayer", "ModelAudioOutput", "ModelDropPath", "Module", "__init__", "attention", "chunk_size_feed_forward", "class", "config", "def", "dim", "drop_path", "drop_path_rate", "else", "eps", "if", "input_resolution", "intermediate", "layer_norm_eps", "layernorm_after", "layernorm_before", "nn", "num_heads", "output", "self", "shift_size", "super", "window_size"], "clap/modeling_clap.py:ClapAudioLayer.set_shift_and_window_size": ["ModelAudioLayer", "Module", "class", "def", "else", "if", "input_resolution", "is_tracing", "jit", "min", "nn", "self", "set_shift_and_window_size", "shift_size", "tensor", "torch", "torch_int", "window_size"], "clap/modeling_clap.py:ClapAudioLayer.get_attn_mask": ["ModelAudioLayer", "Module", "None", "attn_mask", "class", "count", "def", "device", "dtype", "else", "for", "get_attn_mask", "height", "height_slice", "height_slices", "if", "img_mask", "in", "mask_windows", "masked_fill", "nn", "return", "self", "shift_size", "slice", "torch", "unsqueeze", "view", "width", "width_slice", "width_slices", "window_partition", "window_size", "zeros"], "clap/modeling_clap.py:ClapAudioLayer.maybe_pad": ["ModelAudioLayer", "Module", "class", "def", "functional", "height", "hidden_states", "maybe_pad", "nn", "pad", "pad_bottom", "pad_right", "pad_values", "return", "self", "width", "window_size"], "clap/modeling_clap.py:ClapAudioLayer.forward": ["False", "ModelAudioLayer", "Module", "None", "Tensor", "_", "always_partition", "attention", "attention_output", "attention_outputs", "attention_windows", "attn_mask", "batch_size", "bool", "channels", "class", "contiguous", "def", "device", "dims", "drop_path", "dtype", "else", "forward", "get_attn_mask", "height", "height_pad", "hidden_states", "hidden_states_windows", "if", "input_dimensions", "int", "intermediate", "layer_output", "layer_outputs", "layernorm_after", "layernorm_before", "maybe_pad", "nn", "not", "or", "output", "output_attentions", "pad_values", "pass", "return", "roll", "self", "set_shift_and_window_size", "shape", "shift_size", "shifted_hidden_states", "shifted_windows", "shifts", "shortcut", "size", "torch", "tuple", "view", "was_padded", "width", "width_pad", "window_partition", "window_reverse", "window_size"], "clap/modeling_clap.py:ClapAudioStage.__init__": ["False", "GradientCheckpointingLayer", "LayerNorm", "ModelAudioLayer", "ModelAudioStage", "ModuleList", "None", "__init__", "blocks", "class", "config", "def", "depth", "dim", "downsample", "drop_path", "drop_path_rate", "else", "for", "i", "if", "in", "input_resolution", "is", "nn", "norm_layer", "not", "num_heads", "pointing", "range", "self", "shift_size", "super", "window_size"], "clap/modeling_clap.py:ClapAudioStage.forward": ["False", "GradientCheckpointingLayer", "ModelAudioStage", "None", "Tensor", "always_partition", "blocks", "bool", "class", "def", "downsample", "else", "enumerate", "for", "forward", "height", "height_downsampled", "hidden_states", "hidden_states_before_downsampling", "i", "if", "in", "input_dimensions", "int", "is", "layer_module", "layer_outputs", "not", "output_attentions", "output_dimensions", "return", "self", "stage_outputs", "torch", "tuple", "width", "width_downsampled"], "clap/modeling_clap.py:ClapAudioPatchMerging.__init__": ["False", "LayerNorm", "Linear", "ModelAudioPatchMerging", "Module", "None", "__init__", "bias", "class", "def", "dim", "input_resolution", "int", "nn", "norm", "norm_layer", "reduction", "self", "super", "tuple"], "clap/modeling_clap.py:ClapAudioPatchMerging.maybe_pad": ["ModelAudioPatchMerging", "Module", "class", "def", "functional", "height", "if", "input_feature", "maybe_pad", "nn", "or", "pad", "pad_values", "return", "self", "should_pad", "width"], "clap/modeling_clap.py:ClapAudioPatchMerging.forward": ["ModelAudioPatchMerging", "Module", "Tensor", "batch_size", "cat", "class", "def", "dim", "forward", "height", "input_dimensions", "input_feature", "input_feature_0", "input_feature_1", "input_feature_2", "input_feature_3", "int", "maybe_pad", "nn", "norm", "num_channels", "reduction", "return", "self", "shape", "torch", "tuple", "view", "width"], "clap/modeling_clap.py:ClapAudioEncoder.__init__": ["AdaptiveAvgPool1d", "BatchNorm2d", "False", "LayerNorm", "ModelAudioEncoder", "ModelAudioPatchEmbed", "ModelAudioPatchMerging", "ModelAudioStage", "Module", "ModuleList", "None", "__init__", "avgpool", "batch_norm", "class", "config", "cpu", "def", "depth", "depths", "device", "dim", "downsample", "drop_path", "drop_path_rate", "else", "enable_fusion", "for", "freq_ratio", "gradient_checkpointing", "grid_size", "i", "i_layer", "if", "in", "input_resolution", "input_resolutions", "int", "item", "layers", "len", "linspace", "nn", "norm", "num_attention_heads", "num_features", "num_heads", "num_layers", "num_mel_bins", "patch_embed", "patch_embeds_hidden_size", "patch_stride", "range", "self", "spec_size", "sum", "super", "torch", "x"], "clap/modeling_clap.py:ClapAudioEncoder.reshape_mel2img": ["ModelAudioEncoder", "Module", "True", "ValueError", "_", "align_corners", "batch", "be", "bicubic", "channels", "class", "contiguous", "def", "equal", "freq", "freq_length", "freq_ratio", "functional", "if", "input", "int", "interpolate", "less", "mode", "nn", "normalized_input_features", "or", "permute", "raise", "reshape", "reshape_mel2img", "return", "self", "shape", "should", "size", "spec_height", "spec_size", "spec_width", "swin", "than", "the", "time", "time_length", "to", "wav"], "clap/modeling_clap.py:ClapAudioEncoder.forward": ["BaseModelOutputWithPooling", "False", "FloatTensor", "ModelAudioEncoder", "ModelAudioModelOutput", "Module", "None", "True", "_", "all_hidden_states", "all_reshaped_hidden_states", "all_self_attentions", "always_partition", "and", "attentions", "avgpool", "batch_norm", "batch_size", "bool", "c_freq_bin", "class", "contiguous", "def", "depths", "device", "elif", "else", "enable_fusion", "enumerate", "flatten", "for", "forward", "frames_num", "freq_ratio", "freq_shape", "hidden_size", "hidden_states", "hidden_states_before_downsampling", "i", "if", "in", "input_dimensions", "input_features", "input_resolutions", "is", "is_longer", "is_longer_list", "is_longer_list_idx", "last_hidden_state", "latent_output", "layer_module", "layer_outputs", "layers", "len", "n_channels", "n_frequencies", "n_temp", "nn", "norm", "normalized_input_features", "not", "output_attentions", "output_dimensions", "output_hidden_states", "output_hidden_states_before_downsampling", "patch_embed", "patch_stride", "permute", "pooler_output", "reshape", "reshape_mel2img", "reshaped_hidden_state", "return", "return_dict", "self", "shape", "temporal_shape", "to", "torch", "transpose", "tuple", "v", "view", "where"], "clap/modeling_clap.py:ClapProjectionLayer.__init__": ["ACT2FN", "Linear", "ModelAudioConfig", "ModelProjectionLayer", "ModelTextConfig", "Module", "__init__", "activation", "class", "config", "def", "hidden_size", "linear1", "linear2", "nn", "projection_dim", "projection_hidden_act", "self", "super"], "clap/modeling_clap.py:ClapProjectionLayer.forward": ["ModelProjectionLayer", "Module", "activation", "class", "def", "forward", "hidden_states", "linear1", "linear2", "nn", "return", "self"], "clap/modeling_clap.py:ClapTextEmbeddings.__init__": ["Dropout", "Embedding", "LayerNorm", "ModelTextEmbeddings", "Module", "True", "__init__", "arange", "class", "config", "def", "dropout", "dtype", "eps", "expand", "hidden_dropout_prob", "hidden_size", "layer_norm_eps", "long", "max_position_embeddings", "nn", "pad_token_id", "padding_idx", "persistent", "position_embeddings", "position_ids", "register_buffer", "self", "size", "super", "token_type_embeddings", "token_type_ids", "torch", "type_vocab_size", "vocab_size", "word_embeddings", "zeros"], "clap/modeling_clap.py:ClapTextEmbeddings.forward": ["FloatTensor", "LayerNorm", "LongTensor", "ModelTextEmbeddings", "Module", "None", "Tensor", "batch_size", "buffered_token_type_ids", "class", "create_position_ids_from_input_ids", "create_position_ids_from_inputs_embeds", "def", "device", "dim", "dropout", "dtype", "else", "embeddings", "expand", "forward", "gather", "hasattr", "if", "index", "input_ids", "input_shape", "inputs_embeds", "int", "is", "long", "nn", "not", "padding_idx", "past_key_values_length", "position_embeddings", "position_ids", "return", "self", "seq_length", "shape", "size", "token_type_embeddings", "token_type_ids", "torch", "word_embeddings", "zeros"], "clap/modeling_clap.py:ClapTextEmbeddings.create_position_ids_from_inputs_embeds": ["ModelTextEmbeddings", "Module", "arange", "class", "create_position_ids_from_inputs_embeds", "def", "device", "dtype", "expand", "input_shape", "inputs_embeds", "long", "nn", "padding_idx", "position_ids", "return", "sequence_length", "size", "torch", "unsqueeze"], "clap/modeling_clap.py:ClapTextEmbeddings.create_position_ids_from_input_ids": ["ModelTextEmbeddings", "Module", "class", "create_position_ids_from_input_ids", "cumsum", "def", "dim", "incremental_indices", "input_ids", "int", "long", "mask", "ne", "nn", "padding_idx", "past_key_values_length", "return", "torch", "type_as"], "clap/modeling_clap.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value"], "clap/modeling_clap.py:ClapTextSelfAttention.__init__": ["Dropout", "Linear", "ModelTextSelfAttention", "Module", "The", "ValueError", "__init__", "a", "all_head_size", "and", "attention", "attention_dropout", "attention_head_size", "attention_probs_dropout_prob", "class", "config", "def", "dropout", "embedding_size", "f", "hasattr", "heads", "hidden", "hidden_size", "if", "int", "is", "key", "multiple", "nn", "not", "num_attention_heads", "number", "of", "query", "raise", "scaling", "self", "size", "super", "the", "value"], "clap/modeling_clap.py:ClapTextSelfAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "False", "FloatTensor", "ModelTextSelfAttention", "Module", "None", "Tensor", "_attn_implementation", "attention_dropout", "attention_head_size", "attention_interface", "attention_mask", "attn_output", "attn_weights", "bool", "class", "config", "contiguous", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "hidden_shape", "hidden_states", "if", "input_shape", "key", "key_states", "kwargs", "nn", "not", "output_attentions", "outputs", "query", "query_states", "reshape", "return", "scaling", "self", "shape", "torch", "training", "transpose", "tuple", "value", "value_states", "view"], "clap/modeling_clap.py:ClapTextSelfOutput.__init__": ["Dropout", "LayerNorm", "Linear", "ModelTextSelfOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "eps", "hidden_dropout_prob", "hidden_size", "layer_norm_eps", "nn", "self", "super"], "clap/modeling_clap.py:ClapTextSelfOutput.forward": ["LayerNorm", "ModelTextSelfOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "clap/modeling_clap.py:ClapTextAttention.__init__": ["ModelTextAttention", "ModelTextSelfAttention", "ModelTextSelfOutput", "Module", "__init__", "class", "config", "def", "nn", "output", "self", "super"], "clap/modeling_clap.py:ClapTextAttention.forward": ["False", "FloatTensor", "ModelTextAttention", "Module", "None", "Tensor", "attention_mask", "attention_output", "bool", "class", "def", "forward", "hidden_states", "kwargs", "nn", "output", "output_attentions", "outputs", "return", "self", "self_outputs", "torch", "tuple"], "clap/modeling_clap.py:ClapTextIntermediate.__init__": ["ACT2FN", "Linear", "ModelTextIntermediate", "Module", "__init__", "class", "config", "def", "dense", "else", "hidden_act", "hidden_size", "if", "intermediate_act_fn", "intermediate_size", "isinstance", "nn", "self", "str", "super"], "clap/modeling_clap.py:ClapTextIntermediate.forward": ["ModelTextIntermediate", "Module", "Tensor", "class", "def", "dense", "forward", "hidden_states", "intermediate_act_fn", "nn", "return", "self", "torch"], "clap/modeling_clap.py:ClapTextOutput.__init__": ["Dropout", "LayerNorm", "Linear", "ModelTextOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "eps", "hidden_dropout_prob", "hidden_size", "intermediate_size", "layer_norm_eps", "nn", "self", "super"], "clap/modeling_clap.py:ClapTextOutput.forward": ["LayerNorm", "ModelTextOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "clap/modeling_clap.py:ClapTextLayer.__init__": ["GradientCheckpointingLayer", "ModelTextAttention", "ModelTextIntermediate", "ModelTextLayer", "ModelTextOutput", "__init__", "attention", "chunk_size_feed_forward", "class", "config", "def", "intermediate", "output", "self", "seq_len_dim", "super"], "clap/modeling_clap.py:ClapTextLayer.forward": ["False", "FloatTensor", "GradientCheckpointingLayer", "ModelTextLayer", "None", "Tensor", "apply_chunking_to_forward", "attention", "attention_mask", "attention_output", "bool", "chunk_size_feed_forward", "class", "def", "feed_forward_chunk", "forward", "hidden_states", "kwargs", "layer_output", "output_attentions", "outputs", "return", "self", "self_attention_outputs", "seq_len_dim", "torch", "tuple"], "clap/modeling_clap.py:ClapTextLayer.feed_forward_chunk": ["GradientCheckpointingLayer", "ModelTextLayer", "attention_output", "class", "def", "feed_forward_chunk", "intermediate", "intermediate_output", "layer_output", "output", "return", "self"], "clap/modeling_clap.py:ClapTextEncoder.__init__": ["False", "ModelTextEncoder", "ModelTextLayer", "Module", "ModuleList", "__init__", "class", "config", "def", "for", "gradient_checkpointing", "i", "in", "layer", "nn", "num_hidden_layers", "range", "self", "super"], "clap/modeling_clap.py:ClapTextEncoder.forward": ["BaseModelOutput", "False", "FloatTensor", "ModelTextEncoder", "Module", "None", "Tensor", "True", "all_hidden_states", "all_self_attentions", "attention_mask", "attentions", "bool", "class", "def", "else", "enumerate", "for", "forward", "hidden_states", "i", "if", "in", "kwargs", "last_hidden_state", "layer", "layer_module", "layer_outputs", "nn", "output_attentions", "output_hidden_states", "return", "return_dict", "self", "torch", "tuple"], "clap/modeling_clap.py:ClapTextPooler.__init__": ["Linear", "ModelTextPooler", "Module", "Tanh", "__init__", "activation", "class", "config", "def", "dense", "hidden_size", "nn", "self", "super"], "clap/modeling_clap.py:ClapTextPooler.forward": ["ModelTextPooler", "Module", "Tensor", "activation", "class", "def", "dense", "first_token_tensor", "forward", "hidden_states", "nn", "pooled_output", "return", "self", "torch"], "clap/modeling_clap.py:ClapPreTrainedModel._init_weights": ["BatchNorm2d", "Conv2d", "Embedding", "LayerNorm", "Linear", "ModelAudioSelfAttention", "ModelModel", "ModelPreTrainedModel", "ModelTextEmbeddings", "Module", "None", "PreTrainedModel", "_init_weights", "arange", "bias", "class", "config", "constant_", "copy_", "create_relative_position_index", "def", "elif", "expand", "factor", "getattr", "hidden_size", "if", "in_proj_std", "init", "initializer_factor", "is", "isinstance", "log", "logit_scale_a", "logit_scale_init_value", "logit_scale_t", "math", "mean", "module", "nn", "normal_", "not", "num_batches_tracked", "num_hidden_layers", "ones_", "position_embeddings", "position_ids", "relative_position_bias_table", "relative_position_index", "running_mean", "running_var", "self", "shape", "std", "token_type_embeddings", "token_type_ids", "torch", "weight", "zeros_"], "clap/modeling_clap.py:ClapAudioModel.__init__": ["ModelAudioConfig", "ModelAudioEncoder", "ModelAudioModel", "ModelPreTrainedModel", "__init__", "audio_encoder", "class", "config", "def", "post_init", "self", "super"], "clap/modeling_clap.py:ClapAudioModel.get_input_embeddings": ["ModelAudioModel", "ModelPreTrainedModel", "Module", "audio_encoder", "class", "def", "get_input_embeddings", "nn", "patch_embed", "proj", "return", "self"], "clap/modeling_clap.py:ClapAudioModel.forward": ["BaseModelOutputWithPooling", "BoolTensor", "FloatTensor", "ModelAudioModel", "ModelPreTrainedModel", "None", "audio_encoder", "bool", "class", "config", "def", "else", "forward", "if", "input_features", "is", "is_longer", "kwargs", "not", "output_attentions", "output_hidden_states", "r", "return", "return_dict", "self", "torch", "tuple", "use_return_dict"], "clap/modeling_clap.py:ClapTextModel.__init__": ["ModelPreTrainedModel", "ModelTextEmbeddings", "ModelTextEncoder", "ModelTextModel", "ModelTextPooler", "None", "True", "__init__", "add_pooling_layer", "class", "config", "def", "else", "embeddings", "encoder", "if", "pooler", "post_init", "r", "self", "super"], "clap/modeling_clap.py:ClapTextModel.get_input_embeddings": ["ModelPreTrainedModel", "ModelTextModel", "class", "def", "embeddings", "get_input_embeddings", "return", "self", "word_embeddings"], "clap/modeling_clap.py:ClapTextModel.set_input_embeddings": ["ModelPreTrainedModel", "ModelTextModel", "class", "def", "embeddings", "self", "set_input_embeddings", "value", "word_embeddings"], "clap/modeling_clap.py:ClapTextModel.forward": ["BaseModelOutputWithPooling", "BaseModelOutputWithPoolingAndCrossAttentions", "ModelPreTrainedModel", "ModelTextModel", "None", "Tensor", "True", "ValueError", "You", "and", "at", "attention_mask", "attentions", "batch_size", "bool", "both", "buffered_token_type_ids", "buffered_token_type_ids_expanded", "cannot", "class", "config", "def", "device", "dtype", "either", "elif", "else", "embedding_output", "embeddings", "encoder", "encoder_outputs", "expand", "extended_attention_mask", "forward", "get_extended_attention_mask", "hasattr", "have", "hidden_states", "if", "input_ids", "input_shape", "inputs_embeds", "is", "kwargs", "last_hidden_state", "long", "not", "ones", "or", "output_attentions", "output_hidden_states", "pooled_output", "pooler", "pooler_output", "position_ids", "raise", "return", "return_dict", "same", "self", "seq_length", "sequence_output", "size", "specify", "the", "time", "to", "token_type_ids", "torch", "tuple", "use_return_dict", "warn_if_padding_and_no_attention_mask", "zeros"], "clap/modeling_clap.py:ClapModel.__init__": ["ModelAudioConfig", "ModelAudioModel", "ModelConfig", "ModelModel", "ModelPreTrainedModel", "ModelProjectionLayer", "ModelTextConfig", "ModelTextModel", "Parameter", "TypeError", "__init__", "audio_config", "audio_model", "audio_projection", "be", "but", "class", "config", "def", "expected", "f", "if", "is", "isinstance", "log", "logit_scale_a", "logit_scale_init_value", "logit_scale_t", "math", "nn", "not", "of", "post_init", "projection_dim", "raise", "self", "super", "tensor", "text_config", "text_model", "text_projection", "to", "torch", "type"], "clap/modeling_clap.py:ClapModel.get_text_features": ["BaseModelOutputWithPooling", "F", "FloatTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "attention_mask", "class", "def", "dim", "get_text_features", "input_ids", "normalize", "pooler_output", "position_ids", "r", "return", "self", "text_features", "text_model", "text_outputs", "text_projection", "torch"], "clap/modeling_clap.py:ClapModel.get_audio_features": ["BaseModelOutputWithPooling", "F", "FloatTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "attention_mask", "audio_features", "audio_model", "audio_outputs", "audio_projection", "class", "def", "dim", "get_audio_features", "input_features", "is_longer", "normalize", "pooler_output", "r", "return", "self", "torch"], "clap/modeling_clap.py:ClapModel.forward": ["BoolTensor", "FloatTensor", "LongTensor", "ModelModel", "ModelOutput", "ModelPreTrainedModel", "None", "Tensor", "True", "attention_mask", "audio_embeds", "audio_loss", "audio_model", "audio_model_output", "audio_outputs", "audio_projection", "bool", "caption_loss", "class", "config", "contrastive_loss", "def", "dim", "else", "exp", "forward", "if", "input_features", "input_ids", "is", "is_longer", "keepdim", "kwargs", "logit_scale_a", "logit_scale_audio", "logit_scale_t", "logit_scale_text", "logits_per_audio", "logits_per_text", "loss", "matmul", "norm", "not", "output_attentions", "output_hidden_states", "p", "pooler_output", "position_ids", "r", "return", "return_dict", "return_loss", "self", "t", "text_embeds", "text_model", "text_model_output", "text_outputs", "text_projection", "torch", "tuple", "use_return_dict"], "clap/modeling_clap.py:ClapTextModelWithProjection.__init__": ["ModelPreTrainedModel", "ModelProjectionLayer", "ModelTextConfig", "ModelTextModel", "ModelTextModelWithProjection", "__init__", "class", "config", "def", "post_init", "self", "super", "text_model", "text_projection"], "clap/modeling_clap.py:ClapTextModelWithProjection.get_input_embeddings": ["ModelPreTrainedModel", "ModelTextModelWithProjection", "Module", "class", "def", "embeddings", "get_input_embeddings", "nn", "return", "self", "text_model", "word_embeddings"], "clap/modeling_clap.py:ClapTextModelWithProjection.set_input_embeddings": ["ModelPreTrainedModel", "ModelTextModelWithProjection", "class", "def", "embeddings", "self", "set_input_embeddings", "text_model", "value", "word_embeddings"], "clap/modeling_clap.py:ClapTextModelWithProjection.forward": ["ModelPreTrainedModel", "ModelTextModelOutput", "ModelTextModelWithProjection", "None", "Tensor", "True", "attention_mask", "attentions", "bool", "class", "config", "def", "else", "forward", "hidden_states", "if", "input_ids", "is", "kwargs", "last_hidden_state", "not", "output_attentions", "output_hidden_states", "pooled_output", "pooler_output", "position_ids", "r", "return", "return_dict", "self", "text_embeds", "text_model", "text_outputs", "text_projection", "torch", "tuple", "use_return_dict"], "clap/modeling_clap.py:ClapAudioModelWithProjection.__init__": ["ModelAudioConfig", "ModelAudioModel", "ModelAudioModelWithProjection", "ModelPreTrainedModel", "ModelProjectionLayer", "__init__", "audio_model", "audio_projection", "class", "config", "def", "post_init", "self", "super"], "clap/modeling_clap.py:ClapAudioModelWithProjection.get_input_embeddings": ["ModelAudioModelWithProjection", "ModelPreTrainedModel", "Module", "audio_encoder", "audio_model", "class", "def", "get_input_embeddings", "nn", "patch_embed", "proj", "return", "self"], "clap/modeling_clap.py:ClapAudioModelWithProjection.forward": ["BoolTensor", "FloatTensor", "ModelAudioModelOutput", "ModelAudioModelWithProjection", "ModelPreTrainedModel", "None", "True", "attentions", "audio_embeds", "audio_model", "audio_outputs", "audio_projection", "bool", "class", "config", "def", "else", "forward", "hidden_states", "if", "input_features", "is", "is_longer", "kwargs", "last_hidden_state", "not", "output_attentions", "output_hidden_states", "pooled_output", "pooler_output", "r", "return", "return_dict", "self", "torch", "tuple", "use_return_dict"], "clip/modeling_clip.py:contrastive_loss": ["Model_loss", "Tensor", "arange", "cross_entropy", "def", "device", "functional", "len", "logits", "nn", "return", "torch"], "clip/modeling_clip.py:clip_loss": ["Model_loss", "Tensor", "caption_loss", "contrastive_loss", "def", "image_loss", "return", "similarity", "t", "torch"], "clip/modeling_clip.py:_get_vector_norm": ["Tensor", "True", "_get_vector_norm", "def", "dim", "keepdim", "normed_tensor", "pow", "return", "square_tensor", "sum", "sum_tensor", "tensor", "torch"], "clip/modeling_clip.py:CLIPOutput.to_tuple": ["Any", "Model", "ModelOutput", "class", "def", "else", "for", "getattr", "if", "in", "k", "keys", "not", "return", "self", "text_model_output", "to_tuple", "tuple", "vision_model_output"], "clip/modeling_clip.py:CLIPVisionEmbeddings.__init__": ["Conv2d", "Embedding", "False", "Model", "ModelVisionConfig", "Module", "Parameter", "__init__", "arange", "bias", "class", "class_embedding", "config", "def", "embed_dim", "expand", "hidden_size", "image_size", "in_channels", "kernel_size", "nn", "num_channels", "num_patches", "num_positions", "out_channels", "patch_embedding", "patch_size", "persistent", "position_embedding", "position_ids", "randn", "register_buffer", "self", "stride", "super", "torch"], "clip/modeling_clip.py:CLIPVisionEmbeddings.interpolate_pos_encoding": ["False", "Model", "Module", "Tensor", "align_corners", "and", "bicubic", "cat", "class", "class_pos_embed", "def", "dim", "embeddings", "functional", "height", "if", "int", "interpolate", "interpolate_pos_encoding", "is_tracing", "jit", "mode", "new_height", "new_width", "nn", "not", "num_patches", "num_positions", "patch_pos_embed", "patch_size", "permute", "position_embedding", "position_ids", "reshape", "return", "self", "shape", "size", "sqrt_num_positions", "torch", "torch_int", "unsqueeze", "view", "weight", "width"], "clip/modeling_clip.py:CLIPVisionEmbeddings.forward": ["False", "FloatTensor", "Input", "Model", "Module", "Tensor", "ValueError", "_", "and", "batch_size", "cat", "class", "class_embedding", "class_embeds", "def", "dim", "doesn", "dtype", "else", "embeddings", "expand", "f", "flatten", "forward", "height", "if", "image", "image_size", "interpolate_pos_encoding", "match", "model", "nn", "not", "or", "patch_embedding", "patch_embeds", "pixel_values", "position_embedding", "position_ids", "raise", "return", "self", "shape", "size", "t", "target_dtype", "to", "torch", "transpose", "weight", "width"], "clip/modeling_clip.py:CLIPTextEmbeddings.__init__": ["Embedding", "False", "Model", "ModelTextConfig", "Module", "__init__", "arange", "class", "config", "def", "embed_dim", "expand", "hidden_size", "max_position_embeddings", "nn", "persistent", "position_embedding", "position_ids", "register_buffer", "self", "super", "token_embedding", "torch", "vocab_size"], "clip/modeling_clip.py:CLIPTextEmbeddings.forward": ["FloatTensor", "LongTensor", "Model", "Module", "None", "Sequence", "Tensor", "ValueError", "and", "be", "class", "def", "else", "embeddings", "f", "forward", "got", "if", "input_ids", "inputs_embeds", "is", "length", "less", "max_position_embedding", "max_position_embeddings", "must", "nn", "not", "position_embedding", "position_embeddings", "position_ids", "raise", "return", "self", "seq_length", "sequence", "shape", "than", "token_embedding", "torch", "weight"], "clip/modeling_clip.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "softmax", "to", "torch", "training", "transpose", "value"], "clip/modeling_clip.py:CLIPAttention.__init__": ["False", "Linear", "Model", "ModelTextConfig", "ModelVisionConfig", "Module", "ValueError", "__init__", "and", "attention_dropout", "be", "by", "class", "config", "def", "divisible", "dropout", "embed_dim", "f", "got", "head_dim", "hidden_size", "if", "is_causal", "k_proj", "must", "nn", "num_attention_heads", "num_heads", "out_proj", "q_proj", "raise", "scale", "self", "super", "v_proj"], "clip/modeling_clip.py:CLIPAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "Model", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_attn_implementation", "attention_interface", "attention_mask", "attn_output", "attn_weights", "batch_size", "class", "config", "contiguous", "def", "dropout", "eager", "eager_attention_forward", "else", "embed_dim", "forward", "head_dim", "hidden_states", "if", "k_proj", "keys", "kwargs", "nn", "not", "out_proj", "q_proj", "queries", "reshape", "return", "scale", "scaling", "self", "seq_length", "shape", "torch", "training", "transpose", "tuple", "v_proj", "values", "view"], "clip/modeling_clip.py:CLIPMLP.__init__": ["ACT2FN", "Linear", "Model", "Module", "__init__", "activation_fn", "class", "config", "def", "fc1", "fc2", "hidden_act", "hidden_size", "intermediate_size", "nn", "self", "super"], "clip/modeling_clip.py:CLIPMLP.forward": ["Model", "Module", "Tensor", "activation_fn", "class", "def", "fc1", "fc2", "forward", "hidden_states", "nn", "return", "self", "torch"], "clip/modeling_clip.py:CLIPEncoderLayer.__init__": ["GradientCheckpointingLayer", "LayerNorm", "Model", "ModelAttention", "ModelMLP", "ModelTextConfig", "ModelVisionConfig", "__init__", "class", "config", "def", "embed_dim", "eps", "hidden_size", "layer_norm1", "layer_norm2", "layer_norm_eps", "mlp", "nn", "self", "self_attn", "super"], "clip/modeling_clip.py:CLIPEncoderLayer.forward": ["FloatTensor", "GradientCheckpointingLayer", "Model", "Tensor", "TransformersKwargs", "Unpack", "_", "attention_mask", "class", "def", "forward", "hidden_states", "kwargs", "layer_norm1", "layer_norm2", "mlp", "residual", "return", "self", "self_attn", "torch"], "clip/modeling_clip.py:CLIPPreTrainedModel._init_weights": ["LayerNorm", "Linear", "Model", "ModelAttention", "ModelForImageClassification", "ModelMLP", "ModelModel", "ModelTextEmbeddings", "ModelTextModelWithProjection", "ModelVisionEmbeddings", "ModelVisionModelWithProjection", "None", "PreTrainedModel", "_init_weights", "and", "arange", "bias", "class", "class_embedding", "classifier", "config", "copy_", "def", "elif", "embed_dim", "expand", "factor", "fc1", "fc2", "fc_std", "hidden_size", "if", "in_proj_std", "init", "initializer_factor", "initializer_range", "is", "isinstance", "k_proj", "mean", "module", "nn", "normal_", "not", "num_hidden_layers", "num_positions", "ones_", "out_proj", "out_proj_std", "patch_embedding", "position_embedding", "position_ids", "q_proj", "self", "shape", "std", "text_embed_dim", "text_projection", "token_embedding", "torch", "v_proj", "vision_config", "vision_embed_dim", "visual_projection", "weight", "zeros_"], "clip/modeling_clip.py:CLIPEncoder.__init__": ["False", "Model", "ModelConfig", "ModelLayer", "Module", "ModuleList", "_", "__init__", "class", "config", "def", "for", "gradient_checkpointing", "in", "layers", "nn", "num_hidden_layers", "range", "self", "super"], "clip/modeling_clip.py:CLIPEncoder.forward": ["BaseModelOutput", "Model", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "class", "def", "encoder_layer", "for", "forward", "hidden_states", "in", "inputs_embeds", "kwargs", "last_hidden_state", "layers", "nn", "r", "return", "self", "torch"], "clip/modeling_clip.py:CLIPTextTransformer.__init__": ["LayerNorm", "Model", "ModelEncoder", "ModelTextConfig", "ModelTextEmbeddings", "Module", "__init__", "class", "config", "def", "embed_dim", "embeddings", "encoder", "eos_token_id", "eps", "final_layer_norm", "hidden_size", "layer_norm_eps", "nn", "self", "super"], "clip/modeling_clip.py:CLIPTextTransformer.forward": ["BaseModelOutput", "BaseModelOutputWithPooling", "Model", "Module", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "ValueError", "You", "arange", "argmax", "attention_mask", "cache_position", "class", "config", "create_causal_mask", "def", "device", "dim", "dtype", "else", "embeddings", "encoder", "encoder_outputs", "eos_token_id", "final_layer_norm", "forward", "have", "hidden_states", "if", "input_embeds", "input_ids", "input_shape", "inputs_embeds", "int", "is", "is_causal", "kwargs", "last_hidden_state", "nn", "past_key_values", "pooled_output", "pooler_output", "pop", "position_ids", "raise", "return", "self", "shape", "size", "specify", "to", "torch", "view"], "clip/modeling_clip.py:CLIPTextModel.__init__": ["Model", "ModelPreTrainedModel", "ModelTextConfig", "ModelTextTransformer", "__init__", "class", "config", "def", "post_init", "self", "super", "text_model"], "clip/modeling_clip.py:CLIPTextModel.get_input_embeddings": ["Model", "ModelPreTrainedModel", "Module", "class", "def", "embeddings", "get_input_embeddings", "nn", "return", "self", "text_model", "token_embedding"], "clip/modeling_clip.py:CLIPTextModel.set_input_embeddings": ["Model", "ModelPreTrainedModel", "class", "def", "embeddings", "self", "set_input_embeddings", "text_model", "token_embedding", "value"], "clip/modeling_clip.py:CLIPTextModel.forward": ["BaseModelOutputWithPooling", "Model", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "class", "def", "forward", "input_ids", "kwargs", "position_ids", "r", "return", "self", "text_model", "torch"], "clip/modeling_clip.py:CLIPVisionTransformer.__init__": ["LayerNorm", "Model", "ModelEncoder", "ModelVisionConfig", "ModelVisionEmbeddings", "Module", "__init__", "class", "config", "def", "embed_dim", "embeddings", "encoder", "eps", "hidden_size", "layer_norm_eps", "nn", "post_layernorm", "pre_layrnorm", "self", "super"], "clip/modeling_clip.py:CLIPVisionTransformer.forward": ["BaseModelOutput", "BaseModelOutputWithPooling", "False", "FloatTensor", "Model", "Module", "None", "TransformersKwargs", "Unpack", "ValueError", "You", "bool", "class", "def", "embeddings", "encoder", "encoder_outputs", "forward", "have", "hidden_states", "if", "inputs_embeds", "interpolate_pos_encoding", "is", "kwargs", "last_hidden_state", "nn", "pixel_values", "pooled_output", "pooler_output", "post_layernorm", "pre_layrnorm", "raise", "return", "self", "specify", "to", "torch"], "clip/modeling_clip.py:CLIPVisionModel.__init__": ["Model", "ModelPreTrainedModel", "ModelVisionConfig", "ModelVisionTransformer", "__init__", "class", "config", "def", "post_init", "self", "super", "vision_model"], "clip/modeling_clip.py:CLIPVisionModel.get_input_embeddings": ["Model", "ModelPreTrainedModel", "Module", "class", "def", "embeddings", "get_input_embeddings", "nn", "patch_embedding", "return", "self", "vision_model"], "clip/modeling_clip.py:CLIPVisionModel.forward": ["BaseModelOutputWithPooling", "False", "FloatTensor", "Model", "ModelPreTrainedModel", "None", "TransformersKwargs", "Unpack", "bool", "class", "def", "forward", "interpolate_pos_encoding", "kwargs", "pixel_values", "r", "return", "self", "torch", "vision_model"], "clip/modeling_clip.py:CLIPModel.__init__": ["False", "Linear", "Model", "ModelConfig", "ModelPreTrainedModel", "ModelTextConfig", "ModelTextModel", "ModelVisionConfig", "ModelVisionModel", "Parameter", "TypeError", "__init__", "_from_config", "be", "bias", "but", "class", "config", "def", "expected", "f", "hidden_size", "if", "is", "isinstance", "logit_scale", "logit_scale_init_value", "nn", "not", "of", "post_init", "projection_dim", "raise", "self", "super", "tensor", "text_config", "text_embed_dim", "text_model", "text_projection", "to", "torch", "type", "vision_config", "vision_embed_dim", "vision_model", "visual_projection"], "clip/modeling_clip.py:CLIPModel.get_text_features": ["BaseModelOutputWithPooling", "FloatTensor", "Model", "ModelPreTrainedModel", "None", "Tensor", "attention_mask", "class", "def", "get_text_features", "input_ids", "pooled_output", "pooler_output", "position_ids", "r", "return", "self", "text_features", "text_model", "text_outputs", "text_projection", "torch"], "clip/modeling_clip.py:CLIPModel.get_image_features": ["BaseModelOutputWithPooling", "False", "FloatTensor", "Model", "ModelPreTrainedModel", "bool", "class", "def", "get_image_features", "image_features", "interpolate_pos_encoding", "pixel_values", "pooled_output", "pooler_output", "r", "return", "self", "torch", "vision_model", "vision_outputs", "visual_projection"], "clip/modeling_clip.py:CLIPModel.forward": ["BaseModelOutputWithPooling", "False", "FloatTensor", "LongTensor", "Model", "ModelOutput", "ModelPreTrainedModel", "Model_loss", "None", "Tensor", "TransformersKwargs", "Unpack", "_get_vector_norm", "attention_mask", "bool", "class", "def", "device", "exp", "forward", "if", "image_embeds", "input_ids", "interpolate_pos_encoding", "kwargs", "logit_scale", "logits_per_image", "logits_per_text", "loss", "matmul", "pixel_values", "pooler_output", "position_ids", "r", "return", "return_loss", "self", "t", "text_embeds", "text_model", "text_model_output", "text_outputs", "text_projection", "to", "torch", "vision_model", "vision_model_output", "vision_outputs", "visual_projection"], "clip/modeling_clip.py:CLIPTextModelWithProjection.__init__": ["False", "Linear", "Model", "ModelPreTrainedModel", "ModelTextConfig", "ModelTextModel", "__init__", "_from_config", "bias", "class", "config", "def", "hidden_size", "nn", "post_init", "projection_dim", "self", "super", "text_model", "text_projection"], "clip/modeling_clip.py:CLIPTextModelWithProjection.get_input_embeddings": ["Model", "ModelPreTrainedModel", "Module", "class", "def", "embeddings", "get_input_embeddings", "nn", "return", "self", "text_model", "token_embedding"], "clip/modeling_clip.py:CLIPTextModelWithProjection.set_input_embeddings": ["Model", "ModelPreTrainedModel", "class", "def", "embeddings", "self", "set_input_embeddings", "text_model", "token_embedding", "value"], "clip/modeling_clip.py:CLIPTextModelWithProjection.forward": ["BaseModelOutputWithPooling", "Model", "ModelPreTrainedModel", "ModelTextModelOutput", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "class", "def", "forward", "input_ids", "kwargs", "last_hidden_state", "pooled_output", "pooler_output", "position_ids", "r", "return", "self", "text_embeds", "text_model", "text_outputs", "text_projection", "torch"], "clip/modeling_clip.py:CLIPVisionModelWithProjection.__init__": ["False", "Linear", "Model", "ModelPreTrainedModel", "ModelVisionConfig", "ModelVisionModel", "__init__", "_from_config", "bias", "class", "config", "def", "hidden_size", "nn", "post_init", "projection_dim", "self", "super", "vision_model", "visual_projection"], "clip/modeling_clip.py:CLIPVisionModelWithProjection.get_input_embeddings": ["Model", "ModelPreTrainedModel", "Module", "class", "def", "embeddings", "get_input_embeddings", "nn", "patch_embedding", "return", "self", "vision_model"], "clip/modeling_clip.py:CLIPVisionModelWithProjection.forward": ["BaseModelOutputWithPooling", "False", "FloatTensor", "Model", "ModelPreTrainedModel", "ModelVisionModelOutput", "None", "TransformersKwargs", "Unpack", "bool", "class", "def", "forward", "image_embeds", "interpolate_pos_encoding", "kwargs", "last_hidden_state", "pixel_values", "pooled_output", "pooler_output", "r", "return", "self", "torch", "vision_model", "vision_outputs", "visual_projection"], "clip/modeling_clip.py:CLIPForImageClassification.__init__": ["Identity", "Linear", "Model", "ModelConfig", "ModelPreTrainedModel", "ModelVisionModel", "None", "__init__", "_from_config", "class", "classifier", "config", "def", "else", "hidden_size", "if", "nn", "num_labels", "post_init", "self", "super", "vision_config", "vision_model"], "clip/modeling_clip.py:CLIPForImageClassification.forward": ["BaseModelOutputWithPooling", "ImageClassifierOutput", "Model", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "class", "classifier", "config", "def", "dim", "forward", "if", "is", "kwargs", "labels", "last_hidden_state", "logits", "loss", "loss_function", "mean", "not", "outputs", "pixel_values", "r", "return", "self", "sequence_output", "torch", "vision_model"], "clipseg/modeling_clipseg.py:contrastive_loss": ["Model_loss", "Tensor", "arange", "cross_entropy", "def", "device", "functional", "len", "logits", "nn", "return", "torch"], "clipseg/modeling_clipseg.py:clipseg_loss": ["Model_loss", "Tensor", "caption_loss", "contrastive_loss", "def", "image_loss", "return", "similarity", "t", "torch"], "clipseg/modeling_clipseg.py:CLIPSegOutput.to_tuple": ["Any", "Model", "ModelOutput", "class", "def", "else", "for", "getattr", "if", "in", "k", "keys", "not", "return", "self", "text_model_output", "to_tuple", "tuple", "vision_model_output"], "clipseg/modeling_clipseg.py:CLIPSegImageSegmentationOutput.to_tuple": ["Any", "Model", "ModelOutput", "class", "decoder_output", "def", "else", "for", "getattr", "if", "in", "k", "keys", "not", "return", "self", "to_tuple", "tuple", "vision_model_output"], "clipseg/modeling_clipseg.py:CLIPSegVisionEmbeddings.__init__": ["Conv2d", "Embedding", "False", "Model", "ModelVisionConfig", "Module", "Parameter", "__init__", "arange", "bias", "class", "class_embedding", "config", "def", "embed_dim", "expand", "hidden_size", "image_size", "in_channels", "kernel_size", "nn", "num_channels", "num_patches", "num_positions", "out_channels", "patch_embedding", "patch_size", "persistent", "position_embedding", "position_ids", "randn", "register_buffer", "self", "stride", "super", "torch"], "clipseg/modeling_clipseg.py:CLIPSegVisionEmbeddings.interpolate_pos_encoding": ["False", "Model", "Module", "Tensor", "align_corners", "and", "bicubic", "cat", "class", "class_pos_embed", "def", "dim", "embeddings", "functional", "height", "if", "int", "interpolate", "interpolate_pos_encoding", "is_tracing", "jit", "mode", "new_height", "new_width", "nn", "not", "num_patches", "num_positions", "patch_pos_embed", "patch_size", "permute", "position_embedding", "position_ids", "reshape", "return", "self", "shape", "size", "sqrt_num_positions", "torch", "torch_int", "unsqueeze", "view", "weight", "width"], "clipseg/modeling_clipseg.py:CLIPSegVisionEmbeddings.forward": ["FloatTensor", "Input", "Model", "Module", "Tensor", "True", "ValueError", "_", "and", "batch_size", "cat", "class", "class_embedding", "class_embeds", "def", "dim", "doesn", "else", "embeddings", "expand", "f", "flatten", "forward", "height", "if", "image", "image_size", "interpolate_pos_encoding", "match", "model", "nn", "not", "or", "patch_embedding", "patch_embeds", "pixel_values", "position_embedding", "position_ids", "raise", "return", "self", "shape", "size", "t", "torch", "transpose", "width"], "clipseg/modeling_clipseg.py:CLIPSegTextEmbeddings.__init__": ["Embedding", "False", "Model", "ModelTextConfig", "Module", "__init__", "arange", "class", "config", "def", "embed_dim", "expand", "hidden_size", "max_position_embeddings", "nn", "persistent", "position_embedding", "position_ids", "register_buffer", "self", "super", "token_embedding", "torch", "vocab_size"], "clipseg/modeling_clipseg.py:CLIPSegTextEmbeddings.forward": ["FloatTensor", "LongTensor", "Model", "Module", "None", "Sequence", "Tensor", "ValueError", "and", "be", "class", "def", "else", "embeddings", "f", "forward", "got", "if", "input_ids", "inputs_embeds", "is", "length", "less", "max_position_embedding", "max_position_embeddings", "must", "nn", "not", "position_embedding", "position_embeddings", "position_ids", "raise", "return", "self", "seq_length", "sequence", "shape", "than", "token_embedding", "torch", "weight"], "clipseg/modeling_clipseg.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "attention_mask", "attn_output", "attn_weights", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "softmax", "to", "torch", "training", "transpose", "value"], "clipseg/modeling_clipseg.py:CLIPSegAttention.__init__": ["False", "Linear", "Model", "ModelTextConfig", "ModelVisionConfig", "Module", "ValueError", "__init__", "and", "attention_dropout", "be", "by", "class", "config", "def", "divisible", "dropout", "embed_dim", "f", "got", "head_dim", "hidden_size", "if", "is_causal", "k_proj", "must", "nn", "num_attention_heads", "num_heads", "out_proj", "q_proj", "raise", "scale", "self", "super", "v_proj"], "clipseg/modeling_clipseg.py:CLIPSegAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "False", "Model", "Module", "None", "Tensor", "_attn_implementation", "and", "attention_interface", "attention_mask", "attn_output", "attn_weights", "batch_size", "bool", "causal_attention_mask", "class", "config", "contiguous", "def", "dropout", "eager", "eager_attention_forward", "elif", "else", "embed_dim", "flash_attention_2", "forward", "head_dim", "hidden_states", "if", "is", "is_causal", "k_proj", "keys", "nn", "not", "num_heads", "out_proj", "output_attentions", "q_proj", "queries", "reshape", "return", "scale", "scaling", "self", "seq_length", "shape", "torch", "training", "transpose", "tuple", "v_proj", "values", "view"], "clipseg/modeling_clipseg.py:CLIPSegMLP.__init__": ["ACT2FN", "Linear", "Model", "Module", "__init__", "activation_fn", "class", "config", "def", "fc1", "fc2", "hidden_act", "hidden_size", "intermediate_size", "nn", "self", "super"], "clipseg/modeling_clipseg.py:CLIPSegMLP.forward": ["Model", "Module", "Tensor", "activation_fn", "class", "def", "fc1", "fc2", "forward", "hidden_states", "nn", "return", "self", "torch"], "clipseg/modeling_clipseg.py:CLIPSegEncoderLayer.__init__": ["GradientCheckpointingLayer", "LayerNorm", "Model", "ModelAttention", "ModelConfig", "ModelMLP", "__init__", "class", "config", "def", "embed_dim", "eps", "hidden_size", "layer_norm1", "layer_norm2", "layer_norm_eps", "mlp", "nn", "self", "self_attn", "super"], "clipseg/modeling_clipseg.py:CLIPSegEncoderLayer.forward": ["False", "FloatTensor", "GradientCheckpointingLayer", "Model", "None", "Tensor", "attention_mask", "attn_weights", "bool", "causal_attention_mask", "class", "def", "forward", "hidden_states", "if", "layer_norm1", "layer_norm2", "mlp", "output_attentions", "outputs", "residual", "return", "self", "self_attn", "torch", "tuple"], "clipseg/modeling_clipseg.py:CLIPSegPreTrainedModel._init_weights": ["LayerNorm", "Linear", "Model", "ModelAttention", "ModelMLP", "ModelModel", "ModelTextEmbeddings", "ModelVisionEmbeddings", "None", "PreTrainedModel", "_init_weights", "and", "arange", "bias", "class", "class_embedding", "config", "copy_", "def", "elif", "embed_dim", "expand", "factor", "fc1", "fc2", "fc_std", "hidden_size", "if", "in_proj_std", "init", "initializer_factor", "initializer_range", "is", "isinstance", "k_proj", "mean", "module", "nn", "normal_", "not", "num_hidden_layers", "num_positions", "ones_", "out_proj", "out_proj_std", "patch_embedding", "position_embedding", "position_ids", "q_proj", "self", "shape", "std", "text_embed_dim", "text_projection", "token_embedding", "torch", "v_proj", "vision_embed_dim", "visual_projection", "weight", "zeros_"], "clipseg/modeling_clipseg.py:CLIPSegEncoder.__init__": ["False", "Model", "ModelConfig", "ModelLayer", "Module", "ModuleList", "_", "__init__", "class", "config", "def", "for", "gradient_checkpointing", "in", "layers", "nn", "num_hidden_layers", "range", "self", "super"], "clipseg/modeling_clipseg.py:CLIPSegEncoder.forward": ["BaseModelOutput", "Model", "Module", "None", "Tensor", "all_attentions", "attention_mask", "attentions", "bool", "causal_attention_mask", "class", "config", "def", "else", "encoder_layer", "encoder_states", "enumerate", "for", "forward", "hidden_states", "idx", "if", "in", "inputs_embeds", "is", "last_hidden_state", "layer_outputs", "layers", "nn", "not", "output_attentions", "output_hidden_states", "r", "return", "return_dict", "self", "torch", "tuple", "use_return_dict"], "clipseg/modeling_clipseg.py:CLIPSegTextTransformer.__init__": ["LayerNorm", "Model", "ModelEncoder", "ModelTextConfig", "ModelTextEmbeddings", "Module", "__init__", "class", "config", "def", "embed_dim", "embeddings", "encoder", "eos_token_id", "eps", "final_layer_norm", "hidden_size", "layer_norm_eps", "nn", "self", "super"], "clipseg/modeling_clipseg.py:CLIPSegTextTransformer.forward": ["BaseModelOutputWithPooling", "Model", "Module", "None", "Tensor", "ValueError", "You", "_create_4d_causal_attention_mask", "_prepare_4d_attention_mask", "arange", "argmax", "attention_mask", "attentions", "bool", "causal_attention_mask", "class", "config", "def", "device", "dim", "dtype", "else", "embeddings", "encoder", "encoder_outputs", "eos_token_id", "final_layer_norm", "forward", "have", "hidden_states", "if", "input_ids", "input_shape", "inputs_embeds", "int", "is", "last_hidden_state", "nn", "not", "output_attentions", "output_hidden_states", "pooled_output", "pooler_output", "position_ids", "raise", "return", "return_dict", "self", "shape", "size", "specify", "to", "torch", "tuple", "use_return_dict", "view"], "clipseg/modeling_clipseg.py:CLIPSegTextModel.__init__": ["Model", "ModelPreTrainedModel", "ModelTextConfig", "ModelTextTransformer", "__init__", "class", "config", "def", "post_init", "self", "super", "text_model"], "clipseg/modeling_clipseg.py:CLIPSegTextModel.get_input_embeddings": ["Model", "ModelPreTrainedModel", "Module", "class", "def", "embeddings", "get_input_embeddings", "nn", "return", "self", "text_model", "token_embedding"], "clipseg/modeling_clipseg.py:CLIPSegTextModel.set_input_embeddings": ["Model", "ModelPreTrainedModel", "class", "def", "embeddings", "self", "set_input_embeddings", "text_model", "token_embedding", "value"], "clipseg/modeling_clipseg.py:CLIPSegTextModel.forward": ["BaseModelOutputWithPooling", "Model", "ModelPreTrainedModel", "None", "Tensor", "attention_mask", "bool", "class", "def", "forward", "input_ids", "kwargs", "output_attentions", "output_hidden_states", "position_ids", "r", "return", "return_dict", "self", "text_model", "torch", "tuple"], "clipseg/modeling_clipseg.py:CLIPSegVisionTransformer.__init__": ["LayerNorm", "Model", "ModelEncoder", "ModelVisionConfig", "ModelVisionEmbeddings", "Module", "__init__", "class", "config", "def", "embed_dim", "embeddings", "encoder", "eps", "hidden_size", "layer_norm_eps", "nn", "post_layernorm", "pre_layrnorm", "self", "super"], "clipseg/modeling_clipseg.py:CLIPSegVisionTransformer.forward": ["BaseModelOutputWithPooling", "FloatTensor", "Model", "Module", "None", "True", "attentions", "bool", "class", "config", "def", "else", "embeddings", "encoder", "encoder_outputs", "forward", "hidden_states", "if", "inputs_embeds", "interpolate_pos_encoding", "is", "last_hidden_state", "nn", "not", "output_attentions", "output_hidden_states", "pixel_values", "pooled_output", "pooler_output", "post_layernorm", "pre_layrnorm", "return", "return_dict", "self", "torch", "tuple", "use_return_dict"], "clipseg/modeling_clipseg.py:CLIPSegVisionModel.__init__": ["Model", "ModelPreTrainedModel", "ModelVisionConfig", "ModelVisionTransformer", "__init__", "class", "config", "def", "post_init", "self", "super", "vision_model"], "clipseg/modeling_clipseg.py:CLIPSegVisionModel.get_input_embeddings": ["Model", "ModelPreTrainedModel", "Module", "class", "def", "embeddings", "get_input_embeddings", "nn", "patch_embedding", "return", "self", "vision_model"], "clipseg/modeling_clipseg.py:CLIPSegVisionModel.forward": ["BaseModelOutputWithPooling", "FloatTensor", "Model", "ModelPreTrainedModel", "None", "True", "bool", "class", "def", "forward", "interpolate_pos_encoding", "kwargs", "output_attentions", "output_hidden_states", "pixel_values", "r", "return", "return_dict", "self", "torch", "tuple", "vision_model"], "clipseg/modeling_clipseg.py:CLIPSegModel.__init__": ["False", "Linear", "Model", "ModelConfig", "ModelPreTrainedModel", "ModelTextConfig", "ModelTextTransformer", "ModelVisionConfig", "ModelVisionTransformer", "Parameter", "TypeError", "__init__", "_attn_implementation", "be", "bias", "but", "class", "config", "def", "expected", "f", "hidden_size", "if", "is", "isinstance", "logit_scale", "logit_scale_init_value", "nn", "not", "of", "post_init", "projection_dim", "raise", "self", "super", "tensor", "text_config", "text_embed_dim", "text_model", "text_projection", "to", "torch", "type", "vision_config", "vision_embed_dim", "vision_model", "visual_projection"], "clipseg/modeling_clipseg.py:CLIPSegModel.get_text_features": ["BaseModelOutputWithPooling", "FloatTensor", "Model", "ModelPreTrainedModel", "None", "Tensor", "attention_mask", "class", "def", "get_text_features", "input_ids", "pooled_output", "pooler_output", "position_ids", "r", "return", "self", "text_features", "text_model", "text_outputs", "text_projection", "torch"], "clipseg/modeling_clipseg.py:CLIPSegModel.get_image_features": ["BaseModelOutputWithPooling", "FloatTensor", "Model", "ModelPreTrainedModel", "True", "bool", "class", "def", "get_image_features", "image_features", "interpolate_pos_encoding", "pixel_values", "pooled_output", "pooler_output", "r", "return", "self", "torch", "vision_model", "vision_outputs", "visual_projection"], "clipseg/modeling_clipseg.py:CLIPSegModel.forward": ["FloatTensor", "LongTensor", "Model", "ModelOutput", "ModelPreTrainedModel", "Model_loss", "None", "Tensor", "True", "attention_mask", "bool", "class", "config", "def", "dim", "else", "exp", "forward", "if", "image_embeds", "input_ids", "interpolate_pos_encoding", "is", "keepdim", "kwargs", "logit_scale", "logits_per_image", "logits_per_text", "loss", "matmul", "norm", "not", "output", "output_attentions", "output_hidden_states", "p", "pixel_values", "position_ids", "r", "return", "return_dict", "return_loss", "self", "t", "text_embeds", "text_model", "text_model_output", "text_outputs", "text_projection", "torch", "tuple", "use_return_dict", "vision_model", "vision_model_output", "vision_outputs", "visual_projection"], "clipseg/modeling_clipseg.py:CLIPSegDecoderLayer.__init__": ["LayerNorm", "Model", "ModelAttention", "ModelConfig", "ModelMLP", "Module", "__init__", "class", "config", "def", "embed_dim", "eps", "hidden_size", "layer_norm1", "layer_norm2", "layer_norm_eps", "mlp", "nn", "self", "self_attn", "super"], "clipseg/modeling_clipseg.py:CLIPSegDecoderLayer.forward": ["False", "FloatTensor", "Model", "Module", "None", "Tensor", "attention_mask", "attn_weights", "bool", "causal_attention_mask", "class", "def", "forward", "hidden_states", "if", "layer_norm1", "layer_norm2", "mlp", "nn", "output_attentions", "outputs", "residual", "return", "self", "self_attn", "torch", "tuple"], "clipseg/modeling_clipseg.py:CLIPSegDecoder.__init__": ["Conv2d", "ConvTranspose2d", "Linear", "Model", "ModelConfig", "ModelLayer", "ModelPreTrainedModel", "ModuleList", "ReLU", "Sequential", "_", "__init__", "class", "conditional_layer", "config", "copy", "decoder_config", "decoder_intermediate_size", "decoder_num_attention_heads", "deepcopy", "def", "depth", "else", "extract_layers", "film_add", "film_mul", "for", "hidden_act", "hidden_size", "if", "in", "intermediate_size", "kernel_size", "layers", "len", "nn", "num_attention_heads", "padding", "patch_size", "post_init", "projection_dim", "range", "reduce_dim", "reduces", "relu", "self", "stride", "super", "transposed_convolution", "transposed_kernels", "use_complex_transposed_convolution", "vision_config"], "clipseg/modeling_clipseg.py:CLIPSegDecoder.forward": ["Model", "ModelOutput", "ModelPreTrainedModel", "None", "Tensor", "True", "activation", "activations", "all_attentions", "all_hidden_states", "attention_mask", "attentions", "batch_size", "bool", "causal_attention_mask", "class", "conditional_embeddings", "conditional_layer", "def", "else", "enumerate", "film_add", "film_mul", "for", "forward", "hidden_states", "i", "if", "in", "int", "is", "kwargs", "layer", "layer_outputs", "layers", "logits", "math", "not", "output", "output_attentions", "output_hidden_states", "permute", "reduce", "reduces", "return", "return_dict", "self", "shape", "size", "sqrt", "squeeze", "torch", "transposed_convolution", "tuple", "v", "view", "zip"], "clipseg/modeling_clipseg.py:CLIPSegForImageSegmentation.__init__": ["Model", "ModelConfig", "ModelDecoder", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "clip", "config", "decoder", "def", "extract_layers", "post_init", "self", "super"], "clipseg/modeling_clipseg.py:CLIPSegForImageSegmentation.get_conditional_embeddings": ["Invalid", "Make", "Model", "ModelPreTrainedModel", "None", "Tensor", "ValueError", "are", "as", "attention_mask", "batch_size", "be", "class", "clip", "conditional", "conditional_embeddings", "conditional_pixel_values", "def", "either", "elif", "else", "get_conditional_embeddings", "get_image_features", "get_text_features", "if", "images", "input_ids", "int", "is", "len", "many", "no_grad", "not", "or", "pass", "position_ids", "prompt", "provided", "query", "raise", "return", "self", "should", "sure", "texts", "there", "to", "torch", "with"], "clipseg/modeling_clipseg.py:CLIPSegForImageSegmentation.forward": ["BCEWithLogitsLoss", "BaseModelOutputWithPooling", "FloatTensor", "LongTensor", "Make", "Model", "ModelImageSegmentationOutput", "ModelOutput", "ModelPreTrainedModel", "None", "Tensor", "True", "ValueError", "activations", "are", "as", "attention_mask", "attentions", "batch", "batch_size", "bool", "class", "clip", "conditional", "conditional_embeddings", "conditional_pixel_values", "config", "decoder", "decoder_output", "decoder_outputs", "def", "device", "dimension", "else", "embeddings", "extract_layers", "feature", "for", "forward", "get_conditional_embeddings", "hidden_states", "i", "if", "images", "in", "input_ids", "interpolate_pos_encoding", "is", "kwargs", "labels", "last_hidden_state", "logits", "loss", "loss_fn", "many", "matches", "nn", "no_grad", "not", "of", "output", "output_attentions", "output_hidden_states", "pass", "pixel_values", "pooled_output", "pooler_output", "position_ids", "projection_dim", "query", "r", "raise", "return", "return_dict", "self", "shape", "sure", "that", "the", "there", "to", "torch", "tuple", "use_return_dict", "vision_model", "vision_model_output", "vision_outputs", "visual_projection", "with"], "clvp/modeling_clvp.py:contrastive_loss": ["Model_loss", "Tensor", "arange", "cross_entropy", "def", "device", "functional", "len", "logits", "nn", "return", "torch"], "clvp/modeling_clvp.py:clvp_loss": ["Model_loss", "Tensor", "caption_loss", "contrastive_loss", "def", "return", "similarity", "speech_loss", "t", "torch"], "clvp/modeling_clvp.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "clvp/modeling_clvp.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cos", "def", "k", "k_embed", "position_ids", "q", "q_embed", "return", "rotate_half", "sin", "unsqueeze", "unsqueeze_dim", "v", "v_embed"], "clvp/modeling_clvp.py:_pad_extra_bos_eos_tokens": ["None", "True", "_pad_extra_bos_eos_tokens", "add_bos_token", "add_eos_token", "attention_mask", "bos_token_id", "concatenate", "def", "device", "dtype", "each_input_id", "else", "enumerate", "eos_token_id", "for", "functional", "i", "if", "in", "input_ids", "is", "isin_mps_friendly", "min", "modified_input_ids", "nn", "not", "pad", "pad_token_id", "pos", "return", "shape", "sum", "tensor", "torch", "value", "where", "zeros"], "clvp/modeling_clvp.py:ClvpRMSNorm.__init__": ["ModelRMSNorm", "Module", "Parameter", "__init__", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "clvp/modeling_clvp.py:ClvpRMSNorm.forward": ["ModelRMSNorm", "Module", "True", "class", "def", "dtype", "float32", "forward", "hidden_states", "input_dtype", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "clvp/modeling_clvp.py:ClvpRMSNorm.extra_repr": ["ModelRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "variance_epsilon", "weight"], "clvp/modeling_clvp.py:ClvpRotaryPositionalEmbedding.__init__": ["ModelRotaryPositionalEmbedding", "Module", "None", "__init__", "arange", "cached_rotary_positional_embedding", "cached_sequence_length", "class", "config", "def", "dim", "dtype", "float", "int64", "inv_freq", "max", "nn", "num_attention_heads", "projection_dim", "register_buffer", "self", "super", "torch"], "clvp/modeling_clvp.py:ClvpRotaryPositionalEmbedding.forward": ["FloatTensor", "ModelRotaryPositionalEmbedding", "Module", "None", "and", "arange", "cached_rotary_positional_embedding", "cached_sequence_length", "cat", "class", "def", "device", "dim", "einsum", "embeddings", "forward", "freqs", "hidden_states", "i", "if", "ij", "inv_freq", "is", "j", "nn", "not", "return", "self", "sequence_length", "shape", "time_stamps", "torch", "type_as", "unsqueeze"], "clvp/modeling_clvp.py:ClvpSelfAttention.__init__": ["False", "Linear", "ModelSelfAttention", "Module", "None", "ValueError", "__init__", "and", "attention_dropout", "be", "bias", "bool", "by", "class", "config", "def", "divisible", "dropout", "dtype", "embed_dim", "f", "got", "hasattr", "head_dim", "hidden_size", "if", "k_proj", "layer_idx", "max_position_embeddings", "max_positions", "must", "nn", "num_attention_heads", "num_heads", "ones", "out_proj", "persistent", "q_proj", "raise", "register_buffer", "scale", "self", "super", "torch", "tril", "use_attention_bias", "v_proj", "view"], "clvp/modeling_clvp.py:ClvpSelfAttention._shape": ["ModelSelfAttention", "Module", "Tensor", "_shape", "bsz", "class", "contiguous", "def", "head_dim", "int", "nn", "num_heads", "return", "self", "seq_len", "tensor", "torch", "transpose", "view"], "clvp/modeling_clvp.py:ClvpSelfAttention.forward": ["Attention", "Cache", "False", "FloatTensor", "LongTensor", "ModelSelfAttention", "Module", "None", "Tensor", "ValueError", "_", "_shape", "and", "apply_rotary_pos_emb", "attention_mask", "attn_output", "attn_probs", "attn_weights", "be", "bool", "bsz", "but", "cache_position", "cat", "class", "contiguous", "cos", "def", "dim", "dropout", "embed_dim", "f", "forward", "functional", "head_dim", "hidden_states", "if", "is", "k_proj", "key_pass", "key_rot", "key_states", "layer_idx", "mask", "matmul", "must", "nn", "not", "num_heads", "of", "out_proj", "output_attentions", "p", "past_key_values", "position_ids", "provided", "q_proj", "query_pass", "query_rot", "query_states", "raise", "reshape", "return", "rotary_emb_dim", "rotary_pos_emb", "scale", "self", "shape", "should", "sin", "size", "softmax", "squeeze", "src_len", "tgt_len", "torch", "training", "transpose", "tuple", "update", "use_cache", "v_proj", "value_pass", "value_rot", "value_states", "when"], "clvp/modeling_clvp.py:ClvpGatedLinearUnit.__init__": ["ACT2FN", "Linear", "ModelGatedLinearUnit", "Module", "__init__", "activation_fn", "class", "config", "def", "hidden_act", "hidden_size", "intermediate_size", "nn", "proj", "self", "super"], "clvp/modeling_clvp.py:ClvpGatedLinearUnit.forward": ["FloatTensor", "ModelGatedLinearUnit", "Module", "activation_fn", "chunk", "class", "def", "dim", "forward", "gate", "hidden_states", "nn", "proj", "return", "self", "torch"], "clvp/modeling_clvp.py:ClvpEncoderMLP.__init__": ["Dropout", "Linear", "ModelEncoderMLP", "ModelGatedLinearUnit", "Module", "__init__", "class", "config", "def", "dropout", "dropout_layer", "fc1", "fc2", "hidden_size", "intermediate_size", "nn", "self", "super"], "clvp/modeling_clvp.py:ClvpEncoderMLP.forward": ["FloatTensor", "ModelEncoderMLP", "Module", "class", "def", "dropout_layer", "fc1", "fc2", "forward", "hidden_states", "nn", "return", "self", "torch"], "clvp/modeling_clvp.py:ClvpEncoderLayer.__init__": ["ModelConfig", "ModelEncoderLayer", "ModelEncoderMLP", "ModelRMSNorm", "ModelSelfAttention", "Module", "__init__", "class", "config", "def", "embed_dim", "eps", "hidden_size", "input_rmsnorm", "layer_norm_eps", "mlp", "nn", "post_attention_rmsnorm", "self", "self_attn", "super"], "clvp/modeling_clvp.py:ClvpEncoderLayer.forward": ["False", "FloatTensor", "LongTensor", "ModelEncoderLayer", "Module", "None", "attention_mask", "attn_weights", "bool", "class", "def", "forward", "hidden_states", "input_rmsnorm", "mlp", "nn", "output_attentions", "position_ids", "post_attention_rmsnorm", "residual", "return", "rotary_pos_emb", "self", "self_attn", "torch", "tuple"], "clvp/modeling_clvp.py:ClvpSequenceSummary.__init__": ["Callable", "Dropout", "Identity", "Linear", "ModelConfig", "ModelSequenceSummary", "Module", "None", "NotImplementedError", "__init__", "activation", "activation_string", "and", "attn", "class", "config", "def", "else", "first_dropout", "get_activation", "getattr", "hasattr", "hidden_size", "if", "last", "last_dropout", "nn", "num_classes", "num_labels", "raise", "self", "summary", "summary_activation", "summary_first_dropout", "summary_last_dropout", "summary_proj_to_labels", "summary_type", "summary_use_proj", "super"], "clvp/modeling_clvp.py:ClvpSequenceSummary.forward": ["FloatTensor", "LongTensor", "ModelSequenceSummary", "Module", "None", "NotImplementedError", "activation", "attn", "class", "cls_index", "def", "dim", "dtype", "elif", "else", "expand", "first", "first_dropout", "forward", "full_like", "gather", "hidden_states", "if", "is", "last", "last_dropout", "long", "mean", "nn", "output", "raise", "return", "self", "shape", "size", "squeeze", "summary", "summary_type", "torch", "unsqueeze"], "clvp/modeling_clvp.py:ClvpDecoderMLP.__init__": ["ACT2FN", "Conv1D", "Dropout", "ModelDecoderMLP", "Module", "__init__", "act", "activation_function", "c_fc", "c_proj", "class", "config", "def", "dropout", "embed_dim", "hidden_size", "intermediate_size", "nn", "resid_pdrop", "self", "super"], "clvp/modeling_clvp.py:ClvpDecoderMLP.forward": ["FloatTensor", "ModelDecoderMLP", "Module", "None", "act", "c_fc", "c_proj", "class", "def", "dropout", "forward", "hidden_states", "nn", "return", "self", "torch", "tuple"], "clvp/modeling_clvp.py:ClvpDecoderLayer.__init__": ["LayerNorm", "ModelDecoderLayer", "ModelDecoderMLP", "ModelSelfAttention", "Module", "None", "__init__", "attn", "class", "config", "def", "else", "eps", "hidden_size", "if", "inner_dim", "input_layernorm", "is", "layer_idx", "layer_norm_epsilon", "mlp", "n_inner", "nn", "not", "post_attention_layernorm", "self", "super"], "clvp/modeling_clvp.py:ClvpDecoderLayer.forward": ["Cache", "False", "FloatTensor", "LongTensor", "ModelDecoderLayer", "Module", "None", "Tensor", "attention_mask", "attn", "attn_output", "attn_outputs", "bool", "cache_position", "class", "def", "feed_forward_hidden_states", "forward", "hidden_states", "input_layernorm", "mlp", "nn", "output_attentions", "past_key_values", "position_ids", "post_attention_layernorm", "residual", "return", "self", "torch", "tuple", "use_cache"], "clvp/modeling_clvp.py:ClvpConditioningEncoder.__init__": ["Conv1d", "Embedding", "False", "GroupNorm", "ModelConditioningEncoder", "ModelConfig", "ModelSelfAttention", "Module", "ModuleList", "True", "_", "__init__", "affine", "class", "compute_groupnorm_groups", "config", "decoder_config", "def", "eps", "feature_size", "for", "gradient_checkpointing", "group_norms", "hidden_size", "in", "kernel_size", "max_text_tokens", "mel_attn_blocks", "mel_conv", "nn", "num_groups", "num_mel_attn_blocks", "range", "self", "super", "text_config", "text_position_embedding", "text_token_embedding", "vocab_size"], "clvp/modeling_clvp.py:ClvpConditioningEncoder.compute_groupnorm_groups": ["GroupNorm", "ModelConditioningEncoder", "Module", "Number", "Please", "ValueError", "a", "be", "but", "channels", "class", "compute_groupnorm_groups", "consider", "def", "different", "elif", "f", "for", "greater", "groups", "hidden_size", "if", "int", "is", "it", "must", "nn", "of", "raise", "return", "self", "than", "the", "using", "while"], "clvp/modeling_clvp.py:ClvpConditioningEncoder.forward": ["FloatTensor", "Found", "LongTensor", "ModelConditioningEncoder", "Module", "None", "The", "ValueError", "You", "_pad_extra_bos_eos_tokens", "and", "at", "attention_mask", "audios", "batch_size", "be", "bos_token_id", "both", "cannot", "checkpoint", "class", "concat", "cumsum", "def", "device", "dim", "dtype", "either", "elif", "else", "enumerate", "eos_token_id", "f", "for", "forward", "gradient_checkpointing", "group_norms", "have", "i", "if", "in", "input_features", "input_ids", "inputs_embeds", "is", "long", "mel_attn_block", "mel_attn_blocks", "mel_conv", "mel_spec", "must", "nn", "not", "number", "of", "ones", "or", "position_embeds", "position_ids", "raise", "repeat", "residual_mel_spec", "return", "same", "self", "seq_length", "shape", "size", "specify", "text_config", "text_embeds", "text_position_embedding", "text_token_embedding", "texts", "the", "time", "to", "torch", "training", "transpose", "unsqueeze", "utils", "vs"], "clvp/modeling_clvp.py:ClvpPreTrainedModel._init_weights": ["Conv1D", "Conv1d", "Embedding", "GroupNorm", "LayerNorm", "Linear", "ModelConditioningEncoder", "ModelEncoder", "ModelEncoderMLP", "ModelForCausalLM", "ModelModelForConditionalGeneration", "ModelPreTrainedModel", "ModelRMSNorm", "ModelRotaryPositionalEmbedding", "ModelSelfAttention", "Module", "None", "PreTrainedModel", "_init_weights", "arange", "bias", "bool", "c_proj", "class", "config", "constant_", "copy_", "def", "dim", "dtype", "elif", "else", "factor", "fc1", "fc2", "fc_std", "float", "for", "get_text_config", "getattr", "hasattr", "hidden_size", "if", "in", "in_proj_std", "init", "initializer_factor", "initializer_range", "int64", "inv_freq", "is", "isinstance", "logit_scale", "logit_scale_init_value", "math", "max", "max_position_embeddings", "max_positions", "mean", "mel_conv", "module", "name", "named_parameters", "nn", "normal_", "not", "num_attention_heads", "num_hidden_layers", "ones", "ones_", "p", "proj", "projection", "projection_dim", "self", "sqrt", "std", "torch", "tril", "view", "weight", "zeros_"], "clvp/modeling_clvp.py:ClvpEncoder.__init__": ["Embedding", "False", "LayerNorm", "Linear", "ModelConfig", "ModelEncoder", "ModelEncoderLayer", "ModelPreTrainedModel", "ModelRotaryPositionalEmbedding", "ModelSequenceSummary", "ModuleList", "None", "_", "__init__", "bias", "class", "config", "def", "else", "eps", "final_layer_norm", "for", "gradient_checkpointing", "hidden_size", "if", "in", "layer_norm_eps", "layers", "nn", "num_hidden_layers", "post_init", "projection", "projection_dim", "range", "rotary_pos_emb", "self", "sequence_summary", "super", "token_embedding", "use_rotary_embedding", "vocab_size"], "clvp/modeling_clvp.py:ClvpEncoder.get_input_embeddings": ["ModelEncoder", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "return", "self", "token_embedding"], "clvp/modeling_clvp.py:ClvpEncoder.set_input_embeddings": ["ModelEncoder", "ModelPreTrainedModel", "class", "def", "self", "set_input_embeddings", "token_embedding", "value"], "clvp/modeling_clvp.py:ClvpEncoder.forward": ["BaseModelOutput", "LongTensor", "ModelEncoder", "ModelEncoderOutput", "ModelPreTrainedModel", "None", "ValueError", "You", "__call__", "_prepare_4d_attention_mask", "all_attentions", "and", "arange", "at", "attention_mask", "attentions", "bool", "both", "cannot", "checkpoint", "class", "config", "def", "device", "dtype", "either", "elif", "else", "embeds", "encoder_layer", "encoder_states", "enumerate", "final_layer_norm", "for", "forward", "gradient_checkpointing", "have", "hidden_states", "idx", "if", "in", "input_ids", "input_shape", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layer_outputs", "layers", "long", "not", "or", "output_attentions", "output_hidden_states", "pooled_output", "pooler_output", "position_ids", "projection", "r", "raise", "return", "return_dict", "rotary_pos_emb", "same", "self", "sequence_summary", "size", "specify", "the", "time", "to", "token_embedding", "torch", "training", "tuple", "unsqueeze", "use_return_dict", "utils", "v", "view", "warn_if_padding_and_no_attention_mask"], "clvp/modeling_clvp.py:ClvpDecoder.__init__": ["Dropout", "Embedding", "False", "LayerNorm", "ModelDecoder", "ModelDecoderLayer", "ModelPreTrainedModel", "ModuleList", "__init__", "class", "config", "def", "drop", "embd_pdrop", "eps", "for", "gradient_checkpointing", "hidden_size", "i", "in", "input_embeds_layer", "layer_idx", "layer_norm", "layer_norm_epsilon", "layers", "max_position_embeddings", "nn", "num_hidden_layers", "position_embeds_layer", "post_init", "range", "self", "super", "vocab_size"], "clvp/modeling_clvp.py:ClvpDecoder.get_input_embeddings": ["ModelDecoder", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "input_embeds_layer", "return", "self"], "clvp/modeling_clvp.py:ClvpDecoder.set_input_embeddings": ["ModelDecoder", "ModelPreTrainedModel", "class", "def", "input_embeds_layer", "new_embeddings", "self", "set_input_embeddings"], "clvp/modeling_clvp.py:ClvpDecoder.forward": ["BaseModelOutputWithPastAndCrossAttentions", "Cache", "DynamicCache", "False", "FloatTensor", "LongTensor", "ModelDecoder", "ModelPreTrainedModel", "None", "Setting", "Tensor", "True", "ValueError", "You", "__call__", "_prepare_4d_causal_attention_mask", "add_cross_attention", "all_cross_attentions", "all_hidden_states", "all_self_attentions", "and", "arange", "at", "attention_mask", "attentions", "block", "bool", "both", "cache_position", "cannot", "checkpoint", "checkpointing", "class", "config", "cross_attentions", "def", "device", "drop", "dtype", "either", "elif", "else", "enumerate", "for", "forward", "get_seq_length", "gradient", "gradient_checkpointing", "have", "hidden_states", "i", "if", "in", "incompatible", "input_embeds_layer", "input_ids", "input_shape", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layer_norm", "layers", "logger", "long", "not", "or", "output_attentions", "output_hidden_states", "output_shape", "outputs", "past_key_values", "past_key_values_length", "position_embeds", "position_embeds_layer", "position_ids", "raise", "return", "return_dict", "same", "self", "shape", "size", "specify", "the", "time", "to", "token_type_embeds", "token_type_ids", "torch", "training", "tuple", "unsqueeze", "use_cache", "use_return_dict", "utils", "v", "view", "warn_if_padding_and_no_attention_mask", "warning_once", "with"], "clvp/modeling_clvp.py:ClvpModel.__init__": ["ModelDecoder", "ModelDecoderConfig", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "config", "decoder", "def", "post_init", "self", "super"], "clvp/modeling_clvp.py:ClvpModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "decoder", "def", "get_input_embeddings", "input_embeds_layer", "return", "self"], "clvp/modeling_clvp.py:ClvpModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "decoder", "def", "input_embeds_layer", "self", "set_input_embeddings", "value"], "clvp/modeling_clvp.py:ClvpModel.forward": ["BaseModelOutputWithPastAndCrossAttentions", "Cache", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "cross_attentions", "decoder", "decoder_outputs", "def", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "last_hidden_state", "not", "output_attentions", "output_hidden_states", "past_key_values", "position_ids", "return", "return_dict", "self", "token_type_ids", "torch", "tuple", "use_cache", "use_return_dict"], "clvp/modeling_clvp.py:ClvpForCausalLM.__init__": ["GenerationMixin", "LayerNorm", "Linear", "ModelForCausalLM", "ModelModel", "ModelPreTrainedModel", "True", "__init__", "bias", "class", "config", "def", "final_norm", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "vocab_size"], "clvp/modeling_clvp.py:ClvpForCausalLM.get_output_embeddings": ["GenerationMixin", "ModelForCausalLM", "ModelPreTrainedModel", "None", "class", "def", "get_output_embeddings", "return", "self"], "clvp/modeling_clvp.py:ClvpForCausalLM.get_input_embeddings": ["GenerationMixin", "ModelForCausalLM", "ModelPreTrainedModel", "class", "decoder", "def", "get_input_embeddings", "input_embeds_layer", "model", "return", "self"], "clvp/modeling_clvp.py:ClvpForCausalLM.set_input_embeddings": ["GenerationMixin", "ModelForCausalLM", "ModelPreTrainedModel", "class", "decoder", "def", "input_embeds_layer", "model", "new_embeddings", "self", "set_input_embeddings"], "clvp/modeling_clvp.py:ClvpForCausalLM._prepare_model_inputs": ["GenerationMixin", "Make", "ModelForCausalLM", "ModelPreTrainedModel", "None", "Tensor", "ValueError", "_maybe_initialize_input_ids_for_generation", "_prepare_model_inputs", "allowed", "alongside", "and", "arange", "attention_mask", "bos_token_id", "class", "concat", "conditioning_embeds", "config", "cumsum", "decoder", "def", "device", "dict", "dim", "dtype", "either", "elif", "else", "f", "fill_value", "for", "full", "get", "hasattr", "if", "in", "input_embeds_layer", "input_ids", "input_name", "inputs", "inputs_embeds", "inputs_kwarg", "int", "is", "items", "k", "long", "main_input_name", "mel_start_token_embedding", "model", "model_kwargs", "not", "ones", "or", "pass", "passed", "pop", "position_embeds_layer", "position_ids", "raise", "repeat", "return", "self", "shape", "str", "sure", "to", "torch", "tuple", "unsqueeze", "v", "were", "which"], "clvp/modeling_clvp.py:ClvpForCausalLM.prepare_inputs_for_generation": ["False", "GenerationMixin", "ModelForCausalLM", "ModelPreTrainedModel", "None", "and", "cache_position", "class", "conditioning_embeds", "def", "device", "dtype", "if", "input_ids", "input_ids_length", "inputs_embeds", "is", "is_first_iteration", "kwargs", "long", "model_inputs", "not", "past_key_values", "position_ids", "prepare_inputs_for_generation", "return", "self", "shape", "super", "tensor", "torch"], "clvp/modeling_clvp.py:ClvpForCausalLM.forward": ["Cache", "CausalLMOutputWithCrossAttentions", "CrossEntropyLoss", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForCausalLM", "ModelPreTrainedModel", "None", "Tensor", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "contiguous", "cross_attentions", "def", "device", "else", "final_norm", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "lm_head", "lm_logits", "logits", "loss", "loss_fct", "model", "not", "output", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "position_ids", "r", "return", "return_dict", "self", "shift_labels", "shift_logits", "size", "to", "token_type_ids", "torch", "tuple", "use_cache", "use_return_dict", "view"], "clvp/modeling_clvp.py:ClvpModelForConditionalGeneration.__init__": ["GenerationMixin", "ModelConditioningEncoder", "ModelConfig", "ModelDecoderConfig", "ModelEncoder", "ModelEncoderConfig", "ModelForCausalLM", "ModelModelForConditionalGeneration", "ModelPreTrainedModel", "Parameter", "TypeError", "__init__", "be", "but", "class", "conditioning_encoder", "config", "decoder_config", "def", "expected", "f", "if", "is", "isinstance", "logit_scale", "logit_scale_init_value", "nn", "not", "of", "post_init", "raise", "self", "speech_config", "speech_decoder_model", "speech_encoder_model", "super", "tensor", "text_config", "text_encoder_model", "to", "torch", "type"], "clvp/modeling_clvp.py:ClvpModelForConditionalGeneration.fix_speech_decoder_output": ["GenerationMixin", "LongTensor", "ModelModelForConditionalGeneration", "ModelPreTrainedModel", "argmax", "bool", "class", "config", "continue", "decoder_config", "decoder_fixing_codes", "def", "device", "dtype", "each_seq_stop_token_index", "enumerate", "eos_token_id", "fix_speech_decoder_output", "for", "i", "if", "in", "long", "mask", "masked_fill", "return", "self", "shape", "speech_decoder_model", "speech_ids", "stm", "stop_token_indices", "sum", "tensor", "torch", "value", "where"], "clvp/modeling_clvp.py:ClvpModelForConditionalGeneration.get_text_features": ["FloatTensor", "GenerationMixin", "LongTensor", "ModelModelForConditionalGeneration", "ModelPreTrainedModel", "None", "attention_mask", "class", "def", "get_text_features", "input_ids", "inputs_embeds", "outputs", "r", "return", "self", "text_encoder_inputs_embeds", "text_encoder_model", "torch"], "clvp/modeling_clvp.py:ClvpModelForConditionalGeneration.get_speech_features": ["Either", "FloatTensor", "GenerationConfig", "GenerationMixin", "LongTensor", "ModelModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Tensor", "ValueError", "and", "attention_mask", "be", "class", "conditioning_embeds", "conditioning_encoder", "conditioning_encoder_inputs_embeds", "def", "fix_speech_decoder_output", "generate", "generation_config", "get_speech_features", "if", "input_features", "input_ids", "inputs_embeds", "is", "kwargs", "must", "or", "outputs", "provided", "r", "raise", "return", "self", "speech_decoder_model", "speech_encoder_model", "speech_ids", "torch", "update"], "clvp/modeling_clvp.py:ClvpModelForConditionalGeneration.forward": ["False", "FloatTensor", "GenerationMixin", "LongTensor", "ModelModelForConditionalGeneration", "ModelOutput", "ModelPreTrainedModel", "Model_loss", "None", "Tensor", "True", "argmax", "attention_mask", "bool", "cache_position", "class", "conditioning_embeds", "conditioning_encoder", "conditioning_encoder_inputs_embeds", "config", "decoder_hidden_states", "decoder_outputs", "def", "dim", "else", "exp", "fix_speech_decoder_output", "forward", "hidden_states", "if", "input_features", "input_ids", "inputs_embeds", "is", "keepdim", "kwargs", "logit_scale", "logits_per_speech", "logits_per_text", "loss", "matmul", "ndim", "norm", "not", "output", "output_attentions", "output_hidden_states", "p", "r", "return", "return_dict", "return_loss", "self", "speech_decoder_model", "speech_embeds", "speech_encoder_hidden_states", "speech_encoder_model", "speech_ids", "speech_model_output", "speech_outputs", "t", "text_embeds", "text_encoder_hidden_states", "text_encoder_inputs_embeds", "text_encoder_model", "text_model_output", "text_outputs", "torch", "tuple", "use_return_dict"], "clvp/modeling_clvp.py:ClvpModelForConditionalGeneration.generate": ["False", "FloatTensor", "Found", "GenerationConfig", "GenerationMixin", "LongTensor", "Maximum", "ModelModelForConditionalGeneration", "ModelOutput", "ModelPreTrainedModel", "None", "Please", "True", "ValueError", "_pad_extra_bos_eos_tokens", "_validate_model_kwargs", "add_bos_token", "attention_mask", "bool", "bos_token_id", "class", "conditioning_embeds", "conditioning_encoder", "config", "copy", "decoder_config", "decoder_hidden_states", "decoder_outputs", "deepcopy", "def", "dim", "eos_token_id", "exp", "f", "fix_speech_decoder_output", "functional", "generate", "generation_config", "hidden_states", "if", "input_features", "input_ids", "int", "is", "isinstance", "keepdim", "kwargs", "length", "logit_scale", "logits_per_speech", "logits_per_text", "make", "matmul", "max_text_tokens", "maximum", "model_kwargs", "nn", "norm", "not", "of", "output", "output_hidden_states", "p", "pad", "pad_to_max_mel_tokens", "padding_needed", "raise", "reached", "return", "return_dict", "return_dict_in_generate", "self", "sequence", "sequence_length", "sequences", "shape", "speech_decoder_model", "speech_embeds", "speech_encoder_hidden_states", "speech_encoder_model", "speech_ids", "speech_model_output", "speech_outputs", "sure", "t", "text_config", "text_embeds", "text_encoder_hidden_states", "text_encoder_model", "text_model_output", "text_outputs", "that", "the", "torch", "update", "validate", "value"], "codegen/modeling_codegen.py:create_sinusoidal_positions": ["Model_sinusoidal_positions", "Tensor", "arange", "cat", "cos", "def", "dim", "dtype", "einsum", "float", "i", "int", "int64", "inv_freq", "j", "num_pos", "return", "sin", "sinusoid_inp", "torch"], "codegen/modeling_codegen.py:rotate_every_two": ["Model_every_two", "Tensor", "def", "dim", "flatten", "return", "stack", "torch", "x", "x1", "x2"], "codegen/modeling_codegen.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "None", "Tensor", "cos", "def", "repeat_interleave", "return", "rotate_every_two", "sin", "tensor", "torch"], "codegen/modeling_codegen.py:CodeGenAttention.__init__": ["Dropout", "False", "Instantiating", "Linear", "ModelAttention", "Module", "None", "Please", "ValueError", "__class__", "__init__", "__name__", "a", "and", "attn_dropout", "attn_pdrop", "be", "bias", "by", "caching", "call", "class", "config", "create_sinusoidal_positions", "creating", "def", "divisible", "during", "embed_dim", "embed_positions", "errors", "f", "forward", "got", "head_dim", "hidden_size", "if", "is", "layer_idx", "lead", "logger", "make", "math", "max_position_embeddings", "max_positions", "must", "nn", "not", "num_attention_heads", "or", "out_proj", "passing", "persistent", "pos_embd_dim", "provide", "qkv_proj", "raise", "recommended", "register_buffer", "resid_dropout", "resid_pdrop", "rotary_dim", "scale_attn", "self", "sqrt", "super", "sure", "the", "this", "to", "used", "warning_once", "when", "will", "without"], "codegen/modeling_codegen.py:CodeGenAttention._split_heads": ["ModelAttention", "Module", "_split_heads", "class", "def", "dim_head", "mp_num", "n_head", "nn", "reshape", "reshaped", "return", "self", "shape", "x"], "codegen/modeling_codegen.py:CodeGenAttention._merge_heads": ["Input", "ModelAttention", "Module", "ValueError", "_merge_heads", "attn_head_size", "be", "but", "class", "contiguous", "def", "elif", "else", "f", "if", "is", "len", "new_shape", "nn", "num_attention_heads", "of", "one", "permute", "raise", "rank", "return", "self", "shape", "should", "size", "tensor", "view"], "codegen/modeling_codegen.py:CodeGenAttention._attn": ["ModelAttention", "Module", "None", "Softmax", "_attn", "attention_mask", "attn_dropout", "attn_output", "attn_weights", "causal_mask", "class", "def", "dim", "dtype", "float32", "if", "is", "key", "matmul", "nn", "not", "query", "return", "scale_attn", "self", "shape", "to", "torch", "transpose", "value"], "codegen/modeling_codegen.py:CodeGenAttention.forward": ["Cache", "False", "FloatTensor", "LongTensor", "ModelAttention", "Module", "None", "Tensor", "_attn", "_merge_heads", "_split_heads", "apply_rotary_pos_emb", "attention_mask", "attn_output", "attn_weights", "bool", "cache_kwargs", "cache_position", "cat", "class", "cos", "def", "device", "dim", "dtype", "else", "embed_positions", "forward", "head_dim", "hidden_states", "if", "is", "k_pass", "k_rot", "key", "layer_idx", "layer_past", "local_dim", "mp_num", "nn", "not", "num_attention_heads", "out_proj", "output_attentions", "partial_rotation_size", "permute", "position_ids", "q_pass", "q_rot", "qkv", "qkv_proj", "qkv_split", "query", "reshape", "resid_dropout", "return", "rotary_dim", "self", "shape", "sin", "sincos", "split", "to", "torch", "tuple", "update", "use_cache", "value"], "codegen/modeling_codegen.py:CodeGenMLP.__init__": ["ACT2FN", "Dropout", "Linear", "ModelMLP", "Module", "__init__", "act", "activation_function", "class", "config", "def", "dropout", "embed_dim", "fc_in", "fc_out", "intermediate_size", "n_embd", "nn", "resid_pdrop", "self", "super"], "codegen/modeling_codegen.py:CodeGenMLP.forward": ["FloatTensor", "ModelMLP", "Module", "None", "act", "class", "def", "dropout", "fc_in", "fc_out", "forward", "hidden_states", "nn", "return", "self", "torch"], "codegen/modeling_codegen.py:CodeGenBlock.__init__": ["GradientCheckpointingLayer", "LayerNorm", "ModelAttention", "ModelBlock", "ModelMLP", "None", "__init__", "attn", "class", "config", "def", "else", "eps", "if", "inner_dim", "is", "layer_idx", "layer_norm_epsilon", "ln_1", "mlp", "n_embd", "n_inner", "nn", "not", "self", "super"], "codegen/modeling_codegen.py:CodeGenBlock.forward": ["Cache", "False", "FloatTensor", "GradientCheckpointingLayer", "LongTensor", "ModelBlock", "None", "Tensor", "attention_mask", "attn", "attn_outputs", "attn_weights", "bool", "cache_position", "class", "def", "feed_forward_hidden_states", "forward", "hidden_states", "layer_past", "ln_1", "mlp", "output_attentions", "position_ids", "residual", "return", "self", "torch", "tuple", "use_cache"], "codegen/modeling_codegen.py:CodeGenPreTrainedModel._init_weights": ["ModelAttention", "ModelPreTrainedModel", "PreTrainedModel", "_init_weights", "class", "copy_", "create_sinusoidal_positions", "def", "embed_positions", "if", "init", "isinstance", "max_positions", "module", "pos_embd_dim", "self", "super"], "codegen/modeling_codegen.py:CodeGenModel.__init__": ["Dropout", "Embedding", "False", "LayerNorm", "ModelBlock", "ModelModel", "ModelPreTrainedModel", "ModuleList", "__init__", "class", "config", "def", "drop", "embd_pdrop", "embed_dim", "eps", "for", "gradient_checkpointing", "h", "i", "in", "layer_idx", "layer_norm_epsilon", "ln_f", "min", "n_ctx", "n_embd", "n_layer", "nn", "num_attention_heads", "post_init", "range", "rotary_dim", "self", "super", "vocab_size", "wte"], "codegen/modeling_codegen.py:CodeGenModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "return", "self", "wte"], "codegen/modeling_codegen.py:CodeGenModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "new_embeddings", "self", "set_input_embeddings", "wte"], "codegen/modeling_codegen.py:CodeGenModel.forward": ["BaseModelOutputWithPast", "Cache", "DynamicCache", "False", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Setting", "True", "ValueError", "You", "_update_causal_mask", "all_hidden_states", "all_self_attentions", "and", "arange", "attention_mask", "attentions", "block", "bool", "cache_position", "causal_mask", "checkpointing", "class", "config", "def", "device", "drop", "else", "enumerate", "exactly", "for", "forward", "get_seq_length", "gradient", "gradient_checkpointing", "h", "hidden_states", "i", "if", "in", "incompatible", "input_ids", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layer_past", "ln_f", "logger", "must", "not", "of", "one", "or", "output_attentions", "output_hidden_states", "output_shape", "outputs", "past_key_values", "past_seen_tokens", "position_ids", "r", "raise", "return", "return_dict", "self", "seq_length", "shape", "size", "specify", "token_type_embeds", "token_type_ids", "torch", "training", "tuple", "unsqueeze", "use_cache", "use_return_dict", "v", "view", "warning_once", "with", "wte"], "codegen/modeling_codegen.py:CodeGenModel._update_causal_mask": ["AttentionMaskConverter", "BlockMask", "Cache", "False", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "Union", "_attn_implementation", "_ignore_causal_mask_sdpa", "_prepare_4d_causal_attention_mask_with_cache_position", "_unmask_unattended", "_update_causal_mask", "and", "any", "attention_mask", "batch_size", "bool", "cache_position", "causal_mask", "class", "config", "cuda", "def", "device", "dtype", "else", "finfo", "flash_attention_2", "flex_attention", "get_max_cache_shape", "get_seq_length", "if", "in", "input_tensor", "inputs_embeds", "is", "is_compileable", "is_training", "isinstance", "make_flex_block_causal_mask", "min", "min_dtype", "not", "npu", "output_attentions", "past_key_values", "past_key_values_length", "past_seen_tokens", "return", "sdpa", "self", "sequence_length", "shape", "target_length", "torch", "training", "type", "using_compilable_cache", "xpu"], "codegen/modeling_codegen.py:CodeGenModel._prepare_4d_causal_attention_mask_with_cache_position": ["ModelModel", "ModelPreTrainedModel", "None", "Tensor", "_prepare_4d_causal_attention_mask_with_cache_position", "and", "arange", "attention_mask", "batch_size", "cache_position", "causal_mask", "class", "clone", "def", "device", "diagonal", "dim", "dtype", "else", "expand", "fill_value", "finfo", "full", "if", "int", "is", "kwargs", "mask_length", "masked_fill", "min", "min_dtype", "not", "padding_mask", "reshape", "return", "sequence_length", "shape", "target_length", "to", "torch", "triu"], "codegen/modeling_codegen.py:CodeGenForCausalLM.__init__": ["GenerationMixin", "Linear", "ModelForCausalLM", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "config", "def", "lm_head", "n_embd", "nn", "post_init", "self", "super", "transformer", "vocab_size"], "codegen/modeling_codegen.py:CodeGenForCausalLM.forward": ["Cache", "CausalLMOutputWithPast", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForCausalLM", "ModelPreTrainedModel", "None", "Tensor", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "not", "output", "output_attentions", "output_hidden_states", "past_key_values", "position_ids", "r", "return", "return_dict", "self", "slice", "slice_indices", "token_type_ids", "torch", "transformer", "transformer_outputs", "tuple", "use_cache", "use_return_dict", "vocab_size"], "cohere/modeling_cohere.py:CohereLayerNorm.__init__": ["False", "ModelLayerNorm", "Module", "None", "Parameter", "__init__", "bias", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "cohere/modeling_cohere.py:CohereLayerNorm.forward": ["ModelLayerNorm", "Module", "True", "class", "def", "dtype", "float32", "forward", "hidden_states", "input_dtype", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "cohere/modeling_cohere.py:CohereRotaryEmbedding.__init__": ["Callable", "False", "ModelConfig", "ModelRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "cohere/modeling_cohere.py:CohereRotaryEmbedding.compute_default_rope_parameters": ["ModelConfig", "ModelRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "cohere/modeling_cohere.py:CohereRotaryEmbedding.forward": ["False", "ModelRotaryEmbedding", "Module", "None", "and", "attention_scaling", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "repeat_interleave", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "cohere/modeling_cohere.py:CohereMLP.__init__": ["ACT2FN", "False", "Linear", "ModelMLP", "Module", "__init__", "act_fn", "bias", "class", "config", "def", "down_proj", "gate_proj", "hidden_act", "hidden_size", "intermediate_size", "nn", "self", "super", "up_proj"], "cohere/modeling_cohere.py:CohereMLP.forward": ["ModelMLP", "Module", "act_fn", "class", "def", "down_proj", "forward", "gate_proj", "nn", "return", "self", "up_proj", "x"], "cohere/modeling_cohere.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "cohere/modeling_cohere.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value", "value_states"], "cohere/modeling_cohere.py:rotate_half": ["Model_half", "def", "dim", "flatten", "return", "rot_x", "stack", "torch", "x", "x1", "x2"], "cohere/modeling_cohere.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cos", "def", "dtype", "float", "k", "k_embed", "q", "q_embed", "return", "rotate_half", "sin", "to", "unsqueeze", "unsqueeze_dim"], "cohere/modeling_cohere.py:CohereAttention.__init__": ["Linear", "ModelAttention", "ModelConfig", "ModelLayerNorm", "Module", "None", "True", "__init__", "attention_bias", "attention_dropout", "bias", "class", "config", "def", "eps", "getattr", "head_dim", "hidden_size", "if", "int", "is_causal", "k_norm", "k_proj", "layer_idx", "layer_norm_eps", "nn", "num_attention_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "q_norm", "q_proj", "scaling", "self", "super", "use_qk_norm", "v_proj"], "cohere/modeling_cohere.py:CohereAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "FlashAttentionKwargs", "LongTensor", "ModelAttention", "Module", "None", "Tensor", "Unpack", "_attn_implementation", "apply_rotary_pos_emb", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "cache_kwargs", "cache_position", "class", "config", "contiguous", "cos", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "head_dim", "hidden_shape", "hidden_states", "if", "input_shape", "is", "k_norm", "k_proj", "key_states", "kwargs", "layer_idx", "nn", "not", "o_proj", "past_key_values", "position_embeddings", "q_norm", "q_proj", "query_states", "reshape", "return", "scaling", "self", "shape", "sin", "torch", "training", "transpose", "tuple", "update", "use_qk_norm", "v_proj", "value_states", "view"], "cohere/modeling_cohere.py:CohereDecoderLayer.__init__": ["GradientCheckpointingLayer", "ModelAttention", "ModelConfig", "ModelDecoderLayer", "ModelLayerNorm", "ModelMLP", "__init__", "class", "config", "def", "eps", "hidden_size", "input_layernorm", "int", "layer_idx", "layer_norm_eps", "mlp", "self", "self_attn", "super"], "cohere/modeling_cohere.py:CohereDecoderLayer.forward": ["Cache", "False", "FlashAttentionKwargs", "FloatTensor", "GradientCheckpointingLayer", "LongTensor", "ModelDecoderLayer", "None", "Tensor", "Unpack", "_", "attention_mask", "bool", "cache_position", "class", "def", "forward", "hidden_states", "hidden_states_attention", "hidden_states_mlp", "input_layernorm", "kwargs", "mlp", "past_key_values", "position_embeddings", "position_ids", "residual", "return", "self", "self_attn", "torch", "tuple", "use_cache"], "cohere/modeling_cohere.py:CohereModel.__init__": ["Embedding", "False", "ModelConfig", "ModelDecoderLayer", "ModelLayerNorm", "ModelModel", "ModelPreTrainedModel", "ModelRotaryEmbedding", "ModuleList", "__init__", "class", "config", "def", "embed_tokens", "eps", "for", "gradient_checkpointing", "hidden_size", "in", "layer_idx", "layer_norm_eps", "layers", "nn", "norm", "num_hidden_layers", "pad_token_id", "padding_idx", "post_init", "range", "rotary_emb", "self", "super", "vocab_size"], "cohere/modeling_cohere.py:CohereModel.forward": ["BaseModelOutputWithPast", "Cache", "DynamicCache", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "and", "arange", "attention_mask", "bool", "cache_position", "causal_mask", "class", "config", "create_causal_mask", "decoder_layer", "def", "device", "else", "embed_tokens", "exactly", "for", "forward", "get_seq_length", "hidden_states", "if", "in", "input_embeds", "input_ids", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layers", "must", "norm", "not", "num_hidden_layers", "of", "one", "or", "past_key_values", "past_seen_tokens", "position_embeddings", "position_ids", "raise", "return", "rotary_emb", "self", "shape", "specify", "torch", "unsqueeze", "use_cache"], "cohere/modeling_cohere.py:CohereForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "ModelForCausalLM", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "logit_scale", "model", "nn", "post_init", "self", "super", "tie_word_embeddings", "vocab_size"], "cohere/modeling_cohere.py:CohereForCausalLM.forward": ["BaseModelOutputWithPast", "Cache", "CausalLMOutputWithPast", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForCausalLM", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "logit_scale", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "position_ids", "r", "return", "self", "slice", "slice_indices", "torch", "use_cache", "vocab_size"], "cohere2/modeling_cohere2.py:Cohere2RotaryEmbedding.__init__": ["Callable", "False", "ModelConfig", "ModelRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "cohere2/modeling_cohere2.py:Cohere2RotaryEmbedding.compute_default_rope_parameters": ["ModelConfig", "ModelRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "cohere2/modeling_cohere2.py:Cohere2RotaryEmbedding.forward": ["False", "ModelRotaryEmbedding", "Module", "None", "and", "attention_scaling", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "repeat_interleave", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "cohere2/modeling_cohere2.py:Cohere2LayerNorm.__init__": ["False", "ModelLayerNorm", "Module", "None", "Parameter", "__init__", "bias", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "cohere2/modeling_cohere2.py:Cohere2LayerNorm.forward": ["ModelLayerNorm", "Module", "True", "class", "def", "dtype", "float32", "forward", "hidden_states", "input_dtype", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "cohere2/modeling_cohere2.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "cohere2/modeling_cohere2.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value", "value_states"], "cohere2/modeling_cohere2.py:rotate_half": ["Model_half", "def", "dim", "flatten", "return", "rot_x", "stack", "torch", "x", "x1", "x2"], "cohere2/modeling_cohere2.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cos", "def", "dtype", "float", "k", "k_embed", "q", "q_embed", "return", "rotate_half", "sin", "to", "unsqueeze", "unsqueeze_dim"], "cohere2/modeling_cohere2.py:Cohere2Attention.__init__": ["Linear", "ModelAttention", "ModelConfig", "Module", "None", "True", "__init__", "attention_bias", "attention_dropout", "bias", "class", "config", "def", "else", "getattr", "hasattr", "head_dim", "hidden_size", "if", "int", "is_causal", "k_proj", "layer_idx", "layer_type", "layer_types", "nn", "num_attention_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "q_proj", "scaling", "self", "sliding_attention", "sliding_window", "super", "v_proj"], "cohere2/modeling_cohere2.py:Cohere2Attention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "LongTensor", "ModelAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_attn_implementation", "apply_rotary_pos_emb", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "cache_kwargs", "cache_position", "class", "config", "contiguous", "cos", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "head_dim", "hidden_shape", "hidden_states", "if", "input_shape", "is", "k_proj", "key_states", "kwargs", "layer_idx", "nn", "not", "o_proj", "past_key_values", "position_embeddings", "q_proj", "query_states", "reshape", "return", "scaling", "self", "shape", "sin", "sliding_window", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "view"], "cohere2/modeling_cohere2.py:Cohere2MLP.__init__": ["ACT2FN", "False", "Linear", "ModelMLP", "Module", "__init__", "act_fn", "bias", "class", "config", "def", "down_proj", "gate_proj", "hidden_act", "hidden_size", "intermediate_size", "nn", "self", "super", "up_proj"], "cohere2/modeling_cohere2.py:Cohere2MLP.forward": ["ModelMLP", "Module", "act_fn", "class", "def", "down_proj", "forward", "gate_proj", "nn", "return", "self", "up_proj", "x"], "cohere2/modeling_cohere2.py:Cohere2DecoderLayer.__init__": ["GradientCheckpointingLayer", "ModelAttention", "ModelConfig", "ModelDecoderLayer", "ModelLayerNorm", "ModelMLP", "__init__", "attention_type", "class", "config", "def", "eps", "hidden_size", "input_layernorm", "int", "layer_idx", "layer_norm_eps", "layer_types", "mlp", "self", "self_attn", "super"], "cohere2/modeling_cohere2.py:Cohere2DecoderLayer.forward": ["Cache", "False", "FloatTensor", "GradientCheckpointingLayer", "LongTensor", "ModelDecoderLayer", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "attention_mask", "bool", "cache_position", "class", "def", "forward", "hidden_states", "hidden_states_attention", "hidden_states_mlp", "input_layernorm", "kwargs", "mlp", "past_key_values", "position_embeddings", "residual", "return", "self", "self_attn", "torch", "tuple", "use_cache"], "cohere2/modeling_cohere2.py:Cohere2Model.__init__": ["Embedding", "False", "ModelConfig", "ModelDecoderLayer", "ModelLayerNorm", "ModelModel", "ModelPreTrainedModel", "ModelRotaryEmbedding", "ModuleList", "__init__", "class", "config", "def", "embed_tokens", "eps", "for", "gradient_checkpointing", "hidden_size", "in", "layer_idx", "layer_norm_eps", "layers", "nn", "norm", "num_hidden_layers", "pad_token_id", "padding_idx", "post_init", "range", "rotary_emb", "self", "super", "vocab_size"], "cohere2/modeling_cohere2.py:Cohere2Model.forward": ["BaseModelOutputWithPast", "Cache", "DynamicCache", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "and", "arange", "attention_mask", "attention_type", "bool", "cache_position", "causal_mask_mapping", "class", "config", "create_causal_mask", "create_sliding_window_causal_mask", "decoder_layer", "def", "device", "dict", "else", "embed_tokens", "exactly", "for", "forward", "full_attention", "get_seq_length", "hidden_states", "if", "in", "input_embeds", "input_ids", "inputs_embeds", "is", "isinstance", "kwargs", "last_hidden_state", "layers", "mask_kwargs", "must", "norm", "not", "of", "one", "or", "past_key_values", "past_seen_tokens", "position_embeddings", "position_ids", "raise", "return", "rotary_emb", "self", "shape", "sliding_attention", "specify", "torch", "unsqueeze", "use_cache"], "cohere2/modeling_cohere2.py:Cohere2ForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "ModelForCausalLM", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "logit_scale", "model", "nn", "post_init", "self", "super", "tie_word_embeddings", "vocab_size"], "cohere2/modeling_cohere2.py:Cohere2ForCausalLM.forward": ["BaseModelOutputWithPast", "Cache", "CausalLMOutputWithPast", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForCausalLM", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "logit_scale", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "position_ids", "r", "return", "self", "slice", "slice_indices", "torch", "use_cache", "vocab_size"], "cohere2_vision/modeling_cohere2_vision.py:Cohere2VisionMultiModalProjector.__init__": ["Linear", "ModelConfig", "ModelMultiModalProjector", "Module", "SiLU", "True", "__init__", "act", "alignment_intermediate_size", "bias", "class", "config", "def", "downsample_factor", "hidden_size", "intermediate_size", "linear_1", "linear_2", "nn", "self", "super", "text_config", "vision_config"], "cohere2_vision/modeling_cohere2_vision.py:Cohere2VisionMultiModalProjector.pixel_shuffle": ["ModelMultiModalProjector", "Module", "batch_size", "channels", "class", "def", "downsample_factor", "feature_dim", "height", "image_features", "int", "nn", "permute", "pixel_shuffle", "reshape", "return", "self", "seq_length", "shape", "width"], "cohere2_vision/modeling_cohere2_vision.py:Cohere2VisionMultiModalProjector.forward": ["ModelMultiModalProjector", "Module", "act", "chunk", "class", "def", "dim", "forward", "gate", "hidden_states", "image_features", "linear_1", "linear_2", "nn", "pixel_shuffle", "return", "self", "x"], "cohere2_vision/modeling_cohere2_vision.py:Cohere2VisionModel.__init__": ["AutoModel", "ModelConfig", "ModelModel", "ModelMultiModalProjector", "ModelPreTrainedModel", "__init__", "class", "config", "def", "from_config", "language_model", "multi_modal_projector", "post_init", "self", "super", "text_config", "vision_config", "vision_tower"], "cohere2_vision/modeling_cohere2_vision.py:Cohere2VisionModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "language_model", "return", "self"], "cohere2_vision/modeling_cohere2_vision.py:Cohere2VisionModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "language_model", "self", "set_input_embeddings", "value"], "cohere2_vision/modeling_cohere2_vision.py:Cohere2VisionModel.get_image_features": ["FloatTensor", "ModelModel", "ModelPreTrainedModel", "True", "class", "def", "get_image_features", "image_features", "last_hidden_state", "multi_modal_projector", "output_hidden_states", "pixel_values", "return", "selected_image_feature", "self", "torch", "vision_tower"], "cohere2_vision/modeling_cohere2_vision.py:Cohere2VisionModel.get_placeholder_mask": ["FloatTensor", "Image", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "ValueError", "all", "and", "class", "config", "def", "device", "do", "dtype", "else", "expand_as", "f", "features", "get_input_embeddings", "get_placeholder_mask", "if", "image", "image_features", "image_token_id", "input_ids", "inputs_embeds", "is", "long", "match", "n_image_features", "n_image_tokens", "not", "numel", "raise", "return", "self", "shape", "special_image_mask", "sum", "tensor", "to", "tokens", "torch", "unsqueeze"], "cohere2_vision/modeling_cohere2_vision.py:Cohere2VisionModel.forward": ["Cache", "FlashAttentionKwargs", "FloatTensor", "LongTensor", "ModelModel", "ModelModelOutputWithPast", "ModelPreTrainedModel", "None", "Tensor", "Unpack", "ValueError", "You", "attention_mask", "attentions", "bool", "cache_position", "class", "def", "device", "dtype", "else", "exactly", "forward", "get_image_features", "get_input_embeddings", "get_placeholder_mask", "hidden_states", "if", "image_features", "image_hidden_states", "input_ids", "inputs_embeds", "is", "kwargs", "language_model", "last_hidden_state", "masked_scatter", "must", "not", "of", "one", "or", "outputs", "past_key_values", "pixel_values", "position_ids", "raise", "return", "self", "special_image_mask", "specify", "to", "torch", "tuple", "use_cache"], "cohere2_vision/modeling_cohere2_vision.py:Cohere2VisionForConditionalGeneration.__init__": ["False", "GenerationMixin", "Linear", "ModelConfig", "ModelForConditionalGeneration", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "text_config", "vocab_size"], "cohere2_vision/modeling_cohere2_vision.py:Cohere2VisionForConditionalGeneration.get_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "model", "return", "self"], "cohere2_vision/modeling_cohere2_vision.py:Cohere2VisionForConditionalGeneration.set_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "model", "self", "set_input_embeddings", "value"], "cohere2_vision/modeling_cohere2_vision.py:Cohere2VisionForConditionalGeneration.get_output_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "Module", "class", "def", "get_output_embeddings", "lm_head", "nn", "return", "self"], "cohere2_vision/modeling_cohere2_vision.py:Cohere2VisionForConditionalGeneration.get_image_features": ["FloatTensor", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "get_image_features", "model", "pixel_values", "return", "self", "torch"], "cohere2_vision/modeling_cohere2_vision.py:Cohere2VisionForConditionalGeneration.forward": ["Cache", "FloatTensor", "GenerationMixin", "LongTensor", "ModelCausalLMOutputWithPast", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "else", "forward", "hidden_states", "if", "image_hidden_states", "image_sizes", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "outputs", "past_key_values", "pixel_values", "position_ids", "r", "return", "self", "slice", "slice_indices", "text_config", "torch", "tuple", "use_cache", "vocab_size"], "cohere2_vision/modeling_cohere2_vision.py:Cohere2VisionForConditionalGeneration.prepare_inputs_for_generation": ["False", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "True", "attention_mask", "cache_position", "class", "def", "get", "if", "input_ids", "inputs_embeds", "is_first_iteration", "kwargs", "logits_to_keep", "model_inputs", "not", "or", "past_key_values", "pixel_values", "prepare_inputs_for_generation", "return", "self", "super", "use_cache"], "colpali/modeling_colpali.py:ColPaliPreTrainedModel._init_weights": ["Conv2d", "Embedding", "False", "Linear", "ModelPreTrainedModel", "None", "PreTrainedModel", "_init_weights", "_is_hf_initialized", "and", "bias", "class", "config", "def", "elif", "else", "getattr", "hasattr", "if", "init", "initializer_range", "is", "isinstance", "mean", "module", "nn", "normal_", "not", "padding_idx", "self", "std", "text_config", "vlm_config", "weight", "zeros_"], "colpali/modeling_colpali.py:ColPaliForRetrieval.__init__": ["AutoModelForImageTextToText", "Linear", "ModelConfig", "ModelForRetrieval", "ModelPreTrainedModel", "__init__", "class", "config", "def", "embedding_dim", "embedding_proj_layer", "from_config", "hidden_size", "nn", "post_init", "self", "super", "text_config", "vlm", "vlm_config", "vocab_size"], "colpali/modeling_colpali.py:ColPaliForRetrieval.forward": ["FloatTensor", "LongTensor", "ModelForRetrieval", "ModelForRetrievalOutput", "ModelPreTrainedModel", "None", "Tensor", "True", "attention_mask", "attentions", "bool", "class", "config", "def", "dim", "dtype", "else", "embedding_proj_layer", "embeddings", "forward", "hidden_states", "if", "image_hidden_states", "input_ids", "is", "keepdim", "kwargs", "last_hidden_states", "model", "norm", "not", "output_attentions", "output_hidden_states", "past_key_values", "pixel_values", "proj_dtype", "return", "return_dict", "self", "to", "torch", "unsqueeze", "use_return_dict", "vlm", "vlm_hidden_states", "vlm_image_hidden_states", "vlm_output", "weight"], "colpali/modeling_colpali.py:ColPaliForRetrieval.get_input_embeddings": ["ModelForRetrieval", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "return", "self", "vlm"], "colpali/modeling_colpali.py:ColPaliForRetrieval.set_input_embeddings": ["ModelForRetrieval", "ModelPreTrainedModel", "class", "def", "self", "set_input_embeddings", "value", "vlm"], "colpali/modeling_colpali.py:ColPaliForRetrieval.get_output_embeddings": ["ModelForRetrieval", "ModelPreTrainedModel", "class", "def", "get_output_embeddings", "return", "self", "vlm"], "colpali/modeling_colpali.py:ColPaliForRetrieval.set_output_embeddings": ["ModelForRetrieval", "ModelPreTrainedModel", "class", "def", "new_embeddings", "self", "set_output_embeddings", "vlm"], "colpali/modeling_colpali.py:ColPaliForRetrieval.resize_token_embeddings": ["Embedding", "ModelForRetrieval", "ModelPreTrainedModel", "None", "True", "bool", "class", "config", "def", "int", "mean_resizing", "model_embeds", "new_num_tokens", "nn", "num_embeddings", "pad_to_multiple_of", "resize_token_embeddings", "return", "self", "text_config", "vlm", "vlm_config", "vocab_size"], "colqwen2/modeling_colqwen2.py:ColQwen2PreTrainedModel._init_weights": ["Conv2d", "Embedding", "False", "Linear", "ModelPreTrainedModel", "None", "PreTrainedModel", "_init_weights", "_is_hf_initialized", "and", "bias", "class", "config", "def", "elif", "else", "getattr", "hasattr", "if", "init", "initializer_range", "is", "isinstance", "mean", "module", "nn", "normal_", "not", "padding_idx", "self", "std", "text_config", "vlm_config", "weight", "zeros_"], "colqwen2/modeling_colqwen2.py:ColQwen2ForRetrieval.__init__": ["AutoModelForImageTextToText", "Linear", "ModelConfig", "ModelForRetrieval", "ModelPreTrainedModel", "__init__", "class", "config", "def", "embedding_dim", "embedding_proj_layer", "from_config", "hidden_size", "nn", "post_init", "self", "super", "text_config", "vlm", "vlm_config", "vocab_size"], "colqwen2/modeling_colqwen2.py:ColQwen2ForRetrieval.forward": ["Cache", "FloatTensor", "LongTensor", "ModelForRetrieval", "ModelForRetrievalOutput", "ModelPreTrainedModel", "None", "Tensor", "True", "and", "attention_mask", "attentions", "bool", "cache_position", "cat", "class", "config", "def", "device", "dim", "dtype", "else", "embedding_proj_layer", "embeddings", "expand_as", "for", "forward", "get_input_embeddings", "get_rope_index", "grid_thw", "hidden_states", "if", "image_embeds", "image_grid_thw", "image_mask", "image_token_id", "in", "input_ids", "inputs_embeds", "is", "keepdim", "kwargs", "labels", "last_hidden_states", "masked_scatter", "model", "norm", "not", "offset", "offsets", "output_attentions", "output_hidden_states", "past_key_values", "pixel_sequence", "pixel_values", "position_ids", "proj_dtype", "r", "return", "return_dict", "rope_deltas", "self", "to", "torch", "unsqueeze", "use_cache", "use_return_dict", "video_grid_thw", "visual", "vlm", "vlm_config", "vlm_hidden_states", "vlm_output", "weight", "zip"], "colqwen2/modeling_colqwen2.py:ColQwen2ForRetrieval.get_input_embeddings": ["ModelForRetrieval", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "return", "self", "vlm"], "colqwen2/modeling_colqwen2.py:ColQwen2ForRetrieval.set_input_embeddings": ["ModelForRetrieval", "ModelPreTrainedModel", "class", "def", "self", "set_input_embeddings", "value", "vlm"], "colqwen2/modeling_colqwen2.py:ColQwen2ForRetrieval.get_output_embeddings": ["ModelForRetrieval", "ModelPreTrainedModel", "class", "def", "get_output_embeddings", "return", "self", "vlm"], "colqwen2/modeling_colqwen2.py:ColQwen2ForRetrieval.set_output_embeddings": ["ModelForRetrieval", "ModelPreTrainedModel", "class", "def", "new_embeddings", "self", "set_output_embeddings", "vlm"], "colqwen2/modeling_colqwen2.py:ColQwen2ForRetrieval.resize_token_embeddings": ["Embedding", "ModelForRetrieval", "ModelPreTrainedModel", "None", "True", "bool", "class", "config", "def", "int", "mean_resizing", "model_embeds", "new_num_tokens", "nn", "num_embeddings", "pad_to_multiple_of", "resize_token_embeddings", "return", "self", "text_config", "vlm", "vlm_config", "vocab_size"], "conditional_detr/modeling_conditional_detr.py:ConditionalDetrFrozenBatchNorm2d.__init__": ["ModelFrozenBatchNorm2d", "Module", "__init__", "bias", "class", "def", "n", "nn", "ones", "register_buffer", "running_mean", "running_var", "self", "super", "torch", "weight", "zeros"], "conditional_detr/modeling_conditional_detr.py:ConditionalDetrFrozenBatchNorm2d._load_from_state_dict": ["ModelFrozenBatchNorm2d", "Module", "_load_from_state_dict", "class", "def", "del", "error_msgs", "if", "in", "local_metadata", "missing_keys", "nn", "num_batches_tracked", "num_batches_tracked_key", "prefix", "self", "state_dict", "strict", "super", "unexpected_keys"], "conditional_detr/modeling_conditional_detr.py:ConditionalDetrFrozenBatchNorm2d.forward": ["ModelFrozenBatchNorm2d", "Module", "bias", "class", "def", "epsilon", "forward", "nn", "reshape", "return", "rsqrt", "running_mean", "running_var", "scale", "self", "weight", "x"], "conditional_detr/modeling_conditional_detr.py:replace_batch_norm": ["BatchNorm2d", "ModelFrozenBatchNorm2d", "Model_batch_norm", "_modules", "bias", "children", "copy_", "def", "device", "for", "if", "in", "isinstance", "len", "list", "meta", "model", "module", "name", "named_children", "new_module", "nn", "num_features", "r", "running_mean", "running_var", "torch", "weight"], "conditional_detr/modeling_conditional_detr.py:ConditionalDetrConvEncoder.__init__": ["Either", "False", "ModelConvEncoder", "Module", "None", "True", "ValueError", "__init__", "and", "backbone", "backbone_config", "backbone_kwargs", "backbone_model_type", "be", "channels", "class", "config", "copy", "create_model", "def", "dilation", "elif", "else", "feature_info", "features_only", "for", "get", "getattr", "if", "in", "in_chans", "intermediate_channel_sizes", "is", "kwargs", "layer2", "layer3", "layer4", "load_backbone", "model", "model_type", "name", "named_parameters", "nn", "no_grad", "not", "num_channels", "or", "out_indices", "output_stride", "parameter", "pop", "pretrained", "provided", "raise", "replace_batch_norm", "requires_backends", "requires_grad_", "resnet", "self", "should", "stage", "super", "the", "timm", "torch", "use_pretrained_backbone", "use_timm_backbone", "with"], "conditional_detr/modeling_conditional_detr.py:ConditionalDetrConvEncoder.forward": ["ModelConvEncoder", "Module", "None", "Tensor", "append", "bool", "class", "config", "def", "else", "feature_map", "feature_maps", "features", "float", "for", "forward", "functional", "if", "in", "interpolate", "mask", "model", "nn", "out", "pixel_mask", "pixel_values", "return", "self", "shape", "size", "to", "torch", "use_timm_backbone"], "conditional_detr/modeling_conditional_detr.py:ConditionalDetrConvModel.__init__": ["ModelConvModel", "Module", "__init__", "class", "conv_encoder", "def", "nn", "position_embedding", "self", "super"], "conditional_detr/modeling_conditional_detr.py:ConditionalDetrConvModel.forward": ["ModelConvModel", "Module", "append", "class", "conv_encoder", "def", "dtype", "feature_map", "for", "forward", "in", "mask", "nn", "out", "pixel_mask", "pixel_values", "pos", "position_embedding", "return", "self", "to"], "conditional_detr/modeling_conditional_detr.py:ConditionalDetrSinePositionEmbedding.__init__": ["False", "ModelSinePositionEmbedding", "Module", "None", "True", "ValueError", "__init__", "and", "be", "class", "def", "embedding_dim", "if", "is", "math", "nn", "normalize", "not", "passed", "pi", "raise", "scale", "self", "should", "super", "temperature"], "conditional_detr/modeling_conditional_detr.py:ConditionalDetrSinePositionEmbedding.forward": ["ModelSinePositionEmbedding", "Module", "No", "None", "ValueError", "arange", "cat", "class", "cos", "cumsum", "def", "device", "dim", "dim_t", "div", "dtype", "embedding_dim", "flatten", "float", "float32", "floor", "forward", "if", "int64", "is", "mask", "nn", "normalize", "permute", "pixel", "pixel_mask", "pixel_values", "pos", "pos_x", "pos_y", "provided", "raise", "return", "rounding_mode", "scale", "self", "sin", "stack", "temperature", "torch", "x_embed", "y_embed"], "conditional_detr/modeling_conditional_detr.py:ConditionalDetrLearnedPositionEmbedding.__init__": ["Embedding", "ModelLearnedPositionEmbedding", "Module", "__init__", "class", "column_embeddings", "def", "embedding_dim", "nn", "row_embeddings", "self", "super"], "conditional_detr/modeling_conditional_detr.py:ConditionalDetrLearnedPositionEmbedding.forward": ["ModelLearnedPositionEmbedding", "Module", "None", "arange", "cat", "class", "column_embeddings", "def", "device", "dim", "forward", "height", "height_values", "nn", "permute", "pixel_mask", "pixel_values", "pos", "repeat", "return", "row_embeddings", "self", "shape", "torch", "unsqueeze", "width", "width_values", "x_emb", "y_emb"], "conditional_detr/modeling_conditional_detr.py:build_position_encoding": ["ModelLearnedPositionEmbedding", "ModelSinePositionEmbedding", "Model_position_encoding", "Not", "True", "ValueError", "config", "d_model", "def", "elif", "else", "f", "if", "learned", "n_steps", "normalize", "position_embedding", "position_embedding_type", "raise", "return", "sine", "supported"], "conditional_detr/modeling_conditional_detr.py:gen_sine_position_embeddings": ["Model_sine_position_embeddings", "None", "arange", "cat", "cos", "d_model", "def", "device", "dim", "dim_t", "div", "dtype", "flatten", "float32", "floor", "math", "pi", "pos", "pos_tensor", "pos_x", "pos_y", "return", "rounding_mode", "scale", "sin", "stack", "to", "torch", "x_embed", "y_embed"], "conditional_detr/modeling_conditional_detr.py:inverse_sigmoid": ["Model_sigmoid", "clamp", "def", "eps", "log", "max", "min", "return", "torch", "x", "x1", "x2"], "conditional_detr/modeling_conditional_detr.py:DetrAttention.__init__": ["Linear", "ModelAttention", "Module", "True", "ValueError", "__init__", "and", "be", "bias", "bool", "by", "class", "def", "divisible", "dropout", "embed_dim", "f", "float", "got", "head_dim", "if", "int", "k_proj", "must", "nn", "num_heads", "out_proj", "q_proj", "raise", "scaling", "self", "super", "v_proj"], "conditional_detr/modeling_conditional_detr.py:DetrAttention._shape": ["ModelAttention", "Module", "Tensor", "_shape", "batch_size", "class", "contiguous", "def", "head_dim", "int", "nn", "num_heads", "return", "self", "seq_len", "tensor", "torch", "transpose", "view"], "conditional_detr/modeling_conditional_detr.py:DetrAttention.with_pos_embed": ["ModelAttention", "Module", "None", "Tensor", "class", "def", "else", "if", "is", "nn", "object_queries", "return", "self", "tensor", "torch", "with_pos_embed"], "conditional_detr/modeling_conditional_detr.py:DetrAttention.forward": ["Attention", "False", "ModelAttention", "Module", "None", "Tensor", "ValueError", "_shape", "attention_mask", "attn_output", "attn_probs", "attn_weights", "attn_weights_reshaped", "batch_size", "be", "bmm", "bool", "but", "class", "def", "dim", "dropout", "dtype", "else", "embed_dim", "f", "forward", "functional", "head_dim", "hidden_states", "hidden_states_original", "if", "inf", "is", "is_cross_attention", "k_proj", "key_states", "key_value_states", "key_value_states_original", "mask", "masked_fill_", "nn", "not", "num_heads", "object_queries", "of", "out_proj", "output_attentions", "p", "proj_shape", "q_proj", "query_states", "raise", "reshape", "return", "scaling", "self", "should", "size", "softmax", "source_len", "spatial_position_embeddings", "target_len", "torch", "training", "transpose", "tuple", "v_proj", "value_states", "view", "weights", "with_pos_embed", "zeros_like"], "conditional_detr/modeling_conditional_detr.py:ConditionalDetrAttention.__init__": ["Linear", "ModelAttention", "Module", "True", "ValueError", "__init__", "and", "be", "bias", "bool", "by", "class", "def", "divisible", "dropout", "embed_dim", "f", "float", "got", "head_dim", "if", "int", "must", "nn", "num_heads", "out_dim", "out_proj", "raise", "scaling", "self", "super", "v_head_dim"], "conditional_detr/modeling_conditional_detr.py:ConditionalDetrAttention._qk_shape": ["ModelAttention", "Module", "Tensor", "_qk_shape", "batch_size", "class", "contiguous", "def", "head_dim", "int", "nn", "num_heads", "return", "self", "seq_len", "tensor", "torch", "transpose", "view"], "conditional_detr/modeling_conditional_detr.py:ConditionalDetrAttention._v_shape": ["ModelAttention", "Module", "Tensor", "_v_shape", "batch_size", "class", "contiguous", "def", "int", "nn", "num_heads", "return", "self", "seq_len", "tensor", "torch", "transpose", "v_head_dim", "view"], "conditional_detr/modeling_conditional_detr.py:ConditionalDetrAttention.forward": ["Attention", "False", "ModelAttention", "Module", "None", "Tensor", "ValueError", "_", "_qk_shape", "_v_shape", "attention_mask", "attn_output", "attn_probs", "attn_weights", "attn_weights_reshaped", "batch_size", "be", "bmm", "bool", "but", "class", "def", "dim", "dropout", "dtype", "else", "f", "forward", "functional", "head_dim", "hidden_states", "if", "inf", "is", "key_states", "mask", "masked_fill_", "nn", "not", "num_heads", "of", "out_dim", "out_proj", "output_attentions", "p", "proj_shape", "query_states", "raise", "reshape", "return", "scaling", "self", "should", "size", "softmax", "source_len", "target_len", "torch", "training", "transpose", "tuple", "v_head_dim", "v_proj_shape", "value_states", "view", "weights", "zeros_like"], "conditional_detr/modeling_conditional_detr.py:ConditionalDetrEncoderLayer.__init__": ["ACT2FN", "DetrAttention", "LayerNorm", "Linear", "ModelConfig", "ModelEncoderLayer", "Module", "__init__", "activation_dropout", "activation_fn", "activation_function", "attention_dropout", "class", "config", "d_model", "def", "dropout", "embed_dim", "encoder_attention_heads", "encoder_ffn_dim", "fc1", "fc2", "final_layer_norm", "nn", "num_heads", "self", "self_attn", "self_attn_layer_norm", "super"], "conditional_detr/modeling_conditional_detr.py:ConditionalDetrEncoderLayer.forward": ["False", "ModelEncoderLayer", "Module", "None", "Tensor", "activation_dropout", "activation_fn", "any", "attention_mask", "attn_weights", "bool", "clamp", "clamp_value", "class", "def", "dropout", "dtype", "fc1", "fc2", "final_layer_norm", "finfo", "forward", "functional", "hidden_states", "if", "isinf", "isnan", "max", "min", "nn", "object_queries", "or", "output_attentions", "outputs", "p", "residual", "return", "self", "self_attn", "self_attn_layer_norm", "torch", "training"], "conditional_detr/modeling_conditional_detr.py:ConditionalDetrDecoderLayer.__init__": ["ACT2FN", "GradientCheckpointingLayer", "LayerNorm", "Linear", "ModelAttention", "ModelConfig", "ModelDecoderLayer", "__init__", "activation_dropout", "activation_fn", "activation_function", "attention_dropout", "ca_kcontent_proj", "ca_kpos_proj", "ca_qcontent_proj", "ca_qpos_proj", "ca_qpos_sine_proj", "ca_v_proj", "class", "config", "d_model", "decoder_attention_heads", "decoder_ffn_dim", "def", "dropout", "embed_dim", "encoder_attn", "encoder_attn_layer_norm", "fc1", "fc2", "final_layer_norm", "nhead", "nn", "num_heads", "out_dim", "sa_kcontent_proj", "sa_kpos_proj", "sa_qcontent_proj", "sa_qpos_proj", "sa_v_proj", "self", "self_attn", "self_attn_layer_norm", "super"], "conditional_detr/modeling_conditional_detr.py:ConditionalDetrDecoderLayer.forward": ["False", "GradientCheckpointingLayer", "ModelDecoderLayer", "None", "Tensor", "_", "activation_dropout", "activation_fn", "attention_mask", "batch_size", "bool", "ca_kcontent_proj", "ca_kpos_proj", "ca_qcontent_proj", "ca_qpos_proj", "ca_qpos_sine_proj", "ca_v_proj", "cat", "class", "cross_attn_weights", "def", "dim", "dropout", "else", "encoder_attention_mask", "encoder_attn", "encoder_attn_layer_norm", "encoder_hidden_states", "fc1", "fc2", "final_layer_norm", "forward", "functional", "hidden_states", "if", "is", "is_first", "k", "k_content", "k_pos", "key_states", "n_model", "nhead", "nn", "not", "num_queries", "object_queries", "output_attentions", "outputs", "p", "q", "q_content", "q_pos", "query_position_embeddings", "query_sine_embed", "residual", "return", "sa_kcontent_proj", "sa_kpos_proj", "sa_qcontent_proj", "sa_qpos_proj", "sa_v_proj", "self", "self_attn", "self_attn_layer_norm", "self_attn_weights", "shape", "source_len", "torch", "training", "v", "value_states", "view"], "conditional_detr/modeling_conditional_detr.py:MLP.__init__": ["Linear", "Model", "Module", "ModuleList", "__init__", "class", "def", "for", "h", "hidden_dim", "in", "input_dim", "k", "layers", "n", "nn", "num_layers", "output_dim", "self", "super", "zip"], "conditional_detr/modeling_conditional_detr.py:MLP.forward": ["Model", "Module", "class", "def", "else", "enumerate", "for", "forward", "functional", "i", "if", "in", "layer", "layers", "nn", "num_layers", "relu", "return", "self", "x"], "conditional_detr/modeling_conditional_detr.py:ConditionalDetrPreTrainedModel._init_weights": ["Conv2d", "Embedding", "False", "GroupNorm", "LayerNorm", "Linear", "ModelLearnedPositionEmbedding", "ModelMHAttentionMap", "ModelPreTrainedModel", "None", "PreTrainedModel", "_init_weights", "_is_hf_initialized", "and", "bias", "class", "column_embeddings", "config", "def", "elif", "gain", "getattr", "if", "init", "init_std", "init_xavier_std", "is", "isinstance", "k_linear", "mean", "module", "nn", "normal_", "not", "ones_", "padding_idx", "q_linear", "row_embeddings", "self", "std", "uniform_", "weight", "xavier_std", "xavier_uniform_", "zeros_"], "conditional_detr/modeling_conditional_detr.py:ConditionalDetrEncoder.__init__": ["ModelConfig", "ModelEncoder", "ModelEncoderLayer", "ModelPreTrainedModel", "ModuleList", "_", "__init__", "class", "config", "def", "dropout", "encoder_layerdrop", "encoder_layers", "for", "in", "layerdrop", "layers", "nn", "post_init", "range", "self", "super"], "conditional_detr/modeling_conditional_detr.py:ConditionalDetrEncoder.forward": ["BaseModelOutput", "False", "ModelEncoder", "ModelPreTrainedModel", "None", "True", "_prepare_4d_attention_mask", "all_attentions", "attention_mask", "attentions", "class", "config", "def", "dropout", "dropout_probability", "dtype", "else", "encoder_layer", "encoder_states", "enumerate", "for", "forward", "functional", "hidden_states", "i", "if", "in", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layer_outputs", "layerdrop", "layers", "nn", "not", "object_queries", "output_attentions", "output_hidden_states", "p", "r", "rand", "return", "return_dict", "self", "to_drop", "torch", "training", "tuple", "use_return_dict", "v"], "conditional_detr/modeling_conditional_detr.py:ConditionalDetrDecoder.__init__": ["False", "LayerNorm", "MLP", "ModelConfig", "ModelDecoder", "ModelDecoderLayer", "ModelPreTrainedModel", "ModuleList", "None", "_", "__init__", "ca_qpos_proj", "class", "config", "d_model", "decoder_layerdrop", "decoder_layers", "def", "dropout", "for", "gradient_checkpointing", "in", "layer_id", "layerdrop", "layernorm", "layers", "nn", "post_init", "query_scale", "range", "ref_point_head", "self", "super"], "conditional_detr/modeling_conditional_detr.py:ConditionalDetrDecoder.forward": ["ModelDecoder", "ModelDecoderOutput", "ModelPreTrainedModel", "None", "_prepare_4d_attention_mask", "all_cross_attentions", "all_hidden_states", "all_self_attns", "and", "attention_mask", "attentions", "auxiliary_loss", "class", "config", "continue", "cross_attentions", "d_model", "decoder_layer", "def", "dropout_probability", "dtype", "else", "encoder_attention_mask", "encoder_hidden_states", "enumerate", "for", "forward", "gen_sine_position_embeddings", "hidden_states", "idx", "if", "in", "input_shape", "inputs_embeds", "intermediate", "intermediate_hidden_states", "is", "is_first", "kwargs", "last_hidden_state", "layer_outputs", "layerdrop", "layernorm", "layers", "not", "obj_center", "object_queries", "output_attentions", "output_hidden_states", "pos_transformation", "query_position_embeddings", "query_scale", "query_sine_embed", "query_sine_embed_before_transformation", "r", "rand", "ref_point_head", "reference_points", "reference_points_before_sigmoid", "return", "return_dict", "self", "sigmoid", "size", "stack", "tgt_len", "torch", "training", "transpose", "tuple", "use_return_dict", "v"], "conditional_detr/modeling_conditional_detr.py:ConditionalDetrModel.__init__": ["Conv2d", "Embedding", "ModelConfig", "ModelConvEncoder", "ModelConvModel", "ModelDecoder", "ModelEncoder", "ModelModel", "ModelPreTrainedModel", "__init__", "backbone", "build_position_encoding", "class", "config", "d_model", "decoder", "def", "encoder", "input_projection", "intermediate_channel_sizes", "kernel_size", "nn", "num_queries", "object_queries", "post_init", "query_position_embeddings", "self", "super"], "conditional_detr/modeling_conditional_detr.py:ConditionalDetrModel.freeze_backbone": ["False", "ModelModel", "ModelPreTrainedModel", "backbone", "class", "conv_encoder", "def", "for", "freeze_backbone", "in", "model", "name", "named_parameters", "param", "requires_grad_", "self"], "conditional_detr/modeling_conditional_detr.py:ConditionalDetrModel.unfreeze_backbone": ["ModelModel", "ModelPreTrainedModel", "True", "backbone", "class", "conv_encoder", "def", "for", "in", "model", "name", "named_parameters", "param", "requires_grad_", "self", "unfreeze_backbone"], "conditional_detr/modeling_conditional_detr.py:ConditionalDetrModel.forward": ["Backbone", "BaseModelOutput", "FloatTensor", "LongTensor", "ModelModel", "ModelModelOutput", "ModelPreTrainedModel", "None", "ValueError", "and", "attention_mask", "attentions", "backbone", "batch_size", "bool", "class", "config", "cross_attentions", "decoder", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_inputs_embeds", "decoder_outputs", "def", "device", "does", "downsampled", "elif", "else", "encoder", "encoder_attention_mask", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "feature_map", "features", "flatten", "flattened_features", "flattened_mask", "forward", "height", "hidden_states", "if", "input_projection", "inputs_embeds", "intermediate_hidden_states", "is", "isinstance", "kwargs", "last_hidden_state", "len", "mask", "not", "num_channels", "object_queries", "object_queries_list", "ones", "output_attentions", "output_hidden_states", "permute", "pixel", "pixel_mask", "pixel_values", "projected_feature_map", "queries", "query_position_embeddings", "r", "raise", "reference_points", "repeat", "return", "return_dict", "self", "shape", "torch", "tuple", "unsqueeze", "use_return_dict", "weight", "width", "zeros_like"], "conditional_detr/modeling_conditional_detr.py:ConditionalDetrMLPPredictionHead.__init__": ["Linear", "ModelMLPPredictionHead", "Module", "ModuleList", "__init__", "class", "def", "for", "h", "hidden_dim", "in", "input_dim", "k", "layers", "n", "nn", "num_layers", "output_dim", "self", "super", "zip"], "conditional_detr/modeling_conditional_detr.py:ConditionalDetrMLPPredictionHead.forward": ["ModelMLPPredictionHead", "Module", "class", "def", "else", "enumerate", "for", "forward", "functional", "i", "if", "in", "layer", "layers", "nn", "num_layers", "relu", "return", "self", "x"], "conditional_detr/modeling_conditional_detr.py:ConditionalDetrForObjectDetection.__init__": ["Linear", "ModelConfig", "ModelForObjectDetection", "ModelMLPPredictionHead", "ModelModel", "ModelPreTrainedModel", "__init__", "bbox_predictor", "class", "class_labels_classifier", "config", "d_model", "def", "hidden_dim", "input_dim", "model", "nn", "num_labels", "num_layers", "output_dim", "post_init", "self", "super"], "conditional_detr/modeling_conditional_detr.py:ConditionalDetrForObjectDetection._set_aux_loss": ["ModelForObjectDetection", "ModelPreTrainedModel", "_set_aux_loss", "a", "b", "class", "def", "for", "in", "logits", "outputs_class", "outputs_coord", "pred_boxes", "return", "self", "zip"], "conditional_detr/modeling_conditional_detr.py:ConditionalDetrForObjectDetection.forward": ["FloatTensor", "LongTensor", "ModelForObjectDetection", "ModelObjectDetectionOutput", "ModelPreTrainedModel", "None", "append", "auxiliary_loss", "auxiliary_outputs", "bbox_predictor", "bool", "class", "class_labels_classifier", "config", "cross_attentions", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_inputs_embeds", "def", "device", "dict", "else", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "for", "forward", "hs", "if", "in", "inputs_embeds", "intermediate", "intermediate_hidden_states", "inverse_sigmoid", "is", "kwargs", "labels", "last_hidden_state", "list", "logits", "loss", "loss_dict", "loss_function", "lvl", "model", "not", "output", "output_attentions", "output_hidden_states", "outputs", "outputs_class", "outputs_coord", "outputs_coords", "pixel_mask", "pixel_values", "pred_boxes", "r", "range", "reference", "reference_before_sigmoid", "reference_points", "return", "return_dict", "self", "sequence_output", "shape", "sigmoid", "stack", "tmp", "torch", "transpose", "tuple", "use_return_dict"], "conditional_detr/modeling_conditional_detr.py:ConditionalDetrForSegmentation.__init__": ["Model", "ModelConfig", "ModelForObjectDetection", "ModelForSegmentation", "ModelMHAttentionMap", "ModelMaskHeadSmallConv", "ModelPreTrainedModel", "__init__", "backbone", "bbox_attention", "class", "config", "conv_encoder", "d_model", "def", "dropout", "encoder_attention_heads", "hidden_size", "init_xavier_std", "intermediate_channel_sizes", "mask_head", "model", "number_of_heads", "post_init", "self", "std", "super"], "conditional_detr/modeling_conditional_detr.py:ConditionalDetrForSegmentation.forward": ["BaseModelOutput", "FloatTensor", "LongTensor", "Model", "ModelForSegmentation", "ModelPreTrainedModel", "ModelSegmentationOutput", "None", "and", "attention_mask", "attentions", "auxiliary_loss", "auxiliary_outputs", "backbone", "batch_size", "bbox_attention", "bbox_mask", "bbox_predictor", "bool", "class", "class_labels_classifier", "config", "cross_attentions", "d_model", "decoder", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_inputs_embeds", "decoder_outputs", "def", "device", "dict", "elif", "else", "encoder", "encoder_attention_mask", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "feature_map", "features", "flatten", "flattened_features", "flattened_mask", "forward", "height", "hidden_states", "if", "input_projection", "inputs_embeds", "intermediate", "intermediate_hidden_states", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "len", "list", "logits", "loss", "loss_dict", "loss_function", "mask", "mask_head", "memory", "model", "not", "num_channels", "num_queries", "object_queries", "object_queries_list", "ones", "output", "output_attentions", "output_hidden_states", "outputs_class", "outputs_coord", "permute", "pixel_mask", "pixel_values", "pred_boxes", "pred_masks", "projected_feature_map", "queries", "query_position_embeddings", "r", "repeat", "return", "return_dict", "seg_masks", "self", "sequence_output", "shape", "sigmoid", "torch", "tuple", "unsqueeze", "use_return_dict", "view", "weight", "width", "zeros_like"], "conditional_detr/modeling_conditional_detr.py:_expand": ["_expand", "def", "flatten", "int", "length", "repeat", "return", "tensor", "unsqueeze"], "conditional_detr/modeling_conditional_detr.py:ConditionalDetrMaskHeadSmallConv.__init__": ["Conv2d", "GroupNorm", "ModelMaskHeadSmallConv", "Module", "The", "ValueError", "__init__", "a", "adapter1", "adapter2", "adapter3", "as", "attention", "be", "bias", "by", "class", "constant_", "context_dim", "def", "dim", "divisible", "for", "fpn_dims", "gn1", "gn2", "gn3", "gn4", "gn5", "groups", "heads", "hidden_size", "if", "in", "init", "inter_dims", "is", "isinstance", "kaiming_uniform_", "lay1", "lay2", "lay3", "lay4", "lay5", "m", "min", "modules", "must", "nn", "number", "of", "out_lay", "padding", "raise", "self", "set", "super", "the", "to", "weight"], "conditional_detr/modeling_conditional_detr.py:ConditionalDetrMaskHeadSmallConv.forward": ["ModelMaskHeadSmallConv", "Module", "Tensor", "_expand", "adapter1", "adapter2", "adapter3", "bbox_mask", "cat", "class", "cur_fpn", "def", "flatten", "forward", "fpns", "functional", "gn1", "gn2", "gn3", "gn4", "gn5", "if", "interpolate", "lay1", "lay2", "lay3", "lay4", "lay5", "list", "mode", "nearest", "nn", "out_lay", "relu", "return", "self", "shape", "size", "torch", "x"], "conditional_detr/modeling_conditional_detr.py:ConditionalDetrMHAttentionMap.__init__": ["Dropout", "Linear", "ModelMHAttentionMap", "Module", "None", "True", "__init__", "bias", "class", "def", "dropout", "float", "hidden_dim", "k_linear", "nn", "normalize_fact", "num_heads", "q_linear", "query_dim", "self", "std", "super"], "conditional_detr/modeling_conditional_detr.py:ConditionalDetrMHAttentionMap.forward": ["ModelMHAttentionMap", "Module", "None", "Tensor", "bias", "bnchw", "bqnc", "bqnhw", "class", "conv2d", "def", "dim", "dropout", "dtype", "einsum", "finfo", "flatten", "forward", "functional", "hidden_dim", "if", "is", "k", "k_linear", "keys_per_head", "mask", "masked_fill", "min", "nn", "normalize_fact", "not", "num_heads", "q", "q_linear", "queries_per_head", "return", "self", "shape", "size", "softmax", "torch", "unsqueeze", "view", "weight", "weights"], "convbert/modeling_convbert.py:ConvBertEmbeddings.__init__": ["Dropout", "Embedding", "False", "LayerNorm", "ModelEmbeddings", "Module", "__init__", "arange", "class", "config", "def", "dropout", "dtype", "embedding_size", "eps", "expand", "hidden_dropout_prob", "layer_norm_eps", "long", "max_position_embeddings", "nn", "pad_token_id", "padding_idx", "persistent", "position_embeddings", "position_ids", "register_buffer", "self", "size", "super", "token_type_embeddings", "token_type_ids", "torch", "type_vocab_size", "vocab_size", "word_embeddings", "zeros"], "convbert/modeling_convbert.py:ConvBertEmbeddings.forward": ["FloatTensor", "LayerNorm", "LongTensor", "ModelEmbeddings", "Module", "None", "buffered_token_type_ids", "buffered_token_type_ids_expanded", "class", "def", "device", "dropout", "dtype", "else", "embeddings", "expand", "forward", "hasattr", "if", "input_ids", "input_shape", "inputs_embeds", "is", "long", "nn", "not", "position_embeddings", "position_ids", "return", "self", "seq_length", "size", "token_type_embeddings", "token_type_ids", "torch", "word_embeddings", "zeros"], "convbert/modeling_convbert.py:ConvBertPreTrainedModel._init_weights": ["GroupedLinearLayer", "ModelEmbeddings", "ModelPreTrainedModel", "PreTrainedModel", "SeparableModel1D", "_init_weights", "arange", "bias", "class", "config", "copy_", "def", "elif", "expand", "if", "init", "initializer_range", "isinstance", "mean", "module", "normal_", "position_ids", "self", "shape", "std", "super", "token_type_ids", "torch", "weight", "zeros_"], "convbert/modeling_convbert.py:SeparableConv1D.__init__": ["Conv1d", "False", "ModelConv1D", "Module", "Parameter", "__init__", "bias", "class", "config", "data", "def", "depthwise", "groups", "initializer_range", "input_filters", "kernel_size", "kwargs", "mean", "nn", "normal_", "output_filters", "padding", "pointwise", "self", "std", "super", "torch", "weight", "zeros"], "convbert/modeling_convbert.py:SeparableConv1D.forward": ["ModelConv1D", "Module", "Tensor", "bias", "class", "def", "depthwise", "forward", "hidden_states", "nn", "pointwise", "return", "self", "torch", "x"], "convbert/modeling_convbert.py:ConvBertSelfAttention.__init__": ["Dropout", "Linear", "ModelSelfAttention", "Model_kernel_layer", "Model_kernel_size", "Model_out_layer", "Module", "SeparableModel1D", "The", "Unfold", "ValueError", "__init__", "a", "all_head_size", "and", "attention", "attention_head_size", "attention_probs_dropout_prob", "be", "by", "class", "config", "def", "divisible", "dropout", "else", "embedding_size", "f", "hasattr", "head_ratio", "heads", "hidden", "hidden_size", "if", "int", "is", "kernel_size", "key", "key_Model_attn_layer", "multiple", "new_num_attention_heads", "nn", "not", "num_attention_heads", "number", "of", "padding", "query", "raise", "self", "should", "size", "super", "the", "unfold", "value"], "convbert/modeling_convbert.py:ConvBertSelfAttention.forward": ["False", "FloatTensor", "ModelSelfAttention", "Model_attn_layer", "Model_kernel_layer", "Model_kernel_size", "Model_out", "Model_out_layer", "Module", "None", "Tensor", "_", "all_head_size", "attention_head_size", "attention_mask", "attention_probs", "attention_scores", "batch_size", "bool", "cat", "class", "context_layer", "contiguous", "def", "dilation", "dim", "dropout", "else", "encoder_hidden_states", "forward", "functional", "hidden_states", "if", "is", "kernel_size", "key", "key_Model_attn_layer", "key_layer", "math", "matmul", "mixed_key_Model_attn_layer", "mixed_key_layer", "mixed_query_layer", "mixed_value_layer", "multiply", "new_context_layer_shape", "nn", "not", "num_attention_heads", "output_attentions", "outputs", "padding", "permute", "query", "query_layer", "reshape", "return", "self", "seq_length", "shape", "size", "softmax", "sqrt", "stride", "torch", "transpose", "tuple", "unfold", "unsqueeze", "value", "value_layer", "view"], "convbert/modeling_convbert.py:ConvBertSelfOutput.__init__": ["Dropout", "LayerNorm", "Linear", "ModelSelfOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "eps", "hidden_dropout_prob", "hidden_size", "layer_norm_eps", "nn", "self", "super"], "convbert/modeling_convbert.py:ConvBertSelfOutput.forward": ["LayerNorm", "ModelSelfOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "convbert/modeling_convbert.py:ConvBertAttention.__init__": ["ModelAttention", "ModelSelfAttention", "ModelSelfOutput", "Module", "__init__", "class", "config", "def", "nn", "output", "self", "super"], "convbert/modeling_convbert.py:ConvBertAttention.forward": ["False", "FloatTensor", "ModelAttention", "Module", "None", "Tensor", "attention_mask", "attention_output", "bool", "class", "def", "encoder_hidden_states", "forward", "hidden_states", "nn", "output", "output_attentions", "outputs", "return", "self", "self_outputs", "torch", "tuple"], "convbert/modeling_convbert.py:GroupedLinearLayer.__init__": ["ModelLinearLayer", "Module", "Parameter", "__init__", "bias", "class", "def", "empty", "group_in_dim", "group_out_dim", "input_size", "nn", "num_groups", "output_size", "self", "super", "torch", "weight"], "convbert/modeling_convbert.py:GroupedLinearLayer.forward": ["ModelLinearLayer", "Module", "Tensor", "batch_size", "bias", "class", "def", "forward", "group_in_dim", "hidden_states", "list", "matmul", "nn", "num_groups", "output_size", "permute", "reshape", "return", "self", "size", "torch", "weight", "x"], "convbert/modeling_convbert.py:ConvBertIntermediate.__init__": ["ACT2FN", "GroupedLinearLayer", "Linear", "ModelIntermediate", "Module", "__init__", "class", "config", "def", "dense", "else", "hidden_act", "hidden_size", "if", "input_size", "intermediate_act_fn", "intermediate_size", "isinstance", "nn", "num_groups", "output_size", "self", "str", "super"], "convbert/modeling_convbert.py:ConvBertIntermediate.forward": ["ModelIntermediate", "Module", "Tensor", "class", "def", "dense", "forward", "hidden_states", "intermediate_act_fn", "nn", "return", "self", "torch"], "convbert/modeling_convbert.py:ConvBertOutput.__init__": ["Dropout", "GroupedLinearLayer", "LayerNorm", "Linear", "ModelOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "else", "eps", "hidden_dropout_prob", "hidden_size", "if", "input_size", "intermediate_size", "layer_norm_eps", "nn", "num_groups", "output_size", "self", "super"], "convbert/modeling_convbert.py:ConvBertOutput.forward": ["LayerNorm", "ModelOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "convbert/modeling_convbert.py:ConvBertLayer.__init__": ["GradientCheckpointingLayer", "ModelAttention", "ModelIntermediate", "ModelLayer", "ModelOutput", "TypeError", "__init__", "a", "add_cross_attention", "added", "as", "attention", "be", "chunk_size_feed_forward", "class", "config", "cross", "crossattention", "decoder", "def", "f", "if", "intermediate", "is", "is_decoder", "model", "not", "output", "raise", "self", "seq_len_dim", "should", "super", "used"], "convbert/modeling_convbert.py:ConvBertLayer.forward": ["AttributeError", "False", "FloatTensor", "GradientCheckpointingLayer", "If", "ModelLayer", "None", "Tensor", "True", "add_cross_attention", "and", "apply_chunking_to_forward", "are", "attention", "attention_mask", "attention_output", "be", "bool", "by", "chunk_size_feed_forward", "class", "config", "cross", "cross_attention_outputs", "crossattention", "def", "encoder_attention_mask", "encoder_hidden_states", "f", "feed_forward_chunk", "forward", "has", "hasattr", "hidden_states", "if", "instantiated", "is", "is_decoder", "layer_output", "layers", "not", "output_attentions", "outputs", "passed", "raise", "return", "self", "self_attention_outputs", "seq_len_dim", "setting", "to", "torch", "tuple", "with"], "convbert/modeling_convbert.py:ConvBertLayer.feed_forward_chunk": ["GradientCheckpointingLayer", "ModelLayer", "attention_output", "class", "def", "feed_forward_chunk", "intermediate", "intermediate_output", "layer_output", "output", "return", "self"], "convbert/modeling_convbert.py:ConvBertEncoder.__init__": ["False", "ModelEncoder", "ModelLayer", "Module", "ModuleList", "_", "__init__", "class", "config", "def", "for", "gradient_checkpointing", "in", "layer", "nn", "num_hidden_layers", "range", "self", "super"], "convbert/modeling_convbert.py:ConvBertEncoder.forward": ["BaseModelOutputWithCrossAttentions", "False", "FloatTensor", "ModelEncoder", "Module", "None", "Tensor", "True", "add_cross_attention", "all_cross_attentions", "all_hidden_states", "all_self_attentions", "and", "attention_mask", "attentions", "bool", "class", "config", "cross_attentions", "def", "else", "encoder_attention_mask", "encoder_hidden_states", "enumerate", "for", "forward", "hidden_states", "i", "if", "in", "is", "last_hidden_state", "layer", "layer_module", "layer_outputs", "nn", "not", "output_attentions", "output_hidden_states", "return", "return_dict", "self", "torch", "tuple", "v"], "convbert/modeling_convbert.py:ConvBertPredictionHeadTransform.__init__": ["ACT2FN", "LayerNorm", "Linear", "ModelPredictionHeadTransform", "Module", "__init__", "class", "config", "def", "dense", "else", "eps", "hidden_act", "hidden_size", "if", "isinstance", "layer_norm_eps", "nn", "self", "str", "super", "transform_act_fn"], "convbert/modeling_convbert.py:ConvBertPredictionHeadTransform.forward": ["LayerNorm", "ModelPredictionHeadTransform", "Module", "Tensor", "class", "def", "dense", "forward", "hidden_states", "nn", "return", "self", "torch", "transform_act_fn"], "convbert/modeling_convbert.py:ConvBertSequenceSummary.__init__": ["Callable", "Dropout", "Identity", "Linear", "ModelConfig", "ModelSequenceSummary", "Module", "None", "NotImplementedError", "__init__", "activation", "activation_string", "and", "attn", "class", "config", "def", "else", "first_dropout", "get_activation", "getattr", "hasattr", "hidden_size", "if", "last", "last_dropout", "nn", "num_classes", "num_labels", "raise", "self", "summary", "summary_activation", "summary_first_dropout", "summary_last_dropout", "summary_proj_to_labels", "summary_type", "summary_use_proj", "super"], "convbert/modeling_convbert.py:ConvBertSequenceSummary.forward": ["FloatTensor", "LongTensor", "ModelSequenceSummary", "Module", "None", "NotImplementedError", "activation", "attn", "class", "cls_index", "def", "dim", "dtype", "elif", "else", "expand", "first", "first_dropout", "forward", "full_like", "gather", "hidden_states", "if", "is", "last", "last_dropout", "long", "mean", "nn", "output", "raise", "return", "self", "shape", "size", "squeeze", "summary", "summary_type", "torch", "unsqueeze"], "convbert/modeling_convbert.py:ConvBertModel.__init__": ["Linear", "ModelEmbeddings", "ModelEncoder", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "config", "def", "embedding_size", "embeddings", "embeddings_project", "encoder", "hidden_size", "if", "nn", "post_init", "self", "super"], "convbert/modeling_convbert.py:ConvBertModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "return", "self", "word_embeddings"], "convbert/modeling_convbert.py:ConvBertModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "self", "set_input_embeddings", "value", "word_embeddings"], "convbert/modeling_convbert.py:ConvBertModel.forward": ["BaseModelOutputWithCrossAttentions", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "ValueError", "You", "and", "at", "attention_mask", "batch_size", "bool", "both", "buffered_token_type_ids", "buffered_token_type_ids_expanded", "cannot", "class", "config", "def", "device", "dtype", "either", "elif", "else", "embeddings", "embeddings_project", "encoder", "expand", "extended_attention_mask", "forward", "get_extended_attention_mask", "hasattr", "have", "hidden_states", "if", "input_ids", "input_shape", "inputs_embeds", "is", "kwargs", "long", "not", "ones", "or", "output_attentions", "output_hidden_states", "position_ids", "raise", "return", "return_dict", "same", "self", "seq_length", "size", "specify", "the", "time", "to", "token_type_ids", "torch", "tuple", "use_return_dict", "warn_if_padding_and_no_attention_mask", "zeros"], "convbert/modeling_convbert.py:ConvBertGeneratorPredictions.__init__": ["LayerNorm", "Linear", "ModelGeneratorPredictions", "Module", "__init__", "activation", "class", "config", "def", "dense", "embedding_size", "eps", "gelu", "get_activation", "hidden_size", "layer_norm_eps", "nn", "self", "super"], "convbert/modeling_convbert.py:ConvBertGeneratorPredictions.forward": ["FloatTensor", "LayerNorm", "ModelGeneratorPredictions", "Module", "activation", "class", "def", "dense", "forward", "generator_hidden_states", "hidden_states", "nn", "return", "self", "torch"], "convbert/modeling_convbert.py:ConvBertForMaskedLM.__init__": ["Linear", "Model", "ModelForMaskedLM", "ModelGeneratorPredictions", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "config", "def", "embedding_size", "generator_lm_head", "generator_predictions", "nn", "post_init", "self", "super", "vocab_size"], "convbert/modeling_convbert.py:ConvBertForMaskedLM.get_output_embeddings": ["ModelForMaskedLM", "ModelPreTrainedModel", "class", "def", "generator_lm_head", "get_output_embeddings", "return", "self"], "convbert/modeling_convbert.py:ConvBertForMaskedLM.set_output_embeddings": ["ModelForMaskedLM", "ModelPreTrainedModel", "class", "def", "generator_lm_head", "self", "set_output_embeddings", "word_embeddings"], "convbert/modeling_convbert.py:ConvBertForMaskedLM.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "MaskedLMOutput", "Model", "ModelForMaskedLM", "ModelPreTrainedModel", "None", "attention_mask", "attentions", "bool", "class", "config", "def", "else", "forward", "generator_hidden_states", "generator_lm_head", "generator_predictions", "generator_sequence_output", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "nn", "not", "output", "output_attentions", "output_hidden_states", "position_ids", "prediction_scores", "r", "return", "return_dict", "self", "token_type_ids", "torch", "tuple", "use_return_dict", "view", "vocab_size"], "convbert/modeling_convbert.py:ConvBertClassificationHead.__init__": ["Dropout", "Linear", "ModelClassificationHead", "Module", "None", "__init__", "class", "classifier_dropout", "config", "def", "dense", "dropout", "else", "hidden_dropout_prob", "hidden_size", "if", "is", "nn", "not", "num_labels", "out_proj", "self", "super"], "convbert/modeling_convbert.py:ConvBertClassificationHead.forward": ["ACT2FN", "ModelClassificationHead", "Module", "Tensor", "class", "config", "def", "dense", "dropout", "forward", "hidden_act", "hidden_states", "kwargs", "nn", "out_proj", "return", "self", "torch", "x"], "convbert/modeling_convbert.py:ConvBertForSequenceClassification.__init__": ["Model", "ModelClassificationHead", "ModelForSequenceClassification", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "classifier", "config", "def", "num_labels", "post_init", "self", "super"], "convbert/modeling_convbert.py:ConvBertForSequenceClassification.forward": ["BCEWithLogitsLoss", "CrossEntropyLoss", "FloatTensor", "LongTensor", "MSELoss", "Model", "ModelForSequenceClassification", "ModelPreTrainedModel", "None", "SequenceClassifierOutput", "and", "attention_mask", "attentions", "bool", "class", "classifier", "config", "def", "dtype", "elif", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "kwargs", "labels", "logits", "long", "loss", "loss_fct", "multi_label_classification", "not", "num_labels", "or", "output", "output_attentions", "output_hidden_states", "outputs", "position_ids", "problem_type", "r", "regression", "return", "return_dict", "self", "sequence_output", "single_label_classification", "squeeze", "token_type_ids", "torch", "tuple", "use_return_dict", "view"], "convbert/modeling_convbert.py:ConvBertForMultipleChoice.__init__": ["Linear", "Model", "ModelForMultipleChoice", "ModelModel", "ModelPreTrainedModel", "ModelSequenceSummary", "__init__", "class", "classifier", "config", "def", "hidden_size", "nn", "post_init", "self", "sequence_summary", "super"], "convbert/modeling_convbert.py:ConvBertForMultipleChoice.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "Model", "ModelForMultipleChoice", "ModelPreTrainedModel", "MultipleChoiceModelOutput", "None", "attention_mask", "attentions", "bool", "class", "classifier", "config", "def", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "not", "num_choices", "output", "output_attentions", "output_hidden_states", "outputs", "pooled_output", "position_ids", "r", "reshaped_logits", "return", "return_dict", "self", "sequence_output", "sequence_summary", "shape", "size", "token_type_ids", "torch", "tuple", "use_return_dict", "view"], "convbert/modeling_convbert.py:ConvBertForTokenClassification.__init__": ["Dropout", "Linear", "Model", "ModelForTokenClassification", "ModelModel", "ModelPreTrainedModel", "None", "__init__", "class", "classifier", "classifier_dropout", "config", "def", "dropout", "else", "hidden_dropout_prob", "hidden_size", "if", "is", "nn", "not", "num_labels", "post_init", "self", "super"], "convbert/modeling_convbert.py:ConvBertForTokenClassification.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "Model", "ModelForTokenClassification", "ModelPreTrainedModel", "None", "TokenClassifierOutput", "attention_mask", "attentions", "bool", "class", "classifier", "config", "def", "dropout", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "not", "num_labels", "output", "output_attentions", "output_hidden_states", "outputs", "position_ids", "r", "return", "return_dict", "self", "sequence_output", "token_type_ids", "torch", "tuple", "use_return_dict", "view"], "convbert/modeling_convbert.py:ConvBertForQuestionAnswering.__init__": ["Linear", "Model", "ModelForQuestionAnswering", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "config", "def", "hidden_size", "nn", "num_labels", "post_init", "qa_outputs", "self", "super"], "convbert/modeling_convbert.py:ConvBertForQuestionAnswering.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "Model", "ModelForQuestionAnswering", "ModelPreTrainedModel", "None", "QuestionAnsweringModelOutput", "and", "attention_mask", "attentions", "bool", "clamp", "class", "config", "contiguous", "def", "dim", "else", "end_logits", "end_loss", "end_positions", "forward", "hidden_states", "if", "ignore_index", "ignored_index", "input_ids", "inputs_embeds", "is", "kwargs", "len", "logits", "loss", "loss_fct", "not", "output", "output_attentions", "output_hidden_states", "outputs", "position_ids", "qa_outputs", "return", "return_dict", "self", "sequence_output", "size", "split", "squeeze", "start_logits", "start_loss", "start_positions", "token_type_ids", "torch", "total_loss", "tuple", "use_return_dict"], "convnext/modeling_convnext.py:drop_path": ["False", "Model_path", "Model_prob", "Tensor", "bool", "def", "device", "div", "dtype", "float", "floor_", "if", "input", "keep_prob", "ndim", "not", "or", "output", "rand", "random_tensor", "return", "shape", "torch", "training"], "convnext/modeling_convnext.py:ConvNextDropPath.__init__": ["ModelDropPath", "Module", "None", "__init__", "class", "def", "drop_prob", "float", "nn", "self", "super"], "convnext/modeling_convnext.py:ConvNextDropPath.forward": ["ModelDropPath", "Module", "Tensor", "class", "def", "drop_path", "drop_prob", "forward", "hidden_states", "nn", "return", "self", "torch", "training"], "convnext/modeling_convnext.py:ConvNextDropPath.extra_repr": ["ModelDropPath", "Module", "class", "def", "drop_prob", "extra_repr", "f", "nn", "p", "return", "self", "str"], "convnext/modeling_convnext.py:ConvNextLayerNorm.__init__": ["LayerNorm", "ModelLayerNorm", "NotImplementedError", "Unsupported", "__init__", "channels_first", "channels_last", "class", "data", "data_format", "def", "eps", "f", "format", "if", "in", "kwargs", "nn", "normalized_shape", "not", "raise", "self", "super"], "convnext/modeling_convnext.py:ConvNextLayerNorm.forward": ["LayerNorm", "ModelLayerNorm", "Tensor", "channels_first", "class", "data_format", "def", "else", "features", "forward", "if", "nn", "permute", "return", "self", "super", "torch"], "convnext/modeling_convnext.py:ConvNextEmbeddings.__init__": ["Model2d", "ModelEmbeddings", "ModelLayerNorm", "Module", "__init__", "channels_first", "class", "config", "data_format", "def", "eps", "hidden_sizes", "kernel_size", "layernorm", "nn", "num_channels", "patch_embeddings", "patch_size", "self", "stride", "super"], "convnext/modeling_convnext.py:ConvNextEmbeddings.forward": ["FloatTensor", "Make", "ModelEmbeddings", "Module", "Tensor", "ValueError", "channel", "class", "configuration", "def", "dimension", "embeddings", "forward", "if", "in", "layernorm", "match", "nn", "num_channels", "of", "one", "patch_embeddings", "pixel", "pixel_values", "raise", "return", "self", "set", "shape", "sure", "that", "the", "torch", "values", "with"], "convnext/modeling_convnext.py:ConvNextLayer.__init__": ["ACT2FN", "Identity", "Linear", "Model2d", "ModelDropPath", "ModelLayer", "ModelLayerNorm", "Module", "None", "Parameter", "True", "__init__", "act", "class", "config", "def", "dim", "drop_path", "dwModel", "else", "eps", "groups", "hidden_act", "if", "kernel_size", "layer_scale_init_value", "layer_scale_parameter", "layernorm", "nn", "ones", "padding", "pwModel1", "pwModel2", "requires_grad", "self", "super", "torch"], "convnext/modeling_convnext.py:ConvNextLayer.forward": ["ModelLayer", "Module", "None", "Tensor", "act", "class", "def", "drop_path", "dwModel", "features", "forward", "if", "is", "layer_scale_parameter", "layernorm", "nn", "not", "permute", "pwModel1", "pwModel2", "residual", "return", "self", "torch"], "convnext/modeling_convnext.py:ConvNextStage.__init__": ["Model2d", "ModelLayer", "ModelLayerNorm", "ModelStage", "Module", "ModuleList", "None", "__init__", "channels_first", "class", "config", "data_format", "def", "depth", "dim", "downsampling_layer", "drop_path", "drop_path_rates", "else", "eps", "for", "if", "in", "in_channels", "j", "kernel_size", "layers", "nn", "or", "out_channels", "range", "self", "stride", "super"], "convnext/modeling_convnext.py:ConvNextStage.forward": ["ModelStage", "Module", "Tensor", "class", "def", "downsampling_layer", "features", "for", "forward", "in", "layer", "layers", "nn", "return", "self", "torch"], "convnext/modeling_convnext.py:ConvNextEncoder.__init__": ["ModelEncoder", "ModelStage", "Module", "ModuleList", "__init__", "append", "class", "config", "cpu", "def", "depth", "depths", "device", "drop_path_rate", "drop_path_rates", "else", "for", "hidden_sizes", "i", "if", "in", "in_channels", "linspace", "nn", "num_stages", "out_channels", "out_chs", "prev_chs", "range", "self", "split", "stage", "stages", "stride", "sum", "super", "tolist", "torch", "x"], "convnext/modeling_convnext.py:ConvNextEncoder.forward": ["BaseModelOutputWithNoAttention", "False", "ModelEncoder", "Module", "None", "Tensor", "all_hidden_states", "append", "bool", "class", "def", "else", "for", "forward", "hidden_states", "if", "in", "is", "last_hidden_state", "layer_module", "nn", "not", "output_hidden_states", "return", "self", "stages", "torch"], "convnext/modeling_convnext.py:ConvNextPreTrainedModel._init_weights": ["ModelLayer", "ModelPreTrainedModel", "None", "PreTrainedModel", "_init_weights", "class", "config", "constant_", "def", "if", "init", "is", "isinstance", "layer_scale_init_value", "layer_scale_parameter", "module", "not", "self", "super"], "convnext/modeling_convnext.py:ConvNextModel.__init__": ["LayerNorm", "ModelEmbeddings", "ModelEncoder", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "config", "def", "embeddings", "encoder", "eps", "hidden_sizes", "layer_norm_eps", "layernorm", "nn", "post_init", "self", "super"], "convnext/modeling_convnext.py:ConvNextModel.forward": ["BaseModelOutputWithNoAttention", "BaseModelOutputWithPoolingAndNoAttention", "FloatTensor", "ModelModel", "ModelPreTrainedModel", "None", "ValueError", "You", "bool", "class", "config", "def", "embedding_output", "embeddings", "encoder", "encoder_outputs", "forward", "have", "hidden_states", "if", "is", "kwargs", "last_hidden_state", "layernorm", "mean", "output_hidden_states", "pixel_values", "pooled_output", "pooler_output", "raise", "return", "self", "specify", "to", "torch"], "convnext/modeling_convnext.py:ConvNextForImageClassification.__init__": ["Identity", "Linear", "Model", "ModelForImageClassification", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "classifier", "config", "def", "else", "hidden_sizes", "if", "nn", "num_labels", "post_init", "self", "super"], "convnext/modeling_convnext.py:ConvNextForImageClassification.forward": ["BaseModelOutputWithPoolingAndNoAttention", "FloatTensor", "ImageClassifierOutputWithNoAttention", "LongTensor", "Model", "ModelForImageClassification", "ModelPreTrainedModel", "None", "class", "classifier", "config", "def", "forward", "hidden_states", "if", "is", "kwargs", "labels", "logits", "loss", "loss_function", "not", "outputs", "pixel_values", "pooled_logits", "pooled_output", "pooler_output", "r", "return", "self", "torch"], "convnext/modeling_convnext.py:ConvNextBackbone.__init__": ["BackboneMixin", "ModelBackbone", "ModelEmbeddings", "ModelEncoder", "ModelLayerNorm", "ModelPreTrainedModel", "ModuleDict", "__init__", "_init_backbone", "_out_features", "channels", "channels_first", "class", "config", "data_format", "def", "embeddings", "encoder", "for", "hidden_sizes", "hidden_states_norms", "in", "nn", "num_channels", "num_features", "post_init", "self", "stage", "super", "zip"], "convnext/modeling_convnext.py:ConvNextBackbone.forward": ["BackboneMixin", "BackboneOutput", "BaseModelOutputWithPoolingAndNoAttention", "ModelBackbone", "ModelPreTrainedModel", "None", "Tensor", "True", "append", "bool", "class", "config", "def", "else", "embedding_output", "embeddings", "encoder", "feature_maps", "for", "forward", "hidden_state", "hidden_states", "hidden_states_norms", "if", "in", "is", "kwargs", "out_features", "output_hidden_states", "outputs", "pixel_values", "r", "return", "self", "stage", "stage_names", "torch", "tuple", "zip"], "convnextv2/modeling_convnextv2.py:drop_path": ["False", "Model_path", "Model_prob", "Tensor", "bool", "def", "device", "div", "dtype", "float", "floor_", "if", "input", "keep_prob", "ndim", "not", "or", "output", "rand", "random_tensor", "return", "shape", "torch", "training"], "convnextv2/modeling_convnextv2.py:ConvNextV2DropPath.__init__": ["ModelDropPath", "Module", "None", "__init__", "class", "def", "drop_prob", "float", "nn", "self", "super"], "convnextv2/modeling_convnextv2.py:ConvNextV2DropPath.forward": ["ModelDropPath", "Module", "Tensor", "class", "def", "drop_path", "drop_prob", "forward", "hidden_states", "nn", "return", "self", "torch", "training"], "convnextv2/modeling_convnextv2.py:ConvNextV2DropPath.extra_repr": ["ModelDropPath", "Module", "class", "def", "drop_prob", "extra_repr", "f", "nn", "p", "return", "self", "str"], "convnextv2/modeling_convnextv2.py:ConvNextV2GRN.__init__": ["ModelGRN", "Module", "Parameter", "__init__", "bias", "class", "def", "dim", "int", "nn", "self", "super", "torch", "weight", "zeros"], "convnextv2/modeling_convnextv2.py:ConvNextV2GRN.forward": ["FloatTensor", "ModelGRN", "Module", "True", "bias", "class", "def", "dim", "forward", "global_features", "hidden_states", "keepdim", "linalg", "mean", "nn", "norm_features", "ord", "return", "self", "torch", "vector_norm", "weight"], "convnextv2/modeling_convnextv2.py:ConvNextV2LayerNorm.__init__": ["LayerNorm", "ModelLayerNorm", "NotImplementedError", "Unsupported", "__init__", "channels_first", "channels_last", "class", "data", "data_format", "def", "eps", "f", "format", "if", "in", "kwargs", "nn", "normalized_shape", "not", "raise", "self", "super"], "convnextv2/modeling_convnextv2.py:ConvNextV2LayerNorm.forward": ["LayerNorm", "ModelLayerNorm", "Tensor", "channels_first", "class", "data_format", "def", "else", "features", "forward", "if", "nn", "permute", "return", "self", "super", "torch"], "convnextv2/modeling_convnextv2.py:ConvNextV2Embeddings.__init__": ["Model2d", "ModelEmbeddings", "ModelLayerNorm", "Module", "__init__", "channels_first", "class", "config", "data_format", "def", "eps", "hidden_sizes", "kernel_size", "layernorm", "nn", "num_channels", "patch_embeddings", "patch_size", "self", "stride", "super"], "convnextv2/modeling_convnextv2.py:ConvNextV2Embeddings.forward": ["FloatTensor", "Make", "ModelEmbeddings", "Module", "Tensor", "ValueError", "channel", "class", "configuration", "def", "dimension", "embeddings", "forward", "if", "in", "layernorm", "match", "nn", "num_channels", "of", "one", "patch_embeddings", "pixel", "pixel_values", "raise", "return", "self", "set", "shape", "sure", "that", "the", "torch", "values", "with"], "convnextv2/modeling_convnextv2.py:ConvNextV2Layer.__init__": ["ACT2FN", "Identity", "Linear", "Model2d", "ModelDropPath", "ModelGRN", "ModelLayer", "ModelLayerNorm", "Module", "__init__", "act", "class", "config", "def", "dim", "drop_path", "dwModel", "else", "eps", "grn", "groups", "hidden_act", "if", "kernel_size", "layernorm", "nn", "padding", "pwModel1", "pwModel2", "self", "super"], "convnextv2/modeling_convnextv2.py:ConvNextV2Layer.forward": ["ModelLayer", "Module", "Tensor", "act", "class", "def", "drop_path", "dwModel", "features", "forward", "grn", "layernorm", "nn", "permute", "pwModel1", "pwModel2", "residual", "return", "self", "torch"], "convnextv2/modeling_convnextv2.py:ConvNextV2Stage.__init__": ["Model2d", "ModelLayer", "ModelLayerNorm", "ModelStage", "Module", "ModuleList", "None", "__init__", "channels_first", "class", "config", "data_format", "def", "depth", "dim", "downsampling_layer", "drop_path", "drop_path_rates", "else", "eps", "for", "if", "in", "in_channels", "j", "kernel_size", "layers", "nn", "or", "out_channels", "range", "self", "stride", "super"], "convnextv2/modeling_convnextv2.py:ConvNextV2Stage.forward": ["ModelStage", "Module", "Tensor", "class", "def", "downsampling_layer", "features", "for", "forward", "in", "layer", "layers", "nn", "return", "self", "torch"], "convnextv2/modeling_convnextv2.py:ConvNextV2Encoder.__init__": ["ModelEncoder", "ModelStage", "Module", "ModuleList", "__init__", "append", "class", "config", "cpu", "def", "depth", "depths", "device", "drop_path_rate", "drop_path_rates", "else", "for", "hidden_sizes", "i", "if", "in", "in_channels", "linspace", "nn", "num_stages", "out_channels", "out_chs", "prev_chs", "range", "self", "split", "stage", "stages", "stride", "sum", "super", "tolist", "torch", "x"], "convnextv2/modeling_convnextv2.py:ConvNextV2Encoder.forward": ["BaseModelOutputWithNoAttention", "False", "ModelEncoder", "Module", "None", "Tensor", "all_hidden_states", "append", "bool", "class", "def", "else", "for", "forward", "hidden_states", "if", "in", "is", "last_hidden_state", "layer_module", "nn", "not", "output_hidden_states", "return", "self", "stages", "torch"], "convnextv2/modeling_convnextv2.py:ConvNextV2PreTrainedModel._init_weights": ["ModelGRN", "ModelPreTrainedModel", "PreTrainedModel", "_init_weights", "bias", "class", "def", "if", "init", "isinstance", "module", "self", "super", "weight", "zeros_"], "convnextv2/modeling_convnextv2.py:ConvNextV2Model.__init__": ["LayerNorm", "ModelEmbeddings", "ModelEncoder", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "config", "def", "embeddings", "encoder", "eps", "hidden_sizes", "layer_norm_eps", "layernorm", "nn", "post_init", "self", "super"], "convnextv2/modeling_convnextv2.py:ConvNextV2Model.forward": ["BaseModelOutputWithNoAttention", "BaseModelOutputWithPoolingAndNoAttention", "FloatTensor", "ModelModel", "ModelPreTrainedModel", "None", "ValueError", "You", "bool", "class", "config", "def", "embedding_output", "embeddings", "encoder", "encoder_outputs", "forward", "have", "hidden_states", "if", "is", "kwargs", "last_hidden_state", "layernorm", "mean", "output_hidden_states", "pixel_values", "pooled_output", "pooler_output", "raise", "return", "self", "specify", "to", "torch"], "convnextv2/modeling_convnextv2.py:ConvNextV2ForImageClassification.__init__": ["Identity", "Linear", "Model", "ModelForImageClassification", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "classifier", "config", "def", "else", "hidden_sizes", "if", "nn", "num_labels", "post_init", "self", "super"], "convnextv2/modeling_convnextv2.py:ConvNextV2ForImageClassification.forward": ["BaseModelOutputWithPoolingAndNoAttention", "FloatTensor", "ImageClassifierOutputWithNoAttention", "LongTensor", "Model", "ModelForImageClassification", "ModelPreTrainedModel", "None", "class", "classifier", "config", "def", "forward", "hidden_states", "if", "is", "kwargs", "labels", "logits", "loss", "loss_function", "not", "outputs", "pixel_values", "pooled_logits", "pooled_output", "pooler_output", "r", "return", "self", "torch"], "convnextv2/modeling_convnextv2.py:ConvNextV2Backbone.__init__": ["BackboneMixin", "ModelBackbone", "ModelEmbeddings", "ModelEncoder", "ModelLayerNorm", "ModelPreTrainedModel", "ModuleDict", "__init__", "_init_backbone", "_out_features", "channels", "channels_first", "class", "config", "data_format", "def", "embeddings", "encoder", "for", "hidden_sizes", "hidden_states_norms", "in", "nn", "num_channels", "num_features", "post_init", "self", "stage", "super", "zip"], "convnextv2/modeling_convnextv2.py:ConvNextV2Backbone.forward": ["BackboneMixin", "BackboneOutput", "BaseModelOutputWithPoolingAndNoAttention", "ModelBackbone", "ModelPreTrainedModel", "None", "Tensor", "True", "append", "bool", "class", "config", "def", "else", "embedding_output", "embeddings", "encoder", "feature_maps", "for", "forward", "hidden_state", "hidden_states", "hidden_states_norms", "if", "in", "is", "kwargs", "out_features", "output_hidden_states", "outputs", "pixel_values", "r", "return", "self", "stage", "stage_names", "torch", "tuple", "zip"], "cpmant/modeling_cpmant.py:CpmAntLayerNorm.__init__": ["ModelConfig", "ModelLayerNorm", "Module", "Parameter", "__init__", "class", "config", "def", "dim_norm", "empty", "eps", "hidden_size", "nn", "self", "super", "torch", "weight"], "cpmant/modeling_cpmant.py:CpmAntLayerNorm.forward": ["AssertionError", "ModelLayerNorm", "Module", "Tensor", "True", "class", "def", "dim", "dim_norm", "dtype", "eps", "float32", "forward", "hidden_states", "if", "keepdim", "mean", "nn", "old_dtype", "pow", "raise", "return", "rsqrt", "self", "size", "to", "torch", "variance", "weight"], "cpmant/modeling_cpmant.py:CpmAntAttention.__init__": ["Dropout", "False", "Linear", "ModelAttention", "ModelConfig", "Module", "None", "Softmax", "__init__", "attention_out", "bias", "class", "config", "def", "dim", "dim_head", "dim_model", "dropout", "dropout_p", "else", "hidden_size", "if", "is", "layer_idx", "nn", "not", "num_attention_heads", "num_heads", "p", "project_k", "project_q", "project_v", "self", "softmax", "super", "torch"], "cpmant/modeling_cpmant.py:CpmAntAttention.forward": ["BoolTensor", "Cache", "False", "ModelAttention", "Module", "None", "Tensor", "attention_mask", "attention_out", "attn_weights", "batch_size", "bool", "cache_position", "class", "contiguous", "def", "device", "dim_head", "dropout", "dtype", "else", "float", "forward", "hidden_kv", "hidden_q", "if", "inf", "is", "key", "layer_idx", "len_k", "len_q", "masked_fill", "math", "matmul", "nn", "not", "num_heads", "output_attentions", "past_key_values", "permute", "position_bias", "project_k", "project_q", "project_v", "query", "return", "scalar_tensor", "score", "self", "size", "softmax", "sqrt", "tensor", "torch", "transpose", "update", "use_cache", "value", "view"], "cpmant/modeling_cpmant.py:CpmAntSelfAttentionBlock.__init__": ["Dropout", "ModelAttention", "ModelConfig", "ModelLayerNorm", "ModelSelfAttentionBlock", "Module", "None", "__init__", "class", "config", "def", "dropout", "dropout_p", "else", "if", "layer_idx", "layernorm_before_attention", "nn", "self", "self_attention", "super", "torch"], "cpmant/modeling_cpmant.py:CpmAntSelfAttentionBlock.forward": ["Cache", "False", "ModelSelfAttentionBlock", "Module", "None", "Tensor", "attention_mask", "attn_weights", "bool", "cache_position", "class", "def", "dropout", "forward", "hidden_states", "if", "is", "layernorm_before_attention", "nn", "not", "output_attentions", "outputs", "past_key_values", "position_bias", "return", "self", "self_attention", "torch", "use_cache"], "cpmant/modeling_cpmant.py:CpmAntDenseGatedACT.__init__": ["False", "GELU", "Linear", "ModelConfig", "ModelDenseGatedACT", "Module", "__init__", "act", "bias", "class", "config", "def", "dim_ff", "hidden_size", "nn", "self", "super", "torch", "w_0", "w_1"], "cpmant/modeling_cpmant.py:CpmAntDenseGatedACT.forward": ["ModelDenseGatedACT", "Module", "Tensor", "act", "class", "def", "forward", "gate_score", "hidden_states", "nn", "return", "self", "torch", "w_0", "w_1"], "cpmant/modeling_cpmant.py:CpmAntFeedForward.__init__": ["Dropout", "False", "Linear", "ModelConfig", "ModelDenseGatedACT", "ModelFeedForward", "Module", "None", "__init__", "bias", "class", "config", "def", "dim_ff", "dropout", "dropout_p", "else", "hidden_size", "if", "is", "nn", "not", "self", "super", "torch", "w_in", "w_out"], "cpmant/modeling_cpmant.py:CpmAntFeedForward.forward": ["ModelFeedForward", "Module", "None", "Tensor", "class", "def", "dropout", "forward", "hidden_states", "if", "is", "nn", "not", "return", "self", "torch", "w_in", "w_out"], "cpmant/modeling_cpmant.py:CpmAntFFNBlock.__init__": ["Dropout", "ModelConfig", "ModelFFNBlock", "ModelFeedForward", "ModelLayerNorm", "Module", "None", "__init__", "class", "config", "def", "dropout", "dropout_p", "else", "ffn", "if", "layernorm_before_ffn", "nn", "self", "super", "torch"], "cpmant/modeling_cpmant.py:CpmAntFFNBlock.forward": ["ModelFFNBlock", "Module", "None", "Tensor", "class", "def", "dropout", "ffn", "forward", "hidden_states", "if", "is", "layernorm_before_ffn", "ln_outputs", "nn", "not", "outputs", "return", "self", "torch"], "cpmant/modeling_cpmant.py:CpmAntTransformerBlock.__init__": ["ModelConfig", "ModelFFNBlock", "ModelSelfAttentionBlock", "ModelTransformerBlock", "Module", "None", "__init__", "class", "config", "def", "ffn", "layer_idx", "nn", "self", "self_att", "super"], "cpmant/modeling_cpmant.py:CpmAntTransformerBlock.forward": ["Cache", "False", "ModelTransformerBlock", "Module", "None", "Tensor", "attention_mask", "attn_weights", "bool", "cache_position", "class", "def", "ffn", "forward", "hidden_states", "nn", "output_attentions", "past_key_values", "position_bias", "return", "self", "self_att", "torch", "use_cache"], "cpmant/modeling_cpmant.py:CpmAntEncoder.__init__": ["ModelConfig", "ModelEncoder", "ModelLayerNorm", "ModelTransformerBlock", "Module", "ModuleList", "__init__", "class", "config", "def", "for", "i", "in", "layer_idx", "layers", "nn", "num_hidden_layers", "num_layers", "output_layernorm", "range", "self", "super"], "cpmant/modeling_cpmant.py:CpmAntEncoder.forward": ["Cache", "ModelEncoder", "Module", "None", "Tensor", "all_hidden_states", "all_self_attns", "attention_mask", "attn_weights", "bool", "cache_position", "class", "def", "else", "enumerate", "for", "forward", "hidden_states", "i", "if", "in", "layer", "layer_outputs", "layers", "nn", "output_attentions", "output_hidden_states", "output_layernorm", "past_key_values", "position_bias", "return", "self", "torch", "use_cache"], "cpmant/modeling_cpmant.py:CpmAntIntermediate.__init__": ["ACT2FN", "Linear", "ModelIntermediate", "Module", "__init__", "class", "config", "def", "dense", "else", "hidden_act", "hidden_size", "if", "intermediate_act_fn", "intermediate_size", "isinstance", "nn", "self", "str", "super"], "cpmant/modeling_cpmant.py:CpmAntIntermediate.forward": ["ModelIntermediate", "Module", "Tensor", "class", "def", "dense", "forward", "hidden_states", "intermediate_act_fn", "nn", "return", "self", "torch"], "cpmant/modeling_cpmant.py:CpmAntSegmentPositionEmbedding.__init__": ["ModelConfig", "ModelSegmentPositionEmbedding", "Module", "Parameter", "__init__", "class", "config", "def", "empty", "max_distance", "nn", "num_attention_heads", "num_buckets", "num_heads", "num_segments", "position_bias_max_distance", "position_bias_num_buckets", "relative_attention_bias", "segment_types", "self", "super", "torch"], "cpmant/modeling_cpmant.py:CpmAntSegmentPositionEmbedding.forward": ["AssertionError", "F", "ModelSegmentPositionEmbedding", "Module", "None", "Tensor", "_position_bucket", "_segment_relative_position_bucket", "absolute_position_bucket", "and", "arange", "batch", "be", "but", "class", "contiguous", "def", "device", "dtype", "embedding", "embeds", "equal", "f", "forward", "got", "if", "int32", "key_pos", "key_segment", "keylen", "max_distance", "nn", "no_grad", "num_buckets", "or", "permute", "query_pos", "query_segment", "querylen", "raise", "relative_attention_bias", "relative_position_bucket", "return", "self", "should", "size", "to", "torch", "view", "where", "with"], "cpmant/modeling_cpmant.py:CpmAntSegmentPositionEmbedding._segment_relative_position_bucket": ["ModelSegmentPositionEmbedding", "Module", "_segment_relative_position_bucket", "class", "def", "key_segment", "nn", "num_segments", "query_segment", "return", "self"], "cpmant/modeling_cpmant.py:CpmAntSegmentPositionEmbedding._position_bucket": ["ModelSegmentPositionEmbedding", "Module", "_position_bucket", "abs", "class", "def", "float", "full_like", "int32", "is_small", "log", "math", "max_distance", "max_exact", "min", "nn", "num_buckets", "relative_buckets", "relative_position", "relative_position_if_large", "return", "self", "to", "torch", "where"], "cpmant/modeling_cpmant.py:CpmAntOutput.__init__": ["Dropout", "LayerNorm", "Linear", "ModelOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "eps", "hidden_dropout_prob", "hidden_size", "intermediate_size", "layer_norm_eps", "nn", "self", "super"], "cpmant/modeling_cpmant.py:CpmAntOutput.forward": ["LayerNorm", "ModelOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "cpmant/modeling_cpmant.py:CpmAntPreTrainedModel._init_weights": ["ModelLayerNorm", "ModelPreTrainedModel", "ModelSegmentPositionEmbedding", "PreTrainedModel", "_init_weights", "class", "config", "def", "elif", "if", "init", "init_std", "isinstance", "mean", "module", "normal_", "ones_", "relative_attention_bias", "self", "std", "super", "weight"], "cpmant/modeling_cpmant.py:CpmAntModel.__init__": ["Embedding", "ModelConfig", "ModelEncoder", "ModelModel", "ModelPreTrainedModel", "ModelSegmentPositionEmbedding", "__init__", "class", "config", "def", "encoder", "hidden_size", "input_embedding", "nn", "position_bias", "post_init", "prompt_length", "prompt_types", "segment_embedding", "segment_types", "self", "super", "vocab_size"], "cpmant/modeling_cpmant.py:CpmAntModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "input_embedding", "return", "self"], "cpmant/modeling_cpmant.py:CpmAntModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "input_embedding", "kwargs", "self", "set_input_embeddings"], "cpmant/modeling_cpmant.py:CpmAntModel._prepare_attention_mask": ["ModelModel", "ModelPreTrainedModel", "None", "_prepare_attention_mask", "arange", "attention_mask", "batch", "bool", "cat", "class", "context", "def", "device", "dim", "directional_mask_2d", "input_ids", "length", "list", "logical_not", "mask_1d", "ones", "prompt_length", "range", "repeat", "return", "self", "seqlen", "size", "span", "tensor", "torch", "view"], "cpmant/modeling_cpmant.py:CpmAntModel.forward": ["BaseModelOutputWithPast", "Cache", "DynamicCache", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "_prepare_attention_mask", "all_attentions", "all_hidden_states", "and", "arange", "attention", "attention_mask", "attentions", "batch", "bool", "cache_position", "cat", "class", "config", "context", "contiguous", "def", "device", "dim", "dtype", "else", "encoder", "for", "forward", "full", "get_seq_length", "hidden_state", "hidden_states", "if", "in", "input_embedding", "input_ids", "int32", "is", "kwargs", "last_hidden_state", "length", "new_attentions", "new_hidden_states", "not", "output_attentions", "output_hidden_states", "past_key_values", "past_length", "position", "position_bias", "prompt_length", "r", "repeat", "return", "return_dict", "segment", "segment_embedding", "segment_states", "self", "seq_length", "size", "span", "sum", "to", "torch", "tuple", "use_cache", "use_return_dict", "v", "vocab_size", "where", "zeros"], "cpmant/modeling_cpmant.py:CpmAntForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "Model", "ModelConfig", "ModelForCausalLM", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "nn", "post_init", "prompt_length", "prompt_types", "self", "super", "vocab_size"], "cpmant/modeling_cpmant.py:CpmAntForCausalLM.forward": ["Cache", "CausalLMOutputWithPast", "CrossEntropyLoss", "GenerationMixin", "Model", "ModelForCausalLM", "ModelPreTrainedModel", "None", "Tensor", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "else", "forward", "hidden_states", "if", "input_ids", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "logits", "logits_to_keep", "loss", "loss_func", "model_output", "not", "output", "output_attentions", "output_hidden_states", "past_key_values", "r", "return", "return_dict", "self", "size", "slice", "slice_indices", "torch", "tuple", "use_cache", "use_return_dict", "view"], "cpmant/modeling_cpmant.py:CpmAntForCausalLM.get_input_embeddings": ["GenerationMixin", "Model", "ModelForCausalLM", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "input_embedding", "return", "self"], "cpmant/modeling_cpmant.py:CpmAntForCausalLM.set_input_embeddings": ["GenerationMixin", "Model", "ModelForCausalLM", "ModelPreTrainedModel", "class", "def", "embeddings", "input_embedding", "self", "set_input_embeddings"], "csm/modeling_csm.py:CsmRMSNorm.__init__": ["ModelRMSNorm", "Module", "Parameter", "__init__", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "csm/modeling_csm.py:CsmRMSNorm.forward": ["ModelRMSNorm", "Module", "True", "class", "def", "dtype", "float32", "forward", "hidden_states", "input_dtype", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "csm/modeling_csm.py:CsmRMSNorm.extra_repr": ["ModelRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "variance_epsilon", "weight"], "csm/modeling_csm.py:CsmRotaryEmbedding.__init__": ["Callable", "False", "ModelConfig", "ModelRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "csm/modeling_csm.py:CsmRotaryEmbedding.compute_default_rope_parameters": ["ModelConfig", "ModelRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "csm/modeling_csm.py:CsmRotaryEmbedding.forward": ["False", "ModelRotaryEmbedding", "Module", "None", "and", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "csm/modeling_csm.py:CsmMLP.__init__": ["ACT2FN", "Linear", "ModelMLP", "Module", "__init__", "act_fn", "bias", "class", "config", "def", "down_proj", "gate_proj", "hidden_act", "hidden_size", "intermediate_size", "mlp_bias", "nn", "self", "super", "up_proj"], "csm/modeling_csm.py:CsmMLP.forward": ["ModelMLP", "Module", "act_fn", "class", "def", "down_proj", "forward", "gate_proj", "nn", "return", "self", "up_proj", "x"], "csm/modeling_csm.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "csm/modeling_csm.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cos", "def", "k", "k_embed", "q", "q_embed", "return", "rotate_half", "sin", "unsqueeze", "unsqueeze_dim"], "csm/modeling_csm.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "csm/modeling_csm.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value", "value_states"], "csm/modeling_csm.py:CsmAttention.__init__": ["Linear", "ModelAttention", "ModelConfig", "Module", "True", "__init__", "attention_bias", "attention_dropout", "bias", "class", "config", "def", "getattr", "head_dim", "hidden_size", "int", "is_causal", "k_proj", "layer_idx", "nn", "num_attention_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "q_proj", "scaling", "self", "super", "v_proj"], "csm/modeling_csm.py:CsmAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "LongTensor", "ModelAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_attn_implementation", "apply_rotary_pos_emb", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "cache_kwargs", "cache_position", "class", "config", "contiguous", "cos", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "head_dim", "hidden_shape", "hidden_states", "if", "input_shape", "is", "k_proj", "key_states", "kwargs", "layer_idx", "nn", "not", "o_proj", "past_key_values", "position_embeddings", "q_proj", "query_states", "reshape", "return", "scaling", "self", "shape", "sin", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "view"], "csm/modeling_csm.py:CsmDecoderLayer.__init__": ["GradientCheckpointingLayer", "ModelAttention", "ModelConfig", "ModelDecoderLayer", "ModelMLP", "ModelRMSNorm", "__init__", "class", "config", "def", "eps", "hidden_size", "input_layernorm", "int", "layer_idx", "mlp", "post_attention_layernorm", "rms_norm_eps", "self", "self_attn", "super"], "csm/modeling_csm.py:CsmDecoderLayer.forward": ["Cache", "False", "GradientCheckpointingLayer", "LongTensor", "ModelDecoderLayer", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "attention_mask", "bool", "cache_position", "class", "def", "forward", "hidden_states", "input_layernorm", "kwargs", "mlp", "past_key_values", "position_embeddings", "position_ids", "post_attention_layernorm", "residual", "return", "self", "self_attn", "torch", "tuple", "use_cache"], "csm/modeling_csm.py:CsmPreTrainedModel._init_weights": ["ModelBackboneModelEmbeddings", "ModelCodebooksHead", "ModelPreTrainedModel", "PreTrainedModel", "_init_weights", "arange", "audio_tokens_offsets", "class", "config", "copy_", "def", "elif", "for", "i", "if", "in", "init", "initializer_range", "isinstance", "mean", "module", "normal_", "num_codebooks", "range", "self", "std", "super", "torch", "vocab_size", "weight"], "csm/modeling_csm.py:CsmDepthDecoderModel.__init__": ["Embedding", "False", "Linear", "ModelDecoderLayer", "ModelDepthDecoderModel", "ModelPreTrainedModel", "ModelRMSNorm", "ModelRotaryEmbedding", "ModuleList", "__init__", "backbone_hidden_size", "bias", "class", "config", "def", "embed_tokens", "eps", "for", "gradient_checkpointing", "hidden_size", "in", "inputs_embeds_projector", "layer_idx", "layers", "nn", "norm", "num_codebooks", "num_hidden_layers", "pad_token_id", "padding_idx", "post_init", "range", "rms_norm_eps", "rotary_emb", "self", "super", "vocab_size"], "csm/modeling_csm.py:CsmDepthDecoderModel.forward": ["BaseModelOutputWithPast", "Cache", "Custom", "DynamicCache", "FloatTensor", "LongTensor", "Model", "ModelDepthDecoderModel", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "When", "You", "across", "also", "and", "arange", "as", "attention_mask", "automatically", "backbone_last_hidden_state", "batch", "be", "bool", "but", "cache_position", "causal_mask", "clamp", "class", "codebook", "codebook_idxs", "config", "correct", "create_causal_mask", "decoder", "decoder_layer", "def", "depth", "determines", "device", "else", "embed_tokens", "exactly", "first", "for", "forward", "from", "get_seq_length", "hidden_states", "identical", "if", "ignored", "in", "inference", "input_embeds", "input_ids", "input_ids_are_first_codebook", "inputs_embeds", "inputs_embeds_projector", "inputs_seq_length", "is", "is_torchdynamo_compiling", "it", "kwargs", "last_hidden_state", "layers", "logger", "min", "must", "norm", "not", "num_hidden_layers", "of", "offset", "one", "or", "past_key_values", "past_seen_tokens", "position_embeddings", "position_ids", "provided", "r", "raise", "requires", "return", "rotary_emb", "self", "shape", "should", "specify", "the", "them", "to", "token", "torch", "tuple", "unsqueeze", "use_cache", "vocab_size", "warning", "warning_once", "were", "will"], "csm/modeling_csm.py:CsmCodebooksHead.__init__": ["ModelCodebooksHead", "Module", "Parameter", "__init__", "class", "def", "empty", "hidden_size", "nn", "num_codebooks", "self", "super", "torch", "vocab_size", "weight"], "csm/modeling_csm.py:CsmCodebooksHead.forward": ["ModelCodebooksHead", "Module", "None", "T", "arange", "cache_position", "class", "codebook_idx", "codebook_idxs", "codebook_weight", "def", "dim", "else", "for", "forward", "functional", "hidden_states", "if", "in", "is", "linear", "nn", "range", "return", "self", "seq_length", "shape", "stack", "torch", "weight"], "csm/modeling_csm.py:CsmDepthDecoderForCausalLM.__init__": ["GenerationMixin", "ModelCodebooksHead", "ModelDepthDecoderForCausalLM", "ModelDepthDecoderModel", "ModelPreTrainedModel", "__init__", "class", "codebooks_head", "config", "def", "hidden_size", "model", "num_codebooks", "post_init", "self", "super", "vocab_size"], "csm/modeling_csm.py:CsmDepthDecoderForCausalLM.forward": ["Cache", "CausalLMOutputWithPast", "FloatTensor", "GenerationMixin", "LongTensor", "ModelDepthDecoderForCausalLM", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "backbone_last_hidden_state", "bool", "cache_position", "class", "codebooks_head", "config", "contiguous", "def", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "outputs", "past_key_values", "position_ids", "r", "return", "self", "shift_labels", "slice", "slice_indices", "torch", "tuple", "use_cache", "vocab_size"], "csm/modeling_csm.py:CsmDepthDecoderForCausalLM.prepare_inputs_for_generation": ["Cache", "FloatTensor", "GenerationMixin", "LongTensor", "ModelDepthDecoderForCausalLM", "ModelPreTrainedModel", "None", "attention_mask", "backbone_last_hidden_state", "cache_position", "class", "def", "if", "input_ids", "inputs_embeds", "is_first_generation_step", "kwargs", "model_inputs", "not", "past_key_values", "pop", "position_ids", "prepare_inputs_for_generation", "return", "self", "super", "torch"], "csm/modeling_csm.py:CsmBackboneModelEmbeddings.__init__": ["Embedding", "False", "ModelBackboneModelEmbeddings", "Module", "__init__", "arange", "audio_tokens_offsets", "class", "config", "def", "embed_audio_tokens", "hidden_size", "nn", "num_codebooks", "persistent", "register_buffer", "self", "super", "torch", "vocab_size"], "csm/modeling_csm.py:CsmBackboneModelEmbeddings.forward": ["ModelBackboneModelEmbeddings", "Module", "audio_tokens_offsets", "class", "def", "dim", "embed_audio_tokens", "forward", "input_embeds", "input_ids", "nn", "return", "self", "sum"], "csm/modeling_csm.py:CsmBackboneModel.__init__": ["False", "ModelBackboneModel", "ModelBackboneModelEmbeddings", "ModelDecoderLayer", "ModelPreTrainedModel", "ModelRMSNorm", "ModelRotaryEmbedding", "ModuleList", "__init__", "class", "config", "def", "embed_tokens", "eps", "for", "gradient_checkpointing", "hidden_size", "in", "layer_idx", "layers", "nn", "norm", "num_hidden_layers", "pad_token_id", "padding_idx", "post_init", "range", "rms_norm_eps", "rotary_emb", "self", "super", "vocab_size"], "csm/modeling_csm.py:CsmBackboneModel.forward": ["BaseModelOutputWithPast", "Cache", "DynamicCache", "FloatTensor", "LongTensor", "ModelBackboneModel", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "and", "arange", "attention_mask", "bool", "cache_position", "causal_mask", "class", "config", "create_causal_mask", "decoder_layer", "def", "device", "else", "embed_tokens", "exactly", "for", "forward", "get_seq_length", "hidden_states", "if", "in", "input_embeds", "input_ids", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layers", "must", "norm", "not", "num_hidden_layers", "of", "one", "or", "past_key_values", "past_seen_tokens", "position_embeddings", "position_ids", "r", "raise", "return", "rotary_emb", "self", "shape", "specify", "torch", "unsqueeze", "use_cache"], "csm/modeling_csm.py:CsmForConditionalGeneration.__init__": ["AutoModel", "Embedding", "False", "Linear", "ModelBackboneModel", "ModelDepthDecoderForCausalLM", "ModelForConditionalGeneration", "ModelGenerationMixin", "ModelPreTrainedModel", "__init__", "_from_config", "backbone_model", "bias", "class", "codec_config", "codec_model", "config", "def", "depth_decoder", "depth_decoder_config", "embed_text_tokens", "from_config", "hidden_size", "lm_head", "nn", "post_init", "self", "super", "text_vocab_size", "vocab_size"], "csm/modeling_csm.py:CsmForConditionalGeneration.get_input_embeddings": ["ModelForConditionalGeneration", "ModelGenerationMixin", "ModelPreTrainedModel", "backbone_model", "class", "def", "embed_tokens", "get_input_embeddings", "return", "self"], "csm/modeling_csm.py:CsmForConditionalGeneration.set_input_embeddings": ["ModelForConditionalGeneration", "ModelGenerationMixin", "ModelPreTrainedModel", "backbone_model", "class", "def", "embed_tokens", "self", "set_input_embeddings", "value"], "csm/modeling_csm.py:CsmForConditionalGeneration.from_pretrained": ["False", "ModelForConditionalGeneration", "ModelGenerationMixin", "ModelPreTrainedModel", "_from_model_config", "args", "attr", "class", "cls", "def", "delattr", "depth_decoder", "depth_decoder_", "depth_decoder_attrs", "else", "for", "from_pretrained", "generation_config", "get", "if", "in", "items", "kwargs", "len", "loading_info", "model", "output_loading_info", "prefix", "prefix_len", "return", "startswith", "super", "update", "value", "vars"], "csm/modeling_csm.py:CsmForConditionalGeneration.save_pretrained": ["ModelForConditionalGeneration", "ModelGenerationMixin", "ModelPreTrainedModel", "None", "args", "attr", "class", "def", "depth_decoder", "depth_decoder_", "depth_decoder_attrs", "for", "generation_config", "in", "items", "kwargs", "pop", "prefix", "save_pretrained", "self", "setattr", "super", "to_diff_dict", "transformers_version", "value"], "csm/modeling_csm.py:CsmForConditionalGeneration._merge_input_ids_with_input_values": ["ModelForConditionalGeneration", "ModelGenerationMixin", "ModelPreTrainedModel", "None", "Tensor", "True", "_merge_input_ids_with_input_values", "append", "arange", "as_tuple", "audio_batch", "audio_codes", "audio_codes_mask", "audio_embeds", "audio_eos_embeds", "audio_eos_frame_ids", "audio_eos_token_id", "audio_eos_token_mask", "audio_lengths", "audio_token_id", "audio_token_mask", "audio_tokens_list", "backbone_model", "batch_input_values", "batch_input_values_cutoffs", "batched_audio_token_ids", "class", "codebook_eos_token_id", "codebook_ids", "codec_model", "codec_outputs", "config", "def", "depth_decoder_ignore_frames_idxs", "device", "diff", "dtype", "el", "embed_text_tokens", "embed_tokens", "encode", "end_idx", "expand", "for", "functional", "get_audio_codes_mask", "i", "if", "in", "input_ids", "input_values", "input_values_cutoffs", "input_values_mask", "inputs_embeds", "is", "labels", "labels_expanded", "len", "long", "max", "max_audio_frames", "nn", "no_grad", "nonzero", "not", "num_codebooks", "ones", "pad", "range", "repeat", "return", "self", "shape", "squeeze", "stack", "start_idx", "sum", "torch", "transpose", "unsqueeze", "with", "zip"], "csm/modeling_csm.py:CsmForConditionalGeneration.prepare_inputs_for_generation": ["Cache", "FloatTensor", "LongTensor", "ModelForConditionalGeneration", "ModelGenerationMixin", "ModelPreTrainedModel", "None", "_merge_input_ids_with_input_values", "and", "attention_mask", "cache_position", "class", "def", "get", "if", "input_ids", "input_values", "input_values_cutoffs", "inputs_embeds", "is", "kwargs", "labels", "merged_inputs", "model_inputs", "ndim", "not", "past_key_values", "prepare_inputs_for_generation", "return", "self", "super", "torch", "update"], "csm/modeling_csm.py:CsmForConditionalGeneration.forward": ["Cache", "FloatTensor", "LongTensor", "ModelForConditionalGeneration", "ModelGenerationMixin", "ModelOutputWithPast", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "_merge_input_ids_with_input_values", "all", "and", "as_tuple", "attention_mask", "attentions", "backbone_hidden_states", "backbone_labels", "backbone_last_hidden_state", "backbone_last_hidden_states", "backbone_logits", "backbone_loss", "backbone_model", "backbone_outputs", "bool", "cache_position", "class", "config", "def", "depth_decoder", "depth_decoder_attentions", "depth_decoder_hidden_states", "depth_decoder_input_ids", "depth_decoder_labels", "depth_decoder_logits", "depth_decoder_loss", "depth_decoder_outputs", "depth_decoder_past_key_values", "dim", "else", "forward", "functional", "hidden_states", "if", "input_ids", "input_values", "input_values_cutoffs", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "merged_inputs", "ndim", "nn", "nonzero", "not", "num_codebooks", "pad", "past_key_values", "position_ids", "r", "return", "return_dict", "self", "slice", "slice_indices", "torch", "train_idxs", "train_mask", "tuple", "use_cache", "value", "vocab_size"], "ctrl/modeling_ctrl.py:angle_defn": ["Model_defn", "Model_rates", "d_model_size", "def", "i", "pos", "pow", "return", "torch"], "ctrl/modeling_ctrl.py:positional_encoding": ["Model_encoding", "angle_defn", "angle_rads", "arange", "cat", "cos", "cosines", "d_model_size", "def", "dim", "dtype", "int64", "pos_encoding", "position", "return", "sin", "sines", "to", "torch", "unsqueeze"], "ctrl/modeling_ctrl.py:scaled_dot_product_attention": ["Model_attention_logits", "Model_dot_product_attention", "None", "attention_mask", "attention_weights", "def", "dim", "dk", "if", "is", "k", "mask", "matmul", "matmul_qk", "nd", "not", "np", "ns", "output", "permute", "q", "return", "shape", "size", "softmax", "sqrt", "torch", "v"], "ctrl/modeling_ctrl.py:MultiHeadAttention.__init__": ["Linear", "ModelHeadAttention", "Module", "None", "Wk", "Wq", "Wv", "__init__", "class", "d_model_size", "def", "dense", "depth", "int", "layer_idx", "nn", "num_heads", "self", "super"], "ctrl/modeling_ctrl.py:MultiHeadAttention.split_into_heads": ["ModelHeadAttention", "Module", "batch_size", "class", "def", "depth", "nn", "num_heads", "permute", "reshape", "return", "self", "split_into_heads", "x"], "ctrl/modeling_ctrl.py:MultiHeadAttention.forward": ["False", "ModelHeadAttention", "Module", "None", "Wk", "Wq", "Wv", "attention_mask", "attn", "batch_size", "cache_position", "class", "d_model_size", "def", "dense", "forward", "if", "is", "k", "layer_idx", "layer_past", "mask", "nn", "not", "original_size_attention", "output", "output_attentions", "permute", "q", "reshape", "return", "scaled_attention", "scaled_dot_product_attention", "self", "shape", "split_into_heads", "update", "use_cache", "v"], "ctrl/modeling_ctrl.py:point_wise_feed_forward_network": ["Linear", "Model_wise_feed_forward_network", "ReLU", "Sequential", "d_model_size", "def", "dff", "nn", "return"], "ctrl/modeling_ctrl.py:EncoderLayer.__init__": ["Dropout", "LayerNorm", "ModelLayer", "Module", "MultiHeadAttention", "None", "__init__", "class", "d_model_size", "def", "dff", "dropout1", "dropout2", "eps", "ffn", "layer_idx", "layernorm1", "layernorm2", "multi_head_attention", "nn", "num_heads", "point_wise_feed_forward_network", "rate", "self", "super"], "ctrl/modeling_ctrl.py:EncoderLayer.forward": ["False", "ModelLayer", "Module", "None", "attention_mask", "attn_output", "attn_outputs", "cache_position", "class", "def", "dropout1", "dropout2", "ffn", "ffn_output", "forward", "layer_past", "layernorm1", "layernorm2", "mask", "multi_head_attention", "nn", "normed", "out1", "out2", "output_attentions", "outputs", "return", "self", "use_cache", "x"], "ctrl/modeling_ctrl.py:CTRLPreTrainedModel._init_weights": ["Model", "ModelModel", "PreTrainedModel", "_init_weights", "class", "config", "copy_", "d_model_size", "def", "float", "if", "init", "isinstance", "module", "n_positions", "pos_encoding", "positional_encoding", "self", "super", "torch"], "ctrl/modeling_ctrl.py:CTRLModel.__init__": ["Dropout", "Embedding", "EncoderLayer", "False", "LayerNorm", "Model", "ModelPreTrainedModel", "ModuleList", "__init__", "class", "config", "d_model_size", "def", "dff", "dropout", "embd_pdrop", "eps", "float", "for", "h", "i", "in", "layer_idx", "layer_norm_epsilon", "layernorm", "n_embd", "n_head", "n_layer", "n_positions", "nn", "num_layers", "persistent", "pos_encoding", "positional_encoding", "post_init", "range", "register_buffer", "resid_pdrop", "self", "super", "torch", "vocab_size", "w"], "ctrl/modeling_ctrl.py:CTRLModel.get_input_embeddings": ["Model", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "return", "self", "w"], "ctrl/modeling_ctrl.py:CTRLModel.set_input_embeddings": ["Model", "ModelPreTrainedModel", "class", "def", "new_embeddings", "self", "set_input_embeddings", "w"], "ctrl/modeling_ctrl.py:CTRLModel.forward": ["BaseModelOutputWithPast", "Cache", "DynamicCache", "FloatTensor", "LongTensor", "Model", "ModelPreTrainedModel", "None", "Tensor", "ValueError", "You", "all_attentions", "all_hidden_states", "and", "arange", "at", "attention_mask", "attentions", "batch_size", "be", "bool", "both", "cache_position", "cannot", "class", "config", "d_model_size", "def", "defined", "device", "dropout", "dtype", "either", "elif", "else", "enumerate", "finfo", "for", "forward", "get_seq_length", "h", "has", "have", "hidden_states", "i", "if", "in", "input_ids", "input_shape", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layer_past", "layernorm", "long", "mask", "min", "not", "np", "ones", "or", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "past_length", "pos_embeds", "pos_encoding", "position_ids", "r", "raise", "return", "return_dict", "same", "self", "seq_len", "shape", "size", "specify", "sqrt", "the", "time", "to", "token_type_embeds", "token_type_ids", "torch", "triu", "tuple", "unsqueeze", "use_cache", "use_return_dict", "v", "view", "w", "warn_if_padding_and_no_attention_mask"], "ctrl/modeling_ctrl.py:CTRLLMHeadModel.__init__": ["GenerationMixin", "Linear", "Model", "ModelModel", "ModelPreTrainedModel", "True", "__init__", "bias", "class", "config", "def", "lm_head", "n_embd", "nn", "post_init", "self", "super", "transformer", "vocab_size"], "ctrl/modeling_ctrl.py:CTRLLMHeadModel.forward": ["Cache", "CausalLMOutputWithPast", "FloatTensor", "GenerationMixin", "LongTensor", "Model", "ModelPreTrainedModel", "None", "Tensor", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "not", "output", "output_attentions", "output_hidden_states", "past_key_values", "position_ids", "r", "return", "return_dict", "self", "slice", "slice_indices", "token_type_ids", "torch", "transformer", "transformer_outputs", "tuple", "use_cache", "use_return_dict", "vocab_size"], "ctrl/modeling_ctrl.py:CTRLLMHeadModel.prepare_inputs_for_generation": ["False", "GenerationMixin", "Model", "ModelPreTrainedModel", "None", "Warning", "a", "class", "def", "else", "f", "for", "get_seq_length", "if", "in", "input", "input_ids", "is", "is_first_iteration", "items", "key", "kwargs", "model_inputs", "not", "past_key_values", "past_length", "pop", "prepare_inputs_for_generation", "print", "recognized", "remove_prefix_length", "return", "self", "shape", "token_type_ids", "use_cache", "value"], "ctrl/modeling_ctrl.py:CTRLForSequenceClassification.__init__": ["False", "Linear", "Model", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "classifier", "config", "def", "n_embd", "nn", "num_labels", "post_init", "self", "super", "transformer"], "ctrl/modeling_ctrl.py:CTRLForSequenceClassification.forward": ["BCEWithLogitsLoss", "Cache", "Cannot", "CrossEntropyLoss", "FloatTensor", "LongTensor", "MSELoss", "Model", "ModelPreTrainedModel", "None", "Results", "SequenceClassifierOutput", "Tensor", "ValueError", "__class__", "__name__", "and", "arange", "argmax", "attention_mask", "attentions", "batch", "batch_size", "be", "bool", "class", "classifier", "config", "conjunction", "def", "defined", "detect", "device", "dtype", "elif", "else", "f", "forward", "handle", "hidden_states", "if", "in", "input_ids", "inputs_embeds", "int", "int32", "is", "kwargs", "labels", "last_non_pad_token", "logger", "logits", "long", "loss", "loss_fct", "may", "multi_label_classification", "no", "non_pad_mask", "not", "num_labels", "or", "output", "output_attentions", "output_hidden_states", "pad_token_id", "padding", "past_key_values", "pooled_logits", "position_ids", "problem_type", "r", "raise", "regression", "return", "return_dict", "self", "sequence_length", "shape", "single_label_classification", "sizes", "squeeze", "to", "token", "token_indices", "token_type_ids", "tokens", "torch", "transformer", "transformer_outputs", "tuple", "unexpected", "use_cache", "use_return_dict", "using", "view", "warning_once", "will", "with"], "cvt/modeling_cvt.py:drop_path": ["False", "Model_path", "Model_prob", "Tensor", "bool", "def", "device", "div", "dtype", "float", "floor_", "if", "input", "keep_prob", "ndim", "not", "or", "output", "rand", "random_tensor", "return", "shape", "torch", "training"], "cvt/modeling_cvt.py:CvtDropPath.__init__": ["ModelDropPath", "Module", "None", "__init__", "class", "def", "drop_prob", "float", "nn", "self", "super"], "cvt/modeling_cvt.py:CvtDropPath.forward": ["ModelDropPath", "Module", "Tensor", "class", "def", "drop_path", "drop_prob", "forward", "hidden_states", "nn", "return", "self", "torch", "training"], "cvt/modeling_cvt.py:CvtDropPath.extra_repr": ["ModelDropPath", "Module", "class", "def", "drop_prob", "extra_repr", "f", "nn", "p", "return", "self", "str"], "cvt/modeling_cvt.py:CvtEmbeddings.__init__": ["Dropout", "ModelConvEmbeddings", "ModelEmbeddings", "Module", "__init__", "class", "convolution_embeddings", "def", "dropout", "dropout_rate", "embed_dim", "nn", "num_channels", "padding", "patch_size", "self", "stride", "super"], "cvt/modeling_cvt.py:CvtEmbeddings.forward": ["ModelEmbeddings", "Module", "class", "convolution_embeddings", "def", "dropout", "forward", "hidden_state", "nn", "pixel_values", "return", "self"], "cvt/modeling_cvt.py:CvtConvEmbeddings.__init__": ["Conv2d", "Iterable", "LayerNorm", "ModelConvEmbeddings", "Module", "__init__", "abc", "class", "collections", "def", "else", "embed_dim", "if", "isinstance", "kernel_size", "nn", "normalization", "num_channels", "padding", "patch_size", "projection", "self", "stride", "super"], "cvt/modeling_cvt.py:CvtConvEmbeddings.forward": ["ModelConvEmbeddings", "Module", "batch_size", "class", "def", "forward", "height", "hidden_size", "if", "nn", "normalization", "num_channels", "permute", "pixel_values", "projection", "return", "self", "shape", "view", "width"], "cvt/modeling_cvt.py:CvtSelfAttentionConvProjection.__init__": ["BatchNorm2d", "Conv2d", "False", "ModelSelfAttentionConvProjection", "Module", "__init__", "bias", "class", "convolution", "def", "embed_dim", "groups", "kernel_size", "nn", "normalization", "padding", "self", "stride", "super"], "cvt/modeling_cvt.py:CvtSelfAttentionConvProjection.forward": ["ModelSelfAttentionConvProjection", "Module", "class", "convolution", "def", "forward", "hidden_state", "nn", "normalization", "return", "self"], "cvt/modeling_cvt.py:CvtSelfAttentionLinearProjection.forward": ["ModelSelfAttentionLinearProjection", "Module", "batch_size", "class", "def", "forward", "height", "hidden_size", "hidden_state", "nn", "num_channels", "permute", "return", "self", "shape", "view", "width"], "cvt/modeling_cvt.py:CvtSelfAttentionProjection.__init__": ["ModelSelfAttentionConvProjection", "ModelSelfAttentionLinearProjection", "ModelSelfAttentionProjection", "Module", "__init__", "class", "convolution_projection", "def", "dw_bn", "embed_dim", "if", "kernel_size", "linear_projection", "nn", "padding", "projection_method", "self", "stride", "super"], "cvt/modeling_cvt.py:CvtSelfAttentionProjection.forward": ["ModelSelfAttentionProjection", "Module", "class", "convolution_projection", "def", "forward", "hidden_state", "linear_projection", "nn", "return", "self"], "cvt/modeling_cvt.py:CvtSelfAttention.__init__": ["Dropout", "Linear", "ModelSelfAttention", "ModelSelfAttentionProjection", "Module", "True", "__init__", "attention_drop_rate", "avg", "bias", "class", "convolution_projection_key", "convolution_projection_query", "convolution_projection_value", "def", "dropout", "else", "embed_dim", "if", "kernel_size", "kwargs", "linear", "nn", "num_heads", "padding_kv", "padding_q", "projection_key", "projection_method", "projection_query", "projection_value", "qkv_bias", "qkv_projection_method", "scale", "self", "stride_kv", "stride_q", "super", "with_cls_token"], "cvt/modeling_cvt.py:CvtSelfAttention.rearrange_for_multi_head_attention": ["ModelSelfAttention", "Module", "_", "batch_size", "class", "def", "embed_dim", "head_dim", "hidden_size", "hidden_state", "nn", "num_heads", "permute", "rearrange_for_multi_head_attention", "return", "self", "shape", "view"], "cvt/modeling_cvt.py:CvtSelfAttention.forward": ["ModelSelfAttention", "Module", "_", "attention_probs", "attention_score", "batch_size", "bhlk", "bhlt", "bhlv", "bhtk", "bhtv", "cat", "class", "cls_token", "context", "contiguous", "convolution_projection_key", "convolution_projection_query", "convolution_projection_value", "def", "dim", "dropout", "einsum", "embed_dim", "forward", "functional", "head_dim", "height", "hidden_size", "hidden_state", "if", "key", "nn", "num_channels", "num_heads", "permute", "projection_key", "projection_query", "projection_value", "query", "rearrange_for_multi_head_attention", "return", "scale", "self", "shape", "softmax", "split", "torch", "value", "view", "width", "with_cls_token"], "cvt/modeling_cvt.py:CvtSelfOutput.__init__": ["Dropout", "Linear", "ModelSelfOutput", "Module", "__init__", "class", "def", "dense", "drop_rate", "dropout", "embed_dim", "nn", "self", "super"], "cvt/modeling_cvt.py:CvtSelfOutput.forward": ["ModelSelfOutput", "Module", "class", "def", "dense", "dropout", "forward", "hidden_state", "input_tensor", "nn", "return", "self"], "cvt/modeling_cvt.py:CvtAttention.__init__": ["ModelAttention", "ModelSelfAttention", "ModelSelfOutput", "Module", "True", "__init__", "attention", "attention_drop_rate", "class", "def", "drop_rate", "embed_dim", "kernel_size", "nn", "num_heads", "output", "padding_kv", "padding_q", "qkv_bias", "qkv_projection_method", "self", "stride_kv", "stride_q", "super", "with_cls_token"], "cvt/modeling_cvt.py:CvtAttention.forward": ["ModelAttention", "Module", "attention", "attention_output", "class", "def", "forward", "height", "hidden_state", "nn", "output", "return", "self", "self_output", "width"], "cvt/modeling_cvt.py:CvtIntermediate.__init__": ["GELU", "Linear", "ModelIntermediate", "Module", "__init__", "activation", "class", "def", "dense", "embed_dim", "int", "mlp_ratio", "nn", "self", "super"], "cvt/modeling_cvt.py:CvtIntermediate.forward": ["ModelIntermediate", "Module", "activation", "class", "def", "dense", "forward", "hidden_state", "nn", "return", "self"], "cvt/modeling_cvt.py:CvtOutput.__init__": ["Dropout", "Linear", "ModelOutput", "Module", "__init__", "class", "def", "dense", "drop_rate", "dropout", "embed_dim", "int", "mlp_ratio", "nn", "self", "super"], "cvt/modeling_cvt.py:CvtOutput.forward": ["ModelOutput", "Module", "class", "def", "dense", "dropout", "forward", "hidden_state", "input_tensor", "nn", "return", "self"], "cvt/modeling_cvt.py:CvtLayer.__init__": ["Identity", "LayerNorm", "ModelAttention", "ModelDropPath", "ModelIntermediate", "ModelLayer", "ModelOutput", "Module", "True", "__init__", "attention", "attention_drop_rate", "class", "def", "drop_path", "drop_path_rate", "drop_prob", "drop_rate", "else", "embed_dim", "if", "intermediate", "kernel_size", "layernorm_after", "layernorm_before", "mlp_ratio", "nn", "num_heads", "output", "padding_kv", "padding_q", "qkv_bias", "qkv_projection_method", "self", "stride_kv", "stride_q", "super", "with_cls_token"], "cvt/modeling_cvt.py:CvtLayer.forward": ["ModelLayer", "Module", "attention", "attention_output", "class", "def", "drop_path", "forward", "height", "hidden_state", "intermediate", "layer_output", "layernorm_after", "layernorm_before", "nn", "output", "return", "self", "self_attention_output", "width"], "cvt/modeling_cvt.py:CvtStage.__init__": ["ModelEmbeddings", "ModelLayer", "ModelStage", "Module", "Parameter", "Sequential", "_", "__init__", "attention_drop_rate", "class", "cls_token", "config", "cpu", "def", "depth", "device", "drop_path_rate", "drop_path_rates", "drop_rate", "dropout_rate", "else", "embed_dim", "embedding", "for", "if", "in", "item", "kernel_qkv", "kernel_size", "layers", "linspace", "mlp_ratio", "nn", "num_channels", "num_heads", "padding", "padding_kv", "padding_q", "patch_padding", "patch_size", "patch_sizes", "patch_stride", "qkv_bias", "qkv_projection_method", "randn", "range", "self", "stage", "stride", "stride_kv", "stride_q", "super", "torch", "with_cls_token", "x"], "cvt/modeling_cvt.py:CvtStage.forward": ["ModelStage", "Module", "None", "batch_size", "cat", "class", "cls_token", "config", "def", "dim", "embedding", "expand", "for", "forward", "height", "hidden_state", "if", "in", "layer", "layer_outputs", "layers", "nn", "num_channels", "permute", "return", "self", "shape", "split", "stage", "torch", "view", "width"], "cvt/modeling_cvt.py:CvtEncoder.__init__": ["ModelEncoder", "ModelStage", "Module", "ModuleList", "__init__", "append", "class", "config", "def", "depth", "for", "in", "len", "nn", "range", "self", "stage_idx", "stages", "super"], "cvt/modeling_cvt.py:CvtEncoder.forward": ["BaseModelOutputWithCLSToken", "False", "ModelEncoder", "Module", "None", "True", "_", "all_hidden_states", "class", "cls_token", "cls_token_value", "def", "else", "enumerate", "for", "forward", "hidden_state", "hidden_states", "if", "in", "is", "last_hidden_state", "nn", "not", "output_hidden_states", "pixel_values", "return", "return_dict", "self", "stage_module", "stages", "tuple", "v"], "cvt/modeling_cvt.py:CvtPreTrainedModel._init_weights": ["BatchNorm2d", "Conv2d", "LayerNorm", "Linear", "ModelPreTrainedModel", "ModelStage", "None", "PreTrainedModel", "_init_weights", "bias", "class", "cls_token", "config", "def", "elif", "getattr", "if", "init", "initializer_range", "is", "isinstance", "mean", "module", "nn", "not", "num_batches_tracked", "ones_", "running_mean", "running_var", "self", "stage", "std", "trunc_normal_", "weight", "zeros_"], "cvt/modeling_cvt.py:CvtModel.__init__": ["ModelEncoder", "ModelModel", "ModelPreTrainedModel", "True", "__init__", "add_pooling_layer", "class", "config", "def", "encoder", "post_init", "r", "self", "super"], "cvt/modeling_cvt.py:CvtModel.forward": ["BaseModelOutputWithCLSToken", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "ValueError", "You", "bool", "class", "cls_token_value", "config", "def", "else", "encoder", "encoder_outputs", "forward", "have", "hidden_states", "if", "is", "kwargs", "last_hidden_state", "not", "output_hidden_states", "pixel_values", "raise", "return", "return_dict", "self", "sequence_output", "specify", "to", "torch", "tuple", "use_return_dict"], "cvt/modeling_cvt.py:CvtForImageClassification.__init__": ["False", "Identity", "LayerNorm", "Linear", "Model", "ModelForImageClassification", "ModelModel", "ModelPreTrainedModel", "__init__", "add_pooling_layer", "class", "classifier", "config", "def", "else", "embed_dim", "if", "layernorm", "nn", "num_labels", "post_init", "self", "super"], "cvt/modeling_cvt.py:CvtForImageClassification.forward": ["BCEWithLogitsLoss", "CrossEntropyLoss", "ImageClassifierOutputWithNoAttention", "MSELoss", "Model", "ModelForImageClassification", "ModelPreTrainedModel", "None", "Tensor", "and", "batch_size", "bool", "class", "classifier", "cls_token", "config", "def", "dim", "dtype", "elif", "else", "forward", "height", "hidden_states", "if", "int", "is", "kwargs", "labels", "layernorm", "logits", "long", "loss", "loss_fct", "mean", "multi_label_classification", "not", "num_channels", "num_labels", "or", "output", "output_hidden_states", "outputs", "permute", "pixel_values", "problem_type", "r", "regression", "return", "return_dict", "self", "sequence_output", "sequence_output_mean", "shape", "single_label_classification", "squeeze", "torch", "tuple", "use_return_dict", "view", "width"], "cwm/modeling_cwm.py:CwmRotaryEmbedding.__init__": ["Callable", "False", "ModelConfig", "ModelRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "cwm/modeling_cwm.py:CwmRotaryEmbedding.compute_default_rope_parameters": ["ModelConfig", "ModelRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "cwm/modeling_cwm.py:CwmRotaryEmbedding.forward": ["False", "ModelRotaryEmbedding", "Module", "None", "and", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "cwm/modeling_cwm.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "cwm/modeling_cwm.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cos", "def", "k", "k_embed", "q", "q_embed", "return", "rotate_half", "sin", "unsqueeze", "unsqueeze_dim"], "cwm/modeling_cwm.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "cwm/modeling_cwm.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value", "value_states"], "cwm/modeling_cwm.py:CwmAttention.__init__": ["False", "Linear", "ModelAttention", "ModelConfig", "Module", "None", "True", "__init__", "attention_dropout", "bias", "class", "config", "def", "else", "getattr", "hasattr", "head_dim", "hidden_size", "if", "int", "is_causal", "k_proj", "layer_idx", "layer_type", "layer_types", "nn", "num_attention_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "q_proj", "scaling", "self", "sliding_attention", "sliding_window", "super", "torch", "v_proj"], "cwm/modeling_cwm.py:CwmAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "FlashAttentionKwargs", "LongTensor", "ModelAttention", "Module", "None", "Tensor", "Unpack", "_attn_implementation", "apply_rotary_pos_emb", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "cache_kwargs", "cache_position", "class", "config", "contiguous", "cos", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "head_dim", "hidden_shape", "hidden_states", "if", "input_shape", "is", "k_proj", "key_states", "kwargs", "layer_idx", "nn", "not", "o_proj", "past_key_values", "position_embeddings", "q_proj", "query_states", "reshape", "return", "scaling", "self", "shape", "sin", "sliding_window", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "view"], "cwm/modeling_cwm.py:CwmRMSNorm.__init__": ["ModelRMSNorm", "Module", "Parameter", "__init__", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "cwm/modeling_cwm.py:CwmRMSNorm.forward": ["ModelRMSNorm", "Module", "True", "class", "def", "dtype", "float32", "forward", "hidden_states", "input_dtype", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "cwm/modeling_cwm.py:CwmRMSNorm.extra_repr": ["ModelRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "variance_epsilon", "weight"], "cwm/modeling_cwm.py:CwmMLP.__init__": ["ACT2FN", "Linear", "ModelMLP", "Module", "__init__", "act_fn", "bias", "class", "config", "def", "down_proj", "gate_proj", "hidden_act", "hidden_size", "intermediate_size", "mlp_bias", "nn", "self", "super", "up_proj"], "cwm/modeling_cwm.py:CwmMLP.forward": ["ModelMLP", "Module", "act_fn", "class", "def", "down_proj", "forward", "gate_proj", "nn", "return", "self", "up_proj", "x"], "cwm/modeling_cwm.py:CwmDecoderLayer.__init__": ["GradientCheckpointingLayer", "ModelAttention", "ModelConfig", "ModelDecoderLayer", "ModelMLP", "ModelRMSNorm", "__init__", "attention_type", "class", "config", "def", "eps", "hidden_size", "input_layernorm", "int", "layer_idx", "layer_types", "mlp", "post_attention_layernorm", "rms_norm_eps", "self", "self_attn", "super"], "cwm/modeling_cwm.py:CwmDecoderLayer.forward": ["Cache", "False", "GradientCheckpointingLayer", "LongTensor", "ModelDecoderLayer", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "attention_mask", "bool", "cache_position", "class", "def", "forward", "hidden_states", "input_layernorm", "kwargs", "mlp", "past_key_values", "position_embeddings", "position_ids", "post_attention_layernorm", "residual", "return", "self", "self_attn", "torch", "tuple", "use_cache"], "cwm/modeling_cwm.py:CwmModel.__init__": ["Embedding", "False", "ModelConfig", "ModelDecoderLayer", "ModelModel", "ModelPreTrainedModel", "ModelRMSNorm", "ModelRotaryEmbedding", "ModuleList", "__init__", "class", "config", "def", "embed_tokens", "eps", "for", "gradient_checkpointing", "hidden_size", "in", "layer_idx", "layers", "nn", "norm", "num_hidden_layers", "pad_token_id", "padding_idx", "post_init", "range", "rms_norm_eps", "rotary_emb", "self", "super", "torch", "vocab_size"], "cwm/modeling_cwm.py:CwmModel.forward": ["Cache", "DynamicCache", "FloatTensor", "LongTensor", "ModelModel", "ModelModelOutputWithPast", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "and", "arange", "attention_mask", "attention_type", "bool", "cache_position", "causal_mask_mapping", "class", "config", "copy", "create_causal_mask", "create_sliding_window_causal_mask", "decoder_layer", "def", "device", "dict", "else", "embed_tokens", "exactly", "for", "forward", "full_attention", "get_seq_length", "hidden_states", "if", "in", "input_embeds", "input_ids", "inputs_embeds", "is", "isinstance", "kwargs", "last_hidden_state", "layers", "mask_kwargs", "must", "norm", "not", "num_hidden_layers", "of", "one", "or", "past_key_values", "past_seen_tokens", "position_embeddings", "position_ids", "raise", "return", "rotary_emb", "self", "shape", "sliding_attention", "sliding_mask_kwargs", "specify", "torch", "unsqueeze", "use_cache"], "cwm/modeling_cwm.py:CwmForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "ModelForCausalLM", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "vocab_size"], "cwm/modeling_cwm.py:CwmForCausalLM.forward": ["BaseModelOutputWithPast", "Cache", "CausalLMOutputWithPast", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForCausalLM", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "outputs", "past_key_values", "position_ids", "r", "return", "self", "slice", "slice_indices", "torch", "use_cache", "vocab_size"], "d_fine/modeling_d_fine.py:multi_scale_deformable_attention_v2": ["False", "Model_scale_deformable_attention_v2", "Tensor", "_", "align_corners", "append", "arange", "attention_weights", "batch_size", "bilinear", "clamp", "concat", "contiguous", "def", "default", "device", "dim", "discrete", "elif", "enumerate", "flatten", "for", "functional", "grid_sample", "height", "hidden_dim", "if", "in", "int", "int64", "level_id", "list", "method", "mode", "nn", "num_heads", "num_levels", "num_points", "num_points_list", "num_queries", "output", "padding_mode", "permute", "repeat", "reshape", "return", "sampling_coord", "sampling_coord_x", "sampling_coord_y", "sampling_grid_l_", "sampling_grids", "sampling_idx", "sampling_locations", "sampling_value_l_", "sampling_value_list", "shape", "split", "stack", "sum", "tensor", "to", "torch", "transpose", "unsqueeze", "value", "value_l_", "value_list", "value_spatial_shapes", "view", "width", "zeros"], "d_fine/modeling_d_fine.py:DFineMultiscaleDeformableAttention.__init__": ["Linear", "Model", "ModelConfig", "Module", "_", "__init__", "attention_weights", "class", "config", "d_model", "decoder_attention_heads", "decoder_method", "decoder_n_points", "decoder_offset_scale", "def", "dtype", "else", "float32", "for", "if", "in", "isinstance", "list", "ms_deformable_attn_core", "multi_scale_deformable_attention_v2", "n", "n_heads", "n_levels", "n_points", "nn", "num_feature_levels", "num_points_list", "num_points_scale", "offset_scale", "range", "register_buffer", "sampling_offsets", "self", "sum", "super", "tensor", "torch", "total_points"], "d_fine/modeling_d_fine.py:DFineMultiscaleDeformableAttention.forward": ["F", "Last", "Make", "Model", "Module", "None", "Tensor", "ValueError", "_", "align", "and", "attention_mask", "attention_weights", "batch_size", "be", "but", "class", "d_model", "decoder_method", "def", "dim", "dtype", "elif", "else", "encoder", "encoder_hidden_states", "f", "flip", "float", "forward", "get", "hidden", "hidden_states", "if", "instead", "is", "is_torchdynamo_compiling", "length", "masked_fill", "ms_deformable_attn_core", "must", "n_heads", "n_levels", "nn", "not", "num_points_list", "num_points_scale", "num_queries", "of", "offset", "offset_normalizer", "offset_scale", "or", "output", "raise", "reference_points", "reshape", "return", "sampling_locations", "sampling_offsets", "self", "sequence", "sequence_length", "shape", "shapes", "softmax", "spatial", "spatial_shapes", "spatial_shapes_list", "states", "sum", "sure", "tensor", "the", "to", "torch", "tuple", "unsqueeze", "value", "with"], "d_fine/modeling_d_fine.py:DFineGate.__init__": ["LayerNorm", "Linear", "Model", "Module", "__init__", "class", "d_model", "def", "gate", "int", "nn", "norm", "self", "super"], "d_fine/modeling_d_fine.py:DFineGate.forward": ["Model", "Module", "Tensor", "cat", "chunk", "class", "def", "dim", "forward", "gate", "gate1", "gate2", "gate_input", "gates", "hidden_states", "nn", "norm", "return", "second_residual", "self", "sigmoid", "torch"], "d_fine/modeling_d_fine.py:DFineMultiheadAttention.__init__": ["Linear", "Model", "Module", "True", "ValueError", "__init__", "and", "be", "bias", "bool", "by", "class", "def", "divisible", "dropout", "embed_dim", "f", "float", "got", "head_dim", "if", "int", "k_proj", "must", "nn", "num_heads", "out_proj", "q_proj", "raise", "scaling", "self", "super", "v_proj"], "d_fine/modeling_d_fine.py:DFineMultiheadAttention._reshape": ["Model", "Module", "Tensor", "_reshape", "batch_size", "class", "contiguous", "def", "head_dim", "int", "nn", "num_heads", "return", "self", "seq_len", "tensor", "torch", "transpose", "view"], "d_fine/modeling_d_fine.py:DFineMultiheadAttention.with_pos_embed": ["Model", "Module", "None", "Tensor", "class", "def", "else", "if", "is", "nn", "position_embeddings", "return", "self", "tensor", "torch", "with_pos_embed"], "d_fine/modeling_d_fine.py:DFineMultiheadAttention.forward": ["Attention", "False", "Model", "Module", "None", "Tensor", "ValueError", "_reshape", "attention_mask", "attn_output", "attn_probs", "attn_weights", "attn_weights_reshaped", "batch_size", "be", "bmm", "bool", "but", "class", "def", "dim", "dropout", "dtype", "else", "embed_dim", "expand", "f", "forward", "functional", "head_dim", "hidden_states", "hidden_states_original", "if", "inf", "is", "k_proj", "key_states", "mask", "masked_fill_", "nn", "not", "num_heads", "of", "out_proj", "output_attentions", "p", "position_embeddings", "proj_shape", "q_proj", "query_states", "raise", "reshape", "return", "scaling", "self", "should", "size", "softmax", "source_len", "target_len", "torch", "training", "transpose", "tuple", "v_proj", "value_states", "view", "weights", "with_pos_embed", "zeros_like"], "d_fine/modeling_d_fine.py:DFineDecoderLayer.__init__": ["ACT2FN", "LayerNorm", "Linear", "Model", "ModelConfig", "ModelGate", "ModelMultiheadAttention", "ModelMultiscaleDeformableAttention", "Module", "__init__", "activation_dropout", "activation_fn", "attention_dropout", "class", "config", "d_model", "decoder_activation_function", "decoder_attention_heads", "decoder_ffn_dim", "def", "dropout", "embed_dim", "encoder_attn", "eps", "fc1", "fc2", "final_layer_norm", "gateway", "layer_norm_eps", "nn", "num_heads", "self", "self_attn", "self_attn_layer_norm", "super"], "d_fine/modeling_d_fine.py:DFineDecoderLayer.forward": ["Any", "False", "Model", "Module", "None", "Tensor", "activation_dropout", "activation_fn", "attention_mask", "bool", "clamp", "class", "cross_attn_weights", "def", "dropout", "else", "encoder_attention_mask", "encoder_attn", "encoder_hidden_states", "fc1", "fc2", "final_layer_norm", "forward", "functional", "gateway", "hidden_states", "hidden_states_2", "if", "is", "max", "min", "nn", "output_attentions", "outputs", "p", "position_embeddings", "reference_points", "residual", "return", "self", "self_attn", "self_attn_layer_norm", "self_attn_weights", "spatial_shapes", "spatial_shapes_list", "torch", "training", "tuple"], "d_fine/modeling_d_fine.py:DFinePreTrainedModel._init_weights": ["BatchNorm2d", "Conv2d", "LayerNorm", "Linear", "Model", "ModelDecoder", "ModelForObjectDetection", "ModelGate", "ModelLQE", "ModelModel", "ModelMultiscaleDeformableAttention", "None", "PreTrainedModel", "True", "_", "_init_weights", "abs", "and", "arange", "attention_weights", "bbox_embed", "bias", "class", "class_embed", "concat", "config", "constant_", "copy_", "cos", "def", "default_dtype", "denoising_class_embed", "dtype", "enc_score_head", "flatten", "float", "float32", "for", "gate", "get_default_dtype", "getattr", "grid_init", "hasattr", "if", "in", "init", "initializer_bias_prior_prob", "initializer_range", "int64", "is", "isinstance", "keepdim", "layer", "layers", "learn_initial_query", "log", "math", "max", "mean", "module", "n", "n_heads", "nn", "normal_", "not", "num_batches_tracked", "num_denoising", "num_labels", "num_points_list", "num_points_scale", "ones_", "or", "pi", "prior_prob", "range", "reg_conf", "reg_scale", "reshape", "running_mean", "running_var", "sampling_offsets", "scaling", "self", "sin", "stack", "std", "sum", "tensor", "thetas", "tile", "to", "torch", "up", "values", "weight", "weight_embedding", "xavier_uniform_", "zeros_"], "d_fine/modeling_d_fine.py:DFineIntegral.__init__": ["Model", "ModelConfig", "Module", "__init__", "class", "config", "def", "max_num_bins", "nn", "self", "super"], "d_fine/modeling_d_fine.py:DFineIntegral.forward": ["F", "Model", "Module", "Tensor", "_", "batch_size", "class", "def", "device", "dim", "forward", "linear", "max_num_bins", "nn", "num_queries", "pred_corners", "project", "reshape", "return", "self", "shape", "softmax", "to", "torch"], "d_fine/modeling_d_fine.py:inverse_sigmoid": ["Model_sigmoid", "clamp", "def", "eps", "log", "max", "min", "return", "torch", "x", "x1", "x2"], "d_fine/modeling_d_fine.py:weighting_function": ["Model_function", "None", "Tensor", "abs", "cat", "def", "for", "i", "in", "int", "left_values", "max_num_bins", "range", "reg_scale", "return", "right_values", "step", "torch", "up", "upper_bound1", "upper_bound2", "values", "zeros_like"], "d_fine/modeling_d_fine.py:distance2bbox": ["Model", "Tensor", "abs", "bboxes", "bottom_right_x", "bottom_right_y", "corners_to_center_format", "def", "distance", "float", "points", "reg_scale", "return", "stack", "top_left_x", "top_left_y", "torch"], "d_fine/modeling_d_fine.py:DFineDecoder.__init__": ["False", "Model", "ModelConfig", "ModelIntegral", "ModelLQE", "ModelLayer", "ModelMLP", "ModelMLPPredictionHead", "ModelPreTrainedModel", "ModuleList", "None", "Parameter", "_", "__init__", "bbox_embed", "class", "class_embed", "config", "d_model", "decoder_attention_heads", "decoder_layers", "def", "dropout", "else", "eval_idx", "for", "hidden_size", "if", "in", "integral", "layer_scale", "layers", "lqe_layers", "max_num_bins", "nn", "num_head", "num_layers", "post_init", "pre_bbox_head", "query_pos_head", "range", "reg_scale", "requires_grad", "self", "super", "tensor", "torch", "up"], "d_fine/modeling_d_fine.py:DFineDecoder.forward": ["F", "Model", "ModelOutput", "ModelPreTrainedModel", "None", "Tensor", "all_cross_attentions", "all_hidden_states", "all_self_attns", "and", "attentions", "bbox_embed", "clamp", "class", "class_embed", "config", "cross_attentions", "decoder_layer", "def", "detach", "dim", "distance2bbox", "else", "encoder_attention_mask", "encoder_hidden_states", "enumerate", "eval_idx", "for", "forward", "hidden_states", "i", "if", "in", "initial_reference_points", "inputs_embeds", "integral", "inter_ref_bbox", "intermediate", "intermediate_hidden_states", "intermediate_logits", "intermediate_predicted_corners", "intermediate_reference_points", "inverse_sigmoid", "is", "kwargs", "last_hidden_state", "layers", "level_start_index", "lqe_layers", "max", "max_num_bins", "memory_mask", "min", "new_reference_points", "not", "or", "output", "output_attentions", "output_detach", "output_hidden_states", "position_embeddings", "pre_bbox_head", "pred_corners", "pred_corners_undetach", "project", "query_pos_embed", "query_pos_head", "r", "ref_points_detach", "ref_points_initial", "ref_points_input", "reference_points", "reg_scale", "return", "return_dict", "scores", "self", "sigmoid", "spatial_shapes", "spatial_shapes_list", "stack", "torch", "training", "tuple", "unsqueeze", "up", "use_return_dict", "v", "weighting_function"], "d_fine/modeling_d_fine.py:DFineFrozenBatchNorm2d.__init__": ["Model", "Module", "__init__", "bias", "class", "def", "n", "nn", "ones", "register_buffer", "running_mean", "running_var", "self", "super", "torch", "weight", "zeros"], "d_fine/modeling_d_fine.py:DFineFrozenBatchNorm2d._load_from_state_dict": ["Model", "Module", "_load_from_state_dict", "class", "def", "del", "error_msgs", "if", "in", "local_metadata", "missing_keys", "nn", "num_batches_tracked", "num_batches_tracked_key", "prefix", "self", "state_dict", "strict", "super", "unexpected_keys"], "d_fine/modeling_d_fine.py:DFineFrozenBatchNorm2d.forward": ["Model", "Module", "bias", "class", "def", "epsilon", "forward", "nn", "reshape", "return", "rsqrt", "running_mean", "running_var", "scale", "self", "weight", "x"], "d_fine/modeling_d_fine.py:replace_batch_norm": ["BatchNorm2d", "ModelFrozenBatchNorm2d", "Model_batch_norm", "_modules", "bias", "children", "copy_", "def", "device", "for", "if", "in", "isinstance", "len", "list", "meta", "model", "module", "name", "named_children", "new_module", "nn", "num_features", "r", "running_mean", "running_var", "torch", "weight"], "d_fine/modeling_d_fine.py:DFineConvEncoder.__init__": ["Model", "Module", "__init__", "backbone", "channels", "class", "config", "def", "freeze_backbone_batch_norms", "if", "intermediate_channel_sizes", "load_backbone", "model", "nn", "no_grad", "replace_batch_norm", "self", "super", "torch", "with"], "d_fine/modeling_d_fine.py:DFineConvEncoder.forward": ["Model", "Module", "None", "Tensor", "append", "bool", "class", "def", "feature_map", "feature_maps", "features", "float", "for", "forward", "functional", "in", "interpolate", "mask", "model", "nn", "out", "pixel_mask", "pixel_values", "return", "self", "shape", "size", "to", "torch"], "d_fine/modeling_d_fine.py:get_contrastive_denoising_training_group": ["Model_contrastive_denoising_training_group", "None", "attn_mask", "batch_size", "bool", "box_noise_scale", "boxes", "center_to_corners_format", "class_embed", "class_labels", "clip_", "corners_to_center_format", "def", "denoise_positive_idx", "denoising_meta_values", "device", "diff", "dn_num_group", "dn_num_split", "dn_positive_idx", "dtype", "else", "float", "for", "full", "i", "idx_block_end", "idx_block_start", "if", "in", "inf", "input_query_bbox", "input_query_class", "int32", "inverse_sigmoid", "known_bbox", "label_noise_ratio", "len", "mask", "max", "max_gt_num", "min", "n", "negative_gt_mask", "new_label", "nonzero", "num_classes", "num_denoising_queries", "num_ground_truths", "num_groups_denoising_queries", "num_gt", "num_queries", "pad_gt_mask", "positive_gt_mask", "rand_like", "rand_part", "rand_sign", "randint_like", "range", "return", "split", "squeeze", "t", "tarModel_size", "tarModels", "tile", "torch", "torch_int", "where", "zeros"], "d_fine/modeling_d_fine.py:DFineModel.__init__": ["BatchNorm2d", "Conv2d", "Embedding", "False", "Identity", "LayerNorm", "Linear", "Model", "ModelConfig", "ModelConvEncoder", "ModelDecoder", "ModelHybridEncoder", "ModelMLPPredictionHead", "ModelPreTrainedModel", "ModuleList", "Sequential", "_", "__init__", "anchor_image_size", "anchors", "append", "backbone", "batch_norm_eps", "batchnorm", "bias", "class", "config", "conv", "d_model", "decoder", "decoder_in_channels", "decoder_input_proj", "decoder_input_proj_list", "def", "denoising_class_embed", "dtype", "else", "enc_bbox_head", "enc_output", "enc_score_head", "encoder", "encoder_hidden_dim", "encoder_input_proj", "encoder_input_proj_list", "eps", "for", "generate_anchors", "hidden_size", "if", "in", "in_channels", "intermediate_channel_sizes", "kernel_size", "layer_norm_eps", "learn_initial_query", "len", "nn", "num_backbone_outs", "num_denoising", "num_feature_levels", "num_labels", "num_layers", "num_queries", "padding", "padding_idx", "post_init", "range", "self", "stride", "super", "valid_mask", "weight_embedding"], "d_fine/modeling_d_fine.py:DFineModel.freeze_backbone": ["False", "Model", "ModelPreTrainedModel", "backbone", "class", "def", "for", "freeze_backbone", "in", "param", "parameters", "requires_grad_", "self"], "d_fine/modeling_d_fine.py:DFineModel.unfreeze_backbone": ["Model", "ModelPreTrainedModel", "True", "backbone", "class", "def", "for", "in", "param", "parameters", "requires_grad_", "self", "unfreeze_backbone"], "d_fine/modeling_d_fine.py:DFineModel.generate_anchors": ["Model", "ModelPreTrainedModel", "None", "True", "all", "anchor_image_size", "anchors", "append", "arange", "class", "concat", "config", "cpu", "def", "device", "dtype", "end", "enumerate", "eps", "feat_strides", "finfo", "float32", "for", "generate_anchors", "grid_size", "grid_x", "grid_xy", "grid_y", "height", "if", "ij", "in", "indexing", "int", "is", "keepdim", "level", "log", "max", "meshgrid", "ones_like", "reshape", "return", "s", "self", "spatial_shapes", "stack", "tensor", "to", "torch", "unsqueeze", "valid_mask", "wh", "where", "width"], "d_fine/modeling_d_fine.py:DFineModel.forward": ["BaseModelOutput", "F", "FloatTensor", "LongTensor", "Model", "ModelOutput", "ModelPreTrainedModel", "None", "_", "_len_sources", "anchor_image_size", "anchors", "and", "append", "attention_mask", "attentions", "backbone", "batch_size", "bool", "box_noise_scale", "cat", "class", "class_embed", "concat", "config", "cross_attentions", "cumsum", "decoder", "decoder_attentions", "decoder_hidden_states", "decoder_input_proj", "decoder_inputs_embeds", "decoder_outputs", "def", "denoising_bbox_unact", "denoising_class", "denoising_class_embed", "denoising_meta_values", "detach", "device", "dict", "dim", "dn_outputs", "dtype", "elif", "else", "empty", "enc_bbox_head", "enc_output", "enc_outputs", "enc_outputs_class", "enc_outputs_coord_logits", "enc_score_head", "enc_topk_bboxes", "enc_topk_logits", "encoder", "encoder_attention_mask", "encoder_attentions", "encoder_hidden_states", "encoder_input_proj", "encoder_last_hidden_state", "encoder_outputs", "enumerate", "features", "flatten", "for", "forward", "gather", "generate_anchors", "get_contrastive_denoising_training_group", "height", "hidden_states", "i", "if", "in", "index", "init_reference_points", "initial_reference_points", "inputs_embeds", "intermediate_hidden_states", "intermediate_logits", "intermediate_predicted_corners", "intermediate_reference_points", "is", "isinstance", "kwargs", "label_noise_ratio", "labels", "last_hidden_state", "learn_initial_query", "len", "level", "level_start_index", "list", "long", "mask", "max", "memory", "new_zeros", "not", "num_channels", "num_classes", "num_denoising", "num_denoising_queries", "num_feature_levels", "num_labels", "num_queries", "ones", "or", "output_attentions", "output_hidden_states", "output_memory", "pixel_mask", "pixel_values", "prod", "proj_feats", "r", "range", "reference_points", "reference_points_unact", "repeat", "return", "return_dict", "self", "shape", "sigmoid", "source", "source_flatten", "sources", "spatial_shapes", "spatial_shapes_list", "spatial_shapes_tuple", "target", "targets", "tile", "to", "topk", "topk_ind", "torch", "training", "transpose", "tuple", "tuple_outputs", "unsqueeze", "use_return_dict", "valid_mask", "value", "values", "weight_embedding", "width"], "d_fine/modeling_d_fine.py:DFineForObjectDetection.__init__": ["Linear", "Model", "ModelConfig", "ModelMLP", "ModelModel", "ModelPreTrainedModel", "ModuleList", "_", "__init__", "bbox_embed", "class", "class_embed", "config", "d_model", "decoder", "decoder_layers", "def", "else", "eval_idx", "for", "hidden_size", "if", "in", "layer_scale", "max_num_bins", "model", "nn", "num_labels", "num_pred", "post_init", "range", "round", "scaled_dim", "self", "super"], "d_fine/modeling_d_fine.py:DFineForObjectDetection._set_aux_loss": ["Model", "ModelPreTrainedModel", "_set_aux_loss", "a", "b", "class", "def", "for", "in", "logits", "outputs_class", "outputs_coord", "pred_boxes", "return", "self", "zip"], "d_fine/modeling_d_fine.py:DFineForObjectDetection.forward": ["FloatTensor", "LongTensor", "Model", "ModelObjectDetectionOutput", "ModelPreTrainedModel", "None", "auxiliary_outputs", "bool", "class", "config", "cross_attentions", "decoder_attentions", "decoder_hidden_states", "decoder_inputs_embeds", "def", "denoising_meta_values", "device", "dict", "else", "enc_outputs_class", "enc_outputs_coord_logits", "enc_topk_bboxes", "enc_topk_logits", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "forward", "if", "init_reference_points", "initial_reference_points", "inputs_embeds", "intermediate_hidden_states", "intermediate_logits", "intermediate_predicted_corners", "intermediate_reference_points", "is", "kwargs", "labels", "last_hidden_state", "list", "logits", "loss", "loss_dict", "loss_function", "model", "not", "output", "output_attentions", "output_hidden_states", "outputs", "outputs_class", "outputs_coord", "pixel_mask", "pixel_values", "pred_boxes", "predicted_corners", "r", "return", "return_dict", "self", "torch", "training", "tuple", "use_return_dict"], "d_fine/modeling_d_fine.py:DFineMLPPredictionHead.__init__": ["Linear", "Model", "Module", "ModuleList", "__init__", "class", "config", "d_model", "def", "for", "h", "in", "input_dim", "k", "layers", "n", "nn", "num_layers", "output_dim", "self", "super", "zip"], "d_fine/modeling_d_fine.py:DFineMLPPredictionHead.forward": ["Model", "Module", "class", "def", "else", "enumerate", "for", "forward", "functional", "i", "if", "in", "layer", "layers", "nn", "num_layers", "relu", "return", "self", "x"], "d_fine/modeling_d_fine.py:DFineMLP.__init__": ["ACT2CLS", "Linear", "Model", "Module", "ModuleList", "__init__", "act", "class", "def", "for", "hidden_dim", "hidden_dims", "in", "in_dim", "input_dim", "input_dims", "int", "layers", "nn", "num_layers", "out_dim", "output_dim", "output_dims", "relu", "self", "str", "super", "zip"], "d_fine/modeling_d_fine.py:DFineMLP.forward": ["Model", "Module", "Tensor", "act", "class", "def", "else", "enumerate", "for", "forward", "i", "if", "in", "layer", "layers", "nn", "num_layers", "return", "self", "stat_features", "torch"], "d_fine/modeling_d_fine.py:DFineLQE.__init__": ["Model", "ModelConfig", "ModelMLP", "Module", "__init__", "class", "config", "def", "lqe_hidden_dim", "lqe_layers", "max_num_bins", "nn", "reg_conf", "self", "super", "top_prob_values"], "d_fine/modeling_d_fine.py:DFineLQE.forward": ["F", "Model", "Module", "Tensor", "True", "_", "batch_size", "cat", "class", "def", "dim", "forward", "keepdim", "length", "max_num_bins", "mean", "nn", "pred_corners", "prob", "prob_topk", "quality_score", "reg_conf", "reshape", "return", "scores", "self", "size", "softmax", "stat", "top_prob_values", "topk", "torch"], "d_fine/modeling_d_fine.py:DFineConvNormLayer.__init__": ["ACT2CLS", "BatchNorm2d", "Conv2d", "False", "Identity", "Model", "ModelConfig", "Module", "None", "__init__", "activation", "batch_norm_eps", "bias", "class", "config", "conv", "def", "else", "groups", "if", "in_channels", "int", "is", "kernel_size", "nn", "norm", "out_channels", "padding", "self", "str", "stride", "super"], "d_fine/modeling_d_fine.py:DFineConvNormLayer.forward": ["Model", "Module", "activation", "class", "conv", "def", "forward", "hidden_state", "nn", "norm", "return", "self"], "d_fine/modeling_d_fine.py:DFineRepVggBlock.__init__": ["ACT2CLS", "Identity", "Model", "ModelConfig", "ModelConvNormLayer", "Module", "None", "__init__", "activation", "activation_function", "class", "config", "conv1", "conv2", "def", "else", "hidden_channels", "if", "in_channels", "int", "is", "nn", "out_channels", "padding", "self", "super"], "d_fine/modeling_d_fine.py:DFineRepVggBlock.forward": ["Model", "Module", "activation", "class", "conv1", "conv2", "def", "forward", "nn", "return", "self", "x", "y"], "d_fine/modeling_d_fine.py:DFineCSPRepLayer.__init__": ["Identity", "Model", "ModelConfig", "ModelConvNormLayer", "ModelRepVggBlock", "Module", "ModuleList", "_", "__init__", "activation", "activation_function", "bottlenecks", "class", "config", "conv1", "conv2", "conv3", "def", "else", "expansion", "float", "for", "hidden_channels", "if", "in", "in_channels", "int", "nn", "num_blocks", "out_channels", "range", "self", "super"], "d_fine/modeling_d_fine.py:DFineCSPRepLayer.forward": ["Model", "Module", "Tensor", "bottleneck", "bottlenecks", "class", "conv1", "conv2", "conv3", "def", "for", "forward", "hidden_state", "hidden_state_1", "hidden_state_2", "hidden_state_3", "in", "nn", "return", "self", "torch"], "d_fine/modeling_d_fine.py:DFineRepNCSPELAN4.__init__": ["Model", "ModelCSPRepLayer", "ModelConfig", "ModelConvNormLayer", "Module", "__init__", "act", "activation", "class", "config", "conv1", "conv1_dim", "conv2", "conv2_dim", "conv3", "conv3_dim", "conv4", "conv4_dim", "conv_dim", "csp_rep1", "csp_rep2", "def", "encoder_hidden_dim", "hidden_expansion", "int", "nn", "num_blocks", "numb_blocks", "round", "self", "silu", "str", "super"], "d_fine/modeling_d_fine.py:DFineRepNCSPELAN4.forward": ["Model", "Module", "Tensor", "branch1", "branch2", "cat", "class", "conv1", "conv2", "conv3", "conv4", "conv_dim", "csp_rep1", "csp_rep2", "def", "extend", "forward", "input_features", "list", "merged_features", "nn", "return", "self", "split", "split_features", "torch"], "d_fine/modeling_d_fine.py:DFineSCDown.__init__": ["Model", "ModelConfig", "ModelConvNormLayer", "Module", "__init__", "class", "config", "conv1", "conv2", "def", "encoder_hidden_dim", "int", "kernel_size", "nn", "self", "stride", "super"], "d_fine/modeling_d_fine.py:DFineSCDown.forward": ["Model", "Module", "Tensor", "class", "conv1", "conv2", "def", "forward", "input_features", "nn", "return", "self", "torch"], "d_fine/modeling_d_fine.py:DFineEncoderLayer.__init__": ["ACT2FN", "LayerNorm", "Linear", "Model", "ModelConfig", "ModelMultiheadAttention", "Module", "__init__", "activation_dropout", "activation_fn", "class", "config", "def", "dropout", "embed_dim", "encoder_activation_function", "encoder_ffn_dim", "encoder_hidden_dim", "eps", "fc1", "fc2", "final_layer_norm", "layer_norm_eps", "nn", "normalize_before", "num_attention_heads", "num_heads", "self", "self_attn", "self_attn_layer_norm", "super"], "d_fine/modeling_d_fine.py:DFineEncoderLayer.forward": ["False", "Model", "Module", "None", "Tensor", "activation_dropout", "activation_fn", "any", "attention_mask", "attn_weights", "bool", "clamp", "clamp_value", "class", "def", "dropout", "dtype", "fc1", "fc2", "final_layer_norm", "finfo", "forward", "functional", "hidden_states", "if", "isinf", "isnan", "kwargs", "max", "min", "nn", "normalize_before", "not", "or", "output_attentions", "outputs", "p", "position_embeddings", "residual", "return", "self", "self_attn", "self_attn_layer_norm", "torch", "training"], "d_fine/modeling_d_fine.py:DFineEncoder.__init__": ["Model", "ModelConfig", "ModelLayer", "Module", "ModuleList", "_", "__init__", "class", "config", "def", "encoder_layers", "for", "in", "layers", "nn", "range", "self", "super"], "d_fine/modeling_d_fine.py:DFineEncoder.forward": ["False", "Model", "Module", "None", "Tensor", "attention_mask", "bool", "class", "def", "for", "forward", "hidden_states", "in", "layer", "layers", "nn", "output_attentions", "pos_embed", "position_embeddings", "return", "self", "src", "src_mask", "torch"], "d_fine/modeling_d_fine.py:DFineHybridEncoder.__init__": ["Model", "ModelConfig", "ModelConvNormLayer", "ModelEncoder", "ModelRepNCSPELAN4", "ModelSCDown", "Module", "ModuleList", "_", "__init__", "append", "class", "config", "def", "depth_mult", "downsample_convs", "encode_proj_layers", "encoder", "encoder_hidden_dim", "encoder_in_channels", "eval_size", "feat_strides", "for", "fpn_blocks", "fpn_layer", "in", "in_channels", "lateral_convs", "lateral_layer", "len", "nn", "num_blocks", "num_fpn_stages", "numb_blocks", "out_channels", "out_strides", "pan_blocks", "positional_encoding_temperature", "range", "round", "self", "super"], "d_fine/modeling_d_fine.py:DFineHybridEncoder.build_2d_sincos_position_embedding": ["Embed", "Model", "Module", "None", "ValueError", "arange", "be", "build_2d_sincos_position_embedding", "by", "class", "concat", "cos", "cpu", "def", "device", "dim", "dimension", "divisible", "dtype", "embed_dim", "embedding", "flatten", "float32", "for", "grid_h", "grid_w", "height", "if", "indexing", "meshgrid", "must", "nn", "omega", "out_h", "out_w", "pos_dim", "position", "raise", "return", "sin", "temperature", "to", "torch", "torch_int", "width", "xy"], "d_fine/modeling_d_fine.py:DFineHybridEncoder.forward": ["BaseModelOutput", "F", "Model", "Module", "None", "all_attentions", "append", "attention_mask", "attentions", "backbone_feature_map", "build_2d_sincos_position_embedding", "class", "concat", "config", "contiguous", "def", "device", "dim", "downsample_conv", "downsample_convs", "downsampled_feature_map", "dtype", "else", "enc_ind", "encode_proj_layers", "encoder", "encoder_hidden_dim", "encoder_layers", "encoder_states", "enumerate", "eval_size", "flatten", "for", "forward", "fpn_block", "fpn_blocks", "fpn_feature_map", "fpn_feature_maps", "fused_feature_map", "height", "hidden_states", "i", "idx", "if", "in", "inputs_embeds", "interpolate", "is", "last_hidden_state", "lateral_conv", "lateral_convs", "layer_outputs", "level_start_index", "mode", "nearest", "new_fpn_feature_map", "new_pan_feature_map", "nn", "not", "num_fpn_stages", "or", "output_attentions", "output_hidden_states", "pan_block", "pan_blocks", "pan_feature_maps", "permute", "pos_embed", "position_embeddings", "positional_encoding_temperature", "r", "reshape", "return", "return_dict", "reverse", "scale_factor", "self", "shape", "spatial_shapes", "src_flatten", "top_fpn_feature_map", "top_pan_feature_map", "torch", "training", "tuple", "use_return_dict", "v", "valid_ratios", "width", "zip"], "dab_detr/modeling_dab_detr.py:DabDetrFrozenBatchNorm2d.__init__": ["ModelFrozenBatchNorm2d", "Module", "__init__", "bias", "class", "def", "n", "nn", "ones", "register_buffer", "running_mean", "running_var", "self", "super", "torch", "weight", "zeros"], "dab_detr/modeling_dab_detr.py:DabDetrFrozenBatchNorm2d._load_from_state_dict": ["ModelFrozenBatchNorm2d", "Module", "_load_from_state_dict", "class", "def", "del", "error_msgs", "if", "in", "local_metadata", "missing_keys", "nn", "num_batches_tracked", "num_batches_tracked_key", "prefix", "self", "state_dict", "strict", "super", "unexpected_keys"], "dab_detr/modeling_dab_detr.py:DabDetrFrozenBatchNorm2d.forward": ["ModelFrozenBatchNorm2d", "Module", "bias", "class", "def", "epsilon", "forward", "nn", "reshape", "return", "rsqrt", "running_mean", "running_var", "scale", "self", "weight", "x"], "dab_detr/modeling_dab_detr.py:replace_batch_norm": ["BatchNorm2d", "ModelFrozenBatchNorm2d", "Model_batch_norm", "_modules", "bias", "children", "copy_", "def", "device", "for", "if", "in", "isinstance", "len", "list", "meta", "model", "module", "name", "named_children", "new_module", "nn", "num_features", "r", "running_mean", "running_var", "torch", "weight"], "dab_detr/modeling_dab_detr.py:DabDetrConvEncoder.__init__": ["ModelConfig", "ModelConvEncoder", "Module", "__init__", "backbone", "channels", "class", "config", "def", "intermediate_channel_sizes", "load_backbone", "model", "nn", "no_grad", "replace_batch_norm", "self", "super", "torch", "with"], "dab_detr/modeling_dab_detr.py:DabDetrConvEncoder.forward": ["ModelConvEncoder", "Module", "None", "Tensor", "append", "bool", "class", "def", "feature_map", "feature_maps", "features", "float", "for", "forward", "functional", "in", "interpolate", "mask", "model", "nn", "out", "pixel_mask", "pixel_values", "return", "self", "shape", "size", "to", "torch"], "dab_detr/modeling_dab_detr.py:DabDetrConvModel.__init__": ["ModelConvModel", "Module", "__init__", "class", "conv_encoder", "def", "nn", "position_embedding", "self", "super"], "dab_detr/modeling_dab_detr.py:DabDetrConvModel.forward": ["ModelConvModel", "Module", "append", "class", "conv_encoder", "def", "dtype", "feature_map", "for", "forward", "in", "mask", "nn", "out", "pixel_mask", "pixel_values", "pos", "position_embedding", "return", "self", "to"], "dab_detr/modeling_dab_detr.py:DabDetrSinePositionEmbedding.__init__": ["ModelConfig", "ModelSinePositionEmbedding", "Module", "None", "__init__", "class", "config", "def", "embedding_dim", "hidden_size", "if", "is", "math", "nn", "pi", "scale", "self", "sine_position_embedding_scale", "super", "temperature_height", "temperature_width"], "dab_detr/modeling_dab_detr.py:DabDetrSinePositionEmbedding.forward": ["ModelSinePositionEmbedding", "Module", "No", "None", "ValueError", "arange", "cat", "class", "copy_", "cos", "cumsum", "def", "device", "dim", "dim_tx", "dim_ty", "dtype", "embedding_dim", "flatten", "float32", "forward", "if", "is", "mask", "mul_", "nn", "permute", "pixel", "pixel_mask", "pixel_values", "pos", "pos_x", "pos_y", "provided", "raise", "return", "scale", "self", "sin", "stack", "temperature_height", "temperature_width", "torch", "x_embed", "y_embed"], "dab_detr/modeling_dab_detr.py:gen_sine_position_embeddings": ["Model_sine_position_embeddings", "None", "Unknown", "ValueError", "arange", "cat", "cos", "def", "device", "dim", "dim_t", "div", "dtype", "else", "f", "flatten", "float32", "floor", "h_embed", "hidden_size", "if", "math", "pi", "pos", "pos_h", "pos_tensor", "pos_w", "pos_x", "pos_y", "raise", "return", "rounding_mode", "scale", "shape", "sin", "size", "stack", "to", "torch", "w_embed", "x_embed", "y_embed"], "dab_detr/modeling_dab_detr.py:inverse_sigmoid": ["Model_sigmoid", "clamp", "def", "eps", "log", "max", "min", "return", "torch", "x", "x1", "x2"], "dab_detr/modeling_dab_detr.py:DetrAttention.__init__": ["Linear", "ModelAttention", "ModelConfig", "Module", "True", "ValueError", "__init__", "and", "attention_dropout", "be", "bias", "bool", "by", "class", "config", "def", "divisible", "encoder_attention_heads", "f", "got", "head_dim", "hidden_size", "if", "k_proj", "must", "nn", "num_heads", "out_proj", "q_proj", "raise", "scaling", "self", "super", "v_proj"], "dab_detr/modeling_dab_detr.py:DetrAttention.forward": ["False", "ModelAttention", "Module", "None", "Tensor", "ValueError", "attention_dropout", "attention_mask", "attn_output", "attn_weights", "batch_size", "be", "bool", "but", "class", "contiguous", "def", "dim", "dropout", "dtype", "embed_dim", "f", "float32", "forward", "functional", "head_dim", "hidden_states", "hidden_states_original", "if", "is", "k_proj", "key_states", "key_value_states", "matmul", "nn", "not", "num_heads", "object_queries", "of", "out_proj", "output_attentions", "p", "q_len", "q_proj", "query_states", "raise", "reshape", "return", "scaling", "self", "should", "size", "softmax", "to", "torch", "training", "transpose", "tuple", "v_proj", "value_states", "view"], "dab_detr/modeling_dab_detr.py:DabDetrAttention.__init__": ["False", "Linear", "ModelAttention", "ModelConfig", "Module", "True", "ValueError", "__init__", "and", "attention_dropout", "attention_head_dim", "attention_heads", "be", "bias", "bool", "by", "class", "config", "decoder_attention_heads", "def", "divisible", "else", "embed_dim", "f", "got", "hidden_size", "if", "is_cross", "must", "nn", "num_heads", "output_dim", "output_proj", "raise", "scaling", "self", "super", "values_head_dim"], "dab_detr/modeling_dab_detr.py:DabDetrAttention.forward": ["ModelAttention", "Module", "None", "Tensor", "ValueError", "_", "attention_dropout", "attention_head_dim", "attention_heads", "attention_mask", "attn_output", "attn_probs", "attn_weights", "batch_size", "be", "bool", "but", "class", "contiguous", "def", "dim", "dropout", "dtype", "f", "float32", "forward", "functional", "hidden_states", "if", "is", "key_states", "matmul", "nn", "not", "of", "output_attentions", "output_dim", "output_proj", "p", "q_len", "query_states", "raise", "reshape", "return", "scaling", "self", "should", "size", "softmax", "to", "torch", "training", "transpose", "tuple", "value_states", "values_head_dim", "view"], "dab_detr/modeling_dab_detr.py:DabDetrDecoderLayerSelfAttention.__init__": ["LayerNorm", "Linear", "ModelAttention", "ModelConfig", "ModelDecoderLayerSelfAttention", "Module", "__init__", "class", "config", "def", "dropout", "hidden_size", "nn", "self", "self_attn", "self_attn_key_content_proj", "self_attn_key_pos_proj", "self_attn_layer_norm", "self_attn_query_content_proj", "self_attn_query_pos_proj", "self_attn_value_proj", "super"], "dab_detr/modeling_dab_detr.py:DabDetrDecoderLayerSelfAttention.forward": ["ModelDecoderLayerSelfAttention", "Module", "None", "Tensor", "True", "attention_mask", "attn_weights", "bool", "class", "def", "dropout", "forward", "functional", "hidden_states", "key", "key_content", "key_pos", "key_states", "nn", "output_attentions", "p", "query", "query_content", "query_pos", "query_position_embeddings", "residual", "return", "self", "self_attn", "self_attn_key_content_proj", "self_attn_key_pos_proj", "self_attn_layer_norm", "self_attn_query_content_proj", "self_attn_query_pos_proj", "self_attn_value_proj", "torch", "training", "value", "value_states"], "dab_detr/modeling_dab_detr.py:DabDetrDecoderLayerCrossAttention.__init__": ["False", "LayerNorm", "Linear", "ModelAttention", "ModelConfig", "ModelDecoderLayerCrossAttention", "Module", "None", "True", "__init__", "and", "bool", "class", "config", "cross_attn", "cross_attn_key_content_proj", "cross_attn_key_pos_proj", "cross_attn_layer_norm", "cross_attn_query_content_proj", "cross_attn_query_pos_proj", "cross_attn_query_pos_sine_proj", "cross_attn_value_proj", "decoder_attention_heads", "def", "dropout", "hidden_size", "if", "is_cross", "is_first", "keep_query_pos", "nn", "not", "self", "super"], "dab_detr/modeling_dab_detr.py:DabDetrDecoderLayerCrossAttention.forward": ["ModelDecoderLayerCrossAttention", "Module", "None", "Tensor", "_", "attention_mask", "batch_size", "bool", "cat", "class", "cross_attn", "cross_attn_key_content_proj", "cross_attn_key_pos_proj", "cross_attn_layer_norm", "cross_attn_query_content_proj", "cross_attn_query_pos_proj", "cross_attn_query_pos_sine_proj", "cross_attn_value_proj", "cross_attn_weights", "decoder_attention_heads", "def", "dim", "dropout", "else", "encoder_attention_mask", "encoder_hidden_states", "forward", "functional", "height_width", "hidden_states", "if", "is", "is_first", "keep_query_pos", "key", "key_content", "key_pos", "key_states", "n_model", "nn", "not", "num_queries", "object_queries", "or", "output_attentions", "p", "query", "query_content", "query_pos", "query_position_embeddings", "query_sine_embed", "residual", "return", "self", "shape", "torch", "training", "value", "value_states", "view"], "dab_detr/modeling_dab_detr.py:DabDetrDecoderLayerFFN.__init__": ["ACT2FN", "LayerNorm", "Linear", "ModelConfig", "ModelDecoderLayerFFN", "Module", "__init__", "activation_dropout", "activation_fn", "activation_function", "class", "config", "decoder_ffn_dim", "def", "dropout", "fc1", "fc2", "final_layer_norm", "hidden_size", "keep_query_pos", "nn", "self", "super"], "dab_detr/modeling_dab_detr.py:DabDetrDecoderLayerFFN.forward": ["ModelDecoderLayerFFN", "Module", "Tensor", "activation_dropout", "activation_fn", "class", "def", "dropout", "fc1", "fc2", "final_layer_norm", "forward", "functional", "hidden_states", "nn", "p", "residual", "return", "self", "torch", "training"], "dab_detr/modeling_dab_detr.py:DabDetrEncoderLayer.__init__": ["ACT2FN", "DetrAttention", "GradientCheckpointingLayer", "LayerNorm", "Linear", "ModelConfig", "ModelEncoderLayer", "__init__", "activation_fn", "activation_function", "class", "config", "def", "dropout", "encoder_ffn_dim", "fc1", "fc2", "final_layer_norm", "hidden_size", "nn", "self", "self_attn", "self_attn_layer_norm", "super"], "dab_detr/modeling_dab_detr.py:DabDetrEncoderLayer.forward": ["GradientCheckpointingLayer", "ModelEncoderLayer", "None", "Tensor", "activation_fn", "attention_mask", "attn_weights", "bool", "class", "def", "dropout", "fc1", "fc2", "final_layer_norm", "forward", "functional", "hidden_states", "if", "nn", "object_queries", "output_attentions", "outputs", "p", "residual", "return", "self", "self_attn", "self_attn_layer_norm", "torch", "training"], "dab_detr/modeling_dab_detr.py:DabDetrDecoderLayer.__init__": ["False", "GradientCheckpointingLayer", "ModelConfig", "ModelDecoderLayer", "ModelDecoderLayerCrossAttention", "ModelDecoderLayerFFN", "ModelDecoderLayerSelfAttention", "__init__", "bool", "class", "config", "cross_attn", "def", "is_first", "mlp", "self", "self_attn", "super"], "dab_detr/modeling_dab_detr.py:DabDetrDecoderLayer.forward": ["GradientCheckpointingLayer", "ModelDecoderLayer", "None", "Tensor", "attention_mask", "bool", "class", "cross_attn", "cross_attn_weights", "def", "encoder_attention_mask", "encoder_hidden_states", "forward", "hidden_states", "if", "mlp", "object_queries", "output_attentions", "outputs", "query_position_embeddings", "query_sine_embed", "return", "self", "self_attn", "self_attn_weights", "torch"], "dab_detr/modeling_dab_detr.py:DabDetrMLP.__init__": ["Linear", "ModelMLP", "Module", "ModuleList", "__init__", "class", "def", "for", "h", "hidden_dim", "in", "input_dim", "k", "layers", "n", "nn", "num_layers", "output_dim", "self", "super", "zip"], "dab_detr/modeling_dab_detr.py:DabDetrMLP.forward": ["ModelMLP", "Module", "class", "def", "else", "enumerate", "for", "forward", "functional", "i", "if", "in", "input_tensor", "layer", "layers", "nn", "num_layers", "relu", "return", "self"], "dab_detr/modeling_dab_detr.py:DabDetrPreTrainedModel._init_weights": ["Conv2d", "Embedding", "False", "LayerNorm", "Linear", "ModelForObjectDetection", "ModelMHAttentionMap", "ModelPreTrainedModel", "None", "PReLU", "PreTrainedModel", "_init_weights", "_is_hf_initialized", "and", "bbox_predictor", "bias", "bias_value", "class", "class_embed", "config", "constant_", "def", "elif", "gain", "getattr", "if", "init", "init_std", "init_xavier_std", "initializer_bias_prior_prob", "is", "isinstance", "k_linear", "layers", "log", "math", "mean", "module", "nn", "normal_", "not", "num_labels", "ones_", "or", "padding_idx", "prior_prob", "q_linear", "reset_parameters", "self", "std", "weight", "xavier_std", "xavier_uniform_", "zeros_"], "dab_detr/modeling_dab_detr.py:DabDetrEncoder.__init__": ["False", "LayerNorm", "ModelConfig", "ModelEncoder", "ModelEncoderLayer", "ModelMLP", "ModelPreTrainedModel", "ModuleList", "None", "_", "__init__", "class", "config", "def", "dropout", "else", "encoder_layers", "for", "gradient_checkpointing", "hidden_size", "if", "in", "layers", "nn", "norm", "normalize_before", "post_init", "query_scale", "range", "self", "super"], "dab_detr/modeling_dab_detr.py:DabDetrEncoder.forward": ["BaseModelOutput", "ModelEncoder", "ModelPreTrainedModel", "None", "_prepare_4d_attention_mask", "all_attentions", "attention_mask", "attentions", "bool", "class", "config", "def", "dtype", "else", "encoder_layer", "encoder_states", "for", "forward", "hidden_states", "if", "in", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layer_outputs", "layers", "norm", "not", "object_queries", "output_attentions", "output_hidden_states", "pos_scales", "query_scale", "r", "return", "return_dict", "scaled_object_queries", "self", "tuple", "use_return_dict", "v"], "dab_detr/modeling_dab_detr.py:DabDetrDecoder.__init__": ["False", "LayerNorm", "ModelConfig", "ModelDecoder", "ModelDecoderLayer", "ModelMLP", "ModelPreTrainedModel", "ModuleList", "None", "__init__", "bbox_embed", "class", "config", "decoder_layers", "def", "dropout", "for", "gradient_checkpointing", "hidden_size", "in", "is_first", "layer_id", "layernorm", "layers", "nn", "num_layers", "post_init", "query_dim", "query_scale", "range", "ref_anchor_head", "ref_point_head", "self", "super"], "dab_detr/modeling_dab_detr.py:DabDetrDecoder.forward": ["ModelDecoder", "ModelDecoderOutput", "ModelPreTrainedModel", "None", "_prepare_4d_attention_mask", "all_cross_attentions", "all_hidden_states", "all_self_attns", "and", "append", "attentions", "bbox_embed", "bool", "class", "config", "cross_attentions", "decoder_layer", "def", "detach", "dtype", "else", "encoder_attention_mask", "encoder_hidden_states", "enumerate", "for", "forward", "gen_sine_position_embeddings", "hidden_size", "hidden_states", "if", "in", "input_shape", "inputs_embeds", "intermediate", "intermediate_hidden_states", "inverse_sigmoid", "is", "kwargs", "last_hidden_state", "layer_id", "layer_outputs", "layernorm", "layers", "memory_key_padding_mask", "new_reference_points", "not", "num_layers", "obj_center", "object_queries", "output_attentions", "output_hidden_states", "output_intermediate_hidden_states", "output_reference_points", "pos_transformation", "query_dim", "query_pos", "query_position_embeddings", "query_scale", "query_sine_embed", "r", "ref_anchor_head", "ref_point_head", "ref_points", "reference_anchor_size", "reference_points", "return", "return_dict", "self", "sigmoid", "size", "stack", "tgt_len", "torch", "tuple", "unsqueeze", "use_return_dict", "v"], "dab_detr/modeling_dab_detr.py:DabDetrModel.__init__": ["Conv2d", "Embedding", "False", "ModelConfig", "ModelConvEncoder", "ModelConvModel", "ModelDecoder", "ModelEncoder", "ModelModel", "ModelPreTrainedModel", "ModelSinePositionEmbedding", "__init__", "aux_loss", "auxiliary_loss", "backbone", "be", "but", "class", "config", "data", "decoder", "def", "encoder", "f", "hidden_size", "if", "input_projection", "int", "intermediate_channel_sizes", "inverse_sigmoid", "isinstance", "kernel_size", "logger", "nn", "not", "num_patterns", "num_queries", "object_queries", "patterns", "post_init", "query_dim", "query_refpoint_embeddings", "random_refpoints_xy", "requires_grad", "self", "should", "super", "type", "uniform_", "warning", "weight"], "dab_detr/modeling_dab_detr.py:DabDetrModel.freeze_backbone": ["False", "ModelModel", "ModelPreTrainedModel", "backbone", "class", "conv_encoder", "def", "for", "freeze_backbone", "in", "model", "name", "named_parameters", "param", "requires_grad_", "self"], "dab_detr/modeling_dab_detr.py:DabDetrModel.unfreeze_backbone": ["ModelModel", "ModelPreTrainedModel", "True", "backbone", "class", "conv_encoder", "def", "for", "in", "model", "name", "named_parameters", "param", "requires_grad_", "self", "unfreeze_backbone"], "dab_detr/modeling_dab_detr.py:DabDetrModel.forward": ["Backbone", "BaseModelOutput", "FloatTensor", "LongTensor", "ModelModel", "ModelModelOutput", "ModelPreTrainedModel", "None", "ValueError", "_", "and", "attention_mask", "attentions", "backbone", "batch_size", "bool", "class", "config", "cross_attentions", "decoder", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_inputs_embeds", "decoder_outputs", "def", "device", "does", "downsampled", "elif", "else", "encoder", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "feature_map", "features", "flatten", "flattened_features", "flattened_mask", "forward", "height", "hidden_size", "hidden_states", "if", "input_projection", "inputs_embeds", "intermediate_hidden_states", "is", "isinstance", "kwargs", "last_hidden_state", "len", "mask", "memory_key_padding_mask", "not", "num_patterns", "num_queries", "object_queries", "object_queries_list", "ones", "output", "output_attentions", "output_hidden_states", "patterns", "permute", "pixel", "pixel_mask", "pixel_values", "projected_feature_map", "queries", "query_position_embeddings", "query_refpoint_embeddings", "r", "raise", "reference_points", "reference_position_embeddings", "repeat", "return", "return_dict", "self", "shape", "torch", "tuple", "unsqueeze", "use_return_dict", "weight", "width", "zeros"], "dab_detr/modeling_dab_detr.py:DabDetrMHAttentionMap.__init__": ["Dropout", "Linear", "ModelMHAttentionMap", "Module", "None", "True", "__init__", "bias", "class", "def", "dropout", "float", "hidden_dim", "k_linear", "nn", "normalize_fact", "num_heads", "q_linear", "query_dim", "self", "std", "super"], "dab_detr/modeling_dab_detr.py:DabDetrMHAttentionMap.forward": ["ModelMHAttentionMap", "Module", "None", "Tensor", "bias", "bnchw", "bqnc", "bqnhw", "class", "conv2d", "def", "dim", "dropout", "dtype", "einsum", "finfo", "flatten", "forward", "functional", "hidden_dim", "if", "is", "k", "k_linear", "keys_per_head", "mask", "masked_fill", "min", "nn", "normalize_fact", "not", "num_heads", "q", "q_linear", "queries_per_head", "return", "self", "shape", "size", "softmax", "torch", "unsqueeze", "view", "weight", "weights"], "dab_detr/modeling_dab_detr.py:DabDetrForObjectDetection.__init__": ["Linear", "ModelConfig", "ModelForObjectDetection", "ModelMLP", "ModelModel", "ModelPreTrainedModel", "__init__", "auxiliary_loss", "bbox_embed", "bbox_predictor", "class", "class_embed", "config", "decoder", "def", "hidden_size", "model", "nn", "num_labels", "post_init", "query_dim", "self", "super"], "dab_detr/modeling_dab_detr.py:DabDetrForObjectDetection._set_aux_loss": ["ModelForObjectDetection", "ModelPreTrainedModel", "_set_aux_loss", "a", "b", "class", "def", "for", "in", "logits", "outputs_class", "outputs_coord", "pred_boxes", "return", "self", "zip"], "dab_detr/modeling_dab_detr.py:DabDetrForObjectDetection.forward": ["FloatTensor", "LongTensor", "ModelForObjectDetection", "ModelObjectDetectionOutput", "ModelPreTrainedModel", "None", "auxiliary_loss", "auxiliary_outputs", "bbox_predictor", "bbox_with_refinement", "bool", "class", "class_embed", "config", "cross_attentions", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_inputs_embeds", "def", "device", "dict", "else", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "forward", "if", "inputs_embeds", "intermediate_hidden_states", "inverse_sigmoid", "is", "kwargs", "labels", "last_hidden_state", "list", "logits", "loss", "loss_dict", "loss_function", "model", "model_outputs", "not", "output", "output_attentions", "output_hidden_states", "outputs_class", "outputs_coord", "pixel_mask", "pixel_values", "pred_boxes", "query_dim", "r", "reference_before_sigmoid", "reference_points", "return", "return_dict", "self", "sigmoid", "torch", "tuple", "use_return_dict"], "dac/modeling_dac.py:Snake1d.__init__": ["Model", "Module", "Parameter", "__init__", "alpha", "class", "def", "hidden_dim", "nn", "ones", "self", "super", "torch"], "dac/modeling_dac.py:Snake1d.forward": ["Model", "Module", "alpha", "class", "def", "forward", "hidden_states", "nn", "pow", "reciprocal", "reshape", "return", "self", "shape", "sin", "torch"], "dac/modeling_dac.py:DacVectorQuantize.__init__": ["Conv1d", "Embedding", "ModelConfig", "ModelVectorQuantize", "Module", "__init__", "class", "codebook", "codebook_dim", "codebook_size", "config", "def", "hidden_size", "in_proj", "kernel_size", "nn", "out_proj", "self", "super"], "dac/modeling_dac.py:DacVectorQuantize.forward": ["F", "ModelVectorQuantize", "Module", "audio_codes", "class", "codebook_loss", "commitment_loss", "decode_latents", "def", "detach", "forward", "hidden_state", "in_proj", "mean", "mse_loss", "nn", "out_proj", "projected_latents", "quantized_representation", "reduction", "return", "self"], "dac/modeling_dac.py:DacVectorQuantize.decode_latents": ["F", "ModelVectorQuantize", "Module", "True", "batch_size", "class", "codebook", "decode_latents", "def", "dist", "encodings", "hidden_dim", "hidden_states", "indices", "keepdim", "l2_norm", "max", "nn", "normalize", "permute", "pow", "quantized_representation", "reshape", "return", "self", "sequence_length", "shape", "size", "sum", "t", "transpose", "weight"], "dac/modeling_dac.py:DacResidualUnit.__init__": ["Conv1d", "ModelResidualUnit", "Module", "Snake1d", "__init__", "class", "conv1", "conv2", "def", "dilation", "dimension", "int", "kernel_size", "nn", "pad", "padding", "self", "snake1", "snake2", "super"], "dac/modeling_dac.py:DacResidualUnit.forward": ["ModelResidualUnit", "Module", "class", "conv1", "conv2", "def", "forward", "hidden_state", "if", "nn", "output_tensor", "padding", "return", "self", "shape", "snake1", "snake2"], "dac/modeling_dac.py:DacEncoderBlock.__init__": ["Conv1d", "ModelConfig", "ModelEncoderBlock", "ModelResidualUnit", "Module", "Snake1d", "__init__", "ceil", "class", "config", "conv1", "def", "dilation", "dimension", "encoder_hidden_size", "int", "kernel_size", "math", "nn", "padding", "res_unit1", "res_unit2", "res_unit3", "self", "snake1", "stride", "stride_index", "super"], "dac/modeling_dac.py:DacEncoderBlock.forward": ["ModelEncoderBlock", "Module", "class", "conv1", "def", "forward", "hidden_state", "nn", "res_unit1", "res_unit2", "res_unit3", "return", "self", "snake1"], "dac/modeling_dac.py:DacDecoderBlock.__init__": ["ConvTranspose1d", "ModelConfig", "ModelDecoderBlock", "ModelResidualUnit", "Module", "Snake1d", "__init__", "ceil", "class", "config", "conv_t1", "decoder_hidden_size", "def", "dilation", "input_dim", "int", "kernel_size", "math", "nn", "output_dim", "padding", "res_unit1", "res_unit2", "res_unit3", "self", "snake1", "stride", "stride_index", "super"], "dac/modeling_dac.py:DacDecoderBlock.forward": ["ModelDecoderBlock", "Module", "class", "conv_t1", "def", "forward", "hidden_state", "nn", "res_unit1", "res_unit2", "res_unit3", "return", "self", "snake1"], "dac/modeling_dac.py:DacResidualVectorQuantizer.__init__": ["ModelConfig", "ModelResidualVectorQuantizer", "ModelVectorQuantize", "Module", "ModuleList", "__init__", "class", "config", "def", "for", "i", "in", "n_codebooks", "nn", "quantizer_dropout", "quantizers", "range", "self", "super"], "dac/modeling_dac.py:DacResidualVectorQuantizer.forward": ["False", "ModelResidualVectorQuantizer", "Module", "None", "and", "append", "audio_codes", "break", "cat", "class", "codebook_loss", "codebook_loss_i", "commitment_loss", "commitment_loss_i", "def", "device", "dim", "dropout", "else", "enumerate", "fill_value", "for", "forward", "full", "hidden_state", "i", "if", "in", "indices_i", "int", "is", "mask", "n_codebooks", "n_dropout", "n_quantizers", "nn", "not", "ones", "projected_latents", "projected_latents_i", "quantized_representation", "quantized_representation_i", "quantizer", "quantizer_dropout", "quantizers", "randint", "residual", "return", "self", "shape", "stack", "to", "torch", "training"], "dac/modeling_dac.py:DacResidualVectorQuantizer.from_codes": ["ModelResidualVectorQuantizer", "Module", "Tensor", "append", "audio_codes", "cat", "class", "codebook", "def", "dim", "for", "from_codes", "i", "in", "n_codebooks", "nn", "out_proj", "projected_latents", "projected_latents_i", "quantized_representation", "quantizers", "range", "return", "self", "shape", "torch", "transpose"], "dac/modeling_dac.py:DacResidualVectorQuantizer.from_latents": ["ModelResidualVectorQuantizer", "Module", "Tensor", "True", "append", "axis", "cat", "class", "codebook_dim", "codebook_dims_tensor", "codes", "codes_i", "cumsum", "decode_latents", "def", "dim", "dims", "for", "from_latents", "hidden_dim_j", "hidden_dim_k", "i", "in", "keepdims", "latents", "max", "n_codebooks", "nn", "np", "out_proj", "q", "quantized_latents", "quantized_latents_i", "quantized_representation", "quantized_representation_i", "quantizers", "range", "return", "self", "shape", "tensor", "torch", "where"], "dac/modeling_dac.py:DacDecoder.__init__": ["Conv1d", "ModelConfig", "ModelDecoder", "ModelDecoderBlock", "Module", "ModuleList", "Snake1d", "Tanh", "__init__", "block", "channels", "class", "config", "conv1", "conv2", "decoder_hidden_size", "def", "enumerate", "for", "hidden_size", "in", "input_channel", "kernel_size", "nn", "output_dim", "padding", "self", "snake1", "stride", "stride_index", "strides", "super", "tanh", "upsampling_ratios"], "dac/modeling_dac.py:DacDecoder.forward": ["ModelDecoder", "Module", "block", "class", "conv1", "conv2", "def", "for", "forward", "hidden_state", "in", "layer", "nn", "return", "self", "snake1", "tanh"], "dac/modeling_dac.py:DacEncoder.__init__": ["Conv1d", "ModelConfig", "ModelEncoder", "ModelEncoderBlock", "Module", "ModuleList", "Snake1d", "__init__", "block", "class", "config", "conv1", "conv2", "d_model", "def", "downsampling_ratios", "encoder_hidden_size", "enumerate", "for", "hidden_size", "in", "kernel_size", "nn", "padding", "self", "snake1", "stride", "stride_index", "strides", "super"], "dac/modeling_dac.py:DacEncoder.forward": ["ModelEncoder", "Module", "block", "class", "conv1", "conv2", "def", "for", "forward", "hidden_state", "in", "module", "nn", "return", "self", "snake1"], "dac/modeling_dac.py:DacPreTrainedModel._init_weights": ["Conv1d", "ConvTranspose1d", "Embedding", "ModelPreTrainedModel", "PreTrainedAudioTokenizerBase", "Snake1d", "_init_weights", "alpha", "bias", "class", "constant_", "def", "elif", "if", "init", "isinstance", "mean", "module", "nn", "normal_", "ones_", "reset_parameters", "self", "std", "trunc_normal_", "weight"], "dac/modeling_dac.py:DacPreTrainedModel.apply_weight_norm": ["ModelPreTrainedModel", "PreTrainedAudioTokenizerBase", "apply_weight_norm", "block", "class", "conv1", "conv2", "conv_t1", "decoder", "def", "encoder", "for", "hasattr", "if", "in", "in_proj", "layer", "nn", "out_proj", "parametrizations", "quantizer", "quantizers", "res_unit1", "res_unit2", "res_unit3", "self", "utils", "weight_norm"], "dac/modeling_dac.py:DacPreTrainedModel.remove_weight_norm": ["ModelPreTrainedModel", "PreTrainedAudioTokenizerBase", "block", "class", "conv1", "conv2", "conv_t1", "decoder", "def", "encoder", "for", "in", "in_proj", "layer", "nn", "out_proj", "quantizer", "quantizers", "remove_weight_norm", "res_unit1", "res_unit2", "res_unit3", "self", "utils"], "dac/modeling_dac.py:DacModel.__init__": ["ModelConfig", "ModelDecoder", "ModelEncoder", "ModelModel", "ModelPreTrainedModel", "ModelResidualVectorQuantizer", "The", "ValueError", "__init__", "a", "be", "bits_per_codebook", "class", "codebook_size", "config", "decoder", "def", "encoder", "if", "int", "log2", "math", "must", "of", "post_init", "power", "quantizer", "raise", "self", "super"], "dac/modeling_dac.py:DacModel.encode": ["ModelEncoderOutput", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "audio_codes", "bool", "class", "codebook_loss", "codebook_loss_weight", "commitment_loss", "commitment_loss_weight", "config", "def", "else", "encode", "encoder", "if", "input_values", "int", "is", "loss", "n_quantizers", "not", "projected_latents", "quantized_representation", "quantizer", "r", "return", "return_dict", "self", "torch", "tuple"], "dac/modeling_dac.py:DacModel.decode": ["Either", "ModelDecoderOutput", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "ValueError", "and", "audio_codes", "audio_values", "be", "bool", "class", "config", "decode", "decoder", "def", "else", "from_codes", "if", "is", "must", "not", "or", "provided", "quantized_representation", "quantizer", "r", "raise", "return", "return_dict", "self", "squeeze", "torch", "tuple"], "dac/modeling_dac.py:DacModel.forward": ["False", "ModelModel", "ModelOutput", "ModelPreTrainedModel", "None", "Tensor", "audio_codes", "audio_values", "bool", "class", "config", "decode", "def", "else", "encode", "forward", "if", "input_values", "int", "is", "length", "loss", "n_quantizers", "not", "projected_latents", "quantized_representation", "r", "return", "return_dict", "self", "shape", "torch", "tuple"], "data2vec/modeling_data2vec_audio.py:Data2VecAudioConvLayer.__init__": ["ACT2FN", "Conv1d", "GradientCheckpointingLayer", "LayerNorm", "ModelAudioConvLayer", "True", "__init__", "activation", "bias", "class", "config", "conv", "conv_bias", "conv_dim", "conv_kernel", "conv_stride", "def", "elementwise_affine", "else", "feat_extract_activation", "if", "in_conv_dim", "kernel_size", "layer_id", "layer_norm", "nn", "out_conv_dim", "self", "stride", "super"], "data2vec/modeling_data2vec_audio.py:Data2VecAudioConvLayer.forward": ["GradientCheckpointingLayer", "ModelAudioConvLayer", "activation", "class", "conv", "def", "forward", "hidden_states", "layer_norm", "return", "self", "transpose"], "data2vec/modeling_data2vec_audio.py:Data2VecAudioPadLayer.__init__": ["ModelAudioPadLayer", "Module", "__init__", "class", "def", "else", "if", "nn", "num_conv_pos_embeddings", "num_pad_remove", "self", "super"], "data2vec/modeling_data2vec_audio.py:Data2VecAudioPadLayer.forward": ["ModelAudioPadLayer", "Module", "class", "def", "forward", "hidden_states", "if", "nn", "num_pad_remove", "return", "self"], "data2vec/modeling_data2vec_audio.py:Data2VecAudioPositionalConvLayer.__init__": ["ACT2FN", "Conv1d", "False", "LayerNorm", "ModelAudioPadLayer", "ModelAudioPositionalConvLayer", "Module", "__init__", "activation", "class", "config", "conv", "conv_pos_kernel_size", "def", "elementwise_affine", "feat_extract_activation", "groups", "hidden_size", "kernel_size", "layer_norm", "nn", "num_conv_pos_embedding_groups", "padding", "self", "super"], "data2vec/modeling_data2vec_audio.py:Data2VecAudioPositionalConvLayer.forward": ["ModelAudioPositionalConvLayer", "Module", "activation", "class", "conv", "def", "forward", "hidden_states", "layer_norm", "nn", "padding", "return", "self", "transpose"], "data2vec/modeling_data2vec_audio.py:Data2VecAudioPositionalConvEmbedding.__init__": ["ModelAudioPositionalConvEmbedding", "ModelAudioPositionalConvLayer", "Module", "ModuleList", "_", "__init__", "class", "config", "def", "for", "in", "layers", "nn", "num_conv_pos_embeddings", "range", "self", "super"], "data2vec/modeling_data2vec_audio.py:Data2VecAudioPositionalConvEmbedding.forward": ["ModelAudioPositionalConvEmbedding", "Module", "class", "def", "for", "forward", "hidden_states", "in", "layer", "layers", "nn", "return", "self", "transpose"], "data2vec/modeling_data2vec_audio.py:Data2VecAudioFeatureEncoder.__init__": ["False", "ModelAudioConvLayer", "ModelAudioFeatureEncoder", "Module", "ModuleList", "True", "__init__", "_requires_grad", "class", "config", "conv_layers", "def", "for", "gradient_checkpointing", "i", "in", "layer_id", "nn", "num_feat_extract_layers", "range", "self", "super"], "data2vec/modeling_data2vec_audio.py:Data2VecAudioFeatureEncoder._freeze_parameters": ["False", "ModelAudioFeatureEncoder", "Module", "_freeze_parameters", "_requires_grad", "class", "def", "for", "in", "nn", "param", "parameters", "requires_grad", "self"], "data2vec/modeling_data2vec_audio.py:Data2VecAudioFeatureEncoder.forward": ["ModelAudioFeatureEncoder", "Module", "None", "True", "_requires_grad", "and", "class", "conv_layer", "conv_layers", "def", "for", "forward", "hidden_states", "if", "in", "input_values", "nn", "requires_grad", "return", "self", "training"], "data2vec/modeling_data2vec_audio.py:Data2VecAudioFeatureProjection.__init__": ["Dropout", "LayerNorm", "Linear", "ModelAudioFeatureProjection", "Module", "__init__", "class", "config", "conv_dim", "def", "dropout", "eps", "feat_proj_dropout", "hidden_size", "layer_norm", "layer_norm_eps", "nn", "projection", "self", "super"], "data2vec/modeling_data2vec_audio.py:Data2VecAudioFeatureProjection.forward": ["ModelAudioFeatureProjection", "Module", "class", "def", "dropout", "forward", "hidden_states", "layer_norm", "nn", "norm_hidden_states", "projection", "return", "self"], "data2vec/modeling_data2vec_audio.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "contiguous", "def", "dim", "dropout", "float", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "shape", "size", "softmax", "torch", "training", "transpose", "value"], "data2vec/modeling_data2vec_audio.py:Data2VecAudioAttention.__init__": ["False", "Linear", "ModelAudioAttention", "ModelAudioConfig", "Module", "None", "True", "ValueError", "__init__", "and", "be", "bias", "bool", "by", "class", "config", "def", "divisible", "dropout", "embed_dim", "f", "float", "got", "head_dim", "if", "int", "is_causal", "is_decoder", "k_proj", "must", "nn", "num_heads", "out_proj", "q_proj", "raise", "scaling", "self", "super", "v_proj"], "data2vec/modeling_data2vec_audio.py:Data2VecAudioAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "False", "FlashAttentionKwargs", "ModelAudioAttention", "Module", "None", "Tensor", "Unpack", "_attn_implementation", "attention_interface", "attention_mask", "attn_output", "attn_weights", "bool", "bsz", "class", "config", "contiguous", "current_states", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "head_dim", "hidden_states", "if", "is", "is_cross_attention", "k_proj", "key_states", "key_value_states", "kv_input_shape", "kwargs", "nn", "not", "out_proj", "output_attentions", "q_input_shape", "q_proj", "query_states", "reshape", "return", "scaling", "self", "shape", "src_len", "tgt_len", "torch", "training", "transpose", "tuple", "v_proj", "value_states", "view"], "data2vec/modeling_data2vec_audio.py:Data2VecAudioFeedForward.__init__": ["ACT2FN", "Dropout", "Linear", "ModelAudioFeedForward", "Module", "__init__", "activation_dropout", "class", "config", "def", "else", "hidden_act", "hidden_dropout", "hidden_size", "if", "intermediate_act_fn", "intermediate_dense", "intermediate_dropout", "intermediate_size", "isinstance", "nn", "output_dense", "output_dropout", "self", "str", "super"], "data2vec/modeling_data2vec_audio.py:Data2VecAudioFeedForward.forward": ["ModelAudioFeedForward", "Module", "class", "def", "forward", "hidden_states", "intermediate_act_fn", "intermediate_dense", "intermediate_dropout", "nn", "output_dense", "output_dropout", "return", "self"], "data2vec/modeling_data2vec_audio.py:Data2VecAudioEncoderLayer.__init__": ["Dropout", "False", "GradientCheckpointingLayer", "LayerNorm", "ModelAudioAttention", "ModelAudioEncoderLayer", "ModelAudioFeedForward", "__init__", "attention", "attention_dropout", "class", "config", "def", "dropout", "embed_dim", "eps", "feed_forward", "final_layer_norm", "hidden_dropout", "hidden_size", "is_decoder", "layer_norm", "layer_norm_eps", "nn", "num_attention_heads", "num_heads", "self", "super"], "data2vec/modeling_data2vec_audio.py:Data2VecAudioEncoderLayer.forward": ["False", "GradientCheckpointingLayer", "ModelAudioEncoderLayer", "None", "_", "attention", "attention_mask", "attn_residual", "attn_weights", "class", "def", "dropout", "feed_forward", "final_layer_norm", "forward", "hidden_states", "if", "layer_norm", "output_attentions", "outputs", "return", "self"], "data2vec/modeling_data2vec_audio.py:Data2VecAudioEncoder.__init__": ["Dropout", "False", "LayerNorm", "ModelAudioEncoder", "ModelAudioEncoderLayer", "ModelAudioPositionalConvEmbedding", "Module", "ModuleList", "_", "__init__", "class", "config", "def", "dropout", "eps", "for", "gradient_checkpointing", "hidden_dropout", "hidden_size", "in", "layer_norm", "layer_norm_eps", "layers", "nn", "num_hidden_layers", "pos_conv_embed", "range", "self", "super"], "data2vec/modeling_data2vec_audio.py:Data2VecAudioEncoder.forward": ["BaseModelOutput", "False", "ModelAudioEncoder", "Module", "None", "Tensor", "True", "all_hidden_states", "all_self_attentions", "and", "attention_mask", "attentions", "bool", "class", "config", "create_bidirectional_mask", "def", "dropout", "dropout_probability", "else", "expand_attention_mask", "for", "forward", "hidden_states", "if", "in", "input_embeds", "is", "is_deepspeed_zero3_enabled", "is_fsdp_managed_module", "last_hidden_state", "layer", "layer_norm", "layer_outputs", "layerdrop", "layers", "nn", "not", "or", "output_attentions", "output_hidden_states", "pos_conv_embed", "position_embeddings", "rand", "repeat", "return", "return_dict", "self", "shape", "skip_the_layer", "synced_gpus", "tensor", "torch", "training", "tuple", "unsqueeze", "v"], "data2vec/modeling_data2vec_audio.py:Data2VecAudioAdapterLayer.__init__": ["Conv1d", "ModelAudioAdapterLayer", "Module", "__init__", "adapter_kernel_size", "adapter_stride", "class", "config", "conv", "def", "nn", "output_hidden_size", "padding", "self", "stride", "super"], "data2vec/modeling_data2vec_audio.py:Data2VecAudioAdapterLayer.forward": ["ModelAudioAdapterLayer", "Module", "class", "conv", "def", "dim", "forward", "functional", "glu", "hidden_states", "nn", "return", "self"], "data2vec/modeling_data2vec_audio.py:Data2VecAudioAdapter.__init__": ["LayerNorm", "Linear", "ModelAudioAdapter", "ModelAudioAdapterLayer", "Module", "ModuleList", "None", "_", "__init__", "class", "config", "def", "else", "for", "hidden_size", "if", "in", "layerdrop", "layers", "nn", "num_adapter_layers", "output_hidden_size", "proj", "proj_layer_norm", "range", "self", "super"], "data2vec/modeling_data2vec_audio.py:Data2VecAudioAdapter.forward": ["ModelAudioAdapter", "Module", "None", "and", "class", "def", "for", "forward", "hidden_states", "if", "in", "is", "layer", "layerdrop", "layerdrop_prob", "layers", "nn", "not", "np", "or", "proj", "proj_layer_norm", "random", "return", "self", "training", "transpose"], "data2vec/modeling_data2vec_audio.py:Data2VecAudioPreTrainedModel._init_weights": ["Conv1d", "GroupNorm", "LayerNorm", "Linear", "ModelAudioFeatureProjection", "ModelAudioPositionalConvLayer", "ModelAudioPreTrainedModel", "None", "PreTrainedModel", "_init_weights", "a", "b", "bias", "class", "config", "constant_", "conv", "def", "elif", "groups", "if", "in_channels", "in_features", "init", "initializer_range", "is", "isinstance", "k", "kaiming_normal_", "kernel_size", "math", "mean", "module", "nn", "normal_", "not", "ones_", "projection", "self", "sqrt", "std", "uniform_", "weight", "zeros_"], "data2vec/modeling_data2vec_audio.py:Data2VecAudioPreTrainedModel._get_feat_extract_output_lengths": ["LongTensor", "ModelAudioPreTrainedModel", "None", "PreTrainedModel", "_", "_conv_out_length", "_get_feat_extract_output_lengths", "adapter_stride", "add_adapter", "bool", "class", "config", "conv_kernel", "conv_stride", "def", "div", "else", "floor", "for", "if", "in", "input_length", "input_lengths", "int", "is", "kernel_size", "num_adapter_layers", "range", "return", "rounding_mode", "self", "stride", "torch", "zip"], "data2vec/modeling_data2vec_audio.py:Data2VecAudioPreTrainedModel._get_feature_vector_attention_mask": ["LongTensor", "ModelAudioPreTrainedModel", "None", "PreTrainedModel", "_get_feat_extract_output_lengths", "_get_feature_vector_attention_mask", "add_adapter", "arange", "attention_mask", "batch_size", "bool", "class", "cumsum", "def", "device", "dim", "dtype", "feature_vector_length", "flip", "int", "long", "non_padded_lengths", "output_lengths", "return", "self", "shape", "to", "torch", "zeros"], "data2vec/modeling_data2vec_audio.py:_compute_mask_indices": ["False", "LongTensor", "None", "ValueError", "_", "_compute_mask_indices", "and", "append", "arange", "array", "attention_mask", "batch_size", "be", "bigger", "bool", "broadcast_to", "but", "choice", "compute_num_masked_span", "concatenate", "def", "detach", "dtype", "dummy_mask_idx", "else", "epsilon", "f", "float", "for", "got", "has", "if", "in", "input_length", "input_lengths", "int", "int32", "is", "item", "len", "mask_length", "mask_prob", "max", "max_num_masked_span", "min_masks", "ndarray", "not", "np", "num_masked_span", "offsets", "ones", "put_along_axis", "raise", "rand", "random", "range", "replace", "reshape", "return", "sequence_length", "shape", "smaller", "spec_aug_mask", "spec_aug_mask_idx", "spec_aug_mask_idxs", "sum", "than", "to", "tolist", "torch", "tuple", "zeros"], "data2vec/modeling_data2vec_audio.py:Data2VecAudioModel.__init__": ["ModelAudioAdapter", "ModelAudioConfig", "ModelAudioEncoder", "ModelAudioFeatureEncoder", "ModelAudioFeatureProjection", "ModelAudioModel", "ModelAudioPreTrainedModel", "None", "Parameter", "Tensor", "__init__", "adapter", "add_adapter", "class", "config", "def", "else", "encoder", "feature_extractor", "feature_projection", "hidden_size", "if", "mask_feature_prob", "mask_time_prob", "masked_spec_embed", "nn", "or", "post_init", "self", "super", "torch", "uniform_"], "data2vec/modeling_data2vec_audio.py:Data2VecAudioModel.freeze_feature_encoder": ["ModelAudioModel", "ModelAudioPreTrainedModel", "_freeze_parameters", "class", "def", "feature_extractor", "freeze_feature_encoder", "self"], "data2vec/modeling_data2vec_audio.py:Data2VecAudioModel._mask_hidden_states": ["FloatTensor", "LongTensor", "ModelAudioModel", "ModelAudioPreTrainedModel", "None", "True", "_compute_mask_indices", "_mask_hidden_states", "and", "apply_spec_augment", "attention_mask", "batch_size", "bool", "class", "config", "def", "device", "dtype", "elif", "expand", "getattr", "hidden_size", "hidden_states", "if", "is", "mask_feature_indices", "mask_feature_length", "mask_feature_min_masks", "mask_feature_prob", "mask_length", "mask_prob", "mask_time_indices", "mask_time_length", "mask_time_min_masks", "mask_time_prob", "masked_spec_embed", "min_masks", "not", "return", "self", "sequence_length", "size", "tensor", "to", "torch", "training"], "data2vec/modeling_data2vec_audio.py:Data2VecAudioModel.forward": ["False", "FloatTensor", "ModelAudioBaseModelOutput", "ModelAudioModel", "ModelAudioPreTrainedModel", "None", "Tensor", "_get_feature_vector_attention_mask", "_mask_hidden_states", "adapter", "add_adapter", "attention_mask", "attentions", "bool", "class", "config", "def", "else", "encoder", "encoder_outputs", "extract_features", "feature_extractor", "feature_projection", "forward", "hidden_states", "if", "input_values", "is", "kwargs", "last_hidden_state", "mask_time_indices", "not", "output_attentions", "output_hidden_states", "r", "return", "return_dict", "self", "shape", "torch", "transpose", "tuple", "use_return_dict"], "data2vec/modeling_data2vec_audio.py:Data2VecAudioForCTC.__init__": ["Dropout", "Linear", "ModelAudioForCTC", "ModelAudioModel", "ModelAudioPreTrainedModel", "Model_audio", "None", "Please", "ValueError", "You", "__class__", "__init__", "a", "add_adapter", "and", "are", "as", "class", "config", "configuration", "def", "define", "does", "dropout", "else", "f", "final_dropout", "follows", "from_pretrained", "hasattr", "head", "hidden_size", "if", "instantiate", "is", "language", "lm_head", "model", "nn", "not", "of", "or", "output_hidden_size", "post_init", "r", "raise", "s", "self", "size", "super", "that", "the", "to", "trying", "vocab_size", "vocabulary", "with", "your"], "data2vec/modeling_data2vec_audio.py:Data2VecAudioForCTC.freeze_feature_encoder": ["ModelAudioForCTC", "ModelAudioPreTrainedModel", "Model_audio", "_freeze_parameters", "class", "def", "feature_extractor", "freeze_feature_encoder", "self"], "data2vec/modeling_data2vec_audio.py:Data2VecAudioForCTC.forward": ["CausalLMOutput", "False", "Label", "ModelAudioForCTC", "ModelAudioPreTrainedModel", "Model_audio", "None", "Tensor", "ValueError", "_HIDDEN_STATES_START_POSITION", "_get_feat_extract_output_lengths", "and", "attention_mask", "attentions", "backends", "be", "blank", "bool", "class", "config", "ctc_loss", "ctc_loss_reduction", "ctc_zero_infinity", "cudnn", "def", "dim", "dropout", "dtype", "else", "enabled", "f", "flags", "flattened_targets", "float32", "forward", "functional", "hidden_states", "if", "input_lengths", "input_values", "is", "kwargs", "labels", "labels_mask", "lm_head", "log_probs", "log_softmax", "logits", "long", "loss", "masked_select", "max", "must", "nn", "not", "ones_like", "output", "output_attentions", "output_hidden_states", "outputs", "pad_token_id", "r", "raise", "reduction", "return", "return_dict", "self", "sum", "target_lengths", "to", "torch", "transpose", "tuple", "use_return_dict", "values", "vocab_size", "with", "zero_infinity"], "data2vec/modeling_data2vec_audio.py:Data2VecAudioForSequenceClassification.__init__": ["Linear", "ModelAudio", "ModelAudioForSequenceClassification", "ModelAudioModel", "ModelAudioPreTrainedModel", "Model_audio", "Parameter", "Sequence", "True", "ValueError", "__init__", "adapters", "add_adapter", "and", "class", "classification", "classifier", "classifier_proj_size", "config", "def", "does", "hasattr", "hidden_size", "if", "layer_weights", "nn", "not", "num_hidden_layers", "num_labels", "num_layers", "of", "ones", "post_init", "projector", "raise", "self", "super", "support", "the", "torch", "use", "use_weighted_layer_sum"], "data2vec/modeling_data2vec_audio.py:Data2VecAudioForSequenceClassification.freeze_feature_encoder": ["ModelAudioForSequenceClassification", "ModelAudioPreTrainedModel", "Model_audio", "_freeze_parameters", "class", "def", "feature_extractor", "freeze_feature_encoder", "self"], "data2vec/modeling_data2vec_audio.py:Data2VecAudioForSequenceClassification.freeze_base_model": ["False", "ModelAudioForSequenceClassification", "ModelAudioPreTrainedModel", "Model_audio", "class", "def", "for", "freeze_base_model", "in", "param", "parameters", "requires_grad", "self"], "data2vec/modeling_data2vec_audio.py:Data2VecAudioForSequenceClassification.forward": ["CrossEntropyLoss", "ModelAudioForSequenceClassification", "ModelAudioPreTrainedModel", "Model_audio", "None", "SequenceClassifierOutput", "Tensor", "True", "_HIDDEN_STATES_START_POSITION", "_get_feature_vector_attention_mask", "attention_mask", "attentions", "bool", "class", "classifier", "config", "def", "dim", "else", "expand_padding_mask", "forward", "functional", "hidden_states", "if", "input_values", "is", "kwargs", "labels", "layer_weights", "logits", "loss", "loss_fct", "mean", "nn", "norm_weights", "not", "num_labels", "output", "output_attentions", "output_hidden_states", "outputs", "padding_mask", "pooled_output", "projector", "r", "repeat", "return", "return_dict", "self", "shape", "softmax", "stack", "sum", "torch", "tuple", "unsqueeze", "use_return_dict", "use_weighted_layer_sum", "view"], "data2vec/modeling_data2vec_audio.py:Data2VecAudioForAudioFrameClassification.__init__": ["Audio", "Linear", "ModelAudio", "ModelAudioForAudioFrameClassification", "ModelAudioModel", "ModelAudioPreTrainedModel", "Model_audio", "Parameter", "True", "ValueError", "__init__", "adapters", "add_adapter", "and", "class", "classification", "classifier", "config", "def", "does", "frame", "hasattr", "hidden_size", "if", "layer_weights", "nn", "not", "num_hidden_layers", "num_labels", "num_layers", "of", "ones", "post_init", "raise", "self", "super", "support", "the", "torch", "use", "use_weighted_layer_sum"], "data2vec/modeling_data2vec_audio.py:Data2VecAudioForAudioFrameClassification.freeze_feature_encoder": ["ModelAudioForAudioFrameClassification", "ModelAudioPreTrainedModel", "Model_audio", "_freeze_parameters", "class", "def", "feature_extractor", "freeze_feature_encoder", "self"], "data2vec/modeling_data2vec_audio.py:Data2VecAudioForAudioFrameClassification.freeze_base_model": ["False", "ModelAudioForAudioFrameClassification", "ModelAudioPreTrainedModel", "Model_audio", "class", "def", "for", "freeze_base_model", "in", "param", "parameters", "requires_grad", "self"], "data2vec/modeling_data2vec_audio.py:Data2VecAudioForAudioFrameClassification.forward": ["CrossEntropyLoss", "ModelAudioForAudioFrameClassification", "ModelAudioPreTrainedModel", "Model_audio", "None", "Tensor", "TokenClassifierOutput", "True", "_HIDDEN_STATES_START_POSITION", "argmax", "attention_mask", "attentions", "axis", "bool", "class", "classifier", "config", "def", "dim", "else", "forward", "functional", "hidden_states", "if", "input_values", "is", "kwargs", "labels", "layer_weights", "logits", "loss", "loss_fct", "nn", "norm_weights", "not", "num_labels", "output", "output_attentions", "output_hidden_states", "outputs", "r", "return", "return_dict", "self", "softmax", "stack", "sum", "torch", "tuple", "use_return_dict", "use_weighted_layer_sum", "view"], "data2vec/modeling_data2vec_audio.py:AMSoftmaxLoss.__init__": ["CrossEntropyLoss", "Model", "Module", "Parameter", "True", "__init__", "class", "def", "input_dim", "loss", "margin", "nn", "num_labels", "randn", "requires_grad", "scale", "self", "super", "torch", "weight"], "data2vec/modeling_data2vec_audio.py:AMSoftmaxLoss.forward": ["Model", "Module", "bool", "class", "cos_theta", "def", "dim", "flatten", "forward", "functional", "hidden_states", "labels", "logits", "loss", "margin", "mm", "nn", "normalize", "num_labels", "one_hot", "onehot", "psi", "return", "scale", "self", "torch", "weight", "where"], "data2vec/modeling_data2vec_audio.py:TDNNLayer.__init__": ["Linear", "Model", "Module", "ReLU", "__init__", "activation", "class", "config", "def", "dilation", "else", "if", "in_conv_dim", "kernel", "kernel_size", "layer_id", "nn", "out_conv_dim", "self", "super", "tdnn_dilation", "tdnn_dim", "tdnn_kernel"], "data2vec/modeling_data2vec_audio.py:TDNNLayer.forward": ["Detected", "LoRA", "LoraLayer", "Model", "Module", "Tensor", "You", "activation", "applied", "be", "bias", "class", "conv1d", "def", "dilation", "due", "exclude", "forward", "from", "functional", "hidden_states", "if", "in_conv_dim", "is_peft_available", "isinstance", "kernel", "kernel_size", "modules", "nn", "on", "optimization", "out_conv_dim", "return", "s", "self", "should", "t", "target", "to", "torch", "transpose", "view", "warn", "warnings", "weight", "weights", "won"], "data2vec/modeling_data2vec_audio.py:Data2VecAudioForXVector.__init__": ["AMSoftmaxLoss", "Linear", "ModelAudioForXVector", "ModelAudioModel", "ModelAudioPreTrainedModel", "Model_audio", "ModuleList", "Parameter", "TDNNLayer", "__init__", "class", "classifier", "config", "def", "feature_extractor", "for", "hidden_size", "i", "if", "in", "layer_weights", "len", "nn", "num_hidden_layers", "num_labels", "num_layers", "objective", "ones", "post_init", "projector", "range", "self", "super", "tdnn", "tdnn_dim", "tdnn_layers", "torch", "use_weighted_layer_sum", "xvector_output_dim"], "data2vec/modeling_data2vec_audio.py:Data2VecAudioForXVector.freeze_feature_encoder": ["ModelAudioForXVector", "ModelAudioPreTrainedModel", "Model_audio", "_freeze_parameters", "class", "def", "feature_extractor", "freeze_feature_encoder", "self"], "data2vec/modeling_data2vec_audio.py:Data2VecAudioForXVector.freeze_base_model": ["False", "ModelAudioForXVector", "ModelAudioPreTrainedModel", "Model_audio", "class", "def", "for", "freeze_base_model", "in", "param", "parameters", "requires_grad", "self"], "data2vec/modeling_data2vec_audio.py:Data2VecAudioForXVector._get_tdnn_output_lengths": ["LongTensor", "ModelAudioForXVector", "ModelAudioPreTrainedModel", "_conv_out_length", "_get_tdnn_output_lengths", "class", "config", "def", "for", "in", "input_length", "input_lengths", "int", "kernel_size", "return", "self", "stride", "tdnn_kernel", "torch"], "data2vec/modeling_data2vec_audio.py:Data2VecAudioForXVector.forward": ["ModelAudioForXVector", "ModelAudioPreTrainedModel", "Model_audio", "None", "Tensor", "True", "XVectorOutput", "_HIDDEN_STATES_START_POSITION", "_get_feat_extract_output_lengths", "_get_tdnn_output_lengths", "append", "attention_mask", "attentions", "bool", "cat", "class", "classifier", "config", "def", "dim", "else", "embeddings", "enumerate", "feat_extract_output_lengths", "feature_extractor", "for", "forward", "functional", "hidden_states", "i", "if", "in", "input_values", "is", "kwargs", "labels", "layer_weights", "length", "logits", "loss", "mean", "mean_features", "nn", "norm_weights", "not", "objective", "output", "output_attentions", "output_embeddings", "output_hidden_states", "outputs", "projector", "r", "return", "return_dict", "self", "softmax", "stack", "statistic_pooling", "std", "std_features", "sum", "tdnn", "tdnn_layer", "tdnn_output_lengths", "torch", "tuple", "use_return_dict", "use_weighted_layer_sum", "view"], "data2vec/modeling_data2vec_text.py:Data2VecTextEmbeddings.__init__": ["Dropout", "Embedding", "False", "LayerNorm", "ModelTextEmbeddings", "Module", "__init__", "arange", "class", "config", "def", "dropout", "dtype", "eps", "expand", "hidden_dropout_prob", "hidden_size", "layer_norm_eps", "long", "max_position_embeddings", "nn", "pad_token_id", "padding_idx", "persistent", "position_embeddings", "position_ids", "register_buffer", "self", "size", "super", "token_type_embeddings", "token_type_ids", "torch", "type_vocab_size", "vocab_size", "word_embeddings", "zeros"], "data2vec/modeling_data2vec_text.py:Data2VecTextEmbeddings.forward": ["FloatTensor", "LayerNorm", "LongTensor", "ModelTextEmbeddings", "Module", "None", "Tensor", "batch_size", "buffered_token_type_ids", "class", "create_position_ids_from_input_ids", "create_position_ids_from_inputs_embeds", "def", "device", "dim", "dropout", "dtype", "else", "embeddings", "expand", "forward", "gather", "hasattr", "if", "index", "input_ids", "input_shape", "inputs_embeds", "int", "is", "long", "nn", "not", "padding_idx", "past_key_values_length", "position_embeddings", "position_ids", "return", "self", "seq_length", "shape", "size", "token_type_embeddings", "token_type_ids", "torch", "word_embeddings", "zeros"], "data2vec/modeling_data2vec_text.py:Data2VecTextEmbeddings.create_position_ids_from_inputs_embeds": ["ModelTextEmbeddings", "Module", "arange", "class", "create_position_ids_from_inputs_embeds", "def", "device", "dtype", "expand", "input_shape", "inputs_embeds", "long", "nn", "padding_idx", "position_ids", "return", "sequence_length", "size", "torch", "unsqueeze"], "data2vec/modeling_data2vec_text.py:Data2VecTextEmbeddings.create_position_ids_from_input_ids": ["ModelTextEmbeddings", "Module", "class", "create_position_ids_from_input_ids", "cumsum", "def", "dim", "incremental_indices", "input_ids", "int", "long", "mask", "ne", "nn", "padding_idx", "past_key_values_length", "return", "torch", "type_as"], "data2vec/modeling_data2vec_text.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "contiguous", "def", "dim", "dropout", "float", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "shape", "size", "softmax", "torch", "training", "transpose", "value"], "data2vec/modeling_data2vec_text.py:Data2VecTextSelfAttention.__init__": ["Dropout", "False", "Linear", "ModelTextSelfAttention", "Module", "None", "The", "ValueError", "__init__", "a", "all_head_size", "and", "attention", "attention_head_size", "attention_probs_dropout_prob", "class", "config", "def", "dropout", "embedding_size", "f", "hasattr", "heads", "hidden", "hidden_size", "if", "int", "is", "is_causal", "is_decoder", "key", "layer_idx", "multiple", "nn", "not", "num_attention_heads", "number", "of", "query", "raise", "scaling", "self", "size", "super", "the", "value"], "data2vec/modeling_data2vec_text.py:Data2VecTextSelfAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "EncoderDecoderCache", "FloatTensor", "ModelTextSelfAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_attn_implementation", "attention_head_size", "attention_interface", "attention_mask", "attn_output", "attn_weights", "cache_position", "class", "config", "contiguous", "current_past_key_values", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "hidden_shape", "hidden_states", "if", "input_shape", "is", "isinstance", "key", "key_layer", "kwargs", "layer_idx", "nn", "not", "p", "past_key_values", "query", "query_layer", "reshape", "return", "scaling", "self", "self_attention_cache", "shape", "torch", "training", "transpose", "tuple", "update", "value", "value_layer", "view"], "data2vec/modeling_data2vec_text.py:Data2VecTextCrossAttention.__init__": ["Dropout", "False", "Linear", "ModelTextCrossAttention", "Module", "None", "The", "ValueError", "__init__", "a", "all_head_size", "and", "attention", "attention_head_size", "attention_probs_dropout_prob", "class", "config", "def", "dropout", "embedding_size", "f", "hasattr", "heads", "hidden", "hidden_size", "if", "int", "is", "is_causal", "key", "layer_idx", "multiple", "nn", "not", "num_attention_heads", "number", "of", "query", "raise", "scaling", "self", "size", "super", "the", "value"], "data2vec/modeling_data2vec_text.py:Data2VecTextCrossAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "EncoderDecoderCache", "False", "FloatTensor", "ModelTextCrossAttention", "Module", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "_attn_implementation", "and", "attention_head_size", "attention_interface", "attention_mask", "attn_output", "attn_weights", "bsz", "class", "config", "contiguous", "cross_attention_cache", "def", "dropout", "eager", "eager_attention_forward", "else", "encoder_hidden_states", "forward", "get", "hidden_states", "if", "is", "is_updated", "key", "key_layer", "keys", "kv_input_shape", "kwargs", "layer_idx", "layers", "nn", "not", "p", "past_key_values", "q_input_shape", "query", "query_layer", "reshape", "return", "scaling", "self", "shape", "src_len", "tgt_len", "torch", "training", "transpose", "tuple", "update", "value", "value_layer", "values", "view"], "data2vec/modeling_data2vec_text.py:Data2VecTextSelfOutput.__init__": ["Dropout", "LayerNorm", "Linear", "ModelTextSelfOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "eps", "hidden_dropout_prob", "hidden_size", "layer_norm_eps", "nn", "self", "super"], "data2vec/modeling_data2vec_text.py:Data2VecTextSelfOutput.forward": ["LayerNorm", "ModelTextSelfOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "data2vec/modeling_data2vec_text.py:Data2VecTextAttention.__init__": ["False", "ModelTextAttention", "ModelTextCrossAttention", "ModelTextSelfAttention", "ModelTextSelfOutput", "Module", "None", "__init__", "attention_class", "class", "config", "def", "else", "if", "is_causal", "is_cross_attention", "layer_idx", "nn", "output", "self", "super"], "data2vec/modeling_data2vec_text.py:Data2VecTextAttention.forward": ["Cache", "FloatTensor", "ModelTextAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attention_output", "attn_weights", "cache_position", "class", "def", "else", "encoder_attention_mask", "encoder_hidden_states", "forward", "hidden_states", "if", "is_cross_attention", "kwargs", "nn", "not", "output", "past_key_values", "return", "self", "torch", "tuple"], "data2vec/modeling_data2vec_text.py:Data2VecTextIntermediate.__init__": ["ACT2FN", "Linear", "ModelTextIntermediate", "Module", "__init__", "class", "config", "def", "dense", "else", "hidden_act", "hidden_size", "if", "intermediate_act_fn", "intermediate_size", "isinstance", "nn", "self", "str", "super"], "data2vec/modeling_data2vec_text.py:Data2VecTextIntermediate.forward": ["ModelTextIntermediate", "Module", "Tensor", "class", "def", "dense", "forward", "hidden_states", "intermediate_act_fn", "nn", "return", "self", "torch"], "data2vec/modeling_data2vec_text.py:Data2VecTextOutput.__init__": ["Dropout", "LayerNorm", "Linear", "ModelTextOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "eps", "hidden_dropout_prob", "hidden_size", "intermediate_size", "layer_norm_eps", "nn", "self", "super"], "data2vec/modeling_data2vec_text.py:Data2VecTextOutput.forward": ["LayerNorm", "ModelTextOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "data2vec/modeling_data2vec_text.py:Data2VecTextLayer.__init__": ["False", "GradientCheckpointingLayer", "ModelTextAttention", "ModelTextIntermediate", "ModelTextLayer", "ModelTextOutput", "None", "True", "ValueError", "__init__", "a", "add_cross_attention", "added", "as", "attention", "be", "chunk_size_feed_forward", "class", "config", "cross", "crossattention", "decoder", "def", "f", "if", "intermediate", "is", "is_causal", "is_cross_attention", "is_decoder", "layer_idx", "model", "not", "output", "raise", "self", "seq_len_dim", "should", "super", "used"], "data2vec/modeling_data2vec_text.py:Data2VecTextLayer.forward": ["Cache", "FloatTensor", "GradientCheckpointingLayer", "If", "ModelTextLayer", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "ValueError", "_", "add_cross_attention", "and", "apply_chunking_to_forward", "are", "attention", "attention_mask", "attention_output", "be", "by", "cache_position", "chunk_size_feed_forward", "class", "config", "cross", "cross_attention_output", "crossattention", "def", "encoder_attention_mask", "encoder_hidden_states", "f", "feed_forward_chunk", "forward", "has", "hasattr", "hidden_states", "if", "instantiated", "is", "is_decoder", "kwargs", "layer_output", "layers", "not", "passed", "past_key_values", "raise", "return", "self", "self_attention_output", "seq_len_dim", "setting", "to", "torch", "tuple", "with"], "data2vec/modeling_data2vec_text.py:Data2VecTextLayer.feed_forward_chunk": ["GradientCheckpointingLayer", "ModelTextLayer", "attention_output", "class", "def", "feed_forward_chunk", "intermediate", "intermediate_output", "layer_output", "output", "return", "self"], "data2vec/modeling_data2vec_text.py:Data2VecTextPreTrainedModel._init_weights": ["ModelTextEmbeddings", "ModelTextPreTrainedModel", "PreTrainedModel", "_init_weights", "arange", "class", "copy_", "def", "expand", "if", "init", "isinstance", "module", "position_ids", "self", "shape", "super", "token_type_ids", "torch", "zeros_"], "data2vec/modeling_data2vec_text.py:Data2VecTextEncoder.__init__": ["ModelTextEncoder", "ModelTextLayer", "Module", "ModuleList", "__init__", "class", "config", "def", "for", "i", "in", "layer", "layer_idx", "nn", "num_hidden_layers", "range", "self", "super"], "data2vec/modeling_data2vec_text.py:Data2VecTextEncoder.forward": ["BaseModelOutputWithPastAndCrossAttentions", "Cache", "FloatTensor", "ModelTextEncoder", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "bool", "cache_position", "class", "def", "else", "encoder_attention_mask", "encoder_hidden_states", "enumerate", "for", "forward", "hidden_states", "i", "if", "in", "kwargs", "last_hidden_state", "layer", "layer_module", "nn", "past_key_values", "return", "self", "torch", "tuple", "use_cache"], "data2vec/modeling_data2vec_text.py:Data2VecTextPooler.__init__": ["Linear", "ModelTextPooler", "Module", "Tanh", "__init__", "activation", "class", "config", "def", "dense", "hidden_size", "nn", "self", "super"], "data2vec/modeling_data2vec_text.py:Data2VecTextPooler.forward": ["ModelTextPooler", "Module", "Tensor", "activation", "class", "def", "dense", "first_token_tensor", "forward", "hidden_states", "nn", "pooled_output", "return", "self", "torch"], "data2vec/modeling_data2vec_text.py:Data2VecTextModel.__init__": ["False", "ModelTextEmbeddings", "ModelTextEncoder", "ModelTextModel", "ModelTextPooler", "ModelTextPreTrainedModel", "None", "True", "__init__", "add_pooling_layer", "class", "config", "def", "else", "embeddings", "encoder", "gradient_checkpointing", "if", "pooler", "post_init", "r", "self", "super"], "data2vec/modeling_data2vec_text.py:Data2VecTextModel.get_input_embeddings": ["ModelTextModel", "ModelTextPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "return", "self", "word_embeddings"], "data2vec/modeling_data2vec_text.py:Data2VecTextModel.set_input_embeddings": ["ModelTextModel", "ModelTextPreTrainedModel", "class", "def", "embeddings", "self", "set_input_embeddings", "value", "word_embeddings"], "data2vec/modeling_data2vec_text.py:Data2VecTextModel.forward": ["BaseModelOutputWithPoolingAndCrossAttentions", "Cache", "DynamicCache", "EncoderDecoderCache", "False", "ModelTextModel", "ModelTextPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "_create_attention_masks", "and", "arange", "attention_mask", "bool", "cache_position", "class", "config", "def", "device", "else", "embedding_output", "embeddings", "encoder", "encoder_attention_mask", "encoder_hidden_states", "encoder_outputs", "exactly", "forward", "get_seq_length", "if", "input_ids", "inputs_embeds", "is", "is_decoder", "is_encoder_decoder", "kwargs", "last_hidden_state", "must", "not", "of", "one", "or", "past_key_values", "past_key_values_length", "pooled_output", "pooler", "pooler_output", "position_ids", "raise", "return", "self", "seq_length", "sequence_output", "shape", "specify", "token_type_ids", "torch", "tuple", "use_cache"], "data2vec/modeling_data2vec_text.py:Data2VecTextModel._create_attention_masks": ["ModelTextModel", "ModelTextPreTrainedModel", "None", "_create_attention_masks", "attention_mask", "cache_position", "class", "config", "create_bidirectional_mask", "create_causal_mask", "def", "else", "embedding_output", "encoder_attention_mask", "encoder_hidden_states", "if", "input_embeds", "is", "is_decoder", "not", "past_key_values", "return", "self"], "data2vec/modeling_data2vec_text.py:Data2VecTextLMHead.__init__": ["LayerNorm", "Linear", "ModelTextLMHead", "Module", "Parameter", "__init__", "bias", "class", "config", "decoder", "def", "dense", "eps", "hidden_size", "layer_norm", "layer_norm_eps", "nn", "self", "super", "torch", "vocab_size", "zeros"], "data2vec/modeling_data2vec_text.py:Data2VecTextLMHead.forward": ["ModelTextLMHead", "Module", "class", "decoder", "def", "dense", "features", "forward", "gelu", "kwargs", "layer_norm", "nn", "return", "self", "x"], "data2vec/modeling_data2vec_text.py:Data2VecTextClassificationHead.__init__": ["Dropout", "Linear", "ModelTextClassificationHead", "Module", "None", "__init__", "class", "classifier_dropout", "config", "def", "dense", "dropout", "else", "hidden_dropout_prob", "hidden_size", "if", "is", "nn", "not", "num_labels", "out_proj", "self", "super"], "data2vec/modeling_data2vec_text.py:Data2VecTextClassificationHead.forward": ["ModelTextClassificationHead", "Module", "class", "def", "dense", "dropout", "features", "forward", "kwargs", "nn", "out_proj", "return", "self", "tanh", "torch", "x"], "data2vec/modeling_data2vec_text.py:Data2VecTextForCausalLM.__init__": ["False", "GenerationMixin", "If", "ModelTextForCausalLM", "ModelTextLMHead", "ModelTextLMHeadModel", "ModelTextModel", "ModelTextPreTrainedModel", "Model_text", "True", "__init__", "a", "add", "add_pooling_layer", "as", "class", "config", "def", "if", "is_decoder", "lm_head", "logger", "not", "post_init", "self", "standalone", "super", "to", "use", "want", "warning", "you"], "data2vec/modeling_data2vec_text.py:Data2VecTextForCausalLM.get_output_embeddings": ["GenerationMixin", "ModelTextForCausalLM", "ModelTextPreTrainedModel", "class", "decoder", "def", "get_output_embeddings", "lm_head", "return", "self"], "data2vec/modeling_data2vec_text.py:Data2VecTextForCausalLM.set_output_embeddings": ["GenerationMixin", "ModelTextForCausalLM", "ModelTextPreTrainedModel", "class", "decoder", "def", "lm_head", "new_embeddings", "self", "set_output_embeddings"], "data2vec/modeling_data2vec_text.py:Data2VecTextForCausalLM.forward": ["BaseModelOutputWithPoolingAndCrossAttentions", "CausalLMOutputWithCrossAttentions", "False", "FloatTensor", "GenerationMixin", "LongTensor", "ModelTextForCausalLM", "ModelTextPreTrainedModel", "Model_text", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "cross_attentions", "def", "else", "encoder_attention_mask", "encoder_hidden_states", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "not", "outputs", "past_key_values", "position_ids", "r", "return", "return_dict", "self", "slice", "slice_indices", "token_type_ids", "torch", "tuple", "use_cache", "vocab_size"], "data2vec/modeling_data2vec_text.py:Data2VecTextForMaskedLM.__init__": ["False", "If", "ModelTextForMaskedLM", "ModelTextLMHead", "ModelTextModel", "ModelTextPreTrainedModel", "Model_text", "__init__", "add_pooling_layer", "attention", "bi", "class", "config", "def", "directional", "for", "if", "is_decoder", "lm_head", "logger", "make", "post_init", "self", "super", "sure", "to", "use", "want", "warning", "you"], "data2vec/modeling_data2vec_text.py:Data2VecTextForMaskedLM.get_output_embeddings": ["ModelTextForMaskedLM", "ModelTextPreTrainedModel", "class", "decoder", "def", "get_output_embeddings", "lm_head", "return", "self"], "data2vec/modeling_data2vec_text.py:Data2VecTextForMaskedLM.set_output_embeddings": ["ModelTextForMaskedLM", "ModelTextPreTrainedModel", "class", "decoder", "def", "lm_head", "new_embeddings", "self", "set_output_embeddings"], "data2vec/modeling_data2vec_text.py:Data2VecTextForMaskedLM.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "MaskedLMOutput", "ModelTextForMaskedLM", "ModelTextPreTrainedModel", "Model_text", "None", "TransformersKwargs", "True", "Unpack", "attention_mask", "attentions", "class", "config", "def", "device", "encoder_attention_mask", "encoder_hidden_states", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "lm_head", "logits", "loss", "loss_fct", "masked_lm_loss", "not", "outputs", "position_ids", "prediction_scores", "r", "return", "return_dict", "self", "sequence_output", "to", "token_type_ids", "torch", "tuple", "view", "vocab_size"], "data2vec/modeling_data2vec_text.py:Data2VecTextForSequenceClassification.__init__": ["False", "ModelTextClassificationHead", "ModelTextForSequenceClassification", "ModelTextModel", "ModelTextPreTrainedModel", "Model_text", "__init__", "add_pooling_layer", "class", "classifier", "config", "def", "num_labels", "post_init", "self", "super"], "data2vec/modeling_data2vec_text.py:Data2VecTextForSequenceClassification.forward": ["BCEWithLogitsLoss", "CrossEntropyLoss", "FloatTensor", "LongTensor", "MSELoss", "ModelTextForSequenceClassification", "ModelTextPreTrainedModel", "Model_text", "None", "SequenceClassifierOutput", "TransformersKwargs", "True", "Unpack", "and", "attention_mask", "attentions", "class", "classifier", "config", "def", "device", "dtype", "elif", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "kwargs", "labels", "logits", "long", "loss", "loss_fct", "multi_label_classification", "not", "num_labels", "or", "outputs", "position_ids", "problem_type", "r", "regression", "return", "return_dict", "self", "sequence_output", "single_label_classification", "squeeze", "to", "token_type_ids", "torch", "tuple", "view"], "data2vec/modeling_data2vec_text.py:Data2VecTextForMultipleChoice.__init__": ["Dropout", "Linear", "ModelTextForMultipleChoice", "ModelTextModel", "ModelTextPreTrainedModel", "Model_text", "__init__", "class", "classifier", "config", "def", "dropout", "hidden_dropout_prob", "hidden_size", "nn", "post_init", "self", "super"], "data2vec/modeling_data2vec_text.py:Data2VecTextForMultipleChoice.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "ModelTextForMultipleChoice", "ModelTextPreTrainedModel", "Model_text", "MultipleChoiceModelOutput", "None", "TransformersKwargs", "True", "Unpack", "attention_mask", "attentions", "class", "classifier", "def", "device", "dropout", "else", "flat_attention_mask", "flat_input_ids", "flat_inputs_embeds", "flat_position_ids", "flat_token_type_ids", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "not", "num_choices", "outputs", "pooled_output", "position_ids", "r", "reshaped_logits", "return", "return_dict", "self", "shape", "size", "to", "token_type_ids", "torch", "tuple", "view"], "data2vec/modeling_data2vec_text.py:Data2VecTextForTokenClassification.__init__": ["Dropout", "False", "Linear", "ModelTextForTokenClassification", "ModelTextModel", "ModelTextPreTrainedModel", "Model_text", "None", "__init__", "add_pooling_layer", "class", "classifier", "classifier_dropout", "config", "def", "dropout", "else", "hidden_dropout_prob", "hidden_size", "if", "is", "nn", "not", "num_labels", "post_init", "self", "super"], "data2vec/modeling_data2vec_text.py:Data2VecTextForTokenClassification.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "ModelTextForTokenClassification", "ModelTextPreTrainedModel", "Model_text", "None", "TokenClassifierOutput", "TransformersKwargs", "True", "Unpack", "attention_mask", "attentions", "class", "classifier", "def", "device", "dropout", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "not", "num_labels", "outputs", "position_ids", "r", "return", "return_dict", "self", "sequence_output", "to", "token_type_ids", "torch", "tuple", "view"], "data2vec/modeling_data2vec_text.py:Data2VecTextForQuestionAnswering.__init__": ["False", "Linear", "ModelTextForQuestionAnswering", "ModelTextModel", "ModelTextPreTrainedModel", "Model_text", "__init__", "add_pooling_layer", "class", "config", "def", "hidden_size", "nn", "num_labels", "post_init", "qa_outputs", "self", "super"], "data2vec/modeling_data2vec_text.py:Data2VecTextForQuestionAnswering.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "ModelTextForQuestionAnswering", "ModelTextPreTrainedModel", "Model_text", "None", "QuestionAnsweringModelOutput", "TransformersKwargs", "True", "Unpack", "and", "attention_mask", "attentions", "clamp", "class", "contiguous", "def", "dim", "end_logits", "end_loss", "end_positions", "forward", "hidden_states", "if", "ignore_index", "ignored_index", "input_ids", "inputs_embeds", "is", "kwargs", "len", "logits", "loss", "loss_fct", "not", "outputs", "position_ids", "qa_outputs", "return", "return_dict", "self", "sequence_output", "size", "split", "squeeze", "start_logits", "start_loss", "start_positions", "token_type_ids", "torch", "total_loss", "tuple"], "data2vec/modeling_data2vec_vision.py:drop_path": ["False", "Model_path", "Model_prob", "Tensor", "bool", "def", "device", "div", "dtype", "float", "floor_", "if", "input", "keep_prob", "ndim", "not", "or", "output", "rand", "random_tensor", "return", "shape", "torch", "training"], "data2vec/modeling_data2vec_vision.py:Data2VecVisionDropPath.__init__": ["ModelVisionDropPath", "Module", "None", "__init__", "class", "def", "drop_prob", "float", "nn", "self", "super"], "data2vec/modeling_data2vec_vision.py:Data2VecVisionDropPath.forward": ["ModelVisionDropPath", "Module", "Tensor", "class", "def", "drop_path", "drop_prob", "forward", "hidden_states", "nn", "return", "self", "torch", "training"], "data2vec/modeling_data2vec_vision.py:Data2VecVisionDropPath.extra_repr": ["ModelVisionDropPath", "Module", "class", "def", "drop_prob", "extra_repr", "f", "nn", "p", "return", "self", "str"], "data2vec/modeling_data2vec_vision.py:Data2VecVisionEmbeddings.__init__": ["Dropout", "Iterable", "ModelVisionConfig", "ModelVisionEmbeddings", "ModelVisionPatchEmbeddings", "Module", "None", "Parameter", "__init__", "abc", "class", "cls_token", "collections", "config", "def", "dropout", "else", "hidden_dropout_prob", "hidden_size", "if", "image_size", "isinstance", "mask_token", "nn", "num_patches", "patch_embeddings", "patch_size", "position_embeddings", "self", "super", "torch", "use_absolute_position_embeddings", "use_mask_token", "zeros"], "data2vec/modeling_data2vec_vision.py:Data2VecVisionEmbeddings.interpolate_pos_encoding": ["False", "ModelVisionEmbeddings", "Module", "Tensor", "align_corners", "and", "bicubic", "cat", "class", "class_pos_embed", "def", "dim", "embeddings", "functional", "height", "if", "int", "interpolate", "interpolate_pos_encoding", "is_tracing", "jit", "mode", "new_height", "new_width", "nn", "not", "num_patches", "num_positions", "patch_pos_embed", "patch_size", "permute", "position_embeddings", "reshape", "return", "self", "shape", "size", "sqrt_num_positions", "torch", "torch_int", "view", "width"], "data2vec/modeling_data2vec_vision.py:Data2VecVisionEmbeddings.forward": ["BoolTensor", "ModelVisionEmbeddings", "Module", "None", "Tensor", "_", "batch_size", "bool_masked_pos", "cat", "class", "cls_token", "cls_tokens", "def", "dim", "dropout", "embeddings", "expand", "forward", "height", "if", "interpolate_pos_encoding", "is", "mask_token", "mask_tokens", "nn", "not", "patch_embeddings", "patch_height", "patch_width", "pixel_values", "position_embeddings", "return", "self", "seq_len", "shape", "size", "torch", "type_as", "unsqueeze", "w", "width"], "data2vec/modeling_data2vec_vision.py:Data2VecVisionPatchEmbeddings.__init__": ["Conv2d", "Iterable", "ModelVisionPatchEmbeddings", "Module", "__init__", "abc", "class", "collections", "config", "def", "else", "hidden_size", "if", "image_size", "isinstance", "kernel_size", "nn", "num_channels", "num_patches", "patch_shape", "patch_size", "projection", "self", "stride", "super"], "data2vec/modeling_data2vec_vision.py:Data2VecVisionPatchEmbeddings.forward": ["Make", "ModelVisionPatchEmbeddings", "Module", "Tensor", "ValueError", "batch_size", "channel", "class", "configuration", "def", "dimension", "dtype", "embeddings", "flatten", "forward", "height", "if", "in", "match", "nn", "num_channels", "of", "one", "patch_height", "patch_width", "pixel", "pixel_values", "projection", "raise", "return", "self", "set", "shape", "sure", "that", "the", "to", "torch", "transpose", "values", "weight", "width", "with"], "data2vec/modeling_data2vec_vision.py:Data2VecVisionSelfAttention.__init__": ["Dropout", "False", "Linear", "ModelVisionConfig", "ModelVisionRelativePositionBias", "ModelVisionSelfAttention", "Module", "None", "The", "ValueError", "__init__", "a", "all_head_size", "and", "attention", "attention_head_size", "attention_probs_dropout_prob", "bias", "bool", "class", "config", "def", "dropout", "embedding_size", "f", "has_relative_position_bias", "hasattr", "heads", "hidden", "hidden_size", "if", "int", "is", "key", "multiple", "nn", "not", "num_attention_heads", "number", "of", "query", "raise", "relative_position_bias", "self", "size", "super", "the", "tuple", "value", "window_size"], "data2vec/modeling_data2vec_vision.py:Data2VecVisionSelfAttention.forward": ["False", "ModelVisionSelfAttention", "Module", "None", "Tensor", "_", "all_head_size", "attention_head_size", "attention_probs", "attention_scores", "batch_size", "bool", "class", "config", "context_layer", "contiguous", "def", "dim", "dim_size", "dropout", "else", "forward", "functional", "has_relative_position_bias", "height", "hidden_states", "if", "int", "interpolate_pos_encoding", "is", "key", "key_layer", "math", "matmul", "new_context_layer_shape", "nn", "not", "num_attention_heads", "output_attentions", "outputs", "patch_size", "permute", "query", "query_layer", "relative_position_bias", "resolution", "return", "self", "seq_length", "shape", "size", "softmax", "sqrt", "torch", "transpose", "tuple", "value", "value_layer", "view", "width", "window_size"], "data2vec/modeling_data2vec_vision.py:Data2VecVisionSdpaSelfAttention.forward": ["False", "If", "ModelVisionSdpaSelfAttention", "ModelVisionSelfAttention", "None", "Tensor", "The", "True", "_", "__class__", "__name__", "all_head_size", "attention", "attention_head_size", "attention_probs_dropout_prob", "attn_bias", "attn_implementation", "attn_mask", "batch_size", "be", "bool", "class", "config", "context_layer", "contiguous", "def", "dim_size", "does", "dropout_p", "eager", "else", "f", "forward", "functional", "get", "has_relative_position_bias", "height", "hidden_states", "if", "int", "interpolate_pos_encoding", "is", "is_causal", "key", "key_layer", "loading", "logger", "math", "model", "new_context_layer_shape", "nn", "not", "num_attention_heads", "output_attentions", "patch_size", "permute", "please", "query", "query_layer", "relative_position_bias", "resolution", "return", "returned", "scale", "scaled_dot_product_attention", "scaling", "self", "seq_length", "set", "shape", "size", "sqrt", "support", "the", "to", "torch", "training", "transpose", "tuple", "value", "value_layer", "view", "want", "warning_once", "weights", "when", "width", "will", "window_size", "you"], "data2vec/modeling_data2vec_vision.py:Data2VecVisionSelfOutput.__init__": ["Dropout", "Linear", "ModelVisionConfig", "ModelVisionSelfOutput", "Module", "None", "__init__", "class", "config", "def", "dense", "dropout", "hidden_dropout_prob", "hidden_size", "nn", "self", "super"], "data2vec/modeling_data2vec_vision.py:Data2VecVisionSelfOutput.forward": ["ModelVisionSelfOutput", "Module", "None", "Tensor", "class", "def", "dense", "dropout", "forward", "gamma", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "data2vec/modeling_data2vec_vision.py:Data2VecVisionAttention.__init__": ["ModelVisionAttention", "ModelVisionConfig", "ModelVisionSelfOutput", "Model_VISION_SELF_ATTENTION_CLASSES", "Module", "None", "__init__", "_attn_implementation", "attention", "class", "config", "def", "nn", "output", "self", "super", "tuple", "window_size"], "data2vec/modeling_data2vec_vision.py:Data2VecVisionAttention.forward": ["False", "ModelVisionAttention", "ModelVisionRelativePositionBias", "Module", "None", "Optional", "Tensor", "attention", "attention_output", "bool", "class", "def", "forward", "hidden_states", "int", "interpolate_pos_encoding", "nn", "output", "output_attentions", "outputs", "relative_position_bias", "resolution", "return", "self", "self_outputs", "torch", "tuple"], "data2vec/modeling_data2vec_vision.py:Data2VecVisionIntermediate.__init__": ["ACT2FN", "Linear", "ModelVisionConfig", "ModelVisionIntermediate", "Module", "None", "__init__", "class", "config", "def", "dense", "else", "hidden_act", "hidden_size", "if", "intermediate_act_fn", "intermediate_size", "isinstance", "nn", "self", "str", "super"], "data2vec/modeling_data2vec_vision.py:Data2VecVisionIntermediate.forward": ["ModelVisionIntermediate", "Module", "Tensor", "class", "def", "dense", "forward", "hidden_states", "intermediate_act_fn", "nn", "return", "self", "torch"], "data2vec/modeling_data2vec_vision.py:Data2VecVisionOutput.__init__": ["Dropout", "Linear", "ModelVisionConfig", "ModelVisionOutput", "Module", "None", "__init__", "class", "config", "def", "dense", "dropout", "hidden_dropout_prob", "hidden_size", "intermediate_size", "nn", "self", "super"], "data2vec/modeling_data2vec_vision.py:Data2VecVisionOutput.forward": ["ModelVisionOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "nn", "return", "self", "torch"], "data2vec/modeling_data2vec_vision.py:Data2VecVisionLayer.__init__": ["GradientCheckpointingLayer", "Identity", "LayerNorm", "ModelVisionAttention", "ModelVisionConfig", "ModelVisionDropPath", "ModelVisionIntermediate", "ModelVisionLayer", "ModelVisionOutput", "None", "Parameter", "True", "__init__", "attention", "chunk_size_feed_forward", "class", "config", "def", "drop_path", "drop_path_rate", "else", "eps", "float", "hidden_size", "if", "init_values", "intermediate", "lambda_1", "lambda_2", "layer_norm_eps", "layer_scale_init_value", "layernorm_after", "layernorm_before", "nn", "ones", "output", "requires_grad", "self", "seq_len_dim", "super", "torch", "tuple", "window_size"], "data2vec/modeling_data2vec_vision.py:Data2VecVisionLayer.forward": ["False", "GradientCheckpointingLayer", "ModelVisionLayer", "None", "Tensor", "attention", "attention_output", "bool", "class", "def", "drop_path", "forward", "hidden_states", "if", "int", "intermediate", "interpolate_pos_encoding", "is", "lambda_1", "lambda_2", "layer_output", "layernorm_after", "layernorm_before", "not", "output", "output_attentions", "outputs", "relative_position_bias", "resolution", "return", "self", "self_attention_outputs", "torch", "tuple"], "data2vec/modeling_data2vec_vision.py:Data2VecVisionRelativePositionBias.__init__": ["ModelVisionConfig", "ModelVisionRelativePositionBias", "Module", "None", "Parameter", "__init__", "class", "config", "def", "nn", "num_attention_heads", "num_relative_distance", "relative_position_bias_table", "self", "super", "torch", "tuple", "window_size", "zeros"], "data2vec/modeling_data2vec_vision.py:Data2VecVisionRelativePositionBias.generate_relative_position_index": ["ModelVisionRelativePositionBias", "Module", "None", "Tensor", "arange", "class", "contiguous", "coords", "coords_flatten", "def", "dtype", "flatten", "generate_relative_position_index", "grid", "ij", "indexing", "int", "meshgrid", "nn", "num_relative_distance", "permute", "relative_coords", "relative_position_index", "return", "self", "size", "stack", "sum", "torch", "tuple", "window_area", "window_size", "zeros"], "data2vec/modeling_data2vec_vision.py:Data2VecVisionRelativePositionBias.forward": ["False", "ModelVisionRelativePositionBias", "Module", "None", "Tensor", "align_corners", "bilinear", "bool", "cat", "class", "contiguous", "def", "dim_size", "forward", "functional", "generate_relative_position_index", "if", "interpolate", "interpolate_pos_encoding", "mode", "new_height", "new_num_relative_distance", "new_relative_position_bias_table", "new_sub_table", "new_width", "nn", "num_relative_distance", "old_height", "old_num_relative_distance", "old_relative_position_bias_table", "old_sub_table", "old_width", "permute", "relative_position_bias", "relative_position_bias_table", "relative_position_index", "reshape", "return", "self", "size", "squeeze", "torch", "torch_int", "unsqueeze", "view", "window_size"], "data2vec/modeling_data2vec_vision.py:Data2VecVisionEncoder.__init__": ["False", "ModelVisionConfig", "ModelVisionEncoder", "ModelVisionLayer", "ModelVisionRelativePositionBias", "Module", "ModuleList", "None", "__init__", "class", "config", "cpu", "def", "device", "dpr", "drop_path_rate", "else", "for", "gradient_checkpointing", "has_relative_position_bias", "i", "if", "in", "item", "layer", "linspace", "nn", "num_hidden_layers", "range", "relative_position_bias", "self", "super", "torch", "tuple", "use_relative_position_bias", "use_shared_relative_position_bias", "window_size", "x"], "data2vec/modeling_data2vec_vision.py:Data2VecVisionEncoder.forward": ["BaseModelOutput", "False", "ModelVisionEncoder", "Module", "None", "Tensor", "True", "all_hidden_states", "all_self_attentions", "attentions", "bool", "class", "config", "def", "dim_size", "else", "enumerate", "for", "forward", "has_relative_position_bias", "height", "hidden_states", "i", "if", "in", "int", "interpolate_pos_encoding", "is", "last_hidden_state", "layer", "layer_module", "layer_outputs", "nn", "not", "output_attentions", "output_hidden_states", "patch_size", "relative_position_bias", "resolution", "return", "return_dict", "self", "shape", "torch", "tuple", "v", "width", "window_size"], "data2vec/modeling_data2vec_vision.py:Data2VecVisionPreTrainedModel._init_weights": ["ModelVisionEmbeddings", "ModelVisionLayer", "ModelVisionPreTrainedModel", "ModelVisionRelativePositionBias", "None", "PreTrainedModel", "_init_weights", "class", "cls_token", "config", "constant_", "def", "elif", "if", "init", "is", "isinstance", "lambda_1", "lambda_2", "layer_scale_init_value", "mask_token", "module", "not", "position_embeddings", "relative_position_bias_table", "self", "super", "zeros_"], "data2vec/modeling_data2vec_vision.py:Data2VecVisionModel.__init__": ["False", "Identity", "LayerNorm", "ModelVisionConfig", "ModelVisionEmbeddings", "ModelVisionEncoder", "ModelVisionModel", "ModelVisionPooler", "ModelVisionPreTrainedModel", "None", "__init__", "add_pooling_layer", "bool", "class", "config", "def", "else", "embeddings", "encoder", "eps", "hidden_size", "if", "layer_norm_eps", "layernorm", "nn", "patch_embeddings", "patch_shape", "pooler", "post_init", "r", "self", "super", "use_mean_pooling", "window_size"], "data2vec/modeling_data2vec_vision.py:Data2VecVisionModel.get_input_embeddings": ["ModelVisionModel", "ModelVisionPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "patch_embeddings", "return", "self"], "data2vec/modeling_data2vec_vision.py:Data2VecVisionModel.forward": ["BoolTensor", "False", "ModelVisionModel", "ModelVisionModelOutputWithPooling", "ModelVisionPreTrainedModel", "None", "Tensor", "_", "attentions", "bool", "bool_masked_pos", "class", "config", "def", "else", "embedding_output", "embeddings", "encoder", "encoder_outputs", "forward", "head_outputs", "hidden_states", "if", "interpolate_pos_encoding", "is", "kwargs", "last_hidden_state", "layernorm", "not", "output_attentions", "output_hidden_states", "pixel_values", "pooled_output", "pooler", "pooler_output", "r", "resolution", "return", "return_dict", "self", "sequence_output", "shape", "torch", "tuple", "use_return_dict"], "data2vec/modeling_data2vec_vision.py:Data2VecVisionPooler.__init__": ["LayerNorm", "ModelVisionConfig", "ModelVisionPooler", "Module", "None", "__init__", "class", "config", "def", "else", "eps", "hidden_size", "if", "layer_norm_eps", "layernorm", "nn", "self", "super", "use_mean_pooling"], "data2vec/modeling_data2vec_vision.py:Data2VecVisionPooler.forward": ["ModelVisionPooler", "Module", "None", "Tensor", "class", "def", "else", "forward", "hidden_states", "if", "is", "layernorm", "mean", "nn", "not", "patch_tokens", "pooled_output", "return", "self", "torch"], "data2vec/modeling_data2vec_vision.py:Data2VecVisionForImageClassification.__init__": ["Identity", "Linear", "ModelVisionConfig", "ModelVisionForImageClassification", "ModelVisionModel", "ModelVisionPreTrainedModel", "Model_vision", "None", "True", "__init__", "add_pooling_layer", "class", "classifier", "config", "def", "else", "hidden_size", "if", "nn", "num_labels", "post_init", "self", "super"], "data2vec/modeling_data2vec_vision.py:Data2VecVisionForImageClassification.forward": ["False", "ImageClassifierOutput", "ModelVisionForImageClassification", "ModelVisionPreTrainedModel", "Model_vision", "None", "Tensor", "attentions", "bool", "class", "classifier", "config", "def", "else", "forward", "hidden_states", "if", "interpolate_pos_encoding", "is", "kwargs", "labels", "logits", "loss", "loss_function", "not", "output", "output_attentions", "output_hidden_states", "outputs", "pixel_values", "pooled_output", "pooler_output", "r", "return", "return_dict", "self", "torch", "tuple", "use_return_dict"], "data2vec/modeling_data2vec_vision.py:Data2VecVisionConvModule.__init__": ["BatchNorm2d", "Conv2d", "False", "ModelVisionConvModule", "Module", "None", "ReLU", "__init__", "activation", "bias", "bn", "bool", "class", "conv", "def", "dilation", "in_channels", "int", "kernel_size", "nn", "out_channels", "padding", "self", "str", "super", "tuple"], "data2vec/modeling_data2vec_vision.py:Data2VecVisionConvModule.forward": ["ModelVisionConvModule", "Module", "Tensor", "activation", "bn", "class", "conv", "def", "forward", "input", "nn", "output", "return", "self", "torch"], "data2vec/modeling_data2vec_vision.py:Data2VecVisionPyramidPoolingBlock.__init__": ["AdaptiveAvgPool2d", "ModelVisionConvModule", "ModelVisionPyramidPoolingBlock", "Module", "None", "__init__", "add_module", "channels", "class", "def", "enumerate", "for", "i", "in", "in_channels", "int", "kernel_size", "layer", "layers", "nn", "pool_scale", "self", "str", "super"], "data2vec/modeling_data2vec_vision.py:Data2VecVisionPyramidPoolingBlock.forward": ["ModelVisionPyramidPoolingBlock", "Module", "Tensor", "class", "def", "for", "forward", "hidden_state", "in", "input", "layer", "layers", "nn", "return", "self", "torch"], "data2vec/modeling_data2vec_vision.py:Data2VecVisionPyramidPoolingModule.__init__": ["ModelVisionPyramidPoolingBlock", "ModelVisionPyramidPoolingModule", "Module", "None", "__init__", "add_module", "align_corners", "append", "block", "blocks", "bool", "channels", "class", "def", "enumerate", "for", "i", "in", "in_channels", "int", "nn", "pool_scale", "pool_scales", "self", "str", "super", "tuple"], "data2vec/modeling_data2vec_vision.py:Data2VecVisionPyramidPoolingModule.forward": ["ModelVisionPyramidPoolingModule", "Module", "Tensor", "align_corners", "append", "bilinear", "blocks", "class", "def", "for", "forward", "functional", "in", "interpolate", "list", "mode", "nn", "ppm", "ppm_out", "ppm_outs", "return", "self", "size", "torch", "upsampled_ppm_out", "x"], "data2vec/modeling_data2vec_vision.py:Data2VecVisionUperHead.__init__": ["Conv2d", "False", "ModelVisionConfig", "ModelVisionConvModule", "ModelVisionPyramidPoolingModule", "ModelVisionUperHead", "Module", "ModuleList", "None", "__init__", "align_corners", "append", "bottleneck", "channels", "class", "classifier", "config", "def", "for", "fpn_bottleneck", "fpn_conv", "fpn_convs", "hidden_size", "in", "in_channels", "kernel_size", "l_conv", "lateral_convs", "len", "nn", "num_labels", "padding", "pool_scales", "psp_modules", "self", "super"], "data2vec/modeling_data2vec_vision.py:Data2VecVisionUperHead.psp_forward": ["ModelVisionUperHead", "Module", "bottleneck", "cat", "class", "def", "dim", "extend", "inputs", "nn", "output", "psp_forward", "psp_modules", "psp_outs", "return", "self", "torch", "x"], "data2vec/modeling_data2vec_vision.py:Data2VecVisionUperHead.forward": ["ModelVisionUperHead", "Module", "Tensor", "align_corners", "append", "bilinear", "cat", "class", "classifier", "def", "dim", "encoder_hidden_states", "enumerate", "for", "forward", "fpn_bottleneck", "fpn_convs", "fpn_outs", "functional", "i", "in", "interpolate", "lateral_conv", "lateral_convs", "laterals", "len", "mode", "nn", "output", "prev_shape", "psp_forward", "range", "return", "self", "shape", "size", "torch", "used_backbone_levels"], "data2vec/modeling_data2vec_vision.py:Data2VecVisionFCNHead.__init__": ["Conv2d", "Identity", "ModelVisionConfig", "ModelVisionConvModule", "ModelVisionFCNHead", "Module", "None", "Sequential", "__init__", "append", "auxiliary_channels", "auxiliary_concat_input", "auxiliary_num_convs", "channels", "class", "classifier", "concat_input", "config", "conv_cat", "conv_padding", "convs", "def", "dilation", "else", "for", "hidden_size", "i", "if", "in", "in_channels", "in_index", "int", "kernel_size", "nn", "num_convs", "num_labels", "padding", "range", "self", "super", "tuple"], "data2vec/modeling_data2vec_vision.py:Data2VecVisionFCNHead.forward": ["ModelVisionFCNHead", "Module", "Tensor", "cat", "class", "classifier", "concat_input", "conv_cat", "convs", "def", "dim", "encoder_hidden_states", "forward", "hidden_states", "if", "in_index", "nn", "output", "return", "self", "torch"], "data2vec/modeling_data2vec_vision.py:Data2VecVisionForSemanticSegmentation.__init__": ["BatchNorm2d", "ConvTranspose2d", "False", "GELU", "Identity", "MaxPool2d", "ModelVisionConfig", "ModelVisionFCNHead", "ModelVisionForSemanticSegmentation", "ModelVisionModel", "ModelVisionPreTrainedModel", "ModelVisionUperHead", "Model_vision", "None", "One", "Sequential", "ValueError", "__init__", "a", "add_pooling_layer", "architecture", "auxiliary_head", "backbone", "base", "be", "can", "case", "class", "config", "decode_head", "def", "else", "features", "fpn1", "fpn2", "fpn3", "fpn4", "from", "hidden_size", "if", "in", "integers", "kernel_size", "len", "list", "nn", "num_labels", "of", "out_indices", "post_init", "raise", "requires", "self", "sized", "specifying", "stride", "super", "the", "to", "use", "use_auxiliary_head", "which"], "data2vec/modeling_data2vec_vision.py:Data2VecVisionForSemanticSegmentation.compute_loss": ["CrossEntropyLoss", "False", "ModelVisionForSemanticSegmentation", "ModelVisionPreTrainedModel", "None", "align_corners", "auxiliary_logits", "auxiliary_loss", "auxiliary_loss_weight", "bilinear", "class", "compute_loss", "config", "def", "functional", "if", "ignore_index", "interpolate", "is", "labels", "logits", "loss", "loss_fct", "main_loss", "mode", "nn", "not", "return", "self", "semantic_loss_ignore_index", "shape", "size", "upsampled_auxiliary_logits", "upsampled_logits"], "data2vec/modeling_data2vec_vision.py:Data2VecVisionForSemanticSegmentation.forward": ["False", "ModelVisionForSemanticSegmentation", "ModelVisionPreTrainedModel", "Model_vision", "None", "SemanticSegmenterOutput", "Tensor", "The", "True", "ValueError", "and", "attentions", "auxiliary_head", "auxiliary_logits", "batch_size", "be", "bool", "class", "compute_loss", "config", "decode_head", "def", "else", "encoder_hidden_states", "enumerate", "feature", "features", "for", "forward", "fpn1", "fpn2", "fpn3", "fpn4", "greater", "hidden_states", "i", "idx", "if", "image_size", "in", "interpolate_pos_encoding", "is", "kwargs", "labels", "len", "logits", "loss", "not", "num_labels", "number", "of", "one", "ops", "out_indices", "output", "output_attentions", "output_hidden_states", "outputs", "patch_resolution", "patch_size", "permute", "pixel_values", "r", "raise", "range", "reshape", "return", "return_dict", "self", "shape", "should", "than", "torch", "tuple", "use_return_dict", "x"], "dbrx/modeling_dbrx.py:DbrxRotaryEmbedding.__init__": ["Callable", "False", "ModelConfig", "ModelRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "dbrx/modeling_dbrx.py:DbrxRotaryEmbedding.compute_default_rope_parameters": ["ModelConfig", "ModelRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "dbrx/modeling_dbrx.py:DbrxRotaryEmbedding.forward": ["False", "ModelRotaryEmbedding", "Module", "None", "and", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "dbrx/modeling_dbrx.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "dbrx/modeling_dbrx.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cos", "def", "k", "k_embed", "q", "q_embed", "return", "rotate_half", "sin", "unsqueeze", "unsqueeze_dim"], "dbrx/modeling_dbrx.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "dbrx/modeling_dbrx.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value", "value_states"], "dbrx/modeling_dbrx.py:DbrxAttention.__init__": ["False", "Linear", "ModelAttention", "Module", "None", "True", "Wqkv", "__init__", "attention_dropout", "attn_config", "attn_pdrop", "bias", "class", "clip_qkv", "config", "d_model", "def", "head_dim", "hidden_size", "int", "is_causal", "kv_n_heads", "kwargs", "layer_idx", "max_position_embeddings", "max_seq_len", "n_heads", "nn", "num_heads", "num_key_value_groups", "num_key_value_heads", "out_proj", "rope_theta", "scaling", "self", "super"], "dbrx/modeling_dbrx.py:DbrxAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "LongTensor", "ModelAttention", "Module", "None", "Tensor", "Wqkv", "_attn_implementation", "apply_rotary_pos_emb", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "cache_kwargs", "cache_position", "clamp", "class", "clip_qkv", "config", "contiguous", "cos", "def", "dim", "dropout", "eager", "eager_attention_forward", "else", "forward", "head_dim", "hidden_shape", "hidden_size", "hidden_states", "if", "input_shape", "is", "key_states", "kwargs", "layer_idx", "max", "min", "min_val", "nn", "not", "num_key_value_heads", "out_proj", "past_key_values", "position_embeddings", "qkv_states", "query_states", "reshape", "return", "scaling", "self", "shape", "sin", "split", "torch", "training", "transpose", "tuple", "update", "value_states", "view"], "dbrx/modeling_dbrx.py:DbrxExpertGLU.__init__": ["ACT2FN", "ModelExpertGLU", "Module", "Parameter", "__init__", "act_fn_name", "activation_fn", "class", "config", "def", "empty", "ffn_act_fn", "ffn_hidden_size", "get", "hidden_size", "moe_num_experts", "name", "nn", "self", "silu", "super", "torch", "v1", "w1", "w2"], "dbrx/modeling_dbrx.py:DbrxExpertGLU.forward": ["ModelExpertGLU", "Module", "Tensor", "activation_fn", "class", "def", "down_proj", "expert_v1", "expert_w1", "expert_w2", "forward", "gate_proj", "intermediate_states", "matmul", "nn", "return", "self", "t", "torch", "up_proj", "x"], "dbrx/modeling_dbrx.py:DbrxExperts.__init__": ["ModelExpertGLU", "ModelExperts", "Module", "__init__", "class", "config", "def", "ffn_hidden_size", "hidden_size", "mlp", "moe_num_experts", "nn", "num_experts", "self", "super"], "dbrx/modeling_dbrx.py:DbrxExperts.forward": ["ModelExperts", "Module", "None", "Tensor", "batch_size", "class", "def", "device", "dim", "dtype", "expert_hit", "expert_idx", "expert_mask", "ffn_hidden_size", "for", "forward", "functional", "greater", "hidden_size", "hidden_states", "idx", "in", "index_add_", "mlp", "next_states", "nn", "no_grad", "nonzero", "num_classes", "num_experts", "one_hot", "permute", "reshape", "return", "self", "shape", "split_expert_shape", "states", "sum", "token_idx", "top_k_index", "top_k_weights", "torch", "v1", "view", "w1", "w2", "where", "with", "zeros_like"], "dbrx/modeling_dbrx.py:DbrxRouter.__init__": ["False", "Linear", "ModelRouter", "Module", "__init__", "bias", "class", "config", "def", "ffn_hidden_size", "hidden_size", "layer", "moe_jitter_eps", "moe_num_experts", "nn", "self", "super"], "dbrx/modeling_dbrx.py:DbrxRouter.forward": ["LongTensor", "ModelRouter", "Module", "None", "Tensor", "and", "class", "def", "empty_like", "forward", "hidden_states", "if", "is", "layer", "moe_jitter_eps", "nn", "not", "return", "router_logits", "self", "shape", "torch", "training", "tuple", "uniform_", "view"], "dbrx/modeling_dbrx.py:DbrxFFN.__init__": ["ModelExperts", "ModelFFN", "ModelRouter", "Module", "__init__", "class", "config", "def", "experts", "ffn_config", "kwargs", "moe_normalize_expert_weights", "moe_top_k", "nn", "router", "self", "super", "top_k"], "dbrx/modeling_dbrx.py:DbrxFFN.route_tokens_to_experts": ["ModelFFN", "Module", "None", "True", "class", "def", "dim", "dtype", "functional", "if", "is", "keepdim", "moe_normalize_expert_weights", "nn", "norm", "not", "p", "return", "route_tokens_to_experts", "router_indices", "router_logits", "router_top_value", "self", "softmax", "top_k", "topk", "torch"], "dbrx/modeling_dbrx.py:DbrxFFN.forward": ["ModelFFN", "Module", "Tensor", "class", "def", "experts", "forward", "hidden_states", "nn", "output", "return", "route_tokens_to_experts", "router", "router_logits", "self", "top_k_index", "top_k_weights", "torch", "tuple"], "dbrx/modeling_dbrx.py:DbrxNormAttentionNorm.__init__": ["False", "LayerNorm", "ModelAttention", "ModelConfig", "ModelNormAttentionNorm", "Module", "None", "__init__", "attn", "bias", "class", "config", "d_model", "def", "int", "layer_idx", "nn", "norm_1", "norm_2", "resid_pdrop", "self", "super"], "dbrx/modeling_dbrx.py:DbrxNormAttentionNorm.forward": ["Any", "Cache", "LongTensor", "ModelNormAttentionNorm", "Module", "None", "Tensor", "_", "attention_mask", "attn", "cache_position", "class", "def", "dropout", "dtype", "forward", "functional", "hidden_states", "kwargs", "nn", "norm_1", "norm_2", "p", "past_key_values", "position_embeddings", "resid_pdrop", "residual_states", "return", "self", "to", "torch", "training", "tuple"], "dbrx/modeling_dbrx.py:DbrxBlock.__init__": ["GradientCheckpointingLayer", "ModelBlock", "ModelConfig", "ModelFFN", "ModelNormAttentionNorm", "__init__", "class", "config", "d_model", "def", "ffn", "hidden_size", "int", "layer_idx", "norm_attn_norm", "resid_pdrop", "self", "super"], "dbrx/modeling_dbrx.py:DbrxBlock.forward": ["Any", "Cache", "GradientCheckpointingLayer", "LongTensor", "ModelBlock", "None", "Tensor", "attention_mask", "cache_position", "class", "def", "dropout", "ffn", "forward", "functional", "hidden_states", "kwargs", "nn", "norm_attn_norm", "p", "past_key_values", "position_embeddings", "resid_pdrop", "resid_states", "return", "self", "torch", "training"], "dbrx/modeling_dbrx.py:DbrxPreTrainedModel._init_weights": ["ModelExpertGLU", "ModelPreTrainedModel", "Module", "PreTrainedModel", "_init_weights", "class", "config", "def", "if", "init", "initializer_range", "isinstance", "mean", "module", "nn", "normal_", "self", "std", "super", "v1", "w1", "w2"], "dbrx/modeling_dbrx.py:DbrxModel.__init__": ["Embedding", "False", "LayerNorm", "ModelBlock", "ModelConfig", "ModelModel", "ModelPreTrainedModel", "ModelRotaryEmbedding", "ModuleList", "__init__", "bias", "blocks", "class", "config", "d_model", "def", "emb_pdrop", "for", "gradient_checkpointing", "in", "layer_idx", "n_layers", "nn", "norm_f", "pad_token_id", "padding_idx", "post_init", "range", "rotary_emb", "self", "super", "vocab_size", "wte"], "dbrx/modeling_dbrx.py:DbrxModel.get_input_embeddings": ["Embedding", "ModelModel", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "nn", "return", "self", "wte"], "dbrx/modeling_dbrx.py:DbrxModel.set_input_embeddings": ["Embedding", "ModelModel", "ModelPreTrainedModel", "class", "def", "nn", "self", "set_input_embeddings", "value", "wte"], "dbrx/modeling_dbrx.py:DbrxModel.forward": ["Cache", "DynamicCache", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "MoeModelOutputWithPast", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "and", "arange", "attention_mask", "blocks", "bool", "cache_position", "causal_mask", "class", "config", "create_causal_mask", "decoder_layer", "def", "device", "else", "exactly", "for", "forward", "get_seq_length", "hidden_states", "if", "in", "input_embeds", "input_ids", "inputs_embeds", "is", "kwargs", "last_hidden_state", "must", "norm_f", "not", "num_hidden_layers", "of", "one", "or", "past_key_values", "past_seen_tokens", "position_embeddings", "position_ids", "raise", "return", "rotary_emb", "self", "shape", "specify", "torch", "unsqueeze", "use_cache", "wte"], "dbrx/modeling_dbrx.py:load_balancing_loss_func": ["Model_balancing_loss_func", "None", "Tensor", "_", "attention_mask", "batch_size", "cat", "compute_device", "concatenated_gate_logits", "def", "device", "dim", "else", "expand", "expert_attention_mask", "expert_mask", "float", "for", "functional", "gate_logits", "if", "in", "int", "is", "isinstance", "layer_gate", "mean", "nn", "not", "num_experts", "num_hidden_layers", "one_hot", "or", "overall_loss", "r", "reshape", "return", "router_per_expert_attention_mask", "router_prob_per_expert", "routing_weights", "selected_experts", "sequence_length", "shape", "softmax", "sum", "to", "tokens_per_expert", "top_k", "topk", "torch", "tuple", "unsqueeze"], "dbrx/modeling_dbrx.py:DbrxForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "ModelConfig", "ModelForCausalLM", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "ffn_config", "hidden_size", "lm_head", "moe_loss_weight", "moe_num_experts", "moe_top_k", "nn", "num_experts", "num_experts_per_tok", "post_init", "router_aux_loss_coef", "self", "super", "transformer", "vocab_size"], "dbrx/modeling_dbrx.py:DbrxForCausalLM.get_input_embeddings": ["Embedding", "GenerationMixin", "ModelForCausalLM", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "nn", "return", "self", "transformer"], "dbrx/modeling_dbrx.py:DbrxForCausalLM.set_input_embeddings": ["Embedding", "GenerationMixin", "ModelForCausalLM", "ModelPreTrainedModel", "class", "def", "nn", "self", "set_input_embeddings", "transformer", "value"], "dbrx/modeling_dbrx.py:DbrxForCausalLM.get_output_embeddings": ["GenerationMixin", "Linear", "ModelForCausalLM", "ModelPreTrainedModel", "class", "def", "get_output_embeddings", "lm_head", "nn", "return", "self"], "dbrx/modeling_dbrx.py:DbrxForCausalLM.set_output_embeddings": ["GenerationMixin", "Linear", "ModelForCausalLM", "ModelPreTrainedModel", "class", "def", "lm_head", "new_embeddings", "nn", "self", "set_output_embeddings"], "dbrx/modeling_dbrx.py:DbrxForCausalLM.set_decoder": ["GenerationMixin", "ModelForCausalLM", "ModelModel", "ModelPreTrainedModel", "class", "decoder", "def", "self", "set_decoder", "transformer"], "dbrx/modeling_dbrx.py:DbrxForCausalLM.get_decoder": ["GenerationMixin", "ModelForCausalLM", "ModelModel", "ModelPreTrainedModel", "class", "def", "get_decoder", "return", "self", "transformer"], "dbrx/modeling_dbrx.py:DbrxForCausalLM.forward": ["Cache", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForCausalLM", "ModelPreTrainedModel", "MoeCausalLMOutputWithPast", "MoeModelOutputWithPast", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "aux_loss", "bool", "cache_position", "class", "config", "def", "device", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "load_balancing_loss_func", "logits", "logits_to_keep", "loss", "loss_function", "not", "num_experts", "num_experts_per_tok", "output_router_logits", "outputs", "past_key_values", "position_ids", "r", "return", "router_aux_loss_coef", "router_logits", "self", "slice", "slice_indices", "to", "torch", "transformer", "use_cache", "vocab_size"], "deberta/modeling_deberta.py:DebertaLayerNorm.__init__": ["ModelLayerNorm", "Module", "Parameter", "__init__", "bias", "class", "def", "eps", "nn", "ones", "self", "size", "super", "torch", "variance_epsilon", "weight", "zeros"], "deberta/modeling_deberta.py:DebertaLayerNorm.forward": ["ModelLayerNorm", "Module", "True", "bias", "class", "def", "dtype", "float", "forward", "hidden_states", "input_type", "keepdim", "mean", "nn", "pow", "return", "self", "sqrt", "to", "torch", "variance", "variance_epsilon", "weight", "y"], "deberta/modeling_deberta.py:DebertaSelfOutput.__init__": ["Dropout", "LayerNorm", "Linear", "ModelLayerNorm", "ModelSelfOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "hidden_dropout_prob", "hidden_size", "layer_norm_eps", "nn", "self", "super"], "deberta/modeling_deberta.py:DebertaSelfOutput.forward": ["LayerNorm", "ModelSelfOutput", "Module", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self"], "deberta/modeling_deberta.py:build_relative_position": ["Model_relative_position", "None", "arange", "def", "device", "dtype", "k_ids", "key_layer", "key_size", "long", "q_ids", "query_layer", "query_size", "rel_pos_ids", "repeat", "return", "size", "torch", "unsqueeze", "view"], "deberta/modeling_deberta.py:c2p_dynamic_expand": ["Model_dynamic_expand", "Model_pos", "def", "expand", "query_layer", "relative_pos", "return", "size"], "deberta/modeling_deberta.py:p2c_dynamic_expand": ["Model_dynamic_expand", "c2p_pos", "def", "expand", "key_layer", "query_layer", "return", "size"], "deberta/modeling_deberta.py:pos_dynamic_expand": ["Model_dynamic_expand", "Model_index", "def", "expand", "key_layer", "p2c_att", "return", "size"], "deberta/modeling_deberta.py:scaled_size_sqrt": ["Model_size_sqrt", "Tensor", "def", "dtype", "float", "int", "query_layer", "return", "scale_factor", "size", "sqrt", "tensor", "torch"], "deberta/modeling_deberta.py:build_rpos": ["Model_relative_position", "Model_rpos", "Tensor", "def", "else", "if", "key_layer", "query_layer", "relative_pos", "return", "size", "torch"], "deberta/modeling_deberta.py:compute_attention_span": ["Model_attention_span", "Tensor", "def", "int", "key_layer", "max", "max_relative_positions", "min", "query_layer", "return", "size", "tensor", "torch"], "deberta/modeling_deberta.py:uneven_size_corrected": ["Model_size_corrected", "Tensor", "def", "dim", "else", "gather", "if", "index", "key_layer", "p2c_att", "pos_dynamic_expand", "pos_index", "query_layer", "relative_pos", "return", "size", "torch", "unsqueeze"], "deberta/modeling_deberta.py:DisentangledSelfAttention.__init__": ["Dropout", "False", "Linear", "ModelSelfAttention", "Module", "None", "Parameter", "The", "ValueError", "__init__", "a", "all_head_size", "attention", "attention_head_size", "attention_probs_dropout_prob", "bias", "c2p", "class", "config", "def", "dropout", "dtype", "else", "f", "float", "getattr", "head_logits_proj", "head_weights_proj", "heads", "hidden", "hidden_dropout_prob", "hidden_size", "if", "in", "in_proj", "int", "is", "max_position_embeddings", "max_relative_positions", "multiple", "nn", "not", "num_attention_heads", "number", "of", "p2c", "pos_att_type", "pos_dropout", "pos_proj", "pos_q_proj", "q_bias", "raise", "relative_attention", "self", "size", "super", "talking_head", "the", "torch", "v_bias", "zeros"], "deberta/modeling_deberta.py:DisentangledSelfAttention.transpose_for_scores": ["ModelSelfAttention", "Module", "class", "def", "new_x_shape", "nn", "num_attention_heads", "permute", "return", "self", "size", "transpose_for_scores", "view", "x"], "deberta/modeling_deberta.py:DisentangledSelfAttention.forward": ["False", "ModelSelfAttention", "Model_att_bias", "Module", "None", "Tensor", "and", "attention_mask", "attention_probs", "attention_scores", "bool", "cat", "chunk", "class", "context_layer", "contiguous", "def", "dim", "dropout", "dtype", "else", "finfo", "for", "forward", "functional", "head_logits_proj", "head_weights_proj", "hidden_states", "i", "if", "in", "in_proj", "int", "is", "k", "key_layer", "len", "masked_fill", "matmul", "min", "new_context_layer_shape", "nn", "not", "num_attention_heads", "output_attentions", "permute", "pos_att_type", "pos_dropout", "q", "q_bias", "qkvw", "qp", "query_layer", "query_states", "range", "rel_att", "rel_embeddings", "relative_attention", "relative_pos", "return", "scale", "scale_factor", "scaled_size_sqrt", "self", "size", "softmax", "t", "to", "torch", "transpose", "transpose_for_scores", "tuple", "v", "v_bias", "value_layer", "view", "weight", "ws", "x"], "deberta/modeling_deberta.py:DisentangledSelfAttention.disentangled_att_bias": ["ModelSelfAttention", "Model_att_bias", "Module", "None", "Relative", "Tensor", "ValueError", "att_span", "be", "build_relative_position", "build_rpos", "c2p", "c2p_att", "c2p_dynamic_expand", "c2p_pos", "clamp", "class", "compute_attention_span", "def", "device", "dim", "dtype", "elif", "f", "gather", "ids", "if", "in", "index", "int", "is", "key_layer", "long", "matmul", "max_relative_positions", "must", "nn", "of", "or", "p2c", "p2c_att", "p2c_dynamic_expand", "p2c_pos", "pos_att_type", "pos_key_layer", "pos_proj", "pos_q_proj", "pos_query_layer", "position", "query_layer", "r_pos", "raise", "rel_embeddings", "relative_pos", "return", "scale_factor", "scaled_size_sqrt", "score", "self", "to", "torch", "transpose", "transpose_for_scores", "uneven_size_corrected", "unsqueeze"], "deberta/modeling_deberta.py:DebertaEmbeddings.__init__": ["Dropout", "Embedding", "False", "LayerNorm", "Linear", "ModelEmbeddings", "ModelLayerNorm", "Module", "None", "True", "__init__", "arange", "bias", "class", "config", "def", "dropout", "else", "embed_proj", "embedding_size", "expand", "getattr", "hidden_dropout_prob", "hidden_size", "if", "layer_norm_eps", "max_position_embeddings", "nn", "not", "pad_token_id", "padding_idx", "persistent", "position_biased_input", "position_embeddings", "position_ids", "register_buffer", "self", "super", "token_type_embeddings", "torch", "type_vocab_size", "vocab_size", "word_embeddings"], "deberta/modeling_deberta.py:DebertaEmbeddings.forward": ["LayerNorm", "ModelEmbeddings", "Module", "None", "class", "def", "device", "dim", "dropout", "dtype", "else", "embed_proj", "embeddings", "forward", "if", "input_ids", "input_shape", "inputs_embeds", "is", "long", "mask", "nn", "not", "position_biased_input", "position_embeddings", "position_ids", "return", "self", "seq_length", "size", "squeeze", "to", "token_type_embeddings", "token_type_ids", "torch", "unsqueeze", "word_embeddings", "zeros", "zeros_like"], "deberta/modeling_deberta.py:DebertaAttention.__init__": ["DisentangledSelfAttention", "ModelAttention", "ModelSelfOutput", "Module", "__init__", "class", "config", "def", "nn", "output", "self", "super"], "deberta/modeling_deberta.py:DebertaAttention.forward": ["False", "ModelAttention", "Module", "None", "Tensor", "att_matrix", "attention_mask", "attention_output", "bool", "class", "def", "else", "forward", "hidden_states", "if", "is", "nn", "output", "output_attentions", "query_states", "rel_embeddings", "relative_pos", "return", "self", "self_output", "torch", "tuple"], "deberta/modeling_deberta.py:DebertaIntermediate.__init__": ["ACT2FN", "Linear", "ModelIntermediate", "Module", "__init__", "class", "config", "def", "dense", "else", "hidden_act", "hidden_size", "if", "intermediate_act_fn", "intermediate_size", "isinstance", "nn", "self", "str", "super"], "deberta/modeling_deberta.py:DebertaIntermediate.forward": ["ModelIntermediate", "Module", "Tensor", "class", "def", "dense", "forward", "hidden_states", "intermediate_act_fn", "nn", "return", "self", "torch"], "deberta/modeling_deberta.py:DebertaOutput.__init__": ["Dropout", "LayerNorm", "Linear", "ModelLayerNorm", "ModelOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "hidden_dropout_prob", "hidden_size", "intermediate_size", "layer_norm_eps", "nn", "self", "super"], "deberta/modeling_deberta.py:DebertaOutput.forward": ["LayerNorm", "ModelOutput", "Module", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self"], "deberta/modeling_deberta.py:DebertaLayer.__init__": ["GradientCheckpointingLayer", "ModelAttention", "ModelIntermediate", "ModelLayer", "ModelOutput", "__init__", "attention", "class", "config", "def", "intermediate", "output", "self", "super"], "deberta/modeling_deberta.py:DebertaLayer.forward": ["False", "GradientCheckpointingLayer", "ModelLayer", "None", "Tensor", "att_matrix", "attention", "attention_mask", "attention_output", "bool", "class", "def", "else", "forward", "hidden_states", "if", "intermediate", "intermediate_output", "layer_output", "output", "output_attentions", "query_states", "rel_embeddings", "relative_pos", "return", "self", "torch", "tuple"], "deberta/modeling_deberta.py:DebertaEncoder.__init__": ["Embedding", "False", "ModelEncoder", "ModelLayer", "Module", "ModuleList", "_", "__init__", "class", "config", "def", "for", "getattr", "gradient_checkpointing", "hidden_size", "if", "in", "layer", "max_position_embeddings", "max_relative_positions", "nn", "num_hidden_layers", "range", "rel_embeddings", "relative_attention", "self", "super"], "deberta/modeling_deberta.py:DebertaEncoder.get_rel_embedding": ["ModelEncoder", "Module", "None", "class", "def", "else", "get_rel_embedding", "if", "nn", "rel_embeddings", "relative_attention", "return", "self", "weight"], "deberta/modeling_deberta.py:DebertaEncoder.get_attention_mask": ["ModelEncoder", "Module", "attention_mask", "class", "def", "dim", "elif", "extended_attention_mask", "get_attention_mask", "if", "nn", "return", "self", "squeeze", "unsqueeze"], "deberta/modeling_deberta.py:DebertaEncoder.get_rel_pos": ["ModelEncoder", "Module", "None", "and", "build_relative_position", "class", "def", "else", "get_rel_pos", "hidden_states", "if", "is", "nn", "not", "query_states", "relative_attention", "relative_pos", "return", "self"], "deberta/modeling_deberta.py:DebertaEncoder.forward": ["BaseModelOutput", "False", "ModelEncoder", "Module", "None", "Tensor", "True", "all_attentions", "all_hidden_states", "att_m", "attention_mask", "attentions", "bool", "class", "def", "else", "enumerate", "for", "forward", "get_attention_mask", "get_rel_embedding", "get_rel_pos", "hidden_states", "i", "if", "in", "is", "last_hidden_state", "layer", "layer_module", "next_kv", "nn", "not", "output_attentions", "output_hidden_states", "query_states", "rel_embeddings", "relative_pos", "return", "return_dict", "self", "torch", "tuple", "v"], "deberta/modeling_deberta.py:DebertaPreTrainedModel._init_weights": ["DisentangledSelfAttention", "LegacyModelLMPredictionHead", "ModelEmbeddings", "ModelLMPredictionHead", "ModelPreTrainedModel", "PreTrainedModel", "_init_weights", "arange", "bias", "class", "copy_", "def", "elif", "expand", "if", "init", "isinstance", "module", "position_ids", "q_bias", "self", "shape", "super", "torch", "v_bias", "zeros_"], "deberta/modeling_deberta.py:DebertaModel.__init__": ["ModelEmbeddings", "ModelEncoder", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "config", "def", "embeddings", "encoder", "post_init", "self", "super", "z_steps"], "deberta/modeling_deberta.py:DebertaModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "return", "self", "word_embeddings"], "deberta/modeling_deberta.py:DebertaModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "new_embeddings", "self", "set_input_embeddings", "word_embeddings"], "deberta/modeling_deberta.py:DebertaModel.forward": ["BaseModelOutput", "False", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "True", "ValueError", "You", "_", "and", "append", "at", "attention_mask", "attentions", "bool", "both", "cannot", "class", "config", "def", "device", "dtype", "either", "elif", "else", "embedding_output", "embeddings", "encoded_layers", "encoder", "encoder_outputs", "for", "forward", "get_attention_mask", "get_rel_embedding", "get_rel_pos", "have", "hidden_states", "if", "in", "input_ids", "input_shape", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layer", "layers", "long", "mask", "not", "ones", "or", "output_attentions", "output_hidden_states", "position_ids", "query_states", "raise", "range", "rel_embeddings", "rel_pos", "relative_pos", "return", "return_dict", "same", "self", "sequence_output", "size", "specify", "the", "time", "to", "token_type_ids", "torch", "tuple", "use_return_dict", "warn_if_padding_and_no_attention_mask", "z_steps", "zeros"], "deberta/modeling_deberta.py:LegacyDebertaPredictionHeadTransform.__init__": ["ACT2FN", "LayerNorm", "Linear", "ModelModelPredictionHeadTransform", "Module", "__init__", "class", "config", "def", "dense", "else", "embedding_size", "eps", "getattr", "hidden_act", "hidden_size", "if", "isinstance", "layer_norm_eps", "nn", "self", "str", "super", "transform_act_fn"], "deberta/modeling_deberta.py:LegacyDebertaPredictionHeadTransform.forward": ["LayerNorm", "ModelModelPredictionHeadTransform", "Module", "class", "def", "dense", "forward", "hidden_states", "nn", "return", "self", "transform_act_fn"], "deberta/modeling_deberta.py:LegacyDebertaLMPredictionHead.__init__": ["Linear", "ModelModelLMPredictionHead", "ModelModelPredictionHeadTransform", "Module", "Parameter", "True", "__init__", "bias", "class", "config", "decoder", "def", "embedding_size", "getattr", "hidden_size", "nn", "self", "super", "torch", "transform", "vocab_size", "zeros"], "deberta/modeling_deberta.py:LegacyDebertaLMPredictionHead.forward": ["ModelModelLMPredictionHead", "Module", "class", "decoder", "def", "forward", "hidden_states", "nn", "return", "self", "transform"], "deberta/modeling_deberta.py:LegacyDebertaOnlyMLMHead.__init__": ["ModelModelLMPredictionHead", "ModelModelOnlyMLMHead", "Module", "__init__", "class", "config", "def", "nn", "predictions", "self", "super"], "deberta/modeling_deberta.py:LegacyDebertaOnlyMLMHead.forward": ["ModelModelOnlyMLMHead", "Module", "Tensor", "class", "def", "forward", "nn", "prediction_scores", "predictions", "return", "self", "sequence_output", "torch"], "deberta/modeling_deberta.py:DebertaLMPredictionHead.__init__": ["ACT2FN", "LayerNorm", "Linear", "ModelLMPredictionHead", "Module", "Parameter", "True", "__init__", "bias", "class", "config", "def", "dense", "elementwise_affine", "else", "eps", "hidden_act", "hidden_size", "if", "isinstance", "layer_norm_eps", "nn", "self", "str", "super", "torch", "transform_act_fn", "vocab_size", "zeros"], "deberta/modeling_deberta.py:DebertaLMPredictionHead.forward": ["LayerNorm", "ModelLMPredictionHead", "Module", "bias", "class", "def", "dense", "forward", "hidden_states", "matmul", "nn", "return", "self", "t", "torch", "transform_act_fn", "weight", "word_embeddings"], "deberta/modeling_deberta.py:DebertaOnlyMLMHead.__init__": ["ModelLMPredictionHead", "ModelOnlyMLMHead", "Module", "__init__", "class", "config", "def", "lm_head", "nn", "self", "super"], "deberta/modeling_deberta.py:DebertaOnlyMLMHead.forward": ["ModelOnlyMLMHead", "Module", "class", "def", "forward", "lm_head", "nn", "prediction_scores", "return", "self", "sequence_output", "word_embeddings"], "deberta/modeling_deberta.py:DebertaForMaskedLM.__init__": ["LegacyModelOnlyMLMHead", "Model", "ModelForMaskedLM", "ModelModel", "ModelOnlyMLMHead", "ModelPreTrainedModel", "__init__", "_tied_weights_keys", "class", "cls", "config", "def", "else", "embeddings", "if", "legacy", "lm_head", "lm_predictions", "post_init", "self", "super", "weight", "word_embeddings"], "deberta/modeling_deberta.py:DebertaForMaskedLM.get_output_embeddings": ["ModelForMaskedLM", "ModelPreTrainedModel", "class", "cls", "decoder", "def", "dense", "else", "get_output_embeddings", "if", "legacy", "lm_head", "lm_predictions", "predictions", "return", "self"], "deberta/modeling_deberta.py:DebertaForMaskedLM.set_output_embeddings": ["ModelForMaskedLM", "ModelPreTrainedModel", "bias", "class", "cls", "decoder", "def", "dense", "else", "if", "legacy", "lm_head", "lm_predictions", "new_embeddings", "predictions", "self", "set_output_embeddings"], "deberta/modeling_deberta.py:DebertaForMaskedLM.forward": ["CrossEntropyLoss", "MaskedLMOutput", "Model", "ModelForMaskedLM", "ModelPreTrainedModel", "None", "Tensor", "attention_mask", "attentions", "bool", "class", "cls", "config", "def", "else", "embeddings", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "legacy", "lm_predictions", "logits", "loss", "loss_fct", "masked_lm_loss", "not", "output", "output_attentions", "output_hidden_states", "outputs", "position_ids", "prediction_scores", "r", "return", "return_dict", "self", "sequence_output", "token_type_ids", "torch", "tuple", "use_return_dict", "view", "vocab_size", "word_embeddings"], "deberta/modeling_deberta.py:ContextPooler.__init__": ["Dropout", "Linear", "ModelPooler", "Module", "__init__", "class", "config", "def", "dense", "dropout", "nn", "pooler_dropout", "pooler_hidden_size", "self", "super"], "deberta/modeling_deberta.py:ContextPooler.forward": ["ACT2FN", "ModelPooler", "Model_token", "Module", "class", "config", "def", "dense", "dropout", "forward", "hidden_states", "nn", "pooled_output", "pooler_hidden_act", "return", "self"], "deberta/modeling_deberta.py:ContextPooler.output_dim": ["ModelPooler", "Module", "class", "config", "def", "hidden_size", "nn", "output_dim", "return", "self"], "deberta/modeling_deberta.py:DebertaForSequenceClassification.__init__": ["ContextPooler", "Dropout", "Linear", "Model", "ModelForSequenceClassification", "ModelModel", "ModelPreTrainedModel", "None", "__init__", "class", "classifier", "cls_dropout", "config", "def", "drop_out", "dropout", "else", "getattr", "hidden_dropout_prob", "if", "is", "nn", "num_labels", "output_dim", "pooler", "post_init", "self", "super"], "deberta/modeling_deberta.py:DebertaForSequenceClassification.get_input_embeddings": ["Model", "ModelForSequenceClassification", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "return", "self"], "deberta/modeling_deberta.py:DebertaForSequenceClassification.set_input_embeddings": ["Model", "ModelForSequenceClassification", "ModelPreTrainedModel", "class", "def", "new_embeddings", "self", "set_input_embeddings"], "deberta/modeling_deberta.py:DebertaForSequenceClassification.forward": ["BCEWithLogitsLoss", "CrossEntropyLoss", "LogSoftmax", "MSELoss", "Model", "ModelForSequenceClassification", "ModelPreTrainedModel", "None", "SequenceClassifierOutput", "Tensor", "attention_mask", "attentions", "bool", "class", "classifier", "config", "def", "dim", "dropout", "dtype", "elif", "else", "encoder_layer", "expand", "float", "forward", "gather", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "label_index", "labeled_logits", "labels", "log_softmax", "logits", "long", "loss", "loss_fct", "loss_fn", "mean", "multi_label_classification", "nn", "nonzero", "not", "num_labels", "or", "output", "output_attentions", "output_hidden_states", "outputs", "pooled_output", "pooler", "position_ids", "problem_type", "r", "regression", "return", "return_dict", "self", "single_label_classification", "size", "squeeze", "sum", "tensor", "to", "token_type_ids", "torch", "tuple", "use_return_dict", "view"], "deberta/modeling_deberta.py:DebertaForTokenClassification.__init__": ["Dropout", "Linear", "Model", "ModelForTokenClassification", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "classifier", "config", "def", "dropout", "hidden_dropout_prob", "hidden_size", "nn", "num_labels", "post_init", "self", "super"], "deberta/modeling_deberta.py:DebertaForTokenClassification.forward": ["CrossEntropyLoss", "Model", "ModelForTokenClassification", "ModelPreTrainedModel", "None", "Tensor", "TokenClassifierOutput", "attention_mask", "attentions", "bool", "class", "classifier", "config", "def", "dropout", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "not", "num_labels", "output", "output_attentions", "output_hidden_states", "outputs", "position_ids", "r", "return", "return_dict", "self", "sequence_output", "token_type_ids", "torch", "tuple", "use_return_dict", "view"], "deberta/modeling_deberta.py:DebertaForQuestionAnswering.__init__": ["Linear", "Model", "ModelForQuestionAnswering", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "config", "def", "hidden_size", "nn", "num_labels", "post_init", "qa_outputs", "self", "super"], "deberta/modeling_deberta.py:DebertaForQuestionAnswering.forward": ["CrossEntropyLoss", "Model", "ModelForQuestionAnswering", "ModelPreTrainedModel", "None", "QuestionAnsweringModelOutput", "Tensor", "and", "attention_mask", "attentions", "bool", "clamp", "class", "config", "contiguous", "def", "dim", "else", "end_logits", "end_loss", "end_positions", "forward", "hidden_states", "if", "ignore_index", "ignored_index", "input_ids", "inputs_embeds", "is", "kwargs", "len", "logits", "loss", "loss_fct", "not", "output", "output_attentions", "output_hidden_states", "outputs", "position_ids", "qa_outputs", "return", "return_dict", "self", "sequence_output", "size", "split", "squeeze", "start_logits", "start_loss", "start_positions", "token_type_ids", "torch", "total_loss", "tuple", "use_return_dict"], "deberta_v2/modeling_deberta_v2.py:DebertaV2SelfOutput.__init__": ["Dropout", "LayerNorm", "Linear", "ModelSelfOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "hidden_dropout_prob", "hidden_size", "layer_norm_eps", "nn", "self", "super"], "deberta_v2/modeling_deberta_v2.py:DebertaV2SelfOutput.forward": ["LayerNorm", "ModelSelfOutput", "Module", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self"], "deberta_v2/modeling_deberta_v2.py:make_log_bucket_position": ["Model_log_bucket_position", "abs", "abs_pos", "bucket_pos", "bucket_size", "ceil", "def", "int", "log", "log_pos", "max_position", "mid", "relative_pos", "return", "sign", "tensor", "torch", "type_as", "where"], "deberta_v2/modeling_deberta_v2.py:build_relative_position": ["Model_relative_position", "None", "and", "arange", "bucket_size", "def", "device", "dtype", "if", "int", "k_ids", "key_layer", "key_size", "long", "make_log_bucket_position", "max_position", "q_ids", "query_layer", "query_size", "rel_pos_ids", "return", "size", "to", "torch", "unsqueeze"], "deberta_v2/modeling_deberta_v2.py:c2p_dynamic_expand": ["Model_dynamic_expand", "Model_pos", "def", "expand", "query_layer", "relative_pos", "return", "size"], "deberta_v2/modeling_deberta_v2.py:p2c_dynamic_expand": ["Model_dynamic_expand", "c2p_pos", "def", "expand", "key_layer", "query_layer", "return", "size"], "deberta_v2/modeling_deberta_v2.py:pos_dynamic_expand": ["Model_dynamic_expand", "Model_index", "def", "expand", "key_layer", "p2c_att", "return", "size"], "deberta_v2/modeling_deberta_v2.py:scaled_size_sqrt": ["Model_size_sqrt", "Tensor", "def", "dtype", "float", "int", "query_layer", "return", "scale_factor", "size", "sqrt", "tensor", "torch"], "deberta_v2/modeling_deberta_v2.py:build_rpos": ["Model_relative_position", "Model_rpos", "bucket_size", "def", "else", "if", "int", "key_layer", "max_position", "max_relative_positions", "position_buckets", "query_layer", "relative_pos", "return", "size"], "deberta_v2/modeling_deberta_v2.py:DisentangledSelfAttention.__init__": ["Dropout", "False", "Linear", "ModelSelfAttention", "Module", "None", "The", "True", "ValueError", "__init__", "_attention_head_size", "a", "all_head_size", "attention", "attention_head_size", "attention_probs_dropout_prob", "bias", "c2p", "class", "config", "def", "dropout", "else", "f", "getattr", "heads", "hidden", "hidden_dropout_prob", "hidden_size", "if", "in", "is", "key_proj", "max_position_embeddings", "max_relative_positions", "multiple", "nn", "not", "num_attention_heads", "number", "of", "p2c", "pos_att_type", "pos_dropout", "pos_ebd_size", "pos_key_proj", "pos_query_proj", "position_buckets", "query_proj", "raise", "relative_attention", "self", "share_att_key", "size", "super", "the", "value_proj"], "deberta_v2/modeling_deberta_v2.py:DisentangledSelfAttention.transpose_for_scores": ["ModelSelfAttention", "Module", "Tensor", "attention_heads", "class", "contiguous", "def", "new_x_shape", "nn", "permute", "return", "self", "size", "torch", "transpose_for_scores", "view", "x"], "deberta_v2/modeling_deberta_v2.py:DisentangledSelfAttention.forward": ["False", "ModelSelfAttention", "Model_attention_bias", "Module", "None", "attention_mask", "attention_probs", "attention_scores", "bmm", "bool", "c2p", "class", "context_layer", "contiguous", "def", "dim", "dropout", "dtype", "finfo", "forward", "functional", "hidden_states", "if", "in", "is", "key_layer", "key_proj", "masked_fill", "min", "new_context_layer_shape", "nn", "not", "num_attention_heads", "output_attentions", "p2c", "permute", "pos_att_type", "pos_dropout", "query_layer", "query_proj", "query_states", "rel_att", "rel_embeddings", "relative_attention", "relative_pos", "return", "scale", "scale_factor", "scaled_size_sqrt", "self", "size", "softmax", "to", "torch", "transpose", "transpose_for_scores", "value_layer", "value_proj", "view"], "deberta_v2/modeling_deberta_v2.py:DisentangledSelfAttention.disentangled_attention_bias": ["ModelSelfAttention", "Model_attention_bias", "Module", "None", "Relative", "ValueError", "att_span", "be", "bmm", "bucket_size", "build_relative_position", "build_rpos", "c2p", "c2p_att", "c2p_pos", "clamp", "class", "def", "device", "dim", "dtype", "elif", "else", "expand", "f", "gather", "ids", "if", "in", "index", "is", "key_layer", "key_proj", "long", "max_position", "max_relative_positions", "must", "nn", "num_attention_heads", "of", "or", "p2c", "p2c_att", "p2c_pos", "pos_att_type", "pos_ebd_size", "pos_key_layer", "pos_key_proj", "pos_query_layer", "pos_query_proj", "position", "position_buckets", "query_layer", "query_proj", "r_pos", "raise", "rel_embeddings", "relative_pos", "repeat", "return", "scale", "scale_factor", "scaled_size_sqrt", "score", "self", "share_att_key", "size", "squeeze", "to", "torch", "transpose", "transpose_for_scores", "unsqueeze"], "deberta_v2/modeling_deberta_v2.py:DebertaV2Attention.__init__": ["DisentangledSelfAttention", "ModelAttention", "ModelSelfOutput", "Module", "__init__", "class", "config", "def", "nn", "output", "self", "super"], "deberta_v2/modeling_deberta_v2.py:DebertaV2Attention.forward": ["False", "ModelAttention", "Module", "None", "Tensor", "att_matrix", "attention_mask", "attention_output", "bool", "class", "def", "else", "forward", "hidden_states", "if", "is", "nn", "output", "output_attentions", "query_states", "rel_embeddings", "relative_pos", "return", "self", "self_output", "torch", "tuple"], "deberta_v2/modeling_deberta_v2.py:DebertaV2Intermediate.__init__": ["ACT2FN", "Linear", "ModelIntermediate", "Module", "__init__", "class", "config", "def", "dense", "else", "hidden_act", "hidden_size", "if", "intermediate_act_fn", "intermediate_size", "isinstance", "nn", "self", "str", "super"], "deberta_v2/modeling_deberta_v2.py:DebertaV2Intermediate.forward": ["ModelIntermediate", "Module", "Tensor", "class", "def", "dense", "forward", "hidden_states", "intermediate_act_fn", "nn", "return", "self", "torch"], "deberta_v2/modeling_deberta_v2.py:DebertaV2Output.__init__": ["Dropout", "LayerNorm", "Linear", "ModelOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "hidden_dropout_prob", "hidden_size", "intermediate_size", "layer_norm_eps", "nn", "self", "super"], "deberta_v2/modeling_deberta_v2.py:DebertaV2Output.forward": ["LayerNorm", "ModelOutput", "Module", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self"], "deberta_v2/modeling_deberta_v2.py:DebertaV2Layer.__init__": ["GradientCheckpointingLayer", "ModelAttention", "ModelIntermediate", "ModelLayer", "ModelOutput", "__init__", "attention", "class", "config", "def", "intermediate", "output", "self", "super"], "deberta_v2/modeling_deberta_v2.py:DebertaV2Layer.forward": ["False", "GradientCheckpointingLayer", "ModelLayer", "None", "Tensor", "att_matrix", "attention", "attention_mask", "attention_output", "bool", "class", "def", "else", "forward", "hidden_states", "if", "intermediate", "intermediate_output", "layer_output", "output", "output_attentions", "query_states", "rel_embeddings", "relative_pos", "return", "self", "torch", "tuple"], "deberta_v2/modeling_deberta_v2.py:ConvLayer.__init__": ["Dropout", "LayerNorm", "Model", "Model1d", "ModelLayer", "Model_act", "Model_groups", "Model_kernel_size", "Module", "__init__", "class", "config", "def", "dropout", "getattr", "groups", "hidden_dropout_prob", "hidden_size", "kernel_size", "layer_norm_eps", "nn", "padding", "self", "super", "tanh"], "deberta_v2/modeling_deberta_v2.py:ConvLayer.forward": ["ACT2FN", "LayerNorm", "Model", "ModelLayer", "Model_act", "Module", "None", "bool", "class", "contiguous", "def", "dim", "dropout", "dtype", "else", "expand", "forward", "hidden_states", "if", "input_mask", "is", "layer_norm_input", "masked_fill_", "nn", "out", "output", "output_states", "permute", "residual_states", "return", "rmask", "self", "size", "squeeze", "to", "unsqueeze"], "deberta_v2/modeling_deberta_v2.py:DebertaV2Embeddings.__init__": ["Dropout", "Embedding", "False", "LayerNorm", "Linear", "ModelEmbeddings", "Module", "None", "True", "__init__", "arange", "bias", "class", "config", "def", "dropout", "else", "embed_proj", "embedding_size", "expand", "getattr", "hidden_dropout_prob", "hidden_size", "if", "layer_norm_eps", "max_position_embeddings", "nn", "not", "pad_token_id", "padding_idx", "persistent", "position_biased_input", "position_embeddings", "position_ids", "register_buffer", "self", "super", "token_type_embeddings", "torch", "type_vocab_size", "vocab_size", "word_embeddings"], "deberta_v2/modeling_deberta_v2.py:DebertaV2Embeddings.forward": ["LayerNorm", "ModelEmbeddings", "Module", "None", "class", "def", "device", "dim", "dropout", "dtype", "else", "embed_proj", "embeddings", "forward", "if", "input_ids", "input_shape", "inputs_embeds", "is", "long", "mask", "nn", "not", "position_biased_input", "position_embeddings", "position_ids", "return", "self", "seq_length", "size", "squeeze", "to", "token_type_embeddings", "token_type_ids", "torch", "unsqueeze", "word_embeddings", "zeros", "zeros_like"], "deberta_v2/modeling_deberta_v2.py:DebertaV2Encoder.__init__": ["ConvLayer", "Embedding", "False", "LayerNorm", "ModelEncoder", "ModelLayer", "Module", "ModuleList", "None", "True", "_", "__init__", "class", "config", "conv", "conv_kernel_size", "def", "elementwise_affine", "else", "for", "getattr", "gradient_checkpointing", "hidden_size", "if", "in", "layer", "layer_norm", "layer_norm_eps", "lower", "max_position_embeddings", "max_relative_positions", "nn", "none", "norm_rel_ebd", "num_hidden_layers", "pos_ebd_size", "position_buckets", "range", "rel_embeddings", "relative_attention", "self", "split", "strip", "super", "x"], "deberta_v2/modeling_deberta_v2.py:DebertaV2Encoder.get_rel_embedding": ["LayerNorm", "ModelEncoder", "Module", "None", "and", "class", "def", "else", "get_rel_embedding", "if", "in", "is", "layer_norm", "nn", "norm_rel_ebd", "not", "rel_embeddings", "relative_attention", "return", "self", "weight"], "deberta_v2/modeling_deberta_v2.py:DebertaV2Encoder.get_attention_mask": ["ModelEncoder", "Module", "attention_mask", "class", "def", "dim", "elif", "extended_attention_mask", "get_attention_mask", "if", "nn", "return", "self", "squeeze", "unsqueeze"], "deberta_v2/modeling_deberta_v2.py:DebertaV2Encoder.get_rel_pos": ["ModelEncoder", "Module", "None", "and", "bucket_size", "build_relative_position", "class", "def", "else", "get_rel_pos", "hidden_states", "if", "is", "max_position", "max_relative_positions", "nn", "not", "position_buckets", "query_states", "relative_attention", "relative_pos", "return", "self"], "deberta_v2/modeling_deberta_v2.py:DebertaV2Encoder.forward": ["BaseModelOutput", "False", "ModelEncoder", "Module", "None", "Sequence", "Tensor", "True", "all_attentions", "all_hidden_states", "and", "attention_mask", "attentions", "attn_weights", "class", "conv", "def", "dim", "else", "enumerate", "for", "forward", "get_attention_mask", "get_rel_embedding", "get_rel_pos", "hidden_states", "i", "if", "in", "input_mask", "is", "isinstance", "last_hidden_state", "layer", "layer_module", "len", "next_kv", "nn", "not", "output_attentions", "output_hidden_states", "output_states", "query_states", "rel_embeddings", "relative_pos", "return", "return_dict", "self", "sum", "torch", "tuple", "v"], "deberta_v2/modeling_deberta_v2.py:DebertaV2PreTrainedModel._init_weights": ["LegacyModelLMPredictionHead", "ModelEmbeddings", "ModelLMPredictionHead", "ModelPreTrainedModel", "PreTrainedModel", "_init_weights", "arange", "bias", "class", "copy_", "def", "elif", "expand", "if", "init", "isinstance", "module", "position_ids", "self", "shape", "super", "torch", "zeros_"], "deberta_v2/modeling_deberta_v2.py:DebertaV2Model.__init__": ["ModelEmbeddings", "ModelEncoder", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "config", "def", "embeddings", "encoder", "post_init", "self", "super", "z_steps"], "deberta_v2/modeling_deberta_v2.py:DebertaV2Model.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "return", "self", "word_embeddings"], "deberta_v2/modeling_deberta_v2.py:DebertaV2Model.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "new_embeddings", "self", "set_input_embeddings", "word_embeddings"], "deberta_v2/modeling_deberta_v2.py:DebertaV2Model.forward": ["BaseModelOutput", "False", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "True", "ValueError", "You", "_", "and", "append", "at", "attention_mask", "attentions", "bool", "both", "cannot", "class", "config", "def", "device", "dtype", "either", "elif", "else", "embedding_output", "embeddings", "encoded_layers", "encoder", "encoder_outputs", "for", "forward", "get_attention_mask", "get_rel_embedding", "get_rel_pos", "have", "hidden_states", "if", "in", "input_ids", "input_shape", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layer", "layers", "long", "mask", "not", "ones", "or", "output_attentions", "output_hidden_states", "position_ids", "query_states", "raise", "range", "rel_embeddings", "rel_pos", "relative_pos", "return", "return_dict", "same", "self", "sequence_output", "size", "specify", "the", "time", "to", "token_type_ids", "torch", "tuple", "use_return_dict", "warn_if_padding_and_no_attention_mask", "z_steps", "zeros"], "deberta_v2/modeling_deberta_v2.py:LegacyDebertaV2PredictionHeadTransform.__init__": ["ACT2FN", "LayerNorm", "Linear", "ModelModelPredictionHeadTransform", "Module", "__init__", "class", "config", "def", "dense", "else", "embedding_size", "eps", "getattr", "hidden_act", "hidden_size", "if", "isinstance", "layer_norm_eps", "nn", "self", "str", "super", "transform_act_fn"], "deberta_v2/modeling_deberta_v2.py:LegacyDebertaV2PredictionHeadTransform.forward": ["LayerNorm", "ModelModelPredictionHeadTransform", "Module", "class", "def", "dense", "forward", "hidden_states", "nn", "return", "self", "transform_act_fn"], "deberta_v2/modeling_deberta_v2.py:LegacyDebertaV2LMPredictionHead.__init__": ["Linear", "ModelModelLMPredictionHead", "ModelModelPredictionHeadTransform", "Module", "Parameter", "__init__", "bias", "class", "config", "decoder", "def", "embedding_size", "getattr", "hidden_size", "nn", "self", "super", "torch", "transform", "vocab_size", "zeros"], "deberta_v2/modeling_deberta_v2.py:LegacyDebertaV2LMPredictionHead.forward": ["ModelModelLMPredictionHead", "Module", "class", "decoder", "def", "forward", "hidden_states", "nn", "return", "self", "transform"], "deberta_v2/modeling_deberta_v2.py:LegacyDebertaV2OnlyMLMHead.__init__": ["ModelModelLMPredictionHead", "ModelModelOnlyMLMHead", "Module", "__init__", "class", "config", "def", "nn", "predictions", "self", "super"], "deberta_v2/modeling_deberta_v2.py:LegacyDebertaV2OnlyMLMHead.forward": ["ModelModelOnlyMLMHead", "Module", "class", "def", "forward", "nn", "prediction_scores", "predictions", "return", "self", "sequence_output"], "deberta_v2/modeling_deberta_v2.py:DebertaV2LMPredictionHead.__init__": ["ACT2FN", "LayerNorm", "Linear", "ModelLMPredictionHead", "Module", "Parameter", "True", "__init__", "bias", "class", "config", "def", "dense", "elementwise_affine", "else", "eps", "hidden_act", "hidden_size", "if", "isinstance", "layer_norm_eps", "nn", "self", "str", "super", "torch", "transform_act_fn", "vocab_size", "zeros"], "deberta_v2/modeling_deberta_v2.py:DebertaV2LMPredictionHead.forward": ["LayerNorm", "ModelLMPredictionHead", "Module", "bias", "class", "def", "dense", "forward", "hidden_states", "matmul", "nn", "return", "self", "t", "torch", "transform_act_fn", "weight", "word_embeddings"], "deberta_v2/modeling_deberta_v2.py:DebertaV2OnlyMLMHead.__init__": ["ModelLMPredictionHead", "ModelOnlyMLMHead", "Module", "__init__", "class", "config", "def", "lm_head", "nn", "self", "super"], "deberta_v2/modeling_deberta_v2.py:DebertaV2OnlyMLMHead.forward": ["ModelOnlyMLMHead", "Module", "class", "def", "forward", "lm_head", "nn", "prediction_scores", "return", "self", "sequence_output", "word_embeddings"], "deberta_v2/modeling_deberta_v2.py:DebertaV2ForMaskedLM.__init__": ["LegacyModelOnlyMLMHead", "Model", "ModelForMaskedLM", "ModelModel", "ModelOnlyMLMHead", "ModelPreTrainedModel", "__init__", "_tied_weights_keys", "class", "cls", "config", "def", "else", "embeddings", "if", "legacy", "lm_head", "lm_predictions", "post_init", "self", "super", "weight", "word_embeddings"], "deberta_v2/modeling_deberta_v2.py:DebertaV2ForMaskedLM.get_output_embeddings": ["ModelForMaskedLM", "ModelPreTrainedModel", "class", "cls", "decoder", "def", "dense", "else", "get_output_embeddings", "if", "legacy", "lm_head", "lm_predictions", "predictions", "return", "self"], "deberta_v2/modeling_deberta_v2.py:DebertaV2ForMaskedLM.set_output_embeddings": ["ModelForMaskedLM", "ModelPreTrainedModel", "bias", "class", "cls", "decoder", "def", "dense", "else", "if", "legacy", "lm_head", "lm_predictions", "new_embeddings", "predictions", "self", "set_output_embeddings"], "deberta_v2/modeling_deberta_v2.py:DebertaV2ForMaskedLM.forward": ["CrossEntropyLoss", "MaskedLMOutput", "Model", "ModelForMaskedLM", "ModelPreTrainedModel", "None", "Tensor", "attention_mask", "attentions", "bool", "class", "cls", "config", "def", "else", "embeddings", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "legacy", "lm_predictions", "logits", "loss", "loss_fct", "masked_lm_loss", "not", "output", "output_attentions", "output_hidden_states", "outputs", "position_ids", "prediction_scores", "r", "return", "return_dict", "self", "sequence_output", "token_type_ids", "torch", "tuple", "use_return_dict", "view", "vocab_size", "word_embeddings"], "deberta_v2/modeling_deberta_v2.py:ContextPooler.__init__": ["Dropout", "Linear", "ModelPooler", "Module", "__init__", "class", "config", "def", "dense", "dropout", "nn", "pooler_dropout", "pooler_hidden_size", "self", "super"], "deberta_v2/modeling_deberta_v2.py:ContextPooler.forward": ["ACT2FN", "ModelPooler", "Model_token", "Module", "class", "config", "def", "dense", "dropout", "forward", "hidden_states", "nn", "pooled_output", "pooler_hidden_act", "return", "self"], "deberta_v2/modeling_deberta_v2.py:ContextPooler.output_dim": ["ModelPooler", "Module", "class", "config", "def", "hidden_size", "nn", "output_dim", "return", "self"], "deberta_v2/modeling_deberta_v2.py:DebertaV2ForSequenceClassification.__init__": ["ContextPooler", "Dropout", "Linear", "Model", "ModelForSequenceClassification", "ModelModel", "ModelPreTrainedModel", "None", "__init__", "class", "classifier", "cls_dropout", "config", "def", "drop_out", "dropout", "else", "getattr", "hidden_dropout_prob", "if", "is", "nn", "num_labels", "output_dim", "pooler", "post_init", "self", "super"], "deberta_v2/modeling_deberta_v2.py:DebertaV2ForSequenceClassification.get_input_embeddings": ["Model", "ModelForSequenceClassification", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "return", "self"], "deberta_v2/modeling_deberta_v2.py:DebertaV2ForSequenceClassification.set_input_embeddings": ["Model", "ModelForSequenceClassification", "ModelPreTrainedModel", "class", "def", "new_embeddings", "self", "set_input_embeddings"], "deberta_v2/modeling_deberta_v2.py:DebertaV2ForSequenceClassification.forward": ["BCEWithLogitsLoss", "CrossEntropyLoss", "LogSoftmax", "MSELoss", "Model", "ModelForSequenceClassification", "ModelPreTrainedModel", "None", "SequenceClassifierOutput", "Tensor", "attention_mask", "attentions", "bool", "class", "classifier", "config", "def", "dim", "dropout", "dtype", "elif", "else", "encoder_layer", "expand", "float", "forward", "gather", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "label_index", "labeled_logits", "labels", "log_softmax", "logits", "long", "loss", "loss_fct", "loss_fn", "mean", "multi_label_classification", "nn", "nonzero", "not", "num_labels", "or", "output", "output_attentions", "output_hidden_states", "outputs", "pooled_output", "pooler", "position_ids", "problem_type", "r", "regression", "return", "return_dict", "self", "single_label_classification", "size", "squeeze", "sum", "tensor", "to", "token_type_ids", "torch", "tuple", "use_return_dict", "view"], "deberta_v2/modeling_deberta_v2.py:DebertaV2ForTokenClassification.__init__": ["Dropout", "Linear", "Model", "ModelForTokenClassification", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "classifier", "config", "def", "dropout", "hidden_dropout_prob", "hidden_size", "nn", "num_labels", "post_init", "self", "super"], "deberta_v2/modeling_deberta_v2.py:DebertaV2ForTokenClassification.forward": ["CrossEntropyLoss", "Model", "ModelForTokenClassification", "ModelPreTrainedModel", "None", "Tensor", "TokenClassifierOutput", "attention_mask", "attentions", "bool", "class", "classifier", "config", "def", "dropout", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "not", "num_labels", "output", "output_attentions", "output_hidden_states", "outputs", "position_ids", "r", "return", "return_dict", "self", "sequence_output", "token_type_ids", "torch", "tuple", "use_return_dict", "view"], "deberta_v2/modeling_deberta_v2.py:DebertaV2ForQuestionAnswering.__init__": ["Linear", "Model", "ModelForQuestionAnswering", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "config", "def", "hidden_size", "nn", "num_labels", "post_init", "qa_outputs", "self", "super"], "deberta_v2/modeling_deberta_v2.py:DebertaV2ForQuestionAnswering.forward": ["CrossEntropyLoss", "Model", "ModelForQuestionAnswering", "ModelPreTrainedModel", "None", "QuestionAnsweringModelOutput", "Tensor", "and", "attention_mask", "attentions", "bool", "clamp", "class", "config", "contiguous", "def", "dim", "else", "end_logits", "end_loss", "end_positions", "forward", "hidden_states", "if", "ignore_index", "ignored_index", "input_ids", "inputs_embeds", "is", "kwargs", "len", "logits", "loss", "loss_fct", "not", "output", "output_attentions", "output_hidden_states", "outputs", "position_ids", "qa_outputs", "return", "return_dict", "self", "sequence_output", "size", "split", "squeeze", "start_logits", "start_loss", "start_positions", "token_type_ids", "torch", "total_loss", "tuple", "use_return_dict"], "deberta_v2/modeling_deberta_v2.py:DebertaV2ForMultipleChoice.__init__": ["ContextPooler", "Dropout", "Linear", "Model", "ModelForMultipleChoice", "ModelModel", "ModelPreTrainedModel", "None", "__init__", "class", "classifier", "cls_dropout", "config", "def", "drop_out", "dropout", "else", "getattr", "hidden_dropout_prob", "if", "is", "nn", "num_labels", "output_dim", "pooler", "post_init", "self", "super"], "deberta_v2/modeling_deberta_v2.py:DebertaV2ForMultipleChoice.get_input_embeddings": ["Model", "ModelForMultipleChoice", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "return", "self"], "deberta_v2/modeling_deberta_v2.py:DebertaV2ForMultipleChoice.set_input_embeddings": ["Model", "ModelForMultipleChoice", "ModelPreTrainedModel", "class", "def", "new_embeddings", "self", "set_input_embeddings"], "deberta_v2/modeling_deberta_v2.py:DebertaV2ForMultipleChoice.forward": ["CrossEntropyLoss", "Model", "ModelForMultipleChoice", "ModelPreTrainedModel", "MultipleChoiceModelOutput", "None", "Tensor", "attention_mask", "attentions", "bool", "class", "classifier", "config", "def", "dropout", "else", "encoder_layer", "flat_attention_mask", "flat_input_ids", "flat_inputs_embeds", "flat_position_ids", "flat_token_type_ids", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "not", "num_choices", "output", "output_attentions", "output_hidden_states", "outputs", "pooled_output", "pooler", "position_ids", "r", "reshaped_logits", "return", "return_dict", "self", "shape", "size", "token_type_ids", "torch", "tuple", "use_return_dict", "view"], "decision_transformer/modeling_decision_transformer.py:eager_attention_forward": ["Model_attention_forward", "None", "attention_mask", "attn_dropout", "attn_output", "attn_weights", "bias", "causal_mask", "def", "device", "dim", "dtype", "finfo", "float", "full", "functional", "if", "is", "is_cross_attention", "key", "key_length", "kwargs", "layer_idx", "mask_value", "matmul", "min", "module", "nn", "not", "query", "query_length", "return", "scale_attn_by_inverse_layer_idx", "scale_attn_weights", "shape", "size", "softmax", "to", "torch", "transpose", "type", "value", "where"], "decision_transformer/modeling_decision_transformer.py:DecisionTransformerGPT2Attention.__init__": ["Conv1D", "Dropout", "False", "ModelGPT2Attention", "Module", "None", "ValueError", "__init__", "and", "attn_dropout", "attn_pdrop", "be", "bias", "bool", "by", "c_attn", "c_proj", "class", "config", "def", "divisible", "dtype", "else", "embed_dim", "f", "got", "head_dim", "hidden_size", "if", "is_causal", "is_cross_attention", "layer_idx", "max_position_embeddings", "max_positions", "must", "nn", "not", "num_attention_heads", "num_heads", "ones", "persistent", "q_attn", "raise", "register_buffer", "reorder_and_upcast_attn", "resid_dropout", "resid_pdrop", "scale_attn_by_inverse_layer_idx", "scale_attn_weights", "self", "split_size", "super", "torch", "tril", "view"], "decision_transformer/modeling_decision_transformer.py:DecisionTransformerGPT2Attention._upcast_and_reordered_attn": ["Error", "False", "ModelGPT2Attention", "Module", "None", "RuntimeError", "_", "_upcast_and_reordered_attn", "alpha", "attention_mask", "attn_dropout", "attn_output", "attn_weights", "baddbmm", "beta", "bias", "bsz", "causal_mask", "class", "def", "device", "dim", "dk", "does", "dtype", "empty", "enabled", "finfo", "float", "float32", "functional", "have", "if", "is", "is_cross_attention", "k", "k_seq_len", "key", "key_length", "layer_idx", "mask_value", "matmul", "maybe_autocast", "min", "nn", "not", "num_heads", "q", "q_seq_len", "query", "query_length", "raise", "reshape", "return", "scale_attn_by_inverse_layer_idx", "scale_attn_weights", "scale_factor", "self", "size", "softmax", "tensor", "torch", "transpose", "type", "upcasting", "value", "where", "with"], "decision_transformer/modeling_decision_transformer.py:DecisionTransformerGPT2Attention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "EncoderDecoderCache", "False", "FloatTensor", "If", "LongTensor", "ModelGPT2Attention", "Module", "None", "Please", "Tensor", "True", "ValueError", "_attn_implementation", "_upcast_and_reordered_attn", "and", "as", "attention", "attention_interface", "attention_mask", "attn_dropout", "attn_output", "attn_weights", "be", "bool", "c_attn", "c_proj", "cache_position", "class", "config", "contiguous", "cross", "cross_attention_cache", "curr_past_key_values", "def", "defined", "dim", "dropout", "eager", "eager_attention_forward", "else", "encoder_attention_mask", "encoder_hidden_states", "forward", "get", "hasattr", "have", "head_dim", "hidden_states", "if", "instantiate", "is", "is_cross_attention", "is_updated", "isinstance", "key_states", "keys", "kwargs", "layer_idx", "layers", "make", "nn", "not", "or", "output_attentions", "p", "past_key_values", "q_attn", "query_states", "raise", "reorder_and_upcast_attn", "reshape", "resid_dropout", "return", "self", "self_attention_cache", "shape", "shape_kv", "shape_q", "split", "split_size", "sure", "the", "to", "torch", "training", "transpose", "tuple", "update", "used", "using_eager", "value_states", "values", "view", "weights", "with"], "decision_transformer/modeling_decision_transformer.py:DecisionTransformerGPT2MLP.__init__": ["ACT2FN", "Conv1D", "Dropout", "ModelGPT2MLP", "Module", "__init__", "act", "activation_function", "c_fc", "c_proj", "class", "config", "def", "dropout", "embed_dim", "hidden_size", "intermediate_size", "nn", "resid_pdrop", "self", "super"], "decision_transformer/modeling_decision_transformer.py:DecisionTransformerGPT2MLP.forward": ["FloatTensor", "ModelGPT2MLP", "Module", "None", "act", "c_fc", "c_proj", "class", "def", "dropout", "forward", "hidden_states", "nn", "return", "self", "torch", "tuple"], "decision_transformer/modeling_decision_transformer.py:DecisionTransformerGPT2Block.__init__": ["GradientCheckpointingLayer", "LayerNorm", "ModelGPT2Attention", "ModelGPT2Block", "ModelGPT2MLP", "None", "True", "__init__", "add_cross_attention", "attn", "class", "config", "crossattention", "def", "else", "eps", "hidden_size", "if", "inner_dim", "is", "is_cross_attention", "layer_idx", "layer_norm_epsilon", "ln_1", "ln_2", "ln_cross_attn", "mlp", "n_inner", "nn", "not", "self", "super"], "decision_transformer/modeling_decision_transformer.py:DecisionTransformerGPT2Block.forward": ["Cache", "False", "FloatTensor", "GradientCheckpointingLayer", "If", "LongTensor", "ModelGPT2Block", "None", "Tensor", "True", "ValueError", "add_cross_attention", "are", "attention", "attention_mask", "attn", "attn_output", "be", "bool", "by", "cache_position", "class", "config", "cross", "cross_attn_output", "cross_attn_weights", "crossattention", "def", "encoder_attention_mask", "encoder_hidden_states", "f", "feed_forward_hidden_states", "forward", "has", "hasattr", "hidden_states", "if", "instantiated", "is", "kwargs", "layers", "ln_1", "ln_2", "ln_cross_attn", "mlp", "not", "output_attentions", "outputs", "passed", "past_key_values", "raise", "residual", "return", "self", "self_attn_weights", "setting", "to", "torch", "tuple", "use_cache", "with"], "decision_transformer/modeling_decision_transformer.py:DecisionTransformerGPT2PreTrainedModel._init_weights": ["ModelGPT2Attention", "ModelGPT2PreTrainedModel", "PreTrainedModel", "_init_weights", "and", "bias", "bool", "c_proj", "class", "config", "copy_", "def", "dtype", "elif", "for", "if", "in", "init", "initializer_range", "isinstance", "math", "max_position_embeddings", "max_positions", "mean", "module", "n_layer", "name", "named_parameters", "normal_", "ones", "p", "self", "sqrt", "std", "super", "torch", "tril", "view", "weight"], "decision_transformer/modeling_decision_transformer.py:DecisionTransformerGPT2Model.__init__": ["Dropout", "Embedding", "False", "LayerNorm", "ModelGPT2Block", "ModelGPT2Model", "ModelGPT2PreTrainedModel", "ModuleList", "__init__", "class", "config", "def", "drop", "embd_pdrop", "embed_dim", "eps", "for", "gradient_checkpointing", "h", "hidden_size", "i", "in", "layer_idx", "layer_norm_epsilon", "ln_f", "max_position_embeddings", "nn", "num_hidden_layers", "post_init", "range", "self", "super", "vocab_size", "wpe", "wte"], "decision_transformer/modeling_decision_transformer.py:DecisionTransformerGPT2Model.get_input_embeddings": ["ModelGPT2Model", "ModelGPT2PreTrainedModel", "class", "def", "get_input_embeddings", "return", "self", "wte"], "decision_transformer/modeling_decision_transformer.py:DecisionTransformerGPT2Model.set_input_embeddings": ["ModelGPT2Model", "ModelGPT2PreTrainedModel", "class", "def", "new_embeddings", "self", "set_input_embeddings", "wte"], "decision_transformer/modeling_decision_transformer.py:DecisionTransformerGPT2Model.forward": ["BaseModelOutputWithPastAndCrossAttentions", "Cache", "DynamicCache", "EncoderDecoderCache", "False", "FloatTensor", "LongTensor", "ModelGPT2Model", "ModelGPT2PreTrainedModel", "None", "Setting", "Tensor", "True", "ValueError", "You", "_", "add_cross_attention", "all_cross_attentions", "all_hidden_states", "all_self_attentions", "and", "arange", "at", "attention_mask", "attentions", "batch_size", "be", "block", "bool", "both", "cache_position", "cannot", "checkpointing", "class", "config", "cross_attentions", "def", "defined", "device", "drop", "dtype", "either", "elif", "else", "encoder_attention_mask", "encoder_batch_size", "encoder_hidden_shape", "encoder_hidden_states", "encoder_sequence_length", "finfo", "for", "forward", "get_seq_length", "gradient", "gradient_checkpointing", "h", "has", "have", "hidden_states", "if", "in", "incompatible", "input_ids", "input_shape", "inputs_embeds", "invert_attention_mask", "is", "isinstance", "kwargs", "last_hidden_state", "ln_f", "logger", "min", "not", "ones", "or", "output_attentions", "output_hidden_states", "output_shape", "outputs", "past_key_values", "past_seen_tokens", "position_embeds", "position_ids", "raise", "return", "return_dict", "same", "self", "shape", "size", "specify", "the", "time", "to", "token_type_embeds", "token_type_ids", "torch", "training", "tuple", "unsqueeze", "use_cache", "use_return_dict", "v", "view", "warn_if_padding_and_no_attention_mask", "warning_once", "with", "wpe", "wte"], "decision_transformer/modeling_decision_transformer.py:DecisionTransformerModel.__init__": ["Embedding", "LayerNorm", "Linear", "ModelGPT2Model", "ModelModel", "ModelPreTrainedModel", "Sequential", "Tanh", "__init__", "act_dim", "action_tanh", "class", "config", "def", "else", "embed_action", "embed_ln", "embed_return", "embed_state", "embed_timestep", "encoder", "hidden_size", "if", "max_ep_len", "nn", "post_init", "predict_action", "predict_return", "predict_state", "self", "state_dim", "super", "torch"], "decision_transformer/modeling_decision_transformer.py:DecisionTransformerModel.forward": ["FloatTensor", "LongTensor", "ModelModel", "ModelOutput", "ModelPreTrainedModel", "None", "action_embeddings", "action_preds", "actions", "attention_mask", "attentions", "batch_size", "bool", "class", "config", "def", "device", "dim", "dtype", "else", "embed_action", "embed_ln", "embed_return", "embed_state", "embed_timestep", "encoder", "encoder_outputs", "forward", "hidden_size", "hidden_states", "if", "inputs_embeds", "is", "kwargs", "last_hidden_state", "long", "not", "ones", "output_attentions", "output_hidden_states", "permute", "position_ids", "predict_action", "predict_return", "predict_state", "r", "reshape", "return", "return_dict", "return_preds", "returns_embeddings", "returns_to_go", "rewards", "self", "seq_length", "shape", "stack", "stacked_attention_mask", "stacked_inputs", "state_embeddings", "state_preds", "states", "time_embeddings", "timesteps", "torch", "tuple", "use_return_dict", "x", "zeros"], "deepseek_v2/modeling_deepseek_v2.py:DeepseekV2Experts.__init__": ["ACT2FN", "ModelExperts", "Module", "Parameter", "__init__", "act_fn", "class", "config", "def", "down_proj", "empty", "gate_up_proj", "hidden_act", "hidden_dim", "hidden_size", "intermediate_dim", "moe_intermediate_size", "n_routed_experts", "nn", "num_experts", "self", "super", "torch"], "deepseek_v2/modeling_deepseek_v2.py:DeepseekV2Experts.forward": ["ModelExperts", "Module", "None", "Tensor", "act_fn", "chunk", "class", "continue", "current_hidden_states", "current_state", "def", "dim", "down_proj", "dtype", "expert_hit", "expert_idx", "expert_mask", "final_hidden_states", "for", "forward", "functional", "gate", "gate_up_proj", "greater", "hidden_states", "if", "in", "index_add_", "linear", "nn", "no_grad", "nonzero", "num_classes", "num_experts", "one_hot", "permute", "return", "self", "sum", "to", "token_idx", "top_k_index", "top_k_pos", "top_k_weights", "torch", "up", "where", "with", "zeros_like"], "deepseek_v2/modeling_deepseek_v2.py:DeepseekV2Moe.__init__": ["False", "Linear", "ModelConfig", "ModelExperts", "ModelMLP", "ModelMoe", "Module", "None", "__init__", "bias", "class", "config", "def", "experts", "gate", "hidden_size", "if", "intermediate_size", "is", "moe_intermediate_size", "n_group", "n_routed_experts", "n_shared_experts", "nn", "not", "num_experts_per_tok", "num_group", "routed_scaling_factor", "self", "shared_experts", "super", "top_k", "topk_group", "topk_method"], "deepseek_v2/modeling_deepseek_v2.py:DeepseekV2Moe.route_tokens_to_experts": ["False", "ModelMoe", "Module", "batch_size", "bool", "class", "def", "dim", "dtype", "elif", "expand", "float32", "greedy", "group_idx", "group_limited_greedy", "group_mask", "group_scores", "hidden_dim", "if", "k", "masked_fill", "max", "nn", "num_experts", "num_group", "reshape", "return", "route_tokens_to_experts", "routed_scaling_factor", "router_logits", "scatter_", "score_mask", "self", "seq_len", "shape", "softmax", "sorted", "tmp_scores", "top_k", "topk", "topk_group", "topk_idx", "topk_method", "topk_weight", "torch", "unsqueeze", "values", "view", "zeros_like"], "deepseek_v2/modeling_deepseek_v2.py:DeepseekV2Moe.forward": ["ModelMoe", "Module", "Tensor", "class", "def", "experts", "float32", "forward", "functional", "gate", "hidden_states", "linear", "nn", "orig_shape", "residuals", "return", "route_tokens_to_experts", "router_logits", "self", "shape", "shared_experts", "topk_indices", "topk_weights", "torch", "type", "view", "weight"], "deepseek_v2/modeling_deepseek_v2.py:DeepseekV2MLP.__init__": ["ACT2FN", "Linear", "ModelConfig", "ModelMLP", "Module", "None", "__init__", "act_fn", "bias", "class", "config", "def", "down_proj", "else", "gate_proj", "hidden_act", "hidden_size", "if", "intermediate_size", "is", "mlp_bias", "nn", "self", "super", "up_proj"], "deepseek_v2/modeling_deepseek_v2.py:DeepseekV2MLP.forward": ["ModelMLP", "Module", "act_fn", "class", "def", "down_proj", "forward", "gate_proj", "nn", "return", "self", "up_proj", "x"], "deepseek_v2/modeling_deepseek_v2.py:DeepseekV2RMSNorm.__init__": ["ModelRMSNorm", "Module", "Parameter", "__init__", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "deepseek_v2/modeling_deepseek_v2.py:DeepseekV2RMSNorm.forward": ["ModelRMSNorm", "Module", "True", "class", "def", "dtype", "float32", "forward", "hidden_states", "input_dtype", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "deepseek_v2/modeling_deepseek_v2.py:DeepseekV2RMSNorm.extra_repr": ["ModelRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "variance_epsilon", "weight"], "deepseek_v2/modeling_deepseek_v2.py:DeepseekV2RotaryEmbedding.__init__": ["Callable", "False", "ModelConfig", "ModelRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "deepseek_v2/modeling_deepseek_v2.py:DeepseekV2RotaryEmbedding.compute_default_rope_parameters": ["ModelConfig", "ModelRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "deepseek_v2/modeling_deepseek_v2.py:DeepseekV2RotaryEmbedding.forward": ["False", "ModelRotaryEmbedding", "Module", "None", "and", "attention_scaling", "class", "cpu", "def", "device", "device_type", "else", "enabled", "expand", "float", "forward", "freqs", "freqs_cis", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "nn", "ones_like", "polar", "position_ids", "position_ids_expanded", "return", "self", "shape", "str", "to", "torch", "transpose", "type", "with", "x"], "deepseek_v2/modeling_deepseek_v2.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "deepseek_v2/modeling_deepseek_v2.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value", "value_states"], "deepseek_v2/modeling_deepseek_v2.py:apply_rotary_emb": ["Model_rotary_emb", "Tensor", "def", "device", "flatten", "float", "freqs_cis", "reshape", "return", "shape", "to", "torch", "tuple", "type_as", "unsqueeze", "view_as_complex", "view_as_real", "xk", "xk_", "xk_out", "xq", "xq_", "xq_out"], "deepseek_v2/modeling_deepseek_v2.py:DeepseekV2Attention.__init__": ["False", "Linear", "ModelAttention", "ModelConfig", "ModelRMSNorm", "Module", "None", "True", "__init__", "attention_bias", "attention_dropout", "bias", "class", "config", "def", "else", "head_dim", "hidden_size", "if", "int", "is", "is_causal", "kv_a_layernorm", "kv_a_proj_with_mqa", "kv_b_proj", "kv_lora_rank", "layer_idx", "max_position_embeddings", "nn", "num_attention_heads", "num_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "q_a_layernorm", "q_a_proj", "q_b_proj", "q_lora_rank", "q_proj", "qk_head_dim", "qk_nope_head_dim", "qk_rope_head_dim", "scaling", "self", "super", "v_head_dim"], "deepseek_v2/modeling_deepseek_v2.py:DeepseekV2Attention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "F", "LongTensor", "ModelAttention", "Module", "None", "Tensor", "_attn_implementation", "and", "apply_rotary_emb", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "batch_size", "cache_kwargs", "cache_position", "cat", "class", "compressed_kv", "config", "contiguous", "def", "device", "dim", "dropout", "eager", "eager_attention_forward", "else", "expand", "flash_attention_2", "forward", "hidden_states", "if", "is", "k_nope", "k_pe", "key_shape", "key_states", "kv_a_layernorm", "kv_a_proj_with_mqa", "kv_b_proj", "kv_lora_rank", "kwargs", "layer_idx", "nn", "not", "o_proj", "pad", "past_key_values", "position_embeddings", "q", "q_a_layernorm", "q_a_proj", "q_b_proj", "q_lora_rank", "q_nope", "q_pe", "q_proj", "qk_head_dim", "qk_nope_head_dim", "qk_rope_head_dim", "query_shape", "query_states", "reshape", "return", "scaling", "self", "seq_length", "shape", "split", "to", "torch", "training", "transpose", "tuple", "update", "v_head_dim", "value_states", "view"], "deepseek_v2/modeling_deepseek_v2.py:DeepseekV2DecoderLayer.__init__": ["GradientCheckpointingLayer", "ModelAttention", "ModelConfig", "ModelDecoderLayer", "ModelMLP", "ModelMoe", "ModelRMSNorm", "__init__", "class", "config", "def", "else", "eps", "first_k_dense_replace", "hidden_size", "if", "input_layernorm", "int", "layer_idx", "mlp", "post_attention_layernorm", "rms_norm_eps", "self", "self_attn", "super"], "deepseek_v2/modeling_deepseek_v2.py:DeepseekV2DecoderLayer.forward": ["Cache", "False", "GradientCheckpointingLayer", "LongTensor", "ModelDecoderLayer", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "attention_mask", "bool", "cache_position", "class", "def", "forward", "hidden_states", "input_layernorm", "kwargs", "mlp", "past_key_values", "position_embeddings", "position_ids", "post_attention_layernorm", "residual", "return", "self", "self_attn", "torch", "tuple", "use_cache"], "deepseek_v2/modeling_deepseek_v2.py:DeepseekV2PreTrainedModel._init_weights": ["ModelExperts", "ModelPreTrainedModel", "PreTrainedModel", "_init_weights", "class", "config", "def", "down_proj", "gate_up_proj", "if", "init", "initializer_range", "isinstance", "mean", "module", "normal_", "self", "std", "super"], "deepseek_v2/modeling_deepseek_v2.py:DeepseekV2Model.__init__": ["Embedding", "False", "ModelConfig", "ModelDecoderLayer", "ModelModel", "ModelPreTrainedModel", "ModelRMSNorm", "ModelRotaryEmbedding", "ModuleList", "__init__", "class", "config", "def", "embed_tokens", "eps", "for", "gradient_checkpointing", "hidden_size", "in", "layer_idx", "layers", "nn", "norm", "num_hidden_layers", "pad_token_id", "padding_idx", "post_init", "range", "rms_norm_eps", "rotary_emb", "self", "super", "vocab_size"], "deepseek_v2/modeling_deepseek_v2.py:DeepseekV2Model.forward": ["BaseModelOutputWithPast", "Cache", "DynamicCache", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "and", "arange", "attention_mask", "bool", "cache_position", "causal_mask", "class", "config", "create_causal_mask", "decoder_layer", "def", "device", "else", "embed_tokens", "exactly", "for", "forward", "get_seq_length", "hidden_states", "if", "in", "input_embeds", "input_ids", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layers", "must", "norm", "not", "num_hidden_layers", "of", "one", "or", "past_key_values", "past_seen_tokens", "position_embeddings", "position_ids", "raise", "return", "rotary_emb", "self", "shape", "specify", "torch", "unsqueeze", "use_cache"], "deepseek_v2/modeling_deepseek_v2.py:DeepseekV2ForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "ModelForCausalLM", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "vocab_size"], "deepseek_v2/modeling_deepseek_v2.py:DeepseekV2ForCausalLM.forward": ["BaseModelOutputWithPast", "Cache", "CausalLMOutputWithPast", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForCausalLM", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "outputs", "past_key_values", "position_ids", "r", "return", "self", "slice", "slice_indices", "torch", "use_cache", "vocab_size"], "deepseek_v3/modeling_deepseek_v3.py:DeepseekV3RMSNorm.__init__": ["ModelRMSNorm", "Module", "Parameter", "__init__", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "deepseek_v3/modeling_deepseek_v3.py:DeepseekV3RMSNorm.forward": ["ModelRMSNorm", "Module", "True", "class", "def", "dtype", "float32", "forward", "hidden_states", "input_dtype", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "deepseek_v3/modeling_deepseek_v3.py:DeepseekV3RMSNorm.extra_repr": ["ModelRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "variance_epsilon", "weight"], "deepseek_v3/modeling_deepseek_v3.py:DeepseekV3RotaryEmbedding.__init__": ["Callable", "False", "ModelConfig", "ModelRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "deepseek_v3/modeling_deepseek_v3.py:DeepseekV3RotaryEmbedding.compute_default_rope_parameters": ["ModelConfig", "ModelRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "deepseek_v3/modeling_deepseek_v3.py:DeepseekV3RotaryEmbedding.forward": ["False", "ModelRotaryEmbedding", "Module", "None", "and", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "deepseek_v3/modeling_deepseek_v3.py:DeepseekV3MLP.__init__": ["ACT2FN", "False", "Linear", "ModelMLP", "Module", "None", "__init__", "act_fn", "bias", "class", "config", "def", "down_proj", "else", "gate_proj", "hidden_act", "hidden_size", "if", "intermediate_size", "is", "nn", "self", "super", "up_proj"], "deepseek_v3/modeling_deepseek_v3.py:DeepseekV3MLP.forward": ["ModelMLP", "Module", "act_fn", "class", "def", "down_proj", "forward", "gate_proj", "nn", "return", "self", "up_proj", "x"], "deepseek_v3/modeling_deepseek_v3.py:DeepseekV3TopkRouter.__init__": ["ModelTopkRouter", "Module", "Parameter", "__init__", "class", "config", "def", "e_score_correction_bias", "empty", "hidden_size", "n_routed_experts", "nn", "register_buffer", "self", "super", "torch", "weight", "zeros"], "deepseek_v3/modeling_deepseek_v3.py:DeepseekV3TopkRouter.forward": ["F", "ModelTopkRouter", "Module", "class", "config", "def", "float32", "forward", "hidden_size", "hidden_states", "linear", "nn", "return", "router_logits", "self", "torch", "type", "view", "weight"], "deepseek_v3/modeling_deepseek_v3.py:DeepseekV3NaiveMoe.__init__": ["ACT2FN", "ModelNaiveMoe", "Module", "Parameter", "__init__", "act_fn", "class", "config", "def", "down_proj", "empty", "gate_up_proj", "hidden_act", "hidden_dim", "hidden_size", "intermediate_dim", "moe_intermediate_size", "nn", "num_experts", "num_local_experts", "self", "super", "torch"], "deepseek_v3/modeling_deepseek_v3.py:DeepseekV3NaiveMoe.forward": ["ModelNaiveMoe", "Module", "None", "Tensor", "act_fn", "chunk", "class", "continue", "current_hidden_states", "current_state", "def", "dim", "down_proj", "dtype", "expert_hit", "expert_idx", "expert_mask", "final_hidden_states", "for", "forward", "functional", "gate", "gate_up_proj", "greater", "hidden_states", "if", "in", "index_add_", "linear", "nn", "no_grad", "nonzero", "num_classes", "num_experts", "one_hot", "permute", "return", "self", "sum", "to", "token_idx", "top_k_index", "top_k_pos", "top_k_weights", "torch", "up", "where", "with", "zeros_like"], "deepseek_v3/modeling_deepseek_v3.py:DeepseekV3MoE.__init__": ["ModelMLP", "ModelMoE", "ModelNaiveMoe", "ModelTopkRouter", "Module", "__init__", "class", "config", "def", "experts", "gate", "intermediate_size", "moe_intermediate_size", "n_group", "n_routed_experts", "n_shared_experts", "nn", "norm_topk_prob", "num_experts_per_tok", "routed_scaling_factor", "self", "shared_experts", "super", "top_k", "topk_group"], "deepseek_v3/modeling_deepseek_v3.py:DeepseekV3MoE.route_tokens_to_experts": ["False", "ModelMoE", "Module", "True", "bool", "class", "def", "denominator", "dim", "e_score_correction_bias", "expand", "gate", "gather", "group_idx", "group_mask", "group_scores", "if", "k", "keepdim", "masked_fill", "n_group", "n_routed_experts", "nn", "norm_topk_prob", "reshape", "return", "route_tokens_to_experts", "routed_scaling_factor", "router_logits", "router_logits_for_choice", "scatter_", "score_mask", "scores_for_choice", "self", "sigmoid", "sorted", "sum", "top_k", "topk", "topk_group", "topk_indices", "topk_weights", "torch", "unsqueeze", "view", "zeros_like"], "deepseek_v3/modeling_deepseek_v3.py:DeepseekV3MoE.forward": ["ModelMoE", "Module", "class", "def", "experts", "forward", "gate", "hidden_states", "nn", "orig_shape", "residuals", "return", "route_tokens_to_experts", "router_logits", "self", "shape", "shared_experts", "topk_indices", "topk_weights", "view"], "deepseek_v3/modeling_deepseek_v3.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "deepseek_v3/modeling_deepseek_v3.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cos", "def", "k", "k_embed", "q", "q_embed", "return", "rotate_half", "sin", "unsqueeze", "unsqueeze_dim"], "deepseek_v3/modeling_deepseek_v3.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "deepseek_v3/modeling_deepseek_v3.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value", "value_states"], "deepseek_v3/modeling_deepseek_v3.py:apply_rotary_pos_emb_interleave": ["Model_rotary_pos_emb_interleave", "None", "b", "cos", "d", "def", "h", "k", "k_embed", "position_ids", "q", "q_embed", "r", "reshape", "return", "rotate_half", "s", "shape", "sin", "transpose", "unsqueeze", "unsqueeze_dim", "view"], "deepseek_v3/modeling_deepseek_v3.py:yarn_get_mscale": ["Model_get_mscale", "def", "if", "log", "math", "mscale", "return", "scale"], "deepseek_v3/modeling_deepseek_v3.py:DeepseekV3Attention.__init__": ["False", "Linear", "ModelAttention", "ModelConfig", "ModelRMSNorm", "Module", "None", "True", "__init__", "attention_bias", "attention_dropout", "bias", "class", "config", "def", "default", "else", "factor", "get", "hidden_size", "if", "int", "is", "is_causal", "kv_a_layernorm", "kv_a_proj_with_mqa", "kv_b_proj", "kv_lora_rank", "layer_idx", "mscale", "mscale_all_dim", "nn", "num_attention_heads", "num_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "q_a_layernorm", "q_a_proj", "q_b_proj", "q_lora_rank", "q_proj", "qk_head_dim", "qk_nope_head_dim", "qk_rope_head_dim", "rope_parameters", "rope_type", "scaling", "scaling_factor", "self", "super", "v_head_dim", "yarn_get_mscale"], "deepseek_v3/modeling_deepseek_v3.py:DeepseekV3Attention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "F", "FlashAttentionKwargs", "LongTensor", "ModelAttention", "Module", "None", "Tensor", "Unpack", "_attn_implementation", "and", "apply_rotary_pos_emb", "apply_rotary_pos_emb_interleave", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "batch_size", "cache_kwargs", "cache_position", "cat", "class", "compressed_kv", "config", "contiguous", "cos", "def", "dim", "dropout", "eager", "eager_attention_forward", "else", "expand", "flash_attention_2", "forward", "hidden_states", "if", "is", "k_pass", "k_rot", "key_shape", "key_states", "kv_a_layernorm", "kv_a_proj_with_mqa", "kv_b_proj", "kv_lora_rank", "kwargs", "layer_idx", "nn", "not", "o_proj", "pad", "past_key_values", "position_embeddings", "q_a_layernorm", "q_a_proj", "q_b_proj", "q_lora_rank", "q_pass", "q_proj", "q_rot", "q_states", "qk_head_dim", "qk_nope_head_dim", "qk_rope_head_dim", "query_shape", "query_states", "reshape", "return", "rope_interleave", "scaling", "self", "seq_length", "shape", "sin", "split", "torch", "training", "transpose", "tuple", "update", "v_head_dim", "value_states", "view"], "deepseek_v3/modeling_deepseek_v3.py:DeepseekV3DecoderLayer.__init__": ["GradientCheckpointingLayer", "ModelAttention", "ModelConfig", "ModelDecoderLayer", "ModelMLP", "ModelMoE", "ModelRMSNorm", "__init__", "class", "config", "def", "else", "eps", "first_k_dense_replace", "hidden_size", "if", "input_layernorm", "int", "layer_idx", "mlp", "post_attention_layernorm", "rms_norm_eps", "self", "self_attn", "super"], "deepseek_v3/modeling_deepseek_v3.py:DeepseekV3DecoderLayer.forward": ["Cache", "False", "GradientCheckpointingLayer", "LongTensor", "ModelDecoderLayer", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "attention_mask", "bool", "cache_position", "class", "def", "forward", "hidden_states", "input_layernorm", "kwargs", "mlp", "past_key_values", "position_embeddings", "position_ids", "post_attention_layernorm", "residual", "return", "self", "self_attn", "torch", "tuple", "use_cache"], "deepseek_v3/modeling_deepseek_v3.py:DeepseekV3PreTrainedModel._init_weights": ["ModelNaiveMoe", "ModelPreTrainedModel", "ModelTopkRouter", "PreTrainedModel", "_init_weights", "class", "config", "def", "down_proj", "e_score_correction_bias", "elif", "gate_up_proj", "if", "init", "initializer_range", "isinstance", "mean", "module", "normal_", "self", "std", "super", "weight", "zeros_"], "deepseek_v3/modeling_deepseek_v3.py:DeepseekV3Model.__init__": ["Embedding", "False", "ModelConfig", "ModelDecoderLayer", "ModelModel", "ModelPreTrainedModel", "ModelRMSNorm", "ModelRotaryEmbedding", "ModuleList", "__init__", "class", "config", "def", "embed_tokens", "eps", "for", "gradient_checkpointing", "hidden_size", "in", "layer_idx", "layers", "nn", "norm", "num_hidden_layers", "pad_token_id", "padding_idx", "post_init", "range", "rms_norm_eps", "rotary_emb", "self", "super", "vocab_size"], "deepseek_v3/modeling_deepseek_v3.py:DeepseekV3Model.forward": ["BaseModelOutputWithPast", "Cache", "DynamicCache", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "and", "arange", "attention_mask", "bool", "cache_position", "causal_mask", "class", "config", "create_causal_mask", "decoder_layer", "def", "device", "else", "embed_tokens", "exactly", "for", "forward", "get_seq_length", "hidden_states", "if", "in", "input_embeds", "input_ids", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layers", "must", "norm", "not", "num_hidden_layers", "of", "one", "or", "past_key_values", "past_seen_tokens", "position_embeddings", "position_ids", "raise", "return", "rotary_emb", "self", "shape", "specify", "torch", "unsqueeze", "use_cache"], "deepseek_v3/modeling_deepseek_v3.py:DeepseekV3ForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "ModelForCausalLM", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "vocab_size"], "deepseek_v3/modeling_deepseek_v3.py:DeepseekV3ForCausalLM.forward": ["BaseModelOutputWithPast", "Cache", "CausalLMOutputWithPast", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForCausalLM", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "outputs", "past_key_values", "position_ids", "r", "return", "self", "slice", "slice_indices", "torch", "use_cache", "vocab_size"], "deepseek_vl/modeling_deepseek_vl.py:DeepseekVLAligner.__init__": ["GELU", "Linear", "ModelAligner", "Module", "__init__", "activation", "class", "config", "def", "hidden_size", "in_features", "linear1", "linear2", "nn", "out_features", "self", "super", "text_config", "vision_config"], "deepseek_vl/modeling_deepseek_vl.py:DeepseekVLAligner.forward": ["ModelAligner", "Module", "Tensor", "activation", "class", "def", "forward", "linear1", "linear2", "nn", "return", "self", "torch", "vision_encodings", "x"], "deepseek_vl/modeling_deepseek_vl.py:DeepseekVLModel.__init__": ["AutoModel", "False", "ModelAligner", "ModelModel", "ModelPreTrainedModel", "__init__", "aligner", "class", "config", "def", "from_config", "gradient_checkpointing", "language_model", "post_init", "self", "super", "text_config", "vision_config", "vision_model"], "deepseek_vl/modeling_deepseek_vl.py:DeepseekVLModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "language_model", "return", "self"], "deepseek_vl/modeling_deepseek_vl.py:DeepseekVLModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "language_model", "self", "set_input_embeddings", "value"], "deepseek_vl/modeling_deepseek_vl.py:DeepseekVLModel.get_image_features": ["ModelModel", "ModelPreTrainedModel", "aligner", "class", "def", "get_image_features", "image_embeds", "last_hidden_state", "pixel_values", "return", "self", "vision_model"], "deepseek_vl/modeling_deepseek_vl.py:DeepseekVLModel.get_placeholder_mask": ["FloatTensor", "Image", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "ValueError", "all", "and", "class", "config", "def", "device", "do", "dtype", "else", "expand_as", "f", "features", "get_input_embeddings", "get_placeholder_mask", "if", "image", "image_features", "image_token_id", "input_ids", "inputs_embeds", "is", "long", "match", "n_image_features", "n_image_tokens", "not", "numel", "raise", "return", "self", "shape", "special_image_mask", "sum", "tensor", "to", "tokens", "torch", "unsqueeze"], "deepseek_vl/modeling_deepseek_vl.py:DeepseekVLModel.forward": ["Cache", "FloatTensor", "LongTensor", "ModelBaseModelOutputWithPast", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "ValueError", "You", "and", "at", "attention_mask", "attentions", "bool", "both", "cache_position", "cannot", "class", "def", "device", "dtype", "either", "else", "forward", "get_image_features", "get_input_embeddings", "get_placeholder_mask", "hidden_states", "if", "image_attention_mask", "image_embeds", "image_features", "image_hidden_states", "input_ids", "inputs_embeds", "int", "is", "kwargs", "language_model", "last_hidden_state", "lm_output", "logits_to_keep", "masked_scatter", "must", "not", "one", "past_key_values", "pixel_values", "position_ids", "raise", "reshape", "return", "same", "self", "shape", "specify", "the", "time", "to", "torch", "use_cache"], "deepseek_vl/modeling_deepseek_vl.py:DeepseekVLForConditionalGeneration.__init__": ["False", "GenerationMixin", "Linear", "ModelConfig", "ModelForConditionalGeneration", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "text_config", "vocab_size"], "deepseek_vl/modeling_deepseek_vl.py:DeepseekVLForConditionalGeneration.get_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "language_model", "model", "return", "self"], "deepseek_vl/modeling_deepseek_vl.py:DeepseekVLForConditionalGeneration.set_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "language_model", "model", "self", "set_input_embeddings", "value"], "deepseek_vl/modeling_deepseek_vl.py:DeepseekVLForConditionalGeneration.forward": ["Cache", "FloatTensor", "GenerationMixin", "LongTensor", "ModelCausalLMOutputWithPast", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "else", "forward", "hidden_states", "if", "image_hidden_states", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "outputs", "past_key_values", "pixel_values", "position_ids", "r", "return", "self", "slice", "slice_indices", "text_config", "torch", "use_cache", "vocab_size"], "deepseek_vl/modeling_deepseek_vl.py:DeepseekVLForConditionalGeneration.prepare_inputs_for_generation": ["False", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "True", "attention_mask", "cache_position", "class", "def", "get", "if", "input_ids", "inputs_embeds", "is_first_iteration", "kwargs", "logits_to_keep", "model_inputs", "not", "or", "past_key_values", "pixel_values", "prepare_inputs_for_generation", "return", "self", "super", "use_cache"], "deepseek_vl_hybrid/modeling_deepseek_vl_hybrid.py:DeepseekVLHybridLayerNorm.__init__": ["LayerNorm", "ModelLayerNorm", "NotImplementedError", "Unsupported", "__init__", "channels_first", "channels_last", "class", "data", "data_format", "def", "eps", "f", "format", "if", "in", "kwargs", "nn", "normalized_shape", "not", "raise", "self", "super"], "deepseek_vl_hybrid/modeling_deepseek_vl_hybrid.py:DeepseekVLHybridLayerNorm.forward": ["LayerNorm", "ModelLayerNorm", "Tensor", "channels_first", "class", "data_format", "def", "else", "features", "forward", "if", "nn", "permute", "return", "self", "super", "torch"], "deepseek_vl_hybrid/modeling_deepseek_vl_hybrid.py:DeepseekVLSamVisionNeck.__init__": ["Conv2d", "False", "ModelLayerNorm", "ModelVLSamVisionNeck", "Module", "__init__", "bias", "channels_first", "class", "config", "conv1", "conv2", "data_format", "def", "hidden_size", "kernel_size", "layer_norm1", "layer_norm2", "nn", "output_channels", "padding", "self", "super"], "deepseek_vl_hybrid/modeling_deepseek_vl_hybrid.py:DeepseekVLSamVisionNeck.forward": ["ModelVLSamVisionNeck", "Module", "class", "conv1", "conv2", "def", "forward", "hidden_states", "layer_norm1", "layer_norm2", "nn", "permute", "return", "self"], "deepseek_vl_hybrid/modeling_deepseek_vl_hybrid.py:DeepseekVLSamVisionProj.__init__": ["Conv2d", "False", "ModelVLSamVisionProj", "Module", "__init__", "bias", "class", "config", "conv1", "conv2", "def", "int", "kernel_size", "nn", "output_channels", "output_size", "padding", "self", "stride", "super"], "deepseek_vl_hybrid/modeling_deepseek_vl_hybrid.py:DeepseekVLSamVisionProj.forward": ["False", "ModelVLSamVisionProj", "Module", "Tensor", "align_corners", "bilinear", "class", "conv1", "conv2", "def", "features", "forward", "functional", "interpolate", "mode", "nn", "output_size", "return", "self", "size", "torch"], "deepseek_vl_hybrid/modeling_deepseek_vl_hybrid.py:DeepseekVLHybridAligner.__init__": ["GELU", "Linear", "ModelAligner", "ModelConfig", "Module", "__init__", "act", "class", "config", "def", "hidden_size", "high_res_in_channels", "high_res_vision_config", "high_res_vision_proj", "in_channels", "nn", "out_channels", "output_channels", "proj", "self", "super", "text_config", "vision_config", "vision_proj"], "deepseek_vl_hybrid/modeling_deepseek_vl_hybrid.py:DeepseekVLHybridAligner.forward": ["ModelAligner", "Module", "Tensor", "act", "class", "concat", "def", "dim", "encodings", "forward", "high_res_vision_encodings", "high_res_vision_proj", "nn", "proj", "return", "self", "torch", "vision_encodings", "vision_proj"], "deepseek_vl_hybrid/modeling_deepseek_vl_hybrid.py:DeepseekVLHybridPreTrainedModel._init_weights": ["Conv2d", "Linear", "ModelLayerNorm", "ModelModel", "ModelPreTrainedModel", "None", "PreTrainedModel", "_init_weights", "bias", "class", "config", "def", "elif", "fan_out", "high_res_vision_alpha", "if", "init", "initializer_range", "is", "isinstance", "kaiming_normal_", "mean", "mode", "module", "nn", "nonlinearity", "normal_", "not", "ones_", "relu", "self", "std", "text_config", "weight", "zeros_"], "deepseek_vl_hybrid/modeling_deepseek_vl_hybrid.py:DeepseekVLHybridModel.__init__": ["AutoModel", "False", "ModelAligner", "ModelModel", "ModelPreTrainedModel", "ModelVLSamVisionNeck", "ModelVLSamVisionProj", "Parameter", "__init__", "aligner", "class", "config", "def", "from_config", "global_attn_index", "global_attn_indexes", "gradient_checkpointing", "high_res_vision_alpha", "high_res_vision_config", "high_res_vision_model", "high_res_vision_neck", "high_res_vision_proj", "image_size", "language_model", "nn", "output_size", "patch_size", "post_init", "self", "super", "text_config", "torch", "vision_config", "vision_model", "zeros"], "deepseek_vl_hybrid/modeling_deepseek_vl_hybrid.py:DeepseekVLHybridModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "language_model", "return", "self"], "deepseek_vl_hybrid/modeling_deepseek_vl_hybrid.py:DeepseekVLHybridModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "language_model", "self", "set_input_embeddings", "value"], "deepseek_vl_hybrid/modeling_deepseek_vl_hybrid.py:DeepseekVLHybridModel.get_image_features": ["ModelModel", "ModelPreTrainedModel", "aligner", "class", "def", "get_high_res_image_features", "get_image_features", "get_low_res_image_features", "high_res_pixel_values", "high_res_vision_encodings", "images_embeds", "pixel_values", "return", "self", "vision_encodings"], "deepseek_vl_hybrid/modeling_deepseek_vl_hybrid.py:DeepseekVLHybridModel.get_placeholder_mask": ["FloatTensor", "Image", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "ValueError", "all", "and", "class", "config", "def", "device", "do", "dtype", "else", "expand_as", "f", "features", "get_input_embeddings", "get_placeholder_mask", "if", "image", "image_features", "image_token_id", "input_ids", "inputs_embeds", "is", "long", "match", "n_image_features", "n_image_tokens", "not", "numel", "raise", "return", "self", "shape", "special_image_mask", "sum", "tensor", "to", "tokens", "torch", "unsqueeze"], "deepseek_vl_hybrid/modeling_deepseek_vl_hybrid.py:DeepseekVLHybridModel.forward": ["Both", "Cache", "FloatTensor", "LongTensor", "ModelBaseModelOutputWithPast", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "ValueError", "You", "all", "and", "at", "attention_mask", "attentions", "be", "bool", "both", "cache_position", "cannot", "class", "config", "def", "device", "dtype", "either", "else", "expand_as", "forward", "get_image_features", "get_input_embeddings", "hidden_states", "high_res_pixel_values", "if", "image_attention_mask", "image_embeds", "image_features", "image_hidden_states", "image_token_id", "input_ids", "inputs_embeds", "int", "is", "kwargs", "language_model", "last_hidden_state", "lm_output", "logits_to_keep", "long", "masked_scatter", "must", "not", "one", "past_key_values", "pixel_values", "position_ids", "raise", "reshape", "return", "same", "self", "shape", "should", "specified", "specify", "tensor", "the", "time", "to", "torch", "unsqueeze", "use_cache"], "deepseek_vl_hybrid/modeling_deepseek_vl_hybrid.py:DeepseekVLHybridModel.get_low_res_image_features": ["ModelModel", "ModelPreTrainedModel", "class", "def", "get_low_res_image_features", "output", "pixel_values", "return", "self", "vision_model"], "deepseek_vl_hybrid/modeling_deepseek_vl_hybrid.py:DeepseekVLHybridModel.get_high_res_image_features": ["ModelModel", "ModelPreTrainedModel", "True", "class", "def", "get_high_res_image_features", "global_attn_index", "global_hidden_state", "hidden_states", "high_res_vision_alpha", "high_res_vision_model", "high_res_vision_neck", "high_res_vision_proj", "last_hidden_state", "output", "output_hidden_states", "permute", "pixel_values", "reshape", "return", "return_dict", "self", "shape"], "deepseek_vl_hybrid/modeling_deepseek_vl_hybrid.py:DeepseekVLHybridForConditionalGeneration.__init__": ["False", "GenerationMixin", "Linear", "ModelConfig", "ModelForConditionalGeneration", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "text_config", "vocab_size"], "deepseek_vl_hybrid/modeling_deepseek_vl_hybrid.py:DeepseekVLHybridForConditionalGeneration.get_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "language_model", "model", "return", "self"], "deepseek_vl_hybrid/modeling_deepseek_vl_hybrid.py:DeepseekVLHybridForConditionalGeneration.set_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "language_model", "model", "self", "set_input_embeddings", "value"], "deepseek_vl_hybrid/modeling_deepseek_vl_hybrid.py:DeepseekVLHybridForConditionalGeneration.forward": ["Cache", "FloatTensor", "GenerationMixin", "LongTensor", "ModelCausalLMOutputWithPast", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "else", "forward", "hidden_states", "high_res_pixel_values", "if", "image_hidden_states", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "outputs", "past_key_values", "pixel_values", "position_ids", "r", "return", "self", "slice", "slice_indices", "text_config", "torch", "use_cache", "vocab_size"], "deepseek_vl_hybrid/modeling_deepseek_vl_hybrid.py:DeepseekVLHybridForConditionalGeneration.prepare_inputs_for_generation": ["False", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "True", "attention_mask", "cache_position", "class", "def", "get", "high_res_pixel_values", "if", "input_ids", "inputs_embeds", "is_first_iteration", "kwargs", "logits_to_keep", "model_inputs", "not", "or", "past_key_values", "pixel_values", "prepare_inputs_for_generation", "return", "self", "super", "use_cache"], "deformable_detr/modeling_deformable_detr.py:MultiScaleDeformableAttention.forward": ["False", "ModelScaleDeformableAttention", "Module", "Tensor", "_", "align_corners", "append", "attention_weights", "batch_size", "bilinear", "class", "contiguous", "def", "dim", "enumerate", "flatten", "for", "forward", "functional", "grid_sample", "height", "hidden_dim", "im2col_step", "in", "int", "level_id", "level_start_index", "list", "mode", "nn", "num_heads", "num_levels", "num_points", "num_queries", "output", "padding_mode", "reshape", "return", "sampling_grid_l_", "sampling_grids", "sampling_locations", "sampling_value_l_", "sampling_value_list", "self", "shape", "split", "stack", "sum", "torch", "transpose", "tuple", "value", "value_l_", "value_list", "value_spatial_shapes", "value_spatial_shapes_list", "view", "width", "zeros"], "deformable_detr/modeling_deformable_detr.py:inverse_sigmoid": ["Model_sigmoid", "clamp", "def", "eps", "log", "max", "min", "return", "torch", "x", "x1", "x2"], "deformable_detr/modeling_deformable_detr.py:DeformableDetrFrozenBatchNorm2d.__init__": ["ModelFrozenBatchNorm2d", "Module", "__init__", "bias", "class", "def", "n", "nn", "ones", "register_buffer", "running_mean", "running_var", "self", "super", "torch", "weight", "zeros"], "deformable_detr/modeling_deformable_detr.py:DeformableDetrFrozenBatchNorm2d._load_from_state_dict": ["ModelFrozenBatchNorm2d", "Module", "_load_from_state_dict", "class", "def", "del", "error_msgs", "if", "in", "local_metadata", "missing_keys", "nn", "num_batches_tracked", "num_batches_tracked_key", "prefix", "self", "state_dict", "strict", "super", "unexpected_keys"], "deformable_detr/modeling_deformable_detr.py:DeformableDetrFrozenBatchNorm2d.forward": ["ModelFrozenBatchNorm2d", "Module", "bias", "class", "def", "epsilon", "forward", "nn", "reshape", "return", "rsqrt", "running_mean", "running_var", "scale", "self", "weight", "x"], "deformable_detr/modeling_deformable_detr.py:replace_batch_norm": ["BatchNorm2d", "ModelFrozenBatchNorm2d", "Model_batch_norm", "_modules", "bias", "children", "copy_", "def", "device", "for", "if", "in", "isinstance", "len", "list", "meta", "model", "module", "name", "named_children", "new_module", "nn", "num_features", "r", "running_mean", "running_var", "torch", "weight"], "deformable_detr/modeling_deformable_detr.py:DeformableDetrConvEncoder.__init__": ["Either", "False", "ModelConvEncoder", "Module", "None", "True", "ValueError", "__init__", "and", "backbone", "backbone_config", "backbone_kwargs", "backbone_model_type", "be", "channels", "class", "config", "copy", "create_model", "def", "dilation", "elif", "else", "feature_info", "features_only", "for", "get", "getattr", "if", "in", "in_chans", "intermediate_channel_sizes", "is", "kwargs", "layer2", "layer3", "layer4", "load_backbone", "model", "model_type", "name", "named_parameters", "nn", "no_grad", "not", "num_channels", "num_feature_levels", "or", "out_indices", "output_stride", "parameter", "pop", "pretrained", "provided", "raise", "replace_batch_norm", "requires_backends", "requires_grad_", "resnet", "self", "should", "stage", "super", "the", "timm", "torch", "use_pretrained_backbone", "use_timm_backbone", "with"], "deformable_detr/modeling_deformable_detr.py:DeformableDetrConvEncoder.forward": ["ModelConvEncoder", "Module", "None", "Tensor", "append", "bool", "class", "config", "def", "else", "feature_map", "feature_maps", "features", "float", "for", "forward", "functional", "if", "in", "interpolate", "mask", "model", "nn", "out", "pixel_mask", "pixel_values", "return", "self", "shape", "size", "to", "torch", "use_timm_backbone"], "deformable_detr/modeling_deformable_detr.py:DeformableDetrConvModel.__init__": ["ModelConvModel", "Module", "__init__", "class", "conv_encoder", "def", "nn", "position_embedding", "self", "super"], "deformable_detr/modeling_deformable_detr.py:DeformableDetrConvModel.forward": ["ModelConvModel", "Module", "append", "class", "conv_encoder", "def", "dtype", "feature_map", "for", "forward", "in", "mask", "nn", "out", "pixel_mask", "pixel_values", "pos", "position_embedding", "return", "self", "to"], "deformable_detr/modeling_deformable_detr.py:DeformableDetrSinePositionEmbedding.__init__": ["False", "ModelSinePositionEmbedding", "Module", "None", "True", "ValueError", "__init__", "and", "be", "class", "def", "embedding_dim", "if", "is", "math", "nn", "normalize", "not", "passed", "pi", "raise", "scale", "self", "should", "super", "temperature"], "deformable_detr/modeling_deformable_detr.py:DeformableDetrSinePositionEmbedding.forward": ["ModelSinePositionEmbedding", "Module", "No", "None", "ValueError", "arange", "cat", "class", "cos", "cumsum", "def", "device", "dim", "dim_t", "div", "dtype", "embedding_dim", "eps", "flatten", "floor", "forward", "if", "is", "mask", "nn", "normalize", "permute", "pixel", "pixel_mask", "pixel_values", "pos", "pos_x", "pos_y", "provided", "raise", "return", "rounding_mode", "scale", "self", "sin", "stack", "temperature", "torch", "x_embed", "y_embed"], "deformable_detr/modeling_deformable_detr.py:DeformableDetrLearnedPositionEmbedding.__init__": ["Embedding", "ModelLearnedPositionEmbedding", "Module", "__init__", "class", "column_embeddings", "def", "embedding_dim", "nn", "row_embeddings", "self", "super"], "deformable_detr/modeling_deformable_detr.py:DeformableDetrLearnedPositionEmbedding.forward": ["ModelLearnedPositionEmbedding", "Module", "None", "arange", "cat", "class", "column_embeddings", "def", "device", "dim", "forward", "height", "height_values", "nn", "permute", "pixel_mask", "pixel_values", "pos", "repeat", "return", "row_embeddings", "self", "shape", "torch", "unsqueeze", "width", "width_values", "x_emb", "y_emb"], "deformable_detr/modeling_deformable_detr.py:build_position_encoding": ["ModelLearnedPositionEmbedding", "ModelSinePositionEmbedding", "Model_position_encoding", "Not", "True", "ValueError", "config", "d_model", "def", "elif", "else", "f", "if", "learned", "n_steps", "normalize", "position_embedding", "position_embedding_type", "raise", "return", "sine", "supported"], "deformable_detr/modeling_deformable_detr.py:DeformableDetrMultiscaleDeformableAttention.__init__": ["CUDA", "Linear", "ModelConfig", "ModelMultiscaleModelAttention", "Module", "MultiScaleModelAttention", "ValueError", "You", "__init__", "a", "and", "attention", "attention_weights", "attn", "authors", "be", "better", "but", "by", "class", "config", "d", "d_model", "def", "dim_per_head", "dimension", "disable_custom_kernels", "divisible", "each", "efficient", "embed_dim", "f", "got", "head", "if", "im2col_step", "implementation", "in", "int", "is", "make", "more", "must", "n_heads", "n_levels", "n_points", "nn", "not", "num_feature_levels", "num_heads", "of", "output_proj", "power", "raise", "sampling_offsets", "self", "set", "super", "the", "to", "value_proj", "warn", "warnings", "which"], "deformable_detr/modeling_deformable_detr.py:DeformableDetrMultiscaleDeformableAttention.with_pos_embed": ["ModelMultiscaleModelAttention", "Module", "None", "Tensor", "class", "def", "else", "if", "is", "nn", "position_embeddings", "return", "self", "tensor", "torch", "with_pos_embed"], "deformable_detr/modeling_deformable_detr.py:DeformableDetrMultiscaleDeformableAttention.forward": ["F", "False", "Last", "Make", "ModelMultiscaleModelAttention", "Module", "None", "Tensor", "ValueError", "_", "align", "attention_mask", "attention_weights", "attn", "batch_size", "be", "bool", "but", "class", "d_model", "def", "dim", "elif", "else", "encoder", "encoder_attention_mask", "encoder_hidden_states", "f", "float", "for", "forward", "got", "height", "hidden", "hidden_states", "if", "im2col_step", "in", "is", "length", "level_start_index", "masked_fill", "must", "n_heads", "n_levels", "n_points", "nn", "not", "num_coordinates", "num_queries", "of", "offset_normalizer", "or", "output", "output_attentions", "output_proj", "position_embeddings", "raise", "reference_points", "return", "sampling_locations", "sampling_offsets", "self", "sequence", "sequence_length", "shape", "shapes", "softmax", "spatial", "spatial_shapes", "spatial_shapes_list", "stack", "states", "sum", "sure", "the", "to", "torch", "total_elements", "value", "value_proj", "view", "width", "with", "with_pos_embed"], "deformable_detr/modeling_deformable_detr.py:DeformableDetrMultiheadAttention.__init__": ["Linear", "ModelMultiheadAttention", "Module", "True", "ValueError", "__init__", "and", "be", "bias", "bool", "by", "class", "def", "divisible", "dropout", "embed_dim", "f", "float", "got", "head_dim", "if", "int", "k_proj", "must", "nn", "num_heads", "out_proj", "q_proj", "raise", "scaling", "self", "super", "v_proj"], "deformable_detr/modeling_deformable_detr.py:DeformableDetrMultiheadAttention._shape": ["ModelMultiheadAttention", "Module", "Tensor", "_shape", "batch_size", "class", "contiguous", "def", "head_dim", "int", "nn", "num_heads", "return", "self", "seq_len", "tensor", "torch", "transpose", "view"], "deformable_detr/modeling_deformable_detr.py:DeformableDetrMultiheadAttention.with_pos_embed": ["ModelMultiheadAttention", "Module", "None", "Tensor", "class", "def", "else", "if", "is", "nn", "position_embeddings", "return", "self", "tensor", "torch", "with_pos_embed"], "deformable_detr/modeling_deformable_detr.py:DeformableDetrMultiheadAttention.forward": ["Attention", "False", "ModelMultiheadAttention", "Module", "None", "Tensor", "ValueError", "_prepare_4d_attention_mask", "_shape", "attention_mask", "attn_output", "attn_probs", "attn_weights", "attn_weights_reshaped", "batch_size", "be", "bmm", "bool", "but", "class", "def", "dim", "dropout", "dtype", "else", "embed_dim", "f", "forward", "functional", "head_dim", "hidden_states", "hidden_states_original", "if", "inf", "is", "k_proj", "key_states", "mask", "masked_fill_", "nn", "not", "num_heads", "of", "out_proj", "output_attentions", "p", "position_embeddings", "proj_shape", "q_proj", "query_states", "raise", "reshape", "return", "scaling", "self", "should", "size", "softmax", "source_len", "target_len", "torch", "training", "transpose", "tuple", "v_proj", "value_states", "view", "weights", "with_pos_embed", "zeros_like"], "deformable_detr/modeling_deformable_detr.py:DeformableDetrEncoderLayer.__init__": ["ACT2FN", "GradientCheckpointingLayer", "LayerNorm", "Linear", "ModelConfig", "ModelEncoderLayer", "ModelMultiscaleModelAttention", "__init__", "activation_dropout", "activation_fn", "activation_function", "class", "config", "d_model", "def", "dropout", "embed_dim", "encoder_attention_heads", "encoder_ffn_dim", "encoder_n_points", "fc1", "fc2", "final_layer_norm", "n_points", "nn", "num_heads", "self", "self_attn", "self_attn_layer_norm", "super"], "deformable_detr/modeling_deformable_detr.py:DeformableDetrEncoderLayer.forward": ["False", "GradientCheckpointingLayer", "ModelEncoderLayer", "None", "Tensor", "activation_dropout", "activation_fn", "any", "attention_mask", "attn_weights", "bool", "clamp", "clamp_value", "class", "def", "dropout", "dtype", "encoder_attention_mask", "encoder_hidden_states", "fc1", "fc2", "final_layer_norm", "finfo", "forward", "functional", "hidden_states", "if", "isinf", "isnan", "level_start_index", "max", "min", "nn", "or", "output_attentions", "outputs", "p", "position_embeddings", "reference_points", "residual", "return", "self", "self_attn", "self_attn_layer_norm", "spatial_shapes", "spatial_shapes_list", "torch", "training"], "deformable_detr/modeling_deformable_detr.py:DeformableDetrDecoderLayer.__init__": ["ACT2FN", "GradientCheckpointingLayer", "LayerNorm", "Linear", "ModelConfig", "ModelDecoderLayer", "ModelMultiheadAttention", "ModelMultiscaleModelAttention", "__init__", "activation_dropout", "activation_fn", "activation_function", "attention_dropout", "class", "config", "d_model", "decoder_attention_heads", "decoder_ffn_dim", "decoder_n_points", "def", "dropout", "embed_dim", "encoder_attn", "encoder_attn_layer_norm", "fc1", "fc2", "final_layer_norm", "n_points", "nn", "num_heads", "self", "self_attn", "self_attn_layer_norm", "super"], "deformable_detr/modeling_deformable_detr.py:DeformableDetrDecoderLayer.forward": ["False", "GradientCheckpointingLayer", "ModelDecoderLayer", "None", "Tensor", "activation_dropout", "activation_fn", "attention_mask", "bool", "class", "cross_attn_weights", "def", "dropout", "encoder_attention_mask", "encoder_attn", "encoder_attn_layer_norm", "encoder_hidden_states", "fc1", "fc2", "final_layer_norm", "forward", "functional", "hidden_states", "if", "level_start_index", "nn", "output_attentions", "outputs", "p", "position_embeddings", "reference_points", "residual", "return", "second_residual", "self", "self_attn", "self_attn_layer_norm", "self_attn_weights", "spatial_shapes", "spatial_shapes_list", "torch", "training"], "deformable_detr/modeling_deformable_detr.py:DeformableDetrPreTrainedModel._init_weights": ["Conv2d", "Embedding", "False", "Linear", "ModelLearnedPositionEmbedding", "ModelMultiscaleModelAttention", "ModelPreTrainedModel", "None", "PreTrainedModel", "True", "_init_weights", "_is_hf_initialized", "abs", "and", "arange", "attention_weights", "bias", "class", "column_embeddings", "config", "constant_", "copy_", "cos", "def", "default_dtype", "dtype", "elif", "for", "gain", "get_default_dtype", "getattr", "grid_init", "hasattr", "i", "if", "in", "init", "init_std", "int64", "is", "isinstance", "keepdim", "level_embed", "math", "max", "mean", "module", "n_heads", "n_levels", "n_points", "nn", "normal_", "not", "output_proj", "padding_idx", "pi", "range", "reference_points", "repeat", "row_embeddings", "sampling_offsets", "self", "sin", "stack", "std", "thetas", "to", "torch", "two_stage", "uniform_", "value_proj", "view", "weight", "xavier_uniform_", "zeros_"], "deformable_detr/modeling_deformable_detr.py:DeformableDetrEncoder.__init__": ["False", "ModelConfig", "ModelEncoder", "ModelEncoderLayer", "ModelPreTrainedModel", "ModuleList", "_", "__init__", "class", "config", "def", "dropout", "encoder_layers", "for", "gradient_checkpointing", "in", "layers", "nn", "post_init", "range", "self", "super"], "deformable_detr/modeling_deformable_detr.py:DeformableDetrEncoder.get_reference_points": ["ModelEncoder", "ModelPreTrainedModel", "None", "append", "cat", "class", "def", "device", "dtype", "enumerate", "for", "get_reference_points", "height", "ij", "in", "indexing", "level", "linspace", "meshgrid", "ref", "ref_x", "ref_y", "reference_points", "reference_points_list", "reshape", "return", "spatial_shapes", "stack", "torch", "valid_ratios", "width"], "deformable_detr/modeling_deformable_detr.py:DeformableDetrEncoder.forward": ["BaseModelOutput", "ModelEncoder", "ModelPreTrainedModel", "None", "all_attentions", "attention_mask", "attentions", "class", "config", "def", "device", "dropout", "else", "encoder_layer", "encoder_states", "enumerate", "for", "forward", "functional", "get_reference_points", "hidden_states", "i", "if", "in", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layer_outputs", "layers", "level_start_index", "nn", "not", "output_attentions", "output_hidden_states", "p", "position_embeddings", "r", "reference_points", "return", "return_dict", "self", "spatial_shapes", "spatial_shapes_list", "spatial_shapes_tuple", "training", "tuple", "use_return_dict", "v", "valid_ratios"], "deformable_detr/modeling_deformable_detr.py:DeformableDetrDecoder.__init__": ["False", "ModelConfig", "ModelDecoder", "ModelDecoderLayer", "ModelPreTrainedModel", "ModuleList", "None", "_", "__init__", "bbox_embed", "class", "class_embed", "config", "decoder_layers", "def", "dropout", "for", "gradient_checkpointing", "in", "layers", "nn", "post_init", "range", "self", "super"], "deformable_detr/modeling_deformable_detr.py:DeformableDetrDecoder.forward": ["Last", "ModelDecoder", "ModelDecoderOutput", "ModelPreTrainedModel", "None", "Reference", "ValueError", "all_cross_attentions", "all_hidden_states", "all_self_attns", "and", "attentions", "bbox_embed", "be", "but", "cat", "class", "config", "cross_attentions", "decoder_layer", "def", "detach", "dim", "dimension", "elif", "else", "encoder_attention_mask", "encoder_hidden_states", "enumerate", "f", "for", "forward", "got", "hidden_states", "idx", "if", "in", "inputs_embeds", "intermediate", "intermediate_hidden_states", "intermediate_reference_points", "inverse_sigmoid", "is", "kwargs", "last", "last_hidden_state", "layer_outputs", "layers", "level_start_index", "must", "new_reference_points", "not", "num_coordinates", "of", "or", "output_attentions", "output_hidden_states", "points", "position_embeddings", "r", "raise", "reference_points", "reference_points_input", "return", "return_dict", "self", "shape", "sigmoid", "size", "spatial_shapes", "spatial_shapes_list", "stack", "tmp", "torch", "tuple", "use_return_dict", "v", "valid_ratios"], "deformable_detr/modeling_deformable_detr.py:DeformableDetrModel.__init__": ["Conv2d", "Embedding", "GroupNorm", "LayerNorm", "Linear", "ModelConfig", "ModelConvEncoder", "ModelConvModel", "ModelDecoder", "ModelEncoder", "ModelModel", "ModelPreTrainedModel", "ModuleList", "Parameter", "Sequential", "Tensor", "_", "__init__", "append", "backbone", "build_position_encoding", "class", "config", "d_model", "decoder", "def", "else", "enc_output", "enc_output_norm", "encoder", "for", "if", "in", "in_channels", "input_proj", "input_proj_list", "intermediate_channel_sizes", "kernel_size", "len", "level_embed", "nn", "not", "num_backbone_outs", "num_feature_levels", "num_queries", "padding", "pos_trans", "pos_trans_norm", "position_embeddings", "post_init", "query_position_embeddings", "range", "reference_points", "self", "stride", "super", "torch", "two_stage"], "deformable_detr/modeling_deformable_detr.py:DeformableDetrModel.freeze_backbone": ["False", "ModelModel", "ModelPreTrainedModel", "backbone", "class", "conv_encoder", "def", "for", "freeze_backbone", "in", "model", "name", "named_parameters", "param", "requires_grad_", "self"], "deformable_detr/modeling_deformable_detr.py:DeformableDetrModel.unfreeze_backbone": ["ModelModel", "ModelPreTrainedModel", "True", "backbone", "class", "conv_encoder", "def", "for", "in", "model", "name", "named_parameters", "param", "requires_grad_", "self", "unfreeze_backbone"], "deformable_detr/modeling_deformable_detr.py:DeformableDetrModel.get_valid_ratio": ["ModelModel", "ModelPreTrainedModel", "_", "class", "def", "dtype", "float32", "get_valid_ratio", "height", "mask", "return", "self", "shape", "stack", "sum", "to", "torch", "valid_height", "valid_ratio", "valid_ratio_height", "valid_ratio_width", "valid_width", "width"], "deformable_detr/modeling_deformable_detr.py:DeformableDetrModel.get_proposal_pos_embed": ["ModelModel", "ModelPreTrainedModel", "None", "arange", "class", "config", "cos", "d_model", "def", "device", "dim", "dim_t", "div", "dtype", "flatten", "floor", "get_proposal_pos_embed", "math", "num_pos_feats", "pi", "pos", "proposals", "return", "rounding_mode", "scale", "self", "sigmoid", "sin", "stack", "temperature", "torch"], "deformable_detr/modeling_deformable_detr.py:DeformableDetrModel.gen_encoder_output_proposals": ["ModelModel", "ModelPreTrainedModel", "True", "_cur", "all", "append", "batch_size", "cat", "class", "def", "device", "dtype", "enc_output", "enc_output_norm", "enumerate", "expand", "float", "for", "gen_encoder_output_proposals", "grid", "grid_x", "grid_y", "height", "ij", "in", "indexing", "inf", "keepdim", "level", "linspace", "log", "mask_flatten_", "masked_fill", "meshgrid", "object_query", "ones_like", "output_proposals", "output_proposals_valid", "padding_mask", "proposal", "proposals", "return", "scale", "self", "shape", "spatial_shapes", "sum", "torch", "unsqueeze", "valid_height", "valid_width", "view", "width", "width_height"], "deformable_detr/modeling_deformable_detr.py:DeformableDetrModel.forward": ["BaseModelOutput", "FloatTensor", "LongTensor", "ModelModel", "ModelModelOutput", "ModelPreTrainedModel", "No", "None", "ValueError", "_", "_len_sources", "and", "append", "as_tensor", "attention", "attention_mask", "attentions", "backbone", "batch_size", "bbox_embed", "bool", "cat", "class", "class_embed", "config", "cross_attentions", "cumsum", "decoder", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_inputs_embeds", "decoder_outputs", "def", "delta_bbox", "detach", "device", "dim", "dtype", "elif", "else", "enc_outputs", "enc_outputs_class", "enc_outputs_coord_logits", "encoder", "encoder_attention_mask", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "enumerate", "expand", "features", "flatten", "for", "forward", "functional", "gather", "gen_encoder_output_proposals", "get_proposal_pos_embed", "get_valid_ratio", "height", "hidden_states", "if", "in", "init_reference_points", "input_proj", "inputs_embeds", "intermediate_hidden_states", "intermediate_reference_points", "interpolate", "is", "isinstance", "kwargs", "last_hidden_state", "len", "level", "level_embed", "level_start_index", "long", "lvl_pos_embed", "lvl_pos_embed_flatten", "m", "mask", "mask_flatten", "masks", "new_zeros", "nn", "not", "num_channels", "num_feature_levels", "object_query_embedding", "ones", "output_attentions", "output_hidden_states", "output_proposals", "pixel_mask", "pixel_values", "pos_embed", "pos_l", "pos_trans", "pos_trans_norm", "pos_trans_out", "position_embedding", "position_embeddings", "position_embeddings_list", "prod", "provided", "query_embed", "query_embeds", "query_position_embeddings", "r", "raise", "range", "reference_points", "repeat", "return", "return_dict", "self", "shape", "sigmoid", "size", "source", "source_flatten", "sources", "spatial_shape", "spatial_shapes", "spatial_shapes_list", "split", "stack", "target", "to", "topk", "topk_coords_logits", "topk_proposals", "torch", "transpose", "tuple", "tuple_outputs", "two_stage", "two_stage_num_proposals", "unsqueeze", "use_return_dict", "valid_ratios", "value", "view", "was", "weight", "width", "zip"], "deformable_detr/modeling_deformable_detr.py:DeformableDetrMLPPredictionHead.__init__": ["Linear", "ModelMLPPredictionHead", "Module", "ModuleList", "__init__", "class", "def", "for", "h", "hidden_dim", "in", "input_dim", "k", "layers", "n", "nn", "num_layers", "output_dim", "self", "super", "zip"], "deformable_detr/modeling_deformable_detr.py:DeformableDetrMLPPredictionHead.forward": ["ModelMLPPredictionHead", "Module", "class", "def", "else", "enumerate", "for", "forward", "functional", "i", "if", "in", "layer", "layers", "nn", "num_layers", "relu", "return", "self", "x"], "deformable_detr/modeling_deformable_detr.py:DeformableDetrForObjectDetection.__init__": ["Linear", "ModelConfig", "ModelForObjectDetection", "ModelMLPPredictionHead", "ModelModel", "ModelPreTrainedModel", "ModuleList", "_", "__init__", "_tied_weights_keys", "bbox_embed", "class", "class_embed", "config", "d_model", "decoder", "decoder_layers", "def", "else", "for", "hidden_dim", "if", "in", "input_dim", "model", "nn", "num_labels", "num_layers", "num_pred", "output_dim", "post_init", "range", "self", "super", "two_stage", "with_box_refine"], "deformable_detr/modeling_deformable_detr.py:DeformableDetrForObjectDetection.forward": ["FloatTensor", "LongTensor", "ModelForObjectDetection", "ModelObjectDetectionOutput", "ModelPreTrainedModel", "None", "ValueError", "append", "auxiliary_outputs", "bbox_embed", "be", "bool", "but", "class", "class_embed", "config", "cross_attentions", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_inputs_embeds", "def", "delta_bbox", "device", "dict", "dict_outputs", "elif", "else", "enc_outputs_class", "enc_outputs_coord_logits", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "f", "for", "forward", "got", "hidden_states", "if", "in", "init_reference", "init_reference_points", "inputs_embeds", "inter_references", "intermediate_hidden_states", "intermediate_reference_points", "inverse_sigmoid", "is", "kwargs", "labels", "last_hidden_state", "level", "list", "logits", "loss", "loss_dict", "loss_function", "model", "not", "or", "output", "output_attentions", "output_hidden_states", "outputs", "outputs_class", "outputs_classes", "outputs_coord", "outputs_coord_logits", "outputs_coords", "pixel_mask", "pixel_values", "pred_boxes", "r", "raise", "range", "reference", "return", "return_dict", "self", "shape", "should", "sigmoid", "stack", "torch", "tuple", "tuple_outputs", "use_return_dict"], "deit/modeling_deit.py:DeiTEmbeddings.__init__": ["Dropout", "False", "ModelConfig", "ModelEmbeddings", "ModelPatchEmbeddings", "Module", "None", "Parameter", "__init__", "bool", "class", "cls_token", "config", "def", "distillation_token", "dropout", "else", "hidden_dropout_prob", "hidden_size", "if", "mask_token", "nn", "num_patches", "patch_embeddings", "patch_size", "position_embeddings", "self", "super", "torch", "use_mask_token", "zeros"], "deit/modeling_deit.py:DeiTEmbeddings.interpolate_pos_encoding": ["False", "ModelEmbeddings", "Module", "Tensor", "align_corners", "and", "bicubic", "cat", "class", "class_and_dist_pos_embed", "def", "dim", "embeddings", "functional", "height", "if", "int", "interpolate", "interpolate_pos_encoding", "is_tracing", "jit", "mode", "new_height", "new_width", "nn", "not", "num_patches", "num_positions", "patch_pos_embed", "patch_size", "permute", "position_embeddings", "reshape", "return", "self", "shape", "size", "sqrt_num_positions", "torch", "torch_int", "view", "width"], "deit/modeling_deit.py:DeiTEmbeddings.forward": ["BoolTensor", "False", "ModelEmbeddings", "Module", "None", "Tensor", "_", "batch_size", "bool", "bool_masked_pos", "cat", "class", "cls_token", "cls_tokens", "def", "dim", "distillation_token", "distillation_tokens", "dropout", "embeddings", "expand", "forward", "height", "if", "interpolate_pos_encoding", "is", "mask", "mask_token", "mask_tokens", "nn", "not", "patch_embeddings", "pixel_values", "position_embedding", "position_embeddings", "return", "self", "seq_length", "shape", "size", "torch", "type_as", "unsqueeze", "width"], "deit/modeling_deit.py:DeiTPatchEmbeddings.__init__": ["Conv2d", "Iterable", "ModelPatchEmbeddings", "Module", "__init__", "abc", "class", "collections", "config", "def", "else", "hidden_size", "if", "image_size", "isinstance", "kernel_size", "nn", "num_channels", "num_patches", "patch_size", "projection", "self", "stride", "super"], "deit/modeling_deit.py:DeiTPatchEmbeddings.forward": ["Make", "ModelPatchEmbeddings", "Module", "Tensor", "ValueError", "batch_size", "channel", "class", "configuration", "def", "dimension", "flatten", "forward", "height", "if", "in", "match", "nn", "num_channels", "of", "one", "pixel", "pixel_values", "projection", "raise", "return", "self", "set", "shape", "sure", "that", "the", "torch", "transpose", "values", "width", "with", "x"], "deit/modeling_deit.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "contiguous", "def", "dim", "dropout", "float", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "shape", "size", "softmax", "torch", "training", "transpose", "value"], "deit/modeling_deit.py:DeiTSelfAttention.__init__": ["False", "Linear", "ModelConfig", "ModelSelfAttention", "Module", "The", "ValueError", "__init__", "a", "all_head_size", "and", "attention", "attention_head_size", "attention_probs_dropout_prob", "bias", "class", "config", "def", "dropout_prob", "embedding_size", "f", "hasattr", "heads", "hidden", "hidden_size", "if", "int", "is", "is_causal", "key", "multiple", "nn", "not", "num_attention_heads", "number", "of", "qkv_bias", "query", "raise", "scaling", "self", "size", "super", "the", "value"], "deit/modeling_deit.py:DeiTSelfAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "ModelSelfAttention", "Module", "None", "Tensor", "_attn_implementation", "all_head_size", "attention_head_size", "attention_interface", "attention_probs", "batch_size", "class", "config", "context_layer", "def", "dropout", "dropout_prob", "eager", "eager_attention_forward", "else", "forward", "hidden_states", "if", "is_causal", "key", "key_layer", "new_context_layer_shape", "new_shape", "nn", "not", "num_attention_heads", "query", "query_layer", "reshape", "return", "scaling", "self", "shape", "size", "torch", "training", "transpose", "tuple", "value", "value_layer", "view"], "deit/modeling_deit.py:DeiTSelfOutput.__init__": ["Dropout", "Linear", "ModelConfig", "ModelSelfOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "hidden_dropout_prob", "hidden_size", "nn", "self", "super"], "deit/modeling_deit.py:DeiTSelfOutput.forward": ["ModelSelfOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "deit/modeling_deit.py:DeiTAttention.__init__": ["ModelAttention", "ModelConfig", "ModelSelfAttention", "ModelSelfOutput", "Module", "__init__", "attention", "class", "config", "def", "nn", "output", "self", "super"], "deit/modeling_deit.py:DeiTAttention.forward": ["ModelAttention", "Module", "Tensor", "_", "attention", "class", "def", "forward", "hidden_states", "nn", "output", "return", "self", "self_attn_output", "torch"], "deit/modeling_deit.py:DeiTIntermediate.__init__": ["ACT2FN", "Linear", "ModelConfig", "ModelIntermediate", "Module", "__init__", "class", "config", "def", "dense", "else", "hidden_act", "hidden_size", "if", "intermediate_act_fn", "intermediate_size", "isinstance", "nn", "self", "str", "super"], "deit/modeling_deit.py:DeiTIntermediate.forward": ["ModelIntermediate", "Module", "Tensor", "class", "def", "dense", "forward", "hidden_states", "intermediate_act_fn", "nn", "return", "self", "torch"], "deit/modeling_deit.py:DeiTOutput.__init__": ["Dropout", "Linear", "ModelConfig", "ModelOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "hidden_dropout_prob", "hidden_size", "intermediate_size", "nn", "self", "super"], "deit/modeling_deit.py:DeiTOutput.forward": ["ModelOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "deit/modeling_deit.py:DeiTLayer.__init__": ["GradientCheckpointingLayer", "LayerNorm", "ModelAttention", "ModelConfig", "ModelIntermediate", "ModelLayer", "ModelOutput", "__init__", "attention", "chunk_size_feed_forward", "class", "config", "def", "eps", "hidden_size", "intermediate", "layer_norm_eps", "layernorm_after", "layernorm_before", "nn", "output", "self", "seq_len_dim", "super"], "deit/modeling_deit.py:DeiTLayer.forward": ["GradientCheckpointingLayer", "ModelLayer", "Tensor", "attention", "attention_output", "class", "def", "forward", "hidden_states", "hidden_states_norm", "intermediate", "layer_output", "layernorm_after", "layernorm_before", "output", "return", "self", "torch"], "deit/modeling_deit.py:DeiTEncoder.__init__": ["False", "ModelConfig", "ModelEncoder", "ModelLayer", "Module", "ModuleList", "_", "__init__", "class", "config", "def", "for", "gradient_checkpointing", "in", "layer", "nn", "num_hidden_layers", "range", "self", "super"], "deit/modeling_deit.py:DeiTEncoder.forward": ["BaseModelOutput", "ModelEncoder", "Module", "Tensor", "class", "def", "enumerate", "for", "forward", "hidden_states", "i", "in", "last_hidden_state", "layer", "layer_module", "nn", "return", "self", "torch"], "deit/modeling_deit.py:DeiTPreTrainedModel._init_weights": ["Conv2d", "LayerNorm", "Linear", "ModelEmbeddings", "ModelPreTrainedModel", "None", "PreTrainedModel", "_init_weights", "bias", "class", "cls_token", "config", "def", "distillation_token", "elif", "if", "init", "initializer_range", "is", "isinstance", "mask_token", "mean", "module", "nn", "not", "ones_", "position_embeddings", "self", "std", "trunc_normal_", "weight", "zeros_"], "deit/modeling_deit.py:DeiTModel.__init__": ["False", "LayerNorm", "ModelConfig", "ModelEmbeddings", "ModelEncoder", "ModelModel", "ModelPooler", "ModelPreTrainedModel", "None", "True", "__init__", "add_pooling_layer", "bool", "class", "config", "def", "else", "embeddings", "encoder", "eps", "hidden_size", "if", "layer_norm_eps", "layernorm", "nn", "pooler", "post_init", "r", "self", "super", "use_mask_token"], "deit/modeling_deit.py:DeiTModel.get_input_embeddings": ["ModelModel", "ModelPatchEmbeddings", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "patch_embeddings", "return", "self"], "deit/modeling_deit.py:DeiTModel.forward": ["BaseModelOutput", "BaseModelOutputWithPooling", "BoolTensor", "False", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "bool", "bool_masked_pos", "class", "def", "dtype", "else", "embedding_output", "embeddings", "encoder", "encoder_outputs", "expected_dtype", "forward", "have", "if", "interpolate_pos_encoding", "is", "kwargs", "last_hidden_state", "layernorm", "not", "patch_embeddings", "pixel_values", "pooled_output", "pooler", "pooler_output", "projection", "r", "raise", "return", "self", "sequence_output", "specify", "to", "torch", "weight"], "deit/modeling_deit.py:DeiTPooler.__init__": ["ACT2FN", "Linear", "ModelConfig", "ModelPooler", "Module", "__init__", "activation", "class", "config", "def", "dense", "hidden_size", "nn", "pooler_act", "pooler_output_size", "self", "super"], "deit/modeling_deit.py:DeiTPooler.forward": ["ModelPooler", "Module", "Tensor", "activation", "class", "def", "dense", "first_token_tensor", "forward", "hidden_states", "nn", "pooled_output", "return", "self", "torch"], "deit/modeling_deit.py:DeiTForMaskedImageModeling.__init__": ["Conv2d", "False", "Model", "ModelConfig", "ModelForMaskedImageModeling", "ModelModel", "ModelPreTrainedModel", "None", "PixelShuffle", "Sequential", "True", "__init__", "add_pooling_layer", "class", "config", "decoder", "def", "encoder_stride", "hidden_size", "in_channels", "kernel_size", "nn", "num_channels", "out_channels", "post_init", "self", "super", "use_mask_token"], "deit/modeling_deit.py:DeiTForMaskedImageModeling.forward": ["BaseModelOutputWithPooling", "BoolTensor", "False", "MaskedImageModelingOutput", "Model", "ModelForMaskedImageModeling", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "attentions", "batch_size", "bool", "bool_masked_pos", "class", "config", "contiguous", "decoder", "def", "forward", "functional", "height", "hidden_states", "if", "image_size", "int", "interpolate_pos_encoding", "is", "kwargs", "l1_loss", "last_hidden_state", "loss", "mask", "masked_im_loss", "nn", "none", "not", "num_channels", "outputs", "patch_size", "permute", "pixel_values", "r", "reconstructed_pixel_values", "reconstruction", "reconstruction_loss", "reduction", "repeat_interleave", "reshape", "return", "self", "sequence_length", "sequence_output", "shape", "size", "sum", "torch", "unsqueeze", "width"], "deit/modeling_deit.py:DeiTForImageClassification.__init__": ["False", "Identity", "Linear", "Model", "ModelConfig", "ModelForImageClassification", "ModelModel", "ModelPreTrainedModel", "None", "__init__", "add_pooling_layer", "class", "classifier", "config", "def", "else", "hidden_size", "if", "nn", "num_labels", "post_init", "self", "super"], "deit/modeling_deit.py:DeiTForImageClassification.forward": ["BaseModelOutputWithPooling", "False", "ImageClassifierOutput", "Model", "ModelForImageClassification", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "attentions", "bool", "class", "classifier", "config", "def", "forward", "hidden_states", "if", "interpolate_pos_encoding", "is", "kwargs", "labels", "last_hidden_state", "logits", "loss", "loss_function", "not", "outputs", "pixel_values", "r", "return", "self", "sequence_output", "torch"], "deit/modeling_deit.py:DeiTForImageClassificationWithTeacher.__init__": ["False", "Identity", "Linear", "Model", "ModelConfig", "ModelForImageClassificationWithTeacher", "ModelModel", "ModelPreTrainedModel", "None", "__init__", "add_pooling_layer", "class", "cls_classifier", "config", "def", "distillation_classifier", "else", "hidden_size", "if", "nn", "num_labels", "post_init", "self", "super"], "deit/modeling_deit.py:DeiTForImageClassificationWithTeacher.forward": ["BaseModelOutputWithPooling", "False", "Model", "ModelForImageClassificationWithTeacher", "ModelForImageClassificationWithTeacherOutput", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "attentions", "bool", "class", "cls_classifier", "cls_logits", "def", "distillation_classifier", "distillation_logits", "forward", "hidden_states", "interpolate_pos_encoding", "kwargs", "last_hidden_state", "logits", "outputs", "pixel_values", "return", "self", "sequence_output", "torch"], "depth_anything/modeling_depth_anything.py:DepthAnythingReassembleLayer.__init__": ["Conv2d", "ConvTranspose2d", "Identity", "ModelReassembleLayer", "Module", "__init__", "channels", "class", "config", "def", "elif", "factor", "if", "in_channels", "int", "kernel_size", "nn", "out_channels", "padding", "projection", "reassemble_hidden_size", "resize", "self", "stride", "super"], "depth_anything/modeling_depth_anything.py:DepthAnythingReassembleLayer.forward": ["ModelReassembleLayer", "Module", "class", "def", "forward", "hidden_state", "nn", "projection", "resize", "return", "self"], "depth_anything/modeling_depth_anything.py:DepthAnythingReassembleStage.__init__": ["ModelReassembleLayer", "ModelReassembleStage", "Module", "ModuleList", "__init__", "append", "channels", "class", "config", "def", "factor", "for", "in", "layers", "neck_hidden_sizes", "nn", "reassemble_factors", "self", "super", "zip"], "depth_anything/modeling_depth_anything.py:DepthAnythingReassembleStage.forward": ["ModelReassembleStage", "Module", "None", "Tensor", "_", "append", "batch_size", "class", "contiguous", "def", "enumerate", "for", "forward", "hidden_state", "hidden_states", "i", "in", "layers", "list", "nn", "num_channels", "out", "patch_height", "patch_width", "permute", "reshape", "return", "self", "shape", "torch"], "depth_anything/modeling_depth_anything.py:DepthAnythingPreActResidualLayer.__init__": ["Conv2d", "ModelPreActResidualLayer", "Module", "ReLU", "True", "__init__", "activation1", "activation2", "bias", "class", "config", "convolution1", "convolution2", "def", "fusion_hidden_size", "kernel_size", "nn", "padding", "self", "stride", "super"], "depth_anything/modeling_depth_anything.py:DepthAnythingPreActResidualLayer.forward": ["ModelPreActResidualLayer", "Module", "Tensor", "activation1", "activation2", "class", "convolution1", "convolution2", "def", "forward", "hidden_state", "nn", "residual", "return", "self", "torch"], "depth_anything/modeling_depth_anything.py:DepthAnythingFeatureFusionLayer.__init__": ["Conv2d", "ModelFeatureFusionLayer", "ModelPreActResidualLayer", "Module", "True", "__init__", "bias", "class", "config", "def", "fusion_hidden_size", "kernel_size", "nn", "projection", "residual_layer1", "residual_layer2", "self", "super"], "depth_anything/modeling_depth_anything.py:DepthAnythingFeatureFusionLayer.forward": ["False", "ModelFeatureFusionLayer", "Module", "None", "True", "align_corners", "bilinear", "class", "def", "else", "forward", "functional", "hidden_state", "if", "interpolate", "is", "mode", "modifier", "nn", "not", "projection", "residual", "residual_layer1", "residual_layer2", "return", "scale_factor", "self", "shape", "size"], "depth_anything/modeling_depth_anything.py:DepthAnythingFeatureFusionStage.__init__": ["ModelConfig", "ModelFeatureFusionLayer", "ModelFeatureFusionStage", "Module", "ModuleList", "_", "__init__", "append", "class", "config", "def", "for", "in", "layers", "len", "neck_hidden_sizes", "nn", "range", "self", "super"], "depth_anything/modeling_depth_anything.py:DepthAnythingFeatureFusionStage.forward": ["ModelFeatureFusionStage", "Module", "None", "append", "class", "def", "else", "enumerate", "for", "forward", "fused_hidden_state", "fused_hidden_states", "hidden_state", "hidden_states", "idx", "if", "in", "is", "layer", "layers", "len", "nn", "return", "self", "shape", "size", "zip"], "depth_anything/modeling_depth_anything.py:DepthAnythingNeck.__init__": ["Conv2d", "False", "ModelFeatureFusionStage", "ModelNeck", "ModelReassembleStage", "Module", "ModuleList", "__init__", "append", "bias", "channel", "class", "config", "convs", "def", "for", "fusion_hidden_size", "fusion_stage", "in", "kernel_size", "neck_hidden_sizes", "nn", "padding", "reassemble_stage", "self", "super"], "depth_anything/modeling_depth_anything.py:DepthAnythingNeck.forward": ["ModelNeck", "Module", "None", "Tensor", "The", "TypeError", "ValueError", "a", "be", "class", "config", "convs", "def", "enumerate", "equal", "feature", "features", "for", "forward", "fusion_stage", "hidden", "hidden_states", "i", "if", "in", "isinstance", "len", "list", "neck", "neck_hidden_sizes", "nn", "not", "number", "of", "or", "output", "patch_height", "patch_width", "raise", "reassemble_stage", "return", "self", "should", "sizes", "states", "tensors", "the", "to", "torch", "tuple"], "depth_anything/modeling_depth_anything.py:DepthAnythingDepthEstimationHead.__init__": ["Conv2d", "Model", "ModelModelEstimationHead", "Model_estimation_type", "Module", "ReLU", "Sigmoid", "Unknown", "ValueError", "__init__", "activation1", "activation2", "class", "config", "conv1", "conv2", "conv3", "def", "elif", "else", "estimation", "f", "features", "fusion_hidden_size", "head_hidden_size", "head_in_index", "if", "kernel_size", "max_Model", "metric", "nn", "padding", "patch_size", "raise", "relative", "self", "stride", "super", "type"], "depth_anything/modeling_depth_anything.py:DepthAnythingDepthEstimationHead.forward": ["ModelModelEstimationHead", "Module", "Tensor", "True", "activation1", "activation2", "align_corners", "bilinear", "class", "conv1", "conv2", "conv3", "def", "dim", "forward", "functional", "head_in_index", "hidden_states", "int", "interpolate", "list", "max_Model", "mode", "nn", "patch_height", "patch_size", "patch_width", "predicted_Model", "return", "self", "squeeze", "torch"], "depth_anything/modeling_depth_anything.py:DepthAnythingForDepthEstimation.__init__": ["ModelForModelEstimation", "ModelModelEstimationHead", "ModelNeck", "ModelPreTrainedModel", "__init__", "backbone", "class", "config", "def", "head", "load_backbone", "neck", "post_init", "self", "super"], "depth_anything/modeling_depth_anything.py:DepthAnythingForDepthEstimation.forward": ["FloatTensor", "LongTensor", "ModelEstimatorOutput", "ModelForModelEstimation", "ModelPreTrainedModel", "None", "NotImplementedError", "Tensor", "Training", "_", "attentions", "backbone", "bool", "class", "config", "def", "else", "feature_maps", "forward", "forward_with_filtered_kwargs", "head", "height", "hidden_states", "if", "implemented", "is", "kwargs", "labels", "loss", "neck", "not", "output", "output_attentions", "output_hidden_states", "outputs", "patch_height", "patch_size", "patch_width", "pixel_values", "predicted_Model", "r", "raise", "return", "return_dict", "self", "shape", "torch", "tuple", "use_return_dict", "width", "yet"], "depth_pro/modeling_depth_pro.py:split_to_patches": ["F", "Model_to_patches", "Tensor", "batch_size", "def", "float", "height", "if", "int", "kernel_size", "num_channels", "overlap_ratio", "patch_size", "patches", "permute", "pixel_values", "reshape", "return", "shape", "stride", "torch", "torch_int", "unfold", "width"], "depth_pro/modeling_depth_pro.py:reshape_features": ["Model", "Model_features", "Tensor", "def", "hidden_size", "hidden_states", "n_samples", "permute", "return", "seq_len", "shape", "size", "torch", "torch_int"], "depth_pro/modeling_depth_pro.py:merge_patches": ["Model_patches", "Modeld", "Tensor", "_", "append", "batch_size", "box", "box_h", "box_w", "boxes", "boxes_in_row", "cat", "def", "dim", "else", "for", "h", "hidden_size", "i", "if", "in", "int", "min", "n_patches", "n_patches_per_batch", "new_out_size", "out_size", "pad_bottom", "pad_left", "pad_right", "pad_top", "padding", "paddings", "patches", "permute", "range", "reshape", "return", "shape", "sqrt_n_patches_per_batch", "torch", "torch_int", "w"], "depth_pro/modeling_depth_pro.py:reconstruct_feature_maps": ["F", "False", "Model_feature_maps", "Tensor", "align_corners", "batch_size", "bilinear", "def", "features", "float", "hidden_state", "int", "interpolate", "merge_patches", "mode", "output_size", "padding", "reshape_features", "return", "size", "torch", "tuple"], "depth_pro/modeling_depth_pro.py:DepthProPatchEncoder.__init__": ["AutoModel", "ModelConfig", "ModelPatchEncoder", "Module", "__init__", "class", "config", "def", "from_config", "image_model_config", "image_size", "intermediate_feature_dims", "intermediate_hook_ids", "len", "merge_padding_value", "model", "n_intermediate_hooks", "n_scaled_images", "nn", "out_size", "patch_model_config", "patch_size", "scaled_images_feature_dims", "scaled_images_overlap_ratios", "scaled_images_ratios", "self", "super"], "depth_pro/modeling_depth_pro.py:DepthProPatchEncoder.forward": ["F", "False", "Image", "ModelPatchEncoder", "Module", "Tensor", "ValueError", "align_corners", "append", "base_height", "base_width", "batch_size", "be", "bilinear", "cat", "class", "config", "def", "dim", "encodings", "exponent_value", "f", "features", "for", "forward", "height", "hidden_state", "i", "if", "in", "intermediate_features", "intermediate_hook_ids", "interpolate", "is", "len", "list", "log2", "math", "merge_padding_value", "min", "mode", "model", "n_intermediate_hooks", "n_patches_per_scaled_image", "n_scaled_images", "nn", "num_channels", "out_size", "output_height", "output_hidden_states", "output_size", "output_width", "overlap_ratio", "padding", "patch_size", "patches", "pixel_values", "raise", "range", "ratio", "reconstruct_feature_maps", "return", "scale_factor", "scaled", "scaled_images", "scaled_images_features", "scaled_images_last_hidden_state", "scaled_images_overlap_ratios", "scaled_images_ratios", "self", "shape", "size", "small", "split_to_patches", "split_with_sizes", "to", "too", "torch", "torch_int", "when", "width", "with", "x"], "depth_pro/modeling_depth_pro.py:DepthProImageEncoder.__init__": ["AutoModel", "ModelConfig", "ModelImageEncoder", "Module", "__init__", "class", "config", "def", "from_config", "image_model_config", "image_size", "model", "nn", "out_size", "patch_size", "self", "super"], "depth_pro/modeling_depth_pro.py:DepthProImageEncoder.forward": ["F", "False", "ModelImageEncoder", "ModelOutput", "Module", "Tensor", "True", "align_corners", "attentions", "base_height", "base_width", "batch_size", "bilinear", "bool", "class", "config", "def", "encodings", "exponent_value", "features", "forward", "height", "hidden_states", "if", "image_model_config", "image_size", "interpolate", "last_hidden_state", "log2", "math", "mode", "model", "nn", "not", "num_channels", "out_size", "output_attentions", "output_hidden_states", "output_size", "padding", "pixel_values", "reconstruct_feature_maps", "return", "return_dict", "self", "shape", "size", "torch", "torch_int", "tuple", "width"], "depth_pro/modeling_depth_pro.py:DepthProEncoder.__init__": ["ModelConfig", "ModelEncoder", "ModelImageEncoder", "ModelPatchEncoder", "Module", "__init__", "class", "config", "def", "image_encoder", "intermediate_feature_dims", "intermediate_hook_ids", "len", "merge_padding_value", "n_intermediate_hooks", "n_scaled_images", "nn", "patch_encoder", "scaled_images_feature_dims", "scaled_images_overlap_ratios", "scaled_images_ratios", "self", "super"], "depth_pro/modeling_depth_pro.py:DepthProEncoder.forward": ["False", "ModelEncoder", "ModelOutput", "Module", "Tensor", "True", "attentions", "batch_size", "bool", "class", "def", "features", "forward", "height", "hidden_states", "if", "image_encoder", "image_encodings", "image_features", "last_hidden_state", "nn", "not", "num_channels", "output_attentions", "output_hidden_states", "patch_encoder", "patch_features", "pixel_values", "return", "return_dict", "self", "shape", "torch", "tuple", "width"], "depth_pro/modeling_depth_pro.py:DepthProFeatureUpsampleBlock.__init__": ["Conv2d", "ConvTranspose2d", "False", "ModelConfig", "ModelFeatureUpsampleBlock", "Module", "ModuleList", "True", "__init__", "append", "bias", "bool", "class", "config", "def", "else", "for", "i", "if", "in", "in_channels", "input_dims", "int", "intermediate_dims", "kernel_size", "layer", "layers", "n_upsample_layers", "nn", "out_channels", "output_dims", "padding", "proj", "range", "self", "stride", "super", "use_proj"], "depth_pro/modeling_depth_pro.py:DepthProFeatureUpsampleBlock.forward": ["ModelFeatureUpsampleBlock", "Module", "Tensor", "class", "def", "features", "for", "forward", "in", "layer", "layers", "nn", "return", "self", "torch"], "depth_pro/modeling_depth_pro.py:DepthProFeatureUpsample.__init__": ["False", "ModelConfig", "ModelFeatureUpsample", "ModelFeatureUpsampleBlock", "Module", "ModuleList", "True", "__init__", "append", "bias", "block", "class", "config", "def", "else", "enumerate", "feature_dims", "for", "fusion_hidden_size", "hidden_size", "i", "if", "image_block", "image_model_config", "in", "input_dims", "intermediate", "intermediate_dims", "intermediate_feature_dims", "intermediate_hook_ids", "len", "n_intermediate_hooks", "n_scaled_images", "n_upsample_layers", "nn", "output_dims", "patch_model_config", "scaled_images", "scaled_images_feature_dims", "scaled_images_ratios", "self", "super", "use_proj"], "depth_pro/modeling_depth_pro.py:DepthProFeatureUpsample.forward": ["ModelFeatureUpsample", "Module", "Tensor", "class", "def", "features", "for", "forward", "i", "image_block", "in", "intermediate", "list", "n_intermediate_hooks", "n_scaled_images", "nn", "range", "return", "scaled_images", "self", "torch"], "depth_pro/modeling_depth_pro.py:DepthProFeatureProjection.__init__": ["Conv2d", "False", "Identity", "ModelConfig", "ModelFeatureProjection", "Module", "ModuleList", "__init__", "and", "append", "bias", "class", "combined_feature_dims", "config", "def", "else", "enumerate", "for", "fusion_hidden_size", "i", "if", "in", "in_channels", "intermediate_feature_dims", "kernel_size", "len", "nn", "out_channels", "padding", "projections", "scaled_images_feature_dims", "self", "stride", "super"], "depth_pro/modeling_depth_pro.py:DepthProFeatureProjection.forward": ["ModelFeatureProjection", "Module", "Tensor", "append", "class", "def", "enumerate", "features", "for", "forward", "i", "in", "list", "nn", "projected_features", "projection", "projections", "return", "self", "torch", "upsampled_feature"], "depth_pro/modeling_depth_pro.py:DepthProNeck.__init__": ["Conv2d", "ModelConfig", "ModelFeatureProjection", "ModelFeatureUpsample", "ModelNeck", "Module", "True", "__init__", "bias", "class", "config", "def", "feature_projection", "feature_upsample", "fuse_image_with_low_res", "in_channels", "kernel_size", "nn", "out_channels", "padding", "scaled_images_feature_dims", "self", "stride", "super"], "depth_pro/modeling_depth_pro.py:DepthProNeck.forward": ["ModelNeck", "Module", "Tensor", "cat", "class", "def", "dim", "feature_projection", "feature_upsample", "features", "forward", "fuse_image_with_low_res", "global_features", "list", "nn", "return", "self", "torch"], "depth_pro/modeling_depth_pro.py:DepthProPreTrainedModel._init_weights": ["Conv2d", "ConvTranspose2d", "LayerNorm", "Linear", "ModelPreTrainedModel", "None", "PreTrainedModel", "_init_weights", "bias", "class", "config", "def", "elif", "fan_out", "if", "init", "initializer_range", "is", "isinstance", "kaiming_normal_", "mean", "mode", "module", "nn", "nonlinearity", "normal_", "not", "ones_", "relu", "self", "std", "weight", "zeros_"], "depth_pro/modeling_depth_pro.py:DepthProModel.__init__": ["ModelEncoder", "ModelModel", "ModelNeck", "ModelPreTrainedModel", "__init__", "class", "config", "def", "encoder", "neck", "post_init", "self", "super"], "depth_pro/modeling_depth_pro.py:DepthProModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "encoder", "get_input_embeddings", "image_encoder", "model", "return", "self"], "depth_pro/modeling_depth_pro.py:DepthProModel.forward": ["FloatTensor", "ModelModel", "ModelOutput", "ModelPreTrainedModel", "None", "attentions", "bool", "class", "config", "def", "else", "encoder", "encodings", "features", "forward", "hidden_states", "if", "is", "kwargs", "last_hidden_state", "neck", "not", "output_attentions", "output_hidden_states", "pixel_values", "r", "return", "return_dict", "self", "torch", "tuple", "use_return_dict"], "depth_pro/modeling_depth_pro.py:DepthProPreActResidualLayer.__init__": ["BatchNorm2d", "Conv2d", "ModelConfig", "ModelPreActResidualLayer", "Module", "None", "ReLU", "__init__", "activation1", "activation2", "batch_norm1", "batch_norm2", "bias", "class", "config", "convolution1", "convolution2", "def", "else", "fusion_hidden_size", "if", "is", "kernel_size", "nn", "not", "padding", "self", "stride", "super", "use_batch_norm", "use_batch_norm_in_fusion_residual", "use_bias_in_fusion_residual"], "depth_pro/modeling_depth_pro.py:DepthProPreActResidualLayer.forward": ["ModelPreActResidualLayer", "Module", "Tensor", "activation1", "activation2", "batch_norm1", "batch_norm2", "class", "convolution1", "convolution2", "def", "forward", "hidden_state", "if", "nn", "residual", "return", "self", "torch", "use_batch_norm"], "depth_pro/modeling_depth_pro.py:DepthProFeatureFusionLayer.__init__": ["Conv2d", "ConvTranspose2d", "False", "ModelConfig", "ModelFeatureFusionLayer", "ModelPreActResidualLayer", "Module", "True", "__init__", "bias", "bool", "class", "config", "deconv", "def", "fusion_hidden_size", "if", "in_channels", "kernel_size", "nn", "out_channels", "padding", "projection", "residual_layer1", "residual_layer2", "self", "stride", "super", "use_deconv"], "depth_pro/modeling_depth_pro.py:DepthProFeatureFusionLayer.forward": ["ModelFeatureFusionLayer", "Module", "None", "Tensor", "class", "deconv", "def", "forward", "hidden_state", "if", "is", "nn", "not", "projection", "residual", "residual_layer1", "residual_layer2", "return", "self", "torch", "use_deconv"], "depth_pro/modeling_depth_pro.py:DepthProFeatureFusionStage.__init__": ["False", "ModelFeatureFusionLayer", "ModelFeatureFusionStage", "Module", "ModuleList", "_", "__init__", "append", "class", "config", "def", "final", "for", "in", "intermediate", "intermediate_hook_ids", "len", "nn", "num_layers", "range", "scaled_images_ratios", "self", "super", "use_deconv"], "depth_pro/modeling_depth_pro.py:DepthProFeatureFusionStage.forward": ["ModelFeatureFusionStage", "Module", "None", "Tensor", "ValueError", "append", "class", "def", "does", "else", "f", "final", "for", "forward", "fused_hidden_state", "fused_hidden_states", "hidden_state", "hidden_states", "if", "in", "intermediate", "is", "layer", "len", "list", "match", "nn", "not", "num_layers", "raise", "return", "self", "torch", "zip"], "depth_pro/modeling_depth_pro.py:DepthProFovEncoder.__init__": ["AutoModel", "Linear", "ModelConfig", "ModelFovEncoder", "Module", "__init__", "class", "config", "def", "fov_model_config", "from_config", "fusion_hidden_size", "hidden_size", "image_model_config", "image_size", "model", "neck", "nn", "out_size", "patch_size", "self", "super"], "depth_pro/modeling_depth_pro.py:DepthProFovEncoder.forward": ["F", "False", "ModelFovEncoder", "Module", "Tensor", "align_corners", "base_height", "base_width", "batch_size", "bilinear", "class", "config", "def", "encodings", "exponent_value", "features", "forward", "fov_model_config", "height", "hidden_state", "image_size", "interpolate", "log2", "math", "mode", "model", "neck", "nn", "num_channels", "out_size", "output_size", "padding", "pixel_values", "reconstruct_feature_maps", "return", "self", "shape", "size", "torch", "torch_int", "width"], "depth_pro/modeling_depth_pro.py:DepthProFovHead.__init__": ["Conv2d", "ModelConfig", "ModelFovHead", "Module", "ModuleList", "ReLU", "True", "__init__", "append", "ceil", "class", "config", "def", "final_in_channels", "final_kernel_size", "for", "fusion_hidden_size", "i", "image_model_config", "image_size", "in", "in_channels", "kernel_size", "layers", "math", "nn", "num_fov_head_layers", "out_channels", "out_size", "padding", "patch_size", "range", "self", "stride", "super", "torch_int"], "depth_pro/modeling_depth_pro.py:DepthProFovHead.forward": ["F", "False", "ModelFovHead", "Module", "Tensor", "align_corners", "bilinear", "class", "def", "features", "for", "forward", "in", "interpolate", "layer", "layers", "mode", "nn", "out_size", "return", "self", "size", "torch"], "depth_pro/modeling_depth_pro.py:DepthProFovModel.__init__": ["Conv2d", "ModelConfig", "ModelFovEncoder", "ModelFovHead", "ModelFovModel", "Module", "ReLU", "True", "__init__", "activation", "class", "config", "conv", "def", "fov_encoder", "fusion_hidden_size", "head", "inplace", "kernel_size", "nn", "padding", "self", "stride", "super"], "depth_pro/modeling_depth_pro.py:DepthProFovModel.forward": ["ModelFovModel", "Module", "Tensor", "activation", "class", "conv", "def", "flatten", "forward", "fov_encoder", "fov_features", "fov_output", "global_features", "head", "nn", "pixel_values", "return", "self", "torch"], "depth_pro/modeling_depth_pro.py:DepthProDepthEstimationHead.__init__": ["Conv2d", "ConvTranspose2d", "ModelModelEstimationHead", "Module", "ModuleList", "ReLU", "True", "__init__", "bias", "class", "config", "def", "features", "fusion_hidden_size", "in_channels", "kernel_size", "layers", "nn", "out_channels", "padding", "self", "stride", "super"], "depth_pro/modeling_depth_pro.py:DepthProDepthEstimationHead.forward": ["ModelModelEstimationHead", "Module", "Tensor", "class", "def", "dim", "for", "forward", "hidden_states", "in", "layer", "layers", "nn", "predicted_Model", "return", "self", "squeeze", "torch"], "depth_pro/modeling_depth_pro.py:DepthProForDepthEstimation.__init__": ["Model", "ModelFeatureFusionStage", "ModelForModelEstimation", "ModelFovModel", "ModelModel", "ModelModelEstimationHead", "ModelPreTrainedModel", "None", "__init__", "class", "config", "def", "else", "fov_model", "fusion_stage", "head", "if", "is", "not", "post_init", "r", "self", "super", "use_fov_model"], "depth_pro/modeling_depth_pro.py:DepthProForDepthEstimation.forward": ["FloatTensor", "LongTensor", "Model", "ModelForModelEstimation", "ModelModelEstimatorOutput", "ModelPreTrainedModel", "Model_outputs", "None", "NotImplementedError", "Tensor", "Training", "True", "attentions", "bool", "class", "config", "def", "detach", "else", "features", "features_for_fov", "field_of_view", "for", "forward", "fov", "fov_model", "fused_hidden_states", "fusion_stage", "global_features", "head", "hidden_states", "if", "implemented", "in", "is", "kwargs", "labels", "loss", "not", "output_attentions", "output_hidden_states", "outputs", "pixel_values", "predicted_Model", "r", "raise", "return", "return_dict", "self", "torch", "tuple", "use_fov_model", "use_return_dict", "v", "yet"], "detr/modeling_detr.py:DetrFrozenBatchNorm2d.__init__": ["ModelFrozenBatchNorm2d", "Module", "__init__", "bias", "class", "def", "n", "nn", "ones", "register_buffer", "running_mean", "running_var", "self", "super", "torch", "weight", "zeros"], "detr/modeling_detr.py:DetrFrozenBatchNorm2d._load_from_state_dict": ["ModelFrozenBatchNorm2d", "Module", "_load_from_state_dict", "class", "def", "del", "error_msgs", "if", "in", "local_metadata", "missing_keys", "nn", "num_batches_tracked", "num_batches_tracked_key", "prefix", "self", "state_dict", "strict", "super", "unexpected_keys"], "detr/modeling_detr.py:DetrFrozenBatchNorm2d.forward": ["ModelFrozenBatchNorm2d", "Module", "bias", "class", "def", "epsilon", "forward", "nn", "reshape", "return", "rsqrt", "running_mean", "running_var", "scale", "self", "weight", "x"], "detr/modeling_detr.py:replace_batch_norm": ["BatchNorm2d", "ModelFrozenBatchNorm2d", "Model_batch_norm", "_modules", "bias", "children", "copy_", "def", "device", "for", "if", "in", "isinstance", "len", "list", "meta", "model", "module", "name", "named_children", "new_module", "nn", "num_features", "r", "running_mean", "running_var", "torch", "weight"], "detr/modeling_detr.py:DetrConvEncoder.__init__": ["Either", "False", "ModelConvEncoder", "Module", "None", "True", "ValueError", "__init__", "and", "backbone", "backbone_config", "backbone_kwargs", "backbone_model_type", "be", "channels", "class", "config", "copy", "create_model", "def", "dilation", "elif", "else", "feature_info", "features_only", "for", "get", "getattr", "if", "in", "in_chans", "intermediate_channel_sizes", "is", "kwargs", "layer2", "layer3", "layer4", "load_backbone", "model", "model_type", "name", "named_parameters", "nn", "no_grad", "not", "num_channels", "or", "out_indices", "output_stride", "parameter", "pop", "pretrained", "provided", "raise", "replace_batch_norm", "requires_backends", "requires_grad_", "resnet", "self", "should", "stage", "super", "the", "timm", "torch", "use_pretrained_backbone", "use_timm_backbone", "with"], "detr/modeling_detr.py:DetrConvEncoder.forward": ["ModelConvEncoder", "Module", "None", "Tensor", "append", "bool", "class", "config", "def", "else", "feature_map", "feature_maps", "features", "float", "for", "forward", "functional", "if", "in", "interpolate", "mask", "model", "nn", "out", "pixel_mask", "pixel_values", "return", "self", "shape", "size", "to", "torch", "use_timm_backbone"], "detr/modeling_detr.py:DetrConvModel.__init__": ["ModelConvModel", "Module", "__init__", "class", "conv_encoder", "def", "nn", "position_embedding", "self", "super"], "detr/modeling_detr.py:DetrConvModel.forward": ["ModelConvModel", "Module", "append", "class", "conv_encoder", "def", "dtype", "feature_map", "for", "forward", "in", "mask", "nn", "out", "pixel_mask", "pixel_values", "pos", "position_embedding", "return", "self", "to"], "detr/modeling_detr.py:DetrSinePositionEmbedding.__init__": ["False", "ModelSinePositionEmbedding", "Module", "None", "True", "ValueError", "__init__", "and", "be", "class", "def", "embedding_dim", "if", "is", "math", "nn", "normalize", "not", "passed", "pi", "raise", "scale", "self", "should", "super", "temperature"], "detr/modeling_detr.py:DetrSinePositionEmbedding.forward": ["ModelSinePositionEmbedding", "Module", "No", "None", "ValueError", "arange", "cat", "class", "cos", "cumsum", "def", "device", "dim", "dim_t", "div", "dtype", "embedding_dim", "flatten", "float", "float32", "floor", "forward", "if", "int64", "is", "mask", "nn", "normalize", "permute", "pixel", "pixel_mask", "pixel_values", "pos", "pos_x", "pos_y", "provided", "raise", "return", "rounding_mode", "scale", "self", "sin", "stack", "temperature", "torch", "x_embed", "y_embed"], "detr/modeling_detr.py:DetrLearnedPositionEmbedding.__init__": ["Embedding", "ModelLearnedPositionEmbedding", "Module", "__init__", "class", "column_embeddings", "def", "embedding_dim", "nn", "row_embeddings", "self", "super"], "detr/modeling_detr.py:DetrLearnedPositionEmbedding.forward": ["ModelLearnedPositionEmbedding", "Module", "None", "arange", "cat", "class", "column_embeddings", "def", "device", "dim", "forward", "height", "height_values", "nn", "permute", "pixel_mask", "pixel_values", "pos", "repeat", "return", "row_embeddings", "self", "shape", "torch", "unsqueeze", "width", "width_values", "x_emb", "y_emb"], "detr/modeling_detr.py:build_position_encoding": ["ModelLearnedPositionEmbedding", "ModelSinePositionEmbedding", "Model_position_encoding", "Not", "True", "ValueError", "config", "d_model", "def", "elif", "else", "f", "if", "learned", "n_steps", "normalize", "position_embedding", "position_embedding_type", "raise", "return", "sine", "supported"], "detr/modeling_detr.py:DetrAttention.__init__": ["Linear", "ModelAttention", "Module", "True", "ValueError", "__init__", "and", "be", "bias", "bool", "by", "class", "def", "divisible", "dropout", "embed_dim", "f", "float", "got", "head_dim", "if", "int", "k_proj", "must", "nn", "num_heads", "out_proj", "q_proj", "raise", "scaling", "self", "super", "v_proj"], "detr/modeling_detr.py:DetrAttention._shape": ["ModelAttention", "Module", "Tensor", "_shape", "batch_size", "class", "contiguous", "def", "head_dim", "int", "nn", "num_heads", "return", "self", "seq_len", "tensor", "torch", "transpose", "view"], "detr/modeling_detr.py:DetrAttention.with_pos_embed": ["ModelAttention", "Module", "None", "Tensor", "class", "def", "else", "if", "is", "nn", "object_queries", "return", "self", "tensor", "torch", "with_pos_embed"], "detr/modeling_detr.py:DetrAttention.forward": ["Attention", "False", "ModelAttention", "Module", "None", "Tensor", "ValueError", "_shape", "attention_mask", "attn_output", "attn_probs", "attn_weights", "attn_weights_reshaped", "batch_size", "be", "bmm", "bool", "but", "class", "def", "dim", "dropout", "dtype", "else", "embed_dim", "f", "forward", "functional", "head_dim", "hidden_states", "hidden_states_original", "if", "inf", "is", "is_cross_attention", "k_proj", "key_states", "key_value_states", "key_value_states_original", "mask", "masked_fill_", "nn", "not", "num_heads", "object_queries", "of", "out_proj", "output_attentions", "p", "proj_shape", "q_proj", "query_states", "raise", "reshape", "return", "scaling", "self", "should", "size", "softmax", "source_len", "spatial_position_embeddings", "target_len", "torch", "training", "transpose", "tuple", "v_proj", "value_states", "view", "weights", "with_pos_embed", "zeros_like"], "detr/modeling_detr.py:DetrEncoderLayer.__init__": ["ACT2FN", "LayerNorm", "Linear", "ModelAttention", "ModelConfig", "ModelEncoderLayer", "Module", "__init__", "activation_dropout", "activation_fn", "activation_function", "attention_dropout", "class", "config", "d_model", "def", "dropout", "embed_dim", "encoder_attention_heads", "encoder_ffn_dim", "fc1", "fc2", "final_layer_norm", "nn", "num_heads", "self", "self_attn", "self_attn_layer_norm", "super"], "detr/modeling_detr.py:DetrEncoderLayer.forward": ["False", "ModelEncoderLayer", "Module", "None", "Tensor", "activation_dropout", "activation_fn", "any", "attention_mask", "attn_weights", "bool", "clamp", "clamp_value", "class", "def", "dropout", "dtype", "fc1", "fc2", "final_layer_norm", "finfo", "forward", "functional", "hidden_states", "if", "isinf", "isnan", "max", "min", "nn", "object_queries", "or", "output_attentions", "outputs", "p", "residual", "return", "self", "self_attn", "self_attn_layer_norm", "torch", "training"], "detr/modeling_detr.py:DetrDecoderLayer.__init__": ["ACT2FN", "GradientCheckpointingLayer", "LayerNorm", "Linear", "ModelAttention", "ModelConfig", "ModelDecoderLayer", "__init__", "activation_dropout", "activation_fn", "activation_function", "attention_dropout", "class", "config", "d_model", "decoder_attention_heads", "decoder_ffn_dim", "def", "dropout", "embed_dim", "encoder_attn", "encoder_attn_layer_norm", "fc1", "fc2", "final_layer_norm", "nn", "num_heads", "self", "self_attn", "self_attn_layer_norm", "super"], "detr/modeling_detr.py:DetrDecoderLayer.forward": ["False", "GradientCheckpointingLayer", "ModelDecoderLayer", "None", "Tensor", "activation_dropout", "activation_fn", "attention_mask", "bool", "class", "cross_attn_weights", "def", "dropout", "encoder_attention_mask", "encoder_attn", "encoder_attn_layer_norm", "encoder_hidden_states", "fc1", "fc2", "final_layer_norm", "forward", "functional", "hidden_states", "if", "is", "key_value_states", "nn", "not", "object_queries", "output_attentions", "outputs", "p", "query_position_embeddings", "residual", "return", "self", "self_attn", "self_attn_layer_norm", "self_attn_weights", "spatial_position_embeddings", "torch", "training"], "detr/modeling_detr.py:DetrPreTrainedModel._init_weights": ["Conv2d", "Embedding", "False", "GroupNorm", "LayerNorm", "Linear", "ModelLearnedPositionEmbedding", "ModelMHAttentionMap", "ModelPreTrainedModel", "None", "PreTrainedModel", "_init_weights", "_is_hf_initialized", "and", "bias", "class", "column_embeddings", "config", "def", "elif", "gain", "getattr", "if", "init", "init_std", "init_xavier_std", "is", "isinstance", "k_linear", "mean", "module", "nn", "normal_", "not", "ones_", "padding_idx", "q_linear", "row_embeddings", "self", "std", "uniform_", "weight", "xavier_std", "xavier_uniform_", "zeros_"], "detr/modeling_detr.py:DetrEncoder.__init__": ["ModelConfig", "ModelEncoder", "ModelEncoderLayer", "ModelPreTrainedModel", "ModuleList", "_", "__init__", "class", "config", "def", "dropout", "encoder_layerdrop", "encoder_layers", "for", "in", "layerdrop", "layers", "nn", "post_init", "range", "self", "super"], "detr/modeling_detr.py:DetrEncoder.forward": ["BaseModelOutput", "False", "ModelEncoder", "ModelPreTrainedModel", "None", "True", "_prepare_4d_attention_mask", "all_attentions", "attention_mask", "attentions", "class", "config", "def", "dropout", "dropout_probability", "dtype", "else", "encoder_layer", "encoder_states", "enumerate", "for", "forward", "functional", "hidden_states", "i", "if", "in", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layer_outputs", "layerdrop", "layers", "nn", "not", "object_queries", "output_attentions", "output_hidden_states", "p", "r", "rand", "return", "return_dict", "self", "to_drop", "torch", "training", "tuple", "use_return_dict", "v"], "detr/modeling_detr.py:DetrDecoder.__init__": ["False", "LayerNorm", "ModelConfig", "ModelDecoder", "ModelDecoderLayer", "ModelPreTrainedModel", "ModuleList", "_", "__init__", "class", "config", "d_model", "decoder_layerdrop", "decoder_layers", "def", "dropout", "for", "gradient_checkpointing", "in", "layerdrop", "layernorm", "layers", "nn", "post_init", "range", "self", "super"], "detr/modeling_detr.py:DetrDecoder.forward": ["ModelDecoder", "ModelDecoderOutput", "ModelPreTrainedModel", "None", "_prepare_4d_attention_mask", "all_cross_attentions", "all_hidden_states", "all_self_attns", "and", "attention_mask", "attentions", "auxiliary_loss", "class", "combined_attention_mask", "config", "continue", "cross_attentions", "decoder_layer", "def", "dropout_probability", "dtype", "else", "encoder_attention_mask", "encoder_hidden_states", "enumerate", "for", "forward", "hidden_states", "idx", "if", "in", "input_shape", "inputs_embeds", "intermediate", "intermediate_hidden_states", "is", "kwargs", "last_hidden_state", "layer_outputs", "layerdrop", "layernorm", "layers", "not", "object_queries", "output_attentions", "output_hidden_states", "query_position_embeddings", "r", "rand", "return", "return_dict", "self", "size", "stack", "tgt_len", "torch", "training", "tuple", "use_return_dict", "v"], "detr/modeling_detr.py:DetrModel.__init__": ["Conv2d", "Embedding", "ModelConfig", "ModelConvEncoder", "ModelConvModel", "ModelDecoder", "ModelEncoder", "ModelModel", "ModelPreTrainedModel", "__init__", "backbone", "build_position_encoding", "class", "config", "d_model", "decoder", "def", "encoder", "input_projection", "intermediate_channel_sizes", "kernel_size", "nn", "num_queries", "object_queries", "post_init", "query_position_embeddings", "self", "super"], "detr/modeling_detr.py:DetrModel.freeze_backbone": ["False", "ModelModel", "ModelPreTrainedModel", "backbone", "class", "conv_encoder", "def", "for", "freeze_backbone", "in", "model", "name", "named_parameters", "param", "requires_grad_", "self"], "detr/modeling_detr.py:DetrModel.unfreeze_backbone": ["ModelModel", "ModelPreTrainedModel", "True", "backbone", "class", "conv_encoder", "def", "for", "in", "model", "name", "named_parameters", "param", "requires_grad_", "self", "unfreeze_backbone"], "detr/modeling_detr.py:DetrModel.forward": ["Backbone", "BaseModelOutput", "FloatTensor", "LongTensor", "ModelModel", "ModelModelOutput", "ModelPreTrainedModel", "None", "ValueError", "and", "attention_mask", "attentions", "backbone", "batch_size", "bool", "class", "config", "cross_attentions", "decoder", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_inputs_embeds", "decoder_outputs", "def", "device", "does", "downsampled", "elif", "else", "encoder", "encoder_attention_mask", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "feature_map", "features", "flatten", "flattened_features", "flattened_mask", "forward", "height", "hidden_states", "if", "input_projection", "inputs_embeds", "intermediate_hidden_states", "is", "isinstance", "kwargs", "last_hidden_state", "len", "mask", "not", "num_channels", "object_queries", "object_queries_list", "ones", "output_attentions", "output_hidden_states", "permute", "pixel", "pixel_mask", "pixel_values", "projected_feature_map", "queries", "query_position_embeddings", "r", "raise", "repeat", "return", "return_dict", "self", "shape", "torch", "tuple", "unsqueeze", "use_return_dict", "weight", "width", "zeros_like"], "detr/modeling_detr.py:DetrMLPPredictionHead.__init__": ["Linear", "ModelMLPPredictionHead", "Module", "ModuleList", "__init__", "class", "def", "for", "h", "hidden_dim", "in", "input_dim", "k", "layers", "n", "nn", "num_layers", "output_dim", "self", "super", "zip"], "detr/modeling_detr.py:DetrMLPPredictionHead.forward": ["ModelMLPPredictionHead", "Module", "class", "def", "else", "enumerate", "for", "forward", "functional", "i", "if", "in", "layer", "layers", "nn", "num_layers", "relu", "return", "self", "x"], "detr/modeling_detr.py:DetrForObjectDetection.__init__": ["Linear", "ModelConfig", "ModelForObjectDetection", "ModelMLPPredictionHead", "ModelModel", "ModelPreTrainedModel", "__init__", "bbox_predictor", "class", "class_labels_classifier", "config", "d_model", "def", "hidden_dim", "input_dim", "model", "nn", "num_labels", "num_layers", "output_dim", "post_init", "self", "super"], "detr/modeling_detr.py:DetrForObjectDetection.forward": ["FloatTensor", "LongTensor", "ModelForObjectDetection", "ModelObjectDetectionOutput", "ModelPreTrainedModel", "None", "auxiliary_loss", "auxiliary_outputs", "bbox_predictor", "bool", "class", "class_labels_classifier", "config", "cross_attentions", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_inputs_embeds", "def", "device", "dict", "else", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "forward", "if", "inputs_embeds", "intermediate", "intermediate_hidden_states", "is", "kwargs", "labels", "last_hidden_state", "list", "logits", "loss", "loss_dict", "loss_function", "model", "not", "output", "output_attentions", "output_hidden_states", "outputs", "outputs_class", "outputs_coord", "pixel_mask", "pixel_values", "pred_boxes", "r", "return", "return_dict", "self", "sequence_output", "sigmoid", "torch", "tuple", "use_return_dict"], "detr/modeling_detr.py:DetrForSegmentation.__init__": ["Model", "ModelConfig", "ModelForObjectDetection", "ModelForSegmentation", "ModelMHAttentionMap", "ModelMaskHeadSmallConv", "ModelPreTrainedModel", "__init__", "backbone", "bbox_attention", "class", "config", "conv_encoder", "d_model", "def", "dropout", "encoder_attention_heads", "hidden_size", "init_xavier_std", "intermediate_channel_sizes", "mask_head", "model", "number_of_heads", "post_init", "self", "std", "super"], "detr/modeling_detr.py:DetrForSegmentation.forward": ["BaseModelOutput", "FloatTensor", "LongTensor", "Model", "ModelForSegmentation", "ModelPreTrainedModel", "ModelSegmentationOutput", "None", "and", "attention_mask", "attentions", "auxiliary_loss", "auxiliary_outputs", "backbone", "batch_size", "bbox_attention", "bbox_mask", "bbox_predictor", "bool", "class", "class_labels_classifier", "config", "cross_attentions", "d_model", "decoder", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_inputs_embeds", "decoder_outputs", "def", "device", "dict", "elif", "else", "encoder", "encoder_attention_mask", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "feature_map", "features", "flatten", "flattened_features", "flattened_mask", "forward", "height", "hidden_states", "if", "input_projection", "inputs_embeds", "intermediate", "intermediate_hidden_states", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "len", "list", "logits", "loss", "loss_dict", "loss_function", "mask", "mask_head", "memory", "model", "not", "num_channels", "num_queries", "object_queries", "object_queries_list", "ones", "output", "output_attentions", "output_hidden_states", "outputs_class", "outputs_coord", "permute", "pixel_mask", "pixel_values", "pred_boxes", "pred_masks", "projected_feature_map", "queries", "query_position_embeddings", "r", "repeat", "return", "return_dict", "seg_masks", "self", "sequence_output", "shape", "sigmoid", "torch", "tuple", "unsqueeze", "use_return_dict", "view", "weight", "width", "zeros_like"], "detr/modeling_detr.py:_expand": ["_expand", "def", "flatten", "int", "length", "repeat", "return", "tensor", "unsqueeze"], "detr/modeling_detr.py:DetrMaskHeadSmallConv.__init__": ["Conv2d", "GroupNorm", "ModelMaskHeadSmallConv", "Module", "The", "ValueError", "__init__", "a", "adapter1", "adapter2", "adapter3", "as", "attention", "be", "bias", "by", "class", "constant_", "context_dim", "def", "dim", "divisible", "for", "fpn_dims", "gn1", "gn2", "gn3", "gn4", "gn5", "groups", "heads", "hidden_size", "if", "in", "init", "inter_dims", "is", "isinstance", "kaiming_uniform_", "lay1", "lay2", "lay3", "lay4", "lay5", "m", "min", "modules", "must", "nn", "number", "of", "out_lay", "padding", "raise", "self", "set", "super", "the", "to", "weight"], "detr/modeling_detr.py:DetrMaskHeadSmallConv.forward": ["ModelMaskHeadSmallConv", "Module", "Tensor", "_expand", "adapter1", "adapter2", "adapter3", "bbox_mask", "cat", "class", "cur_fpn", "def", "flatten", "forward", "fpns", "functional", "gn1", "gn2", "gn3", "gn4", "gn5", "if", "interpolate", "lay1", "lay2", "lay3", "lay4", "lay5", "list", "mode", "nearest", "nn", "out_lay", "relu", "return", "self", "shape", "size", "torch", "x"], "detr/modeling_detr.py:DetrMHAttentionMap.__init__": ["Dropout", "Linear", "ModelMHAttentionMap", "Module", "None", "True", "__init__", "bias", "class", "def", "dropout", "float", "hidden_dim", "k_linear", "nn", "normalize_fact", "num_heads", "q_linear", "query_dim", "self", "std", "super"], "detr/modeling_detr.py:DetrMHAttentionMap.forward": ["ModelMHAttentionMap", "Module", "None", "Tensor", "bias", "bnchw", "bqnc", "bqnhw", "class", "conv2d", "def", "dim", "dropout", "dtype", "einsum", "finfo", "flatten", "forward", "functional", "hidden_dim", "if", "is", "k", "k_linear", "keys_per_head", "mask", "masked_fill", "min", "nn", "normalize_fact", "not", "num_heads", "q", "q_linear", "queries_per_head", "return", "self", "shape", "size", "softmax", "torch", "unsqueeze", "view", "weight", "weights"], "dia/modeling_dia.py:DiaPreTrainedModel._init_weights": ["ModelMultiChannelEmbedding", "ModelPreTrainedModel", "PreTrainedModel", "_init_weights", "arange", "class", "config", "copy_", "def", "dtype", "if", "init", "isinstance", "long", "module", "num_channels", "offsets", "self", "super", "torch", "vocab_size"], "dia/modeling_dia.py:DiaMultiChannelEmbedding.__init__": ["Embedding", "False", "ModelDecoderConfig", "ModelMultiChannelEmbedding", "Module", "__init__", "arange", "class", "config", "def", "dtype", "embed", "hidden_size", "long", "nn", "num_channels", "offsets", "persistent", "register_buffer", "self", "super", "torch", "vocab_size"], "dia/modeling_dia.py:DiaMultiChannelEmbedding.forward": ["ModelMultiChannelEmbedding", "Module", "Tensor", "audio_codes", "class", "def", "device", "dim", "embed", "embeds", "forward", "hidden_size", "nn", "offsets", "return", "self", "shape", "squeeze", "sum", "to", "tokens", "torch", "view"], "dia/modeling_dia.py:DiaMLP.__init__": ["ACT2FN", "False", "Linear", "ModelMLP", "Module", "__init__", "activation_fn", "bias", "class", "config", "def", "down_proj", "gate_up_proj", "hidden_act", "hidden_size", "intermeModelte_size", "nn", "self", "super"], "dia/modeling_dia.py:DiaMLP.forward": ["FloatTensor", "ModelMLP", "Module", "activation_fn", "chunk", "class", "def", "dim", "down_proj", "forward", "gate", "gate_up_proj", "hidden_states", "nn", "return", "self", "torch", "up_states"], "dia/modeling_dia.py:DiaRMSNorm.__init__": ["ModelRMSNorm", "Module", "Parameter", "__init__", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "dia/modeling_dia.py:DiaRMSNorm.forward": ["ModelRMSNorm", "Module", "True", "class", "def", "dtype", "float32", "forward", "hidden_states", "input_dtype", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "dia/modeling_dia.py:DiaRMSNorm.extra_repr": ["ModelRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "variance_epsilon", "weight"], "dia/modeling_dia.py:DiaRotaryEmbedding.__init__": ["Callable", "False", "ModelConfig", "ModelRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "dia/modeling_dia.py:DiaRotaryEmbedding.compute_default_rope_parameters": ["ModelConfig", "ModelRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "dia/modeling_dia.py:DiaRotaryEmbedding.forward": ["False", "ModelRotaryEmbedding", "Module", "None", "and", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "dia/modeling_dia.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "dia/modeling_dia.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cos", "def", "k", "k_embed", "q", "q_embed", "return", "rotate_half", "sin", "unsqueeze", "unsqueeze_dim"], "dia/modeling_dia.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "dia/modeling_dia.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value", "value_states"], "dia/modeling_dia.py:DiaSelfAttention.__init__": ["False", "Linear", "ModelDecoderConfig", "ModelEncoderConfig", "ModelSelfAttention", "Module", "__init__", "attention_dropout", "bias", "bool", "class", "config", "def", "getattr", "head_dim", "hidden_size", "int", "is_causal", "k_proj", "layer_idx", "nn", "num_attention_heads", "num_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "or", "q_proj", "scaling", "self", "super", "v_proj"], "dia/modeling_dia.py:DiaSelfAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "LongTensor", "ModelSelfAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_attn_implementation", "apply_rotary_pos_emb", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "cache_kwargs", "cache_position", "class", "config", "contiguous", "cos", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "head_dim", "hidden_shape", "hidden_states", "if", "input_shape", "is", "k_proj", "key_states", "kwargs", "layer_idx", "nn", "not", "o_proj", "past_key_values", "position_embeddings", "q_proj", "query_states", "reshape", "return", "scaling", "self", "shape", "sin", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "view"], "dia/modeling_dia.py:DiaCrossAttention.__init__": ["False", "Linear", "ModelCrossAttention", "ModelDecoderConfig", "Module", "__init__", "attention_dropout", "bias", "class", "config", "cross_head_dim", "cross_hidden_size", "cross_num_attention_heads", "cross_num_key_value_heads", "def", "head_dim", "hidden_size", "int", "is_causal", "k_proj", "layer_idx", "nn", "num_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "q_proj", "scaling", "self", "super", "v_proj"], "dia/modeling_dia.py:DiaCrossAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "EncoderDecoderCache", "False", "FlashAttentionKwargs", "ModelCrossAttention", "Module", "None", "Tensor", "True", "Unpack", "_attn_implementation", "and", "attention_interface", "attention_mask", "attn_output", "attn_weights", "class", "config", "contiguous", "cross_attention_cache", "cross_attention_states", "cross_shape", "def", "eager", "eager_attention_forward", "else", "forward", "get", "head_dim", "hidden_shape", "hidden_states", "if", "input_shape", "is", "is_updated", "k_proj", "key_states", "keys", "kwargs", "layer_idx", "layers", "nn", "not", "o_proj", "past_key_values", "q_proj", "query_states", "reshape", "return", "scaling", "self", "shape", "torch", "transpose", "tuple", "update", "v_proj", "value_states", "values", "view"], "dia/modeling_dia.py:DiaEncoderLayer.__init__": ["False", "GradientCheckpointingLayer", "ModelEncoderConfig", "ModelEncoderLayer", "ModelMLP", "ModelRMSNorm", "ModelSelfAttention", "__init__", "class", "config", "def", "eps", "hidden_size", "int", "is_causal", "layer_idx", "mlp", "norm_eps", "post_sa_norm", "pre_sa_norm", "self", "self_attention", "super"], "dia/modeling_dia.py:DiaEncoderLayer.forward": ["FlashAttentionKwargs", "GradientCheckpointingLayer", "ModelEncoderLayer", "None", "Tensor", "Unpack", "attention_mask", "class", "def", "forward", "hidden_states", "kwargs", "mlp", "mlp_out", "normed_states", "position_embeddings", "post_sa_norm", "pre_sa_norm", "residual", "return", "self", "self_attention", "self_attn_output", "self_attn_weights", "torch", "tuple"], "dia/modeling_dia.py:DiaEncoder.__init__": ["Embedding", "ModelEncoder", "ModelEncoderConfig", "ModelEncoderLayer", "ModelPreTrainedModel", "ModelRMSNorm", "ModelRotaryEmbedding", "ModuleList", "__init__", "class", "config", "def", "embedding", "eps", "for", "hidden_size", "in", "layer_idx", "layers", "nn", "norm", "norm_eps", "num_hidden_layers", "post_init", "range", "rotary_emb", "self", "super", "vocab_size"], "dia/modeling_dia.py:DiaEncoder.forward": ["BaseModelOutput", "False", "FlashAttentionKwargs", "ModelEncoder", "ModelPreTrainedModel", "None", "Tensor", "Unpack", "all_attentions", "arange", "attention_mask", "attentions", "bool", "class", "config", "create_bidirectional_mask", "def", "device", "else", "embedding", "encoder_layer", "encoder_states", "for", "forward", "hidden_states", "if", "in", "input_embeds", "input_ids", "kwargs", "last_hidden_state", "layer_outputs", "layers", "norm", "output_attentions", "output_hidden_states", "position_embeddings", "position_ids", "return", "rotary_emb", "self", "shape", "torch", "tuple"], "dia/modeling_dia.py:DiaDecoderLayer.__init__": ["GradientCheckpointingLayer", "ModelCrossAttention", "ModelDecoderConfig", "ModelDecoderLayer", "ModelMLP", "ModelRMSNorm", "ModelSelfAttention", "True", "__init__", "class", "config", "cross_attention", "def", "embed_dim", "eps", "hidden_size", "int", "is_causal", "layer_idx", "mlp", "norm_eps", "pre_ca_norm", "pre_mlp_norm", "pre_sa_norm", "self", "self_attention", "super"], "dia/modeling_dia.py:DiaDecoderLayer.forward": ["EncoderDecoderCache", "GradientCheckpointingLayer", "LongTensor", "ModelDecoderLayer", "None", "Tensor", "attention_mask", "cache_position", "class", "cross_attention", "cross_attn_weights", "cross_states", "def", "encoder_attention_mask", "encoder_hidden_states", "forward", "hidden_states", "if", "isinstance", "kwargs", "mlp", "mlp_out", "normed_states", "past_key_values", "position_embeddings", "pre_ca_norm", "pre_mlp_norm", "pre_sa_norm", "residual", "return", "self", "self_attention", "self_attention_cache", "self_attn_cache", "self_attn_output", "self_attn_weights", "torch", "tuple"], "dia/modeling_dia.py:DiaDecoder.__init__": ["ModelDecoder", "ModelDecoderConfig", "ModelDecoderLayer", "ModelMultiChannelEmbedding", "ModelPreTrainedModel", "ModelRMSNorm", "ModelRotaryEmbedding", "ModuleList", "__init__", "class", "config", "def", "embeddings", "eps", "for", "hidden_size", "in", "layer_idx", "layers", "nn", "norm", "norm_eps", "num_channels", "num_hidden_layers", "post_init", "range", "rotary_emb", "self", "super", "vocab_size"], "dia/modeling_dia.py:DiaDecoder.forward": ["BaseModelOutputWithPastAndCrossAttentions", "EncoderDecoderCache", "False", "FloatTensor", "LongTensor", "ModelDecoder", "ModelPreTrainedModel", "None", "Tensor", "all_cross_attentions", "all_hidden_states", "all_self_attns", "and", "arange", "attention_mask", "attentions", "batch_size", "bool", "cache_position", "class", "config", "create_bidirectional_mask", "create_causal_mask", "cross_attentions", "def", "device", "else", "embeddings", "encoder_attention_mask", "encoder_hidden_states", "for", "forward", "get_seq_length", "hidden_states", "if", "in", "input_embeds", "input_ids", "is", "is_torchdynamo_compiling", "kwargs", "last_hidden_state", "layer", "layer_outputs", "layers", "mask_seq_length", "norm", "not", "ones", "output_attentions", "output_hidden_states", "past_key_values", "past_key_values_length", "position_embeddings", "position_ids", "r", "return", "rotary_emb", "self", "seq_length", "size", "torch", "tuple"], "dia/modeling_dia.py:DiaModel.__init__": ["ModelConfig", "ModelDecoder", "ModelEncoder", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "config", "decoder", "decoder_config", "def", "encoder", "encoder_config", "post_init", "self", "super"], "dia/modeling_dia.py:DiaModel.forward": ["BaseModelOutput", "DynamicCache", "EncoderDecoderCache", "False", "LongTensor", "ModelModel", "ModelPreTrainedModel", "Neither", "None", "Seq2SeqModelOutput", "Setting", "True", "ValueError", "You", "and", "attention_mask", "attentions", "been", "bool", "bos_token_id", "bsz", "cache_position", "cached", "channels", "checkpointing", "class", "config", "cross_attentions", "decoder", "decoder_attention_mask", "decoder_attentions", "decoder_config", "decoder_hidden_states", "decoder_input_ids", "decoder_outputs", "decoder_position_ids", "def", "device", "either", "elif", "else", "encoder", "encoder_attention_mask", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "encodings", "fill_value", "forward", "found", "full", "gradient", "has", "hidden_states", "ids", "if", "incompatible", "input_ids", "is", "is_gradient_checkpointing", "isinstance", "kwargs", "last_hidden_state", "len", "logger", "ndim", "not", "num_channels", "or", "output_attentions", "output_hidden_states", "past_key_values", "position_ids", "provide", "r", "raise", "reshape", "return", "self", "seq_len", "shape", "should", "size", "text", "the", "torch", "training", "transpose", "tuple", "use_cache", "warning_once", "with"], "dia/modeling_dia.py:DiaForConditionalGeneration.__init__": ["False", "ForMaskedLM", "Linear", "ModelConfig", "ModelForConditionalGeneration", "ModelGenerationMixin", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "decoder_config", "def", "hidden_size", "logits_dense", "loss_type", "model", "nn", "num_channels", "post_init", "self", "super", "vocab_size"], "dia/modeling_dia.py:DiaForConditionalGeneration.forward": ["BaseModelOutput", "EncoderDecoderCache", "LongTensor", "ModelForConditionalGeneration", "ModelGenerationMixin", "ModelPreTrainedModel", "None", "Seq2SeqLMOutput", "attention_mask", "audio_logits", "batch_size", "bool", "cache_position", "class", "contiguous", "cross_attentions", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_input_ids", "decoder_position_ids", "def", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "forward", "if", "input_ids", "is", "kwargs", "labels", "last_hidden_state", "logits", "logits_dense", "loss", "loss_function", "model", "not", "num_channels", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "r", "return", "self", "shape", "torch", "transpose", "tuple", "use_cache", "view", "vocab_size"], "diffllama/modeling_diffllama.py:DiffLlamaMLP.__init__": ["ACT2FN", "False", "Linear", "ModelMLP", "Module", "__init__", "act_fn", "bias", "class", "config", "def", "down_proj", "gate_proj", "hidden_act", "hidden_size", "intermediate_size", "nn", "self", "super", "up_proj"], "diffllama/modeling_diffllama.py:DiffLlamaMLP.forward": ["ModelMLP", "Module", "act_fn", "class", "def", "down_proj", "forward", "gate_proj", "nn", "return", "self", "up_proj", "x"], "diffllama/modeling_diffllama.py:DiffLlamaRotaryEmbedding.__init__": ["Callable", "False", "ModelConfig", "ModelRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "diffllama/modeling_diffllama.py:DiffLlamaRotaryEmbedding.compute_default_rope_parameters": ["ModelConfig", "ModelRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "diffllama/modeling_diffllama.py:DiffLlamaRotaryEmbedding.forward": ["False", "ModelRotaryEmbedding", "Module", "None", "and", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "diffllama/modeling_diffllama.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "diffllama/modeling_diffllama.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cos", "def", "k", "k_embed", "q", "q_embed", "return", "rotate_half", "sin", "unsqueeze", "unsqueeze_dim"], "diffllama/modeling_diffllama.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "diffllama/modeling_diffllama.py:lambda_init_fn": ["Model_init_fn", "def", "exp", "layer_idx", "math", "return"], "diffllama/modeling_diffllama.py:DiffLlamaAttention.__init__": ["False", "Instantiating", "Linear", "ModelAttention", "ModelConfig", "Module", "None", "Parameter", "Please", "RMSNorm", "True", "__class__", "__init__", "__name__", "a", "and", "attention_bias", "attention_dropout", "bias", "caching", "call", "class", "config", "creating", "def", "during", "elementwise_affine", "eps", "errors", "f", "forward", "getattr", "groupnorm", "head_dim", "hidden_size", "if", "int", "is", "is_causal", "k_proj", "lambda_init", "lambda_init_fn", "lambda_k1", "lambda_k2", "lambda_q1", "lambda_q2", "lambda_std_dev", "layer_idx", "lead", "logger", "make", "max_position_embeddings", "nn", "normal", "not", "num_attention_heads", "num_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "passing", "provide", "q_proj", "recommended", "rms_norm_eps", "self", "size", "super", "sure", "the", "this", "to", "torch", "used", "v_proj", "warning_once", "when", "will", "without"], "diffllama/modeling_diffllama.py:DiffLlamaAttention.forward": ["Cache", "False", "LongTensor", "ModelAttention", "Module", "None", "Tensor", "_", "apply_rotary_pos_emb", "attention_dropout", "attention_mask", "attn_output", "attn_output1", "attn_output2", "attn_weights", "bool", "bsz", "cache_kwargs", "cache_position", "cat", "causal_mask", "chunk", "class", "contiguous", "cos", "def", "dim", "dropout", "dtype", "exp", "float32", "forward", "functional", "groupnorm", "head_dim", "hidden_states", "if", "is", "k_proj", "key_states", "kwargs", "lambda_1", "lambda_2", "lambda_full", "lambda_init", "lambda_k1", "lambda_k2", "lambda_q1", "lambda_q2", "layer_idx", "math", "matmul", "nn", "not", "num_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "p", "past_key_values", "position_embeddings", "position_ids", "q_len", "q_proj", "query_states", "repeat", "repeat_kv", "reshape", "return", "self", "shape", "sin", "size", "softmax", "sqrt", "sum", "target_len", "to", "torch", "training", "transpose", "tuple", "update", "use_cache", "v_proj", "value_states", "view"], "diffllama/modeling_diffllama.py:DiffLlamaFlashAttention2.__init__": ["ModelAttention", "ModelFlashAttention2", "__init__", "_flash_attn_uses_top_left_mask", "args", "class", "def", "flash_attn_supports_top_left_mask", "kwargs", "self", "super"], "diffllama/modeling_diffllama.py:DiffLlamaFlashAttention2.forward": ["Cache", "False", "LongTensor", "ModelAttention", "ModelFlashAttention2", "None", "StaticCache", "Tensor", "The", "ValueError", "We", "_", "_flash_attention_forward", "_flash_attn_uses_top_left_mask", "_is_quantized", "an", "and", "apply_rotary_pos_emb", "at", "attention_dropout", "attention_mask", "attn_implementation", "attn_output", "attn_output1", "attn_output2", "back", "be", "bool", "bsz", "cache", "cache_kwargs", "cache_position", "cast", "casted", "cat", "chunk", "class", "com", "compatible", "config", "contiguous", "cos", "cpu", "def", "device", "device_type", "dim", "dropout", "dropout_rate", "dtype", "elif", "else", "embedding", "exp", "f", "fact", "flash_attention_2", "float32", "forward", "get_autocast_dtype", "get_autocast_gpu_dtype", "getattr", "github", "groupnorm", "hasattr", "have", "head_dim", "hidden", "hidden_states", "https", "huggingface", "if", "implementation", "in", "input", "input_dtype", "is", "is_autocast_enabled", "is_causal", "isinstance", "issue", "k_proj", "key_states", "lambda_1", "lambda_2", "lambda_full", "lambda_init", "lambda_k1", "lambda_k2", "lambda_q1", "lambda_q2", "layer", "layer_idx", "layers", "logger", "make", "mean", "might", "mps", "norm", "not", "num_heads", "num_key_value_heads", "o_proj", "open", "or", "past_key_values", "position_embeddings", "position_ids", "q_len", "q_proj", "query_states", "raise", "related", "repeat", "reshape", "return", "sdpa", "seems", "self", "silently", "sin", "size", "sliding_window", "states", "static", "sum", "sure", "target_dtype", "the", "this", "time", "to", "torch", "training", "transformers", "transpose", "tuple", "type", "upcasted", "update", "use", "use_cache", "use_top_left_mask", "v_proj", "value_states", "value_states1", "value_states2", "view", "warning_once", "weight", "will", "with", "you"], "diffllama/modeling_diffllama.py:DiffLlamaSdpaAttention.forward": ["Cache", "False", "LongTensor", "ModelAttention", "ModelSdpaAttention", "None", "Tensor", "_", "and", "apply_rotary_pos_emb", "attention_dropout", "attention_mask", "attn_mask", "attn_output", "attn_output1", "attn_output2", "bool", "bsz", "cache_kwargs", "cache_position", "cat", "causal_mask", "chunk", "class", "contiguous", "cos", "cuda", "def", "device", "dim", "dropout_p", "dtype", "else", "exp", "float32", "forward", "functional", "groupnorm", "head_dim", "hidden_states", "if", "is", "is_causal", "k_proj", "key_states", "kwargs", "lambda_1", "lambda_2", "lambda_full", "lambda_init", "lambda_k1", "lambda_k2", "lambda_q1", "lambda_q2", "layer_idx", "nn", "not", "num_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "past_key_values", "position_embeddings", "position_ids", "q_len", "q_proj", "query_states", "repeat", "repeat_kv", "return", "scaled_dot_product_attention", "self", "shape", "sin", "size", "sum", "to", "torch", "training", "transpose", "tuple", "type", "update", "use_cache", "v_proj", "value_states", "view"], "diffllama/modeling_diffllama.py:DiffLlamaRMSNorm.__init__": ["ModelRMSNorm", "Module", "Parameter", "__init__", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "diffllama/modeling_diffllama.py:DiffLlamaRMSNorm.forward": ["ModelRMSNorm", "Module", "True", "class", "def", "dtype", "float32", "forward", "hidden_states", "input_dtype", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "diffllama/modeling_diffllama.py:DiffLlamaRMSNorm.extra_repr": ["ModelRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "variance_epsilon", "weight"], "diffllama/modeling_diffllama.py:DiffLlamaDecoderLayer.__init__": ["GradientCheckpointingLayer", "ModelConfig", "ModelDecoderLayer", "ModelMLP", "ModelRMSNorm", "Model_ATTENTION_CLASSES", "__init__", "_attn_implementation", "class", "config", "def", "eps", "hidden_size", "input_layernorm", "int", "layer_idx", "mlp", "post_attention_layernorm", "rms_norm_eps", "self", "self_attn", "super"], "diffllama/modeling_diffllama.py:DiffLlamaDecoderLayer.forward": ["Cache", "False", "GradientCheckpointingLayer", "LongTensor", "ModelDecoderLayer", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "attention_mask", "bool", "cache_position", "class", "def", "forward", "hidden_states", "input_layernorm", "kwargs", "mlp", "past_key_values", "position_embeddings", "position_ids", "post_attention_layernorm", "residual", "return", "self", "self_attn", "torch", "tuple", "use_cache"], "diffllama/modeling_diffllama.py:DiffLlamaPreTrainedModel._init_weights": ["ModelAttention", "ModelPreTrainedModel", "PreTrainedModel", "_init_weights", "class", "config", "def", "if", "init", "isinstance", "lambda_k1", "lambda_k2", "lambda_q1", "lambda_q2", "lambda_std_dev", "module", "normal_", "self", "super"], "diffllama/modeling_diffllama.py:DiffLlamaModel.__init__": ["Embedding", "False", "ModelConfig", "ModelDecoderLayer", "ModelModel", "ModelPreTrainedModel", "ModelRMSNorm", "ModelRotaryEmbedding", "ModuleList", "__init__", "class", "config", "def", "embed_tokens", "eps", "for", "gradient_checkpointing", "hidden_size", "in", "layer_idx", "layers", "nn", "norm", "num_hidden_layers", "pad_token_id", "padding_idx", "post_init", "range", "rms_norm_eps", "rotary_emb", "self", "super", "vocab_size"], "diffllama/modeling_diffllama.py:DiffLlamaModel.forward": ["BaseModelOutputWithPast", "Cache", "DynamicCache", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "and", "arange", "attention_mask", "bool", "cache_position", "causal_mask", "class", "config", "create_causal_mask", "decoder_layer", "def", "device", "else", "embed_tokens", "exactly", "for", "forward", "get_seq_length", "hidden_states", "if", "in", "input_embeds", "input_ids", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layers", "must", "norm", "not", "num_hidden_layers", "of", "one", "or", "past_key_values", "past_seen_tokens", "position_embeddings", "position_ids", "raise", "return", "rotary_emb", "self", "shape", "specify", "torch", "unsqueeze", "use_cache"], "diffllama/modeling_diffllama.py:DiffLlamaForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "ModelForCausalLM", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "vocab_size"], "diffllama/modeling_diffllama.py:DiffLlamaForCausalLM.forward": ["BaseModelOutputWithPast", "Cache", "CausalLMOutputWithPast", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForCausalLM", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "outputs", "past_key_values", "position_ids", "r", "return", "self", "slice", "slice_indices", "torch", "use_cache", "vocab_size"], "dinat/modeling_dinat.py:DinatEmbeddings.__init__": ["Dropout", "LayerNorm", "ModelEmbeddings", "ModelPatchEmbeddings", "Module", "__init__", "class", "config", "def", "dropout", "embed_dim", "hidden_dropout_prob", "nn", "norm", "patch_embeddings", "self", "super"], "dinat/modeling_dinat.py:DinatEmbeddings.forward": ["FloatTensor", "ModelEmbeddings", "Module", "None", "Tensor", "class", "def", "dropout", "embeddings", "forward", "nn", "norm", "patch_embeddings", "pixel_values", "return", "self", "torch", "tuple"], "dinat/modeling_dinat.py:DinatPatchEmbeddings.__init__": ["Conv2d", "Model", "ModelPatchEmbeddings", "Module", "Sequential", "ValueError", "__init__", "at", "class", "config", "def", "else", "embed_dim", "hidden_size", "if", "kernel_size", "moment", "nn", "num_channels", "of", "only", "padding", "pass", "patch", "patch_size", "projection", "raise", "self", "size", "stride", "super", "supports", "the"], "dinat/modeling_dinat.py:DinatPatchEmbeddings.forward": ["FloatTensor", "Make", "ModelPatchEmbeddings", "Module", "None", "Tensor", "ValueError", "_", "channel", "class", "configuration", "def", "dimension", "embeddings", "forward", "height", "if", "in", "match", "nn", "num_channels", "of", "one", "permute", "pixel", "pixel_values", "projection", "raise", "return", "self", "set", "shape", "sure", "that", "the", "torch", "values", "width", "with"], "dinat/modeling_dinat.py:DinatDownsampler.__init__": ["Conv2d", "False", "LayerNorm", "ModelDownsampler", "Module", "None", "__init__", "bias", "class", "def", "dim", "int", "kernel_size", "nn", "norm", "norm_layer", "padding", "reduction", "self", "stride", "super"], "dinat/modeling_dinat.py:DinatDownsampler.forward": ["ModelDownsampler", "Module", "Tensor", "class", "def", "forward", "input_feature", "nn", "norm", "permute", "reduction", "return", "self", "torch"], "dinat/modeling_dinat.py:drop_path": ["False", "Model_path", "Model_prob", "Tensor", "bool", "def", "device", "div", "dtype", "float", "floor_", "if", "input", "keep_prob", "ndim", "not", "or", "output", "rand", "random_tensor", "return", "shape", "torch", "training"], "dinat/modeling_dinat.py:DinatDropPath.__init__": ["ModelDropPath", "Module", "None", "__init__", "class", "def", "drop_prob", "float", "nn", "self", "super"], "dinat/modeling_dinat.py:DinatDropPath.forward": ["ModelDropPath", "Module", "Tensor", "class", "def", "drop_path", "drop_prob", "forward", "hidden_states", "nn", "return", "self", "torch", "training"], "dinat/modeling_dinat.py:DinatDropPath.extra_repr": ["ModelDropPath", "Module", "class", "def", "drop_prob", "extra_repr", "f", "nn", "p", "return", "self", "str"], "dinat/modeling_dinat.py:NeighborhoodAttention.__init__": ["Dropout", "Linear", "ModelAttention", "Module", "Parameter", "The", "ValueError", "__init__", "a", "all_head_size", "attention", "attention_head_size", "attention_probs_dropout_prob", "bias", "class", "config", "def", "dilation", "dim", "dropout", "f", "heads", "hidden", "if", "int", "is", "kernel_size", "key", "multiple", "nn", "not", "num_attention_heads", "num_heads", "number", "of", "qkv_bias", "query", "raise", "rpb", "self", "size", "super", "the", "torch", "value", "zeros"], "dinat/modeling_dinat.py:NeighborhoodAttention.forward": ["False", "ModelAttention", "Module", "None", "Tensor", "_", "all_head_size", "attention_head_size", "attention_probs", "attention_scores", "batch_size", "bool", "class", "context_layer", "contiguous", "def", "dilation", "dim", "dropout", "else", "forward", "functional", "hidden_states", "if", "kernel_size", "key", "key_layer", "math", "natten2dav", "natten2dqkrpb", "new_context_layer_shape", "nn", "num_attention_heads", "output_attentions", "outputs", "permute", "query", "query_layer", "return", "rpb", "self", "seq_length", "shape", "size", "softmax", "sqrt", "torch", "transpose", "tuple", "value", "value_layer", "view"], "dinat/modeling_dinat.py:NeighborhoodAttentionOutput.__init__": ["Dropout", "Linear", "ModelAttentionOutput", "Module", "__init__", "attention_probs_dropout_prob", "class", "config", "def", "dense", "dim", "dropout", "nn", "self", "super"], "dinat/modeling_dinat.py:NeighborhoodAttentionOutput.forward": ["ModelAttentionOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "dinat/modeling_dinat.py:NeighborhoodAttentionModule.__init__": ["ModelAttention", "ModelAttentionModule", "ModelAttentionOutput", "Module", "__init__", "class", "config", "def", "dilation", "dim", "kernel_size", "nn", "num_heads", "output", "self", "super"], "dinat/modeling_dinat.py:NeighborhoodAttentionModule.forward": ["False", "ModelAttentionModule", "Module", "None", "Tensor", "attention_output", "bool", "class", "def", "forward", "hidden_states", "nn", "output", "output_attentions", "outputs", "return", "self", "self_outputs", "torch", "tuple"], "dinat/modeling_dinat.py:DinatIntermediate.__init__": ["ACT2FN", "Linear", "ModelIntermediate", "Module", "__init__", "class", "config", "def", "dense", "dim", "else", "hidden_act", "if", "int", "intermediate_act_fn", "isinstance", "mlp_ratio", "nn", "self", "str", "super"], "dinat/modeling_dinat.py:DinatIntermediate.forward": ["ModelIntermediate", "Module", "Tensor", "class", "def", "dense", "forward", "hidden_states", "intermediate_act_fn", "nn", "return", "self", "torch"], "dinat/modeling_dinat.py:DinatOutput.__init__": ["Dropout", "Linear", "ModelOutput", "Module", "__init__", "class", "config", "def", "dense", "dim", "dropout", "hidden_dropout_prob", "int", "mlp_ratio", "nn", "self", "super"], "dinat/modeling_dinat.py:DinatOutput.forward": ["ModelOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "nn", "return", "self", "torch"], "dinat/modeling_dinat.py:DinatLayer.__init__": ["Identity", "LayerNorm", "ModelDropPath", "ModelIntermediate", "ModelLayer", "ModelOutput", "Module", "NeighborhoodAttentionModule", "None", "Parameter", "True", "__init__", "attention", "chunk_size_feed_forward", "class", "config", "def", "dilation", "dim", "drop_path", "drop_path_rate", "else", "eps", "if", "intermediate", "kernel_size", "layer_norm_eps", "layer_scale_init_value", "layer_scale_parameters", "layernorm_after", "layernorm_before", "nn", "num_heads", "ones", "output", "requires_grad", "self", "super", "torch", "window_size"], "dinat/modeling_dinat.py:DinatLayer.maybe_pad": ["ModelLayer", "Module", "class", "def", "functional", "height", "hidden_states", "if", "max", "maybe_pad", "nn", "or", "pad", "pad_b", "pad_l", "pad_r", "pad_t", "pad_values", "return", "self", "width", "window_size"], "dinat/modeling_dinat.py:DinatLayer.forward": ["False", "ModelLayer", "Module", "None", "Tensor", "_", "attention", "attention_output", "attention_outputs", "batch_size", "bool", "channels", "class", "contiguous", "def", "drop_path", "else", "forward", "height", "height_pad", "hidden_states", "if", "intermediate", "is", "layer_output", "layer_outputs", "layer_scale_parameters", "layernorm_after", "layernorm_before", "maybe_pad", "nn", "not", "or", "output", "output_attentions", "pad_values", "return", "self", "shape", "shortcut", "size", "torch", "tuple", "was_padded", "width", "width_pad"], "dinat/modeling_dinat.py:DinatStage.__init__": ["False", "LayerNorm", "ModelLayer", "ModelStage", "Module", "ModuleList", "None", "__init__", "class", "config", "def", "depth", "dilation", "dilations", "dim", "downsample", "drop_path_rate", "else", "for", "i", "if", "in", "is", "layers", "nn", "norm_layer", "not", "num_heads", "pointing", "range", "self", "super"], "dinat/modeling_dinat.py:DinatStage.forward": ["False", "ModelStage", "Module", "None", "Tensor", "_", "bool", "class", "def", "downsample", "enumerate", "for", "forward", "height", "hidden_states", "hidden_states_before_downsampling", "i", "if", "in", "is", "layer_module", "layer_outputs", "layers", "nn", "not", "output_attentions", "return", "self", "size", "stage_outputs", "torch", "tuple", "width"], "dinat/modeling_dinat.py:DinatEncoder.__init__": ["ModelDownsampler", "ModelEncoder", "ModelStage", "Module", "ModuleList", "None", "__init__", "class", "config", "cpu", "def", "depth", "depths", "device", "dilations", "dim", "downsample", "dpr", "drop_path_rate", "else", "embed_dim", "for", "i_layer", "if", "in", "int", "item", "len", "levels", "linspace", "nn", "num_heads", "num_levels", "range", "self", "sum", "super", "torch", "x"], "dinat/modeling_dinat.py:DinatEncoder.forward": ["False", "ModelEncoder", "ModelEncoderOutput", "Module", "None", "Tensor", "True", "all_hidden_states", "all_reshaped_hidden_states", "all_self_attentions", "and", "attentions", "bool", "class", "def", "elif", "else", "enumerate", "for", "forward", "hidden_states", "hidden_states_before_downsampling", "i", "if", "in", "is", "last_hidden_state", "layer_module", "layer_outputs", "levels", "nn", "not", "output_attentions", "output_hidden_states", "output_hidden_states_before_downsampling", "permute", "reshaped_hidden_state", "reshaped_hidden_states", "return", "return_dict", "self", "torch", "tuple", "v"], "dinat/modeling_dinat.py:DinatModel.__init__": ["AdaptiveAvgPool1d", "LayerNorm", "ModelEmbeddings", "ModelEncoder", "ModelModel", "ModelPreTrainedModel", "None", "True", "__init__", "add_pooling_layer", "class", "config", "def", "depths", "else", "embed_dim", "embeddings", "encoder", "eps", "if", "int", "layer_norm_eps", "layernorm", "len", "natten", "nn", "num_features", "num_levels", "pooler", "post_init", "r", "requires_backends", "self", "super"], "dinat/modeling_dinat.py:DinatModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "patch_embeddings", "return", "self"], "dinat/modeling_dinat.py:DinatModel.forward": ["FloatTensor", "ModelModel", "ModelModelOutput", "ModelPreTrainedModel", "None", "ValueError", "You", "attentions", "bool", "class", "config", "def", "else", "embedding_output", "embeddings", "encoder", "encoder_outputs", "flatten", "forward", "have", "hidden_states", "if", "is", "kwargs", "last_hidden_state", "layernorm", "not", "output", "output_attentions", "output_hidden_states", "pixel_values", "pooled_output", "pooler", "pooler_output", "raise", "reshaped_hidden_states", "return", "return_dict", "self", "sequence_output", "specify", "to", "torch", "transpose", "tuple", "use_return_dict"], "dinat/modeling_dinat.py:DinatForImageClassification.__init__": ["Identity", "Linear", "Model", "ModelForImageClassification", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "classifier", "config", "def", "else", "if", "natten", "nn", "num_features", "num_labels", "post_init", "requires_backends", "self", "super"], "dinat/modeling_dinat.py:DinatForImageClassification.forward": ["FloatTensor", "LongTensor", "Model", "ModelForImageClassification", "ModelImageClassifierOutput", "ModelPreTrainedModel", "None", "attentions", "bool", "class", "classifier", "config", "def", "else", "forward", "hidden_states", "if", "is", "kwargs", "labels", "logits", "loss", "loss_function", "not", "output", "output_attentions", "output_hidden_states", "outputs", "pixel_values", "pooled_output", "r", "reshaped_hidden_states", "return", "return_dict", "self", "torch", "tuple", "use_return_dict"], "dinat/modeling_dinat.py:DinatBackbone.__init__": ["BackboneMixin", "LayerNorm", "ModelBackbone", "ModelEmbeddings", "ModelEncoder", "ModelPreTrainedModel", "ModuleDict", "__init__", "_init_backbone", "_out_features", "channels", "class", "config", "def", "depths", "embed_dim", "embeddings", "encoder", "for", "hidden_states_norms", "i", "in", "int", "len", "natten", "nn", "num_channels", "num_features", "post_init", "range", "requires_backends", "self", "stage", "super", "zip"], "dinat/modeling_dinat.py:DinatBackbone.get_input_embeddings": ["BackboneMixin", "ModelBackbone", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "patch_embeddings", "return", "self"], "dinat/modeling_dinat.py:DinatBackbone.forward": ["BackboneMixin", "BackboneOutput", "ModelBackbone", "ModelPreTrainedModel", "None", "Tensor", "True", "attentions", "batch_size", "bool", "class", "config", "contiguous", "def", "else", "embedding_output", "embeddings", "encoder", "feature_maps", "for", "forward", "height", "hidden_state", "hidden_states", "hidden_states_norms", "if", "in", "is", "kwargs", "not", "num_channels", "out_features", "output", "output_attentions", "output_hidden_states", "output_hidden_states_before_downsampling", "outputs", "permute", "pixel_values", "r", "reshaped_hidden_states", "return", "return_dict", "self", "shape", "stage", "stage_names", "torch", "use_return_dict", "view", "width", "zip"], "dinov2/modeling_dinov2.py:Dinov2Embeddings.__init__": ["Dropout", "ModelConfig", "ModelEmbeddings", "ModelPatchEmbeddings", "Module", "None", "Parameter", "__init__", "class", "cls_token", "config", "def", "dropout", "hidden_dropout_prob", "hidden_size", "if", "mask_token", "nn", "num_patches", "patch_embeddings", "patch_size", "position_embeddings", "randn", "self", "super", "torch", "use_mask_token", "zeros"], "dinov2/modeling_dinov2.py:Dinov2Embeddings.interpolate_pos_encoding": ["False", "ModelEmbeddings", "Module", "Tensor", "align_corners", "and", "bicubic", "cat", "class", "class_pos_embed", "def", "dim", "dtype", "embeddings", "float32", "functional", "height", "if", "int", "interpolate", "interpolate_pos_encoding", "is_tracing", "jit", "mode", "new_height", "new_width", "nn", "not", "num_patches", "num_positions", "patch_pos_embed", "patch_size", "permute", "position_embeddings", "reshape", "return", "self", "shape", "size", "sqrt_num_positions", "target_dtype", "to", "torch", "torch_int", "view", "width"], "dinov2/modeling_dinov2.py:Dinov2Embeddings.forward": ["ModelEmbeddings", "Module", "None", "Tensor", "_", "and", "batch_size", "bool_masked_pos", "cat", "class", "cls_token", "cls_tokens", "def", "dim", "dropout", "dtype", "embeddings", "expand", "forward", "height", "if", "interpolate_pos_encoding", "is", "mask_token", "nn", "not", "patch_embeddings", "pixel_values", "projection", "return", "self", "shape", "target_dtype", "to", "torch", "unsqueeze", "use_mask_token", "weight", "where", "width"], "dinov2/modeling_dinov2.py:Dinov2PatchEmbeddings.__init__": ["Conv2d", "Iterable", "ModelPatchEmbeddings", "Module", "__init__", "abc", "class", "collections", "config", "def", "else", "hidden_size", "if", "image_size", "isinstance", "kernel_size", "nn", "num_channels", "num_patches", "patch_size", "projection", "self", "stride", "super"], "dinov2/modeling_dinov2.py:Dinov2PatchEmbeddings.forward": ["Expected", "Make", "ModelPatchEmbeddings", "Module", "Tensor", "ValueError", "but", "channel", "class", "configuration", "def", "dimension", "embeddings", "f", "flatten", "forward", "got", "if", "in", "match", "nn", "num_channels", "of", "one", "pixel", "pixel_values", "projection", "raise", "return", "self", "set", "shape", "sure", "that", "the", "torch", "transpose", "values", "with"], "dinov2/modeling_dinov2.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "contiguous", "def", "dim", "dropout", "float", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "shape", "size", "softmax", "torch", "training", "transpose", "value"], "dinov2/modeling_dinov2.py:Dinov2SelfAttention.__init__": ["False", "Linear", "ModelConfig", "ModelSelfAttention", "Module", "The", "ValueError", "__init__", "a", "all_head_size", "and", "attention", "attention_head_size", "attention_probs_dropout_prob", "bias", "class", "config", "def", "dropout_prob", "embedding_size", "f", "hasattr", "heads", "hidden", "hidden_size", "if", "int", "is", "is_causal", "key", "multiple", "nn", "not", "num_attention_heads", "number", "of", "qkv_bias", "query", "raise", "scaling", "self", "size", "super", "the", "value"], "dinov2/modeling_dinov2.py:Dinov2SelfAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "ModelSelfAttention", "Module", "None", "Tensor", "_attn_implementation", "all_head_size", "attention_head_size", "attention_interface", "attention_probs", "batch_size", "class", "config", "context_layer", "def", "dropout", "dropout_prob", "eager", "eager_attention_forward", "else", "forward", "hidden_states", "if", "is_causal", "key", "key_layer", "new_context_layer_shape", "new_shape", "nn", "not", "num_attention_heads", "query", "query_layer", "reshape", "return", "scaling", "self", "shape", "size", "torch", "training", "transpose", "tuple", "value", "value_layer", "view"], "dinov2/modeling_dinov2.py:Dinov2SelfOutput.__init__": ["Dropout", "Linear", "ModelConfig", "ModelSelfOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "hidden_dropout_prob", "hidden_size", "nn", "self", "super"], "dinov2/modeling_dinov2.py:Dinov2SelfOutput.forward": ["ModelSelfOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "dinov2/modeling_dinov2.py:Dinov2Attention.__init__": ["ModelAttention", "ModelConfig", "ModelSelfAttention", "ModelSelfOutput", "Module", "__init__", "attention", "class", "config", "def", "nn", "output", "self", "super"], "dinov2/modeling_dinov2.py:Dinov2Attention.forward": ["ModelAttention", "Module", "Tensor", "_", "attention", "class", "def", "forward", "hidden_states", "nn", "output", "return", "self", "self_attn_output", "torch"], "dinov2/modeling_dinov2.py:Dinov2LayerScale.__init__": ["ModelLayerScale", "Module", "None", "Parameter", "__init__", "class", "config", "def", "hidden_size", "lambda1", "layerscale_value", "nn", "ones", "self", "super", "torch"], "dinov2/modeling_dinov2.py:Dinov2LayerScale.forward": ["ModelLayerScale", "Module", "Tensor", "class", "def", "forward", "hidden_state", "lambda1", "nn", "return", "self", "torch"], "dinov2/modeling_dinov2.py:drop_path": ["False", "Model_path", "Model_prob", "Tensor", "bool", "def", "device", "div", "dtype", "float", "floor_", "if", "input", "keep_prob", "ndim", "not", "or", "output", "rand", "random_tensor", "return", "shape", "torch", "training"], "dinov2/modeling_dinov2.py:Dinov2DropPath.__init__": ["ModelDropPath", "Module", "None", "__init__", "class", "def", "drop_prob", "float", "nn", "self", "super"], "dinov2/modeling_dinov2.py:Dinov2DropPath.forward": ["ModelDropPath", "Module", "Tensor", "class", "def", "drop_path", "drop_prob", "forward", "hidden_states", "nn", "return", "self", "torch", "training"], "dinov2/modeling_dinov2.py:Dinov2DropPath.extra_repr": ["ModelDropPath", "Module", "class", "def", "drop_prob", "extra_repr", "f", "nn", "p", "return", "self", "str"], "dinov2/modeling_dinov2.py:Dinov2MLP.__init__": ["ACT2FN", "Linear", "ModelMLP", "Module", "None", "True", "__init__", "activation", "bias", "class", "config", "def", "else", "fc1", "fc2", "hidden_act", "hidden_features", "hidden_size", "if", "in_features", "int", "isinstance", "mlp_ratio", "nn", "out_features", "self", "str", "super"], "dinov2/modeling_dinov2.py:Dinov2MLP.forward": ["ModelMLP", "Module", "Tensor", "activation", "class", "def", "fc1", "fc2", "forward", "hidden_state", "nn", "return", "self", "torch"], "dinov2/modeling_dinov2.py:Dinov2SwiGLUFFN.__init__": ["Linear", "ModelSwiGLUFFN", "Module", "None", "True", "__init__", "bias", "class", "config", "def", "hidden_features", "hidden_size", "in_features", "int", "mlp_ratio", "nn", "out_features", "self", "super", "weights_in", "weights_out"], "dinov2/modeling_dinov2.py:Dinov2SwiGLUFFN.forward": ["ModelSwiGLUFFN", "Module", "Tensor", "chunk", "class", "def", "dim", "forward", "functional", "hidden", "hidden_state", "nn", "return", "self", "silu", "torch", "weights_in", "weights_out", "x1", "x2"], "dinov2/modeling_dinov2.py:Dinov2Layer.__init__": ["GradientCheckpointingLayer", "Identity", "LayerNorm", "ModelAttention", "ModelConfig", "ModelDropPath", "ModelLayer", "ModelLayerScale", "ModelMLP", "ModelSwiGLUFFN", "None", "__init__", "attention", "class", "config", "def", "drop_path", "drop_path_rate", "else", "eps", "hidden_size", "if", "layer_norm_eps", "layer_scale1", "layer_scale2", "mlp", "nn", "norm1", "norm2", "self", "super", "use_swiglu_ffn"], "dinov2/modeling_dinov2.py:Dinov2Layer.forward": ["GradientCheckpointingLayer", "ModelLayer", "Tensor", "attention", "class", "def", "drop_path", "forward", "hidden_states", "hidden_states_norm", "layer_output", "layer_scale1", "layer_scale2", "mlp", "norm1", "norm2", "return", "self", "self_attention_output", "torch"], "dinov2/modeling_dinov2.py:Dinov2Encoder.__init__": ["False", "ModelConfig", "ModelEncoder", "ModelLayer", "Module", "ModuleList", "_", "__init__", "class", "config", "def", "for", "gradient_checkpointing", "in", "layer", "nn", "num_hidden_layers", "range", "self", "super"], "dinov2/modeling_dinov2.py:Dinov2Encoder.forward": ["BaseModelOutput", "False", "ModelEncoder", "Module", "None", "Tensor", "all_hidden_states", "append", "bool", "class", "def", "else", "enumerate", "for", "forward", "hidden_states", "i", "if", "in", "last_hidden_state", "layer", "layer_module", "nn", "output_hidden_states", "return", "self", "torch", "tuple"], "dinov2/modeling_dinov2.py:Dinov2PreTrainedModel._init_weights": ["Conv2d", "LayerNorm", "Linear", "ModelEmbeddings", "ModelLayerScale", "ModelPreTrainedModel", "None", "PreTrainedModel", "_init_weights", "bias", "class", "cls_token", "config", "constant_", "def", "elif", "if", "init", "initializer_range", "is", "isinstance", "lambda1", "layerscale_value", "mask_token", "mean", "module", "nn", "not", "ones_", "position_embeddings", "self", "std", "trunc_normal_", "use_mask_token", "weight", "zeros_"], "dinov2/modeling_dinov2.py:Dinov2Model.__init__": ["LayerNorm", "ModelConfig", "ModelEmbeddings", "ModelEncoder", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "config", "def", "embeddings", "encoder", "eps", "hidden_size", "layer_norm_eps", "layernorm", "nn", "post_init", "self", "super"], "dinov2/modeling_dinov2.py:Dinov2Model.get_input_embeddings": ["ModelModel", "ModelPatchEmbeddings", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "patch_embeddings", "return", "self"], "dinov2/modeling_dinov2.py:Dinov2Model.forward": ["BaseModelOutput", "BaseModelOutputWithPooling", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "ValueError", "You", "bool", "bool_masked_pos", "class", "config", "def", "embedding_output", "embeddings", "encoder", "encoder_outputs", "forward", "have", "hidden_states", "if", "is", "kwargs", "last_hidden_state", "layernorm", "output_hidden_states", "pixel_values", "pooled_output", "pooler_output", "r", "raise", "return", "self", "sequence_output", "specify", "to", "torch"], "dinov2/modeling_dinov2.py:Dinov2ForImageClassification.__init__": ["Identity", "Linear", "Model", "ModelConfig", "ModelForImageClassification", "ModelModel", "ModelPreTrainedModel", "None", "__init__", "class", "classifier", "config", "def", "else", "hidden_size", "if", "nn", "num_labels", "post_init", "self", "super"], "dinov2/modeling_dinov2.py:Dinov2ForImageClassification.forward": ["BaseModelOutputWithPooling", "ImageClassifierOutput", "Model", "ModelForImageClassification", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "attentions", "cat", "class", "classifier", "cls_token", "config", "def", "dim", "forward", "hidden_states", "if", "is", "kwargs", "labels", "last_hidden_state", "linear_input", "logits", "loss", "loss_function", "mean", "not", "outputs", "patch_tokens", "pixel_values", "r", "return", "self", "sequence_output", "torch"], "dinov2/modeling_dinov2.py:Dinov2Backbone.__init__": ["BackboneMixin", "LayerNorm", "ModelBackbone", "ModelEmbeddings", "ModelEncoder", "ModelPreTrainedModel", "_", "__init__", "_init_backbone", "class", "config", "def", "embeddings", "encoder", "eps", "for", "hidden_size", "in", "layer_norm_eps", "layernorm", "nn", "num_features", "num_hidden_layers", "post_init", "range", "self", "super"], "dinov2/modeling_dinov2.py:Dinov2Backbone.get_input_embeddings": ["BackboneMixin", "ModelBackbone", "ModelPatchEmbeddings", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "patch_embeddings", "return", "self"], "dinov2/modeling_dinov2.py:Dinov2Backbone.forward": ["BackboneMixin", "BackboneOutput", "BaseModelOutput", "ModelBackbone", "ModelPreTrainedModel", "None", "Tensor", "True", "_", "append", "apply_layernorm", "batch_size", "bool", "class", "config", "contiguous", "def", "else", "embedding_output", "embeddings", "encoder", "feature_maps", "for", "forward", "height", "hidden_state", "hidden_states", "if", "in", "is", "kwargs", "layernorm", "out_features", "output", "output_hidden_states", "patch_size", "permute", "pixel_values", "r", "reshape", "reshape_hidden_states", "return", "self", "shape", "stage", "stage_names", "torch", "tuple", "width", "zip"], "dinov2_with_registers/modeling_dinov2_with_registers.py:Dinov2WithRegistersPatchEmbeddings.__init__": ["Conv2d", "Iterable", "ModelPatchEmbeddings", "Module", "__init__", "abc", "class", "collections", "config", "def", "else", "hidden_size", "if", "image_size", "isinstance", "kernel_size", "nn", "num_channels", "num_patches", "patch_size", "projection", "self", "stride", "super"], "dinov2_with_registers/modeling_dinov2_with_registers.py:Dinov2WithRegistersPatchEmbeddings.forward": ["Expected", "Make", "ModelPatchEmbeddings", "Module", "Tensor", "ValueError", "but", "channel", "class", "configuration", "def", "dimension", "embeddings", "f", "flatten", "forward", "got", "if", "in", "match", "nn", "num_channels", "of", "one", "pixel", "pixel_values", "projection", "raise", "return", "self", "set", "shape", "sure", "that", "the", "torch", "transpose", "values", "with"], "dinov2_with_registers/modeling_dinov2_with_registers.py:Dinov2WithRegistersEmbeddings.__init__": ["Dropout", "ModelConfig", "ModelEmbeddings", "ModelPatchEmbeddings", "Module", "None", "Parameter", "__init__", "class", "cls_token", "config", "def", "dropout", "hidden_dropout_prob", "hidden_size", "mask_token", "nn", "num_patches", "num_register_tokens", "patch_embeddings", "patch_size", "position_embeddings", "randn", "register_tokens", "self", "super", "torch", "zeros"], "dinov2_with_registers/modeling_dinov2_with_registers.py:Dinov2WithRegistersEmbeddings.interpolate_pos_encoding": ["False", "ModelEmbeddings", "Module", "Tensor", "True", "ValueError", "Width", "align_corners", "and", "antialias", "bicubic", "cat", "class", "class_pos_embed", "config", "def", "dim", "does", "dtype", "embeddings", "float32", "functional", "height", "if", "int", "interpolate", "interpolate_pos_encoding", "interpolated", "is_tracing", "jit", "match", "mode", "nn", "not", "num_patches", "num_positions", "or", "patch_pos_embed", "patch_size", "permute", "position", "position_embeddings", "raise", "reshape", "return", "self", "shape", "size", "sqrt_num_positions", "target_dtype", "the", "to", "torch", "torch_int", "unsqueeze", "view", "width", "with"], "dinov2_with_registers/modeling_dinov2_with_registers.py:Dinov2WithRegistersEmbeddings.forward": ["ModelEmbeddings", "Module", "None", "Tensor", "_", "batch_size", "bool_masked_pos", "cat", "class", "cls_token", "cls_tokens", "def", "dim", "dropout", "dtype", "embeddings", "expand", "forward", "height", "if", "interpolate_pos_encoding", "is", "mask_token", "nn", "not", "patch_embeddings", "pixel_values", "projection", "register_tokens", "return", "self", "shape", "target_dtype", "to", "torch", "unsqueeze", "weight", "where", "width"], "dinov2_with_registers/modeling_dinov2_with_registers.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "contiguous", "def", "dim", "dropout", "float", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "shape", "size", "softmax", "torch", "training", "transpose", "value"], "dinov2_with_registers/modeling_dinov2_with_registers.py:Dinov2WithRegistersSelfAttention.__init__": ["False", "Linear", "ModelConfig", "ModelSelfAttention", "Module", "The", "ValueError", "__init__", "a", "all_head_size", "and", "attention", "attention_head_size", "attention_probs_dropout_prob", "bias", "class", "config", "def", "dropout_prob", "embedding_size", "f", "hasattr", "heads", "hidden", "hidden_size", "if", "int", "is", "is_causal", "key", "multiple", "nn", "not", "num_attention_heads", "number", "of", "qkv_bias", "query", "raise", "scaling", "self", "size", "super", "the", "value"], "dinov2_with_registers/modeling_dinov2_with_registers.py:Dinov2WithRegistersSelfAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "ModelSelfAttention", "Module", "None", "Tensor", "_attn_implementation", "all_head_size", "attention_head_size", "attention_interface", "attention_probs", "batch_size", "class", "config", "context_layer", "def", "dropout", "dropout_prob", "eager", "eager_attention_forward", "else", "forward", "hidden_states", "if", "is_causal", "key", "key_layer", "new_context_layer_shape", "new_shape", "nn", "not", "num_attention_heads", "query", "query_layer", "reshape", "return", "scaling", "self", "shape", "size", "torch", "training", "transpose", "tuple", "value", "value_layer", "view"], "dinov2_with_registers/modeling_dinov2_with_registers.py:Dinov2WithRegistersSelfOutput.__init__": ["Dropout", "Linear", "ModelConfig", "ModelSelfOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "hidden_dropout_prob", "hidden_size", "nn", "self", "super"], "dinov2_with_registers/modeling_dinov2_with_registers.py:Dinov2WithRegistersSelfOutput.forward": ["ModelSelfOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "dinov2_with_registers/modeling_dinov2_with_registers.py:Dinov2WithRegistersAttention.__init__": ["ModelAttention", "ModelConfig", "ModelSelfAttention", "ModelSelfOutput", "Module", "__init__", "attention", "class", "config", "def", "nn", "output", "self", "super"], "dinov2_with_registers/modeling_dinov2_with_registers.py:Dinov2WithRegistersAttention.forward": ["ModelAttention", "Module", "Tensor", "_", "attention", "class", "def", "forward", "hidden_states", "nn", "output", "return", "self", "self_attn_output", "torch"], "dinov2_with_registers/modeling_dinov2_with_registers.py:Dinov2WithRegistersLayerScale.__init__": ["ModelLayerScale", "Module", "None", "Parameter", "__init__", "class", "config", "def", "hidden_size", "lambda1", "layerscale_value", "nn", "ones", "self", "super", "torch"], "dinov2_with_registers/modeling_dinov2_with_registers.py:Dinov2WithRegistersLayerScale.forward": ["ModelLayerScale", "Module", "Tensor", "class", "def", "forward", "hidden_state", "lambda1", "nn", "return", "self", "torch"], "dinov2_with_registers/modeling_dinov2_with_registers.py:drop_path": ["False", "Model_path", "Model_prob", "Tensor", "bool", "def", "device", "div", "dtype", "float", "floor_", "if", "input", "keep_prob", "ndim", "not", "or", "output", "rand", "random_tensor", "return", "shape", "torch", "training"], "dinov2_with_registers/modeling_dinov2_with_registers.py:Dinov2WithRegistersDropPath.__init__": ["ModelDropPath", "Module", "None", "__init__", "class", "def", "drop_prob", "float", "nn", "self", "super"], "dinov2_with_registers/modeling_dinov2_with_registers.py:Dinov2WithRegistersDropPath.forward": ["ModelDropPath", "Module", "Tensor", "class", "def", "drop_path", "drop_prob", "forward", "hidden_states", "nn", "return", "self", "torch", "training"], "dinov2_with_registers/modeling_dinov2_with_registers.py:Dinov2WithRegistersDropPath.extra_repr": ["ModelDropPath", "Module", "class", "def", "drop_prob", "extra_repr", "f", "nn", "p", "return", "self", "str"], "dinov2_with_registers/modeling_dinov2_with_registers.py:Dinov2WithRegistersMLP.__init__": ["ACT2FN", "Linear", "ModelMLP", "Module", "None", "True", "__init__", "activation", "bias", "class", "config", "def", "else", "fc1", "fc2", "hidden_act", "hidden_features", "hidden_size", "if", "in_features", "int", "isinstance", "mlp_ratio", "nn", "out_features", "self", "str", "super"], "dinov2_with_registers/modeling_dinov2_with_registers.py:Dinov2WithRegistersMLP.forward": ["ModelMLP", "Module", "Tensor", "activation", "class", "def", "fc1", "fc2", "forward", "hidden_state", "nn", "return", "self", "torch"], "dinov2_with_registers/modeling_dinov2_with_registers.py:Dinov2WithRegistersSwiGLUFFN.__init__": ["Linear", "ModelSwiGLUFFN", "Module", "None", "True", "__init__", "bias", "class", "config", "def", "hidden_features", "hidden_size", "in_features", "int", "mlp_ratio", "nn", "out_features", "self", "super", "weights_in", "weights_out"], "dinov2_with_registers/modeling_dinov2_with_registers.py:Dinov2WithRegistersSwiGLUFFN.forward": ["ModelSwiGLUFFN", "Module", "Tensor", "chunk", "class", "def", "dim", "forward", "functional", "hidden", "hidden_state", "nn", "return", "self", "silu", "torch", "weights_in", "weights_out", "x1", "x2"], "dinov2_with_registers/modeling_dinov2_with_registers.py:Dinov2WithRegistersLayer.__init__": ["GradientCheckpointingLayer", "Identity", "LayerNorm", "ModelAttention", "ModelConfig", "ModelDropPath", "ModelLayer", "ModelLayerScale", "ModelMLP", "ModelSwiGLUFFN", "None", "__init__", "attention", "class", "config", "def", "drop_path", "drop_path_rate", "else", "eps", "hidden_size", "if", "layer_norm_eps", "layer_scale1", "layer_scale2", "mlp", "nn", "norm1", "norm2", "self", "super", "use_swiglu_ffn"], "dinov2_with_registers/modeling_dinov2_with_registers.py:Dinov2WithRegistersLayer.forward": ["GradientCheckpointingLayer", "ModelLayer", "Tensor", "attention", "class", "def", "drop_path", "forward", "hidden_states", "hidden_states_norm", "layer_output", "layer_scale1", "layer_scale2", "mlp", "norm1", "norm2", "return", "self", "self_attention_output", "torch"], "dinov2_with_registers/modeling_dinov2_with_registers.py:Dinov2WithRegistersEncoder.__init__": ["False", "ModelConfig", "ModelEncoder", "ModelLayer", "Module", "ModuleList", "_", "__init__", "class", "config", "def", "for", "gradient_checkpointing", "in", "layer", "nn", "num_hidden_layers", "range", "self", "super"], "dinov2_with_registers/modeling_dinov2_with_registers.py:Dinov2WithRegistersEncoder.forward": ["BaseModelOutput", "False", "ModelEncoder", "Module", "None", "Tensor", "all_hidden_states", "append", "bool", "class", "def", "else", "enumerate", "for", "forward", "hidden_states", "i", "if", "in", "last_hidden_state", "layer", "layer_module", "nn", "output_hidden_states", "return", "self", "torch", "tuple"], "dinov2_with_registers/modeling_dinov2_with_registers.py:Dinov2WithRegistersPreTrainedModel._init_weights": ["Conv2d", "LayerNorm", "Linear", "ModelEmbeddings", "ModelLayerScale", "ModelPreTrainedModel", "None", "PreTrainedModel", "_init_weights", "bias", "class", "cls_token", "config", "constant_", "def", "elif", "if", "init", "initializer_range", "is", "isinstance", "lambda1", "layerscale_value", "mask_token", "mean", "module", "nn", "not", "ones_", "position_embeddings", "register_tokens", "self", "std", "trunc_normal_", "weight", "zeros_"], "dinov2_with_registers/modeling_dinov2_with_registers.py:Dinov2WithRegistersModel.__init__": ["LayerNorm", "ModelConfig", "ModelEmbeddings", "ModelEncoder", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "config", "def", "embeddings", "encoder", "eps", "hidden_size", "layer_norm_eps", "layernorm", "nn", "post_init", "self", "super"], "dinov2_with_registers/modeling_dinov2_with_registers.py:Dinov2WithRegistersModel.get_input_embeddings": ["ModelModel", "ModelPatchEmbeddings", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "patch_embeddings", "return", "self"], "dinov2_with_registers/modeling_dinov2_with_registers.py:Dinov2WithRegistersModel.forward": ["BaseModelOutput", "BaseModelOutputWithPooling", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "ValueError", "You", "bool", "bool_masked_pos", "class", "config", "def", "embedding_output", "embeddings", "encoder", "encoder_outputs", "forward", "have", "hidden_states", "if", "is", "kwargs", "last_hidden_state", "layernorm", "output_hidden_states", "pixel_values", "pooled_output", "pooler_output", "r", "raise", "return", "self", "sequence_output", "specify", "to", "torch"], "dinov2_with_registers/modeling_dinov2_with_registers.py:Dinov2WithRegistersForImageClassification.__init__": ["Identity", "Linear", "Model", "ModelConfig", "ModelForImageClassification", "ModelModel", "ModelPreTrainedModel", "None", "__init__", "class", "classifier", "config", "def", "else", "hidden_size", "if", "nn", "num_labels", "post_init", "self", "super"], "dinov2_with_registers/modeling_dinov2_with_registers.py:Dinov2WithRegistersForImageClassification.forward": ["BaseModelOutputWithPooling", "ImageClassifierOutput", "Model", "ModelForImageClassification", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "attentions", "cat", "class", "classifier", "cls_token", "config", "def", "dim", "forward", "hidden_states", "if", "is", "kwargs", "labels", "last_hidden_state", "linear_input", "logits", "loss", "loss_function", "mean", "not", "num_register_tokens", "outputs", "patch_tokens", "pixel_values", "r", "return", "self", "sequence_output", "torch"], "dinov2_with_registers/modeling_dinov2_with_registers.py:Dinov2WithRegistersBackbone.__init__": ["BackboneMixin", "LayerNorm", "ModelBackbone", "ModelEmbeddings", "ModelEncoder", "ModelPreTrainedModel", "_", "__init__", "_init_backbone", "class", "config", "def", "embeddings", "encoder", "eps", "for", "hidden_size", "in", "layer_norm_eps", "layernorm", "nn", "num_features", "num_hidden_layers", "num_register_tokens", "post_init", "range", "self", "super"], "dinov2_with_registers/modeling_dinov2_with_registers.py:Dinov2WithRegistersBackbone.get_input_embeddings": ["BackboneMixin", "ModelBackbone", "ModelPatchEmbeddings", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "patch_embeddings", "return", "self"], "dinov2_with_registers/modeling_dinov2_with_registers.py:Dinov2WithRegistersBackbone.forward": ["BackboneMixin", "BackboneOutput", "BaseModelOutput", "ModelBackbone", "ModelPreTrainedModel", "None", "Tensor", "True", "_", "append", "apply_layernorm", "batch_size", "bool", "class", "config", "contiguous", "def", "else", "embedding_output", "embeddings", "encoder", "feature_maps", "for", "forward", "height", "hidden_state", "hidden_states", "if", "in", "is", "kwargs", "layernorm", "num_register_tokens", "out_features", "output", "output_hidden_states", "patch_size", "permute", "pixel_values", "r", "reshape", "reshape_hidden_states", "return", "self", "shape", "stage", "stage_names", "torch", "tuple", "width", "zip"], "dinov3_convnext/modeling_dinov3_convnext.py:drop_path": ["False", "Model_path", "Model_prob", "Tensor", "bool", "def", "device", "div", "dtype", "float", "floor_", "if", "input", "keep_prob", "ndim", "not", "or", "output", "rand", "random_tensor", "return", "shape", "torch", "training"], "dinov3_convnext/modeling_dinov3_convnext.py:DINOv3ConvNextDropPath.__init__": ["Model", "Module", "None", "__init__", "class", "def", "drop_prob", "float", "nn", "self", "super"], "dinov3_convnext/modeling_dinov3_convnext.py:DINOv3ConvNextDropPath.forward": ["Model", "Module", "Tensor", "class", "def", "drop_path", "drop_prob", "forward", "hidden_states", "nn", "return", "self", "torch", "training"], "dinov3_convnext/modeling_dinov3_convnext.py:DINOv3ConvNextDropPath.extra_repr": ["Model", "Module", "class", "def", "drop_prob", "extra_repr", "f", "nn", "p", "return", "self", "str"], "dinov3_convnext/modeling_dinov3_convnext.py:DINOv3ConvNextLayerNorm.__init__": ["LayerNorm", "Model", "NotImplementedError", "Unsupported", "__init__", "args", "channels_first", "channels_last", "class", "data", "data_format", "def", "f", "format", "if", "in", "kwargs", "nn", "not", "raise", "self", "super"], "dinov3_convnext/modeling_dinov3_convnext.py:DINOv3ConvNextLayerNorm.forward": ["LayerNorm", "Model", "Tensor", "channels_first", "class", "data_format", "def", "else", "features", "forward", "if", "nn", "permute", "return", "self", "super", "torch"], "dinov3_convnext/modeling_dinov3_convnext.py:DINOv3ConvNextLayer.__init__": ["ACT2FN", "Conv2d", "Identity", "Linear", "Model", "ModelConfig", "ModelDropPath", "ModelNorm", "Module", "Parameter", "True", "__init__", "activation_fn", "channels", "class", "config", "def", "depthwise_conv", "drop_path", "else", "eps", "float", "full", "gamma", "groups", "hidden_act", "if", "int", "kernel_size", "layer_norm", "layer_norm_eps", "layer_scale_init_value", "nn", "padding", "pointwise_conv1", "pointwise_conv2", "requires_grad", "self", "super", "torch"], "dinov3_convnext/modeling_dinov3_convnext.py:DINOv3ConvNextLayer.forward": ["Model", "Module", "Tensor", "activation_fn", "class", "def", "depthwise_conv", "drop_path", "features", "forward", "gamma", "layer_norm", "nn", "permute", "pointwise_conv1", "pointwise_conv2", "residual", "return", "self", "torch"], "dinov3_convnext/modeling_dinov3_convnext.py:DINOv3ConvNextStage.__init__": ["Conv2d", "Model", "ModelConfig", "ModelLayer", "ModelLayerNorm", "Module", "ModuleList", "__init__", "channels", "channels_first", "class", "config", "data_format", "def", "depths", "downsample_layers", "drop_path", "drop_path_rate", "drop_path_rates", "else", "eps", "for", "hidden_sizes", "i", "if", "in", "in_channels", "int", "kernel_size", "layer_norm_eps", "layers", "linspace", "nn", "np", "num_channels", "num_previous_layers", "num_stage_layers", "num_total_layers", "out_channels", "range", "self", "stage_idx", "stride", "sum", "super", "tolist"], "dinov3_convnext/modeling_dinov3_convnext.py:DINOv3ConvNextStage.forward": ["Model", "Module", "Tensor", "class", "def", "downsample_layers", "features", "for", "forward", "in", "layer", "layers", "nn", "return", "self", "torch"], "dinov3_convnext/modeling_dinov3_convnext.py:DINOv3ConvNextPreTrainedModel._init_weights": ["Model", "ModelLayer", "None", "PreTrainedModel", "_init_weights", "class", "config", "constant_", "def", "gamma", "if", "init", "is", "isinstance", "layer_scale_init_value", "module", "not", "self", "super"], "dinov3_convnext/modeling_dinov3_convnext.py:DINOv3ConvNextModel.__init__": ["AdaptiveAvgPool2d", "LayerNorm", "Model", "ModelConfig", "ModelPreTrainedModel", "ModelStage", "ModuleList", "__init__", "class", "config", "def", "eps", "for", "hidden_sizes", "in", "layer_norm", "layer_norm_eps", "nn", "num_stages", "pool", "post_init", "range", "self", "stage_idx", "stages", "super"], "dinov3_convnext/modeling_dinov3_convnext.py:DINOv3ConvNextModel.forward": ["BaseModelOutputWithPoolingAndNoAttention", "FloatTensor", "Model", "ModelPreTrainedModel", "None", "all_hidden_states", "append", "bool", "cat", "class", "config", "def", "dim", "else", "flatten", "for", "forward", "hidden_states", "if", "in", "kwargs", "last_hidden_state", "layer_norm", "or", "output_hidden_states", "pixel_values", "pool", "pooled_output", "pooler_output", "return", "self", "stage", "stages", "torch", "transpose", "tuple"], "dinov3_convnext/modeling_dinov3_convnext.py:DINOv3ConvNextBackbone.__init__": ["BackboneMixin", "Model", "ModelConfig", "ModelPreTrainedModel", "ModelStage", "ModuleList", "__init__", "_init_backbone", "class", "config", "def", "for", "hidden_sizes", "in", "list", "nn", "num_channels", "num_features", "num_stages", "post_init", "range", "s", "self", "stages", "super"], "dinov3_convnext/modeling_dinov3_convnext.py:DINOv3ConvNextBackbone.get_input_embeddings": ["BackboneMixin", "Model", "ModelPreTrainedModel", "None", "class", "def", "get_input_embeddings", "return", "self"], "dinov3_convnext/modeling_dinov3_convnext.py:DINOv3ConvNextBackbone.forward": ["BackboneMixin", "BackboneOutput", "FloatTensor", "Model", "ModelPreTrainedModel", "None", "Tensor", "all_hidden_states", "append", "bool", "class", "config", "def", "else", "feature_maps", "for", "forward", "hidden_states", "if", "in", "is", "kwargs", "list", "out_features", "output_hidden_states", "pixel_values", "return", "self", "stage", "stage_names", "stages", "torch", "tuple", "zip"], "dinov3_vit/modeling_dinov3_vit.py:DINOv3ViTEmbeddings.__init__": ["Conv2d", "Model", "ModelConfig", "Module", "Parameter", "__init__", "class", "cls_token", "config", "def", "empty", "hidden_size", "kernel_size", "mask_token", "nn", "num_channels", "num_register_tokens", "patch_embeddings", "patch_size", "randn", "register_tokens", "self", "stride", "super", "torch", "zeros"], "dinov3_vit/modeling_dinov3_vit.py:DINOv3ViTEmbeddings.forward": ["Model", "Module", "None", "Tensor", "batch_size", "bool_masked_pos", "cat", "class", "cls_token", "def", "dim", "dtype", "embeddings", "expand", "flatten", "forward", "if", "is", "mask_token", "nn", "not", "patch_embeddings", "pixel_values", "register_tokens", "return", "self", "shape", "target_dtype", "to", "torch", "transpose", "unsqueeze", "weight", "where"], "dinov3_vit/modeling_dinov3_vit.py:get_patches_center_coordinates": ["Model_patches_center_coordinates", "Tensor", "arange", "coords", "coords_h", "coords_w", "def", "device", "dim", "dtype", "flatten", "ij", "indexing", "int", "meshgrid", "num_patches_h", "num_patches_w", "return", "stack", "torch"], "dinov3_vit/modeling_dinov3_vit.py:augment_patches_center_coordinates": ["Model_patches_center_coordinates", "None", "Tensor", "coords", "def", "device", "dtype", "empty", "exp", "float", "if", "is", "jitter", "jitter_hw", "jitter_range", "log", "not", "np", "rescale", "rescale_hw", "rescale_range", "return", "shift", "shift_hw", "torch", "uniform_"], "dinov3_vit/modeling_dinov3_vit.py:DINOv3ViTRopePositionEmbedding.__init__": ["False", "Model", "ModelConfig", "Module", "__init__", "arange", "base", "class", "config", "def", "dtype", "float32", "head_dim", "hidden_size", "image_size", "inv_freq", "nn", "num_attention_heads", "num_patches_h", "num_patches_w", "patch_size", "persistent", "register_buffer", "rope_theta", "self", "super", "torch"], "dinov3_vit/modeling_dinov3_vit.py:DINOv3ViTRopePositionEmbedding.forward": ["False", "Model", "Module", "None", "Tensor", "_", "and", "angles", "augment_patches_center_coordinates", "class", "config", "cos", "cpu", "def", "device", "device_type", "dtype", "else", "enabled", "flatten", "float32", "forward", "get_patches_center_coordinates", "height", "if", "inv_freq", "isinstance", "jitter", "math", "maybe_autocast", "mps", "nn", "num_patches_h", "num_patches_w", "patch_coords", "patch_size", "pi", "pixel_values", "pos_embed_jitter", "pos_embed_rescale", "pos_embed_shift", "rescale", "return", "self", "shape", "shift", "sin", "str", "tile", "to", "torch", "training", "tuple", "type", "width", "with"], "dinov3_vit/modeling_dinov3_vit.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "dinov3_vit/modeling_dinov3_vit.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "contiguous", "def", "dim", "dropout", "float", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "shape", "size", "softmax", "torch", "training", "transpose", "value"], "dinov3_vit/modeling_dinov3_vit.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "Tensor", "cat", "cos", "def", "dim", "k", "k_patches", "k_prefix_tokens", "kwargs", "num_patches", "num_prefix_tokens", "num_tokens", "q", "q_patches", "q_prefix_tokens", "return", "rotate_half", "shape", "sin", "split", "torch", "tuple"], "dinov3_vit/modeling_dinov3_vit.py:DINOv3ViTAttention.__init__": ["False", "Linear", "Model", "ModelConfig", "Module", "__init__", "attention_dropout", "bias", "class", "config", "def", "dropout", "embed_dim", "head_dim", "hidden_size", "is_causal", "k_proj", "key_bias", "nn", "num_attention_heads", "num_heads", "o_proj", "proj_bias", "q_proj", "query_bias", "scaling", "self", "super", "v_proj", "value_bias"], "dinov3_vit/modeling_dinov3_vit.py:DINOv3ViTAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "Model", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "_attn_implementation", "apply_rotary_pos_emb", "attention_interface", "attention_mask", "attn_output", "attn_weights", "batch_size", "class", "config", "contiguous", "cos", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "head_dim", "hidden_states", "if", "k_proj", "key_states", "kwargs", "nn", "not", "num_heads", "o_proj", "patches", "position_embeddings", "q_proj", "query_states", "reshape", "return", "scaling", "self", "sin", "size", "torch", "training", "transpose", "tuple", "v_proj", "value_states", "view"], "dinov3_vit/modeling_dinov3_vit.py:DINOv3ViTLayerScale.__init__": ["Model", "Module", "None", "Parameter", "__init__", "class", "config", "def", "hidden_size", "lambda1", "layerscale_value", "nn", "ones", "self", "super", "torch"], "dinov3_vit/modeling_dinov3_vit.py:DINOv3ViTLayerScale.forward": ["Model", "Module", "Tensor", "class", "def", "forward", "hidden_state", "lambda1", "nn", "return", "self", "torch"], "dinov3_vit/modeling_dinov3_vit.py:drop_path": ["False", "Model_path", "Model_prob", "Tensor", "bool", "def", "device", "div", "dtype", "float", "floor_", "if", "input", "keep_prob", "ndim", "not", "or", "output", "rand", "random_tensor", "return", "shape", "torch", "training"], "dinov3_vit/modeling_dinov3_vit.py:DINOv3ViTDropPath.__init__": ["Model", "Module", "None", "__init__", "class", "def", "drop_prob", "float", "nn", "self", "super"], "dinov3_vit/modeling_dinov3_vit.py:DINOv3ViTDropPath.forward": ["Model", "Module", "Tensor", "class", "def", "drop_path", "drop_prob", "forward", "hidden_states", "nn", "return", "self", "torch", "training"], "dinov3_vit/modeling_dinov3_vit.py:DINOv3ViTDropPath.extra_repr": ["Model", "Module", "class", "def", "drop_prob", "extra_repr", "f", "nn", "p", "return", "self", "str"], "dinov3_vit/modeling_dinov3_vit.py:DINOv3ViTMLP.__init__": ["ACT2FN", "Linear", "Model", "Module", "__init__", "act_fn", "bias", "class", "config", "def", "down_proj", "hidden_act", "hidden_size", "intermediate_size", "mlp_bias", "nn", "self", "super", "up_proj"], "dinov3_vit/modeling_dinov3_vit.py:DINOv3ViTMLP.forward": ["Model", "Module", "act_fn", "class", "def", "down_proj", "forward", "nn", "return", "self", "up_proj", "x"], "dinov3_vit/modeling_dinov3_vit.py:DINOv3ViTGatedMLP.__init__": ["ACT2FN", "Linear", "Model", "Module", "__init__", "act_fn", "bias", "class", "config", "def", "down_proj", "gate_proj", "hidden_act", "hidden_size", "intermediate_size", "mlp_bias", "nn", "self", "super", "up_proj"], "dinov3_vit/modeling_dinov3_vit.py:DINOv3ViTGatedMLP.forward": ["Model", "Module", "act_fn", "class", "def", "down_proj", "forward", "gate_proj", "nn", "return", "self", "up_proj", "x"], "dinov3_vit/modeling_dinov3_vit.py:DINOv3ViTLayer.__init__": ["GradientCheckpointingLayer", "Identity", "LayerNorm", "Model", "ModelAttention", "ModelConfig", "ModelDropPath", "ModelGatedMLP", "ModelMLP", "ModelScale", "__init__", "attention", "class", "config", "def", "drop_path", "drop_path_rate", "else", "eps", "hidden_size", "if", "layer_norm_eps", "layer_scale1", "layer_scale2", "mlp", "nn", "norm1", "norm2", "self", "super", "use_gated_mlp"], "dinov3_vit/modeling_dinov3_vit.py:DINOv3ViTLayer.forward": ["GradientCheckpointingLayer", "Model", "None", "Tensor", "_", "attention", "attention_mask", "class", "def", "drop_path", "forward", "hidden_states", "layer_scale1", "layer_scale2", "mlp", "norm1", "norm2", "position_embeddings", "residual", "return", "self", "torch", "tuple"], "dinov3_vit/modeling_dinov3_vit.py:DINOv3ViTPreTrainedModel._init_weights": ["Conv2d", "LayerNorm", "Linear", "Model", "ModelEmbeddings", "ModelLayerScale", "ModelRopePositionEmbedding", "None", "PreTrainedModel", "_init_weights", "arange", "base", "bias", "class", "cls_token", "config", "constant_", "copy_", "def", "dtype", "elif", "float32", "head_dim", "if", "init", "initializer_range", "inv_freq", "is", "isinstance", "lambda1", "layerscale_value", "mask_token", "mean", "module", "nn", "not", "num_register_tokens", "ones_", "register_tokens", "self", "std", "torch", "trunc_normal_", "weight", "zeros_"], "dinov3_vit/modeling_dinov3_vit.py:DINOv3ViTModel.__init__": ["False", "LayerNorm", "Model", "ModelConfig", "ModelEmbeddings", "ModelLayer", "ModelPreTrainedModel", "ModelRopePositionEmbedding", "ModuleList", "_", "__init__", "class", "config", "def", "embeddings", "eps", "for", "gradient_checkpointing", "hidden_size", "in", "layer", "layer_norm_eps", "nn", "norm", "num_hidden_layers", "post_init", "range", "rope_embeddings", "self", "super"], "dinov3_vit/modeling_dinov3_vit.py:DINOv3ViTModel.get_input_embeddings": ["Model", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "patch_embeddings", "return", "self"], "dinov3_vit/modeling_dinov3_vit.py:DINOv3ViTModel.forward": ["BaseModelOutputWithPooling", "Model", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "bool_masked_pos", "class", "def", "dtype", "embeddings", "enumerate", "for", "forward", "hidden_states", "i", "in", "kwargs", "last_hidden_state", "layer", "layer_module", "norm", "patch_embeddings", "pixel_values", "pooled_output", "pooler_output", "position_embeddings", "r", "return", "rope_embeddings", "self", "sequence_output", "to", "torch", "weight"], "dinov3_vit/modeling_dinov3_vit.py:DINOv3ViTBackbone.__init__": ["BackboneMixin", "False", "LayerNorm", "Model", "ModelEmbeddings", "ModelLayer", "ModelPreTrainedModel", "ModelRopePositionEmbedding", "ModuleList", "_", "__init__", "_init_backbone", "class", "config", "def", "embeddings", "eps", "for", "gradient_checkpointing", "hidden_size", "in", "layer", "layer_norm_eps", "nn", "norm", "num_features", "num_hidden_layers", "post_init", "range", "rope_embeddings", "self", "super"], "dinov3_vit/modeling_dinov3_vit.py:DINOv3ViTBackbone.get_input_embeddings": ["BackboneMixin", "Model", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "patch_embeddings", "return", "self"], "dinov3_vit/modeling_dinov3_vit.py:DINOv3ViTBackbone.forward": ["BackboneMixin", "BackboneOutput", "Model", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "append", "apply_layernorm", "batch_size", "bool", "class", "config", "contiguous", "def", "dtype", "elif", "else", "embeddings", "enumerate", "feature_maps", "fmap", "for", "forward", "getattr", "hidden_state", "hidden_states", "idx", "if", "image_height", "image_width", "in", "kwargs", "last_hidden_state", "last_stage_idx", "layer", "layer_module", "len", "list", "norm", "num_patches_height", "num_patches_width", "num_prefix", "num_register_tokens", "out_features", "output", "output_hidden_states", "patch_embeddings", "patch_size", "patch_tokens", "permute", "pixel_values", "position_embeddings", "reshape", "reshape_hidden_states", "return", "rope_embeddings", "self", "sequence_output", "shape", "stage_hidden_states", "stage_name", "stage_names", "to", "torch", "tuple", "weight", "zip"], "distilbert/modeling_distilbert.py:create_sinusoidal_embeddings": ["GatheredParameters", "Model_sinusoidal_embeddings", "Tensor", "_Model_sinusoidal_embeddings", "deepspeed", "def", "dim", "distributed", "else", "get_rank", "if", "int", "is_deepspeed_zero3_enabled", "modifier_rank", "n_pos", "out", "return", "torch", "with", "zero"], "distilbert/modeling_distilbert.py:_create_sinusoidal_embeddings": ["False", "FloatTensor", "Tensor", "_create_sinusoidal_embeddings", "array", "cos", "def", "detach_", "dim", "for", "in", "int", "j", "n_pos", "np", "out", "pos", "position_enc", "power", "range", "requires_grad", "return", "sin", "torch"], "distilbert/modeling_distilbert.py:Embeddings.__init__": ["Dropout", "Embedding", "False", "LayerNorm", "Model", "Module", "PreTrainedConfig", "__init__", "arange", "class", "config", "def", "dim", "dropout", "eps", "expand", "max_position_Model", "nn", "pad_token_id", "padding_idx", "persistent", "position_Model", "position_ids", "register_buffer", "self", "super", "torch", "vocab_size", "word_Model"], "distilbert/modeling_distilbert.py:Embeddings.forward": ["LayerNorm", "LongTensor", "Model", "Module", "None", "Tensor", "arange", "class", "def", "device", "dropout", "dtype", "else", "expand_as", "forward", "hasattr", "if", "input_embeds", "input_ids", "is", "long", "nn", "not", "position_Model", "position_ids", "return", "self", "seq_length", "size", "torch", "unsqueeze", "word_Model"], "distilbert/modeling_distilbert.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "contiguous", "def", "dim", "dropout", "float", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "shape", "size", "softmax", "torch", "training", "transpose", "value"], "distilbert/modeling_distilbert.py:DistilBertSelfAttention.__init__": ["Dropout", "False", "Linear", "ModelSelfAttention", "Module", "PreTrainedConfig", "ValueError", "__init__", "attention_dropout", "attention_head_size", "class", "config", "def", "dim", "divide", "dropout", "evenly", "f", "if", "in_features", "is_causal", "k_lin", "must", "n_heads", "nn", "out_features", "out_lin", "p", "q_lin", "raise", "scaling", "self", "super", "v_lin"], "distilbert/modeling_distilbert.py:DistilBertSelfAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "FloatTensor", "ModelSelfAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_attn_implementation", "attention_head_size", "attention_interface", "attention_mask", "attn_output", "attn_weights", "class", "config", "contiguous", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "hidden_shape", "hidden_states", "if", "input_shape", "k_lin", "key_layer", "kwargs", "nn", "not", "out_lin", "p", "q_lin", "query_layer", "reshape", "return", "scaling", "self", "shape", "torch", "training", "transpose", "tuple", "v_lin", "value_layer", "view"], "distilbert/modeling_distilbert.py:FFN.__init__": ["Dropout", "Linear", "Model", "Module", "PreTrainedConfig", "__init__", "activation", "chunk_size_feed_forward", "class", "config", "def", "dim", "dropout", "get_activation", "hidden_dim", "in_features", "lin1", "lin2", "nn", "out_features", "p", "self", "seq_len_dim", "super"], "distilbert/modeling_distilbert.py:FFN.forward": ["Model", "Module", "Tensor", "apply_chunking_to_forward", "chunk_size_feed_forward", "class", "def", "ff_chunk", "forward", "input", "nn", "return", "self", "seq_len_dim", "torch"], "distilbert/modeling_distilbert.py:FFN.ff_chunk": ["Model", "Module", "Tensor", "activation", "class", "def", "dropout", "ff_chunk", "input", "lin1", "lin2", "nn", "return", "self", "torch", "x"], "distilbert/modeling_distilbert.py:TransformerBlock.__init__": ["FFN", "GradientCheckpointingLayer", "LayerNorm", "ModelBlock", "ModelSelfAttention", "PreTrainedConfig", "ValueError", "__init__", "attention", "class", "config", "def", "dim", "divide", "eps", "evenly", "f", "ffn", "if", "must", "n_heads", "nn", "normalized_shape", "output_layer_norm", "raise", "sa_layer_norm", "self", "super"], "distilbert/modeling_distilbert.py:TransformerBlock.forward": ["GradientCheckpointingLayer", "ModelBlock", "ModelsKwargs", "None", "Tensor", "Unpack", "_", "attention", "attention_mask", "attention_output", "class", "def", "ffn", "ffn_output", "forward", "hidden_states", "kwargs", "output_layer_norm", "return", "sa_layer_norm", "self", "torch", "tuple"], "distilbert/modeling_distilbert.py:Transformer.__init__": ["False", "Model", "ModelBlock", "Module", "ModuleList", "PreTrainedConfig", "_", "__init__", "class", "config", "def", "for", "gradient_checkpointing", "in", "layer", "n_layers", "nn", "range", "self", "super"], "distilbert/modeling_distilbert.py:Transformer.forward": ["BaseModelOutput", "Model", "ModelsKwargs", "Module", "None", "Tensor", "Unpack", "attention_mask", "class", "def", "for", "forward", "hidden_states", "in", "kwargs", "last_hidden_state", "layer", "layer_module", "nn", "return", "self", "torch"], "distilbert/modeling_distilbert.py:DistilBertPreTrainedModel._init_weights": ["Embeddings", "ModelPreTrainedModel", "Module", "PreTrainedModel", "_init_weights", "arange", "class", "config", "copy_", "create_sinusoidal_embeddings", "def", "dim", "empty_like", "expand", "if", "init", "isinstance", "max_position_embeddings", "module", "nn", "position_embeddings", "position_ids", "self", "shape", "sinusoidal_pos_embds", "super", "torch", "weight"], "distilbert/modeling_distilbert.py:DistilBertModel.__init__": ["Embeddings", "ModelModel", "ModelPreTrainedModel", "PreTrainedConfig", "Transformer", "__init__", "class", "config", "def", "embeddings", "post_init", "self", "super", "transformer"], "distilbert/modeling_distilbert.py:DistilBertModel.get_position_embeddings": ["Embedding", "ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "get_position_embeddings", "nn", "position_embeddings", "return", "self"], "distilbert/modeling_distilbert.py:DistilBertModel.resize_position_embeddings": ["Embedding", "ModelModel", "ModelPreTrainedModel", "Parameter", "Setting", "class", "clone", "config", "create_sinusoidal_embeddings", "def", "device", "dim", "else", "embeddings", "f", "if", "info", "int", "logger", "max_position_embeddings", "n_pos", "new_num_position_embeddings", "nn", "no_grad", "num_position_embeds_diff", "old_position_embeddings_weight", "out", "position_embeddings", "resize_position_embeddings", "return", "self", "sinusoidal_pos_embds", "to", "torch", "weight", "with"], "distilbert/modeling_distilbert.py:DistilBertModel.get_input_embeddings": ["Embedding", "ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "nn", "return", "self", "word_embeddings"], "distilbert/modeling_distilbert.py:DistilBertModel.set_input_embeddings": ["Embedding", "ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "new_embeddings", "nn", "self", "set_input_embeddings", "word_embeddings"], "distilbert/modeling_distilbert.py:DistilBertModel.forward": ["BaseModelOutput", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "attention_mask", "class", "config", "create_bidirectional_mask", "def", "embeddings", "exactly", "forward", "hidden_states", "if", "input_embeds", "input_ids", "inputs_embeds", "is", "kwargs", "must", "not", "of", "one", "or", "position_ids", "r", "raise", "return", "self", "specify", "torch", "transformer", "tuple"], "distilbert/modeling_distilbert.py:DistilBertForMaskedLM.__init__": ["CrossEntropyLoss", "LayerNorm", "Linear", "Model", "ModelForMaskedLM", "ModelModel", "ModelPreTrainedModel", "PreTrainedConfig", "__init__", "activation", "class", "config", "def", "dim", "eps", "get_activation", "mlm_loss_fct", "nn", "post_init", "self", "super", "vocab_layer_norm", "vocab_projector", "vocab_size", "vocab_transform"], "distilbert/modeling_distilbert.py:DistilBertForMaskedLM.get_position_embeddings": ["Embedding", "Model", "ModelForMaskedLM", "ModelPreTrainedModel", "class", "def", "get_position_embeddings", "nn", "return", "self"], "distilbert/modeling_distilbert.py:DistilBertForMaskedLM.resize_position_embeddings": ["Model", "ModelForMaskedLM", "ModelPreTrainedModel", "class", "def", "int", "new_num_position_embeddings", "resize_position_embeddings", "self"], "distilbert/modeling_distilbert.py:DistilBertForMaskedLM.get_output_embeddings": ["ModelForMaskedLM", "ModelPreTrainedModel", "Module", "class", "def", "get_output_embeddings", "nn", "return", "self", "vocab_projector"], "distilbert/modeling_distilbert.py:DistilBertForMaskedLM.set_output_embeddings": ["ModelForMaskedLM", "ModelPreTrainedModel", "Module", "class", "def", "new_embeddings", "nn", "self", "set_output_embeddings", "vocab_projector"], "distilbert/modeling_distilbert.py:DistilBertForMaskedLM.forward": ["LongTensor", "MaskedLMOutput", "Model", "ModelForMaskedLM", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "activation", "attention_mask", "attentions", "class", "def", "dlbrt_output", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "mlm_loss", "mlm_loss_fct", "not", "position_ids", "prediction_logits", "r", "return", "return_dict", "self", "size", "torch", "tuple", "view", "vocab_layer_norm", "vocab_projector", "vocab_transform"], "distilbert/modeling_distilbert.py:DistilBertForSequenceClassification.__init__": ["Dropout", "Linear", "Model", "ModelForSequenceClassification", "ModelModel", "ModelPreTrainedModel", "PreTrainedConfig", "__init__", "class", "classifier", "config", "def", "dim", "dropout", "nn", "num_labels", "post_init", "pre_classifier", "self", "seq_classif_dropout", "super"], "distilbert/modeling_distilbert.py:DistilBertForSequenceClassification.get_position_embeddings": ["Embedding", "Model", "ModelForSequenceClassification", "ModelPreTrainedModel", "class", "def", "get_position_embeddings", "nn", "return", "self"], "distilbert/modeling_distilbert.py:DistilBertForSequenceClassification.resize_position_embeddings": ["Model", "ModelForSequenceClassification", "ModelPreTrainedModel", "class", "def", "int", "new_num_position_embeddings", "resize_position_embeddings", "self"], "distilbert/modeling_distilbert.py:DistilBertForSequenceClassification.forward": ["BCEWithLogitsLoss", "CrossEntropyLoss", "LongTensor", "MSELoss", "Model", "ModelForSequenceClassification", "ModelPreTrainedModel", "Model_output", "None", "ReLU", "SequenceClassifierOutput", "Tensor", "TransformersKwargs", "True", "Unpack", "and", "attention_mask", "attentions", "class", "classifier", "config", "def", "dropout", "dtype", "elif", "else", "forward", "hidden_state", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "kwargs", "labels", "logits", "long", "loss", "loss_fct", "multi_label_classification", "nn", "not", "num_labels", "or", "pooled_output", "position_ids", "pre_classifier", "problem_type", "r", "regression", "return", "return_dict", "self", "single_label_classification", "squeeze", "torch", "tuple", "view"], "distilbert/modeling_distilbert.py:DistilBertForQuestionAnswering.__init__": ["Dropout", "Linear", "Model", "ModelForQuestionAnswering", "ModelModel", "ModelPreTrainedModel", "PreTrainedConfig", "ValueError", "__init__", "be", "but", "class", "config", "def", "dim", "dropout", "f", "if", "is", "it", "nn", "num_labels", "post_init", "qa_dropout", "qa_outputs", "raise", "self", "should", "super"], "distilbert/modeling_distilbert.py:DistilBertForQuestionAnswering.get_position_embeddings": ["Embedding", "Model", "ModelForQuestionAnswering", "ModelPreTrainedModel", "class", "def", "get_position_embeddings", "nn", "return", "self"], "distilbert/modeling_distilbert.py:DistilBertForQuestionAnswering.resize_position_embeddings": ["Model", "ModelForQuestionAnswering", "ModelPreTrainedModel", "class", "def", "int", "new_num_position_embeddings", "resize_position_embeddings", "self"], "distilbert/modeling_distilbert.py:DistilBertForQuestionAnswering.forward": ["CrossEntropyLoss", "Model", "ModelForQuestionAnswering", "ModelPreTrainedModel", "Model_output", "None", "QuestionAnsweringModelOutput", "Tensor", "TransformersKwargs", "True", "Unpack", "and", "attention_mask", "attentions", "clamp", "class", "contiguous", "def", "dim", "dropout", "end_logits", "end_loss", "end_positions", "forward", "hidden_states", "if", "ignore_index", "ignored_index", "input_ids", "inputs_embeds", "is", "kwargs", "len", "logits", "loss", "loss_fct", "nn", "not", "position_ids", "qa_outputs", "r", "return", "return_dict", "self", "size", "split", "squeeze", "start_logits", "start_loss", "start_positions", "torch", "total_loss", "tuple"], "distilbert/modeling_distilbert.py:DistilBertForTokenClassification.__init__": ["Dropout", "Linear", "Model", "ModelForTokenClassification", "ModelModel", "ModelPreTrainedModel", "PreTrainedConfig", "__init__", "class", "classifier", "config", "def", "dropout", "hidden_size", "nn", "num_labels", "post_init", "self", "super"], "distilbert/modeling_distilbert.py:DistilBertForTokenClassification.get_position_embeddings": ["Embedding", "Model", "ModelForTokenClassification", "ModelPreTrainedModel", "class", "def", "get_position_embeddings", "nn", "return", "self"], "distilbert/modeling_distilbert.py:DistilBertForTokenClassification.resize_position_embeddings": ["Model", "ModelForTokenClassification", "ModelPreTrainedModel", "class", "def", "int", "new_num_position_embeddings", "resize_position_embeddings", "self"], "distilbert/modeling_distilbert.py:DistilBertForTokenClassification.forward": ["CrossEntropyLoss", "LongTensor", "Model", "ModelForTokenClassification", "ModelPreTrainedModel", "None", "Tensor", "TokenClassifierOutput", "TransformersKwargs", "True", "Unpack", "attention_mask", "attentions", "class", "classifier", "def", "dropout", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "not", "num_labels", "outputs", "position_ids", "r", "return", "return_dict", "self", "sequence_output", "torch", "tuple", "view"], "distilbert/modeling_distilbert.py:DistilBertForMultipleChoice.__init__": ["Dropout", "Linear", "Model", "ModelForMultipleChoice", "ModelModel", "ModelPreTrainedModel", "PreTrainedConfig", "__init__", "class", "classifier", "config", "def", "dim", "dropout", "nn", "post_init", "pre_classifier", "self", "seq_classif_dropout", "super"], "distilbert/modeling_distilbert.py:DistilBertForMultipleChoice.get_position_embeddings": ["Embedding", "Model", "ModelForMultipleChoice", "ModelPreTrainedModel", "class", "def", "get_position_embeddings", "nn", "return", "self"], "distilbert/modeling_distilbert.py:DistilBertForMultipleChoice.resize_position_embeddings": ["Model", "ModelForMultipleChoice", "ModelPreTrainedModel", "class", "def", "int", "new_num_position_embeddings", "resize_position_embeddings", "self"], "distilbert/modeling_distilbert.py:DistilBertForMultipleChoice.forward": ["CrossEntropyLoss", "LongTensor", "Model", "ModelForMultipleChoice", "ModelPreTrainedModel", "MultipleChoiceModelOutput", "None", "ReLU", "Tensor", "TransformersKwargs", "True", "Unpack", "attention_mask", "attentions", "class", "classifier", "def", "dropout", "else", "forward", "hidden_state", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "nn", "not", "num_choices", "outputs", "pooled_output", "position_ids", "pre_classifier", "r", "reshaped_logits", "return", "return_dict", "self", "shape", "size", "torch", "tuple", "view"], "doge/modeling_doge.py:DogeRMSNorm.__init__": ["ModelRMSNorm", "Module", "Parameter", "__init__", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "doge/modeling_doge.py:DogeRMSNorm.forward": ["ModelRMSNorm", "Module", "True", "class", "def", "dtype", "float32", "forward", "hidden_states", "input_dtype", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "doge/modeling_doge.py:DogeRMSNorm.extra_repr": ["ModelRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "variance_epsilon", "weight"], "doge/modeling_doge.py:DogeRotaryEmbedding.__init__": ["Callable", "False", "ModelConfig", "ModelRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "doge/modeling_doge.py:DogeRotaryEmbedding.compute_default_rope_parameters": ["ModelConfig", "ModelRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "doge/modeling_doge.py:DogeRotaryEmbedding.forward": ["False", "ModelRotaryEmbedding", "Module", "None", "and", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "doge/modeling_doge.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "doge/modeling_doge.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cos", "def", "k", "k_embed", "q", "q_embed", "return", "rotate_half", "sin", "unsqueeze", "unsqueeze_dim"], "doge/modeling_doge.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "doge/modeling_doge.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value", "value_states"], "doge/modeling_doge.py:flex_attention_forward": ["BlockMask", "Model_attention_forward", "Module", "None", "Tensor", "True", "Union", "attention_mask", "attention_weights", "attn_output", "batch_idx", "block_mask", "causal_mask", "compile_friendly_Model_attention", "contiguous", "def", "dtype", "else", "enable_gqa", "float", "head_idx", "if", "is", "isinstance", "key", "kv_idx", "kwargs", "module", "nn", "not", "q_idx", "query", "return", "return_lse", "scale", "scaling", "score", "score_mod", "shape", "softcap", "tanh", "to", "torch", "transpose", "tuple", "value"], "doge/modeling_doge.py:DogeAttention.__init__": ["A", "Linear", "ModelAttention", "ModelConfig", "ModelRMSNorm", "Module", "None", "Parameter", "__init__", "attention_bias", "attention_dropout", "bias", "class", "config", "def", "dt_proj", "eps", "getattr", "head_dim", "hidden_size", "int", "k_norm", "k_proj", "keep_window_size", "layer_idx", "nn", "num_attention_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "q_norm", "q_proj", "rms_norm_eps", "scaling", "self", "super", "torch", "v_proj", "zeros"], "doge/modeling_doge.py:DogeAttention.forward": ["A", "ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "F", "LongTensor", "ModelAttention", "Module", "None", "Tensor", "_attn_implementation", "apply_rotary_pos_emb", "attention_dropout", "attention_interface", "attention_mask", "attn_mask", "attn_output", "attn_weights", "cache_kwargs", "cache_position", "class", "config", "contiguous", "cos", "def", "dropout", "dt_proj", "dt_states", "eager", "eager_attention_forward", "else", "exp", "forward", "head_dim", "hidden_shape", "hidden_states", "if", "input_shape", "is", "k_norm", "k_proj", "keep_window_size", "key_states", "kwargs", "layer_idx", "nn", "not", "num_key_value_groups", "o_proj", "past_key_values", "position_embeddings", "prepare_dynamic_mask", "q_norm", "q_proj", "query_states", "repeat_kv", "reshape", "return", "scaling", "self", "shape", "sin", "softplus", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "view"], "doge/modeling_doge.py:DogeAttention.prepare_dynamic_mask": ["BlockMask", "False", "ModelAttention", "Module", "None", "Tensor", "True", "active_mask", "and", "attention_mask", "attn_mask", "bool", "class", "def", "device", "dim", "dt_states", "dtype", "expand", "finfo", "hidden_states", "if", "indices", "int", "is", "isinstance", "keep_window_size", "largest", "masked_fill", "min", "min_dtype", "nn", "not", "prepare_dynamic_mask", "return", "scatter", "self", "shape", "sorted", "tensor", "topk", "topk_indices", "torch", "where", "zeros_like"], "doge/modeling_doge.py:DogeMLP.__init__": ["ACT2FN", "Linear", "ModelMLP", "Module", "__init__", "act_fn", "bias", "class", "config", "def", "down_proj", "gate_proj", "hidden_act", "hidden_size", "intermediate_size", "mlp_bias", "nn", "self", "super", "up_proj"], "doge/modeling_doge.py:DogeMLP.forward": ["ModelMLP", "Module", "act_fn", "class", "def", "down_proj", "forward", "gate_proj", "nn", "return", "self", "up_proj", "x"], "doge/modeling_doge.py:DogeCDMoE.__init__": ["ACT2FN", "Embedding", "False", "Linear", "ModelCDMoE", "ModelConfig", "Module", "__init__", "act_fn", "bias", "class", "config", "def", "down_embed", "down_proj", "floor", "gate_proj", "hidden_act", "hidden_size", "intermediate_size", "math", "mlp_bias", "nn", "norm_topk_prob", "num_experts", "num_experts_per_tok", "num_keys", "router_gate", "self", "sqrt", "super", "top_k", "up_embed", "up_proj"], "doge/modeling_doge.py:DogeCDMoE.forward": ["F", "ModelCDMoE", "Module", "Tensor", "True", "_", "act_fn", "all_indices", "all_scores", "bsz", "class", "def", "dim", "down_embed", "down_proj", "experts_states", "experts_weights", "forward", "gate_proj", "gather", "hidden_states", "if", "indices", "indices_x", "indices_y", "keepdim", "kwargs", "matmul", "nn", "norm_topk_prob", "num_keys", "position_indices", "return", "router_gate", "router_logits", "routing_weights", "scores", "scores_x", "scores_y", "self", "seq_len", "shape", "softmax", "sum", "top_k", "topk", "torch", "unsqueeze", "up_embed", "up_proj", "view"], "doge/modeling_doge.py:DogeDecoderLayer.__init__": ["GradientCheckpointingLayer", "ModelAttention", "ModelCDMoE", "ModelConfig", "ModelDecoderLayer", "ModelMLP", "ModelRMSNorm", "None", "Parameter", "__init__", "class", "config", "def", "else", "eps", "hidden_dropout", "hidden_size", "if", "input_layernorm", "input_residual", "int", "is_moe", "layer_idx", "mlp", "nn", "not", "ones", "post_attention_layernorm", "post_attention_residual", "rms_norm_eps", "self", "self_attn", "super", "torch"], "doge/modeling_doge.py:DogeDecoderLayer.forward": ["Cache", "F", "False", "FloatTensor", "GradientCheckpointingLayer", "LongTensor", "ModelDecoderLayer", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "bool", "cache_position", "class", "def", "dropout", "forward", "hidden_dropout", "hidden_states", "input_layernorm", "input_residual", "kwargs", "mlp", "p", "past_key_values", "position_embeddings", "position_ids", "post_attention_layernorm", "post_attention_residual", "residual", "return", "self", "self_attn", "self_attn_weights", "torch", "training", "tuple", "use_cache"], "doge/modeling_doge.py:DogePreTrainedModel._init_weights": ["A", "ModelAttention", "ModelDecoderLayer", "ModelPreTrainedModel", "PreTrainedModel", "_init_weights", "class", "def", "elif", "hasattr", "if", "init", "input_residual", "isinstance", "module", "ones_", "post_attention_residual", "self", "super", "zeros_"], "doge/modeling_doge.py:DogeModel.__init__": ["Embedding", "False", "ModelConfig", "ModelDecoderLayer", "ModelModel", "ModelPreTrainedModel", "ModelRMSNorm", "ModelRotaryEmbedding", "ModuleList", "__init__", "class", "config", "def", "embed_tokens", "eps", "for", "gradient_checkpointing", "hidden_size", "in", "layer_idx", "layers", "nn", "norm", "num_hidden_layers", "pad_token_id", "padding_idx", "post_init", "range", "rms_norm_eps", "rotary_emb", "self", "super", "vocab_size"], "doge/modeling_doge.py:DogeModel.forward": ["Cache", "DynamicCache", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "MoeModelOutputWithPast", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "and", "arange", "attention_mask", "bool", "cache_position", "causal_mask", "class", "config", "create_causal_mask", "create_sliding_window_causal_mask", "decoder_layer", "def", "device", "else", "embed_tokens", "exactly", "for", "forward", "get_seq_length", "hidden_states", "if", "in", "input_embeds", "input_ids", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layers", "mask_function", "must", "norm", "not", "num_hidden_layers", "of", "one", "or", "past_key_values", "past_seen_tokens", "position_embeddings", "position_ids", "raise", "return", "rotary_emb", "self", "shape", "sliding_window", "specify", "torch", "unsqueeze", "use_cache"], "doge/modeling_doge.py:load_balancing_loss_func": ["F", "Model_balancing_loss_func", "None", "Tensor", "_", "all_expert_indices", "all_indices", "all_routing_weights", "all_scores", "append", "attention_mask", "batch_size", "bool", "cat", "compute_device", "compute_dtype", "def", "device", "dim", "dtype", "else", "expand", "expert_attention_mask", "expert_indices", "for", "gate_logits", "gather", "if", "in", "indices_x", "indices_y", "int", "is", "isinstance", "layer_gate_logits", "len", "mean", "not", "num_experts", "num_hidden_layers", "num_keys", "ones_like", "or", "overall_loss", "pad", "position_indices", "r", "reshape", "return", "router_per_expert_attention_mask", "router_prob_per_expert", "routing_weights", "scatter_add_", "scores_x", "scores_y", "sequence_length", "shape", "softmax", "sum", "to", "tokens_per_expert", "top_k", "topk", "torch", "tuple", "unsqueeze", "view", "zeros"], "doge/modeling_doge.py:DogeForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "ModelForCausalLM", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "num_experts", "num_experts_per_tok", "post_init", "router_aux_loss_coef", "self", "super", "vocab_size"], "doge/modeling_doge.py:DogeForCausalLM.forward": ["Cache", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForCausalLM", "ModelPreTrainedModel", "MoeCausalLMOutputWithPast", "MoeModelOutputWithPast", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "aux_loss", "bool", "cache_position", "class", "config", "def", "device", "else", "floor", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "load_balancing_loss_func", "logits", "logits_to_keep", "loss", "loss_function", "math", "model", "not", "num_experts", "num_experts_per_tok", "output_router_logits", "outputs", "past_key_values", "position_ids", "r", "return", "router_aux_loss_coef", "router_logits", "self", "slice", "slice_indices", "sqrt", "to", "torch", "use_cache", "vocab_size"], "donut/modeling_donut_swin.py:window_partition": ["Model_partition", "Model_size", "Models", "batch_size", "contiguous", "def", "height", "input_feature", "num_channels", "permute", "return", "shape", "view", "width"], "donut/modeling_donut_swin.py:window_reverse": ["Model_reverse", "Model_size", "Models", "contiguous", "def", "height", "num_channels", "permute", "return", "shape", "view", "width"], "donut/modeling_donut_swin.py:DonutSwinEmbeddings.__init__": ["Dropout", "False", "LayerNorm", "ModelSwinEmbeddings", "ModelSwinPatchEmbeddings", "Module", "None", "Parameter", "__init__", "class", "config", "def", "dropout", "else", "embed_dim", "grid_size", "hidden_dropout_prob", "if", "mask_token", "nn", "norm", "num_patches", "patch_embeddings", "patch_grid", "patch_size", "position_embeddings", "self", "super", "torch", "use_absolute_embeddings", "use_mask_token", "zeros"], "donut/modeling_donut_swin.py:DonutSwinEmbeddings.interpolate_pos_encoding": ["False", "ModelSwinEmbeddings", "Module", "Tensor", "align_corners", "and", "bicubic", "cat", "class", "class_pos_embed", "def", "dim", "embeddings", "functional", "height", "if", "int", "interpolate", "interpolate_pos_encoding", "is_tracing", "jit", "mode", "new_height", "new_width", "nn", "not", "num_patches", "num_positions", "patch_pos_embed", "patch_size", "permute", "position_embeddings", "reshape", "return", "self", "shape", "size", "sqrt_num_positions", "torch", "torch_int", "view", "width"], "donut/modeling_donut_swin.py:DonutSwinEmbeddings.forward": ["BoolTensor", "False", "FloatTensor", "ModelSwinEmbeddings", "Module", "None", "Tensor", "_", "batch_size", "bool", "bool_masked_pos", "class", "def", "dropout", "else", "embeddings", "expand", "forward", "height", "if", "interpolate_pos_encoding", "is", "mask", "mask_token", "mask_tokens", "nn", "norm", "not", "num_channels", "output_dimensions", "patch_embeddings", "pixel_values", "position_embeddings", "return", "self", "seq_len", "shape", "size", "torch", "tuple", "type_as", "unsqueeze", "width"], "donut/modeling_donut_swin.py:DonutSwinPatchEmbeddings.__init__": ["Conv2d", "Iterable", "ModelSwinPatchEmbeddings", "Module", "__init__", "abc", "class", "collections", "config", "def", "else", "embed_dim", "grid_size", "hidden_size", "if", "image_size", "isinstance", "kernel_size", "nn", "num_channels", "num_patches", "patch_size", "projection", "self", "stride", "super"], "donut/modeling_donut_swin.py:DonutSwinPatchEmbeddings.maybe_pad": ["ModelSwinPatchEmbeddings", "Module", "class", "def", "functional", "height", "if", "maybe_pad", "nn", "pad", "pad_values", "patch_size", "pixel_values", "return", "self", "width"], "donut/modeling_donut_swin.py:DonutSwinPatchEmbeddings.forward": ["FloatTensor", "ModelSwinPatchEmbeddings", "Module", "None", "Tensor", "_", "class", "def", "embeddings", "flatten", "forward", "height", "int", "maybe_pad", "nn", "num_channels", "output_dimensions", "pixel_values", "projection", "return", "self", "shape", "torch", "transpose", "tuple", "width"], "donut/modeling_donut_swin.py:DonutSwinPatchMerging.__init__": ["False", "LayerNorm", "Linear", "ModelSwinPatchMerging", "Module", "None", "__init__", "bias", "class", "def", "dim", "input_resolution", "int", "nn", "norm", "norm_layer", "reduction", "self", "super", "tuple"], "donut/modeling_donut_swin.py:DonutSwinPatchMerging.maybe_pad": ["ModelSwinPatchMerging", "Module", "class", "def", "functional", "height", "if", "input_feature", "maybe_pad", "nn", "or", "pad", "pad_values", "return", "self", "should_pad", "width"], "donut/modeling_donut_swin.py:DonutSwinPatchMerging.forward": ["ModelSwinPatchMerging", "Module", "Tensor", "batch_size", "cat", "class", "def", "dim", "forward", "height", "input_dimensions", "input_feature", "input_feature_0", "input_feature_1", "input_feature_2", "input_feature_3", "int", "maybe_pad", "nn", "norm", "num_channels", "reduction", "return", "self", "shape", "torch", "tuple", "view", "width"], "donut/modeling_donut_swin.py:drop_path": ["False", "Model_path", "Model_prob", "Tensor", "bool", "def", "device", "div", "dtype", "float", "floor_", "if", "input", "keep_prob", "ndim", "not", "or", "output", "rand", "random_tensor", "return", "shape", "torch", "training"], "donut/modeling_donut_swin.py:DonutSwinDropPath.__init__": ["ModelSwinDropPath", "Module", "None", "__init__", "class", "def", "drop_prob", "float", "nn", "self", "super"], "donut/modeling_donut_swin.py:DonutSwinDropPath.forward": ["ModelSwinDropPath", "Module", "Tensor", "class", "def", "drop_path", "drop_prob", "forward", "hidden_states", "nn", "return", "self", "torch", "training"], "donut/modeling_donut_swin.py:DonutSwinDropPath.extra_repr": ["ModelSwinDropPath", "Module", "class", "def", "drop_prob", "extra_repr", "f", "nn", "p", "return", "self", "str"], "donut/modeling_donut_swin.py:DonutSwinSelfAttention.__init__": ["Dropout", "Iterable", "Linear", "ModelSwinSelfAttention", "Module", "Parameter", "The", "ValueError", "__init__", "a", "abc", "all_head_size", "attention", "attention_head_size", "attention_probs_dropout_prob", "bias", "class", "collections", "config", "create_relative_position_index", "def", "dim", "dropout", "else", "f", "heads", "hidden", "if", "int", "is", "isinstance", "key", "multiple", "nn", "not", "num_attention_heads", "num_heads", "number", "of", "qkv_bias", "query", "raise", "register_buffer", "relative_position_bias_table", "relative_position_index", "self", "size", "super", "the", "torch", "value", "window_size", "zeros"], "donut/modeling_donut_swin.py:DonutSwinSelfAttention.forward": ["False", "FloatTensor", "ModelSwinSelfAttention", "Module", "None", "Tensor", "all_head_size", "attention_head_size", "attention_mask", "attention_probs", "attention_scores", "batch_size", "bool", "class", "context_layer", "contiguous", "def", "dim", "dropout", "else", "forward", "functional", "hidden_shape", "hidden_states", "if", "is", "key", "key_layer", "mask_shape", "math", "matmul", "new_context_layer_shape", "nn", "not", "num_attention_heads", "num_channels", "output_attentions", "outputs", "permute", "query", "query_layer", "relative_position_bias", "relative_position_bias_table", "relative_position_index", "return", "self", "shape", "size", "softmax", "sqrt", "torch", "transpose", "tuple", "unsqueeze", "value", "value_layer", "view", "window_size"], "donut/modeling_donut_swin.py:DonutSwinSelfAttention.create_relative_position_index": ["ModelSwinSelfAttention", "Module", "None", "arange", "class", "contiguous", "coords", "coords_flatten", "coords_h", "coords_w", "create_relative_position_index", "def", "flatten", "ij", "indexing", "meshgrid", "nn", "permute", "relative_coords", "relative_position_index", "return", "self", "stack", "sum", "torch", "window_size"], "donut/modeling_donut_swin.py:DonutSwinSelfOutput.__init__": ["Dropout", "Linear", "ModelSwinSelfOutput", "Module", "__init__", "attention_probs_dropout_prob", "class", "config", "def", "dense", "dim", "dropout", "nn", "self", "super"], "donut/modeling_donut_swin.py:DonutSwinSelfOutput.forward": ["ModelSwinSelfOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "donut/modeling_donut_swin.py:DonutSwinAttention.__init__": ["ModelSwinAttention", "ModelSwinSelfAttention", "ModelSwinSelfOutput", "Module", "__init__", "class", "config", "def", "dim", "nn", "num_heads", "output", "self", "super", "window_size"], "donut/modeling_donut_swin.py:DonutSwinAttention.forward": ["False", "FloatTensor", "ModelSwinAttention", "Module", "None", "Tensor", "attention_mask", "attention_output", "bool", "class", "def", "forward", "hidden_states", "nn", "output", "output_attentions", "outputs", "return", "self", "self_outputs", "torch", "tuple"], "donut/modeling_donut_swin.py:DonutSwinIntermediate.__init__": ["ACT2FN", "Linear", "ModelSwinIntermediate", "Module", "__init__", "class", "config", "def", "dense", "dim", "else", "hidden_act", "if", "int", "intermediate_act_fn", "isinstance", "mlp_ratio", "nn", "self", "str", "super"], "donut/modeling_donut_swin.py:DonutSwinIntermediate.forward": ["ModelSwinIntermediate", "Module", "Tensor", "class", "def", "dense", "forward", "hidden_states", "intermediate_act_fn", "nn", "return", "self", "torch"], "donut/modeling_donut_swin.py:DonutSwinOutput.__init__": ["Dropout", "Linear", "ModelSwinOutput", "Module", "__init__", "class", "config", "def", "dense", "dim", "dropout", "hidden_dropout_prob", "int", "mlp_ratio", "nn", "self", "super"], "donut/modeling_donut_swin.py:DonutSwinOutput.forward": ["ModelSwinOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "nn", "return", "self", "torch"], "donut/modeling_donut_swin.py:DonutSwinLayer.__init__": ["Identity", "LayerNorm", "ModelSwinAttention", "ModelSwinDropPath", "ModelSwinIntermediate", "ModelSwinLayer", "ModelSwinOutput", "Module", "__init__", "attention", "chunk_size_feed_forward", "class", "config", "def", "dim", "drop_path", "drop_path_rate", "else", "eps", "if", "input_resolution", "intermediate", "layer_norm_eps", "layernorm_after", "layernorm_before", "nn", "num_heads", "output", "self", "shift_size", "super", "window_size"], "donut/modeling_donut_swin.py:DonutSwinLayer.set_shift_and_window_size": ["ModelSwinLayer", "Module", "class", "def", "else", "if", "input_resolution", "is_tracing", "jit", "min", "nn", "self", "set_shift_and_window_size", "shift_size", "tensor", "torch", "torch_int", "window_size"], "donut/modeling_donut_swin.py:DonutSwinLayer.get_attn_mask": ["ModelSwinLayer", "Module", "None", "attn_mask", "class", "count", "def", "device", "dtype", "else", "for", "get_attn_mask", "height", "height_slice", "height_slices", "if", "img_mask", "in", "mask_windows", "masked_fill", "nn", "return", "self", "shift_size", "slice", "torch", "unsqueeze", "view", "width", "width_slice", "width_slices", "window_partition", "window_size", "zeros"], "donut/modeling_donut_swin.py:DonutSwinLayer.maybe_pad": ["ModelSwinLayer", "Module", "class", "def", "functional", "height", "hidden_states", "maybe_pad", "nn", "pad", "pad_bottom", "pad_right", "pad_values", "return", "self", "width", "window_size"], "donut/modeling_donut_swin.py:DonutSwinLayer.forward": ["False", "ModelSwinLayer", "Module", "None", "Tensor", "_", "always_partition", "attention", "attention_output", "attention_outputs", "attention_windows", "attn_mask", "batch_size", "bool", "channels", "class", "contiguous", "def", "device", "dims", "drop_path", "dtype", "else", "forward", "get_attn_mask", "height", "height_pad", "hidden_states", "hidden_states_windows", "if", "input_dimensions", "int", "intermediate", "layer_output", "layer_outputs", "layernorm_after", "layernorm_before", "maybe_pad", "nn", "not", "or", "output", "output_attentions", "pad_values", "pass", "return", "roll", "self", "set_shift_and_window_size", "shape", "shift_size", "shifted_hidden_states", "shifted_windows", "shifts", "shortcut", "size", "torch", "tuple", "view", "was_padded", "width", "width_pad", "window_partition", "window_reverse", "window_size"], "donut/modeling_donut_swin.py:DonutSwinStage.__init__": ["False", "GradientCheckpointingLayer", "LayerNorm", "ModelSwinLayer", "ModelSwinStage", "ModuleList", "None", "__init__", "blocks", "class", "config", "def", "depth", "dim", "downsample", "drop_path", "drop_path_rate", "else", "for", "i", "if", "in", "input_resolution", "is", "nn", "norm_layer", "not", "num_heads", "pointing", "range", "self", "shift_size", "super", "window_size"], "donut/modeling_donut_swin.py:DonutSwinStage.forward": ["False", "GradientCheckpointingLayer", "ModelSwinStage", "None", "Tensor", "always_partition", "blocks", "bool", "class", "def", "downsample", "else", "enumerate", "for", "forward", "height", "height_downsampled", "hidden_states", "hidden_states_before_downsampling", "i", "if", "in", "input_dimensions", "int", "is", "layer_module", "layer_outputs", "not", "output_attentions", "output_dimensions", "return", "self", "stage_outputs", "torch", "tuple", "width", "width_downsampled"], "donut/modeling_donut_swin.py:DonutSwinEncoder.__init__": ["False", "ModelSwinEncoder", "ModelSwinPatchMerging", "ModelSwinStage", "Module", "ModuleList", "None", "__init__", "class", "config", "cpu", "def", "depth", "depths", "device", "dim", "downsample", "dpr", "drop_path", "drop_path_rate", "else", "embed_dim", "for", "gradient_checkpointing", "grid_size", "i_layer", "if", "in", "input_resolution", "int", "item", "layers", "len", "linspace", "nn", "num_heads", "num_layers", "range", "self", "sum", "super", "torch", "x"], "donut/modeling_donut_swin.py:DonutSwinEncoder.forward": ["False", "ModelSwinEncoder", "ModelSwinEncoderOutput", "Module", "None", "Tensor", "True", "_", "all_hidden_states", "all_reshaped_hidden_states", "all_self_attentions", "always_partition", "and", "attentions", "batch_size", "bool", "class", "def", "elif", "else", "enumerate", "for", "forward", "hidden_size", "hidden_states", "hidden_states_before_downsampling", "i", "if", "in", "input_dimensions", "int", "is", "last_hidden_state", "layer_module", "layer_outputs", "layers", "nn", "not", "output_attentions", "output_dimensions", "output_hidden_states", "output_hidden_states_before_downsampling", "permute", "reshaped_hidden_state", "reshaped_hidden_states", "return", "return_dict", "self", "shape", "torch", "tuple", "v", "view"], "donut/modeling_donut_swin.py:DonutSwinPreTrainedModel._init_weights": ["ModelSwinEmbeddings", "ModelSwinPreTrainedModel", "ModelSwinSelfAttention", "None", "PreTrainedModel", "_init_weights", "class", "copy_", "create_relative_position_index", "def", "elif", "if", "init", "is", "isinstance", "mask_token", "module", "not", "position_embeddings", "relative_position_bias_table", "relative_position_index", "self", "super", "zeros_"], "donut/modeling_donut_swin.py:DonutSwinModel.__init__": ["AdaptiveAvgPool1d", "False", "ModelSwinEmbeddings", "ModelSwinEncoder", "ModelSwinModel", "ModelSwinPreTrainedModel", "None", "True", "__init__", "add_pooling_layer", "class", "config", "def", "depths", "else", "embed_dim", "embeddings", "encoder", "if", "int", "len", "nn", "num_features", "num_layers", "patch_grid", "pooler", "post_init", "r", "self", "super", "use_mask_token"], "donut/modeling_donut_swin.py:DonutSwinModel.get_input_embeddings": ["ModelSwinModel", "ModelSwinPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "patch_embeddings", "return", "self"], "donut/modeling_donut_swin.py:DonutSwinModel.forward": ["BoolTensor", "False", "FloatTensor", "ModelSwinModel", "ModelSwinModelOutput", "ModelSwinPreTrainedModel", "None", "ValueError", "You", "attentions", "bool", "bool_masked_pos", "class", "config", "def", "else", "embedding_output", "embeddings", "encoder", "encoder_outputs", "flatten", "forward", "have", "hidden_states", "if", "input_dimensions", "interpolate_pos_encoding", "is", "kwargs", "last_hidden_state", "not", "output", "output_attentions", "output_hidden_states", "pixel_values", "pooled_output", "pooler", "pooler_output", "r", "raise", "reshaped_hidden_states", "return", "return_dict", "self", "sequence_output", "specify", "to", "torch", "transpose", "tuple", "use_return_dict"], "donut/modeling_donut_swin.py:DonutSwinForImageClassification.__init__": ["Identity", "Linear", "Model", "ModelSwinForImageClassification", "ModelSwinModel", "ModelSwinPreTrainedModel", "__init__", "class", "classifier", "config", "def", "else", "if", "nn", "num_features", "num_labels", "post_init", "self", "super"], "donut/modeling_donut_swin.py:DonutSwinForImageClassification.forward": ["False", "FloatTensor", "LongTensor", "Model", "ModelSwinForImageClassification", "ModelSwinImageClassifierOutput", "ModelSwinPreTrainedModel", "None", "attentions", "bool", "class", "classifier", "config", "def", "else", "forward", "hidden_states", "if", "interpolate_pos_encoding", "is", "kwargs", "labels", "logits", "loss", "loss_function", "not", "output", "output_attentions", "output_hidden_states", "outputs", "pixel_values", "pooled_output", "r", "reshaped_hidden_states", "return", "return_dict", "self", "torch", "tuple", "use_return_dict"], "dots1/modeling_dots1.py:Dots1RMSNorm.__init__": ["ModelRMSNorm", "Module", "None", "Parameter", "__init__", "class", "def", "eps", "float", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "dots1/modeling_dots1.py:Dots1RMSNorm.forward": ["ModelRMSNorm", "Module", "Tensor", "True", "class", "def", "dtype", "float32", "forward", "hidden_states", "input_dtype", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "dots1/modeling_dots1.py:Dots1RMSNorm.extra_repr": ["ModelRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "variance_epsilon", "weight"], "dots1/modeling_dots1.py:Dots1RotaryEmbedding.__init__": ["Callable", "False", "ModelConfig", "ModelRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "dots1/modeling_dots1.py:Dots1RotaryEmbedding.compute_default_rope_parameters": ["ModelConfig", "ModelRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "dots1/modeling_dots1.py:Dots1RotaryEmbedding.forward": ["False", "ModelRotaryEmbedding", "Module", "None", "and", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "dots1/modeling_dots1.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "dots1/modeling_dots1.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cos", "def", "k", "k_embed", "q", "q_embed", "return", "rotate_half", "sin", "unsqueeze", "unsqueeze_dim"], "dots1/modeling_dots1.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "dots1/modeling_dots1.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value", "value_states"], "dots1/modeling_dots1.py:Dots1Attention.__init__": ["Linear", "ModelAttention", "ModelConfig", "ModelRMSNorm", "Module", "None", "True", "__init__", "attention_bias", "attention_dropout", "bias", "class", "config", "def", "else", "eps", "getattr", "hasattr", "head_dim", "hidden_size", "if", "int", "is_causal", "k_norm", "k_proj", "layer_idx", "layer_type", "layer_types", "nn", "num_attention_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "q_norm", "q_proj", "rms_norm_eps", "scaling", "self", "sliding_attention", "sliding_window", "super", "v_proj"], "dots1/modeling_dots1.py:Dots1Attention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "FlashAttentionKwargs", "LongTensor", "ModelAttention", "Module", "None", "Tensor", "Unpack", "_attn_implementation", "apply_rotary_pos_emb", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "cache_kwargs", "cache_position", "class", "config", "contiguous", "cos", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "head_dim", "hidden_shape", "hidden_states", "if", "input_shape", "is", "k_norm", "k_proj", "key_states", "kwargs", "layer_idx", "nn", "not", "o_proj", "past_key_values", "position_embeddings", "q_norm", "q_proj", "query_states", "reshape", "return", "scaling", "self", "shape", "sin", "sliding_window", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "view"], "dots1/modeling_dots1.py:Dots1MLP.__init__": ["ACT2FN", "False", "Linear", "ModelMLP", "Module", "None", "__init__", "act_fn", "bias", "class", "config", "def", "down_proj", "else", "gate_proj", "hidden_act", "hidden_size", "if", "intermediate_size", "is", "nn", "self", "super", "up_proj"], "dots1/modeling_dots1.py:Dots1MLP.forward": ["ModelMLP", "Module", "act_fn", "class", "def", "down_proj", "forward", "gate_proj", "nn", "return", "self", "up_proj", "x"], "dots1/modeling_dots1.py:Dots1TopkRouter.__init__": ["ModelTopkRouter", "Module", "Parameter", "__init__", "class", "config", "def", "e_score_correction_bias", "empty", "hidden_size", "n_routed_experts", "nn", "register_buffer", "self", "super", "torch", "weight", "zeros"], "dots1/modeling_dots1.py:Dots1TopkRouter.forward": ["F", "ModelTopkRouter", "Module", "class", "config", "def", "float32", "forward", "hidden_size", "hidden_states", "linear", "nn", "return", "router_logits", "self", "torch", "type", "view", "weight"], "dots1/modeling_dots1.py:Dots1NaiveMoe.__init__": ["ACT2FN", "ModelNaiveMoe", "Module", "Parameter", "__init__", "act_fn", "class", "config", "def", "down_proj", "empty", "gate_up_proj", "hidden_act", "hidden_dim", "hidden_size", "intermediate_dim", "moe_intermediate_size", "nn", "num_experts", "num_local_experts", "self", "super", "torch"], "dots1/modeling_dots1.py:Dots1NaiveMoe.forward": ["ModelNaiveMoe", "Module", "None", "Tensor", "act_fn", "chunk", "class", "continue", "current_hidden_states", "current_state", "def", "dim", "down_proj", "dtype", "expert_hit", "expert_idx", "expert_mask", "final_hidden_states", "for", "forward", "functional", "gate", "gate_up_proj", "greater", "hidden_states", "if", "in", "index_add_", "linear", "nn", "no_grad", "nonzero", "num_classes", "num_experts", "one_hot", "permute", "return", "self", "sum", "to", "token_idx", "top_k_index", "top_k_pos", "top_k_weights", "torch", "up", "where", "with", "zeros_like"], "dots1/modeling_dots1.py:Dots1MoE.__init__": ["ModelMLP", "ModelMoE", "ModelNaiveMoe", "ModelTopkRouter", "Module", "__init__", "class", "config", "def", "experts", "gate", "intermediate_size", "moe_intermediate_size", "n_group", "n_routed_experts", "n_shared_experts", "nn", "norm_topk_prob", "num_experts_per_tok", "routed_scaling_factor", "self", "shared_experts", "super", "top_k", "topk_group"], "dots1/modeling_dots1.py:Dots1MoE.route_tokens_to_experts": ["False", "ModelMoE", "Module", "True", "bool", "class", "def", "denominator", "dim", "e_score_correction_bias", "expand", "gate", "gather", "group_idx", "group_mask", "group_scores", "if", "k", "keepdim", "masked_fill", "n_group", "n_routed_experts", "nn", "norm_topk_prob", "reshape", "return", "route_tokens_to_experts", "routed_scaling_factor", "router_logits", "router_logits_for_choice", "scatter_", "score_mask", "scores_for_choice", "self", "sigmoid", "sorted", "sum", "top_k", "topk", "topk_group", "topk_indices", "topk_weights", "torch", "unsqueeze", "view", "zeros_like"], "dots1/modeling_dots1.py:Dots1MoE.forward": ["ModelMoE", "Module", "class", "def", "experts", "forward", "gate", "hidden_states", "nn", "orig_shape", "residuals", "return", "route_tokens_to_experts", "router_logits", "self", "shape", "shared_experts", "topk_indices", "topk_weights", "view"], "dots1/modeling_dots1.py:Dots1DecoderLayer.__init__": ["GradientCheckpointingLayer", "ModelAttention", "ModelConfig", "ModelDecoderLayer", "ModelMLP", "ModelMoE", "ModelRMSNorm", "__init__", "attention_type", "class", "config", "def", "else", "eps", "first_k_dense_replace", "hidden_size", "if", "input_layernorm", "int", "layer_idx", "layer_types", "mlp", "post_attention_layernorm", "rms_norm_eps", "self", "self_attn", "super"], "dots1/modeling_dots1.py:Dots1DecoderLayer.forward": ["Cache", "False", "GradientCheckpointingLayer", "LongTensor", "ModelDecoderLayer", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "attention_mask", "bool", "cache_position", "class", "def", "forward", "hidden_states", "input_layernorm", "kwargs", "mlp", "past_key_values", "position_embeddings", "position_ids", "post_attention_layernorm", "residual", "return", "self", "self_attn", "torch", "tuple", "use_cache"], "dots1/modeling_dots1.py:Dots1PreTrainedModel._init_weights": ["ModelNaiveMoe", "ModelPreTrainedModel", "ModelTopkRouter", "PreTrainedModel", "_init_weights", "class", "config", "def", "down_proj", "e_score_correction_bias", "elif", "gate_up_proj", "if", "init", "initializer_range", "isinstance", "mean", "module", "normal_", "self", "std", "super", "weight", "zeros_"], "dots1/modeling_dots1.py:Dots1Model.__init__": ["Embedding", "False", "ModelConfig", "ModelDecoderLayer", "ModelModel", "ModelPreTrainedModel", "ModelRMSNorm", "ModelRotaryEmbedding", "ModuleList", "__init__", "class", "config", "def", "embed_tokens", "eps", "for", "gradient_checkpointing", "has_sliding_layers", "hidden_size", "in", "layer_idx", "layer_types", "layers", "nn", "norm", "num_hidden_layers", "pad_token_id", "padding_idx", "post_init", "range", "rms_norm_eps", "rotary_emb", "self", "sliding_attention", "super", "vocab_size"], "dots1/modeling_dots1.py:Dots1Model.forward": ["BaseModelOutputWithPast", "Cache", "DynamicCache", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "and", "arange", "attention_mask", "attention_type", "bool", "cache_position", "causal_mask_mapping", "class", "config", "create_causal_mask", "create_sliding_window_causal_mask", "decoder_layer", "def", "device", "dict", "else", "embed_tokens", "exactly", "for", "forward", "full_attention", "get_seq_length", "has_sliding_layers", "hidden_states", "if", "in", "input_embeds", "input_ids", "inputs_embeds", "is", "isinstance", "kwargs", "last_hidden_state", "layers", "mask_kwargs", "must", "norm", "not", "num_hidden_layers", "of", "one", "or", "past_key_values", "past_seen_tokens", "position_embeddings", "position_ids", "raise", "return", "rotary_emb", "self", "shape", "sliding_attention", "specify", "torch", "unsqueeze", "use_cache"], "dots1/modeling_dots1.py:Dots1ForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "ModelForCausalLM", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "vocab_size"], "dots1/modeling_dots1.py:Dots1ForCausalLM.forward": ["BaseModelOutputWithPast", "Cache", "CausalLMOutputWithPast", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForCausalLM", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "outputs", "past_key_values", "position_ids", "r", "return", "self", "slice", "slice_indices", "torch", "use_cache", "vocab_size"], "dpr/modeling_dpr.py:DPREncoder.__init__": ["BertModel", "Encoder", "False", "Linear", "Model", "ModelConfig", "ModelPreTrainedModel", "ValueError", "__init__", "add_pooling_layer", "be", "bert_model", "can", "class", "config", "def", "encode_proj", "hidden_size", "if", "nn", "post_init", "projection_dim", "raise", "self", "super", "t", "zero"], "dpr/modeling_dpr.py:DPREncoder.forward": ["BaseModelOutputWithPooling", "False", "Model", "ModelPreTrainedModel", "None", "Tensor", "attention_mask", "attentions", "bert_model", "bool", "class", "def", "encode_proj", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "kwargs", "last_hidden_state", "not", "output_attentions", "output_hidden_states", "outputs", "pooled_output", "pooler_output", "projection_dim", "return", "return_dict", "self", "sequence_output", "token_type_ids", "tuple"], "dpr/modeling_dpr.py:DPREncoder.embeddings_size": ["Model", "ModelPreTrainedModel", "bert_model", "class", "config", "def", "embeddings_size", "encode_proj", "hidden_size", "if", "int", "out_features", "projection_dim", "return", "self"], "dpr/modeling_dpr.py:DPRSpanPredictor.__init__": ["Linear", "Model", "ModelConfig", "ModelEncoder", "ModelPreTrainedModel", "__init__", "class", "config", "def", "embeddings_size", "encoder", "nn", "post_init", "qa_classifier", "qa_outputs", "self", "super"], "dpr/modeling_dpr.py:DPRSpanPredictor.forward": ["False", "Model", "ModelPreTrainedModel", "ModelReaderOutput", "None", "Tensor", "attention_mask", "attentions", "bool", "class", "contiguous", "def", "dim", "else", "encoder", "end_logits", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "logits", "n_passages", "not", "output_attentions", "output_hidden_states", "outputs", "qa_classifier", "qa_outputs", "relevance_logits", "return", "return_dict", "self", "sequence_length", "sequence_output", "size", "split", "squeeze", "start_logits", "tuple", "view"], "dpr/modeling_dpr.py:DPRContextEncoder.__init__": ["Model", "ModelConfig", "ModelEncoder", "ModelPretrainedContextEncoder", "__init__", "class", "config", "ctx_encoder", "def", "post_init", "self", "super"], "dpr/modeling_dpr.py:DPRContextEncoder.forward": ["Model", "ModelOutput", "ModelPretrainedContextEncoder", "None", "Tensor", "ValueError", "You", "and", "at", "attention_mask", "attentions", "bool", "both", "cannot", "class", "config", "ctx_encoder", "def", "device", "dtype", "either", "elif", "else", "forward", "have", "hidden_states", "if", "input_ids", "input_shape", "inputs_embeds", "is", "kwargs", "long", "not", "ones", "or", "output_attentions", "output_hidden_states", "outputs", "pad_token_id", "pooler_output", "r", "raise", "return", "return_dict", "same", "self", "size", "specify", "the", "time", "to", "token_type_ids", "torch", "tuple", "use_return_dict", "zeros"], "dpr/modeling_dpr.py:DPRQuestionEncoder.__init__": ["Model", "ModelConfig", "ModelEncoder", "ModelPretrainedQuestionEncoder", "__init__", "class", "config", "def", "post_init", "question_encoder", "self", "super"], "dpr/modeling_dpr.py:DPRQuestionEncoder.forward": ["Model", "ModelOutput", "ModelPretrainedQuestionEncoder", "None", "Tensor", "ValueError", "You", "and", "at", "attention_mask", "attentions", "bool", "both", "cannot", "class", "config", "def", "device", "dtype", "either", "elif", "else", "forward", "have", "hidden_states", "if", "input_ids", "input_shape", "inputs_embeds", "is", "kwargs", "long", "not", "ones", "or", "output_attentions", "output_hidden_states", "outputs", "pad_token_id", "pooler_output", "question_encoder", "r", "raise", "return", "return_dict", "same", "self", "size", "specify", "the", "time", "to", "token_type_ids", "torch", "tuple", "use_return_dict", "warn_if_padding_and_no_attention_mask", "zeros"], "dpr/modeling_dpr.py:DPRReader.__init__": ["Model", "ModelConfig", "ModelPretrainedReader", "ModelSpanPredictor", "__init__", "class", "config", "def", "post_init", "self", "span_predictor", "super"], "dpr/modeling_dpr.py:DPRReader.forward": ["Model", "ModelOutput", "ModelPretrainedReader", "None", "Tensor", "ValueError", "You", "and", "at", "attention_mask", "bool", "both", "cannot", "class", "config", "def", "device", "either", "elif", "else", "forward", "have", "if", "input_ids", "input_shape", "inputs_embeds", "is", "kwargs", "not", "ones", "or", "output_attentions", "output_hidden_states", "r", "raise", "return", "return_dict", "same", "self", "size", "span_predictor", "specify", "the", "time", "to", "torch", "tuple", "use_return_dict", "warn_if_padding_and_no_attention_mask"], "dpt/modeling_dpt.py:DPTViTHybridEmbeddings.__init__": ["Conv2d", "Expected", "Iterable", "Model", "ModelConfig", "Module", "None", "Parameter", "ValueError", "__init__", "abc", "backbone", "backbone_featmap_shape", "channels", "class", "cls_token", "collections", "config", "def", "else", "f", "feat_map_shape", "feature_dim", "feature_size", "features", "got", "have", "hidden_size", "if", "image_size", "int", "is", "isinstance", "kernel_size", "len", "load_backbone", "nn", "num_channels", "num_patches", "output", "patch_size", "position_embeddings", "projection", "raise", "residual_feature_map_index", "self", "super", "to", "torch", "tuple", "zeros"], "dpt/modeling_dpt.py:DPTViTHybridEmbeddings._resize_pos_embed": ["Model", "Module", "_resize_pos_embed", "bilinear", "cat", "class", "def", "dim", "functional", "grid_size_height", "grid_size_width", "interpolate", "len", "mode", "nn", "old_grid_size", "permute", "posemb", "posemb_grid", "posemb_tok", "reshape", "return", "self", "size", "start_index", "torch", "torch_int"], "dpt/modeling_dpt.py:DPTViTHybridEmbeddings.forward": ["BaseModelOutputWithIntermediateActivations", "False", "Input", "Make", "Model", "Module", "Tensor", "ValueError", "_resize_pos_embed", "backbone", "backbone_output", "batch_size", "bool", "cat", "channel", "class", "cls_token", "cls_tokens", "configuration", "def", "dim", "dimension", "doesn", "embeddings", "expand", "f", "feature_maps", "features", "flatten", "for", "forward", "height", "if", "image", "image_size", "in", "index", "intermediate_activations", "interpolate_pos_encoding", "last_hidden_states", "match", "model", "nn", "not", "num_channels", "of", "one", "or", "output_hidden_states", "patch_size", "pixel", "pixel_values", "position_embeddings", "projection", "raise", "residual_feature_map_index", "return", "self", "set", "shape", "size", "sure", "t", "that", "the", "torch", "transpose", "values", "width", "with"], "dpt/modeling_dpt.py:DPTViTEmbeddings.__init__": ["Dropout", "Model", "ModelViTPatchEmbeddings", "Module", "Parameter", "__init__", "class", "cls_token", "config", "def", "dropout", "hidden_dropout_prob", "hidden_size", "nn", "num_patches", "patch_embeddings", "position_embeddings", "self", "super", "torch", "zeros"], "dpt/modeling_dpt.py:DPTViTEmbeddings._resize_pos_embed": ["Model", "Module", "_resize_pos_embed", "bilinear", "cat", "class", "def", "dim", "functional", "grid_size_height", "grid_size_width", "interpolate", "mode", "nn", "old_grid_size", "permute", "posemb", "posemb_grid", "posemb_tok", "reshape", "return", "self", "size", "start_index", "torch", "torch_int"], "dpt/modeling_dpt.py:DPTViTEmbeddings.forward": ["BaseModelOutputWithIntermediateActivations", "Model", "Module", "Tensor", "_", "_resize_pos_embed", "batch_size", "cat", "class", "cls_token", "cls_tokens", "config", "def", "dim", "dropout", "embeddings", "expand", "forward", "height", "last_hidden_states", "nn", "num_channels", "patch_embeddings", "patch_size", "pixel_values", "position_embeddings", "return", "self", "seq_len", "shape", "size", "torch", "width"], "dpt/modeling_dpt.py:DPTViTPatchEmbeddings.__init__": ["Conv2d", "Iterable", "Model", "ModelConfig", "Module", "__init__", "abc", "class", "collections", "config", "def", "else", "hidden_size", "if", "image_size", "isinstance", "kernel_size", "nn", "num_channels", "num_patches", "patch_size", "projection", "self", "stride", "super"], "dpt/modeling_dpt.py:DPTViTPatchEmbeddings.forward": ["Make", "Model", "Module", "Tensor", "ValueError", "batch_size", "channel", "class", "configuration", "def", "dimension", "embeddings", "flatten", "forward", "height", "if", "in", "match", "nn", "num_channels", "of", "one", "pixel", "pixel_values", "projection", "raise", "return", "self", "set", "shape", "sure", "that", "the", "torch", "transpose", "values", "width", "with"], "dpt/modeling_dpt.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "contiguous", "def", "dim", "dropout", "float", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "shape", "size", "softmax", "torch", "training", "transpose", "value"], "dpt/modeling_dpt.py:DPTSelfAttention.__init__": ["False", "Linear", "Model", "ModelConfig", "Module", "The", "ValueError", "__init__", "a", "all_head_size", "and", "attention", "attention_head_size", "attention_probs_dropout_prob", "bias", "class", "config", "def", "dropout_prob", "embedding_size", "f", "hasattr", "heads", "hidden", "hidden_size", "if", "int", "is", "is_causal", "key", "multiple", "nn", "not", "num_attention_heads", "number", "of", "qkv_bias", "query", "raise", "scaling", "self", "size", "super", "the", "value"], "dpt/modeling_dpt.py:DPTSelfAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "Model", "Module", "None", "Tensor", "_attn_implementation", "all_head_size", "attention_head_size", "attention_interface", "attention_probs", "batch_size", "class", "config", "context_layer", "def", "dropout", "dropout_prob", "eager", "eager_attention_forward", "else", "forward", "hidden_states", "if", "is_causal", "key", "key_layer", "new_context_layer_shape", "new_shape", "nn", "not", "num_attention_heads", "query", "query_layer", "reshape", "return", "scaling", "self", "shape", "size", "torch", "training", "transpose", "tuple", "value", "value_layer", "view"], "dpt/modeling_dpt.py:DPTViTSelfOutput.__init__": ["Dropout", "Linear", "Model", "ModelConfig", "Module", "__init__", "class", "config", "def", "dense", "dropout", "hidden_dropout_prob", "hidden_size", "nn", "self", "super"], "dpt/modeling_dpt.py:DPTViTSelfOutput.forward": ["Model", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "dpt/modeling_dpt.py:DPTViTAttention.__init__": ["Model", "ModelConfig", "ModelSelfAttention", "ModelViTSelfOutput", "Module", "__init__", "attention", "class", "config", "def", "nn", "output", "self", "super"], "dpt/modeling_dpt.py:DPTViTAttention.forward": ["Model", "Module", "Tensor", "_", "attention", "class", "def", "forward", "hidden_states", "nn", "output", "return", "self", "self_attn_output", "torch"], "dpt/modeling_dpt.py:DPTViTIntermediate.__init__": ["ACT2FN", "Linear", "Model", "ModelConfig", "Module", "__init__", "class", "config", "def", "dense", "else", "hidden_act", "hidden_size", "if", "intermediate_act_fn", "intermediate_size", "isinstance", "nn", "self", "str", "super"], "dpt/modeling_dpt.py:DPTViTIntermediate.forward": ["Model", "Module", "Tensor", "class", "def", "dense", "forward", "hidden_states", "intermediate_act_fn", "nn", "return", "self", "torch"], "dpt/modeling_dpt.py:DPTViTOutput.__init__": ["Dropout", "Linear", "Model", "ModelConfig", "Module", "__init__", "class", "config", "def", "dense", "dropout", "hidden_dropout_prob", "hidden_size", "intermediate_size", "nn", "self", "super"], "dpt/modeling_dpt.py:DPTViTOutput.forward": ["Model", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "dpt/modeling_dpt.py:DPTViTLayer.__init__": ["GradientCheckpointingLayer", "LayerNorm", "Model", "ModelConfig", "ModelViTAttention", "ModelViTIntermediate", "ModelViTOutput", "__init__", "attention", "chunk_size_feed_forward", "class", "config", "def", "eps", "hidden_size", "intermediate", "layer_norm_eps", "layernorm_after", "layernorm_before", "nn", "output", "self", "seq_len_dim", "super"], "dpt/modeling_dpt.py:DPTViTLayer.forward": ["GradientCheckpointingLayer", "Model", "Tensor", "attention", "attention_output", "class", "def", "forward", "hidden_states", "hidden_states_norm", "intermediate", "layer_output", "layernorm_after", "layernorm_before", "output", "return", "self", "torch"], "dpt/modeling_dpt.py:DPTViTEncoder.__init__": ["False", "Model", "ModelConfig", "ModelViTLayer", "Module", "ModuleList", "_", "__init__", "class", "config", "def", "for", "gradient_checkpointing", "in", "layer", "nn", "num_hidden_layers", "range", "self", "super"], "dpt/modeling_dpt.py:DPTViTEncoder.forward": ["BaseModelOutput", "False", "Model", "Module", "None", "Tensor", "all_hidden_states", "append", "bool", "class", "def", "else", "enumerate", "for", "forward", "hidden_states", "i", "if", "in", "last_hidden_state", "layer", "layer_module", "nn", "output_hidden_states", "return", "self", "torch", "tuple"], "dpt/modeling_dpt.py:DPTReassembleStage.__init__": ["Model", "Module", "ModuleList", "__init__", "_init_reassemble_Model", "_init_reassemble_Model_hybrid", "class", "config", "def", "else", "if", "is_hybrid", "layers", "neck_ignore_stages", "nn", "self", "super"], "dpt/modeling_dpt.py:DPTReassembleStage._init_reassemble_dpt_hybrid": ["ACT2FN", "Hybrid", "Identity", "Linear", "Model", "ModelReassembleLayer", "Module", "ModuleList", "Readout", "Sequential", "ValueError", "_get_backbone_hidden_size", "_init_reassemble_Model_hybrid", "append", "channels", "class", "config", "def", "elif", "f", "factor", "for", "hidden_act", "hidden_size", "i", "if", "in", "is", "layers", "len", "neck_hidden_sizes", "nn", "not", "project", "r", "raise", "range", "readout_projects", "readout_type", "reassemble_factors", "self", "supported", "type", "zip"], "dpt/modeling_dpt.py:DPTReassembleStage._init_reassemble_dpt": ["ACT2FN", "Linear", "Model", "ModelReassembleLayer", "Module", "ModuleList", "Sequential", "_", "_get_backbone_hidden_size", "_init_reassemble_Model", "append", "channels", "class", "config", "def", "factor", "for", "hidden_act", "hidden_size", "i", "if", "in", "layers", "len", "neck_hidden_sizes", "nn", "project", "range", "readout_projects", "readout_type", "reassemble_factors", "self", "zip"], "dpt/modeling_dpt.py:DPTReassembleStage.forward": ["Model", "Module", "None", "Tensor", "add", "and", "append", "batch_size", "cat", "class", "cls_token", "config", "contiguous", "def", "elif", "else", "enumerate", "expand_as", "feature_shape", "flatten", "for", "forward", "hidden_state", "hidden_states", "i", "if", "in", "is", "layers", "list", "neck_ignore_stages", "nn", "not", "num_channels", "out", "patch_height", "patch_width", "permute", "project", "readout", "readout_projects", "readout_type", "reshape", "return", "self", "sequence_length", "shape", "size", "torch", "torch_int", "unsqueeze"], "dpt/modeling_dpt.py:_get_backbone_hidden_size": ["False", "None", "_get_backbone_hidden_size", "and", "backbone_config", "config", "def", "else", "hidden_size", "if", "is", "is_hybrid", "not", "return"], "dpt/modeling_dpt.py:DPTReassembleLayer.__init__": ["Conv2d", "ConvTranspose2d", "Identity", "Model", "ModelConfig", "Module", "__init__", "_get_backbone_hidden_size", "channels", "class", "config", "def", "elif", "factor", "hidden_size", "if", "in_channels", "int", "kernel_size", "nn", "out_channels", "padding", "projection", "resize", "self", "stride", "super"], "dpt/modeling_dpt.py:DPTReassembleLayer.forward": ["Model", "Module", "class", "def", "forward", "hidden_state", "nn", "projection", "resize", "return", "self"], "dpt/modeling_dpt.py:DPTFeatureFusionStage.__init__": ["Model", "ModelConfig", "ModelFeatureFusionLayer", "Module", "ModuleList", "_", "__init__", "append", "class", "config", "def", "for", "in", "layers", "len", "neck_hidden_sizes", "nn", "range", "self", "super"], "dpt/modeling_dpt.py:DPTFeatureFusionStage.forward": ["Model", "Module", "None", "append", "class", "def", "else", "for", "forward", "fused_hidden_state", "fused_hidden_states", "hidden_state", "hidden_states", "if", "in", "is", "layer", "layers", "nn", "return", "self", "zip"], "dpt/modeling_dpt.py:DPTPreActResidualLayer.__init__": ["BatchNorm2d", "Conv2d", "Model", "ModelConfig", "Module", "None", "ReLU", "__init__", "activation1", "activation2", "batch_norm1", "batch_norm2", "bias", "class", "config", "convolution1", "convolution2", "def", "else", "fusion_hidden_size", "if", "is", "kernel_size", "nn", "not", "padding", "self", "stride", "super", "use_batch_norm", "use_batch_norm_in_fusion_residual", "use_bias_in_fusion_residual"], "dpt/modeling_dpt.py:DPTPreActResidualLayer.forward": ["Model", "Module", "Tensor", "activation1", "activation2", "batch_norm1", "batch_norm2", "class", "convolution1", "convolution2", "def", "forward", "hidden_state", "if", "nn", "residual", "return", "self", "torch", "use_batch_norm"], "dpt/modeling_dpt.py:DPTFeatureFusionLayer.__init__": ["Conv2d", "Model", "ModelConfig", "ModelPreActResidualLayer", "Module", "True", "__init__", "align_corners", "bias", "bool", "class", "config", "def", "fusion_hidden_size", "kernel_size", "nn", "projection", "residual_layer1", "residual_layer2", "self", "super"], "dpt/modeling_dpt.py:DPTFeatureFusionLayer.forward": ["False", "Model", "Module", "None", "Tensor", "align_corners", "bilinear", "class", "def", "forward", "functional", "hidden_state", "if", "interpolate", "is", "mode", "nn", "not", "projection", "residual", "residual_layer1", "residual_layer2", "return", "scale_factor", "self", "shape", "size", "torch"], "dpt/modeling_dpt.py:DPTPreTrainedModel._init_weights": ["Model", "ModelViTEmbeddings", "ModelViTHybridEmbeddings", "PreTrainedModel", "_init_weights", "class", "cls_token", "def", "if", "init", "isinstance", "module", "position_embeddings", "self", "super", "zeros_"], "dpt/modeling_dpt.py:DPTModel.__init__": ["LayerNorm", "Model", "ModelConfig", "ModelPreTrainedModel", "ModelViTEmbeddings", "ModelViTEncoder", "ModelViTHybridEmbeddings", "ModelViTPooler", "None", "True", "__init__", "add_pooling_layer", "bool", "class", "config", "def", "else", "embeddings", "encoder", "eps", "hidden_size", "if", "is_hybrid", "layer_norm_eps", "layernorm", "nn", "pooler", "post_init", "r", "self", "super"], "dpt/modeling_dpt.py:DPTModel.get_input_embeddings": ["Model", "ModelPreTrainedModel", "class", "config", "def", "else", "embeddings", "get_input_embeddings", "if", "is_hybrid", "patch_embeddings", "return", "self"], "dpt/modeling_dpt.py:DPTModel.forward": ["BaseModelOutput", "BaseModelOutputWithIntermediateActivations", "BaseModelOutputWithPoolingAndIntermediateActivations", "FloatTensor", "Model", "ModelPreTrainedModel", "None", "bool", "class", "config", "def", "else", "embedding_last_hidden_states", "embedding_output", "embeddings", "encoder", "encoder_outputs", "forward", "hidden_states", "if", "intermediate_activations", "is", "kwargs", "last_hidden_state", "last_hidden_states", "layernorm", "not", "output_hidden_states", "pixel_values", "pooled_output", "pooler", "pooler_output", "return", "self", "sequence_output", "torch"], "dpt/modeling_dpt.py:DPTViTPooler.__init__": ["ACT2FN", "Linear", "Model", "ModelConfig", "Module", "__init__", "activation", "class", "config", "def", "dense", "hidden_size", "nn", "pooler_act", "pooler_output_size", "self", "super"], "dpt/modeling_dpt.py:DPTViTPooler.forward": ["Model", "Module", "Tensor", "activation", "class", "def", "dense", "first_token_tensor", "forward", "hidden_states", "nn", "pooled_output", "return", "self", "torch"], "dpt/modeling_dpt.py:DPTNeck.__init__": ["Conv2d", "False", "Model", "ModelConfig", "ModelFeatureFusionStage", "ModelReassembleStage", "Module", "ModuleList", "None", "__init__", "and", "append", "backbone_config", "bias", "channel", "class", "config", "convs", "def", "else", "for", "fusion_hidden_size", "fusion_stage", "if", "in", "is", "kernel_size", "model_type", "neck_hidden_sizes", "nn", "not", "padding", "reassemble_stage", "self", "super", "swinv2"], "dpt/modeling_dpt.py:DPTNeck.forward": ["Model", "Module", "None", "Tensor", "The", "TypeError", "ValueError", "a", "be", "class", "config", "convs", "def", "enumerate", "equal", "feature", "features", "for", "forward", "fusion_stage", "hidden", "hidden_states", "i", "if", "in", "int", "is", "isinstance", "len", "list", "neck", "neck_hidden_sizes", "nn", "not", "number", "of", "or", "output", "patch_height", "patch_width", "raise", "reassemble_stage", "return", "self", "should", "sizes", "states", "tensors", "the", "to", "torch", "tuple"], "dpt/modeling_dpt.py:DPTDepthEstimationHead.__init__": ["Conv2d", "Model", "ModelConfig", "Module", "None", "ReLU", "Sequential", "True", "Upsample", "__init__", "add_projection", "align_corners", "bilinear", "class", "config", "def", "features", "fusion_hidden_size", "head", "if", "kernel_size", "mode", "nn", "padding", "projection", "scale_factor", "self", "stride", "super"], "dpt/modeling_dpt.py:DPTDepthEstimationHead.forward": ["Model", "Module", "None", "ReLU", "Tensor", "class", "config", "def", "dim", "forward", "head", "head_in_index", "hidden_states", "if", "is", "list", "nn", "not", "predicted_depth", "projection", "return", "self", "squeeze", "torch"], "dpt/modeling_dpt.py:DPTForDepthEstimation.__init__": ["False", "Model", "ModelDepthEstimationHead", "ModelModel", "ModelNeck", "ModelPreTrainedModel", "None", "__init__", "add_pooling_layer", "and", "backbone", "backbone_config", "class", "config", "def", "else", "head", "if", "is", "is_hybrid", "load_backbone", "neck", "not", "or", "post_init", "self", "super"], "dpt/modeling_dpt.py:DPTForDepthEstimation.forward": ["DepthEstimatorOutput", "False", "FloatTensor", "LongTensor", "Model", "ModelPreTrainedModel", "None", "NotImplementedError", "Training", "True", "_", "and", "attentions", "backbone", "backbone_config", "backbone_hidden_states", "backbone_out_indices", "bool", "class", "config", "def", "else", "enumerate", "extend", "feature", "feature_maps", "for", "forward", "forward_with_filtered_kwargs", "head", "height", "hidden_states", "idx", "if", "implemented", "in", "intermediate_activations", "is", "is_hybrid", "kwargs", "labels", "loss", "neck", "not", "output_hidden_states", "outputs", "patch_height", "patch_size", "patch_width", "pixel_values", "predicted_depth", "r", "raise", "return", "self", "shape", "torch", "width", "yet"], "dpt/modeling_dpt.py:DPTSemanticSegmentationHead.__init__": ["BatchNorm2d", "Conv2d", "Dropout", "False", "Model", "ModelConfig", "Module", "ReLU", "Sequential", "True", "Upsample", "__init__", "align_corners", "bias", "bilinear", "class", "config", "def", "features", "fusion_hidden_size", "head", "kernel_size", "mode", "nn", "num_labels", "padding", "scale_factor", "self", "semantic_classifier_dropout", "super"], "dpt/modeling_dpt.py:DPTSemanticSegmentationHead.forward": ["Model", "Module", "Tensor", "class", "config", "def", "forward", "head", "head_in_index", "hidden_states", "list", "logits", "nn", "return", "self", "torch"], "dpt/modeling_dpt.py:DPTAuxiliaryHead.__init__": ["BatchNorm2d", "Conv2d", "Dropout", "False", "Model", "ModelConfig", "Module", "ReLU", "Sequential", "__init__", "bias", "class", "config", "def", "features", "fusion_hidden_size", "head", "kernel_size", "nn", "num_labels", "padding", "self", "super"], "dpt/modeling_dpt.py:DPTAuxiliaryHead.forward": ["Model", "Module", "Tensor", "class", "def", "forward", "head", "hidden_states", "logits", "nn", "return", "self", "torch"], "dpt/modeling_dpt.py:DPTForSemanticSegmentation.__init__": ["False", "Model", "ModelAuxiliaryHead", "ModelConfig", "ModelModel", "ModelNeck", "ModelPreTrainedModel", "ModelSemanticSegmentationHead", "None", "__init__", "add_pooling_layer", "auxiliary_head", "class", "config", "def", "else", "head", "if", "neck", "post_init", "self", "super", "use_auxiliary_head"], "dpt/modeling_dpt.py:DPTForSemanticSegmentation.forward": ["BaseModelOutputWithPoolingAndIntermediateActivations", "CrossEntropyLoss", "False", "FloatTensor", "LongTensor", "Model", "ModelPreTrainedModel", "None", "SemanticSegmenterOutput", "The", "True", "ValueError", "align_corners", "and", "attentions", "auxiliary_head", "auxiliary_logits", "auxiliary_loss", "auxiliary_loss_weight", "backbone_hidden_states", "backbone_out_indices", "be", "bilinear", "bool", "class", "config", "def", "else", "enumerate", "extend", "feature", "for", "forward", "functional", "greater", "head", "hidden_states", "idx", "if", "ignore_index", "in", "intermediate_activations", "interpolate", "is", "is_hybrid", "kwargs", "labels", "logits", "loss", "loss_fct", "main_loss", "mode", "neck", "nn", "not", "num_labels", "number", "of", "one", "output_hidden_states", "outputs", "pixel_values", "r", "raise", "return", "self", "semantic_loss_ignore_index", "shape", "should", "size", "than", "torch", "upsampled_auxiliary_logits", "upsampled_logits"], "edgetam/modeling_edgetam.py:EdgeTamLayerNorm.__init__": ["LayerNorm", "ModelLayerNorm", "NotImplementedError", "Unsupported", "__init__", "channels_first", "channels_last", "class", "data", "data_format", "def", "eps", "f", "format", "if", "in", "kwargs", "nn", "normalized_shape", "not", "raise", "self", "super"], "edgetam/modeling_edgetam.py:EdgeTamLayerNorm.forward": ["LayerNorm", "ModelLayerNorm", "Tensor", "channels_first", "class", "data_format", "def", "else", "features", "forward", "if", "nn", "permute", "return", "self", "super", "torch"], "edgetam/modeling_edgetam.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "attention_mask", "attn_output", "attn_weights", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "softmax", "to", "torch", "training", "transpose", "value"], "edgetam/modeling_edgetam.py:EdgeTamAttention.__init__": ["False", "Linear", "ModelAttention", "Module", "None", "__init__", "attention_downsample_rate", "class", "config", "def", "downsample_rate", "else", "head_dim", "hidden_size", "if", "internal_dim", "is", "is_causal", "k_proj", "nn", "num_attention_heads", "o_proj", "q_proj", "scaling", "self", "super", "v_proj"], "edgetam/modeling_edgetam.py:EdgeTamAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "ModelAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_attn_implementation", "attention_interface", "attention_mask", "attention_similarity", "attn_output", "attn_weights", "batch_size", "class", "config", "contiguous", "def", "dropout", "eager", "eager_attention_forward", "forward", "head_dim", "if", "is_causal", "k_proj", "key", "kwargs", "new_shape", "nn", "num_attention_heads", "o_proj", "point_batch_size", "q_proj", "query", "reshape", "return", "scaling", "self", "shape", "torch", "transpose", "tuple", "v_proj", "value", "view"], "edgetam/modeling_edgetam.py:EdgeTamTwoWayAttentionBlock.__init__": ["False", "LayerNorm", "ModelAttention", "ModelFeedForward", "ModelMaskDecoderConfig", "ModelTwoWayAttentionBlock", "Module", "__init__", "bool", "class", "config", "cross_attn_image_to_token", "cross_attn_token_to_image", "def", "downsample_rate", "hidden_size", "layer_norm1", "layer_norm2", "layer_norm3", "layer_norm4", "mlp", "mlp_dim", "nn", "num_hidden_layers", "num_layers", "self", "self_attn", "skip_first_layer_pe", "super"], "edgetam/modeling_edgetam.py:EdgeTamTwoWayAttentionBlock.forward": ["ModelTwoWayAttentionBlock", "Module", "Tensor", "TransformersKwargs", "Unpack", "_", "attention_similarity", "attn_out", "class", "cross_attn_image_to_token", "cross_attn_token_to_image", "def", "else", "forward", "if", "key", "key_point_embedding", "keys", "kwargs", "layer_norm1", "layer_norm2", "layer_norm3", "layer_norm4", "mlp", "mlp_out", "nn", "queries", "query", "query_point_embedding", "return", "self", "self_attn", "skip_first_layer_pe", "value"], "edgetam/modeling_edgetam.py:EdgeTamFeedForward.__init__": ["ACT2FN", "False", "Linear", "ModelFeedForward", "Module", "ModuleList", "_", "__init__", "activation", "bool", "class", "def", "for", "hidden_dim", "in", "input_dim", "int", "layers", "nn", "num_layers", "output_dim", "proj_in", "proj_out", "range", "relu", "self", "sigmoid_output", "str", "super"], "edgetam/modeling_edgetam.py:EdgeTamFeedForward.forward": ["F", "ModelFeedForward", "Module", "activation", "class", "def", "for", "forward", "hidden_states", "if", "in", "layer", "layers", "nn", "proj_in", "proj_out", "return", "self", "sigmoid", "sigmoid_output"], "edgetam/modeling_edgetam.py:EdgeTamPreTrainedModel._init_weights": ["ModelModel", "ModelPreTrainedModel", "None", "PreTrainedModel", "_init_weights", "class", "def", "elif", "hasattr", "if", "init", "is", "isinstance", "module", "no_memory_embedding", "normal_", "not", "positional_embedding", "scale", "self", "std", "super", "zeros_"], "edgetam/modeling_edgetam.py:EdgeTamSinePositionEmbedding.__init__": ["False", "ModelSinePositionEmbedding", "Module", "None", "True", "ValueError", "__init__", "and", "be", "bool", "class", "def", "else", "float", "if", "int", "is", "math", "nn", "normalize", "not", "num_pos_feats", "passed", "pi", "raise", "scale", "self", "should", "super", "temperature"], "edgetam/modeling_edgetam.py:EdgeTamSinePositionEmbedding.forward": ["ModelSinePositionEmbedding", "Module", "None", "Size", "Tensor", "arange", "bool", "cat", "class", "cos", "cumsum", "def", "device", "dim", "dim_t", "div", "dtype", "eps", "flatten", "floor", "forward", "if", "int64", "is", "mask", "nn", "normalize", "not_mask", "num_pos_feats", "permute", "pos", "pos_x", "pos_y", "return", "rounding_mode", "scale", "self", "shape", "sin", "stack", "str", "temperature", "to", "torch", "x_embed", "y_embed", "zeros"], "edgetam/modeling_edgetam.py:EdgeTamVisionNeck.__init__": ["Conv2d", "ModelSinePositionEmbedding", "ModelVisionConfig", "ModelVisionNeck", "Module", "ModuleList", "True", "__init__", "append", "backbone_channel_list", "class", "config", "convs", "def", "for", "fpn_hidden_size", "fpn_kernel_size", "fpn_padding", "fpn_stride", "fpn_top_down_levels", "in", "in_channels", "kernel_size", "nn", "normalize", "num_pos_feats", "out_channels", "padding", "position_encoding", "self", "stride", "super"], "edgetam/modeling_edgetam.py:EdgeTamVisionNeck.forward": ["F", "False", "ModelVisionNeck", "Module", "None", "Tensor", "align_corners", "antialias", "class", "convs", "def", "device", "dtype", "else", "float32", "for", "forward", "fpn_hidden_states", "fpn_position_encoding", "fpn_top_down_levels", "hidden_states", "i", "if", "in", "interpolate", "lateral_features", "len", "mode", "n", "nearest", "nn", "not", "or", "permute", "position_encoding", "prev_features", "prev_position_encoding", "range", "return", "scale_factor", "self", "shape", "to", "top_down_features", "torch", "tuple", "weight"], "edgetam/modeling_edgetam.py:EdgeTamVisionModel.__init__": ["AutoModel", "ModelPreTrainedModel", "ModelVisionConfig", "ModelVisionModel", "ModelVisionNeck", "__init__", "backbone", "backbone_config", "class", "config", "def", "from_config", "neck", "num_feature_levels", "post_init", "self", "super"], "edgetam/modeling_edgetam.py:EdgeTamVisionModel.forward": ["FloatTensor", "ModelPreTrainedModel", "ModelVisionEncoderOutput", "ModelVisionModel", "None", "TransformersKwargs", "Unpack", "ValueError", "You", "backbone", "backbone_output", "class", "def", "for", "forward", "fpn_hidden_states", "fpn_position_encoding", "have", "hidden_state", "if", "in", "intermediate_hidden_states", "is", "kwargs", "last_hidden_state", "neck", "num_feature_levels", "permute", "pixel_values", "raise", "return", "self", "specify", "to", "torch", "tuple"], "edgetam/modeling_edgetam.py:EdgeTamPositionalEmbedding.__init__": ["ModelPositionalEmbedding", "ModelPromptEncoderConfig", "Module", "__init__", "class", "config", "def", "hidden_size", "nn", "positional_embedding", "randn", "register_buffer", "scale", "self", "super", "torch"], "edgetam/modeling_edgetam.py:EdgeTamPositionalEmbedding.forward": ["ModelPositionalEmbedding", "Module", "None", "cat", "class", "clone", "coordinates", "cos", "def", "dim", "dtype", "float32", "forward", "if", "input_coords", "input_shape", "is", "nn", "not", "np", "pi", "positional_embedding", "return", "self", "sin", "to", "torch"], "edgetam/modeling_edgetam.py:EdgeTamMaskEmbedding.__init__": ["ACT2FN", "Conv2d", "ModelLayerNorm", "ModelMaskEmbedding", "ModelPromptEncoderConfig", "Module", "__init__", "activation", "channels_first", "class", "config", "conv1", "conv2", "conv3", "data_format", "def", "eps", "hidden_act", "hidden_size", "kernel_size", "layer_norm1", "layer_norm2", "layer_norm_eps", "mask_input_channels", "nn", "self", "stride", "super"], "edgetam/modeling_edgetam.py:EdgeTamMaskEmbedding.forward": ["ModelMaskEmbedding", "Module", "activation", "class", "conv1", "conv2", "conv3", "def", "dense_embeddings", "forward", "hidden_states", "layer_norm1", "layer_norm2", "masks", "nn", "return", "self"], "edgetam/modeling_edgetam.py:EdgeTamPromptEncoder.__init__": ["Embedding", "ModelMaskEmbedding", "ModelPositionalEmbedding", "ModelPromptEncoder", "ModelPromptEncoderConfig", "Module", "__init__", "class", "config", "def", "hidden_size", "image_embedding_size", "image_size", "input_image_size", "mask_embed", "mask_input_size", "nn", "no_mask_embed", "not_a_point_embed", "num_point_embeddings", "patch_size", "point_embed", "self", "shared_embedding", "super"], "edgetam/modeling_edgetam.py:EdgeTamPromptEncoder._embed_points": ["ModelPromptEncoder", "Module", "None", "Tensor", "_embed_points", "bool", "clamp", "class", "constant", "def", "functional", "if", "input_image_size", "input_shape", "labels", "min", "mode", "nn", "not_a_point_embed", "pad", "point_embed", "point_embedding", "points", "return", "self", "shared_embedding", "torch", "unsqueeze", "value", "weight", "where", "zeros_like"], "edgetam/modeling_edgetam.py:EdgeTamPromptEncoder._embed_boxes": ["ModelPromptEncoder", "Module", "Tensor", "_embed_boxes", "boxes", "class", "constant", "coords", "corner_embedding", "def", "expand_as", "functional", "input_image_size", "mode", "nn", "not_a_point_embed", "pad", "point_embed", "return", "self", "shape", "shared_embedding", "torch", "value", "view", "weight"], "edgetam/modeling_edgetam.py:EdgeTamPromptEncoder.forward": ["If", "ModelPromptEncoder", "Module", "None", "Tensor", "ValueError", "_embed_boxes", "_embed_points", "also", "are", "batch_size", "be", "box_embeddings", "cat", "class", "def", "dense_embeddings", "dim", "else", "expand", "forward", "if", "image_embedding_size", "input_boxes", "input_labels", "input_masks", "input_points", "is", "labels", "mask_embed", "must", "nn", "no_mask_embed", "not", "pad", "point_embeddings", "points", "provided", "raise", "reshape", "return", "self", "shape", "sparse_embeddings", "torch", "tuple", "weight"], "edgetam/modeling_edgetam.py:EdgeTamTwoWayTransformer.__init__": ["LayerNorm", "ModelAttention", "ModelMaskDecoderConfig", "ModelTwoWayAttentionBlock", "ModelTwoWayTransformer", "Module", "ModuleList", "__init__", "append", "class", "config", "def", "final_attn_token_to_image", "for", "hidden_size", "i", "in", "layer_norm_final_attn", "layers", "nn", "num_hidden_layers", "range", "self", "skip_first_layer_pe", "super"], "edgetam/modeling_edgetam.py:EdgeTamTwoWayTransformer.forward": ["BaseModelOutput", "ModelTwoWayTransformer", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "_", "an", "attention_similarity", "attn_out", "class", "def", "final_attn_token_to_image", "flatten", "for", "forward", "have", "if", "image_embedding", "image_embeddings", "image_positional_embeddings", "in", "is", "key", "key_point_embedding", "keys", "kwargs", "layer", "layer_norm_final_attn", "layers", "nn", "not", "permute", "point_embeddings", "queries", "query", "query_point_embedding", "raise", "return", "self", "specify", "target_embedding", "to", "tuple", "unsqueeze", "value"], "edgetam/modeling_edgetam.py:EdgeTamMaskDecoder.__init__": ["Conv2d", "ConvTranspose2d", "Embedding", "GELU", "ModelFeedForward", "ModelLayerNorm", "ModelMaskDecoder", "ModelMaskDecoderConfig", "ModelTwoWayTransformer", "Module", "ModuleList", "True", "_", "__init__", "activation", "channels_first", "class", "config", "conv_s0", "conv_s1", "data_format", "def", "dynamic_multimask_stability_delta", "dynamic_multimask_stability_thresh", "dynamic_multimask_via_stability", "for", "hidden_size", "in", "iou_head_depth", "iou_head_hidden_dim", "iou_prediction_head", "iou_token", "kernel_size", "mask_tokens", "mlps_list", "nn", "num_mask_tokens", "num_multimask_outputs", "obj_score_token", "output_hypernetworks_mlps", "pred_obj_score_head", "range", "self", "sigmoid_output", "stride", "super", "transformer", "upscale_conv1", "upscale_conv2", "upscale_layer_norm"], "edgetam/modeling_edgetam.py:EdgeTamMaskDecoder.forward": ["ModelMaskDecoder", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "_dynamic_multimask_via_stability", "activation", "and", "attention_similarity", "batch_size", "bool", "cat", "class", "current_mlp", "def", "dense_prompt_embeddings", "dim", "dtype", "dynamic_multimask_via_stability", "elif", "else", "feat_s0", "feat_s1", "for", "forward", "height", "high_resolution_features", "hyper_in", "hyper_in_list", "i", "if", "image_embeddings", "image_positional_embeddings", "in", "iou_pred", "iou_prediction_head", "iou_token", "iou_token_out", "kwargs", "list", "mask_slice", "mask_tokens", "mask_tokens_out", "masks", "multimask_output", "nn", "not", "num_channels", "num_mask_tokens", "obj_score_token", "object_score_logits", "output_hypernetworks_mlps", "output_tokens", "point_batch_size", "point_embeddings", "pred_obj_score_head", "range", "repeat", "repeat_interleave", "return", "sam_tokens_out", "self", "shape", "slice", "sparse_prompt_embeddings", "stack", "target_embedding", "to", "tokens", "torch", "training", "transformer", "transpose", "tuple", "upscale_conv1", "upscale_conv2", "upscale_layer_norm", "upscaled_embedding", "view", "weight", "width"], "edgetam/modeling_edgetam.py:EdgeTamMaskDecoder._get_stability_scores": ["ModelMaskDecoder", "Module", "_get_stability_scores", "area_i", "area_u", "class", "def", "dim", "dynamic_multimask_stability_delta", "flatten", "float", "mask_logits", "nn", "return", "self", "stability_delta", "stability_scores", "sum", "torch", "where"], "edgetam/modeling_edgetam.py:EdgeTamMaskDecoder._dynamic_multimask_via_stability": ["ModelMaskDecoder", "Module", "None", "_dynamic_multimask_via_stability", "_get_stability_scores", "all_iou_scores", "all_mask_logits", "argmax", "best_multimask_iou_scores", "best_multimask_logits", "best_scores_inds", "best_scores_inds_expanded", "class", "def", "dim", "dynamic_multimask_stability_thresh", "expand", "expand_as", "gather", "iou_scores_out", "is_stable", "mask_logits_out", "multimask_iou_scores", "multimask_logits", "nn", "return", "self", "singlemask_iou_scores", "singlemask_logits", "size", "stability_scores", "torch", "unsqueeze", "where"], "edgetam/modeling_edgetam.py:EdgeTamModel.__init__": ["AutoModel", "ModelConfig", "ModelMaskDecoder", "ModelModel", "ModelPositionalEmbedding", "ModelPreTrainedModel", "ModelPromptEncoder", "Parameter", "__init__", "_attn_implementation", "backbone_feature_sizes", "class", "config", "def", "fpn_hidden_size", "from_config", "hidden_dim", "mask_decoder", "mask_decoder_config", "nn", "no_memory_embedding", "num_feature_levels", "post_init", "prompt_encoder", "prompt_encoder_config", "self", "shared_image_embedding", "super", "torch", "vision_config", "vision_encoder", "zeros"], "edgetam/modeling_edgetam.py:EdgeTamModel.get_image_wide_positional_embeddings": ["ModelModel", "ModelPreTrainedModel", "Tensor", "class", "cumsum", "def", "device", "dim", "dtype", "get_image_wide_positional_embeddings", "grid", "image_embedding_size", "ones", "permute", "positional_embedding", "prompt_encoder", "return", "self", "shared_image_embedding", "size", "stack", "target_device", "target_dtype", "torch", "unsqueeze", "x_embed", "y_embed"], "edgetam/modeling_edgetam.py:EdgeTamModel.get_image_embeddings": ["FloatTensor", "ModelModel", "ModelPreTrainedModel", "Tensor", "TransformersKwargs", "Unpack", "_", "backbone_feature_sizes", "batch_size", "class", "def", "feat", "feat_size", "feature_maps", "for", "get_image_embeddings", "get_image_features", "image_embeddings", "in", "kwargs", "list", "no_memory_embedding", "permute", "pixel_values", "r", "return", "self", "shape", "torch", "view", "zip"], "edgetam/modeling_edgetam.py:EdgeTamModel.get_prompt_embeddings": ["FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "class", "def", "get_prompt_embeddings", "input_boxes", "input_labels", "input_masks", "input_points", "prompt_encoder", "prompt_output", "r", "return", "self", "torch"], "edgetam/modeling_edgetam.py:EdgeTamModel.forward": ["Exactly", "F", "False", "FloatTensor", "Got", "LongTensor", "ModelImageSegmentationOutput", "ModelModel", "ModelPreTrainedModel", "None", "TransformersKwargs", "True", "Unpack", "ValueError", "You", "_", "align_corners", "and", "antialias", "as", "attention_similarity", "backbone_feature_sizes", "batch_size", "be", "bilinear", "bool", "bounding", "box", "boxes", "class", "def", "dense_embeddings", "dense_prompt_embeddings", "device", "dtype", "else", "f", "feat", "feat_size", "feature_maps", "float", "for", "forward", "get_image_features", "get_image_wide_positional_embeddings", "high_resolution_features", "if", "image_embeddings", "image_positional_embeddings", "in", "input", "input_boxes", "input_labels", "input_masks", "input_points", "int", "int32", "interpolate", "iou_scores", "is", "kwargs", "low_res_multimasks", "many", "mask_decoder", "mask_input_size", "mode", "multimask_output", "must", "no_memory_embedding", "not", "object_score_logits", "of", "one", "ones", "ones_like", "or", "per", "permute", "pixel_values", "points", "pred_masks", "prompt_encoder", "provide", "provided", "r", "raise", "repeat", "return", "self", "shape", "should", "size", "sparse_embeddings", "sparse_prompt_embeddings", "target_embedding", "to", "torch", "view", "vision_attentions", "vision_hidden_states", "zeros", "zip"], "edgetam/modeling_edgetam.py:EdgeTamModel.get_image_features": ["FloatTensor", "ModelModel", "ModelPreTrainedModel", "ModelVisionEncoderOutput", "None", "Tensor", "TransformersKwargs", "Unpack", "attentions", "class", "conv_s0", "conv_s1", "def", "feature_map", "feature_map_position_embedding", "feature_maps", "feature_maps_position_embeddings", "flatten", "for", "fpn_hidden_states", "fpn_position_encoding", "get_image_features", "hidden_states", "in", "kwargs", "list", "mask_decoder", "permute", "pixel_values", "r", "return", "self", "torch", "tuple", "vision_encoder", "vision_outputs"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoLayerNorm.__init__": ["LayerNorm", "ModelLayerNorm", "NotImplementedError", "Unsupported", "__init__", "channels_first", "channels_last", "class", "data", "data_format", "def", "eps", "f", "format", "if", "in", "kwargs", "nn", "normalized_shape", "not", "raise", "self", "super"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoLayerNorm.forward": ["LayerNorm", "ModelLayerNorm", "Tensor", "channels_first", "class", "data_format", "def", "else", "features", "forward", "if", "nn", "permute", "return", "self", "super", "torch"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoMemoryFuserCXBlock.__init__": ["ACT2FN", "Conv2d", "GradientCheckpointingLayer", "Linear", "ModelConfig", "ModelLayerNorm", "ModelMemoryFuserCXBlock", "Parameter", "True", "__init__", "activation", "channels_first", "class", "config", "data_format", "def", "depthwise_conv", "eps", "groups", "kernel_size", "layer_norm", "memory_fuser_embed_dim", "memory_fuser_hidden_act", "memory_fuser_intermediate_dim", "memory_fuser_kernel_size", "memory_fuser_layer_scale_init_value", "memory_fuser_padding", "nn", "ones", "padding", "pointwise_conv1", "pointwise_conv2", "requires_grad", "scale", "self", "super", "torch"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoMemoryFuserCXBlock.forward": ["GradientCheckpointingLayer", "ModelMemoryFuserCXBlock", "activation", "class", "def", "depthwise_conv", "forward", "hidden_states", "input", "layer_norm", "permute", "pointwise_conv1", "pointwise_conv2", "return", "scale", "self"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoVisionRotaryEmbedding.__init__": ["Dimension", "False", "ModelConfig", "ModelVisionRotaryEmbedding", "Module", "None", "RoPE", "ValueError", "__init__", "axial", "be", "by", "class", "config", "cos", "create_inv_freq", "def", "dim", "divisible", "else", "end_x", "end_y", "for", "if", "int", "inv_freq", "is", "memory_attention_downsample_rate", "memory_attention_hidden_size", "memory_attention_num_attention_heads", "memory_attention_rope_feat_sizes", "memory_attention_rope_theta", "must", "nn", "persistent", "raise", "register_buffer", "rope_embeddings_cos", "rope_embeddings_sin", "self", "sin", "super"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoVisionRotaryEmbedding.forward": ["ModelVisionRotaryEmbedding", "Module", "Tensor", "class", "def", "forward", "nn", "return", "rope_embeddings_cos", "rope_embeddings_sin", "self", "torch", "tuple"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoVisionRotaryEmbedding.create_inv_freq": ["ModelVisionRotaryEmbedding", "Module", "arange", "cat", "class", "create_inv_freq", "def", "dim", "div", "dtype", "end_x", "end_y", "flattened_indices", "float", "floor", "freqs", "freqs_x", "freqs_y", "inv_freq", "long", "memory_attention_rope_theta", "nn", "outer", "repeat_interleave", "return", "rounding_mode", "self", "torch", "x_positions", "y_positions"], "edgetam_video/modeling_edgetam_video.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "attention_mask", "attn_output", "attn_weights", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "softmax", "to", "torch", "training", "transpose", "value"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoAttention.__init__": ["False", "Linear", "ModelAttention", "Module", "None", "__init__", "attention_downsample_rate", "class", "config", "def", "downsample_rate", "else", "head_dim", "hidden_size", "if", "internal_dim", "is", "is_causal", "k_proj", "nn", "num_attention_heads", "o_proj", "q_proj", "scaling", "self", "super", "v_proj"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "ModelAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_attn_implementation", "attention_interface", "attention_mask", "attention_similarity", "attn_output", "attn_weights", "batch_size", "class", "config", "contiguous", "def", "dropout", "eager", "eager_attention_forward", "forward", "head_dim", "if", "is_causal", "k_proj", "key", "kwargs", "new_shape", "nn", "num_attention_heads", "o_proj", "point_batch_size", "q_proj", "query", "reshape", "return", "scaling", "self", "shape", "torch", "transpose", "tuple", "v_proj", "value", "view"], "edgetam_video/modeling_edgetam_video.py:rotate_pairwise": ["Model_pairwise", "def", "dim", "flatten", "return", "shape", "stack", "start_dim", "torch", "unbind", "view", "x", "x1", "x2"], "edgetam_video/modeling_edgetam_video.py:apply_rotary_pos_emb_2d_self_attn": ["Model_rotary_pos_emb_2d_self_attn", "Tensor", "cos", "def", "float", "k", "k_embed", "q", "q_embed", "return", "rotate_pairwise", "sin", "torch", "tuple", "type_as"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoRoPESelfAttention.__init__": ["False", "Linear", "ModelConfig", "ModelRoPESelfAttention", "Module", "__init__", "class", "config", "def", "dropout_p", "head_dim", "hidden_size", "internal_dim", "is_causal", "k_proj", "memory_attention_downsample_rate", "memory_attention_hidden_size", "memory_attention_num_attention_heads", "memory_attention_rope_dropout", "nn", "num_attention_heads", "o_proj", "q_proj", "scaling", "self", "super", "v_proj"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoRoPESelfAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "FlashAttentionKwargs", "ModelRoPESelfAttention", "Module", "None", "Tensor", "Unpack", "_attn_implementation", "apply_rotary_pos_emb_2d_self_attn", "attention_interface", "attention_mask", "attn_output", "attn_weights", "batch_size", "class", "config", "contiguous", "cos", "def", "dropout", "dropout_p", "eager", "eager_attention_forward", "else", "forward", "head_dim", "if", "is_causal", "k_proj", "key", "kwargs", "new_shape", "nn", "not", "num_attention_heads", "o_proj", "point_batch_size", "position_embeddings", "q_proj", "query", "reshape", "return", "scaling", "self", "shape", "sin", "torch", "training", "transpose", "tuple", "v_proj", "value", "view"], "edgetam_video/modeling_edgetam_video.py:apply_rotary_pos_emb_2d_cross_attn": ["Model_rotary_pos_emb_2d_cross_attn", "Tensor", "batch_size", "cat", "channels_per_head", "cos", "cos_k", "def", "dim", "float", "if", "int", "k", "k_embed", "k_excluded", "k_final", "k_for_rope", "k_grouped", "k_rope_input", "k_seq_len", "k_spatial", "k_spatial_embed", "k_spatial_reshaped", "k_temporal", "k_temporal_reshaped", "num_heads", "num_k_exclude_rope", "num_total_k_tokens", "q", "q_embed", "repeat", "repeat_freqs_k", "reshape", "return", "rotate_pairwise", "shape", "sin", "sin_k", "spatial_tokens", "temporal_tokens", "tokens_per_group", "torch", "tuple", "type_as", "view"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoRoPECrossAttention.__init__": ["False", "Linear", "ModelConfig", "ModelRoPECrossAttention", "Module", "__init__", "class", "config", "def", "dropout_p", "head_dim", "hidden_size", "int", "internal_dim", "is_causal", "k_proj", "kv_in_dim", "memory_attention_downsample_rate", "memory_attention_hidden_size", "memory_attention_num_attention_heads", "memory_attention_rope_dropout", "nn", "num_attention_heads", "o_proj", "q_proj", "scaling", "self", "super", "v_proj"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoRoPECrossAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "FlashAttentionKwargs", "ModelRoPECrossAttention", "Module", "None", "Tensor", "Unpack", "_attn_implementation", "apply_rotary_pos_emb_2d_cross_attn", "attention_interface", "attention_mask", "attn_output", "attn_weights", "batch_size", "class", "config", "contiguous", "cos", "cos_k", "def", "dropout", "dropout_p", "eager", "eager_attention_forward", "else", "forward", "head_dim", "if", "int", "is_causal", "k_proj", "key", "kwargs", "new_shape", "nn", "not", "num_attention_heads", "num_k_exclude_rope", "o_proj", "point_batch_size", "position_embeddings", "position_embeddings_k", "q_proj", "query", "repeat_freqs_k", "reshape", "return", "rope_k_repeat", "scaling", "self", "shape", "sin", "sin_k", "torch", "training", "transpose", "tuple", "v_proj", "value", "view"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoTwoWayAttentionBlock.__init__": ["False", "LayerNorm", "ModelAttention", "ModelFeedForward", "ModelMaskDecoderConfig", "ModelTwoWayAttentionBlock", "Module", "__init__", "bool", "class", "config", "cross_attn_image_to_token", "cross_attn_token_to_image", "def", "downsample_rate", "hidden_size", "layer_norm1", "layer_norm2", "layer_norm3", "layer_norm4", "mlp", "mlp_dim", "nn", "num_hidden_layers", "num_layers", "self", "self_attn", "skip_first_layer_pe", "super"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoTwoWayAttentionBlock.forward": ["ModelTwoWayAttentionBlock", "Module", "Tensor", "TransformersKwargs", "Unpack", "_", "attention_similarity", "attn_out", "class", "cross_attn_image_to_token", "cross_attn_token_to_image", "def", "else", "forward", "if", "key", "key_point_embedding", "keys", "kwargs", "layer_norm1", "layer_norm2", "layer_norm3", "layer_norm4", "mlp", "mlp_out", "nn", "queries", "query", "query_point_embedding", "return", "self", "self_attn", "skip_first_layer_pe", "value"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoPositionEmbeddingSine.__init__": ["False", "ModelPositionEmbeddingSine", "Module", "None", "True", "ValueError", "__init__", "and", "be", "bool", "class", "def", "else", "float", "if", "int", "is", "math", "nn", "normalize", "not", "num_pos_feats", "passed", "pi", "raise", "scale", "self", "should", "super", "temperature"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoPositionEmbeddingSine.forward": ["ModelPositionEmbeddingSine", "Module", "None", "Size", "Tensor", "arange", "bool", "cat", "class", "cos", "cumsum", "def", "device", "dim", "dim_t", "div", "dtype", "eps", "flatten", "floor", "forward", "if", "int64", "is", "mask", "nn", "normalize", "not_mask", "num_pos_feats", "permute", "pos", "pos_x", "pos_y", "return", "rounding_mode", "scale", "self", "shape", "sin", "stack", "str", "temperature", "to", "torch", "x_embed", "y_embed", "zeros"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoMemoryFuser.__init__": ["ModelConfig", "ModelMemoryFuser", "ModelMemoryFuserCXBlock", "Module", "ModuleList", "_", "__init__", "class", "config", "def", "for", "in", "layers", "memory_fuser_num_layers", "nn", "range", "self", "super"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoMemoryFuser.forward": ["ModelMemoryFuser", "Module", "class", "def", "for", "forward", "hidden_states", "in", "layer", "layers", "nn", "return", "self"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoMaskDownSamplerLayer.__init__": ["ACT2FN", "Conv2d", "ModelConfig", "ModelLayerNorm", "ModelMaskDownSamplerLayer", "Module", "__init__", "activation", "channels_first", "class", "config", "conv", "data_format", "def", "eps", "in_channels", "int", "kernel_size", "layer_norm", "mask_downsampler_hidden_act", "mask_downsampler_kernel_size", "mask_downsampler_padding", "mask_downsampler_stride", "nn", "out_channels", "padding", "self", "stride", "super"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoMaskDownSamplerLayer.forward": ["ModelMaskDownSamplerLayer", "Module", "activation", "class", "conv", "def", "forward", "layer_norm", "nn", "return", "self", "x"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoMaskDownSampler.__init__": ["ACT2FN", "Conv2d", "ModelConfig", "ModelMaskDownSampler", "ModelMaskDownSamplerLayer", "Module", "ModuleList", "_", "__init__", "activation", "append", "class", "config", "def", "final_conv", "for", "in", "int", "kernel_size", "layers", "log2", "mask_downsampler_embed_dim", "mask_downsampler_hidden_act", "mask_downsampler_stride", "mask_downsampler_total_stride", "mask_in_chans", "mask_out_chans", "math", "nn", "num_layers", "range", "self", "super"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoMaskDownSampler.forward": ["ModelMaskDownSampler", "Module", "class", "def", "final_conv", "for", "forward", "in", "layer", "layers", "nn", "return", "self", "x"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoMemoryEncoder.__init__": ["Conv2d", "ModelConfig", "ModelMaskDownSampler", "ModelMemoryEncoder", "ModelMemoryFuser", "ModelPositionEmbeddingSine", "Module", "True", "__init__", "class", "config", "def", "feature_projection", "hidden_size", "kernel_size", "mask_downsampler", "memory_encoder_hidden_size", "memory_encoder_output_channels", "memory_fuser", "nn", "normalize", "num_pos_feats", "output_channels", "position_encoding", "projection", "self", "super"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoMemoryEncoder.forward": ["ModelMemoryEncoder", "Module", "Tensor", "class", "def", "device", "dtype", "feature_projection", "forward", "mask_downsampler", "masks", "memory_fuser", "nn", "position_encoding", "projection", "return", "self", "shape", "torch", "tuple", "vision_features", "vision_pos_enc"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoFeedForward.__init__": ["ACT2FN", "False", "Linear", "ModelFeedForward", "Module", "ModuleList", "_", "__init__", "activation", "bool", "class", "def", "for", "hidden_dim", "in", "input_dim", "int", "layers", "nn", "num_layers", "output_dim", "proj_in", "proj_out", "range", "relu", "self", "sigmoid_output", "str", "super"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoFeedForward.forward": ["F", "ModelFeedForward", "Module", "activation", "class", "def", "for", "forward", "hidden_states", "if", "in", "layer", "layers", "nn", "proj_in", "proj_out", "return", "self", "sigmoid", "sigmoid_output"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoPositionalEmbedding.__init__": ["ModelPositionalEmbedding", "ModelPromptEncoderConfig", "Module", "__init__", "class", "config", "def", "hidden_size", "nn", "positional_embedding", "randn", "register_buffer", "scale", "self", "super", "torch"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoPositionalEmbedding.forward": ["ModelPositionalEmbedding", "Module", "None", "cat", "class", "clone", "coordinates", "cos", "def", "dim", "dtype", "float32", "forward", "if", "input_coords", "input_shape", "is", "nn", "not", "np", "pi", "positional_embedding", "return", "self", "sin", "to", "torch"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoPreTrainedModel._init_weights": ["ModelMemoryFuserCXBlock", "ModelModel", "ModelPositionalEmbedding", "ModelPreTrainedModel", "ModelVisionRotaryEmbedding", "None", "PreTrainedModel", "_init_weights", "class", "copy_", "cos", "create_inv_freq", "def", "elif", "if", "init", "inv_freq", "is", "isinstance", "memory_temporal_positional_encoding", "module", "no_memory_positional_encoding", "no_object_pointer", "normal_", "not", "occlusion_spatial_embedding_parameter", "positional_embedding", "rope_embeddings_cos", "rope_embeddings_sin", "scale", "self", "sin", "std", "super", "zeros_"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoInferenceCache.__init__": ["ModelInferenceCache", "__init__", "_vision_features", "class", "cpu", "def", "device", "inference_device", "inference_state_device", "int", "max_vision_features_cache_size", "self", "str", "torch"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoInferenceCache.cache_vision_features": ["ModelInferenceCache", "Tensor", "True", "_vision_features", "and", "cache_vision_features", "cached", "class", "def", "dict", "elif", "else", "features", "for", "frame_idx", "if", "in", "inference_state_device", "int", "isinstance", "items", "key", "keys", "len", "list", "max_vision_features_cache_size", "min", "non_blocking", "pop", "self", "to", "torch", "tuple", "v", "value"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoInferenceCache.get_vision_features": ["ModelInferenceCache", "None", "Tensor", "True", "_vision_features", "and", "cached", "class", "def", "dict", "elif", "else", "for", "frame_idx", "get_vision_features", "if", "in", "inference_device", "int", "isinstance", "items", "key", "list", "moved", "non_blocking", "not", "return", "self", "to", "torch", "tuple", "v", "value"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoInferenceCache.clear_all": ["ModelInferenceCache", "_vision_features", "class", "clear", "clear_all", "def", "self"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoInferenceSession.__init__": ["FloatTensor", "ModelInferenceCache", "ModelInferenceSession", "None", "OrderedDict", "__init__", "_obj_id_to_idx", "_obj_idx_to_id", "cache", "class", "cpu", "def", "device", "dict", "dtype", "else", "enumerate", "float32", "frames_tracked_per_obj", "if", "inference_device", "inference_state_device", "int", "is", "mask_inputs_per_obj", "max_vision_features_cache_size", "not", "obj_ids", "obj_with_new_inputs", "output_dict_per_obj", "point_inputs_per_obj", "processed_frames", "self", "str", "to", "torch", "video", "video_height", "video_storage_device", "video_width"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoInferenceSession.num_frames": ["ModelInferenceSession", "None", "class", "def", "else", "if", "int", "is", "len", "not", "num_frames", "processed_frames", "return", "self"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoInferenceSession.obj_id_to_idx": ["ModelInferenceSession", "None", "_obj_id_to_idx", "_obj_idx_to_id", "class", "cond_frame_outputs", "def", "frames_tracked_per_obj", "get", "if", "int", "is", "len", "list", "mask_inputs_per_obj", "non_cond_frame_outputs", "not", "obj_id", "obj_id_to_idx", "obj_ids", "obj_idx", "output_dict_per_obj", "point_inputs_per_obj", "return", "self"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoInferenceSession.obj_idx_to_id": ["ModelInferenceSession", "_obj_idx_to_id", "class", "def", "int", "obj_idx", "obj_idx_to_id", "return", "self"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoInferenceSession.get_obj_num": ["ModelInferenceSession", "_obj_idx_to_id", "class", "def", "get_obj_num", "int", "len", "return", "self"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoInferenceSession.add_point_inputs": ["False", "ModelInferenceSession", "Tensor", "add_point_inputs", "class", "def", "device_inputs", "dict", "else", "for", "frame_idx", "if", "in", "inference_device", "inputs", "int", "isinstance", "items", "key", "non_blocking", "obj_idx", "point_inputs_per_obj", "self", "to", "torch", "value"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoInferenceSession.remove_point_inputs": ["ModelInferenceSession", "None", "class", "def", "frame_idx", "int", "obj_idx", "point_inputs_per_obj", "pop", "remove_point_inputs", "self"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoInferenceSession.add_mask_inputs": ["ModelInferenceSession", "Tensor", "True", "add_mask_inputs", "class", "def", "dtype", "frame_idx", "inference_device", "inputs", "int", "mask_inputs_per_obj", "non_blocking", "obj_idx", "self", "to", "torch"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoInferenceSession.remove_mask_inputs": ["ModelInferenceSession", "None", "class", "def", "frame_idx", "int", "mask_inputs_per_obj", "obj_idx", "pop", "remove_mask_inputs", "self"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoInferenceSession.store_output": ["ModelInferenceSession", "None", "Tensor", "True", "and", "bool", "class", "cond_frame_outputs", "def", "dict", "elif", "else", "for", "frame_idx", "if", "in", "inference_state_device", "int", "is", "is_conditioning_frame", "isinstance", "items", "key", "non_blocking", "non_cond_frame_outputs", "obj_idx", "object_pointer", "object_score_logits", "output_dict_per_obj", "output_key", "output_value", "return", "self", "storage_key", "store_output", "str", "to", "torch", "value"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoInferenceSession.get_output": ["ModelInferenceSession", "None", "Tensor", "True", "bool", "class", "cond_frame_outputs", "def", "else", "frame_idx", "get", "get_output", "if", "inference_device", "int", "is", "is_conditioning_frame", "isinstance", "non_blocking", "non_cond_frame_outputs", "obj_idx", "out", "output_dict_per_obj", "output_key", "return", "self", "storage_key", "str", "to", "torch", "value"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoInferenceSession.add_new_frame": ["ModelInferenceSession", "None", "Tensor", "True", "add_new_frame", "class", "def", "dim", "dtype", "else", "frame_idx", "if", "int", "is", "len", "non_blocking", "not", "pixel_values", "processed_frames", "return", "self", "squeeze", "to", "torch", "video_storage_device"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoInferenceSession.get_frame": ["ModelInferenceSession", "Tensor", "True", "class", "def", "frame_idx", "get_frame", "inference_device", "int", "non_blocking", "processed_frames", "return", "self", "to", "torch"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoInferenceSession.reset_tracking_data": ["ModelInferenceSession", "_obj_id_to_idx", "_obj_idx_to_id", "class", "clear", "def", "frames_tracked_per_obj", "mask_inputs_per_obj", "obj_ids", "obj_with_new_inputs", "output_dict_per_obj", "point_inputs_per_obj", "reset_tracking_data", "self"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoInferenceSession.reset_inference_session": ["ModelInferenceSession", "_obj_id_to_idx", "_obj_idx_to_id", "cache", "class", "clear", "clear_all", "def", "frames_tracked_per_obj", "mask_inputs_per_obj", "obj_ids", "obj_with_new_inputs", "output_dict_per_obj", "point_inputs_per_obj", "reset_inference_session", "self"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoMemoryAttentionMLP.__init__": ["ACT2FN", "Dropout", "Linear", "ModelConfig", "ModelMemoryAttentionMLP", "Module", "__init__", "act_fn", "class", "config", "def", "down_proj", "dropout", "hidden_size", "intermediate_size", "memory_attention_dropout", "memory_attention_hidden_size", "memory_attention_mlp_hidden_act", "memory_attention_mlp_hidden_size", "nn", "self", "super", "up_proj"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoMemoryAttentionMLP.forward": ["ModelMemoryAttentionMLP", "Module", "act_fn", "class", "def", "down_proj", "dropout", "forward", "nn", "return", "self", "up_proj", "x"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoMemoryAttentionLayer.__init__": ["Dropout", "LayerNorm", "ModelConfig", "ModelMemoryAttentionLayer", "ModelMemoryAttentionMLP", "ModelRoPECrossAttention", "ModelRoPESelfAttention", "Module", "__init__", "class", "config", "cross_attn_image", "def", "dropout1", "dropout2", "dropout3", "hidden_size", "kv_in_dim", "layer_norm1", "layer_norm2", "layer_norm3", "memory_attention_dropout", "memory_attention_hidden_size", "mlp", "nn", "self", "self_attn", "super"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoMemoryAttentionLayer.forward": ["ModelMemoryAttentionLayer", "Module", "None", "Tensor", "_", "class", "cross_attn_image", "def", "dropout1", "dropout2", "dropout3", "forward", "int", "key", "key_point_embedding", "keys", "layer_norm1", "layer_norm2", "layer_norm3", "mlp", "nn", "num_k_exclude_rope", "position_embeddings", "position_embeddings_k", "queries", "query", "return", "rope_k_repeat", "rope_position_embeddings", "rope_position_embeddings_k", "self", "self_attn", "torch", "tuple", "value"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoMemoryAttention.__init__": ["LayerNorm", "ModelConfig", "ModelMemoryAttention", "ModelMemoryAttentionLayer", "ModelVisionRotaryEmbedding", "Module", "ModuleList", "_", "__init__", "class", "config", "def", "end_x", "end_y", "for", "in", "layer_norm", "layers", "memory_attention_hidden_size", "memory_attention_num_layers", "memory_attention_rope_k_sizes", "nn", "range", "rotary_emb", "rotary_emb_k", "self", "super"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoMemoryAttention.forward": ["ModelMemoryAttention", "Module", "None", "Tensor", "class", "current_vision_features", "current_vision_position_embeddings", "def", "else", "for", "forward", "if", "in", "int", "is", "key_point_embedding", "keys", "layer", "layer_norm", "layers", "memory", "memory_posision_embeddings", "ndim", "nn", "normed_output", "not", "num_k_exclude_rope", "num_object_pointer_tokens", "num_spatial_memory_tokens", "output", "queries", "return", "rope_k_repeat", "rope_position_embeddings", "rope_position_embeddings_k", "rotary_emb", "rotary_emb_k", "self", "torch", "transpose", "unsqueeze"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoPerceiverMLP.__init__": ["False", "GELU", "LayerNorm", "Linear", "ModelConfig", "ModelPerceiverMLP", "Module", "__init__", "act_fn", "bias", "class", "config", "def", "down_proj", "hidden_size", "intermediate_size", "layer_norm", "nn", "perceiver_resampler_hidden_size", "perceiver_resampler_mlp_intermediate_size", "self", "super", "up_proj"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoPerceiverMLP.forward": ["ModelPerceiverMLP", "Module", "Tensor", "act_fn", "class", "def", "down_proj", "forward", "hidden_states", "layer_norm", "nn", "return", "self", "torch", "up_proj"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoPerceiverAttention.__init__": ["False", "Linear", "ModelConfig", "ModelPerceiverAttention", "Module", "__init__", "attention_dropout", "bias", "class", "config", "def", "head_dim", "hidden_size", "inner_dim", "is_causal", "k_proj", "nn", "num_attention_heads", "o_proj", "perceiver_resampler_attention_dropout", "perceiver_resampler_attention_head_dim", "perceiver_resampler_hidden_size", "perceiver_resampler_num_attention_heads", "q_proj", "scaling", "self", "super", "v_proj"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoPerceiverAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "ModelPerceiverAttention", "Module", "None", "Tensor", "_", "_attn_implementation", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "batch_size", "class", "config", "contiguous", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "head_dim", "if", "inner_dim", "is", "is_causal", "k_proj", "key", "kwargs", "nn", "not", "num_attention_heads", "o_proj", "pos_encoding", "positional_encoding", "q_proj", "query", "return", "scaling", "self", "seq_len_kv", "seq_len_q", "shape", "torch", "training", "transpose", "v_proj", "value", "view"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoPerceiverEncoderLayer.__init__": ["Dropout", "LayerNorm", "ModelConfig", "ModelPerceiverAttention", "ModelPerceiverEncoderLayer", "ModelPerceiverMLP", "Module", "__init__", "class", "config", "cross_attention", "def", "dropout", "layer_norm_input", "layer_norm_latents", "layer_norm_self", "mlp", "nn", "perceiver_resampler_hidden_dropout", "perceiver_resampler_hidden_size", "self", "self_attention", "self_mlp", "super"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoPerceiverEncoderLayer.forward": ["ModelPerceiverEncoderLayer", "Module", "None", "Tensor", "class", "cross_attention", "cross_attention_output", "def", "dropout", "forward", "input_features", "key", "latents", "layer_norm_input", "layer_norm_latents", "layer_norm_self", "mlp", "mlp_output", "nn", "normalized_input", "normalized_latents", "normalized_latents_self", "positional_encoding", "query", "return", "self", "self_attention", "self_attention_output", "self_mlp", "self_mlp_output", "torch", "value"], "edgetam_video/modeling_edgetam_video.py:window_partition": ["Model_partition", "Model_size", "Models", "batch_size", "contiguous", "def", "functional", "height", "hidden_state", "nn", "num_channels", "pad", "pad_height", "pad_width", "padded_height", "padded_width", "permute", "return", "shape", "view", "width"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoPerceiverResampler.__init__": ["LayerNorm", "ModelConfig", "ModelPerceiverEncoderLayer", "ModelPerceiverResampler", "ModelPositionEmbeddingSine", "Module", "ModuleList", "Parameter", "True", "_", "__init__", "class", "config", "def", "for", "hidden_size", "if", "in", "latents_1d", "latents_2d", "layer_norm", "layers", "nn", "normalize", "num_latents_1d", "num_latents_2d", "num_layers", "num_pos_feats", "perceiver_resampler_hidden_size", "perceiver_resampler_num_latents", "perceiver_resampler_num_latents_2d", "perceiver_resampler_num_layers", "positional_encoding", "randn", "range", "self", "super", "torch"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoPerceiverResampler.forward": ["ModelPerceiverResampler", "Module", "None", "Tensor", "_forward_1d", "_forward_2d", "and", "append", "cat", "class", "combined_latents", "combined_positional_encoding", "def", "dim", "forward", "hidden_states", "if", "is", "latents_1d", "latents_2d", "nn", "not", "num_latents_1d", "num_latents_2d", "output_latents", "output_positional_encodings", "pos_1d", "pos_2d", "positional_encoding", "return", "self", "torch", "tuple"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoPerceiverResampler._forward_1d": ["ModelPerceiverResampler", "Module", "None", "Tensor", "_forward_1d", "batch_size", "class", "def", "expand", "flatten", "flattened_features", "for", "hidden_states", "if", "in", "is", "latents", "latents_1d", "layer", "layer_norm", "layers", "nn", "not", "output_positional_encoding", "permute", "positional_encoding", "positional_features", "return", "self", "shape", "torch", "tuple", "unsqueeze", "zeros_like"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoPerceiverResampler._forward_2d": ["ModelPerceiverResampler", "Module", "None", "Tensor", "_", "_forward_2d", "batch_size", "channels", "class", "def", "device", "dtype", "expand", "flatten", "for", "height", "hidden_states", "in", "int", "latents_2d", "layer", "layer_norm", "layers", "math", "nn", "num_latents_2d", "num_windows_per_dim", "permute", "positional_encoding", "positional_encoding_2d", "return", "self", "shape", "sqrt", "to", "torch", "tuple", "unsqueeze", "view", "width", "window_partition", "window_size", "windowed_features", "windowed_input"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoMaskEmbedding.__init__": ["ACT2FN", "Conv2d", "ModelLayerNorm", "ModelMaskEmbedding", "ModelPromptEncoderConfig", "Module", "__init__", "activation", "channels_first", "class", "config", "conv1", "conv2", "conv3", "data_format", "def", "eps", "hidden_act", "hidden_size", "kernel_size", "layer_norm1", "layer_norm2", "layer_norm_eps", "mask_input_channels", "nn", "self", "stride", "super"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoMaskEmbedding.forward": ["ModelMaskEmbedding", "Module", "activation", "class", "conv1", "conv2", "conv3", "def", "dense_embeddings", "forward", "hidden_states", "layer_norm1", "layer_norm2", "masks", "nn", "return", "self"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoPromptEncoder.__init__": ["Embedding", "ModelMaskEmbedding", "ModelPositionalEmbedding", "ModelPromptEncoder", "ModelPromptEncoderConfig", "Module", "__init__", "class", "config", "def", "hidden_size", "image_embedding_size", "image_size", "input_image_size", "mask_embed", "mask_input_size", "nn", "no_mask_embed", "not_a_point_embed", "num_point_embeddings", "patch_size", "point_embed", "self", "shared_embedding", "super"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoPromptEncoder._embed_points": ["ModelPromptEncoder", "Module", "None", "Tensor", "_embed_points", "bool", "clamp", "class", "constant", "def", "functional", "if", "input_image_size", "input_shape", "labels", "min", "mode", "nn", "not_a_point_embed", "pad", "point_embed", "point_embedding", "points", "return", "self", "shared_embedding", "torch", "unsqueeze", "value", "weight", "where", "zeros_like"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoPromptEncoder._embed_boxes": ["ModelPromptEncoder", "Module", "Tensor", "_embed_boxes", "boxes", "class", "constant", "coords", "corner_embedding", "def", "expand_as", "functional", "input_image_size", "mode", "nn", "not_a_point_embed", "pad", "point_embed", "return", "self", "shape", "shared_embedding", "torch", "value", "view", "weight"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoPromptEncoder.forward": ["If", "ModelPromptEncoder", "Module", "None", "Tensor", "ValueError", "_embed_boxes", "_embed_points", "also", "are", "batch_size", "be", "box_embeddings", "cat", "class", "def", "dense_embeddings", "dim", "else", "expand", "forward", "if", "image_embedding_size", "input_boxes", "input_labels", "input_masks", "input_points", "is", "labels", "mask_embed", "must", "nn", "no_mask_embed", "not", "pad", "point_embeddings", "points", "provided", "raise", "reshape", "return", "self", "shape", "sparse_embeddings", "torch", "tuple", "weight"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoTwoWayTransformer.__init__": ["LayerNorm", "ModelAttention", "ModelMaskDecoderConfig", "ModelTwoWayAttentionBlock", "ModelTwoWayTransformer", "Module", "ModuleList", "__init__", "append", "class", "config", "def", "final_attn_token_to_image", "for", "hidden_size", "i", "in", "layer_norm_final_attn", "layers", "nn", "num_hidden_layers", "range", "self", "skip_first_layer_pe", "super"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoTwoWayTransformer.forward": ["BaseModelOutput", "ModelTwoWayTransformer", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "_", "an", "attention_similarity", "attn_out", "class", "def", "final_attn_token_to_image", "flatten", "for", "forward", "have", "if", "image_embedding", "image_embeddings", "image_positional_embeddings", "in", "is", "key", "key_point_embedding", "keys", "kwargs", "layer", "layer_norm_final_attn", "layers", "nn", "not", "permute", "point_embeddings", "queries", "query", "query_point_embedding", "raise", "return", "self", "specify", "target_embedding", "to", "tuple", "unsqueeze", "value"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoMaskDecoder.__init__": ["Conv2d", "ConvTranspose2d", "Embedding", "GELU", "ModelFeedForward", "ModelLayerNorm", "ModelMaskDecoder", "ModelMaskDecoderConfig", "ModelTwoWayTransformer", "Module", "ModuleList", "True", "_", "__init__", "activation", "channels_first", "class", "config", "conv_s0", "conv_s1", "data_format", "def", "dynamic_multimask_stability_delta", "dynamic_multimask_stability_thresh", "dynamic_multimask_via_stability", "for", "hidden_size", "in", "iou_head_depth", "iou_head_hidden_dim", "iou_prediction_head", "iou_token", "kernel_size", "mask_tokens", "mlps_list", "nn", "num_mask_tokens", "num_multimask_outputs", "obj_score_token", "output_hypernetworks_mlps", "pred_obj_score_head", "range", "self", "sigmoid_output", "stride", "super", "transformer", "upscale_conv1", "upscale_conv2", "upscale_layer_norm"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoMaskDecoder.forward": ["ModelMaskDecoder", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "_dynamic_multimask_via_stability", "activation", "and", "attention_similarity", "batch_size", "bool", "cat", "class", "current_mlp", "def", "dense_prompt_embeddings", "dim", "dtype", "dynamic_multimask_via_stability", "elif", "else", "feat_s0", "feat_s1", "for", "forward", "height", "high_resolution_features", "hyper_in", "hyper_in_list", "i", "if", "image_embeddings", "image_positional_embeddings", "in", "iou_pred", "iou_prediction_head", "iou_token", "iou_token_out", "kwargs", "list", "mask_slice", "mask_tokens", "mask_tokens_out", "masks", "multimask_output", "nn", "not", "num_channels", "num_mask_tokens", "obj_score_token", "object_score_logits", "output_hypernetworks_mlps", "output_tokens", "point_batch_size", "point_embeddings", "pred_obj_score_head", "range", "repeat", "repeat_interleave", "return", "sam_tokens_out", "self", "shape", "slice", "sparse_prompt_embeddings", "stack", "target_embedding", "to", "tokens", "torch", "training", "transformer", "transpose", "tuple", "upscale_conv1", "upscale_conv2", "upscale_layer_norm", "upscaled_embedding", "view", "weight", "width"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoMaskDecoder._get_stability_scores": ["ModelMaskDecoder", "Module", "_get_stability_scores", "area_i", "area_u", "class", "def", "dim", "dynamic_multimask_stability_delta", "flatten", "float", "mask_logits", "nn", "return", "self", "stability_delta", "stability_scores", "sum", "torch", "where"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoMaskDecoder._dynamic_multimask_via_stability": ["ModelMaskDecoder", "Module", "None", "_dynamic_multimask_via_stability", "_get_stability_scores", "all_iou_scores", "all_mask_logits", "argmax", "best_multimask_iou_scores", "best_multimask_logits", "best_scores_inds", "best_scores_inds_expanded", "class", "def", "dim", "dynamic_multimask_stability_thresh", "expand", "expand_as", "gather", "iou_scores_out", "is_stable", "mask_logits_out", "multimask_iou_scores", "multimask_logits", "nn", "return", "self", "singlemask_iou_scores", "singlemask_logits", "size", "stability_scores", "torch", "unsqueeze", "where"], "edgetam_video/modeling_edgetam_video.py:get_1d_sine_pe": ["Model_1d_sine_pe", "arange", "cat", "cos", "def", "device", "dim", "dim_t", "dtype", "float32", "pe_dim", "pos_embed", "pos_inds", "return", "sin", "temperature", "torch", "unsqueeze"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoModel.__init__": ["AutoModel", "Conv2d", "Identity", "Linear", "ModelConfig", "ModelFeedForward", "ModelMaskDecoder", "ModelMemoryAttention", "ModelMemoryEncoder", "ModelModel", "ModelPerceiverResampler", "ModelPositionalEmbedding", "ModelPreTrainedModel", "ModelPromptEncoder", "None", "Parameter", "__init__", "_attn_implementation", "backbone_feature_sizes", "class", "config", "def", "else", "enable_occlusion_spatial_embedding", "enable_temporal_pos_encoding_for_object_pointers", "fpn_hidden_size", "from_config", "hidden_dim", "if", "image_size", "kernel_size", "mask_decoder", "mask_decoder_config", "mask_downsample", "mem_dim", "memory_attention", "memory_encoder", "memory_encoder_output_channels", "memory_temporal_positional_encoding", "nn", "no_memory_embedding", "no_memory_positional_encoding", "no_object_pointer", "num_feature_levels", "num_maskmem", "object_pointer_proj", "occlusion_spatial_embedding_parameter", "post_init", "prompt_encoder", "prompt_encoder_config", "self", "shared_image_embedding", "spatial_perceiver", "stride", "super", "temporal_positional_encoding_projection_layer", "torch", "vision_config", "vision_encoder", "zeros"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "return", "self", "vision_encoder"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoModel.get_image_wide_positional_embeddings": ["ModelModel", "ModelPreTrainedModel", "Tensor", "class", "cumsum", "def", "device", "dim", "dtype", "get_image_wide_positional_embeddings", "grid", "image_embedding_size", "ones", "permute", "positional_embedding", "prompt_encoder", "return", "self", "shared_image_embedding", "size", "stack", "target_device", "target_dtype", "torch", "unsqueeze", "x_embed", "y_embed"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoModel.get_image_embeddings": ["FloatTensor", "ModelModel", "ModelPreTrainedModel", "Tensor", "TransformersKwargs", "Unpack", "_", "backbone_feature_sizes", "batch_size", "class", "def", "feat", "feat_size", "feature_maps", "for", "get_image_embeddings", "get_image_features", "image_embeddings", "in", "kwargs", "list", "no_memory_embedding", "permute", "pixel_values", "r", "return", "self", "shape", "torch", "view", "zip"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoModel.get_prompt_embeddings": ["FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "class", "def", "get_prompt_embeddings", "input_boxes", "input_labels", "input_masks", "input_points", "prompt_encoder", "prompt_output", "r", "return", "self", "torch", "tuple"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoModel.forward": ["False", "ModelInferenceSession", "ModelModel", "ModelPreTrainedModel", "ModelSegmentationOutput", "No", "None", "Tensor", "True", "ValueError", "_run_single_frame_inference", "add", "add_new_frame", "all_object_score_logits", "all_pred_masks", "and", "are", "batch_size", "bool", "cat", "class", "cond_frame_outputs", "copy", "current_out", "def", "dim", "else", "first", "for", "forward", "frame", "frame_idx", "frames_tracked_per_obj", "get", "get_obj_num", "get_output", "has_cond_output", "has_new_inputs", "if", "in", "inference_session", "inputs", "int", "is", "is_conditioning_frame", "is_init_cond_frame", "kwargs", "len", "mask_inputs", "mask_inputs_per_obj", "not", "num_objects", "obj_id", "obj_ids", "obj_idx", "obj_idx_to_id", "obj_with_new_inputs", "object_ids", "object_score_logits", "object_score_logits_per_obj", "objects", "or", "output_dict_per_obj", "output_value", "please", "point_inputs", "point_inputs_per_obj", "pred_masks", "pred_masks_per_obj", "provided", "r", "raise", "range", "remove", "return", "reverse", "run_mem_encoder", "self", "squeeze", "store_output", "streaming", "torch", "tracking"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoModel.get_image_features": ["FloatTensor", "ModelModel", "ModelPreTrainedModel", "ModelVisionEncoderOutput", "None", "Tensor", "TransformersKwargs", "Unpack", "attentions", "class", "conv_s0", "conv_s1", "def", "feature_map", "feature_map_position_embedding", "feature_maps", "feature_maps_position_embeddings", "flatten", "for", "fpn_hidden_states", "fpn_position_encoding", "get_image_features", "hidden_states", "in", "kwargs", "list", "mask_decoder", "permute", "pixel_values", "r", "return", "self", "torch", "tuple", "vision_encoder", "vision_outputs"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoModel._prepare_vision_features": ["ModelInferenceSession", "ModelModel", "ModelPreTrainedModel", "Tensor", "_", "_prepare_vision_features", "batch_size", "cache", "cache_vision_features", "cached_features", "class", "def", "else", "expand", "for", "frame_idx", "get_frame", "get_image_features", "get_vision_features", "if", "image_batch", "in", "inference_session", "int", "list", "pe", "return", "self", "torch", "tuple", "unsqueeze", "vision_feats", "vision_pos_embeds"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoModel._single_frame_forward": ["Exactly", "F", "False", "FloatTensor", "Got", "LongTensor", "ModelImageSegmentationOutput", "ModelModel", "ModelPreTrainedModel", "NO_OBJ_SCORE", "None", "TransformersKwargs", "True", "Unpack", "ValueError", "You", "_", "_single_frame_forward", "align_corners", "and", "antialias", "arange", "argmax", "as", "attention_similarity", "backbone_feature_sizes", "batch_inds", "batch_size", "be", "best_iou_inds", "bilinear", "bool", "bounding", "box", "boxes", "class", "def", "dense_embeddings", "dense_prompt_embeddings", "device", "dim", "dtype", "elif", "else", "f", "feat", "feat_size", "feature_maps", "float", "for", "get_image_features", "get_image_wide_positional_embeddings", "high_res_masks", "high_res_multimasks", "high_resolution_features", "if", "image_embeddings", "image_positional_embeddings", "image_size", "in", "input", "input_boxes", "input_labels", "input_masks", "input_points", "int", "int32", "interpolate", "iou_scores", "is", "is_obj_appearing", "kwargs", "lambda_is_obj_appearing", "low_res_masks", "low_res_multimasks", "many", "mask_decoder", "mask_input_size", "mode", "multimask_output", "must", "no_memory_embedding", "no_object_pointer", "not", "num_objects", "object_batch_inds", "object_pointer", "object_pointer_proj", "object_score_logits", "of", "one", "ones", "ones_like", "or", "per", "permute", "pixel_values", "points", "pred_masks", "prompt_encoder", "provide", "provided", "raise", "repeat", "return", "sam_output_token", "sam_output_tokens", "self", "shape", "should", "size", "sparse_embeddings", "sparse_prompt_embeddings", "squeeze", "target_embedding", "to", "torch", "unsqueeze", "view", "vision_attentions", "vision_hidden_states", "where", "zeros", "zip"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoModel._use_mask_as_output": ["F", "False", "ModelImageSegmentationOutput", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "True", "_single_frame_forward", "_use_mask_as_output", "align_corners", "antialias", "any", "backbone_features", "bilinear", "class", "def", "dim", "dtype", "flatten", "float", "high_res_features", "high_res_masks", "image_embeddings", "input_masks", "interpolate", "iou_scores", "is_obj_appearing", "lambda_is_obj_appearing", "list", "low_res_masks", "mask_downsample", "mask_inputs", "mask_inputs_float", "mode", "new_ones", "no_object_pointer", "object_pointer", "object_score_logits", "out_bias", "out_scale", "pred_masks", "return", "self", "size", "to", "torch"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoModel._select_closest_cond_frames": ["ModelModel", "ModelPreTrainedModel", "None", "_select_closest_cond_frames", "abs", "class", "cond_frame_outputs", "def", "default", "else", "for", "frame_idx", "idx_after", "idx_before", "if", "in", "inds_remain", "is", "items", "key", "lambda", "len", "max", "max_cond_frame_num", "min", "not", "num_remain", "or", "return", "selected_outputs", "self", "sorted", "t", "unselected_outputs", "update", "v", "x"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoModel._gather_memory_frame_outputs": ["False", "ModelInferenceSession", "ModelModel", "ModelPreTrainedModel", "None", "ValueError", "_gather_memory_frame_outputs", "_select_closest_cond_frames", "append", "be", "bool", "cannot", "class", "cond_frame_outputs", "conditioning", "conditioning_outputs", "config", "def", "dict", "else", "empty", "for", "frame_idx", "get", "if", "in", "inference_session", "int", "is_initial_conditioning_frame", "list", "maskmem_features", "max_cond_frame_num", "non_cond_frame_outputs", "not", "num_maskmem", "obj_idx", "out", "output_data", "output_dict_per_obj", "outputs", "previous_frame_idx", "raise", "range", "relative_temporal_offset", "return", "self", "temporal_positions_and_previous_outputs", "track_in_reverse_time", "tuple", "unselected_conditioning_outputs", "values", "when"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoModel._build_memory_attention_inputs": ["ModelModel", "ModelPreTrainedModel", "None", "Tensor", "True", "_build_memory_attention_inputs", "append", "class", "combined_memory_pos_embed", "continue", "def", "device", "dict", "for", "if", "in", "int", "is", "list", "maskmem_features", "maskmem_pos_enc", "memories_to_concatenate", "memory_features", "memory_positional_embeddings_to_concatenate", "memory_temporal_positional_encoding", "non_blocking", "permute", "prev_output_data", "relative_temporal_offset", "return", "self", "spatial_memory_pos_embed", "squeeze", "temporal_positions_and_previous_outputs", "to", "torch", "tuple"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoModel._get_object_pointers": ["False", "ModelInferenceSession", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "_get_object_pointers", "and", "append", "bool", "break", "class", "cond_frame_outputs", "conditioning_outputs", "config", "def", "device", "eligible_conditioning_outputs", "else", "for", "frame_idx", "get", "if", "in", "inference_session", "int", "is", "items", "list", "max_object_pointers_in_encoder", "max_object_pointers_to_use", "min", "non_cond_frame_outputs", "not", "num_total_frames", "obj_idx", "object_pointer", "or", "out", "out_data", "output_dict_per_obj", "pointer_tokens", "range", "ref_frame_idx", "return", "self", "streaming", "t_diff_offset", "temporal_difference", "temporal_idx", "temporal_offsets", "temporal_position_sign_multiplier", "to", "torch", "track_in_reverse_time", "training", "tuple"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoModel._process_object_pointers": ["ModelModel", "ModelPreTrainedModel", "None", "Tensor", "_process_object_pointers", "batch_size", "class", "config", "def", "device", "dim", "dtype", "else", "enable_temporal_pos_encoding_for_object_pointers", "expand", "flatten", "float", "float32", "get_1d_sine_pe", "if", "int", "len", "list", "max_object_pointers_to_use", "max_temporal_diff", "mem_dim", "new_zeros", "normalized_temporal_diffs", "not", "num_channels", "num_splits", "object_pointers", "object_pointers_pos_embed", "permute", "pointer_tokens", "pointer_tpos_dim", "projected_sine_pe", "repeat_interleave", "reshape", "return", "self", "sine_pe", "stack", "temporal_offsets", "temporal_positional_encoding_projection_layer", "tensor", "to", "torch", "tuple", "unsqueeze"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoModel._prepare_memory_conditioned_features": ["False", "ModelInferenceSession", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "_build_memory_attention_inputs", "_gather_memory_frame_outputs", "_get_object_pointers", "_prepare_memory_conditioned_features", "_process_object_pointers", "append", "backbone_feature_sizes", "batch_size", "bool", "cat", "class", "combined_memory", "combined_memory_positional_embeddings", "conditioned_feature_map", "conditioned_feature_map_flat", "current_feature_map", "current_vision_features", "current_vision_position_embeddings", "current_vision_positional_embeddings", "def", "device", "dim", "frame_idx", "height", "hidden_dim", "if", "inference_session", "int", "is", "is_initial_conditioning_frame", "len", "list", "max_object_pointers_to_use", "memories_to_concatenate", "memory", "memory_attention", "memory_posision_embeddings", "memory_positional_embeddings_to_concatenate", "no_memory_embedding", "not", "num_channels", "num_maskmem", "num_object_pointer_tokens", "num_spatial_memory_tokens", "num_total_frames", "obj_idx", "object_pointers", "object_pointers_pos_embed", "permute", "pointer_tokens", "return", "self", "shape", "size", "squeeze", "streaming", "temporal_offsets", "temporal_positions_and_previous_outputs", "torch", "track_in_reverse_time", "view", "width"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoModel._use_multimask": ["ModelModel", "ModelPreTrainedModel", "None", "_use_multimask", "and", "bool", "class", "config", "def", "dict", "else", "if", "is", "is_init_cond_frame", "multimask_max_pt_num", "multimask_min_pt_num", "multimask_output", "multimask_output_for_tracking", "multimask_output_in_sam", "num_pts", "or", "point_inputs", "point_labels", "return", "self", "size"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoModel._run_single_frame_inference": ["Any", "False", "ModelInferenceSession", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "ValueError", "_encode_new_memory", "_prepare_memory_conditioned_features", "_prepare_vision_features", "_run_single_frame_inference", "_single_frame_forward", "_use_mask_as_output", "_use_multimask", "and", "appear", "as", "backbone_feature_sizes", "batch_size", "bool", "class", "current_out", "current_vision_feats", "current_vision_features", "current_vision_pos_embeds", "current_vision_positional_embeddings", "def", "dict", "else", "for", "frame", "frame_idx", "hidden_dim", "high_res_features", "high_res_masks", "if", "image_embeddings", "in", "inference_session", "input", "input_labels", "input_masks", "input_points", "int", "is", "is_init_cond_frame", "is_initial_conditioning_frame", "is_mask_from_pts", "len", "mask_inputs", "maskmem_features", "maskmem_pos_enc", "multimask_output", "not", "num_frames", "num_maskmem", "num_total_frames", "obj_idx", "object_pointer", "object_score_logits", "on", "or", "permute", "pix_feat", "pixel_values", "point_coords", "point_inputs", "point_labels", "pred_masks", "pred_masks_high_res", "prev_sam_mask_logits", "raise", "return", "reverse", "run_mem_encoder", "s", "sam_outputs", "same", "self", "should", "simultaneously", "size", "str", "streaming", "the", "torch", "track_in_reverse_time", "training", "view", "x", "zip"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoModel._encode_new_memory": ["ModelModel", "ModelPreTrainedModel", "None", "Tensor", "_encode_new_memory", "and", "backbone_feature_sizes", "batch_size", "bool", "channels", "class", "config", "current_vision_feats", "def", "dtype", "else", "expand", "float", "height", "hidden_dim", "if", "is", "is_mask_from_pts", "is_obj_appearing", "list", "mask_for_mem", "maskmem_features", "maskmem_pos_enc", "memory_encoder", "not", "object_score_logits", "occlusion_spatial_embedding_parameter", "permute", "pix_feat", "pred_masks_high_res", "return", "self", "shape", "sigmoid", "sigmoid_bias_for_mem_enc", "sigmoid_scale_for_mem_enc", "size", "spatial_perceiver", "to", "torch", "training", "tuple", "view", "width"], "edgetam_video/modeling_edgetam_video.py:EdgeTamVideoModel.propagate_in_video_iterator": ["Cannot", "False", "Iterator", "ModelInferenceSession", "ModelModel", "ModelPreTrainedModel", "ModelSegmentationOutput", "Model_output", "None", "ValueError", "a", "bool", "class", "cond_frame_outputs", "def", "desc", "determine", "disable", "else", "end_frame_idx", "first", "for", "frame", "frame_idx", "frames_with_inputs", "if", "in", "index", "inference", "inference_session", "inputs", "int", "is", "it", "manually", "max", "max_frame_num_to_track", "min", "not", "num_frames", "obj_output_dict", "on", "or", "output_dict_per_obj", "please", "processing_order", "propagate", "propagate_in_video_iterator", "r", "raise", "range", "reverse", "run", "self", "show_progress_bar", "specify", "start_frame_idx", "starting", "the", "tqdm", "values", "video", "with", "yield"], "efficientloftr/modeling_efficientloftr.py:compute_embeddings": ["Model_embeddings", "Tensor", "cumsum", "def", "device", "dtype", "emb", "embed_height", "embed_width", "hidden_size", "i_indices", "int", "inv_freq", "j_indices", "ones", "return", "torch", "unsqueeze", "zeros"], "efficientloftr/modeling_efficientloftr.py:EfficientLoFTRRotaryEmbedding.__init__": ["Callable", "False", "ModelConfig", "ModelRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "if", "inv_freq", "nn", "original_inv_freq", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "efficientloftr/modeling_efficientloftr.py:EfficientLoFTRRotaryEmbedding.compute_default_rope_parameters": ["ModelConfig", "ModelRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "get", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "partial_rotary_factor", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "efficientloftr/modeling_efficientloftr.py:EfficientLoFTRRotaryEmbedding.forward": ["False", "LongTensor", "ModelRotaryEmbedding", "Module", "None", "Tensor", "and", "class", "compute_embeddings", "config", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "embed_height", "embed_width", "enabled", "feats_height", "feats_width", "forward", "hidden_size", "if", "inv_freq", "isinstance", "layer_type", "maybe_autocast", "mps", "nn", "position_ids", "q_aggregation_kernel_size", "q_aggregation_stride", "repeat_interleave", "return", "self", "shape", "sin", "str", "to", "torch", "tuple", "type", "with", "x"], "efficientloftr/modeling_efficientloftr.py:EfficientLoFTRConvNormLayer.__init__": ["ACT2CLS", "BatchNorm2d", "Conv2d", "False", "Identity", "ModelConvNormLayer", "Module", "None", "__init__", "activation", "batch_norm_eps", "bias", "class", "config", "conv", "def", "else", "if", "in_channels", "is", "kernel_size", "nn", "norm", "out_channels", "padding", "self", "stride", "super"], "efficientloftr/modeling_efficientloftr.py:EfficientLoFTRConvNormLayer.forward": ["ModelConvNormLayer", "Module", "activation", "class", "conv", "def", "forward", "hidden_state", "nn", "norm", "return", "self"], "efficientloftr/modeling_efficientloftr.py:EfficientLoFTRRepVGGBlock.__init__": ["ACT2FN", "BatchNorm2d", "GradientCheckpointingLayer", "Identity", "ModelConfig", "ModelConvNormLayer", "ModelRepVGGBlock", "None", "__init__", "activation", "activation_function", "and", "block_idx", "class", "config", "conv1", "conv2", "def", "else", "identity", "if", "in_channels", "int", "is", "kernel_size", "nn", "out_channels", "padding", "self", "stage_block_in_channels", "stage_block_out_channels", "stage_block_stride", "stage_idx", "stride", "super"], "efficientloftr/modeling_efficientloftr.py:EfficientLoFTRRepVGGBlock.forward": ["GradientCheckpointingLayer", "ModelRepVGGBlock", "None", "Tensor", "activation", "class", "conv1", "conv2", "def", "else", "forward", "hidden_states", "identity", "identity_out", "if", "is", "not", "return", "self", "torch"], "efficientloftr/modeling_efficientloftr.py:EfficientLoFTRRepVGGStage.__init__": ["ModelConfig", "ModelRepVGGBlock", "ModelRepVGGStage", "Module", "ModuleList", "__init__", "append", "block_idx", "blocks", "class", "config", "def", "for", "in", "int", "nn", "range", "self", "stage_idx", "stage_num_blocks", "super"], "efficientloftr/modeling_efficientloftr.py:EfficientLoFTRRepVGGStage.forward": ["ModelRepVGGStage", "Module", "Tensor", "block", "blocks", "class", "def", "for", "forward", "hidden_states", "in", "nn", "return", "self", "torch"], "efficientloftr/modeling_efficientloftr.py:EfficientLoFTRepVGG.__init__": ["ModelConfig", "ModelRepVGGStage", "ModelepVGG", "Module", "ModuleList", "__init__", "append", "class", "config", "def", "for", "in", "len", "nn", "range", "self", "stage", "stage_idx", "stage_stride", "stages", "super"], "efficientloftr/modeling_efficientloftr.py:EfficientLoFTRepVGG.forward": ["ModelepVGG", "Module", "Tensor", "append", "class", "def", "for", "forward", "hidden_states", "in", "list", "nn", "outputs", "return", "self", "stage", "stages", "torch"], "efficientloftr/modeling_efficientloftr.py:EfficientLoFTRAggregationLayer.__init__": ["Conv2d", "False", "LayerNorm", "MaxPool2d", "ModelAggregationLayer", "ModelConfig", "Module", "__init__", "bias", "class", "config", "def", "groups", "hidden_size", "kernel_size", "kv_aggregation", "kv_aggregation_kernel_size", "kv_aggregation_stride", "nn", "norm", "padding", "q_aggregation", "q_aggregation_kernel_size", "q_aggregation_stride", "self", "stride", "super", "torch"], "efficientloftr/modeling_efficientloftr.py:EfficientLoFTRAggregationLayer.forward": ["ModelAggregationLayer", "Module", "None", "Tensor", "class", "def", "else", "encoder_hidden_states", "forward", "hidden_states", "if", "is", "is_cross_attention", "kv_aggregation", "kv_states", "nn", "norm", "not", "permute", "q_aggregation", "query_states", "return", "self", "torch", "tuple"], "efficientloftr/modeling_efficientloftr.py:rotate_half": ["Model_half", "def", "dim", "flatten", "return", "rot_x", "stack", "torch", "x", "x1", "x2"], "efficientloftr/modeling_efficientloftr.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cos", "def", "dtype", "float", "k", "k_embed", "q", "q_embed", "return", "rotate_half", "sin", "to", "unsqueeze", "unsqueeze_dim"], "efficientloftr/modeling_efficientloftr.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "efficientloftr/modeling_efficientloftr.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value", "value_states"], "efficientloftr/modeling_efficientloftr.py:EfficientLoFTRAttention.__init__": ["False", "Linear", "ModelAttention", "ModelConfig", "Module", "__init__", "attention_bias", "attention_dropout", "bias", "class", "config", "def", "getattr", "head_dim", "hidden_size", "int", "is_causal", "k_proj", "layer_idx", "nn", "num_attention_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "q_proj", "scaling", "self", "super", "v_proj"], "efficientloftr/modeling_efficientloftr.py:EfficientLoFTRAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "ModelAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_attn_implementation", "apply_rotary_pos_emb", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "batch_size", "class", "config", "contiguous", "cos", "current_states", "def", "dim", "dropout", "eager", "eager_attention_forward", "else", "encoder_hidden_states", "forward", "head_dim", "hidden_states", "if", "input_shape", "is", "k_proj", "key_states", "kwargs", "nn", "not", "o_proj", "position_embeddings", "q_proj", "query_states", "reshape", "return", "scaling", "self", "seq_len", "shape", "sin", "torch", "training", "transpose", "tuple", "unsqueeze_dim", "v_proj", "value_states", "view"], "efficientloftr/modeling_efficientloftr.py:EfficientLoFTRMLP.__init__": ["ACT2FN", "False", "LayerNorm", "Linear", "ModelConfig", "ModelMLP", "Module", "__init__", "activation", "bias", "class", "config", "def", "fc1", "fc2", "hidden_size", "intermediate_size", "layer_norm", "mlp_activation_function", "nn", "self", "super"], "efficientloftr/modeling_efficientloftr.py:EfficientLoFTRMLP.forward": ["ModelMLP", "Module", "Tensor", "activation", "class", "def", "fc1", "fc2", "forward", "hidden_states", "layer_norm", "nn", "return", "self", "torch"], "efficientloftr/modeling_efficientloftr.py:EfficientLoFTRAggregatedAttention.__init__": ["ModelAggregatedAttention", "ModelAggregationLayer", "ModelAttention", "ModelConfig", "ModelMLP", "Module", "__init__", "aggregation", "attention", "class", "config", "def", "int", "layer_idx", "mlp", "nn", "q_aggregation_kernel_size", "self", "super"], "efficientloftr/modeling_efficientloftr.py:EfficientLoFTRAggregatedAttention.forward": ["False", "ModelAggregatedAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "aggregated_encoder_hidden_states", "aggregated_h", "aggregated_hidden_states", "aggregated_w", "aggregation", "align_corners", "attention", "attn_output", "batch_size", "bilinear", "cat", "class", "def", "dim", "embed_dim", "encoder_hidden_states", "forward", "functional", "hidden_states", "intermediate_states", "interpolate", "kwargs", "mlp", "mode", "nn", "output_states", "permute", "position_embeddings", "q_aggregation_kernel_size", "reshape", "return", "scale_factor", "self", "shape", "torch", "tuple"], "efficientloftr/modeling_efficientloftr.py:EfficientLoFTRLocalFeatureTransformerLayer.__init__": ["GradientCheckpointingLayer", "ModelAggregatedAttention", "ModelConfig", "ModelLocalFeatureTransformerLayer", "__init__", "class", "config", "cross_attention", "def", "int", "layer_idx", "self", "self_attention", "super"], "efficientloftr/modeling_efficientloftr.py:EfficientLoFTRLocalFeatureTransformerLayer.forward": ["GradientCheckpointingLayer", "ModelLocalFeatureTransformerLayer", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "batch_size", "class", "cross_attention", "def", "dim", "embed_dim", "features_0", "features_1", "forward", "height", "hidden_states", "kwargs", "position_embeddings", "reshape", "return", "self", "self_attention", "shape", "stack", "torch", "tuple", "width"], "efficientloftr/modeling_efficientloftr.py:EfficientLoFTRLocalFeatureTransformer.__init__": ["ModelConfig", "ModelLocalFeatureTransformer", "ModelLocalFeatureTransformerLayer", "Module", "ModuleList", "__init__", "class", "config", "def", "for", "i", "in", "layer_idx", "layers", "nn", "num_attention_layers", "range", "self", "super"], "efficientloftr/modeling_efficientloftr.py:EfficientLoFTRLocalFeatureTransformer.forward": ["ModelLocalFeatureTransformer", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "class", "def", "for", "forward", "hidden_states", "in", "kwargs", "layer", "layers", "nn", "position_embeddings", "return", "self", "torch", "tuple"], "efficientloftr/modeling_efficientloftr.py:EfficientLoFTROutConvBlock.__init__": ["ACT2CLS", "BatchNorm2d", "Conv2d", "False", "ModelConfig", "ModelOutConvBlock", "Module", "__init__", "activation", "batch_norm", "bias", "class", "config", "def", "hidden_size", "int", "intermediate_size", "kernel_size", "mlp_activation_function", "nn", "out_conv1", "out_conv2", "out_conv3", "padding", "self", "stride", "super"], "efficientloftr/modeling_efficientloftr.py:EfficientLoFTROutConvBlock.forward": ["False", "ModelOutConvBlock", "Module", "Tensor", "activation", "align_corners", "batch_norm", "bilinear", "class", "def", "forward", "functional", "hidden_states", "interpolate", "mode", "nn", "out_conv1", "out_conv2", "out_conv3", "residual_states", "return", "scale_factor", "self", "torch"], "efficientloftr/modeling_efficientloftr.py:EfficientLoFTRFineFusionLayer.__init__": ["Conv2d", "False", "ModelConfig", "ModelFineFusionLayer", "ModelOutConvBlock", "Module", "ModuleList", "__init__", "append", "bias", "class", "config", "def", "fine_fusion_dims", "fine_kernel_size", "for", "i", "in", "kernel_size", "len", "nn", "out_conv", "out_conv_layers", "padding", "range", "self", "stride", "super"], "efficientloftr/modeling_efficientloftr.py:EfficientLoFTRFineFusionLayer.forward_pyramid": ["False", "ModelFineFusionLayer", "Module", "Tensor", "align_corners", "bilinear", "class", "def", "enumerate", "for", "forward_pyramid", "functional", "hidden_states", "i", "in", "interpolate", "layer", "list", "mode", "nn", "out_conv", "out_conv_layers", "residual_states", "return", "scale_factor", "self", "torch"], "efficientloftr/modeling_efficientloftr.py:EfficientLoFTRFineFusionLayer.forward": ["ModelFineFusionLayer", "Module", "Tensor", "_", "batch_size", "class", "coarse_features", "coarse_height", "coarse_width", "def", "embed_dim", "fine_embed_dim", "fine_features", "fine_features_0", "fine_features_1", "fine_height", "fine_kernel_size", "fine_width", "forward", "forward_pyramid", "functional", "int", "kernel_size", "list", "nn", "padding", "permute", "reshape", "residual_features", "return", "reversed", "self", "seq_len", "shape", "stride", "torch", "tuple", "unfold"], "efficientloftr/modeling_efficientloftr.py:EfficientLoFTRPreTrainedModel._init_weights": ["BatchNorm2d", "Conv1d", "Conv2d", "LayerNorm", "Linear", "ModelPreTrainedModel", "ModelRotaryEmbedding", "Module", "None", "PreTrainedModel", "ROPE_INIT_FUNCTIONS", "_", "_init_weights", "bias", "buffer_value", "class", "compute_default_rope_parameters", "config", "copy_", "def", "default", "elif", "else", "getattr", "if", "init", "initializer_range", "inv_freq", "is", "isinstance", "mean", "module", "nn", "normal_", "not", "num_batches_tracked", "ones_", "original_inv_freq", "rope_fn", "rope_type", "running_mean", "running_var", "self", "std", "weight", "zeros_"], "efficientloftr/modeling_efficientloftr.py:EfficientLoFTRPreTrainedModel.extract_one_channel_pixel_values": ["FloatTensor", "ModelPreTrainedModel", "None", "PreTrainedModel", "class", "def", "extract_one_channel_pixel_values", "pixel_values", "return", "self", "torch"], "efficientloftr/modeling_efficientloftr.py:EfficientLoFTRModel.__init__": ["ModelConfig", "ModelLocalFeatureTransformer", "ModelModel", "ModelPreTrainedModel", "ModelRotaryEmbedding", "ModelepVGG", "__init__", "backbone", "class", "config", "def", "local_feature_transformer", "post_init", "rotary_emb", "self", "super"], "efficientloftr/modeling_efficientloftr.py:EfficientLoFTRModel.forward": ["BackboneOutput", "FloatTensor", "Input", "LongTensor", "Model", "ModelModel", "ModelPreTrainedModel", "None", "TransformersKwargs", "Unpack", "ValueError", "_", "a", "backbone", "batch_size", "be", "channels", "class", "coarse_embed_dim", "coarse_features", "coarse_height", "coarse_width", "cos", "def", "expand", "extract_one_channel_pixel_values", "feature_maps", "features", "forward", "height", "if", "is", "kwargs", "labels", "local_feature_transformer", "must", "ndim", "no", "not", "num_channels", "of", "or", "pixel_values", "position_embeddings", "provided", "r", "raise", "reshape", "residual_features", "return", "rotary_emb", "self", "shape", "should", "sin", "size", "tensor", "torch", "trainable", "tuple", "width"], "efficientloftr/modeling_efficientloftr.py:mask_border": ["Model_border", "Tensor", "bool", "border_margin", "def", "float", "if", "int", "return", "tensor", "torch", "value"], "efficientloftr/modeling_efficientloftr.py:create_meshgrid": ["False", "Model_meshgrid", "None", "Tensor", "bool", "def", "device", "dim", "dtype", "grid", "height", "if", "ij", "indexing", "int", "linspace", "meshgrid", "normalized_coordinates", "permute", "return", "stack", "torch", "unsqueeze", "width", "xs", "ys"], "efficientloftr/modeling_efficientloftr.py:spatial_expectation2d": ["Model_expectation2d", "Tensor", "True", "batch_size", "bool", "cat", "create_meshgrid", "def", "device", "dtype", "embed_dim", "expected_x", "expected_y", "grid", "height", "input", "input_flat", "keepdim", "normalized_coordinates", "output", "pos_x", "pos_y", "r", "reshape", "return", "shape", "sum", "to", "torch", "view", "width"], "efficientloftr/modeling_efficientloftr.py:EfficientLoFTRForKeypointMatching.__init__": ["Model", "ModelConfig", "ModelFineFusionLayer", "ModelForKeypointMatching", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "config", "def", "post_init", "refinement_layer", "self", "super"], "efficientloftr/modeling_efficientloftr.py:EfficientLoFTRForKeypointMatching._get_matches_from_scores": ["False", "ModelForKeypointMatching", "ModelPreTrainedModel", "Tensor", "True", "_get_matches_from_scores", "batch_size", "cat", "class", "coarse_matching_border_removal", "coarse_matching_threshold", "config", "def", "dim", "height0", "height1", "keepdim", "mask", "mask_border", "masked_scores", "matching_indices", "matching_scores", "matching_scores_0", "matching_scores_1", "max", "max_0", "max_1", "max_indices_0", "max_indices_1", "reshape", "return", "scores", "self", "shape", "stack", "torch", "tuple", "values", "view", "where", "width0", "width1"], "efficientloftr/modeling_efficientloftr.py:EfficientLoFTRForKeypointMatching._coarse_matching": ["ModelForKeypointMatching", "ModelPreTrainedModel", "Tensor", "_", "_coarse_matching", "_get_matches_from_scores", "batch_size", "class", "coarse_features", "coarse_features_0", "coarse_features_1", "coarse_matching_skip_softmax", "coarse_matching_temperature", "coarse_scale", "confidence", "config", "def", "dim", "else", "embed_dim", "float", "functional", "height", "if", "keypoints", "matched_indices", "matching_scores", "nn", "permute", "reshape", "return", "self", "shape", "similarity", "softmax", "stack", "torch", "transpose", "tuple", "view", "width"], "efficientloftr/modeling_efficientloftr.py:EfficientLoFTRForKeypointMatching._get_first_stage_fine_matching": ["False", "ModelForKeypointMatching", "ModelPreTrainedModel", "None", "Tensor", "_", "_get_first_stage_fine_matching", "batch_size", "class", "coarse_matched_keypoints", "create_meshgrid", "def", "delta_0", "delta_1", "device", "dim", "dtype", "expand", "fine_confidence", "fine_kernel_size", "fine_matches", "fine_matches_0", "fine_matches_1", "fine_scale", "fine_window_size", "float", "gather", "grid", "indices", "indices_0", "indices_1", "int", "max", "normalized_coordinates", "num_keypoints", "reshape", "return", "self", "shape", "squeeze", "stack", "torch", "torch_int", "tuple", "unsqueeze", "values"], "efficientloftr/modeling_efficientloftr.py:EfficientLoFTRForKeypointMatching._get_second_stage_fine_matching": ["ModelForKeypointMatching", "ModelPreTrainedModel", "None", "Tensor", "True", "_", "_get_second_stage_fine_matching", "arange", "batch_indices", "batch_size", "class", "config", "create_meshgrid", "def", "delta", "device", "dim", "fine_confidence", "fine_coordinates_normalized", "fine_kernel_size", "fine_matches", "fine_matches_0", "fine_matches_1", "fine_matching_regress_temperature", "fine_scale", "fine_window_size", "float", "functional", "heatmap", "indices", "indices_0", "indices_1", "indices_1_i", "indices_1_j", "int", "long", "matches_indices", "nn", "normalized_coordinates", "num_keypoints", "reshape", "return", "self", "shape", "softmax", "spatial_expectation2d", "stack", "to", "torch", "torch_int"], "efficientloftr/modeling_efficientloftr.py:EfficientLoFTRForKeypointMatching._fine_matching": ["ModelForKeypointMatching", "ModelPreTrainedModel", "Tensor", "_fine_matching", "_get_first_stage_fine_matching", "_get_second_stage_fine_matching", "batch_size", "class", "coarse_matched_keypoints", "config", "def", "fine_confidence", "fine_coordinates", "fine_embed_dim", "fine_features_0", "fine_features_1", "fine_indices", "fine_kernel_size", "fine_matches", "fine_matching_slice_dim", "fine_scale", "fine_window_size", "first_stage_fine_confidence", "float", "functional", "nn", "num_keypoints", "reshape", "return", "second_stage_fine_confidence", "self", "shape", "softmax", "split", "split_fine_features_0", "split_fine_features_1", "torch", "torch_int", "transpose"], "efficientloftr/modeling_efficientloftr.py:EfficientLoFTRForKeypointMatching.forward": ["BackboneOutput", "FloatTensor", "LongTensor", "Model", "ModelForKeypointMatching", "ModelKeypointMatchingOutput", "ModelPreTrainedModel", "None", "SuperGlue", "TransformersKwargs", "Unpack", "ValueError", "_", "_coarse_matching", "_fine_matching", "arange", "attentions", "batch_indices", "batch_size", "be", "channels", "class", "coarse_embed_dim", "coarse_features", "coarse_height", "coarse_keypoints", "coarse_matched_indices", "coarse_matching_scores", "coarse_scale", "coarse_width", "config", "def", "feature_maps", "features", "fine_features_0", "fine_features_1", "fine_height", "fine_scale", "forward", "height", "hidden_size", "hidden_states", "if", "is", "keypoints", "kwargs", "labels", "loss", "matches", "matching_keypoints", "matching_scores", "model_outputs", "no", "not", "num_keypoints", "pixel_values", "provided", "r", "raise", "refinement_layer", "residual_features", "return", "self", "shape", "should", "torch", "torch_int", "trainable", "width"], "efficientnet/modeling_efficientnet.py:round_filters": ["ModelConfig", "Model_filters", "config", "def", "depth_divisor", "divisor", "if", "int", "max", "new_dim", "num_channels", "r", "return", "width_coefficient"], "efficientnet/modeling_efficientnet.py:correct_pad": ["Model", "Model_pad", "True", "adjust", "bool", "def", "else", "if", "int", "isinstance", "kernel_size", "r", "return", "tuple"], "efficientnet/modeling_efficientnet.py:EfficientNetEmbeddings.__init__": ["ACT2FN", "BatchNorm2d", "Conv2d", "False", "ModelConfig", "ModelEmbeddings", "Module", "ZeroPad2d", "__init__", "activation", "batch_norm_eps", "batch_norm_momentum", "batchnorm", "bias", "class", "config", "convolution", "def", "eps", "hidden_act", "kernel_size", "momentum", "nn", "num_channels", "out_dim", "padding", "round_filters", "self", "stride", "super", "valid"], "efficientnet/modeling_efficientnet.py:EfficientNetEmbeddings.forward": ["ModelEmbeddings", "Module", "Tensor", "activation", "batchnorm", "class", "convolution", "def", "features", "forward", "nn", "padding", "pixel_values", "return", "self", "torch"], "efficientnet/modeling_efficientnet.py:EfficientNetDepthwiseConv2d.__init__": ["Conv2d", "ModelDepthwiseConv2d", "True", "__init__", "bias", "class", "def", "depth_multiplier", "dilation", "groups", "in_channels", "kernel_size", "nn", "out_channels", "padding", "padding_mode", "self", "stride", "super", "zeros"], "efficientnet/modeling_efficientnet.py:EfficientNetExpansionLayer.__init__": ["ACT2FN", "BatchNorm2d", "Conv2d", "False", "ModelConfig", "ModelExpansionLayer", "Module", "__init__", "batch_norm_eps", "bias", "class", "config", "def", "eps", "expand_act", "expand_bn", "expand_conv", "hidden_act", "in_channels", "in_dim", "int", "kernel_size", "nn", "num_features", "out_channels", "out_dim", "padding", "same", "self", "stride", "super"], "efficientnet/modeling_efficientnet.py:EfficientNetExpansionLayer.forward": ["FloatTensor", "ModelExpansionLayer", "Module", "Tensor", "class", "def", "expand_act", "expand_bn", "expand_conv", "forward", "hidden_states", "nn", "return", "self", "torch"], "efficientnet/modeling_efficientnet.py:EfficientNetDepthwiseLayer.__init__": ["ACT2FN", "BatchNorm2d", "False", "ModelConfig", "ModelDepthwiseConv2d", "ModelDepthwiseLayer", "Module", "ZeroPad2d", "__init__", "adjust", "adjust_padding", "batch_norm_eps", "batch_norm_momentum", "bias", "bool", "class", "config", "conv_pad", "correct_pad", "def", "depthwise_act", "depthwise_conv", "depthwise_conv_pad", "depthwise_norm", "else", "eps", "hidden_act", "if", "in_dim", "int", "kernel_size", "momentum", "nn", "num_features", "padding", "same", "self", "stride", "super", "valid"], "efficientnet/modeling_efficientnet.py:EfficientNetDepthwiseLayer.forward": ["FloatTensor", "ModelDepthwiseLayer", "Module", "Tensor", "class", "def", "depthwise_act", "depthwise_conv", "depthwise_conv_pad", "depthwise_norm", "forward", "hidden_states", "if", "nn", "return", "self", "stride", "torch"], "efficientnet/modeling_efficientnet.py:EfficientNetSqueezeExciteLayer.__init__": ["ACT2FN", "AdaptiveAvgPool2d", "Conv2d", "False", "ModelConfig", "ModelSqueezeExciteLayer", "Module", "Sigmoid", "__init__", "act_expand", "act_reduce", "bool", "class", "config", "def", "dim", "dim_se", "else", "expand", "expand_dim", "hidden_act", "if", "in_channels", "in_dim", "int", "kernel_size", "max", "nn", "out_channels", "output_size", "padding", "reduce", "same", "self", "squeeze", "squeeze_expansion_ratio", "super"], "efficientnet/modeling_efficientnet.py:EfficientNetSqueezeExciteLayer.forward": ["FloatTensor", "ModelSqueezeExciteLayer", "Module", "Tensor", "act_expand", "act_reduce", "class", "def", "expand", "forward", "hidden_states", "inputs", "mul", "nn", "reduce", "return", "self", "squeeze", "torch"], "efficientnet/modeling_efficientnet.py:EfficientNetFinalBlockLayer.__init__": ["BatchNorm2d", "Conv2d", "Dropout", "False", "ModelConfig", "ModelFinalBlockLayer", "Module", "__init__", "and", "apply_dropout", "batch_norm_eps", "batch_norm_momentum", "bias", "bool", "class", "config", "def", "drop_rate", "dropout", "eps", "float", "id_skip", "in_channels", "in_dim", "int", "kernel_size", "momentum", "nn", "not", "num_features", "out_channels", "out_dim", "p", "padding", "project_bn", "project_conv", "same", "self", "stride", "super"], "efficientnet/modeling_efficientnet.py:EfficientNetFinalBlockLayer.forward": ["FloatTensor", "ModelFinalBlockLayer", "Module", "Tensor", "apply_dropout", "class", "def", "dropout", "embeddings", "forward", "hidden_states", "if", "nn", "project_bn", "project_conv", "return", "self", "torch"], "efficientnet/modeling_efficientnet.py:EfficientNetBlock.__init__": ["ModelBlock", "ModelConfig", "ModelDepthwiseLayer", "ModelExpansionLayer", "ModelFinalBlockLayer", "ModelSqueezeExciteLayer", "Module", "__init__", "adjust_padding", "bool", "class", "config", "def", "depthwise_conv", "drop_rate", "else", "expand", "expand_dim", "expand_in_dim", "expand_ratio", "expansion", "float", "id_skip", "if", "in_dim", "int", "kernel_size", "nn", "out_dim", "projection", "self", "squeeze_excite", "stride", "super"], "efficientnet/modeling_efficientnet.py:EfficientNetBlock.forward": ["FloatTensor", "ModelBlock", "Module", "Tensor", "class", "def", "depthwise_conv", "embeddings", "expand_ratio", "expansion", "forward", "hidden_states", "if", "nn", "projection", "return", "self", "squeeze_excite", "torch"], "efficientnet/modeling_efficientnet.py:EfficientNetEncoder.__init__": ["ACT2FN", "BatchNorm2d", "Conv2d", "False", "ModelBlock", "ModelConfig", "ModelEncoder", "Module", "ModuleList", "__init__", "adjust_padding", "append", "batch_norm_eps", "batch_norm_momentum", "bias", "block", "blocks", "ceil", "class", "config", "curr_block_num", "def", "depth_coModel", "depthwise_padding", "drop_connect_rate", "drop_rate", "else", "eps", "expand_ratio", "expand_ratios", "for", "hidden_act", "hidden_dim", "i", "id_skip", "if", "in", "in_channels", "in_dim", "int", "j", "kernel_size", "kernel_sizes", "len", "math", "momentum", "n", "nn", "not", "num_base_blocks", "num_block_repeats", "num_blocks", "num_features", "out_channels", "out_dim", "padding", "range", "repeats", "return", "round_filters", "round_repeats", "same", "self", "stride", "strides", "sum", "super", "top_activation", "top_bn", "top_conv"], "efficientnet/modeling_efficientnet.py:EfficientNetEncoder.forward": ["BaseModelOutputWithNoAttention", "False", "FloatTensor", "ModelEncoder", "Module", "None", "True", "all_hidden_states", "block", "blocks", "bool", "class", "def", "else", "for", "forward", "hidden_states", "if", "in", "is", "last_hidden_state", "nn", "not", "output_hidden_states", "return", "return_dict", "self", "top_activation", "top_bn", "top_conv", "torch", "tuple", "v"], "efficientnet/modeling_efficientnet.py:EfficientNetPreTrainedModel._init_weights": ["BatchNorm2d", "Conv2d", "Linear", "ModelPreTrainedModel", "Module", "None", "PreTrainedModel", "_init_weights", "bias", "class", "config", "def", "getattr", "if", "init", "initializer_range", "is", "isinstance", "mean", "module", "nn", "normal_", "not", "num_batches_tracked", "ones_", "running_mean", "running_var", "self", "std", "weight", "zeros_"], "efficientnet/modeling_efficientnet.py:EfficientNetModel.__init__": ["AvgPool2d", "MaxPool2d", "ModelConfig", "ModelEmbeddings", "ModelEncoder", "ModelModel", "ModelPreTrainedModel", "True", "ValueError", "__init__", "be", "ceil_mode", "class", "config", "def", "elif", "else", "embeddings", "encoder", "f", "got", "hidden_dim", "if", "max", "mean", "must", "nn", "of", "one", "pooler", "pooling", "pooling_type", "post_init", "raise", "self", "super"], "efficientnet/modeling_efficientnet.py:EfficientNetModel.forward": ["BaseModelOutputWithPoolingAndNoAttention", "FloatTensor", "ModelModel", "ModelPreTrainedModel", "None", "ValueError", "You", "bool", "class", "config", "def", "else", "embedding_output", "embeddings", "encoder", "encoder_outputs", "forward", "have", "hidden_states", "if", "is", "kwargs", "last_hidden_state", "not", "output_hidden_states", "pixel_values", "pooled_output", "pooler", "pooler_output", "raise", "reshape", "return", "return_dict", "self", "shape", "specify", "to", "torch", "tuple", "use_return_dict"], "efficientnet/modeling_efficientnet.py:EfficientNetForImageClassification.__init__": ["Dropout", "Identity", "Linear", "Model", "ModelForImageClassification", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "classifier", "config", "def", "dropout", "dropout_rate", "else", "hidden_dim", "if", "nn", "num_labels", "p", "post_init", "self", "super"], "efficientnet/modeling_efficientnet.py:EfficientNetForImageClassification.forward": ["FloatTensor", "ImageClassifierOutputWithNoAttention", "LongTensor", "Model", "ModelForImageClassification", "ModelPreTrainedModel", "None", "bool", "class", "classifier", "config", "def", "dropout", "else", "forward", "hidden_states", "if", "is", "kwargs", "labels", "logits", "loss", "loss_function", "not", "output", "output_hidden_states", "outputs", "pixel_values", "pooled_output", "pooler_output", "r", "return", "return_dict", "self", "torch", "tuple", "use_return_dict"], "electra/modeling_electra.py:ElectraEmbeddings.__init__": ["Dropout", "Embedding", "False", "LayerNorm", "ModelEmbeddings", "Module", "__init__", "arange", "class", "config", "def", "dropout", "dtype", "embedding_size", "eps", "expand", "hidden_dropout_prob", "layer_norm_eps", "long", "max_position_embeddings", "nn", "pad_token_id", "padding_idx", "persistent", "position_embeddings", "position_ids", "register_buffer", "self", "size", "super", "token_type_embeddings", "token_type_ids", "torch", "type_vocab_size", "vocab_size", "word_embeddings", "zeros"], "electra/modeling_electra.py:ElectraEmbeddings.forward": ["FloatTensor", "LayerNorm", "LongTensor", "ModelEmbeddings", "Module", "None", "Tensor", "batch_size", "buffered_token_type_ids", "class", "def", "device", "dim", "dropout", "dtype", "else", "embeddings", "expand", "forward", "gather", "hasattr", "if", "index", "input_ids", "input_shape", "inputs_embeds", "int", "is", "long", "nn", "not", "past_key_values_length", "position_embeddings", "position_ids", "return", "self", "seq_length", "shape", "size", "token_type_embeddings", "token_type_ids", "torch", "word_embeddings", "zeros"], "electra/modeling_electra.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "contiguous", "def", "dim", "dropout", "float", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "shape", "size", "softmax", "torch", "training", "transpose", "value"], "electra/modeling_electra.py:ElectraSelfAttention.__init__": ["Dropout", "False", "Linear", "ModelSelfAttention", "Module", "None", "The", "ValueError", "__init__", "a", "all_head_size", "and", "attention", "attention_head_size", "attention_probs_dropout_prob", "class", "config", "def", "dropout", "embedding_size", "f", "hasattr", "heads", "hidden", "hidden_size", "if", "int", "is", "is_causal", "is_decoder", "key", "layer_idx", "multiple", "nn", "not", "num_attention_heads", "number", "of", "query", "raise", "scaling", "self", "size", "super", "the", "value"], "electra/modeling_electra.py:ElectraSelfAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "EncoderDecoderCache", "FloatTensor", "ModelSelfAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_attn_implementation", "attention_head_size", "attention_interface", "attention_mask", "attn_output", "attn_weights", "cache_position", "class", "config", "contiguous", "current_past_key_values", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "hidden_shape", "hidden_states", "if", "input_shape", "is", "isinstance", "key", "key_layer", "kwargs", "layer_idx", "nn", "not", "p", "past_key_values", "query", "query_layer", "reshape", "return", "scaling", "self", "self_attention_cache", "shape", "torch", "training", "transpose", "tuple", "update", "value", "value_layer", "view"], "electra/modeling_electra.py:ElectraCrossAttention.__init__": ["Dropout", "False", "Linear", "ModelCrossAttention", "Module", "None", "The", "ValueError", "__init__", "a", "all_head_size", "and", "attention", "attention_head_size", "attention_probs_dropout_prob", "class", "config", "def", "dropout", "embedding_size", "f", "hasattr", "heads", "hidden", "hidden_size", "if", "int", "is", "is_causal", "key", "layer_idx", "multiple", "nn", "not", "num_attention_heads", "number", "of", "query", "raise", "scaling", "self", "size", "super", "the", "value"], "electra/modeling_electra.py:ElectraCrossAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "EncoderDecoderCache", "False", "FloatTensor", "ModelCrossAttention", "Module", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "_attn_implementation", "and", "attention_head_size", "attention_interface", "attention_mask", "attn_output", "attn_weights", "bsz", "class", "config", "contiguous", "cross_attention_cache", "def", "dropout", "eager", "eager_attention_forward", "else", "encoder_hidden_states", "forward", "get", "hidden_states", "if", "is", "is_updated", "key", "key_layer", "keys", "kv_input_shape", "kwargs", "layer_idx", "layers", "nn", "not", "p", "past_key_values", "q_input_shape", "query", "query_layer", "reshape", "return", "scaling", "self", "shape", "src_len", "tgt_len", "torch", "training", "transpose", "tuple", "update", "value", "value_layer", "values", "view"], "electra/modeling_electra.py:ElectraSelfOutput.__init__": ["Dropout", "LayerNorm", "Linear", "ModelSelfOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "eps", "hidden_dropout_prob", "hidden_size", "layer_norm_eps", "nn", "self", "super"], "electra/modeling_electra.py:ElectraSelfOutput.forward": ["LayerNorm", "ModelSelfOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "electra/modeling_electra.py:ElectraAttention.__init__": ["False", "ModelAttention", "ModelCrossAttention", "ModelSelfAttention", "ModelSelfOutput", "Module", "None", "__init__", "attention_class", "class", "config", "def", "else", "if", "is_causal", "is_cross_attention", "layer_idx", "nn", "output", "self", "super"], "electra/modeling_electra.py:ElectraAttention.forward": ["Cache", "FloatTensor", "ModelAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attention_output", "attn_weights", "cache_position", "class", "def", "else", "encoder_attention_mask", "encoder_hidden_states", "forward", "hidden_states", "if", "is_cross_attention", "kwargs", "nn", "not", "output", "past_key_values", "return", "self", "torch", "tuple"], "electra/modeling_electra.py:ElectraIntermediate.__init__": ["ACT2FN", "Linear", "ModelIntermediate", "Module", "__init__", "class", "config", "def", "dense", "else", "hidden_act", "hidden_size", "if", "intermediate_act_fn", "intermediate_size", "isinstance", "nn", "self", "str", "super"], "electra/modeling_electra.py:ElectraIntermediate.forward": ["ModelIntermediate", "Module", "Tensor", "class", "def", "dense", "forward", "hidden_states", "intermediate_act_fn", "nn", "return", "self", "torch"], "electra/modeling_electra.py:ElectraOutput.__init__": ["Dropout", "LayerNorm", "Linear", "ModelOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "eps", "hidden_dropout_prob", "hidden_size", "intermediate_size", "layer_norm_eps", "nn", "self", "super"], "electra/modeling_electra.py:ElectraOutput.forward": ["LayerNorm", "ModelOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "electra/modeling_electra.py:ElectraLayer.__init__": ["False", "GradientCheckpointingLayer", "ModelAttention", "ModelIntermediate", "ModelLayer", "ModelOutput", "None", "True", "ValueError", "__init__", "a", "add_cross_attention", "added", "as", "attention", "be", "chunk_size_feed_forward", "class", "config", "cross", "crossattention", "decoder", "def", "f", "if", "intermediate", "is", "is_causal", "is_cross_attention", "is_decoder", "layer_idx", "model", "not", "output", "raise", "self", "seq_len_dim", "should", "super", "used"], "electra/modeling_electra.py:ElectraLayer.forward": ["Cache", "FloatTensor", "GradientCheckpointingLayer", "If", "ModelLayer", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "ValueError", "_", "add_cross_attention", "and", "apply_chunking_to_forward", "are", "attention", "attention_mask", "attention_output", "be", "by", "cache_position", "chunk_size_feed_forward", "class", "config", "cross", "cross_attention_output", "crossattention", "def", "encoder_attention_mask", "encoder_hidden_states", "f", "feed_forward_chunk", "forward", "has", "hasattr", "hidden_states", "if", "instantiated", "is", "is_decoder", "kwargs", "layer_output", "layers", "not", "passed", "past_key_values", "raise", "return", "self", "self_attention_output", "seq_len_dim", "setting", "to", "torch", "tuple", "with"], "electra/modeling_electra.py:ElectraLayer.feed_forward_chunk": ["GradientCheckpointingLayer", "ModelLayer", "attention_output", "class", "def", "feed_forward_chunk", "intermediate", "intermediate_output", "layer_output", "output", "return", "self"], "electra/modeling_electra.py:ElectraEncoder.__init__": ["ModelEncoder", "ModelLayer", "Module", "ModuleList", "__init__", "class", "config", "def", "for", "i", "in", "layer", "layer_idx", "nn", "num_hidden_layers", "range", "self", "super"], "electra/modeling_electra.py:ElectraEncoder.forward": ["BaseModelOutputWithPastAndCrossAttentions", "Cache", "FloatTensor", "ModelEncoder", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "bool", "cache_position", "class", "def", "else", "encoder_attention_mask", "encoder_hidden_states", "enumerate", "for", "forward", "hidden_states", "i", "if", "in", "kwargs", "last_hidden_state", "layer", "layer_module", "nn", "past_key_values", "return", "self", "torch", "tuple", "use_cache"], "electra/modeling_electra.py:ElectraDiscriminatorPredictions.__init__": ["Linear", "ModelDiscriminatorPredictions", "Module", "__init__", "activation", "class", "config", "def", "dense", "dense_prediction", "get_activation", "hidden_act", "hidden_size", "nn", "self", "super"], "electra/modeling_electra.py:ElectraDiscriminatorPredictions.forward": ["ModelDiscriminatorPredictions", "Module", "activation", "class", "def", "dense", "dense_prediction", "discriminator_hidden_states", "forward", "hidden_states", "logits", "nn", "return", "self", "squeeze"], "electra/modeling_electra.py:ElectraGeneratorPredictions.__init__": ["LayerNorm", "Linear", "ModelGeneratorPredictions", "Module", "__init__", "activation", "class", "config", "def", "dense", "embedding_size", "eps", "gelu", "get_activation", "hidden_size", "layer_norm_eps", "nn", "self", "super"], "electra/modeling_electra.py:ElectraGeneratorPredictions.forward": ["LayerNorm", "ModelGeneratorPredictions", "Module", "activation", "class", "def", "dense", "forward", "generator_hidden_states", "hidden_states", "nn", "return", "self"], "electra/modeling_electra.py:ElectraPreTrainedModel._init_weights": ["ModelEmbeddings", "ModelPreTrainedModel", "PreTrainedModel", "_init_weights", "arange", "class", "copy_", "def", "expand", "if", "init", "isinstance", "module", "position_ids", "self", "shape", "super", "token_type_ids", "torch", "zeros_"], "electra/modeling_electra.py:ElectraModel.__init__": ["False", "Linear", "ModelEmbeddings", "ModelEncoder", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "config", "def", "embedding_size", "embeddings", "embeddings_project", "encoder", "gradient_checkpointing", "hidden_size", "if", "nn", "post_init", "self", "super"], "electra/modeling_electra.py:ElectraModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "return", "self", "word_embeddings"], "electra/modeling_electra.py:ElectraModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "self", "set_input_embeddings", "value", "word_embeddings"], "electra/modeling_electra.py:ElectraModel.forward": ["BaseModelOutputWithCrossAttentions", "BaseModelOutputWithPastAndCrossAttentions", "DynamicCache", "EncoderDecoderCache", "False", "FloatTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "_create_attention_masks", "and", "arange", "attention_mask", "bool", "cache_position", "class", "config", "def", "device", "else", "embedding_output", "embeddings", "embeddings_project", "encoder", "encoder_attention_mask", "encoder_hidden_states", "encoder_outputs", "exactly", "forward", "get_seq_length", "hasattr", "if", "input_ids", "input_shape", "inputs_embeds", "is", "is_decoder", "is_encoder_decoder", "kwargs", "last_hidden_state", "list", "must", "not", "of", "one", "or", "past_key_values", "past_key_values_length", "position_ids", "raise", "return", "self", "seq_length", "shape", "specify", "token_type_ids", "torch", "tuple", "use_cache"], "electra/modeling_electra.py:ElectraModel._create_attention_masks": ["ModelModel", "ModelPreTrainedModel", "None", "_create_attention_masks", "attention_mask", "cache_position", "class", "config", "create_bidirectional_mask", "create_causal_mask", "def", "else", "embedding_output", "encoder_attention_mask", "encoder_hidden_states", "if", "input_embeds", "is", "is_decoder", "not", "past_key_values", "return", "self"], "electra/modeling_electra.py:ElectraClassificationHead.__init__": ["Dropout", "Linear", "ModelClassificationHead", "Module", "None", "__init__", "activation", "class", "classifier_dropout", "config", "def", "dense", "dropout", "else", "gelu", "get_activation", "hidden_dropout_prob", "hidden_size", "if", "is", "nn", "not", "num_labels", "out_proj", "self", "super"], "electra/modeling_electra.py:ElectraClassificationHead.forward": ["ModelClassificationHead", "Module", "activation", "class", "def", "dense", "dropout", "features", "forward", "kwargs", "nn", "out_proj", "return", "self", "x"], "electra/modeling_electra.py:ElectraSequenceSummary.__init__": ["Callable", "Dropout", "Identity", "Linear", "ModelConfig", "ModelSequenceSummary", "Module", "None", "NotImplementedError", "__init__", "activation", "activation_string", "and", "attn", "class", "config", "def", "else", "first_dropout", "get_activation", "getattr", "hasattr", "hidden_size", "if", "last", "last_dropout", "nn", "num_classes", "num_labels", "raise", "self", "summary", "summary_activation", "summary_first_dropout", "summary_last_dropout", "summary_proj_to_labels", "summary_type", "summary_use_proj", "super"], "electra/modeling_electra.py:ElectraSequenceSummary.forward": ["FloatTensor", "LongTensor", "ModelSequenceSummary", "Module", "None", "NotImplementedError", "activation", "attn", "class", "cls_index", "def", "dim", "dtype", "elif", "else", "expand", "first", "first_dropout", "forward", "full_like", "gather", "hidden_states", "if", "is", "last", "last_dropout", "long", "mean", "nn", "output", "raise", "return", "self", "shape", "size", "squeeze", "summary", "summary_type", "torch", "unsqueeze"], "electra/modeling_electra.py:ElectraForSequenceClassification.__init__": ["Model", "ModelClassificationHead", "ModelForSequenceClassification", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "classifier", "config", "def", "num_labels", "post_init", "self", "super"], "electra/modeling_electra.py:ElectraForSequenceClassification.forward": ["BCEWithLogitsLoss", "CrossEntropyLoss", "MSELoss", "Model", "ModelForSequenceClassification", "ModelPreTrainedModel", "None", "SequenceClassifierOutput", "Tensor", "TransformersKwargs", "True", "Unpack", "and", "attention_mask", "attentions", "class", "classifier", "config", "def", "discriminator_hidden_states", "dtype", "elif", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "kwargs", "labels", "logits", "long", "loss", "loss_fct", "multi_label_classification", "not", "num_labels", "or", "position_ids", "problem_type", "r", "regression", "return", "return_dict", "self", "sequence_output", "single_label_classification", "squeeze", "token_type_ids", "torch", "tuple", "view"], "electra/modeling_electra.py:ElectraForPreTraining.__init__": ["Model", "ModelDiscriminatorPredictions", "ModelForPreTraining", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "config", "def", "discriminator_predictions", "post_init", "self", "super"], "electra/modeling_electra.py:ElectraForPreTraining.forward": ["BCEWithLogitsLoss", "Model", "ModelForPreTraining", "ModelForPreTrainingOutput", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "active_labels", "active_logits", "active_loss", "attention_mask", "attentions", "class", "def", "discriminator_hidden_states", "discriminator_predictions", "discriminator_sequence_output", "else", "float", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "nn", "not", "position_ids", "r", "return", "return_dict", "self", "shape", "token_type_ids", "torch", "tuple", "view"], "electra/modeling_electra.py:ElectraForMaskedLM.__init__": ["Linear", "Model", "ModelForMaskedLM", "ModelGeneratorPredictions", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "config", "def", "embedding_size", "generator_lm_head", "generator_predictions", "nn", "post_init", "self", "super", "vocab_size"], "electra/modeling_electra.py:ElectraForMaskedLM.get_output_embeddings": ["ModelForMaskedLM", "ModelPreTrainedModel", "class", "def", "generator_lm_head", "get_output_embeddings", "return", "self"], "electra/modeling_electra.py:ElectraForMaskedLM.set_output_embeddings": ["ModelForMaskedLM", "ModelPreTrainedModel", "class", "def", "generator_lm_head", "self", "set_output_embeddings", "word_embeddings"], "electra/modeling_electra.py:ElectraForMaskedLM.forward": ["CrossEntropyLoss", "MaskedLMOutput", "Model", "ModelForMaskedLM", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "attention_mask", "attentions", "class", "config", "def", "forward", "generator_hidden_states", "generator_lm_head", "generator_predictions", "generator_sequence_output", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "nn", "not", "position_ids", "prediction_scores", "r", "return", "return_dict", "self", "token_type_ids", "torch", "tuple", "view", "vocab_size"], "electra/modeling_electra.py:ElectraForTokenClassification.__init__": ["Dropout", "Linear", "Model", "ModelForTokenClassification", "ModelModel", "ModelPreTrainedModel", "None", "__init__", "class", "classifier", "classifier_dropout", "config", "def", "dropout", "else", "hidden_dropout_prob", "hidden_size", "if", "is", "nn", "not", "num_labels", "post_init", "self", "super"], "electra/modeling_electra.py:ElectraForTokenClassification.forward": ["CrossEntropyLoss", "Model", "ModelForTokenClassification", "ModelPreTrainedModel", "None", "Tensor", "TokenClassifierOutput", "TransformersKwargs", "True", "Unpack", "attention_mask", "attentions", "class", "classifier", "def", "discriminator_hidden_states", "discriminator_sequence_output", "dropout", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "not", "num_labels", "position_ids", "r", "return", "return_dict", "self", "token_type_ids", "torch", "tuple", "view"], "electra/modeling_electra.py:ElectraForQuestionAnswering.__init__": ["Linear", "Model", "ModelForQuestionAnswering", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "config", "def", "hidden_size", "nn", "num_labels", "post_init", "qa_outputs", "self", "super"], "electra/modeling_electra.py:ElectraForQuestionAnswering.forward": ["CrossEntropyLoss", "Model", "ModelForQuestionAnswering", "ModelPreTrainedModel", "None", "QuestionAnsweringModelOutput", "Tensor", "TransformersKwargs", "True", "Unpack", "and", "attention_mask", "attentions", "clamp", "class", "contiguous", "def", "dim", "discriminator_hidden_states", "end_logits", "end_loss", "end_positions", "forward", "hidden_states", "if", "ignore_index", "ignored_index", "input_ids", "inputs_embeds", "is", "kwargs", "len", "logits", "loss", "loss_fct", "not", "position_ids", "qa_outputs", "return", "return_dict", "self", "sequence_output", "size", "split", "squeeze", "start_logits", "start_loss", "start_positions", "token_type_ids", "torch", "total_loss", "tuple"], "electra/modeling_electra.py:ElectraForMultipleChoice.__init__": ["Linear", "Model", "ModelForMultipleChoice", "ModelModel", "ModelPreTrainedModel", "ModelSequenceSummary", "__init__", "class", "classifier", "config", "def", "hidden_size", "nn", "post_init", "self", "sequence_summary", "super"], "electra/modeling_electra.py:ElectraForMultipleChoice.forward": ["CrossEntropyLoss", "Model", "ModelForMultipleChoice", "ModelPreTrainedModel", "MultipleChoiceModelOutput", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "attention_mask", "attentions", "class", "classifier", "def", "discriminator_hidden_states", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "not", "num_choices", "pooled_output", "position_ids", "r", "reshaped_logits", "return", "return_dict", "self", "sequence_output", "sequence_summary", "shape", "size", "token_type_ids", "torch", "tuple", "view"], "electra/modeling_electra.py:ElectraForCausalLM.__init__": ["GenerationMixin", "If", "Linear", "Model", "ModelForCausalLM", "ModelGeneratorPredictions", "ModelModel", "ModelPreTrainedModel", "True", "__init__", "a", "add", "as", "class", "config", "def", "embedding_size", "generator_lm_head", "generator_predictions", "if", "is_decoder", "logger", "nn", "not", "post_init", "self", "standalone", "super", "to", "use", "vocab_size", "want", "warning", "you"], "electra/modeling_electra.py:ElectraForCausalLM.get_output_embeddings": ["GenerationMixin", "ModelForCausalLM", "ModelPreTrainedModel", "class", "def", "generator_lm_head", "get_output_embeddings", "return", "self"], "electra/modeling_electra.py:ElectraForCausalLM.set_output_embeddings": ["GenerationMixin", "ModelForCausalLM", "ModelPreTrainedModel", "class", "def", "generator_lm_head", "new_embeddings", "self", "set_output_embeddings"], "electra/modeling_electra.py:ElectraForCausalLM.forward": ["BaseModelOutputWithPastAndCrossAttentions", "Cache", "CausalLMOutputWithCrossAttentions", "False", "GenerationMixin", "Model", "ModelForCausalLM", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "cross_attentions", "def", "else", "encoder_attention_mask", "encoder_hidden_states", "forward", "generator_lm_head", "generator_predictions", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "logits", "logits_to_keep", "loss", "loss_function", "not", "outputs", "past_key_values", "position_ids", "r", "return", "return_dict", "self", "slice", "slice_indices", "token_type_ids", "torch", "tuple", "use_cache", "vocab_size"], "emu3/modeling_emu3.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "emu3/modeling_emu3.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cos", "def", "k", "k_embed", "q", "q_embed", "return", "rotate_half", "sin", "unsqueeze", "unsqueeze_dim"], "emu3/modeling_emu3.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "emu3/modeling_emu3.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value", "value_states"], "emu3/modeling_emu3.py:Emu3Attention.__init__": ["Linear", "ModelAttention", "ModelConfig", "Module", "True", "__init__", "attention_bias", "attention_dropout", "bias", "class", "config", "def", "getattr", "head_dim", "hidden_size", "int", "is_causal", "k_proj", "layer_idx", "nn", "num_attention_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "q_proj", "scaling", "self", "super", "v_proj"], "emu3/modeling_emu3.py:Emu3Attention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "LongTensor", "ModelAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_attn_implementation", "apply_rotary_pos_emb", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "cache_kwargs", "cache_position", "class", "config", "contiguous", "cos", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "head_dim", "hidden_shape", "hidden_states", "if", "input_shape", "is", "k_proj", "key_states", "kwargs", "layer_idx", "nn", "not", "o_proj", "past_key_values", "position_embeddings", "q_proj", "query_states", "reshape", "return", "scaling", "self", "shape", "sin", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "view"], "emu3/modeling_emu3.py:Emu3RMSNorm.__init__": ["ModelRMSNorm", "Module", "Parameter", "__init__", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "emu3/modeling_emu3.py:Emu3RMSNorm.forward": ["ModelRMSNorm", "Module", "True", "class", "def", "dtype", "float32", "forward", "hidden_states", "input_dtype", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "emu3/modeling_emu3.py:Emu3RMSNorm.extra_repr": ["ModelRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "variance_epsilon", "weight"], "emu3/modeling_emu3.py:Emu3MLP.__init__": ["ACT2FN", "Linear", "ModelMLP", "Module", "__init__", "act_fn", "bias", "class", "config", "def", "down_proj", "gate_proj", "hidden_act", "hidden_size", "intermediate_size", "mlp_bias", "nn", "self", "super", "up_proj"], "emu3/modeling_emu3.py:Emu3MLP.forward": ["ModelMLP", "Module", "act_fn", "class", "def", "down_proj", "forward", "gate_proj", "nn", "return", "self", "up_proj", "x"], "emu3/modeling_emu3.py:Emu3DecoderLayer.__init__": ["Dropout", "GradientCheckpointingLayer", "ModelAttention", "ModelConfig", "ModelDecoderLayer", "ModelMLP", "ModelRMSNorm", "__init__", "attention_dropout", "class", "config", "def", "dropout", "eps", "hidden_size", "input_layernorm", "int", "layer_idx", "mlp", "nn", "post_attention_layernorm", "rms_norm_eps", "self", "self_attn", "super"], "emu3/modeling_emu3.py:Emu3DecoderLayer.forward": ["Cache", "False", "GradientCheckpointingLayer", "LongTensor", "ModelDecoderLayer", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "attention_mask", "bool", "cache_position", "class", "def", "dropout", "forward", "hidden_states", "input_layernorm", "kwargs", "mlp", "past_key_values", "position_embeddings", "position_ids", "post_attention_layernorm", "residual", "return", "self", "self_attn", "torch", "tuple", "use_cache"], "emu3/modeling_emu3.py:Emu3VQVAEVectorQuantizer.__init__": ["Embedding", "ModelVQVAEConfig", "ModelVQVAEVectorQuantizer", "Module", "__init__", "class", "codebook_size", "config", "data", "def", "embed_dim", "embedding", "nn", "self", "super", "uniform_", "weight"], "emu3/modeling_emu3.py:Emu3VQVAEVectorQuantizer.forward": ["ModelVQVAEVectorQuantizer", "Module", "Tensor", "True", "argmin", "batch_size", "channels", "class", "contiguous", "def", "dim", "distances", "embedding", "embedding_sum", "forward", "height", "hidden_state", "hidden_state_flattened", "hidden_state_sum", "keepdim", "matmul", "min_encoding_indices", "nn", "permute", "return", "self", "shape", "sum", "temporal", "torch", "transpose", "view", "weight", "width"], "emu3/modeling_emu3.py:Emu3VQVAEEncoderConvDownsample.__init__": ["Conv2d", "ModelVQVAEEncoderConvDownsample", "Module", "__init__", "class", "conv", "def", "in_channels", "kernel_size", "nn", "padding", "self", "stride", "super"], "emu3/modeling_emu3.py:Emu3VQVAEEncoderConvDownsample.forward": ["F", "ModelVQVAEEncoderConvDownsample", "Module", "class", "constant", "conv", "def", "forward", "hidden_states", "mode", "nn", "pad", "return", "self", "value"], "emu3/modeling_emu3.py:Emu3VQVAEEncoderConvUpsample.__init__": ["Conv2d", "ModelVQVAEEncoderConvUpsample", "Module", "__init__", "class", "conv", "def", "in_channels", "kernel_size", "nn", "padding", "self", "stride", "super"], "emu3/modeling_emu3.py:Emu3VQVAEEncoderConvUpsample.forward": ["F", "ModelVQVAEEncoderConvUpsample", "Module", "class", "conv", "def", "forward", "hidden_states", "interpolate", "mode", "nearest", "nn", "return", "scale_factor", "self"], "emu3/modeling_emu3.py:Emu3VQVAEConv3d.__init__": ["Conv3d", "ModelVQVAEConv3d", "Module", "__init__", "class", "conv", "def", "for", "in", "in_channel", "int", "kernel_size", "nn", "one_kernel", "one_stride", "out_channel", "pad_size", "padding", "padding_sizes", "self", "stride", "super", "tuple", "zip"], "emu3/modeling_emu3.py:Emu3VQVAEConv3d.forward": ["F", "ModelVQVAEConv3d", "Module", "Tensor", "class", "conv", "def", "forward", "hidden_states", "nn", "pad", "padding", "return", "self", "torch"], "emu3/modeling_emu3.py:Emu3VQVAESpatialNorm.__init__": ["Conv2d", "GroupNorm", "ModelVQVAESpatialNorm", "Module", "True", "__init__", "affine", "class", "conv_b", "conv_y", "def", "eps", "in_channels", "int", "kernel_size", "nn", "norm_layer", "num_channels", "num_groups", "out_channels", "padding", "self", "stride", "super"], "emu3/modeling_emu3.py:Emu3VQVAESpatialNorm.forward": ["F", "ModelVQVAESpatialNorm", "Module", "Tensor", "class", "conv_b", "conv_y", "def", "forward", "hidden_states", "interpolate", "mode", "nearest", "nn", "norm_layer", "quant_states", "return", "self", "shape", "size", "torch"], "emu3/modeling_emu3.py:Emu3VQVAETemporalUpsample.__init__": ["ModelVQVAEConv3d", "ModelVQVAETemporalUpsample", "Module", "__init__", "class", "conv", "def", "in_channel", "int", "kernel_size", "nn", "out_channel", "self", "stride", "super"], "emu3/modeling_emu3.py:Emu3VQVAETemporalUpsample.forward": ["F", "ModelVQVAETemporalUpsample", "Module", "Tensor", "batch_size", "channels", "class", "contiguous", "conv", "def", "forward", "height", "hidden_states", "interpolate", "mode", "nearest", "nn", "permute", "return", "scale_factor", "self", "shape", "temporal", "torch", "view", "width"], "emu3/modeling_emu3.py:Emu3VQVAETemporalDownsample.__init__": ["ModelVQVAEConv3d", "ModelVQVAETemporalDownsample", "Module", "__init__", "class", "conv", "def", "in_channel", "int", "kernel_size", "nn", "out_channel", "self", "stride", "super"], "emu3/modeling_emu3.py:Emu3VQVAETemporalDownsample.forward": ["ModelVQVAETemporalDownsample", "Module", "Tensor", "class", "conv", "def", "forward", "hidden_states", "nn", "return", "self", "torch"], "emu3/modeling_emu3.py:Emu3VQVAETemporalResnetBlock.__init__": ["BatchNorm3d", "Conv3d", "ModelVQVAEConv3d", "ModelVQVAETemporalResnetBlock", "Module", "None", "__init__", "class", "conv1", "conv2", "def", "else", "if", "in_channels", "is", "kernel_size", "nin_shortcut", "nn", "norm1", "norm2", "out_channels", "padding", "self", "stride", "super"], "emu3/modeling_emu3.py:Emu3VQVAETemporalResnetBlock.forward": ["ModelVQVAETemporalResnetBlock", "Module", "class", "conv1", "conv2", "def", "forward", "hidden_states", "if", "in_channels", "nin_shortcut", "nn", "norm1", "norm2", "out_channels", "residual", "return", "self", "sigmoid", "torch"], "emu3/modeling_emu3.py:Emu3VQVAEResnetBlock.__init__": ["Conv2d", "GroupNorm", "ModelVQVAEResnetBlock", "ModelVQVAESpatialNorm", "Module", "None", "True", "__init__", "affine", "class", "conv1", "conv2", "def", "else", "eps", "if", "in_channels", "int", "is", "kernel_size", "nin_shortcut", "nn", "norm1", "norm2", "num_channels", "num_groups", "out_channels", "padding", "quant_channels", "self", "stride", "super"], "emu3/modeling_emu3.py:Emu3VQVAEResnetBlock.forward": ["ModelVQVAEResnetBlock", "Module", "None", "Tensor", "class", "conv1", "conv2", "def", "else", "forward", "hidden_states", "if", "in_channels", "is", "nin_shortcut", "nn", "norm1", "norm2", "norm_args", "out_channels", "quant_channels", "residual", "return", "self", "sigmoid", "torch"], "emu3/modeling_emu3.py:Emu3VQVAEAttentionBlock.__init__": ["False", "Linear", "ModelVQVAEAttentionBlock", "ModelVQVAEConfig", "Module", "ValueError", "__init__", "and", "attention_dropout", "be", "by", "class", "config", "def", "divisible", "dropout", "embed_dim", "f", "got", "head_dim", "hidden_size", "if", "is_causal", "k_proj", "must", "nn", "num_attention_heads", "num_heads", "num_key_value_groups", "out_proj", "q_proj", "raise", "scale", "self", "super", "v_proj"], "emu3/modeling_emu3.py:Emu3VQVAEAttentionBlock.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "ModelVQVAEAttentionBlock", "Module", "None", "Tensor", "_attn_implementation", "attention_interface", "attention_mask", "attn_output", "attn_weights", "batch_size", "class", "config", "contiguous", "def", "dropout", "eager", "eager_attention_forward", "else", "embed_dim", "forward", "head_dim", "hidden_states", "if", "is_causal", "k_proj", "keys", "kwargs", "nn", "not", "num_heads", "out_proj", "q_proj", "queries", "reshape", "return", "scale", "scaling", "self", "seq_length", "shape", "torch", "training", "transpose", "tuple", "v_proj", "values", "view"], "emu3/modeling_emu3.py:Emu3VQVAEGroupNorm.__init__": ["GroupNorm", "ModelVQVAEGroupNorm", "__init__", "class", "def", "kwargs", "nn", "self", "super"], "emu3/modeling_emu3.py:Emu3VQVAEGroupNorm.forward": ["F", "GroupNorm", "ModelVQVAEGroupNorm", "None", "bias", "class", "def", "eps", "forward", "group_norm", "input", "nn", "num_groups", "quant_states", "return", "self", "weight"], "emu3/modeling_emu3.py:Emu3VQVAEMiddleBlock.__init__": ["ModelVQVAEAttentionBlock", "ModelVQVAEGroupNorm", "ModelVQVAEMiddleBlock", "ModelVQVAEResnetBlock", "ModelVQVAESpatialNorm", "Module", "None", "True", "__init__", "affine", "attn_1", "attn_norm", "block_1", "block_2", "class", "config", "def", "else", "eps", "if", "in_channels", "is", "nn", "num_channels", "num_groups", "out_channels", "quant_channels", "self", "super"], "emu3/modeling_emu3.py:Emu3VQVAEMiddleBlock.forward": ["FloatTensor", "ModelVQVAEMiddleBlock", "Module", "None", "attn_1", "attn_norm", "batch_size", "block_1", "block_2", "channels", "class", "def", "forward", "height", "hidden_states", "nn", "permute", "quant_states", "reshape", "residual", "return", "self", "shape", "torch", "transpose", "view", "width"], "emu3/modeling_emu3.py:Emu3VQVAEDownBlock.__init__": ["GroupNorm", "ModelVQVAEAttentionBlock", "ModelVQVAEDownBlock", "ModelVQVAEEncoderConvDownsample", "ModelVQVAEResnetBlock", "Module", "ModuleList", "None", "True", "__init__", "affine", "and", "append", "attn", "attn_norms", "attn_resolutions", "base_channels", "block", "block_in", "block_out", "channel_multiplier", "class", "config", "def", "down", "downsample", "eps", "for", "i_block", "i_level", "if", "in", "in_channel_multiplier", "in_channels", "is", "len", "nn", "not", "num_channels", "num_groups", "num_res_blocks", "num_resolutions", "out_channels", "range", "self", "super", "tuple"], "emu3/modeling_emu3.py:Emu3VQVAEDownBlock.forward": ["FloatTensor", "ModelVQVAEDownBlock", "Module", "attn", "attn_norms", "batch_size", "block", "blocks", "channels", "class", "def", "down", "downsample", "enumerate", "for", "forward", "height", "hidden_states", "i_block", "i_level", "if", "in", "len", "nn", "num_res_blocks", "num_resolutions", "permute", "range", "reshape", "residual", "return", "self", "shape", "torch", "transpose", "view", "width"], "emu3/modeling_emu3.py:Emu3VQVAEUpBlock.__init__": ["ModelVQVAEAttentionBlock", "ModelVQVAEEncoderConvUpsample", "ModelVQVAEResnetBlock", "ModelVQVAESpatialNorm", "ModelVQVAEUpBlock", "Module", "ModuleList", "__init__", "append", "attn", "attn_norms", "attn_resolutions", "base_channels", "block", "block_in", "block_out", "channel_multiplier", "class", "config", "def", "embed_dim", "for", "i_block", "i_level", "if", "in", "in_channels", "insert", "len", "nn", "num_res_blocks", "num_resolutions", "out_channels", "quant_channels", "range", "reversed", "self", "super", "up", "upsample"], "emu3/modeling_emu3.py:Emu3VQVAEUpBlock.forward": ["FloatTensor", "ModelVQVAEUpBlock", "Module", "attn", "attn_norms", "batch_size", "block", "blocks", "channels", "class", "def", "enumerate", "for", "forward", "height", "hidden_states", "i_block", "i_level", "if", "in", "len", "nn", "num_res_blocks", "permute", "quant_states", "range", "reshape", "residual", "return", "self", "shape", "torch", "transpose", "up", "upsample", "view", "width"], "emu3/modeling_emu3.py:Emu3VQVAEEncoder.__init__": ["Conv2d", "GroupNorm", "ModelVQVAEDownBlock", "ModelVQVAEEncoder", "ModelVQVAEMiddleBlock", "ModelVQVAETemporalDownsample", "ModelVQVAETemporalResnetBlock", "Module", "ModuleList", "True", "_", "__init__", "affine", "append", "base_channels", "block_in", "channel_multiplier", "class", "config", "conv", "conv_in", "conv_out", "def", "double_latent", "down_block", "else", "eps", "for", "i", "if", "in", "in_channels", "int", "kernel_size", "latent_channels", "log2", "math", "middle_block", "nn", "norm_out", "num_channels", "num_groups", "num_res_blocks", "out_channels", "padding", "range", "self", "stride", "super", "temporal_down_blocks", "temporal_downsample_factor", "time_conv", "time_res_conv", "time_res_stack", "torch"], "emu3/modeling_emu3.py:Emu3VQVAEEncoder.forward": ["LongTensor", "ModelVQVAEEncoder", "Module", "class", "conv", "conv_in", "conv_out", "def", "down_block", "for", "forward", "hidden_states", "in", "layer", "middle_block", "nn", "norm_out", "permute", "pixel_values", "reshape", "return", "self", "shape", "sigmoid", "temporal_dim", "time_conv", "time_res_stack", "torch"], "emu3/modeling_emu3.py:Emu3VQVAEDecoder.__init__": ["Conv2d", "ModelVQVAEConfig", "ModelVQVAEDecoder", "ModelVQVAEMiddleBlock", "ModelVQVAESpatialNorm", "ModelVQVAETemporalResnetBlock", "ModelVQVAETemporalUpsample", "ModelVQVAEUpBlock", "Module", "ModuleList", "_", "__init__", "append", "base_channels", "block_in", "channel_multiplier", "class", "config", "conv", "conv_in", "conv_out", "def", "embed_dim", "for", "i", "in", "in_channels", "int", "kernel_size", "latent_channels", "log2", "math", "middle_block", "nn", "norm_out", "num_res_blocks", "out_channels", "padding", "quant_channels", "range", "self", "stride", "super", "temp_upsample_block_num", "temporal_downsample_factor", "time_conv", "time_res_conv", "time_res_stack", "up_block"], "emu3/modeling_emu3.py:Emu3VQVAEDecoder.forward": ["ModelVQVAEDecoder", "Module", "Tensor", "cat", "chunk", "class", "conv_in", "conv_out", "def", "dim", "for", "forward", "hidden_quant_states", "hidden_states", "in", "layer", "middle_block", "nn", "norm_out", "permute", "quant_states", "reshape", "return", "self", "shape", "sigmoid", "time_conv", "time_res_stack", "torch", "up_block"], "emu3/modeling_emu3.py:Emu3VQVAE._init_weights": ["BatchNorm2d", "BatchNorm3d", "Conv2d", "Conv3d", "Embedding", "False", "GroupNorm", "Linear", "ModelVQVAE", "None", "PreTrainedModel", "_", "_calculate_fan_in_and_fan_out", "_init_weights", "_is_hf_initialized", "a", "and", "bias", "bound", "class", "constant_", "def", "elif", "else", "fan_in", "fan_out", "getattr", "if", "init", "is", "isinstance", "kaiming_normal_", "kaiming_uniform_", "math", "mode", "module", "nn", "nonlinearity", "normal_", "not", "num_batches_tracked", "ones_", "padding_idx", "relu", "running_mean", "running_var", "self", "sqrt", "torch", "uniform_", "weight", "zeros_"], "emu3/modeling_emu3.py:Emu3VQVAE.__init__": ["ModelVQVAE", "ModelVQVAEConfig", "ModelVQVAEConv3d", "ModelVQVAEDecoder", "ModelVQVAEEncoder", "ModelVQVAEVectorQuantizer", "PreTrainedModel", "__init__", "channel_multiplier", "class", "config", "decoder", "def", "embed_dim", "encoder", "eval", "kernel_size", "latent_channels", "len", "post_init", "post_quant_conv", "quant_conv", "quantize", "self", "spatial_scale_factor", "stride", "super", "vision_spatial_factor"], "emu3/modeling_emu3.py:Emu3VQVAE.encode": ["ModelVQVAE", "PreTrainedModel", "Tensor", "batch_size", "channels", "class", "codes", "config", "def", "else", "encode", "encoder", "for", "height", "hidden_states", "if", "image_sizes", "image_tokens", "in", "int", "is_image", "ndim", "permute", "pixel_values", "quant_conv", "quantize", "repeat", "return", "self", "shape", "single_image", "size", "squeeze", "temporal", "temporal_downsample_factor", "torch", "unsqueeze", "vision_spatial_factor", "width", "zip"], "emu3/modeling_emu3.py:Emu3VQVAE.decode": ["ModelVQVAE", "PreTrainedModel", "Tensor", "batch_size", "channels", "class", "config", "contiguous", "decode", "decoder", "def", "else", "embedding", "flatten", "height", "hidden_states", "if", "is_image", "ndim", "out_channels", "permute", "post_quant", "post_quant_conv", "quant", "quantize", "reshape", "return", "self", "shape", "spatial_scale_factor", "temporal", "temporal_downsample_factor", "torch", "unsqueeze", "video", "view", "width"], "emu3/modeling_emu3.py:Emu3ImageVocabularyMapping.__init__": ["ModelImageVocabularyMapping", "__init__", "class", "def", "eol_token_id", "extra_200", "get", "image", "image_token_id", "self", "vocab_map"], "emu3/modeling_emu3.py:Emu3ImageVocabularyMapping.image_tokens": ["ModelImageVocabularyMapping", "class", "def", "for", "if", "image_tokens", "in", "items", "name", "return", "self", "sorted", "startswith", "token", "val", "visual", "vocab_map"], "emu3/modeling_emu3.py:Emu3ImageVocabularyMapping.image_tokens_str": ["ModelImageVocabularyMapping", "class", "def", "for", "if", "image_tokens_str", "in", "items", "name", "return", "self", "sorted", "startswith", "token", "val", "visual", "vocab_map"], "emu3/modeling_emu3.py:Emu3ImageVocabularyMapping.img2bpe": ["ModelImageVocabularyMapping", "class", "def", "for", "image_tokens_str", "img2bpe", "in", "int", "return", "self", "token", "vocab_map"], "emu3/modeling_emu3.py:Emu3ImageVocabularyMapping.bpe2img": ["ModelImageVocabularyMapping", "bpe2img", "class", "def", "for", "img2bpe", "in", "items", "k", "return", "self", "v"], "emu3/modeling_emu3.py:Emu3ImageVocabularyMapping.bpe2img_mapping_tensor": ["ModelImageVocabularyMapping", "bpe2img", "bpe2img_mapping_tensor", "class", "def", "dtype", "for", "in", "int", "items", "k", "keys", "mapping", "max", "return", "self", "torch", "v", "zeros"], "emu3/modeling_emu3.py:Emu3ImageVocabularyMapping.img2bpe_mapping_tensor": ["ModelImageVocabularyMapping", "class", "def", "dtype", "for", "img2bpe", "img2bpe_mapping_tensor", "in", "int", "items", "k", "keys", "mapping", "max", "return", "self", "torch", "v", "zeros"], "emu3/modeling_emu3.py:Emu3ImageVocabularyMapping.convert_img2bpe": ["ModelImageVocabularyMapping", "Tensor", "cat", "class", "convert_img2bpe", "cpu", "def", "device", "dim", "dtype", "eol_row", "eol_token_id", "img2bpe_mapping_tensor", "img_batch", "img_tokens", "int", "list", "ones", "return", "self", "shape", "to", "torch"], "emu3/modeling_emu3.py:Emu3ImageVocabularyMapping.convert_bpe2img": ["ModelImageVocabularyMapping", "Tensor", "bpe2img_mapping_tensor", "class", "convert_bpe2img", "cpu", "def", "device", "img_batch", "img_tokens", "return", "self", "to", "torch"], "emu3/modeling_emu3.py:Emu3RotaryEmbedding.__init__": ["Callable", "False", "ModelConfig", "ModelRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "emu3/modeling_emu3.py:Emu3RotaryEmbedding.compute_default_rope_parameters": ["ModelConfig", "ModelRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "emu3/modeling_emu3.py:Emu3RotaryEmbedding.forward": ["False", "ModelRotaryEmbedding", "Module", "None", "and", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "emu3/modeling_emu3.py:Emu3TextModel.__init__": ["Embedding", "False", "ModelConfig", "ModelDecoderLayer", "ModelPreTrainedModel", "ModelRMSNorm", "ModelRotaryEmbedding", "ModelTextModel", "ModuleList", "__init__", "class", "config", "def", "embed_tokens", "eps", "for", "gradient_checkpointing", "hidden_size", "in", "layer_idx", "layers", "nn", "norm", "num_hidden_layers", "pad_token_id", "padding_idx", "post_init", "range", "rms_norm_eps", "rotary_emb", "self", "super", "vocab_size"], "emu3/modeling_emu3.py:Emu3TextModel.forward": ["BaseModelOutputWithPast", "Cache", "DynamicCache", "FloatTensor", "LongTensor", "ModelPreTrainedModel", "ModelTextModel", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "and", "arange", "attention_mask", "bool", "cache_position", "causal_mask", "class", "config", "create_causal_mask", "decoder_layer", "def", "device", "else", "embed_tokens", "exactly", "for", "forward", "get_seq_length", "hidden_states", "if", "in", "input_embeds", "input_ids", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layers", "must", "norm", "not", "num_hidden_layers", "of", "one", "or", "past_key_values", "past_seen_tokens", "position_embeddings", "position_ids", "raise", "return", "rotary_emb", "self", "shape", "specify", "torch", "unsqueeze", "use_cache"], "emu3/modeling_emu3.py:Emu3ForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "ModelForCausalLM", "ModelPreTrainedModel", "ModelTextModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "vocab_size"], "emu3/modeling_emu3.py:Emu3ForCausalLM.forward": ["BaseModelOutputWithPast", "Cache", "CausalLMOutputWithPast", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForCausalLM", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "outputs", "past_key_values", "position_ids", "r", "return", "self", "slice", "slice_indices", "torch", "use_cache", "vocab_size"], "emu3/modeling_emu3.py:Emu3Model.__init__": ["ModelImageVocabularyMapping", "ModelModel", "ModelPreTrainedModel", "ModelTextModel", "ModelVQVAE", "__init__", "_from_config", "class", "config", "def", "post_init", "self", "super", "text_config", "text_model", "vocabulary_map", "vocabulary_mapping", "vq_config", "vqmodel"], "emu3/modeling_emu3.py:Emu3Model.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "return", "self", "text_model"], "emu3/modeling_emu3.py:Emu3Model.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "self", "set_input_embeddings", "text_model", "value"], "emu3/modeling_emu3.py:Emu3Model.get_image_tokens": ["FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "bpe_tokens", "bpe_tokens_list", "cat", "class", "convert_img2bpe", "def", "encode", "flatten", "for", "get_image_tokens", "image_sizes", "image_tokens_list", "in", "pixel_values", "return", "self", "tokens", "torch", "vocabulary_mapping", "vqmodel"], "emu3/modeling_emu3.py:Emu3Model.get_image_features": ["FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "class", "def", "for", "get_image_features", "get_image_tokens", "get_input_embeddings", "height", "image_features", "image_sizes", "image_tokens", "in", "pixel_values", "return", "self", "split", "split_sizes", "torch", "vision_spatial_factor", "vqmodel", "width"], "emu3/modeling_emu3.py:Emu3Model.decode_image_tokens": ["LongTensor", "ModelModel", "ModelPreTrainedModel", "class", "convert_bpe2img", "decode", "decode_image_tokens", "def", "height", "image", "image_tokens", "int", "return", "self", "sequences", "torch", "view", "vocabulary_mapping", "vqmodel", "width"], "emu3/modeling_emu3.py:Emu3Model.get_placeholder_mask": ["FloatTensor", "Image", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "ValueError", "all", "and", "class", "def", "device", "do", "dtype", "else", "expand_as", "f", "features", "get_input_embeddings", "get_placeholder_mask", "if", "image", "image_features", "image_token_id", "input_ids", "inputs_embeds", "is", "long", "match", "n_image_features", "n_image_tokens", "not", "numel", "raise", "return", "self", "shape", "special_image_mask", "sum", "tensor", "to", "tokens", "torch", "unsqueeze", "vocabulary_mapping"], "emu3/modeling_emu3.py:Emu3Model.forward": ["Cache", "CausalLMOutputWithPast", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "and", "at", "attention_mask", "bool", "both", "cache_position", "cannot", "cat", "class", "def", "dim", "either", "forward", "get_image_features", "get_input_embeddings", "get_placeholder_mask", "if", "image_embeds", "image_features", "image_sizes", "input_ids", "inputs_embeds", "is", "kwargs", "masked_scatter", "must", "not", "one", "outputs", "past_key_values", "pixel_values", "position_ids", "r", "raise", "return", "same", "self", "special_image_mask", "specify", "text_model", "the", "time", "torch", "tuple", "use_cache"], "emu3/modeling_emu3.py:Emu3ForConditionalGeneration.__init__": ["False", "GenerationMixin", "Linear", "ModelForConditionalGeneration", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "text_config", "vocab_size"], "emu3/modeling_emu3.py:Emu3ForConditionalGeneration.get_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "model", "return", "self"], "emu3/modeling_emu3.py:Emu3ForConditionalGeneration.set_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "model", "self", "set_input_embeddings", "value"], "emu3/modeling_emu3.py:Emu3ForConditionalGeneration.get_output_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "Module", "class", "def", "get_output_embeddings", "lm_head", "nn", "return", "self"], "emu3/modeling_emu3.py:Emu3ForConditionalGeneration.decode_image_tokens": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "decode_image_tokens", "def", "kwargs", "model", "return", "self"], "emu3/modeling_emu3.py:Emu3ForConditionalGeneration.forward": ["Cache", "CausalLMOutputWithPast", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "else", "forward", "hidden_states", "if", "image_sizes", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "outputs", "past_key_values", "pixel_values", "position_ids", "r", "return", "self", "slice", "slice_indices", "text_config", "torch", "tuple", "use_cache", "vocab_size"], "emu3/modeling_emu3.py:Emu3ForConditionalGeneration.prepare_inputs_for_generation": ["False", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "True", "and", "attention_mask", "cache_position", "class", "def", "if", "input_ids", "inputs_embeds", "is_first_iteration", "kwargs", "model_inputs", "not", "past_key_values", "pixel_values", "position_ids", "prepare_inputs_for_generation", "return", "self", "super", "use_cache"], "encodec/modeling_encodec.py:EncodecConv1d.__init__": ["Conv1d", "False", "GroupNorm", "ModelConv1d", "Module", "ValueError", "__init__", "and", "be", "been", "causal", "class", "config", "conv", "def", "dilation", "dtype", "elif", "f", "got", "has", "hasattr", "if", "in", "in_channels", "initialized", "int", "int64", "kernel_size", "logger", "must", "nn", "norm", "norm_type", "not", "of", "one", "out_channels", "pad_mode", "padding_total", "parametrizations", "persistent", "raise", "register_buffer", "self", "stride", "super", "tensor", "time_group_norm", "torch", "use_causal_conv", "utils", "warning", "weight_norm", "with"], "encodec/modeling_encodec.py:EncodecConv1d._get_extra_padding_for_conv1d": ["ModelConv1d", "Module", "Tensor", "_get_extra_padding_for_conv1d", "ceil", "class", "def", "hidden_states", "ideal_length", "int64", "kernel_size", "length", "n_frames", "nn", "padding_total", "return", "self", "shape", "stride", "to", "torch"], "encodec/modeling_encodec.py:EncodecConv1d._pad1d": ["ModelConv1d", "Module", "Tensor", "_pad1d", "class", "def", "end", "extra_pad", "float", "functional", "hidden_states", "if", "int", "length", "max", "max_pad", "mode", "nn", "pad", "padded", "padding_left", "padding_right", "paddings", "reflect", "return", "shape", "str", "torch", "tuple", "value", "zero"], "encodec/modeling_encodec.py:EncodecConv1d.forward": ["ModelConv1d", "Module", "_get_extra_padding_for_conv1d", "_pad1d", "causal", "class", "conv", "def", "else", "extra_padding", "forward", "hidden_states", "if", "mode", "nn", "norm", "norm_type", "pad_mode", "padding_left", "padding_right", "padding_total", "return", "self", "time_group_norm"], "encodec/modeling_encodec.py:EncodecConvTranspose1d.__init__": ["ConvTranspose1d", "GroupNorm", "ModelConvTranspose1d", "Module", "ValueError", "__init__", "be", "causal", "class", "config", "conv", "convolutions", "def", "elif", "f", "for", "got", "hasattr", "if", "in", "in_channels", "int", "kernel_size", "makes", "must", "nn", "norm", "norm_type", "not", "of", "one", "only", "or", "out_channels", "parametrizations", "raise", "self", "sense", "stride", "super", "time_group_norm", "trim_right_ratio", "use_causal_conv", "utils", "weight_norm"], "encodec/modeling_encodec.py:EncodecConvTranspose1d.forward": ["ModelConvTranspose1d", "Module", "causal", "ceil", "class", "conv", "def", "else", "end", "forward", "hidden_states", "if", "kernel_size", "math", "nn", "norm", "norm_type", "padding_left", "padding_right", "padding_total", "return", "self", "shape", "stride", "time_group_norm", "trim_right_ratio"], "encodec/modeling_encodec.py:EncodecLSTM.__init__": ["LSTM", "ModelConfig", "ModelLSTM", "Module", "__init__", "class", "config", "def", "dimension", "int", "lstm", "nn", "num_lstm_layers", "self", "super"], "encodec/modeling_encodec.py:EncodecLSTM.forward": ["ModelLSTM", "Module", "class", "def", "forward", "hidden_states", "lstm", "nn", "permute", "return", "self"], "encodec/modeling_encodec.py:EncodecResnetBlock.__init__": ["ELU", "Identity", "ModelConfig", "ModelConv1d", "ModelResnetBlock", "Module", "ModuleList", "Number", "ValueError", "__init__", "block", "class", "compress", "config", "def", "dilation", "dilations", "dim", "else", "enumerate", "for", "hidden", "i", "if", "in", "in_chs", "int", "kernel", "kernel_size", "kernel_sizes", "len", "list", "match", "nn", "number", "of", "out_chs", "raise", "residual_kernel_size", "self", "shortcut", "should", "sizes", "super", "use_conv_shortcut", "zip"], "encodec/modeling_encodec.py:EncodecResnetBlock.forward": ["ModelResnetBlock", "Module", "block", "class", "def", "for", "forward", "hidden_states", "in", "layer", "nn", "residual", "return", "self", "shortcut"], "encodec/modeling_encodec.py:EncodecEncoder.__init__": ["ELU", "ModelConfig", "ModelConv1d", "ModelEncoder", "ModelLSTM", "ModelResnetBlock", "Module", "ModuleList", "__init__", "audio_channels", "class", "config", "current_scale", "def", "dilation_growth_rate", "for", "hidden_size", "in", "j", "kernel_size", "last_kernel_size", "layers", "model", "nn", "num_filters", "num_residual_layers", "range", "ratio", "reversed", "scaling", "self", "stride", "super", "upsampling_ratios"], "encodec/modeling_encodec.py:EncodecEncoder.forward": ["ModelEncoder", "Module", "class", "def", "for", "forward", "hidden_states", "in", "layer", "layers", "nn", "return", "self"], "encodec/modeling_encodec.py:EncodecDecoder.__init__": ["ELU", "ModelConfig", "ModelConv1d", "ModelConvTranspose1d", "ModelDecoder", "ModelLSTM", "ModelResnetBlock", "Module", "ModuleList", "__init__", "audio_channels", "class", "config", "current_scale", "def", "dilation_growth_rate", "for", "hidden_size", "in", "int", "j", "kernel_size", "last_kernel_size", "layers", "len", "model", "nn", "num_filters", "num_residual_layers", "range", "ratio", "scaling", "self", "stride", "super", "upsampling_ratios"], "encodec/modeling_encodec.py:EncodecDecoder.forward": ["ModelDecoder", "Module", "class", "def", "for", "forward", "hidden_states", "in", "layer", "layers", "nn", "return", "self"], "encodec/modeling_encodec.py:EncodecEuclideanCodebook.__init__": ["ModelConfig", "ModelEuclideanCodebook", "Module", "Tensor", "True", "__init__", "class", "clone", "cluster_size", "codebook_dim", "codebook_size", "config", "def", "embed", "embed_avg", "inited", "nn", "register_buffer", "self", "super", "torch", "zeros"], "encodec/modeling_encodec.py:EncodecEuclideanCodebook.quantize": ["ModelEuclideanCodebook", "Module", "True", "class", "def", "dim", "dist", "embed", "embed_ind", "hidden_states", "indices", "keepdim", "max", "nn", "pow", "quantize", "return", "scaled_states", "self", "sum", "t"], "encodec/modeling_encodec.py:EncodecEuclideanCodebook.encode": ["ModelEuclideanCodebook", "Module", "class", "def", "embed_ind", "encode", "hidden_states", "nn", "quantize", "reshape", "return", "self", "shape", "view"], "encodec/modeling_encodec.py:EncodecEuclideanCodebook.decode": ["ModelEuclideanCodebook", "Module", "class", "decode", "def", "embed", "embed_ind", "embedding", "functional", "nn", "quantize", "return", "self"], "encodec/modeling_encodec.py:EncodecVectorQuantization.__init__": ["ModelConfig", "ModelEuclideanCodebook", "ModelVectorQuantization", "Module", "__init__", "class", "codebook", "config", "def", "nn", "self", "super"], "encodec/modeling_encodec.py:EncodecVectorQuantization.encode": ["ModelVectorQuantization", "Module", "class", "codebook", "def", "embed_in", "encode", "hidden_states", "nn", "permute", "return", "self"], "encodec/modeling_encodec.py:EncodecVectorQuantization.decode": ["ModelVectorQuantization", "Module", "class", "codebook", "decode", "def", "embed_ind", "nn", "permute", "quantize", "return", "self"], "encodec/modeling_encodec.py:EncodecResidualVectorQuantizer.__init__": ["ModelConfig", "ModelResidualVectorQuantizer", "ModelVectorQuantization", "Module", "ModuleList", "_", "__init__", "class", "codebook_size", "config", "def", "for", "frame_rate", "in", "layers", "nn", "num_quantizers", "range", "self", "super"], "encodec/modeling_encodec.py:EncodecResidualVectorQuantizer.get_num_quantizers_for_bandwidth": ["ModelResidualVectorQuantizer", "Module", "None", "and", "bandwidth", "bw_per_q", "class", "codebook_size", "def", "float", "floor", "frame_rate", "get_num_quantizers_for_bandwidth", "if", "int", "is", "log2", "math", "max", "nn", "not", "num_quantizers", "return", "self"], "encodec/modeling_encodec.py:EncodecResidualVectorQuantizer.encode": ["ModelResidualVectorQuantizer", "Module", "None", "Tensor", "all_indices", "append", "bandwidth", "class", "decode", "def", "embeddings", "encode", "float", "for", "get_num_quantizers_for_bandwidth", "in", "indices", "layer", "layers", "nn", "num_quantizers", "out_indices", "quantized", "residual", "return", "self", "stack", "torch"], "encodec/modeling_encodec.py:EncodecResidualVectorQuantizer.decode": ["ModelResidualVectorQuantizer", "Module", "Tensor", "class", "codes", "decode", "def", "device", "enumerate", "for", "i", "in", "indices", "layer", "layers", "nn", "quantized", "quantized_out", "return", "self", "tensor", "torch"], "encodec/modeling_encodec.py:EncodecPreTrainedModel._init_weights": ["Conv1d", "ConvTranspose1d", "GroupNorm", "LSTM", "ModelConv1d", "ModelEuclideanCodebook", "ModelPreTrainedModel", "None", "PreTrainedAudioTokenizerBase", "Tensor", "True", "_init_weights", "a", "b", "bias", "class", "cluster_size", "constant_", "conv", "copy_", "def", "dilation", "dtype", "elif", "embed", "embed_avg", "for", "groups", "if", "in", "in_channels", "init", "inited", "int64", "is", "isinstance", "k", "kaiming_normal_", "kernel_size", "math", "module", "name", "named_parameters", "nn", "not", "ones_", "padding_total", "param", "reset_parameters", "self", "sqrt", "stride", "tensor", "torch", "uniform_", "weight", "xavier_uniform_", "zeros_"], "encodec/modeling_encodec.py:EncodecModel.__init__": ["ModelConfig", "ModelDecoder", "ModelEncoder", "ModelModel", "ModelPreTrainedModel", "ModelResidualVectorQuantizer", "The", "ValueError", "__init__", "a", "be", "bits_per_codebook", "class", "codebook_size", "config", "decoder", "def", "encoder", "if", "int", "log2", "math", "must", "of", "post_init", "power", "quantizer", "raise", "self", "super"], "encodec/modeling_encodec.py:EncodecModel._encode_frame": ["Duration", "ModelModel", "ModelPreTrainedModel", "None", "RuntimeError", "Tensor", "True", "_encode_frame", "and", "bandwidth", "chunk", "chunk_length_s", "class", "codes", "config", "def", "dim", "duration", "embeddings", "encode", "encoder", "f", "float", "frame", "if", "input_values", "is", "keepdim", "length", "longer", "mean", "mono", "normalize", "not", "of", "pow", "quantizer", "raise", "return", "sampling_rate", "scale", "self", "shape", "sqrt", "sum", "than", "torch", "transpose", "tuple", "view"], "encodec/modeling_encodec.py:EncodecModel.encode": ["ModelEncoderOutput", "ModelModel", "ModelPreTrainedModel", "None", "Number", "Select", "Tensor", "This", "ValueError", "_", "_encode_frame", "append", "audio", "bandwidth", "be", "bool", "but", "channels", "chunk_length", "chunk_stride", "class", "config", "def", "doesn", "else", "encode", "encoded_frame", "encoded_frames", "f", "float", "for", "frame", "functional", "got", "if", "in", "input_length", "input_values", "int", "is", "last_frame", "last_frame_pad_length", "mask", "model", "must", "nn", "not", "of", "offset", "one", "ones_like", "or", "pad", "padding_mask", "raise", "range", "return", "return_dict", "scale", "scales", "self", "shape", "stack", "stride", "support", "t", "target_bandwidths", "the", "torch", "tuple", "value", "view"], "encodec/modeling_encodec.py:EncodecModel._linear_overlap_add": ["ModelModel", "ModelPreTrainedModel", "Tensor", "ValueError", "_linear_overlap_add", "abs", "an", "be", "bigger", "cannot", "class", "def", "device", "dtype", "element", "empty", "f", "for", "frame", "frame_length", "frames", "if", "in", "int", "len", "linspace", "list", "min", "minimum", "must", "offset", "out", "raise", "return", "shape", "stride", "sum_weight", "than", "time_vec", "torch", "total_size", "weight", "zero", "zeros"], "encodec/modeling_encodec.py:EncodecModel._decode_frame": ["ModelModel", "ModelPreTrainedModel", "None", "Tensor", "_decode_frame", "class", "codes", "decode", "decoder", "def", "embeddings", "if", "is", "not", "outputs", "quantizer", "return", "scale", "self", "torch", "transpose", "view"], "encodec/modeling_encodec.py:EncodecModel.decode": ["Expected", "LongTensor", "ModelDecoderOutput", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "ValueError", "_decode_frame", "_linear_overlap_add", "and", "append", "audio_codes", "audio_scales", "audio_values", "bool", "chunk_length", "chunk_stride", "class", "config", "decode", "decoded_frames", "def", "else", "enumerate", "f", "for", "frame", "frames", "got", "i", "if", "in", "int", "is", "last_frame_pad_length", "len", "not", "one", "or", "padding_mask", "raise", "return", "return_dict", "scale", "self", "shape", "torch", "tuple", "zip"], "encodec/modeling_encodec.py:EncodecModel.forward": ["BoolTensor", "False", "FloatTensor", "LongTensor", "ModelModel", "ModelOutput", "ModelPreTrainedModel", "None", "Tensor", "ValueError", "You", "and", "audio_codes", "audio_scales", "audio_values", "bandwidth", "bool", "but", "class", "config", "decode", "def", "did", "else", "encode", "float", "forward", "if", "input_values", "int", "is", "last_frame_pad_length", "not", "ones_like", "padding_mask", "r", "raise", "return", "return_dict", "self", "shape", "specified", "specify", "the", "torch", "tuple", "view"], "encoder_decoder/modeling_encoder_decoder.py:shift_tokens_right": ["Make", "Model_tokens_right", "Modeled_input_ids", "None", "Tensor", "ValueError", "attribute", "clone", "configuration", "decoder_start_token_id", "def", "if", "input_ids", "int", "is", "masked_fill_", "model", "new_zeros", "of", "pad_token_id", "raise", "return", "s", "set", "shape", "sure", "the", "to", "torch"], "encoder_decoder/modeling_encoder_decoder.py:EncoderDecoderModel.__init__": ["AutoModel", "AutoModelForCausalLM", "Config", "Either", "GenerationMixin", "GitHub", "Got", "Head", "If", "LM", "Linear", "Model", "ModelConfig", "ModelModel", "Model_hidden_states", "None", "Please", "PreTrainedConfig", "PreTrainedModel", "The", "ValueError", "__class__", "__init__", "_attn_implementation", "a", "an", "and", "be", "by", "class", "com", "config", "config_class", "configuration", "cross_attention_hidden_size", "decoder", "decoder_signature", "def", "discussion", "else", "enc_to_dec_proj", "equal", "f", "following", "for", "forward", "from_Model_configs", "from_config", "get_output_embeddings", "github", "has", "have", "hidden", "hidden_size", "https", "huggingface", "if", "in", "inspect", "is", "isinstance", "issues", "it", "keys", "logger", "model", "nn", "not", "of", "on", "or", "overwritten", "parameters", "passed", "post_init", "prepared", "provided", "r", "raise", "s", "see", "selected", "self", "set", "shared", "should", "signature", "specified", "states", "super", "the", "to", "to_dict", "transformers", "type", "use", "warning", "without"], "encoder_decoder/modeling_encoder_decoder.py:EncoderDecoderModel._init_weights": ["GenerationMixin", "Model", "ModelModel", "PreTrainedModel", "_init_weights", "class", "decoder", "def", "elif", "if", "in", "module", "modules", "self"], "encoder_decoder/modeling_encoder_decoder.py:EncoderDecoderModel.get_input_embeddings": ["GenerationMixin", "Model", "ModelModel", "PreTrainedModel", "class", "def", "get_input_embeddings", "return", "self"], "encoder_decoder/modeling_encoder_decoder.py:EncoderDecoderModel.get_output_embeddings": ["GenerationMixin", "ModelModel", "PreTrainedModel", "class", "decoder", "def", "get_output_embeddings", "return", "self"], "encoder_decoder/modeling_encoder_decoder.py:EncoderDecoderModel.set_output_embeddings": ["GenerationMixin", "ModelModel", "PreTrainedModel", "class", "decoder", "def", "new_embeddings", "return", "self", "set_output_embeddings"], "encoder_decoder/modeling_encoder_decoder.py:EncoderDecoderModel.from_encoder_decoder_pretrained": ["AutoConfig", "AutoModel", "AutoModelForCausalLM", "Cross", "Decoder", "False", "GenerationMixin", "If", "In", "Initializing", "Model", "ModelConfig", "ModelModel", "Model_", "Model_config", "Model_model", "Model_pretrained_model_name_or_path", "None", "PreTrainedModel", "True", "ValueError", "a", "add_cross_attention", "added", "allows", "an", "and", "architecture", "are", "argument", "as", "attention", "attributes", "be", "causal", "class", "cls", "config", "cross", "decoder", "decoder_", "decoder_config", "decoder_model", "decoder_pretrained_model_name_or_path", "def", "defined", "del", "disabled", "do", "f", "for", "from", "from_Model_configs", "from_Model_pretrained", "from_pretrained", "has", "if", "in", "info", "initialize", "initialized", "is", "is_decoder", "items", "key", "kwargs", "kwargs_Model", "kwargs_decoder", "layers", "len", "logger", "make", "mask", "model", "model_args", "not", "of", "or", "order", "pass", "passed", "pop", "r", "raise", "randomly", "return", "return_unused_kwargs", "s", "set", "startswith", "str", "sure", "that", "the", "to", "value", "warning"], "encoder_decoder/modeling_encoder_decoder.py:EncoderDecoderModel.forward": ["BaseModelOutput", "BoolTensor", "Cache", "CrossEntropyLoss", "DEPRECATION_WARNING", "FloatTensor", "FutureWarning", "GenerationMixin", "LongTensor", "Model", "ModelModel", "Model_attention_mask", "Model_attentions", "Model_hidden_states", "Model_last_hidden_state", "Model_outputs", "None", "PreTrainedModel", "Seq2SeqLMOutput", "Tensor", "True", "and", "argument", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "cross_attention_hidden_size", "cross_attentions", "decoder", "decoder_", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_input_ids", "decoder_inputs_embeds", "decoder_outputs", "decoder_start_token_id", "def", "dtype", "elif", "enc_to_dec_proj", "for", "forward", "hidden_size", "hidden_states", "if", "in", "input_ids", "inputs_embeds", "is", "isinstance", "items", "key", "kwargs", "kwargs_Model", "kwargs_decoder", "kwargs_shared", "labels", "last_hidden_state", "len", "logits", "loss", "loss_fct", "not", "num_items_in_batch", "output_attentions", "output_hidden_states", "pad_token_id", "past_key_values", "pop", "r", "reshape", "return", "return_dict", "self", "shift_tokens_right", "startswith", "to", "torch", "tuple", "use_cache", "value", "view", "vocab_size", "warn", "warnings"], "encoder_decoder/modeling_encoder_decoder.py:EncoderDecoderModel.prepare_decoder_input_ids_from_labels": ["GenerationMixin", "ModelModel", "PreTrainedModel", "Tensor", "class", "config", "decoder_start_token_id", "def", "labels", "pad_token_id", "prepare_decoder_input_ids_from_labels", "return", "self", "shift_tokens_right", "torch"], "encoder_decoder/modeling_encoder_decoder.py:EncoderDecoderModel.resize_token_embeddings": ["GenerationMixin", "Model", "ModelModel", "NotImplementedError", "Please", "PreTrainedModel", "Resizing", "args", "class", "decoder", "def", "directly", "embedding", "is", "kwargs", "layers", "methods", "model", "not", "objects", "of", "or", "raise", "resize_token_embeddings", "respective", "self", "supported", "the", "use", "via", "wrapped"], "eomt/modeling_eomt.py:sample_point": ["False", "Model_point", "Tensor", "True", "add_dim", "def", "dim", "functional", "grid_Model", "if", "input_features", "kwargs", "nn", "point_coordinates", "point_features", "return", "squeeze", "torch", "unsqueeze"], "eomt/modeling_eomt.py:pair_wise_dice_loss": ["Model_wise_dice_loss", "None", "T", "Tensor", "def", "denominator", "flatten", "inputs", "labels", "loss", "matmul", "numerator", "return", "sigmoid", "sum", "torch"], "eomt/modeling_eomt.py:pair_wise_sigmoid_cross_entropy_loss": ["BCEWithLogitsLoss", "Model_wise_sigmoid_cross_entropy_loss", "T", "Tensor", "criterion", "cross_entropy_loss_neg", "cross_entropy_loss_pos", "def", "height_and_width", "inputs", "labels", "loss", "loss_neg", "loss_pos", "matmul", "nn", "none", "ones_like", "r", "reduction", "return", "shape", "torch", "zeros_like"], "eomt/modeling_eomt.py:EomtHungarianMatcher.__init__": ["All", "ModelHungarianMatcher", "Module", "ValueError", "__init__", "and", "be", "can", "class", "cost_class", "cost_dice", "cost_mask", "costs", "def", "float", "if", "int", "nn", "num_points", "raise", "self", "super", "t"], "eomt/modeling_eomt.py:EomtHungarianMatcher.forward": ["False", "ModelHungarianMatcher", "Module", "None", "Tensor", "align_corners", "append", "array", "as_tensor", "assigned_indices", "batch_size", "class", "class_labels", "class_queries_logits", "cost_class", "cost_dice", "cost_mask", "cost_matrix", "cpu", "def", "device", "dtype", "for", "forward", "i", "in", "indices", "int64", "j", "linear_sum_assignment", "list", "mask_labels", "masks_queries_logits", "matched_indices", "maximum", "minimum", "nan_to_num", "nn", "np", "num_points", "pair_wise_dice_loss", "pair_wise_sigmoid_cross_entropy_loss", "point_coordinates", "pred_coordinates", "pred_mask", "pred_probs", "rand", "range", "repeat", "return", "sample_point", "self", "shape", "softmax", "squeeze", "target_coordinates", "target_mask", "tensor", "to", "torch", "tuple"], "eomt/modeling_eomt.py:dice_loss": ["Model_loss", "Tensor", "def", "denominator", "flatten", "inputs", "int", "labels", "loss", "num_masks", "numerator", "probs", "r", "return", "sigmoid", "sum"], "eomt/modeling_eomt.py:sigmoid_cross_entropy_loss": ["BCEWithLogitsLoss", "Model_cross_entropy_loss", "Tensor", "criterion", "cross_entropy_loss", "def", "inputs", "int", "labels", "loss", "mean", "nn", "none", "num_masks", "r", "reduction", "return", "sum", "torch"], "eomt/modeling_eomt.py:EomtLoss.__init__": ["ModelConfig", "ModelHungarianMatcher", "ModelLoss", "Module", "__init__", "class", "class_weight", "config", "cost_class", "cost_dice", "cost_mask", "def", "dice_weight", "dict", "empty_weight", "eos_coef", "float", "importance_sample_ratio", "mask_weight", "matcher", "nn", "no_object_weight", "num_labels", "num_points", "ones", "oversample_ratio", "register_buffer", "requires_backends", "scipy", "self", "str", "super", "torch", "train_num_points", "weight_dict"], "eomt/modeling_eomt.py:EomtLoss._max_by_axis": ["ModelLoss", "Module", "_max_by_axis", "class", "def", "enumerate", "for", "in", "index", "int", "item", "list", "max", "maxes", "nn", "return", "self", "sizes", "sublist"], "eomt/modeling_eomt.py:EomtLoss._pad_images_to_max_in_batch": ["False", "ModelLoss", "Module", "Tensor", "_", "_max_by_axis", "_pad_images_to_max_in_batch", "batch_shape", "batch_size", "bool", "class", "copy_", "def", "device", "dtype", "for", "height", "in", "len", "list", "max_size", "nn", "ones", "padded_tensor", "padded_tensors", "padding_mask", "padding_masks", "return", "self", "shape", "tensor", "tensors", "torch", "tuple", "width", "zeros", "zip"], "eomt/modeling_eomt.py:EomtLoss.loss_labels": ["CrossEntropyLoss", "ModelLoss", "Module", "Tensor", "_", "_get_predictions_permutation_indices", "array", "batch_size", "cat", "class", "class_labels", "class_queries_logits", "criterion", "def", "device", "dict", "dtype", "empty_weight", "fill_value", "for", "full", "idx", "in", "indices", "int64", "j", "list", "loss_ce", "loss_cross_entropy", "loss_labels", "losses", "nn", "np", "num_labels", "num_queries", "pred_logits", "pred_logits_transposed", "return", "self", "shape", "str", "target", "target_classes", "target_classes_o", "torch", "transpose", "tuple", "weight", "zip"], "eomt/modeling_eomt.py:EomtLoss.loss_masks": ["False", "ModelLoss", "Module", "None", "Tensor", "_", "_get_predictions_permutation_indices", "_get_targets_permutation_indices", "_pad_images_to_max_in_batch", "align_corners", "array", "calculate_uncertainty", "class", "def", "del", "dice_loss", "dict", "importance_sample_ratio", "indices", "int", "lambda", "list", "logits", "loss_dice", "loss_mask", "loss_masks", "losses", "mask_labels", "masks_queries_logits", "nn", "no_grad", "np", "num_masks", "num_points", "oversample_ratio", "point_coordinates", "point_labels", "point_logits", "pred_masks", "return", "sample_point", "sample_points_using_uncertainty", "self", "sigmoid_cross_entropy_loss", "squeeze", "src_idx", "str", "target_masks", "tgt_idx", "torch", "tuple", "with"], "eomt/modeling_eomt.py:EomtLoss._get_predictions_permutation_indices": ["ModelLoss", "Module", "_", "_get_predictions_permutation_indices", "batch_indices", "cat", "class", "def", "enumerate", "for", "full_like", "i", "in", "indices", "nn", "predictions_indices", "return", "self", "src", "torch"], "eomt/modeling_eomt.py:EomtLoss._get_targets_permutation_indices": ["ModelLoss", "Module", "_", "_get_targets_permutation_indices", "batch_indices", "cat", "class", "def", "enumerate", "for", "full_like", "i", "in", "indices", "nn", "return", "self", "target_indices", "tgt", "torch"], "eomt/modeling_eomt.py:EomtLoss.calculate_uncertainty": ["ModelLoss", "Module", "Tensor", "abs", "calculate_uncertainty", "class", "def", "logits", "nn", "return", "self", "torch", "uncertainty_scores"], "eomt/modeling_eomt.py:EomtLoss.sample_points_using_uncertainty": ["False", "ModelLoss", "Module", "None", "Tensor", "align_corners", "arange", "cat", "class", "def", "device", "dim", "dtype", "float", "idx", "if", "importance_sample_ratio", "int", "k", "logits", "long", "nn", "num_boxes", "num_points", "num_points_sampled", "num_random_points", "num_uncertain_points", "oversample_ratio", "point_coordinates", "point_logits", "point_uncertainties", "rand", "return", "sample_point", "sample_points_using_uncertainty", "self", "shape", "shift", "topk", "torch", "uncertainty_function", "view"], "eomt/modeling_eomt.py:EomtLoss.forward": ["ModelLoss", "Module", "None", "Tensor", "_", "aux_outputs", "auxiliary_predictions", "class", "class_labels", "class_queries_logits", "def", "device", "dict", "enumerate", "f", "for", "forward", "get_num_masks", "idx", "if", "in", "indices", "is", "items", "key", "list", "loss_dict", "loss_labels", "loss_masks", "losses", "mask_labels", "masks_queries_logits", "matcher", "nn", "not", "num_masks", "return", "self", "str", "torch", "update", "value"], "eomt/modeling_eomt.py:EomtLoss.get_num_masks": ["ModelLoss", "Module", "PartialState", "Tensor", "_shared_state", "as_tensor", "clamp", "class", "class_labels", "classes", "def", "device", "dtype", "float", "for", "get_num_masks", "if", "in", "is_accelerate_available", "len", "min", "nn", "num_masks", "num_processes", "reduce", "return", "self", "sum", "torch", "world_size"], "eomt/modeling_eomt.py:EomtPatchEmbeddings.__init__": ["Conv2d", "Iterable", "ModelPatchEmbeddings", "Module", "__init__", "abc", "class", "collections", "config", "def", "else", "hidden_size", "if", "image_size", "isinstance", "kernel_size", "nn", "num_channels", "num_patches", "patch_size", "projection", "self", "stride", "super"], "eomt/modeling_eomt.py:EomtPatchEmbeddings.forward": ["Expected", "Make", "ModelPatchEmbeddings", "Module", "Tensor", "ValueError", "but", "channel", "class", "configuration", "def", "dimension", "embeddings", "f", "flatten", "forward", "got", "if", "in", "match", "nn", "num_channels", "of", "one", "pixel", "pixel_values", "projection", "raise", "return", "self", "set", "shape", "sure", "that", "the", "torch", "transpose", "values", "with"], "eomt/modeling_eomt.py:EomtEmbeddings.__init__": ["Dropout", "Embedding", "False", "ModelConfig", "ModelEmbeddings", "ModelPatchEmbeddings", "Module", "None", "Parameter", "__init__", "arange", "class", "cls_token", "config", "def", "dropout", "expand", "hidden_dropout_prob", "hidden_size", "nn", "num_patches", "num_prefix_tokens", "num_register_tokens", "patch_embeddings", "patch_size", "persistent", "position_embeddings", "position_ids", "randn", "register_buffer", "register_tokens", "self", "super", "torch", "zeros"], "eomt/modeling_eomt.py:EomtEmbeddings.forward": ["ModelEmbeddings", "Module", "Tensor", "_", "batch_size", "cat", "class", "cls_token", "cls_tokens", "def", "dim", "dropout", "dtype", "embeddings", "expand", "forward", "nn", "patch_embeddings", "pixel_values", "position_embeddings", "position_ids", "projection", "register_tokens", "return", "self", "shape", "target_dtype", "to", "torch", "weight"], "eomt/modeling_eomt.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "attention_mask", "attn_output", "attn_weights", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "softmax", "to", "torch", "training", "transpose", "value"], "eomt/modeling_eomt.py:EomtAttention.__init__": ["False", "Linear", "ModelAttention", "Module", "ValueError", "__init__", "and", "attention_dropout", "be", "by", "class", "config", "def", "divisible", "dropout", "embed_dim", "f", "got", "head_dim", "hidden_size", "if", "is_causal", "k_proj", "must", "nn", "num_attention_heads", "num_heads", "out_proj", "q_proj", "raise", "scale", "self", "super", "v_proj"], "eomt/modeling_eomt.py:EomtAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "ModelAttention", "Module", "None", "Tensor", "_attn_implementation", "attention_interface", "attention_mask", "attn_output", "attn_weights", "batch_size", "class", "config", "contiguous", "def", "dropout", "eager", "eager_attention_forward", "else", "embed_dim", "forward", "head_dim", "hidden_states", "if", "is_causal", "k_proj", "keys", "kwargs", "nn", "not", "num_heads", "out_proj", "q_proj", "queries", "reshape", "return", "scale", "scaling", "self", "seq_length", "shape", "torch", "training", "transpose", "tuple", "v_proj", "values", "view"], "eomt/modeling_eomt.py:EomtLayerScale.__init__": ["ModelLayerScale", "Module", "None", "Parameter", "__init__", "class", "config", "def", "hidden_size", "lambda1", "layerscale_value", "nn", "ones", "self", "super", "torch"], "eomt/modeling_eomt.py:EomtLayerScale.forward": ["ModelLayerScale", "Module", "Tensor", "class", "def", "forward", "hidden_state", "lambda1", "nn", "return", "self", "torch"], "eomt/modeling_eomt.py:drop_path": ["False", "Model_path", "Model_prob", "Tensor", "bool", "def", "device", "div", "dtype", "float", "floor_", "if", "input", "keep_prob", "ndim", "not", "or", "output", "rand", "random_tensor", "return", "shape", "torch", "training"], "eomt/modeling_eomt.py:EomtDropPath.__init__": ["ModelDropPath", "Module", "None", "__init__", "class", "def", "drop_prob", "float", "nn", "self", "super"], "eomt/modeling_eomt.py:EomtDropPath.forward": ["ModelDropPath", "Module", "Tensor", "class", "def", "drop_path", "drop_prob", "forward", "hidden_states", "nn", "return", "self", "torch", "training"], "eomt/modeling_eomt.py:EomtDropPath.extra_repr": ["ModelDropPath", "Module", "class", "def", "drop_prob", "extra_repr", "f", "nn", "p", "return", "self", "str"], "eomt/modeling_eomt.py:EomtMLP.__init__": ["ACT2FN", "Linear", "ModelMLP", "Module", "None", "True", "__init__", "activation", "bias", "class", "config", "def", "else", "fc1", "fc2", "hidden_act", "hidden_features", "hidden_size", "if", "in_features", "int", "isinstance", "mlp_ratio", "nn", "out_features", "self", "str", "super"], "eomt/modeling_eomt.py:EomtMLP.forward": ["ModelMLP", "Module", "Tensor", "activation", "class", "def", "fc1", "fc2", "forward", "hidden_state", "nn", "return", "self", "torch"], "eomt/modeling_eomt.py:EomtSwiGLUFFN.__init__": ["Linear", "ModelSwiGLUFFN", "Module", "None", "True", "__init__", "bias", "class", "config", "def", "hidden_features", "hidden_size", "in_features", "int", "mlp_ratio", "nn", "out_features", "self", "super", "weights_in", "weights_out"], "eomt/modeling_eomt.py:EomtSwiGLUFFN.forward": ["ModelSwiGLUFFN", "Module", "Tensor", "chunk", "class", "def", "dim", "forward", "functional", "hidden", "hidden_state", "nn", "return", "self", "silu", "torch", "weights_in", "weights_out", "x1", "x2"], "eomt/modeling_eomt.py:EomtLayer.__init__": ["GradientCheckpointingLayer", "Identity", "LayerNorm", "ModelAttention", "ModelConfig", "ModelDropPath", "ModelLayer", "ModelLayerScale", "ModelMLP", "ModelSwiGLUFFN", "None", "__init__", "attention", "class", "config", "def", "drop_path", "drop_path_rate", "else", "eps", "hidden_size", "if", "layer_norm_eps", "layer_scale1", "layer_scale2", "mlp", "nn", "norm1", "norm2", "self", "super", "use_swiglu_ffn"], "eomt/modeling_eomt.py:EomtLayer.forward": ["GradientCheckpointingLayer", "ModelLayer", "None", "Tensor", "_", "attention", "attention_mask", "class", "def", "drop_path", "forward", "hidden_states", "hidden_states_norm", "layer_output", "layer_scale1", "layer_scale2", "mlp", "norm1", "norm2", "return", "self", "self_attention_output", "torch"], "eomt/modeling_eomt.py:EomtLayerNorm2d.__init__": ["LayerNorm", "ModelLayerNorm2d", "True", "__init__", "affine", "class", "def", "elementwise_affine", "eps", "nn", "num_channels", "self", "super"], "eomt/modeling_eomt.py:EomtLayerNorm2d.forward": ["F", "LayerNorm", "ModelLayerNorm2d", "Tensor", "bias", "class", "def", "eps", "forward", "hidden_state", "layer_norm", "nn", "normalized_shape", "permute", "return", "self", "torch", "weight"], "eomt/modeling_eomt.py:EomtScaleLayer.__init__": ["ACT2FN", "Conv2d", "ConvTranspose2d", "False", "ModelConfig", "ModelLayerNorm2d", "ModelScaleLayer", "Module", "__init__", "activation", "bias", "class", "config", "conv1", "conv2", "def", "groups", "hidden_act", "hidden_size", "kernel_size", "layernorm2d", "nn", "padding", "self", "stride", "super"], "eomt/modeling_eomt.py:EomtScaleLayer.forward": ["ModelScaleLayer", "Module", "Tensor", "activation", "class", "conv1", "conv2", "def", "forward", "hidden_states", "layernorm2d", "nn", "return", "self", "torch"], "eomt/modeling_eomt.py:EomtScaleBlock.__init__": ["ModelConfig", "ModelScaleBlock", "ModelScaleLayer", "Module", "ModuleList", "_", "__init__", "block", "class", "config", "def", "for", "in", "nn", "num_blocks", "num_upscale_blocks", "range", "self", "super"], "eomt/modeling_eomt.py:EomtScaleBlock.forward": ["ModelScaleBlock", "Module", "Tensor", "block", "class", "def", "for", "forward", "hidden_states", "in", "nn", "return", "self", "torch"], "eomt/modeling_eomt.py:EomtMaskHead.__init__": ["ACT2FN", "Linear", "ModelConfig", "ModelMaskHead", "Module", "__init__", "activation", "class", "config", "def", "fc1", "fc2", "fc3", "hidden_act", "hidden_size", "nn", "self", "super"], "eomt/modeling_eomt.py:EomtMaskHead.forward": ["ModelMaskHead", "Module", "Tensor", "activation", "class", "def", "fc1", "fc2", "fc3", "forward", "hidden_states", "nn", "return", "self", "torch"], "eomt/modeling_eomt.py:EomtPreTrainedModel._init_weights": ["Conv2d", "ConvTranspose2d", "Embedding", "False", "LayerNorm", "Linear", "ModelEmbeddings", "ModelForUniversalSegmentation", "ModelLayerScale", "ModelLoss", "ModelPreTrainedModel", "Module", "None", "PreTrainedModel", "_", "_calculate_fan_in_and_fan_out", "_init_weights", "_is_hf_initialized", "a", "and", "arange", "attn_mask_probs", "bias", "bound", "class", "cls_token", "config", "constant_", "copy_", "def", "elif", "else", "empty_weight", "eos_coef", "expand", "fan_in", "getattr", "hasattr", "if", "init", "initializer_range", "is", "isinstance", "kaiming_uniform_", "lambda1", "layerscale_value", "math", "mean", "module", "nn", "normal_", "not", "num_labels", "ones", "ones_", "padding_idx", "position_ids", "register_tokens", "self", "shape", "sqrt", "std", "torch", "trunc_normal_", "uniform_", "weight", "zeros_"], "eomt/modeling_eomt.py:EomtForUniversalSegmentation.__init__": ["Embedding", "LayerNorm", "Linear", "ModelConfig", "ModelEmbeddings", "ModelForUniversalSegmentation", "ModelLayer", "ModelLoss", "ModelMaskHead", "ModelPreTrainedModel", "ModelScaleBlock", "ModuleList", "_", "__init__", "attn_mask_probs", "class", "class_predictor", "class_weight", "config", "criterion", "def", "dice_weight", "dict", "embeddings", "eps", "float", "for", "grid_size", "hidden_size", "image_size", "in", "layer_norm_eps", "layernorm", "layers", "loss_cross_entropy", "loss_dice", "loss_mask", "mask_head", "mask_weight", "nn", "num_blocks", "num_hidden_layers", "num_labels", "num_queries", "ones", "patch_size", "post_init", "query", "range", "register_buffer", "self", "str", "super", "torch", "upscale_block", "weight_dict"], "eomt/modeling_eomt.py:EomtForUniversalSegmentation.get_loss_dict": ["ModelForUniversalSegmentation", "ModelPreTrainedModel", "Tensor", "auxiliary_predictions", "class", "class_labels", "class_queries_logits", "criterion", "def", "dict", "for", "get_loss_dict", "if", "in", "items", "key", "loss", "loss_dict", "loss_key", "mask_labels", "masks_queries_logits", "return", "self", "str", "weight", "weight_dict"], "eomt/modeling_eomt.py:EomtForUniversalSegmentation.get_loss": ["ModelForUniversalSegmentation", "ModelPreTrainedModel", "Tensor", "class", "def", "dict", "get_loss", "loss_dict", "return", "self", "str", "sum", "values"], "eomt/modeling_eomt.py:EomtForUniversalSegmentation.forward": ["F", "ModelForUniversalSegmentation", "ModelForUniversalSegmentationOutput", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "_disable_attention_mask", "and", "attention_mask", "attn_mask_probs", "auxiliary_predictions", "bilinear", "bool", "cat", "class", "class_labels", "class_queries_logits", "class_queries_logits_per_layer", "config", "def", "device", "dim", "dtype", "embeddings", "encoder_start_tokens", "enumerate", "expand", "float", "for", "forward", "get_loss", "get_loss_dict", "grid_size", "have", "hidden_states", "idx", "if", "in", "interpolate", "interpolated_logits", "is", "kwargs", "last_hidden_state", "layer_module", "layernorm", "layers", "list", "loss", "loss_dict", "mask_labels", "masked_fill", "masks_queries_logits", "masks_queries_logits_per_layer", "mode", "norm_hidden_states", "not", "num_attention_heads", "num_blocks", "num_hidden_layers", "num_prefix_tokens", "num_queries", "num_query_tokens", "ones", "or", "patch_offsets", "pixel_values", "predict", "prob", "query", "r", "raise", "return", "self", "sequence_output", "shape", "size", "specify", "to", "torch", "training", "view", "weight", "zip"], "eomt/modeling_eomt.py:EomtForUniversalSegmentation.get_input_embeddings": ["ModelForUniversalSegmentation", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "patch_embeddings", "return", "self"], "eomt/modeling_eomt.py:EomtForUniversalSegmentation.predict": ["ModelForUniversalSegmentation", "ModelPreTrainedModel", "Tensor", "bchw", "bqc", "bqhw", "class", "class_logits", "class_predictor", "config", "def", "einsum", "embeddings", "grid_size", "logits", "mask_head", "mask_logits", "num_prefix_tokens", "num_queries", "predict", "prefix_tokens", "query_tokens", "reshape", "return", "self", "shape", "torch", "transpose", "upscale_block"], "eomt/modeling_eomt.py:EomtForUniversalSegmentation._disable_attention_mask": ["ModelForUniversalSegmentation", "ModelPreTrainedModel", "_disable_attention_mask", "attn_mask", "class", "def", "device", "encoder_start_tokens", "if", "num_query_tokens", "prob", "rand", "random_queries", "return", "shape", "torch"], "ernie/modeling_ernie.py:ErnieEmbeddings.__init__": ["Dropout", "Embedding", "False", "LayerNorm", "ModelEmbeddings", "Module", "__init__", "arange", "class", "config", "def", "dropout", "dtype", "eps", "expand", "hidden_dropout_prob", "hidden_size", "if", "layer_norm_eps", "long", "max_position_embeddings", "nn", "pad_token_id", "padding_idx", "persistent", "position_embeddings", "position_ids", "register_buffer", "self", "size", "super", "task_type_embeddings", "task_type_vocab_size", "token_type_embeddings", "token_type_ids", "torch", "type_vocab_size", "use_task_id", "vocab_size", "word_embeddings", "zeros"], "ernie/modeling_ernie.py:ErnieEmbeddings.forward": ["FloatTensor", "LayerNorm", "LongTensor", "ModelEmbeddings", "Module", "None", "Tensor", "batch_size", "buffered_token_type_ids", "class", "def", "device", "dim", "dropout", "dtype", "else", "embeddings", "expand", "forward", "gather", "hasattr", "if", "index", "input_ids", "input_shape", "inputs_embeds", "int", "is", "long", "nn", "not", "past_key_values_length", "position_embeddings", "position_ids", "return", "self", "seq_length", "shape", "size", "task_type_embeddings", "task_type_ids", "to", "token_type_embeddings", "token_type_ids", "torch", "use_task_id", "word_embeddings", "zeros"], "ernie/modeling_ernie.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "contiguous", "def", "dim", "dropout", "float", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "shape", "size", "softmax", "torch", "training", "transpose", "value"], "ernie/modeling_ernie.py:ErnieSelfAttention.__init__": ["Dropout", "False", "Linear", "ModelSelfAttention", "Module", "None", "The", "ValueError", "__init__", "a", "all_head_size", "and", "attention", "attention_head_size", "attention_probs_dropout_prob", "class", "config", "def", "dropout", "embedding_size", "f", "hasattr", "heads", "hidden", "hidden_size", "if", "int", "is", "is_causal", "is_decoder", "key", "layer_idx", "multiple", "nn", "not", "num_attention_heads", "number", "of", "query", "raise", "scaling", "self", "size", "super", "the", "value"], "ernie/modeling_ernie.py:ErnieSelfAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "EncoderDecoderCache", "FloatTensor", "ModelSelfAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_attn_implementation", "attention_head_size", "attention_interface", "attention_mask", "attn_output", "attn_weights", "cache_position", "class", "config", "contiguous", "current_past_key_values", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "hidden_shape", "hidden_states", "if", "input_shape", "is", "isinstance", "key", "key_layer", "kwargs", "layer_idx", "nn", "not", "p", "past_key_values", "query", "query_layer", "reshape", "return", "scaling", "self", "self_attention_cache", "shape", "torch", "training", "transpose", "tuple", "update", "value", "value_layer", "view"], "ernie/modeling_ernie.py:ErnieCrossAttention.__init__": ["Dropout", "False", "Linear", "ModelCrossAttention", "Module", "None", "The", "ValueError", "__init__", "a", "all_head_size", "and", "attention", "attention_head_size", "attention_probs_dropout_prob", "class", "config", "def", "dropout", "embedding_size", "f", "hasattr", "heads", "hidden", "hidden_size", "if", "int", "is", "is_causal", "key", "layer_idx", "multiple", "nn", "not", "num_attention_heads", "number", "of", "query", "raise", "scaling", "self", "size", "super", "the", "value"], "ernie/modeling_ernie.py:ErnieCrossAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "EncoderDecoderCache", "False", "FloatTensor", "ModelCrossAttention", "Module", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "_attn_implementation", "and", "attention_head_size", "attention_interface", "attention_mask", "attn_output", "attn_weights", "bsz", "class", "config", "contiguous", "cross_attention_cache", "def", "dropout", "eager", "eager_attention_forward", "else", "encoder_hidden_states", "forward", "get", "hidden_states", "if", "is", "is_updated", "key", "key_layer", "keys", "kv_input_shape", "kwargs", "layer_idx", "layers", "nn", "not", "p", "past_key_values", "q_input_shape", "query", "query_layer", "reshape", "return", "scaling", "self", "shape", "src_len", "tgt_len", "torch", "training", "transpose", "tuple", "update", "value", "value_layer", "values", "view"], "ernie/modeling_ernie.py:ErnieSelfOutput.__init__": ["Dropout", "LayerNorm", "Linear", "ModelSelfOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "eps", "hidden_dropout_prob", "hidden_size", "layer_norm_eps", "nn", "self", "super"], "ernie/modeling_ernie.py:ErnieSelfOutput.forward": ["LayerNorm", "ModelSelfOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "ernie/modeling_ernie.py:ErnieAttention.__init__": ["False", "ModelAttention", "ModelCrossAttention", "ModelSelfAttention", "ModelSelfOutput", "Module", "None", "__init__", "attention_class", "class", "config", "def", "else", "if", "is_causal", "is_cross_attention", "layer_idx", "nn", "output", "self", "super"], "ernie/modeling_ernie.py:ErnieAttention.forward": ["Cache", "FloatTensor", "ModelAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attention_output", "attn_weights", "cache_position", "class", "def", "else", "encoder_attention_mask", "encoder_hidden_states", "forward", "hidden_states", "if", "is_cross_attention", "kwargs", "nn", "not", "output", "past_key_values", "return", "self", "torch", "tuple"], "ernie/modeling_ernie.py:ErnieIntermediate.__init__": ["ACT2FN", "Linear", "ModelIntermediate", "Module", "__init__", "class", "config", "def", "dense", "else", "hidden_act", "hidden_size", "if", "intermediate_act_fn", "intermediate_size", "isinstance", "nn", "self", "str", "super"], "ernie/modeling_ernie.py:ErnieIntermediate.forward": ["ModelIntermediate", "Module", "Tensor", "class", "def", "dense", "forward", "hidden_states", "intermediate_act_fn", "nn", "return", "self", "torch"], "ernie/modeling_ernie.py:ErnieOutput.__init__": ["Dropout", "LayerNorm", "Linear", "ModelOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "eps", "hidden_dropout_prob", "hidden_size", "intermediate_size", "layer_norm_eps", "nn", "self", "super"], "ernie/modeling_ernie.py:ErnieOutput.forward": ["LayerNorm", "ModelOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "ernie/modeling_ernie.py:ErnieLayer.__init__": ["False", "GradientCheckpointingLayer", "ModelAttention", "ModelIntermediate", "ModelLayer", "ModelOutput", "None", "True", "ValueError", "__init__", "a", "add_cross_attention", "added", "as", "attention", "be", "chunk_size_feed_forward", "class", "config", "cross", "crossattention", "decoder", "def", "f", "if", "intermediate", "is", "is_causal", "is_cross_attention", "is_decoder", "layer_idx", "model", "not", "output", "raise", "self", "seq_len_dim", "should", "super", "used"], "ernie/modeling_ernie.py:ErnieLayer.forward": ["Cache", "FloatTensor", "GradientCheckpointingLayer", "If", "ModelLayer", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "ValueError", "_", "add_cross_attention", "and", "apply_chunking_to_forward", "are", "attention", "attention_mask", "attention_output", "be", "by", "cache_position", "chunk_size_feed_forward", "class", "config", "cross", "cross_attention_output", "crossattention", "def", "encoder_attention_mask", "encoder_hidden_states", "f", "feed_forward_chunk", "forward", "has", "hasattr", "hidden_states", "if", "instantiated", "is", "is_decoder", "kwargs", "layer_output", "layers", "not", "passed", "past_key_values", "raise", "return", "self", "self_attention_output", "seq_len_dim", "setting", "to", "torch", "tuple", "with"], "ernie/modeling_ernie.py:ErnieLayer.feed_forward_chunk": ["GradientCheckpointingLayer", "ModelLayer", "attention_output", "class", "def", "feed_forward_chunk", "intermediate", "intermediate_output", "layer_output", "output", "return", "self"], "ernie/modeling_ernie.py:ErniePooler.__init__": ["Linear", "ModelPooler", "Module", "Tanh", "__init__", "activation", "class", "config", "def", "dense", "hidden_size", "nn", "self", "super"], "ernie/modeling_ernie.py:ErniePooler.forward": ["ModelPooler", "Module", "Tensor", "activation", "class", "def", "dense", "first_token_tensor", "forward", "hidden_states", "nn", "pooled_output", "return", "self", "torch"], "ernie/modeling_ernie.py:ErniePredictionHeadTransform.__init__": ["ACT2FN", "LayerNorm", "Linear", "ModelPredictionHeadTransform", "Module", "__init__", "class", "config", "def", "dense", "else", "eps", "hidden_act", "hidden_size", "if", "isinstance", "layer_norm_eps", "nn", "self", "str", "super", "transform_act_fn"], "ernie/modeling_ernie.py:ErniePredictionHeadTransform.forward": ["LayerNorm", "ModelPredictionHeadTransform", "Module", "Tensor", "class", "def", "dense", "forward", "hidden_states", "nn", "return", "self", "torch", "transform_act_fn"], "ernie/modeling_ernie.py:ErnieLMPredictionHead.__init__": ["Linear", "ModelLMPredictionHead", "ModelPredictionHeadTransform", "Module", "Parameter", "True", "__init__", "bias", "class", "config", "decoder", "def", "hidden_size", "nn", "self", "super", "torch", "transform", "vocab_size", "zeros"], "ernie/modeling_ernie.py:ErnieLMPredictionHead.forward": ["ModelLMPredictionHead", "Module", "class", "decoder", "def", "forward", "hidden_states", "nn", "return", "self", "transform"], "ernie/modeling_ernie.py:ErnieEncoder.__init__": ["ModelEncoder", "ModelLayer", "Module", "ModuleList", "__init__", "class", "config", "def", "for", "i", "in", "layer", "layer_idx", "nn", "num_hidden_layers", "range", "self", "super"], "ernie/modeling_ernie.py:ErnieEncoder.forward": ["BaseModelOutputWithPastAndCrossAttentions", "Cache", "FloatTensor", "ModelEncoder", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "bool", "cache_position", "class", "def", "else", "encoder_attention_mask", "encoder_hidden_states", "enumerate", "for", "forward", "hidden_states", "i", "if", "in", "kwargs", "last_hidden_state", "layer", "layer_module", "nn", "past_key_values", "return", "self", "torch", "tuple", "use_cache"], "ernie/modeling_ernie.py:ErniePreTrainedModel._init_weights": ["ModelEmbeddings", "ModelLMPredictionHead", "ModelPreTrainedModel", "PreTrainedModel", "_init_weights", "arange", "bias", "class", "copy_", "def", "elif", "expand", "if", "init", "isinstance", "module", "position_ids", "self", "shape", "super", "token_type_ids", "torch", "zeros_"], "ernie/modeling_ernie.py:ErnieModel.__init__": ["False", "ModelEmbeddings", "ModelEncoder", "ModelModel", "ModelPooler", "ModelPreTrainedModel", "None", "True", "__init__", "add_pooling_layer", "class", "config", "def", "else", "embeddings", "encoder", "gradient_checkpointing", "if", "pooler", "post_init", "r", "self", "super"], "ernie/modeling_ernie.py:ErnieModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "return", "self", "word_embeddings"], "ernie/modeling_ernie.py:ErnieModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "self", "set_input_embeddings", "value", "word_embeddings"], "ernie/modeling_ernie.py:ErnieModel.forward": ["BaseModelOutputWithPoolingAndCrossAttentions", "Cache", "DynamicCache", "EncoderDecoderCache", "False", "ModelModel", "ModelPreTrainedModel", "None", "Setting", "Tensor", "TransformersKwargs", "True", "Unpack", "ValueError", "You", "_create_attention_masks", "and", "arange", "at", "attention_mask", "batch_size", "bool", "both", "cache_position", "cannot", "checkpointing", "class", "config", "def", "device", "either", "elif", "else", "embedding_output", "embeddings", "encoder", "encoder_attention_mask", "encoder_hidden_states", "encoder_outputs", "forward", "get_seq_length", "gradient", "gradient_checkpointing", "have", "if", "incompatible", "input_ids", "input_shape", "inputs_embeds", "is", "is_decoder", "is_encoder_decoder", "kwargs", "last_hidden_state", "logger", "not", "or", "past_key_values", "past_key_values_length", "pooled_output", "pooler", "pooler_output", "position_ids", "r", "raise", "return", "same", "self", "seq_length", "sequence_output", "size", "specify", "task_type_ids", "the", "time", "to", "token_type_ids", "torch", "training", "tuple", "use_cache", "warn_if_padding_and_no_attention_mask", "warning_once", "with"], "ernie/modeling_ernie.py:ErnieModel._create_attention_masks": ["ModelModel", "ModelPreTrainedModel", "None", "_create_attention_masks", "attention_mask", "cache_position", "class", "config", "create_bidirectional_mask", "create_causal_mask", "def", "else", "embedding_output", "encoder_attention_mask", "encoder_hidden_states", "if", "input_embeds", "is", "is_decoder", "not", "past_key_values", "return", "self"], "ernie/modeling_ernie.py:ErniePreTrainingHeads.__init__": ["Linear", "ModelLMPredictionHead", "ModelPreTrainingHeads", "Module", "__init__", "class", "config", "def", "hidden_size", "nn", "predictions", "self", "seq_relationship", "super"], "ernie/modeling_ernie.py:ErniePreTrainingHeads.forward": ["ModelPreTrainingHeads", "Module", "class", "def", "forward", "nn", "pooled_output", "prediction_scores", "predictions", "return", "self", "seq_relationship", "seq_relationship_score", "sequence_output"], "ernie/modeling_ernie.py:ErnieForPreTraining.__init__": ["Model", "ModelForPreTraining", "ModelModel", "ModelPreTrainedModel", "ModelPreTrainingHeads", "__init__", "class", "cls", "config", "def", "post_init", "self", "super"], "ernie/modeling_ernie.py:ErnieForPreTraining.get_output_embeddings": ["ModelForPreTraining", "ModelPreTrainedModel", "class", "cls", "decoder", "def", "get_output_embeddings", "predictions", "return", "self"], "ernie/modeling_ernie.py:ErnieForPreTraining.set_output_embeddings": ["ModelForPreTraining", "ModelPreTrainedModel", "bias", "class", "cls", "decoder", "def", "new_embeddings", "predictions", "self", "set_output_embeddings"], "ernie/modeling_ernie.py:ErnieForPreTraining.forward": ["CrossEntropyLoss", "Model", "ModelForPreTraining", "ModelForPreTrainingOutput", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "and", "attention_mask", "attentions", "class", "cls", "config", "def", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "loss", "loss_fct", "masked_lm_loss", "next_sentence_label", "next_sentence_loss", "not", "outputs", "pooled_output", "position_ids", "prediction_logits", "prediction_scores", "r", "return", "return_dict", "self", "seq_relationship_logits", "seq_relationship_score", "sequence_output", "task_type_ids", "token_type_ids", "torch", "total_loss", "tuple", "view", "vocab_size"], "ernie/modeling_ernie.py:ErnieOnlyMLMHead.__init__": ["ModelLMPredictionHead", "ModelOnlyMLMHead", "Module", "__init__", "class", "config", "def", "nn", "predictions", "self", "super"], "ernie/modeling_ernie.py:ErnieOnlyMLMHead.forward": ["ModelOnlyMLMHead", "Module", "Tensor", "class", "def", "forward", "nn", "prediction_scores", "predictions", "return", "self", "sequence_output", "torch"], "ernie/modeling_ernie.py:ErnieForCausalLM.__init__": ["False", "GenerationMixin", "If", "Model", "ModelForCausalLM", "ModelModel", "ModelOnlyMLMHead", "ModelPreTrainedModel", "True", "__init__", "a", "add", "add_pooling_layer", "as", "class", "cls", "config", "def", "if", "is_decoder", "logger", "not", "post_init", "self", "standalone", "super", "to", "use", "want", "warning", "you"], "ernie/modeling_ernie.py:ErnieForCausalLM.get_output_embeddings": ["GenerationMixin", "ModelForCausalLM", "ModelPreTrainedModel", "class", "cls", "decoder", "def", "get_output_embeddings", "predictions", "return", "self"], "ernie/modeling_ernie.py:ErnieForCausalLM.set_output_embeddings": ["GenerationMixin", "ModelForCausalLM", "ModelPreTrainedModel", "bias", "class", "cls", "decoder", "def", "new_embeddings", "predictions", "self", "set_output_embeddings"], "ernie/modeling_ernie.py:ErnieForCausalLM.forward": ["BaseModelOutputWithPoolingAndCrossAttentions", "CausalLMOutputWithCrossAttentions", "False", "GenerationMixin", "Model", "ModelForCausalLM", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "attention_mask", "attentions", "bool", "cache_position", "class", "cls", "config", "cross_attentions", "def", "else", "encoder_attention_mask", "encoder_hidden_states", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "list", "logits", "logits_to_keep", "loss", "loss_function", "not", "outputs", "past_key_values", "position_ids", "r", "return", "return_dict", "self", "slice", "slice_indices", "task_type_ids", "token_type_ids", "torch", "tuple", "use_cache", "vocab_size"], "ernie/modeling_ernie.py:ErnieForMaskedLM.__init__": ["False", "If", "Model", "ModelForMaskedLM", "ModelModel", "ModelOnlyMLMHead", "ModelPreTrainedModel", "__init__", "add_pooling_layer", "attention", "bi", "class", "cls", "config", "def", "directional", "for", "if", "is_decoder", "logger", "make", "post_init", "self", "super", "sure", "to", "use", "want", "warning", "you"], "ernie/modeling_ernie.py:ErnieForMaskedLM.get_output_embeddings": ["ModelForMaskedLM", "ModelPreTrainedModel", "class", "cls", "decoder", "def", "get_output_embeddings", "predictions", "return", "self"], "ernie/modeling_ernie.py:ErnieForMaskedLM.set_output_embeddings": ["ModelForMaskedLM", "ModelPreTrainedModel", "bias", "class", "cls", "decoder", "def", "new_embeddings", "predictions", "self", "set_output_embeddings"], "ernie/modeling_ernie.py:ErnieForMaskedLM.forward": ["CrossEntropyLoss", "MaskedLMOutput", "Model", "ModelForMaskedLM", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "attention_mask", "attentions", "class", "cls", "config", "def", "encoder_attention_mask", "encoder_hidden_states", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "masked_lm_loss", "not", "outputs", "position_ids", "prediction_scores", "r", "return", "return_dict", "self", "sequence_output", "task_type_ids", "token_type_ids", "torch", "tuple", "view", "vocab_size"], "ernie/modeling_ernie.py:ErnieForMaskedLM.prepare_inputs_for_generation": ["ModelForMaskedLM", "ModelPreTrainedModel", "None", "PAD", "The", "ValueError", "attention_mask", "be", "cat", "class", "config", "def", "defined", "device", "dim", "dtype", "dummy_token", "effective_batch_size", "for", "full", "generation", "if", "input_ids", "input_shape", "is", "long", "model_kwargs", "new_zeros", "pad_token_id", "prepare_inputs_for_generation", "raise", "return", "self", "shape", "should", "token", "torch"], "ernie/modeling_ernie.py:ErnieForMaskedLM.can_generate": ["False", "ModelForMaskedLM", "ModelPreTrainedModel", "bool", "can_generate", "class", "cls", "def", "return"], "ernie/modeling_ernie.py:ErnieOnlyNSPHead.__init__": ["Linear", "ModelOnlyNSPHead", "Module", "__init__", "class", "config", "def", "hidden_size", "nn", "self", "seq_relationship", "super"], "ernie/modeling_ernie.py:ErnieOnlyNSPHead.forward": ["ModelOnlyNSPHead", "Module", "class", "def", "forward", "nn", "pooled_output", "return", "self", "seq_relationship", "seq_relationship_score"], "ernie/modeling_ernie.py:ErnieForNextSentencePrediction.__init__": ["Model", "ModelForNextSentencePrediction", "ModelModel", "ModelOnlyNSPHead", "ModelPreTrainedModel", "__init__", "class", "cls", "config", "def", "post_init", "self", "super"], "ernie/modeling_ernie.py:ErnieForNextSentencePrediction.forward": ["CrossEntropyLoss", "Model", "ModelForNextSentencePrediction", "ModelPreTrainedModel", "NextSentencePredictorOutput", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "attention_mask", "attentions", "class", "cls", "def", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "next_sentence_loss", "not", "outputs", "pooled_output", "position_ids", "r", "return", "return_dict", "self", "seq_relationship_scores", "task_type_ids", "token_type_ids", "torch", "tuple", "view"], "ernie/modeling_ernie.py:ErnieForSequenceClassification.__init__": ["Dropout", "Linear", "Model", "ModelForSequenceClassification", "ModelModel", "ModelPreTrainedModel", "None", "__init__", "class", "classifier", "classifier_dropout", "config", "def", "dropout", "else", "hidden_dropout_prob", "hidden_size", "if", "is", "nn", "not", "num_labels", "post_init", "self", "super"], "ernie/modeling_ernie.py:ErnieForSequenceClassification.forward": ["BCEWithLogitsLoss", "CrossEntropyLoss", "MSELoss", "Model", "ModelForSequenceClassification", "ModelPreTrainedModel", "None", "SequenceClassifierOutput", "Tensor", "TransformersKwargs", "True", "Unpack", "and", "attention_mask", "attentions", "class", "classifier", "config", "def", "dropout", "dtype", "elif", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "kwargs", "labels", "logits", "long", "loss", "loss_fct", "multi_label_classification", "not", "num_labels", "or", "outputs", "pooled_output", "position_ids", "problem_type", "r", "regression", "return", "return_dict", "self", "single_label_classification", "squeeze", "task_type_ids", "token_type_ids", "torch", "tuple", "view"], "ernie/modeling_ernie.py:ErnieForMultipleChoice.__init__": ["Dropout", "Linear", "Model", "ModelForMultipleChoice", "ModelModel", "ModelPreTrainedModel", "None", "__init__", "class", "classifier", "classifier_dropout", "config", "def", "dropout", "else", "hidden_dropout_prob", "hidden_size", "if", "is", "nn", "not", "post_init", "self", "super"], "ernie/modeling_ernie.py:ErnieForMultipleChoice.forward": ["CrossEntropyLoss", "Model", "ModelForMultipleChoice", "ModelPreTrainedModel", "MultipleChoiceModelOutput", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "attention_mask", "attentions", "class", "classifier", "def", "dropout", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "not", "num_choices", "outputs", "pooled_output", "position_ids", "r", "reshaped_logits", "return", "return_dict", "self", "shape", "size", "task_type_ids", "token_type_ids", "torch", "tuple", "view"], "ernie/modeling_ernie.py:ErnieForTokenClassification.__init__": ["Dropout", "False", "Linear", "Model", "ModelForTokenClassification", "ModelModel", "ModelPreTrainedModel", "None", "__init__", "add_pooling_layer", "class", "classifier", "classifier_dropout", "config", "def", "dropout", "else", "hidden_dropout_prob", "hidden_size", "if", "is", "nn", "not", "num_labels", "post_init", "self", "super"], "ernie/modeling_ernie.py:ErnieForTokenClassification.forward": ["CrossEntropyLoss", "Model", "ModelForTokenClassification", "ModelPreTrainedModel", "None", "Tensor", "TokenClassifierOutput", "TransformersKwargs", "True", "Unpack", "attention_mask", "attentions", "class", "classifier", "def", "dropout", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "not", "num_labels", "outputs", "position_ids", "r", "return", "return_dict", "self", "sequence_output", "task_type_ids", "token_type_ids", "torch", "tuple", "view"], "ernie/modeling_ernie.py:ErnieForQuestionAnswering.__init__": ["False", "Linear", "Model", "ModelForQuestionAnswering", "ModelModel", "ModelPreTrainedModel", "__init__", "add_pooling_layer", "class", "config", "def", "hidden_size", "nn", "num_labels", "post_init", "qa_outputs", "self", "super"], "ernie/modeling_ernie.py:ErnieForQuestionAnswering.forward": ["CrossEntropyLoss", "Model", "ModelForQuestionAnswering", "ModelPreTrainedModel", "None", "QuestionAnsweringModelOutput", "Tensor", "TransformersKwargs", "True", "Unpack", "and", "attention_mask", "attentions", "clamp", "class", "contiguous", "def", "dim", "end_logits", "end_loss", "end_positions", "forward", "hidden_states", "if", "ignore_index", "ignored_index", "input_ids", "inputs_embeds", "is", "kwargs", "len", "logits", "loss", "loss_fct", "not", "outputs", "position_ids", "qa_outputs", "r", "return", "return_dict", "self", "sequence_output", "size", "split", "squeeze", "start_logits", "start_loss", "start_positions", "task_type_ids", "token_type_ids", "torch", "total_loss", "tuple"], "ernie4_5/modeling_ernie4_5.py:Ernie4_5RotaryEmbedding.__init__": ["Callable", "False", "ModelConfig", "ModelRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "ernie4_5/modeling_ernie4_5.py:Ernie4_5RotaryEmbedding.compute_default_rope_parameters": ["ModelConfig", "ModelRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "ernie4_5/modeling_ernie4_5.py:Ernie4_5RotaryEmbedding.forward": ["False", "ModelRotaryEmbedding", "Module", "None", "and", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "ernie4_5/modeling_ernie4_5.py:Ernie4_5MLP.__init__": ["ACT2FN", "Linear", "ModelConfig", "ModelMLP", "Module", "__init__", "act_fn", "bias", "class", "config", "def", "down_proj", "gate_proj", "hidden_act", "hidden_size", "intermediate_size", "nn", "self", "super", "up_proj", "use_bias"], "ernie4_5/modeling_ernie4_5.py:Ernie4_5MLP.forward": ["ModelMLP", "Module", "act_fn", "class", "def", "down_proj", "forward", "gate_proj", "nn", "return", "self", "up_proj", "x"], "ernie4_5/modeling_ernie4_5.py:rotate_half": ["Model_half", "def", "dim", "flatten", "return", "stack", "torch", "x", "x1", "x2"], "ernie4_5/modeling_ernie4_5.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "ernie4_5/modeling_ernie4_5.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value", "value_states"], "ernie4_5/modeling_ernie4_5.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cos", "def", "dim", "dtype", "float", "k", "k_embed", "original_dtype", "q", "q_embed", "repeat_interleave", "return", "rotate_half", "shape", "sin", "to", "unsqueeze", "unsqueeze_dim"], "ernie4_5/modeling_ernie4_5.py:Ernie4_5Attention.__init__": ["Linear", "ModelAttention", "ModelConfig", "Module", "True", "__init__", "attention_dropout", "bias", "class", "config", "def", "getattr", "head_dim", "hidden_size", "int", "is_causal", "k_proj", "layer_idx", "nn", "num_attention_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "q_proj", "scaling", "self", "super", "use_bias", "v_proj"], "ernie4_5/modeling_ernie4_5.py:Ernie4_5Attention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "LongTensor", "ModelAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_attn_implementation", "apply_rotary_pos_emb", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "cache_kwargs", "cache_position", "class", "config", "contiguous", "cos", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "head_dim", "hidden_shape", "hidden_states", "if", "input_shape", "is", "k_proj", "key_states", "kwargs", "layer_idx", "nn", "not", "o_proj", "past_key_values", "position_embeddings", "q_proj", "query_states", "reshape", "return", "scaling", "self", "shape", "sin", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "view"], "ernie4_5/modeling_ernie4_5.py:Ernie4_5RMSNorm.__init__": ["ModelRMSNorm", "Module", "Parameter", "__init__", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "ernie4_5/modeling_ernie4_5.py:Ernie4_5RMSNorm.forward": ["ModelRMSNorm", "Module", "True", "class", "def", "dtype", "float32", "forward", "hidden_states", "input_dtype", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "ernie4_5/modeling_ernie4_5.py:Ernie4_5RMSNorm.extra_repr": ["ModelRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "variance_epsilon", "weight"], "ernie4_5/modeling_ernie4_5.py:Ernie4_5DecoderLayer.__init__": ["GradientCheckpointingLayer", "ModelAttention", "ModelConfig", "ModelDecoderLayer", "ModelMLP", "ModelRMSNorm", "__init__", "class", "config", "def", "eps", "hidden_size", "input_layernorm", "int", "layer_idx", "mlp", "post_attention_layernorm", "rms_norm_eps", "self", "self_attn", "super"], "ernie4_5/modeling_ernie4_5.py:Ernie4_5DecoderLayer.forward": ["Cache", "False", "GradientCheckpointingLayer", "LongTensor", "ModelDecoderLayer", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "attention_mask", "bool", "cache_position", "class", "def", "forward", "hidden_states", "input_layernorm", "kwargs", "mlp", "past_key_values", "position_embeddings", "position_ids", "post_attention_layernorm", "residual", "return", "self", "self_attn", "torch", "tuple", "use_cache"], "ernie4_5/modeling_ernie4_5.py:Ernie4_5Model.__init__": ["Embedding", "False", "ModelConfig", "ModelDecoderLayer", "ModelModel", "ModelPreTrainedModel", "ModelRMSNorm", "ModelRotaryEmbedding", "ModuleList", "__init__", "class", "config", "def", "embed_tokens", "eps", "for", "gradient_checkpointing", "hidden_size", "in", "layer_idx", "layers", "nn", "norm", "num_hidden_layers", "pad_token_id", "padding_idx", "post_init", "range", "rms_norm_eps", "rotary_emb", "self", "super", "vocab_size"], "ernie4_5/modeling_ernie4_5.py:Ernie4_5Model.forward": ["BaseModelOutputWithPast", "Cache", "DynamicCache", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "and", "arange", "attention_mask", "bool", "cache_position", "causal_mask", "class", "config", "create_causal_mask", "decoder_layer", "def", "device", "else", "embed_tokens", "exactly", "for", "forward", "get_seq_length", "hidden_states", "if", "in", "input_embeds", "input_ids", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layers", "must", "norm", "not", "num_hidden_layers", "of", "one", "or", "past_key_values", "past_seen_tokens", "position_embeddings", "position_ids", "raise", "return", "rotary_emb", "self", "shape", "specify", "torch", "unsqueeze", "use_cache"], "ernie4_5/modeling_ernie4_5.py:Ernie4_5ForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "ModelForCausalLM", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "vocab_size"], "ernie4_5/modeling_ernie4_5.py:Ernie4_5ForCausalLM.forward": ["BaseModelOutputWithPast", "Cache", "CausalLMOutputWithPast", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForCausalLM", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "outputs", "past_key_values", "position_ids", "r", "return", "self", "slice", "slice_indices", "torch", "use_cache", "vocab_size"], "ernie4_5_moe/modeling_ernie4_5_moe.py:Ernie4_5_MoeRMSNorm.__init__": ["ModelRMSNorm", "Module", "Parameter", "__init__", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "ernie4_5_moe/modeling_ernie4_5_moe.py:Ernie4_5_MoeRMSNorm.forward": ["ModelRMSNorm", "Module", "True", "class", "def", "dtype", "float32", "forward", "hidden_states", "input_dtype", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "ernie4_5_moe/modeling_ernie4_5_moe.py:Ernie4_5_MoeRMSNorm.extra_repr": ["ModelRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "variance_epsilon", "weight"], "ernie4_5_moe/modeling_ernie4_5_moe.py:Ernie4_5_MoeMLP.__init__": ["ACT2FN", "Linear", "ModelMLP", "Module", "None", "__init__", "act_fn", "bias", "class", "config", "def", "down_proj", "else", "gate_proj", "hidden_act", "hidden_size", "if", "intermediate_size", "is", "nn", "self", "super", "up_proj", "use_bias"], "ernie4_5_moe/modeling_ernie4_5_moe.py:Ernie4_5_MoeMLP.forward": ["ModelMLP", "Module", "act_fn", "class", "def", "down_proj", "forward", "gate_proj", "nn", "return", "self", "up_proj", "x"], "ernie4_5_moe/modeling_ernie4_5_moe.py:Ernie4_5_MoeRotaryEmbedding.__init__": ["Callable", "False", "ModelConfig", "ModelRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "ernie4_5_moe/modeling_ernie4_5_moe.py:Ernie4_5_MoeRotaryEmbedding.compute_default_rope_parameters": ["ModelConfig", "ModelRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "ernie4_5_moe/modeling_ernie4_5_moe.py:Ernie4_5_MoeRotaryEmbedding.forward": ["False", "ModelRotaryEmbedding", "Module", "None", "and", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "ernie4_5_moe/modeling_ernie4_5_moe.py:rotate_half": ["Model_half", "def", "dim", "flatten", "return", "stack", "torch", "x", "x1", "x2"], "ernie4_5_moe/modeling_ernie4_5_moe.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cos", "def", "dim", "dtype", "float", "k", "k_embed", "original_dtype", "q", "q_embed", "repeat_interleave", "return", "rotate_half", "shape", "sin", "to", "unsqueeze", "unsqueeze_dim"], "ernie4_5_moe/modeling_ernie4_5_moe.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "ernie4_5_moe/modeling_ernie4_5_moe.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value", "value_states"], "ernie4_5_moe/modeling_ernie4_5_moe.py:Ernie4_5_MoeAttention.__init__": ["Linear", "ModelAttention", "ModelConfig", "Module", "True", "__init__", "attention_dropout", "bias", "class", "config", "def", "getattr", "head_dim", "hidden_size", "int", "is_causal", "k_proj", "layer_idx", "nn", "num_attention_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "q_proj", "scaling", "self", "super", "use_bias", "v_proj"], "ernie4_5_moe/modeling_ernie4_5_moe.py:Ernie4_5_MoeAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "LongTensor", "ModelAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_attn_implementation", "apply_rotary_pos_emb", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "cache_kwargs", "cache_position", "class", "config", "contiguous", "cos", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "head_dim", "hidden_shape", "hidden_states", "if", "input_shape", "is", "k_proj", "key_states", "kwargs", "layer_idx", "nn", "not", "o_proj", "past_key_values", "position_embeddings", "q_proj", "query_states", "reshape", "return", "scaling", "self", "shape", "sin", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "view"], "ernie4_5_moe/modeling_ernie4_5_moe.py:Ernie4_5_MoeStatics.__init__": ["False", "ModelStatics", "Module", "Parameter", "__init__", "class", "config", "def", "dtype", "e_score_correction_bias", "float32", "moe_num_experts", "nn", "num_experts", "num_experts_groups", "requires_grad", "self", "super", "torch", "zeros"], "ernie4_5_moe/modeling_ernie4_5_moe.py:Ernie4_5_MoeStatics.forward": ["ModelStatics", "Module", "class", "def", "e_score_correction_bias", "forward", "hidden_states", "nn", "return", "self", "squeeze"], "ernie4_5_moe/modeling_ernie4_5_moe.py:Ernie4_5_MoeExperts.__init__": ["ACT2FN", "ModelExperts", "Module", "Parameter", "__init__", "act_fn", "class", "config", "def", "down_proj", "empty", "gate_up_proj", "hidden_act", "hidden_dim", "hidden_size", "intermediate_dim", "moe_intermediate_size", "moe_num_experts", "nn", "num_experts", "self", "super", "torch"], "ernie4_5_moe/modeling_ernie4_5_moe.py:Ernie4_5_MoeExperts.forward": ["ModelExperts", "Module", "None", "Tensor", "act_fn", "chunk", "class", "continue", "current_hidden_states", "current_state", "def", "dim", "down_proj", "dtype", "expert_hit", "expert_idx", "expert_mask", "final_hidden_states", "for", "forward", "functional", "gate", "gate_up_proj", "greater", "hidden_states", "if", "in", "index_add_", "linear", "nn", "no_grad", "nonzero", "num_classes", "num_experts", "one_hot", "permute", "return", "self", "sum", "to", "token_idx", "top_k_index", "top_k_pos", "top_k_weights", "torch", "up", "where", "with", "zeros_like"], "ernie4_5_moe/modeling_ernie4_5_moe.py:Ernie4_5_MoeTopKRouter.__init__": ["ModelStatics", "ModelTopKRouter", "Module", "Parameter", "__init__", "class", "config", "def", "dtype", "float32", "hidden_size", "moe_k", "moe_norm_min", "moe_num_experts", "moe_statics", "nn", "norm_min", "self", "super", "top_k", "torch", "weight", "zeros"], "ernie4_5_moe/modeling_ernie4_5_moe.py:Ernie4_5_MoeTopKRouter.forward": ["F", "False", "ModelTopKRouter", "Module", "Tensor", "True", "_", "and", "clamp", "class", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "enabled", "float", "forward", "gather", "hidden_states", "if", "index", "isinstance", "keepdim", "linear", "maybe_autocast", "min", "moe_statics", "mps", "nn", "norm_min", "return", "router_logits", "routing_weights", "selected_experts", "self", "softmax", "str", "sum", "to", "top_k", "topk", "torch", "tuple", "type", "weight", "with"], "ernie4_5_moe/modeling_ernie4_5_moe.py:Ernie4_5_MoeSparseMoeBlock.__init__": ["ModelExperts", "ModelMLP", "ModelSparseMoeBlock", "ModelTopKRouter", "Module", "None", "__init__", "class", "config", "def", "experts", "gate", "hidden_dim", "hidden_size", "if", "moe_intermediate_size", "moe_k", "moe_num_experts", "moe_num_shared_experts", "nn", "num_experts", "self", "shared_experts", "super", "top_k"], "ernie4_5_moe/modeling_ernie4_5_moe.py:Ernie4_5_MoeSparseMoeBlock.forward": ["ModelSparseMoeBlock", "Module", "None", "Tensor", "_", "batch_size", "class", "def", "dtype", "experts", "final_hidden_states", "forward", "gate", "hidden_dim", "hidden_states", "if", "is", "nn", "not", "reshape", "return", "self", "sequence_length", "shape", "shared_experts", "shared_output", "to", "top_k_index", "top_k_weights", "torch", "view"], "ernie4_5_moe/modeling_ernie4_5_moe.py:Ernie4_5_MoeDecoderLayer.__init__": ["GradientCheckpointingLayer", "ModelAttention", "ModelDecoderLayer", "ModelMLP", "ModelRMSNorm", "ModelSparseMoeBlock", "__init__", "and", "class", "config", "def", "else", "hidden_size", "if", "input_layernorm", "layer_idx", "mlp", "moe_layer_end_index", "moe_layer_interval", "moe_layer_start_index", "post_attention_layernorm", "rms_norm_eps", "self", "self_attn", "super"], "ernie4_5_moe/modeling_ernie4_5_moe.py:Ernie4_5_MoeDecoderLayer.forward": ["Cache", "False", "GradientCheckpointingLayer", "LongTensor", "ModelDecoderLayer", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "attention_mask", "bool", "cache_position", "class", "def", "forward", "hidden_states", "input_layernorm", "kwargs", "mlp", "past_key_values", "position_embeddings", "position_ids", "post_attention_layernorm", "residual", "return", "self", "self_attn", "torch", "tuple", "use_cache"], "ernie4_5_moe/modeling_ernie4_5_moe.py:Ernie4_5_MoePreTrainedModel._init_weights": ["ModelExperts", "ModelPreTrainedModel", "ModelStatics", "PreTrainedModel", "_init_weights", "class", "config", "def", "down_proj", "e_score_correction_bias", "elif", "gate_up_proj", "if", "init", "initializer_range", "isinstance", "mean", "module", "normal_", "self", "std", "super", "zeros_"], "ernie4_5_moe/modeling_ernie4_5_moe.py:Ernie4_5_MoeModel.__init__": ["Embedding", "False", "ModelConfig", "ModelDecoderLayer", "ModelModel", "ModelPreTrainedModel", "ModelRMSNorm", "ModelRotaryEmbedding", "ModuleList", "__init__", "class", "config", "def", "embed_tokens", "eps", "for", "gradient_checkpointing", "hidden_size", "in", "layer_idx", "layers", "nn", "norm", "num_hidden_layers", "pad_token_id", "padding_idx", "post_init", "range", "rms_norm_eps", "rotary_emb", "self", "super", "vocab_size"], "ernie4_5_moe/modeling_ernie4_5_moe.py:Ernie4_5_MoeModel.forward": ["Cache", "DynamicCache", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "MoeModelOutputWithPast", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "and", "arange", "attention_mask", "bool", "cache_position", "causal_mask", "class", "config", "create_causal_mask", "decoder_layer", "def", "device", "else", "embed_tokens", "exactly", "for", "forward", "get_seq_length", "hidden_states", "if", "in", "input_embeds", "input_ids", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layers", "must", "norm", "not", "num_hidden_layers", "of", "one", "or", "past_key_values", "past_seen_tokens", "position_embeddings", "position_ids", "raise", "return", "rotary_emb", "self", "shape", "specify", "torch", "unsqueeze", "use_cache"], "ernie4_5_moe/modeling_ernie4_5_moe.py:load_balancing_loss_func": ["Model_balancing_loss_func", "None", "Tensor", "_", "attention_mask", "batch_size", "cat", "compute_device", "concatenated_gate_logits", "def", "device", "dim", "else", "expand", "expert_attention_mask", "expert_mask", "float", "for", "functional", "gate_logits", "if", "in", "int", "is", "isinstance", "layer_gate", "mean", "nn", "not", "num_experts", "num_hidden_layers", "one_hot", "or", "overall_loss", "r", "reshape", "return", "router_per_expert_attention_mask", "router_prob_per_expert", "routing_weights", "selected_experts", "sequence_length", "shape", "softmax", "sum", "to", "tokens_per_expert", "top_k", "topk", "torch", "tuple", "unsqueeze"], "ernie4_5_moe/modeling_ernie4_5_moe.py:Ernie4_5_MoeForCausalLM.__init__": ["GenerationMixin", "Linear", "ModelForCausalLM", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "moe_k", "moe_num_experts", "nn", "num_experts", "num_experts_per_tok", "post_init", "router_aux_loss_coef", "self", "super", "use_bias", "vocab_size"], "ernie4_5_moe/modeling_ernie4_5_moe.py:Ernie4_5_MoeForCausalLM.forward": ["Cache", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForCausalLM", "ModelPreTrainedModel", "MoeCausalLMOutputWithPast", "MoeModelOutputWithPast", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "aux_loss", "bool", "cache_position", "class", "config", "def", "device", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "load_balancing_loss_func", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "num_experts", "num_experts_per_tok", "output_router_logits", "outputs", "past_key_values", "position_ids", "r", "return", "router_aux_loss_coef", "router_logits", "self", "slice", "slice_indices", "to", "torch", "use_cache", "vocab_size"], "ernie4_5_vl_moe/modeling_ernie4_5_vl_moe.py:Ernie4_5_VL_MoeTextRotaryEmbedding.__init__": ["Callable", "False", "Model", "ModelTextRotaryEmbedding", "Module", "None", "VL", "ValueError", "__init__", "attention_scaling", "but", "class", "compute_default_rope_parameters", "config", "def", "default", "device", "f", "found", "get", "if", "instead", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "mrope_section", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "raise", "register_buffer", "requires", "rope", "rope_init_fn", "rope_parameters", "rope_type", "self", "super", "the", "type"], "ernie4_5_vl_moe/modeling_ernie4_5_vl_moe.py:Ernie4_5_VL_MoeTextRotaryEmbedding.compute_default_rope_parameters": ["ModelTextConfig", "ModelTextRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "cat", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "empty_like", "float", "get", "getattr", "head_dim", "hidden_size", "hw_dim", "int", "int64", "inv_freq", "inv_freq_3d", "mrope_section", "nn", "num_attention_heads", "or", "return", "rope_parameters", "rope_theta", "seq_len", "t_dim", "to", "torch", "tuple"], "ernie4_5_vl_moe/modeling_ernie4_5_vl_moe.py:Ernie4_5_VL_MoeTextRotaryEmbedding.forward": ["False", "ModelTextRotaryEmbedding", "Module", "None", "and", "attention_scaling", "class", "cos", "cpu", "def", "device", "device_type", "else", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "recomposition_to_3d", "return", "self", "shape", "sin", "str", "to", "transpose", "type", "with", "x"], "ernie4_5_vl_moe/modeling_ernie4_5_vl_moe.py:Ernie4_5_VL_MoeTextRotaryEmbedding.recomposition_to_3d": ["ModelTextRotaryEmbedding", "Module", "cat", "class", "def", "dim", "enumerate", "flatten", "for", "freq", "freq_h", "freq_hw", "freq_hwt", "freq_t", "freq_w", "i", "in", "m", "mrope_section", "nn", "recomposition_to_3d", "repeat_interleave", "return", "self", "split", "stack", "torch"], "ernie4_5_vl_moe/modeling_ernie4_5_vl_moe.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "ernie4_5_vl_moe/modeling_ernie4_5_vl_moe.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value", "value_states"], "ernie4_5_vl_moe/modeling_ernie4_5_vl_moe.py:rotate_half_text": ["Model_half_text", "def", "dim", "flatten", "return", "stack", "torch", "x", "x1", "x2"], "ernie4_5_vl_moe/modeling_ernie4_5_vl_moe.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cos", "def", "dtype", "float", "k", "k_embed", "original_dtype", "q", "q_embed", "return", "rotate_half_text", "sin", "to", "unsqueeze", "unsqueeze_dim"], "ernie4_5_vl_moe/modeling_ernie4_5_vl_moe.py:Ernie4_5_VL_MoeTextAttention.__init__": ["Linear", "ModelConfig", "ModelTextAttention", "Module", "True", "__init__", "attention_dropout", "bias", "class", "config", "def", "getattr", "head_dim", "hidden_size", "int", "is_causal", "k_proj", "layer_idx", "nn", "num_attention_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "q_proj", "scaling", "self", "super", "use_bias", "v_proj"], "ernie4_5_vl_moe/modeling_ernie4_5_vl_moe.py:Ernie4_5_VL_MoeTextAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "LongTensor", "ModelTextAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_attn_implementation", "apply_rotary_pos_emb", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "cache_kwargs", "cache_position", "class", "config", "contiguous", "cos", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "head_dim", "hidden_shape", "hidden_states", "if", "input_shape", "is", "k_proj", "key_states", "kwargs", "layer_idx", "nn", "not", "o_proj", "past_key_values", "position_embeddings", "q_proj", "query_states", "reshape", "return", "scaling", "self", "shape", "sin", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "view"], "ernie4_5_vl_moe/modeling_ernie4_5_vl_moe.py:Ernie4_5_VL_MoeRMSNorm.__init__": ["ModelRMSNorm", "Module", "Parameter", "__init__", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "ernie4_5_vl_moe/modeling_ernie4_5_vl_moe.py:Ernie4_5_VL_MoeRMSNorm.forward": ["ModelRMSNorm", "Module", "True", "class", "def", "dtype", "float32", "forward", "hidden_states", "input_dtype", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "ernie4_5_vl_moe/modeling_ernie4_5_vl_moe.py:Ernie4_5_VL_MoeRMSNorm.extra_repr": ["ModelRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "variance_epsilon", "weight"], "ernie4_5_vl_moe/modeling_ernie4_5_vl_moe.py:Ernie4_5_VL_MoeMLP.__init__": ["ACT2FN", "Linear", "ModelMLP", "Module", "None", "__init__", "act_fn", "bias", "class", "config", "def", "down_proj", "else", "gate_proj", "hidden_act", "hidden_size", "if", "intermediate_size", "is", "nn", "self", "super", "up_proj", "use_bias"], "ernie4_5_vl_moe/modeling_ernie4_5_vl_moe.py:Ernie4_5_VL_MoeMLP.forward": ["ModelMLP", "Module", "act_fn", "class", "def", "down_proj", "forward", "gate_proj", "nn", "return", "self", "up_proj", "x"], "ernie4_5_vl_moe/modeling_ernie4_5_vl_moe.py:Ernie4_5_VL_MoeMoeStatics.__init__": ["False", "ModelMoeStatics", "Module", "Parameter", "__init__", "class", "config", "def", "dtype", "e_score_correction_bias", "float32", "moe_num_experts", "nn", "num_experts", "num_experts_groups", "requires_grad", "self", "super", "torch", "zeros"], "ernie4_5_vl_moe/modeling_ernie4_5_vl_moe.py:Ernie4_5_VL_MoeMoeStatics.forward": ["ModelMoeStatics", "Module", "class", "def", "e_score_correction_bias", "forward", "hidden_states", "nn", "return", "self", "squeeze"], "ernie4_5_vl_moe/modeling_ernie4_5_vl_moe.py:Ernie4_5_VL_MoeMoeTopKRouter.__init__": ["ModelMoeStatics", "ModelMoeTopKRouter", "Module", "Parameter", "__init__", "class", "config", "def", "dtype", "float32", "hidden_size", "moe_k", "moe_norm_min", "moe_num_experts", "moe_statics", "nn", "norm_min", "self", "super", "top_k", "torch", "weight", "zeros"], "ernie4_5_vl_moe/modeling_ernie4_5_vl_moe.py:Ernie4_5_VL_MoeMoeTopKRouter.forward": ["F", "False", "ModelMoeTopKRouter", "Module", "Tensor", "True", "_", "and", "clamp", "class", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "enabled", "float", "forward", "gather", "hidden_states", "if", "index", "isinstance", "keepdim", "linear", "maybe_autocast", "min", "moe_statics", "mps", "nn", "norm_min", "return", "router_logits", "routing_weights", "selected_experts", "self", "softmax", "str", "sum", "to", "top_k", "topk", "torch", "tuple", "type", "weight", "with"], "ernie4_5_vl_moe/modeling_ernie4_5_vl_moe.py:Ernie4_5_VL_MoeMoeExperts.__init__": ["ACT2FN", "ModelMoeExperts", "Module", "None", "Parameter", "__init__", "act_fn", "class", "config", "def", "down_proj", "else", "empty", "gate_up_proj", "hidden_act", "hidden_dim", "hidden_size", "if", "intermediate_dim", "intermediate_size", "is", "moe_intermediate_size", "moe_num_experts", "nn", "num_experts", "self", "super", "torch"], "ernie4_5_vl_moe/modeling_ernie4_5_vl_moe.py:Ernie4_5_VL_MoeMoeExperts.forward": ["ModelMoeExperts", "Module", "None", "Tensor", "act_fn", "chunk", "class", "continue", "current_hidden_states", "current_state", "def", "dim", "down_proj", "dtype", "expert_hit", "expert_idx", "expert_mask", "final_hidden_states", "for", "forward", "functional", "gate", "gate_up_proj", "greater", "hidden_states", "if", "in", "index_add_", "linear", "nn", "no_grad", "nonzero", "num_classes", "num_experts", "one_hot", "permute", "return", "self", "sum", "to", "token_idx", "top_k_index", "top_k_pos", "top_k_weights", "torch", "up", "where", "with", "zeros_like"], "ernie4_5_vl_moe/modeling_ernie4_5_vl_moe.py:Ernie4_5_VL_MoeSparseMoeBlock.__init__": ["ModelMoeExperts", "ModelMoeTopKRouter", "ModelSparseMoeBlock", "Module", "__init__", "class", "config", "def", "experts", "gate", "hidden_dim", "hidden_size", "intermediate_size", "moe_k", "moe_num_experts", "nn", "num_experts", "self", "super", "top_k"], "ernie4_5_vl_moe/modeling_ernie4_5_vl_moe.py:Ernie4_5_VL_MoeSparseMoeBlock.forward": ["ModelSparseMoeBlock", "Module", "Tensor", "class", "def", "experts", "final_hidden_states", "flatten", "forward", "gate", "hidden_dim", "hidden_states", "nn", "return", "router_logits", "self", "top_k_index", "top_k_weights", "torch", "tuple", "view"], "ernie4_5_vl_moe/modeling_ernie4_5_vl_moe.py:Ernie4_5_VL_MoeMoeBlock.__init__": ["ModelMLP", "ModelMoeBlock", "ModelSparseMoeBlock", "Module", "None", "__init__", "class", "config", "def", "if", "intermediate_size", "moe_intermediate_size", "moe_num_experts", "moe_num_shared_experts", "nn", "num_experts", "self", "shared_experts", "super", "text_moe", "vision_moe"], "ernie4_5_vl_moe/modeling_ernie4_5_vl_moe.py:Ernie4_5_VL_MoeMoeBlock.forward": ["IntTensor", "ModelMoeBlock", "Module", "None", "Tensor", "and", "any", "batch_size", "bool", "class", "def", "device", "dtype", "else", "expand", "final_hidden_states", "float", "forward", "hidden_dim", "hidden_states", "if", "is", "moe_mm_token_type_ids", "nn", "not", "num_experts", "reshape", "return", "router_logits", "self", "sequence_length", "shape", "shared_experts", "shared_output", "size", "text_moe", "token_type_ids_router", "token_type_ids_states", "torch", "tuple", "vision_moe", "zeros", "zeros_like"], "ernie4_5_vl_moe/modeling_ernie4_5_vl_moe.py:Ernie4_5_VL_MoeDecoderLayer.__init__": ["GradientCheckpointingLayer", "ModelDecoderLayer", "ModelMLP", "ModelMoeBlock", "ModelRMSNorm", "ModelTextAttention", "__init__", "class", "config", "def", "else", "hidden_size", "if", "input_layernorm", "layer_idx", "mlp", "mlp_layer_types", "post_attention_layernorm", "rms_norm_eps", "self", "self_attn", "sparse", "super"], "ernie4_5_vl_moe/modeling_ernie4_5_vl_moe.py:Ernie4_5_VL_MoeDecoderLayer.forward": ["Cache", "FlashAttentionKwargs", "GradientCheckpointingLayer", "IntTensor", "LongTensor", "ModelDecoderLayer", "ModelMoeBlock", "None", "Tensor", "Unpack", "_", "attention_mask", "cache_position", "class", "def", "else", "forward", "hidden_states", "if", "input_layernorm", "isinstance", "kwargs", "mlp", "moe_mm_token_type_ids", "past_key_values", "position_embeddings", "position_ids", "post_attention_layernorm", "residual", "return", "self", "self_attn", "torch", "tuple"], "ernie4_5_vl_moe/modeling_ernie4_5_vl_moe.py:Ernie4_5_VL_MoePreTrainedModel._init_weights": ["ModelMoeExperts", "ModelMoeTopKRouter", "ModelPreTrainedModel", "ModelVisionRotaryEmbedding", "PreTrainedModel", "_init_weights", "arange", "class", "config", "copy_", "def", "dim", "down_proj", "dtype", "e_score_correction_bias", "elif", "float", "gate_up_proj", "if", "init", "initializer_range", "inv_freq", "isinstance", "mean", "module", "moe_statics", "normal_", "self", "std", "super", "theta", "torch", "weight", "zeros_"], "ernie4_5_vl_moe/modeling_ernie4_5_vl_moe.py:Ernie4_5_VL_MoeTextModel.__init__": ["Embedding", "False", "ModelDecoderLayer", "ModelPreTrainedModel", "ModelRMSNorm", "ModelTextConfig", "ModelTextModel", "ModelTextRotaryEmbedding", "ModuleList", "__init__", "class", "config", "def", "embed_tokens", "eps", "for", "gradient_checkpointing", "hidden_size", "in", "layer_idx", "layers", "nn", "norm", "num_hidden_layers", "pad_token_id", "padding_idx", "post_init", "range", "rms_norm_eps", "rotary_emb", "self", "super", "vocab_size"], "ernie4_5_vl_moe/modeling_ernie4_5_vl_moe.py:Ernie4_5_VL_MoeTextModel.forward": ["Cache", "DynamicCache", "FlashAttentionKwargs", "FloatTensor", "IntTensor", "LongTensor", "ModelPreTrainedModel", "ModelTextModel", "MoeModelOutputWithPast", "None", "Tensor", "Unpack", "ValueError", "You", "and", "arange", "attention_mask", "bool", "cache_position", "class", "config", "create_causal_mask", "decoder_layer", "def", "device", "elif", "else", "embed_tokens", "exactly", "expand", "for", "forward", "get_seq_length", "hidden_states", "if", "in", "input_embeds", "input_ids", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layers", "moe_mm_token_type_ids", "must", "ndim", "norm", "not", "num_hidden_layers", "of", "one", "or", "past_key_values", "past_seen_tokens", "position_embeddings", "position_ids", "r", "raise", "return", "rotary_emb", "self", "shape", "specify", "text_position_ids", "torch", "use_cache", "view"], "ernie4_5_vl_moe/modeling_ernie4_5_vl_moe.py:Ernie4_5VLVisionMLP.__init__": ["ACT2FN", "Linear", "Model_5VLVisionMLP", "Module", "None", "__init__", "act", "class", "def", "dim", "fc1", "fc2", "hidden_act", "hidden_dim", "int", "nn", "self", "str", "super"], "ernie4_5_vl_moe/modeling_ernie4_5_vl_moe.py:Ernie4_5VLVisionMLP.forward": ["Model_5VLVisionMLP", "Module", "Tensor", "act", "class", "def", "fc1", "fc2", "forward", "nn", "return", "self", "torch", "x"], "ernie4_5_vl_moe/modeling_ernie4_5_vl_moe.py:Ernie4_5_VL_MoePatchEmbed.__init__": ["False", "Linear", "ModelPatchEmbed", "Module", "None", "__init__", "bias", "class", "def", "embed_dim", "in_channels", "int", "nn", "patch_size", "proj", "self", "super"], "ernie4_5_vl_moe/modeling_ernie4_5_vl_moe.py:Ernie4_5_VL_MoePatchEmbed.forward": ["ModelPatchEmbed", "Module", "Tensor", "class", "def", "dtype", "forward", "hidden_states", "nn", "proj", "return", "self", "target_dtype", "to", "torch", "weight"], "ernie4_5_vl_moe/modeling_ernie4_5_vl_moe.py:Ernie4_5_VL_MoeVisionRotaryEmbedding.__init__": ["False", "ModelVisionRotaryEmbedding", "Module", "None", "__init__", "arange", "class", "def", "dim", "dtype", "float", "int", "inv_freq", "nn", "persistent", "register_buffer", "self", "super", "theta", "torch"], "ernie4_5_vl_moe/modeling_ernie4_5_vl_moe.py:Ernie4_5_VL_MoeVisionRotaryEmbedding.forward": ["ModelVisionRotaryEmbedding", "Module", "Tensor", "arange", "class", "def", "device", "dtype", "forward", "freqs", "int", "inv_freq", "nn", "outer", "return", "self", "seq", "seqlen", "torch"], "ernie4_5_vl_moe/modeling_ernie4_5_vl_moe.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "ernie4_5_vl_moe/modeling_ernie4_5_vl_moe.py:apply_rotary_pos_emb_vision": ["Model_rotary_pos_emb_vision", "Tensor", "cos", "def", "dtype", "float", "k", "k_embed", "orig_k_dtype", "orig_q_dtype", "q", "q_embed", "return", "rotate_half", "sin", "to", "torch", "tuple", "unsqueeze"], "ernie4_5_vl_moe/modeling_ernie4_5_vl_moe.py:Ernie4_5_VL_MoeVisionAttention.__init__": ["False", "Linear", "ModelVisionAttention", "ModelVisionConfig", "Module", "None", "True", "__init__", "attention_dropout", "bias", "class", "config", "def", "dim", "head_dim", "hidden_size", "is_causal", "nn", "num_heads", "num_key_value_groups", "proj", "qkv", "scaling", "self", "super"], "ernie4_5_vl_moe/modeling_ernie4_5_vl_moe.py:Ernie4_5_VL_MoeVisionAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "False", "ModelVisionAttention", "Module", "None", "Tensor", "_", "_attn_implementation", "apply_rotary_pos_emb_vision", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_outputs", "cat", "class", "config", "contiguous", "cos", "cu_seq_lens_k", "cu_seq_lens_q", "cu_seqlens", "def", "dim", "dropout", "eager", "eager_attention_forward", "else", "flash", "for", "forward", "hidden_states", "if", "in", "is_causal", "k", "key_states", "kwargs", "lengths", "max", "max_length_k", "max_length_q", "max_seqlen", "nn", "not", "num_heads", "permute", "position_embeddings", "proj", "q", "qkv", "query_states", "reshape", "return", "rotary_pos_emb", "scaling", "self", "seq_length", "shape", "sin", "split", "splits", "tensor", "tolist", "torch", "training", "transpose", "tuple", "unbind", "unsqueeze", "v", "value_states", "zip"], "ernie4_5_vl_moe/modeling_ernie4_5_vl_moe.py:Ernie4_5_VL_MoeVisionBlock.__init__": ["GradientCheckpointingLayer", "LayerNorm", "ModelVisionAttention", "ModelVisionBlock", "Model_5VLVisionMLP", "None", "__init__", "attn", "class", "config", "def", "dim", "hidden_act", "hidden_dim", "hidden_size", "intermediate_size", "mlp", "nn", "norm1", "norm2", "rms_norm_eps", "self", "super"], "ernie4_5_vl_moe/modeling_ernie4_5_vl_moe.py:Ernie4_5_VL_MoeVisionBlock.forward": ["GradientCheckpointingLayer", "ModelVisionBlock", "None", "Tensor", "attn", "class", "cu_seqlens", "def", "forward", "hidden_states", "kwargs", "mlp", "norm1", "norm2", "position_embeddings", "return", "rotary_pos_emb", "self", "torch", "tuple"], "ernie4_5_vl_moe/modeling_ernie4_5_vl_moe.py:Ernie4_5_VL_MoeVisionTransformerPretrainedModel.__init__": ["False", "LayerNorm", "ModelPatchEmbed", "ModelPreTrainedModel", "ModelVisionBlock", "ModelVisionRotaryEmbedding", "ModelVisionTransformerPretrainedModel", "ModuleList", "None", "_", "__init__", "blocks", "class", "config", "def", "depth", "embed_dim", "eps", "for", "gradient_checkpointing", "head_dim", "hidden_size", "in", "in_channels", "ln", "nn", "num_heads", "patch_embed", "patch_size", "post_init", "range", "rms_norm_eps", "rotary_pos_emb", "self", "spatial_merge_size", "super"], "ernie4_5_vl_moe/modeling_ernie4_5_vl_moe.py:Ernie4_5_VL_MoeVisionTransformerPretrainedModel.rot_pos_emb": ["ModelPreTrainedModel", "ModelVisionTransformerPretrainedModel", "append", "arange", "cat", "class", "def", "dim", "expand", "flatten", "for", "grid_thw", "h", "hpos_ids", "in", "max", "max_grid_size", "permute", "pos_ids", "repeat", "reshape", "return", "rot_pos_emb", "rotary_pos_emb", "rotary_pos_emb_full", "self", "spatial_merge_size", "stack", "t", "torch", "unsqueeze", "w", "wpos_ids"], "ernie4_5_vl_moe/modeling_ernie4_5_vl_moe.py:Ernie4_5_VL_MoeVisionTransformerPretrainedModel.forward": ["F", "ModelPreTrainedModel", "ModelVisionTransformerPretrainedModel", "Tensor", "block", "blocks", "cat", "class", "cos", "cu_seqlens", "cumsum", "def", "dim", "dtype", "else", "emb", "for", "forward", "grid_thw", "hidden_states", "if", "in", "int32", "is_tracing", "jit", "kwargs", "ln", "pad", "patch_embed", "position_embeddings", "r", "repeat_interleave", "return", "rot_pos_emb", "rotary_pos_emb", "self", "sin", "torch", "value"], "ernie4_5_vl_moe/modeling_ernie4_5_vl_moe.py:Ernie4_5_VL_MoeVisionMLP.__init__": ["GELU", "LayerNorm", "Linear", "ModelVisionMLP", "Module", "__init__", "act_fn", "class", "config", "def", "eps", "fc1", "fc2", "in_dim", "ln", "nn", "out_dim", "rms_norm_eps", "self", "super", "vision_config"], "ernie4_5_vl_moe/modeling_ernie4_5_vl_moe.py:Ernie4_5_VL_MoeVisionMLP.forward": ["ModelVisionMLP", "Module", "act_fn", "class", "def", "fc1", "fc2", "forward", "hidden_states", "ln", "nn", "return", "self"], "ernie4_5_vl_moe/modeling_ernie4_5_vl_moe.py:Ernie4_5_VL_MoeVariableResolutionResamplerModel.__init__": ["Linear", "ModelConfig", "ModelRMSNorm", "ModelVariableResolutionResamplerModel", "ModelVisionMLP", "Module", "__init__", "after_norm", "class", "config", "def", "hidden_size", "in_dim", "mlp", "nn", "out_dim", "rms_norm_eps", "self", "spatial_dim", "spatial_linear", "spatial_merge_size", "super", "temporal_dim", "temporal_linear", "temporal_merge_size", "text_config", "vision_config"], "ernie4_5_vl_moe/modeling_ernie4_5_vl_moe.py:Ernie4_5_VL_MoeVariableResolutionResamplerModel._temporal_slicing": ["ModelVariableResolutionResamplerModel", "Module", "_temporal_slicing", "append", "arange", "batch_offset", "batch_offsets", "cat", "class", "concat", "cumsum", "def", "device", "dim", "dtype", "else", "empty", "first_offset_range", "first_slice_offsets", "flatten", "for", "grid_hw", "grid_hw_after_conv", "grid_t", "grid_thw", "hidden_states", "if", "in", "index", "index_select", "nn", "prod", "range", "return", "second_offset_range", "second_slice_offsets", "self", "size", "spatial_merge_size", "spatial_size", "temporal_offset_even", "temporal_offset_odd", "temporal_size", "to", "tokens_per_img_or_vid", "torch", "zip"], "ernie4_5_vl_moe/modeling_ernie4_5_vl_moe.py:Ernie4_5_VL_MoeVariableResolutionResamplerModel.forward": ["ModelVariableResolutionResamplerModel", "Module", "_temporal_slicing", "after_norm", "class", "def", "forward", "grid_thw", "hidden_states", "mlp", "nn", "reshape", "return", "self", "shape", "spatial_linear", "spatial_merge_size", "temporal_linear"], "ernie4_5_vl_moe/modeling_ernie4_5_vl_moe.py:Ernie4_5_VL_MoeModel.__init__": ["ModelConfig", "ModelModel", "ModelPreTrainedModel", "ModelTextModel", "ModelVariableResolutionResamplerModel", "ModelVisionTransformerPretrainedModel", "None", "__init__", "_from_config", "class", "config", "def", "language_model", "post_init", "resampler_model", "rope_deltas", "self", "super", "text_config", "vision_config", "vision_tower"], "ernie4_5_vl_moe/modeling_ernie4_5_vl_moe.py:Ernie4_5_VL_MoeModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "language_model", "return", "self"], "ernie4_5_vl_moe/modeling_ernie4_5_vl_moe.py:Ernie4_5_VL_MoeModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "language_model", "self", "set_input_embeddings", "value"], "ernie4_5_vl_moe/modeling_ernie4_5_vl_moe.py:Ernie4_5_VL_MoeModel.get_rope_index": ["False", "IntTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "True", "and", "append", "arange", "attention_mask", "cat", "class", "config", "cumsum", "def", "device", "dim", "dtype", "else", "end_idx", "end_index", "enumerate", "expand", "flatten", "for", "get_rope_index", "grid_thw", "group", "groupby", "h", "h_index", "i", "if", "image_grid_thw", "image_index", "in", "input_ids", "input_token_type", "input_type_group", "is", "item", "itertools", "keepdim", "key", "lambda", "len", "list", "llm_grid_h", "llm_grid_t", "llm_grid_w", "llm_pos_ids_list", "llm_positions", "long", "masked_fill_", "max", "max_position_ids", "mm_index", "mm_token_type_ids", "modality_type", "mrope_position_deltas", "not", "ones", "ones_like", "or", "position_ids", "range", "reshape", "return", "self", "shape", "spatial_merge_size", "st_idx", "stack", "start_idx", "start_index", "t", "t_idx", "t_index", "t_merge_size", "temporal_merge_size", "tensor", "text_len", "to", "tolist", "torch", "total_input_ids", "tuple", "unsqueeze", "video_grid_thw", "video_index", "view", "vision_config", "w", "w_index", "x", "zeros", "zeros_like"], "ernie4_5_vl_moe/modeling_ernie4_5_vl_moe.py:Ernie4_5_VL_MoeModel.get_video_features": ["FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "class", "def", "get_video_features", "pixel_values_videos", "prod", "resampler_model", "return", "self", "spatial_merge_size", "split", "split_sizes", "temporal_merge_size", "tolist", "torch", "video_embeds", "video_grid_thw", "vision_tower"], "ernie4_5_vl_moe/modeling_ernie4_5_vl_moe.py:Ernie4_5_VL_MoeModel.get_image_features": ["FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "class", "def", "get_image_features", "image_embeds", "image_grid_thw", "pixel_values", "prod", "resampler_model", "return", "self", "spatial_merge_size", "split", "split_sizes", "tolist", "torch", "vision_tower"], "ernie4_5_vl_moe/modeling_ernie4_5_vl_moe.py:Ernie4_5_VL_MoeModel.get_placeholder_mask": ["FloatTensor", "Image", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "ValueError", "Videos", "all", "and", "class", "config", "def", "device", "do", "dtype", "else", "expand_as", "f", "features", "get_input_embeddings", "get_placeholder_mask", "if", "image", "image_features", "image_token_id", "input_ids", "inputs_embeds", "is", "long", "match", "n_image_tokens", "n_video_tokens", "not", "numel", "raise", "return", "self", "shape", "special_image_mask", "special_video_mask", "sum", "tensor", "to", "tokens", "torch", "unsqueeze", "video", "video_features", "video_token_id"], "ernie4_5_vl_moe/modeling_ernie4_5_vl_moe.py:Ernie4_5_VL_MoeModel.forward": ["Cache", "FloatTensor", "IntTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "MoeModelOutputWithPast", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "_", "attention_mask", "attentions", "bool", "cache_position", "cat", "class", "def", "device", "dim", "dtype", "forward", "get_image_features", "get_input_embeddings", "get_placeholder_mask", "get_position_ids", "get_video_features", "hidden_states", "if", "image_embeds", "image_features", "image_grid_thw", "image_mask", "input_ids", "inputs_embeds", "is", "kwargs", "language_model", "last_hidden_state", "masked_scatter", "mm_token_type_ids", "moe_mm_token_type_ids", "not", "outputs", "past_key_values", "pixel_values", "pixel_values_videos", "position_ids", "r", "return", "return_dict", "rope_deltas", "router_logits", "self", "to", "torch", "tuple", "use_cache", "video_embeds", "video_features", "video_grid_thw", "video_mask"], "ernie4_5_vl_moe/modeling_ernie4_5_vl_moe.py:Ernie4_5_VL_MoeModel.get_position_ids": ["Cache", "Cannot", "FloatTensor", "IntTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "Need", "None", "Tensor", "ValueError", "add", "and", "any", "arange", "attention_mask", "batch_size", "cache_position", "calculate", "class", "def", "delta", "device", "dim", "either", "elif", "else", "expand", "get_position_ids", "get_rope_index", "get_seq_length", "ids", "if", "image_grid_thw", "input", "input_ids", "inputs_embeds", "is", "is_torchdynamo_compiling", "mm_token_type_ids", "model", "not", "or", "past_key_values", "position", "position_ids", "prefill_compiled_stage", "prefill_noncompiled_stage", "raise", "repeat_interleave", "return", "rope_deltas", "self", "seq_length", "shape", "the", "to", "torch", "unsqueeze", "video_grid_thw", "view", "without"], "ernie4_5_vl_moe/modeling_ernie4_5_vl_moe.py:load_balancing_loss_func": ["Model_balancing_loss_func", "None", "Tensor", "_", "attention_mask", "batch_size", "cat", "compute_device", "concatenated_gate_logits", "def", "device", "dim", "else", "expand", "expert_attention_mask", "expert_mask", "float", "for", "functional", "gate_logits", "if", "in", "int", "is", "isinstance", "layer_gate", "mean", "nn", "not", "num_experts", "num_hidden_layers", "one_hot", "or", "overall_loss", "r", "reshape", "return", "router_per_expert_attention_mask", "router_prob_per_expert", "routing_weights", "selected_experts", "sequence_length", "shape", "softmax", "sum", "to", "tokens_per_expert", "top_k", "topk", "torch", "tuple", "unsqueeze"], "ernie4_5_vl_moe/modeling_ernie4_5_vl_moe.py:Ernie4_5_VL_MoeForConditionalGeneration.__init__": ["False", "GenerationMixin", "Linear", "ModelForConditionalGeneration", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "moe_k", "moe_num_experts", "nn", "num_experts", "num_experts_per_tok", "post_init", "router_aux_loss_coef", "self", "super", "text_config", "vocab_size"], "ernie4_5_vl_moe/modeling_ernie4_5_vl_moe.py:Ernie4_5_VL_MoeForConditionalGeneration.get_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "model", "return", "self"], "ernie4_5_vl_moe/modeling_ernie4_5_vl_moe.py:Ernie4_5_VL_MoeForConditionalGeneration.set_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "model", "self", "set_input_embeddings", "value"], "ernie4_5_vl_moe/modeling_ernie4_5_vl_moe.py:Ernie4_5_VL_MoeForConditionalGeneration.get_video_features": ["FloatTensor", "GenerationMixin", "LongTensor", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "class", "def", "get_video_features", "model", "pixel_values_videos", "return", "self", "torch", "video_grid_thw"], "ernie4_5_vl_moe/modeling_ernie4_5_vl_moe.py:Ernie4_5_VL_MoeForConditionalGeneration.get_image_features": ["FloatTensor", "GenerationMixin", "LongTensor", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "class", "def", "get_image_features", "image_grid_thw", "model", "pixel_values", "return", "self", "torch"], "ernie4_5_vl_moe/modeling_ernie4_5_vl_moe.py:Ernie4_5_VL_MoeForConditionalGeneration.forward": ["Cache", "FloatTensor", "GenerationMixin", "IntTensor", "LongTensor", "ModelForConditionalGeneration", "ModelPreTrainedModel", "MoeCausalLMOutputWithPast", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "attention_mask", "attentions", "aux_loss", "bool", "cache_position", "class", "config", "def", "device", "else", "forward", "hidden_states", "if", "image_grid_thw", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "load_balancing_loss_func", "logits", "logits_to_keep", "loss", "loss_function", "mm_token_type_ids", "model", "moe_mm_token_type_ids", "not", "num_experts", "num_experts_per_tok", "output_router_logits", "outputs", "past_key_values", "pixel_values", "pixel_values_videos", "position_ids", "r", "return", "return_dict", "rope_deltas", "router_aux_loss_coef", "router_logits", "self", "slice", "slice_indices", "text_config", "to", "torch", "tuple", "use_cache", "video_grid_thw", "vocab_size"], "ernie4_5_vl_moe/modeling_ernie4_5_vl_moe.py:Ernie4_5_VL_MoeForConditionalGeneration.prepare_inputs_for_generation": ["False", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "True", "and", "attention_mask", "cache_position", "class", "def", "get", "get_position_ids", "if", "image_grid_thw", "input_ids", "inputs_embeds", "is_first_iteration", "kwargs", "mm_token_type_ids", "model", "model_inputs", "moe_mm_token_type_ids", "not", "past_key_values", "pixel_values", "pixel_values_videos", "position_ids", "prepare_inputs_for_generation", "return", "self", "super", "use_cache", "video_grid_thw"], "ernie4_5_vl_moe/modeling_ernie4_5_vl_moe.py:Ernie4_5_VL_MoeForConditionalGeneration._get_image_nums_and_video_nums": ["GenerationMixin", "LongTensor", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Tensor", "_get_image_nums_and_video_nums", "class", "config", "cumsum", "def", "device", "dim", "dtype", "else", "get_input_embeddings", "if", "image_counts", "image_start_token_id", "input_ids", "inputs_embeds", "inside_video", "int", "is", "is_image", "is_video_end", "is_video_start", "long", "not", "return", "self", "standalone_images", "sum", "tensor", "torch", "tuple", "video_counts", "video_end_token_id", "video_level", "video_start_token_id"], "ernie4_5_vl_moe/modeling_ernie4_5_vl_moe.py:Ernie4_5_VL_MoeForConditionalGeneration._expand_inputs_for_generation": ["Any", "False", "GenerationMixin", "If", "LongTensor", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Tensor", "True", "ValueError", "_expand_dict_for_generation", "_expand_dict_for_generation_visual", "_expand_inputs_for_generation", "_get_image_nums_and_video_nums", "_repeat_interleave_samples", "and", "bool", "cache_position", "cat", "class", "def", "defined", "dict", "dict_to_expand", "dim", "elif", "encoder_outputs", "expand_size", "for", "get", "if", "image_grid_thw", "image_nums", "in", "input_ids", "inputs_embeds", "int", "is", "is_encoder_decoder", "isinstance", "key", "lengths", "list", "make", "model_kwargs", "not", "pixel_values", "pixel_values_videos", "prod", "raise", "repeat", "repeat_args", "repeat_interleave", "repeat_times", "result", "return", "sample", "samples", "second_per_grid_ts", "self", "split", "str", "sum", "sure", "that", "torch", "tuple", "video_grid_thw", "video_nums", "visual_keys", "x"], "esm/modeling_esm.py:rotate_half": ["Model_half", "cat", "chunk", "def", "dim", "return", "torch", "x", "x1", "x2"], "esm/modeling_esm.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cos", "def", "return", "rotate_half", "shape", "sin", "x"], "esm/modeling_esm.py:gelu": ["Model", "def", "erf", "math", "return", "sqrt", "torch", "x"], "esm/modeling_esm.py:symmetrize": ["Make", "Model", "contact", "def", "dimensions", "final", "for", "in", "layer", "prediction", "return", "symmetric", "transpose", "two", "used", "x"], "esm/modeling_esm.py:average_product_correct": ["Model", "Model_product_correct", "Perform", "True", "a1", "a12", "a2", "avg", "contact", "correct", "def", "div_", "for", "keepdims", "normalized", "prediction", "product", "return", "sum", "used", "x"], "esm/modeling_esm.py:RotaryEmbedding.__init__": ["ModelEmbedding", "Module", "None", "__init__", "_cos_cached", "_seq_len_cached", "_sin_cached", "arange", "class", "def", "dim", "dtype", "float", "int", "int64", "inv_freq", "nn", "register_buffer", "self", "super", "torch"], "esm/modeling_esm.py:RotaryEmbedding._update_cos_sin_tables": ["ModelEmbedding", "Module", "None", "_cos_cached", "_seq_len_cached", "_sin_cached", "_update_cos_sin_tables", "arange", "cat", "class", "cos", "def", "device", "dim", "emb", "freqs", "if", "inv_freq", "nn", "or", "outer", "return", "self", "seq_dimension", "seq_len", "shape", "sin", "t", "to", "torch", "type_as", "x"], "esm/modeling_esm.py:RotaryEmbedding.forward": ["ModelEmbedding", "Module", "Tensor", "_cos_cached", "_sin_cached", "_update_cos_sin_tables", "apply_Model_pos_emb", "class", "def", "dtype", "forward", "k", "nn", "q", "return", "self", "seq_dimension", "to", "torch", "tuple"], "esm/modeling_esm.py:EsmContactPredictionHead.__init__": ["Linear", "ModelContactPredictionHead", "Module", "Sigmoid", "True", "__init__", "activation", "bias", "class", "def", "eos_idx", "in_features", "int", "nn", "regression", "self", "super"], "esm/modeling_esm.py:EsmContactPredictionHead.forward": ["ModelContactPredictionHead", "Module", "None", "_", "activation", "attentions", "average_product_correct", "batch_size", "class", "def", "device", "eos_idx", "eos_mask", "forward", "heads", "layers", "ne", "nn", "permute", "regression", "return", "self", "seqlen", "size", "squeeze", "symmetrize", "to", "tokens", "unsqueeze", "view", "weight"], "esm/modeling_esm.py:EsmEmbeddings.__init__": ["Dropout", "Embedding", "False", "LayerNorm", "ModelEmbeddings", "Module", "None", "__init__", "absolute", "arange", "class", "config", "def", "dropout", "else", "emb_layer_norm_before", "eps", "expand", "getattr", "hidden_dropout_prob", "hidden_size", "if", "layer_norm", "layer_norm_eps", "mask_token_id", "max_position_embeddings", "nn", "pad_token_id", "padding_idx", "persistent", "position_embedding_type", "position_embeddings", "position_ids", "register_buffer", "self", "super", "token_dropout", "torch", "vocab_size", "word_embeddings"], "esm/modeling_esm.py:EsmEmbeddings.forward": ["ModelEmbeddings", "Module", "None", "absolute", "and", "attention_mask", "class", "create_position_ids_from_input_ids", "create_position_ids_from_inputs_embeds", "def", "dtype", "else", "embeddings", "float", "forward", "if", "input_ids", "inputs_embeds", "is", "layer_norm", "mask_ratio_observed", "mask_ratio_train", "mask_token_id", "masked_fill", "nn", "not", "padding_idx", "position_embedding_type", "position_embeddings", "position_ids", "return", "self", "shape", "src_lengths", "sum", "to", "token_dropout", "unsqueeze", "word_embeddings"], "esm/modeling_esm.py:EsmEmbeddings.create_position_ids_from_inputs_embeds": ["ModelEmbeddings", "Module", "arange", "class", "create_position_ids_from_inputs_embeds", "def", "device", "dtype", "expand", "input_shape", "inputs_embeds", "long", "nn", "padding_idx", "position_ids", "return", "self", "sequence_length", "size", "torch", "unsqueeze"], "esm/modeling_esm.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "contiguous", "def", "dim", "dropout", "float", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "shape", "size", "softmax", "torch", "training", "transpose", "value"], "esm/modeling_esm.py:EsmSelfAttention.__init__": ["False", "Linear", "ModelSelfAttention", "Module", "None", "RotaryEmbedding", "The", "ValueError", "__init__", "a", "absolute", "all_head_size", "and", "attention", "attention_head_size", "attention_probs_dropout_prob", "class", "config", "def", "dim", "dropout", "embedding_size", "f", "getattr", "hasattr", "heads", "hidden", "hidden_size", "if", "int", "is", "is_causal", "is_cross_attention", "is_decoder", "key", "layer_idx", "multiple", "nn", "not", "num_attention_heads", "number", "of", "or", "position_embedding_type", "query", "raise", "rotary", "rotary_embeddings", "scaling", "self", "size", "super", "the", "value"], "esm/modeling_esm.py:EsmSelfAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "FloatTensor", "ModelSelfAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_attn_implementation", "attention_head_size", "attention_interface", "attention_mask", "attn_output", "attn_weights", "batch_size", "class", "config", "contiguous", "current_states", "def", "dropout", "eager", "eager_attention_forward", "else", "encoder_attention_mask", "encoder_hidden_states", "forward", "hidden_shape", "hidden_states", "if", "is", "is_cross_attention", "key", "key_layer", "kwargs", "nn", "not", "position_embedding_type", "query", "query_layer", "reshape", "return", "rotary", "rotary_embeddings", "scaling", "self", "seq_length", "shape", "torch", "training", "transpose", "tuple", "value", "value_layer", "view"], "esm/modeling_esm.py:EsmSelfOutput.__init__": ["Dropout", "Linear", "ModelSelfOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "hidden_dropout_prob", "hidden_size", "nn", "self", "super"], "esm/modeling_esm.py:EsmSelfOutput.forward": ["ModelSelfOutput", "Module", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self"], "esm/modeling_esm.py:EsmAttention.__init__": ["False", "LayerNorm", "ModelAttention", "ModelSelfAttention", "ModelSelfOutput", "Module", "None", "__init__", "class", "config", "def", "eps", "hidden_size", "is_cross_attention", "layer_idx", "layer_norm_eps", "nn", "output", "self", "super"], "esm/modeling_esm.py:EsmAttention.forward": ["LayerNorm", "ModelAttention", "Module", "None", "TransformersKwargs", "Unpack", "_", "attention_mask", "attn_output", "class", "def", "encoder_attention_mask", "encoder_hidden_states", "forward", "hidden_states", "hidden_states_ln", "kwargs", "nn", "output", "return", "self"], "esm/modeling_esm.py:EsmIntermediate.__init__": ["Linear", "ModelIntermediate", "Module", "__init__", "class", "config", "def", "dense", "hidden_size", "intermediate_size", "nn", "self", "super"], "esm/modeling_esm.py:EsmIntermediate.forward": ["ModelIntermediate", "Module", "Tensor", "class", "def", "dense", "forward", "gelu", "hidden_states", "nn", "return", "self", "torch"], "esm/modeling_esm.py:EsmOutput.__init__": ["Dropout", "Linear", "ModelOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "hidden_dropout_prob", "hidden_size", "intermediate_size", "nn", "self", "super"], "esm/modeling_esm.py:EsmOutput.forward": ["ModelOutput", "Module", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self"], "esm/modeling_esm.py:EsmLayer.__init__": ["GradientCheckpointingLayer", "LayerNorm", "ModelAttention", "ModelIntermediate", "ModelLayer", "ModelOutput", "RuntimeError", "True", "__init__", "a", "add_cross_attention", "added", "as", "attention", "be", "chunk_size_feed_forward", "class", "config", "cross", "crossattention", "decoder", "def", "eps", "f", "hidden_size", "if", "intermediate", "is", "is_cross_attention", "is_decoder", "layer_norm_eps", "model", "nn", "not", "output", "raise", "self", "seq_len_dim", "should", "super", "used"], "esm/modeling_esm.py:EsmLayer.forward": ["AttributeError", "GradientCheckpointingLayer", "If", "ModelLayer", "None", "TransformersKwargs", "True", "Unpack", "add_cross_attention", "and", "are", "attention", "attention_mask", "attention_output", "be", "by", "class", "config", "cross", "crossattention", "def", "encoder_attention_mask", "encoder_hidden_states", "f", "feed_forward_chunk", "forward", "has", "hasattr", "hidden_states", "if", "instantiated", "is", "is_decoder", "kwargs", "layer_output", "layers", "not", "passed", "raise", "return", "self", "setting", "to", "with"], "esm/modeling_esm.py:EsmLayer.feed_forward_chunk": ["GradientCheckpointingLayer", "LayerNorm", "ModelLayer", "attention_output", "attention_output_ln", "class", "def", "feed_forward_chunk", "intermediate", "intermediate_output", "layer_output", "output", "return", "self"], "esm/modeling_esm.py:EsmEncoder.__init__": ["False", "LayerNorm", "ModelEncoder", "ModelLayer", "Module", "ModuleList", "_", "__init__", "class", "config", "def", "emb_layer_norm_after", "eps", "for", "gradient_checkpointing", "hidden_size", "in", "layer", "layer_norm_eps", "nn", "num_hidden_layers", "range", "self", "super"], "esm/modeling_esm.py:EsmEncoder.forward": ["BaseModelOutputWithCrossAttentions", "ModelEncoder", "Module", "None", "TransformersKwargs", "Unpack", "attention_mask", "class", "def", "emb_layer_norm_after", "encoder_attention_mask", "encoder_hidden_states", "enumerate", "for", "forward", "hidden_states", "i", "if", "in", "kwargs", "last_hidden_state", "layer", "layer_module", "nn", "return", "self"], "esm/modeling_esm.py:EsmPooler.__init__": ["Linear", "ModelPooler", "Module", "Tanh", "__init__", "activation", "class", "config", "def", "dense", "hidden_size", "nn", "self", "super"], "esm/modeling_esm.py:EsmPooler.forward": ["ModelPooler", "Module", "Tensor", "activation", "class", "def", "dense", "first_token_tensor", "forward", "hidden_states", "nn", "pooled_output", "return", "self", "torch"], "esm/modeling_esm.py:EsmPreTrainedModel._init_weights": ["ModelEmbeddings", "ModelLMHead", "ModelPreTrainedModel", "PreTrainedModel", "RotaryEmbedding", "_init_weights", "arange", "bias", "class", "copy_", "def", "dim", "dtype", "elif", "expand", "float", "if", "init", "int64", "inv_freq", "isinstance", "module", "position_ids", "self", "shape", "super", "torch", "zeros_"], "esm/modeling_esm.py:EsmPreTrainedModel.get_output_embeddings": ["ModelPreTrainedModel", "None", "PreTrainedModel", "class", "def", "get_output_embeddings", "return", "self"], "esm/modeling_esm.py:EsmModel.__init__": ["ModelContactPredictionHead", "ModelEmbeddings", "ModelEncoder", "ModelModel", "ModelPooler", "ModelPreTrainedModel", "None", "True", "__init__", "add_pooling_layer", "bias", "class", "config", "contact_head", "def", "else", "embeddings", "encoder", "if", "in_features", "num_attention_heads", "num_hidden_layers", "pooler", "post_init", "r", "self", "super"], "esm/modeling_esm.py:EsmModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "return", "self", "word_embeddings"], "esm/modeling_esm.py:EsmModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "self", "set_input_embeddings", "value", "word_embeddings"], "esm/modeling_esm.py:EsmModel.forward": ["BaseModelOutputWithPoolingAndCrossAttentions", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "_create_attention_masks", "arange", "attention_mask", "cache_position", "class", "def", "device", "else", "embedding_output", "embeddings", "encoder", "encoder_attention_mask", "encoder_hidden_states", "encoder_outputs", "exactly", "forward", "if", "input_ids", "inputs_embeds", "is", "kwargs", "last_hidden_state", "must", "not", "of", "one", "or", "past_key_values", "pooled_output", "pooler", "pooler_output", "position_ids", "r", "raise", "return", "self", "sequence_output", "shape", "specify", "torch", "tuple"], "esm/modeling_esm.py:EsmModel._create_attention_masks": ["ModelModel", "ModelPreTrainedModel", "None", "_create_attention_masks", "attention_mask", "cache_position", "class", "config", "create_bidirectional_mask", "create_causal_mask", "def", "else", "embedding_output", "encoder_attention_mask", "encoder_hidden_states", "if", "input_embeds", "is", "is_decoder", "not", "past_key_values", "return", "self"], "esm/modeling_esm.py:EsmModel.predict_contacts": ["ModelModel", "ModelPreTrainedModel", "True", "attention_mask", "attentions", "attns", "class", "contact_head", "def", "dim", "output_attentions", "predict_contacts", "return", "return_dict", "self", "stack", "tokens", "torch", "unsqueeze"], "esm/modeling_esm.py:EsmForMaskedLM.__init__": ["False", "If", "Model", "ModelForMaskedLM", "ModelLMHead", "ModelModel", "ModelPreTrainedModel", "__init__", "add_pooling_layer", "attention", "bi", "class", "config", "def", "directional", "for", "if", "is_decoder", "lm_head", "logger", "make", "post_init", "self", "super", "sure", "to", "use", "want", "warning", "you"], "esm/modeling_esm.py:EsmForMaskedLM.get_output_embeddings": ["ModelForMaskedLM", "ModelPreTrainedModel", "class", "decoder", "def", "get_output_embeddings", "lm_head", "return", "self"], "esm/modeling_esm.py:EsmForMaskedLM.set_output_embeddings": ["ModelForMaskedLM", "ModelPreTrainedModel", "class", "decoder", "def", "lm_head", "new_embeddings", "self", "set_output_embeddings"], "esm/modeling_esm.py:EsmForMaskedLM.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "MaskedLMOutput", "Model", "ModelForMaskedLM", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "class", "config", "def", "device", "encoder_attention_mask", "encoder_hidden_states", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "lm_head", "logits", "loss", "loss_fct", "masked_lm_loss", "not", "outputs", "position_ids", "prediction_scores", "r", "return", "self", "sequence_output", "to", "torch", "tuple", "view", "vocab_size"], "esm/modeling_esm.py:EsmForMaskedLM.predict_contacts": ["Model", "ModelForMaskedLM", "ModelPreTrainedModel", "attention_mask", "class", "def", "predict_contacts", "return", "self", "tokens"], "esm/modeling_esm.py:EsmLMHead.__init__": ["False", "LayerNorm", "Linear", "ModelLMHead", "Module", "Parameter", "__init__", "bias", "class", "config", "decoder", "def", "dense", "eps", "hidden_size", "layer_norm", "layer_norm_eps", "nn", "self", "super", "torch", "vocab_size", "zeros"], "esm/modeling_esm.py:EsmLMHead.forward": ["ModelLMHead", "Module", "bias", "class", "decoder", "def", "dense", "features", "forward", "gelu", "kwargs", "layer_norm", "nn", "return", "self", "x"], "esm/modeling_esm.py:EsmForSequenceClassification.__init__": ["False", "Model", "ModelClassificationHead", "ModelForSequenceClassification", "ModelModel", "ModelPreTrainedModel", "__init__", "add_pooling_layer", "class", "classifier", "config", "def", "num_labels", "post_init", "self", "super"], "esm/modeling_esm.py:EsmForSequenceClassification.forward": ["BCEWithLogitsLoss", "CrossEntropyLoss", "FloatTensor", "LongTensor", "MSELoss", "Model", "ModelForSequenceClassification", "ModelPreTrainedModel", "None", "SequenceClassifierOutput", "Tensor", "TransformersKwargs", "Unpack", "and", "attention_mask", "attentions", "class", "classifier", "config", "def", "device", "dtype", "elif", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "kwargs", "labels", "logits", "long", "loss", "loss_fct", "multi_label_classification", "not", "num_labels", "or", "outputs", "position_ids", "problem_type", "r", "regression", "return", "self", "sequence_output", "single_label_classification", "squeeze", "to", "torch", "tuple", "view"], "esm/modeling_esm.py:EsmForTokenClassification.__init__": ["Dropout", "False", "Linear", "Model", "ModelForTokenClassification", "ModelModel", "ModelPreTrainedModel", "__init__", "add_pooling_layer", "class", "classifier", "config", "def", "dropout", "hidden_dropout_prob", "hidden_size", "nn", "num_labels", "post_init", "self", "super"], "esm/modeling_esm.py:EsmForTokenClassification.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "Model", "ModelForTokenClassification", "ModelPreTrainedModel", "None", "Tensor", "TokenClassifierOutput", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "class", "classifier", "def", "device", "dropout", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "not", "num_labels", "outputs", "position_ids", "r", "return", "self", "sequence_output", "to", "torch", "tuple", "view"], "esm/modeling_esm.py:EsmClassificationHead.__init__": ["Dropout", "Linear", "ModelClassificationHead", "Module", "__init__", "class", "config", "def", "dense", "dropout", "hidden_dropout_prob", "hidden_size", "nn", "num_labels", "out_proj", "self", "super"], "esm/modeling_esm.py:EsmClassificationHead.forward": ["ModelClassificationHead", "Module", "class", "def", "dense", "dropout", "features", "forward", "kwargs", "nn", "out_proj", "return", "self", "tanh", "torch", "x"], "esm/modeling_esm.py:create_position_ids_from_input_ids": ["Model_position_ids_from_input_ids", "cumsum", "def", "dim", "incremental_indices", "input_ids", "int", "long", "mask", "ne", "padding_idx", "return", "torch", "type_as"], "esm/modeling_esmfold.py:is_fp16_enabled": ["and", "autocast_dtype", "def", "device_type", "else", "float16", "fp16_enabled", "get_autocast_dtype", "get_autocast_gpu_dtype", "hasattr", "if", "is_autocast_enabled", "is_fp16_enabled", "return", "torch"], "esm/modeling_esmfold.py:is_deepspeed_initialized": ["Exception", "False", "deepspeed", "def", "else", "except", "if", "is_deepspeed_available", "is_deepspeed_initialized", "is_initialized", "return", "try", "utils"], "esm/modeling_esmfold.py:collate_dense_tensors": ["Model_dense_tensors", "RuntimeError", "Samples", "Tensor", "def", "device", "dim", "dimensions", "dtype", "empty", "f", "fill_", "float", "for", "has", "i", "if", "in", "k", "len", "list", "lst", "max", "max_shape", "pad_v", "raise", "range", "result", "result_i", "return", "samples", "shape", "slice", "t", "torch", "tuple", "varying", "x", "zip"], "esm/modeling_esmfold.py:flatten_final_dims": ["Model_final_dims", "Tensor", "def", "int", "no_dims", "reshape", "return", "shape", "t", "torch"], "esm/modeling_esmfold.py:permute_final_dims": ["Model", "Model_final_dims", "Tensor", "def", "first_inds", "for", "i", "in", "inds", "int", "len", "list", "range", "return", "shape", "tensor", "torch", "zero_index"], "esm/modeling_esmfold.py:dict_multimap": ["Model", "Model_multimap", "Models", "all_v", "d", "def", "else", "first", "fn", "for", "if", "in", "isinstance", "items", "k", "new_Model", "return", "v"], "esm/modeling_esmfold.py:EsmFoldLinear.__init__": ["Callable", "Invalid", "Linear", "ModelFoldLinear", "None", "Tensor", "True", "ValueError", "__init__", "bias", "bool", "class", "def", "default", "fill_", "final", "gating", "glorot", "if", "in", "in_dim", "init", "init_fn", "int", "nn", "no_grad", "normal", "not", "out_dim", "raise", "relu", "self", "str", "string", "super", "torch", "with"], "esm/modeling_esmfold.py:EsmFoldLayerNorm.__init__": ["ModelFoldLayerNorm", "Module", "Parameter", "__init__", "bias", "c_in", "class", "def", "eps", "nn", "ones", "self", "super", "torch", "weight", "zeros"], "esm/modeling_esmfold.py:EsmFoldLayerNorm.forward": ["False", "ModelFoldLayerNorm", "Module", "and", "bfloat16", "bias", "c_in", "class", "cuda", "d", "def", "device_type", "dtype", "else", "enabled", "eps", "forward", "functional", "if", "is", "is_deepspeed_initialized", "layer_norm", "maybe_autocast", "nn", "not", "out", "return", "self", "to", "torch", "weight", "with", "x"], "esm/modeling_esmfold.py:softmax_no_cast": ["False", "Model", "Model_no_cast", "Tensor", "and", "bfloat16", "cuda", "d", "def", "device_type", "dim", "dtype", "else", "enabled", "functional", "if", "int", "is", "is_deepspeed_initialized", "maybe_autocast", "nn", "not", "return", "s", "t", "torch", "with"], "esm/modeling_esmfold.py:EsmFoldAttention.__init__": ["False", "ModelFoldAttention", "ModelFoldLinear", "Module", "None", "Sigmoid", "True", "__init__", "bias", "bool", "c_hidden", "c_k", "c_q", "c_v", "class", "def", "final", "gating", "glorot", "if", "init", "int", "linear_g", "linear_k", "linear_o", "linear_q", "linear_v", "nn", "no_heads", "self", "sigmoid", "super"], "esm/modeling_esmfold.py:EsmFoldAttention._prep_qkv": ["ModelFoldAttention", "Module", "Tensor", "_prep_qkv", "c_hidden", "class", "def", "k", "kv_x", "linear_k", "linear_q", "linear_v", "math", "nn", "no_heads", "q", "q_x", "return", "self", "shape", "sqrt", "torch", "transpose", "tuple", "v", "view"], "esm/modeling_esmfold.py:EsmFoldAttention._wrap_up": ["ModelFoldAttention", "Module", "None", "Tensor", "_wrap_up", "class", "def", "flatten_final_dims", "g", "if", "is", "linear_g", "linear_o", "nn", "no_heads", "not", "o", "q_x", "return", "self", "shape", "sigmoid", "torch", "view"], "esm/modeling_esmfold.py:EsmFoldAttention.forward": ["Choose", "False", "For", "If", "ModelFoldAttention", "Module", "None", "Tensor", "ValueError", "_prep_qkv", "_wrap_up", "algorithm", "alternative", "and", "at", "attention", "attn_options", "b", "be", "bias", "biases", "bool", "class", "def", "flash_mask", "for", "forward", "if", "in", "incompatible", "instead", "int", "is", "key", "kv_x", "list", "lma_kv_chunk_size", "lma_q_chunk_size", "masking", "matmul", "most", "must", "nn", "not", "one", "option", "or", "output", "permute_final_dims", "provided", "q_x", "query", "raise", "return", "self", "softmax_no_cast", "specified", "sum", "the", "torch", "transpose", "use", "use_flash", "use_lma", "use_memory_efficient_kernel", "value", "with"], "esm/modeling_esmfold.py:EsmFoldTriangleAttention.__init__": ["False", "LayerNorm", "ModelFoldAttention", "ModelFoldLinear", "ModelFoldTriangleAttention", "Module", "True", "__init__", "bias", "c_hidden", "c_in", "class", "def", "inf", "init", "layer_norm", "linear", "mha", "nn", "no_heads", "normal", "self", "starting", "super"], "esm/modeling_esmfold.py:EsmFoldTriangleAttention._chunk": ["False", "ModelFoldTriangleAttention", "Module", "None", "Tensor", "_chunk", "_out", "biases", "bool", "chunk_layer", "chunk_size", "class", "def", "else", "if", "inplace_safe", "int", "kv_x", "len", "list", "mha", "mha_inputs", "nn", "no_batch_dims", "partial", "q_x", "return", "self", "shape", "torch", "triangle", "use_lma", "use_memory_efficient_kernel", "x"], "esm/modeling_esmfold.py:EsmFoldTriangleAttention.forward": ["False", "ModelFoldTriangleAttention", "Module", "None", "Tensor", "_chunk", "biases", "bool", "chunk_size", "class", "def", "else", "forward", "if", "inf", "inplace_safe", "int", "is", "kv_x", "layer_norm", "linear", "mask", "mask_bias", "mha", "new_ones", "nn", "not", "permute_final_dims", "q_x", "return", "self", "shape", "starting", "torch", "transpose", "triangle_bias", "unsqueeze", "use_lma", "use_memory_efficient_kernel", "x"], "esm/modeling_esmfold.py:EsmFoldTriangleMultiplicativeUpdate.__init__": ["LayerNorm", "ModelFoldLinear", "ModelFoldTriangleMultiplicativeUpdate", "Module", "Sigmoid", "True", "__init__", "_outgoing", "c_hidden", "class", "config", "def", "final", "gating", "init", "layer_norm_in", "layer_norm_out", "linear_a_g", "linear_a_p", "linear_b_g", "linear_b_p", "linear_g", "linear_z", "nn", "pairwise_state_dim", "self", "sigmoid", "super"], "esm/modeling_esmfold.py:EsmFoldTriangleMultiplicativeUpdate._combine_projections": ["ModelFoldTriangleMultiplicativeUpdate", "Module", "None", "Tensor", "_combine_projections", "_inplace_chunk_size", "_outgoing", "a", "a_chunk", "b", "b_chunk", "class", "def", "else", "for", "i", "if", "in", "int", "is", "matmul", "nn", "not", "p", "permute_final_dims", "range", "return", "self", "shape", "torch"], "esm/modeling_esmfold.py:EsmFoldTriangleMultiplicativeUpdate._inference_forward": ["False", "ModelFoldTriangleMultiplicativeUpdate", "Module", "None", "Tensor", "True", "_", "_inference_forward", "_outgoing", "a", "after_half", "after_half_offsets", "and", "b", "b_chunk", "b_chunk_dim", "bias", "bool", "c", "chunked", "class", "clone", "col_dim", "combined_range_with_offsets", "compute_projection", "compute_projection_helper", "copy_", "def", "del", "dim", "else", "empty_slicer", "end", "first_half_slicer", "flip_z_cache_", "for", "g", "g_chunk", "half_n", "i", "i_1", "i_2", "i_range", "if", "in", "initial_offsets", "inplace_chunk_size", "int", "is", "layer_norm_in", "layer_norm_out", "linear_a_g", "linear_a_p", "linear_b_g", "linear_b_p", "linear_g", "linear_p", "linear_z", "list", "mask", "mask_chunk", "matmul", "n", "need_transpose", "new_ones", "new_zeros", "nn", "not", "offset", "out_shape", "p", "pair", "pair_chunk", "permute_final_dims", "quadrant_3", "quadrant_3_slicer", "quadrant_4", "range", "return", "row_dim", "s", "self", "shape", "sigmoid_", "slice", "slice_tensor", "start", "t", "torch", "transpose", "unsqueeze", "with_add", "x", "x_chunk", "z", "z_cache", "z_cache_offset", "z_cache_rotated", "z_cache_shape", "z_cache_slicer", "z_chunk_b", "z_chunk_g", "z_chunk_slicer", "z_slicer", "zip"], "esm/modeling_esmfold.py:EsmFoldTriangleMultiplicativeUpdate.forward": ["False", "ModelFoldTriangleMultiplicativeUpdate", "Module", "None", "Tensor", "_add_with_inplace", "_combine_projections", "_inference_forward", "_inplace_chunk_size", "a", "b", "bool", "class", "cpu", "def", "del", "device", "device_type", "else", "enabled", "float", "forward", "g", "if", "inplace_chunk_size", "inplace_safe", "int", "is", "is_fp16_enabled", "layer_norm_in", "layer_norm_out", "linear_a_g", "linear_a_p", "linear_b_g", "linear_b_p", "linear_g", "linear_z", "mask", "maybe_autocast", "mps", "new_ones", "nn", "return", "self", "shape", "sigmoid", "torch", "type", "unsqueeze", "with", "with_add", "x", "z"], "esm/modeling_esmfold.py:EsmFoldPreTrainedModel._init_weights": ["ModelFoldInvariantPointAttention", "ModelFoldLinear", "ModelFoldPreTrainedModel", "ModelFoldTriangularSelfAttentionBlock", "ModelPreTrainedModel", "None", "_init_weights", "bias", "class", "constant_", "def", "default", "elif", "else", "final", "gain", "gating", "glorot", "head_weights", "if", "init", "init_fn", "is", "isinstance", "kaiming_normal_", "linear", "linear_o", "linear_z", "math", "max", "mha", "mlp", "mlp_pair", "mlp_seq", "module", "no_grad", "nonlinearity", "normal", "normal_", "not", "o_proj", "ones_", "pair_to_sequence", "relu", "scale", "self", "seq_attention", "sequence_to_pair", "shape", "softplus_inverse_1", "sqrt", "std", "super", "torch", "tri_att_end", "tri_att_start", "tri_mul_in", "tri_mul_out", "weight", "with", "xavier_uniform_", "zeros_"], "esm/modeling_esmfold.py:EsmFoldSelfAttention.__init__": ["False", "Linear", "ModelFoldSelfAttention", "Module", "True", "__init__", "assert", "bias", "class", "def", "embed_dim", "g_proj", "gated", "head_width", "if", "init", "nn", "num_heads", "o_proj", "ones_", "proj", "rescale_factor", "self", "super", "weight", "zeros_"], "esm/modeling_esmfold.py:EsmFoldSelfAttention.forward": ["False", "ModelFoldSelfAttention", "Module", "None", "a", "bias", "chunk", "class", "def", "dim", "einsum", "forward", "functional", "g_proj", "gated", "hkc", "hqk", "if", "indices", "inf", "is", "k", "kc", "mask", "masked_fill", "nn", "not", "np", "num_heads", "o_proj", "permute", "proj", "q", "qc", "qhc", "qk", "rescale_factor", "reshape", "return", "self", "shape", "sigmoid", "softmax", "t", "torch", "v", "view", "x", "y"], "esm/modeling_esmfold.py:EsmFoldDropout.__init__": ["Dropout", "ModelFoldDropout", "Module", "__init__", "batch_dim", "class", "def", "dropout", "float", "if", "int", "isinstance", "list", "nn", "r", "self", "super"], "esm/modeling_esmfold.py:EsmFoldDropout.forward": ["ModelFoldDropout", "Module", "None", "Tensor", "batch_dim", "bd", "class", "def", "dropout", "for", "forward", "if", "in", "is", "list", "new_ones", "nn", "not", "return", "self", "shape", "torch", "x"], "esm/modeling_esmfold.py:EsmFoldSequenceToPair.__init__": ["LayerNorm", "Linear", "ModelFoldSequenceToPair", "Module", "True", "__init__", "bias", "class", "def", "init", "inner_dim", "layernorm", "nn", "o_proj", "pairwise_state_dim", "proj", "self", "sequence_state_dim", "super", "zeros_"], "esm/modeling_esmfold.py:EsmFoldSequenceToPair.forward": ["ModelFoldSequenceToPair", "Module", "None", "assert", "cat", "chunk", "class", "def", "diff", "dim", "forward", "k", "layernorm", "len", "nn", "o_proj", "prod", "proj", "q", "return", "s", "self", "sequence_state", "shape", "torch", "x"], "esm/modeling_esmfold.py:EsmFoldPairToSequence.__init__": ["False", "LayerNorm", "Linear", "ModelFoldPairToSequence", "Module", "__init__", "bias", "class", "def", "layernorm", "linear", "nn", "num_heads", "pairwise_state_dim", "self", "super"], "esm/modeling_esmfold.py:EsmFoldPairToSequence.forward": ["ModelFoldPairToSequence", "Module", "assert", "class", "def", "forward", "layernorm", "len", "linear", "nn", "pairwise_bias", "pairwise_state", "return", "self", "shape", "z"], "esm/modeling_esmfold.py:EsmFoldResidueMLP.__init__": ["Dropout", "LayerNorm", "Linear", "ModelFoldResidueMLP", "Module", "ReLU", "Sequential", "__init__", "class", "def", "dropout", "embed_dim", "inner_dim", "mlp", "nn", "self", "super"], "esm/modeling_esmfold.py:EsmFoldResidueMLP.forward": ["ModelFoldResidueMLP", "Module", "class", "def", "forward", "mlp", "nn", "return", "self", "x"], "esm/modeling_esmfold.py:EsmFoldTriangularSelfAttentionBlock.__init__": ["Dropout", "False", "LayerNorm", "ModelFoldDropout", "ModelFoldPairToSequence", "ModelFoldResidueMLP", "ModelFoldSelfAttention", "ModelFoldSequenceToPair", "ModelFoldTriangleAttention", "ModelFoldTriangleMultiplicativeUpdate", "ModelFoldTriangularSelfAttentionBlock", "Module", "True", "__init__", "_outgoing", "class", "col_drop", "config", "def", "drop", "dropout", "gated", "inf", "layernorm_1", "mlp_pair", "mlp_seq", "nn", "pair_to_sequence", "pairwise_head_width", "pairwise_num_heads", "pairwise_state_dim", "row_drop", "self", "seq_attention", "sequence_head_width", "sequence_num_heads", "sequence_state_dim", "sequence_to_pair", "starting", "super", "tri_att_end", "tri_att_start", "tri_mul_in", "tri_mul_out"], "esm/modeling_esmfold.py:EsmFoldTriangularSelfAttentionBlock.forward": ["Got", "ModelFoldTriangularSelfAttentionBlock", "Module", "None", "ValueError", "_", "__kwargs", "a", "and", "batch", "batch_dim", "be", "bias", "chunk_size", "class", "col_drop", "config", "def", "dimension", "dims", "drop", "else", "equal", "f", "forward", "got", "have", "if", "inconsistent", "is", "last", "layernorm_1", "len", "length", "mask", "mlp_pair", "mlp_seq", "nn", "not", "or", "pair_to_sequence", "pairwise_state", "pairwise_state_dim", "raise", "return", "row_drop", "self", "seq_attention", "seq_dim", "sequence", "sequence_state", "sequence_state_dim", "sequence_to_pair", "shape", "should", "size", "tensor", "to", "tri_att_end", "tri_att_start", "tri_mask", "tri_mul_in", "tri_mul_out", "unsqueeze", "y"], "esm/modeling_esmfold.py:EsmCategoricalMixture.__init__": ["ModelCategoricalMixture", "__init__", "bins", "class", "def", "device", "dtype", "end", "linspace", "logits", "param", "self", "start", "torch", "v_bins"], "esm/modeling_esmfold.py:EsmCategoricalMixture.log_prob": ["ModelCategoricalMixture", "None", "abs", "argmin", "class", "def", "dim", "log_prob", "log_softmax", "logits", "ndim", "nll", "return", "self", "squeeze", "take_along_dim", "torch", "true", "true_index", "unsqueeze", "v_bins"], "esm/modeling_esmfold.py:EsmCategoricalMixture.mean": ["ModelCategoricalMixture", "class", "def", "logits", "mean", "return", "self", "softmax", "squeeze", "unsqueeze", "v_bins"], "esm/modeling_esmfold.py:categorical_lddt": ["ModelModelMixture", "Model_lddt", "bins", "def", "logits", "mean", "return"], "esm/modeling_esmfold.py:get_axial_mask": ["Model_axial_mask", "None", "ValueError", "a", "batch_dim", "be", "def", "dims", "expand", "f", "got", "if", "is", "len", "m", "mask", "raise", "reshape", "return", "seq_dim", "shape", "should", "tensor", "unsqueeze"], "esm/modeling_esmfold.py:EsmFoldRelativePosition.__init__": ["Embedding", "ModelFoldRelativePosition", "Module", "__init__", "bins", "class", "config", "def", "embedding", "nn", "pairwise_state_dim", "position_bins", "self", "super", "torch"], "esm/modeling_esmfold.py:EsmFoldRelativePosition.forward": ["False", "ModelFoldRelativePosition", "Module", "None", "ValueError", "and", "be", "bins", "clamp", "class", "def", "diff", "dtype", "embedding", "f", "forward", "has", "have", "if", "inconsistent", "is", "it", "long", "mask", "nn", "not", "output", "raise", "residue_index", "return", "self", "shape", "shapes", "should", "torch"], "esm/modeling_esmfold.py:EsmFoldAngleResnetBlock.__init__": ["ModelFoldAngleResnetBlock", "ModelFoldLinear", "Module", "ReLU", "__init__", "class", "config", "def", "final", "init", "linear_1", "linear_2", "nn", "relu", "resnet_dim", "self", "super"], "esm/modeling_esmfold.py:EsmFoldAngleResnetBlock.forward": ["ModelFoldAngleResnetBlock", "Module", "Tensor", "a", "class", "def", "forward", "linear_1", "linear_2", "nn", "relu", "return", "s_initial", "self", "torch"], "esm/modeling_esmfold.py:EsmFoldAngleResnet.__init__": ["ModelFoldAngleResnet", "ModelFoldAngleResnetBlock", "ModelFoldLinear", "Module", "ModuleList", "ReLU", "_", "__init__", "append", "class", "config", "def", "for", "in", "layer", "layers", "linear_in", "linear_initial", "linear_out", "nn", "num_angles", "num_resnet_blocks", "range", "relu", "resnet_dim", "self", "sequence_dim", "super"], "esm/modeling_esmfold.py:EsmFoldAngleResnet.forward": ["ModelFoldAngleResnet", "Module", "Tensor", "True", "clamp", "class", "config", "def", "dim", "epsilon", "for", "forward", "in", "keepdim", "l", "layers", "linear_in", "linear_initial", "linear_out", "min", "nn", "norm_denom", "relu", "return", "s", "s_initial", "self", "shape", "sqrt", "sum", "torch", "tuple", "unnormalized_s", "view"], "esm/modeling_esmfold.py:EsmFoldInvariantPointAttention.__init__": ["ModelFoldInvariantPointAttention", "ModelFoldLinear", "Module", "Parameter", "Softmax", "Softplus", "__init__", "c_s", "c_z", "class", "concat_out_dim", "config", "def", "dim", "final", "hc", "head_weights", "hidden_dim", "hpkv", "hpq", "init", "ipa_dim", "linear_b", "linear_kv", "linear_kv_points", "linear_out", "linear_q", "linear_q_points", "nn", "num_heads", "num_heads_ipa", "num_qk_points", "num_v_points", "pairwise_dim", "self", "sequence_dim", "softmax", "softplus", "super", "torch", "zeros"], "esm/modeling_esmfold.py:EsmFoldInvariantPointAttention.forward": ["False", "ModelFoldInvariantPointAttention", "Module", "None", "Rigid", "Sequence", "Tensor", "_offload_inference", "_z_reference_list", "a", "apply", "assert", "b", "bool", "cat", "class", "config", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "enabled", "epsilon", "flatten_final_dims", "float", "forward", "getrefcount", "head_weights", "hidden_dim", "if", "inf", "invert_apply", "is_fp16_enabled", "k", "k_pts", "kv", "kv_pts", "len", "linear_b", "linear_kv", "linear_kv_points", "linear_out", "linear_q", "linear_q_points", "mask", "math", "matmul", "maybe_autocast", "mps", "nn", "num_heads", "num_qk_points", "num_v_points", "o", "o_pair", "o_pt", "o_pt_norm", "permute_final_dims", "pt_att", "q", "q_pts", "r", "reshape", "return", "s", "self", "shape", "softmax", "softplus", "split", "sqrt", "square_mask", "stack", "sum", "sys", "to", "torch", "transpose", "type", "unbind", "unsqueeze", "v", "v_pts", "view", "with", "z"], "esm/modeling_esmfold.py:EsmFoldBackboneUpdate.__init__": ["ModelFoldBackboneUpdate", "ModelFoldLinear", "Module", "__init__", "class", "config", "def", "final", "init", "linear", "nn", "self", "sequence_dim", "super"], "esm/modeling_esmfold.py:EsmFoldBackboneUpdate.forward": ["ModelFoldBackboneUpdate", "Module", "Tensor", "class", "def", "forward", "linear", "nn", "return", "s", "self", "torch", "tuple", "update"], "esm/modeling_esmfold.py:EsmFoldStructureModuleTransitionLayer.__init__": ["ModelFoldLinear", "ModelFoldStructureModuleTransitionLayer", "Module", "ReLU", "__init__", "class", "config", "def", "final", "init", "linear_1", "linear_2", "linear_3", "nn", "relu", "self", "sequence_dim", "super"], "esm/modeling_esmfold.py:EsmFoldStructureModuleTransitionLayer.forward": ["ModelFoldStructureModuleTransitionLayer", "Module", "class", "def", "forward", "linear_1", "linear_2", "linear_3", "nn", "relu", "return", "s", "s_initial", "self"], "esm/modeling_esmfold.py:EsmFoldStructureModuleTransition.__init__": ["Dropout", "LayerNorm", "ModelFoldStructureModuleTransition", "ModelFoldStructureModuleTransitionLayer", "Module", "ModuleList", "_", "__init__", "append", "class", "config", "def", "dropout", "dropout_rate", "for", "in", "l", "layer_norm", "layers", "nn", "num_transition_layers", "range", "self", "sequence_dim", "super"], "esm/modeling_esmfold.py:EsmFoldStructureModuleTransition.forward": ["ModelFoldStructureModuleTransition", "Module", "class", "def", "dropout", "for", "forward", "in", "l", "layer_norm", "layers", "nn", "return", "s", "self"], "esm/modeling_esmfold.py:EsmFoldStructureModule.__init__": ["Dropout", "LayerNorm", "ModelFoldAngleResnet", "ModelFoldBackboneUpdate", "ModelFoldInvariantPointAttention", "ModelFoldLinear", "ModelFoldStructureModule", "ModelFoldStructureModuleTransition", "Module", "__init__", "angle_resnet", "bb_update", "class", "config", "def", "dropout_rate", "ipa", "ipa_dropout", "layer_norm_ipa", "layer_norm_s", "layer_norm_z", "linear_in", "nn", "pairwise_dim", "self", "sequence_dim", "super", "transition"], "esm/modeling_esmfold.py:EsmFoldStructureModule.forward": ["False", "ModelFoldStructureModule", "Module", "None", "Rigid", "Rotation", "_offload_inference", "_z_reference_list", "aatype", "all_frames_to_global", "angle_resnet", "angles", "append", "assert", "backb_to_global", "bb_update", "class", "compose_q_update_vec", "config", "cpu", "def", "del", "device", "dict_multimap", "dtype", "evoformer_output_dict", "fmt", "for", "forward", "frames", "frames_and_literature_positions_to_atom14_pos", "get_rot_mats", "get_rots", "get_trans", "getrefcount", "i", "identity", "if", "in", "ipa", "ipa_dropout", "is", "layer_norm_ipa", "layer_norm_s", "layer_norm_z", "linear_in", "mask", "new_ones", "nn", "num_blocks", "outputs", "pair", "positions", "pred_xyz", "preds", "quat", "quats", "range", "return", "rigids", "rot_mats", "s", "s_initial", "scale_translation", "scaled_rigids", "self", "shape", "sidechain_frames", "single", "stack", "states", "stop_rot_gradient", "sys", "to", "to_tensor_4x4", "to_tensor_7", "torch", "torsion_angles_to_frames", "training", "trans_scale_factor", "transition", "unnormalized_angles", "z", "z_reference_list"], "esm/modeling_esmfold.py:EsmFoldStructureModule._init_residue_constants": ["False", "ModelFoldStructureModule", "Module", "_init_residue_constants", "atom_mask", "class", "def", "default_frames", "device", "dtype", "float_dtype", "group_idx", "hasattr", "if", "lit_positions", "nn", "not", "persistent", "register_buffer", "requires_grad", "residue_constants", "restype_atom14_mask", "restype_atom14_rigid_group_positions", "restype_atom14_to_rigid_group", "restype_rigid_group_default_frame", "self", "tensor", "torch"], "esm/modeling_esmfold.py:EsmFoldStructureModule.torsion_angles_to_frames": ["ModelFoldStructureModule", "Module", "_init_residue_constants", "alpha", "class", "def", "default_frames", "device", "dtype", "f", "nn", "r", "return", "self", "torsion_angles_to_frames"], "esm/modeling_esmfold.py:EsmFoldStructureModule.frames_and_literature_positions_to_atom14_pos": ["ModelFoldStructureModule", "Module", "_init_residue_constants", "atom_mask", "class", "def", "default_frames", "device", "dtype", "f", "frames_and_literature_positions_to_atom14_pos", "get_rots", "group_idx", "lit_positions", "nn", "r", "return", "self"], "esm/modeling_esmfold.py:EsmFoldingTrunk.__init__": ["Embedding", "LayerNorm", "Linear", "ModelFoldRelativePosition", "ModelFoldStructureModule", "ModelFoldTriangularSelfAttentionBlock", "ModelFoldingTrunk", "Module", "ModuleList", "_", "__init__", "blocks", "c_s", "c_z", "chunk_size", "class", "config", "def", "detach", "for", "in", "nn", "num_blocks", "pairwise_dim", "pairwise_positional_embedding", "pairwise_state_dim", "range", "recycle_bins", "recycle_disto", "recycle_s_norm", "recycle_z_norm", "self", "sequence_dim", "sequence_state_dim", "structure_module", "super", "trunk2sm_s", "trunk2sm_z", "weight", "zero_"], "esm/modeling_esmfold.py:EsmFoldingTrunk.set_chunk_size": ["ModelFoldingTrunk", "Module", "chunk_size", "class", "def", "nn", "self", "set_chunk_size"], "esm/modeling_esmfold.py:EsmFoldingTrunk.forward": ["ContextManagers", "ModelFoldingTrunk", "Module", "None", "Number", "ValueError", "be", "block", "blocks", "chunk_size", "class", "config", "def", "detach", "device", "distogram", "dtype", "else", "float", "for", "forward", "if", "in", "int64", "is", "mask", "max_recycles", "must", "negative", "nn", "no_grad", "no_recycles", "not", "of", "pair", "pair_feats", "pairwise_positional_embedding", "positions", "raise", "range", "recycle_bins", "recycle_disto", "recycle_idx", "recycle_s", "recycle_s_norm", "recycle_z", "recycle_z_norm", "recycles", "residue_index", "residx", "return", "s", "s_s", "s_s_0", "s_z", "s_z_0", "self", "seq_feats", "shape", "single", "structure", "structure_module", "to", "torch", "true_aa", "trunk2sm_s", "trunk2sm_z", "trunk_iter", "with", "z", "zeros", "zeros_like"], "esm/modeling_esmfold.py:EsmFoldingTrunk.distogram": ["C", "CA", "CB", "ModelFoldingTrunk", "Module", "N", "None", "True", "a", "b", "bins", "boundaries", "c", "chunk", "class", "coords", "cross", "def", "device", "dim", "distogram", "dists", "for", "in", "keepdims", "linspace", "max_bin", "min_bin", "nn", "num_bins", "pow", "return", "squeeze", "sum", "torch", "x"], "esm/modeling_esmfold.py:EsmForProteinFolding._init_weights": ["ModelForProteinFolding", "ModelPreTrainedModel", "_af2_to_Model_from_vocab_list", "_init_weights", "af2_to_Model", "class", "config", "copy_", "def", "if", "init", "isinstance", "module", "self", "super", "vocab_list"], "esm/modeling_esmfold.py:EsmForProteinFolding.__init__": ["Embedding", "False", "LayerNorm", "Linear", "Model", "ModelFoldingTrunk", "ModelForProteinFolding", "ModelModel", "ModelPreTrainedModel", "Model_attns", "Model_dict_cls_idx", "Model_dict_eos_idx", "Model_dict_mask_idx", "Model_dict_padding_idx", "Model_feats", "Model_layers", "Model_s_combine", "Model_s_mlp", "Modelfold_config", "Parameter", "ReLU", "Sequential", "__init__", "_af2_to_Model_from_vocab_list", "add_pooling_layer", "af2_to_Model", "c_s", "c_z", "class", "cls", "config", "def", "distogram_bins", "distogram_head", "embed_aa", "embedding", "eos", "fp16_Model", "half", "hidden_size", "if", "index", "lddt_bins", "lddt_head", "lddt_head_hid_dim", "lm_head", "mask", "mask_idx", "n_tokens_embed", "nn", "num_attention_heads", "num_hidden_layers", "pad", "pad_idx", "padding_idx", "pairwise_state_dim", "post_init", "ptm_head", "register_buffer", "requires_grad_", "residue_constants", "restype_num", "self", "sequence_dim", "sequence_state_dim", "structure_module", "structure_module_config", "super", "torch", "trunk", "trunk_config", "unk_idx", "vocab_list", "zeros"], "esm/modeling_esmfold.py:EsmForProteinFolding._af2_to_esm_from_vocab_list": ["ModelForProteinFolding", "ModelPreTrainedModel", "Model_reorder", "Tensor", "_af2_to_Model_from_vocab_list", "class", "def", "for", "in", "index", "list", "pad", "residue_constants", "restypes_with_x", "return", "str", "tensor", "torch", "v", "vocab_list"], "esm/modeling_esmfold.py:EsmForProteinFolding.forward": ["B", "False", "L", "ModelForProteinFolding", "ModelForProteinFoldingOutput", "ModelPreTrainedModel", "Model_ablate_sequence", "Model_s", "Model_s_combine", "Model_s_mlp", "Modelaa", "Modelfold_config", "None", "Tensor", "aa", "aatype", "af2_idx_to_Model_idx", "angles", "arange", "atom14_atom_exists", "atom37_atom_exists", "attention_mask", "bert_mask", "bins", "bool", "categorical_lddt", "cfg", "class", "compute_language_model_representations", "compute_predicted_aligned_error", "compute_tm", "config", "def", "detach", "device", "dict", "disto_logits", "distogram_bins", "distogram_head", "distogram_logits", "dtype", "else", "embed_aa", "embedding", "expand_as", "for", "forward", "frames", "if", "in", "input_ids", "int", "is", "items", "k", "lddt_bins", "lddt_head", "lm_head", "lm_logits", "make_atom14_masks", "masked_aa", "masking_pattern", "max_bin", "mlm_targets", "new_zeros", "no_bins", "no_recycles", "not", "num_recycles", "ones_like", "output_hidden_states", "pairwise_state_dim", "plddt", "position_ids", "positions", "ptm", "ptm_head", "ptm_logits", "r", "reshape", "residue_index", "return", "s_s", "s_s_0", "s_z", "s_z_0", "self", "shape", "sidechain_frames", "softmax", "squeeze", "states", "structure", "to", "torch", "transpose", "trunk", "unnormalized_angles", "unsqueeze", "update", "v"], "esm/modeling_esmfold.py:EsmForProteinFolding.af2_idx_to_esm_idx": ["ModelForProteinFolding", "ModelPreTrainedModel", "aa", "af2_idx_to_Model_idx", "af2_to_Model", "class", "def", "device", "if", "mask", "masked_fill", "return", "self", "to"], "esm/modeling_esmfold.py:EsmForProteinFolding.compute_language_model_representations": ["B", "L", "Model", "ModelForProteinFolding", "ModelPreTrainedModel", "Model_dict_cls_idx", "Model_dict_eos_idx", "Model_dict_padding_idx", "Model_feats", "Model_hidden_states", "Model_s", "Model_s_combine", "Modelaa", "Modelfold_config", "Tensor", "True", "attention_mask", "bos", "bosi", "bypass_lm", "cat", "class", "compute_language_model_representations", "config", "def", "device", "dim", "eos", "eosi", "hidden_states", "if", "new_full", "next", "output_hidden_states", "parameters", "range", "return", "self", "shape", "size", "stack", "sum", "torch", "zeros"], "esm/modeling_esmfold.py:EsmForProteinFolding.bert_mask": ["ModelForProteinFolding", "ModelPreTrainedModel", "Model_dict_mask_idx", "Modelaa", "aa", "bert_mask", "class", "clone", "def", "mask", "mask_idx", "new_Modelaa", "new_aa", "pattern", "return", "self", "target"], "esm/modeling_esmfold.py:EsmForProteinFolding.infer": ["ModelForProteinFolding", "ModelPreTrainedModel", "None", "True", "aatype", "arange", "argmax", "class", "collate_dense_tensors", "def", "device", "dim", "else", "expand", "for", "forward", "from_numpy", "if", "in", "infer", "is", "isinstance", "len", "list", "lst", "map_unknown_to_x", "mapping", "mask", "ndim", "new_ones", "next", "parameters", "position_ids", "residue_constants", "restype_order_with_x", "return", "self", "seq", "seqs", "sequence", "sequence_to_onehot", "shape", "str", "to", "torch", "unsqueeze"], "esm/modeling_esmfold.py:EsmForProteinFolding.output_to_pdb": ["ModelForProteinFolding", "ModelPreTrainedModel", "OFProtein", "aa", "aatype", "append", "atom14_to_atom37", "atom37_atom_exists", "atom_mask", "atom_positions", "b_factors", "class", "cpu", "def", "dict", "final_atom_mask", "final_atom_positions", "for", "i", "in", "items", "k", "list", "mask", "numpy", "output", "output_to_pdb", "pdbs", "plddt", "positions", "pred", "pred_pos", "range", "resid", "residue_index", "return", "shape", "str", "to", "to_pdb", "v"], "esm/modeling_esmfold.py:EsmForProteinFolding.infer_pdb": ["ModelForProteinFolding", "ModelPreTrainedModel", "args", "assert", "class", "def", "infer", "infer_pdb", "isinstance", "kwargs", "output", "output_to_pdb", "return", "self", "seqs", "str"], "esm/modeling_esmfold.py:EsmForProteinFolding.infer_pdbs": ["ModelForProteinFolding", "ModelPreTrainedModel", "args", "class", "def", "infer", "infer_pdbs", "kwargs", "list", "output", "output_to_pdb", "return", "self", "seqs", "str"], "evolla/modeling_evolla.py:create_position_ids_from_input_ids": ["Model_position_ids_from_input_ids", "cumsum", "def", "dim", "incremental_indices", "input_ids", "int", "long", "mask", "ne", "padding_idx", "return", "torch", "type_as"], "evolla/modeling_evolla.py:EvollaSaProtEmbeddings.__init__": ["Dropout", "Embedding", "False", "LayerNorm", "ModelSaProtEmbeddings", "Module", "None", "__init__", "absolute", "arange", "class", "config", "def", "dropout", "else", "emb_layer_norm_before", "eps", "expand", "getattr", "hidden_dropout_prob", "hidden_size", "if", "layer_norm", "layer_norm_eps", "mask_token_id", "max_position_embeddings", "nn", "pad_token_id", "padding_idx", "persistent", "position_embedding_type", "position_embeddings", "position_ids", "register_buffer", "self", "super", "token_dropout", "torch", "vocab_size", "word_embeddings"], "evolla/modeling_evolla.py:EvollaSaProtEmbeddings.forward": ["ModelSaProtEmbeddings", "Module", "None", "absolute", "and", "attention_mask", "class", "create_position_ids_from_input_ids", "create_position_ids_from_inputs_embeds", "def", "dtype", "else", "embeddings", "float", "forward", "if", "input_ids", "inputs_embeds", "is", "layer_norm", "mask_ratio_observed", "mask_ratio_train", "mask_token_id", "masked_fill", "nn", "not", "padding_idx", "position_embedding_type", "position_embeddings", "position_ids", "return", "self", "shape", "src_lengths", "sum", "to", "token_dropout", "unsqueeze", "word_embeddings"], "evolla/modeling_evolla.py:EvollaSaProtEmbeddings.create_position_ids_from_inputs_embeds": ["ModelSaProtEmbeddings", "Module", "arange", "class", "create_position_ids_from_inputs_embeds", "def", "device", "dtype", "expand", "input_shape", "inputs_embeds", "long", "nn", "padding_idx", "position_ids", "return", "self", "sequence_length", "size", "torch", "unsqueeze"], "evolla/modeling_evolla.py:rotate_half_esm": ["Model_half_esm", "cat", "chunk", "def", "dim", "return", "torch", "x", "x1", "x2"], "evolla/modeling_evolla.py:apply_rotary_pos_emb_esm": ["Model_rotary_pos_emb_esm", "cos", "def", "return", "rotate_half_esm", "shape", "sin", "x"], "evolla/modeling_evolla.py:EvollaSaProtRotaryEmbedding.__init__": ["ModelSaProtRotaryEmbedding", "Module", "None", "__init__", "_cos_cached", "_seq_len_cached", "_sin_cached", "arange", "class", "def", "dim", "dtype", "float", "int", "int64", "inv_freq", "nn", "register_buffer", "self", "super", "torch"], "evolla/modeling_evolla.py:EvollaSaProtRotaryEmbedding._update_cos_sin_tables": ["ModelSaProtRotaryEmbedding", "Module", "None", "_cos_cached", "_seq_len_cached", "_sin_cached", "_update_cos_sin_tables", "arange", "cat", "class", "cos", "def", "device", "dim", "emb", "freqs", "if", "inv_freq", "nn", "or", "outer", "return", "self", "seq_dimension", "seq_len", "shape", "sin", "t", "to", "torch", "type_as", "x"], "evolla/modeling_evolla.py:EvollaSaProtRotaryEmbedding.forward": ["ModelSaProtRotaryEmbedding", "Module", "Tensor", "_cos_cached", "_sin_cached", "_update_cos_sin_tables", "apply_rotary_pos_emb_esm", "class", "def", "dtype", "forward", "k", "nn", "q", "return", "self", "seq_dimension", "to", "torch", "tuple"], "evolla/modeling_evolla.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "contiguous", "def", "dim", "dropout", "float", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "shape", "size", "softmax", "torch", "training", "transpose", "value"], "evolla/modeling_evolla.py:EvollaSaProtSelfAttention.__init__": ["False", "Linear", "ModelSaProtRotaryEmbedding", "ModelSaProtSelfAttention", "Module", "None", "The", "ValueError", "__init__", "a", "absolute", "all_head_size", "and", "attention", "attention_head_size", "attention_probs_dropout_prob", "class", "config", "def", "dim", "dropout", "embedding_size", "f", "getattr", "hasattr", "heads", "hidden", "hidden_size", "if", "int", "is", "is_causal", "is_cross_attention", "is_decoder", "key", "layer_idx", "multiple", "nn", "not", "num_attention_heads", "number", "of", "or", "position_embedding_type", "query", "raise", "rotary", "rotary_embeddings", "scaling", "self", "size", "super", "the", "value"], "evolla/modeling_evolla.py:EvollaSaProtSelfAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "FloatTensor", "ModelSaProtSelfAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_attn_implementation", "attention_head_size", "attention_interface", "attention_mask", "attn_output", "attn_weights", "batch_size", "class", "config", "contiguous", "current_states", "def", "dropout", "eager", "eager_attention_forward", "else", "encoder_attention_mask", "encoder_hidden_states", "forward", "hidden_shape", "hidden_states", "if", "is", "is_cross_attention", "key", "key_layer", "kwargs", "nn", "not", "position_embedding_type", "query", "query_layer", "reshape", "return", "rotary", "rotary_embeddings", "scaling", "self", "seq_length", "shape", "torch", "training", "transpose", "tuple", "value", "value_layer", "view"], "evolla/modeling_evolla.py:EvollaSaProtSelfOutput.__init__": ["Dropout", "Linear", "ModelSaProtSelfOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "hidden_dropout_prob", "hidden_size", "nn", "self", "super"], "evolla/modeling_evolla.py:EvollaSaProtSelfOutput.forward": ["ModelSaProtSelfOutput", "Module", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self"], "evolla/modeling_evolla.py:EvollaSaProtAttention.__init__": ["False", "LayerNorm", "ModelSaProtAttention", "ModelSaProtSelfAttention", "ModelSaProtSelfOutput", "Module", "None", "__init__", "class", "config", "def", "eps", "hidden_size", "is_cross_attention", "layer_idx", "layer_norm_eps", "nn", "output", "self", "super"], "evolla/modeling_evolla.py:EvollaSaProtAttention.forward": ["LayerNorm", "ModelSaProtAttention", "Module", "None", "TransformersKwargs", "Unpack", "_", "attention_mask", "attn_output", "class", "def", "encoder_attention_mask", "encoder_hidden_states", "forward", "hidden_states", "hidden_states_ln", "kwargs", "nn", "output", "return", "self"], "evolla/modeling_evolla.py:gelu": ["Model", "def", "erf", "math", "return", "sqrt", "torch", "x"], "evolla/modeling_evolla.py:EvollaSaProtIntermediate.__init__": ["Linear", "ModelSaProtIntermediate", "Module", "__init__", "class", "config", "def", "dense", "hidden_size", "intermediate_size", "nn", "self", "super"], "evolla/modeling_evolla.py:EvollaSaProtIntermediate.forward": ["ModelSaProtIntermediate", "Module", "Tensor", "class", "def", "dense", "forward", "gelu", "hidden_states", "nn", "return", "self", "torch"], "evolla/modeling_evolla.py:EvollaSaProtOutput.__init__": ["Dropout", "Linear", "ModelSaProtOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "hidden_dropout_prob", "hidden_size", "intermediate_size", "nn", "self", "super"], "evolla/modeling_evolla.py:EvollaSaProtOutput.forward": ["ModelSaProtOutput", "Module", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self"], "evolla/modeling_evolla.py:EvollaSaProtLayer.__init__": ["GradientCheckpointingLayer", "LayerNorm", "ModelSaProtAttention", "ModelSaProtIntermediate", "ModelSaProtLayer", "ModelSaProtOutput", "RuntimeError", "True", "__init__", "a", "add_cross_attention", "added", "as", "attention", "be", "chunk_size_feed_forward", "class", "config", "cross", "crossattention", "decoder", "def", "eps", "f", "hidden_size", "if", "intermediate", "is", "is_cross_attention", "is_decoder", "layer_norm_eps", "model", "nn", "not", "output", "raise", "self", "seq_len_dim", "should", "super", "used"], "evolla/modeling_evolla.py:EvollaSaProtLayer.forward": ["AttributeError", "GradientCheckpointingLayer", "If", "ModelSaProtLayer", "None", "TransformersKwargs", "True", "Unpack", "add_cross_attention", "and", "are", "attention", "attention_mask", "attention_output", "be", "by", "class", "config", "cross", "crossattention", "def", "encoder_attention_mask", "encoder_hidden_states", "f", "feed_forward_chunk", "forward", "has", "hasattr", "hidden_states", "if", "instantiated", "is", "is_decoder", "kwargs", "layer_output", "layers", "not", "passed", "raise", "return", "self", "setting", "to", "with"], "evolla/modeling_evolla.py:EvollaSaProtLayer.feed_forward_chunk": ["GradientCheckpointingLayer", "LayerNorm", "ModelSaProtLayer", "attention_output", "attention_output_ln", "class", "def", "feed_forward_chunk", "intermediate", "intermediate_output", "layer_output", "output", "return", "self"], "evolla/modeling_evolla.py:EvollaSaProtEncoder.__init__": ["False", "LayerNorm", "ModelSaProtEncoder", "ModelSaProtLayer", "Module", "ModuleList", "_", "__init__", "class", "config", "def", "emb_layer_norm_after", "eps", "for", "gradient_checkpointing", "hidden_size", "in", "layer", "layer_norm_eps", "nn", "num_hidden_layers", "range", "self", "super"], "evolla/modeling_evolla.py:EvollaSaProtEncoder.forward": ["BaseModelOutputWithCrossAttentions", "ModelSaProtEncoder", "Module", "None", "TransformersKwargs", "Unpack", "attention_mask", "class", "def", "emb_layer_norm_after", "encoder_attention_mask", "encoder_hidden_states", "enumerate", "for", "forward", "hidden_states", "i", "if", "in", "kwargs", "last_hidden_state", "layer", "layer_module", "nn", "return", "self"], "evolla/modeling_evolla.py:EvollaSaProtPooler.__init__": ["Linear", "ModelSaProtPooler", "Module", "Tanh", "__init__", "activation", "class", "config", "def", "dense", "hidden_size", "nn", "self", "super"], "evolla/modeling_evolla.py:EvollaSaProtPooler.forward": ["ModelSaProtPooler", "Module", "Tensor", "activation", "class", "def", "dense", "first_token_tensor", "forward", "hidden_states", "nn", "pooled_output", "return", "self", "torch"], "evolla/modeling_evolla.py:EvollaSaProtPreTrainedModel._init_weights": ["ModelSaProtPreTrainedModel", "ModelSaProtRotaryEmbedding", "PreTrainedModel", "_init_weights", "arange", "class", "copy_", "def", "dim", "dtype", "float", "if", "init", "int64", "inv_freq", "isinstance", "module", "self", "super", "torch"], "evolla/modeling_evolla.py:EvollaSaProtProteinEncoder.__init__": ["ModelSaProtEmbeddings", "ModelSaProtEncoder", "ModelSaProtPreTrainedModel", "ModelSaProtProteinEncoder", "SaProtConfig", "__init__", "class", "config", "def", "embeddings", "encoder", "post_init", "self", "super"], "evolla/modeling_evolla.py:EvollaSaProtProteinEncoder.get_input_embeddings": ["ModelSaProtPreTrainedModel", "ModelSaProtProteinEncoder", "class", "def", "embeddings", "get_input_embeddings", "return", "self", "word_embeddings"], "evolla/modeling_evolla.py:EvollaSaProtProteinEncoder.set_input_embeddings": ["ModelSaProtPreTrainedModel", "ModelSaProtProteinEncoder", "class", "def", "embeddings", "self", "set_input_embeddings", "value", "word_embeddings"], "evolla/modeling_evolla.py:EvollaSaProtProteinEncoder.forward": ["BaseModelOutputWithPoolingAndCrossAttentions", "ModelSaProtPreTrainedModel", "ModelSaProtProteinEncoder", "None", "Tensor", "attention_mask", "attentions", "batch_size", "class", "config", "create_bidirectional_mask", "cross_attentions", "def", "device", "embeddings", "encoder", "encoder_outputs", "forward", "hidden_states", "if", "input_embeds", "input_ids", "input_shape", "inputs_embeds", "is", "kwargs", "last_hidden_state", "ones", "return", "self", "seq_length", "sequence_output", "size", "torch", "tuple"], "evolla/modeling_evolla.py:EvollaSequenceCompressorAttention.__init__": ["False", "LayerNorm", "Linear", "ModelSequenceCompressorAttention", "Module", "__init__", "bias", "class", "def", "dim", "dim_head", "heads", "inner_dim", "nn", "norm_latents", "norm_media", "scale", "self", "super", "to_kv", "to_out", "to_q"], "evolla/modeling_evolla.py:EvollaSequenceCompressorAttention.forward": ["ModelSequenceCompressorAttention", "Module", "None", "True", "amax", "attn", "bool", "bs", "cat", "chunk", "class", "def", "detach", "device", "dim", "forward", "h", "heads", "k", "keepdim", "kv_input", "latents", "mask", "mask_exp", "masked_fill", "matmul", "nh", "nn", "norm_latents", "norm_media", "okd", "ones", "ones_exp", "out", "permute", "q", "reshape", "return", "scale", "self", "shape", "sim", "size", "skd", "softmax", "to", "to_kv", "to_out", "to_q", "torch", "transpose", "v", "view", "x"], "evolla/modeling_evolla.py:EvollaFeedForward.__init__": ["False", "GELU", "LayerNorm", "Linear", "ModelFeedForward", "Module", "__init__", "activation", "bias", "class", "def", "dim", "fc1", "fc2", "inner_dim", "int", "mult", "nn", "norm", "self", "super"], "evolla/modeling_evolla.py:EvollaFeedForward.forward": ["ModelFeedForward", "Module", "activation", "class", "def", "fc1", "fc2", "forward", "nn", "norm", "return", "self", "x"], "evolla/modeling_evolla.py:EvollaSequenceCompressorResampler.__init__": ["LayerNorm", "Linear", "ModelConfig", "ModelFeedForward", "ModelSequenceCompressorAttention", "ModelSequenceCompressorResampler", "Module", "ModuleList", "Parameter", "True", "_", "__init__", "append", "class", "config", "def", "dim", "dim_head", "for", "heads", "hidden_size", "in", "latents", "layers", "mult", "nn", "norm", "num_latents", "protein_encoder_config", "protein_projector", "protein_repr_dim", "randn", "range", "requires_grad", "resampler_depth", "resampler_dim_head", "resampler_ff_mult", "resampler_heads", "resampler_num_latents", "self", "super", "torch"], "evolla/modeling_evolla.py:EvollaSequenceCompressorResampler.forward": ["ModelSequenceCompressorResampler", "Module", "None", "_", "attn", "b", "bs", "cat", "class", "def", "device", "dim", "dtype", "embeds", "ff", "for", "forward", "in", "latent_mask", "latents", "layers", "mask", "nn", "norm", "num_latents", "ones", "protein_projector", "return", "self", "shape", "to", "torch", "transformed_feature", "view"], "evolla/modeling_evolla.py:EvollaProteinEncoder.__init__": ["ModelConfig", "ModelProteinEncoder", "ModelSaProtProteinEncoder", "ModelSequenceCompressorResampler", "Module", "__init__", "class", "config", "def", "model", "nn", "protein_encoder_config", "self", "sequence_compressor_resampler", "super"], "evolla/modeling_evolla.py:EvollaProteinEncoder.forward": ["FloatTensor", "LongTensor", "ModelProteinEncoder", "ModelProteinEncoderModelOutput", "Module", "attention_mask", "class", "def", "forward", "input_ids", "kwargs", "last_hidden_state", "model", "nn", "protein_embeds", "protein_output", "return", "self", "sequence_compressor_output", "sequence_compressor_resampler", "sequence_repr", "torch"], "evolla/modeling_evolla.py:EvollaSequenceAlignerCrossAttention.__init__": ["Dropout", "Linear", "ModelFeedForward", "ModelRMSNorm", "ModelSequenceAlignerCrossAttention", "Module", "None", "Parameter", "__init__", "aligner_attention_probs_dropout_prob", "aligner_enable_bias", "aligner_ffn_mult", "all_head_size", "attention_head_size", "attention_norm", "attention_probs_dropout_prob", "bias", "class", "config", "def", "dropout", "else", "enable_bias", "ff", "ffn_mult", "gate_attention", "gate_ffw", "hidden_size", "if", "int", "is", "key_msa", "key_protein", "key_structure", "msa_encoder_dim", "nn", "not", "num_attention_heads", "out_proj", "protein_encoder_dim", "query", "scale", "self", "structure_encoder_dim", "super", "tensor", "torch", "value_msa", "value_protein", "value_structure"], "evolla/modeling_evolla.py:EvollaSequenceAlignerCrossAttention.cross_attention": ["At", "ModelSequenceAlignerCrossAttention", "Module", "None", "Softmax", "True", "ValueError", "_", "all_head_size", "amax", "and", "attention", "attention_head_size", "attention_mask", "attention_norm", "attention_probs", "attention_scores", "attn_weights", "be", "bool", "cat", "class", "context_layer", "contiguous", "cross", "cross_attention", "def", "detach", "device", "dim", "dtype", "else", "finfo", "for", "if", "in", "is", "keepdim", "key_layer", "key_layer_msa", "key_layer_protein", "key_layer_structure", "key_msa", "key_protein", "key_structure", "kv_attn_mask", "least", "masked_fill", "matmul", "min", "modality", "msa_key_value_states", "msa_kv_attn_mask", "new_context_layer_shape", "new_key_layer_shape", "new_query_layer_shape", "new_value_layer_shape", "nn", "not", "num_attention_heads", "one", "ones", "out_proj", "permute", "protein_key_value_states", "protein_kv_attn_mask", "provided", "query", "query_attn_mask", "query_layer", "query_states", "raise", "return", "scale", "self", "should", "size", "structure_key_value_states", "structure_kv_attn_mask", "to", "torch", "transpose", "value_layer", "value_layer_msa", "value_layer_protein", "value_layer_structure", "value_msa", "value_protein", "value_structure", "view"], "evolla/modeling_evolla.py:EvollaSequenceAlignerCrossAttention.forward": ["ModelSequenceAlignerCrossAttention", "Module", "None", "T", "and", "any", "bs", "class", "cross_attention", "def", "device", "dim", "else", "expand", "ff", "forward", "gate_attention", "gate_ffw", "hidden_states", "if", "is", "msa_batch_mask", "msa_key_value_states", "msa_kv_attn_mask", "msa_kv_seq_len", "msa_kv_states", "nn", "not", "ones", "or", "past_key_values", "protein_batch_mask", "protein_key_value_states", "protein_kv_attn_mask", "protein_kv_seq_len", "protein_kv_states", "query_attn_mask", "query_states", "residual", "return", "self", "shape", "size", "structure_batch_mask", "structure_key_value_states", "structure_kv_attn_mask", "structure_kv_seq_len", "structure_kv_states", "tanh", "to", "torch"], "evolla/modeling_evolla.py:EvollaRMSNorm.__init__": ["ModelRMSNorm", "Module", "Parameter", "__init__", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "evolla/modeling_evolla.py:EvollaRMSNorm.forward": ["ModelRMSNorm", "Module", "True", "class", "def", "dtype", "float32", "forward", "hidden_states", "input_dtype", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "evolla/modeling_evolla.py:EvollaRMSNorm.extra_repr": ["ModelRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "variance_epsilon", "weight"], "evolla/modeling_evolla.py:EvollaRotaryEmbedding.__init__": ["Callable", "False", "ModelConfig", "ModelRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "evolla/modeling_evolla.py:EvollaRotaryEmbedding.compute_default_rope_parameters": ["ModelConfig", "ModelRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "evolla/modeling_evolla.py:EvollaRotaryEmbedding.forward": ["False", "ModelRotaryEmbedding", "Module", "None", "and", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "evolla/modeling_evolla.py:EvollaMLP.__init__": ["ACT2FN", "Linear", "ModelMLP", "Module", "__init__", "act_fn", "bias", "class", "config", "def", "down_proj", "gate_proj", "hidden_act", "hidden_size", "intermediate_size", "mlp_bias", "nn", "self", "super", "up_proj"], "evolla/modeling_evolla.py:EvollaMLP.forward": ["ModelMLP", "Module", "act_fn", "class", "def", "down_proj", "forward", "gate_proj", "nn", "return", "self", "up_proj", "x"], "evolla/modeling_evolla.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "evolla/modeling_evolla.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cos", "def", "k", "k_embed", "q", "q_embed", "return", "rotate_half", "sin", "unsqueeze", "unsqueeze_dim"], "evolla/modeling_evolla.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "evolla/modeling_evolla.py:EvollaAttention.__init__": ["Linear", "ModelAttention", "ModelConfig", "Module", "True", "__init__", "attention_bias", "attention_dropout", "bias", "class", "config", "def", "getattr", "head_dim", "hidden_size", "int", "is_causal", "k_proj", "layer_idx", "nn", "num_attention_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "q_proj", "scaling", "self", "super", "v_proj"], "evolla/modeling_evolla.py:EvollaAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "LongTensor", "ModelAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_attn_implementation", "apply_rotary_pos_emb", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "cache_kwargs", "cache_position", "class", "config", "contiguous", "cos", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "head_dim", "hidden_shape", "hidden_states", "if", "input_shape", "is", "k_proj", "key_states", "kwargs", "layer_idx", "nn", "not", "o_proj", "past_key_values", "position_embeddings", "q_proj", "query_states", "reshape", "return", "scaling", "self", "shape", "sin", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "view"], "evolla/modeling_evolla.py:EvollaDecoderLayer.__init__": ["GradientCheckpointingLayer", "ModelAttention", "ModelConfig", "ModelDecoderLayer", "ModelMLP", "ModelRMSNorm", "ModelSequenceAlignerCrossAttention", "__init__", "adapter", "aligner_num_add_layers", "class", "config", "def", "eps", "hidden_size", "if", "input_layernorm", "int", "layer_idx", "max", "mlp", "num_hidden_layers", "post_attention_layernorm", "protein_encoder_dim", "rms_norm_eps", "self", "self_attn", "super"], "evolla/modeling_evolla.py:EvollaDecoderLayer.forward": ["Cache", "False", "GradientCheckpointingLayer", "LongTensor", "ModelDecoderLayer", "None", "Tensor", "_", "adapter", "attention_mask", "bool", "cache_position", "class", "def", "forward", "hasattr", "hidden_states", "if", "input_layernorm", "kwargs", "mlp", "msa_batch_mask", "msa_kv_states", "past_key_values", "position_embeddings", "position_ids", "post_attention_layernorm", "protein_batch_mask", "protein_kv_states", "query_attn_mask", "query_states", "residual", "return", "self", "self_attn", "structure_batch_mask", "structure_kv_states", "torch", "tuple", "use_cache"], "evolla/modeling_evolla.py:EvollaPreTrainedModel._init_weights": ["ModelPreTrainedModel", "ModelSequenceAlignerCrossAttention", "ModelSequenceCompressorResampler", "PreTrainedModel", "_init_weights", "attention_norm", "class", "config", "def", "elif", "gate_attention", "gate_ffw", "if", "init", "initializer_range", "isinstance", "latents", "mean", "module", "normal_", "ones_", "self", "std", "super", "weight", "zeros_"], "evolla/modeling_evolla.py:EvollaModel.__init__": ["Embedding", "False", "ModelConfig", "ModelDecoderLayer", "ModelModel", "ModelPreTrainedModel", "ModelProteinEncoder", "ModelRMSNorm", "ModelRotaryEmbedding", "ModuleList", "__init__", "class", "config", "def", "embed_tokens", "eps", "for", "getattr", "gradient_checkpointing", "hidden_size", "in", "layer_idx", "layers", "nn", "norm", "num_hidden_layers", "pad_token_id", "padding_idx", "post_init", "protein_encoder", "range", "rms_norm_eps", "rotary_emb", "self", "super", "vocab_size"], "evolla/modeling_evolla.py:EvollaModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embed_tokens", "get_input_embeddings", "return", "self"], "evolla/modeling_evolla.py:EvollaModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embed_tokens", "self", "set_input_embeddings", "value"], "evolla/modeling_evolla.py:EvollaModel.forward": ["BaseModelOutputWithPast", "Cache", "DynamicCache", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "True", "ValueError", "You", "and", "arange", "attention_mask", "bool", "cache_position", "causal_mask", "class", "config", "create_causal_mask", "decoder_layer", "def", "device", "else", "embed_tokens", "exactly", "for", "forward", "get_seq_length", "hidden_states", "if", "in", "input_embeds", "input_ids", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layers", "msa_batch_mask", "msa_feats", "msa_kv_states", "must", "norm", "not", "of", "one", "or", "output", "past_key_values", "past_seen_tokens", "position_embeddings", "position_ids", "protein_attention_mask", "protein_batch_mask", "protein_encoder", "protein_feats", "protein_input_ids", "protein_kv_states", "protein_outputs", "query_attn_mask", "r", "raise", "return", "rotary_emb", "self", "sequence_compressor_output", "shape", "specify", "structure_batch_mask", "structure_feats", "structure_kv_states", "tensor", "torch", "tuple", "unsqueeze", "use_cache"], "evolla/modeling_evolla.py:EvollaForProteinText2Text.__init__": ["False", "GenerationMixin", "Linear", "ModelForProteinText2Text", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "vocab_size"], "evolla/modeling_evolla.py:EvollaForProteinText2Text.get_input_embeddings": ["GenerationMixin", "ModelForProteinText2Text", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "model", "return", "self"], "evolla/modeling_evolla.py:EvollaForProteinText2Text.set_input_embeddings": ["GenerationMixin", "ModelForProteinText2Text", "ModelPreTrainedModel", "class", "def", "model", "return", "self", "set_input_embeddings", "value"], "evolla/modeling_evolla.py:EvollaForProteinText2Text.forward": ["BaseModelOutputWithPast", "CausalLMOutputWithPast", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForProteinText2Text", "ModelPreTrainedModel", "None", "Tensor", "attention_mask", "attentions", "bool", "class", "def", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "lm_outputs", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "outputs", "past_key_values", "protein_attention_mask", "protein_input_ids", "r", "return", "self", "slice", "slice_indices", "torch", "use_cache", "vocab_size"], "exaone4/modeling_exaone4.py:Exaone4RMSNorm.__init__": ["ModelRMSNorm", "Module", "Parameter", "__init__", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "exaone4/modeling_exaone4.py:Exaone4RMSNorm.forward": ["ModelRMSNorm", "Module", "True", "class", "def", "dtype", "float32", "forward", "hidden_states", "input_dtype", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "exaone4/modeling_exaone4.py:Exaone4RMSNorm.extra_repr": ["ModelRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "variance_epsilon", "weight"], "exaone4/modeling_exaone4.py:Exaone4RotaryEmbedding.__init__": ["Callable", "False", "ModelConfig", "ModelRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "exaone4/modeling_exaone4.py:Exaone4RotaryEmbedding.compute_default_rope_parameters": ["ModelConfig", "ModelRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "exaone4/modeling_exaone4.py:Exaone4RotaryEmbedding.forward": ["False", "ModelRotaryEmbedding", "Module", "None", "and", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "exaone4/modeling_exaone4.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "exaone4/modeling_exaone4.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cos", "def", "k", "k_embed", "q", "q_embed", "return", "rotate_half", "sin", "unsqueeze", "unsqueeze_dim"], "exaone4/modeling_exaone4.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "exaone4/modeling_exaone4.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value", "value_states"], "exaone4/modeling_exaone4.py:Exaone4Attention.__init__": ["False", "Linear", "ModelAttention", "ModelConfig", "ModelRMSNorm", "Module", "None", "True", "__init__", "attention_dropout", "bias", "class", "config", "def", "else", "eps", "getattr", "hasattr", "head_dim", "hidden_size", "if", "int", "is_causal", "is_sliding", "k_norm", "k_proj", "layer_idx", "layer_type", "layer_types", "nn", "num_attention_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "q_norm", "q_proj", "rms_norm_eps", "scaling", "self", "sliding_attention", "sliding_window", "sliding_window_pattern", "super", "v_proj"], "exaone4/modeling_exaone4.py:Exaone4Attention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "LongTensor", "ModelAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_attn_implementation", "apply_rotary_pos_emb", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "cache_kwargs", "cache_position", "class", "config", "contiguous", "cos", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "head_dim", "hidden_shape", "hidden_states", "if", "input_shape", "is", "is_sliding", "k_norm", "k_proj", "key_states", "kwargs", "layer_idx", "nn", "not", "o_proj", "or", "past_key_values", "position_embeddings", "q_norm", "q_proj", "query_states", "reshape", "return", "scaling", "self", "shape", "sin", "sliding_window", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "view"], "exaone4/modeling_exaone4.py:Exaone4MLP.__init__": ["ACT2FN", "False", "Linear", "ModelMLP", "Module", "__init__", "act_fn", "bias", "class", "config", "def", "down_proj", "gate_proj", "hidden_act", "hidden_size", "intermediate_size", "nn", "self", "super", "up_proj"], "exaone4/modeling_exaone4.py:Exaone4MLP.forward": ["ModelMLP", "Module", "act_fn", "class", "def", "down_proj", "forward", "gate_proj", "nn", "return", "self", "up_proj", "x"], "exaone4/modeling_exaone4.py:Exaone4DecoderLayer.__init__": ["GradientCheckpointingLayer", "ModelAttention", "ModelConfig", "ModelDecoderLayer", "ModelMLP", "ModelRMSNorm", "__init__", "class", "config", "def", "eps", "hidden_size", "int", "layer_idx", "mlp", "post_attention_layernorm", "post_feedforward_layernorm", "rms_norm_eps", "self", "self_attn", "super"], "exaone4/modeling_exaone4.py:Exaone4DecoderLayer.forward": ["Cache", "False", "GradientCheckpointingLayer", "LongTensor", "ModelDecoderLayer", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "attention_mask", "bool", "cache_position", "class", "def", "forward", "hidden_states", "kwargs", "mlp", "past_key_values", "position_embeddings", "position_ids", "post_attention_layernorm", "post_feedforward_layernorm", "residual", "return", "self", "self_attn", "torch", "tuple", "use_cache"], "exaone4/modeling_exaone4.py:Exaone4Model.__init__": ["Embedding", "False", "ModelConfig", "ModelDecoderLayer", "ModelModel", "ModelPreTrainedModel", "ModelRMSNorm", "ModelRotaryEmbedding", "ModuleList", "__init__", "class", "config", "def", "embed_tokens", "eps", "for", "gradient_checkpointing", "hidden_size", "in", "layer_idx", "layers", "nn", "norm", "num_hidden_layers", "pad_token_id", "padding_idx", "post_init", "range", "rms_norm_eps", "rotary_emb", "self", "super", "vocab_size"], "exaone4/modeling_exaone4.py:Exaone4Model.forward": ["BaseModelOutputWithPast", "Cache", "DynamicCache", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "and", "arange", "attention_mask", "bool", "cache_position", "causal_mask_mapping", "class", "config", "create_causal_mask", "create_sliding_window_causal_mask", "decoder_layer", "def", "device", "dict", "else", "embed_tokens", "enumerate", "exactly", "for", "forward", "full_attention", "get_seq_length", "hidden_states", "i", "if", "in", "input_embeds", "input_ids", "inputs_embeds", "is", "isinstance", "kwargs", "last_hidden_state", "layer_type", "layer_types", "layers", "mask_kwargs", "must", "norm", "not", "of", "one", "or", "past_key_values", "past_seen_tokens", "position_embeddings", "position_ids", "raise", "return", "rotary_emb", "self", "shape", "sliding_attention", "specify", "torch", "tuple", "unsqueeze", "use_cache"], "exaone4/modeling_exaone4.py:Exaone4ForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "ModelForCausalLM", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "vocab_size"], "exaone4/modeling_exaone4.py:Exaone4ForCausalLM.forward": ["BaseModelOutputWithPast", "Cache", "CausalLMOutputWithPast", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForCausalLM", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "outputs", "past_key_values", "position_ids", "r", "return", "self", "slice", "slice_indices", "torch", "use_cache", "vocab_size"], "falcon/modeling_falcon.py:FalconLinear.forward": ["Linear", "ModelLinear", "None", "T", "Tensor", "bias", "class", "def", "forward", "hidden_states", "if", "input", "is", "nn", "return", "self", "torch", "weight"], "falcon/modeling_falcon.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "falcon/modeling_falcon.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cos", "def", "k", "k_embed", "q", "q_embed", "return", "rotate_half", "sin", "unsqueeze", "unsqueeze_dim"], "falcon/modeling_falcon.py:FalconRotaryEmbedding.__init__": ["Callable", "False", "ModelConfig", "ModelRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "falcon/modeling_falcon.py:FalconRotaryEmbedding.compute_default_rope_parameters": ["ModelConfig", "ModelRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "falcon/modeling_falcon.py:FalconRotaryEmbedding.forward": ["False", "ModelRotaryEmbedding", "Module", "None", "and", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "falcon/modeling_falcon.py:build_alibi_tensor": ["Model_alibi_tensor", "None", "Tensor", "alibi", "arange", "arange_tensor", "attention_mask", "base", "batch_size", "bfloat16", "cat", "closest_power_of_2", "cumsum", "def", "device", "dim", "dtype", "extra_base", "extra_powers", "float32", "floor", "if", "int", "int32", "log2", "math", "min", "num_heads", "num_remaining_heads", "pow", "powers", "reshape", "return", "seq_length", "shape", "slopes", "tensor", "to", "torch"], "falcon/modeling_falcon.py:dropout_add": ["F", "Model", "Model_add", "Tensor", "bool", "def", "float", "out", "p", "prob", "residual", "return", "torch", "training", "x"], "falcon/modeling_falcon.py:FalconAttention.__init__": ["Dropout", "Instantiating", "ModelAttention", "ModelConfig", "ModelLinear", "Module", "None", "Please", "True", "ValueError", "__class__", "__init__", "__name__", "a", "and", "attention_dropout", "be", "beta", "bias", "by", "caching", "call", "class", "config", "creating", "def", "dense", "divisible", "during", "elif", "else", "errors", "f", "forward", "got", "head_dim", "hidden_dropout", "hidden_size", "if", "inv_norm_factor", "is", "is_causal", "layer_idx", "lead", "logger", "make", "math", "max_position_embeddings", "multi_query", "must", "new_decoder_architecture", "nn", "not", "num_attention_heads", "num_heads", "num_kv_heads", "or", "passing", "provide", "qkv_out_dim", "query_key_value", "raise", "recommended", "self", "split_size", "sqrt", "super", "sure", "the", "this", "to", "used", "warning_once", "when", "will", "without"], "falcon/modeling_falcon.py:FalconAttention._split_heads": ["ModelAttention", "Module", "Tensor", "_", "_split_heads", "batch", "batch_size", "broadcast_to", "class", "def", "elif", "else", "flatten", "for", "fused_qkv", "head_dim", "if", "in", "key", "multi_query", "new_decoder_architecture", "nn", "not", "num_heads", "num_kv_heads", "qkv", "query", "return", "self", "seq_len", "seq_length", "shape", "three_times_hidden_size", "torch", "tuple", "value", "view", "x"], "falcon/modeling_falcon.py:FalconAttention._merge_heads": ["ModelAttention", "Module", "Tensor", "_", "_merge_heads", "batch_size", "batch_size_and_num_heads", "class", "def", "head_dim", "nn", "num_heads", "permute", "reshape", "return", "self", "seq_length", "shape", "torch", "view", "x"], "falcon/modeling_falcon.py:FalconAttention.forward": ["Cache", "F", "False", "LongTensor", "ModelAttention", "Module", "None", "Tensor", "_", "_attn_implementation", "_merge_heads", "_split_heads", "alibi", "and", "apply_rotary_pos_emb", "attention_dropout", "attention_logits", "attention_mask", "attention_probs", "attention_probs_reshaped", "attention_scores", "attn_mask", "attn_output", "batch_size", "bfloat16", "bool", "cache_kwargs", "cache_position", "class", "config", "contiguous", "cos", "cuda", "def", "dense", "device", "dim", "dropout_p", "dtype", "else", "flatten", "float16", "float32", "forward", "functional", "fused_qkv", "head_dim", "hidden_states", "if", "input_dtype", "inv_norm_factor", "is", "is_causal", "key_layer", "kv_length", "layer_idx", "layer_past", "math", "matmul_result", "new_decoder_architecture", "nn", "not", "num_heads", "num_kv_heads", "or", "output_attentions", "p", "permute", "position_embeddings", "position_ids", "query_key_value", "query_layer", "query_length", "reshape", "return", "scaled_dot_product_attention", "sdpa", "self", "shape", "sin", "softmax", "sqrt", "to", "torch", "training", "transpose", "tuple", "type", "update", "use_cache", "value_layer", "view"], "falcon/modeling_falcon.py:FalconFlashAttention2.__init__": ["ModelAttention", "ModelFlashAttention2", "__init__", "_flash_attn_uses_top_left_mask", "args", "class", "def", "flash_attn_supports_top_left_mask", "kwargs", "self", "super"], "falcon/modeling_falcon.py:FalconFlashAttention2.forward": ["Cache", "False", "LongTensor", "ModelAttention", "ModelFlashAttention2", "None", "Tensor", "The", "True", "ValueError", "We", "_", "_flash_attention_forward", "_flash_attn_uses_top_left_mask", "_is_quantized", "_split_heads", "alibi", "apply_rotary_pos_emb", "attention_dropout", "attention_mask", "attn_dropout", "attn_output", "attn_weights", "back", "batch_size", "be", "bool", "cache_kwargs", "cache_position", "cast", "casted", "class", "config", "cos", "cpu", "def", "dense", "device", "device_type", "dropout", "dtype", "elif", "else", "embedding", "f", "fact", "float32", "forward", "fused_qkv", "get_autocast_dtype", "get_autocast_gpu_dtype", "hasattr", "have", "head_dim", "hidden", "hidden_states", "if", "in", "input", "input_dtype", "is", "is_autocast_enabled", "is_causal", "key_layer", "layer", "layer_idx", "layer_past", "layers", "logger", "might", "mps", "new_decoder_architecture", "norm", "not", "num_heads", "num_kv_heads", "or", "output_attentions", "position_embeddings", "position_ids", "query_key_value", "query_layer", "query_length", "raise", "related", "reshape", "return", "seems", "self", "shape", "silently", "sin", "states", "supported", "target_dtype", "the", "this", "to", "torch", "training", "transpose", "tuple", "type", "upcasted", "update", "use_cache", "use_flash_attn", "use_top_left_mask", "value_layer", "warning_once", "weight", "when", "will", "you"], "falcon/modeling_falcon.py:FalconMLP.__init__": ["ModelConfig", "ModelLinear", "ModelMLP", "Module", "__init__", "act", "activation", "bias", "class", "config", "def", "dense_4h_to_h", "dense_h_to_4h", "ffn_hidden_size", "get_activation", "hidden_dropout", "hidden_size", "nn", "self", "super"], "falcon/modeling_falcon.py:FalconMLP.forward": ["ModelMLP", "Module", "Tensor", "act", "class", "def", "dense_4h_to_h", "dense_h_to_4h", "forward", "nn", "return", "self", "torch", "x"], "falcon/modeling_falcon.py:FalconDecoderLayer.__init__": ["GradientCheckpointingLayer", "LayerNorm", "ModelConfig", "ModelDecoderLayer", "ModelMLP", "Model_ATTENTION_CLASSES", "None", "__init__", "_attn_implementation", "and", "class", "config", "def", "else", "eps", "hidden_dropout", "hidden_size", "if", "input_layernorm", "is", "layer_idx", "layer_norm_epsilon", "ln_attn", "ln_mlp", "mlp", "new_decoder_architecture", "not", "num_attention_heads", "num_heads", "num_ln_in_parallel_attn", "parallel_attn", "post_attention_layernorm", "self", "self_attention", "super"], "falcon/modeling_falcon.py:FalconDecoderLayer.forward": ["Cache", "False", "GradientCheckpointingLayer", "LongTensor", "ModelDecoderLayer", "None", "Tensor", "alibi", "and", "attention_dropout", "attention_layernorm_out", "attention_mask", "attention_output", "attn_weights", "bool", "cache_position", "class", "config", "def", "dropout_add", "else", "forward", "hidden_dropout", "hidden_states", "if", "input_layernorm", "kwargs", "layer_past", "ln_attn", "ln_mlp", "mlp", "mlp_layernorm_out", "mlp_output", "new_decoder_architecture", "not", "num_ln_in_parallel_attn", "or", "output", "output_attentions", "parallel_attn", "position_embeddings", "position_ids", "post_attention_layernorm", "residual", "return", "self", "self_attention", "torch", "training", "tuple", "use_cache"], "falcon/modeling_falcon.py:FalconPreTrainedModel._init_weights": ["ModelLinear", "ModelPreTrainedModel", "Module", "None", "PreTrainedModel", "_init_weights", "bias", "class", "config", "def", "if", "init", "initializer_range", "is", "isinstance", "mean", "module", "nn", "normal_", "not", "self", "std", "super", "weight", "zeros_"], "falcon/modeling_falcon.py:FalconPreTrainedModel._check_and_enable_sdpa": ["False", "ModelPreTrainedModel", "PreTrainedModel", "_attn_implementation", "_check_and_enable_sdpa", "_is_bettertransformer", "bool", "class", "cls", "config", "def", "getattr", "hard_check_only", "if", "not", "return", "sdpa", "use_bettertransformer"], "falcon/modeling_falcon.py:FalconModel.__init__": ["Embedding", "False", "LayerNorm", "ModelConfig", "ModelDecoderLayer", "ModelModel", "ModelPreTrainedModel", "ModelRotaryEmbedding", "ModuleList", "__init__", "alibi", "class", "config", "def", "embed_dim", "eps", "for", "gradient_checkpointing", "h", "hidden_size", "i", "in", "layer_idx", "layer_norm_epsilon", "ln_f", "nn", "num_attention_heads", "num_heads", "num_hidden_layers", "post_init", "range", "rotary_emb", "self", "super", "use_alibi", "vocab_size", "word_embeddings"], "falcon/modeling_falcon.py:FalconModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "return", "self", "word_embeddings"], "falcon/modeling_falcon.py:FalconModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "Tensor", "class", "def", "new_embeddings", "self", "set_input_embeddings", "torch", "word_embeddings"], "falcon/modeling_falcon.py:FalconModel.forward": ["BaseModelOutputWithPastAndCrossAttentions", "Cache", "DynamicCache", "False", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Setting", "Tensor", "True", "ValueError", "You", "_", "_update_causal_mask", "alibi", "all_hidden_states", "all_self_attentions", "and", "arange", "attention_mask", "attentions", "batch_size", "block", "bool", "build_alibi_tensor", "cache_position", "causal_mask", "checkpointing", "class", "config", "def", "device", "dtype", "else", "enumerate", "exactly", "for", "forward", "get_seq_length", "gradient", "gradient_checkpointing", "h", "hidden_states", "i", "if", "in", "incompatible", "input_ids", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layer_past", "ln_f", "logger", "long", "mask", "must", "not", "num_heads", "of", "one", "ones", "or", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "past_key_values_length", "position_embeddings", "position_ids", "r", "raise", "return", "return_dict", "rotary_emb", "self", "seq_length", "shape", "specify", "torch", "training", "tuple", "unsqueeze", "use_alibi", "use_cache", "use_return_dict", "v", "warning_once", "with", "word_embeddings"], "falcon/modeling_falcon.py:FalconModel._update_causal_mask": ["AttentionMaskConverter", "Cache", "ModelModel", "ModelPreTrainedModel", "None", "StaticCache", "Tensor", "_", "_attn_implementation", "_ignore_causal_mask_sdpa", "_prepare_4d_causal_attention_mask_with_cache_position", "_unmask_unattended", "_update_causal_mask", "alibi", "and", "attention_mask", "batch_size", "bool", "cache_position", "causal_mask", "class", "config", "cuda", "def", "device", "dtype", "else", "finfo", "flash_attention_2", "get_max_cache_shape", "get_seq_length", "hidden_size", "if", "in", "input_tensor", "inputs_embeds", "is", "is_training", "isinstance", "masked_fill", "math", "min", "min_dtype", "not", "npu", "num_heads", "output_attentions", "past_key_values", "past_key_values_length", "past_seen_tokens", "reshape", "return", "sdpa", "self", "sequence_length", "shape", "sqrt", "target_length", "torch", "training", "type", "using_static_cache", "xpu"], "falcon/modeling_falcon.py:FalconModel._prepare_4d_causal_attention_mask_with_cache_position": ["ModelModel", "ModelPreTrainedModel", "None", "Tensor", "_prepare_4d_causal_attention_mask_with_cache_position", "and", "arange", "attention_mask", "batch_size", "cache_position", "causal_mask", "class", "clone", "def", "device", "diagonal", "dim", "dtype", "else", "expand", "fill_value", "finfo", "full", "if", "int", "is", "kwargs", "mask_length", "masked_fill", "min", "min_dtype", "not", "padding_mask", "reshape", "return", "sequence_length", "shape", "target_length", "to", "torch", "triu"], "falcon/modeling_falcon.py:FalconForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "ModelConfig", "ModelForCausalLM", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "nn", "post_init", "self", "super", "transformer", "vocab_size"], "falcon/modeling_falcon.py:FalconForCausalLM.set_output_embeddings": ["GenerationMixin", "ModelForCausalLM", "ModelPreTrainedModel", "Tensor", "class", "def", "lm_head", "new_embeddings", "self", "set_output_embeddings", "torch"], "falcon/modeling_falcon.py:FalconForCausalLM.forward": ["Cache", "CausalLMOutputWithCrossAttentions", "GenerationMixin", "LongTensor", "ModelForCausalLM", "ModelPreTrainedModel", "None", "Tensor", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "lm_head", "lm_logits", "logits", "logits_to_keep", "loss", "loss_function", "not", "output", "output_attentions", "output_hidden_states", "past_key_values", "position_ids", "r", "return", "return_dict", "self", "slice", "slice_indices", "torch", "transformer", "transformer_outputs", "tuple", "use_cache", "use_return_dict", "vocab_size"], "falcon/modeling_falcon.py:FalconForSequenceClassification.__init__": ["False", "Linear", "ModelConfig", "ModelForSequenceClassification", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "nn", "num_labels", "post_init", "score", "self", "super", "transformer"], "falcon/modeling_falcon.py:FalconForSequenceClassification.forward": ["BCEWithLogitsLoss", "Cache", "Cannot", "CrossEntropyLoss", "LongTensor", "MSELoss", "ModelForSequenceClassification", "ModelPreTrainedModel", "None", "Results", "SequenceClassifierOutputWithPast", "Tensor", "ValueError", "__class__", "__name__", "and", "arange", "argmax", "attention_mask", "attentions", "batch", "batch_size", "be", "bool", "class", "config", "conjunction", "def", "defined", "detect", "device", "dtype", "elif", "else", "f", "forward", "handle", "hidden_states", "if", "in", "input_ids", "inputs_embeds", "int", "int32", "is", "kwargs", "labels", "last_non_pad_token", "logger", "logits", "long", "loss", "loss_fct", "may", "multi_label_classification", "no", "non_pad_mask", "not", "num_labels", "or", "output", "output_attentions", "output_hidden_states", "pad_token_id", "padding", "past_key_values", "pooled_logits", "problem_type", "r", "raise", "regression", "return", "return_dict", "score", "self", "shape", "single_label_classification", "sizes", "squeeze", "to", "token", "token_indices", "tokens", "torch", "transformer", "transformer_outputs", "tuple", "unexpected", "use_cache", "use_return_dict", "using", "warning_once", "will", "with"], "falcon/modeling_falcon.py:FalconForTokenClassification.__init__": ["Dropout", "Linear", "ModelConfig", "ModelForTokenClassification", "ModelModel", "ModelPreTrainedModel", "None", "__init__", "class", "classifier", "classifier_dropout", "config", "def", "dropout", "elif", "else", "getattr", "hidden_dropout", "hidden_size", "if", "is", "nn", "not", "num_labels", "post_init", "self", "super", "transformer"], "falcon/modeling_falcon.py:FalconForTokenClassification.forward": ["Cache", "CrossEntropyLoss", "LongTensor", "ModelForTokenClassification", "ModelPreTrainedModel", "None", "Tensor", "TokenClassifierOutput", "attention_mask", "attentions", "batch_size", "bool", "class", "classifier", "config", "def", "dropout", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "not", "num_labels", "output", "output_attentions", "output_hidden_states", "past_key_values", "r", "return", "return_dict", "self", "seq_length", "shape", "torch", "transformer", "transformer_outputs", "tuple", "use_cache", "use_return_dict", "view"], "falcon/modeling_falcon.py:FalconForQuestionAnswering.__init__": ["Linear", "ModelForQuestionAnswering", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "config", "def", "hidden_size", "nn", "post_init", "qa_outputs", "self", "super", "transformer"], "falcon/modeling_falcon.py:FalconForQuestionAnswering.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "ModelForQuestionAnswering", "ModelPreTrainedModel", "None", "QuestionAnsweringModelOutput", "and", "attention_mask", "attentions", "bool", "clamp", "class", "config", "contiguous", "def", "dim", "else", "end_logits", "end_loss", "end_positions", "forward", "hidden_states", "if", "ignore_index", "ignored_index", "input_ids", "inputs_embeds", "is", "kwargs", "len", "logits", "loss", "loss_fct", "not", "output", "output_attentions", "output_hidden_states", "outputs", "qa_outputs", "r", "return", "return_dict", "self", "sequence_output", "size", "split", "squeeze", "start_logits", "start_loss", "start_positions", "torch", "total_loss", "transformer", "tuple", "use_return_dict"], "falcon_h1/modeling_falcon_h1.py:FalconHybridMambaAttentionDynamicCache.__init__": ["False", "ModelConfig", "ModelHybridMambaAttentionDynamicCache", "None", "Tensor", "__init__", "append", "batch_size", "class", "config", "conv_kernel_size", "conv_states", "def", "device", "devices", "dtype", "else", "float16", "for", "has_previous_state", "hidden_size", "i", "if", "in", "int", "intermediate_size", "is", "key_cache", "list", "mamba_d_conv", "mamba_d_head", "mamba_d_ssm", "mamba_d_state", "mamba_expand", "mamba_n_groups", "mamba_n_heads", "not", "num_hidden_layers", "range", "self", "seqlen_offset", "ssm_states", "str", "torch", "transformer_layers", "value_cache", "zeros"], "falcon_h1/modeling_falcon_h1.py:FalconHybridMambaAttentionDynamicCache.__len__": ["ModelHybridMambaAttentionDynamicCache", "__len__", "class", "def", "key_cache", "len", "return", "self"], "falcon_h1/modeling_falcon_h1.py:FalconHybridMambaAttentionDynamicCache.__getitem__": ["ModelHybridMambaAttentionDynamicCache", "__getitem__", "class", "def", "key_cache", "layer_idx", "return", "self", "value_cache"], "falcon_h1/modeling_falcon_h1.py:FalconHybridMambaAttentionDynamicCache.update": ["Any", "ModelHybridMambaAttentionDynamicCache", "None", "Tensor", "_", "append", "cache_kwargs", "cat", "class", "def", "dict", "dim", "elif", "else", "for", "if", "in", "int", "key_cache", "key_states", "layer_idx", "len", "range", "return", "self", "str", "torch", "tuple", "update", "value_cache", "value_states"], "falcon_h1/modeling_falcon_h1.py:FalconHybridMambaAttentionDynamicCache.reorder_cache": ["LongTensor", "ModelHybridMambaAttentionDynamicCache", "beam_idx", "class", "conv_states", "def", "device", "for", "get_seq_length", "if", "in", "index_select", "key_cache", "layer_idx", "len", "range", "reorder_cache", "self", "ssm_states", "to", "torch", "value_cache"], "falcon_h1/modeling_falcon_h1.py:FalconHybridMambaAttentionDynamicCache.get_mask_sizes": ["ModelHybridMambaAttentionDynamicCache", "Tensor", "cache_position", "class", "def", "get_mask_sizes", "get_seq_length", "int", "kv_length", "kv_offset", "layer_idx", "query_length", "return", "self", "shape", "torch", "tuple"], "falcon_h1/modeling_falcon_h1.py:FalconHybridMambaAttentionDynamicCache.get_seq_length": ["ModelHybridMambaAttentionDynamicCache", "None", "class", "def", "else", "get_seq_length", "if", "in", "int", "key_cache", "layer_idx", "len", "not", "or", "return", "self", "shape", "transformer_layers"], "falcon_h1/modeling_falcon_h1.py:FalconHybridMambaAttentionDynamicCache.update_conv_state": ["LongTensor", "ModelHybridMambaAttentionDynamicCache", "Tensor", "cache_position", "clamp", "class", "conv_kernel_size", "conv_state", "conv_states", "def", "device", "dims", "else", "if", "int", "layer_idx", "len", "new_conv_state", "return", "roll", "self", "shifts", "to", "torch", "update_conv_state", "zero_"], "falcon_h1/modeling_falcon_h1.py:FalconHybridMambaAttentionDynamicCache.reset": ["ModelHybridMambaAttentionDynamicCache", "class", "conv_states", "def", "reset", "self", "ssm_states", "zero_"], "falcon_h1/modeling_falcon_h1.py:FalconH1RotaryEmbedding.__init__": ["Callable", "False", "ModelConfig", "ModelRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "falcon_h1/modeling_falcon_h1.py:FalconH1RotaryEmbedding.compute_default_rope_parameters": ["ModelConfig", "ModelRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "falcon_h1/modeling_falcon_h1.py:FalconH1RotaryEmbedding.forward": ["False", "ModelRotaryEmbedding", "Module", "None", "and", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "falcon_h1/modeling_falcon_h1.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "falcon_h1/modeling_falcon_h1.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cos", "def", "k", "k_embed", "q", "q_embed", "return", "rotate_half", "sin", "unsqueeze", "unsqueeze_dim"], "falcon_h1/modeling_falcon_h1.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "falcon_h1/modeling_falcon_h1.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value", "value_states"], "falcon_h1/modeling_falcon_h1.py:FalconH1Attention.__init__": ["Linear", "ModelAttention", "ModelConfig", "Module", "True", "__init__", "attention_bias", "attention_dropout", "bias", "class", "config", "def", "getattr", "head_dim", "hidden_size", "int", "is_causal", "k_proj", "key_multiplier", "layer_idx", "nn", "num_attention_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "q_proj", "scaling", "self", "super", "v_proj"], "falcon_h1/modeling_falcon_h1.py:FalconH1Attention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "FlashAttentionKwargs", "LongTensor", "ModelAttention", "Module", "None", "Tensor", "Unpack", "_attn_implementation", "apply_rotary_pos_emb", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "cache_kwargs", "cache_position", "class", "config", "contiguous", "cos", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "head_dim", "hidden_shape", "hidden_states", "if", "input_shape", "is", "k_proj", "key_multiplier", "key_states", "kwargs", "layer_idx", "nn", "not", "o_proj", "past_key_values", "position_embeddings", "q_proj", "query_states", "reshape", "return", "scaling", "self", "shape", "sin", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "view"], "falcon_h1/modeling_falcon_h1.py:FalconH1RMSNormGated.__init__": ["ModelRMSNormGated", "Module", "Parameter", "True", "__init__", "class", "def", "eps", "hidden_size", "n_groups", "nn", "norm_before_gate", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "falcon_h1/modeling_falcon_h1.py:FalconH1RMSNormGated.forward": ["F", "ModelRMSNormGated", "Module", "None", "True", "and", "batch_size", "class", "def", "dim", "dtype", "else", "float32", "forward", "gate", "hidden_states", "if", "input_dtype", "int", "is", "keepdim", "len", "mean", "n_groups", "nn", "norm_before_gate", "not", "pow", "return", "rsqrt", "self", "seq_len", "shape", "silu", "squeeze", "to", "torch", "variance", "variance_epsilon", "view", "weight"], "falcon_h1/modeling_falcon_h1.py:pad_tensor_by_size": ["Model", "Model_shape", "Model_size", "Model_tensor_by_size", "Tensor", "constant", "def", "else", "functional", "if", "input_tensor", "int", "len", "mode", "nn", "return", "shape", "torch", "value"], "falcon_h1/modeling_falcon_h1.py:reshape_into_chunks": ["Model", "Model_into_chunks", "chunk_size", "def", "else", "if", "input_tensor", "len", "pad_size", "pad_tensor_by_size", "return", "shape"], "falcon_h1/modeling_falcon_h1.py:segment_sum": ["Model_sum", "None", "bool", "chunk_size", "cumsum", "def", "device", "diagonal", "dim", "dtype", "expand", "inf", "input_tensor", "mask", "masked_fill", "ones", "return", "size", "tensor_segsum", "torch", "tril"], "falcon_h1/modeling_falcon_h1.py:apply_mask_to_padding_states": ["Model_mask_to_padding_states", "None", "and", "attention_mask", "def", "dtype", "hidden_states", "if", "is", "not", "return", "shape", "to"], "falcon_h1/modeling_falcon_h1.py:FalconH1Mixer.__init__": ["A", "ACT2FN", "AILab", "A_log", "Conv1d", "D", "Dao", "Falling", "GPU", "Linear", "Model", "ModelConfig", "ModelMixer", "ModelRMSNormGated", "Module", "None", "Parameter", "The", "To", "__init__", "a", "act", "activation", "arange", "available", "back", "be", "because", "bias", "causal", "causal_conv1d_fn", "causal_conv1d_update", "chunk_size", "class", "com", "config", "conv1d", "conv_dim", "conv_kernel_size", "def", "dt_bias", "else", "eps", "fast", "float", "follow", "for", "github", "groups", "groups_time_state_size", "head_dim", "hidden_act", "hidden_size", "https", "if", "implementation", "in_channels", "in_proj", "inf", "install", "int", "intermediate_size", "is", "is_fast_path_available", "kernel_size", "layer_idx", "layer_norm_epsilon", "log", "logger", "mamba", "mamba_chunk_size", "mamba_conv_bias", "mamba_d_conv", "mamba_d_head", "mamba_d_ssm", "mamba_d_state", "mamba_expand", "mamba_n_groups", "mamba_n_heads", "mamba_norm_before_gate", "mamba_proj_bias", "mamba_rms_norm", "model", "n_groups", "naive", "nn", "norm", "norm_before_gate", "not", "num_heads", "of", "on", "one", "ones", "out_channels", "out_proj", "padding", "path", "projection_size", "projectors_bias", "rms_norm_eps", "running", "selective_state_update", "self", "spaces", "ssm_in_multiplier", "ssm_multipliers", "ssm_state_size", "state", "super", "the", "time_step_limit", "time_step_max", "time_step_min", "to", "torch", "use_bias", "use_conv_bias", "used", "warning_once", "when", "will", "zxbcdt_multipliers"], "falcon_h1/modeling_falcon_h1.py:FalconH1Mixer.cuda_kernels_forward": ["A", "A_log", "B", "C", "D", "F", "False", "LongTensor", "ModelHybridMambaAttentionDynamicCache", "ModelMixer", "Module", "None", "Tensor", "True", "_", "act", "activation", "and", "apply_mask_to_padding_states", "attention_mask", "batch_size", "bias", "cache_params", "cache_position", "cat", "causal_conv1d_fn", "causal_conv1d_update", "chunk_size", "class", "conv1d", "conv_dim", "conv_kernel_size", "conv_states", "copy_", "cuda", "cuda_kernels_forward", "d_mlp", "d_to_remove", "def", "device", "dim", "dt", "dt_bias", "dt_limit", "dt_limit_kwargs", "dt_softplus", "dtype", "else", "exp", "expand", "float", "float32", "functional", "gate", "groups_time_state_size", "has_previous_state", "head_dim", "headdim", "hidden_states", "hidden_states_B_C", "hidden_states_reshaped", "if", "in", "in_proj", "inf", "intermediate_size", "is", "layer_idx", "mamba_chunk_scan_combined", "mamba_rms_norm", "mamba_split_conv1d_scan_combined", "mup_vector", "n_groups", "ngroups", "nn", "norm", "norm_before_gate", "not", "num_heads", "or", "out", "out_proj", "outproj_bias", "outproj_weight", "pad", "permute", "projected_states", "return", "return_final_states", "rmsnorm_eps", "rmsnorm_weight", "scan_output", "selective_state_update", "self", "seq_idx", "seq_len", "shape", "silu", "softplus", "split", "squeeze", "ssm_in_multiplier", "ssm_state", "ssm_state_size", "ssm_states", "swish", "time_step", "time_step_limit", "to", "torch", "training", "transpose", "update_conv_state", "use_precomputed_states", "variance_epsilon", "view", "weight", "with", "x", "x0", "z", "z0"], "falcon_h1/modeling_falcon_h1.py:FalconH1Mixer.torch_forward": ["A", "A_cumsum", "A_log", "B", "B_decay", "C", "C_reshaped", "C_times_states", "D", "D_residual", "G", "G_intermediate", "L", "LongTensor", "M", "M_intermediate", "ModelHybridMambaAttentionDynamicCache", "ModelMixer", "Module", "None", "Tensor", "Y_diag", "Y_off", "_", "act", "and", "apply_mask_to_padding_states", "attention_mask", "batch_size", "bias", "bmm", "cache_device", "cache_params", "cache_position", "cat", "chunk_size", "clamp", "class", "contextualized_states", "contiguous", "conv1d", "conv_dim", "conv_kernel_size", "conv_states", "copy_", "cumsum", "dA", "dB", "dBx", "decay_chunk", "decay_states", "def", "device", "dim", "dims", "dt", "dt_bias", "dtype", "else", "exp", "expand", "float", "float32", "for", "functional", "gate", "has_previous_state", "head_dim", "hidden_states", "hidden_states_B_C", "hidden_states_B_C_transposed", "if", "in", "in_proj", "input_states", "intermediate_size", "is", "layer_idx", "mamba_rms_norm", "mup_vector", "n_groups", "new_states", "nn", "norm", "not", "num_heads", "out_proj", "output_size", "pad", "pad_size", "pad_tensor_by_size", "permute", "previous_states", "projected_states", "repeat_interleave", "reshape", "reshape_into_chunks", "return", "roll", "scan_output", "segment_sum", "self", "seq_len", "shape", "shifts", "silu", "softplus", "split", "squeeze", "ssm_in_multiplier", "ssm_state", "ssm_state_size", "ssm_states", "ssm_states_reshaped", "state_decay_out", "state_decay_out_permuted", "states", "sum", "t", "time_step_limit", "to", "torch", "torch_forward", "transpose", "use_conv_bias", "use_precomputed_states", "view", "weight", "y", "zeros_like"], "falcon_h1/modeling_falcon_h1.py:FalconH1Mixer.forward": ["LongTensor", "ModelHybridMambaAttentionDynamicCache", "ModelMixer", "Module", "None", "Tensor", "and", "attention_mask", "cache_params", "cache_position", "class", "cuda", "cuda_kernels_forward", "def", "device", "dtype", "forward", "hidden_states", "if", "in", "in_proj", "is", "is_fast_path_available", "nn", "not", "return", "self", "shape", "to", "torch", "torch_forward", "type", "weight"], "falcon_h1/modeling_falcon_h1.py:FalconH1MLP.__init__": ["ACT2FN", "Linear", "ModelConfig", "ModelMLP", "Module", "__init__", "act_fn", "bias", "class", "config", "def", "down_multiplier", "down_proj", "gate_multiplier", "gate_proj", "hidden_act", "hidden_size", "intermediate_size", "mlp_bias", "mlp_multipliers", "nn", "self", "super", "up_proj"], "falcon_h1/modeling_falcon_h1.py:FalconH1MLP.forward": ["ModelMLP", "Module", "act_fn", "class", "def", "down_multiplier", "down_proj", "forward", "gate_multiplier", "gate_proj", "nn", "return", "self", "up_proj", "x", "y"], "falcon_h1/modeling_falcon_h1.py:FalconH1RMSNorm.__init__": ["ModelRMSNorm", "Module", "Parameter", "__init__", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "falcon_h1/modeling_falcon_h1.py:FalconH1RMSNorm.forward": ["ModelRMSNorm", "Module", "True", "class", "def", "dtype", "float32", "forward", "hidden_states", "input_dtype", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "falcon_h1/modeling_falcon_h1.py:FalconH1RMSNorm.extra_repr": ["ModelRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "variance_epsilon", "weight"], "falcon_h1/modeling_falcon_h1.py:FalconH1DecoderLayer.__init__": ["GradientCheckpointingLayer", "ModelAttention", "ModelConfig", "ModelDecoderLayer", "ModelMLP", "ModelMixer", "ModelRMSNorm", "__init__", "attention_in_multiplier", "attention_out_multiplier", "attn_out_multiplier", "channels_attn", "class", "config", "def", "eps", "feed_forward", "head_dim", "hidden_size", "input_layernorm", "int", "layer_idx", "mamba", "num_attention_heads", "num_key_value_heads", "pre_ff_layernorm", "rms_norm_eps", "self", "self_attn", "ssm_out_multiplier", "super"], "falcon_h1/modeling_falcon_h1.py:FalconH1DecoderLayer.forward": ["False", "FloatTensor", "GradientCheckpointingLayer", "LongTensor", "ModelDecoderLayer", "ModelHybridMambaAttentionDynamicCache", "None", "Tensor", "attention_hidden_states", "attention_in_multiplier", "attention_mask", "attn_out_multiplier", "bool", "cache_params", "cache_position", "class", "def", "feed_forward", "forward", "hidden_states", "if", "input_layernorm", "kwargs", "mamba", "mamba_attention_mask", "mamba_hidden_states", "output_attentions", "outputs", "past_key_values", "position_embeddings", "position_ids", "pre_ff_layernorm", "residual", "return", "self", "self_attn", "self_attn_weights", "ssm_out_multiplier", "torch", "tuple", "use_cache"], "falcon_h1/modeling_falcon_h1.py:compute_mup_vector": ["Model_mup_vector", "None", "config", "def", "else", "groups_time_state_size", "hidden_size", "if", "int", "intermediate_size", "is", "mamba_d_ssm", "mamba_d_state", "mamba_expand", "mamba_n_groups", "mamba_n_heads", "mup_vector", "not", "num_heads", "ones", "return", "ssm_multipliers", "torch", "vector_shape", "zxbcdt_multipliers"], "falcon_h1/modeling_falcon_h1.py:FalconH1PreTrainedModel._init_weights": ["A_log", "D", "ModelMixer", "ModelModel", "ModelPreTrainedModel", "PreTrainedModel", "_init_weights", "arange", "class", "compute_mup_vector", "config", "copy_", "def", "dt_bias", "elif", "for", "if", "in", "init", "isinstance", "layer", "layers", "log", "mamba", "module", "mup_vector", "num_heads", "ones_", "self", "super", "torch"], "falcon_h1/modeling_falcon_h1.py:FalconH1Model.__init__": ["Embedding", "False", "ModelConfig", "ModelDecoderLayer", "ModelModel", "ModelPreTrainedModel", "ModelRMSNorm", "ModelRotaryEmbedding", "ModuleList", "__init__", "_attn_implementation", "append", "class", "clone", "compute_mup_vector", "config", "decoder_layers", "def", "embed_tokens", "embedding_multiplier", "eps", "final_layernorm", "for", "gradient_checkpointing", "hidden_size", "i", "in", "layer", "layer_idx", "layers", "lm_head_multiplier", "mamba", "mup_vector", "nn", "num_hidden_layers", "pad_token_id", "padding_idx", "persistent", "post_init", "range", "register_buffer", "rms_norm_eps", "rotary_emb", "self", "super", "vocab_size"], "falcon_h1/modeling_falcon_h1.py:FalconH1Model.forward": ["BaseModelOutputWithPast", "False", "FloatTensor", "LongTensor", "Model", "ModelHybridMambaAttentionDynamicCache", "ModelModel", "ModelPreTrainedModel", "None", "Setting", "Tensor", "True", "ValueError", "You", "_update_causal_mask", "_update_mamba_mask", "a", "all_hidden_states", "all_self_attns", "an", "and", "arange", "attention_mask", "attentions", "be", "bool", "cache", "cache_position", "causal_mask", "checkpointing", "class", "config", "decoder_layer", "def", "device", "else", "embed_tokens", "embedding_multiplier", "exactly", "final_layernorm", "for", "forward", "gradient", "gradient_checkpointing", "has_previous_state", "hidden_states", "if", "in", "incompatible", "initialized", "input_ids", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layer_outputs", "layers", "logger", "mamba_attention_mask", "mamba_mask", "must", "next_cache", "no", "not", "of", "one", "or", "output_attentions", "output_hidden_states", "past_key_values", "position_embeddings", "position_ids", "provided", "raise", "requires", "return", "returned", "rotary_emb", "self", "shape", "so", "specify", "to", "torch", "training", "tuple", "unsqueeze", "use_cache", "warning_once", "was", "will", "with"], "falcon_h1/modeling_falcon_h1.py:FalconH1Model._update_mamba_mask": ["ModelModel", "ModelPreTrainedModel", "None", "_update_mamba_mask", "all", "and", "attention_mask", "cache_position", "class", "def", "if", "is", "mamba_mask", "not", "or", "return", "self", "torch"], "falcon_h1/modeling_falcon_h1.py:FalconH1Model._update_causal_mask": ["AttentionMaskConverter", "ModelHybridMambaAttentionDynamicCache", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "_attn_implementation", "_ignore_causal_mask_sdpa", "_prepare_4d_causal_attention_mask_with_cache_position", "_unmask_unattended", "_update_causal_mask", "and", "attention_mask", "batch_size", "bool", "cache_position", "causal_mask", "class", "config", "cuda", "def", "device", "dtype", "else", "finfo", "flash_attention_2", "get_seq_length", "if", "in", "input_tensor", "inputs_embeds", "is", "is_training", "isinstance", "min", "min_dtype", "not", "npu", "output_attentions", "past_key_values", "past_key_values_length", "past_seen_tokens", "return", "sdpa", "self", "sequence_length", "shape", "target_length", "torch", "training", "type", "xpu"], "falcon_h1/modeling_falcon_h1.py:FalconH1Model._prepare_4d_causal_attention_mask_with_cache_position": ["ModelModel", "ModelPreTrainedModel", "None", "Tensor", "_prepare_4d_causal_attention_mask_with_cache_position", "and", "arange", "attention_mask", "batch_size", "cache_position", "causal_mask", "class", "clone", "def", "device", "diagonal", "dim", "dtype", "else", "expand", "fill_value", "finfo", "full", "if", "int", "is", "kwargs", "mask_length", "masked_fill", "min", "min_dtype", "not", "padding_attention_mask", "padding_mask", "reshape", "return", "sequence_length", "shape", "target_length", "to", "torch", "triu"], "falcon_h1/modeling_falcon_h1.py:FalconH1ForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "ModelForCausalLM", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "vocab_size"], "falcon_h1/modeling_falcon_h1.py:FalconH1ForCausalLM.forward": ["CausalLMOutputWithPast", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForCausalLM", "ModelHybridMambaAttentionDynamicCache", "ModelPreTrainedModel", "None", "Tensor", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "lm_head", "lm_head_multiplier", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "position_ids", "r", "return", "self", "slice", "slice_indices", "torch", "tuple", "use_cache", "vocab_size"], "falcon_h1/modeling_falcon_h1.py:FalconH1ForCausalLM.prepare_inputs_for_generation": ["False", "GenerationMixin", "ModelForCausalLM", "ModelHybridMambaAttentionDynamicCache", "ModelPreTrainedModel", "None", "True", "and", "attention_mask", "cache_position", "class", "config", "contiguous", "conv1d", "cumsum", "def", "device", "devices", "dtype", "elif", "else", "empty_past_kv", "for", "i", "if", "in", "input_ids", "inputs_embeds", "is", "is_first_iteration", "is_torchdynamo_compiling", "items", "key", "kwargs", "layers", "logits_to_keep", "long", "mamba", "masked_fill_", "model", "model_inputs", "not", "num_hidden_layers", "num_logits_to_keep", "or", "past_key_values", "position_ids", "prepare_inputs_for_generation", "range", "return", "self", "shape", "update", "use_cache", "value", "weight"], "falcon_mamba/modeling_falcon_mamba.py:FalconMambaCache.__init__": ["ModelCache", "None", "PreTrainedConfig", "Tensor", "_", "__init__", "_dtype", "_dynamo", "append", "class", "config", "conv_kernel", "conv_kernel_size", "conv_state", "conv_states", "def", "device", "dtype", "else", "float16", "for", "if", "in", "int", "intermediate_size", "is", "list", "mark_static_address", "max_batch_size", "not", "num_hidden_layers", "range", "self", "ssm_state", "ssm_state_size", "ssm_states", "state_size", "str", "torch", "zeros"], "falcon_mamba/modeling_falcon_mamba.py:FalconMambaCache.update_conv_state": ["LongTensor", "ModelCache", "Tensor", "cache_position", "clamp", "class", "conv_kernel_size", "conv_state", "conv_states", "def", "device", "dims", "dtype", "if", "int", "layer_idx", "new_conv_state", "return", "roll", "self", "shifts", "to", "torch", "update_conv_state", "zero_"], "falcon_mamba/modeling_falcon_mamba.py:FalconMambaCache.update_ssm_state": ["ModelCache", "Tensor", "class", "def", "device", "int", "layer_idx", "new_ssm_state", "return", "self", "ssm_states", "to", "torch", "update_ssm_state", "zero_"], "falcon_mamba/modeling_falcon_mamba.py:FalconMambaCache.reset": ["ModelCache", "class", "conv_states", "def", "for", "in", "layer_idx", "len", "range", "reset", "self", "ssm_states", "zero_"], "falcon_mamba/modeling_falcon_mamba.py:rms_forward": ["Model_forward", "True", "def", "dtype", "float32", "hidden_states", "input_dtype", "keepdim", "mean", "pow", "return", "rsqrt", "to", "torch", "variance", "variance_epsilon"], "falcon_mamba/modeling_falcon_mamba.py:FalconMambaMixer.__init__": ["A", "ACT2FN", "A_log", "Conv1d", "D", "False", "Linear", "Model", "ModelConfig", "ModelMixer", "Model_inner_fn", "Model_ssm", "Module", "None", "Parameter", "True", "__init__", "act", "activation", "arange", "b_c_rms", "bias", "causal", "causal_conv1d", "causal_conv1d_fn", "causal_conv1d_update", "class", "config", "contiguous", "conv1d", "conv_kernel", "conv_kernel_size", "def", "dt_proj", "dt_rms", "dtype", "else", "expand", "float32", "global", "groups", "hidden_act", "hidden_size", "if", "in_channels", "in_proj", "int", "intermediate_size", "is", "kernel_size", "layer_idx", "lazy_load_kernel", "log", "mixer_rms_eps", "nn", "not", "ones", "out_channels", "out_proj", "padding", "persistent", "register_buffer", "requires_grad", "rms_eps", "selective_scan_fn", "selective_state_update", "self", "ssm", "ssm_state_size", "state_size", "super", "time_step_rank", "torch", "use_Modelpy", "use_bias", "use_conv_bias", "warn_slow_implementation", "x_proj"], "falcon_mamba/modeling_falcon_mamba.py:FalconMambaMixer.warn_slow_implementation": ["AILab", "Dao", "Falling", "False", "For", "ImportError", "Mamba", "ModelMixer", "Model_inner_fn", "Module", "None", "The", "To", "True", "all", "alxndrTL", "as", "available", "back", "backend", "because", "but", "causal", "causal_conv1d_fn", "causal_conv1d_update", "class", "com", "conv1d", "def", "else", "fast", "follow", "for", "github", "https", "if", "implementation", "install", "installed", "is", "is_fast_path_available", "is_mambapy_available", "it", "kernels", "logger", "mamba", "mamba_inner_fn", "mambapy", "nn", "not", "of", "one", "or", "package", "path", "pip", "py", "raise", "selective_scan_fn", "selective_state_update", "self", "sequential", "set", "spaces", "state", "the", "to", "use_Modelpy", "use_mambapy", "warn_slow_implementation", "warning_once"], "falcon_mamba/modeling_falcon_mamba.py:FalconMambaMixer.cuda_kernels_forward": ["A", "A_log", "B", "C", "D", "LongTensor", "ModelCache", "ModelMixer", "Model_inner_fn", "Module", "None", "Tensor", "True", "_is_quantized", "activation", "and", "attention_mask", "b_c_dt_rms_eps", "b_c_rms", "b_rms_weight", "bias", "c_rms_weight", "cache_params", "cache_position", "causal_conv1d_fn", "causal_conv1d_update", "chunk", "class", "config", "contextualized_states", "conv1d", "conv_kernel_size", "conv_states", "conv_weights", "cuda_kernels_forward", "def", "delta_bias", "delta_softplus", "dim", "discrete_time_step", "dt_proj", "dt_rms", "dt_rms_weight", "dt_softplus", "else", "exp", "float", "functional", "gate", "hasattr", "hidden_states", "if", "in_proj", "is", "layer_idx", "nn", "not", "out_proj", "pad", "projected_states", "return", "return_last_state", "rms_eps", "rms_forward", "scan_outputs", "selective_scan_fn", "selective_state_update", "self", "shape", "size", "split", "squeeze", "ssm_parameters", "ssm_state", "ssm_state_size", "ssm_states", "time_proj_bias", "time_step", "time_step_rank", "torch", "training", "transpose", "unsqueeze", "update_conv_state", "update_ssm_state", "use_bias", "use_conv_bias", "variance_epsilon", "view", "weight", "x_proj"], "falcon_mamba/modeling_falcon_mamba.py:FalconMambaMixer.slow_forward": ["A", "A_log", "B", "C", "D", "LongTensor", "ModelCache", "ModelMixer", "Module", "None", "_", "act", "and", "append", "attention_mask", "batch_size", "bias", "cache_params", "cache_position", "chunk", "class", "clone", "contextualized_states", "conv1d", "conv_kernel_size", "conv_state", "def", "deltaB_u", "device", "dim", "discrete_A", "discrete_B", "discrete_time_step", "dt_proj", "dtype", "else", "exp", "float", "for", "functional", "gate", "hidden_states", "hs", "i", "if", "in", "in_proj", "input_states", "intermediate_size", "is", "layer_idx", "matmul", "nn", "not", "out_proj", "pad", "projected_states", "pscan", "range", "return", "rms_eps", "rms_forward", "scan_output", "scan_outputs", "self", "seq_len", "shape", "slow_forward", "softplus", "split", "squeeze", "ssm_parameters", "ssm_state", "ssm_state_size", "ssm_states", "stack", "sum", "time_step", "time_step_rank", "to", "torch", "training", "transpose", "unsqueeze", "update_conv_state", "update_ssm_state", "use_Modelpy", "use_conv_bias", "variance_epsilon", "weight", "x_proj", "zeros"], "falcon_mamba/modeling_falcon_mamba.py:FalconMambaMixer.forward": ["LongTensor", "ModelCache", "ModelMixer", "Model_inner_fn", "Module", "None", "all", "and", "attention_mask", "cache_params", "cache_position", "causal_conv1d_fn", "causal_conv1d_update", "class", "cuda", "cuda_kernels_forward", "def", "device", "forward", "hidden_states", "if", "in", "is_fast_path_available", "is_torchdynamo_compiling", "nn", "not", "return", "selective_scan_fn", "selective_state_update", "self", "slow_forward", "torch", "type", "weight", "x_proj"], "falcon_mamba/modeling_falcon_mamba.py:FalconMambaRMSNorm.__init__": ["ModelRMSNorm", "Module", "Parameter", "__init__", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "falcon_mamba/modeling_falcon_mamba.py:FalconMambaRMSNorm.forward": ["ModelRMSNorm", "Module", "class", "def", "device", "forward", "hidden_states", "nn", "return", "rms_forward", "self", "to", "variance_epsilon", "weight"], "falcon_mamba/modeling_falcon_mamba.py:FalconMambaRMSNorm.extra_repr": ["ModelRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "variance_epsilon", "weight"], "falcon_mamba/modeling_falcon_mamba.py:FalconMambaBlock.__init__": ["GradientCheckpointingLayer", "ModelBlock", "ModelMixer", "ModelRMSNorm", "__init__", "class", "config", "def", "eps", "hidden_size", "layer_idx", "layer_norm_epsilon", "mixer", "norm", "residual_in_fp32", "self", "super"], "falcon_mamba/modeling_falcon_mamba.py:FalconMambaBlock.forward": ["GradientCheckpointingLayer", "LongTensor", "ModelBlock", "ModelCache", "None", "attention_mask", "cache_params", "cache_position", "class", "def", "dtype", "float32", "forward", "hidden_states", "if", "mixer", "norm", "residual", "residual_in_fp32", "return", "self", "to", "torch", "weight"], "falcon_mamba/modeling_falcon_mamba.py:FalconMambaPreTrainedModel._init_weights": ["A", "A_log", "D", "Embedding", "Linear", "ModelMixer", "ModelPreTrainedModel", "ModelRMSNorm", "None", "PreTrainedModel", "_init_weights", "a", "arange", "b_c_rms", "bias", "clamp", "class", "config", "constant", "constant_", "contiguous", "conv1d", "copy_", "def", "dt", "dt_init_std", "dt_proj", "dt_rms", "dtype", "elif", "exp", "expand", "expm1", "float32", "if", "init", "initializer_range", "intermediate_size", "inv_dt", "is", "isinstance", "kaiming_uniform_", "log", "math", "min", "module", "nn", "normal_", "not", "num_hidden_layers", "ones_", "out_proj", "p", "rand", "random", "rescale_prenorm_residual", "self", "sqrt", "ssm_state_size", "std", "time_step_floor", "time_step_init_scheme", "time_step_max", "time_step_min", "time_step_rank", "time_step_scale", "torch", "uniform_", "weight", "zeros_"], "falcon_mamba/modeling_falcon_mamba.py:FalconMambaModel.__init__": ["Embedding", "False", "ModelBlock", "ModelModel", "ModelPreTrainedModel", "ModelRMSNorm", "ModuleList", "__init__", "class", "config", "def", "embeddings", "eps", "for", "gradient_checkpointing", "hidden_size", "idx", "in", "layer_idx", "layer_norm_epsilon", "layers", "nn", "norm_f", "num_hidden_layers", "post_init", "range", "self", "super", "vocab_size"], "falcon_mamba/modeling_falcon_mamba.py:FalconMambaModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "return", "self"], "falcon_mamba/modeling_falcon_mamba.py:FalconMambaModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "new_embeddings", "self", "set_input_embeddings"], "falcon_mamba/modeling_falcon_mamba.py:FalconMambaModel.forward": ["False", "LongTensor", "ModelCache", "ModelModel", "ModelOutput", "ModelPreTrainedModel", "None", "True", "ValueError", "You", "a", "all_hidden_states", "and", "arange", "are", "attention_mask", "automatically", "be", "because", "bool", "cache_params", "cache_position", "case", "class", "config", "conv_kernel", "def", "device", "don", "dtype", "elif", "else", "embeddings", "exactly", "for", "forward", "gradient_checkpointing", "have", "hidden_states", "if", "in", "initialized", "input_ids", "inputs_embeds", "is", "it", "kwargs", "last_hidden_state", "layers", "manually", "mixer_block", "must", "norm_f", "not", "of", "one", "or", "output_hidden_states", "pass", "passed", "prefilling", "r", "raise", "return", "return_dict", "self", "size", "specify", "stage", "t", "that", "the", "to", "torch", "training", "tuple", "use_cache", "use_return_dict", "v", "when", "will", "you"], "falcon_mamba/modeling_falcon_mamba.py:FalconMambaForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "ModelForCausalLM", "ModelModel", "ModelPreTrainedModel", "__init__", "backbone", "bias", "class", "config", "def", "hidden_size", "lm_head", "nn", "post_init", "self", "super", "vocab_size"], "falcon_mamba/modeling_falcon_mamba.py:FalconMambaForCausalLM.get_input_embeddings": ["GenerationMixin", "ModelForCausalLM", "ModelPreTrainedModel", "backbone", "class", "def", "get_input_embeddings", "return", "self"], "falcon_mamba/modeling_falcon_mamba.py:FalconMambaForCausalLM.set_input_embeddings": ["GenerationMixin", "ModelForCausalLM", "ModelPreTrainedModel", "backbone", "class", "def", "new_embeddings", "return", "self", "set_input_embeddings"], "falcon_mamba/modeling_falcon_mamba.py:FalconMambaForCausalLM._update_model_kwargs_for_generation": ["Any", "GenerationMixin", "ModelForCausalLM", "ModelOutput", "ModelPreTrainedModel", "None", "True", "_update_model_kwargs_for_generation", "and", "attention_mask", "cache_params", "cache_position", "cat", "class", "def", "dict", "dim", "get", "if", "in", "int", "is", "kwargs", "model_kwargs", "new_ones", "not", "num_new_tokens", "outputs", "return", "self", "shape", "str", "torch", "use_cache"], "falcon_mamba/modeling_falcon_mamba.py:FalconMambaForCausalLM.prepare_inputs_for_generation": ["False", "GenerationMixin", "LongTensor", "ModelCache", "ModelForCausalLM", "ModelPreTrainedModel", "None", "and", "arange", "attention_mask", "backbone", "bool", "cache_params", "cache_position", "class", "config", "contiguous", "conv_kernel", "def", "device", "dtype", "else", "for", "if", "in", "input_ids", "inputs_embeds", "is", "is_first_iteration", "items", "key", "kwargs", "max_batch_size", "model_inputs", "not", "prepare_inputs_for_generation", "return", "self", "size", "torch", "unsqueeze", "update", "use_cache", "value"], "falcon_mamba/modeling_falcon_mamba.py:FalconMambaForCausalLM.forward": ["CrossEntropyLoss", "FloatTensor", "GenerationMixin", "LongTensor", "ModelCache", "ModelCausalLMOutput", "ModelForCausalLM", "ModelPreTrainedModel", "Model_outputs", "None", "Tensor", "attention_mask", "backbone", "bool", "cache_params", "cache_position", "class", "config", "contiguous", "def", "device", "dtype", "else", "float", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "lm_head", "logits", "logits_to_keep", "loss", "loss_fct", "not", "output", "output_hidden_states", "r", "return", "return_dict", "self", "shift_labels", "shift_logits", "size", "slice", "slice_indices", "to", "torch", "tuple", "use_cache", "use_return_dict", "view", "weight"], "fast_vlm/modeling_fast_vlm.py:FastVlmMultiModalProjector.__init__": ["ACT2FN", "Linear", "ModelConfig", "ModelMultiModalProjector", "Module", "__init__", "act", "bias", "class", "config", "def", "hidden_size", "linear_1", "linear_2", "multimodal_projector_bias", "nn", "projector_hidden_act", "self", "super", "text_config", "vision_config"], "fast_vlm/modeling_fast_vlm.py:FastVlmMultiModalProjector.forward": ["ModelMultiModalProjector", "Module", "act", "class", "def", "forward", "hidden_states", "image_features", "linear_1", "linear_2", "nn", "return", "self"], "fast_vlm/modeling_fast_vlm.py:FastVlmModel.__init__": ["AutoModel", "ModelConfig", "ModelModel", "ModelMultiModalProjector", "ModelPreTrainedModel", "__init__", "class", "config", "def", "from_config", "language_model", "multi_modal_projector", "post_init", "self", "super", "text_config", "vision_config", "vision_tower"], "fast_vlm/modeling_fast_vlm.py:FastVlmModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "language_model", "return", "self"], "fast_vlm/modeling_fast_vlm.py:FastVlmModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "language_model", "self", "set_input_embeddings", "value"], "fast_vlm/modeling_fast_vlm.py:FastVlmModel.get_image_features": ["FloatTensor", "ModelModel", "ModelPreTrainedModel", "None", "class", "config", "def", "else", "flatten", "for", "get_image_features", "if", "image_features", "image_outputs", "in", "int", "is", "items", "k", "kwargs", "last_hidden_state", "list", "multi_modal_projector", "not", "permute", "pixel_values", "return", "selected_image_feature", "self", "str", "torch", "v", "vision_feature_layer", "vision_feature_select_strategy", "vision_tower"], "fast_vlm/modeling_fast_vlm.py:FastVlmModel.get_placeholder_mask": ["FloatTensor", "Image", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "ValueError", "all", "and", "class", "config", "def", "device", "do", "dtype", "else", "expand_as", "f", "features", "get_input_embeddings", "get_placeholder_mask", "if", "image", "image_features", "image_token_id", "input_ids", "inputs_embeds", "is", "long", "match", "n_image_features", "n_image_tokens", "not", "numel", "raise", "return", "self", "shape", "special_image_mask", "sum", "tensor", "to", "tokens", "torch", "unsqueeze"], "fast_vlm/modeling_fast_vlm.py:FastVlmModel.forward": ["Cache", "FloatTensor", "LongTensor", "ModelModel", "ModelModelOutputWithPast", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "attention_mask", "attentions", "cache_position", "cat", "class", "config", "def", "device", "dim", "dtype", "else", "exactly", "forward", "get_image_features", "get_input_embeddings", "get_placeholder_mask", "hidden_states", "if", "image_features", "image_hidden_states", "image_sizes", "input_ids", "inputs_embeds", "int", "is", "kwargs", "language_model", "last_hidden_state", "list", "masked_scatter", "must", "not", "of", "one", "or", "outputs", "past_key_values", "pixel_values", "position_ids", "r", "raise", "return", "self", "special_image_mask", "specify", "str", "to", "torch", "tuple", "vision_feature_layer", "vision_feature_select_strategy"], "fast_vlm/modeling_fast_vlm.py:FastVlmForConditionalGeneration.__init__": ["False", "GenerationMixin", "Linear", "ModelConfig", "ModelForConditionalGeneration", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "text_config", "vocab_size"], "fast_vlm/modeling_fast_vlm.py:FastVlmForConditionalGeneration.get_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "model", "return", "self"], "fast_vlm/modeling_fast_vlm.py:FastVlmForConditionalGeneration.set_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "model", "self", "set_input_embeddings", "value"], "fast_vlm/modeling_fast_vlm.py:FastVlmForConditionalGeneration.get_output_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "Module", "class", "def", "get_output_embeddings", "lm_head", "nn", "return", "self"], "fast_vlm/modeling_fast_vlm.py:FastVlmForConditionalGeneration.get_image_features": ["FloatTensor", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "class", "def", "get_image_features", "int", "kwargs", "list", "model", "pixel_values", "return", "self", "str", "torch", "vision_feature_layer", "vision_feature_select_strategy"], "fast_vlm/modeling_fast_vlm.py:FastVlmForConditionalGeneration.forward": ["Cache", "FloatTensor", "GenerationMixin", "LongTensor", "ModelCausalLMOutputWithPast", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "cache_position", "class", "config", "def", "else", "forward", "hidden_states", "if", "image_hidden_states", "image_sizes", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "list", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "outputs", "past_key_values", "pixel_values", "position_ids", "r", "return", "self", "slice", "slice_indices", "str", "text_config", "torch", "tuple", "vision_feature_layer", "vision_feature_select_strategy", "vocab_size"], "fast_vlm/modeling_fast_vlm.py:FastVlmForConditionalGeneration.prepare_inputs_for_generation": ["False", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "True", "attention_mask", "cache_position", "class", "def", "get", "if", "input_ids", "inputs_embeds", "is_first_iteration", "kwargs", "logits_to_keep", "model_inputs", "not", "or", "past_key_values", "pixel_values", "prepare_inputs_for_generation", "return", "self", "super", "use_cache"], "fastspeech2_conformer/modeling_fastspeech2_conformer.py:length_regulator": ["Model_regulator", "ValueError", "be", "def", "device", "dim", "dtype", "duration_labels", "elif", "encoded_embedding", "encoded_embeddings", "enumerate", "eq", "float", "for", "greater", "hidden_states", "i", "if", "in", "long", "max", "max_len", "must", "raise", "repeat_interleave", "repeated", "return", "round", "size", "speaking_speed", "sum", "target_duration", "than", "torch", "zeros", "zip"], "fastspeech2_conformer/modeling_fastspeech2_conformer.py:FastSpeech2ConformerDurationPredictor.__init__": ["Linear", "ModelConfig", "ModelDurationPredictor", "ModelPredictorLayer", "Module", "ModuleList", "__init__", "append", "class", "config", "conv_layers", "def", "duration_predictor_channels", "duration_predictor_dropout_rate", "duration_predictor_kernel_size", "duration_predictor_layers", "else", "for", "hidden_size", "if", "in", "input_channels", "layer", "layer_idx", "linear", "log_domain_offset", "nn", "num_chans", "range", "self", "super"], "fastspeech2_conformer/modeling_fastspeech2_conformer.py:FastSpeech2ConformerDurationPredictor.forward": ["ModelDurationPredictor", "Module", "clamp", "class", "conv_layers", "def", "encoder_hidden_states", "exp", "for", "forward", "hidden_states", "if", "in", "layer", "linear", "log_domain_offset", "long", "min", "nn", "not", "return", "round", "self", "squeeze", "torch", "training", "transpose"], "fastspeech2_conformer/modeling_fastspeech2_conformer.py:FastSpeech2ConformerBatchNormConvLayer.__init__": ["BatchNorm1d", "Conv1d", "Dropout", "False", "ModelBatchNormConvLayer", "Module", "None", "Tanh", "__init__", "activation", "batch_norm", "bias", "class", "config", "conv", "def", "dropout", "else", "if", "in_conv_dim", "kernel_size", "layer_id", "nn", "num_mel_bins", "out_conv_dim", "padding", "self", "speech_decoder_postnet_dropout", "speech_decoder_postnet_kernel", "speech_decoder_postnet_layers", "speech_decoder_postnet_units", "stride", "super"], "fastspeech2_conformer/modeling_fastspeech2_conformer.py:FastSpeech2ConformerBatchNormConvLayer.forward": ["ModelBatchNormConvLayer", "Module", "None", "activation", "batch_norm", "class", "conv", "def", "dropout", "forward", "hidden_states", "if", "is", "nn", "not", "return", "self"], "fastspeech2_conformer/modeling_fastspeech2_conformer.py:FastSpeech2ConformerSpeechDecoderPostnet.__init__": ["Linear", "ModelBatchNormConvLayer", "ModelSpeechDecoderPostnet", "Module", "ModuleList", "__init__", "class", "config", "def", "feat_out", "for", "hidden_size", "i", "in", "layers", "nn", "num_mel_bins", "range", "reduction_factor", "self", "speech_decoder_postnet_layers", "super"], "fastspeech2_conformer/modeling_fastspeech2_conformer.py:FastSpeech2ConformerSpeechDecoderPostnet.forward": ["ModelSpeechDecoderPostnet", "Module", "Tensor", "class", "config", "def", "feat_out", "for", "forward", "hidden_states", "in", "layer", "layer_output", "layers", "nn", "num_mel_bins", "outputs_after_postnet", "outputs_before_postnet", "return", "self", "size", "torch", "transpose", "view"], "fastspeech2_conformer/modeling_fastspeech2_conformer.py:FastSpeech2ConformerPredictorLayer.__init__": ["Conv1d", "Dropout", "LayerNorm", "ModelPredictorLayer", "Module", "ReLU", "__init__", "activation", "class", "conv", "def", "dropout", "dropout_rate", "input_channels", "kernel_size", "layer_norm", "nn", "num_chans", "padding", "self", "stride", "super"], "fastspeech2_conformer/modeling_fastspeech2_conformer.py:FastSpeech2ConformerPredictorLayer.forward": ["ModelPredictorLayer", "Module", "activation", "class", "conv", "def", "dropout", "forward", "hidden_states", "layer_norm", "nn", "return", "self", "transpose"], "fastspeech2_conformer/modeling_fastspeech2_conformer.py:FastSpeech2ConformerVariancePredictor.__init__": ["Linear", "ModelConfig", "ModelPredictorLayer", "ModelVariancePredictor", "Module", "ModuleList", "__init__", "append", "class", "config", "conv_layers", "def", "dropout_rate", "else", "for", "hidden_size", "idx", "if", "in", "input_channels", "kernel_size", "layer", "linear", "nn", "num_chans", "num_layers", "range", "self", "super"], "fastspeech2_conformer/modeling_fastspeech2_conformer.py:FastSpeech2ConformerVariancePredictor.forward": ["ModelVariancePredictor", "Module", "None", "class", "conv_layers", "def", "encoder_hidden_states", "for", "forward", "hidden_states", "if", "in", "is", "layer", "linear", "masked_fill", "nn", "not", "padding_masks", "return", "self", "transpose"], "fastspeech2_conformer/modeling_fastspeech2_conformer.py:FastSpeech2ConformerVarianceEmbedding.__init__": ["Conv1d", "Dropout", "ModelVarianceEmbedding", "Module", "__init__", "class", "conv", "def", "dropout", "dropout_rate", "in_channels", "kernel_size", "nn", "out_channels", "padding", "self", "super"], "fastspeech2_conformer/modeling_fastspeech2_conformer.py:FastSpeech2ConformerVarianceEmbedding.forward": ["ModelVarianceEmbedding", "Module", "class", "conv", "def", "dropout", "forward", "hidden_states", "nn", "return", "self", "transpose"], "fastspeech2_conformer/modeling_fastspeech2_conformer.py:FastSpeech2ConformerAttention.__init__": ["Dropout", "False", "Linear", "ModelAttention", "ModelConfig", "Module", "Parameter", "Tensor", "__init__", "attention_dropout_rate", "bias", "class", "config", "def", "dim_key", "dropout", "head_dim", "hidden_size", "linear_k", "linear_out", "linear_pos", "linear_q", "linear_v", "module_config", "nn", "num_attention_heads", "num_heads", "p", "pos_bias_u", "pos_bias_v", "self", "super", "torch"], "fastspeech2_conformer/modeling_fastspeech2_conformer.py:FastSpeech2ConformerAttention.shift_relative_position_tensor": ["ModelAttention", "Module", "cat", "class", "def", "device", "dim", "dtype", "nn", "pos_tensor", "pos_tensor_padded", "return", "self", "shift_relative_position_tensor", "size", "torch", "view", "view_as", "zero_pad", "zeros"], "fastspeech2_conformer/modeling_fastspeech2_conformer.py:FastSpeech2ConformerAttention.forward": ["Attention", "False", "ModelAttention", "Module", "None", "Tensor", "ValueError", "_", "attention_mask", "attn_output", "attn_weights", "be", "bsz", "bsz_pos", "but", "class", "contiguous", "def", "dim", "dim_key", "dropout", "dtype", "else", "eq", "expected_size", "f", "finfo", "float", "forward", "head_dim", "hidden_states", "if", "is", "key_states", "linear_k", "linear_out", "linear_pos", "linear_q", "linear_v", "mask", "masked_fill", "math", "matmul", "matrix_ac", "matrix_bd", "min", "min_value", "nn", "not", "num_heads", "of", "output_attentions", "permute", "pos_bias_u", "pos_bias_v", "pos_emb", "pos_encoding", "q_len", "query_states", "query_with_bias_u", "query_with_bias_v", "raise", "return", "scores", "self", "shift_relative_position_tensor", "should", "size", "softmax", "sqrt", "torch", "transpose", "tuple", "unsqueeze", "value_states", "view"], "fastspeech2_conformer/modeling_fastspeech2_conformer.py:FastSpeech2ConformerConvolutionModule.__init__": ["ACT2FN", "BatchNorm1d", "Conv1d", "ModelConfig", "ModelConvolutionModule", "Module", "None", "__init__", "activation", "bias", "channels", "class", "config", "conv_kernel_size", "convolution_bias", "def", "depthwise_conv", "else", "get", "getattr", "groups", "hidden_act", "hidden_size", "if", "is", "kernel_size", "module_config", "nn", "norm", "padding", "pointwise_conv1", "pointwise_conv2", "self", "silu", "stride", "super"], "fastspeech2_conformer/modeling_fastspeech2_conformer.py:FastSpeech2ConformerConvolutionModule.forward": ["ModelConvolutionModule", "Module", "None", "activation", "all", "all_masked_rows", "attention_mask", "bool", "class", "def", "depthwise_conv", "dim", "dtype", "else", "forward", "functional", "glu", "hidden_states", "if", "is", "masked_fill", "nn", "norm", "not", "pointwise_conv1", "pointwise_conv2", "return", "self", "torch", "transpose"], "fastspeech2_conformer/modeling_fastspeech2_conformer.py:FastSpeech2ConformerEncoderLayer.__init__": ["Dropout", "LayerNorm", "Linear", "ModelAttention", "ModelConfig", "ModelConvolutionModule", "ModelEncoderLayer", "ModelMultiLayeredConv1d", "Module", "__init__", "class", "concat_after", "concat_linear", "config", "conv_layer_norm", "conv_module", "def", "dropout", "dropout_rate", "else", "feed_forward", "feed_forward_macaron", "ff_layer_norm", "ff_macaron_layer_norm", "ff_scale", "final_layer_norm", "hidden_size", "if", "macaron_style", "module_config", "nn", "normalize_before", "self", "self_attn", "self_attn_layer_norm", "size", "super", "use_cnn_in_conformer", "use_cnn_module", "use_macaron_style_in_conformer"], "fastspeech2_conformer/modeling_fastspeech2_conformer.py:FastSpeech2ConformerEncoderLayer.forward": ["False", "ModelEncoderLayer", "Module", "None", "Tensor", "attention_mask", "attention_output", "attention_scores", "cat", "class", "concat_after", "concat_linear", "conv_layer_norm", "conv_module", "def", "dim", "dropout", "else", "feed_forward", "feed_forward_macaron", "ff_layer_norm", "ff_macaron_layer_norm", "ff_scale", "final_layer_norm", "forward", "hidden_states", "if", "is", "macaron_style", "nn", "normalize_before", "not", "output_attentions", "outputs", "pos_emb", "residual", "return", "self", "self_attn", "self_attn_layer_norm", "torch", "use_cnn_module", "x_concat"], "fastspeech2_conformer/modeling_fastspeech2_conformer.py:FastSpeech2ConformerMultiLayeredConv1d.__init__": ["Conv1d", "Dropout", "ModelConfig", "ModelMultiLayeredConv1d", "Module", "__init__", "class", "config", "conv1", "conv2", "def", "dropout", "dropout_rate", "hidden_channels", "hidden_size", "input_channels", "kernel_size", "linear_units", "module_config", "nn", "padding", "positionwise_conv_kernel_size", "self", "stride", "super"], "fastspeech2_conformer/modeling_fastspeech2_conformer.py:FastSpeech2ConformerMultiLayeredConv1d.forward": ["ModelMultiLayeredConv1d", "Module", "class", "conv1", "conv2", "def", "dropout", "forward", "hidden_states", "nn", "relu", "return", "self", "torch", "transpose"], "fastspeech2_conformer/modeling_fastspeech2_conformer.py:FastSpeech2ConformerRelPositionalEncoding.__init__": ["Dropout", "False", "ModelConfig", "ModelRelPositionalEncoding", "Module", "__init__", "class", "config", "def", "dropout", "embed_dim", "expand", "extend_pos_enc", "hidden_size", "input_scale", "math", "max_len", "module_config", "nn", "p", "persistent", "pos_enc", "positional_dropout_rate", "register_buffer", "self", "sqrt", "super", "tensor", "torch"], "fastspeech2_conformer/modeling_fastspeech2_conformer.py:FastSpeech2ConformerRelPositionalEncoding.extend_pos_enc": ["ModelRelPositionalEncoding", "Module", "None", "arange", "cat", "class", "cos", "def", "device", "dim", "div_term", "dtype", "embed_dim", "exp", "extend_pos_enc", "flip", "float", "if", "int64", "is", "log", "math", "nn", "not", "or", "pos_enc", "pos_enc_negative", "pos_enc_positive", "position", "return", "self", "sin", "size", "to", "torch", "unsqueeze", "x", "zeros"], "fastspeech2_conformer/modeling_fastspeech2_conformer.py:FastSpeech2ConformerRelPositionalEncoding.forward": ["ModelRelPositionalEncoding", "Module", "center_idx", "class", "def", "dropout", "extend_pos_enc", "feature_representation", "forward", "hidden_states", "input_scale", "nn", "pos_emb", "pos_enc", "return", "self", "size"], "fastspeech2_conformer/modeling_fastspeech2_conformer.py:FastSpeech2ConformerEncoder.__init__": ["Embedding", "False", "ModelConfig", "ModelEncoder", "ModelEncoderLayer", "ModelRelPositionalEncoding", "Module", "ModuleList", "None", "_", "__init__", "class", "config", "conformer_layers", "def", "embed", "embedding_dim", "for", "hidden_size", "if", "in", "layers", "module_config", "nn", "num_embeddings", "padding_idx", "pos_enc", "range", "self", "super", "use_encoder_input_layer", "vocab_size"], "fastspeech2_conformer/modeling_fastspeech2_conformer.py:FastSpeech2ConformerEncoder.forward": ["BaseModelOutput", "False", "LongTensor", "ModelEncoder", "Module", "None", "all_hidden_states", "all_self_attentions", "attention_mask", "attentions", "bool", "class", "conformer_layer", "conformer_layers", "def", "else", "embed", "feature_representation", "for", "forward", "hidden_states", "if", "in", "input_tensor", "is", "last_hidden_state", "layer_outputs", "nn", "not", "output_attentions", "output_hidden_states", "pos_emb", "pos_enc", "return", "return_dict", "self", "torch", "tuple", "v"], "fastspeech2_conformer/modeling_fastspeech2_conformer.py:FastSpeech2ConformerLoss.__init__": ["Either", "L1Loss", "MSELoss", "ModelConfig", "ModelLoss", "Module", "True", "ValueError", "__init__", "and", "be", "both", "but", "can", "class", "config", "def", "duration_criterion", "else", "if", "l1_criterion", "log_domain_offset", "mean", "mse_criterion", "nn", "none", "not", "or", "raise", "reduction", "self", "super", "use_masking", "use_weighted_masking"], "fastspeech2_conformer/modeling_fastspeech2_conformer.py:FastSpeech2ConformerLoss.forward": ["False", "ModelLoss", "Module", "None", "True", "class", "def", "dim", "duration_criterion", "duration_labels", "duration_loss", "duration_mask", "duration_outputs", "duration_weights", "energy_labels", "energy_loss", "energy_outputs", "float", "forward", "functional", "if", "is", "keepdim", "l1_criterion", "l1_loss", "log", "log_domain_offset", "masked_select", "mse_criterion", "mul", "nn", "not", "out_weights", "outputs_after_postnet", "outputs_before_postnet", "pad", "pitch_and_energy_masks", "pitch_labels", "pitch_loss", "pitch_outputs", "pitch_weights", "return", "self", "size", "spectrogram_labels", "spectrogram_mask", "sum", "torch", "transpose", "unsqueeze", "use_masking", "use_weighted_masking", "value"], "fastspeech2_conformer/modeling_fastspeech2_conformer.py:FastSpeech2ConformerPreTrainedModel._init_weights": ["BatchNorm1d", "Conv1d", "Embedding", "False", "LayerNorm", "Linear", "ModelAttention", "ModelPreTrainedModel", "ModelRelPositionalEncoding", "None", "PreTrainedModel", "_init_weights", "_is_hf_initialized", "a", "and", "b", "bias", "class", "copy_", "def", "elif", "expand", "extend_pos_enc", "getattr", "groups", "if", "in_channels", "init", "is", "isinstance", "kaiming_normal_", "kernel_size", "key", "math", "max_len", "module", "nn", "normal_", "not", "num_batches_tracked", "ones_", "padding_idx", "pos_bias_u", "pos_bias_v", "pos_enc", "running_mean", "running_var", "self", "size", "sqrt", "std", "tensor", "torch", "uniform_", "weight", "xavier_uniform_", "zeros_"], "fastspeech2_conformer/modeling_fastspeech2_conformer.py:FastSpeech2ConformerPreTrainedModel._set_gradient_checkpointing": ["False", "ModelEncoder", "ModelPreTrainedModel", "PreTrainedModel", "_set_gradient_checkpointing", "class", "def", "gradient_checkpointing", "if", "isinstance", "module", "self", "value"], "fastspeech2_conformer/modeling_fastspeech2_conformer.py:FastSpeech2ConformerModel.__init__": ["Embedding", "False", "Linear", "ModelConfig", "ModelDurationPredictor", "ModelEncoder", "ModelLoss", "ModelModel", "ModelPreTrainedModel", "ModelSpeechDecoderPostnet", "ModelVarianceEmbedding", "ModelVariancePredictor", "None", "True", "__init__", "and", "class", "config", "criterion", "decoder", "decoder_config", "def", "dropout_rate", "duration_predictor", "encoder", "encoder_config", "energy_embed", "energy_embed_dropout", "energy_embed_kernel_size", "energy_predictor", "energy_predictor_channels", "energy_predictor_dropout", "energy_predictor_kernel_size", "energy_predictor_layers", "hidden_size", "if", "is", "kernel_size", "language_id_embedding", "multilingual_model", "multispeaker_model", "nn", "not", "num_chans", "num_languages", "num_layers", "num_mel_bins", "num_speakers", "out_channels", "padding", "pitch_embed", "pitch_embed_dropout", "pitch_embed_kernel_size", "pitch_predictor", "pitch_predictor_channels", "pitch_predictor_dropout", "pitch_predictor_kernel_size", "pitch_predictor_layers", "post_init", "projection", "reduction_factor", "self", "speaker_embed_dim", "speaker_id_embedding", "speech_decoder_postnet", "stop_gradient_from_energy_predictor", "stop_gradient_from_pitch_predictor", "super", "torch", "use_encoder_input_layer", "vocab_size"], "fastspeech2_conformer/modeling_fastspeech2_conformer.py:FastSpeech2ConformerModel.forward": ["All", "FloatTensor", "LongTensor", "ModelModel", "ModelModelOutput", "ModelPreTrainedModel", "None", "ValueError", "and", "any", "attention_mask", "attentions", "audio_feature_predictions", "be", "bool", "cat", "class", "config", "criterion", "decoder", "decoder_attentions", "decoder_hidden_states", "decoder_outputs", "def", "detach", "device", "dim", "duration_labels", "duration_mask", "duration_outputs", "duration_predictions", "duration_predictor", "else", "embedded_energy_curve", "embedded_pitch_curve", "embeddings_expanded", "encoder", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "energy_embed", "energy_labels", "energy_outputs", "energy_predictions", "energy_predictor", "expand", "forward", "functional", "has_missing_labels", "hidden_mask", "hidden_states", "if", "in", "input_ids", "int", "is", "kwargs", "labels", "lang_ids", "language_id_embbedings", "language_id_embedding", "last_hidden_state", "length_dim", "length_regulator", "loss", "loss_duration_mask", "loss_spectrogram_mask", "masked_fill", "mode", "multilingual_model", "multispeaker_model", "must", "nn", "normalize", "not", "ones", "or", "output_attentions", "output_hidden_states", "outputs", "outputs_after_postnet", "outputs_before_postnet", "pitch_embed", "pitch_labels", "pitch_outputs", "pitch_predictions", "pitch_predictor", "postnet_outputs", "projection", "provided", "r", "raise", "reduction_factor", "return", "return_dict", "run", "self", "shape", "size", "speaker_embed_dim", "speaker_embedding", "speaker_id_embedding", "speaker_id_embeddings", "speaker_ids", "speaking_speed", "spectrogram", "spectrogram_labels", "spectrogram_mask", "speech_decoder_postnet", "stop_gradient_from_energy_predictor", "stop_gradient_from_pitch_predictor", "text_masks", "to", "torch", "training", "tuple", "unsqueeze", "use_return_dict", "view"], "fastspeech2_conformer/modeling_fastspeech2_conformer.py:HifiGanResidualBlock.__init__": ["Conv1d", "ModelGanResidualBlock", "Module", "ModuleList", "_", "__init__", "channels", "class", "convs1", "convs2", "def", "dilation", "for", "get_padding", "i", "in", "kernel_size", "leaky_relu_slope", "len", "nn", "padding", "range", "self", "stride", "super"], "fastspeech2_conformer/modeling_fastspeech2_conformer.py:HifiGanResidualBlock.get_padding": ["ModelGanResidualBlock", "Module", "class", "def", "dilation", "get_padding", "kernel_size", "nn", "return", "self"], "fastspeech2_conformer/modeling_fastspeech2_conformer.py:HifiGanResidualBlock.apply_weight_norm": ["ModelGanResidualBlock", "Module", "apply_weight_norm", "class", "convs1", "convs2", "def", "for", "hasattr", "if", "in", "layer", "nn", "parametrizations", "self", "utils", "weight_norm"], "fastspeech2_conformer/modeling_fastspeech2_conformer.py:HifiGanResidualBlock.remove_weight_norm": ["ModelGanResidualBlock", "Module", "class", "convs1", "convs2", "def", "for", "in", "layer", "nn", "remove_weight_norm", "self", "utils"], "fastspeech2_conformer/modeling_fastspeech2_conformer.py:HifiGanResidualBlock.forward": ["ModelGanResidualBlock", "Module", "class", "conv1", "conv2", "convs1", "convs2", "def", "for", "forward", "functional", "hidden_states", "in", "leaky_relu", "leaky_relu_slope", "nn", "residual", "return", "self", "zip"], "fastspeech2_conformer/modeling_fastspeech2_conformer.py:FastSpeech2ConformerHifiGan.__init__": ["Conv1d", "ConvTranspose1d", "HifiGanResidualBlock", "ModelHifiGan", "ModelHifiGanConfig", "ModuleList", "PreTrainedModel", "__init__", "append", "channels", "class", "config", "conv_post", "conv_pre", "def", "dilation", "enumerate", "for", "i", "in", "kernel_size", "leaky_relu_slope", "len", "mean", "model_in_dim", "nn", "num_kernels", "num_upsamples", "ones", "padding", "post_init", "range", "register_buffer", "resblock_dilation_sizes", "resblock_kernel_sizes", "resblocks", "scale", "self", "stride", "super", "torch", "upsample_initial_channel", "upsample_kernel_sizes", "upsample_rate", "upsample_rates", "upsampler", "zeros", "zip"], "fastspeech2_conformer/modeling_fastspeech2_conformer.py:FastSpeech2ConformerHifiGan._init_weights": ["ModelHifiGan", "PreTrainedModel", "_init_weights", "class", "def", "if", "init", "isinstance", "mean", "module", "ones_", "scale", "self", "super", "zeros_"], "fastspeech2_conformer/modeling_fastspeech2_conformer.py:FastSpeech2ConformerHifiGan.apply_weight_norm": ["ModelHifiGan", "PreTrainedModel", "apply_weight_norm", "class", "conv_post", "conv_pre", "def", "for", "hasattr", "if", "in", "layer", "nn", "parametrizations", "resblocks", "self", "upsampler", "utils", "weight_norm"], "fastspeech2_conformer/modeling_fastspeech2_conformer.py:FastSpeech2ConformerHifiGan.remove_weight_norm": ["ModelHifiGan", "PreTrainedModel", "class", "conv_post", "conv_pre", "def", "for", "in", "layer", "nn", "remove_weight_norm", "resblocks", "self", "upsampler", "utils"], "fastspeech2_conformer/modeling_fastspeech2_conformer.py:FastSpeech2ConformerHifiGan.forward": ["FloatTensor", "ModelHifiGan", "PreTrainedModel", "class", "config", "conv_post", "conv_pre", "def", "dim", "else", "for", "forward", "functional", "hidden_states", "i", "if", "in", "is_batched", "j", "kwargs", "leaky_relu", "leaky_relu_slope", "mean", "nn", "normalize_before", "not", "num_kernels", "num_upsamples", "r", "range", "res_state", "resblocks", "return", "scale", "self", "spectrogram", "squeeze", "tanh", "torch", "transpose", "unsqueeze", "upsampler", "view", "waveform"], "fastspeech2_conformer/modeling_fastspeech2_conformer.py:FastSpeech2ConformerWithHifiGan.__init__": ["ModelHifiGan", "ModelModel", "ModelWithHifiGan", "ModelWithHifiGanConfig", "PreTrainedModel", "__init__", "class", "config", "def", "model", "model_config", "post_init", "self", "super", "vocoder", "vocoder_config"], "fastspeech2_conformer/modeling_fastspeech2_conformer.py:FastSpeech2ConformerWithHifiGan.forward": ["FloatTensor", "LongTensor", "ModelModelOutput", "ModelWithHifiGan", "ModelWithHifiGanOutput", "None", "PreTrainedModel", "attention_mask", "bool", "class", "config", "def", "duration_labels", "else", "energy_labels", "forward", "has_missing_labels", "if", "input_ids", "is", "kwargs", "lang_ids", "model", "model_config", "model_outputs", "not", "or", "output_attentions", "output_hidden_states", "pitch_labels", "r", "return", "return_dict", "self", "speaker_embedding", "speaker_ids", "spectrogram", "spectrogram_labels", "torch", "tuple", "use_return_dict", "vocoder", "waveform"], "flaubert/modeling_flaubert.py:create_sinusoidal_embeddings": ["False", "FloatTensor", "Model_sinusoidal_embeddings", "array", "cos", "def", "detach_", "dim", "for", "in", "j", "n_pos", "np", "out", "pos", "position_enc", "power", "range", "requires_grad", "return", "sin", "torch"], "flaubert/modeling_flaubert.py:get_masks": ["False", "Model_masks", "None", "alen", "arange", "assert", "attn_mask", "bs", "causal", "def", "device", "dtype", "else", "if", "is", "item", "lengths", "long", "mask", "max", "not", "or", "padding_mask", "repeat", "return", "size", "slen", "torch"], "flaubert/modeling_flaubert.py:MultiHeadAttention.__init__": ["Linear", "ModelHeadAttention", "Module", "__init__", "assert", "attention_dropout", "class", "config", "def", "dim", "dropout", "head_dim", "int", "k_lin", "layer_id", "layer_idx", "n_heads", "nn", "out_lin", "q_lin", "self", "super", "v_lin"], "flaubert/modeling_flaubert.py:MultiHeadAttention.forward": ["EncoderDecoderCache", "False", "ModelHeadAttention", "Module", "None", "True", "and", "bs", "cache", "cache_position", "class", "context", "contiguous", "cross_attention_cache", "curr_past_key_values", "current_states", "def", "dim", "dropout", "dtype", "else", "expand_as", "finfo", "float", "forward", "functional", "get", "head_dim", "if", "input", "is", "is_cross_attention", "is_updated", "isinstance", "k", "k_lin", "key_cache", "kv", "layer_id", "mask", "mask_reshape", "masked_fill_", "math", "matmul", "min", "n_heads", "nn", "not", "out_lin", "output_attentions", "outputs", "p", "q", "q_lin", "qlen", "return", "scores", "self", "self_attention_cache", "size", "softmax", "sqrt", "torch", "training", "transpose", "type_as", "update", "v", "v_lin", "value_cache", "view", "weights"], "flaubert/modeling_flaubert.py:TransformerFFN.__init__": ["Linear", "ModelFFN", "Module", "__init__", "act", "chunk_size_feed_forward", "class", "config", "def", "dim_hidden", "dropout", "else", "functional", "gelu", "gelu_activation", "if", "in_dim", "lin1", "lin2", "nn", "out_dim", "relu", "self", "seq_len_dim", "super"], "flaubert/modeling_flaubert.py:TransformerFFN.forward": ["ModelFFN", "Module", "apply_chunking_to_forward", "chunk_size_feed_forward", "class", "def", "ff_chunk", "forward", "input", "nn", "return", "self", "seq_len_dim"], "flaubert/modeling_flaubert.py:TransformerFFN.ff_chunk": ["ModelFFN", "Module", "act", "class", "def", "dropout", "ff_chunk", "functional", "input", "lin1", "lin2", "nn", "p", "return", "self", "training", "x"], "flaubert/modeling_flaubert.py:FlaubertPredLayer.__init__": ["AdaptiveLogSoftmaxWithLoss", "False", "Linear", "ModelPredLayer", "Module", "True", "__init__", "asm", "asm_cutoffs", "asm_div_value", "bias", "class", "config", "cutoffs", "def", "dim", "div_value", "else", "emb_dim", "head_bias", "if", "in_features", "is", "n_classes", "n_words", "nn", "pad_index", "proj", "self", "super"], "flaubert/modeling_flaubert.py:FlaubertPredLayer.forward": ["False", "ModelPredLayer", "Module", "None", "_", "asm", "class", "cross_entropy", "def", "else", "forward", "functional", "if", "is", "log_prob", "loss", "mean", "n_words", "nn", "not", "outputs", "proj", "reduction", "return", "scores", "self", "view", "x", "y"], "flaubert/modeling_flaubert.py:FlaubertPoolerStartLogits.__init__": ["Linear", "ModelConfig", "ModelPoolerStartLogits", "Module", "__init__", "class", "config", "def", "dense", "hidden_size", "nn", "self", "super"], "flaubert/modeling_flaubert.py:FlaubertPoolerStartLogits.forward": ["FloatTensor", "ModelPoolerStartLogits", "Module", "None", "class", "def", "dense", "dtype", "else", "float16", "forward", "hidden_states", "if", "is", "nn", "not", "p_mask", "return", "self", "squeeze", "torch", "x"], "flaubert/modeling_flaubert.py:FlaubertPoolerEndLogits.__init__": ["LayerNorm", "Linear", "ModelConfig", "ModelPoolerEndLogits", "Module", "Tanh", "__init__", "activation", "class", "config", "def", "dense_0", "dense_1", "eps", "hidden_size", "layer_norm_eps", "nn", "self", "super"], "flaubert/modeling_flaubert.py:FlaubertPoolerEndLogits.forward": ["FloatTensor", "LayerNorm", "LongTensor", "ModelPoolerEndLogits", "Module", "None", "One", "activation", "assert", "be", "cat", "class", "def", "dense_0", "dense_1", "dim", "dtype", "else", "expand", "float16", "forward", "gather", "hidden_states", "hsz", "if", "is", "nn", "not", "of", "or", "p_mask", "return", "self", "shape", "should", "slen", "squeeze", "start_positions", "start_states", "torch", "x"], "flaubert/modeling_flaubert.py:FlaubertPoolerAnswerClass.__init__": ["False", "Linear", "ModelConfig", "ModelPoolerAnswerClass", "Module", "Tanh", "__init__", "activation", "bias", "class", "config", "def", "dense_0", "dense_1", "hidden_size", "nn", "self", "super"], "flaubert/modeling_flaubert.py:FlaubertPoolerAnswerClass.forward": ["FloatTensor", "LongTensor", "ModelPoolerAnswerClass", "Module", "None", "One", "activation", "assert", "be", "cat", "class", "cls_index", "cls_token_state", "def", "dense_0", "dense_1", "dim", "else", "expand", "forward", "gather", "hidden_states", "hsz", "if", "is", "nn", "not", "of", "or", "return", "self", "shape", "should", "squeeze", "start_positions", "start_states", "torch", "x"], "flaubert/modeling_flaubert.py:FlaubertSQuADHead.__init__": ["ModelConfig", "ModelPoolerAnswerClass", "ModelPoolerEndLogits", "ModelPoolerStartLogits", "ModelSQuADHead", "Module", "__init__", "answer_class", "class", "config", "def", "end_logits", "end_n_top", "nn", "self", "start_logits", "start_n_top", "super"], "flaubert/modeling_flaubert.py:FlaubertSQuADHead.forward": ["BCEWithLogitsLoss", "CrossEntropyLoss", "False", "FloatTensor", "LongTensor", "ModelSQuADHead", "ModelSquadHeadOutput", "Module", "None", "and", "answer_class", "bh", "bl", "blh", "bool", "bsz", "class", "cls_index", "cls_logits", "cls_loss", "def", "dim", "einsum", "else", "end_log_probs", "end_logits", "end_loss", "end_n_top", "end_positions", "end_top_index", "end_top_log_probs", "expand", "expand_as", "for", "forward", "functional", "gather", "hidden_states", "hidden_states_expanded", "hsz", "if", "in", "is", "is_impossible", "loss", "loss_fct", "loss_fct_cls", "nn", "not", "p_mask", "r", "return", "return_dict", "self", "size", "slen", "softmax", "squeeze_", "start_log_probs", "start_logits", "start_loss", "start_n_top", "start_positions", "start_states", "start_top_index", "start_top_index_exp", "start_top_log_probs", "topk", "torch", "total_loss", "tuple", "unsqueeze", "view", "x"], "flaubert/modeling_flaubert.py:FlaubertSequenceSummary.__init__": ["Callable", "Dropout", "Identity", "Linear", "ModelConfig", "ModelSequenceSummary", "Module", "None", "NotImplementedError", "__init__", "activation", "activation_string", "and", "attn", "class", "config", "def", "else", "first_dropout", "get_activation", "getattr", "hasattr", "hidden_size", "if", "last", "last_dropout", "nn", "num_classes", "num_labels", "raise", "self", "summary", "summary_activation", "summary_first_dropout", "summary_last_dropout", "summary_proj_to_labels", "summary_type", "summary_use_proj", "super"], "flaubert/modeling_flaubert.py:FlaubertSequenceSummary.forward": ["FloatTensor", "LongTensor", "ModelSequenceSummary", "Module", "None", "NotImplementedError", "activation", "attn", "class", "cls_index", "def", "dim", "dtype", "elif", "else", "expand", "first", "first_dropout", "forward", "full_like", "gather", "hidden_states", "if", "is", "last", "last_dropout", "long", "mean", "nn", "output", "raise", "return", "self", "shape", "size", "squeeze", "summary", "summary_type", "torch", "unsqueeze"], "flaubert/modeling_flaubert.py:FlaubertPreTrainedModel.dummy_inputs": ["ModelPreTrainedModel", "None", "PreTrainedModel", "and", "attention_mask", "attns_list", "class", "config", "def", "dummy_inputs", "else", "if", "input_ids", "inputs_list", "langs", "langs_list", "n_langs", "return", "self", "tensor", "torch", "use_lang_emb"], "flaubert/modeling_flaubert.py:FlaubertPreTrainedModel._init_weights": ["Embedding", "False", "LayerNorm", "Linear", "ModelModel", "ModelPreTrainedModel", "None", "PreTrainedModel", "_init_weights", "_is_hf_initialized", "and", "arange", "bias", "class", "config", "constant_", "copy_", "create_sinusoidal_embeddings", "def", "emb_dim", "embed_init_std", "empty_like", "expand", "getattr", "if", "init", "init_std", "is", "isinstance", "max_position_embeddings", "mean", "module", "nn", "normal_", "not", "ones_", "out", "padding_idx", "position_embeddings", "position_ids", "self", "shape", "sinusoidal_embeddings", "std", "torch", "weight", "zeros_"], "flaubert/modeling_flaubert.py:FlaubertModel.__init__": ["Currently", "Embedding", "False", "LayerNorm", "Model", "ModelModel", "ModelPreTrainedModel", "ModuleList", "MultiHeadAttention", "NotImplementedError", "TransformerFFN", "__init__", "a", "an", "and", "append", "arange", "as", "assert", "attention_dropout", "attentions", "be", "can", "causal", "class", "config", "def", "dim", "dropout", "emb_dim", "embeddings", "encoder", "eos_index", "eps", "expand", "ffns", "for", "getattr", "hidden_dim", "i", "if", "in", "is_decoder", "is_encoder", "lang_embeddings", "layer_idx", "layer_norm1", "layer_norm2", "layer_norm_emb", "layer_norm_eps", "layerdrop", "max_position_embeddings", "multiple", "must", "n_heads", "n_langs", "n_layers", "n_words", "nn", "not", "of", "only", "pad_index", "padding_idx", "persistent", "position_embeddings", "position_ids", "post_init", "pre_norm", "raise", "range", "register_buffer", "self", "super", "torch", "transformer", "use_lang_emb", "used"], "flaubert/modeling_flaubert.py:FlaubertModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "return", "self"], "flaubert/modeling_flaubert.py:FlaubertModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "new_embeddings", "self", "set_input_embeddings"], "flaubert/modeling_flaubert.py:FlaubertModel.forward": ["BaseModelOutput", "DynamicCache", "EncoderDecoderCache", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "_slen", "and", "arange", "assert", "attention_mask", "attentions", "attn", "attn_mask", "attn_outputs", "bool", "bs", "cache", "cache_position", "causal", "class", "config", "continue", "def", "device", "dict", "dim", "dropout", "dropout_probability", "dtype", "else", "embeddings", "expand", "expand_as", "ffns", "for", "forward", "functional", "get_masks", "get_seq_length", "hasattr", "hidden_states", "i", "if", "in", "input_ids", "inputs_embeds", "is", "item", "kwargs", "lang_embeddings", "langs", "last_hidden_state", "layer_norm1", "layer_norm2", "layer_norm_emb", "layerdrop", "lengths", "long", "mask", "max", "n_langs", "n_layers", "nn", "not", "output_attentions", "output_hidden_states", "p", "pad_index", "padding_mask", "position_embeddings", "position_ids", "pre_norm", "r", "rand", "range", "return", "return_dict", "self", "size", "slen", "str", "sum", "tensor", "tensor_normalized", "to", "token_type_ids", "torch", "training", "tuple", "unsqueeze", "use_lang_emb", "use_return_dict", "v"], "flaubert/modeling_flaubert.py:FlaubertWithLMHeadModel.__init__": ["GenerationMixin", "ModelModel", "ModelPreTrainedModel", "ModelPredLayer", "ModelWithLMHeadModel", "__init__", "class", "config", "def", "post_init", "pred_layer", "self", "super", "transformer"], "flaubert/modeling_flaubert.py:FlaubertWithLMHeadModel.get_output_embeddings": ["GenerationMixin", "ModelPreTrainedModel", "ModelWithLMHeadModel", "class", "def", "get_output_embeddings", "pred_layer", "proj", "return", "self"], "flaubert/modeling_flaubert.py:FlaubertWithLMHeadModel.set_output_embeddings": ["GenerationMixin", "ModelPreTrainedModel", "ModelWithLMHeadModel", "class", "def", "new_embeddings", "pred_layer", "proj", "self", "set_output_embeddings"], "flaubert/modeling_flaubert.py:FlaubertWithLMHeadModel.prepare_inputs_for_generation": ["GenerationMixin", "ModelPreTrainedModel", "ModelWithLMHeadModel", "None", "cat", "class", "config", "def", "device", "dim", "dtype", "effective_batch_size", "else", "full", "full_like", "if", "input_ids", "is", "kwargs", "lang_id", "langs", "long", "mask_token", "mask_token_id", "not", "prepare_inputs_for_generation", "return", "self", "shape", "torch"], "flaubert/modeling_flaubert.py:FlaubertWithLMHeadModel.forward": ["GenerationMixin", "MaskedLMOutput", "ModelPreTrainedModel", "ModelWithLMHeadModel", "None", "Tensor", "attention_mask", "attentions", "bool", "cache", "class", "config", "def", "dict", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "langs", "lengths", "logits", "loss", "not", "output", "output_attentions", "output_hidden_states", "outputs", "position_ids", "pred_layer", "r", "return", "return_dict", "self", "str", "token_type_ids", "torch", "transformer", "transformer_outputs", "tuple", "use_return_dict"], "flaubert/modeling_flaubert.py:FlaubertForSequenceClassification.__init__": ["ModelForSequenceClassification", "ModelModel", "ModelPreTrainedModel", "ModelSequenceSummary", "__init__", "class", "config", "def", "num_labels", "post_init", "self", "sequence_summary", "super", "transformer"], "flaubert/modeling_flaubert.py:FlaubertForSequenceClassification.forward": ["BCEWithLogitsLoss", "CrossEntropyLoss", "MSELoss", "ModelForSequenceClassification", "ModelPreTrainedModel", "None", "SequenceClassifierOutput", "Tensor", "and", "attention_mask", "attentions", "bool", "cache", "class", "config", "def", "dict", "dtype", "elif", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "kwargs", "labels", "langs", "lengths", "logits", "long", "loss", "loss_fct", "multi_label_classification", "not", "num_labels", "or", "output", "output_attentions", "output_hidden_states", "position_ids", "problem_type", "r", "regression", "return", "return_dict", "self", "sequence_summary", "single_label_classification", "squeeze", "str", "token_type_ids", "torch", "transformer", "transformer_outputs", "tuple", "use_return_dict", "view"], "flaubert/modeling_flaubert.py:FlaubertForTokenClassification.__init__": ["Dropout", "Linear", "ModelForTokenClassification", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "classifier", "config", "def", "dropout", "hidden_size", "nn", "num_labels", "post_init", "self", "super", "transformer"], "flaubert/modeling_flaubert.py:FlaubertForTokenClassification.forward": ["CrossEntropyLoss", "ModelForTokenClassification", "ModelPreTrainedModel", "None", "Tensor", "TokenClassifierOutput", "attention_mask", "attentions", "bool", "cache", "class", "classifier", "config", "def", "dict", "dropout", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "langs", "lengths", "logits", "loss", "loss_fct", "not", "num_labels", "output", "output_attentions", "output_hidden_states", "outputs", "position_ids", "r", "return", "return_dict", "self", "sequence_output", "str", "token_type_ids", "torch", "transformer", "tuple", "use_return_dict", "view"], "flaubert/modeling_flaubert.py:FlaubertForQuestionAnsweringSimple.__init__": ["Linear", "ModelForQuestionAnsweringSimple", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "config", "def", "hidden_size", "nn", "num_labels", "post_init", "qa_outputs", "self", "super", "transformer"], "flaubert/modeling_flaubert.py:FlaubertForQuestionAnsweringSimple.forward": ["CrossEntropyLoss", "ModelForQuestionAnsweringSimple", "ModelPreTrainedModel", "None", "QuestionAnsweringModelOutput", "Tensor", "and", "attention_mask", "attentions", "bool", "cache", "clamp", "class", "config", "contiguous", "def", "dict", "dim", "else", "end_logits", "end_loss", "end_positions", "forward", "hidden_states", "if", "ignore_index", "ignored_index", "input_ids", "inputs_embeds", "is", "kwargs", "langs", "len", "lengths", "logits", "loss", "loss_fct", "not", "output", "output_attentions", "output_hidden_states", "position_ids", "qa_outputs", "r", "return", "return_dict", "self", "sequence_output", "size", "split", "squeeze", "start_logits", "start_loss", "start_positions", "str", "token_type_ids", "torch", "total_loss", "transformer", "transformer_outputs", "tuple", "use_return_dict"], "flaubert/modeling_flaubert.py:FlaubertForQuestionAnswering.__init__": ["ModelForQuestionAnswering", "ModelModel", "ModelPreTrainedModel", "ModelSQuADHead", "__init__", "class", "config", "def", "post_init", "qa_outputs", "self", "super", "transformer"], "flaubert/modeling_flaubert.py:FlaubertForQuestionAnswering.forward": ["ModelForQuestionAnswering", "ModelForQuestionAnsweringOutput", "ModelPreTrainedModel", "None", "Tensor", "attention_mask", "attentions", "bool", "cache", "class", "cls_index", "cls_logits", "config", "def", "dict", "else", "end_positions", "end_top_index", "end_top_log_probs", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "is_impossible", "kwargs", "langs", "lengths", "loss", "not", "output", "output_attentions", "output_hidden_states", "outputs", "p_mask", "position_ids", "qa_outputs", "r", "return", "return_dict", "self", "start_positions", "start_top_index", "start_top_log_probs", "str", "token_type_ids", "torch", "transformer", "transformer_outputs", "tuple", "use_return_dict"], "flaubert/modeling_flaubert.py:FlaubertForMultipleChoice.__init__": ["Linear", "ModelForMultipleChoice", "ModelModel", "ModelPreTrainedModel", "ModelSequenceSummary", "__init__", "class", "config", "def", "inputs", "kwargs", "logits_proj", "nn", "num_labels", "post_init", "self", "sequence_summary", "super", "transformer"], "flaubert/modeling_flaubert.py:FlaubertForMultipleChoice.forward": ["CrossEntropyLoss", "Model", "ModelForMultipleChoice", "ModelPreTrainedModel", "MultipleChoiceModelOutput", "None", "Please", "Tensor", "The", "attention", "attention_mask", "attentions", "be", "bool", "cache", "cannot", "choice", "class", "config", "def", "dict", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "instead", "is", "kwargs", "labels", "langs", "lengths", "logger", "logits", "logits_proj", "loss", "loss_fct", "mask", "models", "multiple", "not", "num_choices", "output", "output_attentions", "output_hidden_states", "parameter", "position_ids", "r", "reshaped_logits", "return", "return_dict", "self", "sequence_summary", "shape", "size", "str", "the", "token_type_ids", "torch", "transformer", "transformer_outputs", "tuple", "use", "use_return_dict", "used", "view", "warning", "with"], "flava/modeling_flava.py:FlavaModelOutput.to_tuple": ["Any", "ModelModelOutput", "ModelOutput", "class", "def", "else", "for", "getattr", "if", "image_output", "in", "k", "keys", "multimodal_output", "not", "return", "self", "text_output", "to_tuple", "tuple"], "flava/modeling_flava.py:FlavaLosses.all_none": ["False", "ModelLosses", "ModelOutput", "None", "True", "all_none", "bool", "break", "class", "def", "for", "if", "in", "is", "not", "return", "self", "v", "values"], "flava/modeling_flava.py:FlavaForPreTrainingOutput.to_tuple": ["Any", "ModelForPreTrainingOutput", "ModelOutput", "class", "def", "else", "for", "getattr", "if", "image_masked_output", "image_output", "in", "k", "keys", "multimodal_masked_output", "multimodal_output", "not", "return", "self", "text_masked_output", "text_output", "to_tuple", "transformer_outputs", "tuple"], "flava/modeling_flava.py:FlavaImageEmbeddings.__init__": ["Dropout", "False", "ModelImageConfig", "ModelImageEmbeddings", "Module", "None", "Parameter", "PatchEmbeddings", "__init__", "bool", "class", "cls_token", "config", "def", "dropout", "else", "embed_dim", "hidden_dropout_prob", "hidden_size", "if", "image_size", "mask_token", "nn", "num_channels", "num_patches", "or", "patch_embeddings", "patch_size", "position_embeddings", "self", "super", "torch", "use_mask_token", "zeros"], "flava/modeling_flava.py:FlavaImageEmbeddings.interpolate_pos_encoding": ["False", "ModelImageEmbeddings", "Module", "Tensor", "align_corners", "and", "bicubic", "cat", "class", "class_pos_embed", "def", "dim", "embeddings", "functional", "height", "if", "int", "interpolate", "interpolate_pos_encoding", "is_tracing", "jit", "mode", "new_height", "new_width", "nn", "not", "num_patches", "num_positions", "patch_pos_embed", "patch_size", "permute", "position_embeddings", "reshape", "return", "self", "shape", "size", "sqrt_num_positions", "torch", "torch_int", "view", "width"], "flava/modeling_flava.py:FlavaImageEmbeddings.forward": ["BoolTensor", "False", "ModelImageEmbeddings", "Module", "None", "Tensor", "_", "batch_size", "bool", "bool_masked_pos", "cat", "class", "cls_token", "cls_tokens", "def", "dim", "dropout", "else", "embeddings", "expand", "forward", "height", "if", "interpolate_pos_encoding", "is", "mask", "mask_token", "mask_tokens", "nn", "not", "num_channels", "patch_embeddings", "pixel_values", "position_embeddings", "return", "self", "seq_len", "shape", "size", "torch", "type_as", "unsqueeze", "view", "width"], "flava/modeling_flava.py:PatchEmbeddings.__init__": ["Conv2d", "Iterable", "ModelEmbeddings", "Model_size", "Module", "__init__", "abc", "class", "collections", "def", "embed_dim", "if", "image_size", "int", "isinstance", "kernel_size", "nn", "not", "num_Modeles", "num_channels", "projection", "self", "stride", "super", "tuple"], "flava/modeling_flava.py:PatchEmbeddings.forward": ["False", "Input", "ModelEmbeddings", "Module", "Tensor", "ValueError", "batch_size", "bool", "class", "def", "doesn", "f", "flatten", "forward", "height", "if", "image", "image_size", "interpolate_pos_encoding", "match", "model", "nn", "not", "num_channels", "or", "pixel_values", "projection", "raise", "return", "self", "shape", "size", "t", "torch", "transpose", "width", "x"], "flava/modeling_flava.py:FlavaTextEmbeddings.__init__": ["Dropout", "Embedding", "False", "LayerNorm", "ModelTextEmbeddings", "Module", "__init__", "arange", "class", "config", "def", "dropout", "dtype", "eps", "expand", "hidden_dropout_prob", "hidden_size", "layer_norm_eps", "long", "max_position_embeddings", "nn", "pad_token_id", "padding_idx", "persistent", "position_embeddings", "position_ids", "register_buffer", "self", "size", "super", "token_type_embeddings", "token_type_ids", "torch", "type_vocab_size", "vocab_size", "word_embeddings", "zeros"], "flava/modeling_flava.py:FlavaTextEmbeddings.forward": ["LayerNorm", "ModelTextEmbeddings", "Module", "None", "Tensor", "buffered_token_type_ids", "buffered_token_type_ids_expanded", "class", "def", "device", "dropout", "dtype", "else", "embeddings", "expand", "forward", "hasattr", "if", "input_ids", "input_shape", "inputs_embeds", "is", "long", "nn", "position_embeddings", "position_ids", "return", "self", "seq_length", "size", "token_type_embeddings", "token_type_ids", "torch", "word_embeddings", "zeros"], "flava/modeling_flava.py:FlavaSelfAttention.__init__": ["Dropout", "Linear", "ModelPossibleConfigs", "ModelSelfAttention", "Module", "None", "The", "ValueError", "__init__", "a", "all_head_size", "and", "attention", "attention_head_size", "attention_probs_dropout_prob", "bias", "class", "config", "def", "dropout", "embedding_size", "f", "hasattr", "heads", "hidden", "hidden_size", "if", "int", "is", "key", "multiple", "nn", "not", "num_attention_heads", "number", "of", "qkv_bias", "query", "raise", "self", "size", "super", "the", "value"], "flava/modeling_flava.py:FlavaSelfAttention.forward": ["False", "ModelSelfAttention", "Module", "None", "Tensor", "_", "all_head_size", "attention_head_size", "attention_mask", "attention_probs", "attention_scores", "batch_size", "bool", "class", "context_layer", "contiguous", "def", "dim", "dropout", "else", "forward", "functional", "hidden_states", "if", "is", "key", "key_layer", "math", "matmul", "new_context_layer_shape", "nn", "not", "num_attention_heads", "output_attentions", "outputs", "permute", "query", "query_layer", "return", "self", "seq_length", "shape", "size", "softmax", "sqrt", "torch", "transpose", "tuple", "value", "value_layer", "view"], "flava/modeling_flava.py:FlavaSelfOutput.__init__": ["Dropout", "Linear", "ModelPossibleConfigs", "ModelSelfOutput", "Module", "None", "__init__", "class", "config", "def", "dense", "dropout", "hidden_dropout_prob", "hidden_size", "nn", "self", "super"], "flava/modeling_flava.py:FlavaSelfOutput.forward": ["ModelSelfOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "flava/modeling_flava.py:FlavaAttention.__init__": ["ModelAttention", "ModelPossibleConfigs", "ModelSelfAttention", "ModelSelfOutput", "Module", "None", "__init__", "attention", "class", "config", "def", "nn", "output", "self", "super"], "flava/modeling_flava.py:FlavaAttention.forward": ["False", "ModelAttention", "Module", "None", "Tensor", "attention", "attention_mask", "attention_output", "bool", "class", "def", "forward", "hidden_states", "nn", "output", "output_attentions", "outputs", "return", "self", "self_outputs", "torch", "tuple"], "flava/modeling_flava.py:FlavaIntermediate.__init__": ["ACT2FN", "Linear", "ModelIntermediate", "ModelPossibleConfigs", "Module", "None", "__init__", "class", "config", "def", "dense", "else", "hidden_act", "hidden_size", "if", "intermediate_act_fn", "intermediate_size", "isinstance", "nn", "self", "str", "super"], "flava/modeling_flava.py:FlavaIntermediate.forward": ["ModelIntermediate", "Module", "Tensor", "class", "def", "dense", "forward", "hidden_states", "intermediate_act_fn", "nn", "return", "self", "torch"], "flava/modeling_flava.py:FlavaOutput.__init__": ["Dropout", "Linear", "ModelOutput", "ModelPossibleConfigs", "Module", "None", "__init__", "class", "config", "def", "dense", "dropout", "hidden_dropout_prob", "hidden_size", "intermediate_size", "nn", "self", "super"], "flava/modeling_flava.py:FlavaOutput.forward": ["ModelOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "flava/modeling_flava.py:FlavaLayer.__init__": ["GradientCheckpointingLayer", "LayerNorm", "ModelAttention", "ModelIntermediate", "ModelLayer", "ModelOutput", "ModelPossibleConfigs", "None", "__init__", "attention", "chunk_size_feed_forward", "class", "config", "def", "eps", "hidden_size", "intermediate", "layer_norm_eps", "layernorm_after", "layernorm_before", "nn", "output", "self", "seq_len_dim", "super"], "flava/modeling_flava.py:FlavaLayer.forward": ["False", "GradientCheckpointingLayer", "ModelLayer", "None", "Tensor", "attention", "attention_mask", "attention_output", "bool", "class", "def", "forward", "hidden_states", "intermediate", "layer_output", "layernorm_after", "layernorm_before", "output", "output_attentions", "outputs", "return", "self", "self_attention_outputs", "torch", "tuple"], "flava/modeling_flava.py:FlavaEncoder.__init__": ["False", "ModelConfig", "ModelEncoder", "ModelLayer", "Module", "ModuleList", "None", "_", "__init__", "class", "config", "def", "for", "gradient_checkpointing", "in", "layer", "nn", "num_hidden_layers", "range", "self", "super"], "flava/modeling_flava.py:FlavaEncoder.forward": ["BaseModelOutput", "False", "ModelEncoder", "Module", "None", "Tensor", "True", "all_hidden_states", "all_self_attentions", "attention_mask", "attentions", "bool", "class", "def", "else", "enumerate", "for", "forward", "hidden_states", "i", "if", "in", "is", "last_hidden_state", "layer", "layer_module", "layer_outputs", "nn", "not", "output_attentions", "output_hidden_states", "return", "return_dict", "self", "torch", "tuple", "v"], "flava/modeling_flava.py:FlavaPooler.__init__": ["Linear", "ModelPooler", "ModelPossibleConfigs", "Module", "Tanh", "__init__", "activation", "class", "config", "def", "dense", "hidden_size", "nn", "self", "super"], "flava/modeling_flava.py:FlavaPooler.forward": ["ModelPooler", "Module", "Tensor", "activation", "class", "def", "dense", "first_token_tensor", "forward", "hidden_states", "nn", "pooled_output", "return", "self", "torch"], "flava/modeling_flava.py:FlavaPreTrainedModel._init_weights": ["Conv2d", "LayerNorm", "Linear", "ModelImageEmbeddings", "ModelMaskedPredictionHead", "ModelModel", "ModelMultimodalModel", "ModelPreTrainedModel", "ModelTextEmbeddings", "None", "PreTrainedModel", "_init_weights", "arange", "bias", "class", "cls_token", "config", "constant_", "copy_", "def", "elif", "expand", "if", "init", "is", "isinstance", "logit_scale", "logit_scale_init_value", "mask_token", "module", "nn", "not", "position_embeddings", "position_ids", "self", "shape", "super", "token_type_ids", "torch", "use_cls_token", "zeros_"], "flava/modeling_flava.py:FlavaImageModel.__init__": ["LayerNorm", "ModelEncoder", "ModelImageConfig", "ModelImageEmbeddings", "ModelImageModel", "ModelPooler", "ModelPreTrainedModel", "None", "True", "__init__", "add_pooling_layer", "bool", "class", "config", "def", "else", "embeddings", "encoder", "eps", "hidden_size", "if", "layer_norm_eps", "layernorm", "nn", "pooler", "post_init", "r", "self", "super"], "flava/modeling_flava.py:FlavaImageModel.get_input_embeddings": ["ModelImageModel", "ModelPreTrainedModel", "Module", "class", "def", "embeddings", "get_input_embeddings", "nn", "patch_embeddings", "return", "self"], "flava/modeling_flava.py:FlavaImageModel.set_input_embeddings": ["ModelImageModel", "ModelPreTrainedModel", "Module", "class", "def", "embeddings", "nn", "patch_embeddings", "self", "set_input_embeddings", "value"], "flava/modeling_flava.py:FlavaImageModel.forward": ["BaseModelOutputWithPooling", "BoolTensor", "ModelImageModel", "ModelPreTrainedModel", "None", "Tensor", "ValueError", "You", "attention_mask", "attentions", "bool", "bool_masked_pos", "class", "config", "def", "else", "embedding_output", "embeddings", "encoder", "encoder_outputs", "forward", "have", "hidden_states", "if", "interpolate_pos_encoding", "is", "kwargs", "last_hidden_state", "layernorm", "not", "output_attentions", "output_hidden_states", "pixel_values", "pooled_output", "pooler", "pooler_output", "r", "raise", "return", "return_dict", "self", "sequence_output", "specify", "to", "torch", "tuple", "use_return_dict"], "flava/modeling_flava.py:FlavaTextModel.__init__": ["LayerNorm", "ModelEncoder", "ModelPooler", "ModelPreTrainedModel", "ModelTextConfig", "ModelTextEmbeddings", "ModelTextModel", "None", "True", "__init__", "add_pooling_layer", "bool", "class", "config", "def", "else", "embeddings", "encoder", "eps", "hidden_size", "if", "layer_norm_eps", "layernorm", "nn", "pooler", "post_init", "r", "self", "super"], "flava/modeling_flava.py:FlavaTextModel.get_input_embeddings": ["ModelPreTrainedModel", "ModelTextModel", "PatchEmbeddings", "class", "def", "embeddings", "get_input_embeddings", "return", "self", "word_embeddings"], "flava/modeling_flava.py:FlavaTextModel.set_input_embeddings": ["ModelPreTrainedModel", "ModelTextModel", "Module", "class", "def", "embeddings", "nn", "self", "set_input_embeddings", "value", "word_embeddings"], "flava/modeling_flava.py:FlavaTextModel.forward": ["BaseModelOutputWithPooling", "ModelPreTrainedModel", "ModelTextModel", "None", "Tensor", "ValueError", "You", "attention_mask", "attentions", "bool", "class", "config", "def", "device", "else", "embedding_output", "embeddings", "encoder", "encoder_outputs", "extended_attention_mask", "forward", "get_extended_attention_mask", "have", "hidden_states", "if", "input_ids", "input_shape", "is", "kwargs", "last_hidden_state", "layernorm", "not", "ones", "output_attentions", "output_hidden_states", "pooled_output", "pooler", "pooler_output", "position_ids", "r", "raise", "return", "return_dict", "self", "sequence_output", "size", "specify", "to", "token_type_ids", "torch", "tuple", "use_return_dict"], "flava/modeling_flava.py:FlavaMultimodalModel.__init__": ["LayerNorm", "ModelEncoder", "ModelMultimodalConfig", "ModelMultimodalModel", "ModelPooler", "ModelPreTrainedModel", "None", "Parameter", "True", "__init__", "add_pooling_layer", "class", "cls_token", "config", "def", "else", "encoder", "eps", "hidden_size", "if", "layer_norm_eps", "layernorm", "nn", "pooler", "post_init", "r", "self", "super", "torch", "use_cls_token", "zeros"], "flava/modeling_flava.py:FlavaMultimodalModel.forward": ["BaseModelOutputWithPooling", "ModelMultimodalModel", "ModelPreTrainedModel", "None", "Tensor", "_", "attention_mask", "attentions", "batch_size", "bool", "cat", "class", "cls_token", "cls_tokens", "config", "def", "device", "dim", "else", "encoder", "encoder_outputs", "expand", "extended_attention_mask", "forward", "get_extended_attention_mask", "hidden_states", "if", "is", "kwargs", "last_hidden_state", "layernorm", "not", "ones", "output_attentions", "output_hidden_states", "pooled_output", "pooler", "pooler_output", "r", "return", "return_dict", "self", "seq_length", "sequence_output", "size", "torch", "tuple", "use_cls_token", "use_return_dict"], "flava/modeling_flava.py:FlavaModel.__init__": ["Linear", "ModelConfig", "ModelImageConfig", "ModelImageModel", "ModelModel", "ModelMultimodalConfig", "ModelMultimodalModel", "ModelPreTrainedModel", "ModelTextConfig", "ModelTextModel", "Parameter", "TypeError", "__init__", "be", "but", "class", "config", "def", "expected", "f", "hidden_size", "if", "image_config", "image_hidden_size", "image_model", "image_projection", "image_to_mm_projection", "is", "isinstance", "logit_scale", "logit_scale_init_value", "mm_hidden_size", "multimodal_config", "multimodal_model", "nn", "not", "of", "post_init", "projection_dim", "raise", "self", "super", "tensor", "text_config", "text_hidden_size", "text_model", "text_projection", "text_to_mm_projection", "to", "torch", "type"], "flava/modeling_flava.py:FlavaModel.get_text_features": ["BaseModelOutputWithPooling", "FloatTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "attention_mask", "class", "def", "get_text_features", "input_ids", "last_hidden_state", "pooled_output", "position_ids", "r", "return", "self", "text_features", "text_model", "text_outputs", "text_projection", "token_type_ids", "torch"], "flava/modeling_flava.py:FlavaModel.get_image_features": ["BaseModelOutputWithPooling", "BoolTensor", "FloatTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "attention_mask", "bool", "bool_masked_pos", "class", "def", "get_image_features", "image_features", "image_model", "image_outputs", "image_projection", "interpolate_pos_encoding", "last_hidden_state", "pixel_values", "pooled_output", "r", "return", "self", "torch"], "flava/modeling_flava.py:FlavaModel.forward": ["FloatTensor", "LongTensor", "Model", "ModelModel", "ModelModelOutput", "ModelPreTrainedModel", "None", "Please", "Tensor", "True", "ValueError", "_", "and", "attention_mask", "attention_mask_image", "attention_multimodal", "batch_size", "bool", "bool_masked_pos", "cat", "class", "config", "def", "device", "dim", "else", "forward", "hidden", "if", "image_attention_mask", "image_embeddings", "image_mm_projection", "image_model", "image_output", "image_states", "image_to_mm_projection", "input_ids", "is", "kwargs", "model", "multimodal_embeddings", "multimodal_input", "multimodal_model", "multimodal_output", "not", "ones", "output_attentions", "output_hidden_states", "pixel_values", "position_ids", "r", "raise", "requires", "return", "return_dict", "self", "seq_len", "set", "shape", "skip_multimodal_encoder", "states", "text_embeddings", "text_mm_projection", "text_model", "text_output", "text_states", "text_to_mm_projection", "to", "token_type_ids", "torch", "tuple", "use_cls_token", "work"], "flava/modeling_flava.py:FlavaImageCodebookResPath.__init__": ["Conv2d", "ModelImageCodebookResPath", "Module", "OrderedDict", "ReLU", "Sequential", "__init__", "class", "conv_1", "conv_2", "conv_3", "conv_4", "def", "hid_size", "in_size", "int", "kernel_size", "kwargs", "nn", "out_size", "padding", "path", "relu_1", "relu_2", "relu_3", "relu_4", "self", "super"], "flava/modeling_flava.py:FlavaImageCodebookResPath.forward": ["ModelImageCodebookResPath", "Module", "Tensor", "class", "def", "forward", "nn", "path", "return", "self", "torch", "x"], "flava/modeling_flava.py:FlavaImageCodebookBlock.__init__": ["Conv2d", "Identity", "ModelImageCodebookBlock", "ModelImageCodebookResPath", "Module", "__init__", "class", "def", "else", "id_path", "if", "in_size", "int", "kernel_size", "kwargs", "nn", "num_layers", "out_size", "padding", "post_gain", "res_path", "self", "super"], "flava/modeling_flava.py:FlavaImageCodebookBlock.forward": ["ModelImageCodebookBlock", "Module", "Tensor", "class", "def", "forward", "id_path", "nn", "post_gain", "res_path", "return", "self", "torch", "x"], "flava/modeling_flava.py:FlavaImageCodebookLayerGroup.__init__": ["MaxPool2d", "ModelImageCodebookBlock", "ModelImageCodebookLayerGroup", "Module", "OrderedDict", "Sequential", "True", "__init__", "block_", "blocks", "bool", "class", "def", "else", "f", "for", "group", "i", "if", "in", "in_size", "int", "kernel_size", "nn", "num_blocks", "num_layers", "out_size", "pool", "range", "self", "super", "use_pool"], "flava/modeling_flava.py:FlavaImageCodebookLayerGroup.forward": ["ModelImageCodebookLayerGroup", "Module", "Tensor", "class", "def", "forward", "group", "nn", "return", "self", "torch", "x"], "flava/modeling_flava.py:FlavaImageCodebook.__init__": ["Any", "Conv2d", "False", "ModelImageCodebook", "ModelImageCodebookConfig", "ModelImageCodebookLayerGroup", "ModelPreTrainedModel", "OrderedDict", "ReLU", "Sequential", "__init__", "blocks", "class", "config", "conv", "def", "for", "freeze", "group_1", "group_2", "group_3", "group_4", "hidden_size", "if", "in", "input", "input_channels", "kernel_size", "kwargs", "nn", "num_blocks_per_group", "num_groups", "num_layers", "output", "output_blocks", "padding", "param", "parameters", "post_init", "relu", "requires_grad", "self", "super", "use_pool", "vocab_size"], "flava/modeling_flava.py:FlavaImageCodebook.get_codebook_indices": ["ModelImageCodebook", "ModelPreTrainedModel", "Tensor", "argmax", "axis", "blocks", "class", "def", "f", "get_codebook_indices", "pixel_values", "return", "self", "torch", "z_logits"], "flava/modeling_flava.py:FlavaImageCodebook.get_codebook_probs": ["ModelImageCodebook", "ModelPreTrainedModel", "Softmax", "Tensor", "blocks", "class", "def", "dim", "get_codebook_probs", "nn", "pixel_values", "return", "self", "torch", "z_logits"], "flava/modeling_flava.py:FlavaImageCodebook.forward": ["FloatTensor", "ModelImageCodebook", "ModelPreTrainedModel", "Tensor", "ValueError", "blocks", "built", "but", "channels", "class", "def", "f", "for", "forward", "has", "if", "input", "input_channels", "is", "kwargs", "len", "model", "not", "pixel_values", "raise", "return", "self", "shape", "torch"], "flava/modeling_flava.py:FlavaPredictionHeadTransform.__init__": ["ACT2FN", "LayerNorm", "Linear", "ModelPredictionHeadTransform", "Module", "__init__", "class", "config", "def", "dense", "else", "eps", "hidden_act", "hidden_size", "if", "isinstance", "layer_norm_eps", "nn", "self", "str", "super", "transform_act_fn"], "flava/modeling_flava.py:FlavaPredictionHeadTransform.forward": ["LayerNorm", "ModelPredictionHeadTransform", "Module", "class", "def", "dense", "forward", "hidden_states", "nn", "return", "self", "transform_act_fn"], "flava/modeling_flava.py:FlavaMaskedPredictionHead.__init__": ["Linear", "ModelMaskedPredictionHead", "ModelPredictionHeadTransform", "Module", "None", "Parameter", "True", "__init__", "bias", "class", "config", "decoder", "def", "hidden_size", "if", "is", "nn", "not", "self", "super", "torch", "transform", "vocab_size", "weight", "zeros"], "flava/modeling_flava.py:FlavaMaskedPredictionHead.forward": ["ModelMaskedPredictionHead", "Module", "class", "decoder", "def", "forward", "nn", "return", "self", "transform", "x"], "flava/modeling_flava.py:FlavaITMHead.__init__": ["Linear", "ModelITMHead", "ModelPooler", "Module", "__init__", "class", "config", "def", "hidden_size", "nn", "pooler", "self", "seq_relationship", "super"], "flava/modeling_flava.py:FlavaITMHead.forward": ["ModelITMHead", "Module", "class", "def", "forward", "nn", "pooler", "return", "self", "seq_relationship", "x"], "flava/modeling_flava.py:FlavaGlobalContrastiveHead.__init__": ["ModelGlobalContrastiveHead", "Module", "__init__", "class", "config", "def", "global_backprop_contrastive", "nn", "self", "super"], "flava/modeling_flava.py:FlavaGlobalContrastiveHead.forward": ["ModelGlobalContrastiveHead", "Module", "_", "all_gather", "arange", "cat", "class", "def", "device", "distributed", "else", "exp", "for", "forward", "functional", "get_rank", "get_world_size", "global_backprop_contrastive", "if", "image_embeddings", "image_embeddings_all", "in", "is_available", "is_initialized", "labels", "local_batch_size", "logit_scale", "logits_per_image", "logits_per_text", "matmul", "nn", "not", "or", "range", "return", "self", "size", "temperature", "text_embeddings", "text_embeddings_all", "torch", "transpose", "world_size", "zeros_like"], "flava/modeling_flava.py:FlavaForPreTraining.__init__": ["Model", "ModelConfig", "ModelForPreTraining", "ModelGlobalContrastiveHead", "ModelITMHead", "ModelImageCodebook", "ModelMaskedPredictionHead", "ModelModel", "ModelPreTrainedModel", "Module", "None", "__init__", "and", "ce_ignore_index", "class", "config", "def", "global_contrastive_head", "global_contrastive_weight", "if", "image_codebook", "image_codebook_config", "image_config", "image_vocab_size", "init_codebook", "is", "itm_head", "itm_weight", "mim_head", "mim_weight", "mlm_head", "mlm_weight", "mmm_image_head", "mmm_image_weight", "mmm_text_head", "mmm_text_weight", "nn", "post_init", "r", "self", "skip_unmasked_multimodal_encoder", "super", "text_config", "text_vocab_size", "vocab_size"], "flava/modeling_flava.py:FlavaForPreTraining._resize_to_2d": ["ModelForPreTraining", "ModelPreTrainedModel", "Tensor", "_resize_to_2d", "class", "def", "dim", "if", "return", "self", "size", "torch", "view", "x"], "flava/modeling_flava.py:FlavaForPreTraining.forward": ["AutoProcessor", "Call", "FloatTensor", "LOGIT_SCALE_CLAMP_MAX", "LOGIT_SCALE_CLAMP_MIN", "LongTensor", "MLM", "Model", "ModelForPreTraining", "ModelForPreTrainingOutput", "ModelLosses", "ModelPreTrainedModel", "Model_losses", "Model_masked_output", "Model_output", "None", "OKAY", "Please", "Reinstantiate", "RuntimeError", "Tensor", "This", "True", "ValueError", "_resize_to_2d", "all_none", "and", "any", "are", "attention_mask", "be", "been", "bool", "bool_masked_pos", "but", "calculated", "can", "ce_ignore_index", "clamp_", "class", "codebook", "codebook_pixel_value", "codebook_pixel_values", "config", "contrastive_logits_per_image", "contrastive_logits_per_text", "correctlySetting", "cross_entropy", "custom", "data", "def", "dim", "doing", "else", "end_index", "expected", "for", "forward", "functional", "gc_labels", "gc_loss", "gc_loss_image", "gc_loss_text", "generate", "get_codebook_indices", "global_contrastive", "global_contrastive_head", "global_contrastive_weight", "have", "if", "image", "image_attention_mask", "image_codebook", "image_embedding", "image_embeddings", "image_masked_embeddings", "image_masked_output", "image_output", "image_projection", "image_vocab_size", "in", "inference", "init_codebook", "initialized", "input_ids", "input_ids_masked", "is", "isn", "it", "itm", "itm_head", "itm_labels", "itm_logits", "itm_loss", "itm_weight", "kwargs", "logger", "logit_scale", "logits_per_image", "logits_per_text", "loss", "loss_info", "masked_tokens", "means", "mim", "mim_head", "mim_labels", "mim_labels_filtered", "mim_logits", "mim_loss", "mim_weight", "mlm", "mlm_head", "mlm_labels", "mlm_labels_filtered", "mlm_logits", "mlm_loss", "mlm_weight", "mmm_image", "mmm_image_head", "mmm_image_logits", "mmm_image_loss", "mmm_image_weight", "mmm_text", "mmm_text_head", "mmm_text_logits", "mmm_text_loss", "mmm_text_weight", "model", "multimodal_embeddings", "multimodal_masked_embeddings", "multimodal_masked_output", "multimodal_output", "ne", "new", "nn", "no", "normalize", "not", "on", "or", "output", "output_attentions", "output_hidden_states", "pass", "passed", "pixel_values", "pos_mask", "pos_pairs", "position_ids", "r", "raise", "required", "return", "return_codebook_pixels", "return_dict", "return_loss", "self", "sequence_for_image", "sequence_for_text", "set", "size", "skip_multimodal_encoder", "skip_unmasked_multimodal_encoder", "so", "sum", "t", "text", "text_embedding", "text_embeddings", "text_masked_embeddings", "text_masked_output", "text_output", "text_projection", "text_vocab_size", "that", "the", "this", "to", "to_tuple", "token_type_ids", "torch", "total_loss", "tuple", "unintentional", "unmasked", "use_return_dict", "usually", "values", "view", "warning", "where", "which", "with", "won", "work", "x", "you", "your"], "flex_olmo/modeling_flex_olmo.py:FlexOlmoRMSNorm.__init__": ["ModelRMSNorm", "Module", "Parameter", "__init__", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "flex_olmo/modeling_flex_olmo.py:FlexOlmoRMSNorm.forward": ["ModelRMSNorm", "Module", "True", "class", "def", "dtype", "float32", "forward", "hidden_states", "input_dtype", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "flex_olmo/modeling_flex_olmo.py:FlexOlmoRMSNorm.extra_repr": ["ModelRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "variance_epsilon", "weight"], "flex_olmo/modeling_flex_olmo.py:FlexOlmoRotaryEmbedding.__init__": ["Callable", "False", "ModelConfig", "ModelRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "flex_olmo/modeling_flex_olmo.py:FlexOlmoRotaryEmbedding.compute_default_rope_parameters": ["ModelConfig", "ModelRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "flex_olmo/modeling_flex_olmo.py:FlexOlmoRotaryEmbedding.forward": ["False", "ModelRotaryEmbedding", "Module", "None", "and", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "flex_olmo/modeling_flex_olmo.py:FlexOlmoMLP.__init__": ["ACT2FN", "False", "Linear", "ModelMLP", "Module", "__init__", "act_fn", "bias", "class", "config", "def", "down_proj", "gate_proj", "hidden_act", "hidden_size", "intermediate_size", "nn", "self", "super", "up_proj"], "flex_olmo/modeling_flex_olmo.py:FlexOlmoMLP.forward": ["ModelMLP", "Module", "act_fn", "class", "def", "down_proj", "forward", "gate_proj", "nn", "return", "self", "up_proj", "x"], "flex_olmo/modeling_flex_olmo.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "flex_olmo/modeling_flex_olmo.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value", "value_states"], "flex_olmo/modeling_flex_olmo.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cos", "def", "dtype", "k", "k_embed", "k_type", "q", "q_embed", "q_type", "return", "rotate_half", "sin", "to", "unsqueeze", "unsqueeze_dim"], "flex_olmo/modeling_flex_olmo.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "flex_olmo/modeling_flex_olmo.py:FlexOlmoAttention.__init__": ["Linear", "ModelAttention", "ModelConfig", "ModelRMSNorm", "Module", "None", "True", "__init__", "attention_bias", "attention_dropout", "bias", "class", "config", "def", "getattr", "head_dim", "hidden_size", "int", "is_causal", "k_norm", "k_proj", "layer_idx", "nn", "num_attention_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "q_norm", "q_proj", "rms_norm_eps", "scaling", "self", "super", "v_proj"], "flex_olmo/modeling_flex_olmo.py:FlexOlmoAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "LongTensor", "ModelAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_attn_implementation", "apply_rotary_pos_emb", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "cache_kwargs", "cache_position", "class", "config", "contiguous", "cos", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "head_dim", "hidden_shape", "hidden_states", "if", "input_shape", "is", "k_norm", "k_proj", "key_states", "kwargs", "layer_idx", "nn", "not", "o_proj", "past_key_values", "position_embeddings", "q_norm", "q_proj", "query_states", "reshape", "return", "scaling", "self", "shape", "sin", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "view"], "flex_olmo/modeling_flex_olmo.py:FlexOlmoExperts.__init__": ["ACT2FN", "ModelConfig", "ModelExperts", "Module", "Parameter", "__init__", "act_fn", "class", "config", "def", "down_proj", "empty", "gate_up_proj", "hidden_act", "hidden_dim", "hidden_size", "intermediate_dim", "intermediate_size", "nn", "num_experts", "num_local_experts", "self", "super", "torch"], "flex_olmo/modeling_flex_olmo.py:FlexOlmoExperts.forward": ["ModelExperts", "Module", "None", "Tensor", "act_fn", "chunk", "class", "continue", "current_hidden_states", "current_state", "def", "dim", "down_proj", "dtype", "expert_hit", "expert_idx", "expert_mask", "final_hidden_states", "for", "forward", "functional", "gate", "gate_up_proj", "greater", "hidden_states", "if", "in", "index_add_", "linear", "nn", "no_grad", "nonzero", "num_classes", "num_experts", "one_hot", "permute", "return", "self", "sum", "to", "token_idx", "top_k_index", "top_k_pos", "top_k_weights", "torch", "up", "where", "with", "zeros_like"], "flex_olmo/modeling_flex_olmo.py:FlexOlmoTopKRouter.__init__": ["ModelTopKRouter", "Module", "Parameter", "__init__", "class", "config", "def", "hidden_dim", "hidden_size", "nn", "norm_topk_prob", "num_experts", "num_experts_per_tok", "self", "super", "top_k", "torch", "weight", "zeros"], "flex_olmo/modeling_flex_olmo.py:FlexOlmoTopKRouter.forward": ["F", "ModelTopKRouter", "Module", "True", "class", "def", "dim", "dtype", "float", "forward", "functional", "hidden_dim", "hidden_states", "if", "keepdim", "linear", "nn", "norm_topk_prob", "reshape", "return", "router_indices", "router_logits", "router_scores", "router_top_value", "self", "softmax", "sum", "to", "top_k", "topk", "torch", "weight"], "flex_olmo/modeling_flex_olmo.py:FlexOlmoSparseMoeBlock.__init__": ["ModelExperts", "ModelSparseMoeBlock", "ModelTopKRouter", "Module", "__init__", "class", "config", "def", "experts", "gate", "nn", "self", "super"], "flex_olmo/modeling_flex_olmo.py:FlexOlmoSparseMoeBlock.forward": ["ModelSparseMoeBlock", "Module", "Tensor", "_", "batch_size", "class", "def", "experts", "final_hidden_states", "forward", "gate", "hidden_dim", "hidden_states", "nn", "reshape", "return", "self", "sequence_length", "shape", "top_k_index", "top_k_weights", "torch", "view"], "flex_olmo/modeling_flex_olmo.py:FlexOlmoDecoderLayer.__init__": ["GradientCheckpointingLayer", "ModelAttention", "ModelConfig", "ModelDecoderLayer", "ModelRMSNorm", "ModelSparseMoeBlock", "__init__", "class", "config", "def", "eps", "hidden_size", "int", "layer_idx", "mlp", "post_attention_layernorm", "post_feedforward_layernorm", "rms_norm_eps", "self", "self_attn", "super"], "flex_olmo/modeling_flex_olmo.py:FlexOlmoDecoderLayer.forward": ["Cache", "FloatTensor", "GradientCheckpointingLayer", "LongTensor", "ModelDecoderLayer", "None", "Tensor", "_", "attention_mask", "cache_position", "class", "def", "forward", "hidden_states", "kwargs", "mlp", "past_key_values", "position_embeddings", "position_ids", "post_attention_layernorm", "post_feedforward_layernorm", "residual", "return", "self", "self_attn", "torch", "tuple"], "flex_olmo/modeling_flex_olmo.py:FlexOlmoPreTrainedModel._init_weights": ["ModelExperts", "ModelPreTrainedModel", "ModelTopKRouter", "PreTrainedModel", "_init_weights", "class", "config", "def", "down_proj", "elif", "gate_up_proj", "if", "init", "initializer_range", "isinstance", "mean", "module", "normal_", "self", "std", "super", "weight"], "flex_olmo/modeling_flex_olmo.py:FlexOlmoModel.__init__": ["Embedding", "False", "ModelConfig", "ModelDecoderLayer", "ModelModel", "ModelPreTrainedModel", "ModelRMSNorm", "ModelRotaryEmbedding", "ModuleList", "__init__", "class", "config", "def", "embed_tokens", "eps", "for", "gradient_checkpointing", "hidden_size", "in", "layer_idx", "layers", "nn", "norm", "num_hidden_layers", "pad_token_id", "padding_idx", "post_init", "range", "rms_norm_eps", "rotary_emb", "self", "super", "vocab_size"], "flex_olmo/modeling_flex_olmo.py:FlexOlmoModel.forward": ["Cache", "DynamicCache", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "MoeModelOutputWithPast", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "and", "arange", "attention_mask", "bool", "cache_position", "causal_mask", "class", "config", "create_causal_mask", "decoder_layer", "def", "device", "else", "embed_tokens", "exactly", "for", "forward", "get_seq_length", "hidden_states", "if", "in", "input_embeds", "input_ids", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layers", "must", "norm", "not", "num_hidden_layers", "of", "one", "or", "past_key_values", "past_seen_tokens", "position_embeddings", "position_ids", "raise", "return", "rotary_emb", "self", "shape", "specify", "torch", "unsqueeze", "use_cache"], "flex_olmo/modeling_flex_olmo.py:load_balancing_loss_func": ["Model_balancing_loss_func", "None", "Tensor", "_", "attention_mask", "batch_size", "cat", "compute_device", "concatenated_gate_logits", "def", "device", "dim", "else", "expand", "expert_attention_mask", "expert_mask", "float", "for", "functional", "gate_logits", "if", "in", "int", "is", "isinstance", "layer_gate", "mean", "nn", "not", "num_experts", "num_hidden_layers", "one_hot", "or", "overall_loss", "r", "reshape", "return", "router_per_expert_attention_mask", "router_prob_per_expert", "routing_weights", "selected_experts", "sequence_length", "shape", "softmax", "sum", "to", "tokens_per_expert", "top_k", "topk", "torch", "tuple", "unsqueeze"], "flex_olmo/modeling_flex_olmo.py:FlexOlmoForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "ModelForCausalLM", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "num_experts", "num_experts_per_tok", "post_init", "router_aux_loss_coef", "self", "super", "vocab_size"], "flex_olmo/modeling_flex_olmo.py:FlexOlmoForCausalLM.forward": ["Cache", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForCausalLM", "ModelPreTrainedModel", "MoeCausalLMOutputWithPast", "MoeModelOutputWithPast", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "aux_loss", "bool", "cache_position", "class", "config", "def", "device", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "load_balancing_loss_func", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "num_experts", "num_experts_per_tok", "output_router_logits", "outputs", "past_key_values", "position_ids", "r", "return", "router_aux_loss_coef", "router_logits", "self", "slice", "slice_indices", "to", "torch", "use_cache", "vocab_size"], "florence2/modeling_florence2.py:drop_path": ["False", "Model_path", "Model_prob", "Tensor", "bool", "def", "device", "div", "dtype", "float", "floor_", "if", "input", "keep_prob", "ndim", "not", "or", "output", "rand", "random_tensor", "return", "shape", "torch", "training"], "florence2/modeling_florence2.py:Florence2VisionDropPath.__init__": ["ModelVisionDropPath", "Module", "None", "__init__", "class", "def", "drop_prob", "float", "nn", "self", "super"], "florence2/modeling_florence2.py:Florence2VisionDropPath.forward": ["ModelVisionDropPath", "Module", "Tensor", "class", "def", "drop_path", "drop_prob", "forward", "hidden_states", "nn", "return", "self", "torch", "training"], "florence2/modeling_florence2.py:Florence2VisionDropPath.extra_repr": ["ModelVisionDropPath", "Module", "class", "def", "drop_prob", "extra_repr", "f", "nn", "p", "return", "self", "str"], "florence2/modeling_florence2.py:Florence2VisionLearnedAbsolutePositionEmbedding2D.__init__": ["Embedding", "ModelConfig", "ModelVisionLearnedAbsolutePositionEmbedding2D", "Module", "__init__", "class", "column_embeddings", "config", "def", "embed_dim", "embedding_dim", "max_position_embeddings", "nn", "num_pos", "row_embeddings", "self", "super", "vision_config"], "florence2/modeling_florence2.py:Florence2VisionLearnedAbsolutePositionEmbedding2D.forward": ["ModelVisionLearnedAbsolutePositionEmbedding2D", "Module", "None", "arange", "cat", "class", "column_embeddings", "def", "device", "dim", "forward", "height", "height_values", "nn", "permute", "pixel_mask", "pixel_values", "pos", "repeat", "return", "row_embeddings", "self", "shape", "torch", "unsqueeze", "width", "width_values", "x_emb", "y_emb"], "florence2/modeling_florence2.py:Florence2VisionPositionalEmbeddingCosine1D.__init__": ["ModelConfig", "ModelVisionPositionalEmbeddingCosine1D", "Module", "__init__", "class", "config", "cosine", "def", "embed_dim", "empty", "get_sinusoid_embeddings", "max_positions", "max_seq_len", "max_temporal_embeddings", "nn", "pos_idx_to_embed", "register_buffer", "self", "sine", "super", "torch", "vision_config"], "florence2/modeling_florence2.py:Florence2VisionPositionalEmbeddingCosine1D.get_sinusoid_embeddings": ["ModelVisionPositionalEmbeddingCosine1D", "Module", "arange", "class", "cos", "def", "dtype", "emb", "embed_dim", "exp", "float", "get_sinusoid_embeddings", "half_dim", "int", "int64", "log", "math", "max_positions", "nn", "return", "sin", "torch", "unsqueeze"], "florence2/modeling_florence2.py:Florence2VisionPositionalEmbeddingCosine1D.forward": ["Maximum", "ModelVisionPositionalEmbeddingCosine1D", "Module", "Tensor", "ValueError", "class", "def", "f", "forward", "got", "if", "len_seq", "length", "max_seq_len", "nn", "pos_embeds", "pos_idx_to_embed", "raise", "return", "self", "seq_embeds", "sequence", "size", "torch"], "florence2/modeling_florence2.py:Florence2VisionMLP.__init__": ["ACT2FN", "Linear", "ModelVisionConfig", "ModelVisionMLP", "Module", "__init__", "activation_fn", "activation_function", "class", "config", "def", "embed_dim", "fc1", "fc2", "int", "mlp_ratio", "nn", "self", "stage_idx", "super"], "florence2/modeling_florence2.py:Florence2VisionMLP.forward": ["ModelVisionMLP", "Module", "Tensor", "activation_fn", "class", "def", "fc1", "fc2", "forward", "hidden_states", "nn", "return", "self", "torch"], "florence2/modeling_florence2.py:Florence2VisionConvEmbed.__init__": ["Conv2d", "LayerNorm", "ModelVisionConfig", "ModelVisionConvEmbed", "Module", "__init__", "class", "config", "conv", "def", "dim_norm", "else", "embed_dim", "if", "in_channels", "int", "kernel_size", "nn", "norm", "padding", "patch_padding", "patch_prenorm", "patch_size", "patch_stride", "pre_norm", "self", "stage_idx", "stride", "super"], "florence2/modeling_florence2.py:Florence2VisionConvEmbed.forward": ["ModelVisionConvEmbed", "Module", "Tensor", "and", "class", "conv", "def", "forward", "hidden_states", "if", "nn", "norm", "not", "permute", "pre_norm", "return", "self", "torch"], "florence2/modeling_florence2.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "contiguous", "def", "dim", "dropout", "float", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "shape", "size", "softmax", "torch", "training", "transpose", "value"], "florence2/modeling_florence2.py:Florence2VisionChannelAttention.__init__": ["False", "Linear", "ModelVisionChannelAttention", "ModelVisionConfig", "Module", "__init__", "bias", "class", "config", "def", "dim", "embed_dim", "groups", "int", "is_causal", "nn", "num_groups", "proj", "qkv", "qkv_bias", "self", "stage_idx", "super"], "florence2/modeling_florence2.py:Florence2VisionChannelAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "ModelVisionChannelAttention", "Module", "None", "Tensor", "_", "_attn_implementation", "attention_interface", "attention_mask", "batch_size", "class", "config", "def", "eager", "eager_attention_forward", "forward", "groups", "hidden_size", "hidden_states", "if", "key", "nn", "num_tokens", "permute", "proj", "qkv", "query", "reshape", "return", "scale", "scaling", "self", "shape", "torch", "unbind", "value"], "florence2/modeling_florence2.py:Florence2VisionChannelBlock.__init__": ["Conv2d", "Identity", "LayerNorm", "ModelVisionChannelAttention", "ModelVisionChannelBlock", "ModelVisionConfig", "ModelVisionDropPath", "ModelVisionMLP", "Module", "__init__", "channel_attn", "class", "config", "conv1", "conv2", "def", "dim_in", "drop_path1", "drop_path2", "drop_path_rate", "else", "embed_dim", "ffn", "float", "groups", "if", "int", "kernel_size", "nn", "norm1", "norm2", "padding", "self", "stage_idx", "super"], "florence2/modeling_florence2.py:Florence2VisionChannelBlock.forward": ["ModelVisionChannelBlock", "Module", "Tensor", "batch_size", "channel_attn", "class", "conv1", "conv2", "def", "drop_path1", "drop_path2", "embed_dim", "ffn", "flatten", "forward", "height", "hidden_states", "nn", "norm1", "norm2", "residual", "return", "self", "shape", "torch", "transpose", "view", "width"], "florence2/modeling_florence2.py:Florence2VisionWindowAttention.__init__": ["False", "Linear", "ModelVisionConfig", "ModelVisionWindowAttention", "Module", "__init__", "bias", "class", "config", "def", "dim", "embed_dim", "head_dim", "int", "is_causal", "nn", "num_heads", "proj", "qkv", "qkv_bias", "scale", "self", "stage_idx", "super", "window_size"], "florence2/modeling_florence2.py:Florence2VisionWindowAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "F", "ModelVisionWindowAttention", "Module", "None", "Tensor", "_", "_attn_implementation", "attention_interface", "attention_mask", "batch_size", "class", "config", "contiguous", "def", "eager", "eager_attention_forward", "embed_dim", "forward", "height", "hidden_states", "if", "key", "nn", "num_heads", "num_tokens_per_window", "num_windows_per_batch", "pad", "pad_bottom", "pad_left", "pad_right", "pad_top", "padded_height", "padded_width", "permute", "proj", "qkv", "query", "reshape", "return", "scale", "scaling", "self", "shape", "torch", "unbind", "value", "view", "width", "window_size", "windowed_hidden_states"], "florence2/modeling_florence2.py:Florence2VisionSpatialBlock.__init__": ["Conv2d", "Identity", "LayerNorm", "ModelVisionConfig", "ModelVisionDropPath", "ModelVisionMLP", "ModelVisionSpatialBlock", "ModelVisionWindowAttention", "Module", "__init__", "class", "config", "conv1", "conv2", "def", "drop_path1", "drop_path2", "drop_path_rate", "else", "embed_dim", "ffn", "float", "groups", "if", "int", "kernel_size", "nn", "norm1", "norm2", "padding", "self", "stage_idx", "super", "window_attn"], "florence2/modeling_florence2.py:Florence2VisionSpatialBlock.forward": ["ModelVisionSpatialBlock", "Module", "Tensor", "batch_size", "class", "conv1", "conv2", "def", "drop_path1", "drop_path2", "embed_dim", "ffn", "flatten", "forward", "height", "hidden_states", "nn", "norm1", "norm2", "residual", "return", "self", "shape", "torch", "transpose", "view", "width", "window_attn"], "florence2/modeling_florence2.py:Florence2VisionBlock.__init__": ["ModelVisionBlock", "ModelVisionChannelBlock", "ModelVisionConfig", "ModelVisionSpatialBlock", "Module", "__init__", "channel_block", "channel_drop_path_rate", "class", "config", "def", "drop_path_rate", "float", "int", "nn", "self", "spatial_block", "spatial_drop_path_rate", "stage_idx", "super"], "florence2/modeling_florence2.py:Florence2VisionBlock.forward": ["ModelVisionBlock", "Module", "Tensor", "channel_block", "class", "def", "forward", "hidden_states", "nn", "return", "self", "spatial_block", "torch"], "florence2/modeling_florence2.py:Florence2VisionBackbone.__init__": ["Expected", "ModelVisionBackbone", "ModelVisionBlock", "ModelVisionConfig", "ModelVisionConvEmbed", "ModelVisionPreTrainedModel", "ModuleList", "ValueError", "__init__", "append", "block", "block_idx", "blocks", "channel_drop_path_rate", "class", "config", "conv_embed", "convs", "cpu", "def", "depth_offset", "depths", "device", "dpr", "drop_path_rate", "embed_dim", "f", "for", "if", "in", "item", "len", "linspace", "nn", "not", "num_groups", "num_heads", "num_stages", "post_init", "raise", "range", "self", "spatial_drop_path_rate", "stage_idx", "sum", "super", "torch", "x"], "florence2/modeling_florence2.py:Florence2VisionBackbone.forward": ["ModelVisionBackbone", "ModelVisionPreTrainedModel", "Tensor", "block", "blocks", "class", "conv", "convs", "def", "for", "forward", "hidden_states", "in", "kwargs", "layer", "return", "self", "torch", "zip"], "florence2/modeling_florence2.py:Florence2MultiModalProjector.__init__": ["False", "LayerNorm", "Linear", "ModelConfig", "ModelMultiModalProjector", "ModelVisionLearnedAbsolutePositionEmbedding2D", "ModelVisionPositionalEmbeddingCosine1D", "Module", "__init__", "bias", "class", "config", "def", "embed_dim", "image_position_embed", "image_proj_norm", "image_projection", "nn", "projection_dim", "self", "super", "vision_config", "vision_embedding_dim", "vision_projection_dim", "visual_temporal_embed"], "florence2/modeling_florence2.py:Florence2MultiModalProjector.forward": ["ModelMultiModalProjector", "Module", "cat", "class", "def", "dim", "flatten", "forward", "image_features", "image_position_embed", "image_proj_norm", "image_projection", "mean", "nn", "position_features", "return", "self", "spatial_image_features", "temporal_features", "temporal_image_features", "torch", "transpose", "unsqueeze", "visual_temporal_embed", "visual_token_features"], "florence2/modeling_florence2.py:Florence2PreTrainedModel._init_weights": ["ModelPreTrainedModel", "ModelVisionPositionalEmbeddingCosine1D", "PreTrainedModel", "_init_weights", "class", "copy_", "cosine", "def", "embed_dim", "empty", "get_sinusoid_embeddings", "if", "init", "isinstance", "max_positions", "max_seq_len", "module", "pos_idx_to_embed", "self", "sine", "super", "torch"], "florence2/modeling_florence2.py:Florence2Model.__init__": ["AutoModel", "ModelConfig", "ModelModel", "ModelMultiModalProjector", "ModelPreTrainedModel", "ModelVisionBackbone", "__init__", "class", "config", "def", "from_config", "language_model", "multi_modal_projector", "post_init", "self", "super", "text_config", "vision_config", "vision_tower"], "florence2/modeling_florence2.py:Florence2Model.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "language_model", "return", "self"], "florence2/modeling_florence2.py:Florence2Model.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "language_model", "self", "set_input_embeddings", "value"], "florence2/modeling_florence2.py:Florence2Model.get_image_features": ["ModelModel", "ModelPreTrainedModel", "Tensor", "class", "def", "get_image_features", "image_embeds", "image_features", "kwargs", "multi_modal_projector", "pixel_values", "return", "self", "torch", "vision_tower"], "florence2/modeling_florence2.py:Florence2Model.get_placeholder_mask": ["FloatTensor", "Image", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "ValueError", "all", "and", "class", "config", "def", "device", "do", "dtype", "else", "expand_as", "f", "features", "get_input_embeddings", "get_placeholder_mask", "if", "image", "image_features", "image_token_id", "input_ids", "inputs_embeds", "is", "long", "match", "n_image_features", "n_image_tokens", "not", "numel", "raise", "return", "self", "shape", "special_image_mask", "sum", "tensor", "to", "tokens", "torch", "unsqueeze"], "florence2/modeling_florence2.py:Florence2Model.forward": ["Cache", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "ModelSeq2SeqModelOutput", "None", "Tensor", "True", "ValueError", "You", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "cross_attentions", "decoder", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_input_ids", "decoder_inputs_embeds", "decoder_outputs", "decoder_start_token_id", "def", "device", "dtype", "else", "encoder", "encoder_attention_mask", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "exactly", "forward", "get_image_features", "get_input_embeddings", "get_placeholder_mask", "hidden_states", "if", "image_features", "image_hidden_states", "input_ids", "inputs_embeds", "is", "kwargs", "language_model", "last_hidden_state", "list", "long", "masked_scatter", "must", "not", "of", "one", "ones", "or", "output_attentions", "output_hidden_states", "past_key_values", "pixel_values", "raise", "return", "return_dict", "self", "size", "special_image_mask", "specify", "text_config", "to", "torch", "tuple", "use_cache", "use_return_dict"], "florence2/modeling_florence2.py:Florence2Model.get_encoder": ["ModelModel", "ModelPreTrainedModel", "None", "class", "def", "else", "get_encoder", "if", "is", "language_model", "modality", "return", "self", "super"], "florence2/modeling_florence2.py:shift_tokens_right": ["Model_tokens_right", "Modeled_input_ids", "None", "Tensor", "ValueError", "be", "clone", "config", "decoder_start_token_id", "def", "defined", "has", "if", "input_ids", "int", "is", "masked_fill_", "model", "new_zeros", "pad_token_id", "raise", "return", "self", "shape", "to", "torch"], "florence2/modeling_florence2.py:Florence2ForConditionalGeneration.__init__": ["False", "GenerationMixin", "Linear", "ModelConfig", "ModelForConditionalGeneration", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "text_config", "vocab_size"], "florence2/modeling_florence2.py:Florence2ForConditionalGeneration.get_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "model", "return", "self"], "florence2/modeling_florence2.py:Florence2ForConditionalGeneration.set_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "model", "self", "set_input_embeddings", "value"], "florence2/modeling_florence2.py:Florence2ForConditionalGeneration.get_output_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "Module", "class", "def", "get_output_embeddings", "lm_head", "nn", "return", "self"], "florence2/modeling_florence2.py:Florence2ForConditionalGeneration.get_image_features": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "Tensor", "class", "def", "get_image_features", "kwargs", "model", "pixel_values", "return", "self", "torch"], "florence2/modeling_florence2.py:Florence2ForConditionalGeneration.forward": ["Cache", "False", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForConditionalGeneration", "ModelPreTrainedModel", "ModelSeq2SeqLMOutput", "None", "Tensor", "The", "TransformersKwargs", "True", "Unpack", "and", "argument", "attention_mask", "bool", "cache_position", "changed", "class", "config", "cross_attentions", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_input_ids", "decoder_inputs_embeds", "decoder_start_token_id", "def", "else", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "forward", "hidden_states", "if", "image_hidden_states", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "list", "lm_head", "logger", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "output_attentions", "output_hidden_states", "outputs", "pad_token_id", "past_key_values", "pixel_values", "provided", "r", "return", "return_dict", "self", "shift_tokens_right", "since", "slice", "slice_indices", "text_config", "to", "torch", "tuple", "use_cache", "use_return_dict", "vocab_size", "warning"], "florence2/modeling_florence2.py:Florence2ForConditionalGeneration.prepare_inputs_for_generation": ["False", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "True", "attention_mask", "cache_position", "class", "def", "get", "if", "input_ids", "inputs_embeds", "is_first_iteration", "kwargs", "logits_to_keep", "model_inputs", "not", "or", "past_key_values", "pixel_values", "prepare_inputs_for_generation", "return", "self", "super", "use_cache"], "florence2/modeling_florence2.py:Florence2ForConditionalGeneration.get_placeholder_mask": ["FloatTensor", "GenerationMixin", "LongTensor", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "get_placeholder_mask", "image_features", "input_ids", "inputs_embeds", "model", "return", "self", "torch"], "florence2/modeling_florence2.py:Florence2ForConditionalGeneration._prepare_encoder_decoder_kwargs_for_generation": ["Any", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Tensor", "_prepare_encoder_decoder_kwargs_for_generation", "class", "def", "device", "dict", "dtype", "generation_config", "get_image_features", "get_input_embeddings", "get_placeholder_mask", "if", "image_features", "inputs_embeds", "inputs_tensor", "is", "masked_scatter", "model_input_name", "model_kwargs", "not", "pixel_values", "pop", "return", "self", "special_image_mask", "str", "super", "to", "torch"], "fnet/modeling_fnet.py:_two_dim_matmul": ["_two_dim_matmul", "bij", "bnk", "complex64", "def", "einsum", "jk", "matrix_dim_one", "matrix_dim_two", "ni", "return", "seq_length", "shape", "torch", "type", "x"], "fnet/modeling_fnet.py:two_dim_matmul": ["Model_dim_matmul", "_Model_dim_matmul", "def", "matrix_dim_Model", "matrix_dim_one", "return", "x"], "fnet/modeling_fnet.py:fftn": ["Model", "axis", "def", "fft", "for", "in", "ndim", "out", "range", "return", "reversed", "torch", "x"], "fnet/modeling_fnet.py:FNetEmbeddings.__init__": ["Dropout", "Embedding", "False", "LayerNorm", "Linear", "Model", "Module", "__init__", "arange", "class", "config", "def", "dropout", "dtype", "eps", "expand", "hidden_dropout_prob", "hidden_size", "layer_norm_eps", "long", "max_position_embeddings", "nn", "pad_token_id", "padding_idx", "persistent", "position_embeddings", "position_ids", "projection", "register_buffer", "self", "size", "super", "token_type_embeddings", "token_type_ids", "torch", "type_vocab_size", "vocab_size", "word_embeddings", "zeros"], "fnet/modeling_fnet.py:FNetEmbeddings.forward": ["LayerNorm", "Model", "Module", "None", "buffered_token_type_ids", "buffered_token_type_ids_expanded", "class", "def", "device", "dropout", "dtype", "else", "embeddings", "expand", "forward", "hasattr", "if", "input_ids", "input_shape", "inputs_embeds", "is", "long", "nn", "not", "position_embeddings", "position_ids", "projection", "return", "self", "seq_length", "size", "token_type_embeddings", "token_type_ids", "torch", "word_embeddings", "zeros"], "fnet/modeling_fnet.py:FNetBasicFourierTransform.__init__": ["Model", "Module", "__init__", "_init_fourier_transform", "class", "config", "def", "nn", "self", "super"], "fnet/modeling_fnet.py:FNetBasicFourierTransform._init_fourier_transform": ["DFT", "Model", "Module", "SciPy", "TPU", "Using", "_init_fourier_transform", "and", "calculation", "class", "complex64", "config", "def", "dft", "dft_mat_hidden", "dft_mat_seq", "dim", "dtype", "elif", "else", "fast", "fft", "fftn", "for", "found", "fourier", "fourier_transform", "hidden_size", "if", "instead", "is", "is_scipy_available", "linalg", "logging", "matrix", "matrix_dim_one", "matrix_dim_two", "max_position_embeddings", "needed", "nn", "not", "optimized", "partial", "register_buffer", "self", "tensor", "torch", "tpu_short_seq_length", "transform", "two_dim_matmul", "use_tpu_fourier_optimizations", "warning"], "fnet/modeling_fnet.py:FNetBasicFourierTransform.forward": ["Model", "Module", "class", "def", "forward", "fourier_transform", "hidden_states", "nn", "outputs", "real", "return", "self"], "fnet/modeling_fnet.py:FNetBasicOutput.__init__": ["LayerNorm", "Model", "Module", "__init__", "class", "config", "def", "eps", "hidden_size", "layer_norm_eps", "nn", "self", "super"], "fnet/modeling_fnet.py:FNetBasicOutput.forward": ["LayerNorm", "Model", "Module", "class", "def", "forward", "hidden_states", "input_tensor", "nn", "return", "self"], "fnet/modeling_fnet.py:FNetFourierTransform.__init__": ["Model", "ModelBasicFourierTransform", "ModelBasicOutput", "Module", "__init__", "class", "config", "def", "nn", "output", "self", "super"], "fnet/modeling_fnet.py:FNetFourierTransform.forward": ["Model", "Module", "class", "def", "forward", "fourier_output", "hidden_states", "nn", "output", "outputs", "return", "self", "self_outputs"], "fnet/modeling_fnet.py:FNetIntermediate.__init__": ["ACT2FN", "Linear", "Model", "Module", "__init__", "class", "config", "def", "dense", "else", "hidden_act", "hidden_size", "if", "intermediate_act_fn", "intermediate_size", "isinstance", "nn", "self", "str", "super"], "fnet/modeling_fnet.py:FNetIntermediate.forward": ["Model", "Module", "Tensor", "class", "def", "dense", "forward", "hidden_states", "intermediate_act_fn", "nn", "return", "self", "torch"], "fnet/modeling_fnet.py:FNetOutput.__init__": ["Dropout", "LayerNorm", "Linear", "Model", "Module", "__init__", "class", "config", "def", "dense", "dropout", "eps", "hidden_dropout_prob", "hidden_size", "intermediate_size", "layer_norm_eps", "nn", "self", "super"], "fnet/modeling_fnet.py:FNetOutput.forward": ["LayerNorm", "Model", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "fnet/modeling_fnet.py:FNetLayer.__init__": ["GradientCheckpointingLayer", "Model", "ModelFourierTransform", "ModelIntermediate", "ModelOutput", "__init__", "chunk_size_feed_forward", "class", "config", "def", "fourier", "intermediate", "output", "self", "seq_len_dim", "super"], "fnet/modeling_fnet.py:FNetLayer.forward": ["GradientCheckpointingLayer", "Model", "apply_chunking_to_forward", "chunk_size_feed_forward", "class", "def", "feed_forward_chunk", "forward", "fourier", "fourier_output", "hidden_states", "layer_output", "outputs", "return", "self", "self_fourier_outputs", "seq_len_dim"], "fnet/modeling_fnet.py:FNetLayer.feed_forward_chunk": ["GradientCheckpointingLayer", "Model", "class", "def", "feed_forward_chunk", "fourier_output", "intermediate", "intermediate_output", "layer_output", "output", "return", "self"], "fnet/modeling_fnet.py:FNetEncoder.__init__": ["False", "Model", "ModelLayer", "Module", "ModuleList", "_", "__init__", "class", "config", "def", "for", "gradient_checkpointing", "in", "layer", "nn", "num_hidden_layers", "range", "self", "super"], "fnet/modeling_fnet.py:FNetEncoder.forward": ["BaseModelOutput", "False", "Model", "Module", "None", "True", "all_hidden_states", "class", "def", "else", "enumerate", "for", "forward", "hidden_states", "i", "if", "in", "is", "last_hidden_state", "layer", "layer_module", "layer_outputs", "nn", "not", "output_hidden_states", "return", "return_dict", "self", "tuple", "v"], "fnet/modeling_fnet.py:FNetPooler.__init__": ["Linear", "Model", "Module", "Tanh", "__init__", "activation", "class", "config", "def", "dense", "hidden_size", "nn", "self", "super"], "fnet/modeling_fnet.py:FNetPooler.forward": ["Model", "Module", "Tensor", "activation", "class", "def", "dense", "first_token_tensor", "forward", "hidden_states", "nn", "pooled_output", "return", "self", "torch"], "fnet/modeling_fnet.py:FNetPredictionHeadTransform.__init__": ["ACT2FN", "LayerNorm", "Linear", "Model", "Module", "__init__", "class", "config", "def", "dense", "else", "eps", "hidden_act", "hidden_size", "if", "isinstance", "layer_norm_eps", "nn", "self", "str", "super", "transform_act_fn"], "fnet/modeling_fnet.py:FNetPredictionHeadTransform.forward": ["LayerNorm", "Model", "Module", "Tensor", "class", "def", "dense", "forward", "hidden_states", "nn", "return", "self", "torch", "transform_act_fn"], "fnet/modeling_fnet.py:FNetLMPredictionHead.__init__": ["Linear", "Model", "ModelPredictionHeadTransform", "Module", "Parameter", "__init__", "bias", "class", "config", "decoder", "def", "hidden_size", "nn", "self", "super", "torch", "transform", "vocab_size", "zeros"], "fnet/modeling_fnet.py:FNetLMPredictionHead.forward": ["Model", "Module", "class", "decoder", "def", "forward", "hidden_states", "nn", "return", "self", "transform"], "fnet/modeling_fnet.py:FNetOnlyMLMHead.__init__": ["Model", "ModelLMPredictionHead", "Module", "__init__", "class", "config", "def", "nn", "predictions", "self", "super"], "fnet/modeling_fnet.py:FNetOnlyMLMHead.forward": ["Model", "Module", "class", "def", "forward", "nn", "prediction_scores", "predictions", "return", "self", "sequence_output"], "fnet/modeling_fnet.py:FNetOnlyNSPHead.__init__": ["Linear", "Model", "Module", "__init__", "class", "config", "def", "hidden_size", "nn", "self", "seq_relationship", "super"], "fnet/modeling_fnet.py:FNetOnlyNSPHead.forward": ["Model", "Module", "class", "def", "forward", "nn", "pooled_output", "return", "self", "seq_relationship", "seq_relationship_score"], "fnet/modeling_fnet.py:FNetPreTrainingHeads.__init__": ["Linear", "Model", "ModelLMPredictionHead", "Module", "__init__", "class", "config", "def", "hidden_size", "nn", "predictions", "self", "seq_relationship", "super"], "fnet/modeling_fnet.py:FNetPreTrainingHeads.forward": ["Model", "Module", "class", "def", "forward", "nn", "pooled_output", "prediction_scores", "predictions", "return", "self", "seq_relationship", "seq_relationship_score", "sequence_output"], "fnet/modeling_fnet.py:FNetPreTrainedModel._init_weights": ["Model", "ModelEmbeddings", "PreTrainedModel", "_init_weights", "arange", "class", "copy_", "def", "expand", "if", "init", "isinstance", "module", "position_ids", "self", "shape", "super", "token_type_ids", "torch", "zeros_"], "fnet/modeling_fnet.py:FNetModel.__init__": ["Model", "ModelEmbeddings", "ModelEncoder", "ModelPooler", "ModelPreTrainedModel", "None", "True", "__init__", "add_pooling_layer", "class", "config", "def", "else", "embeddings", "encoder", "if", "pooler", "post_init", "r", "self", "super"], "fnet/modeling_fnet.py:FNetModel.get_input_embeddings": ["Model", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "return", "self", "word_embeddings"], "fnet/modeling_fnet.py:FNetModel.set_input_embeddings": ["Model", "ModelPreTrainedModel", "class", "def", "embeddings", "self", "set_input_embeddings", "value", "word_embeddings"], "fnet/modeling_fnet.py:FNetModel.forward": ["BaseModelOutput", "BaseModelOutputWithPooling", "FloatTensor", "LongTensor", "Model", "ModelConfig", "ModelPreTrainedModel", "None", "TPU", "The", "ValueError", "You", "and", "at", "batch_size", "be", "being", "bool", "both", "buffered_token_type_ids", "buffered_token_type_ids_expanded", "cannot", "class", "config", "def", "device", "dtype", "either", "elif", "else", "embedding_output", "embeddings", "encoder", "encoder_outputs", "equal", "expand", "forward", "hasattr", "have", "hidden_states", "if", "in", "input_ids", "input_shape", "inputs_embeds", "is", "kwargs", "last_hidden_state", "length", "long", "model", "not", "optimizations", "or", "output_hidden_states", "passed", "pooler", "pooler_output", "position_ids", "raise", "return", "return_dict", "same", "self", "seq_length", "sequence", "sequence_output", "set", "should", "size", "specify", "the", "time", "to", "token_type_ids", "torch", "tpu_short_seq_length", "tuple", "use_return_dict", "use_tpu_fourier_optimizations", "using", "when", "zeros"], "fnet/modeling_fnet.py:FNetForPreTraining.__init__": ["Model", "ModelModel", "ModelPreTrainedModel", "ModelPreTrainingHeads", "__init__", "class", "cls", "config", "def", "post_init", "self", "super"], "fnet/modeling_fnet.py:FNetForPreTraining.get_output_embeddings": ["Model", "ModelPreTrainedModel", "class", "cls", "decoder", "def", "get_output_embeddings", "predictions", "return", "self"], "fnet/modeling_fnet.py:FNetForPreTraining.set_output_embeddings": ["Model", "ModelPreTrainedModel", "bias", "class", "cls", "decoder", "def", "new_embeddings", "predictions", "self", "set_output_embeddings"], "fnet/modeling_fnet.py:FNetForPreTraining.forward": ["CrossEntropyLoss", "Model", "ModelOutput", "ModelPreTrainedModel", "None", "Tensor", "and", "bool", "class", "cls", "config", "def", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "loss", "loss_fct", "masked_lm_loss", "next_sentence_label", "next_sentence_loss", "not", "output", "output_hidden_states", "outputs", "pooled_output", "position_ids", "prediction_logits", "prediction_scores", "r", "return", "return_dict", "self", "seq_relationship_logits", "seq_relationship_score", "sequence_output", "token_type_ids", "torch", "total_loss", "tuple", "use_return_dict", "view", "vocab_size"], "fnet/modeling_fnet.py:FNetForMaskedLM.__init__": ["Model", "ModelModel", "ModelOnlyMLMHead", "ModelPreTrainedModel", "__init__", "class", "cls", "config", "def", "post_init", "self", "super"], "fnet/modeling_fnet.py:FNetForMaskedLM.get_output_embeddings": ["Model", "ModelPreTrainedModel", "class", "cls", "decoder", "def", "get_output_embeddings", "predictions", "return", "self"], "fnet/modeling_fnet.py:FNetForMaskedLM.set_output_embeddings": ["Model", "ModelPreTrainedModel", "bias", "class", "cls", "decoder", "def", "new_embeddings", "predictions", "self", "set_output_embeddings"], "fnet/modeling_fnet.py:FNetForMaskedLM.forward": ["CrossEntropyLoss", "MaskedLMOutput", "Model", "ModelPreTrainedModel", "None", "Tensor", "bool", "class", "cls", "config", "def", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "masked_lm_loss", "not", "output", "output_hidden_states", "outputs", "position_ids", "prediction_scores", "r", "return", "return_dict", "self", "sequence_output", "token_type_ids", "torch", "tuple", "use_return_dict", "view", "vocab_size"], "fnet/modeling_fnet.py:FNetForNextSentencePrediction.__init__": ["Model", "ModelModel", "ModelOnlyNSPHead", "ModelPreTrainedModel", "__init__", "class", "cls", "config", "def", "post_init", "self", "super"], "fnet/modeling_fnet.py:FNetForNextSentencePrediction.forward": ["CrossEntropyLoss", "Model", "ModelPreTrainedModel", "NextSentencePredictorOutput", "None", "Tensor", "bool", "class", "cls", "config", "def", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "next_sentence_loss", "not", "output", "output_hidden_states", "outputs", "pooled_output", "position_ids", "r", "return", "return_dict", "self", "seq_relationship_scores", "token_type_ids", "torch", "tuple", "use_return_dict", "view"], "fnet/modeling_fnet.py:FNetForSequenceClassification.__init__": ["Dropout", "Linear", "Model", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "classifier", "config", "def", "dropout", "hidden_dropout_prob", "hidden_size", "nn", "num_labels", "post_init", "self", "super"], "fnet/modeling_fnet.py:FNetForSequenceClassification.forward": ["BCEWithLogitsLoss", "CrossEntropyLoss", "MSELoss", "Model", "ModelPreTrainedModel", "None", "SequenceClassifierOutput", "Tensor", "and", "bool", "class", "classifier", "config", "def", "dropout", "dtype", "elif", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "kwargs", "labels", "logits", "long", "loss", "loss_fct", "multi_label_classification", "not", "num_labels", "or", "output", "output_hidden_states", "outputs", "pooled_output", "position_ids", "problem_type", "r", "regression", "return", "return_dict", "self", "single_label_classification", "squeeze", "token_type_ids", "torch", "tuple", "use_return_dict", "view"], "fnet/modeling_fnet.py:FNetForMultipleChoice.__init__": ["Dropout", "Linear", "Model", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "classifier", "config", "def", "dropout", "hidden_dropout_prob", "hidden_size", "nn", "post_init", "self", "super"], "fnet/modeling_fnet.py:FNetForMultipleChoice.forward": ["CrossEntropyLoss", "Model", "ModelPreTrainedModel", "MultipleChoiceModelOutput", "None", "Tensor", "bool", "class", "classifier", "config", "def", "dropout", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "not", "num_choices", "output", "output_hidden_states", "outputs", "pooled_output", "position_ids", "r", "reshaped_logits", "return", "return_dict", "self", "shape", "size", "token_type_ids", "torch", "tuple", "use_return_dict", "view"], "fnet/modeling_fnet.py:FNetForTokenClassification.__init__": ["Dropout", "Linear", "Model", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "classifier", "config", "def", "dropout", "hidden_dropout_prob", "hidden_size", "nn", "num_labels", "post_init", "self", "super"], "fnet/modeling_fnet.py:FNetForTokenClassification.forward": ["CrossEntropyLoss", "Model", "ModelPreTrainedModel", "None", "Tensor", "TokenClassifierOutput", "bool", "class", "classifier", "config", "def", "dropout", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "not", "num_labels", "output", "output_hidden_states", "outputs", "position_ids", "r", "return", "return_dict", "self", "sequence_output", "token_type_ids", "torch", "tuple", "use_return_dict", "view"], "fnet/modeling_fnet.py:FNetForQuestionAnswering.__init__": ["Linear", "Model", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "config", "def", "hidden_size", "nn", "num_labels", "post_init", "qa_outputs", "self", "super"], "fnet/modeling_fnet.py:FNetForQuestionAnswering.forward": ["CrossEntropyLoss", "Model", "ModelPreTrainedModel", "None", "QuestionAnsweringModelOutput", "Tensor", "and", "bool", "clamp", "class", "config", "contiguous", "def", "dim", "else", "end_logits", "end_loss", "end_positions", "forward", "hidden_states", "if", "ignore_index", "ignored_index", "input_ids", "inputs_embeds", "is", "kwargs", "len", "logits", "loss", "loss_fct", "not", "output", "output_hidden_states", "outputs", "position_ids", "qa_outputs", "return", "return_dict", "self", "sequence_output", "size", "split", "squeeze", "start_logits", "start_loss", "start_positions", "token_type_ids", "torch", "total_loss", "tuple", "use_return_dict"], "focalnet/modeling_focalnet.py:FocalNetEmbeddings.__init__": ["Dropout", "False", "LayerNorm", "ModelEmbeddings", "ModelPatchEmbeddings", "Module", "None", "Parameter", "True", "__init__", "class", "config", "def", "dropout", "else", "embed_dim", "eps", "grid_size", "hidden_dropout_prob", "if", "image_size", "is_stem", "layer_norm_eps", "mask_token", "nn", "norm", "num_channels", "patch_embeddings", "patch_grid", "patch_size", "self", "super", "torch", "use_conv_embed", "use_mask_token", "zeros"], "focalnet/modeling_focalnet.py:FocalNetEmbeddings.forward": ["BoolTensor", "FloatTensor", "ModelEmbeddings", "Module", "None", "Tensor", "_", "batch_size", "bool_masked_pos", "class", "def", "dropout", "embeddings", "expand", "forward", "if", "is", "mask", "mask_token", "mask_tokens", "nn", "norm", "not", "output_dimensions", "patch_embeddings", "pixel_values", "return", "self", "seq_len", "size", "torch", "tuple", "type_as", "unsqueeze"], "focalnet/modeling_focalnet.py:FocalNetPatchEmbeddings.__init__": ["Conv2d", "False", "Iterable", "LayerNorm", "ModelPatchEmbeddings", "Module", "None", "__init__", "abc", "add_norm", "class", "collections", "config", "def", "else", "embed_dim", "eps", "grid_size", "if", "image_size", "is_stem", "isinstance", "kernel_size", "layer_norm_eps", "nn", "norm", "num_channels", "num_patches", "padding", "patch_size", "projection", "self", "stride", "super", "use_conv_embed"], "focalnet/modeling_focalnet.py:FocalNetPatchEmbeddings.maybe_pad": ["ModelPatchEmbeddings", "Module", "class", "def", "functional", "height", "if", "maybe_pad", "nn", "pad", "pad_values", "patch_size", "pixel_values", "return", "self", "width"], "focalnet/modeling_focalnet.py:FocalNetPatchEmbeddings.forward": ["FloatTensor", "Make", "ModelPatchEmbeddings", "Module", "None", "Tensor", "ValueError", "_", "channel", "class", "configuration", "def", "dimension", "embeddings", "flatten", "forward", "height", "if", "in", "int", "is", "match", "maybe_pad", "nn", "norm", "not", "num_channels", "of", "one", "output_dimensions", "pixel", "pixel_values", "projection", "raise", "return", "self", "set", "shape", "sure", "that", "the", "torch", "transpose", "tuple", "values", "width", "with"], "focalnet/modeling_focalnet.py:drop_path": ["False", "Model_path", "Model_prob", "Tensor", "bool", "def", "device", "div", "dtype", "float", "floor_", "if", "input", "keep_prob", "ndim", "not", "or", "output", "rand", "random_tensor", "return", "shape", "torch", "training"], "focalnet/modeling_focalnet.py:FocalNetDropPath.__init__": ["ModelDropPath", "Module", "None", "__init__", "class", "def", "drop_prob", "float", "nn", "self", "super"], "focalnet/modeling_focalnet.py:FocalNetDropPath.forward": ["ModelDropPath", "Module", "Tensor", "class", "def", "drop_path", "drop_prob", "forward", "hidden_states", "nn", "return", "self", "torch", "training"], "focalnet/modeling_focalnet.py:FocalNetDropPath.extra_repr": ["ModelDropPath", "Module", "class", "def", "drop_prob", "extra_repr", "f", "nn", "p", "return", "self", "str"], "focalnet/modeling_focalnet.py:FocalNetModulation.__init__": ["Conv2d", "Dropout", "False", "GELU", "LayerNorm", "Linear", "ModelModulation", "Model_factor", "Model_layers", "Model_level", "Model_levels", "Model_window", "Model_windows", "Module", "ModuleList", "Sequential", "True", "__init__", "activation", "append", "bias", "class", "config", "def", "dim", "eps", "for", "groups", "if", "in", "index", "k", "kernel_size", "kernel_sizes", "layer_norm_eps", "layernorm", "nn", "normalize_modulator", "padding", "projection_context", "projection_dropout", "projection_in", "projection_out", "range", "self", "stride", "super", "use_post_layernorm_in_modulation"], "focalnet/modeling_focalnet.py:FocalNetModulation.forward": ["ModelModulation", "Model_layers", "Model_level", "Module", "True", "activation", "class", "contiguous", "ctx", "ctx_all", "ctx_global", "def", "for", "forward", "gates", "hidden_state", "if", "in", "keepdim", "layernorm", "level", "mean", "modulator", "nn", "normalize_modulator", "num_channels", "permute", "projection_context", "projection_dropout", "projection_in", "projection_out", "q", "range", "return", "self", "shape", "split", "torch", "use_post_layernorm_in_modulation", "x", "x_out"], "focalnet/modeling_focalnet.py:FocalNetMlp.__init__": ["ACT2FN", "Dropout", "Linear", "ModelMlp", "Module", "None", "__init__", "activation", "class", "config", "def", "drop", "fc1", "fc2", "hidden_act", "hidden_features", "in_features", "nn", "or", "out_features", "self", "super"], "focalnet/modeling_focalnet.py:FocalNetMlp.forward": ["ModelMlp", "Module", "activation", "class", "def", "drop", "fc1", "fc2", "forward", "hidden_state", "nn", "return", "self"], "focalnet/modeling_focalnet.py:FocalNetLayer.__init__": ["Identity", "LayerNorm", "ModelDropPath", "ModelLayer", "ModelMlp", "ModelModulation", "Module", "Parameter", "True", "__init__", "class", "config", "def", "dim", "drop", "drop_path", "else", "eps", "gamma_1", "gamma_2", "hidden_dropout_prob", "hidden_features", "if", "in_features", "index", "input_resolution", "int", "layer_norm_eps", "layerscale_value", "mlp", "mlp_hidden_dim", "mlp_ratio", "modulation", "nn", "norm1", "norm2", "ones", "projection_dropout", "requires_grad", "self", "super", "torch", "use_layerscale", "use_post_layernorm"], "focalnet/modeling_focalnet.py:FocalNetLayer.forward": ["ModelLayer", "Module", "_", "batch_size", "class", "def", "drop_path", "else", "forward", "gamma_1", "gamma_2", "height", "hidden_state", "if", "input_dimensions", "mlp", "modulation", "nn", "norm1", "norm2", "not", "num_channels", "return", "self", "shape", "shortcut", "use_post_layernorm", "view", "width"], "focalnet/modeling_focalnet.py:FocalNetStage.__init__": ["False", "GradientCheckpointingLayer", "ModelLayer", "ModelPatchEmbeddings", "ModelStage", "ModuleList", "None", "True", "__init__", "add_norm", "class", "config", "cpu", "def", "depths", "device", "dim", "downsample", "dpr", "drop_path", "drop_path_rate", "else", "embed_dim", "for", "i", "if", "image_size", "in", "index", "input_resolution", "is", "is_stem", "isinstance", "item", "layers", "len", "linspace", "list", "nn", "not", "num_channels", "num_stages", "out_dim", "patch_size", "pointing", "range", "self", "sum", "super", "torch", "use_conv_embed", "x"], "focalnet/modeling_focalnet.py:FocalNetStage.forward": ["GradientCheckpointingLayer", "ModelStage", "None", "Tensor", "class", "def", "downsample", "else", "for", "forward", "height", "hidden_states", "hidden_states_before_downsampling", "if", "in", "input_dimensions", "int", "is", "layer_module", "layers", "not", "output_dimensions", "reshape", "return", "self", "shape", "stage_outputs", "torch", "transpose", "tuple", "width"], "focalnet/modeling_focalnet.py:FocalNetEncoder.__init__": ["False", "ModelEncoder", "ModelStage", "Module", "ModuleList", "__init__", "class", "config", "def", "depths", "for", "gradient_checkpointing", "grid_size", "i_layer", "in", "index", "input_resolution", "len", "nn", "num_stages", "range", "self", "stages", "super"], "focalnet/modeling_focalnet.py:FocalNetEncoder.forward": ["False", "ModelEncoder", "ModelEncoderOutput", "Module", "None", "Tensor", "True", "_", "all_hidden_states", "all_reshaped_hidden_states", "and", "batch_size", "bool", "class", "def", "elif", "else", "enumerate", "for", "forward", "hidden_size", "hidden_states", "hidden_states_before_downsampling", "i", "if", "in", "input_dimensions", "int", "is", "last_hidden_state", "nn", "not", "output_dimensions", "output_hidden_states", "output_hidden_states_before_downsampling", "permute", "reshaped_hidden_state", "reshaped_hidden_states", "return", "return_dict", "self", "shape", "stage_module", "stage_outputs", "stages", "torch", "tuple", "v", "view"], "focalnet/modeling_focalnet.py:FocalNetPreTrainedModel._init_weights": ["ModelEmbeddings", "ModelLayer", "ModelPreTrainedModel", "None", "PreTrainedModel", "_init_weights", "class", "config", "constant_", "def", "elif", "gamma_1", "gamma_2", "if", "init", "is", "isinstance", "layerscale_value", "mask_token", "module", "not", "self", "super", "use_layerscale", "zeros_"], "focalnet/modeling_focalnet.py:FocalNetModel.__init__": ["AdaptiveAvgPool1d", "False", "LayerNorm", "ModelEmbeddings", "ModelEncoder", "ModelModel", "ModelPreTrainedModel", "None", "True", "__init__", "add_pooling_layer", "class", "config", "def", "depths", "else", "embed_dim", "embeddings", "encoder", "eps", "if", "int", "layer_norm_eps", "layernorm", "len", "nn", "num_features", "num_stages", "patch_grid", "pooler", "post_init", "r", "self", "super", "use_mask_token"], "focalnet/modeling_focalnet.py:FocalNetModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "patch_embeddings", "return", "self"], "focalnet/modeling_focalnet.py:FocalNetModel.forward": ["BoolTensor", "FloatTensor", "ModelModel", "ModelModelOutput", "ModelPreTrainedModel", "None", "ValueError", "You", "bool", "bool_masked_pos", "class", "config", "def", "else", "embedding_output", "embeddings", "encoder", "encoder_outputs", "flatten", "forward", "have", "hidden_states", "if", "input_dimensions", "is", "kwargs", "last_hidden_state", "layernorm", "not", "output", "output_hidden_states", "pixel_values", "pooled_output", "pooler", "pooler_output", "r", "raise", "reshaped_hidden_states", "return", "return_dict", "self", "sequence_output", "specify", "to", "torch", "transpose", "tuple", "use_return_dict"], "focalnet/modeling_focalnet.py:FocalNetForMaskedImageModeling.__init__": ["Conv2d", "False", "Model", "ModelForMaskedImageModeling", "ModelModel", "ModelPreTrainedModel", "PixelShuffle", "Sequential", "True", "__init__", "add_pooling_layer", "class", "config", "decoder", "def", "depths", "embed_dim", "encoder_stride", "in_channels", "int", "kernel_size", "len", "nn", "num_channels", "num_features", "num_stages", "out_channels", "post_init", "self", "super", "use_mask_token"], "focalnet/modeling_focalnet.py:FocalNetForMaskedImageModeling.forward": ["BoolTensor", "FloatTensor", "Model", "ModelForMaskedImageModeling", "ModelMaskedImageModelingOutput", "ModelPreTrainedModel", "None", "batch_size", "bool", "bool_masked_pos", "class", "config", "contiguous", "decoder", "def", "else", "floor", "forward", "functional", "height", "hidden_states", "if", "image_size", "is", "kwargs", "l1_loss", "loss", "mask", "masked_im_loss", "math", "nn", "none", "not", "num_channels", "output", "output_hidden_states", "outputs", "patch_size", "pixel_values", "r", "reconstructed_pixel_values", "reconstruction", "reconstruction_loss", "reduction", "repeat_interleave", "reshape", "reshaped_hidden_states", "return", "return_dict", "self", "sequence_length", "sequence_output", "shape", "size", "sum", "torch", "transpose", "tuple", "unsqueeze", "use_return_dict", "width"], "focalnet/modeling_focalnet.py:FocalNetForImageClassification.__init__": ["Identity", "Linear", "Model", "ModelForImageClassification", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "classifier", "config", "def", "else", "if", "nn", "num_features", "num_labels", "post_init", "self", "super"], "focalnet/modeling_focalnet.py:FocalNetForImageClassification.forward": ["FloatTensor", "LongTensor", "Model", "ModelForImageClassification", "ModelImageClassifierOutput", "ModelPreTrainedModel", "None", "bool", "class", "classifier", "config", "def", "else", "forward", "hidden_states", "if", "is", "kwargs", "labels", "logits", "loss", "loss_function", "not", "output", "output_hidden_states", "outputs", "pixel_values", "pooled_output", "r", "reshaped_hidden_states", "return", "return_dict", "self", "torch", "tuple", "use_return_dict"], "focalnet/modeling_focalnet.py:FocalNetBackbone.__init__": ["BackboneMixin", "Model", "ModelBackbone", "ModelConfig", "ModelModel", "ModelPreTrainedModel", "__init__", "_init_backbone", "class", "config", "def", "embed_dim", "hidden_sizes", "num_features", "post_init", "self", "super"], "focalnet/modeling_focalnet.py:FocalNetBackbone.forward": ["BackboneMixin", "BackboneOutput", "Model", "ModelBackbone", "ModelPreTrainedModel", "None", "Tensor", "True", "attentions", "bool", "class", "config", "def", "else", "enumerate", "feature_maps", "for", "forward", "hidden_states", "idx", "if", "in", "is", "kwargs", "not", "out_features", "output", "output_hidden_states", "outputs", "pixel_values", "r", "reshaped_hidden_states", "return", "return_dict", "self", "stage", "stage_names", "torch", "use_return_dict"], "fsmt/modeling_fsmt.py:invert_mask": ["Model_mask", "assert", "attention_mask", "def", "dim", "eq", "return"], "fsmt/modeling_fsmt.py:triu_onnx": ["Model_onnx", "arange", "def", "device", "diagonal", "expand", "if", "l", "mask", "masked_fill", "return", "shape", "torch", "unsqueeze", "x"], "fsmt/modeling_fsmt.py:_prepare_fsmt_decoder_inputs": ["None", "_prepare_Model_decoder_inputs", "bsz", "causal_mask", "causal_mask_dtype", "config", "decoder_input_ids", "decoder_padding_mask", "def", "device", "dtype", "else", "fill_with_neg_inf", "float32", "if", "input_ids", "invert_mask", "is", "make_padding_mask", "pad_token_id", "return", "shift_tokens_right", "size", "tgt_len", "to", "torch", "triu_onnx", "zeros"], "fsmt/modeling_fsmt.py:PretrainedFSMTModel._init_weights": ["Embedding", "False", "Linear", "ModelModel", "ModelModelModel", "None", "SinusoidalPositionalEmbedding", "_init_weights", "_is_hf_initialized", "and", "bias", "class", "config", "copy_", "def", "elif", "get_embedding", "getattr", "if", "init", "init_std", "is", "isinstance", "mean", "module", "nn", "normal_", "not", "padding_idx", "self", "shape", "std", "weight", "zeros_"], "fsmt/modeling_fsmt.py:PretrainedFSMTModel.dummy_inputs": ["ModelModel", "ModelModelModel", "attention_mask", "class", "config", "def", "device", "dummy_inputs", "input_ids", "ne", "pad_token", "pad_token_id", "return", "self", "tensor", "torch"], "fsmt/modeling_fsmt.py:_make_linear_from_emb": ["False", "Linear", "_make_linear_from_emb", "bias", "data", "def", "emb", "emb_size", "lin_layer", "nn", "return", "shape", "vocab_size", "weight"], "fsmt/modeling_fsmt.py:_check_shapes": ["AssertionError", "_check_shapes", "def", "f", "if", "mismatch", "raise", "shape", "shape2", "shape_1"], "fsmt/modeling_fsmt.py:shift_tokens_right": ["Model_tokens_right", "clone", "def", "dim", "gather", "index_of_eos", "input_ids", "masked_fill_", "ne", "pad_token_id", "prev_output_tokens", "return", "squeeze", "sum", "unsqueeze"], "fsmt/modeling_fsmt.py:make_padding_mask": ["Model_padding_mask", "None", "any", "def", "eq", "if", "input_ids", "not", "padding_idx", "padding_mask", "return"], "fsmt/modeling_fsmt.py:EncoderLayer.__init__": ["ACT2FN", "Attention", "LayerNorm", "Linear", "ModelConfig", "ModelLayer", "Model_attention_heads", "Model_ffn_dim", "Module", "__init__", "activation_dropout", "activation_fn", "activation_function", "attention_dropout", "class", "config", "d_model", "def", "dropout", "embed_dim", "fc1", "fc2", "final_layer_norm", "nn", "self", "self_attn", "self_attn_layer_norm", "super"], "fsmt/modeling_fsmt.py:EncoderLayer.forward": ["False", "ModelLayer", "Model_padding_mask", "Module", "activation_dropout", "activation_fn", "attn_weights", "class", "def", "dropout", "fc1", "fc2", "final_layer_norm", "forward", "functional", "key", "key_padding_mask", "nn", "output_attentions", "p", "query", "residual", "return", "self", "self_attn", "self_attn_layer_norm", "training", "x"], "fsmt/modeling_fsmt.py:FSMTEncoder.__init__": ["Embedding", "EncoderLayer", "Model", "ModelConfig", "Module", "ModuleList", "SinusoidalPositionalEmbedding", "_", "__init__", "class", "config", "d_model", "def", "dropout", "else", "embed_dim", "embed_positions", "embed_scale", "embed_tokens", "embedding_dim", "encoder_layerdrop", "encoder_layers", "for", "if", "in", "layerdrop", "layers", "math", "max_position_embeddings", "nn", "pad_token_id", "padding_idx", "range", "scale_embedding", "self", "sqrt", "src_vocab_size", "super"], "fsmt/modeling_fsmt.py:FSMTEncoder.forward": ["BaseModelOutput", "False", "Model", "Module", "None", "Tensor", "True", "ValueError", "You", "all_attentions", "and", "at", "attention_mask", "attentions", "attn", "bool", "both", "cannot", "class", "def", "dropout", "dropout_probability", "either", "elif", "else", "embed_pos", "embed_positions", "embed_scale", "embed_tokens", "encoder_layer", "encoder_states", "enumerate", "eq", "for", "forward", "functional", "have", "hidden_states", "idx", "if", "in", "input_ids", "inputs_embeds", "invert_mask", "is", "last_hidden_state", "layerdrop", "layers", "masked_fill", "nn", "not", "or", "output_attentions", "output_hidden_states", "p", "padding_idx", "position_ids", "raise", "rand", "return", "return_dict", "same", "self", "specify", "the", "time", "to", "torch", "training", "transpose", "tuple", "v", "x"], "fsmt/modeling_fsmt.py:DecoderLayer.__init__": ["ACT2FN", "Attention", "LayerNorm", "Linear", "ModelConfig", "ModelLayer", "Model_attention_heads", "Model_ffn_dim", "Module", "None", "True", "__init__", "activation_dropout", "activation_fn", "activation_function", "attention_dropout", "class", "config", "d_model", "def", "dropout", "embed_dim", "encoder_Model_attention", "encoder_attn", "encoder_attn_layer_norm", "fc1", "fc2", "final_layer_norm", "layer_idx", "nn", "num_heads", "self", "self_attn", "self_attn_layer_norm", "super"], "fsmt/modeling_fsmt.py:DecoderLayer.forward": ["False", "ModelLayer", "Model_padding_mask", "Module", "None", "activation_dropout", "activation_fn", "assert", "attn_mask", "cache_key", "cache_position", "causal_mask", "class", "cross_attn_weights", "def", "dropout", "encoder_attn", "encoder_attn_layer_norm", "encoder_attn_mask", "encoder_hidden_states", "fc1", "fc2", "final_layer_norm", "forward", "functional", "key", "key_padding_mask", "layer_state", "nn", "output_attentions", "p", "query", "residual", "return", "self", "self_attn", "self_attn_layer_norm", "self_attn_weights", "training", "x"], "fsmt/modeling_fsmt.py:FSMTDecoder.__init__": ["DecoderLayer", "Embedding", "False", "Linear", "Model", "ModelConfig", "Module", "ModuleList", "SinusoidalPositionalEmbedding", "__init__", "bias", "class", "config", "d_model", "decoder_layerdrop", "decoder_layers", "def", "dropout", "else", "embed_dim", "embed_positions", "embed_scale", "embed_tokens", "embedding_dim", "for", "i", "if", "in", "layer_idx", "layerdrop", "layers", "math", "max_position_embeddings", "nn", "output_projection", "pad_token_id", "padding_idx", "range", "scale_embedding", "self", "sqrt", "super", "tgt_vocab_size"], "fsmt/modeling_fsmt.py:FSMTDecoder.forward": ["BaseModelOutputWithPastAndCrossAttentions", "Cache", "False", "Model", "Module", "None", "Tensor", "True", "ValueError", "You", "all_cross_attns", "all_hidden_states", "all_self_attns", "and", "at", "attentions", "bool", "both", "cache_position", "cannot", "causal_mask", "class", "continue", "cross_attentions", "decoder_causal_mask", "decoder_input_ids", "decoder_inputs_embeds", "decoder_layer", "decoder_padding_mask", "def", "dropout", "dropout_probability", "either", "elif", "else", "embed_positions", "embed_scale", "embed_tokens", "encoder_attn_mask", "encoder_hidden_states", "encoder_padding_mask", "enumerate", "eq", "for", "forward", "functional", "have", "hidden_states", "idx", "if", "in", "input_ids", "inputs_embeds", "invert_mask", "is", "last_hidden_state", "layer_cross_attn", "layer_self_attn", "layer_state", "layerdrop", "layers", "masked_fill", "nn", "not", "or", "output_attentions", "output_hidden_states", "output_projection", "p", "padding_idx", "past_key_values", "position_ids", "positions", "raise", "rand", "return", "return_dict", "same", "self", "specify", "the", "time", "to", "torch", "training", "transpose", "tuple", "use_cache", "v", "x"], "fsmt/modeling_fsmt.py:_reorder_buffer": ["None", "_reorder_buffer", "attn_cache", "def", "for", "if", "in", "index_select", "input_buffer_k", "is", "items", "k", "new_order", "not", "return"], "fsmt/modeling_fsmt.py:Attention.__init__": ["False", "Linear", "Model", "Module", "None", "True", "__init__", "assert", "be", "bias", "by", "cache_key", "class", "def", "divisible", "dropout", "else", "embed_dim", "encoder_decoder", "encoder_decoder_Model", "head_dim", "if", "k_proj", "layer_idx", "must", "nn", "num_heads", "out_proj", "q_proj", "scaling", "self", "super", "v_proj"], "fsmt/modeling_fsmt.py:Attention.forward": ["Cache", "EncoderDecoderCache", "False", "Model", "Module", "None", "Tensor", "True", "and", "assert", "attn_mask", "attn_output", "attn_probs", "attn_weights", "attn_weights_reshaped", "bmm", "bool", "bsz", "cache_position", "class", "contiguous", "cross_Model_cache", "curr_past_key_values", "current_states", "def", "dim", "dropout", "dtype", "else", "embed_dim", "encoder_decoder_Model", "finfo", "forward", "functional", "get", "head_dim", "if", "is", "is_updated", "isinstance", "k_proj", "key", "key_padding_mask", "key_states", "keys", "layer_idx", "layer_state", "layers", "list", "masked_fill", "min", "nn", "not", "num_heads", "or", "out_proj", "output_Models", "p", "permute", "q_proj", "query", "query_states", "reshape", "reshaped", "return", "scaling", "self", "self_Model_cache", "size", "softmax", "src_len", "tgt_len", "torch", "training", "transpose", "tuple", "unsqueeze", "update", "v_proj", "value_states", "values", "view"], "fsmt/modeling_fsmt.py:fill_with_neg_inf": ["Model_", "Model_with_neg_inf", "def", "dtype", "finfo", "float", "min", "return", "t", "torch", "type_as"], "fsmt/modeling_fsmt.py:_get_shape": ["None", "_get_shape", "def", "getattr", "return", "shape", "t"], "fsmt/modeling_fsmt.py:FSMTModel.__init__": ["Model", "ModelConfig", "ModelDecoder", "ModelEncoder", "PretrainedModel", "__init__", "class", "config", "decoder", "def", "encoder", "post_init", "self", "super"], "fsmt/modeling_fsmt.py:FSMTModel.forward": ["BaseModelOutput", "BoolTensor", "Cache", "DynamicCache", "EncoderDecoderCache", "False", "FloatTensor", "LongTensor", "Make", "Model", "None", "PretrainedModel", "Seq2SeqModelOutput", "Tensor", "ValueError", "_prepare_Model_decoder_inputs", "and", "are", "attention_mask", "attentions", "bool", "cache_position", "causal_mask", "causal_mask_dtype", "class", "config", "cross_attentions", "decoder", "decoder_attention_mask", "decoder_attentions", "decoder_causal_mask", "decoder_hidden_states", "decoder_input_ids", "decoder_inputs_embeds", "decoder_outputs", "decoder_padding_mask", "def", "dtype", "elif", "else", "embed_tokens", "encoder", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "isinstance", "kwargs", "last_hidden_state", "len", "not", "or", "output_attentions", "output_hidden_states", "passed", "past_key_values", "r", "raise", "return", "return_dict", "self", "sure", "that", "torch", "tuple", "use_cache", "use_return_dict", "weight"], "fsmt/modeling_fsmt.py:FSMTModel.get_input_embeddings": ["Model", "PretrainedModel", "class", "def", "embed_tokens", "encoder", "get_input_embeddings", "return", "self"], "fsmt/modeling_fsmt.py:FSMTModel.set_input_embeddings": ["Model", "PretrainedModel", "class", "def", "embed_tokens", "encoder", "self", "set_input_embeddings", "value"], "fsmt/modeling_fsmt.py:FSMTModel.get_output_embeddings": ["Model", "PretrainedModel", "class", "decoder", "def", "embed_tokens", "get_output_embeddings", "return", "self"], "fsmt/modeling_fsmt.py:FSMTModel.set_output_embeddings": ["Model", "PretrainedModel", "class", "decoder", "def", "embed_tokens", "self", "set_output_embeddings", "value"], "fsmt/modeling_fsmt.py:FSMTForConditionalGeneration.__init__": ["GenerationMixin", "Model", "ModelConfig", "ModelModel", "PretrainedModelModel", "__init__", "base_model", "class", "config", "def", "model", "post_init", "self", "super"], "fsmt/modeling_fsmt.py:FSMTForConditionalGeneration.forward": ["BoolTensor", "Cache", "CrossEntropyLoss", "False", "FloatTensor", "GenerationMixin", "LongTensor", "Model", "None", "PretrainedModelModel", "Seq2SeqLMOutput", "Tensor", "attention_mask", "bool", "cache_position", "class", "config", "cross_attentions", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_input_ids", "decoder_inputs_embeds", "def", "else", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "forward", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "lm_logits", "logits", "loss", "loss_fct", "masked_lm_loss", "model", "not", "output", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "r", "return", "return_dict", "self", "tgt_vocab_size", "torch", "tuple", "use_cache", "use_return_dict", "view"], "fsmt/modeling_fsmt.py:FSMTForConditionalGeneration.prepare_decoder_input_ids_from_labels": ["GenerationMixin", "Model", "PretrainedModelModel", "Tensor", "class", "config", "def", "labels", "pad_token_id", "prepare_decoder_input_ids_from_labels", "return", "self", "shift_tokens_right", "torch"], "fsmt/modeling_fsmt.py:FSMTForConditionalGeneration.get_output_embeddings": ["GenerationMixin", "Model", "PretrainedModelModel", "class", "decoder", "def", "embed_tokens", "get_output_embeddings", "model", "return", "self"], "fsmt/modeling_fsmt.py:FSMTForConditionalGeneration.set_output_embeddings": ["GenerationMixin", "Model", "PretrainedModelModel", "class", "decoder", "def", "embed_tokens", "model", "self", "set_output_embeddings", "value"], "fsmt/modeling_fsmt.py:SinusoidalPositionalEmbedding.__init__": ["Embedding", "ModelPositionalEmbedding", "__init__", "class", "def", "embedding_dim", "nn", "num_positions", "padding_idx", "self", "super"], "fsmt/modeling_fsmt.py:SinusoidalPositionalEmbedding.make_weight": ["Embedding", "False", "ModelPositionalEmbedding", "Parameter", "class", "def", "detach_", "device", "dtype", "embedding_dim", "get_embedding", "make_weight", "nn", "num_positions", "padding_idx", "requires_grad", "self", "to", "weight"], "fsmt/modeling_fsmt.py:SinusoidalPositionalEmbedding.get_embedding": ["Embedding", "ModelPositionalEmbedding", "None", "arange", "cat", "class", "cos", "def", "dim", "dtype", "emb", "embedding_dim", "exp", "float", "get_embedding", "half_dim", "if", "int64", "is", "log", "math", "nn", "not", "num_embeddings", "padding_idx", "return", "sin", "torch", "unsqueeze", "view", "zeros"], "fsmt/modeling_fsmt.py:SinusoidalPositionalEmbedding.make_positions": ["Embedding", "ModelPositionalEmbedding", "class", "cumsum", "def", "dim", "int", "long", "make_positions", "mask", "ne", "nn", "padding_idx", "return", "tensor", "torch", "type_as"], "fsmt/modeling_fsmt.py:SinusoidalPositionalEmbedding.forward": ["Any", "Embedding", "ModelPositionalEmbedding", "None", "Tensor", "bsz", "class", "def", "embedding_dim", "forward", "if", "incremental_state", "input", "make_positions", "make_weight", "max_pos", "nn", "padding_idx", "positions", "return", "self", "seq_len", "shape", "size", "super", "timestep", "weight"], "funnel/modeling_funnel.py:FunnelEmbeddings.__init__": ["Dropout", "Embedding", "LayerNorm", "ModelConfig", "ModelEmbeddings", "Module", "None", "__init__", "class", "config", "d_model", "def", "dropout", "eps", "hidden_dropout", "hidden_size", "layer_norm", "layer_norm_eps", "nn", "pad_token_id", "padding_idx", "self", "super", "vocab_size", "word_embeddings"], "funnel/modeling_funnel.py:FunnelEmbeddings.forward": ["ModelEmbeddings", "Module", "None", "Tensor", "class", "def", "dropout", "embeddings", "forward", "if", "input_ids", "inputs_embeds", "is", "layer_norm", "nn", "return", "self", "torch", "word_embeddings"], "funnel/modeling_funnel.py:FunnelAttentionStructure.__init__": ["Dropout", "ModelAttentionStructure", "ModelConfig", "Module", "None", "__init__", "class", "config", "cos_dropout", "def", "hidden_dropout", "nn", "pooling_mult", "self", "sin_dropout", "super"], "funnel/modeling_funnel.py:FunnelAttentionStructure.init_attention_inputs": ["ModelAttentionStructure", "Module", "None", "Tensor", "attention_mask", "class", "cls_mask", "config", "def", "device", "dtype", "else", "functional", "get_position_embeds", "if", "init_attention_inputs", "inputs_embeds", "is", "new_ones", "nn", "not", "pad", "pooling_mult", "position_embeds", "return", "self", "separate_cls", "seq_len", "size", "token_type_ids", "token_type_ids_to_mat", "token_type_mat", "torch", "tuple"], "funnel/modeling_funnel.py:FunnelAttentionStructure.token_type_ids_to_mat": ["ModelAttentionStructure", "Module", "None", "Tensor", "class", "cls_ids", "cls_mat", "cls_token_type_id", "def", "nn", "return", "self", "token_type_ids", "token_type_ids_to_mat", "token_type_mat", "torch"], "funnel/modeling_funnel.py:FunnelAttentionStructure.get_position_embeds": ["ModelAttentionStructure", "Module", "None", "Tensor", "append", "arange", "attention_type", "block_index", "cat", "class", "config", "cos", "cos_dropout", "cos_embed", "cos_embed_d", "d_model", "def", "device", "dim", "dtype", "else", "expand", "factorized", "for", "freq_seq", "gather", "get_position_embeds", "if", "in", "int", "int64", "inv_freq", "list", "nn", "num_blocks", "omega", "phi", "pi", "pooled_pos", "pos", "pos_embed", "pos_seq", "position_embeds_list", "position_embeds_no_pooling", "position_embeds_pooling", "psi", "range", "rel_pos", "rel_pos_id", "relative_pos", "return", "self", "seq_len", "shift", "sin", "sin_dropout", "sin_embed", "sin_embed_d", "sinusoid", "size", "stride", "stride_pool_pos", "to", "torch", "tuple", "zero_offset"], "funnel/modeling_funnel.py:FunnelAttentionStructure.stride_pool_pos": ["ModelAttentionStructure", "Module", "Tensor", "block_index", "cat", "class", "cls_pos", "config", "def", "else", "if", "int", "new_tensor", "nn", "pooled_pos_id", "pos_id", "return", "self", "separate_cls", "stride_pool_pos", "torch", "truncate_seq"], "funnel/modeling_funnel.py:FunnelAttentionStructure.relative_pos": ["ModelAttentionStructure", "Module", "None", "Tensor", "arange", "class", "def", "device", "dtype", "if", "int", "is", "len", "long", "max_dist", "min_dist", "nn", "num_remove", "pooled_pos", "pos", "ref_point", "relative_pos", "return", "self", "shift", "stride", "torch"], "funnel/modeling_funnel.py:FunnelAttentionStructure.stride_pool": ["ModelAttentionStructure", "Module", "None", "Tensor", "and", "ax", "axis", "axis_slice", "cat", "class", "cls_slice", "config", "def", "else", "enc_slice", "for", "if", "in", "int", "is", "isinstance", "list", "ndim", "nn", "return", "self", "separate_cls", "slice", "stride_pool", "tensor", "torch", "truncate_seq", "tuple", "type", "x"], "funnel/modeling_funnel.py:FunnelAttentionStructure.pool_tensor": ["ModelAttentionStructure", "Module", "None", "NotImplementedError", "Tensor", "The", "True", "and", "are", "avg_pool2d", "cat", "ceil_mode", "class", "config", "def", "dim", "elif", "else", "for", "functional", "if", "in", "int", "is", "isinstance", "list", "max", "max_pool2d", "mean", "min", "mode", "modes", "ndim", "nn", "pool_tensor", "raise", "return", "self", "separate_cls", "str", "stride", "suffix", "supported", "tensor", "torch", "truncate_seq", "tuple", "type", "x"], "funnel/modeling_funnel.py:FunnelAttentionStructure.pre_attention_pooling": ["ModelAttentionStructure", "Module", "Tensor", "attention_inputs", "attention_mask", "attention_type", "class", "cls_mask", "config", "def", "else", "factorized", "if", "min", "mode", "nn", "output", "pool_q_only", "pool_tensor", "pooling_mult", "pooling_type", "position_embeds", "pre_attention_pooling", "return", "self", "stride_pool", "token_type_mat", "torch", "tuple"], "funnel/modeling_funnel.py:FunnelAttentionStructure.post_attention_pooling": ["ModelAttentionStructure", "Module", "Tensor", "attention_inputs", "attention_mask", "attention_type", "class", "cls_mask", "config", "def", "factorized", "if", "min", "mode", "nn", "pool_q_only", "pool_tensor", "pooling_mult", "position_embeds", "post_attention_pooling", "return", "self", "stride_pool", "token_type_mat", "torch", "tuple"], "funnel/modeling_funnel.py:_relative_shift_gather": ["Tensor", "_relative_shift_gather", "batch_size", "context_len", "def", "int", "max_rel_len", "n_head", "positional_attn", "reshape", "return", "seq_len", "shape", "shift", "torch"], "funnel/modeling_funnel.py:FunnelRelMultiheadAttention.__init__": ["Dropout", "False", "LayerNorm", "Linear", "ModelConfig", "ModelRelMultiheadAttention", "Module", "None", "Parameter", "__init__", "attention_dropout", "bias", "block_index", "class", "config", "d_head", "d_model", "def", "eps", "hidden_dropout", "int", "k_head", "layer_norm", "layer_norm_eps", "n_head", "nn", "post_proj", "q_head", "r_kernel", "r_r_bias", "r_s_bias", "r_w_bias", "scale", "seg_embed", "self", "super", "torch", "v_head", "zeros"], "funnel/modeling_funnel.py:FunnelRelMultiheadAttention.relative_positional_attention": ["ModelRelMultiheadAttention", "Module", "None", "_relative_shift_gather", "attention_type", "bind", "binh", "block_index", "bnij", "bnit", "class", "cls_mask", "config", "context_len", "def", "dnh", "einsum", "else", "factorized", "if", "is", "jd", "nn", "not", "omega", "phi", "pi", "position_embeds", "positional_attn", "psi", "q_head", "q_r_attention", "q_r_attention_1", "q_r_attention_2", "r", "r_head", "r_kernel", "r_r_bias", "relative_positional_attention", "return", "scale", "self", "shape", "shift", "td", "tnh", "torch", "u", "v", "w_r"], "funnel/modeling_funnel.py:FunnelRelMultiheadAttention.relative_token_type_attention": ["ModelRelMultiheadAttention", "Module", "None", "batch_size", "bind", "bnis", "class", "cls_mask", "context_len", "def", "diff_token_type", "dim", "einsum", "expand", "if", "is", "nn", "not", "q_head", "r_s_bias", "relative_token_type_attention", "return", "same_token_type", "scale", "seg_embed", "self", "seq_len", "shape", "snd", "split", "token_type_attn", "token_type_bias", "token_type_mat", "torch", "where"], "funnel/modeling_funnel.py:FunnelRelMultiheadAttention.forward": ["False", "INF", "ModelRelMultiheadAttention", "Module", "None", "Tensor", "_", "attention_dropout", "attention_inputs", "attention_mask", "attn_out", "attn_prob", "attn_score", "attn_vec", "batch_size", "bind", "bjnd", "bnij", "bool", "class", "cls_mask", "config", "content_score", "context_len", "d_head", "def", "dim", "dtype", "einsum", "else", "float", "forward", "hidden_dropout", "if", "is", "k_head", "key", "layer_norm", "n_head", "nn", "not", "output", "output_attentions", "position_embeds", "positional_attn", "post_proj", "q_head", "query", "r_w_bias", "relative_positional_attention", "relative_token_type_attention", "reshape", "return", "scale", "self", "seq_len", "shape", "softmax", "token_type_attn", "token_type_mat", "torch", "tuple", "v_head", "value", "view"], "funnel/modeling_funnel.py:FunnelPositionwiseFFN.__init__": ["ACT2FN", "Dropout", "LayerNorm", "Linear", "ModelConfig", "ModelPositionwiseFFN", "Module", "None", "__init__", "activation_dropout", "activation_function", "class", "config", "d_inner", "d_model", "def", "dropout", "hidden_act", "hidden_dropout", "layer_norm", "layer_norm_eps", "linear_1", "linear_2", "nn", "self", "super"], "funnel/modeling_funnel.py:FunnelPositionwiseFFN.forward": ["ModelPositionwiseFFN", "Module", "Tensor", "activation_dropout", "activation_function", "class", "def", "dropout", "forward", "h", "hidden", "layer_norm", "linear_1", "linear_2", "nn", "return", "self", "torch"], "funnel/modeling_funnel.py:FunnelLayer.__init__": ["ModelConfig", "ModelLayer", "ModelPositionwiseFFN", "ModelRelMultiheadAttention", "Module", "None", "__init__", "attention", "block_index", "class", "config", "def", "ffn", "int", "nn", "self", "super"], "funnel/modeling_funnel.py:FunnelLayer.forward": ["False", "ModelLayer", "Module", "Tensor", "attention", "attention_inputs", "attn", "bool", "class", "def", "else", "ffn", "forward", "if", "key", "nn", "output", "output_attentions", "query", "return", "self", "torch", "tuple", "value"], "funnel/modeling_funnel.py:FunnelEncoder.__init__": ["ModelAttentionStructure", "ModelConfig", "ModelEncoder", "ModelLayer", "Module", "ModuleList", "None", "_", "__init__", "attention_structure", "block_index", "block_size", "block_sizes", "blocks", "class", "config", "def", "enumerate", "for", "in", "nn", "range", "self", "super"], "funnel/modeling_funnel.py:FunnelEncoder.forward": ["BaseModelOutput", "False", "ModelEncoder", "Module", "None", "Tensor", "True", "all_attentions", "all_hidden_states", "and", "attention_inputs", "attention_mask", "attention_structure", "attentions", "block", "block_index", "block_repeats", "blocks", "bool", "class", "config", "def", "do_pooling", "else", "enumerate", "for", "forward", "hidden", "hidden_states", "if", "in", "init_attention_inputs", "inputs_embeds", "is", "key", "last_hidden_state", "layer", "layer_index", "layer_output", "nn", "not", "output_attentions", "output_hidden_states", "pool_q_only", "pooled_hidden", "pooling_flag", "post_attention_pooling", "pre_attention_pooling", "query", "range", "repeat_index", "return", "return_dict", "self", "separate_cls", "size", "token_type_ids", "torch", "tuple", "type_as", "v", "value"], "funnel/modeling_funnel.py:upsample": ["False", "Model", "Tensor", "True", "bool", "cat", "cls", "def", "dim", "else", "functional", "if", "int", "nn", "output", "pad", "repeat_interleave", "repeats", "return", "separate_cls", "stride", "target_len", "torch", "truncate_seq", "x"], "funnel/modeling_funnel.py:FunnelDecoder.__init__": ["ModelAttentionStructure", "ModelConfig", "ModelDecoder", "ModelLayer", "Module", "ModuleList", "None", "_", "__init__", "attention_structure", "class", "config", "def", "for", "in", "layers", "nn", "num_decoder_layers", "range", "self", "super"], "funnel/modeling_funnel.py:FunnelDecoder.forward": ["BaseModelOutput", "False", "ModelDecoder", "Module", "None", "Tensor", "True", "all_attentions", "all_hidden_states", "attention_inputs", "attention_mask", "attention_structure", "attentions", "block_sizes", "bool", "class", "config", "def", "else", "final_hidden", "first_block_hidden", "for", "forward", "hidden", "hidden_states", "if", "in", "init_attention_inputs", "is", "last_hidden_state", "layer", "layer_output", "layers", "len", "nn", "not", "output_attentions", "output_hidden_states", "return", "return_dict", "self", "separate_cls", "shape", "stride", "target_len", "token_type_ids", "torch", "truncate_seq", "tuple", "upsample", "upsampled_hidden", "v"], "funnel/modeling_funnel.py:FunnelDiscriminatorPredictions.__init__": ["Linear", "ModelConfig", "ModelDiscriminatorPredictions", "Module", "None", "__init__", "class", "config", "d_model", "def", "dense", "dense_prediction", "nn", "self", "super"], "funnel/modeling_funnel.py:FunnelDiscriminatorPredictions.forward": ["ACT2FN", "ModelDiscriminatorPredictions", "Module", "Tensor", "class", "config", "def", "dense", "dense_prediction", "discriminator_hidden_states", "forward", "hidden_act", "hidden_states", "logits", "nn", "return", "self", "squeeze", "torch"], "funnel/modeling_funnel.py:FunnelPreTrainedModel._init_weights": ["Linear", "ModelEmbeddings", "ModelPreTrainedModel", "ModelRelMultiheadAttention", "None", "PreTrainedModel", "__class__", "__name__", "_init_weights", "b", "bias", "class", "classname", "config", "constant_", "def", "elif", "else", "fan_in", "fan_out", "find", "float", "getattr", "if", "init", "initializer_range", "initializer_std", "is", "module", "normal_", "not", "np", "padding_idx", "r_kernel", "r_r_bias", "r_s_bias", "r_w_bias", "seg_embed", "self", "shape", "sqrt", "std", "uniform_", "weight", "word_embeddings", "zeros_"], "funnel/modeling_funnel.py:FunnelClassificationHead.__init__": ["Dropout", "Linear", "ModelClassificationHead", "ModelConfig", "Module", "None", "__init__", "class", "config", "d_model", "def", "dropout", "hidden_dropout", "int", "linear_hidden", "linear_out", "n_labels", "nn", "self", "super"], "funnel/modeling_funnel.py:FunnelClassificationHead.forward": ["ModelClassificationHead", "Module", "Tensor", "class", "def", "dropout", "forward", "hidden", "linear_hidden", "linear_out", "nn", "return", "self", "tanh", "torch"], "funnel/modeling_funnel.py:FunnelBaseModel.__init__": ["ModelBaseModel", "ModelConfig", "ModelEmbeddings", "ModelEncoder", "ModelPreTrainedModel", "None", "__init__", "class", "config", "def", "embeddings", "encoder", "post_init", "self", "super"], "funnel/modeling_funnel.py:FunnelBaseModel.get_input_embeddings": ["Embedding", "ModelBaseModel", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "nn", "return", "self", "word_embeddings"], "funnel/modeling_funnel.py:FunnelBaseModel.set_input_embeddings": ["Embedding", "ModelBaseModel", "ModelPreTrainedModel", "None", "class", "def", "embeddings", "new_embeddings", "nn", "self", "set_input_embeddings", "word_embeddings"], "funnel/modeling_funnel.py:FunnelBaseModel.forward": ["BaseModelOutput", "ModelBaseModel", "ModelPreTrainedModel", "None", "Tensor", "ValueError", "You", "and", "at", "attention_mask", "bool", "both", "cannot", "class", "config", "def", "device", "dtype", "either", "elif", "else", "embeddings", "encoder", "encoder_outputs", "forward", "have", "if", "input_ids", "input_shape", "inputs_embeds", "is", "kwargs", "long", "not", "ones", "or", "output_attentions", "output_hidden_states", "position_ids", "raise", "return", "return_dict", "same", "self", "size", "specify", "the", "time", "to", "token_type_ids", "torch", "tuple", "use_return_dict", "warn_if_padding_and_no_attention_mask", "zeros"], "funnel/modeling_funnel.py:FunnelModel.__init__": ["ModelConfig", "ModelDecoder", "ModelEmbeddings", "ModelEncoder", "ModelModel", "ModelPreTrainedModel", "None", "__init__", "class", "config", "decoder", "def", "embeddings", "encoder", "post_init", "self", "super"], "funnel/modeling_funnel.py:FunnelModel.get_input_embeddings": ["Embedding", "ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "nn", "return", "self", "word_embeddings"], "funnel/modeling_funnel.py:FunnelModel.set_input_embeddings": ["Embedding", "ModelModel", "ModelPreTrainedModel", "None", "class", "def", "embeddings", "new_embeddings", "nn", "self", "set_input_embeddings", "word_embeddings"], "funnel/modeling_funnel.py:FunnelModel.forward": ["BaseModelOutput", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "True", "ValueError", "You", "and", "at", "attention_mask", "attentions", "block_sizes", "bool", "both", "cannot", "class", "config", "decoder", "decoder_outputs", "def", "device", "dtype", "either", "elif", "else", "embeddings", "encoder", "encoder_outputs", "final_hidden", "first_block_hidden", "forward", "have", "hidden_states", "idx", "if", "input_ids", "input_shape", "inputs_embeds", "is", "kwargs", "last_hidden_state", "long", "not", "ones", "or", "output_attentions", "output_hidden_states", "outputs", "raise", "return", "return_dict", "same", "self", "size", "specify", "the", "time", "to", "token_type_ids", "torch", "tuple", "use_return_dict", "warn_if_padding_and_no_attention_mask", "zeros"], "funnel/modeling_funnel.py:FunnelForPreTraining.__init__": ["Model", "ModelConfig", "ModelDiscriminatorPredictions", "ModelForPreTraining", "ModelModel", "ModelPreTrainedModel", "None", "__init__", "class", "config", "def", "discriminator_predictions", "post_init", "self", "super"], "funnel/modeling_funnel.py:FunnelForPreTraining.forward": ["BCEWithLogitsLoss", "Model", "ModelForPreTraining", "ModelForPreTrainingOutput", "ModelPreTrainedModel", "None", "Tensor", "active_labels", "active_logits", "active_loss", "attention_mask", "attentions", "bool", "class", "config", "def", "discriminator_hidden_states", "discriminator_predictions", "discriminator_sequence_output", "else", "float", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "nn", "not", "output", "output_attentions", "output_hidden_states", "r", "return", "return_dict", "self", "shape", "token_type_ids", "torch", "tuple", "use_return_dict", "view"], "funnel/modeling_funnel.py:FunnelForMaskedLM.__init__": ["Linear", "Model", "ModelConfig", "ModelForMaskedLM", "ModelModel", "ModelPreTrainedModel", "None", "__init__", "class", "config", "d_model", "def", "lm_head", "nn", "post_init", "self", "super", "vocab_size"], "funnel/modeling_funnel.py:FunnelForMaskedLM.get_output_embeddings": ["Linear", "ModelForMaskedLM", "ModelPreTrainedModel", "class", "def", "get_output_embeddings", "lm_head", "nn", "return", "self"], "funnel/modeling_funnel.py:FunnelForMaskedLM.set_output_embeddings": ["Embedding", "ModelForMaskedLM", "ModelPreTrainedModel", "None", "class", "def", "lm_head", "new_embeddings", "nn", "self", "set_output_embeddings"], "funnel/modeling_funnel.py:FunnelForMaskedLM.forward": ["CrossEntropyLoss", "MaskedLMOutput", "Model", "ModelForMaskedLM", "ModelPreTrainedModel", "None", "Tensor", "attention_mask", "attentions", "bool", "class", "config", "def", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "last_hidden_state", "lm_head", "logits", "loss", "loss_fct", "masked_lm_loss", "not", "output", "output_attentions", "output_hidden_states", "outputs", "prediction_logits", "r", "return", "return_dict", "self", "token_type_ids", "torch", "tuple", "use_return_dict", "view", "vocab_size"], "funnel/modeling_funnel.py:FunnelForSequenceClassification.__init__": ["Model", "ModelBaseModel", "ModelClassificationHead", "ModelConfig", "ModelForSequenceClassification", "ModelPreTrainedModel", "None", "__init__", "class", "classifier", "config", "def", "num_labels", "post_init", "self", "super"], "funnel/modeling_funnel.py:FunnelForSequenceClassification.forward": ["BCEWithLogitsLoss", "CrossEntropyLoss", "MSELoss", "Model", "ModelForSequenceClassification", "ModelPreTrainedModel", "None", "SequenceClassifierOutput", "Tensor", "and", "attention_mask", "attentions", "bool", "class", "classifier", "config", "def", "dtype", "elif", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "kwargs", "labels", "last_hidden_state", "logits", "long", "loss", "loss_fct", "multi_label_classification", "not", "num_labels", "or", "output", "output_attentions", "output_hidden_states", "outputs", "pooled_output", "problem_type", "r", "regression", "return", "return_dict", "self", "single_label_classification", "squeeze", "token_type_ids", "torch", "tuple", "use_return_dict", "view"], "funnel/modeling_funnel.py:FunnelForMultipleChoice.__init__": ["Model", "ModelBaseModel", "ModelClassificationHead", "ModelConfig", "ModelForMultipleChoice", "ModelPreTrainedModel", "None", "__init__", "class", "classifier", "config", "def", "post_init", "self", "super"], "funnel/modeling_funnel.py:FunnelForMultipleChoice.forward": ["CrossEntropyLoss", "Model", "ModelForMultipleChoice", "ModelPreTrainedModel", "MultipleChoiceModelOutput", "None", "Tensor", "attention_mask", "attentions", "bool", "class", "classifier", "config", "def", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "last_hidden_state", "logits", "loss", "loss_fct", "not", "num_choices", "output", "output_attentions", "output_hidden_states", "outputs", "pooled_output", "r", "reshaped_logits", "return", "return_dict", "self", "shape", "size", "token_type_ids", "torch", "tuple", "use_return_dict", "view"], "funnel/modeling_funnel.py:FunnelForTokenClassification.__init__": ["Dropout", "Linear", "Model", "ModelConfig", "ModelForTokenClassification", "ModelModel", "ModelPreTrainedModel", "None", "__init__", "class", "classifier", "config", "def", "dropout", "hidden_dropout", "hidden_size", "nn", "num_labels", "post_init", "self", "super"], "funnel/modeling_funnel.py:FunnelForTokenClassification.forward": ["CrossEntropyLoss", "Model", "ModelForTokenClassification", "ModelPreTrainedModel", "None", "Tensor", "TokenClassifierOutput", "attention_mask", "attentions", "bool", "class", "classifier", "config", "def", "dropout", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "last_hidden_state", "logits", "loss", "loss_fct", "not", "num_labels", "output", "output_attentions", "output_hidden_states", "outputs", "r", "return", "return_dict", "self", "token_type_ids", "torch", "tuple", "use_return_dict", "view"], "funnel/modeling_funnel.py:FunnelForQuestionAnswering.__init__": ["Linear", "Model", "ModelConfig", "ModelForQuestionAnswering", "ModelModel", "ModelPreTrainedModel", "None", "__init__", "class", "config", "def", "hidden_size", "nn", "num_labels", "post_init", "qa_outputs", "self", "super"], "funnel/modeling_funnel.py:FunnelForQuestionAnswering.forward": ["CrossEntropyLoss", "Model", "ModelForQuestionAnswering", "ModelPreTrainedModel", "None", "QuestionAnsweringModelOutput", "Tensor", "and", "attention_mask", "attentions", "bool", "clamp", "class", "config", "contiguous", "def", "dim", "else", "end_logits", "end_loss", "end_positions", "forward", "hidden_states", "if", "ignore_index", "ignored_index", "input_ids", "inputs_embeds", "is", "kwargs", "last_hidden_state", "len", "logits", "loss", "loss_fct", "not", "output", "output_attentions", "output_hidden_states", "outputs", "qa_outputs", "return", "return_dict", "self", "size", "split", "squeeze", "squeze", "start_logits", "start_loss", "start_positions", "token_type_ids", "torch", "total_loss", "tuple", "use_return_dict"], "fuyu/modeling_fuyu.py:FuyuModel.__init__": ["AutoModel", "False", "Linear", "ModelConfig", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "config", "def", "from_config", "gradient_checkpointing", "hidden_size", "language_model", "nn", "num_channels", "pad_token_id", "padding_idx", "patch_size", "post_init", "self", "super", "text_config", "vision_embed_tokens", "vocab_size"], "fuyu/modeling_fuyu.py:FuyuModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "language_model", "return", "self"], "fuyu/modeling_fuyu.py:FuyuModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "language_model", "self", "set_input_embeddings", "value"], "fuyu/modeling_fuyu.py:FuyuModel.gather_continuous_embeddings": ["Batch", "Got", "ModelModel", "ModelPreTrainedModel", "Number", "Tensor", "True", "ValueError", "and", "as_tuple", "batch", "batch_idx", "class", "clone", "continuous", "continuous_embeddings", "def", "device", "does", "dst_indices", "element", "embeddings", "f", "for", "gather_continuous_embeddings", "ids", "if", "image_patch_input_indices", "in", "len", "list", "match", "must", "nonzero", "not", "number", "of", "output_embeddings", "raise", "range", "return", "self", "shape", "sizes", "src_indices", "to", "token", "torch", "word_embeddings"], "fuyu/modeling_fuyu.py:FuyuModel.get_image_features": ["FloatTensor", "ModelModel", "ModelPreTrainedModel", "class", "def", "dtype", "for", "get_image_features", "in", "kwargs", "patch", "patch_embeddings", "pixel_values", "return", "self", "squeeze", "to", "torch", "vision_embed_tokens", "weight"], "fuyu/modeling_fuyu.py:FuyuModel.get_placeholder_mask": ["FloatTensor", "Image", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "ValueError", "all", "and", "class", "config", "def", "device", "do", "dtype", "else", "expand_as", "f", "features", "get_input_embeddings", "get_placeholder_mask", "if", "image", "image_features", "image_token_id", "input_ids", "inputs_embeds", "is", "long", "match", "n_image_features", "n_image_tokens", "not", "numel", "raise", "return", "self", "shape", "special_image_mask", "sum", "tensor", "to", "tokens", "torch", "unsqueeze"], "fuyu/modeling_fuyu.py:FuyuModel.forward": ["Cache", "CausalLMOutputWithPast", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "ValueError", "You", "_", "and", "arange", "at", "attention_mask", "batch_size", "bool", "both", "cannot", "cat", "class", "config", "def", "device", "dim", "dtype", "either", "elif", "else", "forward", "get_image_features", "get_input_embeddings", "get_placeholder_mask", "get_seq_length", "have", "if", "image_features", "image_patches", "image_patches_indices", "input_ids", "input_is", "inputs_embeds", "is", "kwargs", "language_model", "long", "masked_scatter", "not", "or", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "past_key_values_length", "patch_embeddings", "position_ids", "r", "raise", "return", "return_dict", "same", "self", "seq_length", "shape", "special_image_mask", "specify", "the", "time", "to", "torch", "tuple", "unsqueeze", "use_cache", "use_return_dict"], "fuyu/modeling_fuyu.py:FuyuForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "ModelConfig", "ModelForCausalLM", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "text_config", "vocab_size"], "fuyu/modeling_fuyu.py:FuyuForCausalLM.get_input_embeddings": ["GenerationMixin", "ModelForCausalLM", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "model", "return", "self"], "fuyu/modeling_fuyu.py:FuyuForCausalLM.set_input_embeddings": ["GenerationMixin", "ModelForCausalLM", "ModelPreTrainedModel", "class", "def", "model", "self", "set_input_embeddings", "value"], "fuyu/modeling_fuyu.py:FuyuForCausalLM.forward": ["Cache", "CausalLMOutputWithPast", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForCausalLM", "ModelPreTrainedModel", "None", "Tensor", "True", "attention_mask", "attentions", "bool", "class", "config", "def", "else", "forward", "hidden_states", "if", "image_patches", "image_patches_indices", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "position_ids", "r", "return", "return_dict", "self", "slice", "slice_indices", "text_config", "torch", "tuple", "use_cache", "use_return_dict", "vocab_size"], "fuyu/modeling_fuyu.py:FuyuForCausalLM.prepare_inputs_for_generation": ["False", "GenerationMixin", "ModelForCausalLM", "ModelPreTrainedModel", "None", "True", "and", "attention_mask", "cache_position", "class", "def", "get", "if", "image_patches", "image_patches_indices", "input_ids", "inputs_embeds", "is_first_iteration", "kwargs", "model_inputs", "not", "past_key_values", "prepare_inputs_for_generation", "return", "self", "super", "use_cache"], "gemma/modeling_gemma.py:GemmaRMSNorm.__init__": ["ModelRMSNorm", "Module", "Parameter", "__init__", "class", "def", "dim", "eps", "float", "int", "nn", "self", "super", "torch", "weight", "zeros"], "gemma/modeling_gemma.py:GemmaRMSNorm._norm": ["ModelRMSNorm", "Module", "True", "_norm", "class", "def", "eps", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "torch", "x"], "gemma/modeling_gemma.py:GemmaRMSNorm.forward": ["ModelRMSNorm", "Module", "_norm", "class", "def", "float", "forward", "nn", "output", "return", "self", "type_as", "weight", "x"], "gemma/modeling_gemma.py:GemmaRMSNorm.extra_repr": ["ModelRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "weight"], "gemma/modeling_gemma.py:GemmaMLP.__init__": ["ACT2FN", "False", "Linear", "ModelMLP", "Module", "__init__", "act_fn", "bias", "class", "config", "def", "down_proj", "gate_proj", "hidden_act", "hidden_size", "intermediate_size", "nn", "self", "super", "up_proj"], "gemma/modeling_gemma.py:GemmaMLP.forward": ["ModelMLP", "Module", "act_fn", "class", "def", "down_proj", "forward", "gate_proj", "nn", "return", "self", "up_proj", "x"], "gemma/modeling_gemma.py:GemmaRotaryEmbedding.__init__": ["Callable", "False", "ModelConfig", "ModelRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "gemma/modeling_gemma.py:GemmaRotaryEmbedding.compute_default_rope_parameters": ["ModelConfig", "ModelRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "gemma/modeling_gemma.py:GemmaRotaryEmbedding.forward": ["False", "ModelRotaryEmbedding", "Module", "None", "and", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "gemma/modeling_gemma.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "gemma/modeling_gemma.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cos", "def", "k", "k_embed", "q", "q_embed", "return", "rotate_half", "sin", "unsqueeze", "unsqueeze_dim"], "gemma/modeling_gemma.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "gemma/modeling_gemma.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value", "value_states"], "gemma/modeling_gemma.py:GemmaAttention.__init__": ["False", "Linear", "ModelAttention", "ModelConfig", "Module", "__init__", "attention_bias", "attention_dropout", "bias", "class", "config", "def", "getattr", "head_dim", "hidden_size", "int", "is_causal", "k_proj", "layer_idx", "nn", "not", "num_attention_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "q_proj", "scaling", "self", "super", "use_bidirectional_attention", "v_proj"], "gemma/modeling_gemma.py:GemmaAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "LongTensor", "ModelAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_attn_implementation", "apply_rotary_pos_emb", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "cache_kwargs", "cache_position", "class", "config", "contiguous", "cos", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "head_dim", "hidden_shape", "hidden_states", "if", "input_shape", "is", "k_proj", "key_states", "kwargs", "layer_idx", "nn", "not", "o_proj", "past_key_values", "position_embeddings", "q_proj", "query_states", "reshape", "return", "scaling", "self", "shape", "sin", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "view"], "gemma/modeling_gemma.py:GemmaDecoderLayer.__init__": ["GradientCheckpointingLayer", "ModelAttention", "ModelConfig", "ModelDecoderLayer", "ModelMLP", "ModelRMSNorm", "__init__", "class", "config", "def", "eps", "hidden_size", "input_layernorm", "int", "layer_idx", "mlp", "post_attention_layernorm", "rms_norm_eps", "self", "self_attn", "super"], "gemma/modeling_gemma.py:GemmaDecoderLayer.forward": ["Cache", "False", "GradientCheckpointingLayer", "LongTensor", "ModelDecoderLayer", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "attention_mask", "bool", "cache_position", "class", "def", "forward", "hidden_states", "input_layernorm", "kwargs", "mlp", "past_key_values", "position_embeddings", "position_ids", "post_attention_layernorm", "residual", "return", "self", "self_attn", "torch", "tuple", "use_cache"], "gemma/modeling_gemma.py:GemmaPreTrainedModel._init_weights": ["ModelPreTrainedModel", "PreTrainedModel", "RMSNorm", "__class__", "__name__", "_init_weights", "class", "def", "if", "in", "init", "module", "self", "super", "weight", "zeros_"], "gemma/modeling_gemma.py:GemmaModel.__init__": ["Embedding", "False", "ModelConfig", "ModelDecoderLayer", "ModelModel", "ModelPreTrainedModel", "ModelRMSNorm", "ModelRotaryEmbedding", "ModuleList", "__init__", "class", "config", "def", "embed_tokens", "eps", "for", "gradient_checkpointing", "hidden_size", "in", "layer_idx", "layers", "nn", "norm", "num_hidden_layers", "pad_token_id", "padding_idx", "post_init", "range", "rms_norm_eps", "rotary_emb", "self", "super", "vocab_size"], "gemma/modeling_gemma.py:GemmaModel.forward": ["BaseModelOutputWithPast", "Cache", "DynamicCache", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "and", "arange", "attention_mask", "bool", "cache_position", "causal_mask", "class", "config", "create_causal_mask", "decoder_layer", "def", "device", "dtype", "else", "embed_tokens", "exactly", "for", "forward", "get_seq_length", "hidden_size", "hidden_states", "if", "in", "input_embeds", "input_ids", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layers", "must", "norm", "normalizer", "not", "num_hidden_layers", "of", "one", "or", "past_key_values", "past_seen_tokens", "position_embeddings", "position_ids", "raise", "return", "rotary_emb", "self", "shape", "specify", "tensor", "torch", "unsqueeze", "use_cache"], "gemma/modeling_gemma.py:GemmaForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "ModelForCausalLM", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "vocab_size"], "gemma/modeling_gemma.py:GemmaForCausalLM.forward": ["BaseModelOutputWithPast", "Cache", "CausalLMOutputWithPast", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForCausalLM", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "outputs", "past_key_values", "position_ids", "r", "return", "self", "slice", "slice_indices", "torch", "use_cache", "vocab_size"], "gemma2/modeling_gemma2.py:Gemma2RMSNorm.__init__": ["ModelRMSNorm", "Module", "Parameter", "__init__", "class", "def", "dim", "eps", "float", "int", "nn", "self", "super", "torch", "weight", "zeros"], "gemma2/modeling_gemma2.py:Gemma2RMSNorm._norm": ["ModelRMSNorm", "Module", "True", "_norm", "class", "def", "eps", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "torch", "x"], "gemma2/modeling_gemma2.py:Gemma2RMSNorm.forward": ["ModelRMSNorm", "Module", "_norm", "class", "def", "float", "forward", "nn", "output", "return", "self", "type_as", "weight", "x"], "gemma2/modeling_gemma2.py:Gemma2RMSNorm.extra_repr": ["ModelRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "weight"], "gemma2/modeling_gemma2.py:Gemma2MLP.__init__": ["ACT2FN", "False", "Linear", "ModelMLP", "Module", "__init__", "act_fn", "bias", "class", "config", "def", "down_proj", "gate_proj", "hidden_activation", "hidden_size", "intermediate_size", "nn", "self", "super", "up_proj"], "gemma2/modeling_gemma2.py:Gemma2MLP.forward": ["ModelMLP", "Module", "act_fn", "class", "def", "down_proj", "forward", "gate_proj", "nn", "return", "self", "up_proj", "x"], "gemma2/modeling_gemma2.py:Gemma2RotaryEmbedding.__init__": ["Callable", "False", "ModelConfig", "ModelRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "gemma2/modeling_gemma2.py:Gemma2RotaryEmbedding.compute_default_rope_parameters": ["ModelConfig", "ModelRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "gemma2/modeling_gemma2.py:Gemma2RotaryEmbedding.forward": ["False", "ModelRotaryEmbedding", "Module", "None", "and", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "gemma2/modeling_gemma2.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "gemma2/modeling_gemma2.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cos", "def", "k", "k_embed", "q", "q_embed", "return", "rotate_half", "sin", "unsqueeze", "unsqueeze_dim"], "gemma2/modeling_gemma2.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "gemma2/modeling_gemma2.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "head_dim", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softcap", "softmax", "tanh", "to", "torch", "training", "transpose", "tuple", "value", "value_states"], "gemma2/modeling_gemma2.py:Gemma2Attention.__init__": ["False", "Linear", "ModelAttention", "ModelConfig", "Module", "None", "__init__", "attention_bias", "attention_dropout", "attn_logit_softcapping", "bias", "class", "config", "def", "else", "getattr", "hasattr", "head_dim", "hidden_size", "if", "int", "is_causal", "k_proj", "layer_idx", "layer_type", "layer_types", "nn", "not", "num_attention_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "q_proj", "query_pre_attn_scalar", "scaling", "self", "sliding_attention", "sliding_window", "super", "use_bidirectional_attention", "v_proj"], "gemma2/modeling_gemma2.py:Gemma2Attention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "FlashAttentionKwargs", "LongTensor", "ModelAttention", "Module", "None", "Tensor", "Unpack", "_attn_implementation", "apply_rotary_pos_emb", "attention_dropout", "attention_interface", "attention_mask", "attn_logit_softcapping", "attn_output", "attn_weights", "cache_kwargs", "cache_position", "class", "config", "contiguous", "cos", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "head_dim", "hidden_shape", "hidden_states", "if", "input_shape", "is", "k_proj", "key_states", "kwargs", "layer_idx", "nn", "not", "o_proj", "past_key_values", "position_embeddings", "q_proj", "query_states", "reshape", "return", "scaling", "self", "shape", "sin", "sliding_window", "softcap", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "view"], "gemma2/modeling_gemma2.py:Gemma2DecoderLayer.__init__": ["GradientCheckpointingLayer", "ModelAttention", "ModelConfig", "ModelDecoderLayer", "ModelMLP", "ModelRMSNorm", "__init__", "attention_type", "class", "config", "def", "eps", "hidden_size", "input_layernorm", "int", "layer_idx", "layer_types", "mlp", "post_attention_layernorm", "post_feedforward_layernorm", "pre_feedforward_layernorm", "rms_norm_eps", "self", "self_attn", "super"], "gemma2/modeling_gemma2.py:Gemma2DecoderLayer.forward": ["Cache", "FloatTensor", "GradientCheckpointingLayer", "LongTensor", "ModelDecoderLayer", "None", "Tensor", "_", "attention_mask", "cache_position", "class", "def", "forward", "hidden_states", "input_layernorm", "kwargs", "mlp", "past_key_values", "position_embeddings", "position_ids", "post_attention_layernorm", "post_feedforward_layernorm", "pre_feedforward_layernorm", "residual", "return", "self", "self_attn", "torch", "tuple"], "gemma2/modeling_gemma2.py:Gemma2PreTrainedModel._init_weights": ["ModelPreTrainedModel", "PreTrainedModel", "RMSNorm", "__class__", "__name__", "_init_weights", "class", "def", "if", "in", "init", "module", "self", "super", "weight", "zeros_"], "gemma2/modeling_gemma2.py:Gemma2Model.__init__": ["Embedding", "False", "ModelConfig", "ModelDecoderLayer", "ModelModel", "ModelPreTrainedModel", "ModelRMSNorm", "ModelRotaryEmbedding", "ModuleList", "__init__", "class", "config", "def", "embed_tokens", "eps", "for", "gradient_checkpointing", "hidden_size", "in", "layer_idx", "layers", "nn", "norm", "num_hidden_layers", "pad_token_id", "padding_idx", "post_init", "range", "rms_norm_eps", "rotary_emb", "self", "super", "vocab_size"], "gemma2/modeling_gemma2.py:Gemma2Model.forward": ["BaseModelOutputWithPast", "Cache", "DynamicCache", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "and", "arange", "attention_mask", "attention_type", "bool", "cache_position", "causal_mask_mapping", "class", "config", "create_causal_mask", "create_sliding_window_causal_mask", "decoder_layer", "def", "device", "dict", "dtype", "else", "embed_tokens", "exactly", "for", "forward", "full_attention", "get_seq_length", "hidden_size", "hidden_states", "if", "in", "input_embeds", "input_ids", "inputs_embeds", "is", "isinstance", "kwargs", "last_hidden_state", "layers", "mask_kwargs", "must", "norm", "normalizer", "not", "num_hidden_layers", "of", "one", "or", "past_key_values", "past_seen_tokens", "position_embeddings", "position_ids", "raise", "return", "rotary_emb", "self", "shape", "sliding_attention", "specify", "tensor", "torch", "unsqueeze", "use_cache"], "gemma2/modeling_gemma2.py:Gemma2ForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "ModelForCausalLM", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "vocab_size"], "gemma2/modeling_gemma2.py:Gemma2ForCausalLM.forward": ["BaseModelOutputWithPast", "Cache", "CausalLMOutputWithPast", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForCausalLM", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "else", "final_logit_softcapping", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "outputs", "past_key_values", "position_ids", "r", "return", "self", "slice", "slice_indices", "tanh", "torch", "use_cache", "vocab_size"], "gemma3/modeling_gemma3.py:Gemma3TextScaledWordEmbedding.__init__": ["Embedding", "False", "ModelTextScaledWordEmbedding", "__init__", "class", "def", "embed_scale", "embedding_dim", "float", "int", "nn", "num_embeddings", "padding_idx", "persistent", "register_buffer", "scalar_embed_scale", "self", "super", "tensor", "torch"], "gemma3/modeling_gemma3.py:Gemma3TextScaledWordEmbedding.forward": ["Embedding", "ModelTextScaledWordEmbedding", "Tensor", "class", "def", "dtype", "embed_scale", "forward", "input_ids", "nn", "return", "self", "super", "to", "torch", "weight"], "gemma3/modeling_gemma3.py:Gemma3MLP.__init__": ["ACT2FN", "False", "Linear", "ModelMLP", "ModelTextConfig", "Module", "__init__", "act_fn", "bias", "class", "config", "def", "down_proj", "gate_proj", "hidden_activation", "hidden_size", "intermediate_size", "nn", "self", "super", "up_proj"], "gemma3/modeling_gemma3.py:Gemma3MLP.forward": ["ModelMLP", "Module", "act_fn", "class", "def", "down_proj", "forward", "gate_proj", "nn", "return", "self", "up_proj", "x"], "gemma3/modeling_gemma3.py:Gemma3RMSNorm.__init__": ["ModelRMSNorm", "Module", "Parameter", "__init__", "class", "def", "dim", "eps", "float", "int", "nn", "self", "super", "torch", "weight", "zeros"], "gemma3/modeling_gemma3.py:Gemma3RMSNorm._norm": ["ModelRMSNorm", "Module", "True", "_norm", "class", "def", "eps", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "torch", "x"], "gemma3/modeling_gemma3.py:Gemma3RMSNorm.forward": ["ModelRMSNorm", "Module", "_norm", "class", "def", "float", "forward", "nn", "output", "return", "self", "type_as", "weight", "x"], "gemma3/modeling_gemma3.py:Gemma3RMSNorm.extra_repr": ["ModelRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "weight"], "gemma3/modeling_gemma3.py:Gemma3RotaryEmbedding.__init__": ["Callable", "False", "ModelRotaryEmbedding", "ModelTextConfig", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "_attention_scaling", "_inv_freq", "_original_inv_freq", "class", "clone", "compute_default_rope_parameters", "config", "continue", "curr_attention_scaling", "curr_inv_freq", "def", "default", "device", "f", "for", "if", "in", "is", "layer_type", "layer_types", "list", "max_position_embeddings", "max_seq_len_cached", "nn", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_params", "rope_type", "self", "set", "setattr", "super"], "gemma3/modeling_gemma3.py:Gemma3RotaryEmbedding.compute_default_rope_parameters": ["ModelRotaryEmbedding", "ModelTextConfig", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "layer_type", "nn", "num_attention_heads", "or", "return", "rope_parameters", "rope_theta", "seq_len", "str", "to", "torch", "tuple"], "gemma3/modeling_gemma3.py:Gemma3RotaryEmbedding.forward": ["False", "ModelRotaryEmbedding", "Module", "None", "_attention_scaling", "_inv_freq", "and", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "f", "float", "forward", "freqs", "getattr", "if", "inv_freq", "inv_freq_expanded", "isinstance", "layer_type", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "gemma3/modeling_gemma3.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "gemma3/modeling_gemma3.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cos", "def", "k", "k_embed", "q", "q_embed", "return", "rotate_half", "sin", "unsqueeze", "unsqueeze_dim"], "gemma3/modeling_gemma3.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "gemma3/modeling_gemma3.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "head_dim", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softcap", "softmax", "tanh", "to", "torch", "training", "transpose", "tuple", "value", "value_states"], "gemma3/modeling_gemma3.py:Gemma3Attention.__init__": ["Linear", "ModelAttention", "ModelRMSNorm", "ModelTextConfig", "Module", "None", "__init__", "attention_bias", "attention_dropout", "attn_logit_softcapping", "bias", "class", "config", "def", "dim", "else", "eps", "getattr", "hasattr", "head_dim", "hidden_size", "if", "int", "is_causal", "is_sliding", "k_norm", "k_proj", "layer_idx", "layer_type", "layer_types", "nn", "not", "num_attention_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "q_norm", "q_proj", "query_pre_attn_scalar", "rms_norm_eps", "scaling", "self", "sliding_attention", "sliding_window", "super", "use_bidirectional_attention", "v_proj"], "gemma3/modeling_gemma3.py:Gemma3Attention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "LongTensor", "ModelAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_attn_implementation", "apply_rotary_pos_emb", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "cache_kwargs", "cache_position", "class", "config", "contiguous", "cos", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "head_dim", "hidden_shape", "hidden_states", "if", "input_shape", "is", "k_norm", "k_proj", "key_states", "kwargs", "layer_idx", "nn", "not", "o_proj", "past_key_values", "position_embeddings", "q_norm", "q_proj", "query_states", "reshape", "return", "scaling", "self", "shape", "sin", "sliding_window", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "view"], "gemma3/modeling_gemma3.py:Gemma3DecoderLayer.__init__": ["GradientCheckpointingLayer", "ModelAttention", "ModelDecoderLayer", "ModelMLP", "ModelRMSNorm", "ModelTextConfig", "__init__", "attention_type", "class", "config", "def", "eps", "hidden_size", "input_layernorm", "int", "layer_idx", "layer_types", "mlp", "post_attention_layernorm", "post_feedforward_layernorm", "pre_feedforward_layernorm", "rms_norm_eps", "self", "self_attn", "super"], "gemma3/modeling_gemma3.py:Gemma3DecoderLayer.forward": ["Cache", "FloatTensor", "GradientCheckpointingLayer", "LongTensor", "ModelDecoderLayer", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "attention_mask", "cache_position", "class", "def", "forward", "hidden_states", "input_layernorm", "kwargs", "mlp", "past_key_values", "position_embeddings", "position_ids", "post_attention_layernorm", "post_feedforward_layernorm", "pre_feedforward_layernorm", "residual", "return", "self", "self_attn", "torch", "tuple"], "gemma3/modeling_gemma3.py:Gemma3PreTrainedModel._init_weights": ["ModelMultiModalProjector", "ModelPreTrainedModel", "ModelRotaryEmbedding", "ModelTextScaledWordEmbedding", "PreTrainedModel", "RMSNorm", "ROPE_INIT_FUNCTIONS", "_", "__class__", "__name__", "_init_weights", "_inv_freq", "_original_inv_freq", "class", "compute_default_rope_parameters", "config", "constant_", "copy_", "curr_inv_freq", "def", "default", "elif", "embed_scale", "f", "for", "getattr", "if", "in", "init", "isinstance", "layer_type", "layer_types", "mm_input_projection_weight", "module", "rope_init_fn", "rope_type", "scalar_embed_scale", "self", "super", "weight", "zeros_"], "gemma3/modeling_gemma3.py:_bidirectional_window_overlay": ["Callable", "_bidirectional_window_overlay", "abs", "batch_idx", "bool", "def", "head_idx", "inner_mask", "int", "kv_idx", "q_idx", "return", "sliding_window"], "gemma3/modeling_gemma3.py:Gemma3TextModel.__init__": ["False", "ModelDecoderLayer", "ModelPreTrainedModel", "ModelRMSNorm", "ModelRotaryEmbedding", "ModelTextConfig", "ModelTextModel", "ModelTextScaledWordEmbedding", "ModuleList", "__init__", "class", "config", "def", "embed_scale", "embed_tokens", "eps", "for", "gradient_checkpointing", "hidden_size", "in", "layer_idx", "layers", "nn", "norm", "num_hidden_layers", "pad_token_id", "padding_idx", "post_init", "range", "rms_norm_eps", "rotary_emb", "self", "super", "vocab_size"], "gemma3/modeling_gemma3.py:Gemma3TextModel.forward": ["BaseModelOutputWithPast", "Cache", "DynamicCache", "FloatTensor", "LongTensor", "ModelPreTrainedModel", "ModelTextModel", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "ValueError", "You", "_bidirectional_window_overlay", "and", "arange", "args", "attention_mask", "attention_type", "bool", "cache_position", "causal_mask_mapping", "class", "config", "copy", "create_causal_mask", "create_sliding_window_causal_mask", "decoder_layer", "def", "device", "dict", "dtype", "else", "embed_tokens", "exactly", "for", "forward", "full_attention", "get_seq_length", "hidden_states", "if", "in", "input_embeds", "input_ids", "inputs_embeds", "is", "isinstance", "kwargs", "lambda", "last_hidden_state", "layer_type", "layer_types", "layers", "mask_kwargs", "must", "norm", "not", "num_hidden_layers", "of", "one", "or", "or_mask_function", "past_key_values", "past_seen_tokens", "position_embeddings", "position_ids", "raise", "return", "rotary_emb", "self", "shape", "sliding_attention", "sliding_mask_kwargs", "sliding_window", "specify", "tensor", "torch", "unsqueeze", "use_bidirectional_attention", "use_cache"], "gemma3/modeling_gemma3.py:Gemma3ForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "ModelForCausalLM", "ModelPreTrainedModel", "ModelTextConfig", "ModelTextModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "vocab_size"], "gemma3/modeling_gemma3.py:Gemma3ForCausalLM.forward": ["BaseModelOutputWithPast", "Cache", "CausalLMOutputWithPast", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForCausalLM", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "else", "final_logit_softcapping", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "outputs", "past_key_values", "position_ids", "r", "return", "self", "slice", "slice_indices", "tanh", "torch", "use_cache", "vocab_size"], "gemma3/modeling_gemma3.py:Gemma3MultiModalProjector.__init__": ["AvgPool2d", "ModelConfig", "ModelMultiModalProjector", "ModelRMSNorm", "Module", "Parameter", "__init__", "avg_pool", "class", "config", "def", "eps", "hidden_size", "image_size", "int", "kernel_size", "layer_norm_eps", "mm_input_projection_weight", "mm_soft_emb_norm", "mm_tokens_per_image", "nn", "patch_size", "patches_per_image", "self", "stride", "super", "text_config", "tokens_per_side", "torch", "vision_config", "zeros"], "gemma3/modeling_gemma3.py:Gemma3MultiModalProjector.forward": ["ModelMultiModalProjector", "Module", "Tensor", "_", "avg_pool", "batch_size", "class", "contiguous", "def", "flatten", "forward", "matmul", "mm_input_projection_weight", "mm_soft_emb_norm", "nn", "normed_vision_outputs", "patches_per_image", "pooled_vision_outputs", "projected_vision_outputs", "reshape", "reshaped_vision_outputs", "return", "self", "seq_length", "shape", "torch", "transpose", "type_as", "vision_outputs"], "gemma3/modeling_gemma3.py:token_type_ids_mask_function": ["Callable", "Model_type_ids", "Model_type_ids_at_kv_idx", "Model_type_ids_at_q_idx", "Model_type_ids_mask_function", "None", "Tensor", "batch_idx", "bool", "def", "head_idx", "if", "image_group_ids", "image_group_ids_at_kv_idx", "image_group_ids_at_q_idx", "inner_mask", "int", "is", "is_image_block", "kv_idx", "q_idx", "return", "safe_kv_idx", "safe_q_idx", "same_image_block", "shape", "torch", "where"], "gemma3/modeling_gemma3.py:create_causal_mask_mapping": ["Cache", "False", "FloatTensor", "Model_causal_mask_mapping", "Model_masks_for_generate", "None", "PreTrainedConfig", "Tensor", "ValueError", "a", "and", "as", "attention_mask", "bool", "cache_position", "config", "cumsum", "def", "device", "dict", "dim", "else", "functional", "get_text_config", "if", "image_group_ids", "input", "input_embeds", "int", "is", "is_first_iteration", "is_image", "is_initialized", "is_previous_image", "is_training", "kwargs", "mask_kwargs", "model", "new_image_start", "nn", "not", "or", "or_mask_function", "pad", "past_key_values", "pixel_values", "position_ids", "raise", "required", "return", "to", "token_type_ids", "token_type_ids_mask_function", "torch", "training", "value", "when", "where"], "gemma3/modeling_gemma3.py:Gemma3Model.__init__": ["AutoModel", "ModelConfig", "ModelModel", "ModelMultiModalProjector", "ModelPreTrainedModel", "None", "__init__", "class", "config", "def", "else", "from_config", "if", "is", "language_model", "multi_modal_projector", "not", "pad_token_id", "post_init", "self", "super", "text_config", "vision_config", "vision_tower", "vocab_size"], "gemma3/modeling_gemma3.py:Gemma3Model.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "language_model", "return", "self"], "gemma3/modeling_gemma3.py:Gemma3Model.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "language_model", "self", "set_input_embeddings", "value"], "gemma3/modeling_gemma3.py:Gemma3Model.get_image_features": ["ModelModel", "ModelPreTrainedModel", "Tensor", "class", "def", "get_image_features", "image_features", "last_hidden_state", "multi_modal_projector", "pixel_values", "return", "self", "torch", "vision_outputs", "vision_tower"], "gemma3/modeling_gemma3.py:Gemma3Model.get_placeholder_mask": ["FloatTensor", "Image", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "ValueError", "all", "and", "class", "config", "def", "device", "do", "dtype", "else", "expand_as", "f", "features", "get_input_embeddings", "get_placeholder_mask", "if", "image", "image_features", "image_token_id", "input_ids", "inputs_embeds", "is", "long", "match", "n_image_features", "n_image_tokens", "not", "numel", "raise", "return", "self", "shape", "special_image_mask", "sum", "tensor", "to", "tokens", "torch", "unsqueeze"], "gemma3/modeling_gemma3.py:Gemma3Model.forward": ["Cache", "FloatTensor", "LongTensor", "ModelModel", "ModelModelOutputWithPast", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "ValueError", "You", "and", "arange", "attention_mask", "attentions", "bool", "cache_position", "causal_mask_mapping", "class", "clone", "config", "create_causal_mask_mapping", "def", "device", "dict", "dtype", "else", "exactly", "forward", "get_image_features", "get_input_embeddings", "get_placeholder_mask", "get_seq_length", "hidden_states", "if", "image_features", "image_hidden_states", "image_token_id", "input_ids", "inputs_embeds", "is", "is_training", "isinstance", "labels", "language_model", "last_hidden_state", "llm_input_ids", "lm_kwargs", "masked_scatter", "must", "not", "of", "one", "or", "outputs", "past_key_values", "past_seen_tokens", "pixel_values", "position_ids", "r", "raise", "return", "return_dict", "self", "shape", "special_image_mask", "specify", "to", "token_type_ids", "torch", "training", "tuple", "use_cache", "vocab_size"], "gemma3/modeling_gemma3.py:Gemma3ForConditionalGeneration.__init__": ["False", "GenerationMixin", "Linear", "ModelConfig", "ModelForConditionalGeneration", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "text_config", "vocab_size"], "gemma3/modeling_gemma3.py:Gemma3ForConditionalGeneration.get_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "model", "return", "self"], "gemma3/modeling_gemma3.py:Gemma3ForConditionalGeneration.set_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "model", "self", "set_input_embeddings", "value"], "gemma3/modeling_gemma3.py:Gemma3ForConditionalGeneration.get_image_features": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "get_image_features", "model", "pixel_values", "return", "self"], "gemma3/modeling_gemma3.py:Gemma3ForConditionalGeneration.forward": ["Cache", "CrossEntropyLoss", "FloatTensor", "GenerationMixin", "LongTensor", "ModelCausalLMOutputWithPast", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "contiguous", "def", "device", "else", "flat_labels", "flat_logits", "float", "forward", "hidden_states", "if", "image_hidden_states", "input_ids", "inputs_embeds", "int", "is", "isinstance", "labels", "lm_head", "lm_kwargs", "logits", "logits_to_keep", "loss", "loss_fct", "model", "nn", "not", "outputs", "past_key_values", "pixel_values", "position_ids", "r", "return", "self", "shape", "shift_attention_mask", "shift_labels", "shift_logits", "slice", "slice_indices", "text_config", "to", "token_type_ids", "torch", "tuple", "use_cache", "view", "vocab_size"], "gemma3/modeling_gemma3.py:Gemma3ForConditionalGeneration.prepare_inputs_for_generation": ["False", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "True", "attention_mask", "cache_position", "class", "def", "if", "input_ids", "inputs_embeds", "is_first_iteration", "kwargs", "labels", "logits_to_keep", "model_inputs", "not", "or", "past_key_values", "pixel_values", "position_ids", "prepare_inputs_for_generation", "return", "self", "super", "token_type_ids", "use_cache"], "gemma3/modeling_gemma3.py:Gemma3ForConditionalGeneration.create_masks_for_generate": ["Cache", "False", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "PreTrainedConfig", "Tensor", "attention_mask", "bool", "cache_position", "class", "config", "create_causal_mask_mapping", "create_masks_for_generate", "def", "dict", "for", "if", "in", "input_embeds", "is_first_iteration", "items", "k", "kwargs", "past_key_values", "pixel_values", "position_ids", "return", "token_type_ids", "torch", "v"], "gemma3/modeling_gemma3.py:Gemma3ForSequenceClassification.__init__": ["False", "Linear", "ModelForSequenceClassification", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "model", "nn", "num_labels", "post_init", "score", "self", "super", "text_config"], "gemma3/modeling_gemma3.py:Gemma3ForSequenceClassification.get_input_embeddings": ["ModelForSequenceClassification", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "model", "return", "self"], "gemma3/modeling_gemma3.py:Gemma3ForSequenceClassification.set_input_embeddings": ["ModelForSequenceClassification", "ModelPreTrainedModel", "class", "def", "model", "self", "set_input_embeddings", "value"], "gemma3/modeling_gemma3.py:Gemma3ForSequenceClassification.forward": ["Cache", "Cannot", "FloatTensor", "LongTensor", "ModelForSequenceClassification", "ModelPreTrainedModel", "None", "Results", "SequenceClassifierOutputWithPast", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "__class__", "__name__", "and", "arange", "argmax", "attention_mask", "attentions", "batch", "batch_size", "be", "bool", "class", "config", "conjunction", "def", "defined", "detect", "device", "dtype", "elif", "else", "f", "forward", "handle", "hidden_states", "if", "in", "input_ids", "inputs_embeds", "int32", "is", "kwargs", "labels", "last_hidden_state", "last_non_pad_token", "logger", "logits", "loss", "loss_function", "may", "model", "no", "non_pad_mask", "not", "pad_token_id", "padding", "past_key_values", "pixel_values", "pooled_logits", "position_ids", "r", "raise", "return", "score", "self", "shape", "sizes", "text_config", "to", "token", "token_indices", "token_type_ids", "tokens", "torch", "transformer_outputs", "unexpected", "use_cache", "using", "warning_once", "will", "with"], "gemma3n/modeling_gemma3n.py:Gemma3nRMSNorm.__init__": ["False", "ModelRMSNorm", "Module", "Parameter", "True", "__init__", "bool", "class", "def", "dim", "else", "eps", "float", "if", "int", "nn", "ones", "persistent", "register_buffer", "self", "super", "tensor", "torch", "weight", "with_scale"], "gemma3n/modeling_gemma3n.py:Gemma3nRMSNorm._norm": ["ModelRMSNorm", "Module", "True", "_norm", "class", "def", "eps", "keepdim", "mean", "nn", "pow", "return", "self", "sqrt", "torch", "x"], "gemma3n/modeling_gemma3n.py:Gemma3nRMSNorm.forward": ["ModelRMSNorm", "Module", "Tensor", "_norm", "class", "def", "float", "forward", "nn", "output", "return", "self", "torch", "type_as", "weight", "x"], "gemma3n/modeling_gemma3n.py:Gemma3nRMSNorm.extra_repr": ["ModelRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "weight"], "gemma3n/modeling_gemma3n.py:Gemma3nAudioRelativePositionEmbedding.__init__": ["False", "Linear", "ModelAudioConfig", "ModelAudioRelativePositionEmbedding", "Module", "__init__", "arange", "bias", "channels", "class", "conf_attention_context_left", "conf_attention_context_right", "conf_num_attention_heads", "config", "def", "exp", "float", "head_dim", "hidden_size", "inv_timescales", "log", "log_timescale_increment", "math", "max", "max_backward", "max_forward", "max_timescale", "min_timescale", "nn", "num_heads", "num_timescales", "persistent", "pos_proj", "register_buffer", "self", "super", "torch", "unsqueeze"], "gemma3n/modeling_gemma3n.py:Gemma3nAudioRelativePositionEmbedding._get_timing_signal_1d_pos": ["ModelAudioRelativePositionEmbedding", "Module", "Tensor", "_get_timing_signal_1d_pos", "cat", "class", "cos", "def", "device", "dim", "dtype", "float", "float32", "inv_timescales", "nn", "position", "return", "scaled_time", "self", "sin", "timing_signal", "to", "torch", "type", "unsqueeze"], "gemma3n/modeling_gemma3n.py:Gemma3nAudioRelativePositionEmbedding._relative_shift": ["ModelAudioRelativePositionEmbedding", "Module", "Tensor", "_relative_shift", "batch_size", "class", "def", "functional", "int", "key_context_size", "max_span_plus_1", "nn", "num_heads", "num_query_blocks", "pad", "pad_amount_last_dim", "padding_tuple", "query_block_size", "reshape", "return", "self", "term_bd_before_shift", "term_bd_padded", "term_bd_reshaped", "term_bd_shifted", "term_bd_sliced", "torch"], "gemma3n/modeling_gemma3n.py:Gemma3nAudioRelativePositionEmbedding.forward": ["ModelAudioRelativePositionEmbedding", "Module", "Tensor", "_", "_get_timing_signal_1d_pos", "_relative_shift", "arange", "batch_size", "class", "def", "device", "dtype", "forward", "head_dim", "key_context_size", "keys", "keys_p_t", "matmul", "max_backward", "max_forward", "max_span_plus_1", "nn", "num_heads", "num_query_blocks", "permute", "pos_indices", "pos_proj", "projected_sin_emb", "q_permuted", "q_reshaped", "queries", "queries_p", "query_block_size", "reshape", "return", "s_permuted", "self", "shape", "sin_emb", "sin_emb_timing_signal", "squeeze", "term_ac", "term_bd_shifted", "term_bd_unshifed", "term_bd_unshifed_matmul", "torch", "unsqueeze"], "gemma3n/modeling_gemma3n.py:Gemma3nAudioAttention.__init__": ["False", "Linear", "ModelAudioAttention", "ModelAudioConfig", "ModelAudioRelativePositionEmbedding", "Module", "Parameter", "__init__", "attention_logits_soft_cap", "bias", "chunk_size", "class", "clone", "conf_attention_chunk_size", "conf_attention_context_left", "conf_attention_context_right", "conf_attention_logit_cap", "conf_num_attention_heads", "config", "context_size", "create_local_causal_valid_mask", "def", "detach", "float", "functional", "head_dim", "hidden_size", "k_proj", "local_causal_valid_mask", "max", "max_future_horizon", "max_past_horizon", "nn", "num_heads", "per_dim_scale", "persistent", "q_proj", "q_scale", "r_softplus_0", "register_buffer", "relative_position_embedding", "self", "softcap", "softplus", "super", "tensor", "torch", "v_proj", "zeros"], "gemma3n/modeling_gemma3n.py:Gemma3nAudioAttention.create_local_causal_valid_mask": ["ModelAudioAttention", "Module", "T", "bool", "chunk_size", "class", "context_size", "create_local_causal_valid_mask", "def", "diagonal", "dtype", "local_causal_valid_mask", "lower_causal_mask", "max_future_horizon", "max_past_horizon", "nn", "ones", "return", "self", "torch", "tril", "upper_causal_mask"], "gemma3n/modeling_gemma3n.py:Gemma3nAudioAttention._pad_dim1": ["ModelAudioAttention", "Module", "Tensor", "_", "_pad_dim1", "batch", "cat", "class", "def", "dim", "int", "left", "new_zeros", "nn", "pad_left", "pad_right", "return", "right", "self", "shape", "tail_shape", "torch", "x"], "gemma3n/modeling_gemma3n.py:Gemma3nAudioAttention._convert_to_block": ["ModelAudioAttention", "Module", "Tensor", "_convert_to_block", "_pad_dim1", "b", "chunk_size", "class", "contiguous", "def", "hidden_states", "if", "nn", "num_blocks", "padding_len", "permute_dims", "reshape", "return", "self", "shape", "t", "torch"], "gemma3n/modeling_gemma3n.py:Gemma3nAudioAttention._extract_block_context": ["ModelAudioAttention", "Module", "Tensor", "_extract_block_context", "_pad_dim1", "and", "chunk_size", "class", "context_size", "contiguous", "def", "destination", "dimension", "frame_len", "frame_step", "hidden_states", "if", "max_future_horizon", "max_past_horizon", "movedim", "ndim", "nn", "pad_left", "pad_right", "return", "self", "size", "source", "step", "torch", "unfold", "x_unfolded"], "gemma3n/modeling_gemma3n.py:Gemma3nAudioAttention.forward": ["BoolTensor", "ModelAudioAttention", "Module", "Shape", "Tensor", "ValueError", "_convert_to_block", "_extract_block_context", "after", "and", "b_dim", "batch_size", "bmm", "broadcast_shape", "c_dim", "chunk_size", "class", "condition_from_causality", "condition_from_input_validity", "context_size", "context_vectors", "contiguous", "def", "device", "dim", "dtype", "extracted_valid_mask_blocks", "f", "final_condition_for_where", "finfo", "float32", "forward", "functional", "h_dim", "head_dim", "hidden_states", "if", "is", "k_proj", "key_blocks", "key_states", "local_causal_valid_mask", "logical_and", "logits", "mask", "min", "n_dim", "ndim", "nn", "not", "num_heads", "num_query_blocks", "of", "original_valid_mask", "per_dim_scale", "per_dim_scale_sp", "per_dim_scale_sp_broadcast", "permute", "potential", "prob_bun", "probabilities", "q_proj", "q_scale", "q_time", "qkv_shape", "query_blocks", "query_states", "raise", "relative_position_embedding", "reshape", "result_bmm", "return", "self", "shape", "softcap", "softcap_val", "softmax", "softplus", "tanh", "to", "torch", "u_dim", "unsqueeze", "v_bun", "v_proj", "value_blocks", "value_states", "view", "w_dim", "where"], "gemma3n/modeling_gemma3n.py:Gemma3nAudioCumulativeGroupNorm.__init__": ["ModelAudioCumulativeGroupNorm", "Module", "Parameter", "Sequence", "__init__", "class", "def", "eps", "feature_dims", "float", "int", "len", "nn", "num_channels", "ones", "range", "reduction_axes", "self", "super", "torch", "tuple", "weight"], "gemma3n/modeling_gemma3n.py:Gemma3nAudioCumulativeGroupNorm.forward": ["Input", "ModelAudioCumulativeGroupNorm", "Module", "Tensor", "True", "ValueError", "calc_dtype", "clamp", "class", "cum_count_elements", "cum_mean", "cum_sum_sq_diff", "cum_sum_values", "cum_variance", "cumsum", "def", "dim", "does", "dtype", "elements_in_group_at_t", "eps", "expected", "expected_input_suffix", "f", "feature_dims", "final_output", "float32", "forward", "hidden_states", "if", "input_dtype", "keepdim", "mask_calc", "match", "min", "nn", "normalized_x", "not", "num_channels", "ones_like", "pow", "raise", "reduction_axes", "return", "rsqrt", "safe_cum_count_elements", "scale", "scale_view_shape", "self", "shape", "squared_diff_from_mean", "suffix", "sum", "sum_sq_diff_at_t", "sum_values_at_t", "tensor", "to", "torch", "view", "weight", "x_calc"], "gemma3n/modeling_gemma3n.py:Gemma3nAudioSSCPConvBlock.__init__": ["Conv2d", "False", "ModelAudioConfig", "ModelAudioCumulativeGroupNorm", "ModelAudioSSCPConvBlock", "Module", "ReLU", "__init__", "activation", "bias", "class", "config", "conv", "def", "else", "eps", "f_in_padded", "f_out_conv", "feature_dims", "idx", "if", "in_channels", "input_freq_dim", "int", "kernel_h", "kernel_size", "kernel_w", "manual_padding", "nn", "norm", "num_channels", "out_channels", "padding", "self", "sscp_conv_channel_size", "sscp_conv_group_norm_eps", "sscp_conv_kernel_size", "sscp_conv_stride_size", "stride", "stride_h", "stride_w", "super", "tuple"], "gemma3n/modeling_gemma3n.py:Gemma3nAudioSSCPConvBlock.forward": ["F", "ModelAudioSSCPConvBlock", "Module", "Tensor", "activation", "audio_encodings", "audio_encodings_conv", "audio_encodings_normed", "audio_encodings_padded", "class", "constant", "contiguous", "conv", "def", "dtype", "forward", "manual_padding", "mode", "nn", "norm", "pad", "permute", "return", "self", "to", "torch", "value", "weight", "x_for_norm", "x_normed"], "gemma3n/modeling_gemma3n.py:Gemma3nAudioSubSampleConvProjection.__init__": ["False", "Linear", "ModelAudioConfig", "ModelAudioSSCPConvBlock", "ModelAudioSubSampleConvProjection", "Module", "__init__", "append", "bias", "calculated_block_padding", "calculated_f_out_dims", "class", "config", "conv_0", "conv_1", "current_f_for_block_input", "def", "f_in_padded", "f_out_after_conv", "final_c_out", "final_f_out", "for", "hidden_size", "i", "idx", "in", "input_feat_size", "input_freq_dim", "input_proj_in_features", "input_proj_linear", "kernel_h", "kernel_w", "manual_padding", "manual_padding_tuple", "nn", "pad_f_left", "pad_f_right", "pad_t_bottom", "pad_t_top", "range", "self", "sscp_conv_channel_size", "sscp_conv_kernel_size", "sscp_conv_stride_size", "stride_h", "stride_w", "super"], "gemma3n/modeling_gemma3n.py:Gemma3nAudioSubSampleConvProjection.forward": ["ModelAudioSubSampleConvProjection", "Module", "Tensor", "audio_encodings", "audio_encodings_reshaped", "b", "c_out", "class", "contiguous", "conv_0", "conv_1", "def", "f_out", "forward", "input_proj_linear", "nn", "output", "output_flattened", "permute", "return", "self", "shape", "t_out", "torch", "unsqueeze", "view", "x", "x_permuted"], "gemma3n/modeling_gemma3n.py:Gemma3nAudioConformerAttention.__init__": ["False", "Linear", "ModelAudioAttention", "ModelAudioConfig", "ModelAudioConformerAttention", "ModelRMSNorm", "Module", "__init__", "attn", "bias", "class", "config", "def", "gradient_clipping", "hidden_size", "nn", "persistent", "post", "post_in_features", "post_norm", "pre_attn_norm", "register_buffer", "self", "super", "tensor", "torch"], "gemma3n/modeling_gemma3n.py:Gemma3nAudioConformerAttention.forward": ["BoolTensor", "ModelAudioConformerAttention", "Module", "Tensor", "attn", "audio_encodings", "audio_encodings_attn_out", "audio_encodings_input_to_attn", "audio_encodings_norm", "audio_encodings_reshaped", "audio_mel_mask", "b", "clamp", "class", "def", "forward", "gradient_clipping", "head_dim", "nn", "num_heads", "post", "post_norm", "pre_attn_norm", "reshape", "return", "self", "shape", "t", "torch"], "gemma3n/modeling_gemma3n.py:Gemma3nAudioConformerFeedForward.__init__": ["False", "Linear", "ModelAudioConfig", "ModelAudioConformerFeedForward", "ModelRMSNorm", "Module", "__init__", "bias", "class", "conf_residual_weight", "config", "def", "ffw_layer_1", "ffw_layer_2", "gradient_clipping", "hidden_size", "nn", "persistent", "post_layer_norm", "post_layer_scale", "pre_layer_norm", "register_buffer", "self", "super", "tensor", "torch"], "gemma3n/modeling_gemma3n.py:Gemma3nAudioConformerFeedForward.forward": ["ModelAudioConformerFeedForward", "Module", "Tensor", "audio_encodings", "clamp", "class", "def", "ffw_layer_1", "ffw_layer_2", "forward", "functional", "gradient_clipping", "nn", "post_layer_norm", "post_layer_scale", "pre_layer_norm", "residual", "return", "self", "silu", "torch"], "gemma3n/modeling_gemma3n.py:Gemma3nAudioConformerLightConv1d.__init__": ["Conv1d", "False", "Linear", "ModelAudioConfig", "ModelAudioConformerLightConv1d", "ModelRMSNorm", "Module", "__init__", "bias", "causal_padding", "class", "conf_conv_kernel_size", "config", "conv_norm", "def", "depthwise_conv1d", "eps", "gradient_clipping", "groups", "hidden_size", "in_channels", "kernel_size", "linear_end", "linear_start", "nn", "out_channels", "padding", "persistent", "pre_layer_norm", "register_buffer", "rms_norm_eps", "self", "stride", "super", "tensor", "torch"], "gemma3n/modeling_gemma3n.py:Gemma3nAudioConformerLightConv1d.forward": ["F", "ModelAudioConformerLightConv1d", "Module", "Tensor", "audio_encodings", "audio_encodings_permuted", "audio_encodings_permuted_padded", "audio_encodings_residual", "causal_padding", "clamp", "class", "conv_norm", "def", "depthwise_conv1d", "dim", "forward", "functional", "glu", "gradient_clipping", "linear_end", "linear_start", "nn", "output", "pad", "permute", "pre_layer_norm", "return", "self", "silu", "torch"], "gemma3n/modeling_gemma3n.py:Gemma3nAudioConformerBlock.__init__": ["False", "ModelAudioConfig", "ModelAudioConformerAttention", "ModelAudioConformerBlock", "ModelAudioConformerFeedForward", "ModelAudioConformerLightConv1d", "ModelRMSNorm", "Module", "__init__", "attention", "class", "config", "def", "ffw_layer_end", "ffw_layer_start", "gradient_clipping", "hidden_size", "lconv1d", "nn", "norm", "persistent", "register_buffer", "self", "super", "tensor", "torch"], "gemma3n/modeling_gemma3n.py:Gemma3nAudioConformerBlock.forward": ["BoolTensor", "ModelAudioConformerBlock", "Module", "Tensor", "attention", "audio_encodings", "audio_encodings_for_lconv_input", "audio_mel_mask", "clamp", "class", "def", "dtype", "ffw_layer_end", "ffw_layer_start", "forward", "gradient_clipping", "lconv1d", "nn", "norm", "output", "return", "self", "to", "torch", "unsqueeze", "validity_mask_for_lconv"], "gemma3n/modeling_gemma3n.py:Gemma3nAudioEncoder.__init__": ["ModelAudioConfig", "ModelAudioConformerBlock", "ModelAudioEncoder", "ModelAudioSubSampleConvProjection", "ModuleList", "PreTrainedModel", "_", "__init__", "class", "conf_num_hidden_layers", "config", "conformer", "def", "for", "in", "nn", "post_init", "range", "self", "subsample_conv_projection", "super"], "gemma3n/modeling_gemma3n.py:Gemma3nAudioEncoder.forward": ["BoolTensor", "ModelAudioEncoder", "PreTrainedModel", "Tensor", "and", "arange", "audio_encodings", "audio_mel", "audio_mel_mask", "block", "clamp", "class", "conf_reduction_factor", "config", "conformer", "current_mask", "def", "device", "elif", "expand", "for", "forward", "gather", "if", "in", "indices", "kwargs", "len", "masked_fill", "max", "ndim", "range", "return", "self", "shape", "sscp_conv_stride_size", "stride_pair_idx", "subsample_conv_projection", "t_sub", "time_stride_product", "torch", "tuple", "unsqueeze"], "gemma3n/modeling_gemma3n.py:Gemma3nTextScaledWordEmbedding.__init__": ["Embedding", "False", "ModelTextScaledWordEmbedding", "__init__", "class", "def", "embed_scale", "embedding_dim", "float", "int", "nn", "num_embeddings", "padding_idx", "persistent", "register_buffer", "scalar_embed_scale", "self", "super", "tensor", "torch"], "gemma3n/modeling_gemma3n.py:Gemma3nTextScaledWordEmbedding.forward": ["Embedding", "ModelTextScaledWordEmbedding", "Tensor", "class", "def", "dtype", "embed_scale", "forward", "input_ids", "nn", "return", "self", "super", "to", "torch", "weight"], "gemma3n/modeling_gemma3n.py:Gemma3nTextLaurelBlock.__init__": ["False", "Linear", "ModelRMSNorm", "ModelTextConfig", "ModelTextLaurelBlock", "Module", "__init__", "bias", "class", "config", "def", "eps", "hidden_size", "laurel_rank", "linear_left", "linear_right", "nn", "post_laurel_norm", "rms_norm_eps", "self", "super"], "gemma3n/modeling_gemma3n.py:Gemma3nTextLaurelBlock.forward": ["ModelTextLaurelBlock", "Module", "Tensor", "class", "def", "forward", "hidden_states", "laurel_hidden_states", "linear_left", "linear_right", "nn", "normed_laurel_hidden_states", "post_laurel_norm", "return", "self", "torch"], "gemma3n/modeling_gemma3n.py:Gemma3nTextMLP.__init__": ["ACT2FN", "False", "Linear", "ModelTextConfig", "ModelTextMLP", "Module", "__init__", "act_fn", "activation_sparsity", "activation_sparsity_pattern", "bias", "class", "config", "def", "down_proj", "gate_proj", "hidden_activation", "hidden_size", "int", "intermediate_size", "layer_idx", "nn", "self", "super", "up_proj"], "gemma3n/modeling_gemma3n.py:Gemma3nTextMLP.forward": ["ModelTextMLP", "Module", "Tensor", "_gaussian_topk", "act_fn", "activation_sparsity", "activations", "class", "def", "down_proj", "forward", "gate_proj", "hidden_states", "if", "nn", "return", "self", "torch", "up_proj"], "gemma3n/modeling_gemma3n.py:Gemma3nTextMLP._gaussian_topk": ["False", "ModelTextMLP", "Module", "Normal", "Tensor", "True", "_gaussian_topk", "activation_sparsity", "class", "cutoff_x", "def", "device", "dim", "distributions", "dtype", "float32", "functional", "icdf", "inputs", "inputs_mean", "inputs_std", "keepdim", "mean", "nn", "normal", "normal_dist", "relu", "return", "self", "std", "std_multiplier", "target_sparsity_tensor", "tensor", "torch", "type", "unbiased"], "gemma3n/modeling_gemma3n.py:Gemma3nTextAltUp.__init__": ["False", "Linear", "ModelRMSNorm", "ModelTextAltUp", "ModelTextConfig", "Module", "Parameter", "__init__", "altup_num_inputs", "bias", "class", "config", "correct_output_scale", "correction_coefs", "def", "eps", "hidden_size", "modality_router", "nn", "persistent", "prediction_coefs", "register_buffer", "rms_norm_eps", "router_input_scale", "router_norm", "self", "super", "tensor", "torch", "zeros"], "gemma3n/modeling_gemma3n.py:Gemma3nTextAltUp.compute_router_modalities": ["ModelTextAltUp", "Module", "Tensor", "class", "compute_router_modalities", "def", "float", "modality_router", "nn", "return", "routed", "router_input_scale", "router_inputs", "router_norm", "self", "tanh", "torch", "type_as", "x"], "gemma3n/modeling_gemma3n.py:Gemma3nTextAltUp.predict": ["ModelTextAltUp", "Module", "None", "Tensor", "all_coefs", "altup_active_idx", "altup_coef_clip", "altup_num_inputs", "and", "clamp_", "class", "compute_router_modalities", "config", "contiguous", "data", "def", "hidden_states", "if", "is", "matmul", "modalities", "nn", "not", "permute", "predict", "prediction_coefs", "predictions", "reshape", "return", "self", "shape", "torch", "training", "type_as", "weight"], "gemma3n/modeling_gemma3n.py:Gemma3nTextAltUp.correct": ["ModelTextAltUp", "Module", "None", "Tensor", "activated", "all_coefs", "altup_active_idx", "altup_coef_clip", "altup_num_inputs", "clamp_", "class", "compute_router_modalities", "config", "contiguous", "correct", "corrected", "correction_coefs", "data", "def", "if", "innovation", "is", "modalities", "mul", "nn", "not", "permute", "predictions", "repeat", "return", "self", "torch", "type_as", "unsqueeze", "weight"], "gemma3n/modeling_gemma3n.py:Gemma3nTextAltUp.forward": ["ModelTextAltUp", "Module", "Tensor", "class", "correct_output_scale", "corrected", "def", "forward", "nn", "return", "self", "torch", "type_as"], "gemma3n/modeling_gemma3n.py:Gemma3nTextAltUp.scale_corrected_output": ["ModelTextAltUp", "Module", "Tensor", "class", "corrected", "def", "forward", "nn", "return", "scale_corrected_output", "self", "torch"], "gemma3n/modeling_gemma3n.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "gemma3n/modeling_gemma3n.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "gemma3n/modeling_gemma3n.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "head_dim", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softcap", "softmax", "tanh", "to", "torch", "training", "transpose", "tuple", "value", "value_states"], "gemma3n/modeling_gemma3n.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "Tensor", "cos", "def", "int", "return", "rotate_half", "sin", "torch", "unsqueeze", "unsqueeze_dim", "x"], "gemma3n/modeling_gemma3n.py:Gemma3nTextAttention.__init__": ["False", "Linear", "ModelRMSNorm", "ModelTextAttention", "ModelTextConfig", "Module", "None", "True", "__init__", "attention_bias", "attention_dropout", "bias", "class", "config", "def", "dim", "else", "eps", "first_kv_shared_layer_idx", "getattr", "hasattr", "head_dim", "hidden_size", "if", "index", "int", "is_causal", "is_kv_shared_layer", "is_sliding", "k_norm", "k_proj", "kv_shared_layer_index", "layer_idx", "layer_type", "layer_types", "len", "nn", "num_attention_heads", "num_hidden_layers", "num_key_value_groups", "num_key_value_heads", "num_kv_shared_layers", "o_proj", "prev_layers", "q_norm", "q_proj", "rms_norm_eps", "scaling", "self", "sliding_attention", "sliding_window", "store_full_length_kv", "super", "v_norm", "v_proj", "with_scale"], "gemma3n/modeling_gemma3n.py:Gemma3nTextAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "LongTensor", "ModelTextAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_attn_implementation", "and", "apply_rotary_pos_emb", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "cache_kwargs", "cache_position", "class", "config", "contiguous", "cos", "def", "device", "dropout", "eager", "eager_attention_forward", "else", "forward", "hasattr", "head_dim", "hidden_shape", "hidden_states", "if", "input_shape", "is", "is_kv_shared_layer", "k_norm", "k_proj", "key_states", "kv_shared_layer_index", "kwargs", "layer_idx", "nn", "not", "o_proj", "past_key_values", "position_embeddings", "q_norm", "q_proj", "query_states", "reshape", "return", "scaling", "self", "shape", "shared_layers", "sin", "sliding_window", "store_full_length_kv", "to", "torch", "training", "transpose", "tuple", "unsqueeze_dim", "update", "v_norm", "v_proj", "value_states", "view"], "gemma3n/modeling_gemma3n.py:Gemma3nTextDecoderLayer.__init__": ["ACT2FN", "False", "GradientCheckpointingLayer", "Linear", "ModelRMSNorm", "ModelTextAltUp", "ModelTextAttention", "ModelTextConfig", "ModelTextDecoderLayer", "ModelTextLaurelBlock", "ModelTextMLP", "__init__", "act_fn", "altup", "attention_type", "bias", "class", "config", "def", "eps", "hidden_activation", "hidden_size", "hidden_size_per_layer_input", "input_layernorm", "int", "laurel", "layer_idx", "layer_types", "mlp", "nn", "per_layer_input_gate", "per_layer_projection", "post_attention_layernorm", "post_feedforward_layernorm", "post_per_layer_input_norm", "pre_feedforward_layernorm", "rms_norm_eps", "self", "self_attn", "super"], "gemma3n/modeling_gemma3n.py:Gemma3nTextDecoderLayer.forward": ["Cache", "FloatTensor", "GradientCheckpointingLayer", "LongTensor", "ModelTextDecoderLayer", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "act_fn", "active_prediction", "active_prediction_normed", "altup", "altup_active_idx", "altup_correct_scale", "attention_mask", "attn", "attn_ffw", "attn_ffw_laurel_gated", "attn_ffw_norm", "attn_gated", "attn_laurel", "attn_norm", "cache_position", "class", "clone", "config", "correct", "corrected_predictions", "def", "first_prediction", "forward", "hidden_states", "if", "input_layernorm", "kwargs", "laurel", "laurel_output", "math", "mlp", "multiply", "past_key_values", "per_layer_input", "per_layer_input_gate", "per_layer_projection", "position_embeddings", "position_ids", "post_attention_layernorm", "post_feedforward_layernorm", "post_per_layer_input_norm", "pre_feedforward_layernorm", "predict", "predictions", "return", "scale_corrected_output", "self", "self_attn", "sqrt", "torch", "tuple"], "gemma3n/modeling_gemma3n.py:Gemma3nPreTrainedModel._init_weights": ["ModelAudioAttention", "ModelAudioCumulativeGroupNorm", "ModelAudioRelativePositionEmbedding", "ModelPreTrainedModel", "ModelRotaryEmbedding", "ModelTextAltUp", "ModelTextModel", "ModelTextScaledWordEmbedding", "PreTrainedModel", "ROPE_INIT_FUNCTIONS", "_", "_init_weights", "_inv_freq", "_original_inv_freq", "arange", "attention_logits_soft_cap", "channels", "class", "compute_default_rope_parameters", "config", "constant_", "copy_", "correct_output_scale", "create_local_causal_valid_mask", "curr_inv_freq", "def", "default", "elif", "embed_scale", "exp", "f", "float", "for", "functional", "getattr", "gradient_clipping", "hasattr", "head_dim", "hidden_size", "if", "in", "init", "inv_timescales", "isinstance", "layer_type", "layer_types", "local_causal_valid_mask", "log", "log_timescale_increment", "math", "max", "max_timescale", "min_timescale", "module", "nn", "num_timescales", "ones_", "per_dim_scale", "per_layer_input_scale", "per_layer_projection_scale", "q_scale", "r_softplus_0", "rope_init_fn", "rope_type", "router_input_scale", "scalar_embed_scale", "self", "softcap", "softplus", "sqrt", "super", "tensor", "torch", "unsqueeze", "weight", "zeros_"], "gemma3n/modeling_gemma3n.py:Gemma3nRotaryEmbedding.__init__": ["Callable", "False", "ModelRotaryEmbedding", "ModelTextConfig", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "_attention_scaling", "_inv_freq", "_original_inv_freq", "class", "clone", "compute_default_rope_parameters", "config", "continue", "curr_attention_scaling", "curr_inv_freq", "def", "default", "device", "f", "for", "if", "in", "is", "layer_type", "layer_types", "list", "max_position_embeddings", "max_seq_len_cached", "nn", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_params", "rope_type", "self", "set", "setattr", "super"], "gemma3n/modeling_gemma3n.py:Gemma3nRotaryEmbedding.compute_default_rope_parameters": ["ModelRotaryEmbedding", "ModelTextConfig", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "layer_type", "nn", "num_attention_heads", "or", "return", "rope_parameters", "rope_theta", "seq_len", "str", "to", "torch", "tuple"], "gemma3n/modeling_gemma3n.py:Gemma3nRotaryEmbedding.forward": ["False", "ModelRotaryEmbedding", "Module", "None", "_attention_scaling", "_inv_freq", "and", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "f", "float", "forward", "freqs", "getattr", "if", "inv_freq", "inv_freq_expanded", "isinstance", "layer_type", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "gemma3n/modeling_gemma3n.py:Gemma3nTextModel.__init__": ["False", "Linear", "ModelPreTrainedModel", "ModelRMSNorm", "ModelRotaryEmbedding", "ModelTextConfig", "ModelTextDecoderLayer", "ModelTextModel", "ModelTextScaledWordEmbedding", "ModuleList", "_", "__init__", "altup_num_inputs", "altup_projections", "altup_unembed_projections", "bias", "class", "config", "def", "embed_scale", "embed_tokens", "embed_tokens_per_layer", "eps", "for", "gradient_checkpointing", "hidden_size", "hidden_size_per_layer_input", "in", "layer_idx", "layers", "nn", "norm", "num_hidden_layers", "pad_token_id", "padding_idx", "per_layer_input_scale", "per_layer_model_projection", "per_layer_projection_norm", "per_layer_projection_scale", "persistent", "post_init", "range", "register_buffer", "rms_norm_eps", "rotary_emb", "rsqrt", "self", "super", "tensor", "torch", "vocab_size", "vocab_size_per_layer_input"], "gemma3n/modeling_gemma3n.py:Gemma3nTextModel.forward": ["BaseModelOutputWithPast", "Cache", "DynamicCache", "FloatTensor", "LongTensor", "ModelPreTrainedModel", "ModelTextModel", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "ValueError", "You", "altup_num_inputs", "altup_proj", "altup_projections", "altup_unemb_proj", "altup_unembed_projections", "and", "append", "arange", "attention_mask", "attention_type", "bool", "cache_position", "causal_mask", "causal_mask_mapping", "class", "config", "create_causal_mask", "create_sliding_window_causal_mask", "current_hidden_state", "decoder_layer", "def", "device", "dict", "dim", "dtype", "else", "embed_tokens", "epsilon_tensor", "exactly", "for", "forward", "full_attention", "get_per_layer_inputs", "get_seq_length", "hidden_states", "hidden_states_0", "i", "if", "in", "input_embeds", "input_ids", "inputs_embeds", "is", "isinstance", "keepdim", "kwargs", "last_hidden_state", "layer_idx", "layer_type", "layer_types", "layers", "mask_kwargs", "maximum", "mean", "must", "new_magnitude", "norm", "not", "num_hidden_layers", "of", "one", "or", "past_key_values", "past_seen_tokens", "per_layer_input", "per_layer_inputs", "position_embeddings", "position_ids", "project_per_layer_inputs", "r", "raise", "range", "return", "rotary_emb", "self", "shape", "sliding_attention", "specify", "sqrt", "stack", "target_magnitude", "temp_hidden_states", "tensor", "to", "torch", "unsqueeze", "use_cache"], "gemma3n/modeling_gemma3n.py:Gemma3nTextModel.get_per_layer_inputs": ["LongTensor", "ModelPreTrainedModel", "ModelTextModel", "Tensor", "class", "config", "def", "embed_tokens_per_layer", "get_per_layer_inputs", "hidden_size_per_layer_input", "input_ids", "num_hidden_layers", "reshape", "return", "self", "shape", "torch"], "gemma3n/modeling_gemma3n.py:Gemma3nTextModel.project_per_layer_inputs": ["ModelPreTrainedModel", "ModelTextModel", "None", "Tensor", "class", "config", "def", "device", "dtype", "hidden_size_per_layer_input", "if", "inputs_embeds", "is", "num_hidden_layers", "per_layer_input_scale", "per_layer_inputs", "per_layer_model_projection", "per_layer_projection", "per_layer_projection_norm", "per_layer_projection_scale", "project_per_layer_inputs", "reshape", "return", "self", "shape", "to", "torch"], "gemma3n/modeling_gemma3n.py:Gemma3nForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "ModelForCausalLM", "ModelPreTrainedModel", "ModelTextConfig", "ModelTextModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "vocab_size"], "gemma3n/modeling_gemma3n.py:Gemma3nForCausalLM.forward": ["BaseModelOutputWithPast", "Cache", "CausalLMOutputWithPast", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForCausalLM", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "else", "final_logit_softcapping", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "outputs", "past_key_values", "position_ids", "r", "return", "self", "slice", "slice_indices", "tanh", "torch", "use_cache", "vocab_size"], "gemma3n/modeling_gemma3n.py:Gemma3nMultimodalEmbedder.__init__": ["Embedding", "False", "Linear", "ModelAudioConfig", "ModelMultimodalEmbedder", "ModelRMSNorm", "ModelTextConfig", "ModelVisionConfig", "Module", "__init__", "bias", "class", "def", "embedding", "embedding_post_projection_norm", "embedding_projection", "eps", "hard_embedding_norm", "hidden_size", "multimodal_config", "multimodal_hidden_size", "nn", "rms_norm_eps", "self", "soft_embedding_norm", "super", "text_config", "text_hidden_size", "vocab_offset", "vocab_size", "with_scale"], "gemma3n/modeling_gemma3n.py:Gemma3nMultimodalEmbedder.forward": ["LongTensor", "ModelMultimodalEmbedder", "Module", "None", "Tensor", "ValueError", "You", "class", "def", "else", "emb_norm", "emb_norm_proj", "embedding", "embedding_post_projection_norm", "embedding_projection", "exactly", "forward", "hard_emb", "hard_embedding_norm", "if", "input_ids", "inputs_embeds", "is", "must", "nn", "not", "of", "one", "or", "raise", "return", "self", "soft_embedding_norm", "specify", "torch", "vocab_offset"], "gemma3n/modeling_gemma3n.py:Gemma3nModel.__init__": ["AutoModel", "ModelConfig", "ModelModel", "ModelMultimodalEmbedder", "ModelPreTrainedModel", "None", "__init__", "audio_config", "audio_tower", "class", "config", "def", "else", "embed_audio", "embed_vision", "from_config", "if", "is", "language_model", "not", "pad_token_id", "post_init", "self", "super", "text_config", "vision_config", "vision_tower", "vocab_size", "vocab_size_per_layer_input"], "gemma3n/modeling_gemma3n.py:Gemma3nModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "language_model", "return", "self"], "gemma3n/modeling_gemma3n.py:Gemma3nModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "language_model", "self", "set_input_embeddings", "value"], "gemma3n/modeling_gemma3n.py:Gemma3nModel.get_image_features": ["False", "ModelModel", "ModelPreTrainedModel", "Tensor", "True", "class", "config", "def", "do_pooling", "embed_vision", "get_image_features", "hidden_size", "inputs_embeds", "last_hidden_state", "permute", "pixel_values", "reshape", "return", "return_dict", "self", "shape", "torch", "vision_config", "vision_outputs", "vision_soft_tokens_per_image", "vision_tower"], "gemma3n/modeling_gemma3n.py:Gemma3nModel.get_placeholder_mask": ["Audio", "FloatTensor", "Image", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "ValueError", "all", "and", "audio_features", "audio_token_id", "class", "config", "def", "device", "do", "dtype", "else", "expand_as", "f", "features", "get_input_embeddings", "get_placeholder_mask", "if", "image", "image_features", "image_token_id", "input_ids", "inputs_embeds", "is", "long", "match", "n_audio_tokens", "n_image_tokens", "not", "numel", "raise", "return", "self", "shape", "special_audio_mask", "special_image_mask", "sum", "tensor", "to", "tokens", "torch", "unsqueeze"], "gemma3n/modeling_gemma3n.py:Gemma3nModel.forward": ["Cache", "FloatTensor", "LongTensor", "ModelCausalLMOutputWithPast", "ModelModel", "ModelModelOutputWithPast", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "ValueError", "You", "_", "and", "attention_mask", "attentions", "audio_batch_size", "audio_embed_dim", "audio_embeds", "audio_features", "audio_hidden_states", "audio_input_ids", "audio_mask", "audio_padding_embs", "audio_padding_toks", "audio_seq_len", "audio_soft_tokens_per_image", "bool", "cache_position", "cat", "class", "config", "def", "device", "dim", "dtype", "dummy_audio_token_id", "dummy_vision_token_id", "else", "embed_audio", "embed_vision", "exactly", "expand", "expand_as", "expanded_audio_mask", "expanded_vision_mask", "extra_padding_features", "extra_padding_tokens", "forward", "get_audio_features", "get_image_features", "get_input_embeddings", "get_per_layer_inputs", "get_placeholder_mask", "hidden_states", "if", "image_features", "image_hidden_states", "input_features", "input_features_mask", "input_ids", "inputs_embeds", "is", "labels", "language_model", "last_hidden_state", "lm_kwargs", "logical_and", "long", "masked_scatter", "must", "not", "of", "one", "or", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "per_layer_inputs", "per_layer_inputs_mask", "per_layer_inputs_tokens", "pixel_values", "position_ids", "r", "raise", "return", "return_dict", "self", "shape", "special_audio_mask", "special_image_mask", "specify", "tensor", "to", "token_type_ids", "torch", "unsqueeze", "use_cache", "vision_embeds", "vision_input_ids", "vision_mask", "vocab_offset", "vocab_size", "vocab_size_per_layer_input", "where", "zeros_like"], "gemma3n/modeling_gemma3n.py:Gemma3nModel.get_audio_features": ["ModelModel", "ModelPreTrainedModel", "Tensor", "audio_mask", "audio_outputs", "audio_tower", "class", "def", "embed_audio", "get_audio_features", "input_features", "input_features_mask", "inputs_embeds", "return", "self", "torch", "tuple"], "gemma3n/modeling_gemma3n.py:Gemma3nForConditionalGeneration.__init__": ["False", "GenerationMixin", "Linear", "ModelConfig", "ModelForConditionalGeneration", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "text_config", "vocab_size"], "gemma3n/modeling_gemma3n.py:Gemma3nForConditionalGeneration.get_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "model", "return", "self"], "gemma3n/modeling_gemma3n.py:Gemma3nForConditionalGeneration.set_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "model", "self", "set_input_embeddings", "value"], "gemma3n/modeling_gemma3n.py:Gemma3nForConditionalGeneration.get_image_features": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "get_image_features", "model", "pixel_values", "return", "self"], "gemma3n/modeling_gemma3n.py:Gemma3nForConditionalGeneration.forward": ["Cache", "CrossEntropyLoss", "FloatTensor", "GenerationMixin", "LongTensor", "ModelCausalLMOutputWithPast", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "attention_mask", "attentions", "audio_hidden_states", "bool", "cache_position", "class", "config", "contiguous", "def", "device", "else", "final_logit_softcapping", "flat_labels", "flat_logits", "float", "forward", "get_text_config", "hidden_states", "if", "image_hidden_states", "input_features", "input_features_mask", "input_ids", "inputs_embeds", "int", "is", "isinstance", "labels", "last_hidden_state", "lm_head", "lm_kwargs", "logits", "logits_to_keep", "loss", "loss_fct", "model", "nn", "not", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "pixel_values", "position_ids", "r", "return", "return_dict", "self", "shape", "shift_attention_mask", "shift_labels", "shift_logits", "slice", "slice_indices", "tanh", "text_config", "to", "token_type_ids", "torch", "use_cache", "view", "vocab_size"], "gemma3n/modeling_gemma3n.py:Gemma3nForConditionalGeneration.prepare_inputs_for_generation": ["False", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "True", "attention_mask", "cache_position", "class", "def", "if", "input_features", "input_features_mask", "input_ids", "inputs_embeds", "is_first_iteration", "kwargs", "labels", "logits_to_keep", "model_inputs", "not", "or", "past_key_values", "pixel_values", "position_ids", "prepare_inputs_for_generation", "return", "self", "super", "token_type_ids", "use_cache"], "git/modeling_git.py:token_type_ids_mask_function": ["Callable", "Model_type_ids", "Model_type_ids_at_kv_idx", "Model_type_ids_at_q_idx", "Model_type_ids_mask_function", "None", "Tensor", "batch_idx", "bool", "def", "head_idx", "if", "image_group_ids", "image_group_ids_at_kv_idx", "image_group_ids_at_q_idx", "inner_mask", "int", "is", "is_image_block", "kv_idx", "q_idx", "return", "safe_kv_idx", "safe_q_idx", "same_image_block", "shape", "torch", "where"], "git/modeling_git.py:create_causal_mask_mapping": ["Cache", "False", "FloatTensor", "Model_causal_mask_mapping", "Model_masks_for_generate", "None", "PreTrainedConfig", "Tensor", "ValueError", "a", "and", "as", "attention_mask", "bool", "cache_position", "config", "cumsum", "def", "device", "dict", "dim", "else", "functional", "get_text_config", "if", "image_group_ids", "input", "input_embeds", "int", "is", "is_first_iteration", "is_image", "is_initialized", "is_previous_image", "is_training", "kwargs", "mask_kwargs", "model", "new_image_start", "nn", "not", "or", "or_mask_function", "pad", "past_key_values", "pixel_values", "position_ids", "raise", "required", "return", "to", "token_type_ids", "token_type_ids_mask_function", "torch", "training", "value", "when", "where"], "git/modeling_git.py:GitEmbeddings.__init__": ["Dropout", "Embedding", "False", "LayerNorm", "ModelEmbeddings", "Module", "__init__", "arange", "class", "config", "def", "dropout", "eps", "expand", "hidden_dropout_prob", "hidden_size", "layer_norm_eps", "max_position_embeddings", "nn", "pad_token_id", "padding_idx", "persistent", "position_embeddings", "position_ids", "register_buffer", "self", "super", "torch", "vocab_size", "word_embeddings"], "git/modeling_git.py:GitEmbeddings.forward": ["FloatTensor", "LayerNorm", "LongTensor", "ModelEmbeddings", "Module", "None", "Tensor", "class", "def", "dropout", "else", "embeddings", "forward", "if", "input_ids", "input_shape", "inputs_embeds", "int", "is", "nn", "not", "past_key_values_length", "position_embeddings", "position_ids", "return", "self", "seq_length", "size", "torch", "word_embeddings"], "git/modeling_git.py:GitSelfAttention.__init__": ["Dropout", "Instantiating", "Linear", "ModelSelfAttention", "Module", "None", "Please", "The", "ValueError", "__class__", "__init__", "__name__", "a", "all_head_size", "and", "attention", "attention_head_size", "attention_probs_dropout_prob", "caching", "call", "class", "config", "creating", "def", "dropout", "during", "embedding_size", "errors", "f", "forward", "hasattr", "heads", "hidden", "hidden_size", "if", "image_patch_tokens", "image_size", "int", "is", "key", "layer_idx", "lead", "logger", "make", "multiple", "nn", "not", "num_attention_heads", "num_image_with_embedding", "number", "of", "passing", "patch_size", "provide", "query", "raise", "recommended", "self", "size", "super", "sure", "the", "this", "to", "used", "value", "vision_config", "warning_once", "when", "will", "without"], "git/modeling_git.py:GitSelfAttention.forward": ["Cache", "FloatTensor", "ModelSelfAttention", "Module", "None", "Tensor", "all_head_size", "attention_head_size", "attention_mask", "attention_probs", "attention_scores", "batch_size", "cache_kwargs", "cache_position", "class", "context_layer", "contiguous", "def", "dim", "dropout", "forward", "functional", "hidden_states", "if", "is", "key", "key_layer", "layer_idx", "math", "matmul", "new_context_layer_shape", "nn", "not", "num_attention_heads", "past_key_values", "permute", "query", "query_layer", "return", "self", "shape", "size", "softmax", "sqrt", "torch", "transpose", "tuple", "update", "value", "value_layer", "view"], "git/modeling_git.py:GitSelfOutput.__init__": ["Dropout", "LayerNorm", "Linear", "ModelSelfOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "eps", "hidden_dropout_prob", "hidden_size", "layer_norm_eps", "nn", "self", "super"], "git/modeling_git.py:GitSelfOutput.forward": ["LayerNorm", "ModelSelfOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "git/modeling_git.py:GitAttention.__init__": ["ModelAttention", "ModelSelfOutput", "Model_SELF_ATTENTION_CLASSES", "Module", "None", "__init__", "_attn_implementation", "class", "config", "def", "layer_idx", "nn", "output", "self", "super"], "git/modeling_git.py:GitAttention.forward": ["Cache", "False", "FloatTensor", "ModelAttention", "Module", "None", "Tensor", "attention_mask", "attention_output", "attn_output", "bool", "cache_position", "class", "def", "forward", "hidden_states", "nn", "output", "output_attentions", "past_key_values", "return", "self", "self_attn_weights", "torch", "tuple"], "git/modeling_git.py:GitIntermediate.__init__": ["ACT2FN", "Linear", "ModelIntermediate", "Module", "__init__", "class", "config", "def", "dense", "else", "hidden_act", "hidden_size", "if", "intermediate_act_fn", "intermediate_size", "isinstance", "nn", "self", "str", "super"], "git/modeling_git.py:GitIntermediate.forward": ["ModelIntermediate", "Module", "Tensor", "class", "def", "dense", "forward", "hidden_states", "intermediate_act_fn", "nn", "return", "self", "torch"], "git/modeling_git.py:GitOutput.__init__": ["Dropout", "LayerNorm", "Linear", "ModelOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "eps", "hidden_dropout_prob", "hidden_size", "intermediate_size", "layer_norm_eps", "nn", "self", "super"], "git/modeling_git.py:GitOutput.forward": ["LayerNorm", "ModelOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "git/modeling_git.py:GitLayer.__init__": ["GradientCheckpointingLayer", "ModelAttention", "ModelIntermediate", "ModelLayer", "ModelOutput", "None", "__init__", "attention", "chunk_size_feed_forward", "class", "config", "def", "intermediate", "layer_idx", "output", "self", "seq_len_dim", "super"], "git/modeling_git.py:GitLayer.forward": ["Cache", "False", "FloatTensor", "GradientCheckpointingLayer", "ModelLayer", "None", "Tensor", "apply_chunking_to_forward", "attention", "attention_mask", "attention_output", "bool", "cache_position", "chunk_size_feed_forward", "class", "def", "feed_forward_chunk", "forward", "hidden_states", "layer_output", "output_attentions", "past_key_values", "return", "self", "self_attention_weights", "seq_len_dim", "torch", "tuple"], "git/modeling_git.py:GitLayer.feed_forward_chunk": ["GradientCheckpointingLayer", "ModelLayer", "attention_output", "class", "def", "feed_forward_chunk", "intermediate", "intermediate_output", "layer_output", "output", "return", "self"], "git/modeling_git.py:GitEncoder.__init__": ["False", "ModelEncoder", "ModelLayer", "Module", "ModuleList", "__init__", "class", "config", "def", "for", "gradient_checkpointing", "i", "in", "layer", "nn", "num_hidden_layers", "range", "self", "super"], "git/modeling_git.py:GitEncoder.forward": ["BaseModelOutputWithPast", "Cache", "DynamicCache", "False", "FloatTensor", "ModelEncoder", "Module", "None", "Setting", "Tensor", "True", "all_hidden_states", "all_self_attentions", "and", "attention_mask", "attentions", "bool", "cache_position", "checkpointing", "class", "config", "def", "else", "enumerate", "for", "forward", "gradient", "gradient_checkpointing", "hidden_states", "i", "if", "in", "incompatible", "is", "last_hidden_state", "layer", "layer_module", "layer_outputs", "logger", "nn", "not", "output_attentions", "output_hidden_states", "past_key_values", "return", "return_dict", "self", "torch", "training", "tuple", "use_cache", "v", "warning_once", "with"], "git/modeling_git.py:GitPreTrainedModel._init_weights": ["Embedding", "False", "LayerNorm", "Linear", "ModelEmbeddings", "ModelPreTrainedModel", "ModelVisionEmbeddings", "None", "PreTrainedModel", "_init_weights", "_is_hf_initialized", "and", "arange", "bias", "class", "class_embedding", "config", "copy_", "def", "elif", "expand", "getattr", "if", "init", "initializer_range", "is", "isinstance", "mean", "module", "nn", "normal_", "not", "ones_", "padding_idx", "patch_embedding", "position_embedding", "position_ids", "self", "shape", "std", "torch", "weight", "zeros_"], "git/modeling_git.py:GitVisionEmbeddings.__init__": ["Conv2d", "Embedding", "False", "ModelVisionConfig", "ModelVisionEmbeddings", "Module", "Parameter", "__init__", "arange", "bias", "class", "class_embedding", "config", "def", "embed_dim", "expand", "hidden_size", "image_size", "in_channels", "kernel_size", "nn", "num_channels", "num_patches", "num_positions", "out_channels", "patch_embedding", "patch_size", "persistent", "position_embedding", "position_ids", "randn", "register_buffer", "self", "stride", "super", "torch"], "git/modeling_git.py:GitVisionEmbeddings.interpolate_pos_encoding": ["False", "ModelVisionEmbeddings", "Module", "Tensor", "align_corners", "and", "bicubic", "cat", "class", "class_pos_embed", "def", "dim", "embeddings", "functional", "height", "if", "int", "interpolate", "interpolate_pos_encoding", "is_tracing", "jit", "mode", "new_height", "new_width", "nn", "not", "num_patches", "num_positions", "patch_pos_embed", "patch_size", "permute", "position_embedding", "position_ids", "reshape", "return", "self", "shape", "size", "sqrt_num_positions", "torch", "torch_int", "unsqueeze", "view", "weight", "width"], "git/modeling_git.py:GitVisionEmbeddings.forward": ["False", "FloatTensor", "Input", "ModelVisionEmbeddings", "Module", "Tensor", "ValueError", "_", "and", "batch_size", "cat", "class", "class_embedding", "class_embeds", "def", "dim", "doesn", "dtype", "else", "embeddings", "expand", "f", "flatten", "forward", "height", "if", "image", "image_size", "interpolate_pos_encoding", "match", "model", "nn", "not", "or", "patch_embedding", "patch_embeds", "pixel_values", "position_embedding", "position_ids", "raise", "return", "self", "shape", "size", "t", "target_dtype", "to", "torch", "transpose", "weight", "width"], "git/modeling_git.py:GitVisionMLP.__init__": ["ACT2FN", "Linear", "ModelVisionMLP", "Module", "__init__", "activation_fn", "class", "config", "def", "fc1", "fc2", "hidden_act", "hidden_size", "intermediate_size", "nn", "self", "super"], "git/modeling_git.py:GitVisionMLP.forward": ["ModelVisionMLP", "Module", "Tensor", "activation_fn", "class", "def", "fc1", "fc2", "forward", "hidden_states", "nn", "return", "self", "torch"], "git/modeling_git.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "attention_mask", "attn_output", "attn_weights", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "softmax", "to", "torch", "training", "transpose", "value"], "git/modeling_git.py:GitVisionAttention.__init__": ["False", "Linear", "ModelVisionAttention", "Module", "ValueError", "__init__", "and", "attention_dropout", "be", "by", "class", "config", "def", "divisible", "dropout", "embed_dim", "f", "got", "head_dim", "hidden_size", "if", "is_causal", "k_proj", "must", "nn", "num_attention_heads", "num_heads", "out_proj", "q_proj", "raise", "scale", "self", "super", "v_proj"], "git/modeling_git.py:GitVisionAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "False", "ModelVisionAttention", "Module", "None", "Tensor", "_attn_implementation", "and", "attention_interface", "attention_mask", "attn_output", "attn_weights", "batch_size", "bool", "causal_attention_mask", "class", "config", "contiguous", "def", "dropout", "eager", "eager_attention_forward", "elif", "else", "embed_dim", "flash_attention_2", "forward", "head_dim", "hidden_states", "if", "is", "is_causal", "k_proj", "keys", "nn", "not", "num_heads", "out_proj", "output_attentions", "q_proj", "queries", "reshape", "return", "scale", "scaling", "self", "seq_length", "shape", "torch", "training", "transpose", "tuple", "v_proj", "values", "view"], "git/modeling_git.py:GitVisionEncoderLayer.__init__": ["GradientCheckpointingLayer", "LayerNorm", "ModelVisionAttention", "ModelVisionConfig", "ModelVisionEncoderLayer", "ModelVisionMLP", "__init__", "class", "config", "def", "embed_dim", "eps", "hidden_size", "layer_norm1", "layer_norm2", "layer_norm_eps", "mlp", "nn", "self", "self_attn", "super"], "git/modeling_git.py:GitVisionEncoderLayer.forward": ["False", "FloatTensor", "GradientCheckpointingLayer", "ModelVisionEncoderLayer", "None", "Tensor", "attention_mask", "attn_weights", "bool", "causal_attention_mask", "class", "def", "forward", "hidden_states", "if", "layer_norm1", "layer_norm2", "mlp", "output_attentions", "outputs", "residual", "return", "self", "self_attn", "torch", "tuple"], "git/modeling_git.py:GitVisionEncoder.__init__": ["False", "ModelVisionConfig", "ModelVisionEncoder", "ModelVisionEncoderLayer", "Module", "ModuleList", "_", "__init__", "class", "config", "def", "for", "gradient_checkpointing", "in", "layers", "nn", "num_hidden_layers", "range", "self", "super"], "git/modeling_git.py:GitVisionEncoder.forward": ["BaseModelOutput", "ModelVisionEncoder", "Module", "None", "Tensor", "all_attentions", "attention_mask", "attentions", "bool", "causal_attention_mask", "class", "config", "def", "else", "encoder_layer", "encoder_states", "enumerate", "for", "forward", "hidden_states", "idx", "if", "in", "inputs_embeds", "is", "last_hidden_state", "layer_outputs", "layers", "nn", "not", "output_attentions", "output_hidden_states", "r", "return", "return_dict", "self", "torch", "tuple", "use_return_dict"], "git/modeling_git.py:GitVisionTransformer.__init__": ["LayerNorm", "ModelVisionConfig", "ModelVisionEmbeddings", "ModelVisionEncoder", "ModelVisionTransformer", "Module", "__init__", "class", "config", "def", "embed_dim", "embeddings", "encoder", "eps", "hidden_size", "layer_norm_eps", "nn", "post_layernorm", "pre_layrnorm", "self", "super"], "git/modeling_git.py:GitVisionTransformer.forward": ["BaseModelOutput", "False", "FloatTensor", "ModelVisionTransformer", "Module", "None", "ValueError", "You", "attentions", "bool", "class", "config", "def", "else", "embeddings", "encoder", "encoder_outputs", "forward", "have", "hidden_states", "if", "inputs_embeds", "interpolate_pos_encoding", "is", "last_hidden_state", "nn", "not", "output_attentions", "output_hidden_states", "pixel_values", "post_layernorm", "pre_layrnorm", "raise", "return", "return_dict", "self", "specify", "to", "torch", "tuple", "use_return_dict"], "git/modeling_git.py:GitVisionModel.__init__": ["ModelPreTrainedModel", "ModelVisionConfig", "ModelVisionModel", "ModelVisionTransformer", "__init__", "class", "config", "def", "post_init", "self", "super", "vision_model"], "git/modeling_git.py:GitVisionModel.get_input_embeddings": ["ModelPreTrainedModel", "ModelVisionModel", "Module", "class", "def", "embeddings", "get_input_embeddings", "nn", "patch_embedding", "return", "self", "vision_model"], "git/modeling_git.py:GitVisionModel.forward": ["BaseModelOutput", "False", "FloatTensor", "ModelPreTrainedModel", "ModelVisionModel", "None", "bool", "class", "config", "def", "else", "forward", "if", "interpolate_pos_encoding", "is", "kwargs", "not", "output_attentions", "output_hidden_states", "pixel_values", "r", "return", "return_dict", "self", "torch", "tuple", "use_return_dict", "vision_model"], "git/modeling_git.py:GitProjection.__init__": ["LayerNorm", "Linear", "ModelConfig", "ModelProjection", "Module", "Sequential", "__init__", "class", "config", "def", "eps", "hidden_size", "layer_norm_eps", "nn", "self", "super", "vision_config", "visual_projection"], "git/modeling_git.py:GitProjection.forward": ["ModelProjection", "Module", "Tensor", "class", "def", "embeddings", "forward", "nn", "return", "self", "torch", "visual_projection"], "git/modeling_git.py:GitModel.__init__": ["ModelEmbeddings", "ModelEncoder", "ModelModel", "ModelPreTrainedModel", "ModelProjection", "ModelVisionModel", "None", "Parameter", "ParameterList", "_", "__init__", "class", "config", "def", "embeddings", "encoder", "for", "hidden_size", "if", "image_encoder", "img_temporal_embedding", "in", "is", "nn", "not", "num_image_with_embedding", "post_init", "range", "self", "super", "torch", "vision_config", "visual_projection", "zeros"], "git/modeling_git.py:GitModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "return", "self", "word_embeddings"], "git/modeling_git.py:GitModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "self", "set_input_embeddings", "value", "word_embeddings"], "git/modeling_git.py:GitModel.forward": ["BaseModelOutputWithPast", "BaseModelOutputWithPooling", "Cache", "False", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "ValueError", "You", "and", "append", "arange", "at", "attention_mask", "attentions", "be", "bool", "both", "cache_position", "cannot", "cat", "causal_mask", "class", "config", "create_causal_mask_mapping", "def", "device", "dim", "dtype", "elif", "else", "embedding_output", "embeddings", "encoder", "encoder_outputs", "extended_attention_mask", "for", "forward", "frame_idx", "get_seq_length", "hidden_states", "if", "image_encoder", "image_token_type_ids", "img_temporal_embedding", "in", "input_ids", "inputs_embeds", "int", "interpolate_pos_encoding", "is", "isinstance", "kwargs", "last_hidden_state", "must", "ndim", "not", "of", "ones", "ones_like", "or", "output_attentions", "output_hidden_states", "past_key_values", "past_key_values_length", "pixel_values", "position_ids", "projected_visual_features", "r", "raise", "range", "rank", "repeat", "return", "return_dict", "same", "self", "sequence_output", "shape", "size", "specify", "tensor", "the", "time", "token_type_ids", "torch", "tuple", "use_cache", "use_return_dict", "visual_features", "visual_features_frame", "visual_projection", "zeros_like"], "git/modeling_git.py:GitForCausalLM.__init__": ["GenerationMixin", "Linear", "Model", "ModelForCausalLM", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "config", "def", "hidden_size", "nn", "output", "post_init", "self", "super", "vocab_size"], "git/modeling_git.py:GitForCausalLM.get_output_embeddings": ["GenerationMixin", "ModelForCausalLM", "ModelPreTrainedModel", "class", "def", "get_output_embeddings", "output", "return", "self"], "git/modeling_git.py:GitForCausalLM.set_output_embeddings": ["GenerationMixin", "ModelForCausalLM", "ModelPreTrainedModel", "class", "def", "new_embeddings", "output", "self", "set_output_embeddings"], "git/modeling_git.py:GitForCausalLM.forward": ["Cache", "CausalLMOutputWithPast", "False", "GenerationMixin", "Model", "ModelForCausalLM", "ModelPreTrainedModel", "None", "Tensor", "attention", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "contiguous", "def", "else", "encoder", "forward", "hidden_states", "if", "image_patch_tokens", "input_ids", "inputs_embeds", "int", "interpolate_pos_encoding", "is", "isinstance", "kwargs", "labels", "layer", "loModels", "loModels_to_keep", "loss", "loss_function", "not", "num_image_tokens", "output", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "pixel_values", "position_ids", "r", "return", "return_dict", "self", "shifted_loModels", "slice", "slice_indices", "torch", "tuple", "use_cache", "use_return_dict", "view", "vocab_size"], "git/modeling_git.py:GitForCausalLM.prepare_inputs_for_generation": ["False", "GenerationMixin", "ModelForCausalLM", "ModelPreTrainedModel", "None", "attention_mask", "cache_position", "class", "def", "else", "for", "get_seq_length", "if", "in", "input_ids", "input_shape", "is", "is_first_iteration", "items", "key", "kwargs", "model_inputs", "new_ones", "not", "or", "past_key_values", "past_length", "pixel_values", "prepare_inputs_for_generation", "remove_prefix_length", "return", "self", "shape", "use_cache", "value"], "glm/modeling_glm.py:GlmMLP.__init__": ["ACT2FN", "False", "Linear", "ModelMLP", "Module", "__init__", "activation_fn", "bias", "class", "config", "def", "down_proj", "gate_up_proj", "hidden_act", "hidden_size", "intermediate_size", "nn", "self", "super"], "glm/modeling_glm.py:GlmMLP.forward": ["FloatTensor", "ModelMLP", "Module", "activation_fn", "chunk", "class", "def", "dim", "down_proj", "forward", "gate", "gate_up_proj", "hidden_states", "nn", "return", "self", "torch", "up_states"], "glm/modeling_glm.py:GlmRotaryEmbedding.__init__": ["Callable", "False", "ModelConfig", "ModelRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "glm/modeling_glm.py:GlmRotaryEmbedding.compute_default_rope_parameters": ["ModelConfig", "ModelRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "get", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "partial_rotary_factor", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "glm/modeling_glm.py:GlmRotaryEmbedding.forward": ["False", "ModelRotaryEmbedding", "Module", "None", "and", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "glm/modeling_glm.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "glm/modeling_glm.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value", "value_states"], "glm/modeling_glm.py:rotate_half": ["Model_half", "def", "dim", "flatten", "return", "stack", "torch", "x", "x1", "x2"], "glm/modeling_glm.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cat", "cos", "def", "dim", "k", "k_embed", "k_pass", "k_rot", "q", "q_embed", "q_pass", "q_rot", "repeat_interleave", "return", "rotary_dim", "rotate_half", "shape", "sin", "torch", "unsqueeze", "unsqueeze_dim"], "glm/modeling_glm.py:GlmAttention.__init__": ["False", "Linear", "ModelAttention", "ModelConfig", "Module", "None", "True", "__init__", "attention_bias", "attention_dropout", "bias", "class", "config", "def", "getattr", "head_dim", "hidden_size", "int", "is_causal", "k_proj", "layer_idx", "nn", "num_attention_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "q_proj", "scaling", "self", "super", "v_proj"], "glm/modeling_glm.py:GlmAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "LongTensor", "ModelAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_attn_implementation", "apply_rotary_pos_emb", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "cache_kwargs", "cache_position", "class", "config", "contiguous", "cos", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "head_dim", "hidden_shape", "hidden_states", "if", "input_shape", "is", "k_proj", "key_states", "kwargs", "layer_idx", "nn", "not", "o_proj", "past_key_values", "position_embeddings", "q_proj", "query_states", "reshape", "return", "scaling", "self", "shape", "sin", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "view"], "glm/modeling_glm.py:GlmRMSNorm.__init__": ["ModelRMSNorm", "Module", "Parameter", "__init__", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "glm/modeling_glm.py:GlmRMSNorm.forward": ["ModelRMSNorm", "Module", "True", "class", "def", "dtype", "float32", "forward", "hidden_states", "input_dtype", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "glm/modeling_glm.py:GlmRMSNorm.extra_repr": ["ModelRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "variance_epsilon", "weight"], "glm/modeling_glm.py:GlmDecoderLayer.__init__": ["GradientCheckpointingLayer", "ModelAttention", "ModelConfig", "ModelDecoderLayer", "ModelMLP", "ModelRMSNorm", "__init__", "class", "config", "def", "eps", "hidden_size", "input_layernorm", "int", "layer_idx", "mlp", "post_attention_layernorm", "rms_norm_eps", "self", "self_attn", "super"], "glm/modeling_glm.py:GlmDecoderLayer.forward": ["Cache", "False", "GradientCheckpointingLayer", "LongTensor", "ModelDecoderLayer", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "attention_mask", "bool", "cache_position", "class", "def", "forward", "hidden_states", "input_layernorm", "kwargs", "mlp", "past_key_values", "position_embeddings", "position_ids", "post_attention_layernorm", "residual", "return", "self", "self_attn", "torch", "tuple", "use_cache"], "glm/modeling_glm.py:GlmModel.__init__": ["Embedding", "False", "ModelConfig", "ModelDecoderLayer", "ModelModel", "ModelPreTrainedModel", "ModelRMSNorm", "ModelRotaryEmbedding", "ModuleList", "__init__", "class", "config", "def", "embed_tokens", "eps", "for", "gradient_checkpointing", "hidden_size", "in", "layer_idx", "layers", "nn", "norm", "num_hidden_layers", "pad_token_id", "padding_idx", "post_init", "range", "rms_norm_eps", "rotary_emb", "self", "super", "vocab_size"], "glm/modeling_glm.py:GlmModel.forward": ["BaseModelOutputWithPast", "Cache", "DynamicCache", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "and", "arange", "attention_mask", "bool", "cache_position", "causal_mask", "class", "config", "create_causal_mask", "decoder_layer", "def", "device", "else", "embed_tokens", "exactly", "for", "forward", "get_seq_length", "hidden_states", "if", "in", "input_embeds", "input_ids", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layers", "must", "norm", "not", "num_hidden_layers", "of", "one", "or", "past_key_values", "past_seen_tokens", "position_embeddings", "position_ids", "raise", "return", "rotary_emb", "self", "shape", "specify", "torch", "unsqueeze", "use_cache"], "glm/modeling_glm.py:GlmForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "ModelForCausalLM", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "vocab_size"], "glm/modeling_glm.py:GlmForCausalLM.forward": ["BaseModelOutputWithPast", "Cache", "CausalLMOutputWithPast", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForCausalLM", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "outputs", "past_key_values", "position_ids", "r", "return", "self", "slice", "slice_indices", "torch", "use_cache", "vocab_size"], "glm4/modeling_glm4.py:Glm4MLP.__init__": ["ACT2FN", "False", "Linear", "ModelMLP", "Module", "__init__", "activation_fn", "bias", "class", "config", "def", "down_proj", "gate_up_proj", "hidden_act", "hidden_size", "intermediate_size", "nn", "self", "super"], "glm4/modeling_glm4.py:Glm4MLP.forward": ["FloatTensor", "ModelMLP", "Module", "activation_fn", "chunk", "class", "def", "dim", "down_proj", "forward", "gate", "gate_up_proj", "hidden_states", "nn", "return", "self", "torch", "up_states"], "glm4/modeling_glm4.py:Glm4DecoderLayer.__init__": ["GradientCheckpointingLayer", "ModelAttention", "ModelConfig", "ModelDecoderLayer", "ModelMLP", "ModelRMSNorm", "__init__", "class", "config", "def", "eps", "hidden_size", "input_layernorm", "int", "layer_idx", "mlp", "post_attention_layernorm", "post_mlp_layernorm", "post_self_attn_layernorm", "rms_norm_eps", "self", "self_attn", "super"], "glm4/modeling_glm4.py:Glm4DecoderLayer.forward": ["Cache", "False", "FlashAttentionKwargs", "FloatTensor", "GradientCheckpointingLayer", "LongTensor", "ModelDecoderLayer", "None", "Tensor", "Unpack", "_", "attention_mask", "bool", "cache_position", "class", "def", "forward", "hidden_states", "input_layernorm", "kwargs", "mlp", "past_key_values", "position_embeddings", "position_ids", "post_attention_layernorm", "post_mlp_layernorm", "post_self_attn_layernorm", "residual", "return", "self", "self_attn", "torch", "tuple", "use_cache"], "glm4/modeling_glm4.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "glm4/modeling_glm4.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value", "value_states"], "glm4/modeling_glm4.py:rotate_half": ["Model_half", "def", "dim", "flatten", "return", "stack", "torch", "x", "x1", "x2"], "glm4/modeling_glm4.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cat", "cos", "def", "dim", "k", "k_embed", "k_pass", "k_rot", "q", "q_embed", "q_pass", "q_rot", "repeat_interleave", "return", "rotary_dim", "rotate_half", "shape", "sin", "torch", "unsqueeze", "unsqueeze_dim"], "glm4/modeling_glm4.py:Glm4Attention.__init__": ["False", "Linear", "ModelAttention", "ModelConfig", "Module", "None", "True", "__init__", "attention_bias", "attention_dropout", "bias", "class", "config", "def", "getattr", "head_dim", "hidden_size", "int", "is_causal", "k_proj", "layer_idx", "nn", "num_attention_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "q_proj", "scaling", "self", "super", "v_proj"], "glm4/modeling_glm4.py:Glm4Attention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "LongTensor", "ModelAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_attn_implementation", "apply_rotary_pos_emb", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "cache_kwargs", "cache_position", "class", "config", "contiguous", "cos", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "head_dim", "hidden_shape", "hidden_states", "if", "input_shape", "is", "k_proj", "key_states", "kwargs", "layer_idx", "nn", "not", "o_proj", "past_key_values", "position_embeddings", "q_proj", "query_states", "reshape", "return", "scaling", "self", "shape", "sin", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "view"], "glm4/modeling_glm4.py:Glm4RotaryEmbedding.__init__": ["Callable", "False", "ModelConfig", "ModelRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "glm4/modeling_glm4.py:Glm4RotaryEmbedding.compute_default_rope_parameters": ["ModelConfig", "ModelRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "get", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "partial_rotary_factor", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "glm4/modeling_glm4.py:Glm4RotaryEmbedding.forward": ["False", "ModelRotaryEmbedding", "Module", "None", "and", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "glm4/modeling_glm4.py:Glm4RMSNorm.__init__": ["ModelRMSNorm", "Module", "Parameter", "__init__", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "glm4/modeling_glm4.py:Glm4RMSNorm.forward": ["ModelRMSNorm", "Module", "True", "class", "def", "dtype", "float32", "forward", "hidden_states", "input_dtype", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "glm4/modeling_glm4.py:Glm4RMSNorm.extra_repr": ["ModelRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "variance_epsilon", "weight"], "glm4/modeling_glm4.py:Glm4Model.__init__": ["Embedding", "False", "ModelConfig", "ModelDecoderLayer", "ModelModel", "ModelPreTrainedModel", "ModelRMSNorm", "ModelRotaryEmbedding", "ModuleList", "__init__", "class", "config", "def", "embed_tokens", "eps", "for", "gradient_checkpointing", "hidden_size", "in", "layer_idx", "layers", "nn", "norm", "num_hidden_layers", "pad_token_id", "padding_idx", "post_init", "range", "rms_norm_eps", "rotary_emb", "self", "super", "vocab_size"], "glm4/modeling_glm4.py:Glm4Model.forward": ["BaseModelOutputWithPast", "Cache", "DynamicCache", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "and", "arange", "attention_mask", "bool", "cache_position", "causal_mask", "class", "config", "create_causal_mask", "decoder_layer", "def", "device", "else", "embed_tokens", "exactly", "for", "forward", "get_seq_length", "hidden_states", "if", "in", "input_embeds", "input_ids", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layers", "must", "norm", "not", "num_hidden_layers", "of", "one", "or", "past_key_values", "past_seen_tokens", "position_embeddings", "position_ids", "raise", "return", "rotary_emb", "self", "shape", "specify", "torch", "unsqueeze", "use_cache"], "glm4/modeling_glm4.py:Glm4ForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "ModelForCausalLM", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "vocab_size"], "glm4/modeling_glm4.py:Glm4ForCausalLM.forward": ["BaseModelOutputWithPast", "Cache", "CausalLMOutputWithPast", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForCausalLM", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "outputs", "past_key_values", "position_ids", "r", "return", "self", "slice", "slice_indices", "torch", "tuple", "use_cache", "vocab_size"], "glm46v/modeling_glm46v.py:Glm46VModel.__init__": ["AutoModel", "ModelModel", "ModelPreTrainedModel", "None", "__init__", "class", "config", "def", "from_config", "language_model", "post_init", "rope_deltas", "self", "super", "text_config", "vision_config", "visual"], "glm46v/modeling_glm46v.py:Glm46VModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "language_model", "return", "self"], "glm46v/modeling_glm46v.py:Glm46VModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "language_model", "self", "set_input_embeddings", "value"], "glm46v/modeling_glm46v.py:Glm46VModel.get_rope_index": ["False", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "True", "and", "append", "arange", "attention_mask", "cat", "class", "config", "cumsum", "def", "device", "dim", "dtype", "elif", "else", "end_idx", "end_index", "enumerate", "expand", "flatten", "for", "get_rope_index", "group", "groupby", "h", "h_index", "i", "if", "image", "image_grid_thw", "image_index", "image_token_id", "in", "input_ids", "input_token_type", "input_tokens", "input_type_group", "is", "item", "itertools", "keepdim", "key", "lambda", "len", "list", "llm_grid_h", "llm_grid_t", "llm_grid_w", "llm_pos_ids_list", "llm_positions", "long", "masked_fill_", "max", "max_position_ids", "modality_type", "mrope_position_deltas", "not", "ones", "ones_like", "or", "position_ids", "range", "reshape", "return", "self", "shape", "spatial_merge_size", "st_idx", "stack", "start_idx", "start_index", "t", "t_idx", "t_index", "tensor", "text", "text_len", "to", "token", "tolist", "torch", "total_input_ids", "tuple", "unsqueeze", "video", "video_check_flg", "video_end_token_id", "video_frame_num", "video_grid_thw", "video_group_index", "video_index", "video_start_token_id", "view", "vision_config", "w", "w_index", "x", "zeros"], "glm46v/modeling_glm46v.py:Glm46VModel.get_video_features": ["FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "append", "cat", "class", "def", "dim", "dtype", "flattened_video_grid_thw", "for", "get_video_features", "grid_thw", "h", "in", "item", "pixel_values_videos", "prod", "repeat", "repeated_row", "return", "self", "spatial_merge_size", "split", "split_sizes", "t", "temp_frames_hw", "tensor", "tolist", "torch", "type", "unsqueeze", "video_embeds", "video_grid_thw", "visual", "w"], "glm46v/modeling_glm46v.py:Glm46VModel.get_image_features": ["FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "class", "def", "dtype", "get_image_features", "grid_thw", "image_embeds", "image_grid_thw", "pixel_values", "prod", "return", "self", "spatial_merge_size", "split", "split_sizes", "tolist", "torch", "type", "visual"], "glm46v/modeling_glm46v.py:Glm46VModel.get_placeholder_mask": ["FloatTensor", "Image", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "ValueError", "Videos", "all", "and", "class", "config", "def", "device", "do", "dtype", "else", "expand_as", "f", "features", "get_input_embeddings", "get_placeholder_mask", "if", "image", "image_features", "image_token_id", "input_ids", "inputs_embeds", "is", "long", "match", "n_image_tokens", "n_video_tokens", "not", "numel", "raise", "return", "self", "shape", "special_image_mask", "special_video_mask", "sum", "tensor", "to", "tokens", "torch", "unsqueeze", "video", "video_features", "video_token_id"], "glm46v/modeling_glm46v.py:Glm46VModel.forward": ["Cache", "FloatTensor", "LongTensor", "ModelModel", "ModelModelOutputWithPast", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "_", "add", "and", "arange", "attention_mask", "attention_mask_tensor", "attentions", "batch_size", "cache_position", "cat", "class", "def", "delta", "device", "diagonal", "dict", "dim", "dim1", "dim2", "dtype", "else", "exactly", "expand", "finfo", "forward", "full_attention", "get_image_features", "get_input_embeddings", "get_placeholder_mask", "get_rope_index", "get_seq_length", "get_video_features", "hidden_states", "if", "image_embeds", "image_features", "image_grid_thw", "image_mask", "input_ids", "inputs_embeds", "int", "is", "is_floating_point", "is_torchdynamo_compiling", "isinstance", "kwargs", "language_model", "last_hidden_state", "masked_scatter", "min", "must", "ndim", "not", "of", "one", "or", "outputs", "past_key_values", "pixel_values", "pixel_values_videos", "position_ids", "prefill_compiled_stage", "prefill_noncompiled_stage", "r", "raise", "repeat_interleave", "return", "rope_deltas", "self", "seq_length", "shape", "specify", "to", "torch", "tuple", "unsqueeze", "video_embeds", "video_features", "video_grid_thw", "video_mask", "view"], "glm46v/modeling_glm46v.py:Glm46VForConditionalGeneration.__init__": ["False", "GenerationMixin", "Linear", "ModelForConditionalGeneration", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "text_config", "vocab_size"], "glm46v/modeling_glm46v.py:Glm46VForConditionalGeneration.get_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "model", "return", "self"], "glm46v/modeling_glm46v.py:Glm46VForConditionalGeneration.set_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "model", "self", "set_input_embeddings", "value"], "glm46v/modeling_glm46v.py:Glm46VForConditionalGeneration.get_video_features": ["FloatTensor", "GenerationMixin", "LongTensor", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "class", "def", "get_video_features", "model", "pixel_values_videos", "return", "self", "torch", "video_grid_thw"], "glm46v/modeling_glm46v.py:Glm46VForConditionalGeneration.get_image_features": ["FloatTensor", "GenerationMixin", "LongTensor", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "class", "def", "get_image_features", "image_grid_thw", "model", "pixel_values", "return", "self", "torch"], "glm46v/modeling_glm46v.py:Glm46VForConditionalGeneration.forward": ["Cache", "FloatTensor", "GenerationMixin", "LongTensor", "ModelCausalLMOutputWithPast", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "cache_position", "class", "config", "def", "else", "forward", "hidden_states", "if", "image_grid_thw", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "outputs", "past_key_values", "pixel_values", "pixel_values_videos", "position_ids", "r", "return", "rope_deltas", "self", "slice", "slice_indices", "text_config", "torch", "tuple", "video_grid_thw", "vocab_size"], "glm46v/modeling_glm46v.py:Glm46VForConditionalGeneration.prepare_inputs_for_generation": ["False", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "True", "and", "attention_mask", "cache_position", "class", "def", "if", "image_grid_thw", "input_ids", "inputs_embeds", "is_first_iteration", "kwargs", "model_inputs", "not", "past_key_values", "pixel_values", "pixel_values_videos", "position_ids", "prepare_inputs_for_generation", "return", "self", "super", "use_cache", "video_grid_thw"], "glm46v/modeling_glm46v.py:Glm46VForConditionalGeneration._get_image_nums_and_video_nums": ["GenerationMixin", "LongTensor", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Tensor", "_get_image_nums_and_video_nums", "class", "config", "cumsum", "def", "device", "dim", "dtype", "else", "get_input_embeddings", "if", "image_counts", "image_start_token_id", "input_ids", "inputs_embeds", "inside_video", "int", "is", "is_image", "is_video_end", "is_video_start", "long", "not", "return", "self", "standalone_images", "sum", "tensor", "torch", "tuple", "video_counts", "video_end_token_id", "video_level", "video_start_token_id"], "glm46v/modeling_glm46v.py:Glm46VForConditionalGeneration._expand_inputs_for_generation": ["Any", "False", "GenerationMixin", "If", "LongTensor", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Tensor", "True", "ValueError", "_expand_dict_for_generation", "_expand_dict_for_generation_visual", "_expand_inputs_for_generation", "_get_image_nums_and_video_nums", "_repeat_interleave_samples", "and", "bool", "cache_position", "cat", "class", "def", "defined", "dict", "dict_to_expand", "dim", "elif", "encoder_outputs", "expand_size", "for", "get", "if", "image_grid_thw", "image_nums", "in", "input_ids", "inputs_embeds", "int", "is", "is_encoder_decoder", "isinstance", "key", "lengths", "list", "make", "model_kwargs", "not", "pixel_values", "pixel_values_videos", "prod", "raise", "repeat", "repeat_args", "repeat_interleave", "repeat_times", "result", "return", "sample", "samples", "second_per_grid_ts", "self", "split", "str", "sum", "sure", "that", "torch", "tuple", "video_grid_thw", "video_nums", "visual_keys", "x"], "glm4_moe/modeling_glm4_moe.py:Glm4MoeRotaryEmbedding.__init__": ["Callable", "False", "ModelConfig", "ModelRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "glm4_moe/modeling_glm4_moe.py:Glm4MoeRotaryEmbedding.compute_default_rope_parameters": ["ModelConfig", "ModelRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "get", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "partial_rotary_factor", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "glm4_moe/modeling_glm4_moe.py:Glm4MoeRotaryEmbedding.forward": ["False", "ModelRotaryEmbedding", "Module", "None", "and", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "glm4_moe/modeling_glm4_moe.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "glm4_moe/modeling_glm4_moe.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value", "value_states"], "glm4_moe/modeling_glm4_moe.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "glm4_moe/modeling_glm4_moe.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cat", "cos", "def", "dim", "k", "k_embed", "k_pass", "k_rot", "q", "q_embed", "q_pass", "q_rot", "return", "rotary_dim", "rotate_half", "shape", "sin", "torch", "unsqueeze", "unsqueeze_dim"], "glm4_moe/modeling_glm4_moe.py:Glm4MoeAttention.__init__": ["False", "Linear", "ModelAttention", "ModelConfig", "ModelRMSNorm", "Module", "None", "True", "__init__", "attention_bias", "attention_dropout", "bias", "class", "config", "def", "eps", "getattr", "head_dim", "hidden_size", "if", "int", "is_causal", "k_norm", "k_proj", "layer_idx", "nn", "num_attention_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "q_norm", "q_proj", "rms_norm_eps", "rope_parameters", "scaling", "self", "super", "use_qk_norm", "v_proj"], "glm4_moe/modeling_glm4_moe.py:Glm4MoeAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "FlashAttentionKwargs", "LongTensor", "ModelAttention", "Module", "None", "Tensor", "Unpack", "_attn_implementation", "apply_rotary_pos_emb", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "cache_kwargs", "cache_position", "class", "config", "contiguous", "cos", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "head_dim", "hidden_shape", "hidden_states", "if", "input_shape", "is", "k_norm", "k_proj", "key_states", "kwargs", "layer_idx", "nn", "not", "o_proj", "past_key_values", "position_embeddings", "q_norm", "q_proj", "query_states", "reshape", "return", "scaling", "self", "shape", "sin", "torch", "training", "transpose", "tuple", "update", "use_qk_norm", "v_proj", "value_states", "view"], "glm4_moe/modeling_glm4_moe.py:Glm4MoeMLP.__init__": ["ACT2FN", "False", "Linear", "ModelMLP", "Module", "None", "__init__", "act_fn", "bias", "class", "config", "def", "down_proj", "else", "gate_proj", "hidden_act", "hidden_size", "if", "intermediate_size", "is", "nn", "self", "super", "up_proj"], "glm4_moe/modeling_glm4_moe.py:Glm4MoeMLP.forward": ["ModelMLP", "Module", "act_fn", "class", "def", "down_proj", "forward", "gate_proj", "nn", "return", "self", "up_proj", "x"], "glm4_moe/modeling_glm4_moe.py:Glm4MoeTopkRouter.__init__": ["ModelConfig", "ModelTopkRouter", "Module", "Parameter", "__init__", "class", "config", "def", "dtype", "e_score_correction_bias", "empty", "float32", "hidden_size", "n_group", "n_routed_experts", "nn", "norm_topk_prob", "num_experts_per_tok", "register_buffer", "routed_scaling_factor", "self", "super", "top_k", "topk_group", "torch", "weight", "zeros"], "glm4_moe/modeling_glm4_moe.py:Glm4MoeTopkRouter.forward": ["F", "ModelTopkRouter", "Module", "class", "config", "def", "float32", "forward", "hidden_size", "hidden_states", "linear", "nn", "return", "router_logits", "self", "torch", "type", "view", "weight"], "glm4_moe/modeling_glm4_moe.py:Glm4MoeRMSNorm.__init__": ["ModelRMSNorm", "Module", "Parameter", "__init__", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "glm4_moe/modeling_glm4_moe.py:Glm4MoeRMSNorm.forward": ["ModelRMSNorm", "Module", "True", "class", "def", "dtype", "float32", "forward", "hidden_states", "input_dtype", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "glm4_moe/modeling_glm4_moe.py:Glm4MoeRMSNorm.extra_repr": ["ModelRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "variance_epsilon", "weight"], "glm4_moe/modeling_glm4_moe.py:Glm4MoeNaiveMoe.__init__": ["ACT2FN", "ModelNaiveMoe", "Module", "Parameter", "__init__", "act_fn", "class", "config", "def", "down_proj", "empty", "gate_up_proj", "hidden_act", "hidden_dim", "hidden_size", "intermediate_dim", "moe_intermediate_size", "nn", "num_experts", "num_local_experts", "self", "super", "torch"], "glm4_moe/modeling_glm4_moe.py:Glm4MoeNaiveMoe.forward": ["ModelNaiveMoe", "Module", "None", "Tensor", "act_fn", "chunk", "class", "continue", "current_hidden_states", "current_state", "def", "dim", "down_proj", "dtype", "expert_hit", "expert_idx", "expert_mask", "final_hidden_states", "for", "forward", "functional", "gate", "gate_up_proj", "greater", "hidden_states", "if", "in", "index_add_", "linear", "nn", "no_grad", "nonzero", "num_classes", "num_experts", "one_hot", "permute", "return", "self", "sum", "to", "token_idx", "top_k_index", "top_k_pos", "top_k_weights", "torch", "up", "where", "with", "zeros_like"], "glm4_moe/modeling_glm4_moe.py:Glm4MoeMoE.__init__": ["ModelMLP", "ModelMoE", "ModelNaiveMoe", "ModelTopkRouter", "Module", "__init__", "class", "config", "def", "experts", "gate", "intermediate_size", "moe_intermediate_size", "n_group", "n_routed_experts", "n_shared_experts", "nn", "norm_topk_prob", "num_experts_per_tok", "routed_scaling_factor", "self", "shared_experts", "super", "top_k", "topk_group"], "glm4_moe/modeling_glm4_moe.py:Glm4MoeMoE.route_tokens_to_experts": ["False", "ModelMoE", "Module", "True", "bool", "class", "def", "denominator", "dim", "e_score_correction_bias", "expand", "gate", "gather", "group_idx", "group_mask", "group_scores", "if", "k", "keepdim", "masked_fill", "n_group", "n_routed_experts", "nn", "norm_topk_prob", "reshape", "return", "route_tokens_to_experts", "routed_scaling_factor", "router_logits", "router_logits_for_choice", "scatter_", "score_mask", "scores_for_choice", "self", "sigmoid", "sorted", "sum", "top_k", "topk", "topk_group", "topk_indices", "topk_weights", "torch", "unsqueeze", "view", "zeros_like"], "glm4_moe/modeling_glm4_moe.py:Glm4MoeMoE.forward": ["ModelMoE", "Module", "class", "def", "experts", "forward", "gate", "hidden_states", "nn", "orig_shape", "residuals", "return", "route_tokens_to_experts", "router_logits", "self", "shape", "shared_experts", "topk_indices", "topk_weights", "view"], "glm4_moe/modeling_glm4_moe.py:Glm4MoeDecoderLayer.__init__": ["GradientCheckpointingLayer", "ModelAttention", "ModelConfig", "ModelDecoderLayer", "ModelMLP", "ModelMoE", "ModelRMSNorm", "__init__", "class", "config", "def", "else", "eps", "first_k_dense_replace", "hidden_size", "if", "input_layernorm", "int", "layer_idx", "mlp", "post_attention_layernorm", "rms_norm_eps", "self", "self_attn", "super"], "glm4_moe/modeling_glm4_moe.py:Glm4MoeDecoderLayer.forward": ["Cache", "False", "GradientCheckpointingLayer", "LongTensor", "ModelDecoderLayer", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "attention_mask", "bool", "cache_position", "class", "def", "forward", "hidden_states", "input_layernorm", "kwargs", "mlp", "past_key_values", "position_embeddings", "position_ids", "post_attention_layernorm", "residual", "return", "self", "self_attn", "torch", "tuple", "use_cache"], "glm4_moe/modeling_glm4_moe.py:Glm4MoePreTrainedModel._init_weights": ["ModelNaiveMoe", "ModelPreTrainedModel", "ModelTopkRouter", "PreTrainedModel", "_init_weights", "class", "config", "def", "down_proj", "e_score_correction_bias", "elif", "gate_up_proj", "if", "init", "initializer_range", "isinstance", "mean", "module", "normal_", "self", "std", "super", "weight", "zeros_"], "glm4_moe/modeling_glm4_moe.py:Glm4MoeModel.__init__": ["Embedding", "False", "ModelConfig", "ModelDecoderLayer", "ModelModel", "ModelPreTrainedModel", "ModelRMSNorm", "ModelRotaryEmbedding", "ModuleList", "__init__", "class", "config", "def", "embed_tokens", "eps", "for", "gradient_checkpointing", "hidden_size", "in", "layer_idx", "layers", "nn", "norm", "num_hidden_layers", "pad_token_id", "padding_idx", "post_init", "range", "rms_norm_eps", "rotary_emb", "self", "super", "vocab_size"], "glm4_moe/modeling_glm4_moe.py:Glm4MoeModel.forward": ["BaseModelOutputWithPast", "Cache", "DynamicCache", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "and", "arange", "attention_mask", "bool", "cache_position", "causal_mask", "class", "config", "create_causal_mask", "decoder_layer", "def", "device", "else", "embed_tokens", "exactly", "for", "forward", "get_seq_length", "hidden_states", "if", "in", "input_embeds", "input_ids", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layers", "must", "norm", "not", "num_hidden_layers", "of", "one", "or", "past_key_values", "past_seen_tokens", "position_embeddings", "position_ids", "raise", "return", "rotary_emb", "self", "shape", "specify", "torch", "unsqueeze", "use_cache"], "glm4_moe/modeling_glm4_moe.py:Glm4MoeForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "ModelForCausalLM", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "vocab_size"], "glm4_moe/modeling_glm4_moe.py:Glm4MoeForCausalLM.forward": ["BaseModelOutputWithPast", "Cache", "CausalLMOutputWithPast", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForCausalLM", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "outputs", "past_key_values", "position_ids", "r", "return", "self", "slice", "slice_indices", "torch", "use_cache", "vocab_size"], "glm4_moe_lite/modeling_glm4_moe_lite.py:Glm4MoeLiteRotaryEmbedding.__init__": ["Callable", "False", "ModelConfig", "ModelRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "glm4_moe_lite/modeling_glm4_moe_lite.py:Glm4MoeLiteRotaryEmbedding.compute_default_rope_parameters": ["ModelConfig", "ModelRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "get", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "partial_rotary_factor", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "glm4_moe_lite/modeling_glm4_moe_lite.py:Glm4MoeLiteRotaryEmbedding.forward": ["False", "ModelRotaryEmbedding", "Module", "None", "and", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "glm4_moe_lite/modeling_glm4_moe_lite.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "glm4_moe_lite/modeling_glm4_moe_lite.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cos", "def", "k", "k_embed", "q", "q_embed", "return", "rotate_half", "sin", "unsqueeze", "unsqueeze_dim"], "glm4_moe_lite/modeling_glm4_moe_lite.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "glm4_moe_lite/modeling_glm4_moe_lite.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value", "value_states"], "glm4_moe_lite/modeling_glm4_moe_lite.py:apply_rotary_pos_emb_interleave": ["Model_rotary_pos_emb_interleave", "None", "b", "cos", "d", "def", "h", "k", "k_embed", "position_ids", "q", "q_embed", "r", "reshape", "return", "rotate_half", "s", "shape", "sin", "transpose", "unsqueeze", "unsqueeze_dim", "view"], "glm4_moe_lite/modeling_glm4_moe_lite.py:yarn_get_mscale": ["Model_get_mscale", "def", "if", "log", "math", "mscale", "return", "scale"], "glm4_moe_lite/modeling_glm4_moe_lite.py:Glm4MoeLiteAttention.__init__": ["False", "Linear", "ModelAttention", "ModelConfig", "ModelRMSNorm", "Module", "None", "True", "__init__", "attention_bias", "attention_dropout", "bias", "class", "config", "def", "default", "else", "factor", "get", "hidden_size", "if", "int", "is", "is_causal", "kv_a_layernorm", "kv_a_proj_with_mqa", "kv_b_proj", "kv_lora_rank", "layer_idx", "mscale", "mscale_all_dim", "nn", "num_attention_heads", "num_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "q_a_layernorm", "q_a_proj", "q_b_proj", "q_lora_rank", "q_proj", "qk_head_dim", "qk_nope_head_dim", "qk_rope_head_dim", "rope_parameters", "rope_type", "scaling", "scaling_factor", "self", "super", "v_head_dim", "yarn_get_mscale"], "glm4_moe_lite/modeling_glm4_moe_lite.py:Glm4MoeLiteAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "F", "FlashAttentionKwargs", "LongTensor", "ModelAttention", "Module", "None", "Tensor", "Unpack", "_attn_implementation", "and", "apply_rotary_pos_emb", "apply_rotary_pos_emb_interleave", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "batch_size", "cache_kwargs", "cache_position", "cat", "class", "compressed_kv", "config", "contiguous", "cos", "def", "dim", "dropout", "eager", "eager_attention_forward", "else", "expand", "flash_attention_2", "forward", "hidden_states", "if", "is", "k_pass", "k_rot", "key_shape", "key_states", "kv_a_layernorm", "kv_a_proj_with_mqa", "kv_b_proj", "kv_lora_rank", "kwargs", "layer_idx", "nn", "not", "o_proj", "pad", "past_key_values", "position_embeddings", "q_a_layernorm", "q_a_proj", "q_b_proj", "q_lora_rank", "q_pass", "q_proj", "q_rot", "q_states", "qk_head_dim", "qk_nope_head_dim", "qk_rope_head_dim", "query_shape", "query_states", "reshape", "return", "rope_interleave", "scaling", "self", "seq_length", "shape", "sin", "split", "torch", "training", "transpose", "tuple", "update", "v_head_dim", "value_states", "view"], "glm4_moe_lite/modeling_glm4_moe_lite.py:Glm4MoeLiteMLP.__init__": ["ACT2FN", "False", "Linear", "ModelMLP", "Module", "None", "__init__", "act_fn", "bias", "class", "config", "def", "down_proj", "else", "gate_proj", "hidden_act", "hidden_size", "if", "intermediate_size", "is", "nn", "self", "super", "up_proj"], "glm4_moe_lite/modeling_glm4_moe_lite.py:Glm4MoeLiteMLP.forward": ["ModelMLP", "Module", "act_fn", "class", "def", "down_proj", "forward", "gate_proj", "nn", "return", "self", "up_proj", "x"], "glm4_moe_lite/modeling_glm4_moe_lite.py:Glm4MoeLiteTopkRouter.__init__": ["ModelConfig", "ModelTopkRouter", "Module", "Parameter", "__init__", "class", "config", "def", "dtype", "e_score_correction_bias", "empty", "float32", "hidden_size", "n_group", "n_routed_experts", "nn", "norm_topk_prob", "num_experts_per_tok", "register_buffer", "routed_scaling_factor", "self", "super", "top_k", "topk_group", "torch", "weight", "zeros"], "glm4_moe_lite/modeling_glm4_moe_lite.py:Glm4MoeLiteTopkRouter.forward": ["F", "ModelTopkRouter", "Module", "class", "config", "def", "float32", "forward", "hidden_size", "hidden_states", "linear", "nn", "return", "router_logits", "self", "torch", "type", "view", "weight"], "glm4_moe_lite/modeling_glm4_moe_lite.py:Glm4MoeLiteRMSNorm.__init__": ["ModelRMSNorm", "Module", "Parameter", "__init__", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "glm4_moe_lite/modeling_glm4_moe_lite.py:Glm4MoeLiteRMSNorm.forward": ["ModelRMSNorm", "Module", "True", "class", "def", "dtype", "float32", "forward", "hidden_states", "input_dtype", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "glm4_moe_lite/modeling_glm4_moe_lite.py:Glm4MoeLiteRMSNorm.extra_repr": ["ModelRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "variance_epsilon", "weight"], "glm4_moe_lite/modeling_glm4_moe_lite.py:Glm4MoeLiteNaiveMoe.__init__": ["ACT2FN", "ModelNaiveMoe", "Module", "Parameter", "__init__", "act_fn", "class", "config", "def", "down_proj", "empty", "gate_up_proj", "hidden_act", "hidden_dim", "hidden_size", "intermediate_dim", "moe_intermediate_size", "nn", "num_experts", "num_local_experts", "self", "super", "torch"], "glm4_moe_lite/modeling_glm4_moe_lite.py:Glm4MoeLiteNaiveMoe.forward": ["ModelNaiveMoe", "Module", "None", "Tensor", "act_fn", "chunk", "class", "continue", "current_hidden_states", "current_state", "def", "dim", "down_proj", "dtype", "expert_hit", "expert_idx", "expert_mask", "final_hidden_states", "for", "forward", "functional", "gate", "gate_up_proj", "greater", "hidden_states", "if", "in", "index_add_", "linear", "nn", "no_grad", "nonzero", "num_classes", "num_experts", "one_hot", "permute", "return", "self", "sum", "to", "token_idx", "top_k_index", "top_k_pos", "top_k_weights", "torch", "up", "where", "with", "zeros_like"], "glm4_moe_lite/modeling_glm4_moe_lite.py:Glm4MoeLiteMoE.__init__": ["ModelMLP", "ModelMoE", "ModelNaiveMoe", "ModelTopkRouter", "Module", "__init__", "class", "config", "def", "experts", "gate", "intermediate_size", "moe_intermediate_size", "n_group", "n_routed_experts", "n_shared_experts", "nn", "norm_topk_prob", "num_experts_per_tok", "routed_scaling_factor", "self", "shared_experts", "super", "top_k", "topk_group"], "glm4_moe_lite/modeling_glm4_moe_lite.py:Glm4MoeLiteMoE.route_tokens_to_experts": ["False", "ModelMoE", "Module", "True", "bool", "class", "def", "denominator", "dim", "e_score_correction_bias", "expand", "gate", "gather", "group_idx", "group_mask", "group_scores", "if", "k", "keepdim", "masked_fill", "n_group", "n_routed_experts", "nn", "norm_topk_prob", "reshape", "return", "route_tokens_to_experts", "routed_scaling_factor", "router_logits", "router_logits_for_choice", "scatter_", "score_mask", "scores_for_choice", "self", "sigmoid", "sorted", "sum", "top_k", "topk", "topk_group", "topk_indices", "topk_weights", "torch", "unsqueeze", "view", "zeros_like"], "glm4_moe_lite/modeling_glm4_moe_lite.py:Glm4MoeLiteMoE.forward": ["ModelMoE", "Module", "class", "def", "experts", "forward", "gate", "hidden_states", "nn", "orig_shape", "residuals", "return", "route_tokens_to_experts", "router_logits", "self", "shape", "shared_experts", "topk_indices", "topk_weights", "view"], "glm4_moe_lite/modeling_glm4_moe_lite.py:Glm4MoeLiteDecoderLayer.__init__": ["GradientCheckpointingLayer", "ModelAttention", "ModelConfig", "ModelDecoderLayer", "ModelMLP", "ModelMoE", "ModelRMSNorm", "__init__", "class", "config", "def", "else", "hidden_size", "if", "input_layernorm", "int", "layer_idx", "mlp", "mlp_layer_types", "post_attention_layernorm", "rms_norm_eps", "self", "self_attn", "sparse", "super"], "glm4_moe_lite/modeling_glm4_moe_lite.py:Glm4MoeLiteDecoderLayer.forward": ["Cache", "False", "GradientCheckpointingLayer", "LongTensor", "ModelDecoderLayer", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "attention_mask", "bool", "cache_position", "class", "def", "forward", "hidden_states", "input_layernorm", "kwargs", "mlp", "past_key_values", "position_embeddings", "position_ids", "post_attention_layernorm", "residual", "return", "self", "self_attn", "torch", "tuple", "use_cache"], "glm4_moe_lite/modeling_glm4_moe_lite.py:Glm4MoeLitePreTrainedModel._init_weights": ["ModelNaiveMoe", "ModelPreTrainedModel", "ModelTopkRouter", "PreTrainedModel", "_init_weights", "class", "config", "def", "down_proj", "e_score_correction_bias", "elif", "gate_up_proj", "if", "init", "initializer_range", "isinstance", "mean", "module", "normal_", "self", "std", "super", "weight", "zeros_"], "glm4_moe_lite/modeling_glm4_moe_lite.py:Glm4MoeLiteModel.__init__": ["Embedding", "False", "ModelConfig", "ModelDecoderLayer", "ModelModel", "ModelPreTrainedModel", "ModelRMSNorm", "ModelRotaryEmbedding", "ModuleList", "__init__", "class", "config", "def", "embed_tokens", "eps", "for", "gradient_checkpointing", "hidden_size", "in", "layer_idx", "layers", "nn", "norm", "num_hidden_layers", "pad_token_id", "padding_idx", "post_init", "range", "rms_norm_eps", "rotary_emb", "self", "super", "vocab_size"], "glm4_moe_lite/modeling_glm4_moe_lite.py:Glm4MoeLiteModel.forward": ["BaseModelOutputWithPast", "Cache", "DynamicCache", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "and", "arange", "attention_mask", "bool", "cache_position", "causal_mask", "class", "config", "create_causal_mask", "decoder_layer", "def", "device", "else", "embed_tokens", "exactly", "for", "forward", "get_seq_length", "hidden_states", "if", "in", "input_embeds", "input_ids", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layers", "must", "norm", "not", "num_hidden_layers", "of", "one", "or", "past_key_values", "past_seen_tokens", "position_embeddings", "position_ids", "raise", "return", "rotary_emb", "self", "shape", "specify", "torch", "unsqueeze", "use_cache"], "glm4_moe_lite/modeling_glm4_moe_lite.py:Glm4MoeLiteForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "ModelForCausalLM", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "vocab_size"], "glm4_moe_lite/modeling_glm4_moe_lite.py:Glm4MoeLiteForCausalLM.forward": ["BaseModelOutputWithPast", "Cache", "CausalLMOutputWithPast", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForCausalLM", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "outputs", "past_key_values", "position_ids", "r", "return", "self", "slice", "slice_indices", "torch", "use_cache", "vocab_size"], "glm4v/modeling_glm4v.py:Glm4vRMSNorm.__init__": ["ModelRMSNorm", "Module", "Parameter", "__init__", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "glm4v/modeling_glm4v.py:Glm4vRMSNorm.forward": ["ModelRMSNorm", "Module", "True", "class", "def", "dtype", "float32", "forward", "hidden_states", "input_dtype", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "glm4v/modeling_glm4v.py:Glm4vRMSNorm.extra_repr": ["ModelRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "variance_epsilon", "weight"], "glm4v/modeling_glm4v.py:Glm4VisionMlp.__init__": ["ACT2FN", "False", "Linear", "ModelisionMlp", "Module", "__init__", "act_fn", "bias", "bool", "class", "config", "def", "down_proj", "gate_proj", "hidden_act", "hidden_size", "intermediate_size", "nn", "out_hidden_size", "self", "super", "up_proj"], "glm4v/modeling_glm4v.py:Glm4VisionMlp.forward": ["ModelisionMlp", "Module", "act_fn", "class", "def", "down_proj", "forward", "gate_proj", "hidden_state", "nn", "return", "self", "up_proj"], "glm4v/modeling_glm4v.py:Glm4vVisionPatchEmbed.__init__": ["Conv3d", "ModelVisionConfig", "ModelVisionPatchEmbed", "Module", "None", "__init__", "class", "config", "def", "embed_dim", "hidden_size", "in_channels", "kernel_size", "nn", "patch_size", "proj", "self", "stride", "super", "temporal_patch_size"], "glm4v/modeling_glm4v.py:Glm4vVisionPatchEmbed.forward": ["ModelVisionPatchEmbed", "Module", "Tensor", "class", "def", "dtype", "embed_dim", "forward", "hidden_states", "in_channels", "nn", "patch_size", "proj", "return", "self", "target_dtype", "temporal_patch_size", "to", "torch", "view", "weight"], "glm4v/modeling_glm4v.py:Glm4vVisionRotaryEmbedding.__init__": ["False", "ModelVisionRotaryEmbedding", "Module", "None", "__init__", "arange", "class", "def", "dim", "dtype", "float", "int", "inv_freq", "nn", "persistent", "register_buffer", "self", "super", "theta", "torch"], "glm4v/modeling_glm4v.py:Glm4vVisionRotaryEmbedding.forward": ["ModelVisionRotaryEmbedding", "Module", "Tensor", "arange", "class", "def", "device", "dtype", "forward", "freqs", "int", "inv_freq", "nn", "outer", "return", "self", "seq", "seqlen", "torch"], "glm4v/modeling_glm4v.py:Glm4vVisionPatchMerger.__init__": ["ACT2FN", "False", "GELU", "LayerNorm", "Linear", "ModelVisionPatchMerger", "Module", "None", "__init__", "act1", "act_fn", "bias", "bool", "class", "context_dim", "def", "dim", "down_proj", "gate_proj", "hidden_act", "int", "nn", "post_projection_norm", "proj", "self", "str", "super", "up_proj"], "glm4v/modeling_glm4v.py:Glm4vVisionPatchMerger.forward": ["ModelVisionPatchMerger", "Module", "Tensor", "act1", "act_fn", "class", "def", "down_proj", "forward", "gate_proj", "hidden_state", "nn", "post_projection_norm", "proj", "return", "self", "torch", "up_proj"], "glm4v/modeling_glm4v.py:Glm4vVisionEmbeddings.__init__": ["Embedding", "ModelVisionConfig", "ModelVisionEmbeddings", "Module", "__init__", "bicubic", "class", "config", "def", "embed_dim", "hidden_size", "image_size", "interpolated_method", "nn", "num_patches", "num_positions", "patch_size", "position_embedding", "self", "super"], "glm4v/modeling_glm4v.py:Glm4vVisionEmbeddings.forward": ["F", "False", "ModelVisionEmbeddings", "Module", "Tensor", "adapted_pos_embed", "adapted_pos_embed_fp32", "align_corners", "border", "cat", "class", "def", "device", "dim", "dtype", "embeddings", "float32", "for", "forward", "grid", "grid_sample", "h_coords", "hidden_size", "i", "if", "image_shapes", "in", "int", "interpolated_embed_fp32", "interpolated_method", "isinstance", "len", "lengths", "list", "long", "mode", "nn", "norm_h", "norm_w", "orig_size", "orig_size_sq", "padding_mode", "permute", "pos_embed_2d", "pos_embed_weight", "position_embedding", "range", "repeat", "return", "self", "shape", "squeeze", "stack", "target_h", "target_w", "tensor", "to", "torch", "unsqueeze", "view", "w_coords", "weight"], "glm4v/modeling_glm4v.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "glm4v/modeling_glm4v.py:apply_rotary_pos_emb_vision": ["Model_rotary_pos_emb_vision", "Tensor", "cos", "def", "dtype", "float", "k", "k_embed", "orig_k_dtype", "orig_q_dtype", "q", "q_embed", "return", "rotate_half", "sin", "to", "torch", "tuple", "unsqueeze"], "glm4v/modeling_glm4v.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "glm4v/modeling_glm4v.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value", "value_states"], "glm4v/modeling_glm4v.py:Glm4vVisionAttention.__init__": ["False", "Linear", "ModelVisionAttention", "ModelVisionConfig", "Module", "None", "__init__", "attention_bias", "attention_dropout", "bias", "class", "config", "def", "dim", "head_dim", "hidden_size", "is_causal", "nn", "num_heads", "num_key_value_groups", "proj", "qkv", "scaling", "self", "super"], "glm4v/modeling_glm4v.py:Glm4vVisionAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "False", "ModelVisionAttention", "Module", "None", "Tensor", "_", "_attn_implementation", "apply_rotary_pos_emb_vision", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_outputs", "cat", "class", "config", "contiguous", "cos", "cu_seq_lens_k", "cu_seq_lens_q", "cu_seqlens", "def", "dim", "dropout", "eager", "eager_attention_forward", "else", "flash", "for", "forward", "hidden_states", "if", "in", "is_causal", "k", "key_states", "kwargs", "lengths", "max", "max_length_k", "max_length_q", "max_seqlen", "nn", "not", "num_heads", "permute", "position_embeddings", "proj", "q", "qkv", "query_states", "reshape", "return", "rotary_pos_emb", "scaling", "self", "seq_length", "shape", "sin", "split", "splits", "tensor", "tolist", "torch", "training", "transpose", "tuple", "unbind", "unsqueeze", "v", "value_states", "zip"], "glm4v/modeling_glm4v.py:Glm4vVisionBlock.__init__": ["False", "GradientCheckpointingLayer", "ModelRMSNorm", "ModelVisionAttention", "ModelVisionBlock", "ModelisionMlp", "None", "__init__", "attn", "bias", "class", "config", "def", "eps", "hidden_size", "mlp", "norm1", "norm2", "rms_norm_eps", "self", "super"], "glm4v/modeling_glm4v.py:Glm4vVisionBlock.forward": ["GradientCheckpointingLayer", "ModelVisionBlock", "None", "Tensor", "attn", "class", "cu_seqlens", "def", "forward", "hidden_states", "kwargs", "mlp", "norm1", "norm2", "position_embeddings", "return", "rotary_pos_emb", "self", "torch", "tuple"], "glm4v/modeling_glm4v.py:Glm4vTextRotaryEmbedding.__init__": ["Callable", "False", "ModelTextConfig", "ModelTextRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "get", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "mrope_section", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "glm4v/modeling_glm4v.py:Glm4vTextRotaryEmbedding.compute_default_rope_parameters": ["ModelTextConfig", "ModelTextRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "get", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "partial_rotary_factor", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "glm4v/modeling_glm4v.py:Glm4vTextRotaryEmbedding.forward": ["False", "ModelTextRotaryEmbedding", "Module", "None", "and", "apply_mrope", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "mrope_section", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "glm4v/modeling_glm4v.py:Glm4vTextRotaryEmbedding.apply_mrope": ["ModelTextRotaryEmbedding", "Module", "apply_mrope", "cat", "chunk", "chunks", "class", "def", "dim", "enumerate", "for", "freqs", "i", "in", "mrope_section", "nn", "result", "return", "section", "self", "split", "torch"], "glm4v/modeling_glm4v.py:rotate_half_llm": ["Model_half_llm", "def", "dim", "flatten", "return", "stack", "torch", "x", "x1", "x2"], "glm4v/modeling_glm4v.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cat", "cos", "def", "dim", "k", "k_embed", "k_pass", "k_rot", "q", "q_embed", "q_pass", "q_rot", "repeat_interleave", "return", "rotary_dim", "rotate_half_llm", "shape", "sin", "torch", "unsqueeze", "unsqueeze_dim"], "glm4v/modeling_glm4v.py:Glm4vTextAttention.__init__": ["False", "Linear", "ModelTextAttention", "ModelTextConfig", "Module", "None", "True", "__init__", "attention_dropout", "bias", "class", "config", "def", "head_dim", "hidden_size", "int", "is_causal", "k_proj", "layer_idx", "nn", "num_attention_heads", "num_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "q_proj", "rope_parameters", "scaling", "self", "super", "v_proj"], "glm4v/modeling_glm4v.py:Glm4vTextAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "FlashAttentionKwargs", "LongTensor", "ModelTextAttention", "Module", "None", "Tensor", "Unpack", "_", "_attn_implementation", "apply_rotary_pos_emb", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "bsz", "cache_kwargs", "cache_position", "class", "config", "contiguous", "cos", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "head_dim", "hidden_states", "if", "is", "k_proj", "key_states", "kwargs", "layer_idx", "nn", "not", "o_proj", "past_key_values", "position_embeddings", "q_len", "q_proj", "query_states", "reshape", "return", "scaling", "self", "sin", "size", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "view"], "glm4v/modeling_glm4v.py:Glm4vTextMLP.__init__": ["ACT2FN", "False", "Linear", "ModelTextMLP", "Module", "__init__", "activation_fn", "bias", "class", "config", "def", "down_proj", "gate_up_proj", "hidden_act", "hidden_size", "intermediate_size", "nn", "self", "super"], "glm4v/modeling_glm4v.py:Glm4vTextMLP.forward": ["FloatTensor", "ModelTextMLP", "Module", "activation_fn", "chunk", "class", "def", "dim", "down_proj", "forward", "gate", "gate_up_proj", "hidden_states", "nn", "return", "self", "torch", "up_states"], "glm4v/modeling_glm4v.py:Glm4vTextDecoderLayer.__init__": ["GradientCheckpointingLayer", "ModelRMSNorm", "ModelTextAttention", "ModelTextConfig", "ModelTextDecoderLayer", "ModelTextMLP", "__init__", "class", "config", "def", "eps", "hidden_size", "input_layernorm", "int", "layer_idx", "mlp", "post_attention_layernorm", "post_mlp_layernorm", "post_self_attn_layernorm", "rms_norm_eps", "self", "self_attn", "super"], "glm4v/modeling_glm4v.py:Glm4vTextDecoderLayer.forward": ["Cache", "False", "FloatTensor", "GradientCheckpointingLayer", "LongTensor", "ModelTextDecoderLayer", "None", "Tensor", "_", "attention_mask", "bool", "cache_position", "class", "def", "forward", "hidden_states", "input_layernorm", "kwargs", "mlp", "past_key_values", "position_embeddings", "position_ids", "post_attention_layernorm", "post_mlp_layernorm", "post_self_attn_layernorm", "residual", "return", "self", "self_attn", "torch", "tuple", "use_cache"], "glm4v/modeling_glm4v.py:Glm4vPreTrainedModel._init_weights": ["ModelPreTrainedModel", "ModelVisionRotaryEmbedding", "PreTrainedModel", "_init_weights", "arange", "class", "copy_", "def", "dim", "dtype", "float", "if", "init", "inv_freq", "isinstance", "module", "self", "super", "theta", "torch"], "glm4v/modeling_glm4v.py:Glm4vVisionModel.__init__": ["Conv2d", "False", "ModelPreTrainedModel", "ModelRMSNorm", "ModelVisionBlock", "ModelVisionEmbeddings", "ModelVisionModel", "ModelVisionPatchEmbed", "ModelVisionPatchMerger", "ModelVisionRotaryEmbedding", "ModuleList", "None", "_", "__init__", "blocks", "class", "config", "context_dim", "def", "depth", "dim", "downsample", "embeddings", "eps", "for", "gradient_checkpointing", "head_dim", "hidden_act", "hidden_size", "in", "in_channels", "intermediate_size", "kernel_size", "merger", "nn", "num_heads", "out_channels", "out_hidden_size", "patch_embed", "patch_size", "post_conv_layernorm", "post_init", "post_layernorm", "range", "rms_norm_eps", "rotary_pos_emb", "self", "spatial_merge_size", "stride", "super"], "glm4v/modeling_glm4v.py:Glm4vVisionModel.rot_pos_emb": ["ModelPreTrainedModel", "ModelVisionModel", "append", "arange", "cat", "class", "def", "dim", "expand", "flatten", "for", "grid_thw", "h", "hpos_ids", "in", "max", "max_grid_size", "permute", "pos_ids", "repeat", "reshape", "return", "rot_pos_emb", "rotary_pos_emb", "rotary_pos_emb_full", "self", "spatial_merge_size", "stack", "t", "torch", "unsqueeze", "w", "wpos_ids"], "glm4v/modeling_glm4v.py:Glm4vVisionModel.forward": ["F", "ModelPreTrainedModel", "ModelVisionModel", "Tensor", "blk", "blocks", "cat", "class", "config", "cos", "cu_seqlens", "cumsum", "def", "device", "dim", "downsample", "dtype", "else", "emb", "embeddings", "for", "forward", "grid_thw", "hidden_states", "if", "image_type_ids", "in", "int32", "is_tracing", "jit", "kwargs", "merger", "out_hidden_size", "pad", "patch_embed", "permute", "position_embeddings", "post_conv_layernorm", "post_layernorm", "repeat_interleave", "return", "rot_pos_emb", "rotary_pos_emb", "self", "seqlens", "shape", "sin", "spatial_merge_size", "to", "tolist", "torch", "value", "view"], "glm4v/modeling_glm4v.py:Glm4vTextModel.__init__": ["Embedding", "False", "ModelPreTrainedModel", "ModelRMSNorm", "ModelTextConfig", "ModelTextDecoderLayer", "ModelTextModel", "ModelTextRotaryEmbedding", "ModuleList", "__init__", "class", "config", "def", "embed_tokens", "eps", "for", "gradient_checkpointing", "hidden_size", "in", "layer_idx", "layers", "nn", "norm", "num_hidden_layers", "pad_token_id", "padding_idx", "post_init", "range", "rms_norm_eps", "rotary_emb", "self", "super", "vocab_size"], "glm4v/modeling_glm4v.py:Glm4vTextModel.forward": ["BaseModelOutputWithPast", "Cache", "DynamicCache", "FlashAttentionKwargs", "FloatTensor", "LongTensor", "ModelPreTrainedModel", "ModelTextModel", "None", "Tensor", "Unpack", "ValueError", "You", "and", "arange", "attention_mask", "bool", "cache_position", "causal_mask", "class", "config", "create_causal_mask", "decoder_layer", "def", "device", "elif", "else", "embed_tokens", "exactly", "expand", "for", "forward", "get_seq_length", "hidden_states", "if", "in", "input_embeds", "input_ids", "inputs_embeds", "is", "is_tracing", "jit", "kwargs", "last_hidden_state", "layer_outputs", "layers", "mask_kwargs", "must", "ndim", "norm", "not", "of", "one", "or", "past_key_values", "past_seen_tokens", "position_embeddings", "position_ids", "raise", "return", "rotary_emb", "self", "shape", "specify", "text_position_ids", "torch", "tuple", "use_cache", "view"], "glm4v/modeling_glm4v.py:Glm4vModel.__init__": ["ModelModel", "ModelPreTrainedModel", "ModelTextModel", "ModelVisionModel", "None", "__init__", "_from_config", "class", "config", "def", "language_model", "post_init", "rope_deltas", "self", "super", "text_config", "vision_config", "visual"], "glm4v/modeling_glm4v.py:Glm4vModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "language_model", "return", "self"], "glm4v/modeling_glm4v.py:Glm4vModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "language_model", "self", "set_input_embeddings", "value"], "glm4v/modeling_glm4v.py:Glm4vModel.get_rope_index": ["False", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "True", "and", "append", "arange", "attention_mask", "cat", "class", "config", "cumsum", "def", "device", "dim", "dtype", "elif", "else", "end_idx", "end_index", "enumerate", "expand", "flatten", "for", "get_rope_index", "group", "groupby", "h", "h_index", "i", "if", "image", "image_grid_thw", "image_index", "image_token_id", "in", "input_ids", "input_token_type", "input_tokens", "input_type_group", "is", "item", "itertools", "keepdim", "key", "lambda", "len", "list", "llm_grid_h", "llm_grid_t", "llm_grid_w", "llm_pos_ids_list", "llm_positions", "long", "masked_fill_", "max", "max_position_ids", "modality_type", "mrope_position_deltas", "not", "ones", "ones_like", "or", "position_ids", "range", "reshape", "return", "self", "shape", "spatial_merge_size", "st_idx", "stack", "start_idx", "start_index", "t", "t_idx", "t_index", "tensor", "text", "text_len", "to", "token", "tolist", "torch", "total_input_ids", "tuple", "unsqueeze", "video", "video_check_flg", "video_end_token_id", "video_frame_num", "video_grid_thw", "video_group_index", "video_index", "video_start_token_id", "view", "vision_config", "w", "w_index", "x", "zeros"], "glm4v/modeling_glm4v.py:Glm4vModel.get_video_features": ["FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "append", "cat", "class", "def", "dim", "dtype", "flattened_video_grid_thw", "for", "get_video_features", "grid_thw", "h", "in", "item", "pixel_values_videos", "prod", "repeat", "repeated_row", "return", "self", "spatial_merge_size", "split", "split_sizes", "t", "temp_frames_hw", "tensor", "tolist", "torch", "type", "unsqueeze", "video_embeds", "video_grid_thw", "visual", "w"], "glm4v/modeling_glm4v.py:Glm4vModel.get_image_features": ["FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "class", "def", "dtype", "get_image_features", "grid_thw", "image_embeds", "image_grid_thw", "pixel_values", "prod", "return", "self", "spatial_merge_size", "split", "split_sizes", "tolist", "torch", "type", "visual"], "glm4v/modeling_glm4v.py:Glm4vModel.get_placeholder_mask": ["FloatTensor", "Image", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "ValueError", "Videos", "all", "and", "class", "config", "def", "device", "do", "dtype", "else", "expand_as", "f", "features", "get_input_embeddings", "get_placeholder_mask", "if", "image", "image_features", "image_token_id", "input_ids", "inputs_embeds", "is", "long", "match", "n_image_tokens", "n_video_tokens", "not", "numel", "raise", "return", "self", "shape", "special_image_mask", "special_video_mask", "sum", "tensor", "to", "tokens", "torch", "unsqueeze", "video", "video_features", "video_token_id"], "glm4v/modeling_glm4v.py:Glm4vModel.forward": ["Cache", "FloatTensor", "LongTensor", "ModelModel", "ModelModelOutputWithPast", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "_", "add", "and", "arange", "attention_mask", "attention_mask_tensor", "attentions", "batch_size", "cache_position", "cat", "class", "def", "delta", "device", "diagonal", "dict", "dim", "dim1", "dim2", "dtype", "else", "exactly", "expand", "finfo", "forward", "full_attention", "get_image_features", "get_input_embeddings", "get_placeholder_mask", "get_rope_index", "get_seq_length", "get_video_features", "hidden_states", "if", "image_embeds", "image_features", "image_grid_thw", "image_mask", "input_ids", "inputs_embeds", "int", "is", "is_floating_point", "is_torchdynamo_compiling", "isinstance", "kwargs", "language_model", "last_hidden_state", "masked_scatter", "min", "must", "ndim", "not", "of", "one", "or", "outputs", "past_key_values", "pixel_values", "pixel_values_videos", "position_ids", "prefill_compiled_stage", "prefill_noncompiled_stage", "r", "raise", "repeat_interleave", "return", "rope_deltas", "self", "seq_length", "shape", "specify", "to", "torch", "tuple", "unsqueeze", "video_embeds", "video_features", "video_grid_thw", "video_mask", "view"], "glm4v/modeling_glm4v.py:Glm4vForConditionalGeneration.__init__": ["False", "GenerationMixin", "Linear", "ModelForConditionalGeneration", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "text_config", "vocab_size"], "glm4v/modeling_glm4v.py:Glm4vForConditionalGeneration.get_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "model", "return", "self"], "glm4v/modeling_glm4v.py:Glm4vForConditionalGeneration.set_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "model", "self", "set_input_embeddings", "value"], "glm4v/modeling_glm4v.py:Glm4vForConditionalGeneration.get_video_features": ["FloatTensor", "GenerationMixin", "LongTensor", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "class", "def", "get_video_features", "model", "pixel_values_videos", "return", "self", "torch", "video_grid_thw"], "glm4v/modeling_glm4v.py:Glm4vForConditionalGeneration.get_image_features": ["FloatTensor", "GenerationMixin", "LongTensor", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "class", "def", "get_image_features", "image_grid_thw", "model", "pixel_values", "return", "self", "torch"], "glm4v/modeling_glm4v.py:Glm4vForConditionalGeneration.forward": ["Cache", "FloatTensor", "GenerationMixin", "LongTensor", "ModelCausalLMOutputWithPast", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "cache_position", "class", "config", "def", "else", "forward", "hidden_states", "if", "image_grid_thw", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "outputs", "past_key_values", "pixel_values", "pixel_values_videos", "position_ids", "r", "return", "rope_deltas", "self", "slice", "slice_indices", "text_config", "torch", "tuple", "video_grid_thw", "vocab_size"], "glm4v/modeling_glm4v.py:Glm4vForConditionalGeneration.prepare_inputs_for_generation": ["False", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "True", "and", "attention_mask", "cache_position", "class", "def", "if", "image_grid_thw", "input_ids", "inputs_embeds", "is_first_iteration", "kwargs", "model_inputs", "not", "past_key_values", "pixel_values", "pixel_values_videos", "position_ids", "prepare_inputs_for_generation", "return", "self", "super", "use_cache", "video_grid_thw"], "glm4v/modeling_glm4v.py:Glm4vForConditionalGeneration._get_image_nums_and_video_nums": ["GenerationMixin", "LongTensor", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Tensor", "_get_image_nums_and_video_nums", "class", "config", "cumsum", "def", "device", "dim", "dtype", "else", "get_input_embeddings", "if", "image_counts", "image_start_token_id", "input_ids", "inputs_embeds", "inside_video", "int", "is", "is_image", "is_video_end", "is_video_start", "long", "not", "return", "self", "standalone_images", "sum", "tensor", "torch", "tuple", "video_counts", "video_end_token_id", "video_level", "video_start_token_id"], "glm4v/modeling_glm4v.py:Glm4vForConditionalGeneration._expand_inputs_for_generation": ["Any", "False", "GenerationMixin", "If", "LongTensor", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Tensor", "True", "ValueError", "_expand_dict_for_generation", "_expand_dict_for_generation_visual", "_expand_inputs_for_generation", "_get_image_nums_and_video_nums", "_repeat_interleave_samples", "and", "bool", "cache_position", "cat", "class", "def", "defined", "dict", "dict_to_expand", "dim", "elif", "encoder_outputs", "expand_size", "for", "get", "if", "image_grid_thw", "image_nums", "in", "input_ids", "inputs_embeds", "int", "is", "is_encoder_decoder", "isinstance", "key", "lengths", "list", "make", "model_kwargs", "not", "pixel_values", "pixel_values_videos", "prod", "raise", "repeat", "repeat_args", "repeat_interleave", "repeat_times", "result", "return", "sample", "samples", "second_per_grid_ts", "self", "split", "str", "sum", "sure", "that", "torch", "tuple", "video_grid_thw", "video_nums", "visual_keys", "x"], "glm4v_moe/modeling_glm4v_moe.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "glm4v_moe/modeling_glm4v_moe.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value", "value_states"], "glm4v_moe/modeling_glm4v_moe.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "glm4v_moe/modeling_glm4v_moe.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cat", "cos", "def", "dim", "k", "k_embed", "k_pass", "k_rot", "q", "q_embed", "q_pass", "q_rot", "return", "rotary_dim", "rotate_half", "shape", "sin", "torch", "unsqueeze", "unsqueeze_dim"], "glm4v_moe/modeling_glm4v_moe.py:Glm4vMoeTextAttention.__init__": ["False", "Linear", "ModelTextAttention", "ModelTextConfig", "Module", "None", "True", "__init__", "attention_bias", "attention_dropout", "bias", "class", "config", "def", "getattr", "head_dim", "hidden_size", "int", "is_causal", "k_proj", "layer_idx", "nn", "num_attention_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "q_proj", "rope_parameters", "scaling", "self", "super", "v_proj"], "glm4v_moe/modeling_glm4v_moe.py:Glm4vMoeTextAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "FlashAttentionKwargs", "LongTensor", "ModelTextAttention", "Module", "None", "Tensor", "Unpack", "_attn_implementation", "apply_rotary_pos_emb", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "cache_kwargs", "cache_position", "class", "config", "contiguous", "cos", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "head_dim", "hidden_shape", "hidden_states", "if", "input_shape", "is", "k_proj", "key_states", "kwargs", "layer_idx", "nn", "not", "o_proj", "past_key_values", "position_embeddings", "q_proj", "query_states", "reshape", "return", "scaling", "self", "shape", "sin", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "view"], "glm4v_moe/modeling_glm4v_moe.py:Glm4vMoeTextTopkRouter.__init__": ["ModelTextConfig", "ModelTextTopkRouter", "Module", "Parameter", "__init__", "class", "config", "def", "dtype", "e_score_correction_bias", "empty", "float32", "hidden_size", "n_group", "n_routed_experts", "nn", "norm_topk_prob", "num_experts_per_tok", "register_buffer", "routed_scaling_factor", "self", "super", "top_k", "topk_group", "torch", "weight", "zeros"], "glm4v_moe/modeling_glm4v_moe.py:Glm4vMoeTextTopkRouter.forward": ["F", "ModelTextTopkRouter", "Module", "class", "config", "def", "float32", "forward", "hidden_size", "hidden_states", "linear", "nn", "return", "router_logits", "self", "torch", "type", "view", "weight"], "glm4v_moe/modeling_glm4v_moe.py:Glm4vMoeTextNaiveMoe.__init__": ["ACT2FN", "ModelTextNaiveMoe", "Module", "Parameter", "__init__", "act_fn", "class", "config", "def", "down_proj", "empty", "gate_up_proj", "hidden_act", "hidden_dim", "hidden_size", "intermediate_dim", "moe_intermediate_size", "nn", "num_experts", "num_local_experts", "self", "super", "torch"], "glm4v_moe/modeling_glm4v_moe.py:Glm4vMoeTextNaiveMoe.forward": ["ModelTextNaiveMoe", "Module", "None", "Tensor", "act_fn", "chunk", "class", "continue", "current_hidden_states", "current_state", "def", "dim", "down_proj", "dtype", "expert_hit", "expert_idx", "expert_mask", "final_hidden_states", "for", "forward", "functional", "gate", "gate_up_proj", "greater", "hidden_states", "if", "in", "index_add_", "linear", "nn", "no_grad", "nonzero", "num_classes", "num_experts", "one_hot", "permute", "return", "self", "sum", "to", "token_idx", "top_k_index", "top_k_pos", "top_k_weights", "torch", "up", "where", "with", "zeros_like"], "glm4v_moe/modeling_glm4v_moe.py:Glm4vMoeTextMoE.__init__": ["ModelTextConfig", "ModelTextMLP", "ModelTextMoE", "ModelTextNaiveMoe", "ModelTextTopkRouter", "Module", "__init__", "class", "config", "def", "experts", "gate", "intermediate_size", "moe_intermediate_size", "n_group", "n_routed_experts", "n_shared_experts", "nn", "norm_topk_prob", "num_experts_per_tok", "routed_scaling_factor", "self", "shared_experts", "super", "top_k", "topk_group"], "glm4v_moe/modeling_glm4v_moe.py:Glm4vMoeTextMoE.route_tokens_to_experts": ["False", "ModelTextMoE", "Module", "True", "bool", "class", "def", "denominator", "dim", "e_score_correction_bias", "expand", "gate", "gather", "group_idx", "group_mask", "group_scores", "if", "k", "keepdim", "masked_fill", "n_group", "n_routed_experts", "nn", "norm_topk_prob", "reshape", "return", "route_tokens_to_experts", "routed_scaling_factor", "router_logits", "router_logits_for_choice", "scatter_", "score_mask", "scores_for_choice", "self", "sigmoid", "sorted", "sum", "top_k", "topk", "topk_group", "topk_indices", "topk_weights", "torch", "unsqueeze", "view", "zeros_like"], "glm4v_moe/modeling_glm4v_moe.py:Glm4vMoeTextMoE.forward": ["ModelTextMoE", "Module", "class", "def", "experts", "forward", "gate", "hidden_states", "nn", "orig_shape", "residuals", "return", "route_tokens_to_experts", "router_logits", "self", "shape", "shared_experts", "topk_indices", "topk_weights", "view"], "glm4v_moe/modeling_glm4v_moe.py:Glm4vMoeTextMLP.__init__": ["ACT2FN", "False", "Linear", "ModelTextMLP", "Module", "None", "__init__", "act_fn", "bias", "class", "config", "def", "down_proj", "else", "gate_proj", "hidden_act", "hidden_size", "if", "intermediate_size", "is", "nn", "self", "super", "up_proj"], "glm4v_moe/modeling_glm4v_moe.py:Glm4vMoeTextMLP.forward": ["ModelTextMLP", "Module", "act_fn", "class", "def", "down_proj", "forward", "gate_proj", "nn", "return", "self", "up_proj", "x"], "glm4v_moe/modeling_glm4v_moe.py:Glm4vMoeTextRMSNorm.__init__": ["ModelTextRMSNorm", "Module", "Parameter", "__init__", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "glm4v_moe/modeling_glm4v_moe.py:Glm4vMoeTextRMSNorm.forward": ["ModelTextRMSNorm", "Module", "True", "class", "def", "dtype", "float32", "forward", "hidden_states", "input_dtype", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "glm4v_moe/modeling_glm4v_moe.py:Glm4vMoeTextRMSNorm.extra_repr": ["ModelTextRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "variance_epsilon", "weight"], "glm4v_moe/modeling_glm4v_moe.py:Glm4vMoeTextDecoderLayer.__init__": ["GradientCheckpointingLayer", "ModelTextAttention", "ModelTextConfig", "ModelTextDecoderLayer", "ModelTextMLP", "ModelTextMoE", "ModelTextRMSNorm", "__init__", "class", "config", "def", "else", "eps", "first_k_dense_replace", "hidden_size", "if", "input_layernorm", "int", "layer_idx", "mlp", "post_attention_layernorm", "rms_norm_eps", "self", "self_attn", "super"], "glm4v_moe/modeling_glm4v_moe.py:Glm4vMoeTextDecoderLayer.forward": ["Cache", "False", "GradientCheckpointingLayer", "LongTensor", "ModelTextDecoderLayer", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "attention_mask", "bool", "cache_position", "class", "def", "forward", "hidden_states", "input_layernorm", "kwargs", "mlp", "past_key_values", "position_embeddings", "position_ids", "post_attention_layernorm", "residual", "return", "self", "self_attn", "torch", "tuple", "use_cache"], "glm4v_moe/modeling_glm4v_moe.py:Glm4vMoePreTrainedModel._init_weights": ["ModelPreTrainedModel", "ModelTextNaiveMoe", "ModelTextTopkRouter", "ModelVisionRotaryEmbedding", "PreTrainedModel", "_init_weights", "arange", "class", "config", "copy_", "def", "dim", "down_proj", "dtype", "e_score_correction_bias", "elif", "float", "gate_up_proj", "if", "init", "initializer_range", "inv_freq", "isinstance", "mean", "module", "normal_", "self", "std", "super", "theta", "torch", "weight", "zeros_"], "glm4v_moe/modeling_glm4v_moe.py:Glm4vMoeVisionRotaryEmbedding.__init__": ["False", "ModelVisionRotaryEmbedding", "Module", "None", "__init__", "arange", "class", "def", "dim", "dtype", "float", "int", "inv_freq", "nn", "persistent", "register_buffer", "self", "super", "theta", "torch"], "glm4v_moe/modeling_glm4v_moe.py:Glm4vMoeVisionRotaryEmbedding.forward": ["ModelVisionRotaryEmbedding", "Module", "Tensor", "arange", "class", "def", "device", "dtype", "forward", "freqs", "int", "inv_freq", "nn", "outer", "return", "self", "seq", "seqlen", "torch"], "glm4v_moe/modeling_glm4v_moe.py:Glm4vMoeRMSNorm.__init__": ["ModelRMSNorm", "Module", "Parameter", "__init__", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "glm4v_moe/modeling_glm4v_moe.py:Glm4vMoeRMSNorm.forward": ["ModelRMSNorm", "Module", "True", "class", "def", "dtype", "float32", "forward", "hidden_states", "input_dtype", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "glm4v_moe/modeling_glm4v_moe.py:Glm4vMoeRMSNorm.extra_repr": ["ModelRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "variance_epsilon", "weight"], "glm4v_moe/modeling_glm4v_moe.py:Glm4vMoeisionMlp.__init__": ["ACT2FN", "False", "Linear", "ModelisionMlp", "Module", "__init__", "act_fn", "bias", "bool", "class", "config", "def", "down_proj", "gate_proj", "hidden_act", "hidden_size", "intermediate_size", "nn", "out_hidden_size", "self", "super", "up_proj"], "glm4v_moe/modeling_glm4v_moe.py:Glm4vMoeisionMlp.forward": ["ModelisionMlp", "Module", "act_fn", "class", "def", "down_proj", "forward", "gate_proj", "hidden_state", "nn", "return", "self", "up_proj"], "glm4v_moe/modeling_glm4v_moe.py:Glm4vMoeVisionPatchEmbed.__init__": ["Conv3d", "ModelVisionConfig", "ModelVisionPatchEmbed", "Module", "None", "__init__", "class", "config", "def", "embed_dim", "hidden_size", "in_channels", "kernel_size", "nn", "patch_size", "proj", "self", "stride", "super", "temporal_patch_size"], "glm4v_moe/modeling_glm4v_moe.py:Glm4vMoeVisionPatchEmbed.forward": ["ModelVisionPatchEmbed", "Module", "Tensor", "class", "def", "dtype", "embed_dim", "forward", "hidden_states", "in_channels", "nn", "patch_size", "proj", "return", "self", "target_dtype", "temporal_patch_size", "to", "torch", "view", "weight"], "glm4v_moe/modeling_glm4v_moe.py:Glm4vMoeVisionPatchMerger.__init__": ["ACT2FN", "False", "GELU", "LayerNorm", "Linear", "ModelVisionPatchMerger", "Module", "None", "__init__", "act1", "act_fn", "bias", "bool", "class", "context_dim", "def", "dim", "down_proj", "gate_proj", "hidden_act", "int", "nn", "post_projection_norm", "proj", "self", "str", "super", "up_proj"], "glm4v_moe/modeling_glm4v_moe.py:Glm4vMoeVisionPatchMerger.forward": ["ModelVisionPatchMerger", "Module", "Tensor", "act1", "act_fn", "class", "def", "down_proj", "forward", "gate_proj", "hidden_state", "nn", "post_projection_norm", "proj", "return", "self", "torch", "up_proj"], "glm4v_moe/modeling_glm4v_moe.py:Glm4vMoeVisionEmbeddings.__init__": ["Embedding", "ModelVisionConfig", "ModelVisionEmbeddings", "Module", "__init__", "bicubic", "class", "config", "def", "embed_dim", "hidden_size", "image_size", "interpolated_method", "nn", "num_patches", "num_positions", "patch_size", "position_embedding", "self", "super"], "glm4v_moe/modeling_glm4v_moe.py:Glm4vMoeVisionEmbeddings.forward": ["F", "False", "ModelVisionEmbeddings", "Module", "Tensor", "adapted_pos_embed", "adapted_pos_embed_fp32", "align_corners", "border", "cat", "class", "def", "device", "dim", "dtype", "embeddings", "float32", "for", "forward", "grid", "grid_sample", "h_coords", "hidden_size", "i", "if", "image_shapes", "in", "int", "interpolated_embed_fp32", "interpolated_method", "isinstance", "len", "lengths", "list", "long", "mode", "nn", "norm_h", "norm_w", "orig_size", "orig_size_sq", "padding_mode", "permute", "pos_embed_2d", "pos_embed_weight", "position_embedding", "range", "repeat", "return", "self", "shape", "squeeze", "stack", "target_h", "target_w", "tensor", "to", "torch", "unsqueeze", "view", "w_coords", "weight"], "glm4v_moe/modeling_glm4v_moe.py:apply_rotary_pos_emb_vision": ["Model_rotary_pos_emb_vision", "Tensor", "cos", "def", "dtype", "float", "k", "k_embed", "orig_k_dtype", "orig_q_dtype", "q", "q_embed", "return", "rotate_half", "sin", "to", "torch", "tuple", "unsqueeze"], "glm4v_moe/modeling_glm4v_moe.py:Glm4vMoeVisionAttention.__init__": ["False", "Linear", "ModelVisionAttention", "ModelVisionConfig", "Module", "None", "__init__", "attention_bias", "attention_dropout", "bias", "class", "config", "def", "dim", "head_dim", "hidden_size", "is_causal", "nn", "num_heads", "num_key_value_groups", "proj", "qkv", "scaling", "self", "super"], "glm4v_moe/modeling_glm4v_moe.py:Glm4vMoeVisionAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "False", "ModelVisionAttention", "Module", "None", "Tensor", "_", "_attn_implementation", "apply_rotary_pos_emb_vision", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_outputs", "cat", "class", "config", "contiguous", "cos", "cu_seq_lens_k", "cu_seq_lens_q", "cu_seqlens", "def", "dim", "dropout", "eager", "eager_attention_forward", "else", "flash", "for", "forward", "hidden_states", "if", "in", "is_causal", "k", "key_states", "kwargs", "lengths", "max", "max_length_k", "max_length_q", "max_seqlen", "nn", "not", "num_heads", "permute", "position_embeddings", "proj", "q", "qkv", "query_states", "reshape", "return", "rotary_pos_emb", "scaling", "self", "seq_length", "shape", "sin", "split", "splits", "tensor", "tolist", "torch", "training", "transpose", "tuple", "unbind", "unsqueeze", "v", "value_states", "zip"], "glm4v_moe/modeling_glm4v_moe.py:Glm4vMoeVisionBlock.__init__": ["False", "GradientCheckpointingLayer", "ModelRMSNorm", "ModelVisionAttention", "ModelVisionBlock", "ModelisionMlp", "None", "__init__", "attn", "bias", "class", "config", "def", "eps", "hidden_size", "mlp", "norm1", "norm2", "rms_norm_eps", "self", "super"], "glm4v_moe/modeling_glm4v_moe.py:Glm4vMoeVisionBlock.forward": ["GradientCheckpointingLayer", "ModelVisionBlock", "None", "Tensor", "attn", "class", "cu_seqlens", "def", "forward", "hidden_states", "kwargs", "mlp", "norm1", "norm2", "position_embeddings", "return", "rotary_pos_emb", "self", "torch", "tuple"], "glm4v_moe/modeling_glm4v_moe.py:Glm4vMoeVisionModel.__init__": ["Conv2d", "False", "ModelPreTrainedModel", "ModelRMSNorm", "ModelVisionBlock", "ModelVisionEmbeddings", "ModelVisionModel", "ModelVisionPatchEmbed", "ModelVisionPatchMerger", "ModelVisionRotaryEmbedding", "ModuleList", "None", "_", "__init__", "blocks", "class", "config", "context_dim", "def", "depth", "dim", "downsample", "embeddings", "eps", "for", "gradient_checkpointing", "head_dim", "hidden_act", "hidden_size", "in", "in_channels", "intermediate_size", "kernel_size", "merger", "nn", "num_heads", "out_channels", "out_hidden_size", "patch_embed", "patch_size", "post_conv_layernorm", "post_init", "post_layernorm", "range", "rms_norm_eps", "rotary_pos_emb", "self", "spatial_merge_size", "stride", "super"], "glm4v_moe/modeling_glm4v_moe.py:Glm4vMoeVisionModel.rot_pos_emb": ["ModelPreTrainedModel", "ModelVisionModel", "append", "arange", "cat", "class", "def", "dim", "expand", "flatten", "for", "grid_thw", "h", "hpos_ids", "in", "max", "max_grid_size", "permute", "pos_ids", "repeat", "reshape", "return", "rot_pos_emb", "rotary_pos_emb", "rotary_pos_emb_full", "self", "spatial_merge_size", "stack", "t", "torch", "unsqueeze", "w", "wpos_ids"], "glm4v_moe/modeling_glm4v_moe.py:Glm4vMoeVisionModel.forward": ["F", "ModelPreTrainedModel", "ModelVisionModel", "Tensor", "blk", "blocks", "cat", "class", "config", "cos", "cu_seqlens", "cumsum", "def", "device", "dim", "downsample", "dtype", "else", "emb", "embeddings", "for", "forward", "grid_thw", "hidden_states", "if", "image_type_ids", "in", "int32", "is_tracing", "jit", "kwargs", "merger", "out_hidden_size", "pad", "patch_embed", "permute", "position_embeddings", "post_conv_layernorm", "post_layernorm", "repeat_interleave", "return", "rot_pos_emb", "rotary_pos_emb", "self", "seqlens", "shape", "sin", "spatial_merge_size", "to", "tolist", "torch", "value", "view"], "glm4v_moe/modeling_glm4v_moe.py:Glm4vMoeTextRotaryEmbedding.__init__": ["Callable", "False", "ModelTextConfig", "ModelTextRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "get", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "mrope_section", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "glm4v_moe/modeling_glm4v_moe.py:Glm4vMoeTextRotaryEmbedding.compute_default_rope_parameters": ["ModelTextConfig", "ModelTextRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "get", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "partial_rotary_factor", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "glm4v_moe/modeling_glm4v_moe.py:Glm4vMoeTextRotaryEmbedding.forward": ["False", "ModelTextRotaryEmbedding", "Module", "None", "and", "apply_mrope", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "mrope_section", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "glm4v_moe/modeling_glm4v_moe.py:Glm4vMoeTextRotaryEmbedding.apply_mrope": ["ModelTextRotaryEmbedding", "Module", "apply_mrope", "cat", "chunk", "chunks", "class", "def", "dim", "enumerate", "for", "freqs", "i", "in", "mrope_section", "nn", "result", "return", "section", "self", "split", "torch"], "glm4v_moe/modeling_glm4v_moe.py:Glm4vMoeTextModel.__init__": ["Embedding", "False", "ModelPreTrainedModel", "ModelRMSNorm", "ModelTextConfig", "ModelTextDecoderLayer", "ModelTextModel", "ModelTextRotaryEmbedding", "ModuleList", "__init__", "class", "config", "def", "embed_tokens", "eps", "for", "gradient_checkpointing", "hidden_size", "in", "layer_idx", "layers", "nn", "norm", "num_hidden_layers", "pad_token_id", "padding_idx", "post_init", "range", "rms_norm_eps", "rotary_emb", "self", "super", "vocab_size"], "glm4v_moe/modeling_glm4v_moe.py:Glm4vMoeTextModel.forward": ["Cache", "DynamicCache", "FlashAttentionKwargs", "FloatTensor", "LongTensor", "ModelPreTrainedModel", "ModelTextModel", "MoeModelOutputWithPast", "None", "Tensor", "Unpack", "ValueError", "You", "and", "arange", "attention_mask", "bool", "cache_position", "causal_mask", "class", "config", "create_causal_mask", "decoder_layer", "def", "device", "elif", "else", "embed_tokens", "enumerate", "exactly", "expand", "for", "forward", "get_seq_length", "hidden_states", "i", "if", "in", "input_embeds", "input_ids", "inputs_embeds", "is", "is_tracing", "jit", "kwargs", "last_hidden_state", "layer_outputs", "layers", "mask_kwargs", "must", "ndim", "norm", "not", "num_hidden_layers", "of", "one", "or", "past_key_values", "past_seen_tokens", "position_embeddings", "position_ids", "raise", "return", "rotary_emb", "self", "shape", "specify", "text_position_ids", "torch", "use_cache", "view"], "glm4v_moe/modeling_glm4v_moe.py:Glm4vMoeModel.__init__": ["ModelModel", "ModelPreTrainedModel", "ModelTextModel", "ModelVisionModel", "None", "__init__", "_from_config", "class", "config", "def", "language_model", "post_init", "rope_deltas", "self", "super", "text_config", "vision_config", "visual"], "glm4v_moe/modeling_glm4v_moe.py:Glm4vMoeModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "language_model", "return", "self"], "glm4v_moe/modeling_glm4v_moe.py:Glm4vMoeModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "language_model", "self", "set_input_embeddings", "value"], "glm4v_moe/modeling_glm4v_moe.py:Glm4vMoeModel.get_rope_index": ["False", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "True", "and", "append", "arange", "attention_mask", "cat", "class", "config", "cumsum", "def", "device", "dim", "dtype", "elif", "else", "end_idx", "end_index", "enumerate", "expand", "flatten", "for", "get_rope_index", "group", "groupby", "h", "h_index", "i", "if", "image", "image_grid_thw", "image_index", "image_token_id", "in", "input_ids", "input_token_type", "input_tokens", "input_type_group", "is", "item", "itertools", "keepdim", "key", "lambda", "len", "list", "llm_grid_h", "llm_grid_t", "llm_grid_w", "llm_pos_ids_list", "llm_positions", "long", "masked_fill_", "max", "max_position_ids", "modality_type", "mrope_position_deltas", "not", "ones", "ones_like", "or", "position_ids", "range", "reshape", "return", "self", "shape", "spatial_merge_size", "st_idx", "stack", "start_idx", "start_index", "t", "t_idx", "t_index", "tensor", "text", "text_len", "to", "token", "tolist", "torch", "total_input_ids", "tuple", "unsqueeze", "video", "video_check_flg", "video_end_token_id", "video_frame_num", "video_grid_thw", "video_group_index", "video_index", "video_start_token_id", "view", "vision_config", "w", "w_index", "x", "zeros"], "glm4v_moe/modeling_glm4v_moe.py:Glm4vMoeModel.get_video_features": ["FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "append", "cat", "class", "def", "dim", "dtype", "flattened_video_grid_thw", "for", "get_video_features", "grid_thw", "h", "in", "item", "pixel_values_videos", "prod", "repeat", "repeated_row", "return", "self", "spatial_merge_size", "split", "split_sizes", "t", "temp_frames_hw", "tensor", "tolist", "torch", "type", "unsqueeze", "video_embeds", "video_grid_thw", "visual", "w"], "glm4v_moe/modeling_glm4v_moe.py:Glm4vMoeModel.get_image_features": ["FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "class", "def", "dtype", "get_image_features", "grid_thw", "image_embeds", "image_grid_thw", "pixel_values", "prod", "return", "self", "spatial_merge_size", "split", "split_sizes", "tolist", "torch", "type", "visual"], "glm4v_moe/modeling_glm4v_moe.py:Glm4vMoeModel.get_placeholder_mask": ["FloatTensor", "Image", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "ValueError", "Videos", "all", "and", "class", "config", "def", "device", "do", "dtype", "else", "expand_as", "f", "features", "get_input_embeddings", "get_placeholder_mask", "if", "image", "image_features", "image_token_id", "input_ids", "inputs_embeds", "is", "long", "match", "n_image_tokens", "n_video_tokens", "not", "numel", "raise", "return", "self", "shape", "special_image_mask", "special_video_mask", "sum", "tensor", "to", "tokens", "torch", "unsqueeze", "video", "video_features", "video_token_id"], "glm4v_moe/modeling_glm4v_moe.py:Glm4vMoeModel.forward": ["Cache", "FloatTensor", "LongTensor", "ModelModel", "ModelModelOutputWithPast", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "_", "add", "and", "arange", "attention_mask", "attention_mask_tensor", "attentions", "batch_size", "cache_position", "cat", "class", "def", "delta", "device", "diagonal", "dict", "dim", "dim1", "dim2", "dtype", "else", "exactly", "expand", "finfo", "forward", "full_attention", "get_image_features", "get_input_embeddings", "get_placeholder_mask", "get_rope_index", "get_seq_length", "get_video_features", "hidden_states", "if", "image_embeds", "image_features", "image_grid_thw", "image_mask", "input_ids", "inputs_embeds", "int", "is", "is_floating_point", "is_torchdynamo_compiling", "isinstance", "kwargs", "language_model", "last_hidden_state", "masked_scatter", "min", "must", "ndim", "not", "of", "one", "or", "outputs", "past_key_values", "pixel_values", "pixel_values_videos", "position_ids", "prefill_compiled_stage", "prefill_noncompiled_stage", "r", "raise", "repeat_interleave", "return", "rope_deltas", "self", "seq_length", "shape", "specify", "to", "torch", "tuple", "unsqueeze", "video_embeds", "video_features", "video_grid_thw", "video_mask", "view"], "glm4v_moe/modeling_glm4v_moe.py:load_balancing_loss_func": ["Model_balancing_loss_func", "None", "Tensor", "_", "attention_mask", "batch_size", "cat", "compute_device", "concatenated_gate_logits", "def", "device", "dim", "else", "expand", "expert_attention_mask", "expert_mask", "float", "for", "functional", "gate_logits", "if", "in", "int", "is", "isinstance", "layer_gate", "mean", "nn", "not", "num_experts", "num_hidden_layers", "one_hot", "or", "overall_loss", "r", "reshape", "return", "router_per_expert_attention_mask", "router_prob_per_expert", "routing_weights", "selected_experts", "sequence_length", "shape", "softmax", "sum", "to", "tokens_per_expert", "top_k", "topk", "torch", "tuple", "unsqueeze"], "glm4v_moe/modeling_glm4v_moe.py:Glm4vMoeForConditionalGeneration.__init__": ["False", "GenerationMixin", "Linear", "ModelForConditionalGeneration", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "text_config", "vocab_size"], "glm4v_moe/modeling_glm4v_moe.py:Glm4vMoeForConditionalGeneration.get_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "model", "return", "self"], "glm4v_moe/modeling_glm4v_moe.py:Glm4vMoeForConditionalGeneration.set_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "model", "self", "set_input_embeddings", "value"], "glm4v_moe/modeling_glm4v_moe.py:Glm4vMoeForConditionalGeneration.get_video_features": ["FloatTensor", "GenerationMixin", "LongTensor", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "class", "def", "get_video_features", "model", "pixel_values_videos", "return", "self", "torch", "video_grid_thw"], "glm4v_moe/modeling_glm4v_moe.py:Glm4vMoeForConditionalGeneration.get_image_features": ["FloatTensor", "GenerationMixin", "LongTensor", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "class", "def", "get_image_features", "image_grid_thw", "model", "pixel_values", "return", "self", "torch"], "glm4v_moe/modeling_glm4v_moe.py:Glm4vMoeForConditionalGeneration.forward": ["Cache", "False", "FloatTensor", "GenerationMixin", "LongTensor", "ModelCausalLMOutputWithPast", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "aux_loss", "cache_position", "class", "config", "def", "device", "else", "forward", "get", "hidden_states", "if", "image_grid_thw", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "lm_head", "load_balancing_loss_func", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "num_experts", "num_experts_per_tok", "output_router_logits", "outputs", "past_key_values", "pixel_values", "pixel_values_videos", "position_ids", "r", "return", "rope_deltas", "router_aux_loss_coef", "router_logits", "self", "slice", "slice_indices", "text_config", "to", "torch", "tuple", "video_grid_thw", "vocab_size"], "glm4v_moe/modeling_glm4v_moe.py:Glm4vMoeForConditionalGeneration.prepare_inputs_for_generation": ["False", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "True", "and", "attention_mask", "cache_position", "class", "def", "if", "image_grid_thw", "input_ids", "inputs_embeds", "is_first_iteration", "kwargs", "model_inputs", "not", "past_key_values", "pixel_values", "pixel_values_videos", "position_ids", "prepare_inputs_for_generation", "return", "self", "super", "use_cache", "video_grid_thw"], "glm4v_moe/modeling_glm4v_moe.py:Glm4vMoeForConditionalGeneration._get_image_nums_and_video_nums": ["GenerationMixin", "LongTensor", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Tensor", "_get_image_nums_and_video_nums", "class", "config", "cumsum", "def", "device", "dim", "dtype", "else", "get_input_embeddings", "if", "image_counts", "image_start_token_id", "input_ids", "inputs_embeds", "inside_video", "int", "is", "is_image", "is_video_end", "is_video_start", "long", "not", "return", "self", "standalone_images", "sum", "tensor", "torch", "tuple", "video_counts", "video_end_token_id", "video_level", "video_start_token_id"], "glm4v_moe/modeling_glm4v_moe.py:Glm4vMoeForConditionalGeneration._expand_inputs_for_generation": ["Any", "False", "GenerationMixin", "If", "LongTensor", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Tensor", "True", "ValueError", "_expand_dict_for_generation", "_expand_dict_for_generation_visual", "_expand_inputs_for_generation", "_get_image_nums_and_video_nums", "_repeat_interleave_samples", "and", "bool", "cache_position", "cat", "class", "def", "defined", "dict", "dict_to_expand", "dim", "elif", "encoder_outputs", "expand_size", "for", "get", "if", "image_grid_thw", "image_nums", "in", "input_ids", "inputs_embeds", "int", "is", "is_encoder_decoder", "isinstance", "key", "lengths", "list", "make", "model_kwargs", "not", "pixel_values", "pixel_values_videos", "prod", "raise", "repeat", "repeat_args", "repeat_interleave", "repeat_times", "result", "return", "sample", "samples", "second_per_grid_ts", "self", "split", "str", "sum", "sure", "that", "torch", "tuple", "video_grid_thw", "video_nums", "visual_keys", "x"], "glm_image/modeling_glm_image.py:GlmImageVisionMLP.__init__": ["ACT2FN", "Linear", "ModelVisionMLP", "Module", "__init__", "activation_fn", "class", "config", "def", "fc1", "fc2", "hidden_act", "hidden_size", "intermediate_size", "nn", "self", "super"], "glm_image/modeling_glm_image.py:GlmImageVisionMLP.forward": ["ModelVisionMLP", "Module", "Tensor", "activation_fn", "class", "def", "fc1", "fc2", "forward", "hidden_states", "nn", "return", "self", "torch"], "glm_image/modeling_glm_image.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "glm_image/modeling_glm_image.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value", "value_states"], "glm_image/modeling_glm_image.py:GlmImageVisionAttention.__init__": ["False", "Linear", "ModelVisionAttention", "ModelVisionConfig", "Module", "None", "__init__", "attention_bias", "attention_dropout", "bias", "class", "config", "def", "dim", "head_dim", "hidden_size", "is_causal", "nn", "num_heads", "num_key_value_groups", "proj", "qkv", "scaling", "self", "super"], "glm_image/modeling_glm_image.py:GlmImageVisionAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "False", "ModelVisionAttention", "Module", "None", "Tensor", "_", "_attn_implementation", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_outputs", "cat", "class", "config", "contiguous", "cu_seq_lens_k", "cu_seq_lens_q", "cu_seqlens", "def", "dim", "dropout", "eager", "eager_attention_forward", "else", "flash", "for", "forward", "hidden_states", "if", "in", "is_causal", "k", "key_states", "kwargs", "lengths", "max", "max_length_k", "max_length_q", "max_seqlen", "nn", "not", "num_heads", "permute", "proj", "q", "qkv", "query_states", "reshape", "return", "scaling", "self", "seq_length", "shape", "split", "splits", "tensor", "tolist", "torch", "training", "transpose", "unbind", "unsqueeze", "v", "value_states", "zip"], "glm_image/modeling_glm_image.py:GlmImageVisionPatchEmbed.__init__": ["Conv2d", "ModelVisionConfig", "ModelVisionPatchEmbed", "Module", "None", "__init__", "class", "config", "def", "embed_dim", "hidden_size", "in_channels", "kernel_size", "nn", "patch_size", "proj", "self", "stride", "super"], "glm_image/modeling_glm_image.py:GlmImageVisionPatchEmbed.forward": ["ModelVisionPatchEmbed", "Module", "Tensor", "class", "def", "dtype", "embed_dim", "forward", "hidden_states", "in_channels", "nn", "patch_size", "proj", "return", "self", "target_dtype", "to", "torch", "view", "weight"], "glm_image/modeling_glm_image.py:GlmImageVisionEmbeddings.__init__": ["Embedding", "ModelVisionConfig", "ModelVisionEmbeddings", "Module", "None", "__init__", "bilinear", "class", "config", "def", "embed_dim", "hidden_size", "image_size", "interpolated_method", "nn", "num_patches", "num_positions", "patch_size", "position_embedding", "self", "super"], "glm_image/modeling_glm_image.py:GlmImageVisionEmbeddings.forward": ["F", "False", "ModelVisionEmbeddings", "Module", "Tensor", "adapted_pos_embed", "adapted_pos_embed_fp32", "align_corners", "border", "cat", "class", "def", "device", "dim", "dtype", "embeddings", "float32", "for", "forward", "grid", "grid_sample", "h_coords", "hidden_size", "i", "if", "image_shapes", "in", "int", "interpolated_embed_fp32", "interpolated_method", "isinstance", "len", "lengths", "list", "long", "mode", "nn", "norm_h", "norm_w", "orig_size", "orig_size_sq", "padding_mode", "permute", "pos_embed_2d", "pos_embed_weight", "position_embedding", "range", "repeat", "return", "self", "shape", "squeeze", "stack", "target_h", "target_w", "tensor", "to", "torch", "unsqueeze", "view", "w_coords", "weight"], "glm_image/modeling_glm_image.py:GlmImageVisionBlock.__init__": ["GradientCheckpointingLayer", "LayerNorm", "ModelVisionAttention", "ModelVisionBlock", "ModelVisionConfig", "ModelVisionMLP", "None", "__init__", "attn", "class", "config", "def", "eps", "hidden_size", "layer_norm_eps", "mlp", "nn", "norm1", "norm2", "self", "super"], "glm_image/modeling_glm_image.py:GlmImageVisionBlock.forward": ["GradientCheckpointingLayer", "ModelVisionBlock", "Tensor", "TransformersKwargs", "Unpack", "attn", "class", "cu_seqlens", "def", "forward", "hidden_states", "kwargs", "mlp", "norm1", "norm2", "r", "residual", "return", "self", "torch"], "glm_image/modeling_glm_image.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "glm_image/modeling_glm_image.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cat", "cos", "def", "dim", "k", "k_embed", "k_pass", "k_rot", "q", "q_embed", "q_pass", "q_rot", "return", "rotary_dim", "rotate_half", "shape", "sin", "torch", "unsqueeze", "unsqueeze_dim"], "glm_image/modeling_glm_image.py:GlmImageTextAttention.__init__": ["False", "Linear", "ModelTextAttention", "ModelTextConfig", "Module", "None", "True", "__init__", "attention_bias", "attention_dropout", "bias", "class", "config", "def", "getattr", "head_dim", "hidden_size", "int", "is_causal", "k_proj", "layer_idx", "nn", "num_attention_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "q_proj", "rope_parameters", "scaling", "self", "super", "v_proj"], "glm_image/modeling_glm_image.py:GlmImageTextAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "FlashAttentionKwargs", "LongTensor", "ModelTextAttention", "Module", "None", "Tensor", "Unpack", "_attn_implementation", "apply_rotary_pos_emb", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "cache_kwargs", "cache_position", "class", "config", "contiguous", "cos", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "head_dim", "hidden_shape", "hidden_states", "if", "input_shape", "is", "k_proj", "key_states", "kwargs", "layer_idx", "nn", "not", "o_proj", "past_key_values", "position_embeddings", "q_proj", "query_states", "reshape", "return", "scaling", "self", "shape", "sin", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "view"], "glm_image/modeling_glm_image.py:GlmImageRMSNorm.__init__": ["ModelRMSNorm", "Module", "Parameter", "__init__", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "glm_image/modeling_glm_image.py:GlmImageRMSNorm.forward": ["ModelRMSNorm", "Module", "True", "class", "def", "dtype", "float32", "forward", "hidden_states", "input_dtype", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "glm_image/modeling_glm_image.py:GlmImageRMSNorm.extra_repr": ["ModelRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "variance_epsilon", "weight"], "glm_image/modeling_glm_image.py:GlmImageTextMLP.__init__": ["ACT2FN", "False", "Linear", "ModelTextMLP", "Module", "__init__", "activation_fn", "bias", "class", "config", "def", "down_proj", "gate_up_proj", "hidden_act", "hidden_size", "intermediate_size", "nn", "self", "super"], "glm_image/modeling_glm_image.py:GlmImageTextMLP.forward": ["FloatTensor", "ModelTextMLP", "Module", "activation_fn", "chunk", "class", "def", "dim", "down_proj", "forward", "gate", "gate_up_proj", "hidden_states", "nn", "return", "self", "torch", "up_states"], "glm_image/modeling_glm_image.py:GlmImageTextDecoderLayer.__init__": ["GradientCheckpointingLayer", "ModelRMSNorm", "ModelTextAttention", "ModelTextConfig", "ModelTextDecoderLayer", "ModelTextMLP", "__init__", "class", "config", "def", "eps", "hidden_size", "input_layernorm", "int", "layer_idx", "mlp", "post_attention_layernorm", "post_mlp_layernorm", "post_self_attn_layernorm", "rms_norm_eps", "self", "self_attn", "super"], "glm_image/modeling_glm_image.py:GlmImageTextDecoderLayer.forward": ["Cache", "False", "FloatTensor", "GradientCheckpointingLayer", "LongTensor", "ModelTextDecoderLayer", "None", "Tensor", "_", "attention_mask", "bool", "cache_position", "class", "def", "forward", "hidden_states", "input_layernorm", "kwargs", "mlp", "past_key_values", "position_embeddings", "position_ids", "post_attention_layernorm", "post_mlp_layernorm", "post_self_attn_layernorm", "residual", "return", "self", "self_attn", "torch", "tuple", "use_cache"], "glm_image/modeling_glm_image.py:GlmImagePreTrainedModel._init_weights": ["ModelPreTrainedModel", "PreTrainedModel", "_init_weights", "class", "def", "module", "self", "super"], "glm_image/modeling_glm_image.py:GlmImageVQVAEVectorQuantizer.__init__": ["Embedding", "ModelVQVAEConfig", "ModelVQVAEVectorQuantizer", "Module", "__init__", "beta", "class", "config", "def", "embed_dim", "embedding", "embedding_dim", "getattr", "nn", "num_embeddings", "self", "super"], "glm_image/modeling_glm_image.py:GlmImageVQVAEVectorQuantizer.forward": ["F", "ModelVQVAEVectorQuantizer", "Module", "Tensor", "True", "argmin", "bd", "beta", "bn", "class", "contiguous", "def", "detach", "dim", "distances", "dn", "einsum", "embedding", "embedding_dim", "forward", "hidden_state", "hidden_state_flattened", "hidden_state_quant", "keepdim", "loss", "mean", "min_encoding_indices", "nn", "normalize", "p", "permute", "return", "self", "shape", "sum", "torch", "transpose", "view", "weight"], "glm_image/modeling_glm_image.py:GlmImageVQVAE.__init__": ["Conv2d", "ModelPreTrainedModel", "ModelVQVAE", "ModelVQVAEConfig", "ModelVQVAEVectorQuantizer", "__init__", "class", "config", "def", "embed_dim", "eval", "latent_channels", "nn", "post_init", "post_quant_conv", "quant_conv", "quantize", "self", "super", "torch"], "glm_image/modeling_glm_image.py:GlmImageVQVAE.encode": ["ModelPreTrainedModel", "ModelVQVAE", "class", "def", "emb_loss", "encode", "hidden_states", "indices", "quant", "quant_conv", "quantize", "return", "self"], "glm_image/modeling_glm_image.py:GlmImageVisionModel.__init__": ["False", "ModelPreTrainedModel", "ModelVisionBlock", "ModelVisionConfig", "ModelVisionEmbeddings", "ModelVisionModel", "ModelVisionPatchEmbed", "ModuleList", "None", "_", "__init__", "blocks", "class", "config", "def", "depth", "embeddings", "for", "gradient_checkpointing", "head_dim", "hidden_size", "in", "nn", "num_heads", "patch_embed", "patch_size", "post_init", "range", "self", "spatial_merge_size", "super"], "glm_image/modeling_glm_image.py:GlmImageVisionModel.rot_pos_emb": ["ModelPreTrainedModel", "ModelVisionModel", "append", "arange", "cat", "class", "def", "dim", "expand", "flatten", "for", "grid_thw", "h", "hpos_ids", "in", "permute", "pos_ids", "repeat", "reshape", "return", "rot_pos_emb", "self", "spatial_merge_size", "stack", "t", "torch", "unsqueeze", "w", "wpos_ids"], "glm_image/modeling_glm_image.py:GlmImageVisionModel.forward": ["F", "ModelPreTrainedModel", "ModelVisionModel", "Tensor", "blk", "blocks", "class", "cu_seqlens", "cumsum", "def", "device", "dim", "dtype", "else", "embeddings", "for", "forward", "grid_thw", "hidden_states", "if", "image_type_ids", "in", "int32", "is_tracing", "jit", "kwargs", "pad", "patch_embed", "pixel_values", "repeat_interleave", "return", "rot_pos_emb", "self", "seqlens", "to", "tolist", "torch", "value"], "glm_image/modeling_glm_image.py:GlmImageTextRotaryEmbedding.__init__": ["Callable", "False", "ModelTextConfig", "ModelTextRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "get", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "mrope_section", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "glm_image/modeling_glm_image.py:GlmImageTextRotaryEmbedding.compute_default_rope_parameters": ["ModelTextConfig", "ModelTextRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "get", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "partial_rotary_factor", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "glm_image/modeling_glm_image.py:GlmImageTextRotaryEmbedding.forward": ["False", "ModelTextRotaryEmbedding", "Module", "None", "and", "apply_mrope", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "mrope_section", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "glm_image/modeling_glm_image.py:GlmImageTextRotaryEmbedding.apply_mrope": ["ModelTextRotaryEmbedding", "Module", "apply_mrope", "cat", "chunk", "chunks", "class", "def", "dim", "enumerate", "for", "freqs", "i", "in", "mrope_section", "nn", "result", "return", "section", "self", "split", "torch"], "glm_image/modeling_glm_image.py:GlmImageTextModel.__init__": ["Embedding", "False", "ModelPreTrainedModel", "ModelRMSNorm", "ModelTextConfig", "ModelTextDecoderLayer", "ModelTextModel", "ModelTextRotaryEmbedding", "ModuleList", "__init__", "class", "config", "def", "embed_tokens", "eps", "for", "gradient_checkpointing", "hidden_size", "in", "layer_idx", "layers", "nn", "norm", "num_hidden_layers", "pad_token_id", "padding_idx", "post_init", "range", "rms_norm_eps", "rotary_emb", "self", "super", "vocab_size"], "glm_image/modeling_glm_image.py:GlmImageTextModel.forward": ["BaseModelOutputWithPast", "Cache", "DynamicCache", "FlashAttentionKwargs", "FloatTensor", "LongTensor", "ModelPreTrainedModel", "ModelTextModel", "None", "Tensor", "Unpack", "ValueError", "You", "and", "arange", "attention_mask", "bool", "cache_position", "causal_mask", "class", "config", "create_causal_mask", "decoder_layer", "def", "device", "elif", "else", "embed_tokens", "exactly", "expand", "for", "forward", "get_seq_length", "hidden_states", "if", "in", "input_embeds", "input_ids", "inputs_embeds", "is", "is_tracing", "jit", "kwargs", "last_hidden_state", "layer_outputs", "layers", "mask_kwargs", "must", "ndim", "norm", "not", "of", "one", "or", "past_key_values", "past_seen_tokens", "position_embeddings", "position_ids", "raise", "return", "rotary_emb", "self", "shape", "specify", "text_position_ids", "torch", "tuple", "use_cache", "view"], "glm_image/modeling_glm_image.py:GlmImageModel.__init__": ["ModelModel", "ModelPreTrainedModel", "ModelTextModel", "ModelVQVAE", "ModelVisionModel", "None", "__init__", "_from_config", "class", "config", "def", "language_model", "post_init", "rope_deltas", "self", "super", "text_config", "vision_config", "visual", "vq_config", "vqmodel"], "glm_image/modeling_glm_image.py:GlmImageModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "language_model", "return", "self"], "glm_image/modeling_glm_image.py:GlmImageModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "language_model", "self", "set_input_embeddings", "value"], "glm_image/modeling_glm_image.py:GlmImageModel.get_rope_index": ["LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "_", "_cached_decode_position_ids", "_prefill_len", "and", "append", "arange", "attention_mask", "batch_idx", "batch_size", "cat", "class", "config", "curr_input_ids", "curr_position_ids", "current_pos", "decode_height_list", "decode_pos", "decode_temporal_list", "decode_width_list", "def", "device", "dim", "dtype", "else", "end", "end_position", "expand", "flatten", "for", "full", "get_rope_index", "grid", "grid_idx", "h", "h_grids", "h_indices", "i", "if", "image_end", "image_end_token_id", "image_grid_thw", "image_seq_length", "image_start", "image_start_token_id", "in", "input_ids", "is", "item", "len", "llm_pos_length", "llm_position_ids", "long", "max", "mrope_position_deltas", "not", "num_complete_images", "num_decode_grids", "num_height_grid", "num_width_grid", "ones", "position_height", "position_ids", "position_temporal", "position_width", "prev_image_end", "range", "repeat", "repeat_interleave", "return", "self", "seq_len", "shape", "stack", "start", "sum", "tensor", "text_positions", "to", "torch", "total_tokens", "tuple", "unsqueeze", "vision_position_ids", "w", "w_grids", "w_indices", "where", "zeros", "zip"], "glm_image/modeling_glm_image.py:GlmImageModel.get_image_features": ["FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "class", "def", "dtype", "get_image_features", "grid_thw", "image_embeds", "image_grid_thw", "pixel_values", "prod", "return", "self", "spatial_merge_size", "split", "split_sizes", "tolist", "torch", "type", "visual"], "glm_image/modeling_glm_image.py:GlmImageModel.get_placeholder_mask": ["LongTensor", "ModelModel", "ModelPreTrainedModel", "Number", "VQVAE", "ValueError", "class", "config", "def", "does", "f", "from", "get_placeholder_mask", "if", "image", "image_ids", "image_token_id", "input_ids", "item", "match", "n_image_tokens", "n_placeholder_tokens", "not", "number", "of", "placeholder", "raise", "return", "self", "shape", "special_image_mask", "sum", "tokens", "torch"], "glm_image/modeling_glm_image.py:GlmImageModel.forward": ["Cache", "FloatTensor", "LongTensor", "ModelModel", "ModelModelOutputWithPast", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "_", "_cached_decode_position_ids", "_prefill_len", "and", "attention_mask", "attention_mask_2d", "attentions", "batch_size", "cache_position", "cat", "class", "def", "device", "diagonal", "dim", "dim1", "dim2", "dtype", "else", "exactly", "expand", "finfo", "forward", "get_image_features", "get_image_tokens", "get_input_embeddings", "get_placeholder_mask", "get_rope_index", "hidden_states", "if", "image_embeds", "image_grid_thw", "image_ids", "input_ids", "inputs_embeds", "int", "is", "is_floating_point", "is_prefill_stage", "item", "kwargs", "language_model", "last_hidden_state", "masked_scatter", "min", "must", "ndim", "not", "of", "one", "or", "outputs", "past_key_values", "pixel_values", "position_ids", "r", "raise", "return", "rope_deltas", "self", "seq_length", "shape", "special_image_mask", "specify", "step", "to", "torch", "tuple", "unsqueeze", "view"], "glm_image/modeling_glm_image.py:GlmImageModel.get_image_tokens": ["FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "_", "all_image_toks", "append", "cat", "class", "contiguous", "def", "dim", "encode", "enumerate", "for", "get_image_tokens", "grid_h", "grid_t", "grid_w", "hidden_size", "hidden_states", "hidden_states_list", "hs", "i", "image_grid_thw", "image_toks", "in", "permute", "prod", "return", "self", "shape", "split", "split_sizes", "tolist", "torch", "view", "vqmodel"], "glm_image/modeling_glm_image.py:GlmImageForConditionalGeneration.__init__": ["False", "GenerationMixin", "Linear", "ModelForConditionalGeneration", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "text_config", "vision_vocab_size"], "glm_image/modeling_glm_image.py:GlmImageForConditionalGeneration.get_image_features": ["FloatTensor", "GenerationMixin", "LongTensor", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "class", "def", "get_image_features", "image_grid_thw", "model", "pixel_values", "return", "self", "torch"], "glm_image/modeling_glm_image.py:GlmImageForConditionalGeneration.get_image_tokens": ["FloatTensor", "GenerationMixin", "LongTensor", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "class", "def", "get_image_tokens", "hidden_states", "image_grid_thw", "model", "return", "self", "torch"], "glm_image/modeling_glm_image.py:GlmImageForConditionalGeneration.forward": ["Cache", "FloatTensor", "GenerationMixin", "LongTensor", "ModelCausalLMOutputWithPast", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "cache_position", "class", "config", "def", "else", "forward", "hidden_states", "if", "image_grid_thw", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "outputs", "past_key_values", "pixel_values", "position_ids", "r", "return", "rope_deltas", "self", "slice", "slice_indices", "text_config", "torch", "tuple", "vocab_size"], "glm_image/modeling_glm_image.py:GlmImageForConditionalGeneration.prepare_inputs_for_generation": ["False", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "True", "and", "attention_mask", "cache_position", "class", "def", "if", "image_grid_thw", "input_ids", "inputs_embeds", "is_first_iteration", "kwargs", "model_inputs", "not", "past_key_values", "pixel_values", "position_ids", "prepare_inputs_for_generation", "return", "self", "super", "use_cache"], "glm_image/modeling_glm_image.py:GlmImageForConditionalGeneration._get_image_nums": ["GenerationMixin", "LongTensor", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Tensor", "_get_image_nums", "class", "config", "def", "dim", "image_start_token_id", "input_ids", "is_image", "return", "self", "sum", "torch"], "glm_image/modeling_glm_image.py:GlmImageForConditionalGeneration._expand_inputs_for_generation": ["Any", "False", "GenerationMixin", "If", "LongTensor", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Tensor", "True", "ValueError", "_expand_dict_for_generation", "_expand_dict_for_generation_visual", "_expand_inputs_for_generation", "_get_image_nums", "_repeat_interleave_samples", "and", "bool", "cache_position", "cat", "class", "def", "defined", "dict", "dict_to_expand", "dim", "elif", "encoder_outputs", "expand_size", "for", "get", "if", "image_grid_thw", "image_nums", "in", "input_ids", "int", "is", "is_encoder_decoder", "isinstance", "key", "last_image", "lengths", "list", "make", "model_kwargs", "not", "pixel_values", "prod", "raise", "repeat", "repeat_args", "repeat_interleave", "repeat_times", "result", "return", "sample", "samples", "self", "split", "str", "sum", "sure", "that", "torch", "tuple", "visual_keys", "x"], "glmasr/modeling_glmasr.py:GlmAsrRotaryEmbedding.__init__": ["Callable", "False", "ModelConfig", "ModelRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "glmasr/modeling_glmasr.py:GlmAsrRotaryEmbedding.compute_default_rope_parameters": ["ModelConfig", "ModelRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "get", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "partial_rotary_factor", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "glmasr/modeling_glmasr.py:GlmAsrRotaryEmbedding.forward": ["False", "ModelRotaryEmbedding", "Module", "None", "and", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "glmasr/modeling_glmasr.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "glmasr/modeling_glmasr.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "glmasr/modeling_glmasr.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value", "value_states"], "glmasr/modeling_glmasr.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "None", "cat", "cos", "def", "dim", "k", "k_embed", "k_pass", "k_rot", "position_ids", "q", "q_embed", "q_pass", "q_rot", "return", "rotary_dim", "rotate_half", "shape", "sin", "torch", "unsqueeze", "unsqueeze_dim"], "glmasr/modeling_glmasr.py:GlmAsrAttention.__init__": ["False", "Linear", "ModelAttention", "ModelConfig", "Module", "True", "__init__", "attention_dropout", "bias", "class", "config", "def", "getattr", "head_dim", "hidden_size", "int", "is_causal", "k_proj", "layer_idx", "nn", "num_attention_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "q_proj", "scaling", "self", "super", "v_proj"], "glmasr/modeling_glmasr.py:GlmAsrAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "ModelAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_attn_implementation", "apply_rotary_pos_emb", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "class", "config", "contiguous", "cos", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "head_dim", "hidden_shape", "hidden_states", "if", "input_shape", "k_proj", "key_states", "kwargs", "nn", "not", "o_proj", "position_embeddings", "q_proj", "query_states", "reshape", "return", "scaling", "self", "shape", "sin", "torch", "training", "transpose", "tuple", "v_proj", "value_states", "view"], "glmasr/modeling_glmasr.py:GlmAsrMLP.__init__": ["ACT2FN", "Linear", "ModelMLP", "Module", "__init__", "act_fn", "class", "config", "def", "fc1", "fc2", "hidden_act", "hidden_size", "intermediate_size", "nn", "self", "super"], "glmasr/modeling_glmasr.py:GlmAsrMLP.forward": ["ModelMLP", "Module", "Tensor", "act_fn", "class", "def", "fc1", "fc2", "forward", "hidden_states", "nn", "return", "self", "torch"], "glmasr/modeling_glmasr.py:GlmAsrEncoderLayer.__init__": ["GradientCheckpointingLayer", "LayerNorm", "ModelAttention", "ModelConfig", "ModelEncoderLayer", "ModelMLP", "__init__", "class", "config", "def", "hidden_size", "input_layernorm", "int", "layer_idx", "mlp", "nn", "post_attention_layernorm", "self", "self_attn", "super"], "glmasr/modeling_glmasr.py:GlmAsrEncoderLayer.forward": ["GradientCheckpointingLayer", "ModelEncoderLayer", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "class", "def", "forward", "hidden_states", "input_layernorm", "kwargs", "mlp", "position_embeddings", "post_attention_layernorm", "residual", "return", "self", "self_attn", "torch", "tuple"], "glmasr/modeling_glmasr.py:GlmAsrEncoder.__init__": ["Conv1d", "False", "LayerNorm", "ModelEncoder", "ModelEncoderConfig", "ModelEncoderLayer", "ModelPreTrainedModel", "ModelRotaryEmbedding", "ModuleList", "__init__", "class", "config", "conv1", "conv2", "def", "for", "gradient_checkpointing", "hidden_size", "in", "kernel_size", "layer_idx", "layers", "nn", "norm", "num_hidden_layers", "num_mel_bins", "padding", "post_init", "range", "rotary_emb", "self", "stride", "super"], "glmasr/modeling_glmasr.py:GlmAsrEncoder.forward": ["BaseModelOutput", "ModelEncoder", "ModelPreTrainedModel", "None", "TransformersKwargs", "Unpack", "arange", "class", "conv1", "conv2", "def", "device", "encoder_layer", "for", "forward", "functional", "gelu", "hidden_states", "in", "input_features", "inputs_embeds", "kwargs", "last_hidden_state", "layers", "nn", "norm", "position_embeddings", "position_ids", "return", "rotary_emb", "self", "shape", "torch", "transpose"], "glmasr/modeling_glmasr.py:GlmAsrMultiModalProjector.__init__": ["ACT2FN", "Linear", "ModelConfig", "ModelMultiModalProjector", "Module", "__init__", "act", "audio_config", "class", "config", "def", "hidden_size", "intermediate_size", "linear_1", "linear_2", "nn", "projector_hidden_act", "self", "super", "text_config"], "glmasr/modeling_glmasr.py:GlmAsrMultiModalProjector.forward": ["ModelMultiModalProjector", "Module", "act", "audio_features", "class", "def", "forward", "hidden_states", "linear_1", "linear_2", "nn", "return", "self"], "glmasr/modeling_glmasr.py:GlmAsrForConditionalGeneration.__init__": ["AutoModel", "AutoModelForCausalLM", "GenerationMixin", "ModelForConditionalGeneration", "ModelMultiModalProjector", "ModelPreTrainedModel", "__init__", "audio_config", "audio_tower", "class", "config", "def", "from_config", "language_model", "multi_modal_projector", "post_init", "self", "super", "text_config", "vocab_size"], "glmasr/modeling_glmasr.py:GlmAsrForConditionalGeneration.get_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "language_model", "return", "self"], "glmasr/modeling_glmasr.py:GlmAsrForConditionalGeneration.set_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "language_model", "self", "set_input_embeddings", "value"], "glmasr/modeling_glmasr.py:GlmAsrForConditionalGeneration.get_output_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "get_output_embeddings", "language_model", "return", "self"], "glmasr/modeling_glmasr.py:GlmAsrForConditionalGeneration.set_output_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "language_model", "new_embeddings", "self", "set_output_embeddings"], "glmasr/modeling_glmasr.py:GlmAsrForConditionalGeneration.set_decoder": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "decoder", "def", "language_model", "self", "set_decoder"], "glmasr/modeling_glmasr.py:GlmAsrForConditionalGeneration.get_decoder": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "get_decoder", "language_model", "return", "self"], "glmasr/modeling_glmasr.py:GlmAsrForConditionalGeneration.get_audio_features": ["FloatTensor", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Tensor", "arange", "audio_config", "audio_embeds", "audio_hidden_states", "audio_lengths", "audio_outputs", "audio_tower", "class", "config", "def", "device", "for", "get_audio_features", "in", "input_features", "input_features_mask", "intermediate_size", "kernel_size", "last_hidden_state", "merge_factor", "multi_modal_projector", "padding", "post_lengths", "reshape", "return", "self", "shape", "stride", "sum", "to", "torch", "valid_mask"], "glmasr/modeling_glmasr.py:GlmAsrForConditionalGeneration.forward": ["Cache", "CausalLMOutputWithPast", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "and", "attention_mask", "audio_embeds", "audio_token_id", "audio_token_mask", "bool", "cache_position", "class", "config", "def", "device", "forward", "get_audio_features", "get_input_embeddings", "if", "input_features", "input_features_mask", "input_ids", "inputs_embeds", "int", "is", "kwargs", "labels", "language_model", "logits_to_keep", "masked_scatter", "not", "outputs", "past_key_values", "position_ids", "r", "return", "self", "to", "torch", "unsqueeze", "use_cache"], "glmasr/modeling_glmasr.py:GlmAsrForConditionalGeneration.prepare_inputs_for_generation": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "and", "args", "cache_position", "class", "def", "get", "if", "input_features", "input_features_mask", "is", "kwargs", "model_inputs", "not", "pop", "prepare_inputs_for_generation", "return", "self", "super"], "glpn/modeling_glpn.py:drop_path": ["False", "Model_path", "Model_prob", "Tensor", "bool", "def", "device", "div", "dtype", "float", "floor_", "if", "input", "keep_prob", "ndim", "not", "or", "output", "rand", "random_tensor", "return", "shape", "torch", "training"], "glpn/modeling_glpn.py:GLPNDropPath.__init__": ["Model", "Module", "None", "__init__", "class", "def", "drop_prob", "float", "nn", "self", "super"], "glpn/modeling_glpn.py:GLPNDropPath.forward": ["Model", "Module", "Tensor", "class", "def", "drop_path", "drop_prob", "forward", "hidden_states", "nn", "return", "self", "torch", "training"], "glpn/modeling_glpn.py:GLPNDropPath.extra_repr": ["Model", "Module", "class", "def", "drop_prob", "extra_repr", "f", "nn", "p", "return", "self", "str"], "glpn/modeling_glpn.py:GLPNOverlapPatchEmbeddings.__init__": ["Conv2d", "LayerNorm", "Model", "Module", "__init__", "class", "def", "hidden_size", "kernel_size", "layer_norm", "nn", "num_channels", "padding", "patch_size", "proj", "self", "stride", "super"], "glpn/modeling_glpn.py:GLPNOverlapPatchEmbeddings.forward": ["Model", "Module", "_", "class", "def", "embeddings", "flatten", "forward", "height", "layer_norm", "nn", "pixel_values", "proj", "return", "self", "shape", "transpose", "width"], "glpn/modeling_glpn.py:GLPNEfficientSelfAttention.__init__": ["Conv2d", "Dropout", "LayerNorm", "Linear", "Model", "Module", "The", "ValueError", "__init__", "a", "all_head_size", "attention", "attention_head_size", "attention_probs_dropout_prob", "class", "config", "def", "dropout", "f", "heads", "hidden", "hidden_size", "if", "int", "is", "kernel_size", "key", "layer_norm", "multiple", "nn", "not", "num_attention_heads", "number", "of", "query", "raise", "self", "sequence_reduction_ratio", "size", "sr", "sr_ratio", "stride", "super", "the", "value"], "glpn/modeling_glpn.py:GLPNEfficientSelfAttention.forward": ["False", "Model", "Module", "_", "all_head_size", "attention_head_size", "attention_probs", "attention_scores", "batch_size", "class", "context_layer", "contiguous", "def", "dim", "dropout", "else", "forward", "functional", "height", "hidden_states", "if", "key", "key_layer", "layer_norm", "math", "matmul", "new_context_layer_shape", "nn", "num_attention_heads", "num_channels", "output_attentions", "outputs", "permute", "query", "query_layer", "reshape", "return", "self", "seq_len", "seq_length", "shape", "size", "softmax", "sqrt", "sr", "sr_ratio", "torch", "transpose", "value", "value_layer", "view", "width"], "glpn/modeling_glpn.py:GLPNSelfOutput.__init__": ["Dropout", "Linear", "Model", "Module", "__init__", "class", "config", "def", "dense", "dropout", "hidden_dropout_prob", "hidden_size", "nn", "self", "super"], "glpn/modeling_glpn.py:GLPNSelfOutput.forward": ["Model", "Module", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self"], "glpn/modeling_glpn.py:GLPNAttention.__init__": ["Model", "ModelEfficientSelfAttention", "ModelSelfOutput", "Module", "__init__", "class", "config", "def", "hidden_size", "nn", "num_attention_heads", "output", "self", "sequence_reduction_ratio", "super"], "glpn/modeling_glpn.py:GLPNAttention.forward": ["False", "Model", "Module", "attention_output", "class", "def", "forward", "height", "hidden_states", "nn", "output", "output_attentions", "outputs", "return", "self", "self_outputs", "width"], "glpn/modeling_glpn.py:GLPNDWConv.__init__": ["Conv2d", "Model", "Module", "True", "__init__", "bias", "class", "def", "dim", "dwconv", "groups", "nn", "self", "super"], "glpn/modeling_glpn.py:GLPNDWConv.forward": ["Model", "Module", "batch_size", "class", "def", "dwconv", "flatten", "forward", "height", "hidden_states", "nn", "num_channels", "return", "self", "seq_len", "shape", "transpose", "view", "width"], "glpn/modeling_glpn.py:GLPNMixFFN.__init__": ["ACT2FN", "Dropout", "Linear", "Model", "ModelDWConv", "Module", "None", "__init__", "class", "config", "def", "dense1", "dense2", "dropout", "dwconv", "else", "hidden_act", "hidden_dropout_prob", "hidden_features", "if", "in_features", "intermediate_act_fn", "isinstance", "nn", "or", "out_features", "self", "str", "super"], "glpn/modeling_glpn.py:GLPNMixFFN.forward": ["Model", "Module", "class", "def", "dense1", "dense2", "dropout", "dwconv", "forward", "height", "hidden_states", "intermediate_act_fn", "nn", "return", "self", "width"], "glpn/modeling_glpn.py:GLPNLayer.__init__": ["Identity", "LayerNorm", "Model", "ModelAttention", "ModelDropPath", "ModelMixFFN", "Module", "__init__", "attention", "class", "config", "def", "drop_path", "else", "hidden_features", "hidden_size", "if", "in_features", "int", "layer_norm_1", "layer_norm_2", "mlp", "mlp_hidden_size", "mlp_ratio", "nn", "num_attention_heads", "self", "sequence_reduction_ratio", "super"], "glpn/modeling_glpn.py:GLPNLayer.forward": ["False", "Model", "Module", "attention", "attention_output", "class", "def", "drop_path", "forward", "height", "hidden_states", "layer_norm_1", "layer_norm_2", "layer_output", "mlp", "mlp_output", "nn", "output_attentions", "outputs", "return", "self", "self_attention_outputs", "width"], "glpn/modeling_glpn.py:GLPNEncoder.__init__": ["LayerNorm", "Model", "ModelLayer", "ModelOverlapPatchEmbeddings", "Module", "ModuleList", "__init__", "append", "block", "blocks", "class", "config", "cpu", "cur", "def", "depths", "device", "dpr", "drop_path", "drop_path_rate", "else", "embeddings", "for", "hidden_size", "hidden_sizes", "i", "if", "in", "item", "j", "layer_norm", "layers", "linspace", "mlp_ratio", "mlp_ratios", "nn", "num_attention_heads", "num_channels", "num_encoder_blocks", "patch_embeddings", "patch_size", "patch_sizes", "range", "self", "sequence_reduction_ratio", "sr_ratios", "stride", "strides", "sum", "super", "torch", "x"], "glpn/modeling_glpn.py:GLPNEncoder.forward": ["BaseModelOutput", "False", "Model", "Module", "None", "True", "all_hidden_states", "all_self_attentions", "attentions", "batch_size", "blk", "block", "block_layer", "class", "contiguous", "def", "else", "embedding_layer", "enumerate", "for", "forward", "height", "hidden_states", "i", "idx", "if", "in", "is", "last_hidden_state", "layer_norm", "layer_outputs", "nn", "norm_layer", "not", "output_attentions", "output_hidden_states", "patch_embeddings", "permute", "pixel_values", "reshape", "return", "return_dict", "self", "shape", "tuple", "v", "width", "x", "zip"], "glpn/modeling_glpn.py:GLPNModel.__init__": ["Model", "ModelEncoder", "ModelPreTrainedModel", "__init__", "class", "config", "def", "encoder", "post_init", "self", "super"], "glpn/modeling_glpn.py:GLPNModel.forward": ["BaseModelOutput", "FloatTensor", "Model", "ModelPreTrainedModel", "None", "attentions", "bool", "class", "config", "def", "else", "encoder", "encoder_outputs", "forward", "hidden_states", "if", "is", "kwargs", "last_hidden_state", "not", "output_attentions", "output_hidden_states", "pixel_values", "return", "return_dict", "self", "sequence_output", "torch", "tuple", "use_return_dict"], "glpn/modeling_glpn.py:GLPNSelectiveFeatureFusion.__init__": ["BatchNorm2d", "Conv2d", "Model", "Module", "ReLU", "Sequential", "Sigmoid", "__init__", "class", "convolutional_layer1", "convolutional_layer2", "convolutional_layer3", "def", "in_channel", "in_channels", "int", "kernel_size", "nn", "out_channels", "padding", "self", "sigmoid", "stride", "super"], "glpn/modeling_glpn.py:GLPNSelectiveFeatureFusion.forward": ["Model", "Module", "attn", "cat", "class", "convolutional_layer1", "convolutional_layer2", "convolutional_layer3", "def", "dim", "features", "forward", "global_features", "hybrid_features", "local_features", "nn", "return", "self", "sigmoid", "torch", "unsqueeze"], "glpn/modeling_glpn.py:GLPNDecoderStage.__init__": ["Conv2d", "False", "Identity", "Model", "ModelSelectiveFeatureFusion", "Module", "Upsample", "__init__", "align_corners", "bilinear", "class", "convolution", "def", "else", "fusion", "if", "in_channels", "kernel_size", "mode", "nn", "not", "out_channels", "scale_factor", "self", "should_skip", "super", "upsample"], "glpn/modeling_glpn.py:GLPNDecoderStage.forward": ["Model", "Module", "None", "class", "convolution", "def", "forward", "fusion", "hidden_state", "if", "is", "nn", "not", "residual", "return", "self", "upsample"], "glpn/modeling_glpn.py:GLPNDecoder.__init__": ["False", "Model", "ModelStage", "Module", "ModuleList", "None", "Upsample", "__init__", "align_corners", "bilinear", "class", "config", "decoder_hidden_size", "def", "final_upsample", "for", "fusion", "hidden_size", "hidden_sizes", "in", "mode", "nn", "out_channels", "reserved_hidden_sizes", "scale_factor", "self", "stages", "super"], "glpn/modeling_glpn.py:GLPNDecoder.forward": ["Model", "Module", "None", "Tensor", "append", "class", "def", "final_upsample", "for", "forward", "hidden_state", "hidden_states", "in", "list", "nn", "return", "self", "stage", "stage_hidden_state", "stage_hidden_states", "stages", "torch", "zip"], "glpn/modeling_glpn.py:SiLogLoss.__init__": ["Module", "SiLogLoss", "__init__", "class", "def", "lambd", "nn", "self", "super"], "glpn/modeling_glpn.py:SiLogLoss.forward": ["Module", "SiLogLoss", "class", "def", "detach", "diff_log", "forward", "lambd", "log", "loss", "mean", "nn", "pow", "pred", "return", "self", "sqrt", "target", "torch", "valid_mask"], "glpn/modeling_glpn.py:GLPNDepthEstimationHead.__init__": ["Conv2d", "False", "Model", "Module", "ReLU", "Sequential", "__init__", "channels", "class", "config", "decoder_hidden_size", "def", "head", "inplace", "kernel_size", "nn", "padding", "self", "stride", "super"], "glpn/modeling_glpn.py:GLPNDepthEstimationHead.forward": ["Model", "Module", "Tensor", "class", "config", "def", "dim", "forward", "head", "head_in_index", "hidden_states", "list", "max_depth", "nn", "predicted_depth", "return", "self", "sigmoid", "squeeze", "torch"], "glpn/modeling_glpn.py:GLPNForDepthEstimation.__init__": ["Model", "ModelDecoder", "ModelDepthEstimationHead", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "config", "decoder", "def", "head", "post_init", "self", "super"], "glpn/modeling_glpn.py:GLPNForDepthEstimation.forward": ["DepthEstimatorOutput", "FloatTensor", "Model", "ModelPreTrainedModel", "None", "SiLogLoss", "Tensor", "True", "attentions", "bool", "class", "config", "decoder", "def", "else", "forward", "head", "hidden_states", "if", "is", "kwargs", "labels", "loss", "loss_fct", "not", "out", "output", "output_attentions", "output_hidden_states", "outputs", "pixel_values", "predicted_depth", "r", "return", "return_dict", "self", "torch", "tuple", "use_return_dict"], "got_ocr2/modeling_got_ocr2.py:GotOcr2MLPBlock.__init__": ["ACT2FN", "Linear", "ModelMLPBlock", "Module", "__init__", "act", "class", "config", "def", "hidden_act", "hidden_size", "lin1", "lin2", "mlp_dim", "nn", "self", "super"], "got_ocr2/modeling_got_ocr2.py:GotOcr2MLPBlock.forward": ["ModelMLPBlock", "Module", "Tensor", "act", "class", "def", "forward", "hidden_states", "lin1", "lin2", "nn", "return", "self", "torch"], "got_ocr2/modeling_got_ocr2.py:GotOcr2VisionAttention.__init__": ["Input", "Linear", "ModelVisionAttention", "Module", "None", "Parameter", "ValueError", "__init__", "attention_dropout", "be", "bias", "class", "config", "def", "dropout", "else", "encoding", "head_dim", "hidden_size", "if", "image_size", "input_size", "is", "must", "nn", "num_attention_heads", "patch_size", "positional", "proj", "provided", "qkv", "qkv_bias", "raise", "rel_pos_h", "rel_pos_w", "relative", "scale", "self", "size", "super", "torch", "use_rel_pos", "using", "window_size", "zeros"], "got_ocr2/modeling_got_ocr2.py:GotOcr2VisionAttention.get_rel_pos": ["F", "ModelVisionAttention", "Module", "None", "Tensor", "arange", "class", "def", "get_rel_pos", "int", "interpolate", "k_coords", "k_size", "linear", "long", "max", "max_rel_dist", "mode", "nn", "permute", "q_coords", "q_size", "rel_pos", "rel_pos_resized", "relative_coords", "reshape", "return", "self", "shape", "size", "torch"], "got_ocr2/modeling_got_ocr2.py:GotOcr2VisionAttention.get_decomposed_rel_pos": ["ModelVisionAttention", "Module", "None", "Tensor", "_", "batch_size", "bhwc", "bhwk", "class", "decomposed_rel_pos", "def", "dim", "einsum", "get_decomposed_rel_pos", "get_rel_pos", "hkc", "int", "k_size", "key_height", "key_width", "nn", "q_size", "query", "query_height", "query_width", "rel_h", "rel_pos_h", "rel_pos_w", "rel_w", "relative_position_height", "relative_position_width", "reshape", "reshaped_query", "return", "self", "shape", "torch", "tuple", "wkc"], "got_ocr2/modeling_got_ocr2.py:GotOcr2VisionAttention.forward": ["ModelVisionAttention", "Module", "None", "Tensor", "_", "attn_output", "attn_probs", "attn_weights", "batch_size", "class", "decomposed_rel_pos", "def", "dim", "dropout", "dtype", "float32", "forward", "functional", "get_decomposed_rel_pos", "height", "hidden_states", "if", "key", "nn", "num_attention_heads", "output_attentions", "p", "permute", "proj", "qkv", "query", "rel_pos_h", "rel_pos_w", "reshape", "reshape_as", "return", "scale", "self", "shape", "softmax", "to", "torch", "training", "transpose", "tuple", "unbind", "use_rel_pos", "value", "width"], "got_ocr2/modeling_got_ocr2.py:GotOcr2VisionLayer.__init__": ["GradientCheckpointingLayer", "LayerNorm", "ModelMLPBlock", "ModelVisionAttention", "ModelVisionLayer", "__init__", "attn", "class", "config", "def", "eps", "hidden_size", "layer_norm1", "layer_norm2", "layer_norm_eps", "mlp", "nn", "self", "super", "window_size"], "got_ocr2/modeling_got_ocr2.py:GotOcr2VisionLayer.window_partition": ["F", "GradientCheckpointingLayer", "ModelVisionLayer", "Tensor", "batch_size", "channel", "class", "contiguous", "def", "height", "hidden_states", "int", "pad", "pad_h", "pad_height", "pad_w", "pad_width", "permute", "reshape", "return", "self", "shape", "torch", "tuple", "width", "window_partition", "window_size", "windows"], "got_ocr2/modeling_got_ocr2.py:GotOcr2VisionLayer.window_unpartition": ["GradientCheckpointingLayer", "ModelVisionLayer", "Tensor", "batch_size", "class", "contiguous", "def", "height", "hidden_states", "int", "original_shape", "pad_height", "pad_width", "padding_shape", "permute", "reshape", "return", "self", "shape", "torch", "tuple", "width", "window_size", "window_unpartition", "windows"], "got_ocr2/modeling_got_ocr2.py:GotOcr2VisionLayer.forward": ["FloatTensor", "GradientCheckpointingLayer", "ModelVisionLayer", "Tensor", "attn", "attn_weights", "class", "def", "forward", "height", "hidden_states", "if", "layer_norm1", "layer_norm2", "layernorm_output", "mlp", "padding_shape", "residual", "return", "self", "shape", "torch", "tuple", "width", "window_partition", "window_size", "window_unpartition"], "got_ocr2/modeling_got_ocr2.py:GotOcr2PreTrainedModel._init_weights": ["ModelPreTrainedModel", "ModelVisionAttention", "ModelVisionEncoder", "None", "PreTrainedModel", "_init_weights", "class", "def", "elif", "if", "init", "is", "isinstance", "module", "not", "pos_embed", "rel_pos_h", "rel_pos_w", "self", "super", "use_rel_pos", "zeros_"], "got_ocr2/modeling_got_ocr2.py:GotOcr2PatchEmbeddings.__init__": ["Conv2d", "Iterable", "ModelPatchEmbeddings", "Module", "__init__", "abc", "class", "collections", "config", "def", "else", "hidden_size", "if", "image_size", "isinstance", "kernel_size", "nn", "num_channels", "num_patches", "patch_size", "projection", "self", "stride", "super"], "got_ocr2/modeling_got_ocr2.py:GotOcr2PatchEmbeddings.forward": ["Input", "Make", "ModelPatchEmbeddings", "Module", "ValueError", "batch_size", "channel", "class", "configuration", "def", "dimension", "doesn", "embeddings", "f", "forward", "height", "if", "image", "image_size", "in", "match", "model", "nn", "num_channels", "of", "one", "or", "permute", "pixel", "pixel_values", "projection", "raise", "return", "self", "set", "shape", "size", "sure", "t", "that", "the", "values", "width", "with"], "got_ocr2/modeling_got_ocr2.py:GotOcr2LayerNorm.__init__": ["LayerNorm", "ModelLayerNorm", "NotImplementedError", "Unsupported", "__init__", "channels_first", "channels_last", "class", "data", "data_format", "def", "eps", "f", "format", "if", "in", "kwargs", "nn", "normalized_shape", "not", "raise", "self", "super"], "got_ocr2/modeling_got_ocr2.py:GotOcr2LayerNorm.forward": ["LayerNorm", "ModelLayerNorm", "Tensor", "channels_first", "class", "data_format", "def", "else", "features", "forward", "if", "nn", "permute", "return", "self", "super", "torch"], "got_ocr2/modeling_got_ocr2.py:GotOcr2VisionNeck.__init__": ["Conv2d", "False", "ModelLayerNorm", "ModelVisionConfig", "ModelVisionNeck", "Module", "__init__", "bias", "channels_first", "class", "config", "conv1", "conv2", "data_format", "def", "hidden_size", "kernel_size", "layer_norm1", "layer_norm2", "nn", "output_channels", "padding", "self", "super"], "got_ocr2/modeling_got_ocr2.py:GotOcr2VisionNeck.forward": ["ModelVisionNeck", "Module", "class", "conv1", "conv2", "def", "forward", "hidden_states", "layer_norm1", "layer_norm2", "nn", "permute", "return", "self"], "got_ocr2/modeling_got_ocr2.py:GotOcr2VisionEncoder.__init__": ["False", "ModelPatchEmbeddings", "ModelPreTrainedModel", "ModelVisionConfig", "ModelVisionEncoder", "ModelVisionLayer", "ModelVisionNeck", "ModuleList", "None", "Parameter", "__init__", "append", "class", "config", "def", "else", "for", "global_attn_indexes", "gradient_checkpointing", "hidden_size", "i", "if", "image_size", "in", "layer", "layers", "neck", "nn", "not", "num_hidden_layers", "patch_embed", "patch_size", "pos_embed", "post_init", "range", "self", "super", "torch", "use_abs_pos", "window_size", "zeros"], "got_ocr2/modeling_got_ocr2.py:GotOcr2VisionEncoder.get_input_embeddings": ["ModelPreTrainedModel", "ModelVisionEncoder", "class", "def", "get_input_embeddings", "patch_embed", "return", "self"], "got_ocr2/modeling_got_ocr2.py:GotOcr2VisionEncoder.forward": ["FloatTensor", "ModelPreTrainedModel", "ModelVisionEncoder", "ModelVisionEncoderOutput", "None", "TransformersKwargs", "Unpack", "ValueError", "You", "class", "def", "for", "forward", "have", "hidden_states", "if", "in", "is", "kwargs", "last_hidden_state", "layer_module", "layers", "neck", "not", "patch_embed", "pixel_values", "pos_embed", "raise", "return", "self", "specify", "to", "torch"], "got_ocr2/modeling_got_ocr2.py:GotOcr2MultiModalProjector.__init__": ["Conv2d", "False", "Linear", "ModelConfig", "ModelMultiModalProjector", "Module", "__init__", "bias", "class", "config", "conv_upsampler1", "conv_upsampler2", "def", "hidden_size", "kernel_size", "language_hidden_size", "multimodal_projector", "nn", "output_channels", "padding", "self", "stride", "super", "text_config", "vision_config", "vision_output_channels"], "got_ocr2/modeling_got_ocr2.py:GotOcr2MultiModalProjector.forward": ["ModelMultiModalProjector", "Module", "Tensor", "class", "conv_upsampler1", "conv_upsampler2", "def", "flatten", "forward", "hidden_state", "multimodal_projector", "nn", "permute", "return", "self", "torch", "vision_embeddings"], "got_ocr2/modeling_got_ocr2.py:GotOcr2Model.__init__": ["AutoModel", "ModelConfig", "ModelModel", "ModelMultiModalProjector", "ModelPreTrainedModel", "ModelVisionEncoder", "__init__", "class", "config", "def", "from_config", "language_model", "multi_modal_projector", "post_init", "self", "super", "text_config", "vision_config", "vision_tower"], "got_ocr2/modeling_got_ocr2.py:GotOcr2Model.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "language_model", "return", "self"], "got_ocr2/modeling_got_ocr2.py:GotOcr2Model.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "language_model", "self", "set_input_embeddings", "value"], "got_ocr2/modeling_got_ocr2.py:GotOcr2Model.get_image_features": ["FloatTensor", "ModelModel", "ModelPreTrainedModel", "class", "def", "get_image_features", "image_outputs", "last_hidden_state", "multi_modal_projector", "pixel_values", "return", "self", "torch", "vision_tower"], "got_ocr2/modeling_got_ocr2.py:GotOcr2Model.get_placeholder_mask": ["FloatTensor", "Image", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "ValueError", "all", "and", "class", "config", "def", "device", "do", "dtype", "else", "expand_as", "f", "features", "get_input_embeddings", "get_placeholder_mask", "if", "image", "image_features", "image_token_id", "input_ids", "inputs_embeds", "is", "long", "match", "n_image_features", "n_image_tokens", "not", "numel", "raise", "return", "self", "shape", "special_image_mask", "sum", "tensor", "to", "tokens", "torch", "unsqueeze"], "got_ocr2/modeling_got_ocr2.py:GotOcr2Model.forward": ["Cache", "FloatTensor", "LongTensor", "ModelModel", "ModelModelOutputWithPast", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "ValueError", "You", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "device", "dtype", "else", "exactly", "forward", "get_image_features", "get_input_embeddings", "get_placeholder_mask", "hidden_states", "if", "image_features", "image_hidden_states", "input_ids", "inputs_embeds", "is", "kwargs", "language_model", "last_hidden_state", "masked_scatter", "must", "not", "of", "one", "or", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "pixel_values", "position_ids", "raise", "return", "return_dict", "self", "special_image_mask", "specify", "to", "torch", "tuple", "use_cache", "use_return_dict"], "got_ocr2/modeling_got_ocr2.py:GotOcr2ForConditionalGeneration.__init__": ["False", "GenerationMixin", "Linear", "ModelConfig", "ModelForConditionalGeneration", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "text_config", "vocab_size"], "got_ocr2/modeling_got_ocr2.py:GotOcr2ForConditionalGeneration.get_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "model", "return", "self"], "got_ocr2/modeling_got_ocr2.py:GotOcr2ForConditionalGeneration.set_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "model", "self", "set_input_embeddings", "value"], "got_ocr2/modeling_got_ocr2.py:GotOcr2ForConditionalGeneration.get_output_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "Module", "class", "def", "get_output_embeddings", "lm_head", "nn", "return", "self"], "got_ocr2/modeling_got_ocr2.py:GotOcr2ForConditionalGeneration.get_image_features": ["FloatTensor", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "class", "def", "get_image_features", "int", "kwargs", "list", "model", "pixel_values", "return", "self", "str", "torch", "vision_feature_layer", "vision_feature_select_strategy"], "got_ocr2/modeling_got_ocr2.py:GotOcr2ForConditionalGeneration.forward": ["Cache", "FloatTensor", "GenerationMixin", "LongTensor", "ModelCausalLMOutputWithPast", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "else", "forward", "hidden_states", "if", "image_hidden_states", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "pixel_values", "position_ids", "r", "return", "return_dict", "self", "slice", "slice_indices", "text_config", "torch", "tuple", "use_cache", "use_return_dict", "vocab_size"], "got_ocr2/modeling_got_ocr2.py:GotOcr2ForConditionalGeneration.prepare_inputs_for_generation": ["False", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "True", "attention_mask", "cache_position", "class", "def", "get", "if", "input_ids", "inputs_embeds", "is_first_iteration", "kwargs", "logits_to_keep", "model_inputs", "not", "or", "past_key_values", "pixel_values", "prepare_inputs_for_generation", "return", "self", "super", "use_cache"], "gpt2/modeling_gpt2.py:eager_attention_forward": ["Model_attention_forward", "None", "attention_mask", "attn_dropout", "attn_output", "attn_weights", "bias", "causal_mask", "def", "device", "dim", "dtype", "finfo", "float", "full", "functional", "if", "is", "is_cross_attention", "key", "key_length", "kwargs", "layer_idx", "mask_value", "matmul", "min", "module", "nn", "not", "query", "query_length", "return", "scale_attn_by_inverse_layer_idx", "scale_attn_weights", "shape", "size", "softmax", "to", "torch", "transpose", "type", "value", "where"], "gpt2/modeling_gpt2.py:GPT2Attention.__init__": ["Conv1D", "Dropout", "False", "Model", "Module", "None", "ValueError", "__init__", "and", "attn_dropout", "attn_pdrop", "be", "bias", "bool", "by", "c_attn", "c_proj", "class", "config", "def", "divisible", "dtype", "else", "embed_dim", "f", "got", "head_dim", "hidden_size", "if", "is_causal", "is_cross_attention", "layer_idx", "max_position_embeddings", "max_positions", "must", "nn", "not", "num_attention_heads", "num_heads", "ones", "persistent", "q_attn", "raise", "register_buffer", "reorder_and_upcast_attn", "resid_dropout", "resid_pdrop", "scale_attn_by_inverse_layer_idx", "scale_attn_weights", "self", "split_size", "super", "torch", "tril", "view"], "gpt2/modeling_gpt2.py:GPT2Attention._upcast_and_reordered_attn": ["Error", "False", "Model", "Module", "None", "RuntimeError", "_", "_upcast_and_reordered_attn", "alpha", "attention_mask", "attn_dropout", "attn_output", "attn_weights", "baddbmm", "beta", "bias", "bsz", "causal_mask", "class", "def", "device", "dim", "dk", "does", "dtype", "empty", "enabled", "finfo", "float", "float32", "functional", "have", "if", "is", "is_cross_attention", "k", "k_seq_len", "key", "key_length", "layer_idx", "mask_value", "matmul", "maybe_autocast", "min", "nn", "not", "num_heads", "q", "q_seq_len", "query", "query_length", "raise", "reshape", "return", "scale_attn_by_inverse_layer_idx", "scale_attn_weights", "scale_factor", "self", "size", "softmax", "tensor", "torch", "transpose", "type", "upcasting", "value", "where", "with"], "gpt2/modeling_gpt2.py:GPT2Attention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "EncoderDecoderCache", "False", "FloatTensor", "If", "LongTensor", "Model", "Module", "None", "Please", "Tensor", "True", "ValueError", "_attn_implementation", "_upcast_and_reordered_attn", "and", "as", "attention", "attention_interface", "attention_mask", "attn_dropout", "attn_output", "attn_weights", "be", "bool", "c_attn", "c_proj", "cache_position", "class", "config", "contiguous", "cross", "cross_attention_cache", "curr_past_key_values", "def", "defined", "dim", "dropout", "eager", "eager_attention_forward", "else", "encoder_attention_mask", "encoder_hidden_states", "forward", "get", "hasattr", "have", "head_dim", "hidden_states", "if", "instantiate", "is", "is_cross_attention", "is_updated", "isinstance", "key_states", "keys", "kwargs", "layer_idx", "layers", "make", "nn", "not", "or", "output_attentions", "p", "past_key_values", "q_attn", "query_states", "raise", "reorder_and_upcast_attn", "reshape", "resid_dropout", "return", "self", "self_attention_cache", "shape", "shape_kv", "shape_q", "split", "split_size", "sure", "the", "to", "torch", "training", "transpose", "tuple", "update", "used", "using_eager", "value_states", "values", "view", "weights", "with"], "gpt2/modeling_gpt2.py:GPT2MLP.__init__": ["ACT2FN", "Conv1D", "Dropout", "Model", "Module", "__init__", "act", "activation_function", "c_fc", "c_proj", "class", "config", "def", "dropout", "embed_dim", "hidden_size", "intermediate_size", "nn", "resid_pdrop", "self", "super"], "gpt2/modeling_gpt2.py:GPT2MLP.forward": ["FloatTensor", "Model", "Module", "None", "act", "c_fc", "c_proj", "class", "def", "dropout", "forward", "hidden_states", "nn", "return", "self", "torch", "tuple"], "gpt2/modeling_gpt2.py:GPT2Block.__init__": ["GradientCheckpointingLayer", "LayerNorm", "Model", "ModelAttention", "ModelMLP", "None", "True", "__init__", "add_cross_attention", "attn", "class", "config", "crossattention", "def", "else", "eps", "hidden_size", "if", "inner_dim", "is", "is_cross_attention", "layer_idx", "layer_norm_epsilon", "ln_1", "ln_2", "ln_cross_attn", "mlp", "n_inner", "nn", "not", "self", "super"], "gpt2/modeling_gpt2.py:GPT2Block.forward": ["Cache", "False", "FloatTensor", "GradientCheckpointingLayer", "If", "LongTensor", "Model", "None", "Tensor", "True", "ValueError", "add_cross_attention", "are", "attention", "attention_mask", "attn", "attn_output", "be", "bool", "by", "cache_position", "class", "config", "cross", "cross_attn_output", "cross_attn_weights", "crossattention", "def", "encoder_attention_mask", "encoder_hidden_states", "f", "feed_forward_hidden_states", "forward", "has", "hasattr", "hidden_states", "if", "instantiated", "is", "kwargs", "layers", "ln_1", "ln_2", "ln_cross_attn", "mlp", "not", "output_attentions", "outputs", "passed", "past_key_values", "raise", "residual", "return", "self", "self_attn_weights", "setting", "to", "torch", "tuple", "use_cache", "with"], "gpt2/modeling_gpt2.py:GPT2SequenceSummary.__init__": ["Callable", "Dropout", "Identity", "Linear", "Model", "ModelConfig", "Module", "None", "NotImplementedError", "__init__", "activation", "activation_string", "and", "attn", "class", "config", "def", "else", "first_dropout", "get_activation", "getattr", "hasattr", "hidden_size", "if", "last", "last_dropout", "nn", "num_classes", "num_labels", "raise", "self", "summary", "summary_activation", "summary_first_dropout", "summary_last_dropout", "summary_proj_to_labels", "summary_type", "summary_use_proj", "super"], "gpt2/modeling_gpt2.py:GPT2SequenceSummary.forward": ["FloatTensor", "LongTensor", "Model", "Module", "None", "NotImplementedError", "activation", "attn", "class", "cls_index", "def", "dim", "dtype", "elif", "else", "expand", "first", "first_dropout", "forward", "full_like", "gather", "hidden_states", "if", "is", "last", "last_dropout", "long", "mean", "nn", "output", "raise", "return", "self", "shape", "size", "squeeze", "summary", "summary_type", "torch", "unsqueeze"], "gpt2/modeling_gpt2.py:GPT2PreTrainedModel._init_weights": ["Conv1D", "Embedding", "False", "LayerNorm", "Linear", "Model", "ModelAttention", "None", "PreTrainedModel", "_init_weights", "_is_hf_initialized", "and", "bias", "bool", "c_proj", "class", "config", "copy_", "def", "dtype", "elif", "for", "getattr", "if", "in", "init", "initializer_range", "is", "isinstance", "math", "max_position_embeddings", "max_positions", "mean", "module", "n_layer", "name", "named_parameters", "nn", "normal_", "not", "ones", "ones_", "p", "padding_idx", "self", "sqrt", "std", "torch", "tril", "view", "weight", "zeros_"], "gpt2/modeling_gpt2.py:GPT2Model.__init__": ["Dropout", "Embedding", "False", "LayerNorm", "Model", "ModelBlock", "ModelPreTrainedModel", "ModuleList", "__init__", "_attn_implementation", "class", "config", "def", "drop", "embd_pdrop", "embed_dim", "eps", "for", "gradient_checkpointing", "h", "hidden_size", "i", "in", "layer_idx", "layer_norm_epsilon", "ln_f", "max_position_embeddings", "nn", "num_hidden_layers", "post_init", "range", "self", "super", "vocab_size", "wpe", "wte"], "gpt2/modeling_gpt2.py:GPT2Model.get_input_embeddings": ["Model", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "return", "self", "wte"], "gpt2/modeling_gpt2.py:GPT2Model.set_input_embeddings": ["Model", "ModelPreTrainedModel", "class", "def", "new_embeddings", "self", "set_input_embeddings", "wte"], "gpt2/modeling_gpt2.py:GPT2Model.forward": ["BaseModelOutputWithPastAndCrossAttentions", "Cache", "DynamicCache", "EncoderDecoderCache", "False", "FloatTensor", "LongTensor", "Model", "ModelPreTrainedModel", "None", "Setting", "Tensor", "True", "ValueError", "You", "_", "_attn_implementation", "_prepare_4d_attention_mask_for_sdpa", "_use_sdpa", "add_cross_attention", "all_cross_attentions", "all_hidden_states", "all_self_attentions", "and", "arange", "at", "attention_mask", "attentions", "batch_size", "block", "bool", "both", "cache_position", "cannot", "causal_mask", "checkpointing", "class", "config", "create_causal_mask", "cross_attentions", "def", "device", "drop", "dtype", "either", "elif", "else", "encoder_attention_mask", "encoder_batch_size", "encoder_hidden_shape", "encoder_hidden_states", "encoder_sequence_length", "enumerate", "flash_attention_2", "for", "forward", "get_seq_length", "gradient", "gradient_checkpointing", "h", "have", "hidden_states", "i", "if", "in", "incompatible", "input_embeds", "input_ids", "input_shape", "inputs_embeds", "invert_attention_mask", "is", "isinstance", "kwargs", "last_hidden_state", "ln_f", "logger", "mask", "ndim", "not", "ones", "or", "output_attentions", "output_hidden_states", "output_shape", "outputs", "past_key_values", "past_seen_tokens", "position_embeds", "position_ids", "r", "raise", "return", "return_dict", "same", "sdpa", "self", "shape", "size", "specify", "tgt_len", "the", "time", "to", "token_type_embeds", "token_type_ids", "torch", "training", "tuple", "unsqueeze", "use_cache", "use_return_dict", "v", "view", "warn_if_padding_and_no_attention_mask", "warning_once", "with", "wpe", "wte"], "gpt2/modeling_gpt2.py:GPT2LMHeadModel.__init__": ["False", "GenerationMixin", "Linear", "Model", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "lm_head", "n_embd", "nn", "post_init", "self", "super", "transformer", "vocab_size"], "gpt2/modeling_gpt2.py:GPT2LMHeadModel.forward": ["Cache", "CausalLMOutputWithCrossAttentions", "FloatTensor", "GenerationMixin", "LongTensor", "Model", "ModelPreTrainedModel", "None", "Tensor", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "cross_attentions", "def", "else", "encoder_attention_mask", "encoder_hidden_states", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "not", "output", "output_attentions", "output_hidden_states", "past_key_values", "position_ids", "r", "return", "return_dict", "self", "slice", "slice_indices", "token_type_ids", "torch", "transformer", "transformer_outputs", "tuple", "use_cache", "use_return_dict", "vocab_size"], "gpt2/modeling_gpt2.py:GPT2DoubleHeadsModel.__init__": ["False", "GenerationMixin", "Linear", "Model", "ModelModel", "ModelPreTrainedModel", "ModelSequenceSummary", "__init__", "bias", "class", "config", "def", "lm_head", "multiple_choice_head", "n_embd", "nn", "num_labels", "post_init", "self", "super", "transformer", "vocab_size"], "gpt2/modeling_gpt2.py:GPT2DoubleHeadsModel.forward": ["Cache", "CrossEntropyLoss", "FloatTensor", "GenerationMixin", "LongTensor", "Model", "ModelOutput", "ModelPreTrainedModel", "None", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "contiguous", "def", "device", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "lm_head", "lm_logits", "lm_loss", "logits", "loss", "loss_fct", "mc_labels", "mc_logits", "mc_loss", "mc_token_ids", "multiple_choice_head", "not", "output", "output_attentions", "output_hidden_states", "past_key_values", "position_ids", "r", "return", "return_dict", "self", "shift_labels", "shift_logits", "size", "squeeze", "to", "token_type_ids", "torch", "transformer", "transformer_outputs", "tuple", "use_cache", "use_return_dict", "view"], "gpt2/modeling_gpt2.py:GPT2ForSequenceClassification.__init__": ["False", "Linear", "Model", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "n_embd", "nn", "num_labels", "post_init", "score", "self", "super", "transformer"], "gpt2/modeling_gpt2.py:GPT2ForSequenceClassification.forward": ["BCEWithLogitsLoss", "Cache", "Cannot", "CrossEntropyLoss", "FloatTensor", "LongTensor", "MSELoss", "Model", "ModelPreTrainedModel", "None", "Results", "SequenceClassifierOutputWithPast", "ValueError", "__class__", "__name__", "and", "arange", "argmax", "attention_mask", "attentions", "batch", "batch_size", "be", "bool", "class", "config", "conjunction", "def", "defined", "detect", "device", "dtype", "elif", "else", "f", "forward", "handle", "hidden_states", "if", "in", "input_ids", "inputs_embeds", "int", "int32", "is", "kwargs", "labels", "last_non_pad_token", "logger", "logits", "long", "loss", "loss_fct", "may", "multi_label_classification", "no", "non_pad_mask", "not", "num_labels", "or", "output", "output_attentions", "output_hidden_states", "pad_token_id", "padding", "past_key_values", "pooled_logits", "position_ids", "problem_type", "r", "raise", "regression", "return", "return_dict", "score", "self", "sequence_length", "shape", "single_label_classification", "sizes", "squeeze", "to", "token", "token_indices", "token_type_ids", "tokens", "torch", "transformer", "transformer_outputs", "tuple", "unexpected", "use_cache", "use_return_dict", "using", "view", "warning_once", "will", "with"], "gpt2/modeling_gpt2.py:GPT2ForTokenClassification.__init__": ["Dropout", "Linear", "Model", "ModelModel", "ModelPreTrainedModel", "None", "__init__", "and", "class", "classifier", "classifier_dropout", "config", "def", "dropout", "elif", "else", "hasattr", "hidden_dropout", "hidden_size", "if", "is", "nn", "not", "num_labels", "post_init", "self", "super", "transformer"], "gpt2/modeling_gpt2.py:GPT2ForTokenClassification.forward": ["Cache", "CrossEntropyLoss", "FloatTensor", "LongTensor", "Model", "ModelPreTrainedModel", "None", "TokenClassifierOutput", "attention_mask", "attentions", "bool", "class", "classifier", "config", "def", "device", "dropout", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "not", "num_labels", "output", "output_attentions", "output_hidden_states", "past_key_values", "position_ids", "r", "return", "return_dict", "self", "to", "token_type_ids", "torch", "transformer", "transformer_outputs", "tuple", "use_cache", "use_return_dict", "view"], "gpt2/modeling_gpt2.py:GPT2ForQuestionAnswering.__init__": ["Linear", "Model", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "config", "def", "hidden_size", "nn", "num_labels", "post_init", "qa_outputs", "self", "super", "transformer"], "gpt2/modeling_gpt2.py:GPT2ForQuestionAnswering.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "Model", "ModelPreTrainedModel", "None", "QuestionAnsweringModelOutput", "and", "attention_mask", "attentions", "bool", "clamp", "class", "config", "contiguous", "def", "device", "dim", "else", "end_logits", "end_loss", "end_positions", "forward", "hidden_states", "if", "ignore_index", "ignored_index", "input_ids", "inputs_embeds", "is", "kwargs", "len", "logits", "loss", "loss_fct", "not", "output", "output_attentions", "output_hidden_states", "outputs", "position_ids", "qa_outputs", "r", "return", "return_dict", "self", "sequence_output", "size", "split", "squeeze", "start_logits", "start_loss", "start_positions", "to", "token_type_ids", "torch", "total_loss", "transformer", "tuple", "use_return_dict"], "gpt_bigcode/modeling_gpt_bigcode.py:upcast_masked_softmax": ["Model_masked_softmax", "Tensor", "def", "dim", "dtype", "float", "functional", "input_dtype", "mask", "mask_value", "nn", "return", "scale", "softmax", "softmax_dtype", "to", "torch", "where", "x"], "gpt_bigcode/modeling_gpt_bigcode.py:upcast_softmax": ["Model_softmax", "Tensor", "def", "dim", "dtype", "float", "functional", "input_dtype", "nn", "return", "scale", "softmax", "softmax_dtype", "to", "torch", "x"], "gpt_bigcode/modeling_gpt_bigcode.py:masked_softmax": ["Model_softmax", "Tensor", "def", "dim", "functional", "mask", "mask_value", "nn", "return", "softmax", "torch", "where", "x"], "gpt_bigcode/modeling_gpt_bigcode.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "gpt_bigcode/modeling_gpt_bigcode.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value", "value_states"], "gpt_bigcode/modeling_gpt_bigcode.py:GPTBigCodeAttention.__init__": ["Attention", "Dropout", "False", "Linear", "Model", "Module", "Multi", "None", "NotImplementedError", "Query", "True", "ValueError", "__init__", "and", "attention_softmax_in_fp32", "attn_dropout", "attn_pdrop", "be", "by", "c_attn", "c_proj", "class", "config", "cross_attention", "def", "divisible", "else", "embed_dim", "f", "for", "got", "head_dim", "hidden_size", "if", "is_causal", "is_cross_attention", "kv_dim", "kv_heads", "layer_idx", "mask_value", "multi_query", "must", "nn", "not", "num_attention_heads", "num_heads", "num_key_value_groups", "q_attn", "raise", "resid_dropout", "resid_pdrop", "scale_attention_softmax_in_fp32", "scale_attn_weights", "scaling", "self", "split_size", "super", "supported"], "gpt_bigcode/modeling_gpt_bigcode.py:GPTBigCodeAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "EncoderDecoderCache", "False", "If", "Model", "Module", "None", "Please", "Tensor", "True", "ValueError", "_attn_implementation", "and", "as", "attention", "attention_interface", "attention_mask", "attn_dropout", "attn_output", "attn_weights", "be", "bool", "c_attn", "c_proj", "cache_position", "class", "config", "contiguous", "cross", "cross_attention_cache", "curr_past_key_values", "def", "defined", "dim", "dropout", "eager", "eager_attention_forward", "else", "embed_dim", "encoder_attention_mask", "encoder_hidden_states", "forward", "get", "hasattr", "have", "head_dim", "hidden_states", "if", "input_shape", "instantiate", "is", "is_cross_attention", "is_updated", "isinstance", "key", "keys", "kv_dim", "kwargs", "layer_idx", "layer_past", "layers", "make", "multi_query", "nn", "not", "num_heads", "or", "output_attentions", "q_attn", "query", "raise", "reshape", "resid_dropout", "return", "scaling", "self", "self_attention_cache", "shape", "split", "sure", "the", "to", "torch", "training", "transpose", "tuple", "unsqueeze", "update", "use_cache", "used", "value", "values", "view", "weights", "with"], "gpt_bigcode/modeling_gpt_bigcode.py:GPTBigCodeMLP.__init__": ["ACT2FN", "Dropout", "Linear", "Model", "Module", "__init__", "act", "activation_function", "c_fc", "c_proj", "class", "config", "def", "dropout", "embed_dim", "hidden_size", "intermediate_size", "nn", "resid_pdrop", "self", "super"], "gpt_bigcode/modeling_gpt_bigcode.py:GPTBigCodeMLP.forward": ["FloatTensor", "Model", "Module", "None", "act", "c_fc", "c_proj", "class", "def", "dropout", "forward", "hidden_states", "nn", "return", "self", "torch", "tuple"], "gpt_bigcode/modeling_gpt_bigcode.py:GPTBigCodeBlock.__init__": ["Cross", "GradientCheckpointingLayer", "LayerNorm", "MQA", "Model", "ModelAttention", "ModelMLP", "None", "NotImplementedError", "True", "__init__", "add_cross_attention", "attention", "attn", "class", "config", "crossattention", "def", "else", "eps", "for", "hidden_size", "if", "implemented", "inner_dim", "is", "is_cross_attention", "layer_idx", "layer_norm_epsilon", "ln_1", "ln_2", "ln_cross_attn", "mlp", "multi_query", "n_inner", "nn", "not", "raise", "self", "super"], "gpt_bigcode/modeling_gpt_bigcode.py:GPTBigCodeBlock.forward": ["Cache", "False", "GradientCheckpointingLayer", "If", "Model", "None", "Tensor", "True", "ValueError", "add_cross_attention", "are", "attention", "attention_mask", "attn", "attn_output", "attn_outputs", "be", "bool", "by", "cache_position", "class", "config", "cross", "cross_attn_outputs", "crossattention", "def", "encoder_attention_mask", "encoder_hidden_states", "f", "feed_forward_hidden_states", "forward", "has", "hasattr", "hidden_states", "if", "instantiated", "is", "kwargs", "layer_past", "layers", "ln_1", "ln_2", "ln_cross_attn", "mlp", "not", "output_attentions", "outputs", "passed", "raise", "residual", "return", "self", "setting", "to", "torch", "tuple", "use_cache", "with"], "gpt_bigcode/modeling_gpt_bigcode.py:GPTBigCodePreTrainedModel._init_weights": ["Model", "ModelAttention", "ModelMLP", "ModelModel", "PreTrainedModel", "_init_weights", "bias", "bool", "c_proj", "class", "config", "copy_", "def", "dtype", "elif", "if", "init", "initializer_range", "isinstance", "math", "max_position_embeddings", "max_positions", "mean", "module", "n_layer", "normal_", "ones", "self", "sqrt", "std", "super", "torch", "tril", "weight"], "gpt_bigcode/modeling_gpt_bigcode.py:GPTBigCodeModel.__init__": ["Dropout", "Embedding", "False", "LayerNorm", "Model", "ModelBlock", "ModelPreTrainedModel", "ModuleList", "__init__", "bias", "bool", "class", "config", "def", "drop", "dtype", "embd_pdrop", "embed_dim", "eps", "for", "gradient_checkpointing", "h", "hidden_size", "i", "in", "layer_idx", "layer_norm_epsilon", "ln_f", "max_position_embeddings", "max_positions", "multi_query", "nn", "num_hidden_layers", "ones", "persistent", "post_init", "range", "register_buffer", "self", "super", "torch", "tril", "vocab_size", "wpe", "wte"], "gpt_bigcode/modeling_gpt_bigcode.py:GPTBigCodeModel.get_input_embeddings": ["Model", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "return", "self", "wte"], "gpt_bigcode/modeling_gpt_bigcode.py:GPTBigCodeModel.set_input_embeddings": ["Model", "ModelPreTrainedModel", "class", "def", "new_embeddings", "self", "set_input_embeddings", "wte"], "gpt_bigcode/modeling_gpt_bigcode.py:GPTBigCodeModel.forward": ["BaseModelOutputWithPastAndCrossAttentions", "Cache", "DynamicCache", "Model", "ModelPreTrainedModel", "None", "Tensor", "ValueError", "You", "_attn_implementation", "add_cross_attention", "all_cross_attentions", "all_hidden_states", "all_self_attentions", "and", "arange", "assert", "attention_mask", "attentions", "batch_size", "be", "block", "bool", "cache_position", "causal_mask", "class", "config", "create_causal_mask", "cross_attentions", "def", "defined", "device", "dim", "drop", "either", "elif", "else", "encoder_attention_mask", "encoder_hidden_states", "enumerate", "exactly", "flash_attention_2", "for", "forward", "get_seq_length", "h", "has", "have", "hidden_states", "i", "if", "in", "input_embeds", "input_ids", "input_shape", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layer_past", "ln_f", "multi_query", "must", "not", "of", "one", "or", "output_attentions", "output_hidden_states", "output_shape", "outputs", "past_key_values", "past_seen_tokens", "position_embeds", "position_ids", "r", "raise", "return", "return_dict", "self", "shape", "size", "specify", "to", "token_type_embeds", "token_type_ids", "torch", "tuple", "unsqueeze", "use_cache", "use_return_dict", "view", "wpe", "wte"], "gpt_bigcode/modeling_gpt_bigcode.py:GPTBigCodeForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "Model", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "lm_head", "n_embd", "nn", "post_init", "self", "super", "transformer", "vocab_size"], "gpt_bigcode/modeling_gpt_bigcode.py:GPTBigCodeForCausalLM.forward": ["Cache", "CausalLMOutputWithCrossAttentions", "GenerationMixin", "Model", "ModelPreTrainedModel", "None", "Tensor", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "cross_attentions", "def", "else", "encoder_attention_mask", "encoder_hidden_states", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "not", "output", "output_attentions", "output_hidden_states", "past_key_values", "position_ids", "r", "return", "return_dict", "self", "slice", "slice_indices", "token_type_ids", "torch", "transformer", "transformer_outputs", "tuple", "use_cache", "use_return_dict", "vocab_size"], "gpt_bigcode/modeling_gpt_bigcode.py:GPTBigCodeForSequenceClassification.__init__": ["False", "Linear", "Model", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "n_embd", "nn", "num_labels", "post_init", "score", "self", "super", "transformer"], "gpt_bigcode/modeling_gpt_bigcode.py:GPTBigCodeForSequenceClassification.forward": ["BCEWithLogitsLoss", "Cache", "Cannot", "CrossEntropyLoss", "MSELoss", "Model", "ModelPreTrainedModel", "None", "Results", "SequenceClassifierOutputWithPast", "Tensor", "ValueError", "__class__", "__name__", "and", "arange", "argmax", "attention_mask", "attentions", "batch", "batch_size", "be", "bool", "class", "config", "conjunction", "def", "defined", "detect", "device", "dtype", "elif", "else", "f", "forward", "handle", "hidden_states", "if", "in", "input_ids", "inputs_embeds", "int", "int32", "is", "kwargs", "labels", "last_non_pad_token", "logger", "logits", "long", "loss", "loss_fct", "may", "multi_label_classification", "no", "non_pad_mask", "not", "num_labels", "or", "output", "output_attentions", "output_hidden_states", "pad_token_id", "padding", "past_key_values", "pooled_logits", "position_ids", "problem_type", "r", "raise", "regression", "return", "return_dict", "score", "self", "sequence_length", "shape", "single_label_classification", "sizes", "squeeze", "to", "token", "token_indices", "token_type_ids", "tokens", "torch", "transformer", "transformer_outputs", "tuple", "unexpected", "use_cache", "use_return_dict", "using", "view", "warning_once", "will", "with"], "gpt_bigcode/modeling_gpt_bigcode.py:GPTBigCodeForTokenClassification.__init__": ["Dropout", "Linear", "Model", "ModelModel", "ModelPreTrainedModel", "None", "__init__", "and", "class", "classifier", "classifier_dropout", "config", "def", "dropout", "elif", "else", "hasattr", "hidden_dropout", "hidden_size", "if", "is", "nn", "not", "num_labels", "post_init", "self", "super", "transformer"], "gpt_bigcode/modeling_gpt_bigcode.py:GPTBigCodeForTokenClassification.forward": ["Cache", "CrossEntropyLoss", "Model", "ModelPreTrainedModel", "None", "Tensor", "TokenClassifierOutput", "attention_mask", "attentions", "bool", "class", "classifier", "config", "def", "device", "dropout", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "not", "num_labels", "output", "output_attentions", "output_hidden_states", "past_key_values", "position_ids", "r", "return", "return_dict", "self", "to", "token_type_ids", "torch", "transformer", "transformer_outputs", "tuple", "use_cache", "use_return_dict", "view"], "gpt_neo/modeling_gpt_neo.py:GPTNeoSelfAttention.__init__": ["Dropout", "False", "Linear", "Model", "Module", "None", "True", "ValueError", "__init__", "and", "attention_dropout", "attention_type", "attn_dropout", "be", "bias", "bitwise_xor", "bool", "by", "class", "config", "def", "divisible", "dtype", "embed_dim", "f", "float", "got", "head_dim", "hidden_size", "if", "is_causal", "k_proj", "layer_id", "local", "max_position_embeddings", "max_positions", "must", "nn", "num_heads", "ones", "out_proj", "persistent", "q_proj", "raise", "register_buffer", "resid_dropout", "self", "super", "torch", "tril", "v_proj", "view", "window_size"], "gpt_neo/modeling_gpt_neo.py:GPTNeoSelfAttention._split_heads": ["Model", "Module", "_split_heads", "attn_head_size", "class", "def", "new_shape", "nn", "num_heads", "permute", "return", "self", "size", "tensor", "view"], "gpt_neo/modeling_gpt_neo.py:GPTNeoSelfAttention._merge_heads": ["Model", "Module", "_merge_heads", "attn_head_size", "class", "contiguous", "def", "new_shape", "nn", "num_heads", "permute", "return", "self", "size", "tensor", "view"], "gpt_neo/modeling_gpt_neo.py:GPTNeoSelfAttention._attn": ["Model", "Module", "None", "_attn", "attention_mask", "attn_dropout", "attn_output", "attn_weights", "bias", "causal_mask", "class", "def", "device", "dim", "dtype", "finfo", "float32", "functional", "if", "is", "key", "key_length", "mask_value", "matmul", "min", "nn", "not", "query", "query_length", "return", "self", "shape", "size", "softmax", "tensor", "to", "torch", "transpose", "value", "where"], "gpt_neo/modeling_gpt_neo.py:GPTNeoSelfAttention.forward": ["False", "Model", "Module", "None", "_attn", "_merge_heads", "_split_heads", "attention_mask", "attn_output", "attn_weights", "cache_kwargs", "cache_position", "class", "def", "forward", "head_dim", "hidden_states", "if", "is", "k_proj", "key", "layer_id", "layer_past", "nn", "not", "num_heads", "out_proj", "output_attentions", "q_proj", "query", "resid_dropout", "return", "self", "update", "use_cache", "v_proj", "value"], "gpt_neo/modeling_gpt_neo.py:GPTNeoFlashAttention2.__init__": ["Model", "ModelSelfAttention", "__init__", "_flash_attn_uses_top_left_mask", "args", "class", "def", "flash_attn_supports_top_left_mask", "kwargs", "self", "super"], "gpt_neo/modeling_gpt_neo.py:GPTNeoFlashAttention2.forward": ["False", "Model", "ModelSelfAttention", "None", "The", "We", "_", "_flash_attention_forward", "_flash_attn_uses_top_left_mask", "_is_quantized", "_split_heads", "attention_dropout", "attention_mask", "attn_dropout", "attn_output", "attn_weights_reshaped", "back", "be", "bsz", "cache_kwargs", "cache_position", "cast", "casted", "class", "config", "cpu", "def", "device", "device_type", "dropout", "dtype", "elif", "else", "embedding", "f", "fact", "float32", "forward", "get_autocast_dtype", "get_autocast_gpu_dtype", "hasattr", "have", "head_dim", "hidden", "hidden_states", "if", "in", "input", "is", "is_autocast_enabled", "is_causal", "k_proj", "key", "layer", "layer_id", "layer_past", "layers", "logger", "might", "mps", "norm", "not", "num_heads", "or", "out_proj", "output_attentions", "q_proj", "query", "query_length", "related", "reshape", "resid_dropout", "return", "seems", "self", "shape", "silently", "size", "softmax_scale", "states", "target_dtype", "tgt_len", "the", "this", "to", "torch", "training", "transpose", "type", "upcasted", "update", "use_cache", "use_top_left_mask", "v_proj", "value", "view", "warning_once", "weight", "will", "you"], "gpt_neo/modeling_gpt_neo.py:GPTNeoAttention.__init__": ["Model", "Model_ATTENTION_CLASSES", "Module", "NotImplementedError", "Only", "Select", "__init__", "_attn_implementation", "and", "attention", "attention_layers", "attention_type", "attn", "but", "class", "config", "def", "else", "exist", "f", "from", "global", "got", "if", "in", "layer", "layer_id", "local", "nn", "only", "raise", "self", "super", "types"], "gpt_neo/modeling_gpt_neo.py:GPTNeoAttention.forward": ["False", "Model", "Module", "None", "attention", "attention_mask", "cache_position", "class", "def", "forward", "hidden_states", "layer_past", "nn", "output_attentions", "return", "self", "use_cache"], "gpt_neo/modeling_gpt_neo.py:GPTNeoMLP.__init__": ["ACT2FN", "Dropout", "Linear", "Model", "Module", "__init__", "act", "activation_function", "c_fc", "c_proj", "class", "config", "def", "dropout", "embed_dim", "float", "hidden_size", "intermediate_size", "nn", "resid_dropout", "self", "super"], "gpt_neo/modeling_gpt_neo.py:GPTNeoMLP.forward": ["Model", "Module", "act", "c_fc", "c_proj", "class", "def", "dropout", "forward", "hidden_states", "nn", "return", "self"], "gpt_neo/modeling_gpt_neo.py:GPTNeoBlock.__init__": ["GradientCheckpointingLayer", "LayerNorm", "Model", "ModelAttention", "ModelMLP", "None", "__init__", "attn", "class", "config", "def", "else", "eps", "hidden_size", "if", "inner_dim", "intermediate_size", "is", "layer_id", "layer_norm_epsilon", "ln_1", "ln_2", "mlp", "nn", "not", "self", "super"], "gpt_neo/modeling_gpt_neo.py:GPTNeoBlock.forward": ["False", "GradientCheckpointingLayer", "Model", "None", "attention_mask", "attn", "attn_output", "attn_weights", "cache_position", "class", "def", "feed_forward_hidden_states", "forward", "hidden_states", "layer_past", "ln_1", "ln_2", "mlp", "output_attentions", "residual", "return", "self", "use_cache"], "gpt_neo/modeling_gpt_neo.py:GPTNeoPreTrainedModel._init_weights": ["Model", "ModelSelfAttention", "PreTrainedModel", "_init_weights", "attention_type", "bias", "bitwise_xor", "bool", "class", "config", "copy_", "def", "dtype", "if", "init", "isinstance", "local", "max_position_embeddings", "max_positions", "module", "ones", "self", "super", "torch", "tril", "view", "window_size"], "gpt_neo/modeling_gpt_neo.py:GPTNeoModel.__init__": ["Dropout", "Embedding", "False", "LayerNorm", "Model", "ModelBlock", "ModelPreTrainedModel", "ModuleList", "__init__", "class", "config", "def", "drop", "embed_dim", "embed_dropout", "eps", "float", "for", "gradient_checkpointing", "h", "hidden_size", "i", "in", "layer_id", "layer_norm_epsilon", "ln_f", "max_position_embeddings", "nn", "num_layers", "post_init", "range", "self", "super", "vocab_size", "wpe", "wte"], "gpt_neo/modeling_gpt_neo.py:GPTNeoModel.get_input_embeddings": ["Model", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "return", "self", "wte"], "gpt_neo/modeling_gpt_neo.py:GPTNeoModel.set_input_embeddings": ["Model", "ModelPreTrainedModel", "class", "def", "new_embeddings", "self", "set_input_embeddings", "wte"], "gpt_neo/modeling_gpt_neo.py:GPTNeoModel.forward": ["BaseModelOutputWithPast", "BaseModelOutputWithPastAndCrossAttentions", "Cache", "DynamicCache", "False", "LongTensor", "Model", "ModelPreTrainedModel", "None", "Setting", "Tensor", "True", "ValueError", "You", "_update_causal_mask", "all_hidden_states", "all_self_attentions", "and", "arange", "attention_mask", "attentions", "block", "bool", "cache_position", "causal_mask", "checkpointing", "class", "config", "def", "device", "drop", "else", "enumerate", "exactly", "for", "forward", "get_seq_length", "gradient", "gradient_checkpointing", "h", "hidden_states", "i", "if", "in", "incompatible", "input_ids", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layer_past", "ln_f", "logger", "must", "not", "of", "one", "or", "output_attentions", "output_hidden_states", "output_shape", "outputs", "past_key_values", "past_seen_tokens", "position_embeds", "position_ids", "r", "raise", "return", "return_dict", "self", "seq_length", "shape", "size", "specify", "token_type_embeds", "token_type_ids", "torch", "training", "tuple", "unsqueeze", "use_cache", "use_return_dict", "v", "view", "warning_once", "with", "wpe", "wte"], "gpt_neo/modeling_gpt_neo.py:GPTNeoModel._update_causal_mask": ["AttentionMaskConverter", "BlockMask", "Cache", "False", "Model", "ModelPreTrainedModel", "None", "Tensor", "Union", "_attn_implementation", "_ignore_causal_mask_sdpa", "_prepare_4d_causal_attention_mask_with_cache_position", "_unmask_unattended", "_update_causal_mask", "and", "any", "attention_mask", "batch_size", "bool", "cache_position", "causal_mask", "class", "config", "cuda", "def", "device", "dtype", "else", "finfo", "flash_attention_2", "flex_attention", "get_max_cache_shape", "get_seq_length", "if", "in", "input_tensor", "inputs_embeds", "is", "is_compileable", "is_training", "isinstance", "make_flex_block_causal_mask", "min", "min_dtype", "not", "npu", "output_attentions", "past_key_values", "past_key_values_length", "past_seen_tokens", "return", "sdpa", "self", "sequence_length", "shape", "target_length", "torch", "training", "type", "using_compilable_cache", "xpu"], "gpt_neo/modeling_gpt_neo.py:GPTNeoModel._prepare_4d_causal_attention_mask_with_cache_position": ["Model", "ModelPreTrainedModel", "None", "Tensor", "_prepare_4d_causal_attention_mask_with_cache_position", "and", "arange", "attention_mask", "batch_size", "cache_position", "causal_mask", "class", "clone", "def", "device", "diagonal", "dim", "dtype", "else", "expand", "fill_value", "finfo", "full", "if", "int", "is", "kwargs", "mask_length", "masked_fill", "min", "min_dtype", "not", "padding_mask", "reshape", "return", "sequence_length", "shape", "target_length", "to", "torch", "triu"], "gpt_neo/modeling_gpt_neo.py:GPTNeoForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "Model", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "nn", "post_init", "self", "super", "transformer", "vocab_size"], "gpt_neo/modeling_gpt_neo.py:GPTNeoForCausalLM.forward": ["Cache", "CausalLMOutputWithCrossAttentions", "CausalLMOutputWithPast", "GenerationMixin", "LongTensor", "Model", "ModelPreTrainedModel", "None", "Tensor", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "not", "output", "output_attentions", "output_hidden_states", "past_key_values", "position_ids", "r", "return", "return_dict", "self", "slice", "slice_indices", "token_type_ids", "torch", "transformer", "transformer_outputs", "tuple", "use_cache", "use_return_dict", "vocab_size"], "gpt_neo/modeling_gpt_neo.py:GPTNeoForSequenceClassification.__init__": ["False", "Linear", "Model", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "nn", "num_labels", "post_init", "score", "self", "super", "transformer"], "gpt_neo/modeling_gpt_neo.py:GPTNeoForSequenceClassification.forward": ["BCEWithLogitsLoss", "Cache", "Cannot", "CrossEntropyLoss", "MSELoss", "Model", "ModelPreTrainedModel", "None", "Results", "SequenceClassifierOutputWithPast", "Tensor", "ValueError", "__class__", "__name__", "and", "arange", "argmax", "attention_mask", "attentions", "batch", "batch_size", "be", "bool", "class", "config", "conjunction", "def", "defined", "detect", "device", "dtype", "elif", "else", "f", "forward", "handle", "hidden_states", "if", "in", "input_ids", "inputs_embeds", "int", "int32", "is", "kwargs", "labels", "last_non_pad_token", "logger", "logits", "long", "loss", "loss_fct", "may", "multi_label_classification", "no", "non_pad_mask", "not", "num_labels", "or", "output", "output_attentions", "output_hidden_states", "pad_token_id", "padding", "past_key_values", "pooled_logits", "position_ids", "problem_type", "r", "raise", "regression", "return", "return_dict", "score", "self", "sequence_length", "shape", "single_label_classification", "sizes", "squeeze", "to", "token", "token_indices", "token_type_ids", "tokens", "torch", "transformer", "transformer_outputs", "tuple", "unexpected", "use_cache", "use_return_dict", "using", "view", "warning_once", "will", "with"], "gpt_neo/modeling_gpt_neo.py:GPTNeoForTokenClassification.__init__": ["Dropout", "Linear", "Model", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "classifier", "classifier_dropout", "config", "def", "dropout", "hidden_size", "nn", "num_labels", "post_init", "self", "super", "transformer"], "gpt_neo/modeling_gpt_neo.py:GPTNeoForTokenClassification.forward": ["Cache", "CrossEntropyLoss", "FloatTensor", "LongTensor", "Model", "ModelPreTrainedModel", "None", "TokenClassifierOutput", "attention_mask", "attentions", "bool", "class", "classifier", "config", "def", "device", "dropout", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "not", "num_labels", "output", "output_attentions", "output_hidden_states", "past_key_values", "position_ids", "r", "return", "return_dict", "self", "to", "token_type_ids", "torch", "transformer", "transformer_outputs", "tuple", "use_cache", "use_return_dict", "view"], "gpt_neo/modeling_gpt_neo.py:GPTNeoForQuestionAnswering.__init__": ["Linear", "Model", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "config", "def", "hidden_size", "nn", "num_labels", "post_init", "qa_outputs", "self", "super", "transformer"], "gpt_neo/modeling_gpt_neo.py:GPTNeoForQuestionAnswering.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "Model", "ModelPreTrainedModel", "None", "QuestionAnsweringModelOutput", "and", "attention_mask", "attentions", "bool", "clamp", "class", "config", "contiguous", "def", "dim", "else", "end_logits", "end_loss", "end_positions", "forward", "hidden_states", "if", "ignore_index", "ignored_index", "input_ids", "inputs_embeds", "is", "kwargs", "len", "logits", "loss", "loss_fct", "not", "output", "output_attentions", "output_hidden_states", "outputs", "position_ids", "qa_outputs", "r", "return", "return_dict", "self", "sequence_output", "size", "split", "squeeze", "start_logits", "start_loss", "start_positions", "token_type_ids", "torch", "total_loss", "transformer", "tuple", "use_return_dict"], "gpt_neox/modeling_gpt_neox.py:GPTNeoXMLP.__init__": ["ACT2FN", "Linear", "Model", "Module", "__init__", "act", "class", "config", "def", "dense_4h_to_h", "dense_h_to_4h", "hidden_act", "hidden_size", "intermediate_size", "nn", "self", "super"], "gpt_neox/modeling_gpt_neox.py:GPTNeoXMLP.forward": ["Model", "Module", "act", "class", "def", "dense_4h_to_h", "dense_h_to_4h", "forward", "hidden_states", "nn", "return", "self"], "gpt_neox/modeling_gpt_neox.py:GPTNeoXRotaryEmbedding.__init__": ["Callable", "False", "Model", "ModelConfig", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "gpt_neox/modeling_gpt_neox.py:GPTNeoXRotaryEmbedding.compute_default_rope_parameters": ["Model", "ModelConfig", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "get", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "partial_rotary_factor", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "gpt_neox/modeling_gpt_neox.py:GPTNeoXRotaryEmbedding.forward": ["False", "Model", "Module", "None", "and", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "gpt_neox/modeling_gpt_neox.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "gpt_neox/modeling_gpt_neox.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cat", "cos", "def", "dim", "k", "k_embed", "k_pass", "k_rot", "q", "q_embed", "q_pass", "q_rot", "return", "rotary_dim", "rotate_half", "shape", "sin", "torch", "unsqueeze", "unsqueeze_dim"], "gpt_neox/modeling_gpt_neox.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value"], "gpt_neox/modeling_gpt_neox.py:GPTNeoXAttention.__init__": ["Linear", "Model", "Module", "None", "True", "__init__", "attention_bias", "attention_dropout", "bias", "class", "config", "def", "dense", "get", "head_size", "hidden_size", "int", "is_causal", "layer_idx", "nn", "num_attention_heads", "partial_rotary_factor", "query_key_value", "rope_parameters", "rotary_ndims", "scaling", "self", "super"], "gpt_neox/modeling_gpt_neox.py:GPTNeoXAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "FlashAttentionKwargs", "FloatTensor", "LongTensor", "Model", "Module", "None", "Tensor", "Unpack", "_attn_implementation", "apply_rotary_pos_emb", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "cache_kwargs", "cache_position", "chunk", "class", "config", "contiguous", "cos", "def", "dense", "dim", "dropout", "eager", "eager_attention_forward", "else", "forward", "head_size", "hidden_shape", "hidden_states", "if", "input_shape", "is", "key_states", "kwargs", "layer_idx", "layer_past", "nn", "not", "partial_rotation_size", "position_embeddings", "qkv", "query_key_value", "query_states", "reshape", "return", "rotary_ndims", "scaling", "self", "shape", "sin", "torch", "training", "transpose", "tuple", "update", "value_states", "view"], "gpt_neox/modeling_gpt_neox.py:GPTNeoXLayer.__init__": ["Dropout", "GradientCheckpointingLayer", "LayerNorm", "Model", "ModelAttention", "ModelMLP", "__init__", "attention", "class", "config", "def", "eps", "hidden_dropout", "hidden_size", "input_layernorm", "layer_idx", "layer_norm_eps", "mlp", "nn", "post_attention_dropout", "post_attention_layernorm", "post_mlp_dropout", "self", "super", "use_parallel_residual"], "gpt_neox/modeling_gpt_neox.py:GPTNeoXLayer.forward": ["Cache", "False", "FlashAttentionKwargs", "FloatTensor", "GradientCheckpointingLayer", "LongTensor", "Model", "None", "Tensor", "Unpack", "attention", "attention_mask", "attn_output", "attn_weights", "bool", "cache_position", "class", "def", "else", "forward", "hidden_states", "if", "input_layernorm", "kwargs", "layer_past", "mlp", "mlp_output", "output_attentions", "outputs", "position_embeddings", "position_ids", "post_attention_dropout", "post_attention_layernorm", "post_mlp_dropout", "return", "self", "torch", "tuple", "use_cache", "use_parallel_residual"], "gpt_neox/modeling_gpt_neox.py:GPTNeoXRMSNorm.__init__": ["Model", "Module", "Parameter", "__init__", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "gpt_neox/modeling_gpt_neox.py:GPTNeoXRMSNorm.forward": ["Model", "Module", "True", "class", "def", "dtype", "float32", "forward", "hidden_states", "input_dtype", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "gpt_neox/modeling_gpt_neox.py:GPTNeoXRMSNorm.extra_repr": ["Model", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "variance_epsilon", "weight"], "gpt_neox/modeling_gpt_neox.py:GPTNeoXDecoderLayer.__init__": ["GradientCheckpointingLayer", "Model", "ModelAttention", "ModelConfig", "ModelMLP", "ModelRMSNorm", "__init__", "class", "config", "def", "eps", "hidden_size", "input_layernorm", "int", "layer_idx", "mlp", "post_attention_layernorm", "rms_norm_eps", "self", "self_attn", "super"], "gpt_neox/modeling_gpt_neox.py:GPTNeoXDecoderLayer.forward": ["Cache", "False", "GradientCheckpointingLayer", "LongTensor", "Model", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "attention_mask", "bool", "cache_position", "class", "def", "forward", "hidden_states", "input_layernorm", "kwargs", "mlp", "past_key_values", "position_embeddings", "position_ids", "post_attention_layernorm", "residual", "return", "self", "self_attn", "torch", "tuple", "use_cache"], "gpt_neox/modeling_gpt_neox.py:GPTNeoXModel.__init__": ["Dropout", "Embedding", "False", "LayerNorm", "Model", "ModelLayer", "ModelPreTrainedModel", "ModelRotaryEmbedding", "ModuleList", "__init__", "class", "config", "def", "emb_dropout", "embed_in", "eps", "final_layer_norm", "for", "gradient_checkpointing", "hidden_dropout", "hidden_size", "i", "in", "layer_norm_eps", "layers", "nn", "num_hidden_layers", "post_init", "range", "rotary_emb", "self", "super", "vocab_size"], "gpt_neox/modeling_gpt_neox.py:GPTNeoXModel.forward": ["BaseModelOutputWithPast", "Cache", "DynamicCache", "False", "FloatTensor", "LongTensor", "Model", "ModelPreTrainedModel", "None", "Setting", "TransformersKwargs", "True", "Unpack", "ValueError", "You", "all_attentions", "all_hidden_states", "and", "arange", "attention_mask", "attentions", "bool", "cache_position", "causal_mask", "checkpointing", "class", "config", "create_causal_mask", "def", "device", "else", "emb_dropout", "embed_in", "enumerate", "exactly", "final_layer_norm", "for", "forward", "get_seq_length", "gradient", "gradient_checkpointing", "hidden_states", "i", "if", "in", "incompatible", "input_embeds", "input_ids", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layer", "layer_past", "layers", "logger", "must", "not", "of", "one", "or", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "past_seen_tokens", "position_embeddings", "position_ids", "raise", "return", "rotary_emb", "self", "shape", "specify", "torch", "training", "unsqueeze", "use_cache", "warning_once", "with"], "gpt_neox/modeling_gpt_neox.py:GPTNeoXModel.get_input_embeddings": ["Model", "ModelPreTrainedModel", "class", "def", "embed_in", "get_input_embeddings", "return", "self"], "gpt_neox/modeling_gpt_neox.py:GPTNeoXModel.set_input_embeddings": ["Model", "ModelPreTrainedModel", "class", "def", "embed_in", "self", "set_input_embeddings", "value"], "gpt_neox/modeling_gpt_neox.py:GPTNeoXForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "Model", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "embed_out", "hidden_size", "nn", "post_init", "self", "super", "vocab_size"], "gpt_neox/modeling_gpt_neox.py:GPTNeoXForCausalLM.get_output_embeddings": ["GenerationMixin", "Model", "ModelPreTrainedModel", "class", "def", "embed_out", "get_output_embeddings", "return", "self"], "gpt_neox/modeling_gpt_neox.py:GPTNeoXForCausalLM.set_output_embeddings": ["GenerationMixin", "Model", "ModelPreTrainedModel", "class", "def", "embed_out", "new_embeddings", "self", "set_output_embeddings"], "gpt_neox/modeling_gpt_neox.py:GPTNeoXForCausalLM.forward": ["BaseModelOutputWithPast", "Cache", "CausalLMOutputWithPast", "FloatTensor", "GenerationMixin", "LongTensor", "Model", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "else", "embed_out", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "logits", "logits_to_keep", "loss", "loss_function", "not", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "position_ids", "r", "return", "self", "slice", "slice_indices", "torch", "tuple", "use_cache", "vocab_size"], "gpt_neox/modeling_gpt_neox.py:GPTNeoXForSequenceClassification.__init__": ["False", "Linear", "Model", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "nn", "num_labels", "post_init", "score", "self", "super"], "gpt_neox/modeling_gpt_neox.py:GPTNeoXForSequenceClassification.forward": ["BaseModelOutputWithPast", "Cache", "Cannot", "FloatTensor", "LongTensor", "Model", "ModelPreTrainedModel", "None", "Results", "SequenceClassifierOutputWithPast", "ValueError", "__class__", "__name__", "and", "arange", "argmax", "attention_mask", "attentions", "batch", "batch_size", "be", "bool", "class", "config", "conjunction", "def", "defined", "detect", "device", "dtype", "elif", "else", "f", "forward", "handle", "hidden_states", "if", "in", "input_ids", "inputs_embeds", "int32", "is", "kwargs", "labels", "last_hidden_state", "last_non_pad_token", "logger", "logits", "loss", "loss_function", "may", "no", "non_pad_mask", "not", "output_attentions", "output_hidden_states", "outputs", "pad_token_id", "padding", "past_key_values", "pooled_logits", "position_ids", "r", "raise", "return", "score", "self", "shape", "sizes", "to", "token", "token_indices", "tokens", "torch", "unexpected", "use_cache", "using", "warning_once", "will", "with"], "gpt_neox/modeling_gpt_neox.py:GPTNeoXForTokenClassification.__init__": ["Dropout", "Linear", "Model", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "classifier", "classifier_dropout", "config", "def", "dropout", "hidden_size", "nn", "num_labels", "post_init", "self", "super"], "gpt_neox/modeling_gpt_neox.py:GPTNeoXForTokenClassification.forward": ["BaseModelOutputWithPast", "Cache", "FloatTensor", "LongTensor", "Model", "ModelPreTrainedModel", "None", "TokenClassifierOutput", "attention_mask", "attentions", "bool", "class", "classifier", "config", "def", "dropout", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "last_hidden_state", "logits", "loss", "loss_function", "not", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "position_ids", "r", "return", "self", "token_type_ids", "torch", "use_cache"], "gpt_neox/modeling_gpt_neox.py:GPTNeoXForQuestionAnswering.__init__": ["Linear", "Model", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "config", "def", "hidden_size", "nn", "num_labels", "post_init", "qa_outputs", "self", "super"], "gpt_neox/modeling_gpt_neox.py:GPTNeoXForQuestionAnswering.forward": ["BaseModelOutputWithPast", "FloatTensor", "LongTensor", "Model", "ModelPreTrainedModel", "None", "QuestionAnsweringModelOutput", "and", "attention_mask", "attentions", "bool", "class", "contiguous", "def", "dim", "end_logits", "end_positions", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "last_hidden_state", "logits", "loss", "loss_function", "not", "output_attentions", "output_hidden_states", "outputs", "position_ids", "qa_outputs", "return", "self", "sequence_output", "split", "squeeze", "start_logits", "start_positions", "token_type_ids", "torch"], "gpt_neox_japanese/modeling_gpt_neox_japanese.py:GPTNeoXJapanesePreTrainedModel._init_weights": ["Model", "ModelAttention", "None", "PreTrainedModel", "_init_weights", "class", "def", "dense_bias", "if", "init", "is", "isinstance", "module", "not", "self", "super", "zeros_"], "gpt_neox_japanese/modeling_gpt_neox_japanese.py:GPTNeoXJapaneseRotaryEmbedding.__init__": ["Callable", "False", "Model", "ModelConfig", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "gpt_neox_japanese/modeling_gpt_neox_japanese.py:GPTNeoXJapaneseRotaryEmbedding.compute_default_rope_parameters": ["Model", "ModelConfig", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "gpt_neox_japanese/modeling_gpt_neox_japanese.py:GPTNeoXJapaneseRotaryEmbedding.forward": ["False", "Model", "Module", "None", "and", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "gpt_neox_japanese/modeling_gpt_neox_japanese.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "gpt_neox_japanese/modeling_gpt_neox_japanese.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cos", "def", "k", "k_embed", "q", "q_embed", "return", "rotate_half", "sin", "unsqueeze", "unsqueeze_dim"], "gpt_neox_japanese/modeling_gpt_neox_japanese.py:GPTNeoXJapaneseAttention.__init__": ["Dropout", "False", "Instantiating", "Linear", "Model", "Module", "None", "Parameter", "Please", "__class__", "__init__", "__name__", "a", "and", "attention_dropout", "bias", "caching", "call", "class", "config", "creating", "def", "dense", "dense_bias", "during", "else", "errors", "f", "forward", "get", "head_size", "hidden_size", "if", "int", "is", "layer_idx", "lead", "logger", "make", "math", "nn", "norm_factor", "not", "num_attention_heads", "partial_rotary_factor", "passing", "provide", "query_key_value", "recommended", "rope_parameters", "rotary_ndims", "self", "sqrt", "super", "sure", "the", "this", "to", "torch", "use_bias", "used", "warning_once", "when", "will", "without", "zeros"], "gpt_neox_japanese/modeling_gpt_neox_japanese.py:GPTNeoXJapaneseAttention.forward": ["Cache", "False", "FloatTensor", "LongTensor", "Model", "Module", "None", "Tensor", "_attn", "_merge_heads", "apply_rotary_pos_emb", "attention_mask", "attn_output", "attn_weights", "bool", "cache_kwargs", "cache_position", "cat", "class", "contiguous", "cos", "def", "dense", "dense_bias", "dim", "forward", "head_size", "hidden_states", "if", "is", "key", "key_pass", "key_rot", "layer_idx", "layer_past", "new_qkv_shape", "nn", "not", "num_attention_heads", "output_attentions", "partial_rotation_size", "permute", "position_embeddings", "position_ids", "qkv", "query", "query_key_value", "query_pass", "query_rot", "return", "rotary_ndims", "self", "sin", "size", "torch", "tuple", "update", "use_cache", "value", "view"], "gpt_neox_japanese/modeling_gpt_neox_japanese.py:GPTNeoXJapaneseAttention._split_heads": ["Model", "Module", "_split_heads", "attn_head_size", "class", "cls", "def", "new_shape", "nn", "num_attention_heads", "permute", "return", "size", "tensor", "view"], "gpt_neox_japanese/modeling_gpt_neox_japanese.py:GPTNeoXJapaneseAttention._merge_heads": ["Model", "Module", "_merge_heads", "attn_head_size", "class", "cls", "contiguous", "def", "nn", "num_attention_heads", "permute", "return", "size", "tensor", "view"], "gpt_neox_japanese/modeling_gpt_neox_japanese.py:GPTNeoXJapaneseAttention._attn": ["Model", "Module", "None", "_attn", "alpha", "attention_dropout", "attention_mask", "attention_scores", "attn_head_size", "attn_output", "attn_scores", "attn_weights", "baddbmm", "batch_size", "beta", "causal_mask", "class", "def", "device", "dim", "dtype", "functional", "if", "is", "key", "key_length", "matmul", "nn", "norm_factor", "not", "num_attention_heads", "query", "query_length", "return", "self", "shape", "size", "softmax", "to", "torch", "transpose", "value", "view", "zeros"], "gpt_neox_japanese/modeling_gpt_neox_japanese.py:bias_dropout_add": ["Model", "Model_dropout_add", "None", "Tensor", "bool", "def", "dropout", "float", "functional", "if", "is", "nn", "not", "out", "p", "prob", "residual", "return", "torch", "training", "x"], "gpt_neox_japanese/modeling_gpt_neox_japanese.py:GPTNeoXJapaneseMLP.__init__": ["ACT2FN", "False", "Linear", "Model", "Module", "__init__", "act", "bias", "class", "config", "def", "dense_4h_to_h", "dense_h_to_4h", "hidden_act", "hidden_size", "int", "intermediate_multiple_size", "intermediate_size", "nn", "self", "super"], "gpt_neox_japanese/modeling_gpt_neox_japanese.py:GPTNeoXJapaneseMLP.forward": ["Model", "Module", "act", "class", "def", "dense_4h_to_h", "dense_h_to_4h", "forward", "hidden_states", "intermediate", "nn", "output", "return", "self"], "gpt_neox_japanese/modeling_gpt_neox_japanese.py:GPTNeoXJapaneseLayer.__init__": ["LayerNorm", "Model", "ModelAttention", "ModelMLP", "Module", "__init__", "attention", "class", "config", "def", "eps", "hidden_dropout", "hidden_size", "input_layernorm", "layer_idx", "layer_norm_eps", "layer_number", "mlp", "nn", "num_hidden_layers", "post_attention_layernorm", "self", "super", "use_bias"], "gpt_neox_japanese/modeling_gpt_neox_japanese.py:GPTNeoXJapaneseLayer.forward": ["Cache", "False", "FloatTensor", "LongTensor", "Model", "Module", "None", "Tensor", "attention", "attention_mask", "attn_bias", "attn_output", "attn_weights", "bias", "bias_dropout_add", "bool", "cache_position", "class", "def", "else", "expand_as", "forward", "hidden_dropout", "hidden_states", "if", "input_layernorm", "is", "layer_past", "ln_out", "mlp", "mlp_output", "nn", "not", "output_attentions", "position_embeddings", "position_ids", "post_attention_layernorm", "prob", "residual", "return", "self", "torch", "training", "tuple", "use_cache"], "gpt_neox_japanese/modeling_gpt_neox_japanese.py:GPTNeoXJapaneseModel.__init__": ["Embedding", "LayerNorm", "Model", "ModelLayer", "ModelPreTrainedModel", "ModelRotaryEmbedding", "ModuleList", "__init__", "class", "config", "def", "embed_in", "eps", "final_layer_norm", "for", "hidden_size", "i", "in", "layer_norm_eps", "layer_number", "layers", "nn", "num_hidden_layers", "post_init", "range", "rotary_emb", "self", "super", "vocab_size"], "gpt_neox_japanese/modeling_gpt_neox_japanese.py:GPTNeoXJapaneseModel.get_input_embeddings": ["Model", "ModelPreTrainedModel", "class", "def", "embed_in", "get_input_embeddings", "return", "self"], "gpt_neox_japanese/modeling_gpt_neox_japanese.py:GPTNeoXJapaneseModel.set_input_embeddings": ["Model", "ModelPreTrainedModel", "class", "def", "embed_in", "self", "set_input_embeddings", "value"], "gpt_neox_japanese/modeling_gpt_neox_japanese.py:GPTNeoXJapaneseModel.forward": ["BaseModelOutputWithPast", "Cache", "DynamicCache", "FloatTensor", "LongTensor", "Model", "ModelPreTrainedModel", "None", "ValueError", "You", "_update_causal_mask", "all_attentions", "all_hidden_states", "and", "arange", "attention_mask", "attentions", "bool", "cache_position", "causal_mask", "class", "config", "def", "device", "else", "embed_in", "enumerate", "exactly", "final_layer_norm", "for", "forward", "get_seq_length", "hidden_states", "i", "if", "in", "input_ids", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layer", "layer_past", "layers", "must", "not", "of", "one", "or", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "past_seen_tokens", "position_embeddings", "position_ids", "r", "raise", "return", "return_dict", "rotary_emb", "self", "seq_length", "shape", "specify", "torch", "tuple", "unsqueeze", "use_cache", "use_return_dict", "v"], "gpt_neox_japanese/modeling_gpt_neox_japanese.py:GPTNeoXJapaneseModel._update_causal_mask": ["AttentionMaskConverter", "BlockMask", "Cache", "False", "Model", "ModelPreTrainedModel", "None", "Tensor", "Union", "_attn_implementation", "_ignore_causal_mask_sdpa", "_prepare_4d_causal_attention_mask_with_cache_position", "_unmask_unattended", "_update_causal_mask", "and", "any", "attention_mask", "batch_size", "bool", "cache_position", "causal_mask", "class", "config", "cuda", "def", "device", "dtype", "else", "finfo", "flash_attention_2", "flex_attention", "get_max_cache_shape", "get_seq_length", "if", "in", "input_tensor", "inputs_embeds", "is", "is_compileable", "is_training", "isinstance", "make_flex_block_causal_mask", "min", "min_dtype", "not", "npu", "output_attentions", "past_key_values", "past_key_values_length", "past_seen_tokens", "return", "sdpa", "self", "sequence_length", "shape", "target_length", "torch", "training", "type", "using_compilable_cache", "xpu"], "gpt_neox_japanese/modeling_gpt_neox_japanese.py:GPTNeoXJapaneseModel._prepare_4d_causal_attention_mask_with_cache_position": ["Model", "ModelPreTrainedModel", "None", "Tensor", "_prepare_4d_causal_attention_mask_with_cache_position", "and", "arange", "attention_mask", "batch_size", "cache_position", "causal_mask", "class", "clone", "def", "device", "diagonal", "dim", "dtype", "else", "expand", "fill_value", "finfo", "full", "if", "int", "is", "kwargs", "mask_length", "masked_fill", "min", "min_dtype", "not", "padding_mask", "reshape", "return", "sequence_length", "shape", "target_length", "to", "torch", "triu"], "gpt_neox_japanese/modeling_gpt_neox_japanese.py:GPTNeoXJapaneseForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "Model", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "embed_out", "hidden_size", "nn", "post_init", "self", "super", "vocab_size"], "gpt_neox_japanese/modeling_gpt_neox_japanese.py:GPTNeoXJapaneseForCausalLM.get_output_embeddings": ["GenerationMixin", "Model", "ModelPreTrainedModel", "class", "def", "embed_out", "get_output_embeddings", "return", "self"], "gpt_neox_japanese/modeling_gpt_neox_japanese.py:GPTNeoXJapaneseForCausalLM.set_output_embeddings": ["GenerationMixin", "Model", "ModelPreTrainedModel", "class", "def", "embed_out", "new_embeddings", "self", "set_output_embeddings"], "gpt_neox_japanese/modeling_gpt_neox_japanese.py:GPTNeoXJapaneseForCausalLM.forward": ["Cache", "CausalLMOutputWithPast", "FloatTensor", "GenerationMixin", "LongTensor", "Model", "ModelPreTrainedModel", "None", "Tensor", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "else", "embed_out", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "logits", "logits_to_keep", "loss", "loss_function", "not", "output", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "position_ids", "r", "return", "return_dict", "self", "slice", "slice_indices", "torch", "tuple", "use_cache", "use_return_dict", "vocab_size"], "gpt_oss/modeling_gpt_oss.py:GptOssRMSNorm.__init__": ["ModelRMSNorm", "Module", "Parameter", "__init__", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "gpt_oss/modeling_gpt_oss.py:GptOssRMSNorm.forward": ["ModelRMSNorm", "Module", "True", "class", "def", "dtype", "float32", "forward", "hidden_states", "input_dtype", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "gpt_oss/modeling_gpt_oss.py:GptOssRMSNorm.extra_repr": ["ModelRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "variance_epsilon", "weight"], "gpt_oss/modeling_gpt_oss.py:GptOssExperts.__init__": ["ModelExperts", "Module", "Parameter", "__init__", "alpha", "class", "config", "def", "down_proj", "down_proj_bias", "empty", "expert_dim", "gate_up_proj", "gate_up_proj_bias", "hidden_size", "intermediate_size", "limit", "nn", "num_experts", "num_local_experts", "self", "super", "torch", "zeros"], "gpt_oss/modeling_gpt_oss.py:GptOssExperts.forward": ["ModelExperts", "Module", "None", "Tensor", "alpha", "batch_size", "bmm", "clamp", "class", "continue", "cpu", "current_state", "def", "device", "dim", "down_proj", "down_proj_bias", "dtype", "else", "expert_hit", "expert_idx", "expert_mask", "for", "forward", "full_routing_weights", "functional", "gate", "gate_up", "gate_up_proj", "gate_up_proj_bias", "gated_output", "glu", "greater", "hidden_size", "hidden_states", "if", "in", "index_add_", "limit", "max", "min", "next_states", "nn", "no_grad", "nonzero", "num_classes", "num_experts", "num_tokens", "one_hot", "or", "out", "permute", "repeat", "reshape", "return", "router_indices", "routing_weights", "scatter_", "self", "shape", "sigmoid", "sum", "to", "token_idx", "top_k_pos", "torch", "training", "transpose", "type", "up", "view", "weighted_output", "where", "with", "zeros", "zeros_like"], "gpt_oss/modeling_gpt_oss.py:GptOssTopKRouter.__init__": ["ModelTopKRouter", "Module", "Parameter", "__init__", "bias", "class", "config", "def", "hidden_dim", "hidden_size", "nn", "num_experts", "num_experts_per_tok", "num_local_experts", "self", "super", "top_k", "torch", "weight", "zeros"], "gpt_oss/modeling_gpt_oss.py:GptOssTopKRouter.forward": ["F", "ModelTopKRouter", "Module", "bias", "class", "def", "dim", "dtype", "forward", "functional", "hidden_dim", "hidden_states", "linear", "nn", "reshape", "return", "router_indices", "router_logits", "router_scores", "router_top_value", "self", "softmax", "top_k", "topk", "torch", "weight"], "gpt_oss/modeling_gpt_oss.py:GptOssMLP.__init__": ["ModelExperts", "ModelMLP", "ModelTopKRouter", "Module", "__init__", "class", "config", "def", "experts", "nn", "router", "self", "super"], "gpt_oss/modeling_gpt_oss.py:GptOssMLP.forward": ["ModelMLP", "Module", "_", "class", "def", "experts", "forward", "hidden_states", "nn", "return", "routed_out", "router", "router_indices", "router_scores", "self"], "gpt_oss/modeling_gpt_oss.py:GptOssRotaryEmbedding.__init__": ["Callable", "False", "ModelConfig", "ModelRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "gpt_oss/modeling_gpt_oss.py:GptOssRotaryEmbedding.compute_default_rope_parameters": ["ModelConfig", "ModelRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "gpt_oss/modeling_gpt_oss.py:GptOssRotaryEmbedding.forward": ["False", "ModelRotaryEmbedding", "Module", "None", "and", "attention_scaling", "class", "cos", "cpu", "def", "device", "device_type", "dtype", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "transpose", "type", "with", "x"], "gpt_oss/modeling_gpt_oss.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "gpt_oss/modeling_gpt_oss.py:_apply_rotary_emb": ["Tensor", "_apply_rotary_emb", "cat", "chunk", "cos", "def", "dim", "first_", "first_half", "return", "second_", "second_half", "sin", "torch", "x"], "gpt_oss/modeling_gpt_oss.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "None", "_Model_rotary_emb", "cos", "def", "k", "k_embed", "position_ids", "q", "q_embed", "return", "sin", "unsqueeze", "unsqueeze_dim"], "gpt_oss/modeling_gpt_oss.py:eager_attention_forward": ["F", "Model_attention_forward", "Module", "None", "Tensor", "True", "attention_mask", "attn_output", "attn_weights", "cat", "causal_mask", "combined_logits", "contiguous", "def", "dim", "dropout", "dtype", "expand", "float", "functional", "if", "is", "keepdim", "key", "key_states", "kwargs", "matmul", "max", "module", "nn", "not", "num_key_value_groups", "p", "probs", "query", "repeat_kv", "reshape", "return", "scaling", "scores", "shape", "sinks", "softmax", "to", "torch", "training", "transpose", "value", "value_states", "values"], "gpt_oss/modeling_gpt_oss.py:GptOssAttention.__init__": ["Linear", "ModelAttention", "ModelConfig", "Module", "None", "Parameter", "True", "__init__", "attention_bias", "attention_dropout", "bias", "class", "config", "def", "else", "empty", "getattr", "hasattr", "head_dim", "hidden_size", "if", "int", "is_causal", "k_proj", "layer_idx", "layer_type", "layer_types", "nn", "num_attention_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "q_proj", "scaling", "self", "sinks", "sliding_attention", "sliding_window", "super", "torch", "v_proj"], "gpt_oss/modeling_gpt_oss.py:GptOssAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "LongTensor", "ModelAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_attn_implementation", "apply_rotary_pos_emb", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "cache_kwargs", "cache_position", "class", "config", "contiguous", "cos", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "head_dim", "hidden_shape", "hidden_states", "if", "input_shape", "is", "k_proj", "key_states", "kwargs", "layer_idx", "nn", "not", "o_proj", "past_key_values", "position_embeddings", "q_proj", "query_states", "reshape", "return", "s_aux", "scaling", "self", "shape", "sin", "sinks", "sliding_window", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "view"], "gpt_oss/modeling_gpt_oss.py:GptOssDecoderLayer.__init__": ["GradientCheckpointingLayer", "ModelAttention", "ModelConfig", "ModelDecoderLayer", "ModelMLP", "ModelRMSNorm", "__init__", "attention_type", "class", "config", "def", "eps", "hidden_size", "input_layernorm", "int", "layer_idx", "layer_types", "mlp", "post_attention_layernorm", "rms_norm_eps", "self", "self_attn", "super"], "gpt_oss/modeling_gpt_oss.py:GptOssDecoderLayer.forward": ["Cache", "False", "GradientCheckpointingLayer", "LongTensor", "ModelDecoderLayer", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "attention_mask", "bool", "cache_position", "class", "def", "forward", "hidden_states", "input_layernorm", "kwargs", "mlp", "past_key_values", "position_embeddings", "position_ids", "post_attention_layernorm", "residual", "return", "self", "self_attn", "torch", "tuple", "use_cache"], "gpt_oss/modeling_gpt_oss.py:GptOssPreTrainedModel._init_weights": ["ModelAttention", "ModelExperts", "ModelPreTrainedModel", "ModelTopKRouter", "PreTrainedModel", "_init_weights", "bias", "class", "config", "def", "down_proj", "down_proj_bias", "elif", "gate_up_proj", "gate_up_proj_bias", "if", "init", "initializer_range", "isinstance", "mean", "module", "normal_", "self", "sinks", "std", "super", "weight", "zeros_"], "gpt_oss/modeling_gpt_oss.py:GptOssModel.__init__": ["Embedding", "False", "ModelConfig", "ModelDecoderLayer", "ModelModel", "ModelPreTrainedModel", "ModelRMSNorm", "ModelRotaryEmbedding", "ModuleList", "__init__", "class", "config", "def", "embed_tokens", "eps", "for", "gradient_checkpointing", "hidden_size", "in", "layer_idx", "layers", "nn", "norm", "num_hidden_layers", "pad_token_id", "padding_idx", "post_init", "range", "rms_norm_eps", "rotary_emb", "self", "super", "vocab_size"], "gpt_oss/modeling_gpt_oss.py:GptOssModel.forward": ["Cache", "DynamicCache", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "MoeModelOutputWithPast", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "and", "arange", "attention_mask", "attention_type", "bool", "cache_position", "causal_mask_mapping", "class", "config", "create_causal_mask", "create_sliding_window_causal_mask", "decoder_layer", "def", "device", "dict", "else", "embed_tokens", "exactly", "for", "forward", "full_attention", "get_seq_length", "hidden_states", "if", "in", "input_embeds", "input_ids", "inputs_embeds", "is", "isinstance", "kwargs", "last_hidden_state", "layers", "mask_kwargs", "must", "norm", "not", "of", "one", "or", "past_key_values", "past_seen_tokens", "position_embeddings", "position_ids", "raise", "return", "rotary_emb", "self", "shape", "sliding_attention", "specify", "torch", "unsqueeze", "use_cache"], "gpt_oss/modeling_gpt_oss.py:load_balancing_loss_func": ["Model_balancing_loss_func", "None", "Tensor", "_", "attention_mask", "batch_size", "cat", "compute_device", "concatenated_gate_logits", "def", "device", "dim", "else", "expand", "expert_attention_mask", "expert_mask", "float", "for", "functional", "gate_logits", "if", "in", "int", "is", "isinstance", "layer_gate", "mean", "nn", "not", "num_experts", "num_hidden_layers", "one_hot", "or", "overall_loss", "r", "reshape", "return", "router_per_expert_attention_mask", "router_prob_per_expert", "routing_weights", "selected_experts", "sequence_length", "shape", "softmax", "sum", "to", "tokens_per_expert", "top_k", "topk", "torch", "tuple", "unsqueeze"], "gpt_oss/modeling_gpt_oss.py:GptOssForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "ModelForCausalLM", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "num_experts", "num_experts_per_tok", "num_local_experts", "post_init", "router_aux_loss_coef", "self", "super", "vocab_size"], "gpt_oss/modeling_gpt_oss.py:GptOssForCausalLM.forward": ["Cache", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForCausalLM", "ModelPreTrainedModel", "MoeCausalLMOutputWithPast", "MoeModelOutputWithPast", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "aux_loss", "bool", "cache_position", "class", "config", "def", "device", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "load_balancing_loss_func", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "num_experts", "num_experts_per_tok", "output_router_logits", "outputs", "past_key_values", "position_ids", "r", "return", "router_aux_loss_coef", "router_logits", "self", "slice", "slice_indices", "to", "torch", "use_cache", "vocab_size"], "gptj/modeling_gptj.py:create_sinusoidal_positions": ["Model_sinusoidal_positions", "Tensor", "arange", "cat", "cos", "def", "dim", "dtype", "einsum", "float", "i", "int", "int64", "inv_freq", "j", "num_pos", "return", "sin", "sinusoid_inp", "torch"], "gptj/modeling_gptj.py:get_embed_positions": ["Model_embed_positions", "def", "device", "embed_positions", "position_ids", "repeat", "return", "shape", "to"], "gptj/modeling_gptj.py:rotate_every_two": ["Model_every_two", "Tensor", "def", "dim", "flatten", "return", "stack", "torch", "x", "x1", "x2"], "gptj/modeling_gptj.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "None", "Tensor", "cos", "def", "repeat_interleave", "return", "rotate_every_two", "sin", "tensor", "torch"], "gptj/modeling_gptj.py:GPTJAttention.__init__": ["Dropout", "False", "Instantiating", "Linear", "Model", "Module", "None", "Please", "True", "ValueError", "__class__", "__init__", "__name__", "a", "and", "attn_dropout", "attn_pdrop", "be", "bias", "by", "caching", "call", "class", "config", "create_sinusoidal_positions", "creating", "def", "divisible", "during", "embed_dim", "embed_positions", "errors", "f", "forward", "got", "head_dim", "hidden_size", "if", "is", "is_causal", "k_proj", "layer_idx", "lead", "logger", "make", "math", "max_position_embeddings", "max_positions", "must", "nn", "not", "num_attention_heads", "or", "out_proj", "passing", "persistent", "pos_embd_dim", "provide", "q_proj", "raise", "recommended", "register_buffer", "resid_dropout", "resid_pdrop", "rotary_dim", "scale_attn", "self", "sqrt", "super", "sure", "the", "this", "to", "used", "v_proj", "warning_once", "when", "will", "without"], "gptj/modeling_gptj.py:GPTJAttention._split_heads": ["Input", "Model", "Module", "ValueError", "_split_heads", "attn_head_size", "be", "but", "class", "def", "elif", "else", "f", "if", "is", "len", "new_shape", "nn", "num_attention_heads", "of", "one", "permute", "raise", "rank", "return", "rotary", "self", "shape", "should", "size", "tensor", "view"], "gptj/modeling_gptj.py:GPTJAttention._merge_heads": ["Input", "Model", "Module", "ValueError", "_merge_heads", "attn_head_size", "be", "but", "class", "contiguous", "def", "elif", "else", "f", "if", "is", "len", "new_shape", "nn", "num_attention_heads", "of", "one", "permute", "raise", "rank", "return", "self", "shape", "should", "size", "tensor", "view"], "gptj/modeling_gptj.py:GPTJAttention._attn": ["Model", "Module", "None", "_attn", "attention_mask", "attn_dropout", "attn_output", "attn_weights", "causal_mask", "class", "def", "dim", "dtype", "float32", "functional", "if", "is", "key", "matmul", "nn", "not", "query", "return", "scale_attn", "self", "shape", "softmax", "to", "torch", "transpose", "value"], "gptj/modeling_gptj.py:GPTJAttention._get_embed_positions": ["Model", "Module", "_get_embed_positions", "class", "def", "device", "embed_positions", "if", "nn", "position_ids", "repeat", "return", "self", "shape", "to"], "gptj/modeling_gptj.py:GPTJAttention.forward": ["Cache", "False", "FloatTensor", "LongTensor", "Model", "Module", "None", "Tensor", "True", "_attn", "_get_embed_positions", "_merge_heads", "_split_heads", "apply_rotary_pos_emb", "attention_mask", "attn_output", "attn_weights", "bool", "cache_kwargs", "cache_position", "cat", "class", "cos", "def", "dim", "dtype", "else", "embed_positions", "forward", "gather", "head_dim", "hidden_states", "if", "is", "k_pass", "k_proj", "k_rot", "key", "layer_idx", "layer_past", "nn", "not", "num_attention_heads", "out_proj", "output_attentions", "partial_rotation_size", "permute", "position_ids", "q_pass", "q_proj", "q_rot", "query", "repeat", "repeated_position_ids", "resid_dropout", "return", "rotary_dim", "self", "shape", "sin", "sincos", "split", "to", "torch", "tuple", "unsqueeze", "update", "use_cache", "v_proj", "value"], "gptj/modeling_gptj.py:GPTJFlashAttention2.__init__": ["Model", "ModelAttention", "__init__", "_flash_attn_uses_top_left_mask", "args", "class", "def", "flash_attn_supports_top_left_mask", "kwargs", "self", "super"], "gptj/modeling_gptj.py:GPTJFlashAttention2.forward": ["Cache", "False", "FloatTensor", "LongTensor", "Model", "ModelAttention", "None", "Tensor", "The", "True", "We", "_flash_attention_forward", "_flash_attn_uses_top_left_mask", "_get_embed_positions", "_is_quantized", "_split_heads", "apply_rotary_pos_emb", "attention_dropout", "attention_mask", "attn_output", "attn_pdrop", "attn_weights", "back", "be", "bool", "cache_kwargs", "cache_position", "cast", "casted", "cat", "class", "config", "contiguous", "cos", "cpu", "def", "device", "device_type", "dim", "dropout", "dtype", "elif", "else", "embed_positions", "embedding", "f", "fact", "float32", "forward", "gather", "get_autocast_dtype", "get_autocast_gpu_dtype", "hasattr", "have", "head_dim", "hidden", "hidden_states", "if", "in", "input", "input_dtype", "is", "is_autocast_enabled", "is_causal", "k_pass", "k_proj", "k_rot", "key", "layer", "layer_idx", "layer_past", "layers", "logger", "might", "mps", "norm", "not", "num_attention_heads", "or", "out_proj", "output_attentions", "partial_rotation_size", "permute", "position_ids", "q_pass", "q_proj", "q_rot", "query", "query_length", "related", "repeat", "repeated_position_ids", "reshape", "resid_dropout", "return", "rotary_dim", "seems", "self", "shape", "silently", "sin", "sincos", "split", "states", "target_dtype", "the", "this", "to", "torch", "training", "tuple", "type", "unsqueeze", "upcasted", "update", "use_cache", "use_top_left_mask", "v_proj", "value", "warning_once", "weight", "will", "you"], "gptj/modeling_gptj.py:GPTJMLP.__init__": ["ACT2FN", "Dropout", "Linear", "Model", "Module", "__init__", "act", "activation_function", "class", "config", "def", "dropout", "embed_dim", "fc_in", "fc_out", "intermediate_size", "n_embd", "nn", "resid_pdrop", "self", "super"], "gptj/modeling_gptj.py:GPTJMLP.forward": ["FloatTensor", "Model", "Module", "None", "act", "class", "def", "dropout", "fc_in", "fc_out", "forward", "hidden_states", "nn", "return", "self", "torch"], "gptj/modeling_gptj.py:GPTJBlock.__init__": ["GradientCheckpointingLayer", "LayerNorm", "Model", "ModelMLP", "Model_ATTENTION_CLASSES", "None", "__init__", "_attn_implementation", "attn", "class", "config", "def", "else", "eps", "if", "inner_dim", "is", "layer_idx", "layer_norm_epsilon", "ln_1", "mlp", "n_embd", "n_inner", "nn", "not", "self", "super"], "gptj/modeling_gptj.py:GPTJBlock.forward": ["Cache", "False", "FloatTensor", "GradientCheckpointingLayer", "LongTensor", "Model", "None", "Tensor", "attention_mask", "attn", "attn_outputs", "attn_weights", "bool", "cache_position", "class", "def", "feed_forward_hidden_states", "forward", "hidden_states", "layer_past", "ln_1", "mlp", "output_attentions", "position_ids", "residual", "return", "self", "torch", "tuple", "use_cache"], "gptj/modeling_gptj.py:GPTJPreTrainedModel._init_weights": ["Model", "ModelAttention", "PreTrainedModel", "_init_weights", "class", "copy_", "create_sinusoidal_positions", "def", "embed_positions", "if", "init", "isinstance", "max_positions", "module", "pos_embd_dim", "self", "super"], "gptj/modeling_gptj.py:GPTJModel.__init__": ["Dropout", "Embedding", "False", "LayerNorm", "Model", "ModelBlock", "ModelPreTrainedModel", "ModuleList", "__init__", "class", "config", "def", "drop", "embd_pdrop", "embed_dim", "eps", "for", "gradient_checkpointing", "h", "i", "in", "layer_idx", "layer_norm_epsilon", "ln_f", "n_embd", "n_layer", "nn", "post_init", "range", "self", "super", "vocab_size", "wte"], "gptj/modeling_gptj.py:GPTJModel.get_input_embeddings": ["Model", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "return", "self", "wte"], "gptj/modeling_gptj.py:GPTJModel.set_input_embeddings": ["Model", "ModelPreTrainedModel", "class", "def", "new_embeddings", "self", "set_input_embeddings", "wte"], "gptj/modeling_gptj.py:GPTJModel.forward": ["BaseModelOutputWithPast", "Cache", "DynamicCache", "False", "FloatTensor", "LongTensor", "Model", "ModelPreTrainedModel", "None", "Setting", "True", "ValueError", "You", "_update_causal_mask", "all_hidden_states", "all_self_attentions", "and", "arange", "attention_mask", "attentions", "block", "bool", "cache_position", "causal_mask", "checkpointing", "class", "config", "def", "device", "drop", "else", "enumerate", "exactly", "for", "forward", "get_seq_length", "gradient", "gradient_checkpointing", "h", "hidden_states", "i", "if", "in", "incompatible", "input_ids", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layer_past", "ln_f", "logger", "must", "not", "of", "one", "or", "output_attentions", "output_hidden_states", "output_shape", "outputs", "past_key_values", "past_key_values_length", "position_ids", "r", "raise", "return", "return_dict", "self", "seq_length", "shape", "size", "specify", "token_type_embeds", "token_type_ids", "torch", "training", "tuple", "unsqueeze", "use_cache", "use_return_dict", "v", "view", "warning_once", "with", "wte"], "gptj/modeling_gptj.py:GPTJModel._update_causal_mask": ["AttentionMaskConverter", "BlockMask", "Cache", "False", "Model", "ModelPreTrainedModel", "None", "Tensor", "Union", "_attn_implementation", "_ignore_causal_mask_sdpa", "_prepare_4d_causal_attention_mask_with_cache_position", "_unmask_unattended", "_update_causal_mask", "and", "any", "attention_mask", "batch_size", "bool", "cache_position", "causal_mask", "class", "config", "cuda", "def", "device", "dtype", "else", "finfo", "flash_attention_2", "flex_attention", "get_max_cache_shape", "get_seq_length", "if", "in", "input_tensor", "inputs_embeds", "is", "is_compileable", "is_training", "isinstance", "make_flex_block_causal_mask", "min", "min_dtype", "not", "npu", "output_attentions", "past_key_values", "past_key_values_length", "past_seen_tokens", "return", "sdpa", "self", "sequence_length", "shape", "target_length", "torch", "training", "type", "using_compilable_cache", "xpu"], "gptj/modeling_gptj.py:GPTJModel._prepare_4d_causal_attention_mask_with_cache_position": ["Model", "ModelPreTrainedModel", "None", "Tensor", "_prepare_4d_causal_attention_mask_with_cache_position", "and", "arange", "attention_mask", "batch_size", "cache_position", "causal_mask", "class", "clone", "def", "device", "diagonal", "dim", "dtype", "else", "expand", "fill_value", "finfo", "full", "if", "int", "is", "kwargs", "mask_length", "masked_fill", "min", "min_dtype", "not", "padding_mask", "reshape", "return", "sequence_length", "shape", "target_length", "to", "torch", "triu"], "gptj/modeling_gptj.py:GPTJForCausalLM.__init__": ["GenerationMixin", "Linear", "Model", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "config", "def", "lm_head", "n_embd", "nn", "post_init", "self", "super", "transformer", "vocab_size"], "gptj/modeling_gptj.py:GPTJForCausalLM.forward": ["Cache", "CausalLMOutputWithPast", "FloatTensor", "GenerationMixin", "LongTensor", "Model", "ModelPreTrainedModel", "None", "Tensor", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "not", "output", "output_attentions", "output_hidden_states", "past_key_values", "position_ids", "r", "return", "return_dict", "self", "slice", "slice_indices", "token_type_ids", "torch", "transformer", "transformer_outputs", "tuple", "use_cache", "use_return_dict", "vocab_size"], "gptj/modeling_gptj.py:GPTJForSequenceClassification.__init__": ["False", "Linear", "Model", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "n_embd", "nn", "num_labels", "post_init", "score", "self", "super", "transformer"], "gptj/modeling_gptj.py:GPTJForSequenceClassification.forward": ["BCEWithLogitsLoss", "Cache", "Cannot", "CrossEntropyLoss", "FloatTensor", "LongTensor", "MSELoss", "Model", "ModelPreTrainedModel", "None", "Results", "SequenceClassifierOutputWithPast", "ValueError", "__class__", "__name__", "and", "arange", "argmax", "attention_mask", "attentions", "batch", "batch_size", "be", "bool", "class", "config", "conjunction", "def", "defined", "detect", "device", "dtype", "elif", "else", "f", "forward", "handle", "hidden_states", "if", "in", "input_ids", "inputs_embeds", "int", "int32", "is", "kwargs", "labels", "last_non_pad_token", "logger", "logits", "long", "loss", "loss_fct", "may", "multi_label_classification", "no", "non_pad_mask", "not", "num_labels", "or", "output", "output_attentions", "output_hidden_states", "pad_token_id", "padding", "past_key_values", "pooled_logits", "position_ids", "problem_type", "r", "raise", "regression", "return", "return_dict", "score", "self", "shape", "single_label_classification", "sizes", "squeeze", "to", "token", "token_indices", "token_type_ids", "tokens", "torch", "transformer", "transformer_outputs", "tuple", "unexpected", "use_cache", "use_return_dict", "using", "view", "warning_once", "will", "with"], "gptj/modeling_gptj.py:GPTJForQuestionAnswering.__init__": ["Linear", "Model", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "config", "def", "hidden_size", "nn", "num_labels", "post_init", "qa_outputs", "self", "super", "transformer"], "gptj/modeling_gptj.py:GPTJForQuestionAnswering.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "Model", "ModelPreTrainedModel", "None", "QuestionAnsweringModelOutput", "and", "attention_mask", "attentions", "bool", "clamp", "class", "config", "contiguous", "def", "device", "dim", "else", "end_logits", "end_loss", "end_positions", "forward", "hidden_states", "if", "ignore_index", "ignored_index", "input_ids", "inputs_embeds", "is", "kwargs", "len", "logits", "loss", "loss_fct", "not", "output", "output_attentions", "output_hidden_states", "outputs", "position_ids", "qa_outputs", "r", "return", "return_dict", "self", "sequence_output", "size", "split", "squeeze", "start_logits", "start_loss", "start_positions", "to", "token_type_ids", "torch", "total_loss", "transformer", "tuple", "use_return_dict"], "granite/modeling_granite.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "granite/modeling_granite.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cos", "def", "k", "k_embed", "q", "q_embed", "return", "rotate_half", "sin", "unsqueeze", "unsqueeze_dim"], "granite/modeling_granite.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "granite/modeling_granite.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value", "value_states"], "granite/modeling_granite.py:GraniteAttention.__init__": ["Linear", "ModelAttention", "ModelConfig", "Module", "None", "True", "__init__", "attention_bias", "attention_dropout", "attention_multiplier", "bias", "class", "config", "def", "getattr", "head_dim", "hidden_size", "int", "is_causal", "k_proj", "layer_idx", "nn", "num_attention_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "q_proj", "scaling", "self", "super", "v_proj"], "granite/modeling_granite.py:GraniteAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "LongTensor", "ModelAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_attn_implementation", "apply_rotary_pos_emb", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "cache_kwargs", "cache_position", "class", "config", "contiguous", "cos", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "head_dim", "hidden_shape", "hidden_states", "if", "input_shape", "is", "k_proj", "key_states", "kwargs", "layer_idx", "nn", "not", "o_proj", "past_key_values", "position_embeddings", "q_proj", "query_states", "reshape", "return", "scaling", "self", "shape", "sin", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "view"], "granite/modeling_granite.py:GraniteRMSNorm.__init__": ["ModelRMSNorm", "Module", "Parameter", "__init__", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "granite/modeling_granite.py:GraniteRMSNorm.forward": ["ModelRMSNorm", "Module", "True", "class", "def", "dtype", "float32", "forward", "hidden_states", "input_dtype", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "granite/modeling_granite.py:GraniteRMSNorm.extra_repr": ["ModelRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "variance_epsilon", "weight"], "granite/modeling_granite.py:GraniteMLP.__init__": ["ACT2FN", "Linear", "ModelMLP", "Module", "__init__", "act_fn", "bias", "class", "config", "def", "down_proj", "gate_proj", "hidden_act", "hidden_size", "intermediate_size", "mlp_bias", "nn", "self", "super", "up_proj"], "granite/modeling_granite.py:GraniteMLP.forward": ["ModelMLP", "Module", "act_fn", "class", "def", "down_proj", "forward", "gate_proj", "nn", "return", "self", "up_proj", "x"], "granite/modeling_granite.py:GraniteDecoderLayer.__init__": ["GradientCheckpointingLayer", "ModelAttention", "ModelConfig", "ModelDecoderLayer", "ModelMLP", "ModelRMSNorm", "__init__", "class", "config", "def", "eps", "hidden_size", "input_layernorm", "int", "layer_idx", "mlp", "post_attention_layernorm", "residual_multiplier", "rms_norm_eps", "self", "self_attn", "super"], "granite/modeling_granite.py:GraniteDecoderLayer.forward": ["Cache", "False", "FloatTensor", "GradientCheckpointingLayer", "LongTensor", "ModelDecoderLayer", "None", "Tensor", "attention_mask", "bool", "cache_position", "class", "def", "forward", "hidden_states", "if", "input_layernorm", "kwargs", "mlp", "output_attentions", "outputs", "past_key_values", "position_embeddings", "position_ids", "post_attention_layernorm", "residual", "residual_multiplier", "return", "self", "self_attn", "self_attn_weights", "torch", "tuple", "use_cache"], "granite/modeling_granite.py:GraniteRotaryEmbedding.__init__": ["Callable", "False", "ModelConfig", "ModelRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "granite/modeling_granite.py:GraniteRotaryEmbedding.compute_default_rope_parameters": ["ModelConfig", "ModelRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "granite/modeling_granite.py:GraniteRotaryEmbedding.forward": ["False", "ModelRotaryEmbedding", "Module", "None", "and", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "granite/modeling_granite.py:GraniteModel.__init__": ["Embedding", "False", "ModelConfig", "ModelDecoderLayer", "ModelModel", "ModelPreTrainedModel", "ModelRMSNorm", "ModelRotaryEmbedding", "ModuleList", "__init__", "class", "config", "def", "embed_tokens", "embedding_multiplier", "eps", "for", "gradient_checkpointing", "hidden_size", "in", "layer_idx", "layers", "nn", "norm", "num_hidden_layers", "pad_token_id", "padding_idx", "post_init", "range", "rms_norm_eps", "rotary_emb", "self", "super", "vocab_size"], "granite/modeling_granite.py:GraniteModel.forward": ["BaseModelOutputWithPast", "Cache", "DynamicCache", "False", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Setting", "Tensor", "TransformersKwargs", "True", "Unpack", "ValueError", "You", "all_hidden_states", "all_self_attns", "and", "arange", "attention_mask", "attentions", "bool", "cache_position", "causal_mask", "checkpointing", "class", "config", "create_causal_mask", "decoder_layer", "def", "device", "else", "embed_tokens", "embedding_multiplier", "exactly", "for", "forward", "get_seq_length", "gradient", "gradient_checkpointing", "hidden_states", "if", "in", "incompatible", "input_embeds", "input_ids", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layer_outputs", "layers", "logger", "must", "norm", "not", "num_hidden_layers", "of", "one", "or", "output_attentions", "output_hidden_states", "past_key_values", "past_seen_tokens", "position_embeddings", "position_ids", "raise", "return", "rotary_emb", "self", "shape", "specify", "torch", "training", "unsqueeze", "use_cache", "warning_once", "with"], "granite/modeling_granite.py:GraniteForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "ModelForCausalLM", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "vocab_size"], "granite/modeling_granite.py:GraniteForCausalLM.forward": ["BaseModelOutputWithPast", "Cache", "CausalLMOutputWithPast", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForCausalLM", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "logits", "logits_scaling", "logits_to_keep", "loss", "loss_function", "model", "not", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "position_ids", "r", "return", "self", "slice", "slice_indices", "torch", "use_cache", "vocab_size"], "granite_speech/modeling_granite_speech.py:GraniteSpeechEncoderProjector.__init__": ["AutoModel", "Linear", "ModelConfig", "ModelEncoderProjector", "Module", "Parameter", "__init__", "class", "config", "data", "def", "downsample_rate", "from_config", "hidden_size", "linear", "mean", "nn", "normal_", "num_queries", "projector_config", "qformer", "query", "self", "std", "super", "text_config", "torch", "window_size", "zeros"], "granite_speech/modeling_granite_speech.py:GraniteSpeechEncoderProjector.forward": ["ModelEncoderProjector", "Module", "None", "Tensor", "True", "batch_size", "ceil", "class", "constant", "def", "dim", "downsample_rate", "encoder_attention_mask", "encoder_hidden_states", "forward", "functional", "hidden_states", "last_hidden_state", "linear", "math", "nblocks", "nn", "pad", "qformer", "query", "query_embeds", "query_output", "query_proj", "return", "return_dict", "self", "seq_len", "size", "torch", "view", "window_size"], "granite_speech/modeling_granite_speech.py:GraniteSpeechConformerFeedForward.__init__": ["Dropout", "LayerNorm", "Linear", "ModelConformerFeedForward", "ModelEncoderConfig", "Module", "SiLU", "__init__", "class", "config", "def", "down_proj", "dropout", "feedforward_mult", "hidden_dim", "nn", "pre_norm", "self", "silu", "super", "up_proj"], "granite_speech/modeling_granite_speech.py:GraniteSpeechConformerFeedForward.forward": ["ModelConformerFeedForward", "Module", "Tensor", "class", "def", "down_proj", "dropout", "forward", "hidden_states", "nn", "pre_norm", "return", "self", "silu", "torch", "up_proj"], "granite_speech/modeling_granite_speech.py:GraniteSpeechConformerAttention.__init__": ["Context", "Dropout", "Embedding", "False", "LayerNorm", "Linear", "ModelConformerAttention", "ModelEncoderConfig", "Module", "ValueError", "__init__", "bias", "class", "config", "context_size", "def", "dim_head", "dropout", "either", "exceeds", "hidden_dim", "if", "inner_dim", "is", "less", "max_pos_emb", "nn", "num_heads", "or", "pre_norm", "raise", "rel_pos_emb", "scale", "self", "size", "super", "than", "the", "to_kv", "to_out", "to_q"], "granite_speech/modeling_granite_speech.py:GraniteSpeechConformerAttention.forward": ["F", "MATH", "ModelConformerAttention", "Module", "SDPBackend", "Tensor", "_", "attention", "attention_dists", "attn_mask", "b", "bool", "bsz", "c", "ceil", "chunk", "class", "context_size", "d", "def", "device", "dim", "dropout", "dtype", "einsum", "finfo", "forward", "functional", "h", "hidden_states", "if", "key_states", "m", "mask", "mask_value", "masked_fill_", "math", "max", "nn", "num_blocks", "num_features", "num_heads", "ones", "out", "pad", "pos_attn", "pre_norm", "query_states", "r", "rel_pos_emb", "remainder", "reshape", "return", "scale", "scaled_dot_product_attention", "sdpa_kernel", "self", "shape", "to_kv", "to_out", "to_q", "torch", "transpose", "value_states", "with"], "granite_speech/modeling_granite_speech.py:GraniteSpeechConformerDepthWiseConv1d.__init__": ["Conv1d", "False", "ModelConformerDepthWiseConv1d", "Module", "__init__", "bias", "chan_in", "chan_out", "class", "conv", "def", "groups", "int", "kernel_size", "nn", "pad", "pad_offset", "padding", "self", "super"], "granite_speech/modeling_granite_speech.py:GraniteSpeechConformerDepthWiseConv1d.forward": ["F", "ModelConformerDepthWiseConv1d", "Module", "Tensor", "class", "conv", "def", "forward", "hidden_states", "nn", "pad", "padding", "return", "self", "torch"], "granite_speech/modeling_granite_speech.py:GraniteSpeechConformerConvModule.__init__": ["BatchNorm1d", "Conv1d", "Dropout", "GLU", "LayerNorm", "ModelConformerConvModule", "ModelConformerDepthWiseConv1d", "ModelEncoderConfig", "Module", "SiLU", "__init__", "batch_norm", "class", "config", "conv_expansion_factor", "conv_kernel_size", "def", "depth_conv", "dim", "down_conv", "dropout", "glu", "hidden_dim", "inner_dim", "kernel_size", "nn", "norm", "self", "silu", "super", "up_conv"], "granite_speech/modeling_granite_speech.py:GraniteSpeechConformerConvModule.forward": ["ModelConformerConvModule", "Module", "Tensor", "batch_norm", "class", "def", "depth_conv", "down_conv", "dropout", "forward", "glu", "hidden_states", "nn", "norm", "permute", "return", "self", "silu", "torch", "up_conv"], "granite_speech/modeling_granite_speech.py:GraniteSpeechConformerBlock.__init__": ["LayerNorm", "ModelConformerAttention", "ModelConformerBlock", "ModelConformerConvModule", "ModelConformerFeedForward", "ModelEncoderConfig", "Module", "__init__", "attn", "class", "config", "conv", "def", "ff1", "ff2", "hidden_dim", "nn", "post_norm", "self", "super"], "granite_speech/modeling_granite_speech.py:GraniteSpeechConformerBlock.forward": ["ModelConformerBlock", "Module", "Tensor", "attention_dists", "attn", "class", "conv", "def", "ff1", "ff2", "forward", "hidden_states", "nn", "post_norm", "return", "self", "torch"], "granite_speech/modeling_granite_speech.py:GraniteSpeechCTCEncoder.__init__": ["False", "Linear", "ModelCTCEncoder", "ModelConformerBlock", "ModelEncoderConfig", "Module", "ModuleList", "True", "_", "__init__", "arange", "attention_dists", "bias", "clamp", "class", "config", "context_size", "def", "for", "hidden_dim", "in", "input_dim", "input_linear", "layers", "max_pos_emb", "nn", "num_layers", "out", "out_mid", "output_dim", "persistent", "range", "register_buffer", "relpos_dist", "self", "seq", "super", "torch", "view"], "granite_speech/modeling_granite_speech.py:GraniteSpeechCTCEncoder.forward": ["ModelCTCEncoder", "Module", "Softmax", "Tensor", "attention_dists", "class", "clone", "def", "dim", "enumerate", "for", "forward", "hidden_states", "hidden_states_mid", "idx", "if", "in", "input_linear", "layer", "layers", "nn", "num_layers", "out", "out_mid", "return", "self", "start", "torch"], "granite_speech/modeling_granite_speech.py:GraniteSpeechPreTrainedModel._init_weights": ["ModelCTCEncoder", "ModelEncoderProjector", "ModelPreTrainedModel", "Module", "PreTrainedModel", "_init_weights", "arange", "attention_dists", "clamp", "class", "config", "context_size", "copy_", "def", "elif", "if", "init", "isinstance", "max_pos_emb", "module", "nn", "normal_", "query", "relpos_dist", "self", "seq", "super", "torch", "view"], "granite_speech/modeling_granite_speech.py:GraniteSpeechForConditionalGeneration.__init__": ["AutoModelForCausalLM", "Config", "GenerationMixin", "ModelCTCEncoder", "ModelConfig", "ModelEncoderProjector", "ModelForConditionalGeneration", "ModelPreTrainedModel", "Please", "__init__", "a", "adapter", "and", "are", "audio", "be", "but", "cause", "class", "config", "def", "encoder", "encoder_config", "from_config", "has_lora_adapter", "if", "incorrectly", "indicates", "inputs", "install", "installed", "is", "is_peft_available", "language_model", "logger", "lora", "model", "not", "peft", "perform", "post_init", "present", "projector", "provided", "reload", "self", "should", "super", "text_config", "that", "the", "this", "to", "warning", "when", "will"], "granite_speech/modeling_granite_speech.py:GraniteSpeechForConditionalGeneration.set_decoder": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "decoder", "def", "language_model", "self", "set_decoder"], "granite_speech/modeling_granite_speech.py:GraniteSpeechForConditionalGeneration.get_decoder": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "get_decoder", "language_model", "return", "self"], "granite_speech/modeling_granite_speech.py:GraniteSpeechForConditionalGeneration.set_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "language_model", "self", "set_input_embeddings", "value"], "granite_speech/modeling_granite_speech.py:GraniteSpeechForConditionalGeneration.set_output_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "language_model", "new_embeddings", "self", "set_output_embeddings"], "granite_speech/modeling_granite_speech.py:GraniteSpeechForConditionalGeneration.get_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "language_model", "return", "self"], "granite_speech/modeling_granite_speech.py:GraniteSpeechForConditionalGeneration.get_output_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "get_output_embeddings", "language_model", "return", "self"], "granite_speech/modeling_granite_speech.py:GraniteSpeechForConditionalGeneration.get_audio_features": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "Tensor", "class", "def", "encoder", "encoder_embeds", "get_audio_features", "input_features", "projected_embeds", "projector", "return", "self", "torch"], "granite_speech/modeling_granite_speech.py:GraniteSpeechForConditionalGeneration.forward": ["Cache", "CrossEntropyLoss", "FloatTensor", "GenerationMixin", "LongTensor", "ModelCausalLMOutputWithPast", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Tensor", "ValueError", "You", "and", "at", "attention_mask", "attentions", "audio_embeds", "audio_features", "audio_token_id", "bool", "both", "cache_position", "cannot", "class", "clone", "config", "contiguous", "def", "device", "dtype", "either", "else", "exactly", "forward", "get_audio_features", "get_input_embeddings", "get_merged_audio_embeddings", "hidden_states", "if", "input_features", "input_features_mask", "input_ids", "inputs_embeds", "int", "is", "is_audio_idx", "labels", "language_model", "llm_input_ids", "lm_kwargs", "logits", "logits_to_keep", "loss", "loss_fct", "must", "nn", "not", "of", "one", "or", "output", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "position_ids", "r", "raise", "return", "return_dict", "same", "self", "shape", "shift_attention_mask", "shift_labels", "shift_logits", "size", "specify", "the", "time", "to", "torch", "tuple", "use_cache", "use_return_dict", "view"], "granite_speech/modeling_granite_speech.py:GraniteSpeechForConditionalGeneration.prepare_inputs_for_generation": ["False", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "True", "attention_mask", "cache_position", "class", "def", "get", "if", "input_features", "input_ids", "inputs_embeds", "is_first_iteration", "kwargs", "language_model", "logits_to_keep", "model_inputs", "not", "or", "past_key_values", "prepare_inputs_for_generation", "return", "self", "use_cache"], "granite_speech/modeling_granite_speech.py:GraniteSpeechForConditionalGeneration.get_merged_audio_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Number", "Tensor", "ValueError", "all", "audio", "audio_features", "audio_token_id", "class", "config", "def", "device", "dim", "does", "dtype", "features", "get_input_embeddings", "get_merged_audio_embeddings", "if", "input_features_mask", "input_ids", "inputs_embeds", "int", "is", "is_audio_index", "item", "language_model", "llm_input_ids", "masked_scatter", "match", "not", "number", "of", "raise", "return", "self", "special_audio_mask", "sum", "to", "tokens", "torch", "unsqueeze", "where"], "granite_speech/modeling_granite_speech.py:GraniteSpeechForConditionalGeneration.generate": ["GenerationMixin", "LongTensor", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "_hf_peft_config_loaded", "and", "args", "class", "def", "disable_adapters", "else", "enable_adapters", "generate", "if", "input_features", "is", "is_peft_available", "kwargs", "not", "pop", "return", "self", "super", "torch"], "granite_speech/modeling_granite_speech.py:GraniteSpeechForConditionalGeneration.save_pretrained": ["False", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "_get_adapter_name", "_hf_peft_config_loaded", "adapter_name", "and", "args", "base_model_name_or_path", "class", "def", "if", "is_peft_available", "kwargs", "peft_config", "prev_val", "save_directory", "save_pretrained", "self", "super"], "granite_speech/modeling_granite_speech.py:GraniteSpeechForConditionalGeneration._get_adapter_name": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "_get_adapter_name", "class", "def", "keys", "list", "peft_config", "return", "self"], "granitemoe/modeling_granitemoe.py:GraniteMoeRMSNorm.__init__": ["ModelRMSNorm", "Module", "Parameter", "__init__", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "granitemoe/modeling_granitemoe.py:GraniteMoeRMSNorm.forward": ["ModelRMSNorm", "Module", "True", "class", "def", "dtype", "float32", "forward", "hidden_states", "input_dtype", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "granitemoe/modeling_granitemoe.py:GraniteMoeRMSNorm.extra_repr": ["ModelRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "variance_epsilon", "weight"], "granitemoe/modeling_granitemoe.py:GraniteMoeRotaryEmbedding.__init__": ["Callable", "False", "ModelConfig", "ModelRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "granitemoe/modeling_granitemoe.py:GraniteMoeRotaryEmbedding.compute_default_rope_parameters": ["ModelConfig", "ModelRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "granitemoe/modeling_granitemoe.py:GraniteMoeRotaryEmbedding.forward": ["False", "ModelRotaryEmbedding", "Module", "None", "and", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "granitemoe/modeling_granitemoe.py:GraniteMoeParallelExperts.__init__": ["ModelParallelExperts", "Module", "None", "Parameter", "__init__", "class", "def", "empty", "input_size", "int", "nn", "num_experts", "output_size", "self", "super", "torch", "weight"], "granitemoe/modeling_granitemoe.py:GraniteMoeParallelExperts.forward": ["F", "ModelParallelExperts", "Module", "append", "cat", "class", "def", "dim", "expert_size", "for", "forward", "i", "in", "input_list", "inputs", "linear", "nn", "num_experts", "output_list", "range", "results", "return", "self", "split", "torch", "weight"], "granitemoe/modeling_granitemoe.py:GraniteMoeTopKGating.__init__": ["False", "Linear", "ModelTopKGating", "Module", "__init__", "bias", "class", "def", "input_size", "int", "layer", "nn", "num_experts", "self", "super", "top_k"], "granitemoe/modeling_granitemoe.py:GraniteMoeTopKGating.forward": ["ModelTopKGating", "Module", "_", "batch_gates", "batch_index", "class", "def", "device", "dim", "div", "dtype", "expert_size", "flatten", "float", "forward", "gates", "hidden_states", "index_sorted_experts", "layer", "logits", "long", "nn", "num_experts", "return", "rounding_mode", "scatter", "self", "size", "softmax", "sort", "sum", "tolist", "top_k", "top_k_experts", "top_k_gates", "top_k_indices", "top_k_logits", "topk", "torch", "trunc", "type_as", "zeros"], "granitemoe/modeling_granitemoe.py:GraniteMoeMoE.__init__": ["ACT2FN", "ModelConfig", "ModelMoE", "ModelParallelExperts", "ModelTopKGating", "Module", "__init__", "activation", "class", "config", "def", "hidden_act", "hidden_size", "input_linear", "input_size", "intermediate_size", "nn", "num_experts", "num_experts_per_tok", "num_local_experts", "output_linear", "router", "self", "super", "top_k"], "granitemoe/modeling_granitemoe.py:GraniteMoeMoE.forward": ["ModelMoE", "Module", "None", "_", "activation", "batch_gates", "batch_index", "bsz", "chunk", "chunked_hidden_states", "class", "def", "device", "dim", "dtype", "emb_size", "expert_inputs", "expert_outputs", "expert_size", "forward", "hidden_states", "index_add", "input_linear", "input_size", "layer_input", "layer_output", "length", "nn", "output_linear", "reshape", "return", "router", "self", "size", "torch", "view", "zeros"], "granitemoe/modeling_granitemoe.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "granitemoe/modeling_granitemoe.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cos", "def", "k", "k_embed", "q", "q_embed", "return", "rotate_half", "sin", "unsqueeze", "unsqueeze_dim"], "granitemoe/modeling_granitemoe.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "granitemoe/modeling_granitemoe.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value", "value_states"], "granitemoe/modeling_granitemoe.py:GraniteMoeAttention.__init__": ["Linear", "ModelAttention", "ModelConfig", "Module", "True", "__init__", "attention_bias", "attention_dropout", "attention_multiplier", "bias", "class", "config", "def", "getattr", "head_dim", "hidden_size", "int", "is_causal", "k_proj", "layer_idx", "nn", "num_attention_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "q_proj", "scaling", "self", "super", "v_proj"], "granitemoe/modeling_granitemoe.py:GraniteMoeAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "LongTensor", "ModelAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_attn_implementation", "apply_rotary_pos_emb", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "cache_kwargs", "cache_position", "class", "config", "contiguous", "cos", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "head_dim", "hidden_shape", "hidden_states", "if", "input_shape", "is", "k_proj", "key_states", "kwargs", "layer_idx", "nn", "not", "o_proj", "past_key_values", "position_embeddings", "q_proj", "query_states", "reshape", "return", "scaling", "self", "shape", "sin", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "view"], "granitemoe/modeling_granitemoe.py:GraniteMoeDecoderLayer.__init__": ["GradientCheckpointingLayer", "ModelAttention", "ModelConfig", "ModelDecoderLayer", "ModelMoE", "ModelRMSNorm", "__init__", "block_sparse_moe", "class", "config", "def", "eps", "hidden_size", "input_layernorm", "int", "layer_idx", "post_attention_layernorm", "residual_multiplier", "rms_norm_eps", "self", "self_attn", "super"], "granitemoe/modeling_granitemoe.py:GraniteMoeDecoderLayer.forward": ["Cache", "GradientCheckpointingLayer", "LongTensor", "ModelDecoderLayer", "None", "Tensor", "_", "attention_mask", "block_sparse_moe", "cache_position", "class", "def", "forward", "hidden_states", "input_layernorm", "kwargs", "past_key_values", "position_embeddings", "post_attention_layernorm", "residual", "residual_multiplier", "return", "self", "self_attn", "torch", "tuple"], "granitemoe/modeling_granitemoe.py:GraniteMoePreTrainedModel._init_weights": ["ModelParallelExperts", "ModelPreTrainedModel", "PreTrainedModel", "_init_weights", "class", "config", "def", "if", "init", "initializer_range", "isinstance", "mean", "module", "normal_", "self", "std", "super", "weight"], "granitemoe/modeling_granitemoe.py:GraniteMoeModel.__init__": ["Embedding", "False", "ModelConfig", "ModelDecoderLayer", "ModelModel", "ModelPreTrainedModel", "ModelRMSNorm", "ModelRotaryEmbedding", "ModuleList", "__init__", "class", "config", "def", "embed_tokens", "embedding_multiplier", "eps", "for", "gradient_checkpointing", "hidden_size", "in", "layer_idx", "layers", "nn", "norm", "num_hidden_layers", "pad_token_id", "padding_idx", "post_init", "range", "rms_norm_eps", "rotary_emb", "self", "super", "vocab_size"], "granitemoe/modeling_granitemoe.py:GraniteMoeModel.forward": ["Cache", "DynamicCache", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "MoeModelOutputWithPast", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "and", "arange", "attention_mask", "bool", "cache_position", "causal_mask", "class", "config", "create_causal_mask", "decoder_layer", "def", "device", "else", "embed_tokens", "embedding_multiplier", "exactly", "for", "forward", "get_seq_length", "hidden_states", "if", "in", "input_embeds", "input_ids", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layers", "must", "norm", "not", "num_hidden_layers", "of", "one", "or", "past_key_values", "past_seen_tokens", "position_embeddings", "position_ids", "raise", "return", "rotary_emb", "self", "shape", "specify", "torch", "unsqueeze", "use_cache"], "granitemoe/modeling_granitemoe.py:load_balancing_loss_func": ["Model_balancing_loss_func", "None", "Tensor", "_", "attention_mask", "batch_size", "cat", "compute_device", "concatenated_gate_logits", "def", "device", "dim", "else", "expand", "expert_attention_mask", "expert_mask", "float", "for", "functional", "gate_logits", "if", "in", "int", "is", "isinstance", "layer_gate", "mean", "nn", "not", "num_experts", "num_hidden_layers", "one_hot", "or", "overall_loss", "r", "reshape", "return", "router_per_expert_attention_mask", "router_prob_per_expert", "routing_weights", "selected_experts", "sequence_length", "shape", "softmax", "sum", "to", "tokens_per_expert", "top_k", "topk", "torch", "tuple", "unsqueeze"], "granitemoe/modeling_granitemoe.py:GraniteMoeForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "ModelConfig", "ModelForCausalLM", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "logits_scaling", "model", "nn", "num_experts", "num_experts_per_tok", "num_local_experts", "post_init", "router_aux_loss_coef", "self", "super", "vocab_size"], "granitemoe/modeling_granitemoe.py:GraniteMoeForCausalLM.forward": ["Cache", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForCausalLM", "ModelPreTrainedModel", "MoeCausalLMOutputWithPast", "None", "Tensor", "attention_mask", "attentions", "aux_loss", "bool", "cache_position", "class", "config", "def", "device", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "load_balancing_loss_func", "logits", "logits_scaling", "logits_to_keep", "loss", "loss_function", "model", "not", "num_experts", "num_experts_per_tok", "output_router_logits", "outputs", "past_key_values", "position_ids", "r", "return", "router_aux_loss_coef", "router_logits", "self", "slice", "slice_indices", "to", "torch", "tuple", "vocab_size"], "granitemoehybrid/modeling_granitemoehybrid.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "granitemoehybrid/modeling_granitemoehybrid.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cos", "def", "k", "k_embed", "q", "q_embed", "return", "rotate_half", "sin", "unsqueeze", "unsqueeze_dim"], "granitemoehybrid/modeling_granitemoehybrid.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "granitemoehybrid/modeling_granitemoehybrid.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value", "value_states"], "granitemoehybrid/modeling_granitemoehybrid.py:GraniteMoeHybridAttention.__init__": ["Linear", "ModelAttention", "ModelConfig", "Module", "True", "__init__", "attention_bias", "attention_dropout", "attention_multiplier", "bias", "class", "config", "def", "getattr", "head_dim", "hidden_size", "int", "is_causal", "k_proj", "layer_idx", "nn", "num_attention_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "q_proj", "scaling", "self", "super", "v_proj"], "granitemoehybrid/modeling_granitemoehybrid.py:GraniteMoeHybridAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "LongTensor", "ModelAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_attn_implementation", "apply_rotary_pos_emb", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "cache_kwargs", "cache_position", "class", "config", "contiguous", "cos", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "head_dim", "hidden_shape", "hidden_states", "if", "input_shape", "is", "k_proj", "key_states", "kwargs", "layer_idx", "nn", "not", "o_proj", "past_key_values", "position_embeddings", "q_proj", "query_states", "reshape", "return", "scaling", "self", "shape", "sin", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "view"], "granitemoehybrid/modeling_granitemoehybrid.py:HybridMambaAttentionDynamicCache.__init__": ["False", "ModelConfig", "ModelMambaAttentionDynamicCache", "None", "_", "__init__", "append", "batch_size", "class", "config", "conv_kernel_size", "conv_states", "def", "device", "dtype", "else", "float16", "for", "has_previous_state", "hidden_size", "i", "if", "in", "key_cache", "layers_block_type", "mamba", "mamba_d_conv", "mamba_d_head", "mamba_d_state", "mamba_expand", "mamba_n_groups", "mamba_n_heads", "num_hidden_layers", "range", "self", "ssm_state_size", "ssm_states", "tensor", "torch", "transformer_layers", "value_cache", "zeros"], "granitemoehybrid/modeling_granitemoehybrid.py:HybridMambaAttentionDynamicCache.__len__": ["ModelMambaAttentionDynamicCache", "__len__", "class", "def", "key_cache", "len", "return", "self"], "granitemoehybrid/modeling_granitemoehybrid.py:HybridMambaAttentionDynamicCache.__getitem__": ["ModelMambaAttentionDynamicCache", "__getitem__", "class", "def", "key_cache", "layer_idx", "return", "self", "value_cache"], "granitemoehybrid/modeling_granitemoehybrid.py:HybridMambaAttentionDynamicCache.update": ["Any", "ModelMambaAttentionDynamicCache", "None", "Tensor", "cache_kwargs", "cat", "class", "def", "dict", "dim", "else", "if", "int", "key_cache", "key_states", "layer_idx", "return", "self", "shape", "str", "torch", "tuple", "update", "value_cache", "value_states"], "granitemoehybrid/modeling_granitemoehybrid.py:HybridMambaAttentionDynamicCache.reorder_cache": ["LongTensor", "ModelMambaAttentionDynamicCache", "beam_idx", "class", "conv_states", "def", "device", "for", "get_seq_length", "if", "in", "index_select", "key_cache", "layer_idx", "len", "range", "reorder_cache", "self", "ssm_states", "to", "torch", "value_cache"], "granitemoehybrid/modeling_granitemoehybrid.py:HybridMambaAttentionDynamicCache.get_mask_sizes": ["ModelMambaAttentionDynamicCache", "Tensor", "cache_position", "class", "def", "get_mask_sizes", "get_seq_length", "int", "kv_length", "kv_offset", "layer_idx", "query_length", "return", "self", "shape", "torch", "tuple"], "granitemoehybrid/modeling_granitemoehybrid.py:HybridMambaAttentionDynamicCache.get_seq_length": ["ModelMambaAttentionDynamicCache", "None", "class", "def", "else", "get_seq_length", "if", "in", "int", "key_cache", "layer_idx", "len", "not", "or", "return", "self", "shape", "transformer_layers"], "granitemoehybrid/modeling_granitemoehybrid.py:pad_tensor_by_size": ["Model", "Model_shape", "Model_size", "Model_tensor_by_size", "Tensor", "constant", "def", "else", "functional", "if", "input_tensor", "int", "len", "mode", "nn", "return", "shape", "torch", "value"], "granitemoehybrid/modeling_granitemoehybrid.py:reshape_into_chunks": ["Model", "Model_into_chunks", "chunk_size", "def", "else", "if", "input_tensor", "len", "pad_size", "pad_tensor_by_size", "return", "shape"], "granitemoehybrid/modeling_granitemoehybrid.py:segment_sum": ["Model_sum", "None", "bool", "chunk_size", "cumsum", "def", "device", "diagonal", "dim", "dtype", "expand", "inf", "input_tensor", "mask", "masked_fill", "ones", "return", "size", "tensor_segsum", "torch", "tril"], "granitemoehybrid/modeling_granitemoehybrid.py:apply_mask_to_padding_states": ["Model_mask_to_padding_states", "None", "and", "attention_mask", "def", "dtype", "hidden_states", "if", "is", "not", "return", "shape", "to"], "granitemoehybrid/modeling_granitemoehybrid.py:GraniteMoeHybridMambaLayer.__init__": ["A", "ACT2FN", "AILab", "A_log", "Conv1d", "D", "Dao", "Falling", "GPU", "Linear", "Model", "ModelConfig", "ModelMambaLayer", "ModelRMSNormGated", "Module", "None", "Parameter", "The", "To", "__init__", "a", "act", "activation", "all", "arange", "available", "back", "be", "because", "bias", "causal", "causal_conv1d", "causal_conv1d_fn", "causal_conv1d_update", "chunk_size", "class", "com", "config", "conv1d", "conv_dim", "conv_kernel_size", "def", "dt_bias", "else", "eps", "fast", "float", "follow", "for", "getattr", "github", "global", "groups", "head_dim", "hidden_act", "hidden_size", "https", "if", "implementation", "in_channels", "in_proj", "inf", "install", "int", "intermediate_size", "is", "is_fast_path_available", "kernel_size", "layer_idx", "layer_norm_epsilon", "lazy_load_kernel", "log", "logger", "mamba", "mamba_chunk_scan_combined", "mamba_chunk_size", "mamba_conv_bias", "mamba_d_conv", "mamba_d_head", "mamba_d_state", "mamba_expand", "mamba_n_groups", "mamba_n_heads", "mamba_proj_bias", "mamba_split_conv1d_scan_combined", "mamba_ssm", "model", "n_groups", "naive", "nn", "norm", "not", "num_heads", "of", "on", "one", "ones", "out_channels", "out_proj", "padding", "path", "projection_size", "rms_norm_eps", "running", "selective_state_update", "self", "spaces", "ssm", "ssm_state_size", "state", "super", "the", "time_step_limit", "time_step_max", "time_step_min", "to", "torch", "use_bias", "use_conv_bias", "used", "warning_once", "when", "will"], "granitemoehybrid/modeling_granitemoehybrid.py:GraniteMoeHybridMambaLayer.cuda_kernels_forward": ["A", "A_log", "B", "C", "D", "False", "HybridMambaAttentionDynamicCache", "IntTensor", "LongTensor", "ModelMambaLayer", "Module", "None", "Tensor", "True", "_", "act", "activation", "and", "apply_mask_to_padding_states", "attention_mask", "batch_size", "bias", "cache_params", "cache_position", "causal_conv1d_fn", "causal_conv1d_update", "chunk_size", "class", "conv1d", "conv_dim", "conv_kernel_size", "conv_states", "copy_", "cuda_kernels_forward", "def", "dim", "dt", "dt_bias", "dt_limit", "dt_limit_kwargs", "dt_softplus", "dtype", "else", "exp", "expand", "float", "float32", "functional", "gate", "groups_time_state_size", "has_previous_state", "head_dim", "headdim", "hidden_states", "hidden_states_B_C", "hidden_states_B_C_transposed", "hidden_states_reshaped", "if", "in", "in_proj", "inf", "intermediate_size", "is", "layer_idx", "mamba_chunk_scan_combined", "mamba_split_conv1d_scan_combined", "n_groups", "ngroups", "nn", "norm", "norm_before_gate", "not", "num_heads", "out", "out_proj", "outproj_bias", "outproj_weight", "pad", "projected_states", "return", "return_final_states", "rmsnorm_eps", "rmsnorm_weight", "scan_output", "selective_state_update", "self", "seq_idx", "seq_len", "shape", "silu", "split", "squeeze", "ssm_state", "ssm_state_size", "ssm_states", "swish", "time_step_limit", "to", "torch", "training", "transpose", "use_precomputed_states", "variance_epsilon", "view", "weight", "x", "z"], "granitemoehybrid/modeling_granitemoehybrid.py:GraniteMoeHybridMambaLayer.torch_forward": ["A", "A_cumsum", "A_log", "B", "B_decay", "C", "C_reshaped", "C_times_states", "D", "D_residual", "G", "G_intermediate", "HybridMambaAttentionDynamicCache", "L", "LongTensor", "M", "M_intermediate", "ModelMambaLayer", "Module", "None", "Tensor", "Y_diag", "Y_off", "_", "act", "and", "apply_mask_to_padding_states", "attention_mask", "batch_size", "bias", "bmm", "cache_device", "cache_params", "cache_position", "cat", "chunk_size", "clamp", "class", "contextualized_states", "contiguous", "conv1d", "conv_dim", "conv_kernel_size", "conv_states", "copy_", "cumsum", "dA", "dB", "dBx", "decay_chunk", "decay_states", "def", "device", "dim", "dims", "dt", "dt_bias", "dtype", "else", "exp", "expand", "float", "float32", "for", "functional", "gate", "has_previous_state", "head_dim", "hidden_states", "hidden_states_B_C", "hidden_states_B_C_transposed", "if", "in", "in_proj", "input_states", "intermediate_size", "is", "layer_idx", "n_groups", "new_states", "nn", "norm", "not", "num_heads", "out_proj", "output_size", "pad", "pad_size", "pad_tensor_by_size", "permute", "previous_states", "projected_states", "repeat_interleave", "reshape", "reshape_into_chunks", "return", "roll", "scan_output", "segment_sum", "self", "seq_len", "shape", "shifts", "softplus", "split", "squeeze", "ssm_state", "ssm_state_size", "ssm_states", "ssm_states_reshaped", "state_decay_out", "state_decay_out_permuted", "states", "sum", "t", "time_step_limit", "to", "torch", "torch_forward", "transpose", "use_conv_bias", "use_precomputed_states", "view", "weight", "y", "zeros_like"], "granitemoehybrid/modeling_granitemoehybrid.py:GraniteMoeHybridMambaLayer.forward": ["HybridMambaAttentionDynamicCache", "IntTensor", "LongTensor", "ModelMambaLayer", "Module", "None", "NotImplementedError", "Please", "Tensor", "and", "attention_mask", "cache_params", "cache_position", "causal_conv1d", "class", "cuda", "cuda_kernels_forward", "def", "device", "dtype", "fast", "forward", "hidden_states", "if", "in", "in_proj", "install", "is", "is_fast_path_available", "kwargs", "mamba_ssm", "nn", "not", "path", "raise", "requires", "return", "self", "seq_idx", "shape", "support", "to", "torch", "torch_forward", "type", "weight"], "granitemoehybrid/modeling_granitemoehybrid.py:GraniteMoeHybridRMSNormGated.__init__": ["ModelRMSNormGated", "Module", "Parameter", "__init__", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "granitemoehybrid/modeling_granitemoehybrid.py:GraniteMoeHybridRMSNormGated.forward": ["ModelRMSNormGated", "Module", "None", "True", "class", "def", "dtype", "float32", "forward", "functional", "gate", "hidden_states", "if", "input_dtype", "is", "keepdim", "mean", "nn", "not", "pow", "return", "rsqrt", "self", "silu", "to", "torch", "variance", "variance_epsilon", "weight"], "granitemoehybrid/modeling_granitemoehybrid.py:GraniteMoeHybridMLP.__init__": ["ACT2FN", "False", "Linear", "ModelConfig", "ModelMLP", "Module", "__init__", "activation", "bias", "class", "config", "def", "hidden_act", "hidden_size", "input_linear", "input_size", "nn", "output_linear", "self", "shared_intermediate_size", "super"], "granitemoehybrid/modeling_granitemoehybrid.py:GraniteMoeHybridMLP.forward": ["ModelMLP", "Module", "Tensor", "activation", "chunk", "chunked_hidden_states", "class", "def", "dim", "forward", "hidden_states", "input_linear", "nn", "output_linear", "return", "self", "torch"], "granitemoehybrid/modeling_granitemoehybrid.py:GraniteMoeHybridRotaryEmbedding.__init__": ["Callable", "False", "ModelConfig", "ModelRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "granitemoehybrid/modeling_granitemoehybrid.py:GraniteMoeHybridRotaryEmbedding.compute_default_rope_parameters": ["ModelConfig", "ModelRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "granitemoehybrid/modeling_granitemoehybrid.py:GraniteMoeHybridRotaryEmbedding.forward": ["False", "ModelRotaryEmbedding", "Module", "None", "and", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "granitemoehybrid/modeling_granitemoehybrid.py:GraniteMoeHybridParallelExperts.__init__": ["ModelParallelExperts", "Module", "None", "Parameter", "__init__", "class", "def", "empty", "input_size", "int", "nn", "num_experts", "output_size", "self", "super", "torch", "weight"], "granitemoehybrid/modeling_granitemoehybrid.py:GraniteMoeHybridParallelExperts.forward": ["F", "ModelParallelExperts", "Module", "append", "cat", "class", "def", "dim", "expert_size", "for", "forward", "i", "in", "input_list", "inputs", "linear", "nn", "num_experts", "output_list", "range", "results", "return", "self", "split", "torch", "weight"], "granitemoehybrid/modeling_granitemoehybrid.py:GraniteMoeHybridTopKGating.__init__": ["False", "Linear", "ModelTopKGating", "Module", "__init__", "bias", "class", "def", "input_size", "int", "layer", "nn", "num_experts", "self", "super", "top_k"], "granitemoehybrid/modeling_granitemoehybrid.py:GraniteMoeHybridTopKGating.forward": ["ModelTopKGating", "Module", "_", "batch_gates", "batch_index", "class", "def", "device", "dim", "div", "dtype", "expert_size", "flatten", "float", "forward", "gates", "hidden_states", "index_sorted_experts", "layer", "logits", "long", "nn", "num_experts", "return", "rounding_mode", "scatter", "self", "size", "softmax", "sort", "sum", "tolist", "top_k", "top_k_experts", "top_k_gates", "top_k_indices", "top_k_logits", "topk", "torch", "trunc", "type_as", "zeros"], "granitemoehybrid/modeling_granitemoehybrid.py:GraniteMoeHybridMoE.__init__": ["ACT2FN", "ModelConfig", "ModelMoE", "ModelParallelExperts", "ModelTopKGating", "Module", "__init__", "activation", "class", "config", "def", "hidden_act", "hidden_size", "input_linear", "input_size", "intermediate_size", "nn", "num_experts", "num_experts_per_tok", "num_local_experts", "output_linear", "router", "self", "super", "top_k"], "granitemoehybrid/modeling_granitemoehybrid.py:GraniteMoeHybridMoE.forward": ["ModelMoE", "Module", "None", "_", "activation", "batch_gates", "batch_index", "bsz", "chunk", "chunked_hidden_states", "class", "def", "device", "dim", "dtype", "emb_size", "expert_inputs", "expert_outputs", "expert_size", "forward", "hidden_states", "index_add", "input_linear", "input_size", "layer_input", "layer_output", "length", "nn", "output_linear", "reshape", "return", "router", "self", "size", "torch", "view", "zeros"], "granitemoehybrid/modeling_granitemoehybrid.py:GraniteMoeHybridRMSNorm.__init__": ["ModelRMSNorm", "Module", "Parameter", "__init__", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "granitemoehybrid/modeling_granitemoehybrid.py:GraniteMoeHybridRMSNorm.forward": ["ModelRMSNorm", "Module", "True", "class", "def", "dtype", "float32", "forward", "hidden_states", "input_dtype", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "granitemoehybrid/modeling_granitemoehybrid.py:GraniteMoeHybridRMSNorm.extra_repr": ["ModelRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "variance_epsilon", "weight"], "granitemoehybrid/modeling_granitemoehybrid.py:GraniteMoeHybridDecoderLayer.__init__": ["GradientCheckpointingLayer", "ModelAttention", "ModelConfig", "ModelDecoderLayer", "ModelMLP", "ModelMambaLayer", "ModelMoE", "ModelRMSNorm", "None", "__init__", "block_sparse_moe", "class", "config", "def", "else", "eps", "getattr", "has_experts", "hidden_size", "if", "input_layernorm", "int", "layer_idx", "layer_type", "layers_block_type", "mamba", "num_local_experts", "post_attention_layernorm", "residual_multiplier", "rms_norm_eps", "self", "self_attn", "shared_mlp", "super"], "granitemoehybrid/modeling_granitemoehybrid.py:GraniteMoeHybridDecoderLayer.forward": ["Cache", "False", "FloatTensor", "GradientCheckpointingLayer", "LongTensor", "ModelDecoderLayer", "ModelFlashAttentionKwargs", "None", "Tensor", "Unpack", "_", "attention_mask", "block_sparse_moe", "bool", "cache_params", "cache_position", "class", "def", "else", "forward", "has_experts", "hidden_states", "if", "input_layernorm", "is", "kwargs", "mamba", "moe_hidden_states", "not", "past_key_values", "position_embeddings", "post_attention_layernorm", "residual", "residual_multiplier", "return", "self", "self_attn", "shared_mlp", "torch", "tuple", "use_cache"], "granitemoehybrid/modeling_granitemoehybrid.py:GraniteMoeHybridPreTrainedModel._init_weights": ["A_log", "D", "ModelMambaLayer", "ModelParallelExperts", "ModelPreTrainedModel", "ModelRMSNormGated", "PreTrainedModel", "_init_weights", "arange", "class", "config", "copy_", "def", "dt_bias", "elif", "if", "init", "initializer_range", "isinstance", "log", "mean", "module", "normal_", "num_heads", "ones_", "self", "std", "super", "torch", "weight"], "granitemoehybrid/modeling_granitemoehybrid.py:GraniteMoeHybridModel.__init__": ["Embedding", "False", "ModelConfig", "ModelDecoderLayer", "ModelModel", "ModelPreTrainedModel", "ModelRMSNorm", "ModelRotaryEmbedding", "ModuleList", "None", "__init__", "class", "config", "def", "else", "embed_tokens", "embedding_multiplier", "eps", "for", "gradient_checkpointing", "hidden_size", "if", "in", "layer_idx", "layers", "nn", "norm", "num_hidden_layers", "pad_token_id", "padding_idx", "position_embedding_type", "post_init", "range", "rms_norm_eps", "rope", "rotary_emb", "self", "super", "vocab_size"], "granitemoehybrid/modeling_granitemoehybrid.py:GraniteMoeHybridModel.forward": ["BaseModelOutputWithPast", "Cache", "FloatTensor", "LongTensor", "ModelFlashAttentionKwargs", "ModelModel", "ModelPreTrainedModel", "MoeModelOutputWithPast", "None", "Tensor", "True", "Unpack", "ValueError", "You", "_update_mamba_mask", "and", "arange", "attention_mask", "bool", "cache_position", "causal_mask", "class", "config", "create_causal_mask", "decoder_layer", "def", "device", "else", "embed_tokens", "embedding_multiplier", "exactly", "for", "forward", "get_seq_length", "has_previous_state", "hidden_states", "if", "in", "input_ids", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layer_mask", "layer_type", "layers", "mamba", "mamba_mask", "must", "norm", "not", "of", "one", "or", "past_key_values", "past_seen_tokens", "position_embeddings", "position_ids", "raise", "return", "rotary_emb", "self", "shape", "specify", "torch", "tuple", "unsqueeze", "use_cache"], "granitemoehybrid/modeling_granitemoehybrid.py:GraniteMoeHybridModel._update_mamba_mask": ["ModelModel", "ModelPreTrainedModel", "None", "_update_mamba_mask", "all", "and", "attention_mask", "cache_position", "class", "def", "if", "is", "mamba_mask", "not", "or", "return", "self", "torch"], "granitemoehybrid/modeling_granitemoehybrid.py:load_balancing_loss_func": ["Model_balancing_loss_func", "None", "Tensor", "_", "attention_mask", "batch_size", "cat", "compute_device", "concatenated_gate_logits", "def", "device", "dim", "else", "expand", "expert_attention_mask", "expert_mask", "float", "for", "functional", "gate_logits", "if", "in", "int", "is", "isinstance", "layer_gate", "mean", "nn", "not", "num_experts", "num_hidden_layers", "one_hot", "or", "overall_loss", "r", "reshape", "return", "router_per_expert_attention_mask", "router_prob_per_expert", "routing_weights", "selected_experts", "sequence_length", "shape", "softmax", "sum", "to", "tokens_per_expert", "top_k", "topk", "torch", "tuple", "unsqueeze"], "granitemoehybrid/modeling_granitemoehybrid.py:GraniteMoeHybridForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "ModelConfig", "ModelForCausalLM", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "logits_scaling", "model", "nn", "num_experts", "num_experts_per_tok", "num_local_experts", "post_init", "router_aux_loss_coef", "self", "super", "vocab_size"], "granitemoehybrid/modeling_granitemoehybrid.py:GraniteMoeHybridForCausalLM.forward": ["Cache", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForCausalLM", "ModelPreTrainedModel", "MoeCausalLMOutputWithPast", "None", "Tensor", "attention_mask", "attentions", "aux_loss", "bool", "cache_position", "class", "config", "def", "device", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "load_balancing_loss_func", "logits", "logits_scaling", "logits_to_keep", "loss", "loss_function", "model", "not", "num_experts", "num_experts_per_tok", "output_router_logits", "outputs", "past_key_values", "position_ids", "r", "return", "router_aux_loss_coef", "router_logits", "self", "slice", "slice_indices", "to", "torch", "tuple", "vocab_size"], "granitemoehybrid/modeling_granitemoehybrid.py:GraniteMoeHybridForCausalLM.prepare_inputs_for_generation": ["False", "GenerationMixin", "HybridMambaAttentionDynamicCache", "ModelForCausalLM", "ModelPreTrainedModel", "None", "True", "and", "attention_mask", "cache_position", "class", "config", "contiguous", "cumsum", "def", "device", "dtype", "elif", "else", "empty_past_kv", "for", "if", "in", "input_ids", "inputs_embeds", "is", "is_first_iteration", "items", "key", "kwargs", "long", "masked_fill_", "model_inputs", "not", "or", "past_key_values", "position_ids", "prepare_inputs_for_generation", "return", "self", "shape", "update", "use_cache", "value"], "granitemoeshared/modeling_granitemoeshared.py:GraniteMoeSharedMLP.__init__": ["ACT2FN", "False", "Linear", "ModelConfig", "ModelMLP", "Module", "__init__", "activation", "bias", "class", "config", "def", "hidden_act", "hidden_size", "input_linear", "input_size", "nn", "output_linear", "self", "shared_intermediate_size", "super"], "granitemoeshared/modeling_granitemoeshared.py:GraniteMoeSharedMLP.forward": ["ModelMLP", "Module", "Tensor", "activation", "chunk", "chunked_hidden_states", "class", "def", "dim", "forward", "hidden_states", "input_linear", "nn", "output_linear", "return", "self", "torch"], "granitemoeshared/modeling_granitemoeshared.py:GraniteMoeSharedRMSNorm.__init__": ["ModelRMSNorm", "Module", "Parameter", "__init__", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "granitemoeshared/modeling_granitemoeshared.py:GraniteMoeSharedRMSNorm.forward": ["ModelRMSNorm", "Module", "True", "class", "def", "dtype", "float32", "forward", "hidden_states", "input_dtype", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "granitemoeshared/modeling_granitemoeshared.py:GraniteMoeSharedRMSNorm.extra_repr": ["ModelRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "variance_epsilon", "weight"], "granitemoeshared/modeling_granitemoeshared.py:GraniteMoeSharedParallelExperts.__init__": ["ModelParallelExperts", "Module", "None", "Parameter", "__init__", "class", "def", "empty", "input_size", "int", "nn", "num_experts", "output_size", "self", "super", "torch", "weight"], "granitemoeshared/modeling_granitemoeshared.py:GraniteMoeSharedParallelExperts.forward": ["F", "ModelParallelExperts", "Module", "append", "cat", "class", "def", "dim", "expert_size", "for", "forward", "i", "in", "input_list", "inputs", "linear", "nn", "num_experts", "output_list", "range", "results", "return", "self", "split", "torch", "weight"], "granitemoeshared/modeling_granitemoeshared.py:GraniteMoeSharedTopKGating.__init__": ["False", "Linear", "ModelTopKGating", "Module", "__init__", "bias", "class", "def", "input_size", "int", "layer", "nn", "num_experts", "self", "super", "top_k"], "granitemoeshared/modeling_granitemoeshared.py:GraniteMoeSharedTopKGating.forward": ["ModelTopKGating", "Module", "_", "batch_gates", "batch_index", "class", "def", "device", "dim", "div", "dtype", "expert_size", "flatten", "float", "forward", "gates", "hidden_states", "index_sorted_experts", "layer", "logits", "long", "nn", "num_experts", "return", "rounding_mode", "scatter", "self", "size", "softmax", "sort", "sum", "tolist", "top_k", "top_k_experts", "top_k_gates", "top_k_indices", "top_k_logits", "topk", "torch", "trunc", "type_as", "zeros"], "granitemoeshared/modeling_granitemoeshared.py:GraniteMoeSharedMoE.__init__": ["ACT2FN", "ModelConfig", "ModelMoE", "ModelParallelExperts", "ModelTopKGating", "Module", "__init__", "activation", "class", "config", "def", "hidden_act", "hidden_size", "input_linear", "input_size", "intermediate_size", "nn", "num_experts", "num_experts_per_tok", "num_local_experts", "output_linear", "router", "self", "super", "top_k"], "granitemoeshared/modeling_granitemoeshared.py:GraniteMoeSharedMoE.forward": ["ModelMoE", "Module", "None", "_", "activation", "batch_gates", "batch_index", "bsz", "chunk", "chunked_hidden_states", "class", "def", "device", "dim", "dtype", "emb_size", "expert_inputs", "expert_outputs", "expert_size", "forward", "hidden_states", "index_add", "input_linear", "input_size", "layer_input", "layer_output", "length", "nn", "output_linear", "reshape", "return", "router", "self", "size", "torch", "view", "zeros"], "granitemoeshared/modeling_granitemoeshared.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "granitemoeshared/modeling_granitemoeshared.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cos", "def", "k", "k_embed", "q", "q_embed", "return", "rotate_half", "sin", "unsqueeze", "unsqueeze_dim"], "granitemoeshared/modeling_granitemoeshared.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "granitemoeshared/modeling_granitemoeshared.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value", "value_states"], "granitemoeshared/modeling_granitemoeshared.py:GraniteMoeSharedAttention.__init__": ["Linear", "ModelAttention", "ModelConfig", "Module", "True", "__init__", "attention_bias", "attention_dropout", "attention_multiplier", "bias", "class", "config", "def", "getattr", "head_dim", "hidden_size", "int", "is_causal", "k_proj", "layer_idx", "nn", "num_attention_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "q_proj", "scaling", "self", "super", "v_proj"], "granitemoeshared/modeling_granitemoeshared.py:GraniteMoeSharedAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "LongTensor", "ModelAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_attn_implementation", "apply_rotary_pos_emb", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "cache_kwargs", "cache_position", "class", "config", "contiguous", "cos", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "head_dim", "hidden_shape", "hidden_states", "if", "input_shape", "is", "k_proj", "key_states", "kwargs", "layer_idx", "nn", "not", "o_proj", "past_key_values", "position_embeddings", "q_proj", "query_states", "reshape", "return", "scaling", "self", "shape", "sin", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "view"], "granitemoeshared/modeling_granitemoeshared.py:GraniteMoeSharedDecoderLayer.__init__": ["GradientCheckpointingLayer", "ModelAttention", "ModelConfig", "ModelDecoderLayer", "ModelMLP", "ModelMoE", "ModelRMSNorm", "None", "__init__", "block_sparse_moe", "class", "config", "def", "else", "eps", "hidden_size", "if", "input_layernorm", "int", "layer_idx", "post_attention_layernorm", "residual_multiplier", "rms_norm_eps", "self", "self_attn", "shared_intermediate_size", "shared_mlp", "super"], "granitemoeshared/modeling_granitemoeshared.py:GraniteMoeSharedDecoderLayer.forward": ["Cache", "False", "FloatTensor", "GradientCheckpointingLayer", "LongTensor", "ModelDecoderLayer", "ModelFlashAttentionKwargs", "None", "Tensor", "Unpack", "_", "attention_mask", "block_sparse_moe", "bool", "cache_position", "class", "def", "else", "forward", "hidden_states", "if", "input_layernorm", "is", "kwargs", "moe_hidden_states", "output_attentions", "past_key_values", "position_embeddings", "position_ids", "post_attention_layernorm", "residual", "residual_multiplier", "return", "self", "self_attn", "shared_mlp", "torch", "tuple", "use_cache"], "granitemoeshared/modeling_granitemoeshared.py:GraniteMoeSharedPreTrainedModel._init_weights": ["ModelParallelExperts", "ModelPreTrainedModel", "PreTrainedModel", "_init_weights", "class", "config", "def", "if", "init", "initializer_range", "isinstance", "mean", "module", "normal_", "self", "std", "super", "weight"], "granitemoeshared/modeling_granitemoeshared.py:GraniteMoeSharedRotaryEmbedding.__init__": ["Callable", "False", "ModelConfig", "ModelRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "granitemoeshared/modeling_granitemoeshared.py:GraniteMoeSharedRotaryEmbedding.compute_default_rope_parameters": ["ModelConfig", "ModelRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "granitemoeshared/modeling_granitemoeshared.py:GraniteMoeSharedRotaryEmbedding.forward": ["False", "ModelRotaryEmbedding", "Module", "None", "and", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "granitemoeshared/modeling_granitemoeshared.py:GraniteMoeSharedModel.__init__": ["Embedding", "False", "ModelConfig", "ModelDecoderLayer", "ModelModel", "ModelPreTrainedModel", "ModelRMSNorm", "ModelRotaryEmbedding", "ModuleList", "__init__", "class", "config", "def", "embed_tokens", "embedding_multiplier", "eps", "for", "gradient_checkpointing", "hidden_size", "in", "layer_idx", "layers", "nn", "norm", "num_hidden_layers", "pad_token_id", "padding_idx", "post_init", "range", "rms_norm_eps", "rotary_emb", "self", "super", "vocab_size"], "granitemoeshared/modeling_granitemoeshared.py:GraniteMoeSharedModel.forward": ["Cache", "DynamicCache", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "MoeModelOutputWithPast", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "and", "arange", "attention_mask", "bool", "cache_position", "causal_mask", "class", "config", "create_causal_mask", "decoder_layer", "def", "device", "else", "embed_tokens", "embedding_multiplier", "exactly", "for", "forward", "get_seq_length", "hidden_states", "if", "in", "input_embeds", "input_ids", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layers", "must", "norm", "not", "num_hidden_layers", "of", "one", "or", "past_key_values", "past_seen_tokens", "position_embeddings", "position_ids", "raise", "return", "rotary_emb", "self", "shape", "specify", "torch", "unsqueeze", "use_cache"], "granitemoeshared/modeling_granitemoeshared.py:load_balancing_loss_func": ["Model_balancing_loss_func", "None", "Tensor", "_", "attention_mask", "batch_size", "cat", "compute_device", "concatenated_gate_logits", "def", "device", "dim", "else", "expand", "expert_attention_mask", "expert_mask", "float", "for", "functional", "gate_logits", "if", "in", "int", "is", "isinstance", "layer_gate", "mean", "nn", "not", "num_experts", "num_hidden_layers", "one_hot", "or", "overall_loss", "r", "reshape", "return", "router_per_expert_attention_mask", "router_prob_per_expert", "routing_weights", "selected_experts", "sequence_length", "shape", "softmax", "sum", "to", "tokens_per_expert", "top_k", "topk", "torch", "tuple", "unsqueeze"], "granitemoeshared/modeling_granitemoeshared.py:GraniteMoeSharedForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "ModelConfig", "ModelForCausalLM", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "logits_scaling", "model", "nn", "num_experts", "num_experts_per_tok", "num_local_experts", "post_init", "router_aux_loss_coef", "self", "super", "vocab_size"], "granitemoeshared/modeling_granitemoeshared.py:GraniteMoeSharedForCausalLM.forward": ["Cache", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForCausalLM", "ModelPreTrainedModel", "MoeCausalLMOutputWithPast", "None", "Tensor", "attention_mask", "attentions", "aux_loss", "bool", "cache_position", "class", "config", "def", "device", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "load_balancing_loss_func", "logits", "logits_scaling", "logits_to_keep", "loss", "loss_function", "model", "not", "num_experts", "num_experts_per_tok", "output_router_logits", "outputs", "past_key_values", "position_ids", "r", "return", "router_aux_loss_coef", "router_logits", "self", "slice", "slice_indices", "to", "torch", "tuple", "vocab_size"], "grounding_dino/modeling_grounding_dino.py:MultiScaleDeformableAttention.forward": ["False", "ModelScaleDeformableAttention", "Module", "Tensor", "_", "align_corners", "append", "attention_weights", "batch_size", "bilinear", "class", "contiguous", "def", "dim", "enumerate", "flatten", "for", "forward", "functional", "grid_sample", "height", "hidden_dim", "im2col_step", "in", "int", "level_id", "level_start_index", "list", "mode", "nn", "num_heads", "num_levels", "num_points", "num_queries", "output", "padding_mode", "reshape", "return", "sampling_grid_l_", "sampling_grids", "sampling_locations", "sampling_value_l_", "sampling_value_list", "self", "shape", "split", "stack", "sum", "torch", "transpose", "tuple", "value", "value_l_", "value_list", "value_spatial_shapes", "value_spatial_shapes_list", "view", "width", "zeros"], "grounding_dino/modeling_grounding_dino.py:GroundingDinoFrozenBatchNorm2d.__init__": ["ModelFrozenBatchNorm2d", "Module", "__init__", "bias", "class", "def", "n", "nn", "ones", "register_buffer", "running_mean", "running_var", "self", "super", "torch", "weight", "zeros"], "grounding_dino/modeling_grounding_dino.py:GroundingDinoFrozenBatchNorm2d._load_from_state_dict": ["ModelFrozenBatchNorm2d", "Module", "_load_from_state_dict", "class", "def", "del", "error_msgs", "if", "in", "local_metadata", "missing_keys", "nn", "num_batches_tracked", "num_batches_tracked_key", "prefix", "self", "state_dict", "strict", "super", "unexpected_keys"], "grounding_dino/modeling_grounding_dino.py:GroundingDinoFrozenBatchNorm2d.forward": ["ModelFrozenBatchNorm2d", "Module", "bias", "class", "def", "epsilon", "forward", "nn", "reshape", "return", "rsqrt", "running_mean", "running_var", "scale", "self", "weight", "x"], "grounding_dino/modeling_grounding_dino.py:replace_batch_norm": ["BatchNorm2d", "ModelFrozenBatchNorm2d", "Model_batch_norm", "_modules", "bias", "children", "copy_", "def", "device", "for", "if", "in", "isinstance", "len", "list", "meta", "model", "module", "name", "named_children", "new_module", "nn", "num_features", "r", "running_mean", "running_var", "torch", "weight"], "grounding_dino/modeling_grounding_dino.py:GroundingDinoConvEncoder.__init__": ["Either", "False", "ModelConvEncoder", "Module", "None", "True", "ValueError", "__init__", "and", "backbone", "backbone_config", "backbone_kwargs", "backbone_model_type", "be", "channels", "class", "config", "create_model", "def", "elif", "else", "feature_info", "features_only", "for", "if", "in", "intermediate_channel_sizes", "is", "layer2", "layer3", "layer4", "load_backbone", "model", "model_type", "name", "named_parameters", "nn", "no_grad", "not", "or", "parameter", "pretrained", "provided", "raise", "replace_batch_norm", "requires_backends", "requires_grad_", "resnet", "self", "should", "stage", "super", "the", "timm", "torch", "use_pretrained_backbone", "use_timm_backbone", "with"], "grounding_dino/modeling_grounding_dino.py:GroundingDinoConvEncoder.forward": ["ModelConvEncoder", "Module", "None", "Tensor", "append", "bool", "class", "config", "def", "else", "feature_map", "feature_maps", "features", "float", "for", "forward", "functional", "if", "in", "interpolate", "mask", "model", "nn", "out", "pixel_mask", "pixel_values", "return", "self", "shape", "size", "to", "torch", "use_timm_backbone"], "grounding_dino/modeling_grounding_dino.py:GroundingDinoConvModel.__init__": ["ModelConvModel", "Module", "__init__", "class", "conv_encoder", "def", "nn", "position_embedding", "self", "super"], "grounding_dino/modeling_grounding_dino.py:GroundingDinoConvModel.forward": ["ModelConvModel", "Module", "append", "class", "conv_encoder", "def", "dtype", "feature_map", "for", "forward", "in", "mask", "nn", "out", "pixel_mask", "pixel_values", "pos", "position_embedding", "return", "self", "to"], "grounding_dino/modeling_grounding_dino.py:GroundingDinoSinePositionEmbedding.__init__": ["ModelSinePositionEmbedding", "Module", "__init__", "class", "config", "d_model", "def", "embedding_dim", "math", "nn", "pi", "positional_embedding_temperature", "scale", "self", "super", "temperature"], "grounding_dino/modeling_grounding_dino.py:GroundingDinoSinePositionEmbedding.forward": ["ModelSinePositionEmbedding", "Module", "None", "arange", "cat", "class", "cos", "cumsum", "def", "device", "dim", "dim_t", "div", "dtype", "embedding_dim", "eps", "flatten", "float32", "floor", "forward", "nn", "permute", "pixel_mask", "pixel_values", "pos", "pos_x", "pos_y", "return", "rounding_mode", "scale", "self", "sin", "stack", "temperature", "torch", "x_embed", "y_embed"], "grounding_dino/modeling_grounding_dino.py:GroundingDinoLearnedPositionEmbedding.__init__": ["Embedding", "ModelLearnedPositionEmbedding", "Module", "__init__", "class", "column_embeddings", "config", "d_model", "def", "embedding_dim", "nn", "row_embeddings", "self", "super"], "grounding_dino/modeling_grounding_dino.py:GroundingDinoLearnedPositionEmbedding.forward": ["ModelLearnedPositionEmbedding", "Module", "None", "arange", "cat", "class", "column_embeddings", "def", "device", "dim", "forward", "height", "height_values", "nn", "permute", "pixel_mask", "pixel_values", "pos", "repeat", "return", "row_embeddings", "self", "shape", "torch", "unsqueeze", "width", "width_values", "x_emb", "y_emb"], "grounding_dino/modeling_grounding_dino.py:build_position_encoding": ["ModelLearnedPositionEmbedding", "ModelSinePositionEmbedding", "Model_position_encoding", "Not", "ValueError", "config", "def", "elif", "else", "f", "if", "learned", "position_embedding", "position_embedding_type", "raise", "return", "sine", "supported"], "grounding_dino/modeling_grounding_dino.py:GroundingDinoMultiscaleDeformableAttention.__init__": ["CUDA", "Linear", "ModelConfig", "ModelMultiscaleDeformableAttention", "Module", "MultiScaleDeformableAttention", "ValueError", "You", "__init__", "a", "and", "attention", "attention_weights", "attn", "authors", "be", "better", "but", "by", "class", "config", "d", "d_model", "def", "dim_per_head", "dimension", "disable_custom_kernels", "divisible", "each", "efficient", "embed_dim", "f", "got", "head", "if", "im2col_step", "implementation", "in", "int", "is", "make", "more", "must", "n_heads", "n_levels", "n_points", "nn", "not", "num_feature_levels", "num_heads", "of", "output_proj", "power", "raise", "sampling_offsets", "self", "set", "super", "the", "to", "value_proj", "warn", "warnings", "which"], "grounding_dino/modeling_grounding_dino.py:GroundingDinoMultiscaleDeformableAttention.with_pos_embed": ["ModelMultiscaleDeformableAttention", "Module", "None", "Tensor", "class", "def", "else", "if", "is", "nn", "position_embeddings", "return", "self", "tensor", "torch", "with_pos_embed"], "grounding_dino/modeling_grounding_dino.py:GroundingDinoMultiscaleDeformableAttention.forward": ["F", "False", "Last", "Make", "ModelMultiscaleDeformableAttention", "Module", "None", "Tensor", "ValueError", "_", "align", "attention_mask", "attention_weights", "attn", "batch_size", "be", "bool", "but", "class", "d_model", "def", "dim", "elif", "else", "encoder", "encoder_attention_mask", "encoder_hidden_states", "f", "float", "forward", "got", "hidden", "hidden_states", "if", "im2col_step", "is", "length", "level_start_index", "masked_fill", "must", "n_heads", "n_levels", "n_points", "nn", "not", "num_coordinates", "num_queries", "of", "offset_normalizer", "or", "output", "output_attentions", "output_proj", "position_embeddings", "raise", "reference_points", "return", "sampling_locations", "sampling_offsets", "self", "sequence", "sequence_length", "shape", "shapes", "softmax", "spatial", "spatial_shapes", "spatial_shapes_list", "stack", "states", "sum", "sure", "the", "to", "torch", "value", "value_proj", "view", "with", "with_pos_embed"], "grounding_dino/modeling_grounding_dino.py:GroundingDinoTextEnhancerLayer.__init__": ["ACT2FN", "LayerNorm", "Linear", "ModelMultiheadAttention", "ModelTextEnhancerLayer", "Module", "__init__", "activation", "activation_function", "class", "config", "d_model", "def", "dropout", "encoder_attention_heads", "encoder_ffn_dim", "fc1", "fc2", "layer_norm_after", "layer_norm_before", "layer_norm_eps", "nn", "num_attention_heads", "num_heads", "self", "self_attn", "super", "text_enhancer_dropout"], "grounding_dino/modeling_grounding_dino.py:GroundingDinoTextEnhancerLayer.with_pos_embed": ["ModelTextEnhancerLayer", "Module", "None", "Tensor", "class", "def", "else", "hidden_state", "if", "is", "nn", "position_embeddings", "return", "self", "with_pos_embed"], "grounding_dino/modeling_grounding_dino.py:GroundingDinoTextEnhancerLayer.forward": ["BoolTensor", "FloatTensor", "ModelTextEnhancerLayer", "Module", "None", "True", "activation", "and", "attention_mask", "attention_masks", "attention_output", "attention_weights", "class", "def", "dim", "dropout", "dtype", "fc1", "fc2", "finfo", "forward", "functional", "hidden_states", "if", "keys", "layer_norm_after", "layer_norm_before", "min", "nn", "num_heads", "output_attentions", "p", "position_embeddings", "queries", "repeat", "residual", "return", "self", "self_attn", "shape", "to", "torch", "training", "tuple", "values", "with_pos_embed"], "grounding_dino/modeling_grounding_dino.py:GroundingDinoBiMultiHeadAttention.__init__": ["Linear", "ModelBiMultiHeadAttention", "Module", "ValueError", "__init__", "and", "be", "by", "class", "config", "d_model", "def", "divisible", "dropout", "embed_dim", "encoder_attention_heads", "encoder_ffn_dim", "f", "fusion_dropout", "got", "head_dim", "if", "must", "nn", "num_heads", "out_text_proj", "out_vision_proj", "raise", "scale", "self", "super", "text_dim", "text_proj", "values_text_proj", "values_vision_proj", "vision_dim", "vision_proj"], "grounding_dino/modeling_grounding_dino.py:GroundingDinoBiMultiHeadAttention._reshape": ["ModelBiMultiHeadAttention", "Module", "Tensor", "_reshape", "batch_size", "class", "contiguous", "def", "head_dim", "int", "nn", "num_heads", "return", "self", "seq_len", "tensor", "torch", "transpose", "view"], "grounding_dino/modeling_grounding_dino.py:GroundingDinoBiMultiHeadAttention.forward": ["Attention", "BoolTensor", "F", "FloatTensor", "ModelBiMultiHeadAttention", "Module", "None", "True", "ValueError", "_", "_reshape", "attn_weights", "attn_weights_transposed", "batch_size", "be", "bmm", "but", "clamp", "class", "def", "dim", "dropout", "embed_dim", "f", "flatten", "float", "forward", "head_dim", "if", "inf", "is", "keepdim", "masked_fill_", "max", "min", "nn", "not", "num_heads", "of", "out_text_proj", "out_vision_proj", "p", "proj_shape", "raise", "repeat", "reshape", "return", "scale", "self", "should", "size", "softmax", "src_len", "text_attention_mask", "text_attn_output", "text_attn_probs", "text_attn_weights", "text_features", "text_key_states", "text_proj", "text_value_states", "tgt_len", "torch", "training", "transpose", "tuple", "values_text_proj", "values_vision_proj", "view", "vision_attention_mask", "vision_attn_output", "vision_attn_probs", "vision_attn_weights", "vision_features", "vision_proj", "vision_query_states", "vision_value_states", "weights"], "grounding_dino/modeling_grounding_dino.py:drop_path": ["False", "Model_path", "Model_prob", "Tensor", "bool", "def", "device", "div", "dtype", "float", "floor_", "if", "input", "keep_prob", "ndim", "not", "or", "output", "rand", "random_tensor", "return", "shape", "torch", "training"], "grounding_dino/modeling_grounding_dino.py:GroundingDinoDropPath.__init__": ["ModelDropPath", "Module", "None", "__init__", "class", "def", "drop_prob", "float", "nn", "self", "super"], "grounding_dino/modeling_grounding_dino.py:GroundingDinoDropPath.forward": ["ModelDropPath", "Module", "Tensor", "class", "def", "drop_path", "drop_prob", "forward", "hidden_states", "nn", "return", "self", "torch", "training"], "grounding_dino/modeling_grounding_dino.py:GroundingDinoDropPath.extra_repr": ["ModelDropPath", "Module", "class", "def", "drop_prob", "extra_repr", "f", "nn", "p", "return", "self", "str"], "grounding_dino/modeling_grounding_dino.py:GroundingDinoFusionLayer.__init__": ["Identity", "LayerNorm", "ModelBiMultiHeadAttention", "ModelDropPath", "ModelFusionLayer", "Module", "Parameter", "True", "__init__", "attn", "class", "config", "d_model", "def", "drop_path", "else", "fusion_droppath", "if", "init_values", "layer_norm_eps", "layer_norm_text", "layer_norm_vision", "nn", "ones", "requires_grad", "self", "super", "text_param", "torch", "vision_param"], "grounding_dino/modeling_grounding_dino.py:GroundingDinoFusionLayer.forward": ["BoolTensor", "FloatTensor", "ModelFusionLayer", "Module", "None", "attention_mask_text", "attention_mask_vision", "attn", "class", "def", "delta_t", "delta_v", "drop_path", "forward", "layer_norm_text", "layer_norm_vision", "nn", "return", "self", "text_attention_mask", "text_attn", "text_features", "text_param", "torch", "tuple", "vision_attention_mask", "vision_attn", "vision_features", "vision_param"], "grounding_dino/modeling_grounding_dino.py:GroundingDinoDeformableLayer.__init__": ["ACT2FN", "LayerNorm", "Linear", "ModelConfig", "ModelDeformableLayer", "ModelMultiscaleDeformableAttention", "Module", "__init__", "activation_dropout", "activation_fn", "activation_function", "class", "config", "d_model", "def", "dropout", "embed_dim", "encoder_attention_heads", "encoder_ffn_dim", "encoder_n_points", "fc1", "fc2", "final_layer_norm", "layer_norm_eps", "n_points", "nn", "num_heads", "self", "self_attn", "self_attn_layer_norm", "super"], "grounding_dino/modeling_grounding_dino.py:GroundingDinoDeformableLayer.forward": ["False", "ModelDeformableLayer", "Module", "None", "Tensor", "activation_dropout", "activation_fn", "any", "attention_mask", "attn_weights", "bool", "clamp", "clamp_value", "class", "def", "dropout", "dtype", "encoder_attention_mask", "encoder_hidden_states", "fc1", "fc2", "final_layer_norm", "finfo", "forward", "functional", "hidden_states", "if", "isinf", "isnan", "level_start_index", "max", "min", "nn", "or", "output_attentions", "p", "position_embeddings", "reference_points", "residual", "return", "self", "self_attn", "self_attn_layer_norm", "spatial_shapes", "spatial_shapes_list", "torch", "training"], "grounding_dino/modeling_grounding_dino.py:get_sine_pos_embed": ["Model_sine_pos_embed", "Tensor", "True", "arange", "bool", "cat", "cos", "def", "device", "dim", "dim_t", "div", "dtype", "exchange_xy", "flatten", "float32", "floor", "for", "if", "in", "int", "math", "num_pos_feats", "pi", "pos_tensor", "position_embeddings", "return", "rounding_mode", "scale", "shape", "sin", "sin_x", "sine_func", "split", "stack", "temperature", "torch", "x"], "grounding_dino/modeling_grounding_dino.py:GroundingDinoEncoderLayer.__init__": ["ModelDeformableLayer", "ModelEncoderLayer", "ModelFusionLayer", "ModelTextEnhancerLayer", "Module", "None", "__init__", "class", "config", "d_model", "def", "deformable_layer", "fusion_layer", "nn", "self", "super", "text_enhancer_layer"], "grounding_dino/modeling_grounding_dino.py:GroundingDinoEncoderLayer.get_text_position_embeddings": ["False", "ModelEncoderLayer", "Module", "None", "Tensor", "_", "and", "arange", "batch_size", "class", "d_model", "def", "device", "exchange_xy", "float", "get_sine_pos_embed", "get_text_position_embeddings", "if", "is", "nn", "not", "num_pos_feats", "repeat", "return", "self", "seq_length", "shape", "text_features", "text_position_embedding", "text_position_ids", "torch", "unsqueeze"], "grounding_dino/modeling_grounding_dino.py:GroundingDinoEncoderLayer.forward": ["ModelEncoderLayer", "Module", "None", "Tensor", "attention_mask", "attention_mask_text", "attention_mask_vision", "attention_masks", "class", "def", "deformable_layer", "else", "forward", "fusion_layer", "get_text_position_embeddings", "hidden_states", "if", "int", "is", "key_padding_mask", "level_start_index", "list", "nn", "not", "position_embeddings", "reference_points", "return", "self", "spatial_shapes", "spatial_shapes_list", "text_attention_mask", "text_enhanced_attn", "text_enhancer_layer", "text_features", "text_fused_attn", "text_position_embedding", "text_position_ids", "text_self_attention_masks", "tuple", "vision_deformable_attn", "vision_features", "vision_fused_attn", "vision_position_embedding"], "grounding_dino/modeling_grounding_dino.py:GroundingDinoMultiheadAttention.__init__": ["Dropout", "Linear", "ModelMultiheadAttention", "Module", "None", "The", "ValueError", "__init__", "a", "all_head_size", "and", "attention", "attention_dropout", "attention_head_size", "class", "config", "def", "dropout", "embedding_size", "f", "hasattr", "heads", "hidden", "hidden_size", "if", "int", "is", "key", "multiple", "nn", "not", "num_attention_heads", "number", "of", "out_proj", "query", "raise", "self", "size", "super", "the", "value"], "grounding_dino/modeling_grounding_dino.py:GroundingDinoMultiheadAttention.forward": ["False", "FloatTensor", "ModelMultiheadAttention", "Module", "None", "Tensor", "_", "all_head_size", "attention_head_size", "attention_mask", "attention_probs", "attention_scores", "batch_size", "bool", "class", "context_layer", "contiguous", "def", "dim", "dropout", "else", "forward", "functional", "if", "is", "key", "key_layer", "keys", "math", "matmul", "new_context_layer_shape", "nn", "not", "num_attention_heads", "out_proj", "output_attentions", "outputs", "permute", "queries", "query", "query_layer", "return", "self", "seq_length", "shape", "size", "softmax", "sqrt", "torch", "transpose", "tuple", "value", "value_layer", "values", "view"], "grounding_dino/modeling_grounding_dino.py:GroundingDinoDecoderLayer.__init__": ["ACT2FN", "LayerNorm", "Linear", "ModelConfig", "ModelDecoderLayer", "ModelMultiheadAttention", "ModelMultiscaleDeformableAttention", "Module", "__init__", "activation_dropout", "activation_fn", "activation_function", "class", "config", "d_model", "decoder_attention_heads", "decoder_ffn_dim", "decoder_n_points", "def", "dropout", "embed_dim", "encoder_attn", "encoder_attn_layer_norm", "encoder_attn_text", "encoder_attn_text_layer_norm", "fc1", "fc2", "final_layer_norm", "layer_norm_eps", "n_points", "nn", "num_attention_heads", "num_heads", "self", "self_attn", "self_attn_layer_norm", "super"], "grounding_dino/modeling_grounding_dino.py:GroundingDinoDecoderLayer.with_pos_embed": ["ModelDecoderLayer", "Module", "None", "Tensor", "class", "def", "else", "if", "is", "nn", "position_embeddings", "return", "self", "tensor", "torch", "with_pos_embed"], "grounding_dino/modeling_grounding_dino.py:GroundingDinoDecoderLayer.forward": ["False", "ModelDecoderLayer", "Module", "None", "Tensor", "True", "activation_dropout", "activation_fn", "attention_mask", "bool", "class", "cross_attn_weights", "def", "dropout", "encoder_attention_mask", "encoder_attn", "encoder_attn_layer_norm", "encoder_attn_text", "encoder_attn_text_layer_norm", "encoder_hidden_states", "fc1", "fc2", "final_layer_norm", "forward", "functional", "hidden_states", "if", "keys", "level_start_index", "nn", "output_attentions", "outputs", "p", "position_embeddings", "queries", "reference_points", "residual", "return", "second_residual", "self", "self_attn", "self_attn_layer_norm", "self_attn_mask", "self_attn_weights", "spatial_shapes", "spatial_shapes_list", "text_cross_attn_weights", "text_encoder_attention_mask", "text_encoder_hidden_states", "third_residual", "torch", "training", "values", "vision_encoder_attention_mask", "vision_encoder_hidden_states", "with_pos_embed"], "grounding_dino/modeling_grounding_dino.py:GroundingDinoContrastiveEmbedding.__init__": ["ModelContrastiveEmbedding", "Module", "__init__", "class", "config", "def", "max_text_len", "nn", "self", "super"], "grounding_dino/modeling_grounding_dino.py:GroundingDinoContrastiveEmbedding.forward": ["BoolTensor", "FloatTensor", "ModelContrastiveEmbedding", "Module", "None", "class", "def", "device", "float", "forward", "full", "inf", "masked_fill", "max_text_len", "new_output", "nn", "output", "return", "self", "shape", "text_hidden_state", "text_token_mask", "torch", "transpose", "vision_hidden_state"], "grounding_dino/modeling_grounding_dino.py:GroundingDinoPreTrainedModel._init_weights": ["Conv2d", "Embedding", "False", "GroupNorm", "LayerNorm", "Linear", "ModelBiMultiHeadAttention", "ModelFusionLayer", "ModelLearnedPositionEmbedding", "ModelMLPPredictionHead", "ModelMultiscaleDeformableAttention", "ModelPreTrainedModel", "None", "PreTrainedModel", "True", "_init_weights", "_is_hf_initialized", "abs", "and", "arange", "attention_weights", "bias", "class", "column_embeddings", "config", "constant_", "copy_", "cos", "def", "default_dtype", "dtype", "elif", "for", "gain", "get_default_dtype", "getattr", "grid_init", "hasattr", "i", "if", "in", "init", "init_std", "int64", "is", "isinstance", "keepdim", "layers", "level_embed", "math", "max", "mean", "module", "n_heads", "n_levels", "n_points", "nn", "normal_", "not", "ones_", "out_text_proj", "out_vision_proj", "output_proj", "padding_idx", "pi", "range", "reference_points", "repeat", "row_embeddings", "sampling_offsets", "self", "sin", "stack", "std", "text_param", "text_proj", "thetas", "to", "torch", "two_stage", "uniform_", "value_proj", "values_text_proj", "values_vision_proj", "view", "vision_param", "vision_proj", "weight", "xavier_uniform_", "zeros_"], "grounding_dino/modeling_grounding_dino.py:GroundingDinoPreTrainedModel._set_gradient_checkpointing": ["False", "ModelDecoder", "ModelPreTrainedModel", "PreTrainedModel", "_set_gradient_checkpointing", "class", "def", "gradient_checkpointing", "if", "isinstance", "module", "self", "value"], "grounding_dino/modeling_grounding_dino.py:GroundingDinoEncoder.__init__": ["ModelConfig", "ModelEncoder", "ModelEncoderLayer", "ModelPreTrainedModel", "ModuleList", "_", "__init__", "class", "config", "def", "dropout", "encoder_layers", "for", "in", "layers", "nn", "post_init", "range", "self", "super"], "grounding_dino/modeling_grounding_dino.py:GroundingDinoEncoder.get_reference_points": ["ModelEncoder", "ModelPreTrainedModel", "None", "append", "cat", "class", "def", "device", "dtype", "enumerate", "float32", "for", "get_reference_points", "height", "ij", "in", "indexing", "level", "linspace", "meshgrid", "ref", "ref_x", "ref_y", "reference_points", "reference_points_list", "reshape", "return", "spatial_shapes", "stack", "torch", "valid_ratios", "width"], "grounding_dino/modeling_grounding_dino.py:GroundingDinoEncoder.forward": ["ModelEncoder", "ModelEncoderOutput", "ModelPreTrainedModel", "None", "Tensor", "all_attn_deformable", "all_attn_enhanced_text", "all_attn_fused_text", "all_attn_fused_vision", "all_attns", "attentions", "class", "config", "def", "device", "else", "enc_outputs", "encoder_layer", "encoder_text_states", "encoder_vision_states", "enumerate", "for", "forward", "get_reference_points", "i", "if", "in", "int", "is", "key_padding_mask", "kwargs", "last_hidden_state_text", "last_hidden_state_vision", "layers", "level_start_index", "list", "not", "output_attentions", "output_hidden_states", "r", "reference_points", "return", "return_dict", "self", "spatial_shapes", "spatial_shapes_list", "text_attention_mask", "text_features", "text_hidden_states", "text_position_embedding", "text_position_ids", "text_self_attention_masks", "tuple", "use_return_dict", "v", "valid_ratios", "vision_attention_mask", "vision_features", "vision_hidden_states", "vision_position_embedding"], "grounding_dino/modeling_grounding_dino.py:GroundingDinoDecoder.__init__": ["False", "LayerNorm", "ModelConfig", "ModelDecoder", "ModelDecoderLayer", "ModelMLPPredictionHead", "ModelPreTrainedModel", "ModuleList", "None", "_", "__init__", "bbox_embed", "class", "class_embed", "config", "d_model", "decoder_layers", "def", "dropout", "for", "gradient_checkpointing", "in", "layer_norm", "layer_norm_eps", "layers", "nn", "post_init", "query_dim", "query_scale", "range", "reference_points_head", "self", "super"], "grounding_dino/modeling_grounding_dino.py:GroundingDinoDecoder.forward": ["Last", "ModelDecoder", "ModelDecoderOutput", "ModelPreTrainedModel", "None", "ValueError", "all_attns", "all_cross_attns_text", "all_cross_attns_vision", "all_hidden_states", "all_self_attns", "and", "attentions", "bbox_embed", "be", "but", "cat", "checkpoint", "class", "config", "create_custom_forward", "custom_forward", "d_model", "decoder_attention_heads", "decoder_layer", "def", "detach", "dim", "dtype", "elif", "else", "enumerate", "eps", "f", "finfo", "for", "forward", "get_sine_pos_embed", "got", "gradient_checkpointing", "hidden_states", "idx", "if", "in", "inputs", "inputs_embeds", "intermediate", "intermediate_hidden_states", "intermediate_reference_points", "is", "kwargs", "last_hidden_state", "layer_norm", "layer_outputs", "layers", "level_start_index", "logit", "min", "module", "must", "new_reference_points", "not", "num_coordinates", "num_pos_feats", "num_queries", "of", "or", "output_attentions", "output_hidden_states", "position_embeddings", "query_pos", "r", "raise", "reference_points", "reference_points_head", "reference_points_input", "repeat", "return", "return_dict", "self", "self_attn_mask", "shape", "sigmoid", "spatial_shapes", "spatial_shapes_list", "special", "stack", "text_encoder_attention_mask", "text_encoder_hidden_states", "tmp", "to", "torch", "training", "tuple", "use_return_dict", "utils", "v", "valid_ratios", "vision_encoder_attention_mask", "vision_encoder_hidden_states"], "grounding_dino/modeling_grounding_dino.py:generate_masks_with_special_tokens_and_transfer_map": ["LongTensor", "Model_masks_with_special_tokens_and_transfer_map", "SPECIAL_TOKENS", "Tensor", "True", "arange", "attention_mask", "batch_size", "bool", "col", "def", "device", "else", "eye", "for", "i", "idxs", "if", "in", "input_ids", "logical_or", "long", "nonzero", "num_token", "or", "position_ids", "previous_col", "range", "repeat", "return", "row", "shape", "special_token", "special_tokens_mask", "to", "torch", "tuple", "unsqueeze", "zeros"], "grounding_dino/modeling_grounding_dino.py:GroundingDinoModel.__init__": ["AutoModel", "Conv2d", "Embedding", "False", "GroupNorm", "LayerNorm", "Linear", "ModelConfig", "ModelContrastiveEmbedding", "ModelConvEncoder", "ModelConvModel", "ModelDecoder", "ModelEncoder", "ModelMLPPredictionHead", "ModelModel", "ModelPreTrainedModel", "ModuleList", "None", "Parameter", "Sequential", "Tensor", "_", "__init__", "add_pooling_layer", "and", "append", "backbone", "bbox_embed", "build_position_encoding", "class", "config", "d_model", "decoder", "decoder_bbox_embed_share", "def", "else", "embedding_init_target", "enc_output", "enc_output_norm", "encoder", "encoder_output_bbox_embed", "encoder_output_class_embed", "for", "from_config", "hidden_dim", "hidden_size", "i", "if", "in", "in_channels", "input_dim", "input_proj_list", "input_proj_vision", "intermediate_channel_sizes", "is", "kernel_size", "layer_norm_eps", "len", "level_embed", "nn", "not", "num_backbone_outs", "num_feature_levels", "num_layers", "num_queries", "or", "output_dim", "padding", "position_embeddings", "post_init", "query_position_embeddings", "range", "reference_points", "self", "stride", "super", "text_backbone", "text_config", "text_projection", "torch", "two_stage", "two_stage_bbox_embed_share"], "grounding_dino/modeling_grounding_dino.py:GroundingDinoModel.freeze_backbone": ["False", "ModelModel", "ModelPreTrainedModel", "backbone", "class", "conv_encoder", "def", "for", "freeze_backbone", "in", "model", "name", "named_parameters", "param", "requires_grad_", "self"], "grounding_dino/modeling_grounding_dino.py:GroundingDinoModel.unfreeze_backbone": ["ModelModel", "ModelPreTrainedModel", "True", "backbone", "class", "conv_encoder", "def", "for", "in", "model", "name", "named_parameters", "param", "requires_grad_", "self", "unfreeze_backbone"], "grounding_dino/modeling_grounding_dino.py:GroundingDinoModel.get_valid_ratio": ["ModelModel", "ModelPreTrainedModel", "_", "class", "def", "float", "get_valid_ratio", "height", "mask", "return", "self", "shape", "stack", "sum", "torch", "valid_height", "valid_ratio", "valid_ratio_height", "valid_ratio_width", "valid_width", "width"], "grounding_dino/modeling_grounding_dino.py:GroundingDinoModel.generate_encoder_output_proposals": ["ModelModel", "ModelPreTrainedModel", "True", "all", "append", "batch_size", "cat", "class", "current_position", "def", "device", "dtype", "enc_output", "enc_output_norm", "enumerate", "expand", "float", "float32", "for", "generate_encoder_output_proposals", "grid", "grid_x", "grid_y", "height", "ij", "in", "indexing", "inf", "keepdim", "level", "linspace", "log", "mask_flatten_", "masked_fill", "meshgrid", "object_query", "ones_like", "output_proposals", "output_proposals_valid", "padding_mask", "proposal", "proposals", "return", "scale", "self", "shape", "spatial_shapes", "sum", "torch", "unsqueeze", "valid_height", "valid_width", "view", "width", "width_height"], "grounding_dino/modeling_grounding_dino.py:GroundingDinoModel.forward": ["ModelEncoderOutput", "ModelModel", "ModelModelOutput", "ModelPreTrainedModel", "None", "Tensor", "_len_sources", "and", "append", "as_tensor", "attention_mask", "attentions", "backbone", "batch_size", "bool", "cat", "class", "config", "cumsum", "d_model", "decoder", "decoder_attentions", "decoder_hidden_states", "decoder_outputs", "def", "delta_bbox", "detach", "device", "dim", "dtype", "elif", "else", "embedding_init_target", "enc_outputs", "enc_outputs_class", "enc_outputs_coord_logits", "encoder", "encoder_attentions", "encoder_last_hidden_state_text", "encoder_last_hidden_state_vision", "encoder_logits", "encoder_output_bbox_embed", "encoder_output_class_embed", "encoder_outputs", "encoder_pred_boxes", "encoder_text_hidden_states", "encoder_vision_hidden_states", "enumerate", "feature_maps", "flatten", "float", "for", "forward", "functional", "gather", "generate_encoder_output_proposals", "generate_masks_with_special_tokens_and_transfer_map", "get_valid_ratio", "height", "hidden_states", "if", "in", "init_reference_points", "input_ids", "input_proj_vision", "inputs_embeds", "intermediate_hidden_states", "intermediate_reference_points", "interpolate", "is", "isinstance", "kwargs", "last_hidden_state", "last_hidden_state_text", "last_hidden_state_vision", "len", "level", "level_embed", "level_start_index", "long", "lvl_pos_embed", "lvl_pos_embed_flatten", "m", "mask", "mask_flatten", "masks", "max", "max_text_len", "ndim", "new_zeros", "nn", "not", "num_channels", "num_feature_levels", "num_queries", "object_query_embedding", "ones", "ones_like", "or", "output_attentions", "output_hidden_states", "output_proposals", "pixel_mask", "pixel_values", "pos_embed", "pos_l", "position_embedding", "position_embeddings_list", "position_ids", "prod", "query_embeds", "query_position_embeddings", "r", "range", "reference_points", "repeat", "return", "return_dict", "self", "self_attn_mask", "shape", "sigmoid", "size", "source", "source_flatten", "spatial_shape", "spatial_shapes", "spatial_shapes_list", "squeeze", "stack", "target", "text_attention_mask", "text_backbone", "text_encoder_attention_mask", "text_encoder_hidden_states", "text_features", "text_hidden_states", "text_outputs", "text_position_embedding", "text_position_ids", "text_projection", "text_self_attention_masks", "text_token_mask", "to", "token_type_ids", "topk", "topk_coords_logits", "topk_logits", "topk_proposals", "torch", "transpose", "tuple", "tuple_outputs", "two_stage", "unsqueeze", "use_return_dict", "valid_ratios", "value", "view", "vision_attention_mask", "vision_encoder_attention_mask", "vision_encoder_hidden_states", "vision_features", "vision_hidden_states", "vision_position_embedding", "weight", "width", "zeros_like", "zip"], "grounding_dino/modeling_grounding_dino.py:GroundingDinoMLPPredictionHead.__init__": ["Linear", "ModelMLPPredictionHead", "Module", "ModuleList", "__init__", "class", "def", "for", "h", "hidden_dim", "in", "input_dim", "k", "layers", "n", "nn", "num_layers", "output_dim", "self", "super", "zip"], "grounding_dino/modeling_grounding_dino.py:GroundingDinoMLPPredictionHead.forward": ["ModelMLPPredictionHead", "Module", "class", "def", "else", "enumerate", "for", "forward", "functional", "i", "if", "in", "layer", "layers", "nn", "num_layers", "relu", "return", "self", "x"], "grounding_dino/modeling_grounding_dino.py:build_label_maps": ["F", "FloatTensor", "LongTensor", "Model_label_maps", "None", "SPECIAL_TOKENS", "cumsum", "def", "delimiter_token_masks", "delimiter_tokens", "device", "dim", "for", "in", "input_ids", "int32", "isin", "label_group", "label_groups", "label_map", "label_maps", "logits", "max_seq_len", "num_labels", "pad", "repeat", "return", "shape", "tensor", "to", "torch", "tuple", "unique", "unique_labels", "unsqueeze", "value", "where"], "grounding_dino/modeling_grounding_dino.py:build_text_mask": ["Model_text_mask", "None", "attention_mask", "bool", "def", "device", "dtype", "logits", "return", "seq_len", "shape", "text_mask", "torch", "zeros_like"], "grounding_dino/modeling_grounding_dino.py:GroundingDinoForObjectDetection.__init__": ["ModelConfig", "ModelContrastiveEmbedding", "ModelForObjectDetection", "ModelMLPPredictionHead", "ModelModel", "ModelPreTrainedModel", "ModuleList", "_", "__init__", "_tied_weights_keys", "bbox_embed", "class", "class_embed", "config", "d", "d_model", "decoder", "decoder_bbox_embed_share", "decoder_layers", "def", "del", "for", "hidden_dim", "if", "in", "input_dim", "model", "nn", "not", "num_layers", "output_dim", "post_init", "r", "range", "self", "super"], "grounding_dino/modeling_grounding_dino.py:GroundingDinoForObjectDetection.forward": ["BoolTensor", "FloatTensor", "LongTensor", "ModelEncoderOutput", "ModelForObjectDetection", "ModelObjectDetectionOutput", "ModelPreTrainedModel", "None", "ValueError", "append", "attention_mask", "auxiliary_outputs", "bbox_embed", "be", "bool", "build_label_maps", "build_text_mask", "but", "class", "class_embed", "config", "decoder_attentions", "decoder_hidden_states", "def", "delta_bbox", "device", "dict", "dict_outputs", "elif", "else", "enc_outputs_class", "enc_outputs_coord_logits", "enc_text_hidden_state", "encoder_attentions", "encoder_last_hidden_state_text", "encoder_last_hidden_state_vision", "encoder_logits", "encoder_outputs", "encoder_pred_boxes", "encoder_text_hidden_states", "encoder_vision_hidden_states", "eps", "f", "for", "forward", "got", "hidden_states", "idx", "if", "in", "init_reference_points", "input_ids", "inter_references_points", "intermediate_hidden_states", "intermediate_reference_points", "is", "kwargs", "label_maps", "labels", "last_hidden_state", "level", "list", "logit", "logits", "loss", "loss_dict", "loss_function", "model", "not", "num_levels", "ones_like", "or", "out", "output", "output_attentions", "output_hidden_states", "outputs", "outputs_class", "outputs_classes", "outputs_coord", "outputs_coord_logits", "outputs_coords", "pixel_mask", "pixel_values", "pred_boxes", "r", "raise", "range", "reference", "reference_coordinates", "return", "return_dict", "self", "shape", "should", "sigmoid", "special", "stack", "str", "text_hidden_state", "text_mask", "text_token_mask", "token_type_ids", "torch", "tuple", "use_return_dict", "vision_hidden_state"], "groupvit/modeling_groupvit.py:contrastive_loss": ["Model_loss", "Tensor", "arange", "cross_entropy", "def", "device", "functional", "len", "logits", "nn", "return", "torch"], "groupvit/modeling_groupvit.py:groupvit_loss": ["Model_loss", "Tensor", "caption_loss", "contrastive_loss", "def", "image_loss", "return", "similarity", "t", "torch"], "groupvit/modeling_groupvit.py:hard_softmax": ["Model_softmax", "Tensor", "True", "def", "detach", "dim", "index", "int", "keepdim", "legacy_contiguous_format", "logits", "max", "memory_format", "ret", "return", "scatter_", "softmax", "torch", "y_Model", "y_soft", "zeros_like"], "groupvit/modeling_groupvit.py:gumbel_softmax": ["False", "Model", "Model_dist", "Model_softmax", "Models", "Tensor", "True", "bool", "def", "detach", "device", "dim", "distributions", "dtype", "else", "float", "hard", "if", "index", "int", "keepdim", "legacy_contiguous_format", "logits", "max", "memory_format", "ret", "return", "sample", "scatter_", "shape", "softmax", "tau", "tensor", "torch", "y_hard", "y_soft", "zeros_like"], "groupvit/modeling_groupvit.py:resize_attention_map": ["False", "Model_attention_map", "align_corners", "attentions", "batch_size", "bilinear", "def", "else", "feat_height", "feat_width", "functional", "groups", "height", "if", "int", "interpolate", "mode", "nn", "np", "reshape", "return", "round", "scale", "shape", "size", "width"], "groupvit/modeling_groupvit.py:get_grouping_from_attentions": ["Model_grouping_from_attentions", "None", "append", "attentions", "attn_maps", "attn_masks", "contiguous", "cur_attn_map", "def", "else", "final_grouping", "for", "hw_shape", "if", "in", "is", "no_grad", "permute", "prev_attn_masks", "resize_attention_map", "return", "torch", "with"], "groupvit/modeling_groupvit.py:GroupViTCrossAttentionLayer.__init__": ["LayerNorm", "ModelAttention", "ModelCrossAttentionLayer", "ModelMLP", "ModelVisionConfig", "Module", "__init__", "attn", "class", "config", "def", "eps", "hidden_size", "layer_norm_eps", "mlp", "nn", "norm2", "norm_post", "self", "super"], "groupvit/modeling_groupvit.py:GroupViTCrossAttentionLayer.forward": ["ModelCrossAttentionLayer", "Module", "attn", "class", "def", "encoder_hidden_states", "forward", "key", "mlp", "nn", "norm2", "norm_post", "query", "return", "self", "x"], "groupvit/modeling_groupvit.py:GroupViTAssignAttention.__init__": ["Linear", "ModelAssignAttention", "ModelVisionConfig", "Module", "__init__", "assign_eps", "class", "config", "def", "hidden_size", "k_proj", "nn", "proj", "q_proj", "scale", "self", "super", "v_proj"], "groupvit/modeling_groupvit.py:GroupViTAssignAttention.get_attn": ["ModelAssignAttention", "Module", "True", "and", "attn", "class", "def", "dim", "else", "functional", "get_attn", "gumbel", "gumbel_softmax", "hard", "hard_softmax", "if", "nn", "return", "self", "softmax", "training"], "groupvit/modeling_groupvit.py:GroupViTAssignAttention.forward": ["False", "ModelAssignAttention", "Module", "True", "assign_eps", "attn", "class", "def", "dim", "forward", "get_attn", "gumbel", "hard", "k_proj", "keepdim", "key", "nn", "out", "proj", "q_proj", "query", "raw_attn", "return", "scale", "self", "soft_attn", "sum", "transpose", "v_proj", "value"], "groupvit/modeling_groupvit.py:GroupViTTokenAssign.__init__": ["Iterable", "LayerNorm", "ModelAssignAttention", "ModelCrossAttentionLayer", "ModelMLP", "ModelMixerMLP", "ModelTokenAssign", "ModelVisionConfig", "Module", "__init__", "abc", "assign", "assign_mlp_ratio", "channels_dim", "class", "collections", "config", "def", "else", "eps", "for", "hidden_size", "if", "in", "int", "isinstance", "layer_norm_eps", "mlp_channels", "mlp_inter", "nn", "norm_new_x", "norm_post_tokens", "norm_tokens", "norm_x", "num_Model_token", "num_output_Model", "pre_assign_attn", "self", "super", "tokens_dim", "x"], "groupvit/modeling_groupvit.py:GroupViTTokenAssign.project_group_token": ["ModelTokenAssign", "Model_tokens", "Module", "class", "def", "mlp_inter", "nn", "norm_post_tokens", "project_Model_token", "projected_Model_tokens", "return", "self"], "groupvit/modeling_groupvit.py:GroupViTTokenAssign.forward": ["ModelTokenAssign", "Model_tokens", "Module", "assign", "attention", "class", "def", "forward", "image_tokens", "mlp_channels", "new_image_tokens", "nn", "norm_new_x", "norm_tokens", "norm_x", "pre_assign_attn", "project_Model_token", "projected_Model_tokens", "return", "self"], "groupvit/modeling_groupvit.py:GroupViTModelOutput.to_tuple": ["Any", "ModelModelOutput", "ModelOutput", "class", "def", "else", "for", "getattr", "if", "in", "k", "keys", "not", "return", "self", "text_model_output", "to_tuple", "tuple", "vision_model_output"], "groupvit/modeling_groupvit.py:GroupViTPatchEmbeddings.__init__": ["Conv2d", "Iterable", "ModelPatchEmbeddings", "Module", "__init__", "abc", "class", "collections", "def", "else", "embed_dim", "if", "image_size", "int", "isinstance", "kernel_size", "nn", "num_channels", "num_patches", "patch_size", "projection", "self", "stride", "super", "tuple"], "groupvit/modeling_groupvit.py:GroupViTPatchEmbeddings.forward": ["False", "Input", "ModelPatchEmbeddings", "Module", "Tensor", "ValueError", "batch_size", "bool", "class", "def", "doesn", "f", "flatten", "forward", "height", "if", "image", "image_size", "interpolate_pos_encoding", "match", "model", "nn", "not", "num_channels", "or", "pixel_values", "projection", "raise", "return", "self", "shape", "size", "t", "torch", "transpose", "width", "x"], "groupvit/modeling_groupvit.py:GroupViTVisionEmbeddings.__init__": ["Dropout", "LayerNorm", "ModelPatchEmbeddings", "ModelVisionConfig", "ModelVisionEmbeddings", "Module", "Parameter", "__init__", "class", "config", "def", "dropout", "embed_dim", "eps", "hidden_size", "image_size", "layer_norm_eps", "layernorm", "nn", "num_channels", "num_patches", "patch_embeddings", "patch_size", "position_embeddings", "self", "super", "torch", "zeros"], "groupvit/modeling_groupvit.py:GroupViTVisionEmbeddings.interpolate_pos_encoding": ["False", "ModelVisionEmbeddings", "Module", "Tensor", "align_corners", "and", "bicubic", "class", "def", "dim", "embeddings", "functional", "height", "if", "int", "interpolate", "interpolate_pos_encoding", "is_tracing", "jit", "mode", "new_height", "new_width", "nn", "not", "num_patches", "num_positions", "patch_pos_embed", "patch_size", "permute", "position_embeddings", "reshape", "return", "self", "shape", "size", "sqrt_num_positions", "torch", "torch_int", "view", "width"], "groupvit/modeling_groupvit.py:GroupViTVisionEmbeddings.forward": ["False", "ModelVisionEmbeddings", "Module", "Tensor", "_", "batch_size", "bool", "class", "def", "dropout", "else", "embeddings", "forward", "height", "if", "interpolate_pos_encoding", "layernorm", "nn", "num_channels", "patch_embeddings", "pixel_values", "position_embeddings", "return", "self", "seq_len", "shape", "size", "torch", "width"], "groupvit/modeling_groupvit.py:GroupViTTextEmbeddings.__init__": ["Embedding", "False", "ModelTextConfig", "ModelTextEmbeddings", "Module", "__init__", "arange", "class", "config", "def", "embed_dim", "expand", "hidden_size", "max_position_embeddings", "nn", "persistent", "position_embedding", "position_ids", "register_buffer", "self", "super", "token_embedding", "torch", "vocab_size"], "groupvit/modeling_groupvit.py:GroupViTTextEmbeddings.forward": ["FloatTensor", "LongTensor", "ModelTextEmbeddings", "Module", "None", "Sequence", "Tensor", "ValueError", "and", "be", "class", "def", "else", "embeddings", "f", "forward", "got", "if", "input_ids", "inputs_embeds", "is", "length", "less", "max_position_embedding", "max_position_embeddings", "must", "nn", "not", "position_embedding", "position_embeddings", "position_ids", "raise", "return", "self", "seq_length", "sequence", "shape", "than", "token_embedding", "torch", "weight"], "groupvit/modeling_groupvit.py:GroupViTStage.__init__": ["LayerNorm", "ModelEncoderLayer", "ModelMixerMLP", "ModelStage", "ModelTokenAssign", "ModelVisionConfig", "Model_projector", "Model_token", "Module", "ModuleList", "None", "Parameter", "Sequential", "_", "__init__", "and", "class", "config", "def", "depth", "downsample", "else", "eps", "for", "hidden_size", "if", "in", "int", "layer_norm_eps", "layers", "nn", "num_Model_token", "num_output_Model", "num_prev_Model_token", "range", "self", "super", "torch", "zeros"], "groupvit/modeling_groupvit.py:GroupViTStage.with_group_token": ["ModelStage", "Model_token", "Module", "None", "class", "def", "is", "nn", "not", "return", "self", "with_Model_token"], "groupvit/modeling_groupvit.py:GroupViTStage.split_x": ["ModelStage", "Module", "None", "class", "def", "else", "if", "nn", "num_Model_token", "return", "self", "split_x", "with_Model_token", "x"], "groupvit/modeling_groupvit.py:GroupViTStage.concat_x": ["ModelStage", "Model_token", "Module", "None", "Tensor", "cat", "class", "concat_x", "def", "dim", "if", "is", "nn", "return", "self", "torch", "x"], "groupvit/modeling_groupvit.py:GroupViTStage.forward": ["False", "FloatTensor", "ModelStage", "Model_projector", "Model_token", "Module", "None", "Tensor", "attention", "attention_mask", "bool", "cat_x", "causal_attention_mask", "class", "concat_x", "def", "downsample", "else", "expand", "for", "forward", "hidden_states", "if", "in", "is", "layer", "layer_out", "layers", "nn", "not", "output_attentions", "outputs", "prev_Model_token", "return", "self", "size", "split_x", "torch", "tuple", "with_Model_token", "x"], "groupvit/modeling_groupvit.py:GroupViTMLP.__init__": ["ACT2FN", "Linear", "ModelMLP", "ModelVisionConfig", "Module", "None", "__init__", "activation_fn", "class", "config", "def", "else", "fc1", "fc2", "hidden_act", "hidden_size", "if", "int", "intermediate_size", "is", "nn", "not", "output_size", "self", "super"], "groupvit/modeling_groupvit.py:GroupViTMLP.forward": ["ModelMLP", "Module", "Tensor", "activation_fn", "class", "def", "fc1", "fc2", "forward", "hidden_states", "nn", "return", "self", "torch"], "groupvit/modeling_groupvit.py:GroupViTMixerMLP.forward": ["ModelMLP", "ModelMixerMLP", "class", "def", "forward", "return", "self", "super", "transpose", "x"], "groupvit/modeling_groupvit.py:GroupViTAttention.__init__": ["Linear", "ModelAttention", "Module", "ValueError", "__init__", "and", "attention_dropout", "be", "by", "class", "config", "def", "divisible", "dropout", "embed_dim", "f", "got", "head_dim", "hidden_size", "if", "k_proj", "must", "nn", "num_attention_heads", "num_heads", "out_proj", "q_proj", "raise", "scale", "self", "super", "v_proj"], "groupvit/modeling_groupvit.py:GroupViTAttention._shape": ["ModelAttention", "Module", "Tensor", "_shape", "bsz", "class", "contiguous", "def", "head_dim", "int", "nn", "num_heads", "return", "self", "seq_len", "tensor", "torch", "transpose", "view"], "groupvit/modeling_groupvit.py:GroupViTAttention.forward": ["Attention", "False", "FloatTensor", "ModelAttention", "Module", "None", "Tensor", "ValueError", "_shape", "attention_mask", "attn_output", "attn_probs", "attn_weights", "attn_weights_reshaped", "be", "bmm", "bool", "bsz", "but", "causal_attention_mask", "class", "def", "dim", "dropout", "else", "embed_dim", "encoder_hidden_states", "f", "forward", "functional", "head_dim", "hidden_states", "if", "is", "is_cross_attention", "k_proj", "key_states", "mask", "nn", "not", "num_heads", "of", "out_proj", "output_attentions", "p", "proj_shape", "q_proj", "query_states", "raise", "reshape", "return", "scale", "self", "should", "size", "softmax", "src_len", "tgt_len", "torch", "training", "transpose", "tuple", "v_proj", "value_states", "view", "weights"], "groupvit/modeling_groupvit.py:GroupViTEncoderLayer.__init__": ["GradientCheckpointingLayer", "LayerNorm", "ModelAttention", "ModelConfig", "ModelEncoderLayer", "ModelMLP", "__init__", "class", "config", "def", "embed_dim", "eps", "hidden_size", "layer_norm1", "layer_norm2", "layer_norm_eps", "mlp", "nn", "self", "self_attn", "super"], "groupvit/modeling_groupvit.py:GroupViTEncoderLayer.forward": ["False", "FloatTensor", "GradientCheckpointingLayer", "ModelEncoderLayer", "None", "Tensor", "attention_mask", "attn_weights", "bool", "causal_attention_mask", "class", "def", "forward", "hidden_states", "if", "layer_norm1", "layer_norm2", "mlp", "output_attentions", "outputs", "residual", "return", "self", "self_attn", "torch", "tuple"], "groupvit/modeling_groupvit.py:GroupViTPreTrainedModel._init_weights": ["BatchNorm1d", "Conv2d", "LayerNorm", "Linear", "ModelAttention", "ModelMLP", "ModelPreTrainedModel", "ModelTextEmbeddings", "None", "PreTrainedModel", "_init_weights", "arange", "bias", "class", "config", "copy_", "def", "elif", "embed_dim", "expand", "factor", "fc1", "fc2", "fc_std", "getattr", "hidden_size", "if", "in_proj_std", "init", "init_range", "initializer_factor", "initializer_range", "is", "isinstance", "k_proj", "mean", "module", "nn", "normal_", "not", "num_batches_tracked", "num_hidden_layers", "ones_", "out_proj", "out_proj_std", "position_embedding", "position_ids", "q_proj", "running_mean", "running_var", "self", "shape", "std", "token_embedding", "torch", "v_proj", "weight", "zeros_"], "groupvit/modeling_groupvit.py:GroupViTVisionEncoder.__init__": ["False", "ModelStage", "ModelVisionConfig", "ModelVisionEncoder", "Module", "ModuleList", "None", "__init__", "class", "config", "def", "depth", "depths", "else", "for", "gradient_checkpointing", "i", "if", "in", "len", "nn", "num_Model_token", "num_Model_tokens", "num_output_Model", "num_output_Models", "num_prev_Model_token", "range", "self", "stages", "super"], "groupvit/modeling_groupvit.py:GroupViTVisionEncoder.forward": ["BaseModelOutput", "ModelVisionEncoder", "Model_tokens", "Module", "None", "Tensor", "all_Modelings", "all_hidden_states", "and", "attentions", "bool", "class", "config", "def", "else", "enumerate", "for", "forward", "hidden_states", "i", "if", "in", "is", "last_hidden_state", "layer_outputs", "nn", "not", "output_attentions", "output_hidden_states", "return", "return_dict", "self", "stage", "stages", "torch", "tuple", "use_return_dict", "v"], "groupvit/modeling_groupvit.py:GroupViTTextEncoder.__init__": ["False", "ModelEncoderLayer", "ModelTextConfig", "ModelTextEncoder", "Module", "ModuleList", "_", "__init__", "class", "config", "def", "for", "gradient_checkpointing", "in", "layers", "nn", "num_hidden_layers", "range", "self", "super"], "groupvit/modeling_groupvit.py:GroupViTTextEncoder.forward": ["BaseModelOutput", "ModelTextEncoder", "Module", "None", "Tensor", "all_attentions", "attention_mask", "attentions", "bool", "causal_attention_mask", "class", "config", "def", "else", "encoder_layer", "encoder_states", "enumerate", "for", "forward", "hidden_states", "idx", "if", "in", "inputs_embeds", "is", "last_hidden_state", "layer_outputs", "layers", "nn", "not", "output_attentions", "output_hidden_states", "r", "return", "return_dict", "self", "torch", "tuple", "use_return_dict", "v"], "groupvit/modeling_groupvit.py:GroupViTTextTransformer.__init__": ["LayerNorm", "ModelTextConfig", "ModelTextEmbeddings", "ModelTextEncoder", "ModelTextTransformer", "Module", "__init__", "class", "config", "def", "embed_dim", "embeddings", "encoder", "eos_token_id", "eps", "final_layer_norm", "hidden_size", "layer_norm_eps", "nn", "self", "super"], "groupvit/modeling_groupvit.py:GroupViTTextTransformer.forward": ["BaseModelOutputWithPooling", "ModelTextTransformer", "Module", "None", "Tensor", "ValueError", "You", "_create_4d_causal_attention_mask", "_prepare_4d_attention_mask", "arange", "argmax", "attention_mask", "attentions", "bool", "causal_attention_mask", "class", "config", "def", "device", "dim", "dtype", "else", "embeddings", "encoder", "encoder_outputs", "eos_token_id", "final_layer_norm", "forward", "have", "hidden_states", "if", "input_ids", "input_shape", "inputs_embeds", "int", "is", "last_hidden_state", "nn", "not", "output_attentions", "output_hidden_states", "pooled_output", "pooler_output", "position_ids", "raise", "return", "return_dict", "self", "shape", "size", "specify", "to", "torch", "tuple", "use_return_dict", "view"], "groupvit/modeling_groupvit.py:GroupViTTextModel.__init__": ["ModelPreTrainedModel", "ModelTextConfig", "ModelTextModel", "ModelTextTransformer", "__init__", "class", "config", "def", "post_init", "self", "super", "text_model"], "groupvit/modeling_groupvit.py:GroupViTTextModel.get_input_embeddings": ["ModelPreTrainedModel", "ModelTextModel", "Module", "class", "def", "embeddings", "get_input_embeddings", "nn", "return", "self", "text_model", "token_embedding"], "groupvit/modeling_groupvit.py:GroupViTTextModel.set_input_embeddings": ["ModelPreTrainedModel", "ModelTextModel", "class", "def", "embeddings", "self", "set_input_embeddings", "text_model", "token_embedding", "value"], "groupvit/modeling_groupvit.py:GroupViTTextModel.forward": ["BaseModelOutputWithPooling", "ModelPreTrainedModel", "ModelTextModel", "None", "Tensor", "attention_mask", "bool", "class", "def", "forward", "input_ids", "kwargs", "output_attentions", "output_hidden_states", "position_ids", "r", "return", "return_dict", "self", "text_model", "torch", "tuple"], "groupvit/modeling_groupvit.py:GroupViTVisionTransformer.__init__": ["LayerNorm", "ModelVisionConfig", "ModelVisionEmbeddings", "ModelVisionEncoder", "ModelVisionTransformer", "Module", "__init__", "class", "config", "def", "embed_dim", "embeddings", "encoder", "eps", "hidden_size", "layer_norm_eps", "layernorm", "nn", "self", "super"], "groupvit/modeling_groupvit.py:GroupViTVisionTransformer.forward": ["BaseModelOutputWithPooling", "FloatTensor", "ModelVisionTransformer", "Module", "None", "ValueError", "You", "attentions", "bool", "class", "config", "def", "dim", "else", "embeddings", "encoder", "encoder_outputs", "forward", "have", "hidden_states", "if", "is", "last_hidden_state", "layernorm", "mean", "nn", "not", "output_attentions", "output_hidden_states", "pixel_values", "pooled_output", "pooler_output", "raise", "return", "return_dict", "self", "specify", "to", "torch", "tuple", "use_return_dict"], "groupvit/modeling_groupvit.py:GroupViTVisionModel.__init__": ["ModelPreTrainedModel", "ModelVisionConfig", "ModelVisionModel", "ModelVisionTransformer", "__init__", "class", "config", "def", "post_init", "self", "super", "vision_model"], "groupvit/modeling_groupvit.py:GroupViTVisionModel.get_input_embeddings": ["ModelPatchEmbeddings", "ModelPreTrainedModel", "ModelVisionModel", "class", "def", "embeddings", "get_input_embeddings", "patch_embeddings", "return", "self", "vision_model"], "groupvit/modeling_groupvit.py:GroupViTVisionModel.forward": ["BaseModelOutputWithPooling", "FloatTensor", "ModelPreTrainedModel", "ModelVisionModel", "None", "bool", "class", "def", "forward", "kwargs", "output_attentions", "output_hidden_states", "pixel_values", "r", "return", "return_dict", "self", "torch", "tuple", "vision_model"], "groupvit/modeling_groupvit.py:GroupViTModel.__init__": ["BatchNorm1d", "Linear", "ModelConfig", "ModelModel", "ModelPreTrainedModel", "ModelTextConfig", "ModelTextTransformer", "ModelVisionConfig", "ModelVisionTransformer", "Parameter", "ReLU", "Sequential", "True", "TypeError", "__init__", "be", "bias", "but", "class", "config", "def", "expected", "f", "hidden_size", "if", "inplace", "is", "isinstance", "logit_scale", "logit_scale_init_value", "nn", "not", "of", "post_init", "projection_dim", "projection_intermediate_dim", "raise", "self", "super", "tensor", "text_config", "text_embed_dim", "text_model", "text_projection", "to", "torch", "type", "vision_config", "vision_embed_dim", "vision_model", "visual_projection"], "groupvit/modeling_groupvit.py:GroupViTModel.get_text_features": ["BaseModelOutputWithPooling", "FloatTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "attention_mask", "class", "def", "get_text_features", "input_ids", "pooler_output", "position_ids", "r", "return", "self", "text_features", "text_model", "text_outputs", "text_projection", "torch"], "groupvit/modeling_groupvit.py:GroupViTModel.get_image_features": ["BaseModelOutputWithPooling", "FloatTensor", "ModelModel", "ModelPreTrainedModel", "Tensor", "class", "def", "get_image_features", "image_features", "pixel_values", "pooler_output", "r", "return", "self", "torch", "vision_model", "vision_outputs", "visual_projection"], "groupvit/modeling_groupvit.py:GroupViTModel.forward": ["FloatTensor", "LongTensor", "ModelModel", "ModelModelOutput", "ModelPreTrainedModel", "Model_loss", "Modeling", "None", "Tensor", "True", "attention_mask", "attentions", "bool", "class", "config", "def", "dim", "else", "exp", "flatten_Modeling", "forward", "get_Modeling_from_attentions", "if", "image_Model_embeds", "image_embeds", "input_ids", "is", "keepdim", "kwargs", "logit_scale", "logits_per_image", "logits_per_image_Model", "logits_per_text", "loss", "matmul", "norm", "not", "output", "output_attentions", "output_hidden_states", "output_segmentation", "permute", "pixel_values", "position_ids", "r", "reshape", "return", "return_dict", "return_loss", "seg_logits", "segmentation_logits", "self", "shape", "t", "text_embeds", "text_model", "text_model_output", "text_outputs", "text_projection", "torch", "tuple", "use_return_dict", "vision_model", "vision_model_output", "vision_outputs", "visual_projection"], "helium/modeling_helium.py:HeliumRMSNorm.__init__": ["ModelRMSNorm", "Module", "Parameter", "__init__", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "helium/modeling_helium.py:HeliumRMSNorm.forward": ["ModelRMSNorm", "Module", "True", "class", "def", "dtype", "float32", "forward", "hidden_states", "input_dtype", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "helium/modeling_helium.py:HeliumRMSNorm.extra_repr": ["ModelRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "variance_epsilon", "weight"], "helium/modeling_helium.py:HeliumRotaryEmbedding.__init__": ["Callable", "False", "ModelConfig", "ModelRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "helium/modeling_helium.py:HeliumRotaryEmbedding.compute_default_rope_parameters": ["ModelConfig", "ModelRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "helium/modeling_helium.py:HeliumRotaryEmbedding.forward": ["False", "ModelRotaryEmbedding", "Module", "None", "and", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "helium/modeling_helium.py:HeliumMLP.__init__": ["ACT2FN", "Linear", "ModelMLP", "Module", "__init__", "act_fn", "bias", "class", "config", "def", "down_proj", "gate_proj", "hidden_act", "hidden_size", "intermediate_size", "mlp_bias", "nn", "self", "super", "up_proj"], "helium/modeling_helium.py:HeliumMLP.forward": ["ModelMLP", "Module", "act_fn", "class", "def", "down_proj", "forward", "gate_proj", "nn", "return", "self", "up_proj", "x"], "helium/modeling_helium.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "helium/modeling_helium.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value", "value_states"], "helium/modeling_helium.py:rotate_half": ["Model_half", "def", "dim", "flatten", "return", "stack", "torch", "x", "x1", "x2"], "helium/modeling_helium.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cos", "def", "dim", "k", "k_embed", "q", "q_embed", "repeat_interleave", "return", "rotate_half", "shape", "sin", "unsqueeze", "unsqueeze_dim"], "helium/modeling_helium.py:HeliumAttention.__init__": ["False", "Linear", "ModelAttention", "ModelConfig", "Module", "None", "True", "__init__", "attention_bias", "attention_dropout", "bias", "class", "config", "def", "getattr", "head_dim", "hidden_size", "int", "is_causal", "k_proj", "layer_idx", "math", "nn", "num_attention_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "q_proj", "scaling", "self", "sqrt", "super", "v_proj"], "helium/modeling_helium.py:HeliumAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "LongTensor", "ModelAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_attn_implementation", "apply_rotary_pos_emb", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "cache_kwargs", "cache_position", "class", "config", "contiguous", "cos", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "head_dim", "hidden_shape", "hidden_states", "if", "input_shape", "is", "k_proj", "key_states", "kwargs", "layer_idx", "nn", "not", "o_proj", "past_key_values", "position_embeddings", "q_proj", "query_states", "reshape", "return", "scaling", "self", "shape", "sin", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "view"], "helium/modeling_helium.py:HeliumDecoderLayer.__init__": ["GradientCheckpointingLayer", "ModelAttention", "ModelConfig", "ModelDecoderLayer", "ModelMLP", "ModelRMSNorm", "None", "__init__", "class", "config", "def", "eps", "hidden_size", "input_layernorm", "int", "layer_idx", "mlp", "post_attention_layernorm", "rms_norm_eps", "self", "self_attn", "super"], "helium/modeling_helium.py:HeliumDecoderLayer.forward": ["Cache", "False", "GradientCheckpointingLayer", "LongTensor", "ModelDecoderLayer", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "attention_mask", "bool", "cache_position", "class", "def", "forward", "hidden_states", "input_layernorm", "kwargs", "mlp", "past_key_values", "position_embeddings", "position_ids", "post_attention_layernorm", "residual", "return", "self", "self_attn", "torch", "tuple", "use_cache"], "helium/modeling_helium.py:HeliumModel.__init__": ["Embedding", "False", "ModelConfig", "ModelDecoderLayer", "ModelModel", "ModelPreTrainedModel", "ModelRMSNorm", "ModelRotaryEmbedding", "ModuleList", "__init__", "class", "config", "def", "embed_tokens", "eps", "for", "gradient_checkpointing", "hidden_size", "in", "layer_idx", "layers", "nn", "norm", "num_hidden_layers", "pad_token_id", "padding_idx", "post_init", "range", "rms_norm_eps", "rotary_emb", "self", "super", "vocab_size"], "helium/modeling_helium.py:HeliumModel.forward": ["BaseModelOutputWithPast", "Cache", "DynamicCache", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "and", "arange", "attention_mask", "bool", "cache_position", "causal_mask", "class", "config", "create_causal_mask", "decoder_layer", "def", "device", "else", "embed_tokens", "exactly", "for", "forward", "get_seq_length", "hidden_states", "if", "in", "input_embeds", "input_ids", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layers", "must", "norm", "not", "num_hidden_layers", "of", "one", "or", "past_key_values", "past_seen_tokens", "position_embeddings", "position_ids", "raise", "return", "rotary_emb", "self", "shape", "specify", "torch", "unsqueeze", "use_cache"], "helium/modeling_helium.py:HeliumForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "ModelForCausalLM", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "vocab_size"], "helium/modeling_helium.py:HeliumForCausalLM.forward": ["BaseModelOutputWithPast", "Cache", "CausalLMOutputWithPast", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForCausalLM", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "outputs", "past_key_values", "position_ids", "r", "return", "self", "slice", "slice_indices", "torch", "use_cache", "vocab_size"], "hgnet_v2/modeling_hgnet_v2.py:HGNetV2PreTrainedModel._init_weights": ["BatchNorm", "Model", "PreTrainedModel", "__class__", "__name__", "_init_weights", "bias", "class", "def", "if", "in", "init", "module", "ones_", "running_mean", "running_var", "self", "super", "weight", "zeros_"], "hgnet_v2/modeling_hgnet_v2.py:HGNetV2LearnableAffineBlock.__init__": ["Model", "Module", "Parameter", "True", "__init__", "bias", "bias_value", "class", "def", "float", "nn", "requires_grad", "scale", "scale_value", "self", "super", "tensor", "torch"], "hgnet_v2/modeling_hgnet_v2.py:HGNetV2LearnableAffineBlock.forward": ["Model", "Module", "Tensor", "bias", "class", "def", "forward", "hidden_state", "nn", "return", "scale", "self"], "hgnet_v2/modeling_hgnet_v2.py:HGNetV2ConvLayer.__init__": ["ACT2FN", "BatchNorm2d", "Conv2d", "False", "Identity", "Model", "ModelLearnableAffineBlock", "Module", "None", "__init__", "activation", "and", "bias", "bool", "class", "convolution", "def", "else", "groups", "if", "in_channels", "int", "is", "kernel_size", "lab", "nn", "normalization", "not", "out_channels", "padding", "relu", "self", "str", "stride", "super", "use_learnable_affine_block"], "hgnet_v2/modeling_hgnet_v2.py:HGNetV2ConvLayer.forward": ["Model", "Module", "Tensor", "activation", "class", "convolution", "def", "forward", "hidden_state", "input", "lab", "nn", "normalization", "return", "self"], "hgnet_v2/modeling_hgnet_v2.py:HGNetV2ConvLayerLight.__init__": ["False", "Model", "ModelConvLayer", "Module", "None", "__init__", "activation", "bool", "class", "conv1", "conv2", "def", "groups", "in_channels", "int", "kernel_size", "nn", "out_channels", "self", "super", "use_learnable_affine_block"], "hgnet_v2/modeling_hgnet_v2.py:HGNetV2ConvLayerLight.forward": ["Model", "Module", "Tensor", "class", "conv1", "conv2", "def", "forward", "hidden_state", "nn", "return", "self"], "hgnet_v2/modeling_hgnet_v2.py:HGNetV2Embeddings.__init__": ["MaxPool2d", "Model", "ModelConfig", "ModelConvLayer", "Module", "True", "__init__", "activation", "ceil_mode", "class", "config", "def", "hidden_act", "kernel_size", "nn", "num_channels", "pool", "self", "stem1", "stem2a", "stem2b", "stem3", "stem4", "stem_channels", "stride", "super", "use_learnable_affine_block"], "hgnet_v2/modeling_hgnet_v2.py:HGNetV2Embeddings.forward": ["F", "Make", "Model", "Module", "Tensor", "ValueError", "cat", "channel", "class", "configuration", "def", "dim", "dimension", "emb_stem_2a", "embedding", "forward", "if", "in", "match", "nn", "num_channels", "of", "one", "pad", "pixel", "pixel_values", "pool", "pooled_emb", "raise", "return", "self", "set", "shape", "stem1", "stem2a", "stem2b", "stem3", "stem4", "sure", "that", "the", "torch", "values", "with"], "hgnet_v2/modeling_hgnet_v2.py:HGNetV2BasicLayer.__init__": ["Dropout", "False", "Identity", "Model", "ModelConvLayer", "ModelConvLayerLight", "Module", "ModuleList", "Sequential", "__init__", "aggregation", "aggregation_excitation_conv", "aggregation_squeeze_conv", "append", "block", "bool", "class", "def", "drop_path", "else", "float", "for", "i", "if", "in", "in_channels", "int", "kernel_size", "layer_num", "layers", "light_block", "middle_channels", "nn", "out_channels", "range", "residual", "self", "stride", "super", "temp_in_channels", "total_channels", "use_learnable_affine_block"], "hgnet_v2/modeling_hgnet_v2.py:HGNetV2BasicLayer.forward": ["Model", "Module", "Tensor", "aggregation", "append", "cat", "class", "def", "dim", "drop_path", "for", "forward", "hidden_state", "identity", "if", "in", "layer", "layers", "nn", "output", "residual", "return", "self", "torch"], "hgnet_v2/modeling_hgnet_v2.py:HGNetV2Stage.__init__": ["Identity", "Model", "ModelBasicLayer", "ModelConfig", "ModelConvLayer", "Module", "ModuleList", "None", "__init__", "activation", "append", "blocks", "blocks_list", "class", "config", "def", "downsample", "drop_path", "else", "float", "for", "groups", "i", "if", "in", "in_channels", "int", "kernel_size", "light_block", "mid_channels", "nn", "num_blocks", "num_layers", "out_channels", "range", "residual", "self", "stage_downsample", "stage_in_channels", "stage_index", "stage_kernel_size", "stage_light_block", "stage_mid_channels", "stage_num_blocks", "stage_numb_of_layers", "stage_out_channels", "stride", "super", "use_learnable_affine_block"], "hgnet_v2/modeling_hgnet_v2.py:HGNetV2Stage.forward": ["Model", "Module", "Tensor", "block", "blocks", "class", "def", "downsample", "for", "forward", "hidden_state", "in", "nn", "return", "self"], "hgnet_v2/modeling_hgnet_v2.py:HGNetV2Encoder.__init__": ["Model", "ModelConfig", "ModelStage", "Module", "ModuleList", "__init__", "append", "class", "config", "def", "for", "in", "len", "nn", "range", "resnet_stage", "self", "stage_in_channels", "stage_index", "stages", "super"], "hgnet_v2/modeling_hgnet_v2.py:HGNetV2Encoder.forward": ["BaseModelOutputWithNoAttention", "False", "Model", "Module", "None", "Tensor", "True", "bool", "class", "def", "else", "for", "forward", "hidden_state", "hidden_states", "if", "in", "is", "last_hidden_state", "nn", "not", "output_hidden_states", "return", "return_dict", "self", "stage", "stages", "tuple", "v"], "hgnet_v2/modeling_hgnet_v2.py:HGNetV2Backbone.__init__": ["BackboneMixin", "Model", "ModelConfig", "ModelEmbeddings", "ModelEncoder", "ModelPreTrainedModel", "__init__", "_init_backbone", "class", "config", "def", "depths", "embedder", "embedding_size", "encoder", "hidden_sizes", "num_features", "post_init", "self", "super"], "hgnet_v2/modeling_hgnet_v2.py:HGNetV2Backbone.forward": ["BackboneMixin", "BackboneOutput", "Model", "ModelPreTrainedModel", "None", "Tensor", "True", "attentions", "bool", "class", "config", "def", "else", "embedder", "embedding_output", "encoder", "enumerate", "feature_maps", "for", "forward", "hidden_states", "idx", "if", "in", "is", "kwargs", "not", "out_features", "output", "output_hidden_states", "outputs", "pixel_values", "r", "return", "return_dict", "self", "stage", "stage_names", "use_return_dict"], "hgnet_v2/modeling_hgnet_v2.py:HGNetV2ForImageClassification.__init__": ["AdaptiveAvgPool2d", "Flatten", "Identity", "Linear", "Model", "ModelConfig", "ModelEmbeddings", "ModelEncoder", "ModelPreTrainedModel", "ModuleList", "__init__", "avg_pool", "class", "classifier", "config", "def", "else", "embedder", "encoder", "fc", "flatten", "hidden_sizes", "if", "nn", "num_labels", "post_init", "self", "super"], "hgnet_v2/modeling_hgnet_v2.py:HGNetV2ForImageClassification.forward": ["FloatTensor", "ImageClassifierOutputWithNoAttention", "LongTensor", "Model", "ModelPreTrainedModel", "None", "bool", "class", "classifier", "config", "def", "else", "embedder", "embedding_output", "encoder", "fc", "for", "forward", "hidden_states", "if", "in", "is", "kwargs", "labels", "last_hidden_state", "layer", "logits", "loss", "loss_function", "not", "output", "output_hidden_states", "outputs", "pixel_values", "r", "return", "return_dict", "self", "torch", "use_return_dict"], "hiera/modeling_hiera.py:HieraPatchEmbeddings.__init__": ["Conv2d", "False", "ModelPatchEmbeddings", "Module", "The", "ValueError", "__init__", "be", "bool", "but", "class", "config", "def", "dimensions", "embed_dim", "f", "for", "got", "i", "if", "image", "image_size", "in", "input", "is_mae", "kernel_size", "len", "mask_ratio", "mask_spatial_shape", "masked_unit_size", "nn", "num_channels", "number", "of", "padding", "patch_padding", "patch_size", "patch_stride", "projection", "raise", "s", "self", "should", "spatial_dims", "stride", "super", "the", "tokens_spatial_shape", "zip"], "hiera/modeling_hiera.py:HieraPatchEmbeddings.masked_conv": ["BoolTensor", "FloatTensor", "ModelPatchEmbeddings", "Module", "None", "Tensor", "bool_masked_pos", "class", "def", "float", "functional", "if", "interpolate", "is", "mask_spatial_shape", "masked_conv", "nn", "pixel_values", "projection", "return", "self", "shape", "size", "target_size", "torch", "view"], "hiera/modeling_hiera.py:HieraPatchEmbeddings.random_masking": ["BoolTensor", "FloatTensor", "LongTensor", "ModelPatchEmbeddings", "Module", "None", "argsort", "batch_size", "bool", "bool_masked_pos", "class", "def", "device", "dim", "gather", "ids_restore", "ids_shuffle", "if", "index", "int", "is", "len_keep", "mask_ratio", "mask_spatial_shape", "math", "nn", "noise", "num_windows", "pixel_values", "prod", "rand", "random_masking", "return", "self", "shape", "to", "torch", "tuple", "zeros"], "hiera/modeling_hiera.py:HieraPatchEmbeddings.forward": ["BoolTensor", "FloatTensor", "LongTensor", "ModelPatchEmbeddings", "Module", "None", "Tensor", "bool_masked_pos", "class", "def", "else", "embeddings", "flatten", "forward", "ids_restore", "if", "is_mae", "masked_conv", "nn", "noise", "pixel_values", "random_masking", "return", "self", "torch", "transpose", "tuple"], "hiera/modeling_hiera.py:HieraEmbeddings.__init__": ["False", "ModelConfig", "ModelEmbeddings", "ModelPatchEmbeddings", "Module", "None", "Parameter", "__init__", "bool", "class", "config", "def", "embed_dim", "for", "i", "image_size", "in", "is_mae", "mask_spatial_shape", "masked_unit_size", "math", "nn", "num_tokens", "patch_embeddings", "patch_stride", "position_embeddings", "prod", "s", "self", "super", "tokens_spatial_shape", "torch", "zeros", "zip"], "hiera/modeling_hiera.py:HieraEmbeddings.interpolate_pos_encoding": ["False", "ModelEmbeddings", "Module", "Tensor", "align_corners", "and", "bicubic", "class", "def", "dim", "embeddings", "functional", "height", "if", "int", "interpolate", "interpolate_pos_encoding", "is_tracing", "jit", "mode", "new_height", "new_width", "nn", "not", "num_patches", "num_positions", "patch_stride", "permute", "pos_embeds", "reshape", "return", "self", "shape", "size", "sqrt_num_positions", "torch", "torch_int", "view", "width"], "hiera/modeling_hiera.py:HieraEmbeddings.get_position_embedding": ["FloatTensor", "ModelEmbeddings", "Module", "Tensor", "bool", "class", "def", "else", "embeddings", "get_position_embedding", "height", "if", "int", "interpolate_pos_encoding", "nn", "position_embeddings", "return", "self", "torch", "width"], "hiera/modeling_hiera.py:HieraEmbeddings.forward": ["BoolTensor", "False", "FloatTensor", "LongTensor", "ModelEmbeddings", "Module", "None", "Tensor", "bool", "bool_masked_pos", "class", "def", "embeddings", "forward", "get_position_embedding", "height", "ids_restore", "interpolate_pos_encoding", "nn", "noise", "patch_embeddings", "pixel_values", "return", "self", "shape", "torch", "tuple", "width"], "hiera/modeling_hiera.py:HieraMaskUnitAttention.__init__": ["False", "Linear", "ModelMaskUnitAttention", "Module", "None", "__init__", "bool", "class", "def", "head_dim", "hidden_size", "hidden_size_output", "int", "nn", "num_heads", "proj", "qkv", "query_stride", "scale", "self", "super", "use_mask_unit_attn", "window_size"], "hiera/modeling_hiera.py:HieraMaskUnitAttention.forward": ["False", "ModelMaskUnitAttention", "Module", "None", "Tensor", "_", "attn_output", "attn_weights", "batch_size", "bool", "class", "def", "dim", "else", "forward", "head_dim", "hidden_size_output", "hidden_states", "if", "key", "max", "nn", "num_heads", "num_windows", "output_attentions", "permute", "proj", "qkv", "query", "query_stride", "reshape", "return", "scale", "self", "seq_len", "shape", "softmax", "torch", "transpose", "tuple", "unbind", "use_mask_unit_attn", "value", "values", "view", "window_size"], "hiera/modeling_hiera.py:drop_path": ["False", "Model_path", "Model_prob", "Tensor", "bool", "def", "device", "div", "dtype", "float", "floor_", "if", "input", "keep_prob", "ndim", "not", "or", "output", "rand", "random_tensor", "return", "shape", "torch", "training"], "hiera/modeling_hiera.py:HieraDropPath.__init__": ["ModelDropPath", "Module", "None", "__init__", "class", "def", "drop_prob", "float", "nn", "self", "super"], "hiera/modeling_hiera.py:HieraDropPath.forward": ["ModelDropPath", "Module", "Tensor", "class", "def", "drop_path", "drop_prob", "forward", "hidden_states", "nn", "return", "self", "torch", "training"], "hiera/modeling_hiera.py:HieraDropPath.extra_repr": ["ModelDropPath", "Module", "class", "def", "drop_prob", "extra_repr", "f", "nn", "p", "return", "self", "str"], "hiera/modeling_hiera.py:HieraMlp.__init__": ["ACT2FN", "Linear", "ModelMlp", "Module", "None", "__init__", "activation_fn", "class", "config", "def", "dim", "fc1", "fc2", "hidden_act", "int", "mlp_ratio", "nn", "self", "super"], "hiera/modeling_hiera.py:HieraMlp.forward": ["ModelMlp", "Module", "Tensor", "activation_fn", "class", "def", "fc1", "fc2", "forward", "hidden_states", "nn", "return", "self", "torch"], "hiera/modeling_hiera.py:HieraLayer.__init__": ["False", "Identity", "LayerNorm", "Linear", "ModelDropPath", "ModelLayer", "ModelMaskUnitAttention", "ModelMlp", "Module", "None", "__init__", "attn", "bool", "class", "config", "def", "drop_path", "else", "eps", "float", "hidden_size", "hidden_size_output", "if", "int", "layer_norm_eps", "layernorm_after", "layernorm_before", "mlp", "nn", "num_heads", "proj", "query_stride", "self", "super", "use_mask_unit_attn", "window_size"], "hiera/modeling_hiera.py:HieraLayer.forward": ["False", "ModelLayer", "Module", "None", "Tensor", "_", "attn", "attn_weights", "batch_size", "bool", "class", "def", "dim", "drop_path", "forward", "hidden_size", "hidden_size_output", "hidden_states", "hidden_states_norm", "if", "layernorm_after", "layernorm_before", "max", "mlp", "nn", "output_attentions", "proj", "query_stride", "residual", "return", "self", "seq_len", "shape", "torch", "tuple", "values", "view"], "hiera/modeling_hiera.py:HieraStage.__init__": ["False", "GradientCheckpointingLayer", "ModelLayer", "ModelStage", "ModuleList", "None", "__init__", "and", "bool", "class", "config", "def", "depth", "drop_path", "else", "float", "for", "hidden_size", "hidden_size_output", "i", "if", "in", "int", "is", "layers", "list", "masked_unit_attention", "nn", "not", "num_heads", "or", "previous_stage_used_masked_attention", "query_stride", "range", "self", "stage_num", "super", "use_mask_unit_attn", "window_size"], "hiera/modeling_hiera.py:HieraStage.forward": ["False", "GradientCheckpointingLayer", "ModelStage", "None", "Tensor", "attn_weights", "bool", "class", "def", "enumerate", "for", "forward", "hidden_states", "i", "in", "layer_module", "layers", "output_attentions", "return", "self", "torch", "tuple"], "hiera/modeling_hiera.py:undo_windowing": ["Model_windowing", "Tensor", "batch_size", "def", "for", "hidden_size", "hidden_states", "in", "int", "list", "mask_unit_shape", "mu", "num_mask_units", "permute", "reshape", "return", "s", "shape", "torch", "view", "zip"], "hiera/modeling_hiera.py:HieraEncoder.__init__": ["False", "ModelConfig", "ModelEncoder", "ModelStage", "Module", "ModuleList", "None", "__init__", "append", "class", "config", "cpu", "cumsum", "cumulative_depths", "def", "depth", "depths", "device", "dpr", "drop_path", "drop_path_rate", "else", "embed_dim", "embed_dim_multiplier", "enumerate", "for", "gradient_checkpointing", "hidden_size", "hidden_size_output", "i", "idx_stage", "if", "image_size", "in", "int", "item", "len", "linspace", "masked_unit_area", "masked_unit_attention", "masked_unit_size", "math", "nn", "num_heads", "num_query_pool", "patch_stride", "prod", "query_pool_layer", "query_stride", "query_stride_area", "query_strides", "range", "s", "schedule", "self", "stage", "stage_ends", "stage_num", "stage_size", "stages", "sum", "super", "tensor", "tolist", "torch", "total_depth", "unroll_schedule", "use_mask_unit_attn", "window_size", "x", "zip"], "hiera/modeling_hiera.py:HieraEncoder.reroll": ["BoolTensor", "ModelEncoder", "Module", "None", "Tensor", "batch_size", "bool_masked_pos", "class", "def", "for", "hidden_size", "hidden_states", "i", "if", "in", "int", "is", "len", "mask_unit_shape", "math", "nn", "not", "num_dim", "permute", "prod", "range", "reroll", "reshape", "return", "schedule", "self", "seq_len", "shape", "size", "stage_idx", "strides", "torch", "undo_windowing", "view"], "hiera/modeling_hiera.py:HieraEncoder.forward": ["BaseModelOutput", "BoolTensor", "False", "ModelEncoder", "ModelEncoderOutput", "Module", "None", "Tensor", "True", "all_hidden_states", "all_reshaped_hidden_states", "all_self_attentions", "attentions", "bool", "bool_masked_pos", "class", "def", "else", "enumerate", "for", "forward", "hidden_states", "i", "if", "in", "is", "last_hidden_state", "layer_outputs", "nn", "not", "output_attentions", "output_hidden_states", "reroll", "reshaped_hidden_states", "return", "return_dict", "self", "stage_idx", "stage_module", "stages", "torch", "tuple", "v"], "hiera/modeling_hiera.py:unroll": ["Model", "Tensor", "_", "batch_size", "current_size", "def", "flatten", "for", "hidden_size", "hidden_states", "i", "image_shape", "in", "int", "item", "len", "list", "math", "new_shape", "num_dims", "pair", "patch_stride", "permute", "prod", "range", "reshape", "return", "s", "schedule", "shape", "size", "strides", "torch", "tuple", "view", "zip"], "hiera/modeling_hiera.py:HieraPreTrainedModel._init_weights": ["Conv1d", "Conv2d", "LayerNorm", "Linear", "ModelDecoder", "ModelEmbeddings", "ModelPreTrainedModel", "None", "PreTrainedModel", "_init_weights", "bias", "class", "config", "constant_", "decoder_position_embeddings", "def", "elif", "if", "init", "initializer_range", "is", "isinstance", "layer_norm_init", "mask_token", "module", "nn", "not", "position_embeddings", "self", "std", "trunc_normal_", "weight"], "hiera/modeling_hiera.py:HieraPooler.__init__": ["AdaptiveAvgPool1d", "LayerNorm", "ModelConfig", "ModelPooler", "Module", "__init__", "class", "config", "def", "depths", "embed_dim", "embed_dim_multiplier", "eps", "int", "layer_norm_eps", "layernorm", "len", "nn", "num_features", "pooler", "self", "super"], "hiera/modeling_hiera.py:HieraPooler.forward": ["ModelPooler", "Module", "Tensor", "class", "def", "flatten", "forward", "hidden_states", "layernorm", "nn", "pooled_output", "pooler", "return", "self", "torch", "transpose"], "hiera/modeling_hiera.py:HieraModel.__init__": ["False", "ModelConfig", "ModelEmbeddings", "ModelEncoder", "ModelModel", "ModelPooler", "ModelPreTrainedModel", "None", "True", "__init__", "add_pooling_layer", "bool", "class", "config", "def", "depths", "else", "embed_dim", "embed_dim_multiplier", "embeddings", "encoder", "if", "int", "is_mae", "len", "num_features", "pooler", "post_init", "query_stride", "r", "self", "super", "unroll_schedule"], "hiera/modeling_hiera.py:HieraModel.get_input_embeddings": ["ModelModel", "ModelPatchEmbeddings", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "patch_embeddings", "return", "self"], "hiera/modeling_hiera.py:HieraModel.forward": ["BaseModelOutputWithPooling", "FloatTensor", "ModelModel", "ModelModelOutput", "ModelPreTrainedModel", "None", "Tensor", "ValueError", "You", "_", "attentions", "batch_size", "bool", "bool_masked_pos", "class", "config", "def", "else", "embedding_output", "embeddings", "encoder", "encoder_outputs", "forward", "have", "head_outputs", "hidden_size", "hidden_states", "ids_restore", "if", "image_shape", "interpolate_pos_encoding", "is", "kwargs", "last_hidden_state", "mask_unit_area", "masked_unit_size", "math", "noise", "not", "output_attentions", "output_hidden_states", "patch_stride", "pixel_values", "pooled_output", "pooler", "pooler_output", "positions", "prod", "r", "raise", "reshaped_hidden_states", "return", "return_dict", "schedule", "self", "sequence_output", "shape", "specify", "tile", "to", "torch", "tuple", "unroll", "unroll_schedule", "unsqueeze", "use_return_dict", "view"], "hiera/modeling_hiera.py:HieraDecoder.__init__": ["False", "LayerNorm", "Linear", "ModelConfig", "ModelDecoder", "ModelStage", "Module", "Parameter", "__init__", "class", "config", "decoder_block", "decoder_depth", "decoder_embeddings", "decoder_hidden_size", "decoder_norm", "decoder_num_heads", "decoder_position_embeddings", "decoder_pred", "def", "depth", "depths", "drop_path", "embed_dim", "embed_dim_multiplier", "eps", "for", "hidden_size", "hidden_size_output", "i", "image_size", "in", "int", "layer_norm_eps", "len", "mask_token", "mask_unit_spatial_shape_final", "masked_unit_size", "math", "nn", "num_channels", "num_features", "num_heads", "num_query_pool", "patch_stride", "pred_dim", "pred_stride", "prod", "query_stride", "s", "self", "super", "tokens_spatial_shape", "tokens_spatial_shape_final", "torch", "use_mask_unit_attn", "window_size", "zeros", "zip"], "hiera/modeling_hiera.py:HieraDecoder.forward": ["BoolTensor", "False", "ModelDecoder", "Module", "Tensor", "attn_weights", "batch_size", "bool", "bool_masked_pos", "class", "decoder_block", "decoder_embeddings", "decoder_hidden_size", "decoder_hidden_states", "decoder_norm", "decoder_position_embeddings", "decoder_pred", "def", "device", "dtype", "encoder_hidden_states", "expand", "flatten", "float", "forward", "hidden_states", "mask_token", "mask_tokens", "mask_unit_height", "mask_unit_spatial_shape_final", "mask_unit_width", "nn", "num_mask_units", "output_attentions", "reshape", "return", "self", "shape", "tokens_spatial_shape_final", "torch", "tuple", "undo_windowing", "view", "zeros"], "hiera/modeling_hiera.py:HieraMultiScaleHead.__init__": ["Conv2d", "Identity", "ModelConfig", "ModelMultiScaleHead", "Module", "ModuleList", "__init__", "append", "class", "config", "current_masked_unit_size", "def", "depths", "embed_dim", "embed_dim_multiplier", "for", "i", "idx", "in", "int", "kernel", "kernel_size", "len", "mask_unit_spatial_shape_final", "masked_unit_size", "multi_scale_fusion_heads", "nn", "num_query_pool", "query_stride", "range", "s", "self", "stage_dimensions", "stride", "super", "zip"], "hiera/modeling_hiera.py:HieraMultiScaleHead.apply_fusion_head": ["Identity", "ModelMultiScaleHead", "Module", "Tensor", "apply_fusion_head", "batch_size", "class", "def", "head", "hidden_size", "hidden_states", "if", "isinstance", "mask_unit_height", "mask_unit_height_final", "mask_unit_width", "mask_unit_width_final", "nn", "num_mask_units", "permute", "reshape", "return", "self", "shape", "torch"], "hiera/modeling_hiera.py:HieraMultiScaleHead.forward": ["ModelMultiScaleHead", "Module", "Tensor", "apply_fusion_head", "class", "def", "feature_map", "feature_maps", "for", "forward", "head", "hidden_states", "in", "list", "multi_scale_fusion_heads", "nn", "return", "self", "torch", "zip"], "hiera/modeling_hiera.py:HieraForPreTraining.__init__": ["False", "LayerNorm", "Model", "ModelConfig", "ModelDecoder", "ModelForPreTraining", "ModelModel", "ModelMultiScaleHead", "ModelPreTrainedModel", "None", "True", "__init__", "add_pooling_layer", "class", "config", "decoder", "def", "encoder_norm", "eps", "is_mae", "layer_norm_eps", "multiscale_fusion", "nn", "num_features", "post_init", "pred_stride", "self", "super"], "hiera/modeling_hiera.py:HieraForPreTraining.get_pixel_label_2d": ["BoolTensor", "ModelForPreTraining", "ModelPreTrainedModel", "Tensor", "True", "bool_masked_pos", "class", "config", "def", "dim", "flatten", "get_pixel_label_2d", "if", "keepdim", "label", "mean", "normalize_pixel_loss", "permute", "pixel_values", "pred_stride", "return", "self", "size", "torch", "unfold", "var"], "hiera/modeling_hiera.py:HieraForPreTraining.forward_loss": ["BoolTensor", "ModelForPreTraining", "ModelPreTrainedModel", "Tensor", "bool_masked_pos", "class", "def", "forward_loss", "get_pixel_label_2d", "label", "logits", "loss", "mean", "pixel_values", "return", "self", "torch"], "hiera/modeling_hiera.py:HieraForPreTraining.forward": ["FloatTensor", "Model", "ModelForPreTraining", "ModelForPreTrainingOutput", "ModelPreTrainedModel", "None", "Tensor", "True", "attentions", "bool", "bool_masked_pos", "class", "config", "decoder", "def", "else", "encoder_norm", "feature_maps", "forward", "forward_loss", "fused_hidden_states", "hidden_states", "ids_restore", "ids_to_restore", "if", "interpolate_pos_encoding", "is", "kwargs", "logits", "loss", "multiscale_fusion", "noise", "not", "num_query_pool", "output", "output_attentions", "output_hidden_states", "outputs", "pixel_values", "r", "reshaped_hidden_states", "return", "return_dict", "self", "torch", "tuple", "use_return_dict"], "hiera/modeling_hiera.py:HieraForImageClassification.__init__": ["False", "Identity", "Linear", "Model", "ModelConfig", "ModelForImageClassification", "ModelModel", "ModelPreTrainedModel", "None", "True", "__init__", "add_pooling_layer", "class", "classifier", "config", "def", "else", "if", "is_mae", "nn", "num_features", "num_labels", "post_init", "self", "super"], "hiera/modeling_hiera.py:HieraForImageClassification.forward": ["Model", "ModelForImageClassification", "ModelForImageClassificationOutput", "ModelPreTrainedModel", "None", "Tensor", "attentions", "bool", "class", "classifier", "config", "def", "else", "forward", "hidden_states", "if", "interpolate_pos_encoding", "is", "kwargs", "labels", "logits", "loss", "loss_function", "not", "output", "output_attentions", "output_hidden_states", "outputs", "pixel_values", "pooled_output", "r", "reshaped_hidden_states", "return", "return_dict", "self", "torch", "tuple", "use_return_dict"], "hiera/modeling_hiera.py:HieraBackbone.__init__": ["BackboneMixin", "False", "LayerNorm", "ModelBackbone", "ModelConfig", "ModelEmbeddings", "ModelEncoder", "ModelPreTrainedModel", "ModuleDict", "__init__", "_init_backbone", "_out_features", "channels", "class", "config", "def", "depths", "embed_dim", "embed_dim_multiplier", "embeddings", "encoder", "for", "hidden_states_norms", "i", "in", "int", "is_mae", "len", "nn", "num_channels", "num_features", "post_init", "range", "self", "stage", "super", "zip"], "hiera/modeling_hiera.py:HieraBackbone.get_input_embeddings": ["BackboneMixin", "ModelBackbone", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "patch_embeddings", "return", "self"], "hiera/modeling_hiera.py:HieraBackbone.forward": ["BackboneMixin", "BackboneOutput", "ModelBackbone", "ModelPreTrainedModel", "None", "Tensor", "True", "_", "attentions", "batch_size", "bool", "class", "config", "contiguous", "def", "else", "embedding_output", "embeddings", "encoder", "feature_maps", "for", "forward", "height", "hidden_state", "hidden_states", "hidden_states_norms", "if", "in", "is", "kwargs", "not", "num_channels", "out_features", "output", "output_attentions", "output_hidden_states", "outputs", "permute", "pixel_values", "return", "return_dict", "self", "shape", "stage", "stage_names", "torch", "use_return_dict", "view", "width", "zip"], "hubert/modeling_hubert.py:HubertPositionalConvEmbedding.__init__": ["ACT2FN", "BatchNorm1d", "Conv1d", "GatheredParameters", "ModelPositionalConvEmbedding", "ModelSamePadLayer", "Module", "None", "__init__", "activation", "batch_norm", "class", "config", "conv", "conv_pos_batch_norm", "deepspeed", "def", "dim", "else", "feat_extract_activation", "groups", "hasattr", "hidden_size", "if", "is_deepspeed_zero3_enabled", "kernel_size", "modifier_rank", "name", "nn", "num_conv_pos_embedding_groups", "num_conv_pos_embeddings", "original0", "original1", "padding", "parametrizations", "register_external_parameter", "self", "super", "utils", "weight", "weight_g", "weight_norm", "weight_v", "with", "zero"], "hubert/modeling_hubert.py:HubertPositionalConvEmbedding.forward": ["ModelPositionalConvEmbedding", "Module", "None", "activation", "batch_norm", "class", "conv", "def", "forward", "hidden_states", "if", "is", "nn", "not", "padding", "return", "self", "transpose"], "hubert/modeling_hubert.py:HubertSamePadLayer.__init__": ["ModelSamePadLayer", "Module", "__init__", "class", "def", "else", "if", "nn", "num_conv_pos_embeddings", "num_pad_remove", "self", "super"], "hubert/modeling_hubert.py:HubertSamePadLayer.forward": ["ModelSamePadLayer", "Module", "class", "def", "forward", "hidden_states", "if", "nn", "num_pad_remove", "return", "self"], "hubert/modeling_hubert.py:HubertNoLayerNormConvLayer.__init__": ["ACT2FN", "Conv1d", "GradientCheckpointingLayer", "ModelNoLayerNormConvLayer", "__init__", "activation", "bias", "class", "config", "conv", "conv_bias", "conv_dim", "conv_kernel", "conv_stride", "def", "else", "feat_extract_activation", "if", "in_conv_dim", "kernel_size", "layer_id", "nn", "out_conv_dim", "self", "stride", "super"], "hubert/modeling_hubert.py:HubertNoLayerNormConvLayer.forward": ["GradientCheckpointingLayer", "ModelNoLayerNormConvLayer", "activation", "class", "conv", "def", "forward", "hidden_states", "return", "self"], "hubert/modeling_hubert.py:HubertLayerNormConvLayer.__init__": ["ACT2FN", "Conv1d", "GradientCheckpointingLayer", "LayerNorm", "ModelLayerNormConvLayer", "True", "__init__", "activation", "bias", "class", "config", "conv", "conv_bias", "conv_dim", "conv_kernel", "conv_stride", "def", "elementwise_affine", "else", "feat_extract_activation", "if", "in_conv_dim", "kernel_size", "layer_id", "layer_norm", "nn", "out_conv_dim", "self", "stride", "super"], "hubert/modeling_hubert.py:HubertLayerNormConvLayer.forward": ["GradientCheckpointingLayer", "ModelLayerNormConvLayer", "activation", "class", "conv", "def", "forward", "hidden_states", "layer_norm", "return", "self", "transpose"], "hubert/modeling_hubert.py:HubertGroupNormConvLayer.__init__": ["ACT2FN", "Conv1d", "GradientCheckpointingLayer", "GroupNorm", "ModelGroupNormConvLayer", "True", "__init__", "activation", "affine", "bias", "class", "config", "conv", "conv_bias", "conv_dim", "conv_kernel", "conv_stride", "def", "else", "feat_extract_activation", "if", "in_conv_dim", "kernel_size", "layer_id", "layer_norm", "nn", "num_channels", "num_groups", "out_conv_dim", "self", "stride", "super"], "hubert/modeling_hubert.py:HubertGroupNormConvLayer.forward": ["GradientCheckpointingLayer", "ModelGroupNormConvLayer", "activation", "class", "conv", "def", "forward", "hidden_states", "layer_norm", "return", "self"], "hubert/modeling_hubert.py:HubertFeatureEncoder.__init__": ["False", "ModelFeatureEncoder", "ModelGroupNormConvLayer", "ModelLayerNormConvLayer", "ModelNoLayerNormConvLayer", "Module", "ModuleList", "True", "ValueError", "__init__", "_requires_grad", "be", "but", "class", "config", "conv_layers", "def", "elif", "else", "f", "feat_extract_norm", "for", "gradient_checkpointing", "group", "has", "i", "if", "in", "is", "layer", "layer_id", "nn", "num_feat_extract_layers", "of", "one", "raise", "range", "self", "super", "to"], "hubert/modeling_hubert.py:HubertFeatureEncoder._freeze_parameters": ["False", "ModelFeatureEncoder", "Module", "_freeze_parameters", "_requires_grad", "class", "def", "for", "in", "nn", "param", "parameters", "requires_grad", "self"], "hubert/modeling_hubert.py:HubertFeatureEncoder.forward": ["ModelFeatureEncoder", "Module", "None", "True", "_requires_grad", "and", "class", "conv_layer", "conv_layers", "def", "for", "forward", "hidden_states", "if", "in", "input_values", "nn", "requires_grad", "return", "self", "training"], "hubert/modeling_hubert.py:HubertFeatureProjection.__init__": ["Dropout", "LayerNorm", "Linear", "ModelFeatureProjection", "Module", "__init__", "class", "config", "conv_dim", "def", "dropout", "eps", "feat_proj_dropout", "feat_proj_layer_norm", "hidden_size", "if", "layer_norm", "layer_norm_eps", "nn", "projection", "self", "super"], "hubert/modeling_hubert.py:HubertFeatureProjection.forward": ["ModelFeatureProjection", "Module", "class", "def", "dropout", "feat_proj_layer_norm", "forward", "hidden_states", "if", "layer_norm", "nn", "projection", "return", "self"], "hubert/modeling_hubert.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "contiguous", "def", "dim", "dropout", "float", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "shape", "size", "softmax", "torch", "training", "transpose", "value"], "hubert/modeling_hubert.py:HubertAttention.__init__": ["False", "Linear", "ModelAttention", "ModelConfig", "Module", "None", "True", "ValueError", "__init__", "and", "be", "bias", "bool", "by", "class", "config", "def", "divisible", "dropout", "embed_dim", "f", "float", "got", "head_dim", "if", "int", "is_causal", "is_decoder", "k_proj", "must", "nn", "num_heads", "out_proj", "q_proj", "raise", "scaling", "self", "super", "v_proj"], "hubert/modeling_hubert.py:HubertAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "False", "FlashAttentionKwargs", "ModelAttention", "Module", "None", "Tensor", "Unpack", "_attn_implementation", "attention_interface", "attention_mask", "attn_output", "attn_weights", "bool", "bsz", "class", "config", "contiguous", "current_states", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "head_dim", "hidden_states", "if", "is", "is_cross_attention", "k_proj", "key_states", "key_value_states", "kv_input_shape", "kwargs", "nn", "not", "out_proj", "output_attentions", "q_input_shape", "q_proj", "query_states", "reshape", "return", "scaling", "self", "shape", "src_len", "tgt_len", "torch", "training", "transpose", "tuple", "v_proj", "value_states", "view"], "hubert/modeling_hubert.py:HubertFeedForward.__init__": ["ACT2FN", "Dropout", "Linear", "ModelFeedForward", "Module", "__init__", "activation_dropout", "class", "config", "def", "else", "hidden_act", "hidden_dropout", "hidden_size", "if", "intermediate_act_fn", "intermediate_dense", "intermediate_dropout", "intermediate_size", "isinstance", "nn", "output_dense", "output_dropout", "self", "str", "super"], "hubert/modeling_hubert.py:HubertFeedForward.forward": ["ModelFeedForward", "Module", "class", "def", "forward", "hidden_states", "intermediate_act_fn", "intermediate_dense", "intermediate_dropout", "nn", "output_dense", "output_dropout", "return", "self"], "hubert/modeling_hubert.py:HubertEncoderLayer.__init__": ["Dropout", "False", "GradientCheckpointingLayer", "LayerNorm", "ModelAttention", "ModelEncoderLayer", "ModelFeedForward", "__init__", "attention", "attention_dropout", "class", "config", "def", "dropout", "embed_dim", "eps", "feed_forward", "final_layer_norm", "hidden_dropout", "hidden_size", "is_decoder", "layer_norm", "layer_norm_eps", "nn", "num_attention_heads", "num_heads", "self", "super"], "hubert/modeling_hubert.py:HubertEncoderLayer.forward": ["False", "GradientCheckpointingLayer", "ModelEncoderLayer", "None", "_", "attention", "attention_mask", "attn_residual", "attn_weights", "class", "def", "dropout", "feed_forward", "final_layer_norm", "forward", "hidden_states", "if", "layer_norm", "output_attentions", "outputs", "return", "self"], "hubert/modeling_hubert.py:HubertEncoder.__init__": ["Dropout", "False", "LayerNorm", "ModelEncoder", "ModelEncoderLayer", "ModelPositionalConvEmbedding", "Module", "ModuleList", "_", "__init__", "class", "config", "def", "dropout", "eps", "for", "gradient_checkpointing", "hidden_dropout", "hidden_size", "in", "layer_norm", "layer_norm_eps", "layers", "nn", "num_hidden_layers", "pos_conv_embed", "range", "self", "super"], "hubert/modeling_hubert.py:HubertEncoder.forward": ["BaseModelOutput", "False", "ModelEncoder", "Module", "None", "Tensor", "True", "all_hidden_states", "all_self_attentions", "and", "attention_mask", "attentions", "bool", "class", "config", "create_bidirectional_mask", "def", "dropout", "dropout_probability", "else", "expand_attention_mask", "for", "forward", "hidden_states", "if", "in", "input_embeds", "is", "is_deepspeed_zero3_enabled", "is_fsdp_managed_module", "last_hidden_state", "layer", "layer_norm", "layer_outputs", "layerdrop", "layers", "nn", "not", "or", "output_attentions", "output_hidden_states", "pos_conv_embed", "position_embeddings", "rand", "repeat", "return", "return_dict", "self", "shape", "skip_the_layer", "synced_gpus", "tensor", "torch", "training", "tuple", "unsqueeze", "v"], "hubert/modeling_hubert.py:HubertAttnAdapterLayer.__init__": ["LayerNorm", "Linear", "ModelAttnAdapterLayer", "Module", "ReLU", "__init__", "act_fn", "adapter_attn_dim", "class", "config", "def", "hidden_dim", "hidden_size", "input_dim", "linear_1", "linear_2", "nn", "norm", "self", "super"], "hubert/modeling_hubert.py:HubertAttnAdapterLayer.forward": ["FloatTensor", "ModelAttnAdapterLayer", "Module", "act_fn", "class", "def", "forward", "hidden_states", "linear_1", "linear_2", "nn", "norm", "return", "self", "torch"], "hubert/modeling_hubert.py:HubertEncoderLayerStableLayerNorm.__init__": ["Dropout", "False", "GradientCheckpointingLayer", "LayerNorm", "ModelAttention", "ModelAttnAdapterLayer", "ModelEncoderLayerStableLayerNorm", "ModelFeedForward", "None", "__init__", "adapter_attn_dim", "adapter_layer", "attention", "attention_dropout", "class", "config", "def", "dropout", "else", "embed_dim", "eps", "feed_forward", "final_layer_norm", "getattr", "hidden_dropout", "hidden_size", "if", "is", "is_decoder", "layer_norm", "layer_norm_eps", "nn", "not", "num_attention_heads", "num_heads", "self", "super"], "hubert/modeling_hubert.py:HubertEncoderLayerStableLayerNorm.forward": ["False", "GradientCheckpointingLayer", "ModelEncoderLayerStableLayerNorm", "None", "Tensor", "_", "adapter_layer", "attention", "attention_mask", "attn_residual", "attn_weights", "bool", "class", "def", "dropout", "feed_forward", "final_layer_norm", "forward", "hidden_states", "if", "is", "layer_norm", "not", "output_attentions", "outputs", "return", "self", "torch"], "hubert/modeling_hubert.py:HubertEncoderStableLayerNorm.__init__": ["Dropout", "False", "LayerNorm", "ModelEncoderLayerStableLayerNorm", "ModelEncoderStableLayerNorm", "ModelPositionalConvEmbedding", "Module", "ModuleList", "_", "__init__", "class", "config", "def", "dropout", "eps", "for", "gradient_checkpointing", "hidden_dropout", "hidden_size", "in", "layer_norm", "layer_norm_eps", "layers", "nn", "num_hidden_layers", "pos_conv_embed", "range", "self", "super"], "hubert/modeling_hubert.py:HubertEncoderStableLayerNorm.forward": ["BaseModelOutput", "False", "ModelEncoderStableLayerNorm", "Module", "None", "True", "all_hidden_states", "all_self_attentions", "and", "attention_mask", "attentions", "class", "config", "create_bidirectional_mask", "def", "dropout", "dropout_probability", "else", "expand_attention_mask", "for", "forward", "hidden_states", "if", "in", "input_embeds", "is", "is_deepspeed_zero3_enabled", "is_fsdp_managed_module", "last_hidden_state", "layer", "layer_norm", "layer_outputs", "layerdrop", "layers", "nn", "not", "or", "output_attentions", "output_hidden_states", "pos_conv_embed", "position_embeddings", "rand", "repeat", "return", "return_dict", "self", "shape", "skip_the_layer", "synced_gpus", "torch", "training", "tuple", "unsqueeze", "v"], "hubert/modeling_hubert.py:HubertPreTrainedModel._init_weights": ["BatchNorm1d", "Conv1d", "GatheredParameters", "GroupNorm", "LayerNorm", "Linear", "ModelForSequenceClassification", "ModelModel", "ModelPreTrainedModel", "None", "PreTrainedModel", "_init_weights", "and", "bias", "class", "config", "constant_", "deepspeed", "def", "elif", "else", "getattr", "hasattr", "if", "init", "initializer_range", "is", "is_deepspeed_zero3_enabled", "isinstance", "kaiming_normal_", "layer_weights", "masked_spec_embed", "mean", "modifier_rank", "module", "nn", "normal_", "not", "num_batches_tracked", "num_hidden_layers", "ones_", "running_mean", "running_var", "self", "std", "uniform_", "weight", "weight_g", "weight_v", "with", "zero", "zeros_"], "hubert/modeling_hubert.py:HubertPreTrainedModel._get_feat_extract_output_lengths": ["LongTensor", "ModelPreTrainedModel", "PreTrainedModel", "_conv_out_length", "_get_feat_extract_output_lengths", "class", "config", "conv_kernel", "conv_stride", "def", "div", "floor", "for", "in", "input_length", "input_lengths", "int", "kernel_size", "return", "rounding_mode", "self", "stride", "torch", "zip"], "hubert/modeling_hubert.py:HubertPreTrainedModel._get_feature_vector_attention_mask": ["LongTensor", "ModelPreTrainedModel", "PreTrainedModel", "_get_feat_extract_output_lengths", "_get_feature_vector_attention_mask", "arange", "attention_mask", "batch_size", "bool", "class", "cumsum", "def", "device", "dtype", "feature_vector_length", "flip", "int", "long", "output_lengths", "return", "self", "shape", "sum", "to", "torch", "zeros"], "hubert/modeling_hubert.py:_compute_mask_indices": ["False", "LongTensor", "None", "ValueError", "_", "_compute_mask_indices", "and", "append", "arange", "array", "attention_mask", "batch_size", "be", "bigger", "bool", "broadcast_to", "but", "choice", "compute_num_masked_span", "concatenate", "def", "detach", "dtype", "dummy_mask_idx", "else", "epsilon", "f", "float", "for", "got", "has", "if", "in", "input_length", "input_lengths", "int", "int32", "is", "item", "len", "mask_length", "mask_prob", "max", "max_num_masked_span", "min_masks", "ndarray", "not", "np", "num_masked_span", "offsets", "ones", "put_along_axis", "raise", "rand", "random", "range", "replace", "reshape", "return", "sequence_length", "shape", "smaller", "spec_aug_mask", "spec_aug_mask_idx", "spec_aug_mask_idxs", "sum", "than", "to", "tolist", "torch", "tuple", "zeros"], "hubert/modeling_hubert.py:HubertModel.__init__": ["ModelConfig", "ModelEncoder", "ModelEncoderStableLayerNorm", "ModelFeatureEncoder", "ModelFeatureProjection", "ModelModel", "ModelPreTrainedModel", "Parameter", "Tensor", "__init__", "class", "config", "def", "do_stable_layer_norm", "else", "encoder", "feature_extractor", "feature_projection", "hidden_size", "if", "mask_feature_prob", "mask_time_prob", "masked_spec_embed", "nn", "or", "post_init", "self", "super", "torch", "uniform_"], "hubert/modeling_hubert.py:HubertModel._mask_hidden_states": ["FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "True", "_compute_mask_indices", "_mask_hidden_states", "and", "apply_spec_augment", "attention_mask", "batch_size", "bool", "class", "config", "def", "device", "dtype", "elif", "expand", "getattr", "hidden_size", "hidden_states", "if", "is", "mask_feature_indices", "mask_feature_length", "mask_feature_min_masks", "mask_feature_prob", "mask_length", "mask_prob", "mask_time_indices", "mask_time_length", "mask_time_min_masks", "mask_time_prob", "masked_spec_embed", "min_masks", "not", "return", "self", "sequence_length", "size", "tensor", "to", "torch", "training"], "hubert/modeling_hubert.py:HubertModel.forward": ["BaseModelOutput", "FloatTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "_get_feature_vector_attention_mask", "_mask_hidden_states", "attention_mask", "attentions", "bool", "class", "config", "def", "else", "encoder", "encoder_outputs", "extract_features", "feature_extractor", "feature_projection", "forward", "hidden_states", "if", "input_values", "is", "kwargs", "last_hidden_state", "mask_time_indices", "not", "output_attentions", "output_hidden_states", "r", "return", "return_dict", "self", "shape", "torch", "transpose", "tuple", "use_return_dict"], "hubert/modeling_hubert.py:HubertForCTC.__init__": ["Dropout", "Linear", "Model", "ModelForCTC", "ModelModel", "ModelPreTrainedModel", "None", "Please", "ValueError", "You", "__class__", "__init__", "a", "add_adapter", "and", "are", "as", "class", "config", "configuration", "def", "define", "does", "dropout", "else", "f", "final_dropout", "follows", "from_pretrained", "hasattr", "head", "hidden_size", "if", "instantiate", "is", "language", "lm_head", "model", "nn", "not", "of", "or", "output_hidden_size", "post_init", "r", "raise", "s", "self", "size", "str", "super", "target_lang", "that", "the", "to", "trying", "vocab_size", "vocabulary", "with", "your"], "hubert/modeling_hubert.py:HubertForCTC.tie_weights": ["By", "Cannot", "ModelForCTC", "ModelPreTrainedModel", "None", "True", "ValueError", "adapter_attn_dim", "and", "class", "config", "def", "default", "defined", "elif", "eng", "f", "force_load", "getattr", "if", "info", "is", "kwargs", "load_adapter", "logger", "not", "pass", "raise", "self", "set", "target_lang", "tie_weights", "to"], "hubert/modeling_hubert.py:HubertForCTC.freeze_feature_encoder": ["Model", "ModelForCTC", "ModelPreTrainedModel", "_freeze_parameters", "class", "def", "feature_extractor", "freeze_feature_encoder", "self"], "hubert/modeling_hubert.py:HubertForCTC.freeze_base_model": ["False", "Model", "ModelForCTC", "ModelPreTrainedModel", "class", "def", "for", "freeze_base_model", "in", "param", "parameters", "requires_grad", "self"], "hubert/modeling_hubert.py:HubertForCTC.forward": ["CausalLMOutput", "False", "Label", "Model", "ModelForCTC", "ModelPreTrainedModel", "None", "Tensor", "ValueError", "_HIDDEN_STATES_START_POSITION", "_get_feat_extract_output_lengths", "and", "attention_mask", "attentions", "backends", "be", "blank", "bool", "class", "config", "ctc_loss", "ctc_loss_reduction", "ctc_zero_infinity", "cudnn", "def", "dim", "dropout", "dtype", "else", "enabled", "f", "flags", "flattened_targets", "float32", "forward", "functional", "hidden_states", "if", "input_lengths", "input_values", "is", "kwargs", "labels", "labels_mask", "lm_head", "log_probs", "log_softmax", "logits", "long", "loss", "masked_select", "max", "must", "nn", "not", "ones_like", "output", "output_attentions", "output_hidden_states", "outputs", "pad_token_id", "r", "raise", "reduction", "return", "return_dict", "self", "sum", "target_lengths", "to", "torch", "transpose", "tuple", "use_return_dict", "values", "vocab_size", "with", "zero_infinity"], "hubert/modeling_hubert.py:HubertForSequenceClassification.__init__": ["Linear", "Model", "ModelForSequenceClassification", "ModelModel", "ModelPreTrainedModel", "Parameter", "Sequence", "True", "ValueError", "__init__", "adapters", "add_adapter", "and", "class", "classification", "classifier", "classifier_proj_size", "config", "def", "does", "hasattr", "hidden_size", "if", "layer_weights", "nn", "not", "num_hidden_layers", "num_labels", "num_layers", "of", "ones", "post_init", "projector", "raise", "self", "super", "support", "the", "torch", "use", "use_weighted_layer_sum"], "hubert/modeling_hubert.py:HubertForSequenceClassification.freeze_feature_encoder": ["Model", "ModelForSequenceClassification", "ModelPreTrainedModel", "_freeze_parameters", "class", "def", "feature_extractor", "freeze_feature_encoder", "self"], "hubert/modeling_hubert.py:HubertForSequenceClassification.freeze_base_model": ["False", "Model", "ModelForSequenceClassification", "ModelPreTrainedModel", "class", "def", "for", "freeze_base_model", "in", "param", "parameters", "requires_grad", "self"], "hubert/modeling_hubert.py:HubertForSequenceClassification.forward": ["CrossEntropyLoss", "Model", "ModelForSequenceClassification", "ModelPreTrainedModel", "None", "SequenceClassifierOutput", "Tensor", "True", "_HIDDEN_STATES_START_POSITION", "_get_feature_vector_attention_mask", "attention_mask", "attentions", "bool", "class", "classifier", "config", "def", "dim", "else", "expand_padding_mask", "forward", "functional", "hidden_states", "if", "input_values", "is", "kwargs", "labels", "layer_weights", "logits", "loss", "loss_fct", "mean", "nn", "norm_weights", "not", "num_labels", "output", "output_attentions", "output_hidden_states", "outputs", "padding_mask", "pooled_output", "projector", "r", "repeat", "return", "return_dict", "self", "shape", "softmax", "stack", "sum", "torch", "tuple", "unsqueeze", "use_return_dict", "use_weighted_layer_sum", "view"], "hunyuan_v1_dense/modeling_hunyuan_v1_dense.py:HunYuanDenseV1RMSNorm.__init__": ["ModelYuanDenseV1RMSNorm", "Module", "Parameter", "__init__", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "hunyuan_v1_dense/modeling_hunyuan_v1_dense.py:HunYuanDenseV1RMSNorm.forward": ["ModelYuanDenseV1RMSNorm", "Module", "True", "class", "def", "dtype", "float32", "forward", "hidden_states", "input_dtype", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "hunyuan_v1_dense/modeling_hunyuan_v1_dense.py:HunYuanDenseV1RMSNorm.extra_repr": ["ModelYuanDenseV1RMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "variance_epsilon", "weight"], "hunyuan_v1_dense/modeling_hunyuan_v1_dense.py:HunYuanDenseV1MLP.__init__": ["ACT2FN", "False", "Linear", "ModelYuanDenseV1Config", "ModelYuanDenseV1MLP", "Module", "None", "__init__", "act_fn", "bias", "class", "config", "def", "down_proj", "gate_proj", "hidden_act", "hidden_size", "intermediate_size", "is_shared_mlp", "layer_idx", "nn", "self", "super", "up_proj"], "hunyuan_v1_dense/modeling_hunyuan_v1_dense.py:HunYuanDenseV1MLP.forward": ["ModelYuanDenseV1MLP", "Module", "act_fn", "class", "def", "down_proj", "forward", "gate_proj", "nn", "return", "self", "up_proj", "x"], "hunyuan_v1_dense/modeling_hunyuan_v1_dense.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "hunyuan_v1_dense/modeling_hunyuan_v1_dense.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cos", "def", "k", "k_embed", "q", "q_embed", "return", "rotate_half", "sin", "unsqueeze", "unsqueeze_dim"], "hunyuan_v1_dense/modeling_hunyuan_v1_dense.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "hunyuan_v1_dense/modeling_hunyuan_v1_dense.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value", "value_states"], "hunyuan_v1_dense/modeling_hunyuan_v1_dense.py:HunYuanDenseV1Attention.__init__": ["Linear", "ModelYuanDenseV1Attention", "ModelYuanDenseV1Config", "ModelYuanDenseV1RMSNorm", "Module", "True", "__init__", "attention_bias", "attention_dropout", "bias", "class", "config", "def", "eps", "getattr", "head_dim", "hidden_size", "int", "is_causal", "k_proj", "key_layernorm", "layer_idx", "nn", "num_attention_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "q_proj", "query_layernorm", "rms_norm_eps", "scaling", "self", "super", "v_proj"], "hunyuan_v1_dense/modeling_hunyuan_v1_dense.py:HunYuanDenseV1Attention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "LongTensor", "ModelYuanDenseV1Attention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_attn_implementation", "apply_rotary_pos_emb", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "cache_kwargs", "cache_position", "class", "config", "contiguous", "cos", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "head_dim", "hidden_shape", "hidden_states", "if", "input_shape", "is", "k_proj", "key_layernorm", "key_states", "kwargs", "layer_idx", "nn", "not", "o_proj", "past_key_values", "position_embeddings", "q_proj", "query_layernorm", "query_states", "reshape", "return", "scaling", "self", "shape", "sin", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "view"], "hunyuan_v1_dense/modeling_hunyuan_v1_dense.py:HunYuanDenseV1DecoderLayer.__init__": ["GradientCheckpointingLayer", "ModelYuanDenseV1Attention", "ModelYuanDenseV1Config", "ModelYuanDenseV1DecoderLayer", "ModelYuanDenseV1MLP", "ModelYuanDenseV1RMSNorm", "__init__", "class", "config", "def", "eps", "hidden_size", "input_layernorm", "int", "layer_idx", "mlp", "post_attention_layernorm", "rms_norm_eps", "self", "self_attn", "super"], "hunyuan_v1_dense/modeling_hunyuan_v1_dense.py:HunYuanDenseV1DecoderLayer.forward": ["Cache", "False", "GradientCheckpointingLayer", "LongTensor", "ModelYuanDenseV1DecoderLayer", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "attention_mask", "bool", "cache_position", "class", "def", "forward", "hidden_states", "input_layernorm", "kwargs", "mlp", "past_key_values", "position_embeddings", "position_ids", "post_attention_layernorm", "residual", "return", "self", "self_attn", "torch", "tuple", "use_cache"], "hunyuan_v1_dense/modeling_hunyuan_v1_dense.py:HunYuanDenseV1RotaryEmbedding.__init__": ["Callable", "False", "ModelYuanDenseV1Config", "ModelYuanDenseV1RotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "alpha", "and", "arange", "attention_scaling", "base", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "dim", "dynamic", "else", "float", "get", "head_dim", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_theta", "rope_type", "self", "super", "to", "torch"], "hunyuan_v1_dense/modeling_hunyuan_v1_dense.py:HunYuanDenseV1RotaryEmbedding.compute_default_rope_parameters": ["ModelYuanDenseV1Config", "ModelYuanDenseV1RotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "hunyuan_v1_dense/modeling_hunyuan_v1_dense.py:HunYuanDenseV1RotaryEmbedding.forward": ["False", "ModelYuanDenseV1RotaryEmbedding", "Module", "None", "and", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "hunyuan_v1_dense/modeling_hunyuan_v1_dense.py:HunYuanDenseV1Model.__init__": ["Embedding", "False", "ModelYuanDenseV1Config", "ModelYuanDenseV1DecoderLayer", "ModelYuanDenseV1Model", "ModelYuanDenseV1PreTrainedModel", "ModelYuanDenseV1RMSNorm", "ModelYuanDenseV1RotaryEmbedding", "ModuleList", "__init__", "class", "config", "def", "embed_tokens", "eps", "for", "gradient_checkpointing", "hidden_size", "in", "layer_idx", "layers", "nn", "norm", "num_hidden_layers", "pad_token_id", "padding_idx", "post_init", "range", "rms_norm_eps", "rotary_emb", "self", "super", "vocab_size"], "hunyuan_v1_dense/modeling_hunyuan_v1_dense.py:HunYuanDenseV1Model.forward": ["BaseModelOutputWithPast", "Cache", "DynamicCache", "FloatTensor", "LongTensor", "ModelYuanDenseV1Model", "ModelYuanDenseV1PreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "and", "arange", "attention_mask", "bool", "cache_position", "causal_mask", "class", "config", "create_causal_mask", "decoder_layer", "def", "device", "else", "embed_tokens", "exactly", "for", "forward", "get_seq_length", "hidden_states", "if", "in", "input_embeds", "input_ids", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layers", "must", "norm", "not", "num_hidden_layers", "of", "one", "or", "past_key_values", "past_seen_tokens", "position_embeddings", "position_ids", "raise", "return", "rotary_emb", "self", "shape", "specify", "torch", "unsqueeze", "use_cache"], "hunyuan_v1_dense/modeling_hunyuan_v1_dense.py:HunYuanDenseV1ForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "ModelYuanDenseV1ForCausalLM", "ModelYuanDenseV1Model", "ModelYuanDenseV1PreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "vocab_size"], "hunyuan_v1_dense/modeling_hunyuan_v1_dense.py:HunYuanDenseV1ForCausalLM.forward": ["BaseModelOutputWithPast", "Cache", "CausalLMOutputWithPast", "FloatTensor", "GenerationMixin", "LongTensor", "ModelYuanDenseV1ForCausalLM", "ModelYuanDenseV1PreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "outputs", "past_key_values", "position_ids", "r", "return", "self", "slice", "slice_indices", "torch", "use_cache", "vocab_size"], "hunyuan_v1_moe/modeling_hunyuan_v1_moe.py:HunYuanMoEV1RMSNorm.__init__": ["ModelYuanMoEV1RMSNorm", "Module", "Parameter", "__init__", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "hunyuan_v1_moe/modeling_hunyuan_v1_moe.py:HunYuanMoEV1RMSNorm.forward": ["ModelYuanMoEV1RMSNorm", "Module", "True", "class", "def", "dtype", "float32", "forward", "hidden_states", "input_dtype", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "hunyuan_v1_moe/modeling_hunyuan_v1_moe.py:HunYuanMoEV1RMSNorm.extra_repr": ["ModelYuanMoEV1RMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "variance_epsilon", "weight"], "hunyuan_v1_moe/modeling_hunyuan_v1_moe.py:HunYuanMoEV1MLP.__init__": ["ACT2FN", "False", "Linear", "ModelYuanMoEV1Config", "ModelYuanMoEV1MLP", "Module", "__init__", "act_fn", "bias", "class", "config", "def", "down_proj", "gate_proj", "hidden_act", "hidden_size", "intermediate_size", "nn", "self", "super", "up_proj"], "hunyuan_v1_moe/modeling_hunyuan_v1_moe.py:HunYuanMoEV1MLP.forward": ["ModelYuanMoEV1MLP", "Module", "act_fn", "class", "def", "down_proj", "forward", "gate_proj", "nn", "return", "self", "up_proj", "x"], "hunyuan_v1_moe/modeling_hunyuan_v1_moe.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "hunyuan_v1_moe/modeling_hunyuan_v1_moe.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cos", "def", "k", "k_embed", "q", "q_embed", "return", "rotate_half", "sin", "unsqueeze", "unsqueeze_dim"], "hunyuan_v1_moe/modeling_hunyuan_v1_moe.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "hunyuan_v1_moe/modeling_hunyuan_v1_moe.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value", "value_states"], "hunyuan_v1_moe/modeling_hunyuan_v1_moe.py:HunYuanMoEV1Attention.__init__": ["Linear", "ModelYuanMoEV1Attention", "ModelYuanMoEV1Config", "ModelYuanMoEV1RMSNorm", "Module", "True", "__init__", "attention_bias", "attention_dropout", "bias", "class", "config", "def", "eps", "getattr", "head_dim", "hidden_size", "int", "is_causal", "k_proj", "key_layernorm", "layer_idx", "nn", "num_attention_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "q_proj", "query_layernorm", "rms_norm_eps", "scaling", "self", "super", "v_proj"], "hunyuan_v1_moe/modeling_hunyuan_v1_moe.py:HunYuanMoEV1Attention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "LongTensor", "ModelYuanMoEV1Attention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_attn_implementation", "apply_rotary_pos_emb", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "cache_kwargs", "cache_position", "class", "config", "contiguous", "cos", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "head_dim", "hidden_shape", "hidden_states", "if", "input_shape", "is", "k_proj", "key_layernorm", "key_states", "kwargs", "layer_idx", "nn", "not", "o_proj", "past_key_values", "position_embeddings", "q_proj", "query_layernorm", "query_states", "reshape", "return", "scaling", "self", "shape", "sin", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "view"], "hunyuan_v1_moe/modeling_hunyuan_v1_moe.py:HunYuanMoEV1Gate.__init__": ["False", "Linear", "ModelYuanMoEV1Config", "ModelYuanMoEV1Gate", "Module", "None", "__init__", "bias", "class", "config", "def", "dtype", "else", "float32", "hidden_size", "if", "int", "isinstance", "layer_idx", "nn", "num_experts", "self", "super", "torch", "wg"], "hunyuan_v1_moe/modeling_hunyuan_v1_moe.py:HunYuanMoEV1Gate.forward": ["ModelYuanMoEV1Gate", "Module", "bsz", "class", "def", "dtype", "float", "float32", "forward", "hidden_size", "hidden_states", "if", "logits", "nn", "reshape", "return", "self", "seq_len", "shape", "torch", "weight", "wg"], "hunyuan_v1_moe/modeling_hunyuan_v1_moe.py:HunYuanMoEV1Experts.__init__": ["ACT2FN", "ModelYuanMoEV1Config", "ModelYuanMoEV1Experts", "Module", "Parameter", "__init__", "act_fn", "class", "config", "def", "down_proj", "empty", "gate_up_proj", "hidden_act", "hidden_dim", "hidden_size", "intermediate_dim", "intermediate_size", "nn", "num_experts", "num_local_experts", "self", "super", "torch"], "hunyuan_v1_moe/modeling_hunyuan_v1_moe.py:HunYuanMoEV1Experts.forward": ["ModelYuanMoEV1Experts", "Module", "None", "Tensor", "act_fn", "cModelk", "class", "continue", "current_hidden_states", "current_state", "def", "dim", "down_proj", "dtype", "expert_hit", "expert_idx", "expert_mask", "final_hidden_states", "for", "forward", "functional", "gate", "gate_up_proj", "greater", "hidden_states", "if", "in", "index_add_", "linear", "nn", "no_grad", "nonzero", "num_classes", "num_experts", "one_hot", "permute", "return", "self", "sum", "to", "token_idx", "top_k_index", "top_k_pos", "top_k_weights", "torch", "up", "where", "with", "zeros_like"], "hunyuan_v1_moe/modeling_hunyuan_v1_moe.py:HunYuanMoEV1Moe.__init__": ["ModelYuanMoEV1Config", "ModelYuanMoEV1Experts", "ModelYuanMoEV1Gate", "ModelYuanMoEV1MLP", "ModelYuanMoEV1Moe", "Module", "None", "__init__", "class", "config", "def", "else", "experts", "gate", "if", "int", "isinstance", "layer_idx", "moe_topk", "nn", "num_experts", "self", "shared_mlp", "super", "top_k"], "hunyuan_v1_moe/modeling_hunyuan_v1_moe.py:HunYuanMoEV1Moe.route_tokens_to_experts": ["F", "ModelYuanMoEV1Moe", "Module", "True", "class", "def", "dim", "dtype", "float", "float32", "hidden_states", "keepdim", "nn", "return", "route_tokens_to_experts", "routing_weights", "scatter_", "selected_experts", "self", "softmax", "sum", "to", "top_k", "topk", "torch", "zeros_like"], "hunyuan_v1_moe/modeling_hunyuan_v1_moe.py:HunYuanMoEV1Moe.forward": ["ModelYuanMoEV1Moe", "Module", "Tensor", "batch_size", "class", "def", "experts", "final_hidden_states", "forward", "gate", "hidden_dim", "hidden_states", "hidden_states_mlp", "nn", "reshape", "return", "route_tokens_to_experts", "router_logits", "routing_weights", "selected_experts", "self", "sequence_length", "shape", "shared_mlp", "torch", "view"], "hunyuan_v1_moe/modeling_hunyuan_v1_moe.py:HunYuanMoEV1DecoderLayer.__init__": ["GradientCheckpointingLayer", "ModelYuanMoEV1Attention", "ModelYuanMoEV1Config", "ModelYuanMoEV1DecoderLayer", "ModelYuanMoEV1Moe", "ModelYuanMoEV1RMSNorm", "__init__", "class", "config", "def", "eps", "hidden_size", "input_layernorm", "int", "layer_idx", "mlp", "post_attention_layernorm", "rms_norm_eps", "self", "self_attn", "super"], "hunyuan_v1_moe/modeling_hunyuan_v1_moe.py:HunYuanMoEV1DecoderLayer.forward": ["Cache", "False", "GradientCheckpointingLayer", "LongTensor", "ModelYuanMoEV1DecoderLayer", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "attention_mask", "bool", "cache_position", "class", "def", "forward", "hidden_states", "input_layernorm", "kwargs", "mlp", "past_key_values", "position_embeddings", "position_ids", "post_attention_layernorm", "residual", "return", "self", "self_attn", "torch", "tuple", "use_cache"], "hunyuan_v1_moe/modeling_hunyuan_v1_moe.py:HunYuanMoEV1PreTrainedModel._init_weights": ["ModelYuanMoEV1Experts", "ModelYuanMoEV1PreTrainedModel", "PreTrainedModel", "_init_weights", "class", "config", "def", "down_proj", "gate_up_proj", "if", "init", "initializer_range", "isinstance", "mean", "module", "normal_", "self", "std", "super"], "hunyuan_v1_moe/modeling_hunyuan_v1_moe.py:HunYuanMoEV1RotaryEmbedding.__init__": ["Callable", "False", "ModelYuanMoEV1Config", "ModelYuanMoEV1RotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "alpha", "and", "arange", "attention_scaling", "base", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "dim", "dynamic", "else", "float", "get", "head_dim", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_theta", "rope_type", "self", "super", "to", "torch"], "hunyuan_v1_moe/modeling_hunyuan_v1_moe.py:HunYuanMoEV1RotaryEmbedding.compute_default_rope_parameters": ["ModelYuanMoEV1Config", "ModelYuanMoEV1RotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "hunyuan_v1_moe/modeling_hunyuan_v1_moe.py:HunYuanMoEV1RotaryEmbedding.forward": ["False", "ModelYuanMoEV1RotaryEmbedding", "Module", "None", "and", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "hunyuan_v1_moe/modeling_hunyuan_v1_moe.py:HunYuanMoEV1Model.__init__": ["Embedding", "False", "ModelYuanMoEV1Config", "ModelYuanMoEV1DecoderLayer", "ModelYuanMoEV1Model", "ModelYuanMoEV1PreTrainedModel", "ModelYuanMoEV1RMSNorm", "ModelYuanMoEV1RotaryEmbedding", "ModuleList", "__init__", "class", "config", "def", "embed_tokens", "eps", "for", "gradient_checkpointing", "hidden_size", "in", "layer_idx", "layers", "nn", "norm", "num_hidden_layers", "pad_token_id", "padding_idx", "post_init", "range", "rms_norm_eps", "rotary_emb", "self", "super", "vocab_size"], "hunyuan_v1_moe/modeling_hunyuan_v1_moe.py:HunYuanMoEV1Model.forward": ["BaseModelOutputWithPast", "Cache", "DynamicCache", "FloatTensor", "LongTensor", "ModelYuanMoEV1Model", "ModelYuanMoEV1PreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "and", "arange", "attention_mask", "bool", "cache_position", "causal_mask", "class", "config", "create_causal_mask", "decoder_layer", "def", "device", "else", "embed_tokens", "exactly", "for", "forward", "get_seq_length", "hidden_states", "if", "in", "input_embeds", "input_ids", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layers", "must", "norm", "not", "num_hidden_layers", "of", "one", "or", "past_key_values", "past_seen_tokens", "position_embeddings", "position_ids", "raise", "return", "rotary_emb", "self", "shape", "specify", "torch", "unsqueeze", "use_cache"], "hunyuan_v1_moe/modeling_hunyuan_v1_moe.py:HunYuanMoEV1ForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "ModelYuanMoEV1ForCausalLM", "ModelYuanMoEV1Model", "ModelYuanMoEV1PreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "vocab_size"], "hunyuan_v1_moe/modeling_hunyuan_v1_moe.py:HunYuanMoEV1ForCausalLM.forward": ["BaseModelOutputWithPast", "Cache", "CausalLMOutputWithPast", "FloatTensor", "GenerationMixin", "LongTensor", "ModelYuanMoEV1ForCausalLM", "ModelYuanMoEV1PreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "outputs", "past_key_values", "position_ids", "r", "return", "self", "slice", "slice_indices", "torch", "use_cache", "vocab_size"], "ibert/modeling_ibert.py:IBertEmbeddings.__init__": ["Dropout", "False", "IntLayerNorm", "LayerNorm", "Model", "Module", "QuantAct", "QuantEmbedding", "__init__", "act_bit", "arange", "class", "config", "def", "dropout", "embedding_act_bit", "embedding_bit", "embeddings_act1", "embeddings_act2", "eps", "expand", "force_dequant", "hidden_dropout_prob", "hidden_size", "layer_norm_eps", "ln_input_bit", "ln_output_bit", "max_position_embeddings", "nn", "output_activation", "output_bit", "pad_token_id", "padding_idx", "persistent", "position_embeddings", "position_ids", "quant_mode", "register_buffer", "self", "super", "token_type_embeddings", "torch", "type_vocab_size", "vocab_size", "weight_bit", "word_embeddings"], "ibert/modeling_ibert.py:IBertEmbeddings.forward": ["LayerNorm", "Model", "Module", "None", "class", "create_position_ids_from_input_ids", "create_position_ids_from_inputs_embeds", "def", "device", "dropout", "dtype", "else", "embeddings", "embeddings_act1", "embeddings_scaling_factor", "forward", "identity", "identity_scaling_factor", "if", "input_ids", "input_shape", "inputs_embeds", "inputs_embeds_scaling_factor", "is", "long", "nn", "not", "output_activation", "padding_idx", "past_key_values_length", "position_embeddings", "position_embeddings_scaling_factor", "position_ids", "return", "self", "size", "to", "token_type_embeddings", "token_type_embeddings_scaling_factor", "token_type_ids", "torch", "word_embeddings", "zeros"], "ibert/modeling_ibert.py:IBertEmbeddings.create_position_ids_from_inputs_embeds": ["Model", "Module", "arange", "class", "create_position_ids_from_inputs_embeds", "def", "device", "dtype", "expand", "input_shape", "inputs_embeds", "long", "nn", "padding_idx", "position_ids", "return", "self", "sequence_length", "size", "torch", "unsqueeze"], "ibert/modeling_ibert.py:IBertSelfAttention.__init__": ["Dropout", "IntSoftmax", "Model", "Module", "QuantAct", "QuantLinear", "The", "True", "ValueError", "__init__", "a", "act_bit", "all_head_size", "and", "attention", "attention_head_size", "attention_probs_dropout_prob", "bias", "bias_bit", "class", "config", "def", "dropout", "embedding_size", "f", "force_dequant", "hasattr", "heads", "hidden", "hidden_size", "if", "int", "is", "key", "key_activation", "multiple", "nn", "not", "num_attention_heads", "number", "of", "output_activation", "per_channel", "quant_mode", "query", "query_activation", "raise", "self", "size", "softmax", "super", "the", "value", "value_activation", "weight_bit"], "ibert/modeling_ibert.py:IBertSelfAttention.forward": ["False", "Model", "Module", "None", "_", "all_head_size", "attention_head_size", "attention_mask", "attention_probs", "attention_probs_scaling_factor", "attention_scores", "attention_scores_scaling_factor", "batch_size", "class", "context_layer", "context_layer_scaling_factor", "contiguous", "def", "dropout", "else", "forward", "hidden_states", "hidden_states_scaling_factor", "if", "is", "key", "key_activation", "key_layer", "key_layer_scaling_factor", "math", "matmul", "mixed_key_layer", "mixed_key_layer_scaling_factor", "mixed_query_layer", "mixed_query_layer_scaling_factor", "mixed_value_layer", "mixed_value_layer_scaling_factor", "new_context_layer_shape", "nn", "not", "num_attention_heads", "output_activation", "output_attentions", "output_scaling_factor", "outputs", "permute", "quant_mode", "query", "query_activation", "query_layer", "query_layer_scaling_factor", "return", "scale", "self", "seq_length", "shape", "size", "softmax", "sqrt", "torch", "transpose", "value", "value_activation", "value_layer", "value_layer_scaling_factor", "view"], "ibert/modeling_ibert.py:IBertSelfOutput.__init__": ["Dropout", "IntLayerNorm", "LayerNorm", "Model", "Module", "QuantAct", "QuantLinear", "True", "__init__", "act_bit", "bias", "bias_bit", "class", "config", "def", "dense", "dropout", "eps", "force_dequant", "hidden_dropout_prob", "hidden_size", "layer_norm_eps", "ln_input_act", "ln_input_bit", "ln_output_bit", "nn", "output_activation", "output_bit", "per_channel", "quant_mode", "self", "super", "weight_bit"], "ibert/modeling_ibert.py:IBertSelfOutput.forward": ["LayerNorm", "Model", "Module", "class", "def", "dense", "dropout", "forward", "hidden_states", "hidden_states_scaling_factor", "identity", "identity_scaling_factor", "input_tensor", "input_tensor_scaling_factor", "ln_input_act", "nn", "output_activation", "return", "self"], "ibert/modeling_ibert.py:IBertAttention.__init__": ["Model", "ModelSelfAttention", "ModelSelfOutput", "Module", "__init__", "class", "config", "def", "nn", "output", "quant_mode", "self", "super"], "ibert/modeling_ibert.py:IBertAttention.forward": ["False", "Model", "Module", "None", "attention_mask", "attention_output", "attention_output_scaling_factor", "class", "def", "forward", "hidden_states", "hidden_states_scaling_factor", "nn", "output", "output_attentions", "outputs", "outputs_scaling_factor", "return", "self", "self_outputs", "self_outputs_scaling_factor"], "ibert/modeling_ibert.py:IBertIntermediate.__init__": ["BERT", "I", "IntGELU", "Model", "Module", "QuantAct", "QuantLinear", "True", "ValueError", "__init__", "act_bit", "bias", "bias_bit", "class", "config", "def", "dense", "for", "force_dequant", "gelu", "hidden_act", "hidden_size", "if", "intermediate_act_fn", "intermediate_size", "nn", "only", "output_activation", "per_channel", "quant_mode", "raise", "self", "super", "supports", "weight_bit"], "ibert/modeling_ibert.py:IBertIntermediate.forward": ["Model", "Module", "class", "def", "dense", "forward", "hidden_states", "hidden_states_scaling_factor", "intermediate_act_fn", "nn", "output_activation", "return", "self"], "ibert/modeling_ibert.py:IBertOutput.__init__": ["Dropout", "IntLayerNorm", "LayerNorm", "Model", "Module", "QuantAct", "QuantLinear", "True", "__init__", "act_bit", "bias", "bias_bit", "class", "config", "def", "dense", "dropout", "eps", "force_dequant", "hidden_dropout_prob", "hidden_size", "intermediate_size", "layer_norm_eps", "ln_input_act", "ln_input_bit", "ln_output_bit", "nn", "output_activation", "output_bit", "per_channel", "quant_mode", "self", "super", "weight_bit"], "ibert/modeling_ibert.py:IBertOutput.forward": ["LayerNorm", "Model", "Module", "class", "def", "dense", "dropout", "forward", "hidden_states", "hidden_states_scaling_factor", "identity", "identity_scaling_factor", "input_tensor", "input_tensor_scaling_factor", "ln_input_act", "nn", "output_activation", "return", "self"], "ibert/modeling_ibert.py:IBertLayer.__init__": ["Model", "ModelAttention", "ModelIntermediate", "ModelOutput", "Module", "QuantAct", "__init__", "act_bit", "attention", "class", "config", "def", "intermediate", "nn", "output", "pre_intermediate_act", "pre_output_act", "quant_mode", "self", "seq_len_dim", "super"], "ibert/modeling_ibert.py:IBertLayer.forward": ["False", "Model", "Module", "None", "attention", "attention_mask", "attention_output", "attention_output_scaling_factor", "class", "def", "feed_forward_chunk", "forward", "hidden_states", "hidden_states_scaling_factor", "layer_output", "layer_output_scaling_factor", "nn", "output_attentions", "outputs", "return", "self", "self_attention_outputs", "self_attention_outputs_scaling_factor"], "ibert/modeling_ibert.py:IBertLayer.feed_forward_chunk": ["Model", "Module", "attention_output", "attention_output_scaling_factor", "class", "def", "feed_forward_chunk", "intermediate", "intermediate_output", "intermediate_output_scaling_factor", "layer_output", "layer_output_scaling_factor", "nn", "output", "pre_intermediate_act", "pre_output_act", "return", "self"], "ibert/modeling_ibert.py:IBertEncoder.__init__": ["Model", "ModelLayer", "Module", "ModuleList", "_", "__init__", "class", "config", "def", "for", "in", "layer", "nn", "num_hidden_layers", "quant_mode", "range", "self", "super"], "ibert/modeling_ibert.py:IBertEncoder.forward": ["BaseModelOutputWithPastAndCrossAttentions", "False", "Model", "Module", "None", "True", "all_cross_attentions", "all_hidden_states", "all_self_attentions", "attention_mask", "attentions", "class", "cross_attentions", "def", "else", "enumerate", "for", "forward", "hidden_states", "hidden_states_scaling_factor", "i", "if", "in", "is", "last_hidden_state", "layer", "layer_module", "layer_outputs", "nn", "not", "output_attentions", "output_hidden_states", "return", "return_dict", "self", "tuple", "v"], "ibert/modeling_ibert.py:IBertPooler.__init__": ["Linear", "Model", "Module", "Tanh", "__init__", "activation", "class", "config", "def", "dense", "hidden_size", "nn", "quant_mode", "self", "super"], "ibert/modeling_ibert.py:IBertPooler.forward": ["Model", "Module", "activation", "class", "def", "dense", "first_token_tensor", "forward", "hidden_states", "nn", "pooled_output", "return", "self"], "ibert/modeling_ibert.py:IBertPreTrainedModel._init_weights": ["Embedding", "False", "IntLayerNorm", "LayerNorm", "Linear", "Model", "ModelEmbeddings", "ModelLMHead", "None", "PreTrainedModel", "QuantAct", "QuantEmbedding", "QuantLinear", "_init_weights", "_is_hf_initialized", "act_scaling_factor", "and", "arange", "bias", "bias_integer", "class", "config", "constant_", "copy_", "def", "elif", "expand", "fc_scaling_factor", "getattr", "if", "init", "initializer_range", "is", "isinstance", "mean", "module", "nn", "normal_", "not", "ones_", "padding_idx", "position_ids", "self", "shape", "shift", "std", "torch", "weight", "weight_integer", "weight_scaling_factor", "x_max", "x_min", "zeros_"], "ibert/modeling_ibert.py:IBertPreTrainedModel.resize_token_embeddings": ["BERT", "I", "Model", "None", "NotImplementedError", "PreTrainedModel", "class", "def", "for", "is", "new_num_tokens", "not", "raise", "resize_token_embeddings", "self", "supported"], "ibert/modeling_ibert.py:IBertModel.__init__": ["Model", "ModelEmbeddings", "ModelEncoder", "ModelPooler", "ModelPreTrainedModel", "None", "True", "__init__", "add_pooling_layer", "class", "config", "def", "else", "embeddings", "encoder", "if", "pooler", "post_init", "quant_mode", "r", "self", "super"], "ibert/modeling_ibert.py:IBertModel.get_input_embeddings": ["Model", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "return", "self", "word_embeddings"], "ibert/modeling_ibert.py:IBertModel.set_input_embeddings": ["Model", "ModelPreTrainedModel", "class", "def", "embeddings", "self", "set_input_embeddings", "value", "word_embeddings"], "ibert/modeling_ibert.py:IBertModel.forward": ["BaseModelOutputWithPoolingAndCrossAttentions", "FloatTensor", "LongTensor", "Model", "ModelPreTrainedModel", "None", "Tensor", "ValueError", "You", "and", "at", "attention_mask", "attentions", "batch_size", "bool", "both", "cannot", "class", "config", "cross_attentions", "def", "device", "dtype", "either", "elif", "else", "embedding_output", "embedding_output_scaling_factor", "embeddings", "encoder", "encoder_outputs", "extended_attention_mask", "forward", "get_extended_attention_mask", "have", "hidden_states", "if", "input_ids", "input_shape", "inputs_embeds", "is", "kwargs", "last_hidden_state", "long", "not", "ones", "or", "output_attentions", "output_hidden_states", "pooled_output", "pooler", "pooler_output", "position_ids", "raise", "return", "return_dict", "same", "self", "seq_length", "sequence_output", "size", "specify", "the", "time", "to", "token_type_ids", "torch", "tuple", "use_return_dict", "warn_if_padding_and_no_attention_mask", "zeros"], "ibert/modeling_ibert.py:IBertForMaskedLM.__init__": ["False", "Model", "ModelLMHead", "ModelModel", "ModelPreTrainedModel", "__init__", "add_pooling_layer", "class", "config", "def", "lm_head", "post_init", "self", "super"], "ibert/modeling_ibert.py:IBertForMaskedLM.get_output_embeddings": ["Model", "ModelPreTrainedModel", "class", "decoder", "def", "get_output_embeddings", "lm_head", "return", "self"], "ibert/modeling_ibert.py:IBertForMaskedLM.set_output_embeddings": ["Model", "ModelPreTrainedModel", "bias", "class", "decoder", "def", "lm_head", "new_embeddings", "self", "set_output_embeddings"], "ibert/modeling_ibert.py:IBertForMaskedLM.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "MaskedLMOutput", "Model", "ModelPreTrainedModel", "None", "attention_mask", "attentions", "bool", "class", "config", "def", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "lm_head", "logits", "loss", "loss_fct", "masked_lm_loss", "not", "output", "output_attentions", "output_hidden_states", "outputs", "position_ids", "prediction_scores", "r", "return", "return_dict", "self", "sequence_output", "token_type_ids", "torch", "tuple", "use_return_dict", "view", "vocab_size"], "ibert/modeling_ibert.py:IBertLMHead.__init__": ["LayerNorm", "Linear", "Model", "Module", "Parameter", "__init__", "bias", "class", "config", "decoder", "def", "dense", "eps", "hidden_size", "layer_norm", "layer_norm_eps", "nn", "self", "super", "torch", "vocab_size", "zeros"], "ibert/modeling_ibert.py:IBertLMHead.forward": ["Model", "Module", "class", "decoder", "def", "dense", "features", "forward", "gelu", "kwargs", "layer_norm", "nn", "return", "self", "x"], "ibert/modeling_ibert.py:IBertForSequenceClassification.__init__": ["False", "Model", "ModelClassificationHead", "ModelModel", "ModelPreTrainedModel", "__init__", "add_pooling_layer", "class", "classifier", "config", "def", "num_labels", "post_init", "self", "super"], "ibert/modeling_ibert.py:IBertForSequenceClassification.forward": ["BCEWithLogitsLoss", "CrossEntropyLoss", "FloatTensor", "LongTensor", "MSELoss", "Model", "ModelPreTrainedModel", "None", "SequenceClassifierOutput", "and", "attention_mask", "attentions", "bool", "class", "classifier", "config", "def", "dtype", "elif", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "kwargs", "labels", "logits", "long", "loss", "loss_fct", "multi_label_classification", "not", "num_labels", "or", "output", "output_attentions", "output_hidden_states", "outputs", "position_ids", "problem_type", "r", "regression", "return", "return_dict", "self", "sequence_output", "single_label_classification", "squeeze", "token_type_ids", "torch", "tuple", "use_return_dict", "view"], "ibert/modeling_ibert.py:IBertForMultipleChoice.__init__": ["Dropout", "Linear", "Model", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "classifier", "config", "def", "dropout", "hidden_dropout_prob", "hidden_size", "nn", "post_init", "self", "super"], "ibert/modeling_ibert.py:IBertForMultipleChoice.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "Model", "ModelPreTrainedModel", "MultipleChoiceModelOutput", "None", "attention_mask", "attentions", "bool", "class", "classifier", "config", "def", "dropout", "else", "flat_attention_mask", "flat_input_ids", "flat_inputs_embeds", "flat_position_ids", "flat_token_type_ids", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "not", "num_choices", "output", "output_attentions", "output_hidden_states", "outputs", "pooled_output", "position_ids", "r", "reshaped_logits", "return", "return_dict", "self", "shape", "size", "token_type_ids", "torch", "tuple", "use_return_dict", "view"], "ibert/modeling_ibert.py:IBertForTokenClassification.__init__": ["Dropout", "False", "Linear", "Model", "ModelModel", "ModelPreTrainedModel", "__init__", "add_pooling_layer", "class", "classifier", "config", "def", "dropout", "hidden_dropout_prob", "hidden_size", "nn", "num_labels", "post_init", "self", "super"], "ibert/modeling_ibert.py:IBertForTokenClassification.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "Model", "ModelPreTrainedModel", "None", "TokenClassifierOutput", "attention_mask", "attentions", "bool", "class", "classifier", "config", "def", "dropout", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "not", "num_labels", "output", "output_attentions", "output_hidden_states", "outputs", "position_ids", "r", "return", "return_dict", "self", "sequence_output", "token_type_ids", "torch", "tuple", "use_return_dict", "view"], "ibert/modeling_ibert.py:IBertClassificationHead.__init__": ["Dropout", "Linear", "Model", "Module", "__init__", "class", "config", "def", "dense", "dropout", "hidden_dropout_prob", "hidden_size", "nn", "num_labels", "out_proj", "self", "super"], "ibert/modeling_ibert.py:IBertClassificationHead.forward": ["Model", "Module", "class", "def", "dense", "dropout", "features", "forward", "hidden_states", "kwargs", "nn", "out_proj", "return", "self", "tanh", "torch"], "ibert/modeling_ibert.py:IBertForQuestionAnswering.__init__": ["False", "Linear", "Model", "ModelModel", "ModelPreTrainedModel", "__init__", "add_pooling_layer", "class", "config", "def", "hidden_size", "nn", "num_labels", "post_init", "qa_outputs", "self", "super"], "ibert/modeling_ibert.py:IBertForQuestionAnswering.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "Model", "ModelPreTrainedModel", "None", "QuestionAnsweringModelOutput", "and", "attention_mask", "attentions", "bool", "clamp", "class", "config", "contiguous", "def", "dim", "else", "end_logits", "end_loss", "end_positions", "forward", "hidden_states", "if", "ignore_index", "ignored_index", "input_ids", "inputs_embeds", "is", "kwargs", "len", "logits", "loss", "loss_fct", "not", "output", "output_attentions", "output_hidden_states", "outputs", "position_ids", "qa_outputs", "return", "return_dict", "self", "sequence_output", "size", "split", "squeeze", "start_logits", "start_loss", "start_positions", "token_type_ids", "torch", "total_loss", "tuple", "use_return_dict"], "ibert/modeling_ibert.py:create_position_ids_from_input_ids": ["Model_position_ids_from_input_ids", "cumsum", "def", "dim", "incremental_indices", "input_ids", "int", "long", "mask", "ne", "padding_idx", "past_key_values_length", "return", "torch", "type_as"], "idefics/modeling_idefics.py:expand_inputs_for_generation": ["False", "Model_inputs_for_generation", "Model_size", "Modeled_return_idx", "None", "arange", "attention_mask", "def", "device", "elif", "encoder_outputs", "get", "if", "image_attention_mask", "image_encoder_embeddings", "in", "index_select", "input_ids", "is", "is_encoder_decoder", "model_kwargs", "not", "perceiver_embeddings", "pixel_values", "repeat", "return", "shape", "to", "token_type_ids", "torch", "view"], "idefics/modeling_idefics.py:freeze_model": ["Embedding", "False", "LayerNorm", "Linear", "Model_model", "True", "and", "any", "def", "else", "for", "if", "in", "isinstance", "m", "mapping", "model", "module", "module_exceptions", "module_exceptions_mapped", "modules", "nn", "requires_grad_", "return", "t"], "idefics/modeling_idefics.py:IdeficsDecoupledEmbedding.__init__": ["Embedding", "False", "Got", "ModelDecoupledEmbedding", "None", "ValueError", "__init__", "additional_embedding", "and", "be", "bool", "class", "def", "device", "dtype", "embedding_dim", "f", "if", "is", "kwargs", "must", "nn", "not", "num_additional_embeddings", "num_embeddings", "padding_idx", "partially_freeze", "raise", "requires_grad_", "self", "super", "weight", "within"], "idefics/modeling_idefics.py:IdeficsDecoupledEmbedding.forward": ["Embedding", "F", "ModelDecoupledEmbedding", "additional_embedding", "additional_embeddings", "additional_vocab_indices", "class", "clone", "def", "embedding", "forward", "full_vector", "if", "input_ids", "input_ids_additional_vocab", "nn", "num_additional_embeddings", "num_embeddings", "return", "self", "torch", "weight", "where"], "idefics/modeling_idefics.py:IdeficsDecoupledEmbedding.extra_repr": ["Embedding", "ModelDecoupledEmbedding", "class", "def", "embedding_dim", "extra_repr", "f", "nn", "num_additional_embeddings", "num_embeddings", "partially_freeze", "return", "self", "str"], "idefics/modeling_idefics.py:IdeficsDecoupledLinear.__init__": ["False", "Linear", "ModelDecoupledLinear", "None", "True", "__init__", "additional_fc", "bias", "bool", "class", "def", "device", "dtype", "if", "in_features", "int", "nn", "out_additional_features", "out_features", "partially_freeze", "requires_grad_", "self", "super", "weight"], "idefics/modeling_idefics.py:IdeficsDecoupledLinear.forward": ["F", "Linear", "ModelDecoupledLinear", "Tensor", "additional_fc", "additional_features", "bias", "cat", "class", "def", "forward", "if", "input", "linear", "nn", "out_additional_features", "output", "return", "self", "torch", "weight"], "idefics/modeling_idefics.py:IdeficsDecoupledLinear.extra_repr": ["Linear", "ModelDecoupledLinear", "None", "bias", "class", "def", "extra_repr", "f", "in_features", "is", "nn", "not", "out_additional_features", "out_features", "partially_freeze", "return", "self", "str"], "idefics/modeling_idefics.py:IdeficsRMSNorm.__init__": ["ModelRMSNorm", "Module", "Parameter", "__init__", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "idefics/modeling_idefics.py:IdeficsRMSNorm.forward": ["ModelRMSNorm", "Module", "True", "bfloat16", "class", "def", "dtype", "float16", "float32", "forward", "hidden_states", "if", "in", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "idefics/modeling_idefics.py:IdeficsRMSNorm.extra_repr": ["ModelRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "variance_epsilon", "weight"], "idefics/modeling_idefics.py:IdeficsEmbedding.__init__": ["False", "ModelEmbedding", "Module", "None", "__init__", "_set_cos_sin_cache", "arange", "base", "class", "def", "device", "dim", "dtype", "float", "get_default_dtype", "int64", "inv_freq", "max_position_embeddings", "nn", "persistent", "register_buffer", "self", "seq_len", "super", "to", "torch"], "idefics/modeling_idefics.py:IdeficsEmbedding._set_cos_sin_cache": ["False", "ModelEmbedding", "Module", "_set_cos_sin_cache", "arange", "cat", "class", "cos", "cos_cached", "def", "device", "dim", "dtype", "einsum", "emb", "freqs", "i", "ij", "int64", "inv_freq", "j", "max_seq_len_cached", "nn", "persistent", "register_buffer", "self", "seq_len", "sin", "sin_cached", "t", "to", "torch", "type_as"], "idefics/modeling_idefics.py:IdeficsEmbedding.forward": ["ModelEmbedding", "Module", "None", "_set_cos_sin_cache", "class", "cos_cached", "def", "device", "dtype", "forward", "if", "max_seq_len_cached", "nn", "return", "self", "seq_len", "sin_cached", "to", "torch", "x"], "idefics/modeling_idefics.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "idefics/modeling_idefics.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cos", "def", "k", "k_embed", "position_ids", "q", "q_embed", "return", "rotate_half", "sin", "unsqueeze", "unsqueeze_dim"], "idefics/modeling_idefics.py:IdeficsMLP.__init__": ["ACT2FN", "False", "Linear", "ModelMLP", "Module", "__init__", "act_fn", "bias", "class", "def", "down_proj", "gate_proj", "hidden_act", "hidden_size", "int", "intermediate_size", "nn", "self", "str", "super", "up_proj"], "idefics/modeling_idefics.py:IdeficsMLP.forward": ["ModelMLP", "Module", "act_fn", "class", "def", "down_proj", "forward", "gate_proj", "nn", "return", "self", "up_proj", "x"], "idefics/modeling_idefics.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "attention_mask", "attn_output", "attn_weights", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "softmax", "to", "torch", "training", "transpose", "value"], "idefics/modeling_idefics.py:IdeficsAttention.__init__": ["False", "Instantiating", "Linear", "ModelAttention", "ModelEmbedding", "ModelRMSNorm", "Module", "None", "Please", "PreTrainedConfig", "True", "ValueError", "__class__", "__init__", "__name__", "a", "and", "be", "bias", "bool", "by", "caching", "call", "class", "config", "creating", "def", "divisible", "dropout", "during", "else", "embed_dim", "eps", "errors", "f", "float", "forward", "functional", "got", "hasattr", "head_dim", "hidden_size", "higher", "if", "int", "is", "is_causal", "is_cross_attention", "k_layer_norm", "k_proj", "kv_input_dim", "layer_idx", "lead", "logger", "make", "model", "must", "nn", "not", "num_heads", "o_proj", "or", "passing", "provide", "pytorch", "q_layer_norm", "q_proj", "qk_layer_norms", "raise", "recommended", "requires", "rms_norm_eps", "rotary_emb", "scaled_dot_product_attention", "scaling", "self", "super", "sure", "the", "this", "to", "used", "v_proj", "vision_config", "warning_once", "when", "will", "without"], "idefics/modeling_idefics.py:IdeficsAttention._shape": ["ModelAttention", "Module", "Tensor", "_shape", "bsz", "class", "contiguous", "def", "head_dim", "int", "nn", "num_heads", "return", "self", "seq_len", "tensor", "torch", "transpose", "view"], "idefics/modeling_idefics.py:IdeficsAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "LongTensor", "ModelAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "_attn_implementation", "apply_rotary_pos_emb", "attention_interface", "attention_mask", "attn_output", "attn_weights", "bsz", "cache_kwargs", "cache_position", "class", "config", "contiguous", "cos", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "head_dim", "hidden_states", "if", "is", "is_cross_attention", "k_layer_norm", "k_proj", "key_states", "key_value_states", "kv_len", "kv_seq_len", "kwargs", "layer_idx", "max", "nn", "not", "num_heads", "o_proj", "or", "past_key_values", "position_ids", "q_layer_norm", "q_len", "q_proj", "qk_layer_norms", "query_states", "reshape", "return", "rotary_emb", "scaling", "self", "seq_len", "shape", "sin", "size", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "view"], "idefics/modeling_idefics.py:IdeficsDecoderLayer.__init__": ["GradientCheckpointingLayer", "ModelAttention", "ModelConfig", "ModelDecoderLayer", "ModelMLP", "ModelRMSNorm", "None", "__init__", "class", "config", "def", "dropout", "eps", "hidden_act", "hidden_size", "input_layernorm", "int", "intermediate_size", "layer_idx", "mlp", "num_attention_heads", "num_heads", "post_attention_layernorm", "rms_norm_eps", "self", "self_attn", "super"], "idefics/modeling_idefics.py:IdeficsDecoderLayer.forward": ["Cache", "FloatTensor", "GradientCheckpointingLayer", "LongTensor", "ModelDecoderLayer", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "attention_mask", "cache_position", "class", "def", "dropout", "forward", "functional", "hidden_states", "input_layernorm", "kwargs", "mlp", "nn", "p", "past_key_values", "position_ids", "post_attention_layernorm", "residual", "return", "self", "self_attn", "torch", "training"], "idefics/modeling_idefics.py:IdeficsGatedCrossAttentionLayer.__init__": ["Alpha", "GradientCheckpointingLayer", "ModelAttention", "ModelConfig", "ModelGatedCrossAttentionLayer", "ModelMLP", "ModelRMSNorm", "None", "NotImplementedError", "Parameter", "Tanh", "True", "Unknown", "ValueError", "__init__", "act_cross_attn", "act_dense", "alpha_cross_attn", "alpha_dense", "alpha_initializer", "alpha_type", "alphas_initializer_range", "and", "class", "config", "correctly", "cross_attn", "def", "dropout", "elif", "else", "eps", "f", "float", "for", "gaussian", "hasattr", "hidden_act", "hidden_size", "if", "implemented", "in", "initialization", "initialized", "input_layernorm", "int", "intermediate_size", "is_cross_attention", "layer_idx", "mean", "mlp", "nn", "normal", "not", "num_attention_heads", "num_heads", "ones", "parameters", "post_attention_layernorm", "qk_layer_norms", "raise", "random", "rms_norm_eps", "scheme", "self", "size", "std", "super", "torch", "value", "vector", "yet", "zeros"], "idefics/modeling_idefics.py:IdeficsGatedCrossAttentionLayer.forward": ["Cache", "FloatTensor", "GradientCheckpointingLayer", "Model", "ModelGatedCrossAttentionLayer", "None", "NotImplementedError", "Past", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "_", "act_cross_attn", "act_dense", "alpha_cross_attn", "alpha_dense", "are", "attending", "attention", "attention_mask", "be", "class", "conditioned", "config", "cross", "cross_attention_gate", "cross_attn", "def", "dropout", "features", "for", "forward", "functional", "hidden_states", "if", "image_attention_mask", "image_hidden_states", "images", "implemented", "input_layernorm", "is", "key", "key_value_states", "kwargs", "masked_fill", "mlp", "module", "nn", "no", "not", "on", "out", "p", "past_key_values", "post_attention_layernorm", "r", "raise", "required", "residual", "return", "self", "states", "the", "to", "torch", "training", "value", "visual", "which", "zero"], "idefics/modeling_idefics.py:IdeficsPreTrainedModel._init_weights": ["ModelEmbedding", "ModelGatedCrossAttentionLayer", "ModelPerceiverResampler", "ModelPreTrainedModel", "ModelVisionEmbeddings", "PreTrainedModel", "_init_weights", "alpha_cross_attn", "alpha_dense", "alpha_initializer", "alphas_initializer_range", "arange", "base", "cat", "class", "class_embedding", "config", "copy_", "cos", "cos_cached", "def", "dim", "einsum", "elif", "emb", "expand", "freqs", "gaussian", "i", "if", "ij", "in", "init", "inv_freq", "isinstance", "j", "latents", "max_position_embeddings", "mean", "module", "normal", "normal_", "ones", "ones_", "position_ids", "random", "self", "shape", "sin", "sin_cached", "std", "super", "t", "torch", "type_as", "zeros", "zeros_"], "idefics/modeling_idefics.py:IdeficsModel.__init__": ["False", "ModelConfig", "ModelDecoderLayer", "ModelDecoupledEmbedding", "ModelGatedCrossAttentionLayer", "ModelModel", "ModelPerceiverResampler", "ModelPreTrainedModel", "ModelRMSNorm", "ModelVisionTransformer", "ModuleList", "__init__", "_attn_implementation", "additional_vocab_size", "class", "config", "cross_layer_interval", "def", "embed_dim", "embed_tokens", "embedding_dim", "eps", "for", "freeze_relevant_params", "freeze_text_layers", "gated_cross_attn_layers", "gradient_checkpointing", "hidden_size", "i", "if", "image_size", "in", "layer_idx", "layers", "nn", "norm", "num_additional_embeddings", "num_cross_layers", "num_embeddings", "num_hidden_layers", "pad_token_id", "padding_idx", "partially_freeze", "perceiver_config", "perceiver_resampler", "post_init", "range", "resampler_depth", "resampler_head_dim", "resampler_n_heads", "resampler_n_latents", "rms_norm_eps", "self", "super", "use_resampler", "vision_config", "vision_model", "vocab_size"], "idefics/modeling_idefics.py:IdeficsModel.freeze_relevant_params": ["ModelModel", "ModelPreTrainedModel", "None", "class", "config", "def", "freeze_model", "freeze_relevant_params", "freeze_text_layers", "freeze_text_module_exceptions", "freeze_vision_layers", "freeze_vision_module_exceptions", "if", "is", "module_exceptions", "self", "vision_model"], "idefics/modeling_idefics.py:IdeficsModel.freeze_text_layers": ["ModelModel", "ModelPreTrainedModel", "class", "def", "for", "freeze_model", "freeze_text_layers", "in", "layers", "module", "module_exceptions", "norm", "self"], "idefics/modeling_idefics.py:IdeficsModel.freeze_vision_layers": ["ModelModel", "ModelPreTrainedModel", "class", "def", "freeze_model", "freeze_vision_layers", "module_exceptions", "self", "vision_model"], "idefics/modeling_idefics.py:IdeficsModel.forward": ["Cache", "DynamicCache", "Exactly", "False", "FloatTensor", "If", "LongTensor", "ModelBaseModelOutputWithPast", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "ValueError", "You", "_", "and", "any", "arange", "are", "attention_mask", "batch_size", "be", "bool", "cache_position", "causal_mask", "class", "config", "contiguous", "create_causal_mask", "cross_attention_gate", "cross_attn_block", "cross_layer_interval", "cumsum", "decoder_layer", "def", "device", "dim", "dtype", "elif", "else", "embed_tokens", "enumerate", "exactly", "for", "forward", "gated_cross_attn_layers", "get_seq_length", "has", "hidden_states", "idx", "if", "image_attention_mask", "image_batch_size", "image_encoder_embeddings", "image_hidden_shape", "image_hidden_size", "image_hidden_states", "image_seq_len", "image_sequence_length", "in", "input_embeds", "input_ids", "inputs_embeds", "interpolate_pos_encoding", "invert_attention_mask", "is", "kwargs", "last_hidden_state", "layers", "long", "masked_fill_", "must", "norm", "not", "num_images", "of", "one", "ones", "or", "passed", "past_key_values", "past_key_values_length", "perceiver_embeddings", "perceiver_resampler", "pixel_values", "position_ids", "r", "raise", "repeat", "return", "self", "seq_length", "seq_length_with_past", "shape", "should", "size", "specify", "squeeze", "sum", "text_seq_len", "to", "torch", "tuple", "unsqueeze", "use_cache", "use_resampler", "view", "vision_model", "x"], "idefics/modeling_idefics.py:IdeficsForVisionText2Text.__init__": ["False", "GenerationMixin", "ModelDecoupledLinear", "ModelForVisionText2Text", "ModelModel", "ModelPreTrainedModel", "None", "__init__", "_tied_weights_keys", "additional_embedding", "additional_fc", "additional_vocab_size", "bias", "class", "config", "def", "embed_tokens", "freeze_lm_head", "hidden_size", "if", "in_features", "lm_head", "model", "out_additional_features", "out_features", "partially_freeze", "post_init", "self", "super", "vision_model", "vocab_size", "weight"], "idefics/modeling_idefics.py:IdeficsForVisionText2Text.forward": ["Cache", "False", "FloatTensor", "GenerationMixin", "LongTensor", "ModelBaseModelOutputWithPast", "ModelCausalLMOutputWithPast", "ModelForVisionText2Text", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "else", "forward", "hidden_states", "if", "image_attention_mask", "image_encoder_embeddings", "image_hidden_states", "input_ids", "inputs_embeds", "int", "interpolate_pos_encoding", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "outputs", "past_key_values", "perceiver_embeddings", "pixel_values", "position_ids", "r", "return", "return_dict", "self", "slice", "slice_indices", "torch", "tuple", "use_cache", "vocab_size"], "idefics/modeling_idefics.py:IdeficsForVisionText2Text.prepare_inputs_for_generation": ["False", "GenerationMixin", "ModelForVisionText2Text", "ModelPreTrainedModel", "None", "and", "attention_mask", "cache_position", "class", "config", "def", "else", "if", "image_attention_mask", "image_encoder_embeddings", "image_hidden_states", "images_kwargs", "input_ids", "inputs_embeds", "interpolate_pos_encoding", "is", "kwargs", "model_inputs", "not", "past_key_values", "perceiver_embeddings", "pixel_values", "pop", "position_ids", "prepare_inputs_for_generation", "return", "self", "seq_length", "shape", "super", "use_cache", "use_resampler"], "idefics/modeling_idefics.py:IdeficsForVisionText2Text._update_model_kwargs_for_generation": ["Any", "False", "GenerationMixin", "ModelForVisionText2Text", "ModelOutput", "ModelPreTrainedModel", "True", "_update_model_kwargs_for_generation", "bool", "cat", "class", "def", "dict", "dim", "else", "get", "if", "image_attention_mask", "image_hidden_states", "in", "is_encoder_decoder", "kwargs", "last_mask", "model_kwargs", "outputs", "return", "self", "str", "super", "torch", "unsqueeze", "use_cache"], "idefics2/modeling_idefics2.py:Idefics2VisionEmbeddings.__init__": ["Conv2d", "Embedding", "ModelVisionConfig", "ModelVisionEmbeddings", "Module", "__init__", "class", "config", "def", "embed_dim", "hidden_size", "image_size", "in_channels", "kernel_size", "nn", "num_channels", "num_patches", "num_patches_per_side", "num_positions", "out_channels", "padding", "patch_embedding", "patch_size", "position_embedding", "self", "stride", "super", "valid"], "idefics2/modeling_idefics2.py:Idefics2VisionEmbeddings.forward": ["BoolTensor", "FloatTensor", "ModelVisionEmbeddings", "Module", "None", "Tensor", "True", "_", "arange", "batch_idx", "batch_size", "boundaries", "bucket_coords_h", "bucket_coords_w", "bucketize", "clamp", "class", "def", "device", "dtype", "embeddings", "enumerate", "fill_value", "flatten", "float32", "for", "forward", "fractional_coords_h", "fractional_coords_w", "full", "h_indices", "in", "max", "max_im_h", "max_im_w", "max_nb_patches_h", "max_nb_patches_w", "nb_patches_h", "nb_patches_w", "nn", "num_patches_per_side", "p_attn_mask", "patch_attention_mask", "patch_embedding", "patch_embeds", "patch_size", "pixel_values", "pos_ids", "position_embedding", "position_ids", "return", "right", "self", "shape", "size", "step_h", "step_w", "sum", "to", "torch", "transpose", "view", "w_indices"], "idefics2/modeling_idefics2.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "hasattr", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value"], "idefics2/modeling_idefics2.py:Idefics2VisionAttention.__init__": ["False", "Linear", "ModelVisionAttention", "Module", "ValueError", "__init__", "and", "attention_dropout", "be", "by", "class", "config", "def", "divisible", "dropout", "embed_dim", "f", "got", "head_dim", "hidden_size", "if", "is_causal", "k_proj", "must", "nn", "num_attention_heads", "num_heads", "out_proj", "q_proj", "raise", "scale", "self", "super", "v_proj"], "idefics2/modeling_idefics2.py:Idefics2VisionAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "ModelVisionAttention", "Module", "None", "Tensor", "_attn_implementation", "attention_interface", "attention_mask", "attn_output", "attn_weights", "batch_size", "class", "config", "contiguous", "def", "dropout", "eager", "eager_attention_forward", "else", "embed_dim", "forward", "head_dim", "hidden_states", "if", "is_causal", "k_proj", "keys", "kwargs", "nn", "not", "num_heads", "out_proj", "q_proj", "queries", "reshape", "return", "scale", "scaling", "self", "seq_length", "shape", "torch", "training", "transpose", "tuple", "v_proj", "values", "view"], "idefics2/modeling_idefics2.py:Idefics2VisionMLP.__init__": ["ACT2FN", "Linear", "ModelVisionMLP", "Module", "__init__", "activation_fn", "class", "config", "def", "fc1", "fc2", "hidden_act", "hidden_size", "intermediate_size", "nn", "self", "super"], "idefics2/modeling_idefics2.py:Idefics2VisionMLP.forward": ["ModelVisionMLP", "Module", "Tensor", "activation_fn", "class", "def", "fc1", "fc2", "forward", "hidden_states", "nn", "return", "self", "torch"], "idefics2/modeling_idefics2.py:Idefics2MLP.__init__": ["ACT2FN", "False", "Linear", "ModelMLP", "Module", "__init__", "act_fn", "bias", "class", "def", "down_proj", "gate_proj", "hidden_act", "hidden_size", "int", "intermediate_size", "nn", "output_size", "self", "str", "super", "up_proj"], "idefics2/modeling_idefics2.py:Idefics2MLP.forward": ["ModelMLP", "Module", "act_fn", "class", "def", "down_proj", "forward", "gate_proj", "nn", "return", "self", "up_proj", "x"], "idefics2/modeling_idefics2.py:Idefics2MultiheadAttentionPoolingHead.__init__": ["LayerNorm", "ModelMLP", "ModelMultiheadAttentionPoolingHead", "ModelVisionConfig", "Module", "MultiheadAttention", "Parameter", "True", "__init__", "attention", "batch_first", "class", "config", "def", "eps", "hidden_act", "hidden_size", "intermediate_size", "layer_norm_eps", "layernorm", "mlp", "nn", "num_attention_heads", "output_size", "probe", "randn", "self", "super", "torch"], "idefics2/modeling_idefics2.py:Idefics2MultiheadAttentionPoolingHead.forward": ["ModelMultiheadAttentionPoolingHead", "Module", "attention", "batch_size", "class", "def", "forward", "hidden_state", "layernorm", "mlp", "nn", "probe", "repeat", "residual", "return", "self", "shape"], "idefics2/modeling_idefics2.py:Idefics2EncoderLayer.__init__": ["GradientCheckpointingLayer", "LayerNorm", "ModelEncoderLayer", "ModelVisionAttention", "ModelVisionConfig", "ModelVisionMLP", "__init__", "class", "config", "def", "embed_dim", "eps", "hidden_size", "layer_norm1", "layer_norm2", "layer_norm_eps", "mlp", "nn", "self", "self_attn", "super"], "idefics2/modeling_idefics2.py:Idefics2EncoderLayer.forward": ["FloatTensor", "GradientCheckpointingLayer", "ModelEncoderLayer", "Tensor", "TransformersKwargs", "Unpack", "_", "attention_mask", "class", "def", "forward", "hidden_states", "kwargs", "layer_norm1", "layer_norm2", "mlp", "residual", "return", "self", "self_attn", "torch"], "idefics2/modeling_idefics2.py:Idefics2Encoder.__init__": ["False", "ModelConfig", "ModelEncoder", "ModelEncoderLayer", "Module", "ModuleList", "_", "__init__", "class", "config", "def", "for", "gradient_checkpointing", "in", "layers", "nn", "num_hidden_layers", "range", "self", "super"], "idefics2/modeling_idefics2.py:Idefics2Encoder.forward": ["BaseModelOutput", "ModelEncoder", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "class", "def", "encoder_layer", "for", "forward", "hidden_states", "in", "inputs_embeds", "kwargs", "last_hidden_state", "layers", "nn", "return", "self", "torch"], "idefics2/modeling_idefics2.py:Idefics2PreTrainedModel._init_weights": ["ModelMultiheadAttentionPoolingHead", "ModelPerceiverResampler", "ModelPreTrainedModel", "PreTrainedModel", "_init_weights", "class", "def", "elif", "if", "init", "isinstance", "latents", "module", "normal_", "ones_", "probe", "self", "super"], "idefics2/modeling_idefics2.py:Idefics2VisionTransformer.__init__": ["LayerNorm", "ModelEncoder", "ModelPreTrainedModel", "ModelVisionConfig", "ModelVisionEmbeddings", "ModelVisionTransformer", "__init__", "class", "config", "def", "embed_dim", "embeddings", "encoder", "eps", "hidden_size", "layer_norm_eps", "nn", "post_init", "post_layernorm", "self", "super"], "idefics2/modeling_idefics2.py:Idefics2VisionTransformer.get_input_embeddings": ["ModelPreTrainedModel", "ModelVisionTransformer", "class", "def", "embeddings", "get_input_embeddings", "return", "self"], "idefics2/modeling_idefics2.py:Idefics2VisionTransformer.set_input_embeddings": ["ModelPreTrainedModel", "ModelVisionTransformer", "class", "def", "embeddings", "self", "set_input_embeddings", "value"], "idefics2/modeling_idefics2.py:Idefics2VisionTransformer.forward": ["BaseModelOutput", "BoolTensor", "ModelPreTrainedModel", "ModelVisionTransformer", "None", "TransformersKwargs", "Unpack", "_attn_implementation", "_prepare_4d_attention_mask", "any", "attention_mask", "batch_size", "bool", "class", "config", "def", "device", "dtype", "elif", "embeddings", "encoder", "encoder_outputs", "flash_attention_2", "forward", "hidden_states", "if", "inputs_embeds", "is", "kwargs", "last_hidden_state", "not", "ones", "patch_attention_mask", "patch_size", "pixel_values", "post_layernorm", "r", "return", "self", "size", "to", "torch", "tuple", "view"], "idefics2/modeling_idefics2.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "idefics2/modeling_idefics2.py:Idefics2RMSNorm.__init__": ["ModelRMSNorm", "Module", "Parameter", "__init__", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "idefics2/modeling_idefics2.py:Idefics2RMSNorm.forward": ["ModelRMSNorm", "Module", "True", "class", "def", "dtype", "float32", "forward", "hidden_states", "input_dtype", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "idefics2/modeling_idefics2.py:Idefics2RMSNorm.extra_repr": ["ModelRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "variance_epsilon", "weight"], "idefics2/modeling_idefics2.py:Idefics2PerceiverAttention.__init__": ["False", "Linear", "ModelPerceiverAttention", "Module", "None", "__init__", "attention_dropout", "bias", "class", "config", "def", "head_dim", "hidden_size", "int", "is_causal", "k_proj", "layer_idx", "nn", "num_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "q_proj", "resampler_head_dim", "resampler_n_heads", "scaling", "self", "super", "v_proj"], "idefics2/modeling_idefics2.py:Idefics2PerceiverAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "LongTensor", "ModelPerceiverAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "_attn_implementation", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "bsz", "class", "concat", "config", "context", "def", "dim", "dropout", "eager", "eager_attention_forward", "else", "forward", "getattr", "head_dim", "hidden_states", "if", "is", "is_causal", "k_proj", "keys", "kv_seq_len", "kwargs", "latents", "layer_idx", "nn", "not", "num_heads", "num_key_value_heads", "o_proj", "past_key_values", "position_ids", "q_len", "q_proj", "queries", "reshape", "return", "scaling", "self", "size", "torch", "training", "transpose", "tuple", "update", "v_proj", "values", "view"], "idefics2/modeling_idefics2.py:Idefics2PerceiverLayer.__init__": ["ModelMLP", "ModelPerceiverAttention", "ModelPerceiverLayer", "ModelRMSNorm", "Module", "__init__", "class", "config", "def", "depth", "eps", "hidden_act", "hidden_size", "input_context_norm", "input_latents_norm", "int", "intermediate_size", "layer_idx", "mlp", "n_latents", "nn", "output_size", "post_attention_layernorm", "resampler_depth", "resampler_n_latents", "rms_norm_eps", "self", "self_attn", "super"], "idefics2/modeling_idefics2.py:Idefics2PerceiverLayer.forward": ["Cache", "FloatTensor", "LongTensor", "ModelPerceiverLayer", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "attention_mask", "class", "context", "def", "forward", "input_context_norm", "input_latents_norm", "kwargs", "latents", "mlp", "nn", "past_key_values", "position_ids", "post_attention_layernorm", "residual", "return", "self", "self_attn", "torch"], "idefics2/modeling_idefics2.py:Idefics2PerceiverResampler.__init__": ["ModelPerceiverLayer", "ModelPerceiverResampler", "ModelPreTrainedModel", "ModelRMSNorm", "ModuleList", "None", "Parameter", "__init__", "class", "config", "def", "depth", "eps", "for", "hidden_act", "hidden_size", "idx", "in", "latents", "layers", "n_latents", "nn", "norm", "ones", "post_init", "range", "resampler_depth", "resampler_n_latents", "rms_norm_eps", "self", "super", "torch"], "idefics2/modeling_idefics2.py:Idefics2PerceiverResampler.forward": ["ModelPerceiverResampler", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "_attn_implementation", "_prepare_4d_attention_mask", "attention_mask", "cat", "class", "compressed_context", "config", "context", "def", "device", "dim", "dtype", "else", "expand", "flash_attention_2", "for", "forward", "if", "in", "kwargs", "latent_attention_mask", "latents", "layers", "n_latents", "norm", "ones", "perceiver_layer", "position_ids", "r", "return", "self", "shape", "size", "tgt_len", "torch", "unsqueeze"], "idefics2/modeling_idefics2.py:Idefics2Connector.__init__": ["ModelConnector", "ModelMLP", "ModelPerceiverResampler", "Module", "__init__", "_from_config", "class", "config", "def", "hidden_act", "hidden_size", "intermediate_size", "modality_projection", "nn", "output_size", "perceiver_config", "perceiver_resampler", "self", "super", "text_config", "vision_config"], "idefics2/modeling_idefics2.py:Idefics2Connector.forward": ["ModelConnector", "Module", "attention_mask", "class", "context", "def", "forward", "image_hidden_states", "modality_projection", "nn", "perceiver_resampler", "return", "self"], "idefics2/modeling_idefics2.py:Idefics2Model.__init__": ["AutoModel", "ModelConfig", "ModelConnector", "ModelModel", "ModelPreTrainedModel", "ModelVisionTransformer", "__init__", "_from_config", "class", "config", "connector", "def", "from_config", "image_seq_len", "image_token_id", "pad_token_id", "padding_idx", "perceiver_config", "post_init", "resampler_n_latents", "self", "super", "text_config", "text_model", "vision_config", "vision_model", "vocab_size"], "idefics2/modeling_idefics2.py:Idefics2Model.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "return", "self", "text_model"], "idefics2/modeling_idefics2.py:Idefics2Model.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "self", "set_input_embeddings", "text_model", "value"], "idefics2/modeling_idefics2.py:Idefics2Model.inputs_merger": ["LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "all", "class", "config", "def", "device", "dtype", "else", "expand_as", "get_input_embeddings", "if", "image_hidden_states", "image_token_id", "input_ids", "inputs_embeds", "inputs_merger", "is", "long", "masked_scatter", "return", "self", "special_image_mask", "tensor", "to", "torch", "unsqueeze"], "idefics2/modeling_idefics2.py:Idefics2Model.get_image_features": ["FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "attention_mask", "batch_size", "bool", "class", "config", "connector", "contiguous", "def", "device", "dim", "dimension", "dtype", "else", "get_image_features", "height", "if", "image_hidden_states", "is", "last_hidden_state", "nb_values_per_image", "num_channels", "num_images", "numel", "ones", "patch_attention_mask", "patch_size", "patches_subgrid", "pixel_attention_mask", "pixel_values", "real_images_inds", "return", "self", "shape", "size", "step", "sum", "to", "torch", "unfold", "view", "vision_config", "vision_model", "width"], "idefics2/modeling_idefics2.py:Idefics2Model.forward": ["BoolTensor", "Cache", "DynamicCache", "False", "FlashAttentionKwargs", "FloatTensor", "LongTensor", "ModelBaseModelOutputWithPast", "ModelModel", "ModelPreTrainedModel", "None", "Setting", "Tensor", "True", "Unpack", "ValueError", "You", "_", "and", "at", "attention_mask", "attentions", "batch_size", "bool", "both", "cache_position", "cannot", "checkpointing", "class", "config", "def", "device", "dtype", "either", "elif", "else", "forward", "get_image_features", "get_input_embeddings", "gradient", "gradient_checkpointing", "have", "hidden_states", "if", "image_hidden_states", "incompatible", "input_ids", "inputs_embeds", "inputs_merger", "is", "kwargs", "last_hidden_state", "logger", "not", "or", "outputs", "past_key_values", "pixel_attention_mask", "pixel_values", "position_ids", "r", "raise", "return", "return_dict", "same", "self", "seq_length", "shape", "specify", "text_model", "the", "time", "to", "torch", "training", "tuple", "use_cache", "warning_once", "with"], "idefics2/modeling_idefics2.py:Idefics2ForConditionalGeneration.__init__": ["False", "GenerationMixin", "Linear", "ModelForConditionalGeneration", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "image_token_id", "lm_head", "model", "nn", "post_init", "self", "super", "text_config", "vocab_size"], "idefics2/modeling_idefics2.py:Idefics2ForConditionalGeneration.get_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "model", "return", "self", "text_model"], "idefics2/modeling_idefics2.py:Idefics2ForConditionalGeneration.set_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "model", "self", "set_input_embeddings", "text_model", "value"], "idefics2/modeling_idefics2.py:Idefics2ForConditionalGeneration.get_image_features": ["FloatTensor", "GenerationMixin", "LongTensor", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "class", "def", "get_image_features", "model", "pixel_attention_mask", "pixel_values", "return", "self", "torch"], "idefics2/modeling_idefics2.py:Idefics2ForConditionalGeneration.forward": ["BoolTensor", "Cache", "FloatTensor", "GenerationMixin", "LongTensor", "ModelCausalLMOutputWithPast", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "else", "forward", "hidden_states", "if", "image_hidden_states", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "pixel_attention_mask", "pixel_values", "position_ids", "r", "return", "return_dict", "self", "slice", "slice_indices", "text_config", "torch", "tuple", "use_cache", "use_return_dict", "vocab_size"], "idefics2/modeling_idefics2.py:Idefics2ForConditionalGeneration.prepare_inputs_for_generation": ["False", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "attention_mask", "cache_position", "class", "def", "if", "image_hidden_states", "input_ids", "inputs_embeds", "is", "is_first_iteration", "kwargs", "logits_to_keep", "model_inputs", "not", "or", "past_key_values", "pixel_attention_mask", "pixel_values", "prepare_inputs_for_generation", "return", "self", "super"], "idefics3/modeling_idefics3.py:Idefics3VisionEmbeddings.__init__": ["Conv2d", "Embedding", "ModelVisionConfig", "ModelVisionEmbeddings", "Module", "__init__", "class", "config", "def", "embed_dim", "hidden_size", "image_size", "in_channels", "kernel_size", "nn", "num_channels", "num_patches", "num_patches_per_side", "num_positions", "out_channels", "padding", "patch_embedding", "patch_size", "position_embedding", "self", "stride", "super", "valid"], "idefics3/modeling_idefics3.py:Idefics3VisionEmbeddings.forward": ["BoolTensor", "FloatTensor", "ModelVisionEmbeddings", "Module", "None", "Tensor", "True", "_", "arange", "batch_idx", "batch_size", "boundaries", "bucket_coords_h", "bucket_coords_w", "bucketize", "clamp", "class", "def", "device", "dtype", "embeddings", "enumerate", "fill_value", "flatten", "float32", "for", "forward", "fractional_coords_h", "fractional_coords_w", "full", "h_indices", "in", "max", "max_im_h", "max_im_w", "max_nb_patches_h", "max_nb_patches_w", "nb_patches_h", "nb_patches_w", "nn", "num_patches_per_side", "p_attn_mask", "patch_attention_mask", "patch_embedding", "patch_embeds", "patch_size", "pixel_values", "pos_ids", "position_embedding", "position_ids", "return", "right", "self", "shape", "size", "step_h", "step_w", "sum", "to", "torch", "transpose", "view", "w_indices"], "idefics3/modeling_idefics3.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "attention_mask", "attn_output", "attn_weights", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "softmax", "to", "torch", "training", "transpose", "value"], "idefics3/modeling_idefics3.py:Idefics3VisionAttention.__init__": ["False", "Linear", "ModelVisionAttention", "Module", "ValueError", "__init__", "and", "attention_dropout", "be", "by", "class", "config", "def", "divisible", "dropout", "embed_dim", "f", "got", "head_dim", "hidden_size", "if", "is_causal", "k_proj", "must", "nn", "num_attention_heads", "num_heads", "out_proj", "q_proj", "raise", "scale", "self", "super", "v_proj"], "idefics3/modeling_idefics3.py:Idefics3VisionAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "ModelVisionAttention", "Module", "None", "Tensor", "_attn_implementation", "attention_interface", "attention_mask", "attn_output", "attn_weights", "batch_size", "class", "config", "contiguous", "def", "dropout", "eager", "eager_attention_forward", "else", "embed_dim", "forward", "head_dim", "hidden_states", "if", "is_causal", "k_proj", "keys", "kwargs", "nn", "not", "num_heads", "out_proj", "q_proj", "queries", "reshape", "return", "scale", "scaling", "self", "seq_length", "shape", "torch", "training", "transpose", "tuple", "v_proj", "values", "view"], "idefics3/modeling_idefics3.py:Idefics3VisionMLP.__init__": ["ACT2FN", "Linear", "ModelVisionMLP", "Module", "__init__", "activation_fn", "class", "config", "def", "fc1", "fc2", "hidden_act", "hidden_size", "intermediate_size", "nn", "self", "super"], "idefics3/modeling_idefics3.py:Idefics3VisionMLP.forward": ["ModelVisionMLP", "Module", "Tensor", "activation_fn", "class", "def", "fc1", "fc2", "forward", "hidden_states", "nn", "return", "self", "torch"], "idefics3/modeling_idefics3.py:Idefics3SimpleMLP.__init__": ["False", "Linear", "ModelSimpleMLP", "Module", "__init__", "bias", "class", "config", "def", "hidden_size", "input_size", "nn", "output_size", "proj", "scale_factor", "self", "super", "text_config", "vision_config"], "idefics3/modeling_idefics3.py:Idefics3SimpleMLP.forward": ["ModelSimpleMLP", "Module", "class", "def", "forward", "nn", "proj", "return", "self", "x"], "idefics3/modeling_idefics3.py:Idefics3EncoderLayer.__init__": ["GradientCheckpointingLayer", "LayerNorm", "ModelEncoderLayer", "ModelVisionAttention", "ModelVisionConfig", "ModelVisionMLP", "__init__", "class", "config", "def", "embed_dim", "eps", "hidden_size", "layer_norm1", "layer_norm2", "layer_norm_eps", "mlp", "nn", "self", "self_attn", "super"], "idefics3/modeling_idefics3.py:Idefics3EncoderLayer.forward": ["FloatTensor", "GradientCheckpointingLayer", "ModelEncoderLayer", "Tensor", "TransformersKwargs", "Unpack", "_", "attention_mask", "class", "def", "forward", "hidden_states", "kwargs", "layer_norm1", "layer_norm2", "mlp", "residual", "return", "self", "self_attn", "torch"], "idefics3/modeling_idefics3.py:Idefics3Encoder.__init__": ["False", "ModelConfig", "ModelEncoder", "ModelEncoderLayer", "Module", "ModuleList", "_", "__init__", "class", "config", "def", "for", "gradient_checkpointing", "in", "layers", "nn", "num_hidden_layers", "range", "self", "super"], "idefics3/modeling_idefics3.py:Idefics3Encoder.forward": ["BaseModelOutput", "ModelEncoder", "Module", "None", "Tensor", "attention_mask", "class", "def", "encoder_layer", "for", "forward", "hidden_states", "in", "inputs_embeds", "last_hidden_state", "layer_outputs", "layers", "nn", "return", "self", "torch", "tuple"], "idefics3/modeling_idefics3.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "idefics3/modeling_idefics3.py:Idefics3RMSNorm.__init__": ["ModelRMSNorm", "Module", "Parameter", "__init__", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "idefics3/modeling_idefics3.py:Idefics3RMSNorm.forward": ["ModelRMSNorm", "Module", "True", "class", "def", "dtype", "float32", "forward", "hidden_states", "input_dtype", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "idefics3/modeling_idefics3.py:Idefics3RMSNorm.extra_repr": ["ModelRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "variance_epsilon", "weight"], "idefics3/modeling_idefics3.py:Idefics3Connector.__init__": ["ModelConnector", "ModelSimpleMLP", "Module", "__init__", "class", "config", "def", "modality_projection", "nn", "scale_factor", "self", "super"], "idefics3/modeling_idefics3.py:Idefics3Connector.pixel_shuffle": ["ModelConnector", "Module", "bsz", "class", "def", "embed_dim", "height", "int", "nn", "permute", "pixel_shuffle", "reshape", "return", "scale_factor", "self", "seq", "size", "view", "width", "x"], "idefics3/modeling_idefics3.py:Idefics3Connector.forward": ["ModelConnector", "Module", "class", "def", "forward", "image_hidden_states", "modality_projection", "nn", "pixel_shuffle", "return", "scale_factor", "self"], "idefics3/modeling_idefics3.py:Idefics3VisionTransformer.__init__": ["LayerNorm", "ModelEncoder", "ModelPreTrainedModel", "ModelVisionConfig", "ModelVisionEmbeddings", "ModelVisionTransformer", "__init__", "class", "config", "def", "embed_dim", "embeddings", "encoder", "eps", "hidden_size", "layer_norm_eps", "nn", "patch_size", "post_init", "post_layernorm", "self", "super"], "idefics3/modeling_idefics3.py:Idefics3VisionTransformer.get_input_embeddings": ["ModelPreTrainedModel", "ModelVisionTransformer", "class", "def", "embeddings", "get_input_embeddings", "return", "self"], "idefics3/modeling_idefics3.py:Idefics3VisionTransformer.set_input_embeddings": ["ModelPreTrainedModel", "ModelVisionTransformer", "class", "def", "embeddings", "self", "set_input_embeddings", "value"], "idefics3/modeling_idefics3.py:Idefics3VisionTransformer.forward": ["BaseModelOutput", "BoolTensor", "ModelPreTrainedModel", "ModelVisionTransformer", "None", "TransformersKwargs", "Unpack", "attention_mask", "batch_size", "bool", "class", "config", "create_bidirectional_mask", "def", "device", "dtype", "embeddings", "encoder", "encoder_outputs", "forward", "hidden_states", "if", "input_embeds", "inputs_embeds", "is", "kwargs", "last_hidden_state", "ones", "patch_attention_mask", "patch_size", "pixel_values", "post_layernorm", "return", "self", "size", "to", "torch", "tuple", "view"], "idefics3/modeling_idefics3.py:Idefics3Model.__init__": ["AutoModel", "ModelConfig", "ModelConnector", "ModelModel", "ModelPreTrainedModel", "ModelVisionTransformer", "__init__", "_from_config", "class", "config", "connector", "def", "from_config", "image_seq_len", "image_size", "image_token_id", "int", "pad_token_id", "padding_idx", "patch_size", "post_init", "scale_factor", "self", "super", "text_config", "text_model", "vision_config", "vision_model", "vocab_size"], "idefics3/modeling_idefics3.py:Idefics3Model.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "return", "self", "text_model"], "idefics3/modeling_idefics3.py:Idefics3Model.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "self", "set_input_embeddings", "text_model", "value"], "idefics3/modeling_idefics3.py:Idefics3Model.inputs_merger": ["LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "all", "class", "config", "def", "device", "dtype", "else", "expand_as", "get_input_embeddings", "if", "image_hidden_states", "image_token_id", "input_ids", "inputs_embeds", "inputs_merger", "is", "long", "masked_scatter", "return", "self", "special_image_mask", "tensor", "to", "torch", "unsqueeze"], "idefics3/modeling_idefics3.py:Idefics3Model.get_image_features": ["FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "batch_size", "bool", "class", "config", "connector", "contiguous", "def", "device", "dim", "dimension", "dtype", "else", "get_image_features", "height", "if", "image_hidden_states", "is", "last_hidden_state", "nb_values_per_image", "num_channels", "num_images", "numel", "ones", "patch_attention_mask", "patch_size", "patches_subgrid", "pixel_attention_mask", "pixel_values", "real_images_inds", "return", "self", "shape", "size", "step", "sum", "to", "torch", "unfold", "view", "vision_config", "vision_model", "width"], "idefics3/modeling_idefics3.py:Idefics3Model.forward": ["BoolTensor", "Cache", "DynamicCache", "False", "FlashAttentionKwargs", "FloatTensor", "LongTensor", "ModelBaseModelOutputWithPast", "ModelModel", "ModelPreTrainedModel", "None", "Setting", "Tensor", "True", "Unpack", "ValueError", "You", "_", "and", "at", "attention_mask", "attentions", "batch_size", "bool", "both", "cache_position", "cannot", "checkpointing", "class", "config", "def", "device", "dtype", "either", "elif", "else", "forward", "get_image_features", "get_input_embeddings", "gradient", "gradient_checkpointing", "have", "hidden_states", "if", "image_hidden_states", "incompatible", "input_ids", "inputs_embeds", "inputs_merger", "is", "kwargs", "last_hidden_state", "logger", "not", "or", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "pixel_attention_mask", "pixel_values", "position_ids", "r", "raise", "return", "return_dict", "same", "self", "seq_length", "shape", "specify", "text_model", "the", "time", "to", "torch", "training", "tuple", "use_cache", "use_return_dict", "warning_once", "with"], "idefics3/modeling_idefics3.py:Idefics3ForConditionalGeneration.__init__": ["False", "GenerationMixin", "Linear", "ModelForConditionalGeneration", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "image_token_id", "lm_head", "model", "nn", "post_init", "self", "super", "text_config", "vocab_size"], "idefics3/modeling_idefics3.py:Idefics3ForConditionalGeneration.get_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "model", "return", "self", "text_model"], "idefics3/modeling_idefics3.py:Idefics3ForConditionalGeneration.set_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "model", "self", "set_input_embeddings", "text_model", "value"], "idefics3/modeling_idefics3.py:Idefics3ForConditionalGeneration.get_image_features": ["FloatTensor", "GenerationMixin", "LongTensor", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "class", "def", "get_image_features", "model", "pixel_attention_mask", "pixel_values", "return", "self", "torch"], "idefics3/modeling_idefics3.py:Idefics3ForConditionalGeneration.forward": ["BoolTensor", "Cache", "FloatTensor", "GenerationMixin", "LongTensor", "ModelCausalLMOutputWithPast", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "else", "forward", "hidden_states", "if", "image_hidden_states", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "pixel_attention_mask", "pixel_values", "position_ids", "r", "return", "return_dict", "self", "slice", "slice_indices", "text_config", "torch", "tuple", "use_cache", "use_return_dict", "vocab_size"], "idefics3/modeling_idefics3.py:Idefics3ForConditionalGeneration.prepare_inputs_for_generation": ["False", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "attention_mask", "cache_position", "class", "def", "if", "image_hidden_states", "input_ids", "inputs_embeds", "is", "is_first_iteration", "kwargs", "logits_to_keep", "model_inputs", "not", "or", "past_key_values", "pixel_attention_mask", "pixel_values", "prepare_inputs_for_generation", "return", "self", "super"], "ijepa/modeling_ijepa.py:IJepaPatchEmbeddings.__init__": ["Conv2d", "Iterable", "Model", "ModelConfig", "Module", "__init__", "abc", "class", "collections", "config", "def", "else", "hidden_size", "if", "image_size", "isinstance", "kernel_size", "nn", "num_channels", "num_patches", "patch_size", "projection", "self", "stride", "super"], "ijepa/modeling_ijepa.py:IJepaPatchEmbeddings.forward": ["Expected", "False", "Input", "Make", "Model", "Module", "Tensor", "ValueError", "batch_size", "bool", "but", "channel", "class", "configuration", "def", "dimension", "doesn", "embeddings", "f", "flatten", "forward", "got", "height", "if", "image", "image_size", "in", "interpolate_pos_encoding", "match", "model", "nn", "not", "num_channels", "of", "one", "or", "pixel", "pixel_values", "projection", "raise", "return", "self", "set", "shape", "size", "sure", "t", "that", "the", "torch", "transpose", "values", "width", "with"], "ijepa/modeling_ijepa.py:IJepaEmbeddings.__init__": ["Dropout", "False", "Model", "ModelConfig", "ModelPatchEmbeddings", "Module", "None", "Parameter", "__init__", "bool", "class", "config", "def", "dropout", "else", "hidden_dropout_prob", "hidden_size", "if", "mask_token", "nn", "num_patches", "patch_embeddings", "patch_size", "position_embeddings", "randn", "self", "super", "torch", "use_mask_token", "zeros"], "ijepa/modeling_ijepa.py:IJepaEmbeddings.interpolate_pos_encoding": ["False", "Model", "Module", "Tensor", "align_corners", "and", "bicubic", "class", "def", "dim", "embeddings", "functional", "height", "if", "int", "interpolate", "interpolate_pos_encoding", "is_tracing", "jit", "mode", "new_height", "new_width", "nn", "not", "num_patches", "num_positions", "patch_pos_embed", "patch_size", "permute", "position_embeddings", "reshape", "return", "self", "shape", "size", "sqrt_num_positions", "torch", "torch_int", "view", "width"], "ijepa/modeling_ijepa.py:IJepaEmbeddings.forward": ["BoolTensor", "False", "Model", "Module", "None", "Tensor", "_", "batch_size", "bool", "bool_masked_pos", "class", "def", "dropout", "else", "embeddings", "expand", "forward", "height", "if", "interpolate_pos_encoding", "is", "mask", "mask_token", "mask_tokens", "nn", "not", "patch_embeddings", "pixel_values", "position_embeddings", "return", "self", "seq_length", "shape", "torch", "type_as", "unsqueeze", "width"], "ijepa/modeling_ijepa.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "contiguous", "def", "dim", "dropout", "float", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "shape", "size", "softmax", "torch", "training", "transpose", "value"], "ijepa/modeling_ijepa.py:IJepaSelfAttention.__init__": ["False", "Linear", "Model", "ModelConfig", "Module", "The", "ValueError", "__init__", "a", "all_head_size", "and", "attention", "attention_head_size", "attention_probs_dropout_prob", "bias", "class", "config", "def", "dropout_prob", "embedding_size", "f", "hasattr", "heads", "hidden", "hidden_size", "if", "int", "is", "is_causal", "key", "multiple", "nn", "not", "num_attention_heads", "number", "of", "qkv_bias", "query", "raise", "scaling", "self", "size", "super", "the", "value"], "ijepa/modeling_ijepa.py:IJepaSelfAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "Model", "Module", "None", "Tensor", "_attn_implementation", "all_head_size", "attention_head_size", "attention_interface", "attention_probs", "batch_size", "class", "config", "context_layer", "def", "dropout", "dropout_prob", "eager", "eager_attention_forward", "else", "forward", "hidden_states", "if", "is_causal", "key", "key_layer", "new_context_layer_shape", "new_shape", "nn", "not", "num_attention_heads", "query", "query_layer", "reshape", "return", "scaling", "self", "shape", "size", "torch", "training", "transpose", "tuple", "value", "value_layer", "view"], "ijepa/modeling_ijepa.py:IJepaSelfOutput.__init__": ["Dropout", "Linear", "Model", "ModelConfig", "Module", "__init__", "class", "config", "def", "dense", "dropout", "hidden_dropout_prob", "hidden_size", "nn", "self", "super"], "ijepa/modeling_ijepa.py:IJepaSelfOutput.forward": ["Model", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "ijepa/modeling_ijepa.py:IJepaAttention.__init__": ["Model", "ModelConfig", "ModelSelfAttention", "ModelSelfOutput", "Module", "__init__", "attention", "class", "config", "def", "nn", "output", "self", "super"], "ijepa/modeling_ijepa.py:IJepaAttention.forward": ["Model", "Module", "Tensor", "_", "attention", "class", "def", "forward", "hidden_states", "nn", "output", "return", "self", "self_attn_output", "torch"], "ijepa/modeling_ijepa.py:IJepaIntermediate.__init__": ["ACT2FN", "Linear", "Model", "ModelConfig", "Module", "__init__", "class", "config", "def", "dense", "else", "hidden_act", "hidden_size", "if", "intermediate_act_fn", "intermediate_size", "isinstance", "nn", "self", "str", "super"], "ijepa/modeling_ijepa.py:IJepaIntermediate.forward": ["Model", "Module", "Tensor", "class", "def", "dense", "forward", "hidden_states", "intermediate_act_fn", "nn", "return", "self", "torch"], "ijepa/modeling_ijepa.py:IJepaOutput.__init__": ["Dropout", "Linear", "Model", "ModelConfig", "Module", "__init__", "class", "config", "def", "dense", "dropout", "hidden_dropout_prob", "hidden_size", "intermediate_size", "nn", "self", "super"], "ijepa/modeling_ijepa.py:IJepaOutput.forward": ["Model", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "ijepa/modeling_ijepa.py:IJepaLayer.__init__": ["GradientCheckpointingLayer", "LayerNorm", "Model", "ModelAttention", "ModelConfig", "ModelIntermediate", "ModelOutput", "__init__", "attention", "chunk_size_feed_forward", "class", "config", "def", "eps", "hidden_size", "intermediate", "layer_norm_eps", "layernorm_after", "layernorm_before", "nn", "output", "self", "seq_len_dim", "super"], "ijepa/modeling_ijepa.py:IJepaLayer.forward": ["GradientCheckpointingLayer", "Model", "Tensor", "attention", "attention_output", "class", "def", "forward", "hidden_states", "hidden_states_norm", "intermediate", "layer_output", "layernorm_after", "layernorm_before", "output", "return", "self", "torch"], "ijepa/modeling_ijepa.py:IJepaPreTrainedModel._init_weights": ["Conv2d", "LayerNorm", "Linear", "Model", "ModelEmbeddings", "None", "PreTrainedModel", "_init_weights", "bias", "class", "config", "def", "elif", "if", "init", "initializer_range", "is", "isinstance", "mask_token", "mean", "module", "nn", "not", "ones_", "position_embeddings", "self", "std", "trunc_normal_", "weight", "zeros_"], "ijepa/modeling_ijepa.py:IJepaEncoder.__init__": ["False", "Model", "ModelConfig", "ModelLayer", "Module", "ModuleList", "_", "__init__", "class", "config", "def", "for", "gradient_checkpointing", "in", "layer", "nn", "num_hidden_layers", "range", "self", "super"], "ijepa/modeling_ijepa.py:IJepaEncoder.forward": ["BaseModelOutput", "Model", "Module", "Tensor", "class", "def", "enumerate", "for", "forward", "hidden_states", "i", "in", "last_hidden_state", "layer", "layer_module", "nn", "return", "self", "torch"], "ijepa/modeling_ijepa.py:IJepaPooler.__init__": ["ACT2FN", "Linear", "Model", "ModelConfig", "Module", "__init__", "activation", "class", "config", "def", "dense", "hidden_size", "nn", "pooler_act", "pooler_output_size", "self", "super"], "ijepa/modeling_ijepa.py:IJepaPooler.forward": ["Model", "Module", "Tensor", "activation", "class", "def", "dense", "first_token_tensor", "forward", "hidden_states", "nn", "pooled_output", "return", "self", "torch"], "ijepa/modeling_ijepa.py:IJepaModel.__init__": ["False", "LayerNorm", "Model", "ModelConfig", "ModelEmbeddings", "ModelEncoder", "ModelPooler", "ModelPreTrainedModel", "None", "__init__", "add_pooling_layer", "bool", "class", "config", "def", "else", "embeddings", "encoder", "eps", "hidden_size", "if", "layer_norm_eps", "layernorm", "nn", "pooler", "post_init", "r", "self", "super", "use_mask_token"], "ijepa/modeling_ijepa.py:IJepaModel.get_input_embeddings": ["Model", "ModelPatchEmbeddings", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "patch_embeddings", "return", "self"], "ijepa/modeling_ijepa.py:IJepaModel.forward": ["BaseModelOutput", "BaseModelOutputWithPooling", "BoolTensor", "Model", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "bool", "bool_masked_pos", "class", "def", "dtype", "else", "embedding_output", "embeddings", "encoder", "encoder_outputs", "expected_dtype", "forward", "have", "if", "interpolate_pos_encoding", "is", "kwargs", "last_hidden_state", "layernorm", "not", "patch_embeddings", "pixel_values", "pooled_output", "pooler", "pooler_output", "projection", "r", "raise", "return", "self", "sequence_output", "specify", "to", "torch", "weight"], "ijepa/modeling_ijepa.py:IJepaForImageClassification.__init__": ["False", "Identity", "Linear", "Model", "ModelConfig", "ModelModel", "ModelPreTrainedModel", "__init__", "add_pooling_layer", "class", "classifier", "config", "def", "else", "hidden_size", "if", "nn", "num_labels", "post_init", "self", "super"], "ijepa/modeling_ijepa.py:IJepaForImageClassification.forward": ["BaseModelOutputWithPooling", "ImageClassifierOutput", "Model", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "attentions", "bool", "class", "classifier", "config", "def", "dim", "forward", "hidden_states", "if", "interpolate_pos_encoding", "is", "kwargs", "labels", "last_hidden_state", "logits", "loss", "loss_function", "mean", "not", "outputs", "pixel_values", "r", "return", "self", "sequence_output", "torch"], "imagegpt/modeling_imagegpt.py:ImageGPTLayerNorm.__init__": ["ModelLayerNorm", "Module", "Parameter", "Tensor", "__init__", "class", "def", "eps", "float", "hidden_size", "int", "nn", "self", "super", "torch", "tuple", "weight"], "imagegpt/modeling_imagegpt.py:ImageGPTLayerNorm.forward": ["ModelLayerNorm", "Module", "Tensor", "True", "axis", "class", "def", "eps", "forward", "keepdim", "mean", "nn", "return", "self", "sqrt", "square", "tensor", "torch", "weight"], "imagegpt/modeling_imagegpt.py:ImageGPTAttention.__init__": ["Conv1D", "Dropout", "False", "ModelAttention", "Module", "None", "ValueError", "__init__", "and", "attn_dropout", "attn_pdrop", "be", "bias", "bool", "by", "c_attn", "c_proj", "class", "config", "def", "divisible", "dtype", "else", "embed_dim", "f", "got", "head_dim", "hidden_size", "if", "int", "is_cross_attention", "layer_idx", "max_position_embeddings", "max_positions", "must", "nn", "num_attention_heads", "num_heads", "ones", "persistent", "q_attn", "raise", "register_buffer", "reorder_and_upcast_attn", "resid_dropout", "resid_pdrop", "scale_attn_by_inverse_layer_idx", "scale_attn_weights", "self", "split_size", "super", "torch", "tril", "view"], "imagegpt/modeling_imagegpt.py:ImageGPTAttention._attn": ["ModelAttention", "Module", "None", "Softmax", "_attn", "attention_mask", "attn_dropout", "attn_output", "attn_weights", "bias", "causal_mask", "class", "def", "device", "dim", "dtype", "finfo", "float", "if", "is", "is_cross_attention", "key", "key_length", "layer_idx", "mask_value", "matmul", "min", "nn", "not", "query", "query_length", "return", "scale_attn_by_inverse_layer_idx", "scale_attn_weights", "self", "size", "tensor", "torch", "torch_float", "transpose", "type", "value", "where"], "imagegpt/modeling_imagegpt.py:ImageGPTAttention._upcast_and_reordered_attn": ["Error", "False", "ModelAttention", "Module", "None", "RuntimeError", "Softmax", "_", "_upcast_and_reordered_attn", "alpha", "attention_mask", "attn_dropout", "attn_output", "attn_weights", "baddbmm", "beta", "bias", "bsz", "causal_mask", "class", "def", "device", "dim", "dk", "does", "dtype", "empty", "enabled", "finfo", "float", "float32", "have", "if", "is", "is_cross_attention", "k", "k_seq_len", "key", "key_length", "layer_idx", "mask_value", "matmul", "maybe_autocast", "min", "nn", "not", "num_heads", "q", "q_seq_len", "query", "query_length", "raise", "reshape", "return", "scale_attn_by_inverse_layer_idx", "scale_attn_weights", "scale_factor", "self", "size", "tensor", "torch", "transpose", "type", "upcasting", "value", "where", "with"], "imagegpt/modeling_imagegpt.py:ImageGPTAttention._split_heads": ["ModelAttention", "Module", "_split_heads", "attn_head_size", "class", "def", "new_shape", "nn", "num_heads", "permute", "return", "self", "size", "tensor", "view"], "imagegpt/modeling_imagegpt.py:ImageGPTAttention._merge_heads": ["ModelAttention", "Module", "_merge_heads", "attn_head_size", "class", "contiguous", "def", "new_shape", "nn", "num_heads", "permute", "return", "self", "size", "tensor", "view"], "imagegpt/modeling_imagegpt.py:ImageGPTAttention.forward": ["Cache", "EncoderDecoderCache", "False", "If", "ModelAttention", "Module", "None", "Please", "Tensor", "True", "ValueError", "_", "_attn", "_merge_heads", "_upcast_and_reordered_attn", "and", "as", "attention", "attention_mask", "attn_output", "attn_weights", "be", "bool", "bsz", "c_attn", "c_proj", "cache_position", "class", "cross", "cross_attention_cache", "curr_past_key_values", "current_states", "def", "defined", "dim", "else", "encoder_attention_mask", "encoder_hidden_states", "forward", "get", "hasattr", "have", "head_dim", "hidden_states", "if", "instantiate", "is", "is_cross_attention", "is_updated", "isinstance", "key", "keys", "layer_idx", "layer_past", "layers", "make", "nn", "not", "num_heads", "output_attentions", "q_attn", "query", "raise", "reorder_and_upcast_attn", "resid_dropout", "return", "self", "self_attention_cache", "seq_len", "shape", "split", "split_size", "sure", "the", "to", "torch", "transpose", "tuple", "update", "use_cache", "used", "value", "values", "view", "weights", "with"], "imagegpt/modeling_imagegpt.py:ImageGPTMLP.__init__": ["ACT2FN", "Conv1D", "Dropout", "ModelMLP", "Module", "__init__", "act", "activation_function", "c_fc", "c_proj", "class", "config", "def", "dropout", "embed_dim", "hidden_size", "intermediate_size", "nn", "resid_pdrop", "self", "super"], "imagegpt/modeling_imagegpt.py:ImageGPTMLP.forward": ["ModelMLP", "Module", "Tensor", "act", "c_fc", "c_proj", "class", "def", "dropout", "forward", "hidden_states", "nn", "return", "self", "torch"], "imagegpt/modeling_imagegpt.py:ImageGPTBlock.__init__": ["GradientCheckpointingLayer", "ModelAttention", "ModelBlock", "ModelLayerNorm", "ModelMLP", "None", "True", "__init__", "add_cross_attention", "attn", "class", "config", "crossattention", "def", "else", "eps", "hidden_size", "if", "inner_dim", "is", "is_cross_attention", "layer_idx", "layer_norm_epsilon", "ln_1", "ln_2", "ln_cross_attn", "mlp", "n_inner", "not", "self", "super"], "imagegpt/modeling_imagegpt.py:ImageGPTBlock.forward": ["Cache", "False", "GradientCheckpointingLayer", "If", "ModelBlock", "None", "Tensor", "True", "ValueError", "add_cross_attention", "are", "attention", "attention_mask", "attn", "attn_output", "attn_outputs", "be", "bool", "by", "cache_position", "class", "config", "cross", "cross_attn_outputs", "crossattention", "def", "encoder_attention_mask", "encoder_hidden_states", "f", "feed_forward_hidden_states", "forward", "has", "hasattr", "hidden_states", "if", "instantiated", "is", "layer_past", "layers", "ln_1", "ln_2", "ln_cross_attn", "mlp", "not", "output_attentions", "outputs", "passed", "raise", "residual", "return", "self", "setting", "to", "torch", "tuple", "use_cache", "with"], "imagegpt/modeling_imagegpt.py:ImageGPTPreTrainedModel._init_weights": ["ModelAttention", "ModelPreTrainedModel", "PreTrainedModel", "_init_weights", "and", "bias", "bool", "c_proj", "class", "config", "copy_", "def", "dtype", "elif", "for", "if", "in", "init", "initializer_range", "isinstance", "math", "max_position_embeddings", "max_positions", "mean", "module", "n_layer", "name", "named_parameters", "normal_", "ones", "p", "self", "sqrt", "std", "super", "torch", "tril", "view", "weight"], "imagegpt/modeling_imagegpt.py:ImageGPTModel.__init__": ["Dropout", "Embedding", "False", "ModelBlock", "ModelConfig", "ModelLayerNorm", "ModelModel", "ModelPreTrainedModel", "ModuleList", "__init__", "class", "config", "def", "drop", "embd_pdrop", "embed_dim", "eps", "for", "gradient_checkpointing", "h", "hidden_size", "i", "in", "layer_idx", "layer_norm_epsilon", "ln_f", "max_position_embeddings", "nn", "num_hidden_layers", "post_init", "range", "self", "super", "vocab_size", "wpe", "wte"], "imagegpt/modeling_imagegpt.py:ImageGPTModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "return", "self", "wte"], "imagegpt/modeling_imagegpt.py:ImageGPTModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "new_embeddings", "self", "set_input_embeddings", "wte"], "imagegpt/modeling_imagegpt.py:ImageGPTModel.forward": ["Any", "BaseModelOutputWithPastAndCrossAttentions", "Cache", "DynamicCache", "False", "ModelModel", "ModelPreTrainedModel", "None", "Setting", "Tensor", "True", "ValueError", "You", "_", "add_cross_attention", "all_cross_attentions", "all_hidden_states", "all_self_attentions", "and", "arange", "at", "attention_mask", "attentions", "batch_size", "be", "block", "bool", "both", "cache_position", "cannot", "checkpointing", "class", "config", "cross_attentions", "def", "defined", "device", "drop", "dtype", "either", "elif", "else", "encoder_attention_mask", "encoder_batch_size", "encoder_hidden_shape", "encoder_hidden_states", "encoder_sequence_length", "enumerate", "finfo", "for", "forward", "get_seq_length", "gradient", "gradient_checkpointing", "h", "has", "have", "hidden_states", "i", "if", "in", "incompatible", "input_ids", "input_shape", "inputs_embeds", "invert_attention_mask", "is", "kwargs", "last_hidden_state", "ln_f", "logger", "min", "not", "ones", "or", "output_attentions", "output_hidden_states", "output_shape", "outputs", "past_key_values", "past_seen_tokens", "position_embeds", "position_ids", "r", "raise", "return", "return_dict", "same", "self", "shape", "size", "specify", "the", "time", "to", "token_type_embeds", "token_type_ids", "torch", "training", "tuple", "unsqueeze", "use_cache", "use_return_dict", "v", "view", "warn_if_padding_and_no_attention_mask", "warning_once", "with", "wpe", "wte"], "imagegpt/modeling_imagegpt.py:ImageGPTForCausalImageModeling.__init__": ["False", "GenerationMixin", "Linear", "ModelConfig", "ModelForCausalModelModeling", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "lm_head", "n_embd", "nn", "post_init", "self", "super", "transformer", "vocab_size"], "imagegpt/modeling_imagegpt.py:ImageGPTForCausalImageModeling.forward": ["Any", "Cache", "CausalLMOutputWithCrossAttentions", "CrossEntropyLoss", "GenerationMixin", "ModelForCausalModelModeling", "ModelPreTrainedModel", "None", "Tensor", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "contiguous", "cross_attentions", "def", "else", "encoder_attention_mask", "encoder_hidden_states", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "lm_head", "lm_logits", "logits", "loss", "loss_fct", "not", "output", "output_attentions", "output_hidden_states", "past_key_values", "position_ids", "r", "return", "return_dict", "self", "shift_labels", "shift_logits", "size", "token_type_ids", "torch", "transformer", "transformer_outputs", "tuple", "use_cache", "use_return_dict", "view"], "imagegpt/modeling_imagegpt.py:ImageGPTForImageClassification.__init__": ["False", "Linear", "ModelConfig", "ModelForModelClassification", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "n_embd", "nn", "num_labels", "post_init", "score", "self", "super", "transformer"], "imagegpt/modeling_imagegpt.py:ImageGPTForImageClassification.forward": ["Any", "Cache", "ModelForModelClassification", "ModelPreTrainedModel", "None", "SequenceClassifierOutputWithPast", "Tensor", "attention_mask", "attentions", "bool", "class", "config", "def", "dim", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_function", "mean", "not", "output", "output_attentions", "output_hidden_states", "past_key_values", "pooled_hidden_states", "position_ids", "r", "return", "return_dict", "score", "self", "token_type_ids", "torch", "transformer", "transformer_outputs", "tuple", "use_cache", "use_return_dict"], "informer/modeling_informer.py:InformerFeatureEmbedder.__init__": ["Embedding", "ModelFeatureEmbedder", "Module", "ModuleList", "None", "__init__", "c", "cardinalities", "class", "d", "def", "embedders", "embedding_dims", "for", "in", "int", "len", "list", "nn", "num_features", "self", "super", "zip"], "informer/modeling_informer.py:InformerFeatureEmbedder.forward": ["ModelFeatureEmbedder", "Module", "Tensor", "cat", "cat_feature_slice", "cat_feature_slices", "chunk", "class", "def", "dim", "else", "embed", "embedders", "features", "for", "forward", "if", "in", "nn", "num_features", "return", "self", "squeeze", "torch", "zip"], "informer/modeling_informer.py:InformerStdScaler.__init__": ["ModelConfig", "ModelStdScaler", "Module", "True", "__init__", "class", "config", "def", "dim", "else", "hasattr", "if", "keepdim", "minimum_scale", "nn", "scaling_dim", "self", "super"], "informer/modeling_informer.py:InformerStdScaler.forward": ["ModelStdScaler", "Module", "Tensor", "clamp_min", "class", "data", "def", "denominator", "dim", "forward", "keepdim", "loc", "minimum_scale", "nn", "observed_indicator", "return", "scale", "self", "sqrt", "sum", "torch", "tuple", "variance"], "informer/modeling_informer.py:InformerMeanScaler.__init__": ["ModelConfig", "ModelMeanScaler", "Module", "None", "True", "__init__", "class", "config", "def", "default_scale", "dim", "else", "hasattr", "if", "keepdim", "minimum_scale", "nn", "scaling_dim", "self", "super"], "informer/modeling_informer.py:InformerMeanScaler.forward": ["ModelMeanScaler", "Module", "None", "Tensor", "True", "abs", "batch_observations", "batch_sum", "clamp", "class", "data", "def", "default_scale", "dim", "else", "forward", "if", "is", "keepdim", "min", "minimum_scale", "nn", "not", "num_observed", "observed_indicator", "ones_like", "return", "scale", "scaled_data", "self", "squeeze", "sum", "torch", "ts_sum", "tuple", "where", "zeros_like"], "informer/modeling_informer.py:InformerNOPScaler.__init__": ["ModelConfig", "ModelNOPScaler", "Module", "True", "__init__", "class", "config", "def", "dim", "else", "hasattr", "if", "keepdim", "nn", "scaling_dim", "self", "super"], "informer/modeling_informer.py:InformerNOPScaler.forward": ["False", "ModelNOPScaler", "Module", "None", "Tensor", "class", "data", "def", "dim", "forward", "keepdim", "loc", "mean", "nn", "observed_indicator", "ones_like", "requires_grad", "return", "scale", "self", "torch", "tuple", "zeros_like"], "informer/modeling_informer.py:InformerSinusoidalPositionalEmbedding.__init__": ["Embedding", "ModelSinusoidalPositionalEmbedding", "None", "True", "__init__", "_freeze", "class", "def", "embedding_dim", "int", "nn", "num_positions", "padding_idx", "self", "super"], "informer/modeling_informer.py:InformerSinusoidalPositionalEmbedding.create_weight": ["Embedding", "False", "FloatTensor", "ModelSinusoidalPositionalEmbedding", "array", "class", "cos", "create_weight", "def", "dim", "dtype", "else", "empty", "for", "if", "in", "j", "n_pos", "nn", "np", "out", "pos", "position_enc", "power", "range", "requires_grad", "return", "self", "sentinel", "shape", "sin", "torch", "weight"], "informer/modeling_informer.py:InformerSinusoidalPositionalEmbedding.forward": ["Embedding", "ModelSinusoidalPositionalEmbedding", "None", "Size", "Tensor", "arange", "bsz", "class", "def", "device", "dtype", "forward", "if", "input_ids_shape", "int", "is", "long", "nn", "past_key_values_length", "position_ids", "return", "self", "seq_len", "super", "torch", "weight"], "informer/modeling_informer.py:InformerValueEmbedding.__init__": ["False", "Linear", "ModelValueEmbedding", "Module", "__init__", "bias", "class", "d_model", "def", "feature_size", "in_features", "nn", "out_features", "self", "super", "value_projection"], "informer/modeling_informer.py:InformerValueEmbedding.forward": ["ModelValueEmbedding", "Module", "class", "def", "forward", "nn", "return", "self", "value_projection", "x"], "informer/modeling_informer.py:InformerPreTrainedModel._init_weights": ["ModelPreTrainedModel", "ModelSinusoidalPositionalEmbedding", "Module", "PreTrainedModel", "_init_weights", "class", "copy_", "create_weight", "def", "if", "init", "isinstance", "module", "nn", "self", "super", "weight"], "informer/modeling_informer.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "contiguous", "def", "dim", "dropout", "float", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "shape", "size", "softmax", "torch", "training", "transpose", "value"], "informer/modeling_informer.py:InformerAttention.__init__": ["False", "Instantiating", "Linear", "ModelAttention", "ModelConfig", "Module", "None", "Please", "True", "ValueError", "__class__", "__init__", "__name__", "a", "and", "be", "bias", "bool", "by", "caching", "call", "class", "config", "creating", "decoder", "def", "divisible", "dropout", "during", "embed_dim", "errors", "f", "float", "forward", "got", "head_dim", "if", "int", "is", "is_causal", "is_decoder", "k_proj", "layer_idx", "lead", "logger", "make", "must", "nn", "not", "num_heads", "out_proj", "passing", "provide", "q_proj", "raise", "recommended", "scaling", "self", "super", "sure", "the", "this", "to", "used", "v_proj", "warning_once", "when", "will", "without"], "informer/modeling_informer.py:InformerAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "EncoderDecoderCache", "False", "FlashAttentionKwargs", "ModelAttention", "Module", "None", "Tensor", "True", "Unpack", "_attn_implementation", "and", "attention_interface", "attention_mask", "attn_output", "attn_weights", "bool", "bsz", "cache_position", "class", "config", "contiguous", "cross_attention_cache", "curr_past_key_values", "current_states", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "get", "head_dim", "hidden_states", "if", "is", "is_cross_attention", "is_updated", "isinstance", "k_proj", "key_states", "key_value_states", "keys", "kv_input_shape", "kwargs", "layer_idx", "layers", "nn", "not", "out_proj", "output_attentions", "past_key_values", "q_input_shape", "q_proj", "query_states", "reshape", "return", "scaling", "self", "self_attention_cache", "shape", "src_len", "tgt_len", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "values", "view"], "informer/modeling_informer.py:InformerProbSparseAttention.__init__": ["False", "Linear", "ModelProbSparseAttention", "Module", "None", "True", "ValueError", "__init__", "and", "be", "bias", "bool", "by", "class", "def", "divisible", "dropout", "embed_dim", "f", "factor", "float", "got", "head_dim", "if", "int", "is_decoder", "k_proj", "layer_idx", "must", "nn", "num_heads", "out_proj", "q_proj", "raise", "sampling_factor", "scaling", "self", "super", "v_proj"], "informer/modeling_informer.py:InformerProbSparseAttention._shape": ["ModelProbSparseAttention", "Module", "Tensor", "_shape", "bsz", "class", "contiguous", "def", "head_dim", "int", "nn", "num_heads", "return", "self", "seq_len", "tensor", "torch", "transpose", "view"], "informer/modeling_informer.py:InformerProbSparseAttention.forward": ["Attention", "Cache", "EncoderDecoderCache", "False", "ModelProbSparseAttention", "Module", "None", "Tensor", "True", "ValueError", "_", "_shape", "and", "arange", "astype", "attention_mask", "attn_output", "attn_probs", "attn_weights", "attn_weights_reshaped", "be", "bmm", "bool", "bsz", "but", "cache_position", "ceil", "class", "clone", "context", "cross_attention_cache", "cumsum", "curr_past_key_values", "current_states", "def", "dim", "dim_for_slice", "div", "dropout", "dtype", "else", "embed_dim", "expand", "f", "factor", "float32", "forward", "functional", "get", "head_dim", "hidden_states", "if", "index_sample", "int", "is", "is_cross_attention", "is_decoder", "is_updated", "isinstance", "item", "k_proj", "k_sample", "key_states", "key_states_time_length", "key_value_states", "keys", "kv_input_shape", "layer_idx", "layers", "log1p", "log_key_states_time_length", "log_query_states_time_length", "mask", "max", "mean", "min", "nn", "not", "np", "num_heads", "of", "out_proj", "output_attentions", "p", "past_key_values", "prob_mask", "proj_shape", "q_proj", "q_reduce", "queries_keys_sample", "query_states", "query_states_time_length", "raise", "randint", "reshape", "return", "scaling", "self", "self_attention_cache", "shape", "should", "size", "softmax", "sorted", "sparsity_measurement", "src_len", "sum", "tgt_len", "to", "top_u_sparsity_measurement", "topk", "torch", "training", "transpose", "tuple", "u", "u_part", "unsqueeze", "update", "v_mean_dim_time", "v_proj", "value_states", "values", "view", "weights"], "informer/modeling_informer.py:InformerConvLayer.__init__": ["BatchNorm1d", "Conv1d", "ELU", "GradientCheckpointingLayer", "MaxPool1d", "ModelConvLayer", "__init__", "activation", "c_in", "circular", "class", "def", "downConv", "in_channels", "kernel_size", "maxPool", "nn", "norm", "out_channels", "padding", "padding_mode", "self", "stride", "super"], "informer/modeling_informer.py:InformerConvLayer.forward": ["GradientCheckpointingLayer", "ModelConvLayer", "activation", "class", "def", "downConv", "forward", "maxPool", "norm", "permute", "return", "self", "transpose", "x"], "informer/modeling_informer.py:InformerEncoderLayer.__init__": ["ACT2FN", "GradientCheckpointingLayer", "LayerNorm", "Linear", "ModelAttention", "ModelConfig", "ModelEncoderLayer", "ModelProbSparseAttention", "__init__", "activation_dropout", "activation_fn", "activation_function", "attention_dropout", "attention_type", "class", "config", "d_model", "def", "dropout", "else", "embed_dim", "encoder_attention_heads", "encoder_ffn_dim", "fc1", "fc2", "final_layer_norm", "if", "nn", "num_heads", "prob", "sampling_factor", "self", "self_attn", "self_attn_layer_norm", "super"], "informer/modeling_informer.py:InformerEncoderLayer.forward": ["False", "FloatTensor", "GradientCheckpointingLayer", "ModelEncoderLayer", "None", "activation_dropout", "activation_fn", "and", "any", "attention_mask", "attn_weights", "bool", "clamp", "clamp_value", "class", "def", "dropout", "dtype", "fc1", "fc2", "final_layer_norm", "finfo", "float16", "forward", "functional", "hidden_states", "if", "isinf", "isnan", "max", "min", "nn", "or", "output_attentions", "outputs", "p", "residual", "return", "self", "self_attn", "self_attn_layer_norm", "torch", "training", "tuple"], "informer/modeling_informer.py:InformerDecoderLayer.__init__": ["ACT2FN", "GradientCheckpointingLayer", "LayerNorm", "Linear", "ModelAttention", "ModelConfig", "ModelDecoderLayer", "ModelProbSparseAttention", "None", "True", "__init__", "activation_dropout", "activation_fn", "activation_function", "attention_dropout", "attention_type", "class", "config", "d_model", "decoder_attention_heads", "decoder_ffn_dim", "def", "dropout", "else", "embed_dim", "encoder_attn", "encoder_attn_layer_norm", "fc1", "fc2", "final_layer_norm", "if", "int", "is_decoder", "layer_idx", "nn", "num_heads", "prob", "sampling_factor", "self", "self_attn", "self_attn_layer_norm", "super"], "informer/modeling_informer.py:InformerDecoderLayer.forward": ["Cache", "False", "FloatTensor", "GradientCheckpointingLayer", "ModelDecoderLayer", "None", "Tensor", "True", "activation_dropout", "activation_fn", "attention_mask", "bool", "cache_position", "class", "cross_attn_weights", "def", "dropout", "encoder_attention_mask", "encoder_attn", "encoder_attn_layer_norm", "encoder_hidden_states", "fc1", "fc2", "final_layer_norm", "forward", "functional", "hidden_states", "if", "is", "key_value_states", "nn", "not", "output_attentions", "outputs", "p", "past_key_values", "residual", "return", "self", "self_attn", "self_attn_layer_norm", "self_attn_weights", "torch", "training", "tuple", "use_cache"], "informer/modeling_informer.py:InformerEncoder.__init__": ["False", "LayerNorm", "ModelConfig", "ModelConvLayer", "ModelEncoder", "ModelEncoderLayer", "ModelPreTrainedModel", "ModelSinusoidalPositionalEmbedding", "ModelValueEmbedding", "ModuleList", "None", "The", "ValueError", "_", "__init__", "append", "be", "class", "config", "context_length", "conv_layers", "d_model", "def", "distil", "dropout", "else", "embed_positions", "encoder_layerdrop", "encoder_layers", "feature_size", "for", "gradient_checkpointing", "if", "in", "is", "layerdrop", "layernorm_embedding", "layers", "needs", "nn", "post_init", "prediction_length", "raise", "range", "self", "specified", "super", "to", "value_embedding"], "informer/modeling_informer.py:InformerEncoder.forward": ["BaseModelOutput", "False", "FloatTensor", "ModelEncoder", "ModelPreTrainedModel", "None", "Tensor", "True", "_prepare_4d_attention_mask", "all_attentions", "attention_mask", "attentions", "bool", "class", "config", "conv_layer", "conv_layers", "def", "dropout", "dropout_probability", "dtype", "else", "embed_pos", "embed_positions", "encoder_layer", "encoder_states", "enumerate", "for", "forward", "functional", "hidden_states", "idx", "if", "in", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layer_outputs", "layerdrop", "layernorm_embedding", "layers", "nn", "not", "output", "output_attentions", "output_hidden_states", "p", "r", "rand", "return", "return_dict", "self", "size", "to_drop", "torch", "training", "tuple", "use_return_dict", "v", "value_embedding", "zip"], "informer/modeling_informer.py:InformerDecoder.__init__": ["False", "LayerNorm", "ModelConfig", "ModelDecoder", "ModelDecoderLayer", "ModelPreTrainedModel", "ModelSinusoidalPositionalEmbedding", "ModelValueEmbedding", "ModuleList", "None", "The", "ValueError", "__init__", "be", "class", "config", "context_length", "d_model", "decoder_layerdrop", "decoder_layers", "def", "dropout", "embed_positions", "feature_size", "for", "gradient_checkpointing", "i", "if", "in", "is", "layer_idx", "layerdrop", "layernorm_embedding", "layers", "needs", "nn", "post_init", "prediction_length", "raise", "range", "self", "specified", "super", "to", "value_embedding"], "informer/modeling_informer.py:InformerDecoder.forward": ["BaseModelOutputWithPastAndCrossAttentions", "Cache", "DynamicCache", "EncoderDecoderCache", "False", "FloatTensor", "LongTensor", "ModelDecoder", "ModelPreTrainedModel", "None", "Setting", "Tensor", "True", "all_cross_attentions", "all_hidden_states", "all_self_attns", "and", "arange", "attention_mask", "attentions", "bool", "cache_position", "checkpointing", "class", "config", "context_length", "continue", "create_bidirectional_mask", "create_causal_mask", "cross_attentions", "decoder_layer", "def", "device", "dropout", "dropout_probability", "else", "embed_pos", "embed_positions", "encoder_attention_mask", "encoder_hidden_states", "enumerate", "for", "forward", "functional", "get_seq_length", "gradient", "gradient_checkpointing", "hidden_states", "idx", "if", "in", "incompatible", "input_embeds", "input_shape", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layer_outputs", "layerdrop", "layernorm_embedding", "layers", "logger", "nn", "not", "output_attentions", "output_hidden_states", "p", "past_key_values", "past_key_values_length", "r", "rand", "return", "return_dict", "self", "size", "torch", "training", "tuple", "use_cache", "use_return_dict", "v", "value_embedding", "warning_once", "with"], "informer/modeling_informer.py:InformerModel.__init__": ["ModelConfig", "ModelDecoder", "ModelEncoder", "ModelFeatureEmbedder", "ModelMeanScaler", "ModelModel", "ModelNOPScaler", "ModelPreTrainedModel", "ModelStdScaler", "True", "__init__", "cardinalities", "cardinality", "class", "config", "decoder", "def", "elif", "else", "embedder", "embedding_dimension", "embedding_dims", "encoder", "if", "is", "mean", "num_static_categorical_features", "or", "post_init", "scaler", "scaling", "self", "std", "super"], "informer/modeling_informer.py:InformerModel._past_length": ["ModelModel", "ModelPreTrainedModel", "_past_length", "class", "config", "context_length", "def", "int", "lags_sequence", "max", "return", "self"], "informer/modeling_informer.py:InformerModel.get_lagged_subsequences": ["ModelModel", "ModelPreTrainedModel", "None", "Tensor", "ValueError", "append", "begin_index", "cannot", "class", "config", "def", "dim", "else", "end_index", "f", "for", "found", "further", "get_lagged_subsequences", "go", "history", "if", "in", "indices", "int", "is", "lag", "lag_index", "lagged_values", "lags", "lags_sequence", "length", "max", "only", "raise", "return", "self", "sequence", "sequence_length", "shape", "shift", "stack", "subsequences_length", "than", "torch", "while"], "informer/modeling_informer.py:InformerModel.create_network_inputs": ["ModelModel", "ModelPreTrainedModel", "None", "Tensor", "ValueError", "_", "_past_length", "abs", "and", "cat", "class", "config", "context", "context_length", "create_network_inputs", "def", "dim", "does", "else", "embedded_cat", "embedder", "expand", "expanded_static_feat", "f", "feature", "features", "future_time_features", "future_values", "get_lagged_subsequences", "if", "input", "inputs", "is", "lagged_sequence", "lags_shape", "length", "lengths", "loc", "log", "log1p", "log_abs_loc", "log_scale", "match", "ndim", "not", "observed_context", "ones_like", "past_observed_mask", "past_time_features", "past_values", "prediction_length", "raise", "reshape", "reshaped_lagged_sequence", "return", "scale", "scaler", "self", "sequence", "shape", "squeeze", "squeezed_loc", "squeezed_scale", "static_categorical_features", "static_feat", "static_real_features", "subsequences_length", "time", "time_feat", "torch", "transformer_inputs", "unsqueeze"], "informer/modeling_informer.py:InformerModel.forward": ["BaseModelOutput", "Cache", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Seq2SeqTSModelOutput", "Tensor", "_", "and", "attention_mask", "attentions", "bool", "bsz", "cache_position", "class", "config", "context_length", "create_network_inputs", "cross_attentions", "dec_input", "decoder", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_outputs", "def", "device", "dim", "dtype", "elif", "else", "enc_input", "encoder", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "forward", "future_time_features", "future_values", "hidden_states", "if", "inputs_embeds", "is", "isinstance", "kwargs", "last_hidden_state", "len", "list", "loc", "not", "output_attentions", "output_hidden_states", "past_key_values", "past_observed_mask", "past_time_features", "past_values", "r", "return", "return_dict", "scale", "self", "shape", "size", "static_categorical_features", "static_feat", "static_features", "static_real_features", "torch", "transformer_inputs", "tuple", "use_cache", "use_return_dict", "zeros"], "informer/modeling_informer.py:weighted_average": ["Model_average", "Model_tensor", "None", "Tensor", "clamp", "def", "dim", "else", "if", "input_tensor", "is", "mean", "min", "not", "return", "sum", "sum_weights", "torch", "weights", "where", "zeros_like"], "informer/modeling_informer.py:nll": ["Distribution", "Model", "Tensor", "def", "distributions", "input", "log_prob", "return", "target", "torch"], "informer/modeling_informer.py:InformerForPrediction.__init__": ["ModelConfig", "ModelForPrediction", "ModelModel", "ModelPreTrainedModel", "NegativeBinomialOutput", "NormalOutput", "StudentTOutput", "Unknown", "ValueError", "__init__", "class", "config", "d_model", "def", "dim", "distribution", "distribution_output", "elif", "else", "event_shape", "f", "function", "get_parameter_projection", "if", "input_size", "loss", "model", "negative_binomial", "nll", "normal", "output", "parameter_projection", "post_init", "raise", "self", "student_t", "super", "target_shape"], "informer/modeling_informer.py:InformerForPrediction.output_params": ["ModelForPrediction", "ModelPreTrainedModel", "class", "dec_output", "def", "output_params", "parameter_projection", "return", "self"], "informer/modeling_informer.py:InformerForPrediction.output_distribution": ["Distribution", "ModelForPrediction", "ModelPreTrainedModel", "None", "class", "def", "distribution", "distribution_output", "distributions", "for", "if", "in", "is", "loc", "not", "output_distribution", "p", "params", "return", "scale", "self", "sliced_params", "torch", "trailing_n"], "informer/modeling_informer.py:InformerForPrediction.forward": ["Cache", "False", "FloatTensor", "LongTensor", "ModelForPrediction", "ModelPreTrainedModel", "None", "Seq2SeqTSModelOutput", "Seq2SeqTSPredictionOutput", "Tensor", "_", "bool", "cache_position", "class", "config", "cross_attentions", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "def", "dim", "distribution", "else", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "forward", "future_observed_mask", "future_time_features", "future_values", "if", "is", "keepdim", "kwargs", "len", "list", "loc", "loss", "loss_weights", "min", "model", "not", "ones_like", "output_attentions", "output_distribution", "output_hidden_states", "output_params", "outputs", "params", "past_key_values", "past_observed_mask", "past_time_features", "past_values", "prediction_loss", "r", "return", "return_dict", "scale", "self", "static_categorical_features", "static_features", "static_real_features", "target_shape", "torch", "tuple", "use_cache", "use_return_dict", "weighted_average", "weights"], "informer/modeling_informer.py:InformerForPrediction.generate": ["ModelForPrediction", "ModelPreTrainedModel", "None", "SampleTSPredictionOutput", "Tensor", "True", "append", "bool", "cat", "class", "concat_future_samples", "config", "dec_last_hidden", "dec_output", "decoder", "decoder_input", "def", "dim", "distr", "enc_last_hidden", "encoder_hidden_states", "encoder_last_hidden_state", "expand", "expanded_static_feat", "features", "for", "future_samples", "future_time_features", "future_values", "generate", "get_decoder", "get_lagged_subsequences", "in", "inputs_embeds", "k", "lagged_sequence", "lags_shape", "last_hidden_state", "loc", "model", "next_sample", "num_parallel_samples", "output_attentions", "output_distribution", "output_hidden_states", "outputs", "parameter_projection", "params", "past_observed_mask", "past_time_features", "past_values", "prediction_length", "r", "range", "repeat_interleave", "repeated_enc_last_hidden", "repeated_features", "repeated_loc", "repeated_past_values", "repeated_scale", "repeats", "reshape", "reshaped_lagged_sequence", "return", "return_dict", "sample", "scale", "self", "sequence", "sequences", "shape", "shift", "static_categorical_features", "static_feat", "static_features", "static_real_features", "subsequences_length", "target_shape", "torch", "unsqueeze", "use_cache"], "instructblip/modeling_instructblip.py:InstructBlipForConditionalGenerationModelOutput.to_tuple": ["Any", "ModelForConditionalGenerationModelOutput", "ModelOutput", "class", "def", "else", "for", "getattr", "if", "in", "k", "keys", "language_model_outputs", "not", "qformer_outputs", "return", "self", "to_tuple", "tuple", "vision_outputs"], "instructblip/modeling_instructblip.py:InstructBlipVisionEmbeddings.__init__": ["Conv2d", "ModelVisionConfig", "ModelVisionEmbeddings", "Module", "Parameter", "__init__", "class", "class_embedding", "config", "def", "embed_dim", "hidden_size", "image_size", "in_channels", "kernel_size", "nn", "num_patches", "num_positions", "out_channels", "patch_embedding", "patch_size", "position_embedding", "randn", "self", "stride", "super", "torch"], "instructblip/modeling_instructblip.py:InstructBlipVisionEmbeddings.interpolate_pos_encoding": ["False", "ModelVisionEmbeddings", "Module", "Tensor", "align_corners", "and", "bicubic", "cat", "class", "class_pos_embed", "def", "dim", "embeddings", "functional", "height", "if", "int", "interpolate", "interpolate_pos_encoding", "is_tracing", "jit", "mode", "new_height", "new_width", "nn", "not", "num_patches", "num_positions", "patch_pos_embed", "patch_size", "permute", "position_embedding", "reshape", "return", "self", "shape", "size", "sqrt_num_positions", "torch", "torch_int", "view", "width"], "instructblip/modeling_instructblip.py:InstructBlipVisionEmbeddings.forward": ["False", "FloatTensor", "ModelVisionEmbeddings", "Module", "Tensor", "_", "batch_size", "bool", "cat", "class", "class_embedding", "class_embeds", "def", "dim", "dtype", "else", "embeddings", "expand", "flatten", "forward", "height", "if", "interpolate_pos_encoding", "nn", "patch_embedding", "patch_embeds", "pixel_values", "position_embedding", "return", "self", "shape", "size", "target_dtype", "to", "torch", "transpose", "weight", "width"], "instructblip/modeling_instructblip.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "attention_mask", "attn_output", "attn_weights", "contiguous", "def", "dim", "dropout", "float", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "softmax", "torch", "training", "transpose", "value"], "instructblip/modeling_instructblip.py:InstructBlipAttention.__init__": ["False", "Linear", "ModelAttention", "Module", "None", "Parameter", "ValueError", "__init__", "and", "attention_dropout", "be", "bias", "by", "cat", "class", "config", "def", "divisible", "else", "embed_dim", "f", "got", "head_dim", "hidden_size", "if", "is", "is_causal", "must", "nn", "not", "num_attention_heads", "num_heads", "projection", "q_bias", "qkv", "qkv_bias", "raise", "requires_grad", "scale", "self", "super", "torch", "v_bias", "zeros", "zeros_like"], "instructblip/modeling_instructblip.py:InstructBlipAttention._shape": ["ModelAttention", "Module", "Tensor", "_shape", "bsz", "class", "contiguous", "def", "head_dim", "int", "nn", "num_heads", "return", "self", "seq_len", "tensor", "torch", "transpose", "view"], "instructblip/modeling_instructblip.py:InstructBlipAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "ModelAttention", "Module", "None", "Tensor", "_attn_implementation", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "bsz", "class", "config", "contiguous", "def", "dropout", "eager", "eager_attention_forward", "else", "embed_dim", "forward", "hidden_states", "if", "key_states", "kwargs", "mixed_qkv", "nn", "not", "num_heads", "permute", "projection", "qkv", "query_states", "reshape", "return", "scale", "scaling", "self", "size", "tgt_len", "torch", "training", "tuple", "value_states"], "instructblip/modeling_instructblip.py:InstructBlipMLP.__init__": ["ACT2FN", "Linear", "ModelMLP", "Module", "__init__", "activation_fn", "class", "config", "def", "fc1", "fc2", "hidden_act", "hidden_size", "intermediate_size", "nn", "self", "super"], "instructblip/modeling_instructblip.py:InstructBlipMLP.forward": ["ModelMLP", "Module", "Tensor", "activation_fn", "class", "def", "fc1", "fc2", "forward", "hidden_states", "nn", "return", "self", "torch"], "instructblip/modeling_instructblip.py:InstructBlipEncoderLayer.__init__": ["GradientCheckpointingLayer", "LayerNorm", "ModelAttention", "ModelConfig", "ModelEncoderLayer", "ModelMLP", "__init__", "class", "config", "def", "embed_dim", "eps", "hidden_size", "layer_norm1", "layer_norm2", "layer_norm_eps", "mlp", "nn", "self", "self_attn", "super"], "instructblip/modeling_instructblip.py:InstructBlipEncoderLayer.forward": ["FloatTensor", "GradientCheckpointingLayer", "ModelEncoderLayer", "Tensor", "TransformersKwargs", "Unpack", "_", "class", "def", "forward", "hidden_states", "kwargs", "layer_norm1", "layer_norm2", "mlp", "residual", "return", "self", "self_attn", "torch"], "instructblip/modeling_instructblip.py:InstructBlipPreTrainedModel._init_weights": ["ModelForConditionalGeneration", "ModelModel", "ModelPreTrainedModel", "ModelQFormerEmbeddings", "ModelVisionEmbeddings", "PreTrainedModel", "_init_weights", "arange", "class", "class_embedding", "config", "copy_", "def", "elif", "expand", "factor", "if", "init", "initializer_range", "isinstance", "mean", "module", "position_embedding", "position_ids", "query_tokens", "self", "shape", "std", "super", "torch", "trunc_normal_", "zeros_"], "instructblip/modeling_instructblip.py:InstructBlipEncoder.__init__": ["False", "ModelConfig", "ModelEncoder", "ModelEncoderLayer", "Module", "ModuleList", "_", "__init__", "class", "config", "def", "for", "gradient_checkpointing", "in", "layers", "nn", "num_hidden_layers", "range", "self", "super"], "instructblip/modeling_instructblip.py:InstructBlipEncoder.forward": ["BaseModelOutput", "ModelEncoder", "Module", "TransformersKwargs", "Unpack", "class", "def", "encoder_layer", "for", "forward", "hidden_states", "in", "inputs_embeds", "kwargs", "last_hidden_state", "layers", "nn", "return", "self", "tuple"], "instructblip/modeling_instructblip.py:InstructBlipVisionModel.__init__": ["LayerNorm", "ModelEncoder", "ModelPreTrainedModel", "ModelVisionConfig", "ModelVisionEmbeddings", "ModelVisionModel", "__init__", "class", "config", "def", "embed_dim", "embeddings", "encoder", "eps", "hidden_size", "layer_norm_eps", "nn", "post_init", "post_layernorm", "self", "super"], "instructblip/modeling_instructblip.py:InstructBlipVisionModel.forward": ["BaseModelOutput", "BaseModelOutputWithPooling", "False", "FloatTensor", "ModelPreTrainedModel", "ModelVisionModel", "None", "TransformersKwargs", "Unpack", "ValueError", "You", "bool", "class", "def", "embeddings", "encoder", "encoder_outputs", "forward", "have", "hidden_states", "if", "inputs_embeds", "interpolate_pos_encoding", "is", "kwargs", "last_hidden_state", "pixel_values", "pooled_output", "pooler_output", "post_layernorm", "raise", "return", "self", "specify", "to", "torch", "tuple"], "instructblip/modeling_instructblip.py:InstructBlipVisionModel.get_input_embeddings": ["ModelPreTrainedModel", "ModelVisionModel", "class", "def", "embeddings", "get_input_embeddings", "return", "self"], "instructblip/modeling_instructblip.py:InstructBlipQFormerMultiHeadAttention.__init__": ["Dropout", "False", "Linear", "ModelQFormerMultiHeadAttention", "Module", "The", "ValueError", "__init__", "a", "all_head_size", "and", "attention", "attention_head_size", "attention_probs_dropout_prob", "class", "config", "d", "def", "dropout", "else", "embedding_size", "encoder_hidden_size", "hasattr", "heads", "hidden", "hidden_size", "if", "int", "is", "is_cross_attention", "key", "multiple", "nn", "not", "num_attention_heads", "number", "of", "query", "raise", "save_attention", "self", "size", "super", "the", "value"], "instructblip/modeling_instructblip.py:InstructBlipQFormerMultiHeadAttention.save_attn_gradients": ["ModelQFormerMultiHeadAttention", "Module", "attn_gradients", "class", "def", "nn", "save_attn_gradients", "self"], "instructblip/modeling_instructblip.py:InstructBlipQFormerMultiHeadAttention.get_attn_gradients": ["ModelQFormerMultiHeadAttention", "Module", "attn_gradients", "class", "def", "get_attn_gradients", "nn", "return", "self"], "instructblip/modeling_instructblip.py:InstructBlipQFormerMultiHeadAttention.save_attention_map": ["ModelQFormerMultiHeadAttention", "Module", "attention_map", "class", "def", "nn", "save_attention_map", "self"], "instructblip/modeling_instructblip.py:InstructBlipQFormerMultiHeadAttention.get_attention_map": ["ModelQFormerMultiHeadAttention", "Module", "attention_map", "class", "def", "get_attention_map", "nn", "return", "self"], "instructblip/modeling_instructblip.py:InstructBlipQFormerMultiHeadAttention.transpose_for_scores": ["ModelQFormerMultiHeadAttention", "Module", "attention_head_size", "class", "def", "new_x_shape", "nn", "num_attention_heads", "permute", "return", "self", "size", "transpose_for_scores", "view", "x"], "instructblip/modeling_instructblip.py:InstructBlipQFormerMultiHeadAttention.forward": ["ModelQFormerMultiHeadAttention", "Module", "None", "Softmax", "TransformersKwargs", "Unpack", "all_head_size", "and", "attention_head_size", "attention_mask", "attention_probs", "attention_probs_dropped", "attention_scores", "attention_scores_dtype", "class", "context_layer", "contiguous", "def", "dim", "dropout", "dtype", "else", "encoder_attention_mask", "encoder_hidden_states", "forward", "hidden_states", "if", "is", "is_cross_attention", "key", "key_layer", "kwargs", "math", "matmul", "mixed_query_layer", "new_context_layer_shape", "nn", "not", "permute", "query", "query_layer", "register_hook", "return", "save_attention", "save_attention_map", "save_attn_gradients", "self", "size", "sqrt", "to", "torch", "transpose", "transpose_for_scores", "value", "value_layer", "view"], "instructblip/modeling_instructblip.py:InstructBlipQFormerSelfOutput.__init__": ["Dropout", "LayerNorm", "Linear", "ModelQFormerSelfOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "eps", "hidden_dropout_prob", "hidden_size", "layer_norm_eps", "nn", "self", "super"], "instructblip/modeling_instructblip.py:InstructBlipQFormerSelfOutput.forward": ["LayerNorm", "ModelQFormerSelfOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "instructblip/modeling_instructblip.py:InstructBlipQFormerAttention.__init__": ["False", "ModelQFormerAttention", "ModelQFormerMultiHeadAttention", "ModelQFormerSelfOutput", "Module", "__init__", "attention", "class", "config", "def", "is_cross_attention", "nn", "output", "self", "super"], "instructblip/modeling_instructblip.py:InstructBlipQFormerAttention.forward": ["FloatTensor", "ModelQFormerAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "attention", "attention_mask", "attention_output", "attn_output", "class", "def", "encoder_attention_mask", "encoder_hidden_states", "forward", "hidden_states", "kwargs", "nn", "output", "return", "self", "torch"], "instructblip/modeling_instructblip.py:InstructBlipQFormerIntermediate.__init__": ["ACT2FN", "Linear", "ModelQFormerIntermediate", "Module", "__init__", "class", "config", "def", "dense", "else", "hidden_act", "hidden_size", "if", "intermediate_act_fn", "intermediate_size", "isinstance", "nn", "self", "str", "super"], "instructblip/modeling_instructblip.py:InstructBlipQFormerIntermediate.forward": ["ModelQFormerIntermediate", "Module", "Tensor", "class", "def", "dense", "forward", "hidden_states", "intermediate_act_fn", "nn", "return", "self", "torch"], "instructblip/modeling_instructblip.py:InstructBlipQFormerOutput.__init__": ["Dropout", "LayerNorm", "Linear", "ModelQFormerOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "eps", "hidden_dropout_prob", "hidden_size", "intermediate_size", "layer_norm_eps", "nn", "self", "super"], "instructblip/modeling_instructblip.py:InstructBlipQFormerOutput.forward": ["LayerNorm", "ModelQFormerOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "instructblip/modeling_instructblip.py:InstructBlipQFormerLayer.__init__": ["False", "GradientCheckpointingLayer", "ModelQFormerAttention", "ModelQFormerIntermediate", "ModelQFormerLayer", "ModelQFormerOutput", "True", "__init__", "attention", "chunk_size_feed_forward", "class", "config", "cross_attention_frequency", "crossattention", "def", "else", "has_cross_attention", "if", "intermediate", "intermediate_query", "is_cross_attention", "layer_idx", "output", "output_query", "self", "seq_len_dim", "super"], "instructblip/modeling_instructblip.py:InstructBlipQFormerLayer.forward": ["GradientCheckpointingLayer", "ModelQFormerLayer", "None", "TransformersKwargs", "Unpack", "ValueError", "apply_chunking_to_forward", "attention", "attention_mask", "attention_output", "be", "cat", "chunk_size_feed_forward", "class", "cross", "crossattention", "def", "device", "dim", "else", "encoder_attention_mask", "encoder_hidden_states", "feed_forward_chunk", "feed_forward_chunk_query", "for", "forward", "given", "has_cross_attention", "hidden_states", "if", "is", "kwargs", "layer_output", "layer_output_text", "layers", "must", "query_attention_output", "query_length", "raise", "return", "self", "seq_len_dim", "shape", "to", "torch"], "instructblip/modeling_instructblip.py:InstructBlipQFormerLayer.feed_forward_chunk": ["GradientCheckpointingLayer", "ModelQFormerLayer", "attention_output", "class", "def", "feed_forward_chunk", "intermediate", "intermediate_output", "layer_output", "output", "return", "self"], "instructblip/modeling_instructblip.py:InstructBlipQFormerLayer.feed_forward_chunk_query": ["GradientCheckpointingLayer", "ModelQFormerLayer", "attention_output", "class", "def", "feed_forward_chunk_query", "intermediate_output", "intermediate_query", "layer_output", "output_query", "return", "self"], "instructblip/modeling_instructblip.py:InstructBlipQFormerEncoder.__init__": ["False", "ModelQFormerEncoder", "ModelQFormerLayer", "Module", "ModuleList", "__init__", "class", "config", "def", "for", "gradient_checkpointing", "in", "layer", "layer_idx", "nn", "num_hidden_layers", "range", "self", "super"], "instructblip/modeling_instructblip.py:InstructBlipQFormerEncoder.forward": ["BaseModelOutputWithPastAndCrossAttentions", "ModelQFormerEncoder", "Module", "None", "TransformersKwargs", "Unpack", "attention_mask", "class", "config", "def", "encoder_attention_mask", "encoder_hidden_states", "for", "forward", "hidden_states", "i", "in", "kwargs", "last_hidden_state", "layer", "layer_module", "nn", "num_hidden_layers", "query_length", "range", "return", "self"], "instructblip/modeling_instructblip.py:InstructBlipQFormerEmbeddings.__init__": ["Dropout", "Embedding", "False", "LayerNorm", "ModelQFormerEmbeddings", "Module", "__init__", "arange", "class", "config", "def", "dropout", "eps", "expand", "hidden_dropout_prob", "hidden_size", "layer_norm_eps", "layernorm", "max_position_embeddings", "nn", "pad_token_id", "padding_idx", "persistent", "position_embeddings", "position_ids", "register_buffer", "self", "super", "torch", "vocab_size", "word_embeddings"], "instructblip/modeling_instructblip.py:InstructBlipQFormerEmbeddings.forward": ["ModelQFormerEmbeddings", "Module", "None", "cat", "class", "clone", "def", "device", "dim", "dropout", "dtype", "else", "embeddings", "forward", "if", "input_ids", "is", "layernorm", "nn", "not", "past_key_values_length", "position_embeddings", "position_ids", "query_embeds", "return", "self", "seq_length", "size", "to", "torch", "weight", "word_embeddings"], "instructblip/modeling_instructblip.py:InstructBlipQFormerModel.__init__": ["ModelPreTrainedModel", "ModelQFormerConfig", "ModelQFormerEmbeddings", "ModelQFormerEncoder", "ModelQFormerModel", "__init__", "class", "config", "def", "embeddings", "encoder", "post_init", "self", "super"], "instructblip/modeling_instructblip.py:InstructBlipQFormerModel.get_input_embeddings": ["ModelPreTrainedModel", "ModelQFormerModel", "class", "def", "embeddings", "get_input_embeddings", "return", "self", "word_embeddings"], "instructblip/modeling_instructblip.py:InstructBlipQFormerModel.set_input_embeddings": ["ModelPreTrainedModel", "ModelQFormerModel", "class", "def", "embeddings", "self", "set_input_embeddings", "value", "word_embeddings"], "instructblip/modeling_instructblip.py:InstructBlipQFormerModel.get_extended_attention_mask": ["False", "ModelPreTrainedModel", "ModelQFormerModel", "None", "Tensor", "ValueError", "Wrong", "attention_mask", "bool", "class", "def", "device", "dim", "dtype", "elif", "else", "extended_attention_mask", "f", "for", "get_extended_attention_mask", "has_query", "if", "input_ids", "input_shape", "int", "or", "raise", "return", "self", "shape", "to", "torch", "tuple"], "instructblip/modeling_instructblip.py:InstructBlipQFormerModel.forward": ["BaseModelOutput", "BaseModelOutputWithPoolingAndCrossAttentions", "FloatTensor", "LongTensor", "ModelPreTrainedModel", "ModelQFormerModel", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "_", "and", "attention_mask", "batch_size", "class", "def", "device", "elif", "else", "embedding_output", "embeddings", "encoder", "encoder_attention_mask", "encoder_batch_size", "encoder_extended_attention_mask", "encoder_hidden_shape", "encoder_hidden_states", "encoder_outputs", "encoder_sequence_length", "extended_attention_mask", "for", "forward", "get_extended_attention_mask", "have", "if", "in", "input_ids", "input_shape", "invert_attention_mask", "is", "isinstance", "kwargs", "last_hidden_state", "list", "mask", "not", "ones", "pooled_output", "pooler_output", "position_ids", "query_embeds", "query_length", "r", "raise", "return", "self", "seq_length", "sequence_output", "shape", "size", "specify", "to", "torch", "tuple", "when"], "instructblip/modeling_instructblip.py:InstructBlipModel.__init__": ["AutoModel", "Linear", "ModelConfig", "ModelModel", "ModelPreTrainedModel", "ModelQFormerModel", "ModelVisionModel", "None", "Parameter", "__init__", "_keep_in_fp32_modules", "_no_split_modules", "class", "config", "def", "extend", "from_config", "hidden_size", "if", "is", "language_model", "language_projection", "nn", "not", "num_query_tokens", "post_init", "qformer", "qformer_config", "query_tokens", "self", "super", "text_config", "torch", "vision_config", "vision_model", "zeros"], "instructblip/modeling_instructblip.py:InstructBlipModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "language_model", "return", "self"], "instructblip/modeling_instructblip.py:InstructBlipModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "language_model", "self", "set_input_embeddings", "value"], "instructblip/modeling_instructblip.py:InstructBlipModel._preprocess_accelerate": ["GPU", "ModelModel", "ModelPreTrainedModel", "Please", "The", "True", "_hf_hook", "_preprocess_accelerate", "a", "accelerate", "and", "are", "behavior", "blob", "blog", "class", "com", "contains", "creating", "cuda", "def", "details", "device_count", "device_map", "dictionary", "environment", "for", "github", "hasattr", "hf_device_map", "https", "huggingface", "if", "in", "io_same_device", "is", "language_model", "large", "lead", "len", "logger", "main", "may", "md", "models", "more", "multi", "not", "on", "pass", "r", "refer", "remove", "running", "script", "self", "that", "the", "this", "to", "torch", "unexpected", "using", "warning", "when", "you", "your"], "instructblip/modeling_instructblip.py:InstructBlipModel.get_placeholder_mask": ["FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "all", "class", "config", "def", "device", "dtype", "else", "expand_as", "get_input_embeddings", "get_placeholder_mask", "if", "image_token_id", "input_ids", "inputs_embeds", "is", "long", "return", "self", "special_image_mask", "tensor", "to", "torch", "unsqueeze"], "instructblip/modeling_instructblip.py:InstructBlipModel.forward": ["False", "FlashAttentionKwargs", "FloatTensor", "LongTensor", "ModelForConditionalGenerationModelOutput", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "Unpack", "attention_mask", "bool", "cat", "class", "config", "decoder_attention_mask", "decoder_input_ids", "def", "device", "dim", "dtype", "else", "encoder_attention_mask", "encoder_hidden_states", "expand", "forward", "get_input_embeddings", "get_placeholder_mask", "if", "image_attention_mask", "image_embeds", "input_ids", "inputs_embeds", "interpolate_pos_encoding", "is", "kwargs", "language_model", "language_model_inputs", "language_model_outputs", "language_projection", "long", "masked_scatter", "ones", "ones_like", "outputs", "pixel_values", "qformer", "qformer_attention_mask", "qformer_input_ids", "qformer_outputs", "query_attention_mask", "query_embeds", "query_output", "query_outputs", "query_tokens", "r", "return", "self", "shape", "size", "special_image_mask", "to", "torch", "tuple", "use_decoder_only_language_model", "vision_model", "vision_outputs"], "instructblip/modeling_instructblip.py:InstructBlipForConditionalGeneration.__init__": ["AutoModelForCausalLM", "AutoModelForSeq2SeqLM", "GenerationMixin", "Linear", "ModelConfig", "ModelForConditionalGeneration", "ModelPreTrainedModel", "ModelQFormerModel", "ModelVisionModel", "None", "Parameter", "__init__", "_from_config", "_keep_in_fp32_modules", "_no_split_modules", "class", "config", "def", "else", "extend", "from_config", "hidden_size", "if", "is", "language_model", "language_projection", "nn", "not", "num_query_tokens", "post_init", "qformer", "qformer_config", "query_tokens", "self", "super", "text_config", "torch", "use_decoder_only_language_model", "vision_config", "vision_model", "zeros"], "instructblip/modeling_instructblip.py:InstructBlipForConditionalGeneration.get_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "language_model", "return", "self"], "instructblip/modeling_instructblip.py:InstructBlipForConditionalGeneration.set_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "language_model", "self", "set_input_embeddings", "value"], "instructblip/modeling_instructblip.py:InstructBlipForConditionalGeneration.set_output_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "language_model", "new_embeddings", "self", "set_output_embeddings"], "instructblip/modeling_instructblip.py:InstructBlipForConditionalGeneration.get_output_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "Module", "class", "def", "get_output_embeddings", "language_model", "nn", "return", "self"], "instructblip/modeling_instructblip.py:InstructBlipForConditionalGeneration.get_encoder": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "class", "def", "else", "get_encoder", "if", "is", "language_model", "modality", "return", "self", "super"], "instructblip/modeling_instructblip.py:InstructBlipForConditionalGeneration.get_decoder": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "get_decoder", "language_model", "return", "self"], "instructblip/modeling_instructblip.py:InstructBlipForConditionalGeneration._preprocess_accelerate": ["GPU", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "Please", "The", "True", "_hf_hook", "_preprocess_accelerate", "a", "accelerate", "and", "are", "behavior", "blob", "blog", "class", "com", "contains", "creating", "cuda", "def", "details", "device_count", "device_map", "dictionary", "environment", "for", "github", "hasattr", "hf_device_map", "https", "huggingface", "if", "in", "io_same_device", "is", "language_model", "large", "lead", "len", "logger", "main", "may", "md", "models", "more", "multi", "not", "on", "pass", "r", "refer", "remove", "running", "script", "self", "that", "the", "this", "to", "torch", "unexpected", "using", "warning", "when", "you", "your"], "instructblip/modeling_instructblip.py:InstructBlipForConditionalGeneration.get_image_features": ["False", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "True", "attention_mask", "bool", "cat", "class", "def", "device", "dim", "dtype", "encoder_attention_mask", "encoder_hidden_states", "expand", "get_image_features", "if", "image_attention_mask", "image_embeds", "input_ids", "interpolate_pos_encoding", "is", "language_model_inputs", "language_projection", "long", "ones", "ones_like", "pixel_values", "qformer", "qformer_attention_mask", "qformer_input_ids", "query_attention_mask", "query_embeds", "query_output", "query_outputs", "query_tokens", "return", "return_dict", "self", "shape", "size", "torch", "vision_model", "vision_outputs"], "instructblip/modeling_instructblip.py:InstructBlipForConditionalGeneration.get_placeholder_mask": ["FloatTensor", "GenerationMixin", "LongTensor", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "all", "class", "config", "def", "device", "dtype", "else", "expand_as", "get_input_embeddings", "get_placeholder_mask", "if", "image_token_id", "input_ids", "inputs_embeds", "is", "long", "return", "self", "special_image_mask", "tensor", "to", "torch", "unsqueeze"], "instructblip/modeling_instructblip.py:InstructBlipForConditionalGeneration.forward": ["False", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForConditionalGeneration", "ModelForConditionalGenerationModelOutput", "ModelPreTrainedModel", "None", "TransformersKwargs", "True", "Unpack", "attention_mask", "bool", "class", "config", "decoder_attention_mask", "decoder_input_ids", "def", "device", "dtype", "else", "forward", "get_image_features", "get_input_embeddings", "get_placeholder_mask", "if", "input_ids", "inputs_embeds", "interpolate_pos_encoding", "is", "kwargs", "labels", "language_model", "language_model_inputs", "language_model_outputs", "logits", "loss", "loss_function", "masked_scatter", "not", "ones_like", "outputs", "pixel_values", "qformer_attention_mask", "qformer_input_ids", "qformer_outputs", "query_outputs", "r", "return", "return_dict", "self", "special_image_mask", "text_config", "to", "torch", "tuple", "use_decoder_only_language_model", "vision_outputs", "vocab_size"], "instructblip/modeling_instructblip.py:InstructBlipForConditionalGeneration.generate": ["False", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "True", "_preprocess_accelerate", "attention_mask", "batch_size", "bool", "bos_token_id", "class", "config", "def", "device", "dtype", "generate", "generate_kwargs", "get_image_features", "get_input_embeddings", "get_placeholder_mask", "hasattr", "hf_device_map", "if", "image_token_index", "image_tokens", "input_ids", "inputs", "inputs_embeds", "interpolate_pos_encoding", "is", "is_encoder_decoder", "language_model", "language_model_inputs", "long", "masked_scatter", "not", "num_query_tokens", "ones_like", "outputs", "pixel_values", "qformer_attention_mask", "qformer_input_ids", "query_outputs", "repeat", "return", "return_dict", "self", "shape", "special_image_mask", "start_tokens", "tensor", "text_config", "to", "torch", "vision_outputs"], "instructblipvideo/modeling_instructblipvideo.py:InstructBlipVideoVisionEmbeddings.__init__": ["Conv2d", "ModelVisionConfig", "ModelVisionEmbeddings", "Module", "Parameter", "__init__", "class", "class_embedding", "config", "def", "embed_dim", "hidden_size", "image_size", "in_channels", "kernel_size", "nn", "num_patches", "num_positions", "out_channels", "patch_embedding", "patch_size", "position_embedding", "randn", "self", "stride", "super", "torch"], "instructblipvideo/modeling_instructblipvideo.py:InstructBlipVideoVisionEmbeddings.interpolate_pos_encoding": ["False", "ModelVisionEmbeddings", "Module", "Tensor", "align_corners", "and", "bicubic", "cat", "class", "class_pos_embed", "def", "dim", "embeddings", "functional", "height", "if", "int", "interpolate", "interpolate_pos_encoding", "is_tracing", "jit", "mode", "new_height", "new_width", "nn", "not", "num_patches", "num_positions", "patch_pos_embed", "patch_size", "permute", "position_embedding", "reshape", "return", "self", "shape", "size", "sqrt_num_positions", "torch", "torch_int", "view", "width"], "instructblipvideo/modeling_instructblipvideo.py:InstructBlipVideoVisionEmbeddings.forward": ["False", "FloatTensor", "ModelVisionEmbeddings", "Module", "Tensor", "_", "batch_size", "bool", "cat", "class", "class_embedding", "class_embeds", "def", "dim", "dtype", "else", "embeddings", "expand", "flatten", "forward", "height", "if", "interpolate_pos_encoding", "nn", "patch_embedding", "patch_embeds", "pixel_values", "position_embedding", "return", "self", "shape", "size", "target_dtype", "to", "torch", "transpose", "weight", "width"], "instructblipvideo/modeling_instructblipvideo.py:InstructBlipVideoQFormerEmbeddings.__init__": ["Dropout", "Embedding", "False", "LayerNorm", "ModelQFormerEmbeddings", "Module", "__init__", "arange", "class", "config", "def", "dropout", "eps", "expand", "hidden_dropout_prob", "hidden_size", "layer_norm_eps", "layernorm", "max_position_embeddings", "nn", "pad_token_id", "padding_idx", "persistent", "position_embeddings", "position_ids", "register_buffer", "self", "super", "torch", "vocab_size", "word_embeddings"], "instructblipvideo/modeling_instructblipvideo.py:InstructBlipVideoQFormerEmbeddings.forward": ["ModelQFormerEmbeddings", "Module", "None", "cat", "class", "clone", "def", "device", "dim", "dropout", "dtype", "else", "embeddings", "forward", "if", "input_ids", "is", "layernorm", "nn", "not", "past_key_values_length", "position_embeddings", "position_ids", "query_embeds", "return", "self", "seq_length", "size", "to", "torch", "weight", "word_embeddings"], "instructblipvideo/modeling_instructblipvideo.py:InstructBlipVideoPreTrainedModel._init_weights": ["ModelForConditionalGeneration", "ModelModel", "ModelPreTrainedModel", "ModelQFormerEmbeddings", "ModelVisionEmbeddings", "PreTrainedModel", "_init_weights", "arange", "class", "class_embedding", "config", "copy_", "def", "elif", "expand", "factor", "if", "init", "initializer_range", "isinstance", "mean", "module", "position_embedding", "position_ids", "query_tokens", "self", "shape", "std", "super", "torch", "trunc_normal_", "zeros_"], "instructblipvideo/modeling_instructblipvideo.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "attention_mask", "attn_output", "attn_weights", "contiguous", "def", "dim", "dropout", "float", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "softmax", "torch", "training", "transpose", "value"], "instructblipvideo/modeling_instructblipvideo.py:InstructBlipVideoAttention.__init__": ["False", "Linear", "ModelAttention", "Module", "None", "Parameter", "ValueError", "__init__", "and", "attention_dropout", "be", "bias", "by", "cat", "class", "config", "def", "divisible", "else", "embed_dim", "f", "got", "head_dim", "hidden_size", "if", "is", "is_causal", "must", "nn", "not", "num_attention_heads", "num_heads", "projection", "q_bias", "qkv", "qkv_bias", "raise", "requires_grad", "scale", "self", "super", "torch", "v_bias", "zeros", "zeros_like"], "instructblipvideo/modeling_instructblipvideo.py:InstructBlipVideoAttention._shape": ["ModelAttention", "Module", "Tensor", "_shape", "bsz", "class", "contiguous", "def", "head_dim", "int", "nn", "num_heads", "return", "self", "seq_len", "tensor", "torch", "transpose", "view"], "instructblipvideo/modeling_instructblipvideo.py:InstructBlipVideoAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "ModelAttention", "Module", "None", "Tensor", "_attn_implementation", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "bsz", "class", "config", "contiguous", "def", "dropout", "eager", "eager_attention_forward", "else", "embed_dim", "forward", "hidden_states", "if", "key_states", "kwargs", "mixed_qkv", "nn", "not", "num_heads", "permute", "projection", "qkv", "query_states", "reshape", "return", "scale", "scaling", "self", "size", "tgt_len", "torch", "training", "tuple", "value_states"], "instructblipvideo/modeling_instructblipvideo.py:InstructBlipVideoMLP.__init__": ["ACT2FN", "Linear", "ModelMLP", "Module", "__init__", "activation_fn", "class", "config", "def", "fc1", "fc2", "hidden_act", "hidden_size", "intermediate_size", "nn", "self", "super"], "instructblipvideo/modeling_instructblipvideo.py:InstructBlipVideoMLP.forward": ["ModelMLP", "Module", "Tensor", "activation_fn", "class", "def", "fc1", "fc2", "forward", "hidden_states", "nn", "return", "self", "torch"], "instructblipvideo/modeling_instructblipvideo.py:InstructBlipVideoEncoderLayer.__init__": ["GradientCheckpointingLayer", "LayerNorm", "ModelAttention", "ModelConfig", "ModelEncoderLayer", "ModelMLP", "__init__", "class", "config", "def", "embed_dim", "eps", "hidden_size", "layer_norm1", "layer_norm2", "layer_norm_eps", "mlp", "nn", "self", "self_attn", "super"], "instructblipvideo/modeling_instructblipvideo.py:InstructBlipVideoEncoderLayer.forward": ["FloatTensor", "GradientCheckpointingLayer", "ModelEncoderLayer", "Tensor", "TransformersKwargs", "Unpack", "_", "class", "def", "forward", "hidden_states", "kwargs", "layer_norm1", "layer_norm2", "mlp", "residual", "return", "self", "self_attn", "torch"], "instructblipvideo/modeling_instructblipvideo.py:InstructBlipVideoEncoder.__init__": ["False", "ModelConfig", "ModelEncoder", "ModelEncoderLayer", "Module", "ModuleList", "_", "__init__", "class", "config", "def", "for", "gradient_checkpointing", "in", "layers", "nn", "num_hidden_layers", "range", "self", "super"], "instructblipvideo/modeling_instructblipvideo.py:InstructBlipVideoEncoder.forward": ["BaseModelOutput", "ModelEncoder", "Module", "TransformersKwargs", "Unpack", "class", "def", "encoder_layer", "for", "forward", "hidden_states", "in", "inputs_embeds", "kwargs", "last_hidden_state", "layers", "nn", "return", "self", "tuple"], "instructblipvideo/modeling_instructblipvideo.py:InstructBlipVideoVisionModel.__init__": ["LayerNorm", "ModelEncoder", "ModelPreTrainedModel", "ModelVisionConfig", "ModelVisionEmbeddings", "ModelVisionModel", "__init__", "class", "config", "def", "embed_dim", "embeddings", "encoder", "eps", "hidden_size", "layer_norm_eps", "nn", "post_init", "post_layernorm", "self", "super"], "instructblipvideo/modeling_instructblipvideo.py:InstructBlipVideoVisionModel.forward": ["BaseModelOutput", "BaseModelOutputWithPooling", "False", "FloatTensor", "ModelPreTrainedModel", "ModelVisionModel", "None", "TransformersKwargs", "Unpack", "ValueError", "You", "bool", "class", "def", "embeddings", "encoder", "encoder_outputs", "forward", "have", "hidden_states", "if", "inputs_embeds", "interpolate_pos_encoding", "is", "kwargs", "last_hidden_state", "pixel_values", "pooled_output", "pooler_output", "post_layernorm", "raise", "return", "self", "specify", "to", "torch", "tuple"], "instructblipvideo/modeling_instructblipvideo.py:InstructBlipVideoVisionModel.get_input_embeddings": ["ModelPreTrainedModel", "ModelVisionModel", "class", "def", "embeddings", "get_input_embeddings", "return", "self"], "instructblipvideo/modeling_instructblipvideo.py:InstructBlipVideoQFormerMultiHeadAttention.__init__": ["Dropout", "False", "Linear", "ModelQFormerMultiHeadAttention", "Module", "The", "ValueError", "__init__", "a", "all_head_size", "and", "attention", "attention_head_size", "attention_probs_dropout_prob", "class", "config", "d", "def", "dropout", "else", "embedding_size", "encoder_hidden_size", "hasattr", "heads", "hidden", "hidden_size", "if", "int", "is", "is_cross_attention", "key", "multiple", "nn", "not", "num_attention_heads", "number", "of", "query", "raise", "save_attention", "self", "size", "super", "the", "value"], "instructblipvideo/modeling_instructblipvideo.py:InstructBlipVideoQFormerMultiHeadAttention.save_attn_gradients": ["ModelQFormerMultiHeadAttention", "Module", "attn_gradients", "class", "def", "nn", "save_attn_gradients", "self"], "instructblipvideo/modeling_instructblipvideo.py:InstructBlipVideoQFormerMultiHeadAttention.get_attn_gradients": ["ModelQFormerMultiHeadAttention", "Module", "attn_gradients", "class", "def", "get_attn_gradients", "nn", "return", "self"], "instructblipvideo/modeling_instructblipvideo.py:InstructBlipVideoQFormerMultiHeadAttention.save_attention_map": ["ModelQFormerMultiHeadAttention", "Module", "attention_map", "class", "def", "nn", "save_attention_map", "self"], "instructblipvideo/modeling_instructblipvideo.py:InstructBlipVideoQFormerMultiHeadAttention.get_attention_map": ["ModelQFormerMultiHeadAttention", "Module", "attention_map", "class", "def", "get_attention_map", "nn", "return", "self"], "instructblipvideo/modeling_instructblipvideo.py:InstructBlipVideoQFormerMultiHeadAttention.transpose_for_scores": ["ModelQFormerMultiHeadAttention", "Module", "attention_head_size", "class", "def", "new_x_shape", "nn", "num_attention_heads", "permute", "return", "self", "size", "transpose_for_scores", "view", "x"], "instructblipvideo/modeling_instructblipvideo.py:InstructBlipVideoQFormerMultiHeadAttention.forward": ["ModelQFormerMultiHeadAttention", "Module", "None", "Softmax", "TransformersKwargs", "Unpack", "all_head_size", "and", "attention_head_size", "attention_mask", "attention_probs", "attention_probs_dropped", "attention_scores", "attention_scores_dtype", "class", "context_layer", "contiguous", "def", "dim", "dropout", "dtype", "else", "encoder_attention_mask", "encoder_hidden_states", "forward", "hidden_states", "if", "is", "is_cross_attention", "key", "key_layer", "kwargs", "math", "matmul", "mixed_query_layer", "new_context_layer_shape", "nn", "not", "permute", "query", "query_layer", "register_hook", "return", "save_attention", "save_attention_map", "save_attn_gradients", "self", "size", "sqrt", "to", "torch", "transpose", "transpose_for_scores", "value", "value_layer", "view"], "instructblipvideo/modeling_instructblipvideo.py:InstructBlipVideoQFormerSelfOutput.__init__": ["Dropout", "LayerNorm", "Linear", "ModelQFormerSelfOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "eps", "hidden_dropout_prob", "hidden_size", "layer_norm_eps", "nn", "self", "super"], "instructblipvideo/modeling_instructblipvideo.py:InstructBlipVideoQFormerSelfOutput.forward": ["LayerNorm", "ModelQFormerSelfOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "instructblipvideo/modeling_instructblipvideo.py:InstructBlipVideoQFormerAttention.__init__": ["False", "ModelQFormerAttention", "ModelQFormerMultiHeadAttention", "ModelQFormerSelfOutput", "Module", "__init__", "attention", "class", "config", "def", "is_cross_attention", "nn", "output", "self", "super"], "instructblipvideo/modeling_instructblipvideo.py:InstructBlipVideoQFormerAttention.forward": ["FloatTensor", "ModelQFormerAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "attention", "attention_mask", "attention_output", "attn_output", "class", "def", "encoder_attention_mask", "encoder_hidden_states", "forward", "hidden_states", "kwargs", "nn", "output", "return", "self", "torch"], "instructblipvideo/modeling_instructblipvideo.py:InstructBlipVideoQFormerIntermediate.__init__": ["ACT2FN", "Linear", "ModelQFormerIntermediate", "Module", "__init__", "class", "config", "def", "dense", "else", "hidden_act", "hidden_size", "if", "intermediate_act_fn", "intermediate_size", "isinstance", "nn", "self", "str", "super"], "instructblipvideo/modeling_instructblipvideo.py:InstructBlipVideoQFormerIntermediate.forward": ["ModelQFormerIntermediate", "Module", "Tensor", "class", "def", "dense", "forward", "hidden_states", "intermediate_act_fn", "nn", "return", "self", "torch"], "instructblipvideo/modeling_instructblipvideo.py:InstructBlipVideoQFormerOutput.__init__": ["Dropout", "LayerNorm", "Linear", "ModelQFormerOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "eps", "hidden_dropout_prob", "hidden_size", "intermediate_size", "layer_norm_eps", "nn", "self", "super"], "instructblipvideo/modeling_instructblipvideo.py:InstructBlipVideoQFormerOutput.forward": ["LayerNorm", "ModelQFormerOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "instructblipvideo/modeling_instructblipvideo.py:InstructBlipVideoQFormerLayer.__init__": ["False", "GradientCheckpointingLayer", "ModelQFormerAttention", "ModelQFormerIntermediate", "ModelQFormerLayer", "ModelQFormerOutput", "True", "__init__", "attention", "chunk_size_feed_forward", "class", "config", "cross_attention_frequency", "crossattention", "def", "else", "has_cross_attention", "if", "intermediate", "intermediate_query", "is_cross_attention", "layer_idx", "output", "output_query", "self", "seq_len_dim", "super"], "instructblipvideo/modeling_instructblipvideo.py:InstructBlipVideoQFormerLayer.forward": ["GradientCheckpointingLayer", "ModelQFormerLayer", "None", "TransformersKwargs", "Unpack", "ValueError", "apply_chunking_to_forward", "attention", "attention_mask", "attention_output", "be", "cat", "chunk_size_feed_forward", "class", "cross", "crossattention", "def", "device", "dim", "else", "encoder_attention_mask", "encoder_hidden_states", "feed_forward_chunk", "feed_forward_chunk_query", "for", "forward", "given", "has_cross_attention", "hidden_states", "if", "is", "kwargs", "layer_output", "layer_output_text", "layers", "must", "query_attention_output", "query_length", "raise", "return", "self", "seq_len_dim", "shape", "to", "torch"], "instructblipvideo/modeling_instructblipvideo.py:InstructBlipVideoQFormerLayer.feed_forward_chunk": ["GradientCheckpointingLayer", "ModelQFormerLayer", "attention_output", "class", "def", "feed_forward_chunk", "intermediate", "intermediate_output", "layer_output", "output", "return", "self"], "instructblipvideo/modeling_instructblipvideo.py:InstructBlipVideoQFormerLayer.feed_forward_chunk_query": ["GradientCheckpointingLayer", "ModelQFormerLayer", "attention_output", "class", "def", "feed_forward_chunk_query", "intermediate_output", "intermediate_query", "layer_output", "output_query", "return", "self"], "instructblipvideo/modeling_instructblipvideo.py:InstructBlipVideoQFormerEncoder.__init__": ["False", "ModelQFormerEncoder", "ModelQFormerLayer", "Module", "ModuleList", "__init__", "class", "config", "def", "for", "gradient_checkpointing", "in", "layer", "layer_idx", "nn", "num_hidden_layers", "range", "self", "super"], "instructblipvideo/modeling_instructblipvideo.py:InstructBlipVideoQFormerEncoder.forward": ["BaseModelOutputWithPastAndCrossAttentions", "ModelQFormerEncoder", "Module", "None", "TransformersKwargs", "Unpack", "attention_mask", "class", "config", "def", "encoder_attention_mask", "encoder_hidden_states", "for", "forward", "hidden_states", "i", "in", "kwargs", "last_hidden_state", "layer", "layer_module", "nn", "num_hidden_layers", "query_length", "range", "return", "self"], "instructblipvideo/modeling_instructblipvideo.py:InstructBlipVideoQFormerModel.__init__": ["ModelPreTrainedModel", "ModelQFormerConfig", "ModelQFormerEmbeddings", "ModelQFormerEncoder", "ModelQFormerModel", "__init__", "class", "config", "def", "embeddings", "encoder", "post_init", "self", "super"], "instructblipvideo/modeling_instructblipvideo.py:InstructBlipVideoQFormerModel.get_input_embeddings": ["ModelPreTrainedModel", "ModelQFormerModel", "class", "def", "embeddings", "get_input_embeddings", "return", "self", "word_embeddings"], "instructblipvideo/modeling_instructblipvideo.py:InstructBlipVideoQFormerModel.set_input_embeddings": ["ModelPreTrainedModel", "ModelQFormerModel", "class", "def", "embeddings", "self", "set_input_embeddings", "value", "word_embeddings"], "instructblipvideo/modeling_instructblipvideo.py:InstructBlipVideoQFormerModel.get_extended_attention_mask": ["False", "ModelPreTrainedModel", "ModelQFormerModel", "None", "Tensor", "ValueError", "Wrong", "attention_mask", "bool", "class", "def", "device", "dim", "dtype", "elif", "else", "extended_attention_mask", "f", "for", "get_extended_attention_mask", "has_query", "if", "input_ids", "input_shape", "int", "or", "raise", "return", "self", "shape", "to", "torch", "tuple"], "instructblipvideo/modeling_instructblipvideo.py:InstructBlipVideoQFormerModel.forward": ["BaseModelOutput", "BaseModelOutputWithPoolingAndCrossAttentions", "FloatTensor", "LongTensor", "ModelPreTrainedModel", "ModelQFormerModel", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "_", "and", "attention_mask", "batch_size", "class", "def", "device", "elif", "else", "embedding_output", "embeddings", "encoder", "encoder_attention_mask", "encoder_batch_size", "encoder_extended_attention_mask", "encoder_hidden_shape", "encoder_hidden_states", "encoder_outputs", "encoder_sequence_length", "extended_attention_mask", "for", "forward", "get_extended_attention_mask", "have", "if", "in", "input_ids", "input_shape", "invert_attention_mask", "is", "isinstance", "kwargs", "last_hidden_state", "list", "mask", "not", "ones", "pooled_output", "pooler_output", "position_ids", "query_embeds", "query_length", "r", "raise", "return", "self", "seq_length", "sequence_output", "shape", "size", "specify", "to", "torch", "tuple", "when"], "instructblipvideo/modeling_instructblipvideo.py:InstructBlipVideoForConditionalGenerationModelOutput.to_tuple": ["Any", "ModelForConditionalGenerationModelOutput", "ModelOutput", "class", "def", "else", "for", "getattr", "if", "in", "k", "keys", "language_model_outputs", "not", "qformer_outputs", "return", "self", "to_tuple", "tuple", "vision_outputs"], "instructblipvideo/modeling_instructblipvideo.py:InstructBlipVideoModel.__init__": ["AutoModel", "Linear", "ModelConfig", "ModelModel", "ModelPreTrainedModel", "ModelQFormerModel", "ModelVisionModel", "None", "Parameter", "__init__", "_keep_in_fp32_modules", "_no_split_modules", "class", "config", "def", "extend", "from_config", "hidden_size", "if", "is", "language_model", "language_projection", "nn", "not", "num_query_tokens", "post_init", "qformer", "qformer_config", "query_tokens", "self", "super", "text_config", "torch", "vision_config", "vision_model", "zeros"], "instructblipvideo/modeling_instructblipvideo.py:InstructBlipVideoModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "language_model", "return", "self"], "instructblipvideo/modeling_instructblipvideo.py:InstructBlipVideoModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "language_model", "self", "set_input_embeddings", "value"], "instructblipvideo/modeling_instructblipvideo.py:InstructBlipVideoModel._preprocess_accelerate": ["GPU", "ModelModel", "ModelPreTrainedModel", "Please", "The", "True", "_hf_hook", "_preprocess_accelerate", "a", "accelerate", "and", "are", "behavior", "blob", "blog", "class", "com", "contains", "creating", "cuda", "def", "details", "device_count", "device_map", "dictionary", "environment", "for", "github", "hasattr", "hf_device_map", "https", "huggingface", "if", "in", "io_same_device", "is", "language_model", "large", "lead", "len", "logger", "main", "may", "md", "models", "more", "multi", "not", "on", "pass", "r", "refer", "remove", "running", "script", "self", "that", "the", "this", "to", "torch", "unexpected", "using", "warning", "when", "you", "your"], "instructblipvideo/modeling_instructblipvideo.py:InstructBlipVideoModel.get_placeholder_mask": ["FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "all", "class", "config", "def", "device", "dtype", "else", "expand_as", "get_input_embeddings", "get_placeholder_mask", "if", "image_token_id", "input_ids", "inputs_embeds", "is", "long", "return", "self", "special_image_mask", "tensor", "to", "torch", "unsqueeze"], "instructblipvideo/modeling_instructblipvideo.py:InstructBlipVideoModel.forward": ["False", "FlashAttentionKwargs", "FloatTensor", "LongTensor", "ModelForConditionalGenerationModelOutput", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "Unpack", "all", "attention_mask", "batch_size", "bool", "cat", "channel", "class", "config", "decoder_attention_mask", "decoder_input_ids", "def", "device", "dim", "dtype", "else", "encoder_attention_mask", "encoder_hidden_states", "expand", "expand_as", "forward", "frames", "get_input_embeddings", "height", "if", "image_attention_mask", "image_embeds", "input_ids", "inputs_embeds", "interpolate_pos_encoding", "is", "kwargs", "language_model", "language_model_inputs", "language_model_outputs", "language_projection", "long", "masked_scatter", "not", "num_query_tokens", "ones", "ones_like", "output_attentions", "output_hidden_states", "outputs", "pixel_values", "qformer", "qformer_attention_mask", "qformer_input_ids", "qformer_outputs", "query_attention_mask", "query_embeds", "query_output", "query_outputs", "query_tokens", "r", "repeat_interleave", "reshape", "return", "return_dict", "self", "shape", "size", "special_image_mask", "tensor", "to", "torch", "tuple", "unsqueeze", "use_cache", "use_decoder_only_language_model", "use_return_dict", "video_token_id", "vision_model", "vision_outputs", "width"], "instructblipvideo/modeling_instructblipvideo.py:InstructBlipVideoForConditionalGeneration.__init__": ["AutoModelForCausalLM", "AutoModelForSeq2SeqLM", "GenerationMixin", "Linear", "ModelConfig", "ModelForConditionalGeneration", "ModelPreTrainedModel", "ModelQFormerModel", "ModelVisionModel", "None", "Parameter", "__init__", "_from_config", "_keep_in_fp32_modules", "_no_split_modules", "class", "config", "def", "else", "extend", "from_config", "hidden_size", "if", "is", "language_model", "language_projection", "nn", "not", "num_query_tokens", "post_init", "qformer", "qformer_config", "query_tokens", "self", "super", "text_config", "torch", "use_decoder_only_language_model", "vision_config", "vision_model", "zeros"], "instructblipvideo/modeling_instructblipvideo.py:InstructBlipVideoForConditionalGeneration.get_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "language_model", "return", "self"], "instructblipvideo/modeling_instructblipvideo.py:InstructBlipVideoForConditionalGeneration.set_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "language_model", "self", "set_input_embeddings", "value"], "instructblipvideo/modeling_instructblipvideo.py:InstructBlipVideoForConditionalGeneration.set_output_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "language_model", "new_embeddings", "self", "set_output_embeddings"], "instructblipvideo/modeling_instructblipvideo.py:InstructBlipVideoForConditionalGeneration.get_output_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "Module", "class", "def", "get_output_embeddings", "language_model", "nn", "return", "self"], "instructblipvideo/modeling_instructblipvideo.py:InstructBlipVideoForConditionalGeneration.get_encoder": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "class", "def", "else", "get_encoder", "if", "is", "language_model", "modality", "return", "self", "super"], "instructblipvideo/modeling_instructblipvideo.py:InstructBlipVideoForConditionalGeneration.get_decoder": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "get_decoder", "language_model", "return", "self"], "instructblipvideo/modeling_instructblipvideo.py:InstructBlipVideoForConditionalGeneration._preprocess_accelerate": ["GPU", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "Please", "The", "True", "_hf_hook", "_preprocess_accelerate", "a", "accelerate", "and", "are", "behavior", "blob", "blog", "class", "com", "contains", "creating", "cuda", "def", "details", "device_count", "device_map", "dictionary", "environment", "for", "github", "hasattr", "hf_device_map", "https", "huggingface", "if", "in", "io_same_device", "is", "language_model", "large", "lead", "len", "logger", "main", "may", "md", "models", "more", "multi", "not", "on", "pass", "r", "refer", "remove", "running", "script", "self", "that", "the", "this", "to", "torch", "unexpected", "using", "warning", "when", "you", "your"], "instructblipvideo/modeling_instructblipvideo.py:InstructBlipVideoForConditionalGeneration.get_image_features": ["False", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "bool", "class", "def", "get_image_features", "interpolate_pos_encoding", "pass", "pixel_values", "qformer_attention_mask", "qformer_input_ids", "return_dict", "self", "torch"], "instructblipvideo/modeling_instructblipvideo.py:InstructBlipVideoForConditionalGeneration.get_placeholder_mask": ["FloatTensor", "GenerationMixin", "LongTensor", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "all", "class", "config", "def", "device", "dtype", "else", "expand_as", "get_input_embeddings", "get_placeholder_mask", "if", "input_ids", "inputs_embeds", "is", "long", "return", "self", "special_image_mask", "tensor", "to", "torch", "unsqueeze", "video_token_id"], "instructblipvideo/modeling_instructblipvideo.py:InstructBlipVideoForConditionalGeneration.forward": ["False", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForConditionalGeneration", "ModelForConditionalGenerationModelOutput", "ModelPreTrainedModel", "None", "TransformersKwargs", "True", "Unpack", "attention_mask", "bool", "class", "config", "decoder_attention_mask", "decoder_input_ids", "def", "device", "dtype", "else", "forward", "get_input_embeddings", "get_placeholder_mask", "get_video_features", "if", "input_ids", "inputs_embeds", "interpolate_pos_encoding", "is", "kwargs", "labels", "language_model", "language_model_inputs", "language_model_outputs", "logits", "loss", "loss_function", "masked_scatter", "not", "ones_like", "output_attentions", "output_hidden_states", "outputs", "pixel_values", "qformer_attention_mask", "qformer_input_ids", "qformer_outputs", "query_outputs", "r", "return", "return_dict", "self", "special_image_mask", "text_config", "to", "to_tuple", "torch", "tuple", "use_cache", "use_decoder_only_language_model", "use_return_dict", "vision_outputs", "vocab_size"], "instructblipvideo/modeling_instructblipvideo.py:InstructBlipVideoForConditionalGeneration.generate": ["False", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "True", "_preprocess_accelerate", "attention_mask", "batch_size", "bool", "bos_token_id", "class", "config", "def", "device", "dtype", "generate", "generate_kwargs", "get_input_embeddings", "get_placeholder_mask", "get_video_features", "hasattr", "hf_device_map", "if", "input_ids", "inputs", "inputs_embeds", "interpolate_pos_encoding", "is", "is_encoder_decoder", "language_model", "language_model_inputs", "long", "masked_scatter", "not", "num_query_tokens", "ones_like", "outputs", "pixel_values", "qformer_attention_mask", "qformer_input_ids", "query_outputs", "r", "repeat", "return", "return_dict", "self", "shape", "special_image_mask", "start_tokens", "tensor", "text_config", "to", "torch", "video_token_index", "video_tokens", "vision_outputs"], "instructblipvideo/modeling_instructblipvideo.py:InstructBlipVideoForConditionalGeneration.get_video_features": ["False", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "True", "attention_mask", "batch_size", "bool", "cat", "channel", "class", "config", "def", "device", "dim", "dtype", "encoder_attention_mask", "encoder_hidden_states", "expand", "frames", "get_video_features", "height", "if", "image_attention_mask", "image_embeds", "input_ids", "interpolate_pos_encoding", "is", "language_model_inputs", "language_projection", "long", "num_query_tokens", "ones", "ones_like", "pixel_values", "qformer", "qformer_attention_mask", "qformer_input_ids", "query_attention_mask", "query_embeds", "query_output", "query_outputs", "query_tokens", "repeat_interleave", "reshape", "return", "return_dict", "self", "shape", "size", "torch", "vision_model", "vision_outputs", "width"], "internvl/modeling_internvl.py:InternVLVisionRMSNorm.__init__": ["ModelVisionRMSNorm", "Module", "Parameter", "__init__", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "internvl/modeling_internvl.py:InternVLVisionRMSNorm.forward": ["ModelVisionRMSNorm", "Module", "True", "class", "def", "dtype", "float32", "forward", "hidden_states", "input_dtype", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "internvl/modeling_internvl.py:InternVLVisionRMSNorm.extra_repr": ["ModelVisionRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "variance_epsilon", "weight"], "internvl/modeling_internvl.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "float", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "shape", "softmax", "torch", "training", "transpose", "value", "value_states"], "internvl/modeling_internvl.py:InternVLVisionAttention.__init__": ["Dropout", "False", "Identity", "Linear", "ModelVisionAttention", "ModelVisionConfig", "ModelVisionRMSNorm", "Module", "ValueError", "__init__", "and", "attention_bias", "attention_dropout", "be", "bias", "by", "class", "config", "def", "divisible", "else", "embed_dim", "f", "got", "head_dim", "hidden_size", "if", "is_causal", "k_norm", "k_proj", "must", "nn", "num_attention_heads", "num_heads", "proj_dropout", "projection_dropout", "projection_layer", "q_norm", "q_proj", "qk_norm", "raise", "scale", "self", "super", "use_qk_norm", "v_proj"], "internvl/modeling_internvl.py:InternVLVisionAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "False", "ModelVisionAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "_attn_implementation", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "batch_size", "class", "config", "def", "dropout", "eager", "eager_attention_forward", "else", "embed_dim", "forward", "head_dim", "hidden_states", "if", "is_causal", "k_norm", "k_proj", "key_states", "kwargs", "nn", "not", "num_heads", "output", "projection_dropout", "projection_layer", "q_norm", "q_proj", "query_states", "reshape", "return", "scale", "scaling", "self", "seq_len", "size", "torch", "training", "transpose", "v_proj", "value_states", "view"], "internvl/modeling_internvl.py:InternVLVisionPatchEmbeddings.__init__": ["Conv2d", "ModelVisionPatchEmbeddings", "Module", "__init__", "class", "config", "def", "hidden_size", "image_size", "kernel_size", "nn", "num_channels", "num_patches", "patch_shape", "patch_size", "projection", "self", "stride", "super"], "internvl/modeling_internvl.py:InternVLVisionPatchEmbeddings.forward": ["Make", "ModelVisionPatchEmbeddings", "Module", "Tensor", "ValueError", "batch_size", "channel", "class", "configuration", "def", "dimension", "dtype", "embeddings", "flatten", "forward", "height", "if", "in", "match", "nn", "num_channels", "of", "one", "pixel", "pixel_values", "projection", "raise", "return", "self", "set", "shape", "sure", "that", "the", "to", "torch", "transpose", "values", "weight", "width", "with"], "internvl/modeling_internvl.py:InternVLVisionEmbeddings.__init__": ["Dropout", "Iterable", "ModelVisionConfig", "ModelVisionEmbeddings", "ModelVisionPatchEmbeddings", "Module", "None", "Parameter", "__init__", "abc", "class", "cls_token", "collections", "config", "def", "dropout", "else", "hidden_dropout_prob", "hidden_size", "if", "image_size", "isinstance", "mask_token", "nn", "num_patches", "patch_embeddings", "patch_size", "position_embeddings", "self", "super", "torch", "use_absolute_position_embeddings", "use_mask_token", "zeros"], "internvl/modeling_internvl.py:InternVLVisionEmbeddings.interpolate_pos_encoding": ["False", "ModelVisionEmbeddings", "Module", "Tensor", "align_corners", "and", "bicubic", "cat", "class", "class_pos_embed", "def", "dim", "embeddings", "functional", "height", "if", "int", "interpolate", "interpolate_pos_encoding", "is_tracing", "jit", "mode", "new_height", "new_width", "nn", "not", "num_patches", "num_positions", "patch_pos_embed", "patch_size", "permute", "position_embeddings", "reshape", "return", "self", "shape", "size", "sqrt_num_positions", "torch", "torch_int", "view", "width"], "internvl/modeling_internvl.py:InternVLVisionEmbeddings.forward": ["BoolTensor", "ModelVisionEmbeddings", "Module", "None", "Tensor", "_", "batch_size", "bool_masked_pos", "cat", "class", "cls_token", "cls_tokens", "def", "dim", "dropout", "embeddings", "expand", "forward", "height", "if", "interpolate_pos_encoding", "is", "mask_token", "mask_tokens", "nn", "not", "patch_embeddings", "pixel_values", "position_embeddings", "return", "self", "seq_len", "shape", "size", "torch", "type_as", "unsqueeze", "w", "width"], "internvl/modeling_internvl.py:InternVLVisionMLP.__init__": ["ACT2FN", "Linear", "ModelVisionMLP", "Module", "__init__", "activation_fn", "class", "config", "def", "fc1", "fc2", "hidden_act", "hidden_size", "intermediate_size", "nn", "self", "super"], "internvl/modeling_internvl.py:InternVLVisionMLP.forward": ["ModelVisionMLP", "Module", "Tensor", "activation_fn", "class", "def", "fc1", "fc2", "forward", "hidden_states", "nn", "return", "self", "torch"], "internvl/modeling_internvl.py:InternVLVisionLayer.__init__": ["Dropout", "GradientCheckpointingLayer", "ModelVisionAttention", "ModelVisionConfig", "ModelVisionLayer", "ModelVisionMLP", "NORM2FN", "None", "Parameter", "True", "__init__", "attention", "chunk_size_feed_forward", "class", "config", "def", "dropout", "eps", "hidden_dropout_prob", "hidden_size", "init_values", "lambda_1", "lambda_2", "layer_norm_eps", "layer_scale_init_value", "layernorm_after", "layernorm_before", "mlp", "nn", "norm_type", "ones", "requires_grad", "self", "seq_len_dim", "super", "torch"], "internvl/modeling_internvl.py:InternVLVisionLayer.forward": ["GradientCheckpointingLayer", "ModelVisionLayer", "None", "Tensor", "_", "attention", "attention_output", "class", "def", "dropout", "forward", "hidden_states", "if", "is", "lambda_1", "lambda_2", "layer_output", "layernorm_after", "layernorm_before", "mlp", "not", "return", "self", "torch", "tuple"], "internvl/modeling_internvl.py:InternVLVisionEncoder.__init__": ["False", "ModelVisionConfig", "ModelVisionEncoder", "ModelVisionLayer", "Module", "ModuleList", "None", "__init__", "class", "config", "def", "for", "gradient_checkpointing", "i", "in", "layer", "nn", "num_hidden_layers", "range", "self", "super"], "internvl/modeling_internvl.py:InternVLVisionEncoder.forward": ["BaseModelOutput", "ModelVisionEncoder", "Module", "Tensor", "class", "def", "for", "forward", "hidden_states", "in", "last_hidden_state", "layer", "layer_module", "nn", "return", "self", "torch", "tuple"], "internvl/modeling_internvl.py:InternVLVisionPreTrainedModel._init_weights": ["ModelVisionEmbeddings", "ModelVisionLayer", "ModelVisionPreTrainedModel", "None", "PreTrainedModel", "_init_weights", "class", "cls_token", "config", "constant_", "def", "elif", "if", "init", "is", "isinstance", "lambda_1", "lambda_2", "layer_scale_init_value", "mask_token", "module", "not", "position_embeddings", "self", "super", "zeros_"], "internvl/modeling_internvl.py:InternVLVisionModel.__init__": ["Identity", "LayerNorm", "ModelVisionConfig", "ModelVisionEmbeddings", "ModelVisionEncoder", "ModelVisionModel", "ModelVisionPreTrainedModel", "None", "__init__", "class", "config", "def", "else", "embeddings", "encoder", "eps", "hidden_size", "if", "layer_norm_eps", "layernorm", "nn", "post_init", "self", "super", "use_mean_pooling"], "internvl/modeling_internvl.py:InternVLVisionModel.get_input_embeddings": ["ModelVisionModel", "ModelVisionPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "patch_embeddings", "return", "self"], "internvl/modeling_internvl.py:InternVLVisionModel.forward": ["BoolTensor", "ModelVisionModel", "ModelVisionModelOutputWithPooling", "ModelVisionPreTrainedModel", "None", "Tensor", "attentions", "bool_masked_pos", "class", "def", "embedding_output", "embeddings", "encoder", "encoder_outputs", "forward", "hidden_states", "kwargs", "last_hidden_state", "layernorm", "pixel_values", "r", "return", "self", "sequence_output", "torch", "tuple"], "internvl/modeling_internvl.py:InternVLMultiModalProjector.__init__": ["ACT2FN", "LayerNorm", "Linear", "ModelConfig", "ModelMultiModalProjector", "Module", "__init__", "act", "class", "config", "def", "downsample_ratio", "hidden_size", "int", "layer_norm", "linear_1", "linear_2", "nn", "projector_hidden_act", "self", "super", "text_config", "vision_config"], "internvl/modeling_internvl.py:InternVLMultiModalProjector.forward": ["ModelMultiModalProjector", "Module", "act", "class", "def", "forward", "hidden_states", "image_features", "layer_norm", "linear_1", "linear_2", "nn", "return", "self"], "internvl/modeling_internvl.py:InternVLModel.__init__": ["AutoModel", "ModelConfig", "ModelModel", "ModelMultiModalProjector", "ModelPreTrainedModel", "__init__", "class", "config", "def", "from_config", "language_model", "multi_modal_projector", "post_init", "self", "super", "text_config", "vision_config", "vision_tower"], "internvl/modeling_internvl.py:InternVLModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "language_model", "return", "self"], "internvl/modeling_internvl.py:InternVLModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "language_model", "self", "set_input_embeddings", "value"], "internvl/modeling_internvl.py:InternVLModel.get_image_features": ["FloatTensor", "ModelModel", "ModelPreTrainedModel", "None", "batch_size", "channels", "class", "config", "def", "default", "downsample_ratio", "dtype", "else", "feature_size", "get_image_features", "hidden_states", "if", "int", "is", "kwargs", "last_hidden_state", "list", "multi_modal_projector", "not", "pixel_shuffle", "pixel_values", "reshape", "return", "scale_factor", "self", "shape", "str", "to", "torch", "vision_feature_layer", "vision_feature_select_strategy", "vision_features", "vision_model", "vision_tower"], "internvl/modeling_internvl.py:InternVLModel.get_placeholder_mask": ["FloatTensor", "Image", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "ValueError", "all", "and", "class", "config", "def", "device", "do", "dtype", "else", "expand_as", "f", "features", "get_input_embeddings", "get_placeholder_mask", "if", "image", "image_features", "image_token_id", "input_ids", "inputs_embeds", "is", "long", "match", "n_image_features", "n_image_tokens", "not", "numel", "raise", "return", "self", "shape", "special_image_mask", "sum", "tensor", "to", "tokens", "torch", "unsqueeze"], "internvl/modeling_internvl.py:InternVLModel.forward": ["Cache", "FloatTensor", "LongTensor", "ModelModel", "ModelModelOutputWithPast", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "attention_mask", "attentions", "cache_position", "class", "config", "def", "device", "dtype", "else", "exactly", "forward", "get_image_features", "get_input_embeddings", "get_placeholder_mask", "hidden_states", "if", "image_features", "image_hidden_states", "input_ids", "inputs_embeds", "int", "is", "kwargs", "language_model", "last_hidden_state", "list", "masked_scatter", "must", "not", "of", "one", "or", "outputs", "past_key_values", "pixel_values", "position_ids", "raise", "return", "self", "special_image_mask", "specify", "str", "to", "torch", "tuple", "vision_feature_layer", "vision_feature_select_strategy"], "internvl/modeling_internvl.py:InternVLModel.pixel_shuffle": ["Height", "ModelModel", "ModelPreTrainedModel", "Tensor", "ValueError", "and", "batch_size", "be", "by", "channels", "class", "contiguous", "def", "divisible", "downsampling", "float", "for", "height", "if", "int", "must", "or", "permute", "pixel_shuffle", "proper", "raise", "return", "scale_factor", "self", "size", "torch", "view", "vision_features", "width"], "internvl/modeling_internvl.py:InternVLForConditionalGeneration.__init__": ["False", "GenerationMixin", "Linear", "ModelConfig", "ModelForConditionalGeneration", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "text_config", "vocab_size"], "internvl/modeling_internvl.py:InternVLForConditionalGeneration.get_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "model", "return", "self"], "internvl/modeling_internvl.py:InternVLForConditionalGeneration.set_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "model", "self", "set_input_embeddings", "value"], "internvl/modeling_internvl.py:InternVLForConditionalGeneration.get_output_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "Module", "class", "def", "get_output_embeddings", "lm_head", "nn", "return", "self"], "internvl/modeling_internvl.py:InternVLForConditionalGeneration.get_image_features": ["FloatTensor", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "class", "def", "get_image_features", "int", "kwargs", "list", "model", "pixel_values", "return", "self", "str", "torch", "vision_feature_layer", "vision_feature_select_strategy"], "internvl/modeling_internvl.py:InternVLForConditionalGeneration.forward": ["Cache", "FloatTensor", "GenerationMixin", "LongTensor", "ModelCausalLMOutputWithPast", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "cache_position", "class", "config", "def", "else", "forward", "hidden_states", "if", "image_hidden_states", "image_sizes", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "list", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "outputs", "past_key_values", "pixel_values", "position_ids", "r", "return", "self", "slice", "slice_indices", "str", "text_config", "torch", "tuple", "vision_feature_layer", "vision_feature_select_strategy", "vocab_size"], "internvl/modeling_internvl.py:InternVLForConditionalGeneration.prepare_inputs_for_generation": ["False", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "True", "attention_mask", "cache_position", "class", "def", "get", "if", "input_ids", "inputs_embeds", "is_first_iteration", "kwargs", "logits_to_keep", "model_inputs", "not", "or", "past_key_values", "pixel_values", "prepare_inputs_for_generation", "return", "self", "super", "use_cache"], "jais2/modeling_jais2.py:Jais2MLP.__init__": ["ACT2FN", "Linear", "ModelMLP", "Module", "__init__", "act_fn", "bias", "class", "config", "def", "down_proj", "hidden_act", "hidden_size", "intermediate_size", "mlp_bias", "nn", "self", "super", "up_proj"], "jais2/modeling_jais2.py:Jais2MLP.forward": ["ModelMLP", "Module", "act_fn", "class", "def", "down_proj", "forward", "nn", "return", "self", "up_proj", "x"], "jais2/modeling_jais2.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "jais2/modeling_jais2.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cos", "def", "k", "k_embed", "q", "q_embed", "return", "rotate_half", "sin", "unsqueeze", "unsqueeze_dim"], "jais2/modeling_jais2.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "jais2/modeling_jais2.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value", "value_states"], "jais2/modeling_jais2.py:Jais2Attention.__init__": ["Linear", "ModelAttention", "ModelConfig", "Module", "True", "__init__", "attention_bias", "attention_dropout", "bias", "class", "config", "def", "getattr", "head_dim", "hidden_size", "int", "is_causal", "k_proj", "layer_idx", "nn", "num_attention_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "q_proj", "scaling", "self", "super", "v_proj"], "jais2/modeling_jais2.py:Jais2Attention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "LongTensor", "ModelAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_attn_implementation", "apply_rotary_pos_emb", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "cache_kwargs", "cache_position", "class", "config", "contiguous", "cos", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "head_dim", "hidden_shape", "hidden_states", "if", "input_shape", "is", "k_proj", "key_states", "kwargs", "layer_idx", "nn", "not", "o_proj", "past_key_values", "position_embeddings", "q_proj", "query_states", "reshape", "return", "scaling", "self", "shape", "sin", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "view"], "jais2/modeling_jais2.py:Jais2DecoderLayer.__init__": ["GradientCheckpointingLayer", "LayerNorm", "ModelAttention", "ModelConfig", "ModelDecoderLayer", "ModelMLP", "__init__", "class", "config", "def", "eps", "hidden_size", "input_layernorm", "int", "layer_idx", "layer_norm_eps", "mlp", "nn", "post_attention_layernorm", "self", "self_attn", "super"], "jais2/modeling_jais2.py:Jais2DecoderLayer.forward": ["Cache", "False", "GradientCheckpointingLayer", "LongTensor", "ModelDecoderLayer", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "attention_mask", "bool", "cache_position", "class", "def", "forward", "hidden_states", "input_layernorm", "kwargs", "mlp", "past_key_values", "position_embeddings", "position_ids", "post_attention_layernorm", "residual", "return", "self", "self_attn", "torch", "tuple", "use_cache"], "jais2/modeling_jais2.py:Jais2RotaryEmbedding.__init__": ["Callable", "False", "ModelConfig", "ModelRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "jais2/modeling_jais2.py:Jais2RotaryEmbedding.compute_default_rope_parameters": ["ModelConfig", "ModelRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "jais2/modeling_jais2.py:Jais2RotaryEmbedding.forward": ["False", "ModelRotaryEmbedding", "Module", "None", "and", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "jais2/modeling_jais2.py:Jais2Model.__init__": ["Embedding", "False", "LayerNorm", "ModelConfig", "ModelDecoderLayer", "ModelModel", "ModelPreTrainedModel", "ModelRotaryEmbedding", "ModuleList", "__init__", "class", "config", "def", "embed_tokens", "eps", "for", "gradient_checkpointing", "hidden_size", "in", "layer_idx", "layer_norm_eps", "layers", "nn", "norm", "num_hidden_layers", "pad_token_id", "padding_idx", "post_init", "range", "rotary_emb", "self", "super", "vocab_size"], "jais2/modeling_jais2.py:Jais2Model.forward": ["BaseModelOutputWithPast", "Cache", "DynamicCache", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "and", "arange", "attention_mask", "bool", "cache_position", "causal_mask", "class", "config", "create_causal_mask", "decoder_layer", "def", "device", "else", "embed_tokens", "exactly", "for", "forward", "get_seq_length", "hidden_states", "if", "in", "input_embeds", "input_ids", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layers", "must", "norm", "not", "num_hidden_layers", "of", "one", "or", "past_key_values", "past_seen_tokens", "position_embeddings", "position_ids", "raise", "return", "rotary_emb", "self", "shape", "specify", "torch", "unsqueeze", "use_cache"], "jais2/modeling_jais2.py:Jais2ForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "ModelForCausalLM", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "vocab_size"], "jais2/modeling_jais2.py:Jais2ForCausalLM.forward": ["BaseModelOutputWithPast", "Cache", "CausalLMOutputWithPast", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForCausalLM", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "outputs", "past_key_values", "position_ids", "r", "return", "self", "slice", "slice_indices", "torch", "use_cache", "vocab_size"], "jamba/modeling_jamba.py:JambaRMSNorm.__init__": ["ModelRMSNorm", "Module", "Parameter", "__init__", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "jamba/modeling_jamba.py:JambaRMSNorm.forward": ["ModelRMSNorm", "Module", "True", "class", "def", "dtype", "float32", "forward", "hidden_states", "input_dtype", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "jamba/modeling_jamba.py:JambaRMSNorm.extra_repr": ["ModelRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "variance_epsilon", "weight"], "jamba/modeling_jamba.py:HybridMambaAttentionDynamicCache.__init__": ["False", "ModelMambaAttentionDynamicCache", "None", "_", "__init__", "append", "batch_size", "class", "config", "conv_kernel_size", "conv_states", "def", "device", "dtype", "else", "float16", "for", "has_previous_state", "hidden_size", "i", "if", "in", "intermediate_size", "key_cache", "layers_block_type", "mamba", "mamba_d_conv", "mamba_d_state", "mamba_expand", "num_hidden_layers", "range", "self", "ssm_state_size", "ssm_states", "tensor", "torch", "transformer_layers", "value_cache", "zeros"], "jamba/modeling_jamba.py:HybridMambaAttentionDynamicCache.__len__": ["ModelMambaAttentionDynamicCache", "__len__", "class", "def", "key_cache", "len", "return", "self"], "jamba/modeling_jamba.py:HybridMambaAttentionDynamicCache.__getitem__": ["ModelMambaAttentionDynamicCache", "__getitem__", "class", "def", "key_cache", "layer_idx", "return", "self", "value_cache"], "jamba/modeling_jamba.py:HybridMambaAttentionDynamicCache.update": ["Any", "ModelMambaAttentionDynamicCache", "None", "Tensor", "cache_kwargs", "cat", "class", "def", "dict", "dim", "else", "if", "int", "key_cache", "key_states", "layer_idx", "return", "self", "shape", "str", "torch", "tuple", "update", "value_cache", "value_states"], "jamba/modeling_jamba.py:HybridMambaAttentionDynamicCache.reorder_cache": ["LongTensor", "ModelMambaAttentionDynamicCache", "beam_idx", "class", "conv_states", "def", "device", "for", "get_seq_length", "if", "in", "index_select", "key_cache", "layer_idx", "len", "range", "reorder_cache", "self", "ssm_states", "to", "torch", "value_cache"], "jamba/modeling_jamba.py:HybridMambaAttentionDynamicCache.get_mask_sizes": ["ModelMambaAttentionDynamicCache", "Tensor", "cache_position", "class", "def", "get_mask_sizes", "get_seq_length", "int", "kv_length", "kv_offset", "layer_idx", "query_length", "return", "self", "shape", "torch", "tuple"], "jamba/modeling_jamba.py:HybridMambaAttentionDynamicCache.get_seq_length": ["ModelMambaAttentionDynamicCache", "None", "class", "def", "else", "get_seq_length", "if", "in", "int", "key_cache", "layer_idx", "len", "not", "or", "return", "self", "shape", "transformer_layers"], "jamba/modeling_jamba.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "jamba/modeling_jamba.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cos", "def", "k", "k_embed", "q", "q_embed", "return", "rotate_half", "sin", "unsqueeze", "unsqueeze_dim"], "jamba/modeling_jamba.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "jamba/modeling_jamba.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value", "value_states"], "jamba/modeling_jamba.py:JambaAttention.__init__": ["False", "Linear", "ModelAttention", "ModelConfig", "Module", "True", "__init__", "attention_dropout", "bias", "class", "config", "def", "getattr", "head_dim", "hidden_size", "int", "is_causal", "k_proj", "layer_idx", "nn", "num_attention_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "q_proj", "scaling", "self", "super", "v_proj"], "jamba/modeling_jamba.py:JambaAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "HybridMambaAttentionDynamicCache", "LongTensor", "ModelAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_attn_implementation", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "cache_position", "class", "config", "contiguous", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "head_dim", "hidden_shape", "hidden_states", "if", "input_shape", "is", "k_proj", "key_states", "kwargs", "layer_idx", "nn", "not", "o_proj", "past_key_values", "q_proj", "query_states", "reshape", "return", "scaling", "self", "shape", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "view"], "jamba/modeling_jamba.py:JambaMambaMixer.__init__": ["A", "ACT2FN", "AILab", "A_log", "Conv1d", "D", "Dao", "False", "If", "Linear", "ModelConfig", "ModelMambaMixer", "ModelRMSNorm", "Module", "None", "Parameter", "The", "To", "True", "__init__", "act", "activation", "all", "arange", "available", "b_layernorm", "because", "bias", "c_layernorm", "causal", "causal_conv1d", "causal_conv1d_fn", "causal_conv1d_update", "class", "com", "config", "contiguous", "conv1d", "conv_kernel_size", "def", "dt_layernorm", "dt_proj", "eps", "expand", "fast", "follow", "getattr", "github", "global", "groups", "hidden_act", "hidden_size", "https", "if", "implementation", "in", "in_channels", "in_proj", "install", "intermediate_size", "is", "is_fast_path_available", "kernel_size", "layer_idx", "lazy_load_kernel", "log", "logger", "mamba", "mamba_conv_bias", "mamba_d_conv", "mamba_d_state", "mamba_dt_rank", "mamba_expand", "mamba_inner_fn", "mamba_proj_bias", "mamba_ssm", "model", "naive", "nn", "not", "of", "on", "ones", "out_channels", "out_proj", "padding", "path", "rms_norm_eps", "selective_scan_fn", "selective_state_update", "self", "set", "spaces", "ssm", "ssm_state_size", "state", "super", "the", "time_step_rank", "to", "torch", "use", "use_bias", "use_conv_bias", "use_fast_kernels", "use_mamba_kernels", "want", "warning_once", "x_proj", "you"], "jamba/modeling_jamba.py:JambaMambaMixer.cuda_kernels_forward": ["A", "A_log", "B", "C", "D", "HybridMambaAttentionDynamicCache", "LongTensor", "ModelMambaMixer", "Module", "None", "Tensor", "True", "_", "activation", "and", "attention_mask", "b_layernorm", "batch_size", "bias", "c_layernorm", "cache_params", "causal_conv1d_fn", "causal_conv1d_update", "chunk", "class", "contextualized_states", "conv1d", "conv_kernel_size", "conv_states", "conv_weights", "copy_", "cuda_kernels_forward", "data", "def", "delta_softplus", "dim", "discrete_time_step", "dt_layernorm", "dt_proj", "dt_softplus", "else", "exp", "float", "functional", "gate", "has_previous_state", "hidden_states", "if", "in_proj", "is", "layer_idx", "nn", "no_grad", "not", "out_proj", "pad", "projected_states", "return", "return_last_state", "scan_outputs", "selective_scan_fn", "selective_state_update", "self", "seq_len", "shape", "size", "split", "squeeze", "ssm_parameters", "ssm_state", "ssm_state_size", "ssm_states", "time_proj_bias", "time_step", "time_step_rank", "torch", "transpose", "unsqueeze", "use_precomputed_states", "view", "weight", "with", "x_proj", "zeros_like"], "jamba/modeling_jamba.py:JambaMambaMixer.slow_forward": ["A", "A_log", "B", "C", "D", "HybridMambaAttentionDynamicCache", "LongTensor", "ModelMambaMixer", "Module", "None", "_", "act", "and", "append", "attention_mask", "b_layernorm", "batch_size", "bias", "c_layernorm", "cache_params", "chunk", "class", "clone", "contextualized_states", "conv1d", "conv_kernel_size", "conv_state", "conv_states", "def", "deltaB_u", "device", "dim", "dims", "discrete_A", "discrete_B", "discrete_time_step", "dt_layernorm", "dt_proj", "dtype", "else", "exp", "float", "for", "functional", "gate", "has_previous_state", "hidden_states", "i", "if", "in", "in_proj", "input_states", "intermediate_size", "is", "isinstance", "layer_idx", "matmul", "nn", "not", "out_proj", "pad", "projected_states", "range", "return", "roll", "scan_output", "scan_outputs", "self", "seq_len", "shape", "shifts", "slow_forward", "softplus", "split", "ssm_parameters", "ssm_state", "ssm_state_size", "ssm_states", "stack", "sum", "time_step", "time_step_rank", "to", "torch", "training", "transpose", "unsqueeze", "use_cache", "use_conv_bias", "weight", "x_proj", "zeros"], "jamba/modeling_jamba.py:JambaMambaMixer.forward": ["CUDA", "Fast", "HybridMambaAttentionDynamicCache", "LongTensor", "Make", "Mamba", "ModelMambaMixer", "Module", "None", "ValueError", "a", "and", "are", "attention_mask", "available", "cache_params", "class", "cuda", "cuda_kernels_forward", "def", "device", "forward", "hidden_states", "if", "in", "installed", "is", "is_fast_path_available", "kernels", "mamba", "module", "nn", "not", "on", "or", "raise", "return", "self", "slow_forward", "sure", "that", "the", "they", "to", "torch", "type", "use_fast_kernels", "weight", "x_proj"], "jamba/modeling_jamba.py:JambaMLP.__init__": ["ACT2FN", "False", "Linear", "ModelMLP", "Module", "__init__", "act_fn", "bias", "class", "config", "def", "down_proj", "gate_proj", "hidden_act", "hidden_size", "intermediate_size", "nn", "self", "super", "up_proj"], "jamba/modeling_jamba.py:JambaMLP.forward": ["ModelMLP", "Module", "act_fn", "class", "def", "down_proj", "forward", "gate_proj", "nn", "return", "self", "up_proj", "x"], "jamba/modeling_jamba.py:JambaExperts.__init__": ["ACT2FN", "ModelConfig", "ModelExperts", "Module", "Parameter", "__init__", "act_fn", "class", "config", "def", "down_proj", "empty", "gate_up_proj", "hidden_act", "hidden_dim", "hidden_size", "intermediate_dim", "intermediate_size", "nn", "num_experts", "num_local_experts", "self", "super", "torch"], "jamba/modeling_jamba.py:JambaExperts.forward": ["ModelExperts", "Module", "None", "Tensor", "act_fn", "chunk", "class", "continue", "current_hidden_states", "current_state", "def", "dim", "down_proj", "dtype", "expert_hit", "expert_idx", "expert_mask", "final_hidden_states", "for", "forward", "functional", "gate", "gate_up_proj", "greater", "hidden_states", "if", "in", "index_add_", "linear", "nn", "no_grad", "nonzero", "num_classes", "num_experts", "one_hot", "permute", "return", "self", "sum", "to", "token_idx", "top_k_index", "top_k_pos", "top_k_weights", "torch", "up", "where", "with", "zeros_like"], "jamba/modeling_jamba.py:JambaSparseMoeBlock.__init__": ["False", "Linear", "ModelConfig", "ModelExperts", "ModelSparseMoeBlock", "Module", "__init__", "bias", "class", "config", "def", "experts", "ffn_dim", "hidden_dim", "hidden_size", "intermediate_size", "nn", "num_experts", "num_experts_per_tok", "router", "self", "super", "top_k"], "jamba/modeling_jamba.py:JambaSparseMoeBlock.route_tokens_to_experts": ["ModelSparseMoeBlock", "Module", "class", "def", "dim", "dtype", "float", "functional", "hidden_states", "nn", "return", "route_tokens_to_experts", "router_logits", "routing_weights", "self", "softmax", "to", "top_k", "top_k_index", "top_k_weights", "topk", "torch"], "jamba/modeling_jamba.py:JambaSparseMoeBlock.forward": ["ModelSparseMoeBlock", "Module", "Tensor", "batch_size", "class", "def", "experts", "forward", "hidden_dim", "hidden_states", "nn", "reshape", "return", "route_tokens_to_experts", "router", "router_logits", "self", "sequence_length", "shape", "top_k_index", "top_k_weights", "torch", "view"], "jamba/modeling_jamba.py:JambaAttentionDecoderLayer.__init__": ["GradientCheckpointingLayer", "ModelAttention", "ModelAttentionDecoderLayer", "ModelConfig", "ModelMLP", "ModelRMSNorm", "ModelSparseMoeBlock", "__init__", "class", "config", "def", "else", "eps", "feed_forward", "ffn_layer_class", "hidden_size", "if", "input_layernorm", "int", "layer_idx", "layers_num_experts", "num_experts", "pre_ff_layernorm", "rms_norm_eps", "self", "self_attn", "super"], "jamba/modeling_jamba.py:JambaAttentionDecoderLayer.forward": ["False", "FloatTensor", "GradientCheckpointingLayer", "HybridMambaAttentionDynamicCache", "LongTensor", "ModelAttentionDecoderLayer", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "attention_mask", "bool", "cache_position", "class", "def", "feed_forward", "forward", "hidden_states", "input_layernorm", "kwargs", "past_key_values", "position_ids", "pre_ff_layernorm", "residual", "return", "self", "self_attn", "torch", "use_cache"], "jamba/modeling_jamba.py:JambaMambaDecoderLayer.__init__": ["GradientCheckpointingLayer", "ModelConfig", "ModelMLP", "ModelMambaDecoderLayer", "ModelMambaMixer", "ModelRMSNorm", "ModelSparseMoeBlock", "__init__", "class", "config", "def", "else", "eps", "feed_forward", "ffn_layer_class", "hidden_size", "if", "input_layernorm", "int", "layer_idx", "layers_num_experts", "mamba", "num_experts", "pre_ff_layernorm", "rms_norm_eps", "self", "super"], "jamba/modeling_jamba.py:JambaMambaDecoderLayer.forward": ["FloatTensor", "GradientCheckpointingLayer", "HybridMambaAttentionDynamicCache", "LongTensor", "ModelMambaDecoderLayer", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "cache_params", "class", "def", "feed_forward", "forward", "hidden_states", "input_layernorm", "kwargs", "mamba", "past_key_values", "position_ids", "pre_ff_layernorm", "residual", "return", "self", "torch"], "jamba/modeling_jamba.py:JambaPreTrainedModel._init_weights": ["A", "A_log", "D", "ModelExperts", "ModelMambaMixer", "ModelPreTrainedModel", "None", "PreTrainedModel", "_init_weights", "arange", "class", "config", "contiguous", "copy_", "def", "down_proj", "elif", "expand", "gate_up_proj", "if", "init", "initializer_range", "intermediate_size", "isinstance", "log", "mean", "module", "normal_", "ones_", "self", "ssm_state_size", "std", "super", "torch"], "jamba/modeling_jamba.py:JambaModel.__init__": ["ALL_DECODER_LAYER_TYPES", "Embedding", "False", "ModelConfig", "ModelModel", "ModelPreTrainedModel", "ModelRMSNorm", "ModuleList", "__init__", "append", "class", "config", "decoder_layers", "def", "embed_tokens", "eps", "final_layernorm", "for", "gradient_checkpointing", "hidden_size", "i", "in", "layer_class", "layer_idx", "layers", "layers_block_type", "nn", "num_hidden_layers", "pad_token_id", "padding_idx", "post_init", "range", "rms_norm_eps", "self", "super", "vocab_size"], "jamba/modeling_jamba.py:JambaModel.forward": ["FloatTensor", "HybridMambaAttentionDynamicCache", "LongTensor", "ModelMambaDecoderLayer", "ModelModel", "ModelPreTrainedModel", "MoeModelOutputWithPast", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "ValueError", "You", "_update_mamba_mask", "and", "arange", "attention_mask", "batch_size", "bool", "cache_position", "causal_mask", "class", "config", "create_causal_mask", "decoder_layer", "def", "device", "dtype", "else", "embed_tokens", "exactly", "final_layernorm", "for", "forward", "get_seq_length", "has_previous_state", "hidden_states", "if", "in", "input_embeds", "input_ids", "inputs_embeds", "is", "isinstance", "kwargs", "last_hidden_state", "layer_mask", "layers", "mamba_mask", "must", "not", "of", "one", "or", "past_key_values", "past_seen_tokens", "position_ids", "raise", "return", "self", "shape", "specify", "torch", "unsqueeze", "use_cache"], "jamba/modeling_jamba.py:JambaModel._update_mamba_mask": ["ModelModel", "ModelPreTrainedModel", "None", "_update_mamba_mask", "all", "and", "attention_mask", "cache_position", "class", "def", "if", "is", "mamba_mask", "not", "or", "return", "self", "torch"], "jamba/modeling_jamba.py:load_balancing_loss_func": ["Model_balancing_loss_func", "None", "Tensor", "_", "attention_mask", "batch_size", "cat", "compute_device", "concatenated_gate_logits", "def", "device", "dim", "else", "expand", "expert_attention_mask", "expert_mask", "float", "for", "functional", "gate_logits", "if", "in", "int", "is", "isinstance", "layer_gate", "mean", "nn", "not", "num_experts", "num_hidden_layers", "one_hot", "or", "overall_loss", "r", "reshape", "return", "router_per_expert_attention_mask", "router_prob_per_expert", "routing_weights", "selected_experts", "sequence_length", "shape", "softmax", "sum", "to", "tokens_per_expert", "top_k", "topk", "torch", "tuple", "unsqueeze"], "jamba/modeling_jamba.py:JambaForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "ModelConfig", "ModelForCausalLM", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "num_experts", "num_experts_per_tok", "post_init", "router_aux_loss_coef", "self", "super", "vocab_size"], "jamba/modeling_jamba.py:JambaForCausalLM.forward": ["FloatTensor", "GenerationMixin", "HybridMambaAttentionDynamicCache", "LongTensor", "ModelForCausalLM", "ModelPreTrainedModel", "MoeCausalLMOutputWithPast", "MoeModelOutputWithPast", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "aux_loss", "bool", "cache_position", "class", "config", "def", "device", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "load_balancing_loss_func", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "num_experts", "num_experts_per_tok", "output_router_logits", "outputs", "past_key_values", "position_ids", "r", "return", "router_aux_loss_coef", "router_logits", "self", "slice", "slice_indices", "to", "torch", "use_cache", "vocab_size"], "janus/modeling_janus.py:JanusPreTrainedModel._init_weights": ["ModelPreTrainedModel", "ModelVisionEmbeddings", "PreTrainedModel", "_init_weights", "arange", "class", "copy_", "def", "expand", "if", "init", "isinstance", "module", "position_ids", "self", "shape", "super", "torch"], "janus/modeling_janus.py:JanusVisionEmbeddings.__init__": ["Conv2d", "Embedding", "False", "ModelVisionConfig", "ModelVisionEmbeddings", "Module", "__init__", "arange", "class", "config", "def", "embed_dim", "expand", "hidden_size", "image_size", "in_channels", "kernel_size", "nn", "num_channels", "num_patches", "num_positions", "out_channels", "padding", "patch_embedding", "patch_size", "persistent", "position_embedding", "position_ids", "register_buffer", "self", "stride", "super", "torch", "valid"], "janus/modeling_janus.py:JanusVisionEmbeddings.interpolate_pos_encoding": ["False", "ModelVisionEmbeddings", "Module", "Tensor", "align_corners", "and", "bicubic", "class", "def", "dim", "embeddings", "functional", "height", "if", "int", "interpolate", "interpolate_pos_encoding", "is_tracing", "jit", "mode", "new_height", "new_width", "nn", "not", "num_patches", "num_positions", "patch_pos_embed", "patch_size", "permute", "position_embedding", "position_ids", "reshape", "return", "self", "shape", "size", "sqrt_num_positions", "torch", "torch_int", "unsqueeze", "view", "weight", "width"], "janus/modeling_janus.py:JanusVisionEmbeddings.forward": ["False", "ModelVisionEmbeddings", "Module", "Tensor", "_", "bool", "class", "def", "dtype", "else", "embeddings", "flatten", "forward", "height", "if", "interpolate_pos_encoding", "nn", "patch_embedding", "patch_embeds", "pixel_values", "pos_embeds", "position_embedding", "position_ids", "return", "self", "shape", "target_dtype", "to", "torch", "transpose", "weight", "width"], "janus/modeling_janus.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "janus/modeling_janus.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value", "value_states"], "janus/modeling_janus.py:JanusVisionAttention.__init__": ["Dropout", "False", "Identity", "LayerNorm", "Linear", "ModelVisionAttention", "ModelVisionConfig", "Module", "ValueError", "__init__", "and", "attention_bias", "attention_dropout", "be", "bias", "by", "class", "config", "def", "divisible", "else", "embed_dim", "f", "got", "head_dim", "hidden_size", "if", "is_causal", "k_norm", "k_proj", "must", "nn", "num_attention_heads", "num_heads", "num_key_value_groups", "proj_dropout", "projection_dropout", "projection_layer", "q_norm", "q_proj", "qk_norm", "raise", "scale", "self", "super", "use_qk_norm", "v_proj"], "janus/modeling_janus.py:JanusVisionAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "ModelVisionAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "_attn_implementation", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "batch_size", "class", "config", "def", "dropout", "eager", "eager_attention_forward", "else", "embed_dim", "forward", "head_dim", "hidden_states", "if", "is_causal", "k_norm", "k_proj", "key_states", "kwargs", "nn", "not", "num_heads", "output", "projection_dropout", "projection_layer", "q_norm", "q_proj", "query_states", "reshape", "return", "scale", "scaling", "self", "seq_len", "size", "torch", "training", "transpose", "v_proj", "value_states", "view"], "janus/modeling_janus.py:JanusVisionMLP.__init__": ["ACT2FN", "Dropout", "Linear", "ModelVisionConfig", "ModelVisionMLP", "Module", "__init__", "activation_fn", "class", "config", "def", "dropout1", "dropout2", "fc1", "fc2", "hidden_act", "hidden_dropout_rate", "hidden_size", "int", "intermediate_size", "mlp_ratio", "nn", "self", "super"], "janus/modeling_janus.py:JanusVisionMLP.forward": ["ModelVisionMLP", "Module", "Tensor", "activation_fn", "class", "def", "dropout1", "dropout2", "fc1", "fc2", "forward", "hidden_states", "nn", "return", "self", "torch"], "janus/modeling_janus.py:JanusVisionEncoderLayer.__init__": ["GradientCheckpointingLayer", "LayerNorm", "ModelVisionAttention", "ModelVisionConfig", "ModelVisionEncoderLayer", "ModelVisionMLP", "__init__", "class", "config", "def", "embed_dim", "eps", "hidden_size", "layer_norm1", "layer_norm2", "layer_norm_eps", "mlp", "nn", "self", "self_attn", "super"], "janus/modeling_janus.py:JanusVisionEncoderLayer.forward": ["FloatTensor", "GradientCheckpointingLayer", "ModelVisionEncoderLayer", "Tensor", "TransformersKwargs", "Unpack", "_", "attention_mask", "class", "def", "forward", "hidden_states", "kwargs", "layer_norm1", "layer_norm2", "mlp", "residual", "return", "self", "self_attn", "torch"], "janus/modeling_janus.py:JanusVisionEncoder.__init__": ["False", "ModelVisionConfig", "ModelVisionEncoder", "ModelVisionEncoderLayer", "Module", "ModuleList", "_", "__init__", "class", "config", "def", "for", "gradient_checkpointing", "in", "layers", "nn", "num_hidden_layers", "range", "self", "super"], "janus/modeling_janus.py:JanusVisionEncoder.forward": ["BaseModelOutput", "ModelVisionEncoder", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "class", "def", "encoder_layer", "for", "forward", "hidden_states", "in", "inputs_embeds", "kwargs", "last_hidden_state", "layers", "nn", "return", "self", "torch"], "janus/modeling_janus.py:JanusAttention.__init__": ["False", "Linear", "ModelAttention", "Module", "None", "Parameter", "ValueError", "__init__", "and", "attention_dropout", "be", "bias", "by", "cat", "class", "config", "def", "divisible", "else", "embed_dim", "f", "got", "head_dim", "hidden_size", "if", "is", "is_causal", "must", "nn", "not", "num_attention_heads", "num_heads", "projection", "q_bias", "qkv", "qkv_bias", "raise", "requires_grad", "scale", "self", "super", "torch", "v_bias", "zeros", "zeros_like"], "janus/modeling_janus.py:JanusAttention._shape": ["ModelAttention", "Module", "Tensor", "_shape", "bsz", "class", "contiguous", "def", "head_dim", "int", "nn", "num_heads", "return", "self", "seq_len", "tensor", "torch", "transpose", "view"], "janus/modeling_janus.py:JanusAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "ModelAttention", "Module", "None", "Tensor", "_attn_implementation", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "bsz", "class", "config", "contiguous", "def", "dropout", "eager", "eager_attention_forward", "else", "embed_dim", "forward", "hidden_states", "if", "key_states", "kwargs", "mixed_qkv", "nn", "not", "num_heads", "permute", "projection", "qkv", "query_states", "reshape", "return", "scale", "scaling", "self", "size", "tgt_len", "torch", "training", "tuple", "value_states"], "janus/modeling_janus.py:JanusMLP.__init__": ["ACT2FN", "Linear", "ModelMLP", "Module", "__init__", "activation_fn", "class", "config", "def", "fc1", "fc2", "hidden_act", "hidden_size", "intermediate_size", "nn", "self", "super"], "janus/modeling_janus.py:JanusMLP.forward": ["ModelMLP", "Module", "Tensor", "activation_fn", "class", "def", "fc1", "fc2", "forward", "hidden_states", "nn", "return", "self", "torch"], "janus/modeling_janus.py:JanusEncoderLayer.__init__": ["GradientCheckpointingLayer", "LayerNorm", "ModelAttention", "ModelConfig", "ModelEncoderLayer", "ModelMLP", "__init__", "class", "config", "def", "embed_dim", "eps", "hidden_size", "layer_norm1", "layer_norm2", "layer_norm_eps", "mlp", "nn", "self", "self_attn", "super"], "janus/modeling_janus.py:JanusEncoderLayer.forward": ["FloatTensor", "GradientCheckpointingLayer", "ModelEncoderLayer", "Tensor", "TransformersKwargs", "Unpack", "_", "class", "def", "forward", "hidden_states", "kwargs", "layer_norm1", "layer_norm2", "mlp", "residual", "return", "self", "self_attn", "torch"], "janus/modeling_janus.py:JanusVisionModel.__init__": ["LayerNorm", "ModelPreTrainedModel", "ModelVisionConfig", "ModelVisionEmbeddings", "ModelVisionEncoder", "ModelVisionModel", "__init__", "class", "config", "def", "embed_dim", "embeddings", "encoder", "eps", "hidden_size", "layer_norm_eps", "nn", "post_init", "post_layernorm", "self", "super"], "janus/modeling_janus.py:JanusVisionModel.forward": ["BaseModelOutput", "BaseModelOutputWithPooling", "False", "FloatTensor", "ModelPreTrainedModel", "ModelVisionModel", "None", "TransformersKwargs", "Unpack", "ValueError", "You", "bool", "class", "def", "embeddings", "encoder", "encoder_outputs", "forward", "have", "hidden_states", "if", "inputs_embeds", "interpolate_pos_encoding", "is", "kwargs", "last_hidden_state", "pixel_values", "pooled_output", "pooler_output", "post_layernorm", "raise", "return", "self", "specify", "to", "torch", "tuple"], "janus/modeling_janus.py:JanusVisionModel.get_input_embeddings": ["ModelPreTrainedModel", "ModelVisionModel", "class", "def", "embeddings", "get_input_embeddings", "return", "self"], "janus/modeling_janus.py:JanusVisionAlignerMLP.__init__": ["ACT2FN", "Linear", "ModelVisionAlignerMLP", "ModelVisionConfig", "Module", "ModuleList", "_", "__init__", "activation_fn", "class", "config", "def", "depth", "fc1", "for", "hidden_act", "hidden_layers", "hidden_size", "in", "nn", "projection_dim", "range", "self", "super"], "janus/modeling_janus.py:JanusVisionAlignerMLP.forward": ["ModelVisionAlignerMLP", "Module", "activation_fn", "class", "def", "fc1", "for", "forward", "hidden_layers", "hidden_states", "in", "layer", "nn", "return", "self"], "janus/modeling_janus.py:JanusVQVAEVectorQuantizer.__init__": ["Embedding", "ModelVQVAEConfig", "ModelVQVAEVectorQuantizer", "Module", "__init__", "beta", "class", "config", "def", "embed_dim", "embedding", "embedding_dim", "getattr", "nn", "num_embeddings", "num_patches", "quant_state_dims", "self", "super"], "janus/modeling_janus.py:JanusVQVAEVectorQuantizer.forward": ["ModelVQVAEVectorQuantizer", "Module", "Tensor", "True", "argmin", "bd", "beta", "bn", "class", "contiguous", "def", "detach", "dim", "distances", "dn", "einsum", "embedding", "embedding_dim", "forward", "hidden_state", "hidden_state_flattened", "hidden_state_quant", "keepdim", "loss", "mean", "min_encoding_indices", "nn", "permute", "return", "self", "shape", "sum", "torch", "transpose", "view", "weight"], "janus/modeling_janus.py:JanusVQVAEVectorQuantizer.get_codebook_entry": ["F", "FloatTensor", "LongTensor", "ModelVQVAEVectorQuantizer", "Module", "batch_size", "class", "contiguous", "def", "dim", "emb_dim", "embedding", "get_codebook_entry", "hidden_state_quant", "image_tokens", "int", "nn", "normalize", "p", "permute", "quant_state_dims", "return", "self", "shape", "torch", "view", "weight"], "janus/modeling_janus.py:JanusVQVAEResnetBlock.__init__": ["Conv2d", "Dropout", "False", "GroupNorm", "ModelVQVAEResnetBlock", "Module", "None", "True", "__init__", "affine", "class", "config", "conv1", "conv2", "conv_shortcut", "def", "dropout", "else", "eps", "if", "in_channels", "is", "kernel_size", "nin_shortcut", "nn", "norm1", "norm2", "num_channels", "num_groups", "out_channels", "padding", "self", "stride", "super", "torch", "use_conv_shortcut"], "janus/modeling_janus.py:JanusVQVAEResnetBlock.forward": ["ModelVQVAEResnetBlock", "Module", "class", "conv1", "conv2", "conv_shortcut", "def", "dropout", "else", "forward", "hidden_states", "if", "in_channels", "nin_shortcut", "nn", "norm1", "norm2", "out_channels", "residual", "return", "self", "sigmoid", "torch", "use_conv_shortcut"], "janus/modeling_janus.py:JanusVQVAEAttnBlock.__init__": ["Conv2d", "GroupNorm", "ModelVQVAEAttnBlock", "Module", "True", "__init__", "affine", "class", "def", "eps", "in_channels", "k", "kernel_size", "nn", "norm", "num_channels", "num_groups", "padding", "proj_out", "q", "self", "stride", "super", "torch", "v"], "janus/modeling_janus.py:JanusVQVAEAttnBlock.forward": ["F", "ModelVQVAEAttnBlock", "Module", "attn_output", "attn_weights", "batch_size", "bmm", "channels", "class", "def", "dim", "forward", "height", "hidden_states", "int", "k", "key_states", "nn", "norm", "permute", "proj_out", "q", "query_states", "reshape", "residual", "return", "self", "shape", "softmax", "torch", "v", "value_states", "width"], "janus/modeling_janus.py:JanusVQVAEConvDownsample.__init__": ["Conv2d", "ModelVQVAEConvDownsample", "Module", "__init__", "class", "conv", "def", "in_channels", "kernel_size", "nn", "padding", "self", "stride", "super"], "janus/modeling_janus.py:JanusVQVAEConvDownsample.forward": ["F", "ModelVQVAEConvDownsample", "Module", "class", "constant", "conv", "def", "forward", "hidden_states", "mode", "nn", "pad", "return", "self", "value"], "janus/modeling_janus.py:JanusVQVAEConvUpsample.__init__": ["Conv2d", "ModelVQVAEConvUpsample", "Module", "__init__", "class", "conv", "def", "in_channels", "kernel_size", "nn", "padding", "self", "stride", "super", "torch"], "janus/modeling_janus.py:JanusVQVAEConvUpsample.forward": ["F", "ModelVQVAEConvUpsample", "Module", "class", "conv", "def", "forward", "hidden_states", "interpolate", "mode", "nearest", "nn", "return", "scale_factor", "self"], "janus/modeling_janus.py:JanusVQVAEMidBlock.__init__": ["ModelVQVAEAttnBlock", "ModelVQVAEConfig", "ModelVQVAEMidBlock", "ModelVQVAEResnetBlock", "Module", "__init__", "attn_1", "block_1", "block_2", "channels", "class", "config", "def", "in_channels", "int", "nn", "out_channels", "self", "super"], "janus/modeling_janus.py:JanusVQVAEMidBlock.forward": ["ModelVQVAEMidBlock", "Module", "Tensor", "attn_1", "block_1", "block_2", "class", "def", "forward", "hidden_states", "nn", "return", "self", "torch"], "janus/modeling_janus.py:JanusVQVAEEncoder.__init__": ["Conv2d", "GroupNorm", "ModelVQVAEAttnBlock", "ModelVQVAEConvDownsample", "ModelVQVAEEncoder", "ModelVQVAEMidBlock", "ModelVQVAEResnetBlock", "Module", "ModuleList", "True", "__init__", "affine", "append", "attn", "base_channels", "block", "block_in", "block_out", "channel_multiplier", "class", "config", "conv_in", "conv_out", "def", "double_latent", "down", "downsample", "else", "eps", "for", "i_block", "i_level", "if", "in", "in_channel_multiplier", "in_channels", "kernel_size", "latent_channels", "len", "mid", "nn", "norm_out", "num_channels", "num_groups", "num_res_blocks", "num_resolutions", "out_channels", "padding", "range", "self", "stride", "super", "torch", "tuple"], "janus/modeling_janus.py:JanusVQVAEEncoder.forward": ["LongTensor", "ModelVQVAEEncoder", "Module", "append", "attn", "block", "class", "conv_in", "conv_out", "def", "down", "downsample", "for", "forward", "hidden_state", "hidden_states", "i_block", "i_level", "if", "in", "last_hidden_state", "len", "mid", "nn", "norm_out", "num_res_blocks", "num_resolutions", "pixel_values", "range", "return", "self", "sigmoid", "torch"], "janus/modeling_janus.py:JanusVQVAEDecoder.__init__": ["Conv2d", "GroupNorm", "ModelVQVAEAttnBlock", "ModelVQVAEConvUpsample", "ModelVQVAEDecoder", "ModelVQVAEMidBlock", "ModelVQVAEResnetBlock", "Module", "ModuleList", "True", "__init__", "affine", "append", "attn", "base_channels", "block", "block_in", "block_out", "channel_multiplier", "class", "config", "conv_in", "conv_out", "def", "eps", "for", "i_block", "i_level", "if", "in", "in_channels", "kernel_size", "latent_channels", "len", "mid", "nn", "norm_out", "num_channels", "num_groups", "num_res_blocks", "num_resolutions", "out_channels", "padding", "range", "reversed", "self", "stride", "super", "torch", "up", "upsample"], "janus/modeling_janus.py:JanusVQVAEDecoder.forward": ["FloatTensor", "ModelVQVAEDecoder", "Module", "attn", "block", "class", "conv_in", "conv_out", "def", "for", "forward", "hidden_state", "i_block", "i_level", "if", "in", "len", "mid", "nn", "norm_out", "num_res_blocks", "num_resolutions", "range", "return", "self", "sigmoid", "torch", "up", "upsample"], "janus/modeling_janus.py:JanusVQVAE.__init__": ["Conv2d", "False", "ModelPreTrainedModel", "ModelVQVAE", "ModelVQVAEConfig", "ModelVQVAEDecoder", "ModelVQVAEEncoder", "ModelVQVAEVectorQuantizer", "__init__", "class", "config", "decoder", "def", "embed_dim", "encoder", "eval", "gradient_checkpointing", "latent_channels", "nn", "post_init", "post_quant_conv", "quant_conv", "quantize", "self", "super", "torch"], "janus/modeling_janus.py:JanusVQVAE.encode": ["LongTensor", "ModelPreTrainedModel", "ModelVQVAE", "class", "def", "emb_loss", "encode", "encoder", "hidden_states", "indices", "pixel_values", "quant", "quant_conv", "quantize", "return", "self", "torch"], "janus/modeling_janus.py:JanusVQVAE.decode": ["Expected", "FloatTensor", "LongTensor", "ModelPreTrainedModel", "ModelVQVAE", "ValueError", "batch_size", "but", "class", "codebook_entry", "decode", "decoder", "def", "f", "get_codebook_entry", "got", "have", "hidden_states", "if", "image_tokens", "pixel_values", "post_quant_conv", "quant_state_dims", "quantize", "raise", "return", "self", "shape", "to", "torch"], "janus/modeling_janus.py:JanusVQVAE.forward": ["FloatTensor", "ModelPreTrainedModel", "ModelVQVAE", "ModelVQVAEOutput", "batch_size", "class", "decode", "decoded_pixel_values", "def", "embedding_loss", "encode", "forward", "indices", "kwargs", "pixel_values", "quant", "return", "self", "shape", "torch", "tuple", "view"], "janus/modeling_janus.py:JanusVQVAEAlignerMLP.__init__": ["ACT2FN", "Linear", "ModelVQVAEAlignerMLP", "ModelVQVAEConfig", "Module", "ModuleList", "_", "__init__", "activation_fn", "class", "config", "def", "embed_dim", "fc1", "for", "hidden_act", "hidden_layers", "in", "nn", "num_hidden_layers", "projection_dim", "range", "self", "super"], "janus/modeling_janus.py:JanusVQVAEAlignerMLP.forward": ["ModelVQVAEAlignerMLP", "Module", "activation_fn", "class", "def", "fc1", "for", "forward", "hidden_layers", "hidden_states", "in", "layer", "nn", "return", "self"], "janus/modeling_janus.py:JanusVQVAEHead.__init__": ["ACT2FN", "Linear", "ModelVQVAEConfig", "ModelVQVAEHead", "Module", "__init__", "activation_fn", "class", "config", "def", "hidden_act", "image_token_embed_dim", "nn", "num_embeddings", "proj_out", "projection_dim", "self", "super", "vision_head"], "janus/modeling_janus.py:JanusVQVAEHead.forward": ["ModelVQVAEHead", "Module", "Tensor", "activation_fn", "class", "def", "forward", "hidden_states", "nn", "proj_out", "return", "self", "tensor", "torch", "vision_head"], "janus/modeling_janus.py:JanusModel.__init__": ["AutoModel", "Embedding", "False", "ModelConfig", "ModelModel", "ModelPreTrainedModel", "ModelVQVAE", "ModelVQVAEAlignerMLP", "ModelVQVAEHead", "ModelVisionAlignerMLP", "ModelVisionModel", "__init__", "_from_config", "aligner", "class", "config", "def", "embed_dim", "from_config", "generation_aligner", "generation_embeddings", "generation_head", "gradient_checkpointing", "language_model", "nn", "num_embeddings", "post_init", "self", "super", "text_config", "vision_config", "vision_model", "vq_config", "vqmodel"], "janus/modeling_janus.py:JanusModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "language_model", "return", "self"], "janus/modeling_janus.py:JanusModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "language_model", "self", "set_input_embeddings", "value"], "janus/modeling_janus.py:JanusModel.get_image_features": ["ModelModel", "ModelPreTrainedModel", "aligner", "class", "def", "get_image_features", "image_embeds", "last_hidden_state", "pixel_values", "return", "self", "vision_model"], "janus/modeling_janus.py:JanusModel.get_placeholder_mask": ["FloatTensor", "Image", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "ValueError", "all", "and", "class", "config", "def", "device", "do", "dtype", "else", "expand_as", "f", "features", "get_input_embeddings", "get_placeholder_mask", "if", "image", "image_features", "image_token_id", "input_ids", "inputs_embeds", "is", "long", "match", "n_image_features", "n_image_tokens", "not", "numel", "raise", "return", "self", "shape", "special_image_mask", "sum", "tensor", "to", "tokens", "torch", "unsqueeze"], "janus/modeling_janus.py:JanusModel.forward": ["Cache", "FloatTensor", "LongTensor", "ModelBaseModelOutputWithPast", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "ValueError", "You", "and", "at", "attention_mask", "attentions", "bool", "both", "cache_position", "cannot", "class", "def", "device", "dtype", "either", "else", "forward", "get_image_features", "get_input_embeddings", "get_placeholder_mask", "hidden_states", "if", "image_attention_mask", "image_embeds", "image_features", "image_hidden_states", "input_ids", "inputs_embeds", "int", "is", "kwargs", "language_model", "last_hidden_state", "lm_output", "logits_to_keep", "masked_scatter", "must", "not", "one", "past_key_values", "pixel_values", "position_ids", "raise", "reshape", "return", "same", "self", "shape", "specify", "the", "time", "to", "torch", "use_cache"], "janus/modeling_janus.py:JanusForConditionalGeneration.__init__": ["False", "GenerationMixin", "Linear", "ModelConfig", "ModelForConditionalGeneration", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "text_config", "vocab_size"], "janus/modeling_janus.py:JanusForConditionalGeneration.get_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "language_model", "model", "return", "self"], "janus/modeling_janus.py:JanusForConditionalGeneration.set_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "language_model", "model", "self", "set_input_embeddings", "value"], "janus/modeling_janus.py:JanusForConditionalGeneration.prepare_embeddings_for_image_generation": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "Tensor", "class", "def", "generation_aligner", "generation_embeddings", "hidden_state", "inputs", "model", "prepare_embeddings_for_image_generation", "return", "self", "torch"], "janus/modeling_janus.py:JanusForConditionalGeneration.forward": ["Cache", "FloatTensor", "GenerationMixin", "LongTensor", "ModelCausalLMOutputWithPast", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "else", "forward", "hidden_states", "if", "image_hidden_states", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "outputs", "past_key_values", "pixel_values", "position_ids", "r", "return", "self", "slice", "slice_indices", "text_config", "torch", "use_cache", "vocab_size"], "janus/modeling_janus.py:JanusForConditionalGeneration.prepare_inputs_for_generation": ["False", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "True", "attention_mask", "cache_position", "class", "def", "get", "if", "input_ids", "inputs_embeds", "is_first_iteration", "kwargs", "logits_to_keep", "model_inputs", "not", "or", "past_key_values", "pixel_values", "prepare_inputs_for_generation", "return", "self", "super", "use_cache"], "janus/modeling_janus.py:JanusForConditionalGeneration.decode_image_tokens": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "Tensor", "class", "decode", "decode_image_tokens", "decoded_image", "def", "image_tokens", "model", "permute", "return", "self", "torch", "vqmodel"], "janus/modeling_janus.py:JanusForConditionalGeneration.generate": ["CFG", "ClassifierFreeGuidanceLogitsProcessor", "Ensure", "Expected", "GREEDY_SEARCH", "GenerateDecoderOnlyOutput", "Generation", "GenerationMixin", "GenerationMode", "Got", "Image", "LogitsProcessorList", "LongTensor", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Passing", "SAMPLE", "Setting", "Tensor", "True", "ValueError", "_expand_inputs_for_generation", "_get_cache", "_get_initial_cache_position", "_get_logits_processor", "_prepare_model_inputs", "_prepare_special_tokens", "_update_model_kwargs_for_generation", "_validate_model_kwargs", "activated", "and", "append", "argmax", "attention_mask", "attentions", "batch_size", "be", "beam", "boi_token_id", "bos_token_id", "but", "by", "cache_implementation", "cache_position", "cat", "class", "clone", "config", "copy", "currently", "de", "decoder_attentions", "decoder_hidden_states", "deepcopy", "def", "default", "device", "dim", "do_sample", "dtype", "else", "embeds", "encoder_input_ids", "expand_size", "f", "float", "for", "generate", "generated_tokens", "generation_config", "generation_head", "generation_kwargs", "generation_mode", "get", "get_generation_mode", "get_input_embeddings", "got", "greedy", "guidance_scale", "hidden_state", "hidden_states", "i", "ids", "if", "in", "incompatible", "input", "input_ids", "input_ids_seq_length", "input_tokens", "inputs", "inputs_embeds", "is", "kwargs", "kwargs_has_attention_mask", "language_model", "last_hidden_state", "len", "logger", "logits", "logits_processor", "mask", "masked_fill_", "max", "max_cache_len", "max_length", "mode", "model", "model_input_name", "model_inputs", "model_kwargs", "multinomial", "next_token", "next_token_scores", "not", "num_beams", "num_image_tokens", "num_return_sequences", "num_samples", "of", "one", "or", "output_attentions", "output_hidden_states", "output_logits", "output_scores", "outputs", "pad_token_id", "past_key_values", "pop", "prefix_allowed_tokens_fn", "prepare_embeddings_for_image_generation", "prepare_inputs_for_generation", "probs", "provided", "raise", "range", "raw_logits", "raw_scores", "repeat", "required", "return", "return_dict_in_generate", "sampling", "scores", "search", "self", "seq_len", "sequences", "setting", "shape", "should", "softmax", "squeeze", "static", "super", "supported", "text", "that", "to", "torch", "unsqueeze", "update", "use_cache", "validate", "value", "vision_model", "warning", "zeros"], "jetmoe/modeling_jetmoe.py:JetMoeRMSNorm.__init__": ["ModelRMSNorm", "Module", "Parameter", "__init__", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "jetmoe/modeling_jetmoe.py:JetMoeRMSNorm.forward": ["ModelRMSNorm", "Module", "True", "class", "def", "dtype", "float32", "forward", "hidden_states", "input_dtype", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "jetmoe/modeling_jetmoe.py:JetMoeRMSNorm.extra_repr": ["ModelRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "variance_epsilon", "weight"], "jetmoe/modeling_jetmoe.py:JetMoeRotaryEmbedding.__init__": ["Callable", "False", "ModelConfig", "ModelRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "jetmoe/modeling_jetmoe.py:JetMoeRotaryEmbedding.compute_default_rope_parameters": ["ModelConfig", "ModelRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "jetmoe/modeling_jetmoe.py:JetMoeRotaryEmbedding.forward": ["False", "ModelRotaryEmbedding", "Module", "None", "and", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "jetmoe/modeling_jetmoe.py:JetMoeParallelExperts.__init__": ["ModelParallelExperts", "Module", "None", "Parameter", "__init__", "class", "def", "empty", "input_size", "int", "nn", "num_experts", "output_size", "self", "super", "torch", "weight"], "jetmoe/modeling_jetmoe.py:JetMoeParallelExperts.forward": ["F", "ModelParallelExperts", "Module", "append", "cat", "class", "def", "dim", "expert_size", "for", "forward", "i", "in", "input_list", "inputs", "linear", "nn", "num_experts", "output_list", "range", "results", "return", "self", "split", "torch", "weight"], "jetmoe/modeling_jetmoe.py:JetMoeTopKGating.__init__": ["False", "Linear", "ModelTopKGating", "Module", "__init__", "bias", "class", "def", "input_size", "int", "layer", "nn", "num_experts", "self", "super", "top_k"], "jetmoe/modeling_jetmoe.py:JetMoeTopKGating.forward": ["ModelTopKGating", "Module", "_", "batch_gates", "batch_index", "class", "def", "device", "dim", "div", "dtype", "expert_size", "flatten", "float", "forward", "gates", "hidden_states", "index_sorted_experts", "layer", "logits", "long", "nn", "num_experts", "return", "rounding_mode", "scatter", "self", "size", "softmax", "sort", "sum", "tolist", "top_k", "top_k_experts", "top_k_gates", "top_k_indices", "top_k_logits", "topk", "torch", "trunc", "type_as", "zeros"], "jetmoe/modeling_jetmoe.py:JetMoeMoE.__init__": ["ACT2FN", "ModelConfig", "ModelMoE", "ModelParallelExperts", "ModelTopKGating", "Module", "Parameter", "__init__", "activation", "activation_function", "bias", "class", "config", "def", "empty", "hidden_size", "input_linear", "input_size", "intermediate_size", "nn", "num_experts", "num_experts_per_tok", "num_local_experts", "output_linear", "router", "self", "super", "top_k", "torch"], "jetmoe/modeling_jetmoe.py:JetMoeMoE.forward": ["ModelMoE", "Module", "None", "_", "activation", "batch_gates", "batch_index", "bias", "bsz", "chunk", "chunked_hidden_states", "class", "def", "device", "dim", "dtype", "emb_size", "expert_inputs", "expert_outputs", "expert_size", "forward", "hidden_states", "index_add", "input_linear", "input_size", "layer_input", "layer_output", "length", "nn", "output_linear", "reshape", "return", "router", "router_logits", "self", "size", "torch", "view", "zeros"], "jetmoe/modeling_jetmoe.py:JetMoeMoA.__init__": ["ModelConfig", "ModelMoA", "ModelParallelExperts", "ModelTopKGating", "Module", "Parameter", "__init__", "bias", "class", "config", "def", "empty", "hidden_size", "input_linear", "input_size", "kv_channels", "nn", "num_experts", "num_experts_per_tok", "num_key_value_heads", "num_local_experts", "output_linear", "router", "self", "super", "top_k", "torch"], "jetmoe/modeling_jetmoe.py:JetMoeMoA.map": ["ModelMoA", "Module", "batch_gates", "batch_index", "bsz", "class", "def", "device", "dtype", "emb_size", "expert_inputs", "expert_outputs", "expert_size", "hidden_size", "index_add", "index_sorted_experts", "input_linear", "layer_input", "layer_output", "length", "map", "nn", "reshape", "return", "router", "router_logits", "self", "size", "top_k", "topo_info", "torch", "view", "zeros"], "jetmoe/modeling_jetmoe.py:JetMoeMoA.reduce": ["ModelMoA", "Module", "None", "batch_gates", "batch_index", "bias", "bsz", "class", "def", "device", "dtype", "expert_inputs", "expert_outputs", "expert_size", "hidden_size", "index_add", "index_sorted_experts", "input_size", "k", "layer_input", "layer_output", "length", "nn", "output_linear", "reduce", "reshape", "return", "self", "size", "topo_info", "torch", "view", "zeros"], "jetmoe/modeling_jetmoe.py:JetMoeMoA.forward": ["ModelMoA", "Module", "NotImplementedError", "This", "and", "call", "class", "def", "doesn", "forward", "layer_input", "module", "nn", "raise", "self", "support", "t"], "jetmoe/modeling_jetmoe.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "jetmoe/modeling_jetmoe.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cos", "def", "k", "k_embed", "q", "q_embed", "return", "rotate_half", "sin", "unsqueeze", "unsqueeze_dim"], "jetmoe/modeling_jetmoe.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "jetmoe/modeling_jetmoe.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value", "value_states"], "jetmoe/modeling_jetmoe.py:JetMoeAttention.__init__": ["False", "Instantiating", "Linear", "ModelAttention", "ModelConfig", "ModelMoA", "Module", "None", "Please", "True", "__class__", "__init__", "__name__", "a", "and", "attention_dropout", "bias", "caching", "call", "class", "config", "creating", "def", "during", "errors", "experts", "f", "forward", "head_dim", "hidden_size", "if", "int", "is", "is_causal", "kv_channels", "kv_proj", "kv_projection_size", "layer_idx", "lead", "logger", "make", "nn", "not", "num_attention_heads", "num_experts_per_tok", "num_heads", "num_key_value_groups", "num_key_value_heads", "passing", "provide", "recommended", "scaling", "self", "super", "sure", "the", "this", "to", "top_k", "torch", "used", "warning_once", "when", "will", "without"], "jetmoe/modeling_jetmoe.py:JetMoeAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "LongTensor", "ModelAttention", "Module", "None", "Tensor", "_attn_implementation", "apply_rotary_pos_emb", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "cache_kwargs", "cache_position", "chunk", "class", "config", "cos", "def", "dim", "dropout", "eager", "eager_attention_forward", "else", "experts", "forward", "head_dim", "hidden_shape", "hidden_states", "if", "input_shape", "is", "key_states", "kv_proj", "kwargs", "layer_idx", "map", "nn", "not", "past_key_values", "position_embeddings", "query_states", "reduce", "repeat", "return", "router_logits", "scaling", "self", "shape", "sin", "top_k", "topo_info", "torch", "training", "transpose", "tuple", "update", "value_states", "view"], "jetmoe/modeling_jetmoe.py:JetMoeDecoderLayer.__init__": ["GradientCheckpointingLayer", "ModelAttention", "ModelConfig", "ModelDecoderLayer", "ModelMoE", "ModelRMSNorm", "None", "__init__", "class", "config", "def", "hidden_size", "input_layernorm", "int", "layer_idx", "mlp", "post_attention_layernorm", "self", "self_attention", "super"], "jetmoe/modeling_jetmoe.py:JetMoeDecoderLayer.forward": ["Cache", "False", "GradientCheckpointingLayer", "LongTensor", "ModelDecoderLayer", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "attention_mask", "bool", "cache_position", "class", "def", "forward", "hidden_states", "input_layernorm", "kwargs", "mlp", "past_key_values", "position_embeddings", "position_ids", "post_attention_layernorm", "residual", "return", "self", "self_attention", "torch", "tuple", "use_cache"], "jetmoe/modeling_jetmoe.py:JetMoePreTrainedModel._init_weights": ["ModelMoA", "ModelMoE", "ModelParallelExperts", "ModelPreTrainedModel", "PreTrainedModel", "_init_weights", "bias", "class", "config", "def", "elif", "if", "init", "initializer_range", "isinstance", "mean", "module", "normal_", "self", "std", "super", "weight", "zeros_"], "jetmoe/modeling_jetmoe.py:JetMoeModel.__init__": ["Embedding", "False", "ModelConfig", "ModelDecoderLayer", "ModelModel", "ModelPreTrainedModel", "ModelRMSNorm", "ModelRotaryEmbedding", "ModuleList", "__init__", "_attn_implementation", "class", "config", "def", "embed_tokens", "eps", "for", "gradient_checkpointing", "hidden_size", "in", "layer_idx", "layers", "nn", "norm", "num_hidden_layers", "pad_token_id", "padding_idx", "post_init", "range", "rms_norm_eps", "rotary_emb", "self", "super", "vocab_size"], "jetmoe/modeling_jetmoe.py:JetMoeModel.forward": ["Cache", "DynamicCache", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "MoeModelOutputWithPast", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "and", "arange", "attention_mask", "bool", "cache_position", "causal_mask", "class", "config", "create_causal_mask", "decoder_layer", "def", "device", "else", "embed_tokens", "exactly", "for", "forward", "get_seq_length", "hidden_states", "if", "in", "input_embeds", "input_ids", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layers", "must", "norm", "not", "num_hidden_layers", "of", "one", "or", "past_key_values", "past_seen_tokens", "position_embeddings", "position_ids", "raise", "return", "rotary_emb", "self", "shape", "specify", "torch", "unsqueeze", "use_cache"], "jetmoe/modeling_jetmoe.py:load_balancing_loss_func": ["Model_balancing_loss_func", "None", "Tensor", "_", "attention_mask", "batch_size", "cat", "compute_device", "concatenated_gate_logits", "def", "device", "dim", "else", "expand", "expert_attention_mask", "expert_mask", "float", "for", "functional", "gate_logits", "if", "in", "int", "is", "isinstance", "layer_gate", "mean", "nn", "not", "num_experts", "num_hidden_layers", "one_hot", "or", "overall_loss", "r", "reshape", "return", "router_per_expert_attention_mask", "router_prob_per_expert", "routing_weights", "selected_experts", "sequence_length", "shape", "softmax", "sum", "to", "tokens_per_expert", "top_k", "topk", "torch", "tuple", "unsqueeze"], "jetmoe/modeling_jetmoe.py:JetMoeForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "ModelForCausalLM", "ModelModel", "ModelPreTrainedModel", "__init__", "aux_loss_coef", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "tie_word_embeddings", "vocab_size"], "jetmoe/modeling_jetmoe.py:JetMoeForCausalLM.forward": ["Cache", "False", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForCausalLM", "ModelPreTrainedModel", "MoeCausalLMOutputWithPast", "MoeModelOutputWithPast", "None", "Tensor", "attention_mask", "attentions", "aux_loss", "aux_loss_coef", "bool", "cache_position", "class", "config", "def", "device", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "load_balancing_loss_func", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "num_experts", "num_experts_per_tok", "output_router_logits", "outputs", "past_key_values", "position_ids", "return", "router_logits", "self", "slice", "slice_indices", "to", "torch", "use_cache", "vocab_size"], "kosmos2/modeling_kosmos2.py:_expand_mask": ["None", "Tensor", "_expand_mask", "bool", "bsz", "def", "dtype", "else", "expand", "expanded_mask", "finfo", "if", "int", "inverted_mask", "is", "mask", "masked_fill", "min", "not", "return", "size", "src_len", "tgt_len", "to", "torch"], "kosmos2/modeling_kosmos2.py:_make_causal_mask": ["None", "Size", "_make_causal_mask", "arange", "bsz", "cat", "def", "device", "dim", "dtype", "expand", "finfo", "full", "if", "input_ids_shape", "int", "mask", "mask_cond", "masked_fill_", "min", "past_key_values_length", "return", "size", "tgt_len", "to", "torch", "view", "zeros"], "kosmos2/modeling_kosmos2.py:Kosmos2ModelOutput.to_tuple": ["Any", "ModelModelOutput", "ModelOutput", "class", "def", "else", "for", "getattr", "if", "in", "k", "keys", "not", "return", "self", "text_model_output", "to_tuple", "tuple", "vision_model_output"], "kosmos2/modeling_kosmos2.py:Kosmos2ForConditionalGenerationModelOutput.to_tuple": ["Any", "ModelForConditionalGenerationModelOutput", "ModelOutput", "class", "def", "else", "for", "getattr", "if", "in", "k", "keys", "not", "return", "self", "text_model_output", "to_tuple", "tuple", "vision_model_output"], "kosmos2/modeling_kosmos2.py:Kosmos2VisionEmbeddings.__init__": ["Conv2d", "Embedding", "False", "ModelVisionConfig", "ModelVisionEmbeddings", "Module", "Parameter", "__init__", "arange", "bias", "class", "class_embedding", "config", "def", "embed_dim", "expand", "hidden_size", "image_size", "in_channels", "kernel_size", "nn", "num_channels", "num_patches", "num_positions", "out_channels", "patch_embedding", "patch_size", "persistent", "position_embedding", "position_ids", "randn", "register_buffer", "self", "stride", "super", "torch"], "kosmos2/modeling_kosmos2.py:Kosmos2VisionEmbeddings.interpolate_pos_encoding": ["False", "ModelVisionEmbeddings", "Module", "Tensor", "align_corners", "and", "bicubic", "cat", "class", "class_pos_embed", "def", "dim", "embeddings", "functional", "height", "if", "int", "interpolate", "interpolate_pos_encoding", "is_tracing", "jit", "mode", "new_height", "new_width", "nn", "not", "num_patches", "num_positions", "patch_pos_embed", "patch_size", "permute", "position_embedding", "position_ids", "reshape", "return", "self", "shape", "size", "sqrt_num_positions", "torch", "torch_int", "unsqueeze", "view", "weight", "width"], "kosmos2/modeling_kosmos2.py:Kosmos2VisionEmbeddings.forward": ["False", "FloatTensor", "Input", "ModelVisionEmbeddings", "Module", "Tensor", "ValueError", "_", "and", "batch_size", "cat", "class", "class_embedding", "class_embeds", "def", "dim", "doesn", "dtype", "else", "embeddings", "expand", "f", "flatten", "forward", "height", "if", "image", "image_size", "interpolate_pos_encoding", "match", "model", "nn", "not", "or", "patch_embedding", "patch_embeds", "pixel_values", "position_embedding", "position_ids", "raise", "return", "self", "shape", "size", "t", "target_dtype", "to", "torch", "transpose", "weight", "width"], "kosmos2/modeling_kosmos2.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "attention_mask", "attn_output", "attn_weights", "contiguous", "def", "dim", "dropout", "float", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "softmax", "torch", "training", "transpose", "value"], "kosmos2/modeling_kosmos2.py:Kosmos2VisionAttention.__init__": ["False", "Linear", "ModelVisionAttention", "Module", "ValueError", "__init__", "and", "attention_dropout", "be", "by", "class", "config", "def", "divisible", "dropout", "embed_dim", "f", "got", "head_dim", "hidden_size", "if", "is_causal", "k_proj", "must", "nn", "num_attention_heads", "num_heads", "out_proj", "q_proj", "raise", "scale", "self", "super", "v_proj"], "kosmos2/modeling_kosmos2.py:Kosmos2VisionAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "False", "ModelVisionAttention", "Module", "None", "Tensor", "_attn_implementation", "and", "attention_interface", "attention_mask", "attn_output", "attn_weights", "batch_size", "bool", "causal_attention_mask", "class", "config", "contiguous", "def", "dropout", "eager", "eager_attention_forward", "elif", "else", "embed_dim", "flash_attention_2", "forward", "head_dim", "hidden_states", "if", "is", "is_causal", "k_proj", "keys", "nn", "not", "num_heads", "out_proj", "output_attentions", "q_proj", "queries", "reshape", "return", "scale", "scaling", "self", "seq_length", "shape", "torch", "training", "transpose", "tuple", "v_proj", "values", "view"], "kosmos2/modeling_kosmos2.py:Kosmos2VisionMLP.__init__": ["ACT2FN", "Linear", "ModelVisionMLP", "Module", "__init__", "activation_fn", "class", "config", "def", "fc1", "fc2", "hidden_act", "hidden_size", "intermediate_size", "nn", "self", "super"], "kosmos2/modeling_kosmos2.py:Kosmos2VisionMLP.forward": ["ModelVisionMLP", "Module", "Tensor", "activation_fn", "class", "def", "fc1", "fc2", "forward", "hidden_states", "nn", "return", "self", "torch"], "kosmos2/modeling_kosmos2.py:Kosmos2VisionEncoderLayer.__init__": ["GradientCheckpointingLayer", "LayerNorm", "ModelVisionAttention", "ModelVisionConfig", "ModelVisionEncoderLayer", "ModelVisionMLP", "__init__", "class", "config", "def", "embed_dim", "eps", "hidden_size", "layer_norm1", "layer_norm2", "layer_norm_eps", "mlp", "nn", "self", "self_attn", "super"], "kosmos2/modeling_kosmos2.py:Kosmos2VisionEncoderLayer.forward": ["False", "FloatTensor", "GradientCheckpointingLayer", "ModelVisionEncoderLayer", "None", "Tensor", "attention_mask", "attn_weights", "bool", "causal_attention_mask", "class", "def", "forward", "hidden_states", "if", "layer_norm1", "layer_norm2", "mlp", "output_attentions", "outputs", "residual", "return", "self", "self_attn", "torch", "tuple"], "kosmos2/modeling_kosmos2.py:Kosmos2VisionEncoder.__init__": ["False", "ModelVisionConfig", "ModelVisionEncoder", "ModelVisionEncoderLayer", "Module", "ModuleList", "_", "__init__", "class", "config", "def", "for", "gradient_checkpointing", "in", "layers", "nn", "num_hidden_layers", "range", "self", "super"], "kosmos2/modeling_kosmos2.py:Kosmos2VisionEncoder.forward": ["BaseModelOutput", "ModelVisionEncoder", "Module", "None", "Tensor", "all_attentions", "attention_mask", "attentions", "bool", "causal_attention_mask", "class", "config", "def", "else", "encoder_layer", "encoder_states", "enumerate", "for", "forward", "hidden_states", "idx", "if", "in", "inputs_embeds", "is", "last_hidden_state", "layer_outputs", "layers", "nn", "not", "output_attentions", "output_hidden_states", "r", "return", "return_dict", "self", "torch", "tuple", "use_return_dict"], "kosmos2/modeling_kosmos2.py:Kosmos2VisionTransformer.__init__": ["LayerNorm", "ModelVisionConfig", "ModelVisionEmbeddings", "ModelVisionEncoder", "ModelVisionTransformer", "Module", "__init__", "class", "config", "def", "embed_dim", "embeddings", "encoder", "eps", "hidden_size", "layer_norm_eps", "nn", "post_layernorm", "pre_layrnorm", "self", "super"], "kosmos2/modeling_kosmos2.py:Kosmos2VisionTransformer.forward": ["BaseModelOutputWithPooling", "False", "FloatTensor", "ModelVisionTransformer", "Module", "None", "ValueError", "You", "attentions", "bool", "class", "config", "def", "else", "embeddings", "encoder", "encoder_outputs", "forward", "have", "hidden_states", "if", "inputs_embeds", "interpolate_pos_encoding", "is", "last_hidden_state", "nn", "not", "output_attentions", "output_hidden_states", "pixel_values", "pooled_output", "pooler_output", "post_layernorm", "pre_layrnorm", "raise", "return", "return_dict", "self", "specify", "to", "torch", "tuple", "use_return_dict"], "kosmos2/modeling_kosmos2.py:Kosmos2TextSinusoidalPositionalEmbedding.__init__": ["ModelTextSinusoidalPositionalEmbedding", "Module", "None", "__init__", "class", "def", "embedding_dim", "int", "make_weights", "nn", "num_positions", "offset", "padding_idx", "self", "super"], "kosmos2/modeling_kosmos2.py:Kosmos2TextSinusoidalPositionalEmbedding.make_weights": ["False", "ModelTextSinusoidalPositionalEmbedding", "Module", "None", "class", "def", "device", "dtype", "emb_weights", "embedding_dim", "get_embedding", "hasattr", "if", "int", "make_weights", "nn", "num_embeddings", "padding_idx", "persistent", "register_buffer", "self", "to", "weights"], "kosmos2/modeling_kosmos2.py:Kosmos2TextSinusoidalPositionalEmbedding.get_embedding": ["ModelTextSinusoidalPositionalEmbedding", "Module", "None", "arange", "cat", "class", "cos", "def", "dim", "dtype", "emb", "embedding_dim", "exp", "float", "get_default_dtype", "get_embedding", "half_dim", "if", "int", "int64", "is", "log", "math", "nn", "not", "num_embeddings", "padding_idx", "return", "sin", "to", "torch", "unsqueeze", "view", "zeros"], "kosmos2/modeling_kosmos2.py:Kosmos2TextSinusoidalPositionalEmbedding.forward": ["ModelTextSinusoidalPositionalEmbedding", "Module", "None", "Tensor", "bsz", "class", "create_position_ids_from_input_ids", "create_position_ids_from_inputs_embeds", "def", "detach", "device", "else", "embedding_dim", "forward", "if", "index_select", "input_ids", "inputs_embeds", "int", "is", "make_weights", "max_pos", "nn", "not", "offset", "padding_idx", "past_key_values_length", "position_ids", "return", "self", "seq_len", "shape", "size", "to", "torch", "view", "weights"], "kosmos2/modeling_kosmos2.py:Kosmos2TextSinusoidalPositionalEmbedding.create_position_ids_from_inputs_embeds": ["ModelTextSinusoidalPositionalEmbedding", "Module", "arange", "class", "contiguous", "create_position_ids_from_inputs_embeds", "def", "device", "dtype", "expand", "input_shape", "inputs_embeds", "long", "nn", "padding_idx", "past_key_values_length", "position_ids", "return", "sequence_length", "size", "torch", "unsqueeze"], "kosmos2/modeling_kosmos2.py:Kosmos2TextSinusoidalPositionalEmbedding.create_position_ids_from_input_ids": ["ModelTextSinusoidalPositionalEmbedding", "Module", "class", "create_position_ids_from_input_ids", "cumsum", "def", "dim", "incremental_indices", "input_ids", "int", "long", "mask", "ne", "nn", "padding_idx", "past_key_values_length", "return", "torch", "type_as"], "kosmos2/modeling_kosmos2.py:KosmosTextAttention.__init__": ["False", "LayerNorm", "Linear", "ModelTextAttention", "Module", "None", "True", "ValueError", "__init__", "add_inner_attn_layernorm", "and", "be", "bias", "bool", "by", "class", "config", "def", "divisible", "dropout", "embed_dim", "eps", "f", "float", "got", "head_dim", "if", "inner_attn_ln", "int", "is_causal", "is_decoder", "k_proj", "layer_idx", "layer_norm_eps", "must", "nn", "num_heads", "out_proj", "q_proj", "raise", "scaling", "self", "super", "v_proj"], "kosmos2/modeling_kosmos2.py:KosmosTextAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "EncoderDecoderCache", "False", "ModelTextAttention", "Module", "None", "Tensor", "True", "_attn_implementation", "and", "attention_interface", "attention_mask", "attn_output", "attn_weights", "batch_size", "bool", "cache_position", "class", "config", "contiguous", "cross_attention_cache", "curr_past_key_values", "current_states", "def", "dropout", "eager", "eager_attention_forward", "else", "encoder_hidden_states", "forward", "get", "head_dim", "hidden_states", "if", "inner_attn_ln", "is", "is_cross_attention", "is_updated", "isinstance", "k_proj", "key_states", "keys", "kwargs", "layer_idx", "layers", "nn", "not", "num_heads", "out_proj", "output_attentions", "past_key_values", "q_proj", "query_states", "reshape", "return", "scaling", "self", "self_attention_cache", "seq_length", "shape", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "values", "view"], "kosmos2/modeling_kosmos2.py:Kosmos2TextFFN.__init__": ["ACT2FN", "LayerNorm", "Linear", "ModelTextConfig", "ModelTextFFN", "Module", "__init__", "activation_dropout", "activation_fn", "activation_function", "class", "config", "def", "dropout", "embed_dim", "eps", "fc1", "fc2", "ffn_dim", "ffn_layernorm", "layer_norm_eps", "nn", "self", "super"], "kosmos2/modeling_kosmos2.py:Kosmos2TextFFN.forward": ["ModelTextFFN", "Module", "activation_dropout", "activation_fn", "class", "def", "dropout", "fc1", "fc2", "ffn_layernorm", "forward", "functional", "hidden_states", "nn", "p", "return", "self", "training"], "kosmos2/modeling_kosmos2.py:Kosmos2TextBlock.__init__": ["False", "GradientCheckpointingLayer", "LayerNorm", "ModelTextAttention", "ModelTextBlock", "ModelTextConfig", "ModelTextFFN", "None", "True", "__init__", "add_cross_attention", "add_inner_attn_layernorm", "attention_dropout", "attention_heads", "class", "config", "def", "dropout", "embed_dim", "encoder_attn", "encoder_attn_layer_norm", "eps", "ffn", "final_layer_norm", "if", "is_decoder", "layer_idx", "layer_norm_eps", "nn", "num_heads", "self", "self_attn", "self_attn_layer_norm", "super"], "kosmos2/modeling_kosmos2.py:Kosmos2TextBlock.forward": ["Cache", "False", "FloatTensor", "GradientCheckpointingLayer", "If", "ModelTextBlock", "None", "Tensor", "True", "ValueError", "add_cross_attention", "are", "attention", "attention_mask", "be", "bool", "by", "cache_position", "class", "config", "cross", "cross_attn_weights", "def", "dropout", "encoder_attention_mask", "encoder_attn", "encoder_attn_layer_norm", "encoder_hidden_states", "f", "ffn", "final_layer_norm", "forward", "functional", "has", "hasattr", "hidden_states", "if", "instantiated", "is", "kwargs", "layers", "nn", "not", "output_attentions", "outputs", "p", "passed", "past_key_values", "raise", "residual", "return", "self", "self_attn", "self_attn_layer_norm", "self_attn_weights", "setting", "to", "torch", "training", "tuple", "use_cache", "with"], "kosmos2/modeling_kosmos2.py:Kosmos2TextTransformer.__init__": ["Embedding", "False", "LayerNorm", "ModelTextBlock", "ModelTextConfig", "ModelTextSinusoidalPositionalEmbedding", "ModelTextTransformer", "Module", "ModuleList", "__init__", "class", "config", "def", "dropout", "else", "embed_dim", "embed_positions", "embed_scale", "embed_tokens", "embedding_dim", "for", "gradient_checkpointing", "i", "if", "in", "layer_idx", "layer_norm", "layer_norm_eps", "layerdrop", "layers", "math", "max_position_embeddings", "nn", "num_positions", "pad_token_id", "padding_idx", "range", "scale_embedding", "self", "sqrt", "super", "vocab_size"], "kosmos2/modeling_kosmos2.py:Kosmos2TextTransformer._prepare_decoder_attention_mask": ["ModelTextTransformer", "Module", "None", "_expand_mask", "_make_causal_mask", "_prepare_decoder_attention_mask", "attention_mask", "class", "combined_attention_mask", "def", "device", "dtype", "else", "expanded_attn_mask", "if", "input_shape", "inputs_embeds", "is", "nn", "not", "past_key_values_length", "return", "self", "tgt_len", "to"], "kosmos2/modeling_kosmos2.py:Kosmos2TextTransformer.forward_embedding": ["ModelTextTransformer", "Module", "None", "Tensor", "bool", "class", "def", "device", "dropout", "dtype", "embed_positions", "embed_scale", "embed_tokens", "forward_embedding", "functional", "hidden_states", "if", "image_embeds", "img_input_mask", "input_ids", "inputs_embeds", "int", "is", "nn", "not", "p", "past_key_values_length", "position_ids", "positions", "return", "self", "size", "to", "torch", "training", "view"], "kosmos2/modeling_kosmos2.py:Kosmos2TextTransformer.forward": ["BaseModelOutputWithPastAndCrossAttentions", "Cache", "DynamicCache", "EncoderDecoderCache", "False", "FlashAttentionKwargs", "ModelTextTransformer", "Module", "None", "Setting", "Tensor", "True", "Unpack", "ValueError", "You", "_expand_mask", "_prepare_decoder_attention_mask", "all_cross_attentions", "all_hidden_states", "all_self_attns", "and", "at", "attention_mask", "attentions", "bool", "both", "cache_position", "cannot", "checkpointing", "class", "config", "continue", "cross_attentions", "decoder_layer", "def", "dropout", "dropout_probability", "dtype", "either", "elif", "else", "encoder_attention_mask", "encoder_hidden_states", "enumerate", "for", "forward", "forward_embedding", "functional", "get_seq_length", "gradient", "gradient_checkpointing", "have", "hidden_states", "idx", "if", "image_embeds", "image_embeds_position_mask", "img_input_mask", "in", "incompatible", "input_ids", "input_shape", "inputs_embeds", "is", "is_encoder_decoder", "kwargs", "last_hidden_state", "layer_norm", "layer_outputs", "layerdrop", "layers", "logger", "nn", "not", "or", "output_attentions", "output_hidden_states", "p", "past_key_values", "past_key_values_length", "position_ids", "raise", "rand", "return", "return_dict", "same", "self", "shape", "size", "specify", "tgt_len", "the", "time", "to", "torch", "training", "tuple", "use_cache", "view", "warning_once", "with"], "kosmos2/modeling_kosmos2.py:Kosmos2PreTrainedModel._init_weights": ["LayerNorm", "Linear", "ModelForConditionalGeneration", "ModelImageToTextProjection", "ModelModel", "ModelPreTrainedModel", "ModelTextAttention", "ModelTextFFN", "ModelTextForCausalLM", "ModelTextModel", "ModelTextSinusoidalPositionalEmbedding", "ModelTextTransformer", "ModelVisionAttention", "ModelVisionEmbeddings", "ModelVisionMLP", "ModelVisionModel", "Module", "None", "PreTrainedModel", "_init_weights", "and", "arange", "bias", "class", "class_embedding", "config", "copy_", "def", "dense", "elif", "emb_weights", "embed_dim", "embed_tokens", "embedding_dim", "expand", "factor", "fc1", "fc2", "fc_std", "get_embedding", "hidden_size", "if", "in_proj_std", "init", "init_std", "initializer_factor", "initializer_range", "is", "isinstance", "k_proj", "latent_query", "lm_head", "mean", "module", "nn", "normal_", "not", "num_hidden_layers", "num_positions", "offset", "ones_", "out_proj", "out_proj_std", "padding_idx", "patch_embedding", "position_embedding", "position_ids", "q_proj", "self", "shape", "std", "text_config", "torch", "v_proj", "vision_config", "weight", "weights", "zeros_"], "kosmos2/modeling_kosmos2.py:Kosmos2VisionModel.__init__": ["ModelPreTrainedModel", "ModelVisionConfig", "ModelVisionModel", "ModelVisionTransformer", "__init__", "class", "config", "def", "model", "post_init", "self", "super"], "kosmos2/modeling_kosmos2.py:Kosmos2VisionModel.get_input_embeddings": ["ModelPreTrainedModel", "ModelVisionModel", "Module", "class", "def", "embeddings", "get_input_embeddings", "model", "nn", "patch_embedding", "return", "self"], "kosmos2/modeling_kosmos2.py:Kosmos2VisionModel.forward": ["BaseModelOutputWithPooling", "False", "FloatTensor", "ModelPreTrainedModel", "ModelVisionModel", "None", "bool", "class", "def", "forward", "interpolate_pos_encoding", "kwargs", "model", "output_attentions", "output_hidden_states", "pixel_values", "return", "return_dict", "self", "torch", "tuple"], "kosmos2/modeling_kosmos2.py:Kosmos2TextModel.__init__": ["ModelPreTrainedModel", "ModelTextConfig", "ModelTextModel", "ModelTextTransformer", "__init__", "class", "config", "def", "model", "post_init", "self", "super"], "kosmos2/modeling_kosmos2.py:Kosmos2TextModel.get_input_embeddings": ["ModelPreTrainedModel", "ModelTextModel", "Module", "class", "def", "embed_tokens", "get_input_embeddings", "model", "nn", "return", "self"], "kosmos2/modeling_kosmos2.py:Kosmos2TextModel.forward": ["BaseModelOutputWithPastAndCrossAttentions", "Cache", "FlashAttentionKwargs", "ModelPreTrainedModel", "ModelTextModel", "None", "Tensor", "Unpack", "attention_mask", "bool", "cache_position", "class", "def", "encoder_attention_mask", "encoder_hidden_states", "forward", "image_embeds", "image_embeds_position_mask", "input_ids", "inputs_embeds", "kwargs", "model", "output_attentions", "output_hidden_states", "past_key_values", "position_ids", "r", "return", "return_dict", "self", "torch", "tuple", "use_cache"], "kosmos2/modeling_kosmos2.py:Kosmos2TextForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "ModelPreTrainedModel", "ModelTextConfig", "ModelTextForCausalLM", "ModelTextTransformer", "__init__", "bias", "class", "config", "def", "embed_dim", "in_features", "lm_head", "model", "nn", "out_features", "post_init", "self", "super", "vocab_size"], "kosmos2/modeling_kosmos2.py:Kosmos2TextForCausalLM.get_input_embeddings": ["GenerationMixin", "ModelPreTrainedModel", "ModelTextForCausalLM", "Module", "class", "def", "embed_tokens", "get_input_embeddings", "model", "nn", "return", "self"], "kosmos2/modeling_kosmos2.py:Kosmos2TextForCausalLM.get_output_embeddings": ["GenerationMixin", "ModelPreTrainedModel", "ModelTextForCausalLM", "Module", "class", "def", "get_output_embeddings", "lm_head", "nn", "return", "self"], "kosmos2/modeling_kosmos2.py:Kosmos2TextForCausalLM.forward": ["BaseModelOutputWithPastAndCrossAttentions", "Cache", "CausalLMOutputWithCrossAttentions", "False", "GenerationMixin", "LongTensor", "ModelPreTrainedModel", "ModelTextForCausalLM", "None", "Tensor", "The", "TransformersKwargs", "Unpack", "argument", "attention_mask", "attentions", "bool", "cache_position", "changed", "class", "config", "cross_attentions", "def", "else", "encoder_attention_mask", "encoder_hidden_states", "forward", "hidden_states", "if", "image_embeds", "image_embeds_position_mask", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "logger", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "position_ids", "provided", "r", "return", "self", "since", "slice", "slice_indices", "to", "torch", "tuple", "use_cache", "vocab_size", "warning"], "kosmos2/modeling_kosmos2.py:Kosmos2TextForCausalLM.prepare_inputs_for_generation": ["False", "GenerationMixin", "ModelPreTrainedModel", "ModelTextForCausalLM", "None", "and", "attention_mask", "batch_size", "bool", "cache_position", "cat", "class", "def", "device", "dim", "dtype", "elif", "else", "if", "image_embeds", "image_embeds_position_mask", "input_ids", "inputs_embeds", "is", "is_first_iteration", "mask_len", "model_inputs", "model_kwargs", "not", "past_key_values", "pop", "position_ids", "prepare_inputs_for_generation", "return", "self", "seq_len", "size", "super", "torch", "use_cache", "zeros"], "kosmos2/modeling_kosmos2.py:Kosmos2ImageToTextProjection.__init__": ["False", "Linear", "ModelConfig", "ModelImageToTextProjection", "ModelTextAttention", "Module", "Parameter", "__init__", "add_inner_attn_layernorm", "attention_dropout", "attention_heads", "class", "config", "def", "dense", "dropout", "embed_dim", "hidden_size", "is_decoder", "latent_query", "latent_query_num", "nn", "randn", "self", "super", "text_config", "torch", "vision_config", "x_attn"], "kosmos2/modeling_kosmos2.py:Kosmos2ImageToTextProjection.forward": ["ModelImageToTextProjection", "Module", "None", "attention_mask", "attn_weights", "cat", "class", "def", "dense", "dim", "encoder_hidden_states", "expand", "features", "forward", "hidden_states", "key_value_states", "latent_query", "nn", "output_attentions", "past_key_values", "return", "self", "size", "torch", "unsqueeze", "x_attn"], "kosmos2/modeling_kosmos2.py:Kosmos2Model.__init__": ["ModelConfig", "ModelImageToTextProjection", "ModelModel", "ModelPreTrainedModel", "ModelTextModel", "ModelVisionModel", "__init__", "class", "config", "def", "image_to_text_projection", "post_init", "self", "super", "text_config", "text_model", "vision_config", "vision_model"], "kosmos2/modeling_kosmos2.py:Kosmos2Model.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "Module", "class", "def", "embed_tokens", "get_input_embeddings", "model", "nn", "return", "self", "text_model"], "kosmos2/modeling_kosmos2.py:Kosmos2Model.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embed_tokens", "model", "self", "set_input_embeddings", "text_model", "value"], "kosmos2/modeling_kosmos2.py:Kosmos2Model.get_image_features": ["False", "FloatTensor", "ModelModel", "ModelPreTrainedModel", "None", "bool", "class", "def", "dim", "functional", "get_image_features", "if", "image_embeds", "image_to_text_projection", "interpolate_pos_encoding", "model", "nn", "normalize", "pixel_values", "post_layernorm", "projection_attentions", "return", "return_attentions", "self", "torch", "vision_model", "vision_model_output"], "kosmos2/modeling_kosmos2.py:Kosmos2Model.forward": ["Cache", "False", "FlashAttentionKwargs", "ModelModel", "ModelModelOutput", "ModelPreTrainedModel", "None", "Tensor", "True", "Unpack", "ValueError", "You", "attention_mask", "attentions", "bool", "class", "config", "def", "either", "else", "forward", "get_image_features", "have", "hidden_states", "if", "image_embeds", "image_embeds_position_mask", "input_ids", "inputs_embeds", "interpolate_pos_encoding", "is", "kwargs", "last_hidden_state", "not", "or", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "pixel_values", "position_ids", "projection_attentions", "r", "raise", "return", "return_attentions", "return_dict", "self", "specify", "text_model", "to", "torch", "tuple", "use_cache", "use_return_dict", "vision_model_output"], "kosmos2/modeling_kosmos2.py:Kosmos2ForConditionalGeneration.__init__": ["GenerationMixin", "ModelConfig", "ModelForConditionalGeneration", "ModelImageToTextProjection", "ModelPreTrainedModel", "ModelTextForCausalLM", "ModelVisionModel", "__init__", "class", "config", "def", "image_to_text_projection", "post_init", "self", "super", "text_config", "text_model", "vision_config", "vision_model"], "kosmos2/modeling_kosmos2.py:Kosmos2ForConditionalGeneration.get_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "Module", "class", "def", "embed_tokens", "get_input_embeddings", "model", "nn", "return", "self", "text_model"], "kosmos2/modeling_kosmos2.py:Kosmos2ForConditionalGeneration.set_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "embed_tokens", "model", "self", "set_input_embeddings", "text_model", "value"], "kosmos2/modeling_kosmos2.py:Kosmos2ForConditionalGeneration.get_output_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "Module", "class", "def", "get_output_embeddings", "nn", "return", "self", "text_model"], "kosmos2/modeling_kosmos2.py:Kosmos2ForConditionalGeneration.set_output_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "new_embeddings", "self", "set_output_embeddings", "text_model"], "kosmos2/modeling_kosmos2.py:Kosmos2ForConditionalGeneration.forward": ["Cache", "CausalLMOutputWithCrossAttentions", "GenerationMixin", "LongTensor", "ModelForConditionalGeneration", "ModelForConditionalGenerationModelOutput", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "attention_mask", "attentions", "bool", "class", "config", "def", "dim", "either", "else", "forward", "functional", "have", "hidden_states", "if", "image_embeds", "image_embeds_position_mask", "image_to_text_projection", "input_ids", "inputs_embeds", "int", "is", "kwargs", "labels", "lm_outputs", "logits", "logits_to_keep", "loss", "model", "nn", "normalize", "not", "or", "output_attentions", "output_hidden_states", "past_key_values", "pixel_values", "position_ids", "post_layernorm", "projection_attentions", "r", "raise", "return", "self", "specify", "text_model", "to", "torch", "tuple", "use_cache", "vision_model", "vision_model_output"], "kosmos2/modeling_kosmos2.py:Kosmos2ForConditionalGeneration.generate": ["GenerationMixin", "Make", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Tensor", "ValueError", "allowed", "alongside", "and", "attention_mask", "class", "def", "dim", "either", "f", "functional", "generate", "if", "image_embeds", "image_embeds_position_mask", "image_to_text_projection", "input_ids", "inputs", "inputs_embeds", "is", "kwargs", "model", "nn", "normalize", "not", "or", "output", "pass", "passed", "pixel_values", "pop", "post_layernorm", "projection_attentions", "raise", "return", "self", "sure", "text_model", "to", "torch", "vision_model", "vision_model_output", "were", "which"], "kosmos2_5/modeling_kosmos2_5.py:_expand_mask": ["None", "Tensor", "_expand_mask", "bool", "bsz", "def", "dtype", "else", "expand", "expanded_mask", "finfo", "if", "int", "inverted_mask", "is", "mask", "masked_fill", "min", "not", "return", "size", "src_len", "tgt_len", "to", "torch"], "kosmos2_5/modeling_kosmos2_5.py:Kosmos2_5ModelOutput.to_tuple": ["Any", "ModelModelOutput", "ModelOutput", "class", "def", "else", "for", "getattr", "if", "in", "k", "keys", "return", "self", "to_tuple", "tuple", "vision_model_output"], "kosmos2_5/modeling_kosmos2_5.py:Kosmos2_5ForConditionalGenerationModelOutput.to_tuple": ["Any", "ModelForConditionalGenerationModelOutput", "ModelOutput", "class", "def", "else", "for", "getattr", "if", "in", "k", "keys", "return", "self", "to_tuple", "tuple", "vision_model_output"], "kosmos2_5/modeling_kosmos2_5.py:Kosmos2_5LayerNorm.__init__": ["ModelLayerNorm", "Module", "Parameter", "__init__", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "kosmos2_5/modeling_kosmos2_5.py:Kosmos2_5LayerNorm.forward": ["ModelLayerNorm", "Module", "True", "bfloat16", "class", "def", "dtype", "float16", "float32", "forward", "hidden_states", "if", "in", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "kosmos2_5/modeling_kosmos2_5.py:Kosmos2_5VisionEmbeddings.__init__": ["Dropout", "Embedding", "False", "Linear", "ModelVisionConfig", "ModelVisionEmbeddings", "Module", "None", "__init__", "class", "column_embedder", "config", "def", "dropout", "dropout_rate", "hidden_size", "inplace", "max_num_patches", "nn", "patch_embed_hidden_size", "patch_projection", "row_embedder", "self", "super"], "kosmos2_5/modeling_kosmos2_5.py:Kosmos2_5VisionEmbeddings.forward": ["ModelVisionEmbeddings", "Module", "Tensor", "class", "col_embeddings", "col_indices", "column_embedder", "def", "device", "dropout", "embeddings", "flattened_patches", "forward", "long", "nn", "patch_projection", "return", "row_embedder", "row_embeddings", "row_indices", "self", "to", "torch"], "kosmos2_5/modeling_kosmos2_5.py:Kosmos2_5VisionMlp.__init__": ["ACT2FN", "Dropout", "False", "Linear", "ModelVisionConfig", "ModelVisionMlp", "Module", "__init__", "act", "bias", "class", "config", "def", "dense_act_fn", "dropout", "dropout_rate", "hidden_size", "intermediate_size", "nn", "self", "super", "wi_0", "wi_1", "wo"], "kosmos2_5/modeling_kosmos2_5.py:Kosmos2_5VisionMlp.forward": ["ModelVisionMlp", "Module", "Tensor", "act", "and", "class", "def", "dropout", "dtype", "forward", "hidden_gelu", "hidden_linear", "hidden_states", "if", "int8", "isinstance", "nn", "return", "self", "to", "torch", "weight", "wi_0", "wi_1", "wo"], "kosmos2_5/modeling_kosmos2_5.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value"], "kosmos2_5/modeling_kosmos2_5.py:Kosmos2_5VisionAttention.__init__": ["False", "Linear", "ModelVisionAttention", "Module", "__init__", "attention_dropout", "bias", "class", "config", "def", "dropout", "gradient_checkpointing", "head_dim", "hidden_size", "inner_dim", "is_causal", "key", "n_heads", "nn", "num_attention_heads", "output", "query", "scaling", "self", "super", "value"], "kosmos2_5/modeling_kosmos2_5.py:Kosmos2_5VisionAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "Falling", "False", "ModelVisionAttention", "Module", "None", "This", "TransformersKwargs", "True", "Unpack", "_attn_implementation", "and", "argument", "attention", "attention_interface", "attention_mask", "attn_implementation", "attn_output", "attn_weights", "back", "be", "can", "class", "config", "def", "does", "dropout", "eager", "eager_attention_forward", "else", "forward", "functional", "get", "head_dim", "hidden_shape", "hidden_states", "if", "input_shape", "key", "key_states", "kwargs", "loading", "logger", "model", "nn", "not", "output", "output_attentions", "query", "query_states", "removed", "reshape", "return", "scaled_dot_product_attention", "scaling", "sdpa", "self", "shape", "support", "the", "to", "torch", "training", "transpose", "using", "value", "value_states", "view", "warning", "warning_once", "when"], "kosmos2_5/modeling_kosmos2_5.py:Kosmos2_5VisionLayer.__init__": ["GradientCheckpointingLayer", "ModelLayerNorm", "ModelVisionAttention", "ModelVisionConfig", "ModelVisionLayer", "ModelVisionMlp", "None", "__init__", "attention", "class", "config", "def", "eps", "hidden_size", "layer_norm_eps", "mlp", "pre_attention_layer_norm", "pre_mlp_layer_norm", "self", "super"], "kosmos2_5/modeling_kosmos2_5.py:Kosmos2_5VisionLayer.forward": ["False", "GradientCheckpointingLayer", "ModelVisionLayer", "None", "Tensor", "TransformersKwargs", "Unpack", "attention", "attention_mask", "attention_output", "bool", "class", "def", "forward", "hidden_states", "if", "kwargs", "layer_output", "mlp", "output_attentions", "outputs", "pre_attention_layer_norm", "pre_mlp_layer_norm", "residual", "return", "self", "self_attn_weights", "torch", "tuple"], "kosmos2_5/modeling_kosmos2_5.py:Kosmos2_5VisionEncoder.__init__": ["False", "ModelVisionConfig", "ModelVisionEncoder", "ModelVisionLayer", "Module", "ModuleList", "None", "_", "__init__", "class", "config", "def", "for", "gradient_checkpointing", "in", "layer", "nn", "num_hidden_layers", "range", "self", "super"], "kosmos2_5/modeling_kosmos2_5.py:Kosmos2_5VisionEncoder._prepare_attention_mask": ["ModelVisionEncoder", "Module", "None", "_attn_implementation", "_expand_mask", "_prepare_attention_mask", "and", "attention_mask", "class", "config", "def", "device", "dtype", "expanded_attn_mask", "flash_attention_2", "if", "in", "input_shape", "inputs_embeds", "is", "nn", "not", "return", "self", "tgt_len", "to"], "kosmos2_5/modeling_kosmos2_5.py:Kosmos2_5VisionEncoder.forward": ["BaseModelOutput", "False", "ModelVisionEncoder", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_prepare_attention_mask", "all_hidden_states", "all_self_attentions", "attention_mask", "attentions", "bool", "class", "def", "else", "enumerate", "for", "forward", "hidden_states", "i", "if", "in", "kwargs", "last_hidden_state", "layer", "layer_module", "layer_outputs", "nn", "output_attentions", "output_hidden_states", "return", "self", "shape", "torch"], "kosmos2_5/modeling_kosmos2_5.py:Kosmos2_5TextSinusoidalPositionalEmbedding.__init__": ["ModelTextSinusoidalPositionalEmbedding", "Module", "None", "__init__", "class", "def", "embedding_dim", "int", "make_weights", "nn", "num_positions", "offset", "padding_idx", "self", "super"], "kosmos2_5/modeling_kosmos2_5.py:Kosmos2_5TextSinusoidalPositionalEmbedding.make_weights": ["False", "ModelTextSinusoidalPositionalEmbedding", "Module", "None", "class", "def", "device", "dtype", "emb_weights", "embedding_dim", "get_embedding", "hasattr", "if", "int", "make_weights", "nn", "num_embeddings", "padding_idx", "persistent", "register_buffer", "self", "to", "weights"], "kosmos2_5/modeling_kosmos2_5.py:Kosmos2_5TextSinusoidalPositionalEmbedding.get_embedding": ["ModelTextSinusoidalPositionalEmbedding", "Module", "None", "arange", "cat", "class", "cos", "def", "dim", "dtype", "emb", "embedding_dim", "exp", "float", "get_default_dtype", "get_embedding", "half_dim", "if", "int", "int64", "is", "log", "math", "nn", "not", "num_embeddings", "padding_idx", "return", "sin", "to", "torch", "unsqueeze", "view", "zeros"], "kosmos2_5/modeling_kosmos2_5.py:Kosmos2_5TextSinusoidalPositionalEmbedding.forward": ["ModelTextSinusoidalPositionalEmbedding", "Module", "None", "Tensor", "bsz", "class", "create_position_ids_from_input_ids", "create_position_ids_from_inputs_embeds", "def", "detach", "device", "else", "embedding_dim", "forward", "if", "index_select", "input_ids", "inputs_embeds", "int", "is", "make_weights", "max_pos", "nn", "not", "offset", "padding_idx", "past_key_values_length", "position_ids", "return", "self", "seq_len", "shape", "size", "to", "torch", "view", "weights"], "kosmos2_5/modeling_kosmos2_5.py:Kosmos2_5TextSinusoidalPositionalEmbedding.create_position_ids_from_inputs_embeds": ["ModelTextSinusoidalPositionalEmbedding", "Module", "arange", "class", "contiguous", "create_position_ids_from_inputs_embeds", "def", "device", "dtype", "expand", "input_shape", "inputs_embeds", "long", "nn", "padding_idx", "past_key_values_length", "position_ids", "return", "sequence_length", "size", "torch", "unsqueeze"], "kosmos2_5/modeling_kosmos2_5.py:Kosmos2_5TextSinusoidalPositionalEmbedding.create_position_ids_from_input_ids": ["ModelTextSinusoidalPositionalEmbedding", "Module", "class", "create_position_ids_from_input_ids", "cumsum", "def", "dim", "incremental_indices", "input_ids", "int", "long", "mask", "ne", "nn", "padding_idx", "past_key_values_length", "return", "torch", "type_as"], "kosmos2_5/modeling_kosmos2_5.py:Kosmos2_5TextFFN.__init__": ["ACT2FN", "LayerNorm", "Linear", "ModelTextConfig", "ModelTextFFN", "Module", "__init__", "activation_dropout", "activation_fn", "activation_function", "class", "config", "def", "dropout", "embed_dim", "eps", "fc1", "fc2", "ffn_dim", "ffn_layernorm", "layer_norm_eps", "nn", "self", "super"], "kosmos2_5/modeling_kosmos2_5.py:Kosmos2_5TextFFN.forward": ["ModelTextFFN", "Module", "activation_dropout", "activation_fn", "class", "def", "dropout", "fc1", "fc2", "ffn_layernorm", "forward", "functional", "hidden_states", "nn", "p", "return", "self", "training"], "kosmos2_5/modeling_kosmos2_5.py:Kosmos2_5TextAttention.__init__": ["False", "Linear", "ModelTextAttention", "Module", "None", "True", "ValueError", "__init__", "and", "be", "bias", "bool", "by", "class", "config", "def", "divisible", "dropout", "embed_dim", "f", "float", "got", "head_dim", "if", "int", "is_causal", "is_decoder", "k_proj", "layer_idx", "must", "nn", "num_heads", "out_proj", "q_proj", "raise", "scaling", "self", "super", "v_proj"], "kosmos2_5/modeling_kosmos2_5.py:Kosmos2_5TextAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "Falling", "False", "LongTensor", "ModelTextAttention", "Module", "None", "Tensor", "This", "TransformersKwargs", "True", "Unpack", "_attn_implementation", "and", "argument", "attention", "attention_interface", "attention_mask", "attn_implementation", "attn_output", "attn_weights", "back", "be", "cache_kwargs", "cache_position", "can", "class", "config", "contiguous", "current_hidden_shape", "current_input_shape", "current_states", "def", "does", "dropout", "eager", "eager_attention_forward", "else", "encoder_hidden_states", "forward", "functional", "get", "head_dim", "hidden_shape", "hidden_states", "if", "input_shape", "is", "is_cross_attention", "k_proj", "key_states", "kwargs", "layer_idx", "loading", "logger", "model", "nn", "not", "out_proj", "output_attentions", "past_key_values", "q_proj", "query_states", "removed", "reshape", "return", "scaled_dot_product_attention", "scaling", "sdpa", "self", "shape", "support", "the", "to", "torch", "training", "transpose", "tuple", "update", "using", "v_proj", "value_states", "view", "warning", "warning_once", "when"], "kosmos2_5/modeling_kosmos2_5.py:Kosmos2_5TextBlock.__init__": ["GradientCheckpointingLayer", "LayerNorm", "ModelTextAttention", "ModelTextBlock", "ModelTextConfig", "ModelTextFFN", "True", "__init__", "attention_dropout", "attention_heads", "class", "config", "def", "dropout", "embed_dim", "eps", "ffn", "final_layer_norm", "int", "is_causal", "is_decoder", "layer_idx", "layer_norm_eps", "nn", "num_heads", "self", "self_attn", "self_attn_layer_norm", "super"], "kosmos2_5/modeling_kosmos2_5.py:Kosmos2_5TextBlock.forward": ["Cache", "False", "FloatTensor", "GradientCheckpointingLayer", "LongTensor", "ModelTextBlock", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "attention_mask", "bool", "cache_position", "class", "def", "dropout", "ffn", "final_layer_norm", "forward", "functional", "hidden_states", "if", "kwargs", "nn", "output_attentions", "outputs", "p", "past_key_values", "residual", "return", "self", "self_attn", "self_attn_layer_norm", "self_attn_weights", "torch", "training", "tuple", "use_cache"], "kosmos2_5/modeling_kosmos2_5.py:Kosmos2_5TextTransformer.__init__": ["Embedding", "False", "LayerNorm", "ModelTextBlock", "ModelTextConfig", "ModelTextSinusoidalPositionalEmbedding", "ModelTextTransformer", "Module", "ModuleList", "__init__", "class", "config", "def", "dropout", "else", "embed_dim", "embed_positions", "embed_scale", "embed_tokens", "embedding_dim", "for", "gradient_checkpointing", "if", "in", "layer_idx", "layer_norm", "layer_norm_eps", "layerdrop", "layers", "math", "max_position_embeddings", "nn", "num_positions", "pad_token_id", "padding_idx", "range", "scale_embedding", "segment_emb", "self", "sqrt", "super", "vocab_size"], "kosmos2_5/modeling_kosmos2_5.py:Kosmos2_5TextTransformer._update_causal_mask": ["AttentionMaskConverter", "BlockMask", "Cache", "False", "ModelTextTransformer", "Module", "None", "Tensor", "Union", "_attn_implementation", "_ignore_causal_mask_sdpa", "_prepare_4d_causal_attention_mask_with_cache_position", "_unmask_unattended", "_update_causal_mask", "and", "any", "attention_mask", "batch_size", "bool", "cache_position", "causal_mask", "class", "config", "cuda", "def", "device", "dtype", "else", "finfo", "flash_attention_2", "flex_attention", "get_max_cache_shape", "get_seq_length", "if", "in", "input_tensor", "inputs_embeds", "is", "is_compileable", "is_training", "isinstance", "make_flex_block_causal_mask", "min", "min_dtype", "nn", "not", "npu", "output_attentions", "past_key_values", "past_key_values_length", "past_seen_tokens", "return", "sdpa", "self", "sequence_length", "shape", "target_length", "torch", "training", "type", "using_compilable_cache", "xpu"], "kosmos2_5/modeling_kosmos2_5.py:Kosmos2_5TextTransformer._prepare_4d_causal_attention_mask_with_cache_position": ["ModelTextTransformer", "Module", "None", "Tensor", "_prepare_4d_causal_attention_mask_with_cache_position", "and", "arange", "attention_mask", "batch_size", "cache_position", "causal_mask", "class", "clone", "def", "device", "diagonal", "dim", "dtype", "else", "expand", "fill_value", "finfo", "full", "if", "int", "is", "kwargs", "mask_length", "masked_fill", "min", "min_dtype", "nn", "not", "padding_mask", "reshape", "return", "sequence_length", "shape", "target_length", "to", "torch", "triu"], "kosmos2_5/modeling_kosmos2_5.py:Kosmos2_5TextTransformer.forward": ["BaseModelOutputWithPast", "BaseModelOutputWithPastAndCrossAttentions", "Cache", "DynamicCache", "False", "LongTensor", "ModelTextTransformer", "Module", "None", "Setting", "Tensor", "TransformersKwargs", "True", "Unpack", "ValueError", "You", "_update_causal_mask", "all_hidden_states", "all_self_attns", "and", "arange", "at", "attention_mask", "attentions", "bool", "both", "bsz", "cache_position", "cannot", "causal_mask", "checkpointing", "class", "config", "decoder_layer", "def", "device", "dim", "dropout", "dtype", "either", "else", "embed_positions", "embed_scale", "embed_tokens", "for", "forward", "functional", "get_seq_length", "gradient", "gradient_checkpointing", "hidden_states", "if", "image_embeds", "image_embeds_position_mask", "in", "incompatible", "input_ids", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layer_norm", "layer_outputs", "layers", "logger", "long", "must", "ne", "nn", "not", "one", "output", "output_attentions", "output_hidden_states", "p", "past_key_values", "past_key_values_length", "past_seen_tokens", "position_ids", "positions", "raise", "return", "same", "segment_emb", "segment_embeds", "self", "seq_len", "shape", "size", "specify", "the", "time", "to", "torch", "training", "use_cache", "view", "warning_once", "weight", "with", "zero_emb", "zeros"], "kosmos2_5/modeling_kosmos2_5.py:Kosmos2_5ImageToTextProjection.__init__": ["False", "Linear", "ModelConfig", "ModelImageToTextProjection", "ModelTextAttention", "Module", "Parameter", "__init__", "attention_dropout", "attention_heads", "class", "config", "def", "dense", "dropout", "embed_dim", "hidden_size", "is_causal", "is_decoder", "latent_query", "latent_query_num", "nn", "randn", "self", "super", "text_config", "torch", "vision_config", "x_attn"], "kosmos2_5/modeling_kosmos2_5.py:Kosmos2_5ImageToTextProjection.forward": ["False", "ModelImageToTextProjection", "Module", "None", "attention_mask", "attn_weights", "cat", "class", "def", "dense", "dim", "encoder_hidden_states", "expand", "features", "forward", "hidden_states", "is_causal", "key_value_states", "latent_query", "nn", "output_attentions", "past_key_values", "return", "self", "size", "torch", "unsqueeze", "x_attn"], "kosmos2_5/modeling_kosmos2_5.py:Kosmos2_5PreTrainedModel._init_weights": ["Embedding", "False", "LayerNorm", "Linear", "ModelForConditionalGeneration", "ModelImageToTextProjection", "ModelLayerNorm", "ModelModel", "ModelPreTrainedModel", "ModelTextForCausalLM", "ModelTextModel", "ModelTextSinusoidalPositionalEmbedding", "ModelVisionModel", "None", "PreTrainedModel", "_init_weights", "_is_hf_initialized", "and", "bias", "class", "config", "copy_", "def", "elif", "emb_weights", "embedding_dim", "get_embedding", "getattr", "if", "init", "init_factor", "init_std", "initializer_factor", "initializer_range", "is", "isinstance", "latent_query", "mean", "module", "nn", "normal_", "not", "num_positions", "offset", "ones_", "padding_idx", "self", "std", "text_config", "weight", "weights", "zeros_"], "kosmos2_5/modeling_kosmos2_5.py:Kosmos2_5VisionModel.__init__": ["ModelLayerNorm", "ModelPreTrainedModel", "ModelVisionConfig", "ModelVisionEmbeddings", "ModelVisionEncoder", "ModelVisionModel", "__init__", "class", "config", "def", "embeddings", "encoder", "eps", "hidden_size", "layer_norm_eps", "layernorm", "post_init", "self", "super"], "kosmos2_5/modeling_kosmos2_5.py:Kosmos2_5VisionModel.get_input_embeddings": ["ModelPreTrainedModel", "ModelVisionModel", "class", "def", "embeddings", "get_input_embeddings", "patch_projection", "return", "self"], "kosmos2_5/modeling_kosmos2_5.py:Kosmos2_5VisionModel.forward": ["BaseModelOutput", "BaseModelOutputWithPooling", "ModelPreTrainedModel", "ModelVisionModel", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "attention_mask", "attentions", "bool", "class", "config", "def", "dim", "else", "embedding_output", "embeddings", "encoder", "encoder_outputs", "flattened_patches", "float", "forward", "have", "hidden_states", "if", "is", "kwargs", "last_hidden_state", "layernorm", "not", "output_attentions", "output_hidden_states", "raise", "return", "self", "sequence_output", "specify", "sum", "to", "torch"], "kosmos2_5/modeling_kosmos2_5.py:Kosmos2_5TextModel.__init__": ["ModelPreTrainedModel", "ModelTextConfig", "ModelTextModel", "ModelTextTransformer", "__init__", "class", "config", "def", "model", "post_init", "self", "super"], "kosmos2_5/modeling_kosmos2_5.py:Kosmos2_5TextModel.get_input_embeddings": ["ModelPreTrainedModel", "ModelTextModel", "Module", "class", "def", "embed_tokens", "get_input_embeddings", "model", "nn", "return", "self"], "kosmos2_5/modeling_kosmos2_5.py:Kosmos2_5TextModel.set_input_embeddings": ["ModelPreTrainedModel", "ModelTextModel", "class", "def", "embed_tokens", "model", "self", "set_input_embeddings", "value"], "kosmos2_5/modeling_kosmos2_5.py:Kosmos2_5TextModel.forward": ["BaseModelOutputWithPastAndCrossAttentions", "Cache", "LongTensor", "ModelPreTrainedModel", "ModelTextModel", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "bool", "cache_position", "class", "def", "forward", "image_embeds", "image_embeds_position_mask", "input_ids", "inputs_embeds", "kwargs", "model", "output_attentions", "output_hidden_states", "past_key_values", "position_ids", "r", "return", "self", "torch", "use_cache"], "kosmos2_5/modeling_kosmos2_5.py:Kosmos2_5Model.__init__": ["ModelConfig", "ModelImageToTextProjection", "ModelModel", "ModelPreTrainedModel", "ModelTextModel", "ModelVisionModel", "__init__", "_from_config", "class", "config", "def", "image_to_text_projection", "post_init", "self", "super", "text_config", "text_model", "vision_config", "vision_model"], "kosmos2_5/modeling_kosmos2_5.py:Kosmos2_5Model.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "Module", "class", "def", "embed_tokens", "get_input_embeddings", "model", "nn", "return", "self", "text_model"], "kosmos2_5/modeling_kosmos2_5.py:Kosmos2_5Model.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embed_tokens", "model", "self", "set_input_embeddings", "text_model", "value"], "kosmos2_5/modeling_kosmos2_5.py:Kosmos2_5Model.forward": ["Cache", "LongTensor", "ModelModel", "ModelModelOutput", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "dim", "else", "flattened_patches", "forward", "functional", "height", "hidden_states", "if", "image_embeds", "image_embeds_position_mask", "image_to_text_projection", "input_ids", "inputs_embeds", "is", "kwargs", "last_hidden_state", "nn", "normalize", "not", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "position_ids", "projection_attentions", "r", "return", "self", "text_model", "torch", "use_cache", "vision_model", "vision_model_output", "width"], "kosmos2_5/modeling_kosmos2_5.py:Kosmos2_5TextForCausalLM.__init__": ["False", "Linear", "ModelPreTrainedModel", "ModelTextConfig", "ModelTextForCausalLM", "ModelTextTransformer", "__init__", "bias", "class", "config", "def", "embed_dim", "in_features", "lm_head", "model", "nn", "out_features", "post_init", "self", "super", "vocab_size"], "kosmos2_5/modeling_kosmos2_5.py:Kosmos2_5TextForCausalLM.get_input_embeddings": ["ModelPreTrainedModel", "ModelTextForCausalLM", "Module", "class", "def", "embed_tokens", "get_input_embeddings", "model", "nn", "return", "self"], "kosmos2_5/modeling_kosmos2_5.py:Kosmos2_5TextForCausalLM.set_input_embeddings": ["ModelPreTrainedModel", "ModelTextForCausalLM", "class", "def", "embed_tokens", "model", "self", "set_input_embeddings", "value"], "kosmos2_5/modeling_kosmos2_5.py:Kosmos2_5TextForCausalLM.get_output_embeddings": ["ModelPreTrainedModel", "ModelTextForCausalLM", "Module", "class", "def", "get_output_embeddings", "lm_head", "nn", "return", "self"], "kosmos2_5/modeling_kosmos2_5.py:Kosmos2_5TextForCausalLM.set_output_embeddings": ["ModelPreTrainedModel", "ModelTextForCausalLM", "class", "def", "lm_head", "new_embeddings", "self", "set_output_embeddings"], "kosmos2_5/modeling_kosmos2_5.py:Kosmos2_5TextForCausalLM.forward": ["BaseModelOutputWithPastAndCrossAttentions", "Cache", "CausalLMOutputWithCrossAttentions", "False", "LongTensor", "ModelPreTrainedModel", "ModelTextForCausalLM", "None", "Tensor", "The", "TransformersKwargs", "Unpack", "argument", "attention_mask", "attentions", "bool", "changed", "class", "config", "def", "else", "forward", "hidden_states", "if", "image_embeds", "image_embeds_position_mask", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "logger", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "position_ids", "provided", "r", "return", "self", "since", "slice", "slice_indices", "to", "torch", "use_cache", "vocab_size", "warning"], "kosmos2_5/modeling_kosmos2_5.py:Kosmos2_5TextForCausalLM.prepare_inputs_for_generation": ["False", "ModelPreTrainedModel", "ModelTextForCausalLM", "ModelTextSinusoidalPositionalEmbedding", "None", "attention_mask", "batch_size", "bool", "cache_position", "cat", "class", "config", "create_position_ids_from_input_ids", "def", "device", "dim", "dtype", "elif", "for", "get_seq_length", "if", "image_embeds", "image_embeds_position_mask", "in", "input_ids", "input_shape", "is", "is_first_iteration", "items", "key", "mask_len", "model_inputs", "model_kwargs", "new_ones", "not", "pad_token_id", "padding_idx", "past_key_values", "past_key_values_length", "position_ids", "prepare_inputs_for_generation", "return", "self", "seq_len", "shape", "size", "torch", "use_cache", "value", "zeros"], "kosmos2_5/modeling_kosmos2_5.py:Kosmos2_5ForConditionalGeneration.__init__": ["GenerationMixin", "ModelConfig", "ModelForConditionalGeneration", "ModelImageToTextProjection", "ModelPreTrainedModel", "ModelTextForCausalLM", "ModelVisionModel", "__init__", "class", "config", "def", "image_to_text_projection", "post_init", "self", "super", "text_config", "text_model", "vision_config", "vision_model"], "kosmos2_5/modeling_kosmos2_5.py:Kosmos2_5ForConditionalGeneration.get_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "Module", "class", "def", "embed_tokens", "get_input_embeddings", "model", "nn", "return", "self", "text_model"], "kosmos2_5/modeling_kosmos2_5.py:Kosmos2_5ForConditionalGeneration.set_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "embed_tokens", "model", "self", "set_input_embeddings", "text_model", "value"], "kosmos2_5/modeling_kosmos2_5.py:Kosmos2_5ForConditionalGeneration.get_output_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "Module", "class", "def", "get_output_embeddings", "nn", "return", "self", "text_model"], "kosmos2_5/modeling_kosmos2_5.py:Kosmos2_5ForConditionalGeneration.set_output_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "new_embeddings", "self", "set_output_embeddings", "text_model"], "kosmos2_5/modeling_kosmos2_5.py:Kosmos2_5ForConditionalGeneration.forward": ["Cache", "CausalLMOutputWithCrossAttentions", "GenerationMixin", "LongTensor", "ModelForConditionalGeneration", "ModelForConditionalGenerationModelOutput", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "bool", "class", "config", "def", "dim", "else", "flattened_patches", "forward", "functional", "height", "hidden_states", "if", "image_embeds", "image_embeds_position_mask", "image_to_text_projection", "input_ids", "inputs_embeds", "int", "is", "kwargs", "labels", "last_hidden_state", "lm_outputs", "logits", "logits_to_keep", "loss", "nn", "normalize", "not", "output_attentions", "output_hidden_states", "past_key_values", "position_ids", "projection_attentions", "r", "return", "self", "text_model", "torch", "use_cache", "vision_model", "vision_model_output", "width"], "kosmos2_5/modeling_kosmos2_5.py:Kosmos2_5ForConditionalGeneration.prepare_inputs_for_generation": ["False", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "attention_mask", "cache_position", "class", "def", "flattened_patches", "if", "image_embeds", "image_embeds_position_mask", "input_ids", "is_first_iteration", "model_inputs", "model_kwargs", "not", "or", "past_key_values", "position_ids", "prepare_inputs_for_generation", "return", "self", "text_model", "use_cache"], "kyutai_speech_to_text/modeling_kyutai_speech_to_text.py:KyutaiSpeechToTextFlexibleLinear.__init__": ["ModelFlexibleLinear", "Module", "Parameter", "__init__", "class", "def", "input_size", "nn", "num_layers", "output_size", "randn", "self", "super", "torch", "weight"], "kyutai_speech_to_text/modeling_kyutai_speech_to_text.py:KyutaiSpeechToTextFlexibleLinear.forward": ["ModelFlexibleLinear", "Module", "None", "class", "def", "else", "forward", "if", "index_select", "is", "layer_idx", "matmul", "nn", "not", "return", "selected_weights", "self", "squeeze", "torch", "transpose", "weight", "x"], "kyutai_speech_to_text/modeling_kyutai_speech_to_text.py:KyutaiSpeechToTextPreTrainedModel._init_weights": ["ModelEmbeddings", "ModelFlexibleLinear", "ModelPreTrainedModel", "PreTrainedModel", "_init_weights", "arange", "audio_tokens_offsets", "class", "codebook_vocab_size", "config", "copy_", "def", "functional", "if", "init", "isinstance", "module", "nn", "normal_", "num_codebooks", "pad", "self", "super", "torch", "vocab_size", "weight"], "kyutai_speech_to_text/modeling_kyutai_speech_to_text.py:KyutaiSpeechToTextConv1dPaddingCache.__init__": ["Expected", "ModelConv1dPaddingCache", "None", "NotImplementedError", "True", "ValueError", "__init__", "all", "and", "class", "constant", "convolutions", "def", "elif", "f", "for", "from_args_num_layers", "if", "in", "int", "is", "len", "list", "mode", "not", "num_layers", "or", "other", "padding", "padding_cache", "per_layer_in_channels", "per_layer_is_init", "per_layer_padding", "per_layer_padding_mode", "pop", "raise", "replicate", "self", "str", "supported", "than", "using", "values"], "kyutai_speech_to_text/modeling_kyutai_speech_to_text.py:KyutaiSpeechToTextConv1dPaddingCache.update": ["ModelConv1dPaddingCache", "None", "Tensor", "batch_size", "class", "constant", "current_cache", "def", "device", "dtype", "elif", "else", "empty", "hidden_states", "if", "in_channels", "int", "is", "layer_idx", "ones", "padding", "padding_cache", "padding_mode", "padding_states", "per_layer_in_channels", "per_layer_padding", "per_layer_padding_mode", "replicate", "return", "self", "shape", "torch", "update", "zeros"], "kyutai_speech_to_text/modeling_kyutai_speech_to_text.py:KyutaiSpeechToTextEmbeddings.__init__": ["Embedding", "False", "ModelEmbeddings", "Module", "__init__", "arange", "audio_pad_token_id", "audio_tokens_offsets", "class", "codebook_vocab_size", "config", "def", "embed_tokens", "functional", "hidden_size", "nn", "num_codebooks", "pad", "padding_idx", "persistent", "register_buffer", "self", "super", "torch", "vocab_size"], "kyutai_speech_to_text/modeling_kyutai_speech_to_text.py:KyutaiSpeechToTextEmbeddings.forward": ["ModelEmbeddings", "Module", "audio_tokens_offsets", "class", "def", "dim", "embed_tokens", "forward", "input_ids", "inputs_embeds", "nn", "padding_idx", "return", "self", "sum", "torch", "where"], "kyutai_speech_to_text/modeling_kyutai_speech_to_text.py:KyutaiSpeechToTextRMSNorm.__init__": ["ModelRMSNorm", "Module", "Parameter", "__init__", "class", "def", "dim", "eps", "float", "int", "nn", "ones", "self", "super", "torch", "weight"], "kyutai_speech_to_text/modeling_kyutai_speech_to_text.py:KyutaiSpeechToTextRMSNorm._norm": ["ModelRMSNorm", "Module", "True", "_norm", "class", "def", "eps", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "torch", "x"], "kyutai_speech_to_text/modeling_kyutai_speech_to_text.py:KyutaiSpeechToTextRMSNorm.forward": ["ModelRMSNorm", "Module", "_norm", "class", "def", "float", "forward", "nn", "output", "return", "self", "type_as", "weight", "x"], "kyutai_speech_to_text/modeling_kyutai_speech_to_text.py:KyutaiSpeechToTextRMSNorm.extra_repr": ["ModelRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "weight"], "kyutai_speech_to_text/modeling_kyutai_speech_to_text.py:KyutaiSpeechToTextLinear.__init__": ["False", "Linear", "ModelFlexibleLinear", "ModelLinear", "Module", "__init__", "bias", "class", "def", "else", "if", "input_dim", "linear", "nn", "not", "num_codebooks", "num_layers", "output_dim", "self", "super", "use_flexible_linear"], "kyutai_speech_to_text/modeling_kyutai_speech_to_text.py:KyutaiSpeechToTextLinear.forward": ["ModelLinear", "Module", "None", "class", "def", "else", "forward", "if", "layer_idx", "linear", "nn", "return", "self", "use_flexible_linear", "x"], "kyutai_speech_to_text/modeling_kyutai_speech_to_text.py:KyutaiSpeechToTextRotaryEmbedding.__init__": ["Callable", "False", "ModelConfig", "ModelRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "kyutai_speech_to_text/modeling_kyutai_speech_to_text.py:KyutaiSpeechToTextRotaryEmbedding.compute_default_rope_parameters": ["ModelConfig", "ModelRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "kyutai_speech_to_text/modeling_kyutai_speech_to_text.py:KyutaiSpeechToTextRotaryEmbedding.forward": ["False", "ModelRotaryEmbedding", "Module", "None", "and", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "kyutai_speech_to_text/modeling_kyutai_speech_to_text.py:KyutaiSpeechToTextGatingMLP.__init__": ["ACT2FN", "False", "Linear", "ModelFlexibleLinear", "ModelGatingMLP", "Module", "__init__", "activation_fn", "bias", "class", "config", "def", "else", "fc1", "fc2", "ffn_dim", "hidden_act", "hidden_size", "if", "nn", "num_codebooks", "num_layers", "self", "super", "use_flexible_linear"], "kyutai_speech_to_text/modeling_kyutai_speech_to_text.py:KyutaiSpeechToTextGatingMLP.forward": ["ModelGatingMLP", "Module", "None", "Tensor", "_", "activation_fn", "batch_size", "class", "def", "else", "fc1", "fc2", "forward", "hidden_states", "if", "int", "is", "layer_idx", "nn", "return", "self", "sequence_length", "shape", "torch", "view"], "kyutai_speech_to_text/modeling_kyutai_speech_to_text.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "kyutai_speech_to_text/modeling_kyutai_speech_to_text.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cos", "def", "k", "k_embed", "q", "q_embed", "return", "rotate_half", "sin", "unsqueeze", "unsqueeze_dim"], "kyutai_speech_to_text/modeling_kyutai_speech_to_text.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "kyutai_speech_to_text/modeling_kyutai_speech_to_text.py:KyutaiSpeechToTextAttention.__init__": ["False", "Instantiating", "ModelAttention", "ModelConfig", "ModelLinear", "ModelRotaryEmbedding", "Module", "None", "Please", "True", "ValueError", "__class__", "__init__", "__name__", "a", "and", "attention_dropout", "be", "by", "caching", "call", "class", "config", "creating", "def", "divisible", "during", "errors", "f", "forward", "got", "head_dim", "hidden_size", "if", "int", "is", "is_causal", "k_proj", "layer_idx", "lead", "logger", "make", "math", "max_position_embeddings", "must", "nn", "not", "num_attention_heads", "num_codebooks", "num_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "passing", "provide", "q_proj", "raise", "recommended", "rotary_emb", "scaling", "self", "sqrt", "super", "sure", "the", "this", "to", "use_flexible_linear", "use_rope", "used", "v_proj", "warning_once", "when", "will", "without"], "kyutai_speech_to_text/modeling_kyutai_speech_to_text.py:KyutaiSpeechToTextAttention.forward": ["Cache", "False", "LongTensor", "ModelAttention", "Module", "None", "Tensor", "ValueError", "_", "apply_rotary_pos_emb", "attention_dropout", "attention_mask", "attn_output", "attn_weights", "be", "bool", "bsz", "but", "cache_kwargs", "cache_position", "causal_mask", "class", "contiguous", "cos", "def", "dim", "dropout", "dtype", "else", "f", "float32", "forward", "functional", "head_dim", "hidden_states", "if", "is", "k_proj", "key_states", "layer_idx", "matmul", "nn", "not", "num_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "of", "output_attentions", "p", "past_key_values", "position_ids", "q_len", "q_proj", "query_states", "raise", "repeat_kv", "return", "rotary_emb", "scaling", "self", "shape", "should", "sin", "size", "softmax", "to", "torch", "training", "transpose", "tuple", "update", "use_cache", "v_proj", "value_states", "view"], "kyutai_speech_to_text/modeling_kyutai_speech_to_text.py:KyutaiSpeechToTextFlashAttention2.__init__": ["ModelAttention", "ModelFlashAttention2", "__init__", "_flash_attn_uses_top_left_mask", "args", "class", "def", "flash_attn_supports_top_left_mask", "kwargs", "self", "super"], "kyutai_speech_to_text/modeling_kyutai_speech_to_text.py:KyutaiSpeechToTextFlashAttention2.forward": ["Cache", "False", "LongTensor", "ModelAttention", "ModelFlashAttention2", "None", "StaticCache", "Tensor", "The", "ValueError", "We", "_", "_flash_attention_forward", "_flash_attn_uses_top_left_mask", "_is_quantized", "an", "and", "apply_rotary_pos_emb", "at", "attention_dropout", "attention_mask", "attn_implementation", "attn_output", "attn_weights", "back", "be", "bool", "bsz", "cache", "cache_kwargs", "cache_position", "cast", "casted", "class", "com", "compatible", "config", "contiguous", "cos", "cpu", "def", "device", "device_type", "dropout", "dropout_rate", "dtype", "elif", "else", "embedding", "f", "fact", "flash_attention_2", "float32", "forward", "get_autocast_dtype", "get_autocast_gpu_dtype", "getattr", "github", "hasattr", "have", "head_dim", "hidden", "hidden_states", "https", "huggingface", "if", "implementation", "in", "input", "input_dtype", "is", "is_autocast_enabled", "is_causal", "isinstance", "issue", "k_proj", "key_states", "layer", "layer_idx", "layers", "logger", "make", "mean", "might", "mps", "norm", "not", "num_heads", "num_key_value_heads", "o_proj", "open", "or", "output_attentions", "past_key_values", "position_ids", "q_len", "q_proj", "query_states", "raise", "related", "reshape", "return", "rotary_emb", "sdpa", "seems", "self", "silently", "sin", "size", "sliding_window", "states", "static", "sure", "target_dtype", "the", "this", "time", "to", "torch", "training", "transformers", "transpose", "tuple", "type", "upcasted", "update", "use", "use_cache", "use_top_left_mask", "v_proj", "value_states", "view", "warning_once", "weight", "will", "with", "you"], "kyutai_speech_to_text/modeling_kyutai_speech_to_text.py:KyutaiSpeechToTextSdpaAttention.forward": ["Cache", "False", "If", "LongTensor", "ModelAttention", "ModelSdpaAttention", "None", "Tensor", "The", "True", "_", "__class__", "__name__", "and", "apply_rotary_pos_emb", "attention", "attention_dropout", "attention_mask", "attn_implementation", "attn_mask", "attn_output", "be", "bool", "bsz", "cache_kwargs", "cache_position", "causal_mask", "class", "contiguous", "cos", "cuda", "def", "device", "does", "dropout_p", "eager", "else", "f", "forward", "functional", "get", "head_dim", "hidden_states", "if", "is", "is_causal", "k_proj", "key_states", "kwargs", "layer_idx", "loading", "logger", "model", "nn", "not", "num_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "output_attentions", "past_key_values", "please", "position_ids", "q_len", "q_proj", "query_states", "repeat_kv", "return", "returned", "rotary_emb", "scaled_dot_product_attention", "self", "set", "shape", "sin", "size", "support", "the", "to", "torch", "training", "transpose", "tuple", "type", "update", "use_cache", "v_proj", "value_states", "view", "want", "warning_once", "weights", "when", "will", "you"], "kyutai_speech_to_text/modeling_kyutai_speech_to_text.py:KyutaiSpeechToTextDecoderLayer.__init__": ["GradientCheckpointingLayer", "ModelConfig", "ModelDecoderLayer", "ModelGatingMLP", "ModelRMSNorm", "Model_ATTENTION_CLASSES", "True", "__init__", "_attn_implementation", "bool", "class", "config", "def", "eps", "hidden_size", "input_layernorm", "int", "layer_idx", "mlp", "post_attention_layernorm", "rms_norm_eps", "self", "self_attn", "sliding_window", "super", "use_flexible_linear", "use_rope"], "kyutai_speech_to_text/modeling_kyutai_speech_to_text.py:KyutaiSpeechToTextDecoderLayer.forward": ["Cache", "False", "FloatTensor", "GradientCheckpointingLayer", "LongTensor", "ModelDecoderLayer", "None", "Tensor", "attention_mask", "bool", "cache_position", "class", "def", "else", "forward", "hidden_states", "if", "input_layernorm", "kwargs", "mlp", "not", "output_attentions", "outputs", "past_key_values", "position_ids", "post_attention_layernorm", "residual", "return", "self", "self_attn", "self_attn_weights", "torch", "tuple", "use_cache", "use_flexible_linear"], "kyutai_speech_to_text/modeling_kyutai_speech_to_text.py:KyutaiSpeechToTextModel.__init__": ["False", "ModelDecoderLayer", "ModelEmbeddings", "ModelModel", "ModelPreTrainedModel", "ModelRMSNorm", "ModuleList", "__init__", "class", "config", "def", "embed_tokens", "eps", "for", "gradient_checkpointing", "hidden_size", "in", "layer_idx", "layers", "nn", "norm", "num_hidden_layers", "pad_token_id", "padding_idx", "post_init", "range", "rms_norm_eps", "self", "super", "use_flexible_linear", "vocab_size"], "kyutai_speech_to_text/modeling_kyutai_speech_to_text.py:KyutaiSpeechToTextModel.forward": ["BaseModelOutputWithPast", "Cache", "DynamicCache", "False", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Setting", "Tensor", "True", "_update_causal_mask", "all_hidden_states", "all_self_attns", "and", "arange", "attention_mask", "attentions", "bool", "cache_position", "causal_mask", "checkpointing", "class", "config", "decoder_layer", "def", "device", "else", "embed_tokens", "for", "forward", "get_seq_length", "gradient", "gradient_checkpointing", "hidden_states", "if", "in", "incompatible", "input_ids", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layer_outputs", "layers", "logger", "norm", "not", "output_attentions", "output_hidden_states", "past_key_values", "past_seen_tokens", "position_ids", "return", "return_dict", "self", "shape", "torch", "training", "tuple", "unsqueeze", "use_cache", "use_return_dict", "v", "warning_once", "with"], "kyutai_speech_to_text/modeling_kyutai_speech_to_text.py:KyutaiSpeechToTextModel._update_causal_mask": ["Attention", "AttentionMaskConverter", "BlockMask", "Cache", "False", "Flash", "Make", "Model", "ModelModel", "ModelPreTrainedModel", "None", "StaticCache", "Tensor", "Union", "ValueError", "You", "_attn_implementation", "_ignore_causal_mask_sdpa", "_prepare_4d_causal_attention_mask_with_cache_position", "_unmask_unattended", "_update_causal_mask", "and", "are", "attempting", "attention_mask", "batch_size", "batched", "before", "behaviour", "bool", "cache_position", "call", "causal_mask", "class", "config", "cuda", "def", "device", "dtype", "else", "finfo", "flash_attention_2", "flex_attention", "for", "generation", "get_max_cache_shape", "get_seq_length", "if", "in", "input", "input_tensor", "inputs_embeds", "is", "is_padding_right", "is_training", "isinstance", "item", "lead", "left", "make_flex_block_causal_mask", "may", "min", "min_dtype", "not", "npu", "of", "output_attentions", "padding_side", "past_key_values", "past_key_values_length", "past_seen_tokens", "perform", "raise", "return", "right", "sdpa", "self", "sequence_length", "shape", "size", "sliding_window", "sum", "sure", "target_length", "the", "this", "to", "tokenizer", "tokenizing", "torch", "training", "type", "unexpected", "using_static_cache", "version", "with", "xpu"], "kyutai_speech_to_text/modeling_kyutai_speech_to_text.py:KyutaiSpeechToTextModel._prepare_4d_causal_attention_mask_with_cache_position": ["Cache", "ModelConfig", "ModelModel", "ModelPreTrainedModel", "None", "StaticCache", "Tensor", "True", "_prepare_4d_causal_attention_mask_with_cache_position", "all", "and", "arange", "attention_mask", "batch_size", "bitwise_or_", "cache_position", "causal_mask", "class", "clone", "config", "def", "device", "diagonal_attend_mask", "dim", "dtype", "else", "expand", "fill_value", "finfo", "full", "get_text_config", "getattr", "if", "int", "is", "is_sliding", "is_static_sliding_cache", "isinstance", "mask_length", "masked_fill", "min", "min_dtype", "not", "or", "padding_mask", "past_key_values", "reshape", "return", "sequence_length", "shape", "sliding_attend_mask", "sliding_window", "target_length", "text_config", "to", "torch", "use_sliding_window"], "kyutai_speech_to_text/modeling_kyutai_speech_to_text.py:KyutaiSpeechToTextForConditionalGeneration.__init__": ["AutoModel", "False", "GenerationConfig", "GenerationMixin", "Linear", "ModelForConditionalGeneration", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "codec_config", "codec_model", "config", "def", "from_config", "from_model_config", "generation_config", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "vocab_size"], "kyutai_speech_to_text/modeling_kyutai_speech_to_text.py:KyutaiSpeechToTextForConditionalGeneration.forward": ["BaseModelOutputWithPast", "Cache", "CausalLMOutputWithPast", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "outputs", "past_key_values", "position_ids", "r", "return", "self", "slice", "slice_indices", "torch", "use_cache", "vocab_size"], "kyutai_speech_to_text/modeling_kyutai_speech_to_text.py:KyutaiSpeechToTextForConditionalGeneration._prepare_generation_config": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "_prepare_generation_config", "args", "audio_window_size", "class", "def", "else", "generation_config", "hasattr", "if", "kwargs", "model_kwargs", "return", "self", "super"], "kyutai_speech_to_text/modeling_kyutai_speech_to_text.py:KyutaiSpeechToTextForConditionalGeneration._prepare_model_inputs": ["GenerationMixin", "MethodType", "ModelConv1dPaddingCache", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Tensor", "True", "__func__", "_get_cache", "_mimiconv1d_layer_names", "_prepare_cache_for_generation", "_prepare_model_inputs", "_supports_default_dynamic_cache", "append", "audio_tokens", "audio_window_size", "batch_size", "bos_token_id", "cache_implementation", "cache_methods", "class", "codec_config", "codec_model", "config", "contiguous", "current_window", "def", "device", "dict", "downsample", "dtype", "dynamic", "encoder", "encoder_past_key_values", "expand", "for", "generation_config", "generation_mode", "get", "get_encoded_length", "get_submodule", "getattr", "if", "in", "in_channels", "input_name", "input_values", "inputs", "is", "item", "lambda", "layer_name", "len", "long", "max_cache_length", "method", "model_kwargs", "num_codebooks", "num_layers", "pad_mode", "padding_cache", "padding_total", "past_key_values", "per_layer_in_channels", "per_layer_padding", "per_layer_padding_mode", "return", "self", "setattr", "shape", "sliding_window", "str", "super", "temporary_model_kwargs", "tensor", "torch", "tuple", "types", "x", "zeros"], "kyutai_speech_to_text/modeling_kyutai_speech_to_text.py:KyutaiSpeechToTextForConditionalGeneration.prepare_inputs_for_generation": ["Cache", "FloatTensor", "GenerationMixin", "LongTensor", "ModelConv1dPaddingCache", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Tensor", "args", "audio_bos_token_id", "audio_codes", "audio_tokens", "audio_window_size", "cache_position", "cat", "clamp", "class", "clone", "codec_model", "codec_model_output", "config", "copy_", "current_audio_tokens", "current_audio_tokens_idxs", "current_input_values", "current_window", "def", "device", "dim", "encode", "encoder_past_key_values", "end", "expand", "frame_size", "if", "input_ids", "input_values", "input_values_end_idx", "input_values_start_idx", "int", "is", "kwargs", "min", "model_inputs", "new_audio_tokens", "no_grad", "not", "padding_cache", "padding_mask", "pop", "prepare_inputs_for_generation", "return", "self", "shape", "start", "super", "tensor", "torch", "transpose", "tuple", "unsqueeze", "with"], "kyutai_speech_to_text/modeling_kyutai_speech_to_text.py:KyutaiSpeechToTextForConditionalGeneration.from_pretrained": ["False", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "_from_model_config", "args", "attr", "class", "cls", "codec_", "codec_model", "codec_model_attrs", "def", "delattr", "else", "for", "from_pretrained", "generation_config", "get", "if", "in", "items", "kwargs", "len", "loading_info", "model", "output_loading_info", "prefix", "prefix_len", "return", "startswith", "super", "update", "value", "vars"], "kyutai_speech_to_text/modeling_kyutai_speech_to_text.py:KyutaiSpeechToTextForConditionalGeneration.save_pretrained": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "args", "attr", "class", "codec_", "codec_model", "codec_model_attrs", "def", "for", "generation_config", "in", "items", "kwargs", "pop", "prefix", "save_pretrained", "self", "setattr", "super", "to_diff_dict", "transformers_version", "value"], "kyutai_speech_to_text/modeling_kyutai_speech_to_text.py:KyutaiSpeechToTextForConditionalGeneration.generate": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Setting", "args", "audio", "class", "codec_config", "config", "def", "f", "frame_size", "frames", "generate", "get", "greater", "if", "input_values", "is", "kwargs", "logger", "max_audio_frames", "max_new_tokens", "maximum", "not", "number", "of", "or", "pop", "r", "return", "self", "shape", "super", "than", "the", "to", "warning"], "lasr/modeling_lasr.py:LasrEncoderSubsampling.__init__": ["Conv1d", "Linear", "ModelEncoderConfig", "ModelEncoderSubsampling", "Module", "ReLU", "__init__", "act_fn", "class", "config", "conv_0", "conv_1", "def", "dense_0", "dense_1", "hidden_size", "kernel_size", "nn", "num_mel_bins", "self", "stride", "subsampling_conv_channels", "subsampling_conv_kernel_size", "subsampling_conv_stride", "super"], "lasr/modeling_lasr.py:LasrEncoderSubsampling.forward": ["ModelEncoderSubsampling", "Module", "Tensor", "act_fn", "class", "conv_0", "conv_1", "def", "dense_0", "dense_1", "forward", "hidden_states", "input_features", "nn", "return", "self", "torch", "transpose"], "lasr/modeling_lasr.py:LasrEncoderRotaryEmbedding.__init__": ["Callable", "False", "ModelEncoderConfig", "ModelEncoderRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "lasr/modeling_lasr.py:LasrEncoderRotaryEmbedding.compute_default_rope_parameters": ["ModelEncoderConfig", "ModelEncoderRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "lasr/modeling_lasr.py:LasrEncoderRotaryEmbedding.forward": ["False", "ModelEncoderRotaryEmbedding", "Module", "None", "and", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "lasr/modeling_lasr.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "lasr/modeling_lasr.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cos", "def", "k", "k_embed", "q", "q_embed", "return", "rotate_half", "sin", "unsqueeze", "unsqueeze_dim"], "lasr/modeling_lasr.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "lasr/modeling_lasr.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value", "value_states"], "lasr/modeling_lasr.py:LasrEncoderAttention.__init__": ["False", "Linear", "ModelEncoderAttention", "ModelEncoderConfig", "Module", "__init__", "attention_bias", "attention_dropout", "bias", "class", "config", "def", "getattr", "head_dim", "hidden_size", "int", "is_causal", "k_proj", "layer_idx", "nn", "num_attention_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "q_proj", "scaling", "self", "super", "v_proj"], "lasr/modeling_lasr.py:LasrEncoderAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "ModelEncoderAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_attn_implementation", "apply_rotary_pos_emb", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "class", "config", "contiguous", "cos", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "head_dim", "hidden_shape", "hidden_states", "if", "input_shape", "k_proj", "key_states", "kwargs", "nn", "not", "o_proj", "position_embeddings", "q_proj", "query_states", "reshape", "return", "scaling", "self", "shape", "sin", "torch", "training", "transpose", "tuple", "v_proj", "value_states", "view"], "lasr/modeling_lasr.py:LasrEncoderConvolutionModule.__init__": ["ACT2FN", "BatchNorm1d", "Conv1d", "ModelEncoderConfig", "ModelEncoderConvolutionModule", "Module", "None", "__init__", "activation", "batch_norm_momentum", "bias", "channels", "class", "config", "conv_kernel_size", "convolution_bias", "def", "depthwise_conv", "else", "get", "getattr", "groups", "hidden_act", "hidden_size", "if", "is", "kernel_size", "module_config", "momentum", "nn", "norm", "padding", "pointwise_conv1", "pointwise_conv2", "same", "self", "silu", "stride", "super"], "lasr/modeling_lasr.py:LasrEncoderConvolutionModule.forward": ["ModelEncoderConvolutionModule", "Module", "None", "activation", "all", "all_masked_rows", "attention_mask", "bool", "class", "def", "depthwise_conv", "dim", "dtype", "else", "forward", "functional", "glu", "hidden_states", "if", "is", "masked_fill", "nn", "norm", "not", "pointwise_conv1", "pointwise_conv2", "return", "self", "torch", "transpose"], "lasr/modeling_lasr.py:LasrEncoderFeedForward.__init__": ["ACT2FN", "Linear", "ModelEncoderConfig", "ModelEncoderFeedForward", "Module", "__init__", "activation", "activation_dropout", "attention_bias", "bias", "class", "config", "def", "hidden_act", "hidden_size", "intermediate_size", "linear1", "linear2", "nn", "self", "super"], "lasr/modeling_lasr.py:LasrEncoderFeedForward.forward": ["ModelEncoderFeedForward", "Module", "activation", "activation_dropout", "class", "def", "dropout", "forward", "functional", "hidden_states", "linear1", "linear2", "nn", "p", "return", "self", "training"], "lasr/modeling_lasr.py:LasrEncoderBlock.__init__": ["False", "GradientCheckpointingLayer", "LayerNorm", "ModelEncoderAttention", "ModelEncoderBlock", "ModelEncoderConfig", "ModelEncoderConvolutionModule", "ModelEncoderFeedForward", "__init__", "bias", "class", "config", "conv", "conv_residual_weights", "def", "feed_forward1", "feed_forward2", "feed_forward_residual_weights", "gradient_checkpointing", "hidden_size", "int", "layer_idx", "layer_norm_eps", "nn", "norm_conv", "norm_feed_forward1", "norm_feed_forward2", "norm_out", "norm_self_att", "self", "self_attn", "super"], "lasr/modeling_lasr.py:LasrEncoderBlock.forward": ["GradientCheckpointingLayer", "ModelEncoderBlock", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "attention_mask", "attn_output", "class", "conv", "conv_output", "conv_residual_weights", "def", "feed_forward1", "feed_forward2", "feed_forward_residual_weights", "forward", "hidden_states", "kwargs", "norm_conv", "norm_feed_forward1", "norm_feed_forward2", "norm_out", "norm_self_att", "normalized_hidden_states", "position_embeddings", "residual", "return", "self", "self_attn", "torch"], "lasr/modeling_lasr.py:LasrPreTrainedModel._init_weights": ["ModelPreTrainedModel", "PreTrainedModel", "_init_weights", "class", "def", "module", "self", "super"], "lasr/modeling_lasr.py:LasrPreTrainedModel._get_subsampling_output_length": ["ModelCTCConfig", "ModelPreTrainedModel", "PreTrainedModel", "Tensor", "_", "_get_subsampling_output_length", "class", "config", "def", "else", "encoder_config", "for", "if", "in", "input_lengths", "isinstance", "kernel_size", "num_layers", "range", "return", "self", "stride", "subsampling_conv_kernel_size", "subsampling_conv_stride", "torch"], "lasr/modeling_lasr.py:LasrPreTrainedModel._get_output_attention_mask": ["ModelPreTrainedModel", "None", "PreTrainedModel", "Tensor", "_get_output_attention_mask", "_get_subsampling_output_length", "arange", "attention_mask", "class", "def", "device", "else", "if", "int", "is", "max", "max_length", "not", "output_lengths", "return", "self", "sum", "target_length", "torch"], "lasr/modeling_lasr.py:LasrEncoder.__init__": ["False", "LayerNorm", "ModelEncoder", "ModelEncoderBlock", "ModelEncoderConfig", "ModelEncoderRotaryEmbedding", "ModelEncoderSubsampling", "ModelPreTrainedModel", "ModuleList", "__init__", "bias", "class", "config", "def", "dropout", "dropout_positions", "eps", "for", "gradient_checkpointing", "hidden_size", "in", "layer_idx", "layer_norm_eps", "layerdrop", "layers", "nn", "num_hidden_layers", "out_norm", "post_init", "range", "rotary_emb", "self", "subsampler", "super"], "lasr/modeling_lasr.py:LasrEncoder.forward": ["BaseModelOutput", "False", "ModelEncoder", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "_get_output_attention_mask", "arange", "attention_mask", "class", "config", "cos", "create_bidirectional_mask", "def", "device", "dropout", "dropout_positions", "dropout_probability", "encoder_layer", "for", "forward", "functional", "hidden_states", "if", "in", "input_embeds", "input_features", "is", "kwargs", "last_hidden_state", "layerdrop", "layers", "nn", "not", "out_norm", "p", "position_embeddings", "r", "rand", "return", "rotary_emb", "self", "shape", "sin", "subsampler", "target_length", "to_drop", "torch", "training", "unsqueeze"], "lasr/modeling_lasr.py:LasrForCTC.__init__": ["Conv1d", "ModelCTCConfig", "ModelEncoder", "ModelForCTC", "ModelPreTrainedModel", "__init__", "class", "config", "ctc_head", "def", "encoder", "encoder_config", "hidden_size", "kernel_size", "nn", "post_init", "self", "super", "vocab_size"], "lasr/modeling_lasr.py:LasrForCTC.forward": ["CausalLMOutput", "False", "ModelForCTC", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "_get_subsampling_output_length", "attention_mask", "attentions", "backends", "blank", "class", "config", "ctc_head", "ctc_loss", "ctc_loss_reduction", "ctc_zero_infinity", "cudnn", "def", "dim", "dtype", "else", "enabled", "encoder", "encoder_outputs", "flags", "flattened_targets", "float32", "forward", "functional", "hidden_states", "if", "input_features", "input_lengths", "is", "kwargs", "labels", "labels_mask", "last_hidden_state", "log_probs", "log_softmax", "logits", "long", "loss", "masked_select", "nn", "not", "ones_like", "pad_token_id", "r", "reduction", "return", "self", "sum", "target_lengths", "torch", "transpose", "with", "zero_infinity"], "lasr/modeling_lasr.py:LasrForCTC.generate": ["CausalLMOutput", "False", "LongTensor", "ModelForCTC", "ModelGenerateOutput", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "_get_output_attention_mask", "argmax", "attention_mask", "attentions", "bool", "class", "config", "def", "dim", "forward", "generate", "hidden_states", "if", "input_features", "is", "kwargs", "logits", "not", "outputs", "pad_token_id", "r", "return", "return_dict", "return_dict_in_generate", "self", "sequences", "shape", "target_length", "torch"], "layoutlm/modeling_layoutlm.py:LayoutLMEmbeddings.__init__": ["Dropout", "Embedding", "False", "LayerNorm", "ModelEmbeddings", "ModelLayerNorm", "Module", "__init__", "arange", "class", "config", "def", "dropout", "eps", "expand", "h_position_embeddings", "hidden_dropout_prob", "hidden_size", "layer_norm_eps", "max_2d_position_embeddings", "max_position_embeddings", "nn", "pad_token_id", "padding_idx", "persistent", "position_embeddings", "position_ids", "register_buffer", "self", "super", "token_type_embeddings", "torch", "type_vocab_size", "vocab_size", "w_position_embeddings", "word_embeddings", "x_position_embeddings", "y_position_embeddings"], "layoutlm/modeling_layoutlm.py:LayoutLMEmbeddings.forward": ["IndexError", "LayerNorm", "ModelEmbeddings", "Module", "None", "The", "as", "bbox", "be", "class", "coordinate", "def", "device", "dropout", "dtype", "e", "else", "embeddings", "except", "forward", "from", "h_position_embeddings", "if", "input_ids", "input_shape", "inputs_embeds", "is", "left_position_embeddings", "long", "lower_position_embeddings", "nn", "not", "position_embeddings", "position_ids", "raise", "range", "return", "right_position_embeddings", "self", "seq_length", "should", "size", "token_type_embeddings", "token_type_ids", "torch", "try", "upper_position_embeddings", "values", "w_position_embeddings", "within", "word_embeddings", "words_embeddings", "x_position_embeddings", "y_position_embeddings", "zeros"], "layoutlm/modeling_layoutlm.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value"], "layoutlm/modeling_layoutlm.py:LayoutLMSelfAttention.__init__": ["Dropout", "Linear", "ModelSelfAttention", "Module", "The", "ValueError", "__init__", "a", "all_head_size", "and", "attention", "attention_dropout", "attention_head_size", "attention_probs_dropout_prob", "class", "config", "def", "dropout", "embedding_size", "f", "hasattr", "heads", "hidden", "hidden_size", "if", "int", "is", "key", "multiple", "nn", "not", "num_attention_heads", "number", "of", "query", "raise", "scaling", "self", "size", "super", "the", "value"], "layoutlm/modeling_layoutlm.py:LayoutLMSelfAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "False", "FloatTensor", "ModelSelfAttention", "Module", "None", "Tensor", "_attn_implementation", "attention_dropout", "attention_head_size", "attention_interface", "attention_mask", "attn_output", "attn_weights", "bool", "class", "config", "contiguous", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "hidden_shape", "hidden_states", "if", "input_shape", "key", "key_states", "kwargs", "nn", "not", "output_attentions", "outputs", "query", "query_states", "reshape", "return", "scaling", "self", "shape", "torch", "training", "transpose", "tuple", "value", "value_states", "view"], "layoutlm/modeling_layoutlm.py:LayoutLMSelfOutput.__init__": ["Dropout", "LayerNorm", "Linear", "ModelSelfOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "eps", "hidden_dropout_prob", "hidden_size", "layer_norm_eps", "nn", "self", "super"], "layoutlm/modeling_layoutlm.py:LayoutLMSelfOutput.forward": ["LayerNorm", "ModelSelfOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "layoutlm/modeling_layoutlm.py:LayoutLMAttention.__init__": ["ModelAttention", "ModelSelfAttention", "ModelSelfOutput", "Module", "__init__", "class", "config", "def", "nn", "output", "self", "super"], "layoutlm/modeling_layoutlm.py:LayoutLMAttention.forward": ["False", "FloatTensor", "ModelAttention", "Module", "None", "Tensor", "attention_mask", "attention_output", "bool", "class", "def", "forward", "hidden_states", "kwargs", "nn", "output", "output_attentions", "outputs", "return", "self", "self_outputs", "torch", "tuple"], "layoutlm/modeling_layoutlm.py:LayoutLMIntermediate.__init__": ["ACT2FN", "Linear", "ModelIntermediate", "Module", "__init__", "class", "config", "def", "dense", "else", "hidden_act", "hidden_size", "if", "intermediate_act_fn", "intermediate_size", "isinstance", "nn", "self", "str", "super"], "layoutlm/modeling_layoutlm.py:LayoutLMIntermediate.forward": ["ModelIntermediate", "Module", "Tensor", "class", "def", "dense", "forward", "hidden_states", "intermediate_act_fn", "nn", "return", "self", "torch"], "layoutlm/modeling_layoutlm.py:LayoutLMOutput.__init__": ["Dropout", "LayerNorm", "Linear", "ModelOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "eps", "hidden_dropout_prob", "hidden_size", "intermediate_size", "layer_norm_eps", "nn", "self", "super"], "layoutlm/modeling_layoutlm.py:LayoutLMOutput.forward": ["LayerNorm", "ModelOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "layoutlm/modeling_layoutlm.py:LayoutLMLayer.__init__": ["GradientCheckpointingLayer", "ModelAttention", "ModelIntermediate", "ModelLayer", "ModelOutput", "__init__", "attention", "chunk_size_feed_forward", "class", "config", "def", "intermediate", "output", "self", "seq_len_dim", "super"], "layoutlm/modeling_layoutlm.py:LayoutLMLayer.forward": ["False", "FloatTensor", "GradientCheckpointingLayer", "ModelLayer", "None", "Tensor", "apply_chunking_to_forward", "attention", "attention_mask", "attention_output", "bool", "chunk_size_feed_forward", "class", "def", "feed_forward_chunk", "forward", "hidden_states", "kwargs", "layer_output", "output_attentions", "outputs", "return", "self", "self_attention_outputs", "seq_len_dim", "torch", "tuple"], "layoutlm/modeling_layoutlm.py:LayoutLMLayer.feed_forward_chunk": ["GradientCheckpointingLayer", "ModelLayer", "attention_output", "class", "def", "feed_forward_chunk", "intermediate", "intermediate_output", "layer_output", "output", "return", "self"], "layoutlm/modeling_layoutlm.py:LayoutLMEncoder.__init__": ["False", "ModelEncoder", "ModelLayer", "Module", "ModuleList", "__init__", "class", "config", "def", "for", "gradient_checkpointing", "i", "in", "layer", "nn", "num_hidden_layers", "range", "self", "super"], "layoutlm/modeling_layoutlm.py:LayoutLMEncoder.forward": ["BaseModelOutput", "False", "FloatTensor", "ModelEncoder", "Module", "None", "Tensor", "True", "all_hidden_states", "all_self_attentions", "attention_mask", "attentions", "bool", "class", "def", "else", "enumerate", "for", "forward", "hidden_states", "i", "if", "in", "kwargs", "last_hidden_state", "layer", "layer_module", "layer_outputs", "nn", "output_attentions", "output_hidden_states", "return", "return_dict", "self", "torch", "tuple"], "layoutlm/modeling_layoutlm.py:LayoutLMPooler.__init__": ["Linear", "ModelPooler", "Module", "Tanh", "__init__", "activation", "class", "config", "def", "dense", "hidden_size", "nn", "self", "super"], "layoutlm/modeling_layoutlm.py:LayoutLMPooler.forward": ["ModelPooler", "Module", "Tensor", "activation", "class", "def", "dense", "first_token_tensor", "forward", "hidden_states", "nn", "pooled_output", "return", "self", "torch"], "layoutlm/modeling_layoutlm.py:LayoutLMPredictionHeadTransform.__init__": ["ACT2FN", "LayerNorm", "Linear", "ModelPredictionHeadTransform", "Module", "__init__", "class", "config", "def", "dense", "else", "eps", "hidden_act", "hidden_size", "if", "isinstance", "layer_norm_eps", "nn", "self", "str", "super", "transform_act_fn"], "layoutlm/modeling_layoutlm.py:LayoutLMPredictionHeadTransform.forward": ["LayerNorm", "ModelPredictionHeadTransform", "Module", "Tensor", "class", "def", "dense", "forward", "hidden_states", "nn", "return", "self", "torch", "transform_act_fn"], "layoutlm/modeling_layoutlm.py:LayoutLMLMPredictionHead.__init__": ["Linear", "ModelLMPredictionHead", "ModelPredictionHeadTransform", "Module", "Parameter", "True", "__init__", "bias", "class", "config", "decoder", "def", "hidden_size", "nn", "self", "super", "torch", "transform", "vocab_size", "zeros"], "layoutlm/modeling_layoutlm.py:LayoutLMLMPredictionHead.forward": ["ModelLMPredictionHead", "Module", "class", "decoder", "def", "forward", "hidden_states", "nn", "return", "self", "transform"], "layoutlm/modeling_layoutlm.py:LayoutLMOnlyMLMHead.__init__": ["ModelLMPredictionHead", "ModelOnlyMLMHead", "Module", "__init__", "class", "config", "def", "nn", "predictions", "self", "super"], "layoutlm/modeling_layoutlm.py:LayoutLMOnlyMLMHead.forward": ["ModelOnlyMLMHead", "Module", "Tensor", "class", "def", "forward", "nn", "prediction_scores", "predictions", "return", "self", "sequence_output", "torch"], "layoutlm/modeling_layoutlm.py:LayoutLMPreTrainedModel._init_weights": ["ModelEmbeddings", "ModelLMPredictionHead", "ModelPreTrainedModel", "PreTrainedModel", "_init_weights", "arange", "bias", "class", "copy_", "def", "elif", "expand", "if", "init", "isinstance", "module", "position_ids", "self", "shape", "super", "torch", "zeros_"], "layoutlm/modeling_layoutlm.py:LayoutLMModel.__init__": ["ModelEmbeddings", "ModelEncoder", "ModelModel", "ModelPooler", "ModelPreTrainedModel", "__init__", "class", "config", "def", "embeddings", "encoder", "pooler", "post_init", "self", "super"], "layoutlm/modeling_layoutlm.py:LayoutLMModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "return", "self", "word_embeddings"], "layoutlm/modeling_layoutlm.py:LayoutLMModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "self", "set_input_embeddings", "value", "word_embeddings"], "layoutlm/modeling_layoutlm.py:LayoutLMModel.forward": ["BaseModelOutputWithPooling", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "True", "ValueError", "You", "and", "at", "attention_mask", "attentions", "bbox", "bool", "both", "cannot", "class", "config", "def", "device", "dtype", "either", "elif", "else", "embedding_output", "embeddings", "encoder", "encoder_outputs", "extended_attention_mask", "finfo", "forward", "have", "hidden_states", "if", "input_ids", "input_shape", "inputs_embeds", "is", "kwargs", "last_hidden_state", "long", "min", "not", "ones", "or", "output_attentions", "output_hidden_states", "pooled_output", "pooler", "pooler_output", "position_ids", "r", "raise", "return", "return_dict", "same", "self", "sequence_output", "size", "specify", "the", "time", "to", "token_type_ids", "torch", "tuple", "unsqueeze", "use_return_dict", "warn_if_padding_and_no_attention_mask", "zeros"], "layoutlm/modeling_layoutlm.py:LayoutLMForMaskedLM.__init__": ["Model", "ModelForMaskedLM", "ModelModel", "ModelOnlyMLMHead", "ModelPreTrainedModel", "__init__", "class", "cls", "config", "def", "post_init", "self", "super"], "layoutlm/modeling_layoutlm.py:LayoutLMForMaskedLM.get_input_embeddings": ["Model", "ModelForMaskedLM", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "return", "self", "word_embeddings"], "layoutlm/modeling_layoutlm.py:LayoutLMForMaskedLM.get_output_embeddings": ["ModelForMaskedLM", "ModelPreTrainedModel", "class", "cls", "decoder", "def", "get_output_embeddings", "predictions", "return", "self"], "layoutlm/modeling_layoutlm.py:LayoutLMForMaskedLM.set_output_embeddings": ["ModelForMaskedLM", "ModelPreTrainedModel", "bias", "class", "cls", "decoder", "def", "new_embeddings", "predictions", "self", "set_output_embeddings"], "layoutlm/modeling_layoutlm.py:LayoutLMForMaskedLM.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "MaskedLMOutput", "Model", "ModelForMaskedLM", "ModelPreTrainedModel", "None", "True", "attention_mask", "attentions", "bbox", "bool", "class", "cls", "config", "def", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "masked_lm_loss", "not", "output_attentions", "output_hidden_states", "outputs", "position_ids", "prediction_scores", "r", "return", "return_dict", "self", "sequence_output", "token_type_ids", "torch", "tuple", "use_return_dict", "view", "vocab_size"], "layoutlm/modeling_layoutlm.py:LayoutLMForSequenceClassification.__init__": ["Dropout", "Linear", "Model", "ModelForSequenceClassification", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "classifier", "config", "def", "dropout", "hidden_dropout_prob", "hidden_size", "nn", "num_labels", "post_init", "self", "super"], "layoutlm/modeling_layoutlm.py:LayoutLMForSequenceClassification.get_input_embeddings": ["Model", "ModelForSequenceClassification", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "return", "self", "word_embeddings"], "layoutlm/modeling_layoutlm.py:LayoutLMForSequenceClassification.forward": ["BCEWithLogitsLoss", "CrossEntropyLoss", "FloatTensor", "LongTensor", "MSELoss", "Model", "ModelForSequenceClassification", "ModelPreTrainedModel", "None", "SequenceClassifierOutput", "True", "and", "attention_mask", "attentions", "bbox", "bool", "class", "classifier", "config", "def", "dropout", "dtype", "elif", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "kwargs", "labels", "logits", "long", "loss", "loss_fct", "multi_label_classification", "not", "num_labels", "or", "output_attentions", "output_hidden_states", "outputs", "pooled_output", "position_ids", "problem_type", "r", "regression", "return", "return_dict", "self", "single_label_classification", "squeeze", "token_type_ids", "torch", "tuple", "use_return_dict", "view"], "layoutlm/modeling_layoutlm.py:LayoutLMForTokenClassification.__init__": ["Dropout", "Linear", "Model", "ModelForTokenClassification", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "classifier", "config", "def", "dropout", "hidden_dropout_prob", "hidden_size", "nn", "num_labels", "post_init", "self", "super"], "layoutlm/modeling_layoutlm.py:LayoutLMForTokenClassification.get_input_embeddings": ["Model", "ModelForTokenClassification", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "return", "self", "word_embeddings"], "layoutlm/modeling_layoutlm.py:LayoutLMForTokenClassification.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "Model", "ModelForTokenClassification", "ModelPreTrainedModel", "None", "TokenClassifierOutput", "True", "attention_mask", "attentions", "bbox", "bool", "class", "classifier", "config", "def", "dropout", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "not", "num_labels", "output_attentions", "output_hidden_states", "outputs", "position_ids", "r", "return", "return_dict", "self", "sequence_output", "token_type_ids", "torch", "tuple", "use_return_dict", "view"], "layoutlm/modeling_layoutlm.py:LayoutLMForQuestionAnswering.__init__": ["Linear", "Model", "ModelForQuestionAnswering", "ModelModel", "ModelPreTrainedModel", "True", "__init__", "class", "config", "def", "has_visual_segment_embedding", "hidden_size", "nn", "num_labels", "post_init", "qa_outputs", "r", "self", "super"], "layoutlm/modeling_layoutlm.py:LayoutLMForQuestionAnswering.get_input_embeddings": ["Model", "ModelForQuestionAnswering", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "return", "self", "word_embeddings"], "layoutlm/modeling_layoutlm.py:LayoutLMForQuestionAnswering.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "Model", "ModelForQuestionAnswering", "ModelPreTrainedModel", "None", "QuestionAnsweringModelOutput", "True", "and", "attention_mask", "attentions", "bbox", "bool", "clamp", "class", "config", "contiguous", "def", "dim", "else", "end_logits", "end_loss", "end_positions", "forward", "hidden_states", "if", "ignore_index", "ignored_index", "input_ids", "inputs_embeds", "is", "kwargs", "len", "logits", "loss", "loss_fct", "not", "output_attentions", "output_hidden_states", "outputs", "position_ids", "qa_outputs", "r", "return", "return_dict", "self", "sequence_output", "size", "split", "squeeze", "start_logits", "start_loss", "start_positions", "token_type_ids", "torch", "total_loss", "tuple", "use_return_dict"], "layoutlmv2/modeling_layoutlmv2.py:LayoutLMv2Embeddings.__init__": ["Dropout", "Embedding", "False", "LayerNorm", "ModelEmbeddings", "Module", "__init__", "arange", "class", "config", "coordinate_size", "def", "dropout", "eps", "expand", "h_position_embeddings", "hidden_dropout_prob", "hidden_size", "layer_norm_eps", "max_2d_position_embeddings", "max_position_embeddings", "nn", "pad_token_id", "padding_idx", "persistent", "position_embeddings", "position_ids", "register_buffer", "self", "shape_size", "super", "token_type_embeddings", "torch", "type_vocab_size", "vocab_size", "w_position_embeddings", "word_embeddings", "x_position_embeddings", "y_position_embeddings"], "layoutlmv2/modeling_layoutlmv2.py:LayoutLMv2Embeddings._calc_spatial_position_embeddings": ["IndexError", "ModelEmbeddings", "Module", "The", "_calc_spatial_position_embeddings", "as", "bbox", "be", "cat", "class", "coordinate", "def", "dim", "e", "except", "from", "h_position_embeddings", "left_position_embeddings", "lower_position_embeddings", "nn", "raise", "range", "return", "right_position_embeddings", "self", "should", "spatial_position_embeddings", "torch", "try", "upper_position_embeddings", "values", "w_position_embeddings", "within", "x_position_embeddings", "y_position_embeddings"], "layoutlmv2/modeling_layoutlmv2.py:LayoutLMv2SelfAttention.__init__": ["Dropout", "False", "Linear", "ModelSelfAttention", "Module", "Parameter", "The", "ValueError", "__init__", "a", "all_head_size", "and", "attention", "attention_head_size", "attention_probs_dropout_prob", "bias", "class", "config", "def", "dropout", "else", "embedding_size", "f", "fast_qkv", "has_relative_attention_bias", "has_spatial_attention_bias", "hasattr", "heads", "hidden", "hidden_size", "if", "int", "is", "key", "multiple", "nn", "not", "num_attention_heads", "number", "of", "q_bias", "qkv_linear", "query", "raise", "self", "size", "super", "the", "torch", "v_bias", "value", "zeros"], "layoutlmv2/modeling_layoutlmv2.py:LayoutLMv2SelfAttention.compute_qkv": ["ModelSelfAttention", "Module", "_sz", "chunk", "class", "compute_qkv", "def", "dim", "else", "fast_qkv", "hidden_states", "if", "k", "key", "ndimension", "nn", "q", "q_bias", "qkv", "qkv_linear", "query", "return", "self", "torch", "v", "v_bias", "value", "view"], "layoutlmv2/modeling_layoutlmv2.py:LayoutLMv2SelfAttention.forward": ["False", "ModelSelfAttention", "Module", "None", "_", "all_head_size", "attention_head_size", "attention_mask", "attention_probs", "attention_scores", "batch_size", "bool", "class", "compute_qkv", "context_layer", "contiguous", "def", "dim", "dropout", "dtype", "else", "finfo", "float", "float32", "forward", "functional", "has_relative_attention_bias", "has_spatial_attention_bias", "hidden_states", "if", "key", "key_layer", "masked_fill_", "math", "matmul", "min", "new_context_layer_shape", "nn", "num_attention_heads", "output_attentions", "outputs", "permute", "query", "query_layer", "rel_2d_pos", "rel_pos", "return", "self", "seq_length", "shape", "size", "softmax", "sqrt", "to", "torch", "transpose", "type_as", "value", "value_layer", "view"], "layoutlmv2/modeling_layoutlmv2.py:LayoutLMv2Attention.__init__": ["ModelAttention", "ModelSelfAttention", "ModelSelfOutput", "Module", "__init__", "class", "config", "def", "nn", "output", "self", "super"], "layoutlmv2/modeling_layoutlmv2.py:LayoutLMv2Attention.forward": ["False", "ModelAttention", "Module", "None", "attention_mask", "attention_output", "class", "def", "forward", "hidden_states", "nn", "output", "output_attentions", "outputs", "rel_2d_pos", "rel_pos", "return", "self", "self_outputs"], "layoutlmv2/modeling_layoutlmv2.py:LayoutLMv2SelfOutput.__init__": ["Dropout", "LayerNorm", "Linear", "ModelSelfOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "eps", "hidden_dropout_prob", "hidden_size", "layer_norm_eps", "nn", "self", "super"], "layoutlmv2/modeling_layoutlmv2.py:LayoutLMv2SelfOutput.forward": ["LayerNorm", "ModelSelfOutput", "Module", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self"], "layoutlmv2/modeling_layoutlmv2.py:LayoutLMv2Intermediate.__init__": ["ACT2FN", "Linear", "ModelIntermediate", "Module", "__init__", "class", "config", "def", "dense", "else", "hidden_act", "hidden_size", "if", "intermediate_act_fn", "intermediate_size", "isinstance", "nn", "self", "str", "super"], "layoutlmv2/modeling_layoutlmv2.py:LayoutLMv2Intermediate.forward": ["ModelIntermediate", "Module", "Tensor", "class", "def", "dense", "forward", "hidden_states", "intermediate_act_fn", "nn", "return", "self", "torch"], "layoutlmv2/modeling_layoutlmv2.py:LayoutLMv2Output.__init__": ["Dropout", "LayerNorm", "Linear", "ModelOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "eps", "hidden_dropout_prob", "hidden_size", "intermediate_size", "layer_norm_eps", "nn", "self", "super"], "layoutlmv2/modeling_layoutlmv2.py:LayoutLMv2Output.forward": ["LayerNorm", "ModelOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "layoutlmv2/modeling_layoutlmv2.py:LayoutLMv2Layer.__init__": ["GradientCheckpointingLayer", "ModelAttention", "ModelIntermediate", "ModelLayer", "ModelOutput", "__init__", "attention", "chunk_size_feed_forward", "class", "config", "def", "intermediate", "output", "self", "seq_len_dim", "super"], "layoutlmv2/modeling_layoutlmv2.py:LayoutLMv2Layer.forward": ["False", "GradientCheckpointingLayer", "ModelLayer", "None", "apply_chunking_to_forward", "attention", "attention_mask", "attention_output", "chunk_size_feed_forward", "class", "def", "feed_forward_chunk", "forward", "hidden_states", "layer_output", "output_attentions", "outputs", "rel_2d_pos", "rel_pos", "return", "self", "self_attention_outputs", "seq_len_dim"], "layoutlmv2/modeling_layoutlmv2.py:LayoutLMv2Layer.feed_forward_chunk": ["GradientCheckpointingLayer", "ModelLayer", "attention_output", "class", "def", "feed_forward_chunk", "intermediate", "intermediate_output", "layer_output", "output", "return", "self"], "layoutlmv2/modeling_layoutlmv2.py:relative_position_bucket": ["Model_position", "Model_position_bucket", "True", "abs", "bidirectional", "def", "else", "float", "full_like", "if", "is_small", "log", "long", "math", "max", "max_distance", "max_exact", "min", "n", "num_buckets", "ret", "return", "to", "torch", "val_if_large", "where", "zeros_like"], "layoutlmv2/modeling_layoutlmv2.py:LayoutLMv2Encoder.__init__": ["False", "Linear", "ModelEncoder", "ModelLayer", "Module", "ModuleList", "_", "__init__", "bias", "class", "config", "def", "for", "gradient_checkpointing", "has_relative_attention_bias", "has_spatial_attention_bias", "if", "in", "layer", "max_rel_2d_pos", "max_rel_pos", "nn", "num_attention_heads", "num_hidden_layers", "range", "rel_2d_pos_bins", "rel_pos_bias", "rel_pos_bins", "rel_pos_x_bias", "rel_pos_y_bias", "self", "super"], "layoutlmv2/modeling_layoutlmv2.py:LayoutLMv2Encoder._calculate_1d_position_embeddings": ["ModelEncoder", "Module", "_calculate_1d_position_embeddings", "class", "contiguous", "def", "max_distance", "max_rel_pos", "nn", "no_grad", "num_buckets", "permute", "position_ids", "rel_pos", "rel_pos_bias", "rel_pos_bins", "rel_pos_mat", "relative_position_bucket", "return", "self", "t", "torch", "unsqueeze", "weight", "with"], "layoutlmv2/modeling_layoutlmv2.py:LayoutLMv2Encoder._calculate_2d_position_embeddings": ["ModelEncoder", "Module", "_calculate_2d_position_embeddings", "bbox", "class", "contiguous", "def", "max_distance", "max_rel_2d_pos", "nn", "no_grad", "num_buckets", "permute", "position_coord_x", "position_coord_y", "rel_2d_pos", "rel_2d_pos_bins", "rel_pos_x", "rel_pos_x_2d_mat", "rel_pos_x_bias", "rel_pos_y", "rel_pos_y_2d_mat", "rel_pos_y_bias", "relative_position_bucket", "return", "self", "t", "torch", "unsqueeze", "weight", "with"], "layoutlmv2/modeling_layoutlmv2.py:LayoutLMv2Encoder.forward": ["BaseModelOutput", "False", "ModelEncoder", "Module", "None", "True", "_calculate_1d_position_embeddings", "_calculate_2d_position_embeddings", "all_hidden_states", "all_self_attentions", "attention_mask", "attentions", "bbox", "class", "def", "else", "enumerate", "for", "forward", "has_relative_attention_bias", "has_spatial_attention_bias", "hidden_states", "i", "if", "in", "is", "last_hidden_state", "layer", "layer_module", "layer_outputs", "nn", "not", "output_attentions", "output_hidden_states", "position_ids", "rel_2d_pos", "rel_pos", "return", "return_dict", "self", "tuple", "v"], "layoutlmv2/modeling_layoutlmv2.py:LayoutLMv2PreTrainedModel._init_weights": ["FrozenBatchNorm2d", "MODEL", "ModelEmbeddings", "ModelModel", "ModelPreTrainedModel", "ModelSelfAttention", "ModelVisualBackbone", "PIXEL_MEAN", "PIXEL_STD", "PreTrainedModel", "Tensor", "_init_weights", "arange", "bias", "cfg", "class", "config", "constant_", "copy_", "def", "detectron2", "elif", "eps", "expand", "fast_qkv", "hasattr", "if", "init", "initializer_range", "isinstance", "layers", "len", "mean", "module", "normal_", "num_channels", "ones_", "pixel_mean", "pixel_std", "position_ids", "q_bias", "running_mean", "running_var", "self", "shape", "std", "super", "torch", "v_bias", "view", "visual_segment_embedding", "weight", "zeros_"], "layoutlmv2/modeling_layoutlmv2.py:my_convert_sync_batchnorm": ["FrozenBatchNorm2d", "None", "Parameter", "SyncBatchNorm", "True", "_BatchNorm", "add_module", "affine", "batchnorm", "bias", "child", "convert_sync_batchnorm", "def", "del", "detectron2", "device", "dtype", "eps", "for", "if", "in", "isinstance", "layers", "long", "module", "module_output", "modules", "my_convert_sync_batchnorm", "name", "named_children", "nn", "num_batches_tracked", "num_features", "process_group", "return", "running_mean", "running_var", "tensor", "torch", "track_running_stats", "weight"], "layoutlmv2/modeling_layoutlmv2.py:LayoutLMv2VisualBackbone.__init__": ["AdaptiveAvgPool2d", "AvgPool2d", "FPN", "False", "META_ARCHITECTURE", "META_ARCH_REGISTRY", "MODEL", "ModelVisualBackbone", "Module", "PIXEL_MEAN", "PIXEL_STD", "Tensor", "__init__", "append", "are_deterministic_algorithms_enabled", "assert", "backbone", "backbone_stride", "ceil", "cfg", "channels", "class", "config", "def", "detectron2", "else", "get", "get_detectron2_config", "if", "image_feature_pool_shape", "input_shape", "instead", "isinstance", "len", "logger", "math", "meta_arch", "model", "modeling", "nn", "num_channels", "of", "out_feature_key", "output_shape", "p2", "persistent", "pixel_mean", "pixel_std", "pool", "register_buffer", "self", "stride", "super", "torch", "using", "view", "warning"], "layoutlmv2/modeling_layoutlmv2.py:LayoutLMv2VisualBackbone.forward": ["ModelVisualBackbone", "Module", "backbone", "class", "contiguous", "def", "else", "features", "flatten", "forward", "if", "images", "images_input", "is_tensor", "nn", "out_feature_key", "pixel_mean", "pixel_std", "pool", "return", "self", "start_dim", "tensor", "torch", "transpose"], "layoutlmv2/modeling_layoutlmv2.py:LayoutLMv2VisualBackbone.synchronize_batch_norm": ["Make", "ModelVisualBackbone", "Module", "RuntimeError", "and", "backbone", "be", "by", "can", "class", "cuda", "def", "device_count", "distributed", "divided", "for", "get_rank", "get_world_size", "i", "if", "in", "is", "is_available", "is_initialized", "list", "my_convert_sync_batchnorm", "new_group", "nn", "node_global_ranks", "node_rank", "node_size", "nodes", "not", "number", "of", "process_group", "processes", "properly", "raise", "range", "ranks", "self", "self_rank", "set", "sure", "sync_bn_groups", "synchronize_batch_norm", "the", "torch", "up", "world_size"], "layoutlmv2/modeling_layoutlmv2.py:LayoutLMv2Pooler.__init__": ["Linear", "ModelPooler", "Module", "Tanh", "__init__", "activation", "class", "config", "def", "dense", "hidden_size", "nn", "self", "super"], "layoutlmv2/modeling_layoutlmv2.py:LayoutLMv2Pooler.forward": ["ModelPooler", "Module", "activation", "class", "def", "dense", "first_token_tensor", "forward", "hidden_states", "nn", "pooled_output", "return", "self"], "layoutlmv2/modeling_layoutlmv2.py:LayoutLMv2Model.__init__": ["Dropout", "Embedding", "LayerNorm", "Linear", "ModelEmbeddings", "ModelEncoder", "ModelModel", "ModelPooler", "ModelPreTrainedModel", "ModelVisualBackbone", "Parameter", "__init__", "class", "config", "def", "detectron2", "embeddings", "encoder", "eps", "has_visual_segment_embedding", "hidden_dropout_prob", "hidden_size", "if", "image_feature_pool_shape", "layer_norm_eps", "nn", "pooler", "post_init", "requires_backends", "self", "super", "visual", "visual_LayerNorm", "visual_dropout", "visual_proj", "visual_segment_embedding", "weight"], "layoutlmv2/modeling_layoutlmv2.py:LayoutLMv2Model.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "return", "self", "word_embeddings"], "layoutlmv2/modeling_layoutlmv2.py:LayoutLMv2Model.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "self", "set_input_embeddings", "value", "word_embeddings"], "layoutlmv2/modeling_layoutlmv2.py:LayoutLMv2Model._calc_text_embeddings": ["LayerNorm", "ModelModel", "ModelPreTrainedModel", "None", "_calc_spatial_position_embeddings", "_calc_text_embeddings", "arange", "bbox", "class", "def", "device", "dropout", "dtype", "else", "embeddings", "expand_as", "if", "input_ids", "input_shape", "inputs_embeds", "is", "long", "not", "position_embeddings", "position_ids", "return", "self", "seq_length", "size", "spatial_position_embeddings", "token_type_embeddings", "token_type_ids", "torch", "unsqueeze", "word_embeddings", "zeros_like"], "layoutlmv2/modeling_layoutlmv2.py:LayoutLMv2Model._calc_img_embeddings": ["ModelModel", "ModelPreTrainedModel", "_calc_img_embeddings", "_calc_spatial_position_embeddings", "bbox", "class", "def", "embeddings", "has_visual_segment_embedding", "if", "image", "position_embeddings", "position_ids", "return", "self", "spatial_position_embeddings", "visual", "visual_LayerNorm", "visual_dropout", "visual_embeddings", "visual_proj", "visual_segment_embedding"], "layoutlmv2/modeling_layoutlmv2.py:LayoutLMv2Model._calc_visual_bbox": ["ModelModel", "ModelPreTrainedModel", "_calc_visual_bbox", "arange", "bbox", "class", "config", "def", "device", "dim", "div", "dtype", "final_shape", "floor", "image_feature_pool_shape", "repeat", "return", "rounding_mode", "self", "size", "stack", "torch", "transpose", "view", "visual_bbox", "visual_bbox_x", "visual_bbox_y"], "layoutlmv2/modeling_layoutlmv2.py:LayoutLMv2Model._get_input_shape": ["ModelModel", "ModelPreTrainedModel", "None", "ValueError", "You", "_get_input_shape", "and", "at", "both", "cannot", "class", "def", "either", "elif", "else", "have", "if", "input_ids", "inputs_embeds", "is", "not", "or", "raise", "return", "same", "self", "size", "specify", "the", "time", "to"], "layoutlmv2/modeling_layoutlmv2.py:LayoutLMv2Model.forward": ["BaseModelOutputWithPooling", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Size", "_calc_img_embeddings", "_calc_text_embeddings", "_calc_visual_bbox", "_get_input_shape", "arange", "attention_mask", "attentions", "bbox", "bool", "cat", "class", "config", "def", "device", "dim", "dtype", "else", "embeddings", "encoder", "encoder_outputs", "expand", "extended_attention_mask", "final_attention_mask", "final_bbox", "final_emb", "final_position_ids", "final_shape", "finfo", "forward", "hidden_states", "if", "image", "image_feature_pool_shape", "input_ids", "input_shape", "inputs_embeds", "is", "kwargs", "last_hidden_state", "list", "long", "min", "not", "ones", "output_attentions", "output_hidden_states", "pooled_output", "pooler", "pooler_output", "position_ids", "r", "repeat", "return", "return_dict", "self", "seq_length", "sequence_output", "text_Model_emb", "to", "token_type_ids", "torch", "tuple", "unsqueeze", "use_return_dict", "visual_attention_mask", "visual_bbox", "visual_emb", "visual_position_ids", "visual_shape", "zeros"], "layoutlmv2/modeling_layoutlmv2.py:LayoutLMv2ForSequenceClassification.__init__": ["Dropout", "Linear", "Model", "ModelForSequenceClassification", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "classifier", "config", "def", "dropout", "hidden_dropout_prob", "hidden_size", "nn", "num_labels", "post_init", "self", "super"], "layoutlmv2/modeling_layoutlmv2.py:LayoutLMv2ForSequenceClassification.get_input_embeddings": ["Model", "ModelForSequenceClassification", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "return", "self", "word_embeddings"], "layoutlmv2/modeling_layoutlmv2.py:LayoutLMv2ForSequenceClassification.forward": ["BCEWithLogitsLoss", "CrossEntropyLoss", "FloatTensor", "LongTensor", "MSELoss", "Model", "ModelForSequenceClassification", "ModelPreTrainedModel", "None", "SequenceClassifierOutput", "Size", "ValueError", "You", "_calc_img_embeddings", "_calc_visual_bbox", "and", "arange", "at", "attention_mask", "attentions", "bbox", "bool", "both", "cannot", "cat", "class", "classifier", "cls_final_output", "config", "def", "device", "dim", "dropout", "dtype", "either", "elif", "else", "final_image_embeddings", "final_shape", "forward", "have", "hidden_states", "if", "image", "image_feature_pool_shape", "initial_image_embeddings", "input_ids", "input_shape", "inputs_embeds", "int", "is", "kwargs", "labels", "list", "logits", "long", "loss", "loss_fct", "mean", "multi_label_classification", "not", "num_labels", "or", "output", "output_attentions", "output_hidden_states", "outputs", "pooled_final_image_embeddings", "pooled_initial_image_embeddings", "position_ids", "problem_type", "r", "raise", "regression", "repeat", "return", "return_dict", "same", "self", "seq_length", "sequence_output", "single_label_classification", "size", "specify", "squeeze", "the", "time", "to", "token_type_ids", "torch", "tuple", "use_return_dict", "view", "visual_bbox", "visual_position_ids", "visual_shape", "warn_if_padding_and_no_attention_mask"], "layoutlmv2/modeling_layoutlmv2.py:LayoutLMv2ForTokenClassification.__init__": ["Dropout", "Linear", "Model", "ModelForTokenClassification", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "classifier", "config", "def", "dropout", "hidden_dropout_prob", "hidden_size", "nn", "num_labels", "post_init", "self", "super"], "layoutlmv2/modeling_layoutlmv2.py:LayoutLMv2ForTokenClassification.get_input_embeddings": ["Model", "ModelForTokenClassification", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "return", "self", "word_embeddings"], "layoutlmv2/modeling_layoutlmv2.py:LayoutLMv2ForTokenClassification.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "Model", "ModelForTokenClassification", "ModelPreTrainedModel", "None", "TokenClassifierOutput", "attention_mask", "attentions", "bbox", "bool", "class", "classifier", "config", "def", "dropout", "else", "forward", "hidden_states", "if", "image", "input_ids", "input_shape", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "not", "num_labels", "output", "output_attentions", "output_hidden_states", "outputs", "position_ids", "r", "return", "return_dict", "self", "seq_length", "sequence_output", "size", "token_type_ids", "torch", "tuple", "use_return_dict", "view"], "layoutlmv2/modeling_layoutlmv2.py:LayoutLMv2ForQuestionAnswering.__init__": ["Linear", "Model", "ModelForQuestionAnswering", "ModelModel", "ModelPreTrainedModel", "True", "__init__", "class", "config", "def", "has_visual_segment_embedding", "hidden_size", "nn", "num_labels", "post_init", "qa_outputs", "r", "self", "super"], "layoutlmv2/modeling_layoutlmv2.py:LayoutLMv2ForQuestionAnswering.get_input_embeddings": ["Model", "ModelForQuestionAnswering", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "return", "self", "word_embeddings"], "layoutlmv2/modeling_layoutlmv2.py:LayoutLMv2ForQuestionAnswering.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "Model", "ModelForQuestionAnswering", "ModelPreTrainedModel", "None", "QuestionAnsweringModelOutput", "and", "attention_mask", "attentions", "bbox", "bool", "clamp", "class", "config", "contiguous", "def", "dim", "else", "end_logits", "end_loss", "end_positions", "forward", "hidden_states", "if", "ignore_index", "ignored_index", "image", "input_ids", "input_shape", "inputs_embeds", "is", "kwargs", "len", "logits", "loss", "loss_fct", "not", "output", "output_attentions", "output_hidden_states", "outputs", "position_ids", "qa_outputs", "r", "return", "return_dict", "self", "seq_length", "sequence_output", "size", "split", "squeeze", "start_logits", "start_loss", "start_positions", "token_type_ids", "torch", "total_loss", "tuple", "use_return_dict"], "layoutlmv3/modeling_layoutlmv3.py:LayoutLMv3PatchEmbeddings.__init__": ["Conv2d", "Iterable", "ModelPatchEmbeddings", "Module", "__init__", "abc", "class", "collections", "config", "def", "else", "hidden_size", "if", "image_size", "input_size", "isinstance", "kernel_size", "nn", "num_channels", "patch_shape", "patch_size", "proj", "self", "stride", "super"], "layoutlmv3/modeling_layoutlmv3.py:LayoutLMv3PatchEmbeddings.forward": ["F", "ModelPatchEmbeddings", "Module", "None", "bicubic", "class", "def", "embeddings", "flatten", "forward", "if", "interpolate", "is", "mode", "nn", "not", "patch_height", "patch_shape", "patch_width", "permute", "pixel_values", "position_embedding", "proj", "return", "self", "shape", "size", "transpose", "view"], "layoutlmv3/modeling_layoutlmv3.py:LayoutLMv3TextEmbeddings.__init__": ["Dropout", "Embedding", "False", "LayerNorm", "ModelTextEmbeddings", "Module", "__init__", "arange", "class", "config", "coordinate_size", "def", "dropout", "eps", "expand", "h_position_embeddings", "hidden_dropout_prob", "hidden_size", "layer_norm_eps", "max_2d_position_embeddings", "max_position_embeddings", "nn", "pad_token_id", "padding_idx", "persistent", "position_embeddings", "position_ids", "register_buffer", "self", "shape_size", "super", "token_type_embeddings", "torch", "type_vocab_size", "vocab_size", "w_position_embeddings", "word_embeddings", "x_position_embeddings", "y_position_embeddings"], "layoutlmv3/modeling_layoutlmv3.py:LayoutLMv3TextEmbeddings.calculate_spatial_position_embeddings": ["IndexError", "ModelTextEmbeddings", "Module", "The", "as", "bbox", "be", "calculate_spatial_position_embeddings", "cat", "class", "clip", "coordinate", "def", "dim", "e", "except", "from", "h_position_embeddings", "left_position_embeddings", "lower_position_embeddings", "nn", "raise", "range", "return", "right_position_embeddings", "self", "should", "spatial_position_embeddings", "torch", "try", "upper_position_embeddings", "values", "w_position_embeddings", "within", "x_position_embeddings", "y_position_embeddings"], "layoutlmv3/modeling_layoutlmv3.py:LayoutLMv3TextEmbeddings.create_position_ids_from_input_ids": ["ModelTextEmbeddings", "Module", "class", "create_position_ids_from_input_ids", "cumsum", "def", "dim", "incremental_indices", "input_ids", "int", "long", "mask", "ne", "nn", "padding_idx", "return", "self", "torch", "type_as"], "layoutlmv3/modeling_layoutlmv3.py:LayoutLMv3TextEmbeddings.create_position_ids_from_inputs_embeds": ["ModelTextEmbeddings", "Module", "arange", "class", "create_position_ids_from_inputs_embeds", "def", "device", "dtype", "expand", "input_shape", "inputs_embeds", "long", "nn", "padding_idx", "position_ids", "return", "self", "sequence_length", "size", "torch", "unsqueeze"], "layoutlmv3/modeling_layoutlmv3.py:LayoutLMv3TextEmbeddings.forward": ["LayerNorm", "ModelTextEmbeddings", "Module", "None", "bbox", "calculate_spatial_position_embeddings", "class", "create_position_ids_from_input_ids", "create_position_ids_from_inputs_embeds", "def", "device", "dropout", "dtype", "else", "embeddings", "forward", "if", "input_ids", "input_shape", "inputs_embeds", "is", "long", "nn", "not", "padding_idx", "position_embeddings", "position_ids", "return", "self", "size", "spatial_position_embeddings", "to", "token_type_embeddings", "token_type_ids", "torch", "word_embeddings", "zeros"], "layoutlmv3/modeling_layoutlmv3.py:LayoutLMv3PreTrainedModel._init_weights": ["ModelModel", "ModelPreTrainedModel", "ModelTextEmbeddings", "PreTrainedModel", "_init_weights", "arange", "class", "cls_token", "config", "copy_", "create_visual_bbox", "def", "elif", "expand", "hasattr", "if", "image_size", "init", "isinstance", "module", "pos_embed", "position_ids", "self", "shape", "size", "super", "torch", "visual_bbox", "visual_embed", "zeros_"], "layoutlmv3/modeling_layoutlmv3.py:LayoutLMv3SelfAttention.__init__": ["Dropout", "Linear", "ModelSelfAttention", "Module", "The", "ValueError", "__init__", "a", "all_head_size", "and", "attention", "attention_head_size", "attention_probs_dropout_prob", "class", "config", "def", "dropout", "embedding_size", "f", "has_relative_attention_bias", "has_spatial_attention_bias", "hasattr", "heads", "hidden", "hidden_size", "if", "int", "is", "key", "multiple", "nn", "not", "num_attention_heads", "number", "of", "query", "raise", "self", "size", "super", "the", "value"], "layoutlmv3/modeling_layoutlmv3.py:LayoutLMv3SelfAttention.cogview_attention": ["ModelSelfAttention", "Module", "Softmax", "alpha", "amax", "attention_scores", "class", "cogview_attention", "def", "dim", "max_value", "new_attention_scores", "nn", "return", "scaled_attention_scores", "self", "unsqueeze"], "layoutlmv3/modeling_layoutlmv3.py:LayoutLMv3SelfAttention.forward": ["False", "ModelSelfAttention", "Module", "None", "_", "all_head_size", "and", "attention_head_size", "attention_mask", "attention_probs", "attention_scores", "batch_size", "class", "cogview_attention", "context_layer", "contiguous", "def", "dropout", "elif", "else", "forward", "has_relative_attention_bias", "has_spatial_attention_bias", "hidden_states", "if", "is", "key", "key_layer", "math", "matmul", "new_context_layer_shape", "nn", "not", "num_attention_heads", "output_attentions", "outputs", "permute", "query", "query_layer", "rel_2d_pos", "rel_pos", "return", "self", "seq_length", "shape", "size", "sqrt", "torch", "transpose", "value", "value_layer", "view"], "layoutlmv3/modeling_layoutlmv3.py:LayoutLMv3SelfOutput.__init__": ["Dropout", "LayerNorm", "Linear", "ModelSelfOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "eps", "hidden_dropout_prob", "hidden_size", "layer_norm_eps", "nn", "self", "super"], "layoutlmv3/modeling_layoutlmv3.py:LayoutLMv3SelfOutput.forward": ["LayerNorm", "ModelSelfOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "layoutlmv3/modeling_layoutlmv3.py:LayoutLMv3Attention.__init__": ["ModelAttention", "ModelSelfAttention", "ModelSelfOutput", "Module", "__init__", "class", "config", "def", "nn", "output", "self", "super"], "layoutlmv3/modeling_layoutlmv3.py:LayoutLMv3Attention.forward": ["False", "ModelAttention", "Module", "None", "attention_mask", "attention_output", "class", "def", "forward", "hidden_states", "nn", "output", "output_attentions", "outputs", "rel_2d_pos", "rel_pos", "return", "self", "self_outputs"], "layoutlmv3/modeling_layoutlmv3.py:LayoutLMv3Layer.__init__": ["GradientCheckpointingLayer", "ModelAttention", "ModelIntermediate", "ModelLayer", "ModelOutput", "__init__", "attention", "chunk_size_feed_forward", "class", "config", "def", "intermediate", "output", "self", "seq_len_dim", "super"], "layoutlmv3/modeling_layoutlmv3.py:LayoutLMv3Layer.forward": ["False", "GradientCheckpointingLayer", "ModelLayer", "None", "apply_chunking_to_forward", "attention", "attention_mask", "attention_output", "chunk_size_feed_forward", "class", "def", "feed_forward_chunk", "forward", "hidden_states", "layer_output", "output_attentions", "outputs", "rel_2d_pos", "rel_pos", "return", "self", "self_attention_outputs", "seq_len_dim"], "layoutlmv3/modeling_layoutlmv3.py:LayoutLMv3Layer.feed_forward_chunk": ["GradientCheckpointingLayer", "ModelLayer", "attention_output", "class", "def", "feed_forward_chunk", "intermediate", "intermediate_output", "layer_output", "output", "return", "self"], "layoutlmv3/modeling_layoutlmv3.py:LayoutLMv3Encoder.__init__": ["False", "Linear", "ModelEncoder", "ModelLayer", "Module", "ModuleList", "_", "__init__", "bias", "class", "config", "def", "for", "gradient_checkpointing", "has_relative_attention_bias", "has_spatial_attention_bias", "if", "in", "layer", "max_rel_2d_pos", "max_rel_pos", "nn", "num_attention_heads", "num_hidden_layers", "range", "rel_2d_pos_bins", "rel_pos_bias", "rel_pos_bins", "rel_pos_x_bias", "rel_pos_y_bias", "self", "super"], "layoutlmv3/modeling_layoutlmv3.py:LayoutLMv3Encoder.relative_position_bucket": ["ModelEncoder", "Module", "True", "abs", "bidirectional", "class", "def", "else", "float", "full_like", "if", "is_small", "log", "long", "math", "max", "max_distance", "max_exact", "min", "n", "nn", "num_buckets", "relative_position", "relative_position_bucket", "ret", "return", "self", "to", "torch", "val_if_large", "where", "zeros_like"], "layoutlmv3/modeling_layoutlmv3.py:LayoutLMv3Encoder._cal_1d_pos_emb": ["ModelEncoder", "Module", "_cal_1d_pos_emb", "class", "contiguous", "def", "max_distance", "max_rel_pos", "nn", "no_grad", "num_buckets", "permute", "position_ids", "rel_pos", "rel_pos_bias", "rel_pos_bins", "rel_pos_mat", "relative_position_bucket", "return", "self", "t", "torch", "unsqueeze", "weight", "with"], "layoutlmv3/modeling_layoutlmv3.py:LayoutLMv3Encoder._cal_2d_pos_emb": ["ModelEncoder", "Module", "_cal_2d_pos_emb", "bbox", "class", "contiguous", "def", "max_distance", "max_rel_2d_pos", "nn", "no_grad", "num_buckets", "permute", "position_coord_x", "position_coord_y", "rel_2d_pos", "rel_2d_pos_bins", "rel_pos_x", "rel_pos_x_2d_mat", "rel_pos_x_bias", "rel_pos_y", "rel_pos_y_2d_mat", "rel_pos_y_bias", "relative_position_bucket", "return", "self", "t", "torch", "unsqueeze", "weight", "with"], "layoutlmv3/modeling_layoutlmv3.py:LayoutLMv3Encoder.forward": ["BaseModelOutput", "False", "ModelEncoder", "Module", "None", "True", "_cal_1d_pos_emb", "_cal_2d_pos_emb", "all_hidden_states", "all_self_attentions", "attention_mask", "attentions", "bbox", "class", "def", "else", "enumerate", "for", "forward", "has_relative_attention_bias", "has_spatial_attention_bias", "hidden_states", "i", "if", "in", "is", "last_hidden_state", "layer", "layer_module", "layer_outputs", "nn", "not", "output_attentions", "output_hidden_states", "patch_height", "patch_width", "position_ids", "rel_2d_pos", "rel_pos", "return", "return_dict", "self", "tuple", "v"], "layoutlmv3/modeling_layoutlmv3.py:LayoutLMv3Intermediate.__init__": ["ACT2FN", "Linear", "ModelIntermediate", "Module", "__init__", "class", "config", "def", "dense", "else", "hidden_act", "hidden_size", "if", "intermediate_act_fn", "intermediate_size", "isinstance", "nn", "self", "str", "super"], "layoutlmv3/modeling_layoutlmv3.py:LayoutLMv3Intermediate.forward": ["ModelIntermediate", "Module", "Tensor", "class", "def", "dense", "forward", "hidden_states", "intermediate_act_fn", "nn", "return", "self", "torch"], "layoutlmv3/modeling_layoutlmv3.py:LayoutLMv3Output.__init__": ["Dropout", "LayerNorm", "Linear", "ModelOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "eps", "hidden_dropout_prob", "hidden_size", "intermediate_size", "layer_norm_eps", "nn", "self", "super"], "layoutlmv3/modeling_layoutlmv3.py:LayoutLMv3Output.forward": ["LayerNorm", "ModelOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "layoutlmv3/modeling_layoutlmv3.py:LayoutLMv3Model.__init__": ["Dropout", "False", "LayerNorm", "ModelEncoder", "ModelModel", "ModelPatchEmbeddings", "ModelPreTrainedModel", "ModelTextEmbeddings", "Parameter", "__init__", "class", "cls_token", "config", "create_visual_bbox", "def", "dropout", "embeddings", "encoder", "eps", "has_relative_attention_bias", "has_spatial_attention_bias", "hidden_dropout_prob", "hidden_size", "if", "image_size", "input_size", "int", "layer_norm_eps", "nn", "norm", "or", "p", "patch_embed", "patch_size", "persistent", "pos_drop", "pos_embed", "post_init", "register_buffer", "self", "size", "super", "text_embed", "torch", "visual_bbox", "visual_embed", "zeros"], "layoutlmv3/modeling_layoutlmv3.py:LayoutLMv3Model.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "return", "self", "word_embeddings"], "layoutlmv3/modeling_layoutlmv3.py:LayoutLMv3Model.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "self", "set_input_embeddings", "value", "word_embeddings"], "layoutlmv3/modeling_layoutlmv3.py:LayoutLMv3Model.create_visual_bbox": ["ModelModel", "ModelPreTrainedModel", "arange", "cat", "class", "cls_token_box", "create_visual_bbox", "def", "dim", "div", "image_size", "max_len", "repeat", "return", "rounding_mode", "self", "stack", "tensor", "torch", "transpose", "trunc", "view", "visual_bbox", "visual_bbox_x", "visual_bbox_y"], "layoutlmv3/modeling_layoutlmv3.py:LayoutLMv3Model.calculate_visual_bbox": ["ModelModel", "ModelPreTrainedModel", "batch_size", "calculate_visual_bbox", "class", "def", "device", "dtype", "repeat", "return", "self", "to", "type", "visual_bbox"], "layoutlmv3/modeling_layoutlmv3.py:LayoutLMv3Model.forward_image": ["ModelModel", "ModelPreTrainedModel", "None", "_", "batch_size", "cat", "class", "cls_token", "cls_tokens", "def", "dim", "embeddings", "expand", "forward_image", "if", "is", "norm", "not", "patch_embed", "pixel_values", "pos_drop", "pos_embed", "return", "self", "seq_len", "size", "torch"], "layoutlmv3/modeling_layoutlmv3.py:LayoutLMv3Model.forward": ["BaseModelOutput", "FloatTensor", "LayerNorm", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "ValueError", "You", "arange", "attention_mask", "attentions", "batch_size", "bbox", "bool", "calculate_visual_bbox", "cat", "class", "config", "def", "device", "dim", "dropout", "dtype", "either", "elif", "else", "embedding_output", "embeddings", "encoder", "encoder_outputs", "expand", "expand_as", "extended_attention_mask", "final_bbox", "final_position_ids", "forward", "forward_image", "get_extended_attention_mask", "has_relative_attention_bias", "has_spatial_attention_bias", "have", "hidden_states", "if", "input_ids", "input_shape", "inputs_embeds", "is", "kwargs", "last_hidden_state", "len", "list", "long", "not", "ones", "or", "output_attentions", "output_hidden_states", "patch_height", "patch_size", "patch_width", "pixel_values", "position_ids", "r", "raise", "repeat", "return", "return_dict", "self", "seq_length", "sequence_output", "shape", "size", "specify", "to", "token_type_ids", "torch", "torch_int", "tuple", "unsqueeze", "use_return_dict", "visual_attention_mask", "visual_bbox", "visual_embeddings", "visual_position_ids", "zeros"], "layoutlmv3/modeling_layoutlmv3.py:LayoutLMv3ClassificationHead.__init__": ["Dropout", "False", "Linear", "ModelClassificationHead", "Module", "None", "__init__", "class", "classifier_dropout", "config", "def", "dense", "dropout", "else", "hidden_dropout_prob", "hidden_size", "if", "is", "nn", "not", "num_labels", "out_proj", "pool_feature", "self", "super"], "layoutlmv3/modeling_layoutlmv3.py:LayoutLMv3ClassificationHead.forward": ["ModelClassificationHead", "Module", "class", "def", "dense", "dropout", "forward", "nn", "out_proj", "return", "self", "tanh", "torch", "x"], "layoutlmv3/modeling_layoutlmv3.py:LayoutLMv3ForTokenClassification.__init__": ["Dropout", "False", "Linear", "Model", "ModelClassificationHead", "ModelForTokenClassification", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "classifier", "config", "def", "dropout", "else", "hidden_dropout_prob", "hidden_size", "if", "nn", "num_labels", "pool_feature", "post_init", "self", "super"], "layoutlmv3/modeling_layoutlmv3.py:LayoutLMv3ForTokenClassification.get_input_embeddings": ["Model", "ModelForTokenClassification", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "return", "self"], "layoutlmv3/modeling_layoutlmv3.py:LayoutLMv3ForTokenClassification.set_input_embeddings": ["Model", "ModelForTokenClassification", "ModelPreTrainedModel", "class", "def", "self", "set_input_embeddings", "value"], "layoutlmv3/modeling_layoutlmv3.py:LayoutLMv3ForTokenClassification.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "Model", "ModelForTokenClassification", "ModelPreTrainedModel", "None", "TokenClassifierOutput", "attention_mask", "attentions", "bbox", "bool", "class", "classifier", "config", "def", "dropout", "else", "forward", "hidden_states", "if", "input_ids", "input_shape", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "not", "num_labels", "output", "output_attentions", "output_hidden_states", "outputs", "pixel_values", "position_ids", "r", "return", "return_dict", "self", "seq_length", "sequence_output", "size", "token_type_ids", "torch", "tuple", "use_return_dict", "view"], "layoutlmv3/modeling_layoutlmv3.py:LayoutLMv3ForQuestionAnswering.__init__": ["False", "Model", "ModelClassificationHead", "ModelForQuestionAnswering", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "config", "def", "num_labels", "pool_feature", "post_init", "qa_outputs", "self", "super"], "layoutlmv3/modeling_layoutlmv3.py:LayoutLMv3ForQuestionAnswering.get_input_embeddings": ["Model", "ModelForQuestionAnswering", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "return", "self"], "layoutlmv3/modeling_layoutlmv3.py:LayoutLMv3ForQuestionAnswering.set_input_embeddings": ["Model", "ModelForQuestionAnswering", "ModelPreTrainedModel", "class", "def", "self", "set_input_embeddings", "value"], "layoutlmv3/modeling_layoutlmv3.py:LayoutLMv3ForQuestionAnswering.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "Model", "ModelForQuestionAnswering", "ModelPreTrainedModel", "None", "QuestionAnsweringModelOutput", "and", "attention_mask", "attentions", "bbox", "bool", "clamp", "class", "config", "contiguous", "def", "dim", "else", "end_logits", "end_loss", "end_positions", "forward", "hidden_states", "if", "ignore_index", "ignored_index", "input_ids", "inputs_embeds", "is", "kwargs", "len", "logits", "loss", "loss_fct", "not", "output", "output_attentions", "output_hidden_states", "outputs", "pixel_values", "position_ids", "qa_outputs", "r", "return", "return_dict", "self", "sequence_output", "size", "split", "squeeze", "start_logits", "start_loss", "start_positions", "token_type_ids", "torch", "total_loss", "tuple", "use_return_dict"], "layoutlmv3/modeling_layoutlmv3.py:LayoutLMv3ForSequenceClassification.__init__": ["False", "Model", "ModelClassificationHead", "ModelForSequenceClassification", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "classifier", "config", "def", "num_labels", "pool_feature", "post_init", "self", "super"], "layoutlmv3/modeling_layoutlmv3.py:LayoutLMv3ForSequenceClassification.get_input_embeddings": ["Model", "ModelForSequenceClassification", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "return", "self"], "layoutlmv3/modeling_layoutlmv3.py:LayoutLMv3ForSequenceClassification.set_input_embeddings": ["Model", "ModelForSequenceClassification", "ModelPreTrainedModel", "class", "def", "self", "set_input_embeddings", "value"], "layoutlmv3/modeling_layoutlmv3.py:LayoutLMv3ForSequenceClassification.forward": ["BCEWithLogitsLoss", "CrossEntropyLoss", "FloatTensor", "LongTensor", "MSELoss", "Model", "ModelForSequenceClassification", "ModelPreTrainedModel", "None", "SequenceClassifierOutput", "and", "attention_mask", "attentions", "bbox", "bool", "class", "classifier", "config", "def", "dtype", "elif", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "kwargs", "labels", "logits", "long", "loss", "loss_fct", "multi_label_classification", "not", "num_labels", "or", "output", "output_attentions", "output_hidden_states", "outputs", "pixel_values", "position_ids", "problem_type", "r", "regression", "return", "return_dict", "self", "sequence_output", "single_label_classification", "squeeze", "token_type_ids", "torch", "tuple", "use_return_dict", "view"], "led/modeling_led.py:shift_tokens_right": ["ModeModel_input_ids", "Model_tokens_right", "None", "Tensor", "ValueError", "be", "clone", "config", "decoder_start_token_id", "def", "defined", "has", "if", "input_ids", "int", "is", "masked_fill_", "new_zeros", "pad_token_id", "raise", "return", "shape", "to", "torch"], "led/modeling_led.py:_prepare_4d_attention_mask_inverted": ["None", "Tensor", "_prepare_4d_attention_mask_inverted", "bool", "bsz", "def", "dtype", "else", "expand", "expanded_attention_mask", "expanded_mask", "finfo", "if", "int", "inverted_mask", "is", "mask", "masked_fill", "min", "not", "return", "size", "src_len", "tgt_len", "to", "torch"], "led/modeling_led.py:LEDLearnedPositionalEmbedding.__init__": ["Embedding", "Model", "__init__", "class", "def", "embedding_dim", "int", "nn", "num_embeddings", "self", "super"], "led/modeling_led.py:LEDLearnedPositionalEmbedding.forward": ["Embedding", "Model", "Size", "arange", "bsz", "class", "def", "device", "dtype", "forward", "input_ids_shape", "int", "long", "nn", "past_key_values_length", "positions", "return", "self", "seq_len", "super", "torch", "weight"], "led/modeling_led.py:LEDEncoderSelfAttention.__init__": ["Given", "Linear", "Model", "Module", "The", "ValueError", "__init__", "a", "an", "assert", "attention", "attention_probs_dropout_prob", "attention_window", "be", "class", "config", "def", "dropout", "embed_dim", "even", "f", "for", "has", "head_dim", "heads", "hidden", "hidden_size", "if", "int", "is", "key", "key_global", "layer", "layer_id", "multiple", "nn", "not", "num_attention_heads", "num_heads", "number", "of", "one_sided_attn_window_size", "positive", "query", "query_global", "raise", "self", "size", "super", "the", "to", "value", "value_global"], "led/modeling_led.py:LEDEncoderSelfAttention.forward": ["False", "Model", "Module", "None", "Unexpected", "_compute_attn_output_with_global_indices", "_compute_global_attn_output_from_hidden", "_concat_with_global_key_attn_probs", "_get_global_attn_indices", "_sliding_chunks_matmul_attn_probs_value", "_sliding_chunks_query_key_matmul", "and", "assert", "attention_mask", "attn_output", "attn_probs", "attn_scores", "batch_size", "be", "but", "cat", "class", "contiguous", "def", "del", "diagonal_mask", "dim", "dropout", "dtype", "else", "embed_dim", "f", "finfo", "float32", "float_mask", "forward", "functional", "global_attn_output", "global_attn_probs", "global_key_attn_scores", "has", "have", "head_dim", "hidden_states", "if", "is", "is_global_attn", "is_index_global_attn", "is_index_global_attn_nonzero", "is_index_masked", "is_local_index_global_attn_nonzero", "is_local_index_no_global_attn_nonzero", "key", "key_vectors", "len", "list", "local_attn_probs", "masked_fill", "math", "max_num_global_attn_indices", "min", "new_ones", "nn", "nonzero_global_attn_output", "num_heads", "of", "one_sided_attn_window_size", "output_attentions", "outputs", "p", "query", "query_vectors", "remove_from_windowed_attention_mask", "reshape", "return", "self", "seq_len", "should", "size", "softmax", "sqrt", "torch", "training", "transpose", "type_as", "value", "value_vectors", "view"], "led/modeling_led.py:LEDEncoderSelfAttention._pad_and_transpose_last_two_dims": ["Model", "Module", "_pad_and_transpose_last_two_dims", "class", "def", "functional", "hidden_states_padded", "nn", "pad", "padding", "return", "size", "view"], "led/modeling_led.py:LEDEncoderSelfAttention._pad_and_diagonalize": ["Model", "Module", "_pad_and_diagonalize", "chunked_hidden_states", "class", "def", "functional", "hidden_dim", "nn", "num_chunks", "pad", "return", "size", "total_num_heads", "view", "window_overlap"], "led/modeling_led.py:LEDEncoderSelfAttention._chunk": ["False", "Model", "Module", "_chunk", "as_strided", "bool", "chunk", "chunk_size", "chunk_stride", "class", "def", "device", "div", "empty", "for", "hidden_states", "if", "in", "list", "nn", "not", "onnx_export", "overlapping_chunks", "range", "return", "rounding_mode", "size", "stride", "torch", "trunc", "view", "window_overlap"], "led/modeling_led.py:LEDEncoderSelfAttention._mask_invalid_locations": ["Model", "Module", "None", "Tensor", "_mask_invalid_locations", "affected_seq_len", "beginning_input", "beginning_mask", "beginning_mask_2d", "bool", "class", "def", "dims", "ending_input", "ending_mask", "expand", "flip", "float", "full_like", "inf", "input_tensor", "new_ones", "nn", "size", "torch", "tril", "where"], "led/modeling_led.py:LEDEncoderSelfAttention._sliding_chunks_query_key_matmul": ["False", "Given", "Model", "Module", "Sequence", "Tensor", "_chunk", "_mask_invalid_locations", "_pad_and_transpose_last_two_dims", "_sliding_chunks_query_key_matmul", "assert", "batch_size", "bcxd", "bcxy", "bcyd", "be", "chunks_count", "class", "config", "def", "diagonal_attention_scores", "diagonal_chunked_attention_scores", "div", "einsum", "f", "getattr", "head_dim", "int", "key", "length", "multiple", "new_zeros", "nn", "num_heads", "of", "onnx_export", "padding", "query", "reshape", "return", "rounding_mode", "self", "seq_len", "should", "size", "torch", "transpose", "trunc", "view", "window_overlap"], "led/modeling_led.py:LEDEncoderSelfAttention._sliding_chunks_matmul_attn_probs_value": ["Model", "Module", "Tensor", "_pad_and_diagonalize", "_sliding_chunks_matmul_attn_probs_value", "as_strided", "assert", "attn_probs", "batch_size", "bcdh", "bcwd", "bcwh", "chunked_attn_probs", "chunked_value", "chunked_value_size", "chunked_value_stride", "chunks_count", "class", "context", "def", "div", "einsum", "functional", "head_dim", "int", "nn", "num_heads", "pad", "padded_value", "reshape", "return", "rounding_mode", "self", "seq_len", "size", "stride", "torch", "transpose", "trunc", "value", "view", "window_overlap"], "led/modeling_led.py:LEDEncoderSelfAttention._get_global_attn_indices": ["Model", "Module", "True", "_get_global_attn_indices", "arange", "as_tuple", "class", "def", "device", "dim", "is_index_global_attn", "is_index_global_attn_nonzero", "is_local_index_global_attn", "is_local_index_global_attn_nonzero", "is_local_index_no_global_attn_nonzero", "long", "max", "max_num_global_attn_indices", "nn", "nonzero", "num_global_attn_indices", "return", "sum", "torch", "unsqueeze"], "led/modeling_led.py:LEDEncoderSelfAttention._concat_with_global_key_attn_probs": ["Model", "Module", "_concat_with_global_key_attn_probs", "attn_probs_from_global_key", "batch_size", "blhd", "blhs", "bshd", "class", "def", "dtype", "einsum", "finfo", "head_dim", "is_index_global_attn_nonzero", "is_local_index_global_attn_nonzero", "is_local_index_no_global_attn_nonzero", "key_vectors", "key_vectors_only_global", "max_num_global_attn_indices", "min", "new_zeros", "nn", "num_heads", "query_vectors", "return", "self", "shape", "torch", "transpose"], "led/modeling_led.py:LEDEncoderSelfAttention._compute_attn_output_with_global_indices": ["Model", "Module", "_compute_attn_output_with_global_indices", "_sliding_chunks_matmul_attn_probs_value", "attn_output_only_global", "attn_output_without_global", "attn_probs", "attn_probs_only_global", "attn_probs_without_global", "batch_size", "class", "clone", "contiguous", "def", "head_dim", "is_index_global_attn_nonzero", "is_local_index_global_attn_nonzero", "matmul", "max_num_global_attn_indices", "narrow", "new_zeros", "nn", "num_heads", "one_sided_attn_window_size", "return", "self", "shape", "size", "torch", "transpose", "value_vectors", "value_vectors_only_global"], "led/modeling_led.py:LEDEncoderSelfAttention._compute_global_attn_output_from_hidden": ["Model", "Module", "None", "Size", "_compute_global_attn_output_from_hidden", "assert", "batch_size", "be", "bmm", "but", "class", "contiguous", "def", "dim", "dropout", "dtype", "embed_dim", "f", "finfo", "float32", "functional", "global_attn_hidden_states", "global_attn_output", "global_attn_probs", "global_attn_probs_float", "global_attn_scores", "global_key_vectors", "global_query_vectors_only_global", "global_value_vectors", "has", "have", "head_dim", "hidden_states", "is", "is_index_global_attn_nonzero", "is_index_masked", "is_local_index_global_attn_nonzero", "is_local_index_no_global_attn_nonzero", "key_global", "list", "masked_fill", "math", "max_num_global_attn_indices", "min", "new_zeros", "nn", "num_heads", "p", "query_global", "return", "self", "seq_len", "shape", "should", "size", "softmax", "sqrt", "tensor", "the", "torch", "training", "transpose", "type_as", "value_global", "view", "wrong"], "led/modeling_led.py:LEDEncoderAttention.__init__": ["Linear", "Model", "ModelEncoderSelfAttention", "Module", "__init__", "class", "config", "d_model", "def", "layer_id", "longformer_self_attn", "nn", "output", "self", "super"], "led/modeling_led.py:LEDEncoderAttention.forward": ["False", "Model", "Module", "None", "Tensor", "attention_mask", "attn_output", "bool", "class", "def", "forward", "hidden_states", "is_global_attn", "is_index_global_attn", "is_index_masked", "longformer_self_attn", "nn", "output", "output_attentions", "outputs", "return", "self", "self_outputs", "torch", "tuple"], "led/modeling_led.py:LEDDecoderAttention.__init__": ["False", "Linear", "Model", "Module", "None", "True", "ValueError", "__init__", "and", "be", "bias", "bool", "by", "class", "def", "divisible", "dropout", "embed_dim", "f", "float", "got", "head_dim", "if", "int", "is_decoder", "k_proj", "layer_idx", "must", "nn", "num_heads", "out_proj", "q_proj", "raise", "scaling", "self", "super", "v_proj"], "led/modeling_led.py:LEDDecoderAttention.forward": ["Attention", "Cache", "EncoderDecoderCache", "False", "Model", "Module", "None", "Tensor", "True", "ValueError", "and", "attention_mask", "attn_output", "attn_probs", "attn_weights", "attn_weights_reshaped", "be", "bmm", "bool", "bsz", "but", "cache_position", "class", "cross_attention_cache", "curr_past_key_values", "current_states", "def", "dim", "dropout", "else", "embed_dim", "f", "forward", "functional", "get", "head_dim", "hidden_states", "if", "is", "is_cross_attention", "is_updated", "isinstance", "k_proj", "key_states", "key_value_states", "keys", "layer_idx", "layers", "mask", "nn", "not", "num_heads", "of", "out_proj", "output_attentions", "p", "past_key_values", "proj_shape", "q_proj", "query_states", "raise", "reshape", "return", "scaling", "self", "self_attention_cache", "should", "size", "softmax", "src_len", "tgt_len", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "values", "view", "weights"], "led/modeling_led.py:LEDEncoderLayer.__init__": ["ACT2FN", "GradientCheckpointingLayer", "LayerNorm", "Linear", "Model", "ModelConfig", "ModelEncoderAttention", "__init__", "activation_dropout", "activation_fn", "activation_function", "class", "config", "d_model", "def", "dropout", "embed_dim", "encoder_ffn_dim", "fc1", "fc2", "final_layer_norm", "int", "layer_id", "nn", "self", "self_attn", "self_attn_layer_norm", "super"], "led/modeling_led.py:LEDEncoderLayer.forward": ["False", "GradientCheckpointingLayer", "Model", "None", "Tensor", "activation_dropout", "activation_fn", "and", "any", "attention_mask", "attn_outputs", "clamp", "clamp_value", "class", "def", "dropout", "dtype", "fc1", "fc2", "final_layer_norm", "finfo", "float16", "forward", "functional", "hidden_states", "if", "is_global_attn", "is_index_global_attn", "is_index_masked", "isinf", "isnan", "max", "min", "nn", "or", "output_attentions", "p", "residual", "return", "self", "self_attn", "self_attn_layer_norm", "torch", "training"], "led/modeling_led.py:LEDDecoderLayer.__init__": ["ACT2FN", "GradientCheckpointingLayer", "LayerNorm", "Linear", "Model", "ModelConfig", "ModelDecoderAttention", "None", "True", "__init__", "activation_dropout", "activation_fn", "activation_function", "attention_dropout", "class", "config", "d_model", "decoder_attention_heads", "decoder_ffn_dim", "def", "dropout", "embed_dim", "encoder_attn", "encoder_attn_layer_norm", "fc1", "fc2", "final_layer_norm", "is_decoder", "layer_idx", "nn", "num_heads", "self", "self_attn", "self_attn_layer_norm", "super"], "led/modeling_led.py:LEDDecoderLayer.forward": ["Cache", "False", "GradientCheckpointingLayer", "Model", "None", "Tensor", "True", "activation_dropout", "activation_fn", "attention_mask", "bool", "cache_position", "class", "cross_attn_present_key_value", "cross_attn_weights", "def", "dropout", "encoder_attention_mask", "encoder_attn", "encoder_attn_layer_norm", "encoder_hidden_states", "fc1", "fc2", "final_layer_norm", "forward", "functional", "hidden_states", "if", "is", "key_value_states", "nn", "not", "output_attentions", "outputs", "p", "past_key_values", "present_key_value", "residual", "return", "self", "self_attn", "self_attn_layer_norm", "self_attn_weights", "torch", "training", "use_cache"], "led/modeling_led.py:LEDClassificationHead.__init__": ["Dropout", "Linear", "Model", "Module", "__init__", "class", "def", "dense", "dropout", "float", "inner_dim", "input_dim", "int", "nn", "num_classes", "out_proj", "p", "pooler_dropout", "self", "super"], "led/modeling_led.py:LEDClassificationHead.forward": ["Model", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "nn", "out_proj", "return", "self", "tanh", "torch"], "led/modeling_led.py:LEDPreTrainedModel.dummy_inputs": ["Model", "PreTrainedModel", "attention_mask", "class", "config", "def", "device", "dummy_inputs", "input_ids", "ne", "pad_token", "pad_token_id", "return", "self", "tensor", "torch"], "led/modeling_led.py:LEDPreTrainedModel._init_weights": ["Model", "ModelForConditionalGeneration", "PreTrainedModel", "_init_weights", "class", "def", "final_logits_bias", "if", "init", "isinstance", "module", "self", "super", "zeros_"], "led/modeling_led.py:LEDEncoder.__init__": ["Embedding", "Expected", "False", "LayerNorm", "Model", "ModelConfig", "ModelLayer", "ModelLearnedPositionalEmbedding", "ModelPreTrainedModel", "ModuleList", "ValueError", "__init__", "an", "attention_window", "be", "class", "config", "d_model", "def", "dropout", "else", "embed_dim", "embed_positions", "embed_tokens", "encoder_layerdrop", "encoder_layers", "equal", "even", "f", "for", "given", "gradient_checkpointing", "has", "i", "if", "in", "int", "isinstance", "layerdrop", "layernorm_embedding", "layers", "len", "max_encoder_position_embeddings", "max_source_positions", "nn", "num_hidden_layers", "pad_token_id", "padding_idx", "positive", "post_init", "raise", "range", "self", "should", "super", "to", "value", "vocab_size"], "led/modeling_led.py:LEDEncoder._merge_to_attention_mask": ["Model", "ModelPreTrainedModel", "None", "Tensor", "_merge_to_attention_mask", "attention_mask", "class", "def", "else", "global_attention_mask", "if", "is", "not", "return", "self", "torch"], "led/modeling_led.py:LEDEncoder._pad_to_window_size": ["False", "Given", "Input", "Model", "ModelPreTrainedModel", "None", "Tensor", "ValueError", "_pad_to_window_size", "a", "an", "are", "attention_mask", "attention_window", "automatically", "batch_size", "be", "cat", "class", "config", "def", "dim", "dtype", "else", "embed_tokens", "even", "f", "from", "functional", "ids", "if", "input_ids", "input_ids_padding", "input_shape", "inputs_embeds", "inputs_embeds_padding", "int", "is", "isinstance", "logger", "long", "max", "multiple", "new_full", "nn", "not", "of", "pad", "pad_token_id", "padded", "padding_len", "raise", "return", "self", "seq_len", "shape", "should", "to", "torch", "value", "warning_once"], "led/modeling_led.py:LEDEncoder.forward": ["Model", "ModelBaseModelOutput", "ModelPreTrainedModel", "None", "ValueError", "You", "_merge_to_attention_mask", "_pad_to_window_size", "_prepare_4d_attention_mask_inverted", "all_attentions", "all_global_attentions", "and", "any", "at", "attention_mask", "attentions", "both", "cannot", "class", "config", "def", "device", "dropout", "dropout_probability", "dtype", "either", "elif", "else", "embed_pos", "embed_positions", "embed_tokens", "encoder_layer", "encoder_states", "enumerate", "flatten", "for", "forward", "functional", "global_attention_mask", "global_attentions", "have", "hidden_states", "idx", "if", "in", "input_ids", "input_shape", "inputs_embeds", "is", "is_global_attn", "is_index_global_attn", "is_index_masked", "item", "kwargs", "last_hidden_state", "layer_outputs", "layerdrop", "layernorm_embedding", "layers", "long", "nn", "not", "ones", "or", "output_attentions", "output_hidden_states", "p", "pad_token_id", "padding_len", "r", "raise", "rand", "return", "return_dict", "same", "self", "size", "specify", "state", "the", "time", "to", "torch", "training", "transpose", "tuple", "use_return_dict", "v", "view"], "led/modeling_led.py:LEDDecoder.__init__": ["Embedding", "False", "LayerNorm", "Model", "ModelConfig", "ModelLayer", "ModelLearnedPositionalEmbedding", "ModelPreTrainedModel", "ModuleList", "__init__", "class", "config", "d_model", "decoder_layerdrop", "decoder_layers", "def", "dropout", "embed_positions", "embed_tokens", "for", "gradient_checkpointing", "i", "in", "layer_idx", "layerdrop", "layernorm_embedding", "layers", "max_decoder_position_embeddings", "max_target_positions", "nn", "pad_token_id", "padding_idx", "post_init", "range", "self", "super", "vocab_size"], "led/modeling_led.py:LEDDecoder.forward": ["BaseModelOutputWithPastAndCrossAttentions", "DynamicCache", "EncoderDecoderCache", "False", "Model", "ModelPreTrainedModel", "None", "Setting", "True", "ValueError", "You", "_create_4d_causal_attention_mask", "_prepare_4d_attention_mask_inverted", "all_cross_attentions", "all_hidden_states", "all_self_attns", "and", "at", "attention_mask", "attentions", "both", "cache_position", "cannot", "checkpointing", "class", "combined_attention_mask", "config", "continue", "cross_attentions", "decoder_input_ids", "decoder_inputs_embeds", "decoder_layer", "def", "device", "dropout", "dropout_probability", "dtype", "either", "elif", "else", "embed_positions", "embed_tokens", "encoder_attention_mask", "encoder_hidden_states", "enumerate", "for", "forward", "functional", "get_seq_length", "global_attention_mask", "gradient", "gradient_checkpointing", "have", "hidden_states", "idx", "if", "in", "incompatible", "input_ids", "input_shape", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layer_outputs", "layerdrop", "layernorm_embedding", "layers", "logger", "nn", "not", "or", "output_attentions", "output_hidden_states", "p", "past_key_values", "past_key_values_length", "positions", "r", "raise", "rand", "return", "return_dict", "same", "self", "size", "specify", "tgt_len", "the", "time", "to", "torch", "training", "tuple", "use_cache", "use_return_dict", "v", "view", "warning_once", "with"], "led/modeling_led.py:LEDModel.__init__": ["Embedding", "Model", "ModelConfig", "ModelDecoder", "ModelEncoder", "ModelPreTrainedModel", "__init__", "class", "config", "d_model", "decoder", "def", "encoder", "nn", "pad_token_id", "padding_idx", "post_init", "self", "shared", "super", "vocab_size"], "led/modeling_led.py:LEDModel.get_input_embeddings": ["Model", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "return", "self", "shared"], "led/modeling_led.py:LEDModel.set_input_embeddings": ["Model", "ModelPreTrainedModel", "class", "decoder", "def", "embed_tokens", "encoder", "self", "set_input_embeddings", "shared", "value"], "led/modeling_led.py:LEDModel.forward": ["Cache", "FloatTensor", "LongTensor", "Model", "ModelEncoderBaseModelOutput", "ModelPreTrainedModel", "ModelSeq2SeqModelOutput", "None", "Tensor", "and", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "cross_attentions", "decoder", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_input_ids", "decoder_inputs_embeds", "decoder_outputs", "decoder_start_token_id", "def", "elif", "else", "encoder", "encoder_attention_mask", "encoder_attentions", "encoder_global_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "forward", "global_attention_mask", "global_attentions", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "isinstance", "kwargs", "last_hidden_state", "len", "not", "output_attentions", "output_hidden_states", "pad_token_id", "past_key_values", "r", "return", "return_dict", "self", "shift_tokens_right", "torch", "tuple", "use_cache", "use_return_dict"], "led/modeling_led.py:LEDForConditionalGeneration.__init__": ["False", "GenerationMixin", "Linear", "Model", "ModelConfig", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "d_model", "def", "final_logits_bias", "lm_head", "nn", "num_embeddings", "post_init", "register_buffer", "self", "shared", "super", "torch", "zeros"], "led/modeling_led.py:LEDForConditionalGeneration.resize_token_embeddings": ["Embedding", "GenerationMixin", "Model", "ModelPreTrainedModel", "None", "True", "_resize_final_logits_bias", "bool", "class", "def", "int", "mean_resizing", "new_embeddings", "new_num_tokens", "nn", "pad_to_multiple_of", "resize_token_embeddings", "return", "self", "shape", "super", "weight"], "led/modeling_led.py:LEDForConditionalGeneration._resize_final_logits_bias": ["GenerationMixin", "Model", "ModelPreTrainedModel", "None", "_resize_final_logits_bias", "cat", "class", "def", "device", "dim", "else", "extra_bias", "final_logits_bias", "if", "int", "new_bias", "new_num_tokens", "old_num_tokens", "register_buffer", "self", "shape", "torch", "zeros"], "led/modeling_led.py:LEDForConditionalGeneration.forward": ["Cache", "CrossEntropyLoss", "False", "FloatTensor", "GenerationMixin", "LongTensor", "Model", "ModelPreTrainedModel", "ModelSeq2SeqLMOutput", "None", "Tensor", "The", "and", "argument", "attention_mask", "bool", "cache_position", "changed", "class", "config", "cross_attentions", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_input_ids", "decoder_inputs_embeds", "decoder_start_token_id", "def", "else", "encoder_attentions", "encoder_global_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "final_logits_bias", "forward", "global_attention_mask", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "lm_head", "lm_logits", "logger", "logits", "loss", "loss_fct", "masked_lm_loss", "not", "output", "output_attentions", "output_hidden_states", "outputs", "pad_token_id", "past_key_values", "provided", "r", "return", "return_dict", "self", "shift_tokens_right", "since", "to", "torch", "tuple", "use_cache", "use_return_dict", "view", "vocab_size", "warning"], "led/modeling_led.py:LEDForConditionalGeneration.prepare_decoder_input_ids_from_labels": ["GenerationMixin", "Model", "ModelPreTrainedModel", "Tensor", "class", "config", "decoder_start_token_id", "def", "labels", "pad_token_id", "prepare_decoder_input_ids_from_labels", "return", "self", "shift_tokens_right", "torch"], "led/modeling_led.py:LEDForQuestionAnswering.__init__": ["Linear", "Model", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "config", "def", "hidden_size", "nn", "num_labels", "post_init", "qa_outputs", "self", "super"], "led/modeling_led.py:LEDForQuestionAnswering.forward": ["CrossEntropyLoss", "False", "FloatTensor", "LongTensor", "Model", "ModelPreTrainedModel", "ModelSeq2SeqQuestionAnsweringModelOutput", "None", "Tensor", "and", "attention_mask", "bool", "clamp", "class", "config", "contiguous", "cross_attentions", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_input_ids", "decoder_inputs_embeds", "def", "dim", "else", "encoder_attentions", "encoder_global_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "end_logits", "end_loss", "end_positions", "forward", "global_attention_mask", "if", "ignore_index", "ignored_index", "input_ids", "inputs_embeds", "is", "kwargs", "len", "logits", "loss", "loss_fct", "not", "output", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "qa_outputs", "r", "return", "return_dict", "self", "sequence_output", "size", "split", "squeeze", "start_logits", "start_loss", "start_positions", "torch", "total_loss", "tuple", "use_cache", "use_return_dict"], "levit/modeling_levit.py:LevitConvEmbeddings.__init__": ["BatchNorm2d", "Conv2d", "False", "ModelConvEmbeddings", "Module", "__init__", "batch_norm", "bias", "bn_weight_init", "class", "convolution", "def", "dilation", "groups", "in_channels", "kernel_size", "nn", "out_channels", "padding", "self", "stride", "super"], "levit/modeling_levit.py:LevitConvEmbeddings.forward": ["ModelConvEmbeddings", "Module", "batch_norm", "class", "convolution", "def", "embeddings", "forward", "nn", "return", "self"], "levit/modeling_levit.py:LevitPatchEmbeddings.__init__": ["Hardswish", "ModelConvEmbeddings", "ModelPatchEmbeddings", "Module", "__init__", "activation_layer_1", "activation_layer_2", "activation_layer_3", "class", "config", "def", "embedding_layer_1", "embedding_layer_2", "embedding_layer_3", "embedding_layer_4", "hidden_sizes", "kernel_size", "nn", "num_channels", "padding", "self", "stride", "super"], "levit/modeling_levit.py:LevitPatchEmbeddings.forward": ["Make", "ModelPatchEmbeddings", "Module", "ValueError", "activation_layer_1", "activation_layer_2", "activation_layer_3", "channel", "class", "configuration", "def", "dimension", "embedding_layer_1", "embedding_layer_2", "embedding_layer_3", "embedding_layer_4", "embeddings", "flatten", "forward", "if", "in", "match", "nn", "num_channels", "of", "one", "pixel", "pixel_values", "raise", "return", "self", "set", "shape", "sure", "that", "the", "transpose", "values", "with"], "levit/modeling_levit.py:MLPLayerWithBN.__init__": ["BatchNorm1d", "False", "Linear", "Model", "Module", "__init__", "batch_norm", "bias", "bn_weight_init", "class", "def", "in_features", "input_dim", "linear", "nn", "out_features", "output_dim", "self", "super"], "levit/modeling_levit.py:MLPLayerWithBN.forward": ["Model", "Module", "batch_norm", "class", "def", "flatten", "forward", "hidden_state", "linear", "nn", "reshape_as", "return", "self"], "levit/modeling_levit.py:LevitSubsample.__init__": ["ModelSubsample", "Module", "__init__", "class", "def", "nn", "resolution", "self", "stride", "super"], "levit/modeling_levit.py:LevitSubsample.forward": ["ModelSubsample", "Module", "_", "batch_size", "channels", "class", "def", "forward", "hidden_state", "nn", "reshape", "resolution", "return", "self", "shape", "stride", "view"], "levit/modeling_levit.py:LevitAttention.__init__": ["False", "Hardswish", "LongTensor", "MLPLayerWithBN", "ModelAttention", "Module", "Parameter", "__init__", "abs", "activation", "append", "attention_bias_cache", "attention_bias_idxs", "attention_biases", "attention_offsets", "attention_ratio", "bn_weight_init", "class", "def", "for", "hidden_sizes", "if", "in", "indices", "itertools", "key_dim", "len", "len_points", "list", "nn", "not", "num_attention_heads", "offset", "out_dim_keys_values", "out_dim_projection", "p1", "p2", "persistent", "points", "product", "projection", "queries_keys_values", "range", "register_buffer", "resolution", "scale", "self", "super", "torch", "view", "zeros"], "levit/modeling_levit.py:LevitAttention.train": ["ModelAttention", "Module", "True", "and", "attention_bias_cache", "class", "def", "if", "mode", "nn", "self", "super", "train"], "levit/modeling_levit.py:LevitAttention.get_attention_biases": ["ModelAttention", "Module", "attention_bias_cache", "attention_bias_idxs", "attention_biases", "class", "def", "device", "device_key", "else", "get_attention_biases", "if", "in", "nn", "not", "return", "self", "str", "training"], "levit/modeling_levit.py:LevitAttention.forward": ["ModelAttention", "Module", "_", "activation", "attention", "attention_ratio", "batch_size", "class", "def", "device", "dim", "forward", "get_attention_biases", "hidden_state", "key", "key_dim", "nn", "num_attention_heads", "out_dim_projection", "permute", "projection", "queries_keys_values", "query", "reshape", "return", "scale", "self", "seq_length", "shape", "softmax", "split", "transpose", "value", "view"], "levit/modeling_levit.py:LevitAttentionSubsample.__init__": ["False", "Hardswish", "LongTensor", "MLPLayerWithBN", "ModelAttentionSubsample", "ModelSubsample", "Module", "Parameter", "__init__", "abs", "activation", "append", "attention_bias_cache", "attention_bias_idxs", "attention_biases", "attention_offsets", "attention_ratio", "class", "def", "for", "if", "in", "indices", "input_dim", "itertools", "key_dim", "keys_values", "len", "len_points", "len_points_", "list", "nn", "not", "num_attention_heads", "offset", "out_dim_keys_values", "out_dim_projection", "output_dim", "p1", "p2", "persistent", "points", "points_", "product", "projection", "queries", "queries_subsample", "range", "register_buffer", "resolution_in", "resolution_out", "scale", "self", "size", "stride", "super", "torch", "view", "zeros"], "levit/modeling_levit.py:LevitAttentionSubsample.train": ["ModelAttentionSubsample", "Module", "True", "and", "attention_bias_cache", "class", "def", "if", "mode", "nn", "self", "super", "train"], "levit/modeling_levit.py:LevitAttentionSubsample.get_attention_biases": ["ModelAttentionSubsample", "Module", "attention_bias_cache", "attention_bias_idxs", "attention_biases", "class", "def", "device", "device_key", "else", "get_attention_biases", "if", "in", "nn", "not", "return", "self", "str", "training"], "levit/modeling_levit.py:LevitAttentionSubsample.forward": ["ModelAttentionSubsample", "Module", "_", "activation", "attention", "attention_ratio", "batch_size", "class", "def", "device", "dim", "forward", "get_attention_biases", "hidden_state", "key", "key_dim", "keys_values", "nn", "num_attention_heads", "out_dim_projection", "permute", "projection", "queries", "queries_subsample", "query", "reshape", "resolution_out", "return", "scale", "self", "seq_length", "shape", "softmax", "split", "transpose", "value", "view"], "levit/modeling_levit.py:LevitMLPLayer.__init__": ["Hardswish", "MLPLayerWithBN", "ModelMLPLayer", "Module", "__init__", "activation", "class", "def", "hidden_dim", "input_dim", "linear_down", "linear_up", "nn", "self", "super"], "levit/modeling_levit.py:LevitMLPLayer.forward": ["ModelMLPLayer", "Module", "activation", "class", "def", "forward", "hidden_state", "linear_down", "linear_up", "nn", "return", "self"], "levit/modeling_levit.py:LevitResidualLayer.__init__": ["ModelResidualLayer", "Module", "__init__", "class", "def", "drop_rate", "module", "nn", "self", "super"], "levit/modeling_levit.py:LevitResidualLayer.forward": ["ModelResidualLayer", "Module", "and", "class", "def", "detach", "device", "div", "drop_rate", "else", "forward", "ge_", "hidden_state", "if", "module", "nn", "rand", "return", "rnd", "self", "size", "torch", "training"], "levit/modeling_levit.py:LevitStage.__init__": ["ModelAttention", "ModelAttentionSubsample", "ModelMLPLayer", "ModelResidualLayer", "ModelStage", "Module", "ModuleList", "Subsample", "_", "__init__", "append", "attention_ratio", "class", "config", "def", "depths", "down_ops", "drop_path_rate", "for", "hidden_dim", "hidden_sizes", "idx", "if", "in", "key_dim", "layers", "mlp_ratio", "nn", "num_attention_heads", "range", "resolution_in", "resolution_out", "self", "stride", "super"], "levit/modeling_levit.py:LevitStage.get_resolution": ["ModelStage", "Module", "class", "def", "get_resolution", "nn", "resolution_in", "return", "self"], "levit/modeling_levit.py:LevitStage.forward": ["ModelStage", "Module", "class", "def", "for", "forward", "hidden_state", "in", "layer", "layers", "nn", "return", "self"], "levit/modeling_levit.py:LevitEncoder.__init__": ["ModelEncoder", "ModelStage", "Module", "ModuleList", "__init__", "append", "attention_ratio", "class", "config", "def", "depths", "down_ops", "for", "get_resolution", "hidden_sizes", "image_size", "in", "key_dim", "len", "mlp_ratio", "nn", "num_attention_heads", "patch_size", "range", "resolution", "self", "stage", "stage_idx", "stages", "super"], "levit/modeling_levit.py:LevitEncoder.forward": ["BaseModelOutputWithNoAttention", "False", "ModelEncoder", "Module", "None", "True", "all_hidden_states", "class", "def", "else", "for", "forward", "hidden_state", "hidden_states", "if", "in", "is", "last_hidden_state", "nn", "not", "output_hidden_states", "return", "return_dict", "self", "stage", "stages", "tuple", "v"], "levit/modeling_levit.py:LevitClassificationLayer.__init__": ["BatchNorm1d", "Linear", "ModelClassificationLayer", "Module", "__init__", "batch_norm", "class", "def", "input_dim", "linear", "nn", "output_dim", "self", "super"], "levit/modeling_levit.py:LevitClassificationLayer.forward": ["ModelClassificationLayer", "Module", "batch_norm", "class", "def", "forward", "hidden_state", "linear", "logits", "nn", "return", "self"], "levit/modeling_levit.py:LevitPreTrainedModel._init_weights": ["LongTensor", "ModelAttention", "ModelAttentionSubsample", "ModelPreTrainedModel", "PreTrainedModel", "_init_weights", "attention_bias_idxs", "class", "copy_", "def", "elif", "if", "indices", "init", "isinstance", "len_points", "len_points_", "module", "self", "super", "torch", "view"], "levit/modeling_levit.py:LevitModel.__init__": ["ModelEncoder", "ModelModel", "ModelPatchEmbeddings", "ModelPreTrainedModel", "__init__", "class", "config", "def", "encoder", "patch_embeddings", "post_init", "self", "super"], "levit/modeling_levit.py:LevitModel.forward": ["BaseModelOutputWithPoolingAndNoAttention", "FloatTensor", "ModelModel", "ModelPreTrainedModel", "None", "ValueError", "You", "bool", "class", "config", "def", "dim", "else", "embeddings", "encoder", "encoder_outputs", "forward", "have", "hidden_states", "if", "is", "kwargs", "last_hidden_state", "mean", "not", "output_hidden_states", "patch_embeddings", "pixel_values", "pooled_output", "pooler_output", "raise", "return", "return_dict", "self", "specify", "to", "torch", "tuple", "use_return_dict"], "levit/modeling_levit.py:LevitForImageClassification.__init__": ["Identity", "Model", "ModelClassificationLayer", "ModelForImageClassification", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "classifier", "config", "def", "else", "hidden_sizes", "if", "nn", "num_labels", "post_init", "self", "super", "torch"], "levit/modeling_levit.py:LevitForImageClassification.forward": ["FloatTensor", "ImageClassifierOutputWithNoAttention", "LongTensor", "Model", "ModelForImageClassification", "ModelPreTrainedModel", "None", "bool", "class", "classifier", "config", "def", "else", "forward", "hidden_states", "if", "is", "kwargs", "labels", "logits", "loss", "loss_function", "mean", "not", "output", "output_hidden_states", "outputs", "pixel_values", "r", "return", "return_dict", "self", "sequence_output", "torch", "tuple", "use_return_dict"], "levit/modeling_levit.py:LevitForImageClassificationWithTeacher.__init__": ["Identity", "Model", "ModelClassificationLayer", "ModelForImageClassificationWithTeacher", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "classifier", "classifier_distill", "config", "def", "else", "hidden_sizes", "if", "nn", "num_labels", "post_init", "self", "super", "torch"], "levit/modeling_levit.py:LevitForImageClassificationWithTeacher.forward": ["FloatTensor", "Model", "ModelForImageClassificationWithTeacher", "ModelForImageClassificationWithTeacherOutput", "ModelPreTrainedModel", "None", "bool", "class", "classifier", "classifier_distill", "cls_logits", "config", "def", "distill_logits", "distillation_logits", "else", "forward", "hidden_states", "if", "is", "kwargs", "logits", "mean", "not", "output", "output_hidden_states", "outputs", "pixel_values", "return", "return_dict", "self", "sequence_output", "torch", "tuple", "use_return_dict"], "lfm2/modeling_lfm2.py:Lfm2RMSNorm.__init__": ["ModelRMSNorm", "Module", "Parameter", "__init__", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "lfm2/modeling_lfm2.py:Lfm2RMSNorm.forward": ["ModelRMSNorm", "Module", "True", "class", "def", "dtype", "float32", "forward", "hidden_states", "input_dtype", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "lfm2/modeling_lfm2.py:Lfm2RMSNorm.extra_repr": ["ModelRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "variance_epsilon", "weight"], "lfm2/modeling_lfm2.py:Lfm2RotaryEmbedding.__init__": ["Callable", "False", "ModelConfig", "ModelRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "lfm2/modeling_lfm2.py:Lfm2RotaryEmbedding.compute_default_rope_parameters": ["ModelConfig", "ModelRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "lfm2/modeling_lfm2.py:Lfm2RotaryEmbedding.forward": ["False", "ModelRotaryEmbedding", "Module", "None", "and", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "lfm2/modeling_lfm2.py:Lfm2MLP.__init__": ["False", "Linear", "ModelConfig", "ModelMLP", "Module", "None", "__init__", "bias", "block_auto_adjust_ff_dim", "block_ffn_dim_multiplier", "block_multiple_of", "class", "config", "def", "hidden_size", "if", "int", "intermediate_size", "is", "nn", "not", "self", "super", "w1", "w2", "w3"], "lfm2/modeling_lfm2.py:Lfm2MLP.forward": ["F", "ModelMLP", "Module", "class", "def", "forward", "nn", "return", "self", "silu", "w1", "w2", "w3", "x"], "lfm2/modeling_lfm2.py:Lfm2HybridConvCache.__init__": ["ModelConfig", "ModelHybridConvCache", "None", "Tensor", "_", "__init__", "_dtype", "append", "class", "config", "conv_L_cache", "conv_cache", "conv_state", "def", "device", "dtype", "else", "first_attention_layer", "float32", "for", "full_attention", "hidden_size", "if", "in", "index", "int", "is", "key_cache", "layer_types", "list", "max_batch_size", "not", "num_hidden_layers", "range", "self", "str", "tensor", "torch", "value_cache", "zeros"], "lfm2/modeling_lfm2.py:Lfm2HybridConvCache.update": ["Any", "ModelHybridConvCache", "None", "Tensor", "cache_kwargs", "cat", "class", "def", "dict", "dim", "else", "if", "int", "key_cache", "key_states", "layer_idx", "numel", "return", "self", "str", "torch", "tuple", "update", "value_cache", "value_states"], "lfm2/modeling_lfm2.py:Lfm2HybridConvCache.reorder_cache": ["LongTensor", "ModelHybridConvCache", "beam_idx", "class", "conv_cache", "def", "device", "for", "if", "in", "index_select", "key_cache", "layer_idx", "len", "numel", "range", "reorder_cache", "self", "to", "torch", "value_cache"], "lfm2/modeling_lfm2.py:Lfm2HybridConvCache.get_seq_length": ["ModelHybridConvCache", "None", "class", "def", "else", "first_attention_layer", "full_attention", "get_seq_length", "if", "int", "key_cache", "layer_idx", "layer_types", "len", "numel", "or", "return", "self", "shape"], "lfm2/modeling_lfm2.py:Lfm2HybridConvCache.get_mask_sizes": ["ModelHybridConvCache", "Tensor", "cache_position", "class", "def", "full_mask_kv_offset", "get_mask_sizes", "get_seq_length", "int", "kv_length", "layer_idx", "past_seen_tokens", "query_length", "return", "self", "shape", "torch", "tuple"], "lfm2/modeling_lfm2.py:Lfm2HybridConvCache.crop": ["ModelHybridConvCache", "abs", "class", "crop", "def", "for", "get_seq_length", "idx", "if", "in", "int", "key_cache", "len", "max_length", "numel", "range", "return", "self", "value_cache"], "lfm2/modeling_lfm2.py:Lfm2HybridConvCache.__len__": ["ModelHybridConvCache", "__len__", "class", "def", "int", "key_cache", "len", "return", "self"], "lfm2/modeling_lfm2.py:Lfm2HybridConvCache.reset": ["ModelHybridConvCache", "class", "conv_cache", "def", "for", "in", "layer_idx", "len", "range", "reset", "self", "zero_"], "lfm2/modeling_lfm2.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "lfm2/modeling_lfm2.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cos", "def", "k", "k_embed", "q", "q_embed", "return", "rotate_half", "sin", "unsqueeze", "unsqueeze_dim"], "lfm2/modeling_lfm2.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "lfm2/modeling_lfm2.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value", "value_states"], "lfm2/modeling_lfm2.py:Lfm2Attention.__init__": ["False", "Linear", "ModelAttention", "ModelConfig", "ModelRMSNorm", "Module", "True", "__init__", "bias", "class", "config", "def", "eps", "getattr", "head_dim", "hidden_size", "int", "is_causal", "k_layernorm", "k_proj", "layer_idx", "nn", "norm_eps", "num_attention_heads", "num_key_value_groups", "num_key_value_heads", "out_proj", "q_layernorm", "q_proj", "scaling", "self", "super", "v_proj"], "lfm2/modeling_lfm2.py:Lfm2Attention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "LongTensor", "ModelAttention", "ModelHybridConvCache", "Module", "None", "Tensor", "_attn_implementation", "apply_rotary_pos_emb", "attention_interface", "attention_mask", "attn_output", "attn_weights", "cache_kwargs", "cache_position", "class", "config", "contiguous", "cos", "def", "dropout", "eager", "eager_attention_forward", "forward", "head_dim", "hidden_shape", "hidden_states", "if", "input_shape", "is", "k_layernorm", "k_proj", "key_states", "kwargs", "layer_idx", "nn", "not", "out_proj", "output", "past_key_values", "position_embeddings", "q_layernorm", "q_proj", "query_states", "reshape", "return", "scaling", "self", "shape", "sin", "torch", "transpose", "tuple", "update", "v_proj", "value_states", "view"], "lfm2/modeling_lfm2.py:apply_mask_to_padding_states": ["Model_mask_to_padding_states", "None", "and", "attention_mask", "def", "dtype", "hidden_states", "if", "is", "not", "return", "shape", "to"], "lfm2/modeling_lfm2.py:Lfm2ShortConv.__init__": ["Conv1d", "L_cache", "Linear", "ModelConfig", "ModelShortConv", "Module", "__init__", "bias", "class", "config", "conv", "conv_L_cache", "conv_bias", "def", "groups", "hidden_size", "in_channels", "in_proj", "int", "kernel_size", "layer_idx", "nn", "out_channels", "out_proj", "padding", "self", "super"], "lfm2/modeling_lfm2.py:Lfm2ShortConv.cuda_kernels_forward": ["B", "BCx", "Bx", "C", "L_cache", "LongTensor", "ModelHybridConvCache", "ModelShortConv", "Module", "None", "Tensor", "activation", "and", "apply_mask_to_padding_states", "attention_mask", "bias", "cache_position", "causal_conv1d_fn", "causal_conv1d_update", "chunk", "class", "contiguous", "conv", "conv_cache", "conv_out", "conv_state", "conv_weights", "copy_", "cuda_kernels_forward", "def", "dim", "else", "functional", "if", "in_proj", "is", "layer_idx", "nn", "not", "out_proj", "pad", "past_key_values", "return", "self", "shape", "size", "squeeze", "torch", "transpose", "unsqueeze", "view", "weight", "x", "y"], "lfm2/modeling_lfm2.py:Lfm2ShortConv.slow_forward": ["B", "BCx", "Bx", "C", "L_cache", "LongTensor", "ModelHybridConvCache", "ModelShortConv", "Module", "None", "Tensor", "and", "apply_mask_to_padding_states", "attention_mask", "bias", "cache_position", "chunk", "clamp", "class", "contiguous", "conv", "conv_cache", "conv_out", "conv_state", "copy_", "def", "device", "dim", "dims", "dtype", "else", "functional", "if", "in_proj", "is", "layer_idx", "nn", "not", "out_proj", "pad", "past_key_values", "return", "roll", "self", "seqlen", "shape", "shifts", "slow_forward", "sum", "to", "torch", "transpose", "unsqueeze", "weight", "x", "y"], "lfm2/modeling_lfm2.py:Lfm2ShortConv.forward": ["LongTensor", "ModelHybridConvCache", "ModelShortConv", "Module", "None", "Tensor", "and", "attention_mask", "cache_position", "class", "cuda", "cuda_kernels_forward", "def", "device", "forward", "hidden_states", "if", "in", "is_fast_path_available", "is_torchdynamo_compiling", "nn", "not", "past_key_values", "return", "self", "slow_forward", "torch", "type"], "lfm2/modeling_lfm2.py:Lfm2DecoderLayer.__init__": ["GradientCheckpointingLayer", "ModelAttention", "ModelConfig", "ModelDecoderLayer", "ModelMLP", "ModelRMSNorm", "ModelShortConv", "__init__", "class", "config", "conv", "def", "else", "eps", "feed_forward", "ffn_norm", "full_attention", "hidden_size", "if", "int", "is_attention_layer", "layer_idx", "layer_types", "norm_eps", "operator_norm", "self", "self_attn", "super"], "lfm2/modeling_lfm2.py:Lfm2DecoderLayer.forward": ["GradientCheckpointingLayer", "LongTensor", "ModelDecoderLayer", "ModelHybridConvCache", "None", "Tensor", "_", "attention_mask", "cache_position", "class", "conv", "def", "else", "feed_forward", "ffn_norm", "forward", "hidden_states", "if", "is_attention_layer", "kwargs", "operator_norm", "past_key_values", "position_embeddings", "position_ids", "residual", "return", "self", "self_attn", "torch", "tuple"], "lfm2/modeling_lfm2.py:Lfm2Model.__init__": ["Embedding", "False", "ModelConfig", "ModelDecoderLayer", "ModelModel", "ModelPreTrainedModel", "ModelRMSNorm", "ModelRotaryEmbedding", "ModuleList", "__init__", "class", "config", "def", "embed_tokens", "embedding_norm", "eps", "for", "gradient_checkpointing", "hidden_size", "in", "layer_idx", "layers", "nn", "norm_eps", "num_hidden_layers", "pad_token_id", "padding_idx", "post_init", "range", "rotary_emb", "self", "super", "vocab_size"], "lfm2/modeling_lfm2.py:Lfm2Model.forward": ["BaseModelOutputWithPast", "FloatTensor", "LongTensor", "ModelHybridConvCache", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "and", "arange", "attention_mask", "batch_size", "bool", "cache_position", "causal_mask", "class", "config", "create_causal_mask", "decoder_layer", "def", "device", "dtype", "else", "embed_tokens", "embedding_norm", "exactly", "for", "forward", "get_seq_length", "hidden_states", "if", "in", "input_embeds", "input_ids", "inputs_embeds", "is", "is_attention_layer", "kwargs", "last_hidden_state", "layer_mask", "layers", "linear_attention", "max_batch_size", "must", "not", "num_hidden_layers", "of", "one", "or", "past_key_values", "past_seen_tokens", "position_embeddings", "position_ids", "raise", "return", "rotary_emb", "self", "shape", "specify", "torch", "unsqueeze", "use_cache"], "lfm2/modeling_lfm2.py:Lfm2ForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "ModelForCausalLM", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "vocab_size"], "lfm2/modeling_lfm2.py:Lfm2ForCausalLM.forward": ["BaseModelOutputWithPast", "Cache", "CausalLMOutputWithPast", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForCausalLM", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "outputs", "past_key_values", "position_ids", "r", "return", "self", "slice", "slice_indices", "torch", "use_cache", "vocab_size"], "lfm2_moe/modeling_lfm2_moe.py:Lfm2MoeRMSNorm.__init__": ["ModelRMSNorm", "Module", "Parameter", "__init__", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "lfm2_moe/modeling_lfm2_moe.py:Lfm2MoeRMSNorm.forward": ["ModelRMSNorm", "Module", "True", "class", "def", "dtype", "float32", "forward", "hidden_states", "input_dtype", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "lfm2_moe/modeling_lfm2_moe.py:Lfm2MoeRMSNorm.extra_repr": ["ModelRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "variance_epsilon", "weight"], "lfm2_moe/modeling_lfm2_moe.py:Lfm2MoeRotaryEmbedding.__init__": ["Callable", "False", "ModelConfig", "ModelRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "lfm2_moe/modeling_lfm2_moe.py:Lfm2MoeRotaryEmbedding.compute_default_rope_parameters": ["ModelConfig", "ModelRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "lfm2_moe/modeling_lfm2_moe.py:Lfm2MoeRotaryEmbedding.forward": ["False", "ModelRotaryEmbedding", "Module", "None", "and", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "lfm2_moe/modeling_lfm2_moe.py:Lfm2MoeMLP.__init__": ["False", "Linear", "ModelConfig", "ModelMLP", "Module", "None", "__init__", "bias", "class", "config", "def", "else", "hidden_size", "if", "int", "intermediate_size", "is", "nn", "self", "super", "w1", "w2", "w3"], "lfm2_moe/modeling_lfm2_moe.py:Lfm2MoeMLP.forward": ["F", "ModelMLP", "Module", "class", "def", "forward", "nn", "return", "self", "silu", "w1", "w2", "w3", "x"], "lfm2_moe/modeling_lfm2_moe.py:Lfm2MoeExperts.__init__": ["F", "ModelExperts", "Module", "Parameter", "__init__", "act_fn", "class", "config", "def", "down_proj", "empty", "gate_up_proj", "hidden_dim", "hidden_size", "intermediate_dim", "moe_intermediate_size", "nn", "num_experts", "self", "silu", "super", "torch"], "lfm2_moe/modeling_lfm2_moe.py:Lfm2MoeExperts.forward": ["ModelExperts", "Module", "None", "Tensor", "act_fn", "chunk", "class", "continue", "current_hidden_states", "current_state", "def", "dim", "down_proj", "dtype", "expert_hit", "expert_idx", "expert_mask", "final_hidden_states", "for", "forward", "functional", "gate", "gate_up_proj", "greater", "hidden_states", "if", "in", "index_add_", "linear", "nn", "no_grad", "nonzero", "num_classes", "num_experts", "one_hot", "permute", "return", "self", "sum", "to", "token_idx", "top_k_index", "top_k_pos", "top_k_weights", "torch", "up", "where", "with", "zeros_like"], "lfm2_moe/modeling_lfm2_moe.py:Lfm2MoeSparseMoeBlock.__init__": ["False", "Linear", "ModelExperts", "ModelSparseMoeBlock", "Module", "__init__", "bias", "class", "config", "def", "dtype", "expert_bias", "experts", "float32", "gate", "hidden_size", "if", "nn", "norm_topk_prob", "num_experts", "num_experts_per_tok", "register_buffer", "routed_scaling_factor", "self", "super", "top_k", "torch", "use_expert_bias", "zeros"], "lfm2_moe/modeling_lfm2_moe.py:Lfm2MoeSparseMoeBlock.route_tokens_to_experts": ["ModelSparseMoeBlock", "Module", "True", "_", "class", "def", "dim", "else", "expert_bias", "gather", "if", "index", "k", "keepdim", "nn", "norm_topk_prob", "return", "route_tokens_to_experts", "routed_scaling_factor", "router_logits", "routing_weights", "scores_for_routing", "selected_experts", "self", "sigmoid", "sum", "top_k", "topk", "torch", "type_as", "use_expert_bias"], "lfm2_moe/modeling_lfm2_moe.py:Lfm2MoeSparseMoeBlock.forward": ["ModelSparseMoeBlock", "Module", "Tensor", "batch_size", "class", "def", "experts", "final_hidden_states", "forward", "gate", "hidden_dim", "hidden_states", "hidden_states_reshaped", "nn", "reshape", "return", "route_tokens_to_experts", "router_logits", "routing_weights", "selected_experts", "self", "sequence_length", "shape", "torch", "tuple", "view"], "lfm2_moe/modeling_lfm2_moe.py:Lfm2MoeHybridConvCache.__init__": ["ModelConfig", "ModelHybridConvCache", "None", "Tensor", "_", "__init__", "_dtype", "append", "class", "config", "conv_L_cache", "conv_cache", "conv_state", "def", "device", "dtype", "else", "first_attention_layer", "float32", "for", "full_attention", "hidden_size", "if", "in", "index", "int", "is", "key_cache", "layer_types", "list", "max_batch_size", "not", "num_hidden_layers", "range", "self", "str", "tensor", "torch", "value_cache", "zeros"], "lfm2_moe/modeling_lfm2_moe.py:Lfm2MoeHybridConvCache.update": ["Any", "ModelHybridConvCache", "None", "Tensor", "cache_kwargs", "cat", "class", "def", "dict", "dim", "else", "if", "int", "key_cache", "key_states", "layer_idx", "numel", "return", "self", "str", "torch", "tuple", "update", "value_cache", "value_states"], "lfm2_moe/modeling_lfm2_moe.py:Lfm2MoeHybridConvCache.reorder_cache": ["LongTensor", "ModelHybridConvCache", "beam_idx", "class", "conv_cache", "def", "device", "for", "if", "in", "index_select", "key_cache", "layer_idx", "len", "numel", "range", "reorder_cache", "self", "to", "torch", "value_cache"], "lfm2_moe/modeling_lfm2_moe.py:Lfm2MoeHybridConvCache.get_seq_length": ["ModelHybridConvCache", "None", "class", "def", "else", "first_attention_layer", "full_attention", "get_seq_length", "if", "int", "key_cache", "layer_idx", "layer_types", "len", "numel", "or", "return", "self", "shape"], "lfm2_moe/modeling_lfm2_moe.py:Lfm2MoeHybridConvCache.get_mask_sizes": ["ModelHybridConvCache", "Tensor", "cache_position", "class", "def", "full_mask_kv_offset", "get_mask_sizes", "get_seq_length", "int", "kv_length", "layer_idx", "past_seen_tokens", "query_length", "return", "self", "shape", "torch", "tuple"], "lfm2_moe/modeling_lfm2_moe.py:Lfm2MoeHybridConvCache.crop": ["ModelHybridConvCache", "abs", "class", "crop", "def", "for", "get_seq_length", "idx", "if", "in", "int", "key_cache", "len", "max_length", "numel", "range", "return", "self", "value_cache"], "lfm2_moe/modeling_lfm2_moe.py:Lfm2MoeHybridConvCache.__len__": ["ModelHybridConvCache", "__len__", "class", "def", "int", "key_cache", "len", "return", "self"], "lfm2_moe/modeling_lfm2_moe.py:Lfm2MoeHybridConvCache.reset": ["ModelHybridConvCache", "class", "conv_cache", "def", "for", "in", "layer_idx", "len", "range", "reset", "self", "zero_"], "lfm2_moe/modeling_lfm2_moe.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "lfm2_moe/modeling_lfm2_moe.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cos", "def", "k", "k_embed", "q", "q_embed", "return", "rotate_half", "sin", "unsqueeze", "unsqueeze_dim"], "lfm2_moe/modeling_lfm2_moe.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "lfm2_moe/modeling_lfm2_moe.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value", "value_states"], "lfm2_moe/modeling_lfm2_moe.py:Lfm2MoeAttention.__init__": ["False", "Linear", "ModelAttention", "ModelConfig", "ModelRMSNorm", "Module", "True", "__init__", "bias", "class", "config", "def", "eps", "getattr", "head_dim", "hidden_size", "int", "is_causal", "k_layernorm", "k_proj", "layer_idx", "nn", "norm_eps", "num_attention_heads", "num_key_value_groups", "num_key_value_heads", "out_proj", "q_layernorm", "q_proj", "scaling", "self", "super", "v_proj"], "lfm2_moe/modeling_lfm2_moe.py:Lfm2MoeAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "LongTensor", "ModelAttention", "ModelHybridConvCache", "Module", "None", "Tensor", "_attn_implementation", "apply_rotary_pos_emb", "attention_interface", "attention_mask", "attn_output", "attn_weights", "cache_kwargs", "cache_position", "class", "config", "contiguous", "cos", "def", "dropout", "eager", "eager_attention_forward", "forward", "head_dim", "hidden_shape", "hidden_states", "if", "input_shape", "is", "k_layernorm", "k_proj", "key_states", "kwargs", "layer_idx", "nn", "not", "out_proj", "output", "past_key_values", "position_embeddings", "q_layernorm", "q_proj", "query_states", "reshape", "return", "scaling", "self", "shape", "sin", "torch", "transpose", "tuple", "update", "v_proj", "value_states", "view"], "lfm2_moe/modeling_lfm2_moe.py:apply_mask_to_padding_states": ["Model_mask_to_padding_states", "None", "and", "attention_mask", "def", "dtype", "hidden_states", "if", "is", "not", "return", "shape", "to"], "lfm2_moe/modeling_lfm2_moe.py:Lfm2MoeShortConv.__init__": ["Conv1d", "L_cache", "Linear", "ModelConfig", "ModelShortConv", "Module", "__init__", "bias", "class", "config", "conv", "conv_L_cache", "conv_bias", "def", "groups", "hidden_size", "in_channels", "in_proj", "int", "kernel_size", "layer_idx", "nn", "out_channels", "out_proj", "padding", "self", "super"], "lfm2_moe/modeling_lfm2_moe.py:Lfm2MoeShortConv.cuda_kernels_forward": ["B", "BCx", "Bx", "C", "L_cache", "LongTensor", "ModelHybridConvCache", "ModelShortConv", "Module", "None", "Tensor", "activation", "and", "apply_mask_to_padding_states", "attention_mask", "bias", "cache_position", "causal_conv1d_fn", "causal_conv1d_update", "chunk", "class", "contiguous", "conv", "conv_cache", "conv_out", "conv_state", "conv_weights", "copy_", "cuda_kernels_forward", "def", "dim", "else", "functional", "if", "in_proj", "is", "layer_idx", "nn", "not", "out_proj", "pad", "past_key_values", "return", "self", "shape", "size", "squeeze", "torch", "transpose", "unsqueeze", "view", "weight", "x", "y"], "lfm2_moe/modeling_lfm2_moe.py:Lfm2MoeShortConv.slow_forward": ["B", "BCx", "Bx", "C", "L_cache", "LongTensor", "ModelHybridConvCache", "ModelShortConv", "Module", "None", "Tensor", "and", "apply_mask_to_padding_states", "attention_mask", "bias", "cache_position", "chunk", "clamp", "class", "contiguous", "conv", "conv_cache", "conv_out", "conv_state", "copy_", "def", "device", "dim", "dims", "dtype", "else", "functional", "if", "in_proj", "is", "layer_idx", "nn", "not", "out_proj", "pad", "past_key_values", "return", "roll", "self", "seqlen", "shape", "shifts", "slow_forward", "sum", "to", "torch", "transpose", "unsqueeze", "weight", "x", "y"], "lfm2_moe/modeling_lfm2_moe.py:Lfm2MoeShortConv.forward": ["LongTensor", "ModelHybridConvCache", "ModelShortConv", "Module", "None", "Tensor", "and", "attention_mask", "cache_position", "class", "cuda", "cuda_kernels_forward", "def", "device", "forward", "hidden_states", "if", "in", "is_fast_path_available", "is_torchdynamo_compiling", "nn", "not", "past_key_values", "return", "self", "slow_forward", "torch", "type"], "lfm2_moe/modeling_lfm2_moe.py:Lfm2MoeDecoderLayer.__init__": ["GradientCheckpointingLayer", "ModelAttention", "ModelConfig", "ModelDecoderLayer", "ModelMLP", "ModelRMSNorm", "ModelShortConv", "ModelSparseMoeBlock", "__init__", "class", "config", "conv", "def", "else", "eps", "feed_forward", "ffn_norm", "full_attention", "hidden_size", "if", "int", "intermediate_size", "is_attention_layer", "layer_idx", "layer_types", "norm_eps", "num_dense_layers", "operator_norm", "self", "self_attn", "super"], "lfm2_moe/modeling_lfm2_moe.py:Lfm2MoeDecoderLayer.forward": ["GradientCheckpointingLayer", "LongTensor", "ModelDecoderLayer", "ModelHybridConvCache", "None", "Tensor", "_", "attention_mask", "cache_position", "class", "conv", "def", "else", "feed_forward", "ffn_norm", "forward", "hidden_states", "if", "is_attention_layer", "kwargs", "operator_norm", "past_key_values", "position_embeddings", "position_ids", "residual", "return", "self", "self_attn", "torch", "tuple"], "lfm2_moe/modeling_lfm2_moe.py:Lfm2MoePreTrainedModel._init_weights": ["ModelExperts", "ModelPreTrainedModel", "ModelSparseMoeBlock", "PreTrainedModel", "_init_weights", "class", "config", "def", "down_proj", "elif", "expert_bias", "gate_up_proj", "if", "init", "initializer_range", "isinstance", "mean", "module", "normal_", "self", "std", "super", "use_expert_bias", "zeros_"], "lfm2_moe/modeling_lfm2_moe.py:Lfm2MoeModel.__init__": ["Embedding", "False", "ModelConfig", "ModelDecoderLayer", "ModelModel", "ModelPreTrainedModel", "ModelRMSNorm", "ModelRotaryEmbedding", "ModuleList", "__init__", "class", "config", "def", "embed_tokens", "embedding_norm", "eps", "for", "gradient_checkpointing", "hidden_size", "in", "layer_idx", "layers", "nn", "norm_eps", "num_hidden_layers", "pad_token_id", "padding_idx", "pos_emb", "post_init", "range", "self", "super", "vocab_size"], "lfm2_moe/modeling_lfm2_moe.py:Lfm2MoeModel.forward": ["FloatTensor", "LongTensor", "ModelHybridConvCache", "ModelModel", "ModelPreTrainedModel", "MoeModelOutputWithPast", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "and", "arange", "attention_mask", "batch_size", "bool", "cache_position", "causal_mask", "class", "config", "create_causal_mask", "decoder_layer", "def", "device", "dtype", "else", "embed_tokens", "embedding_norm", "exactly", "for", "forward", "get_seq_length", "hidden_states", "if", "in", "input_embeds", "input_ids", "inputs_embeds", "is", "is_attention_layer", "kwargs", "last_hidden_state", "layer_mask", "layers", "linear_attention", "max_batch_size", "must", "not", "num_hidden_layers", "of", "one", "or", "past_key_values", "past_seen_tokens", "pos_emb", "position_embeddings", "position_ids", "raise", "return", "self", "shape", "specify", "torch", "unsqueeze", "use_cache"], "lfm2_moe/modeling_lfm2_moe.py:Lfm2MoeForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "ModelForCausalLM", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "vocab_size"], "lfm2_moe/modeling_lfm2_moe.py:Lfm2MoeForCausalLM.forward": ["BaseModelOutputWithPast", "Cache", "CausalLMOutputWithPast", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForCausalLM", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "outputs", "past_key_values", "position_ids", "r", "return", "self", "slice", "slice_indices", "torch", "use_cache", "vocab_size"], "lfm2_vl/modeling_lfm2_vl.py:Lfm2VlMultiModalProjector.__init__": ["ACT2FN", "LayerNorm", "Linear", "ModelConfig", "ModelMultiModalProjector", "Module", "None", "__init__", "act", "bias", "class", "config", "def", "downsample_factor", "else", "factor", "hidden_size", "if", "in_channels", "layer_norm", "linear_1", "linear_2", "nn", "projector_bias", "projector_hidden_act", "projector_hidden_size", "projector_use_layernorm", "self", "super", "text_config", "use_layer_norm", "vision_config"], "lfm2_vl/modeling_lfm2_vl.py:Lfm2VlMultiModalProjector.forward": ["ModelMultiModalProjector", "Module", "Tensor", "act", "class", "def", "forward", "hidden_states", "if", "image_features", "layer_norm", "linear_1", "linear_2", "nn", "pixel_unshuffle", "return", "self", "torch", "use_layer_norm"], "lfm2_vl/modeling_lfm2_vl.py:Lfm2VlMultiModalProjector.pixel_unshuffle": ["ModelMultiModalProjector", "Module", "Tensor", "batch_size", "channels", "class", "def", "factor", "height", "hidden_states", "nn", "permute", "pixel_unshuffle", "reshape", "return", "self", "size", "torch", "width"], "lfm2_vl/modeling_lfm2_vl.py:Lfm2VlModel.__init__": ["AutoModel", "ModelConfig", "ModelModel", "ModelMultiModalProjector", "ModelPreTrainedModel", "__init__", "class", "config", "def", "from_config", "language_model", "multi_modal_projector", "post_init", "self", "super", "text_config", "vision_config", "vision_tower"], "lfm2_vl/modeling_lfm2_vl.py:Lfm2VlModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "language_model", "return", "self"], "lfm2_vl/modeling_lfm2_vl.py:Lfm2VlModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "language_model", "self", "set_input_embeddings", "value"], "lfm2_vl/modeling_lfm2_vl.py:Lfm2VlModel.get_image_features": ["FloatTensor", "ModelModel", "ModelPreTrainedModel", "Tensor", "append", "class", "def", "dim", "feature", "feature_org_h", "feature_org_w", "for", "get_image_features", "image_features", "image_outputs", "img_embedding", "img_feature_lengths", "img_idx", "in", "kwargs", "last_hidden_state", "list", "multi_modal_projector", "pixel_attention_mask", "pixel_values", "range", "reshape", "return", "self", "size", "spatial_shapes", "sum", "torch", "unsqueeze", "vision_tower"], "lfm2_vl/modeling_lfm2_vl.py:Lfm2VlModel.get_placeholder_mask": ["FloatTensor", "Image", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "ValueError", "all", "and", "class", "config", "def", "device", "do", "dtype", "else", "expand_as", "f", "features", "get_input_embeddings", "get_placeholder_mask", "if", "image", "image_features", "image_token_id", "input_ids", "inputs_embeds", "is", "long", "match", "n_image_features", "n_image_tokens", "not", "numel", "raise", "return", "self", "shape", "special_image_mask", "sum", "tensor", "to", "tokens", "torch", "unsqueeze"], "lfm2_vl/modeling_lfm2_vl.py:Lfm2VlModel.forward": ["Cache", "FloatTensor", "LongTensor", "ModelModel", "ModelModelOutputWithPast", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "attention_mask", "attentions", "bool", "cache_position", "cat", "class", "def", "device", "dim", "dtype", "else", "exactly", "forward", "get_image_features", "get_input_embeddings", "get_placeholder_mask", "hidden_states", "if", "image_features", "image_hidden_states", "input_ids", "inputs_embeds", "is", "kwargs", "language_model", "last_hidden_state", "masked_scatter", "must", "not", "of", "one", "or", "outputs", "past_key_values", "pixel_attention_mask", "pixel_values", "position_ids", "r", "raise", "return", "self", "spatial_shapes", "special_image_mask", "specify", "to", "torch", "tuple", "use_cache"], "lfm2_vl/modeling_lfm2_vl.py:Lfm2VlForConditionalGeneration.__init__": ["False", "GenerationMixin", "Linear", "ModelConfig", "ModelForConditionalGeneration", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "text_config", "vocab_size"], "lfm2_vl/modeling_lfm2_vl.py:Lfm2VlForConditionalGeneration.get_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "model", "return", "self"], "lfm2_vl/modeling_lfm2_vl.py:Lfm2VlForConditionalGeneration.set_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "model", "self", "set_input_embeddings", "value"], "lfm2_vl/modeling_lfm2_vl.py:Lfm2VlForConditionalGeneration.get_output_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "Module", "class", "def", "get_output_embeddings", "lm_head", "nn", "return", "self"], "lfm2_vl/modeling_lfm2_vl.py:Lfm2VlForConditionalGeneration.get_image_features": ["FloatTensor", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "Tensor", "class", "def", "get_image_features", "kwargs", "model", "pixel_attention_mask", "pixel_values", "return", "self", "spatial_shapes", "torch"], "lfm2_vl/modeling_lfm2_vl.py:Lfm2VlForConditionalGeneration.forward": ["Cache", "FloatTensor", "GenerationMixin", "LongTensor", "ModelCausalLMOutputWithPast", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "else", "forward", "hidden_states", "if", "image_hidden_states", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "outputs", "past_key_values", "pixel_attention_mask", "pixel_values", "position_ids", "r", "return", "self", "slice", "slice_indices", "spatial_shapes", "text_config", "torch", "tuple", "use_cache", "vocab_size"], "lfm2_vl/modeling_lfm2_vl.py:Lfm2VlForConditionalGeneration.prepare_inputs_for_generation": ["False", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "True", "attention_mask", "cache_position", "class", "def", "get", "if", "input_ids", "inputs_embeds", "is_first_iteration", "kwargs", "logits_to_keep", "model_inputs", "not", "or", "past_key_values", "pixel_values", "prepare_inputs_for_generation", "return", "self", "super", "use_cache"], "lightglue/modeling_lightglue.py:LightGluePositionalEncoder.__init__": ["False", "Linear", "ModelConfig", "ModelPositionalEncoder", "Module", "__init__", "bias", "class", "config", "def", "descriptor_dim", "nn", "num_attention_heads", "projector", "self", "super"], "lightglue/modeling_lightglue.py:LightGluePositionalEncoder.forward": ["False", "ModelPositionalEncoder", "Module", "None", "Tensor", "bool", "class", "cos", "cosines", "def", "dim", "else", "embeddings", "forward", "if", "keypoints", "nn", "output", "output_hidden_states", "projected_keypoints", "projector", "repeat_interleave", "return", "self", "sin", "sines", "torch", "tuple"], "lightglue/modeling_lightglue.py:rotate_half": ["Model_half", "def", "dim", "flatten", "return", "rot_x", "stack", "torch", "x", "x1", "x2"], "lightglue/modeling_lightglue.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cos", "def", "dtype", "float", "k", "k_embed", "q", "q_embed", "return", "rotate_half", "sin", "to", "unsqueeze", "unsqueeze_dim"], "lightglue/modeling_lightglue.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "lightglue/modeling_lightglue.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value", "value_states"], "lightglue/modeling_lightglue.py:LightGlueAttention.__init__": ["Linear", "ModelAttention", "ModelConfig", "Module", "True", "__init__", "attention_bias", "attention_dropout", "bias", "class", "config", "def", "getattr", "head_dim", "hidden_size", "int", "is_causal", "k_proj", "layer_idx", "nn", "num_attention_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "q_proj", "scaling", "self", "super", "v_proj"], "lightglue/modeling_lightglue.py:LightGlueAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "FlashAttentionKwargs", "ModelAttention", "Module", "None", "Tensor", "Unpack", "_attn_implementation", "apply_rotary_pos_emb", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "class", "config", "contiguous", "cos", "current_attention_mask", "current_states", "def", "dropout", "eager", "eager_attention_forward", "else", "encoder_attention_mask", "encoder_hidden_states", "forward", "head_dim", "hidden_shape", "hidden_states", "if", "input_shape", "is", "is_cross_attention", "k_proj", "key_states", "kwargs", "nn", "not", "o_proj", "position_embeddings", "q_proj", "query_states", "reshape", "return", "scaling", "self", "shape", "sin", "torch", "training", "transpose", "tuple", "v_proj", "value_states", "view"], "lightglue/modeling_lightglue.py:LightGlueMLP.__init__": ["ACT2FN", "LayerNorm", "Linear", "ModelConfig", "ModelMLP", "Module", "True", "__init__", "activation_fn", "class", "config", "def", "elementwise_affine", "fc1", "fc2", "hidden_act", "hidden_size", "intermediate_size", "layer_norm", "nn", "self", "super"], "lightglue/modeling_lightglue.py:LightGlueMLP.forward": ["ModelMLP", "Module", "Tensor", "activation_fn", "class", "def", "fc1", "fc2", "forward", "hidden_states", "layer_norm", "nn", "return", "self", "torch"], "lightglue/modeling_lightglue.py:LightGlueTransformerLayer.__init__": ["ModelAttention", "ModelConfig", "ModelMLP", "ModelTransformerLayer", "Module", "__init__", "class", "config", "cross_attention", "cross_mlp", "def", "int", "layer_idx", "nn", "self", "self_attention", "self_mlp", "super"], "lightglue/modeling_lightglue.py:LightGlueTransformerLayer.forward": ["False", "ModelTransformerLayer", "Module", "None", "Tensor", "all_attentions", "all_hidden_states", "attention_mask", "attention_output", "batch_size", "bool", "cat", "class", "cross_attention", "cross_attention_hidden_states", "cross_attention_output", "cross_attentions", "cross_intermediate_states", "cross_mlp", "cross_output_states", "def", "descriptor_dim", "descriptors", "dim", "else", "encoder_attention_mask", "encoder_hidden_states", "flip", "forward", "if", "intermediate_states", "is", "keypoints", "nn", "not", "num_keypoints", "output_attentions", "output_hidden_states", "output_states", "position_embeddings", "reshape", "return", "self", "self_attention", "self_attention_descriptors", "self_attention_hidden_states", "self_attentions", "self_mlp", "shape", "torch", "tuple"], "lightglue/modeling_lightglue.py:sigmoid_log_double_softmax": ["Model_log_double_softmax", "Tensor", "batch_size", "certainties", "contiguous", "def", "functional", "logModel", "log_softmax", "matchability0", "matchability1", "new_full", "nn", "num_keypoints_0", "num_keypoints_1", "return", "scores", "scores0", "scores1", "shape", "similarity", "squeeze", "torch", "transpose"], "lightglue/modeling_lightglue.py:LightGlueMatchAssignmentLayer.__init__": ["Linear", "ModelConfig", "ModelMatchAssignmentLayer", "Module", "True", "__init__", "bias", "class", "config", "def", "descriptor_dim", "final_projection", "matchability", "nn", "self", "super"], "lightglue/modeling_lightglue.py:LightGlueMatchAssignmentLayer.forward": ["ModelMatchAssignmentLayer", "Module", "None", "Tensor", "batch_size", "class", "def", "descriptor_dim", "descriptors", "device", "dtype", "final_projection", "finfo", "forward", "if", "is", "m_descriptors", "m_descriptors0", "m_descriptors1", "mask", "mask0", "mask1", "masked_fill", "matchability", "matchability_0", "matchability_1", "min", "nn", "not", "num_keypoints", "reshape", "return", "scores", "self", "shape", "sigmoid_log_double_softmax", "similarity", "tensor", "torch", "transpose", "unsqueeze"], "lightglue/modeling_lightglue.py:LightGlueMatchAssignmentLayer.get_matchability": ["ModelMatchAssignmentLayer", "Module", "Tensor", "class", "def", "descriptors", "functional", "get_matchability", "matchability", "nn", "return", "self", "sigmoid", "squeeze", "torch"], "lightglue/modeling_lightglue.py:LightGlueTokenConfidenceLayer.__init__": ["Linear", "ModelConfig", "ModelTokenConfidenceLayer", "Module", "__init__", "class", "config", "def", "descriptor_dim", "nn", "self", "super", "token"], "lightglue/modeling_lightglue.py:LightGlueTokenConfidenceLayer.forward": ["ModelTokenConfidenceLayer", "Module", "Tensor", "class", "def", "descriptors", "detach", "forward", "functional", "nn", "return", "self", "sigmoid", "squeeze", "token", "torch"], "lightglue/modeling_lightglue.py:get_matches_from_scores": ["Model_matches_from_scores", "None", "Tensor", "_", "arange", "batch_size", "def", "device", "exp", "float", "gather", "indices", "indices0", "indices1", "matches", "matches0", "matches1", "matching_scores", "matching_scores0", "matching_scores1", "max", "max0", "max1", "mutual0", "mutual1", "new_tensor", "reshape", "return", "scores", "shape", "stack", "threshold", "torch", "transpose", "tuple", "valid0", "valid1", "values", "where", "zero"], "lightglue/modeling_lightglue.py:normalize_keypoints": ["Model_keypoints", "None", "Tensor", "def", "device", "dtype", "height", "int", "keypoints", "max", "return", "scale", "shift", "size", "tensor", "torch", "values", "width"], "lightglue/modeling_lightglue.py:LightGlueForKeypointMatching.__init__": ["AutoModelForKeypointDetection", "Identity", "Linear", "ModelConfig", "ModelForKeypointMatching", "ModelMatchAssignmentLayer", "ModelPositionalEncoder", "ModelPreTrainedModel", "ModelTokenConfidenceLayer", "ModelTransformerLayer", "ModuleList", "True", "_", "__init__", "bias", "class", "config", "def", "depth_confidence", "descriptor_decoder_dim", "descriptor_dim", "else", "filter_threshold", "for", "from_config", "i", "if", "in", "input_projection", "keypoint_detector", "keypoint_detector_config", "keypoint_detector_descriptor_dim", "layer_idx", "match_assignment_layers", "nn", "num_hidden_layers", "num_layers", "positional_encoder", "post_init", "range", "self", "super", "token_confidence", "transformer_layers", "trust_remote_code", "width_confidence"], "lightglue/modeling_lightglue.py:LightGlueForKeypointMatching._get_confidence_threshold": ["ModelForKeypointMatching", "ModelPreTrainedModel", "_get_confidence_threshold", "class", "clip", "def", "exp", "float", "int", "layer_index", "np", "num_layers", "return", "self", "threshold"], "lightglue/modeling_lightglue.py:LightGlueForKeypointMatching._keypoint_processing": ["False", "ModelForKeypointMatching", "ModelPreTrainedModel", "None", "Tensor", "_keypoint_processing", "bool", "class", "contiguous", "def", "descriptors", "detach", "input_projection", "keypoint_encoding_output", "keypoints", "output_hidden_states", "positional_encoder", "projected_descriptors", "return", "self", "torch", "tuple"], "lightglue/modeling_lightglue.py:LightGlueForKeypointMatching._get_early_stopped_image_pairs": ["ModelForKeypointMatching", "ModelPreTrainedModel", "Tensor", "_", "_get_confidence_threshold", "_get_early_stopped_image_pairs", "batch_size", "bool", "class", "def", "depth_confidence", "dim", "dtype", "early_stopped_pairs", "else", "float", "if", "int", "keypoint_confidences", "layer_index", "mask", "masked_fill", "num_layers", "num_points", "ones", "ratio_confident", "reshape", "return", "self", "shape", "sum", "threshold", "torch"], "lightglue/modeling_lightglue.py:LightGlueForKeypointMatching._get_keypoint_matching": ["ModelForKeypointMatching", "ModelPreTrainedModel", "None", "_get_keypoint_matching", "class", "def", "descriptors", "early_stops", "filter_threshold", "get_matches_from_scores", "if", "is", "layer_index", "mask", "match_assignment_layers", "matches", "matching_scores", "not", "return", "scores", "self"], "lightglue/modeling_lightglue.py:LightGlueForKeypointMatching._get_pruning_mask": ["ModelForKeypointMatching", "ModelPreTrainedModel", "None", "Tensor", "_get_confidence_threshold", "_get_pruning_mask", "class", "confidences", "def", "if", "int", "is", "keep", "layer_index", "not", "return", "scores", "self", "torch", "width_confidence"], "lightglue/modeling_lightglue.py:LightGlueForKeypointMatching._do_layer_keypoint_pruning": ["False", "ModelForKeypointMatching", "ModelPreTrainedModel", "Tensor", "True", "_", "_do_layer_keypoint_pruning", "_get_pruning_mask", "batch_first", "batch_size", "class", "def", "descriptors", "descriptors_matchability", "for", "get_matchability", "i", "in", "indices", "int", "keypoint_confidences", "keypoints", "layer_index", "mask", "masked_fill", "match_assignment_layers", "pad_sequence", "padding_value", "prune_output", "pruned_descriptors", "pruned_indices", "pruned_keypoints", "pruned_keypoints_0", "pruned_keypoints_1", "pruned_keypoints_mask", "pruned_mask", "pruned_tensor", "range", "return", "self", "shape", "t", "tensor", "torch", "zip"], "lightglue/modeling_lightglue.py:LightGlueForKeypointMatching._concat_early_stopped_outputs": ["ModelForKeypointMatching", "ModelPreTrainedModel", "True", "_concat_early_stopped_outputs", "arange", "batch_first", "class", "def", "early_stops_indices", "final_pruned_keypoints_indices", "final_pruned_keypoints_iterations", "for", "ids", "in", "matches", "matching_scores", "order_indices", "pad_sequence", "padding_value", "return", "self", "shape", "stack", "tensor", "torch"], "lightglue/modeling_lightglue.py:LightGlueForKeypointMatching._do_final_keypoint_pruning": ["ModelForKeypointMatching", "ModelPreTrainedModel", "Tensor", "_", "_do_final_keypoint_pruning", "_matches", "_matching_scores", "batch_size", "clamp", "class", "def", "device", "dtype", "for", "full", "gather", "i", "in", "indices", "indices0", "indices1", "matches", "matches0", "matches1", "matching_scores", "matching_scores0", "matching_scores1", "min", "num_keypoints", "range", "reshape", "return", "self", "shape", "tensor", "torch", "tuple", "where", "zeros"], "lightglue/modeling_lightglue.py:LightGlueForKeypointMatching._match_image_pair": ["ModelForKeypointMatching", "ModelPreTrainedModel", "None", "Tensor", "_", "_concat_early_stopped_outputs", "_do_final_keypoint_pruning", "_do_layer_keypoint_pruning", "_get_early_stopped_image_pairs", "_get_keypoint_matching", "_keypoint_processing", "_match_image_pair", "all", "all_attentions", "all_hidden_states", "and", "any", "arange", "attention", "attention_mask", "batch_size", "bool", "break", "class", "def", "depth_confidence", "descriptors", "device", "dim", "do_early_stop", "do_keypoint_pruning", "dtype", "early_stopped_image_indices", "early_stopped_matches", "early_stopped_matching_scores", "early_stopped_pairs", "early_stops", "early_stops_indices", "else", "expand", "extend", "extended_attention_mask", "final_pruned_keypoints_indices", "final_pruned_keypoints_iterations", "for", "get_extended_attention_mask", "height", "hidden_states", "if", "image_indices", "in", "initial_num_keypoints", "input_shape", "int", "is", "keypoint_1", "keypoint_confidences", "keypoint_detector_descriptor_dim", "keypoint_encoding_output", "keypoints", "keypoints_0", "layer_index", "layer_output", "list", "mask", "matches", "matching_scores", "new_full", "new_zeros", "normalize_keypoints", "not", "num_layers", "num_points", "num_points_per_pair", "ones", "ones_like", "output_attentions", "output_hidden_states", "pruned_keypoints_indices", "pruned_keypoints_iterations", "range", "repeat_interleave", "reshape", "return", "self", "shape", "size", "sum", "tensor", "token_confidence", "torch", "transformer_layers", "tuple", "width", "width_confidence"], "lightglue/modeling_lightglue.py:LightGlueForKeypointMatching.forward": ["FloatTensor", "Input", "LongTensor", "Model", "ModelForKeypointMatching", "ModelKeypointMatchingOutput", "ModelPreTrainedModel", "None", "Union", "ValueError", "_", "_match_image_pair", "a", "absolute_keypoints", "attentions", "batch_size", "be", "bool", "channels", "class", "clone", "config", "def", "descriptors", "else", "forward", "height", "hidden_states", "if", "is", "keypoint_detections", "keypoint_detector", "keypoint_detector_descriptor_dim", "keypoints", "kwargs", "labels", "loss", "mask", "matches", "matching_scores", "must", "ndim", "no", "not", "num_channels", "of", "or", "output_attentions", "output_hidden_states", "pixel_values", "provided", "prune", "raise", "reshape", "return", "self", "shape", "should", "size", "tensor", "to", "torch", "trainable", "tuple", "width"], "lighton_ocr/modeling_lighton_ocr.py:LightOnOcrRMSNorm.__init__": ["ModelRMSNorm", "Module", "Parameter", "__init__", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "lighton_ocr/modeling_lighton_ocr.py:LightOnOcrRMSNorm.forward": ["ModelRMSNorm", "Module", "True", "class", "def", "dtype", "float32", "forward", "hidden_states", "input_dtype", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "lighton_ocr/modeling_lighton_ocr.py:LightOnOcrRMSNorm.extra_repr": ["ModelRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "variance_epsilon", "weight"], "lighton_ocr/modeling_lighton_ocr.py:LightOnOcrPatchMerger.__init__": ["False", "Linear", "ModelConfig", "ModelPatchMerger", "Module", "__init__", "bias", "class", "config", "def", "hidden_size", "merging_layer", "nn", "patch_size", "self", "spatial_merge_size", "super", "vision_config"], "lighton_ocr/modeling_lighton_ocr.py:LightOnOcrPatchMerger.forward": ["ModelPatchMerger", "Module", "Tensor", "append", "cat", "class", "d", "def", "dim", "enumerate", "for", "forward", "functional", "grid", "h", "image_features", "image_grid", "image_index", "image_size", "image_sizes", "image_tokens", "in", "kernel_size", "merging_layer", "nn", "patch_size", "permute", "permuted_tensor", "return", "self", "shape", "spatial_merge_size", "split", "stride", "t", "tokens_per_image", "torch", "unfold", "unsqueeze", "view", "w"], "lighton_ocr/modeling_lighton_ocr.py:LightOnOcrMultiModalProjector.__init__": ["False", "GELU", "Linear", "ModelConfig", "ModelMultiModalProjector", "ModelPatchMerger", "ModelRMSNorm", "Module", "__init__", "act", "bias", "class", "config", "def", "eps", "hidden_size", "linear_1", "linear_2", "nn", "norm", "patch_merger", "rms_norm_eps", "self", "super", "text_config", "vision_config"], "lighton_ocr/modeling_lighton_ocr.py:LightOnOcrMultiModalProjector.forward": ["ModelMultiModalProjector", "Module", "Tensor", "act", "class", "def", "forward", "hidden_states", "image_features", "image_sizes", "linear_1", "linear_2", "nn", "norm", "patch_merger", "return", "self", "torch"], "lighton_ocr/modeling_lighton_ocr.py:LightOnOcrModel.__init__": ["AutoModel", "ModelConfig", "ModelModel", "ModelMultiModalProjector", "ModelPreTrainedModel", "__init__", "class", "config", "def", "from_config", "language_model", "post_init", "self", "super", "text_config", "vision_config", "vision_encoder", "vision_projection"], "lighton_ocr/modeling_lighton_ocr.py:LightOnOcrModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "language_model", "return", "self"], "lighton_ocr/modeling_lighton_ocr.py:LightOnOcrModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "language_model", "self", "set_input_embeddings", "value"], "lighton_ocr/modeling_lighton_ocr.py:LightOnOcrModel.get_image_features": ["ModelModel", "ModelPreTrainedModel", "Tensor", "class", "config", "def", "downsample_ratio", "for", "get_image_features", "height", "image_features", "image_sizes", "in", "last_hidden_state", "list", "patch_size", "pixel_values", "return", "self", "spatial_merge_size", "split", "split_sizes", "squeeze", "torch", "vision_config", "vision_encoder", "vision_projection", "visual_features", "width"], "lighton_ocr/modeling_lighton_ocr.py:LightOnOcrModel.get_placeholder_mask": ["FloatTensor", "Image", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "ValueError", "all", "and", "class", "config", "def", "device", "do", "dtype", "else", "expand_as", "f", "features", "get_input_embeddings", "get_placeholder_mask", "if", "image", "image_features", "image_token_id", "input_ids", "inputs_embeds", "is", "long", "match", "n_image_features", "n_image_tokens", "not", "numel", "raise", "return", "self", "shape", "special_image_mask", "sum", "tensor", "to", "tokens", "torch", "unsqueeze"], "lighton_ocr/modeling_lighton_ocr.py:LightOnOcrModel.forward": ["Cache", "FloatTensor", "LongTensor", "ModelModel", "ModelModelOutputWithPast", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "ValueError", "You", "attention_mask", "attentions", "bool", "cache_position", "cat", "class", "config", "def", "device", "dim", "dtype", "else", "exactly", "forward", "get_image_features", "get_input_embeddings", "get_placeholder_mask", "hidden_states", "if", "image_features", "image_hidden_states", "image_sizes", "input_ids", "inputs_embeds", "is", "kwargs", "language_model", "last_hidden_state", "masked_scatter", "must", "not", "of", "one", "or", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "pixel_values", "position_ids", "raise", "return", "return_dict", "self", "special_image_mask", "specify", "to", "torch", "tuple", "use_cache", "use_return_dict"], "lighton_ocr/modeling_lighton_ocr.py:LightOnOcrForConditionalGeneration.__init__": ["False", "GenerationMixin", "Linear", "ModelConfig", "ModelForConditionalGeneration", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "text_config", "vocab_size"], "lighton_ocr/modeling_lighton_ocr.py:LightOnOcrForConditionalGeneration.get_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "model", "return", "self"], "lighton_ocr/modeling_lighton_ocr.py:LightOnOcrForConditionalGeneration.set_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "model", "self", "set_input_embeddings", "value"], "lighton_ocr/modeling_lighton_ocr.py:LightOnOcrForConditionalGeneration.get_output_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "Module", "class", "def", "get_output_embeddings", "lm_head", "nn", "return", "self"], "lighton_ocr/modeling_lighton_ocr.py:LightOnOcrForConditionalGeneration.get_image_features": ["FloatTensor", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "Tensor", "class", "def", "get_image_features", "image_sizes", "kwargs", "model", "pixel_values", "return", "self", "torch"], "lighton_ocr/modeling_lighton_ocr.py:LightOnOcrForConditionalGeneration.forward": ["Cache", "FloatTensor", "GenerationMixin", "LongTensor", "ModelCausalLMOutputWithPast", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "else", "forward", "hidden_states", "if", "image_hidden_states", "image_sizes", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "pixel_values", "position_ids", "r", "return", "return_dict", "self", "slice", "slice_indices", "text_config", "torch", "tuple", "use_cache", "use_return_dict", "vocab_size"], "lighton_ocr/modeling_lighton_ocr.py:LightOnOcrForConditionalGeneration.prepare_inputs_for_generation": ["False", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "True", "attention_mask", "cache_position", "class", "def", "get", "if", "input_ids", "inputs_embeds", "is_first_iteration", "kwargs", "logits_to_keep", "model_inputs", "not", "or", "past_key_values", "pixel_values", "prepare_inputs_for_generation", "return", "self", "super", "use_cache"], "lilt/modeling_lilt.py:LiltTextEmbeddings.__init__": ["Dropout", "Embedding", "False", "LayerNorm", "ModelTextEmbeddings", "Module", "__init__", "arange", "class", "config", "def", "dropout", "eps", "expand", "hidden_dropout_prob", "hidden_size", "layer_norm_eps", "max_position_embeddings", "nn", "pad_token_id", "padding_idx", "persistent", "position_embeddings", "position_ids", "register_buffer", "self", "super", "token_type_embeddings", "torch", "type_vocab_size", "vocab_size", "word_embeddings"], "lilt/modeling_lilt.py:LiltTextEmbeddings.forward": ["LayerNorm", "ModelTextEmbeddings", "Module", "None", "class", "create_position_ids_from_input_ids", "create_position_ids_from_inputs_embeds", "def", "device", "dropout", "dtype", "else", "embeddings", "forward", "if", "input_ids", "input_shape", "inputs_embeds", "is", "long", "nn", "not", "padding_idx", "position_embeddings", "position_ids", "return", "self", "size", "to", "token_type_embeddings", "token_type_ids", "torch", "word_embeddings", "zeros"], "lilt/modeling_lilt.py:LiltTextEmbeddings.create_position_ids_from_input_ids": ["ModelTextEmbeddings", "Module", "class", "create_position_ids_from_input_ids", "cumsum", "def", "dim", "incremental_indices", "input_ids", "int", "long", "mask", "ne", "nn", "padding_idx", "return", "self", "torch", "type_as"], "lilt/modeling_lilt.py:LiltTextEmbeddings.create_position_ids_from_inputs_embeds": ["ModelTextEmbeddings", "Module", "arange", "class", "create_position_ids_from_inputs_embeds", "def", "device", "dtype", "expand", "input_shape", "inputs_embeds", "long", "nn", "padding_idx", "position_ids", "return", "self", "sequence_length", "size", "torch", "unsqueeze"], "lilt/modeling_lilt.py:LiltLayoutEmbeddings.__init__": ["Dropout", "Embedding", "LayerNorm", "Linear", "ModelLayoutEmbeddings", "Module", "__init__", "box_linear_embeddings", "box_position_embeddings", "channel_shrink_ratio", "class", "config", "def", "dropout", "eps", "h_position_embeddings", "hidden_dropout_prob", "hidden_size", "in_features", "layer_norm_eps", "max_2d_position_embeddings", "max_position_embeddings", "nn", "out_features", "pad_token_id", "padding_idx", "self", "super", "w_position_embeddings", "x_position_embeddings", "y_position_embeddings"], "lilt/modeling_lilt.py:LiltLayoutEmbeddings.forward": ["IndexError", "LayerNorm", "ModelLayoutEmbeddings", "Module", "None", "The", "as", "bbox", "be", "box_linear_embeddings", "box_position_embeddings", "cat", "class", "coordinate", "def", "dim", "dropout", "e", "except", "forward", "from", "h_position_embeddings", "left_position_embeddings", "lower_position_embeddings", "nn", "position_ids", "raise", "range", "return", "right_position_embeddings", "self", "should", "spatial_position_embeddings", "torch", "try", "upper_position_embeddings", "values", "w_position_embeddings", "within", "x_position_embeddings", "y_position_embeddings"], "lilt/modeling_lilt.py:LiltSelfAttention.__init__": ["Dropout", "Linear", "ModelSelfAttention", "Module", "None", "The", "ValueError", "__init__", "a", "all_head_size", "and", "attention", "attention_head_size", "attention_probs_dropout_prob", "channel_shrink_ratio", "class", "config", "def", "dropout", "embedding_size", "f", "hasattr", "heads", "hidden", "hidden_size", "if", "int", "is", "key", "layer_idx", "layout_key", "layout_query", "layout_value", "multiple", "nn", "not", "num_attention_heads", "number", "of", "query", "raise", "self", "size", "super", "the", "value"], "lilt/modeling_lilt.py:LiltSelfAttention.transpose_for_scores": ["ModelSelfAttention", "Module", "attention_head_size", "class", "def", "new_x_shape", "nn", "num_attention_heads", "permute", "r", "return", "self", "size", "transpose_for_scores", "view", "x"], "lilt/modeling_lilt.py:LiltSelfAttention.forward": ["False", "ModelSelfAttention", "Module", "None", "Softmax", "all_head_size", "attention_head_size", "attention_mask", "attention_probs", "attention_scores", "channel_shrink_ratio", "class", "context_layer", "contiguous", "def", "dim", "dropout", "forward", "hidden_states", "if", "is", "key", "key_layer", "layout_attention_probs", "layout_attention_scores", "layout_context_layer", "layout_inputs", "layout_key", "layout_key_layer", "layout_query", "layout_query_layer", "layout_value", "layout_value_layer", "math", "matmul", "mixed_query_layer", "new_context_layer_shape", "nn", "not", "output_attentions", "outputs", "permute", "query", "query_layer", "r", "return", "self", "size", "sqrt", "tmp_attention_scores", "tmp_layout_attention_scores", "torch", "transpose", "transpose_for_scores", "value", "value_layer", "view"], "lilt/modeling_lilt.py:LiltSelfOutput.__init__": ["Dropout", "LayerNorm", "Linear", "ModelSelfOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "eps", "hidden_dropout_prob", "hidden_size", "layer_norm_eps", "nn", "self", "super"], "lilt/modeling_lilt.py:LiltSelfOutput.forward": ["LayerNorm", "ModelSelfOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "lilt/modeling_lilt.py:LiltAttention.__init__": ["ModelAttention", "ModelSelfAttention", "ModelSelfOutput", "Module", "None", "__init__", "channel_shrink_ratio", "class", "config", "def", "hidden_size", "layer_idx", "layout_output", "nn", "ori_hidden_size", "output", "self", "super"], "lilt/modeling_lilt.py:LiltAttention.forward": ["False", "FloatTensor", "ModelAttention", "Module", "None", "Tensor", "attention_mask", "attention_output", "bool", "class", "def", "forward", "hidden_states", "layout_attention_output", "layout_inputs", "layout_output", "nn", "output", "output_attentions", "outputs", "return", "self", "self_outputs", "torch", "tuple"], "lilt/modeling_lilt.py:LiltIntermediate.__init__": ["ACT2FN", "Linear", "ModelIntermediate", "Module", "__init__", "class", "config", "def", "dense", "else", "hidden_act", "hidden_size", "if", "intermediate_act_fn", "intermediate_size", "isinstance", "nn", "self", "str", "super"], "lilt/modeling_lilt.py:LiltIntermediate.forward": ["ModelIntermediate", "Module", "Tensor", "class", "def", "dense", "forward", "hidden_states", "intermediate_act_fn", "nn", "return", "self", "torch"], "lilt/modeling_lilt.py:LiltOutput.__init__": ["Dropout", "LayerNorm", "Linear", "ModelOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "eps", "hidden_dropout_prob", "hidden_size", "intermediate_size", "layer_norm_eps", "nn", "self", "super"], "lilt/modeling_lilt.py:LiltOutput.forward": ["LayerNorm", "ModelOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "lilt/modeling_lilt.py:LiltLayer.__init__": ["GradientCheckpointingLayer", "ModelAttention", "ModelIntermediate", "ModelLayer", "ModelOutput", "None", "__init__", "attention", "channel_shrink_ratio", "chunk_size_feed_forward", "class", "config", "def", "hidden_size", "intermediate", "intermediate_size", "layer_idx", "layout_intermediate", "layout_output", "ori_hidden_size", "ori_intermediate_size", "output", "self", "seq_len_dim", "super"], "lilt/modeling_lilt.py:LiltLayer.forward": ["False", "FloatTensor", "GradientCheckpointingLayer", "ModelLayer", "None", "Tensor", "apply_chunking_to_forward", "attention", "attention_mask", "attention_output", "bool", "chunk_size_feed_forward", "class", "def", "feed_forward_chunk", "forward", "hidden_states", "layer_output", "layout_attention_output", "layout_feed_forward_chunk", "layout_inputs", "layout_layer_output", "output_attentions", "outputs", "return", "self", "self_attention_outputs", "seq_len_dim", "torch", "tuple"], "lilt/modeling_lilt.py:LiltLayer.feed_forward_chunk": ["GradientCheckpointingLayer", "ModelLayer", "attention_output", "class", "def", "feed_forward_chunk", "intermediate", "intermediate_output", "layer_output", "output", "return", "self"], "lilt/modeling_lilt.py:LiltLayer.layout_feed_forward_chunk": ["GradientCheckpointingLayer", "ModelLayer", "attention_output", "class", "def", "intermediate_output", "layer_output", "layout_feed_forward_chunk", "layout_intermediate", "layout_output", "return", "self"], "lilt/modeling_lilt.py:LiltEncoder.__init__": ["ModelEncoder", "ModelLayer", "Module", "ModuleList", "_", "__init__", "class", "config", "def", "for", "in", "layer", "nn", "num_hidden_layers", "range", "self", "super"], "lilt/modeling_lilt.py:LiltEncoder.forward": ["BaseModelOutput", "False", "FloatTensor", "ModelEncoder", "Module", "None", "Tensor", "True", "all_hidden_states", "all_self_attentions", "attention_mask", "attentions", "bool", "class", "def", "else", "enumerate", "for", "forward", "hidden_states", "i", "if", "in", "is", "last_hidden_state", "layer", "layer_module", "layer_outputs", "layout_inputs", "nn", "not", "output_attentions", "output_hidden_states", "return", "return_dict", "self", "torch", "tuple", "v"], "lilt/modeling_lilt.py:LiltPooler.__init__": ["Linear", "ModelPooler", "Module", "Tanh", "__init__", "activation", "class", "config", "def", "dense", "hidden_size", "nn", "self", "super"], "lilt/modeling_lilt.py:LiltPooler.forward": ["ModelPooler", "Module", "Tensor", "activation", "class", "def", "dense", "first_token_tensor", "forward", "hidden_states", "nn", "pooled_output", "return", "self", "torch"], "lilt/modeling_lilt.py:LiltPreTrainedModel._init_weights": ["ModelPreTrainedModel", "ModelTextEmbeddings", "PreTrainedModel", "_init_weights", "arange", "class", "copy_", "def", "expand", "if", "init", "isinstance", "module", "position_ids", "self", "shape", "super", "torch"], "lilt/modeling_lilt.py:LiltModel.__init__": ["ModelEncoder", "ModelLayoutEmbeddings", "ModelModel", "ModelPooler", "ModelPreTrainedModel", "ModelTextEmbeddings", "None", "True", "__init__", "add_pooling_layer", "class", "config", "def", "else", "embeddings", "encoder", "if", "layout_embeddings", "pooler", "post_init", "r", "self", "super"], "lilt/modeling_lilt.py:LiltModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "return", "self", "word_embeddings"], "lilt/modeling_lilt.py:LiltModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "self", "set_input_embeddings", "value", "word_embeddings"], "lilt/modeling_lilt.py:LiltModel.forward": ["BaseModelOutputWithPooling", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "ValueError", "You", "and", "at", "attention_mask", "attentions", "batch_size", "bbox", "bool", "both", "buffered_token_type_ids", "buffered_token_type_ids_expanded", "cannot", "class", "config", "def", "device", "dtype", "either", "elif", "else", "embedding_output", "embeddings", "encoder", "encoder_outputs", "expand", "extended_attention_mask", "forward", "get_extended_attention_mask", "hasattr", "have", "hidden_states", "if", "input_ids", "input_shape", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layout_embedding_output", "layout_embeddings", "long", "not", "ones", "or", "output_attentions", "output_hidden_states", "pooled_output", "pooler", "pooler_output", "position_ids", "r", "raise", "return", "return_dict", "same", "self", "seq_length", "sequence_output", "size", "specify", "the", "time", "to", "token_type_ids", "torch", "tuple", "use_return_dict", "warn_if_padding_and_no_attention_mask", "zeros"], "lilt/modeling_lilt.py:LiltForSequenceClassification.__init__": ["False", "Model", "ModelClassificationHead", "ModelForSequenceClassification", "ModelModel", "ModelPreTrainedModel", "__init__", "add_pooling_layer", "class", "classifier", "config", "def", "num_labels", "post_init", "self", "super"], "lilt/modeling_lilt.py:LiltForSequenceClassification.forward": ["BCEWithLogitsLoss", "CrossEntropyLoss", "FloatTensor", "LongTensor", "MSELoss", "Model", "ModelForSequenceClassification", "ModelPreTrainedModel", "None", "SequenceClassifierOutput", "Tensor", "and", "attention_mask", "attentions", "bbox", "bool", "class", "classifier", "config", "def", "device", "dtype", "elif", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "kwargs", "labels", "logits", "long", "loss", "loss_fct", "multi_label_classification", "not", "num_labels", "or", "output", "output_attentions", "output_hidden_states", "outputs", "position_ids", "problem_type", "r", "regression", "return", "return_dict", "self", "sequence_output", "single_label_classification", "squeeze", "to", "token_type_ids", "torch", "tuple", "use_return_dict", "view"], "lilt/modeling_lilt.py:LiltForTokenClassification.__init__": ["Dropout", "False", "Linear", "Model", "ModelForTokenClassification", "ModelModel", "ModelPreTrainedModel", "None", "__init__", "add_pooling_layer", "class", "classifier", "classifier_dropout", "config", "def", "dropout", "else", "hidden_dropout_prob", "hidden_size", "if", "is", "nn", "not", "num_labels", "post_init", "self", "super"], "lilt/modeling_lilt.py:LiltForTokenClassification.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "Model", "ModelForTokenClassification", "ModelPreTrainedModel", "None", "Tensor", "TokenClassifierOutput", "attention_mask", "attentions", "bbox", "bool", "class", "classifier", "config", "def", "device", "dropout", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "not", "num_labels", "output", "output_attentions", "output_hidden_states", "outputs", "position_ids", "r", "return", "return_dict", "self", "sequence_output", "to", "token_type_ids", "torch", "tuple", "use_return_dict", "view"], "lilt/modeling_lilt.py:LiltClassificationHead.__init__": ["Dropout", "Linear", "ModelClassificationHead", "Module", "None", "__init__", "class", "classifier_dropout", "config", "def", "dense", "dropout", "else", "hidden_dropout_prob", "hidden_size", "if", "is", "nn", "not", "num_labels", "out_proj", "self", "super"], "lilt/modeling_lilt.py:LiltClassificationHead.forward": ["ModelClassificationHead", "Module", "class", "def", "dense", "dropout", "features", "forward", "kwargs", "nn", "out_proj", "return", "self", "tanh", "torch", "x"], "lilt/modeling_lilt.py:LiltForQuestionAnswering.__init__": ["False", "Linear", "Model", "ModelForQuestionAnswering", "ModelModel", "ModelPreTrainedModel", "__init__", "add_pooling_layer", "class", "config", "def", "hidden_size", "nn", "num_labels", "post_init", "qa_outputs", "self", "super"], "lilt/modeling_lilt.py:LiltForQuestionAnswering.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "Model", "ModelForQuestionAnswering", "ModelPreTrainedModel", "None", "QuestionAnsweringModelOutput", "Tensor", "and", "attention_mask", "attentions", "bbox", "bool", "clamp", "class", "config", "contiguous", "def", "dim", "else", "end_logits", "end_loss", "end_positions", "forward", "hidden_states", "if", "ignore_index", "ignored_index", "input_ids", "inputs_embeds", "is", "kwargs", "len", "logits", "loss", "loss_fct", "not", "output", "output_attentions", "output_hidden_states", "outputs", "position_ids", "qa_outputs", "r", "return", "return_dict", "self", "sequence_output", "size", "split", "squeeze", "start_logits", "start_loss", "start_positions", "token_type_ids", "torch", "total_loss", "tuple", "use_return_dict"], "llama/modeling_llama.py:LlamaRMSNorm.__init__": ["ModelRMSNorm", "Module", "Parameter", "__init__", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "llama/modeling_llama.py:LlamaRMSNorm.forward": ["ModelRMSNorm", "Module", "True", "class", "def", "dtype", "float32", "forward", "hidden_states", "input_dtype", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "llama/modeling_llama.py:LlamaRMSNorm.extra_repr": ["ModelRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "variance_epsilon", "weight"], "llama/modeling_llama.py:LlamaRotaryEmbedding.__init__": ["Callable", "False", "ModelConfig", "ModelRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "llama/modeling_llama.py:LlamaRotaryEmbedding.compute_default_rope_parameters": ["ModelConfig", "ModelRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "llama/modeling_llama.py:LlamaRotaryEmbedding.forward": ["False", "ModelRotaryEmbedding", "Module", "None", "and", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "llama/modeling_llama.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "llama/modeling_llama.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cos", "def", "k", "k_embed", "q", "q_embed", "return", "rotate_half", "sin", "unsqueeze", "unsqueeze_dim"], "llama/modeling_llama.py:LlamaMLP.__init__": ["ACT2FN", "Linear", "ModelMLP", "Module", "__init__", "act_fn", "bias", "class", "config", "def", "down_proj", "gate_proj", "hidden_act", "hidden_size", "intermediate_size", "mlp_bias", "nn", "self", "super", "up_proj"], "llama/modeling_llama.py:LlamaMLP.forward": ["ModelMLP", "Module", "act_fn", "class", "def", "down_proj", "forward", "gate_proj", "nn", "return", "self", "up_proj", "x"], "llama/modeling_llama.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "llama/modeling_llama.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value", "value_states"], "llama/modeling_llama.py:LlamaAttention.__init__": ["Linear", "ModelAttention", "ModelConfig", "Module", "True", "__init__", "attention_bias", "attention_dropout", "bias", "class", "config", "def", "getattr", "head_dim", "hidden_size", "int", "is_causal", "k_proj", "layer_idx", "nn", "num_attention_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "q_proj", "scaling", "self", "super", "v_proj"], "llama/modeling_llama.py:LlamaAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "LongTensor", "ModelAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_attn_implementation", "apply_rotary_pos_emb", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "cache_kwargs", "cache_position", "class", "config", "contiguous", "cos", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "head_dim", "hidden_shape", "hidden_states", "if", "input_shape", "is", "k_proj", "key_states", "kwargs", "layer_idx", "nn", "not", "o_proj", "past_key_values", "position_embeddings", "q_proj", "query_states", "reshape", "return", "scaling", "self", "shape", "sin", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "view"], "llama/modeling_llama.py:LlamaDecoderLayer.__init__": ["GradientCheckpointingLayer", "ModelAttention", "ModelConfig", "ModelDecoderLayer", "ModelMLP", "ModelRMSNorm", "__init__", "class", "config", "def", "eps", "hidden_size", "input_layernorm", "int", "layer_idx", "mlp", "post_attention_layernorm", "rms_norm_eps", "self", "self_attn", "super"], "llama/modeling_llama.py:LlamaDecoderLayer.forward": ["Cache", "False", "GradientCheckpointingLayer", "LongTensor", "ModelDecoderLayer", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "attention_mask", "bool", "cache_position", "class", "def", "forward", "hidden_states", "input_layernorm", "kwargs", "mlp", "past_key_values", "position_embeddings", "position_ids", "post_attention_layernorm", "residual", "return", "self", "self_attn", "torch", "tuple", "use_cache"], "llama/modeling_llama.py:LlamaModel.__init__": ["Embedding", "False", "ModelConfig", "ModelDecoderLayer", "ModelModel", "ModelPreTrainedModel", "ModelRMSNorm", "ModelRotaryEmbedding", "ModuleList", "__init__", "class", "config", "def", "embed_tokens", "eps", "for", "gradient_checkpointing", "hidden_size", "in", "layer_idx", "layers", "nn", "norm", "num_hidden_layers", "pad_token_id", "padding_idx", "post_init", "range", "rms_norm_eps", "rotary_emb", "self", "super", "vocab_size"], "llama/modeling_llama.py:LlamaModel.forward": ["BaseModelOutputWithPast", "Cache", "DynamicCache", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "and", "arange", "attention_mask", "bool", "cache_position", "causal_mask", "class", "config", "create_causal_mask", "decoder_layer", "def", "device", "else", "embed_tokens", "exactly", "for", "forward", "get_seq_length", "hidden_states", "if", "in", "input_embeds", "input_ids", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layers", "must", "norm", "not", "num_hidden_layers", "of", "one", "or", "past_key_values", "past_seen_tokens", "position_embeddings", "position_ids", "raise", "return", "rotary_emb", "self", "shape", "specify", "torch", "unsqueeze", "use_cache"], "llama/modeling_llama.py:LlamaForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "ModelForCausalLM", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "vocab_size"], "llama/modeling_llama.py:LlamaForCausalLM.forward": ["BaseModelOutputWithPast", "Cache", "CausalLMOutputWithPast", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForCausalLM", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "outputs", "past_key_values", "position_ids", "r", "return", "self", "slice", "slice_indices", "torch", "use_cache", "vocab_size"], "llama4/modeling_llama4.py:Llama4TextExperts.__init__": ["ACT2FN", "ModelTextConfig", "ModelTextExperts", "Module", "Parameter", "__init__", "act_fn", "class", "config", "def", "down_proj", "empty", "expert_dim", "gate_up_proj", "hidden_act", "hidden_size", "intermediate_size", "nn", "num_experts", "num_local_experts", "self", "super", "torch", "zeros"], "llama4/modeling_llama4.py:Llama4TextExperts.forward": ["ModelTextExperts", "Module", "Tensor", "act_fn", "bmm", "chunk", "class", "def", "dim", "down_proj", "forward", "gate", "gate_up", "gate_up_proj", "hidden_size", "hidden_states", "next_states", "nn", "return", "self", "shape", "torch", "up", "view"], "llama4/modeling_llama4.py:Llama4TextMLP.__init__": ["ACT2FN", "False", "Linear", "ModelTextMLP", "Module", "None", "__init__", "activation_fn", "bias", "class", "config", "def", "down_proj", "gate_proj", "hidden_act", "hidden_size", "if", "intermediate_size", "is", "nn", "self", "super", "up_proj"], "llama4/modeling_llama4.py:Llama4TextMLP.forward": ["ModelTextMLP", "Module", "activation_fn", "class", "def", "down_proj", "forward", "gate_proj", "nn", "return", "self", "up_proj", "x"], "llama4/modeling_llama4.py:Llama4TextL2Norm.__init__": ["ModelTextL2Norm", "Module", "__init__", "class", "def", "eps", "float", "nn", "self", "super", "torch"], "llama4/modeling_llama4.py:Llama4TextL2Norm._norm": ["ModelTextL2Norm", "Module", "True", "_norm", "class", "def", "eps", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "torch", "x"], "llama4/modeling_llama4.py:Llama4TextL2Norm.forward": ["ModelTextL2Norm", "Module", "_norm", "class", "def", "float", "forward", "nn", "return", "self", "torch", "type_as", "x"], "llama4/modeling_llama4.py:Llama4TextL2Norm.extra_repr": ["ModelTextL2Norm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "torch"], "llama4/modeling_llama4.py:Llama4TextRMSNorm.__init__": ["ModelTextRMSNorm", "Module", "Parameter", "__init__", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "weight"], "llama4/modeling_llama4.py:Llama4TextRMSNorm._norm": ["ModelTextRMSNorm", "Module", "True", "_norm", "class", "def", "eps", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "torch", "x"], "llama4/modeling_llama4.py:Llama4TextRMSNorm.forward": ["ModelTextRMSNorm", "Module", "_norm", "class", "def", "float", "forward", "nn", "output", "return", "self", "type_as", "weight", "x"], "llama4/modeling_llama4.py:Llama4TextRMSNorm.extra_repr": ["ModelTextRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "weight"], "llama4/modeling_llama4.py:Llama4Router.__init__": ["False", "Linear", "ModelRouter", "__init__", "bias", "class", "config", "def", "hidden_size", "nn", "num_experts", "num_experts_per_tok", "num_local_experts", "self", "super", "top_k"], "llama4/modeling_llama4.py:Llama4Router.forward": ["Linear", "ModelRouter", "class", "def", "dim", "dtype", "float", "forward", "full_like", "functional", "hidden_states", "inf", "nn", "return", "router_indices", "router_logits", "router_scores", "router_top_value", "scatter_", "self", "sigmoid", "super", "to", "top_k", "topk", "torch"], "llama4/modeling_llama4.py:Llama4TextMoe.__init__": ["ModelRouter", "ModelTextExperts", "ModelTextMLP", "ModelTextMoe", "Module", "__init__", "class", "config", "def", "experts", "hidden_dim", "hidden_size", "nn", "num_experts", "num_experts_per_tok", "num_local_experts", "router", "self", "shared_expert", "super", "top_k"], "llama4/modeling_llama4.py:Llama4TextMoe.forward": ["ModelTextMoe", "Module", "add_", "class", "def", "dim", "experts", "forward", "hidden_dim", "hidden_states", "nn", "out", "repeat", "reshape", "return", "routed_in", "routed_out", "router", "router_logits", "router_scores", "self", "shape", "shared_expert", "sum", "transpose"], "llama4/modeling_llama4.py:Llama4TextRotaryEmbedding.__init__": ["Callable", "False", "ModelTextConfig", "ModelTextRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "llama4/modeling_llama4.py:Llama4TextRotaryEmbedding.compute_default_rope_parameters": ["ModelTextConfig", "ModelTextRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "llama4/modeling_llama4.py:Llama4TextRotaryEmbedding.forward": ["False", "ModelTextRotaryEmbedding", "Module", "None", "and", "attention_scaling", "class", "cpu", "def", "device", "device_type", "else", "enabled", "expand", "float", "forward", "freqs", "freqs_cis", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "nn", "ones_like", "polar", "position_ids", "position_ids_expanded", "return", "self", "shape", "str", "to", "torch", "transpose", "type", "with", "x"], "llama4/modeling_llama4.py:apply_rotary_emb": ["Model_rotary_emb", "None", "Tensor", "def", "flatten", "float", "freqs_cis", "reshape", "return", "shape", "torch", "tuple", "type_as", "view_as_complex", "view_as_real", "xk", "xk_", "xk_out", "xq", "xq_", "xq_out"], "llama4/modeling_llama4.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "llama4/modeling_llama4.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "float", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "torch", "training", "transpose", "value", "value_states"], "llama4/modeling_llama4.py:vision_eager_attention_forward": ["Model_eager_attention_forward", "Module", "None", "Tensor", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "float", "functional", "head_dim", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "torch", "training", "transpose", "value", "value_states"], "llama4/modeling_llama4.py:Llama4TextAttention.__init__": ["Linear", "ModelTextAttention", "ModelTextConfig", "ModelTextL2Norm", "Module", "True", "__init__", "and", "attention_bias", "attention_dropout", "attn_scale", "attn_temperature_tuning", "bias", "class", "config", "def", "floor_scale", "getattr", "head_dim", "hidden_size", "if", "is_causal", "k_proj", "layer_idx", "nn", "no_rope_layers", "num_attention_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "q_proj", "qk_norm", "rms_norm_eps", "scaling", "self", "super", "use_qk_norm", "use_rope", "v_proj"], "llama4/modeling_llama4.py:Llama4TextAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "FlashAttentionKwargs", "LongTensor", "ModelTextAttention", "Module", "None", "Tensor", "Unpack", "_attn_implementation", "and", "apply_rotary_emb", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_scale", "attn_scales", "attn_temperature_tuning", "attn_weights", "cache_kwargs", "cache_position", "class", "config", "contiguous", "def", "device", "dropout", "dtype", "eager", "eager_attention_forward", "else", "expand", "float", "floor", "floor_scale", "forward", "hasattr", "head_dim", "hidden_shape", "hidden_states", "if", "input_shape", "is", "k_proj", "key_states", "kwargs", "layer_idx", "log1p", "nn", "not", "o_proj", "past_key_values", "position_embeddings", "q_proj", "qk_norm", "query_states", "reshape", "return", "scaling", "self", "shape", "to", "torch", "training", "transpose", "tuple", "update", "use_rope", "v_proj", "value_states", "view"], "llama4/modeling_llama4.py:Llama4TextDecoderLayer.__init__": ["GradientCheckpointingLayer", "ModelTextAttention", "ModelTextDecoderLayer", "ModelTextMLP", "ModelTextMoe", "ModelTextRMSNorm", "__init__", "attention_type", "class", "config", "def", "else", "eps", "feed_forward", "hidden_size", "if", "in", "input_layernorm", "intermediate_size", "intermediate_size_mlp", "is_moe_layer", "layer_idx", "layer_types", "moe_layers", "post_attention_layernorm", "rms_norm_eps", "self", "self_attn", "super"], "llama4/modeling_llama4.py:Llama4TextDecoderLayer.forward": ["Cache", "False", "FlashAttentionKwargs", "FloatTensor", "GradientCheckpointingLayer", "LongTensor", "ModelTextDecoderLayer", "None", "Tensor", "Unpack", "_", "attention_mask", "attention_states", "bool", "cache_position", "class", "def", "feed_forward", "forward", "hidden_states", "if", "input_layernorm", "is_moe_layer", "kwargs", "past_key_values", "position_embeddings", "position_ids", "post_attention_layernorm", "residual", "return", "self", "self_attn", "shape", "torch", "tuple", "use_cache", "view"], "llama4/modeling_llama4.py:Llama4PreTrainedModel._init_weights": ["ModelPreTrainedModel", "ModelTextExperts", "ModelVisionModel", "PreTrainedModel", "_init_weights", "class", "class_embedding", "config", "def", "down_proj", "elif", "else", "gate_up_proj", "hasattr", "if", "init", "initializer_range", "isinstance", "mean", "module", "normal_", "positional_embedding_vlm", "scale", "self", "std", "super", "text_config"], "llama4/modeling_llama4.py:Llama4TextModel.__init__": ["Embedding", "False", "ModelPreTrainedModel", "ModelTextConfig", "ModelTextDecoderLayer", "ModelTextModel", "ModelTextRMSNorm", "ModelTextRotaryEmbedding", "ModuleList", "__init__", "class", "config", "def", "embed_tokens", "eps", "for", "gradient_checkpointing", "hidden_size", "in", "layer_idx", "layers", "nn", "norm", "num_hidden_layers", "pad_token_id", "padding_idx", "post_init", "range", "rms_norm_eps", "rotary_emb", "self", "super", "vocab_size"], "llama4/modeling_llama4.py:Llama4TextModel.forward": ["BaseModelOutputWithPast", "Cache", "DynamicCache", "FloatTensor", "LongTensor", "ModelPreTrainedModel", "ModelTextModel", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "and", "arange", "attention_mask", "attention_type", "bool", "cache_position", "causal_mask_mapping", "chunked_attention", "class", "config", "create_causal_mask", "create_chunked_causal_mask", "decoder_layer", "def", "device", "dict", "else", "embed_tokens", "exactly", "for", "forward", "freq_cis", "full_attention", "get_seq_length", "hidden_states", "if", "in", "input_embeds", "input_ids", "inputs_embeds", "is", "isinstance", "kwargs", "last_hidden_state", "layers", "mask_kwargs", "must", "norm", "not", "num_hidden_layers", "of", "one", "or", "past_key_values", "past_seen_tokens", "position_embeddings", "position_ids", "raise", "return", "rotary_emb", "self", "shape", "specify", "to", "torch", "tuple", "unsqueeze", "use_cache", "weight"], "llama4/modeling_llama4.py:Llama4ForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "ModelForCausalLM", "ModelPreTrainedModel", "ModelTextConfig", "ModelTextModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "vocab_size"], "llama4/modeling_llama4.py:Llama4ForCausalLM.forward": ["Cache", "CausalLMOutputWithPast", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForCausalLM", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "outputs", "past_key_values", "position_ids", "r", "return", "self", "slice", "slice_indices", "torch", "tuple", "use_cache", "vocab_size"], "llama4/modeling_llama4.py:Llama4VisionMLP2.__init__": ["False", "GELU", "Linear", "ModelVisionMLP2", "Module", "__init__", "activation_fn", "bias", "class", "config", "def", "dropout", "fc1", "fc2", "hidden_size", "intermediate_size", "nn", "projector_dropout", "projector_input_dim", "projector_output_dim", "self", "super", "torch"], "llama4/modeling_llama4.py:Llama4VisionMLP2.forward": ["F", "ModelVisionMLP2", "Module", "activation_fn", "class", "def", "dropout", "fc1", "fc2", "forward", "hidden_states", "nn", "p", "return", "self", "torch", "training"], "llama4/modeling_llama4.py:Llama4MultiModalProjector.__init__": ["False", "Linear", "ModelMultiModalProjector", "Module", "__init__", "bias", "class", "config", "def", "hidden_size", "linear_1", "nn", "self", "super", "text_config", "vision_config", "vision_output_dim"], "llama4/modeling_llama4.py:Llama4MultiModalProjector.forward": ["ModelMultiModalProjector", "Module", "class", "def", "forward", "hidden_states", "image_features", "linear_1", "nn", "return", "self"], "llama4/modeling_llama4.py:pixel_shuffle": ["Model_shuffle", "batch_size", "channels", "contiguous", "def", "height", "input_tensor", "int", "math", "num_patches", "output_tensor", "patch_size", "permute", "reshaped_tensor", "return", "shape", "shuffle_ratio", "size", "sqrt", "view", "width"], "llama4/modeling_llama4.py:Llama4VisionPixelShuffleMLP.__init__": ["ModelVisionMLP2", "ModelVisionPixelShuffleMLP", "Module", "__init__", "class", "config", "def", "inner_dim", "int", "mlp", "nn", "output_dim", "pixel_shuffle_ratio", "projector_input_dim", "projector_output_dim", "self", "super"], "llama4/modeling_llama4.py:Llama4VisionPixelShuffleMLP.forward": ["ModelVisionPixelShuffleMLP", "Module", "Tensor", "class", "def", "encoded_patches", "forward", "mlp", "nn", "pixel_shuffle", "pixel_shuffle_ratio", "return", "self", "torch"], "llama4/modeling_llama4.py:reshape_for_broadcast": ["Model_for_broadcast", "Tensor", "d", "def", "else", "enumerate", "for", "freqs_ci", "i", "if", "in", "ndim", "or", "query", "return", "shape", "torch", "view"], "llama4/modeling_llama4.py:vision_apply_rotary_emb": ["Model_apply_rotary_emb", "Tensor", "def", "device", "flatten", "float", "freqs_ci", "key", "key_", "key_out", "query", "query_", "query_out", "reshape", "reshape_for_broadcast", "return", "shape", "to", "torch", "tuple", "type_as", "view_as_complex", "view_as_real"], "llama4/modeling_llama4.py:Llama4VisionAttention.__init__": ["Linear", "ModelVisionAttention", "ModelVisionConfig", "Module", "True", "__init__", "attention_dropout", "bias", "class", "config", "def", "embed_dim", "head_dim", "hidden_size", "k_proj", "nn", "num_attention_heads", "num_heads", "num_key_value_groups", "o_proj", "q_proj", "scaling", "self", "super", "v_proj"], "llama4/modeling_llama4.py:Llama4VisionAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "False", "FlashAttentionKwargs", "ModelVisionAttention", "Module", "None", "Tensor", "Unpack", "_attn_implementation", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "class", "config", "contiguous", "def", "dropout", "eager", "else", "flex_attention", "forward", "freqs_ci", "head_dim", "hidden_shape", "hidden_states", "if", "in", "input_shape", "is_causal", "k_proj", "key_states", "kwargs", "nn", "not", "o_proj", "past_key_values", "q_proj", "query_states", "reshape", "return", "scaling", "self", "shape", "torch", "training", "transpose", "tuple", "v_proj", "value_states", "view", "vision_apply_rotary_emb", "vision_eager_attention_forward"], "llama4/modeling_llama4.py:Llama4VisionMLP.__init__": ["GELU", "Linear", "ModelVisionMLP", "Module", "True", "__init__", "activation_fn", "bias", "class", "config", "def", "fc1", "fc2", "hidden_size", "intermediate_size", "nn", "self", "super"], "llama4/modeling_llama4.py:Llama4VisionMLP.forward": ["ModelVisionMLP", "Module", "Tensor", "activation_fn", "class", "def", "fc1", "fc2", "forward", "hidden_states", "nn", "return", "self", "torch"], "llama4/modeling_llama4.py:Llama4VisionEncoderLayer.__init__": ["GradientCheckpointingLayer", "LayerNorm", "ModelVisionAttention", "ModelVisionConfig", "ModelVisionEncoderLayer", "ModelVisionMLP", "__init__", "class", "config", "def", "hidden_size", "input_layernorm", "mlp", "nn", "post_attention_layernorm", "self", "self_attn", "super"], "llama4/modeling_llama4.py:Llama4VisionEncoderLayer.forward": ["GradientCheckpointingLayer", "ModelVisionEncoderLayer", "None", "Tensor", "attention_mask", "attn_weights", "bool", "class", "def", "forward", "freqs_ci", "hidden_state", "if", "input_layernorm", "mlp", "output_attentions", "outputs", "post_attention_layernorm", "residual", "return", "self", "self_attn", "torch"], "llama4/modeling_llama4.py:Llama4VisionEncoder.__init__": ["False", "ModelVisionConfig", "ModelVisionEncoder", "ModelVisionEncoderLayer", "Module", "ModuleList", "_", "__init__", "class", "config", "def", "for", "gradient_checkpointing", "in", "layers", "nn", "num_hidden_layers", "range", "self", "super"], "llama4/modeling_llama4.py:Llama4VisionEncoder.forward": ["BaseModelOutput", "ModelVisionEncoder", "Module", "None", "Tensor", "all_attentions", "attention_mask", "attentions", "bool", "class", "config", "def", "else", "encoder_layer", "encoder_states", "for", "forward", "freqs_ci", "hidden_state", "hidden_states", "if", "in", "is", "last_hidden_state", "layer_outputs", "layers", "nn", "not", "output_attentions", "output_hidden_states", "r", "return", "return_dict", "self", "torch", "tuple", "use_return_dict", "v"], "llama4/modeling_llama4.py:Llama4UnfoldConvolution.__init__": ["False", "Linear", "ModelUnfoldConvolution", "Module", "Unfold", "__init__", "bias", "class", "config", "def", "hidden_size", "if", "int", "isinstance", "kernel_size", "linear", "nn", "num_channels", "patch_size", "self", "stride", "super", "torch", "unfold"], "llama4/modeling_llama4.py:Llama4UnfoldConvolution.forward": ["ModelUnfoldConvolution", "Module", "Tensor", "class", "def", "forward", "hidden_states", "linear", "nn", "permute", "return", "self", "torch", "unfold"], "llama4/modeling_llama4.py:Llama4VisionRotaryEmbedding.__init__": ["ModelVisionRotaryEmbedding", "Module", "None", "__init__", "arange", "cat", "class", "config", "contiguous", "cos", "def", "dim", "dtype", "float", "freq_cis", "freq_dim", "freqs", "freqs_ci", "freqs_x", "freqs_y", "frequencies_x", "frequencies_y", "hidden_size", "idx", "image_size", "img_idx", "int32", "masked_fill", "nn", "num_attention_heads", "patch_size", "repeat_interleave", "reshape", "rope_freq", "rope_theta", "self", "sin", "stack", "super", "torch", "view_as_complex"], "llama4/modeling_llama4.py:Llama4VisionRotaryEmbedding.forward": ["ModelVisionRotaryEmbedding", "Module", "class", "def", "device", "forward", "freqs_ci", "hidden_states", "nn", "return", "self", "to"], "llama4/modeling_llama4.py:Llama4VisionModel.__init__": ["LayerNorm", "ModelPreTrainedModel", "ModelUnfoldConvolution", "ModelVisionConfig", "ModelVisionEncoder", "ModelVisionModel", "ModelVisionPixelShuffleMLP", "ModelVisionRotaryEmbedding", "Parameter", "__init__", "class", "class_embedding", "config", "def", "hidden_size", "image_size", "layernorm_post", "layernorm_pre", "model", "nn", "num_channels", "num_patches", "patch_embedding", "patch_size", "positional_embedding_vlm", "post_init", "randn", "rotary_embedding", "scale", "self", "super", "torch", "vision_adapter"], "llama4/modeling_llama4.py:Llama4VisionModel.get_input_embeddings": ["ModelPreTrainedModel", "ModelVisionModel", "class", "def", "get_input_embeddings", "patch_embedding", "return", "self"], "llama4/modeling_llama4.py:Llama4VisionModel.forward": ["BaseModelOutput", "ModelPreTrainedModel", "ModelVisionModel", "None", "Tensor", "_", "attention_mask", "attentions", "batch_size_times_num_tiles", "bool", "cat", "class", "class_embedding", "config", "def", "device", "dim", "dtype", "else", "expand", "for", "forward", "freqs_ci", "height", "hidden_dim", "hidden_state", "hidden_states", "if", "in", "is", "kwargs", "last_hidden_state", "layernorm_post", "layernorm_pre", "model", "not", "num_channels", "num_chunks", "num_concurrent_media", "num_patches", "output", "output_attentions", "output_hidden_states", "patch_embedding", "pixel_values", "positional_embedding", "positional_embedding_vlm", "r", "reshape", "return", "return_dict", "rotary_embedding", "self", "shape", "to", "torch", "tuple", "use_return_dict", "v", "view", "vision_adapter", "width"], "llama4/modeling_llama4.py:Llama4ForConditionalGeneration.__init__": ["GenerationMixin", "ModelConfig", "ModelForCausalLM", "ModelForConditionalGeneration", "ModelMultiModalProjector", "ModelPreTrainedModel", "ModelVisionModel", "None", "__init__", "class", "config", "def", "else", "if", "is", "language_model", "multi_modal_projector", "not", "pad_token_id", "post_init", "self", "super", "text_config", "vision_config", "vision_model", "vocab_size"], "llama4/modeling_llama4.py:Llama4ForConditionalGeneration.get_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "language_model", "return", "self"], "llama4/modeling_llama4.py:Llama4ForConditionalGeneration.set_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "language_model", "self", "set_input_embeddings", "value"], "llama4/modeling_llama4.py:Llama4ForConditionalGeneration.get_output_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "get_output_embeddings", "language_model", "return", "self"], "llama4/modeling_llama4.py:Llama4ForConditionalGeneration.set_output_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "language_model", "new_embeddings", "self", "set_output_embeddings"], "llama4/modeling_llama4.py:Llama4ForConditionalGeneration.set_decoder": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "decoder", "def", "language_model", "self", "set_decoder"], "llama4/modeling_llama4.py:Llama4ForConditionalGeneration.get_decoder": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "get_decoder", "language_model", "return", "self"], "llama4/modeling_llama4.py:Llama4ForConditionalGeneration.get_image_features": ["False", "FloatTensor", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Unexpected", "ValueError", "class", "def", "default", "f", "feature", "for", "full", "get_image_features", "hidden_state", "if", "image_outputs", "in", "is", "items", "k", "kwargs", "last_hidden_state", "not", "output_hidden_states", "pixel_values", "raise", "return", "select", "self", "str", "strategy", "torch", "v", "vision_feature_select_strategy", "vision_model"], "llama4/modeling_llama4.py:Llama4ForConditionalGeneration.get_placeholder_mask": ["FloatTensor", "GenerationMixin", "Image", "LongTensor", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "ValueError", "all", "and", "class", "config", "def", "device", "do", "dtype", "else", "expand_as", "f", "features", "get_input_embeddings", "get_placeholder_mask", "if", "image", "image_features", "image_token_id", "input_ids", "inputs_embeds", "is", "long", "match", "n_image_tokens", "not", "numel", "raise", "return", "self", "shape", "special_image_mask", "sum", "tensor", "to", "tokens", "torch", "unsqueeze"], "llama4/modeling_llama4.py:Llama4ForConditionalGeneration.forward": ["Cache", "CrossEntropyLoss", "FloatTensor", "GenerationMixin", "LongTensor", "ModelCausalLMOutputWithPast", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "and", "at", "attention_mask", "attentions", "bool", "both", "cache_position", "cannot", "class", "config", "contiguous", "def", "device", "dtype", "either", "else", "exactly", "forward", "get_image_features", "get_input_embeddings", "get_placeholder_mask", "hidden_states", "if", "image_features", "image_hidden_states", "input_ids", "inputs_embeds", "int", "is", "kwargs", "labels", "language_model", "logits", "logits_to_keep", "loss", "loss_fct", "masked_scatter", "multi_modal_projector", "must", "nn", "not", "of", "one", "or", "output", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "pixel_values", "position_ids", "projected_vision_flat", "r", "raise", "return", "return_dict", "same", "self", "shape", "shift_attention_mask", "shift_labels", "shift_logits", "size", "special_image_mask", "specify", "str", "the", "time", "to", "torch", "tuple", "use_cache", "use_return_dict", "view", "vision_config", "vision_feature_select_strategy", "vision_flat"], "llama4/modeling_llama4.py:Llama4ForConditionalGeneration.prepare_inputs_for_generation": ["False", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "True", "attention_mask", "cache_position", "class", "def", "get", "if", "input_ids", "inputs_embeds", "is_first_iteration", "kwargs", "language_model", "logits_to_keep", "model_inputs", "not", "or", "past_key_values", "pixel_values", "prepare_inputs_for_generation", "return", "self", "use_cache"], "llava/modeling_llava.py:LlavaMultiModalProjector.__init__": ["ACT2FN", "Linear", "ModelConfig", "ModelMultiModalProjector", "Module", "__init__", "act", "bias", "class", "config", "def", "else", "hidden_size", "if", "int", "isinstance", "len", "linear_1", "linear_2", "multimodal_projector_bias", "nn", "num_feature_layers", "projector_hidden_act", "self", "super", "text_config", "vision_config", "vision_feature_layer"], "llava/modeling_llava.py:LlavaMultiModalProjector.forward": ["ModelMultiModalProjector", "Module", "act", "class", "def", "forward", "hidden_states", "image_features", "linear_1", "linear_2", "nn", "return", "self"], "llava/modeling_llava.py:LlavaModel.__init__": ["AutoModel", "ModelConfig", "ModelModel", "ModelMultiModalProjector", "ModelPreTrainedModel", "__init__", "class", "config", "def", "from_config", "language_model", "multi_modal_projector", "post_init", "self", "super", "text_config", "vision_config", "vision_tower"], "llava/modeling_llava.py:LlavaModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "language_model", "return", "self"], "llava/modeling_llava.py:LlavaModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "language_model", "self", "set_input_embeddings", "value"], "llava/modeling_llava.py:LlavaModel.get_image_features": ["FloatTensor", "ModelModel", "ModelPreTrainedModel", "None", "True", "Unexpected", "ValueError", "as_tensor", "cat", "class", "config", "def", "default", "device", "dim", "else", "f", "feature", "for", "full", "get_image_features", "hidden_states", "hs", "hs_pool", "if", "image_features", "image_outputs", "image_sizes", "in", "int", "is", "isinstance", "items", "k", "kwargs", "layer_idx", "list", "multi_modal_projector", "not", "output_hidden_states", "patch_size", "pixel_values", "prod", "raise", "return", "select", "selected_image_feature", "self", "split", "split_sizes", "squeeze", "str", "strategy", "tolist", "torch", "v", "vision_feature_layer", "vision_feature_select_strategy", "vision_tower"], "llava/modeling_llava.py:LlavaModel.get_placeholder_mask": ["FloatTensor", "Image", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "ValueError", "all", "and", "class", "config", "def", "device", "do", "dtype", "else", "expand_as", "f", "features", "get_input_embeddings", "get_placeholder_mask", "if", "image", "image_features", "image_token_id", "input_ids", "inputs_embeds", "is", "long", "match", "n_image_features", "n_image_tokens", "not", "numel", "raise", "return", "self", "shape", "special_image_mask", "sum", "tensor", "to", "tokens", "torch", "unsqueeze"], "llava/modeling_llava.py:LlavaModel.forward": ["Cache", "FloatTensor", "LongTensor", "ModelModel", "ModelModelOutputWithPast", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "attention_mask", "attentions", "cache_position", "cat", "class", "config", "def", "device", "dim", "dtype", "else", "exactly", "forward", "get_image_features", "get_input_embeddings", "get_placeholder_mask", "hidden_states", "if", "image_features", "image_hidden_states", "image_sizes", "input_ids", "inputs_embeds", "int", "is", "kwargs", "language_model", "last_hidden_state", "list", "masked_scatter", "must", "not", "of", "one", "or", "outputs", "past_key_values", "pixel_values", "position_ids", "raise", "return", "self", "special_image_mask", "specify", "str", "to", "torch", "tuple", "vision_feature_layer", "vision_feature_select_strategy"], "llava/modeling_llava.py:LlavaForConditionalGeneration.__init__": ["False", "GenerationMixin", "Linear", "ModelConfig", "ModelForConditionalGeneration", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "text_config", "vocab_size"], "llava/modeling_llava.py:LlavaForConditionalGeneration.get_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "model", "return", "self"], "llava/modeling_llava.py:LlavaForConditionalGeneration.set_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "model", "self", "set_input_embeddings", "value"], "llava/modeling_llava.py:LlavaForConditionalGeneration.get_output_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "Module", "class", "def", "get_output_embeddings", "lm_head", "nn", "return", "self"], "llava/modeling_llava.py:LlavaForConditionalGeneration.get_image_features": ["FloatTensor", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "class", "def", "get_image_features", "int", "kwargs", "list", "model", "pixel_values", "return", "self", "str", "torch", "vision_feature_layer", "vision_feature_select_strategy"], "llava/modeling_llava.py:LlavaForConditionalGeneration.forward": ["Cache", "FloatTensor", "GenerationMixin", "LongTensor", "ModelCausalLMOutputWithPast", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "cache_position", "class", "config", "def", "else", "forward", "hidden_states", "if", "image_hidden_states", "image_sizes", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "list", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "outputs", "past_key_values", "pixel_values", "position_ids", "r", "return", "self", "slice", "slice_indices", "str", "text_config", "torch", "tuple", "vision_feature_layer", "vision_feature_select_strategy", "vocab_size"], "llava/modeling_llava.py:LlavaForConditionalGeneration.prepare_inputs_for_generation": ["False", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "True", "attention_mask", "cache_position", "class", "def", "get", "if", "input_ids", "inputs_embeds", "is_first_iteration", "kwargs", "logits_to_keep", "model_inputs", "not", "or", "past_key_values", "pixel_values", "prepare_inputs_for_generation", "return", "self", "super", "use_cache"], "llava_next/modeling_llava_next.py:get_anyres_image_grid_shape": ["Model_anyres_image_grid_shape", "Tensor", "TypeError", "a", "be", "def", "either", "f", "grid_pinpoints", "height", "if", "image_size", "invalid", "isinstance", "list", "lists", "ndarray", "not", "np", "of", "or", "patch_size", "raise", "return", "select_best_resolution", "should", "tensor", "tolist", "torch", "tuple", "tuples", "type", "valid", "width"], "llava_next/modeling_llava_next.py:image_size_to_num_patches": ["Model_size", "Model_size_to_num_patches", "Tensor", "TypeError", "a", "be", "best_resolution", "def", "f", "for", "grid_pinpoints", "height", "i", "if", "in", "int", "invalid", "isinstance", "j", "list", "lists", "ndarray", "not", "np", "num_patches", "of", "or", "patch_size", "raise", "range", "return", "select_best_resolution", "should", "tolist", "torch", "tuple", "tuples", "type", "value", "width", "with"], "llava_next/modeling_llava_next.py:unpad_image": ["Model_image", "Modelded_tensor", "Tensor", "TypeError", "be", "current_aspect_ratio", "current_height", "current_width", "def", "either", "else", "f", "if", "image_size", "int", "invalid", "isinstance", "list", "ndarray", "new_height", "new_width", "not", "np", "or", "original_aspect_ratio", "original_height", "original_size", "original_width", "padding", "raise", "return", "round", "scale_factor", "shape", "should", "tensor", "tolist", "torch", "tuple", "type", "valid"], "llava_next/modeling_llava_next.py:LlavaNextMultiModalProjector.__init__": ["ACT2FN", "Linear", "ModelConfig", "ModelMultiModalProjector", "Module", "__init__", "act", "bias", "class", "config", "def", "else", "hidden_size", "if", "int", "isinstance", "len", "linear_1", "linear_2", "multimodal_projector_bias", "nn", "num_feature_layers", "projector_hidden_act", "self", "super", "text_config", "vision_config", "vision_feature_layer"], "llava_next/modeling_llava_next.py:LlavaNextMultiModalProjector.forward": ["ModelMultiModalProjector", "Module", "act", "class", "def", "forward", "hidden_states", "image_features", "linear_1", "linear_2", "nn", "return", "self"], "llava_next/modeling_llava_next.py:LlavaNextPreTrainedModel._init_weights": ["Linear", "ModelModel", "ModelPreTrainedModel", "None", "PreTrainedModel", "_init_weights", "bias", "class", "config", "def", "elif", "embed_std", "get_text_config", "getattr", "hidden_size", "if", "image_newline", "init", "initializer_range", "is", "isinstance", "math", "mean", "module", "nn", "normal_", "not", "self", "sqrt", "std", "text_config", "weight", "zeros_"], "llava_next/modeling_llava_next.py:LlavaNextModel.__init__": ["AutoModel", "ModelConfig", "ModelModel", "ModelMultiModalProjector", "ModelPreTrainedModel", "None", "Parameter", "__init__", "class", "config", "def", "dtype", "else", "embed_std", "from_config", "hidden_size", "if", "image_newline", "is", "language_model", "math", "multi_modal_projector", "nn", "not", "pad_token_id", "post_init", "randn", "self", "sqrt", "super", "text_config", "torch", "vision_config", "vision_tower", "vocab_size"], "llava_next/modeling_llava_next.py:LlavaNextModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "language_model", "return", "self"], "llava_next/modeling_llava_next.py:LlavaNextModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "language_model", "self", "set_input_embeddings", "value"], "llava_next/modeling_llava_next.py:LlavaNextModel.pack_image_features": ["CLS", "Image", "ModelModel", "ModelPreTrainedModel", "None", "You", "a", "and", "append", "base_image_feature", "be", "cat", "class", "config", "contiguous", "def", "default", "device", "dim", "does", "dtype", "else", "encoder", "enumerate", "expand", "feature", "feature_lens", "flatten", "for", "get_anyres_image_grid_shape", "have", "height", "if", "image_feature", "image_features", "image_grid_pinpoints", "image_idx", "image_newline", "image_size", "image_sizes", "in", "is", "line", "logger", "long", "may", "new_image_features", "not", "np", "num_patch_height", "num_patch_width", "pack_image_features", "patch", "patch_size", "permute", "prod", "provided", "return", "self", "shape", "size", "tensor", "that", "the", "to", "torch", "transpose", "unpad_image", "up", "using", "view", "vision_config", "vision_feature_select_strategy", "visual", "warning_once", "width", "with"], "llava_next/modeling_llava_next.py:LlavaNextModel.get_image_features": ["FloatTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "True", "ValueError", "_pixel_values_list", "be", "cat", "class", "config", "def", "default", "dim", "dimensions", "elif", "else", "expect", "f", "feature_lens", "for", "get_image_features", "grid_pinpoints", "hidden_states", "hs_pool", "if", "image_features", "image_grid_pinpoints", "image_newline", "image_num_patches", "image_size", "image_size_to_num_patches", "image_sizes", "imsize", "in", "int", "is", "isinstance", "layer_idx", "list", "multi_modal_projector", "not", "num_patch", "of", "or", "output_hidden_states", "pack_image_features", "patch_size", "pix_val", "pixel_values", "raise", "return", "selected_image_feature", "self", "shape", "split", "str", "to", "torch", "vision_config", "vision_feature_layer", "vision_feature_select_strategy", "vision_tower", "zip"], "llava_next/modeling_llava_next.py:LlavaNextModel.get_placeholder_mask": ["FloatTensor", "Image", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "ValueError", "all", "and", "class", "config", "def", "device", "do", "dtype", "else", "expand_as", "f", "features", "get_input_embeddings", "get_placeholder_mask", "if", "image", "image_features", "image_token_id", "input_ids", "inputs_embeds", "is", "long", "match", "n_image_tokens", "not", "numel", "raise", "return", "self", "shape", "special_image_mask", "sum", "tensor", "to", "tokens", "torch", "unsqueeze"], "llava_next/modeling_llava_next.py:LlavaNextModel.forward": ["Cache", "FlashAttentionKwargs", "FloatTensor", "LongTensor", "ModelModel", "ModelModelOutputWithPast", "ModelPreTrainedModel", "None", "Tensor", "True", "Unpack", "ValueError", "You", "and", "attention_mask", "attentions", "bool", "cache_position", "cat", "class", "config", "def", "device", "dim", "dtype", "else", "exactly", "forward", "get_image_features", "get_input_embeddings", "get_placeholder_mask", "hidden_states", "if", "image_features", "image_hidden_states", "image_sizes", "input_ids", "inputs_embeds", "int", "is", "kwargs", "language_model", "last_hidden_state", "list", "masked_scatter", "must", "not", "of", "one", "or", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "pixel_values", "position_ids", "r", "raise", "return", "return_dict", "self", "size", "special_image_mask", "specify", "str", "to", "torch", "tuple", "use_cache", "use_return_dict", "vision_feature_layer", "vision_feature_select_strategy"], "llava_next/modeling_llava_next.py:LlavaNextForConditionalGeneration.__init__": ["False", "GenerationMixin", "Linear", "ModelConfig", "ModelForConditionalGeneration", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "text_config", "vocab_size"], "llava_next/modeling_llava_next.py:LlavaNextForConditionalGeneration.get_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "model", "return", "self"], "llava_next/modeling_llava_next.py:LlavaNextForConditionalGeneration.set_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "model", "self", "set_input_embeddings", "value"], "llava_next/modeling_llava_next.py:LlavaNextForConditionalGeneration.get_output_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "Module", "class", "def", "get_output_embeddings", "lm_head", "nn", "return", "self"], "llava_next/modeling_llava_next.py:LlavaNextForConditionalGeneration.pack_image_features": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "class", "def", "image_features", "image_newline", "image_sizes", "model", "pack_image_features", "return", "self", "vision_feature_select_strategy"], "llava_next/modeling_llava_next.py:LlavaNextForConditionalGeneration.get_image_features": ["FloatTensor", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Tensor", "class", "def", "get_image_features", "image_sizes", "int", "list", "model", "pixel_values", "return", "self", "str", "torch", "vision_feature_layer", "vision_feature_select_strategy"], "llava_next/modeling_llava_next.py:LlavaNextForConditionalGeneration.forward": ["Cache", "FloatTensor", "GenerationMixin", "LongTensor", "ModelCausalLMOutputWithPast", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "else", "forward", "hidden_states", "if", "image_hidden_states", "image_sizes", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "list", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "pixel_values", "position_ids", "r", "return", "return_dict", "self", "slice", "slice_indices", "str", "text_config", "torch", "tuple", "use_cache", "vision_feature_layer", "vision_feature_select_strategy", "vocab_size"], "llava_next/modeling_llava_next.py:LlavaNextForConditionalGeneration.prepare_inputs_for_generation": ["False", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "True", "attention_mask", "cache_position", "class", "def", "get", "if", "image_sizes", "input_ids", "inputs_embeds", "is_first_iteration", "kwargs", "logits_to_keep", "model_inputs", "not", "or", "past_key_values", "pixel_values", "prepare_inputs_for_generation", "return", "self", "super", "use_cache"], "llava_next/modeling_llava_next.py:LlavaNextForConditionalGeneration._prepare_4d_causal_attention_mask_with_cache_position": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Tensor", "_prepare_4d_causal_attention_mask_with_cache_position", "and", "arange", "attention_mask", "batch_size", "cache_position", "causal_mask", "class", "clone", "def", "device", "diagonal", "dim", "dtype", "else", "expand", "fill_value", "finfo", "full", "if", "int", "is", "kwargs", "mask_length", "masked_fill", "min", "min_dtype", "not", "padding_mask", "reshape", "return", "sequence_length", "shape", "target_length", "to", "torch", "triu"], "llava_next_video/modeling_llava_next_video.py:LlavaNextVideoPooler.__init__": ["AvgPool2d", "Conv2d", "Has", "MaxPool2d", "ModelPooler", "Module", "Unknown", "ValueError", "__init__", "average", "be", "class", "config", "conv", "def", "elif", "else", "f", "getattr", "hidden_size", "if", "image_size", "in_channels", "kernel_size", "max", "mode", "nn", "of", "one", "out_channels", "patch_size", "pool", "pooling", "raise", "self", "spatial_pool_mode", "spatial_pool_out_channels", "spatial_pool_stride", "stride", "super", "to", "vision_config"], "llava_next_video/modeling_llava_next_video.py:LlavaNextVideoPooler.forward": ["ModelPooler", "Module", "_", "batch_size", "class", "contiguous", "def", "dim", "flatten", "forward", "image_features", "image_features_spatial", "image_features_spatial_pool", "image_size", "int", "math", "nn", "ori_height", "ori_width", "permute", "pool", "return", "self", "shape", "sqrt", "transpose", "view"], "llava_next_video/modeling_llava_next_video.py:LlavaNextVideoMultiModalProjector.__init__": ["ACT2FN", "Linear", "ModelConfig", "ModelMultiModalProjector", "Module", "__init__", "act", "bias", "class", "config", "def", "else", "hidden_size", "if", "int", "isinstance", "len", "linear_1", "linear_2", "multimodal_projector_bias", "nn", "num_feature_layers", "projector_hidden_act", "self", "super", "text_config", "vision_config", "vision_feature_layer"], "llava_next_video/modeling_llava_next_video.py:LlavaNextVideoMultiModalProjector.forward": ["ModelMultiModalProjector", "Module", "act", "class", "def", "forward", "hidden_states", "image_features", "linear_1", "linear_2", "nn", "return", "self"], "llava_next_video/modeling_llava_next_video.py:LlavaNextVideoPreTrainedModel._init_weights": ["Linear", "ModelModel", "ModelPreTrainedModel", "None", "PreTrainedModel", "_init_weights", "bias", "class", "config", "def", "elif", "embed_std", "get_text_config", "getattr", "hidden_size", "if", "image_newline", "init", "initializer_range", "is", "isinstance", "math", "mean", "module", "nn", "normal_", "not", "self", "sqrt", "std", "text_config", "weight", "zeros_"], "llava_next_video/modeling_llava_next_video.py:get_anyres_image_grid_shape": ["Model_anyres_image_grid_shape", "Tensor", "TypeError", "a", "be", "def", "either", "f", "grid_pinpoints", "height", "if", "image_size", "invalid", "isinstance", "list", "lists", "ndarray", "not", "np", "of", "or", "patch_size", "raise", "return", "select_best_resolution", "should", "tensor", "tolist", "torch", "tuple", "tuples", "type", "valid", "width"], "llava_next_video/modeling_llava_next_video.py:image_size_to_num_patches": ["Model_size", "Model_size_to_num_patches", "Tensor", "TypeError", "a", "be", "best_resolution", "def", "f", "for", "grid_pinpoints", "height", "i", "if", "in", "int", "invalid", "isinstance", "j", "list", "lists", "ndarray", "not", "np", "num_patches", "of", "or", "patch_size", "raise", "range", "return", "select_best_resolution", "should", "tolist", "torch", "tuple", "tuples", "type", "value", "width", "with"], "llava_next_video/modeling_llava_next_video.py:unpad_image": ["Model_image", "Modelded_tensor", "Tensor", "TypeError", "be", "current_aspect_ratio", "current_height", "current_width", "def", "either", "else", "f", "if", "image_size", "int", "invalid", "isinstance", "list", "ndarray", "new_height", "new_width", "not", "np", "or", "original_aspect_ratio", "original_height", "original_size", "original_width", "padding", "raise", "return", "round", "scale_factor", "shape", "should", "tensor", "tolist", "torch", "tuple", "type", "valid"], "llava_next_video/modeling_llava_next_video.py:LlavaNextVideoModel.__init__": ["AutoModel", "ModelConfig", "ModelModel", "ModelMultiModalProjector", "ModelPooler", "ModelPreTrainedModel", "None", "Parameter", "__init__", "class", "config", "def", "dtype", "else", "embed_std", "from_config", "hidden_size", "if", "image_newline", "is", "language_model", "math", "multi_modal_projector", "nn", "not", "pad_token_id", "post_init", "randn", "self", "sqrt", "super", "text_config", "torch", "vision_config", "vision_resampler", "vision_tower", "vocab_size"], "llava_next_video/modeling_llava_next_video.py:LlavaNextVideoModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "language_model", "return", "self"], "llava_next_video/modeling_llava_next_video.py:LlavaNextVideoModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "language_model", "self", "set_input_embeddings", "value"], "llava_next_video/modeling_llava_next_video.py:LlavaNextVideoModel.pack_image_features": ["CLS", "Image", "ModelModel", "ModelPreTrainedModel", "None", "You", "a", "and", "append", "base_image_feature", "be", "cat", "class", "config", "contiguous", "def", "default", "device", "dim", "does", "dtype", "else", "encoder", "enumerate", "expand", "feature", "feature_lens", "flatten", "for", "get_anyres_image_grid_shape", "have", "height", "if", "image_feature", "image_features", "image_grid_pinpoints", "image_idx", "image_newline", "image_size", "image_sizes", "in", "is", "line", "logger", "long", "may", "new_image_features", "not", "np", "num_patch_height", "num_patch_width", "pack_image_features", "patch", "patch_size", "permute", "prod", "provided", "return", "self", "shape", "size", "tensor", "that", "the", "to", "torch", "transpose", "unpad_image", "up", "using", "view", "vision_config", "vision_feature_select_strategy", "visual", "warning_once", "width", "with"], "llava_next_video/modeling_llava_next_video.py:LlavaNextVideoModel.get_image_features": ["FloatTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "True", "ValueError", "_pixel_values_list", "be", "cat", "class", "config", "def", "default", "dim", "dimensions", "elif", "else", "expect", "f", "feature_lens", "for", "get_image_features", "grid_pinpoints", "hidden_states", "hs_pool", "if", "image_features", "image_grid_pinpoints", "image_newline", "image_num_patches", "image_size", "image_size_to_num_patches", "image_sizes", "imsize", "in", "int", "is", "isinstance", "layer_idx", "list", "multi_modal_projector", "not", "num_patch", "of", "or", "output_hidden_states", "pack_image_features", "patch_size", "pix_val", "pixel_values", "raise", "return", "selected_image_feature", "self", "shape", "split", "str", "to", "torch", "vision_config", "vision_feature_layer", "vision_feature_select_strategy", "vision_tower", "zip"], "llava_next_video/modeling_llava_next_video.py:LlavaNextVideoModel.get_placeholder_mask": ["FloatTensor", "Image", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "ValueError", "Videos", "all", "and", "class", "config", "def", "device", "do", "dtype", "else", "expand_as", "f", "features", "get_input_embeddings", "get_placeholder_mask", "if", "image", "image_features", "image_token_id", "input_ids", "inputs_embeds", "is", "long", "match", "n_image_tokens", "n_video_tokens", "not", "numel", "raise", "return", "self", "shape", "special_image_mask", "special_video_mask", "sum", "tensor", "to", "tokens", "torch", "unsqueeze", "video_features", "video_token_id"], "llava_next_video/modeling_llava_next_video.py:LlavaNextVideoModel.forward": ["Cache", "FlashAttentionKwargs", "FloatTensor", "LongTensor", "ModelModel", "ModelModelOutputWithPast", "ModelPreTrainedModel", "None", "Tensor", "True", "Unpack", "ValueError", "You", "_", "attention_mask", "attentions", "bool", "cache_position", "cat", "class", "config", "def", "device", "dim", "dtype", "else", "exactly", "feature", "flatten", "for", "forward", "get_image_features", "get_input_embeddings", "get_placeholder_mask", "get_video_features", "hidden_states", "if", "image_features", "image_hidden_states", "image_sizes", "in", "input_ids", "inputs_embeds", "int", "is", "kwargs", "language_model", "last_hidden_state", "list", "long", "masked_scatter", "must", "not", "of", "one", "or", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "pixel_values", "pixel_values_videos", "position_ids", "r", "raise", "return", "return_dict", "self", "size", "special_image_mask", "special_video_mask", "specify", "str", "tensor", "to", "torch", "tuple", "use_cache", "use_return_dict", "video_feature_lens", "video_features", "video_hidden_states", "vision_feature_layer", "vision_feature_select_strategy"], "llava_next_video/modeling_llava_next_video.py:LlavaNextVideoModel.get_video_features": ["FloatTensor", "ModelModel", "ModelPreTrainedModel", "None", "True", "batch_size", "cat", "channels", "class", "config", "def", "default", "dim", "else", "for", "frames", "get_video_features", "height", "hidden_states", "hs_pool", "if", "in", "int", "is", "isinstance", "layer_idx", "list", "multi_modal_projector", "not", "output_hidden_states", "pixel_values", "reshape", "return", "selected_video_features", "self", "shape", "split", "str", "torch", "video_features", "vision_feature_layer", "vision_feature_select_strategy", "vision_resampler", "vision_tower", "width"], "llava_next_video/modeling_llava_next_video.py:LlavaNextVideoForConditionalGeneration.__init__": ["False", "GenerationMixin", "Linear", "ModelConfig", "ModelForConditionalGeneration", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "text_config", "vocab_size"], "llava_next_video/modeling_llava_next_video.py:LlavaNextVideoForConditionalGeneration.get_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "model", "return", "self"], "llava_next_video/modeling_llava_next_video.py:LlavaNextVideoForConditionalGeneration.set_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "model", "self", "set_input_embeddings", "value"], "llava_next_video/modeling_llava_next_video.py:LlavaNextVideoForConditionalGeneration.get_output_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "Module", "class", "def", "get_output_embeddings", "lm_head", "nn", "return", "self"], "llava_next_video/modeling_llava_next_video.py:LlavaNextVideoForConditionalGeneration.pack_image_features": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "class", "def", "image_features", "image_newline", "image_sizes", "model", "pack_image_features", "return", "self", "vision_feature_select_strategy"], "llava_next_video/modeling_llava_next_video.py:LlavaNextVideoForConditionalGeneration.get_image_features": ["FloatTensor", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Tensor", "class", "def", "get_image_features", "image_sizes", "int", "list", "model", "pixel_values", "return", "self", "str", "torch", "vision_feature_layer", "vision_feature_select_strategy"], "llava_next_video/modeling_llava_next_video.py:LlavaNextVideoForConditionalGeneration.forward": ["Cache", "FloatTensor", "GenerationMixin", "LongTensor", "ModelCausalLMOutputWithPast", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "else", "forward", "hidden_states", "if", "image_hidden_states", "image_sizes", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "list", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "pixel_values", "pixel_values_videos", "position_ids", "r", "return", "return_dict", "self", "slice", "slice_indices", "str", "text_config", "torch", "tuple", "use_cache", "use_return_dict", "video_hidden_states", "vision_feature_layer", "vision_feature_select_strategy", "vocab_size"], "llava_next_video/modeling_llava_next_video.py:LlavaNextVideoForConditionalGeneration.prepare_inputs_for_generation": ["False", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "True", "attention_mask", "cache_position", "class", "def", "get", "if", "image_sizes", "input_ids", "inputs_embeds", "is_first_iteration", "kwargs", "logits_to_keep", "model_inputs", "not", "or", "past_key_values", "pixel_values", "pixel_values_videos", "prepare_inputs_for_generation", "return", "self", "super", "use_cache"], "llava_next_video/modeling_llava_next_video.py:LlavaNextVideoForConditionalGeneration._prepare_4d_causal_attention_mask_with_cache_position": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Tensor", "_prepare_4d_causal_attention_mask_with_cache_position", "and", "arange", "attention_mask", "batch_size", "cache_position", "causal_mask", "class", "clone", "def", "device", "diagonal", "dim", "dtype", "else", "expand", "fill_value", "finfo", "full", "if", "int", "is", "kwargs", "mask_length", "masked_fill", "min", "min_dtype", "not", "padding_mask", "reshape", "return", "sequence_length", "shape", "target_length", "to", "torch", "triu"], "llava_next_video/modeling_llava_next_video.py:LlavaNextVideoForConditionalGeneration.get_video_features": ["FloatTensor", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "class", "def", "get_video_features", "int", "list", "model", "pixel_values", "return", "self", "str", "torch", "vision_feature_layer", "vision_feature_select_strategy"], "llava_onevision/modeling_llava_onevision.py:LlavaOnevisionPreTrainedModel._init_weights": ["Linear", "ModelModel", "ModelPreTrainedModel", "None", "PreTrainedModel", "_init_weights", "bias", "class", "config", "def", "elif", "embed_std", "get_text_config", "getattr", "hidden_size", "if", "image_newline", "init", "initializer_range", "is", "isinstance", "math", "mean", "module", "nn", "normal_", "not", "self", "sqrt", "std", "text_config", "weight", "zeros_"], "llava_onevision/modeling_llava_onevision.py:LlavaOnevisionMultiModalProjector.__init__": ["ACT2FN", "Linear", "ModelConfig", "ModelMultiModalProjector", "Module", "__init__", "act", "bias", "class", "config", "def", "else", "hidden_size", "if", "int", "isinstance", "len", "linear_1", "linear_2", "multimodal_projector_bias", "nn", "num_feature_layers", "projector_hidden_act", "self", "super", "text_config", "vision_config", "vision_feature_layer"], "llava_onevision/modeling_llava_onevision.py:LlavaOnevisionMultiModalProjector.forward": ["ModelMultiModalProjector", "Module", "act", "class", "def", "forward", "hidden_states", "image_features", "linear_1", "linear_2", "nn", "return", "self"], "llava_onevision/modeling_llava_onevision.py:get_anyres_image_grid_shape": ["Model_anyres_image_grid_shape", "Tensor", "TypeError", "a", "be", "def", "either", "f", "grid_pinpoints", "height", "if", "image_size", "invalid", "isinstance", "list", "lists", "ndarray", "not", "np", "of", "or", "patch_size", "raise", "return", "select_best_resolution", "should", "tensor", "tolist", "torch", "tuple", "tuples", "type", "valid", "width"], "llava_onevision/modeling_llava_onevision.py:image_size_to_num_patches": ["Model_size", "Model_size_to_num_patches", "Tensor", "TypeError", "a", "be", "best_resolution", "def", "f", "for", "grid_pinpoints", "height", "i", "if", "in", "int", "invalid", "isinstance", "j", "list", "lists", "ndarray", "not", "np", "num_patches", "of", "or", "patch_size", "raise", "range", "return", "select_best_resolution", "should", "tolist", "torch", "tuple", "tuples", "type", "value", "width", "with"], "llava_onevision/modeling_llava_onevision.py:unpad_image": ["Model_image", "Modelded_tensor", "Tensor", "TypeError", "be", "current_aspect_ratio", "current_height", "current_width", "def", "either", "else", "f", "if", "image_size", "int", "invalid", "isinstance", "list", "ndarray", "new_height", "new_width", "not", "np", "or", "original_aspect_ratio", "original_height", "original_size", "original_width", "padding", "raise", "return", "round", "scale_factor", "shape", "should", "tensor", "tolist", "torch", "tuple", "type", "valid"], "llava_onevision/modeling_llava_onevision.py:LlavaOnevisionModel.__init__": ["AutoModel", "ModelModel", "ModelMultiModalProjector", "ModelPreTrainedModel", "None", "Parameter", "__init__", "class", "config", "def", "dtype", "else", "embed_std", "from_config", "hidden_size", "if", "image_newline", "is", "language_model", "math", "multi_modal_projector", "nn", "not", "pad_token_id", "post_init", "randn", "self", "sqrt", "super", "text_config", "torch", "vision_config", "vision_tower", "vocab_size"], "llava_onevision/modeling_llava_onevision.py:LlavaOnevisionModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "language_model", "return", "self"], "llava_onevision/modeling_llava_onevision.py:LlavaOnevisionModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "language_model", "self", "set_input_embeddings", "value"], "llava_onevision/modeling_llava_onevision.py:LlavaOnevisionModel.pack_image_features": ["ModelModel", "ModelPreTrainedModel", "None", "The", "ValueError", "anyres_max_", "anyres_max_9", "append", "base_image_feature", "bilinear", "cat", "channels", "class", "config", "consistent", "contiguous", "curr_height", "curr_width", "def", "device", "dim", "dtype", "else", "enumerate", "expand", "feature_lens", "flatten", "for", "functional", "get_anyres_image_grid_shape", "height", "if", "image", "image_feature", "image_features", "image_grid_pinpoints", "image_idx", "image_newline", "image_size", "image_sizes", "in", "int", "interpolate", "is", "long", "math", "max_num_patches", "mode", "new_image_features", "nn", "not", "num_patch_height", "num_patch_width", "number", "of", "pack_image_features", "patch_size", "patches", "permute", "raise", "ratio", "return", "self", "shape", "size", "sqrt", "strip", "tensor", "the", "to", "torch", "transpose", "unpad_image", "view", "vision_aspect_ratio", "vision_config", "width", "with"], "llava_onevision/modeling_llava_onevision.py:LlavaOnevisionModel.get_image_features": ["FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "True", "ValueError", "_", "_pixel_values_list", "batch_num_images", "be", "cat", "class", "config", "def", "default", "dim", "dimensions", "elif", "else", "expect", "f", "feature_lens", "for", "get_image_features", "grid_pinpoints", "hidden_states", "hs_pool", "if", "image_features", "image_grid_pinpoints", "image_newline", "image_num_patches", "image_size", "image_size_to_num_patches", "image_sizes", "imsize", "in", "int", "is", "isinstance", "layer_idx", "len", "list", "multi_modal_projector", "n", "need_patching", "not", "num_patch", "of", "or", "output_hidden_states", "pack_image_features", "patch_size", "pix_val", "pixel_values", "raise", "range", "return", "selected_image_feature", "self", "shape", "should_patch", "split", "str", "to", "torch", "vision_aspect_ratio", "vision_config", "vision_feature_layer", "vision_feature_select_strategy", "vision_tower", "zip"], "llava_onevision/modeling_llava_onevision.py:LlavaOnevisionModel.get_placeholder_mask": ["FloatTensor", "Image", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "ValueError", "Videos", "all", "and", "class", "config", "def", "device", "do", "dtype", "else", "expand_as", "f", "features", "get_input_embeddings", "get_placeholder_mask", "if", "image", "image_features", "image_token_id", "input_ids", "inputs_embeds", "is", "long", "match", "n_image_tokens", "n_video_tokens", "not", "numel", "raise", "return", "self", "shape", "special_image_mask", "special_video_mask", "sum", "tensor", "to", "tokens", "torch", "unsqueeze", "video_features", "video_token_id"], "llava_onevision/modeling_llava_onevision.py:LlavaOnevisionModel.forward": ["Cache", "FlashAttentionKwargs", "FloatTensor", "LongTensor", "ModelModel", "ModelModelOutputWithPast", "ModelPreTrainedModel", "None", "Tensor", "True", "Unpack", "ValueError", "You", "_", "attention_mask", "attentions", "batch_num_images", "bool", "cache_position", "cat", "class", "config", "def", "device", "dim", "dtype", "else", "exactly", "flatten", "forward", "get_image_features", "get_input_embeddings", "get_placeholder_mask", "get_video_features", "hidden_states", "if", "image_features", "image_hidden_states", "image_newline", "image_sizes", "image_sizes_videos", "input_ids", "inputs_embeds", "int", "is", "kwargs", "language_model", "last_hidden_state", "list", "masked_scatter", "must", "not", "of", "one", "or", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "pixel_values", "pixel_values_videos", "position_ids", "r", "raise", "repeat", "return", "return_dict", "self", "shape", "special_image_mask", "special_video_mask", "specify", "str", "to", "torch", "tuple", "use_cache", "use_return_dict", "video_features", "video_hidden_states", "vision_aspect_ratio", "vision_feature_layer", "vision_feature_select_strategy"], "llava_onevision/modeling_llava_onevision.py:LlavaOnevisionModel.get_video_features": ["FloatTensor", "ModelModel", "ModelPreTrainedModel", "True", "apply_pooling", "batch_size", "cat", "channels", "class", "def", "default", "dim", "else", "for", "frames", "get_video_features", "height", "hidden_states", "hs_pool", "if", "in", "int", "isinstance", "layer_idx", "list", "multi_modal_projector", "output_hidden_states", "pixel_values", "reshape", "return", "selected_video_feature", "self", "shape", "str", "torch", "video_features", "view", "vision_feature_layer", "vision_feature_select_strategy", "vision_tower", "width"], "llava_onevision/modeling_llava_onevision.py:LlavaOnevisionModel.apply_pooling": ["ModelModel", "ModelPreTrainedModel", "apply_pooling", "batch_frames", "bilinear", "ceil", "class", "config", "contiguous", "def", "dim", "functional", "height", "image_features", "image_size", "interpolate", "math", "mode", "nn", "patch_size", "permute", "return", "scaled_shape", "self", "seq_len", "shape", "size", "view", "vision_config", "width"], "llava_onevision/modeling_llava_onevision.py:LlavaOnevisionForConditionalGeneration.__init__": ["False", "GenerationMixin", "Linear", "ModelConfig", "ModelForConditionalGeneration", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "text_config", "vocab_size"], "llava_onevision/modeling_llava_onevision.py:LlavaOnevisionForConditionalGeneration.get_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "model", "return", "self"], "llava_onevision/modeling_llava_onevision.py:LlavaOnevisionForConditionalGeneration.set_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "model", "self", "set_input_embeddings", "value"], "llava_onevision/modeling_llava_onevision.py:LlavaOnevisionForConditionalGeneration.get_output_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "Module", "class", "def", "get_output_embeddings", "lm_head", "nn", "return", "self"], "llava_onevision/modeling_llava_onevision.py:LlavaOnevisionForConditionalGeneration.pack_image_features": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "class", "def", "image_features", "image_newline", "image_sizes", "model", "pack_image_features", "return", "self", "vision_feature_select_strategy"], "llava_onevision/modeling_llava_onevision.py:LlavaOnevisionForConditionalGeneration.get_image_features": ["FloatTensor", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Tensor", "class", "def", "get_image_features", "image_sizes", "int", "list", "model", "pixel_values", "return", "self", "str", "torch", "vision_feature_layer", "vision_feature_select_strategy"], "llava_onevision/modeling_llava_onevision.py:LlavaOnevisionForConditionalGeneration.forward": ["Cache", "FloatTensor", "GenerationMixin", "LongTensor", "ModelCausalLMOutputWithPast", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "attention_mask", "attentions", "batch_num_images", "bool", "cache_position", "class", "config", "def", "else", "forward", "hidden_states", "if", "image_hidden_states", "image_sizes", "image_sizes_videos", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "list", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "pixel_values", "pixel_values_videos", "position_ids", "r", "return", "return_dict", "self", "slice", "slice_indices", "str", "text_config", "torch", "tuple", "use_cache", "use_return_dict", "video_hidden_states", "vision_aspect_ratio", "vision_feature_layer", "vision_feature_select_strategy", "vocab_size"], "llava_onevision/modeling_llava_onevision.py:LlavaOnevisionForConditionalGeneration.prepare_inputs_for_generation": ["False", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "True", "attention_mask", "cache_position", "class", "def", "get", "if", "image_sizes", "image_sizes_videos", "input_ids", "inputs_embeds", "is_first_iteration", "kwargs", "logits_to_keep", "model_inputs", "not", "or", "past_key_values", "pixel_values", "pixel_values_videos", "prepare_inputs_for_generation", "return", "self", "super", "use_cache"], "llava_onevision/modeling_llava_onevision.py:LlavaOnevisionForConditionalGeneration._prepare_4d_causal_attention_mask_with_cache_position": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Tensor", "_prepare_4d_causal_attention_mask_with_cache_position", "and", "arange", "attention_mask", "batch_size", "cache_position", "causal_mask", "class", "clone", "def", "device", "diagonal", "dim", "dtype", "else", "expand", "fill_value", "finfo", "full", "if", "int", "is", "kwargs", "mask_length", "masked_fill", "min", "min_dtype", "not", "padding_mask", "reshape", "return", "sequence_length", "shape", "target_length", "to", "torch", "triu"], "llava_onevision/modeling_llava_onevision.py:LlavaOnevisionForConditionalGeneration.get_video_features": ["FloatTensor", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "class", "def", "get_video_features", "int", "list", "model", "pixel_values", "return", "self", "str", "torch", "vision_feature_layer", "vision_feature_select_strategy"], "longcat_flash/modeling_longcat_flash.py:LongcatFlashRMSNorm.__init__": ["ModelRMSNorm", "Module", "Parameter", "__init__", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "longcat_flash/modeling_longcat_flash.py:LongcatFlashRMSNorm.forward": ["ModelRMSNorm", "Module", "True", "class", "def", "dtype", "float32", "forward", "hidden_states", "input_dtype", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "longcat_flash/modeling_longcat_flash.py:LongcatFlashRMSNorm.extra_repr": ["ModelRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "variance_epsilon", "weight"], "longcat_flash/modeling_longcat_flash.py:LongcatFlashRotaryEmbedding.__init__": ["Callable", "False", "ModelConfig", "ModelRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "longcat_flash/modeling_longcat_flash.py:LongcatFlashRotaryEmbedding.compute_default_rope_parameters": ["ModelConfig", "ModelRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "longcat_flash/modeling_longcat_flash.py:LongcatFlashRotaryEmbedding.forward": ["False", "ModelRotaryEmbedding", "Module", "None", "and", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "longcat_flash/modeling_longcat_flash.py:LongcatFlashMLP.__init__": ["ACT2FN", "False", "Linear", "ModelMLP", "Module", "None", "__init__", "act_fn", "bias", "class", "config", "def", "down_proj", "else", "ffn_hidden_size", "gate_proj", "hidden_act", "hidden_size", "if", "intermediate_size", "is", "nn", "self", "super", "up_proj"], "longcat_flash/modeling_longcat_flash.py:LongcatFlashMLP.forward": ["ModelMLP", "Module", "act_fn", "class", "def", "down_proj", "forward", "gate_proj", "nn", "return", "self", "up_proj", "x"], "longcat_flash/modeling_longcat_flash.py:LongcatFlashTopkRouter.__init__": ["False", "Linear", "ModelTopkRouter", "Module", "__init__", "bias", "class", "classifier", "config", "def", "e_score_correction_bias", "getattr", "hidden_size", "moe_topk", "n_routed_experts", "nn", "or", "register_buffer", "routed_scaling_factor", "router_bias", "self", "super", "top_k", "torch", "zero_expert_num", "zeros"], "longcat_flash/modeling_longcat_flash.py:LongcatFlashTopkRouter.forward": ["F", "ModelTopkRouter", "Module", "class", "classifier", "config", "def", "dim", "dtype", "float32", "forward", "gather", "get_topk_indices", "hidden_size", "hidden_states", "linear", "nn", "return", "routed_scaling_factor", "router_logits", "scores", "self", "softmax", "to", "topk_indices", "topk_weights", "torch", "type", "view", "weight"], "longcat_flash/modeling_longcat_flash.py:LongcatFlashTopkRouter.get_topk_indices": ["False", "ModelTopkRouter", "Module", "class", "def", "dim", "e_score_correction_bias", "get_topk_indices", "k", "n_routed_experts", "nn", "return", "scores", "scores_for_choice", "self", "sorted", "top_k", "topk", "topk_indices", "torch", "unsqueeze", "view"], "longcat_flash/modeling_longcat_flash.py:LongcatFlashExperts.__init__": ["ACT2FN", "ModelExperts", "Module", "None", "Parameter", "__init__", "act_fn", "class", "config", "def", "down_proj", "else", "empty", "expert_ffn_hidden_size", "gate_up_proj", "hidden_act", "hidden_size", "if", "intermediate_size", "n_routed_experts", "nn", "num_routed_experts", "or", "register_parameter", "self", "super", "torch", "total_experts", "zero_expert_num"], "longcat_flash/modeling_longcat_flash.py:LongcatFlashExperts.forward": ["F", "False", "ModelExperts", "Module", "None", "act_fn", "as_tuple", "chunk", "class", "continue", "current_hidden_states", "current_state", "def", "dim", "down_proj", "dtype", "else", "expert_hit", "expert_idx", "expert_idx_tensor", "expert_mask", "final_hidden_states", "for", "forward", "functional", "gate", "gate_up_proj", "greater", "hidden_states", "if", "in", "index_add_", "int", "is", "item", "linear", "nn", "nonzero", "num_classes", "num_routed_experts", "numel", "one_hot", "or", "permute", "return", "selection_idx", "self", "squeeze", "sum", "to", "token_idx", "top_k_index", "top_k_weights", "torch", "total_experts", "up", "where", "zeros_like"], "longcat_flash/modeling_longcat_flash.py:LongcatFlashMoE.__init__": ["ModelExperts", "ModelMoE", "ModelTopkRouter", "Module", "__init__", "class", "config", "def", "expert_ffn_hidden_size", "experts", "intermediate_size", "nn", "router", "self", "super"], "longcat_flash/modeling_longcat_flash.py:LongcatFlashMoE.forward": ["ModelMoE", "Module", "class", "def", "experts", "forward", "hidden_states", "nn", "orig_shape", "return", "router", "self", "shape", "topk_indices", "topk_weights", "view"], "longcat_flash/modeling_longcat_flash.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "longcat_flash/modeling_longcat_flash.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "longcat_flash/modeling_longcat_flash.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value", "value_states"], "longcat_flash/modeling_longcat_flash.py:apply_rotary_pos_emb_interleave": ["Model_rotary_pos_emb_interleave", "None", "b", "cos", "d", "def", "h", "k", "k_embed", "position_ids", "q", "q_embed", "r", "reshape", "return", "rotate_half", "s", "shape", "sin", "transpose", "unsqueeze", "unsqueeze_dim", "view"], "longcat_flash/modeling_longcat_flash.py:yarn_get_mscale": ["Model_get_mscale", "def", "if", "log", "math", "mscale", "return", "scale"], "longcat_flash/modeling_longcat_flash.py:LongcatFlashMLA.__init__": ["False", "Linear", "ModelMLA", "ModelRMSNorm", "Module", "None", "True", "__init__", "attention_bias", "attention_dropout", "bias", "class", "config", "def", "default", "else", "factor", "get", "hidden_size", "if", "int", "is", "is_causal", "kv_a_layernorm", "kv_a_proj_with_mqa", "kv_b_proj", "kv_lora_rank", "layer_idx", "mla_scale_kv_lora", "mla_scale_q_lora", "mscale", "mscale_all_dim", "nn", "num_attention_heads", "num_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "q_a_layernorm", "q_a_proj", "q_b_proj", "q_lora_rank", "q_proj", "qk_head_dim", "qk_nope_head_dim", "qk_rope_head_dim", "rope_parameters", "rope_type", "scaling", "scaling_factor", "self", "super", "v_head_dim", "yarn_get_mscale"], "longcat_flash/modeling_longcat_flash.py:LongcatFlashMLA.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "F", "FlashAttentionKwargs", "LongTensor", "ModelMLA", "Module", "None", "Tensor", "Unpack", "_attn_implementation", "and", "apply_rotary_pos_emb_interleave", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "batch_size", "cache_kwargs", "cache_position", "cat", "class", "compressed_kv", "config", "contiguous", "cos", "def", "dim", "dropout", "eager", "eager_attention_forward", "else", "expand", "flash", "forward", "hidden_states", "if", "in", "is", "k_pass", "k_rot", "key_shape", "key_states", "kv_a_layernorm", "kv_a_proj_with_mqa", "kv_b_proj", "kv_lora_rank", "kwargs", "layer_idx", "mla_scale_kv_lora", "mla_scale_q_lora", "nn", "not", "o_proj", "pad", "past_key_values", "position_embeddings", "q_a_layernorm", "q_a_proj", "q_b_proj", "q_pass", "q_rot", "q_states", "qk_head_dim", "qk_nope_head_dim", "qk_rope_head_dim", "query_shape", "query_states", "reshape", "return", "scaling", "self", "seq_length", "shape", "sin", "split", "torch", "training", "transpose", "tuple", "update", "v_head_dim", "value_states", "view"], "longcat_flash/modeling_longcat_flash.py:LongcatFlashDecoderLayer.__init__": ["GradientCheckpointingLayer", "ModelDecoderLayer", "ModelMLA", "ModelMLP", "ModelMoE", "ModelRMSNorm", "ModuleList", "_", "__init__", "class", "config", "def", "eps", "for", "hidden_size", "i", "in", "input_layernorm", "int", "layer_idx", "mlp", "mlps", "nn", "post_attention_layernorm", "rms_norm_eps", "self", "self_attn", "super"], "longcat_flash/modeling_longcat_flash.py:LongcatFlashDecoderLayer.forward": ["Cache", "False", "FlashAttentionKwargs", "GradientCheckpointingLayer", "LongTensor", "ModelDecoderLayer", "None", "Tensor", "Unpack", "_", "attention_mask", "bool", "cache_position", "class", "def", "forward", "hidden_states", "input_layernorm", "kwargs", "mlp", "mlps", "past_key_values", "position_embeddings", "position_ids", "post_attention_layernorm", "residual", "return", "self", "self_attn", "shortcut_mlp_output", "torch", "tuple", "use_cache"], "longcat_flash/modeling_longcat_flash.py:LongcatFlashPreTrainedModel._init_weights": ["ModelExperts", "ModelPreTrainedModel", "ModelTopkRouter", "None", "PreTrainedModel", "_init_weights", "class", "classifier", "config", "def", "down_proj", "e_score_correction_bias", "gate_up_proj", "if", "init", "initializer_range", "is", "isinstance", "mean", "module", "normal_", "not", "self", "std", "super", "weight", "zeros_"], "longcat_flash/modeling_longcat_flash.py:LongcatFlashModel.__init__": ["Embedding", "False", "ModelDecoderLayer", "ModelModel", "ModelPreTrainedModel", "ModelRMSNorm", "ModelRotaryEmbedding", "ModuleList", "__init__", "class", "config", "def", "embed_tokens", "eps", "for", "gradient_checkpointing", "head_dim", "hidden_size", "in", "layer_idx", "layers", "nn", "norm", "num_hidden_layers", "num_layers", "pad_token_id", "padding_idx", "post_init", "range", "rms_norm_eps", "rotary_emb", "self", "super", "vocab_size"], "longcat_flash/modeling_longcat_flash.py:LongcatFlashModel.forward": ["BaseModelOutputWithPast", "Cache", "DynamicCache", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "and", "arange", "attention_mask", "attentions", "bool", "cache_position", "causal_mask", "class", "config", "create_causal_mask", "decoder_layer", "def", "device", "else", "embed_tokens", "exactly", "for", "forward", "get_seq_length", "hidden_states", "if", "in", "input_embeds", "input_ids", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layers", "must", "norm", "not", "num_layers", "of", "one", "or", "past_key_values", "past_seen_tokens", "position_embeddings", "position_ids", "raise", "return", "rotary_emb", "self", "shape", "specify", "torch", "unsqueeze", "use_cache"], "longcat_flash/modeling_longcat_flash.py:LongcatFlashForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "ModelForCausalLM", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "vocab_size"], "longcat_flash/modeling_longcat_flash.py:LongcatFlashForCausalLM.forward": ["BaseModelOutputWithPast", "Cache", "CausalLMOutputWithPast", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForCausalLM", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "outputs", "past_key_values", "position_ids", "r", "return", "self", "slice", "slice_indices", "torch", "use_cache", "vocab_size"], "longformer/modeling_longformer.py:_get_question_end_index": ["There", "You", "_get_question_end_index", "also", "answering", "assert", "avoid", "batch_size", "be", "consider", "def", "dimensions", "error", "every", "exactly", "f", "for", "forward", "function", "global_attention_mask", "have", "in", "input_ids", "manually", "might", "nonzero", "questions", "return", "sample", "sep_token_id", "sep_token_indices", "separator", "set", "shape", "should", "the", "this", "three", "to", "tokens", "two", "view"], "longformer/modeling_longformer.py:_compute_global_attention_mask": ["True", "_compute_global_attention_mask", "_get_question_end_index", "arange", "attention_mask", "before_sep_token", "bool", "def", "device", "dim", "else", "expand_as", "if", "input_ids", "is", "question_end_index", "return", "sep_token_id", "shape", "to", "torch", "unsqueeze"], "longformer/modeling_longformer.py:create_position_ids_from_input_ids": ["Model_position_ids_from_input_ids", "cumsum", "def", "dim", "incremental_indices", "input_ids", "int", "long", "mask", "ne", "padding_idx", "return", "torch", "type_as"], "longformer/modeling_longformer.py:LongformerEmbeddings.__init__": ["Dropout", "Embedding", "LayerNorm", "ModelEmbeddings", "Module", "__init__", "class", "config", "def", "dropout", "eps", "hidden_dropout_prob", "hidden_size", "layer_norm_eps", "max_position_embeddings", "nn", "pad_token_id", "padding_idx", "position_embeddings", "self", "super", "token_type_embeddings", "type_vocab_size", "vocab_size", "word_embeddings"], "longformer/modeling_longformer.py:LongformerEmbeddings.forward": ["LayerNorm", "ModelEmbeddings", "Module", "None", "class", "create_position_ids_from_input_ids", "create_position_ids_from_inputs_embeds", "def", "device", "dropout", "dtype", "else", "embeddings", "forward", "if", "input_ids", "input_shape", "inputs_embeds", "is", "long", "nn", "not", "padding_idx", "position_embeddings", "position_ids", "return", "self", "size", "to", "token_type_embeddings", "token_type_ids", "torch", "word_embeddings", "zeros"], "longformer/modeling_longformer.py:LongformerEmbeddings.create_position_ids_from_inputs_embeds": ["ModelEmbeddings", "Module", "arange", "class", "create_position_ids_from_inputs_embeds", "def", "device", "dtype", "expand", "input_shape", "inputs_embeds", "long", "nn", "padding_idx", "position_ids", "return", "self", "sequence_length", "size", "torch", "unsqueeze"], "longformer/modeling_longformer.py:LongformerSelfAttention.__init__": ["Given", "Linear", "ModelSelfAttention", "Module", "The", "ValueError", "__init__", "a", "an", "assert", "attention", "attention_probs_dropout_prob", "attention_window", "be", "class", "config", "def", "dropout", "embed_dim", "even", "f", "for", "has", "head_dim", "heads", "hidden", "hidden_size", "if", "int", "is", "key", "key_global", "layer", "layer_id", "multiple", "nn", "not", "num_attention_heads", "num_heads", "number", "of", "one_sided_attn_window_size", "positive", "query", "query_global", "raise", "self", "size", "super", "the", "to", "value", "value_global"], "longformer/modeling_longformer.py:LongformerSelfAttention.forward": ["False", "ModelSelfAttention", "Module", "None", "Unexpected", "_compute_attn_output_with_global_indices", "_compute_global_attn_output_from_hidden", "_concat_with_global_key_attn_probs", "_get_global_attn_indices", "_sliding_chunks_matmul_attn_probs_value", "_sliding_chunks_query_key_matmul", "and", "assert", "attention_mask", "attn_output", "attn_probs", "attn_scores", "batch_size", "be", "but", "cat", "class", "contiguous", "def", "del", "diagonal_mask", "dim", "dropout", "dtype", "else", "embed_dim", "f", "finfo", "float32", "float_mask", "forward", "functional", "global_attn_output", "global_attn_probs", "global_key_attn_scores", "has", "have", "head_dim", "hidden_states", "if", "is", "is_global_attn", "is_index_global_attn", "is_index_global_attn_nonzero", "is_index_masked", "is_local_index_global_attn_nonzero", "is_local_index_no_global_attn_nonzero", "key", "key_vectors", "len", "list", "local_attn_probs", "masked_fill", "math", "max_num_global_attn_indices", "min", "new_ones", "nn", "nonzero_global_attn_output", "num_heads", "of", "one_sided_attn_window_size", "output_attentions", "outputs", "p", "query", "query_vectors", "remove_from_windowed_attention_mask", "reshape", "return", "self", "seq_len", "should", "size", "softmax", "sqrt", "torch", "training", "transpose", "type_as", "value", "value_vectors", "view"], "longformer/modeling_longformer.py:LongformerSelfAttention._pad_and_transpose_last_two_dims": ["ModelSelfAttention", "Module", "_pad_and_transpose_last_two_dims", "class", "def", "functional", "hidden_states_padded", "nn", "pad", "padding", "return", "size", "view"], "longformer/modeling_longformer.py:LongformerSelfAttention._pad_and_diagonalize": ["ModelSelfAttention", "Module", "_pad_and_diagonalize", "chunked_hidden_states", "class", "def", "functional", "hidden_dim", "nn", "num_chunks", "pad", "return", "size", "total_num_heads", "view", "window_overlap"], "longformer/modeling_longformer.py:LongformerSelfAttention._chunk": ["False", "ModelSelfAttention", "Module", "_chunk", "as_strided", "bool", "chunk", "chunk_size", "chunk_stride", "class", "def", "device", "div", "empty", "for", "hidden_states", "if", "in", "list", "nn", "not", "onnx_export", "overlapping_chunks", "range", "return", "rounding_mode", "size", "stride", "torch", "trunc", "view", "window_overlap"], "longformer/modeling_longformer.py:LongformerSelfAttention._mask_invalid_locations": ["ModelSelfAttention", "Module", "None", "Tensor", "_mask_invalid_locations", "affected_seq_len", "beginning_input", "beginning_mask", "beginning_mask_2d", "bool", "class", "def", "dims", "ending_input", "ending_mask", "expand", "flip", "float", "full_like", "inf", "input_tensor", "new_ones", "nn", "size", "torch", "tril", "where"], "longformer/modeling_longformer.py:LongformerSelfAttention._sliding_chunks_query_key_matmul": ["False", "Given", "ModelSelfAttention", "Module", "Sequence", "Tensor", "_chunk", "_mask_invalid_locations", "_pad_and_transpose_last_two_dims", "_sliding_chunks_query_key_matmul", "assert", "batch_size", "bcxd", "bcxy", "bcyd", "be", "chunks_count", "class", "config", "def", "diagonal_attention_scores", "diagonal_chunked_attention_scores", "div", "einsum", "f", "getattr", "head_dim", "int", "key", "length", "multiple", "new_zeros", "nn", "num_heads", "of", "onnx_export", "padding", "query", "reshape", "return", "rounding_mode", "self", "seq_len", "should", "size", "torch", "transpose", "trunc", "view", "window_overlap"], "longformer/modeling_longformer.py:LongformerSelfAttention._sliding_chunks_matmul_attn_probs_value": ["ModelSelfAttention", "Module", "Tensor", "_pad_and_diagonalize", "_sliding_chunks_matmul_attn_probs_value", "as_strided", "assert", "attn_probs", "batch_size", "bcdh", "bcwd", "bcwh", "chunked_attn_probs", "chunked_value", "chunked_value_size", "chunked_value_stride", "chunks_count", "class", "context", "def", "div", "einsum", "functional", "head_dim", "int", "nn", "num_heads", "pad", "padded_value", "reshape", "return", "rounding_mode", "self", "seq_len", "size", "stride", "torch", "transpose", "trunc", "value", "view", "window_overlap"], "longformer/modeling_longformer.py:LongformerSelfAttention._get_global_attn_indices": ["ModelSelfAttention", "Module", "True", "_get_global_attn_indices", "arange", "as_tuple", "class", "def", "device", "dim", "is_index_global_attn", "is_index_global_attn_nonzero", "is_local_index_global_attn", "is_local_index_global_attn_nonzero", "is_local_index_no_global_attn_nonzero", "long", "max", "max_num_global_attn_indices", "nn", "nonzero", "num_global_attn_indices", "return", "sum", "torch", "unsqueeze"], "longformer/modeling_longformer.py:LongformerSelfAttention._concat_with_global_key_attn_probs": ["ModelSelfAttention", "Module", "_concat_with_global_key_attn_probs", "attn_probs_from_global_key", "batch_size", "blhd", "blhs", "bshd", "class", "def", "dtype", "einsum", "finfo", "head_dim", "is_index_global_attn_nonzero", "is_local_index_global_attn_nonzero", "is_local_index_no_global_attn_nonzero", "key_vectors", "key_vectors_only_global", "max_num_global_attn_indices", "min", "new_zeros", "nn", "num_heads", "query_vectors", "return", "self", "shape", "torch", "transpose"], "longformer/modeling_longformer.py:LongformerSelfAttention._compute_attn_output_with_global_indices": ["ModelSelfAttention", "Module", "_compute_attn_output_with_global_indices", "_sliding_chunks_matmul_attn_probs_value", "attn_output_only_global", "attn_output_without_global", "attn_probs", "attn_probs_only_global", "attn_probs_without_global", "batch_size", "class", "clone", "contiguous", "def", "head_dim", "is_index_global_attn_nonzero", "is_local_index_global_attn_nonzero", "matmul", "max_num_global_attn_indices", "narrow", "new_zeros", "nn", "num_heads", "one_sided_attn_window_size", "return", "self", "shape", "size", "torch", "transpose", "value_vectors", "value_vectors_only_global"], "longformer/modeling_longformer.py:LongformerSelfAttention._compute_global_attn_output_from_hidden": ["ModelSelfAttention", "Module", "None", "Size", "_compute_global_attn_output_from_hidden", "assert", "batch_size", "be", "bmm", "but", "class", "contiguous", "def", "dim", "dropout", "dtype", "embed_dim", "f", "finfo", "float32", "functional", "global_attn_hidden_states", "global_attn_output", "global_attn_probs", "global_attn_probs_float", "global_attn_scores", "global_key_vectors", "global_query_vectors_only_global", "global_value_vectors", "has", "have", "head_dim", "hidden_states", "is", "is_index_global_attn_nonzero", "is_index_masked", "is_local_index_global_attn_nonzero", "is_local_index_no_global_attn_nonzero", "key_global", "list", "masked_fill", "math", "max_num_global_attn_indices", "min", "new_zeros", "nn", "num_heads", "p", "query_global", "return", "self", "seq_len", "shape", "should", "size", "softmax", "sqrt", "tensor", "the", "torch", "training", "transpose", "type_as", "value_global", "view", "wrong"], "longformer/modeling_longformer.py:LongformerSelfOutput.__init__": ["Dropout", "LayerNorm", "Linear", "ModelSelfOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "eps", "hidden_dropout_prob", "hidden_size", "layer_norm_eps", "nn", "self", "super"], "longformer/modeling_longformer.py:LongformerSelfOutput.forward": ["LayerNorm", "ModelSelfOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "longformer/modeling_longformer.py:LongformerAttention.__init__": ["ModelAttention", "ModelSelfAttention", "ModelSelfOutput", "Module", "__init__", "class", "config", "def", "layer_id", "nn", "output", "self", "super"], "longformer/modeling_longformer.py:LongformerAttention.forward": ["False", "ModelAttention", "Module", "None", "attention_mask", "attn_output", "class", "def", "forward", "hidden_states", "is_global_attn", "is_index_global_attn", "is_index_masked", "nn", "output", "output_attentions", "outputs", "return", "self", "self_outputs"], "longformer/modeling_longformer.py:LongformerIntermediate.__init__": ["ACT2FN", "Linear", "ModelIntermediate", "Module", "__init__", "class", "config", "def", "dense", "else", "hidden_act", "hidden_size", "if", "intermediate_act_fn", "intermediate_size", "isinstance", "nn", "self", "str", "super"], "longformer/modeling_longformer.py:LongformerIntermediate.forward": ["ModelIntermediate", "Module", "Tensor", "class", "def", "dense", "forward", "hidden_states", "intermediate_act_fn", "nn", "return", "self", "torch"], "longformer/modeling_longformer.py:LongformerOutput.__init__": ["Dropout", "LayerNorm", "Linear", "ModelOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "eps", "hidden_dropout_prob", "hidden_size", "intermediate_size", "layer_norm_eps", "nn", "self", "super"], "longformer/modeling_longformer.py:LongformerOutput.forward": ["LayerNorm", "ModelOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "longformer/modeling_longformer.py:LongformerLayer.__init__": ["GradientCheckpointingLayer", "ModelAttention", "ModelIntermediate", "ModelLayer", "ModelOutput", "__init__", "attention", "chunk_size_feed_forward", "class", "config", "def", "intermediate", "layer_id", "output", "self", "seq_len_dim", "super"], "longformer/modeling_longformer.py:LongformerLayer.forward": ["False", "GradientCheckpointingLayer", "ModelLayer", "None", "apply_chunking_to_forward", "attention", "attention_mask", "attn_output", "chunk_size_feed_forward", "class", "def", "ff_chunk", "forward", "hidden_states", "is_global_attn", "is_index_global_attn", "is_index_masked", "layer_output", "output_attentions", "outputs", "return", "self", "self_attn_outputs", "seq_len_dim"], "longformer/modeling_longformer.py:LongformerLayer.ff_chunk": ["GradientCheckpointingLayer", "ModelLayer", "attn_output", "class", "def", "ff_chunk", "intermediate", "intermediate_output", "layer_output", "output", "return", "self"], "longformer/modeling_longformer.py:LongformerEncoder.__init__": ["False", "ModelEncoder", "ModelLayer", "Module", "ModuleList", "__init__", "class", "config", "def", "for", "gradient_checkpointing", "i", "in", "layer", "layer_id", "nn", "num_hidden_layers", "range", "self", "super"], "longformer/modeling_longformer.py:LongformerEncoder.forward": ["False", "ModelBaseModelOutput", "ModelEncoder", "Module", "None", "True", "all_attentions", "all_global_attentions", "all_hidden_states", "and", "any", "attention_mask", "attentions", "class", "def", "else", "enumerate", "flatten", "for", "forward", "global_attentions", "hidden_states", "idx", "if", "in", "is", "is_global_attn", "is_index_global_attn", "is_index_masked", "item", "last_hidden_state", "layer", "layer_module", "layer_outputs", "nn", "not", "output_attentions", "output_hidden_states", "padding_len", "return", "return_dict", "self", "shape", "state", "transpose", "tuple", "v"], "longformer/modeling_longformer.py:LongformerPooler.__init__": ["Linear", "ModelPooler", "Module", "Tanh", "__init__", "activation", "class", "config", "def", "dense", "hidden_size", "nn", "self", "super"], "longformer/modeling_longformer.py:LongformerPooler.forward": ["ModelPooler", "Module", "Tensor", "activation", "class", "def", "dense", "first_token_tensor", "forward", "hidden_states", "nn", "pooled_output", "return", "self", "torch"], "longformer/modeling_longformer.py:LongformerLMHead.__init__": ["LayerNorm", "Linear", "ModelLMHead", "Module", "Parameter", "__init__", "bias", "class", "config", "decoder", "def", "dense", "eps", "hidden_size", "layer_norm", "layer_norm_eps", "nn", "self", "super", "torch", "vocab_size", "zeros"], "longformer/modeling_longformer.py:LongformerLMHead.forward": ["ModelLMHead", "Module", "class", "decoder", "def", "dense", "features", "forward", "gelu", "kwargs", "layer_norm", "nn", "return", "self", "x"], "longformer/modeling_longformer.py:LongformerModel.__init__": ["Expected", "ModelEmbeddings", "ModelEncoder", "ModelModel", "ModelPooler", "ModelPreTrainedModel", "None", "True", "__init__", "add_pooling_layer", "an", "assert", "attention_window", "be", "class", "config", "def", "else", "embeddings", "encoder", "equal", "even", "f", "given", "has", "if", "int", "isinstance", "len", "num_hidden_layers", "pooler", "positive", "post_init", "r", "self", "should", "super", "to", "value"], "longformer/modeling_longformer.py:LongformerModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "return", "self", "word_embeddings"], "longformer/modeling_longformer.py:LongformerModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "self", "set_input_embeddings", "value", "word_embeddings"], "longformer/modeling_longformer.py:LongformerModel._pad_to_window_size": ["Given", "Input", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "_pad_to_window_size", "a", "an", "are", "assert", "attention_mask", "attention_window", "automatically", "batch_size", "be", "cat", "class", "config", "def", "dim", "dtype", "else", "embeddings", "even", "f", "functional", "ids", "if", "input_ids", "input_ids_padding", "input_shape", "inputs_embeds", "inputs_embeds_padding", "int", "is", "isinstance", "logger", "long", "max", "multiple", "new_full", "nn", "not", "of", "pad", "pad_token_id", "padded", "padding_len", "position_ids", "return", "self", "seq_len", "shape", "should", "to", "token_type_ids", "torch", "value", "warning_once"], "longformer/modeling_longformer.py:LongformerModel._merge_to_attention_mask": ["ModelModel", "ModelPreTrainedModel", "None", "Tensor", "_merge_to_attention_mask", "attention_mask", "class", "def", "else", "global_attention_mask", "if", "is", "not", "return", "self", "torch"], "longformer/modeling_longformer.py:LongformerModel.forward": ["ModelBaseModelOutputWithPooling", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "ValueError", "You", "_merge_to_attention_mask", "_pad_to_window_size", "and", "at", "attention_mask", "attentions", "bool", "both", "cannot", "class", "config", "def", "device", "dtype", "either", "elif", "else", "embedding_output", "embeddings", "encoder", "encoder_outputs", "extended_attention_mask", "forward", "get_extended_attention_mask", "global_attention_mask", "global_attentions", "have", "hidden_states", "if", "input_ids", "input_shape", "inputs_embeds", "is", "kwargs", "last_hidden_state", "long", "not", "ones", "or", "output_attentions", "output_hidden_states", "pad_token_id", "padding_len", "pooled_output", "pooler", "pooler_output", "position_ids", "r", "raise", "return", "return_dict", "same", "self", "sequence_output", "size", "specify", "the", "time", "to", "token_type_ids", "torch", "tuple", "use_return_dict", "warn_if_padding_and_no_attention_mask", "zeros"], "longformer/modeling_longformer.py:LongformerForMaskedLM.__init__": ["False", "Model", "ModelForMaskedLM", "ModelLMHead", "ModelModel", "ModelPreTrainedModel", "__init__", "add_pooling_layer", "class", "config", "def", "lm_head", "post_init", "self", "super"], "longformer/modeling_longformer.py:LongformerForMaskedLM.get_output_embeddings": ["ModelForMaskedLM", "ModelPreTrainedModel", "class", "decoder", "def", "get_output_embeddings", "lm_head", "return", "self"], "longformer/modeling_longformer.py:LongformerForMaskedLM.set_output_embeddings": ["ModelForMaskedLM", "ModelPreTrainedModel", "class", "decoder", "def", "lm_head", "new_embeddings", "self", "set_output_embeddings"], "longformer/modeling_longformer.py:LongformerForMaskedLM.forward": ["CrossEntropyLoss", "Model", "ModelForMaskedLM", "ModelMaskedLMOutput", "ModelPreTrainedModel", "None", "Tensor", "attention_mask", "attentions", "bool", "class", "config", "def", "device", "else", "forward", "global_attention_mask", "global_attentions", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "lm_head", "logits", "loss", "loss_fct", "masked_lm_loss", "not", "output", "output_attentions", "output_hidden_states", "outputs", "position_ids", "prediction_scores", "r", "return", "return_dict", "self", "sequence_output", "to", "token_type_ids", "torch", "tuple", "use_return_dict", "view", "vocab_size"], "longformer/modeling_longformer.py:LongformerForSequenceClassification.__init__": ["False", "Model", "ModelClassificationHead", "ModelForSequenceClassification", "ModelModel", "ModelPreTrainedModel", "__init__", "add_pooling_layer", "class", "classifier", "config", "def", "num_labels", "post_init", "self", "super"], "longformer/modeling_longformer.py:LongformerForSequenceClassification.forward": ["BCEWithLogitsLoss", "CLS", "CrossEntropyLoss", "Initializing", "MSELoss", "Model", "ModelForSequenceClassification", "ModelPreTrainedModel", "ModelSequenceClassifierOutput", "None", "Tensor", "and", "attention", "attention_mask", "attentions", "bool", "class", "classifier", "config", "def", "device", "dtype", "elif", "else", "forward", "global", "global_attention_mask", "global_attentions", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "kwargs", "labels", "logger", "logits", "long", "loss", "loss_fct", "multi_label_classification", "not", "num_labels", "on", "or", "output", "output_attentions", "output_hidden_states", "outputs", "position_ids", "problem_type", "r", "regression", "return", "return_dict", "self", "sequence_output", "single_label_classification", "squeeze", "to", "token", "token_type_ids", "torch", "tuple", "use_return_dict", "view", "warning_once", "zeros_like"], "longformer/modeling_longformer.py:LongformerClassificationHead.__init__": ["Dropout", "Linear", "ModelClassificationHead", "Module", "__init__", "class", "config", "def", "dense", "dropout", "hidden_dropout_prob", "hidden_size", "nn", "num_labels", "out_proj", "self", "super"], "longformer/modeling_longformer.py:LongformerClassificationHead.forward": ["ModelClassificationHead", "Module", "class", "def", "dense", "dropout", "forward", "hidden_states", "kwargs", "nn", "out_proj", "output", "return", "self", "tanh", "torch"], "longformer/modeling_longformer.py:LongformerForQuestionAnswering.__init__": ["False", "Linear", "Model", "ModelForQuestionAnswering", "ModelModel", "ModelPreTrainedModel", "__init__", "add_pooling_layer", "class", "config", "def", "hidden_size", "nn", "num_labels", "post_init", "qa_outputs", "self", "super"], "longformer/modeling_longformer.py:LongformerForQuestionAnswering.forward": ["CrossEntropyLoss", "It", "Model", "ModelForQuestionAnswering", "ModelPreTrainedModel", "ModelQuestionAnsweringModelOutput", "None", "Please", "Tensor", "_compute_global_attention_mask", "and", "attention_mask", "attentions", "automatically", "because", "bool", "clamp", "class", "config", "contiguous", "correctly", "def", "dim", "else", "end_logits", "end_loss", "end_positions", "forward", "generate", "global_attention_mask", "global_attentions", "hidden_states", "if", "ignore_index", "ignored_index", "input_ids", "inputs_embeds", "is", "it", "kwargs", "len", "logger", "logits", "loss", "loss_fct", "make", "not", "output", "output_attentions", "output_hidden_states", "outputs", "position_ids", "possible", "qa_outputs", "r", "return", "return_dict", "self", "sep_token_id", "sequence_output", "set", "size", "split", "squeeze", "start_logits", "start_loss", "start_positions", "sure", "that", "the", "to", "token_type_ids", "torch", "total_loss", "tuple", "use_return_dict", "warning"], "longformer/modeling_longformer.py:LongformerForTokenClassification.__init__": ["Dropout", "False", "Linear", "Model", "ModelForTokenClassification", "ModelModel", "ModelPreTrainedModel", "__init__", "add_pooling_layer", "class", "classifier", "config", "def", "dropout", "hidden_dropout_prob", "hidden_size", "nn", "num_labels", "post_init", "self", "super"], "longformer/modeling_longformer.py:LongformerForTokenClassification.forward": ["CrossEntropyLoss", "Model", "ModelForTokenClassification", "ModelPreTrainedModel", "ModelTokenClassifierOutput", "None", "Tensor", "attention_mask", "attentions", "bool", "class", "classifier", "config", "def", "device", "dropout", "else", "forward", "global_attention_mask", "global_attentions", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "not", "num_labels", "output", "output_attentions", "output_hidden_states", "outputs", "position_ids", "r", "return", "return_dict", "self", "sequence_output", "to", "token_type_ids", "torch", "tuple", "use_return_dict", "view"], "longformer/modeling_longformer.py:LongformerForMultipleChoice.__init__": ["Dropout", "Linear", "Model", "ModelForMultipleChoice", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "classifier", "config", "def", "dropout", "hidden_dropout_prob", "hidden_size", "nn", "post_init", "self", "super"], "longformer/modeling_longformer.py:LongformerForMultipleChoice.forward": ["CrossEntropyLoss", "False", "Initializing", "Model", "ModelForMultipleChoice", "ModelMultipleChoiceModelOutput", "ModelPreTrainedModel", "None", "Tensor", "_compute_global_attention_mask", "and", "attention", "attention_mask", "attentions", "before_sep_token", "bool", "choice", "class", "classifier", "config", "def", "device", "dim", "dropout", "else", "flat_attention_mask", "flat_global_attention_mask", "flat_input_ids", "flat_inputs_embeds", "flat_position_ids", "flat_token_type_ids", "for", "forward", "global", "global_attention_mask", "global_attentions", "hidden_states", "i", "if", "in", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logger", "logits", "loss", "loss_fct", "multiple", "not", "num_choices", "on", "output", "output_attentions", "output_hidden_states", "outputs", "pooled_output", "position_ids", "r", "range", "reshaped_logits", "return", "return_dict", "self", "sep_token_id", "shape", "size", "stack", "to", "token_type_ids", "torch", "tuple", "use_return_dict", "view", "warning_once"], "longt5/modeling_longt5.py:_pad_to_multiple": ["Tensor", "_pad_to_multiple", "all", "block_len", "constant", "def", "dim", "dtype", "functional", "if", "int", "list", "mode", "ndim", "new_shape", "nn", "not", "pad", "pad_len", "pad_value", "return", "shape", "sum", "torch", "value", "x", "zeros"], "longt5/modeling_longt5.py:_split_into_blocks": ["Tensor", "_pad_to_multiple", "_split_into_blocks", "block_len", "def", "device", "dim", "dtype", "empty", "if", "in", "int", "num_blocks", "output_shape", "pad_value", "reshape", "return", "shape", "torch", "x"], "longt5/modeling_longt5.py:_concatenate_3_blocks": ["None", "Tensor", "_concatenate_3_blocks", "append", "block_dim", "blocks_list", "cat", "constant", "def", "dim", "for", "functional", "i", "in", "indices", "int", "list", "mode", "ndim", "nn", "num_blocks", "pad", "pad_value", "range", "return", "sequence_dim", "shape", "slice", "sum", "torch", "tuple", "value", "x"], "longt5/modeling_longt5.py:_make_3block_relative_position_ids": ["Tensor", "_make_3block_relative_position_ids", "arange", "block_len", "center_position_ids", "def", "dtype", "int", "int32", "position_ids", "relative_position_ids", "return", "torch", "unsqueeze"], "longt5/modeling_longt5.py:_mask_local_attention_mask": ["None", "Tensor", "_make_3block_relative_position_ids", "_mask_local_attention_mask", "abs", "block_len", "def", "device", "int", "local_attention_mask", "locality_mask", "logical_and", "relative_position_ids", "return", "to", "torch"], "longt5/modeling_longt5.py:_get_local_attention_mask": ["Tensor", "_3blocked_attention_mask", "_blocked_attention_mask", "_concatenate_3_blocks", "_get_local_attention_mask", "_mask_local_attention_mask", "_split_into_blocks", "attention_mask", "block_dim", "block_len", "def", "device", "dim", "int", "local_attention_mask", "logical_and", "return", "sequence_dim", "to", "torch", "unsqueeze"], "longt5/modeling_longt5.py:_make_global_fixed_block_ids": ["Tensor", "_global_block_ids_lower_bound", "_make_global_fixed_block_ids", "_sequence_block_ids_max", "arange", "attention_mask", "axis", "batch_size", "block_ends", "block_ids", "cumsum", "def", "device", "dim", "dtype", "else", "fixed_block_mask", "floor", "full_blocks", "global_block_ids", "global_block_size", "global_segment_ids", "handle_orphan_tokens", "if", "int", "logical_and", "mask", "max", "num_globals", "ones", "ones_like", "repeat", "return", "seq_len", "shape", "sum", "tensor", "to", "torch", "transpose", "true_block_ends", "tuple", "type", "unsqueeze", "values", "where", "zeros"], "longt5/modeling_longt5.py:_make_side_relative_position_ids": ["None", "Tensor", "_make_global_fixed_block_ids", "_make_side_relative_position_ids", "arange", "attention_mask", "block_ids", "def", "device", "global_block_size", "global_positions", "global_segment_ids", "global_seq_len", "int", "int64", "return", "shape", "side_relative_position", "torch", "type"], "longt5/modeling_longt5.py:_create_global_aggregates": ["Tensor", "_create_global_aggregates", "block_ids", "def", "device", "dtype", "einsum", "functional", "gd", "global_seq_len", "hidden_states", "int", "int64", "nd", "ng", "nn", "one_hot", "one_hot_block_ids", "return", "tensor", "torch", "type", "where"], "longt5/modeling_longt5.py:LongT5LayerNorm.__init__": ["ModelLayerNorm", "Module", "Parameter", "__init__", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "longt5/modeling_longt5.py:LongT5LayerNorm.forward": ["ModelLayerNorm", "Module", "True", "bfloat16", "class", "def", "dtype", "float16", "float32", "forward", "hidden_states", "if", "in", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "longt5/modeling_longt5.py:LongT5DenseActDense.__init__": ["ACT2FN", "Dropout", "False", "Linear", "ModelConfig", "ModelDenseActDense", "Module", "__init__", "act", "bias", "class", "config", "d_ff", "d_model", "def", "dense_act_fn", "dropout", "dropout_rate", "nn", "self", "super", "wi", "wo"], "longt5/modeling_longt5.py:LongT5DenseActDense.forward": ["ModelDenseActDense", "Module", "Tensor", "act", "and", "class", "def", "dropout", "dtype", "forward", "hidden_states", "if", "int8", "isinstance", "nn", "return", "self", "to", "torch", "weight", "wi", "wo"], "longt5/modeling_longt5.py:LongT5DenseGatedActDense.__init__": ["ACT2FN", "Dropout", "False", "Linear", "ModelConfig", "ModelDenseGatedActDense", "Module", "__init__", "act", "bias", "class", "config", "d_ff", "d_model", "def", "dense_act_fn", "dropout", "dropout_rate", "nn", "self", "super", "wi_0", "wi_1", "wo"], "longt5/modeling_longt5.py:LongT5DenseGatedActDense.forward": ["ModelDenseGatedActDense", "Module", "act", "class", "def", "dropout", "forward", "hidden_gelu", "hidden_linear", "hidden_states", "nn", "return", "self", "wi_0", "wi_1", "wo"], "longt5/modeling_longt5.py:LongT5LayerFF.__init__": ["DenseReluDense", "Dropout", "ModelConfig", "ModelDenseActDense", "ModelDenseGatedActDense", "ModelLayerFF", "ModelLayerNorm", "Module", "__init__", "class", "config", "d_model", "def", "dropout", "dropout_rate", "else", "eps", "if", "is_gated_act", "layer_norm", "layer_norm_epsilon", "nn", "self", "super"], "longt5/modeling_longt5.py:LongT5LayerFF.forward": ["DenseReluDense", "ModelLayerFF", "Module", "class", "def", "dropout", "forward", "forwarded_states", "hidden_states", "layer_norm", "nn", "return", "self"], "longt5/modeling_longt5.py:LongT5Attention.__init__": ["Embedding", "False", "Instantiating", "Linear", "ModelAttention", "ModelConfig", "Module", "None", "Please", "__class__", "__init__", "__name__", "a", "and", "bias", "caching", "call", "class", "config", "creating", "d_kv", "d_model", "decoder", "def", "dropout", "dropout_rate", "during", "errors", "f", "forward", "gradient_checkpointing", "has_relative_attention_bias", "if", "inner_dim", "int", "is", "is_decoder", "k", "key_value_proj_dim", "layer_idx", "logger", "make", "n_heads", "nn", "not", "num_heads", "o", "passing", "provide", "q", "recommended", "relative_attention_bias", "relative_attention_max_distance", "relative_attention_num_buckets", "self", "super", "sure", "the", "this", "to", "used", "v", "warning_once", "when", "will", "without"], "longt5/modeling_longt5.py:LongT5Attention._relative_position_bucket": ["Model", "ModelAttention", "Module", "True", "_relative_position_bucket", "abs", "bidirectional", "class", "def", "else", "float", "full_like", "if", "is_small", "log", "math", "max_distance", "max_exact", "min", "nn", "num_buckets", "relative_buckets", "relative_position", "relative_position_if_large", "return", "to", "torch", "where", "zeros_like"], "longt5/modeling_longt5.py:LongT5Attention.compute_bias": ["Model", "ModelAttention", "Module", "None", "_relative_position_bucket", "arange", "bidirectional", "cache_position", "class", "compute_bias", "context_position", "def", "device", "dtype", "else", "if", "is", "is_decoder", "key_length", "max_distance", "memory_position", "nn", "not", "num_buckets", "permute", "query_length", "relative_attention_bias", "relative_attention_max_distance", "relative_attention_num_buckets", "relative_position", "relative_position_bucket", "return", "self", "to", "torch", "unsqueeze", "values", "weight"], "longt5/modeling_longt5.py:LongT5Attention.forward": ["EncoderDecoderCache", "False", "ModelAttention", "Module", "None", "True", "and", "attn_output", "attn_weights", "batch_size", "cache_position", "causal_mask", "class", "compute_bias", "contiguous", "cross_attention_cache", "curr_past_key_values", "current_states", "def", "device", "dim", "dropout", "dtype", "else", "float", "forward", "functional", "get", "gradient_checkpointing", "has_relative_attention_bias", "hidden_states", "if", "inner_dim", "is", "is_cross_attention", "is_updated", "isinstance", "k", "key_length", "key_states", "key_value_proj_dim", "key_value_states", "keys", "layer_idx", "layers", "mask", "matmul", "n_heads", "nn", "not", "o", "output_attentions", "outputs", "p", "past_key_values", "position_bias", "position_bias_masked", "q", "query_length", "query_states", "real_seq_length", "requires_grad", "return", "scores", "self", "self_attention_cache", "seq_length", "shape", "softmax", "torch", "training", "transpose", "type_as", "update", "use_cache", "v", "value_states", "values", "view", "zeros"], "longt5/modeling_longt5.py:LongT5LocalAttention.__init__": ["Embedding", "False", "Linear", "ModelConfig", "ModelLocalAttention", "Module", "None", "__init__", "bias", "block_len", "bool", "class", "config", "d_kv", "d_model", "def", "dropout", "dropout_rate", "gradient_checkpointing", "has_relative_attention_bias", "if", "inner_dim", "is_decoder", "k", "key_value_proj_dim", "local_radius", "n_heads", "nn", "num_heads", "o", "q", "relative_attention_bias", "relative_attention_max_distance", "relative_attention_num_buckets", "self", "super", "v"], "longt5/modeling_longt5.py:LongT5LocalAttention._relative_position_bucket": ["Model", "ModelLocalAttention", "Module", "True", "_relative_position_bucket", "abs", "bidirectional", "class", "def", "else", "float", "full_like", "if", "is_small", "log", "math", "max_distance", "max_exact", "min", "nn", "num_buckets", "relative_buckets", "relative_position", "relative_position_if_large", "return", "to", "torch", "where", "zeros_like"], "longt5/modeling_longt5.py:LongT5LocalAttention.compute_bias": ["Model", "ModelLocalAttention", "Module", "None", "_relative_position_bucket", "arange", "bidirectional", "block_length", "class", "compute_bias", "context_position", "def", "device", "dtype", "else", "if", "int", "is_decoder", "max_distance", "memory_position", "meta", "nn", "not", "num_buckets", "permute", "relative_attention_bias", "relative_attention_max_distance", "relative_attention_num_buckets", "relative_position", "relative_position_bucket", "return", "self", "target_device", "torch", "type", "unsqueeze", "values", "weight"], "longt5/modeling_longt5.py:LongT5LocalAttention.forward": ["False", "ModelLocalAttention", "Module", "None", "True", "_concatenate_3_blocks", "_split_into_blocks", "and", "attn_output", "attn_weights", "batch_size", "block_dim", "block_len", "class", "compute_bias", "contiguous", "def", "device", "dim", "dropout", "dtype", "einsum", "else", "float", "forward", "functional", "gradient_checkpointing", "has_relative_attention_bias", "hidden_states", "hqk", "if", "inner_dim", "is", "k", "key_states", "key_value_proj_dim", "khd", "mask", "n_heads", "nn", "not", "o", "output_attentions", "outputs", "p", "position_bias", "q", "qhd", "query_states", "requires_grad", "return", "scores", "self", "seq_length", "sequence_dim", "shape", "softmax", "states", "torch", "training", "transpose", "type", "type_as", "unshape", "v", "value_states", "view", "where", "zeros"], "longt5/modeling_longt5.py:LongT5TransientGlobalAttention.__init__": ["Embedding", "False", "Linear", "ModelConfig", "ModelLayerNorm", "ModelTransientGlobalAttention", "Module", "None", "__init__", "bias", "block_len", "bool", "class", "config", "d_kv", "d_model", "def", "dropout", "dropout_rate", "eps", "global_block_size", "global_input_layer_norm", "global_relative_attention_bias", "has_relative_attention_bias", "if", "inner_dim", "is_decoder", "k", "key_value_proj_dim", "layer_norm_epsilon", "local_radius", "n_heads", "nn", "num_heads", "o", "q", "relative_attention_bias", "relative_attention_max_distance", "relative_attention_num_buckets", "self", "super", "v"], "longt5/modeling_longt5.py:LongT5TransientGlobalAttention._relative_position_bucket": ["Model", "ModelTransientGlobalAttention", "Module", "True", "_relative_position_bucket", "abs", "bidirectional", "class", "def", "else", "float", "full_like", "if", "is_small", "log", "math", "max_distance", "max_exact", "min", "nn", "num_buckets", "relative_buckets", "relative_position", "relative_position_if_large", "return", "to", "torch", "where", "zeros_like"], "longt5/modeling_longt5.py:LongT5TransientGlobalAttention.compute_bias": ["Model", "ModelTransientGlobalAttention", "Module", "None", "_relative_position_bucket", "arange", "bidirectional", "block_length", "class", "compute_bias", "context_position", "def", "device", "dtype", "else", "if", "int", "is_decoder", "max_distance", "memory_position", "meta", "nn", "not", "num_buckets", "permute", "relative_attention_bias", "relative_attention_max_distance", "relative_attention_num_buckets", "relative_position", "relative_position_bucket", "return", "self", "target_device", "torch", "type", "unsqueeze", "values", "weight"], "longt5/modeling_longt5.py:LongT5TransientGlobalAttention.compute_side_bias": ["ModelTransientGlobalAttention", "Module", "None", "Tensor", "_make_side_relative_position_ids", "_relative_position_bucket", "attention_side_bias", "bidirectional", "class", "compute_side_bias", "def", "eq", "global_block_size", "global_relative_attention_bias", "global_segment_ids", "is_decoder", "mask", "max_distance", "nn", "not", "num_buckets", "permute", "relative_attention_max_distance", "relative_attention_num_buckets", "return", "self", "side_attention_mask", "side_bias", "side_relative_position", "side_relative_position_bucket", "torch", "where"], "longt5/modeling_longt5.py:LongT5TransientGlobalAttention.forward": ["False", "ModelTransientGlobalAttention", "Module", "None", "True", "_concatenate_3_blocks", "_create_global_aggregates", "_get_local_attention_mask", "_global_seq_len", "_make_global_fixed_block_ids", "_split_into_blocks", "and", "attn_output", "attn_weights", "batch_size", "block_dim", "block_ids", "block_len", "cat", "class", "compute_bias", "compute_side_bias", "contiguous", "def", "device", "dim", "dropout", "dtype", "einsum", "else", "float", "forward", "functional", "global_block_size", "global_input_layer_norm", "global_inputs", "global_segment_ids", "gradient_checkpointing", "has_relative_attention_bias", "hidden_states", "hqk", "if", "inner_dim", "is", "k", "key_states", "key_value_proj_dim", "khd", "local_attention_mask", "mask", "n_heads", "ndim", "nn", "not", "o", "ones", "output_attentions", "outputs", "p", "position_bias", "q", "qhd", "query_states", "repeat", "reps", "requires_grad", "return", "scores", "self", "seq_length", "sequence_dim", "shape", "side_key_states", "side_position_bias", "side_value_states", "softmax", "states", "to", "torch", "training", "transpose", "type", "type_as", "unshape", "unsqueeze", "v", "value_states", "view", "where", "zeros"], "longt5/modeling_longt5.py:LongT5LayerSelfAttention.__init__": ["Dropout", "False", "ModelAttention", "ModelLayerNorm", "ModelLayerSelfAttention", "Module", "None", "SelfAttention", "__init__", "class", "config", "d_model", "def", "dropout", "dropout_rate", "eps", "has_relative_attention_bias", "int", "layer_idx", "layer_norm", "layer_norm_epsilon", "nn", "self", "super"], "longt5/modeling_longt5.py:LongT5LayerSelfAttention.forward": ["False", "ModelLayerSelfAttention", "Module", "None", "SelfAttention", "attention_mask", "attention_output", "cache_position", "class", "def", "dropout", "forward", "hidden_states", "layer_norm", "mask", "nn", "normed_hidden_states", "output_attentions", "outputs", "past_key_values", "position_bias", "return", "self", "use_cache"], "longt5/modeling_longt5.py:LongT5LayerLocalSelfAttention.__init__": ["Dropout", "False", "LocalSelfAttention", "ModelLayerLocalSelfAttention", "ModelLayerNorm", "ModelLocalAttention", "Module", "None", "__init__", "class", "config", "d_model", "def", "dropout", "dropout_rate", "eps", "has_relative_attention_bias", "int", "layer_idx", "layer_norm", "layer_norm_epsilon", "nn", "self", "super"], "longt5/modeling_longt5.py:LongT5LayerLocalSelfAttention.forward": ["Any", "False", "LocalSelfAttention", "ModelLayerLocalSelfAttention", "Module", "None", "attention_mask", "attention_output", "class", "def", "dropout", "forward", "hidden_states", "kwargs", "layer_norm", "mask", "nn", "normed_hidden_states", "output_attentions", "outputs", "position_bias", "return", "self"], "longt5/modeling_longt5.py:LongT5LayerTransientGlobalSelfAttention.__init__": ["Dropout", "False", "ModelLayerNorm", "ModelLayerTransientGlobalSelfAttention", "ModelTransientGlobalAttention", "Module", "None", "TransientGlobalSelfAttention", "__init__", "class", "config", "d_model", "def", "dropout", "dropout_rate", "eps", "has_relative_attention_bias", "int", "layer_idx", "layer_norm", "layer_norm_epsilon", "nn", "self", "super"], "longt5/modeling_longt5.py:LongT5LayerTransientGlobalSelfAttention.forward": ["Any", "False", "ModelLayerTransientGlobalSelfAttention", "Module", "None", "TransientGlobalSelfAttention", "attention_mask", "attention_output", "class", "def", "dropout", "forward", "hidden_states", "kwargs", "layer_norm", "mask", "nn", "normed_hidden_states", "output_attentions", "outputs", "position_bias", "return", "self"], "longt5/modeling_longt5.py:LongT5LayerCrossAttention.__init__": ["Dropout", "EncDecAttention", "False", "ModelAttention", "ModelLayerCrossAttention", "ModelLayerNorm", "Module", "None", "__init__", "class", "config", "d_model", "def", "dropout", "dropout_rate", "eps", "has_relative_attention_bias", "int", "layer_idx", "layer_norm", "layer_norm_epsilon", "nn", "self", "super"], "longt5/modeling_longt5.py:LongT5LayerCrossAttention.forward": ["EncDecAttention", "False", "ModelLayerCrossAttention", "Module", "None", "attention_mask", "attention_output", "cache_position", "class", "def", "dropout", "forward", "hidden_states", "key_value_states", "layer_norm", "layer_output", "mask", "nn", "normed_hidden_states", "output_attentions", "outputs", "past_key_values", "position_bias", "query_length", "return", "self", "use_cache"], "longt5/modeling_longt5.py:LongT5Block.__init__": ["False", "For", "GradientCheckpointingLayer", "ModelBlock", "ModelLayerCrossAttention", "ModelLayerFF", "ModelLayerLocalSelfAttention", "ModelLayerSelfAttention", "ModelLayerTransientGlobalSelfAttention", "ModuleList", "None", "ValueError", "__init__", "append", "attention", "attention_layer", "but", "class", "config", "def", "either", "elif", "else", "encoder", "encoder_attention_type", "expected", "f", "global", "got", "has_relative_attention_bias", "if", "int", "is", "is_decoder", "layer", "layer_idx", "local", "mechanism", "nn", "or", "raise", "self", "super", "transient", "type"], "longt5/modeling_longt5.py:LongT5Block.forward": ["False", "GradientCheckpointingLayer", "ModelBlock", "None", "True", "and", "any", "attention_mask", "attention_outputs", "cache_position", "clamp", "clamp_value", "class", "cross_attention_outputs", "def", "do_cross_attention", "dtype", "encoder_attention_mask", "encoder_decoder_position_bias", "encoder_hidden_states", "finfo", "float16", "forward", "hidden_states", "if", "is", "is_decoder", "isinf", "key_value_states", "layer", "max", "min", "not", "output_attentions", "past_key_values", "position_bias", "query_length", "return", "return_dict", "self", "self_attention_outputs", "torch", "use_cache"], "longt5/modeling_longt5.py:LongT5PreTrainedModel.dummy_inputs": ["DUMMY_INPUTS", "DUMMY_MASK", "ModelPreTrainedModel", "PreTrainedModel", "class", "decoder_attention_mask", "decoder_input_ids", "def", "dummy_inputs", "input_ids", "input_mask", "return", "self", "tensor", "torch"], "longt5/modeling_longt5.py:LongT5PreTrainedModel._init_weights": ["ModelAttention", "ModelDenseActDense", "ModelDenseGatedActDense", "ModelEncoderModel", "ModelForConditionalGeneration", "ModelLayerNorm", "ModelLocalAttention", "ModelModel", "ModelPreTrainedModel", "ModelTransientGlobalAttention", "None", "PreTrainedModel", "_init_weights", "and", "bias", "class", "config", "constant_", "d_ff", "d_kv", "d_model", "def", "elif", "factor", "global_relative_attention_bias", "has_relative_attention_bias", "hasattr", "if", "init", "initializer_factor", "is", "isinstance", "k", "key_value_proj_dim", "lm_head", "mean", "module", "n_heads", "normal_", "not", "num_heads", "o", "q", "relative_attention_bias", "self", "shared", "std", "tie_word_embeddings", "v", "weight", "wi", "wi_0", "wi_1", "wo", "zeros_"], "longt5/modeling_longt5.py:LongT5PreTrainedModel._shift_right": ["In", "Model", "ModelPreTrainedModel", "None", "PreTrainedModel", "See", "ValueError", "_shift_right", "be", "class", "clone", "config", "decoder_start_token_id", "def", "defined", "docs", "for", "has", "if", "information", "input_ids", "is", "it", "masked_fill_", "model", "more", "new_zeros", "pad_token_id", "raise", "return", "self", "set", "shape", "shifted_input_ids", "the", "to", "usually"], "longt5/modeling_longt5.py:LongT5Stack.__init__": ["Dropout", "Embedding", "False", "ModelBlock", "ModelLayerNorm", "ModelPreTrainedModel", "ModelStack", "ModuleList", "__init__", "block", "block_len", "bool", "class", "config", "d_model", "def", "dropout", "dropout_rate", "embed_tokens", "eps", "final_layer_norm", "for", "gradient_checkpointing", "has_relative_attention_bias", "i", "in", "is_decoder", "layer_idx", "layer_norm_epsilon", "local_radius", "nn", "num_layers", "post_init", "range", "self", "super", "vocab_size"], "longt5/modeling_longt5.py:LongT5Stack.set_input_embeddings": ["ModelPreTrainedModel", "ModelStack", "class", "def", "embed_tokens", "new_embeddings", "self", "set_input_embeddings"], "longt5/modeling_longt5.py:LongT5Stack.forward": ["BaseModelOutputWithPastAndCrossAttentions", "DynamicCache", "EncoderDecoderCache", "False", "ModelPreTrainedModel", "ModelStack", "None", "Setting", "True", "ValueError", "You", "_", "_get_local_attention_mask", "_update_causal_mask", "all_attentions", "all_cross_attentions", "all_hidden_states", "and", "arange", "assert", "at", "attention_mask", "attentions", "batch_size", "block", "block_len", "both", "cache_position", "cannot", "causal_mask", "checkpointing", "class", "config", "cross_attentions", "decoder_", "def", "device", "dropout", "either", "elif", "else", "embed_tokens", "embeddings", "encoder_attention_mask", "encoder_attention_type", "encoder_batch_size", "encoder_decoder_position_bias", "encoder_extended_attention_mask", "encoder_hidden_shape", "encoder_hidden_states", "encoder_sequence_length", "enumerate", "err_msg_prefix", "f", "final_layer_norm", "for", "forward", "get_seq_length", "gradient", "gradient_checkpointing", "have", "hidden_states", "i", "if", "in", "incompatible", "initialize", "input_ids", "input_shape", "inputs_embeds", "invert_attention_mask", "is", "is_decoder", "is_encoder_decoder", "is_torchdynamo_compiling", "isinstance", "kwargs", "last_hidden_state", "layer_module", "layer_outputs", "local", "logger", "mask_seq_length", "model", "not", "ones", "or", "output_attentions", "output_hidden_states", "past_key_values", "past_key_values_length", "position_bias", "raise", "return", "return_dict", "same", "self", "self_attention_cache", "seq_length", "size", "specify", "the", "time", "to", "token", "torch", "training", "tuple", "use_cache", "use_return_dict", "v", "valid", "view", "warning_once", "with"], "longt5/modeling_longt5.py:LongT5Stack._update_causal_mask": ["AttentionMaskConverter", "BlockMask", "Cache", "False", "ModelPreTrainedModel", "ModelStack", "None", "Tensor", "Union", "_attn_implementation", "_ignore_causal_mask_sdpa", "_prepare_4d_causal_attention_mask_with_cache_position", "_unmask_unattended", "_update_causal_mask", "and", "any", "attention_mask", "batch_size", "bool", "cache_position", "causal_mask", "class", "config", "cuda", "def", "device", "dtype", "else", "finfo", "flash_attention_2", "flex_attention", "get_max_cache_shape", "get_seq_length", "if", "in", "input_tensor", "inputs_embeds", "is", "is_compileable", "is_training", "isinstance", "make_flex_block_causal_mask", "min", "min_dtype", "not", "npu", "output_attentions", "past_key_values", "past_key_values_length", "past_seen_tokens", "return", "sdpa", "self", "sequence_length", "shape", "target_length", "torch", "training", "type", "using_compilable_cache", "xpu"], "longt5/modeling_longt5.py:LongT5Stack._prepare_4d_causal_attention_mask_with_cache_position": ["ModelPreTrainedModel", "ModelStack", "None", "Tensor", "_prepare_4d_causal_attention_mask_with_cache_position", "and", "arange", "attention_mask", "batch_size", "cache_position", "causal_mask", "class", "clone", "def", "device", "diagonal", "dim", "dtype", "else", "expand", "fill_value", "finfo", "full", "if", "int", "is", "kwargs", "mask_length", "masked_fill", "min", "min_dtype", "not", "padding_mask", "reshape", "return", "sequence_length", "shape", "target_length", "to", "torch", "triu"], "longt5/modeling_longt5.py:LongT5Model.__init__": ["Embedding", "False", "ModelConfig", "ModelModel", "ModelPreTrainedModel", "ModelStack", "True", "__init__", "class", "config", "copy", "d_model", "decoder", "decoder_config", "deepcopy", "def", "encoder", "encoder_config", "is_decoder", "nn", "num_decoder_layers", "num_layers", "post_init", "self", "shared", "super", "use_cache", "vocab_size"], "longt5/modeling_longt5.py:LongT5Model.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "return", "self", "shared"], "longt5/modeling_longt5.py:LongT5Model.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "decoder", "def", "encoder", "new_embeddings", "self", "set_input_embeddings", "shared"], "longt5/modeling_longt5.py:LongT5Model.forward": ["BaseModelOutput", "BoolTensor", "Cache", "FloatTensor", "ModelModel", "ModelPreTrainedModel", "Modelensor", "None", "Seq2SeqModelOutput", "Tensor", "and", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "cross_attentions", "decoder", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_input_ids", "decoder_inputs_embeds", "decoder_outputs", "def", "elif", "else", "encoder", "encoder_attention_mask", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "isinstance", "kwargs", "last_hidden_state", "len", "not", "output_attentions", "output_hidden_states", "past_key_values", "r", "return", "return_dict", "self", "torch", "tuple", "use_cache", "use_return_dict"], "longt5/modeling_longt5.py:LongT5ForConditionalGeneration.__init__": ["Embedding", "False", "GenerationMixin", "Linear", "ModelConfig", "ModelForConditionalGeneration", "ModelPreTrainedModel", "ModelStack", "True", "__init__", "bias", "class", "config", "copy", "d_model", "decoder", "decoder_config", "deepcopy", "def", "encoder", "encoder_config", "is_decoder", "lm_head", "model_dim", "nn", "num_decoder_layers", "num_layers", "post_init", "self", "shared", "super", "use_cache", "vocab_size"], "longt5/modeling_longt5.py:LongT5ForConditionalGeneration.get_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "return", "self", "shared"], "longt5/modeling_longt5.py:LongT5ForConditionalGeneration.set_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "decoder", "def", "encoder", "new_embeddings", "self", "set_input_embeddings", "shared"], "longt5/modeling_longt5.py:LongT5ForConditionalGeneration.forward": ["BaseModelOutput", "BoolTensor", "Cache", "CrossEntropyLoss", "FloatTensor", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "Modelensor", "None", "Seq2SeqLMOutput", "Tensor", "_shift_right", "and", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "cross_attentions", "decoder", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_input_ids", "decoder_inputs_embeds", "decoder_outputs", "def", "device", "elif", "else", "encoder", "encoder_attention_mask", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "forward", "hidden_states", "if", "ignore_index", "input_ids", "inputs_embeds", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "len", "lm_head", "lm_logits", "logits", "loss", "loss_fct", "model_dim", "not", "output", "output_attentions", "output_hidden_states", "past_key_values", "r", "return", "return_dict", "self", "sequence_output", "size", "tie_word_embeddings", "to", "torch", "tuple", "use_cache", "use_return_dict", "view"], "longt5/modeling_longt5.py:LongT5ForConditionalGeneration.prepare_decoder_input_ids_from_labels": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "Tensor", "_shift_right", "class", "def", "labels", "prepare_decoder_input_ids_from_labels", "return", "self", "torch"], "longt5/modeling_longt5.py:LongT5EncoderModel.__init__": ["Embedding", "False", "ModelConfig", "ModelEncoderModel", "ModelPreTrainedModel", "ModelStack", "__init__", "class", "config", "copy", "d_model", "deepcopy", "def", "encoder", "encoder_config", "nn", "post_init", "self", "shared", "super", "use_cache", "vocab_size"], "longt5/modeling_longt5.py:LongT5EncoderModel.get_input_embeddings": ["ModelEncoderModel", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "return", "self", "shared"], "longt5/modeling_longt5.py:LongT5EncoderModel.set_input_embeddings": ["ModelEncoderModel", "ModelPreTrainedModel", "class", "def", "encoder", "new_embeddings", "self", "set_input_embeddings", "shared"], "longt5/modeling_longt5.py:LongT5EncoderModel.forward": ["BaseModelOutput", "FloatTensor", "ModelEncoderModel", "ModelPreTrainedModel", "Modelensor", "None", "attention_mask", "bool", "class", "config", "def", "else", "encoder", "encoder_outputs", "forward", "if", "input_ids", "inputs_embeds", "is", "kwargs", "not", "output_attentions", "output_hidden_states", "r", "return", "return_dict", "self", "torch", "tuple", "use_return_dict"], "luke/modeling_luke.py:LukeEmbeddings.__init__": ["Dropout", "Embedding", "LayerNorm", "ModelEmbeddings", "Module", "__init__", "class", "config", "def", "dropout", "eps", "hidden_dropout_prob", "hidden_size", "layer_norm_eps", "max_position_embeddings", "nn", "pad_token_id", "padding_idx", "position_embeddings", "self", "super", "token_type_embeddings", "type_vocab_size", "vocab_size", "word_embeddings"], "luke/modeling_luke.py:LukeEmbeddings.forward": ["LayerNorm", "ModelEmbeddings", "Module", "None", "class", "create_position_ids_from_input_ids", "create_position_ids_from_inputs_embeds", "def", "device", "dropout", "dtype", "else", "embeddings", "forward", "if", "input_ids", "input_shape", "inputs_embeds", "is", "long", "nn", "not", "padding_idx", "position_embeddings", "position_ids", "return", "self", "size", "to", "token_type_embeddings", "token_type_ids", "torch", "word_embeddings", "zeros"], "luke/modeling_luke.py:LukeEmbeddings.create_position_ids_from_inputs_embeds": ["ModelEmbeddings", "Module", "arange", "class", "create_position_ids_from_inputs_embeds", "def", "device", "dtype", "expand", "input_shape", "inputs_embeds", "long", "nn", "padding_idx", "position_ids", "return", "self", "sequence_length", "size", "torch", "unsqueeze"], "luke/modeling_luke.py:LukeEntityEmbeddings.__init__": ["Dropout", "Embedding", "False", "LayerNorm", "Linear", "ModelConfig", "ModelEntityEmbeddings", "Module", "__init__", "bias", "class", "config", "def", "dropout", "entity_emb_size", "entity_embedding_dense", "entity_embeddings", "entity_vocab_size", "eps", "hidden_dropout_prob", "hidden_size", "if", "layer_norm_eps", "max_position_embeddings", "nn", "padding_idx", "position_embeddings", "self", "super", "token_type_embeddings", "type_vocab_size"], "luke/modeling_luke.py:LukeEntityEmbeddings.forward": ["LayerNorm", "LongTensor", "ModelEntityEmbeddings", "Module", "None", "clamp", "class", "config", "def", "dim", "dropout", "embeddings", "entity_emb_size", "entity_embedding_dense", "entity_embeddings", "entity_ids", "forward", "hidden_size", "if", "is", "min", "nn", "position_embedding_mask", "position_embeddings", "position_ids", "return", "self", "sum", "token_type_embeddings", "token_type_ids", "torch", "type_as", "unsqueeze", "zeros_like"], "luke/modeling_luke.py:LukeSelfAttention.__init__": ["Dropout", "Linear", "ModelSelfAttention", "Module", "The", "ValueError", "__init__", "a", "all_head_size", "and", "attention", "attention_head_size", "attention_probs_dropout_prob", "class", "config", "def", "dropout", "e2e_query", "e2w_query", "embedding_size", "f", "hasattr", "heads", "hidden", "hidden_size", "if", "int", "is", "key", "multiple", "nn", "not", "num_attention_heads", "number", "of", "query", "raise", "self", "size", "super", "the", "use_entity_aware_attention", "value", "w2e_query"], "luke/modeling_luke.py:LukeSelfAttention.transpose_for_scores": ["ModelSelfAttention", "Module", "attention_head_size", "class", "def", "new_x_shape", "nn", "num_attention_heads", "permute", "return", "self", "size", "transpose_for_scores", "view", "x"], "luke/modeling_luke.py:LukeSelfAttention.forward": ["False", "ModelSelfAttention", "Module", "None", "all_head_size", "and", "attention_head_size", "attention_mask", "attention_probs", "attention_scores", "cat", "class", "concat_hidden_states", "context_layer", "contiguous", "def", "dim", "dropout", "e2e_attention_scores", "e2e_key_layer", "e2e_query", "e2e_query_layer", "e2w_attention_scores", "e2w_key_layer", "e2w_query", "e2w_query_layer", "else", "entity_attention_scores", "entity_hidden_states", "forward", "functional", "if", "is", "key", "key_layer", "math", "matmul", "new_context_layer_shape", "nn", "not", "output_attentions", "output_entity_hidden_states", "output_word_hidden_states", "outputs", "permute", "query", "query_layer", "return", "self", "size", "softmax", "sqrt", "torch", "transpose", "transpose_for_scores", "use_entity_aware_attention", "value", "value_layer", "view", "w2e_attention_scores", "w2e_key_layer", "w2e_query", "w2e_query_layer", "w2w_attention_scores", "w2w_key_layer", "w2w_query_layer", "word_attention_scores", "word_hidden_states", "word_size"], "luke/modeling_luke.py:LukeSelfOutput.__init__": ["Dropout", "LayerNorm", "Linear", "ModelSelfOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "eps", "hidden_dropout_prob", "hidden_size", "layer_norm_eps", "nn", "self", "super"], "luke/modeling_luke.py:LukeSelfOutput.forward": ["LayerNorm", "ModelSelfOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "luke/modeling_luke.py:LukeAttention.__init__": ["ModelAttention", "ModelSelfAttention", "ModelSelfOutput", "Module", "__init__", "class", "config", "def", "nn", "output", "self", "super"], "luke/modeling_luke.py:LukeAttention.forward": ["False", "ModelAttention", "Module", "None", "attention_mask", "attention_output", "cat", "class", "concat_hidden_states", "concat_self_outputs", "def", "dim", "else", "entity_attention_output", "entity_hidden_states", "forward", "if", "is", "nn", "output", "output_attentions", "outputs", "return", "self", "self_outputs", "size", "torch", "word_attention_output", "word_hidden_states", "word_size"], "luke/modeling_luke.py:LukeIntermediate.__init__": ["ACT2FN", "Linear", "ModelIntermediate", "Module", "__init__", "class", "config", "def", "dense", "else", "hidden_act", "hidden_size", "if", "intermediate_act_fn", "intermediate_size", "isinstance", "nn", "self", "str", "super"], "luke/modeling_luke.py:LukeIntermediate.forward": ["ModelIntermediate", "Module", "Tensor", "class", "def", "dense", "forward", "hidden_states", "intermediate_act_fn", "nn", "return", "self", "torch"], "luke/modeling_luke.py:LukeOutput.__init__": ["Dropout", "LayerNorm", "Linear", "ModelOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "eps", "hidden_dropout_prob", "hidden_size", "intermediate_size", "layer_norm_eps", "nn", "self", "super"], "luke/modeling_luke.py:LukeOutput.forward": ["LayerNorm", "ModelOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "luke/modeling_luke.py:LukeLayer.__init__": ["GradientCheckpointingLayer", "ModelAttention", "ModelIntermediate", "ModelLayer", "ModelOutput", "__init__", "attention", "chunk_size_feed_forward", "class", "config", "def", "intermediate", "output", "self", "seq_len_dim", "super"], "luke/modeling_luke.py:LukeLayer.forward": ["False", "GradientCheckpointingLayer", "ModelLayer", "None", "apply_chunking_to_forward", "attention", "attention_mask", "cat", "chunk_size_feed_forward", "class", "concat_attention_output", "def", "dim", "else", "entity_hidden_states", "entity_layer_output", "feed_forward_chunk", "forward", "if", "is", "layer_output", "output_attentions", "outputs", "return", "self", "self_attention_outputs", "seq_len_dim", "size", "torch", "word_hidden_states", "word_layer_output", "word_size"], "luke/modeling_luke.py:LukeLayer.feed_forward_chunk": ["GradientCheckpointingLayer", "ModelLayer", "attention_output", "class", "def", "feed_forward_chunk", "intermediate", "intermediate_output", "layer_output", "output", "return", "self"], "luke/modeling_luke.py:LukeEncoder.__init__": ["False", "ModelEncoder", "ModelLayer", "Module", "ModuleList", "_", "__init__", "class", "config", "def", "for", "gradient_checkpointing", "in", "layer", "nn", "num_hidden_layers", "range", "self", "super"], "luke/modeling_luke.py:LukeEncoder.forward": ["BaseModelModelOutput", "False", "ModelEncoder", "Module", "None", "True", "all_entity_hidden_states", "all_self_attentions", "all_word_hidden_states", "attention_mask", "attentions", "class", "def", "else", "entity_hidden_states", "entity_last_hidden_state", "enumerate", "for", "forward", "hidden_states", "i", "if", "in", "is", "last_hidden_state", "layer", "layer_module", "layer_outputs", "nn", "not", "output_attentions", "output_hidden_states", "return", "return_dict", "self", "tuple", "v", "word_hidden_states"], "luke/modeling_luke.py:LukePooler.__init__": ["Linear", "ModelPooler", "Module", "Tanh", "__init__", "activation", "class", "config", "def", "dense", "hidden_size", "nn", "self", "super"], "luke/modeling_luke.py:LukePooler.forward": ["ModelPooler", "Module", "Tensor", "activation", "class", "def", "dense", "first_token_tensor", "forward", "hidden_states", "nn", "pooled_output", "return", "self", "torch"], "luke/modeling_luke.py:EntityPredictionHeadTransform.__init__": ["ACT2FN", "LayerNorm", "Linear", "ModelPredictionHeadTransform", "Model_emb_size", "Module", "__init__", "class", "config", "def", "dense", "else", "eps", "hidden_act", "hidden_size", "if", "isinstance", "layer_norm_eps", "nn", "self", "str", "super", "transform_act_fn"], "luke/modeling_luke.py:EntityPredictionHeadTransform.forward": ["LayerNorm", "ModelPredictionHeadTransform", "Module", "class", "def", "dense", "forward", "hidden_states", "nn", "return", "self", "transform_act_fn"], "luke/modeling_luke.py:EntityPredictionHead.__init__": ["False", "Linear", "ModelPredictionHead", "ModelPredictionHeadTransform", "Model_emb_size", "Model_vocab_size", "Module", "Parameter", "__init__", "bias", "class", "config", "decoder", "def", "nn", "self", "super", "torch", "transform", "zeros"], "luke/modeling_luke.py:EntityPredictionHead.forward": ["ModelPredictionHead", "Module", "bias", "class", "decoder", "def", "forward", "hidden_states", "nn", "return", "self", "transform"], "luke/modeling_luke.py:LukePreTrainedModel._init_weights": ["Embedding", "False", "LayerNorm", "Linear", "ModelPreTrainedModel", "Module", "None", "PreTrainedModel", "_init_weights", "_is_hf_initialized", "and", "bias", "class", "config", "def", "elif", "else", "embedding_dim", "getattr", "if", "init", "initializer_range", "is", "isinstance", "mean", "module", "nn", "normal_", "not", "ones_", "padding_idx", "self", "std", "weight", "zeros_"], "luke/modeling_luke.py:LukeModel.__init__": ["ModelConfig", "ModelEmbeddings", "ModelEncoder", "ModelEntityEmbeddings", "ModelModel", "ModelPooler", "ModelPreTrainedModel", "None", "True", "__init__", "add_pooling_layer", "bool", "class", "config", "def", "else", "embeddings", "encoder", "entity_embeddings", "if", "pooler", "post_init", "r", "self", "super"], "luke/modeling_luke.py:LukeModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "return", "self", "word_embeddings"], "luke/modeling_luke.py:LukeModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "self", "set_input_embeddings", "value", "word_embeddings"], "luke/modeling_luke.py:LukeModel.get_entity_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "entity_embeddings", "get_entity_embeddings", "return", "self"], "luke/modeling_luke.py:LukeModel.set_entity_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "entity_embeddings", "self", "set_entity_embeddings", "value"], "luke/modeling_luke.py:LukeModel.forward": ["BaseModelModelOutputWithPooling", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "ValueError", "You", "and", "at", "attention_mask", "attentions", "batch_size", "bool", "both", "cannot", "class", "config", "def", "device", "dtype", "either", "elif", "else", "embeddings", "encoder", "encoder_outputs", "entity_attention_mask", "entity_embedding_output", "entity_embeddings", "entity_hidden_states", "entity_ids", "entity_last_hidden_state", "entity_position_ids", "entity_seq_length", "entity_token_type_ids", "extended_attention_mask", "forward", "get_extended_attention_mask", "have", "hidden_states", "if", "input_ids", "input_shape", "inputs_embeds", "is", "kwargs", "last_hidden_state", "long", "not", "ones", "or", "output_attentions", "output_hidden_states", "pooled_output", "pooler", "pooler_output", "position_ids", "r", "raise", "return", "return_dict", "same", "self", "seq_length", "sequence_output", "size", "specify", "the", "time", "to", "token_type_ids", "torch", "tuple", "use_return_dict", "warn_if_padding_and_no_attention_mask", "word_embedding_output", "zeros"], "luke/modeling_luke.py:LukeModel.get_extended_attention_mask": ["LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "ValueError", "Wrong", "attention_mask", "cat", "class", "def", "dim", "dtype", "elif", "else", "entity_attention_mask", "extended_attention_mask", "f", "finfo", "for", "get_extended_attention_mask", "if", "is", "min", "not", "raise", "return", "self", "shape", "to", "torch", "word_attention_mask"], "luke/modeling_luke.py:create_position_ids_from_input_ids": ["Model_position_ids_from_input_ids", "cumsum", "def", "dim", "incremental_indices", "input_ids", "int", "long", "mask", "ne", "padding_idx", "return", "torch", "type_as"], "luke/modeling_luke.py:LukeLMHead.__init__": ["LayerNorm", "Linear", "ModelLMHead", "Module", "Parameter", "__init__", "bias", "class", "config", "decoder", "def", "dense", "eps", "hidden_size", "layer_norm", "layer_norm_eps", "nn", "self", "super", "torch", "vocab_size", "zeros"], "luke/modeling_luke.py:LukeLMHead.forward": ["ModelLMHead", "Module", "class", "decoder", "def", "dense", "features", "forward", "gelu", "kwargs", "layer_norm", "nn", "return", "self", "x"], "luke/modeling_luke.py:LukeForMaskedLM.__init__": ["CrossEntropyLoss", "EntityPredictionHead", "Model", "ModelForMaskedLM", "ModelLMHead", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "config", "def", "entity_predictions", "lm_head", "loss_fn", "nn", "post_init", "self", "super"], "luke/modeling_luke.py:LukeForMaskedLM.get_output_embeddings": ["ModelForMaskedLM", "ModelPreTrainedModel", "class", "decoder", "def", "get_output_embeddings", "lm_head", "return", "self"], "luke/modeling_luke.py:LukeForMaskedLM.set_output_embeddings": ["ModelForMaskedLM", "ModelPreTrainedModel", "class", "decoder", "def", "lm_head", "new_embeddings", "self", "set_output_embeddings"], "luke/modeling_luke.py:LukeForMaskedLM.forward": ["FloatTensor", "LongTensor", "Model", "ModelForMaskedLM", "ModelMaskedLMOutput", "ModelPreTrainedModel", "None", "True", "attention_mask", "attentions", "bool", "class", "config", "def", "device", "else", "entity_attention_mask", "entity_hidden_states", "entity_ids", "entity_labels", "entity_last_hidden_state", "entity_logits", "entity_position_ids", "entity_predictions", "entity_token_type_ids", "entity_vocab_size", "for", "forward", "hidden_states", "if", "in", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "last_hidden_state", "lm_head", "logits", "loss", "loss_fn", "mep_loss", "mlm_loss", "not", "output_attentions", "output_hidden_states", "outputs", "position_ids", "r", "return", "return_dict", "self", "to", "token_type_ids", "torch", "tuple", "use_return_dict", "v", "view", "vocab_size"], "luke/modeling_luke.py:LukeForEntityClassification.__init__": ["Dropout", "Linear", "Model", "ModelForEntityClassification", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "classifier", "config", "def", "dropout", "hidden_dropout_prob", "hidden_size", "nn", "num_labels", "post_init", "self", "super"], "luke/modeling_luke.py:LukeForEntityClassification.forward": ["EntityClassificationOutput", "FloatTensor", "LongTensor", "Model", "ModelForEntityClassification", "ModelPreTrainedModel", "None", "True", "attention_mask", "attentions", "binary_cross_entropy_with_logits", "bool", "class", "classifier", "config", "cross_entropy", "def", "device", "dropout", "else", "entity_attention_mask", "entity_hidden_states", "entity_ids", "entity_last_hidden_state", "entity_position_ids", "entity_token_type_ids", "feature_vector", "for", "forward", "functional", "hidden_states", "if", "in", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "ndim", "nn", "not", "output_attentions", "output_hidden_states", "outputs", "position_ids", "r", "return", "return_dict", "self", "to", "token_type_ids", "torch", "tuple", "type_as", "use_return_dict", "v", "view"], "luke/modeling_luke.py:LukeForEntityPairClassification.__init__": ["Dropout", "False", "Linear", "Model", "ModelForEntityPairClassification", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "classifier", "config", "def", "dropout", "hidden_dropout_prob", "hidden_size", "nn", "num_labels", "post_init", "self", "super"], "luke/modeling_luke.py:LukeForEntityPairClassification.forward": ["EntityPairClassificationOutput", "FloatTensor", "LongTensor", "Model", "ModelForEntityPairClassification", "ModelPreTrainedModel", "None", "True", "attention_mask", "attentions", "binary_cross_entropy_with_logits", "bool", "cat", "class", "classifier", "config", "cross_entropy", "def", "device", "dim", "dropout", "else", "entity_attention_mask", "entity_hidden_states", "entity_ids", "entity_last_hidden_state", "entity_position_ids", "entity_token_type_ids", "feature_vector", "for", "forward", "functional", "hidden_states", "if", "in", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "ndim", "nn", "not", "output_attentions", "output_hidden_states", "outputs", "position_ids", "r", "return", "return_dict", "self", "to", "token_type_ids", "torch", "tuple", "type_as", "use_return_dict", "v", "view"], "luke/modeling_luke.py:LukeForEntitySpanClassification.__init__": ["Dropout", "Linear", "Model", "ModelForEntitySpanClassification", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "classifier", "config", "def", "dropout", "hidden_dropout_prob", "hidden_size", "nn", "num_labels", "post_init", "self", "super"], "luke/modeling_luke.py:LukeForEntitySpanClassification.forward": ["EntitySpanClassificationOutput", "FloatTensor", "LongTensor", "Model", "ModelForEntitySpanClassification", "ModelPreTrainedModel", "None", "True", "attention_mask", "attentions", "binary_cross_entropy_with_logits", "bool", "cat", "class", "classifier", "config", "cross_entropy", "def", "device", "dim", "dropout", "else", "end_states", "entity_attention_mask", "entity_end_positions", "entity_hidden_states", "entity_ids", "entity_last_hidden_state", "entity_position_ids", "entity_start_positions", "entity_token_type_ids", "expand", "feature_vector", "for", "forward", "functional", "gather", "hidden_size", "hidden_states", "if", "in", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "last_hidden_state", "logits", "loss", "ndim", "nn", "not", "num_labels", "output_attentions", "output_hidden_states", "outputs", "position_ids", "r", "return", "return_dict", "self", "size", "start_states", "to", "token_type_ids", "torch", "tuple", "type_as", "unsqueeze", "use_return_dict", "v", "view"], "luke/modeling_luke.py:LukeForSequenceClassification.__init__": ["Dropout", "Linear", "Model", "ModelForSequenceClassification", "ModelModel", "ModelPreTrainedModel", "None", "__init__", "class", "classifier", "classifier_dropout", "config", "def", "dropout", "else", "hidden_dropout_prob", "hidden_size", "if", "is", "nn", "not", "num_labels", "post_init", "self", "super"], "luke/modeling_luke.py:LukeForSequenceClassification.forward": ["BCEWithLogitsLoss", "CrossEntropyLoss", "FloatTensor", "LongTensor", "MSELoss", "Model", "ModelForSequenceClassification", "ModelPreTrainedModel", "ModelSequenceClassifierOutput", "None", "True", "and", "attention_mask", "attentions", "bool", "class", "classifier", "config", "def", "device", "dropout", "dtype", "elif", "else", "entity_attention_mask", "entity_hidden_states", "entity_ids", "entity_position_ids", "entity_token_type_ids", "for", "forward", "hidden_states", "if", "in", "input_ids", "inputs_embeds", "int", "is", "kwargs", "labels", "logits", "long", "loss", "loss_fct", "multi_label_classification", "not", "num_labels", "or", "output_attentions", "output_hidden_states", "outputs", "pooled_output", "pooler_output", "position_ids", "problem_type", "r", "regression", "return", "return_dict", "self", "single_label_classification", "squeeze", "to", "token_type_ids", "torch", "tuple", "use_return_dict", "v", "view"], "luke/modeling_luke.py:LukeForTokenClassification.__init__": ["Dropout", "False", "Linear", "Model", "ModelForTokenClassification", "ModelModel", "ModelPreTrainedModel", "None", "__init__", "add_pooling_layer", "class", "classifier", "classifier_dropout", "config", "def", "dropout", "else", "hidden_dropout_prob", "hidden_size", "if", "is", "nn", "not", "num_labels", "post_init", "self", "super"], "luke/modeling_luke.py:LukeForTokenClassification.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "Model", "ModelForTokenClassification", "ModelPreTrainedModel", "ModelTokenClassifierOutput", "None", "True", "attention_mask", "attentions", "bool", "class", "classifier", "config", "def", "device", "dropout", "else", "entity_attention_mask", "entity_hidden_states", "entity_ids", "entity_position_ids", "entity_token_type_ids", "for", "forward", "hidden_states", "if", "in", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "last_hidden_state", "logits", "loss", "loss_fct", "not", "num_labels", "output_attentions", "output_hidden_states", "outputs", "position_ids", "r", "return", "return_dict", "self", "sequence_output", "to", "token_type_ids", "torch", "tuple", "use_return_dict", "v", "view"], "luke/modeling_luke.py:LukeForQuestionAnswering.__init__": ["False", "Linear", "Model", "ModelForQuestionAnswering", "ModelModel", "ModelPreTrainedModel", "__init__", "add_pooling_layer", "class", "config", "def", "hidden_size", "nn", "num_labels", "post_init", "qa_outputs", "self", "super"], "luke/modeling_luke.py:LukeForQuestionAnswering.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "Model", "ModelForQuestionAnswering", "ModelPreTrainedModel", "ModelQuestionAnsweringModelOutput", "None", "True", "and", "attention_mask", "attentions", "bool", "clamp_", "class", "config", "def", "dim", "else", "end_logits", "end_loss", "end_positions", "entity_attention_mask", "entity_hidden_states", "entity_ids", "entity_position_ids", "entity_token_type_ids", "for", "forward", "hidden_states", "if", "ignore_index", "ignored_index", "in", "input_ids", "inputs_embeds", "is", "kwargs", "last_hidden_state", "len", "logits", "loss", "loss_fct", "not", "output_attentions", "output_hidden_states", "outputs", "position_ids", "qa_outputs", "r", "return", "return_dict", "self", "sequence_output", "size", "split", "squeeze", "start_logits", "start_loss", "start_positions", "token_type_ids", "torch", "total_loss", "tuple", "use_return_dict", "v"], "luke/modeling_luke.py:LukeForMultipleChoice.__init__": ["Dropout", "Linear", "Model", "ModelForMultipleChoice", "ModelModel", "ModelPreTrainedModel", "None", "__init__", "class", "classifier", "classifier_dropout", "config", "def", "dropout", "else", "hidden_dropout_prob", "hidden_size", "if", "is", "nn", "not", "post_init", "self", "super"], "luke/modeling_luke.py:LukeForMultipleChoice.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "Model", "ModelForMultipleChoice", "ModelMultipleChoiceModelOutput", "ModelPreTrainedModel", "None", "True", "attention_mask", "attentions", "bool", "class", "classifier", "config", "def", "device", "dropout", "else", "entity_attention_mask", "entity_hidden_states", "entity_ids", "entity_position_ids", "entity_token_type_ids", "for", "forward", "hidden_states", "if", "in", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "not", "num_choices", "output_attentions", "output_hidden_states", "outputs", "pooled_output", "pooler_output", "position_ids", "r", "reshaped_logits", "return", "return_dict", "self", "shape", "size", "to", "token_type_ids", "torch", "tuple", "use_return_dict", "v", "view"], "lw_detr/modeling_lw_detr.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value", "value_states"], "lw_detr/modeling_lw_detr.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "lw_detr/modeling_lw_detr.py:LwDetrViTSelfAttention.__init__": ["False", "Linear", "ModelViTConfig", "ModelViTSelfAttention", "Module", "The", "ValueError", "__init__", "a", "all_head_size", "and", "attention", "attention_head_size", "bias", "class", "config", "def", "dropout_prob", "embedding_size", "f", "hasattr", "heads", "hidden", "hidden_size", "if", "int", "is", "is_causal", "key", "multiple", "nn", "not", "num_attention_heads", "num_key_value_groups", "number", "of", "qkv_bias", "query", "raise", "scaling", "self", "size", "super", "the", "value"], "lw_detr/modeling_lw_detr.py:LwDetrViTSelfAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "ModelViTSelfAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_attn_implementation", "all_head_size", "attention_head_size", "attention_interface", "attention_probs", "batch_size", "class", "config", "context_layer", "def", "dropout", "dropout_prob", "eager", "eager_attention_forward", "else", "forward", "hidden_states", "if", "is_causal", "key", "key_layer", "kwargs", "new_context_layer_shape", "new_shape", "nn", "not", "num_attention_heads", "query", "query_layer", "reshape", "return", "scaling", "self", "shape", "size", "torch", "training", "transpose", "tuple", "value", "value_layer", "view"], "lw_detr/modeling_lw_detr.py:LwDetrViTAttention.__init__": ["Linear", "ModelViTAttention", "ModelViTConfig", "ModelViTSelfAttention", "Module", "__init__", "attention", "class", "config", "def", "hidden_size", "nn", "output", "self", "super"], "lw_detr/modeling_lw_detr.py:LwDetrViTAttention.forward": ["ModelViTAttention", "Module", "Tensor", "TransformersKwargs", "Unpack", "_", "attention", "class", "def", "forward", "hidden_states", "kwargs", "nn", "output", "return", "self", "self_attn_output", "torch"], "lw_detr/modeling_lw_detr.py:LwDetrViTMlp.__init__": ["ACT2FN", "Dropout", "Linear", "ModelViTMlp", "Module", "None", "__init__", "act", "class", "config", "def", "drop", "dropout_prob", "fc1", "fc2", "hidden_act", "hidden_features", "in_features", "int", "nn", "self", "super"], "lw_detr/modeling_lw_detr.py:LwDetrViTMlp.forward": ["ModelViTMlp", "Module", "Tensor", "act", "class", "def", "drop", "fc1", "fc2", "forward", "nn", "return", "self", "torch", "x"], "lw_detr/modeling_lw_detr.py:LwDetrViTLayer.__init__": ["GradientCheckpointingLayer", "LayerNorm", "ModelViTAttention", "ModelViTConfig", "ModelViTLayer", "ModelViTMlp", "None", "Parameter", "Tensor", "True", "__init__", "attention", "class", "config", "def", "dim", "eps", "gamma_1", "gamma_2", "hidden_features", "hidden_size", "in", "in_features", "int", "intermediate", "layer_idx", "layer_norm_eps", "layernorm_after", "layernorm_before", "mlp_ratio", "nn", "num_windows", "requires_grad", "self", "super", "torch", "window", "window_block_indices"], "lw_detr/modeling_lw_detr.py:LwDetrViTLayer.forward": ["GradientCheckpointingLayer", "ModelViTLayer", "Tensor", "TransformersKwargs", "Unpack", "attention", "attention_output", "batch_size", "channels", "class", "def", "forward", "gamma_1", "gamma_2", "hidden_states", "hidden_states_norm", "if", "intermediate", "kwargs", "layer_output", "layernorm_after", "layernorm_before", "not", "num_windows", "reshape", "return", "self", "seq_len", "shape", "torch", "window"], "lw_detr/modeling_lw_detr.py:LwDetrViTEncoder.__init__": ["False", "ModelViTConfig", "ModelViTEncoder", "ModelViTLayer", "Module", "ModuleList", "None", "__init__", "class", "config", "def", "for", "gradient_checkpointing", "i", "in", "layer", "nn", "num_hidden_layers", "range", "self", "super"], "lw_detr/modeling_lw_detr.py:LwDetrViTEncoder.forward": ["ModelViTEncoder", "Module", "Tensor", "TransformersKwargs", "Unpack", "append", "class", "def", "enumerate", "for", "forward", "hidden_states", "i", "in", "kwargs", "layer", "layer_module", "list", "list_hidden_states", "nn", "return", "self", "torch"], "lw_detr/modeling_lw_detr.py:LwDetrViTEmbeddings.__init__": ["Conv2d", "Iterable", "ModelViTEmbeddings", "Module", "None", "Parameter", "__init__", "abc", "class", "collections", "config", "def", "else", "hidden_size", "if", "image_size", "isinstance", "kernel_size", "nn", "num_channels", "num_patches", "num_positions", "patch_size", "position_embeddings", "pretrain_image_size", "projection", "self", "stride", "super", "torch", "use_absolute_position_embeddings", "zeros"], "lw_detr/modeling_lw_detr.py:LwDetrViTEmbeddings.get_absolute_positions": ["Absolute", "False", "ModelViTEmbeddings", "Module", "ValueError", "a", "abs_pos_embeddings", "align_corners", "be", "bicubic", "class", "def", "else", "embeddings", "functional", "get_absolute_positions", "has_cls_token", "height", "if", "int", "interpolate", "is_tracing", "jit", "math", "mode", "must", "new_abs_pos_embeddings", "nn", "num_position", "number", "or", "permute", "position", "raise", "reshape", "return", "self", "shape", "size", "sqrt", "square", "torch", "width"], "lw_detr/modeling_lw_detr.py:LwDetrViTEmbeddings.forward": ["Expected", "Make", "ModelViTEmbeddings", "Module", "None", "Tensor", "True", "ValueError", "but", "channel", "class", "configuration", "def", "dimension", "embeddings", "f", "forward", "get_absolute_positions", "got", "if", "in", "is", "match", "nn", "not", "num_channels", "of", "one", "permute", "pixel", "pixel_values", "position_embeddings", "projection", "raise", "return", "self", "set", "shape", "sure", "that", "the", "torch", "values", "with"], "lw_detr/modeling_lw_detr.py:LwDetrViTPreTrainedModel._init_weights": ["Conv2d", "LayerNorm", "Linear", "ModelViTEmbeddings", "ModelViTLayer", "ModelViTPreTrainedModel", "None", "PreTrainedModel", "_init_weights", "bias", "cae_init_values", "class", "config", "constant_", "def", "elif", "gamma_1", "gamma_2", "if", "init", "initializer_range", "is", "isinstance", "mean", "module", "nn", "not", "ones_", "position_embeddings", "self", "std", "trunc_normal_", "weight", "zeros_"], "lw_detr/modeling_lw_detr.py:LwDetrViTBackbone.__init__": ["BackboneMixin", "ModelViTBackbone", "ModelViTEmbeddings", "ModelViTEncoder", "ModelViTPreTrainedModel", "_", "__init__", "_init_backbone", "class", "config", "def", "embeddings", "encoder", "for", "hidden_size", "in", "num_features", "num_hidden_layers", "post_init", "range", "self", "super"], "lw_detr/modeling_lw_detr.py:LwDetrViTBackbone.get_input_embeddings": ["BackboneMixin", "ModelViTBackbone", "ModelViTEmbeddings", "ModelViTPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "projection", "return", "self"], "lw_detr/modeling_lw_detr.py:LwDetrViTBackbone.forward": ["BackboneMixin", "BackboneOutput", "ModelViTBackbone", "ModelViTPreTrainedModel", "Tensor", "TransformersKwargs", "Unpack", "batch_size", "channels", "class", "config", "def", "embedding_output", "embeddings", "encoder", "feature_maps", "for", "forward", "height", "hidden_state", "hidden_states", "if", "in", "kwargs", "num_windows_side", "out_features", "permute", "pixel_values", "r", "reshape", "return", "self", "shape", "stage", "stage_names", "torch", "width", "window_height", "window_width", "zip"], "lw_detr/modeling_lw_detr.py:LwDetrConvNormLayer.__init__": ["ACT2CLS", "BatchNorm2d", "Conv2d", "False", "Identity", "ModelConfig", "ModelConvNormLayer", "Module", "None", "__init__", "activation", "batch_norm_eps", "bias", "class", "config", "conv", "def", "else", "if", "in_channels", "int", "is", "kernel_size", "nn", "norm", "out_channels", "padding", "self", "str", "stride", "super"], "lw_detr/modeling_lw_detr.py:LwDetrConvNormLayer.forward": ["ModelConvNormLayer", "Module", "activation", "class", "conv", "def", "forward", "hidden_state", "nn", "norm", "return", "self"], "lw_detr/modeling_lw_detr.py:LwDetrRepVggBlock.__init__": ["ModelConfig", "ModelConvNormLayer", "ModelRepVggBlock", "Module", "__init__", "activation", "activation_function", "class", "config", "conv1", "conv2", "d_model", "def", "hidden_channels", "hidden_expansion", "int", "nn", "self", "super"], "lw_detr/modeling_lw_detr.py:LwDetrRepVggBlock.forward": ["ModelRepVggBlock", "Module", "Tensor", "class", "conv1", "conv2", "def", "forward", "nn", "return", "self", "torch", "x", "y"], "lw_detr/modeling_lw_detr.py:LwDetrC2FLayer.__init__": ["ModelC2FLayer", "ModelConfig", "ModelConvNormLayer", "ModelRepVggBlock", "Module", "ModuleList", "_", "__init__", "activation", "activation_function", "bottlenecks", "c2f_num_blocks", "class", "config", "conv1", "conv1_out_channels", "conv2", "conv2_in_channels", "d_model", "def", "for", "hidden_channels", "hidden_expansion", "in", "in_channels", "int", "nn", "num_blocks", "out_channels", "range", "self", "super"], "lw_detr/modeling_lw_detr.py:LwDetrC2FLayer.forward": ["ModelC2FLayer", "Module", "Tensor", "all_hidden_states", "append", "bottleneck", "bottlenecks", "cat", "class", "conv1", "conv2", "def", "for", "forward", "hidden_channels", "hidden_states", "in", "list", "nn", "return", "self", "split", "torch"], "lw_detr/modeling_lw_detr.py:LwDetrLayerNorm.__init__": ["LayerNorm", "ModelLayerNorm", "NotImplementedError", "Unsupported", "__init__", "channels_first", "channels_last", "class", "data", "data_format", "def", "eps", "f", "format", "if", "in", "kwargs", "nn", "normalized_shape", "not", "raise", "self", "super"], "lw_detr/modeling_lw_detr.py:LwDetrLayerNorm.forward": ["LayerNorm", "ModelLayerNorm", "Tensor", "channels_first", "class", "data_format", "def", "else", "features", "forward", "if", "nn", "permute", "return", "self", "super", "torch"], "lw_detr/modeling_lw_detr.py:LwDetrSamplingLayer.__init__": ["ConvTranspose2d", "ModelConfig", "ModelConvNormLayer", "ModelSamplingLayer", "Module", "ModuleList", "__init__", "activation", "append", "channel_size", "class", "config", "def", "elif", "else", "float", "if", "int", "kernel_size", "layers", "nn", "relu", "scale", "self", "stride", "super"], "lw_detr/modeling_lw_detr.py:LwDetrSamplingLayer.forward": ["ModelSamplingLayer", "Module", "Tensor", "class", "def", "for", "forward", "hidden_states", "in", "layer", "layers", "nn", "return", "self", "torch"], "lw_detr/modeling_lw_detr.py:LwDetrScaleProjector.__init__": ["ModelC2FLayer", "ModelConfig", "ModelLayerNorm", "ModelSamplingLayer", "ModelScaleProjector", "Module", "ModuleList", "__init__", "append", "backbone_config", "channel_size", "channels_first", "class", "config", "d_model", "data_format", "def", "else", "float", "for", "hidden_size", "if", "in", "intermediate_dim", "intermediate_dims", "layer_norm", "len", "nn", "out_indices", "projector_input_dim", "projector_layer", "sampling_layers", "scale", "self", "super"], "lw_detr/modeling_lw_detr.py:LwDetrScaleProjector.forward": ["ModelScaleProjector", "Module", "Tensor", "append", "cat", "class", "def", "dim", "for", "forward", "hidden_states", "hidden_states_tuple", "in", "layer_norm", "nn", "projector_layer", "return", "sampled_hidden_states", "sampling_layer", "sampling_layers", "self", "torch", "tuple", "zip"], "lw_detr/modeling_lw_detr.py:LwDetrMultiScaleProjector.__init__": ["ModelConfig", "ModelMultiScaleProjector", "ModelScaleProjector", "Module", "ModuleList", "__init__", "class", "config", "def", "for", "in", "nn", "projector_scale_factors", "scale", "scale_factors", "scale_layers", "self", "super"], "lw_detr/modeling_lw_detr.py:LwDetrMultiScaleProjector.forward": ["ModelMultiScaleProjector", "Module", "Tensor", "append", "class", "def", "for", "forward", "hidden_states", "in", "list", "nn", "output_hidden_states", "return", "scale_layer", "scale_layers", "self", "torch", "tuple"], "lw_detr/modeling_lw_detr.py:LwDetrConvEncoder.__init__": ["ModelConfig", "ModelConvEncoder", "ModelMultiScaleProjector", "ModelViTBackbone", "Module", "__init__", "backbone", "backbone_config", "class", "config", "def", "nn", "projector", "self", "super"], "lw_detr/modeling_lw_detr.py:LwDetrConvEncoder.forward": ["ModelConvEncoder", "Module", "None", "Tensor", "append", "backbone", "bool", "class", "def", "feature_map", "feature_maps", "features", "float", "for", "forward", "functional", "in", "interpolate", "mask", "nn", "out", "pixel_mask", "pixel_values", "projector", "return", "self", "shape", "size", "to", "torch"], "lw_detr/modeling_lw_detr.py:LwDetrAttention.__init__": ["False", "Linear", "ModelAttention", "ModelConfig", "Module", "__init__", "attention_bias", "attention_dropout", "bias", "class", "config", "d_model", "decoder_self_attention_heads", "def", "getattr", "head_dim", "int", "is_causal", "k_proj", "layer_idx", "nn", "num_key_value_groups", "o_proj", "q_proj", "scaling", "self", "super", "v_proj"], "lw_detr/modeling_lw_detr.py:LwDetrAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "ModelAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "_attn_implementation", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "batch_size", "cat", "class", "config", "contiguous", "def", "dim", "dropout", "eager", "eager_attention_forward", "else", "forward", "group_detr", "head_dim", "hidden_shape", "hidden_states", "hidden_states_original", "if", "input_shape", "is", "k_proj", "key_states", "kwargs", "nn", "not", "o_proj", "position_embeddings", "q_proj", "query_states", "reshape", "return", "scaling", "self", "seq_len", "shape", "split", "torch", "training", "transpose", "tuple", "v_proj", "value_states", "view"], "lw_detr/modeling_lw_detr.py:MultiScaleDeformableAttention.forward": ["False", "ModelScaleDeformableAttention", "Module", "Tensor", "_", "align_corners", "append", "attention_weights", "batch_size", "bilinear", "class", "contiguous", "def", "dim", "enumerate", "flatten", "for", "forward", "functional", "grid_sample", "height", "hidden_dim", "im2col_step", "in", "int", "level_id", "level_start_index", "list", "mode", "nn", "num_heads", "num_levels", "num_points", "num_queries", "output", "padding_mode", "reshape", "return", "sampling_grid_l_", "sampling_grids", "sampling_locations", "sampling_value_l_", "sampling_value_list", "self", "shape", "split", "stack", "sum", "torch", "transpose", "tuple", "value", "value_l_", "value_list", "value_spatial_shapes", "value_spatial_shapes_list", "view", "width", "zeros"], "lw_detr/modeling_lw_detr.py:LwDetrMultiscaleDeformableAttention.__init__": ["CUDA", "Linear", "ModelConfig", "ModelMultiscaleDeformableAttention", "Module", "MultiScaleDeformableAttention", "ValueError", "You", "__init__", "a", "and", "attention", "attention_weights", "attn", "authors", "be", "better", "but", "by", "class", "config", "d", "d_model", "def", "dim_per_head", "dimension", "disable_custom_kernels", "divisible", "each", "efficient", "embed_dim", "f", "got", "head", "if", "im2col_step", "implementation", "in", "int", "is", "make", "more", "must", "n_heads", "n_levels", "n_points", "nn", "not", "num_feature_levels", "num_heads", "of", "output_proj", "power", "raise", "sampling_offsets", "self", "set", "super", "the", "to", "value_proj", "warn", "warnings", "which"], "lw_detr/modeling_lw_detr.py:LwDetrMultiscaleDeformableAttention.with_pos_embed": ["ModelMultiscaleDeformableAttention", "Module", "None", "Tensor", "class", "def", "else", "if", "is", "nn", "position_embeddings", "return", "self", "tensor", "torch", "with_pos_embed"], "lw_detr/modeling_lw_detr.py:LwDetrMultiscaleDeformableAttention.forward": ["F", "Last", "Make", "ModelMultiscaleDeformableAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "_", "align", "attention_mask", "attention_weights", "attn", "batch_size", "be", "but", "class", "d_model", "def", "dim", "elif", "else", "encoder", "encoder_attention_mask", "encoder_hidden_states", "f", "float", "for", "forward", "got", "height", "hidden", "hidden_states", "if", "im2col_step", "in", "is", "kwargs", "length", "level_start_index", "masked_fill", "must", "n_heads", "n_levels", "n_points", "nn", "not", "num_coordinates", "num_queries", "of", "offset_normalizer", "or", "output", "output_proj", "position_embeddings", "raise", "reference_points", "return", "sampling_locations", "sampling_offsets", "self", "sequence", "sequence_length", "shape", "shapes", "softmax", "spatial", "spatial_shapes", "spatial_shapes_list", "stack", "states", "sum", "sure", "the", "to", "torch", "total_elements", "value", "value_proj", "view", "width", "with", "with_pos_embed"], "lw_detr/modeling_lw_detr.py:LwDetrMLP.__init__": ["ACT2FN", "Linear", "ModelConfig", "ModelMLP", "Module", "__init__", "activation_fn", "class", "config", "d_model", "decoder_activation_function", "decoder_ffn_dim", "def", "dropout", "fc1", "fc2", "nn", "self", "super"], "lw_detr/modeling_lw_detr.py:LwDetrMLP.forward": ["ModelMLP", "Module", "Tensor", "activation_fn", "class", "def", "dropout", "fc1", "fc2", "forward", "functional", "hidden_states", "nn", "p", "residual", "return", "self", "torch", "training"], "lw_detr/modeling_lw_detr.py:LwDetrDecoderLayer.__init__": ["ACT2FN", "GradientCheckpointingLayer", "LayerNorm", "ModelAttention", "ModelConfig", "ModelDecoderLayer", "ModelMLP", "ModelMultiscaleDeformableAttention", "Module", "__init__", "activation_dropout", "activation_fn", "class", "config", "cross_attn", "cross_attn_layer_norm", "d_model", "decoder_activation_function", "decoder_cross_attention_heads", "decoder_n_points", "def", "dropout", "int", "layer_idx", "layer_norm", "mlp", "n_points", "nn", "num_heads", "self", "self_attn", "self_attn_layer_norm"], "lw_detr/modeling_lw_detr.py:LwDetrDecoderLayer.forward": ["GradientCheckpointingLayer", "ModelDecoderLayer", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "class", "cross_attention_output", "cross_attn", "cross_attn_layer_norm", "cross_attn_weights", "def", "dropout", "encoder_attention_mask", "encoder_hidden_states", "forward", "functional", "hidden_states", "kwargs", "layer_norm", "level_start_index", "mlp", "nn", "p", "position_embeddings", "reference_points", "return", "self", "self_attention_output", "self_attn", "self_attn_layer_norm", "self_attn_weights", "spatial_shapes", "spatial_shapes_list", "torch", "training"], "lw_detr/modeling_lw_detr.py:LwDetrPreTrainedModel._init_weights": ["ModelMultiscaleDeformableAttention", "ModelPreTrainedModel", "None", "PreTrainedModel", "True", "_init_weights", "abs", "and", "arange", "attention_weights", "bbox_embed", "bias", "bias_value", "class", "class_embed", "constant_", "copy_", "cos", "def", "dtype", "float", "for", "grid_init", "hasattr", "i", "if", "in", "init", "int64", "is", "isinstance", "keepdim", "layers", "level_embed", "log", "math", "max", "module", "n_heads", "n_levels", "n_points", "normal_", "not", "output_proj", "pi", "prior_prob", "range", "refpoint_embed", "repeat", "sampling_offsets", "self", "sin", "stack", "super", "thetas", "torch", "value_proj", "view", "weight", "xavier_uniform_"], "lw_detr/modeling_lw_detr.py:gen_sine_position_embeddings": ["Model_sine_position_embeddings", "None", "Unknown", "ValueError", "arange", "cat", "cos", "def", "device", "dim", "dim_t", "div", "dtype", "else", "f", "flatten", "float32", "floor", "h_embed", "hidden_size", "if", "math", "pi", "pos", "pos_h", "pos_tensor", "pos_w", "pos_x", "pos_y", "raise", "return", "rounding_mode", "scale", "shape", "sin", "size", "stack", "to", "torch", "w_embed", "x_embed", "y_embed"], "lw_detr/modeling_lw_detr.py:LwDetrDecoder.__init__": ["False", "LayerNorm", "ModelConfig", "ModelDecoder", "ModelDecoderLayer", "ModelMLPPredictionHead", "ModelPreTrainedModel", "ModuleList", "__init__", "class", "config", "d_model", "decoder_layers", "def", "dropout", "for", "gradient_checkpointing", "i", "in", "layernorm", "layers", "nn", "num_layers", "post_init", "range", "ref_point_head", "self", "super"], "lw_detr/modeling_lw_detr.py:LwDetrDecoder.get_reference": ["ModelDecoder", "ModelPreTrainedModel", "None", "cat", "class", "config", "d_model", "def", "gen_sine_position_embeddings", "get_reference", "obj_center", "query_pos", "query_sine_embed", "ref_point_head", "reference_points", "reference_points_inputs", "return", "self", "torch", "valid_ratios"], "lw_detr/modeling_lw_detr.py:LwDetrDecoder.forward": ["ModelDecoder", "ModelDecoderOutput", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "class", "decoder_layer", "def", "encoder_attention_mask", "encoder_hidden_states", "enumerate", "for", "forward", "get_reference", "hidden_states", "idx", "if", "in", "inputs_embeds", "intermediate", "intermediate_hidden_states", "intermediate_reference_points", "is", "kwargs", "last_hidden_state", "layernorm", "layers", "level_start_index", "not", "position_embeddings", "query_pos", "reference_points", "reference_points_inputs", "return", "self", "spatial_shapes", "spatial_shapes_list", "stack", "torch", "valid_ratios"], "lw_detr/modeling_lw_detr.py:refine_bboxes": ["Model_bboxes", "cat", "def", "deltas", "device", "exp", "new_reference_points", "new_reference_points_cxcy", "new_reference_points_wh", "reference_points", "return", "to", "torch"], "lw_detr/modeling_lw_detr.py:LwDetrModel.__init__": ["Embedding", "LayerNorm", "Linear", "ModelConfig", "ModelConvEncoder", "ModelDecoder", "ModelMLPPredictionHead", "ModelModel", "ModelPreTrainedModel", "ModuleList", "_", "__init__", "backbone", "class", "config", "d_model", "decoder", "def", "enc_out_bbox_embed", "enc_out_class_embed", "enc_output", "enc_output_norm", "for", "group_detr", "hidden_dim", "in", "nn", "num_labels", "num_layers", "num_queries", "post_init", "query_feat", "range", "reference_point_embed", "self", "super"], "lw_detr/modeling_lw_detr.py:LwDetrModel.freeze_backbone": ["False", "ModelModel", "ModelPreTrainedModel", "backbone", "class", "conv_encoder", "def", "for", "freeze_backbone", "in", "model", "name", "named_parameters", "param", "requires_grad_", "self"], "lw_detr/modeling_lw_detr.py:LwDetrModel.unfreeze_backbone": ["ModelModel", "ModelPreTrainedModel", "True", "backbone", "class", "conv_encoder", "def", "for", "in", "model", "name", "named_parameters", "param", "requires_grad_", "self", "unfreeze_backbone"], "lw_detr/modeling_lw_detr.py:LwDetrModel.get_valid_ratio": ["ModelModel", "ModelPreTrainedModel", "_", "class", "def", "dtype", "float32", "get_valid_ratio", "height", "mask", "return", "self", "shape", "stack", "sum", "to", "torch", "valid_height", "valid_ratio", "valid_ratio_height", "valid_ratio_width", "valid_width", "width"], "lw_detr/modeling_lw_detr.py:LwDetrModel.get_proposal_pos_embed": ["ModelModel", "ModelPreTrainedModel", "None", "arange", "class", "config", "cos", "d_model", "def", "device", "dim", "dim_t", "div", "dtype", "flatten", "floor", "get_proposal_pos_embed", "math", "num_pos_feats", "pi", "pos", "proposals", "return", "rounding_mode", "scale", "self", "sigmoid", "sin", "stack", "temperature", "torch"], "lw_detr/modeling_lw_detr.py:LwDetrModel.gen_encoder_output_proposals": ["ModelModel", "ModelPreTrainedModel", "True", "_cur", "all", "append", "batch_size", "cat", "class", "def", "device", "dtype", "enc_output", "enumerate", "expand", "float", "for", "gen_encoder_output_proposals", "grid", "grid_x", "grid_y", "height", "ij", "in", "indexing", "inf", "keepdim", "level", "linspace", "mask_flatten_", "masked_fill", "meshgrid", "object_query", "ones_like", "output_proposals", "output_proposals_valid", "padding_mask", "proposal", "proposals", "return", "scale", "self", "shape", "spatial_shapes", "sum", "torch", "unsqueeze", "valid_height", "valid_width", "view", "width", "width_height"], "lw_detr/modeling_lw_detr.py:LwDetrModel.forward": ["FloatTensor", "LongTensor", "ModelModel", "ModelModelOutput", "ModelPreTrainedModel", "No", "None", "TransformersKwargs", "Unpack", "ValueError", "append", "as_tensor", "attention", "backbone", "batch_size", "cat", "class", "config", "cumsum", "d_model", "decoder", "decoder_outputs", "def", "detach", "device", "dim", "dtype", "else", "enc_out_bbox_embed", "enc_out_class_embed", "enc_output", "enc_output_norm", "enc_outputs_class", "enc_outputs_coord_logits", "encoder_attention_mask", "encoder_hidden_states", "enumerate", "expand", "features", "flatten", "for", "forward", "gather", "gen_encoder_output_proposals", "get_valid_ratio", "group_delta_bbox", "group_detr", "group_enc_outputs_class", "group_enc_outputs_coord", "group_id", "group_object_query", "group_object_query_undetach", "group_topk_coords_logits", "group_topk_coords_logits_undetach", "group_topk_proposals", "height", "if", "in", "init_reference_points", "inputs_embeds", "intermediate_hidden_states", "intermediate_reference_points", "is", "kwargs", "last_hidden_state", "level", "level_start_index", "long", "m", "mask", "mask_flatten", "masks", "max", "new_zeros", "num_channels", "num_queries", "object_query_embedding", "object_query_undetach", "ones", "output_proposals", "pixel_mask", "pixel_values", "prod", "provided", "query_feat", "r", "raise", "range", "reference_point_embed", "reference_points", "refine_bboxes", "repeat", "return", "self", "shape", "source", "source_flatten", "sources", "spatial_shape", "spatial_shapes", "spatial_shapes_list", "stack", "target", "topk", "topk_coords_logits", "topk_coords_logits_undetach", "torch", "training", "transpose", "unsqueeze", "valid_ratios", "was", "weight", "width", "zip"], "lw_detr/modeling_lw_detr.py:LwDetrMLPPredictionHead.__init__": ["Linear", "ModelMLPPredictionHead", "Module", "ModuleList", "__init__", "class", "def", "for", "h", "hidden_dim", "in", "input_dim", "k", "layers", "n", "nn", "num_layers", "output_dim", "self", "super", "zip"], "lw_detr/modeling_lw_detr.py:LwDetrMLPPredictionHead.forward": ["ModelMLPPredictionHead", "Module", "class", "def", "else", "enumerate", "for", "forward", "functional", "i", "if", "in", "layer", "layers", "nn", "num_layers", "relu", "return", "self", "x"], "lw_detr/modeling_lw_detr.py:LwDetrForObjectDetection.__init__": ["Linear", "ModelConfig", "ModelForObjectDetection", "ModelMLPPredictionHead", "ModelModel", "ModelPreTrainedModel", "__init__", "bbox_embed", "class", "class_embed", "config", "d_model", "def", "model", "nn", "num_labels", "num_layers", "post_init", "self", "super"], "lw_detr/modeling_lw_detr.py:LwDetrForObjectDetection.forward": ["FloatTensor", "LongTensor", "ModelForObjectDetection", "ModelObjectDetectionOutput", "ModelPreTrainedModel", "None", "TransformersKwargs", "Unpack", "append", "auxiliary_loss", "auxiliary_outputs", "bbox_embed", "cat", "class", "class_embed", "config", "def", "device", "dict", "dim", "else", "enc_out_class_embed", "enc_outputs_boxes_logits", "enc_outputs_class", "enc_outputs_class_logits", "enc_outputs_class_logits_list", "enc_outputs_coord_logits", "for", "forward", "group_detr", "group_index", "group_pred_class", "if", "in", "init_reference_points", "intermediate_hidden_states", "intermediate_reference_points", "is", "kwargs", "labels", "last_hidden_state", "last_hidden_states", "list", "logits", "loss", "loss_dict", "loss_function", "model", "not", "num_queries", "outputs", "outputs_class", "outputs_coord", "outputs_coord_delta", "pixel_mask", "pixel_values", "pred_boxes", "pred_boxes_delta", "pred_class", "r", "range", "refine_bboxes", "return", "self", "split", "torch", "training"], "lxmert/modeling_lxmert.py:GeLU.__init__": ["GeLU", "Module", "__init__", "class", "def", "nn", "self", "super"], "lxmert/modeling_lxmert.py:GeLU.forward": ["GeLU", "Module", "class", "def", "forward", "gelu", "nn", "return", "self", "x"], "lxmert/modeling_lxmert.py:LxmertEmbeddings.__init__": ["Dropout", "Embedding", "LayerNorm", "ModelEmbeddings", "Module", "__init__", "class", "config", "def", "dropout", "eps", "hidden_dropout_prob", "hidden_size", "max_position_embeddings", "nn", "padding_idx", "position_embeddings", "self", "super", "token_type_embeddings", "type_vocab_size", "vocab_size", "word_embeddings"], "lxmert/modeling_lxmert.py:LxmertEmbeddings.forward": ["LayerNorm", "ModelEmbeddings", "Module", "None", "arange", "class", "def", "device", "dropout", "dtype", "else", "embeddings", "expand", "forward", "if", "input_ids", "input_shape", "inputs_embeds", "is", "long", "nn", "not", "position_embeddings", "position_ids", "return", "self", "seq_length", "size", "token_type_embeddings", "token_type_ids", "torch", "unsqueeze", "word_embeddings", "zeros"], "lxmert/modeling_lxmert.py:LxmertAttention.__init__": ["Dropout", "Linear", "ModelAttention", "Module", "None", "The", "ValueError", "__init__", "a", "attention", "attention_head_size", "attention_probs_dropout_prob", "class", "config", "ctx_dim", "def", "dropout", "f", "head_size", "heads", "hidden", "hidden_size", "if", "int", "is", "key", "multiple", "nn", "not", "num_attention_heads", "number", "of", "query", "raise", "self", "size", "super", "the", "value"], "lxmert/modeling_lxmert.py:LxmertAttention.forward": ["False", "ModelAttention", "Module", "None", "_", "attention_head_size", "attention_mask", "attention_probs", "attention_scores", "batch_size", "class", "context", "context_layer", "contiguous", "def", "dim", "dropout", "else", "forward", "functional", "head_size", "hidden_states", "if", "is", "key", "key_layer", "math", "matmul", "new_context_layer_shape", "nn", "not", "num_attention_heads", "output_attentions", "outputs", "permute", "query", "query_layer", "return", "self", "seq_length", "shape", "size", "softmax", "sqrt", "torch", "transpose", "value", "value_layer", "view"], "lxmert/modeling_lxmert.py:LxmertAttentionOutput.__init__": ["Dropout", "LayerNorm", "Linear", "ModelAttentionOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "eps", "hidden_dropout_prob", "hidden_size", "nn", "self", "super"], "lxmert/modeling_lxmert.py:LxmertAttentionOutput.forward": ["LayerNorm", "ModelAttentionOutput", "Module", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self"], "lxmert/modeling_lxmert.py:LxmertCrossAttentionLayer.__init__": ["ModelAttention", "ModelAttentionOutput", "ModelCrossAttentionLayer", "Module", "__init__", "att", "class", "config", "def", "nn", "output", "self", "super"], "lxmert/modeling_lxmert.py:LxmertCrossAttentionLayer.forward": ["False", "ModelCrossAttentionLayer", "Module", "None", "att", "attention_output", "attention_probs", "class", "ctx_att_mask", "ctx_tensor", "def", "else", "forward", "if", "input_tensor", "nn", "output", "output_attentions", "outputs", "return", "self"], "lxmert/modeling_lxmert.py:LxmertSelfAttentionLayer.__init__": ["ModelAttention", "ModelAttentionOutput", "ModelSelfAttentionLayer", "Module", "__init__", "class", "config", "def", "nn", "output", "self", "super"], "lxmert/modeling_lxmert.py:LxmertSelfAttentionLayer.forward": ["False", "ModelSelfAttentionLayer", "Module", "attention_mask", "attention_output", "attention_probs", "class", "def", "else", "forward", "if", "input_tensor", "nn", "output", "output_attentions", "outputs", "return", "self"], "lxmert/modeling_lxmert.py:LxmertIntermediate.__init__": ["ACT2FN", "Linear", "ModelIntermediate", "Module", "__init__", "class", "config", "def", "dense", "hidden_act", "hidden_size", "intermediate_act_fn", "intermediate_size", "nn", "self", "super"], "lxmert/modeling_lxmert.py:LxmertIntermediate.forward": ["ModelIntermediate", "Module", "class", "def", "dense", "forward", "hidden_states", "intermediate_act_fn", "nn", "return", "self"], "lxmert/modeling_lxmert.py:LxmertOutput.__init__": ["Dropout", "LayerNorm", "Linear", "ModelOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "eps", "hidden_dropout_prob", "hidden_size", "intermediate_size", "nn", "self", "super"], "lxmert/modeling_lxmert.py:LxmertOutput.forward": ["LayerNorm", "ModelOutput", "Module", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self"], "lxmert/modeling_lxmert.py:LxmertLayer.__init__": ["ModelIntermediate", "ModelLayer", "ModelOutput", "ModelSelfAttentionLayer", "Module", "__init__", "attention", "class", "config", "def", "intermediate", "nn", "output", "self", "super"], "lxmert/modeling_lxmert.py:LxmertLayer.forward": ["False", "ModelLayer", "Module", "None", "attention", "attention_mask", "attention_output", "class", "def", "forward", "hidden_states", "intermediate", "intermediate_output", "layer_output", "nn", "output", "output_attentions", "outputs", "return", "self"], "lxmert/modeling_lxmert.py:LxmertXLayer.__init__": ["ModelCrossAttentionLayer", "ModelIntermediate", "ModelOutput", "ModelSelfAttentionLayer", "ModelXLayer", "Module", "__init__", "class", "config", "def", "lang_inter", "lang_output", "lang_self_att", "nn", "self", "super", "visn_inter", "visn_output", "visn_self_att", "visual_attention"], "lxmert/modeling_lxmert.py:LxmertXLayer.cross_att": ["False", "ModelXLayer", "Module", "class", "cross_att", "ctx_att_mask", "def", "lang_att_output", "lang_attention_mask", "lang_input", "nn", "output_attentions", "output_x_attentions", "return", "self", "visual_att_output", "visual_attention", "visual_attention_mask", "visual_input"], "lxmert/modeling_lxmert.py:LxmertXLayer.self_att": ["False", "ModelXLayer", "Module", "class", "def", "lang_att_output", "lang_attention_mask", "lang_input", "lang_self_att", "nn", "output_attentions", "return", "self", "self_att", "visn_self_att", "visual_att_output", "visual_attention_mask", "visual_input"], "lxmert/modeling_lxmert.py:LxmertXLayer.output_fc": ["ModelXLayer", "Module", "class", "def", "lang_input", "lang_inter", "lang_inter_output", "lang_output", "nn", "output_fc", "return", "self", "visn_inter", "visn_output", "visual_input", "visual_inter_output", "visual_output"], "lxmert/modeling_lxmert.py:LxmertXLayer.forward": ["False", "ModelXLayer", "Module", "attention_probs", "class", "cross_att", "def", "else", "forward", "if", "lang_att_output", "lang_attention_mask", "lang_feats", "lang_input", "lang_output", "nn", "output_attentions", "output_fc", "output_x_attentions", "return", "self", "self_att", "visual_att_output", "visual_attention_mask", "visual_feats", "visual_input", "visual_output"], "lxmert/modeling_lxmert.py:LxmertVisualFeatureEncoder.__init__": ["Dropout", "LayerNorm", "Linear", "ModelVisualFeatureEncoder", "Module", "__init__", "box_fc", "box_layer_norm", "class", "config", "def", "dropout", "eps", "feat_dim", "hidden_dropout_prob", "hidden_size", "nn", "pos_dim", "self", "super", "visn_fc", "visn_layer_norm", "visual_feat_dim", "visual_pos_dim"], "lxmert/modeling_lxmert.py:LxmertVisualFeatureEncoder.forward": ["ModelVisualFeatureEncoder", "Module", "box_fc", "box_layer_norm", "class", "def", "dropout", "forward", "nn", "output", "return", "self", "visn_fc", "visn_layer_norm", "visual_feats", "visual_pos", "x", "y"], "lxmert/modeling_lxmert.py:LxmertEncoder.__init__": ["ModelEncoder", "ModelLayer", "ModelVisualFeatureEncoder", "ModelXLayer", "Module", "ModuleList", "_", "__init__", "class", "config", "def", "for", "in", "l_layers", "layer", "nn", "num_l_layers", "num_r_layers", "num_x_layers", "r_layers", "range", "self", "super", "visn_fc", "x_layers"], "lxmert/modeling_lxmert.py:LxmertEncoder.forward": ["ModelEncoder", "Module", "None", "class", "config", "cross_encoder_attentions", "def", "else", "for", "forward", "if", "in", "is", "l_outputs", "lang_attention_mask", "lang_encoder_outputs", "lang_feats", "language_attentions", "language_hidden_states", "layer", "layer_module", "nn", "not", "or", "output_attentions", "r_layers", "return", "self", "v_outputs", "vision_attentions", "vision_hidden_states", "visn_fc", "visual_attention_mask", "visual_encoder_outputs", "visual_feats", "visual_pos", "x_layers", "x_outputs"], "lxmert/modeling_lxmert.py:LxmertPooler.__init__": ["Linear", "ModelPooler", "Module", "Tanh", "__init__", "activation", "class", "config", "def", "dense", "hidden_size", "nn", "self", "super"], "lxmert/modeling_lxmert.py:LxmertPooler.forward": ["ModelPooler", "Module", "activation", "class", "def", "dense", "first_token_tensor", "forward", "hidden_states", "nn", "pooled_output", "return", "self"], "lxmert/modeling_lxmert.py:LxmertPredictionHeadTransform.__init__": ["ACT2FN", "LayerNorm", "Linear", "ModelPredictionHeadTransform", "Module", "__init__", "class", "config", "def", "dense", "eps", "hidden_act", "hidden_size", "nn", "self", "super", "transform_act_fn"], "lxmert/modeling_lxmert.py:LxmertPredictionHeadTransform.forward": ["LayerNorm", "ModelPredictionHeadTransform", "Module", "class", "def", "dense", "forward", "hidden_states", "nn", "return", "self", "transform_act_fn"], "lxmert/modeling_lxmert.py:LxmertLMPredictionHead.__init__": ["False", "Linear", "ModelLMPredictionHead", "ModelPredictionHeadTransform", "Module", "Parameter", "__init__", "bias", "class", "config", "decoder", "def", "hidden_size", "nn", "self", "super", "torch", "transform", "vocab_size", "zeros"], "lxmert/modeling_lxmert.py:LxmertLMPredictionHead.forward": ["ModelLMPredictionHead", "Module", "bias", "class", "decoder", "def", "forward", "hidden_states", "nn", "return", "self", "transform"], "lxmert/modeling_lxmert.py:LxmertVisualAnswerHead.__init__": ["GeLU", "LayerNorm", "Linear", "ModelVisualAnswerHead", "Module", "Sequential", "__init__", "class", "config", "def", "eps", "hid_dim", "hidden_size", "logit_fc", "nn", "num_labels", "self", "super"], "lxmert/modeling_lxmert.py:LxmertVisualAnswerHead.forward": ["ModelVisualAnswerHead", "Module", "class", "def", "forward", "hidden_states", "logit_fc", "nn", "return", "self"], "lxmert/modeling_lxmert.py:LxmertVisualObjHead.__init__": ["Linear", "ModelPredictionHeadTransform", "ModelVisualObjHead", "Module", "ModuleDict", "__init__", "attr", "class", "config", "decoder_dict", "def", "feat", "for", "hidden_size", "if", "in", "key", "nn", "num", "num_attr_labels", "num_object_labels", "obj", "self", "shape", "super", "transform", "visual_attr_loss", "visual_feat_dim", "visual_feat_loss", "visual_losses", "visual_obj_loss"], "lxmert/modeling_lxmert.py:LxmertVisualObjHead.forward": ["ModelVisualObjHead", "Module", "class", "decoder_dict", "def", "for", "forward", "hidden_states", "in", "key", "nn", "output", "return", "self", "transform", "visual_losses"], "lxmert/modeling_lxmert.py:LxmertPreTrainingHeads.__init__": ["Linear", "ModelLMPredictionHead", "ModelPreTrainingHeads", "Module", "__init__", "class", "config", "def", "hidden_size", "nn", "predictions", "self", "seq_relationship", "super"], "lxmert/modeling_lxmert.py:LxmertPreTrainingHeads.forward": ["ModelPreTrainingHeads", "Module", "class", "def", "forward", "nn", "pooled_output", "prediction_scores", "predictions", "return", "self", "seq_relationship", "seq_relationship_score", "sequence_output"], "lxmert/modeling_lxmert.py:LxmertPreTrainedModel._init_weights": ["ModelLMPredictionHead", "ModelPreTrainedModel", "PreTrainedModel", "_init_weights", "bias", "class", "def", "if", "init", "isinstance", "module", "self", "super", "zeros_"], "lxmert/modeling_lxmert.py:LxmertModel.__init__": ["ModelEmbeddings", "ModelEncoder", "ModelModel", "ModelPooler", "ModelPreTrainedModel", "__init__", "class", "config", "def", "embeddings", "encoder", "pooler", "post_init", "self", "super"], "lxmert/modeling_lxmert.py:LxmertModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "return", "self", "word_embeddings"], "lxmert/modeling_lxmert.py:LxmertModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "new_embeddings", "self", "set_input_embeddings", "word_embeddings"], "lxmert/modeling_lxmert.py:LxmertModel.forward": ["FloatTensor", "LongTensor", "ModelModel", "ModelModelOutput", "ModelPreTrainedModel", "None", "ValueError", "You", "all_attentions", "and", "at", "attention_mask", "be", "bool", "both", "cannot", "class", "config", "cross_encoder_attentions", "def", "device", "dtype", "either", "elif", "else", "embedding_output", "embeddings", "encoder", "encoder_outputs", "extended_attention_mask", "extended_visual_attention_mask", "finfo", "forward", "have", "hidden_states", "if", "input_ids", "input_shape", "inputs_embeds", "is", "kwargs", "lang_encoder_outputs", "lang_output", "language_attentions", "language_hidden_states", "language_output", "long", "min", "not", "ones", "or", "output_attentions", "output_hidden_states", "pooled_output", "pooler", "r", "raise", "return", "return_dict", "same", "self", "size", "specify", "the", "time", "to", "token_type_ids", "torch", "tuple", "unsqueeze", "use_return_dict", "vision_attentions", "vision_hidden_states", "vision_output", "visual_attention_mask", "visual_encoder_outputs", "visual_feats", "visual_output", "visual_pos", "warn_if_padding_and_no_attention_mask", "zeros"], "lxmert/modeling_lxmert.py:LxmertForPreTraining.__init__": ["CrossEntropyLoss", "Model", "ModelForPreTraining", "ModelModel", "ModelPreTrainedModel", "ModelPreTrainingHeads", "ModelVisualAnswerHead", "ModelVisualObjHead", "SmoothL1Loss", "__init__", "answer_head", "attr", "ce", "class", "cls", "config", "def", "feat", "if", "l2", "loss", "loss_fcts", "none", "num", "num_attr_labels", "num_object_labels", "num_qa_labels", "obj", "obj_predict_head", "post_init", "reduction", "self", "shape", "super", "task_mask_lm", "task_matched", "task_obj_predict", "task_qa", "visual_attr_loss", "visual_ce", "visual_feat_dim", "visual_feat_loss", "visual_loss_normalizer", "visual_losses", "visual_obj_loss"], "lxmert/modeling_lxmert.py:LxmertForPreTraining.resize_token_embeddings": ["Embedding", "ModelForPreTraining", "ModelPreTrainedModel", "None", "True", "_resize_bias", "bias", "bool", "class", "cls", "def", "int", "mean_resizing", "new_embeddings", "new_num_tokens", "nn", "pad_to_multiple_of", "predictions", "resize_token_embeddings", "return", "self", "super"], "lxmert/modeling_lxmert.py:LxmertForPreTraining._resize_bias": ["ModelForPreTraining", "ModelPreTrainedModel", "Parameter", "_resize_bias", "bias", "cat", "class", "def", "device", "else", "extra_bias", "if", "int", "new_bias", "new_num_tokens", "nn", "old_num_tokens", "return", "self", "shape", "torch", "zeros"], "lxmert/modeling_lxmert.py:LxmertForPreTraining.resize_num_qa_labels": ["ModelForPreTraining", "ModelPreTrainedModel", "None", "_resize_qa_labels", "class", "config", "cur_qa_logit_layer", "def", "get_qa_logit_layer", "if", "is", "new_qa_logit_layer", "num_labels", "num_qa_labels", "or", "resize_num_qa_labels", "return", "self"], "lxmert/modeling_lxmert.py:LxmertForPreTraining._resize_qa_labels": ["ModelForPreTraining", "ModelPreTrainedModel", "_get_resized_qa_labels", "_resize_qa_labels", "_set_qa_logit_layer", "class", "cur_qa_logit_layer", "def", "get_qa_logit_layer", "new_qa_logit_layer", "num_labels", "return", "self"], "lxmert/modeling_lxmert.py:LxmertForPreTraining.get_qa_logit_layer": ["ModelForPreTraining", "ModelPreTrainedModel", "Module", "answer_head", "class", "def", "get_qa_logit_layer", "hasattr", "if", "logit_fc", "nn", "return", "self"], "lxmert/modeling_lxmert.py:LxmertForPreTraining._set_qa_logit_layer": ["ModelForPreTraining", "ModelPreTrainedModel", "_set_qa_logit_layer", "answer_head", "class", "def", "logit_fc", "qa_logit_layer", "self"], "lxmert/modeling_lxmert.py:LxmertForPreTraining._get_resized_qa_labels": ["False", "Linear", "ModelForPreTraining", "ModelPreTrainedModel", "None", "_get_resized_qa_labels", "_init_weights", "bias", "class", "cur_qa_labels", "cur_qa_logit_layer", "data", "def", "device", "else", "getattr", "hidden_dim", "if", "is", "min", "new_qa_logit_layer", "nn", "not", "num_labels", "num_labels_to_copy", "return", "self", "size", "to", "weight"], "lxmert/modeling_lxmert.py:LxmertForPreTraining.forward": ["FloatTensor", "LongTensor", "Model", "ModelForPreTraining", "ModelForPreTrainingOutput", "ModelPreTrainedModel", "Model_output", "None", "Tensor", "and", "ans", "answer_head", "answer_loss", "answer_score", "attention_mask", "bool", "ce", "class", "cls", "config", "cross_encoder_attentions", "cross_relationship_score", "def", "device", "dict", "dim", "else", "for", "forward", "if", "in", "input_ids", "inputs_embeds", "is", "items", "key", "key_info", "kwargs", "label", "label_shape", "labels", "lang_output", "lang_prediction_scores", "language_attentions", "language_hidden_states", "loss", "loss_fct_name", "loss_fcts", "mask_conf", "masked_lm_loss", "matched_label", "matched_loss", "mean", "not", "num", "num_qa_labels", "obj_labels", "obj_predict_head", "output", "output_attentions", "output_dim", "output_hidden_states", "pooled_output", "prediction_logits", "question_answering_score", "r", "return", "return_dict", "self", "shape", "str", "task_mask_lm", "task_matched", "task_obj_predict", "task_qa", "tensor", "token_type_ids", "torch", "total_loss", "total_visual_loss", "tuple", "use_return_dict", "view", "vision_attentions", "vision_hidden_states", "visual_attention_mask", "visual_feats", "visual_loss", "visual_loss_fct", "visual_loss_normalizer", "visual_losses", "visual_output", "visual_pos", "visual_prediction_scores", "visual_prediction_scores_dict", "vocab_size", "weight"], "lxmert/modeling_lxmert.py:LxmertForQuestionAnswering.__init__": ["CrossEntropyLoss", "Model", "ModelForQuestionAnswering", "ModelModel", "ModelPreTrainedModel", "ModelVisualAnswerHead", "__init__", "answer_head", "class", "config", "def", "loss", "num_qa_labels", "post_init", "self", "super", "visual_loss_normalizer"], "lxmert/modeling_lxmert.py:LxmertForQuestionAnswering.resize_num_qa_labels": ["ModelForQuestionAnswering", "ModelPreTrainedModel", "None", "_resize_qa_labels", "class", "config", "cur_qa_logit_layer", "def", "get_qa_logit_layer", "if", "is", "new_qa_logit_layer", "num_labels", "num_qa_labels", "or", "resize_num_qa_labels", "return", "self"], "lxmert/modeling_lxmert.py:LxmertForQuestionAnswering._resize_qa_labels": ["ModelForQuestionAnswering", "ModelPreTrainedModel", "_get_resized_qa_labels", "_resize_qa_labels", "_set_qa_logit_layer", "class", "cur_qa_logit_layer", "def", "get_qa_logit_layer", "new_qa_logit_layer", "num_labels", "return", "self"], "lxmert/modeling_lxmert.py:LxmertForQuestionAnswering.get_qa_logit_layer": ["ModelForQuestionAnswering", "ModelPreTrainedModel", "Module", "answer_head", "class", "def", "get_qa_logit_layer", "hasattr", "if", "logit_fc", "nn", "return", "self"], "lxmert/modeling_lxmert.py:LxmertForQuestionAnswering._set_qa_logit_layer": ["ModelForQuestionAnswering", "ModelPreTrainedModel", "_set_qa_logit_layer", "answer_head", "class", "def", "logit_fc", "qa_logit_layer", "self"], "lxmert/modeling_lxmert.py:LxmertForQuestionAnswering._get_resized_qa_labels": ["False", "Linear", "ModelForQuestionAnswering", "ModelPreTrainedModel", "None", "_get_resized_qa_labels", "_init_weights", "bias", "class", "cur_qa_labels", "cur_qa_logit_layer", "data", "def", "device", "else", "getattr", "hidden_dim", "if", "is", "min", "new_qa_logit_layer", "nn", "not", "num_labels", "num_labels_to_copy", "return", "self", "size", "to", "weight"], "lxmert/modeling_lxmert.py:LxmertForQuestionAnswering.forward": ["FloatTensor", "LongTensor", "Model", "ModelForQuestionAnswering", "ModelForQuestionAnsweringOutput", "ModelPreTrainedModel", "Model_output", "None", "Tensor", "answer_head", "answer_score", "attention_mask", "bool", "class", "config", "cross_encoder_attentions", "def", "else", "forward", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "language_attentions", "language_hidden_states", "loss", "not", "num_qa_labels", "output", "output_attentions", "output_hidden_states", "pooled_output", "question_answering_score", "r", "return", "return_dict", "self", "token_type_ids", "torch", "tuple", "use_return_dict", "view", "vision_attentions", "vision_hidden_states", "visual_attention_mask", "visual_feats", "visual_pos"], "m2m_100/modeling_m2m_100.py:shift_tokens_right": ["Model_tokens_right", "Modeled_input_ids", "None", "Tensor", "ValueError", "be", "clone", "config", "decoder_start_token_id", "def", "defined", "has", "if", "input_ids", "int", "is", "masked_fill_", "model", "new_zeros", "pad_token_id", "raise", "return", "self", "shape", "to", "torch"], "m2m_100/modeling_m2m_100.py:M2M100ScaledWordEmbedding.__init__": ["Embedding", "ModelScaledWordEmbedding", "None", "__init__", "class", "def", "embed_scale", "embedding_dim", "float", "int", "nn", "num_embeddings", "padding_idx", "self", "super"], "m2m_100/modeling_m2m_100.py:M2M100ScaledWordEmbedding.forward": ["Embedding", "ModelScaledWordEmbedding", "Tensor", "class", "def", "embed_scale", "forward", "input_ids", "nn", "return", "self", "super", "torch"], "m2m_100/modeling_m2m_100.py:M2M100SinusoidalPositionalEmbedding.__init__": ["ModelSinusoidalPositionalEmbedding", "Module", "None", "__init__", "class", "def", "embedding_dim", "int", "make_weights", "nn", "num_positions", "offset", "padding_idx", "self", "super"], "m2m_100/modeling_m2m_100.py:M2M100SinusoidalPositionalEmbedding.make_weights": ["False", "ModelSinusoidalPositionalEmbedding", "Module", "None", "class", "def", "device", "dtype", "emb_weights", "embedding_dim", "get_embedding", "hasattr", "if", "int", "make_weights", "nn", "num_embeddings", "padding_idx", "persistent", "register_buffer", "self", "to", "weights"], "m2m_100/modeling_m2m_100.py:M2M100SinusoidalPositionalEmbedding.get_embedding": ["ModelSinusoidalPositionalEmbedding", "Module", "None", "arange", "cat", "class", "cos", "def", "dim", "dtype", "emb", "embedding_dim", "exp", "float", "get_default_dtype", "get_embedding", "half_dim", "if", "int", "int64", "is", "log", "math", "nn", "not", "num_embeddings", "padding_idx", "return", "sin", "to", "torch", "unsqueeze", "view", "zeros"], "m2m_100/modeling_m2m_100.py:M2M100SinusoidalPositionalEmbedding.forward": ["ModelSinusoidalPositionalEmbedding", "Module", "None", "Tensor", "bsz", "class", "create_position_ids_from_input_ids", "create_position_ids_from_inputs_embeds", "def", "detach", "device", "else", "embedding_dim", "forward", "if", "index_select", "input_ids", "inputs_embeds", "int", "is", "make_weights", "max_pos", "nn", "not", "offset", "padding_idx", "past_key_values_length", "position_ids", "return", "self", "seq_len", "shape", "size", "to", "torch", "view", "weights"], "m2m_100/modeling_m2m_100.py:M2M100SinusoidalPositionalEmbedding.create_position_ids_from_inputs_embeds": ["ModelSinusoidalPositionalEmbedding", "Module", "arange", "class", "contiguous", "create_position_ids_from_inputs_embeds", "def", "device", "dtype", "expand", "input_shape", "inputs_embeds", "long", "nn", "padding_idx", "past_key_values_length", "position_ids", "return", "sequence_length", "size", "torch", "unsqueeze"], "m2m_100/modeling_m2m_100.py:M2M100SinusoidalPositionalEmbedding.create_position_ids_from_input_ids": ["ModelSinusoidalPositionalEmbedding", "Module", "class", "create_position_ids_from_input_ids", "cumsum", "def", "dim", "incremental_indices", "input_ids", "int", "long", "mask", "ne", "nn", "padding_idx", "past_key_values_length", "return", "torch", "type_as"], "m2m_100/modeling_m2m_100.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "contiguous", "def", "dim", "dropout", "float", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "shape", "size", "softmax", "torch", "training", "transpose", "value"], "m2m_100/modeling_m2m_100.py:M2M100Attention.__init__": ["False", "Instantiating", "Linear", "ModelAttention", "ModelConfig", "Module", "None", "Please", "True", "ValueError", "__class__", "__init__", "__name__", "a", "and", "be", "bias", "bool", "by", "caching", "call", "class", "config", "creating", "decoder", "def", "divisible", "dropout", "during", "embed_dim", "errors", "f", "float", "forward", "got", "head_dim", "if", "int", "is", "is_causal", "is_decoder", "k_proj", "layer_idx", "lead", "logger", "make", "must", "nn", "not", "num_heads", "out_proj", "passing", "provide", "q_proj", "raise", "recommended", "scaling", "self", "super", "sure", "the", "this", "to", "used", "v_proj", "warning_once", "when", "will", "without"], "m2m_100/modeling_m2m_100.py:M2M100Attention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "EncoderDecoderCache", "False", "FlashAttentionKwargs", "ModelAttention", "Module", "None", "Tensor", "True", "Unpack", "_attn_implementation", "and", "attention_interface", "attention_mask", "attn_output", "attn_weights", "bool", "bsz", "cache_position", "class", "config", "contiguous", "cross_attention_cache", "curr_past_key_values", "current_states", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "get", "head_dim", "hidden_states", "if", "is", "is_cross_attention", "is_updated", "isinstance", "k_proj", "key_states", "key_value_states", "keys", "kv_input_shape", "kwargs", "layer_idx", "layers", "nn", "not", "out_proj", "output_attentions", "past_key_values", "q_input_shape", "q_proj", "query_states", "reshape", "return", "scaling", "self", "self_attention_cache", "shape", "src_len", "tgt_len", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "values", "view"], "m2m_100/modeling_m2m_100.py:M2M100EncoderLayer.__init__": ["ACT2FN", "GradientCheckpointingLayer", "LayerNorm", "Linear", "ModelAttention", "ModelConfig", "ModelEncoderLayer", "__init__", "activation_dropout", "activation_fn", "activation_function", "attention_dropout", "class", "config", "d_model", "def", "dropout", "embed_dim", "encoder_attention_heads", "encoder_ffn_dim", "fc1", "fc2", "final_layer_norm", "nn", "num_heads", "self", "self_attn", "self_attn_layer_norm", "super"], "m2m_100/modeling_m2m_100.py:M2M100EncoderLayer.forward": ["False", "GradientCheckpointingLayer", "ModelEncoderLayer", "Tensor", "activation_dropout", "activation_fn", "attention_mask", "attn_weights", "bool", "clamp", "clamp_value", "class", "def", "dropout", "dtype", "fc1", "fc2", "final_layer_norm", "finfo", "float16", "forward", "functional", "hidden_states", "if", "max", "min", "nn", "output_attentions", "p", "residual", "return", "self", "self_attn", "self_attn_layer_norm", "torch", "training"], "m2m_100/modeling_m2m_100.py:M2M100DecoderLayer.__init__": ["ACT2FN", "GradientCheckpointingLayer", "LayerNorm", "Linear", "ModelAttention", "ModelConfig", "ModelDecoderLayer", "None", "True", "__init__", "activation_dropout", "activation_fn", "activation_function", "attention_dropout", "class", "config", "d_model", "decoder_attention_heads", "decoder_ffn_dim", "def", "dropout", "embed_dim", "encoder_attn", "encoder_attn_layer_norm", "fc1", "fc2", "final_layer_norm", "int", "is_causal", "is_decoder", "layer_idx", "nn", "num_heads", "self", "self_attn", "self_attn_layer_norm", "super"], "m2m_100/modeling_m2m_100.py:M2M100DecoderLayer.forward": ["Cache", "False", "GradientCheckpointingLayer", "ModelDecoderLayer", "None", "Tensor", "True", "activation_dropout", "activation_fn", "attention_mask", "bool", "cache_position", "class", "cross_attn_weights", "def", "dropout", "encoder_attention_mask", "encoder_attn", "encoder_attn_layer_norm", "encoder_hidden_states", "fc1", "fc2", "final_layer_norm", "forward", "functional", "hidden_states", "if", "is", "key_value_states", "nn", "not", "output_attentions", "outputs", "p", "past_key_values", "residual", "return", "self", "self_attn", "self_attn_layer_norm", "self_attn_weights", "torch", "training", "use_cache"], "m2m_100/modeling_m2m_100.py:M2M100PreTrainedModel._init_weights": ["ModelPreTrainedModel", "ModelSinusoidalPositionalEmbedding", "PreTrainedModel", "_init_weights", "class", "copy_", "def", "emb_weights", "embedding_dim", "get_embedding", "if", "init", "isinstance", "module", "num_positions", "offset", "padding_idx", "self", "super", "weights"], "m2m_100/modeling_m2m_100.py:M2M100Encoder.__init__": ["False", "LayerNorm", "ModelConfig", "ModelEncoder", "ModelEncoderLayer", "ModelPreTrainedModel", "ModelScaledWordEmbedding", "ModelSinusoidalPositionalEmbedding", "ModuleList", "_", "__init__", "class", "config", "d_model", "def", "dropout", "else", "embed_dim", "embed_positions", "embed_scale", "embed_tokens", "encoder_layerdrop", "encoder_layers", "for", "gradient_checkpointing", "if", "in", "layer_norm", "layerdrop", "layers", "math", "max_position_embeddings", "max_source_positions", "nn", "pad_token_id", "padding_idx", "post_init", "range", "scale_embedding", "self", "sqrt", "super", "vocab_size"], "m2m_100/modeling_m2m_100.py:M2M100Encoder.forward": ["BaseModelOutput", "ModelEncoder", "ModelPreTrainedModel", "None", "Tensor", "ValueError", "You", "all_attentions", "and", "at", "attention_mask", "attentions", "bool", "both", "cannot", "class", "config", "create_bidirectional_mask", "def", "device", "dropout", "dropout_probability", "either", "elif", "else", "embed_pos", "embed_positions", "embed_tokens", "encoder_layer", "encoder_states", "enumerate", "for", "forward", "functional", "have", "hidden_states", "idx", "if", "in", "input_embeds", "input_ids", "input_shape", "inputs_embeds", "is", "is_deepspeed_zero3_enabled", "is_fsdp_managed_module", "kwargs", "last_hidden_state", "layer_norm", "layer_outputs", "layerdrop", "layers", "nn", "not", "or", "output_attentions", "output_hidden_states", "p", "r", "raise", "rand", "return", "return_dict", "same", "self", "size", "skip_the_layer", "specify", "synced_gpus", "the", "time", "to", "torch", "training", "tuple", "use_return_dict", "v", "view", "warn_if_padding_and_no_attention_mask"], "m2m_100/modeling_m2m_100.py:M2M100Decoder.__init__": ["False", "LayerNorm", "ModelConfig", "ModelDecoder", "ModelDecoderLayer", "ModelPreTrainedModel", "ModelScaledWordEmbedding", "ModelSinusoidalPositionalEmbedding", "ModuleList", "__init__", "class", "config", "d_model", "decoder_layerdrop", "decoder_layers", "def", "dropout", "else", "embed_positions", "embed_scale", "embed_tokens", "for", "gradient_checkpointing", "i", "if", "in", "layer_idx", "layer_norm", "layerdrop", "layers", "math", "max_position_embeddings", "max_target_positions", "nn", "pad_token_id", "padding_idx", "post_init", "range", "scale_embedding", "self", "sqrt", "super", "vocab_size"], "m2m_100/modeling_m2m_100.py:M2M100Decoder.forward": ["BaseModelOutputWithPastAndCrossAttentions", "Cache", "DynamicCache", "EncoderDecoderCache", "False", "ModelDecoder", "ModelPreTrainedModel", "None", "Setting", "Tensor", "True", "ValueError", "You", "all_cross_attentions", "all_hidden_states", "all_self_attns", "and", "arange", "at", "attention_mask", "attentions", "batch_size", "bool", "both", "cache_position", "cannot", "checkpointing", "class", "config", "continue", "create_bidirectional_mask", "create_causal_mask", "cross_attentions", "decoder_input_ids", "decoder_inputs_embeds", "decoder_layer", "def", "device", "dropout", "dropout_probability", "either", "elif", "else", "embed_positions", "embed_tokens", "encoder_attention_mask", "encoder_hidden_states", "enumerate", "for", "forward", "functional", "get_seq_length", "gradient", "gradient_checkpointing", "have", "hidden_states", "idx", "if", "in", "incompatible", "input_embeds", "input_ids", "input_shape", "inputs_embeds", "is", "is_deepspeed_zero3_enabled", "is_fsdp_managed_module", "is_torchdynamo_compiling", "isinstance", "kwargs", "last_hidden_state", "layer_norm", "layer_outputs", "layerdrop", "layers", "logger", "mask_seq_length", "nn", "not", "ones", "or", "output_attentions", "output_hidden_states", "p", "past_key_values", "past_key_values_length", "positions", "r", "raise", "rand", "return", "return_dict", "same", "self", "self_attention_cache", "self_attn_cache", "seq_length", "size", "skip_the_layer", "specify", "synced_gpus", "the", "time", "to", "torch", "training", "tuple", "use_cache", "use_return_dict", "v", "view", "warning_once", "with"], "m2m_100/modeling_m2m_100.py:M2M100Model.__init__": ["ModelConfig", "ModelDecoder", "ModelEncoder", "ModelModel", "ModelPreTrainedModel", "ModelScaledWordEmbedding", "__init__", "class", "config", "d_model", "decoder", "def", "else", "embed_scale", "encoder", "if", "math", "pad_token_id", "padding_idx", "post_init", "scale_embedding", "self", "shared", "sqrt", "super", "vocab_size"], "m2m_100/modeling_m2m_100.py:M2M100Model.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "return", "self", "shared"], "m2m_100/modeling_m2m_100.py:M2M100Model.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "decoder", "def", "embed_tokens", "encoder", "self", "set_input_embeddings", "shared", "value"], "m2m_100/modeling_m2m_100.py:M2M100Model.forward": ["BaseModelOutput", "Cache", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Seq2SeqModelOutput", "Tensor", "and", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "cross_attentions", "decoder", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_input_ids", "decoder_inputs_embeds", "decoder_outputs", "def", "elif", "else", "encoder", "encoder_attention_mask", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "isinstance", "kwargs", "last_hidden_state", "len", "not", "output_attentions", "output_hidden_states", "past_key_values", "r", "return", "return_dict", "self", "torch", "tuple", "use_cache", "use_return_dict"], "m2m_100/modeling_m2m_100.py:M2M100ForConditionalGeneration.__init__": ["False", "GenerationMixin", "Linear", "ModelConfig", "ModelForConditionalGeneration", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "d_model", "def", "lm_head", "model", "nn", "num_embeddings", "post_init", "self", "shared", "super"], "m2m_100/modeling_m2m_100.py:M2M100ForConditionalGeneration.forward": ["Cache", "CrossEntropyLoss", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Seq2SeqLMOutput", "Tensor", "attention_mask", "bool", "cache_position", "class", "config", "cross_attentions", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_input_ids", "decoder_inputs_embeds", "decoder_start_token_id", "def", "device", "else", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "forward", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "lm_head", "lm_logits", "logits", "loss", "loss_fct", "masked_lm_loss", "model", "not", "output", "output_attentions", "output_hidden_states", "outputs", "pad_token_id", "past_key_values", "r", "return", "return_dict", "self", "shift_tokens_right", "to", "torch", "tuple", "use_cache", "use_return_dict", "view", "vocab_size"], "mamba/modeling_mamba.py:MambaCache.__init__": ["ModelCache", "None", "PreTrainedConfig", "Tensor", "_", "__init__", "_dtype", "_dynamo", "append", "class", "config", "conv_kernel", "conv_kernel_size", "conv_state", "conv_states", "def", "device", "dtype", "else", "float16", "for", "if", "in", "int", "intermediate_size", "is", "list", "mark_static_address", "max_batch_size", "not", "num_hidden_layers", "range", "self", "ssm_state", "ssm_state_size", "ssm_states", "state_size", "str", "torch", "zeros"], "mamba/modeling_mamba.py:MambaCache.update_conv_state": ["LongTensor", "ModelCache", "Tensor", "cache_position", "clamp", "class", "conv_kernel_size", "conv_state", "conv_states", "def", "device", "dims", "dtype", "if", "int", "layer_idx", "new_conv_state", "return", "roll", "self", "shifts", "to", "torch", "update_conv_state", "zero_"], "mamba/modeling_mamba.py:MambaCache.update_ssm_state": ["ModelCache", "Tensor", "class", "def", "device", "int", "layer_idx", "new_ssm_state", "return", "self", "ssm_states", "to", "torch", "update_ssm_state", "zero_"], "mamba/modeling_mamba.py:MambaCache.reset": ["ModelCache", "class", "conv_states", "def", "for", "in", "layer_idx", "len", "range", "reset", "self", "ssm_states", "zero_"], "mamba/modeling_mamba.py:MambaMixer.__init__": ["A", "ACT2FN", "A_log", "Conv1d", "D", "False", "Linear", "Model", "ModelConfig", "ModelMixer", "Model_inner_fn", "Model_ssm", "Module", "None", "Parameter", "True", "__init__", "act", "activation", "arange", "bias", "causal", "causal_conv1d", "causal_conv1d_fn", "causal_conv1d_update", "class", "config", "contiguous", "conv1d", "conv_kernel", "conv_kernel_size", "def", "dt_proj", "dtype", "else", "expand", "float32", "global", "groups", "hidden_act", "hidden_size", "if", "in_channels", "in_proj", "int", "intermediate_size", "is", "kernel_size", "layer_idx", "lazy_load_kernel", "log", "nn", "not", "ones", "out_channels", "out_proj", "padding", "selective_scan_fn", "selective_state_update", "self", "ssm", "ssm_state_size", "state_size", "super", "time_step_rank", "torch", "use_Modelpy", "use_bias", "use_conv_bias", "warn_slow_implementation", "x_proj"], "mamba/modeling_mamba.py:MambaMixer.warn_slow_implementation": ["AILab", "Dao", "Falling", "False", "For", "ImportError", "Model", "ModelMixer", "Model_inner_fn", "Modelpy", "Module", "None", "The", "To", "True", "all", "alxndrTL", "as", "available", "back", "backend", "because", "but", "causal", "causal_conv1d_fn", "causal_conv1d_update", "class", "com", "conv1d", "def", "else", "fast", "follow", "for", "github", "https", "if", "implementation", "install", "installed", "is", "is_Modelpy_available", "is_fast_path_available", "it", "kernels", "library", "logger", "nn", "not", "of", "one", "or", "package", "path", "pip", "py", "raise", "selective_scan_fn", "selective_state_update", "self", "sequential", "set", "spaces", "state", "the", "to", "use_Modelpy", "using", "warn_slow_implementation", "warning_once"], "mamba/modeling_mamba.py:MambaMixer.cuda_kernels_forward": ["A", "A_log", "B", "C", "D", "LongTensor", "ModelCache", "ModelMixer", "Model_inner_fn", "Module", "None", "Tensor", "True", "activation", "and", "attention_mask", "bias", "cache_params", "cache_position", "causal_conv1d_fn", "causal_conv1d_update", "chunk", "class", "contextualized_states", "conv1d", "conv_kernel_size", "conv_states", "conv_weights", "cuda_kernels_forward", "def", "delta_bias", "delta_softplus", "dim", "discrete_time_step", "dt_proj", "dt_softplus", "else", "exp", "float", "functional", "gate", "hasattr", "hidden_states", "if", "in_proj", "is", "layer_idx", "nn", "not", "out_proj", "pad", "projected_states", "return", "return_last_state", "scan_outputs", "selective_scan_fn", "selective_state_update", "self", "shape", "size", "split", "squeeze", "ssm_parameters", "ssm_state", "ssm_state_size", "ssm_states", "time_proj_bias", "time_step", "time_step_rank", "torch", "training", "transpose", "unsqueeze", "update_conv_state", "update_ssm_state", "use_bias", "use_conv_bias", "view", "weight", "x_proj"], "mamba/modeling_mamba.py:MambaMixer.slow_forward": ["A", "A_log", "B", "C", "D", "LongTensor", "ModelCache", "ModelMixer", "Module", "None", "_", "act", "and", "append", "attention_mask", "batch_size", "bias", "cache_params", "cache_position", "chunk", "class", "clone", "contextualized_states", "conv1d", "conv_kernel_size", "conv_state", "copy_", "def", "deltaB_u", "device", "dim", "discrete_A", "discrete_B", "discrete_time_step", "dt_proj", "dtype", "else", "exp", "float", "for", "functional", "gate", "hidden_states", "hs", "i", "if", "in", "in_proj", "input_states", "intermediate_size", "is", "layer_idx", "matmul", "nn", "not", "out_proj", "pad", "projected_states", "pscan", "range", "return", "scan_output", "scan_outputs", "self", "seq_len", "shape", "slow_forward", "softplus", "split", "squeeze", "ssm_parameters", "ssm_state", "ssm_state_size", "ssm_states", "stack", "sum", "time_step", "time_step_rank", "to", "torch", "training", "transpose", "unsqueeze", "update_conv_state", "use_Modelpy", "use_conv_bias", "weight", "x_proj", "zeros"], "mamba/modeling_mamba.py:MambaMixer.forward": ["LongTensor", "ModelCache", "ModelMixer", "Model_inner_fn", "Module", "None", "all", "and", "attention_mask", "cache_params", "cache_position", "causal_conv1d_fn", "causal_conv1d_update", "class", "cuda", "cuda_kernels_forward", "def", "device", "forward", "hidden_states", "if", "in", "is_fast_path_available", "is_torchdynamo_compiling", "nn", "not", "return", "selective_scan_fn", "selective_state_update", "self", "slow_forward", "torch", "type", "weight", "x_proj"], "mamba/modeling_mamba.py:MambaRMSNorm.__init__": ["ModelRMSNorm", "Module", "Parameter", "__init__", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "mamba/modeling_mamba.py:MambaRMSNorm.forward": ["ModelRMSNorm", "Module", "True", "class", "def", "dtype", "float32", "forward", "hidden_states", "input_dtype", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "mamba/modeling_mamba.py:MambaRMSNorm.extra_repr": ["ModelRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "variance_epsilon", "weight"], "mamba/modeling_mamba.py:MambaBlock.__init__": ["GradientCheckpointingLayer", "ModelBlock", "ModelMixer", "ModelRMSNorm", "__init__", "class", "config", "def", "eps", "hidden_size", "layer_idx", "layer_norm_epsilon", "mixer", "norm", "residual_in_fp32", "self", "super"], "mamba/modeling_mamba.py:MambaBlock.forward": ["GradientCheckpointingLayer", "LongTensor", "ModelBlock", "ModelCache", "None", "attention_mask", "cache_params", "cache_position", "class", "def", "dtype", "float32", "forward", "hidden_states", "if", "mixer", "norm", "residual", "residual_in_fp32", "return", "self", "to", "torch", "weight"], "mamba/modeling_mamba.py:MambaPreTrainedModel._init_weights": ["A", "A_log", "D", "Embedding", "Linear", "ModelMixer", "ModelPreTrainedModel", "ModelRMSNorm", "None", "PreTrainedModel", "_init_weights", "a", "arange", "bias", "clamp", "class", "config", "constant", "constant_", "contiguous", "conv1d", "copy_", "def", "dt", "dt_init_std", "dt_proj", "dtype", "elif", "exp", "expand", "expm1", "float32", "if", "init", "initializer_range", "intermediate_size", "inv_dt", "is", "isinstance", "kaiming_uniform_", "log", "math", "min", "module", "nn", "normal_", "not", "num_hidden_layers", "ones_", "out_proj", "p", "rand", "random", "rescale_prenorm_residual", "self", "sqrt", "ssm_state_size", "std", "time_step_floor", "time_step_init_scheme", "time_step_max", "time_step_min", "time_step_rank", "time_step_scale", "torch", "uniform_", "weight", "zeros_"], "mamba/modeling_mamba.py:MambaModel.__init__": ["Embedding", "False", "ModelBlock", "ModelModel", "ModelPreTrainedModel", "ModelRMSNorm", "ModuleList", "__init__", "_register_load_state_dict_pre_hook", "class", "config", "def", "embeddings", "eps", "for", "gradient_checkpointing", "hidden_size", "idx", "in", "layer_idx", "layer_norm_epsilon", "layers", "load_hook", "nn", "norm_f", "num_hidden_layers", "post_init", "range", "self", "super", "vocab_size"], "mamba/modeling_mamba.py:MambaModel.load_hook": ["ModelModel", "ModelPreTrainedModel", "args", "break", "class", "def", "embedding", "embeddings", "for", "if", "in", "k", "load_hook", "pop", "prefix", "replace", "self", "state_dict"], "mamba/modeling_mamba.py:MambaModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "return", "self"], "mamba/modeling_mamba.py:MambaModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "new_embeddings", "self", "set_input_embeddings"], "mamba/modeling_mamba.py:MambaModel.forward": ["False", "LongTensor", "ModelCache", "ModelModel", "ModelOutput", "ModelPreTrainedModel", "None", "True", "ValueError", "You", "a", "all_hidden_states", "and", "arange", "are", "attention_mask", "automatically", "be", "because", "bool", "cache_params", "cache_position", "case", "class", "config", "conv_kernel", "def", "device", "don", "dtype", "elif", "else", "embeddings", "exactly", "for", "forward", "gradient_checkpointing", "have", "hidden_states", "if", "in", "initialized", "input_ids", "inputs_embeds", "is", "it", "kwargs", "last_hidden_state", "layers", "manually", "mixer_block", "must", "norm_f", "not", "of", "one", "or", "output_hidden_states", "pass", "passed", "prefilling", "r", "raise", "return", "return_dict", "self", "size", "specify", "stage", "t", "that", "the", "to", "torch", "training", "tuple", "use_cache", "use_return_dict", "v", "when", "will", "you"], "mamba/modeling_mamba.py:MambaForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "ModelForCausalLM", "ModelModel", "ModelPreTrainedModel", "__init__", "backbone", "bias", "class", "config", "def", "hidden_size", "lm_head", "nn", "post_init", "self", "super", "vocab_size"], "mamba/modeling_mamba.py:MambaForCausalLM.get_input_embeddings": ["GenerationMixin", "ModelForCausalLM", "ModelPreTrainedModel", "backbone", "class", "def", "get_input_embeddings", "return", "self"], "mamba/modeling_mamba.py:MambaForCausalLM.set_input_embeddings": ["GenerationMixin", "ModelForCausalLM", "ModelPreTrainedModel", "backbone", "class", "def", "new_embeddings", "return", "self", "set_input_embeddings"], "mamba/modeling_mamba.py:MambaForCausalLM._update_model_kwargs_for_generation": ["Any", "GenerationMixin", "ModelForCausalLM", "ModelOutput", "ModelPreTrainedModel", "None", "True", "_update_model_kwargs_for_generation", "and", "attention_mask", "cache_params", "cache_position", "cat", "class", "def", "dict", "dim", "get", "if", "in", "int", "is", "kwargs", "model_kwargs", "new_ones", "not", "num_new_tokens", "outputs", "return", "self", "shape", "str", "torch", "use_cache"], "mamba/modeling_mamba.py:MambaForCausalLM.prepare_inputs_for_generation": ["False", "GenerationMixin", "LongTensor", "ModelCache", "ModelForCausalLM", "ModelPreTrainedModel", "None", "and", "arange", "attention_mask", "backbone", "bool", "cache_params", "cache_position", "class", "config", "contiguous", "conv_kernel", "def", "device", "dtype", "else", "for", "if", "in", "input_ids", "inputs_embeds", "is", "is_first_iteration", "items", "key", "kwargs", "max_batch_size", "model_inputs", "not", "prepare_inputs_for_generation", "return", "self", "size", "torch", "unsqueeze", "update", "use_cache", "value"], "mamba/modeling_mamba.py:MambaForCausalLM.forward": ["CrossEntropyLoss", "FloatTensor", "GenerationMixin", "LongTensor", "ModelCache", "ModelCausalLMOutput", "ModelForCausalLM", "ModelPreTrainedModel", "Model_outputs", "None", "Tensor", "attention_mask", "backbone", "bool", "cache_params", "cache_position", "class", "config", "contiguous", "def", "device", "dtype", "else", "float", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "lm_head", "logits", "logits_to_keep", "loss", "loss_fct", "not", "output", "output_hidden_states", "r", "return", "return_dict", "self", "shift_labels", "shift_logits", "size", "slice", "slice_indices", "to", "torch", "tuple", "use_cache", "use_return_dict", "view", "weight"], "mamba2/modeling_mamba2.py:pad_tensor_by_size": ["Model", "Model_shape", "Model_size", "Model_tensor_by_size", "Tensor", "constant", "def", "else", "functional", "if", "input_tensor", "int", "len", "mode", "nn", "return", "shape", "torch", "value"], "mamba2/modeling_mamba2.py:reshape_into_chunks": ["Model", "Model_into_chunks", "chunk_size", "def", "else", "if", "input_tensor", "len", "pad_size", "pad_tensor_by_size", "return", "shape"], "mamba2/modeling_mamba2.py:segment_sum": ["Model_sum", "None", "bool", "chunk_size", "cumsum", "def", "device", "diagonal", "dim", "dtype", "expand", "inf", "input_tensor", "mask", "masked_fill", "ones", "return", "size", "tensor_segsum", "torch", "tril"], "mamba2/modeling_mamba2.py:apply_mask_to_padding_states": ["Model_mask_to_padding_states", "None", "and", "attention_mask", "def", "dtype", "hidden_states", "if", "is", "not", "return", "shape", "to"], "mamba2/modeling_mamba2.py:Mamba2Cache.__init__": ["ModelCache", "ModelConfig", "None", "__init__", "batch_size", "class", "config", "conv_kernel", "conv_kernel_size", "conv_states", "def", "device", "dtype", "expand", "float16", "head_dim", "hidden_size", "int", "intermediate_size", "n_groups", "num_heads", "num_hidden_layers", "self", "ssm_states", "state_size", "str", "torch", "zeros"], "mamba2/modeling_mamba2.py:Mamba2Cache.update_conv_state": ["False", "ModelCache", "Tensor", "bool", "cache_init", "class", "conv_states", "def", "device", "dims", "else", "if", "int", "layer_idx", "new_conv_state", "return", "roll", "self", "shifts", "to", "torch", "update_conv_state"], "mamba2/modeling_mamba2.py:Mamba2Cache.update_ssm_state": ["ModelCache", "Tensor", "class", "def", "device", "int", "layer_idx", "new_ssm_state", "return", "self", "ssm_states", "to", "torch", "update_ssm_state"], "mamba2/modeling_mamba2.py:Mamba2Cache.reset": ["ModelCache", "class", "conv_states", "def", "reset", "self", "ssm_states", "zero_"], "mamba2/modeling_mamba2.py:MambaRMSNormGated.__init__": ["ModelRMSNormGated", "Module", "Parameter", "__init__", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "mamba2/modeling_mamba2.py:MambaRMSNormGated.forward": ["ModelRMSNormGated", "Module", "None", "True", "class", "def", "dtype", "float32", "forward", "functional", "gate", "hidden_states", "if", "input_dtype", "is", "keepdim", "mean", "nn", "not", "pow", "return", "rsqrt", "self", "silu", "to", "torch", "variance", "variance_epsilon", "weight"], "mamba2/modeling_mamba2.py:Mamba2Mixer.__init__": ["A", "ACT2FN", "AILab", "A_log", "Conv1d", "D", "Dao", "Falling", "Linear", "Model", "ModelConfig", "ModelMixer", "ModelRMSNormGated", "Model_chunk_scan_combined", "Model_split_conv1d_scan_combined", "Model_ssm", "Module", "None", "Parameter", "The", "To", "__init__", "act", "activation", "all", "arange", "available", "back", "because", "bias", "causal", "causal_conv1d", "causal_conv1d_fn", "causal_conv1d_update", "chunk_size", "class", "com", "config", "conv1d", "conv_dim", "conv_kernel", "conv_kernel_size", "def", "dt_bias", "eps", "expand", "fast", "follow", "getattr", "github", "global", "groups", "head_dim", "hidden_act", "hidden_size", "https", "if", "implementation", "in_channels", "in_proj", "install", "int", "intermediate_size", "is", "is_fast_path_available", "kernel_size", "layer_idx", "layer_norm_epsilon", "lazy_load_kernel", "log", "logger", "n_groups", "naive", "nn", "norm", "not", "num_heads", "of", "one", "ones", "out_channels", "out_proj", "padding", "path", "projection_size", "rms_norm", "selective_state_update", "self", "spaces", "ssm", "ssm_state_size", "state", "state_size", "super", "the", "time_step_limit", "time_step_max", "time_step_min", "time_step_rank", "to", "torch", "use_bias", "use_conv_bias", "warning_once"], "mamba2/modeling_mamba2.py:Mamba2Mixer.cuda_kernels_forward": ["A", "A_log", "B", "C", "D", "False", "LongTensor", "ModelCache", "ModelMixer", "Model_chunk_scan_combined", "Model_split_conv1d_scan_combined", "Module", "None", "Tensor", "True", "_", "act", "activation", "and", "apply_mask_to_padding_states", "attention_mask", "batch_size", "bias", "cache_init", "cache_params", "cache_position", "causal_conv1d_fn", "causal_conv1d_update", "chunk_size", "class", "conv1d", "conv_dim", "conv_kernel_size", "conv_states", "cuda_kernels_forward", "d_mlp", "def", "dim", "dt", "dt_bias", "dt_limit", "dt_limit_kwargs", "dt_softplus", "dtype", "else", "exp", "expand", "float", "float32", "functional", "gate", "groups_time_state_size", "head_dim", "headdim", "hidden_states", "hidden_states_B_C", "hidden_states_B_C_transposed", "hidden_states_reshaped", "if", "in", "in_proj", "inf", "intermediate_size", "is", "layer_idx", "n_groups", "new_conv_state", "new_ssm_state", "ngroups", "nn", "norm", "norm_before_gate", "not", "num_heads", "out", "out_proj", "outproj_bias", "outproj_weight", "pad", "projected_states", "return", "return_final_states", "rmsnorm_eps", "rmsnorm_weight", "scan_output", "selective_state_update", "self", "seq_idx", "seq_len", "shape", "silu", "split", "squeeze", "ssm_state", "ssm_state_size", "ssm_states", "swish", "time_step_limit", "to", "torch", "training", "transpose", "update_conv_state", "update_ssm_state", "variance_epsilon", "view", "weight", "x", "z"], "mamba2/modeling_mamba2.py:Mamba2Mixer.torch_forward": ["A", "A_cumsum", "A_log", "B", "B_decay", "C", "C_reshaped", "C_times_states", "D", "D_residual", "False", "G", "G_intermediate", "L", "LongTensor", "M", "M_intermediate", "ModelCache", "ModelMixer", "Module", "None", "Tensor", "True", "Y_diag", "Y_off", "_", "act", "and", "apply_mask_to_padding_states", "attention_mask", "batch_size", "bias", "bmm", "cache_device", "cache_init", "cache_params", "cache_position", "cat", "chunk_size", "clamp", "class", "contextualized_states", "contiguous", "conv1d", "conv_dim", "conv_kernel_size", "conv_states", "cumsum", "dA", "dB", "dBx", "d_mlp", "decay_chunk", "decay_states", "def", "device", "dim", "dt", "dt_bias", "dtype", "else", "exp", "expand", "float", "float32", "for", "functional", "gate", "head_dim", "hidden_states", "hidden_states_B_C", "hidden_states_B_C_transposed", "if", "in", "in_proj", "intermediate_size", "is", "layer_idx", "n_groups", "new_conv_state", "new_ssm_state", "new_states", "nn", "norm", "not", "num_heads", "out_proj", "output_size", "pad", "pad_size", "pad_tensor_by_size", "permute", "previous_states", "projected_states", "repeat_interleave", "reshape", "reshape_into_chunks", "return", "scan_output", "segment_sum", "self", "seq_len", "shape", "softplus", "split", "squeeze", "ssm_state", "ssm_state_size", "ssm_states", "ssm_states_reshaped", "state_decay_out", "state_decay_out_permuted", "states", "sum", "t", "time_step_limit", "to", "torch", "torch_forward", "transpose", "update_conv_state", "update_ssm_state", "use_conv_bias", "view", "weight", "y", "zeros_like"], "mamba2/modeling_mamba2.py:Mamba2Mixer.forward": ["LongTensor", "ModelCache", "ModelMixer", "Module", "None", "Tensor", "and", "attention_mask", "cache_params", "cache_position", "class", "cuda", "cuda_kernels_forward", "def", "device", "forward", "hidden_states", "if", "in", "in_proj", "is_fast_path_available", "nn", "return", "self", "torch", "torch_forward", "type", "weight"], "mamba2/modeling_mamba2.py:Mamba2RMSNorm.__init__": ["ModelRMSNorm", "Module", "Parameter", "__init__", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "mamba2/modeling_mamba2.py:Mamba2RMSNorm.forward": ["ModelRMSNorm", "Module", "True", "class", "def", "dtype", "float32", "forward", "hidden_states", "input_dtype", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "mamba2/modeling_mamba2.py:Mamba2Block.__init__": ["GradientCheckpointingLayer", "ModelBlock", "ModelMixer", "ModelRMSNorm", "__init__", "class", "config", "def", "eps", "hidden_size", "layer_idx", "layer_norm_epsilon", "mixer", "norm", "residual_in_fp32", "self", "super"], "mamba2/modeling_mamba2.py:Mamba2Block.forward": ["GradientCheckpointingLayer", "LongTensor", "ModelBlock", "ModelCache", "None", "Tensor", "attention_mask", "cache_params", "cache_position", "class", "def", "dtype", "float32", "forward", "hidden_states", "if", "mixer", "norm", "residual", "residual_in_fp32", "return", "self", "to", "torch", "weight"], "mamba2/modeling_mamba2.py:Mamba2PreTrainedModel._init_weights": ["A", "A_log", "D", "Embedding", "Linear", "ModelMixer", "ModelPreTrainedModel", "ModelRMSNorm", "ModelRMSNormGated", "None", "PreTrainedModel", "_init_weights", "a", "arange", "bias", "clamp", "class", "config", "conv1d", "copy_", "def", "dt", "dt_bias", "elif", "exp", "expm1", "if", "init", "initializer_range", "inv_dt", "is", "isinstance", "kaiming_uniform_", "log", "math", "min", "module", "nn", "normal_", "not", "num_heads", "num_hidden_layers", "ones_", "out_proj", "p", "rand", "rescale_prenorm_residual", "self", "sqrt", "std", "time_step_floor", "time_step_max", "time_step_min", "torch", "weight", "zeros_"], "mamba2/modeling_mamba2.py:Mamba2Model.__init__": ["Embedding", "False", "ModelBlock", "ModelModel", "ModelPreTrainedModel", "ModelRMSNorm", "ModuleList", "__init__", "_register_load_state_dict_pre_hook", "class", "config", "def", "embeddings", "eps", "for", "gradient_checkpointing", "hidden_size", "idx", "in", "layer_idx", "layer_norm_epsilon", "layers", "load_hook", "nn", "norm_f", "num_hidden_layers", "post_init", "range", "self", "super", "vocab_size"], "mamba2/modeling_mamba2.py:Mamba2Model.load_hook": ["ModelModel", "ModelPreTrainedModel", "args", "break", "class", "def", "embedding", "embeddings", "for", "if", "in", "k", "load_hook", "pop", "prefix", "replace", "self", "state_dict"], "mamba2/modeling_mamba2.py:Mamba2Model.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "return", "self"], "mamba2/modeling_mamba2.py:Mamba2Model.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "new_embeddings", "self", "set_input_embeddings"], "mamba2/modeling_mamba2.py:Mamba2Model.forward": ["False", "LongTensor", "ModelCache", "ModelModel", "ModelOutput", "ModelPreTrainedModel", "None", "Tensor", "True", "ValueError", "You", "a", "all_hidden_states", "and", "arange", "are", "attention_mask", "automatically", "be", "because", "bool", "cache_params", "cache_position", "case", "class", "config", "conv_kernel", "def", "device", "don", "dtype", "elif", "else", "embeddings", "exactly", "for", "forward", "gradient_checkpointing", "have", "hidden_states", "if", "in", "initialized", "input_ids", "inputs_embeds", "is", "it", "kwargs", "last_hidden_state", "layers", "manually", "mixer_block", "must", "norm_f", "not", "of", "one", "or", "output_hidden_states", "pass", "passed", "prefilling", "r", "raise", "return", "return_dict", "self", "size", "specify", "stage", "t", "that", "the", "to", "torch", "training", "tuple", "use_cache", "use_return_dict", "v", "when", "will", "you"], "mamba2/modeling_mamba2.py:Mamba2ForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "ModelForCausalLM", "ModelModel", "ModelPreTrainedModel", "__init__", "backbone", "bias", "class", "config", "def", "hidden_size", "lm_head", "nn", "post_init", "self", "super", "vocab_size"], "mamba2/modeling_mamba2.py:Mamba2ForCausalLM.get_input_embeddings": ["GenerationMixin", "ModelForCausalLM", "ModelPreTrainedModel", "backbone", "class", "def", "get_input_embeddings", "return", "self"], "mamba2/modeling_mamba2.py:Mamba2ForCausalLM.set_input_embeddings": ["GenerationMixin", "ModelForCausalLM", "ModelPreTrainedModel", "backbone", "class", "def", "new_embeddings", "return", "self", "set_input_embeddings"], "mamba2/modeling_mamba2.py:Mamba2ForCausalLM.prepare_inputs_for_generation": ["False", "GenerationMixin", "LongTensor", "ModelCache", "ModelForCausalLM", "ModelPreTrainedModel", "None", "Tensor", "and", "arange", "attention_mask", "backbone", "bool", "cache_params", "cache_position", "class", "config", "contiguous", "conv_kernel", "def", "device", "dtype", "else", "for", "if", "in", "input_ids", "inputs_embeds", "is", "is_first_iteration", "items", "key", "kwargs", "max_batch_size", "model_inputs", "not", "prepare_inputs_for_generation", "return", "self", "size", "torch", "unsqueeze", "update", "use_cache", "value"], "mamba2/modeling_mamba2.py:Mamba2ForCausalLM.forward": ["FloatTensor", "GenerationMixin", "LongTensor", "ModelCache", "ModelCausalLMOutput", "ModelForCausalLM", "ModelPreTrainedModel", "Model_outputs", "None", "Tensor", "attention_mask", "backbone", "bool", "cache_params", "cache_position", "class", "config", "def", "dtype", "else", "float", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "not", "output", "output_hidden_states", "r", "return", "return_dict", "self", "slice", "slice_indices", "to", "torch", "tuple", "use_cache", "use_return_dict", "vocab_size", "weight"], "marian/modeling_marian.py:shift_tokens_right": ["Model_tokens_right", "Modeled_input_ids", "None", "Tensor", "ValueError", "be", "clone", "config", "decoder_start_token_id", "def", "defined", "has", "if", "input_ids", "int", "is", "masked_fill_", "model", "new_zeros", "pad_token_id", "raise", "return", "self", "shape", "to", "torch"], "marian/modeling_marian.py:MarianSinusoidalPositionalEmbedding.__init__": ["Embedding", "ModelSinusoidalPositionalEmbedding", "None", "True", "__init__", "_freeze", "class", "def", "embedding_dim", "int", "nn", "num_positions", "padding_idx", "self", "super"], "marian/modeling_marian.py:MarianSinusoidalPositionalEmbedding.create_weight": ["Embedding", "False", "FloatTensor", "ModelSinusoidalPositionalEmbedding", "array", "class", "cos", "create_weight", "def", "dim", "dtype", "else", "empty", "for", "if", "in", "j", "n_pos", "nn", "np", "out", "pos", "position_enc", "power", "range", "requires_grad", "return", "self", "sentinel", "shape", "sin", "torch", "weight"], "marian/modeling_marian.py:MarianSinusoidalPositionalEmbedding.forward": ["Embedding", "ModelSinusoidalPositionalEmbedding", "None", "Size", "Tensor", "arange", "bsz", "class", "def", "device", "dtype", "forward", "if", "input_ids_shape", "int", "is", "long", "nn", "past_key_values_length", "position_ids", "return", "self", "seq_len", "super", "torch", "weight"], "marian/modeling_marian.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "contiguous", "def", "dim", "dropout", "float", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "shape", "size", "softmax", "torch", "training", "transpose", "value"], "marian/modeling_marian.py:MarianAttention.__init__": ["False", "Instantiating", "Linear", "ModelAttention", "ModelConfig", "Module", "None", "Please", "True", "ValueError", "__class__", "__init__", "__name__", "a", "and", "be", "bias", "bool", "by", "caching", "call", "class", "config", "creating", "decoder", "def", "divisible", "dropout", "during", "embed_dim", "errors", "f", "float", "forward", "got", "head_dim", "if", "int", "is", "is_causal", "is_decoder", "k_proj", "layer_idx", "lead", "logger", "make", "must", "nn", "not", "num_heads", "out_proj", "passing", "provide", "q_proj", "raise", "recommended", "scaling", "self", "super", "sure", "the", "this", "to", "used", "v_proj", "warning_once", "when", "will", "without"], "marian/modeling_marian.py:MarianAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "EncoderDecoderCache", "False", "FlashAttentionKwargs", "ModelAttention", "Module", "None", "Tensor", "True", "Unpack", "_attn_implementation", "and", "attention_interface", "attention_mask", "attn_output", "attn_weights", "bool", "bsz", "cache_position", "class", "config", "contiguous", "cross_attention_cache", "curr_past_key_values", "current_states", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "get", "head_dim", "hidden_states", "if", "is", "is_cross_attention", "is_updated", "isinstance", "k_proj", "key_states", "key_value_states", "keys", "kv_input_shape", "kwargs", "layer_idx", "layers", "nn", "not", "out_proj", "output_attentions", "past_key_values", "q_input_shape", "q_proj", "query_states", "reshape", "return", "scaling", "self", "self_attention_cache", "shape", "src_len", "tgt_len", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "values", "view"], "marian/modeling_marian.py:MarianEncoderLayer.__init__": ["ACT2FN", "GradientCheckpointingLayer", "LayerNorm", "Linear", "ModelAttention", "ModelConfig", "ModelEncoderLayer", "None", "__init__", "activation_dropout", "activation_fn", "activation_function", "attention_dropout", "class", "config", "d_model", "def", "dropout", "embed_dim", "encoder_attention_heads", "encoder_ffn_dim", "fc1", "fc2", "final_layer_norm", "int", "layer_idx", "nn", "num_heads", "self", "self_attn", "self_attn_layer_norm", "super"], "marian/modeling_marian.py:MarianEncoderLayer.forward": ["False", "FloatTensor", "GradientCheckpointingLayer", "ModelEncoderLayer", "None", "activation_dropout", "activation_fn", "and", "any", "attention_mask", "attn_weights", "bool", "clamp", "clamp_value", "class", "def", "dropout", "dtype", "fc1", "fc2", "final_layer_norm", "finfo", "float16", "forward", "functional", "hidden_states", "if", "isinf", "isnan", "max", "min", "nn", "or", "output_attentions", "outputs", "p", "residual", "return", "self", "self_attn", "self_attn_layer_norm", "torch", "training", "tuple"], "marian/modeling_marian.py:MarianDecoderLayer.__init__": ["ACT2FN", "GradientCheckpointingLayer", "LayerNorm", "Linear", "ModelAttention", "ModelConfig", "ModelDecoderLayer", "None", "True", "__init__", "activation_dropout", "activation_fn", "activation_function", "attention_dropout", "class", "config", "d_model", "decoder_attention_heads", "decoder_ffn_dim", "def", "dropout", "embed_dim", "encoder_attn", "encoder_attn_layer_norm", "fc1", "fc2", "final_layer_norm", "int", "is_causal", "is_decoder", "layer_idx", "nn", "num_heads", "self", "self_attn", "self_attn_layer_norm", "super"], "marian/modeling_marian.py:MarianDecoderLayer.forward": ["Cache", "False", "FloatTensor", "GradientCheckpointingLayer", "ModelDecoderLayer", "None", "Tensor", "True", "activation_dropout", "activation_fn", "attention_mask", "bool", "cache_position", "class", "cross_attn_weights", "def", "dropout", "encoder_attention_mask", "encoder_attn", "encoder_attn_layer_norm", "encoder_hidden_states", "fc1", "fc2", "final_layer_norm", "forward", "functional", "hidden_states", "if", "is", "key_value_states", "nn", "not", "output_attentions", "outputs", "p", "past_key_values", "residual", "return", "self", "self_attn", "self_attn_layer_norm", "self_attn_weights", "torch", "training", "tuple", "use_cache"], "marian/modeling_marian.py:MarianPreTrainedModel._init_weights": ["ModelMTModel", "ModelPreTrainedModel", "ModelSinusoidalPositionalEmbedding", "PreTrainedModel", "_init_weights", "class", "copy_", "create_weight", "def", "elif", "final_logits_bias", "if", "init", "isinstance", "module", "self", "super", "weight", "zeros_"], "marian/modeling_marian.py:MarianPreTrainedModel.dummy_inputs": ["ModelPreTrainedModel", "PreTrainedModel", "attention_mask", "class", "config", "decoder_input_ids", "def", "device", "dummy_inputs", "input_ids", "ne", "pad_token", "pad_token_id", "return", "self", "tensor", "torch"], "marian/modeling_marian.py:MarianEncoder.__init__": ["Embedding", "False", "ModelConfig", "ModelEncoder", "ModelEncoderLayer", "ModelPreTrainedModel", "ModelSinusoidalPositionalEmbedding", "ModuleList", "_", "__init__", "class", "config", "d_model", "def", "dropout", "else", "embed_dim", "embed_positions", "embed_scale", "embed_tokens", "encoder_layerdrop", "encoder_layers", "for", "gradient_checkpointing", "if", "in", "layerdrop", "layers", "math", "max_position_embeddings", "max_source_positions", "nn", "pad_token_id", "padding_idx", "post_init", "range", "scale_embedding", "self", "sqrt", "super", "vocab_size"], "marian/modeling_marian.py:MarianEncoder.forward": ["BaseModelOutput", "False", "FloatTensor", "LongTensor", "ModelEncoder", "ModelPreTrainedModel", "None", "Tensor", "True", "ValueError", "You", "all_attentions", "and", "at", "attention_mask", "attentions", "bool", "both", "cannot", "class", "config", "create_bidirectional_mask", "def", "dropout", "dropout_probability", "either", "elif", "else", "embed_pos", "embed_positions", "embed_scale", "embed_tokens", "encoder_layer", "encoder_states", "enumerate", "for", "forward", "functional", "have", "hidden_states", "idx", "if", "in", "input_embeds", "input_ids", "input_shape", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layer_outputs", "layerdrop", "layers", "nn", "not", "or", "output_attentions", "output_hidden_states", "p", "r", "raise", "rand", "return", "return_dict", "same", "self", "size", "specify", "the", "time", "to", "to_drop", "torch", "training", "tuple", "use_return_dict", "v", "view", "warn_if_padding_and_no_attention_mask"], "marian/modeling_marian.py:MarianDecoder.__init__": ["Embedding", "False", "ModelConfig", "ModelDecoder", "ModelDecoderLayer", "ModelPreTrainedModel", "ModelSinusoidalPositionalEmbedding", "ModuleList", "__init__", "class", "config", "d_model", "decoder_layerdrop", "decoder_layers", "decoder_vocab_size", "def", "dropout", "else", "embed_positions", "embed_scale", "embed_tokens", "for", "gradient_checkpointing", "i", "if", "in", "layer_idx", "layerdrop", "layers", "math", "max_position_embeddings", "max_target_positions", "nn", "pad_token_id", "padding_idx", "post_init", "range", "scale_embedding", "self", "sqrt", "super"], "marian/modeling_marian.py:MarianDecoder.forward": ["BaseModelOutputWithPastAndCrossAttentions", "Cache", "DynamicCache", "EncoderDecoderCache", "False", "FloatTensor", "LongTensor", "ModelDecoder", "ModelPreTrainedModel", "None", "Setting", "Tensor", "True", "ValueError", "You", "all_cross_attentions", "all_hidden_states", "all_self_attns", "and", "arange", "attention_mask", "attentions", "batch_size", "bool", "cache_position", "causal_mask", "checkpointing", "class", "config", "continue", "create_bidirectional_mask", "create_causal_mask", "cross_attentions", "decoder_layer", "def", "device", "dropout", "dropout_probability", "elif", "else", "embed_positions", "embed_scale", "embed_tokens", "encoder_attention_mask", "encoder_hidden_states", "enumerate", "exactly", "for", "forward", "functional", "get_seq_length", "gradient", "gradient_checkpointing", "hidden_states", "idx", "if", "in", "incompatible", "input", "input_embeds", "input_ids", "input_shape", "inputs_embeds", "is", "is_encoder_decoder", "is_torchdynamo_compiling", "isinstance", "kwargs", "last_hidden_state", "layer_outputs", "layerdrop", "layers", "logger", "mask_seq_length", "must", "nn", "not", "of", "one", "ones", "or", "output_attentions", "output_hidden_states", "p", "past_key_values", "past_key_values_length", "position_ids", "r", "raise", "rand", "return", "return_dict", "self", "self_attention_cache", "self_attn_cache", "seq_length", "shape", "size", "specify", "torch", "training", "tuple", "use_cache", "use_return_dict", "v", "view", "warning_once", "with"], "marian/modeling_marian.py:MarianModel.__init__": ["Embedding", "ModelConfig", "ModelDecoder", "ModelEncoder", "ModelModel", "ModelPreTrainedModel", "None", "__init__", "_tied_weights_keys", "class", "config", "d_model", "decoder", "def", "else", "embed_tokens", "encoder", "if", "nn", "pad_token_id", "padding_idx", "post_init", "self", "share_encoder_decoder_embeddings", "shared", "super", "vocab_size", "weight"], "marian/modeling_marian.py:MarianModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "get_encoder", "get_input_embeddings", "return", "self"], "marian/modeling_marian.py:MarianModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "config", "decoder", "def", "else", "embed_tokens", "encoder", "if", "self", "set_input_embeddings", "share_encoder_decoder_embeddings", "shared", "value"], "marian/modeling_marian.py:MarianModel.get_decoder_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "Please", "True", "ValueError", "be", "called", "class", "config", "def", "get_decoder", "get_decoder_input_embeddings", "get_input_embeddings", "if", "instead", "is", "not", "raise", "return", "self", "share_encoder_decoder_embeddings", "should", "use"], "marian/modeling_marian.py:MarianModel.set_decoder_input_embeddings": ["In", "ModelModel", "ModelPreTrainedModel", "True", "ValueError", "appropriate", "are", "by", "calling", "class", "config", "decoder", "def", "embed_tokens", "embeddings", "encoder", "if", "input", "is", "meaning", "order", "raise", "self", "set", "set_decoder_input_embeddings", "set_input_embeddings", "share_encoder_decoder_embeddings", "shared", "should", "simply", "the", "to", "value", "with", "you"], "marian/modeling_marian.py:MarianModel.resize_decoder_token_embeddings": ["Embedding", "ModelModel", "ModelPreTrainedModel", "None", "Please", "True", "ValueError", "_get_resized_embeddings", "be", "called", "class", "config", "decoder_vocab_size", "def", "get_decoder_input_embeddings", "if", "instead", "int", "is", "model_embeds", "new_embeddings", "new_num_tokens", "nn", "not", "old_embeddings", "raise", "resize_decoder_token_embeddings", "resize_token_embeddings", "return", "self", "set_decoder_input_embeddings", "share_encoder_decoder_embeddings", "should", "tie_weights", "use"], "marian/modeling_marian.py:MarianModel.forward": ["BaseModelOutput", "Cache", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Seq2SeqModelOutput", "Tensor", "and", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "cross_attentions", "decoder", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_input_ids", "decoder_inputs_embeds", "decoder_outputs", "def", "elif", "else", "encoder", "encoder_attention_mask", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "isinstance", "kwargs", "last_hidden_state", "len", "not", "output_attentions", "output_hidden_states", "past_key_values", "r", "return", "return_dict", "self", "torch", "tuple", "use_cache", "use_return_dict"], "marian/modeling_marian.py:MarianMTModel.__init__": ["False", "GenerationMixin", "Linear", "ModelConfig", "ModelMTModel", "ModelModel", "ModelPreTrainedModel", "__init__", "_tied_weights_keys", "bias", "class", "config", "d_model", "decoder", "decoder_vocab_size", "def", "else", "embed_tokens", "encoder", "final_logits_bias", "if", "lm_head", "model", "nn", "post_init", "register_buffer", "self", "share_encoder_decoder_embeddings", "shared", "super", "target_vocab_size", "torch", "vocab_size", "weight", "zeros"], "marian/modeling_marian.py:MarianMTModel.resize_token_embeddings": ["Embedding", "GenerationMixin", "ModelMTModel", "ModelPreTrainedModel", "None", "True", "_resize_final_logits_bias", "bool", "class", "config", "def", "if", "int", "mean_resizing", "new_embeddings", "new_num_tokens", "nn", "pad_to_multiple_of", "resize_token_embeddings", "return", "self", "share_encoder_decoder_embeddings", "super"], "marian/modeling_marian.py:MarianMTModel._resize_token_embeddings": ["Embedding", "GenerationMixin", "ModelMTModel", "ModelPreTrainedModel", "None", "_get_resized_embeddings", "_get_resized_lm_head", "_resize_token_embeddings", "and", "args", "class", "config", "decoder_vocab_size", "def", "get_input_embeddings", "get_output_embeddings", "if", "int", "is", "new_embeddings", "new_lm_head", "new_num_tokens", "nn", "not", "old_embeddings", "old_lm_head", "pad_to_multiple_of", "return", "self", "set_input_embeddings", "set_output_embeddings", "shape", "share_encoder_decoder_embeddings", "tie_word_embeddings", "weight"], "marian/modeling_marian.py:MarianMTModel.resize_decoder_token_embeddings": ["GenerationMixin", "ModelMTModel", "ModelPreTrainedModel", "None", "Please", "True", "ValueError", "_get_resized_embeddings", "_get_resized_lm_head", "_resize_final_logits_bias", "and", "be", "called", "class", "config", "decoder_vocab_size", "def", "get_decoder_input_embeddings", "get_output_embeddings", "if", "instead", "is", "model", "model_embeds", "new_embeddings", "new_lm_head", "new_num_tokens", "not", "old_embeddings", "old_lm_head", "raise", "resize_decoder_token_embeddings", "resize_token_embeddings", "return", "self", "set_decoder_input_embeddings", "set_output_embeddings", "share_encoder_decoder_embeddings", "should", "tie_weights", "tie_word_embeddings", "use"], "marian/modeling_marian.py:MarianMTModel._resize_final_logits_bias": ["GenerationMixin", "ModelMTModel", "ModelPreTrainedModel", "None", "_resize_final_logits_bias", "cat", "class", "def", "device", "dim", "else", "extra_bias", "final_logits_bias", "if", "int", "new_bias", "new_num_tokens", "old_num_tokens", "register_buffer", "self", "shape", "torch", "zeros"], "marian/modeling_marian.py:MarianMTModel.set_output_embeddings": ["Embedding", "GenerationMixin", "ModelMTModel", "ModelPreTrainedModel", "class", "def", "lm_head", "new_embeddings", "nn", "self", "set_output_embeddings"], "marian/modeling_marian.py:MarianMTModel.forward": ["BaseModelOutput", "Cache", "CrossEntropyLoss", "False", "FloatTensor", "GenerationMixin", "LongTensor", "ModelMTModel", "ModelPreTrainedModel", "None", "Seq2SeqLMOutput", "Tensor", "The", "and", "argument", "attention_mask", "bool", "cache_position", "changed", "class", "config", "cross_attentions", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_input_ids", "decoder_inputs_embeds", "decoder_start_token_id", "decoder_vocab_size", "def", "else", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "final_logits_bias", "forward", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "lm_head", "lm_logits", "logger", "logits", "loss", "loss_fct", "masked_lm_loss", "model", "not", "output", "output_attentions", "output_hidden_states", "outputs", "pad_token_id", "past_key_values", "provided", "r", "return", "return_dict", "self", "shift_tokens_right", "since", "to", "torch", "tuple", "use_cache", "use_return_dict", "view", "warning"], "marian/modeling_marian.py:MarianMTModel.prepare_decoder_input_ids_from_labels": ["GenerationMixin", "ModelMTModel", "ModelPreTrainedModel", "Tensor", "class", "config", "decoder_start_token_id", "def", "labels", "pad_token_id", "prepare_decoder_input_ids_from_labels", "return", "self", "shift_tokens_right", "torch"], "marian/modeling_marian.py:MarianDecoderWrapper.__init__": ["ModelDecoder", "ModelDecoderWrapper", "ModelPreTrainedModel", "__init__", "class", "config", "decoder", "def", "post_init", "self", "super"], "marian/modeling_marian.py:MarianDecoderWrapper.forward": ["ModelDecoderWrapper", "ModelPreTrainedModel", "args", "class", "decoder", "def", "forward", "kwargs", "return", "self"], "marian/modeling_marian.py:MarianForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "ModelDecoderWrapper", "ModelForCausalLM", "ModelPreTrainedModel", "True", "__init__", "bias", "class", "config", "def", "hidden_size", "is_decoder", "is_encoder_decoder", "lm_head", "model", "nn", "post_init", "self", "super", "vocab_size"], "marian/modeling_marian.py:MarianForCausalLM.get_input_embeddings": ["GenerationMixin", "ModelForCausalLM", "ModelPreTrainedModel", "class", "decoder", "def", "embed_tokens", "get_input_embeddings", "model", "return", "self"], "marian/modeling_marian.py:MarianForCausalLM.set_input_embeddings": ["GenerationMixin", "ModelForCausalLM", "ModelPreTrainedModel", "class", "decoder", "def", "embed_tokens", "model", "self", "set_input_embeddings", "value"], "marian/modeling_marian.py:MarianForCausalLM.forward": ["Cache", "CausalLMOutputWithCrossAttentions", "CrossEntropyLoss", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForCausalLM", "ModelPreTrainedModel", "None", "Tensor", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "cross_attentions", "decoder", "def", "device", "else", "encoder_attention_mask", "encoder_hidden_states", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "lm_head", "logits", "logits_to_keep", "loss", "loss_fct", "model", "not", "output", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "r", "return", "return_dict", "self", "slice", "slice_indices", "to", "torch", "tuple", "use_cache", "use_return_dict", "view", "vocab_size"], "markuplm/modeling_markuplm.py:XPathEmbeddings.__init__": ["Dropout", "Embedding", "Linear", "Model", "Module", "ModuleList", "ReLU", "_", "__init__", "activation", "class", "config", "def", "dropout", "for", "hidden_dropout_prob", "hidden_size", "in", "inner2emb", "max_depth", "max_xpath_subs_unit_embeddings", "max_xpath_tag_unit_embeddings", "nn", "range", "self", "super", "xpath_subs_sub_embeddings", "xpath_tag_sub_embeddings", "xpath_unit_hidden_size", "xpath_unitseq2_embeddings", "xpath_unitseq2_inner"], "markuplm/modeling_markuplm.py:XPathEmbeddings.forward": ["Model", "Module", "None", "activation", "append", "cat", "class", "def", "dim", "dropout", "for", "forward", "i", "in", "inner2emb", "max_depth", "nn", "range", "return", "self", "torch", "xpath_embeddings", "xpath_subs_embeddings", "xpath_subs_seq", "xpath_subs_sub_embeddings", "xpath_tag_sub_embeddings", "xpath_tags_embeddings", "xpath_tags_seq", "xpath_unitseq2_inner"], "markuplm/modeling_markuplm.py:MarkupLMEmbeddings.__init__": ["Dropout", "Embedding", "False", "LayerNorm", "ModelEmbeddings", "Module", "XPathEmbeddings", "__init__", "arange", "class", "config", "def", "dropout", "eps", "expand", "hidden_dropout_prob", "hidden_size", "layer_norm_eps", "max_depth", "max_position_embeddings", "nn", "pad_token_id", "padding_idx", "persistent", "position_embeddings", "position_ids", "register_buffer", "self", "super", "token_type_embeddings", "torch", "type_vocab_size", "vocab_size", "word_embeddings", "xpath_embeddings"], "markuplm/modeling_markuplm.py:MarkupLMEmbeddings.create_position_ids_from_inputs_embeds": ["ModelEmbeddings", "Module", "arange", "class", "create_position_ids_from_inputs_embeds", "def", "device", "dtype", "expand", "input_shape", "inputs_embeds", "long", "nn", "padding_idx", "position_ids", "return", "sequence_length", "size", "torch", "unsqueeze"], "markuplm/modeling_markuplm.py:MarkupLMEmbeddings.create_position_ids_from_input_ids": ["ModelEmbeddings", "Module", "class", "create_position_ids_from_input_ids", "cumsum", "def", "dim", "incremental_indices", "input_ids", "int", "long", "mask", "ne", "nn", "padding_idx", "past_key_values_length", "return", "torch", "type_as"], "markuplm/modeling_markuplm.py:MarkupLMEmbeddings.forward": ["LayerNorm", "ModelEmbeddings", "Module", "None", "class", "config", "create_position_ids_from_input_ids", "create_position_ids_from_inputs_embeds", "def", "device", "dropout", "dtype", "else", "embeddings", "forward", "if", "input_ids", "input_shape", "inputs_embeds", "is", "list", "long", "max_depth", "nn", "not", "ones", "padding_idx", "position_embeddings", "position_ids", "return", "self", "size", "subs_pad_id", "tag_pad_id", "token_type_embeddings", "token_type_ids", "torch", "tuple", "word_embeddings", "words_embeddings", "xpath_embeddings", "xpath_subs_seq", "xpath_tags_seq", "zeros"], "markuplm/modeling_markuplm.py:MarkupLMSelfOutput.__init__": ["Dropout", "LayerNorm", "Linear", "ModelSelfOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "eps", "hidden_dropout_prob", "hidden_size", "layer_norm_eps", "nn", "self", "super"], "markuplm/modeling_markuplm.py:MarkupLMSelfOutput.forward": ["LayerNorm", "ModelSelfOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "markuplm/modeling_markuplm.py:MarkupLMIntermediate.__init__": ["ACT2FN", "Linear", "ModelIntermediate", "Module", "__init__", "class", "config", "def", "dense", "else", "hidden_act", "hidden_size", "if", "intermediate_act_fn", "intermediate_size", "isinstance", "nn", "self", "str", "super"], "markuplm/modeling_markuplm.py:MarkupLMIntermediate.forward": ["ModelIntermediate", "Module", "Tensor", "class", "def", "dense", "forward", "hidden_states", "intermediate_act_fn", "nn", "return", "self", "torch"], "markuplm/modeling_markuplm.py:MarkupLMOutput.__init__": ["Dropout", "LayerNorm", "Linear", "ModelOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "eps", "hidden_dropout_prob", "hidden_size", "intermediate_size", "layer_norm_eps", "nn", "self", "super"], "markuplm/modeling_markuplm.py:MarkupLMOutput.forward": ["LayerNorm", "ModelOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "markuplm/modeling_markuplm.py:MarkupLMPooler.__init__": ["Linear", "ModelPooler", "Module", "Tanh", "__init__", "activation", "class", "config", "def", "dense", "hidden_size", "nn", "self", "super"], "markuplm/modeling_markuplm.py:MarkupLMPooler.forward": ["ModelPooler", "Module", "Tensor", "activation", "class", "def", "dense", "first_token_tensor", "forward", "hidden_states", "nn", "pooled_output", "return", "self", "torch"], "markuplm/modeling_markuplm.py:MarkupLMPredictionHeadTransform.__init__": ["ACT2FN", "LayerNorm", "Linear", "ModelPredictionHeadTransform", "Module", "__init__", "class", "config", "def", "dense", "else", "eps", "hidden_act", "hidden_size", "if", "isinstance", "layer_norm_eps", "nn", "self", "str", "super", "transform_act_fn"], "markuplm/modeling_markuplm.py:MarkupLMPredictionHeadTransform.forward": ["LayerNorm", "ModelPredictionHeadTransform", "Module", "Tensor", "class", "def", "dense", "forward", "hidden_states", "nn", "return", "self", "torch", "transform_act_fn"], "markuplm/modeling_markuplm.py:MarkupLMLMPredictionHead.__init__": ["Linear", "ModelLMPredictionHead", "ModelPredictionHeadTransform", "Module", "Parameter", "True", "__init__", "bias", "class", "config", "decoder", "def", "hidden_size", "nn", "self", "super", "torch", "transform", "vocab_size", "zeros"], "markuplm/modeling_markuplm.py:MarkupLMLMPredictionHead.forward": ["ModelLMPredictionHead", "Module", "class", "decoder", "def", "forward", "hidden_states", "nn", "return", "self", "transform"], "markuplm/modeling_markuplm.py:MarkupLMOnlyMLMHead.__init__": ["ModelLMPredictionHead", "ModelOnlyMLMHead", "Module", "__init__", "class", "config", "def", "nn", "predictions", "self", "super"], "markuplm/modeling_markuplm.py:MarkupLMOnlyMLMHead.forward": ["ModelOnlyMLMHead", "Module", "Tensor", "class", "def", "forward", "nn", "prediction_scores", "predictions", "return", "self", "sequence_output", "torch"], "markuplm/modeling_markuplm.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value"], "markuplm/modeling_markuplm.py:MarkupLMSelfAttention.__init__": ["Dropout", "Linear", "ModelSelfAttention", "Module", "The", "ValueError", "__init__", "a", "all_head_size", "and", "attention", "attention_dropout", "attention_head_size", "attention_probs_dropout_prob", "class", "config", "def", "dropout", "embedding_size", "f", "hasattr", "heads", "hidden", "hidden_size", "if", "int", "is", "key", "multiple", "nn", "not", "num_attention_heads", "number", "of", "query", "raise", "scaling", "self", "size", "super", "the", "value"], "markuplm/modeling_markuplm.py:MarkupLMSelfAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "False", "FloatTensor", "ModelSelfAttention", "Module", "None", "Tensor", "_attn_implementation", "attention_dropout", "attention_head_size", "attention_interface", "attention_mask", "attn_output", "attn_weights", "bool", "class", "config", "contiguous", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "hidden_shape", "hidden_states", "if", "input_shape", "key", "key_states", "kwargs", "nn", "not", "output_attentions", "outputs", "query", "query_states", "reshape", "return", "scaling", "self", "shape", "torch", "training", "transpose", "tuple", "value", "value_states", "view"], "markuplm/modeling_markuplm.py:MarkupLMAttention.__init__": ["ModelAttention", "ModelSelfAttention", "ModelSelfOutput", "Module", "__init__", "class", "config", "def", "nn", "output", "self", "super"], "markuplm/modeling_markuplm.py:MarkupLMAttention.forward": ["False", "FloatTensor", "ModelAttention", "Module", "None", "Tensor", "attention_mask", "attention_output", "bool", "class", "def", "forward", "hidden_states", "kwargs", "nn", "output", "output_attentions", "outputs", "return", "self", "self_outputs", "torch", "tuple"], "markuplm/modeling_markuplm.py:MarkupLMLayer.__init__": ["GradientCheckpointingLayer", "ModelAttention", "ModelIntermediate", "ModelLayer", "ModelOutput", "__init__", "attention", "chunk_size_feed_forward", "class", "config", "def", "intermediate", "output", "self", "seq_len_dim", "super"], "markuplm/modeling_markuplm.py:MarkupLMLayer.forward": ["False", "FloatTensor", "GradientCheckpointingLayer", "ModelLayer", "None", "Tensor", "apply_chunking_to_forward", "attention", "attention_mask", "attention_output", "bool", "chunk_size_feed_forward", "class", "def", "feed_forward_chunk", "forward", "hidden_states", "kwargs", "layer_output", "output_attentions", "outputs", "return", "self", "self_attention_outputs", "seq_len_dim", "torch", "tuple"], "markuplm/modeling_markuplm.py:MarkupLMLayer.feed_forward_chunk": ["GradientCheckpointingLayer", "ModelLayer", "attention_output", "class", "def", "feed_forward_chunk", "intermediate", "intermediate_output", "layer_output", "output", "return", "self"], "markuplm/modeling_markuplm.py:MarkupLMEncoder.__init__": ["False", "ModelEncoder", "ModelLayer", "Module", "ModuleList", "__init__", "class", "config", "def", "for", "gradient_checkpointing", "i", "in", "layer", "nn", "num_hidden_layers", "range", "self", "super"], "markuplm/modeling_markuplm.py:MarkupLMEncoder.forward": ["BaseModelOutput", "False", "FloatTensor", "ModelEncoder", "Module", "None", "Tensor", "True", "all_hidden_states", "all_self_attentions", "attention_mask", "attentions", "bool", "class", "def", "else", "enumerate", "for", "forward", "hidden_states", "i", "if", "in", "kwargs", "last_hidden_state", "layer", "layer_module", "layer_outputs", "nn", "output_attentions", "output_hidden_states", "return", "return_dict", "self", "torch", "tuple"], "markuplm/modeling_markuplm.py:MarkupLMPreTrainedModel._init_weights": ["ModelEmbeddings", "ModelLMPredictionHead", "ModelPreTrainedModel", "PreTrainedModel", "_init_weights", "arange", "bias", "class", "copy_", "def", "elif", "expand", "if", "init", "isinstance", "module", "position_ids", "self", "shape", "super", "torch", "zeros_"], "markuplm/modeling_markuplm.py:MarkupLMModel.__init__": ["ModelEmbeddings", "ModelEncoder", "ModelModel", "ModelPooler", "ModelPreTrainedModel", "None", "True", "__init__", "add_pooling_layer", "class", "config", "def", "else", "embeddings", "encoder", "if", "pooler", "post_init", "r", "self", "super"], "markuplm/modeling_markuplm.py:MarkupLMModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "return", "self", "word_embeddings"], "markuplm/modeling_markuplm.py:MarkupLMModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "self", "set_input_embeddings", "value", "word_embeddings"], "markuplm/modeling_markuplm.py:MarkupLMModel.forward": ["BaseModelOutputWithPooling", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "True", "ValueError", "You", "and", "at", "attention_mask", "attentions", "bool", "both", "cannot", "class", "config", "def", "device", "dtype", "either", "elif", "else", "embedding_output", "embeddings", "encoder", "encoder_outputs", "extended_attention_mask", "forward", "have", "hidden_states", "if", "input_ids", "input_shape", "inputs_embeds", "is", "kwargs", "last_hidden_state", "long", "not", "ones", "or", "output_attentions", "output_hidden_states", "pooled_output", "pooler", "pooler_output", "position_ids", "r", "raise", "return", "return_dict", "same", "self", "sequence_output", "size", "specify", "the", "time", "to", "token_type_ids", "torch", "tuple", "unsqueeze", "use_return_dict", "warn_if_padding_and_no_attention_mask", "xpath_subs_seq", "xpath_tags_seq", "zeros"], "markuplm/modeling_markuplm.py:MarkupLMForQuestionAnswering.__init__": ["False", "Linear", "Model", "ModelForQuestionAnswering", "ModelModel", "ModelPreTrainedModel", "__init__", "add_pooling_layer", "class", "config", "def", "hidden_size", "nn", "num_labels", "post_init", "qa_outputs", "self", "super"], "markuplm/modeling_markuplm.py:MarkupLMForQuestionAnswering.forward": ["CrossEntropyLoss", "Model", "ModelForQuestionAnswering", "ModelPreTrainedModel", "None", "QuestionAnsweringModelOutput", "Tensor", "True", "and", "attention_mask", "attentions", "bool", "clamp_", "class", "config", "contiguous", "def", "dim", "else", "end_logits", "end_loss", "end_positions", "forward", "hidden_states", "if", "ignore_index", "ignored_index", "input_ids", "inputs_embeds", "is", "kwargs", "len", "logits", "loss", "loss_fct", "not", "output_attentions", "output_hidden_states", "outputs", "position_ids", "qa_outputs", "r", "return", "return_dict", "self", "sequence_output", "size", "split", "squeeze", "start_logits", "start_loss", "start_positions", "token_type_ids", "torch", "total_loss", "tuple", "use_return_dict", "xpath_subs_seq", "xpath_tags_seq"], "markuplm/modeling_markuplm.py:MarkupLMForTokenClassification.__init__": ["Dropout", "False", "Linear", "Model", "ModelForTokenClassification", "ModelModel", "ModelPreTrainedModel", "None", "__init__", "add_pooling_layer", "class", "classifier", "classifier_dropout", "config", "def", "dropout", "else", "hidden_dropout_prob", "hidden_size", "if", "is", "nn", "not", "num_labels", "post_init", "self", "super"], "markuplm/modeling_markuplm.py:MarkupLMForTokenClassification.forward": ["CrossEntropyLoss", "MaskedLMOutput", "Model", "ModelForTokenClassification", "ModelPreTrainedModel", "None", "Tensor", "TokenClassifierOutput", "True", "attention_mask", "attentions", "bool", "class", "classifier", "config", "def", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "not", "num_labels", "output_attentions", "output_hidden_states", "outputs", "position_ids", "prediction_scores", "r", "return", "return_dict", "self", "sequence_output", "token_type_ids", "torch", "tuple", "use_return_dict", "view", "xpath_subs_seq", "xpath_tags_seq"], "markuplm/modeling_markuplm.py:MarkupLMForSequenceClassification.__init__": ["Dropout", "Linear", "Model", "ModelForSequenceClassification", "ModelModel", "ModelPreTrainedModel", "None", "__init__", "class", "classifier", "classifier_dropout", "config", "def", "dropout", "else", "hidden_dropout_prob", "hidden_size", "if", "is", "nn", "not", "num_labels", "post_init", "self", "super"], "markuplm/modeling_markuplm.py:MarkupLMForSequenceClassification.forward": ["BCEWithLogitsLoss", "CrossEntropyLoss", "MSELoss", "Model", "ModelForSequenceClassification", "ModelPreTrainedModel", "None", "SequenceClassifierOutput", "Tensor", "True", "and", "attention_mask", "attentions", "bool", "class", "classifier", "config", "def", "dropout", "dtype", "elif", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "kwargs", "labels", "logits", "long", "loss", "loss_fct", "multi_label_classification", "not", "num_labels", "or", "output_attentions", "output_hidden_states", "outputs", "pooled_output", "position_ids", "problem_type", "r", "regression", "return", "return_dict", "self", "single_label_classification", "squeeze", "token_type_ids", "torch", "tuple", "use_return_dict", "view", "xpath_subs_seq", "xpath_tags_seq"], "mask2former/modeling_mask2former.py:sample_point": ["False", "Model_point", "Tensor", "True", "add_dim", "def", "dim", "functional", "grid_Model", "if", "input_features", "kwargs", "nn", "point_coordinates", "point_features", "return", "squeeze", "torch", "unsqueeze"], "mask2former/modeling_mask2former.py:dice_loss": ["Model_loss", "Tensor", "def", "denominator", "flatten", "inputs", "int", "labels", "loss", "num_masks", "numerator", "probs", "r", "return", "sigmoid", "sum"], "mask2former/modeling_mask2former.py:sigmoid_cross_entropy_loss": ["BCEWithLogitsLoss", "Model_cross_entropy_loss", "Tensor", "criterion", "cross_entropy_loss", "def", "inputs", "int", "labels", "loss", "mean", "nn", "none", "num_masks", "r", "reduction", "return", "sum", "torch"], "mask2former/modeling_mask2former.py:pair_wise_dice_loss": ["Model_wise_dice_loss", "None", "T", "Tensor", "def", "denominator", "flatten", "inputs", "labels", "loss", "matmul", "numerator", "return", "sigmoid", "sum", "torch"], "mask2former/modeling_mask2former.py:pair_wise_sigmoid_cross_entropy_loss": ["BCEWithLogitsLoss", "Model_wise_sigmoid_cross_entropy_loss", "T", "Tensor", "criterion", "cross_entropy_loss_neg", "cross_entropy_loss_pos", "def", "height_and_width", "inputs", "labels", "loss", "loss_neg", "loss_pos", "matmul", "nn", "none", "ones_like", "r", "reduction", "return", "shape", "torch", "zeros_like"], "mask2former/modeling_mask2former.py:Mask2FormerHungarianMatcher.__init__": ["All", "ModelHungarianMatcher", "Module", "ValueError", "__init__", "and", "be", "can", "class", "cost_Model", "cost_class", "cost_dice", "costs", "def", "float", "if", "int", "nn", "num_points", "raise", "self", "super", "t"], "mask2former/modeling_mask2former.py:Mask2FormerHungarianMatcher.forward": ["False", "ModelHungarianMatcher", "Model_labels", "Models_queries_logits", "Module", "None", "Tensor", "align_corners", "append", "array", "as_tensor", "assigned_indices", "batch_size", "class", "class_labels", "class_queries_logits", "cost_Model", "cost_class", "cost_dice", "cost_matrix", "cpu", "def", "device", "dtype", "for", "forward", "i", "in", "indices", "int64", "j", "linear_sum_assignment", "list", "matched_indices", "maximum", "minimum", "nan_to_num", "nn", "np", "num_points", "pair_wise_dice_loss", "pair_wise_sigmoid_cross_entropy_loss", "point_coordinates", "pred_Model", "pred_coordinates", "pred_probs", "rand", "range", "repeat", "return", "sample_point", "self", "shape", "softmax", "squeeze", "target_Model", "target_coordinates", "tensor", "to", "torch", "tuple"], "mask2former/modeling_mask2former.py:Mask2FormerLoss.__init__": ["ModelConfig", "ModelHungarianMatcher", "ModelLoss", "Model_weight", "Module", "__init__", "class", "class_weight", "config", "cost_Model", "cost_class", "cost_dice", "def", "dice_weight", "dict", "empty_weight", "eos_coef", "float", "importance_sample_ratio", "matcher", "nn", "no_object_weight", "num_labels", "num_points", "ones", "oversample_ratio", "register_buffer", "requires_backends", "scipy", "self", "str", "super", "torch", "train_num_points", "weight_dict"], "mask2former/modeling_mask2former.py:Mask2FormerLoss._max_by_axis": ["ModelLoss", "Module", "_max_by_axis", "class", "def", "enumerate", "for", "in", "index", "int", "item", "list", "max", "maxes", "nn", "return", "self", "sizes", "sublist"], "mask2former/modeling_mask2former.py:Mask2FormerLoss._pad_images_to_max_in_batch": ["False", "ModelLoss", "Module", "Tensor", "_", "_max_by_axis", "_pad_images_to_max_in_batch", "batch_shape", "batch_size", "bool", "class", "copy_", "def", "device", "dtype", "for", "height", "in", "len", "list", "max_size", "nn", "ones", "padded_tensor", "padded_tensors", "padding_Model", "padding_Models", "return", "self", "shape", "tensor", "tensors", "torch", "tuple", "width", "zeros", "zip"], "mask2former/modeling_mask2former.py:Mask2FormerLoss.loss_labels": ["CrossEntropyLoss", "ModelLoss", "Module", "Tensor", "_", "_get_predictions_permutation_indices", "array", "batch_size", "cat", "class", "class_labels", "class_queries_logits", "criterion", "def", "device", "dict", "dtype", "empty_weight", "fill_value", "for", "full", "idx", "in", "indices", "int64", "j", "list", "loss_ce", "loss_cross_entropy", "loss_labels", "losses", "nn", "np", "num_labels", "num_queries", "pred_logits", "pred_logits_transposed", "return", "self", "shape", "str", "target", "target_classes", "target_classes_o", "torch", "transpose", "tuple", "weight", "zip"], "mask2former/modeling_mask2former.py:Mask2FormerLoss.loss_masks": ["False", "ModelLoss", "Model_labels", "Models_queries_logits", "Module", "None", "Tensor", "_", "_get_predictions_permutation_indices", "_get_targets_permutation_indices", "_pad_images_to_max_in_batch", "align_corners", "array", "calculate_uncertainty", "class", "def", "del", "dice_loss", "dict", "importance_sample_ratio", "indices", "int", "lambda", "list", "logits", "loss_Model", "loss_Models", "loss_dice", "losses", "nn", "no_grad", "np", "num_Models", "num_points", "oversample_ratio", "point_coordinates", "point_labels", "point_logits", "pred_Models", "return", "sample_point", "sample_points_using_uncertainty", "self", "sigmoid_cross_entropy_loss", "squeeze", "src_idx", "str", "target_Models", "tgt_idx", "torch", "tuple", "with"], "mask2former/modeling_mask2former.py:Mask2FormerLoss._get_predictions_permutation_indices": ["ModelLoss", "Module", "_", "_get_predictions_permutation_indices", "batch_indices", "cat", "class", "def", "enumerate", "for", "full_like", "i", "in", "indices", "nn", "predictions_indices", "return", "self", "src", "torch"], "mask2former/modeling_mask2former.py:Mask2FormerLoss._get_targets_permutation_indices": ["ModelLoss", "Module", "_", "_get_targets_permutation_indices", "batch_indices", "cat", "class", "def", "enumerate", "for", "full_like", "i", "in", "indices", "nn", "return", "self", "target_indices", "tgt", "torch"], "mask2former/modeling_mask2former.py:Mask2FormerLoss.calculate_uncertainty": ["ModelLoss", "Module", "Tensor", "abs", "calculate_uncertainty", "class", "def", "logits", "nn", "return", "self", "torch", "uncertainty_scores"], "mask2former/modeling_mask2former.py:Mask2FormerLoss.sample_points_using_uncertainty": ["False", "ModelLoss", "Module", "None", "Tensor", "align_corners", "arange", "cat", "class", "def", "device", "dim", "dtype", "float", "idx", "if", "importance_sample_ratio", "int", "k", "logits", "long", "nn", "num_boxes", "num_points", "num_points_sampled", "num_random_points", "num_uncertain_points", "oversample_ratio", "point_coordinates", "point_logits", "point_uncertainties", "rand", "return", "sample_point", "sample_points_using_uncertainty", "self", "shape", "shift", "topk", "torch", "uncertainty_function", "view"], "mask2former/modeling_mask2former.py:Mask2FormerLoss.forward": ["ModelLoss", "Model_labels", "Models_queries_logits", "Module", "None", "Tensor", "_", "aux_outputs", "auxiliary_predictions", "class", "class_labels", "class_queries_logits", "def", "device", "dict", "enumerate", "f", "for", "forward", "get_num_Models", "idx", "if", "in", "indices", "is", "items", "key", "list", "loss_Models", "loss_dict", "loss_labels", "losses", "matcher", "nn", "not", "num_Models", "return", "self", "str", "torch", "update", "value"], "mask2former/modeling_mask2former.py:Mask2FormerLoss.get_num_masks": ["ModelLoss", "Module", "PartialState", "Tensor", "_shared_state", "as_tensor", "clamp", "class", "class_labels", "classes", "def", "device", "dtype", "float", "for", "get_num_Models", "if", "in", "is_accelerate_available", "len", "min", "nn", "num_Models", "num_processes", "reduce", "return", "self", "sum", "torch", "world_size"], "mask2former/modeling_mask2former.py:multi_scale_deformable_attention": ["False", "Model_scale_deformable_attention", "Tensor", "_", "align_corners", "append", "attention_weights", "batch_size", "bilinear", "contiguous", "def", "dim", "enumerate", "flatten", "for", "functional", "grid_sample", "height", "hidden_dim", "in", "level_id", "list", "mode", "nn", "num_heads", "num_levels", "num_points", "num_queries", "output", "padding_mode", "reshape", "return", "sampling_grid_l_", "sampling_grids", "sampling_locations", "sampling_value_l_", "sampling_value_list", "shape", "split", "stack", "sum", "torch", "transpose", "tuple", "value", "value_l_", "value_list", "value_spatial_shapes", "view", "width", "zeros"], "mask2former/modeling_mask2former.py:Mask2FormerSinePositionEmbedding.__init__": ["False", "ModelSinePositionEmbedding", "Module", "None", "True", "ValueError", "__init__", "and", "be", "bool", "class", "def", "else", "float", "if", "int", "is", "math", "nn", "normalize", "not", "num_pos_feats", "passed", "pi", "raise", "scale", "self", "should", "super", "temperature"], "mask2former/modeling_mask2former.py:Mask2FormerSinePositionEmbedding.forward": ["Model", "ModelSinePositionEmbedding", "Module", "None", "Size", "Tensor", "arange", "bool", "cat", "class", "cos", "cumsum", "def", "device", "dim", "dim_t", "div", "dtype", "eps", "flatten", "floor", "forward", "if", "int64", "is", "nn", "normalize", "not_Model", "num_pos_feats", "permute", "pos", "pos_x", "pos_y", "return", "rounding_mode", "scale", "self", "shape", "sin", "stack", "str", "temperature", "to", "torch", "x_embed", "y_embed", "zeros"], "mask2former/modeling_mask2former.py:Mask2FormerPixelDecoderEncoderMultiscaleDeformableAttention.__init__": ["CUDA", "DeformableDetrMultiscaleDeformableAttention", "Linear", "ModelPixelDecoderEncoderMultiscaleDeformableAttention", "Module", "ValueError", "You", "__init__", "a", "and", "attention", "attention_weights", "authors", "be", "better", "but", "by", "class", "d", "d_model", "def", "dim_per_head", "dimension", "divisible", "each", "efficient", "embed_dim", "f", "got", "head", "if", "im2col_step", "implementation", "in", "int", "is", "make", "more", "must", "n_heads", "n_levels", "n_points", "nn", "not", "num_heads", "of", "output_proj", "power", "raise", "sampling_offsets", "self", "set", "super", "the", "to", "value_proj", "warn", "warnings", "which"], "mask2former/modeling_mask2former.py:Mask2FormerPixelDecoderEncoderMultiscaleDeformableAttention.with_pos_embed": ["ModelPixelDecoderEncoderMultiscaleDeformableAttention", "Module", "None", "Tensor", "class", "def", "else", "if", "is", "nn", "position_embeddings", "return", "self", "tensor", "torch", "with_pos_embed"], "mask2former/modeling_mask2former.py:Mask2FormerPixelDecoderEncoderMultiscaleDeformableAttention.forward": ["False", "Last", "Make", "ModelPixelDecoderEncoderMultiscaleDeformableAttention", "Modeled_fill", "Module", "None", "Tensor", "ValueError", "_", "align", "attention_Model", "attention_weights", "batch_size", "be", "bool", "but", "class", "d_model", "def", "device", "dim", "dtype", "elif", "else", "encoder", "encoder_attention_Model", "encoder_hidden_states", "f", "float", "for", "forward", "functional", "got", "height", "hidden", "hidden_states", "if", "in", "is", "length", "level_start_index", "long", "multi_scale_deformable_attention", "must", "n_heads", "n_levels", "n_points", "nn", "not", "num_queries", "of", "offset_normalizer", "or", "output", "output_attentions", "output_proj", "position_embeddings", "raise", "reference_points", "return", "sampling_locations", "sampling_offsets", "self", "sequence", "sequence_length", "shape", "shapes", "softmax", "spatial", "spatial_shapes_list", "states", "sum", "sure", "tensor", "the", "to", "torch", "total_elements", "value", "value_proj", "view", "width", "with", "with_pos_embed"], "mask2former/modeling_mask2former.py:Mask2FormerPixelDecoderEncoderLayer.__init__": ["LayerNorm", "Linear", "ModelConfig", "ModelPixelDecoderEncoderLayer", "ModelPixelDecoderEncoderMultiscaleDeformableAttention", "Module", "__init__", "activation_dropout", "activation_fn", "class", "config", "def", "dropout", "embed_dim", "encoder_feedforward_dim", "fc1", "fc2", "feature_size", "final_layer_norm", "functional", "n_levels", "n_points", "nn", "num_attention_heads", "num_heads", "relu", "self", "self_attn", "self_attn_layer_norm", "super"], "mask2former/modeling_mask2former.py:Mask2FormerPixelDecoderEncoderLayer.forward": ["False", "ModelPixelDecoderEncoderLayer", "Module", "None", "Tensor", "activation_dropout", "activation_fn", "any", "attention_Model", "attn_weights", "bool", "clamp", "clamp_value", "class", "def", "dropout", "dtype", "encoder_attention_Model", "encoder_hidden_states", "fc1", "fc2", "final_layer_norm", "finfo", "forward", "functional", "hidden_states", "if", "isinf", "isnan", "level_start_index", "max", "min", "nn", "or", "output_attentions", "outputs", "p", "position_embeddings", "reference_points", "residual", "return", "self", "self_attn", "self_attn_layer_norm", "spatial_shapes_list", "torch", "training", "transpose"], "mask2former/modeling_mask2former.py:Mask2FormerPixelDecoderEncoderOnly.__init__": ["ModelConfig", "ModelPixelDecoderEncoderLayer", "ModelPixelDecoderEncoderOnly", "Module", "ModuleList", "_", "__init__", "class", "config", "def", "dropout", "encoder_layers", "for", "in", "layers", "nn", "range", "self", "super"], "mask2former/modeling_mask2former.py:Mask2FormerPixelDecoderEncoderOnly.get_reference_points": ["ModelPixelDecoderEncoderOnly", "Module", "None", "append", "cat", "class", "def", "device", "dtype", "enumerate", "for", "get_reference_points", "height", "ij", "in", "indexing", "linspace", "lvl", "meshgrid", "nn", "ref", "ref_x", "ref_y", "reference_points", "reference_points_list", "reshape", "return", "spatial_shapes_list", "stack", "torch", "valid_ratios", "width"], "mask2former/modeling_mask2former.py:Mask2FormerPixelDecoderEncoderOnly.forward": ["BaseModelOutput", "ModelPixelDecoderEncoderOnly", "Module", "None", "all_attentions", "all_hidden_states", "attention_Model", "attentions", "class", "config", "def", "device", "else", "encoder_layer", "enumerate", "for", "forward", "get_reference_points", "hidden_states", "i", "if", "in", "inputs_embeds", "is", "last_hidden_state", "layer_outputs", "layers", "level_start_index", "nn", "not", "output_attentions", "output_hidden_states", "position_embeddings", "r", "reference_points", "return", "return_dict", "self", "spatial_shapes_list", "transpose", "use_return_dict", "valid_ratios"], "mask2former/modeling_mask2former.py:Mask2FormerPixelDecoder.__init__": ["Conv2d", "False", "GroupNorm", "ModelConfig", "ModelPixelDecoder", "ModelPixelDecoderEncoderOnly", "ModelSinePositionEmbedding", "Model_dim", "Model_feature_size", "Model_projection", "Module", "ModuleList", "Parameter", "ReLU", "Sequential", "Tensor", "True", "__init__", "adapter_", "add_module", "append", "bias", "class", "common_stride", "config", "def", "else", "encoder", "enumerate", "f", "feature_channels", "feature_dim", "feature_size", "feature_strides", "for", "idx", "if", "in", "in_channels", "input_projections", "input_projections_list", "int", "kernel_size", "lateral_conv", "lateral_convolutions", "lateral_convs", "layer_", "level_embed", "log2", "min", "nn", "normalize", "np", "num_feature_levels", "num_fpn_levels", "num_pos_feats", "num_pos_features", "output_conv", "output_convolutions", "output_convs", "padding", "position_embedding", "self", "stride", "super", "torch", "transformer_feature_strides", "transformer_in_channels"], "mask2former/modeling_mask2former.py:Mask2FormerPixelDecoder.get_valid_ratio": ["Model", "ModelPixelDecoder", "Module", "_", "class", "def", "dtype", "float32", "get_valid_ratio", "height", "nn", "return", "self", "shape", "stack", "sum", "to", "torch", "valid_height", "valid_ratio", "valid_ratio_height", "valid_ratio_width", "valid_width", "width"], "mask2former/modeling_mask2former.py:Mask2FormerPixelDecoder.forward": ["False", "Model", "ModelPixelDecoder", "ModelPixelDecoderOutput", "Model_features", "Model_projection", "Models", "Models_flat", "Module", "None", "align_corners", "append", "as_tensor", "attention_Model", "attentions", "batch_size", "bilinear", "bool", "cat", "class", "config", "cumsum", "current_fpn", "def", "device", "dim", "dtype", "else", "embed", "encoder", "encoder_output", "encoder_outputs", "enumerate", "feature", "features", "flatten", "for", "forward", "functional", "get_valid_ratio", "height", "i", "idx", "if", "in", "input_embeds", "input_embeds_flat", "input_projections", "inputs_embeds", "interpolate", "is", "last_hidden_state", "lateral_conv", "lateral_convolutions", "level", "level_embed", "level_pos_embed_flat", "level_start_index", "level_start_index_list", "long", "mode", "multi_scale_features", "new_zeros", "nn", "not", "num_cur_levels", "num_feature_levels", "num_fpn_levels", "out", "output_attentions", "output_conv", "output_convolutions", "output_hidden_states", "outputs", "position_embedding", "position_embeddings", "prod", "range", "return", "return_dict", "self", "shape", "size", "spatial_shapes", "spatial_shapes_list", "split", "split_sizes", "stack", "torch", "transpose", "tuple", "valid_ratios", "view", "width", "x", "zeros"], "mask2former/modeling_mask2former.py:Mask2FormerPixelLevelModule.__init__": ["ModelConfig", "ModelPixelDecoder", "ModelPixelLevelModule", "Module", "__init__", "channels", "class", "config", "decoder", "def", "encoder", "feature_channels", "load_backbone", "nn", "self", "super"], "mask2former/modeling_mask2former.py:Mask2FormerPixelLevelModule.forward": ["False", "ModelPixelLevelModule", "ModelPixelLevelModuleOutput", "Model_features", "Module", "None", "Tensor", "backbone_features", "bool", "class", "decoder", "decoder_hidden_states", "decoder_last_hidden_state", "decoder_output", "def", "else", "encoder", "encoder_hidden_states", "encoder_last_hidden_state", "feature_maps", "forward", "if", "multi_scale_features", "nn", "output_hidden_states", "pixel_values", "return", "self", "tuple"], "mask2former/modeling_mask2former.py:Mask2FormerAttention.__init__": ["False", "Linear", "ModelAttention", "Module", "True", "ValueError", "__init__", "and", "be", "bias", "bool", "by", "class", "def", "divisible", "dropout", "embed_dim", "f", "float", "got", "head_dim", "if", "int", "is_decoder", "k_proj", "must", "nn", "num_heads", "out_proj", "q_proj", "raise", "scaling", "self", "super", "v_proj"], "mask2former/modeling_mask2former.py:Mask2FormerAttention._shape": ["ModelAttention", "Module", "Tensor", "_shape", "batch_size", "class", "contiguous", "def", "head_dim", "int", "nn", "num_heads", "return", "self", "seq_len", "tensor", "torch", "transpose", "view"], "mask2former/modeling_mask2former.py:Mask2FormerAttention.with_pos_embed": ["ModelAttention", "Module", "None", "Tensor", "class", "def", "else", "if", "is", "nn", "position_embeddings", "return", "self", "tensor", "torch", "with_pos_embed"], "mask2former/modeling_mask2former.py:Mask2FormerAttention.forward": ["Attention", "False", "Model", "ModelAttention", "Module", "None", "Tensor", "ValueError", "_shape", "attention_Model", "attn_output", "attn_probs", "attn_weights", "attn_weights_reshaped", "batch_size", "be", "bmm", "bool", "but", "class", "def", "dim", "dropout", "else", "embed_dim", "f", "forward", "functional", "head_dim", "hidden_states", "hidden_states_original", "if", "is", "is_cross_attention", "k_proj", "key_states", "key_value_position_embeddings", "key_value_states", "key_value_states_original", "nn", "not", "num_heads", "of", "out_proj", "output_attentions", "p", "permute", "position_embeddings", "proj_shape", "q_proj", "query_states", "raise", "reshape", "return", "scaling", "self", "should", "size", "softmax", "source_len", "target_len", "torch", "training", "transpose", "tuple", "v_proj", "value_states", "view", "weights", "with_pos_embed"], "mask2former/modeling_mask2former.py:Mask2FormerMaskedAttentionDecoderLayer.__init__": ["ACT2FN", "GradientCheckpointingLayer", "LayerNorm", "Linear", "ModelAttention", "ModelConfig", "ModelModeledAttentionDecoderLayer", "MultiheadAttention", "True", "__init__", "activation_dropout", "activation_fn", "activation_function", "class", "config", "cross_attn", "cross_attn_layer_norm", "def", "dim_feedforward", "dropout", "embed_dim", "fc1", "fc2", "final_layer_norm", "hidden_dim", "is_decoder", "nn", "num_attention_heads", "num_heads", "pre_norm", "self", "self_attn", "self_attn_layer_norm", "super"], "mask2former/modeling_mask2former.py:Mask2FormerMaskedAttentionDecoderLayer.with_pos_embed": ["GradientCheckpointingLayer", "ModelModeledAttentionDecoderLayer", "None", "Tensor", "class", "def", "else", "if", "is", "pos", "return", "self", "tensor", "with_pos_embed"], "mask2former/modeling_mask2former.py:Mask2FormerMaskedAttentionDecoderLayer.forward_post": ["False", "GradientCheckpointingLayer", "ModelModeledAttentionDecoderLayer", "None", "Tensor", "True", "activation_dropout", "activation_fn", "attention_Model", "attn_Model", "bool", "class", "cross_attn", "cross_attn_layer_norm", "cross_attn_weights", "def", "dropout", "encoder_attention_Model", "encoder_hidden_states", "fc1", "fc2", "final_layer_norm", "forward_post", "functional", "hidden_states", "if", "int", "key", "key_padding_Model", "level_index", "nn", "output_attentions", "outputs", "p", "position_embeddings", "query", "query_position_embeddings", "residual", "return", "self", "self_attn", "self_attn_layer_norm", "self_attn_weights", "torch", "training", "value", "with_pos_embed"], "mask2former/modeling_mask2former.py:Mask2FormerMaskedAttentionDecoderLayer.forward_pre": ["False", "GradientCheckpointingLayer", "ModelModeledAttentionDecoderLayer", "None", "Tensor", "True", "activation_dropout", "activation_fn", "attention_Model", "attn_Model", "bool", "class", "cross_attn", "cross_attn_layer_norm", "cross_attn_weights", "def", "dropout", "encoder_attention_Model", "encoder_hidden_states", "fc1", "fc2", "final_layer_norm", "forward_pre", "functional", "hidden_states", "if", "int", "key", "key_padding_Model", "level_index", "nn", "output_attentions", "outputs", "p", "position_embeddings", "query", "query_position_embeddings", "residual", "return", "self", "self_attn", "self_attn_layer_norm", "self_attn_weights", "torch", "training", "value", "with_pos_embed"], "mask2former/modeling_mask2former.py:Mask2FormerMaskedAttentionDecoderLayer.forward": ["False", "GradientCheckpointingLayer", "ModelModeledAttentionDecoderLayer", "None", "Tensor", "attention_Model", "bool", "class", "def", "else", "encoder_attention_Model", "encoder_hidden_states", "forward", "forward_post", "forward_pre", "hidden_states", "if", "int", "level_index", "output_attentions", "outputs", "position_embeddings", "pre_norm", "query_position_embeddings", "return", "self", "torch"], "mask2former/modeling_mask2former.py:Mask2FormerMaskedAttentionDecoder.__init__": ["False", "LayerNorm", "ModelConfig", "ModelModelPredictor", "ModelModeledAttentionDecoder", "ModelModeledAttentionDecoderLayer", "Model_feature_size", "Model_predictor", "Module", "ModuleList", "_", "__init__", "class", "config", "decoder_layers", "def", "dropout", "for", "gradient_checkpointing", "hidden_dim", "hidden_size", "in", "layerdrop", "layernorm", "layers", "nn", "num_attention_heads", "num_feature_levels", "num_heads", "range", "self", "super"], "mask2former/modeling_mask2former.py:Mask2FormerMaskedAttentionDecoder.forward": ["ModelModeledAttentionDecoder", "ModelModeledAttentionDecoderOutput", "Model_predictor", "Models_queries_logits", "Module", "None", "Tensor", "all_hidden_states", "and", "attention_Model", "attentions", "bool", "class", "config", "continue", "decoder_layer", "def", "dropout_probability", "dtype", "else", "encoder_attention_Model", "encoder_hidden_states", "enumerate", "feature_size_list", "for", "forward", "hidden_states", "idx", "if", "in", "inputs_embeds", "intermediate", "intermediate_Model_predictions", "intermediate_hidden_states", "is", "last_hidden_state", "layer_outputs", "layerdrop", "layernorm", "layers", "level_index", "list", "multi_stage_positional_embeddings", "nn", "not", "num_feature_levels", "output_attentions", "output_hidden_states", "outputs", "pixel_embeddings", "predicted_Model", "query_position_embeddings", "r", "rand", "return", "return_dict", "self", "shape", "sum", "to", "torch", "training", "transpose", "tuple", "unsqueeze", "use_return_dict", "v", "where"], "mask2former/modeling_mask2former.py:Mask2FormerPredictionBlock.__init__": ["Linear", "ModelPredictionBlock", "Module", "None", "__init__", "activation", "add_module", "class", "def", "enumerate", "for", "i", "in", "in_dim", "int", "layer", "layers", "nn", "out_dim", "self", "str", "super"], "mask2former/modeling_mask2former.py:Mask2FormerPredictionBlock.forward": ["ModelPredictionBlock", "Module", "Tensor", "class", "def", "for", "forward", "hidden_state", "in", "input", "layer", "layers", "nn", "return", "self"], "mask2former/modeling_mask2former.py:Mask2FormerMLPPredictionHead.__init__": ["Identity", "ModelMLPPredictionHead", "ModelPredictionBlock", "Module", "ReLU", "__init__", "activation", "add_module", "append", "class", "def", "else", "enumerate", "for", "hidden_dim", "i", "if", "in", "in_dim", "in_dims", "input_dim", "int", "layer", "layers", "nn", "num_layers", "out_dim", "out_dims", "output_dim", "self", "str", "super", "zip"], "mask2former/modeling_mask2former.py:Mask2FormerMLPPredictionHead.forward": ["ModelMLPPredictionHead", "Module", "Tensor", "class", "def", "for", "forward", "hidden_state", "in", "input", "layer", "layers", "nn", "return", "self"], "mask2former/modeling_mask2former.py:Mask2FormerMaskPredictor.__init__": ["ModelMLPPredictionHead", "ModelModelPredictor", "Model_embedder", "Model_feature_size", "Module", "Tensor", "__init__", "class", "def", "hidden_size", "int", "nn", "num_heads", "self", "super", "torch"], "mask2former/modeling_mask2former.py:Mask2FormerMaskPredictor.forward": ["False", "ModelModelPredictor", "Model_embedder", "Model_embeddings", "Module", "None", "Tensor", "align_corners", "attention_Model", "attention_Model_target_size", "bchw", "bilinear", "bool", "bqc", "bqhw", "class", "def", "detach", "einsum", "flatten", "forward", "functional", "int", "interpolate", "mode", "nn", "num_heads", "outputs", "outputs_Model", "pixel_embeddings", "repeat", "return", "self", "sigmoid", "size", "torch", "transpose", "unsqueeze"], "mask2former/modeling_mask2former.py:Mask2FormerTransformerModule.__init__": ["Conv2d", "Embedding", "ModelConfig", "ModelModeledAttentionDecoder", "ModelSinePositionEmbedding", "ModelTransformerModule", "Module", "Sequential", "True", "_", "__init__", "append", "class", "config", "decoder", "def", "else", "enforce_input_projection", "for", "hidden_dim", "if", "in", "in_features", "input_projections", "int", "kernel_size", "level_embed", "nn", "normalize", "num_feature_levels", "num_pos_feats", "num_queries", "or", "position_embedder", "queries_embedder", "queries_features", "range", "self", "super"], "mask2former/modeling_mask2former.py:Mask2FormerTransformerModule.forward": ["False", "ModelModeledAttentionDecoderOutput", "ModelTransformerModule", "Model_features", "Module", "None", "Tensor", "True", "_", "append", "batch_size", "bool", "class", "decoder", "decoder_output", "def", "device", "dtype", "encoder_hidden_states", "feature_size_list", "flatten", "for", "forward", "i", "in", "input_projections", "inputs_embeds", "level_embed", "list", "multi_scale_features", "multi_stage_features", "multi_stage_positional_embeddings", "nn", "num_feature_levels", "output_attentions", "output_hidden_states", "permute", "pixel_embeddings", "position_embedder", "queries_embedder", "queries_features", "query_embeddings", "query_features", "query_position_embeddings", "range", "repeat", "return", "return_dict", "self", "shape", "size_list", "unsqueeze", "weight"], "mask2former/modeling_mask2former.py:Mask2FormerPreTrainedModel._init_weights": ["BatchNorm2d", "Conv2d", "Embedding", "False", "GroupNorm", "LayerNorm", "Linear", "ModelLoss", "ModelModeledAttentionDecoderLayer", "ModelPixelDecoder", "ModelPixelDecoderEncoderMultiscaleDeformableAttention", "ModelPreTrainedModel", "ModelTransformerModule", "Module", "None", "PreTrainedModel", "Sequential", "True", "_init_weights", "_is_hf_initialized", "abs", "and", "arange", "attention_weights", "bias", "class", "config", "constant_", "copy_", "cos", "cross_attn", "def", "dim", "dtype", "elif", "empty_weight", "eos_coef", "float", "for", "gain", "getattr", "grid_init", "hasattr", "i", "if", "in", "in_proj_bias", "init", "init_std", "init_xavier_std", "input_projection", "input_projections", "int64", "is", "isinstance", "keepdim", "level_embed", "math", "max", "mean", "module", "n_heads", "n_levels", "n_points", "nn", "normal_", "not", "num_batches_tracked", "num_labels", "ones", "ones_", "output_proj", "p", "padding_idx", "parameters", "pi", "range", "reference_points", "repeat", "running_mean", "running_var", "sampling_offsets", "self", "sin", "stack", "std", "thetas", "torch", "value_proj", "view", "weight", "xavier_std", "xavier_uniform_", "zeros_"], "mask2former/modeling_mask2former.py:Mask2FormerModel.__init__": ["ModelConfig", "ModelModel", "ModelPixelLevelModule", "ModelPreTrainedModel", "ModelTransformerModule", "__init__", "class", "config", "def", "feature_size", "in_features", "pixel_level_module", "post_init", "self", "super", "transformer_module"], "mask2former/modeling_mask2former.py:Mask2FormerModel.forward": ["ModelModel", "ModelModelOutput", "ModelPreTrainedModel", "Model_features", "Models_queries_logits", "None", "Tensor", "True", "_", "attentions", "batch_size", "bool", "class", "config", "decoder_hidden_states", "decoder_last_hidden_state", "def", "device", "else", "encoder_hidden_states", "encoder_last_hidden_state", "for", "forward", "height", "hidden_states", "if", "in", "intermediate_hidden_states", "is", "kwargs", "last_hidden_state", "multi_scale_features", "not", "ones", "output", "output_attentions", "output_hidden_states", "pixel_Model", "pixel_decoder_hidden_states", "pixel_decoder_last_hidden_state", "pixel_level_module", "pixel_level_module_output", "pixel_values", "return", "return_dict", "self", "shape", "torch", "transformer_decoder_hidden_states", "transformer_decoder_intermediate_states", "transformer_decoder_last_hidden_state", "transformer_module", "transformer_module_output", "tuple", "use_return_dict", "v", "values", "width"], "mask2former/modeling_mask2former.py:Mask2FormerForUniversalSegmentation.__init__": ["Linear", "ModelConfig", "ModelForUniversalSegmentation", "ModelLoss", "ModelModel", "ModelPreTrainedModel", "Model_weight", "__init__", "class", "class_predictor", "class_weight", "config", "criterion", "def", "dice_weight", "dict", "float", "hidden_dim", "loss_Model", "loss_cross_entropy", "loss_dice", "model", "nn", "num_labels", "post_init", "self", "str", "super", "weight_dict"], "mask2former/modeling_mask2former.py:Mask2FormerForUniversalSegmentation.get_loss_dict": ["ModelForUniversalSegmentation", "ModelPreTrainedModel", "Model_labels", "Models_queries_logits", "Tensor", "auxiliary_predictions", "class", "class_labels", "class_queries_logits", "criterion", "def", "dict", "for", "get_loss_dict", "if", "in", "items", "key", "loss", "loss_dict", "loss_key", "return", "self", "str", "weight", "weight_dict"], "mask2former/modeling_mask2former.py:Mask2FormerForUniversalSegmentation.get_loss": ["ModelForUniversalSegmentation", "ModelPreTrainedModel", "Tensor", "class", "def", "dict", "get_loss", "loss_dict", "return", "self", "str", "sum", "values"], "mask2former/modeling_mask2former.py:Mask2FormerForUniversalSegmentation.get_auxiliary_logits": ["ModelForUniversalSegmentation", "ModelPreTrainedModel", "Models_queries_logits", "Tensor", "append", "aux_binary_Models", "aux_classes", "auxiliary_logits", "class", "class_queries_logits", "classes", "def", "dict", "for", "get_auxiliary_logits", "in", "list", "output_Models", "return", "self", "str", "torch", "zip"], "mask2former/modeling_mask2former.py:Mask2FormerForUniversalSegmentation.forward": ["ModelForUniversalSegmentation", "ModelForUniversalSegmentationOutput", "ModelPreTrainedModel", "Model_labels", "Models_queries_logits", "None", "Tensor", "True", "and", "attentions", "auxiliary_logits", "auxiliary_predictions", "bool", "class", "class_labels", "class_prediction", "class_predictor", "class_queries_logits", "config", "decoder_output", "def", "else", "encoder_hidden_states", "encoder_last_hidden_state", "for", "forward", "get_auxiliary_logits", "get_loss", "get_loss_dict", "if", "in", "is", "kwargs", "list", "loss", "loss_dict", "model", "not", "or", "output", "output_attentions", "output_auxiliary_logits", "output_hidden_states", "outputs", "pixel_Model", "pixel_decoder_hidden_states", "pixel_decoder_last_hidden_state", "pixel_values", "r", "return", "return_dict", "self", "transformer_decoder_hidden_states", "transformer_decoder_intermediate_states", "transformer_decoder_last_hidden_state", "transpose", "tuple", "use_auxiliary_loss", "use_return_dict", "v", "values"], "maskformer/modeling_maskformer.py:upsample_like": ["False", "Model_like", "Modeld", "Tensor", "_", "align_corners", "bilinear", "def", "functional", "height", "interpolate", "like", "mode", "nn", "pixel_values", "return", "shape", "size", "str", "width"], "maskformer/modeling_maskformer.py:dice_loss": ["Model_loss", "Tensor", "def", "denominator", "flatten", "inputs", "int", "labels", "loss", "num_masks", "numerator", "probs", "r", "return", "sigmoid", "sum"], "maskformer/modeling_maskformer.py:sigmoid_focal_loss": ["BCEWithLogitsLoss", "Model", "Model_focal_loss", "Tensor", "alpha", "alpha_t", "criterion", "cross_entropy_loss", "def", "float", "gamma", "if", "inputs", "int", "labels", "loss", "mean", "nn", "none", "num_masks", "p_t", "probs", "r", "reduction", "return", "sum"], "maskformer/modeling_maskformer.py:pair_wise_dice_loss": ["Model_wise_dice_loss", "None", "T", "Tensor", "def", "denominator", "flatten", "inputs", "labels", "loss", "matmul", "numerator", "return", "sigmoid", "sum", "torch"], "maskformer/modeling_maskformer.py:pair_wise_sigmoid_focal_loss": ["BCEWithLogitsLoss", "Model_wise_sigmoid_focal_loss", "T", "Tensor", "ValueError", "alpha", "be", "criterion", "cross_entropy_loss_neg", "cross_entropy_loss_pos", "def", "float", "focal_neg", "focal_pos", "gamma", "height_and_width", "if", "inputs", "labels", "loss", "matmul", "must", "nn", "none", "ones_like", "positive", "prob", "r", "raise", "reduction", "return", "shape", "sigmoid", "torch", "zeros_like"], "maskformer/modeling_maskformer.py:DetrAttention.__init__": ["Linear", "ModelAttention", "Module", "True", "ValueError", "__init__", "and", "be", "bias", "bool", "by", "class", "def", "divisible", "dropout", "embed_dim", "f", "float", "got", "head_dim", "if", "int", "k_proj", "must", "nn", "num_heads", "out_proj", "q_proj", "raise", "scaling", "self", "super", "v_proj"], "maskformer/modeling_maskformer.py:DetrAttention._shape": ["ModelAttention", "Module", "Tensor", "_shape", "batch_size", "class", "contiguous", "def", "head_dim", "int", "nn", "num_heads", "return", "self", "seq_len", "tensor", "torch", "transpose", "view"], "maskformer/modeling_maskformer.py:DetrAttention.with_pos_embed": ["ModelAttention", "Module", "None", "Tensor", "class", "def", "else", "if", "is", "nn", "object_queries", "return", "self", "tensor", "torch", "with_pos_embed"], "maskformer/modeling_maskformer.py:DetrAttention.forward": ["Attention", "False", "ModelAttention", "Module", "None", "Tensor", "ValueError", "_shape", "attention_mask", "attn_output", "attn_probs", "attn_weights", "attn_weights_reshaped", "batch_size", "be", "bmm", "bool", "but", "class", "def", "dim", "dropout", "dtype", "else", "embed_dim", "f", "forward", "functional", "head_dim", "hidden_states", "hidden_states_original", "if", "inf", "is", "is_cross_attention", "k_proj", "key_states", "key_value_states", "key_value_states_original", "mask", "masked_fill_", "nn", "not", "num_heads", "object_queries", "of", "out_proj", "output_attentions", "p", "proj_shape", "q_proj", "query_states", "raise", "reshape", "return", "scaling", "self", "should", "size", "softmax", "source_len", "spatial_position_embeddings", "target_len", "torch", "training", "transpose", "tuple", "v_proj", "value_states", "view", "weights", "with_pos_embed", "zeros_like"], "maskformer/modeling_maskformer.py:DetrDecoderLayer.__init__": ["ACT2FN", "GradientCheckpointingLayer", "LayerNorm", "Linear", "ModelAttention", "ModelConfig", "ModelDecoderLayer", "__init__", "activation_dropout", "activation_fn", "activation_function", "attention_dropout", "class", "config", "d_model", "decoder_attention_heads", "decoder_ffn_dim", "def", "dropout", "embed_dim", "encoder_attn", "encoder_attn_layer_norm", "fc1", "fc2", "final_layer_norm", "nn", "num_heads", "self", "self_attn", "self_attn_layer_norm", "super"], "maskformer/modeling_maskformer.py:DetrDecoderLayer.forward": ["False", "GradientCheckpointingLayer", "ModelDecoderLayer", "None", "Tensor", "activation_dropout", "activation_fn", "attention_mask", "bool", "class", "cross_attn_weights", "def", "dropout", "encoder_attention_mask", "encoder_attn", "encoder_attn_layer_norm", "encoder_hidden_states", "fc1", "fc2", "final_layer_norm", "forward", "functional", "hidden_states", "if", "is", "key_value_states", "nn", "not", "object_queries", "output_attentions", "outputs", "p", "query_position_embeddings", "residual", "return", "self", "self_attn", "self_attn_layer_norm", "self_attn_weights", "spatial_position_embeddings", "torch", "training"], "maskformer/modeling_maskformer.py:DetrDecoder.__init__": ["False", "LayerNorm", "ModelConfig", "ModelDecoder", "ModelDecoderLayer", "Module", "ModuleList", "_", "__init__", "class", "config", "d_model", "decoder_layerdrop", "decoder_layers", "def", "dropout", "for", "gradient_checkpointing", "in", "layerdrop", "layernorm", "layers", "nn", "range", "self", "super"], "maskformer/modeling_maskformer.py:DetrDecoder.forward": ["ModelDecoder", "ModelDecoderOutput", "Module", "None", "_prepare_4d_attention_mask", "all_cross_attentions", "all_hidden_states", "all_self_attns", "and", "attention_mask", "attentions", "auxiliary_loss", "class", "config", "continue", "cross_attentions", "decoder_layer", "def", "dropout_probability", "dtype", "else", "encoder_attention_mask", "encoder_hidden_states", "enumerate", "for", "forward", "hidden_states", "idx", "if", "in", "input_shape", "inputs_embeds", "intermediate", "intermediate_hidden_states", "is", "last_hidden_state", "layer_outputs", "layerdrop", "layernorm", "layers", "nn", "not", "object_queries", "output_attentions", "output_hidden_states", "query_position_embeddings", "r", "rand", "return", "return_dict", "self", "size", "stack", "tgt_len", "torch", "training", "tuple", "use_return_dict", "v"], "maskformer/modeling_maskformer.py:MaskFormerHungarianMatcher.__init__": ["All", "ModelHungarianMatcher", "Module", "ValueError", "__init__", "and", "be", "can", "class", "cost_Model", "cost_class", "cost_dice", "costs", "def", "float", "if", "nn", "raise", "self", "super", "t"], "maskformer/modeling_maskformer.py:MaskFormerHungarianMatcher.forward": ["ModelHungarianMatcher", "Model_labels", "Models_queries_logits", "Module", "None", "Tensor", "append", "array", "as_tensor", "assigned_indices", "class", "class_labels", "class_queries_logits", "cost_Model", "cost_class", "cost_dice", "cost_matrix", "cpu", "def", "dtype", "flatten", "for", "forward", "functional", "i", "in", "indices", "int64", "interpolate", "j", "labels", "linear_sum_assignment", "list", "matched_indices", "mode", "nearest", "nn", "np", "pair_wise_dice_loss", "pair_wise_sigmoid_focal_loss", "pred_Model", "pred_Model_flat", "pred_probs", "preds_Models", "preds_probs", "return", "self", "shape", "size", "softmax", "target_Model", "target_Model_flat", "torch", "tuple", "zip"], "maskformer/modeling_maskformer.py:MaskFormerHungarianMatcher.__repr__": ["Matcher", "ModelHungarianMatcher", "Module", "__class__", "__name__", "__repr__", "_repr_indent", "body", "class", "cost_Model", "cost_class", "cost_dice", "def", "f", "for", "head", "in", "join", "line", "lines", "n", "nn", "return", "self"], "maskformer/modeling_maskformer.py:MaskFormerLoss.__init__": ["ModelHungarianMatcher", "ModelLoss", "Module", "__init__", "class", "def", "dict", "empty_weight", "eos_coef", "float", "int", "matcher", "nn", "num_labels", "ones", "register_buffer", "requires_backends", "scipy", "self", "str", "super", "torch", "weight_dict"], "maskformer/modeling_maskformer.py:MaskFormerLoss._max_by_axis": ["ModelLoss", "Module", "_max_by_axis", "class", "def", "enumerate", "for", "in", "index", "int", "item", "list", "max", "maxes", "nn", "return", "self", "sublist", "the_list"], "maskformer/modeling_maskformer.py:MaskFormerLoss._pad_images_to_max_in_batch": ["False", "ModelLoss", "Module", "Tensor", "_", "_max_by_axis", "_pad_images_to_max_in_batch", "b", "batch_shape", "batch_size", "bool", "class", "copy_", "def", "device", "dtype", "for", "h", "in", "len", "list", "max_size", "nn", "ones", "padded_tensor", "padded_tensors", "padding_Model", "padding_Models", "return", "self", "shape", "tensor", "tensors", "torch", "tuple", "w", "zeros", "zip"], "maskformer/modeling_maskformer.py:MaskFormerLoss.loss_labels": ["CrossEntropyLoss", "ModelLoss", "Module", "Tensor", "_", "_get_predictions_permutation_indices", "array", "batch_size", "cat", "class", "class_labels", "class_queries_logits", "criterion", "def", "device", "dict", "dtype", "empty_weight", "fill_value", "for", "full", "idx", "in", "indices", "int64", "j", "list", "loss_ce", "loss_cross_entropy", "loss_labels", "losses", "nn", "np", "num_labels", "num_queries", "pred_logits", "pred_logits_transposed", "return", "self", "shape", "str", "target", "target_classes", "target_classes_o", "torch", "transpose", "tuple", "weight", "zip"], "maskformer/modeling_maskformer.py:MaskFormerLoss.loss_masks": ["False", "ModelLoss", "Model_labels", "Models_queries_logits", "Module", "None", "Tensor", "_", "_get_predictions_permutation_indices", "_get_targets_permutation_indices", "_pad_images_to_max_in_batch", "align_corners", "array", "bilinear", "class", "def", "dice_loss", "dict", "flatten", "functional", "indices", "int", "interpolate", "list", "loss_Model", "loss_Models", "loss_dice", "losses", "mode", "nn", "np", "num_Models", "pred_Models", "return", "self", "shape", "sigmoid_focal_loss", "size", "src_idx", "str", "target_Models", "tgt_idx", "tuple"], "maskformer/modeling_maskformer.py:MaskFormerLoss._get_predictions_permutation_indices": ["ModelLoss", "Module", "_", "_get_predictions_permutation_indices", "batch_indices", "cat", "class", "def", "enumerate", "for", "full_like", "i", "in", "indices", "nn", "predictions_indices", "return", "self", "src", "torch"], "maskformer/modeling_maskformer.py:MaskFormerLoss._get_targets_permutation_indices": ["ModelLoss", "Module", "_", "_get_targets_permutation_indices", "batch_indices", "cat", "class", "def", "enumerate", "for", "full_like", "i", "in", "indices", "nn", "return", "self", "target_indices", "tgt", "torch"], "maskformer/modeling_maskformer.py:MaskFormerLoss.forward": ["ModelLoss", "Model_labels", "Models_queries_logits", "Module", "None", "Number", "Tensor", "_", "aux_outputs", "auxiliary_predictions", "class", "class_labels", "class_queries_logits", "def", "device", "dict", "enumerate", "f", "for", "forward", "get_num_Models", "idx", "if", "in", "indices", "is", "items", "key", "list", "loss_Models", "loss_dict", "loss_labels", "losses", "matcher", "nn", "not", "num_Models", "return", "self", "str", "update", "value"], "maskformer/modeling_maskformer.py:MaskFormerLoss.get_num_masks": ["ModelLoss", "Module", "PartialState", "Tensor", "_shared_state", "as_tensor", "clamp", "class", "class_labels", "classes", "def", "device", "dtype", "float", "for", "get_num_Models", "if", "in", "is_accelerate_available", "len", "min", "nn", "num_Models", "num_processes", "reduce", "return", "self", "sum", "torch", "world_size"], "maskformer/modeling_maskformer.py:MaskFormerFPNConvLayer.__init__": ["Conv2d", "False", "GroupNorm", "ModelFPNConvLayer", "Module", "ReLU", "True", "__init__", "add_module", "bias", "class", "def", "enumerate", "for", "i", "in", "in_features", "inplace", "int", "kernel_size", "layer", "layers", "nn", "out_features", "padding", "self", "str", "super"], "maskformer/modeling_maskformer.py:MaskFormerFPNConvLayer.forward": ["ModelFPNConvLayer", "Module", "Tensor", "class", "def", "for", "forward", "hidden_state", "in", "input", "layer", "layers", "nn", "return", "self"], "maskformer/modeling_maskformer.py:MaskFormerFPNLayer.__init__": ["Conv2d", "False", "GroupNorm", "ModelFPNConvLayer", "ModelFPNLayer", "Module", "Sequential", "__init__", "bias", "block", "class", "def", "in_features", "int", "kernel_size", "lateral_features", "nn", "padding", "proj", "self", "super"], "maskformer/modeling_maskformer.py:MaskFormerFPNLayer.forward": ["ModelFPNLayer", "Module", "Tensor", "block", "class", "def", "down", "forward", "functional", "interpolate", "left", "mode", "nearest", "nn", "proj", "return", "self", "shape", "size"], "maskformer/modeling_maskformer.py:MaskFormerFPNModel.__init__": ["ModelFPNConvLayer", "ModelFPNLayer", "ModelFPNModel", "Module", "Sequential", "__init__", "class", "def", "feature_size", "for", "in", "in_features", "int", "lateral_width", "lateral_widths", "layers", "list", "nn", "self", "stem", "super"], "maskformer/modeling_maskformer.py:MaskFormerFPNModel.forward": ["ModelFPNModel", "Module", "Tensor", "append", "class", "def", "features", "for", "forward", "fpn_features", "in", "last_feature", "layer", "layers", "left", "list", "nn", "other_features", "output", "return", "self", "stem", "zip"], "maskformer/modeling_maskformer.py:MaskFormerPixelDecoder.__init__": ["Conv2d", "ModelFPNModel", "ModelPixelDecoder", "Model_feature_size", "Model_projection", "Module", "__init__", "args", "class", "def", "feature_size", "fpn", "int", "kernel_size", "kwargs", "nn", "padding", "r", "self", "super"], "maskformer/modeling_maskformer.py:MaskFormerPixelDecoder.forward": ["False", "ModelPixelDecoder", "ModelPixelDecoderOutput", "Model_projection", "Module", "Tensor", "True", "bool", "class", "def", "else", "features", "forward", "fpn", "fpn_features", "hidden_states", "if", "last_feature_projected", "last_hidden_state", "list", "nn", "not", "output_hidden_states", "return", "return_dict", "self", "tuple"], "maskformer/modeling_maskformer.py:MaskFormerSinePositionEmbedding.__init__": ["False", "ModelSinePositionEmbedding", "Module", "None", "True", "ValueError", "__init__", "and", "be", "bool", "class", "def", "else", "float", "if", "int", "is", "math", "nn", "normalize", "not", "num_pos_feats", "passed", "pi", "raise", "scale", "self", "should", "super", "temperature"], "maskformer/modeling_maskformer.py:MaskFormerSinePositionEmbedding.forward": ["Model", "ModelSinePositionEmbedding", "Module", "None", "Size", "Tensor", "arange", "bool", "cat", "class", "cos", "cumsum", "def", "device", "dim", "dim_t", "div", "dtype", "eps", "flatten", "floor", "forward", "if", "int64", "is", "nn", "normalize", "not_Model", "num_pos_feats", "permute", "pos", "pos_x", "pos_y", "return", "rounding_mode", "scale", "self", "shape", "sin", "stack", "str", "temperature", "to", "torch", "x_embed", "y_embed", "zeros"], "maskformer/modeling_maskformer.py:PredictionBlock.__init__": ["Linear", "ModelBlock", "Module", "None", "__init__", "activation", "add_module", "class", "def", "enumerate", "for", "i", "in", "in_dim", "int", "layer", "layers", "nn", "out_dim", "self", "str", "super"], "maskformer/modeling_maskformer.py:PredictionBlock.forward": ["ModelBlock", "Module", "Tensor", "class", "def", "for", "forward", "hidden_state", "in", "input", "layer", "layers", "nn", "return", "self"], "maskformer/modeling_maskformer.py:MaskformerMLPPredictionHead.__init__": ["Identity", "ModelMLPPredictionHead", "Module", "PredictionBlock", "ReLU", "__init__", "activation", "add_module", "append", "class", "def", "else", "enumerate", "for", "hidden_dim", "i", "if", "in", "in_dim", "in_dims", "input_dim", "int", "layer", "layers", "nn", "num_layers", "out_dim", "out_dims", "output_dim", "self", "str", "super", "zip"], "maskformer/modeling_maskformer.py:MaskformerMLPPredictionHead.forward": ["ModelMLPPredictionHead", "Module", "Tensor", "class", "def", "for", "forward", "hidden_state", "in", "input", "layer", "layers", "nn", "return", "self"], "maskformer/modeling_maskformer.py:MaskFormerPixelLevelModule.__init__": ["ModelConfig", "ModelPixelDecoder", "ModelPixelLevelModule", "ModelSwinConfig", "Model_feature_size", "Module", "None", "__init__", "and", "backbone_config", "channels", "class", "config", "decoder", "def", "encoder", "feature_channels", "feature_size", "fpn_feature_size", "from_dict", "getattr", "if", "in_features", "is", "lateral_widths", "load_backbone", "model_type", "nn", "not", "out_features", "self", "stage1", "stage2", "stage3", "stage4", "super", "swin", "to_dict"], "maskformer/modeling_maskformer.py:MaskFormerPixelLevelModule.forward": ["False", "ModelPixelLevelModule", "ModelPixelLevelModuleOutput", "Module", "Tensor", "True", "bool", "class", "decoder", "decoder_hidden_states", "decoder_last_hidden_state", "decoder_output", "def", "else", "encoder", "encoder_hidden_states", "encoder_last_hidden_state", "feature_maps", "features", "forward", "hidden_states", "if", "last_hidden_state", "nn", "not", "output_hidden_states", "outputs", "pixel_values", "return", "return_dict", "self", "tuple"], "maskformer/modeling_maskformer.py:MaskFormerTransformerModule.__init__": ["Conv2d", "DetrDecoder", "Embedding", "ModelConfig", "ModelSinePositionEmbedding", "ModelTransformerModule", "Module", "None", "True", "__init__", "class", "config", "decoder", "decoder_config", "def", "else", "hidden_size", "if", "in_features", "input_projection", "int", "kernel_size", "nn", "normalize", "num_pos_feats", "num_queries", "position_embedder", "queries_embedder", "self", "should_project", "super"], "maskformer/modeling_maskformer.py:MaskFormerTransformerModule.forward": ["DetrDecoderOutput", "False", "ModelTransformerModule", "Module", "None", "Tensor", "True", "attention_Model", "batch_size", "bool", "class", "decoder", "decoder_output", "def", "device", "dtype", "encoder_attention_Model", "encoder_hidden_states", "forward", "height", "if", "image_features", "input_projection", "inputs_embeds", "is", "nn", "not", "num_channels", "object_queries", "output_attentions", "output_hidden_states", "permute", "position_embedder", "queries_embedder", "queries_embeddings", "query_position_embeddings", "repeat", "requires_grad", "requires_grad_", "return", "return_dict", "self", "shape", "torch", "training", "unsqueeze", "view", "weight", "width", "zeros_like"], "maskformer/modeling_maskformer.py:MaskFormerPreTrainedModel._init_weights": ["BatchNorm2d", "Conv2d", "Embedding", "False", "LayerNorm", "Linear", "ModelFPNConvLayer", "ModelFPNLayer", "ModelFPNModel", "ModelLoss", "ModelMLPPredictionHead", "ModelPreTrainedModel", "ModelTransformerModule", "Module", "None", "PreTrainedModel", "_init_weights", "_is_hf_initialized", "and", "bias", "class", "config", "constant_", "copy_", "def", "elif", "empty_weight", "eos_coef", "for", "gain", "get_submodule", "getattr", "if", "in", "init", "init_std", "init_xavier_std", "input_projection", "is", "isinstance", "mean", "module", "modules", "nn", "normal_", "not", "num_batches_tracked", "num_labels", "ones", "ones_", "padding_idx", "proj", "running_mean", "running_var", "self", "std", "stem", "submodule", "torch", "weight", "xavier_std", "xavier_uniform_", "zeros_"], "maskformer/modeling_maskformer.py:MaskFormerModel.__init__": ["ModelConfig", "ModelModel", "ModelPixelLevelModule", "ModelPreTrainedModel", "ModelTransformerModule", "__init__", "channels", "class", "config", "def", "encoder", "in_features", "pixel_level_module", "post_init", "self", "super", "transformer_module"], "maskformer/modeling_maskformer.py:MaskFormerModel.forward": ["ModelModel", "ModelModelOutput", "ModelPreTrainedModel", "None", "Tensor", "ValueError", "You", "_", "attentions", "batch_size", "bool", "class", "config", "def", "device", "else", "encoder_hidden_states", "encoder_last_hidden_state", "for", "forward", "have", "height", "hidden_states", "if", "image_features", "in", "is", "kwargs", "last_hidden_state", "not", "ones", "output", "output_attentions", "output_hidden_states", "pixel_Model", "pixel_decoder_hidden_states", "pixel_decoder_last_hidden_state", "pixel_embeddings", "pixel_level_module", "pixel_level_module_output", "pixel_values", "queries", "r", "raise", "return", "return_dict", "self", "shape", "specify", "to", "torch", "transformer_decoder_hidden_states", "transformer_decoder_last_hidden_state", "transformer_module", "transformer_module_output", "tuple", "use_return_dict", "v", "values", "width"], "maskformer/modeling_maskformer.py:MaskFormerForInstanceSegmentation.__init__": ["Linear", "ModelConfig", "ModelForInstanceSegmentation", "ModelHungarianMatcher", "ModelLoss", "ModelMLPPredictionHead", "ModelModel", "ModelPreTrainedModel", "Model_embedder", "Model_feature_size", "Model_weight", "__init__", "class", "class_predictor", "config", "cost_Model", "cost_class", "cost_dice", "criterion", "cross_entropy_weight", "decoder_config", "def", "dice_weight", "dict", "eos_coef", "float", "hidden_size", "loss_Model", "loss_cross_entropy", "loss_dice", "matcher", "model", "nn", "no_object_weight", "num_labels", "post_init", "self", "str", "super", "weight_dict"], "maskformer/modeling_maskformer.py:MaskFormerForInstanceSegmentation.get_loss_dict": ["ModelForInstanceSegmentation", "ModelPreTrainedModel", "Model_labels", "Models_queries_logits", "Tensor", "auxiliary_logits", "class", "class_labels", "class_queries_logits", "criterion", "def", "dict", "for", "get_loss_dict", "if", "in", "items", "key", "loss", "loss_dict", "loss_key", "return", "self", "str", "weight", "weight_dict"], "maskformer/modeling_maskformer.py:MaskFormerForInstanceSegmentation.get_loss": ["ModelForInstanceSegmentation", "ModelPreTrainedModel", "Tensor", "class", "def", "dict", "get_loss", "loss_dict", "return", "self", "str", "sum", "values"], "maskformer/modeling_maskformer.py:MaskFormerForInstanceSegmentation.get_logits": ["ModelForInstanceSegmentation", "ModelModelOutput", "ModelPreTrainedModel", "Model_embedder", "Model_embeddings", "Models_queries_logits", "Tensor", "append", "aux_binary_Models", "aux_classes", "auxiliary_logits", "bchw", "binaries_Models", "bqc", "bqhw", "class", "class_predictor", "class_queries_logits", "classes", "config", "def", "dict", "einsum", "else", "for", "get_logits", "if", "in", "lbqc", "lbqhw", "list", "outputs", "pixel_decoder_last_hidden_state", "pixel_embeddings", "return", "self", "stack", "stacked_transformer_decoder_outputs", "str", "torch", "transformer_decoder_hidden_states", "transformer_decoder_last_hidden_state", "tuple", "use_auxiliary_loss", "zip"], "maskformer/modeling_maskformer.py:MaskFormerForInstanceSegmentation.forward": ["ModelForInstanceSegmentation", "ModelForInstanceSegmentationOutput", "ModelModelOutput", "ModelPreTrainedModel", "Model_labels", "Models_queries_logits", "None", "Tensor", "and", "attentions", "auxiliary_logits", "bool", "class", "class_labels", "class_queries_logits", "config", "def", "dict", "else", "encoder_hidden_states", "encoder_last_hidden_state", "for", "forward", "get_logits", "get_loss", "get_loss_dict", "hidden_states", "if", "in", "is", "kwargs", "list", "loss", "loss_dict", "model", "not", "or", "output", "output_attentions", "output_auxiliary_logits", "output_hidden_states", "outputs", "pixel_Model", "pixel_decoder_hidden_states", "pixel_decoder_last_hidden_state", "pixel_values", "r", "raw_outputs", "return", "return_dict", "self", "str", "transformer_decoder_hidden_states", "transformer_decoder_last_hidden_state", "tuple", "use_auxiliary_loss", "use_return_dict", "v", "values"], "maskformer/modeling_maskformer_swin.py:window_partition": ["Model_partition", "Model_size", "Models", "batch_size", "contiguous", "def", "height", "input_feature", "num_channels", "permute", "return", "shape", "view", "width"], "maskformer/modeling_maskformer_swin.py:window_reverse": ["Model_reverse", "Model_size", "Models", "contiguous", "def", "height", "num_channels", "permute", "return", "shape", "view", "width"], "maskformer/modeling_maskformer_swin.py:drop_path": ["False", "Model_path", "Model_prob", "Tensor", "bool", "def", "device", "div", "dtype", "float", "floor_", "if", "input", "keep_prob", "ndim", "not", "or", "output", "rand", "random_tensor", "return", "shape", "torch", "training"], "maskformer/modeling_maskformer_swin.py:MaskFormerSwinEmbeddings.__init__": ["Dropout", "LayerNorm", "ModelSwinEmbeddings", "ModelSwinPatchEmbeddings", "Module", "None", "Parameter", "__init__", "class", "config", "def", "dropout", "else", "embed_dim", "grid_size", "hidden_dropout_prob", "if", "nn", "norm", "num_patches", "patch_embeddings", "patch_grid", "patch_size", "position_embeddings", "self", "super", "torch", "use_absolute_embeddings", "zeros"], "maskformer/modeling_maskformer_swin.py:MaskFormerSwinEmbeddings.interpolate_pos_encoding": ["False", "ModelSwinEmbeddings", "Module", "Tensor", "align_corners", "and", "bicubic", "cat", "class", "class_pos_embed", "def", "dim", "embeddings", "functional", "height", "if", "int", "interpolate", "interpolate_pos_encoding", "is_tracing", "jit", "mode", "new_height", "new_width", "nn", "not", "num_patches", "num_positions", "patch_pos_embed", "patch_size", "permute", "position_embeddings", "reshape", "return", "self", "shape", "size", "sqrt_num_positions", "torch", "torch_int", "view", "width"], "maskformer/modeling_maskformer_swin.py:MaskFormerSwinEmbeddings.forward": ["ModelSwinEmbeddings", "Module", "None", "_", "class", "def", "dropout", "else", "embeddings", "forward", "height", "if", "interpolate_pos_encoding", "is", "nn", "norm", "not", "num_channels", "output_dimensions", "patch_embeddings", "pixel_values", "position_embeddings", "return", "self", "shape", "width"], "maskformer/modeling_maskformer_swin.py:MaskFormerSwinPatchEmbeddings.__init__": ["Conv2d", "Iterable", "ModelSwinPatchEmbeddings", "Module", "__init__", "abc", "class", "collections", "config", "def", "else", "embed_dim", "grid_size", "hidden_size", "if", "image_size", "isinstance", "kernel_size", "nn", "num_channels", "num_patches", "patch_size", "projection", "self", "stride", "super"], "maskformer/modeling_maskformer_swin.py:MaskFormerSwinPatchEmbeddings.maybe_pad": ["ModelSwinPatchEmbeddings", "Module", "class", "def", "functional", "height", "if", "maybe_pad", "nn", "pad", "pad_values", "patch_size", "pixel_values", "return", "self", "width"], "maskformer/modeling_maskformer_swin.py:MaskFormerSwinPatchEmbeddings.forward": ["FloatTensor", "ModelSwinPatchEmbeddings", "Module", "None", "Tensor", "_", "class", "def", "embeddings", "flatten", "forward", "height", "int", "maybe_pad", "nn", "num_channels", "output_dimensions", "pixel_values", "projection", "return", "self", "shape", "torch", "transpose", "tuple", "width"], "maskformer/modeling_maskformer_swin.py:MaskFormerSwinPatchMerging.__init__": ["False", "LayerNorm", "Linear", "ModelSwinPatchMerging", "Module", "None", "__init__", "bias", "class", "def", "dim", "input_resolution", "int", "nn", "norm", "norm_layer", "reduction", "self", "super", "tuple"], "maskformer/modeling_maskformer_swin.py:MaskFormerSwinPatchMerging.maybe_pad": ["ModelSwinPatchMerging", "Module", "class", "def", "functional", "height", "if", "input_feature", "maybe_pad", "nn", "or", "pad", "pad_values", "return", "self", "should_pad", "width"], "maskformer/modeling_maskformer_swin.py:MaskFormerSwinPatchMerging.forward": ["ModelSwinPatchMerging", "Module", "Tensor", "batch_size", "cat", "class", "def", "dim", "forward", "height", "input_dimensions", "input_feature", "input_feature_0", "input_feature_1", "input_feature_2", "input_feature_3", "int", "maybe_pad", "nn", "norm", "num_channels", "reduction", "return", "self", "shape", "torch", "tuple", "view", "width"], "maskformer/modeling_maskformer_swin.py:MaskFormerSwinDropPath.__init__": ["ModelSwinDropPath", "Module", "None", "__init__", "class", "def", "drop_prob", "float", "nn", "self", "super"], "maskformer/modeling_maskformer_swin.py:MaskFormerSwinDropPath.forward": ["ModelSwinDropPath", "Module", "Tensor", "class", "def", "drop_path", "drop_prob", "forward", "hidden_states", "nn", "return", "self", "torch", "training"], "maskformer/modeling_maskformer_swin.py:MaskFormerSwinDropPath.extra_repr": ["ModelSwinDropPath", "Module", "class", "def", "drop_prob", "extra_repr", "f", "nn", "p", "return", "self", "str"], "maskformer/modeling_maskformer_swin.py:MaskFormerSwinSelfAttention.__init__": ["Dropout", "Iterable", "Linear", "ModelSwinSelfAttention", "Module", "Parameter", "The", "ValueError", "__init__", "a", "abc", "all_head_size", "attention", "attention_head_size", "attention_probs_dropout_prob", "bias", "class", "collections", "config", "create_relative_position_index", "def", "dim", "dropout", "else", "f", "heads", "hidden", "if", "int", "is", "isinstance", "key", "multiple", "nn", "not", "num_attention_heads", "num_heads", "number", "of", "qkv_bias", "query", "raise", "register_buffer", "relative_position_bias_table", "relative_position_index", "self", "size", "super", "the", "torch", "value", "window_size", "zeros"], "maskformer/modeling_maskformer_swin.py:MaskFormerSwinSelfAttention.forward": ["False", "FloatTensor", "ModelSwinSelfAttention", "Model_shape", "Module", "None", "Tensor", "all_head_size", "attention_Model", "attention_head_size", "attention_probs", "attention_scores", "batch_size", "bool", "class", "context_layer", "contiguous", "def", "dim", "dropout", "else", "forward", "functional", "hidden_shape", "hidden_states", "if", "is", "key", "key_layer", "math", "matmul", "new_context_layer_shape", "nn", "not", "num_attention_heads", "num_channels", "output_attentions", "outputs", "permute", "query", "query_layer", "relative_position_bias", "relative_position_bias_table", "relative_position_index", "return", "self", "shape", "size", "softmax", "sqrt", "torch", "transpose", "tuple", "unsqueeze", "value", "value_layer", "view", "window_size"], "maskformer/modeling_maskformer_swin.py:MaskFormerSwinSelfAttention.create_relative_position_index": ["ModelSwinSelfAttention", "Module", "None", "arange", "class", "contiguous", "coords", "coords_flatten", "coords_h", "coords_w", "create_relative_position_index", "def", "flatten", "ij", "indexing", "meshgrid", "nn", "permute", "relative_coords", "relative_position_index", "return", "self", "stack", "sum", "torch", "window_size"], "maskformer/modeling_maskformer_swin.py:MaskFormerSwinSelfOutput.__init__": ["Dropout", "Linear", "ModelSwinSelfOutput", "Module", "__init__", "attention_probs_dropout_prob", "class", "config", "def", "dense", "dim", "dropout", "nn", "self", "super"], "maskformer/modeling_maskformer_swin.py:MaskFormerSwinSelfOutput.forward": ["ModelSwinSelfOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "maskformer/modeling_maskformer_swin.py:MaskFormerSwinAttention.__init__": ["ModelSwinAttention", "ModelSwinSelfAttention", "ModelSwinSelfOutput", "Module", "__init__", "class", "config", "def", "dim", "nn", "num_heads", "output", "self", "super", "window_size"], "maskformer/modeling_maskformer_swin.py:MaskFormerSwinAttention.forward": ["False", "FloatTensor", "ModelSwinAttention", "Module", "None", "Tensor", "attention_Model", "attention_output", "bool", "class", "def", "forward", "hidden_states", "nn", "output", "output_attentions", "outputs", "return", "self", "self_outputs", "torch", "tuple"], "maskformer/modeling_maskformer_swin.py:MaskFormerSwinIntermediate.__init__": ["ACT2FN", "Linear", "ModelSwinIntermediate", "Module", "__init__", "class", "config", "def", "dense", "dim", "else", "hidden_act", "if", "int", "intermediate_act_fn", "isinstance", "mlp_ratio", "nn", "self", "str", "super"], "maskformer/modeling_maskformer_swin.py:MaskFormerSwinIntermediate.forward": ["ModelSwinIntermediate", "Module", "Tensor", "class", "def", "dense", "forward", "hidden_states", "intermediate_act_fn", "nn", "return", "self", "torch"], "maskformer/modeling_maskformer_swin.py:MaskFormerSwinOutput.__init__": ["Dropout", "Linear", "ModelSwinOutput", "Module", "__init__", "class", "config", "def", "dense", "dim", "dropout", "hidden_dropout_prob", "int", "mlp_ratio", "nn", "self", "super"], "maskformer/modeling_maskformer_swin.py:MaskFormerSwinOutput.forward": ["ModelSwinOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "nn", "return", "self", "torch"], "maskformer/modeling_maskformer_swin.py:MaskFormerSwinLayer.__init__": ["Identity", "LayerNorm", "ModelSwinAttention", "ModelSwinDropPath", "ModelSwinIntermediate", "ModelSwinLayer", "ModelSwinOutput", "Module", "__init__", "attention", "class", "config", "def", "dim", "drop_path", "drop_path_rate", "else", "eps", "if", "input_resolution", "intermediate", "layer_norm_eps", "layernorm_after", "layernorm_before", "nn", "num_heads", "output", "self", "shift_size", "super", "window_size"], "maskformer/modeling_maskformer_swin.py:MaskFormerSwinLayer.get_attn_mask": ["ModelSwinLayer", "Model_windows", "Modeled_fill", "Module", "None", "attn_Model", "class", "count", "def", "else", "for", "get_attn_Model", "height", "height_slice", "height_slices", "if", "img_Model", "in", "input_resolution", "nn", "return", "self", "shift_size", "slice", "torch", "unsqueeze", "view", "width", "width_slice", "width_slices", "window_partition", "window_size", "zeros"], "maskformer/modeling_maskformer_swin.py:MaskFormerSwinLayer.maybe_pad": ["ModelSwinLayer", "Module", "class", "def", "functional", "height", "hidden_states", "maybe_pad", "nn", "pad", "pad_bottom", "pad_left", "pad_right", "pad_top", "pad_values", "return", "self", "width", "window_size"], "maskformer/modeling_maskformer_swin.py:MaskFormerSwinLayer.forward": ["False", "ModelSwinLayer", "Module", "None", "_", "attention", "attention_output", "attention_windows", "attn_Model", "batch_size", "channels", "class", "contiguous", "def", "device", "dim", "dims", "drop_path", "else", "forward", "get_attn_Model", "height", "height_pad", "hidden_states", "hidden_states_windows", "if", "input_dimensions", "intermediate", "is", "layer_output", "layernorm_after", "layernorm_before", "maybe_pad", "nn", "not", "or", "output", "output_attentions", "outputs", "pad_values", "return", "roll", "self", "self_attention_outputs", "shape", "shift_size", "shifted_hidden_states", "shifted_windows", "shifts", "shortcut", "size", "to", "torch", "view", "was_padded", "width", "width_pad", "window_partition", "window_reverse", "window_size"], "maskformer/modeling_maskformer_swin.py:MaskFormerSwinStage.__init__": ["False", "GradientCheckpointingLayer", "LayerNorm", "ModelSwinLayer", "ModelSwinStage", "ModuleList", "None", "__init__", "blocks", "class", "config", "def", "depth", "dim", "downsample", "drop_path", "drop_path_rate", "else", "for", "i", "if", "in", "input_resolution", "is", "nn", "norm_layer", "not", "num_heads", "pointing", "range", "self", "shift_size", "super", "window_size"], "maskformer/modeling_maskformer_swin.py:MaskFormerSwinStage.forward": ["False", "GradientCheckpointingLayer", "ModelSwinStage", "None", "all_hidden_states", "block_hidden_states", "block_module", "blocks", "class", "def", "downsample", "else", "enumerate", "for", "forward", "height", "height_downsampled", "hidden_states", "i", "if", "in", "input_dimensions", "is", "not", "output_attentions", "output_dimensions", "output_hidden_states", "return", "self", "width", "width_downsampled"], "maskformer/modeling_maskformer_swin.py:MaskFormerSwinEncoder.__init__": ["False", "ModelSwinEncoder", "ModelSwinPatchMerging", "ModelSwinStage", "Module", "ModuleList", "None", "__init__", "class", "config", "cpu", "def", "depth", "depths", "device", "dim", "downsample", "dpr", "drop_path", "drop_path_rate", "else", "embed_dim", "for", "gradient_checkpointing", "grid_size", "i_layer", "if", "in", "input_resolution", "int", "item", "layers", "len", "linspace", "nn", "num_heads", "num_layers", "range", "self", "sum", "super", "torch", "x"], "maskformer/modeling_maskformer_swin.py:MaskFormerSwinEncoder.forward": ["False", "ModelSwinBaseModelOutput", "ModelSwinEncoder", "Module", "None", "True", "all_hidden_states", "all_input_dimensions", "all_self_attentions", "attentions", "class", "def", "else", "enumerate", "for", "forward", "hidden_states", "hidden_states_spatial_dimensions", "i", "if", "in", "input_dimensions", "is", "last_hidden_state", "layer_all_hidden_states", "layer_hidden_states", "layer_module", "layers", "nn", "not", "output_attentions", "output_dimensions", "output_hidden_states", "return", "return_dict", "self", "tuple", "v"], "maskformer/modeling_maskformer_swin.py:MaskFormerSwinPreTrainedModel._init_weights": ["ModelSwinEmbeddings", "ModelSwinPreTrainedModel", "ModelSwinSelfAttention", "None", "PreTrainedModel", "_init_weights", "class", "copy_", "create_relative_position_index", "def", "elif", "if", "init", "is", "isinstance", "module", "not", "position_embeddings", "relative_position_bias_table", "relative_position_index", "self", "super", "zeros_"], "maskformer/modeling_maskformer_swin.py:MaskFormerSwinModel.__init__": ["AdaptiveAvgPool1d", "LayerNorm", "ModelSwinEmbeddings", "ModelSwinEncoder", "ModelSwinModel", "ModelSwinPreTrainedModel", "None", "True", "__init__", "add_pooling_layer", "class", "config", "def", "depths", "else", "embed_dim", "embeddings", "encoder", "eps", "if", "int", "layer_norm_eps", "layernorm", "len", "nn", "num_features", "num_layers", "patch_grid", "pooler", "post_init", "self", "super"], "maskformer/modeling_maskformer_swin.py:MaskFormerSwinModel.get_input_embeddings": ["ModelSwinModel", "ModelSwinPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "patch_embeddings", "return", "self"], "maskformer/modeling_maskformer_swin.py:MaskFormerSwinModel.forward": ["False", "ModelSwinModel", "ModelSwinModelOutputWithPooling", "ModelSwinPreTrainedModel", "None", "ValueError", "You", "attentions", "class", "config", "def", "else", "embedding_output", "embeddings", "encoder", "encoder_outputs", "flatten", "forward", "have", "hidden_states", "hidden_states_spatial_dimensions", "if", "input_dimensions", "interpolate_pos_encoding", "is", "kwargs", "last_hidden_state", "layernorm", "not", "output_attentions", "output_hidden_states", "pixel_values", "pooled_output", "pooler", "pooler_output", "raise", "return", "return_dict", "self", "sequence_output", "specify", "to", "torch", "transpose", "tuple", "use_return_dict"], "maskformer/modeling_maskformer_swin.py:MaskFormerSwinBackbone.__init__": ["BackboneMixin", "LayerNorm", "ModelSwinBackbone", "ModelSwinConfig", "ModelSwinModel", "ModelSwinPreTrainedModel", "ModuleList", "This", "ValueError", "__init__", "_init_backbone", "backbone", "class", "config", "def", "depths", "does", "embed_dim", "for", "hidden_states_norms", "i", "if", "in", "int", "len", "model", "nn", "not", "num_channels", "num_features", "out_features", "post_init", "raise", "range", "self", "stem", "super", "support", "the"], "maskformer/modeling_maskformer_swin.py:MaskFormerSwinBackbone.forward": ["BackboneMixin", "BackboneOutput", "ModelSwinBackbone", "ModelSwinPreTrainedModel", "None", "Tensor", "True", "_", "attentions", "batch_size", "bool", "class", "config", "contiguous", "def", "else", "enumerate", "feature_maps", "for", "forward", "height", "hidden_size", "hidden_state", "hidden_state_norm", "hidden_state_permuted", "hidden_state_unpolled", "hidden_states", "hidden_states_norms", "hidden_states_spatial_dimensions", "i", "if", "in", "int", "is", "kwargs", "model", "norm", "not", "out_features", "output", "output_attentions", "output_hidden_states", "outputs", "permute", "pixel_values", "return", "return_dict", "self", "shape", "spatial_dimensions", "stage", "stage_names", "tuple", "use_return_dict", "view", "width", "zip"], "mbart/modeling_mbart.py:shift_tokens_right": ["Model_tokens_right", "None", "Tensor", "ValueError", "be", "clone", "config", "decoder_start_tokens", "def", "defined", "dim", "gather", "has", "if", "index_of_eos", "input_ids", "int", "is", "masked_fill_", "model", "ne", "pad_token_id", "prev_output_tokens", "raise", "return", "self", "squeeze", "sum", "to", "torch", "unsqueeze"], "mbart/modeling_mbart.py:MBartLearnedPositionalEmbedding.__init__": ["Embedding", "Model", "__init__", "class", "def", "embedding_dim", "int", "nn", "num_embeddings", "offset", "self", "super"], "mbart/modeling_mbart.py:MBartLearnedPositionalEmbedding.forward": ["Embedding", "Model", "None", "Tensor", "arange", "bsz", "class", "def", "device", "dtype", "else", "expand", "forward", "if", "input_ids", "int", "is", "long", "nn", "offset", "past_key_values_length", "position_ids", "return", "self", "seq_len", "shape", "super", "torch", "unsqueeze", "weight"], "mbart/modeling_mbart.py:MBartScaledWordEmbedding.__init__": ["Embedding", "Model", "None", "__init__", "class", "def", "embed_scale", "embedding_dim", "float", "int", "nn", "num_embeddings", "padding_idx", "self", "super"], "mbart/modeling_mbart.py:MBartScaledWordEmbedding.forward": ["Embedding", "Model", "Tensor", "class", "def", "embed_scale", "forward", "input_ids", "nn", "return", "self", "super", "torch"], "mbart/modeling_mbart.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "contiguous", "def", "dim", "dropout", "float", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "shape", "size", "softmax", "torch", "training", "transpose", "value"], "mbart/modeling_mbart.py:MBartAttention.__init__": ["False", "Instantiating", "Linear", "Model", "ModelConfig", "Module", "None", "Please", "True", "ValueError", "__class__", "__init__", "__name__", "a", "and", "be", "bias", "bool", "by", "caching", "call", "class", "config", "creating", "decoder", "def", "divisible", "dropout", "during", "embed_dim", "errors", "f", "float", "forward", "got", "head_dim", "if", "int", "is", "is_causal", "is_decoder", "k_proj", "layer_idx", "lead", "logger", "make", "must", "nn", "not", "num_heads", "out_proj", "passing", "provide", "q_proj", "raise", "recommended", "scaling", "self", "super", "sure", "the", "this", "to", "used", "v_proj", "warning_once", "when", "will", "without"], "mbart/modeling_mbart.py:MBartAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "EncoderDecoderCache", "False", "FlashAttentionKwargs", "Model", "Module", "None", "Tensor", "True", "Unpack", "_attn_implementation", "and", "attention_interface", "attention_mask", "attn_output", "attn_weights", "bool", "bsz", "cache_position", "class", "config", "contiguous", "cross_attention_cache", "curr_past_key_values", "current_states", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "get", "head_dim", "hidden_states", "if", "is", "is_cross_attention", "is_updated", "isinstance", "k_proj", "key_states", "key_value_states", "keys", "kv_input_shape", "kwargs", "layer_idx", "layers", "nn", "not", "out_proj", "output_attentions", "past_key_values", "q_input_shape", "q_proj", "query_states", "reshape", "return", "scaling", "self", "self_attention_cache", "shape", "src_len", "tgt_len", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "values", "view"], "mbart/modeling_mbart.py:MBartEncoderLayer.__init__": ["ACT2FN", "GradientCheckpointingLayer", "LayerNorm", "Linear", "Model", "ModelAttention", "ModelConfig", "__init__", "activation_dropout", "activation_fn", "activation_function", "attention_dropout", "class", "config", "d_model", "def", "dropout", "embed_dim", "encoder_attention_heads", "encoder_ffn_dim", "fc1", "fc2", "final_layer_norm", "nn", "num_heads", "self", "self_attn", "self_attn_layer_norm", "super"], "mbart/modeling_mbart.py:MBartEncoderLayer.forward": ["False", "GradientCheckpointingLayer", "Model", "Tensor", "activation_dropout", "activation_fn", "attention_mask", "attn_weights", "bool", "clamp", "clamp_value", "class", "def", "dropout", "dtype", "fc1", "fc2", "final_layer_norm", "finfo", "float16", "forward", "functional", "hidden_states", "if", "max", "min", "nn", "output_attentions", "p", "residual", "return", "self", "self_attn", "self_attn_layer_norm", "torch", "training"], "mbart/modeling_mbart.py:MBartDecoderLayer.__init__": ["ACT2FN", "GradientCheckpointingLayer", "LayerNorm", "Linear", "Model", "ModelAttention", "ModelConfig", "None", "True", "__init__", "activation_dropout", "activation_fn", "activation_function", "attention_dropout", "class", "config", "d_model", "decoder_attention_heads", "decoder_ffn_dim", "def", "dropout", "embed_dim", "encoder_attn", "encoder_attn_layer_norm", "fc1", "fc2", "final_layer_norm", "int", "is_causal", "is_decoder", "layer_idx", "nn", "num_heads", "self", "self_attn", "self_attn_layer_norm", "super"], "mbart/modeling_mbart.py:MBartDecoderLayer.forward": ["Cache", "False", "GradientCheckpointingLayer", "Model", "None", "Tensor", "True", "activation_dropout", "activation_fn", "attention_mask", "bool", "cache_position", "class", "cross_attn_weights", "def", "dropout", "encoder_attention_mask", "encoder_attn", "encoder_attn_layer_norm", "encoder_hidden_states", "fc1", "fc2", "final_layer_norm", "forward", "functional", "hidden_states", "if", "is", "key_value_states", "nn", "not", "output_attentions", "outputs", "p", "past_key_values", "residual", "return", "self", "self_attn", "self_attn_layer_norm", "self_attn_weights", "torch", "training", "use_cache"], "mbart/modeling_mbart.py:MBartClassificationHead.__init__": ["Dropout", "Linear", "Model", "Module", "__init__", "class", "def", "dense", "dropout", "float", "inner_dim", "input_dim", "int", "nn", "num_classes", "out_proj", "p", "pooler_dropout", "self", "super"], "mbart/modeling_mbart.py:MBartClassificationHead.forward": ["Model", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "nn", "out_proj", "return", "self", "tanh", "torch"], "mbart/modeling_mbart.py:MBartPreTrainedModel._init_weights": ["Model", "ModelForConditionalGeneration", "PreTrainedModel", "_init_weights", "class", "def", "final_logits_bias", "if", "init", "isinstance", "module", "self", "super", "zeros_"], "mbart/modeling_mbart.py:MBartPreTrainedModel.dummy_inputs": ["Model", "PreTrainedModel", "attention_mask", "class", "config", "def", "device", "dummy_inputs", "input_ids", "ne", "pad_token", "pad_token_id", "return", "self", "tensor", "torch"], "mbart/modeling_mbart.py:MBartEncoder.__init__": ["False", "LayerNorm", "Model", "ModelConfig", "ModelLayer", "ModelLearnedPositionalEmbedding", "ModelPreTrainedModel", "ModelScaledWordEmbedding", "ModuleList", "_", "__init__", "class", "config", "d_model", "def", "dropout", "else", "embed_dim", "embed_positions", "embed_scale", "embed_tokens", "encoder_layerdrop", "encoder_layers", "for", "gradient_checkpointing", "if", "in", "layer_norm", "layerdrop", "layernorm_embedding", "layers", "math", "max_position_embeddings", "max_source_positions", "nn", "pad_token_id", "padding_idx", "post_init", "range", "scale_embedding", "self", "sqrt", "super", "vocab_size"], "mbart/modeling_mbart.py:MBartEncoder._backward_compatibility_gradient_checkpointing": ["False", "Model", "ModelPreTrainedModel", "_backward_compatibility_gradient_checkpointing", "and", "class", "config", "def", "getattr", "gradient_checkpointing", "gradient_checkpointing_enable", "if", "self", "supports_gradient_checkpointing"], "mbart/modeling_mbart.py:MBartEncoder.forward": ["BaseModelOutput", "False", "FloatTensor", "LongTensor", "Model", "ModelPreTrainedModel", "None", "Tensor", "True", "ValueError", "You", "all_attentions", "and", "at", "attention_mask", "attentions", "bool", "both", "cannot", "class", "config", "create_bidirectional_mask", "def", "device", "dropout", "dropout_probability", "either", "elif", "else", "embed_pos", "embed_positions", "embed_tokens", "encoder_layer", "encoder_states", "enumerate", "for", "forward", "functional", "have", "hidden_states", "idx", "if", "in", "input", "input_embeds", "input_ids", "input_shape", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layer_norm", "layer_outputs", "layerdrop", "layernorm_embedding", "layers", "nn", "not", "or", "output_attentions", "output_hidden_states", "p", "r", "raise", "rand", "return", "return_dict", "same", "self", "shape", "specify", "the", "time", "to", "to_drop", "torch", "training", "tuple", "use_return_dict", "v", "view"], "mbart/modeling_mbart.py:MBartDecoder.__init__": ["False", "LayerNorm", "Model", "ModelConfig", "ModelLayer", "ModelLearnedPositionalEmbedding", "ModelPreTrainedModel", "ModelScaledWordEmbedding", "ModuleList", "__init__", "class", "config", "d_model", "decoder_layerdrop", "decoder_layers", "def", "dropout", "else", "embed_positions", "embed_scale", "embed_tokens", "for", "gradient_checkpointing", "i", "if", "in", "layer_idx", "layer_norm", "layerdrop", "layernorm_embedding", "layers", "math", "max_position_embeddings", "max_target_positions", "nn", "pad_token_id", "padding_idx", "post_init", "range", "scale_embedding", "self", "sqrt", "super", "vocab_size"], "mbart/modeling_mbart.py:MBartDecoder.forward": ["BaseModelOutputWithPastAndCrossAttentions", "Cache", "DynamicCache", "EncoderDecoderCache", "False", "FloatTensor", "LongTensor", "Model", "ModelPreTrainedModel", "None", "Setting", "Tensor", "True", "ValueError", "You", "all_cross_attentions", "all_hidden_states", "all_self_attns", "and", "arange", "at", "attention_mask", "attentions", "batch_size", "bool", "both", "cache_position", "cannot", "causal_mask", "checkpointing", "class", "config", "continue", "create_bidirectional_mask", "create_causal_mask", "cross_attentions", "decoder_input_ids", "decoder_inputs_embeds", "decoder_layer", "def", "device", "dropout", "dropout_probability", "either", "elif", "else", "embed_positions", "embed_tokens", "encoder_attention_mask", "encoder_hidden_states", "enumerate", "for", "forward", "functional", "get_seq_length", "gradient", "gradient_checkpointing", "have", "hidden_states", "idx", "if", "in", "incompatible", "input", "input_embeds", "input_ids", "input_shape", "inputs_embeds", "is", "is_encoder_decoder", "is_torchdynamo_compiling", "isinstance", "kwargs", "last_hidden_state", "layer_norm", "layer_outputs", "layerdrop", "layernorm_embedding", "layers", "logger", "mask_seq_length", "nn", "not", "ones", "or", "output_attentions", "output_hidden_states", "p", "past_key_values", "past_key_values_length", "position_ids", "r", "raise", "rand", "return", "return_dict", "same", "self", "self_attention_cache", "self_attn_cache", "seq_length", "shape", "size", "specify", "the", "time", "to", "torch", "training", "tuple", "use_cache", "use_return_dict", "v", "view", "warning_once", "with"], "mbart/modeling_mbart.py:MBartModel.__init__": ["Model", "ModelConfig", "ModelDecoder", "ModelEncoder", "ModelPreTrainedModel", "ModelScaledWordEmbedding", "__init__", "class", "config", "d_model", "decoder", "def", "else", "embed_scale", "encoder", "if", "math", "pad_token_id", "padding_idx", "post_init", "scale_embedding", "self", "shared", "sqrt", "super", "vocab_size"], "mbart/modeling_mbart.py:MBartModel.get_input_embeddings": ["Model", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "return", "self", "shared"], "mbart/modeling_mbart.py:MBartModel.set_input_embeddings": ["Model", "ModelPreTrainedModel", "class", "decoder", "def", "embed_tokens", "encoder", "self", "set_input_embeddings", "shared", "value"], "mbart/modeling_mbart.py:MBartModel.forward": ["BaseModelOutput", "Cache", "FloatTensor", "LongTensor", "Model", "ModelPreTrainedModel", "None", "Seq2SeqModelOutput", "Tensor", "and", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "cross_attentions", "decoder", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_input_ids", "decoder_inputs_embeds", "decoder_outputs", "def", "elif", "else", "encoder", "encoder_attention_mask", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "isinstance", "kwargs", "last_hidden_state", "len", "not", "output_attentions", "output_hidden_states", "pad_token_id", "past_key_values", "r", "return", "return_dict", "self", "shift_tokens_right", "torch", "tuple", "use_cache", "use_return_dict"], "mbart/modeling_mbart.py:MBartForConditionalGeneration.__init__": ["False", "GenerationMixin", "Linear", "Model", "ModelConfig", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "d_model", "def", "final_logits_bias", "lm_head", "model", "nn", "num_embeddings", "post_init", "register_buffer", "self", "shared", "super", "torch", "zeros"], "mbart/modeling_mbart.py:MBartForConditionalGeneration.resize_token_embeddings": ["Embedding", "GenerationMixin", "Model", "ModelPreTrainedModel", "None", "True", "_resize_final_logits_bias", "bool", "class", "def", "int", "mean_resizing", "new_embeddings", "new_num_tokens", "nn", "pad_to_multiple_of", "resize_token_embeddings", "return", "self", "shape", "super", "weight"], "mbart/modeling_mbart.py:MBartForConditionalGeneration._resize_final_logits_bias": ["GenerationMixin", "Model", "ModelPreTrainedModel", "None", "_resize_final_logits_bias", "cat", "class", "def", "device", "dim", "else", "extra_bias", "final_logits_bias", "if", "int", "new_bias", "new_num_tokens", "old_num_tokens", "register_buffer", "self", "shape", "torch", "zeros"], "mbart/modeling_mbart.py:MBartForConditionalGeneration.forward": ["Cache", "CrossEntropyLoss", "False", "FloatTensor", "GenerationMixin", "LongTensor", "Model", "ModelPreTrainedModel", "None", "Seq2SeqLMOutput", "Tensor", "The", "and", "argument", "attention_mask", "bool", "cache_position", "changed", "class", "config", "cross_attentions", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_input_ids", "decoder_inputs_embeds", "def", "else", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "final_logits_bias", "forward", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "lm_head", "lm_logits", "logger", "logits", "loss", "loss_fct", "masked_lm_loss", "model", "not", "output", "output_attentions", "output_hidden_states", "outputs", "pad_token_id", "past_key_values", "provided", "r", "return", "return_dict", "self", "shift_tokens_right", "since", "to", "torch", "tuple", "use_cache", "use_return_dict", "view", "vocab_size", "warning"], "mbart/modeling_mbart.py:MBartForConditionalGeneration.prepare_decoder_input_ids_from_labels": ["GenerationMixin", "Model", "ModelPreTrainedModel", "Tensor", "class", "config", "def", "labels", "pad_token_id", "prepare_decoder_input_ids_from_labels", "return", "self", "shift_tokens_right", "torch"], "mbart/modeling_mbart.py:MBartForSequenceClassification.__init__": ["Model", "ModelClassificationHead", "ModelConfig", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "classification_head", "classifier_dropout", "config", "d_model", "def", "kwargs", "model", "num_labels", "post_init", "self", "super"], "mbart/modeling_mbart.py:MBartForSequenceClassification.forward": ["All", "BCEWithLogitsLoss", "CrossEntropyLoss", "False", "FloatTensor", "LongTensor", "MSELoss", "Model", "ModelPreTrainedModel", "None", "NotImplementedError", "Passing", "Seq2SeqSequenceClassifierOutput", "Tensor", "ValueError", "__class__", "__name__", "and", "attention_mask", "bool", "cache_position", "class", "classification_head", "config", "cross_attentions", "currently", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_input_ids", "decoder_inputs_embeds", "def", "device", "dtype", "elif", "else", "embeddings", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "eos", "eos_mask", "eos_token_id", "eq", "examples", "f", "for", "forward", "have", "hidden_states", "if", "input", "input_ids", "inputs_embeds", "int", "is", "kwargs", "labels", "len", "list", "logits", "long", "loss", "loss_fct", "model", "multi_label_classification", "must", "not", "num_labels", "number", "of", "or", "output", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "problem_type", "r", "raise", "regression", "return", "return_dict", "same", "self", "sentence_representation", "single_label_classification", "size", "squeeze", "sum", "supported", "the", "to", "tokens", "torch", "tuple", "unique_consecutive", "use_cache", "use_return_dict", "view"], "mbart/modeling_mbart.py:MBartForQuestionAnswering.__init__": ["Linear", "Model", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "config", "def", "hidden_size", "model", "nn", "num_labels", "post_init", "qa_outputs", "self", "super"], "mbart/modeling_mbart.py:MBartForQuestionAnswering.forward": ["CrossEntropyLoss", "False", "FloatTensor", "LongTensor", "Model", "ModelPreTrainedModel", "None", "Seq2SeqQuestionAnsweringModelOutput", "Tensor", "and", "attention_mask", "bool", "cache_position", "clamp", "class", "config", "contiguous", "cross_attentions", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_input_ids", "decoder_inputs_embeds", "def", "dim", "else", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "end_logits", "end_loss", "end_positions", "forward", "if", "ignore_index", "ignored_index", "input_ids", "inputs_embeds", "is", "kwargs", "len", "list", "logits", "loss", "loss_fct", "model", "not", "output", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "qa_outputs", "r", "return", "return_dict", "self", "sequence_output", "size", "split", "squeeze", "start_logits", "start_loss", "start_positions", "torch", "total_loss", "tuple", "use_cache", "use_return_dict"], "mbart/modeling_mbart.py:MBartDecoderWrapper.__init__": ["Model", "ModelDecoder", "ModelPreTrainedModel", "__init__", "class", "config", "decoder", "def", "post_init", "self", "super"], "mbart/modeling_mbart.py:MBartDecoderWrapper.forward": ["Model", "ModelPreTrainedModel", "args", "class", "decoder", "def", "forward", "kwargs", "return", "self"], "mbart/modeling_mbart.py:MBartForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "Model", "ModelDecoderWrapper", "ModelPreTrainedModel", "True", "__init__", "bias", "class", "config", "def", "hidden_size", "is_decoder", "is_encoder_decoder", "lm_head", "model", "nn", "post_init", "self", "super", "vocab_size"], "mbart/modeling_mbart.py:MBartForCausalLM.get_input_embeddings": ["GenerationMixin", "Model", "ModelPreTrainedModel", "class", "decoder", "def", "embed_tokens", "get_input_embeddings", "model", "return", "self"], "mbart/modeling_mbart.py:MBartForCausalLM.set_input_embeddings": ["GenerationMixin", "Model", "ModelPreTrainedModel", "class", "decoder", "def", "embed_tokens", "model", "self", "set_input_embeddings", "value"], "mbart/modeling_mbart.py:MBartForCausalLM.forward": ["Cache", "CausalLMOutputWithCrossAttentions", "CrossEntropyLoss", "FloatTensor", "GenerationMixin", "LongTensor", "Model", "ModelPreTrainedModel", "None", "Tensor", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "cross_attentions", "decoder", "def", "device", "else", "encoder_attention_mask", "encoder_hidden_states", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "lm_head", "logits", "logits_to_keep", "loss", "loss_fct", "model", "not", "output", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "r", "return", "return_dict", "self", "slice", "slice_indices", "to", "torch", "tuple", "use_cache", "use_return_dict", "view", "vocab_size"], "megatron_bert/modeling_megatron_bert.py:MegatronBertEmbeddings.__init__": ["Dropout", "Embedding", "False", "ModelEmbeddings", "Module", "__init__", "arange", "class", "config", "def", "dropout", "expand", "hidden_dropout_prob", "hidden_size", "max_position_embeddings", "nn", "pad_token_id", "padding_idx", "persistent", "position_embeddings", "position_ids", "register_buffer", "self", "super", "token_type_embeddings", "torch", "type_vocab_size", "vocab_size", "word_embeddings"], "megatron_bert/modeling_megatron_bert.py:MegatronBertEmbeddings.forward": ["LongTensor", "ModelEmbeddings", "Module", "None", "Tensor", "class", "def", "device", "dropout", "dtype", "else", "embeddings", "forward", "if", "input_ids", "input_shape", "inputs_embeds", "int", "is", "long", "nn", "not", "past_key_values_length", "position_embeddings", "position_ids", "return", "self", "seq_length", "size", "token_type_embeddings", "token_type_ids", "torch", "word_embeddings", "zeros"], "megatron_bert/modeling_megatron_bert.py:MegatronBertSelfAttention.__init__": ["Dropout", "Linear", "ModelSelfAttention", "Module", "None", "The", "ValueError", "__init__", "a", "all_head_size", "and", "attention", "attention_head_size", "attention_probs_dropout_prob", "class", "config", "def", "dropout", "embedding_size", "f", "hasattr", "heads", "hidden", "hidden_size", "if", "int", "is", "is_decoder", "key", "layer_idx", "multiple", "nn", "not", "num_attention_heads", "number", "of", "query", "raise", "self", "size", "super", "the", "value"], "megatron_bert/modeling_megatron_bert.py:MegatronBertSelfAttention.forward": ["Cache", "EncoderDecoderCache", "False", "FloatTensor", "ModelSelfAttention", "Module", "None", "Tensor", "True", "_", "all_head_size", "and", "attention_head_size", "attention_mask", "attention_probs", "attention_scores", "batch_size", "bool", "cache_position", "class", "context_layer", "contiguous", "cross_attention_cache", "curr_past_key_values", "current_states", "def", "dim", "dropout", "else", "encoder_hidden_states", "forward", "functional", "get", "hidden_states", "if", "is", "is_cross_attention", "is_updated", "isinstance", "key", "key_layer", "keys", "layer_idx", "layers", "math", "matmul", "new_context_layer_shape", "nn", "not", "num_attention_heads", "output_attentions", "past_key_values", "permute", "query", "query_layer", "return", "self", "self_attention_cache", "seq_length", "shape", "size", "softmax", "sqrt", "torch", "transpose", "tuple", "update", "value", "value_layer", "values", "view"], "megatron_bert/modeling_megatron_bert.py:MegatronBertSelfOutput.__init__": ["Dropout", "Linear", "ModelSelfOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "hidden_dropout_prob", "hidden_size", "nn", "self", "super"], "megatron_bert/modeling_megatron_bert.py:MegatronBertSelfOutput.forward": ["ModelSelfOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "nn", "residual", "return", "self", "torch"], "megatron_bert/modeling_megatron_bert.py:MegatronBertAttention.__init__": ["LayerNorm", "ModelAttention", "ModelSelfAttention", "ModelSelfOutput", "Module", "None", "__init__", "class", "config", "def", "eps", "hidden_size", "layer_idx", "layer_norm_eps", "ln", "nn", "output", "self", "super"], "megatron_bert/modeling_megatron_bert.py:MegatronBertAttention.forward": ["Cache", "False", "FloatTensor", "ModelAttention", "Module", "None", "Tensor", "attention_mask", "attention_output", "bool", "cache_position", "class", "def", "encoder_hidden_states", "forward", "hidden_states", "ln", "ln_outputs", "nn", "output", "output_attentions", "outputs", "past_key_values", "return", "self", "self_outputs", "torch", "tuple"], "megatron_bert/modeling_megatron_bert.py:MegatronBertIntermediate.__init__": ["ACT2FN", "Linear", "ModelIntermediate", "Module", "__init__", "class", "config", "def", "dense", "else", "hidden_act", "hidden_size", "if", "intermediate_act_fn", "intermediate_size", "isinstance", "nn", "self", "str", "super"], "megatron_bert/modeling_megatron_bert.py:MegatronBertIntermediate.forward": ["ModelIntermediate", "Module", "Tensor", "class", "def", "dense", "forward", "hidden_states", "intermediate_act_fn", "nn", "return", "self", "torch"], "megatron_bert/modeling_megatron_bert.py:MegatronBertOutput.__init__": ["Dropout", "Linear", "ModelOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "hidden_dropout_prob", "hidden_size", "intermediate_size", "nn", "self", "super"], "megatron_bert/modeling_megatron_bert.py:MegatronBertOutput.forward": ["ModelOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "megatron_bert/modeling_megatron_bert.py:MegatronBertLayer.__init__": ["GradientCheckpointingLayer", "LayerNorm", "ModelAttention", "ModelIntermediate", "ModelLayer", "ModelOutput", "None", "TypeError", "__init__", "a", "add_cross_attention", "added", "as", "attention", "be", "chunk_size_feed_forward", "class", "config", "cross", "crossattention", "decoder", "def", "eps", "f", "hidden_size", "if", "intermediate", "is", "is_decoder", "layer_idx", "layer_norm_eps", "ln", "model", "nn", "not", "output", "raise", "self", "seq_len_dim", "should", "super", "used"], "megatron_bert/modeling_megatron_bert.py:MegatronBertLayer.forward": ["AttributeError", "Cache", "False", "FloatTensor", "GradientCheckpointingLayer", "If", "ModelLayer", "None", "Tensor", "True", "add_cross_attention", "and", "apply_chunking_to_forward", "are", "attention", "attention_mask", "attention_output", "be", "bool", "by", "cache_position", "chunk_size_feed_forward", "class", "config", "cross", "cross_attention_outputs", "crossattention", "def", "encoder_attention_mask", "encoder_hidden_states", "f", "feed_forward_chunk", "forward", "has", "hasattr", "hidden_states", "if", "instantiated", "is", "is_decoder", "layer_output", "layers", "not", "output_attentions", "outputs", "passed", "past_key_values", "raise", "return", "self", "self_attention_outputs", "seq_len_dim", "setting", "to", "torch", "tuple", "with"], "megatron_bert/modeling_megatron_bert.py:MegatronBertLayer.feed_forward_chunk": ["GradientCheckpointingLayer", "ModelLayer", "attention_output", "class", "def", "feed_forward_chunk", "intermediate", "intermediate_output", "layer_output", "ln", "ln_output", "output", "return", "self"], "megatron_bert/modeling_megatron_bert.py:MegatronBertEncoder.__init__": ["False", "LayerNorm", "ModelEncoder", "ModelLayer", "Module", "ModuleList", "__init__", "class", "config", "def", "eps", "for", "gradient_checkpointing", "hidden_size", "i", "in", "layer", "layer_idx", "layer_norm_eps", "ln", "nn", "num_hidden_layers", "range", "self", "super"], "megatron_bert/modeling_megatron_bert.py:MegatronBertEncoder.forward": ["BaseModelOutputWithPastAndCrossAttentions", "Cache", "DynamicCache", "EncoderDecoderCache", "False", "FloatTensor", "ModelEncoder", "Module", "None", "Setting", "Tensor", "True", "add_cross_attention", "all_cross_attentions", "all_hidden_states", "all_self_attentions", "and", "attention_mask", "attentions", "bool", "cache_position", "checkpointing", "class", "config", "cross_attentions", "def", "else", "encoder_attention_mask", "encoder_hidden_states", "enumerate", "for", "forward", "gradient", "gradient_checkpointing", "hidden_states", "i", "if", "in", "incompatible", "is", "last_hidden_state", "layer", "layer_module", "layer_outputs", "ln", "logger", "nn", "not", "output_attentions", "output_hidden_states", "past_key_values", "return", "return_dict", "self", "torch", "training", "tuple", "use_cache", "v", "warning_once", "with"], "megatron_bert/modeling_megatron_bert.py:MegatronBertPooler.__init__": ["Linear", "ModelPooler", "Module", "Tanh", "__init__", "activation", "class", "config", "def", "dense", "hidden_size", "nn", "self", "super"], "megatron_bert/modeling_megatron_bert.py:MegatronBertPooler.forward": ["ModelPooler", "Module", "Tensor", "activation", "class", "def", "dense", "first_token_tensor", "forward", "hidden_states", "nn", "pooled_output", "return", "self", "torch"], "megatron_bert/modeling_megatron_bert.py:MegatronBertPredictionHeadTransform.__init__": ["ACT2FN", "LayerNorm", "Linear", "ModelPredictionHeadTransform", "Module", "__init__", "class", "config", "def", "dense", "else", "eps", "hidden_act", "hidden_size", "if", "isinstance", "layer_norm_eps", "nn", "self", "str", "super", "transform_act_fn"], "megatron_bert/modeling_megatron_bert.py:MegatronBertPredictionHeadTransform.forward": ["LayerNorm", "ModelPredictionHeadTransform", "Module", "Tensor", "class", "def", "dense", "forward", "hidden_states", "nn", "return", "self", "torch", "transform_act_fn"], "megatron_bert/modeling_megatron_bert.py:MegatronBertLMPredictionHead.__init__": ["Linear", "ModelLMPredictionHead", "ModelPredictionHeadTransform", "Module", "Parameter", "True", "__init__", "bias", "class", "config", "decoder", "def", "hidden_size", "nn", "self", "super", "torch", "transform", "vocab_size", "zeros"], "megatron_bert/modeling_megatron_bert.py:MegatronBertLMPredictionHead.forward": ["ModelLMPredictionHead", "Module", "class", "decoder", "def", "forward", "hidden_states", "nn", "return", "self", "transform"], "megatron_bert/modeling_megatron_bert.py:MegatronBertOnlyMLMHead.__init__": ["ModelLMPredictionHead", "ModelOnlyMLMHead", "Module", "__init__", "class", "config", "def", "nn", "predictions", "self", "super"], "megatron_bert/modeling_megatron_bert.py:MegatronBertOnlyMLMHead.forward": ["ModelOnlyMLMHead", "Module", "Tensor", "class", "def", "forward", "nn", "prediction_scores", "predictions", "return", "self", "sequence_output", "torch"], "megatron_bert/modeling_megatron_bert.py:MegatronBertOnlyNSPHead.__init__": ["Linear", "ModelOnlyNSPHead", "Module", "__init__", "class", "config", "def", "hidden_size", "nn", "self", "seq_relationship", "super"], "megatron_bert/modeling_megatron_bert.py:MegatronBertOnlyNSPHead.forward": ["ModelOnlyNSPHead", "Module", "class", "def", "forward", "nn", "pooled_output", "return", "self", "seq_relationship", "seq_relationship_score"], "megatron_bert/modeling_megatron_bert.py:MegatronBertPreTrainingHeads.__init__": ["Linear", "ModelLMPredictionHead", "ModelPreTrainingHeads", "Module", "__init__", "class", "config", "def", "hidden_size", "nn", "predictions", "self", "seq_relationship", "super"], "megatron_bert/modeling_megatron_bert.py:MegatronBertPreTrainingHeads.forward": ["ModelPreTrainingHeads", "Module", "class", "def", "forward", "nn", "pooled_output", "prediction_scores", "predictions", "return", "self", "seq_relationship", "seq_relationship_score", "sequence_output"], "megatron_bert/modeling_megatron_bert.py:MegatronBertPreTrainedModel._init_weights": ["ModelEmbeddings", "ModelLMPredictionHead", "ModelPreTrainedModel", "PreTrainedModel", "_init_weights", "arange", "bias", "class", "copy_", "def", "elif", "expand", "if", "init", "isinstance", "module", "position_ids", "self", "shape", "super", "torch", "zeros_"], "megatron_bert/modeling_megatron_bert.py:MegatronBertModel.__init__": ["ModelEmbeddings", "ModelEncoder", "ModelModel", "ModelPooler", "ModelPreTrainedModel", "None", "True", "__init__", "add_pooling_layer", "class", "config", "def", "else", "embeddings", "encoder", "if", "pooler", "post_init", "r", "self", "super"], "megatron_bert/modeling_megatron_bert.py:MegatronBertModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "return", "self", "word_embeddings"], "megatron_bert/modeling_megatron_bert.py:MegatronBertModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "self", "set_input_embeddings", "value", "word_embeddings"], "megatron_bert/modeling_megatron_bert.py:MegatronBertModel.forward": ["BaseModelOutputWithPoolingAndCrossAttentions", "Cache", "False", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "ValueError", "You", "_", "and", "at", "attention_mask", "attentions", "batch_size", "bool", "both", "cache_position", "cannot", "class", "config", "cross_attentions", "def", "device", "dtype", "either", "elif", "else", "embedding_output", "embeddings", "encoder", "encoder_attention_mask", "encoder_batch_size", "encoder_extended_attention_mask", "encoder_hidden_shape", "encoder_hidden_states", "encoder_outputs", "encoder_sequence_length", "extended_attention_mask", "forward", "get_extended_attention_mask", "get_seq_length", "have", "hidden_states", "if", "input_ids", "input_shape", "inputs_embeds", "invert_attention_mask", "is", "is_decoder", "kwargs", "last_hidden_state", "long", "not", "ones", "or", "output_attentions", "output_hidden_states", "past_key_values", "past_key_values_length", "pooled_output", "pooler", "pooler_output", "position_ids", "raise", "return", "return_dict", "same", "self", "seq_length", "sequence_output", "size", "specify", "the", "time", "to", "token_type_ids", "torch", "tuple", "use_cache", "use_return_dict", "warn_if_padding_and_no_attention_mask", "zeros"], "megatron_bert/modeling_megatron_bert.py:MegatronBertForPreTraining.__init__": ["ModelForPreTraining", "ModelModel", "ModelPreTrainedModel", "ModelPreTrainingHeads", "True", "__init__", "add_binary_head", "bert", "class", "cls", "config", "def", "post_init", "r", "self", "super"], "megatron_bert/modeling_megatron_bert.py:MegatronBertForPreTraining.get_output_embeddings": ["ModelForPreTraining", "ModelPreTrainedModel", "class", "cls", "decoder", "def", "get_output_embeddings", "predictions", "return", "self"], "megatron_bert/modeling_megatron_bert.py:MegatronBertForPreTraining.set_output_embeddings": ["ModelForPreTraining", "ModelPreTrainedModel", "bias", "class", "cls", "decoder", "def", "new_embeddings", "predictions", "self", "set_output_embeddings"], "megatron_bert/modeling_megatron_bert.py:MegatronBertForPreTraining.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "ModelForPreTraining", "ModelForPreTrainingOutput", "ModelPreTrainedModel", "None", "and", "attention_mask", "attentions", "bert", "bool", "class", "cls", "config", "def", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "loss", "loss_fct", "masked_lm_loss", "next_sentence_label", "next_sentence_loss", "not", "output", "output_attentions", "output_hidden_states", "outputs", "pooled_output", "position_ids", "prediction_logits", "prediction_scores", "r", "return", "return_dict", "self", "seq_relationship_logits", "seq_relationship_score", "sequence_output", "token_type_ids", "torch", "total_loss", "tuple", "use_return_dict", "view", "vocab_size"], "megatron_bert/modeling_megatron_bert.py:MegatronBertForCausalLM.__init__": ["False", "GenerationMixin", "If", "ModelForCausalLM", "ModelModel", "ModelOnlyMLMHead", "ModelPreTrainedModel", "True", "__init__", "a", "add", "add_pooling_layer", "as", "bert", "class", "cls", "config", "def", "if", "is_decoder", "logger", "not", "post_init", "self", "standalone", "super", "to", "use", "want", "warning", "you"], "megatron_bert/modeling_megatron_bert.py:MegatronBertForCausalLM.get_output_embeddings": ["GenerationMixin", "ModelForCausalLM", "ModelPreTrainedModel", "class", "cls", "decoder", "def", "get_output_embeddings", "predictions", "return", "self"], "megatron_bert/modeling_megatron_bert.py:MegatronBertForCausalLM.set_output_embeddings": ["GenerationMixin", "ModelForCausalLM", "ModelPreTrainedModel", "bias", "class", "cls", "decoder", "def", "new_embeddings", "predictions", "self", "set_output_embeddings"], "megatron_bert/modeling_megatron_bert.py:MegatronBertForCausalLM.forward": ["Cache", "CausalLMOutputWithCrossAttentions", "False", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForCausalLM", "ModelPreTrainedModel", "None", "Tensor", "attention_mask", "attentions", "bert", "bool", "cache_position", "class", "cls", "config", "cross_attentions", "def", "else", "encoder_attention_mask", "encoder_hidden_states", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "logits", "logits_to_keep", "loss", "loss_function", "not", "output", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "position_ids", "r", "return", "return_dict", "self", "slice", "slice_indices", "token_type_ids", "torch", "tuple", "use_cache", "use_return_dict", "vocab_size"], "megatron_bert/modeling_megatron_bert.py:MegatronBertForMaskedLM.__init__": ["False", "If", "ModelForMaskedLM", "ModelModel", "ModelOnlyMLMHead", "ModelPreTrainedModel", "__init__", "add_pooling_layer", "attention", "bert", "bi", "class", "cls", "config", "def", "directional", "for", "if", "is_decoder", "logger", "make", "post_init", "self", "super", "sure", "to", "use", "want", "warning", "you"], "megatron_bert/modeling_megatron_bert.py:MegatronBertForMaskedLM.get_output_embeddings": ["ModelForMaskedLM", "ModelPreTrainedModel", "class", "cls", "decoder", "def", "get_output_embeddings", "predictions", "return", "self"], "megatron_bert/modeling_megatron_bert.py:MegatronBertForMaskedLM.set_output_embeddings": ["ModelForMaskedLM", "ModelPreTrainedModel", "bias", "class", "cls", "decoder", "def", "new_embeddings", "predictions", "self", "set_output_embeddings"], "megatron_bert/modeling_megatron_bert.py:MegatronBertForMaskedLM.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "MaskedLMOutput", "ModelForMaskedLM", "ModelPreTrainedModel", "None", "attention_mask", "attentions", "bert", "bool", "class", "cls", "config", "def", "else", "encoder_attention_mask", "encoder_hidden_states", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "masked_lm_loss", "not", "output", "output_attentions", "output_hidden_states", "outputs", "position_ids", "prediction_scores", "r", "return", "return_dict", "self", "sequence_output", "token_type_ids", "torch", "tuple", "use_return_dict", "view", "vocab_size"], "megatron_bert/modeling_megatron_bert.py:MegatronBertForMaskedLM.prepare_inputs_for_generation": ["ModelForMaskedLM", "ModelPreTrainedModel", "None", "PAD", "The", "ValueError", "attention_mask", "be", "cat", "class", "config", "def", "defined", "device", "dim", "dtype", "dummy_token", "effective_batch_size", "for", "full", "generation", "if", "input_ids", "input_shape", "is", "long", "model_kwargs", "new_zeros", "pad_token_id", "prepare_inputs_for_generation", "raise", "return", "self", "shape", "should", "token", "torch"], "megatron_bert/modeling_megatron_bert.py:MegatronBertForNextSentencePrediction.__init__": ["ModelForNextSentencePrediction", "ModelModel", "ModelOnlyNSPHead", "ModelPreTrainedModel", "__init__", "bert", "class", "cls", "config", "def", "post_init", "self", "super"], "megatron_bert/modeling_megatron_bert.py:MegatronBertForNextSentencePrediction.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "ModelForNextSentencePrediction", "ModelPreTrainedModel", "NextSentencePredictorOutput", "None", "attention_mask", "attentions", "bert", "bool", "class", "cls", "config", "def", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "next_sentence_loss", "not", "output", "output_attentions", "output_hidden_states", "outputs", "pooled_output", "position_ids", "r", "return", "return_dict", "self", "seq_relationship_scores", "token_type_ids", "torch", "tuple", "use_return_dict", "view"], "megatron_bert/modeling_megatron_bert.py:MegatronBertForSequenceClassification.__init__": ["Dropout", "Linear", "ModelForSequenceClassification", "ModelModel", "ModelPreTrainedModel", "__init__", "bert", "class", "classifier", "config", "def", "dropout", "hidden_dropout_prob", "hidden_size", "nn", "num_labels", "post_init", "self", "super"], "megatron_bert/modeling_megatron_bert.py:MegatronBertForSequenceClassification.forward": ["BCEWithLogitsLoss", "CrossEntropyLoss", "FloatTensor", "LongTensor", "MSELoss", "ModelForSequenceClassification", "ModelPreTrainedModel", "None", "SequenceClassifierOutput", "and", "attention_mask", "attentions", "bert", "bool", "class", "classifier", "config", "def", "dropout", "dtype", "elif", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "kwargs", "labels", "logits", "long", "loss", "loss_fct", "multi_label_classification", "not", "num_labels", "or", "output", "output_attentions", "output_hidden_states", "outputs", "pooled_output", "position_ids", "problem_type", "r", "regression", "return", "return_dict", "self", "single_label_classification", "squeeze", "token_type_ids", "torch", "tuple", "use_return_dict", "view"], "megatron_bert/modeling_megatron_bert.py:MegatronBertForMultipleChoice.__init__": ["Dropout", "Linear", "ModelForMultipleChoice", "ModelModel", "ModelPreTrainedModel", "__init__", "bert", "class", "classifier", "config", "def", "dropout", "hidden_dropout_prob", "hidden_size", "nn", "post_init", "self", "super"], "megatron_bert/modeling_megatron_bert.py:MegatronBertForMultipleChoice.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "ModelForMultipleChoice", "ModelPreTrainedModel", "MultipleChoiceModelOutput", "None", "attention_mask", "attentions", "bert", "bool", "class", "classifier", "config", "def", "dropout", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "not", "num_choices", "output", "output_attentions", "output_hidden_states", "outputs", "pooled_output", "position_ids", "r", "reshaped_logits", "return", "return_dict", "self", "shape", "size", "token_type_ids", "torch", "tuple", "use_return_dict", "view"], "megatron_bert/modeling_megatron_bert.py:MegatronBertForTokenClassification.__init__": ["Dropout", "False", "Linear", "ModelForTokenClassification", "ModelModel", "ModelPreTrainedModel", "__init__", "add_pooling_layer", "bert", "class", "classifier", "config", "def", "dropout", "hidden_dropout_prob", "hidden_size", "nn", "num_labels", "post_init", "self", "super"], "megatron_bert/modeling_megatron_bert.py:MegatronBertForTokenClassification.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "ModelForTokenClassification", "ModelPreTrainedModel", "None", "TokenClassifierOutput", "attention_mask", "attentions", "bert", "bool", "class", "classifier", "config", "def", "dropout", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "not", "num_labels", "output", "output_attentions", "output_hidden_states", "outputs", "position_ids", "r", "return", "return_dict", "self", "sequence_output", "token_type_ids", "torch", "tuple", "use_return_dict", "view"], "megatron_bert/modeling_megatron_bert.py:MegatronBertForQuestionAnswering.__init__": ["False", "Linear", "ModelForQuestionAnswering", "ModelModel", "ModelPreTrainedModel", "__init__", "add_pooling_layer", "bert", "class", "config", "def", "hidden_size", "nn", "num_labels", "post_init", "qa_outputs", "self", "super"], "megatron_bert/modeling_megatron_bert.py:MegatronBertForQuestionAnswering.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "ModelForQuestionAnswering", "ModelPreTrainedModel", "None", "QuestionAnsweringModelOutput", "and", "attention_mask", "attentions", "bert", "bool", "clamp", "class", "config", "contiguous", "def", "dim", "else", "end_logits", "end_loss", "end_positions", "forward", "hidden_states", "if", "ignore_index", "ignored_index", "input_ids", "inputs_embeds", "is", "kwargs", "len", "logits", "loss", "loss_fct", "not", "output", "output_attentions", "output_hidden_states", "outputs", "position_ids", "qa_outputs", "return", "return_dict", "self", "sequence_output", "size", "split", "squeeze", "start_logits", "start_loss", "start_positions", "token_type_ids", "torch", "total_loss", "tuple", "use_return_dict"], "metaclip_2/modeling_metaclip_2.py:MetaClip2TextEmbeddings.__init__": ["Embedding", "False", "ModelTextConfig", "ModelTextEmbeddings", "Module", "__init__", "arange", "class", "config", "def", "embed_dim", "expand", "hidden_size", "max_position_embeddings", "nn", "persistent", "position_embedding", "position_ids", "register_buffer", "self", "super", "token_embedding", "torch", "vocab_size"], "metaclip_2/modeling_metaclip_2.py:MetaClip2TextEmbeddings.forward": ["FloatTensor", "LongTensor", "ModelTextEmbeddings", "Module", "None", "Sequence", "Tensor", "ValueError", "and", "be", "class", "def", "else", "embeddings", "f", "forward", "got", "if", "input_ids", "inputs_embeds", "is", "length", "less", "max_position_embedding", "max_position_embeddings", "must", "nn", "not", "position_embedding", "position_embeddings", "position_ids", "raise", "return", "self", "seq_length", "sequence", "shape", "than", "token_embedding", "torch", "weight"], "metaclip_2/modeling_metaclip_2.py:MetaClip2VisionEmbeddings.__init__": ["Conv2d", "Embedding", "False", "ModelVisionConfig", "ModelVisionEmbeddings", "Module", "Parameter", "__init__", "arange", "bias", "class", "class_embedding", "config", "def", "embed_dim", "expand", "hidden_size", "image_size", "in_channels", "kernel_size", "nn", "num_channels", "num_patches", "num_positions", "out_channels", "patch_embedding", "patch_size", "persistent", "position_embedding", "position_ids", "randn", "register_buffer", "self", "stride", "super", "torch"], "metaclip_2/modeling_metaclip_2.py:MetaClip2VisionEmbeddings.interpolate_pos_encoding": ["False", "ModelVisionEmbeddings", "Module", "Tensor", "align_corners", "and", "bicubic", "cat", "class", "class_pos_embed", "def", "dim", "embeddings", "functional", "height", "if", "int", "interpolate", "interpolate_pos_encoding", "is_tracing", "jit", "mode", "new_height", "new_width", "nn", "not", "num_patches", "num_positions", "patch_pos_embed", "patch_size", "permute", "position_embedding", "position_ids", "reshape", "return", "self", "shape", "size", "sqrt_num_positions", "torch", "torch_int", "unsqueeze", "view", "weight", "width"], "metaclip_2/modeling_metaclip_2.py:MetaClip2VisionEmbeddings.forward": ["False", "FloatTensor", "Input", "ModelVisionEmbeddings", "Module", "Tensor", "ValueError", "_", "and", "batch_size", "cat", "class", "class_embedding", "class_embeds", "def", "dim", "doesn", "dtype", "else", "embeddings", "expand", "f", "flatten", "forward", "height", "if", "image", "image_size", "interpolate_pos_encoding", "match", "model", "nn", "not", "or", "patch_embedding", "patch_embeds", "pixel_values", "position_embedding", "position_ids", "raise", "return", "self", "shape", "size", "t", "target_dtype", "to", "torch", "transpose", "weight", "width"], "metaclip_2/modeling_metaclip_2.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "softmax", "to", "torch", "training", "transpose", "value"], "metaclip_2/modeling_metaclip_2.py:MetaClip2Attention.__init__": ["False", "Linear", "ModelAttention", "ModelTextConfig", "ModelVisionConfig", "Module", "ValueError", "__init__", "and", "attention_dropout", "be", "by", "class", "config", "def", "divisible", "dropout", "embed_dim", "f", "got", "head_dim", "hidden_size", "if", "is_causal", "k_proj", "must", "nn", "num_attention_heads", "num_heads", "out_proj", "q_proj", "raise", "scale", "self", "super", "v_proj"], "metaclip_2/modeling_metaclip_2.py:MetaClip2Attention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "ModelAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_attn_implementation", "attention_interface", "attention_mask", "attn_output", "attn_weights", "batch_size", "class", "config", "contiguous", "def", "dropout", "eager", "eager_attention_forward", "else", "embed_dim", "forward", "head_dim", "hidden_states", "if", "k_proj", "keys", "kwargs", "nn", "not", "out_proj", "q_proj", "queries", "reshape", "return", "scale", "scaling", "self", "seq_length", "shape", "torch", "training", "transpose", "tuple", "v_proj", "values", "view"], "metaclip_2/modeling_metaclip_2.py:MetaClip2MLP.__init__": ["ACT2FN", "Linear", "ModelMLP", "Module", "__init__", "activation_fn", "class", "config", "def", "fc1", "fc2", "hidden_act", "hidden_size", "intermediate_size", "nn", "self", "super"], "metaclip_2/modeling_metaclip_2.py:MetaClip2MLP.forward": ["ModelMLP", "Module", "Tensor", "activation_fn", "class", "def", "fc1", "fc2", "forward", "hidden_states", "nn", "return", "self", "torch"], "metaclip_2/modeling_metaclip_2.py:MetaClip2EncoderLayer.__init__": ["GradientCheckpointingLayer", "LayerNorm", "ModelAttention", "ModelEncoderLayer", "ModelMLP", "ModelTextConfig", "ModelVisionConfig", "__init__", "class", "config", "def", "embed_dim", "eps", "hidden_size", "layer_norm1", "layer_norm2", "layer_norm_eps", "mlp", "nn", "self", "self_attn", "super"], "metaclip_2/modeling_metaclip_2.py:MetaClip2EncoderLayer.forward": ["FloatTensor", "GradientCheckpointingLayer", "ModelEncoderLayer", "Tensor", "TransformersKwargs", "Unpack", "_", "attention_mask", "class", "def", "forward", "hidden_states", "kwargs", "layer_norm1", "layer_norm2", "mlp", "residual", "return", "self", "self_attn", "torch"], "metaclip_2/modeling_metaclip_2.py:MetaClip2PreTrainedModel._init_weights": ["LayerNorm", "Linear", "ModelAttention", "ModelForImageClassification", "ModelMLP", "ModelModel", "ModelPreTrainedModel", "ModelTextEmbeddings", "ModelTextModelWithProjection", "ModelVisionEmbeddings", "ModelVisionModelWithProjection", "None", "PreTrainedModel", "_init_weights", "and", "arange", "bias", "class", "class_embedding", "classifier", "config", "copy_", "def", "elif", "embed_dim", "expand", "factor", "fc1", "fc2", "fc_std", "hidden_size", "if", "in_proj_std", "init", "initializer_factor", "initializer_range", "is", "isinstance", "k_proj", "mean", "module", "nn", "normal_", "not", "num_hidden_layers", "ones_", "out_proj", "out_proj_std", "patch_embedding", "position_embedding", "position_ids", "q_proj", "self", "shape", "std", "text_embed_dim", "text_projection", "token_embedding", "torch", "v_proj", "vision_config", "vision_embed_dim", "visual_projection", "weight", "zeros_"], "metaclip_2/modeling_metaclip_2.py:MetaClip2Encoder.__init__": ["False", "ModelConfig", "ModelEncoder", "ModelEncoderLayer", "Module", "ModuleList", "_", "__init__", "class", "config", "def", "for", "gradient_checkpointing", "in", "layers", "nn", "num_hidden_layers", "range", "self", "super"], "metaclip_2/modeling_metaclip_2.py:MetaClip2Encoder.forward": ["BaseModelOutput", "ModelEncoder", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "class", "def", "encoder_layer", "for", "forward", "hidden_states", "in", "inputs_embeds", "kwargs", "last_hidden_state", "layers", "nn", "r", "return", "self", "torch"], "metaclip_2/modeling_metaclip_2.py:MetaClip2TextTransformer.__init__": ["LayerNorm", "ModelEncoder", "ModelTextConfig", "ModelTextEmbeddings", "ModelTextTransformer", "Module", "__init__", "class", "config", "def", "embed_dim", "embeddings", "encoder", "eos_token_id", "eps", "final_layer_norm", "hidden_size", "layer_norm_eps", "nn", "self", "super"], "metaclip_2/modeling_metaclip_2.py:MetaClip2TextTransformer.forward": ["BaseModelOutput", "BaseModelOutputWithPooling", "ModelTextTransformer", "Module", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "arange", "argmax", "attention_mask", "attentions", "cache_position", "class", "config", "create_causal_mask", "def", "device", "dim", "dtype", "embeddings", "encoder", "encoder_outputs", "eos_token_id", "final_layer_norm", "forward", "hidden_states", "input_embeds", "input_ids", "input_shape", "inputs_embeds", "int", "is_causal", "kwargs", "last_hidden_state", "nn", "past_key_values", "pooled_output", "pooler_output", "pop", "position_ids", "return", "self", "shape", "size", "to", "torch", "view"], "metaclip_2/modeling_metaclip_2.py:MetaClip2TextModel.__init__": ["ModelPreTrainedModel", "ModelTextConfig", "ModelTextModel", "ModelTextTransformer", "__init__", "class", "config", "def", "post_init", "self", "super", "text_model"], "metaclip_2/modeling_metaclip_2.py:MetaClip2TextModel.get_input_embeddings": ["ModelPreTrainedModel", "ModelTextModel", "Module", "class", "def", "embeddings", "get_input_embeddings", "nn", "return", "self", "text_model", "token_embedding"], "metaclip_2/modeling_metaclip_2.py:MetaClip2TextModel.set_input_embeddings": ["ModelPreTrainedModel", "ModelTextModel", "class", "def", "embeddings", "self", "set_input_embeddings", "text_model", "token_embedding", "value"], "metaclip_2/modeling_metaclip_2.py:MetaClip2TextModel.forward": ["BaseModelOutputWithPooling", "ModelPreTrainedModel", "ModelTextModel", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "class", "def", "forward", "input_ids", "kwargs", "position_ids", "r", "return", "self", "text_model", "torch"], "metaclip_2/modeling_metaclip_2.py:MetaClip2TextModelWithProjection.__init__": ["False", "Linear", "ModelPreTrainedModel", "ModelTextConfig", "ModelTextModel", "ModelTextModelWithProjection", "__init__", "_from_config", "bias", "class", "config", "def", "hidden_size", "nn", "post_init", "projection_dim", "self", "super", "text_model", "text_projection"], "metaclip_2/modeling_metaclip_2.py:MetaClip2TextModelWithProjection.get_input_embeddings": ["ModelPreTrainedModel", "ModelTextModelWithProjection", "Module", "class", "def", "embeddings", "get_input_embeddings", "nn", "return", "self", "text_model", "token_embedding"], "metaclip_2/modeling_metaclip_2.py:MetaClip2TextModelWithProjection.set_input_embeddings": ["ModelPreTrainedModel", "ModelTextModelWithProjection", "class", "def", "embeddings", "self", "set_input_embeddings", "text_model", "token_embedding", "value"], "metaclip_2/modeling_metaclip_2.py:MetaClip2TextModelWithProjection.forward": ["BaseModelOutputWithPooling", "ModelPreTrainedModel", "ModelTextModelOutput", "ModelTextModelWithProjection", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "class", "def", "forward", "input_ids", "kwargs", "last_hidden_state", "pooled_output", "pooler_output", "position_ids", "r", "return", "self", "text_embeds", "text_model", "text_outputs", "text_projection", "torch"], "metaclip_2/modeling_metaclip_2.py:MetaClip2Output.to_tuple": ["Any", "ModelOutput", "class", "def", "else", "for", "getattr", "if", "in", "k", "keys", "not", "return", "self", "text_model_output", "to_tuple", "tuple", "vision_model_output"], "metaclip_2/modeling_metaclip_2.py:contrastive_loss": ["Model_loss", "Tensor", "arange", "cross_entropy", "def", "device", "functional", "len", "logits", "nn", "return", "torch"], "metaclip_2/modeling_metaclip_2.py:metaclip_2_loss": ["Model_loss", "Tensor", "caption_loss", "contrastive_loss", "def", "image_loss", "return", "similarity", "t", "torch"], "metaclip_2/modeling_metaclip_2.py:_get_vector_norm": ["Tensor", "True", "_get_vector_norm", "def", "dim", "keepdim", "normed_tensor", "pow", "return", "square_tensor", "sum", "sum_tensor", "tensor", "torch"], "metaclip_2/modeling_metaclip_2.py:MetaClip2Model.__init__": ["False", "Linear", "ModelConfig", "ModelModel", "ModelPreTrainedModel", "ModelTextConfig", "ModelTextModel", "ModelVisionConfig", "ModelVisionModel", "Parameter", "TypeError", "__init__", "_from_config", "be", "bias", "but", "class", "config", "def", "expected", "f", "hidden_size", "if", "is", "isinstance", "logit_scale", "logit_scale_init_value", "nn", "not", "of", "post_init", "projection_dim", "raise", "self", "super", "tensor", "text_config", "text_embed_dim", "text_model", "text_projection", "to", "torch", "type", "vision_config", "vision_embed_dim", "vision_model", "visual_projection"], "metaclip_2/modeling_metaclip_2.py:MetaClip2Model.get_text_features": ["BaseModelOutputWithPooling", "FloatTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "attention_mask", "class", "def", "get_text_features", "input_ids", "pooled_output", "pooler_output", "position_ids", "r", "return", "self", "text_features", "text_model", "text_outputs", "text_projection", "torch"], "metaclip_2/modeling_metaclip_2.py:MetaClip2Model.get_image_features": ["BaseModelOutputWithPooling", "False", "FloatTensor", "ModelModel", "ModelPreTrainedModel", "None", "bool", "class", "def", "get_image_features", "image_features", "interpolate_pos_encoding", "pixel_values", "pooled_output", "pooler_output", "r", "return", "self", "torch", "vision_model", "vision_outputs", "visual_projection"], "metaclip_2/modeling_metaclip_2.py:MetaClip2Model.forward": ["BaseModelOutputWithPooling", "False", "FloatTensor", "LongTensor", "ModelModel", "ModelOutput", "ModelPreTrainedModel", "Model_loss", "None", "Tensor", "TransformersKwargs", "Unpack", "_get_vector_norm", "attention_mask", "bool", "class", "def", "device", "exp", "forward", "if", "image_embeds", "input_ids", "interpolate_pos_encoding", "kwargs", "logit_scale", "logits_per_image", "logits_per_text", "loss", "matmul", "pixel_values", "pooler_output", "position_ids", "r", "return", "return_loss", "self", "t", "text_embeds", "text_model", "text_model_output", "text_outputs", "text_projection", "to", "torch", "vision_model", "vision_model_output", "vision_outputs", "visual_projection"], "metaclip_2/modeling_metaclip_2.py:MetaClip2VisionTransformer.__init__": ["LayerNorm", "ModelEncoder", "ModelVisionConfig", "ModelVisionEmbeddings", "ModelVisionTransformer", "Module", "__init__", "class", "config", "def", "embed_dim", "embeddings", "encoder", "eps", "hidden_size", "layer_norm_eps", "nn", "post_layernorm", "pre_layrnorm", "self", "super"], "metaclip_2/modeling_metaclip_2.py:MetaClip2VisionTransformer.forward": ["BaseModelOutput", "BaseModelOutputWithPooling", "False", "FloatTensor", "ModelVisionTransformer", "Module", "None", "TransformersKwargs", "Unpack", "ValueError", "You", "bool", "class", "def", "embeddings", "encoder", "encoder_outputs", "forward", "have", "hidden_states", "if", "inputs_embeds", "interpolate_pos_encoding", "is", "kwargs", "last_hidden_state", "nn", "pixel_values", "pooled_output", "pooler_output", "post_layernorm", "pre_layrnorm", "raise", "return", "self", "specify", "to", "torch"], "metaclip_2/modeling_metaclip_2.py:MetaClip2VisionModel.__init__": ["ModelPreTrainedModel", "ModelVisionConfig", "ModelVisionModel", "ModelVisionTransformer", "__init__", "class", "config", "def", "post_init", "self", "super", "vision_model"], "metaclip_2/modeling_metaclip_2.py:MetaClip2VisionModel.get_input_embeddings": ["ModelPreTrainedModel", "ModelVisionModel", "Module", "class", "def", "embeddings", "get_input_embeddings", "nn", "patch_embedding", "return", "self", "vision_model"], "metaclip_2/modeling_metaclip_2.py:MetaClip2VisionModel.forward": ["BaseModelOutputWithPooling", "False", "FloatTensor", "ModelPreTrainedModel", "ModelVisionModel", "None", "TransformersKwargs", "Unpack", "bool", "class", "def", "forward", "interpolate_pos_encoding", "kwargs", "pixel_values", "r", "return", "self", "torch", "vision_model"], "metaclip_2/modeling_metaclip_2.py:MetaClip2VisionModelWithProjection.__init__": ["False", "Linear", "ModelPreTrainedModel", "ModelVisionConfig", "ModelVisionModel", "ModelVisionModelWithProjection", "__init__", "_from_config", "bias", "class", "config", "def", "hidden_size", "nn", "post_init", "projection_dim", "self", "super", "vision_model", "visual_projection"], "metaclip_2/modeling_metaclip_2.py:MetaClip2VisionModelWithProjection.get_input_embeddings": ["ModelPreTrainedModel", "ModelVisionModelWithProjection", "Module", "class", "def", "embeddings", "get_input_embeddings", "nn", "patch_embedding", "return", "self", "vision_model"], "metaclip_2/modeling_metaclip_2.py:MetaClip2VisionModelWithProjection.forward": ["BaseModelOutputWithPooling", "False", "FloatTensor", "ModelPreTrainedModel", "ModelVisionModelOutput", "ModelVisionModelWithProjection", "None", "TransformersKwargs", "Unpack", "bool", "class", "def", "forward", "image_embeds", "interpolate_pos_encoding", "kwargs", "last_hidden_state", "pixel_values", "pooled_output", "pooler_output", "r", "return", "self", "torch", "vision_model", "vision_outputs", "visual_projection"], "metaclip_2/modeling_metaclip_2.py:MetaClip2ForImageClassification.__init__": ["Identity", "Linear", "ModelConfig", "ModelForImageClassification", "ModelPreTrainedModel", "ModelVisionModel", "None", "__init__", "_from_config", "class", "classifier", "config", "def", "else", "hidden_size", "if", "nn", "num_labels", "post_init", "self", "super", "vision_config", "vision_model"], "metaclip_2/modeling_metaclip_2.py:MetaClip2ForImageClassification.forward": ["BaseModelOutputWithPooling", "ImageClassifierOutput", "ModelForImageClassification", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "class", "classifier", "config", "def", "dim", "forward", "if", "is", "kwargs", "labels", "last_hidden_state", "logits", "loss", "loss_function", "mean", "not", "outputs", "pixel_values", "r", "return", "self", "sequence_output", "torch", "vision_model"], "mgp_str/modeling_mgp_str.py:drop_path": ["False", "Model_path", "Model_prob", "Tensor", "bool", "def", "device", "div", "dtype", "float", "floor_", "if", "input", "keep_prob", "ndim", "not", "or", "output", "rand", "random_tensor", "return", "shape", "torch", "training"], "mgp_str/modeling_mgp_str.py:MgpstrDropPath.__init__": ["ModelDropPath", "Module", "None", "__init__", "class", "def", "drop_prob", "float", "nn", "self", "super"], "mgp_str/modeling_mgp_str.py:MgpstrDropPath.forward": ["ModelDropPath", "Module", "Tensor", "class", "def", "drop_path", "drop_prob", "forward", "hidden_states", "nn", "return", "self", "torch", "training"], "mgp_str/modeling_mgp_str.py:MgpstrDropPath.extra_repr": ["ModelDropPath", "Module", "class", "def", "drop_prob", "extra_repr", "f", "nn", "p", "return", "self", "str"], "mgp_str/modeling_mgp_str.py:MgpstrEmbeddings.__init__": ["Conv2d", "Dropout", "Iterable", "ModelConfig", "ModelEmbeddings", "Module", "Parameter", "__init__", "abc", "class", "cls_token", "collections", "config", "def", "distilled", "drop_rate", "else", "grid_size", "hidden_size", "if", "image_size", "isinstance", "kernel_size", "nn", "num_channels", "num_patches", "num_tokens", "p", "patch_size", "pos_drop", "pos_embed", "proj", "self", "stride", "super", "torch", "zeros"], "mgp_str/modeling_mgp_str.py:MgpstrEmbeddings.forward": ["Input", "ModelEmbeddings", "Module", "ValueError", "batch_size", "cat", "channel", "class", "cls_token", "cls_tokens", "def", "dim", "doesn", "embedding_output", "expand", "f", "flatten", "forward", "height", "if", "image", "image_size", "match", "model", "nn", "or", "patch_embeddings", "pixel_values", "pos_drop", "pos_embed", "proj", "raise", "return", "self", "shape", "size", "t", "torch", "transpose", "width"], "mgp_str/modeling_mgp_str.py:MgpstrMlp.__init__": ["Dropout", "GELU", "Linear", "ModelConfig", "ModelMlp", "Module", "__init__", "act", "class", "config", "def", "drop", "drop_rate", "fc1", "fc2", "hidden_features", "hidden_size", "nn", "or", "self", "super"], "mgp_str/modeling_mgp_str.py:MgpstrMlp.forward": ["ModelMlp", "Module", "act", "class", "def", "drop", "fc1", "fc2", "forward", "hidden_states", "nn", "return", "self"], "mgp_str/modeling_mgp_str.py:MgpstrAttention.__init__": ["Dropout", "Linear", "ModelAttention", "ModelConfig", "Module", "__init__", "attn_drop", "attn_drop_rate", "bias", "class", "config", "def", "drop_rate", "head_dim", "hidden_size", "nn", "num_attention_heads", "num_heads", "proj", "proj_drop", "qkv", "qkv_bias", "scale", "self", "super"], "mgp_str/modeling_mgp_str.py:MgpstrAttention.forward": ["ModelAttention", "Module", "attention_probs", "attn_drop", "batch_size", "channel", "class", "context_layer", "def", "dim", "forward", "hidden_states", "key", "nn", "num", "num_heads", "permute", "proj", "proj_drop", "qkv", "query", "reshape", "return", "scale", "self", "shape", "softmax", "transpose", "value"], "mgp_str/modeling_mgp_str.py:MgpstrLayer.__init__": ["Identity", "LayerNorm", "ModelAttention", "ModelConfig", "ModelDropPath", "ModelLayer", "ModelMlp", "Module", "None", "__init__", "attn", "class", "config", "def", "drop_path", "else", "eps", "hidden_size", "if", "int", "is", "layer_norm_eps", "mlp", "mlp_hidden_dim", "mlp_ratio", "nn", "norm1", "norm2", "not", "self", "super"], "mgp_str/modeling_mgp_str.py:MgpstrLayer.forward": ["ModelLayer", "Module", "attention_output", "attn", "class", "def", "drop_path", "forward", "hidden_states", "layer_output", "mlp", "nn", "norm1", "norm2", "outputs", "return", "self", "self_attention_outputs"], "mgp_str/modeling_mgp_str.py:MgpstrEncoder.__init__": ["ModelConfig", "ModelEncoder", "ModelLayer", "Module", "Sequential", "__init__", "blocks", "class", "config", "cpu", "def", "device", "dpr", "drop_path", "drop_path_rate", "for", "i", "in", "item", "linspace", "nn", "num_hidden_layers", "range", "self", "super", "torch", "x"], "mgp_str/modeling_mgp_str.py:MgpstrEncoder.forward": ["BaseModelOutput", "False", "ModelEncoder", "Module", "None", "True", "_", "all_hidden_states", "all_self_attentions", "attentions", "blk", "blocks", "class", "def", "else", "enumerate", "for", "forward", "hidden_states", "if", "in", "is", "last_hidden_state", "layer_outputs", "nn", "not", "output_attentions", "output_hidden_states", "return", "return_dict", "self", "tuple", "v"], "mgp_str/modeling_mgp_str.py:MgpstrA3Module.__init__": ["Conv2d", "False", "LayerNorm", "ModelA3Module", "ModelConfig", "Module", "Sequential", "__init__", "bias", "class", "config", "def", "eps", "feat", "groups", "hidden_size", "kernel_size", "layer_norm_eps", "max_token_length", "nn", "norm", "self", "stride", "super", "tokenLearner", "token_norm"], "mgp_str/modeling_mgp_str.py:MgpstrA3Module.forward": ["F", "ModelA3Module", "Module", "a3_out", "attentions", "class", "def", "dim", "einsum", "feat", "flatten", "forward", "hidden_states", "id", "nn", "norm", "return", "sd", "selected", "self", "si", "softmax", "tokenLearner", "token_norm", "torch", "transpose", "unsqueeze"], "mgp_str/modeling_mgp_str.py:MgpstrPreTrainedModel._init_weights": ["Conv2d", "LayerNorm", "Linear", "ModelEmbeddings", "ModelPreTrainedModel", "Module", "None", "PreTrainedModel", "_init_weights", "bias", "class", "cls_token", "config", "def", "elif", "if", "init", "initializer_range", "is", "isinstance", "mean", "module", "nn", "not", "ones_", "pos_embed", "self", "std", "trunc_normal_", "weight", "zeros_"], "mgp_str/modeling_mgp_str.py:MgpstrModel.__init__": ["ModelConfig", "ModelEmbeddings", "ModelEncoder", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "config", "def", "embeddings", "encoder", "post_init", "self", "super"], "mgp_str/modeling_mgp_str.py:MgpstrModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "Module", "class", "def", "embeddings", "get_input_embeddings", "nn", "proj", "return", "self"], "mgp_str/modeling_mgp_str.py:MgpstrModel.forward": ["BaseModelOutput", "FloatTensor", "ModelModel", "ModelPreTrainedModel", "None", "ValueError", "You", "attentions", "bool", "class", "config", "def", "else", "embedding_output", "embeddings", "encoder", "encoder_outputs", "forward", "have", "hidden_states", "if", "is", "kwargs", "last_hidden_state", "not", "output_attentions", "output_hidden_states", "pixel_values", "raise", "return", "return_dict", "self", "specify", "to", "torch", "tuple", "use_return_dict"], "mgp_str/modeling_mgp_str.py:MgpstrForSceneTextRecognition.__init__": ["Linear", "Model", "ModelA3Module", "ModelConfig", "ModelForSceneTextRecognition", "ModelModel", "ModelPreTrainedModel", "None", "__init__", "bpe_a3_module", "bpe_head", "char_a3_module", "char_head", "class", "config", "def", "hidden_size", "nn", "num_bpe_labels", "num_character_labels", "num_labels", "num_wordpiece_labels", "post_init", "self", "super", "wp_a3_module", "wp_head"], "mgp_str/modeling_mgp_str.py:MgpstrForSceneTextRecognition.forward": ["FloatTensor", "Model", "ModelForSceneTextRecognition", "ModelModelOutput", "ModelPreTrainedModel", "None", "a3_attentions", "all_a3_attentions", "all_logits", "attentions", "bool", "bpe_a3_module", "bpe_a3_out", "bpe_attention", "bpe_head", "bpe_logits", "char_a3_module", "char_a3_out", "char_attention", "char_head", "char_logits", "class", "config", "def", "else", "for", "forward", "hidden_states", "if", "in", "is", "kwargs", "logits", "mgp_outputs", "not", "output", "output_a3_attentions", "output_attentions", "output_hidden_states", "outputs", "pixel_values", "r", "return", "return_dict", "self", "sequence_output", "torch", "tuple", "use_return_dict", "wp_a3_module", "wp_a3_out", "wp_attention", "wp_head", "wp_logits"], "mimi/modeling_mimi.py:MimiConv1dPaddingCache.__init__": ["Expected", "ModelConv1dPaddingCache", "None", "NotImplementedError", "True", "ValueError", "__init__", "all", "and", "class", "constant", "convolutions", "def", "elif", "f", "for", "from_args_num_layers", "if", "in", "int", "is", "len", "list", "mode", "not", "num_layers", "or", "other", "padding", "padding_cache", "per_layer_in_channels", "per_layer_is_init", "per_layer_padding", "per_layer_padding_mode", "pop", "raise", "replicate", "self", "str", "supported", "than", "using", "values"], "mimi/modeling_mimi.py:MimiConv1dPaddingCache.update": ["ModelConv1dPaddingCache", "None", "Tensor", "batch_size", "class", "constant", "current_cache", "def", "device", "dtype", "elif", "else", "empty", "hidden_states", "if", "in_channels", "int", "is", "layer_idx", "ones", "padding", "padding_cache", "padding_mode", "padding_states", "per_layer_in_channels", "per_layer_padding", "per_layer_padding_mode", "replicate", "return", "self", "shape", "torch", "update", "zeros"], "mimi/modeling_mimi.py:MimiConv1d.__init__": ["Conv1d", "False", "ModelConv1d", "Module", "None", "True", "__init__", "and", "been", "bias", "bool", "causal", "class", "config", "conv", "def", "dilation", "dtype", "else", "f", "groups", "has", "if", "in_channels", "initialized", "int", "int64", "is", "kernel_size", "layer_idx", "logger", "nn", "out_channels", "pad_mode", "padding_left", "padding_right", "padding_total", "persistent", "register_buffer", "self", "str", "stride", "super", "tensor", "torch", "use_causal_conv", "warning", "with"], "mimi/modeling_mimi.py:MimiConv1d.apply_weight_norm": ["ModelConv1d", "Module", "apply_weight_norm", "class", "conv", "def", "hasattr", "if", "nn", "parametrizations", "self", "utils", "weight_norm"], "mimi/modeling_mimi.py:MimiConv1d.remove_weight_norm": ["ModelConv1d", "Module", "class", "conv", "def", "nn", "remove_weight_norm", "self", "utils"], "mimi/modeling_mimi.py:MimiConv1d._get_extra_padding_for_conv1d": ["ModelConv1d", "Module", "Tensor", "_get_extra_padding_for_conv1d", "ceil", "class", "def", "hidden_states", "ideal_length", "int64", "kernel_size", "length", "n_frames", "nn", "padding_total", "return", "self", "shape", "stride", "to", "torch"], "mimi/modeling_mimi.py:MimiConv1d._pad1d": ["ModelConv1d", "Module", "Tensor", "_pad1d", "class", "def", "end", "extra_pad", "float", "functional", "hidden_states", "if", "int", "length", "max", "max_pad", "mode", "nn", "pad", "padded", "padding_left", "padding_right", "paddings", "reflect", "return", "shape", "str", "torch", "tuple", "value", "zero"], "mimi/modeling_mimi.py:MimiConv1d._get_output_length": ["LongTensor", "ModelConv1d", "Module", "_get_output_length", "causal", "ceil", "class", "conv", "def", "dilation", "else", "extra_padding", "ideal_length", "if", "input_length", "int64", "kernel_size", "n_frames", "nn", "output_length", "padding", "padding_left", "padding_right", "padding_total", "return", "self", "stride", "to", "torch"], "mimi/modeling_mimi.py:MimiConv1d.forward": ["ModelConv1d", "Module", "None", "ValueError", "_get_extra_padding_for_conv1d", "_pad1d", "and", "cat", "causal", "class", "conv", "convolutions", "def", "dim", "elif", "else", "extra_padding", "for", "forward", "hidden_states", "if", "is", "layer_idx", "layer_padding_cache", "mode", "nn", "non", "not", "pad_mode", "padding_cache", "padding_left", "padding_right", "padding_total", "raise", "return", "self", "supported", "torch", "update"], "mimi/modeling_mimi.py:MimiConvTranspose1d.__init__": ["ConvTranspose1d", "ModelConvTranspose1d", "Module", "True", "ValueError", "__init__", "bias", "causal", "ceil", "class", "config", "conv", "convolutions", "def", "else", "for", "groups", "if", "in_channels", "int", "kernel_size", "makes", "math", "nn", "not", "only", "or", "out_channels", "padding_left", "padding_right", "padding_total", "raise", "self", "sense", "stride", "super", "trim_right_ratio", "use_causal_conv"], "mimi/modeling_mimi.py:MimiConvTranspose1d.apply_weight_norm": ["ModelConvTranspose1d", "Module", "apply_weight_norm", "class", "conv", "def", "hasattr", "if", "nn", "parametrizations", "self", "utils", "weight_norm"], "mimi/modeling_mimi.py:MimiConvTranspose1d.remove_weight_norm": ["ModelConvTranspose1d", "Module", "class", "conv", "def", "nn", "remove_weight_norm", "self", "utils"], "mimi/modeling_mimi.py:MimiConvTranspose1d.forward": ["ModelConvTranspose1d", "Module", "class", "conv", "def", "end", "forward", "hidden_states", "nn", "padding_left", "padding_right", "return", "self", "shape"], "mimi/modeling_mimi.py:MimiResnetBlock.__init__": ["ELU", "Identity", "ModelConfig", "ModelConv1d", "ModelResnetBlock", "Module", "ModuleList", "Number", "ValueError", "__init__", "block", "class", "compress", "config", "def", "dilation", "dilations", "dim", "else", "enumerate", "for", "hidden", "i", "if", "in", "in_chs", "int", "kernel", "kernel_size", "kernel_sizes", "len", "list", "match", "nn", "number", "of", "out_chs", "raise", "residual_kernel_size", "self", "shortcut", "should", "sizes", "super", "use_conv_shortcut", "zip"], "mimi/modeling_mimi.py:MimiResnetBlock.forward": ["ModelConv1d", "ModelResnetBlock", "Module", "None", "block", "class", "def", "else", "for", "forward", "hidden_states", "if", "in", "isinstance", "layer", "nn", "padding_cache", "residual", "return", "self", "shortcut"], "mimi/modeling_mimi.py:MimiEncoder.__init__": ["ELU", "ModelConfig", "ModelConv1d", "ModelEncoder", "ModelResnetBlock", "Modelconv1d_layer_names", "Module", "ModuleList", "_Modelconv1d_layer_names", "__init__", "append", "audio_channels", "block", "class", "config", "conv_layer", "current_scale", "def", "dilation_growth_rate", "enumerate", "extend", "f", "for", "get_submodule", "hidden_size", "in", "j", "kernel_size", "last_kernel_size", "layer_idx", "layername", "layers", "len", "model", "nn", "num_filters", "num_residual_layers", "range", "ratio", "reversed", "scaling", "self", "setattr", "stride", "super", "upsampling_ratios"], "mimi/modeling_mimi.py:MimiEncoder.forward": ["ModelConv1d", "ModelEncoder", "ModelResnetBlock", "Module", "None", "class", "def", "else", "for", "forward", "hidden_states", "if", "in", "isinstance", "layer", "layers", "nn", "padding_cache", "return", "self"], "mimi/modeling_mimi.py:MimiLayerScale.__init__": ["ModelLayerScale", "Module", "Parameter", "True", "__init__", "channels", "class", "config", "def", "full", "hidden_size", "initial_scale", "layer_scale_initial_scale", "nn", "requires_grad", "scale", "self", "super", "torch"], "mimi/modeling_mimi.py:MimiLayerScale.forward": ["ModelLayerScale", "Module", "Tensor", "class", "def", "forward", "nn", "return", "scale", "self", "torch", "x"], "mimi/modeling_mimi.py:MimiRotaryEmbedding.__init__": ["Callable", "False", "ModelConfig", "ModelRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "mimi/modeling_mimi.py:MimiRotaryEmbedding.compute_default_rope_parameters": ["ModelConfig", "ModelRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "mimi/modeling_mimi.py:MimiRotaryEmbedding.forward": ["False", "ModelRotaryEmbedding", "Module", "None", "and", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "mimi/modeling_mimi.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "mimi/modeling_mimi.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cos", "def", "k", "k_embed", "q", "q_embed", "return", "rotate_half", "sin", "unsqueeze", "unsqueeze_dim"], "mimi/modeling_mimi.py:MimiMLP.__init__": ["ACT2FN", "False", "Linear", "ModelMLP", "Module", "__init__", "activation_fn", "bias", "class", "config", "def", "fc1", "fc2", "hidden_act", "hidden_size", "intermediate_size", "nn", "self", "super"], "mimi/modeling_mimi.py:MimiMLP.forward": ["ModelMLP", "Module", "Tensor", "activation_fn", "class", "def", "fc1", "fc2", "forward", "hidden_states", "nn", "return", "self", "torch"], "mimi/modeling_mimi.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "mimi/modeling_mimi.py:MimiAttention.__init__": ["Instantiating", "Linear", "ModelAttention", "ModelConfig", "ModelRotaryEmbedding", "Module", "None", "Please", "True", "ValueError", "__class__", "__init__", "__name__", "a", "and", "attention_bias", "attention_dropout", "be", "bias", "by", "caching", "call", "class", "config", "creating", "def", "divisible", "during", "errors", "f", "forward", "got", "head_dim", "hidden_size", "if", "int", "is", "is_causal", "k_proj", "layer_idx", "lead", "logger", "make", "math", "max_position_embeddings", "must", "nn", "not", "num_attention_heads", "num_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "passing", "provide", "q_proj", "raise", "recommended", "rotary_emb", "scaling", "self", "sliding_window", "sqrt", "super", "sure", "the", "this", "to", "used", "v_proj", "warning_once", "when", "will", "without"], "mimi/modeling_mimi.py:MimiAttention.forward": ["Cache", "False", "LongTensor", "ModelAttention", "Module", "None", "Tensor", "ValueError", "_", "apply_rotary_pos_emb", "attention_dropout", "attention_mask", "attn_output", "attn_weights", "be", "bool", "bsz", "but", "cache_kwargs", "cache_position", "causal_mask", "class", "contiguous", "cos", "def", "dim", "dropout", "dtype", "f", "float32", "forward", "functional", "head_dim", "hidden_states", "if", "is", "k_proj", "key_states", "layer_idx", "matmul", "nn", "not", "num_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "of", "output_attentions", "p", "past_key_values", "position_ids", "q_len", "q_proj", "query_states", "raise", "repeat_kv", "return", "rotary_emb", "scaling", "self", "shape", "should", "sin", "size", "softmax", "to", "torch", "training", "transpose", "tuple", "update", "use_cache", "v_proj", "value_states", "view"], "mimi/modeling_mimi.py:MimiFlashAttention2.__init__": ["ModelAttention", "ModelFlashAttention2", "__init__", "_flash_attn_uses_top_left_mask", "args", "class", "def", "flash_attn_supports_top_left_mask", "kwargs", "self", "super"], "mimi/modeling_mimi.py:MimiFlashAttention2.forward": ["Cache", "False", "LongTensor", "ModelAttention", "ModelFlashAttention2", "None", "StaticCache", "Tensor", "The", "ValueError", "We", "_", "_flash_attention_forward", "_flash_attn_uses_top_left_mask", "_is_quantized", "an", "and", "apply_rotary_pos_emb", "at", "attention_dropout", "attention_mask", "attn_implementation", "attn_output", "attn_weights", "back", "be", "bool", "bsz", "cache", "cache_kwargs", "cache_position", "cast", "casted", "class", "com", "compatible", "config", "contiguous", "cos", "cpu", "def", "device", "device_type", "dropout", "dropout_rate", "dtype", "elif", "else", "embedding", "f", "fact", "flash_attention_2", "float32", "forward", "get_autocast_dtype", "get_autocast_gpu_dtype", "getattr", "github", "hasattr", "have", "head_dim", "hidden", "hidden_states", "https", "huggingface", "if", "implementation", "in", "input", "input_dtype", "is", "is_autocast_enabled", "is_causal", "isinstance", "issue", "k_proj", "key_states", "layer", "layer_idx", "layers", "logger", "make", "mean", "might", "mps", "norm", "not", "num_heads", "num_key_value_heads", "o_proj", "open", "or", "output_attentions", "past_key_values", "position_ids", "q_len", "q_proj", "query_states", "raise", "related", "reshape", "return", "rotary_emb", "sdpa", "seems", "self", "silently", "sin", "size", "sliding_window", "states", "static", "sure", "target_dtype", "the", "this", "time", "to", "torch", "training", "transformers", "transpose", "tuple", "type", "upcasted", "update", "use", "use_cache", "use_top_left_mask", "v_proj", "value_states", "view", "warning_once", "weight", "will", "with", "you"], "mimi/modeling_mimi.py:MimiSdpaAttention.forward": ["Cache", "False", "If", "LongTensor", "ModelAttention", "ModelSdpaAttention", "None", "Tensor", "The", "True", "_", "__class__", "__name__", "and", "apply_rotary_pos_emb", "attention", "attention_dropout", "attention_mask", "attn_implementation", "attn_mask", "attn_output", "be", "bool", "bsz", "cache_kwargs", "cache_position", "causal_mask", "class", "contiguous", "cos", "cuda", "def", "device", "does", "dropout_p", "eager", "else", "f", "forward", "functional", "get", "head_dim", "hidden_states", "if", "is", "is_causal", "k_proj", "key_states", "kwargs", "layer_idx", "loading", "logger", "model", "nn", "not", "num_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "output_attentions", "past_key_values", "please", "position_ids", "q_len", "q_proj", "query_states", "repeat_kv", "return", "returned", "rotary_emb", "scaled_dot_product_attention", "self", "set", "shape", "sin", "size", "support", "the", "to", "torch", "training", "transpose", "tuple", "type", "update", "use_cache", "v_proj", "value_states", "view", "want", "warning_once", "weights", "when", "will", "you"], "mimi/modeling_mimi.py:MimiTransformerLayer.__init__": ["GradientCheckpointingLayer", "LayerNorm", "ModelConfig", "ModelLayerScale", "ModelMLP", "ModelTransformerLayer", "Model_ATTENTION_CLASSES", "__init__", "_attn_implementation", "class", "config", "def", "eps", "hidden_size", "input_layernorm", "int", "layer_idx", "mlp", "mlp_layer_scale", "nn", "norm_eps", "post_attention_layernorm", "self", "self_attn", "self_attn_layer_scale", "super"], "mimi/modeling_mimi.py:MimiTransformerLayer.forward": ["Cache", "False", "FloatTensor", "GradientCheckpointingLayer", "LongTensor", "ModelTransformerLayer", "None", "Tensor", "attention_mask", "bool", "cache_position", "class", "def", "forward", "hidden_states", "if", "input_layernorm", "kwargs", "mlp", "mlp_layer_scale", "output_attentions", "outputs", "past_key_values", "position_ids", "post_attention_layernorm", "residual", "return", "self", "self_attn", "self_attn_layer_scale", "self_attn_weights", "torch", "tuple", "use_cache"], "mimi/modeling_mimi.py:MimiTransformerModel.__init__": ["False", "ModelConfig", "ModelTransformerLayer", "ModelTransformerModel", "Module", "ModuleList", "__init__", "_attn_implementation", "class", "config", "def", "for", "gradient_checkpointing", "in", "layer_idx", "layers", "nn", "num_hidden_layers", "range", "self", "super"], "mimi/modeling_mimi.py:MimiTransformerModel.forward": ["BaseModelOutputWithPast", "Cache", "DynamicCache", "False", "LongTensor", "ModelTransformerModel", "Module", "None", "Setting", "Tensor", "True", "all_hidden_states", "all_self_attns", "and", "arange", "attention_mask", "attentions", "bool", "cache_position", "causal_mask", "checkpointing", "class", "config", "create_causal_mask", "decoder_layer", "def", "device", "else", "for", "forward", "get_seq_length", "gradient", "gradient_checkpointing", "hidden_states", "if", "in", "incompatible", "input_embeds", "is", "last_hidden_state", "layer_outputs", "layers", "logger", "nn", "not", "output_attentions", "output_hidden_states", "past_key_values", "past_seen_tokens", "position_ids", "return", "return_dict", "self", "shape", "torch", "training", "tuple", "unsqueeze", "use_cache", "use_return_dict", "v", "warning_once", "with"], "mimi/modeling_mimi.py:MimiDecoder.__init__": ["ELU", "ModelConfig", "ModelConv1d", "ModelConvTranspose1d", "ModelDecoder", "ModelResnetBlock", "Module", "ModuleList", "__init__", "audio_channels", "class", "config", "current_scale", "def", "dilation_growth_rate", "for", "hidden_size", "in", "int", "j", "kernel_size", "last_kernel_size", "layers", "len", "model", "nn", "num_filters", "num_residual_layers", "range", "ratio", "scaling", "self", "stride", "super", "upsampling_ratios"], "mimi/modeling_mimi.py:MimiDecoder.forward": ["ModelDecoder", "Module", "class", "def", "for", "forward", "hidden_states", "in", "layer", "layers", "nn", "return", "self"], "mimi/modeling_mimi.py:MimiEuclideanCodebook.__init__": ["ModelConfig", "ModelEuclideanCodebook", "Module", "None", "True", "__init__", "_embed", "class", "cluster_usage", "codebook_dim", "codebook_size", "config", "def", "dtype", "embed", "embed_sum", "epsilon", "float", "float32", "initialized", "nn", "ones", "register_buffer", "self", "super", "tensor", "torch", "zeros"], "mimi/modeling_mimi.py:MimiEuclideanCodebook.embed": ["ModelEuclideanCodebook", "Module", "None", "Tensor", "_embed", "clamp", "class", "cluster_usage", "def", "embed", "embed_sum", "epsilon", "if", "is", "min", "nn", "return", "self", "torch"], "mimi/modeling_mimi.py:MimiEuclideanCodebook.quantize": ["ModelEuclideanCodebook", "Module", "None", "argmin", "cdist", "class", "def", "dim", "dists", "embed", "embed_ind", "float", "hidden_states", "nn", "p", "quantize", "return", "self", "torch"], "mimi/modeling_mimi.py:MimiEuclideanCodebook.encode": ["ModelEuclideanCodebook", "Module", "class", "def", "embed_ind", "encode", "hidden_states", "nn", "quantize", "reshape", "return", "self", "shape", "view"], "mimi/modeling_mimi.py:MimiEuclideanCodebook.decode": ["ModelEuclideanCodebook", "Module", "class", "decode", "def", "embed", "embed_ind", "embedding", "functional", "nn", "quantize", "return", "self"], "mimi/modeling_mimi.py:MimiVectorQuantization.__init__": ["ModelConfig", "ModelEuclideanCodebook", "ModelVectorQuantization", "Module", "__init__", "class", "codebook", "config", "def", "nn", "self", "super"], "mimi/modeling_mimi.py:MimiVectorQuantization.encode": ["ModelVectorQuantization", "Module", "class", "codebook", "def", "embed_in", "encode", "hidden_states", "nn", "permute", "return", "self"], "mimi/modeling_mimi.py:MimiVectorQuantization.decode": ["ModelVectorQuantization", "Module", "class", "codebook", "decode", "def", "embed_ind", "nn", "permute", "quantize", "return", "self"], "mimi/modeling_mimi.py:MimiResidualVectorQuantizer.__init__": ["Conv1d", "False", "ModelConfig", "ModelResidualVectorQuantizer", "ModelVectorQuantization", "Module", "ModuleList", "None", "_", "__init__", "bias", "class", "codebook_size", "config", "def", "else", "for", "frame_rate", "hidden_size", "if", "in", "input_proj", "int", "is", "layers", "nn", "not", "num_quantizers", "output_proj", "range", "self", "super", "torch", "vector_quantization_hidden_dimension"], "mimi/modeling_mimi.py:MimiResidualVectorQuantizer.encode": ["ModelResidualVectorQuantizer", "Module", "None", "Tensor", "all_indices", "append", "class", "decode", "def", "else", "embeddings", "encode", "for", "if", "in", "indices", "input_proj", "int", "is", "layer", "layers", "nn", "not", "num_quantizers", "out_indices", "quantized", "residual", "return", "self", "stack", "torch"], "mimi/modeling_mimi.py:MimiResidualVectorQuantizer.decode": ["ModelResidualVectorQuantizer", "Module", "None", "Tensor", "class", "codes", "decode", "def", "device", "enumerate", "for", "i", "if", "in", "indices", "is", "layer", "layers", "nn", "not", "output_proj", "quantized", "quantized_out", "return", "self", "tensor", "torch", "transpose"], "mimi/modeling_mimi.py:MimiSplitResidualVectorQuantizer.__init__": ["ModelConfig", "ModelResidualVectorQuantizer", "ModelSplitResidualVectorQuantizer", "Module", "__init__", "acoustic_residual_vector_quantizer", "class", "codebook_size", "config", "def", "frame_rate", "max_num_quantizers", "nn", "num_acoustic_quantizers", "num_quantizers", "num_semantic_quantizers", "self", "semantic_residual_vector_quantizer", "super"], "mimi/modeling_mimi.py:MimiSplitResidualVectorQuantizer.encode": ["ModelSplitResidualVectorQuantizer", "Module", "None", "Tensor", "The", "ValueError", "acoustic_codes", "acoustic_residual_vector_quantizer", "asked", "be", "but", "cat", "class", "codebooks", "codes", "currently", "def", "dim", "e", "else", "embeddings", "encode", "f", "float", "higher", "i", "if", "is", "lower", "max_num_quantizers", "nn", "num_quantizers", "num_semantic_quantizers", "number", "of", "quantizers", "raise", "return", "self", "semantic", "semantic_residual_vector_quantizer", "should", "than", "the", "torch", "total"], "mimi/modeling_mimi.py:MimiSplitResidualVectorQuantizer.decode": ["ModelSplitResidualVectorQuantizer", "Module", "Tensor", "acoustic_residual_vector_quantizer", "class", "codes", "decode", "def", "if", "nn", "num_semantic_quantizers", "quantized_out", "return", "self", "semantic_residual_vector_quantizer", "shape", "torch"], "mimi/modeling_mimi.py:MimiPreTrainedModel._init_weights": ["Conv1d", "ConvTranspose1d", "LayerNorm", "Linear", "ModelConv1d", "ModelEuclideanCodebook", "ModelLayerScale", "ModelPreTrainedModel", "ModelRotaryEmbedding", "None", "PreTrainedModel", "ROPE_INIT_FUNCTIONS", "_", "_init_weights", "a", "b", "bias", "buffer_value", "class", "cluster_usage", "compute_default_rope_parameters", "config", "constant_", "conv", "copy_", "def", "default", "dilation", "elif", "else", "embed_sum", "groups", "if", "in_channels", "init", "initialized", "initializer_range", "inv_freq", "is", "isinstance", "k", "kaiming_normal_", "kernel_size", "layer_scale_initial_scale", "math", "mean", "module", "nn", "normal_", "not", "ones_", "original_inv_freq", "padding_total", "rope_fn", "rope_type", "scale", "self", "sqrt", "std", "stride", "uniform_", "weight", "zeros_"], "mimi/modeling_mimi.py:MimiModel.__init__": ["False", "ModelConfig", "ModelConv1d", "ModelConvTranspose1d", "ModelDecoder", "ModelEncoder", "ModelModel", "ModelPreTrainedModel", "ModelSplitResidualVectorQuantizer", "ModelTransformerModel", "None", "The", "ValueError", "_Modelconv1d_layer_names", "__init__", "a", "be", "bias", "bits_per_codebook", "class", "codebook_size", "config", "decoder", "decoder_transformer", "def", "downsample", "encodec_frame_rate", "encoder", "encoder_transformer", "frame_rate", "groups", "hidden_size", "if", "int", "kernel_size", "layer_idx", "len", "log2", "math", "must", "of", "pad_mode", "post_init", "power", "quantizer", "raise", "replicate", "self", "stride", "super", "upsample", "upsample_groups"], "mimi/modeling_mimi.py:MimiModel._encode_frame": ["Cache", "ModelConv1dPaddingCache", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "_encode_frame", "bool", "class", "codes", "def", "downsample", "elif", "embeddings", "encode", "encoder", "encoder_outputs", "encoder_transformer", "get", "if", "input_values", "int", "len", "num_quantizers", "padding_cache", "padding_mask", "past_key_values", "quantizer", "return", "return_dict", "self", "torch", "transpose", "tuple"], "mimi/modeling_mimi.py:MimiModel.get_encoded_length": ["LongTensor", "ModelModel", "ModelPreTrainedModel", "_Modelconv1d_layer_names", "_get_output_length", "class", "def", "downsample", "encoder", "for", "get_encoded_length", "get_submodule", "in", "input_length", "layer_name", "output_length", "return", "self", "torch"], "mimi/modeling_mimi.py:MimiModel.get_audio_codes_mask": ["ModelModel", "ModelPreTrainedModel", "Tensor", "arange", "audio_codes_mask", "class", "def", "device", "dim", "dims", "else", "encoded_lengths", "expand", "flip", "get_audio_codes_mask", "get_encoded_length", "if", "len", "max", "padding_mask", "padding_side", "return", "right", "self", "str", "sum", "to", "torch", "unsqueeze"], "mimi/modeling_mimi.py:MimiModel.encode": ["Cache", "ModelConv1dPaddingCache", "ModelEncoderOutput", "ModelModel", "ModelPreTrainedModel", "None", "Number", "Tensor", "The", "ValueError", "_", "_Modelconv1d_layer_names", "_encode_frame", "and", "append", "asked", "audio", "be", "bool", "but", "channels", "class", "codebooks", "config", "currently", "def", "downsample", "e", "else", "encode", "encoded_frames", "encoder", "encoder_past_key_values", "f", "float", "for", "get_submodule", "got", "i", "if", "in", "in_channels", "input_length", "input_values", "is", "layer_name", "len", "lower", "must", "not", "num_layers", "num_quantizers", "number", "of", "ones_like", "or", "pad_mode", "padding_cache", "padding_mask", "padding_total", "past_key_values", "per_layer_in_channels", "per_layer_padding", "per_layer_padding_mode", "quantizers", "raise", "return", "return_dict", "self", "shape", "should", "than", "the", "torch", "total", "tuple", "use_streaming"], "mimi/modeling_mimi.py:MimiModel._decode_frame": ["Cache", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "_decode_frame", "bool", "class", "codes", "decode", "decoder", "decoder_outputs", "decoder_transformer", "def", "elif", "embeddings", "get", "if", "len", "outputs", "past_key_values", "quantizer", "return", "return_dict", "self", "torch", "transpose", "upsample"], "mimi/modeling_mimi.py:MimiModel.decode": ["Cache", "ModelDecoderOutput", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "_decode_frame", "and", "audio_codes", "audio_values", "bool", "class", "config", "decode", "decoder_past_key_values", "def", "else", "if", "is", "not", "padding_mask", "past_key_values", "return", "return_dict", "self", "shape", "torch", "tuple"], "mimi/modeling_mimi.py:MimiModel.forward": ["Cache", "ModelModel", "ModelOutput", "ModelPreTrainedModel", "None", "Tensor", "audio_codes", "audio_values", "bool", "class", "config", "decode", "decoder_outputs", "decoder_past_key_values", "def", "elif", "else", "encode", "encoder_outputs", "encoder_past_key_values", "forward", "get", "if", "input_values", "int", "is", "kwargs", "len", "not", "num_quantizers", "ones_like", "padding_mask", "past_key_values", "r", "return", "return_dict", "self", "torch", "tuple"], "minimax/modeling_minimax.py:MiniMaxRMSNorm.__init__": ["ModelRMSNorm", "Module", "Parameter", "__init__", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "minimax/modeling_minimax.py:MiniMaxRMSNorm.forward": ["ModelRMSNorm", "Module", "True", "class", "def", "dtype", "float32", "forward", "hidden_states", "input_dtype", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "minimax/modeling_minimax.py:MiniMaxRMSNorm.extra_repr": ["ModelRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "variance_epsilon", "weight"], "minimax/modeling_minimax.py:MiniMaxCache.__init__": ["DynamicCache", "ModelCache", "Tensor", "__init__", "class", "def", "linear_cache", "list", "self", "super", "torch"], "minimax/modeling_minimax.py:MiniMaxCache.set_linear_cache": ["DynamicCache", "ModelCache", "_", "append", "class", "def", "for", "in", "layer_idx", "len", "linear_cache", "range", "self", "set_linear_cache"], "minimax/modeling_minimax.py:MiniMaxCache.get_linear_cache": ["DynamicCache", "ModelCache", "None", "class", "def", "get_linear_cache", "if", "int", "layer_idx", "len", "linear_cache", "return", "self"], "minimax/modeling_minimax.py:MiniMaxCache.__len__": ["DynamicCache", "ModelCache", "__len__", "class", "def", "len", "linear_cache", "max", "return", "self", "super"], "minimax/modeling_minimax.py:MiniMaxCache.batch_repeat_interleave": ["DynamicCache", "ModelCache", "batch_repeat_interleave", "class", "def", "dim", "else", "for", "if", "in", "int", "layer_idx", "layers", "len", "linear_cache", "range", "repeat_interleave", "repeats", "self"], "minimax/modeling_minimax.py:MiniMaxCache.batch_select_indices": ["DynamicCache", "ModelCache", "Tensor", "batch_select_indices", "class", "def", "else", "for", "if", "in", "indices", "layer_idx", "layers", "len", "linear_cache", "range", "self", "torch"], "minimax/modeling_minimax.py:MiniMaxCache.crop": ["DynamicCache", "ModelCache", "RuntimeError", "class", "crop", "def", "doesnot", "int", "max_length", "method", "raise", "self", "support"], "minimax/modeling_minimax.py:MiniMaxLightningAttention.__init__": ["ACT2FN", "False", "Linear", "ModelConfig", "ModelLightningAttention", "ModelRMSNorm", "Module", "None", "__init__", "act_fn", "bias", "block_size", "class", "config", "decay_factors", "def", "diagonal_decay", "get_slope_rate", "getattr", "head_dim", "hidden_act", "hidden_size", "int", "key_decay", "layer_idx", "nn", "norm", "num_attention_heads", "num_hidden_layers", "or", "out_proj", "output_gate", "qkv_proj", "query_decay", "register_buffer", "self", "slope_rate", "super"], "minimax/modeling_minimax.py:MiniMaxLightningAttention.get_slope_rate": ["ModelLightningAttention", "Module", "None", "arange", "base", "class", "def", "exponent", "factor", "get_slope_rate", "layer_idx", "nn", "num_attention_heads", "num_hidden_layers", "rate", "return", "self", "torch"], "minimax/modeling_minimax.py:MiniMaxLightningAttention.decay_factors": ["ModelLightningAttention", "Module", "None", "arange", "block_size", "block_size_range", "class", "decay_factors", "def", "diagonal_decay", "exp", "float", "inf", "key_decay", "nn", "query_decay", "return", "self", "slope_rate", "torch", "where"], "minimax/modeling_minimax.py:MiniMaxLightningAttention.forward": ["Cache", "F", "FlashAttentionKwargs", "LongTensor", "ModelLightningAttention", "Module", "None", "Tensor", "Unpack", "act_fn", "append", "attention_mask", "attn_output", "attn_output_inter", "attn_output_intra", "attn_weights_inter", "attn_weights_intra", "batch_size", "block_decay", "block_size", "bool", "cache_position", "cat", "class", "current_attn_output", "current_attn_weights_inter", "current_block_size", "current_diagonal_decay", "current_key_decay", "current_key_states", "current_query_decay", "current_query_states", "current_value_states", "def", "diagonal_decay", "dim", "dtype", "else", "end_idx", "exp", "for", "forward", "get_linear_cache", "head_dim", "hidden_size", "hidden_states", "i", "if", "in", "is", "key_decay", "key_states", "kwargs", "layer_idx", "masked_fill", "matmul", "min", "next_attn_weights_inter", "nn", "norm", "not", "num_attention_heads", "num_blocks", "out_proj", "output_gate", "past_key_values", "position_embeddings", "qkv_proj", "qkv_states", "query_decay", "query_states", "range", "ratio", "reshape", "return", "self", "seq_len", "set_linear_cache", "shape", "sigmoid", "slope_rate", "split", "start_idx", "to", "torch", "transpose", "tuple", "unsqueeze", "value_states", "zeros"], "minimax/modeling_minimax.py:MiniMaxRotaryEmbedding.__init__": ["Callable", "False", "ModelConfig", "ModelRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "minimax/modeling_minimax.py:MiniMaxRotaryEmbedding.compute_default_rope_parameters": ["ModelConfig", "ModelRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "minimax/modeling_minimax.py:MiniMaxRotaryEmbedding.forward": ["False", "ModelRotaryEmbedding", "Module", "None", "and", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "minimax/modeling_minimax.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "minimax/modeling_minimax.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cos", "def", "k", "k_embed", "q", "q_embed", "return", "rotate_half", "sin", "unsqueeze", "unsqueeze_dim"], "minimax/modeling_minimax.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "minimax/modeling_minimax.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value", "value_states"], "minimax/modeling_minimax.py:MiniMaxAttention.__init__": ["False", "Linear", "ModelAttention", "ModelConfig", "Module", "None", "True", "__init__", "attention_dropout", "bias", "class", "config", "def", "getattr", "head_dim", "hidden_size", "int", "is_causal", "k_proj", "layer_idx", "nn", "num_attention_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "or", "q_proj", "scaling", "self", "super", "v_proj"], "minimax/modeling_minimax.py:MiniMaxAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "FlashAttentionKwargs", "LongTensor", "ModelAttention", "Module", "None", "Tensor", "Unpack", "_attn_implementation", "apply_rotary_pos_emb", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "cache_kwargs", "cache_position", "class", "config", "contiguous", "cos", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "getattr", "head_dim", "hidden_shape", "hidden_states", "if", "input_shape", "is", "k_proj", "key_states", "kwargs", "layer_idx", "nn", "not", "o_proj", "past_key_values", "position_embeddings", "q_proj", "query_states", "reshape", "return", "scaling", "self", "shape", "sin", "sliding_window", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "view"], "minimax/modeling_minimax.py:MiniMaxTopKRouter.__init__": ["ModelTopKRouter", "Module", "Parameter", "__init__", "class", "config", "def", "empty", "hidden_dim", "hidden_size", "nn", "num_experts", "num_experts_per_tok", "num_local_experts", "self", "super", "top_k", "torch", "weight"], "minimax/modeling_minimax.py:MiniMaxTopKRouter.forward": ["F", "ModelTopKRouter", "Module", "True", "class", "def", "dim", "float", "forward", "functional", "hidden_dim", "hidden_states", "keepdim", "linear", "nn", "reshape", "return", "router_indices", "router_logits", "router_scores", "router_top_value", "self", "softmax", "sum", "top_k", "topk", "torch", "weight"], "minimax/modeling_minimax.py:MiniMaxExperts.__init__": ["ACT2FN", "ModelConfig", "ModelExperts", "Module", "Parameter", "__init__", "act_fn", "class", "config", "def", "down_proj", "empty", "gate_up_proj", "hidden_act", "hidden_dim", "hidden_size", "intermediate_dim", "intermediate_size", "nn", "num_experts", "num_local_experts", "self", "super", "torch"], "minimax/modeling_minimax.py:MiniMaxExperts.forward": ["ModelExperts", "Module", "None", "Tensor", "act_fn", "chunk", "class", "continue", "current_hidden_states", "current_state", "def", "dim", "down_proj", "dtype", "expert_hit", "expert_idx", "expert_mask", "final_hidden_states", "for", "forward", "functional", "gate", "gate_up_proj", "greater", "hidden_states", "if", "in", "index_add_", "linear", "nn", "no_grad", "nonzero", "num_classes", "num_experts", "one_hot", "permute", "return", "self", "sum", "to", "token_idx", "top_k_index", "top_k_pos", "top_k_weights", "torch", "up", "where", "with", "zeros_like"], "minimax/modeling_minimax.py:MiniMaxSparseMoeBlock.__init__": ["ModelExperts", "ModelSparseMoeBlock", "ModelTopKRouter", "Module", "__init__", "class", "config", "def", "experts", "gate", "jitter_noise", "nn", "num_experts_per_tok", "router_jitter_noise", "self", "super", "top_k"], "minimax/modeling_minimax.py:MiniMaxSparseMoeBlock.forward": ["ModelSparseMoeBlock", "Module", "Tensor", "_", "and", "batch_size", "class", "def", "empty_like", "experts", "forward", "gate", "hidden_dim", "hidden_states", "if", "jitter_noise", "nn", "reshape", "return", "self", "sequence_length", "shape", "top_k_index", "top_k_weights", "torch", "training", "tuple", "uniform_", "view"], "minimax/modeling_minimax.py:MiniMaxDecoderLayer.__init__": ["GradientCheckpointingLayer", "ModelAttention", "ModelConfig", "ModelDecoderLayer", "ModelLightningAttention", "ModelRMSNorm", "ModelSparseMoeBlock", "None", "__init__", "attn_alpha_factor", "attn_beta_factor", "class", "config", "def", "else", "eps", "full_attn_alpha_factor", "full_attn_beta_factor", "hasattr", "hidden_size", "if", "input_layernorm", "int", "layer_idx", "layer_type", "layer_types", "linear_attention", "linear_attn_alpha_factor", "linear_attn_beta_factor", "mlp", "mlp_alpha_factor", "mlp_beta_factor", "post_attention_layernorm", "rms_norm_eps", "self", "self_attn", "super"], "minimax/modeling_minimax.py:MiniMaxDecoderLayer.forward": ["Cache", "False", "FlashAttentionKwargs", "FloatTensor", "GradientCheckpointingLayer", "LongTensor", "ModelDecoderLayer", "None", "Tensor", "Unpack", "_", "attention_mask", "attn_alpha_factor", "attn_beta_factor", "bool", "cache_position", "class", "def", "forward", "hidden_states", "input_layernorm", "kwargs", "mlp", "mlp_alpha_factor", "mlp_beta_factor", "past_key_values", "position_embeddings", "position_ids", "post_attention_layernorm", "residual", "return", "self", "self_attn", "torch", "tuple", "use_cache"], "minimax/modeling_minimax.py:MiniMaxPreTrainedModel._init_weights": ["ModelExperts", "ModelLightningAttention", "ModelPreTrainedModel", "ModelTopKRouter", "PreTrainedModel", "_init_weights", "class", "config", "copy_", "decay_factors", "def", "diagonal_decay", "down_proj", "elif", "gate_up_proj", "get_slope_rate", "if", "init", "initializer_range", "isinstance", "key_decay", "mean", "module", "normal_", "query_decay", "self", "slope_rate", "std", "super", "weight"], "minimax/modeling_minimax.py:MiniMaxModel.__init__": ["Embedding", "False", "ModelConfig", "ModelDecoderLayer", "ModelModel", "ModelPreTrainedModel", "ModelRMSNorm", "ModelRotaryEmbedding", "ModuleList", "__init__", "class", "config", "def", "embed_tokens", "eps", "for", "gradient_checkpointing", "hidden_size", "in", "layer_idx", "layers", "nn", "norm", "num_hidden_layers", "pad_token_id", "padding_idx", "post_init", "range", "rms_norm_eps", "rotary_emb", "self", "super", "vocab_size"], "minimax/modeling_minimax.py:MiniMaxModel.forward": ["FloatTensor", "LongTensor", "Model", "ModelCache", "ModelModel", "ModelPreTrainedModel", "MoeModelOutputWithPast", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "and", "arange", "attention_mask", "bool", "cache", "cache_position", "causal_mask", "class", "compatible", "config", "create_causal_mask", "create_sliding_window_causal_mask", "decoder_layer", "def", "device", "elif", "else", "embed_tokens", "exactly", "f", "for", "forward", "full_attention", "get_seq_length", "hidden_states", "if", "in", "input_attention_mask", "input_embeds", "input_ids", "inputs_embeds", "is", "isinstance", "its", "kwargs", "last_hidden_state", "layer_type", "layers", "mask_function", "must", "norm", "not", "of", "one", "or", "own", "past_key_values", "past_seen_tokens", "position_embeddings", "position_ids", "raise", "return", "rotary_emb", "self", "shape", "sliding_window", "specify", "torch", "type", "unsqueeze", "use_cache", "uses", "with"], "minimax/modeling_minimax.py:load_balancing_loss_func": ["Model_balancing_loss_func", "None", "Tensor", "_", "attention_mask", "batch_size", "cat", "compute_device", "concatenated_gate_logits", "def", "device", "dim", "else", "expand", "expert_attention_mask", "expert_mask", "float", "for", "functional", "gate_logits", "if", "in", "int", "is", "isinstance", "layer_gate", "mean", "nn", "not", "num_experts", "num_hidden_layers", "one_hot", "or", "overall_loss", "r", "reshape", "return", "router_per_expert_attention_mask", "router_prob_per_expert", "routing_weights", "selected_experts", "sequence_length", "shape", "softmax", "sum", "to", "tokens_per_expert", "top_k", "topk", "torch", "tuple", "unsqueeze"], "minimax/modeling_minimax.py:MiniMaxForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "ModelForCausalLM", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "num_experts", "num_experts_per_tok", "num_local_experts", "post_init", "router_aux_loss_coef", "self", "super", "vocab_size"], "minimax/modeling_minimax.py:MiniMaxForCausalLM.forward": ["Cache", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForCausalLM", "ModelPreTrainedModel", "MoeCausalLMOutputWithPast", "MoeModelOutputWithPast", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "aux_loss", "bool", "cache_position", "class", "config", "def", "device", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "load_balancing_loss_func", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "num_experts", "num_experts_per_tok", "output_router_logits", "outputs", "past_key_values", "position_ids", "r", "return", "router_aux_loss_coef", "router_logits", "self", "slice", "slice_indices", "to", "torch", "use_cache", "vocab_size"], "minimax_m2/modeling_minimax_m2.py:MiniMaxM2TopKRouter.__init__": ["ModelTopKRouter", "Module", "Parameter", "__init__", "class", "config", "def", "empty", "hidden_dim", "hidden_size", "nn", "num_experts", "num_experts_per_tok", "num_local_experts", "self", "super", "top_k", "torch", "weight"], "minimax_m2/modeling_minimax_m2.py:MiniMaxM2TopKRouter.forward": ["F", "False", "ModelTopKRouter", "Module", "True", "_", "class", "def", "dim", "dtype", "e_score_correction_bias", "float", "forward", "functional", "gather", "hidden_dim", "hidden_states", "keepdim", "linear", "nn", "reshape", "return", "router_logits", "router_scores", "routing_weights", "scores_for_choice", "self", "sigmoid", "sorted", "sum", "to", "top_k", "top_k_index", "top_k_weights", "topk", "torch", "weight"], "minimax_m2/modeling_minimax_m2.py:MiniMaxM2Experts.__init__": ["ACT2FN", "ModelConfig", "ModelExperts", "Module", "Parameter", "__init__", "act_fn", "class", "config", "def", "down_proj", "empty", "gate_up_proj", "hidden_act", "hidden_dim", "hidden_size", "intermediate_dim", "intermediate_size", "nn", "num_experts", "num_local_experts", "self", "super", "torch"], "minimax_m2/modeling_minimax_m2.py:MiniMaxM2Experts.forward": ["ModelExperts", "Module", "None", "Tensor", "act_fn", "chunk", "class", "continue", "current_hidden_states", "current_state", "def", "dim", "down_proj", "dtype", "expert_hit", "expert_idx", "expert_mask", "final_hidden_states", "for", "forward", "functional", "gate", "gate_up_proj", "greater", "hidden_states", "if", "in", "index_add_", "linear", "nn", "no_grad", "nonzero", "num_classes", "num_experts", "one_hot", "permute", "return", "self", "sum", "to", "token_idx", "top_k_index", "top_k_pos", "top_k_weights", "torch", "up", "where", "with", "zeros_like"], "minimax_m2/modeling_minimax_m2.py:MiniMaxM2SparseMoeBlock.__init__": ["ModelExperts", "ModelSparseMoeBlock", "ModelTopKRouter", "Module", "__init__", "class", "config", "def", "e_score_correction_bias", "experts", "gate", "jitter_noise", "nn", "num_experts_per_tok", "num_local_experts", "register_buffer", "router_jitter_noise", "self", "super", "top_k", "torch", "zeros"], "minimax_m2/modeling_minimax_m2.py:MiniMaxM2SparseMoeBlock.forward": ["ModelSparseMoeBlock", "Module", "Tensor", "_", "and", "batch_size", "class", "def", "e_score_correction_bias", "empty_like", "experts", "forward", "gate", "hidden_dim", "hidden_states", "if", "jitter_noise", "nn", "reshape", "return", "self", "sequence_length", "shape", "top_k_index", "top_k_weights", "torch", "training", "tuple", "uniform_", "view"], "minimax_m2/modeling_minimax_m2.py:MiniMaxM2RMSNorm.__init__": ["ModelRMSNorm", "Module", "Parameter", "__init__", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "minimax_m2/modeling_minimax_m2.py:MiniMaxM2RMSNorm.forward": ["ModelRMSNorm", "Module", "True", "class", "def", "dtype", "float32", "forward", "hidden_states", "input_dtype", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "minimax_m2/modeling_minimax_m2.py:MiniMaxM2RMSNorm.extra_repr": ["ModelRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "variance_epsilon", "weight"], "minimax_m2/modeling_minimax_m2.py:MiniMaxM2RotaryEmbedding.__init__": ["Callable", "False", "ModelConfig", "ModelRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "minimax_m2/modeling_minimax_m2.py:MiniMaxM2RotaryEmbedding.compute_default_rope_parameters": ["ModelConfig", "ModelRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "get", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "partial_rotary_factor", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "minimax_m2/modeling_minimax_m2.py:MiniMaxM2RotaryEmbedding.forward": ["False", "ModelRotaryEmbedding", "Module", "None", "and", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "minimax_m2/modeling_minimax_m2.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "minimax_m2/modeling_minimax_m2.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value", "value_states"], "minimax_m2/modeling_minimax_m2.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cat", "cos", "def", "dim", "k", "k_embed", "k_pass", "k_rot", "q", "q_embed", "q_pass", "q_rot", "return", "rotary_dim", "rotate_half", "shape", "sin", "torch", "unsqueeze", "unsqueeze_dim"], "minimax_m2/modeling_minimax_m2.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "minimax_m2/modeling_minimax_m2.py:MiniMaxM2Attention.__init__": ["False", "Linear", "ModelAttention", "ModelConfig", "ModelRMSNorm", "Module", "True", "__init__", "attention_dropout", "bias", "class", "config", "def", "getattr", "head_dim", "hidden_size", "int", "is_causal", "k_norm", "k_proj", "layer_idx", "nn", "num_attention_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "q_norm", "q_proj", "rms_norm_eps", "scaling", "self", "super", "v_proj"], "minimax_m2/modeling_minimax_m2.py:MiniMaxM2Attention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "LongTensor", "ModelAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_attn_implementation", "apply_rotary_pos_emb", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "cache_kwargs", "cache_position", "class", "config", "contiguous", "cos", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "head_dim", "hidden_shape", "hidden_states", "if", "input_shape", "is", "k_norm", "k_proj", "key_states", "kwargs", "layer_idx", "nn", "not", "o_proj", "past_key_values", "position_embeddings", "q_norm", "q_proj", "query_states", "reshape", "return", "scaling", "self", "shape", "sin", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "view"], "minimax_m2/modeling_minimax_m2.py:MiniMaxM2DecoderLayer.__init__": ["GradientCheckpointingLayer", "ModelAttention", "ModelConfig", "ModelDecoderLayer", "ModelRMSNorm", "ModelSparseMoeBlock", "__init__", "class", "config", "def", "eps", "hidden_size", "input_layernorm", "int", "layer_idx", "mlp", "post_attention_layernorm", "rms_norm_eps", "self", "self_attn", "super"], "minimax_m2/modeling_minimax_m2.py:MiniMaxM2DecoderLayer.forward": ["Cache", "GradientCheckpointingLayer", "LongTensor", "ModelDecoderLayer", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "attention_mask", "cache_position", "class", "def", "forward", "hidden_states", "input_layernorm", "kwargs", "mlp", "past_key_values", "position_embeddings", "position_ids", "post_attention_layernorm", "residual", "return", "self", "self_attn", "torch", "tuple"], "minimax_m2/modeling_minimax_m2.py:MiniMaxM2PreTrainedModel._init_weights": ["ModelExperts", "ModelPreTrainedModel", "ModelSparseMoeBlock", "ModelTopKRouter", "PreTrainedModel", "_init_weights", "class", "config", "def", "down_proj", "e_score_correction_bias", "elif", "gate_up_proj", "if", "init", "initializer_range", "isinstance", "mean", "module", "normal_", "self", "std", "super", "weight", "zeros_"], "minimax_m2/modeling_minimax_m2.py:MiniMaxM2Model.__init__": ["Embedding", "False", "ModelConfig", "ModelDecoderLayer", "ModelModel", "ModelPreTrainedModel", "ModelRMSNorm", "ModelRotaryEmbedding", "ModuleList", "__init__", "class", "config", "def", "embed_tokens", "eps", "for", "gradient_checkpointing", "hidden_size", "in", "layer_idx", "layers", "nn", "norm", "num_hidden_layers", "pad_token_id", "padding_idx", "post_init", "range", "rms_norm_eps", "rotary_emb", "self", "super", "vocab_size"], "minimax_m2/modeling_minimax_m2.py:MiniMaxM2Model.forward": ["Cache", "DynamicCache", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "MoeModelOutputWithPast", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "and", "arange", "attention_mask", "bool", "cache_position", "causal_mask", "class", "config", "create_causal_mask", "decoder_layer", "def", "device", "else", "embed_tokens", "exactly", "for", "forward", "get_seq_length", "hidden_states", "if", "in", "input_embeds", "input_ids", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layers", "must", "norm", "not", "num_hidden_layers", "of", "one", "or", "past_key_values", "past_seen_tokens", "position_embeddings", "position_ids", "raise", "return", "rotary_emb", "self", "shape", "specify", "torch", "unsqueeze", "use_cache"], "minimax_m2/modeling_minimax_m2.py:load_balancing_loss_func": ["Model_balancing_loss_func", "None", "Tensor", "_", "attention_mask", "batch_size", "cat", "compute_device", "concatenated_gate_logits", "def", "device", "dim", "else", "expand", "expert_attention_mask", "expert_mask", "float", "for", "functional", "gate_logits", "if", "in", "int", "is", "isinstance", "layer_gate", "mean", "nn", "not", "num_experts", "num_hidden_layers", "one_hot", "or", "overall_loss", "r", "reshape", "return", "router_per_expert_attention_mask", "router_prob_per_expert", "routing_weights", "selected_experts", "sequence_length", "shape", "softmax", "sum", "to", "tokens_per_expert", "top_k", "topk", "torch", "tuple", "unsqueeze"], "minimax_m2/modeling_minimax_m2.py:MiniMaxM2ForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "ModelForCausalLM", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "num_experts", "num_experts_per_tok", "num_local_experts", "post_init", "router_aux_loss_coef", "self", "super", "vocab_size"], "minimax_m2/modeling_minimax_m2.py:MiniMaxM2ForCausalLM.forward": ["Cache", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForCausalLM", "ModelPreTrainedModel", "MoeCausalLMOutputWithPast", "MoeModelOutputWithPast", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "aux_loss", "bool", "cache_position", "class", "config", "def", "device", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "load_balancing_loss_func", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "num_experts", "num_experts_per_tok", "output_router_logits", "outputs", "past_key_values", "position_ids", "r", "return", "router_aux_loss_coef", "router_logits", "self", "slice", "slice_indices", "to", "torch", "use_cache", "vocab_size"], "ministral/modeling_ministral.py:MinistralMLP.__init__": ["ACT2FN", "False", "Linear", "ModelMLP", "Module", "__init__", "act_fn", "bias", "class", "config", "def", "down_proj", "gate_proj", "hidden_act", "hidden_size", "intermediate_size", "nn", "self", "super", "up_proj"], "ministral/modeling_ministral.py:MinistralMLP.forward": ["ModelMLP", "Module", "act_fn", "class", "def", "down_proj", "forward", "gate_proj", "nn", "return", "self", "up_proj", "x"], "ministral/modeling_ministral.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "ministral/modeling_ministral.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cos", "def", "k", "k_embed", "q", "q_embed", "return", "rotate_half", "sin", "unsqueeze", "unsqueeze_dim"], "ministral/modeling_ministral.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "ministral/modeling_ministral.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value", "value_states"], "ministral/modeling_ministral.py:MinistralAttention.__init__": ["False", "Linear", "ModelAttention", "Module", "None", "True", "__init__", "attention_dropout", "bias", "class", "config", "def", "else", "getattr", "hasattr", "head_dim", "hidden_size", "if", "int", "is_causal", "k_proj", "layer_idx", "layer_type", "layer_types", "nn", "num_attention_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "q_proj", "scaling", "self", "sliding_attention", "sliding_window", "super", "v_proj"], "ministral/modeling_ministral.py:MinistralAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "FlashAttentionKwargs", "LongTensor", "ModelAttention", "Module", "None", "Tensor", "Unpack", "_attn_implementation", "apply_rotary_pos_emb", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "cache_kwargs", "cache_position", "class", "config", "contiguous", "cos", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "head_dim", "hidden_shape", "hidden_states", "if", "input_shape", "is", "k_proj", "key_states", "kwargs", "layer_idx", "nn", "not", "o_proj", "past_key_values", "position_embeddings", "q_proj", "query_states", "reshape", "return", "scaling", "self", "shape", "sin", "sliding_window", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "view"], "ministral/modeling_ministral.py:MinistralRMSNorm.__init__": ["ModelRMSNorm", "Module", "None", "Parameter", "__init__", "class", "def", "eps", "float", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "ministral/modeling_ministral.py:MinistralRMSNorm.forward": ["ModelRMSNorm", "Module", "Tensor", "True", "class", "def", "dtype", "float32", "forward", "hidden_states", "input_dtype", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "ministral/modeling_ministral.py:MinistralRMSNorm.extra_repr": ["ModelRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "variance_epsilon", "weight"], "ministral/modeling_ministral.py:MinistralDecoderLayer.__init__": ["GradientCheckpointingLayer", "ModelAttention", "ModelConfig", "ModelDecoderLayer", "ModelMLP", "ModelRMSNorm", "__init__", "attention_type", "class", "config", "def", "eps", "hidden_size", "input_layernorm", "int", "layer_idx", "layer_types", "mlp", "post_attention_layernorm", "rms_norm_eps", "self", "self_attn", "super"], "ministral/modeling_ministral.py:MinistralDecoderLayer.forward": ["Cache", "False", "GradientCheckpointingLayer", "LongTensor", "ModelDecoderLayer", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "attention_mask", "bool", "cache_position", "class", "def", "forward", "hidden_states", "input_layernorm", "kwargs", "mlp", "past_key_values", "position_embeddings", "position_ids", "post_attention_layernorm", "residual", "return", "self", "self_attn", "torch", "tuple", "use_cache"], "ministral/modeling_ministral.py:MinistralRotaryEmbedding.__init__": ["Callable", "False", "ModelConfig", "ModelRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "ministral/modeling_ministral.py:MinistralRotaryEmbedding.compute_default_rope_parameters": ["ModelConfig", "ModelRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "ministral/modeling_ministral.py:MinistralRotaryEmbedding.forward": ["False", "ModelRotaryEmbedding", "Module", "None", "and", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "ministral/modeling_ministral.py:MinistralModel.__init__": ["Embedding", "False", "ModelConfig", "ModelDecoderLayer", "ModelModel", "ModelPreTrainedModel", "ModelRMSNorm", "ModelRotaryEmbedding", "ModuleList", "__init__", "class", "config", "def", "embed_tokens", "eps", "for", "gradient_checkpointing", "hidden_size", "in", "layer_idx", "layers", "nn", "norm", "num_hidden_layers", "pad_token_id", "padding_idx", "post_init", "range", "rms_norm_eps", "rotary_emb", "self", "super", "vocab_size"], "ministral/modeling_ministral.py:MinistralModel.forward": ["BaseModelOutputWithPast", "Cache", "DynamicCache", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "and", "arange", "attention_mask", "attention_type", "bool", "cache_position", "causal_mask_mapping", "class", "config", "create_causal_mask", "create_sliding_window_causal_mask", "decoder_layer", "def", "device", "dict", "else", "embed_tokens", "exactly", "for", "forward", "full_attention", "get_seq_length", "hidden_states", "if", "in", "input_embeds", "input_ids", "inputs_embeds", "is", "isinstance", "kwargs", "last_hidden_state", "layers", "mask_kwargs", "must", "norm", "not", "num_hidden_layers", "of", "one", "or", "past_key_values", "past_seen_tokens", "position_embeddings", "position_ids", "raise", "return", "rotary_emb", "self", "shape", "sliding_attention", "specify", "torch", "unsqueeze", "use_cache"], "ministral/modeling_ministral.py:MinistralForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "ModelForCausalLM", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "vocab_size"], "ministral/modeling_ministral.py:MinistralForCausalLM.forward": ["BaseModelOutputWithPast", "Cache", "CausalLMOutputWithPast", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForCausalLM", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "outputs", "past_key_values", "position_ids", "r", "return", "self", "slice", "slice_indices", "torch", "use_cache", "vocab_size"], "ministral3/modeling_ministral3.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "ministral3/modeling_ministral3.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cos", "def", "k", "k_embed", "q", "q_embed", "return", "rotate_half", "sin", "unsqueeze", "unsqueeze_dim"], "ministral3/modeling_ministral3.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "ministral3/modeling_ministral3.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value", "value_states"], "ministral3/modeling_ministral3.py:_get_llama_4_attn_scale": ["Tensor", "_get_llama_4_attn_scale", "beta", "def", "float", "floor", "int", "log", "max_position_embeddings", "positions_ids", "return", "scaling", "torch", "unsqueeze"], "ministral3/modeling_ministral3.py:Ministral3Attention.__init__": ["False", "Linear", "ModelAttention", "ModelConfig", "Module", "None", "True", "__init__", "attention_dropout", "bias", "class", "config", "def", "getattr", "head_dim", "hidden_size", "int", "is_causal", "k_proj", "layer_idx", "nn", "num_attention_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "or", "q_proj", "scaling", "self", "super", "v_proj"], "ministral3/modeling_ministral3.py:Ministral3Attention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "FlashAttentionKwargs", "LongTensor", "ModelAttention", "Module", "None", "Tensor", "Unpack", "_attn_implementation", "_get_llama_4_attn_scale", "apply_rotary_pos_emb", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "cache_kwargs", "cache_position", "class", "config", "contiguous", "cos", "def", "dropout", "dtype", "eager", "eager_attention_forward", "else", "forward", "get", "getattr", "head_dim", "hidden_shape", "hidden_states", "if", "input_shape", "is", "k_proj", "key_states", "kwargs", "layer_idx", "llama_4_scaling_beta", "nn", "not", "o_proj", "original_max_position_embeddings", "past_key_values", "position_embeddings", "q_proj", "query_states", "reshape", "return", "rope_parameters", "scaling", "self", "shape", "sin", "sliding_window", "to", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "view"], "ministral3/modeling_ministral3.py:Ministral3MLP.__init__": ["ACT2FN", "False", "Linear", "ModelMLP", "Module", "__init__", "act_fn", "bias", "class", "config", "def", "down_proj", "gate_proj", "hidden_act", "hidden_size", "intermediate_size", "nn", "self", "super", "up_proj"], "ministral3/modeling_ministral3.py:Ministral3MLP.forward": ["ModelMLP", "Module", "act_fn", "class", "def", "down_proj", "forward", "gate_proj", "nn", "return", "self", "up_proj", "x"], "ministral3/modeling_ministral3.py:Ministral3RMSNorm.__init__": ["ModelRMSNorm", "Module", "Parameter", "__init__", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "ministral3/modeling_ministral3.py:Ministral3RMSNorm.forward": ["ModelRMSNorm", "Module", "True", "class", "def", "dtype", "float32", "forward", "hidden_states", "input_dtype", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "ministral3/modeling_ministral3.py:Ministral3RMSNorm.extra_repr": ["ModelRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "variance_epsilon", "weight"], "ministral3/modeling_ministral3.py:Ministral3DecoderLayer.__init__": ["GradientCheckpointingLayer", "ModelAttention", "ModelConfig", "ModelDecoderLayer", "ModelMLP", "ModelRMSNorm", "__init__", "class", "config", "def", "eps", "hidden_size", "input_layernorm", "int", "layer_idx", "mlp", "post_attention_layernorm", "rms_norm_eps", "self", "self_attn", "super"], "ministral3/modeling_ministral3.py:Ministral3DecoderLayer.forward": ["Cache", "False", "GradientCheckpointingLayer", "LongTensor", "ModelDecoderLayer", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "attention_mask", "bool", "cache_position", "class", "def", "forward", "hidden_states", "input_layernorm", "kwargs", "mlp", "past_key_values", "position_embeddings", "position_ids", "post_attention_layernorm", "residual", "return", "self", "self_attn", "torch", "tuple", "use_cache"], "ministral3/modeling_ministral3.py:Ministral3RotaryEmbedding.__init__": ["Callable", "False", "ModelConfig", "ModelRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "ministral3/modeling_ministral3.py:Ministral3RotaryEmbedding.compute_default_rope_parameters": ["ModelConfig", "ModelRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "ministral3/modeling_ministral3.py:Ministral3RotaryEmbedding.forward": ["False", "ModelRotaryEmbedding", "Module", "None", "and", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "ministral3/modeling_ministral3.py:Ministral3Model.__init__": ["Embedding", "False", "ModelConfig", "ModelDecoderLayer", "ModelModel", "ModelPreTrainedModel", "ModelRMSNorm", "ModelRotaryEmbedding", "ModuleList", "__init__", "class", "config", "def", "embed_tokens", "eps", "for", "gradient_checkpointing", "hidden_size", "in", "layer_idx", "layers", "nn", "norm", "num_hidden_layers", "pad_token_id", "padding_idx", "post_init", "range", "rms_norm_eps", "rotary_emb", "self", "super", "vocab_size"], "ministral3/modeling_ministral3.py:Ministral3Model.forward": ["BaseModelOutputWithPast", "Cache", "DynamicCache", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "and", "arange", "attention_mask", "bool", "cache_position", "causal_mask", "class", "config", "create_causal_mask", "create_sliding_window_causal_mask", "decoder_layer", "def", "device", "else", "embed_tokens", "exactly", "for", "forward", "get_seq_length", "hidden_states", "if", "in", "input_embeds", "input_ids", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layers", "mask_function", "must", "norm", "not", "num_hidden_layers", "of", "one", "or", "past_key_values", "past_seen_tokens", "position_embeddings", "position_ids", "raise", "return", "rotary_emb", "self", "shape", "sliding_window", "specify", "torch", "unsqueeze", "use_cache"], "ministral3/modeling_ministral3.py:Ministral3ForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "ModelForCausalLM", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "vocab_size"], "ministral3/modeling_ministral3.py:Ministral3ForCausalLM.forward": ["BaseModelOutputWithPast", "Cache", "CausalLMOutputWithPast", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForCausalLM", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "outputs", "past_key_values", "position_ids", "r", "return", "self", "slice", "slice_indices", "torch", "use_cache", "vocab_size"], "mistral/modeling_mistral.py:MistralMLP.__init__": ["ACT2FN", "False", "Linear", "ModelMLP", "Module", "__init__", "act_fn", "bias", "class", "config", "def", "down_proj", "gate_proj", "hidden_act", "hidden_size", "intermediate_size", "nn", "self", "super", "up_proj"], "mistral/modeling_mistral.py:MistralMLP.forward": ["ModelMLP", "Module", "act_fn", "class", "def", "down_proj", "forward", "gate_proj", "nn", "return", "self", "up_proj", "x"], "mistral/modeling_mistral.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "mistral/modeling_mistral.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cos", "def", "k", "k_embed", "q", "q_embed", "return", "rotate_half", "sin", "unsqueeze", "unsqueeze_dim"], "mistral/modeling_mistral.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "mistral/modeling_mistral.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value", "value_states"], "mistral/modeling_mistral.py:MistralAttention.__init__": ["False", "Linear", "ModelAttention", "ModelConfig", "Module", "None", "True", "__init__", "attention_dropout", "bias", "class", "config", "def", "getattr", "head_dim", "hidden_size", "int", "is_causal", "k_proj", "layer_idx", "nn", "num_attention_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "or", "q_proj", "scaling", "self", "super", "v_proj"], "mistral/modeling_mistral.py:MistralAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "FlashAttentionKwargs", "LongTensor", "ModelAttention", "Module", "None", "Tensor", "Unpack", "_attn_implementation", "apply_rotary_pos_emb", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "cache_kwargs", "cache_position", "class", "config", "contiguous", "cos", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "getattr", "head_dim", "hidden_shape", "hidden_states", "if", "input_shape", "is", "k_proj", "key_states", "kwargs", "layer_idx", "nn", "not", "o_proj", "past_key_values", "position_embeddings", "q_proj", "query_states", "reshape", "return", "scaling", "self", "shape", "sin", "sliding_window", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "view"], "mistral/modeling_mistral.py:MistralRMSNorm.__init__": ["ModelRMSNorm", "Module", "Parameter", "__init__", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "mistral/modeling_mistral.py:MistralRMSNorm.forward": ["ModelRMSNorm", "Module", "True", "class", "def", "dtype", "float32", "forward", "hidden_states", "input_dtype", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "mistral/modeling_mistral.py:MistralRMSNorm.extra_repr": ["ModelRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "variance_epsilon", "weight"], "mistral/modeling_mistral.py:MistralDecoderLayer.__init__": ["GradientCheckpointingLayer", "ModelAttention", "ModelConfig", "ModelDecoderLayer", "ModelMLP", "ModelRMSNorm", "__init__", "class", "config", "def", "eps", "hidden_size", "input_layernorm", "int", "layer_idx", "mlp", "post_attention_layernorm", "rms_norm_eps", "self", "self_attn", "super"], "mistral/modeling_mistral.py:MistralDecoderLayer.forward": ["Cache", "False", "GradientCheckpointingLayer", "LongTensor", "ModelDecoderLayer", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "attention_mask", "bool", "cache_position", "class", "def", "forward", "hidden_states", "input_layernorm", "kwargs", "mlp", "past_key_values", "position_embeddings", "position_ids", "post_attention_layernorm", "residual", "return", "self", "self_attn", "torch", "tuple", "use_cache"], "mistral/modeling_mistral.py:MistralRotaryEmbedding.__init__": ["Callable", "False", "ModelConfig", "ModelRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "mistral/modeling_mistral.py:MistralRotaryEmbedding.compute_default_rope_parameters": ["ModelConfig", "ModelRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "mistral/modeling_mistral.py:MistralRotaryEmbedding.forward": ["False", "ModelRotaryEmbedding", "Module", "None", "and", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "mistral/modeling_mistral.py:MistralModel.__init__": ["Embedding", "False", "ModelConfig", "ModelDecoderLayer", "ModelModel", "ModelPreTrainedModel", "ModelRMSNorm", "ModelRotaryEmbedding", "ModuleList", "__init__", "class", "config", "def", "embed_tokens", "eps", "for", "gradient_checkpointing", "hidden_size", "in", "layer_idx", "layers", "nn", "norm", "num_hidden_layers", "pad_token_id", "padding_idx", "post_init", "range", "rms_norm_eps", "rotary_emb", "self", "super", "vocab_size"], "mistral/modeling_mistral.py:MistralModel.forward": ["BaseModelOutputWithPast", "Cache", "DynamicCache", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "and", "arange", "attention_mask", "bool", "cache_position", "causal_mask", "class", "config", "create_causal_mask", "create_sliding_window_causal_mask", "decoder_layer", "def", "device", "else", "embed_tokens", "exactly", "for", "forward", "get_seq_length", "hidden_states", "if", "in", "input_embeds", "input_ids", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layers", "mask_function", "must", "norm", "not", "num_hidden_layers", "of", "one", "or", "past_key_values", "past_seen_tokens", "position_embeddings", "position_ids", "raise", "return", "rotary_emb", "self", "shape", "sliding_window", "specify", "torch", "unsqueeze", "use_cache"], "mistral/modeling_mistral.py:MistralForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "ModelForCausalLM", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "vocab_size"], "mistral/modeling_mistral.py:MistralForCausalLM.forward": ["BaseModelOutputWithPast", "Cache", "CausalLMOutputWithPast", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForCausalLM", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "outputs", "past_key_values", "position_ids", "r", "return", "self", "slice", "slice_indices", "torch", "use_cache", "vocab_size"], "mistral3/modeling_mistral3.py:Mistral3RMSNorm.__init__": ["ModelRMSNorm", "Module", "Parameter", "__init__", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "mistral3/modeling_mistral3.py:Mistral3RMSNorm.forward": ["ModelRMSNorm", "Module", "True", "class", "def", "dtype", "float32", "forward", "hidden_states", "input_dtype", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "mistral3/modeling_mistral3.py:Mistral3RMSNorm.extra_repr": ["ModelRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "variance_epsilon", "weight"], "mistral3/modeling_mistral3.py:Mistral3PatchMerger.__init__": ["False", "Linear", "ModelConfig", "ModelPatchMerger", "Module", "__init__", "bias", "class", "config", "def", "hidden_size", "merging_layer", "nn", "patch_size", "self", "spatial_merge_size", "super", "vision_config"], "mistral3/modeling_mistral3.py:Mistral3PatchMerger.forward": ["ModelPatchMerger", "Module", "Tensor", "append", "cat", "class", "d", "def", "dim", "enumerate", "for", "forward", "functional", "grid", "h", "image_features", "image_grid", "image_index", "image_size", "image_sizes", "image_tokens", "in", "kernel_size", "merging_layer", "nn", "patch_size", "permute", "permuted_tensor", "return", "self", "shape", "spatial_merge_size", "split", "stride", "t", "tokens_per_image", "torch", "unfold", "unsqueeze", "view", "w"], "mistral3/modeling_mistral3.py:Mistral3MultiModalProjector.__init__": ["ACT2FN", "Linear", "ModelConfig", "ModelMultiModalProjector", "ModelPatchMerger", "ModelRMSNorm", "Module", "__init__", "act", "bias", "class", "config", "def", "else", "eps", "hidden_size", "if", "int", "isinstance", "len", "linear_1", "linear_2", "multimodal_projector_bias", "nn", "norm", "num_feature_layers", "patch_merger", "projector_hidden_act", "rms_norm_eps", "self", "super", "text_config", "vision_config", "vision_feature_layer"], "mistral3/modeling_mistral3.py:Mistral3MultiModalProjector.forward": ["ModelMultiModalProjector", "Module", "Tensor", "act", "class", "def", "forward", "hidden_states", "image_features", "image_sizes", "linear_1", "linear_2", "nn", "norm", "patch_merger", "return", "self", "torch"], "mistral3/modeling_mistral3.py:Mistral3Model.__init__": ["AutoModel", "ModelConfig", "ModelModel", "ModelMultiModalProjector", "ModelPreTrainedModel", "__init__", "class", "config", "def", "from_config", "language_model", "multi_modal_projector", "post_init", "self", "super", "text_config", "vision_config", "vision_tower"], "mistral3/modeling_mistral3.py:Mistral3Model.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "language_model", "return", "self"], "mistral3/modeling_mistral3.py:Mistral3Model.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "language_model", "self", "set_input_embeddings", "value"], "mistral3/modeling_mistral3.py:Mistral3Model.get_image_features": ["FloatTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "True", "as_tensor", "cat", "class", "config", "def", "device", "dim", "downsample_ratio", "else", "for", "get_image_features", "hidden_states", "hs_pool", "if", "image_features", "image_outputs", "image_sizes", "in", "int", "is", "isinstance", "items", "k", "kwargs", "layer_idx", "list", "multi_modal_projector", "not", "output_hidden_states", "patch_size", "pixel_values", "prod", "return", "selected_image_feature", "self", "spatial_merge_size", "split", "split_sizes", "squeeze", "tolist", "torch", "v", "vision_feature_layer", "vision_tower"], "mistral3/modeling_mistral3.py:Mistral3Model.get_placeholder_mask": ["FloatTensor", "Image", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "ValueError", "all", "and", "class", "config", "def", "device", "do", "dtype", "else", "expand_as", "f", "features", "get_input_embeddings", "get_placeholder_mask", "if", "image", "image_features", "image_token_id", "input_ids", "inputs_embeds", "is", "long", "match", "n_image_features", "n_image_tokens", "not", "numel", "raise", "return", "self", "shape", "special_image_mask", "sum", "tensor", "to", "tokens", "torch", "unsqueeze"], "mistral3/modeling_mistral3.py:Mistral3Model.forward": ["Cache", "FloatTensor", "LongTensor", "ModelModel", "ModelModelOutputWithPast", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "ValueError", "You", "attention_mask", "attentions", "bool", "cache_position", "cat", "class", "config", "def", "device", "dim", "dtype", "else", "exactly", "forward", "get_image_features", "get_input_embeddings", "get_placeholder_mask", "hidden_states", "if", "image_features", "image_hidden_states", "image_sizes", "input_ids", "inputs_embeds", "int", "is", "kwargs", "language_model", "last_hidden_state", "list", "masked_scatter", "must", "not", "of", "one", "or", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "pixel_values", "position_ids", "raise", "return", "return_dict", "self", "special_image_mask", "specify", "to", "torch", "tuple", "use_cache", "use_return_dict", "vision_feature_layer"], "mistral3/modeling_mistral3.py:Mistral3ForConditionalGeneration.__init__": ["False", "GenerationMixin", "Linear", "ModelConfig", "ModelForConditionalGeneration", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "text_config", "vocab_size"], "mistral3/modeling_mistral3.py:Mistral3ForConditionalGeneration.get_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "model", "return", "self"], "mistral3/modeling_mistral3.py:Mistral3ForConditionalGeneration.set_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "model", "self", "set_input_embeddings", "value"], "mistral3/modeling_mistral3.py:Mistral3ForConditionalGeneration.get_output_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "Module", "class", "def", "get_output_embeddings", "lm_head", "nn", "return", "self"], "mistral3/modeling_mistral3.py:Mistral3ForConditionalGeneration.get_image_features": ["FloatTensor", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Tensor", "class", "def", "get_image_features", "image_sizes", "int", "kwargs", "list", "model", "pixel_values", "return", "self", "torch", "vision_feature_layer"], "mistral3/modeling_mistral3.py:Mistral3ForConditionalGeneration.forward": ["Cache", "FloatTensor", "GenerationMixin", "LongTensor", "ModelCausalLMOutputWithPast", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "else", "forward", "hidden_states", "if", "image_hidden_states", "image_sizes", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "pixel_values", "position_ids", "r", "return", "return_dict", "self", "slice", "slice_indices", "text_config", "torch", "tuple", "use_cache", "use_return_dict", "vocab_size"], "mistral3/modeling_mistral3.py:Mistral3ForConditionalGeneration.prepare_inputs_for_generation": ["False", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "True", "attention_mask", "cache_position", "class", "def", "get", "if", "input_ids", "inputs_embeds", "is_first_iteration", "kwargs", "logits_to_keep", "model_inputs", "not", "or", "past_key_values", "pixel_values", "prepare_inputs_for_generation", "return", "self", "super", "use_cache"], "mixtral/modeling_mixtral.py:MixtralExperts.__init__": ["ACT2FN", "ModelConfig", "ModelExperts", "Module", "Parameter", "__init__", "act_fn", "class", "config", "def", "down_proj", "empty", "gate_up_proj", "hidden_act", "hidden_dim", "hidden_size", "intermediate_dim", "intermediate_size", "nn", "num_experts", "num_local_experts", "self", "super", "torch"], "mixtral/modeling_mixtral.py:MixtralExperts.forward": ["ModelExperts", "Module", "None", "Tensor", "act_fn", "chunk", "class", "continue", "current_hidden_states", "current_state", "def", "dim", "down_proj", "dtype", "expert_hit", "expert_idx", "expert_mask", "final_hidden_states", "for", "forward", "functional", "gate", "gate_up_proj", "greater", "hidden_states", "if", "in", "index_add_", "linear", "nn", "no_grad", "nonzero", "num_classes", "num_experts", "one_hot", "permute", "return", "self", "sum", "to", "token_idx", "top_k_index", "top_k_pos", "top_k_weights", "torch", "up", "where", "with", "zeros_like"], "mixtral/modeling_mixtral.py:MixtralTopKRouter.__init__": ["ModelTopKRouter", "Module", "Parameter", "__init__", "class", "config", "def", "empty", "hidden_dim", "hidden_size", "nn", "num_experts", "num_experts_per_tok", "num_local_experts", "self", "super", "top_k", "torch", "weight"], "mixtral/modeling_mixtral.py:MixtralTopKRouter.forward": ["F", "ModelTopKRouter", "Module", "True", "class", "def", "dim", "float", "forward", "functional", "hidden_dim", "hidden_states", "keepdim", "linear", "nn", "reshape", "return", "router_indices", "router_logits", "router_scores", "router_top_value", "self", "softmax", "sum", "top_k", "topk", "torch", "weight"], "mixtral/modeling_mixtral.py:MixtralSparseMoeBlock.__init__": ["ModelExperts", "ModelSparseMoeBlock", "ModelTopKRouter", "Module", "__init__", "class", "config", "def", "experts", "gate", "jitter_noise", "nn", "num_experts_per_tok", "router_jitter_noise", "self", "super", "top_k"], "mixtral/modeling_mixtral.py:MixtralSparseMoeBlock.forward": ["ModelSparseMoeBlock", "Module", "Tensor", "_", "and", "batch_size", "class", "def", "empty_like", "experts", "forward", "gate", "hidden_dim", "hidden_states", "if", "jitter_noise", "nn", "reshape", "return", "self", "sequence_length", "shape", "top_k_index", "top_k_weights", "torch", "training", "tuple", "uniform_", "view"], "mixtral/modeling_mixtral.py:MixtralRMSNorm.__init__": ["ModelRMSNorm", "Module", "Parameter", "__init__", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "mixtral/modeling_mixtral.py:MixtralRMSNorm.forward": ["ModelRMSNorm", "Module", "True", "class", "def", "dtype", "float32", "forward", "hidden_states", "input_dtype", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "mixtral/modeling_mixtral.py:MixtralRMSNorm.extra_repr": ["ModelRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "variance_epsilon", "weight"], "mixtral/modeling_mixtral.py:MixtralRotaryEmbedding.__init__": ["Callable", "False", "ModelConfig", "ModelRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "mixtral/modeling_mixtral.py:MixtralRotaryEmbedding.compute_default_rope_parameters": ["ModelConfig", "ModelRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "mixtral/modeling_mixtral.py:MixtralRotaryEmbedding.forward": ["False", "ModelRotaryEmbedding", "Module", "None", "and", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "mixtral/modeling_mixtral.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "mixtral/modeling_mixtral.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cos", "def", "k", "k_embed", "q", "q_embed", "return", "rotate_half", "sin", "unsqueeze", "unsqueeze_dim"], "mixtral/modeling_mixtral.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "mixtral/modeling_mixtral.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value", "value_states"], "mixtral/modeling_mixtral.py:MixtralAttention.__init__": ["False", "Linear", "ModelAttention", "ModelConfig", "Module", "None", "True", "__init__", "attention_dropout", "bias", "class", "config", "def", "getattr", "head_dim", "hidden_size", "int", "is_causal", "k_proj", "layer_idx", "nn", "num_attention_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "or", "q_proj", "scaling", "self", "super", "v_proj"], "mixtral/modeling_mixtral.py:MixtralAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "FlashAttentionKwargs", "LongTensor", "ModelAttention", "Module", "None", "Tensor", "Unpack", "_attn_implementation", "apply_rotary_pos_emb", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "cache_kwargs", "cache_position", "class", "config", "contiguous", "cos", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "getattr", "head_dim", "hidden_shape", "hidden_states", "if", "input_shape", "is", "k_proj", "key_states", "kwargs", "layer_idx", "nn", "not", "o_proj", "past_key_values", "position_embeddings", "q_proj", "query_states", "reshape", "return", "scaling", "self", "shape", "sin", "sliding_window", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "view"], "mixtral/modeling_mixtral.py:MixtralDecoderLayer.__init__": ["GradientCheckpointingLayer", "ModelAttention", "ModelConfig", "ModelDecoderLayer", "ModelRMSNorm", "ModelSparseMoeBlock", "__init__", "class", "config", "def", "eps", "hidden_size", "input_layernorm", "int", "layer_idx", "mlp", "post_attention_layernorm", "rms_norm_eps", "self", "self_attn", "super"], "mixtral/modeling_mixtral.py:MixtralDecoderLayer.forward": ["Cache", "GradientCheckpointingLayer", "LongTensor", "ModelDecoderLayer", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "attention_mask", "cache_position", "class", "def", "forward", "hidden_states", "input_layernorm", "kwargs", "mlp", "past_key_values", "position_embeddings", "position_ids", "post_attention_layernorm", "residual", "return", "self", "self_attn", "torch", "tuple"], "mixtral/modeling_mixtral.py:MixtralPreTrainedModel._init_weights": ["ModelExperts", "ModelPreTrainedModel", "ModelTopKRouter", "PreTrainedModel", "_init_weights", "class", "config", "def", "down_proj", "elif", "gate_up_proj", "if", "init", "initializer_range", "isinstance", "mean", "module", "normal_", "self", "std", "super", "weight"], "mixtral/modeling_mixtral.py:MixtralModel.__init__": ["Embedding", "False", "ModelConfig", "ModelDecoderLayer", "ModelModel", "ModelPreTrainedModel", "ModelRMSNorm", "ModelRotaryEmbedding", "ModuleList", "__init__", "class", "config", "def", "embed_tokens", "eps", "for", "gradient_checkpointing", "hidden_size", "in", "layer_idx", "layers", "nn", "norm", "num_hidden_layers", "pad_token_id", "padding_idx", "post_init", "range", "rms_norm_eps", "rotary_emb", "self", "super", "vocab_size"], "mixtral/modeling_mixtral.py:MixtralModel.forward": ["Cache", "DynamicCache", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "MoeModelOutputWithPast", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "and", "arange", "attention_mask", "bool", "cache_position", "causal_mask", "class", "config", "create_causal_mask", "create_sliding_window_causal_mask", "decoder_layer", "def", "device", "else", "embed_tokens", "exactly", "for", "forward", "get_seq_length", "hidden_states", "if", "in", "input_embeds", "input_ids", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layers", "mask_function", "must", "norm", "not", "num_hidden_layers", "of", "one", "or", "past_key_values", "past_seen_tokens", "position_embeddings", "position_ids", "raise", "return", "rotary_emb", "self", "shape", "sliding_window", "specify", "torch", "unsqueeze", "use_cache"], "mixtral/modeling_mixtral.py:load_balancing_loss_func": ["Model_balancing_loss_func", "None", "Tensor", "_", "attention_mask", "batch_size", "cat", "compute_device", "concatenated_gate_logits", "def", "device", "dim", "else", "expand", "expert_attention_mask", "expert_mask", "float", "for", "functional", "gate_logits", "if", "in", "int", "is", "isinstance", "layer_gate", "mean", "nn", "not", "num_experts", "num_hidden_layers", "one_hot", "or", "overall_loss", "r", "reshape", "return", "router_per_expert_attention_mask", "router_prob_per_expert", "routing_weights", "selected_experts", "sequence_length", "shape", "softmax", "sum", "to", "tokens_per_expert", "top_k", "topk", "torch", "tuple", "unsqueeze"], "mixtral/modeling_mixtral.py:MixtralForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "ModelForCausalLM", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "num_experts", "num_experts_per_tok", "num_local_experts", "post_init", "router_aux_loss_coef", "self", "super", "vocab_size"], "mixtral/modeling_mixtral.py:MixtralForCausalLM.forward": ["Cache", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForCausalLM", "ModelPreTrainedModel", "MoeCausalLMOutputWithPast", "MoeModelOutputWithPast", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "aux_loss", "bool", "cache_position", "class", "config", "def", "device", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "load_balancing_loss_func", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "num_experts", "num_experts_per_tok", "output_router_logits", "outputs", "past_key_values", "position_ids", "r", "return", "router_aux_loss_coef", "router_logits", "self", "slice", "slice_indices", "to", "torch", "use_cache", "vocab_size"], "mlcd/modeling_mlcd.py:MLCDMLP.__init__": ["ACT2FN", "Linear", "Model", "Module", "__init__", "activation_fn", "class", "config", "def", "fc1", "fc2", "hidden_act", "hidden_size", "intermediate_size", "nn", "self", "super"], "mlcd/modeling_mlcd.py:MLCDMLP.forward": ["Model", "Module", "Tensor", "activation_fn", "class", "def", "fc1", "fc2", "forward", "hidden_states", "nn", "return", "self", "torch"], "mlcd/modeling_mlcd.py:MLCDRotaryEmbedding.__init__": ["False", "Model", "Module", "None", "__init__", "arange", "class", "def", "dim", "dtype", "float", "int", "inv_freq", "nn", "persistent", "register_buffer", "self", "super", "theta", "torch"], "mlcd/modeling_mlcd.py:MLCDRotaryEmbedding.forward": ["Model", "Module", "Tensor", "arange", "class", "def", "device", "dim", "dtype", "expand", "flatten", "forward", "hpos_ids", "int", "inv_freq", "max", "max_grid_size", "nn", "num_patches_height", "num_patches_width", "outer", "pos_ids", "return", "rotary_pos_emb", "rotary_pos_emb_full", "self", "seq", "stack", "torch", "unsqueeze", "wpos_ids"], "mlcd/modeling_mlcd.py:MLCDVisionEmbeddings.__init__": ["Conv2d", "False", "Model", "ModelVisionConfig", "Module", "Parameter", "__init__", "arange", "bias", "class", "class_embedding", "config", "def", "embed_dim", "expand", "hidden_size", "image_size", "in_channels", "kernel_size", "nn", "num_channels", "num_patches", "num_positions", "out_channels", "patch_embedding", "patch_size", "persistent", "position_ids", "randn", "register_buffer", "self", "stride", "super", "torch"], "mlcd/modeling_mlcd.py:MLCDVisionEmbeddings.interpolate_pos_encoding": ["False", "Model", "Module", "Tensor", "align_corners", "and", "bicubic", "cat", "class", "class_pos_embed", "def", "dim", "embeddings", "functional", "height", "if", "int", "interpolate", "interpolate_pos_encoding", "is_tracing", "jit", "mode", "new_height", "new_width", "nn", "not", "num_patches", "num_positions", "patch_pos_embed", "patch_size", "permute", "position_embedding", "position_ids", "reshape", "return", "self", "shape", "size", "sqrt_num_positions", "torch", "torch_int", "unsqueeze", "view", "weight", "width"], "mlcd/modeling_mlcd.py:MLCDVisionEmbeddings.forward": ["FloatTensor", "Model", "Module", "Tensor", "batch_size", "cat", "class", "class_embedding", "class_embeds", "def", "dim", "dtype", "embeddings", "expand", "flatten", "forward", "nn", "patch_embedding", "patch_embeds", "pixel_values", "return", "self", "shape", "target_dtype", "to", "torch", "transpose", "weight"], "mlcd/modeling_mlcd.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value", "value_states"], "mlcd/modeling_mlcd.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "mlcd/modeling_mlcd.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "mlcd/modeling_mlcd.py:apply_rotary_pos_emb_vision": ["Model_rotary_pos_emb_vision", "Tensor", "cos", "def", "dtype", "float", "k", "k_embed", "orig_k_dtype", "orig_q_dtype", "q", "q_embed", "return", "rotate_half", "sin", "to", "torch", "tuple", "unsqueeze"], "mlcd/modeling_mlcd.py:MLCDAttention.__init__": ["False", "Linear", "Model", "ModelVisionConfig", "Module", "ValueError", "__init__", "and", "attention_dropout", "be", "by", "class", "config", "def", "divisible", "dropout", "embed_dim", "f", "got", "head_dim", "hidden_size", "if", "is_causal", "k_proj", "must", "nn", "num_attention_heads", "num_heads", "num_key_value_groups", "out_proj", "q_proj", "raise", "scale", "self", "super", "v_proj"], "mlcd/modeling_mlcd.py:MLCDAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "Model", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_attn_implementation", "apply_rotary_pos_emb_vision", "attention_interface", "attention_mask", "attn_output", "attn_weights", "batch_size", "class", "config", "contiguous", "cos", "def", "dropout", "eager", "eager_attention_forward", "else", "float", "forward", "head_dim", "hidden_states", "if", "is_causal", "k_proj", "key_states", "kwargs", "nn", "not", "num_heads", "out_proj", "permute", "position_embeddings", "q_proj", "query_states", "reshape", "return", "scale", "scaling", "self", "seq_length", "shape", "sin", "torch", "training", "tuple", "unsqueeze", "v_proj", "value_states", "view"], "mlcd/modeling_mlcd.py:MLCDEncoderLayer.__init__": ["GradientCheckpointingLayer", "LayerNorm", "Model", "ModelAttention", "ModelMLP", "ModelVisionConfig", "__init__", "class", "config", "def", "embed_dim", "eps", "hidden_size", "layer_norm1", "layer_norm2", "layer_norm_eps", "mlp", "nn", "self", "self_attn", "super"], "mlcd/modeling_mlcd.py:MLCDEncoderLayer.forward": ["FloatTensor", "GradientCheckpointingLayer", "Model", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "attention_mask", "class", "def", "forward", "hidden_states", "kwargs", "layer_norm1", "layer_norm2", "mlp", "position_embeddings", "residual", "return", "self", "self_attn", "torch", "tuple"], "mlcd/modeling_mlcd.py:MLCDEncoder.__init__": ["False", "Model", "ModelLayer", "ModelVisionConfig", "Module", "ModuleList", "_", "__init__", "class", "config", "def", "for", "gradient_checkpointing", "in", "layers", "nn", "num_hidden_layers", "range", "self", "super"], "mlcd/modeling_mlcd.py:MLCDEncoder.forward": ["BaseModelOutput", "FloatTensor", "Model", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "class", "def", "encoder_layer", "for", "forward", "hidden_states", "in", "inputs_embeds", "kwargs", "last_hidden_state", "layers", "nn", "position_embeddings", "r", "return", "self", "torch", "tuple"], "mlcd/modeling_mlcd.py:MLCDPreTrainedModel._init_weights": ["LayerNorm", "Linear", "Model", "ModelAttention", "ModelMLP", "ModelRotaryEmbedding", "ModelVisionEmbeddings", "ModelVisionTransformer", "None", "PreTrainedModel", "_init_weights", "and", "arange", "bias", "class", "class_embedding", "class_pos_emb", "config", "copy_", "def", "dim", "dtype", "elif", "embed_dim", "expand", "factor", "fc1", "fc2", "fc_std", "float", "hidden_size", "if", "in_proj_std", "init", "initializer_factor", "initializer_range", "inv_freq", "is", "isinstance", "k_proj", "mean", "module", "nn", "normal_", "not", "num_attention_heads", "num_hidden_layers", "ones_", "out_proj", "out_proj_std", "patch_embedding", "pos_emb_std", "position_ids", "q_proj", "self", "shape", "std", "theta", "torch", "v_proj", "weight", "zeros_"], "mlcd/modeling_mlcd.py:MLCDVisionTransformer.__init__": ["LayerNorm", "Model", "ModelEncoder", "ModelRotaryEmbedding", "ModelVisionConfig", "ModelVisionEmbeddings", "Module", "Parameter", "__init__", "class", "class_pos_emb", "config", "def", "embed_dim", "embeddings", "encoder", "eps", "hidden_size", "layer_norm_eps", "nn", "num_attention_heads", "post_layernorm", "pre_layrnorm", "randn", "self", "super", "torch", "vision_rotary_embedding"], "mlcd/modeling_mlcd.py:MLCDVisionTransformer.forward": ["BaseModelOutputWithPooling", "FloatTensor", "Model", "Module", "None", "TransformersKwargs", "Unpack", "ValueError", "You", "cat", "class", "class_pos_emb", "config", "cos", "def", "device", "dim", "emb", "embeddings", "encoder", "encoder_outputs", "forward", "have", "hidden_states", "if", "inputs_embeds", "is", "kwargs", "last_hidden_state", "nn", "num_patches_height", "num_patches_width", "patch_size", "pixel_values", "pooled_output", "pooler_output", "position_embeddings", "post_layernorm", "pre_layrnorm", "raise", "return", "rotary_pos_emb", "self", "shape", "sin", "specify", "to", "torch", "tuple", "vision_rotary_embedding"], "mlcd/modeling_mlcd.py:MLCDVisionModel.__init__": ["Model", "ModelPreTrainedModel", "ModelVisionConfig", "ModelVisionTransformer", "__init__", "class", "config", "def", "post_init", "self", "super", "vision_model"], "mlcd/modeling_mlcd.py:MLCDVisionModel.get_input_embeddings": ["Model", "ModelPreTrainedModel", "Module", "class", "def", "embeddings", "get_input_embeddings", "nn", "patch_embedding", "return", "self", "vision_model"], "mlcd/modeling_mlcd.py:MLCDVisionModel.forward": ["BaseModelOutputWithPooling", "FloatTensor", "Model", "ModelPreTrainedModel", "None", "TransformersKwargs", "Unpack", "class", "def", "forward", "kwargs", "pixel_values", "r", "return", "self", "torch", "tuple", "vision_model"], "mllama/modeling_mllama.py:_prepare_cross_attention_mask": ["None", "Tensor", "_", "_prepare_cross_attention_mask", "any", "batch_size", "bool", "cross_attention_mask", "def", "dim", "dtype", "finfo", "full_text_row_masked_out_mask", "int", "inverted_cross_attn_mask", "masked_fill", "min", "negative_inf_value", "num_vision_tokens", "repeat_interleave", "return", "shape", "str", "text_total_length", "to", "torch", "tuple", "type_as", "unsqueeze", "view"], "mllama/modeling_mllama.py:_prepare_aspect_ratio_attention_mask": ["Tensor", "_prepare_aspect_ratio_attention_mask", "aspect_ratio_mask", "attention_mask", "batch_size", "def", "dtype", "finfo", "int", "max_num_tiles", "min", "num_patches", "pad_patches", "repeat", "reshape", "return", "shape", "target_length", "to", "torch", "transpose", "unsqueeze", "view"], "mllama/modeling_mllama.py:MllamaPrecomputedAspectRatioEmbedding.__init__": ["Embedding", "ModelPrecomputedAspectRatioEmbedding", "ModelVisionConfig", "Module", "Parameter", "True", "__init__", "bool", "class", "config", "def", "embedding", "gate", "hidden_size", "if", "is_gated", "max_aspect_ratio_id", "max_num_tiles", "nn", "self", "super", "torch", "zeros"], "mllama/modeling_mllama.py:MllamaPrecomputedAspectRatioEmbedding.forward": ["ModelPrecomputedAspectRatioEmbedding", "Module", "Tensor", "aspect_ratio_ids", "class", "def", "embedding", "embeddings", "forward", "gate", "hidden_size", "hidden_state", "if", "is_gated", "max_num_tiles", "nn", "reshape", "return", "self", "tanh", "torch"], "mllama/modeling_mllama.py:MllamaPrecomputedPositionEmbedding.__init__": ["Embedding", "ModelPrecomputedPositionEmbedding", "ModelVisionConfig", "Module", "Parameter", "__init__", "class", "config", "def", "embedding", "gate", "hidden_size", "image_size", "max_aspect_ratio_id", "max_num_tiles", "nn", "num_patches", "patch_size", "position_embedding", "randn", "scale", "self", "super", "tile_embedding", "torch", "zeros"], "mllama/modeling_mllama.py:MllamaPrecomputedPositionEmbedding.forward": ["ModelPrecomputedPositionEmbedding", "Module", "Tensor", "aspect_ratio_ids", "batch_size", "class", "def", "embedding", "forward", "gate", "gated_position_embedding", "gated_tile_position_embedding", "hidden_size", "hidden_state", "max_num_tiles", "nn", "num_patches", "reshape", "return", "self", "shape", "tanh", "tile_embedding", "tile_position_embedding", "torch", "view"], "mllama/modeling_mllama.py:MllamaVisionMLP.__init__": ["ACT2FN", "Linear", "ModelVisionMLP", "Module", "__init__", "activation_fn", "class", "config", "def", "fc1", "fc2", "hidden_act", "hidden_size", "intermediate_size", "nn", "self", "super"], "mllama/modeling_mllama.py:MllamaVisionMLP.forward": ["ModelVisionMLP", "Module", "Tensor", "activation_fn", "class", "def", "fc1", "fc2", "forward", "hidden_states", "nn", "return", "self", "torch"], "mllama/modeling_mllama.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "mllama/modeling_mllama.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value", "value_states"], "mllama/modeling_mllama.py:MllamaVisionAttention.__init__": ["False", "Linear", "ModelVisionAttention", "ModelVisionConfig", "Module", "__init__", "attention_heads", "bias", "class", "config", "def", "embed_dim", "head_dim", "hidden_size", "k_proj", "nn", "num_heads", "num_key_value_groups", "o_proj", "q_proj", "scaling", "self", "super", "v_proj"], "mllama/modeling_mllama.py:MllamaVisionAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "ModelVisionAttention", "Module", "None", "Tensor", "_", "_attn_implementation", "attention_interface", "attention_mask", "attn_output", "attn_weights", "batch_size", "class", "config", "contiguous", "def", "dropout", "eager", "eager_attention_forward", "forward", "head_dim", "hidden_state", "if", "k_proj", "key", "kv_seq_len", "kwargs", "nn", "num_heads", "o_proj", "q_proj", "q_seq_len", "query", "reshape", "return", "scaling", "self", "shape", "torch", "transpose", "tuple", "v_proj", "value", "view"], "mllama/modeling_mllama.py:MllamaVisionEncoderLayer.__init__": ["False", "LayerNorm", "ModelVisionAttention", "ModelVisionConfig", "ModelVisionEncoderLayer", "ModelVisionMLP", "Module", "Parameter", "__init__", "attention_heads", "bool", "class", "config", "def", "eps", "gate_attn", "gate_ffn", "hidden_size", "if", "input_layernorm", "intermediate_size", "is_gated", "math", "mlp", "nn", "norm_eps", "num_attention_heads", "ones", "pi", "post_attention_layernorm", "self", "self_attn", "super", "torch"], "mllama/modeling_mllama.py:MllamaVisionEncoderLayer.forward": ["ModelVisionEncoderLayer", "Module", "None", "Tensor", "attention_mask", "attn_weights", "class", "def", "forward", "gate_attn", "gate_ffn", "hidden_state", "if", "input_layernorm", "is_gated", "mlp", "nn", "post_attention_layernorm", "residual", "return", "self", "self_attn", "tanh", "torch"], "mllama/modeling_mllama.py:MllamaVisionEncoder.__init__": ["False", "ModelVisionConfig", "ModelVisionEncoder", "ModelVisionEncoderLayer", "Module", "ModuleList", "_", "__init__", "class", "config", "def", "for", "gradient_checkpointing", "in", "is_gated", "layers", "nn", "num_layers", "range", "self", "super"], "mllama/modeling_mllama.py:MllamaVisionEncoder.forward": ["BaseModelOutput", "ModelVisionEncoder", "Module", "None", "Tensor", "attention_mask", "class", "def", "encoder_layer", "encoder_states", "for", "forward", "hidden_state", "hidden_states", "in", "last_hidden_state", "layers", "nn", "r", "return", "self", "torch"], "mllama/modeling_mllama.py:MllamaTextRMSNorm.__init__": ["ModelTextRMSNorm", "Module", "Parameter", "__init__", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "mllama/modeling_mllama.py:MllamaTextRMSNorm.forward": ["ModelTextRMSNorm", "Module", "True", "class", "def", "dtype", "float32", "forward", "hidden_states", "input_dtype", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "mllama/modeling_mllama.py:MllamaTextRMSNorm.extra_repr": ["ModelTextRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "variance_epsilon", "weight"], "mllama/modeling_mllama.py:MllamaTextCrossAttention.__init__": ["False", "Linear", "ModelTextConfig", "ModelTextCrossAttention", "ModelTextRMSNorm", "Module", "None", "__init__", "bias", "class", "config", "def", "dropout", "eps", "head_dim", "hidden_size", "int", "k_norm", "k_proj", "layer_idx", "nn", "num_attention_heads", "num_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "q_norm", "q_proj", "rms_norm_eps", "scaling", "self", "super", "v_proj"], "mllama/modeling_mllama.py:MllamaTextCrossAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "Cross", "LongTensor", "ModelTextCrossAttention", "Module", "None", "Tensor", "ValueError", "_", "_attn_implementation", "attention", "attention_interface", "attention_mask", "attn_output", "attn_weights", "bool", "bsz", "cache_position", "cached", "can", "class", "config", "contiguous", "cross_attention_states", "cross_attn_states", "def", "dropout", "eager", "eager_attention_forward", "elif", "else", "find", "for", "forward", "head_dim", "hidden_states", "if", "is", "k_norm", "k_proj", "key", "key_states", "keys", "kwargs", "layer", "layer_idx", "layers", "neither", "nn", "nor", "not", "num_heads", "num_key_value_heads", "o_proj", "past_key_values", "q_len", "q_norm", "q_proj", "query_states", "raise", "reshape", "return", "scaling", "self", "size", "t", "torch", "training", "transpose", "tuple", "update", "use_cache", "v_proj", "value_states", "values", "view"], "mllama/modeling_mllama.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "mllama/modeling_mllama.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cos", "def", "k", "k_embed", "q", "q_embed", "return", "rotate_half", "sin", "unsqueeze", "unsqueeze_dim"], "mllama/modeling_mllama.py:MllamaTextSelfAttention.__init__": ["False", "Linear", "ModelTextConfig", "ModelTextSelfAttention", "Module", "True", "__init__", "bias", "class", "config", "def", "dropout", "head_dim", "hidden_size", "int", "is_causal", "k_proj", "layer_idx", "nn", "num_attention_heads", "num_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "q_proj", "scaling", "self", "super", "v_proj"], "mllama/modeling_mllama.py:MllamaTextSelfAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "ModelTextSelfAttention", "Module", "None", "Tensor", "_", "_attn_implementation", "apply_rotary_pos_emb", "attention_interface", "attention_mask", "attn_output", "attn_weights", "bsz", "cache_kwargs", "cache_position", "class", "config", "contiguous", "cos", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "head_dim", "hidden_states", "if", "is", "k_proj", "key_states", "kwargs", "layer_idx", "nn", "not", "num_heads", "num_key_value_heads", "o_proj", "past_key_values", "position_embeddings", "q_len", "q_proj", "query_states", "reshape", "return", "scaling", "self", "sin", "size", "torch", "training", "transpose", "update", "v_proj", "value_states", "view"], "mllama/modeling_mllama.py:MllamaTextMLP.__init__": ["ACT2FN", "False", "Linear", "ModelTextMLP", "Module", "__init__", "act_fn", "bias", "class", "config", "def", "down_proj", "gate_proj", "hidden_act", "hidden_size", "intermediate_size", "nn", "self", "super", "up_proj"], "mllama/modeling_mllama.py:MllamaTextMLP.forward": ["ModelTextMLP", "Module", "act_fn", "class", "def", "down_proj", "forward", "gate_proj", "nn", "return", "self", "up_proj", "x"], "mllama/modeling_mllama.py:MllamaSelfAttentionDecoderLayer.__init__": ["GradientCheckpointingLayer", "ModelSelfAttentionDecoderLayer", "ModelTextConfig", "ModelTextMLP", "ModelTextRMSNorm", "ModelTextSelfAttention", "__init__", "class", "config", "def", "eps", "hidden_size", "input_layernorm", "int", "layer_idx", "mlp", "post_attention_layernorm", "rms_norm_eps", "self", "self_attn", "super"], "mllama/modeling_mllama.py:MllamaSelfAttentionDecoderLayer.forward": ["Cache", "False", "FlashAttentionKwargs", "FloatTensor", "GradientCheckpointingLayer", "LongTensor", "ModelSelfAttentionDecoderLayer", "None", "Tensor", "Unpack", "attention_mask", "bool", "cache_position", "class", "cross_attention_mask", "cross_attention_states", "def", "forward", "full_text_row_masked_out_mask", "hidden_states", "input_layernorm", "kwargs", "mlp", "past_key_values", "position_embeddings", "position_ids", "post_attention_layernorm", "residual", "return", "self", "self_attn", "self_attn_weights", "torch", "tuple", "use_cache"], "mllama/modeling_mllama.py:MllamaCrossAttentionDecoderLayer.__init__": ["GradientCheckpointingLayer", "ModelCrossAttentionDecoderLayer", "ModelTextConfig", "ModelTextCrossAttention", "ModelTextMLP", "ModelTextRMSNorm", "None", "Parameter", "__init__", "class", "config", "cross_attn", "cross_attn_attn_gate", "cross_attn_mlp_gate", "def", "eps", "hidden_size", "input_layernorm", "int", "layer_idx", "mlp", "nn", "post_attention_layernorm", "rms_norm_eps", "self", "super", "torch", "zeros"], "mllama/modeling_mllama.py:MllamaCrossAttentionDecoderLayer.forward": ["Cache", "False", "FlashAttentionKwargs", "GradientCheckpointingLayer", "LongTensor", "ModelCrossAttentionDecoderLayer", "None", "Tensor", "Unpack", "attention_mask", "attn_weights", "bool", "cache_position", "class", "cross_attention_mask", "cross_attention_states", "cross_attn", "cross_attn_attn_gate", "cross_attn_mlp_gate", "def", "forward", "full_text_row_masked_out_mask", "hidden_states", "if", "input_layernorm", "is", "kwargs", "mlp", "not", "past_key_values", "position_embeddings", "position_ids", "post_attention_layernorm", "residual", "return", "self", "tanh", "torch", "tuple", "use_cache"], "mllama/modeling_mllama.py:MllamaRotaryEmbedding.__init__": ["Callable", "False", "ModelRotaryEmbedding", "ModelTextConfig", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "mllama/modeling_mllama.py:MllamaRotaryEmbedding.compute_default_rope_parameters": ["ModelRotaryEmbedding", "ModelTextConfig", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "mllama/modeling_mllama.py:MllamaRotaryEmbedding.forward": ["False", "ModelRotaryEmbedding", "Module", "None", "and", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "mllama/modeling_mllama.py:MllamaPreTrainedModel._init_weights": ["Conv2d", "Embedding", "False", "LayerNorm", "Linear", "ModelCrossAttentionDecoderLayer", "ModelPreTrainedModel", "ModelPrecomputedAspectRatioEmbedding", "ModelPrecomputedPositionEmbedding", "ModelRotaryEmbedding", "ModelTextRMSNorm", "ModelVisionEncoderLayer", "ModelVisionModel", "None", "PreTrainedModel", "ROPE_INIT_FUNCTIONS", "_", "_init_weights", "_is_hf_initialized", "and", "bias", "buffer_value", "class", "class_embedding", "compute_default_rope_parameters", "config", "copy_", "cross_attn_attn_gate", "cross_attn_mlp_gate", "def", "default", "elif", "else", "embedding", "gate", "gate_attn", "gate_ffn", "get_text_config", "getattr", "if", "init", "initializer_range", "inv_freq", "is", "is_gated", "isinstance", "mean", "module", "nn", "normal_", "not", "ones_", "original_inv_freq", "padding_idx", "rope_fn", "rope_type", "self", "std", "weight", "zeros_"], "mllama/modeling_mllama.py:MllamaPreTrainedModel._update_causal_mask": ["AttentionMaskConverter", "BlockMask", "Cache", "False", "ModelPreTrainedModel", "None", "PreTrainedModel", "Tensor", "Union", "_attn_implementation", "_ignore_causal_mask_sdpa", "_prepare_4d_causal_attention_mask_with_cache_position", "_unmask_unattended", "_update_causal_mask", "and", "any", "attention_mask", "batch_size", "bool", "cache_position", "causal_mask", "class", "config", "cuda", "def", "device", "dtype", "else", "finfo", "flash_attention_2", "flex_attention", "get_max_cache_shape", "get_seq_length", "if", "in", "input_tensor", "inputs_embeds", "is", "is_compileable", "is_training", "isinstance", "make_flex_block_causal_mask", "min", "min_dtype", "not", "npu", "output_attentions", "past_key_values", "past_key_values_length", "past_seen_tokens", "return", "sdpa", "self", "sequence_length", "shape", "target_length", "torch", "training", "type", "using_compilable_cache", "xpu"], "mllama/modeling_mllama.py:MllamaPreTrainedModel._prepare_4d_causal_attention_mask_with_cache_position": ["ModelPreTrainedModel", "None", "PreTrainedModel", "Tensor", "_prepare_4d_causal_attention_mask_with_cache_position", "and", "arange", "attention_mask", "batch_size", "cache_position", "causal_mask", "class", "clone", "def", "device", "diagonal", "dim", "dtype", "else", "expand", "fill_value", "finfo", "full", "if", "int", "is", "kwargs", "mask_length", "masked_fill", "min", "min_dtype", "not", "padding_mask", "reshape", "return", "sequence_length", "shape", "target_length", "to", "torch", "triu"], "mllama/modeling_mllama.py:MllamaVisionModel.__init__": ["Conv2d", "False", "LayerNorm", "ModelPreTrainedModel", "ModelPrecomputedAspectRatioEmbedding", "ModelPrecomputedPositionEmbedding", "ModelVisionConfig", "ModelVisionEncoder", "ModelVisionModel", "Parameter", "True", "__init__", "bias", "class", "class_embedding", "config", "def", "gated_positional_embedding", "global_transformer", "hidden_size", "image_size", "in_channels", "intermediate_layers_indices", "is_gated", "kernel_size", "layernorm_post", "layernorm_pre", "max_num_tiles", "nn", "num_channels", "num_global_layers", "num_hidden_layers", "num_patches", "out_channels", "padding", "patch_embedding", "patch_size", "post_init", "post_tile_positional_embedding", "pre_tile_positional_embedding", "randn", "scale", "self", "stride", "super", "torch", "transformer", "valid"], "mllama/modeling_mllama.py:MllamaVisionModel.get_input_embeddings": ["ModelPreTrainedModel", "ModelVisionModel", "class", "def", "get_input_embeddings", "patch_embedding", "return", "self"], "mllama/modeling_mllama.py:MllamaVisionModel.apply_class_embedding": ["ModelPreTrainedModel", "ModelVisionModel", "Tensor", "_", "apply_class_embedding", "batch_size", "cat", "class", "class_embedding", "def", "dim", "expand", "hidden_size", "hidden_state", "return", "self", "shape", "torch"], "mllama/modeling_mllama.py:MllamaVisionModel.forward": ["BaseModelOutput", "F", "ModelPreTrainedModel", "ModelVisionModel", "None", "Tensor", "_", "_prepare_aspect_ratio_attention_mask", "all_intermediate_hidden_states", "apply_class_embedding", "aspect_ratio_ids", "aspect_ratio_mask", "attention_mask", "batch_size", "cat", "class", "constant", "def", "device", "dim", "dtype", "else", "flatten", "for", "forward", "gated_positional_embedding", "global_output", "global_transformer", "height", "hidden_state", "hidden_states", "i", "if", "in", "intermediate_hidden_states", "intermediate_layers_indices", "kwargs", "last_hidden_state", "layernorm_post", "layernorm_pre", "mode", "num_channels", "num_concurrent_media", "num_padding_patches", "num_patches", "num_tiles", "output", "pad", "padding", "patch_embedding", "patch_embeds", "pixel_values", "post_tile_positional_embedding", "pre_tile_positional_embedding", "r", "reshape", "return", "self", "shape", "slice_index", "stack", "target_device", "target_dtype", "target_length", "to", "torch", "transformer", "transpose", "value", "view", "weight", "width"], "mllama/modeling_mllama.py:MllamaTextModel.__init__": ["Embedding", "False", "ModelCrossAttentionDecoderLayer", "ModelPreTrainedModel", "ModelRotaryEmbedding", "ModelSelfAttentionDecoderLayer", "ModelTextConfig", "ModelTextModel", "ModelTextRMSNorm", "ModuleList", "__init__", "append", "class", "config", "cross_attention_layers", "def", "else", "embed_tokens", "eps", "for", "gradient_checkpointing", "hidden_size", "if", "in", "layer_idx", "layers", "nn", "norm", "num_hidden_layers", "pad_token_id", "padding_idx", "post_init", "range", "rms_norm_eps", "rotary_emb", "self", "super", "vocab_size"], "mllama/modeling_mllama.py:MllamaTextModel.forward": ["BaseModelOutputWithPast", "Cache", "DynamicCache", "FlashAttentionKwargs", "FloatTensor", "LongTensor", "ModelPreTrainedModel", "ModelTextModel", "None", "Tensor", "Unpack", "ValueError", "You", "_update_causal_mask", "and", "arange", "attention_mask", "bool", "cache_position", "causal_mask", "class", "config", "continue", "cross_attention_layers", "cross_attention_mask", "cross_attention_states", "decoder_layer", "def", "device", "else", "embed_tokens", "enumerate", "exactly", "for", "forward", "full_text_row_masked_out_mask", "get_seq_length", "hidden_states", "idx", "if", "in", "input_ids", "inputs_embeds", "is", "is_cross_attention_cache_empty", "is_cross_attention_layer", "kwargs", "last_hidden_state", "layers", "must", "norm", "not", "of", "one", "or", "past_key_values", "past_seen_tokens", "position_embeddings", "position_ids", "r", "raise", "return", "rotary_emb", "self", "shape", "specify", "torch", "tuple", "unsqueeze", "use_cache"], "mllama/modeling_mllama.py:MllamaForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "ModelForCausalLM", "ModelPreTrainedModel", "ModelTextModel", "__init__", "_from_config", "bias", "class", "config", "def", "get_text_config", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "text_config", "vocab_size"], "mllama/modeling_mllama.py:MllamaForCausalLM.forward": ["Cache", "CausalLMOutputWithPast", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForCausalLM", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "bool", "cache_position", "class", "cross_attention_mask", "cross_attention_states", "def", "else", "float", "forward", "full_text_row_masked_out_mask", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "outputs", "past_key_values", "position_ids", "r", "return", "self", "slice", "slice_indices", "torch", "tuple", "use_cache", "vocab_size"], "mllama/modeling_mllama.py:MllamaModel.__init__": ["Linear", "ModelConfig", "ModelModel", "ModelPreTrainedModel", "ModelTextModel", "ModelVisionModel", "None", "True", "__init__", "_from_config", "bias", "class", "config", "def", "else", "hidden_size", "if", "is", "language_model", "max_num_tiles", "multi_modal_projector", "nn", "not", "pad_token_id", "post_init", "self", "super", "text_config", "vision_config", "vision_model", "vision_output_dim", "vocab_size"], "mllama/modeling_mllama.py:MllamaModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "language_model", "return", "self"], "mllama/modeling_mllama.py:MllamaModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "language_model", "self", "set_input_embeddings", "value"], "mllama/modeling_mllama.py:MllamaModel.forward": ["BaseModelOutputWithPast", "Cache", "FlashAttentionKwargs", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "Unpack", "ValueError", "You", "_prepare_cross_attention_mask", "and", "aspect_ratio_ids", "aspect_ratio_mask", "attention_mask", "attentions", "be", "bool", "cache_position", "cannot", "class", "cross_attention_mask", "cross_attention_states", "def", "dtype", "else", "exactly", "forward", "full_text_row_masked_out_mask", "hidden_size", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "language_model", "last_hidden_state", "multi_modal_projector", "must", "not", "num_patches", "num_vision_tokens", "of", "one", "or", "outputs", "past_key_values", "pixel_values", "position_ids", "provided", "r", "raise", "reshape", "return", "self", "shape", "simultaneously", "specify", "torch", "use_cache", "vision_model", "vision_outputs"], "mllama/modeling_mllama.py:MllamaForConditionalGeneration.__init__": ["False", "GenerationMixin", "Linear", "ModelConfig", "ModelForConditionalGeneration", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "text_config", "vocab_size"], "mllama/modeling_mllama.py:MllamaForConditionalGeneration.get_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "model", "return", "self"], "mllama/modeling_mllama.py:MllamaForConditionalGeneration.set_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "model", "self", "set_input_embeddings", "value"], "mllama/modeling_mllama.py:MllamaForConditionalGeneration.forward": ["Cache", "CausalLMOutputWithPast", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "aspect_ratio_ids", "aspect_ratio_mask", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "cross_attention_mask", "cross_attention_states", "def", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "outputs", "past_key_values", "pixel_values", "position_ids", "r", "return", "self", "slice", "slice_indices", "text_config", "torch", "tuple", "use_cache", "vocab_size"], "mllama/modeling_mllama.py:MllamaForConditionalGeneration.prepare_inputs_for_generation": ["False", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "and", "aspect_ratio_ids", "aspect_ratio_mask", "attention_mask", "cache_position", "class", "cross_attention_mask", "def", "if", "input_ids", "inputs_embeds", "is_first_iteration", "kwargs", "logits_to_keep", "model_inputs", "not", "past_key_values", "pixel_values", "position_ids", "prepare_inputs_for_generation", "return", "self", "super", "use_cache"], "mllama/modeling_mllama.py:MllamaForConditionalGeneration._update_model_kwargs_for_generation": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "_update_model_kwargs_for_generation", "cat", "class", "cross_attention_mask", "cross_attention_mask_prev", "def", "dim", "get", "if", "is", "is_encoder_decoder", "kwargs", "model_kwargs", "not", "outputs", "return", "self", "super", "torch"], "mm_grounding_dino/modeling_mm_grounding_dino.py:MMGroundingDinoContrastiveEmbedding.__init__": ["Model", "Module", "Parameter", "__init__", "bias", "class", "config", "def", "max_text_len", "nn", "self", "super", "tensor", "torch"], "mm_grounding_dino/modeling_mm_grounding_dino.py:MMGroundingDinoContrastiveEmbedding.forward": ["BoolTensor", "FloatTensor", "Model", "Module", "None", "bias", "class", "def", "device", "float", "forward", "full", "inf", "masked_fill_", "math", "max_text_len", "new_res", "nn", "res", "return", "self", "shape", "sqrt", "text_hidden_state", "text_token_mask", "torch", "transpose", "vision_hidden_state"], "mm_grounding_dino/modeling_mm_grounding_dino.py:MultiScaleDeformableAttention.forward": ["False", "ModelScaleDeformableAttention", "Module", "Tensor", "_", "align_corners", "append", "attention_weights", "batch_size", "bilinear", "class", "contiguous", "def", "dim", "enumerate", "flatten", "for", "forward", "functional", "grid_sample", "height", "hidden_dim", "im2col_step", "in", "int", "level_id", "level_start_index", "list", "mode", "nn", "num_heads", "num_levels", "num_points", "num_queries", "output", "padding_mode", "reshape", "return", "sampling_grid_l_", "sampling_grids", "sampling_locations", "sampling_value_l_", "sampling_value_list", "self", "shape", "split", "stack", "sum", "torch", "transpose", "tuple", "value", "value_l_", "value_list", "value_spatial_shapes", "value_spatial_shapes_list", "view", "width", "zeros"], "mm_grounding_dino/modeling_mm_grounding_dino.py:MMGroundingDinoLearnedPositionEmbedding.__init__": ["Embedding", "Model", "Module", "__init__", "class", "column_embeddings", "config", "d_model", "def", "embedding_dim", "nn", "row_embeddings", "self", "super"], "mm_grounding_dino/modeling_mm_grounding_dino.py:MMGroundingDinoLearnedPositionEmbedding.forward": ["Model", "Module", "None", "arange", "cat", "class", "column_embeddings", "def", "device", "dim", "forward", "height", "height_values", "nn", "permute", "pixel_mask", "pixel_values", "pos", "repeat", "return", "row_embeddings", "self", "shape", "torch", "unsqueeze", "width", "width_values", "x_emb", "y_emb"], "mm_grounding_dino/modeling_mm_grounding_dino.py:MMGroundingDinoMultiscaleDeformableAttention.__init__": ["CUDA", "Linear", "Model", "ModelConfig", "Module", "MultiScaleDeformableAttention", "ValueError", "You", "__init__", "a", "and", "attention", "attention_weights", "attn", "authors", "be", "better", "but", "by", "class", "config", "d", "d_model", "def", "dim_per_head", "dimension", "disable_custom_kernels", "divisible", "each", "efficient", "embed_dim", "f", "got", "head", "if", "im2col_step", "implementation", "in", "int", "is", "make", "more", "must", "n_heads", "n_levels", "n_points", "nn", "not", "num_feature_levels", "num_heads", "of", "output_proj", "power", "raise", "sampling_offsets", "self", "set", "super", "the", "to", "value_proj", "warn", "warnings", "which"], "mm_grounding_dino/modeling_mm_grounding_dino.py:MMGroundingDinoMultiscaleDeformableAttention.with_pos_embed": ["Model", "Module", "None", "Tensor", "class", "def", "else", "if", "is", "nn", "position_embeddings", "return", "self", "tensor", "torch", "with_pos_embed"], "mm_grounding_dino/modeling_mm_grounding_dino.py:MMGroundingDinoMultiscaleDeformableAttention.forward": ["F", "False", "Last", "Make", "Model", "Module", "None", "Tensor", "ValueError", "_", "align", "attention_mask", "attention_weights", "attn", "batch_size", "be", "bool", "but", "class", "d_model", "def", "dim", "elif", "else", "encoder", "encoder_attention_mask", "encoder_hidden_states", "f", "float", "forward", "got", "hidden", "hidden_states", "if", "im2col_step", "is", "length", "level_start_index", "masked_fill", "must", "n_heads", "n_levels", "n_points", "nn", "not", "num_coordinates", "num_queries", "of", "offset_normalizer", "or", "output", "output_attentions", "output_proj", "position_embeddings", "raise", "reference_points", "return", "sampling_locations", "sampling_offsets", "self", "sequence", "sequence_length", "shape", "shapes", "softmax", "spatial", "spatial_shapes", "spatial_shapes_list", "stack", "states", "sum", "sure", "the", "to", "torch", "value", "value_proj", "view", "with", "with_pos_embed"], "mm_grounding_dino/modeling_mm_grounding_dino.py:MMGroundingDinoBiMultiHeadAttention.__init__": ["Linear", "Model", "Module", "ValueError", "__init__", "and", "be", "by", "class", "config", "d_model", "def", "divisible", "dropout", "embed_dim", "encoder_attention_heads", "encoder_ffn_dim", "f", "fusion_dropout", "got", "head_dim", "if", "must", "nn", "num_heads", "out_text_proj", "out_vision_proj", "raise", "scale", "self", "super", "text_dim", "text_proj", "values_text_proj", "values_vision_proj", "vision_dim", "vision_proj"], "mm_grounding_dino/modeling_mm_grounding_dino.py:MMGroundingDinoBiMultiHeadAttention._reshape": ["Model", "Module", "Tensor", "_reshape", "batch_size", "class", "contiguous", "def", "head_dim", "int", "nn", "num_heads", "return", "self", "seq_len", "tensor", "torch", "transpose", "view"], "mm_grounding_dino/modeling_mm_grounding_dino.py:MMGroundingDinoBiMultiHeadAttention.forward": ["Attention", "BoolTensor", "F", "FloatTensor", "Model", "Module", "None", "True", "ValueError", "_", "_reshape", "attn_weights", "attn_weights_transposed", "batch_size", "be", "bmm", "but", "clamp", "class", "def", "dim", "dropout", "embed_dim", "f", "flatten", "float", "forward", "head_dim", "if", "inf", "is", "keepdim", "masked_fill_", "max", "min", "nn", "not", "num_heads", "of", "out_text_proj", "out_vision_proj", "p", "proj_shape", "raise", "repeat", "reshape", "return", "scale", "self", "should", "size", "softmax", "src_len", "text_attention_mask", "text_attn_output", "text_attn_probs", "text_attn_weights", "text_features", "text_key_states", "text_proj", "text_value_states", "tgt_len", "torch", "training", "transpose", "tuple", "values_text_proj", "values_vision_proj", "view", "vision_attention_mask", "vision_attn_output", "vision_attn_probs", "vision_attn_weights", "vision_features", "vision_proj", "vision_query_states", "vision_value_states", "weights"], "mm_grounding_dino/modeling_mm_grounding_dino.py:drop_path": ["False", "Model_path", "Model_prob", "Tensor", "bool", "def", "device", "div", "dtype", "float", "floor_", "if", "input", "keep_prob", "ndim", "not", "or", "output", "rand", "random_tensor", "return", "shape", "torch", "training"], "mm_grounding_dino/modeling_mm_grounding_dino.py:MMGroundingDinoDropPath.__init__": ["Model", "Module", "None", "__init__", "class", "def", "drop_prob", "float", "nn", "self", "super"], "mm_grounding_dino/modeling_mm_grounding_dino.py:MMGroundingDinoDropPath.forward": ["Model", "Module", "Tensor", "class", "def", "drop_path", "drop_prob", "forward", "hidden_states", "nn", "return", "self", "torch", "training"], "mm_grounding_dino/modeling_mm_grounding_dino.py:MMGroundingDinoDropPath.extra_repr": ["Model", "Module", "class", "def", "drop_prob", "extra_repr", "f", "nn", "p", "return", "self", "str"], "mm_grounding_dino/modeling_mm_grounding_dino.py:MMGroundingDinoFusionLayer.__init__": ["Identity", "LayerNorm", "Model", "ModelBiMultiHeadAttention", "ModelDropPath", "Module", "Parameter", "True", "__init__", "attn", "class", "config", "d_model", "def", "drop_path", "else", "fusion_droppath", "if", "init_values", "layer_norm_eps", "layer_norm_text", "layer_norm_vision", "nn", "ones", "requires_grad", "self", "super", "text_param", "torch", "vision_param"], "mm_grounding_dino/modeling_mm_grounding_dino.py:MMGroundingDinoFusionLayer.forward": ["BoolTensor", "FloatTensor", "Model", "Module", "None", "attention_mask_text", "attention_mask_vision", "attn", "class", "def", "delta_t", "delta_v", "drop_path", "forward", "layer_norm_text", "layer_norm_vision", "nn", "return", "self", "text_attention_mask", "text_attn", "text_features", "text_param", "torch", "tuple", "vision_attention_mask", "vision_attn", "vision_features", "vision_param"], "mm_grounding_dino/modeling_mm_grounding_dino.py:MMGroundingDinoPreTrainedModel._init_weights": ["Conv2d", "Embedding", "False", "GroupNorm", "LayerNorm", "Linear", "Model", "ModelBiMultiHeadAttention", "ModelContrastiveEmbedding", "ModelFusionLayer", "ModelLearnedPositionEmbedding", "ModelMLPPredictionHead", "ModelMultiscaleDeformableAttention", "None", "PreTrainedModel", "True", "_init_weights", "_is_hf_initialized", "abs", "and", "arange", "attention_weights", "bias", "class", "column_embeddings", "config", "constant_", "copy_", "cos", "def", "default_dtype", "dtype", "elif", "for", "gain", "get_default_dtype", "getattr", "grid_init", "hasattr", "i", "if", "in", "init", "init_std", "int64", "is", "isinstance", "keepdim", "layers", "level_embed", "log", "math", "max", "mean", "module", "n_heads", "n_levels", "n_points", "nn", "normal_", "not", "ones_", "out_text_proj", "out_vision_proj", "output_proj", "padding_idx", "pi", "range", "reference_points", "repeat", "row_embeddings", "sampling_offsets", "self", "sin", "stack", "std", "text_param", "text_proj", "thetas", "to", "torch", "two_stage", "uniform_", "value_proj", "values_text_proj", "values_vision_proj", "view", "vision_param", "vision_proj", "weight", "xavier_uniform_", "zeros_"], "mm_grounding_dino/modeling_mm_grounding_dino.py:MMGroundingDinoPreTrainedModel._set_gradient_checkpointing": ["False", "Model", "ModelDecoder", "PreTrainedModel", "_set_gradient_checkpointing", "class", "def", "gradient_checkpointing", "if", "isinstance", "module", "self", "value"], "mm_grounding_dino/modeling_mm_grounding_dino.py:MMGroundingDinoFrozenBatchNorm2d.__init__": ["Model", "Module", "__init__", "bias", "class", "def", "n", "nn", "ones", "register_buffer", "running_mean", "running_var", "self", "super", "torch", "weight", "zeros"], "mm_grounding_dino/modeling_mm_grounding_dino.py:MMGroundingDinoFrozenBatchNorm2d._load_from_state_dict": ["Model", "Module", "_load_from_state_dict", "class", "def", "del", "error_msgs", "if", "in", "local_metadata", "missing_keys", "nn", "num_batches_tracked", "num_batches_tracked_key", "prefix", "self", "state_dict", "strict", "super", "unexpected_keys"], "mm_grounding_dino/modeling_mm_grounding_dino.py:MMGroundingDinoFrozenBatchNorm2d.forward": ["Model", "Module", "bias", "class", "def", "epsilon", "forward", "nn", "reshape", "return", "rsqrt", "running_mean", "running_var", "scale", "self", "weight", "x"], "mm_grounding_dino/modeling_mm_grounding_dino.py:replace_batch_norm": ["BatchNorm2d", "ModelFrozenBatchNorm2d", "Model_batch_norm", "_modules", "bias", "children", "copy_", "def", "device", "for", "if", "in", "isinstance", "len", "list", "meta", "model", "module", "name", "named_children", "new_module", "nn", "num_features", "r", "running_mean", "running_var", "torch", "weight"], "mm_grounding_dino/modeling_mm_grounding_dino.py:MMGroundingDinoConvEncoder.__init__": ["Either", "False", "Model", "Module", "None", "True", "ValueError", "__init__", "and", "backbone", "backbone_config", "backbone_kwargs", "backbone_model_type", "be", "channels", "class", "config", "create_model", "def", "elif", "else", "feature_info", "features_only", "for", "if", "in", "intermediate_channel_sizes", "is", "layer2", "layer3", "layer4", "load_backbone", "model", "model_type", "name", "named_parameters", "nn", "no_grad", "not", "or", "parameter", "pretrained", "provided", "raise", "replace_batch_norm", "requires_backends", "requires_grad_", "resnet", "self", "should", "stage", "super", "the", "timm", "torch", "use_pretrained_backbone", "use_timm_backbone", "with"], "mm_grounding_dino/modeling_mm_grounding_dino.py:MMGroundingDinoConvEncoder.forward": ["Model", "Module", "None", "Tensor", "append", "bool", "class", "config", "def", "else", "feature_map", "feature_maps", "features", "float", "for", "forward", "functional", "if", "in", "interpolate", "mask", "model", "nn", "out", "pixel_mask", "pixel_values", "return", "self", "shape", "size", "to", "torch", "use_timm_backbone"], "mm_grounding_dino/modeling_mm_grounding_dino.py:MMGroundingDinoConvModel.__init__": ["Model", "Module", "__init__", "class", "conv_encoder", "def", "nn", "position_embedding", "self", "super"], "mm_grounding_dino/modeling_mm_grounding_dino.py:MMGroundingDinoConvModel.forward": ["Model", "Module", "append", "class", "conv_encoder", "def", "dtype", "feature_map", "for", "forward", "in", "mask", "nn", "out", "pixel_mask", "pixel_values", "pos", "position_embedding", "return", "self", "to"], "mm_grounding_dino/modeling_mm_grounding_dino.py:MMGroundingDinoMultiheadAttention.__init__": ["Dropout", "Linear", "Model", "Module", "None", "The", "ValueError", "__init__", "a", "all_head_size", "and", "attention", "attention_dropout", "attention_head_size", "class", "config", "def", "dropout", "embedding_size", "f", "hasattr", "heads", "hidden", "hidden_size", "if", "int", "is", "key", "multiple", "nn", "not", "num_attention_heads", "number", "of", "out_proj", "query", "raise", "self", "size", "super", "the", "value"], "mm_grounding_dino/modeling_mm_grounding_dino.py:MMGroundingDinoMultiheadAttention.forward": ["False", "FloatTensor", "Model", "Module", "None", "Tensor", "_", "all_head_size", "attention_head_size", "attention_mask", "attention_probs", "attention_scores", "batch_size", "bool", "class", "context_layer", "contiguous", "def", "dim", "dropout", "else", "forward", "functional", "if", "is", "key", "key_layer", "keys", "math", "matmul", "new_context_layer_shape", "nn", "not", "num_attention_heads", "out_proj", "output_attentions", "outputs", "permute", "queries", "query", "query_layer", "return", "self", "seq_length", "shape", "size", "softmax", "sqrt", "torch", "transpose", "tuple", "value", "value_layer", "values", "view"], "mm_grounding_dino/modeling_mm_grounding_dino.py:MMGroundingDinoTextEnhancerLayer.__init__": ["ACT2FN", "LayerNorm", "Linear", "Model", "ModelMultiheadAttention", "Module", "__init__", "activation", "activation_function", "class", "config", "d_model", "def", "dropout", "encoder_attention_heads", "encoder_ffn_dim", "fc1", "fc2", "layer_norm_after", "layer_norm_before", "layer_norm_eps", "nn", "num_attention_heads", "num_heads", "self", "self_attn", "super", "text_enhancer_dropout"], "mm_grounding_dino/modeling_mm_grounding_dino.py:MMGroundingDinoTextEnhancerLayer.with_pos_embed": ["Model", "Module", "None", "Tensor", "class", "def", "else", "hidden_state", "if", "is", "nn", "position_embeddings", "return", "self", "with_pos_embed"], "mm_grounding_dino/modeling_mm_grounding_dino.py:MMGroundingDinoTextEnhancerLayer.forward": ["BoolTensor", "FloatTensor", "Model", "Module", "None", "True", "activation", "and", "attention_mask", "attention_masks", "attention_output", "attention_weights", "class", "def", "dim", "dropout", "dtype", "fc1", "fc2", "finfo", "forward", "functional", "hidden_states", "if", "keys", "layer_norm_after", "layer_norm_before", "min", "nn", "num_heads", "output_attentions", "p", "position_embeddings", "queries", "repeat", "residual", "return", "self", "self_attn", "shape", "to", "torch", "training", "tuple", "values", "with_pos_embed"], "mm_grounding_dino/modeling_mm_grounding_dino.py:MMGroundingDinoDeformableLayer.__init__": ["ACT2FN", "LayerNorm", "Linear", "Model", "ModelConfig", "ModelMultiscaleDeformableAttention", "Module", "__init__", "activation_dropout", "activation_fn", "activation_function", "class", "config", "d_model", "def", "dropout", "embed_dim", "encoder_attention_heads", "encoder_ffn_dim", "encoder_n_points", "fc1", "fc2", "final_layer_norm", "layer_norm_eps", "n_points", "nn", "num_heads", "self", "self_attn", "self_attn_layer_norm", "super"], "mm_grounding_dino/modeling_mm_grounding_dino.py:MMGroundingDinoDeformableLayer.forward": ["False", "Model", "Module", "None", "Tensor", "activation_dropout", "activation_fn", "any", "attention_mask", "attn_weights", "bool", "clamp", "clamp_value", "class", "def", "dropout", "dtype", "encoder_attention_mask", "encoder_hidden_states", "fc1", "fc2", "final_layer_norm", "finfo", "forward", "functional", "hidden_states", "if", "isinf", "isnan", "level_start_index", "max", "min", "nn", "or", "output_attentions", "p", "position_embeddings", "reference_points", "residual", "return", "self", "self_attn", "self_attn_layer_norm", "spatial_shapes", "spatial_shapes_list", "torch", "training"], "mm_grounding_dino/modeling_mm_grounding_dino.py:get_sine_pos_embed": ["Model_sine_pos_embed", "Tensor", "True", "arange", "bool", "cat", "cos", "def", "device", "dim", "dim_t", "div", "dtype", "exchange_xy", "flatten", "float32", "floor", "for", "if", "in", "int", "math", "num_pos_feats", "pi", "pos_tensor", "position_embeddings", "return", "rounding_mode", "scale", "shape", "sin", "sin_x", "sine_func", "split", "stack", "temperature", "torch", "x"], "mm_grounding_dino/modeling_mm_grounding_dino.py:MMGroundingDinoEncoderLayer.__init__": ["Model", "ModelDeformableLayer", "ModelFusionLayer", "ModelTextEnhancerLayer", "Module", "None", "__init__", "class", "config", "d_model", "def", "deformable_layer", "fusion_layer", "nn", "self", "super", "text_enhancer_layer"], "mm_grounding_dino/modeling_mm_grounding_dino.py:MMGroundingDinoEncoderLayer.get_text_position_embeddings": ["False", "Model", "Module", "None", "Tensor", "_", "and", "arange", "batch_size", "class", "d_model", "def", "device", "exchange_xy", "float", "get_sine_pos_embed", "get_text_position_embeddings", "if", "is", "nn", "not", "num_pos_feats", "repeat", "return", "self", "seq_length", "shape", "text_features", "text_position_embedding", "text_position_ids", "torch", "unsqueeze"], "mm_grounding_dino/modeling_mm_grounding_dino.py:MMGroundingDinoEncoderLayer.forward": ["Model", "Module", "None", "Tensor", "attention_mask", "attention_mask_text", "attention_mask_vision", "attention_masks", "class", "def", "deformable_layer", "else", "forward", "fusion_layer", "get_text_position_embeddings", "hidden_states", "if", "int", "is", "key_padding_mask", "level_start_index", "list", "nn", "not", "position_embeddings", "reference_points", "return", "self", "spatial_shapes", "spatial_shapes_list", "text_attention_mask", "text_enhanced_attn", "text_enhancer_layer", "text_features", "text_fused_attn", "text_position_embedding", "text_position_ids", "text_self_attention_masks", "tuple", "vision_deformable_attn", "vision_features", "vision_fused_attn", "vision_position_embedding"], "mm_grounding_dino/modeling_mm_grounding_dino.py:MMGroundingDinoEncoder.__init__": ["Model", "ModelConfig", "ModelLayer", "ModelPreTrainedModel", "ModuleList", "_", "__init__", "class", "config", "def", "dropout", "encoder_layers", "for", "in", "layers", "nn", "post_init", "range", "self", "super"], "mm_grounding_dino/modeling_mm_grounding_dino.py:MMGroundingDinoEncoder.get_reference_points": ["Model", "ModelPreTrainedModel", "None", "append", "cat", "class", "def", "device", "dtype", "enumerate", "float32", "for", "get_reference_points", "height", "ij", "in", "indexing", "level", "linspace", "meshgrid", "ref", "ref_x", "ref_y", "reference_points", "reference_points_list", "reshape", "return", "spatial_shapes", "stack", "torch", "valid_ratios", "width"], "mm_grounding_dino/modeling_mm_grounding_dino.py:MMGroundingDinoEncoder.forward": ["Model", "ModelOutput", "ModelPreTrainedModel", "None", "Tensor", "all_attn_deformable", "all_attn_enhanced_text", "all_attn_fused_text", "all_attn_fused_vision", "all_attns", "attentions", "class", "config", "def", "device", "else", "enc_outputs", "encoder_layer", "encoder_text_states", "encoder_vision_states", "enumerate", "for", "forward", "get_reference_points", "i", "if", "in", "int", "is", "key_padding_mask", "kwargs", "last_hidden_state_text", "last_hidden_state_vision", "layers", "level_start_index", "list", "not", "output_attentions", "output_hidden_states", "r", "reference_points", "return", "return_dict", "self", "spatial_shapes", "spatial_shapes_list", "text_attention_mask", "text_features", "text_hidden_states", "text_position_embedding", "text_position_ids", "text_self_attention_masks", "tuple", "use_return_dict", "v", "valid_ratios", "vision_attention_mask", "vision_features", "vision_hidden_states", "vision_position_embedding"], "mm_grounding_dino/modeling_mm_grounding_dino.py:MMGroundingDinoDecoderLayer.__init__": ["ACT2FN", "LayerNorm", "Linear", "Model", "ModelConfig", "ModelMultiheadAttention", "ModelMultiscaleDeformableAttention", "Module", "__init__", "activation_dropout", "activation_fn", "activation_function", "class", "config", "d_model", "decoder_attention_heads", "decoder_ffn_dim", "decoder_n_points", "def", "dropout", "embed_dim", "encoder_attn", "encoder_attn_layer_norm", "encoder_attn_text", "encoder_attn_text_layer_norm", "fc1", "fc2", "final_layer_norm", "layer_norm_eps", "n_points", "nn", "num_attention_heads", "num_heads", "self", "self_attn", "self_attn_layer_norm", "super"], "mm_grounding_dino/modeling_mm_grounding_dino.py:MMGroundingDinoDecoderLayer.with_pos_embed": ["Model", "Module", "None", "Tensor", "class", "def", "else", "if", "is", "nn", "position_embeddings", "return", "self", "tensor", "torch", "with_pos_embed"], "mm_grounding_dino/modeling_mm_grounding_dino.py:MMGroundingDinoDecoderLayer.forward": ["False", "Model", "Module", "None", "Tensor", "True", "activation_dropout", "activation_fn", "attention_mask", "bool", "class", "cross_attn_weights", "def", "dropout", "encoder_attention_mask", "encoder_attn", "encoder_attn_layer_norm", "encoder_attn_text", "encoder_attn_text_layer_norm", "encoder_hidden_states", "fc1", "fc2", "final_layer_norm", "forward", "functional", "hidden_states", "if", "keys", "level_start_index", "nn", "output_attentions", "outputs", "p", "position_embeddings", "queries", "reference_points", "residual", "return", "second_residual", "self", "self_attn", "self_attn_layer_norm", "self_attn_mask", "self_attn_weights", "spatial_shapes", "spatial_shapes_list", "text_cross_attn_weights", "text_encoder_attention_mask", "text_encoder_hidden_states", "third_residual", "torch", "training", "values", "vision_encoder_attention_mask", "vision_encoder_hidden_states", "with_pos_embed"], "mm_grounding_dino/modeling_mm_grounding_dino.py:MMGroundingDinoDecoder.__init__": ["False", "LayerNorm", "Model", "ModelConfig", "ModelLayer", "ModelMLPPredictionHead", "ModelPreTrainedModel", "ModuleList", "None", "_", "__init__", "bbox_embed", "class", "class_embed", "config", "d_model", "decoder_layers", "def", "dropout", "for", "gradient_checkpointing", "in", "layer_norm", "layer_norm_eps", "layers", "nn", "post_init", "query_dim", "query_scale", "range", "reference_points_head", "self", "super"], "mm_grounding_dino/modeling_mm_grounding_dino.py:MMGroundingDinoDecoder.forward": ["Last", "Model", "ModelOutput", "ModelPreTrainedModel", "None", "ValueError", "all_attns", "all_cross_attns_text", "all_cross_attns_vision", "all_hidden_states", "all_self_attns", "and", "attentions", "bbox_embed", "be", "but", "cat", "checkpoint", "class", "config", "create_custom_forward", "custom_forward", "d_model", "decoder_attention_heads", "decoder_layer", "def", "detach", "dim", "dtype", "elif", "else", "enumerate", "eps", "f", "finfo", "for", "forward", "get_sine_pos_embed", "got", "gradient_checkpointing", "hidden_states", "idx", "if", "in", "inputs", "inputs_embeds", "intermediate", "intermediate_hidden_states", "intermediate_reference_points", "is", "kwargs", "last_hidden_state", "layer_norm", "layer_outputs", "layers", "level_start_index", "logit", "min", "module", "must", "new_reference_points", "not", "num_coordinates", "num_pos_feats", "num_queries", "of", "or", "output_attentions", "output_hidden_states", "position_embeddings", "query_pos", "r", "raise", "reference_points", "reference_points_head", "reference_points_input", "repeat", "return", "return_dict", "self", "self_attn_mask", "shape", "sigmoid", "spatial_shapes", "spatial_shapes_list", "special", "stack", "text_encoder_attention_mask", "text_encoder_hidden_states", "tmp", "to", "torch", "training", "tuple", "use_return_dict", "utils", "v", "valid_ratios", "vision_encoder_attention_mask", "vision_encoder_hidden_states"], "mm_grounding_dino/modeling_mm_grounding_dino.py:MMGroundingDinoSinePositionEmbedding.__init__": ["Model", "Module", "__init__", "class", "config", "d_model", "def", "embedding_dim", "math", "nn", "pi", "positional_embedding_temperature", "scale", "self", "super", "temperature"], "mm_grounding_dino/modeling_mm_grounding_dino.py:MMGroundingDinoSinePositionEmbedding.forward": ["Model", "Module", "None", "arange", "cat", "class", "cos", "cumsum", "def", "device", "dim", "dim_t", "div", "dtype", "embedding_dim", "eps", "flatten", "float32", "floor", "forward", "nn", "permute", "pixel_mask", "pixel_values", "pos", "pos_x", "pos_y", "return", "rounding_mode", "scale", "self", "sin", "stack", "temperature", "torch", "x_embed", "y_embed"], "mm_grounding_dino/modeling_mm_grounding_dino.py:build_position_encoding": ["ModelLearnedPositionEmbedding", "ModelSinePositionEmbedding", "Model_position_encoding", "Not", "ValueError", "config", "def", "elif", "else", "f", "if", "learned", "position_embedding", "position_embedding_type", "raise", "return", "sine", "supported"], "mm_grounding_dino/modeling_mm_grounding_dino.py:generate_masks_with_special_tokens_and_transfer_map": ["LongTensor", "Model_masks_with_special_tokens_and_transfer_map", "SPECIAL_TOKENS", "Tensor", "True", "arange", "attention_mask", "batch_size", "bool", "col", "def", "device", "else", "eye", "for", "i", "idxs", "if", "in", "input_ids", "logical_or", "long", "nonzero", "num_token", "or", "position_ids", "previous_col", "range", "repeat", "return", "row", "shape", "special_token", "special_tokens_mask", "to", "torch", "tuple", "unsqueeze", "zeros"], "mm_grounding_dino/modeling_mm_grounding_dino.py:MMGroundingDinoModel.__init__": ["AutoModel", "Conv2d", "Embedding", "False", "GroupNorm", "LayerNorm", "Linear", "Model", "ModelConfig", "ModelContrastiveEmbedding", "ModelConvEncoder", "ModelConvModel", "ModelDecoder", "ModelEncoder", "ModelMLPPredictionHead", "ModelPreTrainedModel", "ModuleList", "Parameter", "Sequential", "Tensor", "_", "__init__", "add_pooling_layer", "append", "backbone", "build_position_encoding", "class", "config", "d_model", "decoder", "def", "embedding_init_target", "enc_output", "enc_output_norm", "encoder", "encoder_output_bbox_embed", "encoder_output_class_embed", "for", "from_config", "hidden_dim", "hidden_size", "i", "if", "in", "in_channels", "input_dim", "input_proj_list", "input_proj_vision", "intermediate_channel_sizes", "kernel_size", "layer_norm_eps", "len", "level_embed", "nn", "not", "num_backbone_outs", "num_feature_levels", "num_layers", "num_queries", "or", "output_dim", "padding", "position_embeddings", "post_init", "query_position_embeddings", "range", "self", "stride", "super", "text_backbone", "text_config", "text_projection", "torch", "two_stage"], "mm_grounding_dino/modeling_mm_grounding_dino.py:MMGroundingDinoModel.freeze_backbone": ["False", "Model", "ModelPreTrainedModel", "backbone", "class", "conv_encoder", "def", "for", "freeze_backbone", "in", "model", "name", "named_parameters", "param", "requires_grad_", "self"], "mm_grounding_dino/modeling_mm_grounding_dino.py:MMGroundingDinoModel.unfreeze_backbone": ["Model", "ModelPreTrainedModel", "True", "backbone", "class", "conv_encoder", "def", "for", "in", "model", "name", "named_parameters", "param", "requires_grad_", "self", "unfreeze_backbone"], "mm_grounding_dino/modeling_mm_grounding_dino.py:MMGroundingDinoModel.get_valid_ratio": ["Model", "ModelPreTrainedModel", "_", "class", "def", "float", "get_valid_ratio", "height", "mask", "return", "self", "shape", "stack", "sum", "torch", "valid_height", "valid_ratio", "valid_ratio_height", "valid_ratio_width", "valid_width", "width"], "mm_grounding_dino/modeling_mm_grounding_dino.py:MMGroundingDinoModel.generate_encoder_output_proposals": ["Model", "ModelPreTrainedModel", "True", "all", "append", "batch_size", "cat", "class", "current_position", "def", "device", "dtype", "enc_output", "enc_output_norm", "enumerate", "expand", "float", "float32", "for", "generate_encoder_output_proposals", "grid", "grid_x", "grid_y", "height", "ij", "in", "indexing", "inf", "keepdim", "level", "linspace", "log", "mask_flatten_", "masked_fill", "meshgrid", "object_query", "ones_like", "output_proposals", "output_proposals_valid", "padding_mask", "proposal", "proposals", "return", "scale", "self", "shape", "spatial_shapes", "sum", "torch", "unsqueeze", "valid_height", "valid_width", "view", "width", "width_height"], "mm_grounding_dino/modeling_mm_grounding_dino.py:MMGroundingDinoModel.forward": ["Model", "ModelEncoderOutput", "ModelOutput", "ModelPreTrainedModel", "None", "Tensor", "_len_sources", "and", "append", "as_tensor", "attention_mask", "attentions", "backbone", "batch_size", "bool", "cat", "class", "config", "cumsum", "d_model", "decoder", "decoder_attentions", "decoder_hidden_states", "decoder_outputs", "def", "delta_bbox", "detach", "device", "dim", "dtype", "elif", "else", "embedding_init_target", "enc_outputs", "enc_outputs_class", "enc_outputs_coord_logits", "encoder", "encoder_attentions", "encoder_last_hidden_state_text", "encoder_last_hidden_state_vision", "encoder_logits", "encoder_output_bbox_embed", "encoder_output_class_embed", "encoder_outputs", "encoder_pred_boxes", "encoder_text_hidden_states", "encoder_vision_hidden_states", "enumerate", "feature_maps", "flatten", "float", "for", "forward", "functional", "gather", "generate_encoder_output_proposals", "generate_masks_with_special_tokens_and_transfer_map", "get_valid_ratio", "height", "hidden_states", "if", "in", "init_reference_points", "input_ids", "input_proj_vision", "inputs_embeds", "intermediate_hidden_states", "intermediate_reference_points", "interpolate", "is", "isinstance", "kwargs", "last_hidden_state", "last_hidden_state_text", "last_hidden_state_vision", "len", "level", "level_embed", "level_start_index", "long", "lvl_pos_embed", "lvl_pos_embed_flatten", "m", "mask", "mask_flatten", "masks", "max", "max_text_len", "ndim", "new_zeros", "nn", "not", "num_channels", "num_feature_levels", "num_queries", "object_query_embedding", "ones", "ones_like", "or", "output_attentions", "output_hidden_states", "output_proposals", "pixel_mask", "pixel_values", "pos_embed", "pos_l", "position_embedding", "position_embeddings_list", "position_ids", "prod", "query_embeds", "query_position_embeddings", "r", "range", "reference_points", "repeat", "return", "return_dict", "self", "self_attn_mask", "shape", "sigmoid", "size", "source", "source_flatten", "spatial_shape", "spatial_shapes", "spatial_shapes_list", "squeeze", "stack", "target", "text_attention_mask", "text_backbone", "text_encoder_attention_mask", "text_encoder_hidden_states", "text_features", "text_hidden_states", "text_outputs", "text_position_embedding", "text_position_ids", "text_projection", "text_self_attention_masks", "text_token_mask", "to", "token_type_ids", "topk", "topk_coords_logits", "topk_logits", "topk_proposals", "torch", "transpose", "tuple", "tuple_outputs", "two_stage", "unsqueeze", "use_return_dict", "valid_ratios", "value", "view", "vision_attention_mask", "vision_encoder_attention_mask", "vision_encoder_hidden_states", "vision_features", "vision_hidden_states", "vision_position_embedding", "weight", "width", "zeros_like", "zip"], "mm_grounding_dino/modeling_mm_grounding_dino.py:MMGroundingDinoMLPPredictionHead.__init__": ["Linear", "Model", "Module", "ModuleList", "__init__", "class", "def", "for", "h", "hidden_dim", "in", "input_dim", "k", "layers", "n", "nn", "num_layers", "output_dim", "self", "super", "zip"], "mm_grounding_dino/modeling_mm_grounding_dino.py:MMGroundingDinoMLPPredictionHead.forward": ["Model", "Module", "class", "def", "else", "enumerate", "for", "forward", "functional", "i", "if", "in", "layer", "layers", "nn", "num_layers", "relu", "return", "self", "x"], "mm_grounding_dino/modeling_mm_grounding_dino.py:build_label_maps": ["F", "FloatTensor", "LongTensor", "Model_label_maps", "None", "SPECIAL_TOKENS", "cumsum", "def", "delimiter_token_masks", "delimiter_tokens", "device", "dim", "for", "in", "input_ids", "int32", "isin", "label_group", "label_groups", "label_map", "label_maps", "logits", "max_seq_len", "num_labels", "pad", "repeat", "return", "shape", "tensor", "to", "torch", "tuple", "unique", "unique_labels", "unsqueeze", "value", "where"], "mm_grounding_dino/modeling_mm_grounding_dino.py:build_text_mask": ["Model_text_mask", "None", "attention_mask", "bool", "def", "device", "dtype", "logits", "return", "seq_len", "shape", "text_mask", "torch", "zeros_like"], "mm_grounding_dino/modeling_mm_grounding_dino.py:MMGroundingDinoForObjectDetection.__init__": ["Model", "ModelConfig", "ModelContrastiveEmbedding", "ModelMLPPredictionHead", "ModelModel", "ModelPreTrainedModel", "ModuleList", "_", "__init__", "bbox_embed", "class", "class_embed", "config", "d_model", "decoder", "decoder_layers", "def", "for", "hidden_dim", "in", "input_dim", "model", "nn", "num_layers", "output_dim", "post_init", "range", "self", "super"], "mm_grounding_dino/modeling_mm_grounding_dino.py:MMGroundingDinoForObjectDetection.forward": ["BoolTensor", "FloatTensor", "LongTensor", "Model", "ModelEncoderOutput", "ModelObjectDetectionOutput", "ModelPreTrainedModel", "None", "ValueError", "append", "attention_mask", "auxiliary_outputs", "bbox_embed", "be", "bool", "build_label_maps", "build_text_mask", "but", "class", "class_embed", "config", "decoder_attentions", "decoder_hidden_states", "def", "delta_bbox", "device", "dict", "dict_outputs", "elif", "else", "enc_outputs_class", "enc_outputs_coord_logits", "enc_text_hidden_state", "encoder_attentions", "encoder_last_hidden_state_text", "encoder_last_hidden_state_vision", "encoder_logits", "encoder_outputs", "encoder_pred_boxes", "encoder_text_hidden_states", "encoder_vision_hidden_states", "eps", "f", "for", "forward", "got", "hidden_states", "idx", "if", "in", "init_reference_points", "input_ids", "inter_references_points", "intermediate_hidden_states", "intermediate_reference_points", "is", "kwargs", "label_maps", "labels", "last_hidden_state", "level", "list", "logit", "logits", "loss", "loss_dict", "loss_function", "model", "not", "num_levels", "ones_like", "or", "out", "output", "output_attentions", "output_hidden_states", "outputs", "outputs_class", "outputs_classes", "outputs_coord", "outputs_coord_logits", "outputs_coords", "pixel_mask", "pixel_values", "pred_boxes", "r", "raise", "range", "reference", "reference_coordinates", "return", "return_dict", "self", "shape", "should", "sigmoid", "special", "stack", "str", "text_hidden_state", "text_mask", "text_token_mask", "token_type_ids", "torch", "tuple", "use_return_dict", "vision_hidden_state"], "mobilebert/modeling_mobilebert.py:NoNorm.__init__": ["Module", "NoNorm", "None", "Parameter", "__init__", "bias", "class", "def", "eps", "feat_size", "nn", "ones", "self", "super", "torch", "weight", "zeros"], "mobilebert/modeling_mobilebert.py:NoNorm.forward": ["Module", "NoNorm", "Tensor", "bias", "class", "def", "forward", "input_tensor", "nn", "return", "self", "torch", "weight"], "mobilebert/modeling_mobilebert.py:MobileBertEmbeddings.__init__": ["Dropout", "Embedding", "False", "LayerNorm", "Linear", "ModelEmbeddings", "Module", "NORM2FN", "__init__", "arange", "class", "config", "def", "dropout", "else", "embed_dim_multiplier", "embedded_input_size", "embedding_size", "embedding_transformation", "expand", "hidden_dropout_prob", "hidden_size", "if", "max_position_embeddings", "nn", "normalization_type", "pad_token_id", "padding_idx", "persistent", "position_embeddings", "position_ids", "register_buffer", "self", "super", "token_type_embeddings", "torch", "trigram_input", "type_vocab_size", "vocab_size", "word_embeddings"], "mobilebert/modeling_mobilebert.py:MobileBertEmbeddings.forward": ["FloatTensor", "LayerNorm", "LongTensor", "ModelEmbeddings", "Module", "None", "Tensor", "cat", "class", "def", "device", "dim", "dropout", "dtype", "else", "embedding_size", "embedding_transformation", "embeddings", "forward", "functional", "hidden_size", "if", "input_ids", "input_shape", "inputs_embeds", "is", "long", "nn", "not", "or", "pad", "position_embeddings", "position_ids", "return", "self", "seq_length", "size", "token_type_embeddings", "token_type_ids", "torch", "trigram_input", "value", "word_embeddings", "zeros"], "mobilebert/modeling_mobilebert.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "contiguous", "def", "dim", "dropout", "float", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "shape", "size", "softmax", "torch", "training", "transpose", "value"], "mobilebert/modeling_mobilebert.py:MobileBertSelfAttention.__init__": ["Dropout", "False", "Linear", "ModelSelfAttention", "Module", "__init__", "all_head_size", "attention_head_size", "attention_probs_dropout_prob", "class", "config", "def", "dropout", "else", "hidden_size", "if", "int", "is_causal", "key", "nn", "num_attention_heads", "query", "scaling", "self", "super", "true_hidden_size", "use_bottleneck_attention", "value"], "mobilebert/modeling_mobilebert.py:MobileBertSelfAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "FloatTensor", "ModelSelfAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_attn_implementation", "attention_head_size", "attention_interface", "attention_mask", "attn_output", "attn_weights", "class", "config", "contiguous", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "hidden_shape", "if", "input_shape", "key", "key_layer", "key_tensor", "kwargs", "nn", "not", "p", "query", "query_layer", "query_tensor", "reshape", "return", "scaling", "self", "shape", "torch", "training", "transpose", "tuple", "value", "value_layer", "value_tensor", "view"], "mobilebert/modeling_mobilebert.py:MobileBertSelfOutput.__init__": ["Dropout", "LayerNorm", "Linear", "ModelSelfOutput", "Module", "NORM2FN", "__init__", "class", "config", "def", "dense", "dropout", "eps", "hidden_dropout_prob", "if", "layer_norm_eps", "nn", "normalization_type", "not", "self", "super", "true_hidden_size", "use_bottleneck"], "mobilebert/modeling_mobilebert.py:MobileBertSelfOutput.forward": ["LayerNorm", "ModelSelfOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "if", "layer_outputs", "nn", "not", "residual_tensor", "return", "self", "torch", "use_bottleneck"], "mobilebert/modeling_mobilebert.py:MobileBertAttention.__init__": ["ModelAttention", "ModelSelfAttention", "ModelSelfOutput", "Module", "__init__", "class", "config", "def", "nn", "output", "self", "super"], "mobilebert/modeling_mobilebert.py:MobileBertAttention.forward": ["FloatTensor", "ModelAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attention_output", "attn_weights", "class", "def", "forward", "key_tensor", "kwargs", "layer_input", "nn", "output", "query_tensor", "return", "self", "torch", "tuple", "value_tensor"], "mobilebert/modeling_mobilebert.py:MobileBertIntermediate.__init__": ["ACT2FN", "Linear", "ModelIntermediate", "Module", "__init__", "class", "config", "def", "dense", "else", "hidden_act", "if", "intermediate_act_fn", "intermediate_size", "isinstance", "nn", "self", "str", "super", "true_hidden_size"], "mobilebert/modeling_mobilebert.py:MobileBertIntermediate.forward": ["ModelIntermediate", "Module", "Tensor", "class", "def", "dense", "forward", "hidden_states", "intermediate_act_fn", "nn", "return", "self", "torch"], "mobilebert/modeling_mobilebert.py:OutputBottleneck.__init__": ["Dropout", "LayerNorm", "Linear", "ModelBottleneck", "Module", "NORM2FN", "__init__", "class", "config", "def", "dense", "dropout", "eps", "hidden_dropout_prob", "hidden_size", "layer_norm_eps", "nn", "normalization_type", "self", "super", "true_hidden_size"], "mobilebert/modeling_mobilebert.py:OutputBottleneck.forward": ["LayerNorm", "ModelBottleneck", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "layer_Models", "nn", "residual_tensor", "return", "self", "torch"], "mobilebert/modeling_mobilebert.py:MobileBertOutput.__init__": ["Dropout", "LayerNorm", "Linear", "ModelOutput", "Module", "NORM2FN", "OutputBottleneck", "__init__", "bottleneck", "class", "config", "def", "dense", "dropout", "else", "hidden_dropout_prob", "if", "intermediate_size", "nn", "normalization_type", "not", "self", "super", "true_hidden_size", "use_bottleneck"], "mobilebert/modeling_mobilebert.py:MobileBertOutput.forward": ["LayerNorm", "ModelOutput", "Module", "Tensor", "bottleneck", "class", "def", "dense", "dropout", "else", "forward", "if", "intermediate_states", "layer_output", "nn", "not", "residual_tensor_1", "residual_tensor_2", "return", "self", "torch", "use_bottleneck"], "mobilebert/modeling_mobilebert.py:BottleneckLayer.__init__": ["LayerNorm", "Linear", "ModelLayer", "Module", "NORM2FN", "__init__", "class", "config", "def", "dense", "eps", "hidden_size", "intra_Model_size", "layer_norm_eps", "nn", "normalization_type", "self", "super"], "mobilebert/modeling_mobilebert.py:BottleneckLayer.forward": ["LayerNorm", "ModelLayer", "Module", "Tensor", "class", "def", "dense", "forward", "hidden_states", "layer_input", "nn", "return", "self", "torch"], "mobilebert/modeling_mobilebert.py:Bottleneck.__init__": ["Model", "ModelLayer", "Module", "__init__", "attention", "class", "config", "def", "if", "input", "key_query_shared_Model", "nn", "self", "super", "use_Model_attention"], "mobilebert/modeling_mobilebert.py:Bottleneck.forward": ["Model", "Modeled_hidden_states", "Module", "Tensor", "attention", "class", "def", "elif", "else", "forward", "hidden_states", "if", "input", "key_query_shared_Model", "nn", "return", "self", "shared_attention_input", "torch", "tuple", "use_Model_attention"], "mobilebert/modeling_mobilebert.py:FFNOutput.__init__": ["LayerNorm", "Linear", "Model", "Module", "NORM2FN", "__init__", "class", "config", "def", "dense", "eps", "intermediate_size", "layer_norm_eps", "nn", "normalization_type", "self", "super", "true_hidden_size"], "mobilebert/modeling_mobilebert.py:FFNOutput.forward": ["LayerNorm", "Model", "Module", "Tensor", "class", "def", "dense", "forward", "hidden_states", "layer_outputs", "nn", "residual_tensor", "return", "self", "torch"], "mobilebert/modeling_mobilebert.py:FFNLayer.__init__": ["FFNOutput", "Model", "ModelIntermediate", "Module", "__init__", "class", "config", "def", "intermediate", "nn", "output", "self", "super"], "mobilebert/modeling_mobilebert.py:FFNLayer.forward": ["Model", "Module", "Tensor", "class", "def", "forward", "hidden_states", "intermediate", "intermediate_output", "layer_outputs", "nn", "output", "return", "self", "torch"], "mobilebert/modeling_mobilebert.py:MobileBertLayer.__init__": ["Bottleneck", "FFNLayer", "GradientCheckpointingLayer", "ModelAttention", "ModelIntermediate", "ModelLayer", "ModelOutput", "ModuleList", "_", "__init__", "attention", "bottleneck", "class", "config", "def", "ffn", "for", "if", "in", "intermediate", "nn", "num_feedforward_networks", "output", "range", "self", "super", "use_bottleneck"], "mobilebert/modeling_mobilebert.py:MobileBertLayer.forward": ["FloatTensor", "GradientCheckpointingLayer", "ModelLayer", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "attention", "attention_mask", "attention_output", "bottleneck", "class", "def", "else", "ffn", "ffn_module", "for", "forward", "hidden_states", "if", "in", "intermediate", "intermediate_output", "key_tensor", "kwargs", "layer_input", "layer_output", "num_feedforward_networks", "output", "query_tensor", "return", "self", "self_attention_output", "torch", "tuple", "use_bottleneck", "value_tensor"], "mobilebert/modeling_mobilebert.py:MobileBertEncoder.__init__": ["ModelEncoder", "ModelLayer", "Module", "ModuleList", "_", "__init__", "class", "config", "def", "for", "in", "layer", "nn", "num_hidden_layers", "range", "self", "super"], "mobilebert/modeling_mobilebert.py:MobileBertEncoder.forward": ["BaseModelOutput", "FloatTensor", "ModelEncoder", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "class", "def", "enumerate", "for", "forward", "hidden_states", "i", "in", "kwargs", "last_hidden_state", "layer", "layer_module", "nn", "return", "self", "torch", "tuple"], "mobilebert/modeling_mobilebert.py:MobileBertPooler.__init__": ["Linear", "ModelPooler", "Module", "__init__", "class", "classifier_activation", "config", "def", "dense", "do_activate", "hidden_size", "if", "nn", "self", "super"], "mobilebert/modeling_mobilebert.py:MobileBertPooler.forward": ["ModelPooler", "Module", "Tensor", "class", "def", "dense", "do_activate", "else", "first_token_tensor", "forward", "hidden_states", "if", "nn", "not", "pooled_output", "return", "self", "tanh", "torch"], "mobilebert/modeling_mobilebert.py:MobileBertPredictionHeadTransform.__init__": ["ACT2FN", "LayerNorm", "Linear", "ModelPredictionHeadTransform", "Module", "NORM2FN", "__init__", "class", "config", "def", "dense", "else", "eps", "hidden_act", "hidden_size", "if", "isinstance", "layer_norm", "layer_norm_eps", "nn", "self", "str", "super", "transform_act_fn"], "mobilebert/modeling_mobilebert.py:MobileBertPredictionHeadTransform.forward": ["LayerNorm", "ModelPredictionHeadTransform", "Module", "Tensor", "class", "def", "dense", "forward", "hidden_states", "nn", "return", "self", "torch", "transform_act_fn"], "mobilebert/modeling_mobilebert.py:MobileBertLMPredictionHead.__init__": ["False", "Linear", "ModelLMPredictionHead", "ModelPredictionHeadTransform", "Module", "Parameter", "True", "__init__", "bias", "class", "config", "decoder", "def", "dense", "embedding_size", "hidden_size", "nn", "self", "super", "torch", "transform", "vocab_size", "zeros"], "mobilebert/modeling_mobilebert.py:MobileBertLMPredictionHead.forward": ["ModelLMPredictionHead", "Module", "Tensor", "bias", "cat", "class", "decoder", "def", "dense", "dim", "forward", "hidden_states", "matmul", "nn", "return", "self", "t", "torch", "transform", "weight"], "mobilebert/modeling_mobilebert.py:MobileBertOnlyMLMHead.__init__": ["ModelLMPredictionHead", "ModelOnlyMLMHead", "Module", "__init__", "class", "config", "def", "nn", "predictions", "self", "super"], "mobilebert/modeling_mobilebert.py:MobileBertOnlyMLMHead.forward": ["ModelOnlyMLMHead", "Module", "Tensor", "class", "def", "forward", "nn", "prediction_scores", "predictions", "return", "self", "sequence_output", "torch"], "mobilebert/modeling_mobilebert.py:MobileBertPreTrainingHeads.__init__": ["Linear", "ModelLMPredictionHead", "ModelPreTrainingHeads", "Module", "__init__", "class", "config", "def", "hidden_size", "nn", "predictions", "self", "seq_relationship", "super"], "mobilebert/modeling_mobilebert.py:MobileBertPreTrainingHeads.forward": ["ModelPreTrainingHeads", "Module", "Tensor", "class", "def", "forward", "nn", "pooled_output", "prediction_scores", "predictions", "return", "self", "seq_relationship", "seq_relationship_score", "sequence_output", "torch", "tuple"], "mobilebert/modeling_mobilebert.py:MobileBertPreTrainedModel._init_weights": ["ModelEmbeddings", "ModelLMPredictionHead", "ModelPreTrainedModel", "NoNorm", "PreTrainedModel", "_init_weights", "arange", "bias", "class", "copy_", "def", "elif", "expand", "if", "init", "isinstance", "module", "ones_", "position_ids", "self", "shape", "super", "torch", "weight", "zeros_"], "mobilebert/modeling_mobilebert.py:MobileBertModel.__init__": ["False", "ModelEmbeddings", "ModelEncoder", "ModelModel", "ModelPooler", "ModelPreTrainedModel", "None", "True", "__init__", "add_pooling_layer", "class", "config", "def", "else", "embeddings", "encoder", "gradient_checkpointing", "if", "pooler", "post_init", "r", "self", "super"], "mobilebert/modeling_mobilebert.py:MobileBertModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "return", "self", "word_embeddings"], "mobilebert/modeling_mobilebert.py:MobileBertModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "self", "set_input_embeddings", "value", "word_embeddings"], "mobilebert/modeling_mobilebert.py:MobileBertModel.forward": ["BaseModelOutputWithPooling", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "TransformersKwargs", "Unpack", "ValueError", "You", "attention_mask", "class", "config", "create_bidirectional_mask", "def", "else", "embedding_output", "embeddings", "encoder", "encoder_outputs", "exactly", "forward", "if", "input_embeds", "input_ids", "inputs_embeds", "is", "kwargs", "last_hidden_state", "must", "not", "of", "one", "or", "pooled_output", "pooler", "pooler_output", "position_ids", "raise", "return", "self", "sequence_output", "specify", "token_type_ids", "torch", "tuple"], "mobilebert/modeling_mobilebert.py:MobileBertForPreTraining.__init__": ["Model", "ModelForPreTraining", "ModelModel", "ModelPreTrainedModel", "ModelPreTrainingHeads", "__init__", "class", "cls", "config", "def", "post_init", "self", "super"], "mobilebert/modeling_mobilebert.py:MobileBertForPreTraining.get_output_embeddings": ["ModelForPreTraining", "ModelPreTrainedModel", "class", "cls", "decoder", "def", "get_output_embeddings", "predictions", "return", "self"], "mobilebert/modeling_mobilebert.py:MobileBertForPreTraining.set_output_embeddings": ["ModelForPreTraining", "ModelPreTrainedModel", "bias", "class", "cls", "decoder", "def", "new_embeddings", "predictions", "self", "set_output_embeddings"], "mobilebert/modeling_mobilebert.py:MobileBertForPreTraining.resize_token_embeddings": ["Embedding", "ModelForPreTraining", "ModelPreTrainedModel", "None", "True", "_get_resized_lm_head", "class", "cls", "def", "dense", "int", "new_num_tokens", "nn", "predictions", "resize_token_embeddings", "return", "self", "super", "transposed"], "mobilebert/modeling_mobilebert.py:MobileBertForPreTraining.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "Model", "ModelForPreTraining", "ModelForPreTrainingOutput", "ModelPreTrainedModel", "None", "TransformersKwargs", "True", "Unpack", "and", "attention_mask", "attentions", "class", "cls", "config", "def", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "loss", "loss_fct", "masked_lm_loss", "next_sentence_label", "next_sentence_loss", "not", "outputs", "pooled_output", "position_ids", "prediction_logits", "prediction_scores", "r", "return", "return_dict", "self", "seq_relationship_logits", "seq_relationship_score", "sequence_output", "token_type_ids", "torch", "total_loss", "tuple", "view", "vocab_size"], "mobilebert/modeling_mobilebert.py:MobileBertForMaskedLM.__init__": ["False", "Model", "ModelForMaskedLM", "ModelModel", "ModelOnlyMLMHead", "ModelPreTrainedModel", "__init__", "add_pooling_layer", "class", "cls", "config", "def", "post_init", "self", "super"], "mobilebert/modeling_mobilebert.py:MobileBertForMaskedLM.get_output_embeddings": ["ModelForMaskedLM", "ModelPreTrainedModel", "class", "cls", "decoder", "def", "get_output_embeddings", "predictions", "return", "self"], "mobilebert/modeling_mobilebert.py:MobileBertForMaskedLM.set_output_embeddings": ["ModelForMaskedLM", "ModelPreTrainedModel", "bias", "class", "cls", "decoder", "def", "new_embeddings", "predictions", "self", "set_output_embeddings"], "mobilebert/modeling_mobilebert.py:MobileBertForMaskedLM.resize_token_embeddings": ["Embedding", "ModelForMaskedLM", "ModelPreTrainedModel", "None", "True", "_get_resized_lm_head", "class", "cls", "def", "dense", "int", "new_num_tokens", "nn", "predictions", "resize_token_embeddings", "return", "self", "super", "transposed"], "mobilebert/modeling_mobilebert.py:MobileBertForMaskedLM.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "MaskedLMOutput", "Model", "ModelForMaskedLM", "ModelPreTrainedModel", "None", "TransformersKwargs", "True", "Unpack", "attention_mask", "attentions", "class", "cls", "config", "def", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "masked_lm_loss", "not", "outputs", "position_ids", "prediction_scores", "r", "return", "return_dict", "self", "sequence_output", "token_type_ids", "torch", "tuple", "view", "vocab_size"], "mobilebert/modeling_mobilebert.py:MobileBertOnlyNSPHead.__init__": ["Linear", "ModelOnlyNSPHead", "Module", "__init__", "class", "config", "def", "hidden_size", "nn", "self", "seq_relationship", "super"], "mobilebert/modeling_mobilebert.py:MobileBertOnlyNSPHead.forward": ["ModelOnlyNSPHead", "Module", "Tensor", "class", "def", "forward", "nn", "pooled_output", "return", "self", "seq_relationship", "seq_relationship_score", "torch"], "mobilebert/modeling_mobilebert.py:MobileBertForNextSentencePrediction.__init__": ["Model", "ModelForNextSentencePrediction", "ModelModel", "ModelOnlyNSPHead", "ModelPreTrainedModel", "__init__", "class", "cls", "config", "def", "post_init", "self", "super"], "mobilebert/modeling_mobilebert.py:MobileBertForNextSentencePrediction.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "Model", "ModelForNextSentencePrediction", "ModelPreTrainedModel", "NextSentencePredictorOutput", "None", "TransformersKwargs", "True", "Unpack", "attention_mask", "attentions", "class", "cls", "def", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "next_sentence_loss", "not", "outputs", "pooled_output", "position_ids", "r", "return", "return_dict", "self", "seq_relationship_score", "token_type_ids", "torch", "tuple", "view"], "mobilebert/modeling_mobilebert.py:MobileBertForSequenceClassification.__init__": ["Dropout", "Linear", "Model", "ModelForSequenceClassification", "ModelModel", "ModelPreTrainedModel", "None", "__init__", "class", "classifier", "classifier_dropout", "config", "def", "dropout", "else", "hidden_dropout_prob", "hidden_size", "if", "is", "nn", "not", "num_labels", "post_init", "self", "super"], "mobilebert/modeling_mobilebert.py:MobileBertForSequenceClassification.forward": ["BCEWithLogitsLoss", "CrossEntropyLoss", "MSELoss", "Model", "ModelForSequenceClassification", "ModelPreTrainedModel", "None", "SequenceClassifierOutput", "Tensor", "TransformersKwargs", "True", "Unpack", "and", "attention_mask", "attentions", "class", "classifier", "config", "def", "dropout", "dtype", "elif", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "kwargs", "labels", "logits", "long", "loss", "loss_fct", "multi_label_classification", "not", "num_labels", "or", "outputs", "pooled_output", "position_ids", "problem_type", "r", "regression", "return", "return_dict", "self", "single_label_classification", "squeeze", "token_type_ids", "torch", "tuple", "view"], "mobilebert/modeling_mobilebert.py:MobileBertForQuestionAnswering.__init__": ["False", "Linear", "Model", "ModelForQuestionAnswering", "ModelModel", "ModelPreTrainedModel", "__init__", "add_pooling_layer", "class", "config", "def", "hidden_size", "nn", "num_labels", "post_init", "qa_outputs", "self", "super"], "mobilebert/modeling_mobilebert.py:MobileBertForQuestionAnswering.forward": ["CrossEntropyLoss", "Model", "ModelForQuestionAnswering", "ModelPreTrainedModel", "None", "QuestionAnsweringModelOutput", "Tensor", "TransformersKwargs", "True", "Unpack", "and", "attention_mask", "attentions", "clamp", "class", "contiguous", "def", "dim", "end_logits", "end_loss", "end_positions", "forward", "hidden_states", "if", "ignore_index", "ignored_index", "input_ids", "inputs_embeds", "is", "kwargs", "len", "logits", "loss", "loss_fct", "not", "outputs", "position_ids", "qa_outputs", "return", "return_dict", "self", "sequence_output", "size", "split", "squeeze", "start_logits", "start_loss", "start_positions", "token_type_ids", "torch", "total_loss", "tuple"], "mobilebert/modeling_mobilebert.py:MobileBertForMultipleChoice.__init__": ["Dropout", "Linear", "Model", "ModelForMultipleChoice", "ModelModel", "ModelPreTrainedModel", "None", "__init__", "class", "classifier", "classifier_dropout", "config", "def", "dropout", "else", "hidden_dropout_prob", "hidden_size", "if", "is", "nn", "not", "post_init", "self", "super"], "mobilebert/modeling_mobilebert.py:MobileBertForMultipleChoice.forward": ["CrossEntropyLoss", "Model", "ModelForMultipleChoice", "ModelPreTrainedModel", "MultipleChoiceModelOutput", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "attention_mask", "attentions", "class", "classifier", "def", "dropout", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "not", "num_choices", "outputs", "pooled_output", "position_ids", "r", "reshaped_logits", "return", "return_dict", "self", "shape", "size", "token_type_ids", "torch", "tuple", "view"], "mobilebert/modeling_mobilebert.py:MobileBertForTokenClassification.__init__": ["Dropout", "False", "Linear", "Model", "ModelForTokenClassification", "ModelModel", "ModelPreTrainedModel", "None", "__init__", "add_pooling_layer", "class", "classifier", "classifier_dropout", "config", "def", "dropout", "else", "hidden_dropout_prob", "hidden_size", "if", "is", "nn", "not", "num_labels", "post_init", "self", "super"], "mobilebert/modeling_mobilebert.py:MobileBertForTokenClassification.forward": ["CrossEntropyLoss", "Model", "ModelForTokenClassification", "ModelPreTrainedModel", "None", "Tensor", "TokenClassifierOutput", "TransformersKwargs", "True", "Unpack", "attention_mask", "attentions", "class", "classifier", "def", "dropout", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "not", "num_labels", "outputs", "position_ids", "r", "return", "return_dict", "self", "sequence_output", "token_type_ids", "torch", "tuple", "view"], "mobilenet_v1/modeling_mobilenet_v1.py:apply_tf_padding": ["Conv2d", "Model_tf_padding", "Tensor", "constant", "conv_layer", "def", "else", "features", "functional", "if", "in_height", "in_width", "kernel_height", "kernel_size", "kernel_width", "max", "nn", "pad", "pad_along_height", "pad_along_width", "pad_bottom", "pad_left", "pad_right", "pad_top", "padding", "return", "shape", "stride", "stride_height", "stride_width", "torch"], "mobilenet_v1/modeling_mobilenet_v1.py:MobileNetV1ConvLayer.__init__": ["ACT2FN", "BatchNorm2d", "Conv2d", "False", "Input", "ModelConfig", "ModelConvLayer", "Module", "None", "Output", "True", "ValueError", "__init__", "activation", "affine", "are", "bias", "bool", "by", "channels", "class", "config", "convolution", "def", "divisible", "elif", "else", "eps", "f", "groups", "hidden_act", "if", "in_channels", "int", "isinstance", "kernel_size", "layer_norm_eps", "momentum", "nn", "normalization", "not", "num_features", "out_channels", "padding", "padding_mode", "raise", "self", "str", "stride", "super", "tf_padding", "track_running_stats", "use_activation", "use_normalization", "zeros"], "mobilenet_v1/modeling_mobilenet_v1.py:MobileNetV1ConvLayer.forward": ["ModelConvLayer", "Module", "None", "Tensor", "activation", "apply_tf_padding", "class", "config", "convolution", "def", "features", "forward", "if", "is", "nn", "normalization", "not", "return", "self", "tf_padding", "torch"], "mobilenet_v1/modeling_mobilenet_v1.py:MobileNetV1Model.__init__": ["AdaptiveAvgPool2d", "ModelConfig", "ModelConvLayer", "ModelModel", "ModelPreTrainedModel", "ModuleList", "None", "True", "__init__", "add_pooling_layer", "append", "bool", "class", "config", "conv_stem", "def", "depth", "depth_multiplier", "else", "for", "groups", "i", "if", "in", "in_channels", "int", "kernel_size", "layer", "max", "min_depth", "nn", "num_channels", "or", "out_channels", "pooler", "post_init", "r", "range", "self", "stride", "strides", "super"], "mobilenet_v1/modeling_mobilenet_v1.py:MobileNetV1Model.forward": ["BaseModelOutputWithPoolingAndNoAttention", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "ValueError", "You", "all_hidden_states", "bool", "class", "config", "conv_stem", "def", "else", "enumerate", "flatten", "for", "forward", "have", "hidden_states", "i", "if", "in", "is", "kwargs", "last_hidden_state", "layer", "layer_module", "not", "output_hidden_states", "pixel_values", "pooled_output", "pooler", "pooler_output", "raise", "return", "return_dict", "self", "specify", "start_dim", "to", "torch", "tuple", "use_return_dict", "v"], "mobilenet_v1/modeling_mobilenet_v1.py:MobileNetV1ForImageClassification.__init__": ["Dropout", "Identity", "Linear", "Model", "ModelConfig", "ModelForImageClassification", "ModelModel", "ModelPreTrainedModel", "None", "True", "__init__", "class", "classifier", "classifier_dropout_prob", "config", "convolution", "def", "dropout", "else", "if", "inplace", "last_hidden_size", "layer", "nn", "num_labels", "out_channels", "post_init", "self", "super"], "mobilenet_v1/modeling_mobilenet_v1.py:MobileNetV1ForImageClassification.forward": ["ImageClassifierOutputWithNoAttention", "Model", "ModelForImageClassification", "ModelPreTrainedModel", "None", "Tensor", "bool", "class", "classifier", "config", "def", "dropout", "else", "forward", "hidden_states", "if", "is", "kwargs", "labels", "logits", "loss", "loss_function", "not", "output", "output_hidden_states", "outputs", "pixel_values", "pooled_output", "pooler_output", "r", "return", "return_dict", "self", "torch", "tuple", "use_return_dict"], "mobilenet_v2/modeling_mobilenet_v2.py:make_divisible": ["Model_divisible", "None", "def", "divisor", "if", "int", "is", "max", "min_value", "new_value", "return", "value"], "mobilenet_v2/modeling_mobilenet_v2.py:apply_depth_multiplier": ["ModelConfig", "Model_depth_multiplier", "channels", "config", "def", "depth_divisible_by", "depth_multiplier", "int", "make_divisible", "min_depth", "return", "round"], "mobilenet_v2/modeling_mobilenet_v2.py:apply_tf_padding": ["Conv2d", "Model_tf_padding", "Tensor", "constant", "conv_layer", "def", "dilation", "dilation_height", "dilation_width", "else", "features", "functional", "if", "in_height", "in_width", "int", "kernel_height", "kernel_size", "kernel_width", "max", "nn", "pad", "pad_along_height", "pad_along_width", "pad_bottom", "pad_left", "pad_right", "pad_top", "padding", "return", "shape", "stride", "stride_height", "stride_width", "torch"], "mobilenet_v2/modeling_mobilenet_v2.py:MobileNetV2ConvLayer.__init__": ["ACT2FN", "BatchNorm2d", "Conv2d", "False", "Input", "ModelConfig", "ModelConvLayer", "Module", "None", "Output", "True", "ValueError", "__init__", "activation", "affine", "are", "bias", "bool", "by", "channels", "class", "config", "convolution", "def", "dilation", "divisible", "elif", "else", "eps", "f", "float", "groups", "hidden_act", "if", "in_channels", "int", "is", "isinstance", "kernel_size", "layer_norm_eps", "momentum", "nn", "normalization", "not", "num_features", "out_channels", "padding", "padding_mode", "raise", "self", "str", "stride", "super", "tf_padding", "track_running_stats", "use_activation", "use_normalization", "zeros"], "mobilenet_v2/modeling_mobilenet_v2.py:MobileNetV2ConvLayer.forward": ["ModelConvLayer", "Module", "None", "Tensor", "activation", "apply_tf_padding", "class", "config", "convolution", "def", "features", "forward", "if", "is", "nn", "normalization", "not", "return", "self", "tf_padding", "torch"], "mobilenet_v2/modeling_mobilenet_v2.py:MobileNetV2InvertedResidual.__init__": ["False", "Invalid", "ModelConfig", "ModelConvLayer", "ModelInvertedResidual", "Module", "None", "ValueError", "__init__", "and", "class", "config", "conv_3x3", "def", "depth_divisible_by", "dilation", "expand_1x1", "expand_ratio", "expanded_channels", "f", "groups", "if", "in", "in_channels", "int", "kernel_size", "make_divisible", "min_depth", "nn", "not", "out_channels", "raise", "reduce_1x1", "round", "self", "stride", "super", "use_activation", "use_residual"], "mobilenet_v2/modeling_mobilenet_v2.py:MobileNetV2InvertedResidual.forward": ["ModelInvertedResidual", "Module", "Tensor", "class", "conv_3x3", "def", "else", "expand_1x1", "features", "forward", "if", "nn", "reduce_1x1", "residual", "return", "self", "torch", "use_residual"], "mobilenet_v2/modeling_mobilenet_v2.py:MobileNetV2Stem.__init__": ["False", "ModelConfig", "ModelConvLayer", "ModelStem", "Module", "None", "__init__", "class", "config", "conv_3x3", "def", "else", "expand_1x1", "expanded_channels", "first_conv", "first_layer_is_expansion", "groups", "if", "in_channels", "int", "kernel_size", "nn", "out_channels", "reduce_1x1", "self", "stride", "super", "use_activation"], "mobilenet_v2/modeling_mobilenet_v2.py:MobileNetV2Stem.forward": ["ModelStem", "Module", "None", "Tensor", "class", "conv_3x3", "def", "expand_1x1", "features", "first_conv", "forward", "if", "is", "nn", "not", "reduce_1x1", "return", "self", "torch"], "mobilenet_v2/modeling_mobilenet_v2.py:MobileNetV2Model.__init__": ["AdaptiveAvgPool2d", "ModelConfig", "ModelConvLayer", "ModelInvertedResidual", "ModelModel", "ModelPreTrainedModel", "ModelStem", "ModuleList", "None", "True", "__init__", "add_pooling_layer", "and", "append", "apply_depth_multiplier", "bool", "channels", "class", "config", "conv_1x1", "conv_stem", "current_stride", "def", "depth_multiplier", "dilation", "else", "expanded_channels", "finegrained_output", "for", "i", "if", "in", "in_channels", "kernel_size", "layer", "layer_dilation", "layer_stride", "nn", "num_channels", "out_channels", "output_channels", "output_stride", "pooler", "post_init", "r", "range", "self", "stride", "strides", "super", "x"], "mobilenet_v2/modeling_mobilenet_v2.py:MobileNetV2Model.forward": ["BaseModelOutputWithPoolingAndNoAttention", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "ValueError", "You", "all_hidden_states", "bool", "class", "config", "conv_1x1", "conv_stem", "def", "else", "enumerate", "flatten", "for", "forward", "have", "hidden_states", "i", "if", "in", "is", "kwargs", "last_hidden_state", "layer", "layer_module", "not", "output_hidden_states", "pixel_values", "pooled_output", "pooler", "pooler_output", "raise", "return", "return_dict", "self", "specify", "start_dim", "to", "torch", "tuple", "use_return_dict", "v"], "mobilenet_v2/modeling_mobilenet_v2.py:MobileNetV2ForImageClassification.__init__": ["Dropout", "Identity", "Linear", "Model", "ModelConfig", "ModelForImageClassification", "ModelModel", "ModelPreTrainedModel", "None", "True", "__init__", "class", "classifier", "classifier_dropout_prob", "config", "conv_1x1", "convolution", "def", "dropout", "else", "if", "inplace", "last_hidden_size", "nn", "num_labels", "out_channels", "post_init", "self", "super"], "mobilenet_v2/modeling_mobilenet_v2.py:MobileNetV2ForImageClassification.forward": ["ImageClassifierOutputWithNoAttention", "Model", "ModelForImageClassification", "ModelPreTrainedModel", "None", "Tensor", "bool", "class", "classifier", "config", "def", "dropout", "else", "forward", "hidden_states", "if", "is", "kwargs", "labels", "logits", "loss", "loss_function", "not", "output", "output_hidden_states", "outputs", "pixel_values", "pooled_output", "pooler_output", "r", "return", "return_dict", "self", "torch", "tuple", "use_return_dict"], "mobilenet_v2/modeling_mobilenet_v2.py:MobileNetV2DeepLabV3Plus.__init__": ["AdaptiveAvgPool2d", "Dropout2d", "False", "ModelConfig", "ModelConvLayer", "ModelDeepLabV3Plus", "Module", "None", "True", "__init__", "apply_depth_multiplier", "avg_pool", "bias", "class", "classifier", "classifier_dropout_prob", "config", "conv_aspp", "conv_pool", "conv_projection", "def", "dropout", "in_channels", "kernel_size", "layer_norm_eps", "nn", "num_labels", "out_channels", "output_size", "relu", "self", "stride", "super", "use_activation", "use_normalization"], "mobilenet_v2/modeling_mobilenet_v2.py:MobileNetV2DeepLabV3Plus.forward": ["ModelDeepLabV3Plus", "Module", "Tensor", "True", "align_corners", "avg_pool", "bilinear", "cat", "class", "classifier", "conv_aspp", "conv_pool", "conv_projection", "def", "dim", "dropout", "features", "features_aspp", "features_pool", "forward", "functional", "interpolate", "mode", "nn", "return", "self", "shape", "size", "spatial_size", "torch"], "mobilenet_v2/modeling_mobilenet_v2.py:MobileNetV2ForSemanticSegmentation.__init__": ["False", "Model", "ModelConfig", "ModelDeepLabV3Plus", "ModelForSemanticSegmentation", "ModelModel", "ModelPreTrainedModel", "None", "__init__", "add_pooling_layer", "class", "config", "def", "num_labels", "post_init", "segmentation_head", "self", "super"], "mobilenet_v2/modeling_mobilenet_v2.py:MobileNetV2ForSemanticSegmentation.forward": ["CrossEntropyLoss", "False", "Model", "ModelForSemanticSegmentation", "ModelPreTrainedModel", "None", "SemanticSegmenterOutput", "Tensor", "The", "True", "ValueError", "align_corners", "and", "attentions", "be", "bilinear", "bool", "class", "config", "def", "else", "encoder_hidden_states", "forward", "functional", "greater", "hidden_states", "if", "ignore_index", "interpolate", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "mode", "nn", "not", "num_labels", "number", "of", "one", "output", "output_hidden_states", "outputs", "pixel_values", "r", "raise", "return", "return_dict", "segmentation_head", "self", "semantic_loss_ignore_index", "shape", "should", "size", "than", "torch", "tuple", "upsampled_logits", "use_return_dict"], "mobilevit/modeling_mobilevit.py:make_divisible": ["Model_divisible", "None", "def", "divisor", "if", "int", "is", "max", "min_value", "new_value", "return", "value"], "mobilevit/modeling_mobilevit.py:MobileViTConvLayer.__init__": ["ACT2FN", "BatchNorm2d", "Conv2d", "False", "Input", "ModelConfig", "ModelConvLayer", "Module", "None", "Output", "True", "ValueError", "__init__", "activation", "affine", "are", "bias", "bool", "by", "channels", "class", "config", "convolution", "def", "dilation", "divisible", "elif", "else", "eps", "f", "groups", "hidden_act", "if", "in_channels", "int", "isinstance", "kernel_size", "momentum", "nn", "normalization", "not", "num_features", "out_channels", "padding", "padding_mode", "raise", "self", "str", "stride", "super", "track_running_stats", "use_activation", "use_normalization", "zeros"], "mobilevit/modeling_mobilevit.py:MobileViTConvLayer.forward": ["ModelConvLayer", "Module", "None", "Tensor", "activation", "class", "convolution", "def", "features", "forward", "if", "is", "nn", "normalization", "not", "return", "self", "torch"], "mobilevit/modeling_mobilevit.py:MobileViTInvertedResidual.__init__": ["False", "Invalid", "ModelConfig", "ModelConvLayer", "ModelInvertedResidual", "Module", "None", "ValueError", "__init__", "and", "class", "config", "conv_3x3", "def", "dilation", "expand_1x1", "expand_ratio", "expanded_channels", "f", "groups", "if", "in", "in_channels", "int", "kernel_size", "make_divisible", "nn", "not", "out_channels", "raise", "reduce_1x1", "round", "self", "stride", "super", "use_activation", "use_residual"], "mobilevit/modeling_mobilevit.py:MobileViTInvertedResidual.forward": ["ModelInvertedResidual", "Module", "Tensor", "class", "conv_3x3", "def", "else", "expand_1x1", "features", "forward", "if", "nn", "reduce_1x1", "residual", "return", "self", "torch", "use_residual"], "mobilevit/modeling_mobilevit.py:MobileViTMobileNetLayer.__init__": ["ModelConfig", "ModelInvertedResidual", "ModelModelNetLayer", "Module", "ModuleList", "None", "__init__", "append", "class", "config", "def", "else", "for", "i", "if", "in", "in_channels", "int", "layer", "nn", "num_stages", "out_channels", "range", "self", "stride", "super"], "mobilevit/modeling_mobilevit.py:MobileViTMobileNetLayer.forward": ["ModelModelNetLayer", "Module", "Tensor", "class", "def", "features", "for", "forward", "in", "layer", "layer_module", "nn", "return", "self", "torch"], "mobilevit/modeling_mobilevit.py:MobileViTSelfAttention.__init__": ["Dropout", "Linear", "ModelConfig", "ModelSelfAttention", "Module", "None", "The", "ValueError", "__init__", "a", "all_head_size", "attention", "attention_head_size", "attention_probs_dropout_prob", "bias", "class", "config", "def", "dropout", "f", "heads", "hidden", "hidden_size", "if", "int", "is", "key", "multiple", "nn", "not", "num_attention_heads", "number", "of", "qkv_bias", "query", "raise", "self", "size", "super", "the", "value"], "mobilevit/modeling_mobilevit.py:MobileViTSelfAttention.forward": ["ModelSelfAttention", "Module", "Tensor", "_", "all_head_size", "attention_head_size", "attention_probs", "attention_scores", "batch_size", "class", "context_layer", "contiguous", "def", "dim", "dropout", "forward", "functional", "hidden_states", "key", "key_layer", "math", "matmul", "new_context_layer_shape", "nn", "num_attention_heads", "permute", "query", "query_layer", "return", "self", "seq_length", "shape", "size", "softmax", "sqrt", "torch", "transpose", "value", "value_layer", "view"], "mobilevit/modeling_mobilevit.py:MobileViTSelfOutput.__init__": ["Dropout", "Linear", "ModelConfig", "ModelSelfOutput", "Module", "None", "__init__", "class", "config", "def", "dense", "dropout", "hidden_dropout_prob", "hidden_size", "int", "nn", "self", "super"], "mobilevit/modeling_mobilevit.py:MobileViTSelfOutput.forward": ["ModelSelfOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "nn", "return", "self", "torch"], "mobilevit/modeling_mobilevit.py:MobileViTAttention.__init__": ["ModelAttention", "ModelConfig", "ModelSelfAttention", "ModelSelfOutput", "Module", "None", "__init__", "attention", "class", "config", "def", "hidden_size", "int", "nn", "output", "self", "super"], "mobilevit/modeling_mobilevit.py:MobileViTAttention.forward": ["ModelAttention", "Module", "Tensor", "attention", "attention_output", "class", "def", "forward", "hidden_states", "nn", "output", "return", "self", "self_outputs", "torch"], "mobilevit/modeling_mobilevit.py:MobileViTIntermediate.__init__": ["ACT2FN", "Linear", "ModelConfig", "ModelIntermediate", "Module", "None", "__init__", "class", "config", "def", "dense", "else", "hidden_act", "hidden_size", "if", "int", "intermediate_act_fn", "intermediate_size", "isinstance", "nn", "self", "str", "super"], "mobilevit/modeling_mobilevit.py:MobileViTIntermediate.forward": ["ModelIntermediate", "Module", "Tensor", "class", "def", "dense", "forward", "hidden_states", "intermediate_act_fn", "nn", "return", "self", "torch"], "mobilevit/modeling_mobilevit.py:MobileViTOutput.__init__": ["Dropout", "Linear", "ModelConfig", "ModelOutput", "Module", "None", "__init__", "class", "config", "def", "dense", "dropout", "hidden_dropout_prob", "hidden_size", "int", "intermediate_size", "nn", "self", "super"], "mobilevit/modeling_mobilevit.py:MobileViTOutput.forward": ["ModelOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "mobilevit/modeling_mobilevit.py:MobileViTTransformerLayer.__init__": ["LayerNorm", "ModelAttention", "ModelConfig", "ModelIntermediate", "ModelOutput", "ModelTransformerLayer", "Module", "None", "__init__", "attention", "class", "config", "def", "eps", "hidden_size", "int", "intermediate", "intermediate_size", "layer_norm_eps", "layernorm_after", "layernorm_before", "nn", "output", "self", "super"], "mobilevit/modeling_mobilevit.py:MobileViTTransformerLayer.forward": ["ModelTransformerLayer", "Module", "Tensor", "attention", "attention_output", "class", "def", "forward", "hidden_states", "intermediate", "layer_output", "layernorm_after", "layernorm_before", "nn", "output", "return", "self", "torch"], "mobilevit/modeling_mobilevit.py:MobileViTTransformer.__init__": ["ModelConfig", "ModelTransformer", "ModelTransformerLayer", "Module", "ModuleList", "None", "_", "__init__", "append", "class", "config", "def", "for", "hidden_size", "in", "int", "intermediate_size", "layer", "mlp_ratio", "nn", "num_stages", "range", "self", "super", "transformer_layer"], "mobilevit/modeling_mobilevit.py:MobileViTTransformer.forward": ["ModelTransformer", "Module", "Tensor", "class", "def", "for", "forward", "hidden_states", "in", "layer", "layer_module", "nn", "return", "self", "torch"], "mobilevit/modeling_mobilevit.py:MobileViTLayer.__init__": ["False", "GradientCheckpointingLayer", "LayerNorm", "ModelConfig", "ModelConvLayer", "ModelInvertedResidual", "ModelLayer", "ModelTransformer", "None", "__init__", "class", "config", "conv_1x1", "conv_kernel_size", "conv_kxk", "conv_projection", "def", "dilation", "downsampling_layer", "else", "eps", "fusion", "hidden_size", "if", "in_channels", "int", "kernel_size", "layer_norm_eps", "layernorm", "nn", "num_stages", "out_channels", "patch_height", "patch_size", "patch_width", "self", "stride", "super", "transformer", "use_activation", "use_normalization"], "mobilevit/modeling_mobilevit.py:MobileViTLayer.unfolding": ["False", "GradientCheckpointingLayer", "ModelLayer", "Tensor", "True", "align_corners", "batch_size", "bilinear", "ceil", "channels", "class", "def", "dict", "else", "features", "functional", "if", "info_dict", "int", "interpolate", "is_tracing", "jit", "math", "mode", "new_height", "new_width", "nn", "num_patch_height", "num_patch_width", "num_patches", "num_patches_height", "num_patches_width", "or", "orig_height", "orig_size", "orig_width", "patch_area", "patch_height", "patch_width", "patches", "reshape", "return", "self", "shape", "size", "torch", "torch_int", "transpose", "tuple", "unfolding"], "mobilevit/modeling_mobilevit.py:MobileViTLayer.folding": ["False", "GradientCheckpointingLayer", "ModelLayer", "Tensor", "align_corners", "batch_size", "bilinear", "channels", "class", "contiguous", "def", "dict", "features", "folding", "functional", "if", "info_dict", "int", "interpolate", "mode", "nn", "num_patch_height", "num_patch_width", "num_patches", "num_patches_height", "num_patches_width", "orig_size", "patch_area", "patch_height", "patch_width", "patches", "reshape", "return", "self", "size", "torch", "transpose", "view"], "mobilevit/modeling_mobilevit.py:MobileViTLayer.forward": ["GradientCheckpointingLayer", "ModelLayer", "Tensor", "cat", "class", "conv_1x1", "conv_kxk", "conv_projection", "def", "dim", "downsampling_layer", "features", "folding", "forward", "fusion", "if", "info_dict", "layernorm", "patches", "residual", "return", "self", "torch", "transformer", "unfolding"], "mobilevit/modeling_mobilevit.py:MobileViTEncoder.__init__": ["False", "ModelConfig", "ModelEncoder", "ModelLayer", "ModelModelNetLayer", "Module", "ModuleList", "None", "True", "__init__", "append", "class", "config", "def", "dilate_layer_4", "dilate_layer_5", "dilation", "elif", "gradient_checkpointing", "hidden_size", "hidden_sizes", "if", "in_channels", "layer", "layer_1", "layer_2", "layer_3", "layer_4", "layer_5", "neck_hidden_sizes", "nn", "num_stages", "out_channels", "output_stride", "self", "stride", "super"], "mobilevit/modeling_mobilevit.py:MobileViTEncoder.forward": ["BaseModelOutputWithNoAttention", "False", "ModelEncoder", "Module", "None", "Tensor", "True", "all_hidden_states", "bool", "class", "def", "else", "enumerate", "for", "forward", "hidden_states", "i", "if", "in", "is", "last_hidden_state", "layer", "layer_module", "nn", "not", "output_hidden_states", "return", "return_dict", "self", "torch", "tuple", "v"], "mobilevit/modeling_mobilevit.py:MobileViTPreTrainedModel._init_weights": ["BatchNorm2d", "Conv2d", "LayerNorm", "Linear", "ModelPreTrainedModel", "Module", "None", "PreTrainedModel", "_init_weights", "bias", "class", "config", "def", "elif", "getattr", "if", "init", "initializer_range", "is", "isinstance", "mean", "module", "nn", "normal_", "not", "num_batches_tracked", "ones_", "running_mean", "running_var", "self", "std", "weight", "zeros_"], "mobilevit/modeling_mobilevit.py:MobileViTModel.__init__": ["ModelConfig", "ModelConvLayer", "ModelEncoder", "ModelModel", "ModelPreTrainedModel", "True", "__init__", "bool", "class", "config", "conv_1x1_exp", "conv_stem", "def", "encoder", "expand_output", "if", "in_channels", "kernel_size", "neck_hidden_sizes", "num_channels", "out_channels", "post_init", "r", "self", "stride", "super"], "mobilevit/modeling_mobilevit.py:MobileViTModel.forward": ["BaseModelOutputWithPoolingAndNoAttention", "False", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "ValueError", "You", "bool", "class", "config", "conv_1x1_exp", "conv_stem", "def", "dim", "else", "embedding_output", "encoder", "encoder_outputs", "expand_output", "forward", "have", "hidden_states", "if", "is", "keepdim", "kwargs", "last_hidden_state", "mean", "not", "output", "output_hidden_states", "pixel_values", "pooled_output", "pooler_output", "raise", "return", "return_dict", "self", "specify", "to", "torch", "tuple", "use_return_dict"], "mobilevit/modeling_mobilevit.py:MobileViTForImageClassification.__init__": ["Dropout", "Identity", "Linear", "Model", "ModelConfig", "ModelForImageClassification", "ModelModel", "ModelPreTrainedModel", "None", "True", "__init__", "class", "classifier", "classifier_dropout_prob", "config", "def", "dropout", "else", "if", "inplace", "neck_hidden_sizes", "nn", "num_labels", "post_init", "self", "super"], "mobilevit/modeling_mobilevit.py:MobileViTForImageClassification.forward": ["ImageClassifierOutputWithNoAttention", "Model", "ModelForImageClassification", "ModelPreTrainedModel", "None", "Tensor", "bool", "class", "classifier", "config", "def", "dropout", "else", "forward", "hidden_states", "if", "is", "kwargs", "labels", "logits", "loss", "loss_function", "not", "output", "output_hidden_states", "outputs", "pixel_values", "pooled_output", "pooler_output", "r", "return", "return_dict", "self", "torch", "tuple", "use_return_dict"], "mobilevit/modeling_mobilevit.py:MobileViTASPPPooling.__init__": ["AdaptiveAvgPool2d", "ModelASPPPooling", "ModelConfig", "ModelConvLayer", "Module", "None", "True", "__init__", "class", "config", "conv_1x1", "def", "global_pool", "in_channels", "int", "kernel_size", "nn", "out_channels", "output_size", "relu", "self", "stride", "super", "use_activation", "use_normalization"], "mobilevit/modeling_mobilevit.py:MobileViTASPPPooling.forward": ["False", "ModelASPPPooling", "Module", "Tensor", "align_corners", "bilinear", "class", "conv_1x1", "def", "features", "forward", "functional", "global_pool", "interpolate", "mode", "nn", "return", "self", "shape", "size", "spatial_size", "torch"], "mobilevit/modeling_mobilevit.py:MobileViTASPP.__init__": ["Dropout", "Expected", "ModelASPP", "ModelASPPPooling", "ModelConfig", "ModelConvLayer", "Module", "ModuleList", "None", "ValueError", "__init__", "append", "aspp_dropout_prob", "aspp_out_channels", "atrous_rates", "class", "config", "convs", "def", "dilation", "dropout", "extend", "for", "if", "in", "in_channels", "in_projection", "kernel_size", "len", "neck_hidden_sizes", "nn", "out_channels", "p", "pool_layer", "project", "raise", "rate", "relu", "self", "super", "use_activation", "values"], "mobilevit/modeling_mobilevit.py:MobileViTASPP.forward": ["ModelASPP", "Module", "Tensor", "append", "cat", "class", "conv", "convs", "def", "dim", "dropout", "features", "for", "forward", "in", "nn", "pooled_features", "project", "pyramid", "return", "self", "torch"], "mobilevit/modeling_mobilevit.py:MobileViTDeepLabV3.__init__": ["Dropout2d", "False", "ModelASPP", "ModelConfig", "ModelConvLayer", "ModelDeepLabV3", "Module", "None", "True", "__init__", "aspp", "aspp_out_channels", "bias", "class", "classifier", "classifier_dropout_prob", "config", "def", "dropout", "in_channels", "kernel_size", "nn", "num_labels", "out_channels", "self", "super", "use_activation", "use_normalization"], "mobilevit/modeling_mobilevit.py:MobileViTDeepLabV3.forward": ["ModelDeepLabV3", "Module", "Tensor", "aspp", "class", "classifier", "def", "dropout", "features", "forward", "hidden_states", "nn", "return", "self", "torch"], "mobilevit/modeling_mobilevit.py:MobileViTForSemanticSegmentation.__init__": ["False", "Model", "ModelConfig", "ModelDeepLabV3", "ModelForSemanticSegmentation", "ModelModel", "ModelPreTrainedModel", "None", "__init__", "class", "config", "def", "expand_output", "num_labels", "post_init", "segmentation_head", "self", "super"], "mobilevit/modeling_mobilevit.py:MobileViTForSemanticSegmentation.forward": ["CrossEntropyLoss", "False", "Model", "ModelForSemanticSegmentation", "ModelPreTrainedModel", "None", "SemanticSegmenterOutput", "Tensor", "The", "True", "ValueError", "align_corners", "and", "attentions", "be", "bilinear", "bool", "class", "config", "def", "else", "encoder_hidden_states", "forward", "functional", "greater", "hidden_states", "if", "ignore_index", "interpolate", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "mode", "nn", "not", "num_labels", "number", "of", "one", "output", "output_hidden_states", "outputs", "pixel_values", "r", "raise", "return", "return_dict", "segmentation_head", "self", "semantic_loss_ignore_index", "shape", "should", "size", "than", "torch", "tuple", "upsampled_logits", "use_return_dict"], "mobilevitv2/modeling_mobilevitv2.py:make_divisible": ["Model_divisible", "None", "def", "divisor", "if", "int", "is", "max", "min_value", "new_value", "return", "value"], "mobilevitv2/modeling_mobilevitv2.py:clip": ["Model", "def", "float", "inf", "max", "max_val", "min", "min_val", "return", "value"], "mobilevitv2/modeling_mobilevitv2.py:MobileViTV2ConvLayer.__init__": ["ACT2FN", "BatchNorm2d", "Conv2d", "False", "Input", "ModelConfig", "ModelConvLayer", "Module", "None", "Output", "True", "ValueError", "__init__", "activation", "affine", "are", "bias", "bool", "by", "channels", "class", "config", "convolution", "def", "dilation", "divisible", "elif", "else", "eps", "f", "groups", "hidden_act", "if", "in_channels", "int", "isinstance", "kernel_size", "momentum", "nn", "normalization", "not", "num_features", "out_channels", "padding", "padding_mode", "raise", "self", "str", "stride", "super", "track_running_stats", "use_activation", "use_normalization", "zeros"], "mobilevitv2/modeling_mobilevitv2.py:MobileViTV2ConvLayer.forward": ["ModelConvLayer", "Module", "None", "Tensor", "activation", "class", "convolution", "def", "features", "forward", "if", "is", "nn", "normalization", "not", "return", "self", "torch"], "mobilevitv2/modeling_mobilevitv2.py:MobileViTV2InvertedResidual.__init__": ["False", "Invalid", "ModelConfig", "ModelConvLayer", "ModelInvertedResidual", "Module", "None", "ValueError", "__init__", "and", "class", "config", "conv_3x3", "def", "dilation", "expand_1x1", "expand_ratio", "expanded_channels", "f", "groups", "if", "in", "in_channels", "int", "kernel_size", "make_divisible", "nn", "not", "out_channels", "raise", "reduce_1x1", "round", "self", "stride", "super", "use_activation", "use_residual"], "mobilevitv2/modeling_mobilevitv2.py:MobileViTV2InvertedResidual.forward": ["ModelInvertedResidual", "Module", "Tensor", "class", "conv_3x3", "def", "else", "expand_1x1", "features", "forward", "if", "nn", "reduce_1x1", "residual", "return", "self", "torch", "use_residual"], "mobilevitv2/modeling_mobilevitv2.py:MobileViTV2MobileNetLayer.__init__": ["ModelConfig", "ModelInvertedResidual", "ModelModelNetLayer", "Module", "ModuleList", "None", "__init__", "append", "class", "config", "def", "else", "for", "i", "if", "in", "in_channels", "int", "layer", "nn", "num_stages", "out_channels", "range", "self", "stride", "super"], "mobilevitv2/modeling_mobilevitv2.py:MobileViTV2MobileNetLayer.forward": ["ModelModelNetLayer", "Module", "Tensor", "class", "def", "features", "for", "forward", "in", "layer", "layer_module", "nn", "return", "self", "torch"], "mobilevitv2/modeling_mobilevitv2.py:MobileViTV2LinearSelfAttention.__init__": ["Dropout", "False", "ModelConfig", "ModelConvLayer", "ModelLinearSelfAttention", "Module", "None", "True", "__init__", "attn_dropout", "bias", "class", "config", "def", "embed_dim", "in_channels", "int", "kernel_size", "nn", "out_channels", "out_proj", "p", "qkv_proj", "self", "super", "use_activation", "use_normalization"], "mobilevitv2/modeling_mobilevitv2.py:MobileViTV2LinearSelfAttention.forward": ["ModelLinearSelfAttention", "Module", "Tensor", "True", "attn_dropout", "class", "context_scores", "context_vector", "def", "dim", "embed_dim", "expand_as", "forward", "functional", "hidden_states", "keepdim", "key", "nn", "out", "out_proj", "qkv", "qkv_proj", "query", "relu", "return", "self", "softmax", "split", "split_size_or_sections", "sum", "torch", "value"], "mobilevitv2/modeling_mobilevitv2.py:MobileViTV2FFN.__init__": ["Dropout", "False", "ModelConfig", "ModelConvLayer", "ModelFFN", "Module", "None", "True", "__init__", "bias", "class", "config", "conv1", "conv2", "def", "dropout1", "dropout2", "embed_dim", "ffn_dropout", "ffn_latent_dim", "float", "in_channels", "int", "kernel_size", "nn", "out_channels", "self", "stride", "super", "use_activation", "use_normalization"], "mobilevitv2/modeling_mobilevitv2.py:MobileViTV2FFN.forward": ["ModelFFN", "Module", "Tensor", "class", "conv1", "conv2", "def", "dropout1", "dropout2", "forward", "hidden_states", "nn", "return", "self", "torch"], "mobilevitv2/modeling_mobilevitv2.py:MobileViTV2TransformerLayer.__init__": ["Dropout", "GroupNorm", "ModelConfig", "ModelFFN", "ModelLinearSelfAttention", "ModelTransformerLayer", "Module", "None", "__init__", "attention", "class", "config", "def", "dropout", "dropout1", "embed_dim", "eps", "ffn", "ffn_dropout", "ffn_latent_dim", "float", "int", "layer_norm_eps", "layernorm_after", "layernorm_before", "nn", "num_channels", "num_groups", "p", "self", "super"], "mobilevitv2/modeling_mobilevitv2.py:MobileViTV2TransformerLayer.forward": ["ModelTransformerLayer", "Module", "Tensor", "attention", "attention_output", "class", "def", "ffn", "forward", "hidden_states", "layer_output", "layernorm_1_out", "layernorm_after", "layernorm_before", "nn", "return", "self", "torch"], "mobilevitv2/modeling_mobilevitv2.py:MobileViTV2Transformer.__init__": ["ModelConfig", "ModelTransformer", "ModelTransformerLayer", "Module", "ModuleList", "None", "__init__", "append", "block_idx", "class", "config", "d", "d_model", "def", "embed_dim", "ffn_dims", "ffn_latent_dim", "ffn_multiplier", "for", "in", "int", "layer", "n_layers", "nn", "range", "self", "super", "transformer_layer"], "mobilevitv2/modeling_mobilevitv2.py:MobileViTV2Transformer.forward": ["ModelTransformer", "Module", "Tensor", "class", "def", "for", "forward", "hidden_states", "in", "layer", "layer_module", "nn", "return", "self", "torch"], "mobilevitv2/modeling_mobilevitv2.py:MobileViTV2Layer.__init__": ["False", "GradientCheckpointingLayer", "GroupNorm", "ModelConfig", "ModelConvLayer", "ModelInvertedResidual", "ModelLayer", "ModelTransformer", "None", "True", "__init__", "attn_unit_dim", "class", "cnn_out_dim", "config", "conv_1x1", "conv_kernel_size", "conv_kxk", "conv_projection", "d_model", "def", "dilation", "downsampling_layer", "else", "eps", "groups", "if", "in_channels", "int", "kernel_size", "layer_norm_eps", "layernorm", "n_attn_blocks", "n_layers", "nn", "num_channels", "num_groups", "out_channels", "patch_height", "patch_size", "patch_width", "self", "stride", "super", "transformer", "use_activation", "use_normalization"], "mobilevitv2/modeling_mobilevitv2.py:MobileViTV2Layer.unfolding": ["GradientCheckpointingLayer", "ModelLayer", "Tensor", "batch_size", "class", "def", "feature_map", "functional", "img_height", "img_width", "in_channels", "int", "kernel_size", "nn", "patch_height", "patch_width", "patches", "reshape", "return", "self", "shape", "stride", "torch", "tuple", "unfold", "unfolding"], "mobilevitv2/modeling_mobilevitv2.py:MobileViTV2Layer.folding": ["GradientCheckpointingLayer", "ModelLayer", "Tensor", "batch_size", "class", "def", "feature_map", "fold", "folding", "functional", "in_dim", "int", "kernel_size", "n_patches", "nn", "output_size", "patch_height", "patch_size", "patch_width", "patches", "reshape", "return", "self", "shape", "stride", "torch", "tuple"], "mobilevitv2/modeling_mobilevitv2.py:MobileViTV2Layer.forward": ["GradientCheckpointingLayer", "ModelLayer", "Tensor", "class", "conv_1x1", "conv_kxk", "conv_projection", "def", "downsampling_layer", "features", "folding", "forward", "if", "layernorm", "output_size", "patches", "return", "self", "torch", "transformer", "unfolding"], "mobilevitv2/modeling_mobilevitv2.py:MobileViTV2Encoder.__init__": ["False", "ModelConfig", "ModelEncoder", "ModelLayer", "ModelModelNetLayer", "Module", "ModuleList", "None", "True", "__init__", "append", "attn_unit_dim", "base_attn_unit_dims", "class", "clip", "config", "def", "dilate_layer_4", "dilate_layer_5", "dilation", "divisor", "elif", "gradient_checkpointing", "if", "in_channels", "layer", "layer_0_dim", "layer_1", "layer_1_dim", "layer_2", "layer_2_dim", "layer_3", "layer_3_dim", "layer_4", "layer_4_dim", "layer_5", "layer_5_dim", "make_divisible", "max_val", "min_val", "min_value", "n_attn_blocks", "nn", "num_stages", "out_channels", "output_stride", "self", "stride", "super", "value", "width_multiplier"], "mobilevitv2/modeling_mobilevitv2.py:MobileViTV2Encoder.forward": ["BaseModelOutputWithNoAttention", "False", "ModelEncoder", "Module", "None", "Tensor", "True", "all_hidden_states", "bool", "class", "def", "else", "enumerate", "for", "forward", "hidden_states", "i", "if", "in", "is", "last_hidden_state", "layer", "layer_module", "nn", "not", "output_hidden_states", "return", "return_dict", "self", "torch", "tuple", "v"], "mobilevitv2/modeling_mobilevitv2.py:MobileViTV2PreTrainedModel._init_weights": ["BatchNorm2d", "Conv2d", "GroupNorm", "Linear", "ModelPreTrainedModel", "Module", "None", "PreTrainedModel", "_init_weights", "bias", "class", "config", "def", "elif", "getattr", "if", "init", "initializer_range", "is", "isinstance", "mean", "module", "nn", "normal_", "not", "num_batches_tracked", "ones_", "running_mean", "running_var", "self", "std", "weight", "zeros_"], "mobilevitv2/modeling_mobilevitv2.py:MobileViTV2Model.__init__": ["ModelConfig", "ModelConvLayer", "ModelEncoder", "ModelModel", "ModelPreTrainedModel", "True", "__init__", "bool", "class", "clip", "config", "conv_stem", "def", "divisor", "encoder", "expand_output", "in_channels", "kernel_size", "layer_0_dim", "make_divisible", "max_val", "min_val", "min_value", "num_channels", "out_channels", "post_init", "r", "self", "stride", "super", "use_activation", "use_normalization", "value", "width_multiplier"], "mobilevitv2/modeling_mobilevitv2.py:MobileViTV2Model.forward": ["BaseModelOutputWithPoolingAndNoAttention", "False", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "ValueError", "You", "bool", "class", "config", "conv_stem", "def", "dim", "else", "embedding_output", "encoder", "encoder_outputs", "expand_output", "forward", "have", "hidden_states", "if", "is", "keepdim", "kwargs", "last_hidden_state", "mean", "not", "output", "output_hidden_states", "pixel_values", "pooled_output", "pooler_output", "raise", "return", "return_dict", "self", "specify", "to", "torch", "tuple", "use_return_dict"], "mobilevitv2/modeling_mobilevitv2.py:MobileViTV2ForImageClassification.__init__": ["Identity", "Linear", "Model", "ModelConfig", "ModelForImageClassification", "ModelModel", "ModelPreTrainedModel", "None", "__init__", "class", "classifier", "config", "def", "divisor", "else", "if", "in_features", "make_divisible", "nn", "num_labels", "out_channels", "out_features", "post_init", "self", "super", "width_multiplier"], "mobilevitv2/modeling_mobilevitv2.py:MobileViTV2ForImageClassification.forward": ["ImageClassifierOutputWithNoAttention", "Model", "ModelForImageClassification", "ModelPreTrainedModel", "None", "Tensor", "bool", "class", "classifier", "config", "def", "else", "forward", "hidden_states", "if", "is", "kwargs", "labels", "logits", "loss", "loss_function", "not", "output", "output_hidden_states", "outputs", "pixel_values", "pooled_output", "pooler_output", "r", "return", "return_dict", "self", "torch", "tuple", "use_return_dict"], "mobilevitv2/modeling_mobilevitv2.py:MobileViTV2ASPPPooling.__init__": ["AdaptiveAvgPool2d", "ModelASPPPooling", "ModelConfig", "ModelConvLayer", "Module", "None", "True", "__init__", "class", "config", "conv_1x1", "def", "global_pool", "in_channels", "int", "kernel_size", "nn", "out_channels", "output_size", "relu", "self", "stride", "super", "use_activation", "use_normalization"], "mobilevitv2/modeling_mobilevitv2.py:MobileViTV2ASPPPooling.forward": ["False", "ModelASPPPooling", "Module", "Tensor", "align_corners", "bilinear", "class", "conv_1x1", "def", "features", "forward", "functional", "global_pool", "interpolate", "mode", "nn", "return", "self", "shape", "size", "spatial_size", "torch"], "mobilevitv2/modeling_mobilevitv2.py:MobileViTV2ASPP.__init__": ["Dropout", "Expected", "ModelASPP", "ModelASPPPooling", "ModelConfig", "ModelConvLayer", "Module", "ModuleList", "None", "ValueError", "__init__", "append", "aspp_dropout_prob", "aspp_out_channels", "atrous_rates", "class", "config", "convs", "def", "dilation", "divisor", "dropout", "encoder_out_channels", "extend", "for", "if", "in", "in_channels", "in_projection", "kernel_size", "len", "make_divisible", "nn", "out_channels", "p", "pool_layer", "project", "raise", "rate", "relu", "self", "super", "use_activation", "values", "width_multiplier"], "mobilevitv2/modeling_mobilevitv2.py:MobileViTV2ASPP.forward": ["ModelASPP", "Module", "Tensor", "append", "cat", "class", "conv", "convs", "def", "dim", "dropout", "features", "for", "forward", "in", "nn", "pooled_features", "project", "pyramid", "return", "self", "torch"], "mobilevitv2/modeling_mobilevitv2.py:MobileViTV2DeepLabV3.__init__": ["Dropout2d", "False", "ModelASPP", "ModelConfig", "ModelConvLayer", "ModelDeepLabV3", "Module", "None", "True", "__init__", "aspp", "aspp_out_channels", "bias", "class", "classifier", "classifier_dropout_prob", "config", "def", "dropout", "in_channels", "kernel_size", "nn", "num_labels", "out_channels", "self", "super", "use_activation", "use_normalization"], "mobilevitv2/modeling_mobilevitv2.py:MobileViTV2DeepLabV3.forward": ["ModelDeepLabV3", "Module", "Tensor", "aspp", "class", "classifier", "def", "dropout", "features", "forward", "hidden_states", "nn", "return", "self", "torch"], "mobilevitv2/modeling_mobilevitv2.py:MobileViTV2ForSemanticSegmentation.__init__": ["False", "Model", "ModelConfig", "ModelDeepLabV3", "ModelForSemanticSegmentation", "ModelModel", "ModelPreTrainedModel", "None", "__init__", "class", "config", "def", "expand_output", "num_labels", "post_init", "segmentation_head", "self", "super"], "mobilevitv2/modeling_mobilevitv2.py:MobileViTV2ForSemanticSegmentation.forward": ["CrossEntropyLoss", "False", "Model", "ModelForSemanticSegmentation", "ModelPreTrainedModel", "None", "SemanticSegmenterOutput", "Tensor", "The", "True", "ValueError", "align_corners", "and", "attentions", "be", "bilinear", "bool", "class", "config", "def", "else", "encoder_hidden_states", "forward", "functional", "greater", "hidden_states", "if", "ignore_index", "interpolate", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "mode", "nn", "not", "num_labels", "number", "of", "one", "output", "output_hidden_states", "outputs", "pixel_values", "r", "raise", "return", "return_dict", "segmentation_head", "self", "semantic_loss_ignore_index", "shape", "should", "size", "than", "torch", "tuple", "upsampled_logits", "use_return_dict"], "modernbert/modeling_modernbert.py:ApplyRotaryEmbUnpad.forward": ["False", "Function", "ModelRotaryEmbUnpad", "Model_rotary", "None", "Tensor", "True", "_nheads", "_three", "autograd", "class", "contiguous", "cos", "ctx", "cu_seqlens", "def", "forward", "headdim", "inplace", "int", "interleaved", "max_seqlen", "qk", "qkv", "return", "save_for_backward", "seqlen_offsets", "shape", "sin", "torch", "total_nnz", "view"], "modernbert/modeling_modernbert.py:ApplyRotaryEmbUnpad.backward": ["False", "Function", "ModelRotaryEmbUnpad", "Model_rotary", "None", "True", "_nheads", "_three", "autograd", "backward", "class", "conjugate", "contiguous", "cos", "ctx", "cu_seqlens", "def", "do", "dqk", "headdim", "inplace", "interleaved", "max_seqlen", "return", "saved_tensors", "seqlen_offsets", "shape", "sin", "torch", "total_nnz", "view"], "modernbert/modeling_modernbert.py:apply_rotary_unpadded": ["Model", "ModelRotaryEmbUnpad", "Model_rotary_unpadded", "None", "Tensor", "cos", "cu_seqlens", "def", "int", "max_seqlen", "qkv", "return", "sin", "torch"], "modernbert/modeling_modernbert.py:ModernBertUnpaddedRotaryEmbedding.__init__": ["False", "ModelUnpaddedRotaryEmbedding", "None", "RotaryEmbedding", "__init__", "_update_cos_sin_cache", "and", "base", "class", "def", "device", "dim", "dtype", "float", "if", "int", "interleaved", "is", "max_seqlen", "not", "self", "super", "torch"], "modernbert/modeling_modernbert.py:ModernBertUnpaddedRotaryEmbedding.forward": ["ModelUnpaddedRotaryEmbedding", "None", "RotaryEmbedding", "Tensor", "_cos_cached", "_sin_cached", "_update_cos_sin_cache", "apply_rotary_unpadded", "class", "cu_seqlens", "def", "device", "dtype", "forward", "if", "int", "is", "max_seqlen", "not", "qkv", "return", "self", "torch", "tuple"], "modernbert/modeling_modernbert.py:ModernBertUnpaddedRotaryEmbedding.extra_repr": ["ModelUnpaddedRotaryEmbedding", "RotaryEmbedding", "base", "class", "def", "dim", "extra_repr", "f", "return", "scale_base", "self", "str"], "modernbert/modeling_modernbert.py:ModernBertEmbeddings.__init__": ["Dropout", "Embedding", "LayerNorm", "ModelConfig", "ModelEmbeddings", "Module", "__init__", "bias", "class", "config", "def", "drop", "embedding_dropout", "eps", "hidden_size", "nn", "norm", "norm_bias", "norm_eps", "pad_token_id", "padding_idx", "self", "super", "tok_embeddings", "vocab_size"], "modernbert/modeling_modernbert.py:ModernBertEmbeddings.compiled_embeddings": ["LongTensor", "ModelEmbeddings", "Module", "Tensor", "class", "compiled_embeddings", "def", "drop", "input_ids", "nn", "norm", "return", "self", "tok_embeddings", "torch"], "modernbert/modeling_modernbert.py:ModernBertEmbeddings.forward": ["LongTensor", "ModelEmbeddings", "Module", "None", "Tensor", "class", "compiled_embeddings", "config", "def", "drop", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "nn", "norm", "not", "reference_compile", "return", "self", "tok_embeddings", "torch"], "modernbert/modeling_modernbert.py:ModernBertMLP.__init__": ["ACT2FN", "Dropout", "Linear", "ModelConfig", "ModelMLP", "Module", "Wi", "Wo", "__init__", "act", "bias", "class", "config", "def", "drop", "hidden_activation", "hidden_size", "int", "intermediate_size", "mlp_bias", "mlp_dropout", "nn", "self", "super"], "modernbert/modeling_modernbert.py:ModernBertMLP.forward": ["ModelMLP", "Module", "Tensor", "Wi", "Wo", "act", "chunk", "class", "def", "dim", "drop", "forward", "gate", "hidden_states", "input", "nn", "return", "self", "torch"], "modernbert/modeling_modernbert.py:ModernBertRotaryEmbedding.__init__": ["Callable", "False", "ModelConfig", "ModelRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "_attention_scaling", "_inv_freq", "_original_inv_freq", "class", "clone", "compute_default_rope_parameters", "config", "continue", "curr_attention_scaling", "curr_inv_freq", "def", "default", "device", "f", "for", "if", "in", "is", "layer_type", "layer_types", "list", "max_position_embeddings", "max_seq_len_cached", "nn", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_params", "rope_type", "self", "set", "setattr", "super"], "modernbert/modeling_modernbert.py:ModernBertRotaryEmbedding.compute_default_rope_parameters": ["ModelConfig", "ModelRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "layer_type", "nn", "num_attention_heads", "or", "return", "rope_parameters", "rope_theta", "seq_len", "str", "to", "torch", "tuple"], "modernbert/modeling_modernbert.py:ModernBertRotaryEmbedding.forward": ["False", "ModelRotaryEmbedding", "Module", "None", "_attention_scaling", "_inv_freq", "and", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "f", "float", "forward", "freqs", "getattr", "if", "inv_freq", "inv_freq_expanded", "isinstance", "layer_type", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "modernbert/modeling_modernbert.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "modernbert/modeling_modernbert.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cos", "def", "k", "k_embed", "q", "q_embed", "return", "rotate_half", "sin", "unsqueeze", "unsqueeze_dim"], "modernbert/modeling_modernbert.py:eager_attention_forward": ["False", "LongTensor", "ModelAttention", "Model_attention_forward", "None", "Tensor", "_kwargs", "apply_rotary_pos_emb", "attention_dropout", "attention_mask", "attn_output", "attn_weights", "bool", "bs", "contiguous", "cos", "def", "dim", "dropout", "dtype", "float32", "functional", "head_dim", "if", "int", "key", "local_attention", "matmul", "module", "nn", "output_attentions", "p", "position_embeddings", "position_ids", "qkv", "query", "return", "scale", "sin", "sliding_window_mask", "softmax", "to", "torch", "training", "transpose", "tuple", "unbind", "value", "view"], "modernbert/modeling_modernbert.py:flash_attention_forward": ["ModelAttention", "ModelUnpaddedRotaryEmbedding", "Model_attention_forward", "Model_attn_varlen_qkvpacked_func", "Tensor", "_kwargs", "attention_dropout", "attn", "bfloat16", "bs", "convert_dtype", "cu_seqlens", "def", "deterministic", "deterministic_Model_attn", "dim", "dropout_p", "dtype", "else", "float16", "if", "in", "int", "local_attention", "max_seqlen", "module", "not", "orig_dtype", "qkv", "return", "rotary_emb", "target_dtype", "to", "torch", "training", "tuple", "view", "window_size"], "modernbert/modeling_modernbert.py:sdpa_attention_forward": ["F", "LongTensor", "ModelAttention", "Model_attention_forward", "None", "Tensor", "_kwargs", "apply_rotary_pos_emb", "attention_dropout", "attention_mask", "attn_mask", "attn_output", "bs", "contiguous", "cos", "def", "dim", "dropout_p", "else", "if", "int", "key", "local_attention", "module", "position_embeddings", "position_ids", "qkv", "query", "return", "scaled_dot_product_attention", "sin", "sliding_window_mask", "torch", "training", "transpose", "tuple", "unbind", "value", "view"], "modernbert/modeling_modernbert.py:ModernBertAttention.__init__": ["Dropout", "Identity", "Linear", "ModelAttention", "ModelConfig", "ModelUnpaddedRotaryEmbedding", "Module", "None", "The", "ValueError", "Wo", "Wqkv", "__init__", "_attn_implementation", "a", "all_head_size", "attention", "attention_bias", "attention_dropout", "base", "bias", "class", "config", "def", "deterministic_flash_attn", "dim", "else", "f", "flash_attention_2", "global_attn_every_n_layers", "head_dim", "heads", "hidden", "hidden_size", "if", "int", "is", "layer_id", "layer_type", "layer_types", "local_attention", "max_position_embeddings", "max_seqlen", "multiple", "nn", "not", "num_attention_heads", "num_heads", "number", "of", "out_drop", "raise", "rope_parameters", "rope_parameters_dict", "rope_theta", "rotary_emb", "self", "size", "super", "the"], "modernbert/modeling_modernbert.py:ModernBertAttention.forward": ["False", "ModelAttention", "Model_ATTENTION_FUNCTION", "Module", "None", "Tensor", "Wo", "Wqkv", "_attn_implementation", "all_head_size", "attn_outputs", "bool", "bs", "class", "config", "def", "dim", "else", "flash_attention_2", "forward", "head_dim", "hidden_states", "if", "kwargs", "local_attention", "nn", "num_heads", "out_drop", "output_attentions", "position_embeddings", "qkv", "return", "rotary_emb", "self", "shape", "torch", "view"], "modernbert/modeling_modernbert.py:ModernBertEncoderLayer.__init__": ["GradientCheckpointingLayer", "Identity", "LayerNorm", "ModelAttention", "ModelConfig", "ModelEncoderLayer", "ModelMLP", "None", "__init__", "attention_type", "attn", "attn_norm", "bias", "class", "config", "def", "else", "eps", "hidden_size", "if", "int", "layer_id", "layer_types", "mlp", "mlp_norm", "nn", "norm_bias", "norm_eps", "self", "super"], "modernbert/modeling_modernbert.py:ModernBertEncoderLayer.compiled_mlp": ["GradientCheckpointingLayer", "ModelEncoderLayer", "Tensor", "class", "compiled_mlp", "def", "hidden_states", "mlp", "mlp_norm", "return", "self", "torch"], "modernbert/modeling_modernbert.py:ModernBertEncoderLayer.forward": ["False", "GradientCheckpointingLayer", "LongTensor", "ModelEncoderLayer", "None", "Tensor", "attention_mask", "attn", "attn_norm", "attn_outputs", "bool", "class", "compiled_mlp", "config", "cu_seqlens", "def", "else", "forward", "hidden_states", "if", "int", "max_seqlen", "mlp", "mlp_norm", "mlp_output", "output_attentions", "position_embeddings", "position_ids", "reference_compile", "return", "self", "sliding_window_mask", "torch"], "modernbert/modeling_modernbert.py:ModernBertPreTrainedModel._init_weights": ["LayerNorm", "Linear", "ModelAttention", "ModelEmbeddings", "ModelForMaskedLM", "ModelForMultipleChoice", "ModelForQuestionAnswering", "ModelForSequenceClassification", "ModelForTokenClassification", "ModelMLP", "ModelPreTrainedModel", "ModelPredictionHead", "ModelRotaryEmbedding", "ModelUnpaddedRotaryEmbedding", "Module", "None", "PreTrainedModel", "ROPE_INIT_FUNCTIONS", "Wi", "Wo", "Wqkv", "_", "_compute_inv_freq", "_init_weights", "_inv_freq", "_original_inv_freq", "a", "b", "bias", "class", "classifier", "compute_default_rope_parameters", "config", "copy_", "curr_inv_freq", "cutoff_factor", "decoder", "def", "default", "dense", "elif", "embedding", "f", "final_out", "float", "for", "getattr", "hidden_size", "if", "in", "init", "init_weight", "initializer_cutoff_factor", "initializer_range", "inv_freq", "is", "isinstance", "layer_type", "layer_types", "math", "mean", "module", "nn", "not", "num_hidden_layers", "ones_", "out", "rope_init_fn", "rope_type", "self", "sqrt", "std", "stds", "tok_embeddings", "trunc_normal_", "weight", "zeros_"], "modernbert/modeling_modernbert.py:ModernBertPreTrainedModel._check_and_adjust_attn_implementation": ["False", "ImportError", "ModelPreTrainedModel", "None", "PreTrainedModel", "ValueError", "_check_and_adjust_attn_implementation", "_flash_attn_2_can_dispatch", "and", "attn_implementation", "bool", "class", "def", "else", "except", "flash_attention_2", "if", "is", "is_init_check", "pass", "return", "self", "str", "super", "try"], "modernbert/modeling_modernbert.py:ModernBertPreTrainedModel._maybe_set_compile": ["Compiling", "Falling", "False", "If", "ModelPreTrainedModel", "None", "PreTrainedModel", "_maybe_set_compile", "a", "accelerate", "across", "and", "back", "class", "compile", "compiled", "config", "cpu", "def", "device", "devices", "hasattr", "hf_device_map", "if", "is", "is_triton_available", "len", "logger", "mode", "model", "mps", "non", "not", "reference_compile", "return", "self", "split", "supported", "the", "to", "torch", "type", "using", "warning_once", "will", "with", "work"], "modernbert/modeling_modernbert.py:ModernBertPreTrainedModel.resize_token_embeddings": ["Falling", "False", "ModelPreTrainedModel", "None", "PreTrainedModel", "Resizing", "True", "args", "back", "class", "compile", "compiled", "config", "def", "embeddings", "if", "in", "is", "kwargs", "logger", "mode", "model_embeds", "non", "not", "reference_compile", "resize_token_embeddings", "return", "self", "super", "supported", "to", "token", "torch", "warning_once", "with"], "modernbert/modeling_modernbert.py:_unpad_modernbert_input": ["False", "None", "Tensor", "_unpad_Model_input", "as_tuple", "attention_mask", "batch", "cu_seqlens", "cumsum", "def", "dim", "dtype", "else", "flatten", "functional", "if", "indices", "inputs", "int", "int32", "is", "item", "labels", "max", "max_seqlen_in_batch", "nn", "nonzero", "not", "pad", "position_ids", "rest", "return", "seqlen", "seqlens_in_batch", "shape", "sum", "torch", "tuple", "unpadded_inputs", "unpadded_labels", "unpadded_position_ids", "view"], "modernbert/modeling_modernbert.py:_pad_modernbert_output": ["Tensor", "_", "_pad_Model_output", "batch", "def", "device", "dim", "dtype", "else", "if", "indices", "inputs", "int", "output", "padded_inputs", "rest", "return", "seqlen", "shape", "torch", "view", "zeros"], "modernbert/modeling_modernbert.py:ModernBertModel.__init__": ["False", "LayerNorm", "ModelConfig", "ModelEmbeddings", "ModelEncoderLayer", "ModelModel", "ModelPreTrainedModel", "ModelRotaryEmbedding", "ModuleList", "__init__", "bias", "class", "config", "def", "embeddings", "eps", "final_norm", "for", "gradient_checkpointing", "hidden_size", "in", "layer_id", "layers", "nn", "norm_bias", "norm_eps", "num_hidden_layers", "post_init", "range", "rotary_emb", "self", "super"], "modernbert/modeling_modernbert.py:ModernBertModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "return", "self", "tok_embeddings"], "modernbert/modeling_modernbert.py:ModernBertModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "self", "set_input_embeddings", "tok_embeddings", "value"], "modernbert/modeling_modernbert.py:ModernBertModel.forward": ["BaseModelOutput", "False", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "True", "ValueError", "You", "_", "_attn_implementation", "_maybe_set_compile", "_pad_Model_output", "_unpad_Model_input", "_update_attention_mask", "all_hidden_states", "all_self_attentions", "and", "arange", "attention_mask", "attention_type", "attentions", "batch", "batch_size", "bool", "class", "config", "cu_seqlens", "def", "device", "dim", "dtype", "elif", "else", "embeddings", "encoder_layer", "exactly", "final_norm", "flash_attention_2", "for", "forward", "hidden_states", "hs", "if", "in", "indices", "input_ids", "inputs", "inputs_embeds", "int", "is", "kwargs", "last_hidden_state", "layer_outputs", "layer_type", "layer_types", "layers", "len", "max_seqlen", "must", "no_grad", "not", "of", "one", "ones", "or", "output_attentions", "output_hidden_states", "position_embeddings", "position_ids", "r", "raise", "repad", "return", "return_dict", "rotary_emb", "self", "seq_len", "seqlen", "shape", "sliding_window_mask", "specify", "torch", "tuple", "unsqueeze", "use_return_dict", "v", "warn_if_padding_and_no_attention_mask", "with"], "modernbert/modeling_modernbert.py:ModernBertModel._update_attention_mask": ["Consider", "Falling", "False", "ModelModel", "ModelPreTrainedModel", "Outputting", "Setting", "T", "Tensor", "_attn_implementation", "_prepare_4d_attention_mask", "_update_attention_mask", "abs", "arange", "attention", "attention_mask", "attentions", "attn_implementation", "back", "bool", "class", "config", "def", "device", "distance", "dtype", "eager", "elif", "f", "finfo", "global_attention_mask", "if", "implementation", "is", "local_attention", "logger", "logical_not", "masked_fill", "min", "not", "only", "output_attentions", "return", "rows", "sdpa", "self", "setting", "shape", "sliding_window_mask", "supported", "the", "to", "torch", "unsqueeze", "warning_once", "window_mask", "with"], "modernbert/modeling_modernbert.py:ModernBertPredictionHead.__init__": ["ACT2FN", "LayerNorm", "Linear", "ModelConfig", "ModelPredictionHead", "Module", "__init__", "act", "bias", "class", "classifier_activation", "classifier_bias", "config", "def", "dense", "eps", "hidden_size", "nn", "norm", "norm_bias", "norm_eps", "self", "super"], "modernbert/modeling_modernbert.py:ModernBertPredictionHead.forward": ["ModelPredictionHead", "Module", "Tensor", "act", "class", "def", "dense", "forward", "hidden_states", "nn", "norm", "return", "self", "torch"], "modernbert/modeling_modernbert.py:ModernBertForMaskedLM.__init__": ["Linear", "ModelConfig", "ModelForMaskedLM", "ModelModel", "ModelPreTrainedModel", "ModelPredictionHead", "__init__", "bias", "class", "config", "decoder", "decoder_bias", "def", "head", "hidden_size", "model", "nn", "post_init", "self", "sparse_pred_ignore_index", "sparse_prediction", "super", "vocab_size"], "modernbert/modeling_modernbert.py:ModernBertForMaskedLM.get_output_embeddings": ["ModelForMaskedLM", "ModelPreTrainedModel", "class", "decoder", "def", "get_output_embeddings", "return", "self"], "modernbert/modeling_modernbert.py:ModernBertForMaskedLM.set_output_embeddings": ["Linear", "ModelForMaskedLM", "ModelPreTrainedModel", "class", "decoder", "def", "new_embeddings", "nn", "self", "set_output_embeddings"], "modernbert/modeling_modernbert.py:ModernBertForMaskedLM.compiled_head": ["ModelForMaskedLM", "ModelPreTrainedModel", "Tensor", "class", "compiled_head", "decoder", "def", "head", "output", "return", "self", "torch"], "modernbert/modeling_modernbert.py:ModernBertForMaskedLM.forward": ["LongTensor", "MaskedLMOutput", "ModelForMaskedLM", "ModelPreTrainedModel", "None", "Tensor", "_attn_implementation", "_maybe_set_compile", "_pad_Model_output", "_unpad_Model_input", "and", "append", "attention_mask", "attentions", "batch", "batch_size", "bool", "class", "compiled_head", "config", "cu_seqlens", "decoder", "def", "device", "dim", "dtype", "else", "flash_attention_2", "for", "forward", "getattr", "head", "hidden_states", "hs", "if", "in", "indices", "input_ids", "inputs", "inputs_embeds", "int", "is", "kwargs", "labels", "last_hidden_state", "logits", "loss", "loss_function", "mask_tokens", "max_seqlen", "model", "no_grad", "not", "nullcontext", "ones", "or", "output", "output_attentions", "output_hidden_states", "outputs", "padded_hidden_states", "position_ids", "r", "reference_compile", "repad_logits_with_grad", "return", "return_dict", "self", "seq_len", "seqlen", "shape", "sliding_window_mask", "sparse_pred_ignore_index", "sparse_prediction", "squeeze", "torch", "tuple", "use_return_dict", "view", "vocab_size", "with"], "modernbert/modeling_modernbert.py:ModernBertForSequenceClassification.__init__": ["Dropout", "Linear", "ModelConfig", "ModelForSequenceClassification", "ModelModel", "ModelPreTrainedModel", "ModelPredictionHead", "__init__", "class", "classifier", "classifier_dropout", "config", "def", "drop", "head", "hidden_size", "model", "nn", "num_labels", "post_init", "self", "super", "torch"], "modernbert/modeling_modernbert.py:ModernBertForSequenceClassification.forward": ["BCEWithLogitsLoss", "CrossEntropyLoss", "LongTensor", "MSELoss", "ModelForSequenceClassification", "ModelPreTrainedModel", "None", "SequenceClassifierOutput", "Tensor", "True", "_maybe_set_compile", "and", "attention_mask", "attentions", "batch_size", "bool", "class", "classifier", "classifier_pooling", "cls", "config", "cu_seqlens", "def", "device", "dim", "drop", "dtype", "elif", "else", "forward", "head", "hidden_states", "if", "indices", "input_ids", "inputs_embeds", "int", "is", "keepdim", "kwargs", "labels", "last_hidden_state", "logits", "long", "loss", "loss_fct", "max_seqlen", "mean", "model", "multi_label_classification", "not", "num_labels", "ones", "or", "output", "output_attentions", "output_hidden_states", "outputs", "pooled_output", "position_ids", "problem_type", "r", "regression", "return", "return_dict", "self", "seq_len", "shape", "single_label_classification", "sliding_window_mask", "squeeze", "sum", "torch", "tuple", "unsqueeze", "use_return_dict", "view", "warn_if_padding_and_no_attention_mask"], "modernbert/modeling_modernbert.py:ModernBertForTokenClassification.__init__": ["Dropout", "Linear", "ModelConfig", "ModelForTokenClassification", "ModelModel", "ModelPreTrainedModel", "ModelPredictionHead", "__init__", "class", "classifier", "classifier_dropout", "config", "def", "drop", "head", "hidden_size", "model", "nn", "num_labels", "post_init", "self", "super", "torch"], "modernbert/modeling_modernbert.py:ModernBertForTokenClassification.forward": ["CrossEntropyLoss", "LongTensor", "ModelForTokenClassification", "ModelPreTrainedModel", "None", "Tensor", "TokenClassifierOutput", "_maybe_set_compile", "attention_mask", "attentions", "batch_size", "bool", "class", "classifier", "config", "cu_seqlens", "def", "drop", "else", "forward", "head", "hidden_states", "if", "indices", "input_ids", "inputs_embeds", "int", "is", "kwargs", "labels", "last_hidden_state", "logits", "loss", "loss_fct", "max_seqlen", "model", "not", "num_labels", "output", "output_attentions", "output_hidden_states", "outputs", "position_ids", "r", "return", "return_dict", "self", "seq_len", "sliding_window_mask", "torch", "tuple", "use_return_dict", "view"], "modernbert/modeling_modernbert.py:ModernBertForQuestionAnswering.__init__": ["Dropout", "Linear", "ModelConfig", "ModelForQuestionAnswering", "ModelModel", "ModelPreTrainedModel", "ModelPredictionHead", "__init__", "class", "classifier", "classifier_dropout", "config", "def", "drop", "head", "hidden_size", "model", "nn", "num_labels", "post_init", "self", "super", "torch"], "modernbert/modeling_modernbert.py:ModernBertForQuestionAnswering.forward": ["ModelForQuestionAnswering", "ModelPreTrainedModel", "None", "QuestionAnsweringModelOutput", "Tensor", "_maybe_set_compile", "and", "attention_mask", "attentions", "batch_size", "bool", "class", "classifier", "config", "contiguous", "cu_seqlens", "def", "dim", "drop", "else", "end_logits", "end_positions", "forward", "head", "hidden_states", "if", "indices", "input_ids", "int", "is", "kwargs", "last_hidden_state", "logits", "loss", "loss_function", "max_seqlen", "model", "not", "output", "output_attentions", "output_hidden_states", "outputs", "position_ids", "r", "return", "return_dict", "self", "seq_len", "sliding_window_mask", "split", "squeeze", "start_logits", "start_positions", "torch", "tuple", "use_return_dict"], "modernbert/modeling_modernbert.py:ModernBertForMultipleChoice.__init__": ["Dropout", "Linear", "ModelConfig", "ModelForMultipleChoice", "ModelModel", "ModelPreTrainedModel", "ModelPredictionHead", "__init__", "class", "classifier", "classifier_dropout", "config", "def", "drop", "head", "hidden_size", "model", "nn", "post_init", "self", "super", "torch"], "modernbert/modeling_modernbert.py:ModernBertForMultipleChoice.forward": ["CrossEntropyLoss", "LongTensor", "ModelForMultipleChoice", "ModelPreTrainedModel", "MultipleChoiceModelOutput", "None", "Tensor", "True", "_maybe_set_compile", "arange", "argmax", "attention_mask", "attentions", "batch_size", "bool", "class", "classifier", "classifier_pooling", "cls", "cls_mask", "config", "cu_seqlens", "def", "device", "dim", "drop", "dtype", "elif", "else", "forward", "head", "hidden_states", "if", "indices", "indices_0", "input_ids", "inputs_embeds", "int", "is", "keepdim", "kwargs", "labels", "last_hidden_state", "logits", "long", "loss", "loss_fct", "max_seqlen", "mean", "model", "nn", "not", "num_choices", "num_non_pad_tokens", "output", "output_attentions", "output_hidden_states", "outputs", "pooled_output", "position_ids", "r", "reshaped_logits", "return", "return_dict", "self", "seq_len", "shape", "size", "sliding_window_mask", "sum", "tensor", "to", "torch", "tuple", "unsqueeze", "use_return_dict", "view"], "modernbert_decoder/modeling_modernbert_decoder.py:ModernBertDecoderEmbeddings.__init__": ["Dropout", "Embedding", "LayerNorm", "ModelConfig", "ModelEmbeddings", "Module", "__init__", "bias", "class", "config", "def", "drop", "embedding_dropout", "eps", "hidden_size", "nn", "norm", "norm_bias", "norm_eps", "pad_token_id", "padding_idx", "self", "super", "tok_embeddings", "vocab_size"], "modernbert_decoder/modeling_modernbert_decoder.py:ModernBertDecoderEmbeddings.compiled_embeddings": ["LongTensor", "ModelEmbeddings", "Module", "Tensor", "class", "compiled_embeddings", "def", "drop", "input_ids", "nn", "norm", "return", "self", "tok_embeddings", "torch"], "modernbert_decoder/modeling_modernbert_decoder.py:ModernBertDecoderEmbeddings.forward": ["LongTensor", "ModelEmbeddings", "Module", "None", "Tensor", "class", "compiled_embeddings", "config", "def", "drop", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "nn", "norm", "not", "reference_compile", "return", "self", "tok_embeddings", "torch"], "modernbert_decoder/modeling_modernbert_decoder.py:ModernBertDecoderMLP.__init__": ["ACT2FN", "Dropout", "Linear", "ModelConfig", "ModelMLP", "Module", "Wi", "Wo", "__init__", "act", "bias", "class", "config", "def", "drop", "hidden_activation", "hidden_size", "int", "intermediate_size", "mlp_bias", "mlp_dropout", "nn", "self", "super"], "modernbert_decoder/modeling_modernbert_decoder.py:ModernBertDecoderMLP.forward": ["ModelMLP", "Module", "Tensor", "Wi", "Wo", "act", "chunk", "class", "def", "dim", "drop", "forward", "gate", "hidden_states", "input", "nn", "return", "self", "torch"], "modernbert_decoder/modeling_modernbert_decoder.py:ModernBertDecoderRotaryEmbedding.__init__": ["Callable", "False", "ModelConfig", "ModelRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "_attention_scaling", "_inv_freq", "_original_inv_freq", "class", "clone", "compute_default_rope_parameters", "config", "continue", "curr_attention_scaling", "curr_inv_freq", "def", "default", "device", "f", "for", "if", "in", "is", "layer_type", "layer_types", "list", "max_position_embeddings", "max_seq_len_cached", "nn", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_params", "rope_type", "self", "set", "setattr", "super"], "modernbert_decoder/modeling_modernbert_decoder.py:ModernBertDecoderRotaryEmbedding.compute_default_rope_parameters": ["ModelConfig", "ModelRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "layer_type", "nn", "num_attention_heads", "or", "return", "rope_parameters", "rope_theta", "seq_len", "str", "to", "torch", "tuple"], "modernbert_decoder/modeling_modernbert_decoder.py:ModernBertDecoderRotaryEmbedding.forward": ["False", "ModelRotaryEmbedding", "Module", "None", "_attention_scaling", "_inv_freq", "and", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "f", "float", "forward", "freqs", "getattr", "if", "inv_freq", "inv_freq_expanded", "isinstance", "layer_type", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "modernbert_decoder/modeling_modernbert_decoder.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "modernbert_decoder/modeling_modernbert_decoder.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cos", "def", "k", "k_embed", "q", "q_embed", "return", "rotate_half", "sin", "unsqueeze", "unsqueeze_dim"], "modernbert_decoder/modeling_modernbert_decoder.py:eager_attention_forward": ["ModelAttention", "Model_attention_forward", "None", "Tensor", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "head_dim", "if", "int", "is", "key", "kwargs", "matmul", "module", "nn", "p", "query", "return", "scaling", "shape", "sliding_window", "softmax", "to", "torch", "training", "transpose", "tuple", "value"], "modernbert_decoder/modeling_modernbert_decoder.py:ModernBertDecoderAttention.__init__": ["Dropout", "Linear", "ModelAttention", "ModelConfig", "Module", "None", "The", "True", "ValueError", "Wo", "__init__", "a", "all_head_size", "attention", "attention_bias", "attention_dropout", "bias", "class", "config", "def", "else", "f", "head_dim", "heads", "hidden", "hidden_size", "if", "int", "is", "is_causal", "is_sliding", "k_proj", "layer_idx", "layer_types", "multiple", "nn", "not", "num_attention_heads", "num_heads", "number", "of", "out_drop", "q_proj", "raise", "scaling", "self", "size", "sliding_attention", "sliding_window", "super", "the", "v_proj"], "modernbert_decoder/modeling_modernbert_decoder.py:ModernBertDecoderAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "LongTensor", "ModelAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "Wo", "_attn_implementation", "apply_rotary_pos_emb", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "cache_kwargs", "cache_position", "class", "config", "contiguous", "cos", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "head_dim", "hidden_shape", "hidden_states", "if", "input_shape", "is", "k_proj", "key_states", "kwargs", "layer_idx", "nn", "not", "out_drop", "past_key_values", "position_embeddings", "q_proj", "query_states", "reshape", "return", "scaling", "self", "shape", "sin", "sliding_window", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "view"], "modernbert_decoder/modeling_modernbert_decoder.py:ModernBertDecoderLayer.__init__": ["GradientCheckpointingLayer", "Identity", "LayerNorm", "ModelAttention", "ModelConfig", "ModelLayer", "ModelMLP", "None", "__init__", "attention_type", "attn", "attn_norm", "bias", "class", "config", "def", "else", "eps", "hidden_size", "if", "int", "layer_idx", "layer_types", "mlp", "mlp_norm", "nn", "norm_bias", "norm_eps", "self", "super"], "modernbert_decoder/modeling_modernbert_decoder.py:ModernBertDecoderLayer.forward": ["Cache", "FloatTensor", "GradientCheckpointingLayer", "LongTensor", "ModelLayer", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn", "attn_norm", "attn_outputs", "cache_position", "class", "def", "forward", "hidden_states", "kwargs", "mlp", "mlp_norm", "mlp_output", "past_key_values", "position_embeddings", "residual", "return", "self", "torch", "tuple"], "modernbert_decoder/modeling_modernbert_decoder.py:ModernBertDecoderPredictionHead.__init__": ["ACT2FN", "LayerNorm", "Linear", "ModelConfig", "ModelPredictionHead", "Module", "__init__", "act", "bias", "class", "classifier_activation", "classifier_bias", "config", "def", "dense", "eps", "hidden_size", "nn", "norm", "norm_bias", "norm_eps", "self", "super"], "modernbert_decoder/modeling_modernbert_decoder.py:ModernBertDecoderPredictionHead.forward": ["ModelPredictionHead", "Module", "Tensor", "act", "class", "def", "dense", "forward", "hidden_states", "nn", "norm", "return", "self", "torch"], "modernbert_decoder/modeling_modernbert_decoder.py:ModernBertDecoderPreTrainedModel._init_weights": ["LayerNorm", "Linear", "ModelAttention", "ModelEmbeddings", "ModelForCausalLM", "ModelForSequenceClassification", "ModelMLP", "ModelPreTrainedModel", "ModelPredictionHead", "ModelRotaryEmbedding", "Module", "None", "PreTrainedModel", "ROPE_INIT_FUNCTIONS", "Wi", "Wo", "_", "_init_weights", "_inv_freq", "_original_inv_freq", "a", "b", "bias", "class", "classifier", "compute_default_rope_parameters", "config", "copy_", "curr_inv_freq", "cutoff_factor", "decoder", "def", "default", "dense", "elif", "embedding", "f", "final_out", "float", "for", "getattr", "hidden_size", "if", "in", "init", "init_weight", "initializer_cutoff_factor", "initializer_range", "is", "isinstance", "k_proj", "layer_type", "layer_types", "math", "mean", "module", "nn", "not", "num_hidden_layers", "ones_", "out", "q_proj", "rope_init_fn", "rope_type", "self", "sqrt", "std", "stds", "tok_embeddings", "trunc_normal_", "v_proj", "weight", "zeros_"], "modernbert_decoder/modeling_modernbert_decoder.py:ModernBertDecoderModel.__init__": ["False", "LayerNorm", "ModelConfig", "ModelEmbeddings", "ModelLayer", "ModelModel", "ModelPreTrainedModel", "ModelRotaryEmbedding", "ModuleList", "__init__", "bias", "class", "config", "def", "embeddings", "eps", "final_norm", "for", "gradient_checkpointing", "hidden_size", "in", "layer_idx", "layers", "nn", "norm_bias", "norm_eps", "num_hidden_layers", "post_init", "range", "rotary_emb", "self", "super"], "modernbert_decoder/modeling_modernbert_decoder.py:ModernBertDecoderModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "return", "self", "tok_embeddings"], "modernbert_decoder/modeling_modernbert_decoder.py:ModernBertDecoderModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "self", "set_input_embeddings", "tok_embeddings", "value"], "modernbert_decoder/modeling_modernbert_decoder.py:ModernBertDecoderModel.forward": ["BaseModelOutputWithPast", "Cache", "DynamicCache", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "and", "arange", "attention_mask", "attention_type", "batch_size", "bool", "cache_position", "causal_mask_mapping", "class", "config", "create_causal_mask", "create_sliding_window_causal_mask", "decoder_layer", "def", "device", "dict", "else", "embeddings", "exactly", "expand", "final_norm", "for", "forward", "full_attention", "get_seq_length", "hidden_states", "if", "in", "input_embeds", "input_ids", "inputs_embeds", "is", "isinstance", "kwargs", "last_hidden_state", "layer_type", "layer_types", "layers", "mask_kwargs", "must", "not", "of", "one", "or", "past_key_values", "past_seen_tokens", "position_embeddings", "position_ids", "raise", "return", "rotary_emb", "self", "seq_length", "shape", "sliding_attention", "specify", "torch", "tuple", "unsqueeze", "use_cache", "warn_if_padding_and_no_attention_mask"], "modernbert_decoder/modeling_modernbert_decoder.py:ModernBertDecoderForCausalLM.__init__": ["GenerationMixin", "Linear", "ModelConfig", "ModelForCausalLM", "ModelModel", "ModelPreTrainedModel", "ModelPredictionHead", "__init__", "bias", "class", "config", "decoder", "decoder_bias", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "vocab_size"], "modernbert_decoder/modeling_modernbert_decoder.py:ModernBertDecoderForCausalLM.get_output_embeddings": ["GenerationMixin", "ModelForCausalLM", "ModelPreTrainedModel", "class", "decoder", "def", "get_output_embeddings", "return", "self"], "modernbert_decoder/modeling_modernbert_decoder.py:ModernBertDecoderForCausalLM.set_output_embeddings": ["GenerationMixin", "ModelForCausalLM", "ModelPreTrainedModel", "class", "decoder", "def", "new_embeddings", "self", "set_output_embeddings"], "modernbert_decoder/modeling_modernbert_decoder.py:ModernBertDecoderForCausalLM.forward": ["BaseModelOutputWithPast", "Cache", "CausalLMOutputWithPast", "CrossEntropyLoss", "GenerationMixin", "LongTensor", "ModelForCausalLM", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "bool", "class", "config", "contiguous", "decoder", "def", "device", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "logits", "logits_to_keep", "loss", "loss_fct", "model", "not", "outputs", "past_key_values", "position_ids", "r", "return", "self", "shift_labels", "shift_logits", "slice", "slice_indices", "to", "torch", "tuple", "use_cache", "view", "vocab_size"], "modernbert_decoder/modeling_modernbert_decoder.py:ModernBertDecoderForSequenceClassification.__init__": ["Dropout", "Linear", "ModelConfig", "ModelForSequenceClassification", "ModelModel", "ModelPreTrainedModel", "ModelPredictionHead", "__init__", "bias", "class", "classifier", "classifier_bias", "classifier_dropout", "config", "def", "drop", "head", "hidden_size", "model", "nn", "num_labels", "post_init", "self", "super", "torch"], "modernbert_decoder/modeling_modernbert_decoder.py:ModernBertDecoderForSequenceClassification.forward": ["BCEWithLogitsLoss", "Cache", "Cannot", "CrossEntropyLoss", "LongTensor", "MSELoss", "ModelForSequenceClassification", "ModelPreTrainedModel", "None", "Results", "SequenceClassifierOutputWithPast", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "__class__", "__name__", "and", "arange", "argmax", "attention_mask", "attentions", "batch", "batch_size", "be", "bool", "class", "classifier", "config", "conjunction", "def", "defined", "detect", "device", "drop", "dtype", "elif", "else", "f", "forward", "handle", "head", "hidden_states", "if", "in", "input_ids", "inputs_embeds", "int", "int32", "is", "kwargs", "labels", "last_non_pad_token", "logger", "logits", "long", "loss", "loss_fct", "may", "model", "multi_label_classification", "no", "non_pad_mask", "not", "num_labels", "or", "pad_token_id", "padding", "past_key_values", "pooled_logits", "position_ids", "problem_type", "r", "raise", "regression", "return", "self", "sequence_length", "shape", "single_label_classification", "sizes", "squeeze", "to", "token", "token_indices", "tokens", "torch", "transformer_outputs", "tuple", "unexpected", "use_cache", "using", "view", "warning_once", "will", "with"], "moonshine/modeling_moonshine.py:MoonshineEncoderMLP.__init__": ["ACT2FN", "Linear", "ModelEncoderMLP", "Module", "__init__", "activation_fn", "class", "config", "def", "fc1", "fc2", "hidden_act", "hidden_size", "intermediate_size", "nn", "self", "super"], "moonshine/modeling_moonshine.py:MoonshineEncoderMLP.forward": ["ModelEncoderMLP", "Module", "Tensor", "activation_fn", "class", "def", "fc1", "fc2", "forward", "hidden_states", "nn", "return", "self", "torch"], "moonshine/modeling_moonshine.py:MoonshineDecoderMLP.__init__": ["ACT2FN", "Linear", "ModelDecoderMLP", "Module", "__init__", "activation_fn", "class", "config", "def", "fc1", "fc2", "hidden_act", "hidden_size", "intermediate_size", "nn", "self", "super"], "moonshine/modeling_moonshine.py:MoonshineDecoderMLP.forward": ["ModelDecoderMLP", "Module", "Tensor", "activation_fn", "chunk", "class", "def", "dim", "fc1", "fc2", "forward", "gate", "hidden_states", "nn", "return", "self", "torch"], "moonshine/modeling_moonshine.py:MoonshineRotaryEmbedding.__init__": ["Callable", "False", "ModelConfig", "ModelRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "moonshine/modeling_moonshine.py:MoonshineRotaryEmbedding.compute_default_rope_parameters": ["ModelConfig", "ModelRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "get", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "partial_rotary_factor", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "moonshine/modeling_moonshine.py:MoonshineRotaryEmbedding.forward": ["False", "ModelRotaryEmbedding", "Module", "None", "and", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "moonshine/modeling_moonshine.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "moonshine/modeling_moonshine.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value", "value_states"], "moonshine/modeling_moonshine.py:rotate_half": ["Model_half", "def", "dim", "flatten", "return", "stack", "torch", "x", "x1", "x2"], "moonshine/modeling_moonshine.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cat", "cos", "def", "dim", "k", "k_embed", "k_pass", "k_rot", "q", "q_embed", "q_pass", "q_rot", "repeat_interleave", "return", "rotary_dim", "rotate_half", "shape", "sin", "torch", "unsqueeze", "unsqueeze_dim"], "moonshine/modeling_moonshine.py:MoonshineAttention.__init__": ["False", "Linear", "ModelAttention", "ModelConfig", "Module", "None", "__init__", "attention_bias", "attention_dropout", "bias", "bool", "class", "config", "def", "else", "getattr", "head_dim", "head_dim_padding", "hidden_size", "if", "int", "is", "is_causal", "k_proj", "layer_idx", "nn", "not", "num_attention_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "pad_head_dim_to_multiple_of", "q_proj", "scaling", "self", "super", "target_head_dim", "target_multiple", "update", "v_proj"], "moonshine/modeling_moonshine.py:MoonshineAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "FlashAttentionKwargs", "LongTensor", "ModelAttention", "Module", "None", "Tensor", "True", "Unpack", "_attn_implementation", "and", "apply_rotary_pos_emb", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "bsz", "cache_kwargs", "cache_position", "class", "config", "contiguous", "cos", "cross_attention_cache", "current_states", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "functional", "get", "head_dim", "head_dim_padding", "hidden_states", "if", "is", "is_causal", "is_cross_attention", "is_updated", "k_proj", "key_states", "key_value_states", "keys", "kwargs", "layer_idx", "layers", "nn", "not", "num_key_value_heads", "o_proj", "pad", "past_key_values", "position_embeddings", "q_len", "q_proj", "query_states", "reshape", "return", "scaling", "self", "self_attention_cache", "shape", "sin", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "values", "view"], "moonshine/modeling_moonshine.py:MoonshineEncoderLayer.__init__": ["False", "GradientCheckpointingLayer", "LayerNorm", "ModelAttention", "ModelConfig", "ModelEncoderLayer", "ModelEncoderMLP", "__init__", "bias", "class", "config", "def", "encoder_hidden_act", "encoder_num_attention_heads", "encoder_num_key_value_heads", "hidden_size", "input_layernorm", "int", "is_causal", "layer_idx", "mlp", "nn", "num_attention_heads", "num_key_value_heads", "post_attention_layernorm", "self", "self_attn", "super"], "moonshine/modeling_moonshine.py:MoonshineEncoderLayer.forward": ["Cache", "False", "GradientCheckpointingLayer", "LongTensor", "ModelEncoderLayer", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "attention_mask", "bool", "cache_position", "class", "def", "forward", "hidden_states", "input_layernorm", "kwargs", "mlp", "past_key_values", "position_embeddings", "position_ids", "post_attention_layernorm", "residual", "return", "self", "self_attn", "torch", "tuple", "use_cache"], "moonshine/modeling_moonshine.py:MoonshineDecoderLayer.__init__": ["False", "GradientCheckpointingLayer", "LayerNorm", "ModelAttention", "ModelConfig", "ModelDecoderLayer", "ModelDecoderMLP", "None", "True", "__init__", "bias", "class", "config", "decoder_hidden_act", "decoder_num_attention_heads", "decoder_num_key_value_heads", "def", "encoder_attn", "final_layernorm", "hidden_size", "input_layernorm", "int", "is_causal", "layer_idx", "mlp", "nn", "num_attention_heads", "num_key_value_heads", "post_attention_layernorm", "self", "self_attn", "super"], "moonshine/modeling_moonshine.py:MoonshineDecoderLayer.forward": ["Cache", "False", "FloatTensor", "GradientCheckpointingLayer", "LongTensor", "ModelDecoderLayer", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "attention_mask", "bool", "cache_position", "class", "def", "encoder_attention_mask", "encoder_attn", "encoder_hidden_states", "encoder_position_embeddings", "encoder_position_ids", "final_layernorm", "forward", "hidden_states", "if", "input_layernorm", "is", "key_value_states", "kwargs", "mlp", "not", "past_key_values", "position_embeddings", "position_ids", "post_attention_layernorm", "residual", "return", "self", "self_attn", "torch", "tuple", "use_cache"], "moonshine/modeling_moonshine.py:MoonshinePreTrainedModel._get_feat_extract_output_lengths": ["LongTensor", "ModelPreTrainedModel", "PreTrainedModel", "_get_feat_extract_output_lengths", "class", "def", "input_lengths", "int", "output_conv1_length", "output_conv2_length", "output_conv3_length", "return", "self", "torch"], "moonshine/modeling_moonshine.py:MoonshineEncoder.__init__": ["Conv1d", "False", "GroupNorm", "LayerNorm", "ModelConfig", "ModelEncoder", "ModelEncoderLayer", "ModelPreTrainedModel", "ModelRotaryEmbedding", "ModuleList", "__init__", "bias", "class", "config", "conv1", "conv2", "conv3", "def", "embed_dim", "encoder_num_hidden_layers", "eps", "for", "gradient_checkpointing", "groupnorm", "hidden_size", "idx", "in", "kernel_size", "layer_norm", "layers", "nn", "num_channels", "num_groups", "post_init", "range", "rotary_emb", "self", "stride", "super"], "moonshine/modeling_moonshine.py:MoonshineEncoder.get_input_embeddings": ["ModelEncoder", "ModelPreTrainedModel", "Module", "class", "conv1", "def", "get_input_embeddings", "nn", "return", "self"], "moonshine/modeling_moonshine.py:MoonshineEncoder.set_input_embeddings": ["ModelEncoder", "ModelPreTrainedModel", "Module", "class", "conv1", "def", "nn", "self", "set_input_embeddings", "value"], "moonshine/modeling_moonshine.py:MoonshineEncoder.forward": ["BaseModelOutputWithPast", "FloatTensor", "ModelEncoder", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "_attn_implementation", "_get_feat_extract_output_lengths", "_prepare_4d_attention_mask", "_prepare_4d_attention_mask_for_sdpa", "any", "arange", "attention_mask", "class", "config", "conv1", "conv2", "conv3", "def", "device", "downsample_stride", "dtype", "elif", "else", "encoder_layer", "flash_attention_2", "for", "forward", "functional", "gelu", "groupnorm", "hidden_states", "if", "in", "input_values", "is", "kwargs", "last_hidden_state", "layer_norm", "layers", "mask_len", "nn", "not", "permute", "position_embeddings", "position_ids", "r", "return", "rotary_emb", "sdpa", "self", "shape", "tanh", "torch", "unsqueeze"], "moonshine/modeling_moonshine.py:MoonshineDecoder.__init__": ["Embedding", "False", "LayerNorm", "ModelConfig", "ModelDecoder", "ModelDecoderLayer", "ModelPreTrainedModel", "ModelRotaryEmbedding", "ModuleList", "__init__", "bias", "class", "config", "decoder_num_hidden_layers", "def", "embed_tokens", "for", "gradient_checkpointing", "hidden_size", "idx", "in", "layers", "nn", "norm", "pad_token_id", "padding_idx", "post_init", "range", "rotary_emb", "self", "super", "vocab_size"], "moonshine/modeling_moonshine.py:MoonshineDecoder.forward": ["BaseModelOutputWithPast", "BaseModelOutputWithPastAndCrossAttentions", "Cache", "DynamicCache", "EncoderDecoderCache", "FloatTensor", "LongTensor", "ModelDecoder", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "_attn_implementation", "_prepare_4d_attention_mask", "_prepare_4d_attention_mask_for_sdpa", "and", "any", "arange", "attention_mask", "bool", "cache_position", "causal_mask", "class", "config", "create_causal_mask", "decoder_layer", "def", "device", "downsample_stride", "dtype", "elif", "else", "embed_tokens", "encoder_attention_mask", "encoder_hidden_states", "exactly", "flash_attention_2", "for", "forward", "get_seq_length", "hidden_states", "if", "in", "input_embeds", "input_ids", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layers", "mask_len", "must", "norm", "not", "of", "one", "or", "past_key_values", "past_seen_tokens", "position_embeddings", "position_ids", "r", "raise", "return", "rotary_emb", "sdpa", "self", "shape", "specify", "torch", "tuple", "unsqueeze", "use_cache"], "moonshine/modeling_moonshine.py:_compute_mask_indices": ["False", "LongTensor", "None", "ValueError", "_", "_compute_mask_indices", "and", "append", "arange", "array", "attention_mask", "batch_size", "be", "bigger", "bool", "broadcast_to", "but", "choice", "compute_num_masked_span", "concatenate", "def", "detach", "dtype", "dummy_mask_idx", "else", "epsilon", "f", "float", "for", "got", "has", "if", "in", "input_length", "input_lengths", "int", "int32", "is", "item", "len", "mask_length", "mask_prob", "max", "max_num_masked_span", "min_masks", "ndarray", "not", "np", "num_masked_span", "offsets", "ones", "put_along_axis", "raise", "rand", "random", "range", "replace", "reshape", "return", "sequence_length", "shape", "smaller", "spec_aug_mask", "spec_aug_mask_idx", "spec_aug_mask_idxs", "sum", "than", "to", "tolist", "torch", "tuple", "zeros"], "moonshine/modeling_moonshine.py:MoonshineModel.__init__": ["ModelConfig", "ModelDecoder", "ModelEncoder", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "config", "decoder", "def", "encoder", "post_init", "self", "super"], "moonshine/modeling_moonshine.py:MoonshineModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "decoder", "def", "embed_tokens", "get_input_embeddings", "return", "self"], "moonshine/modeling_moonshine.py:MoonshineModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "decoder", "def", "embed_tokens", "self", "set_input_embeddings", "value"], "moonshine/modeling_moonshine.py:MoonshineModel.freeze_encoder": ["ModelModel", "ModelPreTrainedModel", "_freeze_parameters", "class", "def", "encoder", "freeze_encoder", "self"], "moonshine/modeling_moonshine.py:MoonshineModel._mask_input_features": ["FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "True", "_compute_mask_indices", "_mask_input_features", "and", "apply_spec_augment", "attention_mask", "batch_size", "bool", "class", "config", "def", "device", "dtype", "expand", "getattr", "hidden_size", "if", "input_features", "mask_feature_indices", "mask_feature_length", "mask_feature_min_masks", "mask_feature_prob", "mask_length", "mask_prob", "mask_time_indices", "mask_time_length", "mask_time_min_masks", "mask_time_prob", "min_masks", "not", "return", "self", "sequence_length", "size", "tensor", "torch", "training"], "moonshine/modeling_moonshine.py:MoonshineModel.forward": ["BaseModelOutput", "BaseModelOutputWithPastAndCrossAttentions", "EncoderDecoderCache", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Seq2SeqModelOutput", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "bool", "cache_position", "class", "cross_attentions", "decoder", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_input_ids", "decoder_inputs_embeds", "decoder_outputs", "decoder_position_ids", "def", "encoder", "encoder_attention_mask", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "forward", "hidden_states", "if", "input_ids", "input_values", "inputs_embeds", "is", "kwargs", "last_hidden_state", "past_key_values", "position_ids", "r", "return", "self", "torch", "tuple", "use_cache"], "moonshine/modeling_moonshine.py:shift_tokens_right": ["Model_tokens_right", "Modeled_input_ids", "None", "Tensor", "ValueError", "be", "clone", "config", "decoder_start_token_id", "def", "defined", "has", "if", "input_ids", "int", "is", "masked_fill_", "model", "new_zeros", "pad_token_id", "raise", "return", "self", "shape", "to", "torch"], "moonshine/modeling_moonshine.py:MoonshineForConditionalGeneration.__init__": ["False", "GenerationMixin", "Linear", "ModelConfig", "ModelForConditionalGeneration", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "model", "nn", "post_init", "proj_out", "self", "super", "vocab_size"], "moonshine/modeling_moonshine.py:MoonshineForConditionalGeneration.get_output_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "get_output_embeddings", "proj_out", "return", "self"], "moonshine/modeling_moonshine.py:MoonshineForConditionalGeneration.set_output_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "new_embeddings", "proj_out", "self", "set_output_embeddings"], "moonshine/modeling_moonshine.py:MoonshineForConditionalGeneration.get_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "Module", "class", "def", "get_input_embeddings", "model", "nn", "return", "self"], "moonshine/modeling_moonshine.py:MoonshineForConditionalGeneration.forward": ["EncoderDecoderCache", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Seq2SeqLMOutput", "Seq2SeqModelOutput", "TransformersKwargs", "Unpack", "and", "attention_mask", "bool", "cache_position", "class", "config", "cross_attentions", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_input_ids", "decoder_inputs_embeds", "decoder_position_ids", "decoder_start_token_id", "def", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "forward", "if", "input_values", "is", "kwargs", "labels", "last_hidden_state", "logits", "loss", "loss_function", "model", "not", "outputs", "pad_token_id", "past_key_values", "proj_out", "r", "return", "self", "shift_tokens_right", "torch", "tuple", "use_cache", "vocab_size"], "moshi/modeling_moshi.py:MoshiRMSNorm.__init__": ["ModelRMSNorm", "Module", "Parameter", "__init__", "class", "def", "dim", "eps", "float", "int", "nn", "ones", "self", "super", "torch", "weight"], "moshi/modeling_moshi.py:MoshiRMSNorm._norm": ["ModelRMSNorm", "Module", "True", "_norm", "class", "def", "eps", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "torch", "x"], "moshi/modeling_moshi.py:MoshiRMSNorm.forward": ["ModelRMSNorm", "Module", "_norm", "class", "def", "float", "forward", "nn", "output", "return", "self", "type_as", "weight", "x"], "moshi/modeling_moshi.py:MoshiRMSNorm.extra_repr": ["ModelRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "weight"], "moshi/modeling_moshi.py:MoshiFlexibleLinear.__init__": ["ModelFlexibleLinear", "Module", "Parameter", "__init__", "class", "def", "input_size", "nn", "num_layers", "output_size", "randn", "self", "super", "torch", "weight"], "moshi/modeling_moshi.py:MoshiFlexibleLinear.forward": ["ModelFlexibleLinear", "Module", "None", "class", "def", "else", "forward", "if", "index_select", "is", "layer_idx", "matmul", "nn", "not", "return", "selected_weights", "self", "squeeze", "torch", "transpose", "weight", "x"], "moshi/modeling_moshi.py:MoshiLinear.__init__": ["False", "Linear", "ModelFlexibleLinear", "ModelLinear", "Module", "__init__", "bias", "class", "def", "else", "if", "input_dim", "linear", "nn", "not", "num_codebooks", "num_layers", "output_dim", "self", "super", "use_flexible_linear"], "moshi/modeling_moshi.py:MoshiLinear.forward": ["ModelLinear", "Module", "None", "class", "def", "else", "forward", "if", "layer_idx", "linear", "nn", "return", "self", "use_flexible_linear", "x"], "moshi/modeling_moshi.py:MoshiRotaryEmbedding.__init__": ["Callable", "False", "ModelConfig", "ModelRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "moshi/modeling_moshi.py:MoshiRotaryEmbedding.compute_default_rope_parameters": ["ModelConfig", "ModelRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "moshi/modeling_moshi.py:MoshiRotaryEmbedding.forward": ["False", "ModelRotaryEmbedding", "Module", "None", "and", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "moshi/modeling_moshi.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "moshi/modeling_moshi.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cos", "def", "k", "k_embed", "q", "q_embed", "return", "rotate_half", "sin", "unsqueeze", "unsqueeze_dim"], "moshi/modeling_moshi.py:MoshiGatingMLP.__init__": ["ACT2FN", "False", "Linear", "ModelFlexibleLinear", "ModelGatingMLP", "Module", "__init__", "activation_fn", "bias", "class", "config", "def", "else", "fc1", "fc2", "ffn_dim", "hidden_act", "hidden_size", "if", "nn", "num_codebooks", "num_layers", "self", "super", "use_flexible_linear"], "moshi/modeling_moshi.py:MoshiGatingMLP.forward": ["ModelGatingMLP", "Module", "None", "Tensor", "_", "activation_fn", "batch_size", "class", "def", "else", "fc1", "fc2", "forward", "hidden_states", "if", "int", "is", "layer_idx", "nn", "return", "self", "sequence_length", "shape", "torch", "view"], "moshi/modeling_moshi.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "moshi/modeling_moshi.py:MoshiAttention.__init__": ["False", "Instantiating", "ModelAttention", "ModelConfig", "ModelLinear", "ModelRotaryEmbedding", "Module", "None", "Please", "True", "ValueError", "__class__", "__init__", "__name__", "a", "and", "attention_dropout", "be", "by", "caching", "call", "class", "config", "creating", "def", "divisible", "during", "errors", "f", "forward", "got", "head_dim", "hidden_size", "if", "int", "is", "is_causal", "k_proj", "layer_idx", "lead", "logger", "make", "math", "max_position_embeddings", "must", "nn", "not", "num_attention_heads", "num_codebooks", "num_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "passing", "provide", "q_proj", "raise", "recommended", "rotary_emb", "scaling", "self", "sqrt", "super", "sure", "the", "this", "to", "use_flexible_linear", "use_rope", "used", "v_proj", "warning_once", "when", "will", "without"], "moshi/modeling_moshi.py:MoshiAttention.forward": ["Cache", "False", "LongTensor", "ModelAttention", "Module", "None", "Tensor", "ValueError", "_", "apply_rotary_pos_emb", "attention_dropout", "attention_mask", "attn_output", "attn_weights", "be", "bool", "bsz", "but", "cache_kwargs", "cache_position", "causal_mask", "class", "contiguous", "cos", "def", "dim", "dropout", "dtype", "else", "f", "float32", "forward", "functional", "head_dim", "hidden_states", "if", "is", "k_proj", "key_states", "layer_idx", "matmul", "nn", "not", "num_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "of", "output_attentions", "p", "past_key_values", "position_ids", "q_len", "q_proj", "query_states", "raise", "repeat_kv", "return", "rotary_emb", "scaling", "self", "shape", "should", "sin", "size", "softmax", "to", "torch", "training", "transpose", "tuple", "update", "use_cache", "v_proj", "value_states", "view"], "moshi/modeling_moshi.py:MoshiFlashAttention2.__init__": ["ModelAttention", "ModelFlashAttention2", "__init__", "_flash_attn_uses_top_left_mask", "args", "class", "def", "flash_attn_supports_top_left_mask", "kwargs", "self", "super"], "moshi/modeling_moshi.py:MoshiFlashAttention2.forward": ["Cache", "False", "LongTensor", "ModelAttention", "ModelFlashAttention2", "None", "StaticCache", "Tensor", "The", "ValueError", "We", "_", "_flash_attention_forward", "_flash_attn_uses_top_left_mask", "_is_quantized", "an", "and", "apply_rotary_pos_emb", "at", "attention_dropout", "attention_mask", "attn_implementation", "attn_output", "attn_weights", "back", "be", "bool", "bsz", "cache", "cache_kwargs", "cache_position", "cast", "casted", "class", "com", "compatible", "config", "contiguous", "cos", "cpu", "def", "device", "device_type", "dropout", "dropout_rate", "dtype", "elif", "else", "embedding", "f", "fact", "flash_attention_2", "float32", "forward", "get_autocast_dtype", "get_autocast_gpu_dtype", "getattr", "github", "hasattr", "have", "head_dim", "hidden", "hidden_states", "https", "huggingface", "if", "implementation", "in", "input", "input_dtype", "is", "is_autocast_enabled", "is_causal", "isinstance", "issue", "k_proj", "key_states", "layer", "layer_idx", "layers", "logger", "make", "mean", "might", "mps", "norm", "not", "num_heads", "num_key_value_heads", "o_proj", "open", "or", "output_attentions", "past_key_values", "position_ids", "q_len", "q_proj", "query_states", "raise", "related", "reshape", "return", "rotary_emb", "sdpa", "seems", "self", "silently", "sin", "size", "sliding_window", "states", "static", "sure", "target_dtype", "the", "this", "time", "to", "torch", "training", "transformers", "transpose", "tuple", "type", "upcasted", "update", "use", "use_cache", "use_top_left_mask", "v_proj", "value_states", "view", "warning_once", "weight", "will", "with", "you"], "moshi/modeling_moshi.py:MoshiSdpaAttention.forward": ["Cache", "False", "If", "LongTensor", "ModelAttention", "ModelSdpaAttention", "None", "Tensor", "The", "True", "_", "__class__", "__name__", "and", "apply_rotary_pos_emb", "attention", "attention_dropout", "attention_mask", "attn_implementation", "attn_mask", "attn_output", "be", "bool", "bsz", "cache_kwargs", "cache_position", "causal_mask", "class", "contiguous", "cos", "cuda", "def", "device", "does", "dropout_p", "eager", "else", "f", "forward", "functional", "get", "head_dim", "hidden_states", "if", "is", "is_causal", "k_proj", "key_states", "kwargs", "layer_idx", "loading", "logger", "model", "nn", "not", "num_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "output_attentions", "past_key_values", "please", "position_ids", "q_len", "q_proj", "query_states", "repeat_kv", "return", "returned", "rotary_emb", "scaled_dot_product_attention", "self", "set", "shape", "sin", "size", "support", "the", "to", "torch", "training", "transpose", "tuple", "type", "update", "use_cache", "v_proj", "value_states", "view", "want", "warning_once", "weights", "when", "will", "you"], "moshi/modeling_moshi.py:MoshiDecoderLayer.__init__": ["GradientCheckpointingLayer", "ModelConfig", "ModelDecoderLayer", "ModelGatingMLP", "ModelRMSNorm", "Model_ATTENTION_CLASSES", "True", "__init__", "_attn_implementation", "bool", "class", "config", "def", "eps", "hidden_size", "input_layernorm", "int", "layer_idx", "mlp", "post_attention_layernorm", "rms_norm_eps", "self", "self_attn", "sliding_window", "super", "use_flexible_linear", "use_rope"], "moshi/modeling_moshi.py:MoshiDecoderLayer.forward": ["Cache", "False", "FloatTensor", "GradientCheckpointingLayer", "LongTensor", "ModelDecoderLayer", "None", "Tensor", "attention_mask", "bool", "cache_position", "class", "def", "else", "forward", "hidden_states", "if", "input_layernorm", "kwargs", "mlp", "not", "output_attentions", "outputs", "past_key_values", "position_ids", "post_attention_layernorm", "residual", "return", "self", "self_attn", "self_attn_weights", "torch", "tuple", "use_cache", "use_flexible_linear"], "moshi/modeling_moshi.py:MoshiPreTrainedModel._init_weights": ["ModelFlexibleLinear", "ModelPreTrainedModel", "PreTrainedModel", "_init_weights", "class", "def", "if", "init", "isinstance", "module", "normal_", "self", "super", "weight"], "moshi/modeling_moshi.py:MoshiDepthDecoder.__init__": ["Embedding", "False", "GenerationMixin", "ModelDecoderLayer", "ModelDepthConfig", "ModelDepthDecoder", "ModelFlexibleLinear", "ModelPreTrainedModel", "ModuleList", "True", "_", "__init__", "_attn_implementation", "audio_vocab_size", "class", "config", "def", "embed_tokens", "for", "gradient_checkpointing", "hidden_size", "in", "input_projections", "input_size", "layer_idx", "layers", "lm_heads", "nn", "num_codebooks", "num_hidden_layers", "post_init", "range", "self", "super", "text_embed_tokens", "use_flexible_linear", "use_rope", "vocab_size"], "moshi/modeling_moshi.py:MoshiDepthDecoder.forward": ["BaseModelOutputWithPast", "BoolTensor", "Cache", "CausalLMOutputWithPast", "CrossEntropyLoss", "DynamicCache", "False", "FloatTensor", "GenerationMixin", "LongTensor", "ModelDepthDecoder", "ModelPreTrainedModel", "None", "Setting", "True", "_update_causal_mask", "all_hidden_states", "all_self_attns", "and", "append", "arange", "attention_mask", "attentions", "audio_vocab_size", "bool", "cache_position", "cat", "causal_mask", "checkpointing", "class", "config", "decoder_layer", "def", "device", "dim", "else", "embed_tokens", "float", "for", "forward", "get_seq_length", "gradient", "gradient_checkpointing", "hidden_states", "if", "in", "incompatible", "input_ids", "input_projections", "inputs_embeds", "is", "item", "kwargs", "labels", "last_hidden_state", "layer_outputs", "layers", "lm_heads", "logger", "logits", "loss", "loss_fct", "masked_fill", "not", "output_attentions", "output_hidden_states", "past_key_values", "past_seen_tokens", "position_ids", "position_idx", "reshape", "return", "return_dict", "self", "shape", "text_embed_tokens", "to", "torch", "training", "tuple", "unsqueeze", "use_cache", "use_return_dict", "v", "warning_once", "with"], "moshi/modeling_moshi.py:MoshiDepthDecoder._update_causal_mask": ["Attention", "AttentionMaskConverter", "BlockMask", "Cache", "False", "Flash", "GenerationMixin", "Make", "Model", "ModelDepthDecoder", "ModelPreTrainedModel", "None", "StaticCache", "Tensor", "Union", "ValueError", "You", "_attn_implementation", "_ignore_causal_mask_sdpa", "_prepare_4d_causal_attention_mask_with_cache_position", "_unmask_unattended", "_update_causal_mask", "and", "are", "attempting", "attention_mask", "batch_size", "batched", "before", "behaviour", "bool", "cache_position", "call", "causal_mask", "class", "config", "cuda", "def", "device", "dtype", "else", "finfo", "flash_attention_2", "flex_attention", "for", "generation", "get_max_cache_shape", "get_seq_length", "if", "in", "input", "input_tensor", "inputs_embeds", "is", "is_padding_right", "is_training", "isinstance", "item", "lead", "left", "make_flex_block_causal_mask", "may", "min", "min_dtype", "not", "npu", "of", "output_attentions", "padding_side", "past_key_values", "past_key_values_length", "past_seen_tokens", "perform", "raise", "return", "right", "sdpa", "self", "sequence_length", "shape", "size", "sliding_window", "sum", "sure", "target_length", "the", "this", "to", "tokenizer", "tokenizing", "torch", "training", "type", "unexpected", "using_static_cache", "version", "with", "xpu"], "moshi/modeling_moshi.py:MoshiDepthDecoder._prepare_4d_causal_attention_mask_with_cache_position": ["Cache", "GenerationMixin", "ModelDepthConfig", "ModelDepthDecoder", "ModelPreTrainedModel", "None", "StaticCache", "Tensor", "True", "_prepare_4d_causal_attention_mask_with_cache_position", "all", "and", "arange", "attention_mask", "batch_size", "bitwise_or_", "cache_position", "causal_mask", "class", "clone", "config", "def", "device", "diagonal_attend_mask", "dim", "dtype", "else", "expand", "fill_value", "finfo", "full", "get_text_config", "getattr", "if", "int", "is", "is_sliding", "is_static_sliding_cache", "isinstance", "mask_length", "masked_fill", "min", "min_dtype", "not", "or", "padding_mask", "past_key_values", "reshape", "return", "sequence_length", "shape", "sliding_attend_mask", "sliding_window", "target_length", "text_config", "to", "torch", "use_sliding_window"], "moshi/modeling_moshi.py:MoshiModel.__init__": ["Embedding", "False", "ModelConfig", "ModelDecoderLayer", "ModelModel", "ModelPreTrainedModel", "ModelRMSNorm", "ModuleList", "__init__", "class", "config", "def", "embed_tokens", "eps", "for", "gradient_checkpointing", "hidden_size", "in", "layer_idx", "layers", "nn", "norm", "num_hidden_layers", "pad_token_id", "padding_idx", "post_init", "range", "rms_norm_eps", "self", "super", "use_flexible_linear", "vocab_size"], "moshi/modeling_moshi.py:MoshiModel.forward": ["BaseModelOutputWithPast", "Cache", "DynamicCache", "False", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Setting", "Tensor", "True", "_update_causal_mask", "all_hidden_states", "all_self_attns", "and", "arange", "attention_mask", "attentions", "bool", "cache_position", "causal_mask", "checkpointing", "class", "config", "decoder_layer", "def", "device", "else", "embed_tokens", "for", "forward", "get_seq_length", "gradient", "gradient_checkpointing", "hidden_states", "if", "in", "incompatible", "input_ids", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layer_outputs", "layers", "logger", "norm", "not", "output_attentions", "output_hidden_states", "past_key_values", "past_seen_tokens", "position_ids", "return", "return_dict", "self", "shape", "torch", "training", "tuple", "unsqueeze", "use_cache", "use_return_dict", "v", "warning_once", "with"], "moshi/modeling_moshi.py:MoshiModel._update_causal_mask": ["Attention", "AttentionMaskConverter", "BlockMask", "Cache", "False", "Flash", "Make", "Model", "ModelModel", "ModelPreTrainedModel", "None", "StaticCache", "Tensor", "Union", "ValueError", "You", "_attn_implementation", "_ignore_causal_mask_sdpa", "_prepare_4d_causal_attention_mask_with_cache_position", "_unmask_unattended", "_update_causal_mask", "and", "are", "attempting", "attention_mask", "batch_size", "batched", "before", "behaviour", "bool", "cache_position", "call", "causal_mask", "class", "config", "cuda", "def", "device", "dtype", "else", "finfo", "flash_attention_2", "flex_attention", "for", "generation", "get_max_cache_shape", "get_seq_length", "if", "in", "input", "input_tensor", "inputs_embeds", "is", "is_padding_right", "is_training", "isinstance", "item", "lead", "left", "make_flex_block_causal_mask", "may", "min", "min_dtype", "not", "npu", "of", "output_attentions", "padding_side", "past_key_values", "past_key_values_length", "past_seen_tokens", "perform", "raise", "return", "right", "sdpa", "self", "sequence_length", "shape", "size", "sliding_window", "sum", "sure", "target_length", "the", "this", "to", "tokenizer", "tokenizing", "torch", "training", "type", "unexpected", "using_static_cache", "version", "with", "xpu"], "moshi/modeling_moshi.py:MoshiModel._prepare_4d_causal_attention_mask_with_cache_position": ["Cache", "ModelConfig", "ModelModel", "ModelPreTrainedModel", "None", "StaticCache", "Tensor", "True", "_prepare_4d_causal_attention_mask_with_cache_position", "all", "and", "arange", "attention_mask", "batch_size", "bitwise_or_", "cache_position", "causal_mask", "class", "clone", "config", "def", "device", "diagonal_attend_mask", "dim", "dtype", "else", "expand", "fill_value", "finfo", "full", "get_text_config", "getattr", "if", "int", "is", "is_sliding", "is_static_sliding_cache", "isinstance", "mask_length", "masked_fill", "min", "min_dtype", "not", "or", "padding_mask", "past_key_values", "reshape", "return", "sequence_length", "shape", "sliding_attend_mask", "sliding_window", "target_length", "text_config", "to", "torch", "use_sliding_window"], "moshi/modeling_moshi.py:MoshiForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "ModelForCausalLM", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "vocab_size"], "moshi/modeling_moshi.py:MoshiForCausalLM.forward": ["Cache", "FloatTensor", "GenerationMixin", "LongTensor", "ModelCausalLMOutputWithPast", "ModelForCausalLM", "ModelPreTrainedModel", "None", "Tensor", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "contiguous", "def", "device", "else", "float", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "output", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "position_ids", "r", "return", "return_dict", "self", "shift_labels", "shift_logits", "slice", "slice_indices", "to", "torch", "tuple", "use_cache", "use_return_dict", "view", "vocab_size"], "moshi/modeling_moshi.py:MoshiForConditionalGeneration.__init__": ["AutoModel", "Embedding", "GenerationMixin", "ModelConfig", "ModelDepthDecoder", "ModelForCausalLM", "ModelForConditionalGeneration", "ModelPreTrainedModel", "ModuleList", "_", "__init__", "_from_config", "audio_encoder", "audio_encoder_config", "audio_vocab_size", "class", "config", "decoder", "def", "depth_decoder", "depth_decoder_config", "embed_tokens", "for", "from_config", "hidden_size", "in", "nn", "num_codebooks", "post_init", "range", "self", "super"], "moshi/modeling_moshi.py:MoshiForConditionalGeneration.get_depth_decoder": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "depth_decoder", "get_depth_decoder", "return", "self"], "moshi/modeling_moshi.py:MoshiForConditionalGeneration.forward": ["BoolTensor", "Cache", "FloatTensor", "GenerationMixin", "LongTensor", "ModelConditionalGenerationOutputWithPast", "ModelForConditionalGeneration", "ModelPreTrainedModel", "Model_audio_codes", "Model_input_values", "None", "Seq2SeqLMOutput", "Tensor", "True", "ValueError", "You", "and", "argument", "at", "attention_mask", "attentions", "audio_codes", "audio_encoder", "audio_encoder_", "audio_inputs_embeds", "audio_labels", "audio_logits", "audio_vocab_size", "bool", "bos_token_id", "build_delay_pattern_mask", "cat", "class", "codebook", "config", "decoder", "decoder_", "decoder_last_hidden_state", "decoder_outputs", "def", "depth_attentions", "depth_decoder", "depth_decoder_", "depth_decoder_outputs", "depth_hidden_states", "depth_input_ids", "depth_loss", "depth_past_key_values", "device", "dim", "else", "embed_tokens", "encode", "final_loss", "for", "forward", "hidden_states", "if", "in", "input_ids", "input_values", "inputs_embeds", "is", "items", "kwargs", "kwargs_audio_encoder", "kwargs_decoder", "kwargs_depth_decoder", "labels", "last_hidden_state", "least", "len", "logits", "loss", "max_length", "model", "must", "not", "num_codebooks", "num_quantizers", "of", "one", "output_attentions", "output_hidden_states", "outputs", "pad_token_id", "past_key_values", "provide", "r", "raise", "range", "reshape", "return", "return_dict", "self", "shape", "startswith", "sum", "text_labels", "to", "to_tuple", "torch", "transpose", "tuple", "use_cache", "use_return_dict", "user_audio_codes", "user_input_values", "value", "view"], "moshi/modeling_moshi.py:MoshiForConditionalGeneration._prepare_attention_mask_for_generation": ["Any", "GenerationConfig", "GenerationMixin", "LongTensor", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "_prepare_attention_mask_for_generation", "and", "any", "attention_mask", "attention_mask_from_padding", "can_infer_attention_mask", "class", "def", "default_attention_mask", "device", "dict", "dtype", "eos_token_id", "generation_config", "if", "input_ids", "is", "is_pad_token_in_inputs", "is_pad_token_not_equal_to_eos_token_id", "isin", "kwargs", "long", "ne", "not", "ones", "or", "pad_token_id", "return", "self", "shape", "str", "torch"], "moshi/modeling_moshi.py:MoshiForConditionalGeneration._prepare_inputs_embeds_for_generation": ["False", "FloatTensor", "GenerationConfig", "GenerationMixin", "LongTensor", "ModelForConditionalGeneration", "ModelPreTrainedModel", "Model_audio_codes", "Model_delay_pattern_mask", "Model_input_values", "None", "Tensor", "ValueError", "You", "_prepare_inputs_embeds_for_generation", "and", "apply_delay_pattern_mask", "at", "attention_mask", "audio_codes", "audio_encoder", "audio_inputs_embeds", "audio_vocab_size", "bool", "bos_token_id", "build_delay_pattern_mask", "cat", "class", "codebook", "concat_unconditional_inputs", "config", "decoder", "def", "device", "dim", "elif", "else", "embed_tokens", "encode", "for", "generation_config", "get_unconditional_inputs", "if", "in", "input_ids", "inputs_embeds", "is", "least", "max_length", "model", "must", "not", "num_codebooks", "num_quantizers", "num_samples", "of", "one", "or", "pad_token_id", "provide", "raise", "range", "return", "self", "shape", "sum", "to", "torch", "unconditional_inputs", "user_audio_codes", "user_delay_pattern_mask", "user_input_values"], "moshi/modeling_moshi.py:MoshiForConditionalGeneration.generate": ["FloatTensor", "GenerationMixin", "LongTensor", "ModelConditionalGenerationGenerateOutput", "ModelForConditionalGeneration", "ModelPreTrainedModel", "Model_audio_codes", "Model_delay_pattern_mask", "Model_input_values", "None", "Tensor", "True", "_check_and_maybe_initialize_inputs", "_prepare_attention_mask_for_generation", "_prepare_generated_length", "_prepare_generation_config", "_prepare_inputs_embeds_for_generation", "and", "apply_delay_pattern_mask", "argument", "attention_mask", "audio_codes", "audio_encoder", "audio_encoder_config", "audio_sequences", "audio_values", "audio_vocab_size", "beam_indices", "blank_input_values", "blank_user_audio_codes", "bool", "cache_implementation", "cat", "class", "concat_unconditional_inputs", "config", "cpu", "d", "decode", "def", "depth_decoder", "depth_decoder_", "depth_decoder_config", "depth_decoder_generation_config", "depth_decoder_kwargs_keys", "device", "dim", "disk", "dtype", "else", "encode", "expand", "expanded_beam_indices", "for", "frame_rate", "gather", "generate", "generated_audio_codes", "generation_config", "get", "has_default_max_length", "has_default_min_length", "hasattr", "hf_device_map", "if", "in", "index", "index_select", "input_ids", "input_ids_length", "inputs", "inputs_embeds", "inputs_tensor", "int", "is", "items", "key", "kwargs", "kwargs_depth_decoder", "last_generated_audio_codes", "last_hidden_state", "len", "main_device", "mask", "max", "max_length", "min_length", "model_input_name", "not", "num_beams", "num_codebooks", "num_quantizers", "num_return_sequences", "or", "output_audio_codes", "output_scores", "output_text_ids", "output_values", "outputs", "pop", "repeat_interleave", "reshape", "return", "return_audio_codes", "return_audio_waveforms", "return_dict_in_generate", "sampling_rate", "self", "sequences", "shape", "startswith", "static", "super", "torch", "unsqueeze", "update", "user_audio_codes", "user_delay_pattern_mask", "user_input_values", "value", "values", "view", "zeros"], "moshi/modeling_moshi.py:MoshiForConditionalGeneration.prepare_inputs_for_generation": ["False", "FloatTensor", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "Model_audio_codes", "Model_delay_pattern_mask", "None", "True", "_", "_prepare_inputs_embeds_for_generation", "apply_delay_pattern_mask", "attention_mask", "blank_user_audio_codes", "cache_position", "cat", "class", "def", "depth_decoder", "device", "dim", "generate", "generated_audio_codes", "if", "input_ids", "inputs_embeds", "is", "is_first_iteration", "kwargs", "kwargs_depth_decoder", "last_hidden_state", "logits_to_keep", "model_inputs", "not", "past_key_values", "pop", "position_ids", "prepare_inputs_for_generation", "return", "self", "shape", "super", "to", "torch", "unsqueeze", "use_cache", "user_audio_codes", "user_delay_pattern_mask", "view"], "moshi/modeling_moshi.py:MoshiForConditionalGeneration._update_model_kwargs_for_generation": ["Any", "False", "GenerationMixin", "ModelForConditionalGeneration", "ModelOutput", "ModelPreTrainedModel", "_update_model_kwargs_for_generation", "bool", "class", "def", "dict", "get", "int", "is_encoder_decoder", "last_hidden_state", "model_kwargs", "num_new_tokens", "outputs", "return", "self", "str", "super"], "moshi/modeling_moshi.py:MoshiForConditionalGeneration.get_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "decoder", "def", "get_input_embeddings", "return", "self"], "moshi/modeling_moshi.py:MoshiForConditionalGeneration.set_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "decoder", "def", "self", "set_input_embeddings", "value"], "moshi/modeling_moshi.py:MoshiForConditionalGeneration.get_output_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "decoder", "def", "get_output_embeddings", "return", "self"], "moshi/modeling_moshi.py:MoshiForConditionalGeneration.set_output_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "decoder", "def", "new_embeddings", "self", "set_output_embeddings"], "moshi/modeling_moshi.py:MoshiForConditionalGeneration.freeze_audio_encoder": ["False", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "_requires_grad", "audio_encoder", "class", "def", "for", "freeze_audio_encoder", "in", "param", "parameters", "requires_grad", "self"], "moshi/modeling_moshi.py:MoshiForConditionalGeneration.freeze_depth_decoder": ["False", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "_requires_grad", "class", "def", "depth_decoder", "for", "freeze_depth_decoder", "in", "param", "parameters", "requires_grad", "self"], "moshi/modeling_moshi.py:MoshiForConditionalGeneration.apply_delay_pattern_mask": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "apply_delay_pattern_mask", "class", "decoder_pad_token_mask", "def", "input_ids", "return", "seq_len", "shape", "torch", "where"], "moshi/modeling_moshi.py:MoshiForConditionalGeneration.build_delay_pattern_mask": ["GenerationMixin", "LongTensor", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "bos_token_id", "bsz", "build_delay_pattern_mask", "class", "def", "device", "dtype", "else", "generation_config", "if", "input_ids", "input_ids_shifted", "int", "is", "long", "max_length", "min", "not", "num_codebooks", "ones", "pad_token_id", "pattern_mask", "return", "self", "seq_len", "seq_len_to_keep", "shape", "torch"], "moshi/modeling_moshi.py:MoshiForConditionalGeneration.get_unconditional_inputs": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "ModelUnconditionalInput", "Model_audio_codes", "attention_mask", "audio_vocab_size", "class", "config", "def", "device", "dtype", "get_unconditional_inputs", "input_ids", "int64", "long", "num_codebooks", "num_samples", "ones", "return", "self", "torch", "user_audio_codes", "vocab_size"], "moshi/modeling_moshi.py:MoshiForConditionalGeneration._check_and_maybe_initialize_inputs": ["At", "Check", "False", "GenerationMixin", "Make", "Model", "ModelForConditionalGeneration", "ModelPreTrainedModel", "Model_audio_codes", "Model_input", "Model_input_values", "Model_seq_length", "No", "None", "ValueError", "_check_and_maybe_initialize_inputs", "all", "alongside", "and", "as", "audio", "audio_encoder_config", "been", "ceil", "class", "concat_unconditional_inputs", "config", "def", "docstrings", "doesn", "either", "elif", "else", "for", "frame_rate", "get_unconditional_inputs", "have", "if", "information", "input_ids", "inputs", "inputs_embeds", "is", "least", "length", "math", "more", "not", "of", "one", "one_input_has_been_passed", "or", "other", "others", "passed", "raise", "ratio", "return", "same", "sampling_rate", "self", "sequence", "shape", "sure", "t", "that", "the", "they", "tokens_seq_length", "unconditional_inputs", "use", "user", "user_audio_codes", "user_input", "user_input_values", "user_seq_length"], "mpnet/modeling_mpnet.py:MPNetPreTrainedModel._init_weights": ["Model", "ModelEmbeddings", "ModelLMHead", "PreTrainedModel", "_init_weights", "arange", "bias", "class", "copy_", "def", "elif", "expand", "if", "init", "isinstance", "module", "position_ids", "self", "shape", "super", "torch", "zeros_"], "mpnet/modeling_mpnet.py:MPNetEmbeddings.__init__": ["Dropout", "Embedding", "False", "LayerNorm", "Model", "Module", "__init__", "arange", "class", "config", "def", "dropout", "eps", "expand", "hidden_dropout_prob", "hidden_size", "layer_norm_eps", "max_position_embeddings", "nn", "padding_idx", "persistent", "position_embeddings", "position_ids", "register_buffer", "self", "super", "torch", "vocab_size", "word_embeddings"], "mpnet/modeling_mpnet.py:MPNetEmbeddings.forward": ["LayerNorm", "Model", "Module", "None", "class", "create_position_ids_from_input_ids", "create_position_ids_from_inputs_embeds", "def", "dropout", "else", "embeddings", "forward", "if", "input_ids", "input_shape", "inputs_embeds", "is", "kwargs", "nn", "not", "padding_idx", "position_embeddings", "position_ids", "return", "self", "seq_length", "size", "word_embeddings"], "mpnet/modeling_mpnet.py:MPNetEmbeddings.create_position_ids_from_inputs_embeds": ["Model", "Module", "arange", "class", "create_position_ids_from_inputs_embeds", "def", "device", "dtype", "expand", "input_shape", "inputs_embeds", "long", "nn", "padding_idx", "position_ids", "return", "self", "sequence_length", "size", "torch", "unsqueeze"], "mpnet/modeling_mpnet.py:MPNetSelfAttention.__init__": ["Dropout", "Linear", "Model", "Module", "The", "ValueError", "__init__", "a", "all_head_size", "and", "attention", "attention_head_size", "attention_probs_dropout_prob", "class", "config", "def", "dropout", "embedding_size", "f", "hasattr", "heads", "hidden", "hidden_size", "if", "int", "is", "k", "multiple", "nn", "not", "num_attention_heads", "number", "o", "of", "q", "raise", "self", "size", "super", "the", "v"], "mpnet/modeling_mpnet.py:MPNetSelfAttention.forward": ["False", "Model", "Module", "None", "_", "all_head_size", "attention_head_size", "attention_mask", "attention_probs", "attention_scores", "batch_size", "c", "class", "contiguous", "def", "dim", "dropout", "else", "forward", "functional", "hidden_states", "if", "is", "k", "kwargs", "math", "matmul", "new_c_shape", "nn", "not", "num_attention_heads", "o", "output_attentions", "outputs", "permute", "position_bias", "q", "return", "self", "seq_length", "shape", "size", "softmax", "sqrt", "torch", "transpose", "v", "view"], "mpnet/modeling_mpnet.py:MPNetAttention.__init__": ["Dropout", "LayerNorm", "Model", "ModelSelfAttention", "Module", "__init__", "attn", "class", "config", "def", "dropout", "eps", "hidden_dropout_prob", "hidden_size", "layer_norm_eps", "nn", "self", "super"], "mpnet/modeling_mpnet.py:MPNetAttention.forward": ["False", "LayerNorm", "Model", "Module", "None", "attention_mask", "attention_output", "attn", "class", "def", "dropout", "forward", "hidden_states", "kwargs", "nn", "output_attentions", "outputs", "position_bias", "return", "self", "self_outputs"], "mpnet/modeling_mpnet.py:MPNetIntermediate.__init__": ["ACT2FN", "Linear", "Model", "Module", "__init__", "class", "config", "def", "dense", "else", "hidden_act", "hidden_size", "if", "intermediate_act_fn", "intermediate_size", "isinstance", "nn", "self", "str", "super"], "mpnet/modeling_mpnet.py:MPNetIntermediate.forward": ["Model", "Module", "Tensor", "class", "def", "dense", "forward", "hidden_states", "intermediate_act_fn", "nn", "return", "self", "torch"], "mpnet/modeling_mpnet.py:MPNetOutput.__init__": ["Dropout", "LayerNorm", "Linear", "Model", "Module", "__init__", "class", "config", "def", "dense", "dropout", "eps", "hidden_dropout_prob", "hidden_size", "intermediate_size", "layer_norm_eps", "nn", "self", "super"], "mpnet/modeling_mpnet.py:MPNetOutput.forward": ["LayerNorm", "Model", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "mpnet/modeling_mpnet.py:MPNetLayer.__init__": ["Model", "ModelAttention", "ModelIntermediate", "ModelOutput", "Module", "__init__", "attention", "class", "config", "def", "intermediate", "nn", "output", "self", "super"], "mpnet/modeling_mpnet.py:MPNetLayer.forward": ["False", "Model", "Module", "None", "attention", "attention_mask", "attention_output", "class", "def", "forward", "hidden_states", "intermediate", "intermediate_output", "kwargs", "layer_output", "nn", "output", "output_attentions", "outputs", "position_bias", "return", "self", "self_attention_outputs"], "mpnet/modeling_mpnet.py:MPNetEncoder.__init__": ["Embedding", "Model", "ModelLayer", "Module", "ModuleList", "_", "__init__", "class", "config", "def", "for", "in", "layer", "n_heads", "nn", "num_attention_heads", "num_hidden_layers", "range", "relative_attention_bias", "relative_attention_num_buckets", "self", "super"], "mpnet/modeling_mpnet.py:MPNetEncoder.forward": ["BaseModelOutput", "False", "Model", "Module", "None", "Tensor", "all_attentions", "all_hidden_states", "attention_mask", "attentions", "bool", "class", "compute_position_bias", "def", "else", "enumerate", "for", "forward", "hidden_states", "i", "if", "in", "is", "kwargs", "last_hidden_state", "layer", "layer_module", "layer_outputs", "nn", "not", "output_attentions", "output_hidden_states", "position_bias", "return", "return_dict", "self", "torch", "tuple", "v"], "mpnet/modeling_mpnet.py:MPNetEncoder.compute_position_bias": ["Model", "Module", "None", "arange", "bsz", "class", "compute_position_bias", "context_position", "contiguous", "def", "device", "dtype", "else", "expand", "if", "is", "klen", "long", "memory_position", "nn", "not", "num_buckets", "permute", "position_ids", "qlen", "relative_attention_bias", "relative_position", "relative_position_bucket", "return", "rp_bucket", "self", "size", "to", "torch", "unsqueeze", "values", "x"], "mpnet/modeling_mpnet.py:MPNetEncoder.relative_position_bucket": ["Model", "Module", "abs", "class", "def", "float", "full_like", "is_small", "log", "long", "math", "max_distance", "max_exact", "min", "n", "nn", "num_buckets", "relative_position", "relative_position_bucket", "ret", "return", "to", "torch", "val_if_large", "where"], "mpnet/modeling_mpnet.py:MPNetPooler.__init__": ["Linear", "Model", "Module", "Tanh", "__init__", "activation", "class", "config", "def", "dense", "hidden_size", "nn", "self", "super"], "mpnet/modeling_mpnet.py:MPNetPooler.forward": ["Model", "Module", "Tensor", "activation", "class", "def", "dense", "first_token_tensor", "forward", "hidden_states", "nn", "pooled_output", "return", "self", "torch"], "mpnet/modeling_mpnet.py:MPNetModel.__init__": ["Model", "ModelEmbeddings", "ModelEncoder", "ModelPooler", "ModelPreTrainedModel", "None", "True", "__init__", "add_pooling_layer", "class", "config", "def", "else", "embeddings", "encoder", "if", "pooler", "post_init", "r", "self", "super"], "mpnet/modeling_mpnet.py:MPNetModel.get_input_embeddings": ["Model", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "return", "self", "word_embeddings"], "mpnet/modeling_mpnet.py:MPNetModel.set_input_embeddings": ["Model", "ModelPreTrainedModel", "class", "def", "embeddings", "self", "set_input_embeddings", "value", "word_embeddings"], "mpnet/modeling_mpnet.py:MPNetModel.forward": ["BaseModelOutputWithPooling", "FloatTensor", "LongTensor", "Model", "ModelPreTrainedModel", "None", "Tensor", "ValueError", "You", "and", "at", "attention_mask", "attentions", "bool", "both", "cannot", "class", "config", "def", "device", "either", "elif", "else", "embedding_output", "embeddings", "encoder", "encoder_outputs", "extended_attention_mask", "forward", "get_extended_attention_mask", "have", "hidden_states", "if", "input_ids", "input_shape", "inputs_embeds", "is", "kwargs", "last_hidden_state", "not", "ones", "or", "output_attentions", "output_hidden_states", "pooled_output", "pooler", "pooler_output", "position_ids", "raise", "return", "return_dict", "same", "self", "sequence_output", "size", "specify", "the", "time", "to", "torch", "tuple", "use_return_dict", "warn_if_padding_and_no_attention_mask"], "mpnet/modeling_mpnet.py:MPNetForMaskedLM.__init__": ["False", "Model", "ModelLMHead", "ModelModel", "ModelPreTrainedModel", "__init__", "add_pooling_layer", "class", "config", "def", "lm_head", "post_init", "self", "super"], "mpnet/modeling_mpnet.py:MPNetForMaskedLM.get_output_embeddings": ["Model", "ModelPreTrainedModel", "class", "decoder", "def", "get_output_embeddings", "lm_head", "return", "self"], "mpnet/modeling_mpnet.py:MPNetForMaskedLM.set_output_embeddings": ["Model", "ModelPreTrainedModel", "bias", "class", "decoder", "def", "lm_head", "new_embeddings", "self", "set_output_embeddings"], "mpnet/modeling_mpnet.py:MPNetForMaskedLM.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "MaskedLMOutput", "Model", "ModelPreTrainedModel", "None", "Tensor", "attention_mask", "attentions", "bool", "class", "config", "def", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "lm_head", "logits", "loss", "loss_fct", "masked_lm_loss", "not", "output", "output_attentions", "output_hidden_states", "outputs", "position_ids", "prediction_scores", "r", "return", "return_dict", "self", "sequence_output", "torch", "tuple", "use_return_dict", "view", "vocab_size"], "mpnet/modeling_mpnet.py:MPNetLMHead.__init__": ["LayerNorm", "Linear", "Model", "Module", "Parameter", "True", "__init__", "bias", "class", "config", "decoder", "def", "dense", "eps", "hidden_size", "layer_norm", "layer_norm_eps", "nn", "self", "super", "torch", "vocab_size", "zeros"], "mpnet/modeling_mpnet.py:MPNetLMHead.forward": ["Model", "Module", "class", "decoder", "def", "dense", "features", "forward", "gelu", "kwargs", "layer_norm", "nn", "return", "self", "x"], "mpnet/modeling_mpnet.py:MPNetForSequenceClassification.__init__": ["False", "Model", "ModelClassificationHead", "ModelModel", "ModelPreTrainedModel", "__init__", "add_pooling_layer", "class", "classifier", "config", "def", "num_labels", "post_init", "self", "super"], "mpnet/modeling_mpnet.py:MPNetForSequenceClassification.forward": ["BCEWithLogitsLoss", "CrossEntropyLoss", "FloatTensor", "LongTensor", "MSELoss", "Model", "ModelPreTrainedModel", "None", "SequenceClassifierOutput", "Tensor", "and", "attention_mask", "attentions", "bool", "class", "classifier", "config", "def", "dtype", "elif", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "kwargs", "labels", "logits", "long", "loss", "loss_fct", "multi_label_classification", "not", "num_labels", "or", "output", "output_attentions", "output_hidden_states", "outputs", "position_ids", "problem_type", "r", "regression", "return", "return_dict", "self", "sequence_output", "single_label_classification", "squeeze", "torch", "tuple", "use_return_dict", "view"], "mpnet/modeling_mpnet.py:MPNetForMultipleChoice.__init__": ["Dropout", "Linear", "Model", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "classifier", "config", "def", "dropout", "hidden_dropout_prob", "hidden_size", "nn", "post_init", "self", "super"], "mpnet/modeling_mpnet.py:MPNetForMultipleChoice.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "Model", "ModelPreTrainedModel", "MultipleChoiceModelOutput", "None", "Tensor", "attention_mask", "attentions", "bool", "class", "classifier", "config", "def", "dropout", "else", "flat_attention_mask", "flat_input_ids", "flat_inputs_embeds", "flat_position_ids", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "not", "num_choices", "output", "output_attentions", "output_hidden_states", "outputs", "pooled_output", "position_ids", "r", "reshaped_logits", "return", "return_dict", "self", "shape", "size", "torch", "tuple", "use_return_dict", "view"], "mpnet/modeling_mpnet.py:MPNetForTokenClassification.__init__": ["Dropout", "False", "Linear", "Model", "ModelModel", "ModelPreTrainedModel", "__init__", "add_pooling_layer", "class", "classifier", "config", "def", "dropout", "hidden_dropout_prob", "hidden_size", "nn", "num_labels", "post_init", "self", "super"], "mpnet/modeling_mpnet.py:MPNetForTokenClassification.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "Model", "ModelPreTrainedModel", "None", "Tensor", "TokenClassifierOutput", "attention_mask", "attentions", "bool", "class", "classifier", "config", "def", "dropout", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "not", "num_labels", "output", "output_attentions", "output_hidden_states", "outputs", "position_ids", "r", "return", "return_dict", "self", "sequence_output", "torch", "tuple", "use_return_dict", "view"], "mpnet/modeling_mpnet.py:MPNetClassificationHead.__init__": ["Dropout", "Linear", "Model", "Module", "__init__", "class", "config", "def", "dense", "dropout", "hidden_dropout_prob", "hidden_size", "nn", "num_labels", "out_proj", "self", "super"], "mpnet/modeling_mpnet.py:MPNetClassificationHead.forward": ["Model", "Module", "class", "def", "dense", "dropout", "features", "forward", "kwargs", "nn", "out_proj", "return", "self", "tanh", "torch", "x"], "mpnet/modeling_mpnet.py:MPNetForQuestionAnswering.__init__": ["False", "Linear", "Model", "ModelModel", "ModelPreTrainedModel", "__init__", "add_pooling_layer", "class", "config", "def", "hidden_size", "nn", "num_labels", "post_init", "qa_outputs", "self", "super"], "mpnet/modeling_mpnet.py:MPNetForQuestionAnswering.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "Model", "ModelPreTrainedModel", "None", "QuestionAnsweringModelOutput", "Tensor", "and", "attention_mask", "attentions", "bool", "clamp", "class", "config", "contiguous", "def", "dim", "else", "end_logits", "end_loss", "end_positions", "forward", "hidden_states", "if", "ignore_index", "ignored_index", "input_ids", "inputs_embeds", "is", "kwargs", "len", "logits", "loss", "loss_fct", "not", "output", "output_attentions", "output_hidden_states", "outputs", "position_ids", "qa_outputs", "return", "return_dict", "self", "sequence_output", "size", "split", "squeeze", "start_logits", "start_loss", "start_positions", "torch", "total_loss", "tuple", "use_return_dict"], "mpnet/modeling_mpnet.py:create_position_ids_from_input_ids": ["Model_position_ids_from_input_ids", "cumsum", "def", "dim", "incremental_indices", "input_ids", "int", "long", "mask", "ne", "padding_idx", "return", "torch", "type_as"], "mpt/modeling_mpt.py:build_mpt_alibi_tensor": ["Model_Model_alibi_tensor", "None", "alibi", "alibi_bias_max", "arange", "base", "ceil", "concat", "def", "device", "dim", "dtype", "float", "if", "int32", "int64", "log2", "math", "num_heads", "num_heads_power_of_2", "pow", "r", "return", "sequence_length", "slopes", "squeeze", "torch", "view"], "mpt/modeling_mpt.py:MptAttention.__init__": ["False", "Linear", "ModelAttention", "ModelConfig", "Module", "None", "Wqkv", "__init__", "attn_config", "attn_dropout_p", "attn_pdrop", "bias", "class", "clip_qkv", "config", "def", "head_dim", "hidden_size", "if", "int", "is", "layer_idx", "math", "max_seq_len", "max_seq_length", "n_heads", "nn", "out_proj", "self", "softmax_scale", "sqrt", "super"], "mpt/modeling_mpt.py:MptAttention.forward": ["Cache", "Expecting", "ModelAttention", "Module", "None", "Tensor", "ValueError", "Wqkv", "attention_mask", "attention_scores", "attn_dropout_p", "attn_output", "attn_weights", "batch_size", "be", "cache_kwargs", "cache_position", "chunk", "clamp", "class", "clip_qkv", "context_states", "contiguous", "def", "dim", "dimensions", "dropout", "dtype", "else", "f", "finfo", "float", "forward", "functional", "get_seq_length", "got", "head_dim", "hidden_states", "if", "is", "key_length", "key_states", "layer_idx", "len", "masked_fill", "matmul", "max", "min", "mixed_qkv", "n_heads", "nn", "not", "out_proj", "p", "past_key_values", "permute", "position_bias", "position_bias_key_index", "position_bias_query_index", "query_length", "query_states", "raise", "reshape", "return", "self", "seq_length", "shape", "size", "softmax", "softmax_scale", "to", "torch", "training", "transpose", "update", "value_states", "view"], "mpt/modeling_mpt.py:MptMLP.__init__": ["False", "GELU", "Linear", "ModelConfig", "ModelMLP", "Module", "__init__", "act", "approximate", "attn_config", "attn_pdrop", "bias", "class", "config", "def", "down_proj", "hidden_dropout", "hidden_size", "nn", "none", "self", "super", "up_proj"], "mpt/modeling_mpt.py:MptMLP.forward": ["F", "ModelMLP", "Module", "Tensor", "act", "class", "def", "down_proj", "dropout", "forward", "hidden_dropout", "hidden_states", "intermediate_output", "nn", "output", "p", "residual", "return", "self", "torch", "training", "up_proj"], "mpt/modeling_mpt.py:MptBlock.__init__": ["Dropout", "GradientCheckpointingLayer", "LayerNorm", "ModelAttention", "ModelBlock", "ModelConfig", "ModelMLP", "None", "__init__", "attn", "attn_config", "attn_pdrop", "bias", "class", "config", "def", "dropout_rate", "eps", "ffn", "hidden_size", "int", "layer_idx", "layer_norm_epsilon", "n_heads", "nn", "norm_1", "norm_2", "num_heads", "resid_attn_dropout", "self", "super"], "mpt/modeling_mpt.py:MptBlock.forward": ["Cache", "False", "GradientCheckpointingLayer", "ModelBlock", "None", "Tensor", "attention_mask", "attn", "attn_outputs", "attn_weights", "bool", "cache_position", "class", "def", "ffn", "forward", "hidden_states", "layer_past", "layernorm_output", "norm_1", "norm_2", "output", "output_attentions", "past_key_values", "position_bias", "resid_attn_dropout", "residual", "return", "self", "torch", "use_cache"], "mpt/modeling_mpt.py:MptModel.__init__": ["Embedding", "False", "LayerNorm", "ModelBlock", "ModelConfig", "ModelModel", "ModelPreTrainedModel", "ModuleList", "None", "__init__", "bias", "blocks", "class", "config", "def", "eps", "for", "gradient_checkpointing", "hidden_size", "i", "in", "layer_idx", "layer_norm_epsilon", "n_heads", "n_layers", "nn", "norm_f", "num_heads", "post_init", "range", "self", "super", "vocab_size", "wte"], "mpt/modeling_mpt.py:MptModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "return", "self", "wte"], "mpt/modeling_mpt.py:MptModel.build_mpt_alibi_tensor": ["ModelModel", "ModelPreTrainedModel", "None", "alibi_bias_max", "build_Model_alibi_tensor", "class", "def", "device", "num_heads", "return", "self", "sequence_length"], "mpt/modeling_mpt.py:MptModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "Tensor", "class", "def", "new_embeddings", "self", "set_input_embeddings", "torch", "wte"], "mpt/modeling_mpt.py:MptModel.forward": ["BaseModelOutputWithPastAndCrossAttentions", "Cache", "DynamicCache", "False", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Setting", "Tensor", "True", "ValueError", "You", "_", "_prepare_4d_causal_attention_mask", "alibi", "all_hidden_states", "all_self_attentions", "and", "at", "attention_mask", "attentions", "batch_size", "block", "blocks", "bool", "both", "build_Model_alibi_tensor", "cache_position", "cannot", "causal_mask", "checkpointing", "class", "config", "def", "device", "either", "elif", "else", "for", "forward", "get_seq_length", "gradient", "gradient_checkpointing", "have", "hidden_states", "if", "in", "incompatible", "input_ids", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layer_past", "logger", "max_seq_len", "norm_f", "not", "num_heads", "ones", "or", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "past_key_values_length", "position_bias", "r", "raise", "return", "return_dict", "same", "self", "seq_length", "seq_length_with_past", "shape", "specify", "the", "time", "to", "torch", "training", "tuple", "use_cache", "use_return_dict", "v", "warning_once", "with", "wte"], "mpt/modeling_mpt.py:MptForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "ModelConfig", "ModelForCausalLM", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "nn", "post_init", "self", "super", "transformer", "vocab_size"], "mpt/modeling_mpt.py:MptForCausalLM.set_output_embeddings": ["GenerationMixin", "ModelForCausalLM", "ModelPreTrainedModel", "Tensor", "class", "def", "lm_head", "new_embeddings", "self", "set_output_embeddings", "torch"], "mpt/modeling_mpt.py:MptForCausalLM.forward": ["Cache", "CausalLMOutputWithCrossAttentions", "GenerationMixin", "LongTensor", "ModelForCausalLM", "ModelPreTrainedModel", "None", "Tensor", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "not", "output", "output_attentions", "output_hidden_states", "past_key_values", "r", "return", "return_dict", "self", "slice", "slice_indices", "torch", "transformer", "transformer_outputs", "tuple", "use_cache", "use_return_dict", "vocab_size"], "mpt/modeling_mpt.py:MptForSequenceClassification.__init__": ["False", "Linear", "ModelConfig", "ModelForSequenceClassification", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "nn", "num_labels", "post_init", "score", "self", "super", "transformer"], "mpt/modeling_mpt.py:MptForSequenceClassification.set_output_embeddings": ["ModelForSequenceClassification", "ModelPreTrainedModel", "Tensor", "class", "def", "new_embeddings", "score", "self", "set_output_embeddings", "torch"], "mpt/modeling_mpt.py:MptForSequenceClassification.forward": ["BCEWithLogitsLoss", "Cache", "Cannot", "CrossEntropyLoss", "LongTensor", "MSELoss", "ModelForSequenceClassification", "ModelPreTrainedModel", "None", "Results", "SequenceClassifierOutputWithPast", "Tensor", "ValueError", "__class__", "__name__", "and", "arange", "argmax", "attention_mask", "attentions", "batch", "batch_size", "be", "bool", "class", "config", "conjunction", "def", "defined", "detect", "device", "dtype", "elif", "else", "f", "forward", "handle", "hidden_states", "if", "in", "input_ids", "inputs_embeds", "int", "int32", "is", "kwargs", "labels", "last_non_pad_token", "logger", "logits", "long", "loss", "loss_fct", "may", "multi_label_classification", "no", "non_pad_mask", "not", "num_labels", "or", "output", "output_attentions", "output_hidden_states", "pad_token_id", "padding", "past_key_values", "pooled_logits", "problem_type", "r", "raise", "regression", "return", "return_dict", "score", "self", "shape", "single_label_classification", "sizes", "squeeze", "to", "token", "token_indices", "tokens", "torch", "transformer", "transformer_outputs", "tuple", "unexpected", "use_cache", "use_return_dict", "using", "warning_once", "will", "with"], "mpt/modeling_mpt.py:MptForTokenClassification.__init__": ["Dropout", "Linear", "ModelConfig", "ModelForTokenClassification", "ModelModel", "ModelPreTrainedModel", "None", "__init__", "and", "class", "classifier", "classifier_dropout", "config", "def", "dropout", "elif", "else", "hasattr", "hidden_dropout", "hidden_size", "if", "is", "nn", "not", "num_labels", "post_init", "self", "super", "transformer"], "mpt/modeling_mpt.py:MptForTokenClassification.forward": ["Cache", "CrossEntropyLoss", "LongTensor", "ModelForTokenClassification", "ModelPreTrainedModel", "None", "Tensor", "TokenClassifierOutput", "attention_mask", "attentions", "batch_size", "bool", "class", "classifier", "config", "def", "deprecated_arguments", "device", "dropout", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "labels", "logits", "loss", "loss_fct", "not", "num_labels", "output", "output_attentions", "output_hidden_states", "past_key_values", "r", "return", "return_dict", "self", "seq_length", "shape", "to", "torch", "transformer", "transformer_outputs", "tuple", "use_cache", "use_return_dict", "view"], "mpt/modeling_mpt.py:MptForQuestionAnswering.__init__": ["Linear", "ModelForQuestionAnswering", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "config", "def", "hidden_size", "nn", "post_init", "qa_outputs", "self", "super", "transformer"], "mpt/modeling_mpt.py:MptForQuestionAnswering.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "ModelForQuestionAnswering", "ModelPreTrainedModel", "None", "QuestionAnsweringModelOutput", "and", "attention_mask", "attentions", "bool", "clamp", "class", "config", "contiguous", "def", "dim", "else", "end_logits", "end_loss", "end_positions", "forward", "hidden_states", "if", "ignore_index", "ignored_index", "input_ids", "inputs_embeds", "is", "kwargs", "len", "logits", "loss", "loss_fct", "not", "output", "output_attentions", "output_hidden_states", "outputs", "qa_outputs", "r", "return", "return_dict", "self", "sequence_output", "size", "split", "squeeze", "start_logits", "start_loss", "start_positions", "torch", "total_loss", "transformer", "tuple", "use_return_dict"], "mra/modeling_mra.py:load_cuda_kernels": ["ImportError", "Model", "Model_cuda_kernel", "Model_cuda_kernels", "community", "def", "get_kernel", "global", "if", "install", "installed", "is", "is_kernels_available", "it", "kernels", "not", "pip", "please", "raise", "with"], "mra/modeling_mra.py:sparse_max": ["Model_cuda_kernel", "Model_max", "Model_qk_prod", "None", "The", "ValueError", "a", "be", "contiguous", "def", "dim", "dimension", "dimensional", "if", "index_max", "index_vals", "indices", "int", "key_num_block", "len", "max", "max_vals", "max_vals_scatter", "must", "of", "query_num_block", "raise", "return", "second", "size", "tensor", "the", "third", "transpose", "values"], "mra/modeling_mra.py:sparse_mask": ["Model_mask", "None", "ValueError", "a", "and", "arange", "batch_idx", "batch_size", "be", "block_size", "def", "device", "dimension", "dimensional", "dtype", "have", "if", "in", "indices", "len", "long", "mask", "must", "num_block", "raise", "reshape", "return", "same", "seq_len", "shape", "size", "tensor", "th", "the", "torch", "zero"], "mra/modeling_mra.py:mm_to_sparse": ["Model_cuda_kernel", "The", "ValueError", "_", "a", "batch_size", "be", "block_size", "by", "contiguous", "def", "dense_key", "dense_query", "dim", "dimension", "dimensional", "divisible", "first", "if", "indices", "int", "key_size", "len", "mm_to_sparse", "must", "of", "query_size", "raise", "reshape", "return", "size", "tensor", "third", "transpose"], "mra/modeling_mra.py:sparse_dense_mm": ["Model_cuda_kernel", "Model_dense_mm", "Model_query", "The", "ValueError", "a", "batch_size", "be", "block_size", "by", "contiguous", "def", "dense_key", "dense_qk_prod", "dim", "dimension", "dimensional", "divisible", "equal", "first", "if", "indices", "int", "key_size", "len", "must", "of", "query_num_block", "raise", "reshape", "return", "second", "size", "tensor", "the", "third", "to", "transpose"], "mra/modeling_mra.py:transpose_indices": ["Model_indices", "def", "dim_1_block", "dim_2_block", "div", "floor", "indices", "long", "return", "rounding_mode", "torch"], "mra/modeling_mra.py:MraSampledDenseMatMul.forward": ["Function", "ModelSampledDenseMatMul", "autograd", "block_size", "class", "ctx", "def", "dense_key", "dense_query", "forward", "indices", "mm_to_sparse", "return", "save_for_backward", "sparse_qk_prod", "torch"], "mra/modeling_mra.py:MraSampledDenseMatMul.backward": ["Function", "ModelSampledDenseMatMul", "None", "autograd", "backward", "block_size", "class", "ctx", "def", "dense_key", "dense_query", "grad", "grad_key", "grad_query", "indices", "indices_T", "key_num_block", "query_num_block", "return", "saved_tensors", "size", "sparse_dense_mm", "torch", "transpose", "transpose_indices"], "mra/modeling_mra.py:MraSampledDenseMatMul.operator_call": ["Function", "ModelSampledDenseMatMul", "apply", "autograd", "block_size", "class", "def", "dense_key", "dense_query", "indices", "operator_call", "return", "torch"], "mra/modeling_mra.py:MraSparseDenseMatMul.forward": ["Function", "ModelSparseDenseMatMul", "autograd", "class", "ctx", "def", "dense_key", "forward", "indices", "query_num_block", "return", "save_for_backward", "sparse_dense_mm", "sparse_qk_prod", "sparse_query", "torch"], "mra/modeling_mra.py:MraSparseDenseMatMul.backward": ["Function", "ModelSparseDenseMatMul", "None", "autograd", "backward", "class", "ctx", "def", "dense_key", "grad", "grad_key", "grad_query", "indices", "indices_T", "key_num_block", "mm_to_sparse", "query_num_block", "return", "saved_tensors", "size", "sparse_dense_mm", "sparse_query", "torch", "transpose", "transpose_indices"], "mra/modeling_mra.py:MraSparseDenseMatMul.operator_call": ["Function", "ModelSparseDenseMatMul", "apply", "autograd", "class", "def", "dense_key", "indices", "operator_call", "query_num_block", "return", "sparse_query", "torch"], "mra/modeling_mra.py:MraReduceSum.operator_call": ["ModelReduceSum", "None", "ValueError", "_", "a", "arange", "batch_idx", "batch_size", "be", "block_size", "class", "def", "device", "dim", "dimensional", "div", "dtype", "floor", "global_idxes", "if", "index_add", "indices", "key_num_block", "len", "long", "must", "num_block", "operator_call", "output", "query_num_block", "raise", "reshape", "return", "rounding_mode", "size", "sparse_query", "sum", "temp", "tensor", "torch", "zeros"], "mra/modeling_mra.py:get_low_resolution_logit": ["Model_low_resolution_logit", "None", "True", "batch_size", "block_size", "def", "device", "dim", "dtype", "else", "float", "head_dim", "if", "is", "keepdims", "key", "key_hat", "low_resolution_logit", "low_resolution_logit_row_max", "mask", "math", "matmul", "max", "mean", "not", "num_block_per_row", "ones", "query", "query_hat", "reshape", "return", "seq_len", "size", "sqrt", "sum", "token_count", "torch", "transpose", "value", "value_hat", "values"], "mra/modeling_mra.py:get_block_idxes": ["False", "Model_block_idxes", "None", "True", "ValueError", "_", "a", "approx_mode", "approx_model", "batch_size", "def", "device", "diagonal", "diagonal_mask", "dim", "elif", "else", "f", "float", "full", "high_resolution_mask", "if", "indices", "initial_prior_diagonal_n_blocks", "initial_prior_first_n_blocks", "is", "largest", "low_resolution_logit", "min", "not", "num_blocks", "offset", "ones", "raise", "reshape", "return", "shape", "sorted", "sparse", "temp_mask", "threshold", "top_k_vals", "topk", "torch", "total_blocks_per_row", "tril", "triu", "valid", "value", "values"], "mra/modeling_mra.py:mra2_attention": ["Exception", "ModelReduceSum", "ModelSampledDenseMatMul", "ModelSparseDenseMatMul", "Model_attention", "Model_cuda_kernel", "None", "ValueError", "_", "approx_mode", "batch_size", "be", "block_size", "by", "config", "context_layer", "def", "dim", "divisible", "elif", "else", "exp", "float", "full", "get_block_idxes", "get_low_resolution_logit", "head_dim", "high_resolution_attn", "high_resolution_attn_out", "high_resolution_corr", "high_resolution_logit", "high_resolution_mask", "high_resolution_normalizer", "if", "indices", "initial_prior_diagonal_n_blocks", "initial_prior_first_n_blocks", "is", "key", "length", "log_correction", "low_resolution_attn", "low_resolution_attn_out", "low_resolution_corr", "low_resolution_logit", "low_resolution_logit_normalized", "low_resolution_logit_row_max", "low_resolution_normalizer", "mask", "math", "matmul", "max_vals", "max_vals_scatter", "meta_batch", "must", "no_grad", "not", "num_block_per_row", "num_blocks", "num_head", "operator_call", "or", "query", "raise", "repeat", "requires_grad_", "reshape", "return", "seq_len", "sequence", "size", "sparse", "sparse_mask", "sparse_max", "sqrt", "sum", "the", "token_count", "torch", "value", "value_hat", "with", "zeros_like"], "mra/modeling_mra.py:MraEmbeddings.__init__": ["Dropout", "Embedding", "False", "LayerNorm", "ModelEmbeddings", "Module", "__init__", "arange", "class", "config", "def", "device", "dropout", "dtype", "eps", "expand", "hidden_dropout_prob", "hidden_size", "layer_norm_eps", "long", "max_position_embeddings", "nn", "pad_token_id", "padding_idx", "persistent", "position_embeddings", "position_ids", "register_buffer", "self", "size", "super", "token_type_embeddings", "token_type_ids", "torch", "type_vocab_size", "vocab_size", "word_embeddings", "zeros"], "mra/modeling_mra.py:MraEmbeddings.forward": ["LayerNorm", "ModelEmbeddings", "Module", "None", "buffered_token_type_ids", "buffered_token_type_ids_expanded", "class", "def", "device", "dropout", "dtype", "else", "embeddings", "expand", "forward", "hasattr", "if", "input_ids", "input_shape", "inputs_embeds", "is", "long", "nn", "not", "position_embeddings", "position_ids", "return", "self", "seq_length", "size", "token_type_embeddings", "token_type_ids", "torch", "word_embeddings", "zeros"], "mra/modeling_mra.py:MraSelfAttention.__init__": ["Could", "Dropout", "Exception", "Linear", "ModelSelfAttention", "Model_cuda_kernel", "Module", "None", "The", "ValueError", "__init__", "a", "all_head_size", "and", "approx_mode", "as", "attention", "attention_head_size", "attention_probs_dropout_prob", "block_per_row", "class", "config", "custom", "def", "deformable", "dropout", "e", "embedding_size", "except", "f", "for", "hasattr", "heads", "hidden", "hidden_size", "if", "initial_prior_diagonal_n_blocks", "initial_prior_first_n_blocks", "int", "is", "is_cuda_platform", "is_ninja_available", "is_torch_cuda_available", "kernel", "kernel_loaded", "key", "load", "load_cuda_kernels", "logger", "max_position_embeddings", "min", "multi", "multiple", "nn", "not", "num_attention_heads", "num_block", "number", "of", "query", "raise", "scale", "self", "size", "super", "the", "try", "value", "warning"], "mra/modeling_mra.py:MraSelfAttention.forward": ["Model2_attention", "ModelSelfAttention", "Module", "None", "_", "all_head_size", "approx_mode", "attention_head_size", "attention_mask", "batch_size", "cat", "class", "context_layer", "contiguous", "def", "device", "dim", "float", "forward", "gpu_warp_size", "hidden_states", "if", "initial_prior_diagonal_n_blocks", "initial_prior_first_n_blocks", "int", "key", "key_layer", "new_context_layer_shape", "nn", "num_attention_heads", "num_block", "outputs", "pad_size", "permute", "query", "query_layer", "repeat", "reshape", "return", "self", "seq_len", "shape", "size", "squeeze", "torch", "transpose", "value", "value_layer", "view", "zeros"], "mra/modeling_mra.py:MraSelfOutput.__init__": ["Dropout", "LayerNorm", "Linear", "ModelSelfOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "eps", "hidden_dropout_prob", "hidden_size", "layer_norm_eps", "nn", "self", "super"], "mra/modeling_mra.py:MraSelfOutput.forward": ["LayerNorm", "ModelSelfOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "mra/modeling_mra.py:MraAttention.__init__": ["ModelAttention", "ModelSelfAttention", "ModelSelfOutput", "Module", "__init__", "class", "config", "def", "nn", "output", "self", "super"], "mra/modeling_mra.py:MraAttention.forward": ["ModelAttention", "Module", "None", "attention_mask", "attention_output", "class", "def", "forward", "hidden_states", "nn", "output", "outputs", "return", "self", "self_outputs"], "mra/modeling_mra.py:MraIntermediate.__init__": ["ACT2FN", "Linear", "ModelIntermediate", "Module", "__init__", "class", "config", "def", "dense", "else", "hidden_act", "hidden_size", "if", "intermediate_act_fn", "intermediate_size", "isinstance", "nn", "self", "str", "super"], "mra/modeling_mra.py:MraIntermediate.forward": ["ModelIntermediate", "Module", "Tensor", "class", "def", "dense", "forward", "hidden_states", "intermediate_act_fn", "nn", "return", "self", "torch"], "mra/modeling_mra.py:MraOutput.__init__": ["Dropout", "LayerNorm", "Linear", "ModelOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "eps", "hidden_dropout_prob", "hidden_size", "intermediate_size", "layer_norm_eps", "nn", "self", "super"], "mra/modeling_mra.py:MraOutput.forward": ["LayerNorm", "ModelOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "mra/modeling_mra.py:MraLayer.__init__": ["GradientCheckpointingLayer", "ModelAttention", "ModelIntermediate", "ModelLayer", "ModelOutput", "__init__", "add_cross_attention", "attention", "chunk_size_feed_forward", "class", "config", "def", "intermediate", "output", "self", "seq_len_dim", "super"], "mra/modeling_mra.py:MraLayer.forward": ["GradientCheckpointingLayer", "ModelLayer", "None", "apply_chunking_to_forward", "attention", "attention_mask", "attention_output", "chunk_size_feed_forward", "class", "def", "feed_forward_chunk", "forward", "hidden_states", "layer_output", "outputs", "return", "self", "self_attention_outputs", "seq_len_dim"], "mra/modeling_mra.py:MraLayer.feed_forward_chunk": ["GradientCheckpointingLayer", "ModelLayer", "attention_output", "class", "def", "feed_forward_chunk", "intermediate", "intermediate_output", "layer_output", "output", "return", "self"], "mra/modeling_mra.py:MraEncoder.__init__": ["False", "ModelEncoder", "ModelLayer", "Module", "ModuleList", "_", "__init__", "class", "config", "def", "for", "gradient_checkpointing", "in", "layer", "nn", "num_hidden_layers", "range", "self", "super"], "mra/modeling_mra.py:MraEncoder.forward": ["BaseModelOutputWithCrossAttentions", "False", "ModelEncoder", "Module", "None", "True", "all_hidden_states", "attention_mask", "class", "def", "else", "enumerate", "for", "forward", "hidden_states", "i", "if", "in", "is", "last_hidden_state", "layer", "layer_module", "layer_outputs", "nn", "not", "output_hidden_states", "return", "return_dict", "self", "tuple", "v"], "mra/modeling_mra.py:MraPredictionHeadTransform.__init__": ["ACT2FN", "LayerNorm", "Linear", "ModelPredictionHeadTransform", "Module", "__init__", "class", "config", "def", "dense", "else", "eps", "hidden_act", "hidden_size", "if", "isinstance", "layer_norm_eps", "nn", "self", "str", "super", "transform_act_fn"], "mra/modeling_mra.py:MraPredictionHeadTransform.forward": ["LayerNorm", "ModelPredictionHeadTransform", "Module", "Tensor", "class", "def", "dense", "forward", "hidden_states", "nn", "return", "self", "torch", "transform_act_fn"], "mra/modeling_mra.py:MraLMPredictionHead.__init__": ["Linear", "ModelLMPredictionHead", "ModelPredictionHeadTransform", "Module", "Parameter", "True", "__init__", "bias", "class", "config", "decoder", "def", "hidden_size", "nn", "self", "super", "torch", "transform", "vocab_size", "zeros"], "mra/modeling_mra.py:MraLMPredictionHead.forward": ["ModelLMPredictionHead", "Module", "class", "decoder", "def", "forward", "hidden_states", "nn", "return", "self", "transform"], "mra/modeling_mra.py:MraOnlyMLMHead.__init__": ["ModelLMPredictionHead", "ModelOnlyMLMHead", "Module", "__init__", "class", "config", "def", "nn", "predictions", "self", "super"], "mra/modeling_mra.py:MraOnlyMLMHead.forward": ["ModelOnlyMLMHead", "Module", "Tensor", "class", "def", "forward", "nn", "prediction_scores", "predictions", "return", "self", "sequence_output", "torch"], "mra/modeling_mra.py:MraPreTrainedModel._init_weights": ["ModelEmbeddings", "ModelLMPredictionHead", "ModelPreTrainedModel", "Module", "PreTrainedModel", "_init_weights", "arange", "bias", "class", "copy_", "def", "elif", "expand", "if", "init", "isinstance", "module", "nn", "position_ids", "self", "shape", "super", "token_type_ids", "torch", "zeros_"], "mra/modeling_mra.py:MraModel.__init__": ["ModelEmbeddings", "ModelEncoder", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "config", "def", "embeddings", "encoder", "post_init", "self", "super"], "mra/modeling_mra.py:MraModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "return", "self", "word_embeddings"], "mra/modeling_mra.py:MraModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "self", "set_input_embeddings", "value", "word_embeddings"], "mra/modeling_mra.py:MraModel.forward": ["BaseModelOutputWithCrossAttentions", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "ValueError", "You", "and", "at", "attention_mask", "attentions", "batch_size", "bool", "both", "buffered_token_type_ids", "buffered_token_type_ids_expanded", "cannot", "class", "config", "cross_attentions", "def", "device", "dtype", "either", "elif", "else", "embedding_output", "embeddings", "encoder", "encoder_outputs", "expand", "extended_attention_mask", "forward", "get_extended_attention_mask", "hasattr", "have", "hidden_states", "if", "input_ids", "input_shape", "inputs_embeds", "is", "kwargs", "last_hidden_state", "long", "not", "ones", "or", "output_hidden_states", "position_ids", "raise", "return", "return_dict", "same", "self", "seq_length", "sequence_output", "size", "specify", "the", "time", "to", "token_type_ids", "torch", "tuple", "use_return_dict", "warn_if_padding_and_no_attention_mask", "zeros"], "mra/modeling_mra.py:MraForMaskedLM.__init__": ["Model", "ModelForMaskedLM", "ModelModel", "ModelOnlyMLMHead", "ModelPreTrainedModel", "__init__", "class", "cls", "config", "def", "post_init", "self", "super"], "mra/modeling_mra.py:MraForMaskedLM.get_output_embeddings": ["ModelForMaskedLM", "ModelPreTrainedModel", "class", "cls", "decoder", "def", "get_output_embeddings", "predictions", "return", "self"], "mra/modeling_mra.py:MraForMaskedLM.set_output_embeddings": ["ModelForMaskedLM", "ModelPreTrainedModel", "bias", "class", "cls", "decoder", "def", "new_embeddings", "predictions", "self", "set_output_embeddings"], "mra/modeling_mra.py:MraForMaskedLM.forward": ["CrossEntropyLoss", "MaskedLMOutput", "Model", "ModelForMaskedLM", "ModelPreTrainedModel", "None", "Tensor", "attention_mask", "attentions", "bool", "class", "cls", "config", "def", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "masked_lm_loss", "not", "output", "output_hidden_states", "outputs", "position_ids", "prediction_scores", "r", "return", "return_dict", "self", "sequence_output", "token_type_ids", "torch", "tuple", "use_return_dict", "view", "vocab_size"], "mra/modeling_mra.py:MraClassificationHead.__init__": ["Dropout", "Linear", "ModelClassificationHead", "Module", "__init__", "class", "config", "def", "dense", "dropout", "hidden_dropout_prob", "hidden_size", "nn", "num_labels", "out_proj", "self", "super"], "mra/modeling_mra.py:MraClassificationHead.forward": ["ACT2FN", "ModelClassificationHead", "Module", "class", "config", "def", "dense", "dropout", "features", "forward", "hidden_act", "kwargs", "nn", "out_proj", "return", "self", "x"], "mra/modeling_mra.py:MraForSequenceClassification.__init__": ["Model", "ModelClassificationHead", "ModelForSequenceClassification", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "classifier", "config", "def", "num_labels", "post_init", "self", "super"], "mra/modeling_mra.py:MraForSequenceClassification.forward": ["BCEWithLogitsLoss", "CrossEntropyLoss", "MSELoss", "Model", "ModelForSequenceClassification", "ModelPreTrainedModel", "None", "SequenceClassifierOutput", "Tensor", "and", "attention_mask", "attentions", "bool", "class", "classifier", "config", "def", "dtype", "elif", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "kwargs", "labels", "logits", "long", "loss", "loss_fct", "multi_label_classification", "not", "num_labels", "or", "output", "output_hidden_states", "outputs", "position_ids", "problem_type", "r", "regression", "return", "return_dict", "self", "sequence_output", "single_label_classification", "squeeze", "token_type_ids", "torch", "tuple", "use_return_dict", "view"], "mra/modeling_mra.py:MraForMultipleChoice.__init__": ["Linear", "Model", "ModelForMultipleChoice", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "classifier", "config", "def", "hidden_size", "nn", "post_init", "pre_classifier", "self", "super"], "mra/modeling_mra.py:MraForMultipleChoice.forward": ["CrossEntropyLoss", "Model", "ModelForMultipleChoice", "ModelPreTrainedModel", "MultipleChoiceModelOutput", "None", "ReLU", "Tensor", "attention_mask", "attentions", "bool", "class", "classifier", "config", "def", "else", "forward", "hidden_state", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "nn", "not", "num_choices", "output", "output_hidden_states", "outputs", "pooled_output", "position_ids", "pre_classifier", "r", "reshaped_logits", "return", "return_dict", "self", "shape", "size", "token_type_ids", "torch", "tuple", "use_return_dict", "view"], "mra/modeling_mra.py:MraForTokenClassification.__init__": ["Dropout", "Linear", "Model", "ModelForTokenClassification", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "classifier", "config", "def", "dropout", "hidden_dropout_prob", "hidden_size", "nn", "num_labels", "post_init", "self", "super"], "mra/modeling_mra.py:MraForTokenClassification.forward": ["CrossEntropyLoss", "Model", "ModelForTokenClassification", "ModelPreTrainedModel", "None", "Tensor", "TokenClassifierOutput", "active_labels", "active_logits", "active_loss", "attention_mask", "attentions", "bool", "class", "classifier", "config", "def", "dropout", "else", "forward", "hidden_states", "if", "ignore_index", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "not", "num_labels", "output", "output_hidden_states", "outputs", "position_ids", "r", "return", "return_dict", "self", "sequence_output", "tensor", "token_type_ids", "torch", "tuple", "type_as", "use_return_dict", "view", "where"], "mra/modeling_mra.py:MraForQuestionAnswering.__init__": ["Linear", "Model", "ModelForQuestionAnswering", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "config", "def", "hidden_size", "nn", "num_labels", "post_init", "qa_outputs", "self", "super"], "mra/modeling_mra.py:MraForQuestionAnswering.forward": ["CrossEntropyLoss", "Model", "ModelForQuestionAnswering", "ModelPreTrainedModel", "None", "QuestionAnsweringModelOutput", "Tensor", "and", "attention_mask", "attentions", "bool", "clamp", "class", "config", "def", "dim", "else", "end_logits", "end_loss", "end_positions", "forward", "hidden_states", "if", "ignore_index", "ignored_index", "input_ids", "inputs_embeds", "is", "kwargs", "len", "logits", "loss", "loss_fct", "not", "output", "output_hidden_states", "outputs", "position_ids", "qa_outputs", "return", "return_dict", "self", "sequence_output", "size", "split", "squeeze", "start_logits", "start_loss", "start_positions", "token_type_ids", "torch", "total_loss", "tuple", "use_return_dict"], "mt5/modeling_mt5.py:MT5LayerNorm.__init__": ["Model", "Module", "Parameter", "__init__", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "mt5/modeling_mt5.py:MT5LayerNorm.forward": ["Model", "Module", "True", "bfloat16", "class", "def", "dtype", "float16", "float32", "forward", "hidden_states", "if", "in", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "mt5/modeling_mt5.py:MT5DenseActDense.__init__": ["ACT2FN", "Dropout", "False", "Linear", "Model", "ModelConfig", "Module", "__init__", "act", "bias", "class", "config", "d_ff", "d_model", "def", "dense_act_fn", "dropout", "dropout_rate", "nn", "self", "super", "wi", "wo"], "mt5/modeling_mt5.py:MT5DenseActDense.forward": ["Model", "Module", "Tensor", "act", "and", "class", "def", "dropout", "dtype", "forward", "hidden_states", "if", "int8", "isinstance", "nn", "return", "self", "to", "torch", "weight", "wi", "wo"], "mt5/modeling_mt5.py:MT5DenseGatedActDense.__init__": ["ACT2FN", "Dropout", "False", "Linear", "Model", "ModelConfig", "Module", "__init__", "act", "bias", "class", "config", "d_ff", "d_model", "def", "dense_act_fn", "dropout", "dropout_rate", "nn", "self", "super", "wi_0", "wi_1", "wo"], "mt5/modeling_mt5.py:MT5DenseGatedActDense.forward": ["Model", "Module", "Tensor", "act", "and", "class", "def", "dropout", "dtype", "forward", "hidden_gelu", "hidden_linear", "hidden_states", "if", "int8", "isinstance", "nn", "return", "self", "to", "torch", "weight", "wi_0", "wi_1", "wo"], "mt5/modeling_mt5.py:MT5LayerFF.__init__": ["DenseReluDense", "Dropout", "Model", "ModelConfig", "ModelDenseActDense", "ModelDenseGatedActDense", "ModelLayerNorm", "Module", "__init__", "class", "config", "d_model", "def", "dropout", "dropout_rate", "else", "eps", "if", "is_gated_act", "layer_norm", "layer_norm_epsilon", "nn", "self", "super"], "mt5/modeling_mt5.py:MT5LayerFF.forward": ["DenseReluDense", "Model", "Module", "class", "def", "dropout", "forward", "forwarded_states", "hidden_states", "layer_norm", "nn", "return", "self"], "mt5/modeling_mt5.py:MT5Attention.__init__": ["Embedding", "False", "Instantiating", "Linear", "Model", "ModelConfig", "Module", "None", "Please", "__class__", "__init__", "__name__", "a", "and", "bias", "caching", "call", "class", "config", "creating", "d_kv", "d_model", "decoder", "def", "dropout", "dropout_rate", "during", "errors", "f", "forward", "gradient_checkpointing", "has_relative_attention_bias", "if", "inner_dim", "int", "is", "is_decoder", "k", "key_value_proj_dim", "layer_idx", "logger", "make", "n_heads", "nn", "not", "num_heads", "o", "passing", "provide", "q", "recommended", "relative_attention_bias", "relative_attention_max_distance", "relative_attention_num_buckets", "self", "super", "sure", "the", "this", "to", "used", "v", "warning_once", "when", "will", "without"], "mt5/modeling_mt5.py:MT5Attention._relative_position_bucket": ["Model", "Module", "True", "_relative_position_bucket", "abs", "bidirectional", "class", "def", "else", "float", "full_like", "if", "is_small", "log", "long", "math", "max_distance", "max_exact", "min", "nn", "num_buckets", "relative_buckets", "relative_position", "relative_position_if_large", "return", "to", "torch", "where", "zeros_like"], "mt5/modeling_mt5.py:MT5Attention.compute_bias": ["Model", "Module", "None", "_relative_position_bucket", "arange", "bidirectional", "cache_position", "class", "compute_bias", "context_position", "def", "device", "dtype", "else", "if", "is", "is_decoder", "key_length", "long", "max_distance", "memory_position", "nn", "not", "num_buckets", "permute", "query_length", "relative_attention_bias", "relative_attention_max_distance", "relative_attention_num_buckets", "relative_position", "relative_position_bucket", "return", "self", "to", "torch", "unsqueeze", "values", "weight"], "mt5/modeling_mt5.py:MT5Attention.forward": ["EncoderDecoderCache", "False", "Model", "Module", "None", "True", "and", "attn_output", "attn_weights", "batch_size", "cache_position", "causal_mask", "class", "compute_bias", "contiguous", "cross_attention_cache", "curr_past_key_values", "current_states", "def", "device", "dim", "dropout", "dtype", "else", "float", "forward", "functional", "get", "gradient_checkpointing", "has_relative_attention_bias", "hidden_states", "if", "inner_dim", "is", "is_cross_attention", "is_updated", "isinstance", "k", "key_length", "key_states", "key_value_proj_dim", "key_value_states", "keys", "layer_idx", "layers", "mask", "matmul", "n_heads", "nn", "not", "o", "output_attentions", "outputs", "p", "past_key_values", "position_bias", "position_bias_masked", "q", "query_length", "query_states", "real_seq_length", "requires_grad", "return", "scores", "self", "self_attention_cache", "seq_length", "shape", "softmax", "torch", "training", "transpose", "type_as", "update", "use_cache", "v", "value_states", "values", "view", "zeros"], "mt5/modeling_mt5.py:MT5LayerSelfAttention.__init__": ["Dropout", "False", "Model", "ModelAttention", "ModelLayerNorm", "Module", "None", "SelfAttention", "__init__", "class", "config", "d_model", "def", "dropout", "dropout_rate", "eps", "has_relative_attention_bias", "int", "layer_idx", "layer_norm", "layer_norm_epsilon", "nn", "self", "super"], "mt5/modeling_mt5.py:MT5LayerSelfAttention.forward": ["False", "Model", "Module", "None", "SelfAttention", "attention_mask", "attention_output", "cache_position", "class", "def", "dropout", "forward", "hidden_states", "layer_norm", "mask", "nn", "normed_hidden_states", "output_attentions", "outputs", "past_key_values", "position_bias", "return", "self", "use_cache"], "mt5/modeling_mt5.py:MT5LayerCrossAttention.__init__": ["Dropout", "EncDecAttention", "False", "Model", "ModelAttention", "ModelLayerNorm", "Module", "None", "__init__", "class", "config", "d_model", "def", "dropout", "dropout_rate", "eps", "has_relative_attention_bias", "int", "layer_idx", "layer_norm", "layer_norm_epsilon", "nn", "self", "super"], "mt5/modeling_mt5.py:MT5LayerCrossAttention.forward": ["EncDecAttention", "False", "Model", "Module", "None", "attention_mask", "attention_output", "cache_position", "class", "def", "dropout", "forward", "hidden_states", "key_value_states", "layer_norm", "layer_output", "mask", "nn", "normed_hidden_states", "output_attentions", "outputs", "past_key_values", "position_bias", "query_length", "return", "self", "use_cache"], "mt5/modeling_mt5.py:MT5Block.__init__": ["False", "GradientCheckpointingLayer", "Model", "ModelLayerCrossAttention", "ModelLayerFF", "ModelLayerSelfAttention", "ModuleList", "None", "__init__", "append", "class", "config", "def", "has_relative_attention_bias", "if", "int", "is_decoder", "layer", "layer_idx", "nn", "self", "super"], "mt5/modeling_mt5.py:MT5Block.forward": ["False", "GradientCheckpointingLayer", "Model", "None", "True", "and", "any", "attention_mask", "attention_outputs", "cache_position", "clamp", "clamp_value", "class", "cross_attention_outputs", "def", "do_cross_attention", "dtype", "encoder_attention_mask", "encoder_decoder_position_bias", "encoder_hidden_states", "finfo", "float16", "forward", "hidden_states", "if", "is", "is_decoder", "isinf", "key_value_states", "layer", "max", "min", "not", "output_attentions", "outputs", "past_key_values", "position_bias", "query_length", "return", "return_dict", "self", "self_attention_outputs", "torch", "use_cache", "where"], "mt5/modeling_mt5.py:MT5ClassificationHead.__init__": ["Dropout", "Linear", "Model", "ModelConfig", "Module", "__init__", "class", "classifier_dropout", "config", "d_model", "def", "dense", "dropout", "nn", "num_labels", "out_proj", "p", "self", "super"], "mt5/modeling_mt5.py:MT5ClassificationHead.forward": ["Model", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "nn", "out_proj", "return", "self", "tanh", "torch"], "mt5/modeling_mt5.py:MT5PreTrainedModel.dummy_inputs": ["DUMMY_INPUTS", "DUMMY_MASK", "Model", "PreTrainedModel", "class", "decoder_attention_mask", "decoder_input_ids", "def", "dummy_inputs", "input_ids", "input_mask", "return", "self", "tensor", "torch"], "mt5/modeling_mt5.py:MT5PreTrainedModel._init_weights": ["Model", "ModelAttention", "ModelClassificationHead", "ModelDenseActDense", "ModelDenseGatedActDense", "ModelEncoderModel", "ModelForConditionalGeneration", "ModelForQuestionAnswering", "ModelForTokenClassification", "ModelLayerNorm", "ModelModel", "None", "PreTrainedModel", "_init_weights", "and", "bias", "class", "classifier", "config", "constant_", "d_ff", "d_kv", "d_model", "def", "dense", "elif", "factor", "has_relative_attention_bias", "hasattr", "if", "init", "initializer_factor", "is", "isinstance", "k", "key_value_proj_dim", "lm_head", "mean", "module", "n_heads", "normal_", "not", "num_heads", "o", "out_proj", "q", "qa_outputs", "relative_attention_bias", "self", "shared", "std", "tie_word_embeddings", "v", "weight", "wi", "wi_0", "wi_1", "wo", "zeros_"], "mt5/modeling_mt5.py:MT5PreTrainedModel._shift_right": ["In", "Model", "None", "PreTrainedModel", "See", "ValueError", "_shift_right", "be", "class", "clone", "config", "decoder_start_token_id", "def", "defined", "docs", "for", "has", "if", "information", "input_ids", "is", "it", "masked_fill_", "model", "more", "new_zeros", "pad_token_id", "raise", "return", "self", "set", "shape", "shifted_input_ids", "the", "to", "usually"], "mt5/modeling_mt5.py:MT5Stack.__init__": ["Dropout", "Embedding", "False", "Model", "ModelBlock", "ModelLayerNorm", "ModelPreTrainedModel", "ModuleList", "__init__", "block", "bool", "class", "config", "d_model", "def", "dropout", "dropout_rate", "embed_tokens", "eps", "final_layer_norm", "for", "gradient_checkpointing", "has_relative_attention_bias", "i", "in", "is_decoder", "layer_idx", "layer_norm_epsilon", "nn", "num_layers", "post_init", "range", "self", "super", "vocab_size"], "mt5/modeling_mt5.py:MT5Stack.set_input_embeddings": ["Model", "ModelPreTrainedModel", "class", "def", "embed_tokens", "new_embeddings", "self", "set_input_embeddings"], "mt5/modeling_mt5.py:MT5Stack.forward": ["BaseModelOutputWithPastAndCrossAttentions", "DynamicCache", "EncoderDecoderCache", "False", "Model", "ModelPreTrainedModel", "None", "Setting", "True", "ValueError", "You", "a", "all_attentions", "all_cross_attentions", "all_hidden_states", "and", "arange", "as", "at", "attention_mask", "attentions", "batch_size", "be", "block", "both", "cache_position", "can", "cannot", "checkpointing", "class", "config", "create_bidirectional_mask", "create_causal_mask", "cross_attentions", "decoder", "decoder_", "def", "device", "dropout", "either", "elif", "else", "embed_tokens", "embeddings", "encoder_attention_mask", "encoder_decoder_position_bias", "encoder_extended_attention_mask", "encoder_hidden_states", "err_msg_prefix", "f", "final_layer_norm", "for", "forward", "get_seq_length", "gradient", "gradient_checkpointing", "have", "hidden_states", "if", "in", "incompatible", "initialize", "input_embeds", "input_ids", "input_shape", "inputs_embeds", "is", "is_decoder", "is_encoder_decoder", "isinstance", "kwargs", "last_hidden_state", "layer_module", "layer_outputs", "logger", "model", "not", "only", "or", "output_attentions", "output_hidden_states", "past_key_values", "past_key_values_length", "position_bias", "raise", "return", "return_dict", "same", "self", "self_attention_cache", "seq_length", "set", "size", "specify", "the", "time", "to", "token", "torch", "training", "tuple", "use_cache", "use_return_dict", "used", "v", "valid", "view", "warning_once", "with"], "mt5/modeling_mt5.py:MT5Model.__init__": ["Embedding", "False", "Model", "ModelConfig", "ModelPreTrainedModel", "ModelStack", "True", "__init__", "class", "config", "copy", "d_model", "decoder", "decoder_config", "deepcopy", "def", "encoder", "encoder_config", "is_decoder", "nn", "num_decoder_layers", "num_layers", "post_init", "self", "shared", "super", "use_cache", "vocab_size"], "mt5/modeling_mt5.py:MT5Model.get_input_embeddings": ["Model", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "return", "self", "shared"], "mt5/modeling_mt5.py:MT5Model.set_input_embeddings": ["Model", "ModelPreTrainedModel", "class", "decoder", "def", "encoder", "new_embeddings", "self", "set_input_embeddings", "shared"], "mt5/modeling_mt5.py:MT5Model.forward": ["BaseModelOutput", "BoolTensor", "Cache", "FloatTensor", "LongTensor", "Model", "ModelPreTrainedModel", "None", "Seq2SeqModelOutput", "Tensor", "and", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "cross_attentions", "decoder", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_input_ids", "decoder_inputs_embeds", "decoder_outputs", "def", "elif", "else", "encoder", "encoder_attention_mask", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "isinstance", "kwargs", "last_hidden_state", "len", "not", "output_attentions", "output_hidden_states", "past_key_values", "r", "return", "return_dict", "self", "torch", "tuple", "use_cache", "use_return_dict"], "mt5/modeling_mt5.py:MT5ForConditionalGeneration.__init__": ["Embedding", "False", "GenerationMixin", "Linear", "Model", "ModelConfig", "ModelPreTrainedModel", "ModelStack", "True", "__init__", "bias", "class", "config", "copy", "d_model", "decoder", "decoder_config", "deepcopy", "def", "encoder", "encoder_config", "is_decoder", "lm_head", "model_dim", "nn", "num_decoder_layers", "num_layers", "post_init", "self", "shared", "super", "use_cache", "vocab_size"], "mt5/modeling_mt5.py:MT5ForConditionalGeneration.get_input_embeddings": ["GenerationMixin", "Model", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "return", "self", "shared"], "mt5/modeling_mt5.py:MT5ForConditionalGeneration.set_input_embeddings": ["GenerationMixin", "Model", "ModelPreTrainedModel", "class", "decoder", "def", "encoder", "new_embeddings", "self", "set_input_embeddings", "shared"], "mt5/modeling_mt5.py:MT5ForConditionalGeneration.forward": ["BaseModelOutput", "BoolTensor", "Cache", "CrossEntropyLoss", "FloatTensor", "GenerationMixin", "LongTensor", "Model", "ModelPreTrainedModel", "None", "Seq2SeqLMOutput", "Tensor", "_shift_right", "and", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "cross_attentions", "decoder", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_input_ids", "decoder_inputs_embeds", "decoder_outputs", "def", "device", "elif", "else", "encoder", "encoder_attention_mask", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "forward", "hidden_states", "if", "ignore_index", "input_ids", "inputs_embeds", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "len", "lm_head", "lm_logits", "logits", "loss", "loss_fct", "not", "output", "output_attentions", "output_hidden_states", "past_key_values", "r", "return", "return_dict", "self", "sequence_output", "size", "to", "torch", "tuple", "use_cache", "use_return_dict", "view"], "mt5/modeling_mt5.py:MT5ForConditionalGeneration.prepare_decoder_input_ids_from_labels": ["GenerationMixin", "Model", "ModelPreTrainedModel", "Tensor", "_shift_right", "class", "def", "labels", "prepare_decoder_input_ids_from_labels", "return", "self", "torch"], "mt5/modeling_mt5.py:MT5EncoderModel.__init__": ["Embedding", "False", "Model", "ModelConfig", "ModelPreTrainedModel", "ModelStack", "__init__", "class", "config", "d_model", "def", "encoder", "encoder_config", "is_encoder_decoder", "nn", "post_init", "self", "shared", "super", "use_cache", "vocab_size"], "mt5/modeling_mt5.py:MT5EncoderModel.get_input_embeddings": ["Model", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "return", "self", "shared"], "mt5/modeling_mt5.py:MT5EncoderModel.set_input_embeddings": ["Model", "ModelPreTrainedModel", "class", "def", "encoder", "new_embeddings", "self", "set_input_embeddings", "shared"], "mt5/modeling_mt5.py:MT5EncoderModel.forward": ["BaseModelOutput", "FloatTensor", "LongTensor", "Model", "ModelPreTrainedModel", "None", "attention_mask", "bool", "class", "config", "def", "else", "encoder", "encoder_outputs", "forward", "if", "input_ids", "inputs_embeds", "is", "kwargs", "not", "output_attentions", "output_hidden_states", "r", "return", "return_dict", "self", "torch", "tuple", "use_return_dict"], "mt5/modeling_mt5.py:MT5ForSequenceClassification.__init__": ["Model", "ModelClassificationHead", "ModelConfig", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "classification_head", "config", "def", "post_init", "self", "super", "transformer"], "mt5/modeling_mt5.py:MT5ForSequenceClassification.forward": ["All", "BCEWithLogitsLoss", "CrossEntropyLoss", "False", "FloatTensor", "If", "LongTensor", "MSELoss", "Model", "ModelPreTrainedModel", "None", "NotImplementedError", "Passing", "Please", "Seq2SeqSequenceClassifierOutput", "Tensor", "ValueError", "_", "__class__", "__name__", "_shift_right", "and", "are", "attention_mask", "batch_size", "be", "bool", "cannot", "class", "classification_head", "config", "cross_attentions", "currently", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_input_ids", "decoder_inputs_embeds", "def", "device", "dtype", "either", "elif", "else", "embeddings", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "eos", "eos_mask", "eos_token_id", "eq", "examples", "f", "for", "forward", "have", "hidden_size", "if", "input", "input_ids", "inputs_embeds", "int", "is", "kwargs", "labels", "len", "list", "logits", "long", "loss", "loss_fct", "multi_label_classification", "must", "no", "not", "num_labels", "number", "of", "or", "output", "output_attentions", "output_hidden_states", "outputs", "pass", "passed", "past_key_values", "problem_type", "r", "raise", "regression", "return", "return_dict", "same", "self", "sentence_representation", "sequence_output", "shape", "single_label_classification", "squeeze", "sum", "supported", "the", "to", "tokens", "torch", "transformer", "tuple", "unique_consecutive", "use_cache", "use_return_dict", "view"], "mt5/modeling_mt5.py:MT5ForTokenClassification.__init__": ["Dropout", "Linear", "Model", "ModelConfig", "ModelEncoderModel", "ModelPreTrainedModel", "__init__", "class", "classifier", "classifier_dropout", "config", "def", "dropout", "hidden_size", "nn", "num_labels", "post_init", "self", "super", "transformer"], "mt5/modeling_mt5.py:MT5ForTokenClassification.forward": ["CrossEntropyLoss", "Model", "ModelPreTrainedModel", "None", "Tensor", "TokenClassifierOutput", "attention_mask", "attentions", "bool", "class", "classifier", "config", "def", "dropout", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "not", "num_labels", "output", "output_attentions", "output_hidden_states", "outputs", "r", "return", "return_dict", "self", "torch", "transformer", "tuple", "use_return_dict", "view"], "mt5/modeling_mt5.py:MT5ForQuestionAnswering.__init__": ["Embedding", "False", "Linear", "Model", "ModelConfig", "ModelPreTrainedModel", "ModelStack", "True", "__init__", "class", "config", "copy", "d_model", "decoder", "decoder_config", "deepcopy", "def", "encoder", "encoder_config", "hidden_size", "is_decoder", "model_dim", "nn", "num_decoder_layers", "num_labels", "num_layers", "post_init", "qa_outputs", "self", "shared", "super", "use_cache", "vocab_size"], "mt5/modeling_mt5.py:MT5ForQuestionAnswering.get_input_embeddings": ["Model", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "return", "self", "shared"], "mt5/modeling_mt5.py:MT5ForQuestionAnswering.set_input_embeddings": ["Model", "ModelPreTrainedModel", "class", "decoder", "def", "encoder", "new_embeddings", "self", "set_input_embeddings", "shared"], "mt5/modeling_mt5.py:MT5ForQuestionAnswering.forward": ["BaseModelOutput", "BoolTensor", "CrossEntropyLoss", "False", "FloatTensor", "If", "LongTensor", "Model", "ModelPreTrainedModel", "None", "Please", "Seq2SeqQuestionAnsweringModelOutput", "Tensor", "ValueError", "_shift_right", "and", "are", "attention_mask", "attentions", "be", "bool", "cannot", "clamp", "class", "config", "contiguous", "cross_attentions", "decoder", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_input_ids", "decoder_inputs_embeds", "decoder_outputs", "def", "device", "dim", "either", "elif", "else", "encoder", "encoder_attention_mask", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "end_logits", "end_loss", "end_positions", "forward", "hidden_states", "if", "ignore_index", "ignored_index", "input_ids", "inputs_embeds", "is", "isinstance", "kwargs", "last_hidden_state", "len", "logits", "loss", "loss_fct", "no", "not", "or", "output", "output_attentions", "output_hidden_states", "pass", "passed", "past_key_values", "qa_outputs", "r", "raise", "return", "return_dict", "self", "sequence_output", "size", "split", "squeeze", "start_logits", "start_loss", "start_positions", "to", "torch", "total_loss", "tuple", "use_cache", "use_return_dict"], "musicgen/modeling_musicgen.py:shift_tokens_right": ["Make", "Model_tokens_right", "Modeled_input_ids", "None", "Tensor", "ValueError", "attribute", "clone", "configuration", "decoder_start_token_id", "def", "if", "input_ids", "int", "is", "masked_fill_", "model", "new_zeros", "of", "pad_token_id", "raise", "return", "s", "set", "shape", "sure", "the", "to", "torch", "transpose"], "musicgen/modeling_musicgen.py:MusicgenSinusoidalPositionalEmbedding.__init__": ["ModelSinusoidalPositionalEmbedding", "Module", "__init__", "class", "def", "embedding_dim", "int", "make_weights", "nn", "num_positions", "self", "super"], "musicgen/modeling_musicgen.py:MusicgenSinusoidalPositionalEmbedding.make_weights": ["False", "ModelSinusoidalPositionalEmbedding", "Module", "class", "def", "device", "dtype", "emb_weights", "embedding_dim", "get_embedding", "hasattr", "if", "int", "make_weights", "nn", "num_embeddings", "persistent", "register_buffer", "self", "to", "weights"], "musicgen/modeling_musicgen.py:MusicgenSinusoidalPositionalEmbedding.get_embedding": ["ModelSinusoidalPositionalEmbedding", "Module", "arange", "cat", "class", "cos", "def", "dim", "dtype", "emb", "embedding_dim", "exp", "float", "get_default_dtype", "get_embedding", "half_dim", "if", "int", "int64", "log", "math", "nn", "num_embeddings", "return", "sin", "to", "torch", "unsqueeze", "view", "zeros"], "musicgen/modeling_musicgen.py:MusicgenSinusoidalPositionalEmbedding.forward": ["ModelSinusoidalPositionalEmbedding", "Module", "Tensor", "arange", "bsz", "class", "codebooks", "def", "detach", "device", "embedding_dim", "forward", "if", "index_select", "input_ids", "int", "make_weights", "nn", "past_key_values_length", "position_ids", "return", "self", "seq_len", "size", "to", "torch", "view", "weights"], "musicgen/modeling_musicgen.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "contiguous", "def", "dim", "dropout", "float", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "shape", "size", "softmax", "torch", "training", "transpose", "value"], "musicgen/modeling_musicgen.py:MusicgenAttention.__init__": ["False", "Linear", "ModelAttention", "ModelConfig", "Module", "None", "True", "ValueError", "__init__", "and", "be", "bias", "bool", "by", "class", "config", "def", "divisible", "dropout", "embed_dim", "f", "float", "got", "head_dim", "if", "int", "is_causal", "is_decoder", "k_proj", "layer_idx", "must", "nn", "num_heads", "out_proj", "q_proj", "raise", "scaling", "self", "super", "v_proj"], "musicgen/modeling_musicgen.py:MusicgenAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "EncoderDecoderCache", "False", "FlashAttentionKwargs", "ModelAttention", "Module", "None", "Tensor", "True", "Unpack", "_attn_implementation", "and", "attention_interface", "attention_mask", "attn_output", "attn_weights", "bool", "bsz", "cache_position", "class", "config", "contiguous", "cross_attention_cache", "curr_past_key_values", "current_states", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "get", "head_dim", "hidden_states", "if", "is", "is_cross_attention", "is_updated", "isinstance", "k_proj", "key_states", "key_value_states", "keys", "kv_input_shape", "kwargs", "layer_idx", "layers", "nn", "not", "out_proj", "output_attentions", "past_key_values", "q_input_shape", "q_proj", "query_states", "reshape", "return", "scaling", "self", "self_attention_cache", "shape", "src_len", "tgt_len", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "values", "view"], "musicgen/modeling_musicgen.py:MusicgenDecoderLayer.__init__": ["ACT2FN", "False", "GradientCheckpointingLayer", "LayerNorm", "Linear", "ModelAttention", "ModelDecoderConfig", "ModelDecoderLayer", "None", "True", "__init__", "activation_dropout", "activation_fn", "activation_function", "attention_dropout", "bias", "class", "config", "def", "dropout", "embed_dim", "encoder_attn", "encoder_attn_layer_norm", "fc1", "fc2", "ffn_dim", "final_layer_norm", "hidden_size", "is_causal", "is_decoder", "layer_idx", "nn", "num_attention_heads", "num_heads", "self", "self_attn", "self_attn_layer_norm", "super"], "musicgen/modeling_musicgen.py:MusicgenDecoderLayer.forward": ["Cache", "False", "GradientCheckpointingLayer", "ModelDecoderLayer", "None", "Tensor", "True", "activation_dropout", "activation_fn", "attention_mask", "bool", "cache_position", "class", "cross_attn_weights", "def", "dropout", "encoder_attention_mask", "encoder_attn", "encoder_attn_layer_norm", "encoder_hidden_states", "fc1", "fc2", "final_layer_norm", "forward", "functional", "hidden_states", "if", "is", "key_value_states", "nn", "not", "output_attentions", "outputs", "p", "past_key_values", "residual", "return", "self", "self_attn", "self_attn_layer_norm", "self_attn_weights", "torch", "training", "use_cache"], "musicgen/modeling_musicgen.py:MusicgenPreTrainedModel._init_weights": ["Embedding", "False", "LayerNorm", "Linear", "ModelPreTrainedModel", "ModelSinusoidalPositionalEmbedding", "None", "PreTrainedModel", "_init_weights", "_is_hf_initialized", "and", "bias", "class", "config", "copy_", "def", "elif", "emb_weights", "embedding_dim", "get_embedding", "getattr", "if", "init", "initializer_factor", "is", "isinstance", "mean", "module", "nn", "normal_", "not", "num_positions", "ones_", "padding_idx", "self", "std", "weight", "weights", "zeros_"], "musicgen/modeling_musicgen.py:MusicgenDecoder.__init__": ["Embedding", "False", "LayerNorm", "ModelDecoder", "ModelDecoderConfig", "ModelDecoderLayer", "ModelPreTrainedModel", "ModelSinusoidalPositionalEmbedding", "ModuleList", "_", "__init__", "_attn_implementation", "attn_implementation", "class", "config", "d_model", "def", "dropout", "else", "embed_dim", "embed_positions", "embed_scale", "embed_tokens", "for", "gradient_checkpointing", "hidden_size", "i", "if", "in", "layer_idx", "layer_norm", "layerdrop", "layers", "math", "max_position_embeddings", "max_target_positions", "nn", "num_codebooks", "num_hidden_layers", "post_init", "range", "scale_embedding", "self", "sqrt", "super", "vocab_size"], "musicgen/modeling_musicgen.py:MusicgenDecoder.forward": ["BaseModelOutputWithPastAndCrossAttentions", "Cache", "DynamicCache", "EncoderDecoderCache", "False", "FloatTensor", "LongTensor", "ModelDecoder", "ModelPreTrainedModel", "None", "Setting", "Tensor", "True", "ValueError", "You", "_update_causal_mask", "_update_cross_attn_mask", "all_cross_attentions", "all_hidden_states", "all_self_attns", "and", "at", "attention_mask", "attentions", "bool", "both", "bsz", "cache_position", "cannot", "checkpointing", "class", "codebook", "config", "continue", "cross_attentions", "decoder_input_ids", "decoder_inputs_embeds", "decoder_layer", "def", "device", "dropout", "dropout_probability", "either", "elif", "else", "embed_positions", "embed_tokens", "encoder_attention_mask", "encoder_hidden_states", "enumerate", "for", "forward", "functional", "get_seq_length", "gradient", "gradient_checkpointing", "have", "hidden_states", "idx", "if", "in", "incompatible", "input", "input_ids", "input_shape", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layer_norm", "layer_outputs", "layerdrop", "layers", "logger", "nn", "not", "num_codebooks", "or", "output_attentions", "output_hidden_states", "p", "past_key_values", "past_key_values_length", "positions", "r", "raise", "random", "range", "reshape", "return", "return_dict", "same", "self", "seq_len", "shape", "size", "specify", "sum", "the", "time", "to", "torch", "training", "tuple", "uniform", "use_cache", "use_return_dict", "v", "warning_once", "with"], "musicgen/modeling_musicgen.py:MusicgenDecoder._update_causal_mask": ["ModelDecoder", "ModelPreTrainedModel", "None", "Size", "Tensor", "_attn_implementation", "_prepare_4d_causal_attention_mask", "_prepare_4d_causal_attention_mask_for_sdpa", "_update_causal_mask", "and", "attention_mask", "class", "config", "def", "device", "elif", "else", "flash_attention_2", "flex_attention", "if", "in", "input_shape", "inputs_embeds", "int", "is", "isinstance", "make_flex_block_causal_mask", "not", "ones", "past_key_values_length", "return", "sdpa", "self", "size", "torch"], "musicgen/modeling_musicgen.py:MusicgenDecoder._update_cross_attn_mask": ["False", "ModelDecoder", "ModelPreTrainedModel", "None", "Size", "Tensor", "_attn_implementation", "_prepare_4d_attention_mask", "_prepare_4d_attention_mask_for_sdpa", "_update_cross_attn_mask", "and", "class", "config", "def", "dtype", "elif", "else", "encoder_attention_mask", "encoder_hidden_states", "flash_attention_2", "flex_attention", "if", "in", "input_shape", "inputs_embeds", "is", "is_causal", "isinstance", "make_flex_block_causal_mask", "not", "query_length", "return", "sdpa", "self", "tgt_len", "torch"], "musicgen/modeling_musicgen.py:MusicgenModel.__init__": ["ModelDecoder", "ModelDecoderConfig", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "config", "decoder", "def", "post_init", "self", "super"], "musicgen/modeling_musicgen.py:MusicgenModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "decoder", "def", "embed_tokens", "get_input_embeddings", "return", "self"], "musicgen/modeling_musicgen.py:MusicgenModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "decoder", "def", "embed_tokens", "self", "set_input_embeddings", "value"], "musicgen/modeling_musicgen.py:MusicgenModel.forward": ["BaseModelOutputWithPastAndCrossAttentions", "Cache", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "cross_attentions", "decoder", "decoder_outputs", "def", "else", "encoder_attention_mask", "encoder_hidden_states", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "last_hidden_state", "not", "output_attentions", "output_hidden_states", "past_key_values", "r", "return", "return_dict", "self", "torch", "tuple", "use_cache", "use_return_dict"], "musicgen/modeling_musicgen.py:MusicgenForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "ModelDecoderConfig", "ModelForCausalLM", "ModelModel", "ModelPreTrainedModel", "ModuleList", "_", "__init__", "bias", "class", "config", "def", "for", "hidden_size", "in", "lm_heads", "model", "nn", "num_codebooks", "post_init", "range", "self", "super", "vocab_size"], "musicgen/modeling_musicgen.py:MusicgenForCausalLM.get_input_embeddings": ["GenerationMixin", "ModelForCausalLM", "ModelPreTrainedModel", "class", "decoder", "def", "embed_tokens", "get_input_embeddings", "model", "return", "self"], "musicgen/modeling_musicgen.py:MusicgenForCausalLM.set_input_embeddings": ["GenerationMixin", "ModelForCausalLM", "ModelPreTrainedModel", "class", "decoder", "def", "embed_tokens", "model", "self", "set_input_embeddings", "value"], "musicgen/modeling_musicgen.py:MusicgenForCausalLM.get_output_embeddings": ["GenerationMixin", "ModelForCausalLM", "ModelPreTrainedModel", "class", "def", "get_output_embeddings", "lm_heads", "return", "self"], "musicgen/modeling_musicgen.py:MusicgenForCausalLM.set_output_embeddings": ["GenerationMixin", "ModelForCausalLM", "ModelPreTrainedModel", "class", "def", "lm_heads", "new_embeddings", "self", "set_output_embeddings"], "musicgen/modeling_musicgen.py:MusicgenForCausalLM.forward": ["Cache", "CausalLMOutputWithCrossAttentions", "CrossEntropyLoss", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForCausalLM", "ModelPreTrainedModel", "None", "Tensor", "and", "attention_mask", "attentions", "bool", "bos_token_id", "cache_position", "class", "codebook", "codebook_labels", "codebook_logits", "config", "contiguous", "cross_attentions", "def", "device", "dim", "else", "encoder_attention_mask", "encoder_hidden_states", "for", "forward", "head", "hidden_states", "if", "in", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "lm_heads", "lm_logits", "logits", "loss", "loss_fct", "masked_fill", "model", "not", "num_codebooks", "output", "output_attentions", "output_hidden_states", "outputs", "pad_token_id", "past_key_values", "r", "range", "reshape", "return", "return_dict", "self", "shape", "shift_tokens_right", "stack", "torch", "tuple", "use_cache", "use_return_dict", "view", "zeros"], "musicgen/modeling_musicgen.py:MusicgenForCausalLM.prepare_inputs_for_generation": ["GenerationMixin", "ModelForCausalLM", "ModelPreTrainedModel", "None", "True", "and", "apply_delay_pattern_mask", "attention_mask", "build_delay_pattern_mask", "class", "def", "delay_pattern_mask", "encoder_attention_mask", "encoder_hidden_states", "generation_config", "guidance_scale", "if", "input_ids", "is", "kwargs", "max_length", "not", "pad_token_id", "past_key_values", "prepare_inputs_for_generation", "repeat", "return", "self", "use_cache"], "musicgen/modeling_musicgen.py:MusicgenForCausalLM.build_delay_pattern_mask": ["GenerationMixin", "LongTensor", "ModelForCausalLM", "ModelPreTrainedModel", "None", "audio_channels", "bool", "bsz", "build_delay_pattern_mask", "channel_codebooks", "class", "codebook", "config", "def", "delay_pattern", "device", "diagonal", "dim", "dtype", "else", "first_codebook_ids", "first_start_id", "for", "generation_config", "if", "in", "input_ids", "input_ids_shifted", "int", "is", "len", "long", "mask", "max_length", "min", "nonzero", "not", "num_codebooks", "ones", "pad_token_id", "pattern_mask", "range", "repeat_interleave", "reshape", "return", "self", "seq_len", "shape", "start_ids", "to", "torch", "tril", "triu"], "musicgen/modeling_musicgen.py:MusicgenForCausalLM.apply_delay_pattern_mask": ["GenerationMixin", "ModelForCausalLM", "ModelPreTrainedModel", "apply_delay_pattern_mask", "class", "decoder_pad_token_mask", "def", "input_ids", "return", "seq_len", "shape", "torch", "where"], "musicgen/modeling_musicgen.py:MusicgenForCausalLM.generate": ["BaseStreamer", "ClassifierFreeGuidanceLogitsProcessor", "Ensure", "GREEDY_SEARCH", "GenerationConfig", "GenerationMixin", "GenerationMode", "Got", "LogitsProcessorList", "ModelForCausalLM", "ModelPreTrainedModel", "None", "Optional", "SAMPLE", "StoppingCriteriaList", "Tensor", "ValueError", "_decoder_start_token_tensor", "_expand_inputs_for_generation", "_get_logits_processor", "_get_stopping_criteria", "_pad_token_tensor", "_prepare_attention_mask_for_generation", "_prepare_cache_for_generation", "_prepare_generated_length", "_prepare_model_inputs", "_prepare_special_tokens", "_sample", "_validate_generated_length", "_validate_model_kwargs", "activated", "and", "append", "apply_delay_pattern_mask", "attention_mask", "batch_size", "be", "beam", "bool", "bos_token_id", "build_delay_pattern_mask", "by", "class", "config", "copy", "cpu", "de", "deepcopy", "def", "delay_pattern_mask", "device", "else", "encoder_input_ids", "encoder_outputs", "expand_size", "for", "generate", "generation", "generation_config", "generation_mode", "get", "get_generation_mode", "greedy", "guidance_scale", "has_default_max_length", "has_default_min_length", "if", "in", "incompatible", "input_ids", "input_ids_length", "input_ids_seq_length", "inputs", "inputs_embeds", "inputs_tensor", "is", "is_encoder_decoder", "kwargs", "kwargs_has_attention_mask", "logits_processor", "max_cache_length", "max_length", "min_length", "mode", "model_input_name", "model_kwargs", "not", "num_beams", "num_codebooks", "num_return_sequences", "of", "one", "or", "output_ids", "outputs", "pad_token_id", "prefix_allowed_tokens_fn", "put", "raise", "requires_attention_mask", "reshape", "return", "return_dict_in_generate", "sampling", "search", "self", "sequences", "setting", "shape", "should", "stopping_criteria", "streamer", "synced_gpus", "that", "torch", "update", "use_cache", "validate"], "musicgen/modeling_musicgen.py:MusicgenForConditionalGeneration.__init__": ["AutoModel", "AutoModelForTextEncoding", "Config", "Either", "GenerationMixin", "GitHub", "Got", "Head", "If", "LM", "Linear", "Model", "ModelConfig", "ModelForCausalLM", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Please", "PreTrainedModel", "The", "ValueError", "__class__", "__init__", "_attn_implementation", "_from_config", "a", "all", "and", "audio", "audio_encoder", "be", "by", "class", "com", "config", "config_class", "configuration", "cross_attention_hidden_size", "decoder", "decoder_signature", "def", "discussion", "else", "enc_to_dec_proj", "encoder", "encoder_hidden_states", "equal", "f", "following", "for", "forward", "from_config", "get_output_embeddings", "github", "has", "have", "hidden", "hidden_size", "https", "huggingface", "if", "in", "inspect", "is", "isinstance", "issues", "it", "keys", "logger", "model", "nn", "not", "of", "on", "or", "overwritten", "parameters", "passed", "post_init", "prepared", "provided", "r", "raise", "s", "see", "selected", "self", "set", "shared", "should", "signature", "specified", "states", "super", "text", "text_encoder", "the", "three", "to", "to_dict", "transformers", "type", "use", "warning", "without"], "musicgen/modeling_musicgen.py:MusicgenForConditionalGeneration.get_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "return", "self", "text_encoder"], "musicgen/modeling_musicgen.py:MusicgenForConditionalGeneration.get_output_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "decoder", "def", "get_output_embeddings", "return", "self"], "musicgen/modeling_musicgen.py:MusicgenForConditionalGeneration.set_output_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "decoder", "def", "new_embeddings", "return", "self", "set_output_embeddings"], "musicgen/modeling_musicgen.py:MusicgenForConditionalGeneration.from_sub_models_pretrained": ["AutoConfig", "AutoModel", "Cross", "Decoder", "False", "GenerationMixin", "If", "In", "Initializing", "ModelConfig", "ModelForCausalLM", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "PreTrainedModel", "True", "ValueError", "a", "add_cross_attention", "added", "allows", "an", "and", "architecture", "are", "argument", "as", "attention", "attributes", "audio_encoder", "audio_encoder_", "audio_encoder_model", "audio_encoder_pretrained_model_name_or_path", "be", "causal", "class", "cls", "config", "cross", "decoder", "decoder_", "decoder_config", "decoder_model", "decoder_pretrained_model_name_or_path", "def", "defined", "del", "disabled", "do", "encoder_config", "f", "for", "from", "from_pretrained", "from_sub_models_pretrained", "has", "if", "in", "info", "initialize", "initialized", "is", "is_decoder", "isinstance", "items", "key", "kwargs", "kwargs_audio_encoder", "kwargs_decoder", "kwargs_text_encoder", "layers", "len", "logger", "make", "mask", "model", "model_args", "not", "of", "or", "order", "pass", "passed", "pop", "r", "raise", "randomly", "return", "return_unused_kwargs", "s", "set", "startswith", "str", "sure", "text_encoder", "text_encoder_", "text_encoder_model", "text_encoder_pretrained_model_name_or_path", "that", "the", "to", "value", "warning"], "musicgen/modeling_musicgen.py:MusicgenForConditionalGeneration.forward": ["BaseModelOutput", "BoolTensor", "Cache", "Ensure", "Expected", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Seq2SeqLMOutput", "ValueError", "and", "argument", "attention_mask", "attentions", "audio", "audio_channels", "audio_codes", "audio_encoder", "audio_encoder_", "audio_encoder_outputs", "bool", "bsz", "by", "chunk_length", "chunking", "class", "code", "codebooks", "config", "cross_attention_hidden_size", "cross_attentions", "decoder", "decoder_", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_input_ids", "decoder_inputs_embeds", "decoder_outputs", "decoder_start_token_id", "def", "dim", "disabled", "elif", "else", "enc_to_dec_proj", "encoder", "encoder_attention_mask", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "f", "for", "forward", "frame", "frames", "got", "hidden_size", "hidden_states", "if", "in", "input_ids", "input_values", "inputs_embeds", "is", "isinstance", "items", "kwargs", "kwargs_audio_encoder", "kwargs_decoder", "kwargs_text_encoder", "labels", "last_hidden_state", "len", "logits", "loss", "not", "num_codebooks", "output_attentions", "output_hidden_states", "outputs", "pad_token_id", "padding_mask", "past_key_values", "r", "raise", "repeat_interleave", "reshape", "return", "return_dict", "self", "seq_len", "setting", "shape", "shift_tokens_right", "startswith", "text_encoder", "text_encoder_", "the", "torch", "tuple", "use_cache", "use_return_dict", "value"], "musicgen/modeling_musicgen.py:MusicgenForConditionalGeneration.prepare_inputs_for_generation": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "and", "apply_delay_pattern_mask", "attention_mask", "build_delay_pattern_mask", "cache_position", "class", "decoder", "decoder_attention_mask", "decoder_delay_pattern_mask", "decoder_input_ids", "def", "elif", "else", "encoder_outputs", "generation_config", "guidance_scale", "if", "input_ids", "is", "kwargs", "max_length", "not", "pad_token_id", "past_key_values", "prepare_inputs_for_generation", "repeat", "return", "self", "shape", "use_cache"], "musicgen/modeling_musicgen.py:MusicgenForConditionalGeneration._prepare_decoder_input_ids_for_generation": ["GenerationMixin", "LongTensor", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Tensor", "_get_decoder_start_token_id", "_prepare_decoder_input_ids_for_generation", "all", "and", "batch_size", "bos_token_id", "cat", "class", "decoder", "decoder_attention_mask", "decoder_input_ids", "decoder_input_ids_start", "decoder_start_token_id", "def", "device", "dict", "dim", "dtype", "elif", "else", "if", "in", "input_ids", "int", "is", "item", "long", "model_input_name", "model_kwargs", "not", "num_codebooks", "ones", "ones_like", "pop", "return", "self", "str", "torch", "tuple"], "musicgen/modeling_musicgen.py:MusicgenForConditionalGeneration._prepare_text_encoder_kwargs_for_generation": ["Any", "BaseModelOutput", "GenerationConfig", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Tensor", "True", "_hf_hook", "_prepare_text_encoder_kwargs_for_generation", "and", "any", "argument", "attention_mask", "class", "concatenate", "cross_attn", "decoder_", "def", "dict", "dim", "else", "encoder", "encoder_accepts_wildcard", "encoder_kwargs", "encoder_outputs", "encoder_signature", "for", "forward", "generation_config", "get_encoder", "guidance_scale", "hasattr", "if", "in", "inputs_tensor", "inspect", "io_same_device", "irrelevant_prefix", "is", "items", "kwargs", "last_hidden_state", "main_input_name", "model_input_name", "model_kwargs", "not", "or", "output_attentions", "output_hidden_states", "p", "parameters", "return", "return_dict", "self", "set", "signature", "startswith", "str", "text_encoder", "torch", "use_cache", "value", "zeros_like"], "musicgen/modeling_musicgen.py:MusicgenForConditionalGeneration._prepare_audio_encoder_kwargs_for_generation": ["Ensure", "Expected", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "True", "ValueError", "_hf_hook", "_prepare_audio_encoder_kwargs_for_generation", "any", "argument", "audio", "audio_channels", "audio_codes", "audio_codes_left", "audio_codes_right", "audio_encoder", "audio_encoder_outputs", "audio_encoder_outputs_left", "audio_encoder_outputs_right", "audio_scales", "audio_scales_left", "audio_scales_right", "bsz", "but", "by", "channel", "channels", "chunk_length", "chunking", "class", "code", "codebooks", "config", "cross_attn", "decoder", "decoder_", "decoder_input_ids", "def", "dim", "disabled", "else", "encode", "encoder", "encoder_accepts_wildcard", "encoder_kwargs", "encoder_signature", "example", "f", "for", "forward", "frame", "frames", "get_encoder", "got", "has", "hasattr", "if", "in", "input_values", "inspect", "io_same_device", "irrelevant_prefix", "is", "items", "kwargs", "main_input_name", "modality", "model_input_name", "model_kwargs", "new_ones", "not", "num_codebooks", "or", "outputs", "p", "parameters", "raise", "reshape", "return", "return_dict", "self", "seq_len", "set", "setting", "shape", "signature", "stack", "startswith", "stereo", "str", "the", "torch", "use_cache", "value"], "musicgen/modeling_musicgen.py:MusicgenForConditionalGeneration.prepare_decoder_input_ids_from_labels": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "Tensor", "bos_token_id", "class", "config", "decoder", "def", "labels", "pad_token_id", "prepare_decoder_input_ids_from_labels", "return", "self", "shift_tokens_right", "torch"], "musicgen/modeling_musicgen.py:MusicgenForConditionalGeneration.resize_token_embeddings": ["EncoderDecoderModel", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "NotImplementedError", "Please", "Resizing", "args", "class", "decoder", "def", "directly", "embedding", "encoder", "is", "kwargs", "layers", "methods", "model", "not", "objects", "of", "or", "raise", "resize_token_embeddings", "respective", "self", "supported", "the", "use", "via", "wrapped"], "musicgen/modeling_musicgen.py:MusicgenForConditionalGeneration.freeze_audio_encoder": ["False", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "_requires_grad", "audio_encoder", "class", "def", "for", "freeze_audio_encoder", "in", "param", "parameters", "requires_grad", "self"], "musicgen/modeling_musicgen.py:MusicgenForConditionalGeneration.freeze_text_encoder": ["False", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "_requires_grad", "class", "def", "for", "freeze_text_encoder", "in", "param", "parameters", "requires_grad", "self", "text_encoder"], "musicgen/modeling_musicgen.py:MusicgenForConditionalGeneration._maybe_initialize_input_ids_for_generation": ["GenerationMixin", "LongTensor", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Tensor", "ValueError", "_maybe_initialize_input_ids_for_generation", "are", "batch_size", "be", "bos_token_id", "break", "class", "def", "defined", "device", "dict", "dtype", "encoder_outputs", "for", "get", "has", "if", "in", "input_ids", "inputs", "int", "is", "isinstance", "long", "model_kwargs", "no", "not", "ones", "provided", "raise", "return", "self", "shape", "size", "str", "to", "torch", "value", "values", "when"], "musicgen/modeling_musicgen.py:MusicgenForConditionalGeneration._get_decoder_start_token_id": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "ValueError", "_get_decoder_start_token_id", "be", "bos_token_id", "class", "decoder", "decoder_start_token_id", "def", "defined", "elif", "else", "encoder", "for", "generation", "generation_config", "has", "if", "int", "is", "list", "not", "or", "raise", "return", "self", "to"], "musicgen/modeling_musicgen.py:MusicgenForConditionalGeneration.generate": ["BaseModelOutput", "BaseStreamer", "ClassifierFreeGuidanceLogitsProcessor", "Ensure", "False", "GREEDY_SEARCH", "GenerationConfig", "GenerationMixin", "GenerationMode", "Got", "LogitsProcessorList", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Optional", "SAMPLE", "StoppingCriteriaList", "Tensor", "ValueError", "_bos_token_tensor", "_decoder_start_token_tensor", "_expand_inputs_for_generation", "_extract_generation_mode_kwargs", "_get_logits_processor", "_get_stopping_criteria", "_pad_token_tensor", "_prefill", "_prepare_attention_mask_for_generation", "_prepare_audio_encoder_kwargs_for_generation", "_prepare_decoder_input_ids_for_generation", "_prepare_generated_length", "_prepare_generation_config", "_prepare_model_inputs", "_prepare_special_tokens", "_prepare_text_encoder_kwargs_for_generation", "_sample", "_validate_generation_mode", "_validate_model_kwargs", "activated", "and", "append", "apply_delay_pattern_mask", "attention_mask", "audio_channels", "audio_encoder", "audio_scales", "audio_values", "batch_size", "be", "beam", "bool", "bos_token_id", "build_delay_pattern_mask", "by", "cat", "class", "codec_outputs_left", "codec_outputs_right", "config", "copy", "cpu", "de", "decode", "decoder", "decoder_delay_pattern_mask", "decoder_input_ids", "decoder_start_token_id", "def", "device", "dim", "else", "encoder_input_ids", "encoder_outputs", "expand_size", "for", "generate", "generation", "generation_config", "generation_mode", "generation_mode_kwargs", "get", "get_generation_mode", "greedy", "guidance_scale", "has_default_max_length", "has_default_min_length", "if", "in", "incompatible", "input_ids", "input_ids_length", "input_ids_seq_length", "input_values", "inputs", "inputs_tensor", "is", "is_encoder_decoder", "kwargs", "kwargs_has_attention_mask", "last_hidden_state", "logits_processor", "max_length", "min_length", "mode", "model_input_name", "model_kwargs", "not", "num_beams", "num_codebooks", "num_return_sequences", "of", "one", "or", "output_ids", "output_values", "output_values_left", "output_values_right", "outputs", "pad_token_id", "prefill_outputs", "prefix_allowed_tokens_fn", "put", "raise", "requires_attention_mask", "reshape", "return", "return_dict_in_generate", "sampling", "search", "self", "sequences", "setting", "shape", "should", "stopping_criteria", "streamer", "synced_gpus", "that", "torch", "tuple", "type", "use_cache"], "musicgen/modeling_musicgen.py:MusicgenForConditionalGeneration.get_unconditional_inputs": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "ModelUnconditionalInput", "attention_mask", "class", "config", "def", "device", "dtype", "encoder_outputs", "get_unconditional_inputs", "guidance_scale", "hidden_size", "last_hidden_state", "long", "num_samples", "return", "self", "text_encoder", "torch", "zeros"], "musicgen_melody/modeling_musicgen_melody.py:shift_tokens_right": ["Make", "Model_tokens_right", "Modeled_input_ids", "None", "Tensor", "ValueError", "attribute", "clone", "configuration", "decoder_start_token_id", "def", "if", "input_ids", "int", "is", "masked_fill_", "model", "new_zeros", "of", "pad_token_id", "raise", "return", "s", "set", "shape", "sure", "the", "to", "torch", "transpose"], "musicgen_melody/modeling_musicgen_melody.py:MusicgenMelodySinusoidalPositionalEmbedding.__init__": ["ModelSinusoidalPositionalEmbedding", "Module", "__init__", "class", "def", "embedding_dim", "int", "make_weights", "nn", "num_positions", "self", "super"], "musicgen_melody/modeling_musicgen_melody.py:MusicgenMelodySinusoidalPositionalEmbedding.make_weights": ["False", "ModelSinusoidalPositionalEmbedding", "Module", "class", "def", "device", "dtype", "emb_weights", "embedding_dim", "get_embedding", "hasattr", "if", "int", "make_weights", "nn", "num_embeddings", "persistent", "register_buffer", "self", "to", "weights"], "musicgen_melody/modeling_musicgen_melody.py:MusicgenMelodySinusoidalPositionalEmbedding.get_embedding": ["ModelSinusoidalPositionalEmbedding", "Module", "arange", "cat", "class", "cos", "def", "dim", "dtype", "emb", "embedding_dim", "exp", "float", "get_default_dtype", "get_embedding", "half_dim", "if", "int", "int64", "log", "math", "nn", "num_embeddings", "return", "sin", "to", "torch", "unsqueeze", "view", "zeros"], "musicgen_melody/modeling_musicgen_melody.py:MusicgenMelodySinusoidalPositionalEmbedding.forward": ["ModelSinusoidalPositionalEmbedding", "Module", "Tensor", "_", "arange", "bsz", "class", "def", "detach", "device", "embedding_dim", "forward", "if", "index_select", "inputs_embeds", "int", "make_weights", "nn", "past_key_values_length", "position_ids", "return", "self", "seq_len", "size", "to", "torch", "view", "weights"], "musicgen_melody/modeling_musicgen_melody.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "contiguous", "def", "dim", "dropout", "float", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "shape", "size", "softmax", "torch", "training", "transpose", "value"], "musicgen_melody/modeling_musicgen_melody.py:MusicgenMelodyAttention.__init__": ["False", "Linear", "ModelAttention", "ModelConfig", "Module", "None", "True", "ValueError", "__init__", "and", "be", "bias", "bool", "by", "class", "config", "def", "divisible", "dropout", "embed_dim", "f", "float", "got", "head_dim", "if", "int", "is_causal", "is_decoder", "k_proj", "layer_idx", "must", "nn", "num_heads", "out_proj", "q_proj", "raise", "scaling", "self", "super", "v_proj"], "musicgen_melody/modeling_musicgen_melody.py:MusicgenMelodyAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "EncoderDecoderCache", "False", "FlashAttentionKwargs", "ModelAttention", "Module", "None", "Tensor", "True", "Unpack", "_attn_implementation", "and", "attention_interface", "attention_mask", "attn_output", "attn_weights", "bool", "bsz", "cache_position", "class", "config", "contiguous", "cross_attention_cache", "curr_past_key_values", "current_states", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "get", "head_dim", "hidden_states", "if", "is", "is_cross_attention", "is_updated", "isinstance", "k_proj", "key_states", "key_value_states", "keys", "kv_input_shape", "kwargs", "layer_idx", "layers", "nn", "not", "out_proj", "output_attentions", "past_key_values", "q_input_shape", "q_proj", "query_states", "reshape", "return", "scaling", "self", "self_attention_cache", "shape", "src_len", "tgt_len", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "values", "view"], "musicgen_melody/modeling_musicgen_melody.py:MusicgenMelodyDecoderLayer.__init__": ["ACT2FN", "False", "GradientCheckpointingLayer", "LayerNorm", "Linear", "ModelAttention", "ModelDecoderConfig", "ModelDecoderLayer", "None", "True", "__init__", "activation_dropout", "activation_fn", "activation_function", "attention_dropout", "bias", "class", "config", "def", "dropout", "embed_dim", "fc1", "fc2", "ffn_dim", "final_layer_norm", "hidden_size", "is_causal", "is_decoder", "layer_idx", "nn", "num_attention_heads", "num_heads", "self", "self_attn", "self_attn_layer_norm", "super"], "musicgen_melody/modeling_musicgen_melody.py:MusicgenMelodyDecoderLayer.forward": ["Cache", "False", "GradientCheckpointingLayer", "ModelDecoderLayer", "None", "Tensor", "True", "activation_dropout", "activation_fn", "attention_mask", "bool", "cache_position", "class", "def", "dropout", "fc1", "fc2", "final_layer_norm", "forward", "functional", "hidden_states", "nn", "output_attentions", "p", "past_key_values", "residual", "return", "self", "self_attn", "self_attn_layer_norm", "self_attn_weights", "torch", "training", "use_cache"], "musicgen_melody/modeling_musicgen_melody.py:MusicgenMelodyPreTrainedModel._init_weights": ["Embedding", "False", "LayerNorm", "Linear", "ModelPreTrainedModel", "ModelSinusoidalPositionalEmbedding", "None", "PreTrainedModel", "_init_weights", "_is_hf_initialized", "and", "bias", "class", "config", "copy_", "def", "elif", "emb_weights", "embedding_dim", "get_embedding", "getattr", "if", "init", "initializer_factor", "is", "isinstance", "mean", "module", "nn", "normal_", "not", "num_positions", "ones_", "padding_idx", "self", "std", "weight", "weights", "zeros_"], "musicgen_melody/modeling_musicgen_melody.py:MusicgenMelodyDecoder.__init__": ["Embedding", "False", "LayerNorm", "ModelDecoder", "ModelDecoderConfig", "ModelDecoderLayer", "ModelPreTrainedModel", "ModelSinusoidalPositionalEmbedding", "ModuleList", "_", "__init__", "_attn_implementation", "attn_implementation", "class", "config", "d_model", "def", "dropout", "else", "embed_dim", "embed_positions", "embed_scale", "embed_tokens", "for", "gradient_checkpointing", "hidden_size", "i", "if", "in", "layer_idx", "layer_norm", "layerdrop", "layers", "math", "max_position_embeddings", "max_target_positions", "nn", "num_codebooks", "num_hidden_layers", "post_init", "range", "scale_embedding", "self", "sqrt", "super", "vocab_size"], "musicgen_melody/modeling_musicgen_melody.py:MusicgenMelodyDecoder.forward": ["BaseModelOutputWithPast", "Cache", "DynamicCache", "EncoderDecoderCache", "False", "FloatTensor", "LongTensor", "ModelDecoder", "ModelPreTrainedModel", "None", "Setting", "Tensor", "True", "ValueError", "You", "_update_causal_mask", "all_attentions", "all_hidden_states", "and", "at", "attention_mask", "attentions", "bool", "both", "bsz", "cache_position", "cannot", "cat", "checkpointing", "class", "codebook", "config", "continue", "decoder_input_ids", "decoder_inputs_embeds", "decoder_layer", "def", "device", "dim", "dropout", "dropout_probability", "either", "elif", "else", "embed_positions", "embed_tokens", "encoder_attention_mask", "encoder_hidden_states", "enumerate", "for", "forward", "functional", "get_seq_length", "gradient", "gradient_checkpointing", "have", "hidden_states", "idx", "if", "in", "incompatible", "input", "input_ids", "input_shape", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layer_norm", "layer_outputs", "layerdrop", "layers", "logger", "nn", "not", "num_codebooks", "ones", "or", "output_attentions", "output_hidden_states", "p", "past_key_values", "past_key_values_length", "positions", "r", "raise", "random", "range", "reshape", "return", "return_dict", "same", "self", "seq_len", "shape", "size", "specify", "sum", "the", "time", "to", "torch", "training", "tuple", "uniform", "use_cache", "use_return_dict", "v", "warning_once", "with"], "musicgen_melody/modeling_musicgen_melody.py:MusicgenMelodyDecoder._update_causal_mask": ["ModelDecoder", "ModelPreTrainedModel", "None", "Size", "Tensor", "_attn_implementation", "_prepare_4d_causal_attention_mask", "_prepare_4d_causal_attention_mask_for_sdpa", "_update_causal_mask", "and", "attention_mask", "class", "config", "def", "device", "elif", "else", "flash_attention_2", "flex_attention", "if", "in", "input_shape", "inputs_embeds", "int", "is", "isinstance", "make_flex_block_causal_mask", "not", "ones", "past_key_values_length", "return", "sdpa", "self", "size", "torch"], "musicgen_melody/modeling_musicgen_melody.py:MusicgenMelodyDecoder._update_cross_attn_mask": ["ModelDecoder", "ModelPreTrainedModel", "None", "Size", "Tensor", "_update_cross_attn_mask", "class", "def", "encoder_attention_mask", "encoder_hidden_states", "input_shape", "inputs_embeds", "pass", "self", "torch"], "musicgen_melody/modeling_musicgen_melody.py:MusicgenMelodyModel.__init__": ["ModelDecoder", "ModelDecoderConfig", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "config", "decoder", "def", "post_init", "self", "super"], "musicgen_melody/modeling_musicgen_melody.py:MusicgenMelodyModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "decoder", "def", "embed_tokens", "get_input_embeddings", "return", "self"], "musicgen_melody/modeling_musicgen_melody.py:MusicgenMelodyModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "decoder", "def", "embed_tokens", "self", "set_input_embeddings", "value"], "musicgen_melody/modeling_musicgen_melody.py:MusicgenMelodyModel.forward": ["BaseModelOutputWithPast", "Cache", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "decoder", "decoder_outputs", "def", "else", "encoder_attention_mask", "encoder_hidden_states", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "last_hidden_state", "not", "output_attentions", "output_hidden_states", "past_key_values", "r", "return", "return_dict", "self", "torch", "tuple", "use_cache", "use_return_dict"], "musicgen_melody/modeling_musicgen_melody.py:MusicgenMelodyForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "ModelDecoderConfig", "ModelForCausalLM", "ModelModel", "ModelPreTrainedModel", "ModuleList", "_", "__init__", "bias", "class", "config", "def", "for", "hidden_size", "in", "lm_heads", "model", "nn", "num_codebooks", "post_init", "range", "self", "super", "vocab_size"], "musicgen_melody/modeling_musicgen_melody.py:MusicgenMelodyForCausalLM.get_input_embeddings": ["GenerationMixin", "ModelForCausalLM", "ModelPreTrainedModel", "class", "decoder", "def", "embed_tokens", "get_input_embeddings", "model", "return", "self"], "musicgen_melody/modeling_musicgen_melody.py:MusicgenMelodyForCausalLM.set_input_embeddings": ["GenerationMixin", "ModelForCausalLM", "ModelPreTrainedModel", "class", "decoder", "def", "embed_tokens", "model", "self", "set_input_embeddings", "value"], "musicgen_melody/modeling_musicgen_melody.py:MusicgenMelodyForCausalLM.get_output_embeddings": ["GenerationMixin", "ModelForCausalLM", "ModelPreTrainedModel", "class", "def", "get_output_embeddings", "lm_heads", "return", "self"], "musicgen_melody/modeling_musicgen_melody.py:MusicgenMelodyForCausalLM.set_output_embeddings": ["GenerationMixin", "ModelForCausalLM", "ModelPreTrainedModel", "class", "def", "lm_heads", "new_embeddings", "self", "set_output_embeddings"], "musicgen_melody/modeling_musicgen_melody.py:MusicgenMelodyForCausalLM.forward": ["Cache", "CrossEntropyLoss", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForCausalLM", "ModelOutputWithPast", "ModelPreTrainedModel", "None", "Tensor", "and", "attention_mask", "attentions", "bool", "bos_token_id", "cache_position", "class", "codebook", "codebook_labels", "codebook_logits", "config", "contiguous", "def", "device", "dim", "else", "encoder_attention_mask", "encoder_hidden_states", "for", "forward", "head", "hidden_states", "if", "in", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "lm_heads", "lm_logits", "logits", "loss", "loss_fct", "masked_fill", "model", "not", "num_codebooks", "output", "output_attentions", "output_hidden_states", "outputs", "pad_token_id", "past_key_values", "r", "range", "reshape", "return", "return_dict", "self", "shape", "shift_tokens_right", "stack", "torch", "tuple", "use_cache", "use_return_dict", "view", "zeros"], "musicgen_melody/modeling_musicgen_melody.py:MusicgenMelodyForCausalLM.prepare_inputs_for_generation": ["GenerationMixin", "ModelForCausalLM", "ModelPreTrainedModel", "None", "True", "and", "apply_delay_pattern_mask", "attention_mask", "build_delay_pattern_mask", "class", "concatenate", "def", "delay_pattern_mask", "dim", "encoder_attention_mask", "encoder_hidden_states", "generation_config", "guidance_scale", "if", "input_ids", "is", "kwargs", "max_length", "not", "pad_token_id", "past_key_values", "prepare_inputs_for_generation", "repeat", "return", "self", "torch", "use_cache", "zeros_like"], "musicgen_melody/modeling_musicgen_melody.py:MusicgenMelodyForCausalLM.build_delay_pattern_mask": ["GenerationMixin", "LongTensor", "ModelForCausalLM", "ModelPreTrainedModel", "None", "audio_channels", "bool", "bsz", "build_delay_pattern_mask", "channel_codebooks", "class", "codebook", "config", "def", "delay_pattern", "device", "diagonal", "dim", "dtype", "else", "first_codebook_ids", "first_start_id", "for", "generation_config", "if", "in", "input_ids", "input_ids_shifted", "int", "is", "len", "long", "mask", "max_length", "min", "nonzero", "not", "num_codebooks", "ones", "pad_token_id", "pattern_mask", "range", "repeat_interleave", "reshape", "return", "self", "seq_len", "shape", "start_ids", "to", "torch", "tril", "triu"], "musicgen_melody/modeling_musicgen_melody.py:MusicgenMelodyForCausalLM.apply_delay_pattern_mask": ["GenerationMixin", "ModelForCausalLM", "ModelPreTrainedModel", "apply_delay_pattern_mask", "class", "decoder_pad_token_mask", "def", "input_ids", "return", "seq_len", "shape", "torch", "where"], "musicgen_melody/modeling_musicgen_melody.py:MusicgenMelodyForCausalLM.generate": ["BaseStreamer", "ClassifierFreeGuidanceLogitsProcessor", "Ensure", "GREEDY_SEARCH", "GenerationConfig", "GenerationMixin", "GenerationMode", "Got", "LogitsProcessorList", "ModelForCausalLM", "ModelPreTrainedModel", "None", "Optional", "SAMPLE", "StoppingCriteriaList", "Tensor", "ValueError", "_decoder_start_token_tensor", "_expand_inputs_for_generation", "_get_logits_processor", "_get_stopping_criteria", "_pad_token_tensor", "_prepare_attention_mask_for_generation", "_prepare_generated_length", "_prepare_model_inputs", "_prepare_special_tokens", "_sample", "_validate_model_kwargs", "activated", "and", "append", "apply_delay_pattern_mask", "attention_mask", "batch_size", "be", "beam", "bool", "bos_token_id", "build_delay_pattern_mask", "by", "class", "copy", "cpu", "de", "deepcopy", "def", "delay_pattern_mask", "device", "else", "encoder_input_ids", "encoder_outputs", "expand_size", "for", "generate", "generation", "generation_config", "generation_mode", "get", "get_generation_mode", "greedy", "guidance_scale", "has_default_max_length", "has_default_min_length", "if", "in", "incompatible", "input_ids", "input_ids_length", "input_ids_seq_length", "inputs", "inputs_tensor", "is", "kwargs", "kwargs_has_attention_mask", "logits_processor", "max_length", "min_length", "mode", "model_input_name", "model_kwargs", "not", "num_beams", "num_codebooks", "num_return_sequences", "of", "one", "or", "output_ids", "outputs", "pad_token_id", "prefix_allowed_tokens_fn", "put", "raise", "requires_attention_mask", "reshape", "return", "return_dict_in_generate", "sampling", "search", "self", "sequences", "setting", "shape", "should", "stopping_criteria", "streamer", "synced_gpus", "that", "torch", "update", "use_cache", "validate"], "musicgen_melody/modeling_musicgen_melody.py:MusicgenMelodyForConditionalGeneration.__init__": ["AutoModel", "AutoModelForTextEncoding", "Config", "Either", "GenerationMixin", "Head", "LM", "Linear", "Melody", "Model", "ModelConfig", "ModelForCausalLM", "ModelForConditionalGeneration", "None", "Please", "PreTrainedModel", "The", "ValueError", "__init__", "_attn_implementation", "_from_config", "a", "all", "and", "audio", "audio_enc_to_dec_proj", "audio_encoder", "be", "class", "config", "config_class", "configuration", "decoder", "def", "else", "enc_to_dec_proj", "encoder", "f", "from_config", "get_output_embeddings", "has", "have", "hidden_size", "if", "in", "is", "isinstance", "model", "nn", "not", "num_chroma", "of", "or", "post_init", "provided", "r", "raise", "self", "should", "super", "text", "text_encoder", "three", "to", "type", "use", "without"], "musicgen_melody/modeling_musicgen_melody.py:MusicgenMelodyForConditionalGeneration._init_weights": ["GenerationMixin", "Linear", "ModelForConditionalGeneration", "None", "PreTrainedModel", "_init_weights", "bias", "class", "config", "decoder", "def", "if", "init", "initializer_factor", "is", "isinstance", "mean", "module", "nn", "normal_", "not", "self", "std", "weight", "zeros_"], "musicgen_melody/modeling_musicgen_melody.py:MusicgenMelodyForConditionalGeneration.get_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "PreTrainedModel", "class", "def", "get_input_embeddings", "return", "self", "text_encoder"], "musicgen_melody/modeling_musicgen_melody.py:MusicgenMelodyForConditionalGeneration.get_output_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "PreTrainedModel", "class", "decoder", "def", "get_output_embeddings", "return", "self"], "musicgen_melody/modeling_musicgen_melody.py:MusicgenMelodyForConditionalGeneration.set_output_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "PreTrainedModel", "class", "decoder", "def", "new_embeddings", "return", "self", "set_output_embeddings"], "musicgen_melody/modeling_musicgen_melody.py:MusicgenMelodyForConditionalGeneration.from_sub_models_pretrained": ["AutoConfig", "AutoModel", "Cross", "Decoder", "False", "GenerationMixin", "If", "In", "Initializing", "ModelConfig", "ModelForCausalLM", "ModelForConditionalGeneration", "None", "PreTrainedModel", "True", "ValueError", "a", "add_cross_attention", "added", "allows", "an", "and", "architecture", "are", "argument", "as", "attention", "attributes", "audio_encoder", "audio_encoder_", "audio_encoder_model", "audio_encoder_pretrained_model_name_or_path", "be", "causal", "class", "cls", "config", "cross", "decoder", "decoder_", "decoder_config", "decoder_model", "decoder_pretrained_model_name_or_path", "def", "defined", "del", "disabled", "do", "encoder_config", "f", "for", "from", "from_pretrained", "from_sub_models_pretrained", "has", "if", "in", "info", "initialize", "initialized", "is", "is_decoder", "isinstance", "items", "key", "kwargs", "kwargs_audio_encoder", "kwargs_decoder", "kwargs_text_encoder", "layers", "len", "logger", "make", "mask", "model", "model_args", "not", "of", "or", "order", "pass", "passed", "pop", "r", "raise", "randomly", "return", "return_unused_kwargs", "s", "set", "startswith", "str", "sure", "text_encoder", "text_encoder_", "text_encoder_model", "text_encoder_pretrained_model_name_or_path", "that", "the", "to", "value", "warning"], "musicgen_melody/modeling_musicgen_melody.py:MusicgenMelodyForConditionalGeneration.forward": ["BoolTensor", "Cache", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForConditionalGeneration", "ModelOutputWithPast", "None", "PreTrainedModel", "The", "and", "argument", "attention_mask", "attentions", "audio", "audio_enc_to_dec_proj", "audio_hidden_states", "be", "bool", "bos_token_id", "cat", "ceil", "chroma", "chroma_length", "class", "conditional", "config", "decoder", "decoder_", "decoder_attention_mask", "decoder_input_ids", "decoder_inputs_embeds", "decoder_outputs", "def", "device", "dim", "dtype", "duration", "else", "enc_to_dec_proj", "encoder_hidden_states", "encoder_outputs", "exceeds", "f", "for", "forward", "frames", "hidden_size", "hidden_states", "if", "in", "input_features", "input_ids", "inputs_embeds", "int", "is", "items", "kwargs", "kwargs_decoder", "kwargs_text_encoder", "labels", "len", "length", "logger", "logits", "loss", "math", "maximum", "n_repeat", "not", "num_chroma", "of", "or", "output_attentions", "output_hidden_states", "pad_token_id", "past_key_values", "r", "repeat", "return", "return_dict", "self", "shape", "shift_tokens_right", "signal", "startswith", "text_encoder", "text_encoder_", "the", "to", "torch", "truncated", "tuple", "use_cache", "use_return_dict", "value", "warning", "which", "will", "zeros"], "musicgen_melody/modeling_musicgen_melody.py:MusicgenMelodyForConditionalGeneration.prepare_inputs_for_generation": ["GenerationMixin", "ModelForConditionalGeneration", "None", "PreTrainedModel", "and", "apply_delay_pattern_mask", "attention_mask", "build_delay_pattern_mask", "class", "decoder", "decoder_attention_mask", "decoder_delay_pattern_mask", "decoder_input_ids", "def", "else", "encoder_hidden_states", "generation_config", "get_seq_length", "guidance_scale", "if", "input_ids", "is", "kwargs", "max_length", "not", "pad_token_id", "past_key_values", "past_length", "prepare_inputs_for_generation", "remove_prefix_length", "repeat", "return", "self", "shape", "use_cache"], "musicgen_melody/modeling_musicgen_melody.py:MusicgenMelodyForConditionalGeneration._prepare_decoder_input_ids_for_generation": ["GenerationMixin", "LongTensor", "ModelForConditionalGeneration", "None", "PreTrainedModel", "Tensor", "_get_decoder_start_token_id", "_prepare_decoder_input_ids_for_generation", "all", "and", "batch_size", "bos_token_id", "cat", "class", "decoder", "decoder_attention_mask", "decoder_input_ids", "decoder_input_ids_start", "decoder_start_token_id", "def", "device", "dict", "dim", "dtype", "elif", "else", "if", "in", "input_ids", "int", "is", "item", "long", "model_input_name", "model_kwargs", "not", "num_codebooks", "ones", "ones_like", "pop", "return", "self", "str", "torch", "tuple"], "musicgen_melody/modeling_musicgen_melody.py:MusicgenMelodyForConditionalGeneration._prepare_encoder_hidden_states_kwargs_for_generation": ["Any", "GenerationConfig", "GenerationMixin", "ModelForConditionalGeneration", "None", "PreTrainedModel", "Tensor", "True", "_hf_hook", "_prepare_encoder_hidden_states_kwargs_for_generation", "and", "any", "argument", "attention_mask", "audio_enc_to_dec_proj", "audio_hidden_states", "cat", "ceil", "chroma_length", "class", "concatenate", "config", "decoder", "decoder_", "def", "device", "dict", "dim", "dtype", "else", "enc_to_dec_proj", "encoder", "encoder_accepts_wildcard", "encoder_attention_mask", "encoder_hidden_states", "encoder_kwargs", "encoder_signature", "for", "forward", "generation_config", "get", "get_encoder", "guidance_scale", "hasattr", "hidden_size", "if", "in", "input_features", "inputs_tensor", "inspect", "int", "io_same_device", "irrelevant_prefix", "is", "items", "kwargs", "last_hidden_state", "main_input_name", "math", "model_input_name", "model_kwargs", "n_repeat", "not", "null_audio_hidden_states", "num_chroma", "or", "output_attentions", "output_hidden_states", "p", "parameters", "pop", "repeat", "return", "return_dict", "self", "set", "shape", "signature", "startswith", "str", "text_encoder", "torch", "use_cache", "value", "zeros", "zeros_like"], "musicgen_melody/modeling_musicgen_melody.py:MusicgenMelodyForConditionalGeneration.prepare_decoder_input_ids_from_labels": ["GenerationMixin", "ModelForConditionalGeneration", "PreTrainedModel", "Tensor", "bos_token_id", "class", "config", "decoder", "def", "labels", "pad_token_id", "prepare_decoder_input_ids_from_labels", "return", "self", "shift_tokens_right", "torch"], "musicgen_melody/modeling_musicgen_melody.py:MusicgenMelodyForConditionalGeneration.resize_token_embeddings": ["EncoderDecoderModel", "GenerationMixin", "ModelForConditionalGeneration", "NotImplementedError", "Please", "PreTrainedModel", "Resizing", "args", "class", "decoder", "def", "directly", "embedding", "encoder", "is", "kwargs", "layers", "methods", "model", "not", "objects", "of", "or", "raise", "resize_token_embeddings", "respective", "self", "supported", "the", "use", "via", "wrapped"], "musicgen_melody/modeling_musicgen_melody.py:MusicgenMelodyForConditionalGeneration._maybe_initialize_input_ids_for_generation": ["GenerationMixin", "LongTensor", "ModelForConditionalGeneration", "None", "PreTrainedModel", "Tensor", "ValueError", "_maybe_initialize_input_ids_for_generation", "are", "batch_size", "be", "bos_token_id", "break", "class", "def", "defined", "device", "dict", "dtype", "for", "has", "if", "in", "input_ids", "inputs", "int", "is", "isinstance", "long", "model_kwargs", "no", "not", "ones", "provided", "raise", "return", "self", "shape", "str", "to", "torch", "value", "values", "when"], "musicgen_melody/modeling_musicgen_melody.py:MusicgenMelodyForConditionalGeneration.freeze_audio_encoder": ["False", "GenerationMixin", "ModelForConditionalGeneration", "PreTrainedModel", "_requires_grad", "audio_encoder", "class", "def", "for", "freeze_audio_encoder", "in", "param", "parameters", "requires_grad", "self"], "musicgen_melody/modeling_musicgen_melody.py:MusicgenMelodyForConditionalGeneration.freeze_text_encoder": ["False", "GenerationMixin", "ModelForConditionalGeneration", "PreTrainedModel", "_requires_grad", "class", "def", "for", "freeze_text_encoder", "in", "param", "parameters", "requires_grad", "self", "text_encoder"], "musicgen_melody/modeling_musicgen_melody.py:MusicgenMelodyForConditionalGeneration._get_decoder_start_token_id": ["GenerationMixin", "ModelForConditionalGeneration", "None", "PreTrainedModel", "ValueError", "_get_decoder_start_token_id", "be", "bos_token_id", "class", "decoder", "decoder_start_token_id", "def", "defined", "elif", "else", "encoder", "for", "generation", "generation_config", "has", "if", "int", "is", "list", "not", "or", "raise", "return", "self", "to"], "musicgen_melody/modeling_musicgen_melody.py:MusicgenMelodyForConditionalGeneration.generate": ["BaseStreamer", "ClassifierFreeGuidanceLogitsProcessor", "Ensure", "GREEDY_SEARCH", "GenerationConfig", "GenerationMixin", "GenerationMode", "Got", "LogitsProcessorList", "ModelForConditionalGeneration", "None", "Optional", "PreTrainedModel", "SAMPLE", "StoppingCriteriaList", "Tensor", "ValueError", "_bos_token_tensor", "_decoder_start_token_tensor", "_expand_inputs_for_generation", "_get_logits_processor", "_get_stopping_criteria", "_pad_token_tensor", "_prepare_attention_mask_for_generation", "_prepare_cache_for_generation", "_prepare_decoder_input_ids_for_generation", "_prepare_encoder_hidden_states_kwargs_for_generation", "_prepare_generated_length", "_prepare_model_inputs", "_prepare_special_tokens", "_sample", "_validate_generated_length", "_validate_model_kwargs", "activated", "and", "append", "apply_delay_pattern_mask", "attention_mask", "audio_channels", "audio_encoder", "audio_scales", "audio_values", "batch_size", "be", "beam", "bool", "bos_token_id", "build_delay_pattern_mask", "by", "cat", "class", "codec_outputs_left", "codec_outputs_right", "config", "copy", "cpu", "de", "decode", "decoder", "decoder_delay_pattern_mask", "decoder_start_token_id", "deepcopy", "def", "device", "dim", "else", "encoder_hidden_states", "encoder_input_ids", "encoder_outputs", "expand_size", "for", "generate", "generation", "generation_config", "generation_mode", "get", "get_generation_mode", "greedy", "guidance_scale", "has_default_max_length", "has_default_min_length", "if", "in", "incompatible", "input_ids", "input_ids_length", "input_ids_seq_length", "inputs", "inputs_embeds", "inputs_tensor", "is", "is_encoder_decoder", "kwargs", "kwargs_has_attention_mask", "logits_processor", "max_cache_length", "max_length", "min_length", "mode", "model_input_name", "model_kwargs", "not", "num_beams", "num_codebooks", "num_return_sequences", "of", "one", "or", "output_ids", "output_values", "output_values_left", "output_values_right", "outputs", "pad_token_id", "prefix_allowed_tokens_fn", "put", "raise", "requires_attention_mask", "reshape", "return", "return_dict_in_generate", "sampling", "search", "self", "sequences", "setting", "shape", "should", "stopping_criteria", "streamer", "synced_gpus", "that", "torch", "update", "use_cache", "validate"], "mvp/modeling_mvp.py:shift_tokens_right": ["Model_tokens_right", "Modeled_input_ids", "None", "Tensor", "ValueError", "be", "clone", "config", "decoder_start_token_id", "def", "defined", "has", "if", "input_ids", "int", "is", "masked_fill_", "model", "new_zeros", "pad_token_id", "raise", "return", "self", "shape", "to", "torch"], "mvp/modeling_mvp.py:MvpLearnedPositionalEmbedding.__init__": ["Embedding", "ModelLearnedPositionalEmbedding", "__init__", "class", "def", "embedding_dim", "int", "nn", "num_embeddings", "offset", "self", "super"], "mvp/modeling_mvp.py:MvpLearnedPositionalEmbedding.forward": ["Embedding", "ModelLearnedPositionalEmbedding", "None", "Tensor", "arange", "bsz", "class", "def", "device", "dtype", "else", "expand", "forward", "if", "input_ids", "int", "is", "long", "nn", "offset", "past_key_values_length", "position_ids", "return", "self", "seq_len", "shape", "super", "torch", "unsqueeze", "weight"], "mvp/modeling_mvp.py:MvpAttention.__init__": ["False", "Linear", "ModelAttention", "Module", "None", "True", "ValueError", "__init__", "and", "be", "bias", "bool", "by", "class", "def", "divisible", "dropout", "embed_dim", "f", "float", "got", "head_dim", "if", "int", "is_decoder", "k_proj", "layer_idx", "must", "nn", "num_heads", "out_proj", "q_proj", "raise", "scaling", "self", "super", "v_proj"], "mvp/modeling_mvp.py:MvpAttention.forward": ["Attention", "Cache", "EncoderDecoderCache", "False", "ModelAttention", "Module", "None", "Tensor", "True", "ValueError", "_", "and", "attention_mask", "attn_output", "attn_probs", "attn_prompt", "attn_weights", "attn_weights_reshaped", "be", "bmm", "bool", "bsz", "but", "cache_position", "cat", "class", "cross_attention_cache", "curr_past_key_values", "current_states", "def", "device", "dim", "dropout", "else", "embed_dim", "expand", "f", "forward", "functional", "get", "head_dim", "hidden_states", "if", "is", "is_cross_attention", "is_updated", "isinstance", "k_proj", "key_states", "key_value_states", "keys", "layer_idx", "layers", "mask", "nn", "not", "num_heads", "of", "out_proj", "output_attentions", "p", "past_key_values", "proj_shape", "prompt_mask", "q_proj", "query_states", "raise", "reshape", "return", "scaling", "self", "self_attention_cache", "should", "size", "softmax", "src_len", "tgt_len", "to", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "values", "view", "weights", "zeros"], "mvp/modeling_mvp.py:MvpEncoderLayer.__init__": ["ACT2FN", "GradientCheckpointingLayer", "LayerNorm", "Linear", "ModelAttention", "ModelConfig", "ModelEncoderLayer", "__init__", "activation_dropout", "activation_fn", "activation_function", "attention_dropout", "class", "config", "d_model", "def", "dropout", "embed_dim", "encoder_attention_heads", "encoder_ffn_dim", "fc1", "fc2", "final_layer_norm", "nn", "num_heads", "self", "self_attn", "self_attn_layer_norm", "super"], "mvp/modeling_mvp.py:MvpEncoderLayer.forward": ["False", "FloatTensor", "GradientCheckpointingLayer", "ModelEncoderLayer", "None", "activation_dropout", "activation_fn", "and", "any", "attention_mask", "attn_prompt", "attn_weights", "bool", "clamp", "clamp_value", "class", "def", "dropout", "dtype", "fc1", "fc2", "final_layer_norm", "finfo", "float16", "forward", "functional", "hidden_states", "if", "isinf", "isnan", "max", "min", "nn", "or", "output_attentions", "p", "residual", "return", "self", "self_attn", "self_attn_layer_norm", "self_attn_prompt", "torch", "training", "tuple"], "mvp/modeling_mvp.py:MvpDecoderLayer.__init__": ["ACT2FN", "GradientCheckpointingLayer", "LayerNorm", "Linear", "ModelAttention", "ModelConfig", "ModelDecoderLayer", "None", "True", "__init__", "activation_dropout", "activation_fn", "activation_function", "attention_dropout", "class", "config", "d_model", "decoder_attention_heads", "decoder_ffn_dim", "def", "dropout", "embed_dim", "encoder_attn", "encoder_attn_layer_norm", "fc1", "fc2", "final_layer_norm", "is_decoder", "layer_idx", "nn", "num_heads", "self", "self_attn", "self_attn_layer_norm", "super"], "mvp/modeling_mvp.py:MvpDecoderLayer.forward": ["Cache", "False", "FloatTensor", "GradientCheckpointingLayer", "ModelDecoderLayer", "None", "Tensor", "True", "activation_dropout", "activation_fn", "attention_mask", "attn_prompt", "bool", "cache_position", "class", "cross_attn_prompt", "cross_attn_weights", "def", "dropout", "encoder_attention_mask", "encoder_attn", "encoder_attn_layer_norm", "encoder_hidden_states", "fc1", "fc2", "final_layer_norm", "forward", "functional", "hidden_states", "if", "is", "key_value_states", "nn", "not", "output_attentions", "outputs", "p", "past_key_values", "residual", "return", "self", "self_attn", "self_attn_layer_norm", "self_attn_prompt", "self_attn_weights", "torch", "training", "tuple", "use_cache"], "mvp/modeling_mvp.py:MvpClassificationHead.__init__": ["Dropout", "Linear", "ModelClassificationHead", "Module", "__init__", "class", "def", "dense", "dropout", "float", "inner_dim", "input_dim", "int", "nn", "num_classes", "out_proj", "p", "pooler_dropout", "self", "super"], "mvp/modeling_mvp.py:MvpClassificationHead.forward": ["ModelClassificationHead", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "nn", "out_proj", "return", "self", "tanh", "torch"], "mvp/modeling_mvp.py:MvpPrompt.__init__": ["Dropout", "Embedding", "GELU", "Linear", "ModelPrompt", "Module", "Sequential", "__init__", "class", "config", "d_model", "def", "dropout", "head_dim", "nn", "num_heads", "num_layers", "p", "prompt_embedding", "prompt_length", "prompt_mid_dim", "prompt_trans", "self", "super"], "mvp/modeling_mvp.py:MvpPrompt.forward": ["ModelPrompt", "Module", "Tensor", "class", "def", "dropout", "forward", "head_dim", "nn", "num_heads", "num_layers", "permute", "prompt", "prompt_embedding", "prompt_ids", "prompt_length", "prompt_trans", "return", "self", "split", "torch", "tuple", "view"], "mvp/modeling_mvp.py:MvpPreTrainedModel._init_weights": ["ModelForConditionalGeneration", "ModelPreTrainedModel", "PreTrainedModel", "_init_weights", "class", "def", "final_logits_bias", "if", "init", "isinstance", "module", "self", "super", "zeros_"], "mvp/modeling_mvp.py:MvpPreTrainedModel.dummy_inputs": ["ModelPreTrainedModel", "PreTrainedModel", "attention_mask", "class", "config", "def", "device", "dummy_inputs", "input_ids", "ne", "pad_token", "pad_token_id", "return", "self", "tensor", "torch"], "mvp/modeling_mvp.py:MvpEncoder.__init__": ["Embedding", "False", "LayerNorm", "ModelConfig", "ModelEncoder", "ModelEncoderLayer", "ModelLearnedPositionalEmbedding", "ModelPreTrainedModel", "ModelPrompt", "ModuleList", "None", "_", "__init__", "bool", "class", "config", "d_model", "def", "dropout", "else", "embed_dim", "embed_positions", "embed_scale", "embed_tokens", "encoder_attention_heads", "encoder_layerdrop", "encoder_layers", "for", "gradient_checkpointing", "if", "in", "layerdrop", "layernorm_embedding", "layers", "math", "max_position_embeddings", "max_source_positions", "nn", "pad_token_id", "padding_idx", "post_init", "prompt_length", "range", "scale_embedding", "self", "self_attn_prompt", "sqrt", "super", "use_prompt", "vocab_size"], "mvp/modeling_mvp.py:MvpEncoder.forward": ["BaseModelOutput", "False", "FloatTensor", "LongTensor", "ModelEncoder", "ModelPreTrainedModel", "None", "Tensor", "True", "ValueError", "You", "_prepare_4d_attention_mask", "all_attentions", "and", "arange", "at", "attention_mask", "attentions", "bool", "both", "cannot", "class", "config", "def", "device", "dropout", "dropout_probability", "dtype", "either", "elif", "else", "embed_pos", "embed_positions", "embed_scale", "embed_tokens", "encoder_layer", "encoder_states", "enumerate", "for", "forward", "functional", "have", "hidden_states", "idx", "if", "in", "input", "input_ids", "input_shape", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layer_outputs", "layerdrop", "layernorm_embedding", "layers", "nn", "not", "or", "output_attentions", "output_hidden_states", "p", "prompt_ids", "prompt_length", "r", "raise", "rand", "return", "return_dict", "same", "self", "self_attn_prompt", "shape", "size", "specify", "the", "time", "to", "to_drop", "torch", "training", "tuple", "use_prompt", "use_return_dict", "v", "view"], "mvp/modeling_mvp.py:MvpDecoder.__init__": ["Embedding", "False", "LayerNorm", "ModelConfig", "ModelDecoder", "ModelDecoderLayer", "ModelLearnedPositionalEmbedding", "ModelPreTrainedModel", "ModelPrompt", "ModuleList", "None", "__init__", "bool", "class", "config", "cross_attn_prompt", "d_model", "decoder_attention_heads", "decoder_layerdrop", "decoder_layers", "def", "dropout", "else", "embed_positions", "embed_scale", "embed_tokens", "for", "gradient_checkpointing", "i", "if", "in", "layer_idx", "layerdrop", "layernorm_embedding", "layers", "math", "max_position_embeddings", "max_target_positions", "nn", "pad_token_id", "padding_idx", "post_init", "prompt_length", "range", "scale_embedding", "self", "self_attn_prompt", "sqrt", "super", "use_prompt", "vocab_size"], "mvp/modeling_mvp.py:MvpDecoder.forward": ["BaseModelOutputWithPastAndCrossAttentions", "Cache", "DynamicCache", "EncoderDecoderCache", "False", "FloatTensor", "LongTensor", "ModelDecoder", "ModelPreTrainedModel", "None", "Setting", "Tensor", "True", "ValueError", "You", "_prepare_4d_attention_mask", "_prepare_4d_causal_attention_mask", "all_cross_attentions", "all_hidden_states", "all_self_attns", "and", "arange", "at", "attention_mask", "attentions", "bool", "both", "cache_position", "cannot", "checkpointing", "class", "config", "continue", "cross_attentions", "cross_attn_prompt", "decoder_input_ids", "decoder_inputs_embeds", "decoder_layer", "def", "device", "dropout", "dropout_probability", "dtype", "either", "elif", "else", "embed_positions", "embed_scale", "embed_tokens", "encoder_attention_mask", "encoder_hidden_states", "enumerate", "for", "forward", "functional", "get_seq_length", "gradient", "gradient_checkpointing", "have", "hidden_states", "idx", "if", "in", "incompatible", "input", "input_ids", "input_shape", "inputs_embeds", "is", "is_encoder_decoder", "kwargs", "last_hidden_state", "layer_outputs", "layerdrop", "layernorm_embedding", "layers", "logger", "nn", "not", "or", "output_attentions", "output_hidden_states", "p", "past_key_values", "past_key_values_length", "positions", "prompt_ids", "prompt_length", "r", "raise", "rand", "return", "return_dict", "same", "self", "self_attn_prompt", "shape", "size", "specify", "tgt_len", "the", "time", "to", "torch", "training", "tuple", "use_cache", "use_prompt", "use_return_dict", "v", "view", "warning_once", "with"], "mvp/modeling_mvp.py:MvpModel.__init__": ["Embedding", "ModelConfig", "ModelDecoder", "ModelEncoder", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "config", "d_model", "decoder", "def", "encoder", "nn", "pad_token_id", "padding_idx", "post_init", "self", "shared", "super", "use_prompt", "vocab_size"], "mvp/modeling_mvp.py:MvpModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "return", "self", "shared"], "mvp/modeling_mvp.py:MvpModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "decoder", "def", "embed_tokens", "encoder", "self", "set_input_embeddings", "shared", "value"], "mvp/modeling_mvp.py:MvpModel.set_lightweight_tuning": ["False", "If", "ModelModel", "ModelPreTrainedModel", "True", "assert", "class", "cross_attn_prompt", "decoder", "def", "encoder", "lightweight", "make", "requires_grad_", "self", "self_attn_prompt", "set_lightweight_tuning", "sure", "that", "to", "tuning", "use", "use_prompt", "want", "you"], "mvp/modeling_mvp.py:MvpModel.forward": ["BaseModelOutput", "Cache", "FloatTensor", "If", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Please", "Seq2SeqModelOutput", "Tensor", "ValueError", "and", "are", "attention_mask", "attentions", "be", "bool", "cache_position", "cannot", "class", "config", "cross_attentions", "decoder", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_input_ids", "decoder_inputs_embeds", "decoder_outputs", "decoder_start_token_id", "def", "either", "elif", "else", "encoder", "encoder_attention_mask", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "isinstance", "kwargs", "last_hidden_state", "len", "list", "no", "not", "or", "output_attentions", "output_hidden_states", "pad_token_id", "pass", "passed", "past_key_values", "r", "raise", "return", "return_dict", "self", "shift_tokens_right", "torch", "tuple", "use_cache", "use_return_dict"], "mvp/modeling_mvp.py:MvpForConditionalGeneration.__init__": ["False", "GenerationMixin", "Linear", "ModelConfig", "ModelForConditionalGeneration", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "d_model", "def", "final_logits_bias", "lm_head", "model", "nn", "num_embeddings", "post_init", "register_buffer", "self", "shared", "super", "torch", "zeros"], "mvp/modeling_mvp.py:MvpForConditionalGeneration.resize_token_embeddings": ["Embedding", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "True", "_resize_final_logits_bias", "bool", "class", "def", "int", "mean_resizing", "new_embeddings", "new_num_tokens", "nn", "pad_to_multiple_of", "resize_token_embeddings", "return", "self", "super"], "mvp/modeling_mvp.py:MvpForConditionalGeneration._resize_final_logits_bias": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "_resize_final_logits_bias", "cat", "class", "def", "device", "dim", "else", "extra_bias", "final_logits_bias", "if", "int", "new_bias", "new_num_tokens", "old_num_tokens", "register_buffer", "self", "shape", "torch", "zeros"], "mvp/modeling_mvp.py:MvpForConditionalGeneration.set_lightweight_tuning": ["False", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "lm_head", "model", "requires_grad_", "self", "set_lightweight_tuning"], "mvp/modeling_mvp.py:MvpForConditionalGeneration.forward": ["Cache", "CrossEntropyLoss", "False", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Seq2SeqLMOutput", "Tensor", "The", "and", "argument", "attention_mask", "bool", "cache_position", "changed", "class", "config", "cross_attentions", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_input_ids", "decoder_inputs_embeds", "decoder_start_token_id", "def", "else", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "final_logits_bias", "forward", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "list", "lm_head", "lm_logits", "logger", "logits", "loss", "loss_fct", "masked_lm_loss", "model", "not", "output", "output_attentions", "output_hidden_states", "outputs", "pad_token_id", "past_key_values", "provided", "r", "return", "return_dict", "self", "shift_tokens_right", "since", "to", "torch", "tuple", "use_cache", "use_return_dict", "view", "vocab_size", "warning"], "mvp/modeling_mvp.py:MvpForConditionalGeneration.prepare_decoder_input_ids_from_labels": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "Tensor", "class", "config", "decoder_start_token_id", "def", "labels", "pad_token_id", "prepare_decoder_input_ids_from_labels", "return", "self", "shift_tokens_right", "torch"], "mvp/modeling_mvp.py:MvpForSequenceClassification.__init__": ["ModelClassificationHead", "ModelConfig", "ModelForSequenceClassification", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "classification_head", "classifier_dropout", "config", "d_model", "def", "kwargs", "model", "num_labels", "post_init", "self", "super"], "mvp/modeling_mvp.py:MvpForSequenceClassification.set_lightweight_tuning": ["False", "ModelForSequenceClassification", "ModelPreTrainedModel", "class", "classification_head", "def", "model", "requires_grad_", "self", "set_lightweight_tuning"], "mvp/modeling_mvp.py:MvpForSequenceClassification.forward": ["All", "BCEWithLogitsLoss", "CrossEntropyLoss", "False", "FloatTensor", "LongTensor", "MSELoss", "ModelForSequenceClassification", "ModelPreTrainedModel", "None", "NotImplementedError", "Passing", "Seq2SeqSequenceClassifierOutput", "Tensor", "ValueError", "__class__", "__name__", "and", "attention_mask", "bool", "class", "classification_head", "config", "cross_attentions", "currently", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_input_ids", "decoder_inputs_embeds", "def", "device", "dtype", "elif", "else", "embeddings", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "eos", "eos_mask", "eos_token_id", "eq", "examples", "f", "for", "forward", "have", "hidden_states", "if", "input", "input_ids", "inputs_embeds", "int", "is", "kwargs", "labels", "len", "list", "logits", "long", "loss", "loss_fct", "model", "multi_label_classification", "must", "not", "num_labels", "number", "of", "or", "output", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "problem_type", "r", "raise", "regression", "return", "return_dict", "same", "self", "sentence_representation", "single_label_classification", "size", "squeeze", "sum", "supported", "the", "to", "tokens", "torch", "tuple", "unique_consecutive", "use_cache", "use_return_dict", "view"], "mvp/modeling_mvp.py:MvpForQuestionAnswering.__init__": ["Linear", "ModelForQuestionAnswering", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "config", "def", "hidden_size", "model", "nn", "num_labels", "post_init", "qa_outputs", "self", "super"], "mvp/modeling_mvp.py:MvpForQuestionAnswering.set_lightweight_tuning": ["False", "ModelForQuestionAnswering", "ModelPreTrainedModel", "class", "def", "model", "qa_outputs", "requires_grad_", "self", "set_lightweight_tuning"], "mvp/modeling_mvp.py:MvpForQuestionAnswering.forward": ["CrossEntropyLoss", "False", "FloatTensor", "LongTensor", "ModelForQuestionAnswering", "ModelPreTrainedModel", "None", "Seq2SeqQuestionAnsweringModelOutput", "Tensor", "and", "attention_mask", "bool", "clamp", "class", "config", "contiguous", "cross_attentions", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_input_ids", "decoder_inputs_embeds", "def", "dim", "else", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "end_logits", "end_loss", "end_positions", "forward", "if", "ignore_index", "ignored_index", "input_ids", "inputs_embeds", "is", "kwargs", "len", "list", "logits", "loss", "loss_fct", "model", "not", "output", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "qa_outputs", "r", "return", "return_dict", "self", "sequence_output", "size", "split", "squeeze", "start_logits", "start_loss", "start_positions", "torch", "total_loss", "tuple", "use_cache", "use_return_dict"], "mvp/modeling_mvp.py:MvpDecoderWrapper.__init__": ["ModelDecoder", "ModelDecoderWrapper", "ModelPreTrainedModel", "__init__", "class", "config", "decoder", "def", "post_init", "self", "super"], "mvp/modeling_mvp.py:MvpDecoderWrapper.forward": ["ModelDecoderWrapper", "ModelPreTrainedModel", "args", "class", "decoder", "def", "forward", "kwargs", "return", "self"], "mvp/modeling_mvp.py:MvpForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "ModelDecoderWrapper", "ModelForCausalLM", "ModelPreTrainedModel", "True", "__init__", "bias", "class", "config", "def", "hidden_size", "is_decoder", "is_encoder_decoder", "lm_head", "model", "nn", "post_init", "self", "super", "vocab_size"], "mvp/modeling_mvp.py:MvpForCausalLM.get_input_embeddings": ["GenerationMixin", "ModelForCausalLM", "ModelPreTrainedModel", "class", "decoder", "def", "embed_tokens", "get_input_embeddings", "model", "return", "self"], "mvp/modeling_mvp.py:MvpForCausalLM.set_input_embeddings": ["GenerationMixin", "ModelForCausalLM", "ModelPreTrainedModel", "class", "decoder", "def", "embed_tokens", "model", "self", "set_input_embeddings", "value"], "mvp/modeling_mvp.py:MvpForCausalLM.set_lightweight_tuning": ["False", "GenerationMixin", "ModelForCausalLM", "ModelPreTrainedModel", "class", "def", "lm_head", "model", "requires_grad_", "self", "set_lightweight_tuning"], "mvp/modeling_mvp.py:MvpForCausalLM.forward": ["Cache", "CausalLMOutputWithCrossAttentions", "CrossEntropyLoss", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForCausalLM", "ModelPreTrainedModel", "None", "Tensor", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "cross_attentions", "decoder", "def", "else", "encoder_attention_mask", "encoder_hidden_states", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "lm_head", "logits", "logits_to_keep", "loss", "loss_fct", "model", "not", "output", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "r", "return", "return_dict", "self", "slice", "slice_indices", "torch", "tuple", "use_cache", "use_return_dict", "view", "vocab_size"], "nanochat/modeling_nanochat.py:NanoChatRMSNorm.__init__": ["ModelRMSNorm", "Module", "__init__", "class", "def", "eps", "float", "nn", "self", "super", "torch"], "nanochat/modeling_nanochat.py:NanoChatRMSNorm._norm": ["ModelRMSNorm", "Module", "True", "_norm", "class", "def", "eps", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "torch", "x"], "nanochat/modeling_nanochat.py:NanoChatRMSNorm.forward": ["ModelRMSNorm", "Module", "_norm", "class", "def", "float", "forward", "nn", "return", "self", "torch", "type_as", "x"], "nanochat/modeling_nanochat.py:NanoChatRMSNorm.extra_repr": ["ModelRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "torch"], "nanochat/modeling_nanochat.py:NanoChatRotaryEmbedding.__init__": ["Callable", "False", "ModelConfig", "ModelRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "nanochat/modeling_nanochat.py:NanoChatRotaryEmbedding.compute_default_rope_parameters": ["ModelConfig", "ModelRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "nanochat/modeling_nanochat.py:NanoChatRotaryEmbedding.forward": ["False", "ModelRotaryEmbedding", "Module", "None", "and", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "nanochat/modeling_nanochat.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cos", "def", "k", "k_embed", "q", "q_embed", "return", "rotate_half", "sin", "unsqueeze", "unsqueeze_dim"], "nanochat/modeling_nanochat.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "nanochat/modeling_nanochat.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value", "value_states"], "nanochat/modeling_nanochat.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "nanochat/modeling_nanochat.py:NanoChatAttention.__init__": ["Linear", "ModelAttention", "ModelConfig", "ModelRMSNorm", "Module", "True", "__init__", "attention_bias", "attention_dropout", "bias", "class", "config", "def", "eps", "getattr", "head_dim", "hidden_size", "int", "is_causal", "k_norm", "k_proj", "layer_idx", "nn", "num_attention_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "q_norm", "q_proj", "rms_norm_eps", "scaling", "self", "super", "v_proj"], "nanochat/modeling_nanochat.py:NanoChatAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "LongTensor", "ModelAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_attn_implementation", "apply_rotary_pos_emb", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "cache_kwargs", "cache_position", "class", "config", "contiguous", "cos", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "head_dim", "hidden_shape", "hidden_states", "if", "input_shape", "is", "k_norm", "k_proj", "key_states", "kwargs", "layer_idx", "nn", "not", "o_proj", "past_key_values", "position_embeddings", "q_norm", "q_proj", "query_states", "reshape", "return", "scaling", "self", "shape", "sin", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "view"], "nanochat/modeling_nanochat.py:NanoChatMLP.__init__": ["ACT2FN", "False", "Linear", "ModelMLP", "Module", "__init__", "activation_fn", "bias", "class", "config", "def", "fc1", "fc2", "hidden_act", "hidden_size", "intermediate_size", "nn", "self", "super"], "nanochat/modeling_nanochat.py:NanoChatMLP.forward": ["ModelMLP", "Module", "Tensor", "activation_fn", "class", "def", "fc1", "fc2", "forward", "hidden_states", "nn", "return", "self", "torch"], "nanochat/modeling_nanochat.py:NanoChatDecoderLayer.__init__": ["GradientCheckpointingLayer", "ModelAttention", "ModelConfig", "ModelDecoderLayer", "ModelMLP", "ModelRMSNorm", "__init__", "class", "config", "def", "eps", "hidden_size", "input_layernorm", "int", "layer_idx", "mlp", "post_attention_layernorm", "rms_norm_eps", "self", "self_attn", "super"], "nanochat/modeling_nanochat.py:NanoChatDecoderLayer.forward": ["Cache", "False", "GradientCheckpointingLayer", "LongTensor", "ModelDecoderLayer", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "attention_mask", "bool", "cache_position", "class", "def", "forward", "hidden_states", "input_layernorm", "kwargs", "mlp", "past_key_values", "position_embeddings", "position_ids", "post_attention_layernorm", "residual", "return", "self", "self_attn", "torch", "tuple", "use_cache"], "nanochat/modeling_nanochat.py:NanoChatPreTrainedModel._init_weights": ["ModelAttention", "ModelPreTrainedModel", "Module", "None", "PreTrainedModel", "_init_weights", "class", "config", "def", "if", "init", "initializer_range", "isinstance", "math", "mean", "module", "nn", "normal_", "num_hidden_layers", "o_proj", "self", "sqrt", "std", "super", "weight"], "nanochat/modeling_nanochat.py:NanoChatModel.__init__": ["Embedding", "False", "ModelConfig", "ModelDecoderLayer", "ModelModel", "ModelPreTrainedModel", "ModelRMSNorm", "ModelRotaryEmbedding", "ModuleList", "__init__", "class", "config", "def", "embed_tokens", "eps", "for", "gradient_checkpointing", "hidden_size", "in", "layer_idx", "layers", "nn", "norm", "num_hidden_layers", "pad_token_id", "padding_idx", "post_init", "range", "rms_norm_eps", "rotary_emb", "self", "super", "vocab_size"], "nanochat/modeling_nanochat.py:NanoChatModel.forward": ["BaseModelOutputWithPast", "Cache", "DynamicCache", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "and", "arange", "attention_mask", "bool", "cache_position", "causal_mask", "class", "config", "create_causal_mask", "decoder_layer", "def", "device", "else", "embed_tokens", "exactly", "for", "forward", "get_seq_length", "hidden_states", "if", "in", "input_embeds", "input_ids", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layers", "must", "norm", "not", "num_hidden_layers", "of", "one", "or", "past_key_values", "past_seen_tokens", "position_embeddings", "position_ids", "raise", "return", "rotary_emb", "self", "shape", "specify", "torch", "unsqueeze", "use_cache"], "nanochat/modeling_nanochat.py:NanoChatForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "ModelForCausalLM", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "vocab_size"], "nanochat/modeling_nanochat.py:NanoChatForCausalLM.forward": ["BaseModelOutputWithPast", "Cache", "CausalLMOutputWithPast", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForCausalLM", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "else", "final_logit_softcapping", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "outputs", "past_key_values", "position_ids", "r", "return", "self", "slice", "slice_indices", "tanh", "torch", "use_cache", "vocab_size"], "nemotron/modeling_nemotron.py:_cast_if_autocast_enabled": ["_cast", "_cast_if_autocast_enabled", "amp", "args", "autocast_mode", "def", "device_type", "else", "get_autocast_dtype", "get_autocast_gpu_dtype", "hasattr", "if", "is_autocast_enabled", "not", "return", "target_dtype", "torch"], "nemotron/modeling_nemotron.py:NemotronLayerNorm1P.__init__": ["LayerNorm", "ModelLayerNorm1P", "None", "Size", "True", "__init__", "bias", "bool", "class", "def", "device", "dtype", "elementwise_affine", "eps", "float", "int", "list", "nn", "normalized_shape", "self", "super"], "nemotron/modeling_nemotron.py:NemotronLayerNorm1P.forward": ["F", "False", "LayerNorm", "ModelLayerNorm1P", "Tensor", "_cast_if_autocast_enabled", "args", "bias", "class", "cpu", "def", "device", "device_type", "else", "enabled", "eps", "forward", "if", "input", "layer_norm", "maybe_autocast", "mps", "nn", "normalized_shape", "return", "self", "type", "weight", "with"], "nemotron/modeling_nemotron.py:NemotronRotaryEmbedding.__init__": ["Callable", "False", "ModelConfig", "ModelRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "nemotron/modeling_nemotron.py:NemotronRotaryEmbedding.compute_default_rope_parameters": ["ModelConfig", "ModelRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "get", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "partial_rotary_factor", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "nemotron/modeling_nemotron.py:NemotronRotaryEmbedding.forward": ["False", "ModelRotaryEmbedding", "Module", "None", "and", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "nemotron/modeling_nemotron.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "nemotron/modeling_nemotron.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cat", "cos", "def", "dim", "k", "k_embed", "k_pass", "q", "q_embed", "q_pass", "return", "rot_dim", "rotate_half", "shape", "sin", "torch", "unsqueeze", "unsqueeze_dim"], "nemotron/modeling_nemotron.py:NemotronMLP.__init__": ["ACT2FN", "Linear", "ModelMLP", "Module", "__init__", "act_fn", "bias", "class", "config", "def", "down_proj", "hidden_act", "hidden_size", "intermediate_size", "mlp_bias", "nn", "self", "super", "up_proj"], "nemotron/modeling_nemotron.py:NemotronMLP.forward": ["ModelMLP", "Module", "act_fn", "class", "def", "down_proj", "forward", "nn", "return", "self", "up_proj", "x"], "nemotron/modeling_nemotron.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "nemotron/modeling_nemotron.py:NemotronAttention.__init__": ["Instantiating", "Linear", "ModelAttention", "ModelConfig", "ModelRotaryEmbedding", "Module", "None", "Please", "True", "__class__", "__init__", "__name__", "a", "and", "attention_bias", "attention_dropout", "bias", "caching", "call", "class", "config", "creating", "def", "during", "errors", "f", "forward", "head_dim", "hidden_size", "if", "int", "is", "is_causal", "k_proj", "layer_idx", "lead", "logger", "make", "max_position_embeddings", "nn", "not", "num_attention_heads", "num_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "partial_rotary_factor", "passing", "provide", "q_proj", "recommended", "rope_parameters", "rotary_emb", "self", "super", "sure", "the", "this", "to", "used", "v_proj", "warning_once", "when", "will", "without"], "nemotron/modeling_nemotron.py:NemotronAttention.forward": ["Cache", "False", "LongTensor", "ModelAttention", "Module", "None", "Tensor", "_", "apply_rotary_pos_emb", "attention_dropout", "attention_mask", "attn_output", "attn_weights", "bool", "bsz", "cache_kwargs", "cache_position", "causal_mask", "class", "contiguous", "cos", "def", "dim", "dropout", "dtype", "float32", "forward", "functional", "head_dim", "hidden_states", "if", "is", "k_proj", "key_states", "layer_idx", "math", "matmul", "nn", "not", "num_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "output_attentions", "p", "past_key_values", "position_embeddings", "position_ids", "q_len", "q_proj", "query_states", "repeat_kv", "reshape", "return", "self", "shape", "sin", "size", "softmax", "sqrt", "to", "torch", "training", "transpose", "tuple", "update", "use_cache", "v_proj", "value_states", "view"], "nemotron/modeling_nemotron.py:NemotronFlashAttention2.__init__": ["ModelAttention", "ModelFlashAttention2", "__init__", "_flash_attn_uses_top_left_mask", "args", "class", "def", "flash_attn_supports_top_left_mask", "kwargs", "self", "super"], "nemotron/modeling_nemotron.py:NemotronFlashAttention2.forward": ["Cache", "False", "LongTensor", "ModelAttention", "ModelFlashAttention2", "None", "StaticCache", "Tensor", "The", "ValueError", "We", "_", "_flash_attention_forward", "_flash_attn_uses_top_left_mask", "_is_quantized", "an", "and", "apply_rotary_pos_emb", "at", "attention_dropout", "attention_mask", "attn_implementation", "attn_output", "attn_weights", "back", "be", "bool", "bsz", "cache", "cache_kwargs", "cache_position", "cast", "casted", "class", "com", "compatible", "config", "contiguous", "cos", "cpu", "def", "device", "device_type", "dropout", "dropout_rate", "dtype", "elif", "else", "embedding", "f", "fact", "flash_attention_2", "float32", "forward", "get_autocast_dtype", "get_autocast_gpu_dtype", "getattr", "github", "hasattr", "have", "head_dim", "hidden", "hidden_states", "https", "huggingface", "if", "implementation", "in", "input", "input_dtype", "is", "is_autocast_enabled", "is_causal", "isinstance", "issue", "k_proj", "key_states", "layer", "layer_idx", "layers", "logger", "make", "mean", "might", "mps", "norm", "not", "num_heads", "num_key_value_heads", "o_proj", "open", "or", "output_attentions", "past_key_values", "position_embeddings", "position_ids", "q_len", "q_proj", "query_states", "raise", "related", "reshape", "return", "sdpa", "seems", "self", "silently", "sin", "size", "sliding_window", "states", "static", "sure", "target_dtype", "the", "this", "time", "to", "torch", "training", "transformers", "transpose", "tuple", "type", "upcasted", "update", "use", "use_cache", "use_top_left_mask", "v_proj", "value_states", "view", "warning_once", "weight", "will", "with", "you"], "nemotron/modeling_nemotron.py:NemotronSdpaAttention.forward": ["Cache", "False", "If", "LongTensor", "ModelAttention", "ModelSdpaAttention", "None", "Tensor", "The", "True", "_", "__class__", "__name__", "and", "apply_rotary_pos_emb", "attention", "attention_dropout", "attention_mask", "attn_implementation", "attn_mask", "attn_output", "be", "bool", "bsz", "cache_kwargs", "cache_position", "causal_mask", "class", "contiguous", "cos", "cuda", "def", "device", "does", "dropout_p", "eager", "else", "f", "forward", "functional", "get", "head_dim", "hidden_states", "if", "is", "is_causal", "k_proj", "key_states", "kwargs", "layer_idx", "loading", "logger", "model", "nn", "not", "num_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "output_attentions", "past_key_values", "please", "position_embeddings", "position_ids", "q_len", "q_proj", "query_states", "repeat_kv", "return", "returned", "scaled_dot_product_attention", "self", "set", "shape", "sin", "size", "support", "the", "to", "torch", "training", "transpose", "tuple", "type", "update", "use_cache", "v_proj", "value_states", "view", "want", "warning_once", "weights", "when", "will", "you"], "nemotron/modeling_nemotron.py:NemotronDecoderLayer.__init__": ["GradientCheckpointingLayer", "ModelConfig", "ModelDecoderLayer", "ModelLayerNorm1P", "ModelMLP", "Model_ATTENTION_CLASSES", "__init__", "_attn_implementation", "class", "config", "def", "eps", "hidden_size", "input_layernorm", "int", "layer_idx", "mlp", "norm_eps", "post_attention_layernorm", "self", "self_attn", "super"], "nemotron/modeling_nemotron.py:NemotronDecoderLayer.forward": ["Cache", "False", "FloatTensor", "GradientCheckpointingLayer", "LongTensor", "ModelDecoderLayer", "None", "Tensor", "attention_mask", "bool", "cache_position", "class", "def", "forward", "hidden_states", "if", "input_layernorm", "kwargs", "mlp", "output_attentions", "outputs", "past_key_values", "position_embeddings", "position_ids", "post_attention_layernorm", "residual", "return", "self", "self_attn", "self_attn_weights", "torch", "tuple", "use_cache"], "nemotron/modeling_nemotron.py:NemotronPreTrainedModel._init_weights": ["ModelLayerNorm1P", "ModelPreTrainedModel", "PreTrainedModel", "_init_weights", "bias", "class", "def", "if", "init", "isinstance", "module", "ones_", "self", "super", "weight", "zeros_"], "nemotron/modeling_nemotron.py:NemotronModel.__init__": ["Embedding", "False", "ModelConfig", "ModelDecoderLayer", "ModelLayerNorm1P", "ModelModel", "ModelPreTrainedModel", "ModelRotaryEmbedding", "ModuleList", "__init__", "class", "config", "def", "embed_tokens", "eps", "for", "gradient_checkpointing", "hidden_size", "in", "layer_idx", "layers", "nn", "norm", "norm_eps", "num_hidden_layers", "pad_token_id", "padding_idx", "post_init", "range", "rotary_emb", "self", "super", "vocab_size"], "nemotron/modeling_nemotron.py:NemotronModel.forward": ["BaseModelOutputWithPast", "Cache", "DynamicCache", "False", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Setting", "Tensor", "True", "ValueError", "You", "_update_causal_mask", "all_hidden_states", "all_self_attns", "and", "arange", "attention_mask", "attentions", "bool", "cache_position", "causal_mask", "checkpointing", "class", "config", "decoder_layer", "def", "device", "else", "embed_tokens", "exactly", "for", "forward", "get_seq_length", "gradient", "gradient_checkpointing", "hidden_states", "if", "in", "incompatible", "input_ids", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layer_outputs", "layers", "logger", "must", "norm", "not", "of", "one", "or", "output_attentions", "output_hidden_states", "past_key_values", "past_seen_tokens", "position_embeddings", "position_ids", "raise", "return", "rotary_emb", "self", "shape", "specify", "torch", "training", "unsqueeze", "use_cache", "warning_once", "with"], "nemotron/modeling_nemotron.py:NemotronModel._update_causal_mask": ["AttentionMaskConverter", "BlockMask", "Cache", "False", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "Union", "_attn_implementation", "_ignore_causal_mask_sdpa", "_prepare_4d_causal_attention_mask_with_cache_position", "_unmask_unattended", "_update_causal_mask", "and", "any", "attention_mask", "batch_size", "bool", "cache_position", "causal_mask", "class", "config", "cuda", "def", "device", "dtype", "else", "finfo", "flash_attention_2", "flex_attention", "get_max_cache_shape", "get_seq_length", "if", "in", "input_tensor", "inputs_embeds", "is", "is_compileable", "is_training", "isinstance", "make_flex_block_causal_mask", "min", "min_dtype", "not", "npu", "output_attentions", "past_key_values", "past_key_values_length", "past_seen_tokens", "return", "sdpa", "self", "sequence_length", "shape", "target_length", "torch", "training", "type", "using_compilable_cache", "xpu"], "nemotron/modeling_nemotron.py:NemotronModel._prepare_4d_causal_attention_mask_with_cache_position": ["ModelModel", "ModelPreTrainedModel", "None", "Tensor", "_prepare_4d_causal_attention_mask_with_cache_position", "and", "arange", "attention_mask", "batch_size", "cache_position", "causal_mask", "class", "clone", "def", "device", "diagonal", "dim", "dtype", "else", "expand", "fill_value", "finfo", "full", "if", "int", "is", "kwargs", "mask_length", "masked_fill", "min", "min_dtype", "not", "padding_mask", "reshape", "return", "sequence_length", "shape", "target_length", "to", "torch", "triu"], "nemotron/modeling_nemotron.py:NemotronForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "ModelForCausalLM", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "vocab_size"], "nemotron/modeling_nemotron.py:NemotronForCausalLM.forward": ["BaseModelOutputWithPast", "Cache", "CausalLMOutputWithPast", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForCausalLM", "ModelPreTrainedModel", "None", "Tensor", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "position_ids", "r", "return", "self", "slice", "slice_indices", "torch", "use_cache", "vocab_size"], "nllb_moe/modeling_nllb_moe.py:NllbMoeScaledWordEmbedding.__init__": ["Embedding", "ModelScaledWordEmbedding", "None", "__init__", "class", "def", "embed_scale", "embedding_dim", "float", "int", "nn", "num_embeddings", "padding_idx", "self", "super"], "nllb_moe/modeling_nllb_moe.py:NllbMoeScaledWordEmbedding.forward": ["Embedding", "ModelScaledWordEmbedding", "Tensor", "class", "def", "embed_scale", "forward", "input_ids", "nn", "return", "self", "super", "torch"], "nllb_moe/modeling_nllb_moe.py:NllbMoeSinusoidalPositionalEmbedding.__init__": ["ModelSinusoidalPositionalEmbedding", "Module", "None", "__init__", "class", "def", "embedding_dim", "int", "make_weights", "nn", "num_positions", "offset", "padding_idx", "self", "super"], "nllb_moe/modeling_nllb_moe.py:NllbMoeSinusoidalPositionalEmbedding.make_weights": ["False", "ModelSinusoidalPositionalEmbedding", "Module", "None", "class", "def", "device", "dtype", "emb_weights", "embedding_dim", "get_embedding", "hasattr", "if", "int", "make_weights", "nn", "num_embeddings", "padding_idx", "persistent", "register_buffer", "self", "to", "weights"], "nllb_moe/modeling_nllb_moe.py:NllbMoeSinusoidalPositionalEmbedding.get_embedding": ["ModelSinusoidalPositionalEmbedding", "Module", "None", "arange", "cat", "class", "cos", "def", "dim", "dtype", "emb", "embedding_dim", "exp", "float", "get_default_dtype", "get_embedding", "half_dim", "if", "int", "int64", "is", "log", "math", "nn", "not", "num_embeddings", "padding_idx", "return", "sin", "to", "torch", "unsqueeze", "view", "zeros"], "nllb_moe/modeling_nllb_moe.py:NllbMoeSinusoidalPositionalEmbedding.forward": ["ModelSinusoidalPositionalEmbedding", "Module", "None", "Tensor", "bsz", "class", "create_position_ids_from_input_ids", "create_position_ids_from_inputs_embeds", "def", "detach", "device", "else", "embedding_dim", "forward", "if", "index_select", "input_ids", "inputs_embeds", "int", "is", "make_weights", "max_pos", "nn", "not", "offset", "padding_idx", "past_key_values_length", "position_ids", "return", "self", "seq_len", "shape", "size", "to", "torch", "view", "weights"], "nllb_moe/modeling_nllb_moe.py:NllbMoeSinusoidalPositionalEmbedding.create_position_ids_from_inputs_embeds": ["ModelSinusoidalPositionalEmbedding", "Module", "arange", "class", "contiguous", "create_position_ids_from_inputs_embeds", "def", "device", "dtype", "expand", "input_shape", "inputs_embeds", "long", "nn", "padding_idx", "past_key_values_length", "position_ids", "return", "sequence_length", "size", "torch", "unsqueeze"], "nllb_moe/modeling_nllb_moe.py:NllbMoeSinusoidalPositionalEmbedding.create_position_ids_from_input_ids": ["ModelSinusoidalPositionalEmbedding", "Module", "class", "create_position_ids_from_input_ids", "cumsum", "def", "dim", "incremental_indices", "input_ids", "int", "long", "mask", "ne", "nn", "padding_idx", "past_key_values_length", "return", "torch", "type_as"], "nllb_moe/modeling_nllb_moe.py:NllbMoeTop2Router.__init__": ["Linear", "ModelConfig", "ModelTop2Router", "Module", "__init__", "batch_prioritized_routing", "bias", "class", "classifier", "config", "def", "dtype", "expert_capacity", "getattr", "hidden_size", "moe_eval_capacity_token_fraction", "nn", "normalize_router_prob_before_dropping", "num_experts", "router_bias", "router_dtype", "router_ignore_padding_tokens", "second_expert_policy", "self", "super", "torch"], "nllb_moe/modeling_nllb_moe.py:NllbMoeTop2Router._cast_classifier": ["CB", "ModelTop2Router", "Module", "SCB", "_cast_classifier", "class", "classifier", "def", "dtype", "hasattr", "if", "nn", "not", "or", "r", "self", "to"], "nllb_moe/modeling_nllb_moe.py:NllbMoeTop2Router.normalize_router_probabilities": ["ModelTop2Router", "Module", "clamp", "class", "def", "denom_s", "dim", "dtype", "eps", "finfo", "min", "nn", "normalize_router_probabilities", "return", "router_probs", "self", "sum", "top_1_mask", "top_1_max_probs", "top_2_mask", "top_2_max_probs", "torch"], "nllb_moe/modeling_nllb_moe.py:NllbMoeTop2Router.route_tokens": ["Gumbel", "LongTensor", "ModelTop2Router", "Module", "None", "Tensor", "True", "and", "argmax", "argsort", "batch_prioritized_routing", "bool", "capacity", "ceil", "class", "cumsum", "def", "device", "dim", "distributions", "dtype", "else", "expert_capacity", "float", "float32", "functional", "gates1", "gates2", "gumbel", "if", "importance_scores", "inf", "input_dtype", "is", "keepdim", "len", "locations1", "locations2", "logits_except_top_1", "lt", "masked_fill", "math", "max", "moe_eval_capacity_token_fraction", "nb_tokens", "nn", "non_padding", "normalize_router_prob_before_dropping", "normalize_router_probabilities", "not", "num_classes", "num_experts", "one_hot", "padding_mask", "rand_like", "random", "repeat", "reshape", "return", "route_tokens", "router_ignore_padding_tokens", "router_logits", "router_probs", "rsample", "sampled", "sampling", "second_expert_policy", "self", "shape", "softmax", "sorted_cumsum1", "sorted_cumsum2", "sorted_top_1_mask", "sorted_top_2_mask", "sum", "to", "top_1_expert_index", "top_1_mask", "top_1_max_probs", "top_2_expert_index", "top_2_mask", "top_2_max_probs", "torch", "training", "transpose", "tuple", "unsqueeze"], "nllb_moe/modeling_nllb_moe.py:NllbMoeTop2Router.forward": ["LongTensor", "ModelTop2Router", "Module", "None", "Tensor", "_cast_classifier", "class", "classifier", "def", "dtype", "forward", "hidden_states", "input_dtype", "nn", "padding_mask", "r", "return", "route_tokens", "router_logits", "router_probs", "self", "to", "top_1_mask", "torch", "tuple"], "nllb_moe/modeling_nllb_moe.py:NllbMoeDenseActDense.__init__": ["ACT2FN", "Dropout", "Linear", "ModelConfig", "ModelDenseActDense", "Module", "__init__", "act", "activation_dropout", "activation_function", "class", "config", "d_model", "def", "dropout", "fc1", "fc2", "ffn_dim", "int", "nn", "self", "super"], "nllb_moe/modeling_nllb_moe.py:NllbMoeDenseActDense.forward": ["ModelDenseActDense", "Module", "Tensor", "act", "and", "class", "def", "dropout", "dtype", "fc1", "fc2", "forward", "hidden_states", "if", "int8", "isinstance", "nn", "return", "self", "to", "torch", "uint8", "weight"], "nllb_moe/modeling_nllb_moe.py:NllbMoeExperts.__init__": ["Dropout", "ModelConfig", "ModelDenseActDense", "ModelExperts", "ModuleDict", "__init__", "class", "config", "def", "expert_", "f", "ffn_dim", "for", "idx", "in", "int", "moe_token_dropout", "nn", "num_experts", "range", "self", "super", "token_dropout"], "nllb_moe/modeling_nllb_moe.py:NllbMoeExperts.forward": ["ModelExperts", "ModuleDict", "None", "Tensor", "class", "current_hidden_states", "current_state", "def", "dim", "dtype", "else", "expert_", "expert_hit", "expert_idx", "expert_mask", "f", "final_hidden_states", "for", "forward", "functional", "greater", "hidden_states", "idx", "if", "in", "index_add_", "moe_token_dropout", "nn", "nonzero", "num_classes", "num_experts", "one_hot", "permute", "reshape", "return", "router_mask", "router_probs", "self", "shape", "squeeze", "sum", "to", "token_dropout", "top_x", "torch", "training", "where", "zeros_like"], "nllb_moe/modeling_nllb_moe.py:NllbMoeSparseMLP.__init__": ["ModelConfig", "ModelExperts", "ModelSparseMLP", "ModelTop2Router", "Module", "__init__", "class", "config", "def", "experts", "ffn_dim", "int", "nn", "num_experts", "router", "self", "super"], "nllb_moe/modeling_nllb_moe.py:NllbMoeSparseMLP.forward": ["ModelSparseMLP", "Module", "None", "Tensor", "_", "batch_size", "class", "def", "experts", "forward", "hidden_dim", "hidden_states", "nn", "padding_mask", "reshape", "return", "router", "router_probs", "self", "sequence_length", "shape", "top_1_mask", "torch", "view"], "nllb_moe/modeling_nllb_moe.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "contiguous", "def", "dim", "dropout", "float", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "shape", "size", "softmax", "torch", "training", "transpose", "value"], "nllb_moe/modeling_nllb_moe.py:NllbMoeAttention.__init__": ["False", "Linear", "ModelAttention", "ModelConfig", "Module", "None", "True", "ValueError", "__init__", "and", "be", "bias", "bool", "by", "class", "config", "def", "divisible", "dropout", "embed_dim", "f", "float", "got", "head_dim", "if", "int", "is_causal", "is_decoder", "k_proj", "layer_idx", "must", "nn", "num_heads", "out_proj", "q_proj", "raise", "scaling", "self", "super", "v_proj"], "nllb_moe/modeling_nllb_moe.py:NllbMoeAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "EncoderDecoderCache", "False", "FlashAttentionKwargs", "ModelAttention", "Module", "None", "Tensor", "True", "Unpack", "_attn_implementation", "and", "attention_interface", "attention_mask", "attn_output", "attn_weights", "bsz", "cache_position", "class", "config", "contiguous", "cross_attention_cache", "curr_past_key_values", "current_states", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "get", "head_dim", "hidden_states", "if", "is", "is_cross_attention", "is_updated", "isinstance", "k_proj", "key_states", "key_value_states", "keys", "kv_input_shape", "kwargs", "layer_idx", "layers", "nn", "not", "out_proj", "past_key_values", "q_input_shape", "q_proj", "query_states", "reshape", "return", "scaling", "self", "self_attention_cache", "shape", "src_len", "tgt_len", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "values", "view"], "nllb_moe/modeling_nllb_moe.py:NllbMoeEncoderLayer.__init__": ["Dropout", "False", "GradientCheckpointingLayer", "LayerNorm", "ModelAttention", "ModelConfig", "ModelDenseActDense", "ModelEncoderLayer", "ModelSparseMLP", "__init__", "activation_dropout", "attention_dropout", "attn_dropout", "bool", "class", "config", "d_model", "def", "dropout", "else", "embed_dim", "encoder_attention_heads", "encoder_ffn_dim", "ff_dropout", "ff_layer_norm", "ffn", "ffn_dim", "if", "int", "is_sparse", "layer_idx", "nn", "not", "num_heads", "self", "self_attn", "self_attn_layer_norm", "super"], "nllb_moe/modeling_nllb_moe.py:NllbMoeEncoderLayer.forward": ["GradientCheckpointingLayer", "ModelEncoderLayer", "Tensor", "TransformersKwargs", "Unpack", "_", "and", "any", "attention_mask", "attn_dropout", "clamp", "clamp_value", "class", "def", "dtype", "else", "ff_dropout", "ff_layer_norm", "ffn", "finfo", "float16", "forward", "hidden_states", "if", "is_sparse", "isinf", "isnan", "kwargs", "max", "min", "or", "residual", "return", "self", "self_attn", "self_attn_layer_norm", "torch"], "nllb_moe/modeling_nllb_moe.py:NllbMoeDecoderLayer.__init__": ["ACT2FN", "Dropout", "False", "GradientCheckpointingLayer", "LayerNorm", "ModelAttention", "ModelConfig", "ModelDecoderLayer", "ModelDenseActDense", "ModelSparseMLP", "None", "True", "__init__", "activation_dropout", "activation_fn", "activation_function", "attention_dropout", "attn_dropout", "bool", "class", "config", "cross_attention", "cross_attention_layer_norm", "d_model", "decoder_attention_heads", "decoder_ffn_dim", "def", "dropout", "else", "embed_dim", "ff_dropout", "ff_layer_norm", "ffn", "ffn_dim", "if", "int", "is_decoder", "is_sparse", "layer_idx", "nn", "not", "num_heads", "self", "self_attn", "self_attn_layer_norm", "super"], "nllb_moe/modeling_nllb_moe.py:NllbMoeDecoderLayer.forward": ["Cache", "GradientCheckpointingLayer", "ModelDecoderLayer", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "and", "any", "attention_mask", "attn_dropout", "cache_position", "clamp", "clamp_value", "class", "cross_attention", "cross_attention_layer_norm", "def", "dtype", "else", "encoder_attention_mask", "encoder_hidden_states", "ff_dropout", "ff_layer_norm", "ffn", "finfo", "float16", "forward", "hidden_states", "if", "is", "is_sparse", "isinf", "key_value_states", "kwargs", "max", "min", "not", "past_key_values", "residual", "return", "self", "self_attn", "self_attn_layer_norm", "torch"], "nllb_moe/modeling_nllb_moe.py:NllbMoePreTrainedModel._init_weights": ["ModelPreTrainedModel", "ModelSinusoidalPositionalEmbedding", "PreTrainedModel", "_init_weights", "class", "copy_", "def", "emb_weights", "embedding_dim", "get_embedding", "if", "init", "isinstance", "module", "num_positions", "offset", "padding_idx", "self", "super", "weights"], "nllb_moe/modeling_nllb_moe.py:NllbMoeEncoder.__init__": ["False", "LayerNorm", "ModelConfig", "ModelEncoder", "ModelEncoderLayer", "ModelPreTrainedModel", "ModelScaledWordEmbedding", "ModelSinusoidalPositionalEmbedding", "ModuleList", "__init__", "append", "class", "config", "d_model", "def", "dropout", "else", "embed_dim", "embed_positions", "embed_scale", "embed_tokens", "encoder_layerdrop", "encoder_layers", "encoder_sparse_step", "for", "gradient_checkpointing", "i", "if", "in", "is_sparse", "layer_idx", "layer_norm", "layerdrop", "layers", "math", "max_position_embeddings", "max_source_positions", "nn", "pad_token_id", "padding_idx", "post_init", "range", "scale_embedding", "self", "sparse_step", "sqrt", "super", "vocab_size"], "nllb_moe/modeling_nllb_moe.py:NllbMoeEncoder.forward": ["MoEModelOutput", "ModelEncoder", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "and", "attention_mask", "class", "config", "continue", "create_bidirectional_mask", "def", "device", "dropout", "dropout_probability", "else", "embed_pos", "embed_positions", "embed_tokens", "encoder_layer", "for", "forward", "functional", "hidden_states", "if", "in", "input_embeds", "input_ids", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layer_norm", "layerdrop", "layers", "nn", "p", "rand", "return", "self", "to", "torch", "training"], "nllb_moe/modeling_nllb_moe.py:NllbMoeDecoder.__init__": ["False", "LayerNorm", "ModelConfig", "ModelDecoder", "ModelDecoderLayer", "ModelPreTrainedModel", "ModelScaledWordEmbedding", "ModelSinusoidalPositionalEmbedding", "ModuleList", "__init__", "append", "class", "config", "d_model", "decoder_layerdrop", "decoder_layers", "decoder_sparse_step", "def", "dropout", "else", "embed_positions", "embed_scale", "embed_tokens", "for", "gradient_checkpointing", "i", "if", "in", "is_sparse", "layer_idx", "layer_norm", "layerdrop", "layers", "math", "max_position_embeddings", "max_target_positions", "nn", "pad_token_id", "padding_idx", "post_init", "range", "scale_embedding", "self", "sparse_step", "sqrt", "super", "vocab_size"], "nllb_moe/modeling_nllb_moe.py:NllbMoeDecoder.forward": ["BaseModelOutputWithPastAndCrossAttentions", "Cache", "DynamicCache", "EncoderDecoderCache", "MoEModelOutputWithPastAndCrossAttentions", "ModelDecoder", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "and", "arange", "attention_mask", "bool", "cache_position", "class", "config", "continue", "create_bidirectional_mask", "create_causal_mask", "decoder_layer", "def", "device", "dropout", "dropout_probability", "else", "embed_positions", "embed_tokens", "encoder_attention_mask", "encoder_hidden_states", "enumerate", "for", "forward", "functional", "get_seq_length", "hidden_states", "idx", "if", "in", "input_embeds", "input_ids", "input_shape", "inputs_embeds", "is", "is_deepspeed_zero3_enabled", "is_fsdp_managed_module", "kwargs", "last_hidden_state", "last_hidden_states", "layer_norm", "layerdrop", "layers", "nn", "not", "or", "p", "past_key_values", "past_key_values_length", "positions", "rand", "return", "self", "size", "skip_the_layer", "synced_gpus", "to", "torch", "training", "tuple", "use_cache"], "nllb_moe/modeling_nllb_moe.py:NllbMoeModel.__init__": ["ModelConfig", "ModelDecoder", "ModelEncoder", "ModelModel", "ModelPreTrainedModel", "ModelScaledWordEmbedding", "__init__", "class", "config", "d_model", "decoder", "def", "else", "embed_scale", "encoder", "if", "math", "pad_token_id", "padding_idx", "post_init", "scale_embedding", "self", "shared", "sqrt", "super", "vocab_size"], "nllb_moe/modeling_nllb_moe.py:NllbMoeModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "return", "self", "shared"], "nllb_moe/modeling_nllb_moe.py:NllbMoeModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "decoder", "def", "embed_tokens", "encoder", "self", "set_input_embeddings", "shared", "value"], "nllb_moe/modeling_nllb_moe.py:NllbMoeModel.forward": ["Cache", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Seq2SeqMoEModelOutput", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "bool", "cache_position", "class", "cross_attentions", "decoder", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_input_ids", "decoder_inputs_embeds", "decoder_outputs", "decoder_router_logits", "def", "encoder", "encoder_attention_mask", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "encoder_router_logits", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "last_hidden_state", "past_key_values", "return", "router_logits", "self", "torch", "tuple", "use_cache"], "nllb_moe/modeling_nllb_moe.py:load_balancing_loss_func": ["Model_balancing_loss_func", "None", "Tensor", "_", "attention_mask", "batch_size", "cat", "compute_device", "concatenated_gate_logits", "def", "device", "dim", "else", "expand", "expert_attention_mask", "expert_mask", "float", "for", "functional", "gate_logits", "if", "in", "int", "is", "isinstance", "layer_gate", "mean", "nn", "not", "num_experts", "num_hidden_layers", "one_hot", "or", "overall_loss", "r", "reshape", "return", "router_per_expert_attention_mask", "router_prob_per_expert", "routing_weights", "selected_experts", "sequence_length", "shape", "softmax", "sum", "to", "tokens_per_expert", "top_k", "topk", "torch", "tuple", "unsqueeze"], "nllb_moe/modeling_nllb_moe.py:shift_tokens_right": ["Model_tokens_right", "Modeled_input_ids", "None", "Tensor", "ValueError", "be", "clone", "config", "decoder_start_token_id", "def", "defined", "has", "if", "input_ids", "int", "is", "masked_fill_", "model", "new_zeros", "pad_token_id", "raise", "return", "self", "shape", "to", "torch"], "nllb_moe/modeling_nllb_moe.py:NllbMoeForConditionalGeneration.__init__": ["False", "GenerationMixin", "Linear", "ModelConfig", "ModelForConditionalGeneration", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "d_model", "def", "lm_head", "model", "nn", "num_experts", "post_init", "router_aux_loss_coef", "router_z_loss_coef", "self", "super", "vocab_size"], "nllb_moe/modeling_nllb_moe.py:NllbMoeForConditionalGeneration.forward": ["Cache", "CrossEntropyLoss", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Seq2SeqMoEOutput", "Tensor", "TransformersKwargs", "Unpack", "and", "attention_mask", "aux_loss", "bool", "cache_position", "class", "config", "cross_attentions", "decoder_attention_mask", "decoder_attentions", "decoder_aux_loss", "decoder_hidden_states", "decoder_input_ids", "decoder_inputs_embeds", "decoder_router_logits", "decoder_start_token_id", "def", "else", "encoder_attentions", "encoder_aux_loss", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "encoder_router_logits", "forward", "if", "ignore_index", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "lm_head", "lm_logits", "load_balancing_loss_func", "logits", "loss", "loss_fct", "model", "not", "num_experts", "output_router_logits", "outputs", "pad_token_id", "past_key_values", "return", "router_aux_loss_coef", "self", "shift_tokens_right", "size", "top_k", "torch", "tuple", "use_cache", "view"], "nystromformer/modeling_nystromformer.py:NystromformerEmbeddings.__init__": ["Dropout", "Embedding", "False", "LayerNorm", "ModelEmbeddings", "Module", "__init__", "arange", "class", "config", "def", "device", "dropout", "dtype", "eps", "expand", "hidden_dropout_prob", "hidden_size", "layer_norm_eps", "long", "max_position_embeddings", "nn", "pad_token_id", "padding_idx", "persistent", "position_embeddings", "position_ids", "register_buffer", "self", "size", "super", "token_type_embeddings", "token_type_ids", "torch", "type_vocab_size", "vocab_size", "word_embeddings", "zeros"], "nystromformer/modeling_nystromformer.py:NystromformerEmbeddings.forward": ["LayerNorm", "ModelEmbeddings", "Module", "None", "buffered_token_type_ids", "buffered_token_type_ids_expanded", "class", "def", "device", "dropout", "dtype", "else", "embeddings", "expand", "forward", "hasattr", "if", "input_ids", "input_shape", "inputs_embeds", "is", "long", "nn", "not", "position_embeddings", "position_ids", "return", "self", "seq_length", "size", "token_type_embeddings", "token_type_ids", "torch", "word_embeddings", "zeros"], "nystromformer/modeling_nystromformer.py:NystromformerSelfAttention.__init__": ["Conv2d", "Dropout", "False", "Linear", "ModelSelfAttention", "Module", "None", "The", "ValueError", "__init__", "a", "all_head_size", "and", "attention", "attention_head_size", "attention_probs_dropout_prob", "bias", "class", "config", "conv", "conv_kernel_size", "def", "dropout", "else", "embedding_size", "f", "groups", "hasattr", "heads", "hidden", "hidden_size", "if", "in_channels", "init_option", "int", "inv_coeff_init_option", "inv_init_coeff_option", "is", "kernel_size", "key", "multiple", "nn", "not", "num_attention_heads", "num_landmarks", "number", "of", "original", "out_channels", "padding", "query", "raise", "segment_means_seq_len", "self", "seq_len", "size", "super", "the", "value"], "nystromformer/modeling_nystromformer.py:NystromformerSelfAttention.iterative_inv": ["ModelSelfAttention", "Module", "None", "_", "class", "def", "device", "dim", "else", "eye", "for", "identity", "if", "in", "init_option", "iterative_inv", "key", "key_value", "mat", "matmul", "max", "n_iter", "nn", "original", "range", "return", "self", "size", "sum", "torch", "transpose", "value", "values"], "nystromformer/modeling_nystromformer.py:NystromformerSelfAttention.forward": ["False", "ModelSelfAttention", "Module", "None", "_", "all_head_size", "attention_head_size", "attention_mask", "attention_probs", "attention_scores", "batch_size", "class", "context_layer", "contiguous", "conv", "conv_kernel_size", "def", "dim", "else", "forward", "functional", "hidden_states", "if", "is", "iterative_inv", "k_landmarks", "kernel_1", "kernel_2", "kernel_3", "key", "key_layer", "math", "matmul", "mean", "new_context_layer_shape", "new_value_layer", "nn", "not", "num_attention_heads", "num_landmarks", "output_attentions", "outputs", "permute", "q_landmarks", "query", "query_layer", "reshape", "return", "self", "seq_len", "seq_length", "shape", "size", "softmax", "sqrt", "torch", "transpose", "value", "value_layer", "view"], "nystromformer/modeling_nystromformer.py:NystromformerSelfOutput.__init__": ["Dropout", "LayerNorm", "Linear", "ModelSelfOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "eps", "hidden_dropout_prob", "hidden_size", "layer_norm_eps", "nn", "self", "super"], "nystromformer/modeling_nystromformer.py:NystromformerSelfOutput.forward": ["LayerNorm", "ModelSelfOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "nystromformer/modeling_nystromformer.py:NystromformerAttention.__init__": ["ModelAttention", "ModelSelfAttention", "ModelSelfOutput", "Module", "__init__", "class", "config", "def", "nn", "output", "self", "super"], "nystromformer/modeling_nystromformer.py:NystromformerAttention.forward": ["False", "ModelAttention", "Module", "None", "attention_mask", "attention_output", "class", "def", "forward", "hidden_states", "nn", "output", "output_attentions", "outputs", "return", "self", "self_outputs"], "nystromformer/modeling_nystromformer.py:NystromformerIntermediate.__init__": ["ACT2FN", "Linear", "ModelIntermediate", "Module", "__init__", "class", "config", "def", "dense", "else", "hidden_act", "hidden_size", "if", "intermediate_act_fn", "intermediate_size", "isinstance", "nn", "self", "str", "super"], "nystromformer/modeling_nystromformer.py:NystromformerIntermediate.forward": ["ModelIntermediate", "Module", "Tensor", "class", "def", "dense", "forward", "hidden_states", "intermediate_act_fn", "nn", "return", "self", "torch"], "nystromformer/modeling_nystromformer.py:NystromformerOutput.__init__": ["Dropout", "LayerNorm", "Linear", "ModelOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "eps", "hidden_dropout_prob", "hidden_size", "intermediate_size", "layer_norm_eps", "nn", "self", "super"], "nystromformer/modeling_nystromformer.py:NystromformerOutput.forward": ["LayerNorm", "ModelOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "nystromformer/modeling_nystromformer.py:NystromformerLayer.__init__": ["GradientCheckpointingLayer", "ModelAttention", "ModelIntermediate", "ModelLayer", "ModelOutput", "__init__", "add_cross_attention", "attention", "chunk_size_feed_forward", "class", "config", "def", "intermediate", "output", "self", "seq_len_dim", "super"], "nystromformer/modeling_nystromformer.py:NystromformerLayer.forward": ["False", "GradientCheckpointingLayer", "ModelLayer", "None", "apply_chunking_to_forward", "attention", "attention_mask", "attention_output", "chunk_size_feed_forward", "class", "def", "feed_forward_chunk", "forward", "hidden_states", "layer_output", "output_attentions", "outputs", "return", "self", "self_attention_outputs", "seq_len_dim"], "nystromformer/modeling_nystromformer.py:NystromformerLayer.feed_forward_chunk": ["GradientCheckpointingLayer", "ModelLayer", "attention_output", "class", "def", "feed_forward_chunk", "intermediate", "intermediate_output", "layer_output", "output", "return", "self"], "nystromformer/modeling_nystromformer.py:NystromformerEncoder.__init__": ["False", "ModelEncoder", "ModelLayer", "Module", "ModuleList", "_", "__init__", "class", "config", "def", "for", "gradient_checkpointing", "in", "layer", "nn", "num_hidden_layers", "range", "self", "super"], "nystromformer/modeling_nystromformer.py:NystromformerEncoder.forward": ["BaseModelOutputWithPastAndCrossAttentions", "False", "ModelEncoder", "Module", "None", "Tensor", "True", "all_hidden_states", "all_self_attentions", "attention_mask", "attentions", "bool", "class", "def", "else", "enumerate", "for", "forward", "hidden_states", "i", "if", "in", "is", "last_hidden_state", "layer", "layer_module", "layer_outputs", "nn", "not", "output_attentions", "output_hidden_states", "return", "return_dict", "self", "torch", "tuple", "v"], "nystromformer/modeling_nystromformer.py:NystromformerPredictionHeadTransform.__init__": ["ACT2FN", "LayerNorm", "Linear", "ModelPredictionHeadTransform", "Module", "__init__", "class", "config", "def", "dense", "else", "eps", "hidden_act", "hidden_size", "if", "isinstance", "layer_norm_eps", "nn", "self", "str", "super", "transform_act_fn"], "nystromformer/modeling_nystromformer.py:NystromformerPredictionHeadTransform.forward": ["LayerNorm", "ModelPredictionHeadTransform", "Module", "Tensor", "class", "def", "dense", "forward", "hidden_states", "nn", "return", "self", "torch", "transform_act_fn"], "nystromformer/modeling_nystromformer.py:NystromformerLMPredictionHead.__init__": ["Linear", "ModelLMPredictionHead", "ModelPredictionHeadTransform", "Module", "Parameter", "True", "__init__", "bias", "class", "config", "decoder", "def", "hidden_size", "nn", "self", "super", "torch", "transform", "vocab_size", "zeros"], "nystromformer/modeling_nystromformer.py:NystromformerLMPredictionHead.forward": ["ModelLMPredictionHead", "Module", "class", "decoder", "def", "forward", "hidden_states", "nn", "return", "self", "transform"], "nystromformer/modeling_nystromformer.py:NystromformerOnlyMLMHead.__init__": ["ModelLMPredictionHead", "ModelOnlyMLMHead", "Module", "__init__", "class", "config", "def", "nn", "predictions", "self", "super"], "nystromformer/modeling_nystromformer.py:NystromformerOnlyMLMHead.forward": ["ModelOnlyMLMHead", "Module", "Tensor", "class", "def", "forward", "nn", "prediction_scores", "predictions", "return", "self", "sequence_output", "torch"], "nystromformer/modeling_nystromformer.py:NystromformerPreTrainedModel._init_weights": ["ModelEmbeddings", "ModelPreTrainedModel", "PreTrainedModel", "_init_weights", "arange", "class", "copy_", "def", "expand", "if", "init", "isinstance", "module", "position_ids", "self", "shape", "super", "token_type_ids", "torch", "zeros_"], "nystromformer/modeling_nystromformer.py:NystromformerModel.__init__": ["ModelEmbeddings", "ModelEncoder", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "config", "def", "embeddings", "encoder", "post_init", "self", "super"], "nystromformer/modeling_nystromformer.py:NystromformerModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "return", "self", "word_embeddings"], "nystromformer/modeling_nystromformer.py:NystromformerModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "self", "set_input_embeddings", "value", "word_embeddings"], "nystromformer/modeling_nystromformer.py:NystromformerModel.forward": ["BaseModelOutputWithPastAndCrossAttentions", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "ValueError", "You", "and", "at", "attention_mask", "attentions", "batch_size", "bool", "both", "buffered_token_type_ids", "buffered_token_type_ids_expanded", "cannot", "class", "config", "cross_attentions", "def", "device", "dtype", "either", "elif", "else", "embedding_output", "embeddings", "encoder", "encoder_outputs", "expand", "extended_attention_mask", "forward", "get_extended_attention_mask", "hasattr", "have", "hidden_states", "if", "input_ids", "input_shape", "inputs_embeds", "is", "kwargs", "last_hidden_state", "long", "not", "ones", "or", "output_attentions", "output_hidden_states", "position_ids", "raise", "return", "return_dict", "same", "self", "seq_length", "sequence_output", "size", "specify", "the", "time", "to", "token_type_ids", "torch", "tuple", "use_return_dict", "warn_if_padding_and_no_attention_mask", "zeros"], "nystromformer/modeling_nystromformer.py:NystromformerForMaskedLM.__init__": ["Model", "ModelForMaskedLM", "ModelModel", "ModelOnlyMLMHead", "ModelPreTrainedModel", "__init__", "class", "cls", "config", "def", "post_init", "self", "super"], "nystromformer/modeling_nystromformer.py:NystromformerForMaskedLM.get_output_embeddings": ["ModelForMaskedLM", "ModelPreTrainedModel", "class", "cls", "decoder", "def", "get_output_embeddings", "predictions", "return", "self"], "nystromformer/modeling_nystromformer.py:NystromformerForMaskedLM.set_output_embeddings": ["ModelForMaskedLM", "ModelPreTrainedModel", "bias", "class", "cls", "decoder", "def", "new_embeddings", "predictions", "self", "set_output_embeddings"], "nystromformer/modeling_nystromformer.py:NystromformerForMaskedLM.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "MaskedLMOutput", "Model", "ModelForMaskedLM", "ModelPreTrainedModel", "None", "Tensor", "attention_mask", "attentions", "bool", "class", "cls", "config", "def", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "masked_lm_loss", "not", "output", "output_attentions", "output_hidden_states", "outputs", "position_ids", "prediction_scores", "r", "return", "return_dict", "self", "sequence_output", "token_type_ids", "torch", "tuple", "use_return_dict", "view", "vocab_size"], "nystromformer/modeling_nystromformer.py:NystromformerClassificationHead.__init__": ["Dropout", "Linear", "ModelClassificationHead", "Module", "__init__", "class", "config", "def", "dense", "dropout", "hidden_dropout_prob", "hidden_size", "nn", "num_labels", "out_proj", "self", "super"], "nystromformer/modeling_nystromformer.py:NystromformerClassificationHead.forward": ["ACT2FN", "ModelClassificationHead", "Module", "class", "config", "def", "dense", "dropout", "features", "forward", "hidden_act", "kwargs", "nn", "out_proj", "return", "self", "x"], "nystromformer/modeling_nystromformer.py:NystromformerForSequenceClassification.__init__": ["Model", "ModelClassificationHead", "ModelForSequenceClassification", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "classifier", "config", "def", "num_labels", "post_init", "self", "super"], "nystromformer/modeling_nystromformer.py:NystromformerForSequenceClassification.forward": ["BCEWithLogitsLoss", "CrossEntropyLoss", "FloatTensor", "LongTensor", "MSELoss", "Model", "ModelForSequenceClassification", "ModelPreTrainedModel", "None", "SequenceClassifierOutput", "Tensor", "and", "attention_mask", "attentions", "bool", "class", "classifier", "config", "def", "dtype", "elif", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "kwargs", "labels", "logits", "long", "loss", "loss_fct", "multi_label_classification", "not", "num_labels", "or", "output", "output_attentions", "output_hidden_states", "outputs", "position_ids", "problem_type", "r", "regression", "return", "return_dict", "self", "sequence_output", "single_label_classification", "squeeze", "token_type_ids", "torch", "tuple", "use_return_dict", "view"], "nystromformer/modeling_nystromformer.py:NystromformerForMultipleChoice.__init__": ["Linear", "Model", "ModelForMultipleChoice", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "classifier", "config", "def", "hidden_size", "nn", "post_init", "pre_classifier", "self", "super"], "nystromformer/modeling_nystromformer.py:NystromformerForMultipleChoice.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "Model", "ModelForMultipleChoice", "ModelPreTrainedModel", "MultipleChoiceModelOutput", "None", "ReLU", "Tensor", "attention_mask", "attentions", "bool", "class", "classifier", "config", "def", "else", "forward", "hidden_state", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "nn", "not", "num_choices", "output", "output_attentions", "output_hidden_states", "outputs", "pooled_output", "position_ids", "pre_classifier", "r", "reshaped_logits", "return", "return_dict", "self", "shape", "size", "token_type_ids", "torch", "tuple", "use_return_dict", "view"], "nystromformer/modeling_nystromformer.py:NystromformerForTokenClassification.__init__": ["Dropout", "Linear", "Model", "ModelForTokenClassification", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "classifier", "config", "def", "dropout", "hidden_dropout_prob", "hidden_size", "nn", "num_labels", "post_init", "self", "super"], "nystromformer/modeling_nystromformer.py:NystromformerForTokenClassification.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "Model", "ModelForTokenClassification", "ModelPreTrainedModel", "None", "Tensor", "TokenClassifierOutput", "attention_mask", "attentions", "bool", "class", "classifier", "config", "def", "dropout", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "not", "num_labels", "output", "output_attentions", "output_hidden_states", "outputs", "position_ids", "r", "return", "return_dict", "self", "sequence_output", "token_type_ids", "torch", "tuple", "use_return_dict", "view"], "nystromformer/modeling_nystromformer.py:NystromformerForQuestionAnswering.__init__": ["Linear", "Model", "ModelForQuestionAnswering", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "config", "def", "hidden_size", "nn", "num_labels", "post_init", "qa_outputs", "self", "super"], "nystromformer/modeling_nystromformer.py:NystromformerForQuestionAnswering.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "Model", "ModelForQuestionAnswering", "ModelPreTrainedModel", "None", "QuestionAnsweringModelOutput", "Tensor", "and", "attention_mask", "attentions", "bool", "clamp", "class", "config", "def", "dim", "else", "end_logits", "end_loss", "end_positions", "forward", "hidden_states", "if", "ignore_index", "ignored_index", "input_ids", "inputs_embeds", "is", "kwargs", "len", "logits", "loss", "loss_fct", "not", "output", "output_attentions", "output_hidden_states", "outputs", "position_ids", "qa_outputs", "return", "return_dict", "self", "sequence_output", "size", "split", "squeeze", "start_logits", "start_loss", "start_positions", "token_type_ids", "torch", "total_loss", "tuple", "use_return_dict"], "olmo/modeling_olmo.py:OlmoLayerNorm.__init__": ["ModelLayerNorm", "Module", "None", "__init__", "class", "def", "hidden_size", "int", "nn", "normalized_shape", "self", "super"], "olmo/modeling_olmo.py:OlmoLayerNorm.forward": ["F", "ModelLayerNorm", "Module", "None", "Tensor", "class", "def", "dtype", "eps", "float32", "forward", "hidden_states", "layer_norm", "nn", "normalized_shape", "orig_dtype", "return", "self", "to", "torch"], "olmo/modeling_olmo.py:OlmoMLP.__init__": ["ACT2FN", "False", "Linear", "ModelMLP", "Module", "__init__", "act_fn", "bias", "class", "config", "def", "down_proj", "gate_proj", "hidden_act", "hidden_size", "intermediate_size", "nn", "self", "super", "up_proj"], "olmo/modeling_olmo.py:OlmoMLP.forward": ["ModelMLP", "Module", "act_fn", "class", "def", "down_proj", "forward", "gate_proj", "nn", "return", "self", "up_proj", "x"], "olmo/modeling_olmo.py:OlmoRotaryEmbedding.__init__": ["Callable", "False", "ModelConfig", "ModelRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "olmo/modeling_olmo.py:OlmoRotaryEmbedding.compute_default_rope_parameters": ["ModelConfig", "ModelRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "olmo/modeling_olmo.py:OlmoRotaryEmbedding.forward": ["False", "ModelRotaryEmbedding", "Module", "None", "and", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "olmo/modeling_olmo.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "olmo/modeling_olmo.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "olmo/modeling_olmo.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value", "value_states"], "olmo/modeling_olmo.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cos", "def", "dtype", "k", "k_embed", "k_type", "q", "q_embed", "q_type", "return", "rotate_half", "sin", "to", "unsqueeze", "unsqueeze_dim"], "olmo/modeling_olmo.py:OlmoAttention.__init__": ["Linear", "ModelAttention", "ModelConfig", "Module", "True", "__init__", "attention_bias", "attention_dropout", "bias", "class", "config", "def", "getattr", "head_dim", "hidden_size", "int", "is_causal", "k_proj", "layer_idx", "nn", "num_attention_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "q_proj", "scaling", "self", "super", "v_proj"], "olmo/modeling_olmo.py:OlmoAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "LongTensor", "ModelAttention", "Module", "None", "Tensor", "_attn_implementation", "apply_rotary_pos_emb", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "cache_kwargs", "cache_position", "clamp_", "class", "clip_qkv", "config", "contiguous", "cos", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "head_dim", "hidden_shape", "hidden_states", "if", "input_shape", "is", "k_proj", "key_states", "kwargs", "layer_idx", "max", "min", "nn", "not", "o_proj", "past_key_values", "position_embeddings", "q_proj", "query_states", "reshape", "return", "scaling", "self", "shape", "sin", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "view"], "olmo/modeling_olmo.py:OlmoDecoderLayer.__init__": ["GradientCheckpointingLayer", "ModelAttention", "ModelConfig", "ModelDecoderLayer", "ModelLayerNorm", "ModelMLP", "__init__", "class", "config", "def", "hidden_size", "input_layernorm", "int", "layer_idx", "mlp", "post_attention_layernorm", "self", "self_attn", "super"], "olmo/modeling_olmo.py:OlmoDecoderLayer.forward": ["Cache", "False", "GradientCheckpointingLayer", "LongTensor", "ModelDecoderLayer", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "attention_mask", "bool", "cache_position", "class", "def", "forward", "hidden_states", "input_layernorm", "kwargs", "mlp", "past_key_values", "position_embeddings", "position_ids", "post_attention_layernorm", "residual", "return", "self", "self_attn", "torch", "tuple", "use_cache"], "olmo/modeling_olmo.py:OlmoModel.__init__": ["Embedding", "False", "ModelConfig", "ModelDecoderLayer", "ModelLayerNorm", "ModelModel", "ModelPreTrainedModel", "ModelRotaryEmbedding", "ModuleList", "__init__", "class", "config", "def", "embed_tokens", "for", "gradient_checkpointing", "hidden_size", "in", "layer_idx", "layers", "nn", "norm", "num_hidden_layers", "pad_token_id", "padding_idx", "post_init", "range", "rotary_emb", "self", "super", "vocab_size"], "olmo/modeling_olmo.py:OlmoModel.forward": ["BaseModelOutputWithPast", "Cache", "DynamicCache", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "and", "arange", "attention_mask", "bool", "cache_position", "causal_mask", "class", "config", "create_causal_mask", "decoder_layer", "def", "device", "else", "embed_tokens", "exactly", "for", "forward", "get_seq_length", "hidden_states", "if", "in", "input_embeds", "input_ids", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layers", "must", "norm", "not", "num_hidden_layers", "of", "one", "or", "past_key_values", "past_seen_tokens", "position_embeddings", "position_ids", "raise", "return", "rotary_emb", "self", "shape", "specify", "torch", "unsqueeze", "use_cache"], "olmo/modeling_olmo.py:OlmoForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "ModelForCausalLM", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "vocab_size"], "olmo/modeling_olmo.py:OlmoForCausalLM.forward": ["BaseModelOutputWithPast", "Cache", "CausalLMOutputWithPast", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForCausalLM", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "outputs", "past_key_values", "position_ids", "r", "return", "self", "slice", "slice_indices", "torch", "use_cache", "vocab_size"], "olmo2/modeling_olmo2.py:Olmo2RMSNorm.__init__": ["ModelRMSNorm", "Module", "Parameter", "__init__", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "olmo2/modeling_olmo2.py:Olmo2RMSNorm.forward": ["ModelRMSNorm", "Module", "True", "class", "def", "dtype", "float32", "forward", "hidden_states", "input_dtype", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "olmo2/modeling_olmo2.py:Olmo2RMSNorm.extra_repr": ["ModelRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "variance_epsilon", "weight"], "olmo2/modeling_olmo2.py:Olmo2RotaryEmbedding.__init__": ["Callable", "False", "ModelConfig", "ModelRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "olmo2/modeling_olmo2.py:Olmo2RotaryEmbedding.compute_default_rope_parameters": ["ModelConfig", "ModelRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "olmo2/modeling_olmo2.py:Olmo2RotaryEmbedding.forward": ["False", "ModelRotaryEmbedding", "Module", "None", "and", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "olmo2/modeling_olmo2.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "olmo2/modeling_olmo2.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value", "value_states"], "olmo2/modeling_olmo2.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cos", "def", "dtype", "k", "k_embed", "k_type", "q", "q_embed", "q_type", "return", "rotate_half", "sin", "to", "unsqueeze", "unsqueeze_dim"], "olmo2/modeling_olmo2.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "olmo2/modeling_olmo2.py:Olmo2Attention.__init__": ["Linear", "ModelAttention", "ModelConfig", "ModelRMSNorm", "Module", "None", "True", "__init__", "attention_bias", "attention_dropout", "bias", "class", "config", "def", "getattr", "head_dim", "hidden_size", "int", "is_causal", "k_norm", "k_proj", "layer_idx", "nn", "num_attention_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "q_norm", "q_proj", "rms_norm_eps", "scaling", "self", "super", "v_proj"], "olmo2/modeling_olmo2.py:Olmo2Attention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "LongTensor", "ModelAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_attn_implementation", "apply_rotary_pos_emb", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "cache_kwargs", "cache_position", "class", "config", "contiguous", "cos", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "head_dim", "hidden_shape", "hidden_states", "if", "input_shape", "is", "k_norm", "k_proj", "key_states", "kwargs", "layer_idx", "nn", "not", "o_proj", "past_key_values", "position_embeddings", "q_norm", "q_proj", "query_states", "reshape", "return", "scaling", "self", "shape", "sin", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "view"], "olmo2/modeling_olmo2.py:Olmo2MLP.__init__": ["ACT2FN", "False", "Linear", "ModelMLP", "Module", "__init__", "act_fn", "bias", "class", "config", "def", "down_proj", "gate_proj", "hidden_act", "hidden_size", "intermediate_size", "nn", "self", "super", "up_proj"], "olmo2/modeling_olmo2.py:Olmo2MLP.forward": ["ModelMLP", "Module", "act_fn", "class", "def", "down_proj", "forward", "gate_proj", "nn", "return", "self", "up_proj", "x"], "olmo2/modeling_olmo2.py:Olmo2DecoderLayer.__init__": ["GradientCheckpointingLayer", "ModelAttention", "ModelConfig", "ModelDecoderLayer", "ModelMLP", "ModelRMSNorm", "__init__", "class", "config", "def", "eps", "hidden_size", "int", "layer_idx", "mlp", "post_attention_layernorm", "post_feedforward_layernorm", "rms_norm_eps", "self", "self_attn", "super"], "olmo2/modeling_olmo2.py:Olmo2DecoderLayer.forward": ["Cache", "False", "GradientCheckpointingLayer", "LongTensor", "ModelDecoderLayer", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "attention_mask", "bool", "cache_position", "class", "def", "forward", "hidden_states", "kwargs", "mlp", "past_key_values", "position_embeddings", "position_ids", "post_attention_layernorm", "post_feedforward_layernorm", "residual", "return", "self", "self_attn", "torch", "tuple", "use_cache"], "olmo2/modeling_olmo2.py:Olmo2Model.__init__": ["Embedding", "False", "ModelConfig", "ModelDecoderLayer", "ModelModel", "ModelPreTrainedModel", "ModelRMSNorm", "ModelRotaryEmbedding", "ModuleList", "__init__", "class", "config", "def", "embed_tokens", "eps", "for", "gradient_checkpointing", "hidden_size", "in", "layer_idx", "layers", "nn", "norm", "num_hidden_layers", "pad_token_id", "padding_idx", "post_init", "range", "rms_norm_eps", "rotary_emb", "self", "super", "vocab_size"], "olmo2/modeling_olmo2.py:Olmo2Model.forward": ["BaseModelOutputWithPast", "Cache", "DynamicCache", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "and", "arange", "attention_mask", "bool", "cache_position", "causal_mask", "class", "config", "create_causal_mask", "decoder_layer", "def", "device", "else", "embed_tokens", "exactly", "for", "forward", "get_seq_length", "hidden_states", "if", "in", "input_embeds", "input_ids", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layers", "must", "norm", "not", "num_hidden_layers", "of", "one", "or", "past_key_values", "past_seen_tokens", "position_embeddings", "position_ids", "raise", "return", "rotary_emb", "self", "shape", "specify", "torch", "unsqueeze", "use_cache"], "olmo2/modeling_olmo2.py:Olmo2ForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "ModelForCausalLM", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "vocab_size"], "olmo2/modeling_olmo2.py:Olmo2ForCausalLM.forward": ["BaseModelOutputWithPast", "Cache", "CausalLMOutputWithPast", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForCausalLM", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "outputs", "past_key_values", "position_ids", "r", "return", "self", "slice", "slice_indices", "torch", "use_cache", "vocab_size"], "olmo3/modeling_olmo3.py:Olmo3RMSNorm.__init__": ["ModelRMSNorm", "Module", "Parameter", "__init__", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "olmo3/modeling_olmo3.py:Olmo3RMSNorm.forward": ["ModelRMSNorm", "Module", "True", "class", "def", "dtype", "float32", "forward", "hidden_states", "input_dtype", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "olmo3/modeling_olmo3.py:Olmo3RMSNorm.extra_repr": ["ModelRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "variance_epsilon", "weight"], "olmo3/modeling_olmo3.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "olmo3/modeling_olmo3.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value", "value_states"], "olmo3/modeling_olmo3.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cos", "def", "dtype", "k", "k_embed", "k_type", "q", "q_embed", "q_type", "return", "rotate_half", "sin", "to", "unsqueeze", "unsqueeze_dim"], "olmo3/modeling_olmo3.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "olmo3/modeling_olmo3.py:Olmo3Attention.__init__": ["Linear", "ModelAttention", "ModelConfig", "ModelRMSNorm", "Module", "None", "True", "__init__", "assert", "attention_bias", "attention_dropout", "attention_type", "bias", "class", "config", "def", "else", "getattr", "head_dim", "hidden_size", "if", "int", "is", "is_causal", "k_norm", "k_proj", "layer_idx", "layer_types", "nn", "not", "num_attention_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "q_norm", "q_proj", "rms_norm_eps", "scaling", "self", "sliding_attention", "sliding_window", "super", "v_proj"], "olmo3/modeling_olmo3.py:Olmo3Attention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "LongTensor", "ModelAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_attn_implementation", "apply_rotary_pos_emb", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "cache_kwargs", "cache_position", "class", "config", "contiguous", "cos", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "head_dim", "hidden_shape", "hidden_states", "if", "input_shape", "is", "k_norm", "k_proj", "key_states", "kwargs", "layer_idx", "nn", "not", "o_proj", "past_key_values", "position_embeddings", "q_norm", "q_proj", "query_states", "reshape", "return", "scaling", "self", "shape", "sin", "sliding_window", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "view"], "olmo3/modeling_olmo3.py:Olmo3MLP.__init__": ["ACT2FN", "False", "Linear", "ModelMLP", "Module", "__init__", "act_fn", "bias", "class", "config", "def", "down_proj", "gate_proj", "hidden_act", "hidden_size", "intermediate_size", "nn", "self", "super", "up_proj"], "olmo3/modeling_olmo3.py:Olmo3MLP.forward": ["ModelMLP", "Module", "act_fn", "class", "def", "down_proj", "forward", "gate_proj", "nn", "return", "self", "up_proj", "x"], "olmo3/modeling_olmo3.py:Olmo3DecoderLayer.__init__": ["GradientCheckpointingLayer", "ModelAttention", "ModelConfig", "ModelDecoderLayer", "ModelMLP", "ModelRMSNorm", "__init__", "class", "config", "def", "eps", "hidden_size", "int", "layer_idx", "mlp", "post_attention_layernorm", "post_feedforward_layernorm", "rms_norm_eps", "self", "self_attn", "super"], "olmo3/modeling_olmo3.py:Olmo3DecoderLayer.forward": ["Cache", "False", "GradientCheckpointingLayer", "LongTensor", "ModelDecoderLayer", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "attention_mask", "bool", "cache_position", "class", "def", "forward", "hidden_states", "kwargs", "mlp", "past_key_values", "position_embeddings", "position_ids", "post_attention_layernorm", "post_feedforward_layernorm", "residual", "return", "self", "self_attn", "torch", "tuple", "use_cache"], "olmo3/modeling_olmo3.py:Olmo3RotaryEmbedding.__init__": ["Callable", "False", "ModelConfig", "ModelRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "olmo3/modeling_olmo3.py:Olmo3RotaryEmbedding.compute_default_rope_parameters": ["ModelConfig", "ModelRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "olmo3/modeling_olmo3.py:Olmo3RotaryEmbedding.forward": ["False", "ModelRotaryEmbedding", "Module", "None", "and", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "olmo3/modeling_olmo3.py:Olmo3Model.__init__": ["Embedding", "False", "ModelConfig", "ModelDecoderLayer", "ModelModel", "ModelPreTrainedModel", "ModelRMSNorm", "ModelRotaryEmbedding", "ModuleList", "__init__", "class", "config", "def", "embed_tokens", "eps", "for", "gradient_checkpointing", "hidden_size", "in", "layer_idx", "layers", "nn", "norm", "num_hidden_layers", "pad_token_id", "padding_idx", "post_init", "range", "rms_norm_eps", "rotary_emb", "self", "super", "vocab_size"], "olmo3/modeling_olmo3.py:Olmo3Model.forward": ["BaseModelOutputWithPast", "Cache", "DynamicCache", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "and", "arange", "attention_mask", "attention_type", "bool", "cache_position", "causal_mask_mapping", "class", "config", "create_causal_mask", "create_sliding_window_causal_mask", "decoder_layer", "def", "device", "dict", "else", "embed_tokens", "exactly", "for", "forward", "full_attention", "get_seq_length", "hidden_states", "if", "in", "input_embeds", "input_ids", "inputs_embeds", "is", "isinstance", "kwargs", "last_hidden_state", "layers", "mask_kwargs", "must", "norm", "not", "num_hidden_layers", "of", "one", "or", "past_key_values", "past_seen_tokens", "position_embeddings", "position_ids", "raise", "return", "rotary_emb", "self", "self_attn", "shape", "sliding_attention", "specify", "torch", "unsqueeze", "use_cache"], "olmo3/modeling_olmo3.py:Olmo3ForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "ModelForCausalLM", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "vocab_size"], "olmo3/modeling_olmo3.py:Olmo3ForCausalLM.forward": ["BaseModelOutputWithPast", "Cache", "CausalLMOutputWithPast", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForCausalLM", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "outputs", "past_key_values", "position_ids", "r", "return", "self", "slice", "slice_indices", "torch", "use_cache", "vocab_size"], "olmoe/modeling_olmoe.py:OlmoeRMSNorm.__init__": ["ModelRMSNorm", "Module", "Parameter", "__init__", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "olmoe/modeling_olmoe.py:OlmoeRMSNorm.forward": ["ModelRMSNorm", "Module", "True", "class", "def", "dtype", "float32", "forward", "hidden_states", "input_dtype", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "olmoe/modeling_olmoe.py:OlmoeRMSNorm.extra_repr": ["ModelRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "variance_epsilon", "weight"], "olmoe/modeling_olmoe.py:OlmoeRotaryEmbedding.__init__": ["Callable", "False", "ModelConfig", "ModelRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "olmoe/modeling_olmoe.py:OlmoeRotaryEmbedding.compute_default_rope_parameters": ["ModelConfig", "ModelRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "olmoe/modeling_olmoe.py:OlmoeRotaryEmbedding.forward": ["False", "ModelRotaryEmbedding", "Module", "None", "and", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "olmoe/modeling_olmoe.py:OlmoeMLP.__init__": ["ACT2FN", "False", "Linear", "ModelMLP", "Module", "__init__", "act_fn", "bias", "class", "config", "def", "down_proj", "gate_proj", "hidden_act", "hidden_size", "intermediate_size", "nn", "self", "super", "up_proj"], "olmoe/modeling_olmoe.py:OlmoeMLP.forward": ["ModelMLP", "Module", "act_fn", "class", "def", "down_proj", "forward", "gate_proj", "nn", "return", "self", "up_proj", "x"], "olmoe/modeling_olmoe.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "olmoe/modeling_olmoe.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cos", "def", "k", "k_embed", "q", "q_embed", "return", "rotate_half", "sin", "unsqueeze", "unsqueeze_dim"], "olmoe/modeling_olmoe.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "olmoe/modeling_olmoe.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value", "value_states"], "olmoe/modeling_olmoe.py:OlmoeAttention.__init__": ["Linear", "ModelAttention", "ModelConfig", "ModelRMSNorm", "Module", "None", "True", "__init__", "attention_bias", "attention_dropout", "bias", "class", "config", "def", "eps", "getattr", "head_dim", "hidden_size", "int", "is_causal", "k_norm", "k_proj", "layer_idx", "nn", "num_attention_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "q_norm", "q_proj", "rms_norm_eps", "scaling", "self", "super", "v_proj"], "olmoe/modeling_olmoe.py:OlmoeAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "LongTensor", "ModelAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_attn_implementation", "apply_rotary_pos_emb", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "cache_kwargs", "cache_position", "clamp_", "class", "clip_qkv", "config", "contiguous", "cos", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "getattr", "head_dim", "hidden_shape", "hidden_states", "if", "input_shape", "is", "k_norm", "k_proj", "key_states", "kwargs", "layer_idx", "max", "min", "nn", "not", "o_proj", "past_key_values", "position_embeddings", "q_norm", "q_proj", "query_states", "reshape", "return", "scaling", "self", "shape", "sin", "sliding_window", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "view"], "olmoe/modeling_olmoe.py:OlmoeExperts.__init__": ["ACT2FN", "ModelConfig", "ModelExperts", "Module", "Parameter", "__init__", "act_fn", "class", "config", "def", "down_proj", "empty", "gate_up_proj", "hidden_act", "hidden_dim", "hidden_size", "intermediate_dim", "intermediate_size", "nn", "num_experts", "num_local_experts", "self", "super", "torch"], "olmoe/modeling_olmoe.py:OlmoeExperts.forward": ["ModelExperts", "Module", "None", "Tensor", "act_fn", "chunk", "class", "continue", "current_hidden_states", "current_state", "def", "dim", "down_proj", "dtype", "expert_hit", "expert_idx", "expert_mask", "final_hidden_states", "for", "forward", "functional", "gate", "gate_up_proj", "greater", "hidden_states", "if", "in", "index_add_", "linear", "nn", "no_grad", "nonzero", "num_classes", "num_experts", "one_hot", "permute", "return", "self", "sum", "to", "token_idx", "top_k_index", "top_k_pos", "top_k_weights", "torch", "up", "where", "with", "zeros_like"], "olmoe/modeling_olmoe.py:OlmoeTopKRouter.__init__": ["ModelTopKRouter", "Module", "Parameter", "__init__", "class", "config", "def", "hidden_dim", "hidden_size", "nn", "norm_topk_prob", "num_experts", "num_experts_per_tok", "self", "super", "top_k", "torch", "weight", "zeros"], "olmoe/modeling_olmoe.py:OlmoeTopKRouter.forward": ["F", "ModelTopKRouter", "Module", "True", "class", "def", "dim", "dtype", "float", "forward", "functional", "hidden_dim", "hidden_states", "if", "keepdim", "linear", "nn", "norm_topk_prob", "reshape", "return", "router_indices", "router_logits", "router_scores", "router_top_value", "self", "softmax", "sum", "to", "top_k", "topk", "torch", "weight"], "olmoe/modeling_olmoe.py:OlmoeSparseMoeBlock.__init__": ["ModelExperts", "ModelSparseMoeBlock", "ModelTopKRouter", "Module", "__init__", "class", "config", "def", "experts", "gate", "nn", "self", "super"], "olmoe/modeling_olmoe.py:OlmoeSparseMoeBlock.forward": ["ModelSparseMoeBlock", "Module", "Tensor", "_", "batch_size", "class", "def", "experts", "final_hidden_states", "forward", "gate", "hidden_dim", "hidden_states", "nn", "reshape", "return", "self", "sequence_length", "shape", "top_k_index", "top_k_weights", "torch", "view"], "olmoe/modeling_olmoe.py:OlmoeDecoderLayer.__init__": ["GradientCheckpointingLayer", "ModelAttention", "ModelConfig", "ModelDecoderLayer", "ModelRMSNorm", "ModelSparseMoeBlock", "__init__", "class", "config", "def", "eps", "hidden_size", "input_layernorm", "int", "layer_idx", "mlp", "post_attention_layernorm", "rms_norm_eps", "self", "self_attn", "super"], "olmoe/modeling_olmoe.py:OlmoeDecoderLayer.forward": ["Cache", "False", "GradientCheckpointingLayer", "LongTensor", "ModelDecoderLayer", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "attention_mask", "bool", "cache_position", "class", "def", "forward", "hidden_states", "input_layernorm", "kwargs", "mlp", "past_key_values", "position_embeddings", "position_ids", "post_attention_layernorm", "residual", "return", "self", "self_attn", "torch", "tuple", "use_cache"], "olmoe/modeling_olmoe.py:OlmoePreTrainedModel._init_weights": ["ModelExperts", "ModelPreTrainedModel", "ModelTopKRouter", "PreTrainedModel", "_init_weights", "class", "config", "def", "down_proj", "elif", "gate_up_proj", "if", "init", "initializer_range", "isinstance", "mean", "module", "normal_", "self", "std", "weight"], "olmoe/modeling_olmoe.py:OlmoeModel.__init__": ["Embedding", "False", "ModelConfig", "ModelDecoderLayer", "ModelModel", "ModelPreTrainedModel", "ModelRMSNorm", "ModelRotaryEmbedding", "ModuleList", "__init__", "class", "config", "def", "embed_tokens", "eps", "for", "gradient_checkpointing", "hidden_size", "in", "layer_idx", "layers", "nn", "norm", "num_hidden_layers", "pad_token_id", "padding_idx", "post_init", "range", "rms_norm_eps", "rotary_emb", "self", "super", "vocab_size"], "olmoe/modeling_olmoe.py:OlmoeModel.forward": ["Cache", "DynamicCache", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "MoeModelOutputWithPast", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "and", "arange", "attention_mask", "bool", "cache_position", "causal_mask", "class", "config", "create_causal_mask", "decoder_layer", "def", "device", "else", "embed_tokens", "exactly", "for", "forward", "get_seq_length", "hidden_states", "if", "in", "input_embeds", "input_ids", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layers", "must", "norm", "not", "num_hidden_layers", "of", "one", "or", "past_key_values", "past_seen_tokens", "position_embeddings", "position_ids", "raise", "return", "rotary_emb", "self", "shape", "specify", "torch", "unsqueeze", "use_cache"], "olmoe/modeling_olmoe.py:load_balancing_loss_func": ["Model_balancing_loss_func", "None", "Tensor", "_", "attention_mask", "batch_size", "cat", "compute_device", "concatenated_gate_logits", "def", "device", "dim", "else", "expand", "expert_attention_mask", "expert_mask", "float", "for", "functional", "gate_logits", "if", "in", "int", "is", "isinstance", "layer_gate", "mean", "nn", "not", "num_experts", "num_hidden_layers", "one_hot", "or", "overall_loss", "r", "reshape", "return", "router_per_expert_attention_mask", "router_prob_per_expert", "routing_weights", "selected_experts", "sequence_length", "shape", "softmax", "sum", "to", "tokens_per_expert", "top_k", "topk", "torch", "tuple", "unsqueeze"], "olmoe/modeling_olmoe.py:OlmoeForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "ModelForCausalLM", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "num_experts", "num_experts_per_tok", "post_init", "router_aux_loss_coef", "self", "super", "vocab_size"], "olmoe/modeling_olmoe.py:OlmoeForCausalLM.forward": ["Cache", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForCausalLM", "ModelPreTrainedModel", "MoeCausalLMOutputWithPast", "MoeModelOutputWithPast", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "aux_loss", "bool", "cache_position", "class", "config", "def", "device", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "load_balancing_loss_func", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "num_experts", "num_experts_per_tok", "output_router_logits", "outputs", "past_key_values", "position_ids", "r", "return", "router_aux_loss_coef", "router_logits", "self", "slice", "slice_indices", "to", "torch", "use_cache", "vocab_size"], "omdet_turbo/modeling_omdet_turbo.py:MultiScaleDeformableAttention.forward": ["False", "ModelScaleDeformableAttention", "Module", "Tensor", "_", "align_corners", "append", "attention_weights", "batch_size", "bilinear", "class", "contiguous", "def", "dim", "enumerate", "flatten", "for", "forward", "functional", "grid_sample", "height", "hidden_dim", "im2col_step", "in", "int", "level_id", "level_start_index", "list", "mode", "nn", "num_heads", "num_levels", "num_points", "num_queries", "output", "padding_mode", "reshape", "return", "sampling_grid_l_", "sampling_grids", "sampling_locations", "sampling_value_l_", "sampling_value_list", "self", "shape", "split", "stack", "sum", "torch", "transpose", "tuple", "value", "value_l_", "value_list", "value_spatial_shapes", "value_spatial_shapes_list", "view", "width", "zeros"], "omdet_turbo/modeling_omdet_turbo.py:OmDetTurboLRUCache.__init__": ["ModelLRUCache", "OrderedDict", "__init__", "cache", "capacity", "class", "current_load", "def", "int", "self"], "omdet_turbo/modeling_omdet_turbo.py:OmDetTurboLRUCache.has": ["ModelLRUCache", "bool", "cache", "class", "def", "has", "in", "key", "return", "self"], "omdet_turbo/modeling_omdet_turbo.py:OmDetTurboLRUCache.get": ["ModelLRUCache", "None", "cache", "class", "def", "get", "if", "in", "key", "move_to_end", "not", "return", "self"], "omdet_turbo/modeling_omdet_turbo.py:OmDetTurboLRUCache.put": ["False", "ModelLRUCache", "None", "cache", "capacity", "class", "current_load", "def", "if", "in", "key", "last", "move_to_end", "not", "popitem", "put", "self", "value"], "omdet_turbo/modeling_omdet_turbo.py:OmDetTurboLanguageBackbone.__init__": ["AutoModel", "ModelConfig", "ModelLanguageBackbone", "Module", "Parameter", "__init__", "class", "config", "def", "from_config", "model", "nn", "self", "super", "text_config", "text_projection", "text_projection_in_dim", "text_projection_out_dim", "torch", "zeros"], "omdet_turbo/modeling_omdet_turbo.py:OmDetTurboLanguageBackbone.forward": ["ModelLanguageBackbone", "Module", "None", "ValueError", "arange", "argmax", "class", "def", "dim", "elif", "else", "encode_type", "encoding", "f", "for", "forward", "hidden_states", "if", "is", "item", "mask", "max", "max_len", "max_pooled_output", "model", "nn", "not", "pooled_output", "projected_output", "raise", "required", "return", "self", "shape", "sum", "supported", "task", "text_outputs", "text_projection", "torch", "transpose", "truncated_mask", "truncated_output"], "omdet_turbo/modeling_omdet_turbo.py:OmDetTurboVisionBackbone.__init__": ["LayerNorm", "ModelConfig", "ModelVisionBackbone", "Module", "ModuleList", "__init__", "apply_layernorm_after_vision_backbone", "class", "config", "def", "encoder_in_channels", "eps", "for", "in", "in_channel_dim", "layer_norm_eps", "layer_norms", "load_backbone", "nn", "self", "super", "vision_backbone"], "omdet_turbo/modeling_omdet_turbo.py:OmDetTurboVisionBackbone.forward": ["ModelVisionBackbone", "Module", "apply_layernorm_after_vision_backbone", "class", "contiguous", "def", "feature_maps", "for", "forward", "if", "in", "layer_norm", "layer_norms", "nn", "output", "outputs", "permute", "pixel_values", "return", "self", "vision_backbone", "zip"], "omdet_turbo/modeling_omdet_turbo.py:OmDetTurboMultiscaleDeformableAttention.__init__": ["CUDA", "Linear", "ModelConfig", "ModelMultiscaleDeformableAttention", "Module", "MultiScaleDeformableAttention", "ValueError", "You", "__init__", "a", "and", "attention", "attention_weights", "attn", "authors", "be", "better", "but", "by", "class", "config", "d", "d_model", "def", "dim_per_head", "dimension", "disable_custom_kernels", "divisible", "each", "efficient", "embed_dim", "f", "got", "head", "if", "im2col_step", "implementation", "in", "int", "is", "make", "more", "must", "n_heads", "n_levels", "n_points", "nn", "not", "num_feature_levels", "num_heads", "of", "output_proj", "power", "raise", "sampling_offsets", "self", "set", "super", "the", "to", "value_proj", "warn", "warnings", "which"], "omdet_turbo/modeling_omdet_turbo.py:OmDetTurboMultiscaleDeformableAttention.with_pos_embed": ["ModelMultiscaleDeformableAttention", "Module", "None", "Tensor", "class", "def", "else", "if", "is", "nn", "position_embeddings", "return", "self", "tensor", "torch", "with_pos_embed"], "omdet_turbo/modeling_omdet_turbo.py:OmDetTurboMultiscaleDeformableAttention.forward": ["F", "False", "Last", "Make", "ModelMultiscaleDeformableAttention", "Module", "None", "Tensor", "ValueError", "_", "align", "attention_mask", "attention_weights", "attn", "batch_size", "be", "bool", "but", "class", "d_model", "def", "dim", "elif", "else", "encoder", "encoder_attention_mask", "encoder_hidden_states", "f", "float", "for", "forward", "got", "hidden", "hidden_states", "if", "im2col_step", "in", "is", "length", "level_start_index", "masked_fill", "must", "n_heads", "n_levels", "n_points", "nn", "not", "num_coordinates", "num_queries", "of", "offset_normalizer", "or", "output", "output_attentions", "output_proj", "position_embeddings", "raise", "reference_points", "return", "sampling_locations", "sampling_offsets", "self", "sequence", "sequence_length", "shape", "shapes", "softmax", "spatial", "spatial_shapes", "spatial_shapes_list", "stack", "states", "sum", "sure", "the", "to", "torch", "total_elements", "value", "value_proj", "view", "with", "with_pos_embed"], "omdet_turbo/modeling_omdet_turbo.py:OmDetTurboConvNormLayer.__init__": ["ACT2CLS", "BatchNorm2d", "Conv2d", "False", "Identity", "ModelConvNormLayer", "Module", "None", "__init__", "activation", "batch_norm_eps", "bias", "class", "config", "conv", "def", "else", "if", "in_channels", "is", "kernel_size", "nn", "norm", "out_channels", "padding", "self", "stride", "super"], "omdet_turbo/modeling_omdet_turbo.py:OmDetTurboConvNormLayer.forward": ["ModelConvNormLayer", "Module", "activation", "class", "conv", "def", "forward", "hidden_state", "nn", "norm", "return", "self"], "omdet_turbo/modeling_omdet_turbo.py:OmDetTurboRepVggBlock.__init__": ["ACT2CLS", "Identity", "ModelConfig", "ModelConvNormLayer", "ModelRepVggBlock", "Module", "None", "__init__", "activation", "class", "config", "conv1", "conv2", "csp_activation", "def", "else", "encoder_hidden_dim", "hidden_channels", "hidden_expansion", "if", "int", "is", "nn", "padding", "self", "super"], "omdet_turbo/modeling_omdet_turbo.py:OmDetTurboRepVggBlock.forward": ["ModelRepVggBlock", "Module", "activation", "class", "conv1", "conv2", "def", "forward", "nn", "return", "self", "x", "y"], "omdet_turbo/modeling_omdet_turbo.py:OmDetTurboCSPRepLayer.__init__": ["Identity", "ModelCSPRepLayer", "ModelConfig", "ModelConvNormLayer", "ModelRepVggBlock", "Module", "Sequential", "_", "__init__", "activation", "bottlenecks", "class", "config", "conv1", "conv2", "conv3", "csp_activation", "def", "else", "encoder_hidden_dim", "for", "hidden_channels", "hidden_expansion", "if", "in", "in_channels", "int", "nn", "num_blocks", "out_channels", "range", "self", "super"], "omdet_turbo/modeling_omdet_turbo.py:OmDetTurboCSPRepLayer.forward": ["ModelCSPRepLayer", "Module", "bottlenecks", "class", "conv1", "conv2", "conv3", "def", "forward", "hidden_state", "hidden_state_1", "hidden_state_2", "nn", "return", "self"], "omdet_turbo/modeling_omdet_turbo.py:OmDetTurboMultiheadAttention.__init__": ["Dropout", "Linear", "ModelMultiheadAttention", "Module", "The", "ValueError", "__init__", "a", "all_head_size", "attention", "attention_head_size", "class", "config", "def", "dropout", "f", "heads", "hidden", "hidden_size", "if", "int", "is", "key", "multiple", "nn", "not", "num_attention_heads", "number", "of", "out_proj", "query", "raise", "self", "size", "super", "the", "value"], "omdet_turbo/modeling_omdet_turbo.py:OmDetTurboMultiheadAttention.forward": ["False", "FloatTensor", "ModelMultiheadAttention", "Module", "None", "Tensor", "_", "all_head_size", "attention_head_size", "attention_mask", "attention_probs", "attention_scores", "batch_size", "bool", "class", "context_layer", "contiguous", "def", "dim", "dropout", "else", "forward", "functional", "if", "is", "key", "key_layer", "keys", "math", "matmul", "new_context_layer_shape", "nn", "not", "num_attention_heads", "out_proj", "output_attentions", "outputs", "permute", "queries", "query", "query_layer", "return", "self", "seq_length", "shape", "size", "softmax", "sqrt", "torch", "transpose", "tuple", "value", "value_layer", "values", "view"], "omdet_turbo/modeling_omdet_turbo.py:OmDetTurboEncoderLayer.__init__": ["ACT2FN", "Dropout", "LayerNorm", "Linear", "ModelConfig", "ModelEncoderLayer", "ModelMultiheadAttention", "Module", "__init__", "activation_fn", "class", "config", "def", "dropout", "encoder_dim_feedforward", "encoder_dropout", "encoder_feedforward_activation", "encoder_feedforward_dropout", "encoder_hidden_dim", "eps", "fc1", "fc2", "final_layer_norm", "hidden_size", "layer_norm_eps", "nn", "num_attention_heads", "self", "self_attn", "self_attn_layer_norm", "super"], "omdet_turbo/modeling_omdet_turbo.py:OmDetTurboEncoderLayer.with_pos_embed": ["ModelEncoderLayer", "Module", "None", "class", "def", "else", "if", "is", "nn", "pos_embed", "return", "tensor", "with_pos_embed"], "omdet_turbo/modeling_omdet_turbo.py:OmDetTurboEncoderLayer.forward": ["False", "ModelEncoderLayer", "Module", "None", "Tensor", "activation_fn", "any", "attention_mask", "attentions", "bool", "clamp", "clamp_value", "class", "def", "dropout", "dtype", "else", "encoder_feedforward_dropout", "fc1", "fc2", "final_layer_norm", "finfo", "forward", "hidden_states", "if", "isinf", "isnan", "key", "keys", "max", "min", "nn", "or", "output_attentions", "position_embeddings", "queries", "query", "residual", "return", "self", "self_attn", "self_attn_layer_norm", "torch", "training", "values", "with_pos_embed"], "omdet_turbo/modeling_omdet_turbo.py:OmDetTurboEncoder.__init__": ["ModelConfig", "ModelEncoder", "ModelEncoderLayer", "Module", "ModuleList", "_", "__init__", "class", "config", "def", "encoder_layers", "for", "in", "layers", "nn", "range", "self", "super"], "omdet_turbo/modeling_omdet_turbo.py:OmDetTurboEncoder.forward": ["False", "ModelEncoder", "Module", "None", "Tensor", "attention", "attention_mask", "bool", "class", "def", "else", "for", "forward", "hidden_states", "if", "in", "layer", "layers", "nn", "output_attentions", "pos_embed", "position_embeddings", "return", "self", "src", "src_mask", "torch", "tuple"], "omdet_turbo/modeling_omdet_turbo.py:OmDetTurboHybridEncoder.__init__": ["BatchNorm2d", "Conv2d", "False", "ModelCSPRepLayer", "ModelConfig", "ModelConvNormLayer", "ModelEncoder", "ModelHybridEncoder", "Module", "ModuleList", "Sequential", "_", "__init__", "activation", "append", "bias", "channel_projection_layers", "class", "config", "conv_norm_activation", "def", "downsample_convs", "encoder", "encoder_hidden_dim", "encoder_in_channels", "encoder_projection_indices", "eval_size", "for", "fpn_blocks", "in", "in_channel", "in_channels", "kernel_size", "lateral_convs", "len", "nn", "out_channels", "pan_blocks", "positional_encoding_temperature", "range", "self", "stride", "super"], "omdet_turbo/modeling_omdet_turbo.py:OmDetTurboHybridEncoder.build_2d_sincos_position_embedding": ["Embed", "ModelHybridEncoder", "Module", "None", "ValueError", "arange", "be", "build_2d_sincos_position_embedding", "by", "class", "concat", "cos", "cpu", "def", "device", "dim", "dimension", "divisible", "dtype", "embed_dim", "embedding", "flatten", "float32", "for", "grid_h", "grid_w", "height", "if", "ij", "indexing", "int", "meshgrid", "must", "nn", "omega", "out_h", "out_w", "pos_dim", "position", "raise", "return", "sin", "temperature", "torch", "width"], "omdet_turbo/modeling_omdet_turbo.py:OmDetTurboHybridEncoder.forward": ["F", "ModelEncoderOutput", "ModelHybridEncoder", "Module", "None", "all_attentions", "append", "attentions", "build_2d_sincos_position_embedding", "channel_projection_layers", "class", "concat", "config", "contiguous", "def", "device", "dim", "downsample_convs", "downsample_feat", "dtype", "else", "encoder", "encoder_hidden_dim", "encoder_layer_index", "encoder_projection_indices", "encoder_states", "enumerate", "eval_size", "extracted_states", "feat_high", "feat_low", "feature", "feature_to_project_index", "flatten", "for", "forward", "fpn_blocks", "fpn_feature_maps", "fpn_states", "fps_map", "height", "hidden_states", "i", "idx", "if", "in", "in_channels", "inputs_embeddings", "insert", "interpolate", "is", "last_hidden_state", "lateral_convs", "layer_outputs", "len", "mode", "nearest", "nn", "not", "or", "output_attentions", "output_hidden_states", "pan_blocks", "permute", "pos_embed", "positional_encoding_temperature", "projected_features", "r", "range", "reshape", "return", "return_dict", "scale_factor", "self", "shape", "src_flatten", "to", "torch", "training", "upsample_feat", "use_return_dict", "width"], "omdet_turbo/modeling_omdet_turbo.py:OmDetTurboMLPWithDropout.__init__": ["ACT2FN", "Dropout", "Linear", "ModelMLPWithDropout", "Module", "__init__", "activation", "class", "class_embed_dim", "config", "decoder_activation", "decoder_dropout", "def", "dropout", "linear1", "linear2", "nn", "self", "super", "task_encoder_hidden_dim"], "omdet_turbo/modeling_omdet_turbo.py:OmDetTurboMLPWithDropout.forward": ["ModelMLPWithDropout", "Module", "activation", "class", "def", "dropout", "forward", "linear1", "linear2", "nn", "return", "self", "x"], "omdet_turbo/modeling_omdet_turbo.py:OmDetTurboMLP.__init__": ["Linear", "ModelMLP", "Module", "ModuleList", "__init__", "class", "def", "for", "hidden_dim", "hidden_layers_dims", "in", "in_dim", "input_dim", "layers", "layers_dims", "nn", "num_layers", "out_dim", "output_dim", "self", "super", "zip"], "omdet_turbo/modeling_omdet_turbo.py:OmDetTurboMLP.forward": ["F", "ModelMLP", "Module", "class", "def", "else", "enumerate", "for", "forward", "i", "if", "in", "layer", "layers", "nn", "num_layers", "relu", "return", "self", "x"], "omdet_turbo/modeling_omdet_turbo.py:OmDetTurboResidualLayer.__init__": ["Dropout", "LayerNorm", "ModelResidualLayer", "Module", "__init__", "class", "class_embed_dim", "config", "decoder_dropout", "def", "dropout", "eps", "layer_norm_eps", "nn", "norm1", "self", "super"], "omdet_turbo/modeling_omdet_turbo.py:OmDetTurboResidualLayer.forward": ["ModelResidualLayer", "Module", "class", "def", "dropout", "forward", "nn", "norm1", "return", "self", "x", "y"], "omdet_turbo/modeling_omdet_turbo.py:OmDetTurboTaskEncoder.__init__": ["ModelMLPWithDropout", "ModelResidualLayer", "ModelTaskEncoder", "Module", "__init__", "class", "config", "def", "mlp", "nn", "res1", "self", "super"], "omdet_turbo/modeling_omdet_turbo.py:OmDetTurboTaskEncoder.forward": ["ModelTaskEncoder", "Module", "class", "def", "forward", "mlp", "mlp_out", "nn", "res1", "return", "self", "x"], "omdet_turbo/modeling_omdet_turbo.py:OmDetTurboDeformableTransformerDecoderLayer.__init__": ["ACT2FN", "Dropout", "GradientCheckpointingLayer", "LayerNorm", "Linear", "ModelDeformableTransformerDecoderLayer", "ModelMultiheadAttention", "ModelMultiscaleDeformableAttention", "__init__", "act", "class", "config", "cross_attn", "decoder_activation", "decoder_dim_feedforward", "decoder_dropout", "decoder_hidden_dim", "decoder_num_heads", "decoder_num_points", "def", "dropout", "dropout1", "dropout2", "dropout3", "dropout4", "eps", "hidden_size", "layer_norm_eps", "linear1", "linear2", "n_points", "nn", "norm1", "norm2", "norm3", "num_attention_heads", "num_heads", "output_attentions", "output_hidden_states", "self", "self_attn", "super"], "omdet_turbo/modeling_omdet_turbo.py:OmDetTurboDeformableTransformerDecoderLayer.with_pos_embed": ["GradientCheckpointingLayer", "ModelDeformableTransformerDecoderLayer", "None", "class", "def", "else", "if", "is", "pos", "return", "tensor", "with_pos_embed"], "omdet_turbo/modeling_omdet_turbo.py:OmDetTurboDeformableTransformerDecoderLayer.forward": ["GradientCheckpointingLayer", "ModelDeformableTransformerDecoderLayer", "None", "act", "attention_mask", "cat", "class", "context", "cross_attention", "cross_attn", "decoder_embeddings", "def", "dim", "dropout1", "dropout2", "dropout3", "dropout4", "else", "encoder_hidden_states", "forward", "hidden_states", "if", "is", "key", "level_start_index", "linear1", "linear2", "norm1", "norm2", "norm3", "not", "origin_embedding_len", "output_attentions", "output_hidden_states", "outputs", "padding_mask", "query", "query_position", "reference_points", "residual", "return", "self", "self_attention", "self_attn", "shape", "spatial_shapes", "spatial_shapes_list", "task_features", "torch", "transpose", "unsqueeze", "vision_features", "vision_shapes", "vision_shapes_list", "with_pos_embed"], "omdet_turbo/modeling_omdet_turbo.py:OmDetTurboPreTrainedModel._init_weights": ["BatchNorm2d", "Conv2d", "LayerNorm", "Linear", "ModelDecoder", "ModelEncoderLayer", "ModelLanguageBackbone", "ModelPreTrainedModel", "None", "PreTrainedModel", "_init_weights", "and", "bias", "bound", "channel_projection_layers", "class", "config", "constant_", "decoder_bbox_head", "def", "elif", "encoder_bbox_head", "encoder_vision_features", "fc1", "fc2", "for", "getattr", "hasattr", "if", "in", "init", "init_std", "is", "isinstance", "layer", "layers", "learn_initial_query", "linear_init_", "math", "mean", "mlp", "module", "module_to_init", "nn", "normal_", "not", "num_batches_tracked", "ones_", "query_position_head", "running_mean", "running_var", "self", "shape", "sqrt", "std", "text_projection", "text_projection_in_dim", "tgt_embed", "uniform_", "weight", "xavier_uniform_", "zeros_"], "omdet_turbo/modeling_omdet_turbo.py:OmDetTurboPreTrainedModel._set_gradient_checkpointing": ["False", "ModelDecoder", "ModelPreTrainedModel", "PreTrainedModel", "_set_gradient_checkpointing", "class", "def", "gradient_checkpointing", "if", "isinstance", "module", "self", "value"], "omdet_turbo/modeling_omdet_turbo.py:OmDetTurboPreTrainedModel._get_cache_key_at_index": ["ModelPreTrainedModel", "PreTrainedModel", "_get_cache_key_at_index", "attention_mask", "cache_key", "class", "def", "index", "input_ids", "input_mask", "return", "tolist", "tuple"], "omdet_turbo/modeling_omdet_turbo.py:OmDetTurboPreTrainedModel.get_cached_class_embeddings": ["ModelPreTrainedModel", "None", "PreTrainedModel", "_", "_get_cache_key_at_index", "append", "cache_key", "class", "classes_attention_mask", "classes_input_ids", "def", "device", "else", "emb", "embeddings", "encode_type", "enumerate", "for", "get", "get_cached_class_embeddings", "has", "idx", "idx_to_put", "if", "in", "language_backbone", "language_cache_class", "not_cached_classes", "not_cached_classes_ids", "not_cached_index", "put", "return", "self", "stack", "to", "torch", "total_class_embs", "total_embeddings"], "omdet_turbo/modeling_omdet_turbo.py:OmDetTurboPreTrainedModel.get_cached_task_embeddings": ["F", "ModelPreTrainedModel", "None", "PreTrainedModel", "_", "_get_cache_key_at_index", "append", "cache_key", "cat", "class", "cur_mask", "def", "device", "dim", "else", "emb", "embeddings", "encode_type", "enumerate", "for", "get", "get_cached_task_embeddings", "has", "idx", "idx_to_put", "if", "in", "language_backbone", "language_cache_prompt", "mask", "masks", "max", "max_len", "not_cached_index", "not_cached_index_ids", "not_cached_mask", "not_cached_tasks", "pad", "pad_size", "put", "range", "return", "self", "shape", "stack", "task", "task_feature", "task_mask", "tasks_attention_mask", "tasks_input_ids", "to", "torch", "total_task_features", "total_task_masks", "unsqueeze"], "omdet_turbo/modeling_omdet_turbo.py:OmDetTurboPreTrainedModel.get_language_embedding": ["F", "ModelPreTrainedModel", "PreTrainedModel", "append", "batched_classes_embeddings", "cat", "class", "class_embeddings", "class_embeddings_regrouped", "classes_attention_mask", "classes_input_ids", "classes_structure", "def", "dim", "for", "get_cached_class_embeddings", "get_cached_task_embeddings", "get_language_embedding", "in", "max", "max_class_size", "pad", "pad_size", "return", "self", "size", "start", "task_embeddings", "task_mask", "tasks_attention_mask", "tasks_input_ids", "torch", "unsqueeze"], "omdet_turbo/modeling_omdet_turbo.py:_cosine_similarity_scaled": ["True", "_cosine_similarity_scaled", "a", "b", "bmm", "clamp_min", "def", "dim", "exp", "keepdim", "logit_scale", "logits_per_image", "norm", "return", "torch"], "omdet_turbo/modeling_omdet_turbo.py:get_class_similarity": ["Exception", "Model_class_similarity", "Unknown", "_cosine_similarity_scaled", "bmm", "class_distance_type", "class_logits", "class_proj", "cls_feature", "cosine", "def", "dot", "elif", "else", "f", "if", "log", "logit_scale", "raise", "return", "tensor", "torch"], "omdet_turbo/modeling_omdet_turbo.py:_inverse_sigmoid": ["_inverse_sigmoid", "clamp", "def", "eps", "log", "max", "min", "return", "torch", "x", "x1", "x2"], "omdet_turbo/modeling_omdet_turbo.py:OmDetTurboDecoder.__init__": ["BatchNorm2d", "Conv2d", "Embedding", "False", "LayerNorm", "Linear", "ModelConfig", "ModelDecoder", "ModelDeformableTransformerDecoderLayer", "ModelMLP", "ModelPreTrainedModel", "ModelTaskEncoder", "ModuleList", "Sequential", "_", "__init__", "bias", "channel_projection_layers", "class", "class_distance_type", "class_embed_dim", "config", "decoder_bbox_head", "decoder_class_head", "decoder_hidden_dim", "decoder_num_layers", "def", "encoder_bbox_head", "encoder_class_head", "encoder_vision_features", "eps", "for", "gradient_checkpointing", "hidden_dim", "if", "in", "input_dim", "layer_norm_eps", "layers", "learn_initial_query", "nn", "num_layers", "num_queries", "output_dim", "post_init", "query_position_head", "range", "self", "super", "task_encoder", "task_project", "tgt_embed", "vision_features_channels", "x"], "omdet_turbo/modeling_omdet_turbo.py:OmDetTurboDecoder.generate_anchors": ["ModelDecoder", "ModelPreTrainedModel", "None", "True", "ValueError", "all", "anchors", "append", "arange", "be", "class", "concat", "cpu", "def", "device", "dtype", "end", "enumerate", "eps", "float32", "for", "generate_anchors", "grid_size", "grid_x", "grid_xy", "grid_y", "height", "if", "ij", "in", "indexing", "inf", "is", "keepdim", "level", "log", "meshgrid", "must", "ones_like", "provided", "raise", "reshape", "return", "self", "spatial_shapes", "stack", "tensor", "torch", "unsqueeze", "valid_mask", "valid_wh", "wh", "where", "width"], "omdet_turbo/modeling_omdet_turbo.py:OmDetTurboDecoder._get_encoder_input": ["ModelDecoder", "ModelPreTrainedModel", "_get_encoder_input", "append", "cat", "channel_projection_layers", "class", "cumsum", "def", "device", "dtype", "enumerate", "feat", "flatten", "for", "height", "i", "in", "int64", "level_start_index", "new_vision_features", "new_vision_shapes", "new_vision_shapes_list", "new_zeros", "permute", "prod", "return", "self", "shape", "tensor", "torch", "vision_features", "width"], "omdet_turbo/modeling_omdet_turbo.py:OmDetTurboDecoder._get_decoder_input": ["ModelDecoder", "ModelPreTrainedModel", "None", "_get_decoder_input", "anchors", "arange", "batch_ind", "batch_size", "cat", "class", "class_distance_type", "class_features", "def", "denoise_bboxes", "denoise_embeddings", "detach", "device", "dim", "dtype", "else", "embeddings", "encoder_bbox_head", "encoder_bboxes", "encoder_class_head", "encoder_class_similarity", "encoder_outputs_bboxes", "encoder_vision_features", "end", "generate_anchors", "get_class_similarity", "if", "indices", "is", "learn_initial_query", "len", "max", "not", "num_queries", "original_class_projected", "permute", "predicted_class_features", "reference_points", "repeat", "return", "self", "sigmoid", "tensor", "tgt_embed", "topk", "topk_ind", "torch", "training", "unsqueeze", "valid_mask", "values", "view", "vision_features", "vision_shapes", "weight", "where"], "omdet_turbo/modeling_omdet_turbo.py:OmDetTurboDecoder.forward": ["ModelDecoder", "ModelDecoderOutput", "ModelPreTrainedModel", "None", "_get_decoder_input", "_get_encoder_input", "_inverse_sigmoid", "_prepare_4d_attention_mask", "all_attns", "all_cross_attns", "all_hidden_states", "all_self_attns", "append", "attention_mask", "attentions", "attn_mask_len", "batch_size", "bool", "break", "class", "class_distance_type", "class_features", "class_proj", "cls_feature", "config", "cross_attention", "decoder_bbox_head", "decoder_bboxes", "decoder_class_head", "decoder_classes", "decoder_coords", "decoder_embeddings", "decoder_num_layers", "def", "denoise_bboxes", "denoise_embeddings", "detach", "device", "dtype", "elif", "else", "encoder_bboxes", "encoder_class_logits", "encoder_class_similarity", "encoder_coord_logits", "enumerate", "for", "forward", "fusion_size", "get_class_similarity", "hidden_states", "i", "if", "in", "init_reference_points", "intermediate_reference_points", "is", "key_padding_mask", "kwargs", "last_hidden_state", "last_refined_bbox", "layer", "layers", "level_start_index", "not", "num_queries", "original_class_projected", "output_attentions", "output_hidden_states", "permute", "predicted_class_features", "query_position", "query_position_head", "reference_points", "refined_bbox", "return", "return_dict", "self", "self_attention", "shape", "sigmoid", "src_key_mask", "stack", "task_encoder", "task_features", "task_mask", "task_project", "to", "torch", "training", "tuple", "use_return_dict", "vision_features", "vision_shapes", "vision_shapes_list", "zeros"], "omdet_turbo/modeling_omdet_turbo.py:OmDetTurboForObjectDetection.__init__": ["ModelConfig", "ModelDecoder", "ModelForObjectDetection", "ModelHybridEncoder", "ModelLRUCache", "ModelLanguageBackbone", "ModelPreTrainedModel", "ModelVisionBackbone", "__init__", "cache_size", "class", "config", "decoder", "def", "encoder", "language_backbone", "language_cache_class", "language_cache_prompt", "num_queries", "post_init", "self", "super", "text_config", "vision_backbone", "vocab_size"], "omdet_turbo/modeling_omdet_turbo.py:OmDetTurboForObjectDetection.get_input_embeddings": ["ModelForObjectDetection", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "language_backbone", "model", "return", "self"], "omdet_turbo/modeling_omdet_turbo.py:OmDetTurboForObjectDetection.set_input_embeddings": ["ModelForObjectDetection", "ModelPreTrainedModel", "class", "def", "language_backbone", "model", "self", "set_input_embeddings", "value"], "omdet_turbo/modeling_omdet_turbo.py:OmDetTurboForObjectDetection.resize_token_embeddings": ["Embedding", "ModelForObjectDetection", "ModelPreTrainedModel", "None", "True", "bool", "class", "config", "def", "int", "language_backbone", "mean_resizing", "model", "model_embeds", "new_num_tokens", "nn", "num_embeddings", "pad_to_multiple_of", "resize_token_embeddings", "return", "self", "text_config", "vocab_size"], "omdet_turbo/modeling_omdet_turbo.py:OmDetTurboForObjectDetection.forward": ["FloatTensor", "LongTensor", "ModelForObjectDetection", "ModelObjectDetectionOutput", "ModelPreTrainedModel", "None", "NotImplementedError", "Training", "attentions", "bool", "class", "class_features", "classes_attention_mask", "classes_input_ids", "classes_structure", "config", "decoder", "decoder_attentions", "decoder_class_logits", "decoder_classes", "decoder_coord_logits", "decoder_coords", "decoder_hidden_states", "decoder_outputs", "def", "else", "encoder", "encoder_attentions", "encoder_class_logits", "encoder_coord_logits", "encoder_extracted_states", "encoder_hidden_states", "encoder_outputs", "extracted_states", "for", "forward", "get_language_embedding", "hidden_states", "if", "image_features", "implemented", "in", "init_reference_points", "intermediate_reference_points", "is", "kwargs", "labels", "loss", "not", "output", "output_attentions", "output_hidden_states", "pixel_values", "r", "raise", "return", "return_dict", "self", "task_features", "task_mask", "tasks_attention_mask", "tasks_input_ids", "torch", "tuple", "use_return_dict", "vision_backbone", "yet"], "oneformer/modeling_oneformer.py:_get_clones": ["ModuleList", "N", "_get_clones", "copy", "deepcopy", "def", "for", "i", "in", "module", "nn", "range", "return"], "oneformer/modeling_oneformer.py:multi_scale_deformable_attention": ["False", "Model_scale_deformable_attention", "Tensor", "_", "align_corners", "append", "attention_weights", "batch_size", "bilinear", "contiguous", "def", "dim", "enumerate", "flatten", "for", "functional", "grid_sample", "height", "hidden_dim", "in", "level_id", "list", "mode", "nn", "num_heads", "num_levels", "num_points", "num_queries", "output", "padding_mode", "reshape", "return", "sampling_grid_l_", "sampling_grids", "sampling_locations", "sampling_value_l_", "sampling_value_list", "shape", "split", "stack", "sum", "torch", "transpose", "tuple", "value", "value_l_", "value_list", "value_spatial_shapes", "view", "width", "zeros"], "oneformer/modeling_oneformer.py:dice_loss": ["Model_loss", "Tensor", "def", "denominator", "flatten", "inputs", "int", "labels", "loss", "num_masks", "numerator", "probs", "r", "return", "sigmoid", "sum"], "oneformer/modeling_oneformer.py:sigmoid_cross_entropy_loss": ["BCEWithLogitsLoss", "Model_cross_entropy_loss", "Tensor", "criterion", "cross_entropy_loss", "def", "inputs", "int", "labels", "loss", "mean", "nn", "none", "num_masks", "r", "reduction", "return", "sum", "torch"], "oneformer/modeling_oneformer.py:pair_wise_dice_loss": ["Model_wise_dice_loss", "None", "T", "Tensor", "def", "denominator", "flatten", "inputs", "labels", "loss", "matmul", "numerator", "return", "sigmoid", "sum", "torch"], "oneformer/modeling_oneformer.py:pair_wise_sigmoid_cross_entropy_loss": ["BCEWithLogitsLoss", "Model_wise_sigmoid_cross_entropy_loss", "T", "Tensor", "criterion", "cross_entropy_loss_neg", "cross_entropy_loss_pos", "def", "height_and_width", "inputs", "labels", "loss", "loss_neg", "loss_pos", "matmul", "nn", "none", "ones_like", "r", "reduction", "return", "shape", "torch", "zeros_like"], "oneformer/modeling_oneformer.py:sample_point": ["False", "Model_point", "Tensor", "True", "add_dim", "def", "dim", "functional", "grid_Model", "if", "input_features", "kwargs", "nn", "point_coordinates", "point_features", "return", "squeeze", "torch", "unsqueeze"], "oneformer/modeling_oneformer.py:OneFormerHungarianMatcher.__init__": ["All", "ModelHungarianMatcher", "Module", "ValueError", "__init__", "and", "be", "can", "class", "cost_class", "cost_dice", "cost_mask", "costs", "def", "float", "if", "int", "nn", "num_points", "raise", "self", "super", "t"], "oneformer/modeling_oneformer.py:OneFormerHungarianMatcher.forward": ["False", "ModelHungarianMatcher", "Module", "NModel", "Tensor", "align_corners", "append", "array", "as_tensor", "assigned_indices", "class", "class_labels", "class_queries_logits", "cost_class", "cost_dice", "cost_mask", "cost_matrix", "cpu", "cuda", "def", "device", "device_type", "dtype", "enabled", "float", "for", "forward", "i", "in", "indices", "int64", "j", "labels", "linear_sum_assignment", "list", "mask_labels", "masks_queries_logits", "matched_indices", "maybe_autocast", "nn", "np", "num_points", "num_queries", "pair_wise_dice_loss", "pair_wise_sigmoid_cross_entropy_loss", "point_coords", "pred_mask", "pred_probs", "preds_masks", "preds_probs", "rand", "repeat", "reshape", "return", "sample_point", "self", "shape", "softmax", "squeeze", "target_mask", "to", "torch", "tuple", "with", "zip"], "oneformer/modeling_oneformer.py:OneFormerLoss.__init__": ["ModelHungarianMatcher", "ModelLoss", "Models", "Module", "NModel", "Parameter", "__init__", "class", "contrastive_temperature", "def", "dict", "empty_weight", "eos_coef", "float", "if", "importance_sample_ratio", "int", "is", "log", "logit_scale", "matcher", "nn", "not", "np", "num_classes", "num_points", "oversample_ratio", "register_buffer", "requires_backends", "scipy", "self", "str", "super", "tensor", "torch", "weight_dict"], "oneformer/modeling_oneformer.py:OneFormerLoss._max_by_axis": ["ModelLoss", "Module", "_max_by_axis", "class", "def", "enumerate", "for", "in", "index", "int", "item", "list", "max", "maxes", "nn", "return", "self", "sublist", "the_list"], "oneformer/modeling_oneformer.py:OneFormerLoss._pad_images_to_max_in_batch": ["False", "ModelLoss", "Models", "Module", "Tensor", "_", "_max_by_axis", "_pad_images_to_max_in_batch", "b", "batch_shape", "batch_size", "bool", "class", "copy_", "def", "device", "dtype", "for", "h", "in", "len", "list", "max_size", "nn", "padded_tensor", "padded_tensors", "padding_mask", "padding_masks", "return", "self", "shape", "tensor", "tensors", "torch", "tuple", "w", "zeros", "zip"], "oneformer/modeling_oneformer.py:OneFormerLoss.loss_contrastive": ["ModelLoss", "Module", "Tensor", "arange", "clamp", "class", "contrastive_queries_logits", "cross_entropy", "def", "device", "dim", "exp", "flatten", "float", "functional", "image_queries", "len", "logit_scale", "logits_per_img", "logits_per_text", "loss_contrastive", "loss_img", "loss_text", "losses", "matmul", "max", "nn", "normalize", "return", "self", "t", "text_queries", "torch"], "oneformer/modeling_oneformer.py:OneFormerLoss.loss_labels": ["CrossEntropyLoss", "ModelLoss", "Module", "Tensor", "_", "_get_predictions_permutation_indices", "array", "batch_size", "cat", "class", "class_labels", "class_queries_logits", "criterion", "def", "device", "dict", "dtype", "empty_weight", "fill_value", "for", "full", "idx", "in", "indices", "int64", "j", "list", "loss_ce", "loss_cross_entropy", "loss_labels", "losses", "nn", "np", "num_classes", "num_queries", "pred_logits", "pred_logits_transposed", "return", "self", "shape", "str", "target", "target_classes", "target_classes_o", "torch", "transpose", "tuple", "weight", "zip"], "oneformer/modeling_oneformer.py:OneFormerLoss.loss_masks": ["False", "ModelLoss", "Module", "NModel", "Tensor", "_", "_get_predictions_permutation_indices", "_get_targets_permutation_indices", "_pad_images_to_max_in_batch", "align_corners", "array", "calculate_uncertainty", "class", "def", "del", "dice_loss", "dict", "importance_sample_ratio", "indices", "int", "list", "loss_dice", "loss_mask", "loss_masks", "losses", "mask_labels", "masks_queries_logits", "nn", "no_grad", "np", "num_masks", "num_points", "oversample_ratio", "point_coords", "point_labels", "point_logits", "pred_masks", "return", "sample_point", "sample_points_using_uncertainty", "self", "sigmoid_cross_entropy_loss", "squeeze", "src_idx", "str", "target_masks", "tgt_idx", "torch", "tuple", "with"], "oneformer/modeling_oneformer.py:OneFormerLoss.calculate_uncertainty": ["ModelLoss", "Module", "Tensor", "abs", "calculate_uncertainty", "class", "def", "logits", "nn", "return", "self", "torch", "uncertainty_scores"], "oneformer/modeling_oneformer.py:OneFormerLoss.sample_points_using_uncertainty": ["False", "ModelLoss", "Module", "NModel", "Tensor", "align_corners", "arange", "cat", "class", "def", "device", "dim", "dtype", "float", "idx", "if", "importance_sample_ratio", "int", "k", "logits", "long", "nn", "num_boxes", "num_points", "num_points_sampled", "num_random_points", "num_uncertain_points", "oversample_ratio", "point_coordinates", "point_logits", "point_uncertainties", "rand", "return", "sample_point", "sample_points_using_uncertainty", "self", "shape", "shift", "topk", "torch", "uncertainty_function", "view"], "oneformer/modeling_oneformer.py:OneFormerLoss._get_predictions_permutation_indices": ["ModelLoss", "Module", "_", "_get_predictions_permutation_indices", "batch_indices", "cat", "class", "def", "enumerate", "for", "full_like", "i", "in", "indices", "nn", "predictions_indices", "return", "self", "src", "torch"], "oneformer/modeling_oneformer.py:OneFormerLoss._get_targets_permutation_indices": ["ModelLoss", "Module", "_", "_get_targets_permutation_indices", "batch_indices", "cat", "class", "def", "enumerate", "for", "full_like", "i", "in", "indices", "nn", "return", "self", "target_indices", "tgt", "torch"], "oneformer/modeling_oneformer.py:OneFormerLoss.forward": ["False", "ModelLoss", "Module", "NModel", "Tensor", "True", "_", "aux_outputs", "auxiliary_predictions", "bool", "calculate_contrastive_loss", "class", "class_labels", "class_queries_logits", "contrastive_queries_logits", "def", "device", "dict", "enumerate", "f", "for", "forward", "get_num_masks", "idx", "if", "in", "indices", "is", "items", "key", "list", "loss_contrastive", "loss_dict", "loss_labels", "loss_masks", "losses", "mask_labels", "masks_queries_logits", "matcher", "nn", "not", "num_masks", "return", "self", "str", "text_queries", "update", "value"], "oneformer/modeling_oneformer.py:OneFormerLoss.get_num_masks": ["ModelLoss", "Module", "PartialState", "Tensor", "_shared_state", "as_tensor", "clamp", "class", "class_labels", "classes", "def", "device", "dtype", "float", "for", "get_num_masks", "if", "in", "is_accelerate_available", "len", "min", "nn", "num_masks", "num_processes", "reduce", "return", "self", "sum", "torch", "world_size"], "oneformer/modeling_oneformer.py:OneFormerPixelDecoderEncoderMultiscaleDeformableAttention.__init__": ["CUDA", "DeformableDetrMultiscaleDeformableAttention", "Linear", "ModelPixelDecoderEncoderMultiscaleDeformableAttention", "Module", "ValueError", "You", "__init__", "a", "and", "attention", "attention_weights", "authors", "be", "better", "but", "by", "class", "d", "d_model", "def", "dim_per_head", "dimension", "divisible", "each", "efficient", "embed_dim", "f", "got", "head", "if", "im2col_step", "implementation", "in", "int", "is", "make", "more", "must", "n_heads", "n_levels", "n_points", "nn", "not", "num_heads", "of", "output_proj", "power", "raise", "sampling_offsets", "self", "set", "super", "the", "to", "value_proj", "warn", "warnings", "which"], "oneformer/modeling_oneformer.py:OneFormerPixelDecoderEncoderMultiscaleDeformableAttention.with_pos_embed": ["ModelPixelDecoderEncoderMultiscaleDeformableAttention", "Module", "NModel", "Tensor", "class", "def", "else", "if", "is", "nn", "position_embeddings", "return", "self", "tensor", "torch", "with_pos_embed"], "oneformer/modeling_oneformer.py:OneFormerPixelDecoderEncoderMultiscaleDeformableAttention.forward": ["False", "Last", "Make", "ModelPixelDecoderEncoderMultiscaleDeformableAttention", "Module", "NModel", "Tensor", "ValueError", "_", "align", "attention_mask", "attention_weights", "batch_size", "be", "bool", "but", "class", "d_model", "def", "dim", "elif", "else", "encoder", "encoder_attention_mask", "encoder_hidden_states", "f", "float", "forward", "functional", "got", "hidden", "hidden_states", "if", "is", "length", "level_start_index", "masked_fill", "multi_scale_deformable_attention", "must", "n_heads", "n_levels", "n_points", "nn", "not", "num_queries", "of", "offset_normalizer", "or", "output", "output_attentions", "output_proj", "position_embeddings", "raise", "reference_points", "return", "sampling_locations", "sampling_offsets", "self", "sequence", "sequence_length", "shape", "shapes", "softmax", "spatial", "spatial_shapes", "stack", "states", "sum", "sure", "the", "to", "torch", "value", "value_proj", "view", "with", "with_pos_embed"], "oneformer/modeling_oneformer.py:OneFormerPixelDecoderEncoderLayer.__init__": ["LayerNorm", "Linear", "ModelConfig", "ModelPixelDecoderEncoderLayer", "ModelPixelDecoderEncoderMultiscaleDeformableAttention", "Module", "__init__", "activation_dropout", "activation_fn", "class", "config", "conv_dim", "def", "dropout", "embed_dim", "encoder_feedforward_dim", "eps", "fc1", "fc2", "final_layer_norm", "functional", "is_training", "layer_norm_eps", "n_levels", "n_points", "nn", "num_attention_heads", "num_heads", "relu", "self", "self_attn", "self_attn_layer_norm", "super"], "oneformer/modeling_oneformer.py:OneFormerPixelDecoderEncoderLayer.forward": ["False", "ModelPixelDecoderEncoderLayer", "Module", "NModel", "Tensor", "activation_dropout", "activation_fn", "any", "attention_mask", "attn_weights", "bool", "clamp", "clamp_value", "class", "def", "dropout", "dtype", "encoder_attention_mask", "encoder_hidden_states", "fc1", "fc2", "final_layer_norm", "finfo", "forward", "functional", "hidden_states", "if", "is_training", "isinf", "isnan", "level_start_index", "max", "min", "nn", "or", "output_attentions", "outputs", "p", "position_embeddings", "reference_points", "residual", "return", "self", "self_attn", "self_attn_layer_norm", "spatial_shapes", "torch", "training"], "oneformer/modeling_oneformer.py:OneFormerPixelDecoderEncoderOnly.__init__": ["ModelConfig", "ModelPixelDecoderEncoderLayer", "ModelPixelDecoderEncoderOnly", "Module", "ModuleList", "_", "__init__", "class", "config", "def", "dropout", "encoder_layers", "for", "in", "layers", "nn", "range", "self", "super"], "oneformer/modeling_oneformer.py:OneFormerPixelDecoderEncoderOnly.get_reference_points": ["ModelPixelDecoderEncoderOnly", "Module", "NModel", "append", "cat", "class", "def", "device", "dtype", "enumerate", "for", "get_reference_points", "height", "ij", "in", "indexing", "linspace", "lvl", "meshgrid", "nn", "ref", "ref_x", "ref_y", "reference_points", "reference_points_list", "reshape", "return", "spatial_shapes", "stack", "torch", "valid_ratios", "width"], "oneformer/modeling_oneformer.py:OneFormerPixelDecoderEncoderOnly.forward": ["BaseModelOutput", "ModelPixelDecoderEncoderOnly", "Module", "NModel", "all_attentions", "attention_mask", "attentions", "class", "config", "def", "device", "else", "encoder_layer", "encoder_states", "enumerate", "for", "forward", "get_reference_points", "hidden_states", "i", "if", "in", "inputs_embeds", "is", "last_hidden_state", "layer_outputs", "layers", "level_start_index", "nn", "not", "output_attentions", "output_hidden_states", "position_embeddings", "r", "reference_points", "return", "return_dict", "self", "spatial_shapes", "use_return_dict", "valid_ratios"], "oneformer/modeling_oneformer.py:OneFormerPixelDecoder.__init__": ["Conv2d", "False", "GroupNorm", "ModelConfig", "ModelPixelDecoder", "ModelPixelDecoderEncoderOnly", "ModelSinePositiModelmbedding", "Module", "ModuleList", "Parameter", "ReLU", "Sequential", "Tensor", "True", "__init__", "adapter_", "add_module", "append", "bias", "class", "common_stride", "config", "conv_dim", "def", "else", "encoder", "enumerate", "f", "feature_channels", "for", "idx", "if", "in", "in_channels", "input_projections", "input_projections_list", "int", "kernel_size", "lateral_conv", "lateral_convs", "layer_", "level_embed", "log2", "mask_dim", "mask_projection", "min", "nn", "normalize", "np", "num_feature_levels", "num_fpn_levels", "num_pos_feats", "output_conv", "output_convs", "padding", "position_embedding", "self", "stride", "strides", "super", "torch", "transformer_feature_strides", "transformer_in_channels"], "oneformer/modeling_oneformer.py:OneFormerPixelDecoder.get_valid_ratio": ["ModelPixelDecoder", "Module", "_", "class", "def", "dtype", "float32", "get_valid_ratio", "height", "mask", "nn", "return", "self", "shape", "stack", "sum", "to", "torch", "valid_height", "valid_ratio", "valid_ratio_height", "valid_ratio_width", "valid_width", "width"], "oneformer/modeling_oneformer.py:OneFormerPixelDecoder.forward": ["False", "ModelPixelDecoder", "ModelPixelDecoderOutput", "Module", "NModel", "align_corners", "append", "as_tensor", "attention_mask", "attentions", "batch_size", "bilinear", "bool", "bs", "cat", "class", "config", "cumsum", "cur_fpn", "def", "device", "dim", "dtype", "else", "encoder", "encoder_outputs", "enumerate", "feats", "features", "flatten", "for", "forward", "functional", "get_valid_ratio", "height", "i", "idx", "if", "in", "input_projections", "inputs_embeds", "interpolate", "is", "last_hidden_state", "lateral_conv", "lateral_convs", "level", "level_embed", "level_start_index", "long", "lvl_pos_embed", "lvl_pos_embed_flatten", "m", "mask", "mask_features", "mask_flatten", "mask_projection", "masks", "mode", "multi_scale_features", "new_zeros", "nn", "not", "num_channels", "num_cur_levels", "num_feature_levels", "num_fpn_levels", "o", "out", "output_attentions", "output_conv", "output_convs", "output_hidden_states", "pos_embed", "position_embedding", "position_embeddings", "position_embeddings_list", "prod", "range", "return", "return_dict", "self", "shape", "size", "source", "source_flatten", "sources", "spatial_shape", "spatial_shapes", "split", "split_size_or_sections", "stack", "torch", "transpose", "valid_ratios", "view", "width", "x", "y", "z", "zeros", "zip"], "oneformer/modeling_oneformer.py:OneFormerPixelLevelModule.__init__": ["ModelConfig", "ModelPixelDecoder", "ModelPixelLevelModule", "Module", "__init__", "channels", "class", "config", "decoder", "def", "encoder", "feature_channels", "load_backbModel", "nn", "self", "super"], "oneformer/modeling_oneformer.py:OneFormerPixelLevelModule.forward": ["False", "ModelPixelDecoderOutput", "ModelPixelLevelModule", "ModelPixelLevelModuleOutput", "Module", "Tensor", "bool", "class", "decoder", "decoder_features", "decoder_last_feature", "decoder_output", "def", "encoder", "encoder_features", "feature_maps", "features", "forward", "list", "mask_features", "multi_scale_features", "nn", "output_hidden_states", "pixel_values", "return", "self", "tuple"], "oneformer/modeling_oneformer.py:OneFormerAttention.__init__": ["False", "Linear", "ModelAttention", "Module", "True", "ValueError", "__init__", "and", "be", "bias", "bool", "by", "class", "def", "divisible", "dropout", "embed_dim", "f", "float", "got", "head_dim", "if", "int", "is_decoder", "k_proj", "must", "nn", "num_heads", "out_proj", "q_proj", "raise", "scaling", "self", "super", "v_proj"], "oneformer/modeling_oneformer.py:OneFormerAttention._shape": ["ModelAttention", "Module", "Tensor", "_shape", "batch_size", "class", "contiguous", "def", "head_dim", "int", "nn", "num_heads", "return", "self", "seq_len", "tensor", "torch", "transpose", "view"], "oneformer/modeling_oneformer.py:OneFormerAttention.with_pos_embed": ["ModelAttention", "Module", "NModel", "Tensor", "class", "def", "else", "if", "is", "nn", "position_embeddings", "return", "self", "tensor", "torch", "with_pos_embed"], "oneformer/modeling_oneformer.py:OneFormerAttention.forward": ["Attention", "False", "ModelAttention", "Module", "NModel", "Tensor", "ValueError", "_shape", "attention_mask", "attn_output", "attn_probs", "attn_weights", "attn_weights_reshaped", "batch_size", "be", "bmm", "bool", "but", "class", "def", "dim", "dropout", "else", "embed_dim", "f", "forward", "functional", "head_dim", "hidden_states", "hidden_states_original", "if", "is", "is_cross_attention", "k_proj", "key_states", "key_value_position_embeddings", "key_value_states", "key_value_states_original", "mask", "nn", "not", "num_heads", "of", "out_proj", "output_attentions", "p", "permute", "position_embeddings", "proj_shape", "q_proj", "query_states", "raise", "reshape", "return", "scaling", "self", "should", "size", "softmax", "source_len", "target_len", "torch", "training", "transpose", "tuple", "v_proj", "value_states", "view", "weights", "with_pos_embed"], "oneformer/modeling_oneformer.py:OneFormerTransformerDecoderSelfAttentionLayer.__init__": ["ACT2FN", "Dropout", "False", "LayerNorm", "ModelAttention", "ModelTransformerDecoderSelfAttentionLayer", "Module", "True", "__init__", "activation", "class", "def", "dropout", "embed_dim", "eps", "is_decoder", "layer_norm_eps", "nn", "norm", "normalize_before", "num_heads", "relu", "self", "self_attn", "super"], "oneformer/modeling_oneformer.py:OneFormerTransformerDecoderSelfAttentionLayer.with_pos_embed": ["ModelTransformerDecoderSelfAttentionLayer", "Module", "NModel", "Tensor", "class", "def", "else", "if", "is", "nn", "pos", "return", "self", "tensor", "with_pos_embed"], "oneformer/modeling_oneformer.py:OneFormerTransformerDecoderSelfAttentionLayer.forward_post": ["ModelTransformerDecoderSelfAttentionLayer", "Module", "NModel", "Tensor", "True", "attention_mask", "attention_weights", "class", "def", "dropout", "forward_post", "hidden_states", "nn", "norm", "output", "output2", "output_attentions", "output_key_padding_mask", "output_mask", "position_embeddings", "query_pos", "return", "self", "self_attn"], "oneformer/modeling_oneformer.py:OneFormerTransformerDecoderSelfAttentionLayer.forward_pre": ["ModelTransformerDecoderSelfAttentionLayer", "Module", "NModel", "Tensor", "True", "attention_mask", "attention_weights", "class", "def", "dropout", "forward_pre", "hidden_states", "nn", "norm", "output", "output2", "output_attentions", "output_key_padding_mask", "output_mask", "position_embeddings", "query_pos", "return", "self", "self_attn"], "oneformer/modeling_oneformer.py:OneFormerTransformerDecoderSelfAttentionLayer.forward": ["ModelTransformerDecoderSelfAttentionLayer", "Module", "NModel", "Tensor", "class", "def", "forward", "forward_post", "forward_pre", "if", "nn", "normalize_before", "output", "output_key_padding_mask", "output_mask", "query_pos", "return", "self"], "oneformer/modeling_oneformer.py:OneFormerTransformerDecoderCrossAttentionLayer.__init__": ["ACT2FN", "Dropout", "False", "LayerNorm", "ModelTransformerDecoderCrossAttentionLayer", "Module", "MultiheadAttention", "__init__", "activation", "class", "def", "dropout", "embed_dim", "eps", "layer_norm_eps", "multihead_attn", "nn", "norm", "normalize_before", "num_heads", "relu", "self", "super"], "oneformer/modeling_oneformer.py:OneFormerTransformerDecoderCrossAttentionLayer.with_pos_embed": ["ModelTransformerDecoderCrossAttentionLayer", "Module", "NModel", "Tensor", "class", "def", "else", "if", "is", "nn", "pos", "return", "self", "tensor", "with_pos_embed"], "oneformer/modeling_oneformer.py:OneFormerTransformerDecoderCrossAttentionLayer.forward_post": ["ModelTransformerDecoderCrossAttentionLayer", "Module", "NModel", "Tensor", "attention_weights", "attn_mask", "class", "def", "dropout", "forward_post", "key", "key_padding_mask", "memory", "memory_key_padding_mask", "memory_mask", "multihead_attn", "nn", "norm", "output", "output2", "pos", "query", "query_pos", "return", "self", "value", "with_pos_embed"], "oneformer/modeling_oneformer.py:OneFormerTransformerDecoderCrossAttentionLayer.forward_pre": ["ModelTransformerDecoderCrossAttentionLayer", "Module", "NModel", "Tensor", "attention_weights", "attn_mask", "class", "def", "dropout", "forward_pre", "key", "key_padding_mask", "memory", "memory_key_padding_mask", "memory_mask", "multihead_attn", "nn", "norm", "output", "output2", "pos", "query", "query_pos", "return", "self", "value", "with_pos_embed"], "oneformer/modeling_oneformer.py:OneFormerTransformerDecoderCrossAttentionLayer.forward": ["ModelTransformerDecoderCrossAttentionLayer", "Module", "NModel", "Tensor", "class", "def", "forward", "forward_post", "forward_pre", "if", "memory", "memory_key_padding_mask", "memory_mask", "nn", "normalize_before", "output", "pos", "query_pos", "return", "self"], "oneformer/modeling_oneformer.py:OneFormerTransformerDecoderFFNLayer.__init__": ["ACT2FN", "Dropout", "False", "LayerNorm", "Linear", "ModelTransformerDecoderFFNLayer", "Module", "__init__", "activation", "class", "d_model", "def", "dim_feedforward", "dropout", "eps", "layer_norm_eps", "linear1", "linear2", "nn", "norm", "normalize_before", "relu", "self", "super"], "oneformer/modeling_oneformer.py:OneFormerTransformerDecoderFFNLayer.with_pos_embed": ["ModelTransformerDecoderFFNLayer", "Module", "NModel", "Tensor", "class", "def", "else", "if", "is", "nn", "pos", "return", "self", "tensor", "with_pos_embed"], "oneformer/modeling_oneformer.py:OneFormerTransformerDecoderFFNLayer.forward_post": ["ModelTransformerDecoderFFNLayer", "Module", "activation", "class", "def", "dropout", "forward_post", "linear1", "linear2", "nn", "norm", "output", "output2", "return", "self"], "oneformer/modeling_oneformer.py:OneFormerTransformerDecoderFFNLayer.forward_pre": ["ModelTransformerDecoderFFNLayer", "Module", "activation", "class", "def", "dropout", "forward_pre", "linear1", "linear2", "nn", "norm", "output", "output2", "return", "self"], "oneformer/modeling_oneformer.py:OneFormerTransformerDecoderFFNLayer.forward": ["ModelTransformerDecoderFFNLayer", "Module", "class", "def", "forward", "forward_post", "forward_pre", "if", "nn", "normalize_before", "output", "return", "self"], "oneformer/modeling_oneformer.py:OneFormerMLPPredictionHead.__init__": ["Identity", "ModelMLPPredictionHead", "Module", "PredictionBlock", "ReLU", "Sequential", "__init__", "activation", "append", "class", "def", "else", "enumerate", "for", "hidden_dim", "i", "if", "in", "in_dim", "in_dims", "input_dim", "int", "layers", "nn", "num_layers", "out_dim", "out_dims", "output_dim", "self", "super", "zip"], "oneformer/modeling_oneformer.py:OneFormerMLPPredictionHead.forward": ["ModelMLPPredictionHead", "Module", "Tensor", "class", "def", "forward", "input", "layers", "nn", "return", "self"], "oneformer/modeling_oneformer.py:OneFormerTransformerDecoderLayer.__init__": ["ModelConfig", "ModelTransformerDecoderCrossAttentionLayer", "ModelTransformerDecoderFFNLayer", "ModelTransformerDecoderLayer", "ModelTransformerDecoderSelfAttentionLayer", "Module", "__init__", "class", "config", "cross_attn", "d_model", "def", "dim_feedforward", "dropout", "embed_dim", "ffn", "hidden_dim", "layer_norm_eps", "nn", "normalize_before", "num_attention_heads", "num_feature_levels", "num_heads", "pre_norm", "self", "self_attn", "super"], "oneformer/modeling_oneformer.py:OneFormerTransformerDecoderLayer.forward": ["False", "ModelTransformerDecoderLayer", "Module", "NModel", "Tensor", "attention_mask", "bool", "class", "cross_attn", "cross_attn_weights", "def", "ffn", "forward", "if", "index", "int", "level_index", "list", "memory_key_padding_mask", "memory_mask", "multi_stage_features", "multi_stage_positional_embeddings", "nn", "num_feature_levels", "output", "output_attentions", "output_key_padding_mask", "output_mask", "outputs", "pos", "query_embeddings", "query_pos", "return", "self", "self_attn", "self_attn_weights", "shape", "sum", "torch", "where"], "oneformer/modeling_oneformer.py:OneFormerTransformerDecoderQueryTransformerDecoder.__init__": ["False", "ModelTransformerDecoderQueryTransformerDecoder", "Module", "NModel", "__init__", "_get_clModels", "class", "decoder_layer", "def", "layers", "nn", "norm", "num_layers", "return_intermediate", "self", "super"], "oneformer/modeling_oneformer.py:OneFormerTransformerDecoderQueryTransformerDecoder.forward": ["ModelTransformerDecoderQueryTransformerDecoder", "Module", "NModel", "Tensor", "append", "class", "def", "for", "forward", "if", "in", "intermediate", "is", "layer", "layers", "memory", "memory_key_padding_mask", "memory_mask", "nn", "norm", "not", "output", "output_key_padding_mask", "output_mask", "pop", "pos", "query_pos", "return", "return_intermediate", "self", "stack", "torch", "unsqueeze"], "oneformer/modeling_oneformer.py:OneFormerTransformerDecoderQueryTransformerDecoderLayer.__init__": ["ACT2FN", "Dropout", "False", "LayerNorm", "Linear", "ModelTransformerDecoderQueryTransformerDecoderLayer", "Module", "MultiheadAttention", "__init__", "activation", "class", "d_model", "def", "dim_feedforward", "dropout", "dropout1", "dropout2", "dropout3", "eps", "layer_norm_eps", "linear1", "linear2", "multihead_attn", "nhead", "nn", "norm1", "norm2", "norm3", "normalize_before", "relu", "self", "self_attn", "super"], "oneformer/modeling_oneformer.py:OneFormerTransformerDecoderQueryTransformerDecoderLayer.with_pos_embed": ["ModelTransformerDecoderQueryTransformerDecoderLayer", "Module", "NModel", "Tensor", "class", "def", "else", "if", "is", "nn", "pos", "return", "self", "tensor", "with_pos_embed"], "oneformer/modeling_oneformer.py:OneFormerTransformerDecoderQueryTransformerDecoderLayer.forward_post": ["ModelTransformerDecoderQueryTransformerDecoderLayer", "Module", "NModel", "Tensor", "activation", "attn_mask", "class", "def", "dropout", "dropout1", "dropout2", "dropout3", "forward_post", "k", "key", "key_padding_mask", "linear1", "linear2", "memory", "memory_key_padding_mask", "memory_mask", "multihead_attn", "nn", "norm1", "norm2", "norm3", "output", "output2", "output_key_padding_mask", "output_mask", "pos", "q", "query", "query_pos", "return", "self", "self_attn", "value", "with_pos_embed"], "oneformer/modeling_oneformer.py:OneFormerTransformerDecoderQueryTransformerDecoderLayer.forward_pre": ["ModelTransformerDecoderQueryTransformerDecoderLayer", "Module", "NModel", "Tensor", "activation", "attn_mask", "class", "def", "dropout", "dropout1", "dropout2", "dropout3", "forward_pre", "k", "key", "key_padding_mask", "linear1", "linear2", "memory", "memory_key_padding_mask", "memory_mask", "multihead_attn", "nn", "norm1", "norm2", "norm3", "output", "output2", "output_key_padding_mask", "output_mask", "pos", "q", "query", "query_pos", "return", "self", "self_attn", "value", "with_pos_embed"], "oneformer/modeling_oneformer.py:OneFormerTransformerDecoderQueryTransformerDecoderLayer.forward": ["ModelTransformerDecoderQueryTransformerDecoderLayer", "Module", "NModel", "Tensor", "class", "def", "forward", "forward_post", "forward_pre", "if", "memory", "memory_key_padding_mask", "memory_mask", "nn", "normalize_before", "output", "output_key_padding_mask", "output_mask", "pos", "query_pos", "return", "self"], "oneformer/modeling_oneformer.py:OneFormerTransformerDecoderQueryTransformer.__init__": ["False", "LayerNorm", "ModelTransformerDecoderQueryTransformer", "ModelTransformerDecoderQueryTransformerDecoder", "ModelTransformerDecoderQueryTransformerDecoderLayer", "Module", "__init__", "activation", "class", "d_model", "decoder", "decoder_layer", "decoder_norm", "def", "dim_feedforward", "dropout", "eps", "layer_norm_eps", "nhead", "nn", "normalize_before", "num_decoder_layers", "relu", "return_intermediate", "return_intermediate_dec", "self", "super"], "oneformer/modeling_oneformer.py:OneFormerTransformerDecoderQueryTransformer.forward": ["ModelTransformerDecoderQueryTransformer", "Module", "NModel", "batch_size", "class", "decoder", "def", "else", "flatten", "forward", "if", "is", "mask", "memory_key_padding_mask", "nn", "not", "permute", "pos", "pos_embed", "queries", "query_embed", "query_pos", "repeat", "return", "self", "shape", "src", "task_token", "torch", "transpose", "unsqueeze", "zeros_like"], "oneformer/modeling_oneformer.py:OneFormerTransformerDecoder.__init__": ["Conv2d", "False", "LayerNorm", "Linear", "ModelConfig", "ModelMLPPredictionHead", "ModelTransformerDecoder", "ModelTransformerDecoderLayer", "ModelTransformerDecoderQueryTransformer", "Module", "ModuleList", "_", "__init__", "class", "class_embed", "config", "d_model", "decoder_layers", "decoder_norm", "def", "dim_feedforward", "dropout", "eps", "for", "hidden_dim", "in", "in_channels", "int", "is_training", "kernel_size", "layer_norm_eps", "layers", "mask_dim", "mask_embed", "nhead", "nn", "normalize_before", "num_attention_heads", "num_decoder_layers", "num_feature_levels", "num_heads", "num_labels", "pre_norm", "query_dec_layers", "query_input_projection", "query_transformer", "range", "return_intermediate_dec", "self", "super", "use_auxiliary_loss", "use_task_norm"], "oneformer/modeling_oneformer.py:OneFormerTransformerDecoder.forward": ["Intermediate", "ModelTransformerDecoder", "ModelTransformerDecoderOutput", "Module", "NModel", "ValueError", "_get_aux_predictions", "append", "as", "attention_mask", "attention_mask_target_size", "attentions", "auxiliary_predictions", "cat", "clModel", "class", "contrastive_logits", "decoder", "decoder_norm", "def", "dim", "elements", "else", "enumerate", "for", "forward", "forward_prediction_heads", "have", "if", "in", "index", "intermediate_class_predictions", "intermediate_mask_predictions", "layer", "layer_outputs", "layers", "len", "mask_features", "multi_stage_features", "multi_stage_positional_embeddings", "must", "nn", "not", "num_feature_levels", "number", "object_queries", "of", "output", "output_attentions", "outputs_class", "outputs_mask", "permute", "prediction_class", "prediction_masks", "predictions", "queries", "query_embedder", "query_embeddings", "query_features", "query_input_projection", "query_transformer", "raise", "return", "same", "self", "size_list", "task_token", "the", "torch", "transformer", "use_auxiliary_loss", "use_task_norm", "weight"], "oneformer/modeling_oneformer.py:OneFormerTransformerDecoder.forward_prediction_heads": ["False", "ModelTransformerDecoder", "Module", "align_corners", "attention_mask", "attention_mask_target_size", "bchw", "bilinear", "bool", "bqc", "bqhw", "class", "class_embed", "decoder_norm", "decoder_output", "def", "detach", "einsum", "flatten", "forward_prediction_heads", "functional", "interpolate", "mask_embed", "mask_features", "mode", "nn", "num_heads", "output", "outputs_class", "outputs_mask", "repeat", "return", "self", "sigmoid", "size", "torch", "transpose", "unsqueeze"], "oneformer/modeling_oneformer.py:OneFormerTransformerDecoder._get_aux_predictions": ["ModelTransformerDecoder", "Module", "_get_aux_predictions", "a", "aux_list", "b", "class", "class_queries_logits", "def", "for", "in", "masks_queries_logits", "nn", "outputs_class", "outputs_seg_masks", "return", "self", "tuple", "zip"], "oneformer/modeling_oneformer.py:OneFormerTransformerModule.__init__": ["Conv2d", "Embedding", "ModelConfig", "ModelSinePositiModelmbedding", "ModelTransformerDecoder", "ModelTransformerModule", "Module", "Sequential", "True", "_", "__init__", "append", "class", "config", "decoder", "def", "else", "enforce_input_proj", "for", "hidden_dim", "if", "in", "in_channels", "in_features", "input_projections", "int", "kernel_size", "level_embed", "nn", "normalize", "num_feature_levels", "num_pos_feats", "num_queries", "or", "position_embedder", "queries_embedder", "range", "self", "super"], "oneformer/modeling_oneformer.py:OneFormerTransformerModule.forward": ["False", "ModelTransformerDecoderOutput", "ModelTransformerModule", "Module", "NModel", "Number", "Tensor", "ValueError", "_", "and", "append", "batch_size", "bool", "class", "decoder", "def", "device", "do", "dtype", "elements", "f", "flatten", "for", "forward", "i", "if", "in", "input_projections", "len", "level_embed", "list", "mask_features", "match", "multi_scale_features", "multi_stage_features", "multi_stage_positional_embeddings", "nn", "not", "num_feature_levels", "of", "output_attentions", "permute", "position_embedder", "queries_embedder", "query_embedder", "query_embeddings", "query_features", "raise", "range", "repeat", "return", "self", "shape", "size_list", "task_token", "unsqueeze", "weight"], "oneformer/modeling_oneformer.py:OneFormerSinePositionEmbedding.__init__": ["False", "ModelSinePositiModelmbedding", "Module", "NModel", "True", "ValueError", "__init__", "and", "be", "bool", "class", "def", "else", "float", "if", "int", "is", "math", "nn", "normalize", "not", "num_pos_feats", "passed", "pi", "raise", "scale", "self", "should", "super", "temperature"], "oneformer/modeling_oneformer.py:OneFormerSinePositionEmbedding.forward": ["ModelSinePositiModelmbedding", "Module", "NModel", "Size", "Tensor", "arange", "bool", "cat", "class", "cos", "cumsum", "def", "device", "dim", "dim_t", "div", "dtype", "eps", "flatten", "floor", "forward", "if", "int64", "is", "mask", "nn", "normalize", "not_mask", "num_pos_feats", "permute", "pos", "pos_x", "pos_y", "return", "rounding_mode", "scale", "self", "shape", "sin", "stack", "str", "temperature", "to", "torch", "x_embed", "y_embed", "zeros"], "oneformer/modeling_oneformer.py:PredictionBlock.__init__": ["Linear", "ModelBlock", "Module", "None", "__init__", "activation", "add_module", "class", "def", "enumerate", "for", "i", "in", "in_dim", "int", "layer", "layers", "nn", "out_dim", "self", "str", "super"], "oneformer/modeling_oneformer.py:PredictionBlock.forward": ["ModelBlock", "Module", "Tensor", "class", "def", "for", "forward", "hidden_state", "in", "input", "layer", "layers", "nn", "return", "self"], "oneformer/modeling_oneformer.py:OneFormerTextMapperAttention.__init__": ["Dropout", "False", "Linear", "ModelTextMapperAttention", "Module", "NModel", "__init__", "attn_drop", "bias", "class", "def", "dim", "head_dim", "k_proj", "nn", "num_heads", "or", "proj", "proj_drop", "q_proj", "qk_scale", "qkv_bias", "scale", "self", "super", "v_proj"], "oneformer/modeling_oneformer.py:OneFormerTextMapperAttention.forward": ["ModelTextMapperAttention", "Module", "ValueError", "and", "attn", "batch_size", "bknm", "bmkc", "bnkc", "class", "def", "different", "dim", "einsum", "f", "forward", "have", "if", "k", "k_proj", "k_sequence_length", "keys", "list", "nn", "not", "num_channels", "num_heads", "output", "proj", "proj_drop", "q", "q_proj", "q_sequence_length", "raise", "reshape", "return", "scale", "self", "shape", "shapes", "softmax", "torch", "v", "v_proj", "values"], "oneformer/modeling_oneformer.py:OneFormerTextTransformerDecoderLayer.__init__": ["Dropout", "GELU", "LayerNorm", "Linear", "ModelTextMapperAttention", "ModelTextTransformerDecoderLayer", "Module", "Sequential", "__init__", "class", "cross_attn", "d_model", "def", "dropout", "eps", "layer_norm_eps", "mlp", "nhead", "nn", "norm1", "norm2", "norm3", "proj_drop", "self", "self_attn", "super"], "oneformer/modeling_oneformer.py:OneFormerTextTransformerDecoderLayer.forward": ["ModelTextTransformerDecoderLayer", "Module", "class", "cross_attn", "def", "dropout", "forward", "hidden_state", "k", "mem", "mlp", "nn", "norm1", "norm2", "norm3", "q", "return", "self", "self_attn", "v"], "oneformer/modeling_oneformer.py:OneFormerTextContextDecoder.__init__": ["LayerNorm", "Linear", "ModelTextContextDecoder", "ModelTextTransformerDecoderLayer", "Module", "ModuleList", "Sequential", "_", "__init__", "class", "decoder", "def", "dropout", "eps", "for", "in", "kwargs", "layer_norm_eps", "memory_proj", "nn", "out_proj", "range", "self", "super", "text_proj", "transformer_heads", "transformer_layers", "transformer_width", "visual_dim"], "oneformer/modeling_oneformer.py:OneFormerTextContextDecoder.forward": ["ModelTextContextDecoder", "Module", "class", "decoder", "def", "for", "forward", "hidden_state", "in", "layer", "memory_proj", "nn", "out_proj", "return", "self", "text", "text_proj", "visual"], "oneformer/modeling_oneformer.py:OneFormerTextMLP.__init__": ["ACT2FN", "Linear", "ModelTextMLP", "Module", "NModel", "__init__", "activation_fn", "class", "def", "fc1", "fc2", "hidden_size", "int", "intermediate_size", "nn", "output_size", "quick_gelu", "self", "super"], "oneformer/modeling_oneformer.py:OneFormerTextMLP.forward": ["ModelTextMLP", "Module", "Tensor", "activation_fn", "class", "def", "fc1", "fc2", "forward", "hidden_states", "nn", "return", "self", "torch"], "oneformer/modeling_oneformer.py:OneFormerTextTransformerLayer.__init__": ["GradientCheckpointingLayer", "LayerNorm", "ModelTextMLP", "ModelTextTransformerLayer", "MultiheadAttention", "Tensor", "__init__", "attn_mask", "class", "def", "eps", "heads", "int", "layer_norm1", "layer_norm2", "layer_norm_eps", "mlp", "nn", "self", "self_attn", "super", "torch", "width"], "oneformer/modeling_oneformer.py:OneFormerTextTransformerLayer.forward": ["False", "FloatTensor", "GradientCheckpointingLayer", "ModelTextTransformerLayer", "NModel", "Tensor", "class", "def", "forward", "hidden_states", "key_padding_mask", "layer_norm1", "layer_norm2", "mlp", "need_weights", "residual", "return", "self", "self_attn", "torch"], "oneformer/modeling_oneformer.py:OneFormerTextTransformer.__init__": ["False", "ModelTextTransformer", "ModelTextTransformerLayer", "Module", "NModel", "Sequential", "Tensor", "_", "__init__", "attn_mask", "class", "def", "for", "heads", "in", "int", "layer_norm_eps", "layers", "nn", "num_layers", "range", "self", "super", "torch", "use_checkpoint", "width"], "oneformer/modeling_oneformer.py:OneFormerTextTransformer.forward": ["ModelTextTransformer", "Module", "Tensor", "class", "def", "for", "forward", "hidden_states", "in", "layer", "layers", "nn", "return", "self", "torch"], "oneformer/modeling_oneformer.py:OneFormerTextEncoder.__init__": ["Embedding", "False", "LayerNorm", "ModelTextEncoder", "ModelTextTransformer", "Module", "Parameter", "__init__", "attn_mask", "build_attention_mask", "class", "context_length", "def", "empty", "eps", "heads", "int", "layer_norm_eps", "layers", "ln_final", "nn", "positional_embedding", "self", "super", "token_embedding", "torch", "transformer", "use_checkpoint", "vocab_size", "width"], "oneformer/modeling_oneformer.py:OneFormerTextEncoder.build_attention_mask": ["ModelTextEncoder", "Module", "build_attention_mask", "class", "context_length", "def", "empty", "fill_", "float", "inf", "mask", "nn", "return", "self", "torch", "triu_"], "oneformer/modeling_oneformer.py:OneFormerTextEncoder.forward": ["ModelTextEncoder", "Module", "arange", "argmax", "class", "def", "dim", "forward", "hidden_state", "ln_final", "nn", "permute", "positional_embedding", "return", "self", "shape", "text", "token_embedding", "torch", "transformer"], "oneformer/modeling_oneformer.py:OneFormerTextMapper.__init__": ["Embedding", "ModelConfig", "ModelMLPPredictionHead", "ModelTextEncoder", "ModelTextMapper", "Module", "NModel", "__init__", "class", "config", "context_length", "def", "else", "hidden_dim", "if", "layer_norm_eps", "layers", "nn", "prompt_ctx", "self", "super", "text_encoder", "text_encoder_context_length", "text_encoder_n_ctx", "text_encoder_num_layers", "text_encoder_proj_layers", "text_encoder_vocab_size", "text_encoder_width", "text_projector", "vocab_size", "width"], "oneformer/modeling_oneformer.py:OneFormerTextMapper.forward": ["ModelTextMapper", "Module", "Tensor", "class", "def", "encode_text", "forward", "inputs", "nn", "return", "self", "text_queries"], "oneformer/modeling_oneformer.py:OneFormerTextMapper.encode_text": ["False", "ModelTextMapper", "Module", "NModel", "NModelType", "Number", "True", "ValueError", "_", "batch_size", "be", "cat", "class", "def", "dim", "dimensions", "encode_text", "encoded_text", "hidden_dim", "if", "in", "is", "must", "ndim", "nn", "not", "num_text", "of", "or", "prompt_ctx", "raise", "repeat", "reshape", "return", "self", "shape", "squeeze_dim", "text", "text_encoder", "text_projector", "text_queries", "text_queries_ctx", "torch", "unsqueeze", "weight"], "oneformer/modeling_oneformer.py:OneFormerTaskModel.__init__": ["ModelConfig", "ModelMLPPredictionHead", "ModelTaskModel", "Module", "__init__", "class", "config", "def", "hidden_dim", "nn", "self", "super", "task_mlp", "task_seq_len"], "oneformer/modeling_oneformer.py:OneFormerTaskModel.forward": ["ModelTaskModel", "Module", "Tensor", "class", "def", "forward", "inputs", "nn", "return", "self", "task_mlp", "task_tokens"], "oneformer/modeling_oneformer.py:OneFormerPreTrainedModel._init_weights": ["BatchNorm2d", "Conv2d", "Embedding", "False", "GroupNorm", "LayerNorm", "Linear", "ModelLoss", "ModelMLPPredictionHead", "ModelPixelDecoder", "ModelPixelDecoderEncoderMultiscaleDeformableAttention", "ModelPreTrainedModel", "ModelTextEncoder", "ModelTextTransformer", "ModelTransformerDecoder", "ModelTransformerDecoderLayer", "ModelTransformerDecoderQueryTransformer", "ModelTransformerModule", "Models", "Models_", "Module", "MultiheadAttention", "NModel", "PreTrainedModel", "Sequential", "True", "_init_weights", "_is_hf_initialized", "abs", "and", "arange", "attention_weights", "attn_std", "bias", "class", "config", "constant_", "contrastive_temperature", "copy_", "cos", "def", "dim", "dtype", "elif", "empty_weight", "eos_coef", "fc1", "fc2", "fc_std", "float", "for", "gain", "getattr", "grid_init", "hasattr", "i", "if", "in", "in_proj_bias", "in_proj_weight", "init", "init_std", "init_xavier_std", "input_projection", "input_projections", "int64", "is", "isinstance", "keepdim", "layer", "layers", "level_embed", "log", "logit_scale", "math", "max", "mean", "mlp", "module", "modules", "n_heads", "n_levels", "n_points", "nn", "normal_", "not", "np", "num_batches_tracked", "num_classes", "num_layers", "out_proj", "output_proj", "p", "padding_idx", "parameters", "pi", "positional_embedding", "proj_std", "query_input_projection", "range", "reference_points", "repeat", "running_mean", "running_var", "sampling_offsets", "self", "self_attn", "sin", "stack", "std", "submodule", "thetas", "token_embedding", "torch", "value_proj", "view", "weight", "width", "xavier_std", "xavier_uniform_", "zeros_"], "oneformer/modeling_oneformer.py:OneFormerModel.__init__": ["ModelConfig", "ModelModel", "ModelPixelLevelModule", "ModelPreTrainedModel", "ModelTaskModel", "ModelTextMapper", "ModelTransformerModule", "NModel", "__init__", "class", "config", "conv_dim", "def", "else", "if", "in_features", "is_training", "pixel_level_module", "post_init", "self", "super", "task_encoder", "text_mapper", "transformer_module"], "oneformer/modeling_oneformer.py:OneFormerModel.forward": ["ModelModel", "ModelModelOutput", "ModelPreTrainedModel", "Models", "NModel", "Tensor", "ValueError", "You", "_", "attentions", "auxiliary_predictions", "batch_size", "bool", "class", "config", "contrastive_logits", "decoder_features", "decoder_last_feature", "def", "device", "dtype", "else", "encoder_features", "encoder_hidden_states", "f", "for", "forward", "have", "height", "if", "in", "is", "is_training", "kwargs", "mask_features", "multi_scale_features", "not", "object_queries", "output", "output_attentions", "output_hidden_states", "pixel_decoder_hidden_states", "pixel_level_module", "pixel_level_module_output", "pixel_mask", "pixel_values", "prediction_class", "prediction_masks", "queries", "r", "raise", "return", "return_dict", "self", "shape", "specify", "task_encoder", "task_inputs", "task_token", "text_inputs", "text_mapper", "text_queries", "to", "torch", "transformer_decoder_auxiliary_predictions", "transformer_decoder_class_predictions", "transformer_decoder_contrastive_queries", "transformer_decoder_hidden_states", "transformer_decoder_mask_predictions", "transformer_decoder_object_queries", "transformer_module", "transformer_module_output", "tuple", "use_return_dict", "v", "values", "width"], "oneformer/modeling_oneformer.py:OneFormerForUniversalSegmentation.__init__": ["ModelConfig", "ModelForUniversalSegmentation", "ModelHungarianMatcher", "ModelLoss", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "class_weight", "config", "contrastive_temperature", "contrastive_weight", "cost_class", "cost_dice", "cost_mask", "criterion", "def", "dice_weight", "dict", "eos_coef", "float", "importance_sample_ratio", "loss_contrastive", "loss_cross_entropy", "loss_dice", "loss_mask", "mask_weight", "matcher", "model", "no_object_weight", "num_classes", "num_labels", "num_points", "oversample_ratio", "post_init", "self", "str", "super", "train_num_points", "weight_dict"], "oneformer/modeling_oneformer.py:OneFormerForUniversalSegmentation.get_loss_dict": ["ModelForUniversalSegmentation", "ModelPreTrainedModel", "Tensor", "auxiliary_predictions", "bool", "calculate_contrastive_loss", "class", "class_labels", "class_queries_logits", "contrastive_queries_logits", "criterion", "def", "dict", "for", "get_loss_dict", "if", "in", "items", "key", "loss", "loss_dict", "loss_key", "mask_labels", "masks_queries_logits", "return", "self", "str", "text_queries", "weight", "weight_dict"], "oneformer/modeling_oneformer.py:OneFormerForUniversalSegmentation.get_loss": ["ModelForUniversalSegmentation", "ModelPreTrainedModel", "Tensor", "class", "def", "dict", "get_loss", "loss_dict", "return", "self", "str", "sum", "values"], "oneformer/modeling_oneformer.py:OneFormerForUniversalSegmentation.forward": ["ModelForUniversalSegmentation", "ModelForUniversalSegmentationOutput", "ModelPreTrainedModel", "NModel", "Tensor", "True", "and", "auxiliary_predictions", "bool", "calculate_contrastive_loss", "class", "class_labels", "class_queries_logits", "config", "contrastive_queries_logits", "contrastive_temperature", "def", "dict", "else", "for", "forward", "get_loss", "get_loss_dict", "if", "in", "is", "kwargs", "list", "loss", "loss_dict", "mask_labels", "masks_queries_logits", "model", "not", "or", "output", "output_attentions", "output_auxiliary_logits", "output_hidden_states", "outputs", "pixel_mask", "pixel_values", "r", "return", "return_dict", "self", "str", "task_inputs", "text_inputs", "text_queries", "transformer_decoder_auxiliary_predictions", "transformer_decoder_class_predictions", "transformer_decoder_contrastive_queries", "transformer_decoder_mask_predictions", "tuple", "use_auxiliary_loss", "use_return_dict", "v", "values"], "openai/modeling_openai.py:Attention.__init__": ["Conv1D", "Dropout", "False", "Model", "Module", "ValueError", "__init__", "attn_dropout", "attn_pdrop", "be", "bias", "by", "c_attn", "c_proj", "class", "config", "def", "divisible", "f", "if", "must", "n_head", "n_positions", "n_state", "nn", "nx", "ones", "persistent", "raise", "register_buffer", "resid_dropout", "resid_pdrop", "scale", "self", "shape", "split_size", "super", "torch", "tril", "view"], "openai/modeling_openai.py:Attention._attn": ["False", "Model", "Model_mask", "Module", "None", "_attn", "append", "attn_dropout", "b", "bias", "class", "def", "dim", "functional", "if", "is", "k", "math", "matmul", "nn", "not", "output_Models", "outputs", "q", "return", "scale", "self", "size", "softmax", "sqrt", "torch", "v", "w"], "openai/modeling_openai.py:Attention.merge_heads": ["Model", "Module", "class", "contiguous", "def", "merge_heads", "new_x_shape", "nn", "permute", "return", "self", "size", "view", "x"], "openai/modeling_openai.py:Attention.split_heads": ["False", "Model", "Module", "class", "def", "else", "if", "k", "n_head", "new_x_shape", "nn", "permute", "return", "self", "size", "split_heads", "view", "x"], "openai/modeling_openai.py:Attention.forward": ["False", "Model", "Model_mask", "Module", "None", "True", "_attn", "a", "attn_outputs", "c_attn", "c_proj", "class", "def", "dim", "forward", "k", "key", "merge_heads", "nn", "output_Models", "outputs", "query", "resid_dropout", "return", "self", "split", "split_heads", "split_size", "value", "x"], "openai/modeling_openai.py:MLP.__init__": ["ACT_FNS", "Conv1D", "Dropout", "Model", "Module", "__init__", "act", "afn", "c_fc", "c_proj", "class", "config", "def", "dropout", "n_embd", "n_state", "nn", "nx", "resid_pdrop", "self", "super"], "openai/modeling_openai.py:MLP.forward": ["Model", "Module", "act", "c_fc", "c_proj", "class", "def", "dropout", "forward", "h", "h2", "nn", "return", "self", "x"], "openai/modeling_openai.py:Block.__init__": ["Attention", "False", "LayerNorm", "MLP", "Model", "Module", "__init__", "attn", "class", "config", "def", "eps", "layer_norm_epsilon", "ln_1", "ln_2", "mlp", "n_embd", "n_positions", "nn", "nx", "scale", "self", "super"], "openai/modeling_openai.py:Block.forward": ["False", "Model", "Module", "None", "a", "attention_mask", "attn", "attn_outputs", "class", "def", "forward", "h", "ln_1", "ln_2", "m", "mlp", "n", "nn", "output_attentions", "outputs", "return", "self", "x"], "openai/modeling_openai.py:OpenAIGPTSequenceSummary.__init__": ["Callable", "Dropout", "Identity", "Linear", "ModelGPTConfig", "ModelGPTSequenceSummary", "Module", "None", "NotImplementedError", "__init__", "activation", "activation_string", "and", "attn", "class", "config", "def", "else", "first_dropout", "get_activation", "getattr", "hasattr", "hidden_size", "if", "last", "last_dropout", "nn", "num_classes", "num_labels", "raise", "self", "summary", "summary_activation", "summary_first_dropout", "summary_last_dropout", "summary_proj_to_labels", "summary_type", "summary_use_proj", "super"], "openai/modeling_openai.py:OpenAIGPTSequenceSummary.forward": ["FloatTensor", "LongTensor", "ModelGPTSequenceSummary", "Module", "None", "NotImplementedError", "activation", "attn", "class", "cls_index", "def", "dim", "dtype", "elif", "else", "expand", "first", "first_dropout", "forward", "full_like", "gather", "hidden_states", "if", "is", "last", "last_dropout", "long", "mean", "nn", "output", "raise", "return", "self", "shape", "size", "squeeze", "summary", "summary_type", "torch", "unsqueeze"], "openai/modeling_openai.py:OpenAIGPTPreTrainedModel._init_weights": ["Attention", "ModelGPTModel", "ModelGPTPreTrainedModel", "PreTrainedModel", "_init_weights", "arange", "bias", "class", "config", "copy_", "def", "elif", "if", "init", "isinstance", "module", "n_positions", "ones", "position_ids", "self", "super", "torch", "tril", "view"], "openai/modeling_openai.py:OpenAIGPTModel.__init__": ["Block", "Dropout", "Embedding", "False", "ModelGPTModel", "ModelGPTPreTrainedModel", "ModuleList", "True", "_", "__init__", "arange", "class", "config", "def", "drop", "embd_pdrop", "for", "h", "in", "n_embd", "n_layer", "n_positions", "nn", "persistent", "position_ids", "positions_embed", "post_init", "range", "register_buffer", "scale", "self", "super", "tokens_embed", "torch", "vocab_size"], "openai/modeling_openai.py:OpenAIGPTModel.get_input_embeddings": ["ModelGPTModel", "ModelGPTPreTrainedModel", "class", "def", "get_input_embeddings", "return", "self", "tokens_embed"], "openai/modeling_openai.py:OpenAIGPTModel.set_input_embeddings": ["ModelGPTModel", "ModelGPTPreTrainedModel", "class", "def", "new_embeddings", "self", "set_input_embeddings", "tokens_embed"], "openai/modeling_openai.py:OpenAIGPTModel.forward": ["BaseModelOutput", "FloatTensor", "LongTensor", "ModelGPTModel", "ModelGPTPreTrainedModel", "None", "Tensor", "ValueError", "You", "all_attentions", "all_hidden_states", "and", "at", "attention_mask", "attentions", "block", "bool", "both", "cannot", "class", "config", "def", "drop", "dtype", "either", "elif", "else", "enumerate", "finfo", "for", "forward", "h", "have", "hidden_states", "i", "if", "in", "input_ids", "input_shape", "inputs_embeds", "is", "kwargs", "last_hidden_state", "min", "next", "not", "or", "output_attentions", "output_hidden_states", "output_shape", "outputs", "parameters", "position_embeds", "position_ids", "positions_embed", "raise", "return", "return_dict", "same", "self", "size", "specify", "the", "time", "to", "token_type_embeds", "token_type_ids", "tokens_embed", "torch", "tuple", "unsqueeze", "use_return_dict", "v", "view", "warn_if_padding_and_no_attention_mask"], "openai/modeling_openai.py:OpenAIGPTLMHeadModel.__init__": ["False", "GenerationMixin", "Linear", "ModelGPTLMHeadModel", "ModelGPTModel", "ModelGPTPreTrainedModel", "__init__", "bias", "class", "config", "def", "lm_head", "n_embd", "nn", "post_init", "self", "super", "transformer", "vocab_size"], "openai/modeling_openai.py:OpenAIGPTLMHeadModel.forward": ["CausalLMOutput", "FloatTensor", "GenerationMixin", "LongTensor", "ModelGPTLMHeadModel", "ModelGPTPreTrainedModel", "None", "Tensor", "attention_mask", "attentions", "bool", "class", "config", "def", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "not", "output", "output_attentions", "output_hidden_states", "position_ids", "r", "return", "return_dict", "self", "slice", "slice_indices", "token_type_ids", "torch", "transformer", "transformer_outputs", "tuple", "use_return_dict", "vocab_size"], "openai/modeling_openai.py:OpenAIGPTLMHeadModel.prepare_inputs_for_generation": ["Any", "GenerationMixin", "LongTensor", "ModelGPTLMHeadModel", "ModelGPTPreTrainedModel", "class", "def", "dict", "for", "if", "in", "input_ids", "items", "key", "kwargs", "model_inputs", "not", "prepare_inputs_for_generation", "return", "self", "str", "torch", "value"], "openai/modeling_openai.py:OpenAIGPTDoubleHeadsModel.__init__": ["False", "Linear", "ModelGPTDoubleHeadsModel", "ModelGPTModel", "ModelGPTPreTrainedModel", "ModelGPTSequenceSummary", "__init__", "bias", "class", "config", "def", "lm_head", "multiple_choice_head", "n_embd", "nn", "num_labels", "post_init", "self", "super", "transformer", "vocab_size"], "openai/modeling_openai.py:OpenAIGPTDoubleHeadsModel.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "ModelGPTDoubleHeadsModel", "ModelGPTDoubleHeadsModelOutput", "ModelGPTPreTrainedModel", "None", "Tensor", "attention_mask", "attentions", "bool", "class", "config", "contiguous", "def", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "lm_head", "lm_logits", "lm_loss", "logits", "loss", "loss_fct", "mc_labels", "mc_logits", "mc_loss", "mc_token_ids", "multiple_choice_head", "not", "output", "output_attentions", "output_hidden_states", "position_ids", "r", "return", "return_dict", "self", "shift_labels", "shift_logits", "size", "squeeze", "token_type_ids", "torch", "transformer", "transformer_outputs", "tuple", "use_return_dict", "view"], "openai/modeling_openai.py:OpenAIGPTForSequenceClassification.__init__": ["False", "Linear", "ModelGPTForSequenceClassification", "ModelGPTModel", "ModelGPTPreTrainedModel", "__init__", "bias", "class", "config", "def", "n_embd", "nn", "num_labels", "post_init", "score", "self", "super", "transformer"], "openai/modeling_openai.py:OpenAIGPTForSequenceClassification.forward": ["BCEWithLogitsLoss", "Cannot", "CrossEntropyLoss", "FloatTensor", "LongTensor", "MSELoss", "ModelGPTForSequenceClassification", "ModelGPTPreTrainedModel", "None", "Results", "SequenceClassifierOutput", "Tensor", "ValueError", "__class__", "__name__", "and", "arange", "argmax", "attention_mask", "attentions", "batch", "batch_size", "be", "bool", "class", "config", "conjunction", "def", "defined", "detect", "device", "dtype", "elif", "else", "f", "forward", "handle", "hidden_states", "if", "in", "input_ids", "inputs_embeds", "int", "int32", "is", "kwargs", "labels", "last_non_pad_token", "logger", "logits", "long", "loss", "loss_fct", "may", "multi_label_classification", "no", "non_pad_mask", "not", "num_labels", "or", "output", "output_attentions", "output_hidden_states", "pad_token_id", "padding", "pooled_logits", "position_ids", "problem_type", "r", "raise", "regression", "return", "return_dict", "score", "self", "sequence_length", "shape", "single_label_classification", "sizes", "squeeze", "to", "token", "token_indices", "token_type_ids", "tokens", "torch", "transformer", "transformer_outputs", "tuple", "unexpected", "use_return_dict", "using", "view", "warning_once", "will", "with"], "opt/modeling_opt.py:OPTLearnedPositionalEmbedding.__init__": ["Embedding", "Model", "__init__", "class", "def", "embedding_dim", "int", "nn", "num_embeddings", "offset", "self", "super"], "opt/modeling_opt.py:OPTLearnedPositionalEmbedding.forward": ["Embedding", "LongTensor", "Model", "None", "attention_mask", "class", "cumsum", "def", "dim", "forward", "if", "int", "is", "long", "nn", "offset", "past_key_values_length", "position_ids", "return", "self", "super", "torch"], "opt/modeling_opt.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "attention_mask", "attn_output", "attn_weights", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "softmax", "to", "torch", "training", "transpose", "value"], "opt/modeling_opt.py:OPTAttention.__init__": ["Instantiating", "Linear", "Model", "ModelConfig", "Module", "None", "Please", "True", "ValueError", "__class__", "__init__", "__name__", "a", "and", "attention_dropout", "be", "bias", "by", "caching", "call", "class", "config", "creating", "def", "divisible", "dropout", "during", "embed_dim", "enable_bias", "errors", "f", "forward", "got", "head_dim", "hidden_size", "if", "int", "is", "is_causal", "k_proj", "kwargs", "layer_idx", "lead", "logger", "make", "must", "nn", "not", "num_attention_heads", "num_heads", "out_proj", "passing", "provide", "q_proj", "raise", "recommended", "scaling", "self", "super", "sure", "the", "this", "to", "used", "v_proj", "warning_once", "when", "will", "without"], "opt/modeling_opt.py:OPTAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "False", "Model", "Module", "None", "Tensor", "_", "_attn_implementation", "attention_interface", "attention_mask", "attn_output", "attn_weights", "bool", "bsz", "cache_position", "class", "config", "contiguous", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "head_dim", "hidden_states", "if", "is", "k_proj", "key_states", "kwargs", "layer_idx", "nn", "not", "num_heads", "out_proj", "output_attentions", "past_key_values", "q_proj", "query_states", "reshape", "return", "scaling", "self", "size", "tgt_len", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "view"], "opt/modeling_opt.py:OPTDecoderLayer.__init__": ["ACT2FN", "GradientCheckpointingLayer", "LayerNorm", "Linear", "Model", "ModelAttention", "ModelConfig", "None", "__init__", "activation_fn", "activation_function", "bias", "class", "config", "def", "do_layer_norm_before", "dropout", "elementwise_affine", "embed_dim", "enable_bias", "fc1", "fc2", "ffn_dim", "final_layer_norm", "hidden_size", "int", "layer_idx", "layer_norm_elementwise_affine", "nn", "self", "self_attn", "self_attn_layer_norm", "super"], "opt/modeling_opt.py:OPTDecoderLayer.forward": ["Cache", "False", "FlashAttentionKwargs", "FloatTensor", "GradientCheckpointingLayer", "LongTensor", "Model", "None", "Tensor", "Unpack", "activation_fn", "attention_mask", "bool", "cache_position", "class", "def", "do_layer_norm_before", "dropout", "fc1", "fc2", "final_layer_norm", "forward", "functional", "hidden_states", "hidden_states_shape", "if", "kwargs", "nn", "not", "output_attentions", "outputs", "p", "past_key_values", "position_ids", "reshape", "residual", "return", "self", "self_attn", "self_attn_layer_norm", "self_attn_weights", "shape", "size", "torch", "training", "tuple", "use_cache", "view"], "opt/modeling_opt.py:OPTDecoder.__init__": ["Embedding", "False", "LayerNorm", "Linear", "Model", "ModelConfig", "ModelLayer", "ModelLearnedPositionalEmbedding", "ModelPreTrainedModel", "ModuleList", "None", "__init__", "_remove_final_layer_norm", "and", "bias", "class", "config", "def", "do_layer_norm_before", "dropout", "elementwise_affine", "else", "embed_positions", "embed_tokens", "final_layer_norm", "for", "gradient_checkpointing", "hidden_size", "i", "if", "in", "layer_idx", "layer_norm_elementwise_affine", "layerdrop", "layers", "max_position_embeddings", "max_target_positions", "nn", "not", "num_hidden_layers", "pad_token_id", "padding_idx", "post_init", "project_in", "project_out", "range", "self", "super", "vocab_size", "word_embed_proj_dim"], "opt/modeling_opt.py:OPTDecoder._update_causal_mask": ["AttentionMaskConverter", "BlockMask", "Cache", "False", "Model", "ModelPreTrainedModel", "None", "Tensor", "Union", "_attn_implementation", "_ignore_causal_mask_sdpa", "_prepare_4d_causal_attention_mask_with_cache_position", "_unmask_unattended", "_update_causal_mask", "and", "any", "attention_mask", "batch_size", "bool", "cache_position", "causal_mask", "class", "config", "cuda", "def", "device", "dtype", "else", "finfo", "flash_attention_2", "flex_attention", "get_max_cache_shape", "get_seq_length", "if", "in", "input_tensor", "inputs_embeds", "is", "is_compileable", "is_training", "isinstance", "make_flex_block_causal_mask", "min", "min_dtype", "not", "npu", "output_attentions", "past_key_values", "past_key_values_length", "past_seen_tokens", "return", "sdpa", "self", "sequence_length", "shape", "target_length", "torch", "training", "type", "using_compilable_cache", "xpu"], "opt/modeling_opt.py:OPTDecoder._prepare_4d_causal_attention_mask_with_cache_position": ["Model", "ModelPreTrainedModel", "None", "Tensor", "_prepare_4d_causal_attention_mask_with_cache_position", "and", "arange", "attention_mask", "batch_size", "cache_position", "causal_mask", "class", "clone", "def", "device", "diagonal", "dim", "dtype", "else", "expand", "fill_value", "finfo", "full", "if", "int", "is", "kwargs", "mask_length", "masked_fill", "min", "min_dtype", "not", "padding_mask", "reshape", "return", "sequence_length", "shape", "target_length", "to", "torch", "triu"], "opt/modeling_opt.py:OPTDecoder.forward": ["BaseModelOutputWithPast", "Cache", "DynamicCache", "False", "FlashAttentionKwargs", "FloatTensor", "LongTensor", "Model", "ModelPreTrainedModel", "None", "Setting", "Tensor", "True", "Unpack", "ValueError", "You", "_update_causal_mask", "all_hidden_states", "all_self_attns", "and", "arange", "attention_mask", "attentions", "bool", "cache_position", "causal_mask", "checkpointing", "class", "config", "continue", "cumsum", "decoder_layer", "def", "device", "dim", "dropout_probability", "else", "embed_positions", "embed_tokens", "enumerate", "exactly", "final_layer_norm", "for", "forward", "get_seq_length", "gradient", "gradient_checkpointing", "hidden_states", "idx", "if", "in", "incompatible", "input_ids", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layer_outputs", "layerdrop", "layers", "logger", "long", "must", "not", "of", "one", "ones", "or", "output_attentions", "output_hidden_states", "past_key_values", "past_seen_tokens", "pos_embeds", "position_ids", "project_in", "project_out", "r", "raise", "rand", "return", "return_dict", "self", "seq_length", "shape", "specify", "to", "torch", "training", "tuple", "use_cache", "use_return_dict", "view", "warning_once", "with"], "opt/modeling_opt.py:OPTModel.__init__": ["Model", "ModelConfig", "ModelDecoder", "ModelPreTrainedModel", "__init__", "class", "config", "decoder", "def", "post_init", "self", "super"], "opt/modeling_opt.py:OPTModel.get_input_embeddings": ["Model", "ModelPreTrainedModel", "class", "decoder", "def", "embed_tokens", "get_input_embeddings", "return", "self"], "opt/modeling_opt.py:OPTModel.set_input_embeddings": ["Model", "ModelPreTrainedModel", "class", "decoder", "def", "embed_tokens", "self", "set_input_embeddings", "value"], "opt/modeling_opt.py:OPTModel.forward": ["BaseModelOutputWithPast", "Cache", "FlashAttentionKwargs", "FloatTensor", "LongTensor", "Model", "ModelPreTrainedModel", "None", "Tensor", "True", "Unpack", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "decoder", "decoder_outputs", "def", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "last_hidden_state", "not", "output_attentions", "output_hidden_states", "past_key_values", "position_ids", "return", "return_dict", "self", "torch", "tuple", "use_cache", "use_return_dict"], "opt/modeling_opt.py:OPTForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "Model", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "lm_head", "model", "nn", "post_init", "self", "super", "vocab_size", "word_embed_proj_dim"], "opt/modeling_opt.py:OPTForCausalLM.get_input_embeddings": ["GenerationMixin", "Model", "ModelPreTrainedModel", "class", "decoder", "def", "embed_tokens", "get_input_embeddings", "model", "return", "self"], "opt/modeling_opt.py:OPTForCausalLM.set_input_embeddings": ["GenerationMixin", "Model", "ModelPreTrainedModel", "class", "decoder", "def", "embed_tokens", "model", "self", "set_input_embeddings", "value"], "opt/modeling_opt.py:OPTForCausalLM.forward": ["BaseModelOutputWithPast", "Cache", "CausalLMOutputWithPast", "FloatTensor", "GenerationMixin", "LongTensor", "Model", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "contiguous", "decoder", "def", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "position_ids", "r", "return", "return_dict", "self", "slice", "slice_indices", "torch", "tuple", "use_cache", "use_return_dict", "vocab_size"], "opt/modeling_opt.py:OPTForSequenceClassification.__init__": ["False", "Linear", "Model", "ModelConfig", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "model", "nn", "num_labels", "post_init", "score", "self", "super", "word_embed_proj_dim"], "opt/modeling_opt.py:OPTForSequenceClassification.forward": ["BCEWithLogitsLoss", "Cache", "Cannot", "CrossEntropyLoss", "FloatTensor", "LongTensor", "MSELoss", "Model", "ModelPreTrainedModel", "None", "Results", "SequenceClassifierOutputWithPast", "ValueError", "__class__", "__name__", "and", "arange", "argmax", "attention_mask", "attentions", "batch", "batch_size", "be", "bool", "class", "config", "conjunction", "def", "defined", "detect", "device", "dtype", "elif", "else", "f", "forward", "handle", "hidden_states", "if", "in", "input_ids", "inputs_embeds", "int", "int32", "is", "kwargs", "labels", "last_non_pad_token", "logger", "logits", "long", "loss", "loss_fct", "may", "model", "multi_label_classification", "no", "non_pad_mask", "not", "num_labels", "or", "output", "output_attentions", "output_hidden_states", "pad_token_id", "padding", "past_key_values", "pooled_logits", "position_ids", "problem_type", "r", "raise", "regression", "return", "return_dict", "score", "self", "sequence_length", "shape", "single_label_classification", "sizes", "squeeze", "to", "token", "token_indices", "tokens", "torch", "transformer_outputs", "tuple", "unexpected", "use_cache", "use_return_dict", "using", "view", "warning_once", "will", "with"], "opt/modeling_opt.py:OPTForSequenceClassification.get_input_embeddings": ["Model", "ModelPreTrainedModel", "class", "decoder", "def", "embed_tokens", "get_input_embeddings", "model", "return", "self"], "opt/modeling_opt.py:OPTForSequenceClassification.set_input_embeddings": ["Model", "ModelPreTrainedModel", "class", "decoder", "def", "embed_tokens", "model", "self", "set_input_embeddings", "value"], "opt/modeling_opt.py:OPTForQuestionAnswering.__init__": ["Linear", "Model", "ModelConfig", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "config", "def", "model", "nn", "post_init", "qa_outputs", "self", "super", "word_embed_proj_dim"], "opt/modeling_opt.py:OPTForQuestionAnswering.forward": ["Cache", "CrossEntropyLoss", "FloatTensor", "LongTensor", "Model", "ModelPreTrainedModel", "None", "QuestionAnsweringModelOutput", "and", "attention_mask", "attentions", "bool", "clamp", "class", "config", "contiguous", "def", "device", "dim", "else", "end_logits", "end_loss", "end_positions", "forward", "hidden_states", "if", "ignore_index", "ignored_index", "input_ids", "inputs_embeds", "is", "kwargs", "len", "logits", "loss", "loss_fct", "model", "not", "output", "output_attentions", "output_hidden_states", "past_key_values", "position_ids", "qa_outputs", "r", "return", "return_dict", "self", "size", "split", "squeeze", "start_logits", "start_loss", "start_positions", "to", "torch", "total_loss", "transformer_outputs", "tuple", "use_cache", "use_return_dict"], "opt/modeling_opt.py:OPTForQuestionAnswering.get_input_embeddings": ["Model", "ModelPreTrainedModel", "class", "decoder", "def", "embed_tokens", "get_input_embeddings", "model", "return", "self"], "opt/modeling_opt.py:OPTForQuestionAnswering.set_input_embeddings": ["Model", "ModelPreTrainedModel", "class", "decoder", "def", "embed_tokens", "model", "self", "set_input_embeddings", "value"], "ovis2/modeling_ovis2.py:Ovis2RMSNorm.__init__": ["ModelRMSNorm", "Module", "Parameter", "__init__", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "ovis2/modeling_ovis2.py:Ovis2RMSNorm.forward": ["ModelRMSNorm", "Module", "True", "class", "def", "dtype", "float32", "forward", "hidden_states", "input_dtype", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "ovis2/modeling_ovis2.py:Ovis2RMSNorm.extra_repr": ["ModelRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "variance_epsilon", "weight"], "ovis2/modeling_ovis2.py:Ovis2VisionMLP.__init__": ["ACT2FN", "Linear", "ModelVisionMLP", "Module", "__init__", "act_fn", "bias", "class", "config", "def", "down_proj", "gate_proj", "hidden_act", "hidden_size", "intermediate_size", "mlp_bias", "nn", "self", "super", "up_proj"], "ovis2/modeling_ovis2.py:Ovis2VisionMLP.forward": ["ModelVisionMLP", "Module", "act_fn", "class", "def", "down_proj", "forward", "gate_proj", "nn", "return", "self", "up_proj", "x"], "ovis2/modeling_ovis2.py:Ovis2VisionEmbeddings.__init__": ["Conv2d", "Embedding", "False", "ModelRMSNorm", "ModelVisionConfig", "ModelVisionEmbeddings", "Module", "__init__", "arange", "class", "config", "def", "embed_dim", "expand", "hidden_size", "image_size", "in_channels", "kernel_size", "nn", "num_channels", "num_patches", "num_positions", "out_channels", "padding", "patch_embedding", "patch_size", "persistent", "position_embedding", "position_ids", "register_buffer", "rms_norm", "rms_norm_eps", "self", "stride", "super", "torch", "valid"], "ovis2/modeling_ovis2.py:Ovis2VisionEmbeddings.forward": ["FloatTensor", "ModelVisionEmbeddings", "Module", "Tensor", "class", "def", "dtype", "embeddings", "flatten", "forward", "nn", "patch_embedding", "patch_embeds", "pixel_values", "position_embedding", "position_ids", "return", "rms_norm", "self", "target_dtype", "to", "torch", "transpose", "weight"], "ovis2/modeling_ovis2.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "attention_mask", "attn_output", "attn_weights", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "softmax", "to", "torch", "training", "transpose", "value"], "ovis2/modeling_ovis2.py:Ovis2VisionAttention.__init__": ["False", "Linear", "ModelVisionAttention", "Module", "ValueError", "__init__", "and", "attention_dropout", "be", "bias", "by", "class", "config", "def", "divisible", "dropout", "embed_dim", "f", "got", "head_dim", "hidden_size", "if", "is_causal", "k_proj", "must", "nn", "num_attention_heads", "num_heads", "out_proj", "q_proj", "qkv_bias", "raise", "scale", "self", "super", "v_proj"], "ovis2/modeling_ovis2.py:Ovis2VisionAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "ModelVisionAttention", "Module", "None", "Tensor", "_attn_implementation", "attention_interface", "attention_mask", "attn_output", "attn_weights", "batch_size", "class", "config", "contiguous", "def", "dropout", "eager", "eager_attention_forward", "else", "embed_dim", "forward", "head_dim", "hidden_states", "if", "is_causal", "k_proj", "keys", "kwargs", "nn", "not", "num_heads", "out_proj", "q_proj", "queries", "reshape", "return", "scale", "scaling", "self", "seq_length", "shape", "torch", "training", "transpose", "tuple", "v_proj", "values", "view"], "ovis2/modeling_ovis2.py:Ovis2MLP.__init__": ["ACT2FN", "Linear", "ModelMLP", "Module", "__init__", "act_fn", "bias", "class", "config", "def", "down_proj", "gate_proj", "hidden_act", "hidden_size", "intermediate_size", "mlp_bias", "nn", "self", "super", "up_proj"], "ovis2/modeling_ovis2.py:Ovis2MLP.forward": ["ModelMLP", "Module", "act_fn", "class", "def", "down_proj", "forward", "gate_proj", "nn", "return", "self", "up_proj", "x"], "ovis2/modeling_ovis2.py:Ovis2Attention.__init__": ["False", "Linear", "ModelAttention", "Module", "ValueError", "__init__", "and", "attention_dropout", "be", "bias", "by", "class", "config", "def", "divisible", "dropout", "embed_dim", "f", "got", "head_dim", "hidden_size", "if", "is_causal", "k_proj", "must", "nn", "num_attention_heads", "num_heads", "out_proj", "q_proj", "qkv_bias", "raise", "scale", "self", "super", "v_proj"], "ovis2/modeling_ovis2.py:Ovis2Attention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "ModelAttention", "Module", "None", "Tensor", "_attn_implementation", "attention_interface", "attention_mask", "attn_output", "attn_weights", "batch_size", "class", "config", "contiguous", "def", "dropout", "eager", "eager_attention_forward", "else", "embed_dim", "forward", "head_dim", "hidden_states", "if", "is_causal", "k_proj", "keys", "kwargs", "nn", "not", "num_heads", "out_proj", "q_proj", "queries", "reshape", "return", "scale", "scaling", "self", "seq_length", "shape", "torch", "training", "transpose", "tuple", "v_proj", "values", "view"], "ovis2/modeling_ovis2.py:Ovis2VisionEncoderLayer.__init__": ["GradientCheckpointingLayer", "ModelAttention", "ModelMLP", "ModelRMSNorm", "ModelVisionConfig", "ModelVisionEncoderLayer", "__init__", "attention", "class", "config", "def", "ffn", "hidden_size", "rms_norm1", "rms_norm2", "rms_norm_eps", "self", "super"], "ovis2/modeling_ovis2.py:Ovis2VisionEncoderLayer.forward": ["GradientCheckpointingLayer", "ModelVisionEncoderLayer", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "attention", "attention_mask", "attn_output", "class", "def", "ffn", "forward", "hidden_states", "kwargs", "mlp_output", "norm_hidden_states", "return", "rms_norm1", "rms_norm2", "self", "torch"], "ovis2/modeling_ovis2.py:Ovis2VisionEncoder.__init__": ["False", "ModelVisionConfig", "ModelVisionEncoder", "ModelVisionEncoderLayer", "Module", "ModuleList", "_", "__init__", "class", "config", "def", "for", "gradient_checkpointing", "in", "layers", "nn", "num_hidden_layers", "range", "self", "super"], "ovis2/modeling_ovis2.py:Ovis2VisionEncoder.forward": ["BaseModelOutput", "ModelVisionEncoder", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "class", "def", "encoder_layer", "for", "forward", "hidden_states", "in", "inputs_embeds", "kwargs", "last_hidden_state", "layers", "nn", "return", "self", "torch"], "ovis2/modeling_ovis2.py:Ovis2VisionTransformer.__init__": ["False", "ModelRMSNorm", "ModelVisionConfig", "ModelVisionEmbeddings", "ModelVisionEncoder", "ModelVisionTransformer", "Module", "__init__", "class", "config", "def", "embeddings", "encoder", "gradient_checkpointing", "hidden_size", "nn", "rms_norm", "rms_norm_eps", "self", "super"], "ovis2/modeling_ovis2.py:Ovis2VisionTransformer.forward": ["BaseModelOutput", "ModelVisionTransformer", "Module", "None", "Tensor", "attention_mask", "class", "def", "embeddings", "encoder", "encoder_outputs", "forward", "hidden_states", "inputs_embeds", "kwargs", "last_hidden_state", "nn", "pixel_values", "return", "rms_norm", "self", "torch"], "ovis2/modeling_ovis2.py:Ovis2VisualEmbeddingTable.forward": ["Embedding", "ModelVisualEmbeddingTable", "Tensor", "class", "def", "dtype", "forward", "if", "in", "int16", "int32", "int64", "int8", "long", "matmul", "nn", "return", "self", "super", "torch", "visual_tokens", "weight"], "ovis2/modeling_ovis2.py:Ovis2PreTrainedModel._init_weights": ["ModelPreTrainedModel", "ModelVisionEmbeddings", "PreTrainedModel", "_init_weights", "arange", "class", "copy_", "def", "expand", "if", "init", "isinstance", "module", "position_ids", "self", "shape", "super", "torch"], "ovis2/modeling_ovis2.py:hard_softmax": ["Model_softmax", "Tensor", "True", "def", "detach", "dim", "index", "int", "keepdim", "legacy_contiguous_format", "logits", "max", "memory_format", "ret", "return", "scatter_", "softmax", "torch", "y_Model", "y_soft", "zeros_like"], "ovis2/modeling_ovis2.py:Ovis2VisionModel.__init__": ["False", "LayerNorm", "Linear", "ModelPreTrainedModel", "ModelVisionConfig", "ModelVisionModel", "ModelVisionTransformer", "__init__", "bias", "class", "config", "def", "head_linear", "head_norm", "hidden_size", "hidden_stride", "nn", "num_visual_indicator_tokens", "post_init", "self", "super", "transformer", "vocab_size"], "ovis2/modeling_ovis2.py:Ovis2VisionModel.forward": ["FloatTensor", "ModelPreTrainedModel", "ModelVisionModel", "Tensor", "Token", "True", "ValueError", "a", "be", "class", "config", "constant", "def", "dim", "elif", "forward", "functional", "gumbel_argmax", "gumbel_softmax", "hard", "hard_softmax", "head_linear", "head_norm", "hidden_dim", "hidden_stride", "if", "int", "kwargs", "last_hidden_state", "length", "logits", "math", "must", "nn", "num_images", "outputs", "pad", "pad_size", "perfect", "permute", "pixel_values", "prob_token", "raise", "reshape", "return", "self", "seq_len", "sequence", "shape", "softmax", "sqrt", "sqrt_l", "square", "st_argmax", "tokenize_function", "torch", "transformer", "tuple"], "ovis2/modeling_ovis2.py:Ovis2Model.__init__": ["AutoModel", "ModelConfig", "ModelModel", "ModelPreTrainedModel", "ModelVisionModel", "ModelVisualEmbeddingTable", "__init__", "class", "config", "def", "from_config", "hidden_size", "language_model", "post_init", "self", "super", "text_config", "vision_config", "vision_tower", "visual_embeddings_table", "visual_indicator_token_ids", "visual_vocab_size", "vocab_size"], "ovis2/modeling_ovis2.py:Ovis2Model.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "language_model", "return", "self"], "ovis2/modeling_ovis2.py:Ovis2Model.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "language_model", "self", "set_input_embeddings", "value"], "ovis2/modeling_ovis2.py:Ovis2Model.get_image_features": ["False", "FloatTensor", "ModelModel", "ModelPreTrainedModel", "_", "arange", "batch_size", "cat", "class", "def", "device", "dim", "dtype", "get_image_features", "image_features", "img_seq_len", "layout", "long", "num_visual_indicator_tokens", "padding_tensor", "pixel_values", "requires_grad", "return", "self", "shape", "to", "torch", "vision_tower", "visual_embeddings_table", "visual_indicator", "visual_indicator_features", "visual_vocab_size", "zeros"], "ovis2/modeling_ovis2.py:Ovis2Model.get_placeholder_mask": ["FloatTensor", "Image", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "ValueError", "all", "and", "class", "config", "def", "device", "do", "dtype", "else", "expand_as", "f", "features", "get_input_embeddings", "get_placeholder_mask", "if", "image", "image_features", "image_token_id", "input_ids", "inputs_embeds", "is", "long", "match", "n_image_features", "n_image_tokens", "not", "numel", "raise", "return", "self", "shape", "special_image_mask", "sum", "tensor", "to", "tokens", "torch", "unsqueeze"], "ovis2/modeling_ovis2.py:Ovis2Model.forward": ["Cache", "FloatTensor", "LongTensor", "ModelModel", "ModelModelOutputWithPast", "ModelPreTrainedModel", "None", "Tensor", "True", "ValueError", "You", "all", "any", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "device", "dtype", "else", "enumerate", "exactly", "expand_as", "for", "forward", "get_image_features", "get_input_embeddings", "get_placeholder_mask", "hidden_states", "i", "if", "image_features", "image_hidden_states", "in", "input_ids", "inputs_embeds", "int", "is", "kwargs", "labels", "language_model", "last_hidden_state", "logits_to_keep", "long", "mask", "masked_scatter", "must", "not", "of", "one", "or", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "pixel_values", "position_ids", "raise", "return", "return_dict", "self", "special_image_mask", "specify", "tensor", "to", "torch", "tuple", "use_cache", "visual_indicator_features", "visual_indicator_id", "visual_indicator_token_ids"], "ovis2/modeling_ovis2.py:Ovis2ForConditionalGeneration.__init__": ["False", "GenerationMixin", "Linear", "ModelConfig", "ModelForConditionalGeneration", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "vocab_size"], "ovis2/modeling_ovis2.py:Ovis2ForConditionalGeneration.get_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "model", "return", "self"], "ovis2/modeling_ovis2.py:Ovis2ForConditionalGeneration.set_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "model", "self", "set_input_embeddings", "value"], "ovis2/modeling_ovis2.py:Ovis2ForConditionalGeneration.get_output_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "Module", "class", "def", "get_output_embeddings", "lm_head", "nn", "return", "self"], "ovis2/modeling_ovis2.py:Ovis2ForConditionalGeneration.get_image_features": ["FloatTensor", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "get_image_features", "model", "pixel_values", "return", "self", "torch"], "ovis2/modeling_ovis2.py:Ovis2ForConditionalGeneration.forward": ["Cache", "FloatTensor", "GenerationMixin", "LongTensor", "ModelCausalLMOutputWithPast", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Tensor", "True", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "else", "forward", "hidden_states", "if", "image_hidden_states", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "pixel_values", "position_ids", "r", "return", "return_dict", "self", "slice", "slice_indices", "text_config", "torch", "tuple", "use_cache", "vocab_size"], "ovis2/modeling_ovis2.py:Ovis2ForConditionalGeneration.prepare_inputs_for_generation": ["False", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "True", "attention_mask", "cache_position", "class", "def", "get", "if", "input_ids", "inputs_embeds", "is_first_iteration", "kwargs", "logits_to_keep", "model_inputs", "not", "or", "past_key_values", "pixel_values", "prepare_inputs_for_generation", "return", "self", "super", "use_cache"], "owlv2/modeling_owlv2.py:contrastive_loss": ["Model_loss", "Tensor", "arange", "cross_entropy", "def", "device", "functional", "len", "logits", "nn", "return", "torch"], "owlv2/modeling_owlv2.py:owlv2_loss": ["Model_loss", "Tensor", "caption_loss", "contrastive_loss", "def", "image_loss", "return", "similarity", "t", "torch"], "owlv2/modeling_owlv2.py:Owlv2Output.to_tuple": ["Any", "ModelOutput", "class", "def", "else", "for", "getattr", "if", "in", "k", "keys", "not", "return", "self", "text_model_output", "to_tuple", "tuple", "vision_model_output"], "owlv2/modeling_owlv2.py:_upcast": ["Tensor", "_upcast", "def", "dtype", "else", "float", "float32", "float64", "if", "in", "int", "int32", "int64", "is_floating_point", "return", "t", "torch"], "owlv2/modeling_owlv2.py:box_area": ["Model_area", "Modeles", "Tensor", "_upcast", "def", "return"], "owlv2/modeling_owlv2.py:box_iou": ["Model_area", "Model_iou", "Modeles1", "Modeles2", "None", "area1", "area2", "clamp", "def", "inter", "iou", "left_top", "max", "min", "return", "right_bottom", "torch", "union", "width_height"], "owlv2/modeling_owlv2.py:generalized_box_iou": ["Model_box_iou", "None", "ValueError", "all", "area", "be", "bottom_right", "box_iou", "boxes1", "boxes2", "but", "clamp", "corner", "def", "f", "format", "got", "if", "in", "iou", "max", "min", "must", "not", "raise", "return", "top_left", "torch", "union", "width_height", "x0", "x1", "y0", "y1"], "owlv2/modeling_owlv2.py:Owlv2ObjectDetectionOutput.to_tuple": ["Any", "ModelObjectDetectionOutput", "ModelOutput", "class", "def", "else", "for", "getattr", "if", "in", "k", "keys", "not", "return", "self", "text_model_output", "to_tuple", "tuple", "vision_model_output"], "owlv2/modeling_owlv2.py:Owlv2ImageGuidedObjectDetectionOutput.to_tuple": ["Any", "ModelImageGuidedObjectDetectionOutput", "ModelOutput", "class", "def", "else", "for", "getattr", "if", "in", "k", "keys", "not", "return", "self", "text_model_output", "to_tuple", "tuple", "vision_model_output"], "owlv2/modeling_owlv2.py:Owlv2VisionEmbeddings.__init__": ["Conv2d", "Embedding", "False", "ModelVisionConfig", "ModelVisionEmbeddings", "Module", "Parameter", "__init__", "arange", "bias", "class", "class_embedding", "config", "def", "embed_dim", "expand", "hidden_size", "image_size", "in_channels", "kernel_size", "nn", "num_channels", "num_patches", "num_positions", "out_channels", "patch_embedding", "patch_size", "persistent", "position_embedding", "position_ids", "randn", "register_buffer", "self", "stride", "super", "torch"], "owlv2/modeling_owlv2.py:Owlv2VisionEmbeddings.interpolate_pos_encoding": ["False", "ModelVisionEmbeddings", "Module", "Tensor", "align_corners", "and", "bicubic", "cat", "class", "class_pos_embed", "def", "dim", "embeddings", "functional", "height", "if", "int", "interpolate", "interpolate_pos_encoding", "is_tracing", "jit", "mode", "new_height", "new_width", "nn", "not", "num_patches", "num_positions", "patch_pos_embed", "patch_size", "permute", "position_embedding", "position_ids", "reshape", "return", "self", "shape", "size", "sqrt_num_positions", "torch", "torch_int", "unsqueeze", "view", "weight", "width"], "owlv2/modeling_owlv2.py:Owlv2VisionEmbeddings.forward": ["False", "FloatTensor", "ModelVisionEmbeddings", "Module", "Tensor", "_", "batch_size", "bool", "cat", "class", "class_embedding", "class_embeds", "def", "dim", "else", "embeddings", "expand", "flatten", "forward", "height", "if", "interpolate_pos_encoding", "nn", "patch_embedding", "patch_embeds", "pixel_values", "position_embedding", "position_ids", "return", "self", "shape", "torch", "transpose", "width"], "owlv2/modeling_owlv2.py:Owlv2TextEmbeddings.__init__": ["Embedding", "False", "ModelTextConfig", "ModelTextEmbeddings", "Module", "__init__", "arange", "class", "config", "def", "expand", "hidden_size", "max_position_embeddings", "nn", "persistent", "position_embedding", "position_ids", "register_buffer", "self", "super", "token_embedding", "torch", "vocab_size"], "owlv2/modeling_owlv2.py:Owlv2TextEmbeddings.forward": ["FloatTensor", "LongTensor", "ModelTextEmbeddings", "Module", "None", "Tensor", "class", "def", "else", "embeddings", "forward", "if", "input_ids", "inputs_embeds", "is", "nn", "not", "position_embedding", "position_embeddings", "position_ids", "return", "self", "seq_length", "shape", "token_embedding", "torch"], "owlv2/modeling_owlv2.py:Owlv2Attention.__init__": ["Linear", "ModelAttention", "Module", "ValueError", "__init__", "and", "attention_dropout", "be", "by", "class", "config", "def", "divisible", "dropout", "embed_dim", "f", "got", "head_dim", "hidden_size", "if", "k_proj", "must", "nn", "num_attention_heads", "num_heads", "out_proj", "q_proj", "raise", "scale", "self", "super", "v_proj"], "owlv2/modeling_owlv2.py:Owlv2Attention._shape": ["ModelAttention", "Module", "Tensor", "_shape", "bsz", "class", "contiguous", "def", "head_dim", "int", "nn", "num_heads", "return", "self", "seq_len", "tensor", "torch", "transpose", "view"], "owlv2/modeling_owlv2.py:Owlv2Attention.forward": ["Attention", "False", "ModelAttention", "Module", "None", "Tensor", "ValueError", "_shape", "attention_mask", "attn_output", "attn_probs", "attn_weights", "attn_weights_reshaped", "be", "bmm", "bool", "bsz", "but", "causal_attention_mask", "class", "def", "dim", "dropout", "dtype", "else", "embed_dim", "f", "forward", "functional", "head_dim", "hidden_states", "if", "is", "k_proj", "key_states", "mask", "nn", "not", "num_heads", "of", "out_proj", "output_attentions", "p", "proj_shape", "q_proj", "query_states", "raise", "reshape", "return", "scale", "self", "should", "size", "softmax", "src_len", "tgt_len", "to", "torch", "training", "transpose", "tuple", "v_proj", "value_states", "view", "weights"], "owlv2/modeling_owlv2.py:Owlv2MLP.__init__": ["ACT2FN", "Linear", "ModelMLP", "Module", "__init__", "activation_fn", "class", "config", "def", "fc1", "fc2", "hidden_act", "hidden_size", "intermediate_size", "nn", "self", "super"], "owlv2/modeling_owlv2.py:Owlv2MLP.forward": ["ModelMLP", "Module", "Tensor", "activation_fn", "class", "def", "fc1", "fc2", "forward", "hidden_states", "nn", "return", "self", "torch"], "owlv2/modeling_owlv2.py:Owlv2EncoderLayer.__init__": ["GradientCheckpointingLayer", "LayerNorm", "ModelAttention", "ModelConfig", "ModelEncoderLayer", "ModelMLP", "__init__", "class", "config", "def", "embed_dim", "eps", "hidden_size", "layer_norm1", "layer_norm2", "layer_norm_eps", "mlp", "nn", "self", "self_attn", "super"], "owlv2/modeling_owlv2.py:Owlv2EncoderLayer.forward": ["False", "FloatTensor", "GradientCheckpointingLayer", "ModelEncoderLayer", "None", "Tensor", "attention_mask", "attn_weights", "bool", "causal_attention_mask", "class", "def", "forward", "hidden_states", "if", "layer_norm1", "layer_norm2", "mlp", "output_attentions", "outputs", "residual", "return", "self", "self_attn", "torch", "tuple"], "owlv2/modeling_owlv2.py:Owlv2PreTrainedModel._init_weights": ["LayerNorm", "Linear", "ModelAttention", "ModelForObjectDetection", "ModelMLP", "ModelModel", "ModelPreTrainedModel", "ModelTextEmbeddings", "ModelVisionEmbeddings", "Module", "None", "PreTrainedModel", "_init_weights", "arange", "bias", "box_bias", "class", "class_embedding", "compute_box_bias", "config", "constant_", "copy_", "def", "elif", "embed_dim", "expand", "factor", "fc1", "fc2", "fc_std", "hidden_size", "if", "in_proj_std", "init", "initializer_factor", "initializer_range", "is", "isinstance", "k_proj", "logit_scale", "logit_scale_init_value", "mean", "module", "nn", "normal_", "not", "num_hidden_layers", "num_patches_height", "num_patches_width", "ones_", "out_proj", "out_proj_std", "patch_embedding", "position_embedding", "position_ids", "q_proj", "self", "shape", "std", "text_embed_dim", "text_projection", "token_embedding", "torch", "v_proj", "vision_embed_dim", "visual_projection", "weight", "zeros_"], "owlv2/modeling_owlv2.py:Owlv2Encoder.__init__": ["False", "ModelConfig", "ModelEncoder", "ModelEncoderLayer", "Module", "ModuleList", "_", "__init__", "class", "config", "def", "for", "gradient_checkpointing", "in", "layers", "nn", "num_hidden_layers", "range", "self", "super"], "owlv2/modeling_owlv2.py:Owlv2Encoder.forward": ["BaseModelOutput", "ModelEncoder", "Module", "None", "Tensor", "all_attentions", "attention_mask", "attentions", "bool", "causal_attention_mask", "class", "config", "def", "else", "encoder_layer", "encoder_states", "for", "forward", "hidden_states", "if", "in", "inputs_embeds", "is", "last_hidden_state", "layer_outputs", "layers", "nn", "not", "output_attentions", "output_hidden_states", "r", "return", "return_dict", "self", "torch", "tuple", "use_return_dict", "v"], "owlv2/modeling_owlv2.py:Owlv2TextTransformer.__init__": ["LayerNorm", "ModelEncoder", "ModelTextConfig", "ModelTextEmbeddings", "ModelTextTransformer", "Module", "__init__", "class", "config", "def", "embed_dim", "embeddings", "encoder", "eps", "final_layer_norm", "hidden_size", "layer_norm_eps", "nn", "self", "super"], "owlv2/modeling_owlv2.py:Owlv2TextTransformer.forward": ["BaseModelOutputWithPooling", "ModelTextTransformer", "Module", "None", "Tensor", "_create_4d_causal_attention_mask", "_prepare_4d_attention_mask", "arange", "argmax", "attention_mask", "attentions", "bool", "causal_attention_mask", "class", "config", "def", "device", "dim", "dtype", "else", "embeddings", "encoder", "encoder_outputs", "final_layer_norm", "forward", "hidden_states", "if", "input_ids", "input_shape", "inputs_embeds", "int", "is", "last_hidden_state", "nn", "not", "output_attentions", "output_hidden_states", "pooled_output", "pooler_output", "position_ids", "r", "return", "return_dict", "self", "shape", "size", "to", "torch", "tuple", "use_return_dict", "view"], "owlv2/modeling_owlv2.py:Owlv2TextModel.__init__": ["ModelPreTrainedModel", "ModelTextConfig", "ModelTextModel", "ModelTextTransformer", "__init__", "class", "config", "def", "post_init", "self", "super", "text_model"], "owlv2/modeling_owlv2.py:Owlv2TextModel.get_input_embeddings": ["ModelPreTrainedModel", "ModelTextModel", "Module", "class", "def", "embeddings", "get_input_embeddings", "nn", "return", "self", "text_model", "token_embedding"], "owlv2/modeling_owlv2.py:Owlv2TextModel.set_input_embeddings": ["ModelPreTrainedModel", "ModelTextModel", "class", "def", "embeddings", "self", "set_input_embeddings", "text_model", "token_embedding", "value"], "owlv2/modeling_owlv2.py:Owlv2TextModel.forward": ["BaseModelOutputWithPooling", "ModelPreTrainedModel", "ModelTextModel", "None", "Tensor", "attention_mask", "bool", "class", "def", "forward", "input_ids", "kwargs", "output_attentions", "output_hidden_states", "r", "return", "return_dict", "self", "text_model", "torch", "tuple"], "owlv2/modeling_owlv2.py:Owlv2VisionTransformer.__init__": ["LayerNorm", "ModelEncoder", "ModelVisionConfig", "ModelVisionEmbeddings", "ModelVisionTransformer", "Module", "__init__", "class", "config", "def", "embeddings", "encoder", "eps", "hidden_size", "layer_norm_eps", "nn", "post_layernorm", "pre_layernorm", "self", "super"], "owlv2/modeling_owlv2.py:Owlv2VisionTransformer.forward": ["BaseModelOutputWithPooling", "False", "FloatTensor", "ModelVisionTransformer", "Module", "None", "attentions", "bool", "class", "config", "def", "dtype", "else", "embeddings", "encoder", "encoder_outputs", "expected_input_dtype", "forward", "hidden_states", "if", "inputs_embeds", "interpolate_pos_encoding", "is", "last_hidden_state", "nn", "not", "output_attentions", "output_hidden_states", "patch_embedding", "pixel_values", "pooled_output", "pooler_output", "post_layernorm", "pre_layernorm", "return", "return_dict", "self", "to", "torch", "tuple", "use_return_dict", "weight"], "owlv2/modeling_owlv2.py:Owlv2VisionModel.__init__": ["ModelPreTrainedModel", "ModelVisionConfig", "ModelVisionModel", "ModelVisionTransformer", "__init__", "class", "config", "def", "post_init", "self", "super", "vision_model"], "owlv2/modeling_owlv2.py:Owlv2VisionModel.get_input_embeddings": ["ModelPreTrainedModel", "ModelVisionModel", "Module", "class", "def", "embeddings", "get_input_embeddings", "nn", "patch_embedding", "return", "self", "vision_model"], "owlv2/modeling_owlv2.py:Owlv2VisionModel.forward": ["BaseModelOutputWithPooling", "False", "FloatTensor", "ModelPreTrainedModel", "ModelVisionModel", "None", "bool", "class", "def", "forward", "interpolate_pos_encoding", "kwargs", "output_attentions", "output_hidden_states", "pixel_values", "r", "return", "return_dict", "self", "torch", "tuple", "vision_model"], "owlv2/modeling_owlv2.py:Owlv2Model.__init__": ["False", "Linear", "ModelConfig", "ModelModel", "ModelPreTrainedModel", "ModelTextConfig", "ModelTextTransformer", "ModelVisionConfig", "ModelVisionTransformer", "Parameter", "TypeError", "__init__", "be", "bias", "but", "class", "config", "def", "expected", "f", "hidden_size", "if", "is", "isinstance", "logit_scale", "logit_scale_init_value", "nn", "not", "of", "post_init", "projection_dim", "raise", "self", "super", "tensor", "text_config", "text_embed_dim", "text_model", "text_projection", "to", "torch", "type", "vision_config", "vision_embed_dim", "vision_model", "visual_projection"], "owlv2/modeling_owlv2.py:Owlv2Model.get_text_features": ["BaseModelOutputWithPooling", "FloatTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "attention_mask", "class", "def", "get_text_features", "input_ids", "pooler_output", "r", "return", "self", "text_features", "text_model", "text_outputs", "text_projection", "torch"], "owlv2/modeling_owlv2.py:Owlv2Model.get_image_features": ["BaseModelOutputWithPooling", "False", "FloatTensor", "ModelModel", "ModelPreTrainedModel", "Tensor", "bool", "class", "def", "get_image_features", "image_features", "interpolate_pos_encoding", "pixel_values", "pooler_output", "r", "return", "self", "torch", "vision_model", "vision_outputs", "visual_projection"], "owlv2/modeling_owlv2.py:Owlv2Model.forward": ["False", "FloatTensor", "LongTensor", "ModelModel", "ModelOutput", "ModelPreTrainedModel", "Model_loss", "None", "Tensor", "True", "attention_mask", "bool", "class", "config", "def", "device", "dim", "else", "exp", "forward", "if", "image_embeds", "input_ids", "interpolate_pos_encoding", "is", "keepdim", "kwargs", "linalg", "logit_scale", "logits_per_image", "logits_per_text", "loss", "matmul", "norm", "not", "ord", "output", "output_attentions", "output_hidden_states", "pixel_values", "r", "return", "return_base_image_embeds", "return_dict", "return_loss", "self", "t", "text_embeds", "text_embeds_norm", "text_model", "text_model_output", "text_outputs", "text_projection", "to", "torch", "tuple", "use_return_dict", "vision_model", "vision_model_output", "vision_outputs", "visual_projection"], "owlv2/modeling_owlv2.py:Owlv2BoxPredictionHead.__init__": ["GELU", "Linear", "ModelBoxPredictionHead", "ModelConfig", "Module", "__init__", "class", "config", "def", "dense0", "dense1", "dense2", "gelu", "hidden_size", "int", "nn", "out_dim", "self", "super", "vision_config", "width"], "owlv2/modeling_owlv2.py:Owlv2BoxPredictionHead.forward": ["FloatTensor", "ModelBoxPredictionHead", "Module", "Tensor", "class", "def", "dense0", "dense1", "dense2", "forward", "gelu", "image_features", "nn", "output", "return", "self", "torch"], "owlv2/modeling_owlv2.py:Owlv2ClassPredictionHead.__init__": ["ELU", "Linear", "ModelClassPredictionHead", "ModelConfig", "Module", "__init__", "class", "config", "def", "dense0", "elu", "hidden_size", "logit_scale", "logit_shift", "nn", "out_dim", "query_dim", "self", "super", "text_config", "vision_config"], "owlv2/modeling_owlv2.py:Owlv2ClassPredictionHead.forward": ["FloatTensor", "ModelClassPredictionHead", "Module", "None", "Tensor", "True", "batch_size", "class", "def", "dense0", "device", "dim", "dtype", "einsum", "elu", "finfo", "float32", "forward", "if", "image_class_embeds", "image_embeds", "is", "keepdim", "linalg", "logit_scale", "logit_shift", "min", "ndim", "nn", "norm", "not", "num_patches", "pd", "pq", "pred_logits", "qd", "query_dim", "query_embeds", "query_mask", "return", "self", "shape", "to", "torch", "tuple", "unsqueeze", "where", "zeros"], "owlv2/modeling_owlv2.py:Owlv2ForObjectDetection.__init__": ["False", "LayerNorm", "Model", "ModelBoxPredictionHead", "ModelClassPredictionHead", "ModelConfig", "ModelForObjectDetection", "ModelModel", "ModelPreTrainedModel", "Sigmoid", "__init__", "box_bias", "box_head", "class", "class_head", "compute_box_bias", "config", "def", "eps", "hidden_size", "image_size", "layer_norm", "layer_norm_eps", "nn", "num_patches_height", "num_patches_width", "objectness_head", "out_dim", "patch_size", "persistent", "post_init", "register_buffer", "self", "sigmoid", "super", "vision_config"], "owlv2/modeling_owlv2.py:Owlv2ForObjectDetection.normalize_grid_corner_coordinates": ["ModelForObjectDetection", "ModelPreTrainedModel", "Tensor", "arange", "box_coordinates", "class", "def", "dim", "dtype", "float32", "indexing", "int", "meshgrid", "normalize_grid_corner_coordinates", "num_patches_height", "num_patches_width", "return", "stack", "torch", "view", "x_coordinates", "xx", "xy", "y_coordinates", "yy"], "owlv2/modeling_owlv2.py:Owlv2ForObjectDetection.objectness_predictor": ["FloatTensor", "ModelForObjectDetection", "ModelPreTrainedModel", "class", "def", "detach", "image_features", "objectness_head", "objectness_logits", "objectness_predictor", "return", "self", "torch"], "owlv2/modeling_owlv2.py:Owlv2ForObjectDetection.compute_box_bias": ["ModelForObjectDetection", "ModelPreTrainedModel", "Tensor", "box_bias", "box_coord_bias", "box_coordinates", "box_size", "box_size_bias", "cat", "class", "clip", "compute_box_bias", "def", "dim", "full_like", "int", "log", "log1p", "normalize_grid_corner_coordinates", "num_patches_height", "num_patches_width", "return", "self", "torch"], "owlv2/modeling_owlv2.py:Owlv2ForObjectDetection.box_predictor": ["False", "FloatTensor", "ModelForObjectDetection", "ModelPreTrainedModel", "_", "bool", "box_bias", "box_head", "box_predictor", "class", "compute_box_bias", "def", "device", "else", "feature_map", "if", "image_feats", "interpolate_pos_encoding", "num_patches_height", "num_patches_width", "pred_boxes", "return", "self", "shape", "sigmoid", "to", "torch"], "owlv2/modeling_owlv2.py:Owlv2ForObjectDetection.class_predictor": ["FloatTensor", "ModelForObjectDetection", "ModelPreTrainedModel", "None", "Tensor", "class", "class_head", "class_predictor", "def", "image_class_embeds", "image_feats", "pred_logits", "query_embeds", "query_mask", "return", "self", "torch", "tuple"], "owlv2/modeling_owlv2.py:Owlv2ForObjectDetection.image_text_embedder": ["False", "FloatTensor", "Model", "ModelForObjectDetection", "ModelPreTrainedModel", "None", "Tensor", "True", "_", "attention_mask", "bool", "broadcast_to", "class", "class_token_out", "config", "def", "else", "height", "if", "image_embeds", "image_text_embedder", "input_ids", "interpolate_pos_encoding", "last_hidden_state", "layer_norm", "new_size", "num_patches_height", "num_patches_width", "output_attentions", "output_hidden_states", "outputs", "patch_size", "pixel_values", "post_layernorm", "reshape", "return", "return_dict", "self", "shape", "text_embeds", "torch", "tuple", "vision_config", "vision_model", "vision_model_output", "width"], "owlv2/modeling_owlv2.py:Owlv2ForObjectDetection.image_embedder": ["False", "FloatTensor", "Model", "ModelForObjectDetection", "ModelPreTrainedModel", "None", "True", "_", "bool", "broadcast_to", "class", "class_token_out", "config", "def", "else", "height", "if", "image_embedder", "image_embeds", "interpolate_pos_encoding", "last_hidden_state", "layer_norm", "new_size", "num_patches_height", "num_patches_width", "output_attentions", "output_hidden_states", "patch_size", "pixel_values", "post_layernorm", "reshape", "return", "return_dict", "self", "shape", "torch", "tuple", "vision_config", "vision_model", "vision_outputs", "width"], "owlv2/modeling_owlv2.py:Owlv2ForObjectDetection.embed_image_query": ["False", "FloatTensor", "ModelForObjectDetection", "ModelPreTrainedModel", "None", "_", "all", "append", "argmin", "axis", "best_box_ind", "best_box_indices", "best_class_embeds", "bool", "box_indices", "box_iou", "box_predictor", "center_to_corners_format", "class", "class_embeds", "class_predictor", "d", "def", "device", "each_query_box", "each_query_pred_boxes", "einsum", "else", "embed_image_query", "for", "generalized_box_iou", "i", "id", "if", "in", "interpolate_pos_encoding", "iou_threshold", "ious", "max", "mean", "mean_embeds", "mean_sim", "nonzero", "numel", "pred_boxes", "pred_boxes_as_corners", "pred_boxes_device", "query_embeds", "query_feature_map", "query_image_features", "range", "return", "selected_embeddings", "selected_inds", "self", "shape", "squeeze", "stack", "tensor", "torch"], "owlv2/modeling_owlv2.py:Owlv2ForObjectDetection.image_guided_detection": ["False", "FloatTensor", "ModelForObjectDetection", "ModelImageGuidedObjectDetectionOutput", "ModelPreTrainedModel", "None", "batch_size", "best_box_indices", "bool", "box_predictor", "class", "class_embeds", "class_predictor", "config", "def", "else", "embed_image_query", "feature_map", "for", "hidden_dim", "if", "image_embedder", "image_embeds", "image_feats", "image_guided_detection", "in", "interpolate_pos_encoding", "is", "logits", "not", "num_patches_height", "num_patches_width", "output", "output_attentions", "output_hidden_states", "pixel_values", "pred_logits", "query_embeds", "query_feature_map", "query_image_embeds", "query_image_feats", "query_pixel_values", "query_pred_boxes", "r", "reshape", "return", "return_dict", "self", "shape", "target_pred_boxes", "text_model_output", "to_tuple", "torch", "tuple", "vision_model_output", "vision_outputs", "x"], "owlv2/modeling_owlv2.py:Owlv2ForObjectDetection.forward": ["False", "FloatTensor", "ModelForObjectDetection", "ModelObjectDetectionOutput", "ModelPreTrainedModel", "None", "Tensor", "attention_mask", "batch_size", "bool", "box_predictor", "class", "class_embeds", "class_predictor", "config", "def", "else", "feature_map", "for", "forward", "hidden_dim", "if", "image_embeds", "image_feats", "image_text_embedder", "in", "input_ids", "interpolate_pos_encoding", "is", "kwargs", "logits", "max_text_queries", "not", "num_patches_height", "num_patches_width", "objectness_logits", "objectness_predictor", "output", "output_attentions", "output_hidden_states", "outputs", "pixel_values", "pred_boxes", "pred_logits", "query_embeds", "query_mask", "r", "reshape", "return", "return_dict", "self", "shape", "text_embeds", "text_model_output", "text_outputs", "to_tuple", "torch", "tuple", "vision_model_output", "vision_outputs", "x"], "owlvit/modeling_owlvit.py:contrastive_loss": ["Model_loss", "Tensor", "arange", "cross_entropy", "def", "device", "functional", "len", "logits", "nn", "return", "torch"], "owlvit/modeling_owlvit.py:owlvit_loss": ["Model_loss", "Tensor", "caption_loss", "contrastive_loss", "def", "image_loss", "return", "similarity", "t", "torch"], "owlvit/modeling_owlvit.py:OwlViTOutput.to_tuple": ["Any", "ModelOutput", "class", "def", "else", "for", "getattr", "if", "in", "k", "keys", "not", "return", "self", "text_model_output", "to_tuple", "tuple", "vision_model_output"], "owlvit/modeling_owlvit.py:_upcast": ["Tensor", "_upcast", "def", "dtype", "else", "float", "float32", "float64", "if", "in", "int", "int32", "int64", "is_floating_point", "return", "t", "torch"], "owlvit/modeling_owlvit.py:box_area": ["Model_area", "Modeles", "Tensor", "_upcast", "def", "return"], "owlvit/modeling_owlvit.py:box_iou": ["Model_area", "Model_iou", "Modeles1", "Modeles2", "None", "area1", "area2", "clamp", "def", "inter", "iou", "left_top", "max", "min", "return", "right_bottom", "torch", "union", "width_height"], "owlvit/modeling_owlvit.py:generalized_box_iou": ["Model_box_iou", "None", "ValueError", "all", "area", "be", "bottom_right", "box_iou", "boxes1", "boxes2", "but", "clamp", "corner", "def", "f", "format", "got", "if", "in", "iou", "max", "min", "must", "not", "raise", "return", "top_left", "torch", "union", "width_height", "x0", "x1", "y0", "y1"], "owlvit/modeling_owlvit.py:OwlViTObjectDetectionOutput.to_tuple": ["Any", "ModelObjectDetectionOutput", "ModelOutput", "class", "def", "else", "for", "getattr", "if", "in", "k", "keys", "not", "return", "self", "text_model_output", "to_tuple", "tuple", "vision_model_output"], "owlvit/modeling_owlvit.py:OwlViTImageGuidedObjectDetectionOutput.to_tuple": ["Any", "ModelImageGuidedObjectDetectionOutput", "ModelOutput", "class", "def", "else", "for", "getattr", "if", "in", "k", "keys", "not", "return", "self", "text_model_output", "to_tuple", "tuple", "vision_model_output"], "owlvit/modeling_owlvit.py:OwlViTVisionEmbeddings.__init__": ["Conv2d", "Embedding", "False", "ModelVisionConfig", "ModelVisionEmbeddings", "Module", "Parameter", "__init__", "arange", "bias", "class", "class_embedding", "config", "def", "embed_dim", "expand", "hidden_size", "image_size", "in_channels", "kernel_size", "nn", "num_channels", "num_patches", "num_positions", "out_channels", "patch_embedding", "patch_size", "persistent", "position_embedding", "position_ids", "randn", "register_buffer", "self", "stride", "super", "torch"], "owlvit/modeling_owlvit.py:OwlViTVisionEmbeddings.interpolate_pos_encoding": ["False", "ModelVisionEmbeddings", "Module", "Tensor", "align_corners", "and", "bicubic", "cat", "class", "class_pos_embed", "def", "dim", "embeddings", "functional", "height", "if", "int", "interpolate", "interpolate_pos_encoding", "is_tracing", "jit", "mode", "new_height", "new_width", "nn", "not", "num_patches", "num_positions", "patch_pos_embed", "patch_size", "permute", "position_embedding", "position_ids", "reshape", "return", "self", "shape", "size", "sqrt_num_positions", "torch", "torch_int", "unsqueeze", "view", "weight", "width"], "owlvit/modeling_owlvit.py:OwlViTVisionEmbeddings.forward": ["False", "FloatTensor", "ModelVisionEmbeddings", "Module", "Tensor", "_", "batch_size", "bool", "cat", "class", "class_embedding", "class_embeds", "def", "dim", "else", "embeddings", "expand", "flatten", "forward", "height", "if", "interpolate_pos_encoding", "nn", "patch_embedding", "patch_embeds", "pixel_values", "position_embedding", "position_ids", "return", "self", "shape", "torch", "transpose", "width"], "owlvit/modeling_owlvit.py:OwlViTTextEmbeddings.__init__": ["Embedding", "False", "ModelTextConfig", "ModelTextEmbeddings", "Module", "__init__", "arange", "class", "config", "def", "expand", "hidden_size", "max_position_embeddings", "nn", "persistent", "position_embedding", "position_ids", "register_buffer", "self", "super", "token_embedding", "torch", "vocab_size"], "owlvit/modeling_owlvit.py:OwlViTTextEmbeddings.forward": ["FloatTensor", "LongTensor", "ModelTextEmbeddings", "Module", "None", "Tensor", "class", "def", "else", "embeddings", "forward", "if", "input_ids", "inputs_embeds", "is", "nn", "not", "position_embedding", "position_embeddings", "position_ids", "return", "self", "seq_length", "shape", "token_embedding", "torch"], "owlvit/modeling_owlvit.py:OwlViTAttention.__init__": ["Linear", "ModelAttention", "Module", "ValueError", "__init__", "and", "attention_dropout", "be", "by", "class", "config", "def", "divisible", "dropout", "embed_dim", "f", "got", "head_dim", "hidden_size", "if", "k_proj", "must", "nn", "num_attention_heads", "num_heads", "out_proj", "q_proj", "raise", "scale", "self", "super", "v_proj"], "owlvit/modeling_owlvit.py:OwlViTAttention._shape": ["ModelAttention", "Module", "Tensor", "_shape", "bsz", "class", "contiguous", "def", "head_dim", "int", "nn", "num_heads", "return", "self", "seq_len", "tensor", "torch", "transpose", "view"], "owlvit/modeling_owlvit.py:OwlViTAttention.forward": ["Attention", "False", "ModelAttention", "Module", "None", "Tensor", "ValueError", "_shape", "attention_mask", "attn_output", "attn_probs", "attn_weights", "attn_weights_reshaped", "be", "bmm", "bool", "bsz", "but", "causal_attention_mask", "class", "def", "dim", "dropout", "dtype", "else", "embed_dim", "f", "forward", "functional", "head_dim", "hidden_states", "if", "is", "k_proj", "key_states", "mask", "nn", "not", "num_heads", "of", "out_proj", "output_attentions", "p", "proj_shape", "q_proj", "query_states", "raise", "reshape", "return", "scale", "self", "should", "size", "softmax", "src_len", "tgt_len", "to", "torch", "training", "transpose", "tuple", "v_proj", "value_states", "view", "weights"], "owlvit/modeling_owlvit.py:OwlViTMLP.__init__": ["ACT2FN", "Linear", "ModelMLP", "Module", "__init__", "activation_fn", "class", "config", "def", "fc1", "fc2", "hidden_act", "hidden_size", "intermediate_size", "nn", "self", "super"], "owlvit/modeling_owlvit.py:OwlViTMLP.forward": ["ModelMLP", "Module", "Tensor", "activation_fn", "class", "def", "fc1", "fc2", "forward", "hidden_states", "nn", "return", "self", "torch"], "owlvit/modeling_owlvit.py:OwlViTEncoderLayer.__init__": ["GradientCheckpointingLayer", "LayerNorm", "ModelAttention", "ModelConfig", "ModelEncoderLayer", "ModelMLP", "__init__", "class", "config", "def", "embed_dim", "eps", "hidden_size", "layer_norm1", "layer_norm2", "layer_norm_eps", "mlp", "nn", "self", "self_attn", "super"], "owlvit/modeling_owlvit.py:OwlViTEncoderLayer.forward": ["False", "FloatTensor", "GradientCheckpointingLayer", "ModelEncoderLayer", "None", "Tensor", "attention_mask", "attn_weights", "bool", "causal_attention_mask", "class", "def", "forward", "hidden_states", "if", "layer_norm1", "layer_norm2", "mlp", "output_attentions", "outputs", "residual", "return", "self", "self_attn", "torch", "tuple"], "owlvit/modeling_owlvit.py:OwlViTPreTrainedModel._init_weights": ["LayerNorm", "Linear", "ModelAttention", "ModelForObjectDetection", "ModelMLP", "ModelModel", "ModelPreTrainedModel", "ModelTextEmbeddings", "ModelVisionEmbeddings", "Module", "None", "PreTrainedModel", "_init_weights", "arange", "bias", "box_bias", "class", "class_embedding", "compute_box_bias", "config", "constant_", "copy_", "def", "elif", "embed_dim", "expand", "factor", "fc1", "fc2", "fc_std", "hidden_size", "if", "in_proj_std", "init", "initializer_factor", "initializer_range", "is", "isinstance", "k_proj", "logit_scale", "logit_scale_init_value", "mean", "module", "nn", "normal_", "not", "num_hidden_layers", "num_patches_height", "num_patches_width", "ones_", "out_proj", "out_proj_std", "patch_embedding", "position_embedding", "position_ids", "q_proj", "self", "shape", "std", "text_embed_dim", "text_projection", "token_embedding", "torch", "v_proj", "vision_embed_dim", "visual_projection", "weight", "zeros_"], "owlvit/modeling_owlvit.py:OwlViTEncoder.__init__": ["False", "ModelConfig", "ModelEncoder", "ModelEncoderLayer", "Module", "ModuleList", "_", "__init__", "class", "config", "def", "for", "gradient_checkpointing", "in", "layers", "nn", "num_hidden_layers", "range", "self", "super"], "owlvit/modeling_owlvit.py:OwlViTEncoder.forward": ["BaseModelOutput", "ModelEncoder", "Module", "None", "Tensor", "all_attentions", "attention_mask", "attentions", "bool", "causal_attention_mask", "class", "config", "def", "else", "encoder_layer", "encoder_states", "for", "forward", "hidden_states", "if", "in", "inputs_embeds", "is", "last_hidden_state", "layer_outputs", "layers", "nn", "not", "output_attentions", "output_hidden_states", "r", "return", "return_dict", "self", "torch", "tuple", "use_return_dict", "v"], "owlvit/modeling_owlvit.py:OwlViTTextTransformer.__init__": ["LayerNorm", "ModelEncoder", "ModelTextConfig", "ModelTextEmbeddings", "ModelTextTransformer", "Module", "__init__", "class", "config", "def", "embed_dim", "embeddings", "encoder", "eps", "final_layer_norm", "hidden_size", "layer_norm_eps", "nn", "self", "super"], "owlvit/modeling_owlvit.py:OwlViTTextTransformer.forward": ["BaseModelOutputWithPooling", "ModelTextTransformer", "Module", "None", "Tensor", "_create_4d_causal_attention_mask", "_prepare_4d_attention_mask", "arange", "argmax", "attention_mask", "attentions", "bool", "causal_attention_mask", "class", "config", "def", "device", "dim", "dtype", "else", "embeddings", "encoder", "encoder_outputs", "final_layer_norm", "forward", "hidden_states", "if", "input_ids", "input_shape", "inputs_embeds", "int", "is", "last_hidden_state", "nn", "not", "output_attentions", "output_hidden_states", "pooled_output", "pooler_output", "position_ids", "r", "return", "return_dict", "self", "shape", "size", "to", "torch", "tuple", "use_return_dict", "view"], "owlvit/modeling_owlvit.py:OwlViTTextModel.__init__": ["ModelPreTrainedModel", "ModelTextConfig", "ModelTextModel", "ModelTextTransformer", "__init__", "class", "config", "def", "post_init", "self", "super", "text_model"], "owlvit/modeling_owlvit.py:OwlViTTextModel.get_input_embeddings": ["ModelPreTrainedModel", "ModelTextModel", "Module", "class", "def", "embeddings", "get_input_embeddings", "nn", "return", "self", "text_model", "token_embedding"], "owlvit/modeling_owlvit.py:OwlViTTextModel.set_input_embeddings": ["ModelPreTrainedModel", "ModelTextModel", "class", "def", "embeddings", "self", "set_input_embeddings", "text_model", "token_embedding", "value"], "owlvit/modeling_owlvit.py:OwlViTTextModel.forward": ["BaseModelOutputWithPooling", "ModelPreTrainedModel", "ModelTextModel", "None", "Tensor", "attention_mask", "bool", "class", "def", "forward", "input_ids", "kwargs", "output_attentions", "output_hidden_states", "r", "return", "return_dict", "self", "text_model", "torch", "tuple"], "owlvit/modeling_owlvit.py:OwlViTVisionTransformer.__init__": ["LayerNorm", "ModelEncoder", "ModelVisionConfig", "ModelVisionEmbeddings", "ModelVisionTransformer", "Module", "__init__", "class", "config", "def", "embeddings", "encoder", "eps", "hidden_size", "layer_norm_eps", "nn", "post_layernorm", "pre_layernorm", "self", "super"], "owlvit/modeling_owlvit.py:OwlViTVisionTransformer.forward": ["BaseModelOutputWithPooling", "False", "FloatTensor", "ModelVisionTransformer", "Module", "None", "attentions", "bool", "class", "config", "def", "dtype", "else", "embeddings", "encoder", "encoder_outputs", "expected_input_dtype", "forward", "hidden_states", "if", "inputs_embeds", "interpolate_pos_encoding", "is", "last_hidden_state", "nn", "not", "output_attentions", "output_hidden_states", "patch_embedding", "pixel_values", "pooled_output", "pooler_output", "post_layernorm", "pre_layernorm", "return", "return_dict", "self", "to", "torch", "tuple", "use_return_dict", "weight"], "owlvit/modeling_owlvit.py:OwlViTVisionModel.__init__": ["ModelPreTrainedModel", "ModelVisionConfig", "ModelVisionModel", "ModelVisionTransformer", "__init__", "class", "config", "def", "post_init", "self", "super", "vision_model"], "owlvit/modeling_owlvit.py:OwlViTVisionModel.get_input_embeddings": ["ModelPreTrainedModel", "ModelVisionModel", "Module", "class", "def", "embeddings", "get_input_embeddings", "nn", "patch_embedding", "return", "self", "vision_model"], "owlvit/modeling_owlvit.py:OwlViTVisionModel.forward": ["BaseModelOutputWithPooling", "False", "FloatTensor", "ModelPreTrainedModel", "ModelVisionModel", "None", "bool", "class", "def", "forward", "interpolate_pos_encoding", "kwargs", "output_attentions", "output_hidden_states", "pixel_values", "r", "return", "return_dict", "self", "torch", "tuple", "vision_model"], "owlvit/modeling_owlvit.py:OwlViTModel.__init__": ["False", "Linear", "ModelConfig", "ModelModel", "ModelPreTrainedModel", "ModelTextConfig", "ModelTextTransformer", "ModelVisionConfig", "ModelVisionTransformer", "Parameter", "TypeError", "__init__", "be", "bias", "but", "class", "config", "def", "expected", "f", "hidden_size", "if", "is", "isinstance", "logit_scale", "logit_scale_init_value", "nn", "not", "of", "post_init", "projection_dim", "raise", "self", "super", "tensor", "text_config", "text_embed_dim", "text_model", "text_projection", "to", "torch", "type", "vision_config", "vision_embed_dim", "vision_model", "visual_projection"], "owlvit/modeling_owlvit.py:OwlViTModel.get_text_features": ["BaseModelOutputWithPooling", "FloatTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "attention_mask", "class", "def", "get_text_features", "input_ids", "pooler_output", "r", "return", "self", "text_features", "text_model", "text_outputs", "text_projection", "torch"], "owlvit/modeling_owlvit.py:OwlViTModel.get_image_features": ["BaseModelOutputWithPooling", "False", "FloatTensor", "ModelModel", "ModelPreTrainedModel", "Tensor", "bool", "class", "def", "get_image_features", "image_features", "interpolate_pos_encoding", "pixel_values", "pooler_output", "r", "return", "self", "torch", "vision_model", "vision_outputs", "visual_projection"], "owlvit/modeling_owlvit.py:OwlViTModel.forward": ["False", "FloatTensor", "LongTensor", "ModelModel", "ModelOutput", "ModelPreTrainedModel", "Model_loss", "None", "Tensor", "True", "attention_mask", "bool", "class", "config", "def", "device", "dim", "else", "exp", "forward", "if", "image_embeds", "input_ids", "interpolate_pos_encoding", "is", "keepdim", "kwargs", "linalg", "logit_scale", "logits_per_image", "logits_per_text", "loss", "matmul", "norm", "not", "ord", "output", "output_attentions", "output_hidden_states", "pixel_values", "r", "return", "return_base_image_embeds", "return_dict", "return_loss", "self", "t", "text_embeds", "text_embeds_norm", "text_model", "text_model_output", "text_outputs", "text_projection", "to", "torch", "tuple", "use_return_dict", "vision_model", "vision_model_output", "vision_outputs", "visual_projection"], "owlvit/modeling_owlvit.py:OwlViTBoxPredictionHead.__init__": ["GELU", "Linear", "ModelBoxPredictionHead", "ModelConfig", "Module", "__init__", "class", "config", "def", "dense0", "dense1", "dense2", "gelu", "hidden_size", "int", "nn", "out_dim", "self", "super", "vision_config", "width"], "owlvit/modeling_owlvit.py:OwlViTBoxPredictionHead.forward": ["FloatTensor", "ModelBoxPredictionHead", "Module", "Tensor", "class", "def", "dense0", "dense1", "dense2", "forward", "gelu", "image_features", "nn", "output", "return", "self", "torch"], "owlvit/modeling_owlvit.py:OwlViTClassPredictionHead.__init__": ["ELU", "Linear", "ModelClassPredictionHead", "ModelConfig", "Module", "__init__", "class", "config", "def", "dense0", "elu", "hidden_size", "logit_scale", "logit_shift", "nn", "out_dim", "query_dim", "self", "super", "text_config", "vision_config"], "owlvit/modeling_owlvit.py:OwlViTClassPredictionHead.forward": ["FloatTensor", "ModelClassPredictionHead", "Module", "None", "Tensor", "True", "batch_size", "class", "def", "dense0", "device", "dim", "dtype", "einsum", "elu", "finfo", "float32", "forward", "if", "image_class_embeds", "image_embeds", "is", "keepdim", "linalg", "logit_scale", "logit_shift", "min", "ndim", "nn", "norm", "not", "num_patches", "pd", "pq", "pred_logits", "qd", "query_dim", "query_embeds", "query_mask", "return", "self", "shape", "to", "torch", "tuple", "unsqueeze", "where", "zeros"], "owlvit/modeling_owlvit.py:OwlViTForObjectDetection.__init__": ["False", "LayerNorm", "Model", "ModelBoxPredictionHead", "ModelClassPredictionHead", "ModelConfig", "ModelForObjectDetection", "ModelModel", "ModelPreTrainedModel", "Sigmoid", "__init__", "box_bias", "box_head", "class", "class_head", "compute_box_bias", "config", "def", "eps", "hidden_size", "image_size", "layer_norm", "layer_norm_eps", "nn", "num_patches_height", "num_patches_width", "patch_size", "persistent", "post_init", "register_buffer", "self", "sigmoid", "super", "vision_config"], "owlvit/modeling_owlvit.py:OwlViTForObjectDetection.normalize_grid_corner_coordinates": ["ModelForObjectDetection", "ModelPreTrainedModel", "Tensor", "arange", "box_coordinates", "class", "def", "dim", "dtype", "float32", "indexing", "int", "meshgrid", "normalize_grid_corner_coordinates", "num_patches_height", "num_patches_width", "return", "stack", "torch", "view", "x_coordinates", "xx", "xy", "y_coordinates", "yy"], "owlvit/modeling_owlvit.py:OwlViTForObjectDetection.compute_box_bias": ["ModelForObjectDetection", "ModelPreTrainedModel", "Tensor", "box_bias", "box_coord_bias", "box_coordinates", "box_size", "box_size_bias", "cat", "class", "clip", "compute_box_bias", "def", "dim", "full_like", "int", "log", "log1p", "normalize_grid_corner_coordinates", "num_patches_height", "num_patches_width", "return", "self", "torch"], "owlvit/modeling_owlvit.py:OwlViTForObjectDetection.box_predictor": ["False", "FloatTensor", "ModelForObjectDetection", "ModelPreTrainedModel", "_", "bool", "box_bias", "box_head", "box_predictor", "class", "compute_box_bias", "def", "device", "else", "feature_map", "if", "image_feats", "interpolate_pos_encoding", "num_patches_height", "num_patches_width", "pred_boxes", "return", "self", "shape", "sigmoid", "to", "torch"], "owlvit/modeling_owlvit.py:OwlViTForObjectDetection.class_predictor": ["FloatTensor", "ModelForObjectDetection", "ModelPreTrainedModel", "None", "Tensor", "class", "class_head", "class_predictor", "def", "image_class_embeds", "image_feats", "pred_logits", "query_embeds", "query_mask", "return", "self", "torch", "tuple"], "owlvit/modeling_owlvit.py:OwlViTForObjectDetection.image_text_embedder": ["False", "FloatTensor", "Model", "ModelForObjectDetection", "ModelPreTrainedModel", "None", "Tensor", "True", "_", "attention_mask", "bool", "broadcast_to", "class", "class_token_out", "config", "def", "else", "height", "if", "image_embeds", "image_text_embedder", "input_ids", "interpolate_pos_encoding", "last_hidden_state", "layer_norm", "new_size", "num_patches_height", "num_patches_width", "output_attentions", "output_hidden_states", "outputs", "patch_size", "pixel_values", "post_layernorm", "reshape", "return", "return_dict", "self", "shape", "text_embeds", "torch", "tuple", "vision_config", "vision_model", "vision_model_output", "width"], "owlvit/modeling_owlvit.py:OwlViTForObjectDetection.image_embedder": ["False", "FloatTensor", "Model", "ModelForObjectDetection", "ModelPreTrainedModel", "None", "True", "_", "bool", "broadcast_to", "class", "class_token_out", "config", "def", "else", "height", "if", "image_embedder", "image_embeds", "interpolate_pos_encoding", "last_hidden_state", "layer_norm", "new_size", "num_patches_height", "num_patches_width", "output_attentions", "output_hidden_states", "patch_size", "pixel_values", "post_layernorm", "reshape", "return", "return_dict", "self", "shape", "torch", "tuple", "vision_config", "vision_model", "vision_outputs", "width"], "owlvit/modeling_owlvit.py:OwlViTForObjectDetection.embed_image_query": ["False", "FloatTensor", "ModelForObjectDetection", "ModelPreTrainedModel", "None", "_", "all", "append", "argmin", "axis", "best_box_ind", "best_box_indices", "best_class_embeds", "bool", "box_indices", "box_iou", "box_predictor", "center_to_corners_format", "class", "class_embeds", "class_predictor", "d", "def", "device", "each_query_box", "each_query_pred_boxes", "einsum", "else", "embed_image_query", "for", "generalized_box_iou", "i", "id", "if", "in", "interpolate_pos_encoding", "iou_threshold", "ious", "max", "mean", "mean_embeds", "mean_sim", "nonzero", "numel", "pred_boxes", "pred_boxes_as_corners", "pred_boxes_device", "query_embeds", "query_feature_map", "query_image_features", "range", "return", "selected_embeddings", "selected_inds", "self", "shape", "squeeze", "stack", "tensor", "torch"], "owlvit/modeling_owlvit.py:OwlViTForObjectDetection.image_guided_detection": ["False", "FloatTensor", "ModelForObjectDetection", "ModelImageGuidedObjectDetectionOutput", "ModelPreTrainedModel", "None", "batch_size", "best_box_indices", "bool", "box_predictor", "class", "class_embeds", "class_predictor", "config", "def", "else", "embed_image_query", "feature_map", "for", "hidden_dim", "if", "image_embedder", "image_embeds", "image_feats", "image_guided_detection", "in", "interpolate_pos_encoding", "is", "logits", "not", "num_patches_height", "num_patches_width", "output", "output_attentions", "output_hidden_states", "pixel_values", "pred_logits", "query_embeds", "query_feature_map", "query_image_embeds", "query_image_feats", "query_pixel_values", "query_pred_boxes", "r", "reshape", "return", "return_dict", "self", "shape", "target_pred_boxes", "text_model_output", "to_tuple", "torch", "tuple", "vision_model_output", "vision_outputs", "x"], "owlvit/modeling_owlvit.py:OwlViTForObjectDetection.forward": ["False", "FloatTensor", "ModelForObjectDetection", "ModelObjectDetectionOutput", "ModelPreTrainedModel", "None", "Tensor", "attention_mask", "batch_size", "bool", "box_predictor", "class", "class_embeds", "class_predictor", "config", "def", "else", "feature_map", "for", "forward", "hidden_dim", "if", "image_embeds", "image_feats", "image_text_embedder", "in", "input_ids", "interpolate_pos_encoding", "is", "kwargs", "logits", "max_text_queries", "not", "num_patches_height", "num_patches_width", "output", "output_attentions", "output_hidden_states", "outputs", "pixel_values", "pred_boxes", "pred_logits", "query_embeds", "query_mask", "r", "reshape", "return", "return_dict", "self", "shape", "text_embeds", "text_model_output", "text_outputs", "to_tuple", "torch", "tuple", "vision_model_output", "vision_outputs", "x"], "paddleocr_vl/modeling_paddleocr_vl.py:PaddleOCRProjector.__init__": ["GELUActivation", "LayerNorm", "Linear", "ModelConfig", "ModelOCRProjector", "Module", "True", "__init__", "act", "bias", "class", "config", "def", "eps", "hidden_size", "linear_1", "linear_2", "merge_kernel_size", "nn", "pre_norm", "self", "spatial_merge_size", "super", "text_config", "torch", "vision_config"], "paddleocr_vl/modeling_paddleocr_vl.py:PaddleOCRProjector.forward": ["ModelOCRProjector", "Module", "Tensor", "act", "append", "cat", "class", "d", "def", "dim", "for", "forward", "h", "h_block", "hidden_states", "image_feature", "image_features", "image_features_chunks", "image_grid", "image_grid_thw", "in", "linear_1", "linear_2", "m1", "m2", "merge_kernel_size", "nn", "pre_norm", "processed_features", "prod", "reshape", "return", "self", "shape", "split", "t", "tolist", "torch", "transpose", "w", "w_block", "zip"], "paddleocr_vl/modeling_paddleocr_vl.py:PaddleOCRVisionRotaryEmbedding.__init__": ["False", "ModelOCRVisionRotaryEmbedding", "Module", "None", "__init__", "arange", "class", "def", "dim", "dtype", "float", "int", "inv_freq", "nn", "persistent", "register_buffer", "self", "super", "theta", "torch"], "paddleocr_vl/modeling_paddleocr_vl.py:PaddleOCRVisionRotaryEmbedding.forward": ["ModelOCRVisionRotaryEmbedding", "Module", "Tensor", "arange", "class", "def", "device", "dtype", "forward", "freqs", "int", "inv_freq", "nn", "outer", "return", "self", "seq", "seqlen", "torch"], "paddleocr_vl/modeling_paddleocr_vl.py:PaddleOCRRotaryEmbedding.__init__": ["Callable", "False", "ModelConfig", "ModelOCRRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "paddleocr_vl/modeling_paddleocr_vl.py:PaddleOCRRotaryEmbedding.compute_default_rope_parameters": ["ModelConfig", "ModelOCRRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "paddleocr_vl/modeling_paddleocr_vl.py:PaddleOCRRotaryEmbedding.forward": ["False", "ModelOCRRotaryEmbedding", "Module", "None", "and", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "paddleocr_vl/modeling_paddleocr_vl.py:PaddleOCRMLP.__init__": ["ACT2FN", "Linear", "ModelOCRMLP", "ModelOCRTextConfig", "Module", "__init__", "act_fn", "bias", "class", "config", "def", "down_proj", "gate_proj", "hidden_act", "hidden_size", "intermediate_size", "nn", "self", "super", "up_proj", "use_bias"], "paddleocr_vl/modeling_paddleocr_vl.py:PaddleOCRMLP.forward": ["ModelOCRMLP", "Module", "act_fn", "class", "def", "down_proj", "forward", "gate_proj", "nn", "return", "self", "up_proj", "x"], "paddleocr_vl/modeling_paddleocr_vl.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "paddleocr_vl/modeling_paddleocr_vl.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value", "value_states"], "paddleocr_vl/modeling_paddleocr_vl.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "paddleocr_vl/modeling_paddleocr_vl.py:apply_multimodal_rotary_pos_emb": ["Model_multimodal_rotary_pos_emb", "cat", "cos", "def", "dim", "enumerate", "for", "i", "in", "k", "k_embed", "m", "mrope_section", "q", "q_embed", "return", "rotate_half", "sin", "split", "torch", "unsqueeze", "unsqueeze_dim"], "paddleocr_vl/modeling_paddleocr_vl.py:PaddleOCRAttention.__init__": ["Instantiating", "Linear", "ModelConfig", "ModelOCRAttention", "Module", "None", "Please", "True", "__class__", "__init__", "__name__", "a", "and", "attention_dropout", "bias", "caching", "call", "class", "config", "creating", "def", "during", "else", "errors", "f", "forward", "getattr", "hasattr", "head_dim", "hidden_size", "if", "int", "is", "is_causal", "k_proj", "layer_idx", "layer_type", "layer_types", "logger", "make", "nn", "not", "num_attention_heads", "num_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "passing", "provide", "q_proj", "recommended", "rope_parameters", "scaling", "self", "sliding_attention", "sliding_window", "super", "sure", "the", "this", "to", "use_bias", "used", "v_proj", "warning_once", "when", "will", "without"], "paddleocr_vl/modeling_paddleocr_vl.py:PaddleOCRAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "False", "FlashAttentionKwargs", "LongTensor", "ModelOCRAttention", "Module", "None", "Tensor", "Unpack", "_", "_attn_implementation", "apply_multimodal_rotary_pos_emb", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "bool", "bsz", "cache_kwargs", "cache_position", "class", "config", "contiguous", "cos", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "head_dim", "hidden_states", "if", "is", "k_proj", "key_states", "kwargs", "layer_idx", "mrope_section", "nn", "not", "o_proj", "output_attentions", "past_key_values", "position_embeddings", "position_ids", "q_len", "q_proj", "query_states", "reshape", "return", "rope_parameters", "scaling", "self", "sin", "size", "sliding_window", "torch", "training", "transpose", "tuple", "update", "use_cache", "v_proj", "value_states", "view"], "paddleocr_vl/modeling_paddleocr_vl.py:PaddleOCRRMSNorm.__init__": ["ModelOCRRMSNorm", "Module", "Parameter", "__init__", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "paddleocr_vl/modeling_paddleocr_vl.py:PaddleOCRRMSNorm.forward": ["ModelOCRRMSNorm", "Module", "True", "class", "def", "dtype", "float32", "forward", "hidden_states", "input_dtype", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "paddleocr_vl/modeling_paddleocr_vl.py:PaddleOCRRMSNorm.extra_repr": ["ModelOCRRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "variance_epsilon", "weight"], "paddleocr_vl/modeling_paddleocr_vl.py:PaddleOCRDecoderLayer.__init__": ["GradientCheckpointingLayer", "ModelOCRAttention", "ModelOCRDecoderLayer", "ModelOCRMLP", "ModelOCRRMSNorm", "ModelOCRTextConfig", "__init__", "class", "config", "def", "eps", "hidden_size", "input_layernorm", "int", "layer_idx", "mlp", "post_attention_layernorm", "rms_norm_eps", "self", "self_attn", "super"], "paddleocr_vl/modeling_paddleocr_vl.py:PaddleOCRDecoderLayer.forward": ["Cache", "False", "GradientCheckpointingLayer", "LongTensor", "ModelOCRDecoderLayer", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "attention_mask", "bool", "cache_position", "class", "def", "forward", "hidden_states", "input_layernorm", "kwargs", "mlp", "past_key_values", "position_embeddings", "position_ids", "post_attention_layernorm", "residual", "return", "self", "self_attn", "torch", "tuple", "use_cache"], "paddleocr_vl/modeling_paddleocr_vl.py:PaddleOCRVLPreTrainedModel._init_weights": ["ModelOCRVisionEmbeddings", "ModelOCRVisionRotaryEmbedding", "ModelPreTrainedModel", "PreTrainedModel", "_init_weights", "arange", "class", "copy_", "def", "dim", "dtype", "elif", "expand", "float", "if", "init", "inv_freq", "isinstance", "module", "position_ids", "self", "shape", "super", "theta", "torch"], "paddleocr_vl/modeling_paddleocr_vl.py:PaddleOCRTextModel.__init__": ["Embedding", "False", "ModelOCRDecoderLayer", "ModelOCRRMSNorm", "ModelOCRRotaryEmbedding", "ModelOCRTextConfig", "ModelOCRTextModel", "ModelPreTrainedModel", "ModuleList", "__init__", "class", "config", "def", "embed_tokens", "eps", "for", "gradient_checkpointing", "hidden_size", "in", "layer_idx", "layers", "nn", "norm", "num_hidden_layers", "pad_token_id", "padding_idx", "post_init", "range", "rms_norm_eps", "rotary_emb", "self", "super", "vocab_size"], "paddleocr_vl/modeling_paddleocr_vl.py:PaddleOCRTextModel.forward": ["BaseModelOutputWithPast", "Cache", "DynamicCache", "FloatTensor", "LongTensor", "ModelOCRTextModel", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "and", "arange", "attention_mask", "bool", "cache_position", "causal_mask", "class", "config", "create_causal_mask", "decoder_layer", "def", "device", "elif", "else", "embed_tokens", "exactly", "expand", "for", "forward", "get_seq_length", "hidden_states", "if", "in", "input_embeds", "input_ids", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layers", "must", "ndim", "norm", "not", "num_hidden_layers", "of", "one", "or", "past_key_values", "past_seen_tokens", "position_embeddings", "position_ids", "raise", "return", "rotary_emb", "self", "shape", "specify", "text_position_ids", "torch", "use_cache", "view"], "paddleocr_vl/modeling_paddleocr_vl.py:PaddleOCRVisionModel.__init__": ["ModelOCRVisionConfig", "ModelOCRVisionModel", "ModelOCRVisionTransformer", "ModelPreTrainedModel", "__init__", "class", "config", "def", "post_init", "self", "super", "vision_model"], "paddleocr_vl/modeling_paddleocr_vl.py:PaddleOCRVisionModel.forward": ["BaseModelOutputWithPooling", "FloatTensor", "ModelOCRVisionModel", "ModelPreTrainedModel", "None", "Tensor", "class", "cu_seqlens", "def", "forward", "image_grid_thw", "int", "kwargs", "list", "pixel_values", "return", "self", "torch", "tuple", "vision_model"], "paddleocr_vl/modeling_paddleocr_vl.py:PaddleOCRVisionEmbeddings.__init__": ["Conv2d", "Embedding", "False", "ModelOCRVisionConfig", "ModelOCRVisionEmbeddings", "Module", "__init__", "arange", "class", "config", "def", "embed_dim", "expand", "hidden_size", "image_size", "in_channels", "kernel_size", "nn", "num_channels", "num_patches", "num_positions", "out_channels", "padding", "patch_embedding", "patch_size", "persistent", "position_embedding", "position_ids", "register_buffer", "self", "stride", "super", "torch", "valid"], "paddleocr_vl/modeling_paddleocr_vl.py:PaddleOCRVisionEmbeddings.interpolate_pos_encoding": ["False", "ModelOCRVisionEmbeddings", "Module", "Tensor", "align_corners", "bilinear", "class", "def", "dim", "embeddings", "functional", "height", "int", "interpolate", "interpolate_pos_encoding", "mode", "nn", "num_positions", "patch_pos_embed", "permute", "position_embedding", "reshape", "return", "self", "shape", "size", "sqrt_num_positions", "torch", "torch_int", "unsqueeze", "view", "weight", "width"], "paddleocr_vl/modeling_paddleocr_vl.py:PaddleOCRVisionEmbeddings.forward": ["FloatTensor", "ModelOCRVisionEmbeddings", "Module", "None", "Tensor", "append", "batch_size", "channel", "class", "concat", "def", "dim", "dtype", "embeddings", "end", "flatten", "for", "forward", "h", "height", "image_embeddings", "image_grid", "image_grid_thw", "in", "int", "interpolate_pos_encoding", "list", "nn", "patch_embedding", "patch_embeds", "pixel_values", "position_embedding", "repeat", "reshape", "return", "self", "shape", "squeeze", "squence_len", "start", "t", "target_dtype", "tmp_embeddings", "to", "torch", "tuple", "w", "weight", "width"], "paddleocr_vl/modeling_paddleocr_vl.py:apply_rotary_pos_emb_vision": ["Model_rotary_pos_emb_vision", "Tensor", "cos", "def", "dtype", "float", "k", "k_embed", "orig_k_dtype", "orig_q_dtype", "q", "q_embed", "return", "rotate_half", "sin", "to", "torch", "tuple", "unsqueeze"], "paddleocr_vl/modeling_paddleocr_vl.py:PaddleOCRVisionAttention.__init__": ["False", "Linear", "ModelOCRVisionAttention", "ModelOCRVisionConfig", "Module", "ValueError", "__init__", "and", "attention_dropout", "be", "by", "class", "config", "def", "divisible", "embed_dim", "f", "got", "head_dim", "hidden_size", "if", "is_causal", "k_proj", "must", "nn", "num_attention_heads", "num_heads", "num_key_value_groups", "out_proj", "q_proj", "raise", "scaling", "self", "super", "v_proj"], "paddleocr_vl/modeling_paddleocr_vl.py:PaddleOCRVisionAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "False", "ModelOCRVisionAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_attn_implementation", "append", "apply_rotary_pos_emb_vision", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_outputs", "attn_weight", "attn_weights", "cat", "class", "config", "contiguous", "cos", "cu_seq_lens_k", "cu_seq_lens_q", "cu_seqlens", "def", "dim", "dropout", "eager", "eager_attention_forward", "else", "flash_attention_2", "for", "forward", "head_dim", "hidden_states", "if", "in", "is_causal", "k", "k_proj", "key_states", "kwargs", "lengths", "max", "max_length_k", "max_length_q", "max_seqlen", "nn", "not", "num_heads", "out_proj", "position_embeddings", "q", "q_proj", "query_states", "reshape", "return", "scaling", "self", "seq_length", "shape", "sin", "split", "splits", "tensor", "tolist", "torch", "training", "transpose", "tuple", "unsqueeze", "v", "v_proj", "value_states", "view", "zip"], "paddleocr_vl/modeling_paddleocr_vl.py:PaddleOCRVisionMLP.__init__": ["ACT2FN", "Linear", "ModelOCRVisionConfig", "ModelOCRVisionMLP", "Module", "__init__", "activation_fn", "class", "config", "def", "fc1", "fc2", "hidden_act", "hidden_size", "intermediate_size", "nn", "self", "super"], "paddleocr_vl/modeling_paddleocr_vl.py:PaddleOCRVisionMLP.forward": ["ModelOCRVisionMLP", "Module", "Tensor", "activation_fn", "class", "def", "fc1", "fc2", "forward", "hidden_states", "nn", "return", "self", "torch"], "paddleocr_vl/modeling_paddleocr_vl.py:PaddleOCRVisionEncoderLayer.__init__": ["GradientCheckpointingLayer", "LayerNorm", "ModelOCRVisionAttention", "ModelOCRVisionConfig", "ModelOCRVisionEncoderLayer", "ModelOCRVisionMLP", "__init__", "class", "config", "def", "embed_dim", "eps", "hidden_size", "layer_norm1", "layer_norm2", "layer_norm_eps", "mlp", "nn", "self", "self_attn", "super"], "paddleocr_vl/modeling_paddleocr_vl.py:PaddleOCRVisionEncoderLayer.forward": ["GradientCheckpointingLayer", "ModelOCRVisionEncoderLayer", "Tensor", "TransformersKwargs", "Unpack", "_", "class", "cu_seqlens", "def", "forward", "hidden_states", "kwargs", "layer_norm1", "layer_norm2", "mlp", "position_embeddings", "r", "residual", "return", "self", "self_attn", "torch", "tuple"], "paddleocr_vl/modeling_paddleocr_vl.py:PaddleOCRVisionEncoder.__init__": ["False", "ModelOCRVisionConfig", "ModelOCRVisionEncoder", "ModelOCRVisionEncoderLayer", "ModelOCRVisionRotaryEmbedding", "Module", "ModuleList", "_", "__init__", "class", "config", "def", "embed_dim", "for", "gradient_checkpointing", "head_dim", "hidden_size", "in", "layers", "nn", "num_attention_heads", "num_heads", "num_hidden_layers", "range", "rotary_pos_emb", "self", "super"], "paddleocr_vl/modeling_paddleocr_vl.py:PaddleOCRVisionEncoder.forward": ["BaseModelOutput", "FloatTensor", "ModelOCRVisionEncoder", "Module", "None", "Tensor", "append", "arange", "attention_mask", "class", "concat", "config", "cos", "create_bidirectional_mask", "cu_seqlens", "def", "device", "dim", "encoder_layer", "flatten", "for", "forward", "h", "height_position_ids", "hidden_states", "image_grid_thw", "image_pids", "in", "input_embeds", "inputs_embeds", "int", "last_hidden_state", "layers", "list", "max", "max_grid_size", "nn", "pids", "position_embeddings", "r", "repeat", "return", "rotary_embeddings", "rotary_embeddings_max_grid", "rotary_pos_emb", "sample_hids", "sample_wids", "self", "sin", "split_hids", "split_wids", "stack", "t", "torch", "tuple", "w", "width_position_ids"], "paddleocr_vl/modeling_paddleocr_vl.py:PaddleOCRVisionTransformer.__init__": ["LayerNorm", "ModelOCRVisionConfig", "ModelOCRVisionEmbeddings", "ModelOCRVisionEncoder", "ModelOCRVisionTransformer", "ModelPreTrainedModel", "__init__", "class", "config", "def", "embed_dim", "embeddings", "encoder", "eps", "hidden_size", "layer_norm_eps", "nn", "post_init", "post_layernorm", "self", "super"], "paddleocr_vl/modeling_paddleocr_vl.py:PaddleOCRVisionTransformer.forward": ["BaseModelOutput", "BaseModelOutputWithPooling", "FloatTensor", "ModelOCRVisionTransformer", "ModelPreTrainedModel", "None", "Tensor", "attention_mask", "attentions", "class", "cu_seqlens", "def", "embeddings", "encoder", "encoder_outputs", "forward", "hidden_states", "image_grid_thw", "inputs_embeds", "int", "kwargs", "last_hidden_state", "list", "pixel_values", "pooler_output", "post_layernorm", "return", "self", "torch", "tuple"], "paddleocr_vl/modeling_paddleocr_vl.py:PaddleOCRVLModel.__init__": ["ModelConfig", "ModelModel", "ModelOCRProjector", "ModelOCRTextModel", "ModelOCRVisionModel", "ModelPreTrainedModel", "None", "__init__", "_from_config", "class", "config", "def", "language_model", "post_init", "projector", "rope_deltas", "self", "super", "text_config", "vision_config", "visual"], "paddleocr_vl/modeling_paddleocr_vl.py:PaddleOCRVLModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embed_tokens", "get_input_embeddings", "language_model", "return", "self"], "paddleocr_vl/modeling_paddleocr_vl.py:PaddleOCRVLModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embed_tokens", "language_model", "self", "set_input_embeddings", "value"], "paddleocr_vl/modeling_paddleocr_vl.py:PaddleOCRVLModel.get_rope_index": ["False", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "True", "_", "and", "append", "arange", "argwhere", "attention_mask", "cat", "class", "config", "cumsum", "def", "device", "dim", "dtype", "ed", "ed_image", "ed_video", "else", "enumerate", "expand", "flatten", "for", "get_rope_index", "h", "h_index", "i", "if", "image_grid_thw", "image_index", "image_nums", "image_token_id", "in", "index", "input_ids", "input_tokens", "is", "item", "keepdim", "len", "list", "llm_grid_h", "llm_grid_t", "llm_grid_w", "llm_pos_ids_list", "llm_positions", "long", "masked_fill_", "max", "max_position_ids", "mrope_position_deltas", "not", "ones", "ones_like", "or", "position_ids", "range", "remain_images", "remain_videos", "reshape", "return", "self", "shape", "spatial_merge_size", "squeeze", "st", "st_idx", "stack", "sum", "t", "t_index", "tensor", "text_len", "to", "tolist", "torch", "total_input_ids", "tuple", "unsqueeze", "video_grid_thw", "video_index", "video_nums", "video_token_id", "view", "vision_config", "vision_start_indices", "vision_start_token_id", "vision_tokens", "w", "w_index", "zeros"], "paddleocr_vl/modeling_paddleocr_vl.py:PaddleOCRVLModel.get_video_features": ["FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "class", "def", "dtype", "get_video_features", "grid_thw", "pixel_values_videos", "prod", "return", "self", "spatial_merge_size", "split", "split_sizes", "tolist", "torch", "type", "video_embeds", "video_grid_thw", "visual"], "paddleocr_vl/modeling_paddleocr_vl.py:PaddleOCRVLModel.get_image_features": ["FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "class", "cu_seqlens", "cumsum", "def", "dim", "dtype", "else", "functional", "get_image_features", "if", "image_embeds", "image_grid_thw", "int32", "is_tracing", "jit", "last_hidden_state", "nn", "pad", "pixel_values", "projector", "repeat_interleave", "return", "self", "torch", "type", "unsqueeze", "value", "vision_outputs", "visual"], "paddleocr_vl/modeling_paddleocr_vl.py:PaddleOCRVLModel.get_placeholder_mask": ["FloatTensor", "Image", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "ValueError", "all", "and", "class", "config", "def", "device", "do", "dtype", "else", "expand_as", "f", "features", "get_input_embeddings", "get_placeholder_mask", "if", "image", "image_features", "image_token_id", "input_ids", "inputs_embeds", "is", "long", "match", "n_image_features", "n_image_tokens", "not", "numel", "raise", "return", "self", "shape", "special_image_mask", "sum", "tensor", "to", "tokens", "torch", "unsqueeze"], "paddleocr_vl/modeling_paddleocr_vl.py:PaddleOCRVLModel.forward": ["FloatTensor", "LongTensor", "ModelModel", "ModelModelOutputWithPast", "ModelPreTrainedModel", "None", "Tensor", "_", "arange", "attention_mask", "attentions", "batch_size", "bool", "cache_position", "class", "def", "delta", "device", "dim", "dtype", "else", "embed_tokens", "expand", "forward", "get_image_features", "get_placeholder_mask", "get_rope_index", "get_seq_length", "hidden_states", "if", "image_embeds", "image_features", "image_grid_thw", "image_mask", "input_ids", "inputs_embeds", "is", "kwargs", "language_model", "last_hidden_state", "list", "masked_scatter", "not", "or", "output", "outputs", "past_key_values", "past_key_values_length", "pixel_values", "position_ids", "r", "repeat_interleave", "return", "rope_deltas", "self", "seq_length", "shape", "to", "torch", "tuple", "use_cache", "view"], "paddleocr_vl/modeling_paddleocr_vl.py:PaddleOCRVLForConditionalGeneration.__init__": ["False", "GenerationMixin", "Linear", "ModelForConditionalGeneration", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "text_config", "vocab_size"], "paddleocr_vl/modeling_paddleocr_vl.py:PaddleOCRVLForConditionalGeneration.get_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "model", "return", "self"], "paddleocr_vl/modeling_paddleocr_vl.py:PaddleOCRVLForConditionalGeneration.set_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "model", "self", "set_input_embeddings", "value"], "paddleocr_vl/modeling_paddleocr_vl.py:PaddleOCRVLForConditionalGeneration.get_video_features": ["FloatTensor", "GenerationMixin", "LongTensor", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "class", "def", "get_video_features", "model", "pixel_values_videos", "return", "self", "torch", "video_grid_thw"], "paddleocr_vl/modeling_paddleocr_vl.py:PaddleOCRVLForConditionalGeneration.get_image_features": ["FloatTensor", "GenerationMixin", "LongTensor", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "class", "def", "get_image_features", "image_grid_thw", "model", "pixel_values", "return", "self", "torch"], "paddleocr_vl/modeling_paddleocr_vl.py:PaddleOCRVLForConditionalGeneration.forward": ["Cache", "FloatTensor", "GenerationMixin", "LongTensor", "ModelCausalLMOutputWithPast", "ModelForConditionalGeneration", "ModelModelOutputWithPast", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "else", "forward", "hidden_states", "if", "image_grid_thw", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "outputs", "past_key_values", "pixel_values", "position_ids", "r", "return", "rope_deltas", "self", "slice", "slice_indices", "text_config", "torch", "tuple", "use_cache", "vocab_size"], "paddleocr_vl/modeling_paddleocr_vl.py:PaddleOCRVLForConditionalGeneration.prepare_inputs_for_generation": ["False", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "True", "and", "arange", "attention_mask", "batch_size", "cache_position", "cat", "class", "def", "delta", "device", "dim", "elif", "expand", "expand_as", "get", "get_rope_index", "if", "image_grid_thw", "in", "input_ids", "inputs_embeds", "is", "is_first_iteration", "kwargs", "model", "model_inputs", "not", "or", "past_key_values", "pixel_values", "pixel_values_videos", "position_ids", "prepare_inputs_for_generation", "repeat_interleave", "return", "rope_deltas", "self", "seq_length", "shape", "super", "text_positions", "torch", "use_cache", "video_grid_thw", "view", "vision_positions"], "paddleocr_vl/modeling_paddleocr_vl.py:PaddleOCRVLForConditionalGeneration._get_image_nums_and_video_nums": ["GenerationMixin", "LongTensor", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Tensor", "_get_image_nums_and_video_nums", "class", "config", "def", "device", "dim", "dims", "dtype", "else", "get_input_embeddings", "if", "image_mask", "image_nums", "image_token_id", "input_ids", "inputs_embeds", "is", "long", "not", "return", "roll", "self", "shifts", "sum", "tensor", "torch", "tuple", "video_mask", "video_nums", "video_token_id", "vision_first_mask", "vision_start_mask", "vision_start_token_id"], "paddleocr_vl/modeling_paddleocr_vl.py:PaddleOCRVLForConditionalGeneration._expand_inputs_for_generation": ["Any", "False", "GenerationMixin", "If", "LongTensor", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Tensor", "True", "ValueError", "_expand_dict_for_generation", "_expand_dict_for_generation_visual", "_expand_inputs_for_generation", "_get_image_nums_and_video_nums", "_repeat_interleave_samples", "and", "bool", "cache_position", "cat", "class", "def", "defined", "dict", "dict_to_expand", "dim", "elif", "encoder_outputs", "expand_size", "for", "get", "if", "image_grid_thw", "image_nums", "in", "input_ids", "inputs_embeds", "int", "is", "is_encoder_decoder", "isinstance", "key", "lengths", "list", "make", "model_kwargs", "not", "pixel_values", "pixel_values_videos", "prod", "raise", "repeat", "repeat_args", "repeat_interleave", "repeat_times", "result", "return", "sample", "samples", "second_per_grid_ts", "self", "split", "str", "sum", "sure", "that", "torch", "tuple", "video_grid_thw", "video_nums", "visual_keys", "x"], "paligemma/modeling_paligemma.py:PaliGemmaMultiModalProjector.__init__": ["Linear", "ModelConfig", "ModelMultiModalProjector", "Module", "True", "__init__", "bias", "class", "config", "def", "hidden_size", "linear", "nn", "projection_dim", "self", "super", "vision_config"], "paligemma/modeling_paligemma.py:PaliGemmaMultiModalProjector.forward": ["ModelMultiModalProjector", "Module", "class", "def", "forward", "hidden_states", "image_features", "linear", "nn", "return", "self"], "paligemma/modeling_paligemma.py:token_type_ids_mask_function": ["Callable", "Model_type_ids", "Model_type_ids_at_kv_idx", "Model_type_ids_at_q_idx", "Model_type_ids_mask_function", "None", "Tensor", "batch_idx", "bool", "def", "head_idx", "if", "image_group_ids", "image_group_ids_at_kv_idx", "image_group_ids_at_q_idx", "inner_mask", "int", "is", "is_image_block", "kv_idx", "q_idx", "return", "safe_kv_idx", "safe_q_idx", "same_image_block", "shape", "torch", "where"], "paligemma/modeling_paligemma.py:create_causal_mask_mapping": ["Cache", "False", "FloatTensor", "It", "Model_causal_mask_mapping", "Model_masks_for_generate", "None", "PreTrainedConfig", "Tensor", "The", "True", "ValueError", "We", "a", "and", "as", "attention", "attention_mask", "bad", "bool", "but", "cache_position", "config", "cumsum", "def", "device", "dict", "dim", "else", "full_like", "functional", "get", "get_text_config", "if", "image_group_ids", "input", "input_embeds", "int", "is", "is_first_iteration", "is_image", "is_initialized", "is_previous_image", "is_training", "kwargs", "logger", "mask_kwargs", "masking", "model", "new_image_start", "nn", "not", "ones_like", "or", "or_mask_function", "pad", "passing", "past_key_values", "pixel_values", "position_ids", "prefill", "prevent", "provided", "raise", "recommend", "required", "return", "stage", "the", "to", "token_type_ids", "token_type_ids_mask_function", "torch", "training", "use_cache", "value", "warning_once", "when", "where"], "paligemma/modeling_paligemma.py:PaliGemmaModel.__init__": ["AutoModel", "ModelConfig", "ModelModel", "ModelMultiModalProjector", "ModelPreTrainedModel", "None", "__init__", "class", "config", "def", "dtype", "else", "from_config", "get_text_config", "if", "is", "language_model", "multi_modal_projector", "not", "or", "pad_token_id", "post_init", "self", "super", "text_config", "text_config_dtype", "vision_config", "vision_tower", "vocab_size"], "paligemma/modeling_paligemma.py:PaliGemmaModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "language_model", "return", "self"], "paligemma/modeling_paligemma.py:PaliGemmaModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "language_model", "self", "set_input_embeddings", "value"], "paligemma/modeling_paligemma.py:PaliGemmaModel.get_image_features": ["FloatTensor", "ModelModel", "ModelPreTrainedModel", "class", "config", "def", "get_image_features", "hidden_size", "image_features", "image_outputs", "last_hidden_state", "multi_modal_projector", "pixel_values", "return", "selected_image_feature", "self", "text_config", "torch", "vision_tower"], "paligemma/modeling_paligemma.py:PaliGemmaModel.get_placeholder_mask": ["FloatTensor", "Image", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "ValueError", "all", "and", "class", "config", "def", "device", "do", "dtype", "else", "expand_as", "f", "features", "get_input_embeddings", "get_placeholder_mask", "if", "image", "image_features", "image_token_id", "input_ids", "inputs_embeds", "is", "long", "match", "n_image_features", "n_image_tokens", "not", "numel", "raise", "return", "self", "shape", "special_image_mask", "sum", "tensor", "to", "tokens", "torch", "unsqueeze"], "paligemma/modeling_paligemma.py:PaliGemmaModel.forward": ["Cache", "FlashAttentionKwargs", "FloatTensor", "LongTensor", "ModelModel", "ModelModelOutputWithPast", "ModelPreTrainedModel", "None", "Tensor", "True", "Unpack", "ValueError", "You", "and", "arange", "attention_mask", "attentions", "bool", "cache_position", "causal_mask_mapping", "class", "clone", "config", "create_causal_mask_mapping", "def", "device", "dict", "dtype", "else", "exactly", "forward", "get_image_features", "get_input_embeddings", "get_placeholder_mask", "get_seq_length", "hidden_states", "if", "image_features", "image_hidden_states", "image_token_id", "input_ids", "inputs_embeds", "is", "is_training", "isinstance", "kwargs", "labels", "language_model", "last_hidden_state", "llm_input_ids", "masked_scatter", "must", "not", "of", "one", "or", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "past_seen_tokens", "pixel_values", "position_ids", "r", "raise", "return", "return_dict", "self", "shape", "special_image_mask", "specify", "to", "token_type_ids", "torch", "training", "tuple", "unsqueeze", "use_cache", "use_return_dict", "vocab_size"], "paligemma/modeling_paligemma.py:PaliGemmaForConditionalGeneration.__init__": ["False", "GenerationMixin", "Linear", "ModelConfig", "ModelForConditionalGeneration", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "text_config", "vocab_size"], "paligemma/modeling_paligemma.py:PaliGemmaForConditionalGeneration.get_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "model", "return", "self"], "paligemma/modeling_paligemma.py:PaliGemmaForConditionalGeneration.set_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "model", "self", "set_input_embeddings", "value"], "paligemma/modeling_paligemma.py:PaliGemmaForConditionalGeneration.get_image_features": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "get_image_features", "model", "pixel_values", "return", "self"], "paligemma/modeling_paligemma.py:PaliGemmaForConditionalGeneration.forward": ["Cache", "FloatTensor", "GenerationMixin", "LongTensor", "ModelCausalLMOutputWithPast", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "else", "forward", "hidden_states", "if", "image_hidden_states", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "pixel_values", "position_ids", "r", "return", "return_dict", "self", "slice", "slice_indices", "text_config", "token_type_ids", "torch", "tuple", "use_cache", "use_return_dict", "vocab_size"], "paligemma/modeling_paligemma.py:PaliGemmaForConditionalGeneration.prepare_inputs_for_generation": ["False", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "True", "attention_mask", "cache_position", "class", "def", "get", "if", "input_ids", "inputs_embeds", "is", "is_first_iteration", "kwargs", "labels", "logits_to_keep", "model_inputs", "not", "or", "past_key_values", "pixel_values", "position_ids", "prepare_inputs_for_generation", "return", "self", "super", "token_type_ids", "use_cache"], "paligemma/modeling_paligemma.py:PaliGemmaForConditionalGeneration.create_masks_for_generate": ["Cache", "False", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "PreTrainedConfig", "Tensor", "attention_mask", "bool", "cache_position", "class", "config", "create_causal_mask_mapping", "create_masks_for_generate", "def", "dict", "for", "if", "in", "input_embeds", "is_first_iteration", "items", "k", "kwargs", "past_key_values", "pixel_values", "position_ids", "return", "token_type_ids", "torch", "v"], "parakeet/modeling_parakeet.py:ParakeetEncoderRelPositionalEncoding.__init__": ["False", "ModelEncoderConfig", "ModelEncoderRelPositionalEncoding", "Module", "None", "__init__", "arange", "base", "class", "config", "def", "device", "dtype", "float", "hidden_size", "int64", "inv_freq", "max_position_embeddings", "nn", "persistent", "register_buffer", "self", "super", "to", "torch"], "parakeet/modeling_parakeet.py:ParakeetEncoderRelPositionalEncoding.forward": ["False", "Length", "ModelEncoderRelPositionalEncoding", "Module", "None", "Sequence", "Tensor", "ValueError", "and", "arange", "be", "class", "config", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "enabled", "equal", "expand", "f", "float", "forward", "freqs", "has", "hidden_states", "if", "inv_freq", "inv_freq_expanded", "isinstance", "less", "max_position_embeddings", "maybe_autocast", "mps", "nn", "or", "pos_embed", "position_ids", "position_ids_expanded", "raise", "reshape", "return", "self", "seq_length", "shape", "sin", "stack", "str", "than", "to", "torch", "transpose", "type", "with"], "parakeet/modeling_parakeet.py:ParakeetEncoderFeedForward.__init__": ["ACT2FN", "Linear", "ModelEncoderConfig", "ModelEncoderFeedForward", "Module", "__init__", "activation", "activation_dropout", "attention_bias", "bias", "class", "config", "def", "hidden_act", "hidden_size", "intermediate_size", "linear1", "linear2", "nn", "self", "super"], "parakeet/modeling_parakeet.py:ParakeetEncoderFeedForward.forward": ["ModelEncoderFeedForward", "Module", "activation", "activation_dropout", "class", "def", "dropout", "forward", "functional", "hidden_states", "linear1", "linear2", "nn", "p", "return", "self", "training"], "parakeet/modeling_parakeet.py:ParakeetEncoderConvolutionModule.__init__": ["ACT2FN", "BatchNorm1d", "Conv1d", "ModelEncoderConfig", "ModelEncoderConvolutionModule", "Module", "None", "__init__", "activation", "bias", "channels", "class", "config", "conv_kernel_size", "convolution_bias", "def", "depthwise_conv", "else", "get", "getattr", "groups", "hidden_act", "hidden_size", "if", "is", "kernel_size", "module_config", "nn", "norm", "padding", "pointwise_conv1", "pointwise_conv2", "self", "silu", "stride", "super"], "parakeet/modeling_parakeet.py:ParakeetEncoderConvolutionModule.forward": ["ModelEncoderConvolutionModule", "Module", "None", "activation", "all", "all_masked_rows", "attention_mask", "bool", "class", "def", "depthwise_conv", "dim", "dtype", "else", "forward", "functional", "glu", "hidden_states", "if", "is", "masked_fill", "nn", "norm", "not", "pointwise_conv1", "pointwise_conv2", "return", "self", "torch", "transpose"], "parakeet/modeling_parakeet.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "parakeet/modeling_parakeet.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cos", "def", "k", "k_embed", "q", "q_embed", "return", "rotate_half", "sin", "unsqueeze", "unsqueeze_dim"], "parakeet/modeling_parakeet.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "parakeet/modeling_parakeet.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value", "value_states"], "parakeet/modeling_parakeet.py:ParakeetEncoderAttention.__init__": ["False", "Linear", "ModelEncoderAttention", "ModelEncoderConfig", "Module", "Parameter", "__init__", "attention_bias", "attention_dropout", "bias", "bias_u", "bias_v", "class", "config", "def", "getattr", "head_dim", "hidden_size", "int", "is_causal", "k_proj", "layer_idx", "nn", "num_attention_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "q_proj", "relative_k_proj", "scaling", "self", "super", "torch", "v_proj", "zeros"], "parakeet/modeling_parakeet.py:ParakeetEncoderAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "ModelEncoderAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_attn_implementation", "_rel_shift", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "batch_size", "bias_u", "bias_v", "class", "config", "contiguous", "def", "dropout", "eager", "eager_attention_forward", "else", "float", "forward", "head_dim", "hidden_shape", "hidden_states", "if", "inf", "input_shape", "is", "k_proj", "key", "key_states", "kwargs", "logical_not", "masked_fill_", "matrix_bd", "nn", "not", "num_attention_heads", "o_proj", "permute", "position_embeddings", "q_proj", "query", "query_states", "query_states_with_bias_u", "query_states_with_bias_v", "relative_k_proj", "relative_key_states", "reshape", "return", "scaling", "self", "seq_length", "shape", "torch", "training", "transpose", "tuple", "v_proj", "value", "value_states", "view"], "parakeet/modeling_parakeet.py:ParakeetEncoderAttention._rel_shift": ["ModelEncoderAttention", "Module", "_rel_shift", "attention_scores", "batch_size", "class", "def", "functional", "nn", "num_heads", "pad", "position_length", "query_length", "return", "self", "shape", "view"], "parakeet/modeling_parakeet.py:ParakeetEncoderSubsamplingConv2D.__init__": ["Conv2d", "Linear", "ModelEncoderConfig", "ModelEncoderSubsamplingConv2D", "Module", "ModuleList", "ReLU", "True", "__init__", "append", "bias", "channels", "class", "config", "def", "for", "groups", "hidden_size", "i", "in", "int", "kernel_size", "layers", "linear", "log2", "math", "nn", "num_layers", "num_mel_bins", "out_length", "padding", "range", "self", "stride", "subsampling_conv_channels", "subsampling_conv_kernel_size", "subsampling_conv_stride", "subsampling_factor", "super"], "parakeet/modeling_parakeet.py:ParakeetEncoderSubsamplingConv2D._get_output_length": ["Conv2d", "ModelEncoderSubsamplingConv2D", "Module", "Tensor", "_get_output_length", "and", "class", "conv_layer", "def", "hasattr", "if", "input_lengths", "kernel_size", "nn", "output_lengths", "padding", "return", "self", "stride", "torch"], "parakeet/modeling_parakeet.py:ParakeetEncoderSubsamplingConv2D.forward": ["Conv2d", "ModelEncoderSubsamplingConv2D", "Module", "None", "Tensor", "_get_output_length", "and", "arange", "attention_mask", "channel_mask", "class", "current_lengths", "current_seq_length", "def", "device", "else", "for", "forward", "hidden_states", "if", "in", "input_features", "is", "isinstance", "layer", "layers", "linear", "nn", "not", "reshape", "return", "self", "shape", "sum", "torch", "transpose", "unsqueeze"], "parakeet/modeling_parakeet.py:ParakeetEncoderBlock.__init__": ["False", "GradientCheckpointingLayer", "LayerNorm", "ModelEncoderAttention", "ModelEncoderBlock", "ModelEncoderConfig", "ModelEncoderConvolutionModule", "ModelEncoderFeedForward", "None", "__init__", "class", "config", "conv", "def", "feed_forward1", "feed_forward2", "gradient_checkpointing", "hidden_size", "int", "layer_idx", "nn", "norm_conv", "norm_feed_forward1", "norm_feed_forward2", "norm_out", "norm_self_att", "self", "self_attn", "super"], "parakeet/modeling_parakeet.py:ParakeetEncoderBlock.forward": ["GradientCheckpointingLayer", "ModelEncoderBlock", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "attention_mask", "attn_output", "class", "conv", "conv_output", "def", "feed_forward1", "feed_forward2", "ff2_output", "forward", "hidden_states", "kwargs", "norm_conv", "norm_feed_forward1", "norm_feed_forward2", "norm_out", "norm_self_att", "normalized_hidden_states", "position_embeddings", "residual", "return", "self", "self_attn", "torch"], "parakeet/modeling_parakeet.py:ParakeetPreTrainedModel._init_weights": ["ModelEncoderAttention", "ModelEncoderRelPositionalEncoding", "ModelPreTrainedModel", "PreTrainedModel", "_init_weights", "arange", "bias_u", "bias_v", "class", "config", "copy_", "def", "dtype", "elif", "else", "get_text_config", "getattr", "hasattr", "hidden_size", "if", "init", "initializer_range", "int64", "inv_freq", "isinstance", "mean", "module", "normal_", "self", "std", "super", "torch"], "parakeet/modeling_parakeet.py:ParakeetPreTrainedModel._get_subsampling_output_length": ["ModelCTCConfig", "ModelPreTrainedModel", "PreTrainedModel", "Tensor", "_", "_get_subsampling_output_length", "add_pad", "all_paddings", "class", "config", "def", "div", "dtype", "else", "encoder_config", "float", "floor", "for", "if", "in", "input_lengths", "int", "isinstance", "kernel_size", "lengths", "log2", "math", "num_layers", "range", "return", "self", "stride", "subsampling_conv_kernel_size", "subsampling_conv_stride", "subsampling_factor", "to", "torch"], "parakeet/modeling_parakeet.py:ParakeetPreTrainedModel._get_output_attention_mask": ["ModelPreTrainedModel", "None", "PreTrainedModel", "Tensor", "_get_output_attention_mask", "_get_subsampling_output_length", "arange", "attention_mask", "class", "def", "device", "else", "if", "int", "is", "max", "max_length", "not", "output_lengths", "return", "self", "sum", "target_length", "torch"], "parakeet/modeling_parakeet.py:ParakeetEncoder.__init__": ["False", "ModelEncoder", "ModelEncoderBlock", "ModelEncoderConfig", "ModelEncoderRelPositionalEncoding", "ModelEncoderSubsamplingConv2D", "ModelPreTrainedModel", "ModuleList", "__init__", "class", "config", "def", "dropout", "dropout_positions", "else", "encode_positions", "for", "gradient_checkpointing", "hidden_size", "if", "in", "input_scale", "layer_idx", "layerdrop", "layers", "math", "nn", "num_hidden_layers", "post_init", "range", "scale_input", "self", "sqrt", "subsampling", "super"], "parakeet/modeling_parakeet.py:ParakeetEncoder.forward": ["BaseModelOutput", "False", "ModelEncoder", "ModelEncoderModelOutput", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "_get_output_attention_mask", "attention_mask", "bool", "class", "def", "dropout", "dropout_positions", "dropout_probability", "else", "encode_positions", "encoder_layer", "expand", "for", "forward", "functional", "hidden_states", "if", "in", "input_features", "input_scale", "int", "is", "kwargs", "last_hidden_state", "layerdrop", "layers", "nn", "not", "output_attention_mask", "output_mask", "p", "position_embeddings", "r", "rand", "return", "self", "shape", "subsampling", "target_length", "to_drop", "torch", "training", "transpose", "unsqueeze"], "parakeet/modeling_parakeet.py:ParakeetForCTC.__init__": ["Conv1d", "ModelCTCConfig", "ModelEncoder", "ModelForCTC", "ModelPreTrainedModel", "__init__", "class", "config", "ctc_head", "def", "encoder", "encoder_config", "hidden_size", "kernel_size", "nn", "post_init", "self", "super", "vocab_size"], "parakeet/modeling_parakeet.py:ParakeetForCTC.forward": ["CausalLMOutput", "False", "ModelForCTC", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "_get_subsampling_output_length", "attention_mask", "attentions", "backends", "blank", "class", "config", "ctc_head", "ctc_loss", "ctc_loss_reduction", "ctc_zero_infinity", "cudnn", "def", "dim", "dtype", "else", "enabled", "encoder", "encoder_outputs", "flags", "flattened_targets", "float32", "forward", "functional", "hidden_states", "if", "input_features", "input_lengths", "is", "kwargs", "labels", "labels_mask", "last_hidden_state", "log_probs", "log_softmax", "logits", "long", "loss", "masked_select", "nn", "not", "ones_like", "pad_token_id", "r", "reduction", "return", "self", "sum", "target_lengths", "torch", "transpose", "with", "zero_infinity"], "parakeet/modeling_parakeet.py:ParakeetForCTC.generate": ["CausalLMOutput", "False", "LongTensor", "ModelForCTC", "ModelGenerateOutput", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "_get_output_attention_mask", "argmax", "attention_mask", "attentions", "bool", "class", "config", "def", "dim", "forward", "generate", "hidden_states", "if", "input_features", "is", "kwargs", "logits", "not", "outputs", "pad_token_id", "r", "return", "return_dict", "return_dict_in_generate", "self", "sequences", "shape", "target_length", "torch"], "patchtsmixer/modeling_patchtsmixer.py:PatchTSMixerGatedAttention.__init__": ["Linear", "ModelGatedAttention", "Module", "Softmax", "__init__", "attn_layer", "attn_softmax", "class", "def", "dim", "in_size", "int", "nn", "out_size", "self", "super"], "patchtsmixer/modeling_patchtsmixer.py:PatchTSMixerGatedAttention.forward": ["ModelGatedAttention", "Module", "attn_layer", "attn_softmax", "attn_weight", "class", "def", "forward", "inputs", "nn", "return", "self"], "patchtsmixer/modeling_patchtsmixer.py:PatchTSMixerBatchNorm.__init__": ["BatchNorm1d", "ModelBatchNorm", "ModelConfig", "Module", "__init__", "batchnorm", "class", "config", "d_model", "def", "eps", "nn", "norm_eps", "self", "super"], "patchtsmixer/modeling_patchtsmixer.py:PatchTSMixerBatchNorm.forward": ["ModelBatchNorm", "Module", "Tensor", "batchnorm", "class", "def", "forward", "inputs", "nn", "output", "return", "self", "torch", "transpose"], "patchtsmixer/modeling_patchtsmixer.py:PatchTSMixerPositionalEncoding.__init__": ["ModelConfig", "ModelPositionalEncoding", "Module", "Parameter", "__init__", "_init_pe", "class", "config", "d_model", "def", "else", "if", "nn", "num_Modeles", "position_enc", "self", "super", "torch", "use_positional_encoding", "zeros"], "patchtsmixer/modeling_patchtsmixer.py:PatchTSMixerPositionalEncoding._init_pe": ["Available", "False", "ModelConfig", "ModelPositionalEncoding", "Module", "Parameter", "True", "ValueError", "_init_pe", "a", "and", "arange", "are", "class", "config", "cos", "d_model", "def", "div_term", "elif", "else", "encoder", "exp", "f", "if", "is", "log", "math", "mean", "nn", "not", "num_Modeles", "position", "position_enc", "positional", "positional_encoding_type", "raise", "randn", "random", "requires_grad", "return", "sin", "sincos", "std", "torch", "types", "unsqueeze", "valid", "zeros"], "patchtsmixer/modeling_patchtsmixer.py:PatchTSMixerPositionalEncoding.forward": ["ModelPositionalEncoding", "Model_input", "Module", "Tensor", "class", "def", "forward", "hidden_state", "nn", "position_enc", "return", "self", "torch"], "patchtsmixer/modeling_patchtsmixer.py:PatchTSMixerNormLayer.__init__": ["LayerNorm", "ModelBatchNorm", "ModelConfig", "ModelNormLayer", "Module", "__init__", "batch", "class", "config", "d_model", "def", "else", "eps", "if", "in", "lower", "nn", "norm", "norm_eps", "norm_mlp", "self", "super"], "patchtsmixer/modeling_patchtsmixer.py:PatchTSMixerNormLayer.forward": ["ModelNormLayer", "Module", "Tensor", "batch", "class", "def", "else", "forward", "if", "in", "inputs", "inputs_reshaped", "lower", "nn", "norm", "norm_mlp", "reshape", "return", "self", "shape", "torch"], "patchtsmixer/modeling_patchtsmixer.py:PatchTSMixerMLP.__init__": ["Dropout", "Linear", "ModelMLP", "Module", "__init__", "class", "config", "def", "dropout", "dropout1", "dropout2", "expansion_factor", "fc1", "fc2", "in_features", "nn", "num_hidden", "out_features", "self", "super"], "patchtsmixer/modeling_patchtsmixer.py:PatchTSMixerMLP.forward": ["ModelMLP", "Module", "Tensor", "class", "def", "dropout1", "dropout2", "fc1", "fc2", "forward", "functional", "gelu", "inputs", "nn", "return", "self", "torch"], "patchtsmixer/modeling_patchtsmixer.py:PatchTSMixerChannelFeatureMixerBlock.__init__": ["ModelChannelFeatureMixerBlock", "ModelConfig", "ModelGatedAttention", "ModelMLP", "ModelNormLayer", "Module", "__init__", "class", "config", "def", "gated_attn", "gating_block", "if", "in_features", "in_size", "mlp", "nn", "norm", "num_input_channels", "out_features", "out_size", "self", "super"], "patchtsmixer/modeling_patchtsmixer.py:PatchTSMixerChannelFeatureMixerBlock.forward": ["ModelChannelFeatureMixerBlock", "Module", "Tensor", "class", "def", "forward", "gated_attn", "gating_block", "if", "inputs", "mlp", "nn", "norm", "out", "permute", "residual", "return", "self", "torch"], "patchtsmixer/modeling_patchtsmixer.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "contiguous", "def", "dim", "dropout", "float", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "shape", "size", "softmax", "torch", "training", "transpose", "value"], "patchtsmixer/modeling_patchtsmixer.py:PatchTSMixerAttention.__init__": ["False", "Linear", "ModelAttention", "ModelConfig", "Module", "None", "True", "ValueError", "__init__", "and", "be", "bias", "bool", "by", "class", "config", "def", "divisible", "dropout", "embed_dim", "f", "float", "got", "head_dim", "if", "int", "is_causal", "is_decoder", "k_proj", "must", "nn", "num_heads", "out_proj", "q_proj", "raise", "scaling", "self", "super", "v_proj"], "patchtsmixer/modeling_patchtsmixer.py:PatchTSMixerAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "False", "FlashAttentionKwargs", "ModelAttention", "Module", "None", "Tensor", "Unpack", "_attn_implementation", "attention_interface", "attention_mask", "attn_output", "attn_weights", "bool", "bsz", "class", "config", "contiguous", "current_states", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "head_dim", "hidden_states", "if", "is", "is_cross_attention", "k_proj", "key_states", "key_value_states", "kv_input_shape", "kwargs", "nn", "not", "out_proj", "output_attentions", "q_input_shape", "q_proj", "query_states", "reshape", "return", "scaling", "self", "shape", "src_len", "tgt_len", "torch", "training", "transpose", "tuple", "v_proj", "value_states", "view"], "patchtsmixer/modeling_patchtsmixer.py:PatchMixerBlock.__init__": ["ModelAttention", "ModelConfig", "ModelGatedAttention", "ModelMLP", "ModelMixerBlock", "ModelNormLayer", "Module", "__init__", "class", "config", "d_model", "def", "dropout", "embed_dim", "gated_attn", "gating_block", "if", "in_features", "in_size", "mlp", "nn", "norm", "norm_attn", "num_Modeles", "num_heads", "out_features", "out_size", "self", "self_attn", "self_attn_heads", "self_attn_layer", "super"], "patchtsmixer/modeling_patchtsmixer.py:PatchMixerBlock.forward": ["False", "ModelMixerBlock", "Module", "_", "batch_size", "class", "d_model", "def", "forward", "gated_attn", "gating_block", "hidden_state", "hidden_state_reshaped", "if", "mlp", "n_vars", "nn", "norm", "norm_attn", "num_Modeles", "out", "output_attentions", "reshape", "residual", "return", "self", "self_attn", "self_attn_layer", "shape", "transpose", "x_attn"], "patchtsmixer/modeling_patchtsmixer.py:FeatureMixerBlock.__init__": ["ModelConfig", "ModelGatedAttention", "ModelMLP", "ModelMixerBlock", "ModelNormLayer", "Module", "__init__", "class", "config", "d_model", "def", "gated_attn", "gating_block", "if", "in_Models", "in_size", "mlp", "nn", "norm", "out_Models", "out_size", "self", "super"], "patchtsmixer/modeling_patchtsmixer.py:FeatureMixerBlock.forward": ["ModelMixerBlock", "Module", "Tensor", "class", "def", "forward", "gated_attn", "gating_block", "hidden", "if", "mlp", "nn", "norm", "out", "residual", "return", "self", "torch"], "patchtsmixer/modeling_patchtsmixer.py:PatchTSMixerLayer.__init__": ["FeatureMixerBlock", "ModelChannelFeatureMixerBlock", "ModelConfig", "ModelLayer", "ModelMixerBlock", "Model_mixer", "Module", "__init__", "channel_feature_mixer", "class", "config", "def", "feature_mixer", "if", "mix_channel", "mode", "nn", "self", "super"], "patchtsmixer/modeling_patchtsmixer.py:PatchTSMixerLayer.forward": ["ModelLayer", "Model_mixer", "Module", "Tensor", "channel_feature_mixer", "class", "def", "feature_mixer", "forward", "hidden", "if", "mix_channel", "mode", "nn", "return", "self", "torch"], "patchtsmixer/modeling_patchtsmixer.py:PatchTSMixerBlock.__init__": ["ModelBlock", "ModelConfig", "ModelLayer", "Module", "ModuleList", "_", "__init__", "class", "config", "def", "for", "in", "mixers", "nn", "num_layers", "range", "self", "super"], "patchtsmixer/modeling_patchtsmixer.py:PatchTSMixerBlock.forward": ["False", "ModelBlock", "Module", "None", "all_hidden_states", "append", "bool", "class", "def", "else", "embedding", "for", "forward", "hidden_state", "if", "in", "mixers", "mod", "nn", "output_hidden_states", "return", "self"], "patchtsmixer/modeling_patchtsmixer.py:PatchTSMixerForPredictionHead.__init__": ["Dropout", "Flatten", "Linear", "ModelConfig", "ModelForPredictionHead", "Module", "None", "__init__", "base_forecast_block", "class", "config", "d_model", "def", "distribution_output", "dropout_layer", "else", "flatten", "get_parameter_projection", "head_dropout", "if", "is", "nn", "not", "num_Modeles", "prediction_channel_indices", "prediction_length", "self", "sort", "start_dim", "super"], "patchtsmixer/modeling_patchtsmixer.py:PatchTSMixerForPredictionHead.forward": ["ModelForPredictionHead", "Module", "None", "base_forecast_block", "class", "def", "dropout_layer", "else", "flatten", "for", "forecast", "forward", "hidden_features", "if", "in", "is", "isinstance", "nn", "not", "prediction_channel_indices", "return", "self", "transpose", "tuple", "z"], "patchtsmixer/modeling_patchtsmixer.py:PatchTSMixerLinearHead.__init__": ["Dropout", "Flatten", "Linear", "ModelConfig", "ModelLinearHead", "Module", "None", "__init__", "class", "config", "d_model", "def", "distribution_output", "dropout", "else", "flatten", "get_parameter_projection", "head_aggregation", "head_dropout", "if", "is", "mul_factor", "nn", "num_Modeles", "num_input_channels", "num_targets", "output_range", "projection", "self", "start_dim", "super"], "patchtsmixer/modeling_patchtsmixer.py:PatchTSMixerLinearHead.forward": ["ModelLinearHead", "Module", "None", "and", "avg_pool", "class", "def", "dim", "distribution_output", "dropout", "elif", "flatten", "forward", "head_aggregation", "hidden_features", "if", "is", "max", "max_pool", "mean", "nn", "not", "output_range", "projection", "return", "self", "sigmoid", "torch", "transpose", "use_last", "values"], "patchtsmixer/modeling_patchtsmixer.py:PatchTSMixerPreTrainedModel._init_weights": ["BatchNorm1d", "LayerNorm", "Linear", "ModelBatchNorm", "ModelPositionalEncoding", "ModelPreTrainedModel", "None", "PreTrainedModel", "_init_weights", "batchnorm", "bias", "class", "config", "def", "elif", "getattr", "if", "init", "init_std", "is", "isinstance", "mean", "module", "nn", "normal_", "not", "num_batches_tracked", "ones_", "position_enc", "positional_encoding_type", "random", "running_mean", "running_var", "self", "std", "weight", "zeros_"], "patchtsmixer/modeling_patchtsmixer.py:PatchTSMixerPretrainHead.__init__": ["Dropout", "Linear", "ModelConfig", "ModelPretrainHead", "Model_length", "Module", "__init__", "base_pt_block", "class", "config", "d_model", "def", "dropout_layer", "head_dropout", "nn", "self", "super"], "patchtsmixer/modeling_patchtsmixer.py:PatchTSMixerPretrainHead.forward": ["ModelPretrainHead", "Module", "base_pt_block", "class", "def", "dropout_layer", "forecast", "forward", "hidden_features", "nn", "return", "self"], "patchtsmixer/modeling_patchtsmixer.py:random_masking": ["False", "Mask", "Model_masking", "None", "Tensor", "ValueError", "and", "argsort", "batch_size", "be", "between", "bool", "channel_consistent_masking", "def", "device", "dim", "else", "f", "float", "gather", "has", "ids_restore", "ids_shuffle", "if", "index", "inputs", "inputs_mask", "int", "is", "len_keep", "list", "mask", "mask_ratio", "mask_value", "masked_fill", "noise", "not", "num_channels", "num_features", "ones", "or", "raise", "rand", "ratio", "repeat", "return", "sequence_length", "shape", "to", "torch", "unmasked_channel_indices", "unsqueeze"], "patchtsmixer/modeling_patchtsmixer.py:forecast_masking": ["Model_mask_ratios", "Model_masking", "None", "Tensor", "ValueError", "_", "and", "append", "batch1", "batch2", "batch_size", "be", "bool", "def", "device", "elif", "f", "for", "greater", "if", "in", "inputs", "inputs_mask", "int", "is", "isinstance", "key", "lambda", "less", "list", "mask", "mask_value", "masked_fill", "not", "num_Model_mask_patches", "num_channels", "num_features", "or", "patch_len", "patch_length", "patches", "perm", "raise", "randperm", "ratio", "repeat", "return", "sequence_length", "shape", "should", "sorted", "sum", "t_list", "temp_len", "than", "torch", "total", "total_length", "total_ratio", "unmasked_channel_indices", "unsqueeze", "x", "zeros", "zip"], "patchtsmixer/modeling_patchtsmixer.py:PatchTSMixerPatchify.__init__": ["Model", "ModelConfig", "ModelModelify", "Model_length", "Model_stride", "Module", "Sequence", "ValueError", "__init__", "be", "class", "config", "context_length", "def", "f", "greater", "has", "if", "length", "max", "new_sequence_length", "nn", "num_Modeles", "raise", "self", "sequence_length", "sequence_start", "super", "than", "the", "to"], "patchtsmixer/modeling_patchtsmixer.py:PatchTSMixerPatchify.forward": ["Input", "ModelModelify", "Model_length", "Model_stride", "Module", "Tensor", "ValueError", "class", "configuration", "contiguous", "def", "dimension", "doesn", "f", "forward", "if", "length", "match", "model", "nn", "output", "past_values", "raise", "return", "self", "sequence", "sequence_length", "sequence_start", "shape", "size", "step", "t", "torch", "transpose", "unfold"], "patchtsmixer/modeling_patchtsmixer.py:PatchTSMixerMasking.__init__": ["ModelConfig", "ModelMasking", "Module", "None", "__init__", "channel_consistent_masking", "class", "config", "def", "if", "is", "mask_type", "mask_value", "nn", "not", "num_forecast_mask_Modeles", "random_mask_ratio", "self", "sorted", "super", "unmasked_channel_indices"], "patchtsmixer/modeling_patchtsmixer.py:PatchTSMixerMasking.forward": ["Invalid", "ModelMasking", "Model_input", "Module", "Tensor", "ValueError", "bool", "channel_consistent_masking", "class", "def", "elif", "else", "f", "forecast", "forecast_masking", "forward", "if", "inputs", "mask", "mask_ratio", "mask_type", "mask_value", "masked_input", "nn", "num_forecast_mask_Modeles", "raise", "random", "random_mask_ratio", "random_masking", "return", "self", "torch", "type", "unmasked_channel_indices"], "patchtsmixer/modeling_patchtsmixer.py:PatchTSMixerStdScaler.__init__": ["ModelConfig", "ModelStdScaler", "Module", "True", "__init__", "class", "config", "def", "dim", "else", "hasattr", "if", "keepdim", "minimum_scale", "nn", "scaling_dim", "self", "super"], "patchtsmixer/modeling_patchtsmixer.py:PatchTSMixerStdScaler.forward": ["ModelStdScaler", "Module", "Tensor", "clamp_min", "class", "data", "def", "denominator", "dim", "forward", "keepdim", "loc", "minimum_scale", "nn", "observed_indicator", "return", "scale", "self", "sqrt", "sum", "torch", "tuple", "variance"], "patchtsmixer/modeling_patchtsmixer.py:PatchTSMixerMeanScaler.__init__": ["ModelConfig", "ModelMeanScaler", "Module", "None", "True", "__init__", "class", "config", "def", "default_scale", "dim", "else", "hasattr", "if", "keepdim", "minimum_scale", "nn", "scaling_dim", "self", "super"], "patchtsmixer/modeling_patchtsmixer.py:PatchTSMixerMeanScaler.forward": ["ModelMeanScaler", "Module", "None", "Tensor", "True", "abs", "batch_observations", "batch_sum", "clamp", "class", "data", "def", "default_scale", "dim", "else", "forward", "if", "is", "keepdim", "min", "minimum_scale", "nn", "not", "num_observed", "observed_indicator", "ones_like", "return", "scale", "scaled_data", "self", "squeeze", "sum", "torch", "ts_sum", "tuple", "where", "zeros_like"], "patchtsmixer/modeling_patchtsmixer.py:PatchTSMixerNOPScaler.__init__": ["ModelConfig", "ModelNOPScaler", "Module", "True", "__init__", "class", "config", "def", "dim", "else", "hasattr", "if", "keepdim", "nn", "scaling_dim", "self", "super"], "patchtsmixer/modeling_patchtsmixer.py:PatchTSMixerNOPScaler.forward": ["False", "ModelNOPScaler", "Module", "None", "Tensor", "class", "data", "def", "dim", "forward", "keepdim", "loc", "mean", "nn", "observed_indicator", "ones_like", "requires_grad", "return", "scale", "self", "torch", "tuple", "zeros_like"], "patchtsmixer/modeling_patchtsmixer.py:PatchTSMixerEncoder.__init__": ["Linear", "ModelBlock", "ModelConfig", "ModelEncoder", "ModelPositionalEncoding", "ModelPreTrainedModel", "Model_length", "Modeler", "None", "__init__", "class", "config", "d_model", "def", "else", "if", "mlp_mixer_encoder", "nn", "positional_encoder", "post_init", "self", "super", "use_positional_encoding", "use_return_dict"], "patchtsmixer/modeling_patchtsmixer.py:PatchTSMixerEncoder.forward": ["False", "ModelEncoder", "ModelEncoderOutput", "ModelPreTrainedModel", "Modeler", "Modeles", "None", "Tensor", "bool", "class", "def", "else", "for", "forward", "hidden_states", "if", "in", "is", "kwargs", "last_hidden_state", "mlp_mixer_encoder", "not", "output_hidden_states", "past_values", "positional_encoder", "r", "return", "return_dict", "self", "torch", "tuple", "use_return_dict", "v"], "patchtsmixer/modeling_patchtsmixer.py:PatchTSMixerModel.__init__": ["False", "ModelConfig", "ModelEncoder", "ModelMasking", "ModelMeanScaler", "ModelModel", "ModelModelify", "ModelNOPScaler", "ModelPreTrainedModel", "ModelStdScaler", "Modeling", "None", "True", "__init__", "bool", "class", "config", "def", "elif", "else", "encoder", "if", "is", "mask_input", "masking", "mean", "or", "post_init", "r", "scaler", "scaling", "self", "std", "super", "use_return_dict"], "patchtsmixer/modeling_patchtsmixer.py:PatchTSMixerModel.forward": ["False", "ModelEncoderOutput", "ModelModel", "ModelModelOutput", "ModelPreTrainedModel", "Model_input", "Modeled_x", "Modeling", "None", "Tensor", "bool", "class", "def", "else", "enc_input", "encoder", "encoder_output", "for", "forward", "hidden_states", "if", "in", "is", "isinstance", "kwargs", "last_hidden_state", "loc", "mask", "masking", "not", "observed_mask", "ones_like", "output_hidden_states", "past_values", "r", "return", "return_dict", "scale", "scaled_past_values", "scaler", "self", "torch", "tuple", "use_return_dict", "v"], "patchtsmixer/modeling_patchtsmixer.py:PatchTSMixerForPretraining.__init__": ["ModelConfig", "ModelForPretraining", "ModelModel", "ModelPreTrainedModel", "ModelPretrainHead", "True", "__init__", "class", "config", "def", "head", "if", "mask_input", "masked_loss", "model", "post_init", "self", "super", "use_return_dict"], "patchtsmixer/modeling_patchtsmixer.py:PatchTSMixerForPretraining.forward": ["False", "MSELoss", "ModelForPreTrainingOutput", "ModelForPretraining", "ModelModelOutput", "ModelPreTrainedModel", "Model_input", "None", "Tensor", "True", "and", "bool", "class", "def", "dim", "else", "for", "forward", "head", "hidden_states", "if", "in", "is", "isinstance", "kwargs", "last_hidden_state", "loss", "loss_val", "mask", "masked_loss", "mean", "model", "model_output", "nn", "none", "not", "observed_mask", "output_hidden_states", "past_values", "prediction_outputs", "r", "reduction", "return", "return_dict", "return_loss", "self", "sum", "torch", "tuple", "use_return_dict", "v", "x_hat"], "patchtsmixer/modeling_patchtsmixer.py:nll": ["Distribution", "Model", "Tensor", "def", "distributions", "input", "log_prob", "return", "target", "torch"], "patchtsmixer/modeling_patchtsmixer.py:weighted_average": ["Model_average", "Model_tensor", "None", "Tensor", "clamp", "def", "dim", "else", "if", "input_tensor", "is", "mean", "min", "not", "return", "sum", "sum_weights", "torch", "weights", "where", "zeros_like"], "patchtsmixer/modeling_patchtsmixer.py:PatchTSMixerForPrediction.__init__": ["ModelConfig", "ModelForPrediction", "ModelForPredictionHead", "ModelModel", "ModelPreTrainedModel", "NegativeBinomialOutput", "None", "NormalOutput", "StudentTOutput", "Unknown", "ValueError", "__init__", "class", "config", "def", "dim", "distribution", "distribution_output", "distribution_output_map", "else", "f", "get", "head", "if", "is", "loss", "model", "mse", "negative_binomial", "normal", "not", "num_parallel_samples", "output", "output_class", "post_init", "prediction_channel_indices", "prediction_length", "raise", "self", "student_t", "super", "use_return_dict"], "patchtsmixer/modeling_patchtsmixer.py:PatchTSMixerForPrediction.forward": ["Allowed", "False", "Invalid", "MSELoss", "ModelForPrediction", "ModelForPredictionOutput", "ModelModelOutput", "ModelPreTrainedModel", "None", "Tensor", "True", "ValueError", "and", "bool", "class", "def", "distribution", "distribution_output", "elif", "else", "for", "forward", "function", "future_values", "head", "hidden_states", "if", "in", "is", "isinstance", "kwargs", "last_hidden_state", "loc", "loss", "loss_val", "mean", "model", "model_output", "mse", "nll", "nn", "not", "observed_mask", "output_hidden_states", "past_values", "prediction_channel_indices", "prediction_outputs", "r", "raise", "reduction", "return", "return_dict", "return_loss", "scale", "self", "torch", "tuple", "use_return_dict", "v", "values", "weighted_average", "y_hat"], "patchtsmixer/modeling_patchtsmixer.py:PatchTSMixerForPrediction.generate": ["False", "ModelForPrediction", "ModelPreTrainedModel", "None", "SampleModelPredictionOutput", "Tensor", "_", "class", "def", "dim", "distribution", "distribution_output", "for", "future_values", "generate", "in", "loc", "num_parallel_samples", "observed_mask", "output_hidden_states", "outputs", "past_values", "prediction_outputs", "range", "return", "sample", "samples", "scale", "self", "sequences", "stack", "torch"], "patchtsmixer/modeling_patchtsmixer.py:PatchTSMixerForTimeSeriesClassification.__init__": ["InjectScalerStatistics4D", "ModelConfig", "ModelForTimeSeriesClassification", "ModelLinearHead", "ModelModel", "ModelPreTrainedModel", "None", "True", "__init__", "class", "config", "d_model", "def", "else", "head", "if", "in", "inject_scale", "mean", "model", "num_Modeles", "post_init", "scaling", "self", "std", "super", "use_return_dict"], "patchtsmixer/modeling_patchtsmixer.py:PatchTSMixerForTimeSeriesClassification.forward": ["CrossEntropyLoss", "False", "ModelForTimeSeriesClassification", "ModelForTimeSeriesClassificationOutput", "ModelModelOutput", "ModelPreTrainedModel", "None", "Tensor", "True", "and", "bool", "class", "def", "else", "for", "forward", "head", "hidden_states", "if", "in", "inject_scale", "is", "isinstance", "kwargs", "last_hidden_state", "loc", "loss", "loss_val", "model", "model_output", "nn", "not", "output_hidden_states", "past_values", "prediction_outputs", "r", "return", "return_dict", "return_loss", "scale", "self", "target_values", "torch", "tuple", "use_return_dict", "v", "y_hat"], "patchtsmixer/modeling_patchtsmixer.py:InjectScalerStatistics4D.__init__": ["Linear", "ModelScalerStatistics4D", "Module", "__init__", "class", "d_model", "def", "expansion", "int", "inverse_trans_compression", "inverse_trans_expansion", "map_scale_compression", "map_scale_expansion", "nn", "num_patches", "self", "super"], "patchtsmixer/modeling_patchtsmixer.py:InjectScalerStatistics4D.forward": ["ModelScalerStatistics4D", "Module", "Tensor", "cat", "class", "concat_stats", "def", "dim", "forward", "inputs", "inverse_trans_compression", "inverse_trans_expansion", "loc", "map_scale_compression", "map_scale_expansion", "mean", "nn", "num_patches", "repeat", "return", "scale", "self", "stdev", "torch", "transpose", "unsqueeze"], "patchtsmixer/modeling_patchtsmixer.py:PatchTSMixerForRegression.__init__": ["InjectScalerStatistics4D", "ModelConfig", "ModelForRegression", "ModelLinearHead", "ModelModel", "ModelPreTrainedModel", "NegativeBinomialOutput", "None", "NormalOutput", "StudentTOutput", "True", "Unknown", "ValueError", "__init__", "class", "config", "d_model", "def", "dim", "distribution", "distribution_output", "distribution_output_map", "else", "f", "get", "head", "if", "in", "inject_scale", "is", "loss", "mean", "model", "mse", "negative_binomial", "normal", "not", "num_Modeles", "num_parallel_samples", "num_targets", "output", "output_class", "post_init", "raise", "scaling", "self", "std", "student_t", "super", "use_return_dict"], "patchtsmixer/modeling_patchtsmixer.py:PatchTSMixerForRegression.forward": ["Allowed", "Exception", "False", "Invalid", "MSELoss", "ModelForRegression", "ModelForRegressionOutput", "ModelModelOutput", "ModelPreTrainedModel", "None", "Tensor", "True", "ValueError", "and", "any", "be", "bool", "cannot", "class", "config", "def", "distribution", "distribution_output", "elif", "else", "for", "forward", "function", "head", "hidden_states", "if", "in", "inject_scale", "is", "isinstance", "item", "kwargs", "last_hidden_state", "loc", "loss", "loss_val", "mean", "model", "model_output", "mse", "negative", "negative_binomial", "nll", "nn", "not", "num_targets", "output_hidden_states", "past_values", "r", "raise", "reduction", "regression_outputs", "return", "return_dict", "return_loss", "scale", "self", "target_values", "torch", "tuple", "use_return_dict", "v", "values", "view", "weighted_average", "y_hat"], "patchtsmixer/modeling_patchtsmixer.py:PatchTSMixerForRegression.generate": ["False", "ModelForRegression", "ModelPreTrainedModel", "None", "SampleModelRegressionOutput", "Tensor", "_", "class", "config", "def", "dim", "distribution", "distribution_output", "for", "generate", "in", "num_parallel_samples", "num_targets", "output_hidden_states", "outputs", "past_values", "range", "regression_outputs", "return", "sample", "samples", "self", "sequences", "stack", "target_values", "torch", "view"], "patchtst/modeling_patchtst.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "contiguous", "def", "dim", "dropout", "float", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "shape", "size", "softmax", "torch", "training", "transpose", "value"], "patchtst/modeling_patchtst.py:PatchTSTAttention.__init__": ["False", "Linear", "ModelAttention", "ModelConfig", "Module", "None", "True", "ValueError", "__init__", "and", "be", "bias", "bool", "by", "class", "config", "def", "divisible", "dropout", "embed_dim", "f", "float", "got", "head_dim", "if", "int", "is_causal", "is_decoder", "k_proj", "must", "nn", "num_heads", "out_proj", "q_proj", "raise", "scaling", "self", "super", "v_proj"], "patchtst/modeling_patchtst.py:PatchTSTAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "False", "FlashAttentionKwargs", "ModelAttention", "Module", "None", "Tensor", "Unpack", "_attn_implementation", "attention_interface", "attention_mask", "attn_output", "attn_weights", "bool", "bsz", "class", "config", "contiguous", "current_states", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "head_dim", "hidden_states", "if", "is", "is_cross_attention", "k_proj", "key_states", "key_value_states", "kv_input_shape", "kwargs", "nn", "not", "out_proj", "output_attentions", "q_input_shape", "q_proj", "query_states", "reshape", "return", "scaling", "self", "shape", "src_len", "tgt_len", "torch", "training", "transpose", "tuple", "v_proj", "value_states", "view"], "patchtst/modeling_patchtst.py:PatchTSTBatchNorm.__init__": ["BatchNorm1d", "ModelBatchNorm", "ModelConfig", "Module", "__init__", "batchnorm", "class", "config", "d_model", "def", "eps", "nn", "norm_eps", "self", "super"], "patchtst/modeling_patchtst.py:PatchTSTBatchNorm.forward": ["ModelBatchNorm", "Module", "Tensor", "batchnorm", "class", "def", "forward", "inputs", "nn", "output", "return", "self", "torch", "transpose"], "patchtst/modeling_patchtst.py:random_masking": ["False", "Mask", "Model_masking", "None", "Tensor", "ValueError", "and", "argsort", "batch_size", "be", "between", "bool", "channel_consistent_masking", "def", "device", "dim", "else", "f", "float", "gather", "has", "ids_restore", "ids_shuffle", "if", "index", "inputs", "inputs_mask", "int", "is", "len_keep", "list", "mask", "mask_ratio", "mask_value", "masked_fill", "noise", "not", "num_channels", "num_features", "ones", "or", "raise", "rand", "ratio", "repeat", "return", "sequence_length", "shape", "to", "torch", "unmasked_channel_indices", "unsqueeze"], "patchtst/modeling_patchtst.py:forecast_masking": ["Model_mask_ratios", "Model_masking", "None", "Tensor", "ValueError", "_", "and", "append", "batch1", "batch2", "batch_size", "be", "bool", "def", "device", "elif", "f", "for", "greater", "if", "in", "inputs", "inputs_mask", "int", "is", "isinstance", "key", "lambda", "less", "list", "mask", "mask_value", "masked_fill", "not", "num_Model_mask_patches", "num_channels", "num_features", "or", "patch_len", "patch_length", "patches", "perm", "raise", "randperm", "ratio", "repeat", "return", "sequence_length", "shape", "should", "sorted", "sum", "t_list", "temp_len", "than", "torch", "total", "total_length", "total_ratio", "unmasked_channel_indices", "unsqueeze", "x", "zeros", "zip"], "patchtst/modeling_patchtst.py:PatchTSTPatchify.__init__": ["Model", "ModelConfig", "ModelModelify", "Model_length", "Model_stride", "Module", "Sequence", "ValueError", "__init__", "be", "class", "config", "context_length", "def", "f", "greater", "has", "if", "length", "max", "new_sequence_length", "nn", "num_Modeles", "raise", "self", "sequence_length", "sequence_start", "super", "than", "the", "to"], "patchtst/modeling_patchtst.py:PatchTSTPatchify.forward": ["Input", "ModelModelify", "Model_length", "Model_stride", "Module", "Tensor", "ValueError", "class", "configuration", "contiguous", "def", "dimension", "doesn", "f", "forward", "if", "length", "match", "model", "nn", "output", "past_values", "raise", "return", "self", "sequence", "sequence_length", "sequence_start", "shape", "size", "step", "t", "torch", "transpose", "unfold"], "patchtst/modeling_patchtst.py:PatchTSTMasking.__init__": ["ModelConfig", "ModelMasking", "Module", "None", "__init__", "channel_consistent_masking", "class", "config", "def", "if", "is", "mask_type", "mask_value", "nn", "not", "num_forecast_mask_Modeles", "random_mask_ratio", "self", "sorted", "super", "unmasked_channel_indices"], "patchtst/modeling_patchtst.py:PatchTSTMasking.forward": ["Invalid", "ModelMasking", "Model_input", "Module", "Tensor", "ValueError", "bool", "channel_consistent_masking", "class", "def", "elif", "else", "f", "forecast", "forecast_masking", "forward", "if", "inputs", "mask", "mask_ratio", "mask_type", "mask_value", "masked_input", "nn", "num_forecast_mask_Modeles", "raise", "random", "random_mask_ratio", "random_masking", "return", "self", "torch", "type", "unmasked_channel_indices"], "patchtst/modeling_patchtst.py:PatchTSTEncoderLayer.__init__": ["ACT2CLS", "Dropout", "Identity", "LayerNorm", "Linear", "ModelAttention", "ModelBatchNorm", "ModelConfig", "ModelEncoderLayer", "Module", "Sequential", "ValueError", "__init__", "a", "activation_function", "attention_dropout", "batchnorm", "bias", "channel_attention", "class", "config", "d_model", "def", "dropout", "dropout_path1", "dropout_path2", "dropout_path3", "elif", "else", "embed_dim", "eps", "f", "ff", "ff_dropout", "ffn_dim", "if", "is", "layer", "layernorm", "nn", "norm", "norm_eps", "norm_sublayer1", "norm_sublayer2", "norm_sublayer3", "norm_type", "not", "num_attention_heads", "num_heads", "path_dropout", "pre_norm", "raise", "self", "self_attn", "super", "supported", "type"], "patchtst/modeling_patchtst.py:PatchTSTEncoderLayer.forward": ["ModelEncoderLayer", "Module", "None", "Tensor", "_", "attn_output", "attn_weights", "batch_size", "bool", "channel_attention", "channel_attn_weights", "class", "contiguous", "d_model", "def", "dropout_path1", "dropout_path2", "dropout_path3", "else", "ff", "forward", "hidden_state", "hidden_states", "if", "nn", "norm_sublayer1", "norm_sublayer2", "norm_sublayer3", "num_input_channels", "output_attentions", "outputs", "pre_norm", "reshape", "return", "self", "self_attn", "sequence_length", "shape", "torch", "transpose", "view"], "patchtst/modeling_patchtst.py:PatchTSTPreTrainedModel._init_weights": ["BatchNorm1d", "GatheredParameters", "LayerNorm", "Linear", "ModelPositionalEncoding", "ModelPreTrainedModel", "Model_length", "Model_stride", "Module", "None", "PreTrainedModel", "_init_pe", "_init_weights", "bias", "class", "cls_token", "config", "context_length", "copy_", "deepspeed", "def", "elif", "else", "getattr", "if", "init", "init_std", "is", "is_deepspeed_zero3_enabled", "isinstance", "max", "mean", "modifier_rank", "module", "nn", "normal_", "not", "num_Modeles", "num_batches_tracked", "numel", "ones_", "position_enc", "running_mean", "running_var", "self", "std", "use_cls_token", "weight", "with", "zero", "zeros_"], "patchtst/modeling_patchtst.py:PatchTSTPreTrainedModel._set_gradient_checkpointing": ["False", "ModelEncoder", "ModelPreTrainedModel", "PreTrainedModel", "_set_gradient_checkpointing", "class", "def", "gradient_checkpointing", "if", "isinstance", "module", "self", "value"], "patchtst/modeling_patchtst.py:PatchTSTEmbedding.__init__": ["Linear", "ModelConfig", "ModelEmbedding", "Model_length", "Module", "ModuleList", "_", "__init__", "append", "class", "config", "d_model", "def", "else", "for", "if", "in", "input_embedding", "nn", "num_input_channels", "range", "self", "share_embedding", "super"], "patchtst/modeling_patchtst.py:PatchTSTEmbedding.forward": ["ModelEmbedding", "Model_input", "Module", "Tensor", "The", "ValueError", "as", "batch", "be", "channels", "class", "config", "def", "defined", "dim", "else", "embeddings", "f", "for", "forward", "has", "i", "if", "in", "input", "input_embedding", "nn", "num_input_channels", "number", "of", "raise", "range", "return", "same", "self", "shape", "share_embedding", "stack", "the", "to", "torch"], "patchtst/modeling_patchtst.py:PatchTSTPositionalEncoding.__init__": ["Dropout", "Identity", "ModelConfig", "ModelPositionalEncoding", "Module", "Parameter", "__init__", "_init_pe", "class", "cls_token", "config", "d_model", "def", "else", "if", "int", "nn", "num_Modeles", "num_input_channels", "position_enc", "positional_dropout", "self", "super", "torch", "use_cls_token", "zeros"], "patchtst/modeling_patchtst.py:PatchTSTPositionalEncoding._init_pe": ["Available", "False", "ModelConfig", "ModelPositionalEncoding", "Module", "Parameter", "True", "ValueError", "_init_pe", "a", "and", "arange", "are", "class", "config", "cos", "d_model", "def", "div_term", "elif", "else", "encoder", "exp", "f", "if", "int", "is", "log", "math", "mean", "nn", "not", "num_Modeles", "position", "position_enc", "positional", "positional_encoding_type", "raise", "randn", "random", "requires_grad", "return", "sin", "sincos", "std", "torch", "types", "unsqueeze", "valid", "zeros"], "patchtst/modeling_patchtst.py:PatchTSTPositionalEncoding.forward": ["ModelPositionalEncoding", "Model_input", "Module", "Tensor", "cat", "class", "cls_token", "cls_tokens", "def", "dim", "else", "expand", "forward", "hidden_state", "if", "nn", "num_input_channels", "position_enc", "positional_dropout", "return", "self", "shape", "torch", "use_cls_token"], "patchtst/modeling_patchtst.py:PatchTSTEncoder.__init__": ["False", "ModelConfig", "ModelEmbedding", "ModelEncoder", "ModelEncoderLayer", "ModelPositionalEncoding", "ModelPreTrainedModel", "ModuleList", "__init__", "class", "config", "def", "embedder", "for", "gradient_checkpointing", "i", "in", "int", "layers", "nn", "num_Modeles", "num_hidden_layers", "positional_encoder", "post_init", "range", "self", "super"], "patchtst/modeling_patchtst.py:PatchTSTEncoder.forward": ["BaseModelOutput", "ModelEncoder", "ModelPreTrainedModel", "Model_input", "None", "Tensor", "all_attentions", "attentions", "bool", "class", "config", "def", "else", "embedder", "encoder_layer", "encoder_states", "for", "forward", "hidden_state", "hidden_states", "if", "in", "is", "kwargs", "last_hidden_state", "layer_outputs", "layers", "not", "output_attentions", "output_hidden_states", "positional_encoder", "return", "self", "torch"], "patchtst/modeling_patchtst.py:nll": ["Distribution", "Model", "Tensor", "def", "distributions", "input", "log_prob", "return", "target", "torch"], "patchtst/modeling_patchtst.py:weighted_average": ["Model_average", "Model_tensor", "None", "Tensor", "clamp", "def", "dim", "else", "if", "input_tensor", "is", "mean", "min", "not", "return", "sum", "sum_weights", "torch", "weights", "where", "zeros_like"], "patchtst/modeling_patchtst.py:PatchTSTStdScaler.__init__": ["ModelConfig", "ModelStdScaler", "Module", "True", "__init__", "class", "config", "def", "dim", "else", "hasattr", "if", "keepdim", "minimum_scale", "nn", "scaling_dim", "self", "super"], "patchtst/modeling_patchtst.py:PatchTSTStdScaler.forward": ["ModelStdScaler", "Module", "Tensor", "clamp_min", "class", "data", "def", "denominator", "dim", "forward", "keepdim", "loc", "minimum_scale", "nn", "observed_indicator", "return", "scale", "self", "sqrt", "sum", "torch", "tuple", "variance"], "patchtst/modeling_patchtst.py:PatchTSTMeanScaler.__init__": ["ModelConfig", "ModelMeanScaler", "Module", "None", "True", "__init__", "class", "config", "def", "default_scale", "dim", "else", "hasattr", "if", "keepdim", "minimum_scale", "nn", "scaling_dim", "self", "super"], "patchtst/modeling_patchtst.py:PatchTSTMeanScaler.forward": ["ModelMeanScaler", "Module", "None", "Tensor", "True", "abs", "batch_observations", "batch_sum", "clamp", "class", "data", "def", "default_scale", "dim", "else", "forward", "if", "is", "keepdim", "min", "minimum_scale", "nn", "not", "num_observed", "observed_indicator", "ones_like", "return", "scale", "scaled_data", "self", "squeeze", "sum", "torch", "ts_sum", "tuple", "where", "zeros_like"], "patchtst/modeling_patchtst.py:PatchTSTNOPScaler.__init__": ["ModelConfig", "ModelNOPScaler", "Module", "True", "__init__", "class", "config", "def", "dim", "else", "hasattr", "if", "keepdim", "nn", "scaling_dim", "self", "super"], "patchtst/modeling_patchtst.py:PatchTSTNOPScaler.forward": ["False", "ModelNOPScaler", "Module", "None", "Tensor", "class", "data", "def", "dim", "forward", "keepdim", "loc", "mean", "nn", "observed_indicator", "ones_like", "requires_grad", "return", "scale", "self", "torch", "tuple", "zeros_like"], "patchtst/modeling_patchtst.py:PatchTSTScaler.__init__": ["ModelConfig", "ModelMeanScaler", "ModelNOPScaler", "ModelScaler", "ModelStdScaler", "Module", "True", "__init__", "class", "config", "def", "elif", "else", "if", "is", "mean", "nn", "or", "scaler", "scaling", "self", "std", "super"], "patchtst/modeling_patchtst.py:PatchTSTScaler.forward": ["ModelScaler", "Module", "Tensor", "class", "data", "def", "forward", "loc", "nn", "observed_indicator", "return", "scale", "scaler", "self", "torch", "tuple"], "patchtst/modeling_patchtst.py:PatchTSTModel.__init__": ["Identity", "ModelConfig", "ModelEncoder", "ModelMasking", "ModelModel", "ModelModelify", "ModelPreTrainedModel", "ModelScaler", "Modelifier", "__init__", "class", "config", "def", "do_mask_input", "else", "encoder", "if", "masking", "nn", "num_Modeles", "post_init", "scaler", "self", "super"], "patchtst/modeling_patchtst.py:PatchTSTModel.forward": ["ModelModel", "ModelModelOutput", "ModelPreTrainedModel", "Model_input", "Modeled_values", "Modelifier", "None", "Tensor", "attentions", "bool", "class", "config", "def", "do_mask_input", "else", "encoder", "encoder_output", "for", "forward", "future_values", "hidden_states", "if", "in", "is", "kwargs", "last_hidden_state", "loc", "mask", "masked_values", "masking", "not", "ones_like", "output_attentions", "output_hidden_states", "outputs", "past_observed_mask", "past_values", "r", "return", "return_dict", "scale", "scaled_past_values", "scaler", "self", "torch", "tuple", "use_return_dict", "v"], "patchtst/modeling_patchtst.py:PatchTSTMaskPretrainHead.__init__": ["Dropout", "Identity", "Linear", "ModelConfig", "ModelMaskPretrainHead", "Model_length", "Module", "__init__", "class", "config", "d_model", "def", "dropout", "else", "head_dropout", "if", "linear", "nn", "self", "super", "use_cls_token"], "patchtst/modeling_patchtst.py:PatchTSTMaskPretrainHead.forward": ["ModelMaskPretrainHead", "Module", "Tensor", "class", "def", "dropout", "embedding", "forward", "if", "linear", "nn", "return", "self", "torch", "use_cls_token"], "patchtst/modeling_patchtst.py:PatchTSTForPretraining.__init__": ["ModelConfig", "ModelForPretraining", "ModelMaskPretrainHead", "ModelModel", "ModelPreTrainedModel", "True", "__init__", "class", "config", "def", "do_mask_input", "head", "model", "post_init", "self", "super"], "patchtst/modeling_patchtst.py:PatchTSTForPretraining.forward": ["MSELoss", "ModelForPretraining", "ModelForPretrainingOutput", "ModelPreTrainedModel", "Model_input", "None", "Tensor", "True", "attentions", "bool", "class", "config", "def", "dim", "else", "encoder_states", "forward", "head", "hidden_states", "if", "is", "kwargs", "last_hidden_state", "loss", "loss_val", "mask", "masked_loss", "mean", "model", "model_output", "nn", "none", "not", "output_attentions", "output_hidden_states", "outputs", "past_observed_mask", "past_values", "prediction_output", "r", "reduction", "return", "return_dict", "self", "sum", "torch", "tuple", "use_return_dict", "x_hat"], "patchtst/modeling_patchtst.py:PatchTSTClassificationHead.__init__": ["Dropout", "Flatten", "Identity", "Linear", "ModelClassificationHead", "ModelConfig", "Module", "__init__", "class", "config", "d_model", "def", "dropout", "else", "flatten", "head_dropout", "if", "linear", "nn", "num_input_channels", "num_targets", "pooling_type", "self", "start_dim", "super", "use_cls_token"], "patchtst/modeling_patchtst.py:PatchTSTClassificationHead.forward": ["ModelClassificationHead", "Module", "Tensor", "ValueError", "class", "def", "dim", "dropout", "elif", "else", "embedding", "f", "flatten", "forward", "if", "implemented", "is", "linear", "max", "mean", "nn", "not", "operator", "output", "pooled_embedding", "pooling", "pooling_type", "raise", "return", "self", "torch", "use_cls_token", "values", "yet"], "patchtst/modeling_patchtst.py:PatchTSTForClassification.__init__": ["False", "ModelClassificationHead", "ModelConfig", "ModelForClassification", "ModelModel", "ModelPreTrainedModel", "Setting", "__init__", "class", "config", "def", "do_mask_input", "head", "if", "logger", "model", "parameter", "post_init", "self", "super", "to", "warning"], "patchtst/modeling_patchtst.py:PatchTSTForClassification.forward": ["CrossEntropyLoss", "ModelForClassification", "ModelForClassificationOutput", "ModelPreTrainedModel", "None", "Tensor", "True", "attentions", "bool", "class", "config", "def", "else", "forward", "head", "hidden_states", "if", "is", "kwargs", "last_hidden_state", "loss", "loss_val", "model", "model_output", "nn", "not", "output_attentions", "output_hidden_states", "outputs", "past_observed_mask", "past_values", "prediction_logits", "r", "return", "return_dict", "self", "target_values", "torch", "tuple", "use_return_dict", "y_hat"], "patchtst/modeling_patchtst.py:PatchTSTPredictionHead.__init__": ["Dropout", "Flatten", "Identity", "Linear", "ModelConfig", "ModelPredictionHead", "Module", "ModuleList", "None", "__init__", "append", "class", "config", "d_model", "def", "distribution_output", "dropout", "dropouts", "else", "flatten", "flattens", "for", "get_parameter_projection", "head_dim", "head_dropout", "i", "if", "in", "int", "is", "nn", "not", "num_Modeles", "num_input_channels", "or", "pooling_type", "prediction_length", "projection", "projections", "r", "range", "self", "share_projection", "start_dim", "super", "use_cls_token"], "patchtst/modeling_patchtst.py:PatchTSTPredictionHead.forward": ["ModelPredictionHead", "Module", "Tensor", "append", "class", "def", "dim", "dropout", "dropouts", "elif", "else", "embedding", "flatten", "flattens", "for", "forward", "i", "if", "in", "isinstance", "max", "mean", "nn", "not", "num_input_channels", "output", "pooled_embedding", "pooling_type", "projection", "projections", "range", "return", "self", "share_projection", "stack", "torch", "transpose", "tuple", "use_cls_token", "values", "z"], "patchtst/modeling_patchtst.py:PatchTSTForPrediction.__init__": ["False", "ModelConfig", "ModelForPrediction", "ModelModel", "ModelPreTrainedModel", "ModelPredictionHead", "Modelifier", "NegativeBinomialOutput", "None", "NormalOutput", "Setting", "StudentTOutput", "Unknown", "ValueError", "__init__", "class", "config", "def", "dim", "distribution", "distribution_output", "do_mask_input", "elif", "else", "f", "head", "if", "logger", "loss", "model", "mse", "negative_binomial", "normal", "num_Modeles", "output", "parameter", "post_init", "prediction_length", "raise", "self", "student_t", "super", "to", "warning"], "patchtst/modeling_patchtst.py:PatchTSTForPrediction.forward": ["MSELoss", "ModelForPrediction", "ModelForPredictionOutput", "ModelPreTrainedModel", "None", "Tensor", "True", "attentions", "bool", "class", "config", "def", "distribution", "distribution_output", "else", "forward", "future_values", "head", "hidden_states", "if", "is", "kwargs", "last_hidden_state", "loc", "loss", "loss_val", "mean", "model", "model_output", "nll", "nn", "not", "output_attentions", "output_hidden_states", "outputs", "past_observed_mask", "past_values", "prediction_outputs", "r", "reduction", "return", "return_dict", "scale", "self", "torch", "tuple", "use_return_dict", "weighted_average", "y_hat", "y_hat_out"], "patchtst/modeling_patchtst.py:PatchTSTForPrediction.generate": ["False", "ModelForPrediction", "ModelPreTrainedModel", "None", "SampleModelOutput", "Tensor", "_", "class", "config", "def", "dim", "distribution", "distribution_output", "else", "for", "future_values", "generate", "if", "in", "loc", "num_parallel_samples", "output_hidden_states", "outputs", "past_observed_mask", "past_values", "prediction_outputs", "range", "return", "sample", "samples", "scale", "self", "sequences", "stack", "torch", "unsqueeze"], "patchtst/modeling_patchtst.py:PatchTSTRegressionHead.__init__": ["Dropout", "Flatten", "Identity", "Linear", "ModelConfig", "ModelRegressionHead", "Module", "None", "__init__", "class", "config", "d_model", "def", "distribution_output", "dropout", "else", "flatten", "get_parameter_projection", "head_dim", "head_dropout", "if", "is", "nn", "num_input_channels", "num_targets", "output_range", "pooling_type", "projection", "self", "start_dim", "super", "use_cls_token", "y_range"], "patchtst/modeling_patchtst.py:PatchTSTRegressionHead.forward": ["ModelRegressionHead", "Module", "None", "Tensor", "ValueError", "class", "def", "dim", "distribution_output", "dropout", "elif", "else", "embedding", "f", "flatten", "forward", "if", "implemented", "is", "max", "mean", "nn", "not", "operator", "output", "pooled_embedding", "pooling", "pooling_type", "projection", "raise", "return", "self", "sigmoid", "torch", "use_cls_token", "values", "y_range", "yet"], "patchtst/modeling_patchtst.py:PatchTSTForRegression.__init__": ["False", "ModelConfig", "ModelForRegression", "ModelModel", "ModelPreTrainedModel", "ModelRegressionHead", "NegativeBinomialOutput", "None", "NormalOutput", "Setting", "StudentTOutput", "Unknown", "ValueError", "__init__", "class", "config", "def", "dim", "distribution", "distribution_output", "do_mask_input", "elif", "else", "f", "head", "if", "logger", "loss", "model", "mse", "negative_binomial", "normal", "num_targets", "output", "parameter", "post_init", "raise", "self", "student_t", "super", "to", "warning"], "patchtst/modeling_patchtst.py:PatchTSTForRegression.forward": ["MSELoss", "ModelForRegression", "ModelForRegressionOutput", "ModelPreTrainedModel", "None", "Tensor", "True", "attentions", "bool", "class", "config", "def", "distribution", "distribution_output", "else", "for", "forward", "head", "hidden_states", "if", "in", "is", "item", "kwargs", "last_hidden_state", "loss", "mean", "model", "model_output", "nll", "nn", "not", "num_targets", "output_attentions", "output_hidden_states", "outputs", "past_observed_mask", "past_values", "r", "reduction", "regression_outputs", "return", "return_dict", "self", "target_values", "torch", "tuple", "use_return_dict", "view", "weighted_average", "y_hat"], "patchtst/modeling_patchtst.py:PatchTSTForRegression.generate": ["False", "ModelForRegression", "ModelPreTrainedModel", "None", "SampleModelOutput", "Tensor", "_", "class", "config", "def", "dim", "distribution", "distribution_output", "for", "generate", "in", "num_parallel_samples", "num_targets", "output_hidden_states", "outputs", "past_observed_mask", "past_values", "range", "regression_outputs", "return", "sample", "samples", "self", "sequences", "stack", "target_values", "torch", "view"], "pe_audio/modeling_pe_audio.py:Snake1d.__init__": ["Model", "Module", "Parameter", "__init__", "alpha", "class", "def", "hidden_dim", "nn", "ones", "self", "super", "torch"], "pe_audio/modeling_pe_audio.py:Snake1d.forward": ["Model", "Module", "alpha", "class", "def", "forward", "hidden_states", "nn", "pow", "reciprocal", "reshape", "return", "self", "shape", "sin", "torch"], "pe_audio/modeling_pe_audio.py:PeAudioDacResidualUnit.__init__": ["Conv1d", "ModelDacResidualUnit", "Module", "Snake1d", "__init__", "class", "conv1", "conv2", "def", "dilation", "dimension", "int", "kernel_size", "nn", "pad", "padding", "self", "snake1", "snake2", "super"], "pe_audio/modeling_pe_audio.py:PeAudioDacResidualUnit.forward": ["ModelDacResidualUnit", "Module", "class", "conv1", "conv2", "def", "forward", "hidden_state", "if", "nn", "output_tensor", "padding", "return", "self", "shape", "snake1", "snake2"], "pe_audio/modeling_pe_audio.py:PeAudioDacEncoderBlock.__init__": ["Conv1d", "ModelDacEncoderBlock", "ModelDacResidualUnit", "Module", "PreTrainedConfig", "Snake1d", "__init__", "ceil", "class", "config", "conv1", "def", "dilation", "dimension", "encoder_hidden_size", "int", "kernel_size", "math", "nn", "padding", "res_unit1", "res_unit2", "res_unit3", "self", "snake1", "stride", "stride_index", "super"], "pe_audio/modeling_pe_audio.py:PeAudioDacEncoderBlock.forward": ["ModelDacEncoderBlock", "Module", "class", "conv1", "def", "forward", "hidden_state", "nn", "res_unit1", "res_unit2", "res_unit3", "return", "self", "snake1"], "pe_audio/modeling_pe_audio.py:PeAudioDacEncoder.__init__": ["Conv1d", "ModelDacEncoder", "ModelDacEncoderBlock", "Module", "ModuleList", "PreTrainedConfig", "Snake1d", "__init__", "block", "class", "config", "conv1", "conv2", "d_model", "def", "downsampling_ratios", "encoder_hidden_size", "enumerate", "for", "hidden_size", "in", "kernel_size", "nn", "padding", "self", "snake1", "stride", "stride_index", "strides", "super"], "pe_audio/modeling_pe_audio.py:PeAudioDacEncoder.forward": ["ModelDacEncoder", "Module", "block", "class", "conv1", "conv2", "def", "for", "forward", "hidden_state", "in", "module", "nn", "return", "self", "snake1"], "pe_audio/modeling_pe_audio.py:PeAudioEncoderEmbedder.__init__": ["Conv1d", "Linear", "ModelDacEncoder", "ModelEncoderConfig", "ModelEncoderEmbedder", "Module", "__init__", "bottleneck", "class", "codebook_dim", "config", "dac_config", "dac_encoder", "data_proj", "def", "hidden_size", "nn", "self", "super"], "pe_audio/modeling_pe_audio.py:PeAudioEncoderEmbedder.forward": ["False", "ModelEncoderEmbedder", "Module", "None", "Tensor", "backends", "bottleneck", "class", "codec_features", "config", "cudnn", "dac_config", "dac_encoder", "data_proj", "def", "enabled", "flags", "forward", "hidden_states", "hop_length", "if", "input_values", "inputs_embeds", "is", "nn", "no_grad", "not", "padding_mask", "return", "self", "torch", "transpose", "tuple", "with"], "pe_audio/modeling_pe_audio.py:PeAudioContrastiveHead.__init__": ["False", "LayerNorm", "Linear", "ModelContrastiveHead", "Module", "None", "__init__", "bias", "class", "def", "eps", "in_dim", "int", "layer_norm", "nn", "normalized_shape", "out_dim", "proj", "self", "super"], "pe_audio/modeling_pe_audio.py:PeAudioContrastiveHead.forward": ["FloatTensor", "ModelContrastiveHead", "Module", "Tensor", "class", "def", "forward", "layer_norm", "nn", "proj", "return", "self", "torch", "x"], "pe_audio/modeling_pe_audio.py:PeAudioMaskedGroupNorm.forward": ["False", "GroupNorm", "ModelMaskedGroupNorm", "None", "True", "affine", "batch_size", "bias", "bool", "class", "def", "dim", "eps", "forward", "group_size", "grouped_shape", "hidden_size", "if", "is", "keepdim", "mask", "masked", "mean", "nn", "num_groups", "padding_mask", "padding_mask_grouped", "reshape", "return", "self", "seq_len", "shape", "sqrt", "super", "torch", "unbiased", "var", "view", "weight", "x", "x_grouped", "x_norm"], "pe_audio/modeling_pe_audio.py:PeAudioConvBlock1d.__init__": ["Conv1d", "ModelConvBlock1d", "ModelMaskedGroupNorm", "Module", "SiLU", "__init__", "activation", "class", "config", "def", "groupnorm", "hidden_size", "in_channels", "kernel_size", "nn", "num_channels", "num_groups", "out_channels", "padding", "project", "same", "self", "super"], "pe_audio/modeling_pe_audio.py:PeAudioConvBlock1d.forward": ["ModelConvBlock1d", "Module", "None", "activation", "class", "def", "forward", "groupnorm", "nn", "padding_mask", "project", "return", "self", "x"], "pe_audio/modeling_pe_audio.py:PeAudioResnetBlock1d.__init__": ["ModelConvBlock1d", "ModelResnetBlock1d", "Module", "__init__", "block1", "block2", "class", "config", "def", "nn", "self", "super"], "pe_audio/modeling_pe_audio.py:PeAudioResnetBlock1d.forward": ["ModelResnetBlock1d", "Module", "None", "block1", "block2", "class", "def", "expand_as", "forward", "hidden_states", "if", "is", "nn", "not", "padding_mask", "residual", "return", "self", "transpose", "unsqueeze"], "pe_audio/modeling_pe_audio.py:PeAudioEncoderPatchEmbedder.__init__": ["ModelEncoderPatchEmbedder", "ModelResnetBlock1d", "Module", "Parameter", "__init__", "class", "class_embedding", "config", "def", "hidden_size", "nn", "randn", "resnet_block", "self", "super", "torch"], "pe_audio/modeling_pe_audio.py:PeAudioEncoderPatchEmbedder.forward": ["ModelEncoderPatchEmbedder", "Module", "None", "cat", "class", "class_embedding", "def", "dim", "expand", "forward", "hidden_states", "if", "inputs_embeds", "is", "nn", "not", "padding_mask", "resnet_block", "return", "self", "size", "torch"], "pe_audio/modeling_pe_audio.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "pe_audio/modeling_pe_audio.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value", "value_states"], "pe_audio/modeling_pe_audio.py:stack_freqs": ["Model", "Model_freqs", "Tensor", "cos", "def", "dim", "freqs_cis", "narrow", "return", "sin", "size", "torch", "view"], "pe_audio/modeling_pe_audio.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cos", "def", "flatten", "freqs_cis", "k", "k_", "q", "q_", "reshape", "return", "shape", "sin", "stack_freqs", "sum", "unsqueeze", "unsqueeze_dim"], "pe_audio/modeling_pe_audio.py:PeAudioEncoderRMSNorm.__init__": ["ModelEncoderRMSNorm", "Module", "None", "Parameter", "__init__", "class", "def", "eps", "float", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "pe_audio/modeling_pe_audio.py:PeAudioEncoderRMSNorm.forward": ["ModelEncoderRMSNorm", "Module", "Tensor", "True", "class", "def", "dtype", "float32", "forward", "hidden_states", "input_dtype", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "pe_audio/modeling_pe_audio.py:PeAudioEncoderRMSNorm.extra_repr": ["ModelEncoderRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "variance_epsilon", "weight"], "pe_audio/modeling_pe_audio.py:PeAudioEncoderAttention.__init__": ["False", "Linear", "ModelEncoderAttention", "ModelEncoderRMSNorm", "Module", "None", "__init__", "attention_bias", "attention_dropout", "bias", "class", "config", "def", "else", "eps", "getattr", "hasattr", "head_dim", "hidden_size", "if", "is_causal", "k_norm", "k_proj", "layer_idx", "layer_type", "layer_types", "nn", "num_attention_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "q_norm", "q_proj", "rms_norm_eps", "scaling", "self", "super", "v_proj"], "pe_audio/modeling_pe_audio.py:PeAudioEncoderAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "ModelEncoderAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_attn_implementation", "apply_rotary_pos_emb", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "class", "config", "contiguous", "cos", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "head_dim", "hidden_shape", "hidden_states", "if", "input_shape", "k_norm", "k_proj", "key_states", "kwargs", "nn", "not", "o_proj", "position_embeddings", "q_norm", "q_proj", "query_states", "reshape", "return", "scaling", "self", "shape", "sin", "torch", "training", "transpose", "tuple", "v_proj", "value_states", "view"], "pe_audio/modeling_pe_audio.py:PeAudioEncoderMLP.__init__": ["ACT2FN", "False", "Linear", "ModelEncoderMLP", "Module", "__init__", "act_fn", "bias", "class", "config", "def", "down_proj", "gate_proj", "hidden_act", "hidden_size", "intermediate_size", "nn", "self", "super", "up_proj"], "pe_audio/modeling_pe_audio.py:PeAudioEncoderMLP.forward": ["ModelEncoderMLP", "Module", "act_fn", "class", "def", "down_proj", "forward", "gate_proj", "nn", "return", "self", "up_proj", "x"], "pe_audio/modeling_pe_audio.py:PeAudioEncoderLayer.__init__": ["GradientCheckpointingLayer", "ModelEncoderAttention", "ModelEncoderLayer", "ModelEncoderMLP", "ModelEncoderRMSNorm", "__init__", "class", "config", "def", "eps", "hidden_size", "input_layernorm", "layer_idx", "mlp", "post_attention_layernorm", "rms_norm_eps", "self", "self_attn", "super"], "pe_audio/modeling_pe_audio.py:PeAudioEncoderLayer.forward": ["Cache", "False", "GradientCheckpointingLayer", "LongTensor", "ModelEncoderLayer", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "attention_mask", "bool", "cache_position", "class", "def", "forward", "hidden_states", "input_layernorm", "kwargs", "mlp", "past_key_values", "position_embeddings", "position_ids", "post_attention_layernorm", "residual", "return", "self", "self_attn", "torch", "tuple", "use_cache"], "pe_audio/modeling_pe_audio.py:PeAudioPreTrainedModel._init_weights": ["Conv1d", "ConvTranspose1d", "Embedding", "ModelEncoderPatchEmbedder", "ModelPreTrainedModel", "PreTrainedModel", "Snake1d", "_init_weights", "alpha", "bias", "class", "class_embedding", "config", "constant_", "def", "elif", "else", "embed_dim", "get_text_config", "getattr", "hasattr", "if", "init", "initializer_range", "isinstance", "mean", "module", "nn", "normal_", "ones_", "reset_parameters", "self", "shape", "std", "super", "trunc_normal_", "weight"], "pe_audio/modeling_pe_audio.py:PeAudioEncoderRotaryEmbedding.__init__": ["Callable", "False", "ModelEncoderConfig", "ModelEncoderRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "pe_audio/modeling_pe_audio.py:PeAudioEncoderRotaryEmbedding.compute_default_rope_parameters": ["ModelEncoderConfig", "ModelEncoderRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "pe_audio/modeling_pe_audio.py:PeAudioEncoderRotaryEmbedding.forward": ["False", "ModelEncoderRotaryEmbedding", "Module", "None", "and", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "pe_audio/modeling_pe_audio.py:PeAudioEncoder.__init__": ["False", "Linear", "ModelEncoder", "ModelEncoderConfig", "ModelEncoderEmbedder", "ModelEncoderLayer", "ModelEncoderPatchEmbedder", "ModelEncoderRMSNorm", "ModelEncoderRotaryEmbedding", "ModelPreTrainedModel", "ModuleList", "__init__", "bias", "class", "config", "def", "embedder", "eps", "for", "gradient_checkpointing", "hidden_size", "in", "layer_idx", "layers", "nn", "norm", "num_hidden_layers", "output", "patch_embedder", "post_init", "range", "rms_norm_eps", "rotary_emb", "self", "super"], "pe_audio/modeling_pe_audio.py:PeAudioEncoder.forward": ["BaseModelOutputWithPooling", "ModelEncoder", "ModelEncoderOutput", "ModelPreTrainedModel", "None", "Tensor", "_prepare_4d_attention_mask", "arange", "attention_mask", "class", "config", "def", "device", "dtype", "embedder", "encoder_layer", "for", "forward", "hidden_states", "if", "in", "input_values", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layers", "norm", "not", "num_hidden_layers", "output", "output_mask", "padding_mask", "patch_embedder", "pooler_output", "position_embeddings", "position_ids", "return", "rotary_emb", "self", "shape", "torch", "unsqueeze"], "pe_audio/modeling_pe_audio.py:PeAudioOutput.to_tuple": ["Any", "ModelOutput", "audio_outputs", "class", "def", "else", "for", "getattr", "if", "in", "k", "keys", "not", "return", "self", "text_outputs", "to_tuple", "tuple"], "pe_audio/modeling_pe_audio.py:PeAudioModel.__init__": ["AutoModel", "ModelConfig", "ModelContrastiveHead", "ModelEncoder", "ModelModel", "ModelPreTrainedModel", "Parameter", "__init__", "audio_config", "audio_encoder", "audio_head", "class", "config", "def", "from_config", "hidden_size", "nn", "post_init", "self", "super", "text_audio_head", "text_audio_logit_bias", "text_audio_logit_scale", "text_config", "text_model", "torch", "zeros"], "pe_audio/modeling_pe_audio.py:PeAudioModel.get_text_audio_embeds": ["MaskedLMOutput", "ModelModel", "ModelPreTrainedModel", "None", "True", "attention_mask", "class", "def", "get_text_audio_embeds", "hidden_states", "input_ids", "return", "return_dict", "self", "text_audio_embeds", "text_audio_head", "text_model", "text_outputs"], "pe_audio/modeling_pe_audio.py:PeAudioModel.get_audio_embeds": ["BaseModelOutputWithPooling", "ModelModel", "ModelPreTrainedModel", "None", "True", "audio_embeds", "audio_encoder", "audio_head", "audio_outputs", "class", "def", "get_audio_embeds", "input_values", "padding_mask", "pooler_output", "return", "return_dict", "self"], "pe_audio/modeling_pe_audio.py:PeAudioModel.forward": ["BaseModelOutputWithPooling", "F", "MaskedLMOutput", "ModelModel", "ModelOutput", "ModelPreTrainedModel", "None", "T", "Tensor", "True", "attention_mask", "audio_embeds", "audio_encoder", "audio_head", "audio_outputs", "bool", "class", "def", "device", "eye", "forward", "hidden_states", "if", "input_ids", "input_values", "kwargs", "labels", "logits_audio_text", "logsigmoid", "loss", "output_hidden_states", "padding_mask", "pooler_output", "return", "return_loss", "self", "shape", "sum", "text_audio_embeds", "text_audio_head", "text_audio_logit_bias", "text_audio_logit_scale", "text_model", "text_outputs", "torch"], "pe_audio/modeling_pe_audio.py:PeAudioFrameLevelModel.get_audio_embeds": ["BaseModelOutputWithPooling", "ModelFrameLevelModel", "ModelModel", "None", "True", "audio_embeds", "audio_encoder", "audio_head", "audio_outputs", "class", "def", "get_audio_embeds", "input_values", "last_hidden_state", "padding_mask", "return", "return_dict", "self"], "pe_audio/modeling_pe_audio.py:PeAudioFrameLevelModel.forward": ["BaseModelOutputWithPooling", "F", "MaskedLMOutput", "ModelFrameLevelModel", "ModelModel", "ModelOutput", "None", "T", "Tensor", "True", "attention_mask", "audio_embeds", "audio_encoder", "audio_head", "audio_outputs", "bool", "class", "def", "device", "eye", "forward", "hidden_states", "if", "input_ids", "input_values", "kwargs", "labels", "last_hidden_state", "logits_audio_text", "logsigmoid", "loss", "output_hidden_states", "padding_mask", "return", "return_loss", "self", "shape", "sum", "text_audio_embeds", "text_audio_head", "text_audio_logit_bias", "text_audio_logit_scale", "text_model", "text_outputs", "torch", "transpose"], "pe_audio_video/modeling_pe_audio_video.py:PeAudioVideoMaskedGroupNorm.forward": ["False", "GroupNorm", "ModelMaskedGroupNorm", "None", "True", "affine", "batch_size", "bias", "bool", "class", "def", "dim", "eps", "forward", "group_size", "grouped_shape", "hidden_size", "if", "is", "keepdim", "mask", "masked", "mean", "nn", "num_groups", "padding_mask", "padding_mask_grouped", "reshape", "return", "self", "seq_len", "shape", "sqrt", "super", "torch", "unbiased", "var", "view", "weight", "x", "x_grouped", "x_norm"], "pe_audio_video/modeling_pe_audio_video.py:PeAudioVideoConvBlock1d.__init__": ["Conv1d", "ModelConvBlock1d", "ModelMaskedGroupNorm", "Module", "SiLU", "__init__", "activation", "class", "config", "def", "groupnorm", "hidden_size", "in_channels", "kernel_size", "nn", "num_channels", "num_groups", "out_channels", "padding", "project", "same", "self", "super"], "pe_audio_video/modeling_pe_audio_video.py:PeAudioVideoConvBlock1d.forward": ["ModelConvBlock1d", "Module", "None", "activation", "class", "def", "forward", "groupnorm", "nn", "padding_mask", "project", "return", "self", "x"], "pe_audio_video/modeling_pe_audio_video.py:PeAudioVideoResnetBlock1d.__init__": ["ModelConvBlock1d", "ModelResnetBlock1d", "Module", "__init__", "block1", "block2", "class", "config", "def", "nn", "self", "super"], "pe_audio_video/modeling_pe_audio_video.py:PeAudioVideoResnetBlock1d.forward": ["ModelResnetBlock1d", "Module", "None", "block1", "block2", "class", "def", "expand_as", "forward", "hidden_states", "if", "is", "nn", "not", "padding_mask", "residual", "return", "self", "transpose", "unsqueeze"], "pe_audio_video/modeling_pe_audio_video.py:PeAudioVideoEncoderPatchEmbedder.__init__": ["ModelEncoderPatchEmbedder", "ModelResnetBlock1d", "Module", "Parameter", "__init__", "class", "class_embedding", "config", "def", "hidden_size", "nn", "randn", "resnet_block", "self", "super", "torch"], "pe_audio_video/modeling_pe_audio_video.py:PeAudioVideoEncoderPatchEmbedder.forward": ["ModelEncoderPatchEmbedder", "Module", "None", "cat", "class", "class_embedding", "def", "dim", "expand", "forward", "hidden_states", "if", "inputs_embeds", "is", "nn", "not", "padding_mask", "resnet_block", "return", "self", "size", "torch"], "pe_audio_video/modeling_pe_audio_video.py:PeAudioVideoContrastiveHead.__init__": ["False", "LayerNorm", "Linear", "ModelContrastiveHead", "Module", "None", "__init__", "bias", "class", "def", "eps", "in_dim", "int", "layer_norm", "nn", "normalized_shape", "out_dim", "proj", "self", "super"], "pe_audio_video/modeling_pe_audio_video.py:PeAudioVideoContrastiveHead.forward": ["FloatTensor", "ModelContrastiveHead", "Module", "Tensor", "class", "def", "forward", "layer_norm", "nn", "proj", "return", "self", "torch", "x"], "pe_audio_video/modeling_pe_audio_video.py:PeAudioVideoEncoderEmbedder.__init__": ["AutoModel", "Conv1d", "LayerNorm", "Linear", "ModelEncoderConfig", "ModelEncoderEmbedder", "Module", "__init__", "audio_config", "audio_encoder", "class", "concat_modality_proj", "config", "data_proj", "def", "from_config", "hidden_size", "nn", "self", "super", "video_config", "video_encoder", "video_norm", "video_proj"], "pe_audio_video/modeling_pe_audio_video.py:PeAudioVideoEncoderEmbedder._align_video_hidden_state": ["ModelEncoderEmbedder", "Module", "None", "Tensor", "_align_video_hidden_state", "aligned_hidden_state", "aligned_shape", "all", "and", "audio_hidden_state", "audio_length", "audio_lengths", "class", "def", "dim", "dtype", "else", "enumerate", "for", "functional", "hidden_state", "i", "if", "in", "interpolate", "interpolated_hidden_state", "is", "long", "mode", "nearest", "new_ones", "new_zeros", "nn", "not", "numel", "or", "padding_mask", "padding_mask_videos", "return", "self", "shape", "size", "sum", "torch", "transpose", "video_hidden_state", "video_length", "video_lengths", "zip"], "pe_audio_video/modeling_pe_audio_video.py:PeAudioVideoEncoderEmbedder.forward": ["ModelEncoderEmbedder", "Module", "None", "Tensor", "_align_video_hidden_state", "audio_encoder", "audio_hidden_state", "audio_output", "cat", "class", "concat_modality_proj", "data_proj", "def", "dim", "forward", "input_values", "inputs_embeds", "last_hidden_state", "nn", "output_mask", "padding_mask", "padding_mask_videos", "pixel_values_videos", "return", "self", "torch", "transpose", "video_encoder", "video_hidden_state", "video_norm", "video_output", "video_proj"], "pe_audio_video/modeling_pe_audio_video.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "pe_audio_video/modeling_pe_audio_video.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value", "value_states"], "pe_audio_video/modeling_pe_audio_video.py:stack_freqs": ["Model", "Model_freqs", "Tensor", "cos", "def", "dim", "freqs_cis", "narrow", "return", "sin", "size", "torch", "view"], "pe_audio_video/modeling_pe_audio_video.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cos", "def", "flatten", "freqs_cis", "k", "k_", "q", "q_", "reshape", "return", "shape", "sin", "stack_freqs", "sum", "unsqueeze", "unsqueeze_dim"], "pe_audio_video/modeling_pe_audio_video.py:PeAudioVideoEncoderAttention.__init__": ["False", "Linear", "ModelEncoderAttention", "ModelEncoderRMSNorm", "Module", "None", "__init__", "attention_bias", "attention_dropout", "bias", "class", "config", "def", "else", "eps", "getattr", "hasattr", "head_dim", "hidden_size", "if", "is_causal", "k_norm", "k_proj", "layer_idx", "layer_type", "layer_types", "nn", "num_attention_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "q_norm", "q_proj", "rms_norm_eps", "scaling", "self", "super", "v_proj"], "pe_audio_video/modeling_pe_audio_video.py:PeAudioVideoEncoderAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "ModelEncoderAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_attn_implementation", "apply_rotary_pos_emb", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "class", "config", "contiguous", "cos", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "head_dim", "hidden_shape", "hidden_states", "if", "input_shape", "k_norm", "k_proj", "key_states", "kwargs", "nn", "not", "o_proj", "position_embeddings", "q_norm", "q_proj", "query_states", "reshape", "return", "scaling", "self", "shape", "sin", "torch", "training", "transpose", "tuple", "v_proj", "value_states", "view"], "pe_audio_video/modeling_pe_audio_video.py:PeAudioVideoEncoderMLP.__init__": ["ACT2FN", "False", "Linear", "ModelEncoderMLP", "Module", "__init__", "act_fn", "bias", "class", "config", "def", "down_proj", "gate_proj", "hidden_act", "hidden_size", "intermediate_size", "nn", "self", "super", "up_proj"], "pe_audio_video/modeling_pe_audio_video.py:PeAudioVideoEncoderMLP.forward": ["ModelEncoderMLP", "Module", "act_fn", "class", "def", "down_proj", "forward", "gate_proj", "nn", "return", "self", "up_proj", "x"], "pe_audio_video/modeling_pe_audio_video.py:PeAudioVideoEncoderLayer.__init__": ["GradientCheckpointingLayer", "ModelEncoderAttention", "ModelEncoderLayer", "ModelEncoderMLP", "ModelEncoderRMSNorm", "__init__", "class", "config", "def", "eps", "hidden_size", "input_layernorm", "layer_idx", "mlp", "post_attention_layernorm", "rms_norm_eps", "self", "self_attn", "super"], "pe_audio_video/modeling_pe_audio_video.py:PeAudioVideoEncoderLayer.forward": ["Cache", "False", "GradientCheckpointingLayer", "LongTensor", "ModelEncoderLayer", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "attention_mask", "bool", "cache_position", "class", "def", "forward", "hidden_states", "input_layernorm", "kwargs", "mlp", "past_key_values", "position_embeddings", "position_ids", "post_attention_layernorm", "residual", "return", "self", "self_attn", "torch", "tuple", "use_cache"], "pe_audio_video/modeling_pe_audio_video.py:PeAudioVideoEncoderRMSNorm.__init__": ["ModelEncoderRMSNorm", "Module", "None", "Parameter", "__init__", "class", "def", "eps", "float", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "pe_audio_video/modeling_pe_audio_video.py:PeAudioVideoEncoderRMSNorm.forward": ["ModelEncoderRMSNorm", "Module", "Tensor", "True", "class", "def", "dtype", "float32", "forward", "hidden_states", "input_dtype", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "pe_audio_video/modeling_pe_audio_video.py:PeAudioVideoEncoderRMSNorm.extra_repr": ["ModelEncoderRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "variance_epsilon", "weight"], "pe_audio_video/modeling_pe_audio_video.py:PeAudioVideoEncoderRotaryEmbedding.__init__": ["Callable", "False", "ModelEncoderConfig", "ModelEncoderRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "pe_audio_video/modeling_pe_audio_video.py:PeAudioVideoEncoderRotaryEmbedding.compute_default_rope_parameters": ["ModelEncoderConfig", "ModelEncoderRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "pe_audio_video/modeling_pe_audio_video.py:PeAudioVideoEncoderRotaryEmbedding.forward": ["False", "ModelEncoderRotaryEmbedding", "Module", "None", "and", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "pe_audio_video/modeling_pe_audio_video.py:PeAudioVideoPreTrainedModel._init_weights": ["ModelEncoderPatchEmbedder", "ModelPreTrainedModel", "PreTrainedModel", "_init_weights", "class", "class_embedding", "config", "def", "else", "embed_dim", "get_text_config", "getattr", "hasattr", "if", "init", "initializer_range", "isinstance", "mean", "module", "normal_", "self", "shape", "std", "super"], "pe_audio_video/modeling_pe_audio_video.py:PeAudioVideoEncoder.__init__": ["False", "Linear", "ModelEncoder", "ModelEncoderConfig", "ModelEncoderEmbedder", "ModelEncoderLayer", "ModelEncoderPatchEmbedder", "ModelEncoderRMSNorm", "ModelEncoderRotaryEmbedding", "ModelPreTrainedModel", "ModuleList", "__init__", "bias", "class", "config", "def", "embedder", "eps", "for", "gradient_checkpointing", "hidden_size", "in", "layer_idx", "layers", "nn", "norm", "num_hidden_layers", "output", "patch_embedder", "post_init", "range", "rms_norm_eps", "rotary_emb", "self", "super"], "pe_audio_video/modeling_pe_audio_video.py:PeAudioVideoEncoder.forward": ["ModelEncoder", "ModelEncoderOutput", "ModelPreTrainedModel", "None", "Tensor", "_prepare_4d_attention_mask", "arange", "attention_mask", "audio_model_output", "audio_output", "class", "config", "def", "device", "dtype", "embedder", "encoder_layer", "for", "forward", "hidden_states", "if", "in", "input_values", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layers", "norm", "not", "num_hidden_layers", "output", "padding_mask", "padding_mask_videos", "patch_embedder", "pixel_values_videos", "pooler_output", "position_embeddings", "position_ids", "return", "rotary_emb", "self", "shape", "torch", "unsqueeze", "video_model_output", "video_output"], "pe_audio_video/modeling_pe_audio_video.py:PeAudioVideoOutput.to_tuple": ["Any", "ModelOutput", "class", "def", "else", "endswith", "for", "getattr", "if", "in", "k", "keys", "model_output", "not", "return", "self", "to_tuple", "tuple"], "pe_audio_video/modeling_pe_audio_video.py:PeAudioVideoModel.__init__": ["AutoModel", "ModelConfig", "ModelContrastiveHead", "ModelEncoder", "ModelModel", "ModelPreTrainedModel", "Parameter", "__init__", "audio_config", "audio_hidden_size", "audio_model", "audio_plus_text_head", "audio_plus_text_logit_bias", "audio_plus_text_logit_scale", "audio_video_config", "audio_video_encoder", "audio_video_head", "audio_video_logit_bias", "audio_video_logit_scale", "class", "config", "def", "from_config", "hidden_size", "nn", "post_init", "self", "super", "text_audio_video_head", "text_audio_video_logit_bias", "text_audio_video_logit_scale", "text_config", "text_hidden_size", "text_model", "torch", "video_config", "video_hidden_size", "video_model", "video_plus_text_head", "video_plus_text_logit_bias", "video_plus_text_logit_scale", "zeros"], "pe_audio_video/modeling_pe_audio_video.py:PeAudioVideoModel._contrastive_loss": ["ModelModel", "ModelPreTrainedModel", "Tensor", "_contrastive_loss", "class", "def", "device", "eye", "functional", "labels", "logits", "logsigmoid", "loss", "nn", "return", "self", "shape", "sum", "torch"], "pe_audio_video/modeling_pe_audio_video.py:PeAudioVideoModel.get_text_audio_embeds": ["ModelModel", "ModelPreTrainedModel", "None", "attention_mask", "audio_model", "class", "def", "get_text_audio_embeds", "get_text_embeds", "input_ids", "return", "self"], "pe_audio_video/modeling_pe_audio_video.py:PeAudioVideoModel.get_text_video_embeds": ["ModelModel", "ModelPreTrainedModel", "None", "attention_mask", "class", "def", "get_text_embeds", "get_text_video_embeds", "input_ids", "return", "self", "video_model"], "pe_audio_video/modeling_pe_audio_video.py:PeAudioVideoModel.get_text_audio_video_embeds": ["MaskedLMOutput", "ModelModel", "ModelPreTrainedModel", "None", "True", "attention_mask", "class", "def", "get_text_audio_video_embeds", "hidden_states", "input_ids", "return", "return_dict", "self", "text_audio_video_head", "text_embeds", "text_model", "text_outputs"], "pe_audio_video/modeling_pe_audio_video.py:PeAudioVideoModel.get_audio_embeds": ["ModelModel", "ModelPreTrainedModel", "None", "audio_model", "class", "def", "get_audio_embeds", "input_values", "padding_mask", "return", "self"], "pe_audio_video/modeling_pe_audio_video.py:PeAudioVideoModel.get_video_embeds": ["ModelModel", "ModelPreTrainedModel", "None", "class", "def", "get_video_embeds", "padding_mask_videos", "pixel_values_videos", "return", "self", "video_model"], "pe_audio_video/modeling_pe_audio_video.py:PeAudioVideoModel.get_audio_video_embeds": ["AudioVideoEmbeddings", "False", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "audio_embeds", "audio_head", "audio_model", "audio_model_output", "audio_video_embeds", "audio_video_encoder", "audio_video_head", "audio_video_outputs", "bool", "class", "def", "else", "get_audio_video_embeds", "if", "input_values", "kwargs", "padding_mask", "padding_mask_videos", "pixel_values_videos", "pooler_output", "return", "return_audio_embeds", "return_video_embeds", "self", "torch", "video_embeds", "video_head", "video_model", "video_model_output"], "pe_audio_video/modeling_pe_audio_video.py:PeAudioVideoModel.get_audio_plus_text_embeds": ["MaskedLMOutput", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "True", "attention_mask", "audio_embeds", "audio_encoder", "audio_model", "audio_plus_text_embeds", "audio_plus_text_head", "cat", "class", "def", "dim", "get_audio_plus_text_embeds", "hidden_states", "input_ids", "input_values", "padding_mask", "return", "return_dict", "self", "text_embeds", "text_model", "text_outputs", "torch"], "pe_audio_video/modeling_pe_audio_video.py:PeAudioVideoModel.get_video_plus_text_embeds": ["MaskedLMOutput", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "True", "attention_mask", "cat", "class", "def", "dim", "get_video_plus_text_embeds", "hidden_states", "input_ids", "padding_mask_videos", "pixel_values_videos", "return", "return_dict", "self", "text_embeds", "text_model", "text_outputs", "torch", "video_embeds", "video_encoder", "video_model", "video_plus_text_embeds", "video_plus_text_head"], "pe_audio_video/modeling_pe_audio_video.py:PeAudioVideoModel.forward": ["At", "False", "ModelModel", "ModelOutput", "ModelPreTrainedModel", "None", "T", "Tensor", "True", "ValueError", "_contrastive_loss", "attention_mask", "audio_embeds", "audio_head", "audio_model", "audio_model_output", "audio_outputs", "audio_plus_text_embeds", "audio_plus_text_head", "audio_plus_text_logit_bias", "audio_plus_text_logit_scale", "audio_plus_text_video_loss", "audio_text_loss", "audio_video_embeds", "audio_video_encoder", "audio_video_head", "audio_video_logit_bias", "audio_video_logit_scale", "audio_video_loss", "audio_video_outputs", "audio_video_text_loss", "be", "cat", "class", "def", "dim", "else", "forward", "hidden_states", "if", "input_ids", "input_values", "is", "kwargs", "least", "logits_audio_plus_text_video", "logits_audio_text", "logits_audio_video", "logits_audio_video_text", "logits_video_plus_text_audio", "logits_video_text", "loss", "must", "not", "of", "or", "output_hidden_states", "outputs", "padding_mask", "padding_mask_videos", "pixel_values_videos", "pooler_output", "provided", "raise", "return", "return_dict", "return_loss", "self", "sum", "text_audio_embeds", "text_audio_head", "text_audio_logit_bias", "text_audio_logit_scale", "text_audio_video_embeds", "text_audio_video_head", "text_audio_video_logit_bias", "text_audio_video_logit_scale", "text_embeds", "text_model", "text_outputs", "text_video_embeds", "text_video_head", "text_video_logit_bias", "text_video_logit_scale", "torch", "two", "video_embeds", "video_head", "video_model", "video_model_output", "video_outputs", "video_plus_text_audio_loss", "video_plus_text_embeds", "video_plus_text_head", "video_plus_text_logit_bias", "video_plus_text_logit_scale", "video_text_loss"], "pe_video/modeling_pe_video.py:PeVideoOutput.to_tuple": ["Any", "ModelOutput", "class", "def", "else", "for", "getattr", "if", "in", "k", "keys", "not", "return", "self", "text_outputs", "to_tuple", "tuple", "video_outputs"], "pe_video/modeling_pe_video.py:PeVideoContrastiveHead.__init__": ["False", "LayerNorm", "Linear", "ModelContrastiveHead", "Module", "None", "__init__", "bias", "class", "def", "eps", "in_dim", "int", "layer_norm", "nn", "normalized_shape", "out_dim", "proj", "self", "super"], "pe_video/modeling_pe_video.py:PeVideoContrastiveHead.forward": ["FloatTensor", "ModelContrastiveHead", "Module", "Tensor", "class", "def", "forward", "layer_norm", "nn", "proj", "return", "self", "torch", "x"], "pe_video/modeling_pe_video.py:PeVideoMaskedGroupNorm.forward": ["False", "GroupNorm", "ModelMaskedGroupNorm", "None", "True", "affine", "batch_size", "bias", "bool", "class", "def", "dim", "eps", "forward", "group_size", "grouped_shape", "hidden_size", "if", "is", "keepdim", "mask", "masked", "mean", "nn", "num_groups", "padding_mask", "padding_mask_grouped", "reshape", "return", "self", "seq_len", "shape", "sqrt", "super", "torch", "unbiased", "var", "view", "weight", "x", "x_grouped", "x_norm"], "pe_video/modeling_pe_video.py:PeVideoConvBlock1d.__init__": ["Conv1d", "ModelConvBlock1d", "ModelMaskedGroupNorm", "Module", "SiLU", "__init__", "activation", "class", "config", "def", "groupnorm", "hidden_size", "in_channels", "kernel_size", "nn", "num_channels", "num_groups", "out_channels", "padding", "project", "same", "self", "super"], "pe_video/modeling_pe_video.py:PeVideoConvBlock1d.forward": ["ModelConvBlock1d", "Module", "None", "activation", "class", "def", "forward", "groupnorm", "nn", "padding_mask", "project", "return", "self", "x"], "pe_video/modeling_pe_video.py:PeVideoResnetBlock1d.__init__": ["ModelConvBlock1d", "ModelResnetBlock1d", "Module", "__init__", "block1", "block2", "class", "config", "def", "nn", "self", "super"], "pe_video/modeling_pe_video.py:PeVideoResnetBlock1d.forward": ["ModelResnetBlock1d", "Module", "None", "block1", "block2", "class", "def", "expand_as", "forward", "hidden_states", "if", "is", "nn", "not", "padding_mask", "residual", "return", "self", "transpose", "unsqueeze"], "pe_video/modeling_pe_video.py:PeVideoEncoderPatchEmbedder.__init__": ["ModelEncoderPatchEmbedder", "ModelResnetBlock1d", "Module", "Parameter", "__init__", "class", "class_embedding", "config", "def", "hidden_size", "nn", "randn", "resnet_block", "self", "super", "torch"], "pe_video/modeling_pe_video.py:PeVideoEncoderPatchEmbedder.forward": ["ModelEncoderPatchEmbedder", "Module", "None", "cat", "class", "class_embedding", "def", "dim", "expand", "forward", "hidden_states", "if", "inputs_embeds", "is", "nn", "not", "padding_mask", "resnet_block", "return", "self", "size", "torch"], "pe_video/modeling_pe_video.py:PeVideoEncoderEmbedder.__init__": ["AutoModelForImageClassification", "False", "Linear", "ModelEncoderConfig", "ModelEncoderEmbedder", "Module", "__init__", "bias", "class", "config", "data_proj", "def", "from_config", "hidden_size", "nn", "num_labels", "proj", "self", "super", "vision_config", "vision_model"], "pe_video/modeling_pe_video.py:PeVideoEncoderEmbedder.forward": ["F", "ModelEncoderEmbedder", "Module", "None", "Tensor", "class", "data_proj", "def", "dim", "forward", "input_shape", "inputs_embeds", "logits", "nn", "normalize", "padding_mask", "pixel_values_videos", "proj", "return", "self", "shape", "torch", "tuple", "view", "vision_encoder_outputs", "vision_features", "vision_model"], "pe_video/modeling_pe_video.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "pe_video/modeling_pe_video.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value", "value_states"], "pe_video/modeling_pe_video.py:stack_freqs": ["Model", "Model_freqs", "Tensor", "cos", "def", "dim", "freqs_cis", "narrow", "return", "sin", "size", "torch", "view"], "pe_video/modeling_pe_video.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cos", "def", "flatten", "freqs_cis", "k", "k_", "q", "q_", "reshape", "return", "shape", "sin", "stack_freqs", "sum", "unsqueeze", "unsqueeze_dim"], "pe_video/modeling_pe_video.py:PeVideoEncoderRMSNorm.__init__": ["ModelEncoderRMSNorm", "Module", "None", "Parameter", "__init__", "class", "def", "eps", "float", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "pe_video/modeling_pe_video.py:PeVideoEncoderRMSNorm.forward": ["ModelEncoderRMSNorm", "Module", "Tensor", "True", "class", "def", "dtype", "float32", "forward", "hidden_states", "input_dtype", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "pe_video/modeling_pe_video.py:PeVideoEncoderRMSNorm.extra_repr": ["ModelEncoderRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "variance_epsilon", "weight"], "pe_video/modeling_pe_video.py:PeVideoEncoderAttention.__init__": ["False", "Linear", "ModelEncoderAttention", "ModelEncoderRMSNorm", "Module", "None", "__init__", "attention_bias", "attention_dropout", "bias", "class", "config", "def", "else", "eps", "getattr", "hasattr", "head_dim", "hidden_size", "if", "is_causal", "k_norm", "k_proj", "layer_idx", "layer_type", "layer_types", "nn", "num_attention_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "q_norm", "q_proj", "rms_norm_eps", "scaling", "self", "super", "v_proj"], "pe_video/modeling_pe_video.py:PeVideoEncoderAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "ModelEncoderAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_attn_implementation", "apply_rotary_pos_emb", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "class", "config", "contiguous", "cos", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "head_dim", "hidden_shape", "hidden_states", "if", "input_shape", "k_norm", "k_proj", "key_states", "kwargs", "nn", "not", "o_proj", "position_embeddings", "q_norm", "q_proj", "query_states", "reshape", "return", "scaling", "self", "shape", "sin", "torch", "training", "transpose", "tuple", "v_proj", "value_states", "view"], "pe_video/modeling_pe_video.py:PeVideoEncoderMLP.__init__": ["ACT2FN", "False", "Linear", "ModelEncoderMLP", "Module", "__init__", "act_fn", "bias", "class", "config", "def", "down_proj", "gate_proj", "hidden_act", "hidden_size", "intermediate_size", "nn", "self", "super", "up_proj"], "pe_video/modeling_pe_video.py:PeVideoEncoderMLP.forward": ["ModelEncoderMLP", "Module", "act_fn", "class", "def", "down_proj", "forward", "gate_proj", "nn", "return", "self", "up_proj", "x"], "pe_video/modeling_pe_video.py:PeVideoEncoderLayer.__init__": ["GradientCheckpointingLayer", "ModelEncoderAttention", "ModelEncoderLayer", "ModelEncoderMLP", "ModelEncoderRMSNorm", "__init__", "class", "config", "def", "eps", "hidden_size", "input_layernorm", "layer_idx", "mlp", "post_attention_layernorm", "rms_norm_eps", "self", "self_attn", "super"], "pe_video/modeling_pe_video.py:PeVideoEncoderLayer.forward": ["Cache", "False", "GradientCheckpointingLayer", "LongTensor", "ModelEncoderLayer", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "attention_mask", "bool", "cache_position", "class", "def", "forward", "hidden_states", "input_layernorm", "kwargs", "mlp", "past_key_values", "position_embeddings", "position_ids", "post_attention_layernorm", "residual", "return", "self", "self_attn", "torch", "tuple", "use_cache"], "pe_video/modeling_pe_video.py:PeVideoPreTrainedModel._init_weights": ["ModelEncoderPatchEmbedder", "ModelPreTrainedModel", "PreTrainedModel", "_init_weights", "class", "class_embedding", "config", "def", "else", "embed_dim", "get_text_config", "getattr", "hasattr", "if", "init", "initializer_range", "isinstance", "mean", "module", "normal_", "self", "shape", "std", "super"], "pe_video/modeling_pe_video.py:PeVideoEncoderRotaryEmbedding.__init__": ["Callable", "False", "ModelEncoderConfig", "ModelEncoderRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "pe_video/modeling_pe_video.py:PeVideoEncoderRotaryEmbedding.compute_default_rope_parameters": ["ModelEncoderConfig", "ModelEncoderRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "pe_video/modeling_pe_video.py:PeVideoEncoderRotaryEmbedding.forward": ["False", "ModelEncoderRotaryEmbedding", "Module", "None", "and", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "pe_video/modeling_pe_video.py:PeVideoEncoder.__init__": ["False", "Linear", "ModelEncoder", "ModelEncoderConfig", "ModelEncoderEmbedder", "ModelEncoderLayer", "ModelEncoderPatchEmbedder", "ModelEncoderRMSNorm", "ModelEncoderRotaryEmbedding", "ModelPreTrainedModel", "ModuleList", "__init__", "bias", "class", "config", "def", "embedder", "eps", "for", "gradient_checkpointing", "hidden_size", "in", "layer_idx", "layers", "nn", "norm", "num_hidden_layers", "output", "patch_embedder", "post_init", "range", "rms_norm_eps", "rotary_emb", "self", "super"], "pe_video/modeling_pe_video.py:PeVideoEncoder.forward": ["BaseModelOutputWithPooling", "ModelEncoder", "ModelPreTrainedModel", "None", "Tensor", "_prepare_4d_attention_mask", "arange", "attention_mask", "class", "config", "def", "device", "dtype", "embedder", "encoder_layer", "for", "forward", "hidden_states", "if", "in", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layers", "norm", "not", "num_hidden_layers", "output", "padding_mask", "padding_mask_videos", "patch_embedder", "pixel_values_videos", "pooler_output", "position_embeddings", "position_ids", "return", "rotary_emb", "self", "shape", "torch", "unsqueeze"], "pe_video/modeling_pe_video.py:PeVideoModel.__init__": ["AutoModel", "ModelConfig", "ModelContrastiveHead", "ModelEncoder", "ModelModel", "ModelPreTrainedModel", "Parameter", "__init__", "class", "config", "def", "from_config", "hidden_size", "nn", "post_init", "self", "super", "text_config", "text_model", "text_video_head", "text_video_logit_bias", "text_video_logit_scale", "torch", "video_config", "video_encoder", "video_head", "zeros"], "pe_video/modeling_pe_video.py:PeVideoModel.get_text_features": ["MaskedLMOutput", "ModelModel", "ModelPreTrainedModel", "None", "True", "attention_mask", "class", "def", "get_text_features", "input_ids", "last_hidden_state", "return", "return_dict", "self", "text_features", "text_model", "text_outputs", "text_video_head"], "pe_video/modeling_pe_video.py:PeVideoModel.get_video_features": ["BaseModelOutputWithPooling", "ModelModel", "ModelPreTrainedModel", "None", "True", "class", "def", "get_video_features", "padding_mask_videos", "pixel_values_videos", "pooler_output", "return", "return_dict", "self", "video_encoder", "video_features", "video_head", "video_outputs"], "pe_video/modeling_pe_video.py:PeVideoModel.forward": ["BaseModelOutputWithPooling", "F", "MaskedLMOutput", "ModelModel", "ModelOutput", "ModelPreTrainedModel", "None", "T", "Tensor", "True", "attention_mask", "bool", "class", "def", "device", "eye", "forward", "hidden_states", "if", "input_ids", "kwargs", "labels", "logits_video_text", "logsigmoid", "loss", "output_hidden_states", "padding_mask_videos", "pixel_values_videos", "pooler_output", "return", "return_loss", "self", "shape", "sum", "text_model", "text_outputs", "text_video_embeds", "text_video_head", "text_video_logit_bias", "text_video_logit_scale", "torch", "video_embeds", "video_encoder", "video_head", "video_outputs"], "pegasus/modeling_pegasus.py:shift_tokens_right": ["Model_tokens_right", "Modeled_input_ids", "None", "Tensor", "ValueError", "be", "clone", "config", "decoder_start_token_id", "def", "defined", "has", "if", "input_ids", "int", "is", "masked_fill_", "model", "new_zeros", "pad_token_id", "raise", "return", "self", "shape", "to", "torch"], "pegasus/modeling_pegasus.py:PegasusSinusoidalPositionalEmbedding.__init__": ["Embedding", "ModelSinusoidalPositionalEmbedding", "None", "True", "__init__", "_freeze", "class", "def", "embedding_dim", "int", "nn", "num_positions", "padding_idx", "self", "super"], "pegasus/modeling_pegasus.py:PegasusSinusoidalPositionalEmbedding.create_weight": ["Embedding", "False", "FloatTensor", "ModelSinusoidalPositionalEmbedding", "array", "class", "cos", "create_weight", "def", "dim", "dtype", "else", "empty", "for", "if", "in", "j", "n_pos", "nn", "np", "out", "pos", "position_enc", "power", "range", "requires_grad", "return", "self", "sentinel", "shape", "sin", "torch", "weight"], "pegasus/modeling_pegasus.py:PegasusSinusoidalPositionalEmbedding.forward": ["Embedding", "ModelSinusoidalPositionalEmbedding", "None", "Size", "Tensor", "arange", "bsz", "class", "def", "device", "dtype", "forward", "if", "input_ids_shape", "int", "is", "long", "nn", "past_key_values_length", "position_ids", "return", "self", "seq_len", "super", "torch", "weight"], "pegasus/modeling_pegasus.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "contiguous", "def", "dim", "dropout", "float", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "shape", "size", "softmax", "torch", "training", "transpose", "value"], "pegasus/modeling_pegasus.py:PegasusAttention.__init__": ["False", "Instantiating", "Linear", "ModelAttention", "ModelConfig", "Module", "None", "Please", "True", "ValueError", "__class__", "__init__", "__name__", "a", "and", "be", "bias", "bool", "by", "caching", "call", "class", "config", "creating", "decoder", "def", "divisible", "dropout", "during", "embed_dim", "errors", "f", "float", "forward", "got", "head_dim", "if", "int", "is", "is_causal", "is_decoder", "k_proj", "layer_idx", "lead", "logger", "make", "must", "nn", "not", "num_heads", "out_proj", "passing", "provide", "q_proj", "raise", "recommended", "scaling", "self", "super", "sure", "the", "this", "to", "used", "v_proj", "warning_once", "when", "will", "without"], "pegasus/modeling_pegasus.py:PegasusAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "EncoderDecoderCache", "False", "FlashAttentionKwargs", "ModelAttention", "Module", "None", "Tensor", "True", "Unpack", "_attn_implementation", "and", "attention_interface", "attention_mask", "attn_output", "attn_weights", "bool", "bsz", "cache_position", "class", "config", "contiguous", "cross_attention_cache", "curr_past_key_values", "current_states", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "get", "head_dim", "hidden_states", "if", "is", "is_cross_attention", "is_updated", "isinstance", "k_proj", "key_states", "key_value_states", "keys", "kv_input_shape", "kwargs", "layer_idx", "layers", "nn", "not", "out_proj", "output_attentions", "past_key_values", "q_input_shape", "q_proj", "query_states", "reshape", "return", "scaling", "self", "self_attention_cache", "shape", "src_len", "tgt_len", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "values", "view"], "pegasus/modeling_pegasus.py:PegasusEncoderLayer.__init__": ["ACT2FN", "GradientCheckpointingLayer", "LayerNorm", "Linear", "ModelAttention", "ModelConfig", "ModelEncoderLayer", "__init__", "activation_dropout", "activation_fn", "activation_function", "attention_dropout", "class", "config", "d_model", "def", "dropout", "embed_dim", "encoder_attention_heads", "encoder_ffn_dim", "fc1", "fc2", "final_layer_norm", "nn", "num_heads", "self", "self_attn", "self_attn_layer_norm", "super"], "pegasus/modeling_pegasus.py:PegasusEncoderLayer.forward": ["False", "GradientCheckpointingLayer", "ModelEncoderLayer", "Tensor", "activation_dropout", "activation_fn", "attention_mask", "attn_weights", "bool", "clamp", "clamp_value", "class", "def", "dropout", "dtype", "fc1", "fc2", "final_layer_norm", "finfo", "float16", "forward", "functional", "hidden_states", "if", "max", "min", "nn", "output_attentions", "p", "residual", "return", "self", "self_attn", "self_attn_layer_norm", "torch", "training"], "pegasus/modeling_pegasus.py:PegasusDecoderLayer.__init__": ["ACT2FN", "GradientCheckpointingLayer", "LayerNorm", "Linear", "ModelAttention", "ModelConfig", "ModelDecoderLayer", "None", "True", "__init__", "activation_dropout", "activation_fn", "activation_function", "attention_dropout", "class", "config", "d_model", "decoder_attention_heads", "decoder_ffn_dim", "def", "dropout", "embed_dim", "encoder_attn", "encoder_attn_layer_norm", "fc1", "fc2", "final_layer_norm", "int", "is_causal", "is_decoder", "layer_idx", "nn", "num_heads", "self", "self_attn", "self_attn_layer_norm", "super"], "pegasus/modeling_pegasus.py:PegasusDecoderLayer.forward": ["Cache", "False", "GradientCheckpointingLayer", "ModelDecoderLayer", "None", "Tensor", "True", "activation_dropout", "activation_fn", "attention_mask", "bool", "cache_position", "class", "cross_attn_weights", "def", "dropout", "encoder_attention_mask", "encoder_attn", "encoder_attn_layer_norm", "encoder_hidden_states", "fc1", "fc2", "final_layer_norm", "forward", "functional", "hidden_states", "if", "is", "key_value_states", "nn", "not", "output_attentions", "outputs", "p", "past_key_values", "residual", "return", "self", "self_attn", "self_attn_layer_norm", "self_attn_weights", "torch", "training", "use_cache"], "pegasus/modeling_pegasus.py:PegasusPreTrainedModel._init_weights": ["ModelForConditionalGeneration", "ModelPreTrainedModel", "ModelSinusoidalPositionalEmbedding", "PreTrainedModel", "_init_weights", "class", "copy_", "create_weight", "def", "elif", "final_logits_bias", "if", "init", "isinstance", "module", "self", "super", "weight", "zeros_"], "pegasus/modeling_pegasus.py:PegasusEncoder.__init__": ["Embedding", "False", "LayerNorm", "ModelConfig", "ModelEncoder", "ModelEncoderLayer", "ModelPreTrainedModel", "ModelSinusoidalPositionalEmbedding", "ModuleList", "_", "__init__", "class", "config", "d_model", "def", "dropout", "else", "embed_dim", "embed_positions", "embed_scale", "embed_tokens", "encoder_layerdrop", "encoder_layers", "for", "gradient_checkpointing", "if", "in", "layer_norm", "layerdrop", "layers", "math", "max_position_embeddings", "max_source_positions", "nn", "pad_token_id", "padding_idx", "post_init", "range", "scale_embedding", "self", "sqrt", "super", "vocab_size"], "pegasus/modeling_pegasus.py:PegasusEncoder.resize_position_embeddings": ["ModelEncoder", "ModelPreTrainedModel", "ModelSinusoidalPositionalEmbedding", "Setting", "class", "config", "copy_", "create_weight", "d_model", "def", "device", "embed_positions", "f", "info", "init", "int", "logger", "max_position_embeddings", "new_num_position_embeddings", "padding_idx", "resize_position_embeddings", "self", "to", "weight"], "pegasus/modeling_pegasus.py:PegasusEncoder.get_position_embeddings": ["Embedding", "ModelEncoder", "ModelPreTrainedModel", "class", "def", "embed_positions", "get_position_embeddings", "nn", "return", "self"], "pegasus/modeling_pegasus.py:PegasusEncoder.forward": ["BaseModelOutput", "False", "ModelEncoder", "ModelPreTrainedModel", "None", "True", "ValueError", "You", "all_attentions", "and", "at", "attention_mask", "attentions", "both", "cannot", "class", "config", "create_bidirectional_mask", "def", "dropout", "dropout_probability", "either", "elif", "else", "embed_pos", "embed_positions", "embed_scale", "embed_tokens", "encoder_layer", "encoder_states", "enumerate", "for", "forward", "functional", "have", "hidden_states", "idx", "if", "in", "input_embeds", "input_ids", "input_shape", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layer_norm", "layer_outputs", "layerdrop", "layers", "nn", "not", "or", "output_attentions", "output_hidden_states", "p", "r", "raise", "rand", "return", "return_dict", "same", "self", "size", "specify", "the", "time", "to", "to_drop", "torch", "training", "tuple", "use_return_dict", "v", "view", "warn_if_padding_and_no_attention_mask"], "pegasus/modeling_pegasus.py:PegasusDecoder.__init__": ["Embedding", "False", "LayerNorm", "ModelConfig", "ModelDecoder", "ModelDecoderLayer", "ModelPreTrainedModel", "ModelSinusoidalPositionalEmbedding", "ModuleList", "__init__", "class", "config", "d_model", "decoder_layerdrop", "decoder_layers", "def", "dropout", "else", "embed_positions", "embed_scale", "embed_tokens", "for", "gradient_checkpointing", "i", "if", "in", "layer_idx", "layer_norm", "layerdrop", "layers", "math", "max_position_embeddings", "max_target_positions", "nn", "pad_token_id", "padding_idx", "post_init", "range", "scale_embedding", "self", "sqrt", "super", "vocab_size"], "pegasus/modeling_pegasus.py:PegasusDecoder.resize_position_embeddings": ["ModelDecoder", "ModelPreTrainedModel", "ModelSinusoidalPositionalEmbedding", "Setting", "class", "config", "copy_", "create_weight", "d_model", "def", "device", "embed_positions", "f", "info", "init", "int", "logger", "max_position_embeddings", "new_num_position_embeddings", "padding_idx", "resize_position_embeddings", "self", "to", "weight"], "pegasus/modeling_pegasus.py:PegasusDecoder.get_position_embeddings": ["Embedding", "ModelDecoder", "ModelPreTrainedModel", "class", "def", "embed_positions", "get_position_embeddings", "nn", "return", "self"], "pegasus/modeling_pegasus.py:PegasusDecoder.forward": ["BaseModelOutputWithPastAndCrossAttentions", "DynamicCache", "EncoderDecoderCache", "False", "ModelDecoder", "ModelPreTrainedModel", "None", "Setting", "True", "ValueError", "You", "all_cross_attentions", "all_hidden_states", "all_self_attns", "and", "arange", "at", "attention_mask", "attentions", "batch_size", "both", "cache_position", "cannot", "causal_mask", "checkpointing", "class", "config", "continue", "create_bidirectional_mask", "create_causal_mask", "cross_attentions", "decoder_input_ids", "decoder_inputs_embeds", "decoder_layer", "def", "device", "dropout", "dropout_probability", "either", "elif", "else", "embed_positions", "embed_scale", "embed_tokens", "encoder_attention_mask", "encoder_hidden_states", "enumerate", "for", "forward", "functional", "get_seq_length", "gradient", "gradient_checkpointing", "have", "hidden_states", "idx", "if", "in", "incompatible", "input", "input_embeds", "input_ids", "input_shape", "inputs_embeds", "is", "is_encoder_decoder", "is_torchdynamo_compiling", "isinstance", "kwargs", "last_hidden_state", "layer_norm", "layer_outputs", "layerdrop", "layers", "logger", "mask_seq_length", "nn", "not", "ones", "or", "output_attentions", "output_hidden_states", "p", "past_key_values", "past_key_values_length", "position_ids", "positions", "r", "raise", "rand", "return", "return_dict", "same", "self", "self_attention_cache", "self_attn_cache", "seq_length", "shape", "size", "specify", "the", "time", "to", "torch", "training", "tuple", "use_cache", "use_return_dict", "v", "view", "warning_once", "with"], "pegasus/modeling_pegasus.py:PegasusModel.__init__": ["Embedding", "ModelConfig", "ModelDecoder", "ModelEncoder", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "config", "d_model", "decoder", "def", "encoder", "nn", "pad_token_id", "padding_idx", "post_init", "self", "shared", "super", "vocab_size"], "pegasus/modeling_pegasus.py:PegasusModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "return", "self", "shared"], "pegasus/modeling_pegasus.py:PegasusModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "decoder", "def", "embed_tokens", "encoder", "self", "set_input_embeddings", "shared", "value"], "pegasus/modeling_pegasus.py:PegasusModel.resize_position_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "config", "decoder", "def", "encoder", "int", "max_position_embeddings", "new_num_position_embeddings", "resize_position_embeddings", "self"], "pegasus/modeling_pegasus.py:PegasusModel.get_position_embeddings": ["Embedding", "ModelModel", "ModelPreTrainedModel", "class", "decoder", "def", "encoder", "get_position_embeddings", "nn", "return", "self", "tuple"], "pegasus/modeling_pegasus.py:PegasusModel.forward": ["BaseModelOutput", "Cache", "FloatTensor", "ModelModel", "ModelPreTrainedModel", "None", "Seq2SeqModelOutput", "Tensor", "and", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "cross_attentions", "decoder", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_input_ids", "decoder_inputs_embeds", "decoder_outputs", "def", "elif", "else", "encoder", "encoder_attention_mask", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "isinstance", "kwargs", "last_hidden_state", "len", "not", "output_attentions", "output_hidden_states", "past_key_values", "r", "return", "return_dict", "self", "torch", "tuple", "use_cache", "use_return_dict"], "pegasus/modeling_pegasus.py:PegasusForConditionalGeneration.__init__": ["False", "GenerationMixin", "Linear", "ModelConfig", "ModelForConditionalGeneration", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "d_model", "def", "final_logits_bias", "lm_head", "model", "nn", "num_embeddings", "post_init", "register_buffer", "self", "shared", "super", "torch", "zeros"], "pegasus/modeling_pegasus.py:PegasusForConditionalGeneration.resize_token_embeddings": ["Embedding", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "True", "_resize_final_logits_bias", "bool", "class", "def", "int", "mean_resizing", "new_embeddings", "new_num_tokens", "nn", "pad_to_multiple_of", "resize_token_embeddings", "return", "self", "shape", "super", "weight"], "pegasus/modeling_pegasus.py:PegasusForConditionalGeneration._resize_final_logits_bias": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "_resize_final_logits_bias", "cat", "class", "def", "device", "dim", "else", "extra_bias", "final_logits_bias", "if", "int", "new_bias", "new_num_tokens", "old_num_tokens", "register_buffer", "self", "shape", "torch", "zeros"], "pegasus/modeling_pegasus.py:PegasusForConditionalGeneration.resize_position_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "config", "decoder", "def", "encoder", "int", "max_position_embeddings", "model", "new_num_position_embeddings", "resize_position_embeddings", "self"], "pegasus/modeling_pegasus.py:PegasusForConditionalGeneration.get_position_embeddings": ["Embedding", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "decoder", "def", "encoder", "get_position_embeddings", "model", "nn", "return", "self", "tuple"], "pegasus/modeling_pegasus.py:PegasusForConditionalGeneration.forward": ["Cache", "CrossEntropyLoss", "False", "FloatTensor", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Seq2SeqLMOutput", "Tensor", "The", "and", "argument", "attention_mask", "bool", "cache_position", "changed", "class", "config", "cross_attentions", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_input_ids", "decoder_inputs_embeds", "decoder_start_token_id", "def", "else", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "final_logits_bias", "forward", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "lm_head", "lm_logits", "logger", "logits", "loss", "loss_fct", "masked_lm_loss", "model", "not", "output", "output_attentions", "output_hidden_states", "outputs", "pad_token_id", "past_key_values", "provided", "r", "return", "return_dict", "self", "shift_tokens_right", "since", "to", "torch", "tuple", "use_cache", "use_return_dict", "view", "vocab_size", "warning"], "pegasus/modeling_pegasus.py:PegasusForConditionalGeneration.prepare_decoder_input_ids_from_labels": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "Tensor", "class", "config", "decoder_start_token_id", "def", "labels", "pad_token_id", "prepare_decoder_input_ids_from_labels", "return", "self", "shift_tokens_right", "torch"], "pegasus/modeling_pegasus.py:PegasusDecoderWrapper.__init__": ["ModelDecoder", "ModelDecoderWrapper", "ModelPreTrainedModel", "__init__", "class", "config", "decoder", "def", "post_init", "self", "super"], "pegasus/modeling_pegasus.py:PegasusDecoderWrapper.forward": ["ModelDecoderWrapper", "ModelPreTrainedModel", "args", "class", "decoder", "def", "forward", "kwargs", "return", "self"], "pegasus/modeling_pegasus.py:PegasusForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "ModelDecoderWrapper", "ModelForCausalLM", "ModelPreTrainedModel", "True", "__init__", "bias", "class", "config", "copy", "deepcopy", "def", "hidden_size", "is_decoder", "is_encoder_decoder", "lm_head", "model", "nn", "post_init", "self", "super", "vocab_size"], "pegasus/modeling_pegasus.py:PegasusForCausalLM.get_input_embeddings": ["GenerationMixin", "ModelForCausalLM", "ModelPreTrainedModel", "class", "decoder", "def", "embed_tokens", "get_input_embeddings", "model", "return", "self"], "pegasus/modeling_pegasus.py:PegasusForCausalLM.set_input_embeddings": ["GenerationMixin", "ModelForCausalLM", "ModelPreTrainedModel", "class", "decoder", "def", "embed_tokens", "model", "self", "set_input_embeddings", "value"], "pegasus/modeling_pegasus.py:PegasusForCausalLM.get_position_embeddings": ["Embedding", "GenerationMixin", "ModelForCausalLM", "ModelPreTrainedModel", "class", "decoder", "def", "get_position_embeddings", "model", "nn", "return", "self"], "pegasus/modeling_pegasus.py:PegasusForCausalLM.resize_position_embeddings": ["GenerationMixin", "ModelForCausalLM", "ModelPreTrainedModel", "class", "config", "decoder", "def", "int", "max_position_embeddings", "model", "new_num_position_embeddings", "resize_position_embeddings", "self"], "pegasus/modeling_pegasus.py:PegasusForCausalLM.forward": ["Cache", "CausalLMOutputWithCrossAttentions", "CrossEntropyLoss", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForCausalLM", "ModelPreTrainedModel", "None", "Tensor", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "cross_attentions", "decoder", "def", "device", "else", "encoder_attention_mask", "encoder_hidden_states", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "lm_head", "logits", "logits_to_keep", "loss", "loss_fct", "model", "not", "output", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "r", "return", "return_dict", "self", "slice", "slice_indices", "to", "torch", "tuple", "use_cache", "use_return_dict", "view", "vocab_size"], "pegasus_x/modeling_pegasus_x.py:shift_tokens_right": ["Model_tokens_right", "Modeled_input_ids", "None", "Tensor", "ValueError", "be", "clone", "config", "decoder_start_token_id", "def", "defined", "has", "if", "input_ids", "int", "is", "masked_fill_", "model", "new_zeros", "pad_token_id", "raise", "return", "self", "shape", "to", "torch"], "pegasus_x/modeling_pegasus_x.py:PegasusXScaledWordEmbedding.__init__": ["Embedding", "ModelScaledWordEmbedding", "None", "__init__", "class", "def", "embed_scale", "embedding_dim", "float", "int", "nn", "num_embeddings", "padding_idx", "self", "super"], "pegasus_x/modeling_pegasus_x.py:PegasusXScaledWordEmbedding.forward": ["Embedding", "ModelScaledWordEmbedding", "Tensor", "class", "def", "embed_scale", "forward", "input_ids", "nn", "return", "self", "super", "torch"], "pegasus_x/modeling_pegasus_x.py:PegasusXSinusoidalPositionalEmbedding.__init__": ["ModelSinusoidalPositionalEmbedding", "Module", "__init__", "class", "def", "embed_dim", "int", "max_scale", "nn", "self", "super"], "pegasus_x/modeling_pegasus_x.py:PegasusXSinusoidalPositionalEmbedding.forward": ["ModelSinusoidalPositionalEmbedding", "Module", "None", "Tensor", "arange", "batch_size", "class", "cos", "def", "device", "div_term", "dtype", "embed_dim", "exp", "expand", "float", "forward", "half_d_feature", "if", "input_embeds", "int", "int64", "is", "log", "long", "max_scale", "nn", "np", "past_key_values_length", "pe", "position_ids", "return", "self", "seq_len", "shape", "sin", "torch", "type_as", "zeros"], "pegasus_x/modeling_pegasus_x.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "contiguous", "def", "dim", "dropout", "float", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "shape", "size", "softmax", "torch", "training", "transpose", "value"], "pegasus_x/modeling_pegasus_x.py:PegasusXAttention.__init__": ["False", "Instantiating", "Linear", "ModelAttention", "ModelConfig", "Module", "None", "Please", "True", "ValueError", "__class__", "__init__", "__name__", "a", "and", "be", "bias", "bool", "by", "caching", "call", "class", "config", "creating", "decoder", "def", "divisible", "dropout", "during", "embed_dim", "errors", "f", "float", "forward", "got", "head_dim", "if", "int", "is", "is_causal", "is_decoder", "k_proj", "layer_idx", "lead", "logger", "make", "must", "nn", "not", "num_heads", "out_proj", "passing", "provide", "q_proj", "raise", "recommended", "scaling", "self", "super", "sure", "the", "this", "to", "used", "v_proj", "warning_once", "when", "will", "without"], "pegasus_x/modeling_pegasus_x.py:PegasusXAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "EncoderDecoderCache", "False", "FlashAttentionKwargs", "ModelAttention", "Module", "None", "Tensor", "True", "Unpack", "_attn_implementation", "and", "attention_interface", "attention_mask", "attn_output", "attn_weights", "bool", "bsz", "cache_position", "class", "config", "contiguous", "cross_attention_cache", "curr_past_key_values", "current_states", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "get", "head_dim", "hidden_states", "if", "is", "is_cross_attention", "is_updated", "isinstance", "k_proj", "key_states", "key_value_states", "keys", "kv_input_shape", "kwargs", "layer_idx", "layers", "nn", "not", "out_proj", "output_attentions", "past_key_values", "q_input_shape", "q_proj", "query_states", "reshape", "return", "scaling", "self", "self_attention_cache", "shape", "src_len", "tgt_len", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "values", "view"], "pegasus_x/modeling_pegasus_x.py:PegasusXGlobalLocalAttention.__init__": ["False", "Linear", "ModelGlobalLocalAttention", "Module", "ValueError", "__init__", "and", "be", "bias", "block_size", "bool", "by", "class", "def", "divisible", "dropout", "embed_dim", "f", "float", "got", "head_dim", "if", "int", "is_decoder", "k_proj", "must", "nn", "num_heads", "out_proj", "q_proj", "raise", "scaling", "self", "super", "v_proj"], "pegasus_x/modeling_pegasus_x.py:PegasusXGlobalLocalAttention._shape": ["ModelGlobalLocalAttention", "Module", "Tensor", "_shape", "bsz", "class", "contiguous", "def", "head_dim", "int", "nn", "num_heads", "return", "self", "seq_len", "tensor", "torch", "transpose", "view"], "pegasus_x/modeling_pegasus_x.py:PegasusXGlobalLocalAttention.forward": ["DimensionInfo", "False", "ModelGlobalLocalAttention", "Module", "None", "Tensor", "_shape", "attention_mask", "attn_probs", "batch_size", "block_size", "bool", "bsz", "class", "compute_global_attention_representations", "compute_local_attention_representations", "contiguous", "def", "dim", "dim_per_head", "else", "forward", "global", "global_attn_output", "global_attn_probs", "global_hidden_states", "global_k", "global_len", "global_q", "global_v", "head_dim", "hidden_dim", "if", "k_proj", "local", "local_attn_output", "local_attn_probs", "local_k", "local_q", "local_v", "mask", "nn", "num_blocks", "num_heads", "out_proj", "output_attentions", "padded_seq_len", "permute", "q_proj", "return", "scaling", "self", "seq_len", "shape", "token_hidden_states", "torch", "transpose", "tuple", "v_proj", "view"], "pegasus_x/modeling_pegasus_x.py:PegasusXGlobalLocalAttention.compute_global_attention_representations": ["BHGF", "BHGX", "BHXF", "DimensionInfo", "ModelGlobalLocalAttention", "Module", "None", "attn_output", "attn_probs", "attn_weights", "cat", "class", "compute_global_attention_representations", "def", "dim", "dropout", "einsum", "extended_mask", "functional", "global_and_local_k", "global_and_local_v", "global_k", "global_len", "global_q", "global_v", "local_k", "local_v", "mask", "nn", "p", "pad", "return", "self", "softmax", "torch", "training", "value"], "pegasus_x/modeling_pegasus_x.py:PegasusXGlobalLocalAttention.compute_local_attention_representations": ["BHGF", "BHNKF", "BHNKG", "BHNKX", "BHNXF", "DimensionInfo", "ModelGlobalLocalAttention", "Module", "None", "attn_output", "attn_probs", "attn_weights", "batch_size", "block_size", "blocked_local2global", "blocked_local2local", "blocked_local_k", "blocked_local_q", "blocked_local_v", "cat", "class", "compute_local_attention_representations", "def", "dim", "dim_per_head", "dropout", "einsum", "extended_mask", "functional", "global_k", "global_len", "global_v", "local2global_attn_output", "local2global_attn_probs", "local2local_attn_output", "local2local_attn_probs", "local_k", "local_q", "local_v", "mask", "nn", "num_blocks", "num_heads", "p", "pad", "return", "self", "softmax", "torch", "training", "value", "view"], "pegasus_x/modeling_pegasus_x.py:PegasusXEncoderLayer.__init__": ["ACT2FN", "GradientCheckpointingLayer", "LayerNorm", "Linear", "ModelConfig", "ModelEncoderLayer", "ModelGlobalLocalAttention", "__init__", "activation_dropout", "activation_fn", "activation_function", "attention_dropout", "block_size", "bool", "class", "config", "d_model", "def", "dropout", "embed_dim", "encoder_attention_heads", "encoder_ffn_dim", "fc1", "fc2", "final_layer_norm", "global_self_attn_layer_norm", "nn", "num_heads", "self", "self_attn", "self_attn_layer_norm", "stagger_blocks_this_layer", "super"], "pegasus_x/modeling_pegasus_x.py:PegasusXEncoderLayer.forward": ["False", "GradientCheckpointingLayer", "ModelEncoderLayer", "Tensor", "activation_dropout", "activation_fn", "attention_mask", "attn_weights", "block_size", "bool", "class", "def", "dropout", "fc1", "fc2", "final_layer_norm", "forward", "functional", "global_hidden_states", "global_residual", "global_self_attn_layer_norm", "hidden_states", "if", "nn", "output_attentions", "outputs", "p", "pad_local_tokens", "padded_hidden_states", "residual", "return", "self", "self_attn", "self_attn_layer_norm", "stagger_blocks_this_layer", "token_hidden_states", "torch", "training", "unpad_local_tokens"], "pegasus_x/modeling_pegasus_x.py:PegasusXEncoderLayer.pad_local_tokens": ["GradientCheckpointingLayer", "ModelEncoderLayer", "attention_mask", "block_size", "class", "cls", "def", "dtype", "finfo", "functional", "hidden_states", "mask_min_value", "min", "nn", "pad", "pad_local_tokens", "pad_size", "padded_hidden_states", "padded_mask", "return", "torch", "value"], "pegasus_x/modeling_pegasus_x.py:PegasusXEncoderLayer.unpad_local_tokens": ["GradientCheckpointingLayer", "ModelEncoderLayer", "block_size", "class", "cls", "def", "pad_size", "padded_hidden_states", "return", "unpad_local_tokens"], "pegasus_x/modeling_pegasus_x.py:PegasusXDecoderLayer.__init__": ["ACT2FN", "False", "GradientCheckpointingLayer", "LayerNorm", "Linear", "ModelAttention", "ModelConfig", "ModelDecoderLayer", "None", "True", "__init__", "activation_dropout", "activation_fn", "activation_function", "attention_dropout", "bias", "class", "config", "d_model", "decoder_attention_heads", "decoder_ffn_dim", "def", "dropout", "embed_dim", "encoder_attn", "encoder_attn_layer_norm", "fc1", "fc2", "final_layer_norm", "int", "is_decoder", "layer_idx", "nn", "num_heads", "self", "self_attn", "self_attn_layer_norm", "super"], "pegasus_x/modeling_pegasus_x.py:PegasusXDecoderLayer.forward": ["Cache", "False", "GradientCheckpointingLayer", "ModelDecoderLayer", "None", "Tensor", "True", "activation_dropout", "activation_fn", "attention_mask", "bool", "cache_position", "class", "cross_attn_weights", "def", "dropout", "encoder_attention_mask", "encoder_attn", "encoder_attn_layer_norm", "encoder_hidden_states", "fc1", "fc2", "final_layer_norm", "forward", "functional", "hidden_states", "if", "is", "key_value_states", "nn", "not", "output_attentions", "outputs", "p", "past_key_values", "residual", "return", "self", "self_attn", "self_attn_layer_norm", "self_attn_weights", "torch", "training", "use_cache"], "pegasus_x/modeling_pegasus_x.py:PegasusXEncoder.__init__": ["Embedding", "False", "LayerNorm", "ModelConfig", "ModelEncoder", "ModelEncoderLayer", "ModelPreTrainedModel", "ModelScaledWordEmbedding", "ModelSinusoidalPositionalEmbedding", "ModuleList", "__init__", "and", "class", "config", "d_model", "def", "dropout", "else", "embed_dim", "embed_global", "embed_positions", "embed_scale", "embed_tokens", "encoder_layerdrop", "encoder_layers", "for", "gradient_checkpointing", "i", "if", "in", "layer_norm", "layerdrop", "layers", "math", "max_position_embeddings", "max_source_positions", "nn", "num_global_tokens", "pad_token_id", "padding_idx", "post_init", "range", "scale_embedding", "self", "sqrt", "stagger_blocks_this_layer", "stagger_local_blocks", "super", "vocab_size"], "pegasus_x/modeling_pegasus_x.py:PegasusXEncoder.resize_position_embeddings": ["ModelEncoder", "ModelPreTrainedModel", "ModelSinusoidalPositionalEmbedding", "Setting", "class", "config", "d_model", "def", "device", "embed_positions", "f", "info", "int", "logger", "max_position_embeddings", "new_num_position_embeddings", "resize_position_embeddings", "self", "to"], "pegasus_x/modeling_pegasus_x.py:PegasusXEncoder.get_position_embeddings": ["Embedding", "ModelEncoder", "ModelPreTrainedModel", "class", "def", "embed_positions", "get_position_embeddings", "nn", "return", "self"], "pegasus_x/modeling_pegasus_x.py:PegasusXEncoder.forward": ["BaseModelOutput", "False", "ModelEncoder", "ModelPreTrainedModel", "None", "True", "ValueError", "You", "_", "all_attentions", "and", "arange", "at", "attention_mask", "attentions", "batch_size", "block_size", "bool", "both", "cannot", "class", "config", "def", "device", "dropout", "dropout_probability", "dtype", "either", "elif", "else", "embed_global", "embed_pos", "embed_positions", "embed_tokens", "encoder_layer", "encoder_states", "enumerate", "expand", "finfo", "for", "forward", "functional", "global_hidden_states", "have", "hidden_states", "idx", "if", "in", "input_ids", "input_shape", "inputs_embeds", "inverted_mask", "is", "kwargs", "last_hidden_state", "layer_norm", "layer_outputs", "layerdrop", "layers", "mask_min_value", "masked_fill", "min", "nn", "not", "num_global_tokens", "ones", "or", "output_attentions", "output_hidden_states", "p", "pad", "pad_len", "r", "raise", "rand", "return", "return_dict", "same", "self", "seq_len", "shape", "size", "specify", "the", "time", "to", "to_drop", "torch", "training", "tuple", "use_return_dict", "v", "value", "view", "warn_if_padding_and_no_attention_mask"], "pegasus_x/modeling_pegasus_x.py:PegasusXDecoder.__init__": ["False", "LayerNorm", "ModelConfig", "ModelDecoder", "ModelDecoderLayer", "ModelPreTrainedModel", "ModelScaledWordEmbedding", "ModelSinusoidalPositionalEmbedding", "ModuleList", "__init__", "class", "config", "d_model", "decoder_layerdrop", "decoder_layers", "def", "dropout", "else", "embed_positions", "embed_scale", "embed_tokens", "for", "gradient_checkpointing", "i", "if", "in", "layer_idx", "layer_norm", "layerdrop", "layers", "math", "max_position_embeddings", "max_target_positions", "nn", "pad_token_id", "padding_idx", "post_init", "range", "scale_embedding", "self", "sqrt", "super", "vocab_size"], "pegasus_x/modeling_pegasus_x.py:PegasusXDecoder.forward": ["BaseModelOutputWithPastAndCrossAttentions", "DynamicCache", "EncoderDecoderCache", "False", "ModelDecoder", "ModelPreTrainedModel", "None", "Setting", "True", "ValueError", "You", "all_cross_attentions", "all_hidden_states", "all_self_attns", "and", "arange", "at", "attention_mask", "attentions", "batch_size", "both", "cache_position", "cannot", "causal_mask", "checkpointing", "class", "config", "continue", "create_bidirectional_mask", "create_causal_mask", "cross_attentions", "decoder_input_ids", "decoder_inputs_embeds", "decoder_layer", "def", "device", "dropout", "dropout_probability", "either", "elif", "else", "embed_positions", "embed_tokens", "encoder_attention_mask", "encoder_hidden_states", "enumerate", "for", "forward", "functional", "get_seq_length", "gradient", "gradient_checkpointing", "have", "hidden_states", "idx", "if", "in", "incompatible", "input", "input_embeds", "input_ids", "input_shape", "inputs_embeds", "is", "is_torchdynamo_compiling", "isinstance", "kwargs", "last_hidden_state", "layer_norm", "layer_outputs", "layerdrop", "layers", "logger", "mask_seq_length", "nn", "not", "ones", "or", "output_attentions", "output_hidden_states", "p", "past_key_values", "past_key_values_length", "position_ids", "r", "raise", "rand", "return", "return_dict", "same", "self", "self_attention_cache", "self_attn_cache", "seq_length", "shape", "size", "specify", "the", "time", "to", "torch", "training", "tuple", "unsqueeze", "use_cache", "use_return_dict", "v", "view", "warning_once", "with"], "pegasus_x/modeling_pegasus_x.py:PegasusXModel.__init__": ["ModelConfig", "ModelDecoder", "ModelEncoder", "ModelModel", "ModelPreTrainedModel", "ModelScaledWordEmbedding", "__init__", "class", "config", "d_model", "decoder", "def", "else", "embed_scale", "encoder", "if", "math", "pad_token_id", "padding_idx", "post_init", "scale_embedding", "self", "shared", "sqrt", "super", "vocab_size"], "pegasus_x/modeling_pegasus_x.py:PegasusXModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "return", "self", "shared"], "pegasus_x/modeling_pegasus_x.py:PegasusXModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "decoder", "def", "embed_tokens", "encoder", "self", "set_input_embeddings", "shared", "value"], "pegasus_x/modeling_pegasus_x.py:PegasusXModel.resize_position_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "config", "decoder", "def", "encoder", "int", "max_position_embeddings", "new_num_position_embeddings", "resize_position_embeddings", "self"], "pegasus_x/modeling_pegasus_x.py:PegasusXModel.get_position_embeddings": ["Embedding", "ModelModel", "ModelPreTrainedModel", "class", "decoder", "def", "encoder", "get_position_embeddings", "nn", "return", "self", "tuple"], "pegasus_x/modeling_pegasus_x.py:PegasusXModel.forward": ["BaseModelOutput", "Cache", "FloatTensor", "ModelModel", "ModelPreTrainedModel", "None", "Seq2SeqModelOutput", "Tensor", "and", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "cross_attentions", "decoder", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_input_ids", "decoder_inputs_embeds", "decoder_outputs", "def", "elif", "else", "encoder", "encoder_attention_mask", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "isinstance", "kwargs", "last_hidden_state", "len", "not", "output_attentions", "output_hidden_states", "past_key_values", "r", "return", "return_dict", "self", "torch", "tuple", "use_cache", "use_return_dict"], "pegasus_x/modeling_pegasus_x.py:PegasusXForConditionalGeneration.__init__": ["False", "GenerationMixin", "Linear", "ModelConfig", "ModelForConditionalGeneration", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "d_model", "def", "lm_head", "model", "nn", "num_embeddings", "post_init", "self", "shared", "super"], "pegasus_x/modeling_pegasus_x.py:PegasusXForConditionalGeneration.resize_position_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "config", "decoder", "def", "encoder", "int", "max_position_embeddings", "model", "new_num_position_embeddings", "resize_position_embeddings", "self"], "pegasus_x/modeling_pegasus_x.py:PegasusXForConditionalGeneration.get_position_embeddings": ["Embedding", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "decoder", "def", "encoder", "get_position_embeddings", "model", "nn", "return", "self", "tuple"], "pegasus_x/modeling_pegasus_x.py:PegasusXForConditionalGeneration.forward": ["Cache", "CrossEntropyLoss", "False", "FloatTensor", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Seq2SeqLMOutput", "Tensor", "The", "and", "argument", "attention_mask", "bool", "cache_position", "changed", "class", "config", "cross_attentions", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_input_ids", "decoder_inputs_embeds", "decoder_start_token_id", "def", "else", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "forward", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "lm_head", "lm_logits", "logger", "logits", "loss", "loss_fct", "masked_lm_loss", "model", "not", "output", "output_attentions", "output_hidden_states", "outputs", "pad_token_id", "past_key_values", "provided", "r", "return", "return_dict", "self", "shift_tokens_right", "since", "to", "torch", "tuple", "use_cache", "use_return_dict", "view", "vocab_size", "warning"], "pegasus_x/modeling_pegasus_x.py:PegasusXForConditionalGeneration.prepare_decoder_input_ids_from_labels": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "Tensor", "class", "config", "decoder_start_token_id", "def", "labels", "pad_token_id", "prepare_decoder_input_ids_from_labels", "return", "self", "shift_tokens_right", "torch"], "pegasus_x/modeling_pegasus_x.py:PegasusXDecoderWrapper.__init__": ["ModelDecoder", "ModelDecoderWrapper", "ModelPreTrainedModel", "__init__", "class", "config", "decoder", "def", "post_init", "self", "super"], "pegasus_x/modeling_pegasus_x.py:PegasusXDecoderWrapper.forward": ["ModelDecoderWrapper", "ModelPreTrainedModel", "args", "class", "decoder", "def", "forward", "kwargs", "return", "self"], "perceiver/modeling_perceiver.py:PerceiverEmbeddings.__init__": ["ModelEmbeddings", "Module", "Parameter", "__init__", "class", "config", "d_latents", "def", "latents", "nn", "num_latents", "randn", "self", "super", "torch"], "perceiver/modeling_perceiver.py:PerceiverEmbeddings.forward": ["ModelEmbeddings", "Module", "batch_size", "class", "def", "expand", "forward", "int", "latents", "nn", "return", "self"], "perceiver/modeling_perceiver.py:PerceiverSelfAttention.__init__": ["Dropout", "False", "Identity", "LayerNorm", "Linear", "ModelSelfAttention", "Module", "None", "ValueError", "__init__", "attention_probs_dropout_prob", "be", "by", "class", "config", "def", "divisible", "dropout", "else", "f", "if", "is", "is_cross_attention", "key", "kv_dim", "layernorm1", "layernorm2", "must", "nn", "num_heads", "q_dim", "qk_channels", "qk_channels_per_head", "query", "raise", "self", "super", "v_channels", "v_channels_per_head", "value"], "perceiver/modeling_perceiver.py:PerceiverSelfAttention.transpose_for_scores": ["ModelSelfAttention", "Module", "channels_per_head", "class", "def", "new_x_shape", "nn", "num_heads", "permute", "return", "self", "size", "transpose_for_scores", "view", "x"], "perceiver/modeling_perceiver.py:PerceiverSelfAttention.forward": ["False", "FloatTensor", "ModelSelfAttention", "Module", "None", "Softmax", "Tensor", "_", "attention_mask", "attention_probs", "attention_scores", "batch_size", "bool", "class", "context_layer", "contiguous", "def", "dim", "dropout", "else", "forward", "hidden_states", "hiddens", "if", "inputs", "inputs_mask", "is", "is_cross_attention", "key", "keys", "layernorm1", "layernorm2", "math", "matmul", "new_context_layer_shape", "nn", "not", "num_heads", "output_attentions", "outputs", "permute", "q_head_dim", "qk_channels_per_head", "queries", "query", "return", "self", "seq_len", "shape", "size", "sqrt", "torch", "transpose", "transpose_for_scores", "tuple", "v_channels_per_head", "v_head_dim", "value", "values", "view"], "perceiver/modeling_perceiver.py:PerceiverSelfOutput.__init__": ["Linear", "ModelSelfOutput", "Module", "__init__", "class", "config", "def", "dense", "input_channels", "nn", "output_channels", "self", "super"], "perceiver/modeling_perceiver.py:PerceiverSelfOutput.forward": ["ModelSelfOutput", "Module", "Tensor", "class", "def", "dense", "forward", "hidden_states", "nn", "return", "self", "torch"], "perceiver/modeling_perceiver.py:PerceiverAttention.__init__": ["False", "ModelAttention", "ModelSelfAttention", "ModelSelfOutput", "Module", "None", "True", "Unknown", "ValueError", "__init__", "and", "class", "config", "cross_attention_shape_for_attention", "def", "elif", "else", "f", "for", "if", "input_channels", "is", "is_cross_attention", "kv", "kv_dim", "nn", "num_heads", "output", "output_channels", "q", "q_dim", "qk_channels", "raise", "self", "super", "use_query_residual", "v_channels", "value"], "perceiver/modeling_perceiver.py:PerceiverAttention.forward": ["False", "FloatTensor", "ModelAttention", "Module", "None", "Tensor", "attention_mask", "attention_output", "bool", "class", "def", "forward", "hidden_states", "if", "inputs", "inputs_mask", "nn", "output", "output_attentions", "outputs", "return", "self", "self_outputs", "torch", "tuple", "use_query_residual"], "perceiver/modeling_perceiver.py:PerceiverMLP.__init__": ["ACT2FN", "Linear", "ModelMLP", "Module", "__init__", "class", "config", "def", "dense1", "dense2", "else", "hidden_act", "if", "input_size", "intermediate_act_fn", "isinstance", "nn", "self", "str", "super", "widening_factor"], "perceiver/modeling_perceiver.py:PerceiverMLP.forward": ["ModelMLP", "Module", "Tensor", "class", "def", "dense1", "dense2", "forward", "hidden_states", "intermediate_act_fn", "nn", "return", "self", "torch"], "perceiver/modeling_perceiver.py:PerceiverLayer.__init__": ["False", "LayerNorm", "ModelAttention", "ModelLayer", "ModelMLP", "Module", "None", "True", "__init__", "attention", "chunk_size_feed_forward", "class", "config", "def", "input_size", "is_cross_attention", "kv_dim", "layernorm", "mlp", "nn", "num_heads", "q_dim", "qk_channels", "self", "seq_len_dim", "super", "use_query_residual", "v_channels", "widening_factor"], "perceiver/modeling_perceiver.py:PerceiverLayer.forward": ["False", "FloatTensor", "ModelLayer", "Module", "None", "Tensor", "apply_chunking_to_forward", "attention", "attention_mask", "attention_output", "attention_outputs", "bool", "chunk_size_feed_forward", "class", "def", "feed_forward_chunk", "forward", "hidden_states", "inputs", "inputs_mask", "layer_output", "nn", "output_attentions", "outputs", "return", "self", "seq_len_dim", "torch", "tuple"], "perceiver/modeling_perceiver.py:PerceiverLayer.feed_forward_chunk": ["ModelLayer", "Module", "attention_output", "class", "def", "feed_forward_chunk", "layer_output", "layernorm", "mlp", "nn", "return", "self"], "perceiver/modeling_perceiver.py:PerceiverEncoder.__init__": ["False", "ModelEncoder", "ModelLayer", "Module", "ModuleList", "None", "True", "ValueError", "_", "__init__", "append", "be", "by", "class", "config", "cross_attention", "cross_attention_widening_factor", "d_latents", "def", "divisible", "f", "for", "if", "in", "is_cross_attention", "kv_dim", "layer", "must", "nn", "num_cross_attend_heads", "num_cross_attention_heads", "num_heads", "num_self_attend_heads", "num_self_attends_per_block", "num_self_attention_heads", "num_z_channels", "q_dim", "qk_channels", "raise", "range", "self", "self_attends", "self_attention_layers", "self_attention_widening_factor", "super", "use_query_residual", "v_channels", "widening_factor"], "perceiver/modeling_perceiver.py:PerceiverEncoder.forward": ["BaseModelOutputWithCrossAttentions", "False", "FloatTensor", "ModelEncoder", "Module", "None", "Tensor", "True", "_", "all_cross_attentions", "all_hidden_states", "all_self_attentions", "attention_mask", "attentions", "bool", "class", "config", "cross_attention", "cross_attentions", "def", "else", "enumerate", "for", "forward", "hidden_states", "i", "if", "in", "inputs", "inputs_mask", "is", "last_hidden_state", "layer_module", "layer_outputs", "nn", "not", "num_blocks", "output_attentions", "output_hidden_states", "range", "return", "return_dict", "self", "self_attends", "torch", "tuple", "v"], "perceiver/modeling_perceiver.py:PerceiverPreTrainedModel._init_weights": ["BatchNorm2d", "Conv2d", "Embedding", "False", "LayerNorm", "Linear", "ModelPreTrainedModel", "ModelTrainablePositionEncoding", "None", "ParameterDict", "PreTrainedModel", "_init_weights", "_is_hf_initialized", "and", "bias", "class", "config", "def", "elif", "for", "getattr", "hasattr", "if", "in", "init", "initializer_range", "is", "isinstance", "latents", "mean", "modality", "module", "nn", "normal_", "not", "num_batches_tracked", "ones_", "padding_idx", "position_embeddings", "running_mean", "running_var", "self", "std", "weight", "zeros_"], "perceiver/modeling_perceiver.py:PerceiverModel.__init__": ["ModelAbstractDecoder", "ModelEmbeddings", "ModelEncoder", "ModelModel", "ModelPreTrainedModel", "None", "Optional", "PostprocessorType", "PreprocessorType", "__init__", "class", "config", "d_model", "decoder", "def", "else", "embeddings", "encoder", "if", "input_preprocessor", "is", "kv_dim", "not", "num_channels", "output_postprocessor", "post_init", "r", "self", "super"], "perceiver/modeling_perceiver.py:PerceiverModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "latents", "return", "self"], "perceiver/modeling_perceiver.py:PerceiverModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "latents", "self", "set_input_embeddings", "value"], "perceiver/modeling_perceiver.py:PerceiverModel.forward": ["False", "FloatTensor", "Last", "Make", "ModelModel", "ModelModelOutput", "ModelPreTrainedModel", "None", "Tensor", "ValueError", "_", "and", "appropriately", "attention_mask", "attentions", "audio", "batch_size", "bool", "class", "config", "correspond", "cross_attentions", "d_model", "decoder", "decoder_outputs", "decoder_query", "def", "device", "dict", "dimension", "doesn", "else", "embedding_output", "embeddings", "encoder", "encoder_outputs", "extended_attention_mask", "f", "forward", "hidden_states", "if", "image", "input_preprocessor", "inputs", "inputs_mask", "inputs_without_pos", "interpolate_pos_encoding", "invert_attention_mask", "is", "kwargs", "label", "last_hidden_state", "logits", "modality_sizes", "not", "of", "ones", "output_attentions", "output_hidden_states", "output_modality_sizes", "output_postprocessor", "query_mask", "r", "raise", "return", "return_dict", "self", "seq_length", "sequence_output", "set", "shape", "size", "str", "subsampled_output_points", "subsampled_points", "sure", "t", "the", "to", "torch", "tuple", "use_return_dict", "z"], "perceiver/modeling_perceiver.py:PerceiverForMaskedLM.__init__": ["False", "Model", "ModelBasicDecoder", "ModelConfig", "ModelEmbeddingDecoder", "ModelForMaskedLM", "ModelModel", "ModelPreTrainedModel", "ModelTextPreprocessor", "__init__", "class", "config", "d_latents", "decoder", "def", "embedding_decoder", "final_project", "index_dims", "input_preprocessor", "max_position_embeddings", "num_channels", "num_heads", "output_index_dims", "output_num_channels", "post_init", "qk_channels", "self", "super", "text_preprocessor", "trainable_position_encoding_kwargs", "trainable_position_encoding_kwargs_decoder", "use_query_residual", "v_channels"], "perceiver/modeling_perceiver.py:PerceiverForMaskedLM.forward": ["CrossEntropyLoss", "Model", "ModelForMaskedLM", "ModelMaskedLMOutput", "ModelPreTrainedModel", "None", "Tensor", "ValueError", "You", "and", "attention_mask", "attentions", "bool", "both", "cannot", "class", "config", "cross_attentions", "def", "elif", "else", "embedding_decoder", "embedding_layer", "embeddings", "forward", "hidden_states", "if", "input_ids", "input_preprocessor", "inputs", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "masked_lm_loss", "not", "output", "output_attentions", "output_hidden_states", "outputs", "r", "raise", "return", "return_dict", "self", "torch", "tuple", "use", "use_return_dict", "view", "vocab_size"], "perceiver/modeling_perceiver.py:PerceiverForSequenceClassification.__init__": ["Model", "ModelClassificationDecoder", "ModelForSequenceClassification", "ModelModel", "ModelPreTrainedModel", "ModelTextPreprocessor", "True", "__init__", "class", "config", "d_latents", "decoder", "def", "index_dims", "input_preprocessor", "num_channels", "num_labels", "post_init", "self", "super", "trainable_position_encoding_kwargs", "trainable_position_encoding_kwargs_decoder", "use_query_residual"], "perceiver/modeling_perceiver.py:PerceiverForSequenceClassification.forward": ["BCEWithLogitsLoss", "CrossEntropyLoss", "MSELoss", "Model", "ModelClassifierOutput", "ModelForSequenceClassification", "ModelPreTrainedModel", "None", "Tensor", "ValueError", "You", "and", "attention_mask", "attentions", "bool", "both", "cannot", "class", "config", "cross_attentions", "def", "dtype", "elif", "else", "forward", "hidden_states", "if", "input_ids", "inputs", "int", "is", "kwargs", "labels", "logits", "long", "loss", "loss_fct", "multi_label_classification", "not", "num_labels", "or", "output", "output_attentions", "output_hidden_states", "outputs", "problem_type", "r", "raise", "regression", "return", "return_dict", "self", "single_label_classification", "squeeze", "torch", "tuple", "use", "use_return_dict", "view"], "perceiver/modeling_perceiver.py:PerceiverForImageClassificationLearned.__init__": ["Model", "ModelClassificationDecoder", "ModelForImageClassificationLearned", "ModelImagePreprocessor", "ModelModel", "ModelPreTrainedModel", "True", "__init__", "class", "concat", "concat_or_add_pos", "config", "conv1x1", "d_latents", "decoder", "def", "image_size", "index_dims", "input_preprocessor", "num_channels", "num_labels", "out_channels", "position_encoding_type", "post_init", "prep_type", "project_pos_dim", "self", "spatial_downsample", "super", "trainable", "trainable_position_encoding_kwargs", "trainable_position_encoding_kwargs_decoder", "trainable_position_encoding_kwargs_preprocessor", "use_query_residual"], "perceiver/modeling_perceiver.py:PerceiverForImageClassificationLearned.forward": ["False", "Model", "ModelClassifierOutput", "ModelForImageClassificationLearned", "ModelPreTrainedModel", "None", "Tensor", "ValueError", "You", "and", "attention_mask", "attentions", "bool", "both", "cannot", "class", "config", "cross_attentions", "def", "elif", "else", "forward", "hidden_states", "if", "inputs", "interpolate_pos_encoding", "is", "kwargs", "labels", "logits", "loss", "loss_function", "not", "output", "output_attentions", "output_hidden_states", "outputs", "pixel_values", "r", "raise", "return", "return_dict", "self", "torch", "tuple", "use", "use_return_dict"], "perceiver/modeling_perceiver.py:PerceiverForImageClassificationFourier.__init__": ["False", "Model", "ModelClassificationDecoder", "ModelForImageClassificationFourier", "ModelImagePreprocessor", "ModelModel", "ModelPreTrainedModel", "True", "__init__", "class", "concat_pos", "config", "d_latents", "decoder", "def", "fourier_position_encoding_kwargs", "fourier_position_encoding_kwargs_preprocessor", "index_dims", "input_preprocessor", "max_resolution", "num_bands", "num_channels", "num_labels", "pixels", "post_init", "prep_type", "self", "sine_only", "spatial_downsample", "super", "trainable_position_encoding_kwargs", "trainable_position_encoding_kwargs_decoder", "use_query_residual"], "perceiver/modeling_perceiver.py:PerceiverForImageClassificationFourier.forward": ["Model", "ModelClassifierOutput", "ModelForImageClassificationFourier", "ModelPreTrainedModel", "None", "Tensor", "ValueError", "You", "and", "attention_mask", "attentions", "bool", "both", "cannot", "class", "config", "cross_attentions", "def", "elif", "else", "forward", "hidden_states", "if", "inputs", "is", "kwargs", "labels", "logits", "loss", "loss_function", "not", "output", "output_attentions", "output_hidden_states", "outputs", "pixel_values", "r", "raise", "return", "return_dict", "self", "torch", "tuple", "use", "use_return_dict"], "perceiver/modeling_perceiver.py:PerceiverForImageClassificationConvProcessing.__init__": ["False", "Model", "ModelClassificationDecoder", "ModelForImageClassificationConvProcessing", "ModelImagePreprocessor", "ModelModel", "ModelPreTrainedModel", "True", "__init__", "class", "concat_pos", "config", "conv", "d_latents", "decoder", "def", "fourier", "fourier_position_encoding_kwargs", "fourier_position_encoding_kwargs_preprocessor", "index_dims", "input_preprocessor", "max_resolution", "num_bands", "num_channels", "num_labels", "position_encoding_type", "post_init", "prep_type", "self", "sine_only", "spatial_downsample", "super", "trainable_position_encoding_kwargs", "trainable_position_encoding_kwargs_decoder", "use_query_residual"], "perceiver/modeling_perceiver.py:PerceiverForImageClassificationConvProcessing.forward": ["Model", "ModelClassifierOutput", "ModelForImageClassificationConvProcessing", "ModelPreTrainedModel", "None", "Tensor", "ValueError", "You", "and", "attention_mask", "attentions", "bool", "both", "cannot", "class", "config", "cross_attentions", "def", "elif", "else", "forward", "hidden_states", "if", "inputs", "is", "kwargs", "labels", "logits", "loss", "loss_function", "not", "output", "output_attentions", "output_hidden_states", "outputs", "pixel_values", "r", "raise", "return", "return_dict", "self", "torch", "tuple", "use", "use_return_dict"], "perceiver/modeling_perceiver.py:PerceiverForOpticalFlow.__init__": ["False", "Model", "ModelForOpticalFlow", "ModelImagePreprocessor", "ModelModel", "ModelOpticalFlowDecoder", "ModelPreTrainedModel", "True", "__init__", "class", "concat_pos", "config", "conv_after_patching", "conv_after_patching_in_channels", "decoder", "def", "fourier", "fourier_position_encoding_kwargs", "fourier_position_encoding_kwargs_decoder", "fourier_position_encoding_kwargs_preprocessor", "image_preprocessor", "input_preprocessor", "max_resolution", "num_bands", "num_channels", "output_image_shape", "output_num_channels", "patches", "position_encoding_type", "post_init", "prep_type", "rescale_factor", "self", "sine_only", "spatial_downsample", "super", "temporal_downsample", "train_size", "use_query_residual"], "perceiver/modeling_perceiver.py:PerceiverForOpticalFlow.forward": ["Model", "ModelClassifierOutput", "ModelForOpticalFlow", "ModelPreTrainedModel", "None", "NotImplementedError", "Optical", "Tensor", "attention_mask", "attentions", "bool", "class", "config", "cross_attentions", "def", "else", "flow", "forward", "hidden_states", "if", "inputs", "is", "kwargs", "labels", "logits", "loss", "not", "output", "output_attentions", "output_hidden_states", "outputs", "r", "raise", "return", "return_dict", "self", "supported", "torch", "training", "tuple", "use_return_dict", "yet"], "perceiver/modeling_perceiver.py:PerceiverForMultimodalAutoencoding.__init__": ["False", "Model", "ModelAudioPostprocessor", "ModelAudioPreprocessor", "ModelBasicDecoder", "ModelBasicVideoAutoencodingDecoder", "ModelClassificationDecoder", "ModelClassificationPostprocessor", "ModelConfig", "ModelForMultimodalAutoencoding", "ModelImagePreprocessor", "ModelModel", "ModelMultimodalDecoder", "ModelMultimodalPostprocessor", "ModelMultimodalPreprocessor", "ModelOneHotPreprocessor", "ModelPreTrainedModel", "ModelProjectionPostprocessor", "None", "True", "__init__", "_label_trainable_num_channels", "audio", "audio_samples_per_frame", "class", "concat_pos", "concat_preprocessed_input", "config", "decoder", "def", "fourier", "fourier_position_encoding_kwargs", "image", "image_decoder", "image_size", "in_channels", "index_dims", "input_preprocessor", "label", "mask_probs", "max_resolution", "min_padding_size", "modalities", "n_audio_samples", "num_bands", "num_channels", "num_frames", "num_outputs", "out_channels", "output_index_dims", "output_num_channels", "output_postprocessor", "output_shape", "patches", "position_encoding_only", "position_encoding_type", "post_init", "prep_type", "samples_per_patch", "self", "sine_only", "spatial_downsample", "super", "temporal_downsample", "trainable", "trainable_position_encoding_kwargs", "use_query_residual"], "perceiver/modeling_perceiver.py:PerceiverForMultimodalAutoencoding.forward": ["Model", "ModelClassifierOutput", "ModelForMultimodalAutoencoding", "ModelPreTrainedModel", "Multimodal", "None", "NotImplementedError", "Tensor", "attention_mask", "attentions", "autoencoding", "bool", "class", "config", "cross_attentions", "def", "dict", "else", "forward", "hidden_states", "if", "inputs", "is", "kwargs", "labels", "logits", "loss", "not", "output", "output_attentions", "output_hidden_states", "outputs", "r", "raise", "return", "return_dict", "self", "str", "subsampled_output_points", "supported", "torch", "training", "tuple", "use_return_dict", "yet"], "perceiver/modeling_perceiver.py:build_position_encoding": ["Identity", "Linear", "Make", "ModelFourierPositionEncoding", "ModelTrainablePositionEncoding", "Model_position_encoding", "None", "Unknown", "ValueError", "def", "elif", "else", "encoding", "f", "fourier", "fourier_position_encoding_kwargs", "if", "nn", "not", "out_channels", "output_pos_enc", "pass", "position", "position_encoding_type", "positions_projection", "project_pos_dim", "raise", "return", "sure", "to", "trainable", "trainable_position_encoding_kwargs", "type"], "perceiver/modeling_perceiver.py:PerceiverAbstractDecoder.decoder_query": ["ABCMeta", "ModelAbstractDecoder", "Module", "None", "NotImplementedError", "abc", "class", "decoder_query", "def", "inputs", "inputs_without_pos", "metaclass", "modality_sizes", "nn", "raise", "self", "subsampled_points"], "perceiver/modeling_perceiver.py:PerceiverAbstractDecoder.num_query_channels": ["ABCMeta", "ModelAbstractDecoder", "Module", "NotImplementedError", "abc", "class", "def", "metaclass", "nn", "num_query_channels", "raise", "self"], "perceiver/modeling_perceiver.py:PerceiverAbstractDecoder.forward": ["ABCMeta", "ModelAbstractDecoder", "Module", "None", "NotImplementedError", "abc", "class", "def", "forward", "metaclass", "nn", "query", "query_mask", "raise", "self", "z"], "perceiver/modeling_perceiver.py:PerceiverProjectionDecoder.__init__": ["Linear", "ModelAbstractDecoder", "ModelProjectionDecoder", "__init__", "class", "classifier", "config", "d_latents", "def", "nn", "num_labels", "self", "super"], "perceiver/modeling_perceiver.py:PerceiverProjectionDecoder.decoder_query": ["ModelAbstractDecoder", "ModelProjectionDecoder", "None", "class", "decoder_query", "def", "inputs", "inputs_without_pos", "modality_sizes", "return", "self", "subsampled_points"], "perceiver/modeling_perceiver.py:PerceiverProjectionDecoder.forward": ["FloatTensor", "ModelAbstractDecoder", "ModelProjectionDecoder", "None", "Tensor", "class", "classifier", "def", "dim", "forward", "logits", "mean", "query", "query_mask", "return", "self", "torch", "z"], "perceiver/modeling_perceiver.py:PerceiverBasicDecoder.__init__": ["False", "Identity", "Linear", "ModelAbstractDecoder", "ModelBasicDecoder", "ModelConfig", "ModelLayer", "None", "True", "__init__", "bool", "build_position_encoding", "class", "concat_preprocessed_input", "config", "d_latents", "decoding_cross_attention", "def", "else", "final_layer", "final_project", "if", "int", "is", "is_cross_attention", "kv_dim", "nn", "none", "not", "num_channels", "num_heads", "output_index_dims", "output_num_channels", "output_position_encodings", "position_encoding_kwargs", "position_encoding_only", "position_encoding_type", "positions_projection", "q_dim", "qk_channels", "self", "str", "subsampled_index_dims", "super", "trainable", "use_query_residual", "v_channels", "widening_factor"], "perceiver/modeling_perceiver.py:PerceiverBasicDecoder.num_query_channels": ["ModelAbstractDecoder", "ModelBasicDecoder", "ValueError", "You", "calculate", "cannot", "channels", "class", "decoder", "def", "final_project", "if", "in", "int", "is", "none", "num_channels", "num_query_channels", "number", "of", "output_num_channels", "output_position_encodings", "output_size", "position_encoding_kwargs", "position_encoding_only", "position_encoding_type", "project_pos_dim", "query", "raise", "return", "self", "set", "to", "when"], "perceiver/modeling_perceiver.py:PerceiverBasicDecoder.decoder_query": ["ModelAbstractDecoder", "ModelBasicDecoder", "None", "True", "Value", "ValueError", "You", "batch_size", "broadcast_to", "cannot", "cat", "class", "concat_preprocessed_input", "construct", "cpu", "decoder", "decoder_query", "def", "device", "dim", "dtype", "elif", "else", "for", "fourier", "from_numpy", "if", "in", "index_dims", "indices", "inputs", "inputs_without_pos", "is", "modality_sizes", "none", "not", "np", "output_index_dims", "output_position_encodings", "pos", "pos_emb", "position_encoding_type", "positions_projection", "queries", "raise", "required", "reshape", "return", "self", "set", "shape", "stack", "subsampled_points", "tensor", "to", "torch", "trainable", "unravel_index", "when", "x"], "perceiver/modeling_perceiver.py:PerceiverBasicDecoder.forward": ["False", "FloatTensor", "ModelAbstractDecoder", "ModelBasicDecoder", "ModelDecoderOutput", "None", "Tensor", "attention_mask", "bool", "class", "cross_attentions", "decoding_cross_attention", "def", "else", "final_layer", "forward", "if", "inputs", "inputs_mask", "layer_outputs", "logits", "output", "output_attentions", "query", "query_mask", "return", "self", "torch", "z"], "perceiver/modeling_perceiver.py:PerceiverClassificationDecoder.__init__": ["ModelAbstractDecoder", "ModelBasicDecoder", "ModelClassificationDecoder", "__init__", "class", "config", "decoder", "decoder_kwargs", "def", "num_labels", "output_index_dims", "output_num_channels", "self", "super"], "perceiver/modeling_perceiver.py:PerceiverClassificationDecoder.num_query_channels": ["ModelAbstractDecoder", "ModelClassificationDecoder", "class", "decoder", "def", "int", "num_query_channels", "return", "self"], "perceiver/modeling_perceiver.py:PerceiverClassificationDecoder.decoder_query": ["ModelAbstractDecoder", "ModelClassificationDecoder", "None", "class", "decoder", "decoder_query", "def", "inputs", "inputs_without_pos", "modality_sizes", "return", "self", "subsampled_points"], "perceiver/modeling_perceiver.py:PerceiverClassificationDecoder.forward": ["False", "FloatTensor", "ModelAbstractDecoder", "ModelClassificationDecoder", "ModelDecoderOutput", "None", "Tensor", "bool", "class", "cross_attentions", "decoder", "decoder_outputs", "def", "forward", "logits", "output_attentions", "query", "query_mask", "return", "self", "torch", "z"], "perceiver/modeling_perceiver.py:PerceiverOpticalFlowDecoder.__init__": ["ModelAbstractDecoder", "ModelBasicDecoder", "ModelOpticalFlowDecoder", "__init__", "class", "config", "decoder", "decoder_kwargs", "def", "output_image_shape", "output_num_channels", "rescale_factor", "self", "super"], "perceiver/modeling_perceiver.py:PerceiverOpticalFlowDecoder.num_query_channels": ["ModelAbstractDecoder", "ModelOpticalFlowDecoder", "class", "decoder", "def", "int", "num_query_channels", "return", "self"], "perceiver/modeling_perceiver.py:PerceiverOpticalFlowDecoder.decoder_query": ["FlowDecoder", "ModelAbstractDecoder", "ModelOpticalFlowDecoder", "None", "ValueError", "class", "decoder_query", "def", "doesn", "if", "inputs", "inputs_without_pos", "is", "modality_sizes", "not", "raise", "return", "self", "subsampled_points", "subsampling", "support", "t", "yet"], "perceiver/modeling_perceiver.py:PerceiverOpticalFlowDecoder.forward": ["False", "FloatTensor", "ModelAbstractDecoder", "ModelDecoderOutput", "ModelOpticalFlowDecoder", "None", "Tensor", "bool", "class", "cross_attentions", "decoder", "decoder_outputs", "def", "forward", "list", "logits", "output_attentions", "output_image_shape", "preds", "query", "query_mask", "rescale_factor", "reshape", "return", "self", "shape", "torch", "z"], "perceiver/modeling_perceiver.py:PerceiverBasicVideoAutoencodingDecoder.__init__": ["Expected", "ModelAbstractDecoder", "ModelBasicDecoder", "ModelBasicVideoAutoencodingDecoder", "ModelConfig", "None", "ValueError", "__init__", "class", "config", "decoder", "decoder_kwargs", "def", "f", "got", "if", "int", "len", "list", "output_index_dims", "output_num_channels", "output_shape", "position_encoding_type", "raise", "rank", "self", "str", "super"], "perceiver/modeling_perceiver.py:PerceiverBasicVideoAutoencodingDecoder.num_query_channels": ["ModelAbstractDecoder", "ModelBasicVideoAutoencodingDecoder", "class", "decoder", "def", "int", "num_query_channels", "return", "self"], "perceiver/modeling_perceiver.py:PerceiverBasicVideoAutoencodingDecoder.decoder_query": ["ModelAbstractDecoder", "ModelBasicVideoAutoencodingDecoder", "None", "class", "decoder", "decoder_query", "def", "inputs", "inputs_without_pos", "modality_sizes", "return", "self", "subsampled_points"], "perceiver/modeling_perceiver.py:PerceiverBasicVideoAutoencodingDecoder.forward": ["FloatTensor", "ModelAbstractDecoder", "ModelBasicVideoAutoencodingDecoder", "ModelDecoderOutput", "None", "Tensor", "class", "cross_attentions", "decoder", "decoder_outputs", "def", "forward", "logits", "output_shape", "query", "query_mask", "reshape", "return", "self", "shape", "torch", "z"], "perceiver/modeling_perceiver.py:restructure": ["Mapping", "ModalitySizeType", "Model", "Tensor", "def", "for", "in", "index", "inp", "inputs", "keys", "modality", "modality_sizes", "outputs", "return", "size", "sorted", "str", "torch"], "perceiver/modeling_perceiver.py:PerceiverMultimodalDecoder.__init__": ["ModelAbstractDecoder", "ModelBasicDecoder", "ModelConfig", "ModelMultimodalDecoder", "ModuleDict", "None", "Parameter", "ParameterDict", "__init__", "class", "config", "decoder", "decoder_kwargs", "def", "dict", "for", "in", "int", "items", "min_padding_size", "modalities", "modality", "nn", "none", "num_channels", "num_outputs", "num_query_channels", "output_index_dims", "output_num_channels", "padding", "position_encoding_type", "randn", "self", "str", "subsampled_index_dims", "super", "torch"], "perceiver/modeling_perceiver.py:PerceiverMultimodalDecoder.num_query_channels": ["ModelAbstractDecoder", "ModelMultimodalDecoder", "_", "class", "common_channel_size", "decoder", "def", "for", "in", "int", "items", "max", "max_channel_size", "min_padding_size", "modalities", "num_query_channels", "return", "self"], "perceiver/modeling_perceiver.py:PerceiverMultimodalDecoder.decoder_query": ["ModelAbstractDecoder", "ModelMultimodalDecoder", "None", "broadcast_to", "cat", "class", "decoder", "decoder_queries", "decoder_query", "def", "dim", "embed", "for", "get", "if", "in", "input_without_pos", "inputs", "inputs_without_pos", "is", "items", "keys", "modalities", "modality", "modality_sizes", "not", "np", "num_query_channels", "or", "padding", "pos", "prod", "query", "reshape", "restructure", "return", "self", "shape", "sorted", "subsampled_points", "torch", "x"], "perceiver/modeling_perceiver.py:PerceiverMultimodalDecoder.forward": ["False", "FloatTensor", "ModelAbstractDecoder", "ModelMultimodalDecoder", "None", "Tensor", "bool", "class", "decoder", "decoder_outputs", "def", "forward", "output_attentions", "query", "query_mask", "return", "self", "torch", "z"], "perceiver/modeling_perceiver.py:space_to_depth": ["Frames", "Model_to_depth", "Tensor", "ValueError", "batch", "batch_size", "be", "channels", "contiguous", "def", "elif", "else", "frames", "height", "if", "int", "len", "num_channels", "of", "or", "permute", "raise", "rank", "return", "shape", "should", "spatial_block_size", "temporal_block_size", "time", "torch", "view", "width"], "perceiver/modeling_perceiver.py:Conv2dSamePadding.__init__": ["Model", "ModelSamePadding", "ZeroPad2d", "__add__", "__init__", "args", "class", "def", "for", "in", "k", "kernel_size", "kwargs", "nn", "reduce", "self", "super", "zero_pad_2d"], "perceiver/modeling_perceiver.py:Conv2dSamePadding.forward": ["Model", "ModelSamePadding", "_conv_forward", "bias", "class", "def", "forward", "input", "nn", "return", "self", "weight", "zero_pad_2d"], "perceiver/modeling_perceiver.py:Conv2DDownsample.__init__": ["BatchNorm2d", "False", "Identity", "MaxPool2d", "Model", "ModelDDownsample", "ModeldSamePadding", "Module", "ReLU", "True", "__init__", "batchnorm", "bias", "bool", "class", "def", "else", "if", "in_channels", "int", "kernel_size", "max_pool", "nn", "num_features", "num_layers", "out_channels", "relu", "self", "stride", "super", "use_batchnorm"], "perceiver/modeling_perceiver.py:Conv2DDownsample.forward": ["Model", "ModelDDownsample", "Module", "Tensor", "batchnorm", "class", "def", "forward", "inputs", "max_pool", "nn", "out", "relu", "return", "self", "torch"], "perceiver/modeling_perceiver.py:generate_fourier_features": ["False", "Model_fourier_features", "None", "True", "batch_size", "cat", "concat_pos", "cos", "def", "dim", "else", "end", "expand", "for", "freq_bands", "if", "in", "linspace", "max_resolution", "min_freq", "np", "num_bands", "per_pos_features", "pi", "pos", "prod", "res", "reshape", "return", "shape", "sin", "sine_only", "stack", "start", "steps", "torch"], "perceiver/modeling_perceiver.py:build_linear_positions": ["Model_linear_positions", "_linspace", "array_index_grid", "def", "dim", "dim_ranges", "dtype", "end", "float32", "for", "ij", "in", "index_dims", "indexing", "linspace", "meshgrid", "n_xels_per_dim", "output_range", "return", "stack", "start", "steps", "torch"], "perceiver/modeling_perceiver.py:PerceiverAbstractPositionEncoding.num_dimensions": ["ABCMeta", "ModelAbstractPositionEncoding", "Module", "NotImplementedError", "abc", "class", "def", "int", "metaclass", "nn", "num_dimensions", "raise", "self"], "perceiver/modeling_perceiver.py:PerceiverAbstractPositionEncoding.output_size": ["ABCMeta", "ModelAbstractPositionEncoding", "Module", "NotImplementedError", "abc", "args", "class", "def", "int", "kwargs", "metaclass", "nn", "output_size", "raise", "self"], "perceiver/modeling_perceiver.py:PerceiverAbstractPositionEncoding.forward": ["ABCMeta", "ModelAbstractPositionEncoding", "Module", "NotImplementedError", "abc", "batch_size", "class", "def", "forward", "metaclass", "nn", "pos", "raise", "self"], "perceiver/modeling_perceiver.py:PerceiverTrainablePositionEncoding.__init__": ["ModelAbstractPositionEncoding", "ModelTrainablePositionEncoding", "Parameter", "__init__", "_index_dims", "_num_channels", "class", "def", "index_dim", "index_dims", "nn", "np", "num_channels", "position_embeddings", "prod", "randn", "self", "super", "torch"], "perceiver/modeling_perceiver.py:PerceiverTrainablePositionEncoding.num_dimensions": ["ModelAbstractPositionEncoding", "ModelTrainablePositionEncoding", "_index_dims", "class", "def", "if", "int", "isinstance", "len", "num_dimensions", "return", "self"], "perceiver/modeling_perceiver.py:PerceiverTrainablePositionEncoding.output_size": ["ModelAbstractPositionEncoding", "ModelTrainablePositionEncoding", "_num_channels", "args", "class", "def", "int", "kwargs", "output_size", "return", "self"], "perceiver/modeling_perceiver.py:PerceiverTrainablePositionEncoding.interpolate_pos_encoding": ["False", "ModelAbstractPositionEncoding", "ModelTrainablePositionEncoding", "Tensor", "_num_channels", "align_corners", "and", "bicubic", "class", "def", "functional", "height", "if", "int", "interpolate", "interpolate_pos_encoding", "is_tracing", "jit", "mode", "new_height", "new_width", "nn", "not", "num_positions", "permute", "position_embeddings", "reshape", "return", "self", "shape", "size", "squeeze", "torch", "torch_int", "width"], "perceiver/modeling_perceiver.py:PerceiverTrainablePositionEncoding.forward": ["False", "ModelAbstractPositionEncoding", "ModelTrainablePositionEncoding", "None", "Size", "Tensor", "batch_size", "bool", "class", "def", "expand", "forward", "height", "if", "input_size", "int", "interpolate_pos_encoding", "is", "not", "position_embeddings", "return", "self", "torch", "width"], "perceiver/modeling_perceiver.py:_check_or_build_spatial_positions": ["None", "Spatial", "ValueError", "_check_or_build_spatial_positions", "batch_size", "build_linear_positions", "def", "dimensions", "else", "expand", "features", "have", "if", "index_dims", "is", "len", "np", "number", "of", "pos", "prod", "raise", "reshape", "return", "shape", "the", "torch", "wrong"], "perceiver/modeling_perceiver.py:PerceiverFourierPositionEncoding.__init__": ["False", "ModelAbstractPositionEncoding", "ModelFourierPositionEncoding", "True", "__init__", "class", "concat_pos", "def", "max_resolution", "num_bands", "self", "sine_only", "super"], "perceiver/modeling_perceiver.py:PerceiverFourierPositionEncoding.num_dimensions": ["ModelAbstractPositionEncoding", "ModelFourierPositionEncoding", "class", "def", "int", "len", "max_resolution", "num_dimensions", "return", "self"], "perceiver/modeling_perceiver.py:PerceiverFourierPositionEncoding.output_size": ["ModelAbstractPositionEncoding", "ModelFourierPositionEncoding", "class", "concat_pos", "def", "encoding_size", "if", "len", "max_resolution", "not", "num_bands", "num_dimensions", "num_dims", "output_size", "return", "self", "sine_only"], "perceiver/modeling_perceiver.py:PerceiverFourierPositionEncoding.forward": ["FloatTensor", "ModelAbstractPositionEncoding", "ModelFourierPositionEncoding", "None", "_check_or_build_spatial_positions", "batch_size", "class", "concat_pos", "def", "device", "dtype", "forward", "fourier_pos_enc", "generate_fourier_features", "index_dims", "int", "list", "max_resolution", "num_bands", "pos", "return", "self", "sine_only", "to", "torch"], "perceiver/modeling_perceiver.py:AbstractPreprocessor.num_channels": ["ModelPreprocessor", "Module", "NotImplementedError", "class", "def", "int", "nn", "num_channels", "raise", "self"], "perceiver/modeling_perceiver.py:PerceiverTextPreprocessor.__init__": ["AbstractPreprocessor", "Embedding", "ModelConfig", "ModelTextPreprocessor", "None", "__init__", "class", "config", "d_model", "def", "embedding_dim", "embeddings", "max_position_embeddings", "nn", "num_embeddings", "position_embeddings", "self", "super", "vocab_size"], "perceiver/modeling_perceiver.py:PerceiverTextPreprocessor.num_channels": ["AbstractPreprocessor", "ModelTextPreprocessor", "class", "config", "d_model", "def", "int", "num_channels", "return", "self"], "perceiver/modeling_perceiver.py:PerceiverTextPreprocessor.forward": ["AbstractPreprocessor", "False", "LongTensor", "ModelTextPreprocessor", "None", "Tensor", "True", "arange", "bool", "class", "def", "device", "embeddings", "embeddings_without_pos", "forward", "inputs", "interpolate_pos_encoding", "network_input_is_1d", "pos", "position_embeddings", "position_ids", "return", "self", "seq_length", "shape", "torch"], "perceiver/modeling_perceiver.py:PerceiverEmbeddingDecoder.__init__": ["ModelConfig", "ModelEmbeddingDecoder", "Module", "None", "Parameter", "__init__", "bias", "class", "config", "def", "nn", "self", "super", "torch", "vocab_size", "zeros"], "perceiver/modeling_perceiver.py:PerceiverEmbeddingDecoder.forward": ["ModelEmbeddingDecoder", "Module", "Tensor", "batch_size", "bias", "class", "d_model", "def", "embedding_layer", "forward", "hidden_states", "matmul", "nn", "output", "reshape", "return", "self", "seq_len", "shape", "torch", "transpose", "vocab_size", "weight"], "perceiver/modeling_perceiver.py:PerceiverMultimodalPostprocessor.__init__": ["False", "Mapping", "ModelMultimodalPostprocessor", "Module", "ModuleDict", "PostprocessorType", "__init__", "bool", "class", "def", "input_is_dict", "modalities", "nn", "self", "str", "super"], "perceiver/modeling_perceiver.py:PerceiverMultimodalPostprocessor.forward": ["Mapping", "Modality", "ModelMultimodalPostprocessor", "Module", "None", "Tensor", "ValueError", "a", "be", "class", "def", "dictionary", "for", "forward", "if", "in", "input", "input_is_dict", "inputs", "is", "items", "modalities", "modality", "modality_sizes", "nn", "not", "outputs", "pos", "postprocessor", "raise", "restructure", "return", "self", "should", "sizes", "specified", "str", "torch"], "perceiver/modeling_perceiver.py:PerceiverClassificationPostprocessor.__init__": ["Linear", "ModelClassificationPostprocessor", "ModelConfig", "Module", "None", "__init__", "class", "classifier", "config", "def", "in_channels", "int", "nn", "num_labels", "self", "super"], "perceiver/modeling_perceiver.py:PerceiverClassificationPostprocessor.forward": ["ModelClassificationPostprocessor", "Module", "None", "Tensor", "class", "classifier", "def", "forward", "inputs", "logits", "modality_sizes", "nn", "pos", "return", "self", "torch"], "perceiver/modeling_perceiver.py:PerceiverAudioPostprocessor.__init__": ["Invalid", "Linear", "ModelAudioPostprocessor", "ModelConfig", "Module", "None", "ValueError", "__init__", "class", "classifier", "config", "def", "if", "in_channels", "int", "nn", "patches", "postproc_type", "raise", "samples_per_patch", "self", "str", "super"], "perceiver/modeling_perceiver.py:PerceiverAudioPostprocessor.forward": ["ModelAudioPostprocessor", "Module", "None", "Tensor", "class", "classifier", "def", "forward", "inputs", "logits", "modality_sizes", "nn", "pos", "reshape", "return", "self", "shape", "torch"], "perceiver/modeling_perceiver.py:PerceiverProjectionPostprocessor.__init__": ["Linear", "ModelProjectionPostprocessor", "Module", "None", "__init__", "class", "classifier", "def", "in_channels", "int", "nn", "out_channels", "self", "super"], "perceiver/modeling_perceiver.py:PerceiverProjectionPostprocessor.forward": ["ModelProjectionPostprocessor", "Module", "None", "Tensor", "class", "classifier", "def", "forward", "inputs", "logits", "modality_sizes", "nn", "pos", "return", "self", "torch"], "perceiver/modeling_perceiver.py:PerceiverImagePreprocessor.__init__": ["AbstractPreprocessor", "Conv1x1", "Conv2DDownsample", "Conv2d", "False", "Identity", "Invalid", "Linear", "ModelImagePreprocessor", "Only", "Prep_type", "True", "ValueError", "__init__", "add", "and", "bool", "build_position_encoding", "class", "concat", "concat_or_add_pos", "config", "conv", "conv1x1", "conv2d_use_batchnorm", "conv_after_patches", "conv_after_patching", "conv_after_patching_in_channels", "convnet", "convnet_1x1", "convnet_num_layers", "convnet_num_layers_is_int", "def", "does", "downsample", "downsampling", "elif", "else", "expected", "f", "for", "fourier", "if", "in", "in_channels", "int", "invalid", "is", "kernel_size", "log", "math", "nn", "not", "np", "num_layers", "of", "or", "out_channels", "patches", "pixels", "position_embeddings", "position_encoding_kwargs", "position_encoding_type", "positions_projection", "powers", "prep_type", "project_pos_dim", "raise", "round", "self", "spatial", "spatial_downsample", "str", "stride", "super", "temporal", "temporal_downsample", "time", "use_batchnorm", "value", "with"], "perceiver/modeling_perceiver.py:PerceiverImagePreprocessor.num_channels": ["AbstractPreprocessor", "ModelImagePreprocessor", "add", "ceil", "class", "concat_or_add_pos", "conv", "conv1x1", "conv_after_patching", "def", "elif", "else", "if", "in", "in_channels", "inp_dim", "int", "is_temporal", "math", "not", "num_channels", "num_dimensions", "or", "out_channels", "output_size", "patches", "pixels", "pos_dim", "position_embeddings", "prep_type", "project_pos_dim", "return", "self", "spatial_downsample", "temporal_downsample"], "perceiver/modeling_perceiver.py:PerceiverImagePreprocessor._build_network_inputs": ["AbstractPreprocessor", "False", "ModelImagePreprocessor", "Tensor", "True", "_build_network_inputs", "add", "and", "batch_size", "bool", "cat", "class", "concat", "concat_or_add_pos", "def", "device", "dim", "dtype", "elif", "fourier", "if", "index_dims", "indices", "input_size", "inputs", "inputs_with_pos", "interpolate_pos_encoding", "len", "list", "network_input_is_1d", "not", "np", "pos_enc", "position_embeddings", "position_encoding_type", "positions_projection", "prod", "reshape", "return", "self", "sh", "shape", "torch", "trainable"], "perceiver/modeling_perceiver.py:PerceiverImagePreprocessor.forward": ["AbstractPreprocessor", "False", "ModelImagePreprocessor", "None", "Tensor", "True", "Unsupported", "ValueError", "_build_network_inputs", "and", "bool", "class", "conv", "conv1x1", "conv_after_patches", "convnet", "convnet_1x1", "data", "def", "dim", "elif", "else", "for", "format", "forward", "if", "inputs", "inputs_without_pos", "interpolate_pos_encoding", "modality_sizes", "ndim", "network_input_is_1d", "patches", "permute", "pixels", "pos", "prep_type", "raise", "return", "self", "shape", "space_to_depth", "spatial_block_size", "spatial_downsample", "squeeze", "temporal_block_size", "temporal_downsample", "torch"], "perceiver/modeling_perceiver.py:PerceiverOneHotPreprocessor.__init__": ["AbstractPreprocessor", "ModelConfig", "ModelOneHotPreprocessor", "None", "__init__", "class", "config", "def", "self", "super"], "perceiver/modeling_perceiver.py:PerceiverOneHotPreprocessor.num_channels": ["AbstractPreprocessor", "ModelOneHotPreprocessor", "class", "config", "def", "int", "num_channels", "num_labels", "return", "self"], "perceiver/modeling_perceiver.py:PerceiverOneHotPreprocessor.forward": ["AbstractPreprocessor", "ModelOneHotPreprocessor", "None", "Tensor", "True", "bool", "class", "def", "forward", "inputs", "network_input_is_1d", "pos", "return", "self", "torch"], "perceiver/modeling_perceiver.py:PerceiverAudioPreprocessor.__init__": ["AbstractPreprocessor", "Concat_or_pos", "ModelAudioPreprocessor", "Prep_type", "ValueError", "__init__", "add", "be", "build_position_encoding", "can", "class", "concat", "concat_or_add_pos", "config", "def", "f", "fourier", "if", "in", "int", "invalid", "is", "not", "only", "or", "out_channels", "patches", "position_embeddings", "position_encoding_kwargs", "position_encoding_type", "positions_projection", "prep_type", "project_pos_dim", "raise", "samples_per_patch", "self", "str", "super"], "perceiver/modeling_perceiver.py:PerceiverAudioPreprocessor.num_channels": ["AbstractPreprocessor", "ModelAudioPreprocessor", "add", "class", "concat_or_add_pos", "def", "else", "if", "int", "num_channels", "output_size", "pos_dim", "position_embeddings", "project_pos_dim", "return", "samples_per_patch", "self"], "perceiver/modeling_perceiver.py:PerceiverAudioPreprocessor._build_network_inputs": ["AbstractPreprocessor", "ModelAudioPreprocessor", "_build_network_inputs", "add", "batch_size", "cat", "class", "concat", "concat_or_add_pos", "def", "device", "dim", "dtype", "elif", "fourier", "if", "index_dims", "inputs", "inputs_with_pos", "pos_enc", "position_embeddings", "position_encoding_type", "positions_projection", "return", "self", "shape", "torch", "trainable"], "perceiver/modeling_perceiver.py:PerceiverAudioPreprocessor.forward": ["AbstractPreprocessor", "False", "ModelAudioPreprocessor", "None", "Tensor", "True", "_build_network_inputs", "bool", "class", "def", "forward", "inputs", "inputs_without_pos", "interpolate_pos_encoding", "modality_sizes", "network_input_is_1d", "pos", "reshape", "return", "samples_per_patch", "self", "shape", "torch"], "perceiver/modeling_perceiver.py:PerceiverMultimodalPreprocessor.__init__": ["AbstractPreprocessor", "Mapping", "ModelMultimodalPreprocessor", "ModuleDict", "None", "Parameter", "ParameterDict", "PreprocessorType", "_", "__init__", "class", "def", "else", "float", "for", "if", "in", "int", "is", "items", "mask", "mask_probs", "min_padding_size", "modalities", "modality", "nn", "not", "num_channels", "padding", "preprocessor", "randn", "self", "str", "super", "torch"], "perceiver/modeling_perceiver.py:PerceiverMultimodalPreprocessor.num_channels": ["AbstractPreprocessor", "ModelMultimodalPreprocessor", "_", "class", "common_channel_size", "def", "for", "in", "int", "items", "max", "max_channel_size", "min_padding_size", "modalities", "num_channels", "processor", "return", "self"], "perceiver/modeling_perceiver.py:PerceiverMultimodalPreprocessor.forward": ["AbstractPreprocessor", "False", "Mapping", "ModelMultimodalPreprocessor", "None", "PreprocessorOutputType", "Tensor", "True", "_", "batch_size", "bernoulli", "bool", "broadcast_to", "cat", "class", "def", "device", "dim", "expand", "final_inputs", "for", "forward", "full", "if", "in", "inputs", "inputs_without_pos", "interpolate_pos_encoding", "items", "k", "keys", "mask", "mask_prob", "mask_probs", "mask_token", "modalities", "modality", "modality_sizes", "network_input_is_1d", "num_channels", "num_samples", "output", "output_padded", "padded", "padded_ls", "padding", "pos", "pos_enc", "preprocessor", "return", "self", "shape", "sorted", "str", "to", "torch", "unsqueeze"], "perception_lm/modeling_perception_lm.py:PerceptionLMAdaptiveAvgPooling.__init__": ["ModelAdaptiveAvgPooling", "Module", "__init__", "class", "def", "nn", "pooling_ratio", "self", "super"], "perception_lm/modeling_perception_lm.py:PerceptionLMAdaptiveAvgPooling.forward": ["F", "ModelAdaptiveAvgPooling", "Module", "ValueError", "a", "adaptive_avg_pool2d", "b", "be", "c", "class", "def", "expected", "f", "flatten", "forward", "h", "hidden_states", "if", "int", "is", "math", "nn", "num_tokens", "number", "permute", "pooling_ratio", "raise", "reshape", "return", "self", "shape", "sqrt", "square", "to", "transpose"], "perception_lm/modeling_perception_lm.py:PerceptionLMMultiModalProjector.__init__": ["GELU", "Identity", "Linear", "ModelAdaptiveAvgPooling", "ModelConfig", "ModelMultiModalProjector", "Module", "True", "__init__", "bias", "class", "config", "def", "else", "embed_dim", "gelu", "hidden_size", "if", "in_features", "input_size", "linear_1", "linear_2", "model_args", "nn", "out_features", "output_size", "pooling", "projector_pooling_ratio", "self", "super", "text_config", "vision_config"], "perception_lm/modeling_perception_lm.py:PerceptionLMMultiModalProjector.forward": ["ModelMultiModalProjector", "Module", "class", "def", "features", "forward", "gelu", "linear_1", "linear_2", "nn", "permute", "pooling", "return", "self"], "perception_lm/modeling_perception_lm.py:PerceptionLMModel.__init__": ["AutoModel", "ModelConfig", "ModelModel", "ModelMultiModalProjector", "ModelPreTrainedModel", "__init__", "class", "config", "def", "from_config", "language_model", "multi_modal_projector", "post_init", "self", "super", "text_config", "vision_config", "vision_tower"], "perception_lm/modeling_perception_lm.py:PerceptionLMModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "language_model", "return", "self"], "perception_lm/modeling_perception_lm.py:PerceptionLMModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "language_model", "self", "set_input_embeddings", "value"], "perception_lm/modeling_perception_lm.py:PerceptionLMModel.get_image_features": ["FloatTensor", "ModelModel", "ModelPreTrainedModel", "class", "config", "def", "flatten", "get_image_features", "if", "image_features", "image_outputs", "kwargs", "last_hidden_state", "multi_modal_projector", "pixel_values", "return", "self", "torch", "vision_tower", "vision_use_cls_token"], "perception_lm/modeling_perception_lm.py:PerceptionLMModel.get_placeholder_mask": ["FloatTensor", "Image", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "ValueError", "Videos", "all", "and", "class", "config", "def", "device", "do", "dtype", "else", "expand_as", "f", "features", "get_input_embeddings", "get_placeholder_mask", "if", "image", "image_features", "image_token_id", "input_ids", "inputs_embeds", "is", "long", "match", "n_image_tokens", "n_video_tokens", "not", "numel", "raise", "return", "self", "size", "special_image_mask", "special_video_mask", "sum", "tensor", "to", "tokens", "torch", "unsqueeze", "video_features", "video_token_id"], "perception_lm/modeling_perception_lm.py:PerceptionLMModel.forward": ["Cache", "FloatTensor", "LongTensor", "ModelModel", "ModelModelOutputWithPast", "ModelPreTrainedModel", "None", "Tensor", "True", "ValueError", "You", "_", "and", "at", "attention_mask", "attentions", "bool", "both", "cache_position", "cannot", "class", "config", "def", "device", "dtype", "either", "else", "exactly", "forward", "get_image_features", "get_input_embeddings", "get_placeholder_mask", "hidden_states", "if", "image_features", "image_hidden_states", "input_ids", "inputs_embeds", "int", "is", "language_model", "last_hidden_state", "lm_kwargs", "logits_to_keep", "masked_scatter", "must", "not", "of", "one", "or", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "pixel_values", "pixel_values_videos", "position_ids", "raise", "return", "return_dict", "same", "self", "special_image_mask", "special_video_mask", "specify", "the", "time", "to", "torch", "tuple", "use_cache", "video_features", "video_hidden_states"], "perception_lm/modeling_perception_lm.py:PerceptionLMForConditionalGeneration.__init__": ["False", "GenerationMixin", "Linear", "ModelConfig", "ModelForConditionalGeneration", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "text_config", "vocab_size"], "perception_lm/modeling_perception_lm.py:PerceptionLMForConditionalGeneration.get_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "model", "return", "self"], "perception_lm/modeling_perception_lm.py:PerceptionLMForConditionalGeneration.set_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "model", "self", "set_input_embeddings", "value"], "perception_lm/modeling_perception_lm.py:PerceptionLMForConditionalGeneration.get_output_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "Module", "class", "def", "get_output_embeddings", "lm_head", "nn", "return", "self"], "perception_lm/modeling_perception_lm.py:PerceptionLMForConditionalGeneration.forward": ["Cache", "FloatTensor", "GenerationMixin", "LongTensor", "ModelCausalLMOutputWithPast", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Tensor", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "else", "forward", "hidden_states", "if", "image_hidden_states", "input_ids", "inputs_embeds", "int", "is", "isinstance", "labels", "lm_head", "lm_kwargs", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "pixel_values", "pixel_values_videos", "position_ids", "r", "return", "self", "slice", "slice_indices", "text_config", "torch", "tuple", "use_cache", "video_hidden_states", "vocab_size"], "perception_lm/modeling_perception_lm.py:PerceptionLMForConditionalGeneration.prepare_inputs_for_generation": ["False", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "True", "attention_mask", "cache_position", "class", "def", "get", "if", "input_ids", "inputs_embeds", "is_first_iteration", "kwargs", "logits_to_keep", "model_inputs", "not", "or", "past_key_values", "pixel_values", "pixel_values_videos", "prepare_inputs_for_generation", "return", "self", "super", "use_cache"], "persimmon/modeling_persimmon.py:PersimmonRotaryEmbedding.__init__": ["Callable", "False", "ModelConfig", "ModelRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "persimmon/modeling_persimmon.py:PersimmonRotaryEmbedding.compute_default_rope_parameters": ["ModelConfig", "ModelRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "get", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "partial_rotary_factor", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "persimmon/modeling_persimmon.py:PersimmonRotaryEmbedding.forward": ["False", "ModelRotaryEmbedding", "Module", "None", "and", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "persimmon/modeling_persimmon.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "persimmon/modeling_persimmon.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cos", "def", "k", "k_embed", "q", "q_embed", "return", "rotate_half", "sin", "unsqueeze", "unsqueeze_dim"], "persimmon/modeling_persimmon.py:PersimmonMLP.__init__": ["ACT2FN", "Linear", "ModelMLP", "Module", "__init__", "act", "class", "config", "def", "dense_4h_to_h", "dense_h_to_4h", "hidden_act", "hidden_size", "intermediate_size", "nn", "self", "super"], "persimmon/modeling_persimmon.py:PersimmonMLP.forward": ["ModelMLP", "Module", "act", "class", "def", "dense_4h_to_h", "dense_h_to_4h", "forward", "hidden_states", "nn", "return", "self"], "persimmon/modeling_persimmon.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value"], "persimmon/modeling_persimmon.py:PersimmonAttention.__init__": ["Dropout", "Instantiating", "LayerNorm", "Linear", "ModelAttention", "ModelConfig", "Module", "None", "Please", "True", "ValueError", "__class__", "__init__", "__name__", "a", "and", "attention_dropout", "be", "bias", "by", "caching", "call", "class", "config", "creating", "def", "dense", "divisible", "during", "elementwise_affine", "eps", "errors", "f", "forward", "got", "head_dim", "hidden_size", "if", "int", "is", "is_causal", "k_layernorm", "layer_idx", "layer_norm_eps", "lead", "logger", "make", "must", "nn", "not", "num_attention_heads", "num_heads", "partial_rotary_factor", "passing", "provide", "q_layernorm", "qk_layernorm", "query_key_value", "raise", "recommended", "rope_parameters", "rotary_ndims", "scaling", "self", "super", "sure", "the", "this", "to", "used", "warning_once", "when", "will", "without"], "persimmon/modeling_persimmon.py:PersimmonAttention._split_heads": ["ModelAttention", "Module", "Tensor", "_split_heads", "batch_size", "class", "def", "fused_qkv", "head_dim", "nn", "num_heads", "return", "self", "seq_length", "shape", "three_times_hidden_size", "torch", "tuple", "view"], "persimmon/modeling_persimmon.py:PersimmonAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "False", "FlashAttentionKwargs", "LongTensor", "ModelAttention", "Module", "None", "Tensor", "Unpack", "_", "_attn_implementation", "_split_heads", "apply_rotary_pos_emb", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "bool", "bsz", "cache_kwargs", "cache_position", "cat", "class", "config", "cos", "def", "dense", "dim", "dropout", "eager", "eager_attention_forward", "else", "forward", "fused_qkv", "hidden_states", "if", "is", "k_layernorm", "key_pass", "key_rot", "key_states", "kwargs", "layer_idx", "nn", "not", "output_attentions", "partial_rotation_size", "past_key_values", "position_embeddings", "position_ids", "q_layernorm", "q_len", "qk_layernorm", "query_key_value", "query_pass", "query_rot", "query_states", "reshape", "return", "rotary_ndims", "scaling", "self", "sin", "size", "torch", "training", "transpose", "tuple", "update", "use_cache", "value_states"], "persimmon/modeling_persimmon.py:PersimmonDecoderLayer.__init__": ["Dropout", "GradientCheckpointingLayer", "LayerNorm", "ModelAttention", "ModelConfig", "ModelDecoderLayer", "ModelMLP", "__init__", "class", "config", "def", "dropout", "eps", "hidden_dropout", "hidden_size", "input_layernorm", "int", "layer_idx", "layer_norm_eps", "mlp", "nn", "post_attention_layernorm", "self", "self_attn", "super"], "persimmon/modeling_persimmon.py:PersimmonDecoderLayer.forward": ["Cache", "False", "FlashAttentionKwargs", "FloatTensor", "GradientCheckpointingLayer", "LongTensor", "ModelDecoderLayer", "None", "Tensor", "Unpack", "attention_mask", "bool", "cache_position", "class", "def", "dropout", "forward", "hidden_states", "if", "input_layernorm", "kwargs", "mlp", "output_attentions", "outputs", "past_key_values", "position_embeddings", "position_ids", "post_attention_layernorm", "residual", "return", "self", "self_attn", "self_attn_weights", "torch", "tuple", "use_cache"], "persimmon/modeling_persimmon.py:PersimmonModel.__init__": ["Embedding", "False", "LayerNorm", "ModelConfig", "ModelDecoderLayer", "ModelModel", "ModelPreTrainedModel", "ModelRotaryEmbedding", "ModuleList", "__init__", "class", "config", "def", "embed_tokens", "eps", "final_layernorm", "for", "gradient_checkpointing", "hidden_size", "in", "layer_idx", "layer_norm_eps", "layers", "nn", "num_hidden_layers", "pad_token_id", "padding_idx", "post_init", "range", "rotary_emb", "self", "super", "vocab_size"], "persimmon/modeling_persimmon.py:PersimmonModel.forward": ["BaseModelOutputWithPast", "Cache", "DynamicCache", "False", "FlashAttentionKwargs", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Setting", "Tensor", "True", "Unpack", "ValueError", "You", "_update_causal_mask", "all_hidden_states", "all_self_attns", "and", "arange", "attention_mask", "attentions", "bool", "cache_position", "causal_mask", "checkpointing", "class", "config", "decoder_layer", "def", "device", "else", "embed_tokens", "exactly", "final_layernorm", "for", "forward", "get_seq_length", "gradient", "gradient_checkpointing", "hidden_states", "if", "in", "incompatible", "input_ids", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layer_outputs", "layers", "logger", "must", "not", "of", "one", "or", "output_attentions", "output_hidden_states", "past_key_values", "past_seen_tokens", "position_embeddings", "position_ids", "raise", "return", "rotary_emb", "self", "shape", "specify", "torch", "training", "unsqueeze", "use_cache", "warning_once", "with"], "persimmon/modeling_persimmon.py:PersimmonModel._update_causal_mask": ["AttentionMaskConverter", "BlockMask", "Cache", "False", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "Union", "_attn_implementation", "_ignore_causal_mask_sdpa", "_prepare_4d_causal_attention_mask_with_cache_position", "_unmask_unattended", "_update_causal_mask", "and", "any", "attention_mask", "batch_size", "bool", "cache_position", "causal_mask", "class", "config", "cuda", "def", "device", "dtype", "else", "finfo", "flash_attention_2", "flex_attention", "get_max_cache_shape", "get_seq_length", "if", "in", "input_tensor", "inputs_embeds", "is", "is_compileable", "is_training", "isinstance", "make_flex_block_causal_mask", "min", "min_dtype", "not", "npu", "output_attentions", "past_key_values", "past_key_values_length", "past_seen_tokens", "return", "sdpa", "self", "sequence_length", "shape", "target_length", "torch", "training", "type", "using_compilable_cache", "xpu"], "persimmon/modeling_persimmon.py:PersimmonModel._prepare_4d_causal_attention_mask_with_cache_position": ["ModelModel", "ModelPreTrainedModel", "None", "Tensor", "_prepare_4d_causal_attention_mask_with_cache_position", "and", "arange", "attention_mask", "batch_size", "cache_position", "causal_mask", "class", "clone", "def", "device", "diagonal", "dim", "dtype", "else", "expand", "fill_value", "finfo", "full", "if", "int", "is", "kwargs", "mask_length", "masked_fill", "min", "min_dtype", "not", "padding_mask", "reshape", "return", "sequence_length", "shape", "target_length", "to", "torch", "triu"], "persimmon/modeling_persimmon.py:PersimmonForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "ModelForCausalLM", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "vocab_size"], "persimmon/modeling_persimmon.py:PersimmonForCausalLM.forward": ["BaseModelOutputWithPast", "Cache", "CausalLMOutputWithPast", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForCausalLM", "ModelPreTrainedModel", "None", "Tensor", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "position_ids", "r", "return", "self", "slice", "slice_indices", "torch", "use_cache", "vocab_size"], "phi/modeling_phi.py:PhiRotaryEmbedding.__init__": ["Callable", "False", "ModelConfig", "ModelRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "phi/modeling_phi.py:PhiRotaryEmbedding.compute_default_rope_parameters": ["ModelConfig", "ModelRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "get", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "partial_rotary_factor", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "phi/modeling_phi.py:PhiRotaryEmbedding.forward": ["False", "ModelRotaryEmbedding", "Module", "None", "and", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "phi/modeling_phi.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "phi/modeling_phi.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cos", "def", "k", "k_embed", "q", "q_embed", "return", "rotate_half", "sin", "unsqueeze", "unsqueeze_dim"], "phi/modeling_phi.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "phi/modeling_phi.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value", "value_states"], "phi/modeling_phi.py:PhiAttention.__init__": ["LayerNorm", "Linear", "ModelAttention", "ModelConfig", "Module", "True", "__init__", "attention_dropout", "bias", "class", "config", "def", "dense", "elementwise_affine", "eps", "getattr", "head_dim", "hidden_size", "if", "int", "is_causal", "k_layernorm", "k_proj", "layer_idx", "layer_norm_eps", "nn", "num_attention_heads", "num_key_value_groups", "num_key_value_heads", "partial_rotary_factor", "q_layernorm", "q_proj", "qk_layernorm", "rope_parameters", "rotary_ndims", "scaling", "self", "super", "v_proj"], "phi/modeling_phi.py:PhiAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "LongTensor", "ModelAttention", "Module", "None", "Tensor", "_attn_implementation", "apply_rotary_pos_emb", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "cache_kwargs", "cache_position", "cat", "class", "config", "contiguous", "cos", "def", "dense", "dim", "dropout", "eager", "eager_attention_forward", "else", "forward", "head_dim", "hidden_shape", "hidden_states", "if", "input_shape", "is", "k_layernorm", "k_proj", "key_pass", "key_rot", "key_states", "kwargs", "layer_idx", "nn", "not", "past_key_values", "position_embeddings", "q_layernorm", "q_proj", "qk_layernorm", "query_pass", "query_rot", "query_states", "reshape", "return", "rotary_ndims", "scaling", "self", "shape", "sin", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "view"], "phi/modeling_phi.py:PhiMLP.__init__": ["ACT2FN", "Linear", "ModelMLP", "Module", "__init__", "activation_fn", "class", "config", "def", "fc1", "fc2", "hidden_act", "hidden_size", "intermediate_size", "nn", "self", "super"], "phi/modeling_phi.py:PhiMLP.forward": ["ModelMLP", "Module", "Tensor", "activation_fn", "class", "def", "fc1", "fc2", "forward", "hidden_states", "nn", "return", "self", "torch"], "phi/modeling_phi.py:PhiDecoderLayer.__init__": ["Dropout", "GradientCheckpointingLayer", "LayerNorm", "ModelAttention", "ModelConfig", "ModelDecoderLayer", "ModelMLP", "__init__", "class", "config", "def", "eps", "hidden_size", "input_layernorm", "int", "layer_idx", "layer_norm_eps", "mlp", "nn", "resid_dropout", "resid_pdrop", "self", "self_attn", "super"], "phi/modeling_phi.py:PhiDecoderLayer.forward": ["Cache", "False", "FloatTensor", "GradientCheckpointingLayer", "LongTensor", "ModelDecoderLayer", "None", "Tensor", "attention_mask", "attn_outputs", "bool", "cache_position", "class", "def", "feed_forward_hidden_states", "forward", "hidden_states", "if", "input_layernorm", "kwargs", "mlp", "output_attentions", "outputs", "past_key_values", "position_embeddings", "position_ids", "resid_dropout", "residual", "return", "self", "self_attn", "self_attn_weights", "torch", "tuple", "use_cache"], "phi/modeling_phi.py:PhiModel.__init__": ["Dropout", "Embedding", "False", "LayerNorm", "ModelConfig", "ModelDecoderLayer", "ModelModel", "ModelPreTrainedModel", "ModelRotaryEmbedding", "ModuleList", "__init__", "class", "config", "def", "embd_pdrop", "embed_dropout", "embed_tokens", "eps", "final_layernorm", "for", "gradient_checkpointing", "hidden_size", "in", "layer_idx", "layer_norm_eps", "layers", "nn", "num_hidden_layers", "pad_token_id", "padding_idx", "post_init", "range", "rotary_emb", "self", "super", "vocab_size"], "phi/modeling_phi.py:PhiModel.forward": ["BaseModelOutputWithPast", "Cache", "DynamicCache", "False", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Setting", "Tensor", "TransformersKwargs", "True", "Unpack", "ValueError", "You", "all_hidden_states", "all_self_attns", "and", "arange", "attention_mask", "attentions", "bool", "cache_position", "causal_mask", "checkpointing", "class", "config", "create_causal_mask", "decoder_layer", "def", "device", "else", "embed_dropout", "embed_tokens", "exactly", "final_layernorm", "for", "forward", "get_seq_length", "gradient", "gradient_checkpointing", "hidden_states", "if", "in", "incompatible", "input_embeds", "input_ids", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layer_outputs", "layers", "logger", "must", "not", "num_hidden_layers", "of", "one", "or", "output_attentions", "output_hidden_states", "past_key_values", "past_seen_tokens", "position_embeddings", "position_ids", "raise", "return", "rotary_emb", "self", "shape", "specify", "torch", "training", "unsqueeze", "use_cache", "warning_once", "with"], "phi/modeling_phi.py:PhiForCausalLM.__init__": ["GenerationMixin", "Linear", "ModelForCausalLM", "ModelModel", "ModelPreTrainedModel", "True", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "vocab_size"], "phi/modeling_phi.py:PhiForCausalLM.forward": ["BaseModelOutputWithPast", "Cache", "CausalLMOutputWithPast", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForCausalLM", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "outputs", "past_key_values", "position_ids", "r", "return", "self", "slice", "slice_indices", "torch", "use_cache", "vocab_size"], "phi3/modeling_phi3.py:Phi3MLP.__init__": ["ACT2FN", "False", "Linear", "ModelMLP", "Module", "__init__", "activation_fn", "bias", "class", "config", "def", "down_proj", "gate_up_proj", "hidden_act", "hidden_size", "intermediate_size", "nn", "self", "super"], "phi3/modeling_phi3.py:Phi3MLP.forward": ["FloatTensor", "ModelMLP", "Module", "activation_fn", "chunk", "class", "def", "dim", "down_proj", "forward", "gate", "gate_up_proj", "hidden_states", "nn", "return", "self", "torch", "up_states"], "phi3/modeling_phi3.py:Phi3RotaryEmbedding.__init__": ["Callable", "False", "ModelConfig", "ModelRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "phi3/modeling_phi3.py:Phi3RotaryEmbedding.compute_default_rope_parameters": ["ModelConfig", "ModelRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "get", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "partial_rotary_factor", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "phi3/modeling_phi3.py:Phi3RotaryEmbedding.forward": ["False", "ModelRotaryEmbedding", "Module", "None", "and", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "phi3/modeling_phi3.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "phi3/modeling_phi3.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "phi3/modeling_phi3.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value", "value_states"], "phi3/modeling_phi3.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cat", "cos", "def", "dim", "k", "k_embed", "k_pass", "k_rot", "q", "q_embed", "q_pass", "q_rot", "return", "rotary_dim", "rotate_half", "shape", "sin", "torch", "unsqueeze", "unsqueeze_dim"], "phi3/modeling_phi3.py:Phi3Attention.__init__": ["False", "Linear", "ModelAttention", "ModelConfig", "Module", "None", "True", "__init__", "attention_dropout", "bias", "class", "config", "def", "getattr", "head_dim", "hidden_size", "int", "is_causal", "layer_idx", "nn", "num_attention_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "op_size", "qkv_proj", "scaling", "self", "super"], "phi3/modeling_phi3.py:Phi3Attention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "FlashAttentionKwargs", "LongTensor", "ModelAttention", "Module", "None", "Tensor", "Unpack", "_attn_implementation", "apply_rotary_pos_emb", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "cache_kwargs", "cache_position", "class", "config", "contiguous", "cos", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "getattr", "head_dim", "hidden_shape", "hidden_states", "if", "input_shape", "is", "key_states", "kwargs", "layer_idx", "nn", "not", "num_attention_heads", "num_key_value_heads", "o_proj", "past_key_values", "position_embeddings", "qkv", "qkv_proj", "query_pos", "query_states", "reshape", "return", "scaling", "self", "shape", "sin", "sliding_window", "torch", "training", "transpose", "tuple", "update", "value_states", "view"], "phi3/modeling_phi3.py:Phi3RMSNorm.__init__": ["ModelRMSNorm", "Module", "Parameter", "__init__", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "phi3/modeling_phi3.py:Phi3RMSNorm.forward": ["ModelRMSNorm", "Module", "True", "class", "def", "dtype", "float32", "forward", "hidden_states", "input_dtype", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "phi3/modeling_phi3.py:Phi3RMSNorm.extra_repr": ["ModelRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "variance_epsilon", "weight"], "phi3/modeling_phi3.py:Phi3DecoderLayer.__init__": ["Dropout", "GradientCheckpointingLayer", "ModelAttention", "ModelConfig", "ModelDecoderLayer", "ModelMLP", "ModelRMSNorm", "__init__", "class", "config", "def", "eps", "hidden_size", "input_layernorm", "int", "layer_idx", "mlp", "nn", "post_attention_layernorm", "resid_attn_dropout", "resid_mlp_dropout", "resid_pdrop", "rms_norm_eps", "self", "self_attn", "super"], "phi3/modeling_phi3.py:Phi3DecoderLayer.forward": ["Cache", "False", "FlashAttentionKwargs", "FloatTensor", "GradientCheckpointingLayer", "LongTensor", "ModelDecoderLayer", "None", "Tensor", "Unpack", "attention_mask", "bool", "cache_position", "class", "def", "forward", "hidden_states", "input_layernorm", "kwargs", "mlp", "past_key_values", "position_embeddings", "position_ids", "post_attention_layernorm", "resid_attn_dropout", "resid_mlp_dropout", "residual", "return", "self", "self_attn", "self_attn_weights", "torch", "tuple", "use_cache"], "phi3/modeling_phi3.py:Phi3Model.__init__": ["Embedding", "False", "ModelConfig", "ModelDecoderLayer", "ModelModel", "ModelPreTrainedModel", "ModelRMSNorm", "ModelRotaryEmbedding", "ModuleList", "__init__", "class", "config", "def", "embed_tokens", "eps", "for", "gradient_checkpointing", "hidden_size", "in", "layer_idx", "layers", "nn", "norm", "num_hidden_layers", "pad_token_id", "padding_idx", "post_init", "range", "rms_norm_eps", "rotary_emb", "self", "super", "vocab_size"], "phi3/modeling_phi3.py:Phi3Model.forward": ["BaseModelOutputWithPast", "Cache", "DynamicCache", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "and", "arange", "attention_mask", "bool", "cache_position", "causal_mask", "class", "config", "create_causal_mask", "create_sliding_window_causal_mask", "decoder_layer", "def", "device", "else", "embed_tokens", "exactly", "for", "forward", "get_seq_length", "hidden_states", "if", "in", "input_embeds", "input_ids", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layers", "mask_function", "must", "norm", "not", "num_hidden_layers", "of", "one", "or", "past_key_values", "past_seen_tokens", "position_embeddings", "position_ids", "raise", "return", "rotary_emb", "self", "shape", "sliding_window", "specify", "torch", "unsqueeze", "use_cache"], "phi3/modeling_phi3.py:Phi3ForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "ModelForCausalLM", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "vocab_size"], "phi3/modeling_phi3.py:Phi3ForCausalLM.forward": ["BaseModelOutputWithPast", "Cache", "CausalLMOutputWithPast", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForCausalLM", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "outputs", "past_key_values", "position_ids", "r", "return", "self", "slice", "slice_indices", "torch", "use_cache", "vocab_size"], "phi3/modeling_phi3.py:Phi3ForCausalLM.prepare_inputs_for_generation": ["GenerationMixin", "ModelForCausalLM", "ModelPreTrainedModel", "None", "True", "and", "attention_mask", "cache_position", "class", "config", "def", "hasattr", "if", "input_ids", "inputs_embeds", "kwargs", "logits_to_keep", "model_inputs", "original_max_position_embeddings", "past_key_values", "past_length", "position_ids", "prepare_inputs_for_generation", "return", "self", "shape", "super", "use_cache"], "phi4_multimodal/modeling_phi4_multimodal.py:Phi4MultimodalVisionMLP.__init__": ["ACT2FN", "Linear", "ModelVisionMLP", "Module", "__init__", "activation_fn", "class", "config", "def", "fc1", "fc2", "hidden_act", "hidden_size", "intermediate_size", "nn", "self", "super"], "phi4_multimodal/modeling_phi4_multimodal.py:Phi4MultimodalVisionMLP.forward": ["ModelVisionMLP", "Module", "Tensor", "activation_fn", "class", "def", "fc1", "fc2", "forward", "hidden_states", "nn", "return", "self", "torch"], "phi4_multimodal/modeling_phi4_multimodal.py:simple_eager_attention_forward": ["Model_eager_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key_states", "kwargs", "matmul", "module", "nn", "not", "p", "query_states", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value_states"], "phi4_multimodal/modeling_phi4_multimodal.py:Phi4MultimodalVisionAttention.__init__": ["Linear", "ModelVisionAttention", "ModelVisionConfig", "Module", "True", "__init__", "attention_dropout", "class", "config", "def", "embed_dim", "head_dim", "hidden_size", "is_causal", "k_proj", "nn", "num_attention_heads", "num_heads", "out_proj", "q_proj", "scaling", "self", "super", "v_proj"], "phi4_multimodal/modeling_phi4_multimodal.py:Phi4MultimodalVisionAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "ModelVisionAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_attn_implementation", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "class", "config", "def", "dropout", "eager", "else", "forward", "head_dim", "hidden_shape", "hidden_states", "if", "input_shape", "k_proj", "key_states", "kwargs", "nn", "not", "out_proj", "q_proj", "query_states", "reshape", "return", "scaling", "self", "shape", "simple_eager_attention_forward", "torch", "training", "transpose", "tuple", "v_proj", "value_states", "view"], "phi4_multimodal/modeling_phi4_multimodal.py:Phi4MultimodalVisionEncoderLayer.__init__": ["GradientCheckpointingLayer", "LayerNorm", "ModelVisionAttention", "ModelVisionConfig", "ModelVisionEncoderLayer", "ModelVisionMLP", "__init__", "class", "config", "def", "embed_dim", "eps", "hidden_size", "layer_norm1", "layer_norm2", "layer_norm_eps", "mlp", "nn", "self", "self_attn", "super"], "phi4_multimodal/modeling_phi4_multimodal.py:Phi4MultimodalVisionEncoderLayer.forward": ["FloatTensor", "GradientCheckpointingLayer", "ModelVisionEncoderLayer", "Tensor", "TransformersKwargs", "Unpack", "_", "attention_mask", "class", "def", "forward", "hidden_states", "kwargs", "layer_norm1", "layer_norm2", "mlp", "residual", "return", "self", "self_attn", "torch"], "phi4_multimodal/modeling_phi4_multimodal.py:Phi4MultimodalVisionEncoder.__init__": ["False", "ModelVisionConfig", "ModelVisionEncoder", "ModelVisionEncoderLayer", "Module", "ModuleList", "_", "__init__", "class", "config", "def", "for", "gradient_checkpointing", "in", "layers", "nn", "num_hidden_layers", "range", "self", "super"], "phi4_multimodal/modeling_phi4_multimodal.py:Phi4MultimodalVisionEncoder.forward": ["BaseModelOutput", "ModelVisionEncoder", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "class", "def", "encoder_layer", "for", "forward", "hidden_states", "in", "inputs_embeds", "kwargs", "last_hidden_state", "layers", "nn", "return", "self", "torch"], "phi4_multimodal/modeling_phi4_multimodal.py:variance_scaling_": ["Model", "Model_scaling_", "ValueError", "_calculate_fan_in_and_fan_out", "bound", "def", "denom", "distribution", "elif", "else", "f", "fan_avg", "fan_in", "fan_out", "if", "init", "invalid", "math", "mode", "normal", "normal_", "raise", "sqrt", "std", "tensor", "trunc_normal_", "truncated_normal", "uniform", "uniform_"], "phi4_multimodal/modeling_phi4_multimodal.py:lecun_normal_": ["Model_normal_", "def", "distribution", "fan_in", "mode", "tensor", "truncated_normal", "variance_scaling_"], "phi4_multimodal/modeling_phi4_multimodal.py:default_flax_embed_init": ["Model_flax_embed_init", "def", "distribution", "fan_in", "mode", "normal", "tensor", "variance_scaling_"], "phi4_multimodal/modeling_phi4_multimodal.py:Phi4MultimodalVisionPreTrainedModel._init_weights": ["Conv2d", "Embedding", "LayerNorm", "Linear", "ModelVisionAttention", "ModelVisionConfig", "ModelVisionEmbeddings", "ModelVisionMLP", "ModelVisionMultiheadAttentionPoolingHead", "ModelVisionPreTrainedModel", "None", "PreTrainedModel", "_init_weights", "attention", "bias", "class", "config", "def", "default_flax_embed_init", "elif", "else", "fc1", "fc2", "hidden_size", "if", "in_proj_bias", "in_proj_weight", "init", "is", "isinstance", "k_proj", "lecun_normal_", "module", "nn", "normal_", "not", "np", "ones_", "out_proj", "position_embedding", "probe", "q_proj", "self", "sqrt", "std", "v_proj", "weight", "width", "zeros_"], "phi4_multimodal/modeling_phi4_multimodal.py:Phi4MultimodalVisionEmbeddings.__init__": ["Conv2d", "Embedding", "ModelVisionConfig", "ModelVisionEmbeddings", "Module", "__init__", "class", "config", "def", "hidden_size", "image_size", "in_channels", "kernel_size", "nn", "num_channels", "num_patches_per_side", "out_channels", "padding", "patch_embedding", "patch_size", "position_embedding", "self", "stride", "super", "valid"], "phi4_multimodal/modeling_phi4_multimodal.py:Phi4MultimodalVisionEmbeddings.interpolate_pos_encoding": ["False", "ModelVisionEmbeddings", "Module", "Tensor", "align_corners", "and", "bicubic", "class", "def", "dim", "embeddings", "functional", "height", "if", "int", "interpolate", "interpolate_pos_encoding", "is_tracing", "jit", "mode", "new_height", "new_width", "nn", "not", "num_patches", "num_positions", "patch_pos_embed", "patch_size", "permute", "position_embedding", "position_ids", "reshape", "return", "self", "shape", "size", "sqrt_num_positions", "torch", "torch_int", "unsqueeze", "view", "weight", "width"], "phi4_multimodal/modeling_phi4_multimodal.py:Phi4MultimodalVisionEmbeddings.forward": ["BoolTensor", "FloatTensor", "ModelVisionEmbeddings", "Module", "None", "Tensor", "True", "arange", "batch_idx", "batch_size", "boundaries", "bucket_coords_h", "bucket_coords_w", "bucketize", "class", "cpu", "def", "device", "embeddings", "enumerate", "fill_value", "flatten", "for", "forward", "fractional_coords_h", "fractional_coords_w", "full", "in", "max_im_h", "max_im_w", "max_nb_patches_h", "max_nb_patches_w", "nb_patches_h", "nb_patches_w", "nn", "num_patches_per_side", "p_attn_mask", "patch_attention_mask", "patch_embedding", "patch_embeds", "patch_size", "pixel_values", "pos_ids", "position_embedding", "position_ids", "return", "right", "self", "size", "sum", "to", "torch", "transpose", "view", "weight"], "phi4_multimodal/modeling_phi4_multimodal.py:Phi4MultimodalVisionMultiheadAttentionPoolingHead.__init__": ["LayerNorm", "ModelVisionConfig", "ModelVisionMLP", "ModelVisionMultiheadAttentionPoolingHead", "Module", "MultiheadAttention", "Parameter", "True", "__init__", "attention", "batch_first", "class", "config", "def", "eps", "hidden_size", "layer_norm_eps", "layernorm", "mlp", "nn", "num_attention_heads", "probe", "randn", "self", "super", "torch"], "phi4_multimodal/modeling_phi4_multimodal.py:Phi4MultimodalVisionMultiheadAttentionPoolingHead.forward": ["ModelVisionMultiheadAttentionPoolingHead", "Module", "attention", "attention_mask", "batch_size", "class", "def", "forward", "hidden_state", "key", "key_padding_mask", "layernorm", "mlp", "nn", "probe", "query", "repeat", "residual", "return", "self", "shape", "value"], "phi4_multimodal/modeling_phi4_multimodal.py:Phi4MultimodalVisionModel.__init__": ["LayerNorm", "ModelVisionConfig", "ModelVisionEmbeddings", "ModelVisionEncoder", "ModelVisionModel", "ModelVisionMultiheadAttentionPoolingHead", "ModelVisionPreTrainedModel", "__init__", "class", "config", "def", "embeddings", "encoder", "eps", "head", "hidden_size", "layer_norm_eps", "nn", "post_init", "post_layernorm", "self", "super"], "phi4_multimodal/modeling_phi4_multimodal.py:Phi4MultimodalVisionModel.get_input_embeddings": ["ModelVisionModel", "ModelVisionPreTrainedModel", "Module", "class", "def", "embeddings", "get_input_embeddings", "nn", "patch_embedding", "return", "self"], "phi4_multimodal/modeling_phi4_multimodal.py:Phi4MultimodalVisionModel.forward": ["BaseModelOutput", "BaseModelOutputWithPooling", "BoolTensor", "ModelVisionModel", "ModelVisionPreTrainedModel", "None", "TransformersKwargs", "Unpack", "_attn_implementation", "_prepare_4d_attention_mask", "any", "attention_mask", "batch_size", "bool", "class", "config", "def", "device", "dtype", "else", "embeddings", "encoder", "encoder_outputs", "flash_attention_2", "forward", "head", "hidden_state", "hidden_states", "if", "inputs_embeds", "is", "kwargs", "last_hidden_state", "not", "ones", "patch_attention_mask", "patch_size", "pixel_values", "pooled_output", "pooler_output", "post_layernorm", "return", "self", "size", "torch", "view"], "phi4_multimodal/modeling_phi4_multimodal.py:Phi4MultimodalImageEmbedding.__init__": ["AvgPool2d", "Dropout", "Linear", "ModelConfig", "ModelImageEmbedding", "ModelVisionModel", "Module", "Parameter", "ReflectionPad2d", "__init__", "_from_config", "class", "config", "crop_size", "def", "drop", "embd_pdrop", "feature_layer", "global_img_feature_extensor", "hidden_size", "if", "image_dim_out", "image_size", "image_token_compression", "img_processor", "img_processor_padding", "img_projection_down", "img_projection_up", "kernel_size", "layer_idx", "n_patches", "nn", "num_img_tokens", "patch_size", "self", "stride", "sub_img_feature_extensor", "super", "torch", "vision_config", "zeros"], "phi4_multimodal/modeling_phi4_multimodal.py:Phi4MultimodalImageEmbedding.get_img_features": ["FloatTensor", "ModelImageEmbedding", "Module", "None", "True", "attention_mask", "class", "def", "get_img_features", "getattr", "hidden_states", "if", "image_token_compression", "img_embeds", "img_feature", "img_processor", "img_processor_output", "img_processor_padding", "int", "is", "layer_idx", "math", "nn", "not", "output_hidden_states", "patch_attention_mask", "patch_feature", "permute", "return", "self", "size", "sqrt", "torch", "view", "width"], "phi4_multimodal/modeling_phi4_multimodal.py:Phi4MultimodalImageEmbedding.forward": ["False", "FloatTensor", "LongTensor", "ModelImageEmbedding", "Module", "None", "Tensor", "True", "accumulate", "append", "area_ratio", "as_tuple", "attention_mask", "base_feat_size", "batch_size", "bias", "bool", "cat", "class", "config", "contiguous", "crop_size", "def", "device", "device_type", "dim", "drop", "dtype", "else", "embeddings", "enabled", "flatten", "for", "forward", "functional", "gelu", "get_img_features", "global_img", "global_img_feature_extensor", "height", "height_ratio", "idx", "if", "image_attention_mask", "image_dim_out", "image_embeds", "image_pixel_values", "image_sizes", "image_token_id", "img_feature_proj", "img_features", "img_processor", "img_projection_down", "img_projection_up", "img_set_tensor", "in", "index_put", "indices", "input_ids", "inputs_embeds", "int", "is", "item", "maybe_autocast", "merged_img_set_tensor", "nn", "no_grad", "nonzero", "not", "np", "output_img", "output_imgs", "patch_embedding", "positions_tuple", "range", "repeat", "reshape", "reshaped_image_attention_mask", "return", "self", "shape", "sqrt", "squeeze", "sub_img", "sub_img_feature_extensor", "sum", "target_device", "target_dtype", "temporary_extensor", "to", "torch", "transpose", "type", "useful_height", "useful_width", "values", "view", "vision_config", "weight", "width", "width_ratio", "with"], "phi4_multimodal/modeling_phi4_multimodal.py:Phi4MultimodalAudioMLP.__init__": ["ACT2FN", "Dropout", "LayerNorm", "Linear", "ModelAudioConfig", "ModelAudioMLP", "Module", "__init__", "act_fn", "activation", "class", "config", "def", "down_proj", "dropout", "dropout_rate", "gate_up_proj", "hidden_size", "intermediate_size", "layer_norm", "nn", "self", "super"], "phi4_multimodal/modeling_phi4_multimodal.py:Phi4MultimodalAudioMLP.forward": ["ModelAudioMLP", "Module", "act_fn", "chunk", "class", "def", "dim", "down_proj", "dropout", "forward", "gate", "gate_up_proj", "hidden_states", "layer_norm", "nn", "out", "return", "self", "up_states"], "phi4_multimodal/modeling_phi4_multimodal.py:Phi4MultimodalAudioAttention.__init__": ["Linear", "ModelAudioAttention", "ModelAudioConfig", "Module", "True", "__init__", "attention_dropout", "bias", "class", "config", "def", "dropout_rate", "getattr", "head_dim", "hidden_size", "is_causal", "k_proj", "nn", "num_attention_heads", "o_proj", "q_proj", "scaling", "self", "super", "v_proj"], "phi4_multimodal/modeling_phi4_multimodal.py:Phi4MultimodalAudioAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "ModelAudioAttention", "Module", "Tensor", "_", "_attn_implementation", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "class", "config", "contiguous", "def", "dropout", "eager", "else", "forward", "head_dim", "hidden_shape", "hidden_states", "if", "input_shape", "k_proj", "key_states", "kwargs", "nn", "not", "o_proj", "q_proj", "query_states", "reshape", "return", "scaling", "self", "shape", "simple_eager_attention_forward", "torch", "training", "transpose", "v_proj", "value_states", "view"], "phi4_multimodal/modeling_phi4_multimodal.py:Phi4MultimodalAudioDepthWiseSeparableConv1d.__init__": ["Conv1d", "ModelAudioConfig", "ModelAudioDepthWiseSeparableConv1d", "Module", "__init__", "class", "config", "def", "depthwise_multiplier", "depthwise_separable_out_channel", "dw_conv", "groups", "hidden_size", "int", "kernel_size", "nn", "padding", "pw_conv", "self", "super"], "phi4_multimodal/modeling_phi4_multimodal.py:Phi4MultimodalAudioDepthWiseSeparableConv1d.forward": ["ModelAudioDepthWiseSeparableConv1d", "Module", "class", "def", "dw_conv", "forward", "hidden_states", "nn", "pw_conv", "return", "self"], "phi4_multimodal/modeling_phi4_multimodal.py:Phi4MultimodalAudioGluPointWiseConv.__init__": ["ACT2FN", "Conv1d", "ModelAudioConfig", "ModelAudioGluPointWiseConv", "Module", "Parameter", "__init__", "b1", "b2", "class", "config", "conv_glu_type", "def", "ext_pw_conv_1d", "ext_pw_out_channel", "glu_act", "hidden_size", "kernel_size", "nn", "output_dim", "self", "stride", "super", "torch", "zeros"], "phi4_multimodal/modeling_phi4_multimodal.py:Phi4MultimodalAudioGluPointWiseConv.forward": ["ModelAudioGluPointWiseConv", "Module", "b1", "b2", "class", "def", "ext_pw_conv_1d", "forward", "glu_act", "hidden_states", "nn", "out", "output_dim", "permute", "return", "self"], "phi4_multimodal/modeling_phi4_multimodal.py:Phi4MultimodalAudioConvModule.__init__": ["ACT2FN", "Conv1d", "Dropout", "LayerNorm", "ModelAudioConfig", "ModelAudioConvModule", "ModelAudioDepthWiseSeparableConv1d", "ModelAudioGluPointWiseConv", "Module", "__init__", "act", "class", "config", "conv_activation", "def", "dropout", "dropout_rate", "dw_sep_conv_1d", "ext_pw_conv_1d", "ext_pw_out_channel", "glu", "hidden_size", "kernel_size", "layer_norm", "nn", "padding", "self", "stride", "super"], "phi4_multimodal/modeling_phi4_multimodal.py:Phi4MultimodalAudioConvModule.forward": ["ModelAudioConvModule", "Module", "Tensor", "act", "class", "def", "dropout", "dw_sep_conv_1d", "ext_pw_conv_1d", "forward", "glu", "hidden_states", "if", "kernel_size", "layer_norm", "nn", "out", "permute", "return", "self", "torch"], "phi4_multimodal/modeling_phi4_multimodal.py:Phi4MultimodalAudioConformerEncoderLayer.__init__": ["LayerNorm", "ModelAudioAttention", "ModelAudioConfig", "ModelAudioConformerEncoderLayer", "ModelAudioConvModule", "ModelAudioMLP", "Module", "__init__", "class", "config", "conv", "def", "feed_forward_in", "feed_forward_out", "hidden_size", "layer_norm", "layer_norm_att", "nn", "self", "self_attn", "super"], "phi4_multimodal/modeling_phi4_multimodal.py:Phi4MultimodalAudioConformerEncoderLayer.forward": ["ModelAudioConformerEncoderLayer", "Module", "Tensor", "attention_mask", "class", "conv", "def", "feed_forward_in", "feed_forward_out", "forward", "hidden_states", "layer_norm", "layer_norm_att", "nn", "out", "residual", "return", "self", "self_attn", "torch"], "phi4_multimodal/modeling_phi4_multimodal.py:Phi4MultimodalAudioNemoConvSubsampling.__init__": ["ACT2FN", "Conv2d", "Linear", "ModelAudioConfig", "ModelAudioNemoConvSubsampling", "Module", "Sequential", "_", "__init__", "act_fn", "class", "config", "conv", "conv_channels", "def", "extend", "for", "groups", "hidden_size", "in", "int", "kernel_size", "layers", "log2", "math", "nemo_activation", "nemo_conv_channels", "nemo_final_size", "nn", "out", "padding", "range", "sampling_num", "self", "stride", "subsampling_factor", "super", "time_reduction", "torch"], "phi4_multimodal/modeling_phi4_multimodal.py:Phi4MultimodalAudioNemoConvSubsampling.forward": ["ModelAudioNemoConvSubsampling", "Module", "None", "Tensor", "_", "arange", "arange_", "b", "ceil", "class", "conv", "def", "device", "expand", "feature_lens", "forward", "hidden_states", "if", "is", "mask", "max_audio_length", "nn", "out", "pad_mask", "padding_length", "reshape", "return", "self", "shape", "size", "subsampling_factor", "sum", "t", "torch", "transpose", "unsqueeze"], "phi4_multimodal/modeling_phi4_multimodal.py:Phi4MultimodalAudioRelativeAttentionBias.__init__": ["Embedding", "ModelAudioConfig", "ModelAudioRelativeAttentionBias", "Module", "__init__", "bias_max_distance", "bias_symmetric", "bias_values", "class", "config", "def", "if", "max_distance", "nn", "not", "num_attention_heads", "num_buckets", "self", "super", "symmetric"], "phi4_multimodal/modeling_phi4_multimodal.py:Phi4MultimodalAudioRelativeAttentionBias.forward": ["ModelAudioRelativeAttentionBias", "Module", "None", "abs", "arange", "att_bias", "bias_idx", "bias_values", "class", "context_position", "def", "device", "dtype", "else", "forward", "if", "long", "masked_fill", "max_distance", "max_pos", "memory_position", "nn", "num_buckets", "permute", "relative_position", "return", "self", "size", "symmetric", "torch", "unsqueeze", "x"], "phi4_multimodal/modeling_phi4_multimodal.py:Phi4MultimodalAudioMeanVarianceNormLayer.__init__": ["ModelAudioConfig", "ModelAudioMeanVarianceNormLayer", "Module", "__init__", "class", "config", "def", "global_invstd", "global_mean", "input_size", "nn", "ones", "register_buffer", "self", "super", "torch", "zeros"], "phi4_multimodal/modeling_phi4_multimodal.py:Phi4MultimodalAudioMeanVarianceNormLayer.forward": ["ModelAudioMeanVarianceNormLayer", "Module", "class", "def", "forward", "global_invstd", "global_mean", "nn", "return", "self", "x"], "phi4_multimodal/modeling_phi4_multimodal.py:Phi4MultimodalAudioPreTrainedModel._init_weights": ["ModelAudioGluPointWiseConv", "ModelAudioMeanVarianceNormLayer", "ModelAudioPreTrainedModel", "PreTrainedModel", "_init_weights", "b1", "b2", "class", "def", "elif", "global_invstd", "global_mean", "if", "init", "isinstance", "module", "ones_", "self", "super", "zeros_"], "phi4_multimodal/modeling_phi4_multimodal.py:unfold_tensor": ["D", "F", "Model", "Model_tensor", "None", "_", "contiguous", "def", "kernel_size", "max_seq_len", "new_bsz", "permute", "return", "shape", "slen", "stride", "tensor", "transpose", "view"], "phi4_multimodal/modeling_phi4_multimodal.py:adaptive_enc_mask": ["Model_enc_mask", "Tensor", "arange", "boundary_left", "boundary_right", "chunk_start_idx", "def", "end_pad", "expand", "functional", "idx", "idx_left", "idx_right", "left_window", "len", "long", "mask_left", "mask_right", "nn", "nonzero", "pad", "return", "right_window", "seq_range", "seq_range_expand", "start_pad", "torch", "unsqueeze", "value", "x_len"], "phi4_multimodal/modeling_phi4_multimodal.py:Phi4MultimodalAudioModel.__init__": ["False", "ModelAudioConfig", "ModelAudioConformerEncoderLayer", "ModelAudioMeanVarianceNormLayer", "ModelAudioModel", "ModelAudioNemoConvSubsampling", "ModelAudioPreTrainedModel", "ModelAudioRelativeAttentionBias", "ModuleList", "_", "__init__", "class", "config", "def", "embed", "encoder_embedding", "encoders", "for", "gradient_checkpointing", "in", "nn", "num_blocks", "post_init", "range", "relative_attention_bias_layer", "self", "super"], "phi4_multimodal/modeling_phi4_multimodal.py:Phi4MultimodalAudioModel._streaming_mask": ["ModelAudioModel", "ModelAudioPreTrainedModel", "_streaming_mask", "adaptive_enc_mask", "and", "arange", "batch_size", "chunk_size", "chunk_start_idx", "class", "def", "enc_streaming_mask", "expand", "if", "insert", "left_chunk", "left_window", "np", "rand", "random", "return", "self", "seq_len", "training", "unsqueeze"], "phi4_multimodal/modeling_phi4_multimodal.py:Phi4MultimodalAudioModel.forward_embeddings": ["ModelAudioModel", "ModelAudioPreTrainedModel", "None", "The", "ValueError", "Your", "_streaming_mask", "after", "and", "batch_size", "ceil", "chunk_size", "class", "config", "def", "device", "elif", "else", "embed", "enc_streaming_mask", "f", "feature", "forward_embeddings", "hidden_states", "hs_mask", "if", "input", "invalid", "is", "left_chunk", "length", "masks", "math", "not", "raise", "reduction", "return", "self", "seq_len", "sequence", "shape", "short", "streaming_mask", "time", "time_reduction", "to", "too"], "phi4_multimodal/modeling_phi4_multimodal.py:Phi4MultimodalAudioModel.calculate_hs_mask": ["ModelAudioModel", "ModelAudioPreTrainedModel", "None", "_streaming_mask", "arange", "batch_size", "calculate_hs_mask", "chunk_size", "class", "config", "def", "device", "enc_streaming_mask", "expand", "feature_lens", "hidden_states", "if", "is", "left_chunk", "mask", "max_audio_length", "pad_mask", "padding_length", "return", "self", "shape", "size", "sum", "to", "torch", "unsqueeze"], "phi4_multimodal/modeling_phi4_multimodal.py:Phi4MultimodalAudioModel.forward": ["F", "False", "ModelAudioModel", "ModelAudioPreTrainedModel", "None", "Tensor", "True", "_", "attention_mask", "bool", "bs", "calculate_hs_mask", "chunk_pad_size", "class", "constant", "def", "device", "else", "embed_dim", "encoder_embedding", "encoders", "extra_padded_subsamlped_pad_mask", "float", "for", "forward", "forward_embeddings", "hidden_states", "hidden_states_pad", "hs_mask", "if", "in", "is", "kwargs", "layer", "mask", "masks_unfold", "max_seq_len", "not", "pad", "relative_attention_bias", "relative_attention_bias_layer", "reshape", "return", "self", "seq_len", "shape", "squeeze", "subsampled_pad_mask", "to", "torch", "unfold_tensor", "unfolded", "unsqueeze"], "phi4_multimodal/modeling_phi4_multimodal.py:Phi4MultimodalAudioEmbedding.__init__": ["Dropout", "Linear", "ModelAudioEmbedding", "ModelAudioModel", "ModelConfig", "Module", "__init__", "_from_config", "audio_config", "class", "config", "def", "down_proj_for_speech", "down_proj_for_vision_speech", "downsample_rate", "drop", "embd_pdrop", "encoder", "feature_layer", "hidden_size", "layer_idx", "nn", "self", "super", "up_proj_for_speech", "up_proj_for_vision_speech"], "phi4_multimodal/modeling_phi4_multimodal.py:Phi4MultimodalAudioEmbedding.forward": ["False", "FloatTensor", "LongTensor", "ModelAudioEmbedding", "Module", "None", "Tensor", "True", "accumulate", "as_tuple", "audio_attention_mask", "audio_config", "audio_embed_sizes", "audio_embeds", "audio_encoder_hidden_states", "audio_input_features", "audio_projection_mode", "audio_token_id", "bias", "cat", "class", "config", "def", "device", "device_type", "dim", "down_proj", "down_proj_for_speech", "down_proj_for_vision_speech", "drop", "dtype", "else", "enabled", "encoder", "for", "forward", "functional", "gelu", "i", "if", "in", "index_put", "indices", "input_ids", "inputs_embeds", "len", "maybe_autocast", "merged_audio_embeds", "nn", "no_grad", "nonzero", "positions_tuple", "range", "return", "self", "speech", "target_device", "target_dtype", "to", "torch", "type", "up_proj", "up_proj_for_speech", "up_proj_for_vision_speech", "values", "with"], "phi4_multimodal/modeling_phi4_multimodal.py:Phi4MultimodalRMSNorm.__init__": ["ModelRMSNorm", "Module", "Parameter", "__init__", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "phi4_multimodal/modeling_phi4_multimodal.py:Phi4MultimodalRMSNorm.forward": ["ModelRMSNorm", "Module", "True", "class", "def", "dtype", "float32", "forward", "hidden_states", "input_dtype", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "phi4_multimodal/modeling_phi4_multimodal.py:Phi4MultimodalRMSNorm.extra_repr": ["ModelRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "variance_epsilon", "weight"], "phi4_multimodal/modeling_phi4_multimodal.py:Phi4MultimodalMLP.__init__": ["ACT2FN", "False", "Linear", "ModelMLP", "Module", "__init__", "activation_fn", "bias", "class", "config", "def", "down_proj", "gate_up_proj", "hidden_act", "hidden_size", "intermediate_size", "nn", "self", "super"], "phi4_multimodal/modeling_phi4_multimodal.py:Phi4MultimodalMLP.forward": ["FloatTensor", "ModelMLP", "Module", "activation_fn", "chunk", "class", "def", "dim", "down_proj", "forward", "gate", "gate_up_proj", "hidden_states", "nn", "return", "self", "torch", "up_states"], "phi4_multimodal/modeling_phi4_multimodal.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "phi4_multimodal/modeling_phi4_multimodal.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "phi4_multimodal/modeling_phi4_multimodal.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value", "value_states"], "phi4_multimodal/modeling_phi4_multimodal.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cat", "cos", "def", "dim", "k", "k_embed", "k_pass", "k_rot", "q", "q_embed", "q_pass", "q_rot", "return", "rotary_dim", "rotate_half", "shape", "sin", "torch", "unsqueeze", "unsqueeze_dim"], "phi4_multimodal/modeling_phi4_multimodal.py:Phi4MultimodalAttention.__init__": ["False", "Linear", "ModelAttention", "ModelConfig", "Module", "None", "True", "__init__", "attention_dropout", "bias", "class", "config", "def", "getattr", "head_dim", "hidden_size", "int", "is_causal", "layer_idx", "nn", "num_attention_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "op_size", "qkv_proj", "scaling", "self", "super"], "phi4_multimodal/modeling_phi4_multimodal.py:Phi4MultimodalAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "FlashAttentionKwargs", "LongTensor", "ModelAttention", "Module", "None", "Tensor", "Unpack", "_attn_implementation", "apply_rotary_pos_emb", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "cache_kwargs", "cache_position", "class", "config", "contiguous", "cos", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "getattr", "head_dim", "hidden_shape", "hidden_states", "if", "input_shape", "is", "key_states", "kwargs", "layer_idx", "nn", "not", "num_attention_heads", "num_key_value_heads", "o_proj", "past_key_values", "position_embeddings", "qkv", "qkv_proj", "query_pos", "query_states", "reshape", "return", "scaling", "self", "shape", "sin", "sliding_window", "torch", "training", "transpose", "tuple", "update", "value_states", "view"], "phi4_multimodal/modeling_phi4_multimodal.py:Phi4MultimodalDecoderLayer.__init__": ["Dropout", "GradientCheckpointingLayer", "ModelAttention", "ModelConfig", "ModelDecoderLayer", "ModelMLP", "ModelRMSNorm", "__init__", "class", "config", "def", "eps", "hidden_size", "input_layernorm", "int", "layer_idx", "mlp", "nn", "post_attention_layernorm", "resid_attn_dropout", "resid_mlp_dropout", "resid_pdrop", "rms_norm_eps", "self", "self_attn", "super"], "phi4_multimodal/modeling_phi4_multimodal.py:Phi4MultimodalDecoderLayer.forward": ["Cache", "False", "FlashAttentionKwargs", "FloatTensor", "GradientCheckpointingLayer", "LongTensor", "ModelDecoderLayer", "None", "Tensor", "Unpack", "attention_mask", "bool", "cache_position", "class", "def", "forward", "hidden_states", "input_layernorm", "kwargs", "mlp", "past_key_values", "position_embeddings", "position_ids", "post_attention_layernorm", "resid_attn_dropout", "resid_mlp_dropout", "residual", "return", "self", "self_attn", "self_attn_weights", "torch", "tuple", "use_cache"], "phi4_multimodal/modeling_phi4_multimodal.py:Phi4MultimodalFeatureEmbedding.__init__": ["ModelAudioEmbedding", "ModelConfig", "ModelFeatureEmbedding", "ModelImageEmbedding", "Module", "None", "__init__", "audio_config", "audio_embed", "audio_token_id", "class", "config", "def", "image_embed", "image_token_id", "nn", "self", "super", "vision_config"], "phi4_multimodal/modeling_phi4_multimodal.py:Phi4MultimodalFeatureEmbedding.forward": ["FloatTensor", "LongTensor", "ModelFeatureEmbedding", "Module", "None", "Tensor", "and", "any", "audio_attention_mask", "audio_embed", "audio_embed_sizes", "audio_embeds", "audio_input_features", "audio_projection_mode", "audio_token_id", "class", "config", "def", "elif", "else", "forward", "if", "image_attention_mask", "image_embed", "image_embeds", "image_pixel_values", "image_position_mask", "image_sizes", "image_token_id", "input_ids", "inputs_embeds", "is", "nn", "no_grad", "non_image_position_mask", "not", "return", "self", "speech", "torch", "unsqueeze", "vision", "vision_config", "with"], "phi4_multimodal/modeling_phi4_multimodal.py:Phi4MultimodalPreTrainedModel._init_weights": ["ModelImageEmbedding", "ModelPreTrainedModel", "PreTrainedModel", "_init_weights", "class", "def", "global_img_feature_extensor", "if", "init", "isinstance", "module", "self", "sub_img_feature_extensor", "super", "zeros_"], "phi4_multimodal/modeling_phi4_multimodal.py:Phi4MultimodalRotaryEmbedding.__init__": ["Callable", "False", "ModelConfig", "ModelRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "phi4_multimodal/modeling_phi4_multimodal.py:Phi4MultimodalRotaryEmbedding.compute_default_rope_parameters": ["ModelConfig", "ModelRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "get", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "partial_rotary_factor", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "phi4_multimodal/modeling_phi4_multimodal.py:Phi4MultimodalRotaryEmbedding.forward": ["False", "ModelRotaryEmbedding", "Module", "None", "and", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "phi4_multimodal/modeling_phi4_multimodal.py:Phi4MultimodalModel.__init__": ["Dropout", "Embedding", "False", "ModelConfig", "ModelDecoderLayer", "ModelFeatureEmbedding", "ModelModel", "ModelPreTrainedModel", "ModelRMSNorm", "ModelRotaryEmbedding", "ModuleList", "__init__", "class", "config", "def", "embd_pdrop", "embed_dropout", "embed_tokens", "embed_tokens_extend", "eps", "for", "gradient_checkpointing", "hidden_size", "in", "layer_idx", "layers", "nn", "norm", "num_hidden_layers", "pad_token_id", "padding_idx", "post_init", "range", "rms_norm_eps", "rotary_emb", "self", "super", "vocab_size"], "phi4_multimodal/modeling_phi4_multimodal.py:Phi4MultimodalModel.forward": ["BaseModelOutputWithPast", "Cache", "DynamicCache", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "ValueError", "You", "and", "arange", "attention_mask", "audio_attention_mask", "audio_embed_sizes", "audio_input_features", "bool", "cache_position", "causal_mask", "class", "config", "create_causal_mask", "create_sliding_window_causal_mask", "decoder_layer", "def", "device", "else", "embed_tokens", "embed_tokens_extend", "exactly", "for", "forward", "get_seq_length", "hidden_states", "if", "image_attention_mask", "image_pixel_values", "image_sizes", "in", "input_embeds", "input_ids", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layers", "mask_function", "must", "norm", "not", "of", "one", "or", "output_attentions", "output_hidden_states", "past_key_values", "past_seen_tokens", "position_embeddings", "position_ids", "r", "raise", "return", "rotary_emb", "self", "shape", "sliding_window", "specify", "torch", "unsqueeze", "use_cache"], "phi4_multimodal/modeling_phi4_multimodal.py:Phi4MultimodalForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "ModelForCausalLM", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "vocab_size"], "phi4_multimodal/modeling_phi4_multimodal.py:Phi4MultimodalForCausalLM.forward": ["BaseModelOutputWithPast", "Cache", "CausalLMOutputWithPast", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForCausalLM", "ModelPreTrainedModel", "None", "Tensor", "attention_mask", "attentions", "audio_attention_mask", "audio_embed_sizes", "audio_input_features", "bool", "cache_position", "class", "config", "def", "else", "forward", "hidden_states", "if", "image_attention_mask", "image_pixel_values", "image_sizes", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "position_ids", "r", "return", "self", "slice", "slice_indices", "torch", "use_cache", "vocab_size"], "phi4_multimodal/modeling_phi4_multimodal.py:Phi4MultimodalForCausalLM.prepare_inputs_for_generation": ["GenerationMixin", "ModelForCausalLM", "ModelPreTrainedModel", "None", "True", "and", "attention_mask", "audio_attention_mask", "audio_embed_sizes", "audio_input_features", "cache_position", "class", "config", "def", "if", "image_attention_mask", "image_pixel_values", "image_sizes", "input_ids", "inputs_embeds", "kwargs", "logits_to_keep", "model_inputs", "original_max_position_embeddings", "past_key_values", "past_length", "position_ids", "prepare_inputs_for_generation", "return", "rope_parameters", "self", "shape", "super", "use_cache"], "phimoe/modeling_phimoe.py:PhimoeRotaryEmbedding.__init__": ["Callable", "False", "ModelConfig", "ModelRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "phimoe/modeling_phimoe.py:PhimoeRotaryEmbedding.compute_default_rope_parameters": ["ModelConfig", "ModelRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "phimoe/modeling_phimoe.py:PhimoeRotaryEmbedding.forward": ["False", "ModelRotaryEmbedding", "Module", "None", "ValueError", "__class__", "__name__", "and", "attention_scaling", "but", "cat", "class", "config", "cos", "cpu", "def", "default", "device", "device_type", "dim", "does", "dtype", "else", "emb", "enabled", "expand", "f", "float", "forward", "freqs", "got", "if", "inv_freq", "inv_freq_expanded", "is", "isinstance", "layer", "layer_type", "long_mscale", "max", "maybe_autocast", "mps", "mscale", "nn", "not", "original_max_position_embeddings", "position_ids", "position_ids_expanded", "raise", "return", "rope_init_fn", "rope_parameters", "rope_type", "self", "seq_len", "shape", "short_mscale", "sin", "str", "support", "to", "torch", "transpose", "type", "types", "with", "x"], "phimoe/modeling_phimoe.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "phimoe/modeling_phimoe.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cos", "def", "k", "k_embed", "q", "q_embed", "return", "rotate_half", "sin", "unsqueeze", "unsqueeze_dim"], "phimoe/modeling_phimoe.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "phimoe/modeling_phimoe.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value", "value_states"], "phimoe/modeling_phimoe.py:PhimoeAttention.__init__": ["Linear", "ModelAttention", "ModelConfig", "Module", "True", "__init__", "attention_bias", "attention_dropout", "bias", "class", "config", "def", "getattr", "head_dim", "hidden_size", "int", "is_causal", "k_proj", "layer_idx", "nn", "num_attention_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "q_proj", "scaling", "self", "super", "v_proj"], "phimoe/modeling_phimoe.py:PhimoeAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "LongTensor", "ModelAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_attn_implementation", "apply_rotary_pos_emb", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "cache_kwargs", "cache_position", "class", "config", "contiguous", "cos", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "head_dim", "hidden_shape", "hidden_states", "if", "input_shape", "is", "k_proj", "key_states", "kwargs", "layer_idx", "nn", "not", "o_proj", "past_key_values", "position_embeddings", "q_proj", "query_states", "reshape", "return", "scaling", "self", "shape", "sin", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "view"], "phimoe/modeling_phimoe.py:PhimoeMultiplier.forward": ["Function", "ModelMultiplier", "Tensor", "autograd", "class", "ctx", "def", "forward", "mask_for_one", "masked_gates", "multiplier", "return", "save_for_backward", "scores", "selected_experts", "torch"], "phimoe/modeling_phimoe.py:PhimoeMultiplier.backward": ["Function", "ModelMultiplier", "None", "Tensor", "autograd", "backward", "class", "ctx", "def", "dim", "grad_at_output", "grad_at_scores_expanded", "index", "masked_gates", "mul", "multiplier", "return", "saved_tensors", "scatter_add_", "selected_experts", "src", "torch"], "phimoe/modeling_phimoe.py:PhimoeExperts.__init__": ["ACT2FN", "ModelConfig", "ModelExperts", "Module", "Parameter", "__init__", "act_fn", "class", "config", "def", "down_proj", "empty", "gate_up_proj", "hidden_act", "hidden_dim", "hidden_size", "intermediate_dim", "intermediate_size", "nn", "num_experts", "num_local_experts", "self", "super", "torch"], "phimoe/modeling_phimoe.py:PhimoeExperts.forward": ["ModelExperts", "Module", "None", "Tensor", "act_fn", "chunk", "class", "continue", "current_hidden_states", "current_state", "def", "dim", "down_proj", "dtype", "expert_hit", "expert_idx", "expert_mask", "final_hidden_states", "for", "forward", "functional", "gate", "gate_up_proj", "greater", "hidden_states", "if", "in", "index_add_", "linear", "nn", "no_grad", "nonzero", "num_classes", "num_experts", "one_hot", "permute", "return", "self", "sum", "to", "token_idx", "top_k_index", "top_k_pos", "top_k_weights", "torch", "up", "where", "with", "zeros_like"], "phimoe/modeling_phimoe.py:sparsemixer": ["Model", "ModelMultiplier", "True", "abs", "add", "alpha", "apply", "clamp", "concat", "def", "dim", "else", "empty_like", "exponential_", "factor", "float", "gather", "if", "index", "inf", "jitter_eps", "keepdim", "legacy_contiguous_format", "log", "logical_or", "mask_for_one", "mask_for_one_top2", "mask_logits_threshold", "masked_fill", "masked_gates", "masked_gates_top2", "masked_scores", "max", "max_ind", "max_scores", "memory_format", "min", "multiplier", "multiplier_o", "multiplier_top2", "multiplier_top2_o", "no_grad", "rand_like", "return", "scatter", "scores", "selected_experts", "selected_experts_top2", "softmax", "top_k", "torch", "training", "type_as", "uniform_", "unsqueeze", "with"], "phimoe/modeling_phimoe.py:PhimoeTopKRouter.__init__": ["False", "Linear", "ModelConfig", "ModelTopKRouter", "__init__", "bias", "class", "config", "def", "hidden_size", "input_jitter_noise", "nn", "num_local_experts", "router_jitter_noise", "self", "super"], "phimoe/modeling_phimoe.py:PhimoeTopKRouter.forward": ["Linear", "ModelTopKRouter", "Tensor", "and", "class", "def", "empty_like", "forward", "hidden_states", "if", "input_jitter_noise", "jitter_eps", "nn", "return", "router_jitter_noise", "router_logits", "routing_weights", "scatter_", "selected_experts", "self", "sparsemixer", "super", "torch", "training", "tuple", "uniform_", "zeros_like"], "phimoe/modeling_phimoe.py:PhimoeSparseMoeBlock.__init__": ["ModelExperts", "ModelSparseMoeBlock", "ModelTopKRouter", "Module", "__init__", "class", "config", "def", "experts", "ffn_dim", "hidden_dim", "hidden_size", "input_jitter_noise", "intermediate_size", "nn", "num_experts", "num_experts_per_tok", "num_local_experts", "router", "self", "super", "top_k"], "phimoe/modeling_phimoe.py:PhimoeSparseMoeBlock.forward": ["ModelSparseMoeBlock", "Module", "Tensor", "and", "batch_size", "class", "def", "empty_like", "experts", "final_hidden_states", "forward", "hidden_dim", "hidden_states", "if", "input_jitter_noise", "nn", "reshape", "return", "router", "routing_weights", "selected_experts", "self", "sequence_length", "shape", "torch", "training", "uniform_"], "phimoe/modeling_phimoe.py:PhimoeRMSNorm.__init__": ["ModelRMSNorm", "Module", "Parameter", "__init__", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "phimoe/modeling_phimoe.py:PhimoeRMSNorm.forward": ["ModelRMSNorm", "Module", "True", "class", "def", "dtype", "float32", "forward", "hidden_states", "input_dtype", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "phimoe/modeling_phimoe.py:PhimoeRMSNorm.extra_repr": ["ModelRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "variance_epsilon", "weight"], "phimoe/modeling_phimoe.py:PhimoeDecoderLayer.__init__": ["GradientCheckpointingLayer", "ModelAttention", "ModelConfig", "ModelDecoderLayer", "ModelRMSNorm", "ModelSparseMoeBlock", "__init__", "class", "config", "def", "eps", "hidden_size", "input_layernorm", "int", "layer_idx", "mlp", "post_attention_layernorm", "rms_norm_eps", "self", "self_attn", "super"], "phimoe/modeling_phimoe.py:PhimoeDecoderLayer.forward": ["Cache", "GradientCheckpointingLayer", "LongTensor", "ModelDecoderLayer", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "attention_mask", "cache_position", "class", "def", "forward", "hidden_states", "input_layernorm", "kwargs", "mlp", "past_key_values", "position_embeddings", "position_ids", "post_attention_layernorm", "residual", "return", "self", "self_attn", "torch", "tuple"], "phimoe/modeling_phimoe.py:PhimoePreTrainedModel._init_weights": ["ModelExperts", "ModelPreTrainedModel", "ModelTopKRouter", "PreTrainedModel", "_init_weights", "class", "config", "def", "down_proj", "elif", "gate_up_proj", "if", "init", "initializer_range", "isinstance", "mean", "module", "normal_", "self", "std", "super", "weight"], "phimoe/modeling_phimoe.py:PhimoeModel.__init__": ["Embedding", "False", "LayerNorm", "ModelConfig", "ModelDecoderLayer", "ModelModel", "ModelPreTrainedModel", "ModelRotaryEmbedding", "ModuleList", "True", "__init__", "class", "config", "def", "elementwise_affine", "embed_tokens", "eps", "for", "gradient_checkpointing", "hidden_size", "in", "layer_idx", "layers", "nn", "norm", "num_hidden_layers", "pad_token_id", "padding_idx", "post_init", "range", "rms_norm_eps", "rotary_emb", "self", "super", "vocab_size"], "phimoe/modeling_phimoe.py:PhimoeModel.forward": ["Cache", "DynamicCache", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "MoeModelOutputWithPast", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "and", "arange", "attention_mask", "bool", "cache_position", "causal_mask", "class", "config", "create_causal_mask", "create_sliding_window_causal_mask", "decoder_layer", "def", "device", "else", "embed_tokens", "exactly", "for", "forward", "get_seq_length", "hidden_states", "if", "in", "input_embeds", "input_ids", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layers", "mask_function", "must", "norm", "not", "num_hidden_layers", "of", "one", "or", "past_key_values", "past_seen_tokens", "position_embeddings", "position_ids", "raise", "return", "rotary_emb", "self", "shape", "sliding_window", "specify", "torch", "unsqueeze", "use_cache"], "phimoe/modeling_phimoe.py:load_balancing_loss_func": ["Model_balancing_loss_func", "None", "Tensor", "_", "attention_mask", "batch_size", "cat", "compute_device", "concatenated_gate_logits", "def", "device", "dim", "else", "expand", "expert_attention_mask", "expert_mask", "float", "for", "functional", "gate_logits", "if", "in", "int", "is", "isinstance", "layer_gate", "mean", "nn", "not", "num_experts", "num_hidden_layers", "one_hot", "or", "overall_loss", "r", "reshape", "return", "router_per_expert_attention_mask", "router_prob_per_expert", "routing_weights", "selected_experts", "sequence_length", "shape", "softmax", "sum", "to", "tokens_per_expert", "top_k", "topk", "torch", "tuple", "unsqueeze"], "phimoe/modeling_phimoe.py:PhimoeForCausalLM.__init__": ["GenerationMixin", "Linear", "ModelForCausalLM", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "lm_head_bias", "model", "nn", "num_experts", "num_experts_per_tok", "num_local_experts", "post_init", "router_aux_loss_coef", "self", "super", "vocab_size"], "phimoe/modeling_phimoe.py:PhimoeForCausalLM.forward": ["Cache", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForCausalLM", "ModelPreTrainedModel", "MoeCausalLMOutputWithPast", "MoeModelOutputWithPast", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "aux_loss", "bool", "cache_position", "class", "config", "def", "device", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "load_balancing_loss_func", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "num_experts", "num_experts_per_tok", "output_router_logits", "outputs", "past_key_values", "position_ids", "r", "return", "router_aux_loss_coef", "router_logits", "self", "slice", "slice_indices", "to", "torch", "use_cache", "vocab_size"], "phimoe/modeling_phimoe.py:PhimoeForCausalLM.prepare_inputs_for_generation": ["GenerationMixin", "ModelForCausalLM", "ModelPreTrainedModel", "None", "True", "and", "attention_mask", "cache_position", "class", "config", "def", "hasattr", "if", "input_ids", "inputs_embeds", "kwargs", "logits_to_keep", "model_inputs", "original_max_position_embeddings", "past_key_values", "past_length", "position_ids", "prepare_inputs_for_generation", "return", "self", "shape", "super", "use_cache"], "pix2struct/modeling_pix2struct.py:Pix2StructLayerNorm.__init__": ["ModelLayerNorm", "Module", "Parameter", "__init__", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "pix2struct/modeling_pix2struct.py:Pix2StructLayerNorm.forward": ["ModelLayerNorm", "Module", "True", "bfloat16", "class", "def", "dtype", "float16", "float32", "forward", "hidden_states", "if", "in", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "pix2struct/modeling_pix2struct.py:Pix2StructVisionEmbeddings.__init__": ["Dropout", "Embedding", "Linear", "ModelConfig", "ModelVisionEmbeddings", "Module", "None", "__init__", "class", "column_embedder", "config", "def", "dropout", "dropout_rate", "hidden_size", "nn", "patch_embed_hidden_size", "patch_projection", "row_embedder", "self", "seq_len", "super"], "pix2struct/modeling_pix2struct.py:Pix2StructVisionEmbeddings.forward": ["ModelVisionEmbeddings", "Module", "Tensor", "class", "col_embeddings", "col_indices", "column_embedder", "def", "dropout", "embeddings", "flattened_patches", "forward", "long", "nn", "patch_projection", "return", "row_embedder", "row_embeddings", "row_indices", "self", "torch"], "pix2struct/modeling_pix2struct.py:Pix2StructVisionAttention.__init__": ["False", "Linear", "ModelVisionAttention", "Module", "__init__", "attention_dropout", "bias", "class", "config", "d_kv", "def", "dropout", "gradient_checkpointing", "hidden_size", "inner_dim", "key", "key_value_proj_dim", "n_heads", "nn", "num_attention_heads", "output", "query", "self", "super", "value"], "pix2struct/modeling_pix2struct.py:Pix2StructVisionAttention.forward": ["False", "ModelVisionAttention", "Module", "None", "True", "and", "attention_mask", "attn_output", "attn_weights", "batch_size", "class", "contiguous", "def", "device", "dim", "dropout", "dtype", "elif", "finfo", "float32", "forward", "functional", "gradient_checkpointing", "hidden_states", "if", "inner_dim", "is", "is_torchdynamo_compiling", "key", "key_states", "key_value_proj_dim", "masked_fill", "matmul", "max", "min", "n_heads", "nn", "not", "ones", "output", "output_attentions", "outputs", "p", "position_bias", "position_bias_masked", "query", "query_states", "requires_grad", "return", "scores", "self", "seq_length", "shape", "softmax", "states", "tensor", "to", "to_projection_shape", "torch", "training", "transpose", "type_as", "value", "value_states", "view", "zeros"], "pix2struct/modeling_pix2struct.py:Pix2StructVisionMlp.__init__": ["ACT2FN", "Dropout", "False", "Linear", "ModelVisionConfig", "ModelVisionMlp", "Module", "__init__", "act", "bias", "class", "config", "d_ff", "def", "dense_act_fn", "dropout", "dropout_rate", "hidden_size", "nn", "self", "super", "wi_0", "wi_1", "wo"], "pix2struct/modeling_pix2struct.py:Pix2StructVisionMlp.forward": ["ModelVisionMlp", "Module", "Tensor", "act", "and", "class", "def", "dropout", "dtype", "forward", "hidden_gelu", "hidden_linear", "hidden_states", "if", "int8", "isinstance", "nn", "return", "self", "to", "torch", "weight", "wi_0", "wi_1", "wo"], "pix2struct/modeling_pix2struct.py:Pix2StructVisionLayer.__init__": ["GradientCheckpointingLayer", "ModelConfig", "ModelLayerNorm", "ModelVisionAttention", "ModelVisionLayer", "ModelVisionMlp", "None", "__init__", "attention", "chunk_size_feed_forward", "class", "config", "def", "eps", "hidden_size", "layer_norm_eps", "mlp", "pre_attention_layer_norm", "pre_mlp_layer_norm", "self", "seq_len_dim", "super"], "pix2struct/modeling_pix2struct.py:Pix2StructVisionLayer.forward": ["False", "GradientCheckpointingLayer", "ModelVisionLayer", "None", "Tensor", "attention", "attention_mask", "attention_output", "bool", "class", "def", "forward", "hidden_states", "layer_output", "mlp", "output_attentions", "outputs", "pre_attention_layer_norm", "pre_mlp_layer_norm", "residual", "return", "self", "self_attention_outputs", "torch", "tuple"], "pix2struct/modeling_pix2struct.py:Pix2StructVisionEncoder.__init__": ["False", "ModelVisionConfig", "ModelVisionEncoder", "ModelVisionLayer", "Module", "ModuleList", "None", "_", "__init__", "class", "config", "def", "for", "gradient_checkpointing", "in", "layer", "nn", "num_hidden_layers", "range", "self", "super"], "pix2struct/modeling_pix2struct.py:Pix2StructVisionEncoder.forward": ["BaseModelOutput", "False", "ModelVisionEncoder", "Module", "None", "Tensor", "True", "all_hidden_states", "all_self_attentions", "attention_mask", "attentions", "bool", "class", "def", "else", "enumerate", "for", "forward", "hidden_states", "i", "if", "in", "is", "last_hidden_state", "layer", "layer_module", "layer_outputs", "nn", "not", "output_attentions", "output_hidden_states", "return", "return_dict", "self", "torch", "tuple", "v"], "pix2struct/modeling_pix2struct.py:Pix2StructPreTrainedModel.dummy_inputs": ["DUMMY_INPUTS", "DUMMY_MASK", "ModelPreTrainedModel", "PreTrainedModel", "class", "decoder_attention_mask", "decoder_input_ids", "def", "dummy_inputs", "input_ids", "input_mask", "return", "self", "tensor", "torch"], "pix2struct/modeling_pix2struct.py:Pix2StructPreTrainedModel._init_weights": ["Conv2d", "Embedding", "False", "Linear", "ModelConfig", "ModelLayerNorm", "ModelPreTrainedModel", "ModelTextAttention", "ModelTextDenseGatedActDense", "ModelTextModel", "None", "PreTrainedModel", "_init_weights", "_is_hf_initialized", "and", "bias", "class", "config", "constant_", "d_ff", "d_kv", "def", "elif", "else", "factor", "getattr", "has_relative_attention_bias", "hasattr", "hidden_size", "if", "init", "initializer_factor", "initializer_range", "is", "isinstance", "key", "key_value_proj_dim", "lm_head", "mean", "module", "n_heads", "nn", "normal_", "not", "num_heads", "ones_", "output", "padding_idx", "query", "relative_attention_bias", "self", "std", "text_config", "trunc_normal_", "value", "weight", "wi_0", "wi_1", "wo", "zeros_"], "pix2struct/modeling_pix2struct.py:Pix2StructPreTrainedModel._shift_right": ["In", "Model", "ModelPreTrainedModel", "None", "PreTrainedModel", "See", "ValueError", "_shift_right", "be", "class", "clone", "config", "decoder_start_token_id", "def", "defined", "docs", "for", "has", "if", "information", "input_ids", "is", "it", "masked_fill_", "model", "more", "new_zeros", "pad_token_id", "raise", "return", "self", "set", "shape", "shifted_input_ids", "the", "to", "usually"], "pix2struct/modeling_pix2struct.py:Pix2StructVisionModel.__init__": ["ModelLayerNorm", "ModelPreTrainedModel", "ModelVisionConfig", "ModelVisionEmbeddings", "ModelVisionEncoder", "ModelVisionModel", "__init__", "class", "config", "def", "embeddings", "encoder", "eps", "hidden_size", "layer_norm_eps", "layernorm", "post_init", "self", "super"], "pix2struct/modeling_pix2struct.py:Pix2StructVisionModel.get_input_embeddings": ["ModelPreTrainedModel", "ModelVisionModel", "class", "def", "embeddings", "get_input_embeddings", "patch_projection", "return", "self"], "pix2struct/modeling_pix2struct.py:Pix2StructVisionModel.forward": ["BaseModelOutput", "BaseModelOutputWithPooling", "ModelPreTrainedModel", "ModelVisionModel", "None", "Tensor", "ValueError", "You", "attention_mask", "attentions", "bool", "class", "config", "def", "dim", "else", "embedding_output", "embeddings", "encoder", "encoder_outputs", "flattened_patches", "float", "forward", "have", "head_outputs", "hidden_states", "if", "is", "kwargs", "last_hidden_state", "layernorm", "not", "output_attentions", "output_hidden_states", "r", "raise", "return", "return_dict", "self", "sequence_output", "specify", "sum", "to", "torch", "tuple", "use_return_dict"], "pix2struct/modeling_pix2struct.py:Pix2StructTextDenseGatedActDense.__init__": ["ACT2FN", "Dropout", "False", "Linear", "ModelTextConfig", "ModelTextDenseGatedActDense", "Module", "__init__", "act", "bias", "class", "config", "d_ff", "def", "dense_act_fn", "dropout", "dropout_rate", "hidden_size", "nn", "self", "super", "wi_0", "wi_1", "wo"], "pix2struct/modeling_pix2struct.py:Pix2StructTextDenseGatedActDense.forward": ["ModelTextDenseGatedActDense", "Module", "Tensor", "act", "and", "class", "def", "dropout", "dtype", "forward", "hidden_gelu", "hidden_linear", "hidden_states", "if", "int8", "isinstance", "nn", "return", "self", "to", "torch", "weight", "wi_0", "wi_1", "wo"], "pix2struct/modeling_pix2struct.py:Pix2StructTextLayerFF.__init__": ["DenseReluDense", "Dropout", "ModelLayerNorm", "ModelTextConfig", "ModelTextDenseGatedActDense", "ModelTextLayerFF", "Module", "__init__", "class", "config", "def", "dropout", "dropout_rate", "eps", "hidden_size", "layer_norm", "layer_norm_epsilon", "nn", "self", "super"], "pix2struct/modeling_pix2struct.py:Pix2StructTextLayerFF.forward": ["DenseReluDense", "ModelTextLayerFF", "Module", "class", "def", "dropout", "forward", "forwarded_states", "hidden_states", "layer_norm", "nn", "return", "self"], "pix2struct/modeling_pix2struct.py:Pix2StructTextAttention.__init__": ["Embedding", "False", "Instantiating", "Linear", "ModelTextAttention", "ModelTextConfig", "Module", "None", "Please", "__class__", "__init__", "__name__", "a", "and", "bias", "caching", "call", "class", "config", "creating", "d_kv", "decoder", "def", "dropout", "dropout_rate", "during", "errors", "f", "forward", "gradient_checkpointing", "has_relative_attention_bias", "hidden_size", "if", "inner_dim", "int", "is", "key", "key_value_proj_dim", "layer_idx", "logger", "make", "n_heads", "nn", "not", "num_heads", "output", "passing", "provide", "query", "recommended", "relative_attention_bias", "relative_attention_max_distance", "relative_attention_num_buckets", "self", "super", "sure", "the", "this", "to", "used", "value", "warning_once", "when", "will", "without"], "pix2struct/modeling_pix2struct.py:Pix2StructTextAttention._relative_position_bucket": ["ModelTextAttention", "Module", "True", "_relative_position_bucket", "abs", "bidirectional", "class", "def", "else", "float", "full_like", "if", "is_small", "log", "long", "math", "max_distance", "max_exact", "min", "nn", "num_buckets", "relative_buckets", "relative_position", "relative_position_if_large", "return", "to", "torch", "where", "zeros_like"], "pix2struct/modeling_pix2struct.py:Pix2StructTextAttention.compute_bias": ["False", "ModelTextAttention", "Module", "None", "_relative_position_bucket", "arange", "bidirectional", "cache_position", "class", "compute_bias", "context_position", "def", "device", "dtype", "else", "if", "is", "key_length", "long", "max_distance", "memory_position", "nn", "num_buckets", "permute", "query_length", "relative_attention_bias", "relative_attention_max_distance", "relative_attention_num_buckets", "relative_position", "relative_position_bucket", "return", "self", "to", "torch", "unsqueeze", "values", "weight"], "pix2struct/modeling_pix2struct.py:Pix2StructTextAttention.forward": ["EncoderDecoderCache", "False", "ModelTextAttention", "Module", "None", "True", "and", "attn_output", "attn_weights", "batch_size", "cache_position", "causal_mask", "class", "compute_bias", "contiguous", "cross_attention_cache", "curr_past_key_values", "current_states", "def", "device", "dim", "dropout", "dtype", "else", "float", "forward", "functional", "get", "gradient_checkpointing", "has_relative_attention_bias", "hidden_states", "if", "inner_dim", "is", "is_cross_attention", "is_updated", "isinstance", "key", "key_length", "key_states", "key_value_proj_dim", "key_value_states", "keys", "layer_idx", "layers", "mask", "matmul", "n_heads", "nn", "not", "output", "output_attentions", "outputs", "p", "past_key_values", "position_bias", "position_bias_masked", "query", "query_length", "query_states", "real_seq_length", "requires_grad", "return", "scores", "self", "self_attention_cache", "seq_length", "shape", "softmax", "torch", "training", "transpose", "type_as", "update", "use_cache", "value", "value_states", "values", "view", "zeros"], "pix2struct/modeling_pix2struct.py:Pix2StructTextLayerSelfAttention.__init__": ["Dropout", "False", "ModelLayerNorm", "ModelTextAttention", "ModelTextLayerSelfAttention", "Module", "None", "__init__", "attention", "class", "config", "def", "dropout", "dropout_rate", "eps", "has_relative_attention_bias", "hidden_size", "int", "layer_idx", "layer_norm", "layer_norm_epsilon", "nn", "self", "super"], "pix2struct/modeling_pix2struct.py:Pix2StructTextLayerSelfAttention.forward": ["False", "ModelTextLayerSelfAttention", "Module", "None", "attention", "attention_mask", "attention_output", "cache_position", "class", "def", "dropout", "forward", "hidden_states", "layer_norm", "mask", "nn", "normed_hidden_states", "output_attentions", "outputs", "past_key_values", "position_bias", "return", "self", "use_cache"], "pix2struct/modeling_pix2struct.py:Pix2StructTextLayerCrossAttention.__init__": ["Dropout", "False", "ModelLayerNorm", "ModelTextAttention", "ModelTextLayerCrossAttention", "Module", "None", "__init__", "attention", "class", "config", "def", "dropout", "dropout_rate", "eps", "has_relative_attention_bias", "hidden_size", "int", "layer_idx", "layer_norm", "layer_norm_epsilon", "nn", "self", "super"], "pix2struct/modeling_pix2struct.py:Pix2StructTextLayerCrossAttention.forward": ["False", "ModelTextLayerCrossAttention", "Module", "None", "attention", "attention_mask", "attention_output", "cache_position", "class", "def", "dropout", "forward", "hidden_states", "key_value_states", "layer_norm", "layer_output", "mask", "nn", "normed_hidden_states", "output_attentions", "outputs", "past_key_values", "position_bias", "query_length", "return", "self", "use_cache"], "pix2struct/modeling_pix2struct.py:Pix2StructTextBlock.__init__": ["False", "GradientCheckpointingLayer", "ModelTextBlock", "ModelTextLayerCrossAttention", "ModelTextLayerFF", "ModelTextLayerSelfAttention", "None", "__init__", "class", "config", "def", "encoder_decoder_attention", "has_relative_attention_bias", "int", "layer_idx", "mlp", "self", "self_attention", "super"], "pix2struct/modeling_pix2struct.py:Pix2StructTextBlock.forward": ["False", "GradientCheckpointingLayer", "ModelTextBlock", "None", "True", "and", "any", "attention_mask", "attention_outputs", "cache_position", "clamp", "clamp_value", "class", "cross_attention_outputs", "def", "do_cross_attention", "dtype", "encoder_attention_mask", "encoder_decoder_attention", "encoder_decoder_position_bias", "encoder_hidden_states", "finfo", "float16", "forward", "hidden_states", "if", "is", "isinf", "key_value_states", "max", "min", "mlp", "not", "output_attentions", "outputs", "past_key_values", "position_bias", "query_length", "return", "return_dict", "self", "self_attention", "self_attention_outputs", "torch", "use_cache"], "pix2struct/modeling_pix2struct.py:Pix2StructTextModel.__init__": ["Dropout", "Embedding", "False", "Linear", "ModelLayerNorm", "ModelPreTrainedModel", "ModelTextBlock", "ModelTextModel", "ModuleList", "__init__", "bias", "bool", "class", "config", "def", "dropout", "dropout_rate", "embed_tokens", "eps", "final_layer_norm", "for", "gradient_checkpointing", "has_relative_attention_bias", "hidden_size", "i", "in", "layer", "layer_idx", "layer_norm_epsilon", "lm_head", "nn", "num_layers", "post_init", "range", "self", "super", "vocab_size"], "pix2struct/modeling_pix2struct.py:Pix2StructTextModel.set_input_embeddings": ["ModelPreTrainedModel", "ModelTextModel", "class", "def", "embed_tokens", "new_embeddings", "self", "set_input_embeddings"], "pix2struct/modeling_pix2struct.py:Pix2StructTextModel.forward": ["Cache", "CausalLMOutputWithCrossAttentions", "CrossEntropyLoss", "DynamicCache", "EncoderDecoderCache", "False", "FloatTensor", "LongTensor", "ModelPreTrainedModel", "ModelTextModel", "None", "Setting", "True", "ValueError", "You", "_", "_update_causal_mask", "all_attentions", "all_cross_attentions", "all_hidden_states", "and", "arange", "assert", "at", "attention_mask", "attentions", "batch_size", "bool", "both", "cache_position", "cannot", "causal_mask", "checkpointing", "class", "config", "contiguous", "cross_attentions", "decoder_input_ids", "decoder_inputs_embeds", "def", "device", "dropout", "dtype", "either", "elif", "else", "embed_tokens", "embeddings", "encoder_attention_mask", "encoder_batch_size", "encoder_decoder_position_bias", "encoder_extended_attention_mask", "encoder_hidden_shape", "encoder_hidden_states", "encoder_sequence_length", "enumerate", "final_layer_norm", "finfo", "for", "forward", "get_seq_length", "gradient", "gradient_checkpointing", "have", "hidden_states", "i", "if", "ignore_index", "in", "incompatible", "initialize", "input_ids", "input_shape", "inputs_embeds", "invert_attention_mask", "is", "is_decoder", "is_encoder_decoder", "isinstance", "kwargs", "labels", "layer", "layer_module", "layer_outputs", "lm_head", "logger", "logits", "loss", "loss_fct", "mask_seq_length", "mean", "min", "model", "nn", "not", "ones", "or", "output_attentions", "output_hidden_states", "past_key_values", "past_key_values_length", "position_bias", "r", "raise", "reduction", "return", "return_dict", "same", "self", "self_attention_cache", "seq_length", "size", "specify", "the", "time", "to", "token", "torch", "training", "tuple", "use_cache", "use_return_dict", "v", "valid", "view", "warning", "with"], "pix2struct/modeling_pix2struct.py:Pix2StructTextModel._update_causal_mask": ["AttentionMaskConverter", "BlockMask", "Cache", "False", "ModelPreTrainedModel", "ModelTextModel", "None", "Tensor", "Union", "_attn_implementation", "_ignore_causal_mask_sdpa", "_prepare_4d_causal_attention_mask_with_cache_position", "_unmask_unattended", "_update_causal_mask", "and", "any", "attention_mask", "batch_size", "bool", "cache_position", "causal_mask", "class", "config", "cuda", "def", "device", "dtype", "else", "finfo", "flash_attention_2", "flex_attention", "get_max_cache_shape", "get_seq_length", "if", "in", "input_tensor", "inputs_embeds", "is", "is_compileable", "is_training", "isinstance", "make_flex_block_causal_mask", "min", "min_dtype", "not", "npu", "output_attentions", "past_key_values", "past_key_values_length", "past_seen_tokens", "return", "sdpa", "self", "sequence_length", "shape", "target_length", "torch", "training", "type", "using_compilable_cache", "xpu"], "pix2struct/modeling_pix2struct.py:Pix2StructTextModel._prepare_4d_causal_attention_mask_with_cache_position": ["ModelPreTrainedModel", "ModelTextModel", "None", "Tensor", "_prepare_4d_causal_attention_mask_with_cache_position", "and", "arange", "attention_mask", "batch_size", "cache_position", "causal_mask", "class", "clone", "def", "device", "diagonal", "dim", "dtype", "else", "expand", "fill_value", "finfo", "full", "if", "int", "is", "kwargs", "mask_length", "masked_fill", "min", "min_dtype", "not", "padding_mask", "reshape", "return", "sequence_length", "shape", "target_length", "to", "torch", "triu"], "pix2struct/modeling_pix2struct.py:Pix2StructForConditionalGeneration.__init__": ["GenerationMixin", "ModelConfig", "ModelForConditionalGeneration", "ModelPreTrainedModel", "ModelTextModel", "ModelVisionModel", "__init__", "class", "config", "decoder", "def", "encoder", "is_vqa", "post_init", "self", "super", "text_config", "vision_config"], "pix2struct/modeling_pix2struct.py:Pix2StructForConditionalGeneration.get_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "decoder", "def", "get_input_embeddings", "return", "self"], "pix2struct/modeling_pix2struct.py:Pix2StructForConditionalGeneration.set_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "decoder", "def", "new_embeddings", "self", "set_input_embeddings"], "pix2struct/modeling_pix2struct.py:Pix2StructForConditionalGeneration.get_output_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "Module", "class", "decoder", "def", "get_output_embeddings", "nn", "return", "self"], "pix2struct/modeling_pix2struct.py:Pix2StructForConditionalGeneration.set_output_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "decoder", "def", "new_embeddings", "self", "set_output_embeddings"], "pix2struct/modeling_pix2struct.py:Pix2StructForConditionalGeneration.forward": ["BaseModelOutput", "BoolTensor", "Cache", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Seq2SeqLMOutput", "Seq2SeqModelOutput", "Tensor", "_shift_right", "and", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "cross_attentions", "decoder", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_input_ids", "decoder_inputs_embeds", "decoder_outputs", "def", "elif", "else", "encoder", "encoder_attention_mask", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "flattened_patches", "float", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "len", "logits", "loss", "ne", "not", "output_attentions", "output_hidden_states", "pad_token_id", "past_key_values", "r", "return", "return_dict", "self", "text_config", "torch", "tuple", "use_cache", "use_return_dict"], "pixio/modeling_pixio.py:PixioPatchEmbeddings.__init__": ["Conv2d", "Iterable", "ModelConfig", "ModelPatchEmbeddings", "Module", "__init__", "abc", "class", "collections", "config", "def", "else", "hidden_size", "if", "image_size", "isinstance", "kernel_size", "nn", "num_channels", "num_patches", "patch_size", "projection", "self", "stride", "super"], "pixio/modeling_pixio.py:PixioPatchEmbeddings.forward": ["Expected", "False", "Input", "Make", "ModelPatchEmbeddings", "Module", "Tensor", "ValueError", "batch_size", "bool", "but", "channel", "class", "configuration", "def", "dimension", "doesn", "embeddings", "f", "flatten", "forward", "got", "height", "if", "image", "image_size", "in", "interpolate_pos_encoding", "match", "model", "nn", "not", "num_channels", "of", "one", "or", "pixel", "pixel_values", "projection", "raise", "return", "self", "set", "shape", "size", "sure", "t", "that", "the", "torch", "transpose", "values", "width", "with"], "pixio/modeling_pixio.py:PixioEmbeddings.__init__": ["Dropout", "ModelConfig", "ModelEmbeddings", "ModelPatchEmbeddings", "Module", "None", "Parameter", "__init__", "class", "cls_token", "config", "def", "dropout", "hidden_dropout_prob", "hidden_size", "mask_token", "n_cls_tokens", "nn", "num_patches", "patch_embeddings", "patch_size", "position_embeddings", "randn", "self", "super", "torch"], "pixio/modeling_pixio.py:PixioEmbeddings.interpolate_pos_encoding": ["False", "ModelEmbeddings", "Module", "Tensor", "align_corners", "and", "bicubic", "cat", "class", "class_pos_embed", "def", "dim", "dtype", "embeddings", "float32", "functional", "height", "if", "int", "interpolate", "interpolate_pos_encoding", "is_tracing", "mode", "n_cls_tokens", "new_height", "new_width", "nn", "not", "num_patches", "num_positions", "patch_pos_embed", "patch_size", "permute", "position_embeddings", "reshape", "return", "self", "shape", "size", "sqrt_num_positions", "target_dtype", "to", "torch", "view", "width"], "pixio/modeling_pixio.py:PixioEmbeddings.forward": ["ModelEmbeddings", "Module", "Tensor", "_", "batch_size", "cat", "class", "cls_token", "cls_tokens", "def", "dim", "dropout", "dtype", "embeddings", "expand", "forward", "height", "interpolate_pos_encoding", "nn", "patch_embeddings", "pixel_values", "projection", "return", "self", "shape", "target_dtype", "to", "torch", "weight", "width"], "pixio/modeling_pixio.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "contiguous", "def", "dim", "dropout", "float", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "shape", "size", "softmax", "torch", "training", "transpose", "value"], "pixio/modeling_pixio.py:PixioSelfAttention.__init__": ["False", "Linear", "ModelConfig", "ModelSelfAttention", "Module", "The", "ValueError", "__init__", "a", "all_head_size", "and", "attention", "attention_head_size", "attention_probs_dropout_prob", "bias", "class", "config", "def", "dropout_prob", "embedding_size", "f", "hasattr", "heads", "hidden", "hidden_size", "if", "int", "is", "is_causal", "key", "multiple", "nn", "not", "num_attention_heads", "number", "of", "qkv_bias", "query", "raise", "scaling", "self", "size", "super", "the", "value"], "pixio/modeling_pixio.py:PixioSelfAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "ModelSelfAttention", "Module", "None", "Tensor", "_attn_implementation", "all_head_size", "attention_head_size", "attention_interface", "attention_probs", "batch_size", "class", "config", "context_layer", "def", "dropout", "dropout_prob", "eager", "eager_attention_forward", "else", "forward", "hidden_states", "if", "is_causal", "key", "key_layer", "new_context_layer_shape", "new_shape", "nn", "not", "num_attention_heads", "query", "query_layer", "reshape", "return", "scaling", "self", "shape", "size", "torch", "training", "transpose", "tuple", "value", "value_layer", "view"], "pixio/modeling_pixio.py:PixioSelfOutput.__init__": ["Dropout", "Linear", "ModelConfig", "ModelSelfOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "hidden_dropout_prob", "hidden_size", "nn", "self", "super"], "pixio/modeling_pixio.py:PixioSelfOutput.forward": ["ModelSelfOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "pixio/modeling_pixio.py:PixioAttention.__init__": ["ModelAttention", "ModelConfig", "ModelSelfAttention", "ModelSelfOutput", "Module", "__init__", "attention", "class", "config", "def", "nn", "output", "self", "super"], "pixio/modeling_pixio.py:PixioAttention.forward": ["ModelAttention", "Module", "Tensor", "_", "attention", "class", "def", "forward", "hidden_states", "nn", "output", "return", "self", "self_attn_output", "torch"], "pixio/modeling_pixio.py:drop_path": ["False", "Model_path", "Model_prob", "Tensor", "bool", "def", "device", "div", "dtype", "float", "floor_", "if", "input", "keep_prob", "ndim", "not", "or", "output", "rand", "random_tensor", "return", "shape", "torch", "training"], "pixio/modeling_pixio.py:PixioDropPath.__init__": ["ModelDropPath", "Module", "None", "__init__", "class", "def", "drop_prob", "float", "nn", "self", "super"], "pixio/modeling_pixio.py:PixioDropPath.forward": ["ModelDropPath", "Module", "Tensor", "class", "def", "drop_path", "drop_prob", "forward", "hidden_states", "nn", "return", "self", "torch", "training"], "pixio/modeling_pixio.py:PixioDropPath.extra_repr": ["ModelDropPath", "Module", "class", "def", "drop_prob", "extra_repr", "f", "nn", "p", "return", "self", "str"], "pixio/modeling_pixio.py:PixioMLP.__init__": ["ACT2FN", "Linear", "ModelMLP", "Module", "None", "True", "__init__", "activation", "bias", "class", "config", "def", "else", "fc1", "fc2", "hidden_act", "hidden_features", "hidden_size", "if", "in_features", "int", "isinstance", "mlp_ratio", "nn", "out_features", "self", "str", "super"], "pixio/modeling_pixio.py:PixioMLP.forward": ["ModelMLP", "Module", "Tensor", "activation", "class", "def", "fc1", "fc2", "forward", "hidden_state", "nn", "return", "self", "torch"], "pixio/modeling_pixio.py:PixioLayer.__init__": ["GradientCheckpointingLayer", "Identity", "LayerNorm", "ModelAttention", "ModelConfig", "ModelDropPath", "ModelLayer", "ModelMLP", "None", "__init__", "attention", "class", "config", "def", "drop_path", "drop_path_rate", "else", "eps", "hidden_size", "if", "layer_norm_eps", "mlp", "nn", "norm1", "norm2", "self", "super"], "pixio/modeling_pixio.py:PixioLayer.forward": ["GradientCheckpointingLayer", "ModelLayer", "Tensor", "attention", "class", "def", "drop_path", "forward", "hidden_states", "hidden_states_norm", "layer_output", "mlp", "norm1", "norm2", "return", "self", "self_attention_output", "torch"], "pixio/modeling_pixio.py:PixioEncoder.__init__": ["False", "ModelConfig", "ModelEncoder", "ModelLayer", "Module", "ModuleList", "_", "__init__", "class", "config", "def", "for", "gradient_checkpointing", "in", "layer", "nn", "num_hidden_layers", "range", "self", "super"], "pixio/modeling_pixio.py:PixioEncoder.forward": ["BaseModelOutput", "False", "ModelEncoder", "Module", "None", "Tensor", "all_hidden_states", "append", "bool", "class", "def", "else", "enumerate", "for", "forward", "hidden_states", "i", "if", "in", "last_hidden_state", "layer", "layer_module", "nn", "output_hidden_states", "return", "self", "torch", "tuple"], "pixio/modeling_pixio.py:PixioPreTrainedModel._init_weights": ["Conv2d", "LayerNorm", "Linear", "ModelEmbeddings", "ModelPreTrainedModel", "None", "PreTrainedModel", "_init_weights", "bias", "class", "cls_token", "config", "def", "elif", "if", "init", "initializer_range", "is", "isinstance", "mask_token", "mean", "module", "nn", "not", "ones_", "position_embeddings", "self", "std", "trunc_normal_", "weight", "zeros_"], "pixio/modeling_pixio.py:PixioModel.__init__": ["LayerNorm", "ModelConfig", "ModelEmbeddings", "ModelEncoder", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "config", "def", "embeddings", "encoder", "eps", "hidden_size", "layer_norm_eps", "layernorm", "nn", "post_init", "self", "super"], "pixio/modeling_pixio.py:PixioModel.get_input_embeddings": ["ModelModel", "ModelPatchEmbeddings", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "patch_embeddings", "return", "self"], "pixio/modeling_pixio.py:PixioModel.forward": ["BaseModelOutput", "BaseModelOutputWithPooling", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "ValueError", "You", "bool", "class", "config", "def", "dim", "embedding_output", "embeddings", "encoder", "encoder_outputs", "forward", "have", "hidden_states", "if", "is", "kwargs", "last_hidden_state", "layernorm", "mean", "n_cls_tokens", "output_hidden_states", "pixel_values", "pooled_output", "pooler_output", "raise", "return", "self", "sequence_output", "specify", "to", "torch"], "pixio/modeling_pixio.py:PixioBackbone.__init__": ["BackboneMixin", "LayerNorm", "ModelBackbone", "ModelEmbeddings", "ModelEncoder", "ModelPreTrainedModel", "_", "__init__", "_init_backbone", "class", "config", "def", "embeddings", "encoder", "eps", "for", "hidden_size", "in", "layer_norm_eps", "layernorm", "nn", "num_features", "num_hidden_layers", "post_init", "range", "self", "super"], "pixio/modeling_pixio.py:PixioBackbone.get_input_embeddings": ["BackboneMixin", "ModelBackbone", "ModelPatchEmbeddings", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "patch_embeddings", "return", "self"], "pixio/modeling_pixio.py:PixioBackbone.forward": ["BackboneMixin", "BackboneOutput", "BaseModelOutput", "ModelBackbone", "ModelPreTrainedModel", "None", "Tensor", "True", "_", "append", "apply_layernorm", "batch_size", "bool", "class", "config", "contiguous", "def", "else", "embedding_output", "embeddings", "encoder", "feature_maps", "for", "forward", "height", "hidden_state", "hidden_states", "if", "in", "is", "kwargs", "layernorm", "n_cls_tokens", "out_features", "output", "output_hidden_states", "patch_size", "permute", "pixel_values", "r", "reshape", "reshape_hidden_states", "return", "self", "shape", "stage", "stage_names", "torch", "tuple", "width", "zip"], "pixtral/modeling_pixtral.py:position_ids_in_meshgrid": ["Model_ids_in_meshgrid", "Models", "append", "arange", "cat", "chunk", "def", "dim", "for", "h_grid", "height", "ids", "ij", "in", "indexing", "max_width", "mesh", "meshgrid", "patch", "patch_embeds_list", "reshape", "return", "shape", "stack", "torch", "v_grid", "width"], "pixtral/modeling_pixtral.py:PixtralRotaryEmbedding.__init__": ["Callable", "False", "ModelRotaryEmbedding", "ModelVisionConfig", "Module", "None", "RoPE", "ValueError", "__class__", "__init__", "__name__", "attention_scaling", "but", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "does", "f", "got", "if", "inv_freq", "layer_type", "nn", "non", "not", "original_inv_freq", "persistent", "raise", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super", "support"], "pixtral/modeling_pixtral.py:PixtralRotaryEmbedding.compute_default_rope_parameters": ["ModelRotaryEmbedding", "ModelVisionConfig", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "cat", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "float", "freqs", "freqs_h", "freqs_w", "getattr", "h", "head_dim", "hidden_size", "image_size", "int", "inv_freq", "max_patches_per_side", "nn", "num_attention_heads", "or", "outer", "patch_size", "repeat", "reshape", "return", "rope_parameters", "rope_theta", "seq_len", "torch", "tuple", "w"], "pixtral/modeling_pixtral.py:PixtralRotaryEmbedding.forward": ["False", "ModelRotaryEmbedding", "Module", "and", "class", "cos", "cpu", "def", "device", "device_type", "dtype", "else", "emb", "enabled", "forward", "freqs", "if", "inv_freq", "isinstance", "maybe_autocast", "mps", "nn", "position_ids", "return", "self", "sin", "str", "to", "type", "with", "x"], "pixtral/modeling_pixtral.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "pixtral/modeling_pixtral.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cos", "def", "k", "k_embed", "q", "q_embed", "return", "rotate_half", "sin", "unsqueeze", "unsqueeze_dim"], "pixtral/modeling_pixtral.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "attention_mask", "attn_output", "attn_weights", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "softmax", "to", "torch", "training", "transpose", "value"], "pixtral/modeling_pixtral.py:PixtralAttention.__init__": ["False", "Linear", "ModelAttention", "Module", "__init__", "attention_dropout", "bias", "class", "config", "def", "dropout", "embed_dim", "head_dim", "hidden_size", "is_causal", "k_proj", "nn", "num_attention_heads", "num_heads", "o_proj", "q_proj", "scaling", "self", "super", "v_proj"], "pixtral/modeling_pixtral.py:PixtralAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "False", "FlashAttentionKwargs", "ModelAttention", "Module", "None", "Tensor", "True", "Unpack", "_", "_attn_implementation", "apply_rotary_pos_emb", "attention_interface", "attention_mask", "attn_output", "attn_weights", "batch_size", "bool", "class", "config", "contiguous", "cos", "def", "device", "dropout", "eager", "eager_attention_forward", "else", "flash_attention_2", "forward", "head_dim", "hidden_states", "if", "k_proj", "key_states", "kwargs", "nn", "non_blocking", "not", "num_heads", "o_proj", "output_attentions", "patches", "position_embeddings", "position_ids", "q_proj", "query_states", "reshape", "return", "scaling", "self", "sin", "size", "to", "torch", "training", "transpose", "tuple", "unsqueeze_dim", "v_proj", "value_states", "view"], "pixtral/modeling_pixtral.py:PixtralMLP.__init__": ["ACT2FN", "False", "Linear", "ModelMLP", "Module", "__init__", "act_fn", "bias", "class", "config", "def", "down_proj", "gate_proj", "hidden_act", "hidden_size", "intermediate_size", "nn", "self", "super", "up_proj"], "pixtral/modeling_pixtral.py:PixtralMLP.forward": ["ModelMLP", "Module", "act_fn", "class", "def", "down_proj", "forward", "gate_proj", "nn", "return", "self", "up_proj", "x"], "pixtral/modeling_pixtral.py:PixtralRMSNorm.__init__": ["ModelRMSNorm", "Module", "Parameter", "__init__", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "pixtral/modeling_pixtral.py:PixtralRMSNorm.forward": ["ModelRMSNorm", "Module", "True", "class", "def", "dtype", "float32", "forward", "hidden_states", "input_dtype", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "pixtral/modeling_pixtral.py:PixtralRMSNorm.extra_repr": ["ModelRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "variance_epsilon", "weight"], "pixtral/modeling_pixtral.py:PixtralAttentionLayer.__init__": ["GradientCheckpointingLayer", "ModelAttention", "ModelAttentionLayer", "ModelMLP", "ModelRMSNorm", "__init__", "attention", "attention_norm", "class", "config", "def", "eps", "feed_forward", "ffn_norm", "hidden_size", "self", "super"], "pixtral/modeling_pixtral.py:PixtralAttentionLayer.forward": ["FlashAttentionKwargs", "FloatTensor", "GradientCheckpointingLayer", "ModelAttentionLayer", "None", "Tensor", "Unpack", "attention", "attention_mask", "attention_norm", "attn_weights", "bool", "class", "def", "feed_forward", "ffn_norm", "forward", "hidden_states", "if", "kwargs", "output_attentions", "outputs", "position_embeddings", "residual", "return", "self", "torch", "tuple"], "pixtral/modeling_pixtral.py:PixtralTransformer.__init__": ["False", "ModelAttentionLayer", "ModelTransformer", "Module", "ModuleList", "_", "__init__", "append", "class", "config", "def", "for", "gradient_checkpointing", "in", "layers", "nn", "num_hidden_layers", "range", "self", "super", "torch"], "pixtral/modeling_pixtral.py:PixtralTransformer.forward": ["BaseModelOutput", "FlashAttentionKwargs", "ModelTransformer", "Module", "None", "Tensor", "Unpack", "all_attentions", "attention_mask", "attentions", "bool", "class", "config", "def", "else", "encoder_layer", "encoder_states", "for", "forward", "hidden_states", "if", "in", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layer_outputs", "layers", "nn", "not", "output_attentions", "output_hidden_states", "position_embeddings", "r", "return", "return_dict", "self", "torch", "tuple", "use_return_dict", "v"], "pixtral/modeling_pixtral.py:generate_block_attention_mask": ["Model_block_attention_mask", "None", "block_end_idx", "block_start_idx", "causal_mask", "cumsum", "d_min", "def", "device", "dtype", "end", "expand", "fill_value", "finfo", "for", "full", "in", "min", "patch_embeds_list", "return", "seq_len", "shape", "start", "tensor", "torch", "zip"], "pixtral/modeling_pixtral.py:PixtralVisionModel.__init__": ["Conv2d", "False", "ModelPreTrainedModel", "ModelRMSNorm", "ModelRotaryEmbedding", "ModelTransformer", "ModelVisionModel", "__init__", "bias", "class", "config", "def", "eps", "hidden_size", "in_channels", "kernel_size", "ln_pre", "nn", "num_channels", "out_channels", "patch_conv", "patch_positional_embedding", "patch_size", "post_init", "self", "stride", "super", "transformer"], "pixtral/modeling_pixtral.py:PixtralVisionModel.get_input_embeddings": ["ModelPreTrainedModel", "ModelVisionModel", "class", "def", "get_input_embeddings", "patch_conv", "return", "self"], "pixtral/modeling_pixtral.py:PixtralVisionModel.forward": ["BaseModelOutput", "FlashAttentionKwargs", "ModelPreTrainedModel", "ModelVisionModel", "None", "T", "Tensor", "True", "Unpack", "_", "_attn_implementation", "args", "attention_mask", "batch_size", "bool", "cat", "class", "config", "def", "dim", "else", "embed", "flash_attention_2", "flatten", "for", "forward", "generate_block_attention_mask", "height", "if", "image_size", "image_sizes", "in", "is", "kwargs", "ln_pre", "max_width", "output_attentions", "output_hidden_states", "p", "patch_conv", "patch_embeds", "patch_embeds_list", "patch_positional_embedding", "patch_size", "pixel_values", "position_embeddings", "position_ids", "position_ids_in_meshgrid", "return", "return_dict", "self", "shape", "size", "torch", "transformer", "tuple", "unsqueeze", "width", "zip"], "plbart/modeling_plbart.py:PLBartScaledWordEmbedding.__init__": ["Embedding", "Model", "None", "__init__", "class", "def", "embed_scale", "embedding_dim", "float", "int", "nn", "num_embeddings", "padding_idx", "self", "super"], "plbart/modeling_plbart.py:PLBartScaledWordEmbedding.forward": ["Embedding", "Model", "Tensor", "class", "def", "embed_scale", "forward", "input_ids", "nn", "return", "self", "super", "torch"], "plbart/modeling_plbart.py:PLBartPreTrainedModel._init_weights": ["Model", "ModelForConditionalGeneration", "PreTrainedModel", "_init_weights", "class", "def", "final_logits_bias", "if", "init", "isinstance", "module", "self", "super", "zeros_"], "plbart/modeling_plbart.py:PLBartLearnedPositionalEmbedding.__init__": ["Embedding", "Model", "__init__", "class", "def", "embedding_dim", "int", "nn", "num_embeddings", "offset", "self", "super"], "plbart/modeling_plbart.py:PLBartLearnedPositionalEmbedding.forward": ["Embedding", "Model", "None", "Tensor", "arange", "bsz", "class", "def", "device", "dtype", "else", "expand", "forward", "if", "input_ids", "int", "is", "long", "nn", "offset", "past_key_values_length", "position_ids", "return", "self", "seq_len", "shape", "super", "torch", "unsqueeze", "weight"], "plbart/modeling_plbart.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "contiguous", "def", "dim", "dropout", "float", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "shape", "size", "softmax", "torch", "training", "transpose", "value"], "plbart/modeling_plbart.py:PLBartAttention.__init__": ["False", "Instantiating", "Linear", "Model", "ModelConfig", "Module", "None", "Please", "True", "ValueError", "__class__", "__init__", "__name__", "a", "and", "be", "bias", "bool", "by", "caching", "call", "class", "config", "creating", "decoder", "def", "divisible", "dropout", "during", "embed_dim", "errors", "f", "float", "forward", "got", "head_dim", "if", "int", "is", "is_causal", "is_decoder", "k_proj", "layer_idx", "lead", "logger", "make", "must", "nn", "not", "num_heads", "out_proj", "passing", "provide", "q_proj", "raise", "recommended", "scaling", "self", "super", "sure", "the", "this", "to", "used", "v_proj", "warning_once", "when", "will", "without"], "plbart/modeling_plbart.py:PLBartAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "EncoderDecoderCache", "False", "FlashAttentionKwargs", "Model", "Module", "None", "Tensor", "True", "Unpack", "_attn_implementation", "and", "attention_interface", "attention_mask", "attn_output", "attn_weights", "bool", "bsz", "cache_position", "class", "config", "contiguous", "cross_attention_cache", "curr_past_key_values", "current_states", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "get", "head_dim", "hidden_states", "if", "is", "is_cross_attention", "is_updated", "isinstance", "k_proj", "key_states", "key_value_states", "keys", "kv_input_shape", "kwargs", "layer_idx", "layers", "nn", "not", "out_proj", "output_attentions", "past_key_values", "q_input_shape", "q_proj", "query_states", "reshape", "return", "scaling", "self", "self_attention_cache", "shape", "src_len", "tgt_len", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "values", "view"], "plbart/modeling_plbart.py:PLBartEncoderLayer.__init__": ["ACT2FN", "GradientCheckpointingLayer", "LayerNorm", "Linear", "Model", "ModelAttention", "ModelConfig", "None", "__init__", "activation_dropout", "activation_fn", "activation_function", "attention_dropout", "class", "config", "d_model", "def", "dropout", "embed_dim", "encoder_attention_heads", "encoder_ffn_dim", "fc1", "fc2", "final_layer_norm", "int", "layer_idx", "nn", "num_heads", "self", "self_attn", "self_attn_layer_norm", "super"], "plbart/modeling_plbart.py:PLBartEncoderLayer.forward": ["False", "FloatTensor", "GradientCheckpointingLayer", "Model", "None", "activation_dropout", "activation_fn", "and", "any", "attention_mask", "attn_weights", "bool", "clamp", "clamp_value", "class", "def", "dropout", "dtype", "fc1", "fc2", "final_layer_norm", "finfo", "float16", "forward", "functional", "hidden_states", "if", "isinf", "isnan", "max", "min", "nn", "or", "output_attentions", "outputs", "p", "residual", "return", "self", "self_attn", "self_attn_layer_norm", "torch", "training", "tuple"], "plbart/modeling_plbart.py:PLBartEncoder.__init__": ["False", "LayerNorm", "Model", "ModelConfig", "ModelLayer", "ModelLearnedPositionalEmbedding", "ModelPreTrainedModel", "ModelScaledWordEmbedding", "ModuleList", "__init__", "class", "config", "d_model", "def", "dropout", "else", "embed_dim", "embed_positions", "embed_scale", "embed_tokens", "encoder_layerdrop", "encoder_layers", "for", "gradient_checkpointing", "i", "if", "in", "layer_idx", "layerdrop", "layernorm_embedding", "layers", "math", "max_position_embeddings", "max_source_positions", "nn", "pad_token_id", "padding_idx", "post_init", "range", "scale_embedding", "self", "sqrt", "super", "vocab_size"], "plbart/modeling_plbart.py:PLBartEncoder.forward": ["BaseModelOutput", "False", "FloatTensor", "LongTensor", "Model", "ModelPreTrainedModel", "None", "Tensor", "True", "ValueError", "You", "all_attentions", "and", "at", "attention_mask", "attentions", "bool", "both", "cannot", "class", "config", "create_bidirectional_mask", "def", "device", "dropout", "dropout_probability", "either", "elif", "else", "embed_pos", "embed_positions", "embed_tokens", "encoder_layer", "encoder_states", "enumerate", "for", "forward", "functional", "have", "hidden_states", "idx", "if", "in", "input", "input_embeds", "input_ids", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layer_outputs", "layerdrop", "layernorm_embedding", "layers", "nn", "not", "or", "output_attentions", "output_hidden_states", "p", "r", "raise", "rand", "return", "return_dict", "same", "self", "shape", "specify", "the", "time", "to", "to_drop", "torch", "training", "tuple", "use_return_dict", "v", "view"], "plbart/modeling_plbart.py:PLBartDecoderLayer.__init__": ["ACT2FN", "GradientCheckpointingLayer", "LayerNorm", "Linear", "Model", "ModelAttention", "ModelConfig", "None", "True", "__init__", "activation_dropout", "activation_fn", "activation_function", "attention_dropout", "class", "config", "d_model", "decoder_attention_heads", "decoder_ffn_dim", "def", "dropout", "embed_dim", "encoder_attn", "encoder_attn_layer_norm", "fc1", "fc2", "final_layer_norm", "int", "is_causal", "is_decoder", "layer_idx", "nn", "num_heads", "self", "self_attn", "self_attn_layer_norm", "super"], "plbart/modeling_plbart.py:PLBartDecoderLayer.forward": ["Cache", "False", "FloatTensor", "GradientCheckpointingLayer", "Model", "None", "Tensor", "True", "activation_dropout", "activation_fn", "attention_mask", "bool", "cache_position", "class", "cross_attn_weights", "def", "dropout", "encoder_attention_mask", "encoder_attn", "encoder_attn_layer_norm", "encoder_hidden_states", "fc1", "fc2", "final_layer_norm", "forward", "functional", "hidden_states", "if", "is", "key_value_states", "nn", "not", "output_attentions", "outputs", "p", "past_key_values", "residual", "return", "self", "self_attn", "self_attn_layer_norm", "self_attn_weights", "torch", "training", "tuple", "use_cache"], "plbart/modeling_plbart.py:PLBartDecoder.__init__": ["False", "LayerNorm", "Model", "ModelConfig", "ModelLayer", "ModelLearnedPositionalEmbedding", "ModelPreTrainedModel", "ModelScaledWordEmbedding", "ModuleList", "__init__", "class", "config", "d_model", "decoder_layerdrop", "decoder_layers", "def", "dropout", "else", "embed_positions", "embed_scale", "embed_tokens", "for", "gradient_checkpointing", "i", "if", "in", "layer_idx", "layerdrop", "layernorm_embedding", "layers", "math", "max_position_embeddings", "max_target_positions", "nn", "pad_token_id", "padding_idx", "post_init", "range", "scale_embedding", "self", "sqrt", "super", "vocab_size"], "plbart/modeling_plbart.py:PLBartDecoder.forward": ["BaseModelOutputWithPastAndCrossAttentions", "Cache", "DynamicCache", "EncoderDecoderCache", "False", "FloatTensor", "LongTensor", "Model", "ModelPreTrainedModel", "None", "Setting", "Tensor", "True", "ValueError", "You", "all_cross_attentions", "all_hidden_states", "all_self_attns", "and", "arange", "at", "attention_mask", "attentions", "batch_size", "bool", "both", "cache_position", "cannot", "checkpointing", "class", "config", "continue", "create_bidirectional_mask", "create_causal_mask", "cross_attentions", "decoder_input_ids", "decoder_inputs_embeds", "decoder_layer", "def", "device", "dropout", "dropout_probability", "either", "elif", "else", "embed_positions", "embed_tokens", "encoder_attention_mask", "encoder_hidden_states", "enumerate", "for", "forward", "functional", "get_seq_length", "gradient", "gradient_checkpointing", "have", "hidden_states", "idx", "if", "in", "incompatible", "input", "input_embeds", "input_ids", "input_shape", "inputs_embeds", "is", "is_encoder_decoder", "is_torchdynamo_compiling", "isinstance", "kwargs", "last_hidden_state", "layer_outputs", "layerdrop", "layernorm_embedding", "layers", "logger", "mask_seq_length", "nn", "not", "ones", "or", "output_attentions", "output_hidden_states", "p", "past_key_values", "past_key_values_length", "position_ids", "positions", "r", "raise", "rand", "return", "return_dict", "same", "self", "self_attention_cache", "self_attn_cache", "seq_length", "shape", "size", "specify", "the", "time", "to", "torch", "training", "tuple", "use_cache", "use_return_dict", "v", "view", "warning_once", "with"], "plbart/modeling_plbart.py:shift_tokens_right": ["Model_tokens_right", "None", "Tensor", "ValueError", "be", "clone", "config", "decoder_start_tokens", "def", "defined", "dim", "gather", "has", "if", "index_of_eos", "input_ids", "int", "is", "masked_fill_", "model", "ne", "pad_token_id", "prev_output_tokens", "raise", "return", "self", "squeeze", "sum", "to", "torch", "unsqueeze"], "plbart/modeling_plbart.py:PLBartModel.__init__": ["Model", "ModelConfig", "ModelDecoder", "ModelEncoder", "ModelPreTrainedModel", "ModelScaledWordEmbedding", "__init__", "class", "config", "d_model", "decoder", "def", "else", "embed_scale", "encoder", "if", "math", "pad_token_id", "padding_idx", "post_init", "scale_embedding", "self", "shared", "sqrt", "super", "vocab_size"], "plbart/modeling_plbart.py:PLBartModel.get_input_embeddings": ["Model", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "return", "self", "shared"], "plbart/modeling_plbart.py:PLBartModel.set_input_embeddings": ["Model", "ModelPreTrainedModel", "class", "decoder", "def", "embed_tokens", "encoder", "self", "set_input_embeddings", "shared", "value"], "plbart/modeling_plbart.py:PLBartModel.forward": ["BaseModelOutput", "Cache", "FloatTensor", "LongTensor", "Model", "ModelPreTrainedModel", "None", "Seq2SeqModelOutput", "Tensor", "and", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "cross_attentions", "decoder", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_input_ids", "decoder_inputs_embeds", "decoder_outputs", "def", "elif", "else", "encoder", "encoder_attention_mask", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "isinstance", "kwargs", "last_hidden_state", "len", "list", "not", "output_attentions", "output_hidden_states", "pad_token_id", "past_key_values", "r", "return", "return_dict", "self", "shift_tokens_right", "torch", "tuple", "use_cache", "use_return_dict"], "plbart/modeling_plbart.py:PLBartForConditionalGeneration.__init__": ["False", "GenerationMixin", "Linear", "Model", "ModelConfig", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "d_model", "def", "final_logits_bias", "lm_head", "model", "nn", "num_embeddings", "post_init", "register_buffer", "self", "shared", "super", "torch", "zeros"], "plbart/modeling_plbart.py:PLBartForConditionalGeneration.resize_token_embeddings": ["Embedding", "GenerationMixin", "Model", "ModelPreTrainedModel", "None", "True", "_resize_final_logits_bias", "bool", "class", "def", "int", "mean_resizing", "new_embeddings", "new_num_tokens", "nn", "pad_to_multiple_of", "resize_token_embeddings", "return", "self", "shape", "super", "weight"], "plbart/modeling_plbart.py:PLBartForConditionalGeneration._resize_final_logits_bias": ["GenerationMixin", "Model", "ModelPreTrainedModel", "None", "_resize_final_logits_bias", "cat", "class", "def", "device", "dim", "else", "extra_bias", "final_logits_bias", "if", "int", "new_bias", "new_num_tokens", "old_num_tokens", "register_buffer", "self", "shape", "torch", "zeros"], "plbart/modeling_plbart.py:PLBartForConditionalGeneration.forward": ["Cache", "CrossEntropyLoss", "FloatTensor", "GenerationMixin", "LongTensor", "Model", "ModelPreTrainedModel", "None", "Seq2SeqLMOutput", "Tensor", "and", "attention_mask", "bool", "cache_position", "class", "config", "cross_attentions", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_input_ids", "decoder_inputs_embeds", "def", "device", "else", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "final_logits_bias", "forward", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "list", "lm_head", "lm_logits", "logits", "loss", "loss_fct", "masked_lm_loss", "model", "not", "output", "output_attentions", "output_hidden_states", "outputs", "pad_token_id", "past_key_values", "r", "return", "return_dict", "self", "shift_tokens_right", "to", "torch", "tuple", "use_cache", "use_return_dict", "view", "vocab_size"], "plbart/modeling_plbart.py:PLBartForConditionalGeneration.prepare_decoder_input_ids_from_labels": ["GenerationMixin", "Model", "ModelPreTrainedModel", "Tensor", "class", "config", "def", "labels", "pad_token_id", "prepare_decoder_input_ids_from_labels", "return", "self", "shift_tokens_right", "torch"], "plbart/modeling_plbart.py:PLBartClassificationHead.__init__": ["Dropout", "Linear", "Model", "Module", "__init__", "class", "def", "dense", "dropout", "float", "inner_dim", "input_dim", "int", "nn", "num_classes", "out_proj", "p", "pooler_dropout", "self", "super"], "plbart/modeling_plbart.py:PLBartClassificationHead.forward": ["Model", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "nn", "out_proj", "return", "self", "tanh", "torch"], "plbart/modeling_plbart.py:PLBartForSequenceClassification.__init__": ["Model", "ModelClassificationHead", "ModelConfig", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "classification_head", "classifier_dropout", "config", "d_model", "def", "kwargs", "model", "num_labels", "post_init", "self", "super"], "plbart/modeling_plbart.py:PLBartForSequenceClassification.forward": ["All", "BCEWithLogitsLoss", "CrossEntropyLoss", "False", "FloatTensor", "LongTensor", "MSELoss", "Model", "ModelPreTrainedModel", "None", "NotImplementedError", "Passing", "Seq2SeqSequenceClassifierOutput", "Tensor", "ValueError", "__class__", "__name__", "and", "attention_mask", "bool", "cache_position", "class", "classification_head", "config", "cross_attentions", "currently", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_input_ids", "decoder_inputs_embeds", "def", "device", "dtype", "elif", "else", "embeddings", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "eos", "eos_mask", "eos_token_id", "eq", "examples", "f", "for", "forward", "have", "hidden_states", "if", "input", "input_ids", "inputs_embeds", "int", "is", "kwargs", "labels", "len", "list", "logits", "long", "loss", "loss_fct", "model", "multi_label_classification", "must", "not", "num_labels", "number", "of", "or", "output", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "problem_type", "r", "raise", "regression", "return", "return_dict", "same", "self", "sentence_representation", "single_label_classification", "size", "squeeze", "sum", "supported", "the", "to", "tokens", "torch", "tuple", "unique_consecutive", "use_cache", "use_return_dict", "view"], "plbart/modeling_plbart.py:PLBartDecoderWrapper.__init__": ["Model", "ModelDecoder", "ModelPreTrainedModel", "__init__", "class", "config", "decoder", "def", "post_init", "self", "super"], "plbart/modeling_plbart.py:PLBartDecoderWrapper.forward": ["Model", "ModelPreTrainedModel", "args", "class", "decoder", "def", "forward", "kwargs", "return", "self"], "plbart/modeling_plbart.py:PLBartForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "Model", "ModelDecoderWrapper", "ModelPreTrainedModel", "True", "__init__", "bias", "class", "config", "def", "hidden_size", "is_decoder", "is_encoder_decoder", "lm_head", "model", "nn", "post_init", "self", "super", "vocab_size"], "plbart/modeling_plbart.py:PLBartForCausalLM.get_input_embeddings": ["GenerationMixin", "Model", "ModelPreTrainedModel", "class", "decoder", "def", "embed_tokens", "get_input_embeddings", "model", "return", "self"], "plbart/modeling_plbart.py:PLBartForCausalLM.set_input_embeddings": ["GenerationMixin", "Model", "ModelPreTrainedModel", "class", "decoder", "def", "embed_tokens", "model", "self", "set_input_embeddings", "value"], "plbart/modeling_plbart.py:PLBartForCausalLM.forward": ["Cache", "CausalLMOutputWithCrossAttentions", "CrossEntropyLoss", "FloatTensor", "GenerationMixin", "LongTensor", "Model", "ModelPreTrainedModel", "None", "Tensor", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "cross_attentions", "decoder", "def", "device", "else", "encoder_attention_mask", "encoder_hidden_states", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "lm_head", "logits", "logits_to_keep", "loss", "loss_fct", "model", "not", "output", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "r", "return", "return_dict", "self", "slice", "slice_indices", "to", "torch", "tuple", "use_cache", "use_return_dict", "view", "vocab_size"], "poolformer/modeling_poolformer.py:drop_path": ["False", "Model_path", "Model_prob", "Tensor", "bool", "def", "device", "div", "dtype", "float", "floor_", "if", "input", "keep_prob", "ndim", "not", "or", "output", "rand", "random_tensor", "return", "shape", "torch", "training"], "poolformer/modeling_poolformer.py:PoolFormerDropPath.__init__": ["ModelDropPath", "Module", "None", "__init__", "class", "def", "drop_prob", "float", "nn", "self", "super"], "poolformer/modeling_poolformer.py:PoolFormerDropPath.forward": ["ModelDropPath", "Module", "Tensor", "class", "def", "drop_path", "drop_prob", "forward", "hidden_states", "nn", "return", "self", "torch", "training"], "poolformer/modeling_poolformer.py:PoolFormerDropPath.extra_repr": ["ModelDropPath", "Module", "class", "def", "drop_prob", "extra_repr", "f", "nn", "p", "return", "self", "str"], "poolformer/modeling_poolformer.py:PoolFormerEmbeddings.__init__": ["Conv2d", "Identity", "Iterable", "ModelEmbeddings", "Module", "None", "__init__", "abc", "class", "collections", "def", "else", "hidden_size", "if", "isinstance", "kernel_size", "nn", "norm", "norm_layer", "num_channels", "padding", "patch_size", "projection", "self", "stride", "super"], "poolformer/modeling_poolformer.py:PoolFormerEmbeddings.forward": ["ModelEmbeddings", "Module", "class", "def", "embeddings", "forward", "nn", "norm", "pixel_values", "projection", "return", "self"], "poolformer/modeling_poolformer.py:PoolFormerGroupNorm.__init__": ["GroupNorm", "ModelGroupNorm", "__init__", "class", "def", "kwargs", "nn", "num_channels", "self", "super"], "poolformer/modeling_poolformer.py:PoolFormerPooling.__init__": ["AvgModel2d", "False", "Model", "ModelModeling", "Model_size", "Module", "__init__", "class", "count_include_pad", "def", "nn", "padding", "self", "stride", "super"], "poolformer/modeling_poolformer.py:PoolFormerPooling.forward": ["Model", "ModelModeling", "Module", "class", "def", "forward", "hidden_states", "nn", "return", "self"], "poolformer/modeling_poolformer.py:PoolFormerOutput.__init__": ["ACT2FN", "Conv2d", "ModelDropPath", "ModelOutput", "Module", "__init__", "act_fn", "class", "config", "conv1", "conv2", "def", "drop", "dropout_prob", "else", "hidden_act", "hidden_size", "if", "intermediate_size", "isinstance", "nn", "self", "str", "super"], "poolformer/modeling_poolformer.py:PoolFormerOutput.forward": ["ModelOutput", "Module", "act_fn", "class", "conv1", "conv2", "def", "drop", "forward", "hidden_states", "nn", "return", "self"], "poolformer/modeling_poolformer.py:PoolFormerLayer.__init__": ["Identity", "ModelDropPath", "ModelGroupNorm", "ModelLayer", "ModelModeling", "ModelOutput", "Model_size", "Modeling", "Module", "Parameter", "True", "__init__", "after_norm", "before_norm", "class", "config", "def", "drop_path", "else", "hidden_size", "if", "intermediate_size", "layer_scale_1", "layer_scale_2", "layer_scale_init_value", "nn", "num_channels", "ones", "output", "requires_grad", "self", "super", "torch", "use_layer_scale"], "poolformer/modeling_poolformer.py:PoolFormerLayer.forward": ["ModelLayer", "Modeling", "Modeling_output", "Module", "after_norm", "before_norm", "class", "def", "drop_path", "else", "forward", "hidden_states", "if", "layer_output", "layer_scale_1", "layer_scale_2", "nn", "output", "outputs", "return", "scaled_op", "self", "unsqueeze", "use_layer_scale"], "poolformer/modeling_poolformer.py:PoolFormerEncoder.__init__": ["ModelEmbeddings", "ModelEncoder", "ModelLayer", "Model_size", "Module", "ModuleList", "__init__", "append", "block", "blocks", "class", "config", "cpu", "cur", "def", "depths", "device", "dpr", "drop_path", "drop_path_rate", "else", "embeddings", "for", "hidden_size", "hidden_sizes", "i", "if", "in", "int", "intermediate_size", "item", "j", "layers", "linspace", "mlp_ratio", "nn", "num_channels", "num_encoder_blocks", "padding", "patch_embeddings", "patch_size", "patch_sizes", "range", "self", "stride", "strides", "sum", "super", "torch", "x"], "poolformer/modeling_poolformer.py:PoolFormerEncoder.forward": ["BaseModelOutputWithNoAttention", "False", "ModelEncoder", "Module", "None", "True", "_", "all_hidden_states", "blk", "block", "block_layer", "class", "def", "else", "embedding_layer", "enumerate", "for", "forward", "hidden_states", "idx", "if", "in", "is", "last_hidden_state", "layer_outputs", "layers", "nn", "not", "output_hidden_states", "patch_embeddings", "pixel_values", "return", "return_dict", "self", "tuple", "v", "zip"], "poolformer/modeling_poolformer.py:PoolFormerPreTrainedModel._init_weights": ["ModelLayer", "ModelPreTrainedModel", "PreTrainedModel", "_init_weights", "class", "config", "constant_", "def", "hasattr", "if", "init", "isinstance", "layer_scale_1", "layer_scale_2", "layer_scale_init_value", "module", "self", "super"], "poolformer/modeling_poolformer.py:PoolFormerModel.__init__": ["ModelEncoder", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "config", "def", "encoder", "post_init", "self", "super"], "poolformer/modeling_poolformer.py:PoolFormerModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "encoder", "get_input_embeddings", "patch_embeddings", "return", "self"], "poolformer/modeling_poolformer.py:PoolFormerModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "encoder", "patch_embeddings", "self", "set_input_embeddings", "value"], "poolformer/modeling_poolformer.py:PoolFormerModel.forward": ["BaseModelOutputWithNoAttention", "FloatTensor", "ModelModel", "ModelPreTrainedModel", "None", "ValueError", "You", "bool", "class", "config", "def", "else", "encoder", "encoder_outputs", "forward", "have", "hidden_states", "if", "is", "kwargs", "last_hidden_state", "not", "output_hidden_states", "pixel_values", "raise", "return", "return_dict", "self", "sequence_output", "specify", "to", "torch", "tuple", "use_return_dict"], "poolformer/modeling_poolformer.py:PoolFormerFinalPooler.__init__": ["Linear", "ModelFinalModeler", "Module", "__init__", "class", "config", "def", "dense", "hidden_size", "nn", "self", "super"], "poolformer/modeling_poolformer.py:PoolFormerFinalPooler.forward": ["ModelFinalModeler", "Module", "class", "def", "dense", "forward", "hidden_states", "nn", "output", "return", "self"], "poolformer/modeling_poolformer.py:PoolFormerForImageClassification.__init__": ["Identity", "Linear", "Model", "ModelForImageClassification", "ModelGroupNorm", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "classifier", "config", "def", "else", "hidden_sizes", "if", "nn", "norm", "num_labels", "post_init", "self", "super"], "poolformer/modeling_poolformer.py:PoolFormerForImageClassification.get_input_embeddings": ["Model", "ModelForImageClassification", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "return", "self"], "poolformer/modeling_poolformer.py:PoolFormerForImageClassification.set_input_embeddings": ["Model", "ModelForImageClassification", "ModelPreTrainedModel", "class", "def", "self", "set_input_embeddings", "value"], "poolformer/modeling_poolformer.py:PoolFormerForImageClassification.forward": ["FloatTensor", "ImageClassifierOutputWithNoAttention", "LongTensor", "Model", "ModelForImageClassification", "ModelPreTrainedModel", "None", "bool", "class", "classifier", "config", "def", "else", "forward", "hidden_states", "if", "is", "kwargs", "labels", "logits", "loss", "loss_function", "mean", "norm", "not", "output", "output_hidden_states", "outputs", "pixel_values", "r", "return", "return_dict", "self", "sequence_output", "torch", "tuple", "use_return_dict"], "pop2piano/modeling_pop2piano.py:Pop2PianoLayerNorm.__init__": ["ModelLayerNorm", "Module", "Parameter", "__init__", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "pop2piano/modeling_pop2piano.py:Pop2PianoLayerNorm.forward": ["ModelLayerNorm", "Module", "True", "bfloat16", "class", "def", "dtype", "float16", "float32", "forward", "hidden_states", "if", "in", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "pop2piano/modeling_pop2piano.py:Pop2PianoDenseActDense.__init__": ["ACT2FN", "Dropout", "False", "Linear", "ModelConfig", "ModelDenseActDense", "Module", "__init__", "act", "bias", "class", "config", "d_ff", "d_model", "def", "dense_act_fn", "dropout", "dropout_rate", "nn", "self", "super", "wi", "wo"], "pop2piano/modeling_pop2piano.py:Pop2PianoDenseActDense.forward": ["ModelDenseActDense", "Module", "Tensor", "act", "and", "class", "def", "dropout", "dtype", "forward", "hidden_states", "if", "int8", "isinstance", "nn", "return", "self", "to", "torch", "weight", "wi", "wo"], "pop2piano/modeling_pop2piano.py:Pop2PianoDenseGatedActDense.__init__": ["ACT2FN", "Dropout", "False", "Linear", "ModelConfig", "ModelDenseGatedActDense", "Module", "__init__", "act", "bias", "class", "config", "d_ff", "d_model", "def", "dense_act_fn", "dropout", "dropout_rate", "nn", "self", "super", "wi_0", "wi_1", "wo"], "pop2piano/modeling_pop2piano.py:Pop2PianoDenseGatedActDense.forward": ["ModelDenseGatedActDense", "Module", "Tensor", "act", "and", "class", "def", "dropout", "dtype", "forward", "hidden_gelu", "hidden_linear", "hidden_states", "if", "int8", "isinstance", "nn", "return", "self", "to", "torch", "weight", "wi_0", "wi_1", "wo"], "pop2piano/modeling_pop2piano.py:Pop2PianoLayerFF.__init__": ["DenseReluDense", "Dropout", "ModelConfig", "ModelDenseActDense", "ModelDenseGatedActDense", "ModelLayerFF", "ModelLayerNorm", "Module", "__init__", "class", "config", "d_model", "def", "dropout", "dropout_rate", "else", "eps", "if", "is_gated_act", "layer_norm", "layer_norm_epsilon", "nn", "self", "super"], "pop2piano/modeling_pop2piano.py:Pop2PianoLayerFF.forward": ["DenseReluDense", "ModelLayerFF", "Module", "class", "def", "dropout", "forward", "forwarded_states", "hidden_states", "layer_norm", "nn", "return", "self"], "pop2piano/modeling_pop2piano.py:Pop2PianoAttention.__init__": ["Embedding", "False", "Instantiating", "Linear", "ModelAttention", "ModelConfig", "Module", "None", "Please", "__class__", "__init__", "__name__", "a", "and", "bias", "caching", "call", "class", "config", "creating", "d_kv", "d_model", "decoder", "def", "dropout", "dropout_rate", "during", "errors", "f", "forward", "gradient_checkpointing", "has_relative_attention_bias", "if", "inner_dim", "int", "is", "is_decoder", "k", "key_value_proj_dim", "layer_idx", "logger", "make", "n_heads", "nn", "not", "num_heads", "o", "passing", "provide", "q", "recommended", "relative_attention_bias", "relative_attention_max_distance", "relative_attention_num_buckets", "self", "super", "sure", "the", "this", "to", "used", "v", "warning_once", "when", "will", "without"], "pop2piano/modeling_pop2piano.py:Pop2PianoAttention._relative_position_bucket": ["ModelAttention", "Module", "True", "_relative_position_bucket", "abs", "bidirectional", "class", "def", "else", "float", "full_like", "if", "is_small", "log", "long", "math", "max_distance", "max_exact", "min", "nn", "num_buckets", "relative_buckets", "relative_position", "relative_position_if_large", "return", "to", "torch", "where", "zeros_like"], "pop2piano/modeling_pop2piano.py:Pop2PianoAttention.compute_bias": ["ModelAttention", "Module", "None", "_relative_position_bucket", "arange", "bidirectional", "cache_position", "class", "compute_bias", "context_position", "def", "device", "dtype", "else", "if", "is", "is_decoder", "key_length", "long", "max_distance", "memory_position", "nn", "not", "num_buckets", "permute", "query_length", "relative_attention_bias", "relative_attention_max_distance", "relative_attention_num_buckets", "relative_position", "relative_position_bucket", "return", "self", "to", "torch", "unsqueeze", "values", "weight"], "pop2piano/modeling_pop2piano.py:Pop2PianoAttention.forward": ["EncoderDecoderCache", "False", "ModelAttention", "Module", "None", "True", "and", "attn_output", "attn_weights", "batch_size", "cache_position", "causal_mask", "class", "compute_bias", "contiguous", "cross_attention_cache", "curr_past_key_values", "current_states", "def", "device", "dim", "dropout", "dtype", "else", "float", "forward", "functional", "get", "gradient_checkpointing", "has_relative_attention_bias", "hidden_states", "if", "inner_dim", "is", "is_cross_attention", "is_updated", "isinstance", "k", "key_length", "key_states", "key_value_proj_dim", "key_value_states", "keys", "layer_idx", "layers", "mask", "matmul", "n_heads", "nn", "not", "o", "output_attentions", "outputs", "p", "past_key_values", "position_bias", "position_bias_masked", "q", "query_length", "query_states", "real_seq_length", "requires_grad", "return", "scores", "self", "self_attention_cache", "seq_length", "shape", "softmax", "torch", "training", "transpose", "type_as", "update", "use_cache", "v", "value_states", "values", "view", "zeros"], "pop2piano/modeling_pop2piano.py:Pop2PianoLayerSelfAttention.__init__": ["Dropout", "False", "ModelAttention", "ModelLayerNorm", "ModelLayerSelfAttention", "Module", "None", "SelfAttention", "__init__", "class", "config", "d_model", "def", "dropout", "dropout_rate", "eps", "has_relative_attention_bias", "int", "layer_idx", "layer_norm", "layer_norm_epsilon", "nn", "self", "super"], "pop2piano/modeling_pop2piano.py:Pop2PianoLayerSelfAttention.forward": ["False", "ModelLayerSelfAttention", "Module", "None", "SelfAttention", "attention_mask", "attention_output", "cache_position", "class", "def", "dropout", "forward", "hidden_states", "layer_norm", "mask", "nn", "normed_hidden_states", "output_attentions", "outputs", "past_key_values", "position_bias", "return", "self", "use_cache"], "pop2piano/modeling_pop2piano.py:Pop2PianoLayerCrossAttention.__init__": ["Dropout", "EncDecAttention", "False", "ModelAttention", "ModelLayerCrossAttention", "ModelLayerNorm", "Module", "None", "__init__", "class", "config", "d_model", "def", "dropout", "dropout_rate", "eps", "has_relative_attention_bias", "int", "layer_idx", "layer_norm", "layer_norm_epsilon", "nn", "self", "super"], "pop2piano/modeling_pop2piano.py:Pop2PianoLayerCrossAttention.forward": ["EncDecAttention", "False", "ModelLayerCrossAttention", "Module", "None", "attention_mask", "attention_output", "cache_position", "class", "def", "dropout", "forward", "hidden_states", "key_value_states", "layer_norm", "layer_output", "mask", "nn", "normed_hidden_states", "output_attentions", "outputs", "past_key_values", "position_bias", "query_length", "return", "self", "use_cache"], "pop2piano/modeling_pop2piano.py:Pop2PianoBlock.__init__": ["False", "GradientCheckpointingLayer", "ModelBlock", "ModelLayerCrossAttention", "ModelLayerFF", "ModelLayerSelfAttention", "ModuleList", "None", "__init__", "append", "class", "config", "def", "has_relative_attention_bias", "if", "int", "is_decoder", "layer", "layer_idx", "nn", "self", "super"], "pop2piano/modeling_pop2piano.py:Pop2PianoBlock.forward": ["False", "GradientCheckpointingLayer", "ModelBlock", "None", "True", "and", "any", "attention_mask", "attention_outputs", "cache_position", "clamp", "clamp_value", "class", "cross_attention_outputs", "def", "do_cross_attention", "dtype", "encoder_attention_mask", "encoder_decoder_position_bias", "encoder_hidden_states", "finfo", "float16", "forward", "hidden_states", "if", "is", "is_decoder", "isinf", "key_value_states", "layer", "max", "min", "not", "output_attentions", "outputs", "past_key_values", "position_bias", "query_length", "return", "return_dict", "self", "self_attention_outputs", "torch", "use_cache", "where"], "pop2piano/modeling_pop2piano.py:Pop2PianoPreTrainedModel._init_weights": ["ModelAttention", "ModelConcatEmbeddingToMel", "ModelDenseActDense", "ModelDenseGatedActDense", "ModelForConditionalGeneration", "ModelLayerNorm", "ModelPreTrainedModel", "None", "PreTrainedModel", "_init_weights", "and", "bias", "class", "config", "constant_", "d_ff", "d_kv", "d_model", "def", "elif", "embedding", "factor", "has_relative_attention_bias", "hasattr", "if", "init", "initializer_factor", "is", "isinstance", "k", "key_value_proj_dim", "lm_head", "mean", "module", "n_heads", "normal_", "not", "num_heads", "o", "q", "relative_attention_bias", "self", "shared", "std", "v", "weight", "wi", "wi_0", "wi_1", "wo", "zeros_"], "pop2piano/modeling_pop2piano.py:Pop2PianoPreTrainedModel._shift_right": ["In", "Model", "ModelPreTrainedModel", "None", "PreTrainedModel", "ValueError", "_shift_right", "be", "class", "clone", "config", "decoder_start_token_id", "def", "defined", "has", "if", "input_ids", "is", "it", "masked_fill_", "model", "new_zeros", "pad_token_id", "raise", "return", "self", "set", "shape", "shifted_input_ids", "the", "to", "usually"], "pop2piano/modeling_pop2piano.py:Pop2PianoStack.__init__": ["Dropout", "Embedding", "False", "ModelBlock", "ModelLayerNorm", "ModelPreTrainedModel", "ModelStack", "ModuleList", "__init__", "block", "bool", "class", "config", "d_model", "def", "dropout", "dropout_rate", "embed_tokens", "eps", "final_layer_norm", "for", "gradient_checkpointing", "has_relative_attention_bias", "i", "in", "is_decoder", "layer_idx", "layer_norm_epsilon", "nn", "num_layers", "post_init", "range", "self", "super", "vocab_size"], "pop2piano/modeling_pop2piano.py:Pop2PianoStack.set_input_embeddings": ["ModelPreTrainedModel", "ModelStack", "class", "def", "embed_tokens", "new_embeddings", "self", "set_input_embeddings"], "pop2piano/modeling_pop2piano.py:Pop2PianoStack.forward": ["BaseModelOutputWithPastAndCrossAttentions", "DynamicCache", "EncoderDecoderCache", "False", "ModelPreTrainedModel", "ModelStack", "None", "Setting", "True", "ValueError", "You", "_", "_update_causal_mask", "a", "all_attentions", "all_cross_attentions", "all_hidden_states", "and", "arange", "as", "at", "attention_mask", "attentions", "batch_size", "be", "block", "both", "cache_position", "can", "cannot", "causal_mask", "checkpointing", "class", "config", "cross_attentions", "decoder", "decoder_", "def", "device", "dropout", "dtype", "either", "elif", "else", "embed_tokens", "embeddings", "encoder_attention_mask", "encoder_batch_size", "encoder_decoder_position_bias", "encoder_extended_attention_mask", "encoder_hidden_shape", "encoder_hidden_states", "encoder_sequence_length", "enumerate", "err_msg_prefix", "f", "final_layer_norm", "finfo", "for", "forward", "get_seq_length", "gradient", "gradient_checkpointing", "have", "hidden_states", "i", "if", "in", "incompatible", "initialize", "input_ids", "input_shape", "inputs_embeds", "invert_attention_mask", "is", "is_decoder", "is_encoder_decoder", "is_torchdynamo_compiling", "isinstance", "kwargs", "last_hidden_state", "layer_module", "layer_outputs", "logger", "mask_seq_length", "min", "model", "not", "ones", "only", "or", "output_attentions", "output_hidden_states", "past_key_values", "past_key_values_length", "position_bias", "raise", "return", "return_dict", "same", "self", "self_attention_cache", "seq_length", "set", "size", "specify", "the", "time", "to", "token", "torch", "training", "tuple", "use_cache", "use_return_dict", "used", "v", "valid", "view", "warning_once", "with"], "pop2piano/modeling_pop2piano.py:Pop2PianoStack._update_causal_mask": ["AttentionMaskConverter", "BlockMask", "Cache", "False", "ModelPreTrainedModel", "ModelStack", "None", "Tensor", "Union", "_attn_implementation", "_ignore_causal_mask_sdpa", "_prepare_4d_causal_attention_mask_with_cache_position", "_unmask_unattended", "_update_causal_mask", "and", "any", "attention_mask", "batch_size", "bool", "cache_position", "causal_mask", "class", "config", "cuda", "def", "device", "dtype", "else", "finfo", "flash_attention_2", "flex_attention", "get_max_cache_shape", "get_seq_length", "if", "in", "input_tensor", "inputs_embeds", "is", "is_compileable", "is_training", "isinstance", "make_flex_block_causal_mask", "min", "min_dtype", "not", "npu", "output_attentions", "past_key_values", "past_key_values_length", "past_seen_tokens", "return", "sdpa", "self", "sequence_length", "shape", "target_length", "torch", "training", "type", "using_compilable_cache", "xpu"], "pop2piano/modeling_pop2piano.py:Pop2PianoStack._prepare_4d_causal_attention_mask_with_cache_position": ["ModelPreTrainedModel", "ModelStack", "None", "Tensor", "_prepare_4d_causal_attention_mask_with_cache_position", "and", "arange", "attention_mask", "batch_size", "cache_position", "causal_mask", "class", "clone", "def", "device", "diagonal", "dim", "dtype", "else", "expand", "fill_value", "finfo", "full", "if", "int", "is", "kwargs", "mask_length", "masked_fill", "min", "min_dtype", "not", "padding_mask", "reshape", "return", "sequence_length", "shape", "target_length", "to", "torch", "triu"], "pop2piano/modeling_pop2piano.py:Pop2PianoConcatEmbeddingToMel.__init__": ["Embedding", "ModelConcatEmbeddingToMel", "Module", "__init__", "class", "composer_vocab_size", "config", "d_model", "def", "embedding", "embedding_dim", "nn", "num_embeddings", "self", "super"], "pop2piano/modeling_pop2piano.py:Pop2PianoConcatEmbeddingToMel.forward": ["ModelConcatEmbeddingToMel", "Module", "cat", "class", "composer_embedding", "def", "dim", "embedding", "embedding_offset", "feature", "forward", "index_shifted", "index_value", "inputs_embeds", "nn", "return", "self", "torch", "unsqueeze"], "pop2piano/modeling_pop2piano.py:Pop2PianoForConditionalGeneration.__init__": ["Embedding", "False", "GenerationMixin", "Linear", "ModelConcatEmbeddingToMel", "ModelConfig", "ModelForConditionalGeneration", "ModelPreTrainedModel", "ModelStack", "True", "__init__", "bias", "class", "config", "copy", "d_model", "decoder", "decoder_config", "deepcopy", "def", "encoder", "encoder_config", "is_decoder", "lm_head", "mel_conditioner", "model_dim", "nn", "num_decoder_layers", "num_layers", "post_init", "self", "shared", "super", "use_cache", "vocab_size"], "pop2piano/modeling_pop2piano.py:Pop2PianoForConditionalGeneration.get_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "return", "self", "shared"], "pop2piano/modeling_pop2piano.py:Pop2PianoForConditionalGeneration.set_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "decoder", "def", "encoder", "new_embeddings", "self", "set_input_embeddings", "shared"], "pop2piano/modeling_pop2piano.py:Pop2PianoForConditionalGeneration.get_mel_conditioner_outputs": ["Composer", "FloatTensor", "GenerationConfig", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Please", "ValueError", "a", "attention_mask", "axis", "bool", "choose", "class", "composer", "composer_to_feature_token", "composer_value", "concatenate", "def", "device", "embedding_offset", "f", "feature", "from", "generation_config", "get_mel_conditioner_outputs", "if", "in", "index_value", "input_features", "is", "keys", "list", "mel_conditioner", "min", "not", "raise", "received", "repeat", "return", "self", "shape", "str", "tensor", "torch", "values", "view"], "pop2piano/modeling_pop2piano.py:Pop2PianoForConditionalGeneration.forward": ["BaseModelOutput", "BoolTensor", "Both", "Cache", "CrossEntropyLoss", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Please", "Seq2SeqLMOutput", "Tensor", "ValueError", "_shift_right", "and", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "cross_attentions", "decoder", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_input_ids", "decoder_inputs_embeds", "decoder_outputs", "def", "elif", "else", "encoder", "encoder_attention_mask", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "forward", "hidden_states", "if", "ignore_index", "input_features", "input_ids", "inputs_embeds", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "len", "lm_head", "lm_logits", "logits", "loss", "loss_fct", "model_dim", "not", "of", "one", "only", "output", "output_attentions", "output_hidden_states", "past_key_values", "provide", "r", "raise", "received", "return", "return_dict", "self", "sequence_output", "size", "them", "tie_word_embeddings", "torch", "tuple", "use_cache", "use_return_dict", "view"], "pop2piano/modeling_pop2piano.py:Pop2PianoForConditionalGeneration.generate": ["Found", "GenerationMixin", "Model", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Please", "ValueError", "a", "and", "as", "attention_mask", "be", "blob", "class", "co", "composer", "composer1", "composer_to_feature_token", "composer_vocab_size", "config", "def", "dict", "f", "found", "generate", "generation_config", "get_mel_conditioner_outputs", "hasattr", "https", "huggingface", "if", "in", "input_features", "inputs", "inputs_embeds", "is", "json", "keys", "kwargs", "len", "like", "main", "must", "not", "number", "of", "parse", "raise", "refer", "return", "same", "self", "super", "sweetcocoa", "that", "the", "to", "update", "vs", "was"], "pop2piano/modeling_pop2piano.py:Pop2PianoForConditionalGeneration.prepare_decoder_input_ids_from_labels": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "Tensor", "_shift_right", "class", "def", "labels", "prepare_decoder_input_ids_from_labels", "return", "self", "torch"], "prompt_depth_anything/modeling_prompt_depth_anything.py:PromptDepthAnythingLayer.__init__": ["Conv2d", "ModelConfig", "ModelLayer", "Module", "ReLU", "True", "__init__", "activation1", "activation2", "bias", "class", "config", "convolution1", "convolution2", "convolution3", "def", "fusion_hidden_size", "kernel_size", "nn", "padding", "self", "stride", "super"], "prompt_depth_anything/modeling_prompt_depth_anything.py:PromptDepthAnythingLayer.forward": ["ModelLayer", "Model_depth", "Module", "Tensor", "activation1", "activation2", "class", "convolution1", "convolution2", "convolution3", "def", "forward", "hidden_state", "nn", "return", "self", "torch"], "prompt_depth_anything/modeling_prompt_depth_anything.py:PromptDepthAnythingPreActResidualLayer.__init__": ["Conv2d", "ModelPreActResidualLayer", "Module", "ReLU", "True", "__init__", "activation1", "activation2", "bias", "class", "config", "convolution1", "convolution2", "def", "fusion_hidden_size", "kernel_size", "nn", "padding", "self", "stride", "super"], "prompt_depth_anything/modeling_prompt_depth_anything.py:PromptDepthAnythingPreActResidualLayer.forward": ["ModelPreActResidualLayer", "Module", "Tensor", "activation1", "activation2", "class", "convolution1", "convolution2", "def", "forward", "hidden_state", "nn", "residual", "return", "self", "torch"], "prompt_depth_anything/modeling_prompt_depth_anything.py:PromptDepthAnythingFeatureFusionLayer.__init__": ["Conv2d", "ModelConfig", "ModelFeatureFusionLayer", "ModelLayer", "ModelPreActResidualLayer", "Model_depth_layer", "Module", "True", "__init__", "bias", "class", "config", "def", "fusion_hidden_size", "kernel_size", "nn", "projection", "residual_layer1", "residual_layer2", "self", "super"], "prompt_depth_anything/modeling_prompt_depth_anything.py:PromptDepthAnythingFeatureFusionLayer.forward": ["False", "ModelFeatureFusionLayer", "Model_depth", "Model_depth_layer", "Module", "None", "True", "align_corners", "bilinear", "class", "def", "else", "forward", "functional", "hidden_state", "if", "interpolate", "is", "mode", "modifier", "nn", "not", "projection", "res", "residual", "residual_layer1", "residual_layer2", "return", "scale_factor", "self", "shape", "size"], "prompt_depth_anything/modeling_prompt_depth_anything.py:PromptDepthAnythingFeatureFusionStage.__init__": ["ModelConfig", "ModelFeatureFusionLayer", "ModelFeatureFusionStage", "Module", "ModuleList", "_", "__init__", "append", "class", "config", "def", "for", "in", "layers", "len", "neck_hidden_sizes", "nn", "range", "self", "super"], "prompt_depth_anything/modeling_prompt_depth_anything.py:PromptDepthAnythingFeatureFusionStage.forward": ["ModelFeatureFusionStage", "Model_depth", "Module", "None", "append", "class", "def", "else", "enumerate", "for", "forward", "fused_hidden_state", "fused_hidden_states", "hidden_state", "hidden_states", "idx", "if", "in", "is", "layer", "layers", "len", "nn", "return", "self", "shape", "size", "zip"], "prompt_depth_anything/modeling_prompt_depth_anything.py:PromptDepthAnythingDepthEstimationHead.__init__": ["Conv2d", "ModelDepthEstimationHead", "Module", "ReLU", "Sigmoid", "Unknown", "ValueError", "__init__", "activation1", "activation2", "class", "config", "conv1", "conv2", "conv3", "def", "depth", "depth_estimation_type", "elif", "else", "estimation", "f", "features", "fusion_hidden_size", "head_hidden_size", "head_in_index", "if", "kernel_size", "max_depth", "metric", "nn", "padding", "patch_size", "raise", "relative", "self", "stride", "super", "type"], "prompt_depth_anything/modeling_prompt_depth_anything.py:PromptDepthAnythingDepthEstimationHead.forward": ["ModelDepthEstimationHead", "Module", "Tensor", "True", "activation1", "activation2", "align_corners", "bilinear", "class", "conv1", "conv2", "conv3", "def", "dim", "forward", "functional", "hidden_states", "int", "interpolate", "list", "mode", "nn", "patch_height", "patch_size", "patch_width", "predicted_depth", "return", "self", "squeeze", "target_height", "target_width", "torch", "torch_int"], "prompt_depth_anything/modeling_prompt_depth_anything.py:PromptDepthAnythingReassembleLayer.__init__": ["Conv2d", "ConvTranspose2d", "Identity", "ModelConfig", "ModelReassembleLayer", "Module", "__init__", "channels", "class", "config", "def", "elif", "factor", "if", "in_channels", "int", "kernel_size", "nn", "out_channels", "padding", "projection", "reassemble_hidden_size", "resize", "self", "stride", "super", "torch_int"], "prompt_depth_anything/modeling_prompt_depth_anything.py:PromptDepthAnythingReassembleLayer.forward": ["ModelReassembleLayer", "Module", "class", "def", "forward", "hidden_state", "nn", "projection", "resize", "return", "self"], "prompt_depth_anything/modeling_prompt_depth_anything.py:PromptDepthAnythingReassembleStage.__init__": ["ModelReassembleLayer", "ModelReassembleStage", "Module", "ModuleList", "__init__", "append", "channels", "class", "config", "def", "factor", "for", "in", "layers", "neck_hidden_sizes", "nn", "reassemble_factors", "self", "super", "zip"], "prompt_depth_anything/modeling_prompt_depth_anything.py:PromptDepthAnythingReassembleStage.forward": ["ModelReassembleStage", "Module", "None", "Tensor", "_", "append", "batch_size", "class", "contiguous", "def", "enumerate", "for", "forward", "hidden_state", "hidden_states", "i", "in", "layers", "list", "nn", "num_channels", "out", "patch_height", "patch_width", "permute", "reshape", "return", "self", "shape", "torch"], "prompt_depth_anything/modeling_prompt_depth_anything.py:PromptDepthAnythingNeck.__init__": ["Conv2d", "False", "ModelFeatureFusionStage", "ModelNeck", "ModelReassembleStage", "Module", "ModuleList", "__init__", "append", "bias", "channel", "class", "config", "convs", "def", "for", "fusion_hidden_size", "fusion_stage", "in", "kernel_size", "neck_hidden_sizes", "nn", "padding", "reassemble_stage", "self", "super"], "prompt_depth_anything/modeling_prompt_depth_anything.py:PromptDepthAnythingNeck.forward": ["ModelNeck", "Model_depth", "Module", "None", "Tensor", "The", "TypeError", "ValueError", "a", "be", "class", "config", "convs", "def", "enumerate", "equal", "feature", "features", "for", "forward", "fusion_stage", "hidden", "hidden_states", "i", "if", "in", "int", "isinstance", "len", "list", "neck", "neck_hidden_sizes", "nn", "not", "number", "of", "or", "output", "patch_height", "patch_width", "raise", "reassemble_stage", "return", "self", "should", "sizes", "states", "tensors", "the", "to", "torch", "tuple"], "prompt_depth_anything/modeling_prompt_depth_anything.py:PromptDepthAnythingForDepthEstimation.__init__": ["ModelDepthEstimationHead", "ModelForDepthEstimation", "ModelNeck", "ModelPreTrainedModel", "__init__", "backbone", "class", "config", "def", "head", "load_backbone", "neck", "post_init", "self", "super"], "prompt_depth_anything/modeling_prompt_depth_anything.py:PromptDepthAnythingForDepthEstimation.forward": ["DepthEstimatorOutput", "FloatTensor", "LongTensor", "ModelForDepthEstimation", "ModelPreTrainedModel", "Model_depth", "None", "NotImplementedError", "Tensor", "Training", "_", "attentions", "backbone", "batch_size", "bool", "class", "config", "def", "depth_max", "depth_min", "device", "dim", "else", "feature_maps", "forward", "forward_with_filtered_kwargs", "head", "height", "hidden_states", "if", "implemented", "is", "kwargs", "labels", "loss", "max", "min", "neck", "not", "output", "output_attentions", "output_hidden_states", "outputs", "patch_height", "patch_size", "patch_width", "pixel_values", "predicted_depth", "r", "raise", "reshape", "return", "return_dict", "self", "shape", "squeeze", "to", "torch", "tuple", "use_return_dict", "values", "view", "width", "yet"], "prophetnet/modeling_prophetnet.py:softmax": ["False", "Model", "def", "dim", "dtype", "else", "float", "float32", "functional", "hidden_state", "if", "nn", "onnx_trace", "return", "torch"], "prophetnet/modeling_prophetnet.py:ngram_attention_bias": ["False", "Model", "Model_attention_bias", "cat", "clone", "def", "detach", "device", "dim", "dtype", "fill_diagonal_", "finfo", "for", "in", "left_block", "min", "ones", "range", "return", "right_block", "sequence_length", "stream_idx", "torch", "triu_", "wrap"], "prophetnet/modeling_prophetnet.py:compute_relative_buckets": ["False", "Model_relative_buckets", "abs", "def", "else", "float", "if", "int", "inv_relative_positions", "is_bidirectional", "is_small", "log", "lt", "math", "max", "max_distance", "max_exact", "min", "num_buckets", "ones_like", "rel_positions_bucket", "relative_positions", "return", "torch", "val_if_large", "where", "zeros_like"], "prophetnet/modeling_prophetnet.py:compute_all_stream_relative_buckets": ["False", "Model_all_stream_relative_buckets", "Model_relative_buckets", "cat", "def", "dim", "is_bidirectional", "main_relative_position_buckets", "main_stream_relative_positions", "max_distance", "num_buckets", "position_ids", "predict_relative_position_buckets", "predicting_stream_relative_positions", "repeat", "return", "size", "torch", "unsqueeze"], "prophetnet/modeling_prophetnet.py:ProphetNetPreTrainedModel._shift_right": ["In", "Model", "ModelPreTrainedModel", "None", "PreTrainedModel", "See", "Verify", "_shift_right", "all", "assert", "be", "class", "clone", "config", "decoder_start_token_id", "def", "defined", "docs", "for", "has", "information", "input_ids", "is", "it", "item", "masked_fill_", "model", "more", "new_zeros", "not", "only", "pad_token_id", "positive", "return", "self", "set", "shape", "shifted_input_ids", "that", "the", "to", "torch", "usually", "values"], "prophetnet/modeling_prophetnet.py:ProphetNetPositionalEmbeddings.__init__": ["Embedding", "ModelConfig", "ModelPositionalEmbeddings", "None", "__init__", "class", "config", "def", "hidden_size", "max_length", "max_position_embeddings", "nn", "pad_token_id", "self", "super"], "prophetnet/modeling_prophetnet.py:ProphetNetPositionalEmbeddings.forward": ["Embedding", "If", "ModelPositionalEmbeddings", "None", "and", "assert", "attention_mask", "be", "clamp", "class", "computed", "cumsum", "def", "device", "dim", "dtype", "else", "forward", "get_seq_length", "if", "inputs_shape", "int", "is", "long", "max_length", "nn", "not", "num_input_ids", "ones", "or", "padding_idx", "past_key_values", "position_ids", "pre", "prev_num_input_ids", "return", "self", "set", "should", "super", "then", "torch", "type_as"], "prophetnet/modeling_prophetnet.py:ProphetNetPositionalEmbeddings._forward": ["Embedding", "ModelPositionalEmbeddings", "_forward", "class", "def", "forward", "nn", "position_ids", "return", "self", "super"], "prophetnet/modeling_prophetnet.py:ProphetNetAttention.__init__": ["Linear", "ModelAttention", "ModelConfig", "Module", "None", "__init__", "and", "assert", "attention_dropout", "be", "by", "class", "config", "def", "divisible", "dropout", "head_dim", "hidden_size", "int", "key_proj", "layer_idx", "must", "nn", "num_attn_heads", "num_decoder_attention_heads", "num_encoder_attention_heads", "out_proj", "query_proj", "self", "super", "value_proj"], "prophetnet/modeling_prophetnet.py:ProphetNetAttention.forward": ["Attention", "Cache", "EncoderDecoderCache", "False", "ModelAttention", "Module", "None", "Size", "Tensor", "True", "ValueError", "and", "assert", "attention_dropout", "attention_mask", "attn_output", "attn_probs", "attn_weights", "attn_weights_reshaped", "batch_size", "be", "bool", "bsij", "bsik", "bsjk", "but", "cache_position", "class", "cross_attention_cache", "curr_past_key_values", "current_states", "def", "dim", "dropout", "einsum", "else", "expected_shape", "f", "forward", "functional", "get", "have", "head_dim", "hidden", "hidden_size", "hidden_states", "if", "is", "is_cross_attention", "is_updated", "isinstance", "key_proj", "key_states", "key_value_states", "keys", "layer_idx", "layers", "list", "mask", "nn", "not", "num_attn_heads", "of", "out_proj", "output_attentions", "p", "past_key_values", "query_proj", "query_states", "raise", "reshape", "return", "self", "self_attention_cache", "shape", "should", "size", "softmax", "src_len", "states", "tgt_len", "torch", "training", "transpose", "tuple", "update", "value_proj", "value_states", "values", "view", "weights"], "prophetnet/modeling_prophetnet.py:ProphetNetFeedForward.__init__": ["ACT2FN", "Linear", "ModelConfig", "ModelFeedForward", "Module", "__init__", "activation_dropout", "activation_fn", "activation_function", "class", "config", "def", "dropout", "ffn_dim", "hidden_size", "int", "intermediate", "nn", "output", "self", "super"], "prophetnet/modeling_prophetnet.py:ProphetNetFeedForward.forward": ["ModelFeedForward", "Module", "activation_dropout", "activation_fn", "class", "def", "dropout", "forward", "functional", "hidden_states", "intermediate", "nn", "output", "p", "return", "self", "training"], "prophetnet/modeling_prophetnet.py:ProphetNetNgramSelfAttention.__init__": ["False", "Linear", "ModelConfig", "ModelNgramSelfAttention", "Module", "None", "__init__", "assert", "attention_dropout", "be", "by", "class", "config", "def", "divisible", "dropout", "head_dim", "hidden_size", "key_proj", "layer_idx", "must", "ngram", "nn", "num_attn_heads", "num_buckets", "num_decoder_attention_heads", "onnx_trace", "out_proj", "query_proj", "relative_max_distance", "relative_pos_embeddings", "self", "super", "value_proj"], "prophetnet/modeling_prophetnet.py:ProphetNetNgramSelfAttention._shape": ["ModelNgramSelfAttention", "Module", "_shape", "batch_size", "class", "contiguous", "def", "head_dim", "nn", "num_attn_heads", "return", "self", "seq_len", "tensor", "transpose", "view"], "prophetnet/modeling_prophetnet.py:ProphetNetNgramSelfAttention.prepare_for_onnx_export_": ["ModelNgramSelfAttention", "Module", "True", "class", "def", "nn", "onnx_trace", "prepare_for_onnx_export_", "self"], "prophetnet/modeling_prophetnet.py:ProphetNetNgramSelfAttention.forward": ["Cache", "EncoderDecoderCache", "ModelNgramSelfAttention", "Module", "None", "_shape", "assert", "attention_dropout", "attention_mask", "attn_output", "batch_size", "be", "bncs", "bnhsc", "bnhtc", "bnhts", "bntc", "bnts", "but", "cache_position", "cat", "chunk", "class", "curr_past_key_values", "def", "dim", "dropout", "dtype", "einsum", "else", "extended_predict_attention_mask", "f", "for", "forward", "functional", "get_main_relative_pos_embeddings", "get_predict_relative_pos_embeddings", "head_dim", "hidden_size", "hidden_states", "hidden_states_list", "hidden_states_predict_list", "if", "in", "is", "isinstance", "key", "key_proj", "key_states", "key_states_list", "layer_idx", "list", "main_attn_output", "main_attn_probs", "main_attn_weights", "main_hidden_states", "main_key_states", "main_query_states", "main_relative_pos_embeddings", "main_relative_position_buckets", "main_value_states", "ngram", "ngram_sequence_length", "nn", "not", "num_attn_heads", "of", "onnx_trace", "out_proj", "p", "past_key_values", "permute", "position_ids", "predict_attn_output", "predict_attn_probs", "predict_attn_weights", "predict_hidden_states", "predict_key_states", "predict_key_states_list", "predict_query_states", "predict_query_states_list", "predict_relative_pos_embeddings", "predict_relative_position_buckets", "predict_value_states", "predict_value_states_list", "proj_shape", "query_proj", "query_states", "query_states_list", "reshape", "return", "self", "self_attention_cache", "sequence_length", "shape", "should", "size", "softmax", "stack", "to", "torch", "training", "transpose", "type_as", "unsqueeze", "update", "v_p", "value_proj", "value_states", "value_states_list", "view"], "prophetnet/modeling_prophetnet.py:ProphetNetNgramSelfAttention.get_main_relative_pos_embeddings": ["False", "ModelNgramSelfAttention", "Module", "None", "arange", "attn_weights", "batch_size", "class", "compute_relative_buckets", "def", "device", "dim", "gather", "get_main_relative_pos_embeddings", "hidden_states", "if", "index", "is", "long", "main_relative_pos_embeddings", "main_relative_position_buckets", "nn", "num_attn_heads", "num_buckets", "permute", "position_ids", "rel_pos_embeddings", "relative_max_distance", "relative_pos_embeddings", "relative_positions", "repeat", "reshape", "return", "self", "sequence_length", "shape", "size", "src_len", "tgt_len", "to", "torch", "unsqueeze", "view"], "prophetnet/modeling_prophetnet.py:ProphetNetNgramSelfAttention.get_predict_relative_pos_embeddings": ["False", "ModelNgramSelfAttention", "Module", "None", "They", "arange", "are", "assert", "attn_weights", "batch_size", "be", "class", "compute_relative_buckets", "def", "device", "dim", "format", "gather", "get_predict_relative_pos_embeddings", "hidden_states", "if", "incorrect", "index", "is", "key_sequence_length", "long", "ngram", "nn", "num_attn_heads", "num_buckets", "of", "permute", "position_ids", "predict_relative_pos_embeddings", "predict_relative_position_buckets", "rel_pos_embeddings", "relative_max_distance", "relative_pos_embeddings", "relative_positions", "repeat", "reshape", "return", "self", "sequence_length", "shape", "should", "size", "the", "to", "torch", "transpose", "unsqueeze", "view"], "prophetnet/modeling_prophetnet.py:ProphetNetEncoderLayer.__init__": ["GradientCheckpointingLayer", "LayerNorm", "ModelAttention", "ModelConfig", "ModelEncoderLayer", "ModelFeedForward", "__init__", "class", "config", "def", "encoder_ffn_dim", "feed_forward", "feed_forward_layer_norm", "hidden_size", "num_encoder_attention_heads", "self", "self_attn", "self_attn_layer_norm", "super"], "prophetnet/modeling_prophetnet.py:ProphetNetEncoderLayer.forward": ["False", "GradientCheckpointingLayer", "ModelEncoderLayer", "attention_mask", "attention_output", "attn_weights", "bool", "class", "def", "feed_forward", "feed_forward_layer_norm", "feed_forward_output", "forward", "hidden_states", "if", "output_attentions", "outputs", "return", "self", "self_attn", "self_attn_layer_norm"], "prophetnet/modeling_prophetnet.py:ProphetNetDecoderLayer.__init__": ["GradientCheckpointingLayer", "LayerNorm", "ModelAttention", "ModelConfig", "ModelDecoderLayer", "ModelFeedForward", "ModelNgramSelfAttention", "None", "__init__", "add_cross_attention", "class", "config", "cross_attn", "cross_attn_layer_norm", "decoder_ffn_dim", "def", "feed_forward", "feed_forward_layer_norm", "hidden_size", "if", "layer_idx", "num_decoder_attention_heads", "self", "self_attn", "self_attn_layer_norm", "super"], "prophetnet/modeling_prophetnet.py:ProphetNetDecoderLayer.forward": ["False", "GradientCheckpointingLayer", "ModelDecoderLayer", "None", "Tensor", "True", "attention_mask", "attention_output", "bool", "cache_position", "class", "cross_attn", "cross_attn_layer_norm", "cross_attn_weights", "def", "encoder_attn_mask", "encoder_hidden_states", "extended_predict_attention_mask", "feed_forward", "feed_forward_layer_norm", "feed_forward_output", "forward", "hidden_states", "if", "is", "key_value_states", "main_relative_position_buckets", "ngram_attention_output", "not", "output_attentions", "outputs", "past_key_values", "position_ids", "predict_relative_position_buckets", "return", "self", "self_attn", "self_attn_layer_norm", "self_attn_weights", "self_attn_weights_ngram", "torch", "use_cache"], "prophetnet/modeling_prophetnet.py:ProphetNetEncoder.__init__": ["Embedding", "False", "LayerNorm", "ModelConfig", "ModelEncoder", "ModelEncoderLayer", "ModelPositionalEmbeddings", "ModelPreTrainedModel", "ModuleList", "_", "__init__", "class", "config", "def", "embeddings_layer_norm", "for", "gradient_checkpointing", "hidden_size", "in", "layers", "nn", "num_encoder_layers", "pad_token_id", "padding_idx", "position_embeddings", "post_init", "range", "self", "super", "vocab_size", "word_embeddings"], "prophetnet/modeling_prophetnet.py:ProphetNetEncoder.get_input_embeddings": ["ModelEncoder", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "return", "self", "word_embeddings"], "prophetnet/modeling_prophetnet.py:ProphetNetEncoder.set_input_embeddings": ["ModelEncoder", "ModelPreTrainedModel", "class", "def", "self", "set_input_embeddings", "value", "word_embeddings"], "prophetnet/modeling_prophetnet.py:ProphetNetEncoder.forward": ["BaseModelOutput", "Either", "Make", "ModelEncoder", "ModelPreTrainedModel", "None", "Tensor", "ValueError", "all_attentions", "and", "attention_mask", "attentions", "be", "bool", "class", "config", "def", "device", "dropout", "dtype", "elif", "else", "embeddings_layer_norm", "encoder_hidden_states", "encoder_layer", "enumerate", "extended_attention_mask", "finfo", "for", "forward", "functional", "has", "hidden_states", "idx", "if", "in", "input_ids", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layer_outputs", "layers", "min", "nn", "not", "num_encoder_attention_heads", "only", "or", "output_attentions", "output_hidden_states", "p", "pass", "passed", "position_embeddings", "position_ids", "r", "raise", "repeat", "return", "return_dict", "self", "shape", "sure", "to", "torch", "training", "tuple", "use_return_dict", "v", "word_embeddings"], "prophetnet/modeling_prophetnet.py:ProphetNetDecoder.__init__": ["Embedding", "False", "LayerNorm", "ModelConfig", "ModelDecoder", "ModelDecoderLayer", "ModelPositionalEmbeddings", "ModelPreTrainedModel", "ModuleList", "None", "__init__", "class", "config", "def", "dropout", "embeddings_layer_norm", "for", "gradient_checkpointing", "hidden_size", "i", "in", "layer_idx", "layers", "max_position_embeddings", "max_target_positions", "ngram", "ngram_embeddings", "nn", "num_buckets", "num_decoder_layers", "pad_token_id", "padding_idx", "position_embeddings", "post_init", "range", "relative_max_distance", "self", "super", "vocab_size", "word_embeddings"], "prophetnet/modeling_prophetnet.py:ProphetNetDecoder.get_input_embeddings": ["ModelDecoder", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "return", "self", "word_embeddings"], "prophetnet/modeling_prophetnet.py:ProphetNetDecoder.set_input_embeddings": ["ModelDecoder", "ModelPreTrainedModel", "class", "def", "self", "set_input_embeddings", "value", "word_embeddings"], "prophetnet/modeling_prophetnet.py:ProphetNetDecoder.forward": ["At", "Cache", "DynamicCache", "Either", "EncoderDecoderCache", "False", "Make", "ModelDecoder", "ModelDecoderModelOutput", "ModelPreTrainedModel", "None", "Setting", "Tensor", "True", "ValueError", "_forward", "add_cross_attention", "all_cross_attns", "all_main_stream_attns", "all_main_stream_hidden_states", "all_ngram_stream_attns", "all_ngram_stream_hidden_states", "and", "assert", "attention_mask", "attentions", "batch_size", "be", "bool", "cache_position", "cat", "checkpointing", "class", "compute_buffered_relative_buckets", "config", "cross_attentions", "decoder_input_ids", "decoder_inputs_embeds", "decoder_layer", "def", "device", "dropout", "dtype", "elif", "else", "embeddings_layer_norm", "encoder_attention_mask", "encoder_attn_mask", "encoder_hidden_states", "enumerate", "extended_attention_mask", "extended_encoder_attention_mask", "extended_predict_attention_mask", "finfo", "for", "forward", "functional", "get_seq_length", "gradient", "gradient_checkpointing", "has", "hidden_states", "hidden_states_ngram", "idx", "if", "in", "incompatible", "input_ids", "inputs_embeds", "is", "is_encoder_decoder", "kwargs", "last_hidden_state", "last_hidden_state_ngram", "layer_outputs", "layers", "length", "logger", "main_relative_position_buckets", "main_stream_pos_embed", "min", "moment", "ngram", "ngram_attentions", "ngram_embeddings", "ngram_hidden_states", "nn", "not", "num_decoder_attention_heads", "of", "only", "or", "output_attentions", "output_hidden_states", "p", "pass", "passed", "past_key_values", "past_key_values_length", "position_embeddings", "position_ids", "predict_relative_position_buckets", "predicting_stream_pos_embed", "prepare_attention_mask", "prepare_predict_attention_mask", "r", "raise", "range", "repeat", "return", "return_dict", "self", "sequence_length", "shape", "size", "supported", "sure", "the", "to", "torch", "training", "tuple", "use_cache", "use_return_dict", "v", "warning_once", "weight", "with", "word_embeddings"], "prophetnet/modeling_prophetnet.py:ProphetNetDecoder.compute_buffered_relative_buckets": ["ModelDecoder", "ModelPreTrainedModel", "arange", "batch_size", "cat", "class", "compute_all_stream_relative_buckets", "compute_buffered_relative_buckets", "def", "device", "main_relative_buckets", "max_target_positions", "num_buckets", "position_ids", "predict_relative_buckets", "relative_max_distance", "repeat", "return", "self", "sequence_length", "shape", "to", "torch"], "prophetnet/modeling_prophetnet.py:ProphetNetDecoder.prepare_attention_mask": ["ModelDecoder", "ModelPreTrainedModel", "None", "attention_mask", "batch_size", "causal_mask", "class", "config", "def", "device", "dtype", "else", "expand", "extended_attention_mask", "extended_causal_mask", "finfo", "full", "hidden_states", "if", "is", "min", "not", "num_decoder_attention_heads", "prepare_attention_mask", "return", "self", "seq_length", "shape", "to", "torch", "triu"], "prophetnet/modeling_prophetnet.py:ProphetNetDecoder.prepare_predict_attention_mask": ["ModelDecoder", "ModelPreTrainedModel", "None", "attention_mask", "batch_size", "cat", "class", "config", "def", "device", "dim", "dtype", "else", "expand", "extended_attention_mask", "extended_predict_attention_mask", "extended_predict_causal_mask", "finfo", "hidden_states", "if", "is", "max_target_positions", "min", "ngram", "ngram_attention_bias", "not", "num_decoder_attention_heads", "predict_causal_mask", "prepare_predict_attention_mask", "return", "self", "seq_length", "shape", "to", "torch", "zeros_like"], "prophetnet/modeling_prophetnet.py:ProphetNetModel.__init__": ["Embedding", "False", "ModelConfig", "ModelDecoder", "ModelEncoder", "ModelModel", "ModelPreTrainedModel", "True", "__init__", "class", "config", "copy", "decoder", "decoder_config", "deepcopy", "def", "encoder", "encoder_config", "hidden_size", "is_decoder", "nn", "pad_token_id", "padding_idx", "post_init", "self", "super", "use_cache", "vocab_size", "word_embeddings"], "prophetnet/modeling_prophetnet.py:ProphetNetModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "return", "self", "word_embeddings"], "prophetnet/modeling_prophetnet.py:ProphetNetModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "decoder", "def", "encoder", "self", "set_input_embeddings", "value", "word_embeddings"], "prophetnet/modeling_prophetnet.py:ProphetNetModel.forward": ["BoolTensor", "Cache", "ModelModel", "ModelPreTrainedModel", "ModelSeq2SeqModelOutput", "None", "Tensor", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "cross_attentions", "decoder", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_input_ids", "decoder_inputs_embeds", "decoder_ngram_attentions", "decoder_ngram_hidden_states", "decoder_outputs", "def", "else", "encoder", "encoder_attention_mask", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "forward", "hidden_states", "hidden_states_ngram", "if", "input_ids", "inputs_embeds", "is", "kwargs", "last_hidden_state", "last_hidden_state_ngram", "ngram_attentions", "not", "output_attentions", "output_hidden_states", "past_key_values", "r", "return", "return_dict", "self", "torch", "tuple", "use_cache", "use_return_dict"], "prophetnet/modeling_prophetnet.py:ProphetNetForConditionalGeneration.__init__": ["False", "GenerationMixin", "Linear", "Model", "ModelConfig", "ModelForConditionalGeneration", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "disable_ngram_loss", "hidden_size", "lm_head", "nn", "pad_token_id", "padding_idx", "post_init", "self", "super", "vocab_size"], "prophetnet/modeling_prophetnet.py:ProphetNetForConditionalGeneration.get_input_embeddings": ["GenerationMixin", "Model", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "return", "self", "word_embeddings"], "prophetnet/modeling_prophetnet.py:ProphetNetForConditionalGeneration.forward": ["BoolTensor", "Cache", "GenerationMixin", "Model", "ModelForConditionalGeneration", "ModelPreTrainedModel", "ModelSeq2SeqLMOutput", "None", "Tensor", "_compute_loss", "_shift_right", "all_logits", "and", "attention_mask", "batch_size", "bool", "cache_position", "class", "config", "contiguous", "cross_attentions", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_input_ids", "decoder_inputs_embeds", "decoder_ngram_attentions", "decoder_ngram_hidden_states", "def", "else", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "for", "forward", "if", "in", "input_ids", "inputs_embeds", "is", "is_contiguous", "kwargs", "labels", "lm_head", "logits", "logits_ngram", "loss", "ngram", "not", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "predict_logits", "predicting_streams", "r", "return", "return_dict", "self", "sequence_length", "shape", "torch", "tuple", "use_cache", "use_return_dict", "v", "view"], "prophetnet/modeling_prophetnet.py:ProphetNetForConditionalGeneration._compute_loss": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "True", "_compute_loss", "and", "break", "class", "config", "contiguous", "def", "dim", "disable_ngram_loss", "dtype", "eps", "eps_i", "expend_targets", "fill_", "float32", "for", "functional", "i", "if", "ignore_index", "in", "keepdim", "labels", "log_softmax", "logits", "loss", "lprobs", "mean", "ne", "new_zeros", "ngram", "nll_loss", "nn", "non_masked_tokens", "range", "reduction", "return", "self", "size", "smooth_loss", "sum", "torch", "transpose", "view"], "prophetnet/modeling_prophetnet.py:ProphetNetForConditionalGeneration.prepare_decoder_input_ids_from_labels": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "Tensor", "_shift_right", "class", "def", "labels", "prepare_decoder_input_ids_from_labels", "return", "self", "torch"], "prophetnet/modeling_prophetnet.py:ProphetNetForConditionalGeneration.get_encoder": ["GenerationMixin", "Model", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "class", "def", "else", "encoder", "get_encoder", "if", "is", "modality", "return", "self", "super"], "prophetnet/modeling_prophetnet.py:ProphetNetForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "Model", "ModelConfig", "ModelDecoderWrapper", "ModelForCausalLM", "ModelPreTrainedModel", "True", "__init__", "bias", "class", "config", "copy", "deepcopy", "def", "disable_ngram_loss", "hidden_size", "is_decoder", "is_encoder_decoder", "lm_head", "nn", "pad_token_id", "padding_idx", "post_init", "self", "super", "vocab_size"], "prophetnet/modeling_prophetnet.py:ProphetNetForCausalLM.get_input_embeddings": ["GenerationMixin", "Model", "ModelForCausalLM", "ModelPreTrainedModel", "class", "decoder", "def", "get_input_embeddings", "return", "self", "word_embeddings"], "prophetnet/modeling_prophetnet.py:ProphetNetForCausalLM.set_input_embeddings": ["GenerationMixin", "Model", "ModelForCausalLM", "ModelPreTrainedModel", "class", "decoder", "def", "self", "set_input_embeddings", "value", "word_embeddings"], "prophetnet/modeling_prophetnet.py:ProphetNetForCausalLM.forward": ["Cache", "GenerationMixin", "Model", "ModelDecoderLMOutput", "ModelForCausalLM", "ModelPreTrainedModel", "None", "Tensor", "_compute_loss", "all_logits", "attention_mask", "attentions", "batch_size", "bool", "class", "config", "cross_attentions", "decoder", "def", "else", "encoder_attention_mask", "encoder_hidden_states", "for", "forward", "hidden_states", "hidden_states_ngram", "if", "in", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "lm_head", "logits", "logits_ngram", "loss", "ngram", "ngram_attentions", "not", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "predict_logits", "predicting_streams", "r", "return", "return_dict", "self", "sequence_length", "shape", "torch", "tuple", "use_cache", "use_return_dict", "v", "view"], "prophetnet/modeling_prophetnet.py:ProphetNetForCausalLM._compute_loss": ["GenerationMixin", "ModelForCausalLM", "ModelPreTrainedModel", "True", "_compute_loss", "and", "break", "class", "config", "contiguous", "def", "dim", "disable_ngram_loss", "dtype", "eps", "eps_i", "expend_targets", "fill_", "float32", "for", "functional", "i", "if", "ignore_index", "in", "keepdim", "labels", "log_softmax", "logits", "loss", "lprobs", "mean", "ne", "new_zeros", "ngram", "nll_loss", "nn", "non_masked_tokens", "range", "reduction", "return", "self", "size", "smooth_loss", "sum", "torch", "transpose", "view"], "prophetnet/modeling_prophetnet.py:ProphetNetForCausalLM.prepare_inputs_for_generation": ["False", "GenerationMixin", "ModelForCausalLM", "ModelPreTrainedModel", "None", "and", "attention_mask", "cache_position", "class", "def", "for", "if", "in", "input_ids", "is", "is_first_iteration", "items", "key", "kwargs", "model_inputs", "new_ones", "not", "past_key_values", "pop", "prepare_inputs_for_generation", "return", "self", "shape", "use_cache", "value"], "prophetnet/modeling_prophetnet.py:ProphetNetDecoderWrapper.__init__": ["Embedding", "ModelConfig", "ModelDecoder", "ModelDecoderWrapper", "ModelPreTrainedModel", "__init__", "class", "config", "decoder", "def", "hidden_size", "nn", "pad_token_id", "padding_idx", "post_init", "self", "super", "vocab_size", "word_embeddings"], "prophetnet/modeling_prophetnet.py:ProphetNetDecoderWrapper.forward": ["ModelDecoderWrapper", "ModelPreTrainedModel", "args", "class", "decoder", "def", "forward", "kwargs", "return", "self"], "pvt/modeling_pvt.py:drop_path": ["False", "Model_path", "Model_prob", "Tensor", "bool", "def", "device", "div", "dtype", "float", "floor_", "if", "input", "keep_prob", "ndim", "not", "or", "output", "rand", "random_tensor", "return", "shape", "torch", "training"], "pvt/modeling_pvt.py:PvtDropPath.__init__": ["ModelDropPath", "Module", "None", "__init__", "class", "def", "drop_prob", "float", "nn", "self", "super"], "pvt/modeling_pvt.py:PvtDropPath.forward": ["ModelDropPath", "Module", "Tensor", "class", "def", "drop_path", "drop_prob", "forward", "hidden_states", "nn", "return", "self", "torch", "training"], "pvt/modeling_pvt.py:PvtDropPath.extra_repr": ["ModelDropPath", "Module", "class", "def", "drop_prob", "extra_repr", "f", "nn", "p", "return", "self", "str"], "pvt/modeling_pvt.py:PvtPatchEmbeddings.__init__": ["Conv2d", "Dropout", "False", "Iterable", "LayerNorm", "ModelConfig", "ModelPatchEmbeddings", "Module", "None", "Parameter", "__init__", "abc", "bool", "class", "cls_token", "collections", "config", "def", "dropout", "else", "eps", "hidden_dropout_prob", "hidden_size", "if", "image_size", "int", "isinstance", "kernel_size", "layer_norm", "layer_norm_eps", "nn", "num_channels", "num_patches", "p", "patch_size", "position_embeddings", "projection", "randn", "self", "stride", "super", "torch", "zeros"], "pvt/modeling_pvt.py:PvtPatchEmbeddings.interpolate_pos_encoding": ["F", "ModelPatchEmbeddings", "Module", "Tensor", "and", "bilinear", "class", "config", "def", "embeddings", "height", "if", "image_size", "int", "interpolate", "interpolate_pos_encoding", "interpolated_embeddings", "is_tracing", "jit", "mode", "nn", "not", "num_patches", "permute", "position_embeddings", "reshape", "return", "self", "size", "torch", "width"], "pvt/modeling_pvt.py:PvtPatchEmbeddings.forward": ["Make", "ModelPatchEmbeddings", "Module", "None", "Tensor", "ValueError", "_", "batch_size", "cat", "channel", "class", "cls_token", "configuration", "def", "dim", "dimension", "dropout", "else", "embeddings", "expand", "flatten", "forward", "height", "if", "in", "int", "interpolate_pos_encoding", "is", "layer_norm", "match", "nn", "not", "num_channels", "of", "one", "patch_embed", "pixel", "pixel_values", "position_embeddings", "projection", "raise", "return", "self", "set", "shape", "sure", "that", "the", "torch", "transpose", "tuple", "values", "width", "with"], "pvt/modeling_pvt.py:PvtSelfOutput.__init__": ["Dropout", "Linear", "ModelConfig", "ModelSelfOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "hidden_dropout_prob", "hidden_size", "int", "nn", "self", "super"], "pvt/modeling_pvt.py:PvtSelfOutput.forward": ["ModelSelfOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "nn", "return", "self", "torch"], "pvt/modeling_pvt.py:PvtEfficientSelfAttention.__init__": ["Conv2d", "Dropout", "LayerNorm", "Linear", "ModelConfig", "ModelEfficientSelfAttention", "Module", "The", "ValueError", "__init__", "a", "all_head_size", "attention", "attention_head_size", "attention_probs_dropout_prob", "bias", "class", "config", "def", "dropout", "eps", "f", "float", "heads", "hidden", "hidden_size", "if", "int", "is", "kernel_size", "key", "layer_norm", "layer_norm_eps", "multiple", "nn", "not", "num_attention_heads", "number", "of", "qkv_bias", "query", "raise", "self", "sequence_reduction", "sequences_reduction_ratio", "size", "stride", "super", "the", "value"], "pvt/modeling_pvt.py:PvtEfficientSelfAttention.transpose_for_scores": ["ModelEfficientSelfAttention", "Module", "Tensor", "attention_head_size", "class", "def", "hidden_states", "int", "new_shape", "nn", "num_attention_heads", "permute", "return", "self", "size", "torch", "transpose_for_scores", "view"], "pvt/modeling_pvt.py:PvtEfficientSelfAttention.forward": ["False", "ModelEfficientSelfAttention", "Module", "Tensor", "all_head_size", "attention_head_size", "attention_probs", "attention_scores", "batch_size", "bool", "class", "context_layer", "contiguous", "def", "dim", "dropout", "else", "forward", "functional", "height", "hidden_states", "if", "int", "key", "key_layer", "layer_norm", "math", "matmul", "new_context_layer_shape", "nn", "num_channels", "output_attentions", "outputs", "permute", "query", "query_layer", "reshape", "return", "self", "seq_len", "sequence_reduction", "sequences_reduction_ratio", "shape", "size", "softmax", "sqrt", "torch", "transpose", "transpose_for_scores", "tuple", "value", "value_layer", "view", "width"], "pvt/modeling_pvt.py:PvtAttention.__init__": ["ModelAttention", "ModelConfig", "ModelEfficientSelfAttention", "ModelSelfOutput", "Module", "__init__", "class", "config", "def", "float", "hidden_size", "int", "nn", "num_attention_heads", "output", "self", "sequences_reduction_ratio", "super"], "pvt/modeling_pvt.py:PvtAttention.forward": ["False", "ModelAttention", "Module", "Tensor", "attention_output", "bool", "class", "def", "forward", "height", "hidden_states", "int", "nn", "output", "output_attentions", "outputs", "return", "self", "self_outputs", "torch", "tuple", "width"], "pvt/modeling_pvt.py:PvtFFN.__init__": ["ACT2FN", "Dropout", "Linear", "ModelConfig", "ModelFFN", "Module", "None", "__init__", "class", "config", "def", "dense1", "dense2", "dropout", "else", "hidden_act", "hidden_dropout_prob", "hidden_features", "if", "in_features", "int", "intermediate_act_fn", "is", "isinstance", "nn", "not", "out_features", "self", "str", "super"], "pvt/modeling_pvt.py:PvtFFN.forward": ["ModelFFN", "Module", "Tensor", "class", "def", "dense1", "dense2", "dropout", "forward", "hidden_states", "intermediate_act_fn", "nn", "return", "self", "torch"], "pvt/modeling_pvt.py:PvtLayer.__init__": ["Identity", "LayerNorm", "ModelAttention", "ModelConfig", "ModelDropPath", "ModelFFN", "ModelLayer", "Module", "__init__", "attention", "class", "config", "def", "drop_path", "else", "eps", "float", "hidden_features", "hidden_size", "if", "in_features", "int", "layer_norm_1", "layer_norm_2", "layer_norm_eps", "mlp", "mlp_hidden_size", "mlp_ratio", "nn", "num_attention_heads", "self", "sequences_reduction_ratio", "super"], "pvt/modeling_pvt.py:PvtLayer.forward": ["False", "ModelLayer", "Module", "Tensor", "attention", "attention_output", "bool", "class", "def", "drop_path", "forward", "height", "hidden_states", "int", "layer_norm_1", "layer_norm_2", "layer_output", "mlp", "mlp_output", "nn", "output_attentions", "outputs", "return", "self", "self_attention_outputs", "torch", "width"], "pvt/modeling_pvt.py:PvtEncoder.__init__": ["LayerNorm", "ModelConfig", "ModelEncoder", "ModelLayer", "ModelPatchEmbeddings", "Module", "ModuleList", "__init__", "append", "block", "blocks", "class", "cls_token", "config", "cpu", "cur", "def", "depths", "device", "drop_path", "drop_path_decays", "drop_path_rate", "else", "embeddings", "eps", "for", "hidden_size", "hidden_sizes", "i", "if", "image_size", "in", "j", "layer_norm", "layer_norm_eps", "layers", "linspace", "mlp_ratio", "mlp_ratios", "nn", "num_attention_heads", "num_channels", "num_encoder_blocks", "patch_embeddings", "patch_size", "patch_sizes", "range", "self", "sequence_reduction_ratios", "sequences_reduction_ratio", "stride", "strides", "sum", "super", "tolist", "torch"], "pvt/modeling_pvt.py:PvtEncoder.forward": ["BaseModelOutput", "False", "FloatTensor", "ModelEncoder", "Module", "None", "True", "all_hidden_states", "all_self_attentions", "attentions", "batch_size", "block", "block_layer", "bool", "class", "contiguous", "def", "else", "embedding_layer", "enumerate", "for", "forward", "height", "hidden_states", "idx", "if", "in", "is", "last_hidden_state", "layer_norm", "layer_outputs", "len", "nn", "not", "num_blocks", "output_attentions", "output_hidden_states", "patch_embeddings", "permute", "pixel_values", "reshape", "return", "return_dict", "self", "shape", "torch", "tuple", "v", "width", "zip"], "pvt/modeling_pvt.py:PvtPreTrainedModel._init_weights": ["Conv2d", "LayerNorm", "Linear", "ModelPatchEmbeddings", "ModelPreTrainedModel", "Module", "None", "PreTrainedModel", "_init_weights", "bias", "class", "cls_token", "config", "def", "elif", "if", "init", "initializer_range", "is", "isinstance", "mean", "module", "nn", "not", "ones_", "position_embeddings", "self", "std", "trunc_normal_", "weight", "zeros_"], "pvt/modeling_pvt.py:PvtModel.__init__": ["ModelConfig", "ModelEncoder", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "config", "def", "encoder", "post_init", "self", "super"], "pvt/modeling_pvt.py:PvtModel.forward": ["BaseModelOutput", "FloatTensor", "ModelModel", "ModelPreTrainedModel", "None", "attentions", "bool", "class", "config", "def", "else", "encoder", "encoder_outputs", "forward", "hidden_states", "if", "is", "kwargs", "last_hidden_state", "not", "output_attentions", "output_hidden_states", "pixel_values", "return", "return_dict", "self", "sequence_output", "torch", "tuple", "use_return_dict"], "pvt/modeling_pvt.py:PvtForImageClassification.__init__": ["Identity", "Linear", "Model", "ModelConfig", "ModelForImageClassification", "ModelModel", "ModelPreTrainedModel", "None", "__init__", "class", "classifier", "config", "def", "else", "hidden_sizes", "if", "nn", "num_labels", "post_init", "self", "super"], "pvt/modeling_pvt.py:PvtForImageClassification.forward": ["ImageClassifierOutput", "Model", "ModelForImageClassification", "ModelPreTrainedModel", "None", "Tensor", "attentions", "bool", "class", "classifier", "config", "def", "else", "forward", "hidden_states", "if", "is", "kwargs", "labels", "logits", "loss", "loss_function", "not", "output", "output_attentions", "output_hidden_states", "outputs", "pixel_values", "r", "return", "return_dict", "self", "sequence_output", "torch", "tuple", "use_return_dict"], "pvt_v2/modeling_pvt_v2.py:drop_path": ["False", "Model_path", "Model_prob", "Tensor", "bool", "def", "device", "div", "dtype", "float", "floor_", "if", "input", "keep_prob", "ndim", "not", "or", "output", "rand", "random_tensor", "return", "shape", "torch", "training"], "pvt_v2/modeling_pvt_v2.py:PvtV2DropPath.__init__": ["ModelDropPath", "Module", "None", "__init__", "class", "def", "drop_prob", "float", "nn", "self", "super"], "pvt_v2/modeling_pvt_v2.py:PvtV2DropPath.forward": ["ModelDropPath", "Module", "Tensor", "class", "def", "drop_path", "drop_prob", "forward", "hidden_states", "nn", "return", "self", "torch", "training"], "pvt_v2/modeling_pvt_v2.py:PvtV2DropPath.extra_repr": ["ModelDropPath", "Module", "class", "def", "drop_prob", "extra_repr", "f", "nn", "p", "return", "self", "str"], "pvt_v2/modeling_pvt_v2.py:PvtV2OverlapPatchEmbeddings.__init__": ["Conv2d", "LayerNorm", "ModelConfig", "ModelOverlapPatchEmbeddings", "Module", "__init__", "class", "config", "def", "else", "eps", "hidden_size", "hidden_sizes", "if", "int", "isinstance", "kernel_size", "layer_idx", "layer_norm", "layer_norm_eps", "nn", "num_channels", "padding", "patch_size", "patch_sizes", "proj", "self", "stride", "strides", "super"], "pvt_v2/modeling_pvt_v2.py:PvtV2OverlapPatchEmbeddings.forward": ["ModelOverlapPatchEmbeddings", "Module", "_", "class", "def", "embeddings", "flatten", "forward", "height", "layer_norm", "nn", "pixel_values", "proj", "return", "self", "shape", "transpose", "width"], "pvt_v2/modeling_pvt_v2.py:PvtV2DepthWiseConv.__init__": ["Conv2d", "ModelConfig", "ModelDepthWiseConv", "Module", "True", "__init__", "bias", "class", "config", "def", "dim", "dwconv", "groups", "int", "nn", "self", "super"], "pvt_v2/modeling_pvt_v2.py:PvtV2DepthWiseConv.forward": ["ModelDepthWiseConv", "Module", "batch_size", "class", "def", "dwconv", "flatten", "forward", "height", "hidden_states", "nn", "num_channels", "return", "self", "seq_len", "shape", "transpose", "view", "width"], "pvt_v2/modeling_pvt_v2.py:PvtV2SelfAttention.__init__": ["AdaptiveAvgPool2d", "Conv2d", "Dropout", "GELU", "LayerNorm", "Linear", "ModelConfig", "ModelSelfAttention", "Module", "The", "ValueError", "__init__", "a", "act", "all_head_size", "attention", "attention_head_size", "attention_probs_dropout_prob", "attn_drop", "bias", "class", "config", "def", "elif", "eps", "f", "heads", "hidden", "hidden_dropout_prob", "hidden_size", "if", "int", "is", "kernel_size", "key", "layer_norm", "layer_norm_eps", "linear_attention", "multiple", "nn", "not", "num_attention_heads", "number", "of", "pool", "proj", "proj_drop", "qkv_bias", "query", "raise", "self", "size", "spatial_reduction", "spatial_reduction_ratio", "stride", "super", "the", "value"], "pvt_v2/modeling_pvt_v2.py:PvtV2SelfAttention.transpose_for_scores": ["ModelSelfAttention", "Module", "Tensor", "attention_head_size", "class", "def", "hidden_states", "new_shape", "nn", "num_attention_heads", "permute", "return", "self", "size", "torch", "transpose_for_scores", "view"], "pvt_v2/modeling_pvt_v2.py:PvtV2SelfAttention.forward": ["False", "ModelSelfAttention", "Module", "Tensor", "act", "attention_head_size", "attention_probs", "attention_scores", "attn_drop", "batch_size", "bool", "class", "context_layer", "def", "dim", "elif", "else", "forward", "functional", "height", "hidden_states", "if", "int", "key", "key_layer", "layer_norm", "linear_attention", "math", "matmul", "nn", "num_channels", "output_attentions", "outputs", "permute", "pool", "proj", "proj_drop", "query", "query_layer", "reshape", "return", "self", "seq_len", "shape", "softmax", "spatial_reduction", "spatial_reduction_ratio", "sqrt", "torch", "transpose", "transpose_for_scores", "tuple", "value", "value_layer", "width"], "pvt_v2/modeling_pvt_v2.py:PvtV2ConvFeedForwardNetwork.__init__": ["ACT2FN", "Dropout", "Identity", "Linear", "ModelConfig", "ModelConvFeedForwardNetwork", "ModelDepthWiseConv", "Module", "None", "ReLU", "__init__", "class", "config", "def", "dense1", "dense2", "dropout", "dwconv", "else", "hidden_act", "hidden_dropout_prob", "hidden_features", "if", "in_features", "int", "intermediate_act_fn", "is", "isinstance", "linear_attention", "nn", "not", "out_features", "relu", "self", "str", "super"], "pvt_v2/modeling_pvt_v2.py:PvtV2ConvFeedForwardNetwork.forward": ["ModelConvFeedForwardNetwork", "Module", "Tensor", "class", "def", "dense1", "dense2", "dropout", "dwconv", "forward", "height", "hidden_states", "intermediate_act_fn", "nn", "relu", "return", "self", "torch", "width"], "pvt_v2/modeling_pvt_v2.py:PvtV2BlockLayer.__init__": ["Identity", "LayerNorm", "ModelBlockLayer", "ModelConfig", "ModelConvFeedForwardNetwork", "ModelDropPath", "ModelSelfAttention", "Module", "__init__", "attention", "class", "config", "def", "drop_path", "else", "eps", "float", "hidden_features", "hidden_size", "hidden_sizes", "if", "in_features", "int", "layer_idx", "layer_norm_1", "layer_norm_2", "layer_norm_eps", "mlp", "mlp_hidden_size", "mlp_ratio", "mlp_ratios", "nn", "num_attention_heads", "self", "spatial_reduction_ratio", "sr_ratios", "super"], "pvt_v2/modeling_pvt_v2.py:PvtV2BlockLayer.forward": ["False", "ModelBlockLayer", "Module", "Tensor", "attention", "attention_output", "bool", "class", "def", "drop_path", "forward", "height", "hidden_states", "int", "layer_norm_1", "layer_norm_2", "layer_output", "mlp", "mlp_output", "nn", "output_attentions", "outputs", "return", "self", "self_attention_outputs", "torch", "width"], "pvt_v2/modeling_pvt_v2.py:PvtV2EncoderLayer.__init__": ["GradientCheckpointingLayer", "LayerNorm", "ModelBlockLayer", "ModelConfig", "ModelEncoderLayer", "ModelOverlapPatchEmbeddings", "ModuleList", "__init__", "append", "block_idx", "block_layers", "blocks", "class", "config", "cpu", "def", "depths", "device", "drop_path", "drop_path_decays", "drop_path_rate", "eps", "for", "hidden_sizes", "in", "int", "layer_idx", "layer_norm", "layer_norm_eps", "linspace", "nn", "patch_embedding", "range", "self", "sum", "super", "tolist", "torch"], "pvt_v2/modeling_pvt_v2.py:PvtV2EncoderLayer.forward": ["GradientCheckpointingLayer", "ModelEncoderLayer", "None", "all_self_attentions", "block", "blocks", "class", "def", "else", "for", "forward", "height", "hidden_states", "if", "in", "layer_norm", "layer_outputs", "output_attentions", "outputs", "patch_embedding", "return", "self", "width"], "pvt_v2/modeling_pvt_v2.py:PvtV2Encoder.__init__": ["False", "ModelConfig", "ModelEncoder", "ModelEncoderLayer", "Module", "ModuleList", "__init__", "class", "config", "def", "for", "gradient_checkpointing", "i", "in", "layers", "nn", "num_encoder_blocks", "range", "self", "super"], "pvt_v2/modeling_pvt_v2.py:PvtV2Encoder.forward": ["BaseModelOutput", "False", "FloatTensor", "ModelEncoder", "Module", "None", "True", "all_hidden_states", "all_self_attentions", "attentions", "batch_size", "bool", "class", "contiguous", "def", "else", "enumerate", "for", "forward", "height", "hidden_states", "idx", "if", "in", "is", "last_hidden_state", "layer", "layer_output", "layers", "nn", "not", "output_attentions", "output_hidden_states", "outputs", "permute", "pixel_values", "reshape", "return", "return_dict", "self", "shape", "torch", "tuple", "v", "width"], "pvt_v2/modeling_pvt_v2.py:PvtV2PreTrainedModel._init_weights": ["Conv2d", "LayerNorm", "Linear", "ModelPreTrainedModel", "None", "PreTrainedModel", "_init_weights", "bias", "class", "config", "def", "elif", "fan_out", "groups", "if", "init", "initializer_range", "is", "isinstance", "kernel_size", "math", "mean", "module", "nn", "normal_", "not", "ones_", "out_channels", "self", "sqrt", "std", "trunc_normal_", "weight", "zeros_"], "pvt_v2/modeling_pvt_v2.py:PvtV2Model.__init__": ["ModelConfig", "ModelEncoder", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "config", "def", "encoder", "post_init", "self", "super"], "pvt_v2/modeling_pvt_v2.py:PvtV2Model.forward": ["BaseModelOutput", "FloatTensor", "ModelModel", "ModelPreTrainedModel", "None", "attentions", "bool", "class", "config", "def", "else", "encoder", "encoder_outputs", "forward", "hidden_states", "if", "is", "kwargs", "last_hidden_state", "not", "output_attentions", "output_hidden_states", "pixel_values", "return", "return_dict", "self", "sequence_output", "torch", "tuple", "use_return_dict"], "pvt_v2/modeling_pvt_v2.py:PvtV2ForImageClassification.__init__": ["Identity", "Linear", "Model", "ModelConfig", "ModelForImageClassification", "ModelModel", "ModelPreTrainedModel", "None", "__init__", "class", "classifier", "config", "def", "else", "hidden_sizes", "if", "nn", "num_labels", "post_init", "self", "super"], "pvt_v2/modeling_pvt_v2.py:PvtV2ForImageClassification.forward": ["ImageClassifierOutput", "Model", "ModelForImageClassification", "ModelPreTrainedModel", "None", "Tensor", "attentions", "batch_size", "bool", "class", "classifier", "config", "def", "dim", "else", "forward", "hidden_sizes", "hidden_states", "if", "is", "kwargs", "labels", "logits", "loss", "loss_function", "mean", "not", "output", "output_attentions", "output_hidden_states", "outputs", "permute", "pixel_values", "r", "reshape", "return", "return_dict", "self", "sequence_output", "shape", "torch", "tuple", "use_return_dict"], "pvt_v2/modeling_pvt_v2.py:PvtV2Backbone.__init__": ["BackboneMixin", "ModelBackbone", "ModelConfig", "ModelModel", "__init__", "_init_backbone", "class", "config", "def", "hidden_sizes", "num_features", "self", "super"], "pvt_v2/modeling_pvt_v2.py:PvtV2Backbone.forward": ["BackboneMixin", "BackboneOutput", "FloatTensor", "ModelBackbone", "ModelModel", "None", "True", "attentions", "bool", "class", "config", "def", "else", "encoder", "enumerate", "feature_maps", "for", "forward", "hidden_states", "idx", "if", "in", "is", "kwargs", "not", "out_features", "output", "output_attentions", "output_hidden_states", "outputs", "pixel_values", "r", "return", "return_dict", "self", "stage", "stage_names", "torch", "use_return_dict"], "qwen2/modeling_qwen2.py:Qwen2MLP.__init__": ["ACT2FN", "False", "Linear", "ModelMLP", "Module", "__init__", "act_fn", "bias", "class", "config", "def", "down_proj", "gate_proj", "hidden_act", "hidden_size", "intermediate_size", "nn", "self", "super", "up_proj"], "qwen2/modeling_qwen2.py:Qwen2MLP.forward": ["ModelMLP", "Module", "act_fn", "class", "def", "down_proj", "forward", "gate_proj", "nn", "return", "self", "up_proj", "x"], "qwen2/modeling_qwen2.py:Qwen2RotaryEmbedding.__init__": ["Callable", "False", "ModelConfig", "ModelRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "qwen2/modeling_qwen2.py:Qwen2RotaryEmbedding.compute_default_rope_parameters": ["ModelConfig", "ModelRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "qwen2/modeling_qwen2.py:Qwen2RotaryEmbedding.forward": ["False", "ModelRotaryEmbedding", "Module", "None", "and", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "qwen2/modeling_qwen2.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "qwen2/modeling_qwen2.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cos", "def", "k", "k_embed", "q", "q_embed", "return", "rotate_half", "sin", "unsqueeze", "unsqueeze_dim"], "qwen2/modeling_qwen2.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "qwen2/modeling_qwen2.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value", "value_states"], "qwen2/modeling_qwen2.py:Qwen2Attention.__init__": ["False", "Linear", "ModelAttention", "ModelConfig", "Module", "None", "True", "__init__", "attention_dropout", "bias", "class", "config", "def", "else", "getattr", "hasattr", "head_dim", "hidden_size", "if", "int", "is_causal", "k_proj", "layer_idx", "layer_type", "layer_types", "nn", "num_attention_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "q_proj", "scaling", "self", "sliding_attention", "sliding_window", "super", "v_proj"], "qwen2/modeling_qwen2.py:Qwen2Attention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "FlashAttentionKwargs", "LongTensor", "ModelAttention", "Module", "None", "Tensor", "Unpack", "_attn_implementation", "apply_rotary_pos_emb", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "cache_kwargs", "cache_position", "class", "config", "contiguous", "cos", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "head_dim", "hidden_shape", "hidden_states", "if", "input_shape", "is", "k_proj", "key_states", "kwargs", "layer_idx", "nn", "not", "o_proj", "past_key_values", "position_embeddings", "q_proj", "query_states", "reshape", "return", "scaling", "self", "shape", "sin", "sliding_window", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "view"], "qwen2/modeling_qwen2.py:Qwen2RMSNorm.__init__": ["ModelRMSNorm", "Module", "None", "Parameter", "__init__", "class", "def", "eps", "float", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "qwen2/modeling_qwen2.py:Qwen2RMSNorm.forward": ["ModelRMSNorm", "Module", "Tensor", "True", "class", "def", "dtype", "float32", "forward", "hidden_states", "input_dtype", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "qwen2/modeling_qwen2.py:Qwen2RMSNorm.extra_repr": ["ModelRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "variance_epsilon", "weight"], "qwen2/modeling_qwen2.py:Qwen2DecoderLayer.__init__": ["GradientCheckpointingLayer", "ModelAttention", "ModelConfig", "ModelDecoderLayer", "ModelMLP", "ModelRMSNorm", "__init__", "attention_type", "class", "config", "def", "eps", "hidden_size", "input_layernorm", "int", "layer_idx", "layer_types", "mlp", "post_attention_layernorm", "rms_norm_eps", "self", "self_attn", "super"], "qwen2/modeling_qwen2.py:Qwen2DecoderLayer.forward": ["Cache", "False", "GradientCheckpointingLayer", "LongTensor", "ModelDecoderLayer", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "attention_mask", "bool", "cache_position", "class", "def", "forward", "hidden_states", "input_layernorm", "kwargs", "mlp", "past_key_values", "position_embeddings", "position_ids", "post_attention_layernorm", "residual", "return", "self", "self_attn", "torch", "tuple", "use_cache"], "qwen2/modeling_qwen2.py:Qwen2Model.__init__": ["Embedding", "False", "ModelConfig", "ModelDecoderLayer", "ModelModel", "ModelPreTrainedModel", "ModelRMSNorm", "ModelRotaryEmbedding", "ModuleList", "__init__", "class", "config", "def", "embed_tokens", "eps", "for", "gradient_checkpointing", "has_sliding_layers", "hidden_size", "in", "layer_idx", "layer_types", "layers", "nn", "norm", "num_hidden_layers", "pad_token_id", "padding_idx", "post_init", "range", "rms_norm_eps", "rotary_emb", "self", "sliding_attention", "super", "vocab_size"], "qwen2/modeling_qwen2.py:Qwen2Model.forward": ["BaseModelOutputWithPast", "Cache", "DynamicCache", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "and", "arange", "attention_mask", "attention_type", "bool", "cache_position", "causal_mask_mapping", "class", "config", "create_causal_mask", "create_sliding_window_causal_mask", "decoder_layer", "def", "device", "dict", "else", "embed_tokens", "exactly", "for", "forward", "full_attention", "get_seq_length", "has_sliding_layers", "hidden_states", "if", "in", "input_embeds", "input_ids", "inputs_embeds", "is", "isinstance", "kwargs", "last_hidden_state", "layers", "mask_kwargs", "must", "norm", "not", "num_hidden_layers", "of", "one", "or", "past_key_values", "past_seen_tokens", "position_embeddings", "position_ids", "raise", "return", "rotary_emb", "self", "shape", "sliding_attention", "specify", "torch", "unsqueeze", "use_cache"], "qwen2/modeling_qwen2.py:Qwen2ForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "ModelForCausalLM", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "vocab_size"], "qwen2/modeling_qwen2.py:Qwen2ForCausalLM.forward": ["BaseModelOutputWithPast", "Cache", "CausalLMOutputWithPast", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForCausalLM", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "outputs", "past_key_values", "position_ids", "r", "return", "self", "slice", "slice_indices", "torch", "use_cache", "vocab_size"], "qwen2_5_omni/modeling_qwen2_5_omni.py:kaiser_sinc_filter1d": ["False", "Model_sinc_filter1d", "Model_window", "arange", "attenuation", "beta", "cutoff", "def", "delta_f", "dtype", "elif", "else", "float32", "half_size", "half_width", "if", "is_even", "kernel_size", "math", "normalized_filter", "periodic", "pi", "return", "sinc", "sinc_filter", "sum", "time_indices", "torch", "view", "zeros"], "qwen2_5_omni/modeling_qwen2_5_omni.py:Qwen2_5OmniPreTrainedModel._init_weights": ["DownSample1d", "Model_5OmniPreTrainedModel", "Model_5_VisionRotaryEmbedding", "PreTrainedModel", "SinusoidsPositionEmbedding", "UpSample1d", "_init_weights", "arange", "cat", "channels", "class", "copy_", "cos", "cutoff", "def", "dim", "dtype", "elif", "exp", "filter", "filter_tensor", "float", "half_width", "if", "init", "inv_freq", "inv_timescales", "isinstance", "kaiser_sinc_filter1d", "kernel_size", "length", "log", "log_timescale_increment", "max_timescale", "module", "newaxis", "np", "positional_embedding", "ratio", "scaled_time", "self", "sin", "super", "theta", "torch"], "qwen2_5_omni/modeling_qwen2_5_omni.py:Qwen2_5OmniPreTrainedModelForConditionalGeneration._prepare_4d_causal_attention_mask_with_cache_position": ["Model_5OmniPreTrainedModel", "Model_5OmniPreTrainedModelForConditionalGeneration", "None", "Tensor", "_prepare_4d_causal_attention_mask_with_cache_position", "and", "arange", "attention_mask", "batch_size", "cache_position", "causal_mask", "class", "clone", "def", "device", "diagonal", "dim", "dtype", "else", "expand", "fill_value", "float", "full", "if", "int", "is", "mask_length", "masked_fill", "min_dtype", "not", "padding_mask", "reshape", "return", "self", "sequence_length", "shape", "target_length", "torch", "triu"], "qwen2_5_omni/modeling_qwen2_5_omni.py:Qwen2_5OmniPreTrainedModelForConditionalGeneration.get_llm_pos_ids_for_vision": ["Model_5OmniPreTrainedModel", "Model_5OmniPreTrainedModelForConditionalGeneration", "Tensor", "_llm_pos_ids", "append", "arange", "cat", "class", "def", "dim", "expand", "flatten", "get_llm_pos_ids_for_vision", "grid_hs", "grid_ws", "h_index", "int", "len", "list", "llm_grid_h", "llm_grid_w", "llm_pos_ids", "llm_pos_ids_list", "long", "return", "self", "spatial_merge_size", "stack", "start_idx", "t_index", "torch", "view", "vision_idx", "w_index"], "qwen2_5_omni/modeling_qwen2_5_omni.py:Qwen2_5OmniPreTrainedModelForConditionalGeneration.get_chunked_index": ["Model_5OmniPreTrainedModel", "Model_5OmniPreTrainedModelForConditionalGeneration", "Tensor", "_iter", "class", "current_chunk", "def", "get_chunked_index", "i", "if", "int", "len", "list", "remove_index", "return", "self", "start_idx", "token_indices", "tokens_per_chunk", "torch", "tuple", "while", "yield"], "qwen2_5_omni/modeling_qwen2_5_omni.py:Qwen2_5OmniPreTrainedModelForConditionalGeneration.get_rope_index": ["False", "LongTensor", "Model_5OmniPreTrainedModel", "Model_5OmniPreTrainedModelForConditionalGeneration", "None", "Tensor", "True", "_", "and", "append", "arange", "argwhere", "attention_mask", "audio_chunk_index", "audio_chunk_indexes", "audio_idx", "audio_len", "audio_llm_pos_ids", "audio_nums", "audio_seqlens", "audio_start_token_id", "audio_token_id", "bool", "bos_len", "cat", "class", "config", "cpu", "cumsum", "def", "device", "dim", "dtype", "ed_audio", "ed_image", "ed_video", "elif", "else", "enumerate", "eos_len", "expand", "float", "for", "get_chunked_index", "get_llm_pos_ids_for_vision", "get_rope_index", "grid_hs", "grid_t", "grid_ws", "i", "if", "image_grid_thw", "image_idx", "image_len", "image_nums", "image_token_id", "in", "index", "input_ids", "input_tokens", "int", "is", "j", "keepdim", "len", "list", "llm_pos_ids", "llm_pos_ids_list", "llm_positions", "long", "masked_fill_", "max", "max_position_ids", "min", "min_ed", "mrope_position_deltas", "multimodal_nums", "not", "ones", "or", "position_id_per_seconds", "position_ids", "prod", "range", "remain_audios", "remain_images", "remain_videos", "reshape", "return", "second_per_grids", "seconds_per_chunk", "self", "shape", "spatial_merge_size", "squeeze", "st", "st_idx", "sub_len", "sum", "t_index", "t_ntoken_per_chunk", "tensor", "text_len", "to", "tolist", "torch", "total_input_ids", "tuple", "unsqueeze", "use_audio_in_video", "video_chunk_index", "video_chunk_indexes", "video_grid_thw", "video_idx", "video_len", "video_llm_pos_ids", "video_nums", "video_token_id", "view", "vision_start_indices", "vision_start_token_id", "vision_tokens"], "qwen2_5_omni/modeling_qwen2_5_omni.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "qwen2_5_omni/modeling_qwen2_5_omni.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value", "value_states"], "qwen2_5_omni/modeling_qwen2_5_omni.py:Qwen2_5OmniAudioAttention.__init__": ["False", "Linear", "Model_5OmniAudioAttention", "Model_5OmniAudioEncoderConfig", "Module", "True", "ValueError", "__init__", "and", "attention_dropout", "be", "bias", "by", "class", "config", "d_model", "def", "divisible", "dropout", "embed_dim", "encoder_attention_heads", "f", "got", "head_dim", "if", "is_causal", "is_decoder", "k_proj", "must", "nn", "num_heads", "num_key_value_groups", "out_proj", "q_proj", "raise", "scaling", "self", "super", "v_proj"], "qwen2_5_omni/modeling_qwen2_5_omni.py:Qwen2_5OmniAudioAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "False", "Model_5OmniAudioAttention", "Module", "None", "Tensor", "_", "_attn_implementation", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "class", "config", "contiguous", "cu_seq_lens_k", "cu_seq_lens_q", "cu_seqlens", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "hidden_states", "if", "is_causal", "k_proj", "key_states", "kwargs", "max", "max_length_k", "max_length_q", "max_seqlen", "nn", "not", "num_heads", "out_proj", "q_proj", "query_states", "reshape", "return", "scaling", "self", "seq_length", "size", "torch", "training", "transpose", "tuple", "unsqueeze", "v_proj", "value_states"], "qwen2_5_omni/modeling_qwen2_5_omni.py:Qwen2_5OmniAudioEncoderLayer.__init__": ["ACT2FN", "GradientCheckpointingLayer", "LayerNorm", "Linear", "Model_5OmniAudioAttention", "Model_5OmniAudioEncoderConfig", "Model_5OmniAudioEncoderLayer", "__init__", "activation_dropout", "activation_fn", "activation_function", "class", "config", "d_model", "def", "dropout", "embed_dim", "encoder_ffn_dim", "fc1", "fc2", "final_layer_norm", "nn", "self", "self_attn", "self_attn_layer_norm", "super"], "qwen2_5_omni/modeling_qwen2_5_omni.py:Qwen2_5OmniAudioEncoderLayer.forward": ["GradientCheckpointingLayer", "Model_5OmniAudioEncoderLayer", "None", "Tensor", "activation_fn", "attention_mask", "clamp", "clamp_value", "class", "cu_seqlens", "def", "dtype", "fc1", "fc2", "final_layer_norm", "finfo", "float16", "forward", "hidden_states", "if", "kwargs", "max", "min", "outputs", "residual", "return", "self", "self_attn", "self_attn_layer_norm", "torch"], "qwen2_5_omni/modeling_qwen2_5_omni.py:SinusoidsPositionEmbedding.__init__": ["False", "ModelPositionEmbedding", "Module", "ValueError", "__init__", "arange", "cat", "channels", "class", "cos", "def", "dim", "even", "exp", "float", "if", "input", "inv_timescales", "length", "log", "log_timescale_increment", "max_timescale", "needs", "newaxis", "nn", "np", "persistent", "positional_embedding", "raise", "register_buffer", "scaled_time", "self", "sin", "super", "torch"], "qwen2_5_omni/modeling_qwen2_5_omni.py:SinusoidsPositionEmbedding.forward": ["ModelPositionEmbedding", "Module", "class", "def", "forward", "int", "nn", "positional_embedding", "return", "self", "seqlen"], "qwen2_5_omni/modeling_qwen2_5_omni.py:Qwen2_5OmniAudioEncoder.__init__": ["AvgPool1d", "Conv1d", "Embedding", "False", "LayerNorm", "Linear", "Model_5OmniAudioEncoder", "Model_5OmniAudioEncoderConfig", "Model_5OmniAudioEncoderLayer", "Model_5OmniPreTrainedModel", "ModuleList", "SinusoidsPositionEmbedding", "_", "__init__", "audio_bos_eos_token", "avg_pooler", "class", "config", "conv1", "conv2", "d_model", "def", "dropout", "else", "embed_dim", "embed_scale", "encoder_layers", "for", "gradient_checkpointing", "if", "in", "kernel_size", "layers", "ln_post", "math", "max_source_positions", "n_window", "nn", "num_mel_bins", "output_dim", "padding", "positional_embedding", "post_init", "proj", "range", "scale_embedding", "self", "sqrt", "stride", "super"], "qwen2_5_omni/modeling_qwen2_5_omni.py:Qwen2_5OmniAudioEncoder._freeze_parameters": ["False", "Model_5OmniAudioEncoder", "Model_5OmniPreTrainedModel", "_freeze_parameters", "_requires_grad", "class", "def", "for", "in", "param", "parameters", "requires_grad", "self"], "qwen2_5_omni/modeling_qwen2_5_omni.py:Qwen2_5OmniAudioEncoder.get_input_embeddings": ["Model_5OmniAudioEncoder", "Model_5OmniPreTrainedModel", "Module", "class", "conv1", "def", "get_input_embeddings", "nn", "return", "self"], "qwen2_5_omni/modeling_qwen2_5_omni.py:Qwen2_5OmniAudioEncoder.set_input_embeddings": ["Model_5OmniAudioEncoder", "Model_5OmniPreTrainedModel", "Module", "class", "conv1", "def", "nn", "self", "set_input_embeddings", "value"], "qwen2_5_omni/modeling_qwen2_5_omni.py:Qwen2_5OmniAudioEncoder._prepare_attention_mask": ["Model_5OmniAudioEncoder", "Model_5OmniPreTrainedModel", "None", "Tensor", "_attn_implementation", "_prepare_attention_mask", "attention_mask", "class", "config", "cu_seqlens", "def", "device", "dtype", "finfo", "flash_attention_2", "for", "full", "i", "if", "in", "inputs_tensor", "len", "min", "range", "return", "self", "seq_length", "shape", "torch"], "qwen2_5_omni/modeling_qwen2_5_omni.py:Qwen2_5OmniAudioEncoder.forward": ["BaseModelOutput", "F", "Model_5OmniAudioEncoder", "Model_5OmniPreTrainedModel", "None", "_prepare_attention_mask", "aftercnn_lens", "append", "attention_mask", "avg_pooler", "cat", "ceil", "chunk_lengths", "chunk_list", "chunk_num", "class", "conv1", "conv2", "cu_seqlens", "cumsum", "def", "device", "dim", "dtype", "each_audio_states", "encoder_layer", "feature_lens", "for", "forward", "functional", "gelu", "hidden_states", "hidden_states_list", "in", "input_features", "int32", "kwargs", "last_hidden_state", "layer_outputs", "layers", "ln_post", "long", "n_window", "nn", "pad", "padded_and_mask_function", "padded_embed", "padded_feature", "padded_mask", "padded_mask_after_cnn", "padding_side", "padding_value", "positional_embedding", "proj", "r", "return", "right", "self", "shape", "split", "sum", "tail_chunk_index", "tensor", "to", "token_audio", "token_audio_list", "tolist", "torch", "transpose", "transpose_", "unsqueeze", "value", "where", "zeros"], "qwen2_5_omni/modeling_qwen2_5_omni.py:Qwen2_5OmniAudioEncoder.padded_and_mask_function": ["Model_5OmniAudioEncoder", "Model_5OmniPreTrainedModel", "batch_mask", "batch_mask_after_cnn", "bool", "class", "def", "device", "dim", "dtype", "enumerate", "feature_lens_after_cnn", "fill_value", "for", "full", "i", "in", "len", "length", "long", "max", "max_len", "max_len_after_cnn", "padded_and_mask_function", "padded_tensor", "padding_side", "padding_value", "return", "right", "self", "shape", "size", "tensor_len", "tensor_list", "torch", "unsqueeze", "zeros"], "qwen2_5_omni/modeling_qwen2_5_omni.py:Qwen2_5OmniAudioEncoder._get_feat_extract_output_lengths": ["LongTensor", "Model_5OmniAudioEncoder", "Model_5OmniPreTrainedModel", "_get_feat_extract_output_lengths", "class", "def", "input_lengths", "output_lengths", "return", "self", "torch"], "qwen2_5_omni/modeling_qwen2_5_omni.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "qwen2_5_omni/modeling_qwen2_5_omni.py:apply_rotary_pos_emb_vision": ["Model_rotary_pos_emb_vision", "Tensor", "cos", "def", "dtype", "float", "freqs", "orig_dtype", "output", "repeat", "return", "rotate_half", "sin", "tensor", "to", "torch", "unsqueeze"], "qwen2_5_omni/modeling_qwen2_5_omni.py:Qwen2_5OmniVisionAttention.__init__": ["False", "Linear", "Model_5OmniVisionAttention", "Model_5OmniVisionEncoderConfig", "Module", "None", "True", "__init__", "attention_dropout", "bias", "class", "config", "def", "dim", "head_dim", "hidden_size", "is_causal", "k", "nn", "num_heads", "num_key_value_groups", "proj", "q", "scaling", "self", "super", "v"], "qwen2_5_omni/modeling_qwen2_5_omni.py:Qwen2_5OmniVisionAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "False", "Model_5OmniVisionAttention", "Module", "None", "Tensor", "_", "_attn_implementation", "apply_rotary_pos_emb_vision", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_outputs", "cat", "class", "config", "contiguous", "cu_seq_lens_k", "cu_seq_lens_q", "cu_seqlens", "def", "dim", "dropout", "eager", "eager_attention_forward", "else", "flash_attention_2", "for", "forward", "hidden_states", "if", "in", "is_causal", "k", "key_states", "kwargs", "lengths", "max", "max_length_k", "max_length_q", "max_seqlen", "nn", "not", "num_heads", "proj", "q", "query_states", "reshape", "return", "rotary_pos_emb", "scaling", "self", "seq_length", "shape", "split", "splits", "squeeze", "tensor", "tolist", "torch", "training", "transpose", "unsqueeze", "v", "value_states", "zip"], "qwen2_5_omni/modeling_qwen2_5_omni.py:Qwen2_5OmniMLP.__init__": ["ACT2FN", "False", "Linear", "Model_5OmniMLP", "Module", "__init__", "act_fn", "bias", "bool", "class", "config", "def", "down_proj", "gate_proj", "hidden_act", "hidden_size", "intermediate_size", "nn", "self", "super", "up_proj"], "qwen2_5_omni/modeling_qwen2_5_omni.py:Qwen2_5OmniMLP.forward": ["Model_5OmniMLP", "Module", "act_fn", "class", "def", "down_proj", "forward", "gate_proj", "hidden_state", "nn", "return", "self", "up_proj"], "qwen2_5_omni/modeling_qwen2_5_omni.py:Qwen2_5OmniVisionBlock.__init__": ["GradientCheckpointingLayer", "ModelRMSNorm", "Model_5OmniMLP", "Model_5OmniVisionAttention", "Model_5OmniVisionBlock", "Model_5OmniVisionEncoderConfig", "None", "True", "__init__", "attn", "bias", "class", "config", "def", "eps", "hidden_size", "mlp", "norm1", "norm2", "self", "super"], "qwen2_5_omni/modeling_qwen2_5_omni.py:Qwen2_5OmniVisionBlock.forward": ["GradientCheckpointingLayer", "Model_5OmniVisionBlock", "None", "Tensor", "attn", "class", "cu_seqlens", "def", "forward", "hidden_states", "kwargs", "mlp", "norm1", "norm2", "return", "rotary_pos_emb", "self", "torch"], "qwen2_5_omni/modeling_qwen2_5_omni.py:Qwen2_5_VisionRotaryEmbedding.__init__": ["False", "Model_5_VisionRotaryEmbedding", "Module", "None", "__init__", "arange", "class", "def", "dim", "dtype", "float", "int", "inv_freq", "nn", "persistent", "register_buffer", "self", "super", "theta", "torch"], "qwen2_5_omni/modeling_qwen2_5_omni.py:Qwen2_5_VisionRotaryEmbedding.forward": ["Model_5_VisionRotaryEmbedding", "Module", "Tensor", "arange", "class", "def", "device", "dtype", "forward", "freqs", "int", "inv_freq", "nn", "outer", "return", "self", "seq", "seqlen", "torch"], "qwen2_5_omni/modeling_qwen2_5_omni.py:Qwen2_5_VisionPatchEmbed.__init__": ["Conv3d", "False", "Model_5_VisionPatchEmbed", "Module", "None", "__init__", "bias", "class", "def", "embed_dim", "in_channels", "int", "kernel_size", "nn", "patch_size", "proj", "self", "stride", "super", "temporal_patch_size"], "qwen2_5_omni/modeling_qwen2_5_omni.py:Qwen2_5_VisionPatchEmbed.forward": ["Model_5_VisionPatchEmbed", "Module", "Tensor", "class", "def", "dtype", "embed_dim", "forward", "hidden_states", "in_channels", "nn", "patch_size", "proj", "return", "self", "target_dtype", "temporal_patch_size", "to", "torch", "view", "weight"], "qwen2_5_omni/modeling_qwen2_5_omni.py:Qwen2_5OmniPatchMerger.__init__": ["GELU", "Linear", "ModelRMSNorm", "Model_5OmniPatchMerger", "Module", "None", "Sequential", "__init__", "class", "context_dim", "def", "dim", "eps", "hidden_size", "int", "ln_q", "mlp", "nn", "self", "spatial_merge_size", "super"], "qwen2_5_omni/modeling_qwen2_5_omni.py:Qwen2_5OmniPatchMerger.forward": ["Model_5OmniPatchMerger", "Module", "Tensor", "class", "def", "forward", "hidden_size", "ln_q", "mlp", "nn", "return", "self", "torch", "view", "x"], "qwen2_5_omni/modeling_qwen2_5_omni.py:Qwen2_5OmniVisionEncoder.__init__": ["False", "Model_5OmniPatchMerger", "Model_5OmniPreTrainedModel", "Model_5OmniVisionBlock", "Model_5OmniVisionEncoder", "Model_5OmniVisionEncoderConfig", "Model_5_VisionPatchEmbed", "Model_5_VisionRotaryEmbedding", "ModuleList", "None", "_", "__init__", "blocks", "class", "config", "context_dim", "def", "depth", "dim", "embed_dim", "for", "fullatt_block_indexes", "gradient_checkpointing", "head_dim", "hidden_size", "in", "in_channels", "inputs", "kwargs", "merger", "nn", "num_heads", "out_hidden_size", "patch_embed", "patch_size", "post_init", "range", "rotary_pos_emb", "self", "spatial_merge_size", "spatial_merge_unit", "super", "temporal_patch_size", "window_size"], "qwen2_5_omni/modeling_qwen2_5_omni.py:Qwen2_5OmniVisionEncoder.rot_pos_emb": ["Model_5OmniPreTrainedModel", "Model_5OmniVisionEncoder", "append", "arange", "cat", "class", "def", "dim", "expand", "flatten", "for", "grid_thw", "h", "hpos_ids", "in", "max", "max_grid_size", "permute", "pos_ids", "repeat", "reshape", "return", "rot_pos_emb", "rotary_pos_emb", "rotary_pos_emb_full", "self", "spatial_merge_size", "stack", "t", "torch", "unsqueeze", "w", "wpos_ids"], "qwen2_5_omni/modeling_qwen2_5_omni.py:Qwen2_5OmniVisionEncoder.get_window_index": ["F", "Model_5OmniPreTrainedModel", "Model_5OmniVisionEncoder", "append", "arange", "cat", "class", "constant", "cu_seqlens_tmp", "cu_window_seqlens", "cumsum", "def", "dim", "extend", "for", "get_window_index", "grid_h", "grid_t", "grid_thw", "grid_w", "in", "index", "index_new", "index_padded", "item", "list", "llm_grid_h", "llm_grid_w", "num_windows_h", "num_windows_w", "pad", "pad_h", "pad_w", "patch_size", "permute", "reshape", "return", "self", "seqlens", "spatial_merge_size", "spatial_merge_unit", "sum", "tolist", "torch", "vit_merger_window_size", "window_index", "window_index_id", "window_size"], "qwen2_5_omni/modeling_qwen2_5_omni.py:Qwen2_5OmniVisionEncoder.forward": ["F", "Model_5OmniPreTrainedModel", "Model_5OmniVisionEncoder", "Tensor", "_", "argsort", "blk", "blocks", "class", "cu_seqlens", "cu_seqlens_now", "cu_window_seqlens", "cumsum", "def", "device", "dim", "dtype", "else", "enumerate", "for", "forward", "fullatt_block_indexes", "get_window_index", "grid_thw", "hidden_states", "if", "in", "int32", "is_tracing", "jit", "kwargs", "layer_num", "merger", "pad", "patch_embed", "repeat_interleave", "reshape", "return", "reverse_indices", "rot_pos_emb", "rotary_pos_emb", "self", "seq_len", "size", "spatial_merge_unit", "tensor", "torch", "unique_consecutive", "value", "window_index"], "qwen2_5_omni/modeling_qwen2_5_omni.py:Qwen2_5OmniRotaryEmbedding.__init__": ["Callable", "False", "Model_5OmniRotaryEmbedding", "Model_5OmniThinkerConfig", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "qwen2_5_omni/modeling_qwen2_5_omni.py:Qwen2_5OmniRotaryEmbedding.compute_default_rope_parameters": ["Model_5OmniConfig", "Model_5OmniRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "qwen2_5_omni/modeling_qwen2_5_omni.py:Qwen2_5OmniRotaryEmbedding.forward": ["False", "Model_5OmniRotaryEmbedding", "Module", "None", "and", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "qwen2_5_omni/modeling_qwen2_5_omni.py:apply_multimodal_rotary_pos_emb": ["Model_multimodal_rotary_pos_emb", "cat", "cos", "def", "dim", "enumerate", "for", "i", "in", "k", "k_embed", "m", "mrope_section", "q", "q_embed", "return", "rotate_half", "sin", "split", "torch", "unsqueeze", "unsqueeze_dim"], "qwen2_5_omni/modeling_qwen2_5_omni.py:Qwen2_5OmniAttention.__init__": ["False", "Instantiating", "Linear", "Model_5OmniAttention", "Model_5OmniConfig", "Module", "None", "Please", "True", "__class__", "__init__", "__name__", "a", "and", "attention_dropout", "bias", "caching", "call", "class", "config", "creating", "def", "during", "else", "errors", "f", "forward", "getattr", "hasattr", "head_dim", "hidden_size", "if", "int", "is", "is_causal", "k_proj", "layer_idx", "layer_type", "layer_types", "logger", "make", "nn", "not", "num_attention_heads", "num_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "passing", "provide", "q_proj", "recommended", "rope_parameters", "scaling", "self", "sliding_attention", "sliding_window", "super", "sure", "the", "this", "to", "used", "v_proj", "warning_once", "when", "will", "without"], "qwen2_5_omni/modeling_qwen2_5_omni.py:Qwen2_5OmniAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "False", "FlashAttentionKwargs", "LongTensor", "Model_5OmniAttention", "Module", "None", "Tensor", "Unpack", "_", "_attn_implementation", "apply_multimodal_rotary_pos_emb", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "bool", "bsz", "cache_kwargs", "cache_position", "class", "config", "contiguous", "cos", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "head_dim", "hidden_states", "if", "is", "k_proj", "key_states", "kwargs", "layer_idx", "mrope_section", "nn", "not", "o_proj", "output_attentions", "past_key_values", "position_embeddings", "position_ids", "q_len", "q_proj", "query_states", "reshape", "return", "rope_parameters", "scaling", "self", "sin", "size", "sliding_window", "torch", "training", "transpose", "tuple", "update", "use_cache", "v_proj", "value_states", "view"], "qwen2_5_omni/modeling_qwen2_5_omni.py:Qwen2MLP.__init__": ["ACT2FN", "False", "Linear", "ModelMLP", "Module", "__init__", "act_fn", "bias", "bool", "class", "config", "def", "down_proj", "gate_proj", "hidden_act", "hidden_size", "intermediate_size", "nn", "self", "super", "up_proj"], "qwen2_5_omni/modeling_qwen2_5_omni.py:Qwen2MLP.forward": ["ModelMLP", "Module", "act_fn", "class", "def", "down_proj", "forward", "gate_proj", "hidden_state", "nn", "return", "self", "up_proj"], "qwen2_5_omni/modeling_qwen2_5_omni.py:Qwen2_5OmniDecoderLayer.__init__": ["Attention", "GradientCheckpointingLayer", "ModelMLP", "ModelRMSNorm", "Model_5OmniAttention", "Model_5OmniDecoderLayer", "Model_5OmniTextConfig", "Sliding", "Window", "__init__", "_attn_implementation", "and", "attention_type", "be", "but", "class", "config", "def", "enabled", "encountered", "eps", "f", "flash_attention_2", "for", "hidden_size", "if", "implemented", "input_layernorm", "int", "is", "layer_idx", "layer_types", "logger", "may", "mlp", "not", "post_attention_layernorm", "results", "rms_norm_eps", "self", "self_attn", "super", "unexpected", "use_sliding_window", "warning_once"], "qwen2_5_omni/modeling_qwen2_5_omni.py:Qwen2_5OmniDecoderLayer.forward": ["Cache", "False", "FlashAttentionKwargs", "FloatTensor", "GradientCheckpointingLayer", "LongTensor", "Model_5OmniDecoderLayer", "None", "Tensor", "Unpack", "attention_mask", "bool", "cache_position", "class", "def", "forward", "hidden_states", "if", "input_layernorm", "kwargs", "mlp", "output_attentions", "outputs", "past_key_values", "position_embeddings", "position_ids", "post_attention_layernorm", "residual", "return", "self", "self_attn", "self_attn_weights", "torch", "tuple", "use_cache"], "qwen2_5_omni/modeling_qwen2_5_omni.py:Qwen2_5OmniThinkerTextModel.__init__": ["Embedding", "False", "ModelRMSNorm", "Model_5OmniDecoderLayer", "Model_5OmniPreTrainedModel", "Model_5OmniRotaryEmbedding", "Model_5OmniTextConfig", "Model_5OmniThinkerTextModel", "ModuleList", "__init__", "_attn_implementation", "class", "config", "def", "embed_tokens", "eps", "for", "gradient_checkpointing", "has_sliding_layers", "hidden_size", "in", "layer_idx", "layer_types", "layers", "nn", "norm", "num_hidden_layers", "pad_token_id", "padding_idx", "post_init", "range", "rms_norm_eps", "rotary_emb", "self", "sliding_attention", "super", "vocab_size"], "qwen2_5_omni/modeling_qwen2_5_omni.py:Qwen2_5OmniThinkerTextModel.forward": ["BaseModelOutputWithPast", "Cache", "DynamicCache", "False", "FlashAttentionKwargs", "FloatTensor", "LongTensor", "Model_5OmniPreTrainedModel", "Model_5OmniThinkerTextModel", "None", "Setting", "Tensor", "True", "Unpack", "ValueError", "You", "all_hidden_states", "all_self_attns", "and", "arange", "attention_mask", "attention_type", "attentions", "bool", "cache_position", "causal_mask_mapping", "checkpointing", "class", "config", "create_causal_mask", "create_sliding_window_causal_mask", "decoder_layer", "def", "device", "dict", "elif", "else", "embed_tokens", "exactly", "expand", "for", "forward", "full_attention", "get_seq_length", "gradient", "gradient_checkpointing", "has_sliding_layers", "hidden_states", "if", "in", "incompatible", "input_embeds", "input_ids", "inputs_embeds", "is", "is_tracing", "isinstance", "jit", "kwargs", "last_hidden_state", "layer_outputs", "layers", "logger", "mask_kwargs", "must", "ndim", "norm", "not", "of", "one", "or", "output_attentions", "output_hidden_states", "past_key_values", "past_seen_tokens", "position_embeddings", "position_ids", "raise", "return", "return_dict", "rotary_emb", "self", "shape", "sliding_attention", "specify", "text_position_ids", "torch", "training", "tuple", "use_cache", "use_return_dict", "v", "view", "warning_once", "with"], "qwen2_5_omni/modeling_qwen2_5_omni.py:Qwen2_5OmniThinkerForConditionalGeneration.__init__": ["False", "GenerationMixin", "Linear", "Model_5OmniAudioEncoder", "Model_5OmniPreTrainedModelForConditionalGeneration", "Model_5OmniThinkerConfig", "Model_5OmniThinkerForConditionalGeneration", "Model_5OmniThinkerTextModel", "Model_5OmniVisionEncoder", "None", "__init__", "_from_config", "audio_config", "audio_tower", "bias", "class", "config", "def", "else", "hidden_size", "if", "is", "lm_head", "model", "nn", "not", "pad_token_id", "post_init", "rope_deltas", "self", "spatial_merge_size", "super", "text_config", "vision_config", "visual", "vocab_size"], "qwen2_5_omni/modeling_qwen2_5_omni.py:Qwen2_5OmniThinkerForConditionalGeneration.get_input_embeddings": ["GenerationMixin", "Model_5OmniPreTrainedModelForConditionalGeneration", "Model_5OmniThinkerForConditionalGeneration", "class", "def", "get_input_embeddings", "model", "return", "self"], "qwen2_5_omni/modeling_qwen2_5_omni.py:Qwen2_5OmniThinkerForConditionalGeneration.set_input_embeddings": ["GenerationMixin", "Model_5OmniPreTrainedModelForConditionalGeneration", "Model_5OmniThinkerForConditionalGeneration", "class", "def", "model", "self", "set_input_embeddings", "value"], "qwen2_5_omni/modeling_qwen2_5_omni.py:Qwen2_5OmniThinkerForConditionalGeneration.get_video_features": ["FloatTensor", "GenerationMixin", "LongTensor", "Model_5OmniPreTrainedModelForConditionalGeneration", "Model_5OmniThinkerForConditionalGeneration", "None", "class", "def", "dtype", "get_video_features", "grid_thw", "pixel_values_videos", "return", "self", "torch", "type", "video_embeds", "video_grid_thw", "visual"], "qwen2_5_omni/modeling_qwen2_5_omni.py:Qwen2_5OmniThinkerForConditionalGeneration.get_image_features": ["FloatTensor", "GenerationMixin", "LongTensor", "Model_5OmniPreTrainedModelForConditionalGeneration", "Model_5OmniThinkerForConditionalGeneration", "None", "class", "def", "dtype", "get_image_features", "grid_thw", "image_embeds", "image_grid_thw", "pixel_values", "return", "self", "torch", "type", "visual"], "qwen2_5_omni/modeling_qwen2_5_omni.py:Qwen2_5OmniThinkerForConditionalGeneration.get_audio_features": ["FloatTensor", "GenerationMixin", "LongTensor", "Model_5OmniPreTrainedModelForConditionalGeneration", "Model_5OmniThinkerForConditionalGeneration", "None", "ValueError", "_get_feat_extract_output_lengths", "aftercnn_lens", "audio_feat_lengths", "audio_feature_lengths", "audio_features", "audio_output_lengths", "audio_outputs", "audio_tower", "bool", "class", "def", "dim", "else", "feature_attention_mask", "feature_lens", "get_audio_features", "if", "input_features", "is", "last_hidden_state", "length", "match", "not", "of", "permute", "raise", "return", "self", "shape", "should", "sum", "tolist", "torch"], "qwen2_5_omni/modeling_qwen2_5_omni.py:Qwen2_5OmniThinkerForConditionalGeneration.get_placeholder_mask": ["FloatTensor", "GenerationMixin", "Image", "LongTensor", "Model_5OmniPreTrainedModelForConditionalGeneration", "Model_5OmniThinkerForConditionalGeneration", "None", "ValueError", "Videos", "all", "and", "audio_token_id", "class", "config", "def", "device", "do", "dtype", "else", "expand_as", "f", "features", "get_input_embeddings", "get_placeholder_mask", "if", "image", "image_features", "image_token_id", "input_ids", "inputs_embeds", "is", "long", "match", "n_image_tokens", "n_video_tokens", "not", "numel", "raise", "return", "self", "shape", "special_audio_mask", "special_image_mask", "special_video_mask", "sum", "tensor", "to", "tokens", "torch", "unsqueeze", "video", "video_features", "video_token_id"], "qwen2_5_omni/modeling_qwen2_5_omni.py:Qwen2_5OmniThinkerForConditionalGeneration.forward": ["Cache", "FloatTensor", "GenerationMixin", "LongTensor", "Model_5OmniPreTrainedModelForConditionalGeneration", "Model_5OmniThinkerCausalLMOutputWithPast", "Model_5OmniThinkerForConditionalGeneration", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "add", "and", "arange", "attention_mask", "attentions", "audio_feature_lengths", "audio_features", "audio_mask", "batch_size", "bool", "cache_position", "class", "config", "def", "delta", "delta0", "device", "dim", "dtype", "else", "expand", "feature_attention_mask", "forward", "get_audio_features", "get_image_features", "get_input_embeddings", "get_placeholder_mask", "get_rope_index", "get_seq_length", "get_text_config", "get_video_features", "hidden_states", "if", "image_embeds", "image_features", "image_grid_thw", "image_mask", "input_features", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "lm_head", "logits", "loss", "loss_function", "masked_scatter", "model", "not", "or", "output", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "past_key_values_length", "pixel_values", "pixel_values_videos", "position_ids", "r", "return", "return_dict", "rope_deltas", "self", "seq_length", "shape", "sum", "to", "torch", "tuple", "unsqueeze", "use_audio_in_video", "use_cache", "use_return_dict", "video_embeds", "video_features", "video_grid_thw", "video_mask", "video_second_per_grid", "view", "vocab_size"], "qwen2_5_omni/modeling_qwen2_5_omni.py:Qwen2_5OmniThinkerForConditionalGeneration.prepare_inputs_for_generation": ["False", "GenerationMixin", "Model_5OmniPreTrainedModelForConditionalGeneration", "Model_5OmniThinkerForConditionalGeneration", "None", "True", "and", "attention_mask", "cache_position", "class", "def", "feature_attention_mask", "if", "image_grid_thw", "input_features", "input_ids", "inputs_embeds", "is_first_iteration", "kwargs", "model_inputs", "not", "past_key_values", "pixel_values", "pixel_values_videos", "position_ids", "prepare_inputs_for_generation", "return", "self", "super", "use_audio_in_video", "use_cache", "video_grid_thw", "video_second_per_grid"], "qwen2_5_omni/modeling_qwen2_5_omni.py:Qwen2_5OmniTalkerModel.__init__": ["Embedding", "False", "ModelRMSNorm", "Model_5OmniDecoderLayer", "Model_5OmniPreTrainedModel", "Model_5OmniRotaryEmbedding", "Model_5OmniTalkerConfig", "Model_5OmniTalkerModel", "ModuleList", "__init__", "_attn_implementation", "class", "config", "def", "embed_tokens", "embedding_size", "eps", "for", "gradient_checkpointing", "has_sliding_layers", "hidden_size", "in", "layer_idx", "layer_types", "layers", "nn", "norm", "num_hidden_layers", "pad_token_id", "padding_idx", "post_init", "range", "rms_norm_eps", "rotary_emb", "self", "sliding_attention", "super", "vocab_size"], "qwen2_5_omni/modeling_qwen2_5_omni.py:Qwen2_5OmniTalkerModel.forward": ["BaseModelOutputWithPast", "Cache", "DynamicCache", "False", "FlashAttentionKwargs", "FloatTensor", "LongTensor", "Model_5OmniPreTrainedModel", "Model_5OmniTalkerModel", "None", "Setting", "Tensor", "True", "Unpack", "ValueError", "You", "all_hidden_states", "all_self_attns", "and", "arange", "attention_mask", "attention_type", "attentions", "bool", "cache_position", "causal_mask_mapping", "checkpointing", "class", "config", "create_causal_mask", "create_sliding_window_causal_mask", "decoder_layer", "def", "device", "dict", "elif", "else", "embed_tokens", "exactly", "expand", "for", "forward", "full_attention", "get_seq_length", "gradient", "gradient_checkpointing", "has_sliding_layers", "hidden_states", "if", "in", "incompatible", "input_embeds", "input_ids", "inputs_embeds", "is", "is_tracing", "isinstance", "jit", "kwargs", "last_hidden_state", "layer_outputs", "layers", "logger", "mask_kwargs", "must", "ndim", "norm", "not", "of", "one", "or", "output_attentions", "output_hidden_states", "past_key_values", "past_seen_tokens", "position_embeddings", "position_ids", "raise", "return", "return_dict", "rotary_emb", "self", "shape", "sliding_attention", "specify", "text_position_ids", "torch", "training", "tuple", "use_cache", "use_return_dict", "v", "view", "warning_once", "with"], "qwen2_5_omni/modeling_qwen2_5_omni.py:Qwen2_5OmniTalkerForConditionalGeneration.__init__": ["False", "GenerationMixin", "Linear", "Model_5OmniPreTrainedModelForConditionalGeneration", "Model_5OmniTalkerConfig", "Model_5OmniTalkerForConditionalGeneration", "Model_5OmniTalkerModel", "None", "__init__", "bias", "class", "codebook_size", "codec_bos_token", "codec_eos_token", "codec_head", "codec_mask_token", "codec_pad_token", "config", "def", "embedding_size", "hidden_size", "model", "nn", "post_init", "rope_deltas", "self", "spatial_merge_size", "super", "text_bos_token", "text_eos_token", "text_pad_token", "thinker_to_talker_proj", "tts_codec_end_token_id", "tts_codec_mask_token_id", "tts_codec_pad_token_id", "tts_codec_start_token_id", "tts_text_end_token_id", "tts_text_pad_token_id", "tts_text_start_token_id", "vocab_size"], "qwen2_5_omni/modeling_qwen2_5_omni.py:Qwen2_5OmniTalkerForConditionalGeneration.get_input_embeddings": ["GenerationMixin", "Model_5OmniPreTrainedModelForConditionalGeneration", "Model_5OmniTalkerForConditionalGeneration", "class", "def", "get_input_embeddings", "model", "return", "self"], "qwen2_5_omni/modeling_qwen2_5_omni.py:Qwen2_5OmniTalkerForConditionalGeneration.set_input_embeddings": ["GenerationMixin", "Model_5OmniPreTrainedModelForConditionalGeneration", "Model_5OmniTalkerForConditionalGeneration", "class", "def", "model", "self", "set_input_embeddings", "value"], "qwen2_5_omni/modeling_qwen2_5_omni.py:Qwen2_5OmniTalkerForConditionalGeneration.forward": ["Cache", "FloatTensor", "GenerationMixin", "LongTensor", "Model_5OmniPreTrainedModelForConditionalGeneration", "Model_5OmniTalkerCausalLMOutputWithPast", "Model_5OmniTalkerForConditionalGeneration", "None", "Tensor", "_", "add", "and", "arange", "attention_mask", "attentions", "audio_feature_lengths", "batch_size", "bool", "cache_position", "class", "codec_bos_token", "codec_embeds", "codec_head", "codec_pad_token", "config", "def", "delta", "device", "dtype", "else", "expand", "float", "forward", "get_input_embeddings", "get_rope_index", "get_seq_length", "hidden_states", "if", "image_grid_thw", "input_ids", "input_text_ids", "inputs_embeds", "is", "kwargs", "logits", "long", "loss", "model", "not", "or", "output", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "past_key_values_length", "position_ids", "r", "return", "return_dict", "rope_deltas", "self", "seq_length", "shape", "talker_lm_input", "tensor", "thinker_reply_part", "thinker_to_talker_proj", "to", "torch", "tuple", "unsqueeze", "use_audio_in_video", "use_cache", "use_return_dict", "video_grid_thw", "video_second_per_grid", "view"], "qwen2_5_omni/modeling_qwen2_5_omni.py:Qwen2_5OmniTalkerForConditionalGeneration._get_initial_cache_position": ["GenerationMixin", "Model_5OmniPreTrainedModelForConditionalGeneration", "Model_5OmniTalkerForConditionalGeneration", "_get_initial_cache_position", "class", "def", "device", "inputs_embeds", "model_kwargs", "pop", "return", "self", "seq_length", "super"], "qwen2_5_omni/modeling_qwen2_5_omni.py:Qwen2_5OmniTalkerForConditionalGeneration.prepare_inputs_for_generation": ["False", "GenerationMixin", "Model_5OmniPreTrainedModelForConditionalGeneration", "Model_5OmniTalkerForConditionalGeneration", "None", "True", "attention_mask", "audio_feature_attention_mask", "audio_feature_lengths", "cache_position", "class", "def", "image_grid_thw", "input_audio_features", "input_ids", "input_text_ids", "inputs_embeds", "kwargs", "model_inputs", "past_key_values", "pixel_values", "pixel_values_videos", "position_ids", "prepare_inputs_for_generation", "return", "self", "super", "thinker_reply_part", "use_audio_in_video", "use_cache", "video_grid_thw", "video_second_per_grid"], "qwen2_5_omni/modeling_qwen2_5_omni.py:Qwen2_5OmniTalkerForConditionalGeneration._update_model_kwargs_for_generation": ["Any", "False", "GenerationMixin", "ModelOutput", "Model_5OmniPreTrainedModelForConditionalGeneration", "Model_5OmniTalkerForConditionalGeneration", "None", "_update_model_kwargs_for_generation", "bool", "class", "def", "dict", "getattr", "if", "int", "is", "is_encoder_decoder", "model_kwargs", "not", "num_new_tokens", "outputs", "return", "self", "str", "super", "thinker_reply_part"], "qwen2_5_omni/modeling_qwen2_5_omni.py:Qwen2_5OmniDiTRotaryEmbedding.__init__": ["Callable", "False", "Model_5OmniDiTConfig", "Model_5OmniDiTRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "qwen2_5_omni/modeling_qwen2_5_omni.py:Qwen2_5OmniDiTRotaryEmbedding.compute_default_rope_parameters": ["Model_5OmniDiTConfig", "Model_5OmniDiTRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "qwen2_5_omni/modeling_qwen2_5_omni.py:Qwen2_5OmniDiTRotaryEmbedding.forward": ["False", "Model_5OmniDiTRotaryEmbedding", "Module", "None", "and", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "qwen2_5_omni/modeling_qwen2_5_omni.py:TimeDelayNetBlock.__init__": ["Conv1d", "ModelDelayNetBlock", "Module", "ReLU", "__init__", "activation", "class", "conv", "def", "dilation", "in_channels", "kernel_size", "nn", "out_channels", "padding", "padding_mode", "reflect", "same", "self", "super"], "qwen2_5_omni/modeling_qwen2_5_omni.py:TimeDelayNetBlock.forward": ["ModelDelayNetBlock", "Module", "Tensor", "activation", "class", "conv", "def", "forward", "hidden_states", "nn", "return", "self", "torch"], "qwen2_5_omni/modeling_qwen2_5_omni.py:Res2NetBlock.__init__": ["ModelNetBlock", "Module", "ModuleList", "TimeDelayNetBlock", "__init__", "blocks", "class", "def", "dilation", "for", "hidden_channel", "i", "in", "in_channel", "in_channels", "kernel_size", "nn", "out_channels", "range", "scale", "self", "super", "torch"], "qwen2_5_omni/modeling_qwen2_5_omni.py:Res2NetBlock.forward": ["ModelNetBlock", "Module", "append", "blocks", "cat", "chunk", "class", "def", "dim", "elif", "else", "enumerate", "for", "forward", "hidden_part", "hidden_states", "i", "if", "in", "nn", "output", "output_part", "outputs", "return", "scale", "self", "torch"], "qwen2_5_omni/modeling_qwen2_5_omni.py:SqueezeExcitationBlock.__init__": ["Conv1d", "ModelExcitationBlock", "Module", "ReLU", "Sigmoid", "True", "__init__", "class", "conv1", "conv2", "def", "in_channels", "inplace", "kernel_size", "nn", "out_channels", "padding", "padding_mode", "reflect", "relu", "same", "se_channels", "self", "sigmoid", "super"], "qwen2_5_omni/modeling_qwen2_5_omni.py:SqueezeExcitationBlock.forward": ["ModelExcitationBlock", "Module", "True", "class", "conv1", "conv2", "def", "dim", "forward", "hidden_states", "hidden_states_mean", "keepdim", "mean", "nn", "relu", "return", "self", "sigmoid"], "qwen2_5_omni/modeling_qwen2_5_omni.py:AttentiveStatisticsPooling.__init__": ["Conv1d", "ModelStatisticsPooling", "Module", "Tanh", "TimeDelayNetBlock", "__init__", "attention_channels", "channels", "class", "conv", "def", "eps", "in_channels", "kernel_size", "nn", "out_channels", "padding", "padding_mode", "reflect", "same", "self", "super", "tanh", "tdnn"], "qwen2_5_omni/modeling_qwen2_5_omni.py:AttentiveStatisticsPooling._length_to_mask": ["ModelStatisticsPooling", "Module", "None", "_length_to_mask", "arange", "as_tensor", "class", "def", "device", "dtype", "expand", "if", "is", "item", "len", "length", "long", "mask", "max", "max_len", "nn", "return", "self", "torch", "unsqueeze"], "qwen2_5_omni/modeling_qwen2_5_omni.py:AttentiveStatisticsPooling._compute_statistics": ["ModelStatisticsPooling", "Module", "_compute_statistics", "clamp", "class", "def", "dim", "eps", "m", "mean", "nn", "pow", "return", "self", "sqrt", "std", "sum", "torch", "unsqueeze", "x"], "qwen2_5_omni/modeling_qwen2_5_omni.py:AttentiveStatisticsPooling.forward": ["F", "ModelStatisticsPooling", "Module", "True", "_compute_statistics", "_length_to_mask", "attention", "cat", "class", "conv", "def", "device", "dim", "dtype", "float", "forward", "hidden_states", "inf", "keepdim", "lengths", "mask", "masked_fill", "max_len", "mean", "nn", "ones", "pooled_stats", "repeat", "return", "self", "seq_length", "shape", "softmax", "std", "sum", "tanh", "tdnn", "torch", "total", "unsqueeze"], "qwen2_5_omni/modeling_qwen2_5_omni.py:SqueezeExcitationRes2NetBlock.__init__": ["ModelExcitationBlock", "ModelExcitationRes2NetBlock", "Module", "Res2NetBlock", "TimeDelayNetBlock", "__init__", "class", "def", "dilation", "in_channels", "kernel_size", "nn", "out_channels", "res2net_block", "res2net_scale", "se_block", "se_channels", "self", "super", "tdnn1", "tdnn2"], "qwen2_5_omni/modeling_qwen2_5_omni.py:SqueezeExcitationRes2NetBlock.forward": ["ModelExcitationRes2NetBlock", "Module", "class", "def", "forward", "hidden_state", "nn", "res2net_block", "residual", "return", "se_block", "self", "tdnn1", "tdnn2"], "qwen2_5_omni/modeling_qwen2_5_omni.py:ECAPA_TimeDelayNet.__init__": ["AttentiveStatisticsPooling", "Conv1d", "Model_TimeDelayNet", "Module", "ModuleList", "Qwen2_5OmniDiTConfig", "SqueezeExcitationRes2NetBlock", "TimeDelayNetBlock", "ValueError", "__init__", "and", "append", "asp", "attention_channels", "blocks", "channels", "class", "config", "def", "dilation", "enc_attention_channels", "enc_channels", "enc_dilations", "enc_dim", "enc_kernel_sizes", "enc_res2net_scale", "enc_se_channels", "fc", "for", "have", "i", "if", "in", "in_channels", "kernel_size", "len", "length", "mel_dim", "mfa", "nn", "or", "out_channels", "padding", "padding_mode", "raise", "range", "reflect", "res2net_scale", "same", "se_channels", "self", "should", "super", "torch"], "qwen2_5_omni/modeling_qwen2_5_omni.py:ECAPA_TimeDelayNet.forward": ["Model_TimeDelayNet", "Module", "append", "asp", "blocks", "cat", "class", "def", "dim", "fc", "for", "forward", "hidden_states", "hidden_states_list", "in", "layer", "mfa", "nn", "return", "self", "squeeze", "torch", "transpose"], "qwen2_5_omni/modeling_qwen2_5_omni.py:DiTInputEmbedding.__init__": ["DiTInputEmbedding", "ECAPA_TimeDelayNet", "Linear", "Module", "Qwen2_5OmniDiTConfig", "__init__", "class", "config", "def", "emb_dim", "enc_dim", "enc_emb_dim", "hidden_size", "mel_dim", "nn", "proj", "self", "spk_encoder", "super"], "qwen2_5_omni/modeling_qwen2_5_omni.py:DiTInputEmbedding.forward": ["DiTInputEmbedding", "False", "Module", "None", "Tensor", "True", "apply_cfg", "bool", "cat", "class", "code_embed", "code_embed_uncond", "condition_vector", "def", "dim", "drop_audio_cond", "elif", "forward", "hidden_states", "if", "nn", "proj", "repeat", "return", "self", "size", "speaker_embedding", "spk_encoder", "torch", "unsqueeze", "zeros_like"], "qwen2_5_omni/modeling_qwen2_5_omni.py:DiTCodecEmbedding.__init__": ["DiTCodecEmbedding", "Embedding", "Module", "__init__", "class", "codec_dim", "codec_embed", "codec_num_embeds", "def", "nn", "repeats", "self", "super"], "qwen2_5_omni/modeling_qwen2_5_omni.py:DiTCodecEmbedding.forward": ["DiTCodecEmbedding", "False", "Module", "class", "code", "code_embed", "codec_embed", "def", "dim", "drop_code", "forward", "if", "nn", "repeat_interleave", "repeats", "return", "self", "torch", "zeros_like"], "qwen2_5_omni/modeling_qwen2_5_omni.py:Qwen2_5_OmniAdaLayerNormZero.__init__": ["False", "LayerNorm", "Linear", "ModelAdaLayerNormZero", "Module", "SiLU", "__init__", "class", "def", "dim", "elementwise_affine", "eps", "linear", "nn", "norm", "self", "silu", "super"], "qwen2_5_omni/modeling_qwen2_5_omni.py:Qwen2_5_OmniAdaLayerNormZero.forward": ["ModelAdaLayerNormZero", "Module", "None", "chunk", "class", "def", "dim", "emb", "forward", "gate_mlp", "gate_msa", "hidden_states", "linear", "nn", "norm", "return", "scale_mlp", "scale_msa", "self", "shift_mlp", "shift_msa", "silu", "torch"], "qwen2_5_omni/modeling_qwen2_5_omni.py:Qwen2_5_OmniAdaLayerNormZero_Final.__init__": ["False", "LayerNorm", "Linear", "ModelAdaLayerNormZero_Final", "Module", "SiLU", "__init__", "class", "def", "dim", "elementwise_affine", "eps", "linear", "nn", "norm", "self", "silu", "super"], "qwen2_5_omni/modeling_qwen2_5_omni.py:Qwen2_5_OmniAdaLayerNormZero_Final.forward": ["ModelAdaLayerNormZero_Final", "Module", "None", "chunk", "class", "def", "dim", "emb", "forward", "hidden_states", "linear", "nn", "norm", "return", "scale", "self", "shift", "silu", "torch"], "qwen2_5_omni/modeling_qwen2_5_omni.py:DiTMLP.__init__": ["DiTMLP", "Dropout", "GELU", "Linear", "Module", "ModuleList", "__init__", "approximate", "class", "def", "dim", "dropout", "ff", "inner_dim", "int", "mult", "nn", "self", "super", "tanh"], "qwen2_5_omni/modeling_qwen2_5_omni.py:DiTMLP.forward": ["DiTMLP", "Module", "class", "def", "ff", "for", "forward", "hidden_states", "in", "layer", "nn", "return", "self"], "qwen2_5_omni/modeling_qwen2_5_omni.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cos", "def", "dim", "k", "k_embed", "q", "q_embed", "reshape", "return", "rotate_half_codec", "shape", "sin", "stack", "torch", "unbind", "unsqueeze", "unsqueeze_dim", "x", "x1", "x2"], "qwen2_5_omni/modeling_qwen2_5_omni.py:DiTAttention.__init__": ["DiTAttention", "Dropout", "False", "Linear", "Module", "ModuleList", "Qwen2_5OmniDiTConfig", "__init__", "class", "config", "def", "dim", "dropout", "head_dim", "heads", "hidden_size", "inner_dim", "is_causal", "nn", "num_attention_heads", "self", "super", "to_k", "to_out", "to_q", "to_v"], "qwen2_5_omni/modeling_qwen2_5_omni.py:DiTAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "DiTAttention", "False", "Module", "None", "Tensor", "_", "_attn_implementation", "apply_rotary_pos_emb", "attention_interface", "attention_mask", "attention_output", "attention_weights", "batch_size", "class", "config", "cos", "def", "dtype", "forward", "head_dim", "heads", "hidden_states", "inner_dim", "is_causal", "key", "nn", "position_embeddings", "query", "reshape", "return", "self", "shape", "sin", "to", "to_k", "to_out", "to_q", "to_v", "torch", "transpose", "value", "view"], "qwen2_5_omni/modeling_qwen2_5_omni.py:SinusPositionEmbedding.__init__": ["ModelPositionEmbedding", "Module", "__init__", "class", "def", "dim", "nn", "self", "super"], "qwen2_5_omni/modeling_qwen2_5_omni.py:SinusPositionEmbedding.forward": ["ModelPositionEmbedding", "Module", "arange", "cat", "class", "cos", "def", "device", "dim", "emb", "exp", "float", "forward", "half_dim", "hidden_states", "log", "math", "nn", "return", "scale", "self", "sin", "torch", "type_as", "unsqueeze"], "qwen2_5_omni/modeling_qwen2_5_omni.py:DiTTimestepEmbedding.__init__": ["DiTTimestepEmbedding", "Linear", "Module", "ModuleList", "SiLU", "SinusPositionEmbedding", "__init__", "class", "def", "dim", "freq_embed_dim", "nn", "self", "super", "time_embed", "time_mlp"], "qwen2_5_omni/modeling_qwen2_5_omni.py:DiTTimestepEmbedding.forward": ["DiTTimestepEmbedding", "Module", "class", "def", "dtype", "for", "forward", "in", "layer", "nn", "return", "self", "time_embed", "time_hidden", "time_mlp", "timestep", "to"], "qwen2_5_omni/modeling_qwen2_5_omni.py:DiTDecoderLayer.__init__": ["DiTAttention", "DiTDecoderLayer", "DiTMLP", "False", "LayerNorm", "ModelAdaLayerNormZero", "Module", "Qwen2_5OmniDiTConfig", "__init__", "attn", "attn_norm", "class", "config", "def", "dim", "dropout", "elementwise_affine", "eps", "ff", "ff_mult", "ff_norm", "hidden_size", "look_ahead_block", "look_backward_block", "mult", "nn", "self", "super"], "qwen2_5_omni/modeling_qwen2_5_omni.py:DiTDecoderLayer.forward": ["DiTDecoderLayer", "Module", "None", "attention_mask", "attn", "attn_norm", "attn_output", "block_diff", "class", "def", "emb", "ff", "ff_norm", "ff_output", "float", "forward", "gate_mlp", "gate_msa", "hidden_states", "look_ahead_block", "look_backward_block", "nn", "norm", "position_embeddings", "return", "scale_mlp", "self", "shift_mlp", "timestep", "unsqueeze"], "qwen2_5_omni/modeling_qwen2_5_omni.py:SnakeBeta.__init__": ["ModelBeta", "Module", "Parameter", "__init__", "alpha", "beta", "class", "def", "in_features", "nn", "no_div_by_zero", "self", "super", "torch", "zeros"], "qwen2_5_omni/modeling_qwen2_5_omni.py:SnakeBeta.forward": ["ModelBeta", "Module", "alpha", "beta", "class", "def", "exp", "forward", "hidden_states", "nn", "no_div_by_zero", "pow", "return", "self", "sin", "torch", "unsqueeze"], "qwen2_5_omni/modeling_qwen2_5_omni.py:UpSample1d.__init__": ["False", "Module", "None", "UpSample1d", "__init__", "class", "cutoff", "def", "else", "filter", "half_width", "if", "int", "is", "kaiser_sinc_filter1d", "kernel_size", "nn", "pad", "pad_left", "pad_right", "persistent", "ratio", "register_buffer", "self", "stride", "super"], "qwen2_5_omni/modeling_qwen2_5_omni.py:UpSample1d.forward": ["F", "Module", "UpSample1d", "channels", "class", "conv_transpose1d", "def", "expand", "filter", "forward", "groups", "hidden_states", "mode", "nn", "pad", "pad_left", "pad_right", "ratio", "replicate", "return", "self", "shape", "stride"], "qwen2_5_omni/modeling_qwen2_5_omni.py:DownSample1d.__init__": ["A", "False", "Minimum", "ModelSample1d", "Module", "None", "ValueError", "__init__", "above", "be", "class", "cutoff", "def", "does", "even", "filter", "half_width", "if", "int", "kaiser_sinc_filter1d", "kernel_size", "larger", "make", "must", "nn", "not", "pad_left", "pad_right", "persistent", "raise", "ratio", "register_buffer", "self", "sense", "stride", "super", "than", "zero"], "qwen2_5_omni/modeling_qwen2_5_omni.py:DownSample1d.forward": ["F", "ModelSample1d", "Module", "channels", "class", "conv1d", "def", "expand", "filter", "forward", "groups", "hidden_states", "mode", "nn", "out", "pad", "pad_left", "pad_right", "replicate", "return", "self", "shape", "stride"], "qwen2_5_omni/modeling_qwen2_5_omni.py:TorchActivation1d.__init__": ["Activation", "DownSample1d", "ModelActivation1d", "Module", "TypeError", "UpSample1d", "__init__", "act", "activation", "be", "callable", "class", "def", "down_kernel_size", "down_ratio", "downsample", "function", "if", "int", "must", "nn", "not", "raise", "self", "super", "up_kernel_size", "up_ratio", "upsample"], "qwen2_5_omni/modeling_qwen2_5_omni.py:TorchActivation1d.forward": ["ModelActivation1d", "Module", "act", "class", "def", "downsample", "forward", "hidden_states", "nn", "return", "self", "upsample"], "qwen2_5_omni/modeling_qwen2_5_omni.py:AMPBlock.__init__": ["Conv1d", "Model", "Module", "ModuleList", "SnakeBeta", "TorchActivation1d", "_", "__init__", "_get_padding", "activation", "activations", "channels", "class", "convs1", "convs2", "def", "dilation", "for", "in", "kernel_size", "len", "nn", "num_layers", "padding", "range", "self", "super", "torch"], "qwen2_5_omni/modeling_qwen2_5_omni.py:AMPBlock._get_padding": ["Model", "Module", "_get_padding", "class", "def", "dilation", "int", "kernel_size", "nn", "return", "self", "torch"], "qwen2_5_omni/modeling_qwen2_5_omni.py:AMPBlock.forward": ["Model", "Module", "act1", "act2", "activations", "acts1", "acts2", "class", "conv1", "conv2", "convs1", "convs2", "def", "for", "forward", "hidden_states", "in", "nn", "residual", "return", "self", "torch", "zip"], "qwen2_5_omni/modeling_qwen2_5_omni.py:Qwen2_5OmniToken2WavBigVGANModel.__init__": ["AMPBlock", "Conv1d", "ConvTranspose1d", "False", "Model_5OmniBigVGANConfig", "Model_5OmniPreTrainedModel", "Model_5OmniToken2WavBigVGANModel", "ModuleList", "SnakeBeta", "TorchActivation1d", "__init__", "activation", "activation_post", "bias", "class", "config", "conv_post", "conv_pre", "def", "dilation", "enumerate", "for", "in", "kernel_size", "layer_idx", "len", "mel_dim", "nn", "num_residual_blocks", "num_upsample_layers", "padding", "post_init", "range", "resblock_dilation_sizes", "resblock_kernel_sizes", "resblocks", "self", "stride", "super", "ups", "upsample_initial_channel", "upsample_kernel_sizes", "upsample_rates", "zip"], "qwen2_5_omni/modeling_qwen2_5_omni.py:Qwen2_5OmniToken2WavBigVGANModel.normalize_spectrogram": ["Model_5OmniPreTrainedModel", "Model_5OmniToken2WavBigVGANModel", "clamp", "class", "def", "max_value", "min_db", "normalize_spectrogram", "return", "self", "spectrogram", "torch"], "qwen2_5_omni/modeling_qwen2_5_omni.py:Qwen2_5OmniToken2WavBigVGANModel.amplitude_to_db": ["Model_5OmniPreTrainedModel", "Model_5OmniToken2WavBigVGANModel", "amplitude", "amplitude_to_db", "clamp", "class", "def", "device", "dtype", "exp", "log", "log10", "min", "min_db_level", "min_level", "np", "return", "self", "tensor", "torch"], "qwen2_5_omni/modeling_qwen2_5_omni.py:Qwen2_5OmniToken2WavBigVGANModel.process_mel_spectrogram": ["Model_5OmniPreTrainedModel", "Model_5OmniToken2WavBigVGANModel", "amplitude_spectrum", "amplitude_to_db", "class", "decibel_spectrum", "def", "exp", "mel_spectrogram", "normalize_spectrogram", "process_mel_spectrogram", "return", "self", "torch"], "qwen2_5_omni/modeling_qwen2_5_omni.py:Qwen2_5OmniToken2WavBigVGANModel.forward": ["Model_5OmniPreTrainedModel", "Model_5OmniToken2WavBigVGANModel", "activation_post", "block_index", "clamp", "class", "conv_post", "conv_pre", "cpu", "def", "for", "forward", "hidden_representation", "in", "kwargs", "layer_index", "max", "mel_spectrogram", "min", "num_residual_blocks", "num_upsample_layers", "output_waveform", "process_mel_spectrogram", "processed_spectrogram", "range", "resblocks", "residual_output", "return", "self", "squeeze", "sum", "torch", "ups"], "qwen2_5_omni/modeling_qwen2_5_omni.py:RungeKutta4ODESolver.__init__": ["ModelKutta4ODESolver", "__init__", "_one_third", "_two_thirds", "class", "def", "function", "initial_value", "self"], "qwen2_5_omni/modeling_qwen2_5_omni.py:RungeKutta4ODESolver._rk4_step": ["ModelKutta4ODESolver", "None", "_one_third", "_rk4_step", "_two_thirds", "class", "def", "else", "function", "function_value_start", "if", "is", "k1", "k2", "k3", "k4", "not", "return", "self", "time_end", "time_start", "time_step", "value_start"], "qwen2_5_omni/modeling_qwen2_5_omni.py:RungeKutta4ODESolver._compute_step": ["ModelKutta4ODESolver", "_compute_step", "_rk4_step", "class", "def", "function", "function_value_start", "return", "self", "time_end", "time_start", "time_step", "value_start"], "qwen2_5_omni/modeling_qwen2_5_omni.py:RungeKutta4ODESolver._linear_interpolation": ["ModelKutta4ODESolver", "_linear_interpolation", "class", "def", "if", "return", "self", "time_end", "time_point", "time_start", "value_end", "value_start", "weight"], "qwen2_5_omni/modeling_qwen2_5_omni.py:RungeKutta4ODESolver.integrate": ["ModelKutta4ODESolver", "_", "_compute_step", "_linear_interpolation", "and", "class", "current_index", "current_value", "def", "delta_value", "device", "dtype", "empty", "for", "function", "in", "initial_value", "integrate", "len", "next_value", "return", "self", "shape", "solution", "time_end", "time_points", "time_start", "time_step", "torch", "while", "zip"], "qwen2_5_omni/modeling_qwen2_5_omni.py:Qwen2_5OmniToken2WavDiTModel.__init__": ["DiTCodecEmbedding", "DiTDecoderLayer", "DiTInputEmbedding", "DiTTimestepEmbedding", "Linear", "ModelAdaLayerNormZero_Final", "Model_5OmniDiTConfig", "Model_5OmniDiTRotaryEmbedding", "Model_5OmniPreTrainedModel", "Model_5OmniToken2WavDiTModel", "ModuleList", "__init__", "append", "block_size", "class", "config", "def", "else", "emb_dim", "for", "hidden_size", "i", "if", "in", "input_embed", "layers", "look_ahead_block", "look_ahead_layers", "look_backward_block", "look_backward_layers", "mel_dim", "nn", "norm_out", "num_attention_heads", "num_embeds", "num_hidden_layers", "post_init", "proj_out", "range", "repeats", "rotary_embed", "self", "super", "text_embed", "time_embed", "transformer_blocks"], "qwen2_5_omni/modeling_qwen2_5_omni.py:Qwen2_5OmniToken2WavDiTModel._create_block_diff": ["Model_5OmniPreTrainedModel", "Model_5OmniToken2WavDiTModel", "_create_block_diff", "arange", "batch", "block_diff", "block_i", "block_indices", "block_j", "block_size", "class", "def", "device", "expand", "hidden_states", "num_attention_heads", "return", "self", "seq_len", "shape", "torch", "unsqueeze"], "qwen2_5_omni/modeling_qwen2_5_omni.py:Qwen2_5OmniToken2WavDiTModel.forward": ["False", "Model_5OmniPreTrainedModel", "Model_5OmniToken2WavDiTModel", "None", "True", "_create_block_diff", "apply_cfg", "arange", "batch_size", "block_diff", "blockwise_difference", "class", "code_embed_uncond", "condition_vector", "def", "device", "drop_audio_cond", "drop_audio_conditioning", "drop_code", "else", "for", "forward", "hidden_states", "if", "in", "input_embed", "kwargs", "ndim", "norm_out", "output", "position_embeddings", "position_ids", "proj_out", "quantized_code", "repeat", "return", "rotary_embed", "self", "shape", "speaker_embedding", "text_embed", "text_embedding", "text_embedding_unconditioned", "time_embed", "time_embedding", "time_step", "torch", "transformer_block", "transformer_blocks"], "qwen2_5_omni/modeling_qwen2_5_omni.py:Qwen2_5OmniToken2WavDiTModel.sample": ["False", "Model_5OmniPreTrainedModel", "Model_5OmniToken2WavDiTModel", "None", "Only", "Provide", "Requested", "RungeKutta4ODESolver", "True", "ValueError", "apply_cfg", "batch", "batch_size", "chunk", "class", "condition_vector", "conditioning_vector", "config", "cos", "currently", "def", "device", "dim", "dit_config", "drop_audio_conditioning", "drop_code", "dtype", "exceeds", "f", "function", "generated_mel_spectrogram", "generated_waveform", "guidance_scale", "guided_prediction", "hidden_states", "if", "initial_state", "initial_time", "initial_value", "integrate", "is", "length", "linspace", "max_position_embeddings", "maximum_duration", "mel", "mel_dim", "model_output", "not", "null_prediction", "num_steps", "ode_function", "ode_solver", "permute", "pi", "prediction", "quantized_code", "raise", "randn", "reference_mel_spectrogram", "repeat", "repeats", "return", "sample", "self", "shape", "shorter", "size", "solution_trajectory", "speaker_embedding", "supported", "sway_coefficient", "time_embedding", "time_step", "torch", "unsqueeze"], "qwen2_5_omni/modeling_qwen2_5_omni.py:Qwen2_5OmniToken2WavModel.__init__": ["Model_5OmniPreTrainedModel", "Model_5OmniToken2WavBigVGANModel", "Model_5OmniToken2WavConfig", "Model_5OmniToken2WavDiTModel", "Model_5OmniToken2WavModel", "__init__", "_attn_implementation", "_from_config", "and", "attention", "attn_impl", "attn_implementation", "back", "bf16", "bigvgan_config", "but", "class", "code2wav_bigvgan_model", "code2wav_dit_model", "config", "def", "dit_config", "does", "eager", "elif", "fall", "fallback", "flash_attention_2", "fp16", "fp32", "if", "implementation", "inference", "logger", "must", "not", "of", "only", "post_init", "sdpa", "self", "super", "support", "supports", "to", "warning_once", "will", "with"], "qwen2_5_omni/modeling_qwen2_5_omni.py:Qwen2_5OmniToken2WavModel.forward": ["Model_5OmniPreTrainedModel", "Model_5OmniToken2WavModel", "class", "code", "code2wav_bigvgan_model", "code2wav_dit_model", "conditioning", "def", "forward", "guidance_scale", "kwargs", "mel_spectrogram", "num_steps", "reference_mel", "return", "sample", "self", "sway_coefficient", "waveform"], "qwen2_5_omni/modeling_qwen2_5_omni.py:Qwen2_5OmniForConditionalGeneration.__init__": ["GenerationMixin", "Model_5OmniForConditionalGeneration", "Model_5OmniPreTrainedModel", "Model_5OmniThinkerForConditionalGeneration", "__init__", "class", "config", "def", "enable_audio_output", "enable_talker", "has_talker", "if", "post_init", "self", "speaker_map", "super", "thinker", "thinker_config"], "qwen2_5_omni/modeling_qwen2_5_omni.py:Qwen2_5OmniForConditionalGeneration.enable_talker": ["GenerationMixin", "Model_5OmniForConditionalGeneration", "Model_5OmniPreTrainedModel", "Model_5OmniTalkerForConditionalGeneration", "Model_5OmniToken2WavModel", "True", "class", "config", "def", "enable_talker", "float", "has_talker", "self", "talker", "talker_config", "token2wav", "token2wav_config"], "qwen2_5_omni/modeling_qwen2_5_omni.py:Qwen2_5OmniForConditionalGeneration.load_speakers": ["GenerationMixin", "Model_5OmniForConditionalGeneration", "Model_5OmniPreTrainedModel", "Speaker", "True", "check_torch_load_is_safe", "class", "def", "f", "for", "in", "info", "items", "key", "keys", "list", "load", "load_speakers", "loaded", "logger", "path", "self", "speaker_map", "torch", "value", "weights_only"], "qwen2_5_omni/modeling_qwen2_5_omni.py:Qwen2_5OmniForConditionalGeneration.disable_talker": ["False", "GenerationMixin", "Model_5OmniForConditionalGeneration", "Model_5OmniPreTrainedModel", "class", "def", "del", "disable_talker", "has_talker", "hasattr", "if", "self", "talker", "token2wav"], "qwen2_5_omni/modeling_qwen2_5_omni.py:Qwen2_5OmniForConditionalGeneration.from_pretrained": ["False", "GenerationMixin", "Model_5OmniForConditionalGeneration", "Model_5OmniPreTrainedModel", "None", "True", "ValueError", "cache_dir", "cached_file", "class", "cls", "config", "def", "f", "force_download", "from_pretrained", "if", "ignore_mismatched_sizes", "is", "kwargs", "load_speakers", "local_files_only", "main", "model", "model_args", "pop", "pretrained_model_name_or_path", "proxies", "pt", "raise", "return", "revision", "spk_dict", "spk_path", "subfolder", "super", "token", "use_safetensors", "weights_only"], "qwen2_5_omni/modeling_qwen2_5_omni.py:Qwen2_5OmniForConditionalGeneration.generate": ["Cannot", "Chelsie", "False", "GenerationMixin", "Model", "Model_5OmniForConditionalGeneration", "Model_5OmniPreTrainedModel", "None", "NotImplementedError", "Omni", "Tensor", "True", "Use", "ValueError", "and", "attention_mask", "audio", "audio_feature_lengths", "audio_ids_mask", "audio_mask", "audio_mask_tensor", "audio_token_index", "available", "batched", "bool", "bos_token", "cat", "class", "clone", "codec_bos_token", "codec_mask_token", "codec_pad_token", "cond", "conditioning", "config", "currently", "def", "device", "dim", "do_sample", "does", "dtype", "elif", "else", "embeds_to_talker", "enable", "enable_talker", "eos_embedding", "eos_token", "eos_token_id", "expand_as", "f", "feature_attention_mask", "fill_value", "float", "for", "full_like", "generate", "generate_audio", "generation_mode", "get", "get_input_embeddings", "has_talker", "hidden_states", "if", "image_ids_mask", "image_mask", "image_mask_tensor", "image_token_index", "in", "inference", "initialized", "input_features", "input_ids", "input_text_ids", "inputs_embeds", "int", "is", "is_tensor", "items", "k", "key", "keys", "kwargs", "len", "list", "long", "masked_scatter_", "max_new_tokens", "method", "module", "new_ones", "not", "or", "output", "output_hidden_states", "pad_embedding", "pad_token", "pixel_values", "pixel_values_videos", "pop", "processed_thinker_hidden", "r", "raise", "ref_mel", "reference_mel", "repetition_penalty", "return", "return_audio", "return_dict_in_generate", "self", "sequences", "set", "shape", "shared_kwargs", "size", "speaker", "speaker_map", "speaker_params", "speakers", "startswith", "str", "sum", "support", "suppress_tokens", "talker", "talker_", "talker_attention_mask", "talker_do_sample", "talker_eos_token_id", "talker_generate_codes", "talker_input_ids", "talker_input_text_ids", "talker_inputs_embeds", "talker_kwargs", "talker_max_new_tokens", "talker_repetition_penalty", "talker_result", "talker_temperature", "talker_text_bos_embed", "talker_text_bos_token", "talker_top_k", "talker_top_p", "temperature", "tensor", "text", "text_eos_token", "text_pad_token", "thinker", "thinker_", "thinker_config", "thinker_embed_tokens", "thinker_generate_ids", "thinker_hidden_states", "thinker_kwargs", "thinker_max_new_tokens", "thinker_reply_part", "thinker_result", "thinker_token_embeds", "to", "token2wav", "token2wav_", "token2wav_kwargs", "token_hidden_states", "top_k", "top_p", "torch", "unsqueeze", "use", "use_audio_in_video", "v", "value", "video_ids_mask", "video_mask", "video_mask_tensor", "video_token_index", "wav", "when", "with", "zeros"], "qwen2_5_vl/modeling_qwen2_5_vl.py:Qwen2_5_VLMLP.__init__": ["ACT2FN", "False", "Linear", "ModelMLP", "Module", "__init__", "act_fn", "bias", "bool", "class", "config", "def", "down_proj", "gate_proj", "hidden_act", "hidden_size", "intermediate_size", "nn", "self", "super", "up_proj"], "qwen2_5_vl/modeling_qwen2_5_vl.py:Qwen2_5_VLMLP.forward": ["ModelMLP", "Module", "act_fn", "class", "def", "down_proj", "forward", "gate_proj", "hidden_state", "nn", "return", "self", "up_proj"], "qwen2_5_vl/modeling_qwen2_5_vl.py:Qwen2_5_VisionPatchEmbed.__init__": ["Conv3d", "False", "Model_5_VisionPatchEmbed", "Module", "None", "__init__", "bias", "class", "def", "embed_dim", "in_channels", "int", "kernel_size", "nn", "patch_size", "proj", "self", "stride", "super", "temporal_patch_size"], "qwen2_5_vl/modeling_qwen2_5_vl.py:Qwen2_5_VisionPatchEmbed.forward": ["Model_5_VisionPatchEmbed", "Module", "Tensor", "class", "def", "dtype", "embed_dim", "forward", "hidden_states", "in_channels", "nn", "patch_size", "proj", "return", "self", "target_dtype", "temporal_patch_size", "to", "torch", "view", "weight"], "qwen2_5_vl/modeling_qwen2_5_vl.py:Qwen2_5_VisionRotaryEmbedding.__init__": ["False", "Model_5_VisionRotaryEmbedding", "Module", "None", "__init__", "arange", "class", "def", "dim", "dtype", "float", "int", "inv_freq", "nn", "persistent", "register_buffer", "self", "super", "theta", "torch"], "qwen2_5_vl/modeling_qwen2_5_vl.py:Qwen2_5_VisionRotaryEmbedding.forward": ["Model_5_VisionRotaryEmbedding", "Module", "Tensor", "arange", "class", "def", "device", "dtype", "forward", "freqs", "int", "inv_freq", "nn", "outer", "return", "self", "seq", "seqlen", "torch"], "qwen2_5_vl/modeling_qwen2_5_vl.py:Qwen2_5_VLPatchMerger.__init__": ["GELU", "Linear", "ModelPatchMerger", "ModelRMSNorm", "Module", "None", "Sequential", "__init__", "class", "context_dim", "def", "dim", "eps", "hidden_size", "int", "ln_q", "mlp", "nn", "self", "spatial_merge_size", "super"], "qwen2_5_vl/modeling_qwen2_5_vl.py:Qwen2_5_VLPatchMerger.forward": ["ModelPatchMerger", "Module", "Tensor", "class", "def", "forward", "hidden_size", "ln_q", "mlp", "nn", "return", "self", "torch", "view", "x"], "qwen2_5_vl/modeling_qwen2_5_vl.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "qwen2_5_vl/modeling_qwen2_5_vl.py:apply_rotary_pos_emb_vision": ["Model_rotary_pos_emb_vision", "Tensor", "cos", "def", "dtype", "float", "k", "k_embed", "orig_k_dtype", "orig_q_dtype", "q", "q_embed", "return", "rotate_half", "sin", "to", "torch", "tuple", "unsqueeze"], "qwen2_5_vl/modeling_qwen2_5_vl.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "qwen2_5_vl/modeling_qwen2_5_vl.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value", "value_states"], "qwen2_5_vl/modeling_qwen2_5_vl.py:Qwen2_5_VLVisionAttention.__init__": ["False", "Linear", "ModelVisionAttention", "ModelVisionConfig", "Module", "None", "True", "__init__", "attention_dropout", "bias", "class", "config", "def", "dim", "head_dim", "hidden_size", "is_causal", "nn", "num_heads", "num_key_value_groups", "proj", "qkv", "scaling", "self", "super"], "qwen2_5_vl/modeling_qwen2_5_vl.py:Qwen2_5_VLVisionAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "False", "ModelVisionAttention", "Module", "None", "Tensor", "_", "_attn_implementation", "apply_rotary_pos_emb_vision", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_outputs", "cat", "class", "config", "contiguous", "cos", "cu_seq_lens_k", "cu_seq_lens_q", "cu_seqlens", "def", "dim", "dropout", "eager", "eager_attention_forward", "else", "flash", "for", "forward", "hidden_states", "if", "in", "is_causal", "k", "key_states", "kwargs", "lengths", "max", "max_length_k", "max_length_q", "max_seqlen", "nn", "not", "num_heads", "permute", "position_embeddings", "proj", "q", "qkv", "query_states", "reshape", "return", "rotary_pos_emb", "scaling", "self", "seq_length", "shape", "sin", "split", "splits", "tensor", "tolist", "torch", "training", "transpose", "tuple", "unbind", "unsqueeze", "v", "value_states", "zip"], "qwen2_5_vl/modeling_qwen2_5_vl.py:Qwen2_5_VLVisionBlock.__init__": ["GradientCheckpointingLayer", "ModelMLP", "ModelRMSNorm", "ModelVisionAttention", "ModelVisionBlock", "None", "True", "__init__", "attn", "attn_implementation", "bias", "class", "config", "def", "eps", "hidden_size", "mlp", "norm1", "norm2", "sdpa", "self", "str", "super"], "qwen2_5_vl/modeling_qwen2_5_vl.py:Qwen2_5_VLVisionBlock.forward": ["GradientCheckpointingLayer", "ModelVisionBlock", "None", "Tensor", "attn", "class", "cu_seqlens", "def", "forward", "hidden_states", "kwargs", "mlp", "norm1", "norm2", "position_embeddings", "return", "rotary_pos_emb", "self", "torch", "tuple"], "qwen2_5_vl/modeling_qwen2_5_vl.py:Qwen2_5_VLPreTrainedModel._init_weights": ["ModelPreTrainedModel", "Model_5_VisionRotaryEmbedding", "PreTrainedModel", "_init_weights", "arange", "class", "copy_", "def", "dim", "dtype", "float", "if", "init", "inv_freq", "isinstance", "module", "self", "super", "theta", "torch"], "qwen2_5_vl/modeling_qwen2_5_vl.py:Qwen2_5_VisionTransformerPretrainedModel.__init__": ["False", "ModelPatchMerger", "ModelPreTrainedModel", "ModelVisionBlock", "Model_5_VisionPatchEmbed", "Model_5_VisionRotaryEmbedding", "Model_5_VisionTransformerPretrainedModel", "ModuleList", "None", "_", "__init__", "blocks", "class", "config", "context_dim", "def", "depth", "dim", "embed_dim", "for", "fullatt_block_indexes", "gradient_checkpointing", "head_dim", "hidden_size", "in", "in_channels", "inputs", "kwargs", "merger", "nn", "num_heads", "out_hidden_size", "patch_embed", "patch_size", "post_init", "range", "rotary_pos_emb", "self", "spatial_merge_size", "spatial_merge_unit", "super", "temporal_patch_size", "window_size"], "qwen2_5_vl/modeling_qwen2_5_vl.py:Qwen2_5_VisionTransformerPretrainedModel.rot_pos_emb": ["ModelPreTrainedModel", "Model_5_VisionTransformerPretrainedModel", "append", "arange", "cat", "class", "def", "dim", "expand", "flatten", "for", "grid_thw", "h", "hpos_ids", "in", "max", "max_grid_size", "permute", "pos_ids", "repeat", "reshape", "return", "rot_pos_emb", "rotary_pos_emb", "rotary_pos_emb_full", "self", "spatial_merge_size", "stack", "t", "torch", "unsqueeze", "w", "wpos_ids"], "qwen2_5_vl/modeling_qwen2_5_vl.py:Qwen2_5_VisionTransformerPretrainedModel.get_window_index": ["F", "ModelPreTrainedModel", "Model_5_VisionTransformerPretrainedModel", "append", "arange", "cat", "class", "constant", "cu_seqlens_tmp", "cu_window_seqlens", "cumsum", "def", "dim", "extend", "for", "get_window_index", "grid_h", "grid_t", "grid_thw", "grid_w", "in", "index", "index_new", "index_padded", "item", "list", "llm_grid_h", "llm_grid_w", "num_windows_h", "num_windows_w", "pad", "pad_h", "pad_w", "patch_size", "permute", "reshape", "return", "self", "seqlens", "spatial_merge_size", "spatial_merge_unit", "sum", "tolist", "torch", "vit_merger_window_size", "window_index", "window_index_id", "window_size"], "qwen2_5_vl/modeling_qwen2_5_vl.py:Qwen2_5_VisionTransformerPretrainedModel.forward": ["F", "ModelPreTrainedModel", "Model_5_VisionTransformerPretrainedModel", "Tensor", "_", "argsort", "blk", "blocks", "cat", "class", "cos", "cu_seqlens", "cu_seqlens_now", "cu_window_seqlens", "cumsum", "def", "device", "dim", "dtype", "else", "emb", "enumerate", "for", "forward", "fullatt_block_indexes", "get_window_index", "grid_thw", "hidden_states", "if", "in", "int32", "is_tracing", "jit", "kwargs", "layer_num", "merger", "pad", "patch_embed", "position_embeddings", "repeat_interleave", "reshape", "return", "reverse_indices", "rot_pos_emb", "rotary_pos_emb", "self", "seq_len", "sin", "size", "spatial_merge_unit", "tensor", "torch", "unique_consecutive", "value", "window_index"], "qwen2_5_vl/modeling_qwen2_5_vl.py:Qwen2_5_VLRotaryEmbedding.__init__": ["Callable", "False", "ModelConfig", "ModelRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "qwen2_5_vl/modeling_qwen2_5_vl.py:Qwen2_5_VLRotaryEmbedding.compute_default_rope_parameters": ["ModelConfig", "ModelRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "qwen2_5_vl/modeling_qwen2_5_vl.py:Qwen2_5_VLRotaryEmbedding.forward": ["False", "ModelRotaryEmbedding", "Module", "None", "and", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "qwen2_5_vl/modeling_qwen2_5_vl.py:Qwen2MLP.__init__": ["ACT2FN", "False", "Linear", "ModelMLP", "Module", "__init__", "act_fn", "bias", "class", "config", "def", "down_proj", "gate_proj", "hidden_act", "hidden_size", "intermediate_size", "nn", "self", "super", "up_proj"], "qwen2_5_vl/modeling_qwen2_5_vl.py:Qwen2MLP.forward": ["ModelMLP", "Module", "act_fn", "class", "def", "down_proj", "forward", "gate_proj", "nn", "return", "self", "up_proj", "x"], "qwen2_5_vl/modeling_qwen2_5_vl.py:apply_multimodal_rotary_pos_emb": ["Model_multimodal_rotary_pos_emb", "cat", "cos", "def", "dim", "enumerate", "for", "i", "in", "k", "k_embed", "m", "mrope_section", "q", "q_embed", "return", "rotate_half", "sin", "split", "torch", "unsqueeze", "unsqueeze_dim"], "qwen2_5_vl/modeling_qwen2_5_vl.py:Qwen2_5_VLAttention.__init__": ["False", "Instantiating", "Linear", "ModelAttention", "ModelTextConfig", "Module", "None", "Please", "True", "ValueError", "__class__", "__init__", "__name__", "a", "and", "attention_dropout", "be", "bias", "by", "caching", "call", "class", "config", "creating", "def", "divisible", "during", "else", "errors", "f", "forward", "got", "hasattr", "head_dim", "hidden_size", "if", "int", "is", "is_causal", "k_proj", "layer_idx", "layer_type", "layer_types", "logger", "make", "must", "nn", "not", "num_attention_heads", "num_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "passing", "provide", "q_proj", "raise", "recommended", "rope_parameters", "scaling", "self", "sliding_attention", "sliding_window", "super", "sure", "the", "this", "to", "used", "v_proj", "warning_once", "when", "will", "without"], "qwen2_5_vl/modeling_qwen2_5_vl.py:Qwen2_5_VLAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "False", "FlashAttentionKwargs", "LongTensor", "ModelAttention", "Module", "None", "Tensor", "Unpack", "_", "_attn_implementation", "apply_multimodal_rotary_pos_emb", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "bool", "bsz", "cache_kwargs", "cache_position", "class", "config", "contiguous", "cos", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "head_dim", "hidden_states", "if", "is", "k_proj", "key_states", "kwargs", "layer_idx", "mrope_section", "nn", "not", "o_proj", "output_attentions", "past_key_values", "position_embeddings", "position_ids", "q_len", "q_proj", "query_states", "reshape", "return", "rope_parameters", "scaling", "self", "sin", "size", "sliding_window", "torch", "training", "transpose", "tuple", "update", "use_cache", "v_proj", "value_states", "view"], "qwen2_5_vl/modeling_qwen2_5_vl.py:Qwen2_5_VLDecoderLayer.__init__": ["Attention", "GradientCheckpointingLayer", "ModelAttention", "ModelDecoderLayer", "ModelMLP", "ModelRMSNorm", "ModelTextConfig", "Sliding", "Window", "__init__", "_attn_implementation", "and", "attention_type", "be", "but", "class", "config", "def", "enabled", "encountered", "eps", "f", "flash_attention_2", "for", "hidden_size", "if", "implemented", "input_layernorm", "int", "is", "layer_idx", "layer_types", "logger", "may", "mlp", "not", "post_attention_layernorm", "results", "rms_norm_eps", "self", "self_attn", "super", "unexpected", "use_sliding_window", "warning_once"], "qwen2_5_vl/modeling_qwen2_5_vl.py:Qwen2_5_VLDecoderLayer.forward": ["Cache", "False", "FlashAttentionKwargs", "FloatTensor", "GradientCheckpointingLayer", "LongTensor", "ModelDecoderLayer", "None", "Tensor", "Unpack", "attention_mask", "bool", "cache_position", "class", "def", "forward", "hidden_states", "if", "input_layernorm", "kwargs", "mlp", "output_attentions", "outputs", "past_key_values", "position_embeddings", "position_ids", "post_attention_layernorm", "residual", "return", "self", "self_attn", "self_attn_weights", "torch", "tuple", "use_cache"], "qwen2_5_vl/modeling_qwen2_5_vl.py:Qwen2_5_VLTextModel.__init__": ["Embedding", "False", "ModelDecoderLayer", "ModelPreTrainedModel", "ModelRMSNorm", "ModelRotaryEmbedding", "ModelTextConfig", "ModelTextModel", "ModuleList", "__init__", "_attn_implementation", "class", "config", "def", "embed_tokens", "eps", "for", "gradient_checkpointing", "has_sliding_layers", "hidden_size", "in", "layer_idx", "layer_types", "layers", "nn", "norm", "num_hidden_layers", "pad_token_id", "padding_idx", "post_init", "range", "rms_norm_eps", "rotary_emb", "self", "sliding_attention", "super", "vocab_size"], "qwen2_5_vl/modeling_qwen2_5_vl.py:Qwen2_5_VLTextModel.forward": ["BaseModelOutputWithPast", "Cache", "DynamicCache", "False", "FlashAttentionKwargs", "FloatTensor", "LongTensor", "ModelPreTrainedModel", "ModelTextModel", "None", "Setting", "Tensor", "True", "Unpack", "ValueError", "You", "all_hidden_states", "all_self_attns", "and", "arange", "attention_mask", "attention_type", "attentions", "bool", "cache_position", "causal_mask_mapping", "checkpointing", "class", "config", "create_causal_mask", "create_sliding_window_causal_mask", "decoder_layer", "def", "device", "dict", "elif", "else", "embed_tokens", "exactly", "expand", "for", "forward", "full_attention", "get_seq_length", "gradient", "gradient_checkpointing", "has_sliding_layers", "hidden_states", "if", "in", "incompatible", "input_embeds", "input_ids", "inputs_embeds", "is", "is_tracing", "isinstance", "jit", "kwargs", "last_hidden_state", "layer_outputs", "layers", "logger", "mask_kwargs", "must", "ndim", "norm", "not", "of", "one", "or", "output_attentions", "output_hidden_states", "past_key_values", "past_seen_tokens", "position_embeddings", "position_ids", "raise", "return", "return_dict", "rotary_emb", "self", "shape", "sliding_attention", "specify", "text_position_ids", "torch", "training", "tuple", "use_cache", "use_return_dict", "v", "view", "warning_once", "with"], "qwen2_5_vl/modeling_qwen2_5_vl.py:Qwen2_5_VLModel.__init__": ["ModelModel", "ModelPreTrainedModel", "ModelTextModel", "Model_5_VisionTransformerPretrainedModel", "None", "__init__", "_from_config", "class", "config", "def", "language_model", "post_init", "rope_deltas", "self", "super", "text_config", "vision_config", "visual"], "qwen2_5_vl/modeling_qwen2_5_vl.py:Qwen2_5_VLModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "language_model", "return", "self"], "qwen2_5_vl/modeling_qwen2_5_vl.py:Qwen2_5_VLModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "language_model", "self", "set_input_embeddings", "value"], "qwen2_5_vl/modeling_qwen2_5_vl.py:Qwen2_5_VLModel.get_rope_index": ["False", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "True", "_", "and", "append", "arange", "argwhere", "as_tensor", "attention_mask", "cat", "class", "config", "cumsum", "def", "device", "dim", "dtype", "ed", "ed_image", "ed_video", "else", "enumerate", "expand", "expanded_range", "flatten", "for", "get_rope_index", "h", "h_index", "i", "if", "image_grid_thw", "image_index", "image_nums", "image_token_id", "in", "index", "input_ids", "input_tokens", "is", "item", "keepdim", "len", "list", "llm_grid_h", "llm_grid_t", "llm_grid_w", "llm_pos_ids_list", "llm_positions", "long", "masked_fill_", "max", "max_position_ids", "mrope_position_deltas", "not", "ones", "or", "position_ids", "range", "range_tensor", "remain_images", "remain_videos", "reshape", "return", "second_per_grid_t", "second_per_grid_ts", "self", "shape", "spatial_merge_size", "squeeze", "st", "st_idx", "stack", "sum", "t", "t_index", "tensor", "text_len", "time_tensor", "time_tensor_long", "to", "tokens_per_second", "tolist", "torch", "total_input_ids", "tuple", "unsqueeze", "video_grid_thw", "video_index", "video_nums", "video_token_id", "view", "vision_config", "vision_start_indices", "vision_start_token_id", "vision_tokens", "w", "w_index", "zeros"], "qwen2_5_vl/modeling_qwen2_5_vl.py:Qwen2_5_VLModel.get_video_features": ["FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "class", "def", "dtype", "get_video_features", "grid_thw", "pixel_values_videos", "prod", "return", "self", "spatial_merge_size", "split", "split_sizes", "tolist", "torch", "type", "video_embeds", "video_grid_thw", "visual"], "qwen2_5_vl/modeling_qwen2_5_vl.py:Qwen2_5_VLModel.get_image_features": ["FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "class", "def", "dtype", "get_image_features", "grid_thw", "image_embeds", "image_grid_thw", "pixel_values", "prod", "return", "self", "spatial_merge_size", "split", "split_sizes", "tolist", "torch", "type", "visual"], "qwen2_5_vl/modeling_qwen2_5_vl.py:Qwen2_5_VLModel.get_placeholder_mask": ["FloatTensor", "Image", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "ValueError", "Videos", "all", "and", "class", "config", "def", "device", "do", "dtype", "else", "expand_as", "f", "features", "get_input_embeddings", "get_placeholder_mask", "if", "image", "image_features", "image_token_id", "input_ids", "inputs_embeds", "is", "long", "match", "n_image_tokens", "n_video_tokens", "not", "numel", "raise", "return", "self", "shape", "special_image_mask", "special_video_mask", "sum", "tensor", "to", "tokens", "torch", "unsqueeze", "video", "video_features", "video_token_id"], "qwen2_5_vl/modeling_qwen2_5_vl.py:Qwen2_5_VLModel.forward": ["Cache", "FloatTensor", "LongTensor", "ModelModel", "ModelModelOutputWithPast", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "_", "arange", "attention_mask", "attentions", "batch_size", "bool", "cache_position", "cat", "class", "config", "def", "delta", "device", "dim", "dtype", "else", "expand", "forward", "get_image_features", "get_input_embeddings", "get_placeholder_mask", "get_rope_index", "get_seq_length", "get_video_features", "hidden_states", "if", "image_embeds", "image_features", "image_grid_thw", "image_mask", "input_ids", "inputs_embeds", "is", "kwargs", "language_model", "last_hidden_state", "masked_scatter", "not", "or", "output", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "past_key_values_length", "pixel_values", "pixel_values_videos", "position_ids", "r", "repeat_interleave", "return", "return_dict", "rope_deltas", "second_per_grid_ts", "self", "seq_length", "shape", "to", "to_tuple", "torch", "tuple", "use_cache", "use_return_dict", "video_embeds", "video_features", "video_grid_thw", "video_mask", "view"], "qwen2_5_vl/modeling_qwen2_5_vl.py:Qwen2_5_VLForConditionalGeneration.__init__": ["False", "GenerationMixin", "Linear", "ModelForConditionalGeneration", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "text_config", "vocab_size"], "qwen2_5_vl/modeling_qwen2_5_vl.py:Qwen2_5_VLForConditionalGeneration.get_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "model", "return", "self"], "qwen2_5_vl/modeling_qwen2_5_vl.py:Qwen2_5_VLForConditionalGeneration.set_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "model", "self", "set_input_embeddings", "value"], "qwen2_5_vl/modeling_qwen2_5_vl.py:Qwen2_5_VLForConditionalGeneration.get_video_features": ["FloatTensor", "GenerationMixin", "LongTensor", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "class", "def", "get_video_features", "model", "pixel_values_videos", "return", "self", "torch", "video_grid_thw"], "qwen2_5_vl/modeling_qwen2_5_vl.py:Qwen2_5_VLForConditionalGeneration.get_image_features": ["FloatTensor", "GenerationMixin", "LongTensor", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "class", "def", "get_image_features", "image_grid_thw", "model", "pixel_values", "return", "self", "torch"], "qwen2_5_vl/modeling_qwen2_5_vl.py:Qwen2_5_VLForConditionalGeneration.forward": ["Cache", "FloatTensor", "GenerationMixin", "LongTensor", "ModelCausalLMOutputWithPast", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "else", "forward", "hidden_states", "if", "image_grid_thw", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "pixel_values", "pixel_values_videos", "position_ids", "r", "return", "return_dict", "rope_deltas", "second_per_grid_ts", "self", "slice", "slice_indices", "text_config", "torch", "tuple", "use_cache", "video_grid_thw", "vocab_size"], "qwen2_5_vl/modeling_qwen2_5_vl.py:Qwen2_5_VLForConditionalGeneration.prepare_inputs_for_generation": ["False", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "True", "and", "arange", "attention_mask", "batch_size", "cache_position", "cat", "class", "def", "delta", "device", "dim", "elif", "expand", "expand_as", "get", "get_rope_index", "if", "image_grid_thw", "in", "input_ids", "inputs_embeds", "is", "is_first_iteration", "kwargs", "model", "model_inputs", "not", "or", "past_key_values", "pixel_values", "pixel_values_videos", "position_ids", "prepare_inputs_for_generation", "repeat_interleave", "return", "rope_deltas", "second_per_grid_ts", "self", "seq_length", "shape", "super", "text_positions", "torch", "use_cache", "video_grid_thw", "view", "vision_positions"], "qwen2_5_vl/modeling_qwen2_5_vl.py:Qwen2_5_VLForConditionalGeneration._get_image_nums_and_video_nums": ["GenerationMixin", "LongTensor", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Tensor", "_get_image_nums_and_video_nums", "class", "config", "def", "device", "dim", "dims", "dtype", "else", "get_input_embeddings", "if", "image_mask", "image_nums", "image_token_id", "input_ids", "inputs_embeds", "is", "long", "not", "return", "roll", "self", "shifts", "sum", "tensor", "torch", "tuple", "video_mask", "video_nums", "video_token_id", "vision_first_mask", "vision_start_mask", "vision_start_token_id"], "qwen2_5_vl/modeling_qwen2_5_vl.py:Qwen2_5_VLForConditionalGeneration._expand_inputs_for_generation": ["Any", "False", "GenerationMixin", "If", "LongTensor", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Tensor", "True", "ValueError", "_expand_dict_for_generation", "_expand_dict_for_generation_visual", "_expand_inputs_for_generation", "_get_image_nums_and_video_nums", "_repeat_interleave_samples", "and", "bool", "cache_position", "cat", "class", "def", "defined", "dict", "dict_to_expand", "dim", "elif", "encoder_outputs", "expand_size", "for", "get", "if", "image_grid_thw", "image_nums", "in", "input_ids", "inputs_embeds", "int", "is", "is_encoder_decoder", "isinstance", "key", "lengths", "list", "make", "model_kwargs", "not", "pixel_values", "pixel_values_videos", "prod", "raise", "repeat", "repeat_args", "repeat_interleave", "repeat_times", "result", "return", "sample", "samples", "second_per_grid_ts", "self", "split", "str", "sum", "sure", "that", "torch", "tuple", "video_grid_thw", "video_nums", "visual_keys", "x"], "qwen2_audio/modeling_qwen2_audio.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "and", "attention_mask", "attn_output", "attn_weights", "contiguous", "def", "dim", "dropout", "float", "functional", "if", "is", "key", "kwargs", "matmul", "module", "ndim", "nn", "not", "p", "query", "return", "scaling", "shape", "size", "softmax", "torch", "training", "transpose", "value"], "qwen2_audio/modeling_qwen2_audio.py:Qwen2AudioAttention.__init__": ["False", "Instantiating", "Linear", "ModelAttention", "ModelConfig", "Module", "None", "Please", "True", "ValueError", "__class__", "__init__", "__name__", "a", "and", "be", "bias", "bool", "by", "caching", "call", "class", "config", "creating", "decoder", "def", "divisible", "dropout", "during", "embed_dim", "errors", "f", "float", "forward", "got", "head_dim", "if", "int", "is", "is_causal", "is_decoder", "k_proj", "layer_idx", "logger", "make", "must", "nn", "not", "num_heads", "out_proj", "passing", "provide", "q_proj", "raise", "recommended", "scaling", "self", "super", "sure", "the", "this", "to", "used", "v_proj", "warning_once", "when", "will", "without"], "qwen2_audio/modeling_qwen2_audio.py:Qwen2AudioAttention._shape": ["ModelAttention", "Module", "Tensor", "_shape", "bsz", "class", "contiguous", "def", "head_dim", "int", "nn", "num_heads", "return", "self", "seq_len", "tensor", "torch", "transpose", "view"], "qwen2_audio/modeling_qwen2_audio.py:Qwen2AudioAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "False", "ModelAttention", "Module", "None", "Tensor", "_", "_attn_implementation", "_shape", "attention_interface", "attention_mask", "attn_output", "attn_weights", "bool", "bsz", "class", "config", "contiguous", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "hidden_states", "if", "k_proj", "key_states", "kwargs", "nn", "not", "out_proj", "output_attentions", "q_proj", "query_states", "reshape", "return", "scaling", "self", "size", "tgt_len", "torch", "training", "tuple", "v_proj", "value_states"], "qwen2_audio/modeling_qwen2_audio.py:Qwen2AudioEncoderLayer.__init__": ["ACT2FN", "GradientCheckpointingLayer", "LayerNorm", "Linear", "ModelAttention", "ModelConfig", "ModelEncoderLayer", "__init__", "activation_dropout", "activation_fn", "activation_function", "attention_dropout", "class", "config", "d_model", "def", "dropout", "embed_dim", "encoder_attention_heads", "encoder_ffn_dim", "fc1", "fc2", "final_layer_norm", "nn", "num_heads", "self", "self_attn", "self_attn_layer_norm", "super"], "qwen2_audio/modeling_qwen2_audio.py:Qwen2AudioEncoderLayer.forward": ["False", "GradientCheckpointingLayer", "ModelEncoderLayer", "Tensor", "activation_dropout", "activation_fn", "attention_mask", "attn_weights", "bool", "clamp", "clamp_value", "class", "def", "dropout", "dtype", "fc1", "fc2", "final_layer_norm", "finfo", "float16", "forward", "functional", "hidden_states", "if", "max", "min", "nn", "output_attentions", "p", "residual", "return", "self", "self_attn", "self_attn_layer_norm", "torch", "training"], "qwen2_audio/modeling_qwen2_audio.py:Qwen2AudioEncoder.__init__": ["AvgPool1d", "Conv1d", "Embedding", "False", "LayerNorm", "ModelEncoder", "ModelEncoderConfig", "ModelEncoderLayer", "ModelPreTrainedModel", "ModuleList", "_", "__init__", "avg_pooler", "class", "config", "conv1", "conv2", "d_model", "def", "dropout", "else", "embed_dim", "embed_positions", "embed_scale", "encoder_layerdrop", "encoder_layers", "for", "gradient_checkpointing", "if", "in", "kernel_size", "layer_norm", "layerdrop", "layers", "math", "max_source_positions", "nn", "num_mel_bins", "pad_token_id", "padding", "padding_idx", "post_init", "range", "requires_grad_", "scale_embedding", "self", "sqrt", "stride", "super"], "qwen2_audio/modeling_qwen2_audio.py:Qwen2AudioEncoder._freeze_parameters": ["False", "ModelEncoder", "ModelPreTrainedModel", "_freeze_parameters", "_requires_grad", "class", "def", "for", "in", "param", "parameters", "requires_grad", "self"], "qwen2_audio/modeling_qwen2_audio.py:Qwen2AudioEncoder.get_input_embeddings": ["ModelEncoder", "ModelPreTrainedModel", "Module", "class", "conv1", "def", "get_input_embeddings", "nn", "return", "self"], "qwen2_audio/modeling_qwen2_audio.py:Qwen2AudioEncoder.set_input_embeddings": ["ModelEncoder", "ModelPreTrainedModel", "Module", "class", "conv1", "def", "nn", "self", "set_input_embeddings", "value"], "qwen2_audio/modeling_qwen2_audio.py:Qwen2AudioEncoder.forward": ["BaseModelOutput", "False", "Make", "Model", "ModelEncoder", "ModelPreTrainedModel", "None", "True", "ValueError", "all_attentions", "attention_mask", "attentions", "avg_pooler", "be", "but", "class", "config", "conv1", "conv2", "def", "device", "dropout", "dropout_probability", "dtype", "else", "embed_pos", "embed_positions", "encoder_layer", "encoder_states", "enumerate", "expected_seq_length", "expects", "f", "features", "for", "forward", "found", "functional", "gelu", "hidden_states", "idx", "if", "in", "input", "input_features", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layer_norm", "layer_outputs", "layerdrop", "layers", "length", "max_source_positions", "mel", "nn", "not", "of", "output_attentions", "output_hidden_states", "p", "pad", "permute", "r", "raise", "rand", "return", "return_dict", "self", "shape", "stride", "sure", "the", "to", "to_drop", "torch", "training", "tuple", "use_return_dict", "v", "weight"], "qwen2_audio/modeling_qwen2_audio.py:Qwen2AudioEncoder._get_feat_extract_output_lengths": ["LongTensor", "ModelEncoder", "ModelPreTrainedModel", "_get_feat_extract_output_lengths", "class", "def", "input_lengths", "output_lengths", "return", "self", "torch"], "qwen2_audio/modeling_qwen2_audio.py:Qwen2AudioMultiModalProjector.__init__": ["Linear", "ModelConfig", "ModelMultiModalProjector", "Module", "True", "__init__", "audio_config", "bias", "class", "config", "d_model", "def", "hidden_size", "linear", "nn", "self", "super", "text_config"], "qwen2_audio/modeling_qwen2_audio.py:Qwen2AudioMultiModalProjector.forward": ["ModelMultiModalProjector", "Module", "audio_features", "class", "def", "forward", "hidden_states", "linear", "nn", "return", "self"], "qwen2_audio/modeling_qwen2_audio.py:Qwen2AudioForConditionalGeneration.__init__": ["AutoModel", "AutoModelForCausalLM", "GenerationMixin", "ModelConfig", "ModelForConditionalGeneration", "ModelMultiModalProjector", "ModelPreTrainedModel", "None", "__init__", "_padding_side", "audio_config", "audio_tower", "class", "config", "def", "else", "from_config", "if", "is", "language_model", "left", "multi_modal_projector", "not", "pad_token_id", "post_init", "self", "super", "text_config", "vocab_size"], "qwen2_audio/modeling_qwen2_audio.py:Qwen2AudioForConditionalGeneration.padding_side": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "ValueError", "_padding_side", "class", "def", "f", "if", "in", "is", "left", "not", "or", "padding_side", "raise", "right", "self", "str"], "qwen2_audio/modeling_qwen2_audio.py:Qwen2AudioForConditionalGeneration.get_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "language_model", "return", "self"], "qwen2_audio/modeling_qwen2_audio.py:Qwen2AudioForConditionalGeneration.set_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "language_model", "self", "set_input_embeddings", "value"], "qwen2_audio/modeling_qwen2_audio.py:Qwen2AudioForConditionalGeneration.get_output_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "get_output_embeddings", "language_model", "return", "self"], "qwen2_audio/modeling_qwen2_audio.py:Qwen2AudioForConditionalGeneration.set_output_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "language_model", "new_embeddings", "self", "set_output_embeddings"], "qwen2_audio/modeling_qwen2_audio.py:Qwen2AudioForConditionalGeneration.set_decoder": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "decoder", "def", "language_model", "self", "set_decoder"], "qwen2_audio/modeling_qwen2_audio.py:Qwen2AudioForConditionalGeneration.get_decoder": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "get_decoder", "language_model", "return", "self"], "qwen2_audio/modeling_qwen2_audio.py:Qwen2AudioForConditionalGeneration._merge_input_ids_with_audio_features": ["False", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "The", "This", "True", "ValueError", "_left_padding", "_merge_input_ids_with_audio_features", "_right_padding", "and", "any", "arange", "are", "attention_mask", "audio", "audio_features", "audio_features_mask", "audio_to_overwrite", "audio_token_id", "batch", "batch_indices", "batch_size", "bool", "both", "breaks", "class", "config", "contiguous", "correct", "cumsum", "def", "device", "dim", "dtype", "elif", "else", "embed_dim", "expand", "f", "final_attention_mask", "final_embedding", "final_input_ids", "final_labels", "full", "full_like", "generation", "given", "has", "if", "ignore_index", "indexing", "input", "input_ids", "inputs_embeds", "invalid", "is", "labels", "left", "left_padding", "long", "masked_audio_features", "masked_fill_", "max", "max_audio_tokens", "max_token_num", "model", "nb_audio_pad", "new_token_positions", "non_audio_indices", "not", "num_audio_tokens", "num_audios", "num_special_audio_tokens", "number", "of", "pad_token_id", "padding_side", "position_ids", "prevents", "provided", "raise", "reshape", "return", "self", "seq_indices", "sequence_length", "shape", "side", "special_audio_token_mask", "sum", "target_device", "text_to_overwrite", "the", "to", "token_placeholder_num", "tokens", "torch", "unsqueeze", "val", "view", "where", "while", "wrong", "zero", "zeros", "zeros_like"], "qwen2_audio/modeling_qwen2_audio.py:Qwen2AudioForConditionalGeneration.forward": ["Audio", "Cache", "CrossEntropyLoss", "Expanding", "FloatTensor", "GenerationMixin", "LongTensor", "Model", "ModelCausalLMOutputWithPast", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Tensor", "ValueError", "_", "_get_feat_extract_output_lengths", "_merge_input_ids_with_audio_features", "and", "arange", "attention_mask", "attentions", "audio", "audio_attention_mask", "audio_attention_mask_2d", "audio_feat_lengths", "audio_features", "audio_features_mask", "audio_output_lengths", "audio_outputs", "audio_token_id", "audio_tokens", "audio_tower", "batch_size", "be", "bool", "cache_position", "class", "config", "contiguous", "create_bidirectional_mask", "def", "device", "do", "done", "dtype", "dummy_embeds", "else", "embed_dim", "expand", "expand_as", "f", "feature_attention_mask", "features", "for", "forward", "get_input_embeddings", "hidden_states", "if", "in", "input_embeds", "input_features", "input_ids", "inputs", "inputs_embeds", "is", "item", "kwargs", "labels", "language_model", "last_hidden_state", "legacy_processing", "lengths_expand", "logger", "logits", "long", "loss", "loss_fct", "masked_scatter", "match", "max_audio_tokens", "max_mel_seq_len", "max_seq_len", "multi_modal_projector", "n_audio_features", "n_audio_tokens", "nn", "not", "num_audios", "output", "output_attentions", "output_hidden_states", "outputs", "padding_mask", "past_key_values", "position_ids", "processing", "r", "raise", "return", "return_dict", "selected_audio_feature", "self", "seq_range", "shape", "shift_attention_mask", "shift_labels", "shift_logits", "should", "size", "special_audio_mask", "sum", "target_device", "to", "tokens", "torch", "tuple", "unsqueeze", "use_cache", "use_return_dict", "view", "warning_once", "zeros"], "qwen2_audio/modeling_qwen2_audio.py:Qwen2AudioForConditionalGeneration.prepare_inputs_for_generation": ["False", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "True", "args", "class", "def", "get", "if", "input_features", "is_first_iteration", "kwargs", "model_inputs", "not", "or", "pop", "prepare_inputs_for_generation", "return", "self", "super", "use_cache"], "qwen2_moe/modeling_qwen2_moe.py:Qwen2MoeRMSNorm.__init__": ["ModelRMSNorm", "Module", "Parameter", "__init__", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "qwen2_moe/modeling_qwen2_moe.py:Qwen2MoeRMSNorm.forward": ["ModelRMSNorm", "Module", "True", "class", "def", "dtype", "float32", "forward", "hidden_states", "input_dtype", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "qwen2_moe/modeling_qwen2_moe.py:Qwen2MoeRMSNorm.extra_repr": ["ModelRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "variance_epsilon", "weight"], "qwen2_moe/modeling_qwen2_moe.py:Qwen2MoeRotaryEmbedding.__init__": ["Callable", "False", "ModelConfig", "ModelRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "qwen2_moe/modeling_qwen2_moe.py:Qwen2MoeRotaryEmbedding.compute_default_rope_parameters": ["ModelConfig", "ModelRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "qwen2_moe/modeling_qwen2_moe.py:Qwen2MoeRotaryEmbedding.forward": ["False", "ModelRotaryEmbedding", "Module", "None", "and", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "qwen2_moe/modeling_qwen2_moe.py:Qwen2MoeMLP.__init__": ["ACT2FN", "False", "Linear", "ModelMLP", "Module", "None", "__init__", "act_fn", "bias", "class", "config", "def", "down_proj", "else", "gate_proj", "hidden_act", "hidden_size", "if", "intermediate_size", "is", "nn", "self", "super", "up_proj"], "qwen2_moe/modeling_qwen2_moe.py:Qwen2MoeMLP.forward": ["ModelMLP", "Module", "act_fn", "class", "def", "down_proj", "forward", "gate_proj", "nn", "return", "self", "up_proj", "x"], "qwen2_moe/modeling_qwen2_moe.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "qwen2_moe/modeling_qwen2_moe.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cos", "def", "k", "k_embed", "q", "q_embed", "return", "rotate_half", "sin", "unsqueeze", "unsqueeze_dim"], "qwen2_moe/modeling_qwen2_moe.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "qwen2_moe/modeling_qwen2_moe.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value", "value_states"], "qwen2_moe/modeling_qwen2_moe.py:Qwen2MoeAttention.__init__": ["False", "Linear", "ModelAttention", "ModelConfig", "Module", "True", "__init__", "attention_dropout", "bias", "class", "config", "def", "getattr", "head_dim", "hidden_size", "if", "int", "is_causal", "k_proj", "layer_idx", "layer_types", "nn", "num_attention_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "q_proj", "qkv_bias", "scaling", "self", "sliding_attention", "sliding_window", "super", "v_proj"], "qwen2_moe/modeling_qwen2_moe.py:Qwen2MoeAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "LongTensor", "ModelAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_attn_implementation", "apply_rotary_pos_emb", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "cache_kwargs", "cache_position", "class", "config", "contiguous", "cos", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "head_dim", "hidden_shape", "hidden_states", "if", "input_shape", "is", "k_proj", "key_states", "kwargs", "layer_idx", "nn", "not", "o_proj", "past_key_values", "position_embeddings", "q_proj", "query_states", "reshape", "return", "scaling", "self", "shape", "sin", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "view"], "qwen2_moe/modeling_qwen2_moe.py:Qwen2MoeExperts.__init__": ["ACT2FN", "ModelExperts", "Module", "Parameter", "__init__", "act_fn", "class", "config", "def", "down_proj", "empty", "gate_up_proj", "hidden_act", "hidden_dim", "hidden_size", "intermediate_dim", "moe_intermediate_size", "nn", "num_experts", "self", "super", "torch"], "qwen2_moe/modeling_qwen2_moe.py:Qwen2MoeExperts.forward": ["ModelExperts", "Module", "None", "Tensor", "act_fn", "chunk", "class", "continue", "current_hidden_states", "current_state", "def", "dim", "down_proj", "dtype", "expert_hit", "expert_idx", "expert_mask", "final_hidden_states", "for", "forward", "functional", "gate", "gate_up_proj", "greater", "hidden_states", "if", "in", "index_add_", "linear", "nn", "no_grad", "nonzero", "num_classes", "num_experts", "one_hot", "permute", "return", "self", "sum", "to", "token_idx", "top_k_index", "top_k_pos", "top_k_weights", "torch", "up", "where", "with", "zeros_like"], "qwen2_moe/modeling_qwen2_moe.py:Qwen2MoeTopKRouter.__init__": ["ModelTopKRouter", "Module", "Parameter", "__init__", "class", "config", "def", "hidden_dim", "hidden_size", "nn", "norm_topk_prob", "num_experts", "num_experts_per_tok", "self", "super", "top_k", "torch", "weight", "zeros"], "qwen2_moe/modeling_qwen2_moe.py:Qwen2MoeTopKRouter.forward": ["F", "ModelTopKRouter", "Module", "True", "class", "def", "dim", "dtype", "float", "forward", "functional", "hidden_dim", "hidden_states", "if", "keepdim", "linear", "nn", "norm_topk_prob", "reshape", "return", "router_indices", "router_logits", "router_scores", "router_top_value", "self", "softmax", "sum", "to", "top_k", "topk", "torch", "weight"], "qwen2_moe/modeling_qwen2_moe.py:Qwen2MoeSparseMoeBlock.__init__": ["False", "Linear", "ModelExperts", "ModelMLP", "ModelSparseMoeBlock", "ModelTopKRouter", "Module", "__init__", "bias", "class", "config", "def", "experts", "gate", "hidden_size", "intermediate_size", "nn", "self", "shared_expert", "shared_expert_gate", "shared_expert_intermediate_size", "super", "torch"], "qwen2_moe/modeling_qwen2_moe.py:Qwen2MoeSparseMoeBlock.forward": ["F", "ModelSparseMoeBlock", "Module", "Tensor", "_", "batch_size", "class", "def", "expert_output", "experts", "forward", "gate", "hidden_dim", "hidden_states", "hidden_states_reshaped", "nn", "reshape", "return", "routing_weights", "selected_experts", "self", "sequence_length", "shape", "shared_expert", "shared_expert_gate", "shared_expert_output", "sigmoid", "torch", "tuple", "view"], "qwen2_moe/modeling_qwen2_moe.py:Qwen2MoeDecoderLayer.__init__": ["GradientCheckpointingLayer", "ModelAttention", "ModelConfig", "ModelDecoderLayer", "ModelMLP", "ModelRMSNorm", "ModelSparseMoeBlock", "__init__", "and", "class", "config", "decoder_sparse_step", "def", "else", "eps", "hidden_size", "if", "in", "input_layernorm", "int", "intermediate_size", "layer_idx", "mlp", "mlp_only_layers", "not", "num_experts", "post_attention_layernorm", "rms_norm_eps", "self", "self_attn", "super"], "qwen2_moe/modeling_qwen2_moe.py:Qwen2MoeDecoderLayer.forward": ["Cache", "False", "GradientCheckpointingLayer", "LongTensor", "ModelDecoderLayer", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "attention_mask", "bool", "cache_position", "class", "def", "forward", "hidden_states", "input_layernorm", "kwargs", "mlp", "past_key_values", "position_embeddings", "position_ids", "post_attention_layernorm", "residual", "return", "self", "self_attn", "torch", "tuple", "use_cache"], "qwen2_moe/modeling_qwen2_moe.py:Qwen2MoePreTrainedModel._init_weights": ["ModelExperts", "ModelPreTrainedModel", "ModelTopKRouter", "PreTrainedModel", "_init_weights", "class", "config", "def", "down_proj", "elif", "gate_up_proj", "if", "init", "initializer_range", "isinstance", "mean", "module", "normal_", "self", "std", "super", "weight"], "qwen2_moe/modeling_qwen2_moe.py:Qwen2MoeModel.__init__": ["Embedding", "False", "ModelConfig", "ModelDecoderLayer", "ModelModel", "ModelPreTrainedModel", "ModelRMSNorm", "ModelRotaryEmbedding", "ModuleList", "__init__", "class", "config", "def", "embed_tokens", "eps", "for", "gradient_checkpointing", "hidden_size", "in", "layer_idx", "layers", "nn", "norm", "num_hidden_layers", "pad_token_id", "padding_idx", "post_init", "range", "rms_norm_eps", "rotary_emb", "self", "super", "vocab_size"], "qwen2_moe/modeling_qwen2_moe.py:Qwen2MoeModel.forward": ["Cache", "DynamicCache", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "MoeModelOutputWithPast", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "and", "arange", "attention_mask", "bool", "cache_position", "causal_mask_mapping", "class", "config", "create_causal_mask", "create_sliding_window_causal_mask", "decoder_layer", "def", "device", "dict", "else", "embed_tokens", "enumerate", "exactly", "for", "forward", "full_attention", "get_seq_length", "hidden_states", "i", "if", "in", "input_embeds", "input_ids", "inputs_embeds", "is", "isinstance", "kwargs", "last_hidden_state", "layer_types", "layers", "mask_kwargs", "must", "norm", "not", "num_hidden_layers", "of", "one", "or", "past_key_values", "past_seen_tokens", "position_embeddings", "position_ids", "raise", "return", "rotary_emb", "self", "shape", "sliding_attention", "specify", "torch", "unsqueeze", "use_cache"], "qwen2_moe/modeling_qwen2_moe.py:load_balancing_loss_func": ["Model_balancing_loss_func", "None", "Tensor", "_", "attention_mask", "batch_size", "cat", "compute_device", "concatenated_gate_logits", "def", "device", "dim", "else", "expand", "expert_attention_mask", "expert_mask", "float", "for", "functional", "gate_logits", "if", "in", "int", "is", "isinstance", "layer_gate", "mean", "nn", "not", "num_experts", "num_hidden_layers", "one_hot", "or", "overall_loss", "r", "reshape", "return", "router_per_expert_attention_mask", "router_prob_per_expert", "routing_weights", "selected_experts", "sequence_length", "shape", "softmax", "sum", "to", "tokens_per_expert", "top_k", "topk", "torch", "tuple", "unsqueeze"], "qwen2_moe/modeling_qwen2_moe.py:Qwen2MoeForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "ModelForCausalLM", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "num_experts", "num_experts_per_tok", "post_init", "router_aux_loss_coef", "self", "super", "vocab_size"], "qwen2_moe/modeling_qwen2_moe.py:Qwen2MoeForCausalLM.forward": ["Cache", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForCausalLM", "ModelPreTrainedModel", "MoeCausalLMOutputWithPast", "MoeModelOutputWithPast", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "aux_loss", "bool", "cache_position", "class", "config", "def", "device", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "load_balancing_loss_func", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "num_experts", "num_experts_per_tok", "output_router_logits", "outputs", "past_key_values", "position_ids", "r", "return", "router_aux_loss_coef", "router_logits", "self", "slice", "slice_indices", "to", "torch", "use_cache", "vocab_size"], "qwen2_vl/modeling_qwen2_vl.py:Qwen2VLRotaryEmbedding.__init__": ["Callable", "False", "ModelConfig", "ModelRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "qwen2_vl/modeling_qwen2_vl.py:Qwen2VLRotaryEmbedding.compute_default_rope_parameters": ["ModelConfig", "ModelRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "qwen2_vl/modeling_qwen2_vl.py:Qwen2VLRotaryEmbedding.forward": ["False", "ModelRotaryEmbedding", "Module", "None", "and", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "qwen2_vl/modeling_qwen2_vl.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "qwen2_vl/modeling_qwen2_vl.py:apply_multimodal_rotary_pos_emb": ["Model_multimodal_rotary_pos_emb", "cat", "cos", "def", "dim", "enumerate", "for", "i", "in", "k", "k_embed", "m", "mrope_section", "q", "q_embed", "return", "rotate_half", "sin", "split", "torch", "unsqueeze", "unsqueeze_dim"], "qwen2_vl/modeling_qwen2_vl.py:apply_rotary_pos_emb_vision": ["Model_rotary_pos_emb_vision", "Tensor", "cos", "def", "dtype", "float", "k", "k_embed", "orig_k_dtype", "orig_q_dtype", "q", "q_embed", "return", "rotate_half", "sin", "to", "torch", "tuple", "unsqueeze"], "qwen2_vl/modeling_qwen2_vl.py:VisionRotaryEmbedding.__init__": ["False", "ModelRotaryEmbedding", "Module", "None", "__init__", "arange", "class", "def", "dim", "dtype", "float", "int", "inv_freq", "nn", "persistent", "register_buffer", "self", "super", "theta", "torch"], "qwen2_vl/modeling_qwen2_vl.py:VisionRotaryEmbedding.forward": ["ModelRotaryEmbedding", "Module", "Tensor", "arange", "class", "def", "device", "dtype", "forward", "freqs", "int", "inv_freq", "nn", "outer", "return", "self", "seq", "seqlen", "torch"], "qwen2_vl/modeling_qwen2_vl.py:PatchEmbed.__init__": ["Conv3d", "False", "ModelEmbed", "Model_size", "Module", "None", "__init__", "bias", "class", "def", "embed_dim", "in_channels", "int", "kernel_size", "nn", "proj", "self", "stride", "super", "temporal_Model_size"], "qwen2_vl/modeling_qwen2_vl.py:PatchEmbed.forward": ["ModelEmbed", "Model_size", "Module", "Tensor", "class", "def", "dtype", "embed_dim", "forward", "hidden_states", "in_channels", "nn", "proj", "return", "self", "target_dtype", "temporal_Model_size", "to", "torch", "view", "weight"], "qwen2_vl/modeling_qwen2_vl.py:PatchMerger.__init__": ["GELU", "LayerNorm", "Linear", "ModelMerger", "Module", "None", "Sequential", "__init__", "class", "context_dim", "def", "dim", "eps", "hidden_size", "int", "ln_q", "mlp", "nn", "self", "spatial_merge_size", "super"], "qwen2_vl/modeling_qwen2_vl.py:PatchMerger.forward": ["ModelMerger", "Module", "Tensor", "class", "def", "forward", "hidden_size", "ln_q", "mlp", "nn", "return", "self", "torch", "view", "x"], "qwen2_vl/modeling_qwen2_vl.py:VisionMlp.__init__": ["ACT2FN", "Linear", "ModelMlp", "Module", "None", "__init__", "act", "class", "def", "dim", "fc1", "fc2", "hidden_act", "hidden_dim", "int", "nn", "self", "str", "super"], "qwen2_vl/modeling_qwen2_vl.py:VisionMlp.forward": ["ModelMlp", "Module", "Tensor", "act", "class", "def", "fc1", "fc2", "forward", "nn", "return", "self", "torch", "x"], "qwen2_vl/modeling_qwen2_vl.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "qwen2_vl/modeling_qwen2_vl.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value", "value_states"], "qwen2_vl/modeling_qwen2_vl.py:VisionAttention.__init__": ["False", "Linear", "ModelAttention", "ModelModelConfig", "Module", "None", "True", "__init__", "attention_dropout", "bias", "class", "config", "def", "dim", "embed_dim", "head_dim", "is_causal", "nn", "num_heads", "num_key_value_groups", "proj", "qkv", "scaling", "self", "super"], "qwen2_vl/modeling_qwen2_vl.py:VisionAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "False", "ModelAttention", "Module", "None", "Tensor", "_", "_attn_implementation", "apply_rotary_pos_emb_Model", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_outputs", "cat", "class", "config", "contiguous", "cos", "cu_seq_lens_k", "cu_seq_lens_q", "cu_seqlens", "def", "dim", "dropout", "eager", "eager_attention_forward", "else", "flash", "for", "forward", "hidden_states", "if", "in", "is_causal", "k", "key_states", "kwargs", "lengths", "max", "max_length_k", "max_length_q", "max_seqlen", "nn", "not", "num_heads", "permute", "position_embeddings", "proj", "q", "qkv", "query_states", "reshape", "return", "rotary_pos_emb", "scaling", "self", "seq_length", "shape", "sin", "split", "splits", "tensor", "tolist", "torch", "training", "transpose", "tuple", "unbind", "unsqueeze", "v", "value_states", "zip"], "qwen2_vl/modeling_qwen2_vl.py:Qwen2VLVisionBlock.__init__": ["GradientCheckpointingLayer", "LayerNorm", "ModelVisionBlock", "None", "VisionAttention", "VisionMlp", "__init__", "attn", "attn_implementation", "class", "config", "def", "dim", "embed_dim", "eps", "hidden_act", "hidden_dim", "int", "mlp", "mlp_hidden_dim", "mlp_ratio", "norm1", "norm2", "sdpa", "self", "str", "super"], "qwen2_vl/modeling_qwen2_vl.py:Qwen2VLVisionBlock.forward": ["GradientCheckpointingLayer", "ModelVisionBlock", "None", "Tensor", "attn", "class", "cu_seqlens", "def", "forward", "hidden_states", "kwargs", "mlp", "norm1", "norm2", "position_embeddings", "return", "rotary_pos_emb", "self", "torch", "tuple"], "qwen2_vl/modeling_qwen2_vl.py:Qwen2MLP.__init__": ["ACT2FN", "False", "Linear", "ModelMLP", "Module", "__init__", "act_fn", "bias", "class", "config", "def", "down_proj", "gate_proj", "hidden_act", "hidden_size", "intermediate_size", "nn", "self", "super", "up_proj"], "qwen2_vl/modeling_qwen2_vl.py:Qwen2MLP.forward": ["ModelMLP", "Module", "act_fn", "class", "def", "down_proj", "forward", "gate_proj", "nn", "return", "self", "up_proj", "x"], "qwen2_vl/modeling_qwen2_vl.py:Qwen2VLAttention.__init__": ["False", "Instantiating", "Linear", "ModelAttention", "ModelTextConfig", "Module", "None", "Please", "True", "ValueError", "__class__", "__init__", "__name__", "a", "and", "attention_dropout", "be", "bias", "by", "caching", "call", "class", "config", "creating", "def", "divisible", "during", "else", "errors", "f", "forward", "got", "hasattr", "head_dim", "hidden_size", "if", "int", "is", "is_causal", "k_proj", "layer_idx", "layer_type", "layer_types", "logger", "make", "must", "nn", "not", "num_attention_heads", "num_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "passing", "provide", "q_proj", "raise", "recommended", "rope_parameters", "scaling", "self", "sliding_attention", "sliding_window", "super", "sure", "the", "this", "to", "used", "v_proj", "warning_once", "when", "will", "without"], "qwen2_vl/modeling_qwen2_vl.py:Qwen2VLAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "False", "FlashAttentionKwargs", "LongTensor", "ModelAttention", "Module", "None", "Tensor", "Unpack", "_", "_attn_implementation", "apply_multimodal_rotary_pos_emb", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "bool", "bsz", "cache_kwargs", "cache_position", "class", "config", "contiguous", "cos", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "head_dim", "hidden_states", "if", "is", "k_proj", "key_states", "kwargs", "layer_idx", "mrope_section", "nn", "not", "o_proj", "output_attentions", "past_key_values", "position_embeddings", "position_ids", "q_len", "q_proj", "query_states", "reshape", "return", "rope_parameters", "scaling", "self", "sin", "size", "sliding_window", "torch", "training", "transpose", "tuple", "update", "use_cache", "v_proj", "value_states", "view"], "qwen2_vl/modeling_qwen2_vl.py:Qwen2VLDecoderLayer.__init__": ["Attention", "GradientCheckpointingLayer", "ModelAttention", "ModelDecoderLayer", "ModelMLP", "ModelRMSNorm", "ModelTextConfig", "Sliding", "Window", "__init__", "_attn_implementation", "and", "attention_type", "be", "but", "class", "config", "def", "enabled", "encountered", "eps", "f", "flash_attention_2", "for", "hidden_size", "if", "implemented", "input_layernorm", "int", "is", "layer_idx", "layer_types", "logger", "may", "mlp", "not", "post_attention_layernorm", "results", "rms_norm_eps", "self", "self_attn", "super", "unexpected", "use_sliding_window", "warning_once"], "qwen2_vl/modeling_qwen2_vl.py:Qwen2VLDecoderLayer.forward": ["Cache", "False", "FlashAttentionKwargs", "FloatTensor", "GradientCheckpointingLayer", "LongTensor", "ModelDecoderLayer", "None", "Tensor", "Unpack", "attention_mask", "bool", "cache_position", "class", "def", "forward", "hidden_states", "if", "input_layernorm", "kwargs", "mlp", "output_attentions", "outputs", "past_key_values", "position_embeddings", "position_ids", "post_attention_layernorm", "residual", "return", "self", "self_attn", "self_attn_weights", "torch", "tuple", "use_cache"], "qwen2_vl/modeling_qwen2_vl.py:Qwen2VLPreTrainedModel._init_weights": ["ModelPreTrainedModel", "PreTrainedModel", "VisionRotaryEmbedding", "_init_weights", "arange", "class", "copy_", "def", "dim", "dtype", "float", "if", "init", "inv_freq", "isinstance", "module", "self", "super", "theta", "torch"], "qwen2_vl/modeling_qwen2_vl.py:Qwen2VisionTransformerPretrainedModel.__init__": ["False", "ModelPreTrainedModel", "ModelVisionBlock", "ModelVisionTransformerPretrainedModel", "ModuleList", "None", "PatchEmbed", "PatchMerger", "VisionRotaryEmbedding", "_", "__init__", "blocks", "class", "config", "context_dim", "def", "depth", "dim", "embed_dim", "for", "gradient_checkpointing", "head_dim", "hidden_size", "in", "in_channels", "merger", "nn", "num_heads", "patch_embed", "patch_size", "post_init", "range", "rotary_pos_emb", "self", "spatial_merge_size", "super", "temporal_patch_size"], "qwen2_vl/modeling_qwen2_vl.py:Qwen2VisionTransformerPretrainedModel.get_dtype": ["ModelPreTrainedModel", "ModelVisionTransformerPretrainedModel", "blocks", "class", "def", "dtype", "fc2", "get_dtype", "mlp", "return", "self", "torch", "weight"], "qwen2_vl/modeling_qwen2_vl.py:Qwen2VisionTransformerPretrainedModel.get_device": ["ModelPreTrainedModel", "ModelVisionTransformerPretrainedModel", "blocks", "class", "def", "device", "fc2", "get_device", "mlp", "return", "self", "torch", "weight"], "qwen2_vl/modeling_qwen2_vl.py:Qwen2VisionTransformerPretrainedModel.rot_pos_emb": ["ModelPreTrainedModel", "ModelVisionTransformerPretrainedModel", "append", "arange", "cat", "class", "def", "dim", "expand", "flatten", "for", "grid_thw", "h", "hpos_ids", "in", "max", "max_grid_size", "permute", "pos_ids", "repeat", "reshape", "return", "rot_pos_emb", "rotary_pos_emb", "rotary_pos_emb_full", "self", "spatial_merge_size", "stack", "t", "torch", "unsqueeze", "w", "wpos_ids"], "qwen2_vl/modeling_qwen2_vl.py:Qwen2VisionTransformerPretrainedModel.forward": ["F", "ModelPreTrainedModel", "ModelVisionTransformerPretrainedModel", "Tensor", "blk", "blocks", "cat", "class", "cos", "cu_seqlens", "cumsum", "def", "dim", "dtype", "else", "emb", "for", "forward", "grid_thw", "hidden_states", "if", "in", "int32", "is_tracing", "jit", "kwargs", "merger", "pad", "patch_embed", "position_embeddings", "r", "repeat_interleave", "return", "rot_pos_emb", "rotary_pos_emb", "self", "sin", "torch", "value"], "qwen2_vl/modeling_qwen2_vl.py:Qwen2VLTextModel.__init__": ["Embedding", "False", "ModelDecoderLayer", "ModelPreTrainedModel", "ModelRMSNorm", "ModelRotaryEmbedding", "ModelTextConfig", "ModelTextModel", "ModuleList", "__init__", "_attn_implementation", "class", "config", "def", "embed_tokens", "eps", "for", "gradient_checkpointing", "has_sliding_layers", "hidden_size", "in", "layer_idx", "layer_types", "layers", "nn", "norm", "num_hidden_layers", "pad_token_id", "padding_idx", "post_init", "range", "rms_norm_eps", "rotary_emb", "self", "sliding_attention", "super", "vocab_size"], "qwen2_vl/modeling_qwen2_vl.py:Qwen2VLTextModel.forward": ["BaseModelOutputWithPast", "Cache", "DynamicCache", "False", "FlashAttentionKwargs", "FloatTensor", "LongTensor", "ModelPreTrainedModel", "ModelTextModel", "None", "Setting", "Tensor", "True", "Unpack", "ValueError", "You", "all_hidden_states", "all_self_attns", "and", "arange", "attention_mask", "attention_type", "attentions", "bool", "cache_position", "causal_mask_mapping", "checkpointing", "class", "config", "create_causal_mask", "create_sliding_window_causal_mask", "decoder_layer", "def", "device", "dict", "elif", "else", "embed_tokens", "exactly", "expand", "for", "forward", "full_attention", "get_seq_length", "gradient", "gradient_checkpointing", "has_sliding_layers", "hidden_states", "if", "in", "incompatible", "input_embeds", "input_ids", "inputs_embeds", "is", "is_tracing", "isinstance", "jit", "kwargs", "last_hidden_state", "layer_outputs", "layers", "logger", "mask_kwargs", "must", "ndim", "norm", "not", "of", "one", "or", "output_attentions", "output_hidden_states", "past_key_values", "past_seen_tokens", "position_embeddings", "position_ids", "raise", "return", "return_dict", "rotary_emb", "self", "shape", "sliding_attention", "specify", "text_position_ids", "torch", "training", "tuple", "use_cache", "use_return_dict", "v", "view", "warning_once", "with"], "qwen2_vl/modeling_qwen2_vl.py:Qwen2VLModel.__init__": ["ModelConfig", "ModelModel", "ModelPreTrainedModel", "ModelTextModel", "ModelVisionTransformerPretrainedModel", "None", "__init__", "_from_config", "class", "config", "def", "language_model", "post_init", "rope_deltas", "self", "super", "text_config", "vision_config", "visual"], "qwen2_vl/modeling_qwen2_vl.py:Qwen2VLModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "language_model", "return", "self"], "qwen2_vl/modeling_qwen2_vl.py:Qwen2VLModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "language_model", "self", "set_input_embeddings", "value"], "qwen2_vl/modeling_qwen2_vl.py:Qwen2VLModel.get_rope_index": ["False", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "True", "_", "and", "append", "arange", "argwhere", "attention_mask", "cat", "class", "config", "cumsum", "def", "device", "dim", "dtype", "ed", "ed_image", "ed_video", "else", "enumerate", "expand", "flatten", "for", "get_rope_index", "h", "h_index", "i", "if", "image_grid_thw", "image_index", "image_nums", "image_token_id", "in", "index", "input_ids", "input_tokens", "is", "item", "keepdim", "len", "list", "llm_grid_h", "llm_grid_t", "llm_grid_w", "llm_pos_ids_list", "llm_positions", "long", "masked_fill_", "max", "max_position_ids", "mrope_position_deltas", "not", "ones", "ones_like", "or", "position_ids", "range", "remain_images", "remain_videos", "reshape", "return", "self", "shape", "spatial_merge_size", "squeeze", "st", "st_idx", "stack", "sum", "t", "t_index", "tensor", "text_len", "to", "tolist", "torch", "total_input_ids", "tuple", "unsqueeze", "video_grid_thw", "video_index", "video_nums", "video_token_id", "view", "vision_config", "vision_start_indices", "vision_start_token_id", "vision_tokens", "w", "w_index", "zeros"], "qwen2_vl/modeling_qwen2_vl.py:Qwen2VLModel.get_video_features": ["FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "class", "def", "dtype", "get_video_features", "grid_thw", "pixel_values_videos", "prod", "return", "self", "spatial_merge_size", "split", "split_sizes", "tolist", "torch", "type", "video_embeds", "video_grid_thw", "visual"], "qwen2_vl/modeling_qwen2_vl.py:Qwen2VLModel.get_image_features": ["FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "class", "def", "dtype", "get_image_features", "grid_thw", "image_embeds", "image_grid_thw", "pixel_values", "prod", "return", "self", "spatial_merge_size", "split", "split_sizes", "tolist", "torch", "type", "visual"], "qwen2_vl/modeling_qwen2_vl.py:Qwen2VLModel.get_placeholder_mask": ["FloatTensor", "Image", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "ValueError", "Videos", "all", "and", "class", "config", "def", "device", "do", "dtype", "else", "expand_as", "f", "features", "get_input_embeddings", "get_placeholder_mask", "if", "image", "image_features", "image_token_id", "input_ids", "inputs_embeds", "is", "long", "match", "n_image_tokens", "n_video_tokens", "not", "numel", "raise", "return", "self", "shape", "special_image_mask", "special_video_mask", "sum", "tensor", "to", "tokens", "torch", "unsqueeze", "video", "video_features", "video_token_id"], "qwen2_vl/modeling_qwen2_vl.py:Qwen2VLModel.forward": ["Cache", "FloatTensor", "LongTensor", "ModelModel", "ModelModelOutputWithPast", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "_", "arange", "attention_mask", "attentions", "batch_size", "bool", "cache_position", "cat", "class", "config", "def", "delta", "device", "dim", "dtype", "else", "expand", "forward", "get_image_features", "get_input_embeddings", "get_placeholder_mask", "get_rope_index", "get_seq_length", "get_video_features", "hidden_states", "if", "image_embeds", "image_features", "image_grid_thw", "image_mask", "input_ids", "inputs_embeds", "is", "kwargs", "language_model", "last_hidden_state", "masked_scatter", "not", "or", "output", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "past_key_values_length", "pixel_values", "pixel_values_videos", "position_ids", "r", "repeat_interleave", "return", "return_dict", "rope_deltas", "self", "seq_length", "shape", "to", "to_tuple", "torch", "tuple", "use_cache", "use_return_dict", "video_embeds", "video_features", "video_grid_thw", "video_mask", "view"], "qwen2_vl/modeling_qwen2_vl.py:Qwen2VLForConditionalGeneration.__init__": ["False", "GenerationMixin", "Linear", "ModelForConditionalGeneration", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "text_config", "vocab_size"], "qwen2_vl/modeling_qwen2_vl.py:Qwen2VLForConditionalGeneration.get_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "model", "return", "self"], "qwen2_vl/modeling_qwen2_vl.py:Qwen2VLForConditionalGeneration.set_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "model", "self", "set_input_embeddings", "value"], "qwen2_vl/modeling_qwen2_vl.py:Qwen2VLForConditionalGeneration.get_video_features": ["FloatTensor", "GenerationMixin", "LongTensor", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "class", "def", "get_video_features", "model", "pixel_values_videos", "return", "self", "torch", "video_grid_thw"], "qwen2_vl/modeling_qwen2_vl.py:Qwen2VLForConditionalGeneration.get_image_features": ["FloatTensor", "GenerationMixin", "LongTensor", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "class", "def", "get_image_features", "image_grid_thw", "model", "pixel_values", "return", "self", "torch"], "qwen2_vl/modeling_qwen2_vl.py:Qwen2VLForConditionalGeneration.forward": ["Cache", "FloatTensor", "GenerationMixin", "LongTensor", "ModelCausalLMOutputWithPast", "ModelForConditionalGeneration", "ModelModelOutputWithPast", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "else", "forward", "hidden_states", "if", "image_grid_thw", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "pixel_values", "pixel_values_videos", "position_ids", "r", "return", "return_dict", "rope_deltas", "self", "slice", "slice_indices", "text_config", "torch", "tuple", "use_cache", "video_grid_thw", "vocab_size"], "qwen2_vl/modeling_qwen2_vl.py:Qwen2VLForConditionalGeneration.prepare_inputs_for_generation": ["False", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "True", "and", "arange", "attention_mask", "batch_size", "cache_position", "cat", "class", "def", "delta", "device", "dim", "elif", "expand", "expand_as", "get", "get_rope_index", "if", "image_grid_thw", "in", "input_ids", "inputs_embeds", "is", "is_first_iteration", "kwargs", "model", "model_inputs", "not", "or", "past_key_values", "pixel_values", "pixel_values_videos", "position_ids", "prepare_inputs_for_generation", "repeat_interleave", "return", "rope_deltas", "self", "seq_length", "shape", "super", "text_positions", "torch", "use_cache", "video_grid_thw", "view", "vision_positions"], "qwen2_vl/modeling_qwen2_vl.py:Qwen2VLForConditionalGeneration._get_image_nums_and_video_nums": ["GenerationMixin", "LongTensor", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Tensor", "_get_image_nums_and_video_nums", "class", "config", "def", "device", "dim", "dims", "dtype", "else", "get_input_embeddings", "if", "image_mask", "image_nums", "image_token_id", "input_ids", "inputs_embeds", "is", "long", "not", "return", "roll", "self", "shifts", "sum", "tensor", "torch", "tuple", "video_mask", "video_nums", "video_token_id", "vision_first_mask", "vision_start_mask", "vision_start_token_id"], "qwen2_vl/modeling_qwen2_vl.py:Qwen2VLForConditionalGeneration._expand_inputs_for_generation": ["Any", "False", "GenerationMixin", "If", "LongTensor", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Tensor", "True", "ValueError", "_expand_dict_for_generation", "_expand_dict_for_generation_visual", "_expand_inputs_for_generation", "_get_image_nums_and_video_nums", "_repeat_interleave_samples", "and", "bool", "cache_position", "cat", "class", "def", "defined", "dict", "dict_to_expand", "dim", "elif", "encoder_outputs", "expand_size", "for", "get", "if", "image_grid_thw", "image_nums", "in", "input_ids", "inputs_embeds", "int", "is", "is_encoder_decoder", "isinstance", "key", "lengths", "list", "make", "model_kwargs", "not", "pixel_values", "pixel_values_videos", "prod", "raise", "repeat", "repeat_args", "repeat_interleave", "repeat_times", "result", "return", "sample", "samples", "second_per_grid_ts", "self", "split", "str", "sum", "sure", "that", "torch", "tuple", "video_grid_thw", "video_nums", "visual_keys", "x"], "qwen3/modeling_qwen3.py:Qwen3RMSNorm.__init__": ["ModelRMSNorm", "Module", "None", "Parameter", "__init__", "class", "def", "eps", "float", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "qwen3/modeling_qwen3.py:Qwen3RMSNorm.forward": ["ModelRMSNorm", "Module", "Tensor", "True", "class", "def", "dtype", "float32", "forward", "hidden_states", "input_dtype", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "qwen3/modeling_qwen3.py:Qwen3RMSNorm.extra_repr": ["ModelRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "variance_epsilon", "weight"], "qwen3/modeling_qwen3.py:Qwen3MLP.__init__": ["ACT2FN", "False", "Linear", "ModelMLP", "Module", "__init__", "act_fn", "bias", "class", "config", "def", "down_proj", "gate_proj", "hidden_act", "hidden_size", "intermediate_size", "nn", "self", "super", "up_proj"], "qwen3/modeling_qwen3.py:Qwen3MLP.forward": ["ModelMLP", "Module", "act_fn", "class", "def", "down_proj", "forward", "gate_proj", "nn", "return", "self", "up_proj", "x"], "qwen3/modeling_qwen3.py:Qwen3RotaryEmbedding.__init__": ["Callable", "False", "ModelConfig", "ModelRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "qwen3/modeling_qwen3.py:Qwen3RotaryEmbedding.compute_default_rope_parameters": ["ModelConfig", "ModelRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "qwen3/modeling_qwen3.py:Qwen3RotaryEmbedding.forward": ["False", "ModelRotaryEmbedding", "Module", "None", "and", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "qwen3/modeling_qwen3.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "qwen3/modeling_qwen3.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cos", "def", "k", "k_embed", "q", "q_embed", "return", "rotate_half", "sin", "unsqueeze", "unsqueeze_dim"], "qwen3/modeling_qwen3.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "qwen3/modeling_qwen3.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value", "value_states"], "qwen3/modeling_qwen3.py:Qwen3Attention.__init__": ["Linear", "ModelAttention", "ModelConfig", "ModelRMSNorm", "Module", "None", "True", "__init__", "attention_bias", "attention_dropout", "bias", "class", "config", "def", "else", "eps", "getattr", "hasattr", "head_dim", "hidden_size", "if", "int", "is_causal", "k_norm", "k_proj", "layer_idx", "layer_type", "layer_types", "nn", "num_attention_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "q_norm", "q_proj", "rms_norm_eps", "scaling", "self", "sliding_attention", "sliding_window", "super", "v_proj"], "qwen3/modeling_qwen3.py:Qwen3Attention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "FlashAttentionKwargs", "LongTensor", "ModelAttention", "Module", "None", "Tensor", "Unpack", "_attn_implementation", "apply_rotary_pos_emb", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "cache_kwargs", "cache_position", "class", "config", "contiguous", "cos", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "head_dim", "hidden_shape", "hidden_states", "if", "input_shape", "is", "k_norm", "k_proj", "key_states", "kwargs", "layer_idx", "nn", "not", "o_proj", "past_key_values", "position_embeddings", "q_norm", "q_proj", "query_states", "reshape", "return", "scaling", "self", "shape", "sin", "sliding_window", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "view"], "qwen3/modeling_qwen3.py:Qwen3DecoderLayer.__init__": ["GradientCheckpointingLayer", "ModelAttention", "ModelConfig", "ModelDecoderLayer", "ModelMLP", "ModelRMSNorm", "__init__", "attention_type", "class", "config", "def", "eps", "hidden_size", "input_layernorm", "int", "layer_idx", "layer_types", "mlp", "post_attention_layernorm", "rms_norm_eps", "self", "self_attn", "super"], "qwen3/modeling_qwen3.py:Qwen3DecoderLayer.forward": ["Cache", "False", "GradientCheckpointingLayer", "LongTensor", "ModelDecoderLayer", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "attention_mask", "bool", "cache_position", "class", "def", "forward", "hidden_states", "input_layernorm", "kwargs", "mlp", "past_key_values", "position_embeddings", "position_ids", "post_attention_layernorm", "residual", "return", "self", "self_attn", "torch", "tuple", "use_cache"], "qwen3/modeling_qwen3.py:Qwen3Model.__init__": ["Embedding", "False", "ModelConfig", "ModelDecoderLayer", "ModelModel", "ModelPreTrainedModel", "ModelRMSNorm", "ModelRotaryEmbedding", "ModuleList", "__init__", "class", "config", "def", "embed_tokens", "eps", "for", "gradient_checkpointing", "has_sliding_layers", "hidden_size", "in", "layer_idx", "layer_types", "layers", "nn", "norm", "num_hidden_layers", "pad_token_id", "padding_idx", "post_init", "range", "rms_norm_eps", "rotary_emb", "self", "sliding_attention", "super", "vocab_size"], "qwen3/modeling_qwen3.py:Qwen3Model.forward": ["BaseModelOutputWithPast", "Cache", "DynamicCache", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "and", "arange", "attention_mask", "attention_type", "bool", "cache_position", "causal_mask_mapping", "class", "config", "create_causal_mask", "create_sliding_window_causal_mask", "decoder_layer", "def", "device", "dict", "else", "embed_tokens", "exactly", "for", "forward", "full_attention", "get_seq_length", "has_sliding_layers", "hidden_states", "if", "in", "input_embeds", "input_ids", "inputs_embeds", "is", "isinstance", "kwargs", "last_hidden_state", "layers", "mask_kwargs", "must", "norm", "not", "num_hidden_layers", "of", "one", "or", "past_key_values", "past_seen_tokens", "position_embeddings", "position_ids", "raise", "return", "rotary_emb", "self", "shape", "sliding_attention", "specify", "torch", "unsqueeze", "use_cache"], "qwen3/modeling_qwen3.py:Qwen3ForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "ModelForCausalLM", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "vocab_size"], "qwen3/modeling_qwen3.py:Qwen3ForCausalLM.forward": ["BaseModelOutputWithPast", "Cache", "CausalLMOutputWithPast", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForCausalLM", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "outputs", "past_key_values", "position_ids", "r", "return", "self", "slice", "slice_indices", "torch", "use_cache", "vocab_size"], "qwen3_moe/modeling_qwen3_moe.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "qwen3_moe/modeling_qwen3_moe.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cos", "def", "k", "k_embed", "q", "q_embed", "return", "rotate_half", "sin", "unsqueeze", "unsqueeze_dim"], "qwen3_moe/modeling_qwen3_moe.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "qwen3_moe/modeling_qwen3_moe.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value", "value_states"], "qwen3_moe/modeling_qwen3_moe.py:Qwen3MoeAttention.__init__": ["Linear", "ModelAttention", "ModelConfig", "ModelRMSNorm", "Module", "None", "True", "__init__", "attention_bias", "attention_dropout", "bias", "class", "config", "def", "eps", "getattr", "head_dim", "hidden_size", "int", "is_causal", "k_norm", "k_proj", "layer_idx", "nn", "num_attention_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "q_norm", "q_proj", "rms_norm_eps", "scaling", "self", "sliding_window", "super", "v_proj"], "qwen3_moe/modeling_qwen3_moe.py:Qwen3MoeAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "FlashAttentionKwargs", "LongTensor", "ModelAttention", "Module", "None", "Tensor", "Unpack", "_attn_implementation", "apply_rotary_pos_emb", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "cache_kwargs", "cache_position", "class", "config", "contiguous", "cos", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "head_dim", "hidden_shape", "hidden_states", "if", "input_shape", "is", "k_norm", "k_proj", "key_states", "kwargs", "layer_idx", "nn", "not", "o_proj", "past_key_values", "position_embeddings", "q_norm", "q_proj", "query_states", "reshape", "return", "scaling", "self", "shape", "sin", "sliding_window", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "view"], "qwen3_moe/modeling_qwen3_moe.py:Qwen3MoeMLP.__init__": ["ACT2FN", "False", "Linear", "ModelMLP", "Module", "None", "__init__", "act_fn", "bias", "class", "config", "def", "down_proj", "else", "gate_proj", "hidden_act", "hidden_size", "if", "intermediate_size", "is", "nn", "self", "super", "up_proj"], "qwen3_moe/modeling_qwen3_moe.py:Qwen3MoeMLP.forward": ["ModelMLP", "Module", "act_fn", "class", "def", "down_proj", "forward", "gate_proj", "nn", "return", "self", "up_proj", "x"], "qwen3_moe/modeling_qwen3_moe.py:Qwen3MoeExperts.__init__": ["ACT2FN", "ModelExperts", "Module", "Parameter", "__init__", "act_fn", "class", "config", "def", "down_proj", "empty", "gate_up_proj", "hidden_act", "hidden_dim", "hidden_size", "intermediate_dim", "moe_intermediate_size", "nn", "num_experts", "self", "super", "torch"], "qwen3_moe/modeling_qwen3_moe.py:Qwen3MoeExperts.forward": ["ModelExperts", "Module", "None", "Tensor", "act_fn", "chunk", "class", "continue", "current_hidden_states", "current_state", "def", "dim", "down_proj", "dtype", "expert_hit", "expert_idx", "expert_mask", "final_hidden_states", "for", "forward", "functional", "gate", "gate_up_proj", "greater", "hidden_states", "if", "in", "index_add_", "linear", "nn", "no_grad", "nonzero", "num_classes", "num_experts", "one_hot", "permute", "return", "self", "sum", "to", "token_idx", "top_k_index", "top_k_pos", "top_k_weights", "torch", "up", "where", "with", "zeros_like"], "qwen3_moe/modeling_qwen3_moe.py:Qwen3MoeTopKRouter.__init__": ["ModelTopKRouter", "Module", "Parameter", "__init__", "class", "config", "def", "hidden_dim", "hidden_size", "nn", "norm_topk_prob", "num_experts", "num_experts_per_tok", "self", "super", "top_k", "torch", "weight", "zeros"], "qwen3_moe/modeling_qwen3_moe.py:Qwen3MoeTopKRouter.forward": ["F", "ModelTopKRouter", "Module", "True", "class", "def", "dim", "dtype", "float", "forward", "functional", "hidden_dim", "hidden_states", "if", "keepdim", "linear", "nn", "norm_topk_prob", "reshape", "return", "router_indices", "router_logits", "router_scores", "router_top_value", "self", "softmax", "sum", "to", "top_k", "topk", "torch", "weight"], "qwen3_moe/modeling_qwen3_moe.py:Qwen3MoeSparseMoeBlock.__init__": ["ModelConfig", "ModelExperts", "ModelSparseMoeBlock", "ModelTopKRouter", "Module", "__init__", "class", "config", "def", "experts", "gate", "nn", "self", "super"], "qwen3_moe/modeling_qwen3_moe.py:Qwen3MoeSparseMoeBlock.forward": ["ModelSparseMoeBlock", "Module", "Tensor", "_", "batch_size", "class", "def", "experts", "final_hidden_states", "forward", "gate", "hidden_dim", "hidden_states", "hidden_states_reshaped", "nn", "reshape", "return", "routing_weights", "selected_experts", "self", "sequence_length", "shape", "torch", "tuple", "view"], "qwen3_moe/modeling_qwen3_moe.py:Qwen3MoeRMSNorm.__init__": ["ModelRMSNorm", "Module", "Parameter", "__init__", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "qwen3_moe/modeling_qwen3_moe.py:Qwen3MoeRMSNorm.forward": ["ModelRMSNorm", "Module", "True", "class", "def", "dtype", "float32", "forward", "hidden_states", "input_dtype", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "qwen3_moe/modeling_qwen3_moe.py:Qwen3MoeRMSNorm.extra_repr": ["ModelRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "variance_epsilon", "weight"], "qwen3_moe/modeling_qwen3_moe.py:Qwen3MoeDecoderLayer.__init__": ["GradientCheckpointingLayer", "ModelAttention", "ModelConfig", "ModelDecoderLayer", "ModelMLP", "ModelRMSNorm", "ModelSparseMoeBlock", "__init__", "and", "class", "config", "decoder_sparse_step", "def", "else", "eps", "hidden_size", "if", "in", "input_layernorm", "int", "intermediate_size", "layer_idx", "mlp", "mlp_only_layers", "not", "num_experts", "post_attention_layernorm", "rms_norm_eps", "self", "self_attn", "super"], "qwen3_moe/modeling_qwen3_moe.py:Qwen3MoeDecoderLayer.forward": ["Cache", "False", "GradientCheckpointingLayer", "LongTensor", "ModelDecoderLayer", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "attention_mask", "bool", "cache_position", "class", "def", "forward", "hidden_states", "input_layernorm", "kwargs", "mlp", "past_key_values", "position_embeddings", "position_ids", "post_attention_layernorm", "residual", "return", "self", "self_attn", "torch", "tuple", "use_cache"], "qwen3_moe/modeling_qwen3_moe.py:Qwen3MoePreTrainedModel._init_weights": ["ModelExperts", "ModelPreTrainedModel", "ModelTopKRouter", "PreTrainedModel", "_init_weights", "class", "config", "def", "down_proj", "elif", "gate_up_proj", "if", "init", "initializer_range", "isinstance", "mean", "module", "normal_", "self", "std", "super", "weight"], "qwen3_moe/modeling_qwen3_moe.py:Qwen3MoeRotaryEmbedding.__init__": ["Callable", "False", "ModelConfig", "ModelRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "qwen3_moe/modeling_qwen3_moe.py:Qwen3MoeRotaryEmbedding.compute_default_rope_parameters": ["ModelConfig", "ModelRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "qwen3_moe/modeling_qwen3_moe.py:Qwen3MoeRotaryEmbedding.forward": ["False", "ModelRotaryEmbedding", "Module", "None", "and", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "qwen3_moe/modeling_qwen3_moe.py:Qwen3MoeModel.__init__": ["Embedding", "False", "ModelConfig", "ModelDecoderLayer", "ModelModel", "ModelPreTrainedModel", "ModelRMSNorm", "ModelRotaryEmbedding", "ModuleList", "__init__", "class", "config", "def", "embed_tokens", "eps", "for", "gradient_checkpointing", "hidden_size", "in", "layer_idx", "layers", "nn", "norm", "num_hidden_layers", "pad_token_id", "padding_idx", "post_init", "range", "rms_norm_eps", "rotary_emb", "self", "super", "vocab_size"], "qwen3_moe/modeling_qwen3_moe.py:Qwen3MoeModel.forward": ["Cache", "DynamicCache", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "MoeModelOutputWithPast", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "and", "arange", "attention_mask", "bool", "cache_position", "causal_mask", "class", "config", "create_causal_mask", "create_sliding_window_causal_mask", "decoder_layer", "def", "device", "else", "embed_tokens", "exactly", "for", "forward", "get_seq_length", "hidden_states", "if", "in", "input_embeds", "input_ids", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layers", "mask_function", "must", "norm", "not", "num_hidden_layers", "of", "one", "or", "past_key_values", "past_seen_tokens", "position_embeddings", "position_ids", "raise", "return", "rotary_emb", "self", "shape", "sliding_window", "specify", "torch", "unsqueeze", "use_cache"], "qwen3_moe/modeling_qwen3_moe.py:load_balancing_loss_func": ["Model_balancing_loss_func", "None", "Tensor", "_", "attention_mask", "batch_size", "cat", "compute_device", "concatenated_gate_logits", "def", "device", "dim", "else", "expand", "expert_attention_mask", "expert_mask", "float", "for", "functional", "gate_logits", "if", "in", "int", "is", "isinstance", "layer_gate", "mean", "nn", "not", "num_experts", "num_hidden_layers", "one_hot", "or", "overall_loss", "r", "reshape", "return", "router_per_expert_attention_mask", "router_prob_per_expert", "routing_weights", "selected_experts", "sequence_length", "shape", "softmax", "sum", "to", "tokens_per_expert", "top_k", "topk", "torch", "tuple", "unsqueeze"], "qwen3_moe/modeling_qwen3_moe.py:Qwen3MoeForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "ModelForCausalLM", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "num_experts", "num_experts_per_tok", "post_init", "router_aux_loss_coef", "self", "super", "vocab_size"], "qwen3_moe/modeling_qwen3_moe.py:Qwen3MoeForCausalLM.forward": ["Cache", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForCausalLM", "ModelPreTrainedModel", "MoeCausalLMOutputWithPast", "MoeModelOutputWithPast", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "aux_loss", "bool", "cache_position", "class", "config", "def", "device", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "load_balancing_loss_func", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "num_experts", "num_experts_per_tok", "output_router_logits", "outputs", "past_key_values", "position_ids", "r", "return", "router_aux_loss_coef", "router_logits", "self", "slice", "slice_indices", "to", "torch", "use_cache", "vocab_size"], "qwen3_next/modeling_qwen3_next.py:Qwen3NextRMSNormGated.__init__": ["ModelRMSNormGated", "Module", "Parameter", "__init__", "class", "def", "eps", "hidden_size", "kwargs", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "qwen3_next/modeling_qwen3_next.py:Qwen3NextRMSNormGated.forward": ["F", "ModelRMSNormGated", "Module", "None", "True", "class", "def", "dtype", "float32", "forward", "gate", "hidden_states", "input_dtype", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "silu", "to", "torch", "variance", "variance_epsilon", "weight"], "qwen3_next/modeling_qwen3_next.py:Qwen3NextDynamicCache.__init__": ["ModelConfig", "ModelDynamicCache", "None", "_", "__init__", "class", "config", "conv_states", "def", "for", "full_attention", "i", "if", "in", "index", "key_cache", "last_linear_layer", "layer_types", "len", "linear_attention", "num_hidden_layers", "range", "recurrent_states", "self", "super", "transformer_layers", "value_cache"], "qwen3_next/modeling_qwen3_next.py:Qwen3NextDynamicCache.__len__": ["ModelDynamicCache", "__len__", "class", "def", "layer_types", "len", "return", "self"], "qwen3_next/modeling_qwen3_next.py:Qwen3NextDynamicCache.update": ["Any", "ModelDynamicCache", "None", "Tensor", "cache_kwargs", "cat", "class", "def", "dict", "dim", "else", "if", "int", "is", "key_cache", "key_states", "layer_idx", "return", "self", "str", "torch", "tuple", "update", "value_cache", "value_states"], "qwen3_next/modeling_qwen3_next.py:Qwen3NextDynamicCache.reorder_cache": ["LongTensor", "ModelDynamicCache", "None", "beam_idx", "class", "conv_states", "def", "device", "for", "if", "in", "index_select", "is", "key_cache", "layer_idx", "len", "not", "range", "recurrent_states", "reorder_cache", "self", "to", "torch", "value_cache"], "qwen3_next/modeling_qwen3_next.py:Qwen3NextDynamicCache.get_seq_length": ["ModelDynamicCache", "None", "class", "def", "else", "get_seq_length", "if", "in", "int", "is", "key_cache", "layer_idx", "len", "not", "or", "return", "self", "shape", "transformer_layers"], "qwen3_next/modeling_qwen3_next.py:Qwen3NextDynamicCache.get_mask_sizes": ["ModelDynamicCache", "Tensor", "cache_position", "class", "def", "get_mask_sizes", "get_seq_length", "int", "kv_length", "kv_offset", "layer_idx", "past_seen_tokens", "query_length", "return", "self", "shape", "torch", "tuple"], "qwen3_next/modeling_qwen3_next.py:Qwen3NextDynamicCache.has_previous_state": ["ModelDynamicCache", "None", "class", "conv_states", "def", "has_previous_state", "is", "last_linear_layer", "not", "return", "self"], "qwen3_next/modeling_qwen3_next.py:Qwen3NextRotaryEmbedding.__init__": ["Callable", "False", "ModelConfig", "ModelRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "qwen3_next/modeling_qwen3_next.py:Qwen3NextRotaryEmbedding.compute_default_rope_parameters": ["ModelConfig", "ModelRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "get", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "partial_rotary_factor", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "qwen3_next/modeling_qwen3_next.py:Qwen3NextRotaryEmbedding.forward": ["False", "ModelRotaryEmbedding", "Module", "None", "and", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "qwen3_next/modeling_qwen3_next.py:Qwen3NextRMSNorm.__init__": ["ModelRMSNorm", "Module", "Parameter", "__init__", "class", "def", "dim", "eps", "float", "int", "nn", "self", "super", "torch", "weight", "zeros"], "qwen3_next/modeling_qwen3_next.py:Qwen3NextRMSNorm._norm": ["ModelRMSNorm", "Module", "True", "_norm", "class", "def", "eps", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "torch", "x"], "qwen3_next/modeling_qwen3_next.py:Qwen3NextRMSNorm.forward": ["ModelRMSNorm", "Module", "_norm", "class", "def", "float", "forward", "nn", "output", "return", "self", "type_as", "weight", "x"], "qwen3_next/modeling_qwen3_next.py:Qwen3NextRMSNorm.extra_repr": ["ModelRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "weight"], "qwen3_next/modeling_qwen3_next.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "qwen3_next/modeling_qwen3_next.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cat", "cos", "def", "dim", "k", "k_embed", "k_pass", "k_rot", "q", "q_embed", "q_pass", "q_rot", "return", "rotary_dim", "rotate_half", "shape", "sin", "torch", "unsqueeze", "unsqueeze_dim"], "qwen3_next/modeling_qwen3_next.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "qwen3_next/modeling_qwen3_next.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value", "value_states"], "qwen3_next/modeling_qwen3_next.py:Qwen3NextAttention.__init__": ["Linear", "ModelAttention", "ModelConfig", "ModelRMSNorm", "Module", "True", "__init__", "attention_bias", "attention_dropout", "bias", "class", "config", "def", "eps", "getattr", "head_dim", "hidden_size", "int", "is_causal", "k_norm", "k_proj", "layer_idx", "nn", "num_attention_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "q_norm", "q_proj", "rms_norm_eps", "scaling", "self", "super", "v_proj"], "qwen3_next/modeling_qwen3_next.py:Qwen3NextAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "FlashAttentionKwargs", "LongTensor", "ModelAttention", "Module", "None", "Tensor", "Unpack", "_attn_implementation", "apply_rotary_pos_emb", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "cache_kwargs", "cache_position", "chunk", "class", "config", "contiguous", "cos", "def", "dim", "dropout", "eager", "eager_attention_forward", "else", "forward", "gate", "head_dim", "hidden_shape", "hidden_states", "if", "input_shape", "is", "k_norm", "k_proj", "key_states", "kwargs", "layer_idx", "nn", "not", "o_proj", "past_key_values", "position_embeddings", "q_norm", "q_proj", "query_states", "reshape", "return", "scaling", "self", "shape", "sigmoid", "sin", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "view"], "qwen3_next/modeling_qwen3_next.py:apply_mask_to_padding_states": ["Model_mask_to_padding_states", "None", "and", "attention_mask", "def", "dtype", "hidden_states", "if", "is", "not", "return", "shape", "to"], "qwen3_next/modeling_qwen3_next.py:torch_causal_conv1d_update": ["F", "Model", "Model_causal_conv1d_update", "None", "_", "activation", "bias", "cat", "conv1d", "conv_state", "copy_", "def", "dim", "dtype", "groups", "hidden_size", "hidden_states", "hidden_states_new", "out", "padding", "return", "seq_len", "shape", "silu", "state_len", "to", "unsqueeze", "weight"], "qwen3_next/modeling_qwen3_next.py:l2norm": ["FloatTensor", "Model", "True", "def", "dim", "eps", "float", "int", "inv_norm", "keepdim", "return", "rsqrt", "sum", "torch", "x"], "qwen3_next/modeling_qwen3_next.py:torch_chunk_gated_delta_rule": ["F", "False", "Model", "Model_chunk_gated_delta_rule", "None", "attn", "attn_inter", "batch_size", "beta", "bool", "chunk_size", "clone", "contiguous", "core_attn_out", "cumsum", "decay_mask", "def", "device", "diagonal", "dim", "dtype", "else", "eps", "exp", "eye", "float", "float32", "for", "g", "i", "if", "in", "initial_dtype", "initial_state", "is", "k_beta", "k_cumdecay", "k_head_dim", "k_i", "key", "l2norm", "last_recurrent_state", "mask", "masked_fill", "masked_fill_", "not", "num_heads", "ones", "output_final_state", "pad", "pad_size", "q_i", "query", "range", "reshape", "return", "row", "scale", "sequence_length", "shape", "sub", "sum", "to", "total_sequence_length", "transpose", "tril", "triu", "unsqueeze", "use_qk_l2norm_in_kernel", "v_beta", "v_head_dim", "v_i", "v_new", "v_prime", "value", "x", "zeros", "zeros_like"], "qwen3_next/modeling_qwen3_next.py:torch_recurrent_gated_delta_rule": ["False", "Model", "Model_recurrent_gated_delta_rule", "None", "batch_size", "beta", "beta_t", "contiguous", "core_attn_out", "def", "delta", "dim", "dtype", "else", "eps", "exp", "float32", "for", "g", "g_t", "i", "if", "in", "initial_dtype", "initial_state", "is", "k_head_dim", "k_t", "key", "kv_mem", "l2norm", "last_recurrent_state", "not", "num_heads", "output_final_state", "q_t", "query", "range", "return", "scale", "sequence_length", "shape", "sum", "to", "transpose", "unsqueeze", "use_qk_l2norm_in_kernel", "v_head_dim", "v_t", "value", "x", "zeros"], "qwen3_next/modeling_qwen3_next.py:Qwen3NextGatedDeltaNet.__init__": ["A", "ACT2FN", "AILab", "A_log", "Conv1d", "Dao", "Falling", "False", "FusedRMSNormGated", "Linear", "ModelConfig", "ModelGatedDeltaNet", "ModelRMSNormGated", "Module", "None", "Parameter", "The", "To", "__init__", "act", "activation", "attention", "available", "back", "because", "bias", "causal", "causal_conv1d_fn", "causal_conv1d_update", "chunk_gated_delta_rule", "class", "com", "config", "conv1d", "conv_dim", "conv_kernel_size", "cuda", "current_device", "def", "device", "dt_bias", "dtype", "else", "empty", "eps", "fast", "fla", "flash", "follow", "fused_recurrent_gated_delta_rule", "get_default_dtype", "github", "groups", "head_k_dim", "head_v_dim", "hidden_act", "hidden_size", "https", "if", "implementation", "in_channels", "in_proj_ba", "in_proj_qkvz", "install", "installed", "int", "is", "is_fast_path_available", "kernel_size", "key_dim", "layer_idx", "layer_norm_epsilon", "library", "linear", "linear_conv_kernel_dim", "linear_key_head_dim", "linear_num_key_heads", "linear_num_value_heads", "linear_value_head_dim", "log", "logger", "nn", "norm", "not", "num_k_heads", "num_v_heads", "of", "one", "ones", "or", "org", "out_channels", "out_proj", "padding", "path", "projection_size_ba", "projection_size_qkvz", "recurrent_gated_delta_rule", "required", "rms_norm_eps", "self", "super", "the", "to", "torch", "torch_causal_conv1d_update", "torch_chunk_gated_delta_rule", "torch_recurrent_gated_delta_rule", "uniform_", "value_dim", "warning_once"], "qwen3_next/modeling_qwen3_next.py:Qwen3NextGatedDeltaNet.fix_query_key_value_ordering": ["ModelGatedDeltaNet", "Module", "a", "b", "class", "def", "dim", "fix_query_key_value_ordering", "head_k_dim", "head_v_dim", "key", "mixed_ba", "mixed_qkvz", "new_tensor_shape_ba", "new_tensor_shape_qkvz", "nn", "num_k_heads", "num_v_heads", "query", "reshape", "return", "self", "size", "split", "split_arg_list_ba", "split_arg_list_qkvz", "torch", "value", "view", "z"], "qwen3_next/modeling_qwen3_next.py:Qwen3NextGatedDeltaNet.forward": ["A_log", "F", "LongTensor", "ModelDynamicCache", "ModelGatedDeltaNet", "Module", "None", "Tensor", "True", "_", "a", "activation", "and", "apply_mask_to_padding_states", "attention_mask", "b", "batch_size", "beta", "bias", "cache_params", "cache_position", "cat", "causal_conv1d_fn", "causal_conv1d_update", "chunk_gated_delta_rule", "class", "conv1d", "conv_kernel_size", "conv_state", "conv_states", "core_attn_out", "def", "dim", "dt_bias", "else", "exp", "fix_query_key_value_ordering", "float", "for", "forward", "g", "has_previous_state", "head_k_dim", "head_v_dim", "hidden_states", "if", "in", "in_proj_ba", "in_proj_qkvz", "initial_state", "is", "key", "key_dim", "last_recurrent_state", "layer_idx", "mixed_qkv", "nn", "norm", "not", "num_k_heads", "num_v_heads", "out_proj", "output", "output_final_state", "pad", "projected_states_ba", "projected_states_qkvz", "query", "recurrent_gated_delta_rule", "recurrent_state", "recurrent_states", "repeat_interleave", "reshape", "return", "self", "seq_idx", "seq_len", "shape", "sigmoid", "silu", "softplus", "split", "squeeze", "torch", "transpose", "use_precomputed_states", "use_qk_l2norm_in_kernel", "value", "value_dim", "weight", "x", "z", "z_shape_og"], "qwen3_next/modeling_qwen3_next.py:Qwen3NextMLP.__init__": ["ACT2FN", "False", "Linear", "ModelMLP", "Module", "None", "__init__", "act_fn", "bias", "class", "config", "def", "down_proj", "else", "gate_proj", "hidden_act", "hidden_size", "if", "intermediate_size", "is", "nn", "self", "super", "up_proj"], "qwen3_next/modeling_qwen3_next.py:Qwen3NextMLP.forward": ["ModelMLP", "Module", "act_fn", "class", "def", "down_proj", "forward", "gate_proj", "nn", "return", "self", "up_proj", "x"], "qwen3_next/modeling_qwen3_next.py:Qwen3NextExperts.__init__": ["ACT2FN", "ModelExperts", "Module", "Parameter", "__init__", "act_fn", "class", "config", "def", "down_proj", "empty", "gate_up_proj", "hidden_act", "hidden_dim", "hidden_size", "intermediate_dim", "moe_intermediate_size", "nn", "num_experts", "self", "super", "torch"], "qwen3_next/modeling_qwen3_next.py:Qwen3NextExperts.forward": ["ModelExperts", "Module", "None", "Tensor", "act_fn", "chunk", "class", "continue", "current_hidden_states", "current_state", "def", "dim", "down_proj", "dtype", "expert_hit", "expert_idx", "expert_mask", "final_hidden_states", "for", "forward", "functional", "gate", "gate_up_proj", "greater", "hidden_states", "if", "in", "index_add_", "linear", "nn", "no_grad", "nonzero", "num_classes", "num_experts", "one_hot", "permute", "return", "self", "sum", "to", "token_idx", "top_k_index", "top_k_pos", "top_k_weights", "torch", "up", "where", "with", "zeros_like"], "qwen3_next/modeling_qwen3_next.py:Qwen3NextTopKRouter.__init__": ["ModelTopKRouter", "Module", "Parameter", "__init__", "class", "config", "def", "hidden_dim", "hidden_size", "nn", "norm_topk_prob", "num_experts", "num_experts_per_tok", "self", "super", "top_k", "torch", "weight", "zeros"], "qwen3_next/modeling_qwen3_next.py:Qwen3NextTopKRouter.forward": ["F", "ModelTopKRouter", "Module", "True", "class", "def", "dim", "dtype", "float", "forward", "functional", "hidden_dim", "hidden_states", "if", "keepdim", "linear", "nn", "norm_topk_prob", "reshape", "return", "router_indices", "router_logits", "router_scores", "router_top_value", "self", "softmax", "sum", "to", "top_k", "topk", "torch", "weight"], "qwen3_next/modeling_qwen3_next.py:Qwen3NextSparseMoeBlock.__init__": ["False", "Linear", "ModelExperts", "ModelMLP", "ModelSparseMoeBlock", "ModelTopKRouter", "Module", "__init__", "bias", "class", "config", "def", "experts", "gate", "hidden_size", "intermediate_size", "nn", "self", "shared_expert", "shared_expert_gate", "shared_expert_intermediate_size", "super", "torch"], "qwen3_next/modeling_qwen3_next.py:Qwen3NextSparseMoeBlock.forward": ["F", "ModelSparseMoeBlock", "Module", "Tensor", "_", "batch_size", "class", "def", "expert_output", "experts", "forward", "gate", "hidden_dim", "hidden_states", "hidden_states_reshaped", "nn", "reshape", "return", "routing_weights", "selected_experts", "self", "sequence_length", "shape", "shared_expert", "shared_expert_gate", "shared_expert_output", "sigmoid", "torch", "tuple", "view"], "qwen3_next/modeling_qwen3_next.py:Qwen3NextDecoderLayer.__init__": ["GradientCheckpointingLayer", "ModelAttention", "ModelConfig", "ModelDecoderLayer", "ModelGatedDeltaNet", "ModelMLP", "ModelRMSNorm", "ModelSparseMoeBlock", "__init__", "and", "class", "config", "decoder_sparse_step", "def", "elif", "else", "eps", "full_attention", "hidden_size", "if", "in", "input_layernorm", "int", "intermediate_size", "layer_idx", "layer_type", "layer_types", "linear_attention", "linear_attn", "mlp", "mlp_only_layers", "not", "num_experts", "post_attention_layernorm", "rms_norm_eps", "self", "self_attn", "super"], "qwen3_next/modeling_qwen3_next.py:Qwen3NextDecoderLayer.forward": ["Cache", "FlashAttentionKwargs", "FloatTensor", "GradientCheckpointingLayer", "LongTensor", "ModelDecoderLayer", "None", "Tensor", "Unpack", "_", "attention_mask", "cache_params", "cache_position", "class", "def", "elif", "forward", "full_attention", "hidden_states", "if", "input_layernorm", "isinstance", "kwargs", "layer_type", "linear_attention", "linear_attn", "mlp", "past_key_values", "position_embeddings", "position_ids", "post_attention_layernorm", "residual", "return", "self", "self_attn", "torch", "tuple"], "qwen3_next/modeling_qwen3_next.py:Qwen3NextPreTrainedModel._init_weights": ["A_log", "ModelExperts", "ModelGatedDeltaNet", "ModelPreTrainedModel", "ModelRMSNorm", "ModelSparseMoeBlock", "PreTrainedModel", "_init_weights", "class", "config", "copy_", "def", "down_proj", "dt_bias", "elif", "empty_like", "gate", "gate_up_proj", "if", "init", "initializer_range", "isinstance", "log_", "mean", "module", "normal_", "ones_", "self", "std", "super", "torch", "uniform_", "weight", "zeros_"], "qwen3_next/modeling_qwen3_next.py:Qwen3NextModel.__init__": ["Embedding", "False", "ModelConfig", "ModelDecoderLayer", "ModelModel", "ModelPreTrainedModel", "ModelRMSNorm", "ModelRotaryEmbedding", "ModuleList", "__init__", "class", "config", "def", "embed_tokens", "eps", "for", "gradient_checkpointing", "hidden_size", "in", "layer_idx", "layers", "nn", "norm", "num_hidden_layers", "pad_token_id", "post_init", "range", "rms_norm_eps", "rotary_emb", "self", "super", "vocab_size"], "qwen3_next/modeling_qwen3_next.py:Qwen3NextModel.forward": ["Cache", "FloatTensor", "LongTensor", "ModelDynamicCache", "ModelModel", "ModelPreTrainedModel", "MoeModelOutputWithPast", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "_update_linear_attn_mask", "and", "arange", "attention_mask", "bool", "cache_position", "causal_mask", "class", "config", "create_causal_mask", "decoder_layer", "def", "device", "else", "embed_tokens", "exactly", "for", "forward", "get_seq_length", "hidden_states", "if", "in", "input_embeds", "input_ids", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layer_mask", "layer_type", "layers", "linear_attention", "linear_attn_mask", "must", "norm", "not", "num_hidden_layers", "of", "one", "or", "past_key_values", "past_seen_tokens", "position_embeddings", "position_ids", "raise", "return", "rotary_emb", "self", "shape", "specify", "torch", "unsqueeze", "use_cache"], "qwen3_next/modeling_qwen3_next.py:Qwen3NextModel._update_linear_attn_mask": ["ModelModel", "ModelPreTrainedModel", "None", "_update_linear_attn_mask", "all", "and", "attention_mask", "cache_position", "class", "def", "if", "is", "linear_attn_mask", "not", "or", "return", "self", "torch"], "qwen3_next/modeling_qwen3_next.py:load_balancing_loss_func": ["Model_balancing_loss_func", "None", "Tensor", "_", "attention_mask", "batch_size", "cat", "compute_device", "concatenated_gate_logits", "def", "device", "dim", "else", "expand", "expert_attention_mask", "expert_mask", "float", "for", "functional", "gate_logits", "if", "in", "int", "is", "isinstance", "layer_gate", "mean", "nn", "not", "num_experts", "num_hidden_layers", "one_hot", "or", "overall_loss", "r", "reshape", "return", "router_per_expert_attention_mask", "router_prob_per_expert", "routing_weights", "selected_experts", "sequence_length", "shape", "softmax", "sum", "to", "tokens_per_expert", "top_k", "topk", "torch", "tuple", "unsqueeze"], "qwen3_next/modeling_qwen3_next.py:Qwen3NextForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "ModelForCausalLM", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "num_experts", "num_experts_per_tok", "post_init", "router_aux_loss_coef", "self", "super", "vocab_size"], "qwen3_next/modeling_qwen3_next.py:Qwen3NextForCausalLM.forward": ["FloatTensor", "GenerationMixin", "LongTensor", "ModelDynamicCache", "ModelForCausalLM", "ModelPreTrainedModel", "MoeCausalLMOutputWithPast", "MoeModelOutputWithPast", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "aux_loss", "bool", "cache_position", "class", "config", "def", "device", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "load_balancing_loss_func", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "num_experts", "num_experts_per_tok", "output_router_logits", "outputs", "past_key_values", "position_ids", "r", "return", "router_aux_loss_coef", "router_logits", "self", "slice", "slice_indices", "to", "torch", "use_cache", "vocab_size"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:SinusoidsPositionEmbedding.__init__": ["False", "ModelPositionEmbedding", "Module", "ValueError", "__init__", "arange", "cat", "channels", "class", "cos", "def", "dim", "even", "exp", "float", "if", "input", "inv_timescales", "length", "log", "log_timescale_increment", "max_timescale", "needs", "newaxis", "nn", "np", "persistent", "positional_embedding", "raise", "register_buffer", "scaled_time", "self", "sin", "super", "torch"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:SinusoidsPositionEmbedding.forward": ["ModelPositionEmbedding", "Module", "class", "def", "forward", "int", "nn", "positional_embedding", "return", "self", "seqlen"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoePreTrainedModel._init_weights": ["ModelCode2Wav", "ModelPreTrainedModel", "ModelThinkerTextSparseMoeBlock", "ModelVisionRotaryEmbedding", "PreTrainedModel", "SinusoidsPositionEmbedding", "_init_weights", "arange", "cat", "channels", "class", "code_offset", "codebook_size", "config", "copy_", "cos", "def", "dim", "down_proj", "dtype", "elif", "exp", "experts", "float", "gate", "gate_up_proj", "if", "init", "initializer_range", "inv_freq", "inv_timescales", "isinstance", "length", "log", "log_timescale_increment", "max_timescale", "mean", "module", "newaxis", "normal_", "np", "num_quantizers", "positional_embedding", "scaled_time", "self", "sin", "std", "super", "theta", "torch", "view", "weight"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:_get_feat_extract_output_lengths": ["_get_feat_extract_output_lengths", "def", "feat_lengths", "input_lengths", "input_lengths_leave", "output_lengths", "return"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoePreTrainedModelForConditionalGeneration._prepare_4d_causal_attention_mask_with_cache_position": ["ModelPreTrainedModel", "ModelPreTrainedModelForConditionalGeneration", "None", "Tensor", "_prepare_4d_causal_attention_mask_with_cache_position", "and", "arange", "attention_mask", "batch_size", "cache_position", "causal_mask", "class", "clone", "def", "device", "diagonal", "dim", "dtype", "else", "expand", "fill_value", "float", "full", "if", "int", "is", "mask_length", "masked_fill", "min_dtype", "not", "padding_mask", "reshape", "return", "self", "sequence_length", "shape", "target_length", "torch", "triu"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoePreTrainedModelForConditionalGeneration.get_llm_pos_ids_for_vision": ["ModelPreTrainedModel", "ModelPreTrainedModelForConditionalGeneration", "Tensor", "_llm_pos_ids", "append", "arange", "cat", "class", "def", "dim", "expand", "flatten", "float", "get_llm_pos_ids_for_vision", "grid_hs", "grid_ws", "h_index", "int", "len", "list", "llm_grid_h", "llm_grid_w", "llm_pos_ids", "llm_pos_ids_list", "return", "self", "spatial_merge_size", "stack", "start_idx", "t_index", "torch", "view", "vision_idx", "w_index"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoePreTrainedModelForConditionalGeneration.get_chunked_index": ["ModelPreTrainedModel", "ModelPreTrainedModelForConditionalGeneration", "Tensor", "_iter", "class", "current_chunk", "def", "get_chunked_index", "i", "if", "int", "len", "list", "remove_index", "return", "self", "start_idx", "token_indices", "tokens_per_chunk", "torch", "tuple", "while", "yield"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoePreTrainedModelForConditionalGeneration.get_rope_index": ["False", "LongTensor", "ModelPreTrainedModel", "ModelPreTrainedModelForConditionalGeneration", "None", "Tensor", "True", "_", "_get_feat_extract_output_lengths", "and", "append", "arange", "argwhere", "attention_mask", "audio_data_index", "audio_idx", "audio_len", "audio_llm_pos_ids", "audio_nums", "audio_seqlens", "audio_start_token_id", "audio_token_id", "bool", "bos_len", "cat", "class", "config", "cpu", "cumsum", "def", "device", "dim", "dtype", "ed_audio_start", "ed_vision_start", "elif", "else", "enumerate", "eos_len", "expand", "float", "for", "get_llm_pos_ids_for_vision", "get_rope_index", "grid_hs", "grid_t", "grid_ws", "i", "if", "image_grid_thw", "image_idx", "image_len", "image_nums", "image_token_id", "in", "index", "input_ids", "input_tokens", "int", "is", "item", "keepdim", "len", "list", "llm_pos_ids", "llm_pos_ids_list", "llm_positions", "masked_fill_", "max", "max_position_ids", "min", "min_ed", "mrope_position_deltas", "multimodal_nums", "not", "or", "position_id_per_seconds", "position_ids", "prod", "range", "remain_audios", "remain_images", "remain_videos", "reshape", "return", "second_per_grids", "self", "shape", "spatial_merge_size", "squeeze", "st", "st_idx", "sum", "t_index", "tensor", "text_len", "to", "tolist", "torch", "total_input_ids", "tuple", "unsqueeze", "use_audio_in_video", "video_data_index", "video_grid_thw", "video_idx", "video_len", "video_llm_pos_ids", "video_nums", "video_token_id", "view", "vision_start_indices", "vision_start_token_id", "vision_tokens", "while", "zeros"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value", "value_states"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeAudioAttention.__init__": ["False", "Linear", "ModelAudioAttention", "Module", "True", "ValueError", "__init__", "and", "attention_dropout", "be", "bias", "by", "class", "config", "d_model", "def", "divisible", "dropout", "embed_dim", "encoder_attention_heads", "f", "got", "head_dim", "if", "is_causal", "is_decoder", "k_proj", "must", "nn", "num_heads", "num_key_value_groups", "out_proj", "q_proj", "raise", "scaling", "self", "super", "v_proj"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeAudioAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "False", "ModelAudioAttention", "Module", "None", "Tensor", "_", "_attn_implementation", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "class", "config", "contiguous", "cu_seq_lens_k", "cu_seq_lens_q", "cu_seqlens", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "hidden_states", "if", "is_causal", "k_proj", "key_states", "kwargs", "max", "max_length_k", "max_length_q", "max_seqlen", "nn", "not", "num_heads", "out_proj", "q_proj", "query_states", "reshape", "return", "scaling", "self", "seq_length", "size", "torch", "training", "transpose", "tuple", "unsqueeze", "v_proj", "value_states"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeAudioEncoderLayer.__init__": ["ACT2FN", "GradientCheckpointingLayer", "LayerNorm", "Linear", "ModelAudioAttention", "ModelAudioEncoderConfig", "ModelAudioEncoderLayer", "__init__", "activation_dropout", "activation_fn", "activation_function", "class", "config", "d_model", "def", "dropout", "embed_dim", "encoder_ffn_dim", "fc1", "fc2", "final_layer_norm", "nn", "self", "self_attn", "self_attn_layer_norm", "super"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeAudioEncoderLayer.forward": ["GradientCheckpointingLayer", "ModelAudioEncoderLayer", "None", "Tensor", "activation_fn", "attention_mask", "clamp", "clamp_value", "class", "cu_seqlens", "def", "dtype", "fc1", "fc2", "final_layer_norm", "finfo", "float16", "forward", "hidden_states", "if", "kwargs", "max", "min", "outputs", "residual", "return", "self", "self_attn", "self_attn_layer_norm", "torch"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeAudioEncoder.__init__": ["ACT2FN", "Conv2d", "False", "LayerNorm", "Linear", "ModelAudioEncoder", "ModelAudioEncoderConfig", "ModelAudioEncoderLayer", "ModelPreTrainedModel", "ModuleList", "SinusoidsPositionEmbedding", "_", "__init__", "act", "activation_function", "bias", "class", "config", "conv2d1", "conv2d2", "conv2d3", "conv_chunksize", "conv_out", "d_model", "def", "downsample_hidden_size", "dropout", "else", "embed_dim", "embed_scale", "encoder_layers", "for", "gradient_checkpointing", "if", "in", "layers", "ln_post", "math", "max_source_positions", "n_window", "n_window_infer", "nn", "num_mel_bins", "output_dim", "padding", "positional_embedding", "post_init", "proj1", "proj2", "range", "scale_embedding", "self", "sqrt", "super"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeAudioEncoder._freeze_parameters": ["False", "ModelAudioEncoder", "ModelPreTrainedModel", "_freeze_parameters", "_requires_grad", "class", "def", "for", "in", "param", "parameters", "requires_grad", "self"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeAudioEncoder.get_input_embeddings": ["ModelAudioEncoder", "ModelPreTrainedModel", "Module", "class", "conv2d1", "def", "get_input_embeddings", "nn", "return", "self"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeAudioEncoder.set_input_embeddings": ["ModelAudioEncoder", "ModelPreTrainedModel", "class", "conv2d1", "def", "self", "set_input_embeddings", "value"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeAudioEncoder._prepare_attention_mask": ["ModelAudioEncoder", "ModelPreTrainedModel", "None", "Tensor", "_attn_implementation", "_prepare_attention_mask", "attention_mask", "class", "config", "cu_seqlens", "def", "device", "dtype", "finfo", "flash_attention_2", "for", "full", "i", "if", "in", "inputs_tensor", "len", "min", "range", "return", "self", "seq_length", "shape", "torch"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeAudioEncoder.forward": ["BaseModelOutput", "F", "ModelAudioEncoder", "ModelPreTrainedModel", "None", "T", "True", "_get_feat_extract_output_lengths", "act", "aftercnn_lens", "append", "b", "batch_first", "bool", "c", "cat", "ceil", "chunk", "chunk_lengths", "chunk_list", "chunk_num", "class", "cnn_len", "contiguous", "conv2d1", "conv2d2", "conv2d3", "conv_chunksize", "conv_out", "cu_chunk_lens", "cu_seqlens", "cumsum", "def", "device", "dim", "dtype", "encoder_layer", "f", "feature_lens", "feature_lens_after_cnn", "for", "forward", "gelu", "hidden_states", "if", "in", "input_features", "int32", "kwargs", "last_hidden_state", "layer_outputs", "layers", "length", "ln_post", "long", "n_window", "n_window_infer", "nn", "ones", "pad", "pad_sequence", "padded_embed", "padded_embeds", "padded_feature", "padded_mask_after_cnn", "permute", "positional_embedding", "proj1", "proj2", "r", "remainder", "return", "rnn", "self", "shape", "size", "split", "sum", "t", "tail_chunk_index", "tensor", "to", "tolist", "torch", "transpose", "unsqueeze", "utils", "value", "view", "window_aftercnn"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeAudioEncoder.padded_and_mask_function": ["ModelAudioEncoder", "ModelPreTrainedModel", "batch_mask", "batch_mask_after_cnn", "bool", "class", "def", "device", "dim", "dtype", "enumerate", "feature_lens_after_cnn", "fill_value", "for", "full", "i", "in", "len", "length", "long", "max", "max_len", "max_len_after_cnn", "padded_and_mask_function", "padded_tensor", "padding_side", "padding_value", "return", "right", "self", "shape", "size", "tensor_len", "tensor_list", "torch", "unsqueeze", "zeros"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeAudioEncoder._get_feat_extract_output_lengths": ["LongTensor", "ModelAudioEncoder", "ModelPreTrainedModel", "_get_feat_extract_output_lengths", "class", "def", "input_lengths", "output_lengths", "return", "self", "torch"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:apply_rotary_pos_emb_vision": ["Model_rotary_pos_emb_vision", "Tensor", "cos", "def", "dtype", "float", "k", "k_embed", "orig_k_dtype", "orig_q_dtype", "q", "q_embed", "return", "rotate_half", "sin", "to", "torch", "tuple", "unsqueeze"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeVisionAttention.__init__": ["False", "Linear", "ModelVisionAttention", "ModelVisionEncoderConfig", "Module", "None", "True", "__init__", "attention_dropout", "bias", "class", "config", "def", "dim", "head_dim", "hidden_size", "is_causal", "nn", "num_heads", "num_key_value_groups", "proj", "qkv", "scaling", "self", "super"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeVisionAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "False", "ModelVisionAttention", "Module", "None", "Tensor", "_", "_attn_implementation", "apply_rotary_pos_emb_vision", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_outputs", "cat", "class", "config", "contiguous", "cos", "cu_seq_lens_k", "cu_seq_lens_q", "cu_seqlens", "def", "dim", "dropout", "eager", "eager_attention_forward", "else", "flash", "for", "forward", "hidden_states", "if", "in", "is_causal", "k", "key_states", "kwargs", "lengths", "max", "max_length_k", "max_length_q", "max_seqlen", "nn", "not", "num_heads", "permute", "position_embeddings", "proj", "q", "qkv", "query_states", "reshape", "return", "rotary_pos_emb", "scaling", "self", "seq_length", "shape", "sin", "split", "splits", "tensor", "tolist", "torch", "training", "transpose", "tuple", "unbind", "unsqueeze", "v", "value_states", "zip"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeVisionPatchMerger.__init__": ["False", "GELU", "LayerNorm", "Linear", "ModelVisionEncoderConfig", "ModelVisionPatchMerger", "Module", "ModuleList", "None", "__init__", "class", "config", "def", "else", "eps", "hidden_size", "if", "ln_q", "mlp", "nn", "out_hidden_size", "self", "spatial_merge_size", "super", "use_postshuffle_norm"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeVisionPatchMerger.forward": ["ModelVisionPatchMerger", "Module", "Tensor", "class", "def", "else", "for", "forward", "hidden", "hidden_size", "if", "in", "layer", "ln_q", "mlp", "nn", "return", "self", "torch", "use_postshuffle_norm", "view"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeVisionRotaryEmbedding.__init__": ["False", "ModelVisionRotaryEmbedding", "Module", "None", "__init__", "arange", "class", "def", "dim", "dtype", "float", "int", "inv_freq", "nn", "persistent", "register_buffer", "self", "super", "theta", "torch"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeVisionRotaryEmbedding.forward": ["ModelVisionRotaryEmbedding", "Module", "Tensor", "arange", "class", "def", "device", "dtype", "forward", "freqs", "int", "inv_freq", "nn", "outer", "return", "self", "seq", "seqlen", "torch"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeVisionMLP.__init__": ["ACT2FN", "Linear", "ModelVisionMLP", "Module", "True", "__init__", "act_fn", "bias", "class", "config", "def", "hidden_act", "hidden_size", "intermediate_size", "linear_fc1", "linear_fc2", "nn", "self", "super"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeVisionMLP.forward": ["ModelVisionMLP", "Module", "act_fn", "class", "def", "forward", "hidden_state", "linear_fc1", "linear_fc2", "nn", "return", "self"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeVisionPatchEmbed.__init__": ["Conv3d", "ModelVisionPatchEmbed", "Module", "None", "True", "__init__", "bias", "class", "config", "def", "embed_dim", "hidden_size", "in_channels", "kernel_size", "nn", "patch_size", "proj", "self", "stride", "super", "temporal_patch_size"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeVisionPatchEmbed.forward": ["ModelVisionPatchEmbed", "Module", "Tensor", "class", "def", "dtype", "embed_dim", "forward", "hidden_states", "in_channels", "nn", "patch_size", "proj", "return", "self", "target_dtype", "temporal_patch_size", "to", "torch", "view", "weight"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeVisionBlock.__init__": ["GradientCheckpointingLayer", "LayerNorm", "ModelVisionAttention", "ModelVisionBlock", "ModelVisionMLP", "None", "__init__", "attn", "attn_implementation", "class", "config", "def", "eps", "hidden_size", "mlp", "nn", "norm1", "norm2", "sdpa", "self", "str", "super"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeVisionBlock.forward": ["GradientCheckpointingLayer", "ModelVisionBlock", "None", "Tensor", "attn", "class", "cu_seqlens", "def", "forward", "hidden_states", "kwargs", "mlp", "norm1", "norm2", "position_embeddings", "return", "rotary_pos_emb", "self", "torch", "tuple"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeVisionEncoder.__init__": ["Embedding", "False", "ModelPreTrainedModel", "ModelVisionBlock", "ModelVisionEncoder", "ModelVisionPatchEmbed", "ModelVisionPatchMerger", "ModelVisionRotaryEmbedding", "ModuleList", "None", "True", "_", "__init__", "blocks", "class", "config", "deepstack_visual_indexes", "def", "depth", "for", "gradient_checkpointing", "head_dim", "hidden_size", "in", "inputs", "int", "kwargs", "len", "merger", "merger_list", "nn", "num_grid_per_side", "num_heads", "num_position_embeddings", "patch_embed", "patch_size", "pos_embed", "post_init", "range", "rotary_pos_emb", "self", "spatial_merge_size", "spatial_merge_unit", "super", "use_postshuffle_norm"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeVisionEncoder.rot_pos_emb": ["ModelPreTrainedModel", "ModelVisionEncoder", "None", "Tensor", "arange", "block_cols", "block_rows", "class", "col_idx", "coords", "def", "device", "dim", "dtype", "embeddings", "empty", "expand", "flatten", "for", "freq_table", "grid_thw", "height", "if", "in", "int", "intra_col", "intra_row", "item", "long", "max", "max_hw", "merge_size", "merged_h", "merged_w", "num_frames", "num_tokens", "offset", "pos_ids", "prod", "repeat", "reshape", "return", "rot_pos_emb", "rotary_pos_emb", "row_idx", "self", "shape", "spatial_merge_size", "stack", "sum", "torch", "total_tokens", "width"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeVisionEncoder.fast_pos_embed_interpolate": ["ModelPreTrainedModel", "ModelVisionEncoder", "None", "T", "_", "append", "base_h", "base_h_ceil", "cat", "class", "clip", "config", "def", "device", "dh", "dtype", "dw", "extend", "fast_pos_embed_interpolate", "flatten", "for", "grid_hs", "grid_thw", "grid_ts", "grid_ws", "h", "h_idxs", "h_idxs_ceil", "h_idxs_floor", "i", "idx_list", "idx_tensor", "in", "indices", "int", "linspace", "long", "max", "merge_size", "num_grid_per_side", "patch_pos_embeds", "patch_pos_embeds_permute", "permute", "pos_embed", "pos_embeds", "range", "repeat", "return", "self", "spatial_merge_size", "split", "t", "tensor", "to", "tolist", "torch", "view", "w", "w_idxs", "w_idxs_ceil", "w_idxs_floor", "weight", "weight_list", "weight_tensor", "weights", "zip"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeVisionEncoder.forward": ["F", "ModelPreTrainedModel", "ModelVisionEncoder", "Tensor", "_", "append", "blk", "blocks", "cat", "class", "cos", "cu_seqlens", "cumsum", "deepstack_feature", "deepstack_feature_lists", "deepstack_merger_list", "deepstack_visual_indexes", "def", "dim", "dtype", "else", "emb", "enumerate", "fast_pos_embed_interpolate", "for", "forward", "grid_thw", "hidden_states", "if", "in", "index", "int32", "is_tracing", "jit", "kwargs", "layer_num", "merger", "pad", "patch_embed", "pos_embeds", "position_embeddings", "repeat_interleave", "reshape", "return", "rot_pos_emb", "rotary_pos_emb", "self", "seq_len", "sin", "size", "torch", "value"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeVisionEncoder.deepstack_merger_list": ["ModelPreTrainedModel", "ModelVisionEncoder", "class", "deepstack_merger_list", "def", "merger_list", "return", "self"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeThinkerTextRotaryEmbedding.__init__": ["Callable", "False", "ModelTextConfig", "ModelThinkerTextRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "get", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "mrope_section", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeThinkerTextRotaryEmbedding.compute_default_rope_parameters": ["ModelTextConfig", "ModelThinkerTextRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeThinkerTextRotaryEmbedding.forward": ["False", "ModelThinkerTextRotaryEmbedding", "Module", "None", "and", "apply_interleaved_mrope", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "mrope_section", "ndim", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeThinkerTextRotaryEmbedding.apply_interleaved_mrope": ["ModelThinkerTextRotaryEmbedding", "Module", "apply_interleaved_mrope", "class", "def", "dim", "enumerate", "for", "freqs", "freqs_t", "idx", "in", "length", "mrope_section", "nn", "offset", "return", "self", "slice", "start"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeThinkerTextExperts.__init__": ["ACT2FN", "ModelThinkerConfig", "ModelThinkerTextExperts", "Module", "Parameter", "__init__", "act_fn", "class", "config", "def", "down_proj", "empty", "gate_up_proj", "hidden_act", "hidden_dim", "hidden_size", "intermediate_dim", "moe_intermediate_size", "nn", "num_experts", "self", "super", "torch"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeThinkerTextExperts.forward": ["ModelThinkerTextExperts", "Module", "None", "Tensor", "act_fn", "chunk", "class", "continue", "current_hidden_states", "current_state", "def", "dim", "down_proj", "dtype", "expert_hit", "expert_idx", "expert_mask", "final_hidden_states", "for", "forward", "functional", "gate", "gate_up_proj", "greater", "hidden_states", "if", "in", "index_add_", "linear", "nn", "no_grad", "nonzero", "num_classes", "num_experts", "one_hot", "permute", "return", "self", "sum", "to", "token_idx", "top_k_index", "top_k_pos", "top_k_weights", "torch", "up", "where", "with", "zeros_like"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeThinkerTextTopKRouter.__init__": ["ModelThinkerTextTopKRouter", "Module", "Parameter", "__init__", "class", "config", "def", "hidden_dim", "hidden_size", "nn", "norm_topk_prob", "num_experts", "num_experts_per_tok", "self", "super", "top_k", "torch", "weight", "zeros"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeThinkerTextTopKRouter.forward": ["F", "ModelThinkerTextTopKRouter", "Module", "True", "class", "def", "dim", "dtype", "float", "forward", "functional", "hidden_dim", "hidden_states", "if", "keepdim", "linear", "nn", "norm_topk_prob", "reshape", "return", "router_indices", "router_logits", "router_scores", "router_top_value", "self", "softmax", "sum", "to", "top_k", "topk", "torch", "weight"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeThinkerTextSparseMoeBlock.__init__": ["ModelThinkerConfig", "ModelThinkerTextExperts", "ModelThinkerTextSparseMoeBlock", "ModelThinkerTextTopKRouter", "Module", "__init__", "class", "config", "def", "experts", "gate", "nn", "self", "super"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeThinkerTextSparseMoeBlock.forward": ["ModelThinkerTextSparseMoeBlock", "Module", "Tensor", "_", "batch_size", "class", "def", "experts", "final_hidden_states", "forward", "gate", "hidden_dim", "hidden_states", "hidden_states_reshaped", "nn", "reshape", "return", "routing_weights", "selected_experts", "self", "sequence_length", "shape", "torch", "tuple", "view"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeThinkerTextRMSNorm.__init__": ["ModelThinkerTextRMSNorm", "Module", "Parameter", "__init__", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeThinkerTextRMSNorm.forward": ["ModelThinkerTextRMSNorm", "Module", "True", "class", "def", "dtype", "float32", "forward", "hidden_states", "input_dtype", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeThinkerTextRMSNorm.extra_repr": ["ModelThinkerTextRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "variance_epsilon", "weight"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cos", "def", "k", "k_embed", "q", "q_embed", "return", "rotate_half", "sin", "unsqueeze", "unsqueeze_dim"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeThinkerTextAttention.__init__": ["Linear", "ModelThinkerTextAttention", "ModelThinkerTextRMSNorm", "Module", "None", "True", "__init__", "attention_bias", "attention_dropout", "bias", "class", "config", "def", "eps", "getattr", "head_dim", "hidden_size", "is_causal", "k_norm", "k_proj", "layer_idx", "nn", "num_attention_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "q_norm", "q_proj", "rms_norm_eps", "scaling", "self", "sliding_window", "super", "v_proj"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeThinkerTextAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "FlashAttentionKwargs", "LongTensor", "ModelThinkerTextAttention", "Module", "None", "Tensor", "Unpack", "_attn_implementation", "apply_rotary_pos_emb", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "cache_kwargs", "cache_position", "class", "config", "contiguous", "cos", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "head_dim", "hidden_shape", "hidden_states", "if", "input_shape", "is", "k_norm", "k_proj", "key_states", "kwargs", "layer_idx", "nn", "not", "o_proj", "past_key_values", "position_embeddings", "q_norm", "q_proj", "query_states", "reshape", "return", "scaling", "self", "shape", "sin", "sliding_window", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "view"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeThinkerTextMLP.__init__": ["ACT2FN", "False", "Linear", "ModelThinkerTextMLP", "Module", "None", "__init__", "act_fn", "bias", "class", "config", "def", "down_proj", "else", "gate_proj", "hidden_act", "hidden_size", "if", "intermediate_size", "is", "nn", "self", "super", "up_proj"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeThinkerTextMLP.forward": ["ModelThinkerTextMLP", "Module", "act_fn", "class", "def", "down_proj", "forward", "gate_proj", "nn", "return", "self", "up_proj", "x"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeThinkerTextDecoderLayer.__init__": ["GradientCheckpointingLayer", "ModelThinkerTextAttention", "ModelThinkerTextDecoderLayer", "ModelThinkerTextMLP", "ModelThinkerTextRMSNorm", "ModelThinkerTextSparseMoeBlock", "__init__", "and", "class", "config", "decoder_sparse_step", "def", "else", "eps", "hidden_size", "if", "in", "input_layernorm", "intermediate_size", "layer_idx", "mlp", "mlp_only_layers", "not", "num_experts", "post_attention_layernorm", "rms_norm_eps", "self", "self_attn", "super"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeThinkerTextDecoderLayer.forward": ["Cache", "False", "GradientCheckpointingLayer", "LongTensor", "ModelThinkerTextDecoderLayer", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "attention_mask", "bool", "cache_position", "class", "def", "forward", "hidden_states", "input_layernorm", "kwargs", "mlp", "past_key_values", "position_embeddings", "position_ids", "post_attention_layernorm", "residual", "return", "self", "self_attn", "torch", "tuple", "use_cache"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeThinkerTextPreTrainedModel._init_weights": ["ModelThinkerTextExperts", "ModelThinkerTextPreTrainedModel", "ModelThinkerTextTopKRouter", "PreTrainedModel", "_init_weights", "class", "config", "def", "down_proj", "elif", "gate_up_proj", "if", "init", "initializer_range", "isinstance", "mean", "module", "normal_", "self", "std", "super", "weight"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeTextRMSNorm.__init__": ["ModelTextRMSNorm", "Module", "Parameter", "__init__", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeTextRMSNorm.forward": ["ModelTextRMSNorm", "Module", "True", "class", "def", "dtype", "float32", "forward", "hidden_states", "input_dtype", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeTextRMSNorm.extra_repr": ["ModelTextRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "variance_epsilon", "weight"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeThinkerTextModel.__init__": ["Embedding", "False", "ModelPreTrainedModel", "ModelTextConfig", "ModelTextRMSNorm", "ModelThinkerTextDecoderLayer", "ModelThinkerTextModel", "ModelThinkerTextRotaryEmbedding", "ModuleList", "__init__", "class", "config", "def", "embed_tokens", "eps", "for", "gradient_checkpointing", "hidden_size", "in", "layer_idx", "layers", "nn", "norm", "num_hidden_layers", "pad_token_id", "padding_idx", "post_init", "range", "rms_norm_eps", "rotary_emb", "self", "super", "vocab_size"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeThinkerTextModel.forward": ["BaseModelOutputWithPast", "Cache", "DynamicCache", "FlashAttentionKwargs", "FloatTensor", "LongTensor", "ModelPreTrainedModel", "ModelThinkerTextModel", "None", "Tensor", "Unpack", "ValueError", "You", "_deepstack_process", "and", "arange", "attention_mask", "bool", "cache_position", "class", "config", "create_causal_mask", "decoder_layer", "deepstack_visual_embeds", "def", "device", "elif", "else", "embed_tokens", "enumerate", "exactly", "expand", "for", "forward", "get_seq_length", "hidden_states", "if", "in", "input_embeds", "input_ids", "inputs_embeds", "is", "is_tracing", "jit", "kwargs", "last_hidden_state", "layer_idx", "layer_outputs", "layers", "len", "list", "must", "ndim", "norm", "not", "of", "one", "or", "past_key_values", "past_seen_tokens", "position_embeddings", "position_ids", "r", "raise", "range", "return", "rotary_emb", "self", "shape", "specify", "text_position_ids", "torch", "tuple", "use_cache", "view", "visual_pos_masks"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeThinkerTextModel._deepstack_process": ["ModelPreTrainedModel", "ModelThinkerTextModel", "Tensor", "_deepstack_process", "class", "clone", "def", "device", "dtype", "hidden_states", "local_this", "return", "self", "to", "torch", "visual_embeds", "visual_pos_masks"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:load_balancing_loss_func": ["Model_balancing_loss_func", "None", "Tensor", "_", "attention_mask", "batch_size", "cat", "compute_device", "concatenated_gate_logits", "def", "device", "dim", "else", "expand", "expert_attention_mask", "expert_mask", "float", "for", "functional", "gate_logits", "if", "in", "int", "is", "isinstance", "layer_gate", "mean", "nn", "not", "num_experts", "num_hidden_layers", "one_hot", "or", "overall_loss", "r", "reshape", "return", "router_per_expert_attention_mask", "router_prob_per_expert", "routing_weights", "selected_experts", "sequence_length", "shape", "softmax", "sum", "to", "tokens_per_expert", "top_k", "topk", "torch", "tuple", "unsqueeze"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeThinkerForConditionalGeneration.__init__": ["False", "Linear", "ModelAudioEncoder", "ModelThinkerForConditionalGeneration", "ModelThinkerTextModel", "ModelVisionEncoder", "None", "__init__", "_from_config", "audio_config", "audio_tower", "bias", "class", "config", "def", "else", "hidden_size", "if", "is", "lm_head", "model", "nn", "not", "num_experts", "num_experts_per_tok", "pad_token_id", "post_init", "rope_deltas", "router_aux_loss_coef", "self", "spatial_merge_size", "super", "text_config", "vision_config", "visual", "vocab_size"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeThinkerForConditionalGeneration.get_input_embeddings": ["ModelThinkerForConditionalGeneration", "class", "def", "get_input_embeddings", "model", "return", "self"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeThinkerForConditionalGeneration.set_input_embeddings": ["ModelThinkerForConditionalGeneration", "class", "def", "model", "self", "set_input_embeddings", "value"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeThinkerForConditionalGeneration.get_video_features": ["FloatTensor", "LongTensor", "ModelThinkerForConditionalGeneration", "None", "class", "def", "dtype", "get_video_features", "grid_thw", "pixel_values_videos", "return", "self", "torch", "type", "video_embeds", "video_grid_thw", "visual"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeThinkerForConditionalGeneration.get_image_features": ["FloatTensor", "LongTensor", "ModelThinkerForConditionalGeneration", "None", "class", "def", "dtype", "get_image_features", "grid_thw", "image_embeds", "image_grid_thw", "pixel_values", "return", "self", "torch", "type", "visual"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeThinkerForConditionalGeneration.get_audio_features": ["FloatTensor", "LongTensor", "ModelThinkerForConditionalGeneration", "None", "audio_feature_lengths", "audio_features", "audio_outputs", "audio_tower", "bool", "class", "def", "dim", "else", "feature_attention_mask", "feature_lens", "get_audio_features", "if", "input_features", "is", "last_hidden_state", "not", "permute", "return", "self", "sum", "torch"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeThinkerForConditionalGeneration.get_placeholder_mask": ["FloatTensor", "Image", "LongTensor", "ModelThinkerForConditionalGeneration", "None", "ValueError", "Videos", "all", "and", "audio_token_id", "class", "config", "def", "device", "do", "dtype", "else", "expand_as", "f", "features", "get_input_embeddings", "get_placeholder_mask", "if", "image", "image_features", "image_token_id", "input_ids", "inputs_embeds", "is", "long", "match", "n_image_tokens", "n_video_tokens", "not", "numel", "raise", "return", "self", "shape", "special_audio_mask", "special_image_mask", "special_video_mask", "sum", "tensor", "to", "tokens", "torch", "unsqueeze", "video", "video_features", "video_token_id"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeThinkerForConditionalGeneration.forward": ["ModelThinkerCausalLMOutputWithPast", "ModelThinkerForConditionalGeneration", "None", "_", "add", "and", "arange", "attention_mask", "attentions", "audio_feature_lengths", "audio_features", "audio_mask", "aux_loss", "batch_size", "bool", "cache_position", "class", "config", "deepstack_visual_embeds", "def", "delta", "delta0", "device", "dim", "dtype", "elif", "else", "embed_joint", "expand", "feature_attention_mask", "for", "forward", "get_audio_features", "get_image_features", "get_input_embeddings", "get_placeholder_mask", "get_rope_index", "get_seq_length", "get_text_config", "get_video_features", "hidden_states", "if", "image_embeds", "image_embeds_multiscale", "image_features", "image_grid_thw", "image_mask", "image_mask_joint", "img_embed", "in", "input_features", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "lm_head", "load_balancing_loss_func", "logits", "loss", "loss_function", "masked_scatter", "model", "new_zeros", "not", "num_experts", "num_experts_per_tok", "or", "output_router_logits", "outputs", "past_key_values", "past_key_values_length", "pixel_values", "pixel_values_videos", "position_ids", "r", "return", "rope_deltas", "router_aux_loss_coef", "router_logits", "self", "seq_length", "shape", "sum", "text_config", "to", "torch", "tuple", "unsqueeze", "use_audio_in_video", "use_cache", "vid_embed", "video_embeds", "video_embeds_multiscale", "video_features", "video_grid_thw", "video_mask", "video_mask_joint", "video_second_per_grid", "view", "visual_embeds_multiscale", "visual_embeds_multiscale_joint", "visual_pos_masks", "vocab_size", "zip"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeThinkerForConditionalGeneration.prepare_inputs_for_generation": ["False", "ModelThinkerForConditionalGeneration", "None", "True", "and", "attention_mask", "cache_position", "class", "def", "feature_attention_mask", "if", "image_grid_thw", "input_features", "input_ids", "inputs_embeds", "is_first_iteration", "kwargs", "model_inputs", "not", "past_key_values", "pixel_values", "pixel_values_videos", "position_ids", "prepare_inputs_for_generation", "return", "self", "super", "use_audio_in_video", "use_cache", "video_grid_thw", "video_second_per_grid"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeTalkerResizeMLP.__init__": ["ACT2FN", "Linear", "ModelTalkerConfig", "ModelTalkerResizeMLP", "Module", "True", "__init__", "act_fn", "bias", "class", "config", "def", "hidden_act", "hidden_size", "intermediate_size", "linear_fc1", "linear_fc2", "nn", "self", "super", "text_config", "thinker_hidden_size"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeTalkerResizeMLP.forward": ["ModelTalkerResizeMLP", "Module", "act_fn", "class", "def", "forward", "hidden_state", "linear_fc1", "linear_fc2", "nn", "return", "self"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeRMSNorm.__init__": ["ModelRMSNorm", "Module", "None", "Parameter", "__init__", "class", "def", "eps", "float", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeRMSNorm.forward": ["ModelRMSNorm", "Module", "Tensor", "True", "class", "def", "dtype", "float32", "forward", "hidden_states", "input_dtype", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeRMSNorm.extra_repr": ["ModelRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "variance_epsilon", "weight"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeTalkerCodePredictorAttention.__init__": ["Linear", "ModelConfig", "ModelRMSNorm", "ModelTalkerCodePredictorAttention", "Module", "None", "True", "__init__", "attention_bias", "attention_dropout", "bias", "class", "config", "def", "else", "eps", "getattr", "hasattr", "head_dim", "hidden_size", "if", "int", "is_causal", "k_norm", "k_proj", "layer_idx", "layer_type", "layer_types", "nn", "num_attention_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "q_norm", "q_proj", "rms_norm_eps", "scaling", "self", "sliding_attention", "sliding_window", "super", "v_proj"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeTalkerCodePredictorAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "FlashAttentionKwargs", "LongTensor", "ModelTalkerCodePredictorAttention", "Module", "None", "Tensor", "Unpack", "_attn_implementation", "apply_rotary_pos_emb", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "cache_kwargs", "cache_position", "class", "config", "contiguous", "cos", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "head_dim", "hidden_shape", "hidden_states", "if", "input_shape", "is", "k_norm", "k_proj", "key_states", "kwargs", "layer_idx", "nn", "not", "o_proj", "past_key_values", "position_embeddings", "q_norm", "q_proj", "query_states", "reshape", "return", "scaling", "self", "shape", "sin", "sliding_window", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "view"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeMLP.__init__": ["ACT2FN", "False", "Linear", "ModelMLP", "Module", "__init__", "act_fn", "bias", "class", "config", "def", "down_proj", "gate_proj", "hidden_act", "hidden_size", "intermediate_size", "nn", "self", "super", "up_proj"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeMLP.forward": ["ModelMLP", "Module", "act_fn", "class", "def", "down_proj", "forward", "gate_proj", "nn", "return", "self", "up_proj", "x"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeTalkerCodePredictorDecoderLayer.__init__": ["GradientCheckpointingLayer", "ModelMLP", "ModelRMSNorm", "ModelTalkerCodePredictorAttention", "ModelTalkerCodePredictorDecoderLayer", "__init__", "attention_type", "class", "config", "def", "eps", "hidden_size", "input_layernorm", "layer_idx", "layer_types", "mlp", "post_attention_layernorm", "rms_norm_eps", "self", "self_attn", "super"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeTalkerCodePredictorDecoderLayer.forward": ["Cache", "False", "GradientCheckpointingLayer", "LongTensor", "ModelTalkerCodePredictorDecoderLayer", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "attention_mask", "bool", "cache_position", "class", "def", "forward", "hidden_states", "input_layernorm", "kwargs", "mlp", "past_key_values", "position_embeddings", "position_ids", "post_attention_layernorm", "residual", "return", "self", "self_attn", "torch", "tuple", "use_cache"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeRotaryEmbedding.__init__": ["Callable", "False", "ModelConfig", "ModelRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeRotaryEmbedding.compute_default_rope_parameters": ["ModelConfig", "ModelRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeRotaryEmbedding.forward": ["False", "ModelRotaryEmbedding", "Module", "None", "and", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeTalkerCodePredictorModel.__init__": ["Embedding", "False", "ModelPreTrainedModel", "ModelRMSNorm", "ModelRotaryEmbedding", "ModelTalkerCodePredictorConfig", "ModelTalkerCodePredictorDecoderLayer", "ModelTalkerCodePredictorModel", "ModuleList", "_", "__init__", "class", "codec_embedding", "config", "def", "eps", "for", "gradient_checkpointing", "has_sliding_layers", "hidden_size", "in", "layer_idx", "layer_types", "layers", "nn", "norm", "num_code_groups", "num_hidden_layers", "pad_token_id", "padding_idx", "post_init", "range", "rms_norm_eps", "rotary_emb", "self", "sliding_attention", "super", "vocab_size"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeTalkerCodePredictorModel.forward": ["BaseModelOutputWithPast", "Cache", "DynamicCache", "FloatTensor", "LongTensor", "ModelPreTrainedModel", "ModelTalkerCodePredictorModel", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "and", "arange", "attention_mask", "attention_type", "be", "bool", "cache_position", "causal_mask_mapping", "class", "config", "create_causal_mask", "decoder_layer", "def", "device", "dict", "else", "expected", "for", "forward", "full_attention", "get_seq_length", "hidden_states", "if", "in", "input_embeds", "input_ids", "inputs_embeds", "is", "isinstance", "kwargs", "last_hidden_state", "layers", "mask_kwargs", "norm", "not", "num_hidden_layers", "past_key_values", "past_seen_tokens", "position_embeddings", "position_ids", "raise", "return", "rotary_emb", "self", "shape", "to", "torch", "unsqueeze", "use_cache"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeTalkerCodePredictorModel.get_input_embeddings": ["ModelPreTrainedModel", "ModelTalkerCodePredictorModel", "class", "codec_embedding", "def", "get_input_embeddings", "return", "self"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeTalkerCodePredictorModelForConditionalGeneration.__init__": ["False", "GenerationMixin", "Linear", "ModelPreTrainedModel", "ModelTalkerCodePredictorConfig", "ModelTalkerCodePredictorModel", "ModelTalkerCodePredictorModelForConditionalGeneration", "ModuleList", "_", "__init__", "_from_config", "bias", "class", "config", "def", "for", "hidden_size", "in", "lm_head", "model", "nn", "num_code_groups", "post_init", "range", "self", "super", "vocab_size"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeTalkerCodePredictorModelForConditionalGeneration.forward": ["BaseModelOutputWithPast", "CausalLMOutputWithPast", "GenerationMixin", "ModelPreTrainedModel", "ModelTalkerCodePredictorModelForConditionalGeneration", "ModelTalkerCodePredictorOutputWithPast", "None", "and", "attention_mask", "attentions", "cache_position", "class", "config", "def", "else", "forward", "generation_steps", "get_input_embeddings", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "last_hidden_state", "lm_head", "logits", "loss", "loss_function", "model", "not", "outputs", "past_key_values", "position_ids", "r", "return", "self", "shape", "use_cache", "vocab_size"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeTalkerCodePredictorModelForConditionalGeneration.get_input_embeddings": ["GenerationMixin", "ModelPreTrainedModel", "ModelTalkerCodePredictorModelForConditionalGeneration", "class", "def", "get_input_embeddings", "model", "return", "self"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeTalkerCodePredictorModelForConditionalGeneration._update_model_kwargs_for_generation": ["False", "GenerationMixin", "ModelPreTrainedModel", "ModelTalkerCodePredictorModelForConditionalGeneration", "_update_model_kwargs_for_generation", "class", "def", "generation_steps", "is_encoder_decoder", "model_kwargs", "num_new_tokens", "outputs", "return", "self", "super"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeTalkerTextMLP.__init__": ["ACT2FN", "False", "Linear", "ModelTalkerTextMLP", "Module", "None", "__init__", "act_fn", "bias", "class", "config", "def", "down_proj", "else", "gate_proj", "hidden_act", "hidden_size", "if", "intermediate_size", "is", "nn", "self", "super", "up_proj"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeTalkerTextMLP.forward": ["ModelTalkerTextMLP", "Module", "act_fn", "class", "def", "down_proj", "forward", "gate_proj", "nn", "return", "self", "up_proj", "x"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeTalkerTextExperts.__init__": ["ACT2FN", "ModelTalkerTextExperts", "Module", "Parameter", "__init__", "act_fn", "class", "config", "def", "down_proj", "empty", "gate_up_proj", "hidden_act", "hidden_dim", "hidden_size", "intermediate_dim", "moe_intermediate_size", "nn", "num_experts", "self", "super", "torch"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeTalkerTextExperts.forward": ["ModelTalkerTextExperts", "Module", "None", "Tensor", "act_fn", "chunk", "class", "continue", "current_hidden_states", "current_state", "def", "dim", "down_proj", "dtype", "expert_hit", "expert_idx", "expert_mask", "final_hidden_states", "for", "forward", "functional", "gate", "gate_up_proj", "greater", "hidden_states", "if", "in", "index_add_", "linear", "nn", "no_grad", "nonzero", "num_classes", "num_experts", "one_hot", "permute", "return", "self", "sum", "to", "token_idx", "top_k_index", "top_k_pos", "top_k_weights", "torch", "up", "where", "with", "zeros_like"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeTalkerTextTopKRouter.__init__": ["ModelTalkerTextTopKRouter", "Module", "Parameter", "__init__", "class", "config", "def", "hidden_dim", "hidden_size", "nn", "norm_topk_prob", "num_experts", "num_experts_per_tok", "self", "super", "top_k", "torch", "weight", "zeros"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeTalkerTextTopKRouter.forward": ["F", "ModelTalkerTextTopKRouter", "Module", "True", "class", "def", "dim", "dtype", "float", "forward", "functional", "hidden_dim", "hidden_states", "if", "keepdim", "linear", "nn", "norm_topk_prob", "reshape", "return", "router_indices", "router_logits", "router_scores", "router_top_value", "self", "softmax", "sum", "to", "top_k", "topk", "torch", "weight"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeTalkerTextSparseMoeBlock.__init__": ["False", "Linear", "ModelTalkerTextExperts", "ModelTalkerTextMLP", "ModelTalkerTextSparseMoeBlock", "ModelTalkerTextTopKRouter", "Module", "__init__", "bias", "class", "config", "def", "experts", "gate", "hidden_size", "intermediate_size", "nn", "self", "shared_expert", "shared_expert_gate", "shared_expert_intermediate_size", "super", "torch"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeTalkerTextSparseMoeBlock.forward": ["F", "ModelTalkerTextSparseMoeBlock", "Module", "Tensor", "_", "batch_size", "class", "def", "expert_output", "experts", "forward", "gate", "hidden_dim", "hidden_states", "hidden_states_reshaped", "nn", "reshape", "return", "routing_weights", "selected_experts", "self", "sequence_length", "shape", "shared_expert", "shared_expert_gate", "shared_expert_output", "sigmoid", "torch", "tuple", "view"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeTalkerDecoderLayer.__init__": ["GradientCheckpointingLayer", "ModelTalkerDecoderLayer", "ModelTalkerTextSparseMoeBlock", "ModelThinkerTextAttention", "ModelThinkerTextMLP", "ModelThinkerTextRMSNorm", "ModelThinkerTextSparseMoeBlock", "__init__", "and", "class", "config", "decoder_sparse_step", "def", "else", "eps", "hidden_size", "if", "in", "input_layernorm", "intermediate_size", "layer_idx", "mlp", "mlp_only_layers", "not", "num_experts", "post_attention_layernorm", "rms_norm_eps", "self", "self_attn", "super"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeTalkerDecoderLayer.forward": ["Cache", "False", "GradientCheckpointingLayer", "LongTensor", "ModelTalkerDecoderLayer", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "attention_mask", "bool", "cache_position", "class", "def", "forward", "hidden_states", "input_layernorm", "kwargs", "mlp", "past_key_values", "position_embeddings", "position_ids", "post_attention_layernorm", "residual", "return", "self", "self_attn", "torch", "tuple", "use_cache"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeTalkerModel.__init__": ["Embedding", "False", "ModelPreTrainedModel", "ModelTalkerDecoderLayer", "ModelTalkerModel", "ModelTalkerRotaryEmbedding", "ModelTalkerTextConfig", "ModelTextRMSNorm", "ModuleList", "__init__", "class", "codec_embedding", "config", "def", "eps", "for", "gradient_checkpointing", "hidden_size", "in", "layer_idx", "layers", "nn", "norm", "num_hidden_layers", "pad_token_id", "padding_idx", "post_init", "range", "rms_norm_eps", "rotary_emb", "self", "super", "vocab_size"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeTalkerModel.forward": ["BaseModelOutputWithPast", "Cache", "DynamicCache", "FlashAttentionKwargs", "FloatTensor", "LongTensor", "ModelPreTrainedModel", "ModelTalkerModel", "None", "Tensor", "Unpack", "ValueError", "You", "_deepstack_process", "and", "arange", "attention_mask", "bool", "cache_position", "class", "config", "create_causal_mask", "decoder_layer", "deepstack_visual_embeds", "def", "device", "elif", "else", "embed_tokens", "enumerate", "exactly", "expand", "for", "forward", "get_seq_length", "hidden_states", "if", "in", "input_embeds", "input_ids", "inputs_embeds", "is", "is_tracing", "jit", "kwargs", "last_hidden_state", "layer_idx", "layer_outputs", "layers", "len", "list", "must", "ndim", "norm", "not", "of", "one", "or", "past_key_values", "past_seen_tokens", "position_embeddings", "position_ids", "r", "raise", "range", "return", "rotary_emb", "self", "shape", "specify", "text_position_ids", "torch", "tuple", "use_cache", "view", "visual_pos_masks"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeTalkerModel._deepstack_process": ["ModelPreTrainedModel", "ModelTalkerModel", "Tensor", "_deepstack_process", "class", "clone", "def", "device", "dtype", "hidden_states", "local_this", "return", "self", "to", "torch", "visual_embeds", "visual_pos_masks"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeTalkerModel.get_input_embeddings": ["ModelPreTrainedModel", "ModelTalkerModel", "class", "codec_embedding", "def", "get_input_embeddings", "return", "self"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeTalkerForConditionalGeneration.__init__": ["False", "GenerationMixin", "Linear", "ModelTalkerCodePredictorModelForConditionalGeneration", "ModelTalkerConfig", "ModelTalkerForConditionalGeneration", "ModelTalkerModel", "ModelTalkerResizeMLP", "ModelThinkerTextPreTrainedModel", "None", "__init__", "_from_config", "bias", "class", "code_predictor", "code_predictor_config", "codec_head", "config", "def", "hidden_projection", "hidden_size", "model", "nn", "num_experts", "num_experts_per_tok", "post_init", "rope_deltas", "router_aux_loss_coef", "self", "spatial_merge_size", "super", "text_config", "text_projection", "vocab_size"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeTalkerForConditionalGeneration.forward": ["GenerationMixin", "ModelTalkerForConditionalGeneration", "ModelTalkerOutputWithPast", "ModelThinkerTextPreTrainedModel", "MoeCausalLMOutputWithPast", "MoeModelOutputWithPast", "None", "add", "and", "arange", "attention_mask", "audio_feature_lengths", "aux_loss", "batch_size", "cache_position", "class", "codec_head", "config", "def", "delta", "delta0", "device", "dim", "else", "expand", "forward", "generation_step", "get_rope_index", "get_seq_length", "hidden_states", "if", "image_grid_thw", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "last_hidden_state", "load_balancing_loss_func", "logits", "loss", "loss_function", "model", "not", "num_experts", "num_experts_per_tok", "or", "output_router_logits", "outputs", "past_key_values", "past_key_values_length", "position_ids", "r", "residual_codes", "return", "rope_deltas", "router_aux_loss_coef", "router_logits", "self", "seq_length", "shape", "sum", "talker_input_ids", "to", "torch", "trailing_text_hidden", "tts_pad_embed", "unsqueeze", "use_audio_in_video", "use_cache", "video_grid_thw", "video_second_per_grid", "view", "vocab_size"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeTalkerForConditionalGeneration.get_rope_index": ["False", "GenerationMixin", "LongTensor", "ModelPreTrainedModelForConditionalGeneration", "ModelTalkerForConditionalGeneration", "ModelThinkerTextPreTrainedModel", "None", "Tensor", "attention_mask", "audio_seqlens", "bool", "class", "def", "get_rope_index", "image_grid_thw", "input_ids", "return", "second_per_grids", "self", "torch", "tuple", "use_audio_in_video", "video_grid_thw"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeTalkerForConditionalGeneration.get_llm_pos_ids_for_vision": ["GenerationMixin", "ModelPreTrainedModelForConditionalGeneration", "ModelTalkerForConditionalGeneration", "ModelThinkerTextPreTrainedModel", "Tensor", "class", "def", "get_llm_pos_ids_for_vision", "grid_hs", "grid_ws", "int", "list", "return", "self", "spatial_merge_size", "start_idx", "t_index", "torch", "vision_idx"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeTalkerForConditionalGeneration.get_input_embeddings": ["GenerationMixin", "ModelTalkerForConditionalGeneration", "ModelThinkerTextPreTrainedModel", "class", "def", "get_input_embeddings", "model", "return", "self"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeTalkerForConditionalGeneration._update_model_kwargs_for_generation": ["False", "GenerationMixin", "ModelTalkerForConditionalGeneration", "ModelThinkerTextPreTrainedModel", "_update_model_kwargs_for_generation", "class", "def", "generation_step", "hidden_states", "is_encoder_decoder", "model_kwargs", "num_new_tokens", "outputs", "return", "self", "super"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeTalkerForConditionalGeneration.prepare_inputs_for_generation": ["False", "GenerationMixin", "ModelTalkerForConditionalGeneration", "ModelThinkerTextPreTrainedModel", "None", "True", "and", "attention_mask", "cache_position", "cat", "class", "code_predictor", "codec_hiddens", "config", "def", "device", "dim", "do_sample", "else", "for", "generate", "generation_step", "get", "get_input_embeddings", "hid", "hidden_states", "if", "in", "input_ids", "inputs", "inputs_embeds", "is_first_iteration", "keepdim", "kwargs", "last_id_hidden", "last_residual_hidden", "max_new_tokens", "mid_residual_hiddens", "not", "num_code_groups", "output_hidden_states", "past_hidden", "past_key_values", "pop", "position_ids", "predictor_result", "prepare_inputs_for_generation", "residual_codes", "return", "return_dict_in_generate", "self", "sequences", "shape", "sum", "super", "to", "top_k", "top_p", "torch", "trailing_text_hidden", "tts_pad_embed", "unsqueeze", "use_cache"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeCausalConvNet.__init__": ["Conv1d", "ModelCausalConvNet", "Module", "__init__", "class", "conv", "def", "dilation", "groups", "in_channels", "kernel_size", "nn", "out_channels", "padding", "self", "stride", "super"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeCausalConvNet._get_extra_padding_for_conv1d": ["ModelCausalConvNet", "Module", "Tensor", "_get_extra_padding_for_conv1d", "ceil", "class", "def", "hidden_state", "ideal_length", "int", "kernel_size", "length", "math", "n_frames", "nn", "padding", "return", "self", "shape", "stride", "torch"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeCausalConvNet.forward": ["F", "ModelCausalConvNet", "Module", "_get_extra_padding_for_conv1d", "class", "constant", "contiguous", "conv", "def", "extra_padding", "forward", "hidden_state", "mode", "nn", "pad", "padding", "return", "self", "value"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeCausalTransConvNet.__init__": ["ConvTranspose1d", "ModelCausalTransConvNet", "Module", "__init__", "ceil", "class", "conv", "def", "in_channels", "kernel_size", "left_pad", "math", "nn", "out_channels", "pad", "right_pad", "self", "stride", "super"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeCausalTransConvNet.forward": ["ModelCausalTransConvNet", "Module", "class", "contiguous", "conv", "def", "forward", "hidden_state", "left_pad", "nn", "return", "right_pad", "self", "shape"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeConvNeXtBlock.__init__": ["GELU", "LayerNorm", "Linear", "ModelCausalConvNet", "ModelConvNeXtBlock", "Module", "Parameter", "__init__", "act", "class", "def", "dilation", "dim", "dwconv", "eps", "gamma", "groups", "int", "kernel_size", "nn", "norm", "ones", "pwconv1", "pwconv2", "self", "super", "torch"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeConvNeXtBlock.forward": ["ModelConvNeXtBlock", "Module", "act", "class", "def", "dwconv", "forward", "gamma", "hidden_states", "input", "nn", "norm", "permute", "pwconv1", "pwconv2", "return", "self"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeCode2WavAttention.__init__": ["Identity", "Linear", "ModelCode2WavAttention", "ModelCode2WavConfig", "Module", "None", "True", "__init__", "attention_bias", "attention_dropout", "bias", "class", "config", "def", "else", "getattr", "hasattr", "head_dim", "hidden_size", "if", "is_causal", "k_norm", "k_proj", "layer_idx", "layer_type", "layer_types", "nn", "num_attention_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "q_norm", "q_proj", "scaling", "self", "sliding_window", "super", "v_proj"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeCode2WavAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "FlashAttentionKwargs", "LongTensor", "ModelCode2WavAttention", "Module", "None", "Tensor", "Unpack", "_attn_implementation", "apply_rotary_pos_emb", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "cache_kwargs", "cache_position", "class", "config", "contiguous", "cos", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "head_dim", "hidden_shape", "hidden_states", "if", "input_shape", "is", "k_norm", "k_proj", "key_states", "kwargs", "layer_idx", "nn", "not", "o_proj", "past_key_values", "position_embeddings", "q_norm", "q_proj", "query_states", "reshape", "return", "scaling", "self", "shape", "sin", "sliding_window", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "view"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeCode2WavMlp.__init__": ["ACT2FN", "False", "Linear", "ModelCode2WavMlp", "Module", "__init__", "act_fn", "bias", "class", "config", "def", "down_proj", "gate_proj", "hidden_act", "hidden_size", "intermediate_size", "nn", "self", "super", "up_proj"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeCode2WavMlp.forward": ["ModelCode2WavMlp", "Module", "act_fn", "class", "def", "down_proj", "forward", "gate_proj", "nn", "return", "self", "up_proj", "x"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeCode2WavRMSNorm.__init__": ["ModelCode2WavRMSNorm", "Module", "None", "Parameter", "__init__", "class", "def", "eps", "float", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeCode2WavRMSNorm.forward": ["ModelCode2WavRMSNorm", "Module", "Tensor", "True", "class", "def", "dtype", "float32", "forward", "hidden_states", "input_dtype", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeCode2WavRMSNorm.extra_repr": ["ModelCode2WavRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "variance_epsilon", "weight"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeCode2WavLayerScale.__init__": ["ModelCode2WavLayerScale", "Module", "Parameter", "True", "__init__", "channels", "class", "config", "def", "full", "hidden_size", "initial_scale", "layer_scale_initial_scale", "nn", "requires_grad", "scale", "self", "super", "torch"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeCode2WavLayerScale.forward": ["ModelCode2WavLayerScale", "Module", "Tensor", "class", "def", "forward", "nn", "return", "scale", "self", "torch", "x"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeCode2WavTransformerLayer.__init__": ["GradientCheckpointingLayer", "ModelCode2WavAttention", "ModelCode2WavConfig", "ModelCode2WavLayerScale", "ModelCode2WavMlp", "ModelCode2WavRMSNorm", "ModelCode2WavTransformerLayer", "__init__", "attention_type", "class", "config", "def", "hidden_size", "input_layernorm", "layer_idx", "mlp", "mlp_layer_scale", "post_attention_layernorm", "rms_norm_eps", "self", "self_attn", "self_attn_layer_scale", "sliding_attention", "super"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeCode2WavTransformerLayer.forward": ["Cache", "False", "FloatTensor", "GradientCheckpointingLayer", "LongTensor", "ModelCode2WavTransformerLayer", "None", "Tensor", "_", "attention_mask", "bool", "cache_position", "class", "def", "forward", "hidden_states", "input_layernorm", "kwargs", "mlp", "mlp_layer_scale", "past_key_values", "position_ids", "post_attention_layernorm", "residual", "return", "self", "self_attn", "self_attn_layer_scale", "torch", "tuple", "use_cache"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeCode2WavTransformerModel.__init__": ["False", "ModelCode2WavConfig", "ModelCode2WavTransformerLayer", "ModelCode2WavTransformerModel", "ModelPreTrainedModel", "ModelRMSNorm", "ModelRotaryEmbedding", "ModuleList", "__init__", "class", "config", "def", "eps", "for", "gradient_checkpointing", "has_sliding_layers", "hidden_size", "in", "layer_idx", "layer_types", "layers", "nn", "norm", "num_hidden_layers", "post_init", "range", "rms_norm_eps", "rotary_emb", "self", "sliding_attention", "sliding_window", "super", "window_size"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeCode2WavTransformerModel.forward": ["BaseModelOutputWithPast", "DynamicCache", "ModelCode2WavTransformerModel", "ModelPreTrainedModel", "None", "ValueError", "You", "and", "arange", "attention_mask", "attention_type", "cache_position", "causal_mask_mapping", "class", "config", "create_causal_mask", "create_sliding_window_causal_mask", "decoder_layer", "def", "device", "dict", "else", "embed_tokens", "exactly", "expected", "for", "forward", "full_attention", "get_seq_length", "has_sliding_layers", "hidden_states", "if", "in", "input_embeds", "input_ids", "inputs_embeds", "is", "isinstance", "kwargs", "last_hidden_state", "layers", "mask_kwargs", "must", "norm", "not", "num_hidden_layers", "of", "one", "or", "past_key_values", "past_seen_tokens", "position_embeddings", "position_ids", "raise", "return", "rotary_emb", "self", "shape", "sliding_attention", "specify", "torch", "unsqueeze", "use_cache"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:SnakeBeta.__init__": ["ModelBeta", "Module", "Parameter", "__init__", "alpha", "beta", "class", "def", "in_features", "nn", "no_div_by_zero", "self", "super", "torch", "zeros"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:SnakeBeta.forward": ["ModelBeta", "Module", "alpha", "beta", "class", "def", "exp", "forward", "hidden_states", "nn", "no_div_by_zero", "pow", "return", "self", "sin", "torch", "unsqueeze"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeCode2WavDecoderResidualUnit.__init__": ["ModelCausalConvNet", "ModelCode2WavDecoderResidualUnit", "Module", "SnakeBeta", "__init__", "act1", "act2", "class", "conv1", "conv2", "def", "dilation", "dim", "int", "kernel_size", "nn", "self", "super"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeCode2WavDecoderResidualUnit.forward": ["ModelCode2WavDecoderResidualUnit", "Module", "act1", "act2", "class", "conv1", "conv2", "def", "forward", "hidden_state", "nn", "residual", "return", "self"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeCode2WavDecoderBlock.__init__": ["ModelCausalTransConvNet", "ModelCode2WavConfig", "ModelCode2WavDecoderBlock", "ModelCode2WavDecoderResidualUnit", "ModelPreTrainedModel", "ModuleList", "SnakeBeta", "__init__", "append", "block", "class", "config", "decoder_dim", "def", "dilation", "for", "in", "in_dim", "layer_idx", "nn", "out_dim", "post_init", "self", "super", "upsample_rate", "upsample_rates"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeCode2WavDecoderBlock.forward": ["ModelCode2WavDecoderBlock", "ModelPreTrainedModel", "block", "class", "def", "for", "forward", "hidden", "in", "kwargs", "return", "self"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeCode2Wav.__init__": ["Embedding", "False", "ModelCausalConvNet", "ModelCausalTransConvNet", "ModelCode2Wav", "ModelCode2WavConfig", "ModelCode2WavDecoderBlock", "ModelCode2WavTransformerModel", "ModelConvNeXtBlock", "ModelPreTrainedModel", "ModuleList", "SnakeBeta", "__init__", "_from_config", "append", "arange", "class", "code_embedding", "code_offset", "codebook_size", "config", "decoder", "decoder_dim", "def", "factor", "for", "hidden_size", "i", "in", "len", "nn", "np", "num_quantizers", "output_dim", "persistent", "post_init", "pre_transformer", "prod", "range", "register_buffer", "self", "super", "torch", "total_upsample", "upsample", "upsample_rates", "upsampling_ratios", "view"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeCode2Wav.forward": ["Expected", "ModelCode2Wav", "ModelPreTrainedModel", "ValueError", "block", "blocks", "clamp", "class", "code_embedding", "code_offset", "codes", "config", "decoder", "def", "f", "for", "forward", "got", "hidden", "if", "in", "inputs_embeds", "kwargs", "last_hidden_state", "layer", "max", "mean", "min", "num_quantizers", "of", "permute", "pre_transformer", "raise", "return", "self", "shape", "upsample", "wav"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeCode2Wav.chunked_decode": ["ModelCode2Wav", "ModelPreTrainedModel", "append", "cat", "chunk_size", "chunked_decode", "class", "codes", "codes_chunk", "context_size", "def", "dim", "else", "end_index", "if", "left_context_size", "min", "return", "self", "shape", "start_index", "torch", "total_upsample", "wav_chunk", "wavs", "while"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeForConditionalGeneration.__init__": ["GenerationMixin", "ModelConfig", "ModelForConditionalGeneration", "ModelPreTrainedModel", "ModelThinkerForConditionalGeneration", "__init__", "_from_config", "class", "config", "def", "enable_audio_output", "enable_talker", "has_talker", "if", "post_init", "self", "super", "thinker", "thinker_config"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeForConditionalGeneration.enable_talker": ["GenerationMixin", "ModelCode2Wav", "ModelForConditionalGeneration", "ModelPreTrainedModel", "ModelTalkerForConditionalGeneration", "_from_config", "class", "code2wav", "code2wav_config", "config", "def", "enable_talker", "self", "talker", "talker_config"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeForConditionalGeneration.disable_talker": ["False", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "code2wav", "def", "del", "disable_talker", "has_talker", "hasattr", "if", "self", "talker"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeForConditionalGeneration._get_talker_user_parts": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "_get_talker_user_parts", "any", "class", "config", "def", "device", "dtype", "empty", "hidden_projection", "hidden_size", "if", "im_start_index", "mm_hidden", "multimodal_mask", "return", "segment_end_index", "self", "talker", "talker_config", "text_config", "text_projection", "thinker_embed", "thinker_hidden", "to", "torch", "user_mm_mask", "user_talker_part", "user_text_hidden", "user_thinker_embed", "user_thinker_hidden_mm"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeForConditionalGeneration._get_talker_assistant_parts": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "_get_talker_assistant_parts", "assistant_codec_hidden", "assistant_hidden", "assistant_text_hidden", "cat", "class", "codec_bos_id", "codec_nothink_id", "codec_pad_id", "codec_special_tokens", "codec_think_bos_id", "codec_think_eos_id", "config", "def", "device", "dim", "dtype", "expand", "fill_value", "full", "get_input_embeddings", "hidden_size", "im_start_index", "input_embeds", "input_ids", "long", "return", "segment_end_index", "self", "shape", "speaker_id", "talker", "talker_config", "tensor", "text_config", "text_projection", "thinker_embed", "to", "torch", "trailing_text_hidden", "tts_bos_embed", "tts_eos_embed", "tts_pad_embed", "tts_pad_token_id", "zeros"], "qwen3_omni_moe/modeling_qwen3_omni_moe.py:Qwen3OmniMoeForConditionalGeneration.generate": ["AssertionError", "Cannot", "Ethan", "Expect", "False", "GenerationMixin", "Model", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "NotImplementedError", "Omni", "Speaker", "Tensor", "True", "Use", "ValueError", "_get_talker_assistant_parts", "_get_talker_user_parts", "accept_hidden_layer", "after", "and", "append", "assistant", "assistant_token_id", "attention_mask", "audio", "audio_feature_lengths", "audio_token_id", "batched", "bool", "cat", "chunk", "chunk_size", "chunked_decode", "class", "code2wav", "codec_eos_token_id", "config", "continue", "currently", "def", "device", "dim", "do_sample", "does", "dtype", "elif", "else", "embed", "enable", "enable_talker", "eos_token_id", "f", "feature_attention_mask", "float", "for", "generate", "generate_audio", "get", "get_input_embeddings", "has_talker", "hid", "hidden_states", "i", "id", "if", "im_start", "im_start_index", "im_start_indexes", "im_start_token_id", "image_grid_thw", "image_token_id", "implemented", "in", "inference", "initialized", "input_features", "input_ids", "inputs_embeds", "int", "is", "items", "key", "kwargs", "left_context_size", "len", "lower", "max_new_tokens", "method", "module", "multimodal_mask", "nonzero", "not", "or", "output", "output_hidden_states", "raise", "range", "repetition_penalty", "return", "return_audio", "return_dict_in_generate", "role", "role_token", "segment_end_index", "self", "sequences", "set", "shape", "shared_kwargs", "speaker", "speaker_id", "squeeze", "stack", "startswith", "str", "sum", "support", "suppress_tokens", "system", "system_token_id", "talker", "talker_", "talker_assistant_embeds", "talker_assistant_ids", "talker_codes", "talker_config", "talker_do_sample", "talker_input_embed", "talker_input_embeds", "talker_input_id", "talker_input_ids", "talker_kwargs", "talker_max_new_tokens", "talker_repetition_penalty", "talker_result", "talker_special_tokens", "talker_supppressed_tokens", "talker_temperature", "talker_top_k", "talker_top_p", "talker_user_part", "talker_wavs", "temperature", "tensor", "text_config", "text_projection", "thinker", "thinker_", "thinker_config", "thinker_embed", "thinker_eos_token_id", "thinker_hidden", "thinker_kwargs", "thinker_max_new_tokens", "thinker_result", "to", "token2wav_", "token2wav_kwargs", "top_k", "top_p", "torch", "trailing_text_hidden", "transpose", "tts_bos_embed", "tts_bos_token_id", "tts_eos_embed", "tts_eos_token_id", "tts_pad_embed", "tts_pad_token_id", "use", "use_audio_in_video", "user", "user_token_id", "value", "video_grid_thw", "video_second_per_grid", "video_token_id", "vocab_size", "when", "with"], "qwen3_vl/modeling_qwen3_vl.py:Qwen3VLVisionMLP.__init__": ["ACT2FN", "Linear", "ModelVisionMLP", "Module", "True", "__init__", "act_fn", "bias", "class", "config", "def", "hidden_act", "hidden_size", "intermediate_size", "linear_fc1", "linear_fc2", "nn", "self", "super"], "qwen3_vl/modeling_qwen3_vl.py:Qwen3VLVisionMLP.forward": ["ModelVisionMLP", "Module", "act_fn", "class", "def", "forward", "hidden_state", "linear_fc1", "linear_fc2", "nn", "return", "self"], "qwen3_vl/modeling_qwen3_vl.py:Qwen3VLVisionPatchEmbed.__init__": ["Conv3d", "ModelVisionPatchEmbed", "Module", "None", "True", "__init__", "bias", "class", "config", "def", "embed_dim", "hidden_size", "in_channels", "kernel_size", "nn", "patch_size", "proj", "self", "stride", "super", "temporal_patch_size"], "qwen3_vl/modeling_qwen3_vl.py:Qwen3VLVisionPatchEmbed.forward": ["ModelVisionPatchEmbed", "Module", "Tensor", "class", "def", "dtype", "embed_dim", "forward", "hidden_states", "in_channels", "nn", "patch_size", "proj", "return", "self", "target_dtype", "temporal_patch_size", "to", "torch", "view", "weight"], "qwen3_vl/modeling_qwen3_vl.py:Qwen3VLVisionRotaryEmbedding.__init__": ["False", "ModelVisionRotaryEmbedding", "Module", "None", "__init__", "arange", "class", "def", "dim", "dtype", "float", "int", "inv_freq", "nn", "persistent", "register_buffer", "self", "super", "theta", "torch"], "qwen3_vl/modeling_qwen3_vl.py:Qwen3VLVisionRotaryEmbedding.forward": ["ModelVisionRotaryEmbedding", "Module", "Tensor", "arange", "class", "def", "device", "dtype", "forward", "freqs", "int", "inv_freq", "nn", "outer", "return", "self", "seq", "seqlen", "torch"], "qwen3_vl/modeling_qwen3_vl.py:Qwen3VLVisionPatchMerger.__init__": ["False", "GELU", "LayerNorm", "Linear", "ModelVisionConfig", "ModelVisionPatchMerger", "Module", "None", "__init__", "act_fn", "class", "config", "def", "else", "eps", "hidden_size", "if", "linear_fc1", "linear_fc2", "nn", "norm", "out_hidden_size", "self", "spatial_merge_size", "super", "use_postshuffle_norm"], "qwen3_vl/modeling_qwen3_vl.py:Qwen3VLVisionPatchMerger.forward": ["ModelVisionPatchMerger", "Module", "Tensor", "act_fn", "class", "def", "else", "forward", "hidden_size", "if", "linear_fc1", "linear_fc2", "nn", "norm", "return", "self", "torch", "use_postshuffle_norm", "view", "x"], "qwen3_vl/modeling_qwen3_vl.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "qwen3_vl/modeling_qwen3_vl.py:apply_rotary_pos_emb_vision": ["Model_rotary_pos_emb_vision", "Tensor", "cos", "def", "dtype", "float", "k", "k_embed", "orig_k_dtype", "orig_q_dtype", "q", "q_embed", "return", "rotate_half", "sin", "to", "torch", "tuple", "unsqueeze"], "qwen3_vl/modeling_qwen3_vl.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "qwen3_vl/modeling_qwen3_vl.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value", "value_states"], "qwen3_vl/modeling_qwen3_vl.py:Qwen3VLVisionAttention.__init__": ["False", "Linear", "ModelVisionAttention", "ModelVisionConfig", "Module", "None", "True", "__init__", "attention_dropout", "bias", "class", "config", "def", "dim", "head_dim", "hidden_size", "is_causal", "nn", "num_heads", "num_key_value_groups", "proj", "qkv", "scaling", "self", "super"], "qwen3_vl/modeling_qwen3_vl.py:Qwen3VLVisionAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "False", "ModelVisionAttention", "Module", "None", "Tensor", "_", "_attn_implementation", "apply_rotary_pos_emb_vision", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_outputs", "cat", "class", "config", "contiguous", "cos", "cu_seq_lens_k", "cu_seq_lens_q", "cu_seqlens", "def", "dim", "dropout", "eager", "eager_attention_forward", "else", "flash", "for", "forward", "hidden_states", "if", "in", "is_causal", "k", "key_states", "kwargs", "lengths", "max", "max_length_k", "max_length_q", "max_seqlen", "nn", "not", "num_heads", "permute", "position_embeddings", "proj", "q", "qkv", "query_states", "reshape", "return", "rotary_pos_emb", "scaling", "self", "seq_length", "shape", "sin", "split", "splits", "tensor", "tolist", "torch", "training", "transpose", "tuple", "unbind", "unsqueeze", "v", "value_states", "zip"], "qwen3_vl/modeling_qwen3_vl.py:Qwen3VLVisionBlock.__init__": ["GradientCheckpointingLayer", "LayerNorm", "ModelVisionAttention", "ModelVisionBlock", "ModelVisionMLP", "None", "__init__", "attn", "attn_implementation", "class", "config", "def", "eps", "hidden_size", "mlp", "nn", "norm1", "norm2", "sdpa", "self", "str", "super"], "qwen3_vl/modeling_qwen3_vl.py:Qwen3VLVisionBlock.forward": ["GradientCheckpointingLayer", "ModelVisionBlock", "None", "Tensor", "attn", "class", "cu_seqlens", "def", "forward", "hidden_states", "kwargs", "mlp", "norm1", "norm2", "position_embeddings", "return", "rotary_pos_emb", "self", "torch", "tuple"], "qwen3_vl/modeling_qwen3_vl.py:Qwen3VLTextRotaryEmbedding.__init__": ["Callable", "False", "ModelTextConfig", "ModelTextRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "get", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "mrope_section", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "qwen3_vl/modeling_qwen3_vl.py:Qwen3VLTextRotaryEmbedding.compute_default_rope_parameters": ["ModelTextConfig", "ModelTextRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "qwen3_vl/modeling_qwen3_vl.py:Qwen3VLTextRotaryEmbedding.forward": ["False", "ModelTextRotaryEmbedding", "Module", "None", "and", "apply_interleaved_mrope", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "mrope_section", "ndim", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "qwen3_vl/modeling_qwen3_vl.py:Qwen3VLTextRotaryEmbedding.apply_interleaved_mrope": ["ModelTextRotaryEmbedding", "Module", "apply_interleaved_mrope", "class", "def", "dim", "enumerate", "for", "freqs", "freqs_t", "idx", "in", "length", "mrope_section", "nn", "offset", "return", "self", "slice", "start"], "qwen3_vl/modeling_qwen3_vl.py:Qwen3VLTextRMSNorm.__init__": ["ModelTextRMSNorm", "Module", "None", "Parameter", "__init__", "class", "def", "eps", "float", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "qwen3_vl/modeling_qwen3_vl.py:Qwen3VLTextRMSNorm.forward": ["ModelTextRMSNorm", "Module", "Tensor", "True", "class", "def", "dtype", "float32", "forward", "hidden_states", "input_dtype", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "qwen3_vl/modeling_qwen3_vl.py:Qwen3VLTextRMSNorm.extra_repr": ["ModelTextRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "variance_epsilon", "weight"], "qwen3_vl/modeling_qwen3_vl.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cos", "def", "k", "k_embed", "q", "q_embed", "return", "rotate_half", "sin", "unsqueeze", "unsqueeze_dim"], "qwen3_vl/modeling_qwen3_vl.py:Qwen3VLTextAttention.__init__": ["Linear", "ModelTextAttention", "ModelTextConfig", "ModelTextRMSNorm", "Module", "None", "True", "__init__", "attention_bias", "attention_dropout", "bias", "class", "config", "def", "else", "eps", "getattr", "hasattr", "head_dim", "hidden_size", "if", "int", "is_causal", "k_norm", "k_proj", "layer_idx", "layer_type", "layer_types", "nn", "num_attention_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "q_norm", "q_proj", "rms_norm_eps", "scaling", "self", "super", "v_proj"], "qwen3_vl/modeling_qwen3_vl.py:Qwen3VLTextAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "FlashAttentionKwargs", "LongTensor", "ModelTextAttention", "Module", "None", "Tensor", "Unpack", "_attn_implementation", "apply_rotary_pos_emb", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "cache_kwargs", "cache_position", "class", "config", "contiguous", "cos", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "head_dim", "hidden_shape", "hidden_states", "if", "input_shape", "is", "k_norm", "k_proj", "key_states", "kwargs", "layer_idx", "nn", "not", "o_proj", "past_key_values", "position_embeddings", "q_norm", "q_proj", "query_states", "reshape", "return", "scaling", "self", "shape", "sin", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "view"], "qwen3_vl/modeling_qwen3_vl.py:Qwen3VLTextMLP.__init__": ["ACT2FN", "False", "Linear", "ModelTextMLP", "Module", "__init__", "act_fn", "bias", "class", "config", "def", "down_proj", "gate_proj", "hidden_act", "hidden_size", "intermediate_size", "nn", "self", "super", "up_proj"], "qwen3_vl/modeling_qwen3_vl.py:Qwen3VLTextMLP.forward": ["ModelTextMLP", "Module", "act_fn", "class", "def", "down_proj", "forward", "gate_proj", "nn", "return", "self", "up_proj", "x"], "qwen3_vl/modeling_qwen3_vl.py:Qwen3VLTextDecoderLayer.__init__": ["GradientCheckpointingLayer", "ModelTextAttention", "ModelTextConfig", "ModelTextDecoderLayer", "ModelTextMLP", "ModelTextRMSNorm", "__init__", "class", "config", "def", "eps", "hidden_size", "input_layernorm", "int", "layer_idx", "mlp", "post_attention_layernorm", "rms_norm_eps", "self", "self_attn", "super"], "qwen3_vl/modeling_qwen3_vl.py:Qwen3VLTextDecoderLayer.forward": ["Cache", "False", "GradientCheckpointingLayer", "LongTensor", "ModelTextDecoderLayer", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "attention_mask", "bool", "cache_position", "class", "def", "forward", "hidden_states", "input_layernorm", "kwargs", "mlp", "past_key_values", "position_embeddings", "position_ids", "post_attention_layernorm", "residual", "return", "self", "self_attn", "torch", "tuple", "use_cache"], "qwen3_vl/modeling_qwen3_vl.py:Qwen3VLPreTrainedModel._init_weights": ["ModelPreTrainedModel", "ModelVisionRotaryEmbedding", "PreTrainedModel", "_init_weights", "arange", "class", "copy_", "def", "dim", "dtype", "float", "if", "init", "inv_freq", "isinstance", "module", "self", "super", "theta", "torch"], "qwen3_vl/modeling_qwen3_vl.py:Qwen3VLVisionModel.__init__": ["Embedding", "False", "ModelPreTrainedModel", "ModelVisionBlock", "ModelVisionModel", "ModelVisionPatchEmbed", "ModelVisionPatchMerger", "ModelVisionRotaryEmbedding", "ModuleList", "None", "True", "_", "__init__", "blocks", "class", "config", "deepstack_merger_list", "deepstack_visual_indexes", "def", "depth", "for", "gradient_checkpointing", "head_dim", "hidden_size", "in", "inputs", "int", "kwargs", "len", "merger", "nn", "num_grid_per_side", "num_heads", "num_position_embeddings", "patch_embed", "patch_size", "pos_embed", "post_init", "range", "rotary_pos_emb", "self", "spatial_merge_size", "spatial_merge_unit", "super", "use_postshuffle_norm"], "qwen3_vl/modeling_qwen3_vl.py:Qwen3VLVisionModel.rot_pos_emb": ["ModelPreTrainedModel", "ModelVisionModel", "None", "Tensor", "arange", "block_cols", "block_rows", "class", "col_idx", "coords", "def", "device", "dim", "dtype", "embeddings", "empty", "expand", "flatten", "for", "freq_table", "grid_thw", "height", "if", "in", "int", "intra_col", "intra_row", "item", "long", "max", "max_hw", "merge_size", "merged_h", "merged_w", "num_frames", "num_tokens", "offset", "pos_ids", "prod", "repeat", "reshape", "return", "rot_pos_emb", "rotary_pos_emb", "row_idx", "self", "shape", "spatial_merge_size", "stack", "sum", "torch", "total_tokens", "width"], "qwen3_vl/modeling_qwen3_vl.py:Qwen3VLVisionModel.fast_pos_embed_interpolate": ["ModelPreTrainedModel", "ModelVisionModel", "None", "T", "_", "append", "base_h", "base_h_ceil", "cat", "class", "clip", "config", "def", "device", "dh", "dtype", "dw", "extend", "fast_pos_embed_interpolate", "flatten", "for", "grid_hs", "grid_thw", "grid_ts", "grid_ws", "h", "h_idxs", "h_idxs_ceil", "h_idxs_floor", "i", "idx_list", "idx_tensor", "in", "indices", "int", "linspace", "long", "max", "merge_size", "num_grid_per_side", "patch_pos_embeds", "patch_pos_embeds_permute", "permute", "pos_embed", "pos_embeds", "range", "repeat", "return", "self", "spatial_merge_size", "split", "t", "tensor", "to", "tolist", "torch", "view", "w", "w_idxs", "w_idxs_ceil", "w_idxs_floor", "weight", "weight_list", "weight_tensor", "weights", "zip"], "qwen3_vl/modeling_qwen3_vl.py:Qwen3VLVisionModel.forward": ["F", "ModelPreTrainedModel", "ModelVisionModel", "Tensor", "_", "append", "blk", "blocks", "cat", "class", "cos", "cu_seqlens", "cumsum", "deepstack_feature", "deepstack_feature_lists", "deepstack_merger_list", "deepstack_visual_indexes", "def", "dim", "dtype", "else", "emb", "enumerate", "fast_pos_embed_interpolate", "for", "forward", "grid_thw", "hidden_states", "if", "in", "index", "int32", "is_tracing", "jit", "kwargs", "layer_num", "merger", "pad", "patch_embed", "pos_embeds", "position_embeddings", "repeat_interleave", "reshape", "return", "rot_pos_emb", "rotary_pos_emb", "self", "seq_len", "sin", "size", "torch", "value"], "qwen3_vl/modeling_qwen3_vl.py:Qwen3VLTextModel.__init__": ["Embedding", "False", "ModelPreTrainedModel", "ModelTextConfig", "ModelTextDecoderLayer", "ModelTextModel", "ModelTextRMSNorm", "ModelTextRotaryEmbedding", "ModuleList", "__init__", "class", "config", "def", "embed_tokens", "eps", "for", "gradient_checkpointing", "hidden_size", "in", "layer_idx", "layers", "nn", "norm", "num_hidden_layers", "pad_token_id", "padding_idx", "post_init", "range", "rms_norm_eps", "rotary_emb", "self", "super", "vocab_size"], "qwen3_vl/modeling_qwen3_vl.py:Qwen3VLTextModel.forward": ["BaseModelOutputWithPast", "Cache", "DynamicCache", "FlashAttentionKwargs", "FloatTensor", "LongTensor", "ModelPreTrainedModel", "ModelTextModel", "None", "Tensor", "Unpack", "ValueError", "You", "_deepstack_process", "and", "arange", "attention_mask", "bool", "cache_position", "class", "config", "create_causal_mask", "decoder_layer", "deepstack_visual_embeds", "def", "device", "elif", "else", "embed_tokens", "enumerate", "exactly", "expand", "for", "forward", "get_seq_length", "hidden_states", "if", "in", "input_embeds", "input_ids", "inputs_embeds", "is", "is_tracing", "jit", "kwargs", "last_hidden_state", "layer_idx", "layer_outputs", "layers", "len", "list", "must", "ndim", "norm", "not", "of", "one", "or", "past_key_values", "past_seen_tokens", "position_embeddings", "position_ids", "r", "raise", "range", "return", "rotary_emb", "self", "shape", "specify", "text_position_ids", "torch", "tuple", "use_cache", "view", "visual_pos_masks"], "qwen3_vl/modeling_qwen3_vl.py:Qwen3VLTextModel._deepstack_process": ["ModelPreTrainedModel", "ModelTextModel", "Tensor", "_deepstack_process", "class", "clone", "def", "device", "dtype", "hidden_states", "local_this", "return", "self", "to", "torch", "visual_embeds", "visual_pos_masks"], "qwen3_vl/modeling_qwen3_vl.py:Qwen3VLModel.__init__": ["ModelModel", "ModelPreTrainedModel", "ModelTextModel", "ModelVisionModel", "None", "__init__", "_from_config", "class", "config", "def", "language_model", "post_init", "rope_deltas", "self", "super", "text_config", "vision_config", "visual"], "qwen3_vl/modeling_qwen3_vl.py:Qwen3VLModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "language_model", "return", "self"], "qwen3_vl/modeling_qwen3_vl.py:Qwen3VLModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "language_model", "self", "set_input_embeddings", "value"], "qwen3_vl/modeling_qwen3_vl.py:Qwen3VLModel.get_rope_index": ["False", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "True", "_", "and", "append", "arange", "argwhere", "attention_mask", "cat", "class", "config", "cumsum", "def", "device", "dim", "dtype", "ed", "ed_image", "ed_video", "else", "enumerate", "expand", "flatten", "for", "get_rope_index", "h", "h_index", "i", "if", "image_grid_thw", "image_index", "image_nums", "image_token_id", "in", "index", "input_ids", "input_tokens", "is", "item", "keepdim", "len", "list", "llm_grid_h", "llm_grid_t", "llm_grid_w", "llm_pos_ids_list", "llm_positions", "long", "masked_fill_", "max", "max_position_ids", "mrope_position_deltas", "not", "ones", "ones_like", "or", "position_ids", "range", "remain_images", "remain_videos", "repeat_interleave", "reshape", "return", "self", "shape", "spatial_merge_size", "squeeze", "st", "st_idx", "stack", "sum", "t", "t_index", "tensor", "text_len", "to", "tolist", "torch", "total_input_ids", "tuple", "unsqueeze", "video_grid_thw", "video_index", "video_nums", "video_token_id", "view", "vision_config", "vision_start_indices", "vision_start_token_id", "vision_tokens", "w", "w_index", "zeros"], "qwen3_vl/modeling_qwen3_vl.py:Qwen3VLModel.get_video_features": ["FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "class", "def", "get_image_features", "get_video_features", "pixel_values_videos", "return", "self", "torch", "video_grid_thw"], "qwen3_vl/modeling_qwen3_vl.py:Qwen3VLModel.get_image_features": ["FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "class", "deepstack_image_embeds", "def", "dtype", "get_image_features", "grid_thw", "image_embeds", "image_grid_thw", "pixel_values", "prod", "return", "self", "spatial_merge_size", "split", "split_sizes", "tolist", "torch", "type", "visual"], "qwen3_vl/modeling_qwen3_vl.py:Qwen3VLModel.get_placeholder_mask": ["FloatTensor", "Image", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "ValueError", "Videos", "all", "and", "class", "config", "def", "device", "do", "dtype", "else", "expand_as", "f", "features", "get_input_embeddings", "get_placeholder_mask", "if", "image", "image_features", "image_token_id", "input_ids", "inputs_embeds", "is", "long", "match", "n_image_tokens", "n_video_tokens", "not", "numel", "raise", "return", "self", "shape", "special_image_mask", "special_video_mask", "sum", "tensor", "to", "tokens", "torch", "unsqueeze", "video", "video_features", "video_token_id"], "qwen3_vl/modeling_qwen3_vl.py:Qwen3VLModel.forward": ["Cache", "FloatTensor", "LongTensor", "ModelModel", "ModelModelOutputWithPast", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "_", "add", "and", "append", "arange", "attention_mask", "batch_size", "cache_position", "cat", "class", "deepstack_image_embeds", "deepstack_video_embeds", "deepstack_visual_embeds", "def", "delta", "device", "dim", "dtype", "elif", "else", "embed_joint", "exactly", "expand", "for", "forward", "get_image_features", "get_input_embeddings", "get_placeholder_mask", "get_rope_index", "get_seq_length", "get_video_features", "if", "image_embeds", "image_features", "image_grid_thw", "image_mask", "image_mask_joint", "img_embed", "in", "input_ids", "inputs_embeds", "is", "kwargs", "language_model", "last_hidden_state", "masked_scatter", "must", "new_zeros", "not", "of", "one", "or", "outputs", "past_key_values", "past_key_values_length", "pixel_values", "pixel_values_videos", "position_ids", "r", "raise", "repeat_interleave", "return", "rope_deltas", "self", "seq_length", "shape", "specify", "sum", "to", "torch", "tuple", "unsqueeze", "vid_embed", "video_embeds", "video_features", "video_grid_thw", "video_mask", "video_mask_joint", "view", "visual_pos_masks", "zip"], "qwen3_vl/modeling_qwen3_vl.py:Qwen3VLForConditionalGeneration.__init__": ["False", "GenerationMixin", "Linear", "ModelForConditionalGeneration", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "text_config", "vocab_size"], "qwen3_vl/modeling_qwen3_vl.py:Qwen3VLForConditionalGeneration.get_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "model", "return", "self"], "qwen3_vl/modeling_qwen3_vl.py:Qwen3VLForConditionalGeneration.set_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "model", "self", "set_input_embeddings", "value"], "qwen3_vl/modeling_qwen3_vl.py:Qwen3VLForConditionalGeneration.get_video_features": ["FloatTensor", "GenerationMixin", "LongTensor", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "class", "def", "get_video_features", "model", "pixel_values_videos", "return", "self", "torch", "video_grid_thw"], "qwen3_vl/modeling_qwen3_vl.py:Qwen3VLForConditionalGeneration.get_image_features": ["FloatTensor", "GenerationMixin", "LongTensor", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "class", "def", "get_image_features", "image_grid_thw", "model", "pixel_values", "return", "self", "torch"], "qwen3_vl/modeling_qwen3_vl.py:Qwen3VLForConditionalGeneration.forward": ["Cache", "FloatTensor", "GenerationMixin", "LongTensor", "ModelCausalLMOutputWithPast", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "cache_position", "class", "config", "def", "else", "forward", "hidden_states", "if", "image_grid_thw", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "outputs", "past_key_values", "pixel_values", "pixel_values_videos", "position_ids", "r", "return", "rope_deltas", "self", "slice", "slice_indices", "text_config", "torch", "tuple", "video_grid_thw", "vocab_size"], "qwen3_vl/modeling_qwen3_vl.py:Qwen3VLForConditionalGeneration.prepare_inputs_for_generation": ["False", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "True", "and", "arange", "attention_mask", "batch_size", "cache_position", "cat", "class", "def", "delta", "device", "dim", "elif", "expand", "expand_as", "get", "get_rope_index", "if", "image_grid_thw", "in", "input_ids", "inputs_embeds", "is", "is_first_iteration", "kwargs", "model", "model_inputs", "not", "or", "past_key_values", "pixel_values", "pixel_values_videos", "position_ids", "prepare_inputs_for_generation", "repeat_interleave", "return", "rope_deltas", "self", "seq_length", "shape", "super", "text_positions", "torch", "use_cache", "video_grid_thw", "view", "vision_positions"], "qwen3_vl/modeling_qwen3_vl.py:Qwen3VLForConditionalGeneration._get_image_nums_and_video_nums": ["GenerationMixin", "LongTensor", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Tensor", "_get_image_nums_and_video_nums", "class", "config", "def", "device", "dim", "dims", "dtype", "else", "get_input_embeddings", "if", "image_mask", "image_nums", "image_token_id", "input_ids", "inputs_embeds", "is", "long", "not", "return", "roll", "self", "shifts", "sum", "tensor", "torch", "tuple", "video_mask", "video_nums", "video_token_id", "vision_first_mask", "vision_start_mask", "vision_start_token_id"], "qwen3_vl/modeling_qwen3_vl.py:Qwen3VLForConditionalGeneration._expand_inputs_for_generation": ["Any", "False", "GenerationMixin", "If", "LongTensor", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Tensor", "True", "ValueError", "_expand_dict_for_generation", "_expand_dict_for_generation_visual", "_expand_inputs_for_generation", "_get_image_nums_and_video_nums", "_repeat_interleave_samples", "and", "bool", "cache_position", "cat", "class", "cumsum", "cumulative_frame_counts", "cumulative_token_video_counts", "def", "defined", "dict", "dict_to_expand", "diff", "dim", "elif", "encoder_outputs", "expand_size", "for", "get", "if", "image_grid_thw", "image_nums", "in", "input_ids", "inputs_embeds", "int", "is", "is_encoder_decoder", "isinstance", "key", "lengths", "list", "make", "model_kwargs", "new_ones", "not", "pixel_values", "pixel_values_videos", "prod", "raise", "repeat", "repeat_args", "repeat_interleave", "repeat_times", "result", "return", "sample", "samples", "searchsorted", "self", "split", "str", "sum", "sure", "that", "torch", "tuple", "video_boundary_indices", "video_grid_thw", "video_nums", "visual_keys", "x"], "qwen3_vl_moe/modeling_qwen3_vl_moe.py:Qwen3VLMoeTextRMSNorm.__init__": ["ModelTextRMSNorm", "Module", "Parameter", "__init__", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "qwen3_vl_moe/modeling_qwen3_vl_moe.py:Qwen3VLMoeTextRMSNorm.forward": ["ModelTextRMSNorm", "Module", "True", "class", "def", "dtype", "float32", "forward", "hidden_states", "input_dtype", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "qwen3_vl_moe/modeling_qwen3_vl_moe.py:Qwen3VLMoeTextRMSNorm.extra_repr": ["ModelTextRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "variance_epsilon", "weight"], "qwen3_vl_moe/modeling_qwen3_vl_moe.py:Qwen3VLMoeTextExperts.__init__": ["ACT2FN", "ModelTextExperts", "Module", "Parameter", "__init__", "act_fn", "class", "config", "def", "down_proj", "empty", "gate_up_proj", "hidden_act", "hidden_dim", "hidden_size", "intermediate_dim", "moe_intermediate_size", "nn", "num_experts", "self", "super", "torch"], "qwen3_vl_moe/modeling_qwen3_vl_moe.py:Qwen3VLMoeTextExperts.forward": ["ModelTextExperts", "Module", "None", "Tensor", "act_fn", "chunk", "class", "continue", "current_hidden_states", "current_state", "def", "dim", "down_proj", "dtype", "expert_hit", "expert_idx", "expert_mask", "final_hidden_states", "for", "forward", "functional", "gate", "gate_up_proj", "greater", "hidden_states", "if", "in", "index_add_", "linear", "nn", "no_grad", "nonzero", "num_classes", "num_experts", "one_hot", "permute", "return", "self", "sum", "to", "token_idx", "top_k_index", "top_k_pos", "top_k_weights", "torch", "up", "where", "with", "zeros_like"], "qwen3_vl_moe/modeling_qwen3_vl_moe.py:Qwen3VLMoeTextTopKRouter.__init__": ["ModelTextTopKRouter", "Module", "Parameter", "__init__", "class", "config", "def", "hidden_dim", "hidden_size", "nn", "num_experts", "num_experts_per_tok", "self", "super", "top_k", "torch", "weight", "zeros"], "qwen3_vl_moe/modeling_qwen3_vl_moe.py:Qwen3VLMoeTextTopKRouter.forward": ["F", "ModelTextTopKRouter", "Module", "True", "class", "def", "dim", "dtype", "float", "forward", "functional", "hidden_dim", "hidden_states", "keepdim", "linear", "nn", "reshape", "return", "router_indices", "router_logits", "router_scores", "router_top_value", "self", "softmax", "sum", "to", "top_k", "topk", "torch", "weight"], "qwen3_vl_moe/modeling_qwen3_vl_moe.py:Qwen3VLMoeTextSparseMoeBlock.__init__": ["ModelTextConfig", "ModelTextExperts", "ModelTextSparseMoeBlock", "ModelTextTopKRouter", "Module", "__init__", "class", "config", "def", "experts", "gate", "nn", "self", "super"], "qwen3_vl_moe/modeling_qwen3_vl_moe.py:Qwen3VLMoeTextSparseMoeBlock.forward": ["ModelTextSparseMoeBlock", "Module", "Tensor", "_", "batch_size", "class", "def", "experts", "final_hidden_states", "forward", "gate", "hidden_dim", "hidden_states", "hidden_states_reshaped", "nn", "reshape", "return", "routing_weights", "selected_experts", "self", "sequence_length", "shape", "torch", "tuple", "view"], "qwen3_vl_moe/modeling_qwen3_vl_moe.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "qwen3_vl_moe/modeling_qwen3_vl_moe.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "qwen3_vl_moe/modeling_qwen3_vl_moe.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value", "value_states"], "qwen3_vl_moe/modeling_qwen3_vl_moe.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cos", "def", "k", "k_embed", "q", "q_embed", "return", "rotate_half", "sin", "unsqueeze", "unsqueeze_dim"], "qwen3_vl_moe/modeling_qwen3_vl_moe.py:Qwen3VLMoeTextAttention.__init__": ["Linear", "ModelTextAttention", "ModelTextConfig", "ModelTextRMSNorm", "Module", "None", "True", "__init__", "attention_bias", "attention_dropout", "bias", "class", "config", "def", "else", "eps", "getattr", "hasattr", "head_dim", "hidden_size", "if", "int", "is_causal", "k_norm", "k_proj", "layer_idx", "layer_type", "layer_types", "nn", "num_attention_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "q_norm", "q_proj", "rms_norm_eps", "scaling", "self", "super", "v_proj"], "qwen3_vl_moe/modeling_qwen3_vl_moe.py:Qwen3VLMoeTextAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "FlashAttentionKwargs", "LongTensor", "ModelTextAttention", "Module", "None", "Tensor", "Unpack", "_attn_implementation", "apply_rotary_pos_emb", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "cache_kwargs", "cache_position", "class", "config", "contiguous", "cos", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "head_dim", "hidden_shape", "hidden_states", "if", "input_shape", "is", "k_norm", "k_proj", "key_states", "kwargs", "layer_idx", "nn", "not", "o_proj", "past_key_values", "position_embeddings", "q_norm", "q_proj", "query_states", "reshape", "return", "scaling", "self", "shape", "sin", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "view"], "qwen3_vl_moe/modeling_qwen3_vl_moe.py:Qwen3VLMoeTextMLP.__init__": ["ACT2FN", "False", "Linear", "ModelTextMLP", "Module", "None", "__init__", "act_fn", "bias", "class", "config", "def", "down_proj", "else", "gate_proj", "hidden_act", "hidden_size", "if", "intermediate_size", "is", "nn", "self", "super", "up_proj"], "qwen3_vl_moe/modeling_qwen3_vl_moe.py:Qwen3VLMoeTextMLP.forward": ["ModelTextMLP", "Module", "act_fn", "class", "def", "down_proj", "forward", "gate_proj", "nn", "return", "self", "up_proj", "x"], "qwen3_vl_moe/modeling_qwen3_vl_moe.py:Qwen3VLMoeTextDecoderLayer.__init__": ["GradientCheckpointingLayer", "ModelTextAttention", "ModelTextConfig", "ModelTextDecoderLayer", "ModelTextMLP", "ModelTextRMSNorm", "ModelTextSparseMoeBlock", "__init__", "and", "class", "config", "decoder_sparse_step", "def", "else", "eps", "hidden_size", "if", "in", "input_layernorm", "int", "intermediate_size", "layer_idx", "mlp", "mlp_only_layers", "not", "num_experts", "post_attention_layernorm", "rms_norm_eps", "self", "self_attn", "super"], "qwen3_vl_moe/modeling_qwen3_vl_moe.py:Qwen3VLMoeTextDecoderLayer.forward": ["Cache", "False", "GradientCheckpointingLayer", "LongTensor", "ModelTextDecoderLayer", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "attention_mask", "bool", "cache_position", "class", "def", "forward", "hidden_states", "input_layernorm", "kwargs", "mlp", "past_key_values", "position_embeddings", "position_ids", "post_attention_layernorm", "residual", "return", "self", "self_attn", "torch", "tuple", "use_cache"], "qwen3_vl_moe/modeling_qwen3_vl_moe.py:Qwen3VLMoePreTrainedModel._init_weights": ["ModelPreTrainedModel", "ModelTextExperts", "ModelTextTopKRouter", "ModelVisionRotaryEmbedding", "PreTrainedModel", "_init_weights", "arange", "class", "config", "copy_", "def", "dim", "down_proj", "dtype", "elif", "else", "float", "gate_up_proj", "get_text_config", "getattr", "hasattr", "if", "init", "initializer_range", "inv_freq", "isinstance", "mean", "module", "normal_", "self", "std", "super", "theta", "torch", "weight"], "qwen3_vl_moe/modeling_qwen3_vl_moe.py:Qwen3VLMoeVisionRotaryEmbedding.__init__": ["False", "ModelVisionRotaryEmbedding", "Module", "None", "__init__", "arange", "class", "def", "dim", "dtype", "float", "int", "inv_freq", "nn", "persistent", "register_buffer", "self", "super", "theta", "torch"], "qwen3_vl_moe/modeling_qwen3_vl_moe.py:Qwen3VLMoeVisionRotaryEmbedding.forward": ["ModelVisionRotaryEmbedding", "Module", "Tensor", "arange", "class", "def", "device", "dtype", "forward", "freqs", "int", "inv_freq", "nn", "outer", "return", "self", "seq", "seqlen", "torch"], "qwen3_vl_moe/modeling_qwen3_vl_moe.py:Qwen3VLMoeVisionMLP.__init__": ["ACT2FN", "Linear", "ModelVisionMLP", "Module", "True", "__init__", "act_fn", "bias", "class", "config", "def", "hidden_act", "hidden_size", "intermediate_size", "linear_fc1", "linear_fc2", "nn", "self", "super"], "qwen3_vl_moe/modeling_qwen3_vl_moe.py:Qwen3VLMoeVisionMLP.forward": ["ModelVisionMLP", "Module", "act_fn", "class", "def", "forward", "hidden_state", "linear_fc1", "linear_fc2", "nn", "return", "self"], "qwen3_vl_moe/modeling_qwen3_vl_moe.py:Qwen3VLMoeVisionPatchEmbed.__init__": ["Conv3d", "ModelVisionPatchEmbed", "Module", "None", "True", "__init__", "bias", "class", "config", "def", "embed_dim", "hidden_size", "in_channels", "kernel_size", "nn", "patch_size", "proj", "self", "stride", "super", "temporal_patch_size"], "qwen3_vl_moe/modeling_qwen3_vl_moe.py:Qwen3VLMoeVisionPatchEmbed.forward": ["ModelVisionPatchEmbed", "Module", "Tensor", "class", "def", "dtype", "embed_dim", "forward", "hidden_states", "in_channels", "nn", "patch_size", "proj", "return", "self", "target_dtype", "temporal_patch_size", "to", "torch", "view", "weight"], "qwen3_vl_moe/modeling_qwen3_vl_moe.py:Qwen3VLMoeVisionPatchMerger.__init__": ["False", "GELU", "LayerNorm", "Linear", "ModelVisionConfig", "ModelVisionPatchMerger", "Module", "None", "__init__", "act_fn", "class", "config", "def", "else", "eps", "hidden_size", "if", "linear_fc1", "linear_fc2", "nn", "norm", "out_hidden_size", "self", "spatial_merge_size", "super", "use_postshuffle_norm"], "qwen3_vl_moe/modeling_qwen3_vl_moe.py:Qwen3VLMoeVisionPatchMerger.forward": ["ModelVisionPatchMerger", "Module", "Tensor", "act_fn", "class", "def", "else", "forward", "hidden_size", "if", "linear_fc1", "linear_fc2", "nn", "norm", "return", "self", "torch", "use_postshuffle_norm", "view", "x"], "qwen3_vl_moe/modeling_qwen3_vl_moe.py:apply_rotary_pos_emb_vision": ["Model_rotary_pos_emb_vision", "Tensor", "cos", "def", "dtype", "float", "k", "k_embed", "orig_k_dtype", "orig_q_dtype", "q", "q_embed", "return", "rotate_half", "sin", "to", "torch", "tuple", "unsqueeze"], "qwen3_vl_moe/modeling_qwen3_vl_moe.py:Qwen3VLMoeVisionAttention.__init__": ["False", "Linear", "ModelVisionAttention", "ModelVisionConfig", "Module", "None", "True", "__init__", "attention_dropout", "bias", "class", "config", "def", "dim", "head_dim", "hidden_size", "is_causal", "nn", "num_heads", "num_key_value_groups", "proj", "qkv", "scaling", "self", "super"], "qwen3_vl_moe/modeling_qwen3_vl_moe.py:Qwen3VLMoeVisionAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "False", "ModelVisionAttention", "Module", "None", "Tensor", "_", "_attn_implementation", "apply_rotary_pos_emb_vision", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_outputs", "cat", "class", "config", "contiguous", "cos", "cu_seq_lens_k", "cu_seq_lens_q", "cu_seqlens", "def", "dim", "dropout", "eager", "eager_attention_forward", "else", "flash", "for", "forward", "hidden_states", "if", "in", "is_causal", "k", "key_states", "kwargs", "lengths", "max", "max_length_k", "max_length_q", "max_seqlen", "nn", "not", "num_heads", "permute", "position_embeddings", "proj", "q", "qkv", "query_states", "reshape", "return", "rotary_pos_emb", "scaling", "self", "seq_length", "shape", "sin", "split", "splits", "tensor", "tolist", "torch", "training", "transpose", "tuple", "unbind", "unsqueeze", "v", "value_states", "zip"], "qwen3_vl_moe/modeling_qwen3_vl_moe.py:Qwen3VLMoeVisionBlock.__init__": ["GradientCheckpointingLayer", "LayerNorm", "ModelVisionAttention", "ModelVisionBlock", "ModelVisionMLP", "None", "__init__", "attn", "attn_implementation", "class", "config", "def", "eps", "hidden_size", "mlp", "nn", "norm1", "norm2", "sdpa", "self", "str", "super"], "qwen3_vl_moe/modeling_qwen3_vl_moe.py:Qwen3VLMoeVisionBlock.forward": ["GradientCheckpointingLayer", "ModelVisionBlock", "None", "Tensor", "attn", "class", "cu_seqlens", "def", "forward", "hidden_states", "kwargs", "mlp", "norm1", "norm2", "position_embeddings", "return", "rotary_pos_emb", "self", "torch", "tuple"], "qwen3_vl_moe/modeling_qwen3_vl_moe.py:Qwen3VLMoeVisionModel.__init__": ["Embedding", "False", "ModelPreTrainedModel", "ModelVisionBlock", "ModelVisionModel", "ModelVisionPatchEmbed", "ModelVisionPatchMerger", "ModelVisionRotaryEmbedding", "ModuleList", "None", "True", "_", "__init__", "blocks", "class", "config", "deepstack_merger_list", "deepstack_visual_indexes", "def", "depth", "for", "gradient_checkpointing", "head_dim", "hidden_size", "in", "inputs", "int", "kwargs", "len", "merger", "nn", "num_grid_per_side", "num_heads", "num_position_embeddings", "patch_embed", "patch_size", "pos_embed", "post_init", "range", "rotary_pos_emb", "self", "spatial_merge_size", "spatial_merge_unit", "super", "use_postshuffle_norm"], "qwen3_vl_moe/modeling_qwen3_vl_moe.py:Qwen3VLMoeVisionModel.rot_pos_emb": ["ModelPreTrainedModel", "ModelVisionModel", "None", "Tensor", "arange", "block_cols", "block_rows", "class", "col_idx", "coords", "def", "device", "dim", "dtype", "embeddings", "empty", "expand", "flatten", "for", "freq_table", "grid_thw", "height", "if", "in", "int", "intra_col", "intra_row", "item", "long", "max", "max_hw", "merge_size", "merged_h", "merged_w", "num_frames", "num_tokens", "offset", "pos_ids", "prod", "repeat", "reshape", "return", "rot_pos_emb", "rotary_pos_emb", "row_idx", "self", "shape", "spatial_merge_size", "stack", "sum", "torch", "total_tokens", "width"], "qwen3_vl_moe/modeling_qwen3_vl_moe.py:Qwen3VLMoeVisionModel.fast_pos_embed_interpolate": ["ModelPreTrainedModel", "ModelVisionModel", "None", "T", "_", "append", "base_h", "base_h_ceil", "cat", "class", "clip", "config", "def", "device", "dh", "dtype", "dw", "extend", "fast_pos_embed_interpolate", "flatten", "for", "grid_hs", "grid_thw", "grid_ts", "grid_ws", "h", "h_idxs", "h_idxs_ceil", "h_idxs_floor", "i", "idx_list", "idx_tensor", "in", "indices", "int", "linspace", "long", "max", "merge_size", "num_grid_per_side", "patch_pos_embeds", "patch_pos_embeds_permute", "permute", "pos_embed", "pos_embeds", "range", "repeat", "return", "self", "spatial_merge_size", "split", "t", "tensor", "to", "tolist", "torch", "view", "w", "w_idxs", "w_idxs_ceil", "w_idxs_floor", "weight", "weight_list", "weight_tensor", "weights", "zip"], "qwen3_vl_moe/modeling_qwen3_vl_moe.py:Qwen3VLMoeVisionModel.forward": ["F", "ModelPreTrainedModel", "ModelVisionModel", "Tensor", "_", "append", "blk", "blocks", "cat", "class", "cos", "cu_seqlens", "cumsum", "deepstack_feature", "deepstack_feature_lists", "deepstack_merger_list", "deepstack_visual_indexes", "def", "dim", "dtype", "else", "emb", "enumerate", "fast_pos_embed_interpolate", "for", "forward", "grid_thw", "hidden_states", "if", "in", "index", "int32", "is_tracing", "jit", "kwargs", "layer_num", "merger", "pad", "patch_embed", "pos_embeds", "position_embeddings", "repeat_interleave", "reshape", "return", "rot_pos_emb", "rotary_pos_emb", "self", "seq_len", "sin", "size", "torch", "value"], "qwen3_vl_moe/modeling_qwen3_vl_moe.py:Qwen3VLMoeTextRotaryEmbedding.__init__": ["Callable", "False", "ModelTextConfig", "ModelTextRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "get", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "mrope_section", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "qwen3_vl_moe/modeling_qwen3_vl_moe.py:Qwen3VLMoeTextRotaryEmbedding.compute_default_rope_parameters": ["ModelTextConfig", "ModelTextRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "qwen3_vl_moe/modeling_qwen3_vl_moe.py:Qwen3VLMoeTextRotaryEmbedding.forward": ["False", "ModelTextRotaryEmbedding", "Module", "None", "and", "apply_interleaved_mrope", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "mrope_section", "ndim", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "qwen3_vl_moe/modeling_qwen3_vl_moe.py:Qwen3VLMoeTextRotaryEmbedding.apply_interleaved_mrope": ["ModelTextRotaryEmbedding", "Module", "apply_interleaved_mrope", "class", "def", "dim", "enumerate", "for", "freqs", "freqs_t", "idx", "in", "length", "mrope_section", "nn", "offset", "return", "self", "slice", "start"], "qwen3_vl_moe/modeling_qwen3_vl_moe.py:Qwen3VLMoeTextModel.__init__": ["Embedding", "False", "ModelPreTrainedModel", "ModelTextConfig", "ModelTextDecoderLayer", "ModelTextModel", "ModelTextRMSNorm", "ModelTextRotaryEmbedding", "ModuleList", "__init__", "class", "config", "def", "embed_tokens", "eps", "for", "gradient_checkpointing", "hidden_size", "in", "layer_idx", "layers", "nn", "norm", "num_hidden_layers", "pad_token_id", "padding_idx", "post_init", "range", "rms_norm_eps", "rotary_emb", "self", "super", "vocab_size"], "qwen3_vl_moe/modeling_qwen3_vl_moe.py:Qwen3VLMoeTextModel.forward": ["BaseModelOutputWithPast", "Cache", "DynamicCache", "FlashAttentionKwargs", "FloatTensor", "LongTensor", "ModelPreTrainedModel", "ModelTextModel", "None", "Tensor", "Unpack", "ValueError", "You", "_deepstack_process", "and", "arange", "attention_mask", "bool", "cache_position", "class", "config", "create_causal_mask", "decoder_layer", "deepstack_visual_embeds", "def", "device", "elif", "else", "embed_tokens", "enumerate", "exactly", "expand", "for", "forward", "get_seq_length", "hidden_states", "if", "in", "input_embeds", "input_ids", "inputs_embeds", "is", "is_tracing", "jit", "kwargs", "last_hidden_state", "layer_idx", "layer_outputs", "layers", "len", "list", "must", "ndim", "norm", "not", "of", "one", "or", "past_key_values", "past_seen_tokens", "position_embeddings", "position_ids", "r", "raise", "range", "return", "rotary_emb", "self", "shape", "specify", "text_position_ids", "torch", "tuple", "use_cache", "view", "visual_pos_masks"], "qwen3_vl_moe/modeling_qwen3_vl_moe.py:Qwen3VLMoeTextModel._deepstack_process": ["ModelPreTrainedModel", "ModelTextModel", "Tensor", "_deepstack_process", "class", "clone", "def", "device", "dtype", "hidden_states", "local_this", "return", "self", "to", "torch", "visual_embeds", "visual_pos_masks"], "qwen3_vl_moe/modeling_qwen3_vl_moe.py:Qwen3VLMoeModel.__init__": ["ModelModel", "ModelPreTrainedModel", "ModelTextModel", "ModelVisionModel", "None", "__init__", "_from_config", "class", "config", "def", "language_model", "post_init", "rope_deltas", "self", "super", "text_config", "vision_config", "visual"], "qwen3_vl_moe/modeling_qwen3_vl_moe.py:Qwen3VLMoeModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "language_model", "return", "self"], "qwen3_vl_moe/modeling_qwen3_vl_moe.py:Qwen3VLMoeModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "language_model", "self", "set_input_embeddings", "value"], "qwen3_vl_moe/modeling_qwen3_vl_moe.py:Qwen3VLMoeModel.get_rope_index": ["False", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "True", "_", "and", "append", "arange", "argwhere", "attention_mask", "cat", "class", "config", "cumsum", "def", "device", "dim", "dtype", "ed", "ed_image", "ed_video", "else", "enumerate", "expand", "flatten", "for", "get_rope_index", "h", "h_index", "i", "if", "image_grid_thw", "image_index", "image_nums", "image_token_id", "in", "index", "input_ids", "input_tokens", "is", "item", "keepdim", "len", "list", "llm_grid_h", "llm_grid_t", "llm_grid_w", "llm_pos_ids_list", "llm_positions", "long", "masked_fill_", "max", "max_position_ids", "mrope_position_deltas", "not", "ones", "ones_like", "or", "position_ids", "range", "remain_images", "remain_videos", "repeat_interleave", "reshape", "return", "self", "shape", "spatial_merge_size", "squeeze", "st", "st_idx", "stack", "sum", "t", "t_index", "tensor", "text_len", "to", "tolist", "torch", "total_input_ids", "tuple", "unsqueeze", "video_grid_thw", "video_index", "video_nums", "video_token_id", "view", "vision_config", "vision_start_indices", "vision_start_token_id", "vision_tokens", "w", "w_index", "zeros"], "qwen3_vl_moe/modeling_qwen3_vl_moe.py:Qwen3VLMoeModel.get_video_features": ["FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "class", "def", "get_image_features", "get_video_features", "pixel_values_videos", "return", "self", "torch", "video_grid_thw"], "qwen3_vl_moe/modeling_qwen3_vl_moe.py:Qwen3VLMoeModel.get_image_features": ["FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "class", "deepstack_image_embeds", "def", "dtype", "get_image_features", "grid_thw", "image_embeds", "image_grid_thw", "pixel_values", "prod", "return", "self", "spatial_merge_size", "split", "split_sizes", "tolist", "torch", "type", "visual"], "qwen3_vl_moe/modeling_qwen3_vl_moe.py:Qwen3VLMoeModel.get_placeholder_mask": ["FloatTensor", "Image", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "ValueError", "Videos", "all", "and", "class", "config", "def", "device", "do", "dtype", "else", "expand_as", "f", "features", "get_input_embeddings", "get_placeholder_mask", "if", "image", "image_features", "image_token_id", "input_ids", "inputs_embeds", "is", "long", "match", "n_image_tokens", "n_video_tokens", "not", "numel", "raise", "return", "self", "shape", "special_image_mask", "special_video_mask", "sum", "tensor", "to", "tokens", "torch", "unsqueeze", "video", "video_features", "video_token_id"], "qwen3_vl_moe/modeling_qwen3_vl_moe.py:Qwen3VLMoeModel.forward": ["Cache", "FloatTensor", "LongTensor", "ModelModel", "ModelModelOutputWithPast", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "_", "add", "and", "append", "arange", "attention_mask", "batch_size", "cache_position", "cat", "class", "deepstack_image_embeds", "deepstack_video_embeds", "deepstack_visual_embeds", "def", "delta", "device", "dim", "dtype", "elif", "else", "embed_joint", "exactly", "expand", "for", "forward", "get_image_features", "get_input_embeddings", "get_placeholder_mask", "get_rope_index", "get_seq_length", "get_video_features", "if", "image_embeds", "image_features", "image_grid_thw", "image_mask", "image_mask_joint", "img_embed", "in", "input_ids", "inputs_embeds", "is", "kwargs", "language_model", "last_hidden_state", "masked_scatter", "must", "new_zeros", "not", "of", "one", "or", "outputs", "past_key_values", "past_key_values_length", "pixel_values", "pixel_values_videos", "position_ids", "r", "raise", "repeat_interleave", "return", "rope_deltas", "self", "seq_length", "shape", "specify", "sum", "to", "torch", "tuple", "unsqueeze", "vid_embed", "video_embeds", "video_features", "video_grid_thw", "video_mask", "video_mask_joint", "view", "visual_pos_masks", "zip"], "qwen3_vl_moe/modeling_qwen3_vl_moe.py:load_balancing_loss_func": ["Model_balancing_loss_func", "None", "Tensor", "_", "attention_mask", "batch_size", "cat", "compute_device", "concatenated_gate_logits", "def", "device", "dim", "else", "expand", "expert_attention_mask", "expert_mask", "float", "for", "functional", "gate_logits", "if", "in", "int", "is", "isinstance", "layer_gate", "mean", "nn", "not", "num_experts", "num_hidden_layers", "one_hot", "or", "overall_loss", "r", "reshape", "return", "router_per_expert_attention_mask", "router_prob_per_expert", "routing_weights", "selected_experts", "sequence_length", "shape", "softmax", "sum", "to", "tokens_per_expert", "top_k", "topk", "torch", "tuple", "unsqueeze"], "qwen3_vl_moe/modeling_qwen3_vl_moe.py:Qwen3VLMoeForConditionalGeneration.__init__": ["False", "GenerationMixin", "Linear", "ModelForConditionalGeneration", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "text_config", "vocab_size"], "qwen3_vl_moe/modeling_qwen3_vl_moe.py:Qwen3VLMoeForConditionalGeneration.get_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "model", "return", "self"], "qwen3_vl_moe/modeling_qwen3_vl_moe.py:Qwen3VLMoeForConditionalGeneration.set_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "model", "self", "set_input_embeddings", "value"], "qwen3_vl_moe/modeling_qwen3_vl_moe.py:Qwen3VLMoeForConditionalGeneration.get_video_features": ["FloatTensor", "GenerationMixin", "LongTensor", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "class", "def", "get_video_features", "model", "pixel_values_videos", "return", "self", "torch", "video_grid_thw"], "qwen3_vl_moe/modeling_qwen3_vl_moe.py:Qwen3VLMoeForConditionalGeneration.get_image_features": ["FloatTensor", "GenerationMixin", "LongTensor", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "class", "def", "get_image_features", "image_grid_thw", "model", "pixel_values", "return", "self", "torch"], "qwen3_vl_moe/modeling_qwen3_vl_moe.py:Qwen3VLMoeForConditionalGeneration.forward": ["Cache", "False", "FloatTensor", "GenerationMixin", "LongTensor", "ModelCausalLMOutputWithPast", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "aux_loss", "cache_position", "class", "config", "def", "device", "else", "forward", "get", "hidden_states", "if", "image_grid_thw", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "lm_head", "load_balancing_loss_func", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "num_experts", "num_experts_per_tok", "output_router_logits", "outputs", "past_key_values", "pixel_values", "pixel_values_videos", "position_ids", "r", "return", "rope_deltas", "router_aux_loss_coef", "router_logits", "self", "slice", "slice_indices", "text_config", "to", "torch", "tuple", "video_grid_thw", "vocab_size"], "qwen3_vl_moe/modeling_qwen3_vl_moe.py:Qwen3VLMoeForConditionalGeneration.prepare_inputs_for_generation": ["False", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "True", "and", "arange", "attention_mask", "batch_size", "cache_position", "cat", "class", "def", "delta", "device", "dim", "elif", "expand", "expand_as", "get", "get_rope_index", "if", "image_grid_thw", "in", "input_ids", "inputs_embeds", "is", "is_first_iteration", "kwargs", "model", "model_inputs", "not", "or", "past_key_values", "pixel_values", "pixel_values_videos", "position_ids", "prepare_inputs_for_generation", "repeat_interleave", "return", "rope_deltas", "self", "seq_length", "shape", "super", "text_positions", "torch", "use_cache", "video_grid_thw", "view", "vision_positions"], "qwen3_vl_moe/modeling_qwen3_vl_moe.py:Qwen3VLMoeForConditionalGeneration._get_image_nums_and_video_nums": ["GenerationMixin", "LongTensor", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Tensor", "_get_image_nums_and_video_nums", "class", "config", "def", "device", "dim", "dims", "dtype", "else", "get_input_embeddings", "if", "image_mask", "image_nums", "image_token_id", "input_ids", "inputs_embeds", "is", "long", "not", "return", "roll", "self", "shifts", "sum", "tensor", "torch", "tuple", "video_mask", "video_nums", "video_token_id", "vision_first_mask", "vision_start_mask", "vision_start_token_id"], "qwen3_vl_moe/modeling_qwen3_vl_moe.py:Qwen3VLMoeForConditionalGeneration._expand_inputs_for_generation": ["Any", "False", "GenerationMixin", "If", "LongTensor", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Tensor", "True", "ValueError", "_expand_dict_for_generation", "_expand_dict_for_generation_visual", "_expand_inputs_for_generation", "_get_image_nums_and_video_nums", "_repeat_interleave_samples", "and", "bool", "cache_position", "cat", "class", "cumsum", "cumulative_frame_counts", "cumulative_token_video_counts", "def", "defined", "dict", "dict_to_expand", "diff", "dim", "elif", "encoder_outputs", "expand_size", "for", "get", "if", "image_grid_thw", "image_nums", "in", "input_ids", "inputs_embeds", "int", "is", "is_encoder_decoder", "isinstance", "key", "lengths", "list", "make", "model_kwargs", "new_ones", "not", "pixel_values", "pixel_values_videos", "prod", "raise", "repeat", "repeat_args", "repeat_interleave", "repeat_times", "result", "return", "sample", "samples", "searchsorted", "self", "split", "str", "sum", "sure", "that", "torch", "tuple", "video_boundary_indices", "video_grid_thw", "video_nums", "visual_keys", "x"], "rag/modeling_rag.py:RagPreTrainedModel.from_pretrained_question_encoder_generator": ["AutoConfig", "AutoModel", "AutoModelForSeq2SeqLM", "If", "ModelConfig", "ModelPreTrainedModel", "ModelRetriever", "None", "PreTrainedModel", "True", "a", "an", "argument", "as", "assert", "be", "class", "cls", "config", "def", "defined", "del", "for", "from_pretrained", "from_pretrained_question_encoder_generator", "from_question_encoder_generator_configs", "generator", "generator_", "generator_config", "generator_model", "generator_pretrained_model_name_or_path", "get", "has", "if", "in", "is", "items", "key", "kwargs", "kwargs_generator", "kwargs_question_encoder", "len", "model", "not", "pop", "question_encoder", "question_encoder_", "question_encoder_config", "question_encoder_pretrained_model_name_or_path", "r", "retriever", "return", "return_unused_kwargs", "startswith", "str", "to", "value"], "rag/modeling_rag.py:RagModel.__init__": ["AutoModel", "AutoModelForSeq2SeqLM", "Either", "False", "ModelConfig", "ModelModel", "ModelPreTrainedModel", "ModelRetriever", "None", "PreTrainedConfig", "PreTrainedModel", "__init__", "a", "an", "and", "assert", "be", "but", "class", "config", "config_class", "configuration", "context_encoder_training", "ctx_encoder", "def", "else", "f", "from_config", "from_question_encoder_generator_configs", "generator", "has", "if", "is", "isinstance", "kwargs", "not", "of", "or", "post_init", "provided", "question_encoder", "r", "retriever", "self", "should", "super", "to", "type"], "rag/modeling_rag.py:RagModel.forward": ["Alternatively", "BoolTensor", "Cache", "FloatTensor", "LongTensor", "Make", "ModelModel", "ModelPreTrainedModel", "None", "RetrievAugLMOutput", "Tensor", "The", "True", "a", "and", "are", "assert", "attention_mask", "attentions", "be", "bmm", "bool", "but", "can", "class", "config", "context_attention_mask", "context_encoder_training", "context_input_ids", "cpu", "cross_attentions", "ctx_encoder", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_input_ids", "def", "detach", "device", "dim", "dimension", "doc_ids", "doc_scores", "dtype", "else", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "f", "first", "float32", "forward", "function", "gen_outputs", "generator", "generator_cross_attentions", "generator_dec_attentions", "generator_dec_hidden_states", "generator_enc_attentions", "generator_enc_hidden_states", "generator_enc_last_hidden_state", "has_to_retrieve", "hidden_states", "if", "input_ids", "int", "is", "kwargs", "logits", "multiple", "n_docs", "no", "not", "numpy", "of", "or", "output_attentions", "output_hidden_states", "output_retrieved", "passed", "passing", "past_key_values", "pooler_output", "prefix", "pt", "question_enc_attentions", "question_enc_hidden_states", "question_enc_outputs", "question_encoder", "question_encoder_last_hidden_state", "r", "repeat_interleave", "retrieved_doc_attention_mask", "retrieved_doc_embeds", "retrieved_doc_ids", "retrieved_doc_input_ids", "retriever", "retriever_outputs", "return", "return_dict", "return_tensors", "self", "set", "set_retriever", "shape", "should", "squeeze", "sure", "that", "the", "to", "tokenized_doc_attention_mask", "tokenized_doc_ids", "torch", "transpose", "tuple", "unsqueeze", "use_cache", "using", "view", "when", "you"], "rag/modeling_rag.py:RagSequenceForGeneration.__init__": ["Either", "Model", "ModelConfig", "ModelModel", "ModelPreTrainedModel", "ModelRetriever", "ModelSequenceForGeneration", "None", "PreTrainedConfig", "PreTrainedModel", "__init__", "a", "an", "and", "assert", "be", "class", "config", "configuration", "def", "encoder", "from_question_encoder_generator_configs", "generator", "has", "if", "is", "kwargs", "not", "or", "post_init", "provided", "question_encoder", "r", "retriever", "self", "super", "to"], "rag/modeling_rag.py:RagSequenceForGeneration.set_retriever": ["Model", "ModelPreTrainedModel", "ModelRetriever", "ModelSequenceForGeneration", "class", "def", "retriever", "self", "set_retriever"], "rag/modeling_rag.py:RagSequenceForGeneration.set_context_encoder_for_training": ["Model", "ModelPreTrainedModel", "ModelSequenceForGeneration", "PreTrainedModel", "True", "class", "context_encoder_training", "ctx_encoder", "def", "self", "set_context_encoder_for_training"], "rag/modeling_rag.py:RagSequenceForGeneration.forward": ["BoolTensor", "Cache", "False", "FloatTensor", "LongTensor", "Model", "ModelPreTrainedModel", "ModelSequenceForGeneration", "None", "RetrievAugLMMarginOutput", "Tensor", "attention_mask", "bool", "class", "config", "context_attention_mask", "context_input_ids", "decoder_attention_mask", "decoder_input_ids", "def", "doc_scores", "else", "encoder_outputs", "epsilon", "exclude_bos_score", "forward", "generator_cross_attentions", "generator_dec_attentions", "generator_dec_hidden_states", "generator_enc_attentions", "generator_enc_hidden_states", "generator_enc_last_hidden_state", "get_nll", "if", "input_ids", "int", "is", "kwargs", "label_smoothing", "labels", "logits", "loss", "n_docs", "not", "output_attentions", "output_hidden_states", "output_retrieved", "outputs", "past_key_values", "question_enc_attentions", "question_enc_hidden_states", "question_encoder_last_hidden_state", "r", "reduce_loss", "retrieved_doc_embeds", "retrieved_doc_ids", "return", "self", "torch", "tuple", "use_cache"], "rag/modeling_rag.py:RagSequenceForGeneration.retriever": ["Model", "ModelPreTrainedModel", "ModelSequenceForGeneration", "class", "def", "retriever", "return", "self"], "rag/modeling_rag.py:RagSequenceForGeneration.generator": ["Model", "ModelPreTrainedModel", "ModelSequenceForGeneration", "class", "def", "generator", "return", "self"], "rag/modeling_rag.py:RagSequenceForGeneration.question_encoder": ["Model", "ModelPreTrainedModel", "ModelSequenceForGeneration", "class", "def", "question_encoder", "return", "self"], "rag/modeling_rag.py:RagSequenceForGeneration.generate": ["Alternatively", "At", "FloatTensor", "LongTensor", "Make", "ModelPreTrainedModel", "ModelSequenceForGeneration", "None", "True", "_cat_and_pad", "a", "and", "append", "are", "assert", "attention_mask", "batch_size", "be", "bool", "can", "class", "config", "context_attention_mask", "context_input_ids", "cpu", "def", "detach", "device", "do_deduplication", "doc_scores", "dtype", "else", "exclude_bos_score", "float32", "for", "function", "generate", "generator", "generator_input_ids", "given", "hypos", "if", "in", "index", "individual_attention_mask", "individual_doc_scores", "individual_input_ids", "input_ids", "int", "is", "k", "labels", "least", "list", "loss", "model_kwargs", "must", "n_docs", "new_input_ids", "no", "not", "num_beams", "num_candidates", "num_doc_return_sequences", "num_return_sequences", "numpy", "of", "one", "or", "output_sequences", "outputs", "pad_token_id", "passed", "prefix", "pt", "question_encoder", "question_hidden_states", "range", "repeat", "retriever", "return", "return_tensors", "self", "set", "set_retriever", "shape", "stack", "str", "sure", "that", "the", "to", "tolist", "top_cand_inds", "topk", "torch", "using", "values", "you"], "rag/modeling_rag.py:RagSequenceForGeneration.get_nll": ["False", "ModelPreTrainedModel", "ModelSequenceForGeneration", "Model_logprobs", "None", "True", "_mask_pads", "all", "and", "any", "assert", "bos_token_id", "cat", "class", "config", "def", "dim", "doc_logprobs", "doc_scores", "else", "eps_i", "epsilon", "eq", "exclude_bos_score", "fill_", "first_token_scores", "functional", "gather", "generator", "get_nll", "if", "index", "is", "keepdim", "ll", "log_softmax", "logsumexp", "loss", "masked_fill_", "n_docs", "new", "nll_loss", "nn", "not", "or", "pad_mask", "pad_token_id", "reduce_loss", "remainder", "repeat", "return", "second_token_scores", "self", "seq_logits", "seq_logprobs", "shape", "size", "smooth_loss", "smooth_obj", "squeeze", "sum", "target", "torch", "unsqueeze", "use_bos", "view"], "rag/modeling_rag.py:RagSequenceForGeneration._cat_and_pad": ["ModelPreTrainedModel", "ModelSequenceForGeneration", "_cat_and_pad", "class", "def", "fill_", "for", "in", "ind", "max", "new", "output", "pad_token_id", "return", "shape", "sum", "t", "tensors"], "rag/modeling_rag.py:RagTokenForGeneration.__init__": ["Either", "GenerationMixin", "Model", "ModelConfig", "ModelModel", "ModelPreTrainedModel", "ModelRetriever", "ModelTokenForGeneration", "None", "PreTrainedConfig", "PreTrainedModel", "__init__", "a", "an", "and", "assert", "be", "class", "config", "configuration", "def", "encoder", "from_question_encoder_generator_configs", "generator", "has", "if", "is", "kwargs", "not", "or", "post_init", "provided", "question_encoder", "r", "retriever", "self", "super", "to"], "rag/modeling_rag.py:RagTokenForGeneration.set_retriever": ["GenerationMixin", "Model", "ModelPreTrainedModel", "ModelRetriever", "ModelTokenForGeneration", "class", "def", "retriever", "self", "set_retriever"], "rag/modeling_rag.py:RagTokenForGeneration.set_context_encoder_for_training": ["GenerationMixin", "Model", "ModelPreTrainedModel", "ModelTokenForGeneration", "PreTrainedModel", "True", "class", "context_encoder_training", "ctx_encoder", "def", "self", "set_context_encoder_for_training"], "rag/modeling_rag.py:RagTokenForGeneration.prepare_inputs_for_generation": ["GenerationMixin", "ModelPreTrainedModel", "ModelTokenForGeneration", "None", "True", "attention_mask", "class", "context_attention_mask", "decoder_input_ids", "def", "do_marginalize", "doc_scores", "encoder_outputs", "if", "input_ids", "is", "kwargs", "n_docs", "not", "past_key_values", "prepare_inputs_for_generation", "return", "self", "use_cache"], "rag/modeling_rag.py:RagTokenForGeneration.retriever": ["GenerationMixin", "Model", "ModelPreTrainedModel", "ModelTokenForGeneration", "class", "def", "retriever", "return", "self"], "rag/modeling_rag.py:RagTokenForGeneration.generator": ["GenerationMixin", "Model", "ModelPreTrainedModel", "ModelTokenForGeneration", "class", "def", "generator", "return", "self"], "rag/modeling_rag.py:RagTokenForGeneration.question_encoder": ["GenerationMixin", "Model", "ModelPreTrainedModel", "ModelTokenForGeneration", "class", "def", "question_encoder", "return", "self"], "rag/modeling_rag.py:RagTokenForGeneration._reorder_cache": ["EncoderDecoderCache", "GenerationMixin", "ModelPreTrainedModel", "ModelTokenForGeneration", "_reorder_cache", "_reorder_stacked", "beam_idx", "class", "cross_attention_cache", "cross_attention_k", "cross_attention_v", "def", "device", "else", "for", "hidden_states", "idx", "if", "in", "index_select", "isinstance", "keys", "layers", "len", "n_docs", "new_order", "new_tuple", "past_key_values", "range", "reordered_past", "result", "return", "self_attention_cache", "self_attention_k", "self_attention_v", "shape", "to", "type", "values", "view", "x"], "rag/modeling_rag.py:RagTokenForGeneration.marginalize": ["GenerationMixin", "ModelPreTrainedModel", "ModelTokenForGeneration", "None", "class", "config", "def", "dim", "doc_logprobs", "doc_scores", "else", "functional", "if", "is", "log_prob_sum", "log_softmax", "logsumexp", "marginalize", "n_docs", "nn", "not", "return", "self", "seq_logits", "seq_logprobs", "shape", "size", "torch", "unsqueeze", "view"], "rag/modeling_rag.py:RagTokenForGeneration.forward": ["BoolTensor", "Cache", "False", "FloatTensor", "GenerationMixin", "LongTensor", "Model", "ModelPreTrainedModel", "ModelTokenForGeneration", "None", "RetrievAugLMMarginOutput", "Tensor", "assert", "attention_mask", "bool", "class", "config", "context_attention_mask", "context_input_ids", "decoder_attention_mask", "decoder_input_ids", "def", "do_marginalize", "doc_scores", "else", "encoder_outputs", "epsilon", "forward", "generator_cross_attentions", "generator_dec_attentions", "generator_dec_hidden_states", "generator_enc_attentions", "generator_enc_hidden_states", "generator_enc_last_hidden_state", "get_nll", "if", "input_ids", "int", "is", "kwargs", "label_smoothing", "labels", "logits", "loss", "marginalize", "n_docs", "not", "output_attentions", "output_hidden_states", "output_retrieved", "outputs", "past_key_values", "question_enc_attentions", "question_enc_hidden_states", "question_encoder_last_hidden_state", "r", "reduce_loss", "retrieved_doc_embeds", "retrieved_doc_ids", "return", "self", "torch", "tuple", "use_cache"], "rag/modeling_rag.py:RagTokenForGeneration.generate": ["BEAM_SAMPLE", "BEAM_SEARCH", "Callable", "False", "FloatTensor", "GENERATION_MODES_MAPPING", "GREEDY_SEARCH", "GenerationConfig", "GenerationMixin", "GenerationMode", "LogitsProcessorList", "LongTensor", "Model", "ModelPreTrainedModel", "ModelTokenForGeneration", "None", "Please", "SAMPLE", "StoppingCriteriaList", "Tensor", "The", "True", "ValueError", "_extract_generation_mode_kwargs", "_get_logits_processor", "_get_stopping_criteria", "_prefill", "_prepare_cache_for_generation", "_prepare_generation_config", "_prepare_special_tokens", "_validate_generation_mode", "_validate_model_kwargs", "a", "and", "assert", "attention_mask", "batch_size", "be", "bmm", "but", "check", "class", "compatible", "config", "context_attention_mask", "context_input_ids", "copy", "cpu", "decoder_start_token_id", "decoding_method", "def", "detach", "device", "dim", "dimension", "doc_scores", "dtype", "else", "encoder", "encoder_input_ids", "encoder_outputs", "expand", "extend_enc_output", "f", "first", "float32", "full", "generate", "generation", "generation_config", "generation_mode", "generation_mode_kwargs", "generator", "get", "get_encoder", "get_generation_mode", "getattr", "if", "in", "input_ids", "input_ids_seq_length", "int", "is", "kwargs", "kwargs_has_attention_mask", "last_hidden_state", "list", "logits_processor", "long", "max_cache_length", "max_length", "model", "model_kwargs", "multiple", "n_docs", "next", "not", "num_beams", "numpy", "of", "out", "parameters", "prefill_outputs", "prefix", "prefix_allowed_tokens_fn", "prepared_logits_processor", "prepared_stopping_criteria", "pt", "question_encoder", "question_hidden_states", "raise", "repeat_interleave", "reshape", "retrieved_doc_embeds", "retriever", "return", "return_dict", "return_tensors", "self", "shape", "should", "squeeze", "stopping_criteria", "tensor", "to", "torch", "transpose", "type", "unsqueeze", "with", "your"], "rag/modeling_rag.py:RagTokenForGeneration._temporary_reorder_cache": ["GenerationMixin", "ModelPreTrainedModel", "ModelTokenForGeneration", "_reorder_cache", "_temporary_reorder_cache", "beam_idx", "class", "def", "past_key_values", "return", "self"], "rag/modeling_rag.py:RagTokenForGeneration.get_input_embeddings": ["GenerationMixin", "Model", "ModelPreTrainedModel", "ModelTokenForGeneration", "class", "def", "generator", "get_input_embeddings", "return", "self"], "rag/modeling_rag.py:RagTokenForGeneration.get_output_embeddings": ["GenerationMixin", "Model", "ModelPreTrainedModel", "ModelTokenForGeneration", "class", "def", "generator", "get_output_embeddings", "return", "self"], "rag/modeling_rag.py:RagTokenForGeneration.set_output_embeddings": ["GenerationMixin", "Model", "ModelPreTrainedModel", "ModelTokenForGeneration", "class", "def", "generator", "new_embeddings", "return", "self", "set_output_embeddings"], "rag/modeling_rag.py:RagTokenForGeneration.shift_tokens_right": ["GenerationMixin", "ModelPreTrainedModel", "ModelTokenForGeneration", "None", "class", "clone", "config", "decoder_start_token_id", "def", "if", "input_ids", "is", "new_zeros", "return", "self", "shape", "shift_tokens_right", "shifted_input_ids", "start_token_id"], "rag/modeling_rag.py:RagTokenForGeneration.get_nll": ["False", "GenerationMixin", "ModelPreTrainedModel", "ModelTokenForGeneration", "Model_logprobs", "None", "True", "_mask_pads", "any", "assert", "cat", "class", "config", "def", "dim", "doc_scores", "else", "eps_i", "epsilon", "eq", "fill_", "gather", "generator", "get_nll", "if", "index", "is", "keepdim", "ll", "loss", "marginalize", "masked_fill_", "n_docs", "new", "nll_loss", "not", "pad_mask", "pad_token_id", "reduce_loss", "return", "self", "seq_logits", "shape", "size", "smooth_loss", "smooth_obj", "squeeze", "sum", "target", "torch", "unsqueeze"], "recurrent_gemma/modeling_recurrent_gemma.py:RecurrentGemmaRMSNorm.__init__": ["ModelRMSNorm", "Module", "Parameter", "__init__", "class", "def", "dim", "eps", "float", "int", "nn", "self", "super", "torch", "weight", "zeros"], "recurrent_gemma/modeling_recurrent_gemma.py:RecurrentGemmaRMSNorm._norm": ["ModelRMSNorm", "Module", "True", "_norm", "class", "def", "eps", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "torch", "x"], "recurrent_gemma/modeling_recurrent_gemma.py:RecurrentGemmaRMSNorm.forward": ["ModelRMSNorm", "Module", "_norm", "class", "def", "float", "forward", "nn", "output", "return", "self", "type_as", "weight", "x"], "recurrent_gemma/modeling_recurrent_gemma.py:RecurrentGemmaRMSNorm.extra_repr": ["ModelRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "weight"], "recurrent_gemma/modeling_recurrent_gemma.py:RecurrentGemmaRotaryEmbedding.__init__": ["Callable", "False", "ModelConfig", "ModelRotaryEmbedding", "Module", "None", "RoPE", "ValueError", "__init__", "attention_scaling", "but", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "does", "f", "got", "if", "inv_freq", "nn", "not", "original_inv_freq", "other", "persistent", "raise", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super", "support", "than", "types"], "recurrent_gemma/modeling_recurrent_gemma.py:RecurrentGemmaRotaryEmbedding.compute_default_rope_parameters": ["ModelConfig", "ModelRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "get", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "partial_rotary_factor", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "recurrent_gemma/modeling_recurrent_gemma.py:RecurrentGemmaRotaryEmbedding.forward": ["False", "ModelRotaryEmbedding", "Module", "None", "and", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "recurrent_gemma/modeling_recurrent_gemma.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "recurrent_gemma/modeling_recurrent_gemma.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cos", "def", "k", "k_embed", "q", "q_embed", "return", "rotate_half", "sin", "unsqueeze", "unsqueeze_dim"], "recurrent_gemma/modeling_recurrent_gemma.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "recurrent_gemma/modeling_recurrent_gemma.py:RecurrentGemmaSdpaAttention.__init__": ["Linear", "ModelConfig", "ModelRotaryEmbedding", "ModelSdpaAttention", "Module", "True", "__init__", "attention_bias", "attention_dropout", "bias", "class", "config", "def", "head_dim", "hidden_size", "k_proj", "nn", "num_attention_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "partial_rotary_factor", "q_proj", "rotary_emb", "self", "super", "v_proj"], "recurrent_gemma/modeling_recurrent_gemma.py:RecurrentGemmaSdpaAttention.forward": ["False", "LongTensor", "ModelSdpaAttention", "Module", "None", "Tensor", "_", "_update_cache", "and", "apply_rotary_pos_emb", "attention_dropout", "attention_mask", "attn_mask", "attn_output", "bool", "bsz", "cache_kwargs", "cache_position", "cat", "causal_mask", "chunk", "class", "contiguous", "cos", "def", "dim", "dropout_p", "else", "forward", "functional", "hasattr", "head_dim", "hidden_size", "hidden_states", "if", "int", "is", "k_proj", "key_pass", "key_rot", "key_states", "nn", "not", "num_attention_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "partial_rotary_factor", "position_ids", "q_len", "q_proj", "query_pass", "query_rot", "query_states", "repeat_kv", "return", "rotary_emb", "scale", "scaled_dot_product_attention", "self", "shape", "sin", "size", "torch", "training", "transpose", "tuple", "use_cache", "v_proj", "value_states", "view"], "recurrent_gemma/modeling_recurrent_gemma.py:RecurrentGemmaSdpaAttention._setup_cache": ["ModelSdpaAttention", "Module", "None", "_setup_cache", "and", "attention_window_size", "batch_size", "cache_shape", "class", "config", "def", "device", "dtype", "else", "float32", "head_dim", "if", "is", "key_states", "nn", "not", "num_key_value_heads", "self", "torch", "value_states", "zeros"], "recurrent_gemma/modeling_recurrent_gemma.py:RecurrentGemmaSdpaAttention._update_cache": ["ModelSdpaAttention", "Module", "_update_cache", "attention_window_size", "cache_kwargs", "cache_position", "clamp", "class", "config", "cumsum", "def", "device", "dtype", "else", "get", "if", "indices", "int", "k_out", "key_states", "long", "nn", "ones", "return", "self", "shape", "slicing", "to", "to_shift", "torch", "v_out", "value_states"], "recurrent_gemma/modeling_recurrent_gemma.py:SqrtBoundDerivative.forward": ["Function", "Model", "ModelBoundDerivative", "Tensor", "autograd", "class", "ctx", "def", "forward", "return", "save_for_backward", "torch", "x"], "recurrent_gemma/modeling_recurrent_gemma.py:SqrtBoundDerivative.backward": ["Function", "Model", "ModelBoundDerivative", "Tensor", "_MAX_Model_GRADIENT", "autograd", "backward", "class", "clip", "clipped_x_times_4", "ctx", "def", "grad_output", "min", "return", "saved_tensors", "torch", "x"], "recurrent_gemma/modeling_recurrent_gemma.py:RecurrentGemmaRglru.__init__": ["ModelRglru", "Model_gate_bias", "Model_gate_weight", "Model_param", "Model_states", "Module", "None", "Parameter", "__init__", "block_width", "class", "config", "def", "empty", "input_gate_bias", "input_gate_weight", "lru_width", "nn", "num_attention_heads", "self", "super", "torch"], "recurrent_gemma/modeling_recurrent_gemma.py:RecurrentGemmaRglru.forward": ["ModelRglru", "Model_gate", "Model_gate_bias", "Model_gate_weight", "Model_param", "Model_states", "Module", "None", "SqrtBoundDerivative", "Tensor", "_rnn_scan", "a_square", "activations", "apply", "baddbmm", "batch_size", "block_width", "class", "def", "dtype", "exp", "forward", "functional", "gated_inputs", "hidden_states", "if", "input_gate", "input_gate_bias", "input_gate_weight", "is_tracing", "log_Model_gate", "lru_width", "multiplier", "nn", "normalized_x", "not", "num_attention_heads", "permute", "position_ids", "res", "reset", "reshape", "reshape_act", "return", "self", "seq_len", "shape", "sigmoid", "softplus", "torch", "transpose", "tuple", "type"], "recurrent_gemma/modeling_recurrent_gemma.py:RecurrentGemmaRglru._rnn_scan": ["ModelRglru", "Model_gate", "Model_states", "Module", "None", "Tensor", "_rnn_scan", "acc_dtype", "class", "contextualized_states", "def", "device", "dtype", "else", "float32", "for", "hidden_states", "if", "in", "is", "nn", "range", "reset", "return", "self", "shape", "t", "to", "torch", "tuple", "type", "zeros", "zeros_like"], "recurrent_gemma/modeling_recurrent_gemma.py:RecurrentGemmaRecurrentBlock.__init__": ["ACT2FN", "Conv1d", "Linear", "ModelModelBlock", "ModelRglru", "Module", "None", "__init__", "act_fn", "class", "config", "conv1d_state", "conv1d_width", "conv_1d", "def", "groups", "hidden_activation", "hidden_size", "in_features", "kernel_size", "linear_out", "linear_x", "linear_y", "lru_width", "nn", "out_features", "padding", "rg_lru", "self", "super"], "recurrent_gemma/modeling_recurrent_gemma.py:RecurrentGemmaRecurrentBlock.forward": ["ModelModelBlock", "Model_states", "Module", "None", "Tensor", "True", "_", "act_fn", "attention_mask", "batch_size", "bias", "bool", "cache_position", "cat", "class", "conv1d_state", "conv1d_width", "conv_1d", "conv_state", "def", "device", "dict", "dim", "dtype", "else", "float32", "forward", "functional", "hidden_size", "hidden_states", "if", "input_states", "is", "linear_out", "linear_x", "linear_y", "lru_width", "nn", "or", "pad", "position_ids", "return", "rg_lru", "self", "seq_len", "shape", "str", "sum", "torch", "transpose", "tuple", "unsqueeze", "use_cache", "weight", "x_branch", "y_branch", "zeros"], "recurrent_gemma/modeling_recurrent_gemma.py:RecurrentGemmaRecurrentBlock._setup_cache": ["ModelModelBlock", "Model_states", "Module", "_setup_cache", "batch", "class", "conv1d_state", "conv1d_width", "def", "device", "dtype", "float32", "hidden_size", "lru_width", "nn", "rg_lru", "self", "torch", "zeros"], "recurrent_gemma/modeling_recurrent_gemma.py:RecurrentGemmaMlp.__init__": ["ACT2FN", "Linear", "ModelMlp", "Module", "True", "__init__", "act_fn", "bias", "class", "config", "def", "down_proj", "gate_proj", "hidden_activation", "hidden_size", "intermediate_size", "nn", "self", "super", "up_proj"], "recurrent_gemma/modeling_recurrent_gemma.py:RecurrentGemmaMlp.forward": ["ModelMlp", "Module", "act_fn", "class", "def", "down_proj", "forward", "gate", "gate_proj", "hidden_states", "nn", "return", "self", "up_proj"], "recurrent_gemma/modeling_recurrent_gemma.py:RecurrentGemmaDecoderLayer.__init__": ["GradientCheckpointingLayer", "ModelDecoderLayer", "ModelMlp", "ModelRMSNorm", "TEMPORAL_BLOCK_CLASSES", "__init__", "channel_pre_norm", "class", "config", "def", "eps", "hidden_size", "layer_idx", "layers_block_type", "mlp_block", "rms_norm_eps", "self", "super", "temporal_block", "temporal_pre_norm"], "recurrent_gemma/modeling_recurrent_gemma.py:RecurrentGemmaDecoderLayer.forward": ["GradientCheckpointingLayer", "ModelDecoderLayer", "None", "Tensor", "activations", "attention_mask", "bool", "cache_position", "channel_pre_norm", "class", "def", "dict", "forward", "hidden_states", "inputs_normalized", "mlp_block", "position_ids", "raw_activations", "residual", "return", "self", "str", "temporal_block", "temporal_pre_norm", "torch", "tuple", "use_cache"], "recurrent_gemma/modeling_recurrent_gemma.py:RecurrentGemmaPreTrainedModel._init_weights": ["Conv1d", "Embedding", "False", "Linear", "ModelModel", "ModelModelBlock", "ModelPreTrainedModel", "ModelRMSNorm", "ModelRglru", "ModelSdpaAttention", "Model_gate_bias", "Model_gate_weight", "Model_param", "None", "PreTrainedModel", "_init_weights", "_is_hf_initialized", "and", "bias", "class", "config", "constant_", "conv1d_width", "copy_", "def", "elif", "empty_like", "exp_", "final_w_init_variance_scale", "getattr", "hidden_size", "if", "init", "input_gate_bias", "input_gate_weight", "is", "isinstance", "k_proj", "linear_out", "linear_x", "linear_y", "log_", "lru_width", "math", "mean", "module", "mul_", "neg_", "nn", "normal_", "normalizer", "not", "num_attention_heads", "o_proj", "padding_idx", "q_proj", "self", "sqrt", "std", "sub_", "torch", "uniform_", "v_proj", "w_init_variance_scale", "weight", "zeros_"], "recurrent_gemma/modeling_recurrent_gemma.py:RecurrentGemmaPreTrainedModel._setup_cache": ["ModelPreTrainedModel", "PreTrainedModel", "_setup_cache", "batch", "class", "config", "def", "device", "dtype", "for", "getattr", "in", "layer", "layers", "model", "self", "temporal_block"], "recurrent_gemma/modeling_recurrent_gemma.py:RecurrentGemmaPreTrainedModel.reset_cache": ["ModelPreTrainedModel", "PreTrainedModel", "batch", "class", "def", "device", "dtype", "pass", "reset_cache", "self"], "recurrent_gemma/modeling_recurrent_gemma.py:RecurrentGemmaModel.__init__": ["Embedding", "False", "ModelConfig", "ModelDecoderLayer", "ModelModel", "ModelPreTrainedModel", "ModelRMSNorm", "ModuleList", "__init__", "bfloat16", "class", "config", "def", "dtype", "embed_tokens", "eps", "final_norm", "for", "gradient_checkpointing", "hidden_size", "in", "layer_idx", "layers", "nn", "normalizer", "num_hidden_layers", "pad_token_id", "padding_idx", "persistent", "post_init", "range", "register_buffer", "rms_norm_eps", "self", "super", "tensor", "torch", "vocab_size"], "recurrent_gemma/modeling_recurrent_gemma.py:RecurrentGemmaModel.forward": ["BaseModelOutputWithNoAttention", "False", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Setting", "Tensor", "True", "ValueError", "You", "_setup_cache", "_update_causal_mask", "all_hidden_states", "and", "arange", "attention_mask", "bool", "cache_position", "causal_mask", "checkpointing", "class", "config", "def", "device", "dtype", "else", "embed_tokens", "enumerate", "exactly", "final_norm", "for", "forward", "gradient", "gradient_checkpointing", "hidden_states", "i", "if", "in", "incompatible", "input_ids", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layers", "logger", "must", "normalizer", "not", "of", "one", "or", "output_hidden_states", "position_ids", "raise", "residual_block", "return", "return_dict", "self", "shape", "specify", "torch", "training", "tuple", "type", "unsqueeze", "use_cache", "use_return_dict", "v", "warning_once", "with"], "recurrent_gemma/modeling_recurrent_gemma.py:RecurrentGemmaModel._update_causal_mask": ["AttentionMaskConverter", "ModelModel", "ModelPreTrainedModel", "None", "_unmask_unattended", "_update_causal_mask", "and", "arange", "attention_mask", "attention_window_size", "cache_position", "causal_mask", "class", "clone", "config", "cuda", "def", "device", "diagonal", "dim", "dtype", "eq", "expand", "fill_value", "finfo", "full", "if", "in", "input_tensor", "is", "mask_length", "masked_fill", "max", "min", "min_dtype", "not", "npu", "padding_mask", "reshape", "return", "self", "sequence_length", "shape", "target_length", "torch", "triu", "type", "xpu"], "recurrent_gemma/modeling_recurrent_gemma.py:RecurrentGemmaForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "ModelForCausalLM", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "vocab_size"], "recurrent_gemma/modeling_recurrent_gemma.py:RecurrentGemmaForCausalLM.forward": ["CausalLMOutput", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForCausalLM", "ModelPreTrainedModel", "None", "Tensor", "True", "attention_mask", "bool", "cache_position", "cap", "class", "config", "def", "else", "forward", "functional", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "lm_head", "logits", "logits_soft_cap", "logits_to_keep", "loss", "loss_function", "model", "nn", "not", "output", "output_hidden_states", "outputs", "position_ids", "r", "return", "return_dict", "self", "slice", "slice_indices", "tanh", "torch", "tuple", "use_cache", "use_return_dict", "vocab_size"], "reformer/modeling_reformer.py:ReformerDynamicCache.__init__": ["Iterable", "ModelDynamicCache", "None", "Tensor", "__init__", "_distributed_cache_data", "_seen_tokens", "append", "buckets", "buckets_cache", "class", "def", "for", "if", "in", "is", "list", "not", "self", "states", "states_cache", "torch"], "reformer/modeling_reformer.py:ReformerDynamicCache.__len__": ["ModelDynamicCache", "__len__", "class", "def", "len", "return", "self", "states_cache"], "reformer/modeling_reformer.py:ReformerDynamicCache.update": ["Any", "ModelDynamicCache", "None", "Tensor", "_seen_tokens", "append", "buckets", "buckets_cache", "cache_kwargs", "cat", "class", "def", "device", "dict", "dim", "else", "if", "int", "is", "layer_idx", "len", "not", "return", "self", "shape", "states", "states_cache", "str", "tensor", "torch", "tuple", "update"], "reformer/modeling_reformer.py:ReformerDynamicCache.get_seq_length": ["ModelDynamicCache", "None", "class", "def", "get_seq_length", "int", "layer_idx", "return", "self"], "reformer/modeling_reformer.py:ReformerDynamicCache.get_start_idx": ["ModelDynamicCache", "class", "def", "get_start_idx", "if", "int", "len", "return", "self", "shape", "states_cache"], "reformer/modeling_reformer.py:ReformerDynamicCache.reorder_cache": ["ModelDynamicCache", "None", "and", "beam_idx", "buckets_cache", "class", "def", "device", "for", "if", "in", "index_select", "is", "layer_idx", "len", "not", "numel", "range", "reorder_cache", "self", "states_cache", "to"], "reformer/modeling_reformer.py:_stable_argsort": ["_stable_argsort", "arange", "argsort", "def", "device", "dim", "expand", "return", "scale_offset", "scaled_vector", "shape", "torch", "vector", "view"], "reformer/modeling_reformer.py:_get_least_common_mult_chunk_len": ["NotImplementedError", "Only", "Select", "_get_least_common_mult_chunk_len", "and", "attn", "attn_layers", "attn_types", "attn_types_set", "but", "config", "def", "elif", "else", "exist", "f", "from", "if", "layer", "lcm", "len", "local", "local_attn_chunk_length", "lsh", "lsh_attn_chunk_length", "np", "only", "raise", "return", "set", "types"], "reformer/modeling_reformer.py:_get_min_chunk_len": ["NotImplementedError", "Only", "Select", "_get_min_chunk_len", "and", "attn", "attn_layers", "attn_types", "attn_types_set", "but", "config", "def", "elif", "else", "exist", "f", "from", "if", "layer", "len", "local", "local_attn_chunk_length", "lsh", "lsh_attn_chunk_length", "min", "only", "raise", "return", "set", "types"], "reformer/modeling_reformer.py:AxialPositionEmbeddings.__init__": ["Make", "ModelPositionEmbeddings", "Model_pos_embd_dim", "Model_pos_embds", "Model_pos_embds_dim", "Model_pos_shape", "Module", "Parameter", "ParameterList", "ValueError", "__init__", "_get_least_common_mult_chunk_len", "append", "ax_shape", "axis", "class", "config", "def", "dropout", "dtype", "enumerate", "f", "factors", "float32", "for", "hidden_dropout_prob", "hidden_size", "if", "in", "least_common_mult_chunk_length", "len", "nn", "ones", "raise", "self", "sum", "super", "sure", "that", "to", "torch", "tuple", "weights"], "reformer/modeling_reformer.py:AxialPositionEmbeddings.forward": ["Got", "If", "Make", "ModelPositionEmbeddings", "Model_pos_shape", "Module", "True", "ValueError", "You", "at", "batch_size", "broadcasted_weights", "cat", "changing", "class", "config", "consider", "def", "dim", "dropout", "dropout2d", "dropped_transposed_weights", "dropped_weights", "else", "expand", "f", "factors", "for", "forward", "functional", "i", "if", "in", "index_select", "is", "item", "least", "least_common_mult_chunk_length", "length", "make", "max", "max_position_id", "might", "mul", "multiply", "nn", "or", "p", "padding", "position_encodings", "position_ids", "prod", "raise", "range", "reduce", "required_pos_encodings_columns", "reshape", "return", "self", "sequence", "sequence_length", "shape", "sure", "that", "to", "torch", "training", "transpose", "transposed_weights", "unsqueeze", "want", "weight", "weights", "your"], "reformer/modeling_reformer.py:PositionEmbeddings.__init__": ["Embedding", "ModelEmbeddings", "Module", "__init__", "class", "config", "def", "dropout", "embedding", "hidden_dropout_prob", "hidden_size", "max_Model_embeddings", "nn", "self", "super"], "reformer/modeling_reformer.py:PositionEmbeddings.forward": ["ModelEmbeddings", "Model_embeddings", "Model_ids", "Module", "class", "def", "dropout", "embedding", "forward", "functional", "nn", "p", "return", "self", "training"], "reformer/modeling_reformer.py:ReformerEmbeddings.__init__": ["AxialPositionEmbeddings", "Embedding", "ModelEmbeddings", "Module", "PositionEmbeddings", "__init__", "axial_pos_embds", "class", "config", "def", "dropout", "else", "hidden_dropout_prob", "hidden_size", "if", "max_position_embeddings", "nn", "position_embeddings", "self", "super", "vocab_size", "word_embeddings"], "reformer/modeling_reformer.py:ReformerEmbeddings.forward": ["Length", "ModelEmbeddings", "Module", "None", "Sequence", "ValueError", "arange", "be", "class", "config", "def", "device", "dropout", "dtype", "else", "embeddings", "equal", "expand", "f", "forward", "functional", "has", "if", "input_ids", "input_shape", "inputs_embeds", "is", "less", "long", "max_position_embeddings", "nn", "not", "or", "p", "position_embeddings", "position_ids", "raise", "return", "self", "seq_length", "shape", "size", "start_idx_pos_encodings", "than", "to", "torch", "training", "unsqueeze", "word_embeddings"], "reformer/modeling_reformer.py:EfficientAttentionMixin._look_adjacent": ["ModelAttentionMixin", "_look_adjacent", "and", "append", "cat", "class", "def", "dim", "else", "for", "i", "if", "in", "num_chunks_after", "num_chunks_before", "range", "return", "self", "slices", "torch", "vectors"], "reformer/modeling_reformer.py:EfficientAttentionMixin._split_hidden_size_dim": ["ModelAttentionMixin", "_split_hidden_size_dim", "attn_head_size", "class", "def", "new_x_shape", "num_attn_heads", "return", "self", "size", "transpose", "view", "x"], "reformer/modeling_reformer.py:EfficientAttentionMixin._merge_hidden_size_dims": ["ModelAttentionMixin", "_merge_hidden_size_dims", "attn_head_size", "class", "def", "num_attn_heads", "permute", "reshape", "return", "self", "size", "torch", "x"], "reformer/modeling_reformer.py:EfficientAttentionMixin._split_seq_length_dim_to": ["Input", "ModelAttentionMixin", "None", "ValueError", "_split_seq_length_dim_to", "attn_head_size", "batch_size", "be", "but", "class", "def", "dim_factor_1", "dim_factor_2", "elif", "else", "f", "if", "is", "len", "num_attn_heads", "of", "one", "raise", "rank", "reshape", "return", "self", "shape", "should", "split_dim_shape", "torch", "vector", "vectors"], "reformer/modeling_reformer.py:LSHSelfAttention.__init__": ["EfficientAttentionMixin", "False", "Linear", "Model", "Module", "None", "__init__", "all_head_size", "attention_head_size", "bias", "chunk_length", "class", "config", "def", "dropout", "hash_seed", "hidden_size", "is_decoder", "layer_idx", "lsh_attention_probs_dropout_prob", "lsh_attn_chunk_length", "lsh_num_chunks_after", "lsh_num_chunks_before", "mask_value_float16", "mask_value_float32", "max_position_embeddings", "nn", "num_attention_heads", "num_buckets", "num_chunks_after", "num_chunks_before", "num_hashes", "persistent", "query_key", "register_buffer", "self", "self_mask_value_float16", "self_mask_value_float32", "super", "tensor", "torch", "value"], "reformer/modeling_reformer.py:LSHSelfAttention.forward": ["At", "EfficientAttentionMixin", "False", "If", "Make", "Model", "ModelOutput", "Module", "None", "ReverseSort", "True", "_attend", "_gather_by_expansion", "_get_relevant_hid_states_and_buckets", "_get_sorted_bucket_idx_and_undo_sorted_bucket_idx", "_hash_vectors", "_len_and_dim_norm", "_merge_hidden_size_dims", "_query_per_attn_head", "_set_num_buckets", "_split_hidden_size_dim", "_split_seq_length_dim_to", "_value_per_attn_head", "a", "and", "apply", "arange", "are", "assert", "at", "attention_head_size", "attention_mask", "attention_probs", "auto", "batch_size", "be", "buckets", "buckets_cache", "but", "cache_position", "cat", "chunk_length", "class", "config", "def", "del", "device", "dim", "do_standard_self_attention", "elif", "else", "equals", "exists_cache", "exp", "f", "forward", "generation", "have", "hidden_states", "if", "input", "int", "is", "keepdim", "key_value_hidden_states", "key_vectors", "kwargs", "language", "last", "layer_idx", "len", "length", "logits", "logsumexp", "make", "moment", "nn", "not", "np", "num_attention_heads", "num_buckets", "num_chunks_after", "num_chunks_before", "num_hashes", "numel", "of", "one", "only", "or", "out_vectors", "output_attentions", "passed", "past_buckets", "past_buckets_states", "past_states", "possible", "probs_vectors", "query_key", "query_key_vectors", "query_vectors", "regressive", "repeat", "return", "self", "sequence", "sequence_length", "set", "shape", "should", "sorted_bucket_idx", "sorted_bucket_idx_per_hash", "sqrt", "sqrt_num", "states_cache", "sum", "sure", "that", "the", "time", "to", "torch", "undo_sorted_bucket_idx", "unsqueeze", "use_cache", "value", "value_vectors", "view", "when", "word"], "reformer/modeling_reformer.py:LSHSelfAttention._query_per_attn_head": ["EfficientAttentionMixin", "Model", "Module", "_query_per_attn_head", "ahr", "attention_head_size", "balh", "balr", "class", "def", "einsum", "hidden_size", "hidden_states", "nn", "num_attention_heads", "per_head_query_key", "query_key", "query_key_vectors", "reshape", "return", "self", "torch", "transpose", "weight"], "reformer/modeling_reformer.py:LSHSelfAttention._value_per_attn_head": ["EfficientAttentionMixin", "Model", "Module", "_value_per_attn_head", "ahr", "attention_head_size", "balh", "balr", "class", "def", "einsum", "hidden_size", "hidden_states", "nn", "num_attention_heads", "per_head_value", "reshape", "return", "self", "torch", "transpose", "value", "value_vectors", "weight"], "reformer/modeling_reformer.py:LSHSelfAttention._hash_vectors": ["EfficientAttentionMixin", "False", "Model", "Module", "None", "The", "There", "_hash_vectors", "an", "and", "arange", "argmax", "assert", "attention_mask", "batch_size", "be", "bmhtr", "bmtd", "bool", "bucket_factor", "buckets", "buckets_mask", "but", "cat", "class", "cur_product", "cur_sum", "def", "detach", "device", "dim", "dtype", "einsum", "elif", "else", "end_dim", "even", "expand", "f", "flatten", "for", "hash_seed", "if", "in", "increase_num_buckets", "int", "is", "isinstance", "item", "len", "long", "manual_seed", "mdhr", "nn", "not", "num_attention_heads", "num_bucket", "num_buckets", "num_hashes", "number", "of", "offset_buckets", "offsets", "or", "randn", "random_rotations", "return", "rotated_vectors", "rotated_vectors_factor", "rotation_size", "rotations_shape", "self", "shape", "should", "start_dim", "sum", "tensor", "to", "torch", "vectors", "view", "where"], "reformer/modeling_reformer.py:LSHSelfAttention._get_sorted_bucket_idx_and_undo_sorted_bucket_idx": ["EfficientAttentionMixin", "Model", "Module", "_get_sorted_bucket_idx_and_undo_sorted_bucket_idx", "_stable_argsort", "arange", "buckets", "class", "def", "device", "dim", "expand", "indices", "new", "nn", "no_grad", "num_hashes", "return", "scatter_", "self", "sequence_length", "shape", "size", "sorted_bucket_idx", "torch", "undo_sorted_bucket_idx", "view", "with"], "reformer/modeling_reformer.py:LSHSelfAttention._set_num_buckets": ["EfficientAttentionMixin", "Model", "Module", "Setting", "_set_num_buckets", "bit_length", "chunk_length", "class", "config", "def", "f", "if", "int", "is", "logger", "max", "max_position_embeddings", "nn", "not", "num_buckets", "num_buckets_limit", "num_buckets_pow_2", "self", "sequence_length", "set", "to", "warning"], "reformer/modeling_reformer.py:LSHSelfAttention._attend": ["EfficientAttentionMixin", "Model", "Module", "None", "True", "_attend", "_compute_attn_mask", "_look_adjacent", "_split_seq_length_dim_to", "and", "arange", "attention_mask", "attention_probs", "chunk_length", "class", "def", "del", "device", "dim", "do_standard_self_attention", "dropout", "dtype", "elif", "else", "end_dim", "exp", "expand", "flatten", "float16", "functional", "half", "if", "is", "keepdim", "key_value_bucket_idx", "key_vectors", "logits", "logsumexp", "long", "mask", "mask_value", "mask_value_float16", "mask_value_float32", "matmul", "max", "ndim", "ne", "new_ones", "nn", "not", "num_attention_heads", "num_chunks_after", "num_chunks_before", "ones_like", "out_vectors", "p", "query_bucket_idx", "query_key_dots", "query_vectors", "return", "self", "self_mask", "self_mask_value", "self_mask_value_float16", "self_mask_value_float32", "shape", "sorted_bucket_idx_per_hash", "squeeze", "start_dim", "to", "torch", "training", "transpose", "unsqueeze", "use_cache", "value_vectors", "where"], "reformer/modeling_reformer.py:LSHSelfAttention._compute_attn_mask": ["EfficientAttentionMixin", "Model", "Module", "None", "True", "_compute_attn_mask", "attention_mask", "bool", "causal_mask", "class", "def", "device", "do_standard_self_attention", "else", "expand", "gather", "ge", "if", "is", "is_decoder", "key_indices", "nn", "not", "query_indices", "query_key_dot_shape", "return", "self", "shape", "to", "torch", "unsqueeze"], "reformer/modeling_reformer.py:LSHSelfAttention._get_relevant_hid_states_and_buckets": ["EfficientAttentionMixin", "Model", "Module", "There", "_expand_to_indices_in_relevant_chunk", "_get_relevant_hid_states_and_buckets", "_hash_vectors", "_stable_argsort", "arange", "are", "assert", "attention_mask", "batch_size", "be", "bucket_idx", "bucket_idx_batch_offset", "but", "cat", "chunk_length", "class", "concat_buckets", "def", "device", "dim", "div", "dtype", "else", "f", "floor", "has", "have", "hidden_size", "hidden_states", "if", "increase_num_buckets", "index_select", "int", "isinstance", "long", "max", "max_bucket", "mul", "nn", "nonzero", "num_attention_heads", "num_buckets", "num_chunks_after", "num_chunks_before", "num_hashes", "offset", "past_buckets", "past_states", "query_buckets", "query_vectors", "reduce", "relevant_bucket_idx", "relevant_bucket_idx_chunk", "relevant_bucket_idx_chunk_all_batch", "relevant_hidden_states", "reshape", "return", "rounding_mode", "self", "sequence_length", "shape", "should", "there", "torch", "transpose", "tuple", "unsqueeze"], "reformer/modeling_reformer.py:LSHSelfAttention._expand_to_indices_in_relevant_chunk": ["EfficientAttentionMixin", "Model", "Module", "_expand_to_indices_in_relevant_chunk", "arange", "chunk_length", "chunk_sequence_indices", "class", "clone", "def", "device", "dtype", "expand", "expanded_start_indices", "flatten", "indices", "long", "nn", "num_chunks_after", "num_chunks_before", "return", "self", "sequence_length", "shape", "start_indices_chunk", "torch", "total_chunk_size", "unsqueeze"], "reformer/modeling_reformer.py:LSHSelfAttention._len_and_dim_norm": ["EfficientAttentionMixin", "Model", "Module", "_len_and_dim_norm", "_len_norm", "class", "def", "nn", "return", "self", "sqrt_num", "vectors"], "reformer/modeling_reformer.py:LSHSelfAttention._len_norm": ["EfficientAttentionMixin", "Model", "Module", "True", "_len_norm", "class", "def", "epsilon", "keepdim", "mean", "nn", "norm_x", "return", "rsqrt", "self", "torch", "variance", "x"], "reformer/modeling_reformer.py:LSHSelfAttention._gather_by_expansion": ["EfficientAttentionMixin", "Model", "Module", "_gather_by_expansion", "attention_head_size", "class", "def", "expand", "expanded_idxs", "gather", "idxs", "nn", "num_hashes", "repeat", "return", "self", "torch", "unsqueeze", "vectors"], "reformer/modeling_reformer.py:ReverseSort.forward": ["Function", "ModelSort", "class", "ctx", "def", "expand", "expanded_undo_sort_indices", "forward", "gather", "logits", "no_grad", "out_vectors", "return", "shape", "sorted_bucket_idx", "torch", "undo_sorted_bucket_idx", "unsqueeze", "with"], "reformer/modeling_reformer.py:ReverseSort.backward": ["Function", "ModelSort", "None", "backward", "class", "ctx", "def", "expand", "expanded_sort_indices", "gather", "grad_logits", "grad_out_vectors", "return", "shape", "sorted_bucket_idx", "torch", "unsqueeze"], "reformer/modeling_reformer.py:LocalSelfAttention.__init__": ["EfficientAttentionMixin", "False", "Linear", "ModelSelfAttention", "Model_attention_probs_dropout_prob", "Model_attn_chunk_length", "Model_num_chunks_after", "Model_num_chunks_before", "Module", "None", "__init__", "all_head_size", "attention_head_size", "bias", "chunk_length", "class", "config", "def", "dropout", "hidden_size", "is_decoder", "key", "layer_idx", "mask_value_float16", "mask_value_float32", "nn", "num_attention_heads", "num_chunks_after", "num_chunks_before", "pad_token_id", "persistent", "query", "register_buffer", "self", "super", "tensor", "torch", "value"], "reformer/modeling_reformer.py:LocalSelfAttention.forward": ["EfficientAttentionMixin", "False", "If", "ModelSelfAttention", "ModelSelfAttentionOutput", "Module", "None", "There", "True", "_compute_attn_mask", "_look_adjacent", "_merge_hidden_size_dims", "_retrieve_relevant_hidden_states", "_split_hidden_size_dim", "_split_seq_length_dim_to", "an", "and", "arange", "are", "assert", "attention_head_size", "attention_mask", "attention_probs", "batch_size", "be", "buckets", "buckets_cache", "but", "caching", "cat", "chunk_length", "class", "config", "def", "del", "device", "dim", "do_standard_self_attention", "dropout", "dtype", "else", "end_dim", "error", "exp", "f", "flatten", "float16", "forward", "functional", "half", "hidden_states", "hidden_states_and_buckets", "if", "indices", "is", "keepdim", "key", "key_indices", "key_value_hidden_states", "key_vectors", "kwargs", "last", "layer_idx", "len", "logits", "logsumexp", "make", "mask", "mask_value", "mask_value_float16", "mask_value_float32", "matmul", "nn", "not", "np", "num_attention_heads", "num_chunks_after", "num_chunks_before", "numel", "of", "out_vectors", "output_attentions", "p", "past_buckets", "past_buckets_states", "past_states", "query", "query_indices", "query_key_dots", "query_key_vectors", "query_vectors", "repeat", "return", "seems", "self", "sequence_length", "set", "shape", "should", "sqrt", "start_dim", "states_cache", "sure", "to", "torch", "training", "transpose", "use", "use_cache", "value", "value_vectors", "when", "where"], "reformer/modeling_reformer.py:LocalSelfAttention._compute_attn_mask": ["EfficientAttentionMixin", "ModelSelfAttention", "Module", "None", "True", "_compute_attn_mask", "_look_adjacent", "_split_seq_length_dim_to", "attention_mask", "bool", "causal_mask", "chunk_length", "class", "def", "device", "do_standard_self_attention", "else", "expand", "ge", "if", "is", "is_decoder", "key_indices", "nn", "not", "num_chunks_after", "num_chunks_before", "query_indices", "query_key_dots_shape", "return", "self", "to", "torch", "unsqueeze"], "reformer/modeling_reformer.py:LocalSelfAttention._retrieve_relevant_hidden_states": ["EfficientAttentionMixin", "ModelSelfAttention", "Module", "_retrieve_relevant_hidden_states", "chunk_length", "class", "def", "nn", "num_chunks_before", "previous_hidden_states", "return", "shape", "start_position"], "reformer/modeling_reformer.py:ReformerSelfOutput.__init__": ["False", "Linear", "ModelSelfOutput", "Module", "__init__", "all_head_size", "attention_head_size", "bias", "class", "config", "def", "dense", "dropout", "hidden_dropout_prob", "hidden_size", "nn", "num_attention_heads", "self", "super"], "reformer/modeling_reformer.py:ReformerSelfOutput.forward": ["ModelSelfOutput", "Module", "class", "def", "dense", "dropout", "forward", "functional", "hidden_states", "nn", "p", "return", "self", "training"], "reformer/modeling_reformer.py:ReformerAttention.__init__": ["LSHSelfAttention", "LayerNorm", "LocalSelfAttention", "ModelAttention", "ModelSelfOutput", "Module", "NotImplementedError", "Only", "Select", "__init__", "and", "attn", "attn_layers", "but", "class", "config", "def", "elif", "else", "eps", "exist", "f", "from", "got", "hidden_size", "if", "layer", "layer_id", "layer_idx", "layer_norm", "layer_norm_eps", "len", "local", "lsh", "nn", "only", "output", "raise", "self", "self_attention", "set", "super", "types"], "reformer/modeling_reformer.py:ReformerAttention.forward": ["AttentionOutput", "False", "ModelAttention", "Module", "None", "and", "attention_mask", "attention_output", "attention_probs", "buckets", "buckets_cache", "cache_position", "class", "def", "else", "forward", "hasattr", "hidden_states", "if", "is", "layer_id", "layer_norm", "len", "nn", "not", "num_hashes", "orig_sequence_length", "output", "output_attentions", "past_buckets_states", "return", "self", "self_attention", "self_attention_outputs", "states", "states_cache", "update", "use_cache"], "reformer/modeling_reformer.py:ReformerFeedForwardDense.__init__": ["ACT2FN", "Linear", "ModelFeedForwardDense", "Module", "__init__", "act_fn", "class", "config", "def", "dense", "dropout", "else", "feed_forward_size", "hidden_act", "hidden_dropout_prob", "hidden_size", "if", "isinstance", "nn", "self", "str", "super"], "reformer/modeling_reformer.py:ReformerFeedForwardDense.forward": ["ModelFeedForwardDense", "Module", "act_fn", "class", "def", "dense", "dropout", "forward", "functional", "hidden_states", "nn", "p", "return", "self", "training"], "reformer/modeling_reformer.py:ReformerFeedForwardOutput.__init__": ["Linear", "ModelFeedForwardOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "feed_forward_size", "hidden_dropout_prob", "hidden_size", "nn", "self", "super"], "reformer/modeling_reformer.py:ReformerFeedForwardOutput.forward": ["ModelFeedForwardOutput", "Module", "class", "def", "dense", "dropout", "forward", "functional", "hidden_states", "nn", "p", "return", "self", "training"], "reformer/modeling_reformer.py:ChunkReformerFeedForward.__init__": ["LayerNorm", "ModelFeedForwardDense", "ModelFeedForwardOutput", "ModelModelFeedForward", "Model_size_feed_forward", "Module", "__init__", "class", "config", "def", "dense", "eps", "hidden_size", "layer_norm", "layer_norm_eps", "nn", "output", "self", "seq_len_dim", "super"], "reformer/modeling_reformer.py:ChunkReformerFeedForward.forward": ["ModelModelFeedForward", "Model_size_feed_forward", "Module", "apply_Modeling_to_forward", "attention_output", "class", "def", "forward", "forward_Model", "nn", "return", "self", "seq_len_dim"], "reformer/modeling_reformer.py:ChunkReformerFeedForward.forward_chunk": ["ModelModelFeedForward", "Module", "class", "def", "dense", "forward_Model", "hidden_states", "layer_norm", "nn", "output", "return", "self"], "reformer/modeling_reformer.py:ReformerLayer.__init__": ["ChunkModelFeedForward", "ModelAttention", "ModelLayer", "Module", "None", "__init__", "attention", "attention_seed", "class", "config", "def", "feed_forward", "feed_forward_seed", "layer_id", "nn", "self", "super"], "reformer/modeling_reformer.py:ReformerLayer._init_attention_seed": ["ModelLayer", "Module", "_init_attention_seed", "and", "attention_seed", "class", "cuda", "current_device", "def", "default_generators", "device_idx", "else", "hasattr", "if", "int", "len", "manual_seed", "maxsize", "nn", "seed", "self", "sys", "torch"], "reformer/modeling_reformer.py:ReformerLayer._init_feed_forward_seed": ["ModelLayer", "Module", "_init_feed_forward_seed", "and", "class", "cuda", "current_device", "def", "default_generators", "device_idx", "else", "feed_forward_seed", "hasattr", "if", "int", "len", "manual_seed", "maxsize", "nn", "seed", "self", "sys", "torch"], "reformer/modeling_reformer.py:ReformerLayer.forward": ["False", "ModelLayer", "ModelOutput", "Module", "None", "_init_attention_seed", "_init_feed_forward_seed", "attention", "attention_mask", "attention_probs", "attn_output", "attn_outputs", "buckets", "class", "def", "del", "feed_forward", "forward", "hidden_states", "if", "nn", "no_grad", "num_hashes", "orig_sequence_length", "output_attentions", "past_buckets_states", "prev_attn_output", "return", "self", "torch", "training", "use_cache", "with"], "reformer/modeling_reformer.py:ReformerLayer.backward_pass": ["If", "ModelBackwardOutput", "ModelLayer", "ModelModel", "Module", "None", "True", "and", "assert", "attention", "attention_mask", "attention_seed", "attn_output", "backward", "backward_pass", "buckets", "class", "def", "del", "detach", "enable_grad", "feed_forward", "feed_forward_seed", "grad", "grad_attn_output", "grad_hidden_states", "hidden_states", "into", "its", "make", "manual_seed", "mode", "model", "next_attn_output", "nn", "no_grad", "output", "put", "requires_grad", "res_hidden_states", "retain_graph", "return", "self", "sure", "the", "to", "torch", "train", "training", "use", "variations", "want", "with", "you"], "reformer/modeling_reformer.py:_ReversibleFunction.forward": ["Function", "True", "_ReversibleFunction", "all_attentions", "all_buckets", "all_hidden_states", "append", "attention_mask", "attention_probs", "attn_output", "buckets", "cat", "chunk", "class", "ctx", "def", "detach", "dim", "for", "forward", "hidden_states", "if", "in", "is", "layer", "layer_outputs", "layers", "num_hashes", "orig_sequence_length", "output_attentions", "output_hidden_states", "past_buckets_states", "prev_attn_output", "return", "save_for_backward", "torch", "use_cache"], "reformer/modeling_reformer.py:_ReversibleFunction.backward": ["Function", "ModelBackwardOutput", "None", "_ReversibleFunction", "after", "all_buckets", "assert", "attention_mask", "attn_output", "backpropagation", "backward", "backward_pass", "be", "buckets", "cat", "chunk", "class", "ctx", "def", "del", "dim", "empty", "enumerate", "for", "grad_attn_output", "grad_hidden_states", "have", "hidden_states", "idx", "in", "layer", "layers", "next_attn_output", "output", "return", "saved_tensors", "to", "torch"], "reformer/modeling_reformer.py:ReformerEncoder.__init__": ["LayerNorm", "ModelEncoder", "ModelLayer", "Module", "ModuleList", "__init__", "class", "config", "def", "dropout", "eps", "for", "hidden_dropout_prob", "hidden_size", "i", "in", "layer_norm", "layer_norm_eps", "layers", "nn", "num_hidden_layers", "range", "self", "super"], "reformer/modeling_reformer.py:ReformerEncoder.forward": ["False", "ModelDynamicCache", "ModelEncoder", "ModelEncoderOutput", "Module", "None", "_ReversibleFunction", "all_attentions", "all_hidden_states", "and", "apply", "attention_mask", "cat", "class", "def", "dim", "dropout", "else", "forward", "functional", "hidden_states", "if", "is", "layer_norm", "layers", "next_cache", "nn", "num_hashes", "orig_sequence_length", "output_attentions", "output_hidden_states", "p", "past_buckets_states", "return", "self", "torch", "training", "use_cache"], "reformer/modeling_reformer.py:ReformerOnlyLMHead.__init__": ["False", "Linear", "ModelOnlyLMHead", "Module", "Parameter", "__init__", "bias", "chunk_size_lm_head", "class", "config", "decoder", "def", "hidden_size", "nn", "self", "seq_len_dim", "super", "torch", "vocab_size", "zeros"], "reformer/modeling_reformer.py:ReformerOnlyLMHead.forward": ["ModelOnlyLMHead", "Module", "apply_chunking_to_forward", "chunk_size_lm_head", "class", "def", "forward", "forward_chunk", "hidden_states", "nn", "return", "self", "seq_len_dim"], "reformer/modeling_reformer.py:ReformerOnlyLMHead.forward_chunk": ["ModelOnlyLMHead", "Module", "class", "decoder", "def", "forward_chunk", "hidden_states", "nn", "return", "self"], "reformer/modeling_reformer.py:ReformerPreTrainedModel.dummy_inputs": ["DUMMY_INPUTS", "DUMMY_MASK", "ModelPreTrainedModel", "PreTrainedModel", "attention_mask", "class", "def", "dummy_inputs", "input_ids", "input_mask", "return", "self", "tensor", "torch"], "reformer/modeling_reformer.py:ReformerPreTrainedModel._init_weights": ["AxialPositionEmbeddings", "LSHSelfAttention", "LocalSelfAttention", "ModelPreTrainedModel", "PreTrainedModel", "_init_weights", "axial_norm_std", "class", "config", "constant_", "def", "elif", "for", "if", "in", "init", "isinstance", "mask_value_float16", "mask_value_float32", "module", "normal_", "self", "self_mask_value_float16", "self_mask_value_float32", "std", "super", "weight", "weights"], "reformer/modeling_reformer.py:ReformerModel.__init__": ["ModelEmbeddings", "ModelEncoder", "ModelModel", "ModelPreTrainedModel", "Select", "__init__", "assert", "at", "attn", "attn_layers", "class", "config", "def", "embeddings", "empty", "encoder", "form", "is", "layer", "least", "local", "lsh", "num_hidden_layers", "one", "post_init", "self", "super"], "reformer/modeling_reformer.py:ReformerModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "return", "self", "word_embeddings"], "reformer/modeling_reformer.py:ReformerModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "self", "set_input_embeddings", "value", "word_embeddings"], "reformer/modeling_reformer.py:ReformerModel.forward": ["If", "ModelDynamicCache", "ModelModel", "ModelModelOutput", "ModelPreTrainedModel", "None", "Please", "Tensor", "True", "ValueError", "You", "_get_least_common_mult_chunk_len", "_get_min_chunk_len", "_pad_to_mult_of_chunk_length", "a", "all_attentions", "all_hidden_states", "and", "assert", "at", "attention_mask", "attentions", "batch_size", "be", "bool", "both", "but", "can", "cannot", "chunk_length", "class", "common", "config", "consider", "def", "device", "either", "elif", "else", "embedding_output", "embeddings", "encoder", "encoder_outputs", "f", "for", "forward", "get_start_idx", "got", "has", "have", "hidden_states", "if", "in", "inference", "input", "input_ids", "input_shape", "inputs_embeds", "int", "is", "kwargs", "last_hidden_state", "least", "least_common_mult_chunk_length", "len", "length", "min_chunk_length", "multiple", "must_pad_to_match_chunk_length", "not", "num_hashes", "of", "only", "or", "orig_sequence_length", "output_attentions", "output_hidden_states", "padded_seq_length", "padding", "padding_length", "past_buckets_states", "position_ids", "r", "raise", "return", "return_dict", "same", "self", "sequence", "sequence_length", "sequence_output", "shape", "size", "specify", "start_idx_pos_encodings", "the", "time", "to", "torch", "training", "tuple", "use_cache", "use_return_dict", "used", "v", "warn_if_padding_and_no_attention_mask"], "reformer/modeling_reformer.py:ReformerModel._pad_to_mult_of_chunk_length": ["Input", "ModelModel", "ModelPreTrainedModel", "None", "_pad_to_mult_of_chunk_length", "a", "arange", "are", "attention_mask", "automatically", "be", "bool", "cat", "chunk_length", "class", "config", "def", "device", "dim", "dtype", "else", "expand", "f", "from", "full", "get_input_embeddings", "ids", "if", "input_ids", "input_shape", "inputs_embeds", "is", "logger", "long", "multiple", "not", "of", "ones", "pad_attention_mask", "pad_token_id", "padded", "padded_input_ids", "padded_inputs_embeds", "padded_position_ids", "padded_seq_length", "padding_length", "position_ids", "return", "self", "size", "to", "torch", "unsqueeze", "warning_once", "zeros"], "reformer/modeling_reformer.py:ReformerModelWithLMHead.__init__": ["GenerationMixin", "If", "Model", "ModelModel", "ModelModelWithLMHead", "ModelOnlyLMHead", "ModelPreTrainedModel", "True", "__init__", "and", "assert", "attn_layers", "causal", "class", "config", "def", "enabled", "f", "in", "is", "is_decoder", "lm_head", "local", "local_num_chunks_after", "lsh", "lsh_num_chunks_after", "make", "mask", "not", "or", "post_init", "self", "set", "super", "sure", "that", "to", "use", "want", "you"], "reformer/modeling_reformer.py:ReformerModelWithLMHead.get_output_embeddings": ["GenerationMixin", "ModelModelWithLMHead", "ModelPreTrainedModel", "class", "decoder", "def", "get_output_embeddings", "lm_head", "return", "self"], "reformer/modeling_reformer.py:ReformerModelWithLMHead.set_output_embeddings": ["GenerationMixin", "ModelModelWithLMHead", "ModelPreTrainedModel", "bias", "class", "decoder", "def", "lm_head", "new_embeddings", "self", "set_output_embeddings"], "reformer/modeling_reformer.py:ReformerModelWithLMHead.forward": ["CausalLMOutput", "GenerationMixin", "Model", "ModelModelWithLMHead", "ModelModelWithLMHeadOutput", "ModelPreTrainedModel", "Model_outputs", "None", "Tensor", "attention_mask", "attentions", "bool", "class", "config", "def", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "list", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "not", "num_hashes", "output", "output_attentions", "output_hidden_states", "past_buckets_states", "position_ids", "r", "return", "return_dict", "self", "slice", "slice_indices", "torch", "tuple", "use_cache", "use_return_dict", "vocab_size"], "reformer/modeling_reformer.py:ReformerModelWithLMHead.prepare_inputs_for_generation": ["False", "GenerationMixin", "ModelModelWithLMHead", "ModelPreTrainedModel", "None", "Warning", "a", "attention_mask", "class", "def", "f", "for", "if", "in", "input", "input_ids", "is", "is_first_iteration", "items", "key", "kwargs", "model_inputs", "not", "num_hashes", "past_buckets_states", "past_key_values", "pop", "prepare_inputs_for_generation", "print", "recognized", "return", "self", "use_cache", "value"], "reformer/modeling_reformer.py:ReformerForMaskedLM.__init__": ["False", "If", "Model", "ModelForMaskedLM", "ModelModel", "ModelOnlyLMHead", "ModelPreTrainedModel", "__init__", "assert", "attention", "bi", "class", "config", "def", "directional", "for", "is_decoder", "lm_head", "make", "not", "post_init", "self", "super", "sure", "to", "use", "want", "you"], "reformer/modeling_reformer.py:ReformerForMaskedLM.get_output_embeddings": ["ModelForMaskedLM", "ModelPreTrainedModel", "class", "decoder", "def", "get_output_embeddings", "lm_head", "return", "self"], "reformer/modeling_reformer.py:ReformerForMaskedLM.set_output_embeddings": ["ModelForMaskedLM", "ModelPreTrainedModel", "bias", "class", "decoder", "def", "lm_head", "new_embeddings", "self", "set_output_embeddings"], "reformer/modeling_reformer.py:ReformerForMaskedLM.forward": ["CrossEntropyLoss", "False", "MaskedLMOutput", "Model", "ModelForMaskedLM", "ModelPreTrainedModel", "Model_outputs", "None", "Tensor", "attention_mask", "attentions", "bool", "class", "config", "def", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "kwargs", "labels", "lm_head", "logits", "loss", "loss_fct", "masked_lm_loss", "not", "num_hashes", "output", "output_attentions", "output_hidden_states", "position_ids", "r", "return", "return_dict", "self", "sequence_output", "torch", "tuple", "use_cache", "use_return_dict", "view", "vocab_size"], "reformer/modeling_reformer.py:ReformerForSequenceClassification.__init__": ["Model", "ModelClassificationHead", "ModelForSequenceClassification", "ModelModel", "ModelPreTrainedModel", "True", "You", "__init__", "causal", "class", "classification", "classifier", "config", "def", "disable", "for", "if", "is", "is_decoder", "logger", "masking", "might", "num_labels", "post_init", "self", "sequence", "super", "to", "want", "warning"], "reformer/modeling_reformer.py:ReformerForSequenceClassification.forward": ["BCEWithLogitsLoss", "CrossEntropyLoss", "MSELoss", "Model", "ModelForSequenceClassification", "ModelPreTrainedModel", "None", "SequenceClassifierOutput", "Tensor", "and", "attention_mask", "attentions", "bool", "class", "classifier", "config", "def", "dtype", "elif", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "kwargs", "labels", "logits", "long", "loss", "loss_fct", "multi_label_classification", "not", "num_hashes", "num_labels", "or", "output", "output_attentions", "output_hidden_states", "outputs", "position_ids", "problem_type", "r", "regression", "return", "return_dict", "self", "sequence_output", "single_label_classification", "squeeze", "torch", "tuple", "use_return_dict", "view"], "reformer/modeling_reformer.py:ReformerClassificationHead.__init__": ["Dropout", "Linear", "ModelClassificationHead", "Module", "None", "__init__", "class", "classifier_dropout", "config", "def", "dense", "dropout", "else", "hidden_dropout_prob", "hidden_size", "if", "is", "nn", "not", "num_labels", "out_proj", "self", "super"], "reformer/modeling_reformer.py:ReformerClassificationHead.forward": ["ModelClassificationHead", "Module", "class", "def", "dense", "dropout", "forward", "hidden_states", "kwargs", "nn", "out_proj", "return", "self", "tanh", "torch"], "reformer/modeling_reformer.py:ReformerForQuestionAnswering.__init__": ["Linear", "Model", "ModelForQuestionAnswering", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "config", "def", "hidden_size", "nn", "num_labels", "post_init", "qa_outputs", "self", "super"], "reformer/modeling_reformer.py:ReformerForQuestionAnswering.forward": ["CrossEntropyLoss", "False", "Model", "ModelForQuestionAnswering", "ModelPreTrainedModel", "Model_outputs", "None", "QuestionAnsweringModelOutput", "Tensor", "and", "attention_mask", "attentions", "bool", "clamp", "class", "config", "contiguous", "def", "dim", "else", "end_logits", "end_loss", "end_positions", "forward", "hidden_states", "if", "ignore_index", "ignored_index", "input_ids", "inputs_embeds", "int", "is", "kwargs", "len", "logits", "loss", "loss_fct", "not", "num_hashes", "output", "output_attentions", "output_hidden_states", "position_ids", "qa_outputs", "r", "return", "return_dict", "self", "sequence_output", "size", "split", "squeeze", "start_logits", "start_loss", "start_positions", "torch", "total_loss", "tuple", "use_cache", "use_return_dict"], "regnet/modeling_regnet.py:RegNetConvLayer.__init__": ["ACT2FN", "BatchNorm2d", "Conv2d", "False", "Identity", "ModelConvLayer", "Module", "None", "__init__", "activation", "bias", "class", "convolution", "def", "else", "groups", "if", "in_channels", "int", "is", "kernel_size", "nn", "normalization", "not", "out_channels", "padding", "relu", "self", "str", "stride", "super"], "regnet/modeling_regnet.py:RegNetConvLayer.forward": ["ModelConvLayer", "Module", "activation", "class", "convolution", "def", "forward", "hidden_state", "nn", "normalization", "return", "self"], "regnet/modeling_regnet.py:RegNetEmbeddings.__init__": ["ModelConfig", "ModelConvLayer", "ModelEmbeddings", "Module", "__init__", "activation", "class", "config", "def", "embedder", "embedding_size", "hidden_act", "kernel_size", "nn", "num_channels", "self", "stride", "super"], "regnet/modeling_regnet.py:RegNetEmbeddings.forward": ["Make", "ModelEmbeddings", "Module", "ValueError", "channel", "class", "configuration", "def", "dimension", "embedder", "forward", "hidden_state", "if", "in", "match", "nn", "num_channels", "of", "one", "pixel", "pixel_values", "raise", "return", "self", "set", "shape", "sure", "that", "the", "values", "with"], "regnet/modeling_regnet.py:RegNetShortCut.__init__": ["BatchNorm2d", "Conv2d", "False", "ModelShortCut", "Module", "__init__", "bias", "class", "convolution", "def", "in_channels", "int", "kernel_size", "nn", "normalization", "out_channels", "self", "stride", "super"], "regnet/modeling_regnet.py:RegNetShortCut.forward": ["ModelShortCut", "Module", "Tensor", "class", "convolution", "def", "forward", "hidden_state", "input", "nn", "normalization", "return", "self"], "regnet/modeling_regnet.py:RegNetSELayer.__init__": ["AdaptiveAvgPool2d", "Conv2d", "ModelSELayer", "Module", "ReLU", "Sequential", "Sigmoid", "__init__", "attention", "class", "def", "in_channels", "int", "kernel_size", "nn", "pooler", "reduced_channels", "self", "super"], "regnet/modeling_regnet.py:RegNetSELayer.forward": ["ModelSELayer", "Module", "attention", "class", "def", "forward", "hidden_state", "nn", "pooled", "pooler", "return", "self"], "regnet/modeling_regnet.py:RegNetXLayer.__init__": ["ACT2FN", "Identity", "ModelConfig", "ModelConvLayer", "ModelShortCut", "ModelXLayer", "Module", "None", "Sequential", "__init__", "activation", "class", "config", "def", "else", "groups", "groups_width", "hidden_act", "if", "in_channels", "int", "kernel_size", "layer", "max", "nn", "or", "out_channels", "self", "shortcut", "should_apply_shortcut", "stride", "super"], "regnet/modeling_regnet.py:RegNetXLayer.forward": ["ModelXLayer", "Module", "activation", "class", "def", "forward", "hidden_state", "layer", "nn", "residual", "return", "self", "shortcut"], "regnet/modeling_regnet.py:RegNetYLayer.__init__": ["ACT2FN", "Identity", "ModelConfig", "ModelConvLayer", "ModelSELayer", "ModelShortCut", "ModelYLayer", "Module", "None", "Sequential", "__init__", "activation", "class", "config", "def", "else", "groups", "groups_width", "hidden_act", "if", "in_channels", "int", "kernel_size", "layer", "max", "nn", "or", "out_channels", "reduced_channels", "round", "self", "shortcut", "should_apply_shortcut", "stride", "super"], "regnet/modeling_regnet.py:RegNetYLayer.forward": ["ModelYLayer", "Module", "activation", "class", "def", "forward", "hidden_state", "layer", "nn", "residual", "return", "self", "shortcut"], "regnet/modeling_regnet.py:RegNetStage.__init__": ["ModelConfig", "ModelStage", "ModelXLayer", "ModelYLayer", "Module", "Sequential", "_", "__init__", "class", "config", "def", "depth", "else", "for", "if", "in", "in_channels", "int", "layer", "layer_type", "layers", "nn", "out_channels", "range", "self", "stride", "super", "x"], "regnet/modeling_regnet.py:RegNetStage.forward": ["ModelStage", "Module", "class", "def", "forward", "hidden_state", "layers", "nn", "return", "self"], "regnet/modeling_regnet.py:RegNetEncoder.__init__": ["ModelConfig", "ModelEncoder", "ModelStage", "Module", "ModuleList", "__init__", "append", "class", "config", "def", "depth", "depths", "downsample_in_first_stage", "else", "embedding_size", "for", "hidden_sizes", "if", "in", "in_channels", "in_out_channels", "nn", "out_channels", "self", "stages", "stride", "super", "zip"], "regnet/modeling_regnet.py:RegNetEncoder.forward": ["BaseModelOutputWithNoAttention", "False", "ModelEncoder", "Module", "None", "Tensor", "True", "bool", "class", "def", "else", "for", "forward", "hidden_state", "hidden_states", "if", "in", "is", "last_hidden_state", "nn", "not", "output_hidden_states", "return", "return_dict", "self", "stage_module", "stages", "tuple", "v"], "regnet/modeling_regnet.py:RegNetPreTrainedModel._init_weights": ["BatchNorm2d", "Conv2d", "GroupNorm", "Linear", "ModelPreTrainedModel", "None", "PreTrainedModel", "_", "_calculate_fan_in_and_fan_out", "_init_weights", "a", "bias", "bound", "class", "constant_", "def", "elif", "else", "fan_in", "fan_out", "getattr", "if", "init", "is", "isinstance", "kaiming_normal_", "kaiming_uniform_", "math", "mode", "module", "nn", "nonlinearity", "not", "num_batches_tracked", "ones_", "relu", "running_mean", "running_var", "self", "sqrt", "torch", "uniform_", "weight", "zeros_"], "regnet/modeling_regnet.py:RegNetModel.__init__": ["AdaptiveAvgPool2d", "ModelEmbeddings", "ModelEncoder", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "config", "def", "embedder", "encoder", "nn", "pooler", "post_init", "self", "super"], "regnet/modeling_regnet.py:RegNetModel.forward": ["BaseModelOutputWithPoolingAndNoAttention", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "bool", "class", "config", "def", "else", "embedder", "embedding_output", "encoder", "encoder_outputs", "forward", "hidden_states", "if", "is", "kwargs", "last_hidden_state", "not", "output_hidden_states", "pixel_values", "pooled_output", "pooler", "pooler_output", "return", "return_dict", "self", "use_return_dict"], "regnet/modeling_regnet.py:RegNetForImageClassification.__init__": ["Flatten", "Identity", "Linear", "Model", "ModelForImageClassification", "ModelModel", "ModelPreTrainedModel", "Sequential", "__init__", "class", "classifier", "config", "def", "else", "hidden_sizes", "if", "nn", "num_labels", "post_init", "self", "super"], "regnet/modeling_regnet.py:RegNetForImageClassification.forward": ["FloatTensor", "ImageClassifierOutputWithNoAttention", "LongTensor", "Model", "ModelForImageClassification", "ModelPreTrainedModel", "None", "bool", "class", "classifier", "config", "def", "else", "forward", "hidden_states", "if", "is", "kwargs", "labels", "logits", "loss", "loss_function", "not", "output", "output_hidden_states", "outputs", "pixel_values", "pooled_output", "pooler_output", "r", "return", "return_dict", "self", "torch", "use_return_dict"], "rembert/modeling_rembert.py:RemBertEmbeddings.__init__": ["Dropout", "Embedding", "False", "LayerNorm", "ModelEmbeddings", "Module", "__init__", "arange", "class", "config", "def", "dropout", "eps", "expand", "hidden_dropout_prob", "input_embedding_size", "layer_norm_eps", "max_position_embeddings", "nn", "pad_token_id", "padding_idx", "persistent", "position_embeddings", "position_ids", "register_buffer", "self", "super", "token_type_embeddings", "torch", "type_vocab_size", "vocab_size", "word_embeddings"], "rembert/modeling_rembert.py:RemBertEmbeddings.forward": ["FloatTensor", "LayerNorm", "LongTensor", "ModelEmbeddings", "Module", "None", "Tensor", "class", "def", "device", "dropout", "dtype", "else", "embeddings", "forward", "if", "input_ids", "input_shape", "inputs_embeds", "int", "is", "long", "nn", "not", "past_key_values_length", "position_embeddings", "position_ids", "return", "self", "seq_length", "size", "token_type_embeddings", "token_type_ids", "torch", "word_embeddings", "zeros"], "rembert/modeling_rembert.py:RemBertPooler.__init__": ["Linear", "ModelPooler", "Module", "Tanh", "__init__", "activation", "class", "config", "def", "dense", "hidden_size", "nn", "self", "super"], "rembert/modeling_rembert.py:RemBertPooler.forward": ["ModelPooler", "Module", "Tensor", "activation", "class", "def", "dense", "first_token_tensor", "forward", "hidden_states", "nn", "pooled_output", "return", "self", "torch"], "rembert/modeling_rembert.py:RemBertSelfAttention.__init__": ["Dropout", "Linear", "ModelSelfAttention", "Module", "None", "The", "ValueError", "__init__", "a", "all_head_size", "and", "attention", "attention_head_size", "attention_probs_dropout_prob", "class", "config", "def", "dropout", "embedding_size", "f", "hasattr", "heads", "hidden", "hidden_size", "if", "int", "is", "is_decoder", "key", "layer_idx", "multiple", "nn", "not", "num_attention_heads", "number", "of", "query", "raise", "self", "size", "super", "the", "value"], "rembert/modeling_rembert.py:RemBertSelfAttention.forward": ["Cache", "EncoderDecoderCache", "False", "FloatTensor", "ModelSelfAttention", "Module", "None", "Tensor", "True", "_", "all_head_size", "and", "attention_head_size", "attention_mask", "attention_probs", "attention_scores", "batch_size", "bool", "cache_position", "class", "context_layer", "contiguous", "cross_attention_cache", "curr_past_key_values", "current_states", "def", "dim", "dropout", "else", "encoder_hidden_states", "forward", "functional", "get", "hidden_states", "if", "is", "is_cross_attention", "is_updated", "isinstance", "key", "key_layer", "keys", "layer_idx", "layers", "math", "matmul", "new_context_layer_shape", "nn", "not", "num_attention_heads", "output_attentions", "past_key_values", "permute", "query", "query_layer", "return", "self", "self_attention_cache", "seq_length", "shape", "size", "softmax", "sqrt", "torch", "transpose", "tuple", "update", "value", "value_layer", "values", "view"], "rembert/modeling_rembert.py:RemBertSelfOutput.__init__": ["Dropout", "LayerNorm", "Linear", "ModelSelfOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "eps", "hidden_dropout_prob", "hidden_size", "layer_norm_eps", "nn", "self", "super"], "rembert/modeling_rembert.py:RemBertSelfOutput.forward": ["LayerNorm", "ModelSelfOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "rembert/modeling_rembert.py:RemBertAttention.__init__": ["ModelAttention", "ModelSelfAttention", "ModelSelfOutput", "Module", "None", "__init__", "class", "config", "def", "layer_idx", "nn", "output", "self", "super"], "rembert/modeling_rembert.py:RemBertAttention.forward": ["Cache", "False", "FloatTensor", "ModelAttention", "Module", "None", "Tensor", "attention_mask", "attention_output", "bool", "cache_position", "class", "def", "encoder_hidden_states", "forward", "hidden_states", "nn", "output", "output_attentions", "outputs", "past_key_values", "return", "self", "self_outputs", "torch", "tuple"], "rembert/modeling_rembert.py:RemBertIntermediate.__init__": ["ACT2FN", "Linear", "ModelIntermediate", "Module", "__init__", "class", "config", "def", "dense", "else", "hidden_act", "hidden_size", "if", "intermediate_act_fn", "intermediate_size", "isinstance", "nn", "self", "str", "super"], "rembert/modeling_rembert.py:RemBertIntermediate.forward": ["ModelIntermediate", "Module", "Tensor", "class", "def", "dense", "forward", "hidden_states", "intermediate_act_fn", "nn", "return", "self", "torch"], "rembert/modeling_rembert.py:RemBertOutput.__init__": ["Dropout", "LayerNorm", "Linear", "ModelOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "eps", "hidden_dropout_prob", "hidden_size", "intermediate_size", "layer_norm_eps", "nn", "self", "super"], "rembert/modeling_rembert.py:RemBertOutput.forward": ["LayerNorm", "ModelOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "rembert/modeling_rembert.py:RemBertLayer.__init__": ["GradientCheckpointingLayer", "ModelAttention", "ModelIntermediate", "ModelLayer", "ModelOutput", "None", "ValueError", "__init__", "a", "add_cross_attention", "added", "as", "attention", "be", "chunk_size_feed_forward", "class", "config", "cross", "crossattention", "decoder", "def", "f", "if", "intermediate", "is", "is_decoder", "layer_idx", "model", "not", "output", "raise", "self", "seq_len_dim", "should", "super", "used"], "rembert/modeling_rembert.py:RemBertLayer.forward": ["Cache", "False", "FloatTensor", "GradientCheckpointingLayer", "If", "ModelLayer", "None", "Tensor", "True", "ValueError", "add_cross_attention", "and", "apply_chunking_to_forward", "are", "attention", "attention_mask", "attention_output", "be", "bool", "by", "cache_position", "chunk_size_feed_forward", "class", "config", "cross", "cross_attention_outputs", "crossattention", "def", "encoder_attention_mask", "encoder_hidden_states", "f", "feed_forward_chunk", "forward", "has", "hasattr", "hidden_states", "if", "instantiated", "is", "is_decoder", "layer_output", "layers", "not", "output_attentions", "outputs", "passed", "past_key_values", "raise", "return", "self", "self_attention_outputs", "seq_len_dim", "setting", "to", "torch", "tuple", "with"], "rembert/modeling_rembert.py:RemBertLayer.feed_forward_chunk": ["GradientCheckpointingLayer", "ModelLayer", "attention_output", "class", "def", "feed_forward_chunk", "intermediate", "intermediate_output", "layer_output", "output", "return", "self"], "rembert/modeling_rembert.py:RemBertEncoder.__init__": ["False", "Linear", "ModelEncoder", "ModelLayer", "Module", "ModuleList", "__init__", "class", "config", "def", "embedding_hidden_mapping_in", "for", "gradient_checkpointing", "hidden_size", "i", "in", "input_embedding_size", "layer", "layer_idx", "nn", "num_hidden_layers", "range", "self", "super"], "rembert/modeling_rembert.py:RemBertEncoder.forward": ["BaseModelOutputWithPastAndCrossAttentions", "Cache", "DynamicCache", "EncoderDecoderCache", "False", "FloatTensor", "ModelEncoder", "Module", "None", "Setting", "Tensor", "True", "add_cross_attention", "all_cross_attentions", "all_hidden_states", "all_self_attentions", "and", "attention_mask", "attentions", "bool", "cache_position", "checkpointing", "class", "config", "cross_attentions", "def", "else", "embedding_hidden_mapping_in", "encoder_attention_mask", "encoder_hidden_states", "enumerate", "for", "forward", "gradient", "gradient_checkpointing", "hidden_states", "i", "if", "in", "incompatible", "is", "last_hidden_state", "layer", "layer_module", "layer_outputs", "logger", "nn", "not", "output_attentions", "output_hidden_states", "past_key_values", "return", "return_dict", "self", "torch", "training", "tuple", "use_cache", "v", "warning_once", "with"], "rembert/modeling_rembert.py:RemBertPredictionHeadTransform.__init__": ["ACT2FN", "LayerNorm", "Linear", "ModelPredictionHeadTransform", "Module", "__init__", "class", "config", "def", "dense", "else", "eps", "hidden_act", "hidden_size", "if", "isinstance", "layer_norm_eps", "nn", "self", "str", "super", "transform_act_fn"], "rembert/modeling_rembert.py:RemBertPredictionHeadTransform.forward": ["LayerNorm", "ModelPredictionHeadTransform", "Module", "Tensor", "class", "def", "dense", "forward", "hidden_states", "nn", "return", "self", "torch", "transform_act_fn"], "rembert/modeling_rembert.py:RemBertLMPredictionHead.__init__": ["ACT2FN", "LayerNorm", "Linear", "ModelLMPredictionHead", "Module", "__init__", "activation", "class", "config", "decoder", "def", "dense", "eps", "hidden_act", "hidden_size", "layer_norm_eps", "nn", "output_embedding_size", "self", "super", "vocab_size"], "rembert/modeling_rembert.py:RemBertLMPredictionHead.forward": ["LayerNorm", "ModelLMPredictionHead", "Module", "Tensor", "activation", "class", "decoder", "def", "dense", "forward", "hidden_states", "nn", "return", "self", "torch"], "rembert/modeling_rembert.py:RemBertOnlyMLMHead.__init__": ["ModelLMPredictionHead", "ModelOnlyMLMHead", "Module", "__init__", "class", "config", "def", "nn", "predictions", "self", "super"], "rembert/modeling_rembert.py:RemBertOnlyMLMHead.forward": ["ModelOnlyMLMHead", "Module", "Tensor", "class", "def", "forward", "nn", "prediction_scores", "predictions", "return", "self", "sequence_output", "torch"], "rembert/modeling_rembert.py:RemBertPreTrainedModel._init_weights": ["ModelEmbeddings", "ModelPreTrainedModel", "PreTrainedModel", "_init_weights", "arange", "class", "copy_", "def", "expand", "if", "init", "isinstance", "module", "position_ids", "self", "shape", "super", "torch"], "rembert/modeling_rembert.py:RemBertModel.__init__": ["ModelEmbeddings", "ModelEncoder", "ModelModel", "ModelPooler", "ModelPreTrainedModel", "None", "True", "__init__", "add_pooling_layer", "class", "config", "def", "else", "embeddings", "encoder", "if", "pooler", "post_init", "r", "self", "super"], "rembert/modeling_rembert.py:RemBertModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "return", "self", "word_embeddings"], "rembert/modeling_rembert.py:RemBertModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "self", "set_input_embeddings", "value", "word_embeddings"], "rembert/modeling_rembert.py:RemBertModel.forward": ["BaseModelOutputWithPoolingAndCrossAttentions", "Cache", "False", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "ValueError", "You", "_", "and", "at", "attention_mask", "attentions", "batch_size", "bool", "both", "cache_position", "cannot", "class", "config", "cross_attentions", "def", "device", "dtype", "either", "elif", "else", "embedding_output", "embeddings", "encoder", "encoder_attention_mask", "encoder_batch_size", "encoder_extended_attention_mask", "encoder_hidden_shape", "encoder_hidden_states", "encoder_outputs", "encoder_sequence_length", "extended_attention_mask", "forward", "get_extended_attention_mask", "get_seq_length", "have", "hidden_states", "if", "input_ids", "input_shape", "inputs_embeds", "invert_attention_mask", "is", "is_decoder", "kwargs", "last_hidden_state", "long", "not", "ones", "or", "output_attentions", "output_hidden_states", "past_key_values", "past_key_values_length", "pooled_output", "pooler", "pooler_output", "position_ids", "raise", "return", "return_dict", "same", "self", "seq_length", "sequence_output", "size", "specify", "the", "time", "to", "token_type_ids", "torch", "tuple", "use_cache", "use_return_dict", "warn_if_padding_and_no_attention_mask", "zeros"], "rembert/modeling_rembert.py:RemBertForMaskedLM.__init__": ["False", "If", "Model", "ModelForMaskedLM", "ModelModel", "ModelOnlyMLMHead", "ModelPreTrainedModel", "__init__", "add_pooling_layer", "attention", "bi", "class", "cls", "config", "def", "directional", "for", "if", "is_decoder", "logger", "make", "post_init", "self", "super", "sure", "to", "use", "want", "warning", "you"], "rembert/modeling_rembert.py:RemBertForMaskedLM.get_output_embeddings": ["ModelForMaskedLM", "ModelPreTrainedModel", "class", "cls", "decoder", "def", "get_output_embeddings", "predictions", "return", "self"], "rembert/modeling_rembert.py:RemBertForMaskedLM.set_output_embeddings": ["ModelForMaskedLM", "ModelPreTrainedModel", "class", "cls", "decoder", "def", "new_embeddings", "predictions", "self", "set_output_embeddings"], "rembert/modeling_rembert.py:RemBertForMaskedLM.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "MaskedLMOutput", "Model", "ModelForMaskedLM", "ModelPreTrainedModel", "None", "attention_mask", "attentions", "bool", "class", "cls", "config", "def", "else", "encoder_attention_mask", "encoder_hidden_states", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "masked_lm_loss", "not", "output", "output_attentions", "output_hidden_states", "outputs", "position_ids", "prediction_scores", "r", "return", "return_dict", "self", "sequence_output", "token_type_ids", "torch", "tuple", "use_return_dict", "view", "vocab_size"], "rembert/modeling_rembert.py:RemBertForMaskedLM.prepare_inputs_for_generation": ["False", "ModelForMaskedLM", "ModelPreTrainedModel", "None", "PAD", "The", "assert", "attention_mask", "be", "cat", "class", "config", "def", "defined", "device", "dim", "dtype", "dummy_token", "effective_batch_size", "for", "full", "generation", "input_ids", "input_shape", "is", "is_first_iteration", "long", "model_kwargs", "new_zeros", "not", "pad_token_id", "prepare_inputs_for_generation", "return", "self", "shape", "should", "token", "torch"], "rembert/modeling_rembert.py:RemBertForMaskedLM.can_generate": ["False", "ModelForMaskedLM", "ModelPreTrainedModel", "bool", "can_generate", "class", "cls", "def", "return"], "rembert/modeling_rembert.py:RemBertForCausalLM.__init__": ["False", "GenerationMixin", "If", "Model", "ModelForCausalLM", "ModelModel", "ModelOnlyMLMHead", "ModelPreTrainedModel", "True", "__init__", "a", "add", "add_pooling_layer", "as", "class", "cls", "config", "def", "if", "is_decoder", "logger", "not", "post_init", "self", "standalone", "super", "to", "use", "want", "warning", "you"], "rembert/modeling_rembert.py:RemBertForCausalLM.get_output_embeddings": ["GenerationMixin", "ModelForCausalLM", "ModelPreTrainedModel", "class", "cls", "decoder", "def", "get_output_embeddings", "predictions", "return", "self"], "rembert/modeling_rembert.py:RemBertForCausalLM.set_output_embeddings": ["GenerationMixin", "ModelForCausalLM", "ModelPreTrainedModel", "class", "cls", "decoder", "def", "new_embeddings", "predictions", "self", "set_output_embeddings"], "rembert/modeling_rembert.py:RemBertForCausalLM.forward": ["Cache", "CausalLMOutputWithCrossAttentions", "FloatTensor", "GenerationMixin", "LongTensor", "Model", "ModelForCausalLM", "ModelPreTrainedModel", "None", "Tensor", "attention_mask", "attentions", "bool", "class", "cls", "config", "cross_attentions", "def", "else", "encoder_attention_mask", "encoder_hidden_states", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "logits", "logits_to_keep", "loss", "loss_function", "not", "output", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "position_ids", "r", "return", "return_dict", "self", "slice", "slice_indices", "token_type_ids", "torch", "tuple", "use_cache", "use_return_dict", "vocab_size"], "rembert/modeling_rembert.py:RemBertForSequenceClassification.__init__": ["Dropout", "Linear", "Model", "ModelForSequenceClassification", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "classifier", "classifier_dropout_prob", "config", "def", "dropout", "hidden_size", "nn", "num_labels", "post_init", "self", "super"], "rembert/modeling_rembert.py:RemBertForSequenceClassification.forward": ["BCEWithLogitsLoss", "CrossEntropyLoss", "FloatTensor", "LongTensor", "MSELoss", "Model", "ModelForSequenceClassification", "ModelPreTrainedModel", "None", "SequenceClassifierOutput", "and", "attention_mask", "attentions", "bool", "class", "classifier", "config", "def", "dropout", "dtype", "elif", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "kwargs", "labels", "logits", "long", "loss", "loss_fct", "multi_label_classification", "not", "num_labels", "or", "output", "output_attentions", "output_hidden_states", "outputs", "pooled_output", "position_ids", "problem_type", "r", "regression", "return", "return_dict", "self", "single_label_classification", "squeeze", "token_type_ids", "torch", "tuple", "use_return_dict", "view"], "rembert/modeling_rembert.py:RemBertForMultipleChoice.__init__": ["Dropout", "Linear", "Model", "ModelForMultipleChoice", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "classifier", "classifier_dropout_prob", "config", "def", "dropout", "hidden_size", "nn", "post_init", "self", "super"], "rembert/modeling_rembert.py:RemBertForMultipleChoice.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "Model", "ModelForMultipleChoice", "ModelPreTrainedModel", "MultipleChoiceModelOutput", "None", "attention_mask", "attentions", "bool", "class", "classifier", "config", "def", "dropout", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "not", "num_choices", "output", "output_attentions", "output_hidden_states", "outputs", "pooled_output", "position_ids", "r", "reshaped_logits", "return", "return_dict", "self", "shape", "size", "token_type_ids", "torch", "tuple", "use_return_dict", "view"], "rembert/modeling_rembert.py:RemBertForTokenClassification.__init__": ["Dropout", "False", "Linear", "Model", "ModelForTokenClassification", "ModelModel", "ModelPreTrainedModel", "__init__", "add_pooling_layer", "class", "classifier", "classifier_dropout_prob", "config", "def", "dropout", "hidden_size", "nn", "num_labels", "post_init", "self", "super"], "rembert/modeling_rembert.py:RemBertForTokenClassification.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "Model", "ModelForTokenClassification", "ModelPreTrainedModel", "None", "TokenClassifierOutput", "attention_mask", "attentions", "bool", "class", "classifier", "config", "def", "dropout", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "not", "num_labels", "output", "output_attentions", "output_hidden_states", "outputs", "position_ids", "r", "return", "return_dict", "self", "sequence_output", "token_type_ids", "torch", "tuple", "use_return_dict", "view"], "rembert/modeling_rembert.py:RemBertForQuestionAnswering.__init__": ["False", "Linear", "Model", "ModelForQuestionAnswering", "ModelModel", "ModelPreTrainedModel", "__init__", "add_pooling_layer", "class", "config", "def", "hidden_size", "nn", "num_labels", "post_init", "qa_outputs", "self", "super"], "rembert/modeling_rembert.py:RemBertForQuestionAnswering.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "Model", "ModelForQuestionAnswering", "ModelPreTrainedModel", "None", "QuestionAnsweringModelOutput", "and", "attention_mask", "attentions", "bool", "clamp_", "class", "config", "def", "dim", "else", "end_logits", "end_loss", "end_positions", "forward", "hidden_states", "if", "ignore_index", "ignored_index", "input_ids", "inputs_embeds", "is", "kwargs", "len", "logits", "loss", "loss_fct", "not", "output", "output_attentions", "output_hidden_states", "outputs", "position_ids", "qa_outputs", "return", "return_dict", "self", "sequence_output", "size", "split", "squeeze", "start_logits", "start_loss", "start_positions", "token_type_ids", "torch", "total_loss", "tuple", "use_return_dict"], "resnet/modeling_resnet.py:ResNetConvLayer.__init__": ["ACT2FN", "BatchNorm2d", "Conv2d", "False", "Identity", "ModelConvLayer", "Module", "None", "__init__", "activation", "bias", "class", "convolution", "def", "else", "if", "in_channels", "int", "is", "kernel_size", "nn", "normalization", "not", "out_channels", "padding", "relu", "self", "str", "stride", "super"], "resnet/modeling_resnet.py:ResNetConvLayer.forward": ["ModelConvLayer", "Module", "Tensor", "activation", "class", "convolution", "def", "forward", "hidden_state", "input", "nn", "normalization", "return", "self"], "resnet/modeling_resnet.py:ResNetEmbeddings.__init__": ["MaxPool2d", "ModelConfig", "ModelConvLayer", "ModelEmbeddings", "Module", "__init__", "activation", "class", "config", "def", "embedder", "embedding_size", "hidden_act", "kernel_size", "nn", "num_channels", "padding", "pooler", "self", "stride", "super"], "resnet/modeling_resnet.py:ResNetEmbeddings.forward": ["Make", "ModelEmbeddings", "Module", "Tensor", "ValueError", "channel", "class", "configuration", "def", "dimension", "embedder", "embedding", "forward", "if", "in", "match", "nn", "num_channels", "of", "one", "pixel", "pixel_values", "pooler", "raise", "return", "self", "set", "shape", "sure", "that", "the", "values", "with"], "resnet/modeling_resnet.py:ResNetShortCut.__init__": ["BatchNorm2d", "Conv2d", "False", "ModelShortCut", "Module", "__init__", "bias", "class", "convolution", "def", "in_channels", "int", "kernel_size", "nn", "normalization", "out_channels", "self", "stride", "super"], "resnet/modeling_resnet.py:ResNetShortCut.forward": ["ModelShortCut", "Module", "Tensor", "class", "convolution", "def", "forward", "hidden_state", "input", "nn", "normalization", "return", "self"], "resnet/modeling_resnet.py:ResNetBasicLayer.__init__": ["ACT2FN", "Identity", "ModelBasicLayer", "ModelConvLayer", "ModelShortCut", "Module", "None", "Sequential", "__init__", "activation", "class", "def", "else", "if", "in_channels", "int", "layer", "nn", "or", "out_channels", "relu", "self", "shortcut", "should_apply_shortcut", "str", "stride", "super"], "resnet/modeling_resnet.py:ResNetBasicLayer.forward": ["ModelBasicLayer", "Modelidual", "Module", "activation", "class", "def", "forward", "hidden_state", "layer", "nn", "return", "self", "shortcut"], "resnet/modeling_resnet.py:ResNetBottleNeckLayer.__init__": ["ACT2FN", "False", "Identity", "ModelBottleNeckLayer", "ModelConvLayer", "ModelShortCut", "Module", "None", "Sequential", "__init__", "activation", "bool", "class", "def", "downsample_in_bottleneck", "else", "if", "in_channels", "int", "kernel_size", "layer", "nn", "not", "or", "out_channels", "reduces_channels", "reduction", "relu", "self", "shortcut", "should_apply_shortcut", "str", "stride", "super"], "resnet/modeling_resnet.py:ResNetBottleNeckLayer.forward": ["ModelBottleNeckLayer", "Modelidual", "Module", "activation", "class", "def", "forward", "hidden_state", "layer", "nn", "return", "self", "shortcut"], "resnet/modeling_resnet.py:ResNetStage.__init__": ["ModelBasicLayer", "ModelBottleNeckLayer", "ModelConfig", "ModelStage", "Module", "Sequential", "_", "__init__", "activation", "bottleneck", "class", "config", "def", "depth", "downsample_in_bottleneck", "else", "first_layer", "for", "hidden_act", "if", "in", "in_channels", "int", "layer", "layer_type", "layers", "nn", "out_channels", "range", "self", "stride", "super"], "resnet/modeling_resnet.py:ResNetStage.forward": ["ModelStage", "Module", "Tensor", "class", "def", "for", "forward", "hidden_state", "in", "input", "layer", "layers", "nn", "return", "self"], "resnet/modeling_resnet.py:ResNetEncoder.__init__": ["ModelConfig", "ModelEncoder", "ModelStage", "Module", "ModuleList", "__init__", "append", "class", "config", "def", "depth", "depths", "downsample_in_first_stage", "else", "embedding_size", "for", "hidden_sizes", "if", "in", "in_channels", "in_out_channels", "nn", "out_channels", "self", "stages", "stride", "super", "zip"], "resnet/modeling_resnet.py:ResNetEncoder.forward": ["BaseModelOutputWithNoAttention", "False", "ModelEncoder", "Module", "None", "Tensor", "True", "bool", "class", "def", "else", "for", "forward", "hidden_state", "hidden_states", "if", "in", "is", "last_hidden_state", "nn", "not", "output_hidden_states", "return", "return_dict", "self", "stage_module", "stages", "tuple", "v"], "resnet/modeling_resnet.py:ResNetPreTrainedModel._init_weights": ["BatchNorm", "Conv2d", "Linear", "ModelPreTrainedModel", "None", "PreTrainedModel", "_", "__class__", "__name__", "_calculate_fan_in_and_fan_out", "_init_weights", "a", "bias", "bound", "class", "def", "elif", "else", "fan_in", "fan_out", "getattr", "if", "in", "init", "is", "isinstance", "kaiming_normal_", "kaiming_uniform_", "math", "mode", "module", "nn", "nonlinearity", "not", "num_batches_tracked", "ones_", "relu", "running_mean", "running_var", "self", "sqrt", "torch", "uniform_", "weight", "zeros_"], "resnet/modeling_resnet.py:ResNetModel.__init__": ["AdaptiveAvgPool2d", "ModelEmbeddings", "ModelEncoder", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "config", "def", "embedder", "encoder", "nn", "pooler", "post_init", "self", "super"], "resnet/modeling_resnet.py:ResNetModel.forward": ["BaseModelOutputWithPoolingAndNoAttention", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "bool", "class", "config", "def", "else", "embedder", "embedding_output", "encoder", "encoder_outputs", "forward", "hidden_states", "if", "is", "kwargs", "last_hidden_state", "not", "output_hidden_states", "pixel_values", "pooled_output", "pooler", "pooler_output", "return", "return_dict", "self", "use_return_dict"], "resnet/modeling_resnet.py:ResNetForImageClassification.__init__": ["Flatten", "Identity", "Linear", "Model", "ModelForImageClassification", "ModelModel", "ModelPreTrainedModel", "Sequential", "__init__", "class", "classifier", "config", "def", "else", "hidden_sizes", "if", "nn", "num_labels", "post_init", "self", "super"], "resnet/modeling_resnet.py:ResNetForImageClassification.forward": ["FloatTensor", "ImageClassifierOutputWithNoAttention", "LongTensor", "Model", "ModelForImageClassification", "ModelPreTrainedModel", "None", "bool", "class", "classifier", "config", "def", "else", "forward", "hidden_states", "if", "is", "kwargs", "labels", "logits", "loss", "loss_function", "not", "output", "output_hidden_states", "outputs", "pixel_values", "pooled_output", "pooler_output", "r", "return", "return_dict", "self", "torch", "use_return_dict"], "resnet/modeling_resnet.py:ResNetBackbone.__init__": ["BackboneMixin", "ModelBackbone", "ModelEmbeddings", "ModelEncoder", "ModelPreTrainedModel", "__init__", "_init_backbone", "class", "config", "def", "embedder", "embedding_size", "encoder", "hidden_sizes", "num_featuModel", "post_init", "self", "super"], "resnet/modeling_resnet.py:ResNetBackbone.forward": ["BackboneMixin", "BackboneOutput", "ModelBackbone", "ModelPreTrainedModel", "None", "Tensor", "True", "attentions", "bool", "class", "config", "def", "else", "embedder", "embedding_output", "encoder", "enumerate", "feature_maps", "for", "forward", "hidden_states", "idx", "if", "in", "is", "kwargs", "not", "out_featuModel", "output", "output_hidden_states", "outputs", "pixel_values", "r", "return", "return_dict", "self", "stage", "stage_names", "use_return_dict"], "roberta/modeling_roberta.py:RobertaEmbeddings.__init__": ["Dropout", "Embedding", "False", "LayerNorm", "ModelEmbeddings", "Module", "__init__", "arange", "class", "config", "def", "dropout", "dtype", "eps", "expand", "hidden_dropout_prob", "hidden_size", "layer_norm_eps", "long", "max_position_embeddings", "nn", "pad_token_id", "padding_idx", "persistent", "position_embeddings", "position_ids", "register_buffer", "self", "size", "super", "token_type_embeddings", "token_type_ids", "torch", "type_vocab_size", "vocab_size", "word_embeddings", "zeros"], "roberta/modeling_roberta.py:RobertaEmbeddings.forward": ["FloatTensor", "LayerNorm", "LongTensor", "ModelEmbeddings", "Module", "None", "Tensor", "batch_size", "buffered_token_type_ids", "class", "create_position_ids_from_input_ids", "create_position_ids_from_inputs_embeds", "def", "device", "dim", "dropout", "dtype", "else", "embeddings", "expand", "forward", "gather", "hasattr", "if", "index", "input_ids", "input_shape", "inputs_embeds", "int", "is", "long", "nn", "not", "padding_idx", "past_key_values_length", "position_embeddings", "position_ids", "return", "self", "seq_length", "shape", "size", "token_type_embeddings", "token_type_ids", "torch", "word_embeddings", "zeros"], "roberta/modeling_roberta.py:RobertaEmbeddings.create_position_ids_from_inputs_embeds": ["ModelEmbeddings", "Module", "arange", "class", "create_position_ids_from_inputs_embeds", "def", "device", "dtype", "expand", "input_shape", "inputs_embeds", "long", "nn", "padding_idx", "position_ids", "return", "sequence_length", "size", "torch", "unsqueeze"], "roberta/modeling_roberta.py:RobertaEmbeddings.create_position_ids_from_input_ids": ["ModelEmbeddings", "Module", "class", "create_position_ids_from_input_ids", "cumsum", "def", "dim", "incremental_indices", "input_ids", "int", "long", "mask", "ne", "nn", "padding_idx", "past_key_values_length", "return", "torch", "type_as"], "roberta/modeling_roberta.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "contiguous", "def", "dim", "dropout", "float", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "shape", "size", "softmax", "torch", "training", "transpose", "value"], "roberta/modeling_roberta.py:RobertaSelfAttention.__init__": ["Dropout", "False", "Linear", "ModelSelfAttention", "Module", "None", "The", "ValueError", "__init__", "a", "all_head_size", "and", "attention", "attention_head_size", "attention_probs_dropout_prob", "class", "config", "def", "dropout", "embedding_size", "f", "hasattr", "heads", "hidden", "hidden_size", "if", "int", "is", "is_causal", "is_decoder", "key", "layer_idx", "multiple", "nn", "not", "num_attention_heads", "number", "of", "query", "raise", "scaling", "self", "size", "super", "the", "value"], "roberta/modeling_roberta.py:RobertaSelfAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "EncoderDecoderCache", "FloatTensor", "ModelSelfAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_attn_implementation", "attention_head_size", "attention_interface", "attention_mask", "attn_output", "attn_weights", "cache_position", "class", "config", "contiguous", "current_past_key_values", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "hidden_shape", "hidden_states", "if", "input_shape", "is", "isinstance", "key", "key_layer", "kwargs", "layer_idx", "nn", "not", "p", "past_key_values", "query", "query_layer", "reshape", "return", "scaling", "self", "self_attention_cache", "shape", "torch", "training", "transpose", "tuple", "update", "value", "value_layer", "view"], "roberta/modeling_roberta.py:RobertaCrossAttention.__init__": ["Dropout", "False", "Linear", "ModelCrossAttention", "Module", "None", "The", "ValueError", "__init__", "a", "all_head_size", "and", "attention", "attention_head_size", "attention_probs_dropout_prob", "class", "config", "def", "dropout", "embedding_size", "f", "hasattr", "heads", "hidden", "hidden_size", "if", "int", "is", "is_causal", "key", "layer_idx", "multiple", "nn", "not", "num_attention_heads", "number", "of", "query", "raise", "scaling", "self", "size", "super", "the", "value"], "roberta/modeling_roberta.py:RobertaCrossAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "EncoderDecoderCache", "False", "FloatTensor", "ModelCrossAttention", "Module", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "_attn_implementation", "and", "attention_head_size", "attention_interface", "attention_mask", "attn_output", "attn_weights", "bsz", "class", "config", "contiguous", "cross_attention_cache", "def", "dropout", "eager", "eager_attention_forward", "else", "encoder_hidden_states", "forward", "get", "hidden_states", "if", "is", "is_updated", "key", "key_layer", "keys", "kv_input_shape", "kwargs", "layer_idx", "layers", "nn", "not", "p", "past_key_values", "q_input_shape", "query", "query_layer", "reshape", "return", "scaling", "self", "shape", "src_len", "tgt_len", "torch", "training", "transpose", "tuple", "update", "value", "value_layer", "values", "view"], "roberta/modeling_roberta.py:RobertaSelfOutput.__init__": ["Dropout", "LayerNorm", "Linear", "ModelSelfOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "eps", "hidden_dropout_prob", "hidden_size", "layer_norm_eps", "nn", "self", "super"], "roberta/modeling_roberta.py:RobertaSelfOutput.forward": ["LayerNorm", "ModelSelfOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "roberta/modeling_roberta.py:RobertaAttention.__init__": ["False", "ModelAttention", "ModelCrossAttention", "ModelSelfAttention", "ModelSelfOutput", "Module", "None", "__init__", "attention_class", "class", "config", "def", "else", "if", "is_causal", "is_cross_attention", "layer_idx", "nn", "output", "self", "super"], "roberta/modeling_roberta.py:RobertaAttention.forward": ["Cache", "FloatTensor", "ModelAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attention_output", "attn_weights", "cache_position", "class", "def", "else", "encoder_attention_mask", "encoder_hidden_states", "forward", "hidden_states", "if", "is_cross_attention", "kwargs", "nn", "not", "output", "past_key_values", "return", "self", "torch", "tuple"], "roberta/modeling_roberta.py:RobertaIntermediate.__init__": ["ACT2FN", "Linear", "ModelIntermediate", "Module", "__init__", "class", "config", "def", "dense", "else", "hidden_act", "hidden_size", "if", "intermediate_act_fn", "intermediate_size", "isinstance", "nn", "self", "str", "super"], "roberta/modeling_roberta.py:RobertaIntermediate.forward": ["ModelIntermediate", "Module", "Tensor", "class", "def", "dense", "forward", "hidden_states", "intermediate_act_fn", "nn", "return", "self", "torch"], "roberta/modeling_roberta.py:RobertaOutput.__init__": ["Dropout", "LayerNorm", "Linear", "ModelOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "eps", "hidden_dropout_prob", "hidden_size", "intermediate_size", "layer_norm_eps", "nn", "self", "super"], "roberta/modeling_roberta.py:RobertaOutput.forward": ["LayerNorm", "ModelOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "roberta/modeling_roberta.py:RobertaLayer.__init__": ["False", "GradientCheckpointingLayer", "ModelAttention", "ModelIntermediate", "ModelLayer", "ModelOutput", "None", "True", "ValueError", "__init__", "a", "add_cross_attention", "added", "as", "attention", "be", "chunk_size_feed_forward", "class", "config", "cross", "crossattention", "decoder", "def", "f", "if", "intermediate", "is", "is_causal", "is_cross_attention", "is_decoder", "layer_idx", "model", "not", "output", "raise", "self", "seq_len_dim", "should", "super", "used"], "roberta/modeling_roberta.py:RobertaLayer.forward": ["Cache", "FloatTensor", "GradientCheckpointingLayer", "If", "ModelLayer", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "ValueError", "_", "add_cross_attention", "and", "apply_chunking_to_forward", "are", "attention", "attention_mask", "attention_output", "be", "by", "cache_position", "chunk_size_feed_forward", "class", "config", "cross", "cross_attention_output", "crossattention", "def", "encoder_attention_mask", "encoder_hidden_states", "f", "feed_forward_chunk", "forward", "has", "hasattr", "hidden_states", "if", "instantiated", "is", "is_decoder", "kwargs", "layer_output", "layers", "not", "passed", "past_key_values", "raise", "return", "self", "self_attention_output", "seq_len_dim", "setting", "to", "torch", "tuple", "with"], "roberta/modeling_roberta.py:RobertaLayer.feed_forward_chunk": ["GradientCheckpointingLayer", "ModelLayer", "attention_output", "class", "def", "feed_forward_chunk", "intermediate", "intermediate_output", "layer_output", "output", "return", "self"], "roberta/modeling_roberta.py:RobertaPreTrainedModel._init_weights": ["ModelEmbeddings", "ModelLMHead", "ModelPreTrainedModel", "PreTrainedModel", "_init_weights", "arange", "bias", "class", "copy_", "def", "elif", "expand", "if", "init", "isinstance", "module", "position_ids", "self", "shape", "super", "token_type_ids", "torch", "zeros_"], "roberta/modeling_roberta.py:RobertaEncoder.__init__": ["ModelEncoder", "ModelLayer", "Module", "ModuleList", "__init__", "class", "config", "def", "for", "i", "in", "layer", "layer_idx", "nn", "num_hidden_layers", "range", "self", "super"], "roberta/modeling_roberta.py:RobertaEncoder.forward": ["BaseModelOutputWithPastAndCrossAttentions", "Cache", "FloatTensor", "ModelEncoder", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "bool", "cache_position", "class", "def", "else", "encoder_attention_mask", "encoder_hidden_states", "enumerate", "for", "forward", "hidden_states", "i", "if", "in", "kwargs", "last_hidden_state", "layer", "layer_module", "nn", "past_key_values", "return", "self", "torch", "tuple", "use_cache"], "roberta/modeling_roberta.py:RobertaPooler.__init__": ["Linear", "ModelPooler", "Module", "Tanh", "__init__", "activation", "class", "config", "def", "dense", "hidden_size", "nn", "self", "super"], "roberta/modeling_roberta.py:RobertaPooler.forward": ["ModelPooler", "Module", "Tensor", "activation", "class", "def", "dense", "first_token_tensor", "forward", "hidden_states", "nn", "pooled_output", "return", "self", "torch"], "roberta/modeling_roberta.py:RobertaModel.__init__": ["False", "ModelEmbeddings", "ModelEncoder", "ModelModel", "ModelPooler", "ModelPreTrainedModel", "None", "True", "__init__", "add_pooling_layer", "class", "config", "def", "else", "embeddings", "encoder", "gradient_checkpointing", "if", "pooler", "post_init", "r", "self", "super"], "roberta/modeling_roberta.py:RobertaModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "return", "self", "word_embeddings"], "roberta/modeling_roberta.py:RobertaModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "self", "set_input_embeddings", "value", "word_embeddings"], "roberta/modeling_roberta.py:RobertaModel.forward": ["BaseModelOutputWithPoolingAndCrossAttentions", "Cache", "DynamicCache", "EncoderDecoderCache", "False", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "_create_attention_masks", "and", "arange", "attention_mask", "bool", "cache_position", "class", "config", "def", "device", "else", "embedding_output", "embeddings", "encoder", "encoder_attention_mask", "encoder_hidden_states", "encoder_outputs", "exactly", "forward", "get_seq_length", "if", "input_ids", "inputs_embeds", "is", "is_decoder", "is_encoder_decoder", "kwargs", "last_hidden_state", "must", "not", "of", "one", "or", "past_key_values", "past_key_values_length", "pooled_output", "pooler", "pooler_output", "position_ids", "raise", "return", "self", "seq_length", "sequence_output", "shape", "specify", "token_type_ids", "torch", "tuple", "use_cache"], "roberta/modeling_roberta.py:RobertaModel._create_attention_masks": ["ModelModel", "ModelPreTrainedModel", "None", "_create_attention_masks", "attention_mask", "cache_position", "class", "config", "create_bidirectional_mask", "create_causal_mask", "def", "else", "embedding_output", "encoder_attention_mask", "encoder_hidden_states", "if", "input_embeds", "is", "is_decoder", "not", "past_key_values", "return", "self"], "roberta/modeling_roberta.py:RobertaForCausalLM.__init__": ["False", "GenerationMixin", "If", "Model", "ModelForCausalLM", "ModelLMHead", "ModelLMHeadModel", "ModelModel", "ModelPreTrainedModel", "True", "__init__", "a", "add", "add_pooling_layer", "as", "class", "config", "def", "if", "is_decoder", "lm_head", "logger", "not", "post_init", "self", "standalone", "super", "to", "use", "want", "warning", "you"], "roberta/modeling_roberta.py:RobertaForCausalLM.get_output_embeddings": ["GenerationMixin", "ModelForCausalLM", "ModelPreTrainedModel", "class", "decoder", "def", "get_output_embeddings", "lm_head", "return", "self"], "roberta/modeling_roberta.py:RobertaForCausalLM.set_output_embeddings": ["GenerationMixin", "ModelForCausalLM", "ModelPreTrainedModel", "class", "decoder", "def", "lm_head", "new_embeddings", "self", "set_output_embeddings"], "roberta/modeling_roberta.py:RobertaForCausalLM.forward": ["BaseModelOutputWithPoolingAndCrossAttentions", "CausalLMOutputWithCrossAttentions", "False", "FloatTensor", "GenerationMixin", "LongTensor", "Model", "ModelForCausalLM", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "cross_attentions", "def", "else", "encoder_attention_mask", "encoder_hidden_states", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "not", "outputs", "past_key_values", "position_ids", "r", "return", "return_dict", "self", "slice", "slice_indices", "token_type_ids", "torch", "tuple", "use_cache", "vocab_size"], "roberta/modeling_roberta.py:RobertaForMaskedLM.__init__": ["False", "If", "Model", "ModelForMaskedLM", "ModelLMHead", "ModelModel", "ModelPreTrainedModel", "__init__", "add_pooling_layer", "attention", "bi", "class", "config", "def", "directional", "for", "if", "is_decoder", "lm_head", "logger", "make", "post_init", "self", "super", "sure", "to", "use", "want", "warning", "you"], "roberta/modeling_roberta.py:RobertaForMaskedLM.get_output_embeddings": ["ModelForMaskedLM", "ModelPreTrainedModel", "class", "decoder", "def", "get_output_embeddings", "lm_head", "return", "self"], "roberta/modeling_roberta.py:RobertaForMaskedLM.set_output_embeddings": ["ModelForMaskedLM", "ModelPreTrainedModel", "class", "decoder", "def", "lm_head", "new_embeddings", "self", "set_output_embeddings"], "roberta/modeling_roberta.py:RobertaForMaskedLM.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "MaskedLMOutput", "Model", "ModelForMaskedLM", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "attention_mask", "attentions", "class", "config", "def", "device", "encoder_attention_mask", "encoder_hidden_states", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "lm_head", "logits", "loss", "loss_fct", "masked_lm_loss", "not", "outputs", "position_ids", "prediction_scores", "r", "return", "return_dict", "self", "sequence_output", "to", "token_type_ids", "torch", "tuple", "view", "vocab_size"], "roberta/modeling_roberta.py:RobertaLMHead.__init__": ["LayerNorm", "Linear", "ModelLMHead", "Module", "Parameter", "__init__", "bias", "class", "config", "decoder", "def", "dense", "eps", "hidden_size", "layer_norm", "layer_norm_eps", "nn", "self", "super", "torch", "vocab_size", "zeros"], "roberta/modeling_roberta.py:RobertaLMHead.forward": ["ModelLMHead", "Module", "class", "decoder", "def", "dense", "features", "forward", "gelu", "kwargs", "layer_norm", "nn", "return", "self", "x"], "roberta/modeling_roberta.py:RobertaForSequenceClassification.__init__": ["False", "Model", "ModelClassificationHead", "ModelForSequenceClassification", "ModelModel", "ModelPreTrainedModel", "__init__", "add_pooling_layer", "class", "classifier", "config", "def", "num_labels", "post_init", "self", "super"], "roberta/modeling_roberta.py:RobertaForSequenceClassification.forward": ["BCEWithLogitsLoss", "CrossEntropyLoss", "FloatTensor", "LongTensor", "MSELoss", "Model", "ModelForSequenceClassification", "ModelPreTrainedModel", "None", "SequenceClassifierOutput", "Tensor", "TransformersKwargs", "True", "Unpack", "and", "attention_mask", "attentions", "class", "classifier", "config", "def", "device", "dtype", "elif", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "kwargs", "labels", "logits", "long", "loss", "loss_fct", "multi_label_classification", "not", "num_labels", "or", "outputs", "position_ids", "problem_type", "r", "regression", "return", "return_dict", "self", "sequence_output", "single_label_classification", "squeeze", "to", "token_type_ids", "torch", "tuple", "view"], "roberta/modeling_roberta.py:RobertaForMultipleChoice.__init__": ["Dropout", "Linear", "Model", "ModelForMultipleChoice", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "classifier", "config", "def", "dropout", "hidden_dropout_prob", "hidden_size", "nn", "post_init", "self", "super"], "roberta/modeling_roberta.py:RobertaForMultipleChoice.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "Model", "ModelForMultipleChoice", "ModelPreTrainedModel", "MultipleChoiceModelOutput", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "attention_mask", "attentions", "class", "classifier", "def", "device", "dropout", "else", "flat_attention_mask", "flat_input_ids", "flat_inputs_embeds", "flat_position_ids", "flat_token_type_ids", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "not", "num_choices", "outputs", "pooled_output", "position_ids", "r", "reshaped_logits", "return", "return_dict", "self", "shape", "size", "to", "token_type_ids", "torch", "tuple", "view"], "roberta/modeling_roberta.py:RobertaForTokenClassification.__init__": ["Dropout", "False", "Linear", "Model", "ModelForTokenClassification", "ModelModel", "ModelPreTrainedModel", "None", "__init__", "add_pooling_layer", "class", "classifier", "classifier_dropout", "config", "def", "dropout", "else", "hidden_dropout_prob", "hidden_size", "if", "is", "nn", "not", "num_labels", "post_init", "self", "super"], "roberta/modeling_roberta.py:RobertaForTokenClassification.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "Model", "ModelForTokenClassification", "ModelPreTrainedModel", "None", "Tensor", "TokenClassifierOutput", "TransformersKwargs", "True", "Unpack", "attention_mask", "attentions", "class", "classifier", "def", "device", "dropout", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "not", "num_labels", "outputs", "position_ids", "r", "return", "return_dict", "self", "sequence_output", "to", "token_type_ids", "torch", "tuple", "view"], "roberta/modeling_roberta.py:RobertaClassificationHead.__init__": ["Dropout", "Linear", "ModelClassificationHead", "Module", "None", "__init__", "class", "classifier_dropout", "config", "def", "dense", "dropout", "else", "hidden_dropout_prob", "hidden_size", "if", "is", "nn", "not", "num_labels", "out_proj", "self", "super"], "roberta/modeling_roberta.py:RobertaClassificationHead.forward": ["ModelClassificationHead", "Module", "class", "def", "dense", "dropout", "features", "forward", "kwargs", "nn", "out_proj", "return", "self", "tanh", "torch", "x"], "roberta/modeling_roberta.py:RobertaForQuestionAnswering.__init__": ["False", "Linear", "Model", "ModelForQuestionAnswering", "ModelModel", "ModelPreTrainedModel", "__init__", "add_pooling_layer", "class", "config", "def", "hidden_size", "nn", "num_labels", "post_init", "qa_outputs", "self", "super"], "roberta/modeling_roberta.py:RobertaForQuestionAnswering.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "Model", "ModelForQuestionAnswering", "ModelPreTrainedModel", "None", "QuestionAnsweringModelOutput", "Tensor", "TransformersKwargs", "True", "Unpack", "and", "attention_mask", "attentions", "clamp", "class", "contiguous", "def", "dim", "end_logits", "end_loss", "end_positions", "forward", "hidden_states", "if", "ignore_index", "ignored_index", "input_ids", "inputs_embeds", "is", "kwargs", "len", "logits", "loss", "loss_fct", "not", "outputs", "position_ids", "qa_outputs", "r", "return", "return_dict", "self", "sequence_output", "size", "split", "squeeze", "start_logits", "start_loss", "start_positions", "token_type_ids", "torch", "total_loss", "tuple"], "roberta_prelayernorm/modeling_roberta_prelayernorm.py:RobertaPreLayerNormEmbeddings.__init__": ["Dropout", "Embedding", "False", "LayerNorm", "ModelEmbeddings", "Module", "__init__", "arange", "class", "config", "def", "dropout", "dtype", "eps", "expand", "hidden_dropout_prob", "hidden_size", "layer_norm_eps", "long", "max_position_embeddings", "nn", "pad_token_id", "padding_idx", "persistent", "position_embeddings", "position_ids", "register_buffer", "self", "size", "super", "token_type_embeddings", "token_type_ids", "torch", "type_vocab_size", "vocab_size", "word_embeddings", "zeros"], "roberta_prelayernorm/modeling_roberta_prelayernorm.py:RobertaPreLayerNormEmbeddings.forward": ["FloatTensor", "LayerNorm", "LongTensor", "ModelEmbeddings", "Module", "None", "Tensor", "batch_size", "buffered_token_type_ids", "class", "create_position_ids_from_input_ids", "create_position_ids_from_inputs_embeds", "def", "device", "dim", "dropout", "dtype", "else", "embeddings", "expand", "forward", "gather", "hasattr", "if", "index", "input_ids", "input_shape", "inputs_embeds", "int", "is", "long", "nn", "not", "padding_idx", "past_key_values_length", "position_embeddings", "position_ids", "return", "self", "seq_length", "shape", "size", "token_type_embeddings", "token_type_ids", "torch", "word_embeddings", "zeros"], "roberta_prelayernorm/modeling_roberta_prelayernorm.py:RobertaPreLayerNormEmbeddings.create_position_ids_from_inputs_embeds": ["ModelEmbeddings", "Module", "arange", "class", "create_position_ids_from_inputs_embeds", "def", "device", "dtype", "expand", "input_shape", "inputs_embeds", "long", "nn", "padding_idx", "position_ids", "return", "sequence_length", "size", "torch", "unsqueeze"], "roberta_prelayernorm/modeling_roberta_prelayernorm.py:RobertaPreLayerNormEmbeddings.create_position_ids_from_input_ids": ["ModelEmbeddings", "Module", "class", "create_position_ids_from_input_ids", "cumsum", "def", "dim", "incremental_indices", "input_ids", "int", "long", "mask", "ne", "nn", "padding_idx", "past_key_values_length", "return", "torch", "type_as"], "roberta_prelayernorm/modeling_roberta_prelayernorm.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "contiguous", "def", "dim", "dropout", "float", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "shape", "size", "softmax", "torch", "training", "transpose", "value"], "roberta_prelayernorm/modeling_roberta_prelayernorm.py:RobertaPreLayerNormSelfAttention.__init__": ["Dropout", "False", "Linear", "ModelSelfAttention", "Module", "None", "The", "ValueError", "__init__", "a", "all_head_size", "and", "attention", "attention_head_size", "attention_probs_dropout_prob", "class", "config", "def", "dropout", "embedding_size", "f", "hasattr", "heads", "hidden", "hidden_size", "if", "int", "is", "is_causal", "is_decoder", "key", "layer_idx", "multiple", "nn", "not", "num_attention_heads", "number", "of", "query", "raise", "scaling", "self", "size", "super", "the", "value"], "roberta_prelayernorm/modeling_roberta_prelayernorm.py:RobertaPreLayerNormSelfAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "EncoderDecoderCache", "FloatTensor", "ModelSelfAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_attn_implementation", "attention_head_size", "attention_interface", "attention_mask", "attn_output", "attn_weights", "cache_position", "class", "config", "contiguous", "current_past_key_values", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "hidden_shape", "hidden_states", "if", "input_shape", "is", "isinstance", "key", "key_layer", "kwargs", "layer_idx", "nn", "not", "p", "past_key_values", "query", "query_layer", "reshape", "return", "scaling", "self", "self_attention_cache", "shape", "torch", "training", "transpose", "tuple", "update", "value", "value_layer", "view"], "roberta_prelayernorm/modeling_roberta_prelayernorm.py:RobertaPreLayerNormCrossAttention.__init__": ["Dropout", "False", "Linear", "ModelCrossAttention", "Module", "None", "The", "ValueError", "__init__", "a", "all_head_size", "and", "attention", "attention_head_size", "attention_probs_dropout_prob", "class", "config", "def", "dropout", "embedding_size", "f", "hasattr", "heads", "hidden", "hidden_size", "if", "int", "is", "is_causal", "key", "layer_idx", "multiple", "nn", "not", "num_attention_heads", "number", "of", "query", "raise", "scaling", "self", "size", "super", "the", "value"], "roberta_prelayernorm/modeling_roberta_prelayernorm.py:RobertaPreLayerNormCrossAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "EncoderDecoderCache", "False", "FloatTensor", "ModelCrossAttention", "Module", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "_attn_implementation", "and", "attention_head_size", "attention_interface", "attention_mask", "attn_output", "attn_weights", "bsz", "class", "config", "contiguous", "cross_attention_cache", "def", "dropout", "eager", "eager_attention_forward", "else", "encoder_hidden_states", "forward", "get", "hidden_states", "if", "is", "is_updated", "key", "key_layer", "keys", "kv_input_shape", "kwargs", "layer_idx", "layers", "nn", "not", "p", "past_key_values", "q_input_shape", "query", "query_layer", "reshape", "return", "scaling", "self", "shape", "src_len", "tgt_len", "torch", "training", "transpose", "tuple", "update", "value", "value_layer", "values", "view"], "roberta_prelayernorm/modeling_roberta_prelayernorm.py:RobertaPreLayerNormSelfOutput.__init__": ["Dropout", "Linear", "ModelSelfOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "hidden_dropout_prob", "hidden_size", "nn", "self", "super"], "roberta_prelayernorm/modeling_roberta_prelayernorm.py:RobertaPreLayerNormSelfOutput.forward": ["ModelSelfOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "roberta_prelayernorm/modeling_roberta_prelayernorm.py:RobertaPreLayerNormAttention.__init__": ["False", "LayerNorm", "ModelAttention", "ModelCrossAttention", "ModelSelfAttention", "ModelSelfOutput", "Module", "None", "__init__", "attention_class", "class", "config", "def", "else", "eps", "hidden_size", "if", "is_causal", "is_cross_attention", "layer_idx", "layer_norm_eps", "nn", "output", "self", "super"], "roberta_prelayernorm/modeling_roberta_prelayernorm.py:RobertaPreLayerNormAttention.forward": ["FloatTensor", "LayerNorm", "ModelAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attention_output", "attn_weights", "cache_position", "class", "def", "else", "encoder_attention_mask", "encoder_hidden_states", "forward", "hidden_states", "hidden_states_pre_layer_norm", "if", "is_cross_attention", "kwargs", "nn", "not", "output", "past_key_values", "return", "self", "torch", "tuple"], "roberta_prelayernorm/modeling_roberta_prelayernorm.py:RobertaPreLayerNormIntermediate.__init__": ["ACT2FN", "LayerNorm", "Linear", "ModelIntermediate", "Module", "__init__", "class", "config", "def", "dense", "else", "eps", "hidden_act", "hidden_size", "if", "intermediate_act_fn", "intermediate_size", "isinstance", "layer_norm_eps", "nn", "self", "str", "super"], "roberta_prelayernorm/modeling_roberta_prelayernorm.py:RobertaPreLayerNormIntermediate.forward": ["LayerNorm", "ModelIntermediate", "Module", "Tensor", "class", "def", "dense", "forward", "hidden_states", "intermediate_act_fn", "nn", "return", "self", "torch"], "roberta_prelayernorm/modeling_roberta_prelayernorm.py:RobertaPreLayerNormOutput.__init__": ["Dropout", "Linear", "ModelOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "hidden_dropout_prob", "hidden_size", "intermediate_size", "nn", "self", "super"], "roberta_prelayernorm/modeling_roberta_prelayernorm.py:RobertaPreLayerNormOutput.forward": ["ModelOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "roberta_prelayernorm/modeling_roberta_prelayernorm.py:RobertaPreLayerNormLayer.__init__": ["False", "GradientCheckpointingLayer", "ModelAttention", "ModelIntermediate", "ModelLayer", "ModelOutput", "None", "True", "ValueError", "__init__", "a", "add_cross_attention", "added", "as", "attention", "be", "chunk_size_feed_forward", "class", "config", "cross", "crossattention", "decoder", "def", "f", "if", "intermediate", "is", "is_causal", "is_cross_attention", "is_decoder", "layer_idx", "model", "not", "output", "raise", "self", "seq_len_dim", "should", "super", "used"], "roberta_prelayernorm/modeling_roberta_prelayernorm.py:RobertaPreLayerNormLayer.forward": ["Cache", "FloatTensor", "GradientCheckpointingLayer", "If", "ModelLayer", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "ValueError", "_", "add_cross_attention", "and", "apply_chunking_to_forward", "are", "attention", "attention_mask", "attention_output", "be", "by", "cache_position", "chunk_size_feed_forward", "class", "config", "cross", "cross_attention_output", "crossattention", "def", "encoder_attention_mask", "encoder_hidden_states", "f", "feed_forward_chunk", "forward", "has", "hasattr", "hidden_states", "if", "instantiated", "is", "is_decoder", "kwargs", "layer_output", "layers", "not", "passed", "past_key_values", "raise", "return", "self", "self_attention_output", "seq_len_dim", "setting", "to", "torch", "tuple", "with"], "roberta_prelayernorm/modeling_roberta_prelayernorm.py:RobertaPreLayerNormLayer.feed_forward_chunk": ["GradientCheckpointingLayer", "ModelLayer", "attention_output", "class", "def", "feed_forward_chunk", "intermediate", "intermediate_output", "layer_output", "output", "return", "self"], "roberta_prelayernorm/modeling_roberta_prelayernorm.py:RobertaPreLayerNormEncoder.__init__": ["ModelEncoder", "ModelLayer", "Module", "ModuleList", "__init__", "class", "config", "def", "for", "i", "in", "layer", "layer_idx", "nn", "num_hidden_layers", "range", "self", "super"], "roberta_prelayernorm/modeling_roberta_prelayernorm.py:RobertaPreLayerNormEncoder.forward": ["BaseModelOutputWithPastAndCrossAttentions", "Cache", "FloatTensor", "ModelEncoder", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "bool", "cache_position", "class", "def", "else", "encoder_attention_mask", "encoder_hidden_states", "enumerate", "for", "forward", "hidden_states", "i", "if", "in", "kwargs", "last_hidden_state", "layer", "layer_module", "nn", "past_key_values", "return", "self", "torch", "tuple", "use_cache"], "roberta_prelayernorm/modeling_roberta_prelayernorm.py:RobertaPreLayerNormPooler.__init__": ["Linear", "ModelPooler", "Module", "Tanh", "__init__", "activation", "class", "config", "def", "dense", "hidden_size", "nn", "self", "super"], "roberta_prelayernorm/modeling_roberta_prelayernorm.py:RobertaPreLayerNormPooler.forward": ["ModelPooler", "Module", "Tensor", "activation", "class", "def", "dense", "first_token_tensor", "forward", "hidden_states", "nn", "pooled_output", "return", "self", "torch"], "roberta_prelayernorm/modeling_roberta_prelayernorm.py:RobertaPreLayerNormPreTrainedModel._init_weights": ["ModelEmbeddings", "ModelLMHead", "ModelPreTrainedModel", "PreTrainedModel", "_init_weights", "arange", "bias", "class", "copy_", "def", "elif", "expand", "if", "init", "isinstance", "module", "position_ids", "self", "shape", "super", "token_type_ids", "torch", "zeros_"], "roberta_prelayernorm/modeling_roberta_prelayernorm.py:RobertaPreLayerNormModel.__init__": ["False", "LayerNorm", "ModelEmbeddings", "ModelEncoder", "ModelModel", "ModelPooler", "ModelPreTrainedModel", "None", "True", "__init__", "add_pooling_layer", "class", "config", "def", "else", "embeddings", "encoder", "eps", "gradient_checkpointing", "hidden_size", "if", "layer_norm_eps", "nn", "pooler", "post_init", "r", "self", "super"], "roberta_prelayernorm/modeling_roberta_prelayernorm.py:RobertaPreLayerNormModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "return", "self", "word_embeddings"], "roberta_prelayernorm/modeling_roberta_prelayernorm.py:RobertaPreLayerNormModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "self", "set_input_embeddings", "value", "word_embeddings"], "roberta_prelayernorm/modeling_roberta_prelayernorm.py:RobertaPreLayerNormModel.forward": ["BaseModelOutputWithPoolingAndCrossAttentions", "DynamicCache", "EncoderDecoderCache", "False", "FloatTensor", "LayerNorm", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "_create_attention_masks", "and", "arange", "attention_mask", "bool", "cache_position", "class", "config", "def", "device", "else", "embedding_output", "embeddings", "encoder", "encoder_attention_mask", "encoder_hidden_states", "encoder_outputs", "exactly", "forward", "get_seq_length", "if", "input_ids", "input_shape", "inputs_embeds", "is", "is_decoder", "is_encoder_decoder", "kwargs", "last_hidden_state", "list", "must", "not", "of", "one", "or", "past_key_values", "past_key_values_length", "pooled_output", "pooler", "pooler_output", "position_ids", "r", "raise", "return", "self", "seq_length", "sequence_output", "shape", "specify", "token_type_ids", "torch", "tuple", "use_cache"], "roberta_prelayernorm/modeling_roberta_prelayernorm.py:RobertaPreLayerNormModel._create_attention_masks": ["ModelModel", "ModelPreTrainedModel", "None", "_create_attention_masks", "attention_mask", "cache_position", "class", "config", "create_bidirectional_mask", "create_causal_mask", "def", "else", "embedding_output", "encoder_attention_mask", "encoder_hidden_states", "if", "input_embeds", "is", "is_decoder", "not", "past_key_values", "return", "self"], "roberta_prelayernorm/modeling_roberta_prelayernorm.py:RobertaPreLayerNormForCausalLM.__init__": ["False", "GenerationMixin", "If", "Model", "ModelForCausalLM", "ModelLMHead", "ModelLMHeadModel", "ModelModel", "ModelPreTrainedModel", "True", "__init__", "a", "add", "add_pooling_layer", "as", "class", "config", "def", "if", "is_decoder", "lm_head", "logger", "not", "post_init", "self", "standalone", "super", "to", "use", "want", "warning", "you"], "roberta_prelayernorm/modeling_roberta_prelayernorm.py:RobertaPreLayerNormForCausalLM.get_output_embeddings": ["GenerationMixin", "ModelForCausalLM", "ModelPreTrainedModel", "class", "decoder", "def", "get_output_embeddings", "lm_head", "return", "self"], "roberta_prelayernorm/modeling_roberta_prelayernorm.py:RobertaPreLayerNormForCausalLM.set_output_embeddings": ["GenerationMixin", "ModelForCausalLM", "ModelPreTrainedModel", "class", "decoder", "def", "lm_head", "new_embeddings", "self", "set_output_embeddings"], "roberta_prelayernorm/modeling_roberta_prelayernorm.py:RobertaPreLayerNormForCausalLM.forward": ["BaseModelOutputWithPoolingAndCrossAttentions", "CausalLMOutputWithCrossAttentions", "False", "FloatTensor", "GenerationMixin", "LongTensor", "Model", "ModelForCausalLM", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "cross_attentions", "def", "else", "encoder_attention_mask", "encoder_hidden_states", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "not", "outputs", "past_key_values", "position_ids", "r", "return", "return_dict", "self", "slice", "slice_indices", "token_type_ids", "torch", "tuple", "use_cache", "vocab_size"], "roberta_prelayernorm/modeling_roberta_prelayernorm.py:RobertaPreLayerNormForMaskedLM.__init__": ["False", "If", "Model", "ModelForMaskedLM", "ModelLMHead", "ModelModel", "ModelPreTrainedModel", "__init__", "add_pooling_layer", "attention", "bi", "class", "config", "def", "directional", "for", "if", "is_decoder", "lm_head", "logger", "make", "post_init", "self", "super", "sure", "to", "use", "want", "warning", "you"], "roberta_prelayernorm/modeling_roberta_prelayernorm.py:RobertaPreLayerNormForMaskedLM.get_output_embeddings": ["ModelForMaskedLM", "ModelPreTrainedModel", "class", "decoder", "def", "get_output_embeddings", "lm_head", "return", "self"], "roberta_prelayernorm/modeling_roberta_prelayernorm.py:RobertaPreLayerNormForMaskedLM.set_output_embeddings": ["ModelForMaskedLM", "ModelPreTrainedModel", "class", "decoder", "def", "lm_head", "new_embeddings", "self", "set_output_embeddings"], "roberta_prelayernorm/modeling_roberta_prelayernorm.py:RobertaPreLayerNormForMaskedLM.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "MaskedLMOutput", "Model", "ModelForMaskedLM", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "attention_mask", "attentions", "class", "config", "def", "device", "encoder_attention_mask", "encoder_hidden_states", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "lm_head", "logits", "loss", "loss_fct", "masked_lm_loss", "not", "outputs", "position_ids", "prediction_scores", "r", "return", "return_dict", "self", "sequence_output", "to", "token_type_ids", "torch", "tuple", "view", "vocab_size"], "roberta_prelayernorm/modeling_roberta_prelayernorm.py:RobertaPreLayerNormLMHead.__init__": ["LayerNorm", "Linear", "ModelLMHead", "Module", "Parameter", "__init__", "bias", "class", "config", "decoder", "def", "dense", "eps", "hidden_size", "layer_norm", "layer_norm_eps", "nn", "self", "super", "torch", "vocab_size", "zeros"], "roberta_prelayernorm/modeling_roberta_prelayernorm.py:RobertaPreLayerNormLMHead.forward": ["ModelLMHead", "Module", "class", "decoder", "def", "dense", "features", "forward", "gelu", "kwargs", "layer_norm", "nn", "return", "self", "x"], "roberta_prelayernorm/modeling_roberta_prelayernorm.py:RobertaPreLayerNormForSequenceClassification.__init__": ["False", "Model", "ModelClassificationHead", "ModelForSequenceClassification", "ModelModel", "ModelPreTrainedModel", "__init__", "add_pooling_layer", "class", "classifier", "config", "def", "num_labels", "post_init", "self", "super"], "roberta_prelayernorm/modeling_roberta_prelayernorm.py:RobertaPreLayerNormForSequenceClassification.forward": ["BCEWithLogitsLoss", "CrossEntropyLoss", "FloatTensor", "LongTensor", "MSELoss", "Model", "ModelForSequenceClassification", "ModelPreTrainedModel", "None", "SequenceClassifierOutput", "Tensor", "TransformersKwargs", "True", "Unpack", "and", "attention_mask", "attentions", "class", "classifier", "config", "def", "device", "dtype", "elif", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "kwargs", "labels", "logits", "long", "loss", "loss_fct", "multi_label_classification", "not", "num_labels", "or", "outputs", "position_ids", "problem_type", "r", "regression", "return", "return_dict", "self", "sequence_output", "single_label_classification", "squeeze", "to", "token_type_ids", "torch", "tuple", "view"], "roberta_prelayernorm/modeling_roberta_prelayernorm.py:RobertaPreLayerNormForMultipleChoice.__init__": ["Dropout", "Linear", "Model", "ModelForMultipleChoice", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "classifier", "config", "def", "dropout", "hidden_dropout_prob", "hidden_size", "nn", "post_init", "self", "super"], "roberta_prelayernorm/modeling_roberta_prelayernorm.py:RobertaPreLayerNormForMultipleChoice.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "Model", "ModelForMultipleChoice", "ModelPreTrainedModel", "MultipleChoiceModelOutput", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "attention_mask", "attentions", "class", "classifier", "def", "device", "dropout", "else", "flat_attention_mask", "flat_input_ids", "flat_inputs_embeds", "flat_position_ids", "flat_token_type_ids", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "not", "num_choices", "outputs", "pooled_output", "position_ids", "r", "reshaped_logits", "return", "return_dict", "self", "shape", "size", "to", "token_type_ids", "torch", "tuple", "view"], "roberta_prelayernorm/modeling_roberta_prelayernorm.py:RobertaPreLayerNormForTokenClassification.__init__": ["Dropout", "False", "Linear", "Model", "ModelForTokenClassification", "ModelModel", "ModelPreTrainedModel", "None", "__init__", "add_pooling_layer", "class", "classifier", "classifier_dropout", "config", "def", "dropout", "else", "hidden_dropout_prob", "hidden_size", "if", "is", "nn", "not", "num_labels", "post_init", "self", "super"], "roberta_prelayernorm/modeling_roberta_prelayernorm.py:RobertaPreLayerNormForTokenClassification.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "Model", "ModelForTokenClassification", "ModelPreTrainedModel", "None", "Tensor", "TokenClassifierOutput", "TransformersKwargs", "True", "Unpack", "attention_mask", "attentions", "class", "classifier", "def", "device", "dropout", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "not", "num_labels", "outputs", "position_ids", "r", "return", "return_dict", "self", "sequence_output", "to", "token_type_ids", "torch", "tuple", "view"], "roberta_prelayernorm/modeling_roberta_prelayernorm.py:RobertaPreLayerNormClassificationHead.__init__": ["Dropout", "Linear", "ModelClassificationHead", "Module", "None", "__init__", "class", "classifier_dropout", "config", "def", "dense", "dropout", "else", "hidden_dropout_prob", "hidden_size", "if", "is", "nn", "not", "num_labels", "out_proj", "self", "super"], "roberta_prelayernorm/modeling_roberta_prelayernorm.py:RobertaPreLayerNormClassificationHead.forward": ["ModelClassificationHead", "Module", "class", "def", "dense", "dropout", "features", "forward", "kwargs", "nn", "out_proj", "return", "self", "tanh", "torch", "x"], "roberta_prelayernorm/modeling_roberta_prelayernorm.py:RobertaPreLayerNormForQuestionAnswering.__init__": ["False", "Linear", "Model", "ModelForQuestionAnswering", "ModelModel", "ModelPreTrainedModel", "__init__", "add_pooling_layer", "class", "config", "def", "hidden_size", "nn", "num_labels", "post_init", "qa_outputs", "self", "super"], "roberta_prelayernorm/modeling_roberta_prelayernorm.py:RobertaPreLayerNormForQuestionAnswering.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "Model", "ModelForQuestionAnswering", "ModelPreTrainedModel", "None", "QuestionAnsweringModelOutput", "Tensor", "TransformersKwargs", "True", "Unpack", "and", "attention_mask", "attentions", "clamp", "class", "contiguous", "def", "dim", "end_logits", "end_loss", "end_positions", "forward", "hidden_states", "if", "ignore_index", "ignored_index", "input_ids", "inputs_embeds", "is", "kwargs", "len", "logits", "loss", "loss_fct", "not", "outputs", "position_ids", "qa_outputs", "r", "return", "return_dict", "self", "sequence_output", "size", "split", "squeeze", "start_logits", "start_loss", "start_positions", "token_type_ids", "torch", "total_loss", "tuple"], "roc_bert/modeling_roc_bert.py:RoCBertEmbeddings.__init__": ["Dropout", "Embedding", "False", "LayerNorm", "Linear", "ModelEmbeddings", "Module", "None", "__init__", "arange", "class", "concat_input", "config", "def", "device", "dropout", "dtype", "else", "enable_pronunciation", "enable_shape", "eps", "expand", "hidden_dropout_prob", "hidden_size", "if", "input_dim", "layer_norm_eps", "long", "map_inputs_layer", "max_position_embeddings", "nn", "pad_token_id", "padding_idx", "persistent", "position_embeddings", "position_ids", "pronunciation_dim", "pronunciation_embed", "pronunciation_embed_dim", "pronunciation_vocab_size", "register_buffer", "self", "shape_dim", "shape_embed", "shape_embed_dim", "shape_vocab_size", "size", "super", "token_type_embeddings", "token_type_ids", "torch", "type_vocab_size", "vocab_size", "word_embeddings", "zeros"], "roc_bert/modeling_roc_bert.py:RoCBertEmbeddings.forward": ["LayerNorm", "ModelEmbeddings", "Module", "None", "and", "batch_size", "buffered_token_type_ids", "cat", "class", "clone", "def", "denominator", "device", "dim", "dropout", "dtype", "else", "embedding_in", "embedding_pronunciation", "embedding_shape", "embeddings", "enable_pronunciation", "enable_shape", "expand", "forward", "gather", "hasattr", "if", "index", "input_ids", "input_pronunciation_ids", "input_shape", "input_shape_ids", "inputs_embeds", "is", "long", "map_inputs_layer", "nn", "not", "past_key_values_length", "position_embeddings", "position_ids", "pronunciation_embed", "return", "self", "seq_length", "shape", "shape_embed", "size", "token_type_embeddings", "token_type_ids", "torch", "word_embeddings", "zeros"], "roc_bert/modeling_roc_bert.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "contiguous", "def", "dim", "dropout", "float", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "shape", "size", "softmax", "torch", "training", "transpose", "value"], "roc_bert/modeling_roc_bert.py:RoCBertSelfAttention.__init__": ["Dropout", "False", "Linear", "ModelSelfAttention", "Module", "None", "The", "ValueError", "__init__", "a", "all_head_size", "and", "attention", "attention_head_size", "attention_probs_dropout_prob", "class", "config", "def", "dropout", "embedding_size", "f", "hasattr", "heads", "hidden", "hidden_size", "if", "int", "is", "is_causal", "is_decoder", "key", "layer_idx", "multiple", "nn", "not", "num_attention_heads", "number", "of", "query", "raise", "scaling", "self", "size", "super", "the", "value"], "roc_bert/modeling_roc_bert.py:RoCBertSelfAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "EncoderDecoderCache", "FloatTensor", "ModelSelfAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_attn_implementation", "attention_head_size", "attention_interface", "attention_mask", "attn_output", "attn_weights", "cache_position", "class", "config", "contiguous", "current_past_key_values", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "hidden_shape", "hidden_states", "if", "input_shape", "is", "isinstance", "key", "key_layer", "kwargs", "layer_idx", "nn", "not", "p", "past_key_values", "query", "query_layer", "reshape", "return", "scaling", "self", "self_attention_cache", "shape", "torch", "training", "transpose", "tuple", "update", "value", "value_layer", "view"], "roc_bert/modeling_roc_bert.py:RoCBertCrossAttention.__init__": ["Dropout", "False", "Linear", "ModelCrossAttention", "Module", "None", "The", "ValueError", "__init__", "a", "all_head_size", "and", "attention", "attention_head_size", "attention_probs_dropout_prob", "class", "config", "def", "dropout", "embedding_size", "f", "hasattr", "heads", "hidden", "hidden_size", "if", "int", "is", "is_causal", "key", "layer_idx", "multiple", "nn", "not", "num_attention_heads", "number", "of", "query", "raise", "scaling", "self", "size", "super", "the", "value"], "roc_bert/modeling_roc_bert.py:RoCBertCrossAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "EncoderDecoderCache", "False", "FloatTensor", "ModelCrossAttention", "Module", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "_attn_implementation", "and", "attention_head_size", "attention_interface", "attention_mask", "attn_output", "attn_weights", "bsz", "class", "config", "contiguous", "cross_attention_cache", "def", "dropout", "eager", "eager_attention_forward", "else", "encoder_hidden_states", "forward", "get", "hidden_states", "if", "is", "is_updated", "key", "key_layer", "keys", "kv_input_shape", "kwargs", "layer_idx", "layers", "nn", "not", "p", "past_key_values", "q_input_shape", "query", "query_layer", "reshape", "return", "scaling", "self", "shape", "src_len", "tgt_len", "torch", "training", "transpose", "tuple", "update", "value", "value_layer", "values", "view"], "roc_bert/modeling_roc_bert.py:RoCBertSelfOutput.__init__": ["Dropout", "LayerNorm", "Linear", "ModelSelfOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "eps", "hidden_dropout_prob", "hidden_size", "layer_norm_eps", "nn", "self", "super"], "roc_bert/modeling_roc_bert.py:RoCBertSelfOutput.forward": ["LayerNorm", "ModelSelfOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "roc_bert/modeling_roc_bert.py:RoCBertAttention.__init__": ["False", "ModelAttention", "ModelCrossAttention", "ModelSelfAttention", "ModelSelfOutput", "Module", "None", "__init__", "attention_class", "class", "config", "def", "else", "if", "is_causal", "is_cross_attention", "layer_idx", "nn", "output", "self", "super"], "roc_bert/modeling_roc_bert.py:RoCBertAttention.forward": ["Cache", "FloatTensor", "ModelAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attention_output", "attn_weights", "cache_position", "class", "def", "else", "encoder_attention_mask", "encoder_hidden_states", "forward", "hidden_states", "if", "is_cross_attention", "kwargs", "nn", "not", "output", "past_key_values", "return", "self", "torch", "tuple"], "roc_bert/modeling_roc_bert.py:RoCBertIntermediate.__init__": ["ACT2FN", "Linear", "ModelIntermediate", "Module", "__init__", "class", "config", "def", "dense", "else", "hidden_act", "hidden_size", "if", "intermediate_act_fn", "intermediate_size", "isinstance", "nn", "self", "str", "super"], "roc_bert/modeling_roc_bert.py:RoCBertIntermediate.forward": ["ModelIntermediate", "Module", "Tensor", "class", "def", "dense", "forward", "hidden_states", "intermediate_act_fn", "nn", "return", "self", "torch"], "roc_bert/modeling_roc_bert.py:RoCBertOutput.__init__": ["Dropout", "LayerNorm", "Linear", "ModelOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "eps", "hidden_dropout_prob", "hidden_size", "intermediate_size", "layer_norm_eps", "nn", "self", "super"], "roc_bert/modeling_roc_bert.py:RoCBertOutput.forward": ["LayerNorm", "ModelOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "roc_bert/modeling_roc_bert.py:RoCBertLayer.__init__": ["False", "GradientCheckpointingLayer", "ModelAttention", "ModelIntermediate", "ModelLayer", "ModelOutput", "None", "True", "ValueError", "__init__", "a", "add_cross_attention", "added", "as", "attention", "be", "chunk_size_feed_forward", "class", "config", "cross", "crossattention", "decoder", "def", "f", "if", "intermediate", "is", "is_causal", "is_cross_attention", "is_decoder", "layer_idx", "model", "not", "output", "raise", "self", "seq_len_dim", "should", "super", "used"], "roc_bert/modeling_roc_bert.py:RoCBertLayer.forward": ["Cache", "FloatTensor", "GradientCheckpointingLayer", "If", "ModelLayer", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "ValueError", "_", "add_cross_attention", "and", "apply_chunking_to_forward", "are", "attention", "attention_mask", "attention_output", "be", "by", "cache_position", "chunk_size_feed_forward", "class", "config", "cross", "cross_attention_output", "crossattention", "def", "encoder_attention_mask", "encoder_hidden_states", "f", "feed_forward_chunk", "forward", "has", "hasattr", "hidden_states", "if", "instantiated", "is", "is_decoder", "kwargs", "layer_output", "layers", "not", "passed", "past_key_values", "raise", "return", "self", "self_attention_output", "seq_len_dim", "setting", "to", "torch", "tuple", "with"], "roc_bert/modeling_roc_bert.py:RoCBertLayer.feed_forward_chunk": ["GradientCheckpointingLayer", "ModelLayer", "attention_output", "class", "def", "feed_forward_chunk", "intermediate", "intermediate_output", "layer_output", "output", "return", "self"], "roc_bert/modeling_roc_bert.py:RoCBertEncoder.__init__": ["ModelEncoder", "ModelLayer", "Module", "ModuleList", "__init__", "class", "config", "def", "for", "i", "in", "layer", "layer_idx", "nn", "num_hidden_layers", "range", "self", "super"], "roc_bert/modeling_roc_bert.py:RoCBertEncoder.forward": ["BaseModelOutputWithPastAndCrossAttentions", "Cache", "FloatTensor", "ModelEncoder", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "bool", "cache_position", "class", "def", "else", "encoder_attention_mask", "encoder_hidden_states", "enumerate", "for", "forward", "hidden_states", "i", "if", "in", "kwargs", "last_hidden_state", "layer", "layer_module", "nn", "past_key_values", "return", "self", "torch", "tuple", "use_cache"], "roc_bert/modeling_roc_bert.py:RoCBertPooler.__init__": ["Linear", "ModelPooler", "Module", "Tanh", "__init__", "activation", "class", "config", "def", "dense", "hidden_size", "nn", "self", "super"], "roc_bert/modeling_roc_bert.py:RoCBertPooler.forward": ["ModelPooler", "Module", "Tensor", "activation", "class", "def", "dense", "first_token_tensor", "forward", "hidden_states", "nn", "pooled_output", "return", "self", "torch"], "roc_bert/modeling_roc_bert.py:RoCBertPredictionHeadTransform.__init__": ["ACT2FN", "LayerNorm", "Linear", "ModelPredictionHeadTransform", "Module", "__init__", "class", "config", "def", "dense", "else", "eps", "hidden_act", "hidden_size", "if", "isinstance", "layer_norm_eps", "nn", "self", "str", "super", "transform_act_fn"], "roc_bert/modeling_roc_bert.py:RoCBertPredictionHeadTransform.forward": ["LayerNorm", "ModelPredictionHeadTransform", "Module", "Tensor", "class", "def", "dense", "forward", "hidden_states", "nn", "return", "self", "torch", "transform_act_fn"], "roc_bert/modeling_roc_bert.py:RoCBertLMPredictionHead.__init__": ["Linear", "ModelLMPredictionHead", "ModelPredictionHeadTransform", "Module", "Parameter", "True", "__init__", "bias", "class", "config", "decoder", "def", "hidden_size", "nn", "self", "super", "torch", "transform", "vocab_size", "zeros"], "roc_bert/modeling_roc_bert.py:RoCBertLMPredictionHead.forward": ["ModelLMPredictionHead", "Module", "class", "decoder", "def", "forward", "hidden_states", "nn", "return", "self", "transform"], "roc_bert/modeling_roc_bert.py:RoCBertOnlyMLMHead.__init__": ["ModelLMPredictionHead", "ModelOnlyMLMHead", "Module", "__init__", "class", "config", "def", "nn", "predictions", "self", "super"], "roc_bert/modeling_roc_bert.py:RoCBertOnlyMLMHead.forward": ["ModelOnlyMLMHead", "Module", "Tensor", "class", "def", "forward", "nn", "prediction_scores", "predictions", "return", "self", "sequence_output", "torch"], "roc_bert/modeling_roc_bert.py:RoCBertPreTrainedModel._init_weights": ["ModelEmbeddings", "ModelLMPredictionHead", "ModelPreTrainedModel", "PreTrainedModel", "_init_weights", "arange", "bias", "class", "copy_", "def", "elif", "expand", "if", "init", "isinstance", "module", "position_ids", "self", "shape", "super", "token_type_ids", "torch", "zeros_"], "roc_bert/modeling_roc_bert.py:RoCBertModel.__init__": ["False", "ModelEmbeddings", "ModelEncoder", "ModelModel", "ModelPooler", "ModelPreTrainedModel", "None", "True", "__init__", "add_pooling_layer", "class", "config", "def", "else", "embeddings", "encoder", "gradient_checkpointing", "if", "pooler", "post_init", "r", "self", "super"], "roc_bert/modeling_roc_bert.py:RoCBertModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "return", "self", "word_embeddings"], "roc_bert/modeling_roc_bert.py:RoCBertModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "self", "set_input_embeddings", "value", "word_embeddings"], "roc_bert/modeling_roc_bert.py:RoCBertModel.get_pronunciation_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "get_pronunciation_embeddings", "pronunciation_embed", "return", "self"], "roc_bert/modeling_roc_bert.py:RoCBertModel.set_pronunciation_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "pronunciation_embed", "self", "set_pronunciation_embeddings", "value"], "roc_bert/modeling_roc_bert.py:RoCBertModel.get_shape_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "get_shape_embeddings", "return", "self", "shape_embed"], "roc_bert/modeling_roc_bert.py:RoCBertModel.set_shape_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "self", "set_shape_embeddings", "shape_embed", "value"], "roc_bert/modeling_roc_bert.py:RoCBertModel.forward": ["BaseModelOutputWithPoolingAndCrossAttentions", "Cache", "DynamicCache", "EncoderDecoderCache", "False", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "_create_attention_masks", "and", "arange", "attention_mask", "bool", "cache_position", "class", "config", "def", "device", "else", "embedding_output", "embeddings", "encoder", "encoder_attention_mask", "encoder_hidden_states", "encoder_outputs", "exactly", "forward", "get_seq_length", "if", "input_ids", "input_pronunciation_ids", "input_shape", "input_shape_ids", "inputs_embeds", "is", "is_decoder", "kwargs", "last_hidden_state", "must", "not", "of", "one", "or", "past_key_values", "past_key_values_length", "pooled_output", "pooler", "pooler_output", "position_ids", "r", "raise", "return", "self", "seq_length", "sequence_output", "shape", "specify", "token_type_ids", "torch", "tuple", "use_cache"], "roc_bert/modeling_roc_bert.py:RoCBertModel._create_attention_masks": ["ModelModel", "ModelPreTrainedModel", "None", "_create_attention_masks", "attention_mask", "cache_position", "class", "config", "create_bidirectional_mask", "create_causal_mask", "def", "else", "embedding_output", "encoder_attention_mask", "encoder_hidden_states", "if", "input_embeds", "is", "is_decoder", "not", "past_key_values", "return", "self"], "roc_bert/modeling_roc_bert.py:RoCBertForPreTraining.__init__": ["Model", "ModelForPreTraining", "ModelModel", "ModelOnlyMLMHead", "ModelPreTrainedModel", "__init__", "class", "cls", "config", "def", "post_init", "self", "super"], "roc_bert/modeling_roc_bert.py:RoCBertForPreTraining.get_output_embeddings": ["ModelForPreTraining", "ModelPreTrainedModel", "class", "cls", "decoder", "def", "get_output_embeddings", "predictions", "return", "self"], "roc_bert/modeling_roc_bert.py:RoCBertForPreTraining.set_output_embeddings": ["ModelForPreTraining", "ModelPreTrainedModel", "bias", "class", "cls", "decoder", "def", "new_embeddings", "predictions", "self", "set_output_embeddings"], "roc_bert/modeling_roc_bert.py:RoCBertForPreTraining.forward": ["CrossEntropyLoss", "MaskedLMOutput", "Model", "ModelForPreTraining", "ModelPreTrainedModel", "None", "T", "Tensor", "TransformersKwargs", "True", "Unpack", "_", "attack_attention_mask", "attack_input_ids", "attack_input_pronunciation_ids", "attack_input_shape_ids", "attack_output", "attack_pooled_output", "attack_pooled_output_norm", "attack_token_type_ids", "attention_mask", "attentions", "batch_labels", "batch_size", "class", "clone", "cls", "config", "contrastive_loss", "def", "device", "dim", "else", "forward", "functional", "hidden_states", "if", "input_ids", "input_pronunciation_ids", "input_shape_ids", "inputs_embeds", "is", "kwargs", "labels_attention_mask", "labels_input_ids", "labels_input_pronunciation_ids", "labels_input_shape_ids", "labels_output", "labels_pooled_output", "labels_pooled_output_norm", "labels_token_type_ids", "list", "logits", "loss", "loss_fct", "masked_lm_loss", "matmul", "nn", "normalize", "not", "outputs", "pad_token_id", "pooled_output", "pooled_output_norm", "position_ids", "prediction_scores", "r", "range", "return", "return_dict", "self", "sequence_output", "shape", "sim_matrix", "sim_matrix_target", "target_inputs", "tensor", "token_type_ids", "torch", "tuple", "view", "vocab_size"], "roc_bert/modeling_roc_bert.py:RoCBertForMaskedLM.__init__": ["False", "If", "Model", "ModelForMaskedLM", "ModelModel", "ModelOnlyMLMHead", "ModelPreTrainedModel", "__init__", "add_pooling_layer", "attention", "bi", "class", "cls", "config", "def", "directional", "for", "if", "is_decoder", "logger", "make", "post_init", "self", "super", "sure", "to", "use", "want", "warning", "you"], "roc_bert/modeling_roc_bert.py:RoCBertForMaskedLM.get_output_embeddings": ["ModelForMaskedLM", "ModelPreTrainedModel", "class", "cls", "decoder", "def", "get_output_embeddings", "predictions", "return", "self"], "roc_bert/modeling_roc_bert.py:RoCBertForMaskedLM.set_output_embeddings": ["ModelForMaskedLM", "ModelPreTrainedModel", "bias", "class", "cls", "decoder", "def", "new_embeddings", "predictions", "self", "set_output_embeddings"], "roc_bert/modeling_roc_bert.py:RoCBertForMaskedLM.forward": ["CrossEntropyLoss", "MaskedLMOutput", "Model", "ModelForMaskedLM", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "attention_mask", "attentions", "class", "cls", "config", "def", "encoder_attention_mask", "encoder_hidden_states", "forward", "hidden_states", "if", "input_ids", "input_pronunciation_ids", "input_shape_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "masked_lm_loss", "not", "outputs", "position_ids", "prediction_scores", "r", "return", "return_dict", "self", "sequence_output", "token_type_ids", "torch", "tuple", "view", "vocab_size"], "roc_bert/modeling_roc_bert.py:RoCBertForMaskedLM.prepare_inputs_for_generation": ["ModelForMaskedLM", "ModelPreTrainedModel", "None", "PAD", "The", "ValueError", "attention_mask", "be", "cat", "class", "config", "def", "defined", "device", "dim", "dtype", "dummy_token", "effective_batch_size", "for", "full", "generation", "if", "input_ids", "input_pronunciation_ids", "input_shape", "input_shape_ids", "is", "long", "model_kwargs", "new_zeros", "not", "pad_token_id", "prepare_inputs_for_generation", "raise", "return", "self", "shape", "should", "token", "torch"], "roc_bert/modeling_roc_bert.py:RoCBertForMaskedLM.can_generate": ["False", "ModelForMaskedLM", "ModelPreTrainedModel", "bool", "can_generate", "class", "cls", "def", "return"], "roc_bert/modeling_roc_bert.py:RoCBertForCausalLM.__init__": ["False", "GenerationMixin", "If", "Model", "ModelForCausalLM", "ModelModel", "ModelOnlyMLMHead", "ModelPreTrainedModel", "RoCModelForCausalLM", "True", "__init__", "a", "add", "add_pooling_layer", "as", "class", "cls", "config", "def", "if", "is_decoder", "logger", "not", "post_init", "self", "standalone", "super", "to", "use", "want", "warning", "you"], "roc_bert/modeling_roc_bert.py:RoCBertForCausalLM.get_output_embeddings": ["GenerationMixin", "ModelForCausalLM", "ModelPreTrainedModel", "class", "cls", "decoder", "def", "get_output_embeddings", "predictions", "return", "self"], "roc_bert/modeling_roc_bert.py:RoCBertForCausalLM.set_output_embeddings": ["GenerationMixin", "ModelForCausalLM", "ModelPreTrainedModel", "bias", "class", "cls", "decoder", "def", "new_embeddings", "predictions", "self", "set_output_embeddings"], "roc_bert/modeling_roc_bert.py:RoCBertForCausalLM.forward": ["BaseModelOutputWithPoolingAndCrossAttentions", "CausalLMOutputWithCrossAttentions", "GenerationMixin", "Model", "ModelForCausalLM", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "attention_mask", "attentions", "bool", "cache_position", "class", "cls", "config", "cross_attentions", "def", "else", "encoder_attention_mask", "encoder_hidden_states", "forward", "hidden_states", "if", "input_ids", "input_pronunciation_ids", "input_shape_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "list", "logits", "logits_to_keep", "loss", "loss_function", "not", "outputs", "past_key_values", "position_ids", "r", "return", "return_dict", "self", "slice", "slice_indices", "token_type_ids", "torch", "tuple", "use_cache", "vocab_size"], "roc_bert/modeling_roc_bert.py:RoCBertForCausalLM.prepare_inputs_for_generation": ["GenerationMixin", "ModelForCausalLM", "ModelPreTrainedModel", "None", "attention_mask", "class", "def", "if", "input_ids", "input_pronunciation_ids", "input_shape_ids", "is", "model_inputs", "model_kwargs", "not", "past_key_values", "prepare_inputs_for_generation", "return", "self", "super"], "roc_bert/modeling_roc_bert.py:RoCBertForSequenceClassification.__init__": ["Dropout", "Linear", "Model", "ModelForSequenceClassification", "ModelModel", "ModelPreTrainedModel", "None", "__init__", "class", "classifier", "classifier_dropout", "config", "def", "dropout", "else", "hidden_dropout_prob", "hidden_size", "if", "is", "nn", "not", "num_labels", "post_init", "self", "super"], "roc_bert/modeling_roc_bert.py:RoCBertForSequenceClassification.forward": ["BCEWithLogitsLoss", "CrossEntropyLoss", "MSELoss", "Model", "ModelForSequenceClassification", "ModelPreTrainedModel", "None", "SequenceClassifierOutput", "Tensor", "TransformersKwargs", "True", "Unpack", "and", "attention_mask", "attentions", "class", "classifier", "config", "def", "dropout", "dtype", "elif", "else", "forward", "hidden_states", "if", "input_ids", "input_pronunciation_ids", "input_shape_ids", "inputs_embeds", "int", "is", "kwargs", "labels", "logits", "long", "loss", "loss_fct", "multi_label_classification", "not", "num_labels", "or", "outputs", "pooled_output", "position_ids", "problem_type", "r", "regression", "return", "return_dict", "self", "single_label_classification", "squeeze", "token_type_ids", "torch", "tuple", "view"], "roc_bert/modeling_roc_bert.py:RoCBertForMultipleChoice.__init__": ["Dropout", "Linear", "Model", "ModelForMultipleChoice", "ModelModel", "ModelPreTrainedModel", "None", "__init__", "class", "classifier", "classifier_dropout", "config", "def", "dropout", "else", "hidden_dropout_prob", "hidden_size", "if", "is", "nn", "not", "post_init", "self", "super"], "roc_bert/modeling_roc_bert.py:RoCBertForMultipleChoice.forward": ["CrossEntropyLoss", "Model", "ModelForMultipleChoice", "ModelPreTrainedModel", "MultipleChoiceModelOutput", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "attention_mask", "attentions", "class", "classifier", "def", "dropout", "else", "forward", "hidden_states", "if", "input_ids", "input_pronunciation_ids", "input_shape_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "not", "num_choices", "outputs", "pooled_output", "position_ids", "r", "reshaped_logits", "return", "return_dict", "self", "shape", "size", "token_type_ids", "torch", "tuple", "view"], "roc_bert/modeling_roc_bert.py:RoCBertForTokenClassification.__init__": ["Dropout", "False", "Linear", "Model", "ModelForTokenClassification", "ModelModel", "ModelPreTrainedModel", "None", "__init__", "add_pooling_layer", "class", "classifier", "classifier_dropout", "config", "def", "dropout", "else", "hidden_dropout_prob", "hidden_size", "if", "is", "nn", "not", "num_labels", "post_init", "self", "super"], "roc_bert/modeling_roc_bert.py:RoCBertForTokenClassification.forward": ["CrossEntropyLoss", "Model", "ModelForTokenClassification", "ModelPreTrainedModel", "None", "Tensor", "TokenClassifierOutput", "TransformersKwargs", "True", "Unpack", "attention_mask", "attentions", "class", "classifier", "def", "dropout", "forward", "hidden_states", "if", "input_ids", "input_pronunciation_ids", "input_shape_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "not", "num_labels", "outputs", "position_ids", "r", "return", "return_dict", "self", "sequence_output", "token_type_ids", "torch", "tuple", "view"], "roc_bert/modeling_roc_bert.py:RoCBertForQuestionAnswering.__init__": ["False", "Linear", "Model", "ModelForQuestionAnswering", "ModelModel", "ModelPreTrainedModel", "__init__", "add_pooling_layer", "class", "config", "def", "hidden_size", "nn", "num_labels", "post_init", "qa_outputs", "self", "super"], "roc_bert/modeling_roc_bert.py:RoCBertForQuestionAnswering.forward": ["CrossEntropyLoss", "Model", "ModelForQuestionAnswering", "ModelPreTrainedModel", "None", "QuestionAnsweringModelOutput", "Tensor", "TransformersKwargs", "True", "Unpack", "and", "attention_mask", "attentions", "clamp", "class", "def", "dim", "end_logits", "end_loss", "end_positions", "forward", "hidden_states", "if", "ignore_index", "ignored_index", "input_ids", "input_pronunciation_ids", "input_shape_ids", "inputs_embeds", "is", "kwargs", "len", "logits", "loss", "loss_fct", "not", "outputs", "position_ids", "qa_outputs", "r", "return", "return_dict", "self", "sequence_output", "size", "split", "squeeze", "start_logits", "start_loss", "start_positions", "token_type_ids", "torch", "total_loss", "tuple"], "roformer/modeling_roformer.py:RoFormerSinusoidalPositionalEmbedding.__init__": ["Embedding", "ModelSinusoidalPositionalEmbedding", "None", "True", "__init__", "_freeze", "class", "def", "embedding_dim", "int", "nn", "num_positions", "padding_idx", "self", "super"], "roformer/modeling_roformer.py:RoFormerSinusoidalPositionalEmbedding.create_weight": ["Embedding", "False", "FloatTensor", "ModelSinusoidalPositionalEmbedding", "array", "class", "cos", "create_weight", "def", "dim", "dtype", "else", "empty", "for", "if", "in", "j", "n_pos", "nn", "np", "out", "pos", "position_enc", "power", "range", "requires_grad", "return", "self", "sentinel", "shape", "sin", "torch", "weight"], "roformer/modeling_roformer.py:RoFormerSinusoidalPositionalEmbedding.forward": ["Embedding", "ModelSinusoidalPositionalEmbedding", "None", "Size", "Tensor", "arange", "bsz", "class", "def", "device", "dtype", "forward", "if", "input_ids_shape", "int", "is", "long", "nn", "past_key_values_length", "position_ids", "return", "self", "seq_len", "super", "torch", "weight"], "roformer/modeling_roformer.py:RoFormerEmbeddings.__init__": ["Dropout", "Embedding", "LayerNorm", "ModelEmbeddings", "Module", "__init__", "class", "config", "def", "dropout", "embedding_size", "eps", "hidden_dropout_prob", "layer_norm_eps", "nn", "pad_token_id", "padding_idx", "self", "super", "token_type_embeddings", "type_vocab_size", "vocab_size", "word_embeddings"], "roformer/modeling_roformer.py:RoFormerEmbeddings.forward": ["LayerNorm", "ModelEmbeddings", "Module", "None", "class", "def", "device", "dropout", "dtype", "else", "embeddings", "forward", "if", "input_ids", "input_shape", "inputs_embeds", "is", "long", "nn", "not", "return", "self", "size", "token_type_embeddings", "token_type_ids", "torch", "word_embeddings", "zeros"], "roformer/modeling_roformer.py:RoFormerSelfAttention.__init__": ["Dropout", "Linear", "ModelSelfAttention", "Module", "None", "The", "ValueError", "__init__", "a", "all_head_size", "and", "attention", "attention_head_size", "attention_probs_dropout_prob", "class", "config", "def", "dropout", "embedding_size", "f", "hasattr", "heads", "hidden", "hidden_size", "if", "int", "is", "is_decoder", "key", "layer_idx", "multiple", "nn", "not", "num_attention_heads", "number", "of", "query", "raise", "rotary_value", "self", "size", "super", "the", "value"], "roformer/modeling_roformer.py:RoFormerSelfAttention.forward": ["EncoderDecoderCache", "False", "ModelSelfAttention", "Module", "None", "True", "_", "all_head_size", "and", "apply_rotary_position_embeddings", "attention_head_size", "attention_mask", "attention_probs", "attention_scores", "batch_size", "cache_position", "class", "context_layer", "contiguous", "cross_attention_cache", "curr_past_key_values", "current_states", "def", "dim", "dropout", "else", "encoder_hidden_states", "forward", "functional", "get", "hidden_states", "if", "is", "is_cross_attention", "is_updated", "isinstance", "key", "key_layer", "keys", "layer_idx", "layers", "math", "matmul", "new_context_layer_shape", "nn", "not", "num_attention_heads", "output_attentions", "past_key_values", "permute", "query", "query_layer", "return", "rotary_value", "self", "self_attention_cache", "seq_length", "shape", "sinusoidal_pos", "size", "softmax", "sqrt", "torch", "transpose", "update", "value", "value_layer", "values", "view"], "roformer/modeling_roformer.py:RoFormerSelfAttention.apply_rotary_position_embeddings": ["ModelSelfAttention", "Module", "None", "apply_rotary_position_embeddings", "chunk", "class", "cos", "cos_pos", "def", "dim", "if", "is", "key_layer", "nn", "not", "query_layer", "reshape_as", "return", "rotate_half_key_layer", "rotate_half_query_layer", "rotate_half_value_layer", "sin", "sin_pos", "sinusoidal_pos", "stack", "torch", "value_layer"], "roformer/modeling_roformer.py:RoFormerSelfOutput.__init__": ["Dropout", "LayerNorm", "Linear", "ModelSelfOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "eps", "hidden_dropout_prob", "hidden_size", "layer_norm_eps", "nn", "self", "super"], "roformer/modeling_roformer.py:RoFormerSelfOutput.forward": ["LayerNorm", "ModelSelfOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "roformer/modeling_roformer.py:RoFormerAttention.__init__": ["ModelAttention", "ModelSelfAttention", "ModelSelfOutput", "Module", "None", "__init__", "class", "config", "def", "layer_idx", "nn", "output", "self", "super"], "roformer/modeling_roformer.py:RoFormerAttention.forward": ["False", "ModelAttention", "Module", "None", "attention_mask", "attention_output", "cache_position", "class", "def", "encoder_hidden_states", "forward", "hidden_states", "nn", "output", "output_attentions", "outputs", "past_key_values", "return", "self", "self_outputs", "sinusoidal_pos"], "roformer/modeling_roformer.py:RoFormerIntermediate.__init__": ["ACT2FN", "Linear", "ModelIntermediate", "Module", "__init__", "class", "config", "def", "dense", "else", "hidden_act", "hidden_size", "if", "intermediate_act_fn", "intermediate_size", "isinstance", "nn", "self", "str", "super"], "roformer/modeling_roformer.py:RoFormerIntermediate.forward": ["ModelIntermediate", "Module", "Tensor", "class", "def", "dense", "forward", "hidden_states", "intermediate_act_fn", "nn", "return", "self", "torch"], "roformer/modeling_roformer.py:RoFormerOutput.__init__": ["Dropout", "LayerNorm", "Linear", "ModelOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "eps", "hidden_dropout_prob", "hidden_size", "intermediate_size", "layer_norm_eps", "nn", "self", "super"], "roformer/modeling_roformer.py:RoFormerOutput.forward": ["LayerNorm", "ModelOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "roformer/modeling_roformer.py:RoFormerLayer.__init__": ["GradientCheckpointingLayer", "ModelAttention", "ModelIntermediate", "ModelLayer", "ModelOutput", "None", "ValueError", "__init__", "a", "add_cross_attention", "added", "as", "attention", "be", "chunk_size_feed_forward", "class", "config", "cross", "crossattention", "decoder", "def", "f", "if", "intermediate", "is", "is_decoder", "layer_idx", "model", "not", "output", "raise", "self", "seq_len_dim", "should", "super", "used"], "roformer/modeling_roformer.py:RoFormerLayer.forward": ["False", "GradientCheckpointingLayer", "If", "ModelLayer", "None", "True", "ValueError", "add_cross_attention", "and", "apply_chunking_to_forward", "are", "attention", "attention_mask", "attention_output", "be", "by", "cache_position", "chunk_size_feed_forward", "class", "config", "cross", "cross_attention_outputs", "crossattention", "def", "encoder_attention_mask", "encoder_hidden_states", "f", "feed_forward_chunk", "forward", "has", "hasattr", "hidden_states", "if", "instantiated", "is", "is_decoder", "layer_output", "layers", "not", "output_attentions", "outputs", "passed", "past_key_values", "raise", "return", "self", "self_attention_outputs", "seq_len_dim", "setting", "sinusoidal_pos", "to", "with"], "roformer/modeling_roformer.py:RoFormerLayer.feed_forward_chunk": ["GradientCheckpointingLayer", "ModelLayer", "attention_output", "class", "def", "feed_forward_chunk", "intermediate", "intermediate_output", "layer_output", "output", "return", "self"], "roformer/modeling_roformer.py:RoFormerEncoder.__init__": ["False", "ModelEncoder", "ModelLayer", "ModelSinusoidalPositionalEmbedding", "Module", "ModuleList", "__init__", "class", "config", "def", "embed_positions", "for", "gradient_checkpointing", "hidden_size", "i", "in", "layer", "layer_idx", "max_position_embeddings", "nn", "num_attention_heads", "num_hidden_layers", "range", "self", "super"], "roformer/modeling_roformer.py:RoFormerEncoder.forward": ["BaseModelOutputWithPastAndCrossAttentions", "False", "ModelEncoder", "Module", "None", "Setting", "True", "add_cross_attention", "all_cross_attentions", "all_hidden_states", "all_self_attentions", "and", "attention_mask", "attentions", "cache_position", "checkpointing", "class", "config", "cross_attentions", "def", "else", "embed_positions", "encoder_attention_mask", "encoder_hidden_states", "enumerate", "for", "forward", "get_seq_length", "gradient", "gradient_checkpointing", "hidden_states", "i", "if", "in", "incompatible", "is", "last_hidden_state", "layer", "layer_module", "layer_outputs", "logger", "nn", "not", "output_attentions", "output_hidden_states", "past_key_values", "past_key_values_length", "return", "return_dict", "self", "shape", "sinusoidal_pos", "training", "tuple", "use_cache", "v", "warning_once", "with"], "roformer/modeling_roformer.py:RoFormerSequenceSummary.__init__": ["Callable", "Dropout", "Identity", "Linear", "ModelConfig", "ModelSequenceSummary", "Module", "None", "NotImplementedError", "__init__", "activation", "activation_string", "and", "attn", "class", "config", "def", "else", "first_dropout", "get_activation", "getattr", "hasattr", "hidden_size", "if", "last", "last_dropout", "nn", "num_classes", "num_labels", "raise", "self", "summary", "summary_activation", "summary_first_dropout", "summary_last_dropout", "summary_proj_to_labels", "summary_type", "summary_use_proj", "super"], "roformer/modeling_roformer.py:RoFormerSequenceSummary.forward": ["FloatTensor", "LongTensor", "ModelSequenceSummary", "Module", "None", "NotImplementedError", "activation", "attn", "class", "cls_index", "def", "dim", "dtype", "elif", "else", "expand", "first", "first_dropout", "forward", "full_like", "gather", "hidden_states", "if", "is", "last", "last_dropout", "long", "mean", "nn", "output", "raise", "return", "self", "shape", "size", "squeeze", "summary", "summary_type", "torch", "unsqueeze"], "roformer/modeling_roformer.py:RoFormerPredictionHeadTransform.__init__": ["ACT2FN", "LayerNorm", "Linear", "ModelPredictionHeadTransform", "Module", "__init__", "class", "config", "def", "dense", "else", "embedding_size", "eps", "hidden_act", "hidden_size", "if", "isinstance", "layer_norm_eps", "nn", "self", "str", "super", "transform_act_fn"], "roformer/modeling_roformer.py:RoFormerPredictionHeadTransform.forward": ["LayerNorm", "ModelPredictionHeadTransform", "Module", "class", "def", "dense", "forward", "hidden_states", "nn", "return", "self", "transform_act_fn"], "roformer/modeling_roformer.py:RoFormerLMPredictionHead.__init__": ["Linear", "ModelLMPredictionHead", "ModelPredictionHeadTransform", "Module", "Parameter", "True", "__init__", "bias", "class", "config", "decoder", "def", "embedding_size", "nn", "self", "super", "torch", "transform", "vocab_size", "zeros"], "roformer/modeling_roformer.py:RoFormerLMPredictionHead.forward": ["ModelLMPredictionHead", "Module", "class", "decoder", "def", "forward", "hidden_states", "nn", "return", "self", "transform"], "roformer/modeling_roformer.py:RoFormerOnlyMLMHead.__init__": ["ModelLMPredictionHead", "ModelOnlyMLMHead", "Module", "__init__", "class", "config", "def", "nn", "predictions", "self", "super"], "roformer/modeling_roformer.py:RoFormerOnlyMLMHead.forward": ["ModelOnlyMLMHead", "Module", "Tensor", "class", "def", "forward", "nn", "prediction_scores", "predictions", "return", "self", "sequence_output", "torch"], "roformer/modeling_roformer.py:RoFormerPreTrainedModel._init_weights": ["ModelLMPredictionHead", "ModelPreTrainedModel", "ModelSinusoidalPositionalEmbedding", "PreTrainedModel", "_init_weights", "bias", "class", "copy_", "create_weight", "def", "elif", "if", "init", "isinstance", "module", "self", "super", "weight", "zeros_"], "roformer/modeling_roformer.py:RoFormerModel.__init__": ["Linear", "ModelEmbeddings", "ModelEncoder", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "config", "def", "embedding_size", "embeddings", "embeddings_project", "encoder", "hidden_size", "if", "nn", "post_init", "self", "super"], "roformer/modeling_roformer.py:RoFormerModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "return", "self", "word_embeddings"], "roformer/modeling_roformer.py:RoFormerModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "self", "set_input_embeddings", "value", "word_embeddings"], "roformer/modeling_roformer.py:RoFormerModel.forward": ["BaseModelOutputWithPastAndCrossAttentions", "Cache", "DynamicCache", "EncoderDecoderCache", "False", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "ValueError", "You", "_", "and", "at", "attention_mask", "attentions", "batch_size", "bool", "both", "cache_position", "cannot", "class", "config", "cross_attentions", "def", "device", "dtype", "either", "elif", "else", "embedding_output", "embeddings", "embeddings_project", "encoder", "encoder_attention_mask", "encoder_batch_size", "encoder_extended_attention_mask", "encoder_hidden_shape", "encoder_hidden_states", "encoder_outputs", "encoder_sequence_length", "extended_attention_mask", "forward", "get_extended_attention_mask", "get_seq_length", "hasattr", "have", "hidden_states", "if", "input_ids", "input_shape", "inputs_embeds", "invert_attention_mask", "is", "is_decoder", "is_encoder_decoder", "kwargs", "last_hidden_state", "long", "not", "ones", "or", "output_attentions", "output_hidden_states", "past_key_values", "past_key_values_length", "raise", "return", "return_dict", "same", "self", "seq_length", "sequence_output", "size", "specify", "the", "time", "to", "token_type_ids", "torch", "tuple", "use_cache", "use_return_dict", "warn_if_padding_and_no_attention_mask", "zeros"], "roformer/modeling_roformer.py:RoFormerForMaskedLM.__init__": ["False", "If", "Model", "ModelForMaskedLM", "ModelModel", "ModelOnlyMLMHead", "ModelPreTrainedModel", "__init__", "attention", "bi", "class", "cls", "config", "def", "directional", "for", "if", "is_decoder", "logger", "make", "post_init", "self", "super", "sure", "to", "use", "want", "warning", "you"], "roformer/modeling_roformer.py:RoFormerForMaskedLM.get_output_embeddings": ["ModelForMaskedLM", "ModelPreTrainedModel", "class", "cls", "decoder", "def", "get_output_embeddings", "predictions", "return", "self"], "roformer/modeling_roformer.py:RoFormerForMaskedLM.set_output_embeddings": ["ModelForMaskedLM", "ModelPreTrainedModel", "bias", "class", "cls", "decoder", "def", "new_embeddings", "predictions", "self", "set_output_embeddings"], "roformer/modeling_roformer.py:RoFormerForMaskedLM.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "MaskedLMOutput", "Model", "ModelForMaskedLM", "ModelPreTrainedModel", "None", "Tensor", "attention_mask", "attentions", "bool", "class", "cls", "config", "def", "else", "encoder_attention_mask", "encoder_hidden_states", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "masked_lm_loss", "not", "output", "output_attentions", "output_hidden_states", "outputs", "prediction_scores", "r", "return", "return_dict", "self", "sequence_output", "token_type_ids", "torch", "tuple", "use_return_dict", "view", "vocab_size"], "roformer/modeling_roformer.py:RoFormerForMaskedLM.prepare_inputs_for_generation": ["ModelForMaskedLM", "ModelPreTrainedModel", "None", "PAD", "The", "assert", "attention_mask", "be", "cat", "class", "config", "def", "defined", "device", "dim", "dtype", "dummy_token", "effective_batch_size", "for", "full", "generation", "input_ids", "input_shape", "is", "long", "model_kwargs", "new_zeros", "not", "pad_token_id", "prepare_inputs_for_generation", "return", "self", "shape", "should", "token", "torch"], "roformer/modeling_roformer.py:RoFormerForCausalLM.__init__": ["GenerationMixin", "If", "Model", "ModelForCausalLM", "ModelModel", "ModelOnlyMLMHead", "ModelPreTrainedModel", "True", "__init__", "a", "add", "as", "class", "cls", "config", "def", "if", "is_decoder", "logger", "not", "post_init", "self", "standalone", "super", "to", "use", "want", "warning", "you"], "roformer/modeling_roformer.py:RoFormerForCausalLM.get_output_embeddings": ["GenerationMixin", "ModelForCausalLM", "ModelPreTrainedModel", "class", "cls", "decoder", "def", "get_output_embeddings", "predictions", "return", "self"], "roformer/modeling_roformer.py:RoFormerForCausalLM.set_output_embeddings": ["GenerationMixin", "ModelForCausalLM", "ModelPreTrainedModel", "bias", "class", "cls", "decoder", "def", "new_embeddings", "predictions", "self", "set_output_embeddings"], "roformer/modeling_roformer.py:RoFormerForCausalLM.forward": ["Cache", "CausalLMOutputWithCrossAttentions", "FloatTensor", "GenerationMixin", "LongTensor", "Model", "ModelForCausalLM", "ModelPreTrainedModel", "None", "Tensor", "attention_mask", "attentions", "bool", "cache_position", "class", "cls", "config", "cross_attentions", "def", "else", "encoder_attention_mask", "encoder_hidden_states", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "logits", "logits_to_keep", "loss", "loss_function", "not", "output", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "r", "return", "return_dict", "self", "slice", "slice_indices", "token_type_ids", "torch", "tuple", "use_cache", "use_return_dict", "vocab_size"], "roformer/modeling_roformer.py:RoFormerClassificationHead.__init__": ["Dropout", "Linear", "ModelClassificationHead", "Module", "__init__", "class", "config", "def", "dense", "dropout", "hidden_dropout_prob", "hidden_size", "nn", "num_labels", "out_proj", "self", "super"], "roformer/modeling_roformer.py:RoFormerClassificationHead.forward": ["ACT2FN", "ModelClassificationHead", "Module", "class", "config", "def", "dense", "dropout", "features", "forward", "hidden_act", "kwargs", "nn", "out_proj", "return", "self", "x"], "roformer/modeling_roformer.py:RoFormerForSequenceClassification.__init__": ["Model", "ModelClassificationHead", "ModelForSequenceClassification", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "classifier", "config", "def", "num_labels", "post_init", "self", "super"], "roformer/modeling_roformer.py:RoFormerForSequenceClassification.forward": ["BCEWithLogitsLoss", "CrossEntropyLoss", "FloatTensor", "LongTensor", "MSELoss", "Model", "ModelForSequenceClassification", "ModelPreTrainedModel", "None", "SequenceClassifierOutput", "Tensor", "and", "attention_mask", "attentions", "bool", "class", "classifier", "config", "def", "dtype", "elif", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "kwargs", "labels", "logits", "long", "loss", "loss_fct", "multi_label_classification", "not", "num_labels", "or", "output", "output_attentions", "output_hidden_states", "outputs", "problem_type", "r", "regression", "return", "return_dict", "self", "sequence_output", "single_label_classification", "squeeze", "token_type_ids", "torch", "tuple", "use_return_dict", "view"], "roformer/modeling_roformer.py:RoFormerForMultipleChoice.__init__": ["Linear", "Model", "ModelForMultipleChoice", "ModelModel", "ModelPreTrainedModel", "ModelSequenceSummary", "__init__", "class", "classifier", "config", "def", "hidden_size", "nn", "post_init", "self", "sequence_summary", "super"], "roformer/modeling_roformer.py:RoFormerForMultipleChoice.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "Model", "ModelForMultipleChoice", "ModelPreTrainedModel", "MultipleChoiceModelOutput", "None", "Tensor", "attention_mask", "attentions", "bool", "class", "classifier", "config", "def", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "not", "num_choices", "output", "output_attentions", "output_hidden_states", "outputs", "pooled_output", "r", "reshaped_logits", "return", "return_dict", "self", "sequence_output", "sequence_summary", "shape", "size", "token_type_ids", "torch", "tuple", "use_return_dict", "view"], "roformer/modeling_roformer.py:RoFormerForTokenClassification.__init__": ["Dropout", "Linear", "Model", "ModelForTokenClassification", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "classifier", "config", "def", "dropout", "hidden_dropout_prob", "hidden_size", "nn", "num_labels", "post_init", "self", "super"], "roformer/modeling_roformer.py:RoFormerForTokenClassification.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "Model", "ModelForTokenClassification", "ModelPreTrainedModel", "None", "Tensor", "TokenClassifierOutput", "attention_mask", "attentions", "bool", "class", "classifier", "config", "def", "dropout", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "not", "num_labels", "output", "output_attentions", "output_hidden_states", "outputs", "r", "return", "return_dict", "self", "sequence_output", "token_type_ids", "torch", "tuple", "use_return_dict", "view"], "roformer/modeling_roformer.py:RoFormerForQuestionAnswering.__init__": ["Linear", "Model", "ModelForQuestionAnswering", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "config", "def", "hidden_size", "nn", "num_labels", "post_init", "qa_outputs", "self", "super"], "roformer/modeling_roformer.py:RoFormerForQuestionAnswering.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "Model", "ModelForQuestionAnswering", "ModelPreTrainedModel", "None", "QuestionAnsweringModelOutput", "Tensor", "and", "attention_mask", "attentions", "bool", "clamp", "class", "config", "def", "dim", "else", "end_logits", "end_loss", "end_positions", "forward", "hidden_states", "if", "ignore_index", "ignored_index", "input_ids", "inputs_embeds", "is", "kwargs", "len", "logits", "loss", "loss_fct", "not", "output", "output_attentions", "output_hidden_states", "outputs", "qa_outputs", "return", "return_dict", "self", "sequence_output", "size", "split", "squeeze", "start_logits", "start_loss", "start_positions", "token_type_ids", "torch", "total_loss", "tuple", "use_return_dict"], "rt_detr/modeling_rt_detr.py:MultiScaleDeformableAttention.forward": ["False", "ModelScaleDeformableAttention", "Module", "Tensor", "_", "align_corners", "append", "attention_weights", "batch_size", "bilinear", "class", "contiguous", "def", "dim", "enumerate", "flatten", "for", "forward", "functional", "grid_sample", "height", "hidden_dim", "im2col_step", "in", "int", "level_id", "level_start_index", "list", "mode", "nn", "num_heads", "num_levels", "num_points", "num_queries", "output", "padding_mode", "reshape", "return", "sampling_grid_l_", "sampling_grids", "sampling_locations", "sampling_value_l_", "sampling_value_list", "self", "shape", "split", "stack", "sum", "torch", "transpose", "tuple", "value", "value_l_", "value_list", "value_spatial_shapes", "value_spatial_shapes_list", "view", "width", "zeros"], "rt_detr/modeling_rt_detr.py:_get_clones": ["ModuleList", "N", "_get_clones", "def", "for", "i", "in", "nn", "partial_module", "range", "return"], "rt_detr/modeling_rt_detr.py:inverse_sigmoid": ["Model_sigmoid", "clamp", "def", "eps", "log", "max", "min", "return", "torch", "x", "x1", "x2"], "rt_detr/modeling_rt_detr.py:RTDetrFrozenBatchNorm2d.__init__": ["Model", "Module", "__init__", "bias", "class", "def", "n", "nn", "ones", "register_buffer", "running_mean", "running_var", "self", "super", "torch", "weight", "zeros"], "rt_detr/modeling_rt_detr.py:RTDetrFrozenBatchNorm2d._load_from_state_dict": ["Model", "Module", "_load_from_state_dict", "class", "def", "del", "error_msgs", "if", "in", "local_metadata", "missing_keys", "nn", "num_batches_tracked", "num_batches_tracked_key", "prefix", "self", "state_dict", "strict", "super", "unexpected_keys"], "rt_detr/modeling_rt_detr.py:RTDetrFrozenBatchNorm2d.forward": ["Model", "Module", "bias", "class", "def", "epsilon", "forward", "nn", "reshape", "return", "rsqrt", "running_mean", "running_var", "scale", "self", "weight", "x"], "rt_detr/modeling_rt_detr.py:replace_batch_norm": ["BatchNorm2d", "ModelFrozenBatchNorm2d", "Model_batch_norm", "_modules", "bias", "children", "copy_", "def", "device", "for", "if", "in", "isinstance", "len", "list", "meta", "model", "module", "name", "named_children", "new_module", "nn", "num_features", "r", "running_mean", "running_var", "torch", "weight"], "rt_detr/modeling_rt_detr.py:get_contrastive_denoising_training_group": ["Model_contrastive_denoising_training_group", "None", "attn_mask", "batch_size", "bool", "box_noise_scale", "boxes", "center_to_corners_format", "class_embed", "class_labels", "clip_", "corners_to_center_format", "def", "denoise_positive_idx", "denoising_meta_values", "device", "diff", "dn_num_group", "dn_num_split", "dn_positive_idx", "dtype", "else", "float", "for", "full", "i", "idx_block_end", "idx_block_start", "if", "in", "inf", "input_query_bbox", "input_query_class", "int32", "inverse_sigmoid", "known_bbox", "label_noise_ratio", "len", "mask", "max", "max_gt_num", "min", "n", "negative_gt_mask", "new_label", "nonzero", "num_classes", "num_denoising_queries", "num_ground_truths", "num_groups_denoising_queries", "num_gt", "num_queries", "pad_gt_mask", "positive_gt_mask", "rand_like", "rand_part", "rand_sign", "randint_like", "range", "return", "split", "squeeze", "t", "tarModel_size", "tarModels", "tile", "torch", "torch_int", "where", "zeros"], "rt_detr/modeling_rt_detr.py:RTDetrConvEncoder.__init__": ["Model", "Module", "__init__", "backbone", "channels", "class", "config", "def", "freeze_backbone_batch_norms", "if", "intermediate_channel_sizes", "load_backbone", "model", "nn", "no_grad", "replace_batch_norm", "self", "super", "torch", "with"], "rt_detr/modeling_rt_detr.py:RTDetrConvEncoder.forward": ["Model", "Module", "None", "Tensor", "append", "bool", "class", "def", "feature_map", "feature_maps", "features", "float", "for", "forward", "functional", "in", "interpolate", "mask", "model", "nn", "out", "pixel_mask", "pixel_values", "return", "self", "shape", "size", "to", "torch"], "rt_detr/modeling_rt_detr.py:RTDetrConvNormLayer.__init__": ["ACT2CLS", "BatchNorm2d", "Conv2d", "False", "Identity", "Model", "Module", "None", "__init__", "activation", "batch_norm_eps", "bias", "class", "config", "conv", "def", "else", "if", "in_channels", "is", "kernel_size", "nn", "norm", "out_channels", "padding", "self", "stride", "super"], "rt_detr/modeling_rt_detr.py:RTDetrConvNormLayer.forward": ["Model", "Module", "activation", "class", "conv", "def", "forward", "hidden_state", "nn", "norm", "return", "self"], "rt_detr/modeling_rt_detr.py:RTDetrEncoderLayer.__init__": ["ACT2FN", "LayerNorm", "Linear", "Model", "ModelConfig", "ModelMultiheadAttention", "Module", "__init__", "activation_dropout", "activation_fn", "class", "config", "def", "dropout", "embed_dim", "encoder_activation_function", "encoder_ffn_dim", "encoder_hidden_dim", "eps", "fc1", "fc2", "final_layer_norm", "layer_norm_eps", "nn", "normalize_before", "num_attention_heads", "num_heads", "self", "self_attn", "self_attn_layer_norm", "super"], "rt_detr/modeling_rt_detr.py:RTDetrEncoderLayer.forward": ["False", "Model", "Module", "None", "Tensor", "activation_dropout", "activation_fn", "any", "attention_mask", "attn_weights", "bool", "clamp", "clamp_value", "class", "def", "dropout", "dtype", "fc1", "fc2", "final_layer_norm", "finfo", "forward", "functional", "hidden_states", "if", "isinf", "isnan", "kwargs", "max", "min", "nn", "normalize_before", "not", "or", "output_attentions", "outputs", "p", "position_embeddings", "residual", "return", "self", "self_attn", "self_attn_layer_norm", "torch", "training"], "rt_detr/modeling_rt_detr.py:RTDetrRepVggBlock.__init__": ["ACT2CLS", "Identity", "Model", "ModelConfig", "ModelConvNormLayer", "Module", "None", "__init__", "activation", "activation_function", "class", "config", "conv1", "conv2", "def", "else", "encoder_hidden_dim", "hidden_channels", "hidden_expansion", "if", "int", "is", "nn", "padding", "self", "super"], "rt_detr/modeling_rt_detr.py:RTDetrRepVggBlock.forward": ["Model", "Module", "activation", "class", "conv1", "conv2", "def", "forward", "nn", "return", "self", "x", "y"], "rt_detr/modeling_rt_detr.py:RTDetrCSPRepLayer.__init__": ["Identity", "Model", "ModelConfig", "ModelConvNormLayer", "ModelRepVggBlock", "Module", "Sequential", "_", "__init__", "activation", "activation_function", "bottlenecks", "class", "config", "conv1", "conv2", "conv3", "def", "else", "encoder_hidden_dim", "for", "hidden_channels", "hidden_expansion", "if", "in", "in_channels", "int", "nn", "num_blocks", "out_channels", "range", "self", "super"], "rt_detr/modeling_rt_detr.py:RTDetrCSPRepLayer.forward": ["Model", "Module", "bottlenecks", "class", "conv1", "conv2", "conv3", "def", "forward", "hidden_state", "hidden_state_1", "hidden_state_2", "nn", "return", "self"], "rt_detr/modeling_rt_detr.py:RTDetrMultiscaleDeformableAttention.__init__": ["CUDA", "Linear", "Model", "ModelConfig", "Module", "MultiScaleDeformableAttention", "ValueError", "You", "__init__", "a", "and", "attention", "attention_weights", "attn", "authors", "be", "better", "but", "by", "class", "config", "d", "d_model", "def", "dim_per_head", "dimension", "disable_custom_kernels", "divisible", "each", "efficient", "embed_dim", "f", "got", "head", "if", "im2col_step", "implementation", "in", "int", "is", "make", "more", "must", "n_heads", "n_levels", "n_points", "nn", "not", "num_feature_levels", "num_heads", "of", "output_proj", "power", "raise", "sampling_offsets", "self", "set", "super", "the", "to", "value_proj", "warn", "warnings", "which"], "rt_detr/modeling_rt_detr.py:RTDetrMultiscaleDeformableAttention.with_pos_embed": ["Model", "Module", "None", "Tensor", "class", "def", "else", "if", "is", "nn", "position_embeddings", "return", "self", "tensor", "torch", "with_pos_embed"], "rt_detr/modeling_rt_detr.py:RTDetrMultiscaleDeformableAttention.forward": ["F", "False", "Last", "Make", "Model", "Module", "None", "Tensor", "ValueError", "_", "align", "attention_mask", "attention_weights", "attn", "batch_size", "be", "bool", "but", "class", "d_model", "def", "dim", "elif", "else", "encoder", "encoder_attention_mask", "encoder_hidden_states", "f", "float", "for", "forward", "got", "height", "hidden", "hidden_states", "if", "im2col_step", "in", "is", "length", "level_start_index", "masked_fill", "must", "n_heads", "n_levels", "n_points", "nn", "not", "num_coordinates", "num_queries", "of", "offset_normalizer", "or", "output", "output_attentions", "output_proj", "position_embeddings", "raise", "reference_points", "return", "sampling_locations", "sampling_offsets", "self", "sequence", "sequence_length", "shape", "shapes", "softmax", "spatial", "spatial_shapes", "spatial_shapes_list", "stack", "states", "sum", "sure", "the", "to", "torch", "total_elements", "value", "value_proj", "view", "width", "with", "with_pos_embed"], "rt_detr/modeling_rt_detr.py:RTDetrMultiheadAttention.__init__": ["Linear", "Model", "Module", "True", "ValueError", "__init__", "and", "be", "bias", "bool", "by", "class", "def", "divisible", "dropout", "embed_dim", "f", "float", "got", "head_dim", "if", "int", "k_proj", "must", "nn", "num_heads", "out_proj", "q_proj", "raise", "scaling", "self", "super", "v_proj"], "rt_detr/modeling_rt_detr.py:RTDetrMultiheadAttention._reshape": ["Model", "Module", "Tensor", "_reshape", "batch_size", "class", "contiguous", "def", "head_dim", "int", "nn", "num_heads", "return", "self", "seq_len", "tensor", "torch", "transpose", "view"], "rt_detr/modeling_rt_detr.py:RTDetrMultiheadAttention.with_pos_embed": ["Model", "Module", "None", "Tensor", "class", "def", "else", "if", "is", "nn", "position_embeddings", "return", "self", "tensor", "torch", "with_pos_embed"], "rt_detr/modeling_rt_detr.py:RTDetrMultiheadAttention.forward": ["Attention", "False", "Model", "Module", "None", "Tensor", "ValueError", "_reshape", "attention_mask", "attn_output", "attn_probs", "attn_weights", "attn_weights_reshaped", "batch_size", "be", "bmm", "bool", "but", "class", "def", "dim", "dropout", "dtype", "else", "embed_dim", "expand", "f", "forward", "functional", "head_dim", "hidden_states", "hidden_states_original", "if", "inf", "is", "k_proj", "key_states", "mask", "masked_fill_", "nn", "not", "num_heads", "of", "out_proj", "output_attentions", "p", "position_embeddings", "proj_shape", "q_proj", "query_states", "raise", "reshape", "return", "scaling", "self", "should", "size", "softmax", "source_len", "target_len", "torch", "training", "transpose", "tuple", "v_proj", "value_states", "view", "weights", "with_pos_embed", "zeros_like"], "rt_detr/modeling_rt_detr.py:RTDetrDecoderLayer.__init__": ["ACT2FN", "LayerNorm", "Linear", "Model", "ModelConfig", "ModelMultiheadAttention", "ModelMultiscaleDeformableAttention", "Module", "__init__", "activation_dropout", "activation_fn", "attention_dropout", "class", "config", "d_model", "decoder_activation_function", "decoder_attention_heads", "decoder_ffn_dim", "decoder_n_points", "def", "dropout", "embed_dim", "encoder_attn", "encoder_attn_layer_norm", "eps", "fc1", "fc2", "final_layer_norm", "layer_norm_eps", "n_points", "nn", "num_heads", "self", "self_attn", "self_attn_layer_norm", "super"], "rt_detr/modeling_rt_detr.py:RTDetrDecoderLayer.forward": ["False", "Model", "Module", "None", "Tensor", "activation_dropout", "activation_fn", "attention_mask", "bool", "class", "cross_attn_weights", "def", "dropout", "encoder_attention_mask", "encoder_attn", "encoder_attn_layer_norm", "encoder_hidden_states", "fc1", "fc2", "final_layer_norm", "forward", "functional", "hidden_states", "if", "level_start_index", "nn", "output_attentions", "outputs", "p", "position_embeddings", "reference_points", "residual", "return", "second_residual", "self", "self_attn", "self_attn_layer_norm", "self_attn_weights", "spatial_shapes", "spatial_shapes_list", "torch", "training"], "rt_detr/modeling_rt_detr.py:RTDetrPreTrainedModel._init_weights": ["BatchNorm2d", "Conv2d", "LayerNorm", "Linear", "Model", "ModelForObjectDetection", "ModelModel", "ModelMultiscaleDeformableAttention", "None", "PreTrainedModel", "True", "_init_weights", "abs", "and", "arange", "attention_weights", "bbox_embed", "bias", "class", "class_embed", "config", "constant_", "copy_", "cos", "decoder", "def", "default_dtype", "denoising_class_embed", "dtype", "elif", "enc_score_head", "float", "for", "get_default_dtype", "getattr", "grid_init", "hasattr", "i", "if", "in", "init", "initializer_bias_prior_prob", "initializer_range", "int64", "is", "isinstance", "keepdim", "layer", "layers", "learn_initial_query", "log", "math", "max", "mean", "model", "module", "n_heads", "n_levels", "n_points", "nn", "normal_", "not", "num_batches_tracked", "num_denoising", "num_labels", "ones_", "or", "output_proj", "pi", "prior_prob", "range", "repeat", "running_mean", "running_var", "sampling_offsets", "self", "sin", "stack", "std", "thetas", "to", "torch", "value_proj", "view", "weight", "weight_embedding", "xavier_uniform_", "zeros_"], "rt_detr/modeling_rt_detr.py:RTDetrEncoder.__init__": ["Model", "ModelConfig", "ModelLayer", "Module", "ModuleList", "_", "__init__", "class", "config", "def", "encoder_layers", "for", "in", "layers", "nn", "range", "self", "super"], "rt_detr/modeling_rt_detr.py:RTDetrEncoder.forward": ["False", "Model", "Module", "None", "Tensor", "attention_mask", "bool", "class", "def", "for", "forward", "hidden_states", "in", "layer", "layers", "nn", "output_attentions", "pos_embed", "position_embeddings", "return", "self", "src", "src_mask", "torch"], "rt_detr/modeling_rt_detr.py:RTDetrHybridEncoder.__init__": ["Model", "ModelCSPRepLayer", "ModelConfig", "ModelConvNormLayer", "ModelEncoder", "Module", "ModuleList", "_", "__init__", "activation", "activation_function", "append", "class", "config", "def", "downsample_conv", "downsample_convs", "encode_proj_layers", "encoder", "encoder_hidden_dim", "encoder_in_channels", "eval_size", "feat_strides", "for", "fpn_block", "fpn_blocks", "in", "in_channels", "kernel_size", "lateral_conv", "lateral_convs", "len", "nn", "num_fpn_stages", "num_pan_stages", "out_channels", "out_strides", "pan_block", "pan_blocks", "positional_encoding_temperature", "range", "self", "stride", "super"], "rt_detr/modeling_rt_detr.py:RTDetrHybridEncoder.build_2d_sincos_position_embedding": ["Embed", "Model", "Module", "None", "ValueError", "arange", "be", "build_2d_sincos_position_embedding", "by", "class", "concat", "cos", "cpu", "def", "device", "dim", "dimension", "divisible", "dtype", "embed_dim", "embedding", "flatten", "float32", "for", "grid_h", "grid_w", "height", "if", "indexing", "meshgrid", "must", "nn", "omega", "out_h", "out_w", "pos_dim", "position", "raise", "return", "sin", "temperature", "to", "torch", "torch_int", "width", "xy"], "rt_detr/modeling_rt_detr.py:RTDetrHybridEncoder.forward": ["BaseModelOutput", "F", "Model", "Module", "None", "all_attentions", "append", "attention_mask", "attentions", "backbone_feature_map", "build_2d_sincos_position_embedding", "class", "concat", "config", "contiguous", "def", "device", "dim", "downsample_conv", "downsample_convs", "downsampled_feature_map", "dtype", "else", "enc_ind", "encode_proj_layers", "encoder", "encoder_hidden_dim", "encoder_layers", "encoder_states", "enumerate", "eval_size", "flatten", "for", "forward", "fpn_block", "fpn_blocks", "fpn_feature_map", "fpn_feature_maps", "fused_feature_map", "height", "hidden_states", "i", "idx", "if", "in", "inputs_embeds", "interpolate", "is", "last_hidden_state", "lateral_conv", "lateral_convs", "layer_outputs", "level_start_index", "mode", "nearest", "new_fpn_feature_map", "new_pan_feature_map", "nn", "not", "num_fpn_stages", "or", "output_attentions", "output_hidden_states", "pan_block", "pan_blocks", "pan_feature_maps", "permute", "pos_embed", "position_embeddings", "positional_encoding_temperature", "r", "reshape", "return", "return_dict", "reverse", "scale_factor", "self", "shape", "spatial_shapes", "src_flatten", "top_fpn_feature_map", "top_pan_feature_map", "torch", "training", "tuple", "use_return_dict", "v", "valid_ratios", "width", "zip"], "rt_detr/modeling_rt_detr.py:RTDetrDecoder.__init__": ["Model", "ModelConfig", "ModelLayer", "ModelMLPPredictionHead", "ModelPreTrainedModel", "ModuleList", "None", "_", "__init__", "bbox_embed", "class", "class_embed", "config", "d_model", "decoder_layers", "def", "dropout", "for", "in", "layers", "nn", "num_layers", "post_init", "query_pos_head", "range", "self", "super"], "rt_detr/modeling_rt_detr.py:RTDetrDecoder.forward": ["F", "Model", "ModelOutput", "ModelPreTrainedModel", "None", "all_cross_attentions", "all_hidden_states", "all_self_attns", "and", "attentions", "bbox_embed", "class", "class_embed", "config", "cross_attentions", "decoder_layer", "def", "detach", "dim", "else", "encoder_attention_mask", "encoder_hidden_states", "enumerate", "for", "forward", "hidden_states", "idx", "if", "in", "inputs_embeds", "intermediate", "intermediate_hidden_states", "intermediate_logits", "intermediate_reference_points", "inverse_sigmoid", "is", "kwargs", "last_hidden_state", "layer_outputs", "layers", "level_start_index", "logits", "new_reference_points", "not", "output_attentions", "output_hidden_states", "position_embeddings", "predicted_corners", "query_pos_head", "r", "reference_points", "reference_points_input", "return", "return_dict", "self", "sigmoid", "spatial_shapes", "spatial_shapes_list", "stack", "torch", "tuple", "unsqueeze", "use_return_dict", "v", "valid_ratios"], "rt_detr/modeling_rt_detr.py:RTDetrMLPPredictionHead.__init__": ["Linear", "Model", "Module", "ModuleList", "__init__", "class", "config", "d_model", "def", "for", "h", "in", "input_dim", "k", "layers", "n", "nn", "num_layers", "output_dim", "self", "super", "zip"], "rt_detr/modeling_rt_detr.py:RTDetrMLPPredictionHead.forward": ["Model", "Module", "class", "def", "else", "enumerate", "for", "forward", "functional", "i", "if", "in", "layer", "layers", "nn", "num_layers", "relu", "return", "self", "x"], "rt_detr/modeling_rt_detr.py:RTDetrModel.__init__": ["BatchNorm2d", "Conv2d", "Embedding", "False", "LayerNorm", "Linear", "Model", "ModelConfig", "ModelConvEncoder", "ModelDecoder", "ModelHybridEncoder", "ModelMLPPredictionHead", "ModelPreTrainedModel", "ModuleList", "Sequential", "_", "__init__", "anchor_image_size", "anchors", "append", "backbone", "batch_norm_eps", "bias", "class", "config", "d_model", "decoder", "decoder_in_channels", "decoder_input_proj", "decoder_input_proj_list", "def", "denoising_class_embed", "dtype", "enc_bbox_head", "enc_output", "enc_score_head", "encoder", "encoder_hidden_dim", "encoder_input_proj", "encoder_input_proj_list", "eps", "for", "generate_anchors", "if", "in", "in_channels", "intermediate_channel_sizes", "kernel_size", "layer_norm_eps", "learn_initial_query", "len", "nn", "num_backbone_outs", "num_denoising", "num_feature_levels", "num_labels", "num_layers", "num_queries", "padding", "padding_idx", "post_init", "range", "self", "stride", "super", "valid_mask", "weight_embedding"], "rt_detr/modeling_rt_detr.py:RTDetrModel.freeze_backbone": ["False", "Model", "ModelPreTrainedModel", "backbone", "class", "def", "for", "freeze_backbone", "in", "param", "parameters", "requires_grad_", "self"], "rt_detr/modeling_rt_detr.py:RTDetrModel.unfreeze_backbone": ["Model", "ModelPreTrainedModel", "True", "backbone", "class", "def", "for", "in", "param", "parameters", "requires_grad_", "self", "unfreeze_backbone"], "rt_detr/modeling_rt_detr.py:RTDetrModel.generate_anchors": ["Model", "ModelPreTrainedModel", "None", "True", "all", "anchor_image_size", "anchors", "append", "arange", "class", "concat", "config", "cpu", "def", "device", "dtype", "end", "enumerate", "eps", "feat_strides", "finfo", "float32", "for", "generate_anchors", "grid_size", "grid_x", "grid_xy", "grid_y", "height", "if", "ij", "in", "indexing", "int", "is", "keepdim", "level", "log", "max", "meshgrid", "ones_like", "reshape", "return", "s", "self", "spatial_shapes", "stack", "tensor", "to", "torch", "unsqueeze", "valid_mask", "wh", "where", "width"], "rt_detr/modeling_rt_detr.py:RTDetrModel.forward": ["BaseModelOutput", "F", "FloatTensor", "LongTensor", "Model", "ModelOutput", "ModelPreTrainedModel", "None", "_", "_len_sources", "anchor_image_size", "anchors", "and", "append", "attention_mask", "attentions", "backbone", "batch_size", "bool", "box_noise_scale", "cat", "class", "class_embed", "concat", "config", "cross_attentions", "cumsum", "decoder", "decoder_attentions", "decoder_hidden_states", "decoder_input_proj", "decoder_inputs_embeds", "decoder_outputs", "def", "denoising_bbox_unact", "denoising_class", "denoising_class_embed", "denoising_meta_values", "detach", "device", "dict", "dim", "dn_outputs", "dtype", "elif", "else", "empty", "enc_bbox_head", "enc_output", "enc_outputs", "enc_outputs_class", "enc_outputs_coord_logits", "enc_score_head", "enc_topk_bboxes", "enc_topk_logits", "encoder", "encoder_attention_mask", "encoder_attentions", "encoder_hidden_states", "encoder_input_proj", "encoder_last_hidden_state", "encoder_outputs", "enumerate", "features", "flatten", "for", "forward", "gather", "generate_anchors", "get_contrastive_denoising_training_group", "height", "hidden_states", "i", "if", "in", "index", "init_reference_points", "initial_reference_points", "inputs_embeds", "intermediate_hidden_states", "intermediate_logits", "intermediate_predicted_corners", "intermediate_reference_points", "is", "isinstance", "kwargs", "label_noise_ratio", "labels", "last_hidden_state", "learn_initial_query", "len", "level", "level_start_index", "list", "long", "mask", "max", "memory", "new_zeros", "not", "num_channels", "num_classes", "num_denoising", "num_denoising_queries", "num_feature_levels", "num_labels", "num_queries", "ones", "or", "output_attentions", "output_hidden_states", "output_memory", "pixel_mask", "pixel_values", "prod", "proj_feats", "r", "range", "reference_points", "reference_points_unact", "repeat", "return", "return_dict", "self", "shape", "sigmoid", "source", "source_flatten", "sources", "spatial_shapes", "spatial_shapes_list", "spatial_shapes_tuple", "target", "targets", "tile", "to", "topk", "topk_ind", "torch", "training", "transpose", "tuple", "tuple_outputs", "unsqueeze", "use_return_dict", "valid_mask", "value", "values", "weight_embedding", "width"], "rt_detr/modeling_rt_detr.py:RTDetrForObjectDetection.__init__": ["Linear", "Model", "ModelConfig", "ModelMLPPredictionHead", "ModelModel", "ModelPreTrainedModel", "ModuleList", "_", "__init__", "bbox_embed", "class", "class_embed", "config", "d_model", "decoder", "decoder_layers", "def", "for", "in", "model", "nn", "num_labels", "num_layers", "num_pred", "post_init", "range", "self", "super", "torch"], "rt_detr/modeling_rt_detr.py:RTDetrForObjectDetection._set_aux_loss": ["Model", "ModelPreTrainedModel", "_set_aux_loss", "a", "b", "class", "def", "for", "in", "logits", "outputs_class", "outputs_coord", "pred_boxes", "return", "self", "zip"], "rt_detr/modeling_rt_detr.py:RTDetrForObjectDetection.forward": ["FloatTensor", "LongTensor", "Model", "ModelObjectDetectionOutput", "ModelPreTrainedModel", "None", "auxiliary_outputs", "bool", "class", "config", "cross_attentions", "decoder_attentions", "decoder_hidden_states", "decoder_inputs_embeds", "def", "denoising_meta_values", "device", "dict", "else", "enc_outputs_class", "enc_outputs_coord_logits", "enc_topk_bboxes", "enc_topk_logits", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "forward", "if", "init_reference_points", "initial_reference_points", "inputs_embeds", "intermediate_hidden_states", "intermediate_logits", "intermediate_predicted_corners", "intermediate_reference_points", "is", "kwargs", "labels", "last_hidden_state", "list", "logits", "loss", "loss_dict", "loss_function", "model", "not", "output", "output_attentions", "output_hidden_states", "outputs", "outputs_class", "outputs_coord", "pixel_mask", "pixel_values", "pred_boxes", "predicted_corners", "r", "return", "return_dict", "self", "torch", "training", "tuple", "use_return_dict"], "rt_detr/modeling_rt_detr_resnet.py:RTDetrResNetConvLayer.__init__": ["ACT2FN", "BatchNorm2d", "Conv2d", "False", "Identity", "Model", "Module", "None", "__init__", "activation", "bias", "class", "convolution", "def", "else", "if", "in_channels", "int", "is", "kernel_size", "nn", "normalization", "not", "out_channels", "padding", "relu", "self", "str", "stride", "super"], "rt_detr/modeling_rt_detr_resnet.py:RTDetrResNetConvLayer.forward": ["Model", "Module", "Tensor", "activation", "class", "convolution", "def", "forward", "hidden_state", "input", "nn", "normalization", "return", "self"], "rt_detr/modeling_rt_detr_resnet.py:RTDetrResNetEmbeddings.__init__": ["MaxPool2d", "Model", "ModelResNetConfig", "ModelResNetConvLayer", "Module", "Sequential", "__init__", "activation", "class", "config", "def", "embedder", "embedding_size", "hidden_act", "kernel_size", "nn", "num_channels", "padding", "pooler", "self", "stride", "super"], "rt_detr/modeling_rt_detr_resnet.py:RTDetrResNetEmbeddings.forward": ["Make", "Model", "Module", "Tensor", "ValueError", "channel", "class", "configuration", "def", "dimension", "embedder", "embedding", "forward", "if", "in", "match", "nn", "num_channels", "of", "one", "pixel", "pixel_values", "pooler", "raise", "return", "self", "set", "shape", "sure", "that", "the", "values", "with"], "rt_detr/modeling_rt_detr_resnet.py:RTDetrResNetShortCut.__init__": ["BatchNorm2d", "Conv2d", "False", "Model", "Module", "__init__", "bias", "class", "convolution", "def", "in_channels", "int", "kernel_size", "nn", "normalization", "out_channels", "self", "stride", "super"], "rt_detr/modeling_rt_detr_resnet.py:RTDetrResNetShortCut.forward": ["Model", "Module", "Tensor", "class", "convolution", "def", "forward", "hidden_state", "input", "nn", "normalization", "return", "self"], "rt_detr/modeling_rt_detr_resnet.py:RTDetrResNetBasicLayer.__init__": ["ACT2FN", "AvgPool2d", "False", "Identity", "Model", "ModelResNetConfig", "ModelResNetConvLayer", "ModelResNetShortCut", "Module", "None", "Sequential", "True", "__init__", "activation", "bool", "ceil_mode", "class", "config", "def", "else", "hidden_act", "if", "in_channels", "int", "layer", "nn", "out_channels", "self", "shortcut", "should_apply_shortcut", "stride", "super"], "rt_detr/modeling_rt_detr_resnet.py:RTDetrResNetBasicLayer.forward": ["Model", "Module", "activation", "class", "def", "forward", "hidden_state", "layer", "nn", "residual", "return", "self", "shortcut"], "rt_detr/modeling_rt_detr_resnet.py:RTDetrResNetBottleNeckLayer.__init__": ["ACT2FN", "AvgPool2d", "Identity", "Model", "ModelResNetConfig", "ModelResNetConvLayer", "ModelResNetShortCut", "Module", "None", "Sequential", "True", "__init__", "activation", "ceil_mode", "class", "config", "def", "downsample_in_bottleneck", "else", "hidden_act", "if", "in_channels", "int", "kernel_size", "layer", "nn", "not", "or", "out_channels", "reduces_channels", "reduction", "self", "shortcut", "should_apply_shortcut", "stride", "super"], "rt_detr/modeling_rt_detr_resnet.py:RTDetrResNetBottleNeckLayer.forward": ["Model", "Module", "activation", "class", "def", "forward", "hidden_state", "layer", "nn", "residual", "return", "self", "shortcut"], "rt_detr/modeling_rt_detr_resnet.py:RTDetrResNetStage.__init__": ["Model", "ModelResNetBasicLayer", "ModelResNetBottleNeckLayer", "ModelResNetConfig", "Module", "Sequential", "True", "_", "__init__", "bottleneck", "class", "config", "def", "depth", "else", "first_layer", "for", "if", "in", "in_channels", "int", "layer", "layer_type", "layers", "nn", "out_channels", "range", "self", "should_apply_shortcut", "stride", "super"], "rt_detr/modeling_rt_detr_resnet.py:RTDetrResNetStage.forward": ["Model", "Module", "Tensor", "class", "def", "for", "forward", "hidden_state", "in", "input", "layer", "layers", "nn", "return", "self"], "rt_detr/modeling_rt_detr_resnet.py:RTDetrResNetEncoder.__init__": ["Model", "ModelResNetConfig", "ModelResNetStage", "Module", "ModuleList", "__init__", "append", "class", "config", "def", "depth", "depths", "downsample_in_first_stage", "else", "embedding_size", "for", "hidden_sizes", "if", "in", "in_channels", "in_out_channels", "nn", "out_channels", "self", "stages", "stride", "super", "zip"], "rt_detr/modeling_rt_detr_resnet.py:RTDetrResNetEncoder.forward": ["BaseModelOutputWithNoAttention", "False", "Model", "Module", "None", "Tensor", "True", "bool", "class", "def", "else", "for", "forward", "hidden_state", "hidden_states", "if", "in", "is", "last_hidden_state", "nn", "not", "output_hidden_states", "return", "return_dict", "self", "stage_module", "stages", "tuple", "v"], "rt_detr/modeling_rt_detr_resnet.py:RTDetrResNetPreTrainedModel._init_weights": ["BatchNorm", "Conv2d", "Linear", "Model", "None", "PreTrainedModel", "_", "__class__", "__name__", "_calculate_fan_in_and_fan_out", "_init_weights", "a", "bias", "bound", "class", "def", "elif", "else", "fan_in", "fan_out", "getattr", "if", "in", "init", "is", "isinstance", "kaiming_normal_", "kaiming_uniform_", "math", "mode", "module", "nn", "nonlinearity", "not", "num_batches_tracked", "ones_", "relu", "running_mean", "running_var", "self", "sqrt", "torch", "uniform_", "weight", "zeros_"], "rt_detr/modeling_rt_detr_resnet.py:RTDetrResNetBackbone.__init__": ["BackboneMixin", "Model", "ModelResNetEmbeddings", "ModelResNetEncoder", "ModelResNetPreTrainedModel", "__init__", "_init_backbone", "class", "config", "def", "embedder", "embedding_size", "encoder", "hidden_sizes", "num_features", "post_init", "self", "super"], "rt_detr/modeling_rt_detr_resnet.py:RTDetrResNetBackbone.forward": ["BackboneMixin", "BackboneOutput", "Model", "ModelResNetPreTrainedModel", "None", "Tensor", "True", "attentions", "bool", "class", "config", "def", "else", "embedder", "embedding_output", "encoder", "enumerate", "feature_maps", "for", "forward", "hidden_states", "idx", "if", "in", "is", "kwargs", "not", "out_features", "output", "output_hidden_states", "outputs", "pixel_values", "r", "return", "return_dict", "self", "stage", "stage_names", "use_return_dict"], "rt_detr_v2/modeling_rt_detr_v2.py:multi_scale_deformable_attention_v2": ["False", "Model_scale_deformable_attention_v2", "Tensor", "_", "align_corners", "append", "arange", "attention_weights", "batch_size", "bilinear", "clamp", "concat", "contiguous", "def", "default", "device", "dim", "discrete", "elif", "enumerate", "flatten", "for", "functional", "grid_sample", "height", "hidden_dim", "if", "in", "int", "int64", "level_id", "list", "method", "mode", "nn", "num_heads", "num_levels", "num_points", "num_points_list", "num_queries", "output", "padding_mode", "permute", "repeat", "reshape", "return", "sampling_coord", "sampling_coord_x", "sampling_coord_y", "sampling_grid_l_", "sampling_grids", "sampling_idx", "sampling_locations", "sampling_value_l_", "sampling_value_list", "shape", "split", "stack", "sum", "tensor", "to", "torch", "transpose", "unsqueeze", "value", "value_l_", "value_list", "value_spatial_shapes", "view", "width", "zeros"], "rt_detr_v2/modeling_rt_detr_v2.py:RTDetrV2MultiscaleDeformableAttention.__init__": ["CUDA", "Linear", "Model", "ModelConfig", "Module", "ValueError", "You", "_", "__init__", "a", "and", "attention", "attention_weights", "authors", "be", "better", "but", "by", "class", "config", "d", "d_model", "decoder_attention_heads", "decoder_method", "decoder_n_levels", "decoder_n_points", "decoder_offset_scale", "def", "dim_per_head", "dimension", "divisible", "dtype", "each", "efficient", "embed_dim", "f", "float32", "for", "got", "head", "if", "im2col_step", "implementation", "in", "is", "make", "method", "more", "must", "n", "n_heads", "n_levels", "n_points", "n_points_list", "n_points_scale", "nn", "not", "num_heads", "of", "offset_scale", "output_proj", "power", "raise", "range", "register_buffer", "sampling_offsets", "self", "set", "super", "tensor", "the", "to", "torch", "value_proj", "warn", "warnings", "which"], "rt_detr_v2/modeling_rt_detr_v2.py:RTDetrV2MultiscaleDeformableAttention.forward": ["F", "False", "Last", "Make", "Model", "Module", "None", "Tensor", "ValueError", "_", "align", "and", "attention_mask", "attention_weights", "batch_size", "be", "bool", "but", "class", "d_model", "def", "dim", "dtype", "elif", "else", "encoder", "encoder_attention_mask", "encoder_hidden_states", "f", "float", "forward", "got", "hidden", "hidden_states", "if", "is", "is_torchdynamo_compiling", "length", "level_start_index", "masked_fill", "method", "multi_scale_deformable_attention_v2", "must", "n_heads", "n_levels", "n_points", "n_points_list", "n_points_scale", "nn", "not", "num_queries", "of", "offset", "offset_normalizer", "offset_scale", "or", "output", "output_attentions", "output_proj", "position_embeddings", "raise", "reference_points", "return", "sampling_locations", "sampling_offsets", "self", "sequence", "sequence_length", "shape", "shapes", "softmax", "spatial", "spatial_shapes", "spatial_shapes_list", "stack", "states", "sum", "sure", "the", "to", "torch", "unsqueeze", "value", "value_proj", "view", "with"], "rt_detr_v2/modeling_rt_detr_v2.py:RTDetrV2MultiheadAttention.__init__": ["Linear", "Model", "Module", "True", "ValueError", "__init__", "and", "be", "bias", "bool", "by", "class", "def", "divisible", "dropout", "embed_dim", "f", "float", "got", "head_dim", "if", "int", "k_proj", "must", "nn", "num_heads", "out_proj", "q_proj", "raise", "scaling", "self", "super", "v_proj"], "rt_detr_v2/modeling_rt_detr_v2.py:RTDetrV2MultiheadAttention._reshape": ["Model", "Module", "Tensor", "_reshape", "batch_size", "class", "contiguous", "def", "head_dim", "int", "nn", "num_heads", "return", "self", "seq_len", "tensor", "torch", "transpose", "view"], "rt_detr_v2/modeling_rt_detr_v2.py:RTDetrV2MultiheadAttention.with_pos_embed": ["Model", "Module", "None", "Tensor", "class", "def", "else", "if", "is", "nn", "position_embeddings", "return", "self", "tensor", "torch", "with_pos_embed"], "rt_detr_v2/modeling_rt_detr_v2.py:RTDetrV2MultiheadAttention.forward": ["Attention", "False", "Model", "Module", "None", "Tensor", "ValueError", "_reshape", "attention_mask", "attn_output", "attn_probs", "attn_weights", "attn_weights_reshaped", "batch_size", "be", "bmm", "bool", "but", "class", "def", "dim", "dropout", "dtype", "else", "embed_dim", "expand", "f", "forward", "functional", "head_dim", "hidden_states", "hidden_states_original", "if", "inf", "is", "k_proj", "key_states", "mask", "masked_fill_", "nn", "not", "num_heads", "of", "out_proj", "output_attentions", "p", "position_embeddings", "proj_shape", "q_proj", "query_states", "raise", "reshape", "return", "scaling", "self", "should", "size", "softmax", "source_len", "target_len", "torch", "training", "transpose", "tuple", "v_proj", "value_states", "view", "weights", "with_pos_embed", "zeros_like"], "rt_detr_v2/modeling_rt_detr_v2.py:RTDetrV2DecoderLayer.__init__": ["ACT2FN", "LayerNorm", "Linear", "Model", "ModelConfig", "ModelMultiheadAttention", "ModelMultiscaleDeformableAttention", "Module", "__init__", "activation_dropout", "activation_fn", "attention_dropout", "class", "config", "d_model", "decoder_activation_function", "decoder_attention_heads", "decoder_ffn_dim", "def", "dropout", "embed_dim", "encoder_attn", "encoder_attn_layer_norm", "eps", "fc1", "fc2", "final_layer_norm", "layer_norm_eps", "nn", "num_heads", "self", "self_attn", "self_attn_layer_norm", "super"], "rt_detr_v2/modeling_rt_detr_v2.py:RTDetrV2DecoderLayer.forward": ["False", "Model", "Module", "None", "Tensor", "activation_dropout", "activation_fn", "attention_mask", "bool", "class", "cross_attn_weights", "def", "dropout", "encoder_attention_mask", "encoder_attn", "encoder_attn_layer_norm", "encoder_hidden_states", "fc1", "fc2", "final_layer_norm", "forward", "functional", "hidden_states", "if", "level_start_index", "nn", "output_attentions", "outputs", "p", "position_embeddings", "reference_points", "residual", "return", "second_residual", "self", "self_attn", "self_attn_layer_norm", "self_attn_weights", "spatial_shapes", "spatial_shapes_list", "torch", "training"], "rt_detr_v2/modeling_rt_detr_v2.py:RTDetrV2PreTrainedModel._init_weights": ["BatchNorm2d", "Conv2d", "LayerNorm", "Linear", "Model", "ModelForObjectDetection", "ModelModel", "ModelMultiscaleDeformableAttention", "None", "PreTrainedModel", "True", "_", "_init_weights", "abs", "and", "arange", "attention_weights", "bbox_embed", "bias", "class", "class_embed", "config", "constant_", "copy_", "cos", "decoder", "def", "default_dtype", "denoising_class_embed", "dtype", "elif", "enc_score_head", "float", "float32", "for", "get_default_dtype", "getattr", "grid_init", "hasattr", "i", "if", "in", "init", "initializer_bias_prior_prob", "initializer_range", "int64", "is", "isinstance", "keepdim", "layer", "layers", "learn_initial_query", "log", "math", "max", "mean", "model", "module", "n", "n_heads", "n_levels", "n_points", "n_points_list", "n_points_scale", "nn", "normal_", "not", "num_batches_tracked", "num_denoising", "num_labels", "ones_", "or", "output_proj", "pi", "prior_prob", "range", "repeat", "running_mean", "running_var", "sampling_offsets", "self", "sin", "stack", "std", "tensor", "thetas", "to", "torch", "value_proj", "view", "weight", "weight_embedding", "xavier_uniform_", "zeros_"], "rt_detr_v2/modeling_rt_detr_v2.py:inverse_sigmoid": ["Model_sigmoid", "clamp", "def", "eps", "log", "max", "min", "return", "torch", "x", "x1", "x2"], "rt_detr_v2/modeling_rt_detr_v2.py:RTDetrV2Decoder.__init__": ["Model", "ModelConfig", "ModelLayer", "ModelMLPPredictionHead", "ModelPreTrainedModel", "ModuleList", "None", "_", "__init__", "bbox_embed", "class", "class_embed", "config", "d_model", "decoder_layers", "def", "dropout", "for", "in", "layers", "nn", "num_layers", "post_init", "query_pos_head", "range", "self", "super"], "rt_detr_v2/modeling_rt_detr_v2.py:RTDetrV2Decoder.forward": ["F", "Model", "ModelOutput", "ModelPreTrainedModel", "None", "all_cross_attentions", "all_hidden_states", "all_self_attns", "and", "attentions", "bbox_embed", "class", "class_embed", "config", "cross_attentions", "decoder_layer", "def", "detach", "dim", "else", "encoder_attention_mask", "encoder_hidden_states", "enumerate", "for", "forward", "hidden_states", "idx", "if", "in", "inputs_embeds", "intermediate", "intermediate_hidden_states", "intermediate_logits", "intermediate_reference_points", "inverse_sigmoid", "is", "kwargs", "last_hidden_state", "layer_outputs", "layers", "level_start_index", "logits", "new_reference_points", "not", "output_attentions", "output_hidden_states", "position_embeddings", "predicted_corners", "query_pos_head", "r", "reference_points", "reference_points_input", "return", "return_dict", "self", "sigmoid", "spatial_shapes", "spatial_shapes_list", "stack", "torch", "tuple", "unsqueeze", "use_return_dict", "v", "valid_ratios"], "rt_detr_v2/modeling_rt_detr_v2.py:RTDetrV2FrozenBatchNorm2d.__init__": ["Model", "Module", "__init__", "bias", "class", "def", "n", "nn", "ones", "register_buffer", "running_mean", "running_var", "self", "super", "torch", "weight", "zeros"], "rt_detr_v2/modeling_rt_detr_v2.py:RTDetrV2FrozenBatchNorm2d._load_from_state_dict": ["Model", "Module", "_load_from_state_dict", "class", "def", "del", "error_msgs", "if", "in", "local_metadata", "missing_keys", "nn", "num_batches_tracked", "num_batches_tracked_key", "prefix", "self", "state_dict", "strict", "super", "unexpected_keys"], "rt_detr_v2/modeling_rt_detr_v2.py:RTDetrV2FrozenBatchNorm2d.forward": ["Model", "Module", "bias", "class", "def", "epsilon", "forward", "nn", "reshape", "return", "rsqrt", "running_mean", "running_var", "scale", "self", "weight", "x"], "rt_detr_v2/modeling_rt_detr_v2.py:replace_batch_norm": ["BatchNorm2d", "ModelFrozenBatchNorm2d", "Model_batch_norm", "_modules", "bias", "children", "copy_", "def", "device", "for", "if", "in", "isinstance", "len", "list", "meta", "model", "module", "name", "named_children", "new_module", "nn", "num_features", "r", "running_mean", "running_var", "torch", "weight"], "rt_detr_v2/modeling_rt_detr_v2.py:RTDetrV2ConvEncoder.__init__": ["Model", "Module", "__init__", "backbone", "channels", "class", "config", "def", "freeze_backbone_batch_norms", "if", "intermediate_channel_sizes", "load_backbone", "model", "nn", "no_grad", "replace_batch_norm", "self", "super", "torch", "with"], "rt_detr_v2/modeling_rt_detr_v2.py:RTDetrV2ConvEncoder.forward": ["Model", "Module", "None", "Tensor", "append", "bool", "class", "def", "feature_map", "feature_maps", "features", "float", "for", "forward", "functional", "in", "interpolate", "mask", "model", "nn", "out", "pixel_mask", "pixel_values", "return", "self", "shape", "size", "to", "torch"], "rt_detr_v2/modeling_rt_detr_v2.py:RTDetrV2ConvNormLayer.__init__": ["ACT2CLS", "BatchNorm2d", "Conv2d", "False", "Identity", "Model", "Module", "None", "__init__", "activation", "batch_norm_eps", "bias", "class", "config", "conv", "def", "else", "if", "in_channels", "is", "kernel_size", "nn", "norm", "out_channels", "padding", "self", "stride", "super"], "rt_detr_v2/modeling_rt_detr_v2.py:RTDetrV2ConvNormLayer.forward": ["Model", "Module", "activation", "class", "conv", "def", "forward", "hidden_state", "nn", "norm", "return", "self"], "rt_detr_v2/modeling_rt_detr_v2.py:RTDetrV2EncoderLayer.__init__": ["ACT2FN", "LayerNorm", "Linear", "Model", "ModelConfig", "ModelMultiheadAttention", "Module", "__init__", "activation_dropout", "activation_fn", "class", "config", "def", "dropout", "embed_dim", "encoder_activation_function", "encoder_ffn_dim", "encoder_hidden_dim", "eps", "fc1", "fc2", "final_layer_norm", "layer_norm_eps", "nn", "normalize_before", "num_attention_heads", "num_heads", "self", "self_attn", "self_attn_layer_norm", "super"], "rt_detr_v2/modeling_rt_detr_v2.py:RTDetrV2EncoderLayer.forward": ["False", "Model", "Module", "None", "Tensor", "activation_dropout", "activation_fn", "any", "attention_mask", "attn_weights", "bool", "clamp", "clamp_value", "class", "def", "dropout", "dtype", "fc1", "fc2", "final_layer_norm", "finfo", "forward", "functional", "hidden_states", "if", "isinf", "isnan", "kwargs", "max", "min", "nn", "normalize_before", "not", "or", "output_attentions", "outputs", "p", "position_embeddings", "residual", "return", "self", "self_attn", "self_attn_layer_norm", "torch", "training"], "rt_detr_v2/modeling_rt_detr_v2.py:RTDetrV2RepVggBlock.__init__": ["ACT2CLS", "Identity", "Model", "ModelConfig", "ModelConvNormLayer", "Module", "None", "__init__", "activation", "activation_function", "class", "config", "conv1", "conv2", "def", "else", "encoder_hidden_dim", "hidden_channels", "hidden_expansion", "if", "int", "is", "nn", "padding", "self", "super"], "rt_detr_v2/modeling_rt_detr_v2.py:RTDetrV2RepVggBlock.forward": ["Model", "Module", "activation", "class", "conv1", "conv2", "def", "forward", "nn", "return", "self", "x", "y"], "rt_detr_v2/modeling_rt_detr_v2.py:RTDetrV2CSPRepLayer.__init__": ["Identity", "Model", "ModelConfig", "ModelConvNormLayer", "ModelRepVggBlock", "Module", "Sequential", "_", "__init__", "activation", "activation_function", "bottlenecks", "class", "config", "conv1", "conv2", "conv3", "def", "else", "encoder_hidden_dim", "for", "hidden_channels", "hidden_expansion", "if", "in", "in_channels", "int", "nn", "num_blocks", "out_channels", "range", "self", "super"], "rt_detr_v2/modeling_rt_detr_v2.py:RTDetrV2CSPRepLayer.forward": ["Model", "Module", "bottlenecks", "class", "conv1", "conv2", "conv3", "def", "forward", "hidden_state", "hidden_state_1", "hidden_state_2", "nn", "return", "self"], "rt_detr_v2/modeling_rt_detr_v2.py:RTDetrV2Encoder.__init__": ["Model", "ModelConfig", "ModelLayer", "Module", "ModuleList", "_", "__init__", "class", "config", "def", "encoder_layers", "for", "in", "layers", "nn", "range", "self", "super"], "rt_detr_v2/modeling_rt_detr_v2.py:RTDetrV2Encoder.forward": ["False", "Model", "Module", "None", "Tensor", "attention_mask", "bool", "class", "def", "for", "forward", "hidden_states", "in", "layer", "layers", "nn", "output_attentions", "pos_embed", "position_embeddings", "return", "self", "src", "src_mask", "torch"], "rt_detr_v2/modeling_rt_detr_v2.py:RTDetrV2HybridEncoder.__init__": ["Model", "ModelCSPRepLayer", "ModelConfig", "ModelConvNormLayer", "ModelEncoder", "Module", "ModuleList", "_", "__init__", "activation", "activation_function", "append", "class", "config", "def", "downsample_conv", "downsample_convs", "encode_proj_layers", "encoder", "encoder_hidden_dim", "encoder_in_channels", "eval_size", "feat_strides", "for", "fpn_block", "fpn_blocks", "in", "in_channels", "kernel_size", "lateral_conv", "lateral_convs", "len", "nn", "num_fpn_stages", "num_pan_stages", "out_channels", "out_strides", "pan_block", "pan_blocks", "positional_encoding_temperature", "range", "self", "stride", "super"], "rt_detr_v2/modeling_rt_detr_v2.py:RTDetrV2HybridEncoder.build_2d_sincos_position_embedding": ["Embed", "Model", "Module", "None", "ValueError", "arange", "be", "build_2d_sincos_position_embedding", "by", "class", "concat", "cos", "cpu", "def", "device", "dim", "dimension", "divisible", "dtype", "embed_dim", "embedding", "flatten", "float32", "for", "grid_h", "grid_w", "height", "if", "indexing", "meshgrid", "must", "nn", "omega", "out_h", "out_w", "pos_dim", "position", "raise", "return", "sin", "temperature", "to", "torch", "torch_int", "width", "xy"], "rt_detr_v2/modeling_rt_detr_v2.py:RTDetrV2HybridEncoder.forward": ["BaseModelOutput", "F", "Model", "Module", "None", "all_attentions", "append", "attention_mask", "attentions", "backbone_feature_map", "build_2d_sincos_position_embedding", "class", "concat", "config", "contiguous", "def", "device", "dim", "downsample_conv", "downsample_convs", "downsampled_feature_map", "dtype", "else", "enc_ind", "encode_proj_layers", "encoder", "encoder_hidden_dim", "encoder_layers", "encoder_states", "enumerate", "eval_size", "flatten", "for", "forward", "fpn_block", "fpn_blocks", "fpn_feature_map", "fpn_feature_maps", "fused_feature_map", "height", "hidden_states", "i", "idx", "if", "in", "inputs_embeds", "interpolate", "is", "last_hidden_state", "lateral_conv", "lateral_convs", "layer_outputs", "level_start_index", "mode", "nearest", "new_fpn_feature_map", "new_pan_feature_map", "nn", "not", "num_fpn_stages", "or", "output_attentions", "output_hidden_states", "pan_block", "pan_blocks", "pan_feature_maps", "permute", "pos_embed", "position_embeddings", "positional_encoding_temperature", "r", "reshape", "return", "return_dict", "reverse", "scale_factor", "self", "shape", "spatial_shapes", "src_flatten", "top_fpn_feature_map", "top_pan_feature_map", "torch", "training", "tuple", "use_return_dict", "v", "valid_ratios", "width", "zip"], "rt_detr_v2/modeling_rt_detr_v2.py:get_contrastive_denoising_training_group": ["Model_contrastive_denoising_training_group", "None", "attn_mask", "batch_size", "bool", "box_noise_scale", "boxes", "center_to_corners_format", "class_embed", "class_labels", "clip_", "corners_to_center_format", "def", "denoise_positive_idx", "denoising_meta_values", "device", "diff", "dn_num_group", "dn_num_split", "dn_positive_idx", "dtype", "else", "float", "for", "full", "i", "idx_block_end", "idx_block_start", "if", "in", "inf", "input_query_bbox", "input_query_class", "int32", "inverse_sigmoid", "known_bbox", "label_noise_ratio", "len", "mask", "max", "max_gt_num", "min", "n", "negative_gt_mask", "new_label", "nonzero", "num_classes", "num_denoising_queries", "num_ground_truths", "num_groups_denoising_queries", "num_gt", "num_queries", "pad_gt_mask", "positive_gt_mask", "rand_like", "rand_part", "rand_sign", "randint_like", "range", "return", "split", "squeeze", "t", "tarModel_size", "tarModels", "tile", "torch", "torch_int", "where", "zeros"], "rt_detr_v2/modeling_rt_detr_v2.py:RTDetrV2Model.__init__": ["BatchNorm2d", "Conv2d", "Embedding", "False", "LayerNorm", "Linear", "Model", "ModelConfig", "ModelConvEncoder", "ModelDecoder", "ModelHybridEncoder", "ModelMLPPredictionHead", "ModelPreTrainedModel", "ModuleList", "Sequential", "_", "__init__", "anchor_image_size", "anchors", "append", "backbone", "batch_norm_eps", "bias", "class", "config", "d_model", "decoder", "decoder_in_channels", "decoder_input_proj", "decoder_input_proj_list", "def", "denoising_class_embed", "dtype", "enc_bbox_head", "enc_output", "enc_score_head", "encoder", "encoder_hidden_dim", "encoder_input_proj", "encoder_input_proj_list", "eps", "for", "generate_anchors", "if", "in", "in_channels", "intermediate_channel_sizes", "kernel_size", "layer_norm_eps", "learn_initial_query", "len", "nn", "num_backbone_outs", "num_denoising", "num_feature_levels", "num_labels", "num_layers", "num_queries", "padding", "padding_idx", "post_init", "range", "self", "stride", "super", "valid_mask", "weight_embedding"], "rt_detr_v2/modeling_rt_detr_v2.py:RTDetrV2Model.freeze_backbone": ["False", "Model", "ModelPreTrainedModel", "backbone", "class", "def", "for", "freeze_backbone", "in", "param", "parameters", "requires_grad_", "self"], "rt_detr_v2/modeling_rt_detr_v2.py:RTDetrV2Model.unfreeze_backbone": ["Model", "ModelPreTrainedModel", "True", "backbone", "class", "def", "for", "in", "param", "parameters", "requires_grad_", "self", "unfreeze_backbone"], "rt_detr_v2/modeling_rt_detr_v2.py:RTDetrV2Model.generate_anchors": ["Model", "ModelPreTrainedModel", "None", "True", "all", "anchor_image_size", "anchors", "append", "arange", "class", "concat", "config", "cpu", "def", "device", "dtype", "end", "enumerate", "eps", "feat_strides", "finfo", "float32", "for", "generate_anchors", "grid_size", "grid_x", "grid_xy", "grid_y", "height", "if", "ij", "in", "indexing", "int", "is", "keepdim", "level", "log", "max", "meshgrid", "ones_like", "reshape", "return", "s", "self", "spatial_shapes", "stack", "tensor", "to", "torch", "unsqueeze", "valid_mask", "wh", "where", "width"], "rt_detr_v2/modeling_rt_detr_v2.py:RTDetrV2Model.forward": ["BaseModelOutput", "F", "FloatTensor", "LongTensor", "Model", "ModelOutput", "ModelPreTrainedModel", "None", "_", "_len_sources", "anchor_image_size", "anchors", "and", "append", "attention_mask", "attentions", "backbone", "batch_size", "bool", "box_noise_scale", "cat", "class", "class_embed", "concat", "config", "cross_attentions", "cumsum", "decoder", "decoder_attentions", "decoder_hidden_states", "decoder_input_proj", "decoder_inputs_embeds", "decoder_outputs", "def", "denoising_bbox_unact", "denoising_class", "denoising_class_embed", "denoising_meta_values", "detach", "device", "dict", "dim", "dn_outputs", "dtype", "elif", "else", "empty", "enc_bbox_head", "enc_output", "enc_outputs", "enc_outputs_class", "enc_outputs_coord_logits", "enc_score_head", "enc_topk_bboxes", "enc_topk_logits", "encoder", "encoder_attention_mask", "encoder_attentions", "encoder_hidden_states", "encoder_input_proj", "encoder_last_hidden_state", "encoder_outputs", "enumerate", "features", "flatten", "for", "forward", "gather", "generate_anchors", "get_contrastive_denoising_training_group", "height", "hidden_states", "i", "if", "in", "index", "init_reference_points", "initial_reference_points", "inputs_embeds", "intermediate_hidden_states", "intermediate_logits", "intermediate_predicted_corners", "intermediate_reference_points", "is", "isinstance", "kwargs", "label_noise_ratio", "labels", "last_hidden_state", "learn_initial_query", "len", "level", "level_start_index", "list", "long", "mask", "max", "memory", "new_zeros", "not", "num_channels", "num_classes", "num_denoising", "num_denoising_queries", "num_feature_levels", "num_labels", "num_queries", "ones", "or", "output_attentions", "output_hidden_states", "output_memory", "pixel_mask", "pixel_values", "prod", "proj_feats", "r", "range", "reference_points", "reference_points_unact", "repeat", "return", "return_dict", "self", "shape", "sigmoid", "source", "source_flatten", "sources", "spatial_shapes", "spatial_shapes_list", "spatial_shapes_tuple", "target", "targets", "tile", "to", "topk", "topk_ind", "torch", "training", "transpose", "tuple", "tuple_outputs", "unsqueeze", "use_return_dict", "valid_mask", "value", "values", "weight_embedding", "width"], "rt_detr_v2/modeling_rt_detr_v2.py:RTDetrV2MLPPredictionHead.__init__": ["Linear", "Model", "Module", "ModuleList", "__init__", "class", "config", "d_model", "def", "for", "h", "in", "input_dim", "k", "layers", "n", "nn", "num_layers", "output_dim", "self", "super", "zip"], "rt_detr_v2/modeling_rt_detr_v2.py:RTDetrV2MLPPredictionHead.forward": ["Model", "Module", "class", "def", "else", "enumerate", "for", "forward", "functional", "i", "if", "in", "layer", "layers", "nn", "num_layers", "relu", "return", "self", "x"], "rt_detr_v2/modeling_rt_detr_v2.py:RTDetrV2ForObjectDetection.__init__": ["Linear", "Model", "ModelConfig", "ModelMLPPredictionHead", "ModelModel", "ModelPreTrainedModel", "ModuleList", "_", "__init__", "bbox_embed", "class", "class_embed", "config", "d_model", "decoder", "decoder_layers", "def", "for", "in", "model", "nn", "num_labels", "num_layers", "post_init", "range", "self", "super", "torch"], "rt_detr_v2/modeling_rt_detr_v2.py:RTDetrV2ForObjectDetection._set_aux_loss": ["Model", "ModelPreTrainedModel", "_set_aux_loss", "a", "b", "class", "def", "for", "in", "logits", "outputs_class", "outputs_coord", "pred_boxes", "return", "self", "zip"], "rt_detr_v2/modeling_rt_detr_v2.py:RTDetrV2ForObjectDetection.forward": ["FloatTensor", "LongTensor", "Model", "ModelObjectDetectionOutput", "ModelPreTrainedModel", "None", "auxiliary_outputs", "bool", "class", "config", "cross_attentions", "decoder_attentions", "decoder_hidden_states", "decoder_inputs_embeds", "def", "denoising_meta_values", "device", "dict", "else", "enc_outputs_class", "enc_outputs_coord_logits", "enc_topk_bboxes", "enc_topk_logits", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "forward", "if", "init_reference_points", "initial_reference_points", "inputs_embeds", "intermediate_hidden_states", "intermediate_logits", "intermediate_predicted_corners", "intermediate_reference_points", "is", "kwargs", "labels", "last_hidden_state", "list", "logits", "loss", "loss_dict", "loss_function", "model", "not", "output", "output_attentions", "output_hidden_states", "outputs", "outputs_class", "outputs_coord", "pixel_mask", "pixel_values", "pred_boxes", "predicted_corners", "r", "return", "return_dict", "self", "torch", "training", "tuple", "use_return_dict"], "rwkv/modeling_rwkv.py:load_wkv_cuda_kernel": ["ImportError", "Model", "Model_cuda_kernel", "Model_wkv_cuda_kernel", "community", "context_length", "def", "get_kernel", "global", "if", "install", "installed", "is", "is_kernels_available", "it", "kernels", "max_seq_length", "not", "pip", "please", "raise", "with"], "rwkv/modeling_rwkv.py:RwkvLinearAttention.forward": ["CUDA", "Calling", "Cannot", "False", "Function", "ModelLinearAttention", "Model_cuda_kernel", "None", "The", "ValueError", "a", "all", "and", "at", "attention", "autograd", "batch", "batch_size", "be", "bfloat16", "cat", "chunk", "class", "contiguous", "contiguous_format", "ctx", "cuda", "def", "device", "devices", "dim", "dtype", "else", "empty_like", "exp", "f", "float", "float16", "float32", "for", "forward", "forward_bf16", "forward_func", "forward_with_state", "forward_with_state_bf16", "hidden", "hidden_size", "if", "in", "input_dtype", "is", "kernel", "key", "max_seq_length", "maximum", "memory_format", "min", "model", "multiple", "needs", "not", "of", "on", "or", "output", "process", "product", "raise", "requires", "return", "return_state", "round", "s", "same", "save_for_backward", "seq_len", "size", "squeeze", "state", "tensors", "the", "this", "time", "time_decay", "time_first", "to", "tokens", "torch", "type", "unsqueeze", "use", "value", "with", "wkv", "zeros"], "rwkv/modeling_rwkv.py:RwkvLinearAttention.backward": ["Function", "ModelLinearAttention", "Model_cuda_kernel", "None", "autograd", "backward", "backward_bf16", "backward_func", "bfloat16", "class", "contiguous", "contiguous_format", "ctx", "def", "dtype", "else", "empty_like", "float", "float16", "float32", "g_key", "g_output", "g_state", "g_time_decay", "g_time_first", "g_value", "if", "input_dtype", "key", "memory_format", "output", "return", "saved_tensors", "time_decay", "time_first", "to", "torch", "value"], "rwkv/modeling_rwkv.py:rwkv_linear_attention_cpu": ["False", "Model_linear_attention_cpu", "None", "_", "current_index", "current_key", "current_value", "def", "den_state", "denominator", "dtype", "e1", "e2", "else", "exp", "float", "float32", "for", "if", "in", "is", "key", "max_for_output", "max_for_state", "max_state", "maximum", "not", "num_state", "numerator", "or", "output", "range", "return", "return_state", "seq_length", "size", "state", "time_decay", "time_first", "to", "torch", "value", "zeros_like"], "rwkv/modeling_rwkv.py:rwkv_linear_attention": ["False", "ModelLinearAttention", "Model_cuda_kernel", "Model_linear_attention", "Model_linear_attention_cpu", "None", "any", "apply", "cuda", "def", "device", "else", "for", "if", "in", "is", "key", "no_cuda", "one_token", "or", "return", "return_state", "size", "state", "t", "time_decay", "time_first", "type", "value"], "rwkv/modeling_rwkv.py:RwkvSelfAttention.__init__": ["CUDA", "Could", "Exception", "False", "Linear", "Model", "ModelSelfAttention", "Model_cuda_kernel", "Module", "None", "Parameter", "ZeroPad2d", "__init__", "and", "attention", "attention_hidden_size", "bias", "class", "config", "context_length", "custom", "def", "else", "empty", "except", "for", "hidden_size", "if", "info", "is", "is_ninja_available", "is_torch_cuda_available", "kernel", "kernel_loaded", "key", "layer_id", "load", "load_wkv_cuda_kernel", "logger", "max_seq_length", "nn", "not", "output", "receptance", "self", "super", "the", "time_decay", "time_first", "time_mix_key", "time_mix_receptance", "time_mix_value", "time_shift", "torch", "try", "value"], "rwkv/modeling_rwkv.py:RwkvSelfAttention.extract_key_value": ["ModelSelfAttention", "Module", "None", "and", "class", "def", "else", "extract_key_value", "hidden", "if", "is", "key", "layer_id", "nn", "not", "receptance", "return", "self", "shifted", "sigmoid", "size", "state", "time_mix_key", "time_mix_receptance", "time_mix_value", "time_shift", "torch", "value"], "rwkv/modeling_rwkv.py:RwkvSelfAttention.forward": ["False", "Model", "ModelSelfAttention", "Model_linear_attention", "Module", "None", "class", "def", "else", "extract_key_value", "for", "forward", "hidden", "if", "in", "is", "key", "layer_id", "layer_state", "nn", "not", "output", "receptance", "return", "return_state", "s", "self", "state", "time_decay", "time_first", "tuple", "use_cache", "value"], "rwkv/modeling_rwkv.py:RwkvFeedForward.__init__": ["False", "Linear", "ModelFeedForward", "Module", "None", "Parameter", "ZeroPad2d", "__init__", "bias", "class", "config", "def", "else", "empty", "hidden_size", "if", "intermediate_size", "is", "key", "layer_id", "nn", "not", "receptance", "self", "super", "time_mix_key", "time_mix_receptance", "time_shift", "torch", "value"], "rwkv/modeling_rwkv.py:RwkvFeedForward.forward": ["ModelFeedForward", "Module", "None", "and", "class", "def", "else", "forward", "hidden", "if", "is", "key", "layer_id", "nn", "not", "receptance", "relu", "return", "self", "shifted", "sigmoid", "size", "square", "state", "time_mix_key", "time_mix_receptance", "time_shift", "torch", "value"], "rwkv/modeling_rwkv.py:RwkvBlock.__init__": ["GradientCheckpointingLayer", "LayerNorm", "ModelBlock", "ModelFeedForward", "ModelSelfAttention", "__init__", "attention", "class", "config", "def", "eps", "feed_forward", "hidden_size", "if", "layer_id", "layer_norm_epsilon", "ln1", "ln2", "nn", "pre_ln", "self", "super"], "rwkv/modeling_rwkv.py:RwkvBlock.forward": ["False", "GradientCheckpointingLayer", "ModelBlock", "None", "attention", "class", "def", "else", "feed_forward", "forward", "hidden", "if", "layer_id", "ln1", "ln2", "output_attentions", "outputs", "pre_ln", "return", "self", "state", "use_cache"], "rwkv/modeling_rwkv.py:RwkvPreTrainedModel._init_weights": ["Embedding", "LayerNorm", "Linear", "ModelFeedForward", "ModelPreTrainedModel", "ModelSelfAttention", "Module", "None", "PreTrainedModel", "_init_weights", "and", "attention_hidden_size", "bias", "class", "config", "copy_", "decay_speed", "def", "device", "dtype", "elif", "for", "gain", "h", "hidden_size", "i", "if", "in", "init", "is", "isinstance", "layer_id", "log", "math", "max", "module", "nn", "not", "num_hidden_layers", "ones_", "ones_like", "orthogonal_", "pow", "range", "ratio_0_to_1", "ratio_1_to_almost0", "scale", "self", "shape", "sqrt", "tensor", "time_decay", "time_first", "time_mix_key", "time_mix_receptance", "time_mix_value", "time_weight", "torch", "vocab_size", "weight", "zeros_", "zigzag"], "rwkv/modeling_rwkv.py:RwkvModel.__init__": ["Embedding", "False", "LayerNorm", "ModelBlock", "ModelModel", "ModelPreTrainedModel", "ModuleList", "__init__", "blocks", "class", "config", "def", "embeddings", "for", "gradient_checkpointing", "hidden_size", "idx", "in", "layer_id", "layers_are_rescaled", "ln_out", "nn", "num_hidden_layers", "post_init", "range", "self", "super", "vocab_size"], "rwkv/modeling_rwkv.py:RwkvModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "return", "self"], "rwkv/modeling_rwkv.py:RwkvModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "new_embeddings", "self", "set_input_embeddings"], "rwkv/modeling_rwkv.py:RwkvModel.forward": ["False", "FloatTensor", "LongTensor", "ModelModel", "ModelOutput", "ModelPreTrainedModel", "None", "Setting", "True", "ValueError", "You", "_rescale_layers", "all_hidden_states", "all_self_attentions", "and", "at", "attention_mask", "attentions", "block", "blocks", "bool", "both", "but", "cannot", "checkpointing", "class", "config", "def", "device", "dtype", "either", "elif", "else", "embeddings", "enumerate", "float32", "for", "forward", "gradient", "gradient_checkpointing", "have", "hidden_size", "hidden_states", "i", "idx", "if", "in", "incompatible", "input_ids", "inputs_embeds", "is", "it", "kwargs", "last_hidden_state", "layers_are_rescaled", "list", "ln_out", "logger", "model", "not", "num_hidden_layers", "or", "output_attentions", "output_hidden_states", "passed", "r", "raise", "range", "rescale_every", "return", "return_dict", "same", "self", "shape", "size", "specify", "state", "the", "this", "time", "to", "torch", "training", "tuple", "unused", "use_cache", "use_return_dict", "warning_once", "was", "with", "x", "zeros"], "rwkv/modeling_rwkv.py:RwkvModel._rescale_layers": ["ModelModel", "ModelPreTrainedModel", "SCB", "_bnb_4bit_dequantize_and_rescale", "_rescale_layers", "attention", "block", "block_id", "blocks", "class", "config", "def", "div_", "elif", "else", "enumerate", "feed_forward", "for", "hasattr", "if", "in", "int", "layers_are_rescaled", "mul_", "no_grad", "not", "output", "quant_state", "rescale_every", "return", "self", "torch", "training", "value", "weight", "with"], "rwkv/modeling_rwkv.py:RwkvModel._bnb_4bit_dequantize_and_rescale": ["False", "ImportError", "ModelModel", "ModelPreTrainedModel", "Params4bit", "Please", "_bnb_4bit_dequantize_and_rescale", "bitsandbytes", "block_id", "bnb", "class", "config", "cpu", "data", "def", "dequant_weights", "dequantize_4bit", "device", "div_", "functional", "if", "install", "int", "is_bitsandbytes_available", "method", "nn", "not", "quant_state", "quant_weight", "r", "raise", "requires_grad", "rescale_every", "self", "setattr", "target_layer", "this", "to", "use", "weight"], "rwkv/modeling_rwkv.py:RwkvForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "Model", "ModelForCausalLM", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "head", "hidden_size", "nn", "post_init", "self", "super", "vocab_size"], "rwkv/modeling_rwkv.py:RwkvForCausalLM.get_output_embeddings": ["GenerationMixin", "ModelForCausalLM", "ModelPreTrainedModel", "class", "def", "get_output_embeddings", "head", "return", "self"], "rwkv/modeling_rwkv.py:RwkvForCausalLM.set_output_embeddings": ["GenerationMixin", "ModelForCausalLM", "ModelPreTrainedModel", "class", "def", "head", "new_embeddings", "self", "set_output_embeddings"], "rwkv/modeling_rwkv.py:RwkvForCausalLM.prepare_inputs_for_generation": ["GenerationMixin", "ModelForCausalLM", "ModelPreTrainedModel", "None", "and", "class", "def", "else", "for", "if", "in", "input_ids", "inputs_embeds", "is", "items", "key", "kwargs", "model_inputs", "not", "prepare_inputs_for_generation", "return", "self", "state", "unsqueeze", "use_cache", "value"], "rwkv/modeling_rwkv.py:RwkvForCausalLM.forward": ["FloatTensor", "GenerationMixin", "LongTensor", "Model", "ModelCausalLMOutput", "ModelForCausalLM", "ModelPreTrainedModel", "Model_outputs", "None", "Tensor", "attention_mask", "attentions", "bool", "class", "config", "def", "else", "forward", "head", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "list", "logits", "logits_to_keep", "loss", "loss_function", "not", "output", "output_attentions", "output_hidden_states", "r", "return", "return_dict", "self", "slice", "slice_indices", "state", "torch", "tuple", "use_cache", "use_return_dict", "vocab_size"], "sam/modeling_sam.py:SamPatchEmbeddings.__init__": ["Conv2d", "Iterable", "ModelPatchEmbeddings", "Module", "__init__", "abc", "class", "collections", "config", "def", "else", "hidden_size", "if", "image_size", "isinstance", "kernel_size", "nn", "num_channels", "num_patches", "patch_size", "projection", "self", "stride", "super"], "sam/modeling_sam.py:SamPatchEmbeddings.forward": ["Input", "Make", "ModelPatchEmbeddings", "Module", "ValueError", "batch_size", "channel", "class", "configuration", "def", "dimension", "doesn", "embeddings", "f", "forward", "height", "if", "image", "image_size", "in", "match", "model", "nn", "num_channels", "of", "one", "or", "permute", "pixel", "pixel_values", "projection", "raise", "return", "self", "set", "shape", "size", "sure", "t", "that", "the", "values", "width", "with"], "sam/modeling_sam.py:SamMLPBlock.__init__": ["ACT2FN", "Linear", "ModelMLPBlock", "Module", "__init__", "act", "class", "config", "def", "hidden_act", "hidden_size", "lin1", "lin2", "mlp_dim", "nn", "self", "super"], "sam/modeling_sam.py:SamMLPBlock.forward": ["ModelMLPBlock", "Module", "Tensor", "act", "class", "def", "forward", "hidden_states", "lin1", "lin2", "nn", "return", "self", "torch"], "sam/modeling_sam.py:SamLayerNorm.__init__": ["LayerNorm", "ModelLayerNorm", "NotImplementedError", "Unsupported", "__init__", "channels_first", "channels_last", "class", "data", "data_format", "def", "eps", "f", "format", "if", "in", "kwargs", "nn", "normalized_shape", "not", "raise", "self", "super"], "sam/modeling_sam.py:SamLayerNorm.forward": ["LayerNorm", "ModelLayerNorm", "Tensor", "channels_first", "class", "data_format", "def", "else", "features", "forward", "if", "nn", "permute", "return", "self", "super", "torch"], "sam/modeling_sam.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "attention_mask", "attn_output", "attn_weights", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "softmax", "to", "torch", "training", "transpose", "value"], "sam/modeling_sam.py:SamAttention.__init__": ["False", "Linear", "ModelAttention", "Module", "None", "ValueError", "__init__", "attention_downModelple_rate", "class", "config", "def", "divide", "downModelple_rate", "else", "hidden_size", "if", "internal_dim", "is", "is_causal", "k_proj", "must", "nn", "num_attention_heads", "out_proj", "q_proj", "raise", "scaling", "self", "super", "v_proj"], "sam/modeling_sam.py:SamAttention._separate_heads": ["ModelAttention", "Module", "Tensor", "_separate_heads", "batch", "c_per_head", "channel", "class", "def", "hidden_states", "int", "n_tokens", "nn", "num_attention_heads", "point_batch_size", "reshape", "return", "self", "shape", "transpose"], "sam/modeling_sam.py:SamAttention._recombine_heads": ["ModelAttention", "Module", "Tensor", "_recombine_heads", "batch", "c_per_head", "class", "def", "hidden_states", "int", "n_heads", "n_tokens", "nn", "point_batch_size", "reshape", "return", "self", "shape"], "sam/modeling_sam.py:SamAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "ModelAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_attn_implementation", "_recombine_heads", "_separate_heads", "attention_interface", "attention_mask", "attention_similarity", "attn_output", "attn_weights", "class", "config", "def", "dropout", "eager", "eager_attention_forward", "forward", "if", "is_causal", "k_proj", "key", "kwargs", "nn", "num_attention_heads", "out_proj", "point_batch_size", "q_proj", "query", "return", "scaling", "self", "shape", "v_proj", "value"], "sam/modeling_sam.py:SamTwoWayAttentionBlock.__init__": ["False", "LayerNorm", "ModelAttention", "ModelMLPBlock", "ModelTwoWayAttentionBlock", "Module", "__init__", "attention_downModelple_rate", "bool", "class", "config", "cross_attn_image_to_token", "cross_attn_token_to_image", "def", "downModelple_rate", "eps", "hidden_size", "int", "layer_norm1", "layer_norm2", "layer_norm3", "layer_norm4", "layer_norm_eps", "mlp", "nn", "self", "self_attn", "skip_first_layer_pe", "super"], "sam/modeling_sam.py:SamTwoWayAttentionBlock.forward": ["ModelTwoWayAttentionBlock", "Module", "Tensor", "TransformersKwargs", "Unpack", "_", "attention_similarity", "attn_out", "class", "cross_attn_image_to_token", "cross_attn_token_to_image", "def", "else", "forward", "if", "key", "key_point_embedding", "keys", "kwargs", "layer_norm1", "layer_norm2", "layer_norm3", "layer_norm4", "mlp", "mlp_out", "nn", "queries", "query", "query_point_embedding", "return", "self", "self_attn", "skip_first_layer_pe", "value"], "sam/modeling_sam.py:SamTwoWayTransformer.__init__": ["LayerNorm", "ModelAttention", "ModelMaskDecoderConfig", "ModelTwoWayAttentionBlock", "ModelTwoWayTransformer", "Module", "ModuleList", "__init__", "append", "class", "config", "def", "final_attn_token_to_image", "for", "hidden_size", "i", "in", "layer_norm_final_attn", "layers", "nn", "num_hidden_layers", "range", "self", "skip_first_layer_pe", "super"], "sam/modeling_sam.py:SamTwoWayTransformer.forward": ["BaseModelOutput", "ModelTwoWayTransformer", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "_", "an", "attention_similarity", "attn_out", "class", "def", "final_attn_token_to_image", "flatten", "for", "forward", "have", "if", "image_embedding", "image_embeddings", "image_positional_embeddings", "in", "is", "key", "key_point_embedding", "keys", "kwargs", "layer", "layer_norm_final_attn", "layers", "nn", "not", "permute", "point_embeddings", "queries", "query", "query_point_embedding", "raise", "return", "self", "specify", "target_embedding", "to", "tuple", "unsqueeze", "value"], "sam/modeling_sam.py:SamFeedForward.__init__": ["False", "Linear", "ModelFeedForward", "Module", "ModuleList", "ReLU", "_", "__init__", "activation", "bool", "class", "def", "for", "hidden_dim", "in", "input_dim", "int", "layers", "nn", "num_layers", "output_dim", "proj_in", "proj_out", "range", "self", "sigmoid_output", "super"], "sam/modeling_sam.py:SamFeedForward.forward": ["F", "ModelFeedForward", "Module", "activation", "class", "def", "for", "forward", "hidden_states", "if", "in", "layer", "layers", "nn", "proj_in", "proj_out", "return", "self", "sigmoid", "sigmoid_output"], "sam/modeling_sam.py:SamMaskDecoder.__init__": ["ConvTranspose2d", "Embedding", "GELU", "ModelFeedForward", "ModelLayerNorm", "ModelMaskDecoder", "ModelMaskDecoderConfig", "ModelTwoWayTransformer", "Module", "ModuleList", "_", "__init__", "activation", "channels_first", "class", "config", "data_format", "def", "for", "hidden_size", "in", "iou_head_depth", "iou_head_hidden_dim", "iou_prediction_head", "iou_token", "kernel_size", "mask_tokens", "mlps_list", "nn", "num_mask_tokens", "num_multimask_outputs", "output_hypernetworks_mlps", "range", "self", "stride", "super", "transformer", "upscale_conv1", "upscale_conv2", "upscale_layer_norm"], "sam/modeling_sam.py:SamMaskDecoder.forward": ["ModelMaskDecoder", "Module", "None", "Tensor", "_", "activation", "attention_similarity", "batch_size", "bool", "cat", "class", "current_mlp", "def", "dense_prompt_embeddings", "dim", "dtype", "else", "for", "forward", "height", "hyper_in", "hyper_in_list", "i", "if", "image_embeddings", "image_positional_embeddings", "in", "iou_pred", "iou_prediction_head", "iou_token", "iou_token_out", "is", "mask_slice", "mask_tokens", "mask_tokens_out", "masks", "multimask_output", "nn", "not", "num_channels", "num_mask_tokens", "output_hypernetworks_mlps", "output_tokens", "point_batch_size", "point_embedding", "point_embeddings", "range", "repeat", "repeat_interleave", "reshape", "return", "self", "shape", "slice", "sparse_prompt_embeddings", "stack", "target_embedding", "to", "tokens", "torch", "transformer", "transpose", "tuple", "upscale_conv1", "upscale_conv2", "upscale_layer_norm", "upscaled_embedding", "weight", "width"], "sam/modeling_sam.py:SamPositionalEmbedding.__init__": ["ModelPositionalEmbedding", "Module", "__init__", "class", "config", "def", "nn", "num_pos_feats", "positional_embedding", "randn", "register_buffer", "scale", "self", "super", "torch"], "sam/modeling_sam.py:SamPositionalEmbedding.forward": ["ModelPositionalEmbedding", "Module", "None", "cat", "class", "clone", "coordinates", "cos", "def", "dim", "dtype", "forward", "if", "input_coords", "input_shape", "is", "nn", "not", "np", "pi", "positional_embedding", "return", "self", "sin", "to", "torch"], "sam/modeling_sam.py:SamMaskEmbedding.__init__": ["ACT2FN", "Conv2d", "ModelLayerNorm", "ModelMaskEmbedding", "ModelPromptEncoderConfig", "Module", "__init__", "activation", "channels_first", "class", "config", "conv1", "conv2", "conv3", "data_format", "def", "eps", "hidden_act", "hidden_size", "kernel_size", "layer_norm1", "layer_norm2", "layer_norm_eps", "mask_input_channels", "nn", "self", "stride", "super"], "sam/modeling_sam.py:SamMaskEmbedding.forward": ["ModelMaskEmbedding", "Module", "activation", "class", "conv1", "conv2", "conv3", "def", "dense_embeddings", "forward", "hidden_states", "layer_norm1", "layer_norm2", "masks", "nn", "return", "self"], "sam/modeling_sam.py:SamPromptEncoder.__init__": ["Embedding", "ModelConfig", "ModelMaskEmbedding", "ModelPositionalEmbedding", "ModelPromptEncoder", "Module", "ModuleList", "__init__", "class", "config", "def", "for", "hidden_size", "i", "image_embedding_size", "image_size", "in", "input_image_size", "mask_embed", "nn", "no_mask_embed", "not_a_point_embed", "num_point_embeddings", "point_embed", "prompt_encoder_config", "range", "self", "shared_embedding", "super", "vision_config"], "sam/modeling_sam.py:SamPromptEncoder._embed_points": ["ModelPromptEncoder", "Module", "None", "Tensor", "_embed_points", "bool", "cat", "class", "def", "device", "dim", "if", "input_image_size", "input_shape", "labels", "nn", "not_a_point_embed", "ones", "pad", "padding_label", "padding_point", "point_embed", "point_embedding", "points", "return", "self", "shape", "shared_embedding", "target_labels_shape", "target_point_shape", "torch", "weight", "where", "zeros", "zeros_like"], "sam/modeling_sam.py:SamPromptEncoder._embed_boxes": ["ModelPromptEncoder", "Module", "Tensor", "_embed_boxes", "batch_size", "boxes", "class", "coords", "corner_embedding", "def", "input_image_size", "input_shape", "nb_boxes", "nn", "point_embed", "reshape", "return", "self", "shape", "shared_embedding", "torch", "weight"], "sam/modeling_sam.py:SamPromptEncoder.forward": ["If", "ModelPromptEncoder", "Module", "None", "Tensor", "ValueError", "_embed_boxes", "_embed_points", "also", "are", "batch_size", "be", "box_embeddings", "cat", "class", "def", "dense_embeddings", "dim", "else", "expand", "forward", "if", "image_embedding_size", "input_boxes", "input_labels", "input_masks", "input_points", "is", "labels", "mask_embed", "must", "nn", "no_mask_embed", "not", "pad", "point_embeddings", "points", "provided", "raise", "reshape", "return", "self", "shape", "sparse_embeddings", "torch", "tuple", "weight"], "sam/modeling_sam.py:SamVisionAttention.__init__": ["Input", "Linear", "ModelVisionAttention", "Module", "None", "Parameter", "ValueError", "__init__", "attention_dropout", "be", "bias", "class", "config", "def", "dropout", "else", "encoding", "head_dim", "hidden_size", "if", "image_size", "input_size", "is", "must", "nn", "num_attention_heads", "patch_size", "positional", "proj", "provided", "qkv", "qkv_bias", "raise", "rel_pos_h", "rel_pos_w", "relative", "scale", "self", "size", "super", "torch", "use_rel_pos", "using", "window_size", "zeros"], "sam/modeling_sam.py:SamVisionAttention.get_rel_pos": ["F", "ModelVisionAttention", "Module", "None", "Tensor", "arange", "class", "def", "get_rel_pos", "int", "interpolate", "k_coords", "k_size", "linear", "long", "max", "max_rel_dist", "mode", "nn", "permute", "q_coords", "q_size", "rel_pos", "rel_pos_resized", "relative_coords", "reshape", "return", "self", "shape", "size", "torch"], "sam/modeling_sam.py:SamVisionAttention.get_decomposed_rel_pos": ["ModelVisionAttention", "Module", "None", "Tensor", "_", "batch_size", "bhwc", "bhwk", "class", "decomposed_rel_pos", "def", "dim", "einsum", "get_decomposed_rel_pos", "get_rel_pos", "hkc", "int", "k_size", "key_height", "key_width", "nn", "q_size", "query", "query_height", "query_width", "rel_h", "rel_pos_h", "rel_pos_w", "rel_w", "relative_position_height", "relative_position_width", "reshape", "reshaped_query", "return", "self", "shape", "torch", "tuple", "wkc"], "sam/modeling_sam.py:SamVisionAttention.forward": ["ModelVisionAttention", "Module", "None", "Tensor", "_", "attn_output", "attn_probs", "attn_weights", "batch_size", "class", "decomposed_rel_pos", "def", "dim", "dropout", "dtype", "float32", "forward", "functional", "get_decomposed_rel_pos", "height", "hidden_states", "if", "key", "nn", "num_attention_heads", "output_attentions", "p", "permute", "proj", "qkv", "query", "rel_pos_h", "rel_pos_w", "reshape", "reshape_as", "return", "scale", "self", "shape", "softmax", "to", "torch", "training", "transpose", "tuple", "unbind", "use_rel_pos", "value", "width"], "sam/modeling_sam.py:SamVisionSdpaAttention.__init__": ["ModelVisionAttention", "ModelVisionSdpaAttention", "__init__", "class", "config", "def", "self", "super", "window_size"], "sam/modeling_sam.py:SamVisionSdpaAttention.forward": ["False", "If", "ModelVisionAttention", "ModelVisionSdpaAttention", "None", "Tensor", "The", "True", "_", "__class__", "__name__", "attention", "attn_bias", "attn_implementation", "attn_mask", "attn_output", "batch_size", "be", "class", "decomposed_rel_pos", "def", "does", "eager", "f", "forward", "functional", "get", "get_decomposed_rel_pos", "height", "hidden_states", "if", "key", "loading", "logger", "model", "nn", "not", "num_attention_heads", "output_attentions", "permute", "please", "proj", "qkv", "query", "rel_pos_h", "rel_pos_w", "reshape", "return", "returned", "scaled_dot_product_attention", "self", "set", "shape", "support", "the", "to", "torch", "unbind", "use_rel_pos", "value", "view", "want", "warning_once", "weights", "when", "width", "will", "you"], "sam/modeling_sam.py:SamVisionLayer.__init__": ["GradientCheckpointingLayer", "LayerNorm", "ModelMLPBlock", "ModelVisionLayer", "Model_VISION_ATTENTION_CLASSES", "__init__", "_attn_implementation", "attn", "class", "config", "def", "eps", "hidden_size", "layer_norm1", "layer_norm2", "layer_norm_eps", "mlp", "nn", "self", "super", "window_size"], "sam/modeling_sam.py:SamVisionLayer.window_partition": ["F", "GradientCheckpointingLayer", "ModelVisionLayer", "Tensor", "batch_size", "channel", "class", "contiguous", "def", "height", "hidden_states", "int", "pad", "pad_h", "pad_height", "pad_w", "pad_width", "permute", "reshape", "return", "self", "shape", "torch", "tuple", "width", "window_partition", "window_size", "windows"], "sam/modeling_sam.py:SamVisionLayer.window_unpartition": ["GradientCheckpointingLayer", "ModelVisionLayer", "Tensor", "batch_size", "class", "contiguous", "def", "height", "hidden_states", "int", "original_shape", "pad_height", "pad_width", "padding_shape", "permute", "reshape", "return", "self", "shape", "torch", "tuple", "width", "window_size", "window_unpartition", "windows"], "sam/modeling_sam.py:SamVisionLayer.forward": ["FloatTensor", "GradientCheckpointingLayer", "ModelVisionLayer", "Tensor", "attn", "attn_weights", "class", "def", "forward", "height", "hidden_states", "if", "layer_norm1", "layer_norm2", "layernorm_output", "mlp", "padding_shape", "residual", "return", "self", "shape", "torch", "tuple", "width", "window_partition", "window_size", "window_unpartition"], "sam/modeling_sam.py:SamVisionNeck.__init__": ["Conv2d", "False", "ModelLayerNorm", "ModelVisionConfig", "ModelVisionNeck", "Module", "__init__", "bias", "channels_first", "class", "config", "conv1", "conv2", "data_format", "def", "hidden_size", "kernel_size", "layer_norm1", "layer_norm2", "nn", "output_channels", "padding", "self", "super"], "sam/modeling_sam.py:SamVisionNeck.forward": ["ModelVisionNeck", "Module", "class", "conv1", "conv2", "def", "forward", "hidden_states", "layer_norm1", "layer_norm2", "nn", "permute", "return", "self"], "sam/modeling_sam.py:SamPreTrainedModel._init_weights": ["ModelPositionalEmbedding", "ModelPreTrainedModel", "ModelVisionAttention", "ModelVisionEncoder", "Module", "PreTrainedModel", "_init_weights", "class", "config", "def", "elif", "if", "init", "isinstance", "module", "nn", "normal_", "pos_embed", "positional_embedding", "rel_pos_h", "rel_pos_w", "scale", "self", "std", "super", "use_abs_pos", "use_rel_pos", "zeros_"], "sam/modeling_sam.py:SamVisionEncoder.__init__": ["False", "ModelPatchEmbeddings", "ModelPreTrainedModel", "ModelVisionConfig", "ModelVisionEncoder", "ModelVisionLayer", "ModelVisionNeck", "ModuleList", "None", "Parameter", "__init__", "append", "class", "config", "def", "else", "for", "global_attn_indexes", "gradient_checkpointing", "hidden_size", "i", "if", "image_size", "in", "layer", "layers", "neck", "nn", "not", "num_hidden_layers", "patch_embed", "patch_size", "pos_embed", "post_init", "range", "self", "super", "torch", "use_abs_pos", "window_size", "zeros"], "sam/modeling_sam.py:SamVisionEncoder.get_input_embeddings": ["ModelPreTrainedModel", "ModelVisionEncoder", "class", "def", "get_input_embeddings", "patch_embed", "return", "self"], "sam/modeling_sam.py:SamVisionEncoder.forward": ["FloatTensor", "ModelPreTrainedModel", "ModelVisionEncoder", "ModelVisionEncoderOutput", "None", "TransformersKwargs", "Unpack", "ValueError", "You", "class", "def", "for", "forward", "have", "hidden_states", "if", "in", "is", "kwargs", "last_hidden_state", "layer_module", "layers", "neck", "not", "patch_embed", "pixel_values", "pos_embed", "raise", "return", "self", "specify", "to", "torch"], "sam/modeling_sam.py:SamVisionModel.__init__": ["ModelPreTrainedModel", "ModelVisionConfig", "ModelVisionEncoder", "ModelVisionModel", "__init__", "class", "config", "def", "post_init", "self", "super", "vision_encoder"], "sam/modeling_sam.py:SamVisionModel.get_input_embeddings": ["ModelPreTrainedModel", "ModelVisionModel", "Module", "class", "def", "get_input_embeddings", "nn", "patch_embed", "return", "self", "vision_encoder"], "sam/modeling_sam.py:SamVisionModel.forward": ["FloatTensor", "ModelPreTrainedModel", "ModelVisionEncoderOutput", "ModelVisionModel", "None", "TransformersKwargs", "Unpack", "class", "def", "forward", "kwargs", "pixel_values", "return", "self", "torch", "tuple", "vision_encoder"], "sam/modeling_sam.py:SamModel.__init__": ["ModelConfig", "ModelMaskDecoder", "ModelModel", "ModelPositionalEmbedding", "ModelPreTrainedModel", "ModelPromptEncoder", "ModelVisionEncoder", "__init__", "_attn_implementation", "class", "config", "def", "mask_decoder", "mask_decoder_config", "post_init", "prompt_encoder", "self", "shared_image_embedding", "super", "vision_config", "vision_encoder"], "sam/modeling_sam.py:SamModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "return", "self", "vision_encoder"], "sam/modeling_sam.py:SamModel.get_image_wide_positional_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "config", "cumsum", "def", "device", "dim", "dtype", "get_image_wide_positional_embeddings", "grid", "image_embedding_size", "ones", "permute", "positional_embedding", "prompt_encoder_config", "return", "self", "shared_image_embedding", "size", "stack", "target_device", "target_dtype", "torch", "unsqueeze", "x_embed", "y_embed"], "sam/modeling_sam.py:SamModel.get_image_embeddings": ["ModelModel", "ModelPreTrainedModel", "TransformersKwargs", "Unpack", "class", "def", "get_image_embeddings", "image_embeddings", "kwargs", "pixel_values", "r", "return", "self", "vision_encoder", "vision_output"], "sam/modeling_sam.py:SamModel.get_prompt_embeddings": ["FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "class", "def", "get_prompt_embeddings", "input_boxes", "input_labels", "input_masks", "input_points", "prompt_encoder", "prompt_output", "r", "return", "self", "torch"], "sam/modeling_sam.py:SamModel.forward": ["Either", "FloatTensor", "Got", "LongTensor", "ModelImageSegmentationOutput", "ModelModel", "ModelPreTrainedModel", "ModelVisionEncoderOutput", "Modele", "None", "Of", "Only", "The", "TransformersKwargs", "True", "Unpack", "ValueError", "You", "a", "and", "as", "attention_similarity", "attentions", "batch", "batch_size", "be", "bool", "bounding", "box", "box_batch_size", "boxes", "can", "class", "def", "dense_embeddings", "dense_prompt_embeddings", "device", "dtype", "else", "embeddings", "f", "for", "forward", "get_image_wide_positional_embeddings", "got", "hidden_states", "if", "image", "image_embeddings", "image_positional_embeddings", "input", "input_boxes", "input_labels", "input_masks", "input_points", "int", "iou_predictions", "iou_scores", "is", "kwargs", "last_hidden_state", "len", "low_res_masks", "make", "many", "mask_decoder", "multimask_output", "multiple", "must", "nb_boxes", "nb_points_per_image", "not", "num_points_per_image", "of", "one", "ones_like", "or", "pass", "passed", "per", "pixel_values", "point_batch_size", "points", "pred_masks", "prompt_encoder", "provide", "provided", "r", "raise", "repeat", "respectively", "return", "self", "shape", "should", "size", "sparse_embeddings", "sparse_prompt_embeddings", "sure", "target_embedding", "tensor", "that", "the", "to", "torch", "vision_attentions", "vision_encoder", "vision_hidden_states", "vision_outputs", "want", "you"], "sam2/modeling_sam2.py:Sam2PatchEmbeddings.__init__": ["Conv2d", "ModelHieraDetConfig", "ModelPatchEmbeddings", "Module", "__init__", "class", "config", "def", "hidden_size", "kernel_size", "nn", "num_channels", "padding", "patch_kernel_size", "patch_padding", "patch_stride", "projection", "self", "stride", "super"], "sam2/modeling_sam2.py:Sam2PatchEmbeddings.forward": ["ModelPatchEmbeddings", "Module", "_", "class", "def", "dtype", "embeddings", "forward", "height", "nn", "num_channels", "permute", "pixel_values", "projection", "return", "self", "shape", "to", "weight", "width"], "sam2/modeling_sam2.py:Sam2SinePositionEmbedding.__init__": ["False", "ModelSinePositionEmbedding", "Module", "None", "True", "ValueError", "__init__", "and", "be", "bool", "class", "def", "else", "float", "if", "int", "is", "math", "nn", "normalize", "not", "num_pos_feats", "passed", "pi", "raise", "scale", "self", "should", "super", "temperature"], "sam2/modeling_sam2.py:Sam2SinePositionEmbedding.forward": ["ModelSinePositionEmbedding", "Module", "None", "Size", "Tensor", "arange", "bool", "cat", "class", "cos", "cumsum", "def", "device", "dim", "dim_t", "div", "dtype", "eps", "flatten", "floor", "forward", "if", "int64", "is", "mask", "nn", "normalize", "not_mask", "num_pos_feats", "permute", "pos", "pos_x", "pos_y", "return", "rounding_mode", "scale", "self", "shape", "sin", "stack", "str", "temperature", "to", "torch", "x_embed", "y_embed", "zeros"], "sam2/modeling_sam2.py:Sam2VisionNeck.__init__": ["Conv2d", "ModelSinePositionEmbedding", "ModelVisionConfig", "ModelVisionNeck", "Module", "ModuleList", "True", "__init__", "append", "backbone_channel_list", "class", "config", "convs", "def", "for", "fpn_hidden_size", "fpn_kernel_size", "fpn_padding", "fpn_stride", "fpn_top_down_levels", "in", "in_channels", "kernel_size", "nn", "normalize", "num_pos_feats", "out_channels", "padding", "position_encoding", "self", "stride", "super"], "sam2/modeling_sam2.py:Sam2VisionNeck.forward": ["F", "False", "ModelVisionNeck", "Module", "None", "Tensor", "align_corners", "antialias", "class", "convs", "def", "device", "dtype", "else", "float32", "for", "forward", "fpn_hidden_states", "fpn_position_encoding", "fpn_top_down_levels", "hidden_states", "i", "if", "in", "interpolate", "lateral_features", "len", "mode", "n", "nearest", "nn", "not", "or", "permute", "position_encoding", "prev_features", "prev_position_encoding", "range", "return", "scale_factor", "self", "shape", "to", "top_down_features", "torch", "tuple", "weight"], "sam2/modeling_sam2.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "attention_mask", "attn_output", "attn_weights", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "softmax", "to", "torch", "training", "transpose", "value"], "sam2/modeling_sam2.py:do_pool": ["False", "None", "Tensor", "ceil_mode", "def", "do_pool", "functional", "if", "int", "is", "kernel_size", "max_pool2d", "nn", "permute", "query_stride", "return", "stride", "torch", "x"], "sam2/modeling_sam2.py:Sam2MultiScaleAttention.__init__": ["False", "Linear", "ModelHieraDetConfig", "ModelMultiScaleAttention", "Module", "None", "__init__", "class", "config", "def", "dim", "dim_out", "head_dim", "int", "is_causal", "nn", "num_attention_heads", "proj", "qkv", "query_stride", "scale", "self", "super", "tuple"], "sam2/modeling_sam2.py:Sam2MultiScaleAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "ModelMultiScaleAttention", "Module", "None", "Tensor", "_", "_attn_implementation", "attention_interface", "attention_mask", "attn_output", "attn_weights", "batch_size", "class", "config", "def", "dim", "do_pool", "dtype", "eager", "eager_attention_forward", "float32", "forward", "functional", "height", "hidden_states", "if", "is_causal", "key", "kwargs", "nn", "num_attention_heads", "proj", "qkv", "query", "query_stride", "reshape", "return", "scale", "scaling", "self", "shape", "softmax", "to", "torch", "transpose", "unbind", "value", "width"], "sam2/modeling_sam2.py:Sam2FeedForward.__init__": ["ACT2FN", "False", "Linear", "ModelFeedForward", "Module", "ModuleList", "_", "__init__", "activation", "bool", "class", "def", "for", "hidden_dim", "in", "input_dim", "int", "layers", "nn", "num_layers", "output_dim", "proj_in", "proj_out", "range", "relu", "self", "sigmoid_output", "str", "super"], "sam2/modeling_sam2.py:Sam2FeedForward.forward": ["F", "ModelFeedForward", "Module", "activation", "class", "def", "for", "forward", "hidden_states", "if", "in", "layer", "layers", "nn", "proj_in", "proj_out", "return", "self", "sigmoid", "sigmoid_output"], "sam2/modeling_sam2.py:window_partition": ["Model_partition", "Model_size", "Models", "batch_size", "contiguous", "def", "functional", "height", "hidden_state", "nn", "num_channels", "pad", "pad_height", "pad_width", "padded_height", "padded_width", "permute", "return", "shape", "view", "width"], "sam2/modeling_sam2.py:window_unpartition": ["Model_size", "Model_unpartition", "Models", "batch_size", "contiguous", "def", "height", "height_width", "hidden_state", "pad_height_width", "padded_height", "padded_width", "permute", "return", "shape", "view", "width"], "sam2/modeling_sam2.py:Sam2MultiScaleBlock.__init__": ["GradientCheckpointingLayer", "LayerNorm", "Linear", "ModelFeedForward", "ModelHieraDetConfig", "ModelMultiScaleAttention", "ModelMultiScaleBlock", "None", "__init__", "activation", "and", "attn", "block_idx", "class", "config", "def", "dim", "dim_out", "else", "embed_dim_per_stage", "eps", "global_attention_blocks", "hidden_act", "if", "in", "int", "layer_norm1", "layer_norm2", "layer_norm_eps", "mlp", "mlp_ratio", "nn", "num_attention_heads", "num_attention_heads_per_stage", "num_layers", "num_query_pool_stages", "proj", "query_stride", "self", "stage_idx", "super", "total_block_idx", "window_size", "window_size_per_stage"], "sam2/modeling_sam2.py:Sam2MultiScaleBlock.forward": ["FloatTensor", "GradientCheckpointingLayer", "H", "ModelMultiScaleBlock", "Tensor", "TransformersKwargs", "Unpack", "W", "attn", "attn_output", "class", "def", "dim", "dim_out", "do_pool", "forward", "hidden_states", "if", "kwargs", "layer_norm1", "layer_norm2", "layernorm_output", "mlp", "pad_h", "pad_hw", "pad_w", "proj", "query_stride", "residual", "return", "self", "shape", "torch", "window_partition", "window_size", "window_unpartition"], "sam2/modeling_sam2.py:Sam2PreTrainedModel._init_weights": ["ModelHieraDetModel", "ModelModel", "ModelPositionalEmbedding", "ModelPreTrainedModel", "None", "PreTrainedModel", "_init_weights", "class", "def", "elif", "if", "init", "is", "isinstance", "module", "no_memory_embedding", "normal_", "not", "pos_embed", "pos_embed_window", "positional_embedding", "scale", "self", "std", "super", "zeros_"], "sam2/modeling_sam2.py:Sam2HieraDetModel.__init__": ["ModelHieraDetConfig", "ModelHieraDetModel", "ModelMultiScaleBlock", "ModelPatchEmbeddings", "ModelPreTrainedModel", "ModuleList", "Parameter", "__init__", "append", "block", "block_idx", "blocks", "blocks_per_stage", "class", "config", "cumsum", "def", "enumerate", "for", "hidden_size", "in", "nn", "np", "patch_embed", "pos_embed", "pos_embed_window", "post_init", "range", "self", "stage_ends", "stage_idx", "super", "tolist", "torch", "total_block_idx", "window_positional_embedding_background_size", "window_size_per_stage", "zeros"], "sam2/modeling_sam2.py:Sam2HieraDetModel.get_input_embeddings": ["ModelHieraDetModel", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "patch_embed", "return", "self"], "sam2/modeling_sam2.py:Sam2HieraDetModel._get_pos_embed": ["F", "ModelHieraDetModel", "ModelPreTrainedModel", "Tensor", "_get_pos_embed", "bicubic", "class", "def", "for", "h", "hw", "in", "int", "interpolate", "mode", "permute", "pos_embed", "pos_embed_window", "return", "self", "shape", "size", "tile", "torch", "tuple", "w", "window_embed", "x", "y", "zip"], "sam2/modeling_sam2.py:Sam2HieraDetModel.forward": ["FloatTensor", "ModelHieraDetModel", "ModelHieraDetModelOutput", "ModelPreTrainedModel", "None", "TransformersKwargs", "Unpack", "ValueError", "You", "_get_pos_embed", "block_module", "blocks", "class", "def", "enumerate", "for", "forward", "have", "hidden_states", "i", "if", "in", "intermediate_hidden_states", "is", "kwargs", "last_hidden_state", "patch_embed", "pixel_values", "raise", "return", "self", "shape", "specify", "stage_ends", "to", "torch", "tuple"], "sam2/modeling_sam2.py:Sam2VisionModel.__init__": ["AutoModel", "ModelPreTrainedModel", "ModelVisionConfig", "ModelVisionModel", "ModelVisionNeck", "__init__", "backbone", "backbone_config", "class", "config", "def", "from_config", "neck", "num_feature_levels", "post_init", "self", "super"], "sam2/modeling_sam2.py:Sam2VisionModel.get_input_embeddings": ["ModelPreTrainedModel", "ModelVisionModel", "backbone", "class", "def", "get_input_embeddings", "return", "self"], "sam2/modeling_sam2.py:Sam2VisionModel.forward": ["FloatTensor", "ModelPreTrainedModel", "ModelVisionEncoderOutput", "ModelVisionModel", "None", "TransformersKwargs", "Unpack", "ValueError", "You", "backbone", "backbone_output", "class", "def", "forward", "fpn_hidden_states", "fpn_position_encoding", "have", "hidden_states", "if", "intermediate_hidden_states", "is", "kwargs", "last_hidden_state", "neck", "num_feature_levels", "pixel_values", "raise", "return", "self", "specify", "to", "torch", "tuple"], "sam2/modeling_sam2.py:Sam2PositionalEmbedding.__init__": ["ModelPositionalEmbedding", "ModelPromptEncoderConfig", "Module", "__init__", "class", "config", "def", "hidden_size", "nn", "positional_embedding", "randn", "register_buffer", "scale", "self", "super", "torch"], "sam2/modeling_sam2.py:Sam2PositionalEmbedding.forward": ["ModelPositionalEmbedding", "Module", "None", "cat", "class", "clone", "coordinates", "cos", "def", "dim", "dtype", "float32", "forward", "if", "input_coords", "input_shape", "is", "nn", "not", "np", "pi", "positional_embedding", "return", "self", "sin", "to", "torch"], "sam2/modeling_sam2.py:Sam2MaskEmbedding.__init__": ["ACT2FN", "Conv2d", "ModelLayerNorm", "ModelMaskEmbedding", "ModelPromptEncoderConfig", "Module", "__init__", "activation", "channels_first", "class", "config", "conv1", "conv2", "conv3", "data_format", "def", "eps", "hidden_act", "hidden_size", "kernel_size", "layer_norm1", "layer_norm2", "layer_norm_eps", "mask_input_channels", "nn", "self", "stride", "super"], "sam2/modeling_sam2.py:Sam2MaskEmbedding.forward": ["ModelMaskEmbedding", "Module", "activation", "class", "conv1", "conv2", "conv3", "def", "dense_embeddings", "forward", "hidden_states", "layer_norm1", "layer_norm2", "masks", "nn", "return", "self"], "sam2/modeling_sam2.py:Sam2PromptEncoder.__init__": ["Embedding", "ModelMaskEmbedding", "ModelPositionalEmbedding", "ModelPromptEncoder", "ModelPromptEncoderConfig", "Module", "__init__", "class", "config", "def", "hidden_size", "image_embedding_size", "image_size", "input_image_size", "mask_embed", "mask_input_size", "nn", "no_mask_embed", "not_a_point_embed", "num_point_embeddings", "patch_size", "point_embed", "self", "shared_embedding", "super"], "sam2/modeling_sam2.py:Sam2PromptEncoder._embed_points": ["ModelPromptEncoder", "Module", "None", "Tensor", "_embed_points", "bool", "clamp", "class", "constant", "def", "functional", "if", "input_image_size", "input_shape", "labels", "min", "mode", "nn", "not_a_point_embed", "pad", "point_embed", "point_embedding", "points", "return", "self", "shared_embedding", "torch", "unsqueeze", "value", "weight", "where", "zeros_like"], "sam2/modeling_sam2.py:Sam2PromptEncoder._embed_boxes": ["ModelPromptEncoder", "Module", "Tensor", "_embed_boxes", "boxes", "class", "constant", "coords", "corner_embedding", "def", "expand_as", "functional", "input_image_size", "mode", "nn", "not_a_point_embed", "pad", "point_embed", "return", "self", "shape", "shared_embedding", "torch", "value", "view", "weight"], "sam2/modeling_sam2.py:Sam2PromptEncoder.forward": ["If", "ModelPromptEncoder", "Module", "None", "Tensor", "ValueError", "_embed_boxes", "_embed_points", "also", "are", "batch_size", "be", "box_embeddings", "cat", "class", "def", "dense_embeddings", "dim", "else", "expand", "forward", "if", "image_embedding_size", "input_boxes", "input_labels", "input_masks", "input_points", "is", "labels", "mask_embed", "must", "nn", "no_mask_embed", "not", "pad", "point_embeddings", "points", "provided", "raise", "reshape", "return", "self", "shape", "sparse_embeddings", "torch", "tuple", "weight"], "sam2/modeling_sam2.py:Sam2Attention.__init__": ["False", "Linear", "ModelAttention", "Module", "None", "__init__", "attention_downModelple_rate", "class", "config", "def", "downModelple_rate", "else", "head_dim", "hidden_size", "if", "internal_dim", "is", "is_causal", "k_proj", "nn", "num_attention_heads", "o_proj", "q_proj", "scaling", "self", "super", "v_proj"], "sam2/modeling_sam2.py:Sam2Attention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "ModelAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_attn_implementation", "attention_interface", "attention_mask", "attention_similarity", "attn_output", "attn_weights", "batch_size", "class", "config", "contiguous", "def", "dropout", "eager", "eager_attention_forward", "forward", "head_dim", "if", "is_causal", "k_proj", "key", "kwargs", "new_shape", "nn", "num_attention_heads", "o_proj", "point_batch_size", "q_proj", "query", "reshape", "return", "scaling", "self", "shape", "torch", "transpose", "tuple", "v_proj", "value", "view"], "sam2/modeling_sam2.py:Sam2TwoWayAttentionBlock.__init__": ["False", "LayerNorm", "ModelAttention", "ModelFeedForward", "ModelMaskDecoderConfig", "ModelTwoWayAttentionBlock", "Module", "__init__", "bool", "class", "config", "cross_attn_image_to_token", "cross_attn_token_to_image", "def", "downModelple_rate", "hidden_size", "layer_norm1", "layer_norm2", "layer_norm3", "layer_norm4", "mlp", "mlp_dim", "nn", "num_hidden_layers", "num_layers", "self", "self_attn", "skip_first_layer_pe", "super"], "sam2/modeling_sam2.py:Sam2TwoWayAttentionBlock.forward": ["ModelTwoWayAttentionBlock", "Module", "Tensor", "TransformersKwargs", "Unpack", "_", "attention_similarity", "attn_out", "class", "cross_attn_image_to_token", "cross_attn_token_to_image", "def", "else", "forward", "if", "key", "key_point_embedding", "keys", "kwargs", "layer_norm1", "layer_norm2", "layer_norm3", "layer_norm4", "mlp", "mlp_out", "nn", "queries", "query", "query_point_embedding", "return", "self", "self_attn", "skip_first_layer_pe", "value"], "sam2/modeling_sam2.py:Sam2TwoWayTransformer.__init__": ["LayerNorm", "ModelAttention", "ModelMaskDecoderConfig", "ModelTwoWayAttentionBlock", "ModelTwoWayTransformer", "Module", "ModuleList", "__init__", "append", "class", "config", "def", "final_attn_token_to_image", "for", "hidden_size", "i", "in", "layer_norm_final_attn", "layers", "nn", "num_hidden_layers", "range", "self", "skip_first_layer_pe", "super"], "sam2/modeling_sam2.py:Sam2TwoWayTransformer.forward": ["BaseModelOutput", "ModelTwoWayTransformer", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "_", "an", "attention_similarity", "attn_out", "class", "def", "final_attn_token_to_image", "flatten", "for", "forward", "have", "if", "image_embedding", "image_embeddings", "image_positional_embeddings", "in", "is", "key", "key_point_embedding", "keys", "kwargs", "layer", "layer_norm_final_attn", "layers", "nn", "not", "permute", "point_embeddings", "queries", "query", "query_point_embedding", "raise", "return", "self", "specify", "target_embedding", "to", "tuple", "unsqueeze", "value"], "sam2/modeling_sam2.py:Sam2LayerNorm.__init__": ["LayerNorm", "ModelLayerNorm", "NotImplementedError", "Unsupported", "__init__", "channels_first", "channels_last", "class", "data", "data_format", "def", "eps", "f", "format", "if", "in", "kwargs", "nn", "normalized_shape", "not", "raise", "self", "super"], "sam2/modeling_sam2.py:Sam2LayerNorm.forward": ["LayerNorm", "ModelLayerNorm", "Tensor", "channels_first", "class", "data_format", "def", "else", "features", "forward", "if", "nn", "permute", "return", "self", "super", "torch"], "sam2/modeling_sam2.py:Sam2MaskDecoder.__init__": ["Conv2d", "ConvTranspose2d", "Embedding", "GELU", "ModelFeedForward", "ModelLayerNorm", "ModelMaskDecoder", "ModelMaskDecoderConfig", "ModelTwoWayTransformer", "Module", "ModuleList", "True", "_", "__init__", "activation", "channels_first", "class", "config", "conv_s0", "conv_s1", "data_format", "def", "dynamic_multimask_stability_delta", "dynamic_multimask_stability_thresh", "dynamic_multimask_via_stability", "for", "hidden_size", "in", "iou_head_depth", "iou_head_hidden_dim", "iou_prediction_head", "iou_token", "kernel_size", "mask_tokens", "mlps_list", "nn", "num_mask_tokens", "num_multimask_outputs", "obj_score_token", "output_hypernetworks_mlps", "pred_obj_score_head", "range", "self", "sigmoid_output", "stride", "super", "transformer", "upscale_conv1", "upscale_conv2", "upscale_layer_norm"], "sam2/modeling_sam2.py:Sam2MaskDecoder.forward": ["ModelMaskDecoder", "Model_tokens_out", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "_dynamic_multimask_via_stability", "activation", "and", "attention_similarity", "batch_size", "bool", "cat", "class", "current_mlp", "def", "dense_prompt_embeddings", "dim", "dtype", "dynamic_multimask_via_stability", "elif", "else", "feat_s0", "feat_s1", "for", "forward", "height", "high_resolution_features", "hyper_in", "hyper_in_list", "i", "if", "image_embeddings", "image_positional_embeddings", "in", "iou_pred", "iou_prediction_head", "iou_token", "iou_token_out", "kwargs", "list", "mask_slice", "mask_tokens", "mask_tokens_out", "masks", "multimask_output", "nn", "not", "num_channels", "num_mask_tokens", "obj_score_token", "object_score_logits", "output_hypernetworks_mlps", "output_tokens", "point_batch_size", "point_embeddings", "pred_obj_score_head", "range", "repeat", "repeat_interleave", "return", "self", "shape", "slice", "sparse_prompt_embeddings", "stack", "target_embedding", "to", "tokens", "torch", "training", "transformer", "transpose", "tuple", "upscale_conv1", "upscale_conv2", "upscale_layer_norm", "upscaled_embedding", "view", "weight", "width"], "sam2/modeling_sam2.py:Sam2MaskDecoder._get_stability_scores": ["ModelMaskDecoder", "Module", "_get_stability_scores", "area_i", "area_u", "class", "def", "dim", "dynamic_multimask_stability_delta", "flatten", "float", "mask_logits", "nn", "return", "self", "stability_delta", "stability_scores", "sum", "torch", "where"], "sam2/modeling_sam2.py:Sam2MaskDecoder._dynamic_multimask_via_stability": ["ModelMaskDecoder", "Module", "None", "_dynamic_multimask_via_stability", "_get_stability_scores", "all_iou_scores", "all_mask_logits", "argmax", "best_multimask_iou_scores", "best_multimask_logits", "best_scores_inds", "best_scores_inds_expanded", "class", "def", "dim", "dynamic_multimask_stability_thresh", "expand", "expand_as", "gather", "iou_scores_out", "is_stable", "mask_logits_out", "multimask_iou_scores", "multimask_logits", "nn", "return", "self", "singlemask_iou_scores", "singlemask_logits", "size", "stability_scores", "torch", "unsqueeze", "where"], "sam2/modeling_sam2.py:Sam2Model.__init__": ["AutoModel", "ModelConfig", "ModelMaskDecoder", "ModelModel", "ModelPositionalEmbedding", "ModelPreTrainedModel", "ModelPromptEncoder", "Parameter", "__init__", "_attn_implementation", "backbone_feature_sizes", "class", "config", "def", "fpn_hidden_size", "from_config", "hidden_dim", "mask_decoder", "mask_decoder_config", "nn", "no_memory_embedding", "num_feature_levels", "post_init", "prompt_encoder", "prompt_encoder_config", "self", "shared_image_embedding", "super", "torch", "vision_config", "vision_encoder", "zeros"], "sam2/modeling_sam2.py:Sam2Model.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "return", "self", "vision_encoder"], "sam2/modeling_sam2.py:Sam2Model.get_image_wide_positional_embeddings": ["ModelModel", "ModelPreTrainedModel", "Tensor", "class", "cumsum", "def", "device", "dim", "dtype", "get_image_wide_positional_embeddings", "grid", "image_embedding_size", "ones", "permute", "positional_embedding", "prompt_encoder", "return", "self", "shared_image_embedding", "size", "stack", "target_device", "target_dtype", "torch", "unsqueeze", "x_embed", "y_embed"], "sam2/modeling_sam2.py:Sam2Model.get_image_embeddings": ["FloatTensor", "ModelModel", "ModelPreTrainedModel", "Tensor", "TransformersKwargs", "Unpack", "_", "backbone_feature_sizes", "batch_size", "class", "def", "feat", "feat_size", "feature_maps", "for", "get_image_embeddings", "get_image_features", "image_embeddings", "in", "kwargs", "list", "no_memory_embedding", "permute", "pixel_values", "r", "return", "self", "shape", "torch", "view", "zip"], "sam2/modeling_sam2.py:Sam2Model.get_prompt_embeddings": ["FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "class", "def", "get_prompt_embeddings", "input_boxes", "input_labels", "input_masks", "input_points", "prompt_encoder", "prompt_output", "r", "return", "self", "torch"], "sam2/modeling_sam2.py:Sam2Model.forward": ["Exactly", "F", "False", "FloatTensor", "Got", "LongTensor", "ModelImageSegmentationOutput", "ModelModel", "ModelPreTrainedModel", "None", "TransformersKwargs", "True", "Unpack", "ValueError", "You", "_", "align_corners", "and", "antialias", "as", "attention_similarity", "backbone_feature_sizes", "batch_size", "be", "bilinear", "bool", "bounding", "box", "boxes", "class", "def", "dense_embeddings", "dense_prompt_embeddings", "device", "dtype", "else", "f", "feat", "feat_size", "feature_maps", "float", "for", "forward", "get_image_features", "get_image_wide_positional_embeddings", "high_resolution_features", "if", "image_embeddings", "image_positional_embeddings", "in", "input", "input_boxes", "input_labels", "input_masks", "input_points", "int", "int32", "interpolate", "iou_scores", "is", "kwargs", "low_res_multimasks", "many", "mask_decoder", "mask_input_size", "mode", "multimask_output", "must", "no_memory_embedding", "not", "object_score_logits", "of", "one", "ones", "ones_like", "or", "per", "permute", "pixel_values", "points", "pred_masks", "prompt_encoder", "provide", "provided", "r", "raise", "repeat", "return", "self", "shape", "should", "size", "sparse_embeddings", "sparse_prompt_embeddings", "target_embedding", "to", "torch", "view", "vision_attentions", "vision_hidden_states", "zeros", "zip"], "sam2/modeling_sam2.py:Sam2Model.get_image_features": ["FloatTensor", "ModelModel", "ModelPreTrainedModel", "ModelVisionEncoderOutput", "None", "Tensor", "TransformersKwargs", "Unpack", "attentions", "class", "conv_s0", "conv_s1", "def", "feature_map", "feature_map_position_embedding", "feature_maps", "feature_maps_position_embeddings", "flatten", "for", "fpn_hidden_states", "fpn_position_encoding", "get_image_features", "hidden_states", "in", "kwargs", "list", "mask_decoder", "permute", "pixel_values", "r", "return", "self", "torch", "tuple", "vision_encoder", "vision_outputs"], "sam2_video/modeling_sam2_video.py:Sam2VideoInferenceCache.__init__": ["ModelInferenceCache", "__init__", "_vision_features", "class", "cpu", "def", "device", "inference_device", "inference_state_device", "int", "max_vision_features_cache_size", "self", "str", "torch"], "sam2_video/modeling_sam2_video.py:Sam2VideoInferenceCache.cache_vision_features": ["ModelInferenceCache", "Tensor", "True", "_vision_features", "and", "cache_vision_features", "cached", "class", "def", "dict", "elif", "else", "features", "for", "frame_idx", "if", "in", "inference_state_device", "int", "isinstance", "items", "key", "keys", "len", "list", "max_vision_features_cache_size", "min", "non_blocking", "pop", "self", "to", "torch", "tuple", "v", "value"], "sam2_video/modeling_sam2_video.py:Sam2VideoInferenceCache.get_vision_features": ["ModelInferenceCache", "None", "Tensor", "True", "_vision_features", "and", "cached", "class", "def", "dict", "elif", "else", "for", "frame_idx", "get_vision_features", "if", "in", "inference_device", "int", "isinstance", "items", "key", "list", "moved", "non_blocking", "not", "return", "self", "to", "torch", "tuple", "v", "value"], "sam2_video/modeling_sam2_video.py:Sam2VideoInferenceCache.clear_all": ["ModelInferenceCache", "_vision_features", "class", "clear", "clear_all", "def", "self"], "sam2_video/modeling_sam2_video.py:Sam2VideoInferenceSession.__init__": ["FloatTensor", "ModelInferenceCache", "ModelInferenceSession", "None", "OrderedDict", "__init__", "_obj_id_to_idx", "_obj_idx_to_id", "cache", "class", "cpu", "def", "device", "dict", "dtype", "else", "enumerate", "float32", "frames_tracked_per_obj", "if", "inference_device", "inference_state_device", "int", "is", "mask_inputs_per_obj", "max_vision_features_cache_size", "not", "obj_ids", "obj_with_new_inputs", "output_dict_per_obj", "point_inputs_per_obj", "processed_frames", "self", "str", "to", "torch", "video", "video_height", "video_storage_device", "video_width"], "sam2_video/modeling_sam2_video.py:Sam2VideoInferenceSession.num_frames": ["ModelInferenceSession", "None", "class", "def", "else", "if", "int", "is", "len", "not", "num_frames", "processed_frames", "return", "self"], "sam2_video/modeling_sam2_video.py:Sam2VideoInferenceSession.obj_id_to_idx": ["ModelInferenceSession", "None", "_obj_id_to_idx", "_obj_idx_to_id", "class", "cond_frame_outputs", "def", "frames_tracked_per_obj", "get", "if", "int", "is", "len", "list", "mask_inputs_per_obj", "non_cond_frame_outputs", "not", "obj_id", "obj_id_to_idx", "obj_ids", "obj_idx", "output_dict_per_obj", "point_inputs_per_obj", "return", "self"], "sam2_video/modeling_sam2_video.py:Sam2VideoInferenceSession.obj_idx_to_id": ["ModelInferenceSession", "_obj_idx_to_id", "class", "def", "int", "obj_idx", "obj_idx_to_id", "return", "self"], "sam2_video/modeling_sam2_video.py:Sam2VideoInferenceSession.get_obj_num": ["ModelInferenceSession", "_obj_idx_to_id", "class", "def", "get_obj_num", "int", "len", "return", "self"], "sam2_video/modeling_sam2_video.py:Sam2VideoInferenceSession.add_point_inputs": ["False", "ModelInferenceSession", "Tensor", "add_point_inputs", "class", "def", "device_inputs", "dict", "else", "for", "frame_idx", "if", "in", "inference_device", "inputs", "int", "isinstance", "items", "key", "non_blocking", "obj_idx", "point_inputs_per_obj", "self", "to", "torch", "value"], "sam2_video/modeling_sam2_video.py:Sam2VideoInferenceSession.remove_point_inputs": ["ModelInferenceSession", "None", "class", "def", "frame_idx", "int", "obj_idx", "point_inputs_per_obj", "pop", "remove_point_inputs", "self"], "sam2_video/modeling_sam2_video.py:Sam2VideoInferenceSession.add_mask_inputs": ["ModelInferenceSession", "Tensor", "True", "add_mask_inputs", "class", "def", "dtype", "frame_idx", "inference_device", "inputs", "int", "mask_inputs_per_obj", "non_blocking", "obj_idx", "self", "to", "torch"], "sam2_video/modeling_sam2_video.py:Sam2VideoInferenceSession.remove_mask_inputs": ["ModelInferenceSession", "None", "class", "def", "frame_idx", "int", "mask_inputs_per_obj", "obj_idx", "pop", "remove_mask_inputs", "self"], "sam2_video/modeling_sam2_video.py:Sam2VideoInferenceSession.store_output": ["ModelInferenceSession", "None", "Tensor", "True", "and", "bool", "class", "cond_frame_outputs", "def", "dict", "elif", "else", "for", "frame_idx", "if", "in", "inference_state_device", "int", "is", "is_conditioning_frame", "isinstance", "items", "key", "non_blocking", "non_cond_frame_outputs", "obj_idx", "object_pointer", "object_score_logits", "output_dict_per_obj", "output_key", "output_value", "return", "self", "storage_key", "store_output", "str", "to", "torch", "value"], "sam2_video/modeling_sam2_video.py:Sam2VideoInferenceSession.get_output": ["ModelInferenceSession", "None", "Tensor", "True", "bool", "class", "cond_frame_outputs", "def", "else", "frame_idx", "get", "get_output", "if", "inference_device", "int", "is", "is_conditioning_frame", "isinstance", "non_blocking", "non_cond_frame_outputs", "obj_idx", "out", "output_dict_per_obj", "output_key", "return", "self", "storage_key", "str", "to", "torch", "value"], "sam2_video/modeling_sam2_video.py:Sam2VideoInferenceSession.add_new_frame": ["ModelInferenceSession", "None", "Tensor", "True", "add_new_frame", "class", "def", "dim", "dtype", "else", "frame_idx", "if", "int", "is", "len", "non_blocking", "not", "pixel_values", "processed_frames", "return", "self", "squeeze", "to", "torch", "video_storage_device"], "sam2_video/modeling_sam2_video.py:Sam2VideoInferenceSession.get_frame": ["ModelInferenceSession", "Tensor", "True", "class", "def", "frame_idx", "get_frame", "inference_device", "int", "non_blocking", "processed_frames", "return", "self", "to", "torch"], "sam2_video/modeling_sam2_video.py:Sam2VideoInferenceSession.reset_tracking_data": ["ModelInferenceSession", "_obj_id_to_idx", "_obj_idx_to_id", "class", "clear", "def", "frames_tracked_per_obj", "mask_inputs_per_obj", "obj_ids", "obj_with_new_inputs", "output_dict_per_obj", "point_inputs_per_obj", "reset_tracking_data", "self"], "sam2_video/modeling_sam2_video.py:Sam2VideoInferenceSession.reset_inference_session": ["ModelInferenceSession", "_obj_id_to_idx", "_obj_idx_to_id", "cache", "class", "clear", "clear_all", "def", "frames_tracked_per_obj", "mask_inputs_per_obj", "obj_ids", "obj_with_new_inputs", "output_dict_per_obj", "point_inputs_per_obj", "reset_inference_session", "self"], "sam2_video/modeling_sam2_video.py:Sam2VideoLayerNorm.__init__": ["LayerNorm", "ModelLayerNorm", "NotImplementedError", "Unsupported", "__init__", "channels_first", "channels_last", "class", "data", "data_format", "def", "eps", "f", "format", "if", "in", "kwargs", "nn", "normalized_shape", "not", "raise", "self", "super"], "sam2_video/modeling_sam2_video.py:Sam2VideoLayerNorm.forward": ["LayerNorm", "ModelLayerNorm", "Tensor", "channels_first", "class", "data_format", "def", "else", "features", "forward", "if", "nn", "permute", "return", "self", "super", "torch"], "sam2_video/modeling_sam2_video.py:Sam2VideoPositionEmbeddingSine.__init__": ["False", "ModelPositionEmbeddingSine", "Module", "None", "True", "ValueError", "__init__", "and", "be", "bool", "class", "def", "else", "float", "if", "int", "is", "math", "nn", "normalize", "not", "num_pos_feats", "passed", "pi", "raise", "scale", "self", "should", "super", "temperature"], "sam2_video/modeling_sam2_video.py:Sam2VideoPositionEmbeddingSine.forward": ["ModelPositionEmbeddingSine", "Module", "None", "Size", "Tensor", "arange", "bool", "cat", "class", "cos", "cumsum", "def", "device", "dim", "dim_t", "div", "dtype", "eps", "flatten", "floor", "forward", "if", "int64", "is", "mask", "nn", "normalize", "not_mask", "num_pos_feats", "permute", "pos", "pos_x", "pos_y", "return", "rounding_mode", "scale", "self", "shape", "sin", "stack", "str", "temperature", "to", "torch", "x_embed", "y_embed", "zeros"], "sam2_video/modeling_sam2_video.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "attention_mask", "attn_output", "attn_weights", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "softmax", "to", "torch", "training", "transpose", "value"], "sam2_video/modeling_sam2_video.py:Sam2VideoAttention.__init__": ["False", "Linear", "ModelAttention", "Module", "None", "__init__", "attention_downModelple_rate", "class", "config", "def", "downModelple_rate", "else", "head_dim", "hidden_size", "if", "internal_dim", "is", "is_causal", "k_proj", "nn", "num_attention_heads", "o_proj", "q_proj", "scaling", "self", "super", "v_proj"], "sam2_video/modeling_sam2_video.py:Sam2VideoAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "ModelAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_attn_implementation", "attention_interface", "attention_mask", "attention_similarity", "attn_output", "attn_weights", "batch_size", "class", "config", "contiguous", "def", "dropout", "eager", "eager_attention_forward", "forward", "head_dim", "if", "is_causal", "k_proj", "key", "kwargs", "new_shape", "nn", "num_attention_heads", "o_proj", "point_batch_size", "q_proj", "query", "reshape", "return", "scaling", "self", "shape", "torch", "transpose", "tuple", "v_proj", "value", "view"], "sam2_video/modeling_sam2_video.py:Sam2VideoTwoWayAttentionBlock.__init__": ["False", "LayerNorm", "ModelAttention", "ModelFeedForward", "ModelMaskDecoderConfig", "ModelTwoWayAttentionBlock", "Module", "__init__", "bool", "class", "config", "cross_attn_image_to_token", "cross_attn_token_to_image", "def", "downModelple_rate", "hidden_size", "layer_norm1", "layer_norm2", "layer_norm3", "layer_norm4", "mlp", "mlp_dim", "nn", "num_hidden_layers", "num_layers", "self", "self_attn", "skip_first_layer_pe", "super"], "sam2_video/modeling_sam2_video.py:Sam2VideoTwoWayAttentionBlock.forward": ["ModelTwoWayAttentionBlock", "Module", "Tensor", "TransformersKwargs", "Unpack", "_", "attention_similarity", "attn_out", "class", "cross_attn_image_to_token", "cross_attn_token_to_image", "def", "else", "forward", "if", "key", "key_point_embedding", "keys", "kwargs", "layer_norm1", "layer_norm2", "layer_norm3", "layer_norm4", "mlp", "mlp_out", "nn", "queries", "query", "query_point_embedding", "return", "self", "self_attn", "skip_first_layer_pe", "value"], "sam2_video/modeling_sam2_video.py:Sam2VideoFeedForward.__init__": ["ACT2FN", "False", "Linear", "ModelFeedForward", "Module", "ModuleList", "_", "__init__", "activation", "bool", "class", "def", "for", "hidden_dim", "in", "input_dim", "int", "layers", "nn", "num_layers", "output_dim", "proj_in", "proj_out", "range", "relu", "self", "sigmoid_output", "str", "super"], "sam2_video/modeling_sam2_video.py:Sam2VideoFeedForward.forward": ["F", "ModelFeedForward", "Module", "activation", "class", "def", "for", "forward", "hidden_states", "if", "in", "layer", "layers", "nn", "proj_in", "proj_out", "return", "self", "sigmoid", "sigmoid_output"], "sam2_video/modeling_sam2_video.py:Sam2VideoPreTrainedModel._init_weights": ["ModelMemoryFuserCXBlock", "ModelModel", "ModelPositionalEmbedding", "ModelPreTrainedModel", "ModelVisionRotaryEmbedding", "None", "PreTrainedModel", "_init_weights", "class", "copy_", "cos", "create_inv_freq", "def", "elif", "if", "init", "inv_freq", "is", "isinstance", "memory_temporal_positional_encoding", "module", "no_memory_positional_encoding", "no_object_pointer", "normal_", "not", "occlusion_spatial_embedding_parameter", "positional_embedding", "rope_embeddings_cos", "rope_embeddings_sin", "scale", "self", "sin", "std", "super", "zeros_"], "sam2_video/modeling_sam2_video.py:Sam2VideoVisionRotaryEmbedding.__init__": ["Dimension", "False", "ModelConfig", "ModelVisionRotaryEmbedding", "Module", "RoPE", "ValueError", "__init__", "axial", "be", "by", "class", "config", "cos", "create_inv_freq", "def", "dim", "divisible", "end_x", "end_y", "for", "if", "inv_freq", "memory_attention_downModelple_rate", "memory_attention_hidden_size", "memory_attention_num_attention_heads", "memory_attention_rope_feat_sizes", "memory_attention_rope_theta", "must", "nn", "persistent", "raise", "register_buffer", "rope_embeddings_cos", "rope_embeddings_sin", "self", "sin", "super"], "sam2_video/modeling_sam2_video.py:Sam2VideoVisionRotaryEmbedding.forward": ["ModelVisionRotaryEmbedding", "Module", "Tensor", "class", "def", "forward", "nn", "return", "rope_embeddings_cos", "rope_embeddings_sin", "self", "torch", "tuple"], "sam2_video/modeling_sam2_video.py:Sam2VideoVisionRotaryEmbedding.create_inv_freq": ["ModelVisionRotaryEmbedding", "Module", "arange", "cat", "class", "create_inv_freq", "def", "dim", "div", "dtype", "end_x", "end_y", "flattened_indices", "float", "floor", "freqs", "freqs_x", "freqs_y", "inv_freq", "long", "memory_attention_rope_theta", "nn", "outer", "repeat_interleave", "return", "rounding_mode", "self", "torch", "x_positions", "y_positions"], "sam2_video/modeling_sam2_video.py:rotate_pairwise": ["Model_pairwise", "def", "dim", "flatten", "return", "shape", "stack", "start_dim", "torch", "unbind", "view", "x", "x1", "x2"], "sam2_video/modeling_sam2_video.py:apply_rotary_pos_emb_2d": ["False", "Model_rotary_pos_emb_2d", "Tensor", "and", "bool", "cat", "cos", "cos_k", "def", "dim", "else", "float", "if", "int", "k", "k_embed", "k_pass", "k_rot", "num_k_exclude_rope", "q", "q_embed", "repeat", "repeat_factor", "repeat_freqs_k", "return", "rotate_pairwise", "shape", "sin", "sin_k", "torch", "tuple", "type_as"], "sam2_video/modeling_sam2_video.py:Sam2VideoRoPEAttention.__init__": ["False", "Linear", "ModelConfig", "ModelRoPEAttention", "Module", "None", "__init__", "class", "config", "def", "dropout_p", "else", "head_dim", "hidden_size", "if", "int", "internal_dim", "is", "is_causal", "k_proj", "kv_in_dim", "memory_attention_downModelple_rate", "memory_attention_hidden_size", "memory_attention_num_attention_heads", "memory_attention_rope_dropout", "nn", "not", "num_attention_heads", "o_proj", "q_proj", "rope_k_repeat", "scaling", "self", "super", "v_proj"], "sam2_video/modeling_sam2_video.py:Sam2VideoRoPEAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "FlashAttentionKwargs", "ModelRoPEAttention", "Module", "None", "Tensor", "Unpack", "_attn_implementation", "apply_rotary_pos_emb_2d", "attention_interface", "attention_mask", "attn_output", "attn_weights", "batch_size", "class", "config", "contiguous", "cos", "def", "dropout", "dropout_p", "eager", "eager_attention_forward", "else", "forward", "head_dim", "if", "int", "is_causal", "k_proj", "key", "kwargs", "new_shape", "nn", "not", "num_attention_heads", "num_k_exclude_rope", "o_proj", "point_batch_size", "position_embeddings", "q_proj", "query", "repeat_freqs_k", "reshape", "return", "rope_k_repeat", "scaling", "self", "shape", "sin", "torch", "training", "transpose", "tuple", "v_proj", "value", "view"], "sam2_video/modeling_sam2_video.py:Sam2VideoMemoryAttentionLayer.__init__": ["ACT2FN", "Dropout", "LayerNorm", "Linear", "ModelConfig", "ModelMemoryAttentionLayer", "ModelRoPEAttention", "Module", "True", "__init__", "activation", "class", "config", "cross_attn_image", "def", "dropout", "dropout1", "dropout2", "dropout3", "hidden_size", "kv_in_dim", "layer_norm1", "layer_norm2", "layer_norm3", "linear1", "linear2", "memory_attention_dropout", "memory_attention_feed_forward_hidden_act", "memory_attention_feed_forward_hidden_size", "memory_attention_hidden_size", "nn", "rope_k_repeat", "self", "self_attn", "super"], "sam2_video/modeling_sam2_video.py:Sam2VideoMemoryAttentionLayer.forward": ["ModelMemoryAttentionLayer", "Module", "Tensor", "_", "activation", "class", "cross_attn_image", "def", "dropout", "dropout1", "dropout2", "dropout3", "forward", "int", "key", "key_point_embedding", "keys", "layer_norm1", "layer_norm2", "layer_norm3", "linear1", "linear2", "nn", "num_k_exclude_rope", "position_embeddings", "queries", "query", "return", "rope_position_embeddings", "self", "self_attn", "torch", "tuple", "value"], "sam2_video/modeling_sam2_video.py:Sam2VideoMemoryAttention.__init__": ["LayerNorm", "ModelConfig", "ModelMemoryAttention", "ModelMemoryAttentionLayer", "ModelVisionRotaryEmbedding", "Module", "ModuleList", "_", "__init__", "class", "config", "def", "for", "in", "layer_norm", "layers", "memory_attention_hidden_size", "memory_attention_num_layers", "nn", "range", "rotary_emb", "self", "super"], "sam2_video/modeling_sam2_video.py:Sam2VideoMemoryAttention.forward": ["ModelMemoryAttention", "Module", "None", "Tensor", "class", "current_vision_features", "current_vision_position_embeddings", "def", "else", "for", "forward", "if", "in", "int", "is", "key_point_embedding", "keys", "layer", "layer_norm", "layers", "memory", "memory_posision_embeddings", "ndim", "nn", "normed_output", "not", "num_k_exclude_rope", "num_object_pointer_tokens", "output", "queries", "return", "rope_position_embeddings", "rotary_emb", "self", "torch", "transpose", "unsqueeze"], "sam2_video/modeling_sam2_video.py:Sam2VideoMemoryFuserCXBlock.__init__": ["ACT2FN", "Conv2d", "GradientCheckpointingLayer", "Linear", "ModelConfig", "ModelLayerNorm", "ModelMemoryFuserCXBlock", "Parameter", "True", "__init__", "activation", "channels_first", "class", "config", "data_format", "def", "depthwise_conv", "eps", "groups", "kernel_size", "layer_norm", "memory_fuser_embed_dim", "memory_fuser_hidden_act", "memory_fuser_intermediate_dim", "memory_fuser_kernel_size", "memory_fuser_layer_scale_init_value", "memory_fuser_padding", "nn", "ones", "padding", "pointwise_conv1", "pointwise_conv2", "requires_grad", "scale", "self", "super", "torch"], "sam2_video/modeling_sam2_video.py:Sam2VideoMemoryFuserCXBlock.forward": ["GradientCheckpointingLayer", "ModelMemoryFuserCXBlock", "activation", "class", "def", "depthwise_conv", "forward", "hidden_states", "input", "layer_norm", "permute", "pointwise_conv1", "pointwise_conv2", "return", "scale", "self"], "sam2_video/modeling_sam2_video.py:Sam2VideoMemoryFuser.__init__": ["ModelConfig", "ModelMemoryFuser", "ModelMemoryFuserCXBlock", "Module", "ModuleList", "_", "__init__", "class", "config", "def", "for", "in", "layers", "memory_fuser_num_layers", "nn", "range", "self", "super"], "sam2_video/modeling_sam2_video.py:Sam2VideoMemoryFuser.forward": ["ModelMemoryFuser", "Module", "class", "def", "for", "forward", "hidden_states", "in", "layer", "layers", "nn", "return", "self"], "sam2_video/modeling_sam2_video.py:Sam2VideoMaskDownSamplerLayer.__init__": ["ACT2FN", "Conv2d", "ModelConfig", "ModelLayerNorm", "ModelMaskDownModelplerLayer", "Module", "__init__", "activation", "channels_first", "class", "config", "conv", "data_format", "def", "eps", "in_channels", "int", "kernel_size", "layer_norm", "mask_downModelpler_hidden_act", "mask_downModelpler_kernel_size", "mask_downModelpler_padding", "mask_downModelpler_stride", "nn", "out_channels", "padding", "self", "stride", "super"], "sam2_video/modeling_sam2_video.py:Sam2VideoMaskDownSamplerLayer.forward": ["ModelMaskDownModelplerLayer", "Module", "activation", "class", "conv", "def", "forward", "layer_norm", "nn", "return", "self", "x"], "sam2_video/modeling_sam2_video.py:Sam2VideoMaskDownSampler.__init__": ["ACT2FN", "Conv2d", "ModelConfig", "ModelMaskDownModelpler", "ModelMaskDownModelplerLayer", "Module", "ModuleList", "_", "__init__", "activation", "append", "class", "config", "def", "final_conv", "for", "in", "int", "kernel_size", "layers", "log2", "mask_downModelpler_embed_dim", "mask_downModelpler_hidden_act", "mask_downModelpler_stride", "mask_downModelpler_total_stride", "mask_in_chans", "mask_out_chans", "math", "nn", "num_layers", "range", "self", "super"], "sam2_video/modeling_sam2_video.py:Sam2VideoMaskDownSampler.forward": ["ModelMaskDownModelpler", "Module", "class", "def", "final_conv", "for", "forward", "in", "layer", "layers", "nn", "return", "self", "x"], "sam2_video/modeling_sam2_video.py:Sam2VideoMemoryEncoder.__init__": ["Conv2d", "ModelConfig", "ModelMaskDownModelpler", "ModelMemoryEncoder", "ModelMemoryFuser", "ModelPositionEmbeddingSine", "Module", "True", "__init__", "class", "config", "def", "feature_projection", "hidden_size", "kernel_size", "mask_downModelpler", "memory_encoder_hidden_size", "memory_encoder_output_channels", "memory_fuser", "nn", "normalize", "num_pos_feats", "output_channels", "position_encoding", "projection", "self", "super"], "sam2_video/modeling_sam2_video.py:Sam2VideoMemoryEncoder.forward": ["ModelMemoryEncoder", "Module", "Tensor", "class", "def", "device", "dtype", "feature_projection", "forward", "mask_downModelpler", "masks", "memory_fuser", "nn", "position_encoding", "projection", "return", "self", "shape", "torch", "tuple", "vision_features", "vision_pos_enc"], "sam2_video/modeling_sam2_video.py:Sam2VideoPositionalEmbedding.__init__": ["ModelPositionalEmbedding", "ModelPromptEncoderConfig", "Module", "__init__", "class", "config", "def", "hidden_size", "nn", "positional_embedding", "randn", "register_buffer", "scale", "self", "super", "torch"], "sam2_video/modeling_sam2_video.py:Sam2VideoPositionalEmbedding.forward": ["ModelPositionalEmbedding", "Module", "None", "cat", "class", "clone", "coordinates", "cos", "def", "dim", "dtype", "float32", "forward", "if", "input_coords", "input_shape", "is", "nn", "not", "np", "pi", "positional_embedding", "return", "self", "sin", "to", "torch"], "sam2_video/modeling_sam2_video.py:Sam2VideoMaskEmbedding.__init__": ["ACT2FN", "Conv2d", "ModelLayerNorm", "ModelMaskEmbedding", "ModelPromptEncoderConfig", "Module", "__init__", "activation", "channels_first", "class", "config", "conv1", "conv2", "conv3", "data_format", "def", "eps", "hidden_act", "hidden_size", "kernel_size", "layer_norm1", "layer_norm2", "layer_norm_eps", "mask_input_channels", "nn", "self", "stride", "super"], "sam2_video/modeling_sam2_video.py:Sam2VideoMaskEmbedding.forward": ["ModelMaskEmbedding", "Module", "activation", "class", "conv1", "conv2", "conv3", "def", "dense_embeddings", "forward", "hidden_states", "layer_norm1", "layer_norm2", "masks", "nn", "return", "self"], "sam2_video/modeling_sam2_video.py:Sam2VideoPromptEncoder.__init__": ["Embedding", "ModelMaskEmbedding", "ModelPositionalEmbedding", "ModelPromptEncoder", "ModelPromptEncoderConfig", "Module", "__init__", "class", "config", "def", "hidden_size", "image_embedding_size", "image_size", "input_image_size", "mask_embed", "mask_input_size", "nn", "no_mask_embed", "not_a_point_embed", "num_point_embeddings", "patch_size", "point_embed", "self", "shared_embedding", "super"], "sam2_video/modeling_sam2_video.py:Sam2VideoPromptEncoder._embed_points": ["ModelPromptEncoder", "Module", "None", "Tensor", "_embed_points", "bool", "clamp", "class", "constant", "def", "functional", "if", "input_image_size", "input_shape", "labels", "min", "mode", "nn", "not_a_point_embed", "pad", "point_embed", "point_embedding", "points", "return", "self", "shared_embedding", "torch", "unsqueeze", "value", "weight", "where", "zeros_like"], "sam2_video/modeling_sam2_video.py:Sam2VideoPromptEncoder._embed_boxes": ["ModelPromptEncoder", "Module", "Tensor", "_embed_boxes", "boxes", "class", "constant", "coords", "corner_embedding", "def", "expand_as", "functional", "input_image_size", "mode", "nn", "not_a_point_embed", "pad", "point_embed", "return", "self", "shape", "shared_embedding", "torch", "value", "view", "weight"], "sam2_video/modeling_sam2_video.py:Sam2VideoPromptEncoder.forward": ["If", "ModelPromptEncoder", "Module", "None", "Tensor", "ValueError", "_embed_boxes", "_embed_points", "also", "are", "batch_size", "be", "box_embeddings", "cat", "class", "def", "dense_embeddings", "dim", "else", "expand", "forward", "if", "image_embedding_size", "input_boxes", "input_labels", "input_masks", "input_points", "is", "labels", "mask_embed", "must", "nn", "no_mask_embed", "not", "pad", "point_embeddings", "points", "provided", "raise", "reshape", "return", "self", "shape", "sparse_embeddings", "torch", "tuple", "weight"], "sam2_video/modeling_sam2_video.py:Sam2VideoTwoWayTransformer.__init__": ["LayerNorm", "ModelAttention", "ModelMaskDecoderConfig", "ModelTwoWayAttentionBlock", "ModelTwoWayTransformer", "Module", "ModuleList", "__init__", "append", "class", "config", "def", "final_attn_token_to_image", "for", "hidden_size", "i", "in", "layer_norm_final_attn", "layers", "nn", "num_hidden_layers", "range", "self", "skip_first_layer_pe", "super"], "sam2_video/modeling_sam2_video.py:Sam2VideoTwoWayTransformer.forward": ["BaseModelOutput", "ModelTwoWayTransformer", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "_", "an", "attention_similarity", "attn_out", "class", "def", "final_attn_token_to_image", "flatten", "for", "forward", "have", "if", "image_embedding", "image_embeddings", "image_positional_embeddings", "in", "is", "key", "key_point_embedding", "keys", "kwargs", "layer", "layer_norm_final_attn", "layers", "nn", "not", "permute", "point_embeddings", "queries", "query", "query_point_embedding", "raise", "return", "self", "specify", "target_embedding", "to", "tuple", "unsqueeze", "value"], "sam2_video/modeling_sam2_video.py:Sam2VideoMaskDecoder.__init__": ["Conv2d", "ConvTranspose2d", "Embedding", "GELU", "ModelFeedForward", "ModelLayerNorm", "ModelMaskDecoder", "ModelMaskDecoderConfig", "ModelTwoWayTransformer", "Module", "ModuleList", "True", "_", "__init__", "activation", "channels_first", "class", "config", "conv_s0", "conv_s1", "data_format", "def", "dynamic_multimask_stability_delta", "dynamic_multimask_stability_thresh", "dynamic_multimask_via_stability", "for", "hidden_size", "in", "iou_head_depth", "iou_head_hidden_dim", "iou_prediction_head", "iou_token", "kernel_size", "mask_tokens", "mlps_list", "nn", "num_mask_tokens", "num_multimask_outputs", "obj_score_token", "output_hypernetworks_mlps", "pred_obj_score_head", "range", "self", "sigmoid_output", "stride", "super", "transformer", "upscale_conv1", "upscale_conv2", "upscale_layer_norm"], "sam2_video/modeling_sam2_video.py:Sam2VideoMaskDecoder.forward": ["ModelMaskDecoder", "Model_tokens_out", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "_dynamic_multimask_via_stability", "activation", "and", "attention_similarity", "batch_size", "bool", "cat", "class", "current_mlp", "def", "dense_prompt_embeddings", "dim", "dtype", "dynamic_multimask_via_stability", "elif", "else", "feat_s0", "feat_s1", "for", "forward", "height", "high_resolution_features", "hyper_in", "hyper_in_list", "i", "if", "image_embeddings", "image_positional_embeddings", "in", "iou_pred", "iou_prediction_head", "iou_token", "iou_token_out", "kwargs", "list", "mask_slice", "mask_tokens", "mask_tokens_out", "masks", "multimask_output", "nn", "not", "num_channels", "num_mask_tokens", "obj_score_token", "object_score_logits", "output_hypernetworks_mlps", "output_tokens", "point_batch_size", "point_embeddings", "pred_obj_score_head", "range", "repeat", "repeat_interleave", "return", "self", "shape", "slice", "sparse_prompt_embeddings", "stack", "target_embedding", "to", "tokens", "torch", "training", "transformer", "transpose", "tuple", "upscale_conv1", "upscale_conv2", "upscale_layer_norm", "upscaled_embedding", "view", "weight", "width"], "sam2_video/modeling_sam2_video.py:Sam2VideoMaskDecoder._get_stability_scores": ["ModelMaskDecoder", "Module", "_get_stability_scores", "area_i", "area_u", "class", "def", "dim", "dynamic_multimask_stability_delta", "flatten", "float", "mask_logits", "nn", "return", "self", "stability_delta", "stability_scores", "sum", "torch", "where"], "sam2_video/modeling_sam2_video.py:Sam2VideoMaskDecoder._dynamic_multimask_via_stability": ["ModelMaskDecoder", "Module", "None", "_dynamic_multimask_via_stability", "_get_stability_scores", "all_iou_scores", "all_mask_logits", "argmax", "best_multimask_iou_scores", "best_multimask_logits", "best_scores_inds", "best_scores_inds_expanded", "class", "def", "dim", "dynamic_multimask_stability_thresh", "expand", "expand_as", "gather", "iou_scores_out", "is_stable", "mask_logits_out", "multimask_iou_scores", "multimask_logits", "nn", "return", "self", "singlemask_iou_scores", "singlemask_logits", "size", "stability_scores", "torch", "unsqueeze", "where"], "sam2_video/modeling_sam2_video.py:get_1d_sine_pe": ["Model_1d_sine_pe", "arange", "cat", "cos", "def", "device", "dim", "dim_t", "dtype", "float32", "pe_dim", "pos_embed", "pos_inds", "return", "sin", "temperature", "torch", "unsqueeze"], "sam2_video/modeling_sam2_video.py:Sam2VideoModel.__init__": ["AutoModel", "Conv2d", "Identity", "Linear", "ModelConfig", "ModelFeedForward", "ModelMaskDecoder", "ModelMemoryAttention", "ModelMemoryEncoder", "ModelModel", "ModelPositionalEmbedding", "ModelPreTrainedModel", "ModelPromptEncoder", "None", "Parameter", "__init__", "_attn_implementation", "backbone_feature_sizes", "class", "config", "def", "else", "enable_occlusion_spatial_embedding", "enable_temporal_pos_encoding_for_object_pointers", "fpn_hidden_size", "from_config", "hidden_dim", "if", "image_size", "kernel_size", "mask_decoder", "mask_decoder_config", "mask_downModelple", "mem_dim", "memory_attention", "memory_encoder", "memory_encoder_output_channels", "memory_temporal_positional_encoding", "nn", "no_memory_embedding", "no_memory_positional_encoding", "no_object_pointer", "num_feature_levels", "num_maskmem", "object_pointer_proj", "occlusion_spatial_embedding_parameter", "post_init", "prompt_encoder", "prompt_encoder_config", "self", "shared_image_embedding", "stride", "super", "temporal_positional_encoding_projection_layer", "torch", "vision_config", "vision_encoder", "zeros"], "sam2_video/modeling_sam2_video.py:Sam2VideoModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "return", "self", "vision_encoder"], "sam2_video/modeling_sam2_video.py:Sam2VideoModel.get_image_wide_positional_embeddings": ["ModelModel", "ModelPreTrainedModel", "Tensor", "class", "cumsum", "def", "device", "dim", "dtype", "get_image_wide_positional_embeddings", "grid", "image_embedding_size", "ones", "permute", "positional_embedding", "prompt_encoder", "return", "self", "shared_image_embedding", "size", "stack", "target_device", "target_dtype", "torch", "unsqueeze", "x_embed", "y_embed"], "sam2_video/modeling_sam2_video.py:Sam2VideoModel.get_image_embeddings": ["FloatTensor", "ModelModel", "ModelPreTrainedModel", "Tensor", "TransformersKwargs", "Unpack", "_", "backbone_feature_sizes", "batch_size", "class", "def", "feat", "feat_size", "feature_maps", "for", "get_image_embeddings", "get_image_features", "image_embeddings", "in", "kwargs", "list", "no_memory_embedding", "permute", "pixel_values", "r", "return", "self", "shape", "torch", "view", "zip"], "sam2_video/modeling_sam2_video.py:Sam2VideoModel.get_prompt_embeddings": ["FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "class", "def", "get_prompt_embeddings", "input_boxes", "input_labels", "input_masks", "input_points", "prompt_encoder", "prompt_output", "r", "return", "self", "torch", "tuple"], "sam2_video/modeling_sam2_video.py:Sam2VideoModel.forward": ["False", "ModelInferenceSession", "ModelModel", "ModelPreTrainedModel", "ModelSegmentationOutput", "No", "None", "Tensor", "True", "ValueError", "_batch_encode_memories", "_run_single_frame_inference", "add", "add_new_frame", "all_object_score_logits", "all_pred_masks", "and", "append", "are", "batch_size", "bool", "cat", "class", "cond_frame_outputs", "copy", "current_out", "def", "dim", "else", "first", "for", "forward", "frame", "frame_idx", "frames_tracked_per_obj", "get", "get_obj_num", "get_output", "has_cond_output", "has_new_inputs", "high_res_masks", "high_res_masks_for_memory", "if", "in", "inference_session", "inputs", "int", "is", "is_conditioning_frame", "is_init_cond_frame", "is_mask_from_pts_per_obj", "kwargs", "len", "mask_inputs", "mask_inputs_per_obj", "not", "num_maskmem", "num_objects", "obj_id", "obj_ids", "obj_idx", "obj_idx_to_id", "obj_with_new_inputs", "object_ids", "object_score_logits", "object_score_logits_for_memory", "object_score_logits_per_obj", "objects", "objects_needing_memory_encoding", "or", "output_dict_per_obj", "output_value", "please", "point_inputs", "point_inputs_per_obj", "pred_masks", "pred_masks_per_obj", "provided", "r", "raise", "range", "remove", "return", "reverse", "run_mem_encoder", "self", "squeeze", "store_output", "streaming", "torch", "tracking"], "sam2_video/modeling_sam2_video.py:Sam2VideoModel.get_image_features": ["FloatTensor", "ModelModel", "ModelPreTrainedModel", "ModelVisionEncoderOutput", "None", "Tensor", "TransformersKwargs", "Unpack", "attentions", "class", "conv_s0", "conv_s1", "def", "feature_map", "feature_map_position_embedding", "feature_maps", "feature_maps_position_embeddings", "flatten", "for", "fpn_hidden_states", "fpn_position_encoding", "get_image_features", "hidden_states", "in", "kwargs", "list", "mask_decoder", "permute", "pixel_values", "r", "return", "self", "torch", "tuple", "vision_encoder", "vision_outputs"], "sam2_video/modeling_sam2_video.py:Sam2VideoModel._prepare_vision_features": ["ModelInferenceSession", "ModelModel", "ModelPreTrainedModel", "Tensor", "_", "_prepare_vision_features", "batch_size", "cache", "cache_vision_features", "cached_features", "class", "def", "else", "expand", "for", "frame_idx", "get_frame", "get_image_features", "get_vision_features", "if", "image_batch", "in", "inference_session", "int", "list", "pe", "return", "self", "torch", "tuple", "unsqueeze", "vision_feats", "vision_pos_embeds"], "sam2_video/modeling_sam2_video.py:Sam2VideoModel._single_frame_forward": ["Exactly", "F", "False", "FloatTensor", "Got", "LongTensor", "ModelImageSegmentationOutput", "ModelModel", "ModelPreTrainedModel", "Model_output_token", "Model_output_tokens", "NO_OBJ_SCORE", "None", "TransformersKwargs", "True", "Unpack", "ValueError", "You", "_", "_single_frame_forward", "align_corners", "and", "antialias", "arange", "argmax", "as", "attention_similarity", "backbone_feature_sizes", "batch_inds", "batch_size", "be", "best_iou_inds", "bilinear", "bool", "bounding", "box", "boxes", "class", "def", "dense_embeddings", "dense_prompt_embeddings", "device", "dim", "dtype", "elif", "else", "f", "feat", "feat_size", "feature_maps", "float", "for", "get_image_features", "get_image_wide_positional_embeddings", "high_res_masks", "high_res_multimasks", "high_resolution_features", "if", "image_embeddings", "image_positional_embeddings", "image_size", "in", "input", "input_boxes", "input_labels", "input_masks", "input_points", "int", "int32", "interpolate", "iou_scores", "is", "is_obj_appearing", "kwargs", "lambda_is_obj_appearing", "low_res_masks", "low_res_multimasks", "many", "mask_decoder", "mask_input_size", "mode", "multimask_output", "must", "no_memory_embedding", "no_object_pointer", "not", "num_objects", "object_batch_inds", "object_pointer", "object_pointer_proj", "object_score_logits", "of", "one", "ones", "ones_like", "or", "per", "permute", "pixel_values", "points", "pred_masks", "prompt_encoder", "provide", "provided", "raise", "repeat", "return", "self", "shape", "should", "size", "sparse_embeddings", "sparse_prompt_embeddings", "squeeze", "target_embedding", "to", "torch", "unsqueeze", "view", "vision_attentions", "vision_hidden_states", "where", "zeros", "zip"], "sam2_video/modeling_sam2_video.py:Sam2VideoModel._use_mask_as_output": ["F", "False", "ModelImageSegmentationOutput", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "True", "_single_frame_forward", "_use_mask_as_output", "align_corners", "antialias", "any", "backbone_features", "bilinear", "class", "def", "dim", "dtype", "flatten", "float", "high_res_features", "high_res_masks", "if", "image_embeddings", "image_size", "input_masks", "interpolate", "iou_scores", "is_obj_appearing", "lambda_is_obj_appearing", "list", "low_res_masks", "mask_downModelple", "mask_input_size", "mask_inputs", "mask_inputs_float", "mode", "new_ones", "no_object_pointer", "object_pointer", "object_score_logits", "out_bias", "out_scale", "pred_masks", "prompt_encoder", "return", "self", "shape", "size", "to", "torch", "unsqueeze"], "sam2_video/modeling_sam2_video.py:Sam2VideoModel._select_closest_cond_frames": ["ModelModel", "ModelPreTrainedModel", "None", "_select_closest_cond_frames", "abs", "class", "cond_frame_outputs", "def", "default", "else", "for", "frame_idx", "idx_after", "idx_before", "if", "in", "inds_remain", "is", "items", "key", "lambda", "len", "max", "max_cond_frame_num", "min", "not", "num_remain", "or", "return", "selected_outputs", "self", "sorted", "t", "unselected_outputs", "update", "v", "x"], "sam2_video/modeling_sam2_video.py:Sam2VideoModel._gather_memory_frame_outputs": ["False", "ModelInferenceSession", "ModelModel", "ModelPreTrainedModel", "None", "ValueError", "_gather_memory_frame_outputs", "_select_closest_cond_frames", "append", "be", "bool", "cannot", "class", "cond_frame_outputs", "conditioning", "conditioning_outputs", "config", "def", "dict", "else", "empty", "for", "frame_idx", "get", "if", "in", "inference_session", "int", "is_initial_conditioning_frame", "list", "maskmem_features", "max_cond_frame_num", "non_cond_frame_outputs", "not", "num_maskmem", "obj_idx", "out", "output_data", "output_dict_per_obj", "outputs", "previous_frame_idx", "raise", "range", "relative_temporal_offset", "return", "self", "temporal_positions_and_previous_outputs", "track_in_reverse_time", "tuple", "unselected_conditioning_outputs", "values", "when"], "sam2_video/modeling_sam2_video.py:Sam2VideoModel._build_memory_attention_inputs": ["ModelModel", "ModelPreTrainedModel", "None", "Tensor", "True", "_build_memory_attention_inputs", "append", "class", "combined_memory_pos_embed", "continue", "def", "device", "dict", "for", "if", "in", "int", "is", "list", "maskmem_features", "maskmem_pos_enc", "memories_to_concatenate", "memory_features", "memory_positional_embeddings_to_concatenate", "memory_temporal_positional_encoding", "non_blocking", "prev_output_data", "relative_temporal_offset", "return", "self", "spatial_memory_pos_embed", "temporal_positions_and_previous_outputs", "to", "torch", "tuple"], "sam2_video/modeling_sam2_video.py:Sam2VideoModel._get_object_pointers": ["False", "ModelInferenceSession", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "_get_object_pointers", "and", "append", "bool", "break", "class", "cond_frame_outputs", "conditioning_outputs", "config", "def", "device", "eligible_conditioning_outputs", "else", "for", "frame_idx", "get", "if", "in", "inference_session", "int", "is", "items", "list", "max_object_pointers_in_encoder", "max_object_pointers_to_use", "min", "non_cond_frame_outputs", "not", "num_total_frames", "obj_idx", "object_pointer", "or", "out", "out_data", "output_dict_per_obj", "pointer_tokens", "range", "ref_frame_idx", "return", "self", "streaming", "t_diff_offset", "temporal_difference", "temporal_idx", "temporal_offsets", "temporal_position_sign_multiplier", "to", "torch", "track_in_reverse_time", "training", "tuple"], "sam2_video/modeling_sam2_video.py:Sam2VideoModel._process_object_pointers": ["ModelModel", "ModelPreTrainedModel", "None", "Tensor", "_process_object_pointers", "batch_size", "class", "config", "def", "device", "dim", "dtype", "else", "enable_temporal_pos_encoding_for_object_pointers", "expand", "flatten", "float", "float32", "get_1d_sine_pe", "if", "int", "len", "list", "max_object_pointers_to_use", "max_temporal_diff", "mem_dim", "new_zeros", "normalized_temporal_diffs", "not", "num_channels", "num_splits", "object_pointers", "object_pointers_pos_embed", "permute", "pointer_tokens", "pointer_tpos_dim", "projected_sine_pe", "repeat_interleave", "reshape", "return", "self", "sine_pe", "stack", "temporal_offsets", "temporal_positional_encoding_projection_layer", "tensor", "to", "torch", "tuple", "unsqueeze"], "sam2_video/modeling_sam2_video.py:Sam2VideoModel._prepare_memory_conditioned_features": ["False", "ModelInferenceSession", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "_build_memory_attention_inputs", "_gather_memory_frame_outputs", "_get_object_pointers", "_prepare_memory_conditioned_features", "_process_object_pointers", "append", "backbone_feature_sizes", "batch_size", "bool", "cat", "class", "combined_memory", "combined_memory_positional_embeddings", "conditioned_feature_map", "conditioned_feature_map_flat", "current_feature_map", "current_vision_features", "current_vision_position_embeddings", "current_vision_positional_embeddings", "def", "device", "dim", "frame_idx", "height", "hidden_dim", "if", "inference_session", "int", "is", "is_initial_conditioning_frame", "list", "max_object_pointers_to_use", "memories_to_concatenate", "memory", "memory_attention", "memory_posision_embeddings", "memory_positional_embeddings_to_concatenate", "no_memory_embedding", "not", "num_channels", "num_maskmem", "num_object_pointer_tokens", "num_total_frames", "obj_idx", "object_pointers", "object_pointers_pos_embed", "permute", "pointer_tokens", "return", "self", "shape", "size", "squeeze", "streaming", "temporal_offsets", "temporal_positions_and_previous_outputs", "torch", "track_in_reverse_time", "view", "width"], "sam2_video/modeling_sam2_video.py:Sam2VideoModel._use_multimask": ["ModelModel", "ModelPreTrainedModel", "None", "_use_multimask", "and", "bool", "class", "config", "def", "dict", "else", "if", "is", "is_init_cond_frame", "multimask_max_pt_num", "multimask_min_pt_num", "multimask_output", "multimask_output_for_tracking", "multimask_output_in_Model", "num_pts", "or", "point_inputs", "point_labels", "return", "self", "size"], "sam2_video/modeling_sam2_video.py:Sam2VideoModel._run_single_frame_inference": ["Any", "False", "ModelInferenceSession", "ModelModel", "ModelPreTrainedModel", "Model_outputs", "Modele", "None", "Tensor", "ValueError", "_prepare_memory_conditioned_features", "_prepare_vision_features", "_run_single_frame_inference", "_single_frame_forward", "_use_mask_as_output", "_use_multimask", "and", "appear", "as", "backbone_feature_sizes", "batch_size", "bool", "class", "current_out", "current_vision_feats", "current_vision_features", "current_vision_pos_embeds", "current_vision_positional_embeddings", "def", "dict", "else", "for", "frame", "frame_idx", "hidden_dim", "high_res_features", "high_res_masks", "if", "image_embeddings", "in", "inference_session", "input", "input_labels", "input_masks", "input_points", "int", "is", "is_init_cond_frame", "is_initial_conditioning_frame", "len", "mask_inputs", "multimask_output", "not", "num_frames", "num_total_frames", "obj_idx", "object_pointer", "object_score_logits", "on", "permute", "pix_feat", "pixel_values", "point_coords", "point_inputs", "point_labels", "pred_masks", "prev_Model_mask_logits", "raise", "return", "reverse", "s", "self", "should", "simultaneously", "size", "str", "streaming", "the", "torch", "track_in_reverse_time", "training", "view", "x", "zip"], "sam2_video/modeling_sam2_video.py:Sam2VideoModel._encode_new_memory": ["F", "False", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "True", "_encode_new_memory", "align_corners", "and", "antialias", "backbone_feature_sizes", "batch_size", "bfloat16", "bilinear", "bool", "channels", "class", "config", "current_vision_feats", "def", "dtype", "else", "expand", "flatten", "float", "height", "hidden_dim", "if", "interpolate", "is", "is_mask_from_pts", "is_obj_appearing", "list", "mask_for_mem", "mask_input_size", "mask_input_size_h", "mask_input_size_w", "mask_mem_size_h", "mask_mem_size_w", "maskmem_features", "maskmem_pos_enc", "memory_encoder", "mode", "not", "object_score_logits", "occlusion_spatial_embedding_parameter", "permute", "pix_feat", "pred_masks_high_res", "prompt_encoder", "return", "self", "shape", "sigmoid", "sigmoid_bias_for_mem_enc", "sigmoid_scale_for_mem_enc", "size", "to", "torch", "training", "tuple", "view", "width"], "sam2_video/modeling_sam2_video.py:Sam2VideoModel._batch_encode_memories": ["ModelInferenceSession", "ModelModel", "ModelPreTrainedModel", "Tensor", "_", "_batch_encode_memories", "_encode_new_memory", "_prepare_vision_features", "any", "batch_size", "bool", "cat", "class", "cond_frame_outputs", "current_vision_feats", "def", "dim", "else", "enumerate", "expand", "expanded_vision_feats", "for", "frame_idx", "high_res_masks_batched", "high_res_masks_for_memory", "i", "if", "in", "inference_session", "int", "is_mask_from_pts", "is_mask_from_pts_per_obj", "len", "list", "maskmem_features", "maskmem_features_batched", "maskmem_pos_enc", "maskmem_pos_enc_batched", "non_cond_frame_outputs", "not", "obj_idx", "object_score_logits", "object_score_logits_batched", "object_score_logits_for_memory", "objects_needing_memory_encoding", "output_dict", "output_dict_per_obj", "pred_masks_high_res", "return", "self", "storage_key", "torch"], "sam2_video/modeling_sam2_video.py:Sam2VideoModel.propagate_in_video_iterator": ["Cannot", "False", "Iterator", "ModelInferenceSession", "ModelModel", "ModelPreTrainedModel", "ModelSegmentationOutput", "Model_output", "None", "ValueError", "a", "bool", "class", "cond_frame_outputs", "def", "desc", "determine", "disable", "else", "end_frame_idx", "first", "for", "frame", "frame_idx", "frames_with_inputs", "if", "in", "index", "inference", "inference_session", "inputs", "int", "is", "it", "manually", "max", "max_frame_num_to_track", "min", "not", "num_frames", "obj_output_dict", "on", "or", "output_dict_per_obj", "please", "processing_order", "propagate", "propagate_in_video_iterator", "r", "raise", "range", "reverse", "run", "self", "show_progress_bar", "specify", "start_frame_idx", "starting", "the", "tqdm", "values", "video", "with", "yield"], "sam3/modeling_sam3.py:inverse_sigmoid": ["Model_sigmoid", "Tensor", "clamp", "def", "eps", "float", "log", "max", "min", "return", "torch", "x", "x1", "x2"], "sam3/modeling_sam3.py:concat_padded_sequences": ["False", "Model_padded_sequences", "Modelenated_mask", "Modelenated_sequence", "None", "actual_seq1_lengths", "actual_seq2_lengths", "arange", "assert", "batch_size", "batch_size2", "bool", "def", "device", "dim", "dtype", "expand", "final_lengths", "hidden_size", "hidden_size2", "if", "index", "mask1", "mask2", "max_length", "repeat", "return", "return_index", "scatter", "seq1", "seq1_length", "seq2", "seq2_length", "shape", "size", "sum", "torch", "zeros"], "sam3/modeling_sam3.py:box_cxcywh_to_xyxy": ["Model_cxcywh_to_xyxy", "b", "def", "dim", "h", "return", "stack", "torch", "unbind", "w", "x", "x_c", "y_c"], "sam3/modeling_sam3.py:Sam3MLP.__init__": ["ACT2FN", "Dropout", "Linear", "ModelMLP", "ModelViTConfig", "Module", "__init__", "activation_fn", "class", "config", "def", "dropout", "fc1", "fc2", "hidden_act", "hidden_dropout", "hidden_size", "intermediate_size", "nn", "self", "super"], "sam3/modeling_sam3.py:Sam3MLP.forward": ["ModelMLP", "Module", "Tensor", "activation_fn", "class", "def", "dropout", "fc1", "fc2", "forward", "hidden_states", "nn", "return", "self", "torch"], "sam3/modeling_sam3.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "contiguous", "def", "dim", "dropout", "float", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "shape", "size", "softmax", "torch", "training", "transpose", "value"], "sam3/modeling_sam3.py:Sam3Attention.__init__": ["False", "Linear", "ModelAttention", "Module", "__init__", "class", "config", "def", "head_dim", "hidden_size", "is_causal", "k_proj", "nn", "num_attention_heads", "o_proj", "q_proj", "scaling", "self", "super", "v_proj"], "sam3/modeling_sam3.py:Sam3Attention.forward": ["ALL_ATTENTION_FUNCTIONS", "Attention", "Callable", "Flash", "ModelAttention", "Module", "None", "SDPA", "Tensor", "TransformersKwargs", "Unpack", "_attn_implementation", "additive", "and", "attention", "attention_interface", "attention_mask", "attn_output", "attn_weights", "back", "batch_size", "because", "bias", "bool", "class", "config", "contiguous", "cross", "def", "does", "dropout", "dtype", "eager", "eager_attention_forward", "falling", "flash", "for", "forward", "head_dim", "if", "in", "is", "is_causal", "k_proj", "key", "key_len", "kwargs", "logger", "masks", "nn", "not", "num_attention_heads", "o_proj", "position", "q_proj", "query", "query_len", "relative", "reshape", "return", "scaling", "sdpa", "self", "shape", "support", "to", "torch", "transpose", "tuple", "v_proj", "value", "view", "warning_once"], "sam3/modeling_sam3.py:Sam3ViTRotaryEmbedding.__init__": ["Dimension", "False", "ModelViTConfig", "ModelViTRotaryEmbedding", "Module", "RoPE", "ValueError", "__init__", "arange", "axial", "be", "by", "cat", "class", "config", "cos", "def", "dim", "div", "divisible", "dtype", "end_x", "end_y", "flattened_indices", "float", "floor", "for", "freqs", "freqs_x", "freqs_y", "hidden_size", "if", "int", "inv_freq", "long", "must", "nn", "num_attention_heads", "outer", "persistent", "raise", "register_buffer", "repeat_interleave", "rope_embeddings_cos", "rope_embeddings_sin", "rope_theta", "rounding_mode", "scale", "self", "sin", "super", "torch", "x_positions", "y_positions"], "sam3/modeling_sam3.py:Sam3ViTRotaryEmbedding.forward": ["ModelViTRotaryEmbedding", "Module", "Tensor", "class", "def", "forward", "nn", "return", "rope_embeddings_cos", "rope_embeddings_sin", "self", "torch", "tuple"], "sam3/modeling_sam3.py:rotate_pairwise": ["Model_pairwise", "def", "dim", "flatten", "return", "shape", "stack", "start_dim", "torch", "unbind", "view", "x", "x1", "x2"], "sam3/modeling_sam3.py:apply_rotary_pos_emb_2d": ["Model_rotary_pos_emb_2d", "Tensor", "cos", "def", "float", "k", "k_embed", "q", "q_embed", "return", "rotate_pairwise", "sin", "torch", "tuple", "type_as"], "sam3/modeling_sam3.py:Sam3ViTRoPEAttention.__init__": ["False", "Linear", "ModelViTConfig", "ModelViTRoPEAttention", "Module", "__init__", "attention_dropout", "class", "config", "def", "head_dim", "hidden_size", "is_causal", "k_proj", "nn", "num_attention_heads", "o_proj", "q_proj", "scaling", "self", "super", "v_proj"], "sam3/modeling_sam3.py:Sam3ViTRoPEAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "ModelViTRoPEAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "_attn_implementation", "apply_rotary_pos_emb_2d", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "batch_size", "class", "config", "contiguous", "cos", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "head_dim", "height", "hidden_states", "if", "is_causal", "k_proj", "key", "kwargs", "new_shape", "nn", "not", "num_attention_heads", "o_proj", "position_embeddings", "q_proj", "query", "reshape", "return", "scaling", "self", "seq_len", "shape", "sin", "torch", "training", "transpose", "tuple", "v_proj", "value", "view", "width"], "sam3/modeling_sam3.py:Sam3ViTPatchEmbeddings.__init__": ["Conv2d", "False", "Iterable", "ModelViTConfig", "ModelViTPatchEmbeddings", "Module", "__init__", "bias", "class", "config", "def", "else", "hidden_size", "if", "image_size", "isinstance", "kernel_size", "nn", "num_channels", "num_patches", "patch_size", "pretrain_image_size", "projection", "self", "stride", "super"], "sam3/modeling_sam3.py:Sam3ViTPatchEmbeddings.forward": ["ModelViTPatchEmbeddings", "Module", "Tensor", "class", "def", "dtype", "embeddings", "flatten", "forward", "nn", "pixel_values", "projection", "return", "self", "to", "torch", "transpose", "weight"], "sam3/modeling_sam3.py:Sam3ViTEmbeddings.__init__": ["Dropout", "ModelViTConfig", "ModelViTEmbeddings", "ModelViTPatchEmbeddings", "Module", "Parameter", "__init__", "class", "config", "def", "dropout", "hidden_dropout", "hidden_size", "nn", "num_patches", "patch_embeddings", "patch_size", "position_embeddings", "randn", "self", "super", "torch"], "sam3/modeling_sam3.py:Sam3ViTEmbeddings._tile_position_embeddings": ["ModelViTEmbeddings", "Module", "Tensor", "_tile_position_embeddings", "and", "class", "def", "height", "hidden_size", "if", "int", "is_tracing", "jit", "nn", "not", "permute", "pos_embed", "position_embeddings", "pretrain_size", "repeat_h", "repeat_w", "reshape", "return", "self", "shape", "tile", "torch", "width"], "sam3/modeling_sam3.py:Sam3ViTEmbeddings.forward": ["False", "ModelViTEmbeddings", "Module", "Tensor", "_tile_position_embeddings", "bool", "class", "def", "dropout", "embeddings", "forward", "height", "height_patches", "interpolate_pos_encoding", "nn", "patch_embeddings", "patch_size", "pixel_values", "position_embeddings", "return", "self", "shape", "torch", "width", "width_patches"], "sam3/modeling_sam3.py:window_partition": ["Model_partition", "Model_size", "Models", "batch_size", "contiguous", "def", "functional", "height", "hidden_state", "nn", "num_channels", "pad", "pad_height", "pad_width", "padded_height", "padded_width", "permute", "return", "shape", "view", "width"], "sam3/modeling_sam3.py:window_unpartition": ["Model_size", "Model_unpartition", "Models", "batch_size", "contiguous", "def", "height", "height_width", "hidden_state", "pad_height_width", "padded_height", "padded_width", "permute", "return", "shape", "view", "width"], "sam3/modeling_sam3.py:Sam3ViTLayerScale.__init__": ["ModelViTLayerScale", "Module", "None", "Parameter", "__init__", "class", "config", "def", "hidden_size", "lambda1", "layer_scale_init_value", "nn", "ones", "self", "super", "torch"], "sam3/modeling_sam3.py:Sam3ViTLayerScale.forward": ["ModelViTLayerScale", "Module", "Tensor", "class", "def", "forward", "hidden_state", "lambda1", "nn", "return", "self", "torch"], "sam3/modeling_sam3.py:Sam3ViTLayer.__init__": ["Dropout", "GradientCheckpointingLayer", "LayerNorm", "ModelMLP", "ModelViTConfig", "ModelViTLayer", "ModelViTRoPEAttention", "ModelViTRotaryEmbedding", "None", "__init__", "attention", "class", "config", "def", "dropout", "else", "end_x", "end_y", "eps", "hidden_dropout", "hidden_size", "if", "image_size", "input_size", "int", "isinstance", "layer_norm1", "layer_norm2", "layer_norm_eps", "list", "mlp", "nn", "patch_size", "rotary_emb", "rotary_input_size", "rotary_scale", "scale", "self", "super", "tuple", "window_size"], "sam3/modeling_sam3.py:Sam3ViTLayer.forward": ["GradientCheckpointingLayer", "ModelViTLayer", "Tensor", "TransformersKwargs", "Unpack", "_", "attention", "class", "def", "dropout", "forward", "height", "hidden_states", "if", "kwargs", "layer_norm1", "layer_norm2", "mlp", "pad_height_width", "position_embeddings", "residual", "return", "rotary_emb", "self", "shape", "torch", "width", "window_partition", "window_size", "window_unpartition"], "sam3/modeling_sam3.py:Sam3PreTrainedModel._init_weights": ["ModelPreTrainedModel", "ModelViTEmbeddings", "ModelViTRotaryEmbedding", "PreTrainedModel", "_init_weights", "arange", "cat", "class", "config", "copy_", "cos", "def", "dim", "div", "dtype", "elif", "end_x", "end_y", "flattened_indices", "float", "floor", "freqs", "freqs_x", "freqs_y", "if", "init", "initializer_range", "inv_freq", "isinstance", "long", "mean", "module", "normal_", "outer", "position_embeddings", "repeat_interleave", "rope_embeddings_cos", "rope_embeddings_sin", "rope_theta", "rounding_mode", "scale", "self", "sin", "std", "super", "torch", "x_positions", "y_positions"], "sam3/modeling_sam3.py:Sam3ViTModel.__init__": ["LayerNorm", "ModelPreTrainedModel", "ModelViTConfig", "ModelViTEmbeddings", "ModelViTLayer", "ModelViTModel", "ModuleList", "__init__", "class", "config", "def", "else", "embeddings", "eps", "for", "global_attn_indexes", "hidden_size", "i", "if", "in", "layer_norm", "layer_norm_eps", "layers", "nn", "not", "num_hidden_layers", "post_init", "range", "self", "super", "window_size"], "sam3/modeling_sam3.py:Sam3ViTModel.get_input_embeddings": ["ModelPreTrainedModel", "ModelViTModel", "ModelViTPatchEmbeddings", "class", "def", "embeddings", "get_input_embeddings", "patch_embeddings", "return", "self"], "sam3/modeling_sam3.py:Sam3ViTModel.forward": ["BaseModelOutput", "ModelPreTrainedModel", "ModelViTModel", "Tensor", "TransformersKwargs", "Unpack", "batch_size", "class", "config", "def", "embeddings", "for", "forward", "height", "hidden_size", "hidden_states", "in", "kwargs", "last_hidden_state", "layer", "layer_norm", "layers", "patch_size", "pixel_values", "return", "self", "shape", "torch", "view", "width"], "sam3/modeling_sam3.py:Sam3SinePositionEmbedding.__init__": ["False", "ModelSinePositionEmbedding", "Module", "None", "True", "ValueError", "__init__", "and", "be", "bool", "class", "def", "else", "float", "if", "int", "is", "math", "nn", "normalize", "not", "num_pos_feats", "passed", "pi", "raise", "scale", "self", "should", "super", "temperature"], "sam3/modeling_sam3.py:Sam3SinePositionEmbedding.encode_1d_positions": ["ModelSinePositionEmbedding", "Module", "None", "Tensor", "arange", "class", "cos", "def", "device", "dim", "dim_t", "dtype", "encode_1d_positions", "flatten", "int64", "nn", "num_pos_feats", "pos_x", "pos_y", "return", "scale", "self", "sin", "stack", "temperature", "to", "torch", "tuple", "x", "x_embed", "y", "y_embed"], "sam3/modeling_sam3.py:Sam3SinePositionEmbedding.encode_boxes": ["Expected", "ModelSinePositionEmbedding", "Module", "None", "Tensor", "arange", "assert", "box", "boxes", "cat", "class", "coordinates", "cos", "def", "device", "dim", "dim_t", "div", "dtype", "encode_boxes", "f", "flatten", "floor", "got", "h", "h_embed", "int64", "nn", "num_pos_feats", "pos", "pos_h", "pos_w", "pos_x", "pos_y", "return", "rounding_mode", "scale", "self", "shape", "sin", "size", "stack", "temperature", "to", "torch", "w", "w_embed", "x", "x_embed", "y", "y_embed"], "sam3/modeling_sam3.py:Sam3SinePositionEmbedding.forward": ["ModelSinePositionEmbedding", "Module", "None", "Size", "Tensor", "arange", "bool", "cat", "class", "cos", "cumsum", "def", "device", "dim", "dim_t", "div", "dtype", "eps", "flatten", "floor", "forward", "if", "int64", "is", "mask", "nn", "normalize", "not_mask", "num_pos_feats", "permute", "pos", "pos_x", "pos_y", "return", "rounding_mode", "scale", "self", "shape", "sin", "stack", "str", "temperature", "to", "torch", "x_embed", "y_embed", "zeros"], "sam3/modeling_sam3.py:Sam3FPNLayer.__init__": ["Conv2d", "ConvTranspose2d", "GELU", "MaxPool2d", "ModelFPNLayer", "Module", "ModuleList", "NotImplementedError", "__init__", "append", "class", "def", "elif", "else", "f", "float", "fpn_dim", "if", "in_channels", "int", "intermediate_channels", "is", "kernel_size", "nn", "not", "out_channels", "padding", "proj1", "proj2", "raise", "scale_factor", "scale_layers", "self", "stride", "super", "supported", "yet"], "sam3/modeling_sam3.py:Sam3FPNLayer.forward": ["ModelFPNLayer", "Module", "Tensor", "class", "def", "dtype", "for", "forward", "hidden_states", "in", "layer", "nn", "proj1", "proj2", "return", "scale_layers", "self", "to", "torch", "weight"], "sam3/modeling_sam3.py:Sam3VisionNeck.__init__": ["ModelFPNLayer", "ModelSinePositionEmbedding", "ModelVisionConfig", "ModelVisionNeck", "Module", "ModuleList", "True", "__init__", "backbone_config", "class", "config", "def", "for", "fpn_dim", "fpn_hidden_size", "fpn_layers", "hidden_size", "in", "in_channels", "nn", "normalize", "num_pos_feats", "position_encoding", "scale", "scale_factor", "scale_factors", "self", "super"], "sam3/modeling_sam3.py:Sam3VisionNeck.forward": ["ModelVisionNeck", "Module", "Tensor", "class", "def", "device", "dtype", "for", "forward", "fpn_hidden_states", "fpn_layer", "fpn_layers", "fpn_output", "fpn_position_encoding", "hidden_states", "in", "nn", "pos_enc", "position_encoding", "return", "self", "shape", "torch", "tuple"], "sam3/modeling_sam3.py:Sam3VisionModel.__init__": ["AutoModel", "ModelPreTrainedModel", "ModelVisionConfig", "ModelVisionModel", "ModelVisionNeck", "__init__", "backbone", "backbone_config", "class", "config", "def", "from_config", "neck", "post_init", "self", "super"], "sam3/modeling_sam3.py:Sam3VisionModel.get_input_embeddings": ["ModelPreTrainedModel", "ModelVisionModel", "backbone", "class", "def", "get_input_embeddings", "return", "self"], "sam3/modeling_sam3.py:Sam3VisionModel.forward": ["FloatTensor", "ModelPreTrainedModel", "ModelVisionEncoderOutput", "ModelVisionModel", "None", "TransformersKwargs", "Unpack", "ValueError", "You", "backbone", "backbone_config", "backbone_output", "batch_size", "class", "config", "def", "forward", "fpn_hidden_states", "fpn_position_encoding", "have", "height", "hidden_states", "hidden_states_spatial", "if", "is", "kwargs", "last_hidden_state", "neck", "patch_size", "permute", "pixel_values", "raise", "return", "self", "shape", "specify", "to", "torch", "tuple", "view", "width"], "sam3/modeling_sam3.py:Sam3GeometryEncoderLayer.__init__": ["Dropout", "LayerNorm", "ModelAttention", "ModelGeometryEncoderConfig", "ModelGeometryEncoderLayer", "ModelMLP", "Module", "__init__", "class", "config", "cross_attn", "def", "dropout", "hidden_size", "layer_norm1", "layer_norm2", "layer_norm3", "mlp", "nn", "self", "self_attn", "super"], "sam3/modeling_sam3.py:Sam3GeometryEncoderLayer.forward": ["ModelGeometryEncoderLayer", "Module", "Tensor", "TransformersKwargs", "Unpack", "_", "attention_mask", "class", "cross_attn", "def", "dropout", "forward", "hidden_states", "key", "kwargs", "layer_norm1", "layer_norm2", "layer_norm3", "mlp", "nn", "prompt_feats", "prompt_mask", "query", "residual", "return", "self", "self_attn", "value", "vision_feats", "vision_pos_encoding"], "sam3/modeling_sam3.py:Sam3GeometryEncoder.__init__": ["Conv2d", "Embedding", "LayerNorm", "Linear", "ModelGeometryEncoder", "ModelGeometryEncoderConfig", "ModelGeometryEncoderLayer", "ModelSinePositionEmbedding", "Module", "ModuleList", "True", "_", "__init__", "boxes_direct_project", "boxes_pool_project", "boxes_pos_enc_project", "class", "cls_embed", "config", "def", "final_proj", "for", "hidden_size", "in", "label_embed", "layers", "nn", "normalize", "num_layers", "num_pos_feats", "output_layer_norm", "position_encoding", "prompt_layer_norm", "range", "roi_size", "self", "super", "vision_layer_norm"], "sam3/modeling_sam3.py:Sam3GeometryEncoder._encode_box_coordinates": ["ModelGeometryEncoder", "Module", "None", "Tensor", "_encode_box_coordinates", "cat", "center_x", "center_y", "class", "def", "dim", "encode_1d_positions", "height", "nn", "pos", "pos_x", "pos_y", "position_encoding", "return", "self", "torch", "width"], "sam3/modeling_sam3.py:Sam3GeometryEncoder._encode_boxes": ["ModelGeometryEncoder", "Modelpled_features", "Module", "_encode_box_coordinates", "_encode_boxes", "batch_size", "bfloat16", "box_cxcywh_to_xyxy", "box_height", "box_width", "boxes", "boxes_direct_project", "boxes_embed", "boxes_labels", "boxes_mask", "boxes_pool_project", "boxes_pos_enc_project", "boxes_xyxy", "center_x", "center_y", "class", "def", "device", "dtype", "else", "flatten", "float16", "height", "hidden_size", "if", "label_embed", "long", "nn", "num_boxes", "ops", "pooled_projection", "pos_enc", "pos_projection", "return", "roi_align", "roi_size", "scale", "self", "shape", "tensor", "to", "torch", "torchvision", "unbind", "view", "vision_features", "width"], "sam3/modeling_sam3.py:Sam3GeometryEncoder.forward": ["ModelGeometryEncoder", "ModelGeometryEncoderOutput", "Module", "None", "Tensor", "_encode_boxes", "attention_mask", "batch_size", "box_embeddings", "box_labels", "box_mask", "class", "cls_embed", "cls_mask", "concat_padded_sequences", "config", "create_bidirectional_mask", "def", "device", "dtype", "else", "expand", "final_proj", "flatten", "for", "forward", "hidden_size", "if", "img_feats", "img_feats_last", "img_pos_embeds", "in", "input_embeds", "is", "last_hidden_state", "layer", "layers", "nn", "normalized_img_feats", "not", "ones", "output_layer_norm", "permute", "prompt_attention_mask", "prompt_embeds", "prompt_feats", "prompt_layer_norm", "prompt_mask", "return", "self", "shape", "torch", "transpose", "tuple", "unsqueeze", "view", "vision_feats", "vision_feats_flat", "vision_layer_norm", "vision_pos_embeds", "vision_pos_embeds_flat", "vision_pos_encoding", "weight", "zeros_like"], "sam3/modeling_sam3.py:Sam3DetrEncoderLayer.__init__": ["Dropout", "LayerNorm", "ModelAttention", "ModelDETREncoderConfig", "ModelDetrEncoderLayer", "ModelMLP", "Module", "__init__", "class", "config", "cross_attn", "def", "dropout", "hidden_size", "layer_norm1", "layer_norm2", "layer_norm3", "mlp", "nn", "self", "self_attn", "super"], "sam3/modeling_sam3.py:Sam3DetrEncoderLayer.forward": ["ModelDetrEncoderLayer", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "attention_mask", "class", "cross_attn", "def", "dropout", "forward", "hidden_states", "hidden_states_with_pos", "key", "kwargs", "layer_norm1", "layer_norm2", "layer_norm3", "mlp", "nn", "prompt_cross_attn_mask", "prompt_feats", "query", "residual", "return", "self", "self_attn", "value", "vision_feats", "vision_pos_encoding"], "sam3/modeling_sam3.py:Sam3DetrEncoder.__init__": ["ModelDETREncoderConfig", "ModelDetrEncoder", "ModelDetrEncoderLayer", "ModelPreTrainedModel", "ModuleList", "_", "__init__", "class", "config", "def", "for", "hidden_size", "in", "layers", "nn", "num_layers", "post_init", "range", "self", "super"], "sam3/modeling_sam3.py:Sam3DetrEncoder._prepare_multilevel_features": ["ModelDetrEncoder", "ModelPreTrainedModel", "Tensor", "_prepare_multilevel_features", "append", "cat", "class", "def", "device", "dim", "dtype", "features", "features_flattened", "flatten", "for", "height", "in", "list", "long", "pos_embed", "pos_embeds_flattened", "return", "self", "shape", "spatial_shapes", "tensor", "torch", "transpose", "vision_features", "vision_pos_embeds", "width", "zip"], "sam3/modeling_sam3.py:Sam3DetrEncoder.forward": ["ModelDETREncoderOutput", "ModelDetrEncoder", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "_prepare_multilevel_features", "attention_mask", "batch_size", "class", "config", "create_bidirectional_mask", "def", "dim", "else", "encoder_hidden_states", "enumerate", "features_flattened", "for", "forward", "height", "hidden_states", "i", "if", "in", "input_embeds", "int", "is", "kwargs", "last_hidden_state", "layer", "layers", "list", "not", "permute", "pos_embeds_flattened", "prompt_cross_attn_mask", "prompt_feats", "reshape", "return", "self", "shape", "spatial_shapes", "spatial_sizes", "text_features", "text_mask", "torch", "tuple", "vision_features", "vision_pos_embeds", "vision_pos_encoding", "width"], "sam3/modeling_sam3.py:Sam3DecoderMLP.__init__": ["Linear", "ModelDecoderMLP", "Module", "None", "Only", "ValueError", "__init__", "class", "def", "elif", "else", "f", "got", "hidden_dim", "if", "input_dim", "int", "layer1", "layer2", "layer3", "layers", "nn", "num_layers", "or", "output_dim", "raise", "self", "super", "supported"], "sam3/modeling_sam3.py:Sam3DecoderMLP.forward": ["F", "ModelDecoderMLP", "Module", "None", "Tensor", "class", "def", "else", "forward", "if", "is", "layer1", "layer2", "layer3", "nn", "not", "relu", "return", "self", "torch", "x"], "sam3/modeling_sam3.py:Sam3DetrDecoderLayer.__init__": ["Dropout", "LayerNorm", "ModelAttention", "ModelDETRDecoderConfig", "ModelDetrDecoderLayer", "ModelMLP", "Module", "__init__", "class", "config", "def", "dropout", "hidden_size", "mlp", "mlp_dropout", "mlp_layer_norm", "nn", "self", "self_attn", "self_attn_dropout", "self_attn_layer_norm", "super", "text_cross_attn", "text_cross_attn_dropout", "text_cross_attn_layer_norm", "vision_cross_attn", "vision_cross_attn_dropout", "vision_cross_attn_layer_norm"], "sam3/modeling_sam3.py:Sam3DetrDecoderLayer.forward": ["F", "ModelDetrDecoderLayer", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "attention_mask", "attn_output", "class", "constant", "def", "forward", "hidden_states", "key", "key_with_pos", "kwargs", "mlp", "mlp_dropout", "mlp_layer_norm", "mode", "nn", "pad", "query", "query_pos", "query_with_pos", "residual", "return", "self", "self_attn", "self_attn_dropout", "self_attn_layer_norm", "text_cross_attn", "text_cross_attn_dropout", "text_cross_attn_layer_norm", "text_cross_attn_mask", "text_features", "torch", "value", "vision_cross_attn", "vision_cross_attn_dropout", "vision_cross_attn_layer_norm", "vision_cross_attn_mask", "vision_features", "vision_pos_encoding"], "sam3/modeling_sam3.py:Sam3DetrDecoder.__init__": ["Embedding", "False", "LayerNorm", "ModelDETRDecoderConfig", "ModelDecoderMLP", "ModelDetrDecoder", "ModelDetrDecoderLayer", "ModelPreTrainedModel", "ModelSinePositionEmbedding", "ModuleList", "_", "__init__", "box_head", "box_rpb_embed_x", "box_rpb_embed_y", "clamp_presence_logit_max_val", "class", "config", "def", "for", "hidden_size", "in", "layers", "nn", "normalize", "num_attention_heads", "num_layers", "num_pos_feats", "num_queries", "output_layer_norm", "position_encoding", "post_init", "presence_head", "presence_layer_norm", "presence_token", "query_embed", "range", "ref_point_head", "reference_points", "self", "super"], "sam3/modeling_sam3.py:Sam3DetrDecoder._get_coords": ["ModelDetrDecoder", "ModelPreTrainedModel", "Tensor", "_get_coords", "arange", "class", "coords_h", "coords_w", "def", "device", "dtype", "height", "return", "self", "torch", "tuple", "width"], "sam3/modeling_sam3.py:Sam3DetrDecoder._get_rpb_matrix": ["ModelDetrDecoder", "ModelPreTrainedModel", "Tensor", "_", "_get_coords", "_get_rpb_matrix", "abs", "batch_size", "box_cxcywh_to_xyxy", "box_rpb_embed_x", "box_rpb_embed_y", "boxes_xyxy", "class", "contiguous", "coords_h", "coords_w", "def", "deltas_x", "deltas_x_log", "deltas_y", "deltas_y_log", "device", "dtype", "flatten", "height", "log2", "math", "num_queries", "permute", "reference_boxes", "reshape", "return", "rpb_matrix", "self", "shape", "sign", "spatial_shape", "torch", "tuple", "unsqueeze", "view", "width"], "sam3/modeling_sam3.py:Sam3DetrDecoder.forward": ["F", "ModelDETRDecoderOutput", "ModelDetrDecoder", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "_get_rpb_matrix", "and", "append", "attention_mask", "batch_size", "box_head", "cat", "clamp", "clamp_presence_logit_max_val", "class", "config", "constant", "create_bidirectional_mask", "def", "delta_boxes", "detach", "dim", "encode_boxes", "encoder_hidden_states", "expand", "for", "forward", "hidden_states", "if", "in", "input_embeds", "intermediate_boxes", "intermediate_hidden_states", "intermediate_outputs", "intermediate_presence_logits", "inverse_sigmoid", "is", "kwargs", "layer", "layers", "max", "min", "mode", "new_reference_boxes", "not", "output_layer_norm", "pad", "position_encoding", "presence_head", "presence_hidden", "presence_layer_norm", "presence_logits", "presence_token", "query_embed", "query_embeds", "query_hidden_states", "query_pos", "query_sine_embed", "ref_point_head", "reference_boxes", "reference_boxes_before_sigmoid", "reference_points", "reference_points_input", "return", "rpb_matrix", "self", "shape", "sigmoid", "spatial_shape", "spatial_shapes", "squeeze", "stack", "text_cross_attn_mask", "text_features", "text_mask", "torch", "tuple", "unsqueeze", "value", "vision_cross_attn_mask", "vision_features", "vision_pos_encoding", "weight"], "sam3/modeling_sam3.py:Sam3DotProductScoring.__init__": ["Dropout", "LayerNorm", "Linear", "ModelConfig", "ModelDecoderMLP", "ModelDotProductScoring", "Module", "True", "__init__", "clamp_logits", "clamp_max_val", "class", "config", "def", "detr_decoder_config", "dropout", "float", "hidden_dim", "hidden_size", "input_dim", "intermediate_size", "nn", "np", "num_layers", "output_dim", "projection_dim", "query_proj", "scale", "self", "sqrt", "super", "text_mlp", "text_mlp_dropout", "text_mlp_out_norm", "text_proj"], "sam3/modeling_sam3.py:Sam3DotProductScoring._pool_text_features": ["ModelDotProductScoring", "Module", "None", "Tensor", "_pool_text_features", "clamp", "class", "def", "dim", "dtype", "if", "is", "is_valid", "mean", "min", "nn", "num_valid", "pooled_text", "return", "self", "sum", "text_features", "text_mask", "to", "torch", "unsqueeze"], "sam3/modeling_sam3.py:Sam3DotProductScoring.forward": ["ModelDotProductScoring", "Module", "None", "Tensor", "_pool_text_features", "clamp", "clamp_logits", "clamp_max_val", "class", "decoder_hidden_states", "def", "forward", "if", "matmul", "max", "min", "nn", "orig_text_features", "pooled_text", "proj_queries", "proj_text", "query_proj", "return", "scale", "scores", "self", "text_features", "text_mask", "text_mlp", "text_mlp_dropout", "text_mlp_out_norm", "text_proj", "torch", "unsqueeze"], "sam3/modeling_sam3.py:Sam3MaskEmbedder.__init__": ["Linear", "ModelMaskDecoderConfig", "ModelMaskEmbedder", "Module", "ModuleList", "ReLU", "__init__", "activation", "class", "config", "def", "hidden_size", "layers", "nn", "self", "super"], "sam3/modeling_sam3.py:Sam3MaskEmbedder.forward": ["ModelMaskEmbedder", "Module", "Tensor", "activation", "class", "def", "enumerate", "for", "forward", "hidden_states", "i", "if", "in", "layer", "layers", "len", "nn", "queries", "return", "self", "torch"], "sam3/modeling_sam3.py:Sam3PixelDecoder.__init__": ["Conv2d", "GroupNorm", "ModelMaskDecoderConfig", "ModelPixelDecoder", "Module", "ModuleList", "_", "__init__", "class", "config", "conv_layers", "def", "for", "hidden_size", "in", "kernel_size", "nn", "norms", "num_upModelpling_stages", "out_channels", "padding", "range", "self", "stride", "super"], "sam3/modeling_sam3.py:Sam3PixelDecoder.forward": ["F", "ModelPixelDecoder", "Module", "Tensor", "backbone_feat", "backbone_features", "class", "conv_layers", "def", "enumerate", "for", "forward", "in", "interpolate", "layer_idx", "list", "mode", "nearest", "nn", "norms", "prev_fpn", "relu", "return", "reversed", "self", "shape", "size", "torch"], "sam3/modeling_sam3.py:Sam3MaskDecoder.__init__": ["Conv2d", "Dropout", "LayerNorm", "ModelAttention", "ModelMaskDecoder", "ModelMaskDecoderConfig", "ModelMaskEmbedder", "ModelPixelDecoder", "ModelPreTrainedModel", "__init__", "class", "config", "def", "dropout", "hidden_size", "instance_projection", "kernel_size", "mask_embedder", "nn", "out_channels", "pixel_decoder", "post_init", "prompt_cross_attn", "prompt_cross_attn_dropout", "prompt_cross_attn_norm", "self", "semantic_projection", "super"], "sam3/modeling_sam3.py:Sam3MaskDecoder.forward": ["ModelMaskDecoder", "ModelMaskDecoderOutput", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "_embed_pixels", "attention_mask", "attn_output", "backbone_features", "bchw", "bqc", "bqhw", "class", "config", "create_bidirectional_mask", "cross_attn_mask", "decoder_queries", "def", "dict", "einsum", "encoder_hidden_states", "forward", "if", "input_embeds", "instance_embeds", "instance_projection", "is", "key", "kwargs", "list", "mask_embedder", "mask_embeddings", "normed_hidden_states", "not", "pixel_embed", "pred_masks", "prompt_cross_attn", "prompt_cross_attn_dropout", "prompt_cross_attn_norm", "prompt_features", "prompt_mask", "query", "residual", "return", "self", "semantic_projection", "semantic_seg", "str", "torch", "value"], "sam3/modeling_sam3.py:Sam3MaskDecoder._embed_pixels": ["ModelMaskDecoder", "ModelPreTrainedModel", "Tensor", "_", "_embed_pixels", "backbone_features", "backbone_visual_feats", "batch_size", "class", "clone", "def", "encoder_hidden_states", "encoder_visual_embed", "feat", "for", "height", "hidden_size", "in", "list", "pixel_decoder", "pixel_embed", "reshape", "return", "self", "shape", "spatial_dim", "torch", "transpose", "width"], "sam3/modeling_sam3.py:Sam3Model.__init__": ["CLIPTextModelWithProjection", "Linear", "ModelConfig", "ModelDetrDecoder", "ModelDetrEncoder", "ModelDotProductScoring", "ModelGeometryEncoder", "ModelMaskDecoder", "ModelModel", "ModelPreTrainedModel", "ModelVisionModel", "None", "__init__", "_attn_implementation", "and", "class", "config", "def", "detector_config", "detr_decoder", "detr_decoder_config", "detr_encoder", "detr_encoder_config", "dict", "dot_product_scoring", "geometry_encoder", "geometry_encoder_config", "hasattr", "hidden_size", "if", "is", "isinstance", "mask_decoder", "mask_decoder_config", "nn", "not", "post_init", "self", "super", "text_config", "text_encoder", "text_projection", "vision_config", "vision_encoder", "vocab_size"], "sam3/modeling_sam3.py:Sam3Model.get_text_features": ["FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "class", "def", "get_text_features", "input_ids", "kwargs", "last_hidden_state", "r", "return", "self", "text_encoder", "text_features", "text_projection", "torch"], "sam3/modeling_sam3.py:Sam3Model.get_vision_features": ["FloatTensor", "ModelModel", "ModelPreTrainedModel", "ModelVisionEncoderOutput", "TransformersKwargs", "Unpack", "class", "def", "get_vision_features", "kwargs", "pixel_values", "r", "return", "self", "torch", "vision_encoder", "vision_outputs"], "sam3/modeling_sam3.py:Sam3Model.forward": ["FloatTensor", "LongTensor", "ModelImageSegmentationOutput", "ModelModel", "ModelPreTrainedModel", "ModelVisionEncoderOutput", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "all_box_offsets", "all_pred_boxes", "all_pred_boxes_cxcywh", "all_pred_logits", "and", "attention_mask", "attentions", "backbone_features", "batch_size", "bool", "box_cxcywh_to_xyxy", "box_embeddings", "box_head", "box_labels", "box_mask", "cat", "class", "combined_prompt_features", "combined_prompt_mask", "decoder_hidden_states", "decoder_outputs", "decoder_queries", "decoder_reference_boxes", "def", "detr_decoder", "detr_decoder_attentions", "detr_encoder", "detr_encoder_attentions", "device", "dim", "dot_product_scoring", "dtype", "elif", "else", "encoder_hidden_states", "encoder_outputs", "exactly", "forward", "fpn_hidden_states", "fpn_position_encoding", "geo_valid_mask", "geometry_encoder", "geometry_outputs", "geometry_prompt_features", "geometry_prompt_mask", "get_text_features", "has_geometry_prompts", "hidden_states", "if", "img_feats", "img_pos_embeds", "input_boxes", "input_boxes_labels", "input_ids", "intermediate_hidden_states", "inverse_sigmoid", "is", "kwargs", "last_hidden_state", "list", "long", "mask_decoder", "mask_decoder_attentions", "mask_outputs", "must", "not", "numel", "of", "one", "ones", "ones_like", "or", "pixel_values", "pos_embeds_flattened", "pred_boxes", "pred_logits", "pred_masks", "presence_logits", "prompt_features", "prompt_mask", "r", "raise", "reference_boxes", "reference_boxes_inv_sig", "repeat", "return", "self", "semantic_seg", "shape", "sigmoid", "spatial_shapes", "specify", "squeeze", "text_embeds", "text_features", "text_mask", "text_valid_mask", "torch", "vision_attentions", "vision_embeds", "vision_encoder", "vision_features", "vision_hidden_states", "vision_outputs", "vision_pos_embeds", "vision_pos_encoding", "where", "zeros"], "sam3_tracker/modeling_sam3_tracker.py:Sam3TrackerFeedForward.__init__": ["ACT2FN", "False", "Linear", "ModelFeedForward", "Module", "ModuleList", "_", "__init__", "activation", "bool", "class", "def", "for", "hidden_dim", "in", "input_dim", "int", "layers", "nn", "num_layers", "output_dim", "proj_in", "proj_out", "range", "relu", "self", "sigmoid_output", "str", "super"], "sam3_tracker/modeling_sam3_tracker.py:Sam3TrackerFeedForward.forward": ["F", "ModelFeedForward", "Module", "activation", "class", "def", "for", "forward", "hidden_states", "if", "in", "layer", "layers", "nn", "proj_in", "proj_out", "return", "self", "sigmoid", "sigmoid_output"], "sam3_tracker/modeling_sam3_tracker.py:Sam3TrackerPreTrainedModel._init_weights": ["ModelModel", "ModelPositionalEmbedding", "ModelPreTrainedModel", "None", "PreTrainedModel", "_init_weights", "class", "def", "elif", "if", "init", "is", "isinstance", "module", "no_memory_embedding", "normal_", "not", "positional_embedding", "scale", "self", "std", "super", "zeros_"], "sam3_tracker/modeling_sam3_tracker.py:Sam3TrackerPositionalEmbedding.__init__": ["ModelPositionalEmbedding", "ModelPromptEncoderConfig", "Module", "__init__", "class", "config", "def", "hidden_size", "nn", "positional_embedding", "randn", "register_buffer", "scale", "self", "super", "torch"], "sam3_tracker/modeling_sam3_tracker.py:Sam3TrackerPositionalEmbedding.forward": ["ModelPositionalEmbedding", "Module", "None", "cat", "class", "clone", "coordinates", "cos", "def", "dim", "dtype", "float32", "forward", "if", "input_coords", "input_shape", "is", "nn", "not", "np", "pi", "positional_embedding", "return", "self", "sin", "to", "torch"], "sam3_tracker/modeling_sam3_tracker.py:Sam3TrackerMaskEmbedding.__init__": ["ACT2FN", "Conv2d", "ModelLayerNorm", "ModelMaskEmbedding", "ModelPromptEncoderConfig", "Module", "__init__", "activation", "channels_first", "class", "config", "conv1", "conv2", "conv3", "data_format", "def", "eps", "hidden_act", "hidden_size", "kernel_size", "layer_norm1", "layer_norm2", "layer_norm_eps", "mask_input_channels", "nn", "self", "stride", "super"], "sam3_tracker/modeling_sam3_tracker.py:Sam3TrackerMaskEmbedding.forward": ["ModelMaskEmbedding", "Module", "activation", "class", "conv1", "conv2", "conv3", "def", "dense_embeddings", "forward", "hidden_states", "layer_norm1", "layer_norm2", "masks", "nn", "return", "self"], "sam3_tracker/modeling_sam3_tracker.py:Sam3TrackerPromptEncoder.__init__": ["Embedding", "ModelMaskEmbedding", "ModelPositionalEmbedding", "ModelPromptEncoder", "ModelPromptEncoderConfig", "Module", "__init__", "class", "config", "def", "hidden_size", "image_embedding_size", "image_size", "input_image_size", "mask_embed", "mask_input_size", "nn", "no_mask_embed", "not_a_point_embed", "num_point_embeddings", "patch_size", "point_embed", "self", "shared_embedding", "super"], "sam3_tracker/modeling_sam3_tracker.py:Sam3TrackerPromptEncoder._embed_points": ["ModelPromptEncoder", "Module", "None", "Tensor", "_embed_points", "bool", "clamp", "class", "constant", "def", "functional", "if", "input_image_size", "input_shape", "labels", "min", "mode", "nn", "not_a_point_embed", "pad", "point_embed", "point_embedding", "points", "return", "self", "shared_embedding", "torch", "unsqueeze", "value", "weight", "where", "zeros_like"], "sam3_tracker/modeling_sam3_tracker.py:Sam3TrackerPromptEncoder._embed_boxes": ["ModelPromptEncoder", "Module", "Tensor", "_embed_boxes", "boxes", "class", "constant", "coords", "corner_embedding", "def", "expand_as", "functional", "input_image_size", "mode", "nn", "not_a_point_embed", "pad", "point_embed", "return", "self", "shape", "shared_embedding", "torch", "value", "view", "weight"], "sam3_tracker/modeling_sam3_tracker.py:Sam3TrackerPromptEncoder.forward": ["If", "ModelPromptEncoder", "Module", "None", "Tensor", "ValueError", "_embed_boxes", "_embed_points", "also", "are", "batch_size", "be", "box_embeddings", "cat", "class", "def", "dense_embeddings", "dim", "else", "expand", "forward", "if", "image_embedding_size", "input_boxes", "input_labels", "input_masks", "input_points", "is", "labels", "mask_embed", "must", "nn", "no_mask_embed", "not", "pad", "point_embeddings", "points", "provided", "raise", "reshape", "return", "self", "shape", "sparse_embeddings", "torch", "tuple", "weight"], "sam3_tracker/modeling_sam3_tracker.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "attention_mask", "attn_output", "attn_weights", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "softmax", "to", "torch", "training", "transpose", "value"], "sam3_tracker/modeling_sam3_tracker.py:Sam3TrackerAttention.__init__": ["False", "Linear", "ModelAttention", "Module", "None", "__init__", "attention_downModelple_rate", "class", "config", "def", "downModelple_rate", "else", "head_dim", "hidden_size", "if", "internal_dim", "is", "is_causal", "k_proj", "nn", "num_attention_heads", "o_proj", "q_proj", "scaling", "self", "super", "v_proj"], "sam3_tracker/modeling_sam3_tracker.py:Sam3TrackerAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "ModelAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_attn_implementation", "attention_interface", "attention_mask", "attention_similarity", "attn_output", "attn_weights", "batch_size", "class", "config", "contiguous", "def", "dropout", "eager", "eager_attention_forward", "forward", "head_dim", "if", "is_causal", "k_proj", "key", "kwargs", "new_shape", "nn", "num_attention_heads", "o_proj", "point_batch_size", "q_proj", "query", "reshape", "return", "scaling", "self", "shape", "torch", "transpose", "tuple", "v_proj", "value", "view"], "sam3_tracker/modeling_sam3_tracker.py:Sam3TrackerTwoWayAttentionBlock.__init__": ["False", "LayerNorm", "ModelAttention", "ModelFeedForward", "ModelMaskDecoderConfig", "ModelTwoWayAttentionBlock", "Module", "__init__", "bool", "class", "config", "cross_attn_image_to_token", "cross_attn_token_to_image", "def", "downModelple_rate", "hidden_size", "layer_norm1", "layer_norm2", "layer_norm3", "layer_norm4", "mlp", "mlp_dim", "nn", "num_hidden_layers", "num_layers", "self", "self_attn", "skip_first_layer_pe", "super"], "sam3_tracker/modeling_sam3_tracker.py:Sam3TrackerTwoWayAttentionBlock.forward": ["ModelTwoWayAttentionBlock", "Module", "Tensor", "TransformersKwargs", "Unpack", "_", "attention_similarity", "attn_out", "class", "cross_attn_image_to_token", "cross_attn_token_to_image", "def", "else", "forward", "if", "key", "key_point_embedding", "keys", "kwargs", "layer_norm1", "layer_norm2", "layer_norm3", "layer_norm4", "mlp", "mlp_out", "nn", "queries", "query", "query_point_embedding", "return", "self", "self_attn", "skip_first_layer_pe", "value"], "sam3_tracker/modeling_sam3_tracker.py:Sam3TrackerTwoWayTransformer.__init__": ["LayerNorm", "ModelAttention", "ModelMaskDecoderConfig", "ModelTwoWayAttentionBlock", "ModelTwoWayTransformer", "Module", "ModuleList", "__init__", "append", "class", "config", "def", "final_attn_token_to_image", "for", "hidden_size", "i", "in", "layer_norm_final_attn", "layers", "nn", "num_hidden_layers", "range", "self", "skip_first_layer_pe", "super"], "sam3_tracker/modeling_sam3_tracker.py:Sam3TrackerTwoWayTransformer.forward": ["BaseModelOutput", "ModelTwoWayTransformer", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "_", "an", "attention_similarity", "attn_out", "class", "def", "final_attn_token_to_image", "flatten", "for", "forward", "have", "if", "image_embedding", "image_embeddings", "image_positional_embeddings", "in", "is", "key", "key_point_embedding", "keys", "kwargs", "layer", "layer_norm_final_attn", "layers", "nn", "not", "permute", "point_embeddings", "queries", "query", "query_point_embedding", "raise", "return", "self", "specify", "target_embedding", "to", "tuple", "unsqueeze", "value"], "sam3_tracker/modeling_sam3_tracker.py:Sam3TrackerLayerNorm.__init__": ["LayerNorm", "ModelLayerNorm", "NotImplementedError", "Unsupported", "__init__", "channels_first", "channels_last", "class", "data", "data_format", "def", "eps", "f", "format", "if", "in", "kwargs", "nn", "normalized_shape", "not", "raise", "self", "super"], "sam3_tracker/modeling_sam3_tracker.py:Sam3TrackerLayerNorm.forward": ["LayerNorm", "ModelLayerNorm", "Tensor", "channels_first", "class", "data_format", "def", "else", "features", "forward", "if", "nn", "permute", "return", "self", "super", "torch"], "sam3_tracker/modeling_sam3_tracker.py:Sam3TrackerMaskDecoder.__init__": ["Conv2d", "ConvTranspose2d", "Embedding", "GELU", "ModelFeedForward", "ModelLayerNorm", "ModelMaskDecoder", "ModelMaskDecoderConfig", "ModelTwoWayTransformer", "Module", "ModuleList", "True", "_", "__init__", "activation", "channels_first", "class", "config", "conv_s0", "conv_s1", "data_format", "def", "dynamic_multimask_stability_delta", "dynamic_multimask_stability_thresh", "dynamic_multimask_via_stability", "for", "hidden_size", "in", "iou_head_depth", "iou_head_hidden_dim", "iou_prediction_head", "iou_token", "kernel_size", "mask_tokens", "mlps_list", "nn", "num_mask_tokens", "num_multimask_outputs", "obj_score_token", "output_hypernetworks_mlps", "pred_obj_score_head", "range", "self", "sigmoid_output", "stride", "super", "transformer", "upscale_conv1", "upscale_conv2", "upscale_layer_norm"], "sam3_tracker/modeling_sam3_tracker.py:Sam3TrackerMaskDecoder.forward": ["ModelMaskDecoder", "Model_tokens_out", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "_dynamic_multimask_via_stability", "activation", "and", "attention_similarity", "batch_size", "bool", "cat", "class", "current_mlp", "def", "dense_prompt_embeddings", "dim", "dtype", "dynamic_multimask_via_stability", "elif", "else", "feat_s0", "feat_s1", "for", "forward", "height", "high_resolution_features", "hyper_in", "hyper_in_list", "i", "if", "image_embeddings", "image_positional_embeddings", "in", "iou_pred", "iou_prediction_head", "iou_token", "iou_token_out", "kwargs", "list", "mask_slice", "mask_tokens", "mask_tokens_out", "masks", "multimask_output", "nn", "not", "num_channels", "num_mask_tokens", "obj_score_token", "object_score_logits", "output_hypernetworks_mlps", "output_tokens", "point_batch_size", "point_embeddings", "pred_obj_score_head", "range", "repeat", "repeat_interleave", "return", "self", "shape", "slice", "sparse_prompt_embeddings", "stack", "target_embedding", "to", "tokens", "torch", "training", "transformer", "transpose", "tuple", "upscale_conv1", "upscale_conv2", "upscale_layer_norm", "upscaled_embedding", "view", "weight", "width"], "sam3_tracker/modeling_sam3_tracker.py:Sam3TrackerMaskDecoder._get_stability_scores": ["ModelMaskDecoder", "Module", "_get_stability_scores", "area_i", "area_u", "class", "def", "dim", "dynamic_multimask_stability_delta", "flatten", "float", "mask_logits", "nn", "return", "self", "stability_delta", "stability_scores", "sum", "torch", "where"], "sam3_tracker/modeling_sam3_tracker.py:Sam3TrackerMaskDecoder._dynamic_multimask_via_stability": ["ModelMaskDecoder", "Module", "None", "_dynamic_multimask_via_stability", "_get_stability_scores", "all_iou_scores", "all_mask_logits", "argmax", "best_multimask_iou_scores", "best_multimask_logits", "best_scores_inds", "best_scores_inds_expanded", "class", "def", "dim", "dynamic_multimask_stability_thresh", "expand", "expand_as", "gather", "iou_scores_out", "is_stable", "mask_logits_out", "multimask_iou_scores", "multimask_logits", "nn", "return", "self", "singlemask_iou_scores", "singlemask_logits", "size", "stability_scores", "torch", "unsqueeze", "where"], "sam3_tracker/modeling_sam3_tracker.py:Sam3TrackerModel.__init__": ["AutoModel", "ModelConfig", "ModelMaskDecoder", "ModelModel", "ModelPositionalEmbedding", "ModelPreTrainedModel", "ModelPromptEncoder", "None", "Parameter", "__init__", "_attn_implementation", "and", "backbone_feature_sizes", "class", "config", "def", "dict", "fpn_hidden_size", "from_config", "hasattr", "hidden_dim", "if", "is", "isinstance", "mask_decoder", "mask_decoder_config", "nn", "no_memory_embedding", "not", "post_init", "prompt_encoder", "prompt_encoder_config", "self", "shared_image_embedding", "super", "torch", "tracker_config", "vision_config", "vision_encoder", "zeros"], "sam3_tracker/modeling_sam3_tracker.py:Sam3TrackerModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "return", "self", "vision_encoder"], "sam3_tracker/modeling_sam3_tracker.py:Sam3TrackerModel.get_image_wide_positional_embeddings": ["ModelModel", "ModelPreTrainedModel", "Tensor", "class", "cumsum", "def", "device", "dim", "dtype", "get_image_wide_positional_embeddings", "grid", "image_embedding_size", "ones", "permute", "positional_embedding", "prompt_encoder", "return", "self", "shared_image_embedding", "size", "stack", "target_device", "target_dtype", "torch", "unsqueeze", "x_embed", "y_embed"], "sam3_tracker/modeling_sam3_tracker.py:Sam3TrackerModel.get_image_embeddings": ["FloatTensor", "ModelModel", "ModelPreTrainedModel", "Tensor", "TransformersKwargs", "Unpack", "_", "backbone_feature_sizes", "batch_size", "class", "def", "feat", "feat_size", "feature_maps", "for", "get_image_embeddings", "get_image_features", "image_embeddings", "in", "kwargs", "list", "no_memory_embedding", "permute", "pixel_values", "r", "return", "self", "shape", "torch", "view", "zip"], "sam3_tracker/modeling_sam3_tracker.py:Sam3TrackerModel.get_prompt_embeddings": ["FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "class", "def", "get_prompt_embeddings", "input_boxes", "input_labels", "input_masks", "input_points", "prompt_encoder", "prompt_output", "r", "return", "self", "torch"], "sam3_tracker/modeling_sam3_tracker.py:Sam3TrackerModel.forward": ["Exactly", "F", "False", "FloatTensor", "Got", "LongTensor", "ModelImageSegmentationOutput", "ModelModel", "ModelPreTrainedModel", "None", "TransformersKwargs", "True", "Unpack", "ValueError", "You", "_", "align_corners", "and", "antialias", "as", "attention_similarity", "backbone_feature_sizes", "batch_size", "be", "bilinear", "bool", "bounding", "box", "boxes", "class", "def", "dense_embeddings", "dense_prompt_embeddings", "device", "dtype", "else", "f", "feat", "feat_size", "feature_maps", "float", "for", "forward", "get_image_features", "get_image_wide_positional_embeddings", "high_resolution_features", "if", "image_embeddings", "image_positional_embeddings", "in", "input", "input_boxes", "input_labels", "input_masks", "input_points", "int", "int32", "interpolate", "iou_scores", "is", "kwargs", "low_res_multimasks", "many", "mask_decoder", "mask_input_size", "mode", "multimask_output", "must", "no_memory_embedding", "not", "object_score_logits", "of", "one", "ones", "ones_like", "or", "per", "permute", "pixel_values", "points", "pred_masks", "prompt_encoder", "provide", "provided", "r", "raise", "repeat", "return", "self", "shape", "should", "size", "sparse_embeddings", "sparse_prompt_embeddings", "target_embedding", "to", "torch", "view", "vision_attentions", "vision_hidden_states", "zeros", "zip"], "sam3_tracker/modeling_sam3_tracker.py:Sam3TrackerModel.get_image_features": ["FloatTensor", "ModelModel", "ModelPreTrainedModel", "ModelVisionEncoderOutput", "None", "Tensor", "TransformersKwargs", "Unpack", "attentions", "class", "conv_s0", "conv_s1", "def", "feature_map", "feature_map_position_embedding", "feature_maps", "feature_maps_position_embeddings", "flatten", "for", "fpn_hidden_states", "fpn_position_encoding", "get_image_features", "hidden_states", "in", "kwargs", "list", "mask_decoder", "permute", "pixel_values", "r", "return", "self", "torch", "tuple", "vision_encoder", "vision_outputs"], "sam3_tracker_video/modeling_sam3_tracker_video.py:Sam3TrackerVideoInferenceCache.__init__": ["ModelInferenceCache", "__init__", "_vision_features", "class", "cpu", "def", "device", "inference_device", "inference_state_device", "int", "max_vision_features_cache_size", "self", "str", "torch"], "sam3_tracker_video/modeling_sam3_tracker_video.py:Sam3TrackerVideoInferenceCache.cache_vision_features": ["ModelInferenceCache", "Tensor", "True", "_vision_features", "and", "cache_vision_features", "cached", "class", "def", "dict", "elif", "else", "features", "for", "frame_idx", "if", "in", "inference_state_device", "int", "isinstance", "items", "key", "keys", "len", "list", "max_vision_features_cache_size", "min", "non_blocking", "pop", "self", "to", "torch", "tuple", "v", "value"], "sam3_tracker_video/modeling_sam3_tracker_video.py:Sam3TrackerVideoInferenceCache.get_vision_features": ["ModelInferenceCache", "None", "Tensor", "True", "_vision_features", "and", "cached", "class", "def", "dict", "elif", "else", "for", "frame_idx", "get_vision_features", "if", "in", "inference_device", "int", "isinstance", "items", "key", "list", "moved", "non_blocking", "not", "return", "self", "to", "torch", "tuple", "v", "value"], "sam3_tracker_video/modeling_sam3_tracker_video.py:Sam3TrackerVideoInferenceCache.clear_all": ["ModelInferenceCache", "_vision_features", "class", "clear", "clear_all", "def", "self"], "sam3_tracker_video/modeling_sam3_tracker_video.py:Sam3TrackerVideoInferenceSession.__init__": ["FloatTensor", "ModelInferenceCache", "ModelInferenceSession", "None", "OrderedDict", "__init__", "_obj_id_to_idx", "_obj_idx_to_id", "cache", "class", "cpu", "def", "device", "dict", "dtype", "else", "enumerate", "float32", "frames_tracked_per_obj", "if", "inference_device", "inference_state_device", "int", "is", "mask_inputs_per_obj", "max_vision_features_cache_size", "not", "obj_ids", "obj_with_new_inputs", "output_dict_per_obj", "point_inputs_per_obj", "processed_frames", "self", "str", "to", "torch", "video", "video_height", "video_storage_device", "video_width"], "sam3_tracker_video/modeling_sam3_tracker_video.py:Sam3TrackerVideoInferenceSession.num_frames": ["ModelInferenceSession", "None", "class", "def", "else", "if", "int", "is", "len", "not", "num_frames", "processed_frames", "return", "self"], "sam3_tracker_video/modeling_sam3_tracker_video.py:Sam3TrackerVideoInferenceSession.obj_id_to_idx": ["ModelInferenceSession", "None", "_obj_id_to_idx", "_obj_idx_to_id", "class", "cond_frame_outputs", "def", "frames_tracked_per_obj", "get", "if", "int", "is", "len", "list", "mask_inputs_per_obj", "non_cond_frame_outputs", "not", "obj_id", "obj_id_to_idx", "obj_ids", "obj_idx", "output_dict_per_obj", "point_inputs_per_obj", "return", "self"], "sam3_tracker_video/modeling_sam3_tracker_video.py:Sam3TrackerVideoInferenceSession.obj_idx_to_id": ["ModelInferenceSession", "_obj_idx_to_id", "class", "def", "int", "obj_idx", "obj_idx_to_id", "return", "self"], "sam3_tracker_video/modeling_sam3_tracker_video.py:Sam3TrackerVideoInferenceSession.get_obj_num": ["ModelInferenceSession", "_obj_idx_to_id", "class", "def", "get_obj_num", "int", "len", "return", "self"], "sam3_tracker_video/modeling_sam3_tracker_video.py:Sam3TrackerVideoInferenceSession.add_point_inputs": ["False", "ModelInferenceSession", "Tensor", "add_point_inputs", "class", "def", "device_inputs", "dict", "else", "for", "frame_idx", "if", "in", "inference_device", "inputs", "int", "isinstance", "items", "key", "non_blocking", "obj_idx", "point_inputs_per_obj", "self", "to", "torch", "value"], "sam3_tracker_video/modeling_sam3_tracker_video.py:Sam3TrackerVideoInferenceSession.remove_point_inputs": ["ModelInferenceSession", "None", "class", "def", "frame_idx", "int", "obj_idx", "point_inputs_per_obj", "pop", "remove_point_inputs", "self"], "sam3_tracker_video/modeling_sam3_tracker_video.py:Sam3TrackerVideoInferenceSession.add_mask_inputs": ["ModelInferenceSession", "Tensor", "True", "add_mask_inputs", "class", "def", "dtype", "frame_idx", "inference_device", "inputs", "int", "mask_inputs_per_obj", "non_blocking", "obj_idx", "self", "to", "torch"], "sam3_tracker_video/modeling_sam3_tracker_video.py:Sam3TrackerVideoInferenceSession.remove_mask_inputs": ["ModelInferenceSession", "None", "class", "def", "frame_idx", "int", "mask_inputs_per_obj", "obj_idx", "pop", "remove_mask_inputs", "self"], "sam3_tracker_video/modeling_sam3_tracker_video.py:Sam3TrackerVideoInferenceSession.store_output": ["ModelInferenceSession", "None", "Tensor", "True", "and", "bool", "class", "cond_frame_outputs", "def", "dict", "elif", "else", "for", "frame_idx", "if", "in", "inference_state_device", "int", "is", "is_conditioning_frame", "isinstance", "items", "key", "non_blocking", "non_cond_frame_outputs", "obj_idx", "object_pointer", "object_score_logits", "output_dict_per_obj", "output_key", "output_value", "return", "self", "storage_key", "store_output", "str", "to", "torch", "value"], "sam3_tracker_video/modeling_sam3_tracker_video.py:Sam3TrackerVideoInferenceSession.get_output": ["ModelInferenceSession", "None", "Tensor", "True", "bool", "class", "cond_frame_outputs", "def", "else", "frame_idx", "get", "get_output", "if", "inference_device", "int", "is", "is_conditioning_frame", "isinstance", "non_blocking", "non_cond_frame_outputs", "obj_idx", "out", "output_dict_per_obj", "output_key", "return", "self", "storage_key", "str", "to", "torch", "value"], "sam3_tracker_video/modeling_sam3_tracker_video.py:Sam3TrackerVideoInferenceSession.add_new_frame": ["ModelInferenceSession", "None", "Tensor", "True", "add_new_frame", "class", "def", "dim", "dtype", "else", "frame_idx", "if", "int", "is", "len", "non_blocking", "not", "pixel_values", "processed_frames", "return", "self", "squeeze", "to", "torch", "video_storage_device"], "sam3_tracker_video/modeling_sam3_tracker_video.py:Sam3TrackerVideoInferenceSession.get_frame": ["ModelInferenceSession", "Tensor", "True", "class", "def", "frame_idx", "get_frame", "inference_device", "int", "non_blocking", "processed_frames", "return", "self", "to", "torch"], "sam3_tracker_video/modeling_sam3_tracker_video.py:Sam3TrackerVideoInferenceSession.reset_tracking_data": ["ModelInferenceSession", "_obj_id_to_idx", "_obj_idx_to_id", "class", "clear", "def", "frames_tracked_per_obj", "mask_inputs_per_obj", "obj_ids", "obj_with_new_inputs", "output_dict_per_obj", "point_inputs_per_obj", "reset_tracking_data", "self"], "sam3_tracker_video/modeling_sam3_tracker_video.py:Sam3TrackerVideoInferenceSession.reset_inference_session": ["ModelInferenceSession", "_obj_id_to_idx", "_obj_idx_to_id", "cache", "class", "clear", "clear_all", "def", "frames_tracked_per_obj", "mask_inputs_per_obj", "obj_ids", "obj_with_new_inputs", "output_dict_per_obj", "point_inputs_per_obj", "reset_inference_session", "self"], "sam3_tracker_video/modeling_sam3_tracker_video.py:Sam3TrackerVideoLayerNorm.__init__": ["LayerNorm", "ModelLayerNorm", "NotImplementedError", "Unsupported", "__init__", "channels_first", "channels_last", "class", "data", "data_format", "def", "eps", "f", "format", "if", "in", "kwargs", "nn", "normalized_shape", "not", "raise", "self", "super"], "sam3_tracker_video/modeling_sam3_tracker_video.py:Sam3TrackerVideoLayerNorm.forward": ["LayerNorm", "ModelLayerNorm", "Tensor", "channels_first", "class", "data_format", "def", "else", "features", "forward", "if", "nn", "permute", "return", "self", "super", "torch"], "sam3_tracker_video/modeling_sam3_tracker_video.py:Sam3TrackerVideoPositionEmbeddingSine.__init__": ["False", "ModelPositionEmbeddingSine", "Module", "None", "True", "ValueError", "__init__", "and", "be", "bool", "class", "def", "else", "float", "if", "int", "is", "math", "nn", "normalize", "not", "num_pos_feats", "passed", "pi", "raise", "scale", "self", "should", "super", "temperature"], "sam3_tracker_video/modeling_sam3_tracker_video.py:Sam3TrackerVideoPositionEmbeddingSine.forward": ["ModelPositionEmbeddingSine", "Module", "None", "Size", "Tensor", "arange", "bool", "cat", "class", "cos", "cumsum", "def", "device", "dim", "dim_t", "div", "dtype", "eps", "flatten", "floor", "forward", "if", "int64", "is", "mask", "nn", "normalize", "not_mask", "num_pos_feats", "permute", "pos", "pos_x", "pos_y", "return", "rounding_mode", "scale", "self", "shape", "sin", "stack", "str", "temperature", "to", "torch", "x_embed", "y_embed", "zeros"], "sam3_tracker_video/modeling_sam3_tracker_video.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "attention_mask", "attn_output", "attn_weights", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "softmax", "to", "torch", "training", "transpose", "value"], "sam3_tracker_video/modeling_sam3_tracker_video.py:Sam3TrackerVideoAttention.__init__": ["False", "Linear", "ModelAttention", "Module", "None", "__init__", "attention_downModelple_rate", "class", "config", "def", "downModelple_rate", "else", "head_dim", "hidden_size", "if", "internal_dim", "is", "is_causal", "k_proj", "nn", "num_attention_heads", "o_proj", "q_proj", "scaling", "self", "super", "v_proj"], "sam3_tracker_video/modeling_sam3_tracker_video.py:Sam3TrackerVideoAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "ModelAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_attn_implementation", "attention_interface", "attention_mask", "attention_similarity", "attn_output", "attn_weights", "batch_size", "class", "config", "contiguous", "def", "dropout", "eager", "eager_attention_forward", "forward", "head_dim", "if", "is_causal", "k_proj", "key", "kwargs", "new_shape", "nn", "num_attention_heads", "o_proj", "point_batch_size", "q_proj", "query", "reshape", "return", "scaling", "self", "shape", "torch", "transpose", "tuple", "v_proj", "value", "view"], "sam3_tracker_video/modeling_sam3_tracker_video.py:Sam3TrackerVideoTwoWayAttentionBlock.__init__": ["False", "LayerNorm", "ModelAttention", "ModelFeedForward", "ModelMaskDecoderConfig", "ModelTwoWayAttentionBlock", "Module", "__init__", "bool", "class", "config", "cross_attn_image_to_token", "cross_attn_token_to_image", "def", "downModelple_rate", "hidden_size", "layer_norm1", "layer_norm2", "layer_norm3", "layer_norm4", "mlp", "mlp_dim", "nn", "num_hidden_layers", "num_layers", "self", "self_attn", "skip_first_layer_pe", "super"], "sam3_tracker_video/modeling_sam3_tracker_video.py:Sam3TrackerVideoTwoWayAttentionBlock.forward": ["ModelTwoWayAttentionBlock", "Module", "Tensor", "TransformersKwargs", "Unpack", "_", "attention_similarity", "attn_out", "class", "cross_attn_image_to_token", "cross_attn_token_to_image", "def", "else", "forward", "if", "key", "key_point_embedding", "keys", "kwargs", "layer_norm1", "layer_norm2", "layer_norm3", "layer_norm4", "mlp", "mlp_out", "nn", "queries", "query", "query_point_embedding", "return", "self", "self_attn", "skip_first_layer_pe", "value"], "sam3_tracker_video/modeling_sam3_tracker_video.py:Sam3TrackerVideoFeedForward.__init__": ["ACT2FN", "False", "Linear", "ModelFeedForward", "Module", "ModuleList", "_", "__init__", "activation", "bool", "class", "def", "for", "hidden_dim", "in", "input_dim", "int", "layers", "nn", "num_layers", "output_dim", "proj_in", "proj_out", "range", "relu", "self", "sigmoid_output", "str", "super"], "sam3_tracker_video/modeling_sam3_tracker_video.py:Sam3TrackerVideoFeedForward.forward": ["F", "ModelFeedForward", "Module", "activation", "class", "def", "for", "forward", "hidden_states", "if", "in", "layer", "layers", "nn", "proj_in", "proj_out", "return", "self", "sigmoid", "sigmoid_output"], "sam3_tracker_video/modeling_sam3_tracker_video.py:Sam3TrackerVideoPreTrainedModel._init_weights": ["ModelMemoryFuserCXBlock", "ModelModel", "ModelPositionalEmbedding", "ModelPreTrainedModel", "ModelVisionRotaryEmbedding", "None", "PreTrainedModel", "_init_weights", "class", "copy_", "cos", "create_inv_freq", "def", "elif", "if", "init", "inv_freq", "is", "isinstance", "memory_temporal_positional_encoding", "module", "no_memory_positional_encoding", "no_object_pointer", "normal_", "not", "occlusion_spatial_embedding_parameter", "positional_embedding", "rope_embeddings_cos", "rope_embeddings_sin", "scale", "self", "sin", "std", "super", "zeros_"], "sam3_tracker_video/modeling_sam3_tracker_video.py:Sam3TrackerVideoVisionRotaryEmbedding.__init__": ["Dimension", "False", "ModelConfig", "ModelVisionRotaryEmbedding", "Module", "RoPE", "ValueError", "__init__", "axial", "be", "by", "class", "config", "cos", "create_inv_freq", "def", "dim", "divisible", "end_x", "end_y", "for", "if", "inv_freq", "memory_attention_downModelple_rate", "memory_attention_hidden_size", "memory_attention_num_attention_heads", "memory_attention_rope_feat_sizes", "memory_attention_rope_theta", "must", "nn", "persistent", "raise", "register_buffer", "rope_embeddings_cos", "rope_embeddings_sin", "self", "sin", "super"], "sam3_tracker_video/modeling_sam3_tracker_video.py:Sam3TrackerVideoVisionRotaryEmbedding.forward": ["ModelVisionRotaryEmbedding", "Module", "Tensor", "class", "def", "forward", "nn", "return", "rope_embeddings_cos", "rope_embeddings_sin", "self", "torch", "tuple"], "sam3_tracker_video/modeling_sam3_tracker_video.py:Sam3TrackerVideoVisionRotaryEmbedding.create_inv_freq": ["ModelVisionRotaryEmbedding", "Module", "arange", "cat", "class", "create_inv_freq", "def", "dim", "div", "dtype", "end_x", "end_y", "flattened_indices", "float", "floor", "freqs", "freqs_x", "freqs_y", "inv_freq", "long", "memory_attention_rope_theta", "nn", "outer", "repeat_interleave", "return", "rounding_mode", "self", "torch", "x_positions", "y_positions"], "sam3_tracker_video/modeling_sam3_tracker_video.py:rotate_pairwise": ["Model_pairwise", "def", "dim", "flatten", "return", "shape", "stack", "start_dim", "torch", "unbind", "view", "x", "x1", "x2"], "sam3_tracker_video/modeling_sam3_tracker_video.py:apply_rotary_pos_emb_2d": ["False", "Model_rotary_pos_emb_2d", "Tensor", "and", "bool", "cat", "cos", "cos_k", "def", "dim", "else", "float", "if", "int", "k", "k_embed", "k_pass", "k_rot", "num_k_exclude_rope", "q", "q_embed", "repeat", "repeat_factor", "repeat_freqs_k", "return", "rotate_pairwise", "shape", "sin", "sin_k", "torch", "tuple", "type_as"], "sam3_tracker_video/modeling_sam3_tracker_video.py:Sam3TrackerVideoRoPEAttention.__init__": ["False", "Linear", "ModelConfig", "ModelRoPEAttention", "Module", "None", "__init__", "class", "config", "def", "dropout_p", "else", "head_dim", "hidden_size", "if", "int", "internal_dim", "is", "is_causal", "k_proj", "kv_in_dim", "memory_attention_downModelple_rate", "memory_attention_hidden_size", "memory_attention_num_attention_heads", "memory_attention_rope_dropout", "nn", "not", "num_attention_heads", "o_proj", "q_proj", "rope_k_repeat", "scaling", "self", "super", "v_proj"], "sam3_tracker_video/modeling_sam3_tracker_video.py:Sam3TrackerVideoRoPEAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "FlashAttentionKwargs", "ModelRoPEAttention", "Module", "None", "Tensor", "Unpack", "_attn_implementation", "apply_rotary_pos_emb_2d", "attention_interface", "attention_mask", "attn_output", "attn_weights", "batch_size", "class", "config", "contiguous", "cos", "def", "dropout", "dropout_p", "eager", "eager_attention_forward", "else", "forward", "head_dim", "if", "int", "is_causal", "k_proj", "key", "kwargs", "new_shape", "nn", "not", "num_attention_heads", "num_k_exclude_rope", "o_proj", "point_batch_size", "position_embeddings", "q_proj", "query", "repeat_freqs_k", "reshape", "return", "rope_k_repeat", "scaling", "self", "shape", "sin", "torch", "training", "transpose", "tuple", "v_proj", "value", "view"], "sam3_tracker_video/modeling_sam3_tracker_video.py:Sam3TrackerVideoMemoryAttentionLayer.__init__": ["ACT2FN", "Dropout", "LayerNorm", "Linear", "ModelConfig", "ModelMemoryAttentionLayer", "ModelRoPEAttention", "Module", "True", "__init__", "activation", "class", "config", "cross_attn_image", "def", "dropout", "dropout1", "dropout2", "dropout3", "hidden_size", "kv_in_dim", "layer_norm1", "layer_norm2", "layer_norm3", "linear1", "linear2", "memory_attention_dropout", "memory_attention_feed_forward_hidden_act", "memory_attention_feed_forward_hidden_size", "memory_attention_hidden_size", "nn", "rope_k_repeat", "self", "self_attn", "super"], "sam3_tracker_video/modeling_sam3_tracker_video.py:Sam3TrackerVideoMemoryAttentionLayer.forward": ["ModelMemoryAttentionLayer", "Module", "Tensor", "_", "activation", "class", "cross_attn_image", "def", "dropout", "dropout1", "dropout2", "dropout3", "forward", "int", "key", "key_point_embedding", "keys", "layer_norm1", "layer_norm2", "layer_norm3", "linear1", "linear2", "nn", "num_k_exclude_rope", "position_embeddings", "queries", "query", "return", "rope_position_embeddings", "self", "self_attn", "torch", "tuple", "value"], "sam3_tracker_video/modeling_sam3_tracker_video.py:Sam3TrackerVideoMemoryAttention.__init__": ["LayerNorm", "ModelConfig", "ModelMemoryAttention", "ModelMemoryAttentionLayer", "ModelVisionRotaryEmbedding", "Module", "ModuleList", "_", "__init__", "class", "config", "def", "for", "in", "layer_norm", "layers", "memory_attention_hidden_size", "memory_attention_num_layers", "nn", "range", "rotary_emb", "self", "super"], "sam3_tracker_video/modeling_sam3_tracker_video.py:Sam3TrackerVideoMemoryAttention.forward": ["ModelMemoryAttention", "Module", "None", "Tensor", "class", "current_vision_features", "current_vision_position_embeddings", "def", "else", "for", "forward", "if", "in", "int", "is", "key_point_embedding", "keys", "layer", "layer_norm", "layers", "memory", "memory_posision_embeddings", "ndim", "nn", "normed_output", "not", "num_k_exclude_rope", "num_object_pointer_tokens", "output", "queries", "return", "rope_position_embeddings", "rotary_emb", "self", "torch", "transpose", "unsqueeze"], "sam3_tracker_video/modeling_sam3_tracker_video.py:Sam3TrackerVideoMemoryFuserCXBlock.__init__": ["ACT2FN", "Conv2d", "GradientCheckpointingLayer", "Linear", "ModelConfig", "ModelLayerNorm", "ModelMemoryFuserCXBlock", "Parameter", "True", "__init__", "activation", "channels_first", "class", "config", "data_format", "def", "depthwise_conv", "eps", "groups", "kernel_size", "layer_norm", "memory_fuser_embed_dim", "memory_fuser_hidden_act", "memory_fuser_intermediate_dim", "memory_fuser_kernel_size", "memory_fuser_layer_scale_init_value", "memory_fuser_padding", "nn", "ones", "padding", "pointwise_conv1", "pointwise_conv2", "requires_grad", "scale", "self", "super", "torch"], "sam3_tracker_video/modeling_sam3_tracker_video.py:Sam3TrackerVideoMemoryFuserCXBlock.forward": ["GradientCheckpointingLayer", "ModelMemoryFuserCXBlock", "activation", "class", "def", "depthwise_conv", "forward", "hidden_states", "input", "layer_norm", "permute", "pointwise_conv1", "pointwise_conv2", "return", "scale", "self"], "sam3_tracker_video/modeling_sam3_tracker_video.py:Sam3TrackerVideoMemoryFuser.__init__": ["ModelConfig", "ModelMemoryFuser", "ModelMemoryFuserCXBlock", "Module", "ModuleList", "_", "__init__", "class", "config", "def", "for", "in", "layers", "memory_fuser_num_layers", "nn", "range", "self", "super"], "sam3_tracker_video/modeling_sam3_tracker_video.py:Sam3TrackerVideoMemoryFuser.forward": ["ModelMemoryFuser", "Module", "class", "def", "for", "forward", "hidden_states", "in", "layer", "layers", "nn", "return", "self"], "sam3_tracker_video/modeling_sam3_tracker_video.py:Sam3TrackerVideoMaskDownSamplerLayer.__init__": ["ACT2FN", "Conv2d", "ModelConfig", "ModelLayerNorm", "ModelMaskDownModelplerLayer", "Module", "__init__", "activation", "channels_first", "class", "config", "conv", "data_format", "def", "eps", "in_channels", "int", "kernel_size", "layer_norm", "mask_downModelpler_hidden_act", "mask_downModelpler_kernel_size", "mask_downModelpler_padding", "mask_downModelpler_stride", "nn", "out_channels", "padding", "self", "stride", "super"], "sam3_tracker_video/modeling_sam3_tracker_video.py:Sam3TrackerVideoMaskDownSamplerLayer.forward": ["ModelMaskDownModelplerLayer", "Module", "activation", "class", "conv", "def", "forward", "layer_norm", "nn", "return", "self", "x"], "sam3_tracker_video/modeling_sam3_tracker_video.py:Sam3TrackerVideoMaskDownSampler.__init__": ["ACT2FN", "Conv2d", "ModelConfig", "ModelMaskDownModelpler", "ModelMaskDownModelplerLayer", "Module", "ModuleList", "_", "__init__", "activation", "append", "class", "config", "def", "final_conv", "for", "in", "int", "kernel_size", "layers", "log2", "mask_downModelpler_embed_dim", "mask_downModelpler_hidden_act", "mask_downModelpler_stride", "mask_downModelpler_total_stride", "mask_in_chans", "mask_out_chans", "math", "nn", "num_layers", "range", "self", "super"], "sam3_tracker_video/modeling_sam3_tracker_video.py:Sam3TrackerVideoMaskDownSampler.forward": ["ModelMaskDownModelpler", "Module", "class", "def", "final_conv", "for", "forward", "in", "layer", "layers", "nn", "return", "self", "x"], "sam3_tracker_video/modeling_sam3_tracker_video.py:Sam3TrackerVideoMemoryEncoder.__init__": ["Conv2d", "ModelConfig", "ModelMaskDownModelpler", "ModelMemoryEncoder", "ModelMemoryFuser", "ModelPositionEmbeddingSine", "Module", "True", "__init__", "class", "config", "def", "feature_projection", "hidden_size", "kernel_size", "mask_downModelpler", "memory_encoder_hidden_size", "memory_encoder_output_channels", "memory_fuser", "nn", "normalize", "num_pos_feats", "output_channels", "position_encoding", "projection", "self", "super"], "sam3_tracker_video/modeling_sam3_tracker_video.py:Sam3TrackerVideoMemoryEncoder.forward": ["ModelMemoryEncoder", "Module", "Tensor", "class", "def", "device", "dtype", "feature_projection", "forward", "mask_downModelpler", "masks", "memory_fuser", "nn", "position_encoding", "projection", "return", "self", "shape", "torch", "tuple", "vision_features", "vision_pos_enc"], "sam3_tracker_video/modeling_sam3_tracker_video.py:Sam3TrackerVideoPositionalEmbedding.__init__": ["ModelPositionalEmbedding", "ModelPromptEncoderConfig", "Module", "__init__", "class", "config", "def", "hidden_size", "nn", "positional_embedding", "randn", "register_buffer", "scale", "self", "super", "torch"], "sam3_tracker_video/modeling_sam3_tracker_video.py:Sam3TrackerVideoPositionalEmbedding.forward": ["ModelPositionalEmbedding", "Module", "None", "cat", "class", "clone", "coordinates", "cos", "def", "dim", "dtype", "float32", "forward", "if", "input_coords", "input_shape", "is", "nn", "not", "np", "pi", "positional_embedding", "return", "self", "sin", "to", "torch"], "sam3_tracker_video/modeling_sam3_tracker_video.py:Sam3TrackerVideoMaskEmbedding.__init__": ["ACT2FN", "Conv2d", "ModelLayerNorm", "ModelMaskEmbedding", "ModelPromptEncoderConfig", "Module", "__init__", "activation", "channels_first", "class", "config", "conv1", "conv2", "conv3", "data_format", "def", "eps", "hidden_act", "hidden_size", "kernel_size", "layer_norm1", "layer_norm2", "layer_norm_eps", "mask_input_channels", "nn", "self", "stride", "super"], "sam3_tracker_video/modeling_sam3_tracker_video.py:Sam3TrackerVideoMaskEmbedding.forward": ["ModelMaskEmbedding", "Module", "activation", "class", "conv1", "conv2", "conv3", "def", "dense_embeddings", "forward", "hidden_states", "layer_norm1", "layer_norm2", "masks", "nn", "return", "self"], "sam3_tracker_video/modeling_sam3_tracker_video.py:Sam3TrackerVideoPromptEncoder.__init__": ["Embedding", "ModelMaskEmbedding", "ModelPositionalEmbedding", "ModelPromptEncoder", "ModelPromptEncoderConfig", "Module", "__init__", "class", "config", "def", "hidden_size", "image_embedding_size", "image_size", "input_image_size", "mask_embed", "mask_input_size", "nn", "no_mask_embed", "not_a_point_embed", "num_point_embeddings", "patch_size", "point_embed", "self", "shared_embedding", "super"], "sam3_tracker_video/modeling_sam3_tracker_video.py:Sam3TrackerVideoPromptEncoder._embed_points": ["ModelPromptEncoder", "Module", "None", "Tensor", "_embed_points", "bool", "clamp", "class", "constant", "def", "functional", "if", "input_image_size", "input_shape", "labels", "min", "mode", "nn", "not_a_point_embed", "pad", "point_embed", "point_embedding", "points", "return", "self", "shared_embedding", "torch", "unsqueeze", "value", "weight", "where", "zeros_like"], "sam3_tracker_video/modeling_sam3_tracker_video.py:Sam3TrackerVideoPromptEncoder._embed_boxes": ["ModelPromptEncoder", "Module", "Tensor", "_embed_boxes", "boxes", "class", "constant", "coords", "corner_embedding", "def", "expand_as", "functional", "input_image_size", "mode", "nn", "not_a_point_embed", "pad", "point_embed", "return", "self", "shape", "shared_embedding", "torch", "value", "view", "weight"], "sam3_tracker_video/modeling_sam3_tracker_video.py:Sam3TrackerVideoPromptEncoder.forward": ["If", "ModelPromptEncoder", "Module", "None", "Tensor", "ValueError", "_embed_boxes", "_embed_points", "also", "are", "batch_size", "be", "box_embeddings", "cat", "class", "def", "dense_embeddings", "dim", "else", "expand", "forward", "if", "image_embedding_size", "input_boxes", "input_labels", "input_masks", "input_points", "is", "labels", "mask_embed", "must", "nn", "no_mask_embed", "not", "pad", "point_embeddings", "points", "provided", "raise", "reshape", "return", "self", "shape", "sparse_embeddings", "torch", "tuple", "weight"], "sam3_tracker_video/modeling_sam3_tracker_video.py:Sam3TrackerVideoTwoWayTransformer.__init__": ["LayerNorm", "ModelAttention", "ModelMaskDecoderConfig", "ModelTwoWayAttentionBlock", "ModelTwoWayTransformer", "Module", "ModuleList", "__init__", "append", "class", "config", "def", "final_attn_token_to_image", "for", "hidden_size", "i", "in", "layer_norm_final_attn", "layers", "nn", "num_hidden_layers", "range", "self", "skip_first_layer_pe", "super"], "sam3_tracker_video/modeling_sam3_tracker_video.py:Sam3TrackerVideoTwoWayTransformer.forward": ["BaseModelOutput", "ModelTwoWayTransformer", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "_", "an", "attention_similarity", "attn_out", "class", "def", "final_attn_token_to_image", "flatten", "for", "forward", "have", "if", "image_embedding", "image_embeddings", "image_positional_embeddings", "in", "is", "key", "key_point_embedding", "keys", "kwargs", "layer", "layer_norm_final_attn", "layers", "nn", "not", "permute", "point_embeddings", "queries", "query", "query_point_embedding", "raise", "return", "self", "specify", "target_embedding", "to", "tuple", "unsqueeze", "value"], "sam3_tracker_video/modeling_sam3_tracker_video.py:Sam3TrackerVideoMaskDecoder.__init__": ["Conv2d", "ConvTranspose2d", "Embedding", "GELU", "ModelFeedForward", "ModelLayerNorm", "ModelMaskDecoder", "ModelMaskDecoderConfig", "ModelTwoWayTransformer", "Module", "ModuleList", "True", "_", "__init__", "activation", "channels_first", "class", "config", "conv_s0", "conv_s1", "data_format", "def", "dynamic_multimask_stability_delta", "dynamic_multimask_stability_thresh", "dynamic_multimask_via_stability", "for", "hidden_size", "in", "iou_head_depth", "iou_head_hidden_dim", "iou_prediction_head", "iou_token", "kernel_size", "mask_tokens", "mlps_list", "nn", "num_mask_tokens", "num_multimask_outputs", "obj_score_token", "output_hypernetworks_mlps", "pred_obj_score_head", "range", "self", "sigmoid_output", "stride", "super", "transformer", "upscale_conv1", "upscale_conv2", "upscale_layer_norm"], "sam3_tracker_video/modeling_sam3_tracker_video.py:Sam3TrackerVideoMaskDecoder.forward": ["ModelMaskDecoder", "Model_tokens_out", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "_dynamic_multimask_via_stability", "activation", "and", "attention_similarity", "batch_size", "bool", "cat", "class", "current_mlp", "def", "dense_prompt_embeddings", "dim", "dtype", "dynamic_multimask_via_stability", "elif", "else", "feat_s0", "feat_s1", "for", "forward", "height", "high_resolution_features", "hyper_in", "hyper_in_list", "i", "if", "image_embeddings", "image_positional_embeddings", "in", "iou_pred", "iou_prediction_head", "iou_token", "iou_token_out", "kwargs", "list", "mask_slice", "mask_tokens", "mask_tokens_out", "masks", "multimask_output", "nn", "not", "num_channels", "num_mask_tokens", "obj_score_token", "object_score_logits", "output_hypernetworks_mlps", "output_tokens", "point_batch_size", "point_embeddings", "pred_obj_score_head", "range", "repeat", "repeat_interleave", "return", "self", "shape", "slice", "sparse_prompt_embeddings", "stack", "target_embedding", "to", "tokens", "torch", "training", "transformer", "transpose", "tuple", "upscale_conv1", "upscale_conv2", "upscale_layer_norm", "upscaled_embedding", "view", "weight", "width"], "sam3_tracker_video/modeling_sam3_tracker_video.py:Sam3TrackerVideoMaskDecoder._get_stability_scores": ["ModelMaskDecoder", "Module", "_get_stability_scores", "area_i", "area_u", "class", "def", "dim", "dynamic_multimask_stability_delta", "flatten", "float", "mask_logits", "nn", "return", "self", "stability_delta", "stability_scores", "sum", "torch", "where"], "sam3_tracker_video/modeling_sam3_tracker_video.py:Sam3TrackerVideoMaskDecoder._dynamic_multimask_via_stability": ["ModelMaskDecoder", "Module", "None", "_dynamic_multimask_via_stability", "_get_stability_scores", "all_iou_scores", "all_mask_logits", "argmax", "best_multimask_iou_scores", "best_multimask_logits", "best_scores_inds", "best_scores_inds_expanded", "class", "def", "dim", "dynamic_multimask_stability_thresh", "expand", "expand_as", "gather", "iou_scores_out", "is_stable", "mask_logits_out", "multimask_iou_scores", "multimask_logits", "nn", "return", "self", "singlemask_iou_scores", "singlemask_logits", "size", "stability_scores", "torch", "unsqueeze", "where"], "sam3_tracker_video/modeling_sam3_tracker_video.py:get_1d_sine_pe": ["Model_1d_sine_pe", "arange", "cat", "cos", "def", "device", "dim", "dim_t", "dtype", "float32", "pe_dim", "pos_embed", "pos_inds", "return", "sin", "temperature", "torch", "unsqueeze"], "sam3_tracker_video/modeling_sam3_tracker_video.py:Sam3TrackerVideoModel.__init__": ["AutoModel", "Conv2d", "False", "Identity", "Linear", "ModelConfig", "ModelFeedForward", "ModelMaskDecoder", "ModelMemoryAttention", "ModelMemoryEncoder", "ModelModel", "ModelPositionalEmbedding", "ModelPreTrainedModel", "ModelPromptEncoder", "None", "Parameter", "__init__", "_attn_implementation", "and", "backbone_feature_sizes", "bool", "class", "config", "def", "dict", "else", "enable_occlusion_spatial_embedding", "enable_temporal_pos_encoding_for_object_pointers", "fpn_hidden_size", "from_config", "hasattr", "hidden_dim", "if", "image_size", "is", "isinstance", "kernel_size", "mask_decoder", "mask_decoder_config", "mask_downModelple", "mem_dim", "memory_attention", "memory_encoder", "memory_encoder_output_channels", "memory_temporal_positional_encoding", "nn", "no_memory_embedding", "no_memory_positional_encoding", "no_object_pointer", "not", "num_maskmem", "object_pointer_proj", "occlusion_spatial_embedding_parameter", "post_init", "prompt_encoder", "prompt_encoder_config", "r", "remove_vision_encoder", "self", "shared_image_embedding", "stride", "super", "temporal_positional_encoding_projection_layer", "torch", "tracker_config", "vision_config", "vision_encoder", "zeros"], "sam3_tracker_video/modeling_sam3_tracker_video.py:Sam3TrackerVideoModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "return", "self", "vision_encoder"], "sam3_tracker_video/modeling_sam3_tracker_video.py:Sam3TrackerVideoModel.get_image_wide_positional_embeddings": ["ModelModel", "ModelPreTrainedModel", "Tensor", "class", "cumsum", "def", "device", "dim", "dtype", "get_image_wide_positional_embeddings", "grid", "image_embedding_size", "ones", "permute", "positional_embedding", "prompt_encoder", "return", "self", "shared_image_embedding", "size", "stack", "target_device", "target_dtype", "torch", "unsqueeze", "x_embed", "y_embed"], "sam3_tracker_video/modeling_sam3_tracker_video.py:Sam3TrackerVideoModel.get_image_embeddings": ["FloatTensor", "ModelModel", "ModelPreTrainedModel", "Tensor", "TransformersKwargs", "Unpack", "_", "backbone_feature_sizes", "batch_size", "class", "def", "feat", "feat_size", "feature_maps", "for", "get_image_embeddings", "get_image_features", "image_embeddings", "in", "kwargs", "list", "no_memory_embedding", "permute", "pixel_values", "r", "return", "self", "shape", "torch", "view", "zip"], "sam3_tracker_video/modeling_sam3_tracker_video.py:Sam3TrackerVideoModel.get_prompt_embeddings": ["FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "class", "def", "get_prompt_embeddings", "input_boxes", "input_labels", "input_masks", "input_points", "prompt_encoder", "prompt_output", "r", "return", "self", "torch", "tuple"], "sam3_tracker_video/modeling_sam3_tracker_video.py:Sam3TrackerVideoModel.forward": ["False", "ModelInferenceSession", "ModelModel", "ModelPreTrainedModel", "ModelSegmentationOutput", "No", "None", "Tensor", "True", "ValueError", "_batch_encode_memories", "_run_single_frame_inference", "add", "add_new_frame", "all_object_score_logits", "all_pred_masks", "and", "append", "are", "batch_size", "bool", "cat", "class", "cond_frame_outputs", "copy", "current_out", "def", "dim", "else", "first", "for", "forward", "frame", "frame_idx", "frames_tracked_per_obj", "get", "get_obj_num", "get_output", "has_cond_output", "has_new_inputs", "high_res_masks", "high_res_masks_for_memory", "if", "in", "inference_session", "inputs", "int", "is", "is_conditioning_frame", "is_init_cond_frame", "is_mask_from_pts_per_obj", "kwargs", "len", "mask_inputs", "mask_inputs_per_obj", "not", "num_maskmem", "num_objects", "obj_id", "obj_ids", "obj_idx", "obj_idx_to_id", "obj_with_new_inputs", "object_ids", "object_score_logits", "object_score_logits_for_memory", "object_score_logits_per_obj", "objects", "objects_needing_memory_encoding", "or", "output_dict_per_obj", "output_value", "please", "point_inputs", "point_inputs_per_obj", "pred_masks", "pred_masks_per_obj", "provided", "r", "raise", "range", "remove", "return", "reverse", "run_mem_encoder", "self", "squeeze", "store_output", "streaming", "torch", "tracking"], "sam3_tracker_video/modeling_sam3_tracker_video.py:Sam3TrackerVideoModel.get_image_features": ["FloatTensor", "ModelModel", "ModelPreTrainedModel", "ModelVisionEncoderOutput", "None", "Tensor", "TransformersKwargs", "Unpack", "attentions", "class", "conv_s0", "conv_s1", "def", "feature_map", "feature_map_position_embedding", "feature_maps", "feature_maps_position_embeddings", "flatten", "for", "fpn_hidden_states", "fpn_position_encoding", "get_image_features", "hidden_states", "in", "kwargs", "list", "mask_decoder", "permute", "pixel_values", "r", "return", "self", "torch", "tuple", "vision_encoder", "vision_outputs"], "sam3_tracker_video/modeling_sam3_tracker_video.py:Sam3TrackerVideoModel._prepare_vision_features": ["ModelInferenceSession", "ModelModel", "ModelPreTrainedModel", "Tensor", "_", "_prepare_vision_features", "batch_size", "cache", "cache_vision_features", "cached_features", "class", "def", "else", "expand", "for", "frame_idx", "get_frame", "get_image_features", "get_vision_features", "if", "image_batch", "in", "inference_session", "int", "list", "pe", "return", "self", "torch", "tuple", "unsqueeze", "vision_feats", "vision_pos_embeds"], "sam3_tracker_video/modeling_sam3_tracker_video.py:Sam3TrackerVideoModel._single_frame_forward": ["Exactly", "F", "False", "FloatTensor", "Got", "LongTensor", "ModelImageSegmentationOutput", "ModelModel", "ModelPreTrainedModel", "Model_output_token", "Model_output_tokens", "NO_OBJ_SCORE", "None", "TransformersKwargs", "True", "Unpack", "ValueError", "You", "_", "_single_frame_forward", "align_corners", "and", "antialias", "arange", "argmax", "as", "attention_similarity", "backbone_feature_sizes", "batch_inds", "batch_size", "be", "best_iou_inds", "bilinear", "bool", "bounding", "box", "boxes", "class", "def", "dense_embeddings", "dense_prompt_embeddings", "device", "dim", "dtype", "elif", "else", "f", "feat", "feat_size", "feature_maps", "float", "for", "get_image_features", "get_image_wide_positional_embeddings", "high_res_masks", "high_res_multimasks", "high_resolution_features", "if", "image_embeddings", "image_positional_embeddings", "image_size", "in", "input", "input_boxes", "input_labels", "input_masks", "input_points", "int", "int32", "interpolate", "iou_scores", "is", "is_obj_appearing", "kwargs", "lambda_is_obj_appearing", "low_res_masks", "low_res_multimasks", "many", "mask_decoder", "mask_input_size", "mode", "multimask_output", "must", "no_memory_embedding", "no_object_pointer", "not", "num_objects", "object_batch_inds", "object_pointer", "object_pointer_proj", "object_score_logits", "of", "one", "ones", "ones_like", "or", "per", "permute", "pixel_values", "points", "pred_masks", "prompt_encoder", "provide", "provided", "raise", "repeat", "return", "self", "shape", "should", "size", "sparse_embeddings", "sparse_prompt_embeddings", "squeeze", "target_embedding", "to", "torch", "unsqueeze", "view", "vision_attentions", "vision_hidden_states", "where", "zeros", "zip"], "sam3_tracker_video/modeling_sam3_tracker_video.py:Sam3TrackerVideoModel._use_mask_as_output": ["F", "False", "ModelImageSegmentationOutput", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "True", "_single_frame_forward", "_use_mask_as_output", "align_corners", "antialias", "any", "backbone_features", "bilinear", "class", "def", "dim", "dtype", "flatten", "float", "high_res_features", "high_res_masks", "if", "image_embeddings", "image_size", "input_masks", "interpolate", "iou_scores", "is_obj_appearing", "lambda_is_obj_appearing", "list", "low_res_masks", "mask_downModelple", "mask_input_size", "mask_inputs", "mask_inputs_float", "mode", "new_ones", "no_object_pointer", "object_pointer", "object_score_logits", "out_bias", "out_scale", "pred_masks", "prompt_encoder", "return", "self", "shape", "size", "to", "torch", "unsqueeze"], "sam3_tracker_video/modeling_sam3_tracker_video.py:Sam3TrackerVideoModel._select_closest_cond_frames": ["ModelModel", "ModelPreTrainedModel", "None", "_select_closest_cond_frames", "abs", "class", "cond_frame_outputs", "def", "default", "else", "for", "frame_idx", "idx_after", "idx_before", "if", "in", "inds_remain", "is", "items", "key", "lambda", "len", "max", "max_cond_frame_num", "min", "not", "num_remain", "or", "return", "selected_outputs", "self", "sorted", "t", "unselected_outputs", "update", "v", "x"], "sam3_tracker_video/modeling_sam3_tracker_video.py:Sam3TrackerVideoModel._gather_memory_frame_outputs": ["False", "ModelInferenceSession", "ModelModel", "ModelPreTrainedModel", "None", "ValueError", "_gather_memory_frame_outputs", "_select_closest_cond_frames", "append", "be", "bool", "cannot", "class", "cond_frame_outputs", "conditioning", "conditioning_outputs", "config", "def", "dict", "else", "empty", "for", "frame_idx", "get", "if", "in", "inference_session", "int", "is_initial_conditioning_frame", "list", "maskmem_features", "max_cond_frame_num", "non_cond_frame_outputs", "not", "num_maskmem", "obj_idx", "out", "output_data", "output_dict_per_obj", "outputs", "previous_frame_idx", "raise", "range", "relative_temporal_offset", "return", "self", "temporal_positions_and_previous_outputs", "track_in_reverse_time", "tuple", "unselected_conditioning_outputs", "values", "when"], "sam3_tracker_video/modeling_sam3_tracker_video.py:Sam3TrackerVideoModel._build_memory_attention_inputs": ["ModelModel", "ModelPreTrainedModel", "None", "Tensor", "True", "_build_memory_attention_inputs", "append", "class", "combined_memory_pos_embed", "continue", "def", "device", "dict", "for", "if", "in", "int", "is", "list", "maskmem_features", "maskmem_pos_enc", "memories_to_concatenate", "memory_features", "memory_positional_embeddings_to_concatenate", "memory_temporal_positional_encoding", "non_blocking", "prev_output_data", "relative_temporal_offset", "return", "self", "spatial_memory_pos_embed", "temporal_positions_and_previous_outputs", "to", "torch", "tuple"], "sam3_tracker_video/modeling_sam3_tracker_video.py:Sam3TrackerVideoModel._get_object_pointers": ["False", "ModelInferenceSession", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "_get_object_pointers", "and", "append", "bool", "break", "class", "cond_frame_outputs", "conditioning_outputs", "config", "def", "device", "eligible_conditioning_outputs", "else", "for", "frame_idx", "get", "if", "in", "inference_session", "int", "is", "items", "list", "max_object_pointers_in_encoder", "max_object_pointers_to_use", "min", "non_cond_frame_outputs", "not", "num_total_frames", "obj_idx", "object_pointer", "or", "out", "out_data", "output_dict_per_obj", "pointer_tokens", "range", "ref_frame_idx", "return", "self", "streaming", "t_diff_offset", "temporal_difference", "temporal_idx", "temporal_offsets", "temporal_position_sign_multiplier", "to", "torch", "track_in_reverse_time", "training", "tuple"], "sam3_tracker_video/modeling_sam3_tracker_video.py:Sam3TrackerVideoModel._process_object_pointers": ["ModelModel", "ModelPreTrainedModel", "None", "Tensor", "_process_object_pointers", "batch_size", "class", "config", "def", "device", "dim", "dtype", "else", "enable_temporal_pos_encoding_for_object_pointers", "expand", "flatten", "float", "float32", "get_1d_sine_pe", "if", "int", "len", "list", "max_object_pointers_to_use", "max_temporal_diff", "mem_dim", "new_zeros", "normalized_temporal_diffs", "not", "num_channels", "num_splits", "object_pointers", "object_pointers_pos_embed", "permute", "pointer_tokens", "pointer_tpos_dim", "projected_sine_pe", "repeat_interleave", "reshape", "return", "self", "sine_pe", "stack", "temporal_offsets", "temporal_positional_encoding_projection_layer", "tensor", "to", "torch", "tuple", "unsqueeze"], "sam3_tracker_video/modeling_sam3_tracker_video.py:Sam3TrackerVideoModel._prepare_memory_conditioned_features": ["False", "ModelInferenceSession", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "_build_memory_attention_inputs", "_gather_memory_frame_outputs", "_get_object_pointers", "_prepare_memory_conditioned_features", "_process_object_pointers", "append", "backbone_feature_sizes", "batch_size", "bool", "cat", "class", "combined_memory", "combined_memory_positional_embeddings", "conditioned_feature_map", "conditioned_feature_map_flat", "current_feature_map", "current_vision_features", "current_vision_position_embeddings", "current_vision_positional_embeddings", "def", "device", "dim", "frame_idx", "height", "hidden_dim", "if", "inference_session", "int", "is", "is_initial_conditioning_frame", "list", "max_object_pointers_to_use", "memories_to_concatenate", "memory", "memory_attention", "memory_posision_embeddings", "memory_positional_embeddings_to_concatenate", "no_memory_embedding", "not", "num_channels", "num_maskmem", "num_object_pointer_tokens", "num_total_frames", "obj_idx", "object_pointers", "object_pointers_pos_embed", "permute", "pointer_tokens", "return", "self", "shape", "size", "squeeze", "streaming", "temporal_offsets", "temporal_positions_and_previous_outputs", "torch", "track_in_reverse_time", "view", "width"], "sam3_tracker_video/modeling_sam3_tracker_video.py:Sam3TrackerVideoModel._use_multimask": ["ModelModel", "ModelPreTrainedModel", "None", "_use_multimask", "and", "bool", "class", "config", "def", "dict", "else", "if", "is", "is_init_cond_frame", "multimask_max_pt_num", "multimask_min_pt_num", "multimask_output", "multimask_output_for_tracking", "multimask_output_in_Model", "num_pts", "or", "point_inputs", "point_labels", "return", "self", "size"], "sam3_tracker_video/modeling_sam3_tracker_video.py:Sam3TrackerVideoModel._run_single_frame_inference": ["Any", "False", "ModelInferenceSession", "ModelModel", "ModelPreTrainedModel", "Model_outputs", "Modele", "None", "Tensor", "ValueError", "_prepare_memory_conditioned_features", "_prepare_vision_features", "_run_single_frame_inference", "_single_frame_forward", "_use_mask_as_output", "_use_multimask", "and", "appear", "as", "backbone_feature_sizes", "batch_size", "bool", "class", "current_out", "current_vision_feats", "current_vision_features", "current_vision_pos_embeds", "current_vision_positional_embeddings", "def", "dict", "else", "for", "frame", "frame_idx", "hidden_dim", "high_res_features", "high_res_masks", "if", "image_embeddings", "in", "inference_session", "input", "input_labels", "input_masks", "input_points", "int", "is", "is_init_cond_frame", "is_initial_conditioning_frame", "len", "mask_inputs", "multimask_output", "not", "num_frames", "num_total_frames", "obj_idx", "object_pointer", "object_score_logits", "on", "permute", "pix_feat", "pixel_values", "point_coords", "point_inputs", "point_labels", "pred_masks", "prev_Model_mask_logits", "raise", "return", "reverse", "s", "self", "should", "simultaneously", "size", "str", "streaming", "the", "torch", "track_in_reverse_time", "training", "view", "x", "zip"], "sam3_tracker_video/modeling_sam3_tracker_video.py:Sam3TrackerVideoModel._encode_new_memory": ["F", "False", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "True", "_encode_new_memory", "align_corners", "and", "antialias", "backbone_feature_sizes", "batch_size", "bfloat16", "bilinear", "bool", "channels", "class", "config", "current_vision_feats", "def", "dtype", "else", "expand", "flatten", "float", "height", "hidden_dim", "if", "interpolate", "is", "is_mask_from_pts", "is_obj_appearing", "list", "mask_for_mem", "mask_input_size", "mask_input_size_h", "mask_input_size_w", "mask_mem_size_h", "mask_mem_size_w", "maskmem_features", "maskmem_pos_enc", "memory_encoder", "mode", "not", "object_score_logits", "occlusion_spatial_embedding_parameter", "permute", "pix_feat", "pred_masks_high_res", "prompt_encoder", "return", "self", "shape", "sigmoid", "sigmoid_bias_for_mem_enc", "sigmoid_scale_for_mem_enc", "size", "to", "torch", "training", "tuple", "view", "width"], "sam3_tracker_video/modeling_sam3_tracker_video.py:Sam3TrackerVideoModel._batch_encode_memories": ["ModelInferenceSession", "ModelModel", "ModelPreTrainedModel", "Tensor", "_", "_batch_encode_memories", "_encode_new_memory", "_prepare_vision_features", "any", "batch_size", "bool", "cat", "class", "cond_frame_outputs", "current_vision_feats", "def", "dim", "else", "enumerate", "expand", "expanded_vision_feats", "for", "frame_idx", "high_res_masks_batched", "high_res_masks_for_memory", "i", "if", "in", "inference_session", "int", "is_mask_from_pts", "is_mask_from_pts_per_obj", "len", "list", "maskmem_features", "maskmem_features_batched", "maskmem_pos_enc", "maskmem_pos_enc_batched", "non_cond_frame_outputs", "not", "obj_idx", "object_score_logits", "object_score_logits_batched", "object_score_logits_for_memory", "objects_needing_memory_encoding", "output_dict", "output_dict_per_obj", "pred_masks_high_res", "return", "self", "storage_key", "torch"], "sam3_tracker_video/modeling_sam3_tracker_video.py:Sam3TrackerVideoModel.propagate_in_video_iterator": ["Cannot", "False", "Iterator", "ModelInferenceSession", "ModelModel", "ModelPreTrainedModel", "ModelSegmentationOutput", "Model_output", "None", "ValueError", "a", "bool", "class", "cond_frame_outputs", "def", "desc", "determine", "disable", "else", "end_frame_idx", "first", "for", "frame", "frame_idx", "frames_with_inputs", "if", "in", "index", "inference", "inference_session", "inputs", "int", "is", "it", "manually", "max", "max_frame_num_to_track", "min", "not", "num_frames", "obj_output_dict", "on", "or", "output_dict_per_obj", "please", "processing_order", "propagate", "propagate_in_video_iterator", "r", "raise", "range", "reverse", "run", "self", "show_progress_bar", "specify", "start_frame_idx", "starting", "the", "tqdm", "values", "video", "with", "yield"], "sam3_video/modeling_sam3_video.py:_load_cv_utils_kernel_once": ["Exception", "Failed", "False", "Install", "NMS", "None", "_load_cv_utils_kernel_once", "and", "as", "be", "better", "community", "cuda", "cv_utils", "cv_utils_kernel", "def", "e", "except", "f", "filling", "for", "get_kernel", "global", "hole", "if", "install", "installed", "is", "is_kernels_available", "it", "kernel", "kernels", "library", "load", "logger", "mask", "may", "not", "pip", "post", "processing", "quality", "removal", "return", "setup", "skipped", "sprinkle", "supported", "to", "torch", "try", "warning_once", "will", "with", "your"], "sam3_video/modeling_sam3_video.py:Sam3VideoInferenceCache.__init__": ["ModelInferenceCache", "__init__", "_vision_features", "class", "cpu", "def", "device", "inference_device", "inference_state_device", "int", "max_vision_features_cache_size", "self", "str", "torch"], "sam3_video/modeling_sam3_video.py:Sam3VideoInferenceCache.cache_vision_features": ["ModelInferenceCache", "Tensor", "True", "_vision_features", "and", "cache_vision_features", "cached", "class", "def", "dict", "elif", "else", "features", "for", "frame_idx", "if", "in", "inference_state_device", "int", "isinstance", "items", "key", "keys", "len", "list", "max_vision_features_cache_size", "min", "non_blocking", "pop", "self", "to", "torch", "tuple", "v", "value"], "sam3_video/modeling_sam3_video.py:Sam3VideoInferenceCache.get_vision_features": ["ModelInferenceCache", "None", "Tensor", "True", "_vision_features", "and", "cached", "class", "def", "dict", "elif", "else", "for", "frame_idx", "get_vision_features", "if", "in", "inference_device", "int", "isinstance", "items", "key", "list", "moved", "non_blocking", "not", "return", "self", "to", "torch", "tuple", "v", "value"], "sam3_video/modeling_sam3_video.py:Sam3VideoInferenceCache.clear_all": ["ModelInferenceCache", "_vision_features", "class", "clear", "clear_all", "def", "self"], "sam3_video/modeling_sam3_video.py:Sam3VideoInferenceSession.__init__": ["FloatTensor", "ModelInferenceCache", "ModelInferenceSession", "None", "OrderedDict", "__init__", "_obj_id_to_idx", "_obj_idx_to_id", "cache", "class", "cpu", "def", "defaultdict", "device", "dict", "dtype", "else", "enumerate", "float32", "frames_tracked_per_obj", "hotstart_removed_obj_ids", "if", "inference_device", "inference_state_device", "int", "is", "list", "mask_inputs_per_obj", "max_obj_id", "max_vision_features_cache_size", "not", "obj_first_frame_idx", "obj_id_to_last_occluded", "obj_id_to_prompt_id", "obj_id_to_score", "obj_id_to_tracker_score_frame_wise", "obj_ids", "output_buffer", "output_dict_per_obj", "overlap_pair_to_frame_inds", "point_inputs_per_obj", "processed_frames", "prompt_attention_masks", "prompt_embeddings", "prompt_input_ids", "prompts", "removed_obj_ids", "self", "set", "str", "suppressed_obj_ids", "to", "torch", "trk_keep_alive", "unmatched_frame_inds", "video", "video_height", "video_storage_device", "video_width"], "sam3_video/modeling_sam3_video.py:Sam3VideoInferenceSession.num_frames": ["ModelInferenceSession", "None", "class", "def", "else", "if", "int", "is", "len", "not", "num_frames", "processed_frames", "return", "self"], "sam3_video/modeling_sam3_video.py:Sam3VideoInferenceSession.add_prompt": ["ModelInferenceSession", "add_prompt", "class", "def", "for", "if", "in", "int", "items", "len", "prompt_id", "prompt_text", "prompts", "return", "self", "str", "text"], "sam3_video/modeling_sam3_video.py:Sam3VideoInferenceSession.obj_id_to_idx": ["ModelInferenceSession", "_obj_id_to_idx", "_obj_idx_to_id", "append", "class", "cond_frame_outputs", "def", "frames_tracked_per_obj", "if", "in", "int", "len", "mask_inputs_per_obj", "non_cond_frame_outputs", "not", "obj_id", "obj_id_to_idx", "obj_ids", "obj_idx", "output_dict_per_obj", "point_inputs_per_obj", "return", "self"], "sam3_video/modeling_sam3_video.py:Sam3VideoInferenceSession.obj_idx_to_id": ["ModelInferenceSession", "_obj_idx_to_id", "class", "def", "int", "obj_idx", "obj_idx_to_id", "return", "self"], "sam3_video/modeling_sam3_video.py:Sam3VideoInferenceSession.get_obj_num": ["ModelInferenceSession", "_obj_idx_to_id", "class", "def", "get_obj_num", "int", "len", "return", "self"], "sam3_video/modeling_sam3_video.py:Sam3VideoInferenceSession.add_mask_inputs": ["ModelInferenceSession", "Tensor", "True", "add_mask_inputs", "class", "def", "dtype", "frame_idx", "inference_device", "inputs", "int", "mask_inputs_per_obj", "non_blocking", "obj_idx", "self", "to", "torch"], "sam3_video/modeling_sam3_video.py:Sam3VideoInferenceSession.remove_mask_inputs": ["ModelInferenceSession", "None", "class", "def", "frame_idx", "int", "mask_inputs_per_obj", "obj_idx", "pop", "remove_mask_inputs", "self"], "sam3_video/modeling_sam3_video.py:Sam3VideoInferenceSession.remove_object": ["All", "Cannot", "False", "ModelInferenceSession", "None", "RuntimeError", "_map_keys", "_obj_id_to_idx", "_obj_idx_to_id", "append", "as", "bool", "class", "container", "copy", "def", "dict", "doesn", "exist", "existing", "f", "for", "frames_tracked_per_obj", "get", "id", "ids", "if", "in", "int", "is", "it", "k", "len", "list", "mask_inputs_per_obj", "new_kvs", "new_obj_ids", "new_obj_inds", "not", "obj_id", "obj_id_to_prompt_id", "obj_ids", "object", "old_idx", "old_idx_to_new_idx", "old_obj_ids", "old_obj_idx_to_rm", "old_obj_inds", "output_dict_per_obj", "point_inputs_per_obj", "pop", "raise", "range", "remain_old_obj_inds", "remove", "remove_object", "reset_inference_session", "return", "self", "strict", "t", "update", "v", "zip"], "sam3_video/modeling_sam3_video.py:Sam3VideoInferenceSession.store_output": ["ModelInferenceSession", "None", "Tensor", "True", "and", "bool", "class", "cond_frame_outputs", "def", "dict", "elif", "else", "for", "frame_idx", "if", "in", "inference_state_device", "int", "is", "is_conditioning_frame", "isinstance", "items", "key", "non_blocking", "non_cond_frame_outputs", "obj_idx", "object_pointer", "object_score_logits", "output_dict_per_obj", "output_key", "output_value", "return", "self", "storage_key", "store_output", "str", "to", "torch", "value"], "sam3_video/modeling_sam3_video.py:Sam3VideoInferenceSession.get_output": ["ModelInferenceSession", "None", "Tensor", "True", "bool", "class", "cond_frame_outputs", "def", "else", "frame_idx", "get", "get_output", "if", "inference_device", "int", "is", "is_conditioning_frame", "isinstance", "non_blocking", "non_cond_frame_outputs", "obj_idx", "out", "output_dict_per_obj", "output_key", "return", "self", "storage_key", "str", "to", "torch", "value"], "sam3_video/modeling_sam3_video.py:Sam3VideoInferenceSession.add_new_frame": ["ModelInferenceSession", "None", "Tensor", "True", "add_new_frame", "class", "def", "dim", "dtype", "else", "frame_idx", "if", "int", "is", "len", "non_blocking", "not", "pixel_values", "processed_frames", "return", "self", "squeeze", "to", "torch", "video_storage_device"], "sam3_video/modeling_sam3_video.py:Sam3VideoInferenceSession.get_frame": ["ModelInferenceSession", "Tensor", "True", "class", "def", "frame_idx", "get_frame", "inference_device", "int", "non_blocking", "processed_frames", "return", "self", "to", "torch"], "sam3_video/modeling_sam3_video.py:Sam3VideoInferenceSession.reset_tracking_data": ["ModelInferenceSession", "_obj_id_to_idx", "_obj_idx_to_id", "class", "clear", "def", "frames_tracked_per_obj", "obj_id_to_prompt_id", "obj_ids", "output_dict_per_obj", "reset_tracking_data", "self"], "sam3_video/modeling_sam3_video.py:Sam3VideoInferenceSession.reset_inference_session": ["ModelInferenceSession", "_obj_id_to_idx", "_obj_idx_to_id", "cache", "class", "clear", "clear_all", "def", "frames_tracked_per_obj", "obj_id_to_prompt_id", "obj_ids", "output_dict_per_obj", "reset_inference_session", "self"], "sam3_video/modeling_sam3_video.py:Sam3VideoInferenceSession.reset_state": ["ModelInferenceSession", "OrderedDict", "_obj_id_to_idx", "_obj_idx_to_id", "cache", "class", "clear", "clear_all", "def", "defaultdict", "dict", "frames_tracked_per_obj", "list", "max_obj_id", "obj_first_frame_idx", "obj_id_to_last_occluded", "obj_id_to_prompt_id", "obj_id_to_score", "obj_id_to_tracker_score_frame_wise", "obj_ids", "output_buffer", "output_dict_per_obj", "overlap_pair_to_frame_inds", "prompt_attention_masks", "prompt_embeddings", "prompt_input_ids", "prompts", "removed_obj_ids", "reset_state", "self", "set", "suppressed_obj_ids", "trk_keep_alive", "unmatched_frame_inds"], "sam3_video/modeling_sam3_video.py:Sam3VideoModel.__init__": ["AutoModel", "ModelConfig", "ModelModel", "ModelPreTrainedModel", "ModelVisionNeck", "True", "__init__", "assoc_iou_thresh", "class", "config", "decrease_trk_keep_alive_for_empty_masklets", "def", "det_nms_thresh", "detector_config", "detector_model", "eval", "fill_hole_area", "from_config", "high_conf_thresh", "high_iou_thresh", "hotstart_delay", "hotstart_dup_thresh", "hotstart_unmatch_thresh", "init_trk_keep_alive", "low_res_mask_size", "max_num_objects", "max_trk_keep_alive", "min_trk_keep_alive", "new_det_thresh", "post_init", "recondition_every_nth_frame", "recondition_on_trk_masks", "remove_vision_encoder", "score_threshold_detection", "self", "super", "suppress_overlapping_based_on_recent_occlusion_threshold", "suppress_unmatched_only_within_hotstart", "tracker_config", "tracker_model", "tracker_neck", "trk_assoc_iou_thresh", "vision_config"], "sam3_video/modeling_sam3_video.py:Sam3VideoModel.get_vision_features_for_tracker": ["ModelModel", "ModelPreTrainedModel", "Tensor", "batch_size", "class", "conv_s0", "conv_s1", "def", "feature_map", "feature_map_position_embedding", "feature_maps", "feature_maps_position_embeddings", "flatten", "for", "fpn_hidden_states", "fpn_position_encoding", "get_vision_features_for_tracker", "height", "hidden_states", "hidden_states_spatial", "image_embedding_size", "in", "last_hidden_state", "list", "mask_decoder", "permute", "prompt_encoder", "return", "self", "shape", "torch", "tracker_model", "tracker_neck", "view", "vision_embeds", "width"], "sam3_video/modeling_sam3_video.py:Sam3VideoModel.run_detection": ["ModelInferenceSession", "ModelModel", "ModelPreTrainedModel", "No", "Please", "Tensor", "ValueError", "add", "all_detections", "attention_mask", "available", "bbox", "class", "def", "det_nms_thresh", "det_out", "detection", "detector_model", "detector_outputs", "else", "first", "for", "get_text_features", "if", "in", "inference_session", "input_ids", "iou_threshold", "keep", "keys", "list", "mask", "nms_masks", "not", "pos_pred_idx", "pred_boxes", "pred_boxes_xyxy", "pred_logits", "pred_masks", "pred_probs", "presence_logits", "presence_scores", "prob_threshold", "prompt_attention_masks", "prompt_embeddings", "prompt_id", "prompt_ids", "prompt_input_ids", "prompts", "raise", "return", "run_detection", "run_nms", "score_threshold_detection", "scores", "self", "session", "sigmoid", "text_embeds", "the", "to", "torch", "vision_embeds", "where"], "sam3_video/modeling_sam3_video.py:Sam3VideoModel.run_tracker_propagation": ["False", "H_mask", "ModelInferenceSession", "ModelModel", "ModelPreTrainedModel", "True", "W_mask", "append", "bool", "cat", "class", "def", "device", "dim", "else", "fill_hole_area", "fill_holes", "fill_holes_in_mask_scores", "frame_idx", "if", "inference_session", "int", "len", "low_res_mask_size", "low_res_masks", "low_res_masks_list", "max_area", "obj_ids", "obj_scores", "obj_scores_list", "object_score_logits", "out", "out_low_res_masks", "out_obj_scores", "pred_masks", "remove_sprinkles", "return", "reverse", "run_mem_encoder", "run_tracker_propagation", "self", "squeeze", "torch", "tracker_model", "unsqueeze", "zeros"], "sam3_video/modeling_sam3_video.py:Sam3VideoModel._associate_det_trk": ["ModelModel", "ModelPreTrainedModel", "Tensor", "_associate_det_trk", "any", "argmax", "assoc_iou_thresh", "class", "def", "det_idx", "det_is_high_conf", "det_is_high_conf_and_iou", "det_is_high_iou", "det_masks", "det_masks_binary", "det_matches_any_trk", "det_max_iou", "det_prompt_ids", "det_scores", "det_to_matched_trk_obj_ids", "det_to_max_iou_trk_idx", "device", "dim", "dtype", "elif", "else", "empty", "empty_trk_obj_ids", "for", "high_conf_and_iou_mask", "high_conf_thresh", "high_iou_thresh", "if", "in", "int", "iou_threshold", "iou_threshold_trk", "ious", "is_new_det", "item", "list", "long", "mask_iou", "matched_trk_mask", "max", "new_det_out_inds", "new_det_thresh", "prompt_match", "range", "return", "self", "size", "tensor", "tolist", "torch", "trk_assoc_iou_thresh", "trk_id_to_max_iou_high_conf_det", "trk_idx", "trk_is_matched", "trk_is_nonempty", "trk_is_unmatched", "trk_masks", "trk_masks_binary", "trk_obj_id", "trk_obj_ids", "trk_obj_ids_tensor", "trk_prompt_ids", "unmatched_trk_obj_ids", "unsqueeze", "where", "zeros_like"], "sam3_video/modeling_sam3_video.py:Sam3VideoModel._process_hotstart": ["Any", "False", "ModelInferenceSession", "ModelModel", "ModelPreTrainedModel", "Removing", "Suppressing", "_process_hotstart", "add", "and", "another", "append", "at", "being", "bool", "class", "continue", "debug", "decrease_trk_keep_alive_for_empty_masklets", "def", "det_to_matched_trk_obj_ids", "dict", "due", "else", "empty_trk_obj_ids", "extra_metadata", "f", "first_appear_obj_id", "first_obj_id", "for", "frame", "frame_idx", "frame_indices", "frames", "hotstart_delay", "hotstart_diff", "hotstart_dup_thresh", "hotstart_unmatch_thresh", "if", "in", "inference_session", "info", "init_trk_keep_alive", "int", "is", "is_within_hotstart", "it", "items", "key", "lambda", "len", "list", "logger", "matched_trk_obj_ids", "matched_trks", "matched_trks_per_det", "max", "max_trk_keep_alive", "min", "min_trk_keep_alive", "new_det_obj_ids", "not", "obj_first_frame_idx", "obj_id", "obj_ids_newly_removed", "object", "or", "overlap_pair_to_frame_inds", "overlaps", "removed_obj_ids", "return", "reverse", "self", "set", "since", "str", "streaming", "suppress_unmatched_only_within_hotstart", "suppressed_obj_ids", "to", "track", "trk_keep_alive", "unmatched", "unmatched_frame_inds", "unmatched_trk_obj_ids", "update", "values", "with", "x"], "sam3_video/modeling_sam3_video.py:Sam3VideoModel.run_memory_encoder": ["False", "ModelInferenceSession", "ModelModel", "ModelPreTrainedModel", "Tensor", "_encode_new_memory", "cache", "cached_features", "class", "current_vision_feats", "def", "frame_idx", "get_vision_features", "high_res_masks", "inference_session", "int", "is_mask_from_pts", "maskmem_features", "maskmem_pos_enc", "object_score_logits", "pred_masks_high_res", "return", "run_memory_encoder", "self", "torch", "tracker_model", "vision_feats"], "sam3_video/modeling_sam3_video.py:Sam3VideoModel._prepare_recondition_masks": ["ModelInferenceSession", "ModelModel", "ModelPreTrainedModel", "Tensor", "_prepare_recondition_masks", "add", "class", "continue", "def", "det_idx", "det_out", "dict", "else", "for", "frame_idx", "high_conf_thresh", "if", "in", "inference_session", "int", "items", "mask", "new_mask", "obj_id_to_idx", "obj_idx", "obj_score", "recondition_on_trk_masks", "reconditioned_masks", "reconditioned_obj_ids", "return", "self", "set", "str", "tracker_obj_scores_global", "trk_id_to_max_iou_high_conf_det", "trk_masks", "trk_obj_id", "unsqueeze"], "sam3_video/modeling_sam3_video.py:Sam3VideoModel._get_objects_to_suppress_based_on_most_recently_occluded": ["False", "ModelModel", "ModelPreTrainedModel", "Tensor", "_get_objects_to_suppress_based_on_most_recently_occluded", "any", "binary_low_res_masks", "bool", "class", "cmp_op", "def", "device", "diagonal", "dim", "dtype", "else", "gt", "if", "int", "iou", "last_occ_expanded_i", "last_occ_expanded_j", "last_occluded", "len", "list", "lt", "mask_iou", "mask_iou_thresh", "not", "obj_ids", "overlapping_pairs", "return", "reverse", "self", "size", "suppress_i_mask", "suppress_j_mask", "suppress_overlapping_based_on_recent_occlusion_threshold", "to_suppress", "torch", "triu", "unsqueeze", "zeros"], "sam3_video/modeling_sam3_video.py:Sam3VideoModel._suppress_overlapping_based_on_recent_occlusion": ["ALWAYS_OCCLUDED", "Any", "False", "ModelInferenceSession", "ModelModel", "ModelPreTrainedModel", "NEVER_OCCLUDED", "NO_OBJ_LOGIT", "Tensor", "True", "_get_objects_to_suppress_based_on_most_recently_occluded", "_suppress_overlapping_based_on_recent_occlusion", "any", "as_tuple", "batch_size", "binary_tracker_low_res_masks_global", "bool", "cat", "class", "clone", "continue", "def", "device", "dict", "dim", "dtype", "else", "enumerate", "fill_value", "for", "frame_idx", "full", "get", "idx", "if", "in", "inference_session", "int", "is_obj_occluded", "is_obj_occluded_or_suppressed", "last_occluded_new", "last_occluded_prev", "long", "nonzero", "not", "numel", "obj_id", "obj_id_to_last_occluded", "obj_id_to_prompt_id", "obj_ids", "obj_ids_global", "obj_ids_newly_removed", "obj_idx", "prompt_id", "prompt_ids_global", "prompt_indices", "prompt_last_occ", "prompt_mask", "prompt_masks", "prompt_obj_ids", "prompt_suppress", "return", "reverse", "self", "set", "size", "sorted", "str", "tensor", "to_suppress", "tolist", "torch", "tracker_low_res_masks_global", "tracker_metadata_new", "unique", "unique_prompts", "zeros"], "sam3_video/modeling_sam3_video.py:Sam3VideoModel._apply_non_overlapping_constraints": ["ModelModel", "ModelPreTrainedModel", "None", "True", "_apply_non_overlapping_constraints", "arange", "argmax", "batch_obj_inds", "batch_size", "clamp", "class", "def", "device", "dim", "if", "keep", "keepdim", "max", "max_obj_inds", "pred_masks", "return", "self", "size", "torch", "where"], "sam3_video/modeling_sam3_video.py:Sam3VideoModel._suppress_shrinked_masks": ["ModelModel", "ModelPreTrainedModel", "None", "_suppress_shrinked_masks", "area_after", "area_before", "area_ratio", "clamp", "class", "def", "dim", "expand_as", "keep", "keep_mask", "max", "min", "new_pred_masks", "pred_masks", "pred_masks_after", "return", "self", "shrink_threshold", "sum", "torch", "where"], "sam3_video/modeling_sam3_video.py:Sam3VideoModel._suppress_object_pw_area_shrinkage": ["ModelModel", "ModelPreTrainedModel", "Modele", "None", "True", "ValueError", "_suppress_object_pw_area_shrinkage", "_suppress_object_pw_area_shrinkage_impl", "as", "as_tuple", "class", "clone", "continue", "def", "device", "dtype", "for", "have", "if", "in", "indices", "int", "is", "len", "length", "list", "long", "must", "nonzero", "numel", "pred_masks", "pred_masks_grouped", "prompt_id", "prompt_ids", "prompt_ids_tensor", "raise", "return", "self", "size", "sorted", "tensor", "the", "torch", "unique"], "sam3_video/modeling_sam3_video.py:Sam3VideoModel._suppress_object_pw_area_shrinkage_impl": ["ModelModel", "ModelPreTrainedModel", "_apply_non_overlapping_constraints", "_suppress_object_pw_area_shrinkage_impl", "_suppress_shrinked_masks", "class", "def", "if", "pixel_level_non_overlapping_masks", "pred_masks", "return", "self", "size"], "sam3_video/modeling_sam3_video.py:Sam3VideoModel._tracker_update_memories": ["ModelInferenceSession", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "_suppress_object_pw_area_shrinkage", "_tracker_update_memories", "any", "class", "cond_frame_outputs", "continue", "current_out", "def", "dict", "dim", "float", "for", "frame_idx", "high_res_masks", "if", "in", "inference_session", "int", "is", "items", "len", "low_res_masks", "maskmem_features", "maskmem_features_batched", "maskmem_pos_enc", "maskmem_pos_enc_batched", "non_cond_frame_outputs", "not", "num_objects", "obj_id", "obj_id_to_prompt_id", "obj_ids", "obj_idx", "object_score_logits", "object_score_logits_batched", "output_dict", "output_dict_per_obj", "pop", "prompt_ids_for_objects", "range", "recond_mask", "reconditioned_masks", "return", "run_memory_encoder", "self", "storage_key", "torch", "unsqueeze", "where"], "sam3_video/modeling_sam3_video.py:Sam3VideoModel.run_tracker_update_planning_phase": ["False", "ModelInferenceSession", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "True", "_", "_associate_det_trk", "_prepare_recondition_masks", "_process_hotstart", "_suppress_overlapping_based_on_recent_occlusion", "_tracker_update_memories", "and", "batch_size", "bool", "class", "deepcopy", "def", "det_idx", "det_idx_to_prompt_id", "det_mask_preds", "det_masks", "det_out", "det_prompt_ids", "det_scores", "det_to_matched_trk_obj_ids", "device", "dict", "dtype", "else", "empty", "empty_trk_obj_ids", "extra_metadata", "extra_metadata_new", "f", "float", "for", "frame_idx", "hitting", "i", "idx", "if", "in", "inference_session", "int", "k", "largest", "len", "list", "logger", "long", "low_res_masks", "mask", "max", "max_num_objects", "max_obj_id", "new_det_inds_tensor", "new_det_num", "new_det_num_to_keep", "new_det_obj_ids", "new_det_out_inds", "new_det_scores", "new_det_start_obj_id", "not", "num_obj_dropped_due_to_limit", "numel", "obj_first_frame_idx", "obj_id", "obj_id_to_last_occluded", "obj_id_to_prompt_id", "obj_id_to_score", "obj_id_to_tracker_score_frame_wise", "obj_ids", "obj_ids_newly_removed", "overlap_pair_to_frame_inds", "pop", "prev_obj_num", "prompt_id", "range", "recondition_every_nth_frame", "reconditioned_masks", "reconditioned_obj_ids", "removed_obj_ids", "return", "reverse", "run_tracker_update_planning_phase", "scores", "scores_for_new_dets", "self", "set", "should_recondition_periodic", "size", "str", "streaming", "suppress_overlapping_based_on_recent_occlusion_threshold", "suppressed_obj_ids", "tensor", "tolist", "top_inds", "topk", "torch", "tracker_low_res_masks_global", "tracker_metadata_new", "tracker_obj_scores_global", "tracker_update_plan", "trk_id_to_max_iou_high_conf_det", "trk_keep_alive", "trk_masks", "trk_obj_ids", "trk_prompt_ids", "unmatched_frame_inds", "unmatched_trk_obj_ids", "update", "updated_obj_ids", "warning", "with", "zip"], "sam3_video/modeling_sam3_video.py:Sam3VideoModel._tracker_add_new_objects": ["False", "ModelInferenceSession", "ModelModel", "ModelPreTrainedModel", "Tensor", "True", "_tracker_add_new_objects", "add_mask_inputs", "bool", "class", "def", "for", "frame_idx", "in", "inference_session", "int", "list", "mask", "new_obj_ids", "new_obj_masks", "obj_id", "obj_id_to_idx", "obj_idx", "obj_with_new_inputs", "reverse", "run_mem_encoder", "self", "tracker_model", "unsqueeze", "zip"], "sam3_video/modeling_sam3_video.py:Sam3VideoModel.run_tracker_update_execution_phase": ["False", "ModelInferenceSession", "ModelModel", "ModelPreTrainedModel", "Tensor", "_tracker_add_new_objects", "bool", "class", "def", "det_out", "dict", "dtype", "for", "frame_idx", "if", "in", "inference_session", "int", "len", "list", "long", "mask", "new_det_masks", "new_det_obj_ids", "new_det_out_inds", "new_det_out_inds_t", "new_obj_ids", "new_obj_masks", "obj_id", "obj_ids_newly_removed", "remove_object", "reverse", "run_tracker_update_execution_phase", "self", "set", "str", "strict", "tensor", "torch", "tracker_update_plan"], "sam3_video/modeling_sam3_video.py:Sam3VideoModel.build_outputs": ["ModelInferenceSession", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "True", "and", "build_outputs", "class", "def", "det_idx", "det_mask", "det_out", "device", "dict", "dtype", "existing_masklet_obj_ids", "fill_hole_area", "fill_holes", "fill_holes_in_mask_scores", "for", "get", "if", "in", "inference_session", "int", "is", "len", "list", "long", "mask", "max_area", "new_det_low_res_masks", "new_det_obj_ids", "new_det_out_inds", "new_det_out_inds_t", "not", "obj_id", "obj_id_to_mask", "obj_ids", "reconditioned_obj_ids", "remove_sprinkles", "return", "self", "set", "squeeze", "str", "tensor", "torch", "tracker_low_res_masks_global", "tracker_update_plan", "trk_id_to_max_iou_high_conf_det", "unsqueeze", "zip"], "sam3_video/modeling_sam3_video.py:Sam3VideoModel._merge_detections_from_prompts": ["ModelInferenceSession", "ModelModel", "ModelPreTrainedModel", "Tensor", "_merge_detections_from_prompts", "all_detections", "append", "bbox", "cat", "class", "def", "det_idx", "det_idx_to_prompt_id", "det_out", "device", "dict", "else", "for", "i", "if", "in", "inference_device", "inference_session", "int", "items", "len", "low_res_mask_size", "mask", "merged_bboxes", "merged_det_out", "merged_masks", "merged_scores", "num_dets", "prompt_id", "range", "return", "scores", "self", "str", "torch", "tuple", "zeros"], "sam3_video/modeling_sam3_video.py:Sam3VideoModel._det_track_one_frame": ["False", "ModelInferenceSession", "ModelModel", "ModelPreTrainedModel", "_det_track_one_frame", "_merge_detections_from_prompts", "all_detections", "bool", "build_outputs", "cache", "cache_vision_features", "class", "def", "det_idx_to_prompt_id", "det_out", "detector_model", "dict", "frame_idx", "get_frame", "get_vision_features", "get_vision_features_for_tracker", "if", "inference_session", "int", "obj_id_to_mask", "obj_id_to_score", "obj_id_to_tracker_score_frame_wise", "obj_ids", "pixel_values", "reconditioned_obj_ids", "return", "reverse", "run_detection", "run_tracker_propagation", "run_tracker_update_execution_phase", "run_tracker_update_planning_phase", "self", "shape", "sigmoid", "streaming", "tolist", "tracker_low_res_masks_global", "tracker_metadata_new", "tracker_obj_ids", "tracker_obj_scores_global", "tracker_update_plan", "unsqueeze", "update", "vision_embeds", "vision_feats", "vision_pos_embeds", "zip"], "sam3_video/modeling_sam3_video.py:Sam3VideoModel.forward": ["False", "ModelInferenceSession", "ModelModel", "ModelPreTrainedModel", "ModelSegmentationOutput", "None", "Tensor", "ValueError", "_", "_det_track_one_frame", "add_new_frame", "be", "bool", "class", "def", "extra_metadata", "for", "forward", "frame", "frame_idx", "if", "inference_session", "int", "is", "kwargs", "list", "max_obj_id", "must", "not", "obj_first_frame_idx", "obj_id_to_last_occluded", "obj_id_to_mask", "obj_id_to_score", "obj_id_to_tracker_score", "obj_id_to_tracker_score_frame_wise", "obj_ids", "object_ids", "overlap_pair_to_frame_inds", "provided", "r", "raise", "removed_obj_ids", "return", "reverse", "self", "streaming", "suppressed_obj_ids", "torch", "tracker_metadata_new", "trk_keep_alive", "unmatched_frame_inds", "when"], "sam3_video/modeling_sam3_video.py:Sam3VideoModel._get_processing_order": ["False", "ModelInferenceSession", "ModelModel", "ModelPreTrainedModel", "None", "_get_processing_order", "bool", "class", "def", "else", "end_frame_idx", "if", "inference_session", "int", "is", "max", "max_frame_num_to_track", "min", "num_frames", "processing_order", "range", "return", "reverse", "self", "start_frame_idx"], "sam3_video/modeling_sam3_video.py:Sam3VideoModel.propagate_in_video_iterator": ["False", "ModelInferenceSession", "ModelModel", "ModelPreTrainedModel", "None", "_get_processing_order", "append", "class", "def", "elif", "else", "end_frame_idx", "for", "frame_idx", "from", "hotstart_buffer", "hotstart_delay", "hotstart_removed_obj_ids", "if", "in", "inference_session", "len", "max_frame_num_to_track", "out", "processing_order", "propagate_in_video_iterator", "removed_obj_ids", "reverse", "self", "start_frame_idx", "tqdm", "update", "yield", "yield_list"], "sam3_video/modeling_sam3_video.py:fast_diag_box_iou": ["Model_diag_box_iou", "area1", "area2", "box1_XY", "box1_xy", "box2_XY", "box2_xy", "boxes1", "boxes2", "clamp", "def", "inter", "iou", "lt", "max", "min", "prod", "rb", "return", "torch", "union"], "sam3_video/modeling_sam3_video.py:mask_iou": ["H", "M", "Model_iou", "N", "Tensor", "W", "_", "clamp", "def", "dim", "float", "gt_Models", "gt_flat", "intersection", "ious", "min", "pred_Models", "pred_flat", "return", "shape", "sum", "torch", "union", "view"], "sam3_video/modeling_sam3_video.py:nms_masks": ["Exception", "Failed", "Model", "Model_masks", "Tensor", "True", "_load_cv_utils_kernel_once", "as", "be", "cumsum", "cv_utils_kernel", "def", "dim", "e", "except", "f", "float", "generic_Model", "if", "iou_threshold", "ious", "is_valid", "isin", "keep", "kept_inds", "kernels", "library", "logger", "mask_iou", "masks_binary", "not", "numel", "post", "pred_masks", "pred_probs", "prob_threshold", "probs", "processing", "return", "run", "skipped", "to", "torch", "try", "use_iou_matrix", "using", "valid_inds", "warning_once", "where", "will"], "sam3_video/modeling_sam3_video.py:fill_holes_in_mask_scores": ["Model_holes", "Model_holes_in_mask_scores", "True", "_", "_get_connected_components_with_padding", "areas_bg", "areas_fg", "bg_area_thresh", "clamp_", "def", "dim", "dtype", "fg_area_thresh", "floor_divide_", "if", "int32", "keepdim", "mask", "mask_bg", "mask_fg", "max", "max_area", "remove_sprinkles", "return", "small_components_bg", "small_components_fg", "sum", "torch", "where"], "sam3_video/modeling_sam3_video.py:_get_connected_components_with_padding": ["Exception", "F", "Failed", "H", "Hole", "True", "W", "_", "_get_connected_components_with_padding", "_load_cv_utils_kernel_once", "and", "as", "be", "cc_2d", "components", "compute", "connected", "constant", "contiguous", "counts", "cv_utils_kernel", "def", "dtype", "e", "else", "except", "f", "fill_value", "filling", "full_like", "get_counts", "if", "int32", "kernels", "labels", "library", "logger", "mask", "mask_pad", "mode", "not", "pad", "pad_h", "pad_w", "removal", "return", "shape", "skipped", "sprinkle", "to", "torch", "try", "uint8", "using", "value", "warning_once", "will", "zeros_like"], "sam_hq/modeling_sam_hq.py:SamHQVisionAttention.__init__": ["Input", "Linear", "ModelVisionAttention", "Module", "None", "Parameter", "ValueError", "__init__", "attention_dropout", "be", "bias", "class", "config", "def", "dropout", "else", "encoding", "head_dim", "hidden_size", "if", "image_size", "input_size", "is", "must", "nn", "num_attention_heads", "patch_size", "positional", "proj", "provided", "qkv", "qkv_bias", "raise", "rel_pos_h", "rel_pos_w", "relative", "scale", "self", "size", "super", "torch", "use_rel_pos", "using", "window_size", "zeros"], "sam_hq/modeling_sam_hq.py:SamHQVisionAttention.get_rel_pos": ["F", "ModelVisionAttention", "Module", "None", "Tensor", "arange", "class", "def", "get_rel_pos", "int", "interpolate", "k_coords", "k_size", "linear", "long", "max", "max_rel_dist", "mode", "nn", "permute", "q_coords", "q_size", "rel_pos", "rel_pos_resized", "relative_coords", "reshape", "return", "self", "shape", "size", "torch"], "sam_hq/modeling_sam_hq.py:SamHQVisionAttention.get_decomposed_rel_pos": ["ModelVisionAttention", "Module", "None", "Tensor", "_", "batch_size", "bhwc", "bhwk", "class", "decomposed_rel_pos", "def", "dim", "einsum", "get_decomposed_rel_pos", "get_rel_pos", "hkc", "int", "k_size", "key_height", "key_width", "nn", "q_size", "query", "query_height", "query_width", "rel_h", "rel_pos_h", "rel_pos_w", "rel_w", "relative_position_height", "relative_position_width", "reshape", "reshaped_query", "return", "self", "shape", "torch", "tuple", "wkc"], "sam_hq/modeling_sam_hq.py:SamHQVisionAttention.forward": ["ModelVisionAttention", "Module", "None", "Tensor", "_", "attn_output", "attn_probs", "attn_weights", "batch_size", "class", "decomposed_rel_pos", "def", "dim", "dropout", "dtype", "float32", "forward", "functional", "get_decomposed_rel_pos", "height", "hidden_states", "if", "key", "nn", "num_attention_heads", "output_attentions", "p", "permute", "proj", "qkv", "query", "rel_pos_h", "rel_pos_w", "reshape", "reshape_as", "return", "scale", "self", "shape", "softmax", "to", "torch", "training", "transpose", "tuple", "unbind", "use_rel_pos", "value", "width"], "sam_hq/modeling_sam_hq.py:SamHQMLPBlock.__init__": ["ACT2FN", "Linear", "ModelMLPBlock", "Module", "__init__", "act", "class", "config", "def", "hidden_act", "hidden_size", "lin1", "lin2", "mlp_dim", "nn", "self", "super"], "sam_hq/modeling_sam_hq.py:SamHQMLPBlock.forward": ["ModelMLPBlock", "Module", "Tensor", "act", "class", "def", "forward", "hidden_states", "lin1", "lin2", "nn", "return", "self", "torch"], "sam_hq/modeling_sam_hq.py:SamHQVisionSdpaAttention.__init__": ["ModelVisionAttention", "ModelVisionSdpaAttention", "__init__", "class", "config", "def", "self", "super", "window_size"], "sam_hq/modeling_sam_hq.py:SamHQVisionSdpaAttention.forward": ["False", "If", "ModelVisionAttention", "ModelVisionSdpaAttention", "None", "Tensor", "The", "True", "_", "__class__", "__name__", "attention", "attn_bias", "attn_implementation", "attn_mask", "attn_output", "batch_size", "be", "class", "decomposed_rel_pos", "def", "does", "eager", "f", "forward", "functional", "get", "get_decomposed_rel_pos", "height", "hidden_states", "if", "key", "loading", "logger", "model", "nn", "not", "num_attention_heads", "output_attentions", "permute", "please", "proj", "qkv", "query", "rel_pos_h", "rel_pos_w", "reshape", "return", "returned", "scaled_dot_product_attention", "self", "set", "shape", "support", "the", "to", "torch", "unbind", "use_rel_pos", "value", "view", "want", "warning_once", "weights", "when", "width", "will", "you"], "sam_hq/modeling_sam_hq.py:SamHQVisionLayer.__init__": ["GradientCheckpointingLayer", "LayerNorm", "ModelMLPBlock", "ModelVisionLayer", "Model_VISION_ATTENTION_CLASSES", "__init__", "_attn_implementation", "attn", "class", "config", "def", "eps", "hidden_size", "layer_norm1", "layer_norm2", "layer_norm_eps", "mlp", "nn", "self", "super", "window_size"], "sam_hq/modeling_sam_hq.py:SamHQVisionLayer.window_partition": ["F", "GradientCheckpointingLayer", "ModelVisionLayer", "Tensor", "batch_size", "channel", "class", "contiguous", "def", "height", "hidden_states", "int", "pad", "pad_h", "pad_height", "pad_w", "pad_width", "permute", "reshape", "return", "self", "shape", "torch", "tuple", "width", "window_partition", "window_size", "windows"], "sam_hq/modeling_sam_hq.py:SamHQVisionLayer.window_unpartition": ["GradientCheckpointingLayer", "ModelVisionLayer", "Tensor", "batch_size", "class", "contiguous", "def", "height", "hidden_states", "int", "original_shape", "pad_height", "pad_width", "padding_shape", "permute", "reshape", "return", "self", "shape", "torch", "tuple", "width", "window_size", "window_unpartition", "windows"], "sam_hq/modeling_sam_hq.py:SamHQVisionLayer.forward": ["FloatTensor", "GradientCheckpointingLayer", "ModelVisionLayer", "Tensor", "attn", "attn_weights", "class", "def", "forward", "height", "hidden_states", "if", "layer_norm1", "layer_norm2", "layernorm_output", "mlp", "padding_shape", "residual", "return", "self", "shape", "torch", "tuple", "width", "window_partition", "window_size", "window_unpartition"], "sam_hq/modeling_sam_hq.py:SamHQPositionalEmbedding.__init__": ["ModelPositionalEmbedding", "Module", "__init__", "class", "config", "def", "nn", "num_pos_feats", "positional_embedding", "randn", "register_buffer", "scale", "self", "super", "torch"], "sam_hq/modeling_sam_hq.py:SamHQPositionalEmbedding.forward": ["ModelPositionalEmbedding", "Module", "None", "cat", "class", "clone", "coordinates", "cos", "def", "dim", "dtype", "forward", "if", "input_coords", "input_shape", "is", "nn", "not", "np", "pi", "positional_embedding", "return", "self", "sin", "to", "torch"], "sam_hq/modeling_sam_hq.py:SamHQPreTrainedModel._init_weights": ["ModelPositionalEmbedding", "ModelPreTrainedModel", "ModelVisionAttention", "ModelVisionEncoder", "Module", "PreTrainedModel", "_init_weights", "class", "config", "def", "elif", "if", "init", "isinstance", "module", "nn", "normal_", "pos_embed", "positional_embedding", "rel_pos_h", "rel_pos_w", "scale", "self", "std", "super", "use_abs_pos", "use_rel_pos", "zeros_"], "sam_hq/modeling_sam_hq.py:SamHQPatchEmbeddings.__init__": ["Conv2d", "Iterable", "ModelPatchEmbeddings", "Module", "__init__", "abc", "class", "collections", "config", "def", "else", "hidden_size", "if", "image_size", "isinstance", "kernel_size", "nn", "num_channels", "num_patches", "patch_size", "projection", "self", "stride", "super"], "sam_hq/modeling_sam_hq.py:SamHQPatchEmbeddings.forward": ["Input", "Make", "ModelPatchEmbeddings", "Module", "ValueError", "batch_size", "channel", "class", "configuration", "def", "dimension", "doesn", "embeddings", "f", "forward", "height", "if", "image", "image_size", "in", "match", "model", "nn", "num_channels", "of", "one", "or", "permute", "pixel", "pixel_values", "projection", "raise", "return", "self", "set", "shape", "size", "sure", "t", "that", "the", "values", "width", "with"], "sam_hq/modeling_sam_hq.py:SamHQVisionNeck.__init__": ["Conv2d", "False", "ModelLayerNorm", "ModelVisionConfig", "ModelVisionNeck", "Module", "__init__", "bias", "channels_first", "class", "config", "conv1", "conv2", "data_format", "def", "hidden_size", "kernel_size", "layer_norm1", "layer_norm2", "nn", "output_channels", "padding", "self", "super"], "sam_hq/modeling_sam_hq.py:SamHQVisionNeck.forward": ["ModelVisionNeck", "Module", "class", "conv1", "conv2", "def", "forward", "hidden_states", "layer_norm1", "layer_norm2", "nn", "permute", "return", "self"], "sam_hq/modeling_sam_hq.py:SamHQVisionEncoder.__init__": ["False", "ModelPatchEmbeddings", "ModelPreTrainedModel", "ModelVisionConfig", "ModelVisionEncoder", "ModelVisionLayer", "ModelVisionNeck", "ModuleList", "None", "Parameter", "__init__", "append", "class", "config", "def", "else", "for", "global_attn_indexes", "gradient_checkpointing", "hidden_size", "i", "if", "image_size", "in", "layer", "layers", "neck", "nn", "not", "num_hidden_layers", "patch_embed", "patch_size", "pos_embed", "post_init", "range", "self", "super", "torch", "use_abs_pos", "window_size", "zeros"], "sam_hq/modeling_sam_hq.py:SamHQVisionEncoder.get_input_embeddings": ["ModelPreTrainedModel", "ModelVisionEncoder", "class", "def", "get_input_embeddings", "patch_embed", "return", "self"], "sam_hq/modeling_sam_hq.py:SamHQVisionEncoder.forward": ["FloatTensor", "ModelPreTrainedModel", "ModelVisionEncoder", "ModelVisionEncoderOutput", "None", "TransformersKwargs", "Unpack", "ValueError", "You", "and", "append", "class", "def", "for", "forward", "hasattr", "have", "hidden_states", "if", "in", "intermediate_embeddings", "is", "kwargs", "last_hidden_state", "layer_module", "layers", "neck", "not", "patch_embed", "pixel_values", "pos_embed", "raise", "return", "self", "specify", "to", "torch", "tuple", "window_size"], "sam_hq/modeling_sam_hq.py:SamHQLayerNorm.__init__": ["LayerNorm", "ModelLayerNorm", "NotImplementedError", "Unsupported", "__init__", "channels_first", "channels_last", "class", "data", "data_format", "def", "eps", "f", "format", "if", "in", "kwargs", "nn", "normalized_shape", "not", "raise", "self", "super"], "sam_hq/modeling_sam_hq.py:SamHQLayerNorm.forward": ["LayerNorm", "ModelLayerNorm", "Tensor", "channels_first", "class", "data_format", "def", "else", "features", "forward", "if", "nn", "permute", "return", "self", "super", "torch"], "sam_hq/modeling_sam_hq.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "attention_mask", "attn_output", "attn_weights", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "softmax", "to", "torch", "training", "transpose", "value"], "sam_hq/modeling_sam_hq.py:SamHQAttention.__init__": ["False", "Linear", "ModelAttention", "Module", "None", "ValueError", "__init__", "attention_downModelple_rate", "class", "config", "def", "divide", "downModelple_rate", "else", "hidden_size", "if", "internal_dim", "is", "is_causal", "k_proj", "must", "nn", "num_attention_heads", "out_proj", "q_proj", "raise", "scaling", "self", "super", "v_proj"], "sam_hq/modeling_sam_hq.py:SamHQAttention._separate_heads": ["ModelAttention", "Module", "Tensor", "_separate_heads", "batch", "c_per_head", "channel", "class", "def", "hidden_states", "int", "n_tokens", "nn", "num_attention_heads", "point_batch_size", "reshape", "return", "self", "shape", "transpose"], "sam_hq/modeling_sam_hq.py:SamHQAttention._recombine_heads": ["ModelAttention", "Module", "Tensor", "_recombine_heads", "batch", "c_per_head", "class", "def", "hidden_states", "int", "n_heads", "n_tokens", "nn", "point_batch_size", "reshape", "return", "self", "shape"], "sam_hq/modeling_sam_hq.py:SamHQAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "ModelAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_attn_implementation", "_recombine_heads", "_separate_heads", "attention_interface", "attention_mask", "attention_similarity", "attn_output", "attn_weights", "class", "config", "def", "dropout", "eager", "eager_attention_forward", "forward", "if", "is_causal", "k_proj", "key", "kwargs", "nn", "num_attention_heads", "out_proj", "point_batch_size", "q_proj", "query", "return", "scaling", "self", "shape", "v_proj", "value"], "sam_hq/modeling_sam_hq.py:SamHQTwoWayAttentionBlock.__init__": ["False", "LayerNorm", "ModelAttention", "ModelMLPBlock", "ModelTwoWayAttentionBlock", "Module", "__init__", "attention_downModelple_rate", "bool", "class", "config", "cross_attn_image_to_token", "cross_attn_token_to_image", "def", "downModelple_rate", "eps", "hidden_size", "int", "layer_norm1", "layer_norm2", "layer_norm3", "layer_norm4", "layer_norm_eps", "mlp", "nn", "self", "self_attn", "skip_first_layer_pe", "super"], "sam_hq/modeling_sam_hq.py:SamHQTwoWayAttentionBlock.forward": ["ModelTwoWayAttentionBlock", "Module", "Tensor", "TransformersKwargs", "Unpack", "_", "attention_similarity", "attn_out", "class", "cross_attn_image_to_token", "cross_attn_token_to_image", "def", "else", "forward", "if", "key", "key_point_embedding", "keys", "kwargs", "layer_norm1", "layer_norm2", "layer_norm3", "layer_norm4", "mlp", "mlp_out", "nn", "queries", "query", "query_point_embedding", "return", "self", "self_attn", "skip_first_layer_pe", "value"], "sam_hq/modeling_sam_hq.py:SamHQTwoWayTransformer.__init__": ["LayerNorm", "ModelAttention", "ModelMaskDecoderConfig", "ModelTwoWayAttentionBlock", "ModelTwoWayTransformer", "Module", "ModuleList", "__init__", "append", "class", "config", "def", "final_attn_token_to_image", "for", "hidden_size", "i", "in", "layer_norm_final_attn", "layers", "nn", "num_hidden_layers", "range", "self", "skip_first_layer_pe", "super"], "sam_hq/modeling_sam_hq.py:SamHQTwoWayTransformer.forward": ["BaseModelOutput", "ModelTwoWayTransformer", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "_", "an", "attention_similarity", "attn_out", "class", "def", "final_attn_token_to_image", "flatten", "for", "forward", "have", "if", "image_embedding", "image_embeddings", "image_positional_embeddings", "in", "is", "key", "key_point_embedding", "keys", "kwargs", "layer", "layer_norm_final_attn", "layers", "nn", "not", "permute", "point_embeddings", "queries", "query", "query_point_embedding", "raise", "return", "self", "specify", "target_embedding", "to", "tuple", "unsqueeze", "value"], "sam_hq/modeling_sam_hq.py:SamHQFeedForward.__init__": ["False", "Linear", "ModelFeedForward", "Module", "ModuleList", "ReLU", "_", "__init__", "activation", "bool", "class", "def", "for", "hidden_dim", "in", "input_dim", "int", "layers", "nn", "num_layers", "output_dim", "proj_in", "proj_out", "range", "self", "sigmoid_output", "super"], "sam_hq/modeling_sam_hq.py:SamHQFeedForward.forward": ["F", "ModelFeedForward", "Module", "activation", "class", "def", "for", "forward", "hidden_states", "if", "in", "layer", "layers", "nn", "proj_in", "proj_out", "return", "self", "sigmoid", "sigmoid_output"], "sam_hq/modeling_sam_hq.py:SamHQMaskDecoder.__init__": ["Conv2d", "ConvTranspose2d", "Embedding", "GELU", "ModelFeedForward", "ModelLayerNorm", "ModelMaskDecoder", "ModelMaskDecoderConfig", "ModelTwoWayTransformer", "Module", "ModuleList", "_", "__init__", "activation", "channels_first", "class", "compress_vit_conv1", "compress_vit_conv2", "compress_vit_norm", "config", "data_format", "def", "encoder_conv1", "encoder_conv2", "encoder_norm", "for", "hidden_size", "hq_mask_mlp", "hq_token", "in", "iou_head_depth", "iou_head_hidden_dim", "iou_prediction_head", "iou_token", "kernel_size", "mask_conv1", "mask_conv2", "mask_norm", "mask_tokens", "mlps_list", "nn", "num_mask_tokens", "num_multimask_outputs", "output_hypernetworks_mlps", "padding", "range", "self", "stride", "super", "transformer", "upscale_conv1", "upscale_conv2", "upscale_layer_norm", "vit_dim"], "sam_hq/modeling_sam_hq.py:SamHQMaskDecoder.forward": ["ModelMMaskDecoderOutputs", "ModelMaskDecoder", "Module", "None", "Tensor", "True", "_", "activation", "and", "attention_similarity", "batch_size", "bool", "cat", "class", "compress_vit_conv1", "compress_vit_conv2", "compress_vit_norm", "compressed_vit_features", "contiguous", "current_mlp", "def", "dense_prompt_embeddings", "descending", "dim", "dtype", "elif", "else", "embed_encode", "encoder_conv1", "encoder_conv2", "encoder_norm", "expand", "for", "forward", "gather", "has_intermediate", "height", "hq_features", "hq_mask_mlp", "hq_token", "hq_token_only", "hyper_in", "hyper_in_list", "if", "image_embeddings", "image_positional_embeddings", "in", "intermediate_embeddings", "iou_pred", "iou_pred_sorted", "iou_prediction_head", "iou_token", "iou_token_out", "is", "len", "list", "mask_conv1", "mask_conv2", "mask_norm", "mask_slice", "mask_token_index", "mask_tokens", "mask_tokens_out", "masks", "masks_Model", "masks_hq", "multimask_output", "nn", "not", "num_channels", "num_mask_tokens", "output_hypernetworks_mlps", "output_tokens", "permute", "point_batch_size", "point_embedding", "point_embeddings", "range", "repeat", "repeat_interleave", "reshape", "return", "self", "shape", "slice", "sort", "sort_indices", "sparse_prompt_embeddings", "stack", "target_embedding", "to", "tokens", "torch", "transformer", "transpose", "upscale_conv1", "upscale_conv2", "upscale_layer_norm", "upscaled_embedding", "upscaled_embedding_hq", "vit_features", "weight", "width"], "sam_hq/modeling_sam_hq.py:SamHQVisionModel.__init__": ["ModelPreTrainedModel", "ModelVisionConfig", "ModelVisionEncoder", "ModelVisionModel", "__init__", "class", "config", "def", "post_init", "self", "super", "vision_encoder"], "sam_hq/modeling_sam_hq.py:SamHQVisionModel.get_input_embeddings": ["ModelPreTrainedModel", "ModelVisionModel", "Module", "class", "def", "get_input_embeddings", "nn", "patch_embed", "return", "self", "vision_encoder"], "sam_hq/modeling_sam_hq.py:SamHQVisionModel.forward": ["FloatTensor", "ModelPreTrainedModel", "ModelVisionEncoderOutput", "ModelVisionModel", "None", "TransformersKwargs", "Unpack", "class", "def", "forward", "kwargs", "pixel_values", "return", "self", "torch", "tuple", "vision_encoder"], "sam_hq/modeling_sam_hq.py:SamHQMaskEmbedding.__init__": ["ACT2FN", "Conv2d", "ModelLayerNorm", "ModelMaskEmbedding", "ModelPromptEncoderConfig", "Module", "__init__", "activation", "channels_first", "class", "config", "conv1", "conv2", "conv3", "data_format", "def", "eps", "hidden_act", "hidden_size", "kernel_size", "layer_norm1", "layer_norm2", "layer_norm_eps", "mask_input_channels", "nn", "self", "stride", "super"], "sam_hq/modeling_sam_hq.py:SamHQMaskEmbedding.forward": ["ModelMaskEmbedding", "Module", "activation", "class", "conv1", "conv2", "conv3", "def", "dense_embeddings", "forward", "hidden_states", "layer_norm1", "layer_norm2", "masks", "nn", "return", "self"], "sam_hq/modeling_sam_hq.py:SamHQPromptEncoder.__init__": ["Embedding", "ModelConfig", "ModelMaskEmbedding", "ModelPositionalEmbedding", "ModelPromptEncoder", "Module", "ModuleList", "__init__", "class", "config", "def", "for", "hidden_size", "i", "image_embedding_size", "image_size", "in", "input_image_size", "mask_embed", "nn", "no_mask_embed", "not_a_point_embed", "num_point_embeddings", "point_embed", "prompt_encoder_config", "range", "self", "shared_embedding", "super", "vision_config"], "sam_hq/modeling_sam_hq.py:SamHQPromptEncoder._embed_points": ["ModelPromptEncoder", "Module", "None", "Tensor", "_embed_points", "bool", "cat", "class", "def", "device", "dim", "if", "input_image_size", "input_shape", "labels", "nn", "not_a_point_embed", "ones", "pad", "padding_label", "padding_point", "point_embed", "point_embedding", "points", "return", "self", "shape", "shared_embedding", "target_labels_shape", "target_point_shape", "torch", "weight", "where", "zeros", "zeros_like"], "sam_hq/modeling_sam_hq.py:SamHQPromptEncoder._embed_boxes": ["ModelPromptEncoder", "Module", "Tensor", "_embed_boxes", "batch_size", "boxes", "class", "coords", "corner_embedding", "def", "input_image_size", "input_shape", "nb_boxes", "nn", "point_embed", "reshape", "return", "self", "shape", "shared_embedding", "torch", "weight"], "sam_hq/modeling_sam_hq.py:SamHQPromptEncoder.forward": ["If", "ModelPromptEncoder", "Module", "None", "Tensor", "ValueError", "_embed_boxes", "_embed_points", "also", "are", "batch_size", "be", "box_embeddings", "cat", "class", "def", "dense_embeddings", "dim", "else", "expand", "forward", "if", "image_embedding_size", "input_boxes", "input_labels", "input_masks", "input_points", "is", "labels", "mask_embed", "must", "nn", "no_mask_embed", "not", "pad", "point_embeddings", "points", "provided", "raise", "reshape", "return", "self", "shape", "sparse_embeddings", "torch", "tuple", "weight"], "sam_hq/modeling_sam_hq.py:SamHQModel.__init__": ["ModelMaskDecoder", "ModelModel", "ModelPositionalEmbedding", "ModelPreTrainedModel", "ModelPromptEncoder", "ModelVisionEncoder", "__init__", "_attn_implementation", "class", "config", "def", "mask_decoder", "mask_decoder_config", "post_init", "prompt_encoder", "self", "shared_image_embedding", "super", "vision_config", "vision_encoder"], "sam_hq/modeling_sam_hq.py:SamHQModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "return", "self", "vision_encoder"], "sam_hq/modeling_sam_hq.py:SamHQModel.get_image_wide_positional_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "config", "cumsum", "def", "device", "dim", "dtype", "get_image_wide_positional_embeddings", "grid", "image_embedding_size", "ones", "permute", "positional_embedding", "prompt_encoder_config", "return", "self", "shared_image_embedding", "size", "stack", "target_device", "target_dtype", "torch", "unsqueeze", "x_embed", "y_embed"], "sam_hq/modeling_sam_hq.py:SamHQModel.get_image_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "get_image_embeddings", "image_embeddings", "intermediate_embeddings", "pixel_values", "r", "return", "self", "vision_encoder", "vision_output"], "sam_hq/modeling_sam_hq.py:SamHQModel.get_prompt_embeddings": ["FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "class", "def", "get_prompt_embeddings", "input_boxes", "input_labels", "input_masks", "input_points", "prompt_encoder", "prompt_output", "r", "return", "self", "torch"], "sam_hq/modeling_sam_hq.py:SamHQModel.forward": ["Either", "False", "FloatTensor", "Got", "LongTensor", "ModelImageSegmentationOutput", "ModelModel", "ModelPreTrainedModel", "None", "Of", "Only", "Tensor", "The", "TransformersKwargs", "True", "Unpack", "ValueError", "You", "a", "and", "as", "attention_similarity", "attentions", "batch_size", "be", "bool", "bounding", "box", "box_batch_size", "boxes", "can", "class", "def", "dense_embeddings", "dense_prompt_embeddings", "device", "dict", "dtype", "else", "f", "forward", "get_image_wide_positional_embeddings", "got", "hidden_states", "hq_token_only", "if", "image_embeddings", "image_positional_embeddings", "input", "input_boxes", "input_labels", "input_masks", "input_points", "int", "intermediate_embeddings", "iou_scores", "is", "kwargs", "last_hidden_state", "len", "list", "many", "mask_decoder", "mask_decoder_output", "multimask_output", "must", "nb_boxes", "nb_points_per_image", "not", "of", "one", "ones_like", "or", "per", "pixel_values", "point_batch_size", "points", "pred_masks", "prompt_encoder", "provide", "provided", "r", "raise", "repeat", "return", "self", "shape", "should", "sparse_embeddings", "sparse_prompt_embeddings", "str", "target_embedding", "tensor", "torch", "vision_attentions", "vision_encoder", "vision_hidden_states", "vision_outputs"], "seamless_m4t/modeling_seamless_m4t.py:shift_tokens_right": ["Model_tokens_right", "Modeled_input_ids", "None", "Tensor", "ValueError", "be", "clone", "config", "decoder_start_token_id", "def", "defined", "has", "if", "input_ids", "int", "is", "masked_fill_", "model", "new_zeros", "pad_token_id", "raise", "return", "self", "shape", "to", "torch"], "seamless_m4t/modeling_seamless_m4t.py:_compute_new_attention_mask": ["Tensor", "_compute_new_attention_mask", "arange", "batch_size", "bool_mask", "def", "device", "expand", "hidden_states", "indices", "mask", "mask_seq_len", "masked_fill", "new_ones", "return", "seq_lens", "shape", "torch", "unsqueeze"], "seamless_m4t/modeling_seamless_m4t.py:format_speech_generation_kwargs": ["Model_speech_generation_kwargs", "def", "elif", "else", "for", "generation_config", "if", "in", "items", "key", "kwargs", "kwargs_speech", "kwargs_text", "len", "not", "return", "speech_", "startswith", "text_", "value"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TConformerPositionalConvEmbedding.__init__": ["ACT2FN", "Conv1d", "GatheredParameters", "ModelConformerPositionalConvEmbedding", "ModelConformerSamePadLayer", "Module", "__init__", "activation", "class", "config", "conv", "deepspeed", "def", "dim", "else", "groups", "hasattr", "hidden_size", "if", "is_deepspeed_zero3_enabled", "kernel_size", "modifier_rank", "name", "nn", "num_conv_pos_embedding_groups", "num_conv_pos_embeddings", "original0", "original1", "padding", "parametrizations", "register_external_parameter", "self", "speech_encoder_hidden_act", "super", "utils", "weight", "weight_g", "weight_norm", "weight_v", "with", "zero"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TConformerPositionalConvEmbedding.forward": ["ModelConformerPositionalConvEmbedding", "Module", "activation", "class", "conv", "def", "forward", "hidden_states", "nn", "padding", "return", "self", "transpose"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TConformerRotaryPositionalEmbedding.__init__": ["ModelConformerRotaryPositionalEmbedding", "Module", "None", "__init__", "arange", "base", "cached_rotary_positional_embedding", "cached_sequence_length", "class", "config", "def", "dim", "dtype", "float", "hidden_size", "int64", "inv_freq", "nn", "register_buffer", "rotary_embedding_base", "self", "speech_encoder_attention_heads", "super", "torch"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TConformerRotaryPositionalEmbedding.forward": ["ModelConformerRotaryPositionalEmbedding", "Module", "None", "and", "arange", "cached_rotary_positional_embedding", "cached_sequence_length", "cat", "class", "cos", "cos_embeddings", "def", "dim", "einsum", "embeddings", "forward", "freqs", "hidden_states", "i", "if", "ij", "inv_freq", "is", "j", "nn", "not", "return", "self", "sequence_length", "shape", "sin", "sin_embeddings", "stack", "time_stamps", "torch", "type_as"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TConformerRelPositionalEmbedding.__init__": ["False", "ModelConformerRelPositionalEmbedding", "Module", "__init__", "class", "config", "d_model", "def", "expand", "extend_pe", "hidden_size", "max_len", "max_source_positions", "nn", "pe", "persistent", "register_buffer", "self", "super", "tensor", "torch"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TConformerRelPositionalEmbedding.extend_pe": ["ModelConformerRelPositionalEmbedding", "Module", "None", "arange", "cat", "class", "cos", "d_model", "def", "device", "dim", "div_term", "dtype", "exp", "extend_pe", "flip", "float", "if", "int64", "is", "log", "math", "nn", "not", "or", "pe", "pe_negative", "pe_positive", "position", "return", "self", "sin", "size", "to", "torch", "unsqueeze", "x", "zeros"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TConformerRelPositionalEmbedding.forward": ["ModelConformerRelPositionalEmbedding", "Module", "Tensor", "class", "def", "end_idx", "extend_pe", "forward", "hidden_states", "nn", "pe", "relative_position_embeddings", "return", "self", "size", "start_idx", "torch"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TConformerSamePadLayer.__init__": ["ModelConformerSamePadLayer", "Module", "__init__", "class", "def", "else", "if", "nn", "num_conv_pos_embeddings", "num_pad_remove", "self", "super"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TConformerSamePadLayer.forward": ["ModelConformerSamePadLayer", "Module", "class", "def", "forward", "hidden_states", "if", "nn", "num_pad_remove", "return", "self"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TConformerFeatureProjection.__init__": ["Dropout", "LayerNorm", "Linear", "ModelConformerFeatureProjection", "Module", "__init__", "class", "config", "def", "dropout", "eps", "feature_projection_input_dim", "hidden_size", "layer_norm", "layer_norm_eps", "nn", "projection", "self", "speech_encoder_dropout", "super"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TConformerFeatureProjection.forward": ["ModelConformerFeatureProjection", "Module", "class", "def", "dropout", "forward", "hidden_states", "layer_norm", "nn", "norm_hidden_states", "projection", "return", "self"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TConformerFeedForward.__init__": ["ACT2FN", "Dropout", "Linear", "ModelConformerFeedForward", "Module", "None", "__init__", "act_fn", "class", "config", "def", "dropout", "else", "hidden_size", "if", "intermediate_act_fn", "intermediate_dense", "intermediate_dropout", "is", "isinstance", "nn", "not", "output_dense", "output_dropout", "self", "speech_encoder_dropout", "speech_encoder_hidden_act", "speech_encoder_intermediate_size", "str", "super"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TConformerFeedForward.forward": ["ModelConformerFeedForward", "Module", "class", "def", "forward", "hidden_states", "intermediate_act_fn", "intermediate_dense", "intermediate_dropout", "nn", "output_dense", "output_dropout", "return", "self"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TConformerConvolutionModule.__init__": ["ACT2FN", "BatchNorm1d", "Conv1d", "Dropout", "False", "GLU", "LayerNorm", "ModelConformerConvolutionModule", "Module", "SAME", "ValueError", "__init__", "a", "activation", "batch_norm", "be", "bias", "class", "config", "conv_depthwise_kernel_size", "def", "depthwise_conv", "dim", "dropout", "for", "glu", "groups", "hidden_size", "if", "kernel_size", "layer_norm", "nn", "number", "odd", "padding", "pointwise_conv1", "pointwise_conv2", "raise", "same", "self", "should", "speech_encoder_dropout", "speech_encoder_hidden_act", "stride", "super"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TConformerConvolutionModule.forward": ["ModelConformerConvolutionModule", "Module", "None", "activation", "attention_mask", "batch_norm", "bool", "class", "def", "depthwise_conv", "dropout", "forward", "glu", "hidden_states", "if", "is", "layer_norm", "masked_fill", "nn", "not", "pointwise_conv1", "pointwise_conv2", "return", "self", "transpose", "unsqueeze"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TConformerSelfAttention.__init__": ["Dropout", "False", "Linear", "ModelConformerSelfAttention", "Module", "None", "Parameter", "True", "__init__", "bias", "class", "config", "def", "dropout", "else", "head_size", "hidden_size", "if", "linear_k", "linear_out", "linear_pos", "linear_q", "linear_v", "nn", "num_heads", "p", "pos_bias_u", "pos_bias_v", "position_embeddings_type", "relative", "self", "speech_encoder_attention_heads", "speech_encoder_dropout", "super", "torch", "use_position_embeddings", "zeros"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TConformerSelfAttention.forward": ["False", "ModelConformerSelfAttention", "Module", "None", "Tensor", "ValueError", "_apply_relative_embeddings", "_apply_rotary_embedding", "attention_mask", "batch_size", "be", "bool", "class", "def", "defined", "dim", "dropout", "else", "forward", "has", "head_size", "hidden_size", "hidden_states", "if", "is", "key", "linear_k", "linear_out", "linear_q", "linear_v", "math", "matmul", "nn", "not", "num_heads", "output_attentions", "position_embeddings_type", "probs", "query", "query_key_states", "raise", "relative", "relative_position_embeddings", "reshape", "return", "rotary", "scores", "self", "sequence_length", "size", "softmax", "sqrt", "to", "torch", "transpose", "tuple", "value", "value_states", "view", "when"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TConformerSelfAttention._apply_rotary_embedding": ["ModelConformerSelfAttention", "Module", "_apply_rotary_embedding", "batch_size", "cat", "class", "cos", "def", "dim", "head_size", "hidden_size", "hidden_states", "ndim", "nn", "num_heads", "relative_position_embeddings", "return", "rotated_states", "rotated_states_begin", "rotated_states_end", "self", "sequence_length", "sin", "size", "torch", "transpose", "view"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TConformerSelfAttention._apply_relative_embeddings": ["ModelConformerSelfAttention", "Module", "_apply_relative_embeddings", "cat", "class", "def", "device", "dim", "dtype", "head_size", "key", "linear_pos", "math", "matmul", "nn", "num_heads", "pos_bias_u", "pos_bias_v", "proj_relative_position_embeddings", "q_with_bias_u", "q_with_bias_v", "query", "relative_position_embeddings", "return", "scores", "scores_ac", "scores_bd", "scores_bd_padded", "scores_bd_padded_shape", "self", "shape", "size", "sqrt", "torch", "transpose", "view", "view_as", "zero_pad", "zeros"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TConformerEncoderLayer.__init__": ["Dropout", "GradientCheckpointingLayer", "LayerNorm", "ModelConformerConvolutionModule", "ModelConformerEncoderLayer", "ModelConformerFeedForward", "ModelConformerSelfAttention", "__init__", "class", "config", "conv_module", "def", "dropout", "embed_dim", "ffn1", "ffn1_layer_norm", "ffn2", "ffn2_layer_norm", "final_layer_norm", "hidden_size", "nn", "self", "self_attn", "self_attn_dropout", "self_attn_layer_norm", "speech_encoder_dropout", "super"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TConformerEncoderLayer.forward": ["False", "GradientCheckpointingLayer", "ModelConformerEncoderLayer", "None", "Tensor", "attention_mask", "attn_weigts", "bool", "class", "conv_attention_mask", "conv_module", "def", "ffn1", "ffn1_layer_norm", "ffn2", "ffn2_layer_norm", "final_layer_norm", "forward", "hidden_states", "output_attentions", "relative_position_embeddings", "residual", "return", "self", "self_attn", "self_attn_dropout", "self_attn_layer_norm", "torch"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TConformerEncoder.__init__": ["Dropout", "False", "LayerNorm", "ModelConformerEncoder", "ModelConformerEncoderLayer", "ModelConformerRelPositionalEmbedding", "ModelConformerRotaryPositionalEmbedding", "Module", "ModuleList", "None", "_", "__init__", "class", "config", "def", "dropout", "elif", "else", "embed_positions", "eps", "for", "gradient_checkpointing", "hidden_size", "if", "in", "layer_norm", "layer_norm_eps", "layers", "nn", "position_embeddings_type", "range", "relative", "rotary", "self", "speech_encoder_dropout", "speech_encoder_layers", "super"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TConformerEncoder.forward": ["BaseModelOutput", "False", "ModelConformerEncoder", "Module", "None", "True", "all_hidden_states", "all_self_attentions", "and", "attention_mask", "attentions", "bool", "class", "config", "conv_attention_mask", "def", "dropout", "dropout_probability", "dtype", "else", "embed_positions", "enumerate", "expand", "finfo", "for", "forward", "hidden_states", "i", "if", "in", "is", "is_deepspeed_zero3_enabled", "is_fsdp_managed_module", "last_hidden_state", "layer", "layer_norm", "layer_outputs", "layers", "masked_fill", "min", "nn", "not", "or", "output_attentions", "output_hidden_states", "rand", "relative_position_embeddings", "return", "return_dict", "self", "shape", "skip_the_layer", "speech_encoder_layerdrop", "synced_gpus", "to", "torch", "training", "tuple", "unsqueeze", "v"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TConformerAdapterLayer.__init__": ["Conv1d", "Dropout", "False", "GLU", "LayerNorm", "ModelConformerAdapterLayer", "ModelConformerFeedForward", "ModelConformerSelfAttention", "Module", "__init__", "act_fn", "activation", "adaptor_dropout", "adaptor_kernel_size", "adaptor_stride", "class", "config", "def", "dim", "dropout", "embed_dim", "ffn", "ffn_layer_norm", "hidden_size", "kernel_size", "nn", "padding", "relu", "residual_conv", "residual_layer_norm", "self", "self_attn", "self_attn_conv", "self_attn_dropout", "self_attn_layer_norm", "stride", "super", "use_position_embeddings"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TConformerAdapterLayer._compute_sub_sample_lengths_from_attention_mask": ["ModelConformerAdapterLayer", "Module", "_compute_sub_sample_lengths_from_attention_mask", "attention_mask", "class", "def", "floor", "int", "kernel_size", "nn", "pad", "return", "self", "seq_lens", "size", "stride", "sum"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TConformerAdapterLayer.forward": ["False", "ModelConformerAdapterLayer", "Module", "None", "Tensor", "_compute_new_attention_mask", "_compute_sub_sample_lengths_from_attention_mask", "_prepare_4d_attention_mask", "activation", "attention_mask", "attn_weights", "bool", "class", "def", "device", "dtype", "ffn", "ffn_layer_norm", "forward", "hidden_states", "if", "is", "nn", "not", "output_attentions", "residual", "residual_conv", "residual_layer_norm", "return", "self", "self_attn", "self_attn_conv", "self_attn_dropout", "self_attn_layer_norm", "seq_lens", "sub_sampled_lengths", "to", "torch", "transpose"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TConformerAdapter.__init__": ["ModelConformerAdapter", "ModelConformerAdapterLayer", "Module", "ModuleList", "_", "__init__", "class", "config", "def", "for", "in", "layers", "nn", "num_adapter_layers", "range", "self", "super"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TConformerAdapter.forward": ["ModelConformerAdapter", "Module", "attention_mask", "class", "def", "for", "forward", "hidden_states", "in", "layer", "layers", "nn", "return", "self"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TScaledWordEmbedding.__init__": ["Embedding", "ModelScaledWordEmbedding", "None", "__init__", "class", "def", "embed_scale", "embedding_dim", "float", "int", "nn", "num_embeddings", "padding_idx", "self", "super"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TScaledWordEmbedding.forward": ["Embedding", "ModelScaledWordEmbedding", "Tensor", "class", "def", "embed_scale", "forward", "input_ids", "nn", "return", "self", "super", "torch"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TSinusoidalPositionalEmbedding.__init__": ["ModelSinusoidalPositionalEmbedding", "Module", "None", "__init__", "class", "def", "embedding_dim", "int", "make_weights", "nn", "num_positions", "offset", "padding_idx", "self", "super"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TSinusoidalPositionalEmbedding.make_weights": ["False", "ModelSinusoidalPositionalEmbedding", "Module", "None", "class", "def", "device", "dtype", "emb_weights", "embedding_dim", "get_embedding", "hasattr", "if", "int", "make_weights", "nn", "num_embeddings", "padding_idx", "persistent", "register_buffer", "self", "to", "weights"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TSinusoidalPositionalEmbedding.get_embedding": ["ModelSinusoidalPositionalEmbedding", "Module", "None", "arange", "cat", "class", "cos", "def", "dim", "dtype", "emb", "embedding_dim", "exp", "float", "get_default_dtype", "get_embedding", "half_dim", "if", "int", "int64", "is", "log", "math", "nn", "not", "num_embeddings", "padding_idx", "return", "sin", "to", "torch", "unsqueeze", "view", "zeros"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TSinusoidalPositionalEmbedding.forward": ["ModelSinusoidalPositionalEmbedding", "Module", "None", "Tensor", "bsz", "class", "create_position_ids_from_input_ids", "create_position_ids_from_inputs_embeds", "def", "detach", "device", "else", "embedding_dim", "forward", "if", "index_select", "input_ids", "inputs_embeds", "int", "is", "make_weights", "max_pos", "nn", "not", "offset", "padding_idx", "past_key_values_length", "position_ids", "return", "self", "seq_len", "shape", "size", "to", "torch", "view", "weights"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TSinusoidalPositionalEmbedding.create_position_ids_from_inputs_embeds": ["ModelSinusoidalPositionalEmbedding", "Module", "arange", "class", "contiguous", "create_position_ids_from_inputs_embeds", "def", "device", "dtype", "expand", "input_shape", "inputs_embeds", "long", "nn", "padding_idx", "past_key_values_length", "position_ids", "return", "sequence_length", "size", "torch", "unsqueeze"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TSinusoidalPositionalEmbedding.create_position_ids_from_input_ids": ["ModelSinusoidalPositionalEmbedding", "Module", "class", "create_position_ids_from_input_ids", "cumsum", "def", "dim", "incremental_indices", "input_ids", "int", "long", "mask", "ne", "nn", "padding_idx", "past_key_values_length", "return", "torch", "type_as"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TAttention.__init__": ["False", "Instantiating", "Linear", "ModelAttention", "ModelConfig", "Module", "None", "Please", "True", "ValueError", "__class__", "__init__", "__name__", "a", "and", "be", "bias", "bool", "by", "caching", "call", "class", "config", "creating", "decoder", "def", "divisible", "dropout", "during", "embed_dim", "errors", "f", "float", "forward", "got", "head_dim", "if", "int", "is", "is_causal", "is_decoder", "k_proj", "layer_idx", "lead", "logger", "make", "must", "nn", "not", "num_heads", "out_proj", "passing", "provide", "q_proj", "raise", "recommended", "scaling", "self", "super", "sure", "the", "this", "to", "used", "v_proj", "warning_once", "when", "will", "without"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TAttention.forward": ["Attention", "Cache", "EncoderDecoderCache", "False", "ModelAttention", "Module", "None", "Tensor", "True", "ValueError", "_", "and", "attention_mask", "attn_output", "attn_probs", "attn_weights", "attn_weights_reshaped", "be", "bmm", "bool", "bsz", "but", "cache_position", "class", "cross_attention_cache", "curr_past_key_values", "current_states", "def", "dim", "dropout", "else", "embed_dim", "encoder_hidden_states", "f", "forward", "functional", "get", "head_dim", "hidden_states", "if", "is", "is_cross_attention", "is_updated", "isinstance", "k_proj", "key_states", "keys", "layer_idx", "layers", "mask", "nn", "not", "num_heads", "of", "out_proj", "output_attentions", "p", "past_key_values", "proj_shape", "q_proj", "query_states", "raise", "reshape", "return", "scaling", "self", "self_attention_cache", "should", "size", "softmax", "src_len", "tgt_len", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "values", "view", "weights"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TFeedForwardNetwork.__init__": ["ACT2FN", "Dropout", "Linear", "ModelConfig", "ModelFeedForwardNetwork", "Module", "__init__", "act", "activation_dropout", "activation_function", "class", "config", "def", "dropout", "fc1", "fc2", "ffn_dim", "hidden_size", "int", "nn", "self", "super"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TFeedForwardNetwork.forward": ["ModelFeedForwardNetwork", "Module", "Tensor", "act", "and", "class", "def", "dropout", "dtype", "fc1", "fc2", "forward", "hidden_states", "if", "int8", "isinstance", "nn", "return", "self", "to", "torch", "uint8", "weight"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TEncoderLayer.__init__": ["Dropout", "GradientCheckpointingLayer", "LayerNorm", "ModelAttention", "ModelConfig", "ModelEncoderLayer", "ModelFeedForwardNetwork", "None", "__init__", "activation_dropout", "attention_dropout", "attn_dropout", "class", "config", "def", "dropout", "else", "embed_dim", "encoder_attention_heads", "encoder_ffn_dim", "ffn", "ffn_dim", "ffn_dropout", "ffn_layer_norm", "hidden_size", "if", "is", "nn", "num_heads", "self", "self_attn", "self_attn_layer_norm", "super"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TEncoderLayer.forward": ["False", "GradientCheckpointingLayer", "ModelEncoderLayer", "Tensor", "attention_mask", "attn_dropout", "attn_weights", "bool", "class", "def", "ffn", "ffn_dropout", "ffn_layer_norm", "forward", "hidden_states", "if", "output_attentions", "outputs", "residual", "return", "self", "self_attn", "self_attn_layer_norm", "torch"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TDecoderLayer.__init__": ["ACT2FN", "Dropout", "GradientCheckpointingLayer", "LayerNorm", "ModelAttention", "ModelConfig", "ModelDecoderLayer", "ModelFeedForwardNetwork", "None", "True", "__init__", "activation_dropout", "activation_fn", "activation_function", "attention_dropout", "attn_dropout", "class", "config", "cross_attention", "cross_attention_layer_norm", "decoder_attention_heads", "decoder_ffn_dim", "def", "dropout", "else", "embed_dim", "ffn", "ffn_dim", "ffn_dropout", "ffn_layer_norm", "hidden_size", "if", "is", "is_decoder", "layer_idx", "nn", "num_heads", "self", "self_attn", "self_attn_layer_norm", "super"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TDecoderLayer.forward": ["Cache", "False", "GradientCheckpointingLayer", "ModelDecoderLayer", "None", "Tensor", "True", "attention_mask", "attn_dropout", "bool", "cache_position", "class", "cross_attention", "cross_attention_layer_norm", "cross_attn_weights", "def", "encoder_attention_mask", "encoder_hidden_states", "ffn", "ffn_dropout", "ffn_layer_norm", "forward", "hidden_states", "if", "is", "not", "output_attentions", "past_key_values", "residual", "return", "self", "self_attn", "self_attn_layer_norm", "self_attn_weights", "torch", "use_cache"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TPreTrainedModel._init_weights": ["BatchNorm1d", "Conv1d", "Embedding", "False", "LayerNorm", "Linear", "ModelConformerFeatureProjection", "ModelConformerPositionalConvEmbedding", "ModelConformerRelPositionalEmbedding", "ModelConformerRotaryPositionalEmbedding", "ModelConformerSelfAttention", "ModelPreTrainedModel", "ModelSinusoidalPositionalEmbedding", "Module", "None", "PreTrainedModel", "_init_weights", "_is_hf_initialized", "a", "and", "arange", "b", "base", "bias", "class", "config", "constant_", "conv", "copy_", "def", "dim", "dtype", "elif", "emb_weights", "embedding_dim", "expand", "extend_pe", "float", "get_embedding", "getattr", "groups", "hasattr", "hidden_size", "if", "in_channels", "in_features", "init", "initializer_range", "int64", "inv_freq", "is", "isinstance", "k", "kaiming_normal_", "kernel_size", "math", "max_len", "mean", "module", "nn", "normal_", "not", "num_batches_tracked", "num_positions", "offset", "ones_", "padding_idx", "pe", "pos_bias_u", "pos_bias_v", "projection", "rotary_embedding_base", "running_mean", "running_var", "self", "speech_encoder_attention_heads", "sqrt", "std", "tensor", "torch", "uniform_", "weight", "weights", "xavier_uniform_", "zeros_"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TPreTrainedModel._compute_sub_sample_lengths_from_attention_mask": ["ModelPreTrainedModel", "PreTrainedModel", "_compute_sub_sample_lengths_from_attention_mask", "adaptor_kernel_size", "adaptor_stride", "attention_mask", "class", "config", "def", "floor", "int", "kernel_size", "pad", "return", "self", "seq_lens", "size", "stride", "sum"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TPreTrainedModel.compute_last_hidden_states_per_sample": ["ModelPreTrainedModel", "None", "PreTrainedModel", "Tensor", "beam_indices", "beam_indices_mask", "class", "clone", "compute_last_hidden_states_per_sample", "concat", "def", "dim", "expand", "for", "gather", "hidden_states", "if", "in", "is", "last_hidden_states", "long", "max", "max_beam_length", "return", "self", "shape", "sum", "torch", "tuple", "unsqueeze"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TSpeechEncoder.__init__": ["LayerNorm", "ModelConfig", "ModelConformerAdapter", "ModelConformerEncoder", "ModelConformerFeatureProjection", "ModelConformerFeedForward", "ModelPreTrainedModel", "ModelSpeechEncoder", "None", "__init__", "act_fn", "adapter", "add_adapter", "class", "config", "def", "dropout", "else", "encoder", "feature_projection", "hidden_size", "if", "inner_layer_norm", "intermediate_ffn", "nn", "post_init", "relu", "self", "super"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TSpeechEncoder.forward": ["ModelPreTrainedModel", "ModelSpeechEncoder", "None", "Tensor", "ValueError", "Wav2Vec2BaseModelOutput", "adapter", "attention_mask", "attentions", "bool", "class", "config", "def", "else", "encoder", "encoder_outputs", "expanded_hidden_states", "feature_projection", "forward", "hidden_states", "if", "inner_layer_norm", "input_features", "intermediate_ffn", "is", "kwargs", "last_hidden_state", "not", "output_attentions", "output_hidden_states", "raise", "return", "return_dict", "self", "torch", "tuple", "use_return_dict"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TEncoder.__init__": ["Embedding", "False", "LayerNorm", "ModelConfig", "ModelEncoder", "ModelEncoderLayer", "ModelPreTrainedModel", "ModelScaledWordEmbedding", "ModelSinusoidalPositionalEmbedding", "ModuleList", "None", "_", "__init__", "append", "bool", "class", "config", "def", "dropout", "else", "embed_dim", "embed_positions", "embed_scale", "embed_tokens", "encoder_attention_heads", "encoder_ffn_dim", "encoder_layerdrop", "encoder_layers", "for", "gradient_checkpointing", "hidden_size", "if", "in", "is", "is_t2u_encoder", "layer_norm", "layerdrop", "layers", "math", "max_position_embeddings", "max_source_positions", "nn", "not", "pad_token_id", "padding_idx", "post_init", "r", "range", "scale_embedding", "self", "sqrt", "super", "vocab_size", "weight"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TEncoder.forward": ["BaseModelOutput", "False", "FloatTensor", "LongTensor", "ModelEncoder", "ModelPreTrainedModel", "None", "Pass", "Tensor", "True", "ValueError", "You", "_prepare_4d_attention_mask", "all_attentions", "and", "at", "attention_mask", "attentions", "bool", "both", "cannot", "class", "config", "def", "device", "dropout", "dropout_probability", "dtype", "either", "elif", "else", "embed_pos", "embed_positions", "embed_tokens", "encoder", "encoder_layer", "encoder_states", "enumerate", "for", "forward", "functional", "have", "hidden_states", "idx", "if", "in", "input", "input_ids", "input_shape", "inputs_embeds", "instead", "is", "is_t2u_encoder", "kwargs", "last_hidden_state", "layer_norm", "layer_outputs", "layerdrop", "layers", "model", "nn", "not", "of", "or", "output_attentions", "output_hidden_states", "p", "pass", "r", "raise", "rand", "return", "return_dict", "same", "self", "shape", "specify", "text_to_units", "the", "time", "to", "to_drop", "torch", "training", "tuple", "use_return_dict", "v", "view"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TDecoder.__init__": ["Embedding", "False", "LayerNorm", "ModelConfig", "ModelDecoder", "ModelDecoderLayer", "ModelPreTrainedModel", "ModelScaledWordEmbedding", "ModelSinusoidalPositionalEmbedding", "ModuleList", "None", "__init__", "append", "class", "config", "decoder_attention_heads", "decoder_ffn_dim", "decoder_layerdrop", "decoder_layers", "def", "dropout", "else", "embed_positions", "embed_scale", "embed_tokens", "embedding_dim", "for", "gradient_checkpointing", "hidden_size", "i", "if", "in", "is", "layer_idx", "layer_norm", "layerdrop", "layers", "math", "max_position_embeddings", "max_target_positions", "nn", "not", "num_embeddings", "pad_token_id", "padding_idx", "post_init", "r", "range", "scale_embedding", "self", "sqrt", "super", "vocab_size", "weight"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TDecoder.forward": ["BaseModelOutputWithPastAndCrossAttentions", "Cache", "DynamicCache", "EncoderDecoderCache", "False", "FloatTensor", "LongTensor", "ModelDecoder", "ModelPreTrainedModel", "None", "Setting", "Tensor", "True", "ValueError", "You", "_prepare_4d_attention_mask", "_prepare_4d_causal_attention_mask", "all_cross_attentions", "all_hidden_states", "all_self_attns", "and", "at", "attention_mask", "attentions", "bool", "both", "cache_position", "cannot", "checkpointing", "class", "config", "continue", "cross_attentions", "decoder_input_ids", "decoder_inputs_embeds", "decoder_layer", "def", "device", "dropout", "dropout_probability", "dtype", "either", "elif", "else", "embed_positions", "embed_tokens", "encoder_attention_mask", "encoder_hidden_states", "enumerate", "for", "forward", "functional", "get_seq_length", "gradient", "gradient_checkpointing", "have", "hidden_states", "idx", "if", "in", "incompatible", "input", "input_ids", "input_shape", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layer_norm", "layer_outputs", "layerdrop", "layers", "logger", "nn", "not", "or", "output_attentions", "output_hidden_states", "p", "past_key_values", "past_key_values_length", "positions", "raise", "rand", "return", "return_dict", "same", "self", "size", "specify", "tgt_len", "the", "time", "to", "torch", "training", "tuple", "use_cache", "use_return_dict", "v", "view", "warning_once", "with"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TTextToUnitModel.__init__": ["Embedding", "ModelConfig", "ModelDecoder", "ModelEncoder", "ModelPreTrainedModel", "ModelTextToUnitModel", "None", "True", "__init__", "class", "config", "decoder", "def", "embed_tokens_decoder", "encoder", "is_t2u_encoder", "nn", "post_init", "r", "self", "super"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TTextToUnitModel.forward": ["BaseModelOutput", "Cache", "FloatTensor", "LongTensor", "ModelPreTrainedModel", "ModelTextToUnitModel", "None", "Seq2SeqModelOutput", "Tensor", "and", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "cross_attentions", "decoder", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_input_ids", "decoder_inputs_embeds", "decoder_outputs", "def", "elif", "else", "encoder", "encoder_attention_mask", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "isinstance", "kwargs", "last_hidden_state", "len", "not", "output_attentions", "output_hidden_states", "past_key_values", "return", "return_dict", "self", "torch", "tuple", "use_cache", "use_return_dict"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TTextToUnitForConditionalGeneration.__init__": ["Embedding", "False", "GenerationMixin", "Linear", "ModelConfig", "ModelPreTrainedModel", "ModelTextToUnitForConditionalGeneration", "ModelTextToUnitModel", "None", "__init__", "__setattr__", "bias", "class", "config", "copy", "deepcopy", "def", "embed_tokens_decoder", "for", "hidden_size", "if", "in", "items", "lm_head", "model", "nn", "param", "post_init", "r", "self", "startswith", "super", "t2u_", "t2u_vocab_size", "to_dict", "val"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TTextToUnitForConditionalGeneration.get_encoder": ["GenerationMixin", "ModelPreTrainedModel", "ModelTextToUnitForConditionalGeneration", "class", "def", "encoder", "get_encoder", "model", "return", "self"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TTextToUnitForConditionalGeneration.get_decoder": ["GenerationMixin", "ModelPreTrainedModel", "ModelTextToUnitForConditionalGeneration", "class", "decoder", "def", "get_decoder", "model", "return", "self"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TTextToUnitForConditionalGeneration.get_input_embeddings": ["GenerationMixin", "ModelPreTrainedModel", "ModelTextToUnitForConditionalGeneration", "class", "decoder", "def", "embed_tokens", "get_input_embeddings", "model", "return", "self"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TTextToUnitForConditionalGeneration.set_input_embeddings": ["GenerationMixin", "ModelPreTrainedModel", "ModelTextToUnitForConditionalGeneration", "class", "decoder", "def", "embed_tokens", "model", "self", "set_input_embeddings", "value"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TTextToUnitForConditionalGeneration.forward": ["Cache", "CrossEntropyLoss", "False", "FloatTensor", "GenerationMixin", "LongTensor", "ModelPreTrainedModel", "ModelTextToUnitForConditionalGeneration", "None", "Seq2SeqLMOutput", "Tensor", "The", "and", "argument", "attention_mask", "bool", "cache_position", "changed", "class", "config", "cross_attentions", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_input_ids", "decoder_inputs_embeds", "def", "device", "else", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "forward", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "lm_head", "lm_logits", "logger", "logits", "loss", "loss_fct", "masked_lm_loss", "model", "not", "output", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "provided", "r", "return", "return_dict", "self", "shift_tokens_right", "since", "t2u_decoder_start_token_id", "t2u_pad_token_id", "to", "torch", "tuple", "use_cache", "use_return_dict", "view", "vocab_size", "warning"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TTextToUnitForConditionalGeneration.prepare_decoder_input_ids_from_labels": ["GenerationMixin", "ModelPreTrainedModel", "ModelTextToUnitForConditionalGeneration", "Tensor", "class", "config", "def", "labels", "prepare_decoder_input_ids_from_labels", "return", "self", "shift_tokens_right", "t2u_decoder_start_token_id", "t2u_pad_token_id", "torch"], "seamless_m4t/modeling_seamless_m4t.py:HifiGanResidualBlock.__init__": ["Conv1d", "ModelGanResidualBlock", "Module", "ModuleList", "_", "__init__", "channels", "class", "convs1", "convs2", "def", "dilation", "for", "get_padding", "i", "in", "kernel_size", "leaky_relu_slope", "len", "nn", "padding", "range", "self", "stride", "super"], "seamless_m4t/modeling_seamless_m4t.py:HifiGanResidualBlock.get_padding": ["ModelGanResidualBlock", "Module", "class", "def", "dilation", "get_padding", "kernel_size", "nn", "return", "self"], "seamless_m4t/modeling_seamless_m4t.py:HifiGanResidualBlock.apply_weight_norm": ["ModelGanResidualBlock", "Module", "apply_weight_norm", "class", "convs1", "convs2", "def", "for", "hasattr", "if", "in", "layer", "nn", "parametrizations", "self", "utils", "weight_norm"], "seamless_m4t/modeling_seamless_m4t.py:HifiGanResidualBlock.remove_weight_norm": ["ModelGanResidualBlock", "Module", "class", "convs1", "convs2", "def", "for", "in", "layer", "nn", "remove_weight_norm", "self", "utils"], "seamless_m4t/modeling_seamless_m4t.py:HifiGanResidualBlock.forward": ["ModelGanResidualBlock", "Module", "class", "conv1", "conv2", "convs1", "convs2", "def", "for", "forward", "functional", "hidden_states", "in", "leaky_relu", "leaky_relu_slope", "nn", "residual", "return", "self", "zip"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TVariancePredictor.__init__": ["Conv1d", "Dropout", "LayerNorm", "Linear", "ModelVariancePredictor", "Module", "ReLU", "__init__", "activation_function", "class", "config", "conv1", "conv2", "def", "dropout_module", "embed_dim", "kernel_size", "ln1", "ln2", "nn", "p", "padding", "proj", "self", "super", "unit_embed_dim", "var_pred_dropout", "variance_predictor_kernel_size"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TVariancePredictor.forward": ["ModelVariancePredictor", "Module", "Tensor", "activation_function", "class", "conv1", "conv2", "def", "dim", "dropout_module", "forward", "hidden_states", "ln1", "ln2", "nn", "proj", "return", "self", "squeeze", "transpose"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4THifiGan.__init__": ["Conv1d", "ConvTranspose1d", "HifiGanResidualBlock", "ModelConfig", "ModelHifiGan", "Module", "ModuleList", "__init__", "append", "channels", "class", "config", "conv_post", "conv_pre", "def", "dilation", "enumerate", "for", "i", "in", "kernel_size", "lang_embed_dim", "leaky_relu_slope", "len", "model_in_dim", "nn", "num_kernels", "num_upsamples", "padding", "range", "resblock_dilation_sizes", "resblock_kernel_sizes", "resblocks", "self", "spkr_embed_dim", "stride", "super", "unit_embed_dim", "upsample_initial_channel", "upsample_kernel_sizes", "upsample_rate", "upsample_rates", "upsampler", "zip"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4THifiGan.forward": ["FloatTensor", "ModelHifiGan", "Module", "class", "conv_post", "conv_pre", "def", "for", "forward", "functional", "hidden_states", "i", "in", "input_embeds", "j", "leaky_relu", "leaky_relu_slope", "nn", "num_kernels", "num_upsamples", "r", "range", "res_state", "resblocks", "return", "self", "squeeze", "tanh", "torch", "upsampler", "waveform"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TCodeHifiGan.__init__": ["Embedding", "ModelCodeHifiGan", "ModelHifiGan", "ModelVariancePredictor", "PreTrainedModel", "__init__", "class", "config", "def", "dur_predictor", "hifi_gan", "lang_embed_dim", "language_embedding", "nn", "pad_token_id", "post_init", "self", "speaker_embedding", "spkr_embed_dim", "super", "t2u_pad_token_id", "unit_embed_dim", "unit_embedding", "unit_hifi_gan_vocab_size", "vocoder_num_langs", "vocoder_num_spkrs"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TCodeHifiGan._get_dur_output_lengths": ["ModelCodeHifiGan", "PreTrainedModel", "_get_dur_output_lengths", "clamp", "class", "cumsum", "cumulative_dur_out", "def", "dim", "dur_out", "gather", "index", "input_ids", "pad_token_id", "return", "self", "shape", "squeeze", "sum", "torch", "unit_lengths", "unsqueeze"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TCodeHifiGan._get_output_hifigan_lengths": ["LongTensor", "ModelCodeHifiGan", "PreTrainedModel", "_conv_out_length", "_get_output_hifigan_lengths", "_transpose_conv_out_length", "class", "config", "def", "dil", "dilation", "div", "enumerate", "floor", "for", "i", "in", "input_length", "input_lengths", "int", "kernel_size", "len", "pad", "range", "resblock_dilation_sizes", "resblock_kernel_sizes", "return", "rounding_mode", "self", "stride", "torch", "upsample_kernel_sizes", "upsample_rate", "upsample_rates", "zip"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TCodeHifiGan.forward": ["LongTensor", "ModelCodeHifiGan", "PreTrainedModel", "Tensor", "True", "_get_dur_output_lengths", "_get_output_hifigan_lengths", "and", "batch_first", "cat", "clamp", "class", "def", "dim", "dur_out", "dur_predictor", "duration", "else", "expm1", "for", "forward", "hidden_state", "hidden_states", "hifi_gan", "if", "in", "input_ids", "kwargs", "lang", "lang_id", "language_embedding", "lengths", "log_dur_pred", "logger", "long", "min", "nn", "pad_sequence", "repeat", "repeat_interleave", "return", "rnn", "round", "self", "shape", "size", "speaker_embedding", "spkr", "spkr_id", "torch", "training", "transpose", "tuple", "unit_embedding", "unit_lengths", "utils", "view", "warning", "zip"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TCodeHifiGan.apply_weight_norm": ["ModelCodeHifiGan", "PreTrainedModel", "apply_weight_norm", "class", "conv_post", "conv_pre", "def", "for", "hasattr", "hifi_gan", "if", "in", "layer", "nn", "parametrizations", "resblocks", "self", "upsampler", "utils", "weight_norm"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TCodeHifiGan.remove_weight_norm": ["ModelCodeHifiGan", "PreTrainedModel", "class", "conv_post", "conv_pre", "def", "for", "hifi_gan", "in", "layer", "nn", "remove_weight_norm", "resblocks", "self", "upsampler", "utils"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TForTextToText.__init__": ["Embedding", "False", "GenerationMixin", "Linear", "ModelConfig", "ModelDecoder", "ModelEncoder", "ModelForTextToText", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "nn", "pad_token_id", "post_init", "self", "shared", "super", "text_decoder", "text_encoder", "vocab_size"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TForTextToText.get_encoder": ["GenerationMixin", "ModelForTextToText", "ModelPreTrainedModel", "class", "def", "get_encoder", "return", "self", "text_encoder"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TForTextToText.get_decoder": ["GenerationMixin", "ModelForTextToText", "ModelPreTrainedModel", "class", "def", "get_decoder", "return", "self", "text_decoder"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TForTextToText.get_input_embeddings": ["GenerationMixin", "ModelForTextToText", "ModelPreTrainedModel", "class", "def", "embed_tokens", "get_input_embeddings", "return", "self", "text_decoder"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TForTextToText.set_input_embeddings": ["GenerationMixin", "ModelForTextToText", "ModelPreTrainedModel", "class", "def", "embed_tokens", "self", "set_input_embeddings", "shared", "text_decoder", "text_encoder", "value"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TForTextToText.forward": ["BaseModelOutput", "Cache", "CrossEntropyLoss", "False", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForTextToText", "ModelPreTrainedModel", "None", "Seq2SeqLMOutput", "Tensor", "The", "and", "argument", "attention_mask", "attentions", "bool", "changed", "class", "config", "cross_attentions", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_input_ids", "decoder_inputs_embeds", "decoder_outputs", "decoder_start_token_id", "def", "device", "elif", "else", "encoder_attention_mask", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "len", "lm_head", "lm_logits", "logger", "logits", "loss", "loss_fct", "masked_lm_loss", "not", "output", "output_attentions", "output_hidden_states", "outputs", "pad_token_id", "past_key_values", "provided", "r", "return", "return_dict", "self", "shift_tokens_right", "since", "text_decoder", "text_encoder", "to", "torch", "tuple", "use_cache", "use_return_dict", "view", "vocab_size", "warning"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TForTextToText.generate": ["False", "GenerationMixin", "ModelForTextToText", "ModelPreTrainedModel", "None", "ValueError", "__", "batch_size", "class", "decoder_input_ids", "def", "device", "else", "f", "generate", "generation_config", "get", "hasattr", "if", "in", "input_ids", "inputs_embeds", "is", "kwargs", "len", "logger", "logits_processor", "not", "pop", "prefix_allowed_tokens_fn", "raise", "replace", "return", "self", "stopping_criteria", "super", "synced_gpus", "tensor", "text_decoder_input_ids", "text_decoder_lang_to_code_id", "text_tgt_lang_id", "tgt_lang", "torch", "warning"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TForSpeechToText.__init__": ["Embedding", "False", "GenerationMixin", "Linear", "ModelConfig", "ModelDecoder", "ModelForSpeechToText", "ModelPreTrainedModel", "ModelSpeechEncoder", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "nn", "pad_token_id", "post_init", "self", "shared", "speech_encoder", "super", "text_decoder", "vocab_size"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TForSpeechToText.get_encoder": ["GenerationMixin", "ModelForSpeechToText", "ModelPreTrainedModel", "class", "def", "get_encoder", "return", "self", "speech_encoder"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TForSpeechToText.get_decoder": ["GenerationMixin", "ModelForSpeechToText", "ModelPreTrainedModel", "class", "def", "get_decoder", "return", "self", "text_decoder"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TForSpeechToText.get_input_embeddings": ["GenerationMixin", "ModelForSpeechToText", "ModelPreTrainedModel", "class", "def", "embed_tokens", "get_input_embeddings", "return", "self", "text_decoder"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TForSpeechToText.set_input_embeddings": ["GenerationMixin", "ModelForSpeechToText", "ModelPreTrainedModel", "class", "def", "embed_tokens", "self", "set_input_embeddings", "text_decoder", "value"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TForSpeechToText.forward": ["BaseModelOutput", "Cache", "CrossEntropyLoss", "False", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForSpeechToText", "ModelPreTrainedModel", "None", "Seq2SeqLMOutput", "Tensor", "The", "_compute_new_attention_mask", "_compute_sub_sample_lengths_from_attention_mask", "and", "argument", "attention_mask", "attentions", "bool", "changed", "class", "config", "cross_attentions", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_input_ids", "decoder_inputs_embeds", "decoder_outputs", "decoder_start_token_id", "def", "device", "elif", "else", "encoder_attention_mask", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "forward", "hidden_states", "if", "input_features", "input_ids", "inputs_embeds", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "len", "lm_head", "lm_logits", "logger", "logits", "loss", "loss_fct", "masked_lm_loss", "not", "output", "output_attentions", "output_hidden_states", "outputs", "pad_token_id", "past_key_values", "provided", "r", "return", "return_dict", "self", "seq_lens", "shift_tokens_right", "since", "speech_encoder", "sub_sampled_lengths", "text_decoder", "to", "torch", "tuple", "use_cache", "use_return_dict", "view", "vocab_size", "warning"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TForSpeechToText.generate": ["False", "GenerationMixin", "ModelForSpeechToText", "ModelPreTrainedModel", "None", "ValueError", "__", "batch_size", "class", "decoder_input_ids", "def", "device", "else", "encoder_outputs", "f", "generate", "generation_config", "get", "hasattr", "if", "in", "input_embeds", "input_features", "inputs", "is", "kwargs", "last_hidden_state", "len", "logger", "logits_processor", "not", "pop", "prefix_allowed_tokens_fn", "raise", "replace", "return", "self", "stopping_criteria", "super", "synced_gpus", "tensor", "text_decoder_input_ids", "text_decoder_lang_to_code_id", "text_tgt_lang_id", "tgt_lang", "torch", "warning"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TForTextToSpeech.__init__": ["Embedding", "False", "GenerationMixin", "Linear", "ModelCodeHifiGan", "ModelConfig", "ModelDecoder", "ModelEncoder", "ModelForTextToSpeech", "ModelPreTrainedModel", "ModelTextToUnitForConditionalGeneration", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "nn", "pad_token_id", "post_init", "self", "shared", "super", "t2u_model", "text_decoder", "text_encoder", "vocab_size", "vocoder"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TForTextToSpeech.get_encoder": ["GenerationMixin", "ModelForTextToSpeech", "ModelPreTrainedModel", "class", "def", "get_encoder", "return", "self", "text_encoder"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TForTextToSpeech.get_decoder": ["GenerationMixin", "ModelForTextToSpeech", "ModelPreTrainedModel", "class", "def", "get_decoder", "return", "self", "text_decoder"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TForTextToSpeech.get_input_embeddings": ["GenerationMixin", "ModelForTextToSpeech", "ModelPreTrainedModel", "class", "def", "embed_tokens", "get_input_embeddings", "return", "self", "text_decoder"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TForTextToSpeech.set_input_embeddings": ["GenerationMixin", "ModelForTextToSpeech", "ModelPreTrainedModel", "class", "def", "embed_tokens", "self", "set_input_embeddings", "shared", "text_decoder", "text_encoder", "value"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TForTextToSpeech.forward": ["BaseModelOutput", "Cache", "CrossEntropyLoss", "False", "FloatTensor", "GenerationMixin", "If", "It", "LongTensor", "ModelForTextToSpeech", "ModelForTextToText", "ModelPreTrainedModel", "ModelTextToUnitForConditionalGeneration", "None", "Seq2SeqLMOutput", "Tensor", "The", "This", "and", "argument", "as", "attention_mask", "attentions", "bool", "cache_position", "changed", "class", "config", "cross_attentions", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_input_ids", "decoder_inputs_embeds", "decoder_outputs", "decoder_start_token_id", "def", "device", "doesn", "elif", "else", "encoder_attention_mask", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "forward", "generate", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "len", "lm_head", "lm_logits", "logger", "logits", "loss", "loss_fct", "masked_lm_loss", "method", "model", "not", "output", "output_attentions", "output_hidden_states", "outputs", "pad_token_id", "past_key_values", "provided", "r", "return", "return_dict", "same", "self", "shift_tokens_right", "since", "speech", "t", "text", "text_decoder", "text_encoder", "the", "to", "torch", "tuple", "unit", "use", "use_cache", "use_return_dict", "view", "vocab_size", "want", "warning", "you"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TForTextToSpeech.generate": ["GenerationMixin", "ModelForTextToSpeech", "ModelGenerationOutput", "ModelPreTrainedModel", "None", "Tensor", "True", "ValueError", "You", "__", "_compute_new_attention_mask", "a", "arange", "argmax", "attention_mask", "batch_size", "bool", "class", "clone", "config", "decoder_input_ids", "def", "detach", "device", "elif", "else", "encoder_attention_mask", "encoder_hidden_states", "f", "for", "format_speech_generation_kwargs", "generate", "generation_config", "get", "getattr", "idx_most_probable_sequences_per_batch", "if", "in", "input_ids", "inputs_embeds", "int", "is", "key", "kwargs", "kwargs_speech", "kwargs_text", "lang_code_to_id", "lang_id", "last_hidden_state", "len", "must", "not", "num_return_sequences", "output_hidden_states", "output_scores", "output_unit_ids", "pad_token_id", "raise", "replace", "return", "return_dict_in_generate", "return_intermediate_token_ids", "self", "seq_lens", "sequences", "sequences_scores", "shape", "specify", "speech", "spkr_id", "str", "sum", "super", "t2u_decoder_input_ids", "t2u_eos_token_id", "t2u_input_embeds", "t2u_lang_code_to_id", "t2u_model", "t2u_model_attention_mask", "t2u_pad_token_id", "t2u_tgt_lang_id", "tensor", "text_decoder", "text_decoder_input_ids", "text_decoder_lang_to_code_id", "text_generation_output", "text_tgt_lang_id", "tgt_lang", "to", "torch", "translated", "unit_ids", "unit_sequences", "view", "vocoder", "vocoder_lang_code_to_id", "vocoder_offset", "vocoder_tgt_lang_id", "waveform", "waveform_lengths", "where"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TForSpeechToSpeech.__init__": ["Embedding", "False", "GenerationMixin", "Linear", "ModelCodeHifiGan", "ModelDecoder", "ModelForSpeechToSpeech", "ModelPreTrainedModel", "ModelSpeechEncoder", "ModelTextToUnitForConditionalGeneration", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "nn", "pad_token_id", "post_init", "self", "shared", "speech_encoder", "super", "t2u_model", "text_decoder", "vocab_size", "vocoder"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TForSpeechToSpeech.get_encoder": ["GenerationMixin", "ModelForSpeechToSpeech", "ModelPreTrainedModel", "class", "def", "get_encoder", "return", "self", "speech_encoder"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TForSpeechToSpeech.get_decoder": ["GenerationMixin", "ModelForSpeechToSpeech", "ModelPreTrainedModel", "class", "def", "get_decoder", "return", "self", "text_decoder"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TForSpeechToSpeech.get_input_embeddings": ["GenerationMixin", "ModelForSpeechToSpeech", "ModelPreTrainedModel", "class", "def", "embed_tokens", "get_input_embeddings", "return", "self", "text_decoder"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TForSpeechToSpeech.set_input_embeddings": ["GenerationMixin", "ModelForSpeechToSpeech", "ModelPreTrainedModel", "class", "def", "embed_tokens", "self", "set_input_embeddings", "text_decoder", "value"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TForSpeechToSpeech.forward": ["BaseModelOutput", "Cache", "CrossEntropyLoss", "False", "FloatTensor", "GenerationMixin", "If", "It", "LongTensor", "ModelForSpeechToSpeech", "ModelForSpeechToText", "ModelPreTrainedModel", "None", "Seq2SeqLMOutput", "Tensor", "The", "This", "_compute_new_attention_mask", "_compute_sub_sample_lengths_from_attention_mask", "and", "argument", "as", "attention_mask", "attentions", "bool", "changed", "class", "config", "cross_attentions", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_input_ids", "decoder_inputs_embeds", "decoder_outputs", "decoder_start_token_id", "def", "device", "doesn", "elif", "else", "encoder_attention_mask", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "forward", "generate", "hidden_states", "if", "input_features", "input_ids", "inputs_embeds", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "len", "lm_head", "lm_logits", "logger", "logits", "loss", "loss_fct", "masked_lm_loss", "method", "not", "output", "output_attentions", "output_hidden_states", "outputs", "pad_token_id", "past_key_values", "provided", "r", "return", "return_dict", "same", "self", "seq_lens", "shift_tokens_right", "since", "speech", "speech_encoder", "sub_sampled_lengths", "t", "t2u_model", "text_decoder", "the", "to", "torch", "tuple", "use", "use_cache", "use_return_dict", "view", "vocab_size", "want", "warning", "you"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TForSpeechToSpeech.generate": ["GenerationMixin", "ModelForSpeechToSpeech", "ModelGenerationOutput", "ModelPreTrainedModel", "None", "Tensor", "True", "ValueError", "You", "__", "_compute_new_attention_mask", "_compute_sub_sample_lengths_from_attention_mask", "a", "arange", "argmax", "attention_mask", "batch_size", "bool", "class", "clone", "config", "decoder_input_ids", "def", "detach", "device", "elif", "else", "encoder_attention_mask", "encoder_hidden_states", "f", "for", "format_speech_generation_kwargs", "generate", "generation_config", "get", "getattr", "hidden_states", "idx_most_probable_sequences_per_batch", "if", "in", "input_features", "input_ids", "inputs_embeds", "int", "is", "key", "kwargs", "kwargs_speech", "kwargs_text", "lang_code_to_id", "lang_id", "last_hidden_state", "len", "must", "not", "num_return_sequences", "output_hidden_states", "output_scores", "output_unit_ids", "pad_token_id", "raise", "replace", "return", "return_dict_in_generate", "return_intermediate_token_ids", "self", "seq_lens", "sequences", "sequences_scores", "shape", "specify", "speech", "speech_encoder", "spkr_id", "str", "sub_sampled_lengths", "sum", "super", "t2u_decoder_input_ids", "t2u_eos_token_id", "t2u_input_embeds", "t2u_lang_code_to_id", "t2u_model", "t2u_model_attention_mask", "t2u_pad_token_id", "t2u_tgt_lang_id", "tensor", "text_decoder", "text_decoder_input_ids", "text_decoder_lang_to_code_id", "text_generation_output", "text_tgt_lang_id", "tgt_lang", "to", "torch", "translated", "unit_ids", "unit_sequences", "view", "vocoder", "vocoder_lang_code_to_id", "vocoder_offset", "vocoder_tgt_lang_id", "waveform", "waveform_lengths", "where"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TModel.__init__": ["Embedding", "False", "GenerationMixin", "Linear", "ModelCodeHifiGan", "ModelDecoder", "ModelEncoder", "ModelModel", "ModelPreTrainedModel", "ModelSpeechEncoder", "ModelTextToUnitForConditionalGeneration", "__init__", "bias", "class", "config", "current_modality", "def", "hidden_size", "if", "input_features", "lm_head", "main_input_name", "nn", "pad_token_id", "post_init", "r", "self", "shared", "speech", "speech_encoder", "super", "t2u_model", "text", "text_decoder", "text_encoder", "vocab_size", "vocoder"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TModel.set_modality": ["GenerationMixin", "It", "ModelModel", "ModelPreTrainedModel", "ValueError", "a", "be", "class", "current_modality", "def", "elif", "else", "f", "if", "input_features", "input_ids", "is", "main_input_name", "modality", "must", "not", "or", "raise", "self", "set_modality", "speech", "text", "valid"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TModel.get_encoder": ["GenerationMixin", "ModelModel", "ModelPreTrainedModel", "class", "current_modality", "def", "else", "get_encoder", "if", "return", "self", "speech_encoder", "text", "text_encoder"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TModel.get_input_embeddings": ["GenerationMixin", "ModelModel", "ModelPreTrainedModel", "class", "def", "embed_tokens", "get_input_embeddings", "return", "self", "text_decoder"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TModel.set_input_embeddings": ["GenerationMixin", "ModelModel", "ModelPreTrainedModel", "class", "def", "embed_tokens", "self", "set_input_embeddings", "shared", "text_decoder", "text_encoder", "value"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TModel.forward": ["BaseModelOutput", "Cache", "CrossEntropyLoss", "False", "FloatTensor", "GenerationMixin", "If", "LongTensor", "Make", "ModelForSpeechToText", "ModelForTextToText", "ModelModel", "ModelPreTrainedModel", "None", "Seq2SeqLMOutput", "Tensor", "The", "This", "ValueError", "_compute_new_attention_mask", "_compute_sub_sample_lengths_from_attention_mask", "all", "and", "are", "argument", "as", "at", "attention_mask", "attentions", "be", "been", "bool", "but", "calls", "changed", "class", "config", "cross_attentions", "current_modality", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_input_ids", "decoder_inputs_embeds", "decoder_outputs", "decoder_start_token_id", "def", "depending", "device", "elif", "else", "empty", "encoder_attention_mask", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "exclusive", "forward", "generate", "given", "has", "hidden_states", "if", "ignored", "in", "input", "input_features", "input_ids", "inputs_embeds", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "least", "len", "lm_head", "lm_logits", "logger", "logits", "loss", "loss_fct", "masked_lm_loss", "method", "modality", "mutually", "not", "of", "on", "one", "or", "output", "output_attentions", "output_hidden_states", "outputs", "pad_token_id", "past_key_values", "priority", "provided", "r", "raise", "return", "return_dict", "same", "self", "seq_lens", "set_modality", "shift_tokens_right", "since", "speech", "speech_encoder", "sub_sampled_lengths", "sure", "text", "text_decoder", "text_encoder", "that", "the", "them", "through", "to", "torch", "tuple", "use", "use_cache", "use_return_dict", "used", "view", "vocab_size", "want", "warning", "will", "you"], "seamless_m4t/modeling_seamless_m4t.py:SeamlessM4TModel.generate": ["GenerationMixin", "Make", "ModelGenerationOutput", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "True", "ValueError", "You", "__", "_compute_new_attention_mask", "_compute_sub_sample_lengths_from_attention_mask", "a", "all", "and", "arange", "are", "argmax", "at", "attention_mask", "batch_size", "be", "bool", "both", "class", "clone", "config", "current_modality", "decoder_input_ids", "def", "detach", "device", "elif", "else", "empty", "encoder", "encoder_attention_mask", "encoder_hidden_states", "f", "for", "format_speech_generation_kwargs", "generate", "generate_speech", "generation_config", "get", "getattr", "hidden_states", "idx_most_probable_sequences_per_batch", "if", "in", "input_features", "input_ids", "inputs_embeds", "int", "is", "key", "kwargs", "kwargs_speech", "kwargs_text", "lang_code_to_id", "lang_id", "last_hidden_state", "least", "len", "logger", "must", "non", "not", "num_return_sequences", "of", "one", "output_hidden_states", "output_scores", "output_unit_ids", "pad_token_id", "priority", "raise", "replace", "return", "return_dict_in_generate", "return_intermediate_token_ids", "self", "seq_lens", "sequences", "sequences_scores", "set_modality", "shape", "specify", "speech", "speech_encoder", "spkr_id", "str", "sub_sampled_lengths", "sum", "super", "sure", "t2u_decoder_input_ids", "t2u_eos_token_id", "t2u_input_embeds", "t2u_lang_code_to_id", "t2u_model", "t2u_model_attention_mask", "t2u_pad_token_id", "t2u_tgt_lang_id", "tensor", "text", "text_decoder", "text_decoder_input_ids", "text_decoder_lang_to_code_id", "text_generation_output", "text_tgt_lang_id", "tgt_lang", "the", "them", "through", "to", "torch", "translated", "unit_ids", "unit_sequences", "use", "used", "view", "vocoder", "vocoder_lang_code_to_id", "vocoder_offset", "vocoder_tgt_lang_id", "want", "warning", "waveform", "waveform_lengths", "where", "will", "you"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:shift_tokens_right": ["Model_tokens_right", "Modeled_input_ids", "None", "Tensor", "ValueError", "be", "clone", "config", "decoder_start_token_id", "def", "defined", "has", "if", "input_ids", "int", "is", "masked_fill_", "model", "new_zeros", "pad_token_id", "raise", "return", "self", "shape", "to", "torch"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:_compute_new_attention_mask": ["Tensor", "_compute_new_attention_mask", "arange", "batch_size", "bool_mask", "def", "device", "expand", "hidden_states", "indices", "mask", "mask_seq_len", "masked_fill", "new_ones", "return", "seq_lens", "shape", "torch", "unsqueeze"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:format_speech_generation_kwargs": ["Model_speech_generation_kwargs", "def", "elif", "else", "for", "generation_config", "if", "in", "items", "key", "kwargs", "kwargs_speech", "kwargs_text", "len", "not", "return", "speech_", "startswith", "text_", "value"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2ConformerFeatureProjection.__init__": ["Dropout", "LayerNorm", "Linear", "ModelConformerFeatureProjection", "Module", "__init__", "class", "config", "def", "dropout", "eps", "feature_projection_input_dim", "hidden_size", "layer_norm", "layer_norm_eps", "nn", "projection", "self", "speech_encoder_dropout", "super"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2ConformerFeatureProjection.forward": ["ModelConformerFeatureProjection", "Module", "class", "def", "dropout", "dtype", "forward", "hidden_states", "layer_norm", "nn", "norm_hidden_states", "projection", "return", "self", "to", "weight"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2ConformerFeedForward.__init__": ["ACT2FN", "Dropout", "Linear", "ModelConformerFeedForward", "Module", "None", "__init__", "act_fn", "class", "config", "def", "dropout", "else", "hidden_size", "if", "intermediate_act_fn", "intermediate_dense", "intermediate_dropout", "is", "isinstance", "nn", "not", "output_dense", "output_dropout", "self", "speech_encoder_dropout", "speech_encoder_hidden_act", "speech_encoder_intermediate_size", "str", "super"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2ConformerFeedForward.forward": ["ModelConformerFeedForward", "Module", "class", "def", "forward", "hidden_states", "intermediate_act_fn", "intermediate_dense", "intermediate_dropout", "nn", "output_dense", "output_dropout", "return", "self"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2ConformerConvolutionModule.__init__": ["ACT2FN", "Conv1d", "Dropout", "False", "GLU", "LayerNorm", "ModelConformerConvolutionModule", "Module", "SAME", "ValueError", "__init__", "a", "activation", "be", "bias", "class", "config", "conv_depthwise_kernel_size", "def", "depthwise_conv", "depthwise_layer_norm", "dim", "dropout", "for", "glu", "groups", "hidden_size", "if", "kernel_size", "layer_norm", "nn", "number", "odd", "padding", "pointwise_conv1", "pointwise_conv2", "raise", "self", "should", "speech_encoder_dropout", "speech_encoder_hidden_act", "stride", "super"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2ConformerConvolutionModule.forward": ["ModelConformerConvolutionModule", "Module", "None", "activation", "attention_mask", "bool", "class", "def", "depthwise_conv", "depthwise_layer_norm", "dropout", "forward", "functional", "glu", "hidden_states", "if", "is", "kernel_size", "layer_norm", "masked_fill", "nn", "not", "pad", "pointwise_conv1", "pointwise_conv2", "return", "self", "torch", "transpose", "unsqueeze"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2ConformerSelfAttention.__init__": ["Dropout", "Embedding", "Linear", "ModelConformerSelfAttention", "Module", "None", "True", "__init__", "class", "config", "def", "distance_embedding", "dropout", "else", "head_size", "hidden_size", "if", "left_max_position_embeddings", "linear_k", "linear_out", "linear_q", "linear_v", "nn", "num_heads", "num_positions", "p", "position_embeddings_type", "relative_key", "right_max_position_embeddings", "self", "speech_encoder_attention_heads", "speech_encoder_dropout", "super", "use_position_embeddings"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2ConformerSelfAttention.forward": ["False", "ModelConformerSelfAttention", "Module", "None", "Tensor", "arange", "attention_mask", "attn_output", "attn_weights", "batch_size", "bhld", "bhlr", "bool", "clamp", "class", "def", "device", "dim", "distance", "distance_embedding", "dropout", "dtype", "einsum", "forward", "head_size", "hidden_size", "hidden_states", "if", "is", "key", "key_length", "left_max_position_embeddings", "linear_k", "linear_out", "linear_q", "linear_v", "long", "lrd", "math", "matmul", "nn", "not", "num_heads", "output_attentions", "position_embeddings_type", "position_ids_l", "position_ids_r", "positional_embedding", "query", "query_key_states", "query_length", "relative_key", "relative_position_attn_weights", "reshape", "return", "right_max_position_embeddings", "self", "sequence_length", "shape", "size", "softmax", "sqrt", "to", "torch", "transpose", "tuple", "value", "value_states", "view"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2ConformerEncoderLayer.__init__": ["Dropout", "GradientCheckpointingLayer", "LayerNorm", "ModelConformerConvolutionModule", "ModelConformerEncoderLayer", "ModelConformerFeedForward", "ModelConformerSelfAttention", "__init__", "class", "config", "conv_module", "def", "dropout", "embed_dim", "ffn1", "ffn1_layer_norm", "ffn2", "ffn2_layer_norm", "final_layer_norm", "hidden_size", "nn", "self", "self_attn", "self_attn_dropout", "self_attn_layer_norm", "speech_encoder_dropout", "super"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2ConformerEncoderLayer.forward": ["False", "GradientCheckpointingLayer", "ModelConformerEncoderLayer", "None", "Tensor", "attention_mask", "attn_weights", "bool", "class", "conv_attention_mask", "conv_module", "def", "ffn1", "ffn1_layer_norm", "ffn2", "ffn2_layer_norm", "final_layer_norm", "forward", "hidden_states", "output_attentions", "residual", "return", "self", "self_attn", "self_attn_dropout", "self_attn_layer_norm", "torch"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2ConformerEncoder.__init__": ["Dropout", "False", "LayerNorm", "ModelConformerEncoder", "ModelConformerEncoderLayer", "Module", "ModuleList", "_", "__init__", "class", "config", "def", "dropout", "eps", "for", "gradient_checkpointing", "hidden_size", "in", "layer_norm", "layer_norm_eps", "layers", "nn", "range", "self", "speech_encoder_dropout", "speech_encoder_layers", "super"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2ConformerEncoder._apply_chunk_attention": ["ModelConformerEncoder", "Module", "None", "_apply_chunk_attention", "arange", "attention_mask", "bool", "chunk_indices", "chunk_mask", "clamp_", "class", "config", "def", "device", "div", "dtype", "else", "end_indices", "expand", "full_like", "hidden_states", "if", "indices", "is", "long", "max", "min", "nn", "return", "self", "sequence_len", "shape", "speech_encoder_chunk_size", "speech_encoder_left_chunk_num", "start_indices", "to", "torch", "unsqueeze"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2ConformerEncoder.forward": ["BaseModelOutput", "False", "ModelConformerEncoder", "Module", "None", "True", "_apply_chunk_attention", "all_hidden_states", "all_self_attentions", "and", "attention_mask", "attentions", "bool", "class", "config", "conv_attention_mask", "def", "dropout", "dropout_probability", "dtype", "else", "enumerate", "expand", "finfo", "for", "forward", "hidden_states", "i", "if", "in", "is", "is_deepspeed_zero3_enabled", "is_fsdp_managed_module", "last_hidden_state", "layer", "layer_norm", "layer_outputs", "layers", "masked_fill", "min", "nn", "not", "or", "output_attentions", "output_hidden_states", "rand", "return", "return_dict", "self", "shape", "skip_the_layer", "speech_encoder_chunk_size", "speech_encoder_layerdrop", "synced_gpus", "to", "torch", "training", "tuple", "unsqueeze", "v"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2ConformerAdapterLayer.__init__": ["Conv1d", "Dropout", "False", "GLU", "LayerNorm", "ModelConformerAdapterLayer", "ModelConformerFeedForward", "ModelConformerSelfAttention", "Module", "__init__", "act_fn", "activation", "adaptor_dropout", "adaptor_kernel_size", "adaptor_stride", "class", "config", "def", "dim", "dropout", "embed_dim", "ffn", "ffn_layer_norm", "hidden_size", "kernel_size", "nn", "padding", "relu", "residual_conv", "residual_layer_norm", "self", "self_attn", "self_attn_conv", "self_attn_dropout", "self_attn_layer_norm", "stride", "super", "use_position_embeddings"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2ConformerAdapterLayer._compute_sub_sample_lengths_from_attention_mask": ["ModelConformerAdapterLayer", "Module", "_compute_sub_sample_lengths_from_attention_mask", "attention_mask", "class", "def", "floor", "int", "kernel_size", "nn", "pad", "return", "self", "seq_lens", "size", "stride", "sum"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2ConformerAdapterLayer.forward": ["False", "ModelConformerAdapterLayer", "Module", "None", "Tensor", "_compute_new_attention_mask", "_compute_sub_sample_lengths_from_attention_mask", "_prepare_4d_attention_mask", "activation", "attention_mask", "attn_weights", "bool", "class", "def", "device", "dtype", "ffn", "ffn_layer_norm", "forward", "hidden_states", "if", "is", "nn", "not", "output_attentions", "residual", "residual_conv", "residual_layer_norm", "return", "self", "self_attn", "self_attn_conv", "self_attn_dropout", "self_attn_layer_norm", "seq_lens", "sub_sampled_lengths", "to", "torch", "transpose"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2ConformerAdapter.__init__": ["ModelConformerAdapter", "ModelConformerAdapterLayer", "Module", "ModuleList", "_", "__init__", "class", "config", "def", "for", "in", "layers", "nn", "num_adapter_layers", "range", "self", "super"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2ConformerAdapter.forward": ["ModelConformerAdapter", "Module", "attention_mask", "class", "def", "for", "forward", "hidden_states", "in", "layer", "layers", "nn", "return", "self"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2ScaledWordEmbedding.__init__": ["Embedding", "ModelScaledWordEmbedding", "None", "__init__", "class", "def", "embed_scale", "embedding_dim", "float", "int", "nn", "num_embeddings", "padding_idx", "self", "super"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2ScaledWordEmbedding.forward": ["Embedding", "ModelScaledWordEmbedding", "Tensor", "class", "def", "embed_scale", "forward", "input_ids", "nn", "return", "self", "super", "torch"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2SinusoidalPositionalEmbedding.__init__": ["ModelSinusoidalPositionalEmbedding", "Module", "None", "__init__", "class", "def", "embedding_dim", "int", "make_weights", "nn", "num_positions", "offset", "padding_idx", "self", "super"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2SinusoidalPositionalEmbedding.make_weights": ["False", "ModelSinusoidalPositionalEmbedding", "Module", "None", "class", "def", "device", "dtype", "emb_weights", "embedding_dim", "get_embedding", "hasattr", "if", "int", "make_weights", "nn", "num_embeddings", "padding_idx", "persistent", "register_buffer", "self", "to", "weights"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2SinusoidalPositionalEmbedding.get_embedding": ["ModelSinusoidalPositionalEmbedding", "Module", "None", "arange", "cat", "class", "cos", "def", "dim", "dtype", "emb", "embedding_dim", "exp", "float", "get_default_dtype", "get_embedding", "half_dim", "if", "int", "int64", "is", "log", "math", "nn", "not", "num_embeddings", "padding_idx", "return", "sin", "to", "torch", "unsqueeze", "view", "zeros"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2SinusoidalPositionalEmbedding.forward": ["ModelSinusoidalPositionalEmbedding", "Module", "None", "Tensor", "bsz", "class", "create_position_ids_from_input_ids", "create_position_ids_from_inputs_embeds", "def", "detach", "device", "else", "embedding_dim", "forward", "if", "index_select", "input_ids", "inputs_embeds", "int", "is", "make_weights", "max_pos", "nn", "not", "offset", "padding_idx", "past_key_values_length", "position_ids", "return", "self", "seq_len", "shape", "size", "to", "torch", "view", "weights"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2SinusoidalPositionalEmbedding.create_position_ids_from_inputs_embeds": ["ModelSinusoidalPositionalEmbedding", "Module", "arange", "class", "contiguous", "create_position_ids_from_inputs_embeds", "def", "device", "dtype", "expand", "input_shape", "inputs_embeds", "long", "nn", "padding_idx", "past_key_values_length", "position_ids", "return", "sequence_length", "size", "torch", "unsqueeze"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2SinusoidalPositionalEmbedding.create_position_ids_from_input_ids": ["ModelSinusoidalPositionalEmbedding", "Module", "class", "create_position_ids_from_input_ids", "cumsum", "def", "dim", "incremental_indices", "input_ids", "int", "long", "mask", "ne", "nn", "padding_idx", "past_key_values_length", "return", "torch", "type_as"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2Attention.__init__": ["False", "Instantiating", "Linear", "ModelAttention", "ModelConfig", "Module", "None", "Please", "True", "ValueError", "__class__", "__init__", "__name__", "a", "and", "be", "bias", "bool", "by", "caching", "call", "class", "config", "creating", "decoder", "def", "divisible", "dropout", "during", "embed_dim", "errors", "f", "float", "forward", "got", "head_dim", "if", "int", "is", "is_causal", "is_decoder", "k_proj", "layer_idx", "lead", "logger", "make", "must", "nn", "not", "num_heads", "out_proj", "passing", "provide", "q_proj", "raise", "recommended", "scaling", "self", "super", "sure", "the", "this", "to", "used", "v_proj", "warning_once", "when", "will", "without"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2Attention.forward": ["Cache", "EncoderDecoderCache", "False", "ModelAttention", "Module", "None", "Tensor", "True", "and", "attention_mask", "attention_scores", "attn_output", "attn_weights", "batch_size", "bool", "cache_position", "class", "context_states", "contiguous", "cross_attention_cache", "curr_past_key_values", "current_states", "def", "dim", "dropout", "else", "encoder_hidden_states", "float", "forward", "functional", "get", "head_dim", "hidden_states", "if", "is", "is_cross_attention", "is_updated", "isinstance", "k_proj", "key_states", "keys", "layer_idx", "layers", "matmul", "nn", "not", "num_heads", "out_proj", "output_attentions", "p", "past_key_values", "permute", "q_proj", "query_states", "reshape", "return", "scaling", "self", "self_attention_cache", "seq_length", "shape", "softmax", "torch", "training", "transpose", "tuple", "type_as", "update", "v_proj", "value_states", "values", "view"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2FeedForwardNetwork.__init__": ["ACT2FN", "Dropout", "Linear", "ModelConfig", "ModelFeedForwardNetwork", "Module", "__init__", "act", "activation_dropout", "activation_function", "class", "config", "def", "dropout", "fc1", "fc2", "ffn_dim", "hidden_size", "int", "nn", "self", "super"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2FeedForwardNetwork.forward": ["ModelFeedForwardNetwork", "Module", "Tensor", "act", "and", "class", "def", "dropout", "dtype", "fc1", "fc2", "forward", "hidden_states", "if", "int8", "isinstance", "nn", "return", "self", "to", "torch", "uint8", "weight"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2EncoderLayer.__init__": ["Dropout", "GradientCheckpointingLayer", "LayerNorm", "ModelAttention", "ModelConfig", "ModelEncoderLayer", "ModelFeedForwardNetwork", "None", "__init__", "activation_dropout", "attention_dropout", "attn_dropout", "class", "config", "def", "dropout", "else", "embed_dim", "encoder_attention_heads", "encoder_ffn_dim", "ffn", "ffn_dim", "ffn_dropout", "ffn_layer_norm", "hidden_size", "if", "is", "nn", "num_heads", "self", "self_attn", "self_attn_layer_norm", "super"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2EncoderLayer.forward": ["False", "GradientCheckpointingLayer", "ModelEncoderLayer", "Tensor", "attention_mask", "attn_dropout", "attn_weights", "bool", "class", "def", "ffn", "ffn_dropout", "ffn_layer_norm", "forward", "hidden_states", "if", "output_attentions", "outputs", "residual", "return", "self", "self_attn", "self_attn_layer_norm", "torch"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2DecoderLayer.__init__": ["ACT2FN", "Dropout", "GradientCheckpointingLayer", "LayerNorm", "ModelAttention", "ModelConfig", "ModelDecoderLayer", "ModelFeedForwardNetwork", "None", "True", "__init__", "activation_dropout", "activation_fn", "activation_function", "attention_dropout", "attn_dropout", "class", "config", "cross_attention", "cross_attention_layer_norm", "decoder_attention_heads", "decoder_ffn_dim", "def", "dropout", "else", "embed_dim", "ffn", "ffn_dim", "ffn_dropout", "ffn_layer_norm", "hidden_size", "if", "is", "is_decoder", "layer_idx", "nn", "num_heads", "self", "self_attn", "self_attn_layer_norm", "super"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2DecoderLayer.forward": ["Cache", "False", "GradientCheckpointingLayer", "ModelDecoderLayer", "None", "Tensor", "True", "attention_mask", "attn_dropout", "bool", "cache_position", "class", "cross_attention", "cross_attention_layer_norm", "cross_attn_weights", "def", "encoder_attention_mask", "encoder_hidden_states", "ffn", "ffn_dropout", "ffn_layer_norm", "forward", "hidden_states", "if", "is", "not", "output_attentions", "past_key_values", "residual", "return", "self", "self_attn", "self_attn_layer_norm", "self_attn_weights", "torch", "use_cache"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2TextToUnitDecoderLayer.__init__": ["ACT2FN", "Conv1d", "Dropout", "GradientCheckpointingLayer", "LayerNorm", "ModelAttention", "ModelConfig", "ModelTextToUnitDecoderLayer", "None", "True", "__init__", "activation_fn", "activation_function", "attention_dropout", "class", "config", "conv1", "conv2", "conv_dropout", "conv_layer_norm", "decoder_attention_heads", "decoder_ffn_dim", "def", "dropout", "else", "embed_dim", "hidden_size", "if", "is", "is_decoder", "kernel_size", "nn", "num_heads", "padding", "same", "self", "self_attn", "self_attn_layer_norm", "stride", "super"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2TextToUnitDecoderLayer.forward": ["False", "GradientCheckpointingLayer", "ModelTextToUnitDecoderLayer", "None", "Tensor", "activation_fn", "attention_mask", "bool", "class", "conv1", "conv2", "conv_dropout", "conv_layer_norm", "def", "forward", "hidden_states", "if", "is", "masked_fill", "not", "output_attentions", "padding_mask", "residual", "return", "self", "self_attn", "self_attn_layer_norm", "self_attn_weights", "torch", "transpose", "unsqueeze"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2PreTrainedModel._init_weights": ["Conv1d", "ConvTranspose1d", "Embedding", "False", "LayerNorm", "Linear", "ModelConformerFeatureProjection", "ModelConformerSelfAttention", "ModelPreTrainedModel", "ModelSinusoidalPositionalEmbedding", "ModelTextToUnitDecoder", "Module", "None", "PreTrainedModel", "_init_weights", "_is_hf_initialized", "a", "and", "b", "bias", "class", "config", "copy_", "def", "elif", "emb_weights", "embedding_dim", "get_embedding", "getattr", "groups", "hasattr", "if", "in_channels", "in_features", "init", "initializer_range", "is", "isinstance", "k", "kaiming_normal_", "kernel_size", "math", "mean", "module", "nn", "normal_", "not", "num_positions", "offset", "ones_", "padding_idx", "pos_bias_u", "pos_bias_v", "pos_emb_alpha", "pos_emb_alpha_char", "projection", "self", "sqrt", "std", "uniform_", "weight", "weights", "xavier_uniform_", "zeros_"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2PreTrainedModel._compute_sub_sample_lengths_from_attention_mask": ["ModelPreTrainedModel", "PreTrainedModel", "_compute_sub_sample_lengths_from_attention_mask", "adaptor_kernel_size", "adaptor_stride", "attention_mask", "class", "config", "def", "floor", "int", "kernel_size", "pad", "return", "self", "seq_lens", "size", "stride", "sum"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2PreTrainedModel._indices_to_subwords": ["ModelPreTrainedModel", "PreTrainedModel", "ValueError", "_indices_to_subwords", "append", "batch_id", "batch_size", "class", "def", "for", "generation_config", "get", "hasattr", "i", "id_to_text", "if", "in", "input_ids", "item", "not", "raise", "range", "return", "self", "sequence_len", "shape", "str", "subword", "subwords", "subwords_batch"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2PreTrainedModel._count_character_length_in_subword": ["False", "ModelPreTrainedModel", "PreTrainedModel", "_", "_count_character_length_in_subword", "and", "batch_id", "batch_size", "break", "char_count_per_id", "char_len", "class", "def", "elif", "else", "enumerate", "for", "i", "if", "in", "input_ids", "is_next_start_with_space", "is_punc", "isalpha", "isnumeric", "len", "merge_space_with_prev_subword", "ne", "new_zeros", "not", "pad_token_id", "range", "return", "self", "shape", "size", "space", "subword", "subword_idx", "subword_indices", "subword_lens", "subwords", "subwords_batch", "sum", "unk_token_id", "zip"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2PreTrainedModel._get_char_input_ids": ["ModelPreTrainedModel", "PreTrainedModel", "ValueError", "_get_char_input_ids", "batch_id", "batch_size", "ch", "char_count_per_id", "char_ids", "char_seq_len", "char_seqs", "char_to_id", "class", "def", "else", "fill_", "for", "generation_config", "get", "hasattr", "if", "in", "input_ids", "int", "item", "len", "list", "max", "max_len", "ne", "new_zeros", "not", "pad_token_id", "raise", "range", "return", "self", "shape", "subword", "subword_idx", "subword_indices", "subword_lens", "subwords", "subwords_batch", "sum", "tensor", "to", "torch", "total", "unk_token_id", "zip"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2PreTrainedModel._hard_upsample": ["ModelPreTrainedModel", "PreTrainedModel", "True", "_hard_upsample", "and", "batch_first", "class", "def", "dim", "duration", "durations", "else", "for", "hidden_state", "hidden_states", "if", "in", "logger", "nn", "pad_sequence", "repeat_interleave", "return", "rnn", "self", "shape", "size", "torch", "training", "utils", "view", "warning_once", "zip"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2SpeechEncoder.__init__": ["LayerNorm", "ModelConfig", "ModelConformerAdapter", "ModelConformerEncoder", "ModelConformerFeatureProjection", "ModelConformerFeedForward", "ModelPreTrainedModel", "ModelSpeechEncoder", "None", "__init__", "act_fn", "adapter", "add_adapter", "class", "config", "def", "dropout", "else", "encoder", "feature_projection", "hidden_size", "if", "inner_layer_norm", "intermediate_ffn", "nn", "post_init", "relu", "self", "super"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2SpeechEncoder.forward": ["ModelPreTrainedModel", "ModelSpeechEncoder", "None", "Tensor", "ValueError", "Wav2Vec2BaseModelOutput", "adapter", "attention_mask", "attentions", "bool", "class", "config", "def", "else", "encoder", "encoder_outputs", "expanded_hidden_states", "feature_projection", "forward", "hidden_states", "if", "inner_layer_norm", "input_features", "intermediate_ffn", "is", "kwargs", "last_hidden_state", "not", "output_attentions", "output_hidden_states", "raise", "return", "return_dict", "self", "torch", "tuple", "use_return_dict"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2Encoder.__init__": ["Embedding", "False", "LayerNorm", "ModelConfig", "ModelEncoder", "ModelEncoderLayer", "ModelPreTrainedModel", "ModelScaledWordEmbedding", "ModelSinusoidalPositionalEmbedding", "ModuleList", "None", "_", "__init__", "append", "bool", "class", "config", "def", "dropout", "else", "embed_dim", "embed_positions", "embed_scale", "embed_tokens", "encoder_attention_heads", "encoder_ffn_dim", "encoder_layerdrop", "encoder_layers", "for", "gradient_checkpointing", "hidden_size", "if", "in", "is", "is_t2u_encoder", "layer_norm", "layerdrop", "layers", "math", "max_position_embeddings", "max_source_positions", "nn", "not", "pad_token_id", "padding_idx", "post_init", "r", "range", "scale_embedding", "self", "sqrt", "super", "vocab_size", "weight"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2Encoder.forward": ["BaseModelOutput", "False", "FloatTensor", "LongTensor", "ModelEncoder", "ModelPreTrainedModel", "None", "Pass", "Tensor", "True", "ValueError", "You", "_prepare_4d_attention_mask", "all_attentions", "and", "at", "attention_mask", "attentions", "bool", "both", "cannot", "class", "config", "def", "device", "dropout", "dropout_probability", "dtype", "either", "elif", "else", "embed_pos", "embed_positions", "embed_tokens", "encoder", "encoder_layer", "encoder_states", "enumerate", "for", "forward", "functional", "have", "hidden_states", "idx", "if", "in", "input", "input_ids", "input_shape", "inputs_embeds", "instead", "is", "is_t2u_encoder", "kwargs", "last_hidden_state", "layer_norm", "layer_outputs", "layerdrop", "layers", "model", "nn", "not", "of", "or", "output_attentions", "output_hidden_states", "p", "pass", "r", "raise", "rand", "return", "return_dict", "same", "self", "shape", "specify", "text_to_units", "the", "time", "to", "to_drop", "torch", "training", "tuple", "use_return_dict", "v", "view"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2Decoder.__init__": ["Embedding", "False", "LayerNorm", "ModelConfig", "ModelDecoder", "ModelDecoderLayer", "ModelPreTrainedModel", "ModelScaledWordEmbedding", "ModelSinusoidalPositionalEmbedding", "ModuleList", "None", "__init__", "append", "class", "config", "decoder_attention_heads", "decoder_ffn_dim", "decoder_layerdrop", "decoder_layers", "def", "dropout", "else", "embed_positions", "embed_scale", "embed_tokens", "embedding_dim", "for", "gradient_checkpointing", "hidden_size", "i", "if", "in", "is", "layer_idx", "layer_norm", "layerdrop", "layers", "math", "max_position_embeddings", "max_target_positions", "nn", "not", "num_embeddings", "pad_token_id", "padding_idx", "post_init", "r", "range", "scale_embedding", "self", "sqrt", "super", "vocab_size", "weight"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2Decoder.forward": ["BaseModelOutputWithPastAndCrossAttentions", "Cache", "DynamicCache", "EncoderDecoderCache", "False", "FloatTensor", "LongTensor", "ModelDecoder", "ModelPreTrainedModel", "None", "Setting", "Tensor", "True", "ValueError", "You", "_prepare_4d_attention_mask", "_prepare_4d_causal_attention_mask", "all_cross_attentions", "all_hidden_states", "all_self_attns", "and", "at", "attention_mask", "attentions", "bool", "both", "cache_position", "cannot", "checkpointing", "class", "config", "continue", "cross_attentions", "decoder_input_ids", "decoder_inputs_embeds", "decoder_layer", "def", "device", "dropout", "dropout_probability", "dtype", "either", "elif", "else", "embed_positions", "embed_tokens", "encoder_attention_mask", "encoder_hidden_states", "enumerate", "for", "forward", "functional", "get_seq_length", "gradient", "gradient_checkpointing", "have", "hidden_states", "idx", "if", "in", "incompatible", "input", "input_ids", "input_shape", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layer_norm", "layer_outputs", "layerdrop", "layers", "logger", "nn", "not", "or", "output_attentions", "output_hidden_states", "p", "past_key_values", "past_key_values_length", "positions", "raise", "rand", "return", "return_dict", "same", "self", "size", "specify", "tgt_len", "the", "time", "to", "torch", "training", "tuple", "use_cache", "use_return_dict", "v", "view", "warning_once", "with"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2TextToUnitDecoder.__init__": ["Embedding", "False", "LayerNorm", "ModelConfig", "ModelPreTrainedModel", "ModelSinusoidalPositionalEmbedding", "ModelTextToUnitDecoder", "ModelTextToUnitDecoderLayer", "ModelVariancePredictor", "ModuleList", "None", "Parameter", "_", "__init__", "append", "char_vocab_size", "class", "config", "decoder_attention_heads", "decoder_ffn_dim", "decoder_layerdrop", "decoder_layers", "def", "dropout", "duration_predictor", "else", "embed_char", "embed_char_positions", "embed_positions", "embed_scale", "embed_tokens", "embedding_dim", "for", "gradient_checkpointing", "hidden_size", "if", "in", "is", "layer_norm", "layerdrop", "layers", "math", "max_position_embeddings", "max_target_positions", "nn", "not", "num_embeddings", "ones", "pad_token_id", "padding_idx", "pos_emb_alpha", "pos_emb_alpha_char", "post_init", "r", "range", "scale_embedding", "self", "sqrt", "super", "torch", "variance_pred_dropout", "variance_predictor_embed_dim", "variance_predictor_hidden_dim", "variance_predictor_kernel_size", "vocab_size", "weight"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2TextToUnitDecoder.forward": ["FloatTensor", "LongTensor", "ModelPreTrainedModel", "ModelTextToUnitDecoder", "ModelTextToUnitDecoderOutput", "None", "_compute_new_attention_mask", "_hard_upsample", "_prepare_4d_attention_mask", "all_hidden_states", "all_self_attns", "attention_mask", "attentions", "bool", "char_count_per_id", "char_hidden_states", "char_input_ids", "char_padding_mask", "char_positions", "clamp", "class", "config", "continue", "decoder_layer", "def", "dropout", "dropout_probability", "dtype", "dur_out", "duration_predictor", "else", "embed_char", "embed_char_positions", "embed_positions", "embed_scale", "encoder_hidden_states", "enumerate", "expm1", "for", "forward", "functional", "hidden_states", "idx", "if", "in", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layer_norm", "layer_outputs", "layerdrop", "layers", "log_dur_pred", "long", "masked_fill", "min", "nn", "not", "output_attentions", "output_hidden_states", "p", "padding_mask", "pos_emb_alpha", "pos_emb_alpha_char", "positions", "r", "rand", "return", "return_dict", "round", "self", "sum", "torch", "training", "tuple", "use_return_dict", "v"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2TextToUnitModel.__init__": ["Embedding", "ModelConfig", "ModelEncoder", "ModelPreTrainedModel", "ModelTextToUnitDecoder", "ModelTextToUnitModel", "None", "True", "__init__", "class", "config", "decoder", "def", "embed_tokens_decoder", "encoder", "is_t2u_encoder", "nn", "post_init", "r", "self", "super"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2TextToUnitModel.forward": ["BaseModelOutput", "FloatTensor", "LongTensor", "ModelPreTrainedModel", "ModelTextToUnitModel", "ModelTextToUnitOutput", "None", "Seq2SeqModelOutput", "Tensor", "and", "attention_mask", "attentions", "bool", "char_count_per_id", "char_input_ids", "class", "config", "decoder", "decoder_attentions", "decoder_hidden_states", "decoder_outputs", "def", "elif", "else", "encoder", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "isinstance", "kwargs", "last_hidden_state", "len", "not", "output_attentions", "output_hidden_states", "padding_mask", "return", "return_dict", "self", "torch", "tuple", "use_return_dict"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2TextToUnitForConditionalGeneration.__init__": ["Embedding", "False", "GenerationMixin", "Linear", "ModelConfig", "ModelPreTrainedModel", "ModelTextToUnitForConditionalGeneration", "ModelTextToUnitModel", "None", "__init__", "__setattr__", "bias", "class", "config", "copy", "deepcopy", "def", "embed_tokens_decoder", "for", "hidden_size", "if", "in", "items", "lm_head", "model", "nn", "param", "post_init", "r", "self", "startswith", "super", "t2u_", "t2u_vocab_size", "to_dict", "val"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2TextToUnitForConditionalGeneration.get_encoder": ["GenerationMixin", "ModelPreTrainedModel", "ModelTextToUnitForConditionalGeneration", "class", "def", "encoder", "get_encoder", "model", "return", "self"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2TextToUnitForConditionalGeneration.get_decoder": ["GenerationMixin", "ModelPreTrainedModel", "ModelTextToUnitForConditionalGeneration", "class", "decoder", "def", "get_decoder", "model", "return", "self"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2TextToUnitForConditionalGeneration.get_input_embeddings": ["GenerationMixin", "ModelPreTrainedModel", "ModelTextToUnitForConditionalGeneration", "class", "decoder", "def", "embed_tokens", "get_input_embeddings", "model", "return", "self"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2TextToUnitForConditionalGeneration.set_input_embeddings": ["GenerationMixin", "ModelPreTrainedModel", "ModelTextToUnitForConditionalGeneration", "class", "decoder", "def", "embed_tokens", "model", "self", "set_input_embeddings", "value"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2TextToUnitForConditionalGeneration.forward": ["CrossEntropyLoss", "FloatTensor", "GenerationMixin", "LongTensor", "ModelPreTrainedModel", "ModelTextToUnitForConditionalGeneration", "ModelTextToUnitOutput", "None", "Seq2SeqLMOutput", "Tensor", "attention_mask", "bool", "char_count_per_id", "char_input_ids", "class", "config", "decoder_attentions", "decoder_hidden_states", "def", "device", "else", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "forward", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "last_hidden_state", "lm_head", "lm_logits", "loss", "loss_fct", "masked_lm_loss", "model", "not", "output", "output_attentions", "output_hidden_states", "outputs", "padding_mask", "r", "return", "return_dict", "self", "to", "torch", "tuple", "use_return_dict", "view", "vocab_size"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:HifiGanResidualBlock.__init__": ["Conv1d", "ModelGanResidualBlock", "Module", "ModuleList", "_", "__init__", "channels", "class", "convs1", "convs2", "def", "dilation", "for", "get_padding", "i", "in", "kernel_size", "leaky_relu_slope", "len", "nn", "padding", "range", "self", "stride", "super"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:HifiGanResidualBlock.get_padding": ["ModelGanResidualBlock", "Module", "class", "def", "dilation", "get_padding", "kernel_size", "nn", "return", "self"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:HifiGanResidualBlock.apply_weight_norm": ["ModelGanResidualBlock", "Module", "apply_weight_norm", "class", "convs1", "convs2", "def", "for", "hasattr", "if", "in", "layer", "nn", "parametrizations", "self", "utils", "weight_norm"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:HifiGanResidualBlock.remove_weight_norm": ["ModelGanResidualBlock", "Module", "class", "convs1", "convs2", "def", "for", "in", "layer", "nn", "remove_weight_norm", "self", "utils"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:HifiGanResidualBlock.forward": ["ModelGanResidualBlock", "Module", "class", "conv1", "conv2", "convs1", "convs2", "def", "for", "forward", "functional", "hidden_states", "in", "leaky_relu", "leaky_relu_slope", "nn", "residual", "return", "self", "zip"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2VariancePredictor.__init__": ["Conv1d", "Dropout", "LayerNorm", "Linear", "ModelVariancePredictor", "Module", "ReLU", "__init__", "activation_function", "class", "conv1", "conv2", "def", "dropout_module", "embed_dim", "hidden_dim", "kernel_size", "ln1", "ln2", "nn", "p", "padding", "proj", "same", "self", "super", "var_pred_dropout"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2VariancePredictor.forward": ["ModelVariancePredictor", "Module", "None", "Tensor", "activation_function", "bool", "class", "conv1", "conv2", "def", "dim", "dropout_module", "forward", "hidden_states", "if", "is", "ln1", "ln2", "masked_fill", "nn", "not", "padding_mask", "proj", "return", "self", "squeeze", "transpose", "unsqueeze"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2HifiGan.__init__": ["Conv1d", "ConvTranspose1d", "HifiGanResidualBlock", "ModelConfig", "ModelHifiGan", "Module", "ModuleList", "__init__", "append", "channels", "class", "config", "conv_post", "conv_pre", "def", "dilation", "enumerate", "for", "i", "in", "kernel_size", "lang_embed_dim", "leaky_relu_slope", "len", "model_in_dim", "nn", "num_kernels", "num_upsamples", "padding", "range", "resblock_dilation_sizes", "resblock_kernel_sizes", "resblocks", "self", "spkr_embed_dim", "stride", "super", "unit_embed_dim", "upsample_initial_channel", "upsample_kernel_sizes", "upsample_rate", "upsample_rates", "upsampler", "zip"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2HifiGan.forward": ["FloatTensor", "ModelHifiGan", "Module", "class", "conv_post", "conv_pre", "def", "for", "forward", "functional", "hidden_states", "i", "in", "input_embeds", "j", "leaky_relu", "leaky_relu_slope", "nn", "num_kernels", "num_upsamples", "r", "range", "res_state", "resblocks", "return", "self", "squeeze", "tanh", "torch", "upsampler", "waveform"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2CodeHifiGan.__init__": ["Embedding", "ModelCodeHifiGan", "ModelHifiGan", "ModelVariancePredictor", "PreTrainedModel", "__init__", "class", "config", "def", "dur_predictor", "embed_dim", "hifi_gan", "kernel_size", "lang_embed_dim", "language_embedding", "nn", "pad_token_id", "post_init", "self", "speaker_embedding", "spkr_embed_dim", "super", "t2u_pad_token_id", "unit_embed_dim", "unit_embedding", "unit_hifi_gan_vocab_size", "var_pred_dropout", "variance_predictor_kernel_size", "vocoder_num_langs", "vocoder_num_spkrs"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2CodeHifiGan._get_dur_output_lengths": ["ModelCodeHifiGan", "PreTrainedModel", "_get_dur_output_lengths", "clamp", "class", "cumsum", "cumulative_dur_out", "def", "dim", "dur_out", "gather", "index", "input_ids", "pad_token_id", "return", "self", "shape", "squeeze", "sum", "torch", "unit_lengths", "unsqueeze"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2CodeHifiGan._get_output_hifigan_lengths": ["LongTensor", "ModelCodeHifiGan", "PreTrainedModel", "_conv_out_length", "_get_output_hifigan_lengths", "_transpose_conv_out_length", "class", "config", "def", "dil", "dilation", "div", "enumerate", "floor", "for", "i", "in", "input_length", "input_lengths", "int", "kernel_size", "len", "pad", "range", "resblock_dilation_sizes", "resblock_kernel_sizes", "return", "rounding_mode", "self", "stride", "torch", "upsample_kernel_sizes", "upsample_rate", "upsample_rates", "zip"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2CodeHifiGan.forward": ["LongTensor", "ModelCodeHifiGan", "PreTrainedModel", "Tensor", "True", "_get_dur_output_lengths", "_get_output_hifigan_lengths", "and", "batch_first", "cat", "clamp", "class", "def", "dim", "dur_out", "dur_predictor", "duration", "else", "expm1", "for", "forward", "hidden_state", "hidden_states", "hifi_gan", "if", "in", "input_ids", "kwargs", "lang", "lang_id", "language_embedding", "lengths", "log_dur_pred", "logger", "long", "min", "nn", "pad_sequence", "repeat", "repeat_interleave", "return", "rnn", "round", "self", "shape", "size", "speaker_embedding", "speaker_id", "spkr", "torch", "training", "transpose", "tuple", "unit_embedding", "unit_lengths", "utils", "view", "warning", "zip"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2CodeHifiGan.apply_weight_norm": ["ModelCodeHifiGan", "PreTrainedModel", "apply_weight_norm", "class", "conv_post", "conv_pre", "def", "for", "hasattr", "hifi_gan", "if", "in", "layer", "nn", "parametrizations", "resblocks", "self", "upsampler", "utils", "weight_norm"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2CodeHifiGan.remove_weight_norm": ["ModelCodeHifiGan", "PreTrainedModel", "class", "conv_post", "conv_pre", "def", "for", "hifi_gan", "in", "layer", "nn", "remove_weight_norm", "resblocks", "self", "upsampler", "utils"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2ForTextToText.__init__": ["Embedding", "False", "GenerationMixin", "Linear", "ModelConfig", "ModelDecoder", "ModelEncoder", "ModelForTextToText", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "nn", "pad_token_id", "post_init", "self", "shared", "super", "text_decoder", "text_encoder", "vocab_size"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2ForTextToText.get_encoder": ["GenerationMixin", "ModelForTextToText", "ModelPreTrainedModel", "class", "def", "get_encoder", "return", "self", "text_encoder"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2ForTextToText.get_decoder": ["GenerationMixin", "ModelForTextToText", "ModelPreTrainedModel", "class", "def", "get_decoder", "return", "self", "text_decoder"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2ForTextToText.get_input_embeddings": ["GenerationMixin", "ModelForTextToText", "ModelPreTrainedModel", "class", "def", "embed_tokens", "get_input_embeddings", "return", "self", "text_decoder"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2ForTextToText.set_input_embeddings": ["GenerationMixin", "ModelForTextToText", "ModelPreTrainedModel", "class", "def", "embed_tokens", "self", "set_input_embeddings", "shared", "text_decoder", "text_encoder", "value"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2ForTextToText.forward": ["BaseModelOutput", "Cache", "CrossEntropyLoss", "False", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForTextToText", "ModelPreTrainedModel", "None", "Seq2SeqLMOutput", "Tensor", "The", "and", "argument", "attention_mask", "attentions", "bool", "changed", "class", "config", "cross_attentions", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_input_ids", "decoder_inputs_embeds", "decoder_outputs", "decoder_start_token_id", "def", "device", "elif", "else", "encoder_attention_mask", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "len", "lm_head", "lm_logits", "logger", "logits", "loss", "loss_fct", "masked_lm_loss", "not", "output", "output_attentions", "output_hidden_states", "outputs", "pad_token_id", "past_key_values", "provided", "r", "return", "return_dict", "self", "shift_tokens_right", "since", "text_decoder", "text_encoder", "to", "torch", "tuple", "use_cache", "use_return_dict", "view", "vocab_size", "warning"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2ForTextToText.generate": ["False", "GenerationMixin", "ModelForTextToText", "ModelPreTrainedModel", "None", "ValueError", "__", "batch_size", "class", "decoder_input_ids", "def", "device", "else", "f", "generate", "generation_config", "get", "hasattr", "if", "in", "input_ids", "inputs_embeds", "is", "kwargs", "len", "logger", "logits_processor", "not", "pop", "prefix_allowed_tokens_fn", "raise", "replace", "return", "self", "stopping_criteria", "super", "synced_gpus", "tensor", "text_decoder_input_ids", "text_decoder_lang_to_code_id", "text_tgt_lang_id", "tgt_lang", "torch", "warning"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2ForSpeechToText.__init__": ["Embedding", "False", "GenerationMixin", "Linear", "ModelConfig", "ModelDecoder", "ModelForSpeechToText", "ModelPreTrainedModel", "ModelSpeechEncoder", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "nn", "pad_token_id", "post_init", "self", "shared", "speech_encoder", "super", "text_decoder", "vocab_size"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2ForSpeechToText.get_encoder": ["GenerationMixin", "ModelForSpeechToText", "ModelPreTrainedModel", "class", "def", "get_encoder", "return", "self", "speech_encoder"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2ForSpeechToText.get_decoder": ["GenerationMixin", "ModelForSpeechToText", "ModelPreTrainedModel", "class", "def", "get_decoder", "return", "self", "text_decoder"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2ForSpeechToText.get_input_embeddings": ["GenerationMixin", "ModelForSpeechToText", "ModelPreTrainedModel", "class", "def", "embed_tokens", "get_input_embeddings", "return", "self", "text_decoder"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2ForSpeechToText.set_input_embeddings": ["GenerationMixin", "ModelForSpeechToText", "ModelPreTrainedModel", "class", "def", "embed_tokens", "self", "set_input_embeddings", "text_decoder", "value"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2ForSpeechToText.forward": ["BaseModelOutput", "Cache", "CrossEntropyLoss", "False", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForSpeechToText", "ModelPreTrainedModel", "None", "Seq2SeqLMOutput", "Tensor", "The", "_compute_new_attention_mask", "_compute_sub_sample_lengths_from_attention_mask", "and", "argument", "attention_mask", "attentions", "bool", "changed", "class", "config", "cross_attentions", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_input_ids", "decoder_inputs_embeds", "decoder_outputs", "decoder_start_token_id", "def", "device", "elif", "else", "encoder_attention_mask", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "forward", "hidden_states", "if", "input_features", "input_ids", "inputs_embeds", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "len", "lm_head", "lm_logits", "logger", "logits", "loss", "loss_fct", "masked_lm_loss", "not", "output", "output_attentions", "output_hidden_states", "outputs", "pad_token_id", "past_key_values", "provided", "r", "return", "return_dict", "self", "seq_lens", "shift_tokens_right", "since", "speech_encoder", "sub_sampled_lengths", "text_decoder", "to", "torch", "tuple", "use_cache", "use_return_dict", "view", "vocab_size", "warning"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2ForSpeechToText.generate": ["False", "GenerationMixin", "ModelForSpeechToText", "ModelPreTrainedModel", "None", "ValueError", "__", "batch_size", "class", "decoder_input_ids", "def", "device", "else", "encoder_outputs", "f", "generate", "generation_config", "get", "hasattr", "if", "in", "input_embeds", "input_features", "inputs", "is", "kwargs", "last_hidden_state", "len", "logger", "logits_processor", "not", "pop", "prefix_allowed_tokens_fn", "raise", "replace", "return", "self", "stopping_criteria", "super", "synced_gpus", "tensor", "text_decoder_input_ids", "text_decoder_lang_to_code_id", "text_tgt_lang_id", "tgt_lang", "torch", "warning"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2ForTextToSpeech.__init__": ["Embedding", "False", "GenerationMixin", "Linear", "ModelCodeHifiGan", "ModelConfig", "ModelDecoder", "ModelEncoder", "ModelForTextToSpeech", "ModelPreTrainedModel", "ModelTextToUnitForConditionalGeneration", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "nn", "pad_token_id", "post_init", "self", "shared", "super", "t2u_model", "text_decoder", "text_encoder", "vocab_size", "vocoder"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2ForTextToSpeech.get_encoder": ["GenerationMixin", "ModelForTextToSpeech", "ModelPreTrainedModel", "class", "def", "get_encoder", "return", "self", "text_encoder"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2ForTextToSpeech.get_decoder": ["GenerationMixin", "ModelForTextToSpeech", "ModelPreTrainedModel", "class", "def", "get_decoder", "return", "self", "text_decoder"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2ForTextToSpeech.get_input_embeddings": ["GenerationMixin", "ModelForTextToSpeech", "ModelPreTrainedModel", "class", "def", "embed_tokens", "get_input_embeddings", "return", "self", "text_decoder"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2ForTextToSpeech.set_input_embeddings": ["GenerationMixin", "ModelForTextToSpeech", "ModelPreTrainedModel", "class", "def", "embed_tokens", "self", "set_input_embeddings", "shared", "text_decoder", "text_encoder", "value"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2ForTextToSpeech.forward": ["BaseModelOutput", "Cache", "CrossEntropyLoss", "False", "FloatTensor", "GenerationMixin", "If", "It", "LongTensor", "ModelForTextToSpeech", "ModelForTextToText", "ModelPreTrainedModel", "ModelTextToUnitForConditionalGeneration", "None", "Seq2SeqLMOutput", "Tensor", "The", "This", "and", "argument", "as", "attention_mask", "attentions", "bool", "cache_position", "changed", "class", "config", "cross_attentions", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_input_ids", "decoder_inputs_embeds", "decoder_outputs", "decoder_start_token_id", "def", "device", "doesn", "elif", "else", "encoder_attention_mask", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "forward", "generate", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "len", "lm_head", "lm_logits", "logger", "logits", "loss", "loss_fct", "masked_lm_loss", "method", "model", "not", "output", "output_attentions", "output_hidden_states", "outputs", "pad_token_id", "past_key_values", "provided", "r", "return", "return_dict", "same", "self", "shift_tokens_right", "since", "speech", "t", "text", "text_decoder", "text_encoder", "the", "to", "torch", "tuple", "unit", "use", "use_cache", "use_return_dict", "view", "vocab_size", "want", "warning", "you"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2ForTextToSpeech.generate": ["False", "GenerationMixin", "ModelForTextToSpeech", "ModelGenerationOutput", "ModelPreTrainedModel", "None", "Tensor", "True", "ValueError", "You", "__", "_compute_new_attention_mask", "_count_character_length_in_subword", "_get_char_input_ids", "_indices_to_subwords", "a", "argmax", "attention_mask", "batch_size", "bool", "cat", "char_count_per_id", "char_input_ids", "class", "clone", "config", "decoder_input_ids", "def", "detach", "device", "dim", "do_sample", "elif", "else", "encoder_attention_mask", "encoder_hidden_states", "eos_token_id", "f", "for", "format_speech_generation_kwargs", "functional", "generate", "generation_config", "get", "getattr", "if", "in", "input_ids", "inputs_embeds", "int", "is", "key", "kwargs", "kwargs_speech", "kwargs_text", "lang_code_to_id", "lang_id", "last_hidden_state", "len", "masked_fill", "multinomial", "must", "new_zeros", "nn", "not", "num_return_sequences", "num_samples", "or", "output_hidden_states", "output_scores", "output_unit_ids", "pad_token_id", "pad_zero", "padding_mask", "probs", "raise", "repeat_interleave", "replace", "replace_mask", "reshape", "return", "return_dict_in_generate", "return_intermediate_token_ids", "self", "seq_lens", "sequences", "shape", "softmax", "speaker_id", "specify", "speech", "str", "sum", "super", "t2u_char_count_per_id", "t2u_char_input_ids", "t2u_eos_token_id", "t2u_input_embeds", "t2u_input_ids", "t2u_lang_code_to_id", "t2u_logits", "t2u_model", "t2u_model_attention_mask", "t2u_output", "t2u_pad_token_id", "t2u_subwords", "temperature", "tensor", "text_decoder", "text_decoder_input_ids", "text_decoder_lang_to_code_id", "text_generation_output", "text_tgt_lang_id", "tgt_lang", "to", "torch", "translated", "unit_ids", "unit_sequences", "view", "vocoder", "vocoder_lang_code_to_id", "vocoder_offset", "vocoder_tgt_lang_id", "waveform", "waveform_lengths", "where"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2ForSpeechToSpeech.__init__": ["Embedding", "False", "GenerationMixin", "Linear", "ModelCodeHifiGan", "ModelDecoder", "ModelForSpeechToSpeech", "ModelPreTrainedModel", "ModelSpeechEncoder", "ModelTextToUnitForConditionalGeneration", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "nn", "pad_token_id", "post_init", "self", "shared", "speech_encoder", "super", "t2u_model", "text_decoder", "vocab_size", "vocoder"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2ForSpeechToSpeech.get_encoder": ["GenerationMixin", "ModelForSpeechToSpeech", "ModelPreTrainedModel", "class", "def", "get_encoder", "return", "self", "speech_encoder"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2ForSpeechToSpeech.get_decoder": ["GenerationMixin", "ModelForSpeechToSpeech", "ModelPreTrainedModel", "class", "def", "get_decoder", "return", "self", "text_decoder"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2ForSpeechToSpeech.get_input_embeddings": ["GenerationMixin", "ModelForSpeechToSpeech", "ModelPreTrainedModel", "class", "def", "embed_tokens", "get_input_embeddings", "return", "self", "text_decoder"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2ForSpeechToSpeech.set_input_embeddings": ["GenerationMixin", "ModelForSpeechToSpeech", "ModelPreTrainedModel", "class", "def", "embed_tokens", "self", "set_input_embeddings", "text_decoder", "value"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2ForSpeechToSpeech.forward": ["BaseModelOutput", "Cache", "CrossEntropyLoss", "False", "FloatTensor", "GenerationMixin", "If", "It", "LongTensor", "ModelForSpeechToSpeech", "ModelForSpeechToText", "ModelPreTrainedModel", "None", "Seq2SeqLMOutput", "Tensor", "The", "This", "_compute_new_attention_mask", "_compute_sub_sample_lengths_from_attention_mask", "and", "argument", "as", "attention_mask", "attentions", "bool", "changed", "class", "config", "cross_attentions", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_input_ids", "decoder_inputs_embeds", "decoder_outputs", "decoder_start_token_id", "def", "device", "doesn", "elif", "else", "encoder_attention_mask", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "forward", "generate", "hidden_states", "if", "input_features", "input_ids", "inputs_embeds", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "len", "lm_head", "lm_logits", "logger", "logits", "loss", "loss_fct", "masked_lm_loss", "method", "not", "output", "output_attentions", "output_hidden_states", "outputs", "pad_token_id", "past_key_values", "provided", "r", "return", "return_dict", "same", "self", "seq_lens", "shift_tokens_right", "since", "speech", "speech_encoder", "sub_sampled_lengths", "t", "t2u_model", "text_decoder", "the", "to", "torch", "tuple", "use", "use_cache", "use_return_dict", "view", "vocab_size", "want", "warning", "you"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2ForSpeechToSpeech.generate": ["False", "GenerationMixin", "ModelForSpeechToSpeech", "ModelGenerationOutput", "ModelPreTrainedModel", "None", "Tensor", "True", "ValueError", "You", "__", "_compute_new_attention_mask", "_compute_sub_sample_lengths_from_attention_mask", "_count_character_length_in_subword", "_get_char_input_ids", "_indices_to_subwords", "a", "argmax", "attention_mask", "batch_size", "bool", "cat", "char_count_per_id", "char_input_ids", "class", "clone", "config", "decoder_input_ids", "def", "detach", "device", "dim", "do_sample", "elif", "else", "encoder_attention_mask", "encoder_hidden_states", "eos_token_id", "f", "for", "format_speech_generation_kwargs", "functional", "generate", "generation_config", "get", "getattr", "hidden_states", "if", "in", "input_features", "input_ids", "inputs_embeds", "int", "is", "key", "kwargs", "kwargs_speech", "kwargs_text", "lang_code_to_id", "lang_id", "last_hidden_state", "len", "masked_fill", "multinomial", "must", "new_zeros", "nn", "not", "num_return_sequences", "num_samples", "or", "output_hidden_states", "output_scores", "output_unit_ids", "pad_token_id", "pad_zero", "padding_mask", "probs", "raise", "repeat_interleave", "replace", "replace_mask", "reshape", "return", "return_dict_in_generate", "return_intermediate_token_ids", "self", "seq_lens", "sequences", "shape", "softmax", "speaker_id", "specify", "speech", "speech_encoder", "str", "sub_sampled_lengths", "sum", "super", "t2u_char_count_per_id", "t2u_char_input_ids", "t2u_eos_token_id", "t2u_input_embeds", "t2u_input_ids", "t2u_lang_code_to_id", "t2u_logits", "t2u_model", "t2u_model_attention_mask", "t2u_output", "t2u_pad_token_id", "t2u_subwords", "temperature", "tensor", "text_decoder", "text_decoder_input_ids", "text_decoder_lang_to_code_id", "text_generation_output", "text_tgt_lang_id", "tgt_lang", "to", "torch", "translated", "unit_ids", "unit_sequences", "view", "vocoder", "vocoder_lang_code_to_id", "vocoder_offset", "vocoder_tgt_lang_id", "waveform", "waveform_lengths", "where"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2Model.__init__": ["Embedding", "False", "GenerationMixin", "Linear", "ModelCodeHifiGan", "ModelDecoder", "ModelEncoder", "ModelModel", "ModelPreTrainedModel", "ModelSpeechEncoder", "ModelTextToUnitForConditionalGeneration", "__init__", "bias", "class", "config", "current_modality", "def", "hidden_size", "if", "input_features", "lm_head", "main_input_name", "nn", "pad_token_id", "post_init", "r", "self", "shared", "speech", "speech_encoder", "super", "t2u_model", "text", "text_decoder", "text_encoder", "vocab_size", "vocoder"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2Model.set_modality": ["GenerationMixin", "It", "ModelModel", "ModelPreTrainedModel", "ValueError", "a", "be", "class", "current_modality", "def", "elif", "else", "f", "if", "input_features", "input_ids", "is", "main_input_name", "modality", "must", "not", "or", "raise", "self", "set_modality", "speech", "text", "valid"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2Model.get_encoder": ["GenerationMixin", "ModelModel", "ModelPreTrainedModel", "class", "current_modality", "def", "else", "get_encoder", "if", "return", "self", "speech_encoder", "text", "text_encoder"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2Model.get_input_embeddings": ["GenerationMixin", "ModelModel", "ModelPreTrainedModel", "class", "def", "embed_tokens", "get_input_embeddings", "return", "self", "text_decoder"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2Model.set_input_embeddings": ["GenerationMixin", "ModelModel", "ModelPreTrainedModel", "class", "def", "embed_tokens", "self", "set_input_embeddings", "shared", "text_decoder", "text_encoder", "value"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2Model.forward": ["BaseModelOutput", "Cache", "CrossEntropyLoss", "False", "FloatTensor", "GenerationMixin", "If", "LongTensor", "Make", "ModelForSpeechToText", "ModelForTextToText", "ModelModel", "ModelPreTrainedModel", "None", "Seq2SeqLMOutput", "Tensor", "The", "This", "ValueError", "_compute_new_attention_mask", "_compute_sub_sample_lengths_from_attention_mask", "all", "and", "are", "argument", "as", "at", "attention_mask", "attentions", "be", "been", "bool", "but", "calls", "changed", "class", "config", "cross_attentions", "current_modality", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_input_ids", "decoder_inputs_embeds", "decoder_outputs", "decoder_start_token_id", "def", "depending", "device", "elif", "else", "empty", "encoder_attention_mask", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "exclusive", "forward", "generate", "given", "has", "hidden_states", "if", "ignored", "in", "input", "input_features", "input_ids", "inputs_embeds", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "least", "len", "lm_head", "lm_logits", "logger", "logits", "loss", "loss_fct", "masked_lm_loss", "method", "modality", "mutually", "not", "of", "on", "one", "or", "output", "output_attentions", "output_hidden_states", "outputs", "pad_token_id", "past_key_values", "priority", "provided", "r", "raise", "return", "return_dict", "same", "self", "seq_lens", "set_modality", "shift_tokens_right", "since", "speech", "speech_encoder", "sub_sampled_lengths", "sure", "text", "text_decoder", "text_encoder", "that", "the", "them", "through", "to", "torch", "tuple", "use", "use_cache", "use_return_dict", "used", "view", "vocab_size", "want", "warning", "will", "you"], "seamless_m4t_v2/modeling_seamless_m4t_v2.py:SeamlessM4Tv2Model.generate": ["False", "GenerationMixin", "Make", "ModelGenerationOutput", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "True", "ValueError", "You", "__", "_compute_new_attention_mask", "_compute_sub_sample_lengths_from_attention_mask", "_count_character_length_in_subword", "_get_char_input_ids", "_indices_to_subwords", "a", "all", "and", "are", "argmax", "at", "attention_mask", "batch_size", "be", "bool", "both", "cat", "char_count_per_id", "char_input_ids", "class", "clone", "config", "current_modality", "decoder_input_ids", "def", "detach", "device", "dim", "do_sample", "elif", "else", "empty", "encoder", "encoder_attention_mask", "encoder_hidden_states", "eos_token_id", "f", "for", "format_speech_generation_kwargs", "functional", "generate", "generate_speech", "generation_config", "get", "getattr", "hidden_states", "if", "in", "input_features", "input_ids", "inputs_embeds", "int", "is", "key", "keys_to_check", "kwargs", "kwargs_speech", "kwargs_text", "lang_code_to_id", "lang_id", "last_hidden_state", "least", "len", "logger", "masked_fill", "multinomial", "must", "new_zeros", "nn", "non", "not", "num_return_sequences", "num_samples", "of", "one", "or", "output_hidden_states", "output_scores", "output_unit_ids", "pad_token_id", "pad_zero", "padding_mask", "priority", "probs", "raise", "repeat_interleave", "replace", "replace_mask", "reshape", "return", "return_dict_in_generate", "return_intermediate_token_ids", "self", "seq_lens", "sequences", "set_modality", "shape", "softmax", "speaker_id", "specify", "speech", "speech_encoder", "str", "sub_sampled_lengths", "sum", "super", "sure", "t2u_char_count_per_id", "t2u_char_input_ids", "t2u_eos_token_id", "t2u_input_embeds", "t2u_input_ids", "t2u_lang_code_to_id", "t2u_logits", "t2u_model", "t2u_model_attention_mask", "t2u_output", "t2u_pad_token_id", "t2u_subwords", "temperature", "tensor", "text", "text_decoder", "text_decoder_input_ids", "text_decoder_lang_to_code_id", "text_generation_output", "text_tgt_lang_id", "tgt_lang", "the", "them", "through", "to", "torch", "translated", "unit_ids", "unit_sequences", "use", "used", "view", "vocoder", "vocoder_lang_code_to_id", "vocoder_offset", "vocoder_tgt_lang_id", "want", "warning", "waveform", "waveform_lengths", "where", "will", "you"], "seed_oss/modeling_seed_oss.py:SeedOssRMSNorm.__init__": ["ModelRMSNorm", "Module", "Parameter", "__init__", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "seed_oss/modeling_seed_oss.py:SeedOssRMSNorm.forward": ["ModelRMSNorm", "Module", "True", "class", "def", "dtype", "float32", "forward", "hidden_states", "input_dtype", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "seed_oss/modeling_seed_oss.py:SeedOssRMSNorm.extra_repr": ["ModelRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "variance_epsilon", "weight"], "seed_oss/modeling_seed_oss.py:SeedOssMLP.__init__": ["ACT2FN", "Linear", "ModelMLP", "Module", "__init__", "act_fn", "bias", "class", "config", "def", "down_proj", "gate_proj", "hidden_act", "hidden_size", "intermediate_size", "mlp_bias", "nn", "residual_dropout", "self", "super", "up_proj"], "seed_oss/modeling_seed_oss.py:SeedOssMLP.forward": ["ModelMLP", "Module", "act_fn", "class", "def", "down_proj", "dropout", "forward", "functional", "gate_proj", "nn", "p", "residual_dropout", "return", "self", "training", "up_proj", "x"], "seed_oss/modeling_seed_oss.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "seed_oss/modeling_seed_oss.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cos", "def", "k", "k_embed", "q", "q_embed", "return", "rotate_half", "sin", "unsqueeze", "unsqueeze_dim"], "seed_oss/modeling_seed_oss.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "seed_oss/modeling_seed_oss.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value", "value_states"], "seed_oss/modeling_seed_oss.py:SeedOssAttention.__init__": ["Linear", "ModelAttention", "ModelConfig", "Module", "True", "__init__", "attention_bias", "attention_dropout", "attention_out_bias", "bias", "class", "config", "def", "head_dim", "hidden_size", "int", "is_causal", "k_proj", "layer_idx", "nn", "num_attention_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "q_proj", "residual_dropout", "scaling", "self", "super", "v_proj"], "seed_oss/modeling_seed_oss.py:SeedOssAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "LongTensor", "ModelAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_attn_implementation", "apply_rotary_pos_emb", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "cache_kwargs", "cache_position", "class", "config", "contiguous", "cos", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "functional", "head_dim", "hidden_shape", "hidden_states", "if", "input_shape", "is", "k_proj", "key_states", "kwargs", "layer_idx", "nn", "not", "o_proj", "p", "past_key_values", "position_embeddings", "q_proj", "query_states", "reshape", "residual_dropout", "return", "scaling", "self", "shape", "sin", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "view"], "seed_oss/modeling_seed_oss.py:SeedOssDecoderLayer.__init__": ["GradientCheckpointingLayer", "ModelAttention", "ModelConfig", "ModelDecoderLayer", "ModelMLP", "ModelRMSNorm", "__init__", "class", "config", "def", "eps", "hidden_size", "input_layernorm", "int", "layer_idx", "mlp", "post_attention_layernorm", "rms_norm_eps", "self", "self_attn", "super"], "seed_oss/modeling_seed_oss.py:SeedOssDecoderLayer.forward": ["Cache", "False", "GradientCheckpointingLayer", "LongTensor", "ModelDecoderLayer", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "attention_mask", "bool", "cache_position", "class", "def", "forward", "hidden_states", "input_layernorm", "kwargs", "mlp", "past_key_values", "position_embeddings", "position_ids", "post_attention_layernorm", "residual", "return", "self", "self_attn", "torch", "tuple", "use_cache"], "seed_oss/modeling_seed_oss.py:SeedOssRotaryEmbedding.__init__": ["Callable", "False", "ModelConfig", "ModelRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "seed_oss/modeling_seed_oss.py:SeedOssRotaryEmbedding.compute_default_rope_parameters": ["ModelConfig", "ModelRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "seed_oss/modeling_seed_oss.py:SeedOssRotaryEmbedding.forward": ["False", "ModelRotaryEmbedding", "Module", "None", "and", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "seed_oss/modeling_seed_oss.py:SeedOssModel.__init__": ["Embedding", "False", "ModelConfig", "ModelDecoderLayer", "ModelModel", "ModelPreTrainedModel", "ModelRMSNorm", "ModelRotaryEmbedding", "ModuleList", "__init__", "class", "config", "def", "embed_tokens", "eps", "for", "gradient_checkpointing", "hidden_size", "in", "layer_idx", "layers", "nn", "norm", "num_hidden_layers", "pad_token_id", "padding_idx", "post_init", "range", "rms_norm_eps", "rotary_emb", "self", "super", "vocab_size"], "seed_oss/modeling_seed_oss.py:SeedOssModel.forward": ["BaseModelOutputWithPast", "Cache", "DynamicCache", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "and", "arange", "attention_mask", "bool", "cache_position", "causal_mask", "class", "config", "create_causal_mask", "decoder_layer", "def", "device", "else", "embed_tokens", "exactly", "for", "forward", "get_seq_length", "hidden_states", "if", "in", "input_embeds", "input_ids", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layers", "must", "norm", "not", "num_hidden_layers", "of", "one", "or", "past_key_values", "past_seen_tokens", "position_embeddings", "position_ids", "raise", "return", "rotary_emb", "self", "shape", "specify", "torch", "unsqueeze", "use_cache"], "seed_oss/modeling_seed_oss.py:SeedOssForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "ModelForCausalLM", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "vocab_size"], "seed_oss/modeling_seed_oss.py:SeedOssForCausalLM.forward": ["BaseModelOutputWithPast", "Cache", "CausalLMOutputWithPast", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForCausalLM", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "outputs", "past_key_values", "position_ids", "r", "return", "self", "slice", "slice_indices", "torch", "use_cache", "vocab_size"], "segformer/modeling_segformer.py:drop_path": ["False", "Model_path", "Model_prob", "Tensor", "bool", "def", "device", "div", "dtype", "float", "floor_", "if", "input", "keep_prob", "ndim", "not", "or", "output", "rand", "random_tensor", "return", "shape", "torch", "training"], "segformer/modeling_segformer.py:SegformerDropPath.__init__": ["ModelDropPath", "Module", "None", "__init__", "class", "def", "drop_prob", "float", "nn", "self", "super"], "segformer/modeling_segformer.py:SegformerDropPath.forward": ["ModelDropPath", "Module", "Tensor", "class", "def", "drop_path", "drop_prob", "forward", "hidden_states", "nn", "return", "self", "torch", "training"], "segformer/modeling_segformer.py:SegformerDropPath.extra_repr": ["ModelDropPath", "Module", "class", "def", "drop_prob", "extra_repr", "f", "nn", "p", "return", "self", "str"], "segformer/modeling_segformer.py:SegformerOverlapPatchEmbeddings.__init__": ["Conv2d", "LayerNorm", "ModelOverlapPatchEmbeddings", "Module", "__init__", "class", "def", "hidden_size", "kernel_size", "layer_norm", "nn", "num_channels", "padding", "patch_size", "proj", "self", "stride", "super"], "segformer/modeling_segformer.py:SegformerOverlapPatchEmbeddings.forward": ["ModelOverlapPatchEmbeddings", "Module", "_", "class", "def", "embeddings", "flatten", "forward", "height", "layer_norm", "nn", "pixel_values", "proj", "return", "self", "shape", "transpose", "width"], "segformer/modeling_segformer.py:SegformerEfficientSelfAttention.__init__": ["Conv2d", "Dropout", "LayerNorm", "Linear", "ModelEfficientSelfAttention", "Module", "The", "ValueError", "__init__", "a", "all_head_size", "attention", "attention_head_size", "attention_probs_dropout_prob", "class", "config", "def", "dropout", "f", "heads", "hidden", "hidden_size", "if", "int", "is", "kernel_size", "key", "layer_norm", "multiple", "nn", "not", "num_attention_heads", "number", "of", "query", "raise", "self", "sequence_reduction_ratio", "size", "sr", "sr_ratio", "stride", "super", "the", "value"], "segformer/modeling_segformer.py:SegformerEfficientSelfAttention.forward": ["False", "ModelEfficientSelfAttention", "Module", "_", "all_head_size", "attention_head_size", "attention_probs", "attention_scores", "batch_size", "class", "context_layer", "contiguous", "def", "dim", "dropout", "else", "forward", "functional", "height", "hidden_states", "if", "key", "key_layer", "layer_norm", "math", "matmul", "new_context_layer_shape", "nn", "num_attention_heads", "num_channels", "output_attentions", "outputs", "permute", "query", "query_layer", "reshape", "return", "self", "seq_len", "seq_length", "shape", "size", "softmax", "sqrt", "sr", "sr_ratio", "torch", "transpose", "value", "value_layer", "view", "width"], "segformer/modeling_segformer.py:SegformerSelfOutput.__init__": ["Dropout", "Linear", "ModelSelfOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "hidden_dropout_prob", "hidden_size", "nn", "self", "super"], "segformer/modeling_segformer.py:SegformerSelfOutput.forward": ["ModelSelfOutput", "Module", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self"], "segformer/modeling_segformer.py:SegformerAttention.__init__": ["ModelAttention", "ModelEfficientSelfAttention", "ModelSelfOutput", "Module", "__init__", "class", "config", "def", "hidden_size", "nn", "num_attention_heads", "output", "self", "sequence_reduction_ratio", "super"], "segformer/modeling_segformer.py:SegformerAttention.forward": ["False", "ModelAttention", "Module", "attention_output", "class", "def", "forward", "height", "hidden_states", "nn", "output", "output_attentions", "outputs", "return", "self", "self_outputs", "width"], "segformer/modeling_segformer.py:SegformerDWConv.__init__": ["Conv2d", "ModelDWConv", "Module", "True", "__init__", "bias", "class", "def", "dim", "dwconv", "groups", "nn", "self", "super"], "segformer/modeling_segformer.py:SegformerDWConv.forward": ["ModelDWConv", "Module", "batch_size", "class", "def", "dwconv", "flatten", "forward", "height", "hidden_states", "nn", "num_channels", "return", "self", "seq_len", "shape", "transpose", "view", "width"], "segformer/modeling_segformer.py:SegformerMixFFN.__init__": ["ACT2FN", "Dropout", "Linear", "ModelDWConv", "ModelMixFFN", "Module", "None", "__init__", "class", "config", "def", "dense1", "dense2", "dropout", "dwconv", "else", "hidden_act", "hidden_dropout_prob", "hidden_features", "if", "in_features", "intermediate_act_fn", "isinstance", "nn", "or", "out_features", "self", "str", "super"], "segformer/modeling_segformer.py:SegformerMixFFN.forward": ["ModelMixFFN", "Module", "class", "def", "dense1", "dense2", "dropout", "dwconv", "forward", "height", "hidden_states", "intermediate_act_fn", "nn", "return", "self", "width"], "segformer/modeling_segformer.py:SegformerLayer.__init__": ["Identity", "LayerNorm", "ModelAttention", "ModelDropPath", "ModelLayer", "ModelMixFFN", "Module", "__init__", "attention", "class", "config", "def", "drop_path", "else", "hidden_features", "hidden_size", "if", "in_features", "int", "layer_norm_1", "layer_norm_2", "mlp", "mlp_hidden_size", "mlp_ratio", "nn", "num_attention_heads", "self", "sequence_reduction_ratio", "super"], "segformer/modeling_segformer.py:SegformerLayer.forward": ["False", "ModelLayer", "Module", "attention", "attention_output", "class", "def", "drop_path", "forward", "height", "hidden_states", "layer_norm_1", "layer_norm_2", "layer_output", "mlp", "mlp_output", "nn", "output_attentions", "outputs", "return", "self", "self_attention_outputs", "width"], "segformer/modeling_segformer.py:SegformerEncoder.__init__": ["LayerNorm", "ModelEncoder", "ModelLayer", "ModelOverlapPatchEmbeddings", "Module", "ModuleList", "__init__", "append", "block", "blocks", "class", "config", "cpu", "cur", "def", "depths", "device", "drop_path", "drop_path_decays", "drop_path_rate", "else", "embeddings", "for", "hidden_size", "hidden_sizes", "i", "if", "in", "item", "j", "layer_norm", "layers", "linspace", "mlp_ratio", "mlp_ratios", "nn", "num_attention_heads", "num_channels", "num_encoder_blocks", "patch_embeddings", "patch_size", "patch_sizes", "range", "self", "sequence_reduction_ratio", "sr_ratios", "stride", "strides", "sum", "super", "torch", "x"], "segformer/modeling_segformer.py:SegformerEncoder.forward": ["BaseModelOutput", "False", "FloatTensor", "ModelEncoder", "Module", "None", "True", "all_hidden_states", "all_self_attentions", "and", "attentions", "batch_size", "blk", "block", "block_layer", "bool", "class", "config", "contiguous", "def", "else", "embedding_layer", "enumerate", "for", "forward", "height", "hidden_states", "i", "idx", "if", "in", "is", "last_hidden_state", "layer_norm", "layer_outputs", "len", "nn", "norm_layer", "not", "or", "output_attentions", "output_hidden_states", "patch_embeddings", "permute", "pixel_values", "reshape", "reshape_last_stage", "return", "return_dict", "self", "shape", "torch", "tuple", "v", "width", "x", "zip"], "segformer/modeling_segformer.py:SegformerModel.__init__": ["ModelEncoder", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "config", "def", "encoder", "post_init", "self", "super"], "segformer/modeling_segformer.py:SegformerModel.forward": ["BaseModelOutput", "FloatTensor", "ModelModel", "ModelPreTrainedModel", "None", "attentions", "bool", "class", "config", "def", "else", "encoder", "encoder_outputs", "forward", "hidden_states", "if", "is", "kwargs", "last_hidden_state", "not", "output_attentions", "output_hidden_states", "pixel_values", "return", "return_dict", "self", "sequence_output", "torch", "tuple", "use_return_dict"], "segformer/modeling_segformer.py:SegformerForImageClassification.__init__": ["Linear", "Model", "ModelForImageClassification", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "classifier", "config", "def", "hidden_sizes", "nn", "num_labels", "post_init", "self", "super"], "segformer/modeling_segformer.py:SegformerForImageClassification.forward": ["FloatTensor", "LongTensor", "Model", "ModelForImageClassification", "ModelImageClassifierOutput", "ModelPreTrainedModel", "None", "attentions", "batch_size", "bool", "class", "classifier", "config", "def", "dim", "else", "forward", "hidden_sizes", "hidden_states", "if", "is", "kwargs", "labels", "logits", "loss", "loss_function", "mean", "not", "output", "output_attentions", "output_hidden_states", "outputs", "permute", "pixel_values", "r", "reshape", "reshape_last_stage", "return", "return_dict", "self", "sequence_output", "shape", "torch", "tuple", "use_return_dict"], "segformer/modeling_segformer.py:SegformerMLP.__init__": ["Linear", "ModelConfig", "ModelMLP", "Module", "__init__", "class", "config", "decoder_hidden_size", "def", "input_dim", "nn", "proj", "self", "super"], "segformer/modeling_segformer.py:SegformerMLP.forward": ["ModelMLP", "Module", "Tensor", "class", "def", "flatten", "forward", "hidden_states", "nn", "proj", "return", "self", "torch", "transpose"], "segformer/modeling_segformer.py:SegformerDecodeHead.__init__": ["BatchNorm2d", "Conv2d", "Dropout", "False", "ModelDecodeHead", "ModelMLP", "Module", "ModuleList", "ReLU", "__init__", "activation", "append", "batch_norm", "bias", "class", "classifier", "classifier_dropout_prob", "config", "decoder_hidden_size", "def", "dropout", "for", "hidden_sizes", "i", "in", "in_channels", "input_dim", "kernel_size", "linear_c", "linear_fuse", "mlp", "mlps", "nn", "num_encoder_blocks", "num_labels", "out_channels", "range", "self", "super"], "segformer/modeling_segformer.py:SegformerDecodeHead.forward": ["False", "FloatTensor", "ModelDecodeHead", "Module", "Tensor", "activation", "align_corners", "all_hidden_states", "and", "batch_norm", "batch_size", "bilinear", "cat", "class", "classifier", "config", "contiguous", "def", "dim", "dropout", "encoder_hidden_state", "encoder_hidden_states", "for", "forward", "functional", "height", "hidden_states", "if", "in", "int", "interpolate", "is", "kwargs", "linear_c", "linear_fuse", "logits", "math", "mlp", "mode", "ndim", "nn", "permute", "reshape", "reshape_last_stage", "return", "self", "shape", "size", "sqrt", "torch", "width", "zip"], "segformer/modeling_segformer.py:SegformerForSemanticSegmentation.__init__": ["Model", "ModelDecodeHead", "ModelForSemanticSegmentation", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "config", "decode_head", "def", "post_init", "self", "super"], "segformer/modeling_segformer.py:SegformerForSemanticSegmentation.forward": ["BCEWithLogitsLoss", "CrossEntropyLoss", "False", "FloatTensor", "LongTensor", "Model", "ModelForSemanticSegmentation", "ModelPreTrainedModel", "None", "Number", "SemanticSegmenterOutput", "True", "ValueError", "align_corners", "and", "attentions", "be", "bilinear", "bool", "class", "config", "decode_head", "def", "elif", "else", "encoder_hidden_states", "f", "float", "forward", "functional", "hidden_states", "if", "ignore_index", "interpolate", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "mean", "mode", "nn", "none", "not", "num_labels", "of", "output", "output_attentions", "output_hidden_states", "outputs", "pixel_values", "r", "raise", "reduction", "return", "return_dict", "self", "semantic_loss_ignore_index", "shape", "should", "size", "squeeze", "torch", "tuple", "upsampled_logits", "use_return_dict", "valid_mask"], "seggpt/modeling_seggpt.py:SegGptPatchEmbeddings.__init__": ["Conv2d", "Iterable", "ModelPatchEmbeddings", "Module", "__init__", "abc", "class", "collections", "config", "def", "else", "hidden_size", "if", "image_size", "isinstance", "kernel_size", "nn", "num_channels", "num_patches", "patch_size", "projection", "self", "stride", "super"], "seggpt/modeling_seggpt.py:SegGptPatchEmbeddings.forward": ["Input", "Make", "ModelPatchEmbeddings", "Module", "ValueError", "batch_size", "channel", "class", "configuration", "def", "dimension", "doesn", "embeddings", "f", "forward", "height", "if", "image", "image_size", "in", "match", "model", "nn", "num_channels", "of", "one", "or", "permute", "pixel", "pixel_values", "projection", "raise", "return", "self", "set", "shape", "size", "sure", "t", "that", "the", "values", "width", "with"], "seggpt/modeling_seggpt.py:SegGptEmbeddings.__init__": ["Dropout", "ModelConfig", "ModelEmbeddings", "ModelPatchEmbeddings", "Modelment_token_input", "Modelment_token_prompt", "Module", "None", "Parameter", "__init__", "class", "config", "def", "dropout", "hidden_dropout_prob", "hidden_size", "mask_token", "nn", "num_positions", "patch_embeddings", "patch_size", "position_embeddings", "pretrain_image_size", "randn", "self", "super", "torch", "type_token_instance", "type_token_semantic", "zeros"], "seggpt/modeling_seggpt.py:SegGptEmbeddings.interpolate_pos_encoding": ["F", "False", "ModelEmbeddings", "Module", "Tensor", "align_corners", "bicubic", "class", "def", "else", "height", "if", "int", "interpolate", "interpolate_pos_encoding", "is_tracing", "jit", "mode", "nn", "num_patches", "or", "patch_pos_embed", "permute", "position_embeddings", "pretrain_patch_size", "reshape", "return", "self", "shape", "size", "torch", "torch_int", "width"], "seggpt/modeling_seggpt.py:SegGptEmbeddings.forward": ["BoolTensor", "Embedding", "ModelEmbeddings", "Modelment_token_input", "Modelment_token_prompt", "Module", "None", "Tensor", "ValueError", "_", "batch_size", "be", "bool_masked_pos", "but", "cat", "class", "def", "dim", "either", "elif", "else", "embedding_type", "embeddings", "expand", "f", "forward", "got", "if", "input_embeddings", "instance", "interpolate_pos_encoding", "is", "mask_token", "nn", "not", "or", "patch_embeddings", "patch_height", "patch_width", "pixel_values", "pos_embed", "prompt_embeddings", "prompt_pixel_values", "raise", "reshape", "return", "self", "semantic", "shape", "should", "str", "torch", "type", "type_as", "type_embedding", "type_token_instance", "type_token_semantic", "unsqueeze", "w"], "seggpt/modeling_seggpt.py:SegGptAttention.__init__": ["Input", "Iterable", "Linear", "ModelAttention", "Module", "None", "Parameter", "ValueError", "__init__", "abc", "be", "bias", "class", "collections", "config", "def", "else", "encoding", "head_dim", "hidden_size", "if", "image_size", "input_size", "is", "isinstance", "must", "nn", "num_attention_heads", "patch_size", "positional", "proj", "provided", "qkv", "qkv_bias", "raise", "rel_pos_h", "rel_pos_w", "relative", "scale", "self", "size", "super", "torch", "use_relative_position_embeddings", "using", "zeros"], "seggpt/modeling_seggpt.py:SegGptAttention.get_rel_pos": ["F", "ModelAttention", "Module", "None", "Tensor", "arange", "class", "def", "get_rel_pos", "int", "interpolate", "k_coords", "k_size", "linear", "long", "max", "max_rel_dist", "mode", "nn", "permute", "q_coords", "q_size", "rel_pos", "rel_pos_resized", "relative_coords", "reshape", "return", "self", "shape", "size", "torch"], "seggpt/modeling_seggpt.py:SegGptAttention.add_decomposed_rel_pos": ["ModelAttention", "Module", "None", "Tensor", "_", "add_decomposed_rel_pos", "attn", "batch_size", "bhwc", "bhwk", "class", "def", "dim", "einsum", "get_rel_pos", "hkc", "int", "k_size", "key_height", "key_width", "nn", "q_size", "query", "query_height", "query_width", "rel_h", "rel_pos_h", "rel_pos_w", "rel_w", "relative_position_height", "relative_position_width", "reshape", "reshaped_query", "return", "self", "shape", "torch", "tuple", "wkc"], "seggpt/modeling_seggpt.py:SegGptAttention.forward": ["False", "ModelAttention", "Module", "None", "Tensor", "_", "add_decomposed_rel_pos", "attn_output", "attn_weights", "attn_weights_reshaped", "batch_size", "class", "def", "dim", "dtype", "else", "float32", "forward", "functional", "height", "hidden_states", "if", "key", "nn", "num_attention_heads", "output_attentions", "permute", "proj", "qkv", "query", "rel_pos_h", "rel_pos_w", "reshape", "return", "scale", "self", "shape", "softmax", "to", "torch", "transpose", "unbind", "use_relative_position_embeddings", "value", "view", "width"], "seggpt/modeling_seggpt.py:SegGptMlp.__init__": ["ACT2FN", "Linear", "ModelMlp", "Module", "__init__", "act", "class", "config", "def", "hidden_act", "hidden_size", "lin1", "lin2", "mlp_dim", "nn", "self", "super"], "seggpt/modeling_seggpt.py:SegGptMlp.forward": ["ModelMlp", "Module", "Tensor", "act", "class", "def", "forward", "hidden_states", "lin1", "lin2", "nn", "return", "self", "torch"], "seggpt/modeling_seggpt.py:drop_path": ["False", "Model_path", "Model_prob", "Tensor", "bool", "def", "device", "div", "dtype", "float", "floor_", "if", "input", "keep_prob", "ndim", "not", "or", "output", "rand", "random_tensor", "return", "shape", "torch", "training"], "seggpt/modeling_seggpt.py:SegGptDropPath.__init__": ["ModelDropPath", "Module", "None", "__init__", "class", "def", "drop_prob", "float", "nn", "self", "super"], "seggpt/modeling_seggpt.py:SegGptDropPath.forward": ["ModelDropPath", "Module", "Tensor", "class", "def", "drop_path", "drop_prob", "forward", "hidden_states", "nn", "return", "self", "torch", "training"], "seggpt/modeling_seggpt.py:SegGptDropPath.extra_repr": ["ModelDropPath", "Module", "class", "def", "drop_prob", "extra_repr", "f", "nn", "p", "return", "self", "str"], "seggpt/modeling_seggpt.py:SegGptLayer.__init__": ["GradientCheckpointingLayer", "Identity", "LayerNorm", "ModelAttention", "ModelConfig", "ModelDropPath", "ModelLayer", "ModelMlp", "None", "__init__", "attention", "class", "config", "def", "drop_path", "drop_path_rate", "else", "eps", "float", "hidden_size", "if", "layer_norm_eps", "layernorm_after", "layernorm_before", "mlp", "nn", "self", "super"], "seggpt/modeling_seggpt.py:SegGptLayer.forward": ["False", "GradientCheckpointingLayer", "ModelLayer", "Tensor", "True", "and", "attention", "attention_output", "bool", "cat", "class", "def", "dim", "drop_path", "else", "ensemble_cond", "expand_as", "feature_ensemble", "forward", "hidden_states", "if", "inputs", "int", "keepdim", "layernorm_after", "layernorm_before", "mean", "mlp", "num_prompts", "output_attentions", "outputs", "prompt", "reshape", "residual", "return", "self", "self_attention_outputs", "shape", "split", "torch", "tuple"], "seggpt/modeling_seggpt.py:SegGptEncoder.__init__": ["False", "LayerNorm", "ModelConfig", "ModelEncoder", "ModelLayer", "Module", "ModuleList", "None", "__init__", "class", "config", "cpu", "def", "device", "dpr", "drop_path_rate", "eps", "for", "gradient_checkpointing", "hidden_size", "i", "in", "item", "layer_norm_eps", "layernorm", "layers", "linspace", "nn", "num_hidden_layers", "range", "self", "super", "torch", "x"], "seggpt/modeling_seggpt.py:SegGptEncoder.forward": ["False", "ModelEncoder", "ModelEncoderOutput", "Module", "None", "Tensor", "True", "all_hidden_states", "all_self_attentions", "append", "attentions", "bool", "class", "config", "def", "else", "ensemble_cond", "enumerate", "feature_ensemble", "for", "forward", "hidden_states", "i", "if", "in", "intermediate_hidden_state_indices", "intermediate_hidden_states", "is", "last_hidden_state", "layer_module", "layer_outputs", "layernorm", "layers", "merge_index", "nn", "not", "output_attentions", "output_hidden_states", "return", "return_dict", "self", "shape", "torch", "tuple", "v"], "seggpt/modeling_seggpt.py:SegGptLayerNorm.__init__": ["LayerNorm", "ModelLayerNorm", "NotImplementedError", "Unsupported", "__init__", "channels_first", "channels_last", "class", "data", "data_format", "def", "eps", "f", "format", "if", "in", "kwargs", "nn", "normalized_shape", "not", "raise", "self", "super"], "seggpt/modeling_seggpt.py:SegGptLayerNorm.forward": ["LayerNorm", "ModelLayerNorm", "Tensor", "channels_first", "class", "data_format", "def", "else", "features", "forward", "if", "nn", "permute", "return", "self", "super", "torch"], "seggpt/modeling_seggpt.py:SegGptDecoderHead.__init__": ["ACT2FN", "Conv2d", "ModelDecoderHead", "ModelLayerNorm", "Module", "True", "__init__", "act_fct", "bias", "channels_first", "class", "config", "conv", "data_format", "decoder_hidden_size", "def", "eps", "head", "hidden_act", "kernel_size", "layer_norm_eps", "layernorm", "nn", "normalized_shape", "padding", "self", "super"], "seggpt/modeling_seggpt.py:SegGptDecoderHead.forward": ["FloatTensor", "ModelDecoderHead", "Module", "act_fct", "class", "conv", "def", "forward", "head", "hidden_states", "layernorm", "nn", "return", "self", "torch"], "seggpt/modeling_seggpt.py:SegGptDecoder.__init__": ["Linear", "ModelDecoder", "ModelDecoderHead", "Module", "True", "__init__", "bias", "class", "config", "decoder_embed", "decoder_hidden_size", "decoder_pred", "def", "hidden_size", "intermediate_hidden_state_indices", "len", "nn", "patch_size", "self", "super"], "seggpt/modeling_seggpt.py:SegGptDecoder._reshape_hidden_states": ["FloatTensor", "ModelDecoder", "Module", "_", "_reshape_hidden_states", "batch_size", "class", "decoder_hidden_size", "def", "hidden_states", "nn", "patch_height", "patch_size", "patch_width", "permute", "reshape", "return", "self", "shape", "torch"], "seggpt/modeling_seggpt.py:SegGptDecoder.forward": ["FloatTensor", "ModelDecoder", "Module", "_reshape_hidden_states", "class", "decoder_embed", "decoder_pred", "def", "forward", "hidden_states", "nn", "return", "self", "torch"], "seggpt/modeling_seggpt.py:SegGptPreTrainedModel._init_weights": ["Conv2d", "LayerNorm", "Linear", "ModelAttention", "ModelEmbeddings", "ModelLayerNorm", "ModelPreTrainedModel", "Modelment_token_input", "Modelment_token_prompt", "Module", "None", "PreTrainedModel", "_init_weights", "bias", "class", "config", "def", "elif", "if", "init", "initializer_range", "is", "isinstance", "mask_token", "mean", "module", "nn", "normal_", "not", "ones_", "position_embeddings", "rel_pos_h", "rel_pos_w", "self", "std", "trunc_normal_", "type_token_instance", "type_token_semantic", "weight", "zeros_"], "seggpt/modeling_seggpt.py:SegGptModel.__init__": ["ModelConfig", "ModelEmbeddings", "ModelEncoder", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "config", "def", "embeddings", "encoder", "post_init", "self", "super"], "seggpt/modeling_seggpt.py:SegGptModel.get_input_embeddings": ["ModelModel", "ModelPatchEmbeddings", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "patch_embeddings", "return", "self"], "seggpt/modeling_seggpt.py:SegGptModel.forward": ["BoolTensor", "False", "FloatTensor", "If", "It", "Labels", "ModelEncoderOutput", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "a", "and", "be", "bool", "bool_masked_pos", "bool_masked_pos_ones", "bool_masked_pos_zeros", "but", "cat", "class", "config", "def", "default", "device", "dim", "dtype", "else", "embedding_output", "embedding_type", "embeddings", "encoder", "encoder_outputs", "expected_dtype", "feature_ensemble", "forward", "if", "is", "kwargs", "labels", "logger", "make", "model", "not", "num_patches", "ones", "output_attentions", "output_hidden_states", "patch_embeddings", "pixel_values", "projection", "prompt_masks", "prompt_pixel_values", "provide", "provided", "r", "re", "return", "return_dict", "self", "set", "str", "sure", "the", "to", "torch", "training", "tuple", "unsqueeze", "use_return_dict", "value", "warning_once", "weight", "were", "will", "you", "zeros"], "seggpt/modeling_seggpt.py:patchify": ["Model", "Tensor", "batch_size", "def", "height", "int", "num_channels", "patch_height", "patch_size", "patch_width", "permute", "reshape", "return", "shape", "tensor", "torch", "width"], "seggpt/modeling_seggpt.py:unpatchify": ["Model", "Number", "Tensor", "ValueError", "and", "batch_size", "def", "does", "f", "height", "if", "int", "match", "not", "of", "patch", "patch_height", "patch_size", "patch_width", "patches", "permute", "raise", "reshape", "return", "shape", "tensor", "torch", "width"], "seggpt/modeling_seggpt.py:SegGptLoss.__init__": ["ModelLoss", "Module", "__init__", "beta", "class", "config", "def", "nn", "patch_size", "self", "super"], "seggpt/modeling_seggpt.py:SegGptLoss.forward": ["BoolTensor", "F", "FloatTensor", "ModelLoss", "Module", "None", "beta", "bool_masked_pos", "cat", "class", "def", "dim", "forward", "ground_truth", "labels", "loss", "mask", "nn", "none", "patch_size", "pred_masks", "prompt_masks", "reduction", "repeat", "return", "self", "shape", "smooth_l1_loss", "sum", "torch", "unpatchify"], "seggpt/modeling_seggpt.py:SegGptForImageSegmentation.__init__": ["ModelConfig", "ModelDecoder", "ModelForImageModelmentation", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "config", "decoder", "def", "model", "post_init", "self", "super"], "seggpt/modeling_seggpt.py:SegGptForImageSegmentation.forward": ["BoolTensor", "FloatTensor", "ModelForImageModelmentation", "ModelImageModelmentationOutput", "ModelLoss", "ModelPreTrainedModel", "None", "Tensor", "attentions", "bool", "bool_masked_pos", "bool_masked_pos_ones", "bool_masked_pos_zeros", "cat", "class", "config", "decoder", "def", "device", "dim", "dtype", "else", "embedding_type", "embeddings", "feature_ensemble", "forward", "hidden_states", "idx", "if", "intermediate_hidden_states", "is", "kwargs", "labels", "loss", "loss_fn", "model", "not", "num_patches", "ones", "output", "output_attentions", "output_hidden_states", "outputs", "patch_embeddings", "pixel_values", "pred_masks", "prompt_masks", "prompt_pixel_values", "r", "return", "return_dict", "self", "str", "torch", "tuple", "unsqueeze", "use_return_dict", "zeros"], "sew/modeling_sew.py:SEWNoLayerNormConvLayer.__init__": ["ACT2FN", "Conv1d", "GradientCheckpointingLayer", "Model", "__init__", "activation", "bias", "class", "config", "conv", "conv_bias", "conv_dim", "conv_kernel", "conv_stride", "def", "else", "feat_extract_activation", "if", "in_conv_dim", "kernel_size", "layer_id", "nn", "out_conv_dim", "self", "stride", "super"], "sew/modeling_sew.py:SEWNoLayerNormConvLayer.forward": ["GradientCheckpointingLayer", "Model", "activation", "class", "conv", "def", "forward", "hidden_states", "return", "self"], "sew/modeling_sew.py:SEWLayerNormConvLayer.__init__": ["ACT2FN", "Conv1d", "GradientCheckpointingLayer", "LayerNorm", "Model", "True", "__init__", "activation", "bias", "class", "config", "conv", "conv_bias", "conv_dim", "conv_kernel", "conv_stride", "def", "elementwise_affine", "else", "feat_extract_activation", "if", "in_conv_dim", "kernel_size", "layer_id", "layer_norm", "nn", "out_conv_dim", "self", "stride", "super"], "sew/modeling_sew.py:SEWLayerNormConvLayer.forward": ["GradientCheckpointingLayer", "Model", "activation", "class", "conv", "def", "forward", "hidden_states", "layer_norm", "return", "self", "transpose"], "sew/modeling_sew.py:SEWGroupNormConvLayer.__init__": ["ACT2FN", "Conv1d", "GradientCheckpointingLayer", "GroupNorm", "Model", "True", "__init__", "activation", "affine", "bias", "class", "config", "conv", "conv_bias", "conv_dim", "conv_kernel", "conv_stride", "def", "else", "feat_extract_activation", "if", "in_conv_dim", "kernel_size", "layer_id", "layer_norm", "nn", "num_channels", "num_groups", "out_conv_dim", "self", "stride", "super"], "sew/modeling_sew.py:SEWGroupNormConvLayer.forward": ["GradientCheckpointingLayer", "Model", "activation", "class", "conv", "def", "forward", "hidden_states", "layer_norm", "return", "self"], "sew/modeling_sew.py:SEWPositionalConvEmbedding.__init__": ["ACT2FN", "Conv1d", "GatheredParameters", "Model", "ModelSamePadLayer", "Module", "__init__", "activation", "class", "config", "conv", "deepspeed", "def", "dim", "else", "feat_extract_activation", "groups", "hasattr", "hidden_size", "if", "is_deepspeed_zero3_enabled", "kernel_size", "modifier_rank", "name", "nn", "num_conv_pos_embedding_groups", "num_conv_pos_embeddings", "original0", "original1", "padding", "parametrizations", "register_external_parameter", "self", "squeeze_factor", "stride", "super", "utils", "weight", "weight_g", "weight_norm", "weight_v", "with", "zero"], "sew/modeling_sew.py:SEWPositionalConvEmbedding.forward": ["Model", "Module", "activation", "class", "conv", "def", "forward", "hidden_states", "nn", "padding", "return", "self"], "sew/modeling_sew.py:SEWSamePadLayer.__init__": ["Model", "Module", "__init__", "class", "def", "else", "if", "nn", "num_conv_pos_embeddings", "num_pad_remove", "self", "super"], "sew/modeling_sew.py:SEWSamePadLayer.forward": ["Model", "Module", "class", "def", "forward", "hidden_states", "if", "nn", "num_pad_remove", "return", "self"], "sew/modeling_sew.py:SEWUpsampling.__init__": ["ACT2FN", "Linear", "Model", "Module", "__init__", "activation", "class", "config", "def", "feat_extract_activation", "hidden_size", "nn", "projection", "self", "squeeze_factor", "super"], "sew/modeling_sew.py:SEWUpsampling.forward": ["Model", "Module", "activation", "bsz", "class", "def", "forward", "hidden_states", "if", "nn", "projection", "reshape", "return", "self", "size", "squeeze_factor", "src_embed_dim", "src_len", "tgt_embed_dim", "tgt_len"], "sew/modeling_sew.py:SEWFeatureEncoder.__init__": ["False", "Model", "ModelGroupNormConvLayer", "ModelLayerNormConvLayer", "ModelNoLayerNormConvLayer", "Module", "ModuleList", "True", "ValueError", "__init__", "_requires_grad", "be", "but", "class", "config", "conv_layers", "def", "elif", "else", "f", "feat_extract_norm", "for", "gradient_checkpointing", "group", "has", "i", "if", "in", "is", "layer", "layer_id", "nn", "num_feat_extract_layers", "of", "one", "raise", "range", "self", "super", "to"], "sew/modeling_sew.py:SEWFeatureEncoder._freeze_parameters": ["False", "Model", "Module", "_freeze_parameters", "_requires_grad", "class", "def", "for", "in", "nn", "param", "parameters", "requires_grad", "self"], "sew/modeling_sew.py:SEWFeatureEncoder.forward": ["Model", "Module", "None", "True", "_requires_grad", "and", "class", "conv_layer", "conv_layers", "def", "for", "forward", "hidden_states", "if", "in", "input_values", "nn", "requires_grad", "return", "self", "training"], "sew/modeling_sew.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "contiguous", "def", "dim", "dropout", "float", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "shape", "size", "softmax", "torch", "training", "transpose", "value"], "sew/modeling_sew.py:SEWAttention.__init__": ["False", "Linear", "Model", "ModelConfig", "Module", "None", "True", "ValueError", "__init__", "and", "be", "bias", "bool", "by", "class", "config", "def", "divisible", "dropout", "embed_dim", "f", "float", "got", "head_dim", "if", "int", "is_causal", "is_decoder", "k_proj", "must", "nn", "num_heads", "out_proj", "q_proj", "raise", "scaling", "self", "super", "v_proj"], "sew/modeling_sew.py:SEWAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "False", "FlashAttentionKwargs", "Model", "Module", "None", "Tensor", "Unpack", "_attn_implementation", "attention_interface", "attention_mask", "attn_output", "attn_weights", "bool", "bsz", "class", "config", "contiguous", "current_states", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "head_dim", "hidden_states", "if", "is", "is_cross_attention", "k_proj", "key_states", "key_value_states", "kv_input_shape", "kwargs", "nn", "not", "out_proj", "output_attentions", "q_input_shape", "q_proj", "query_states", "reshape", "return", "scaling", "self", "shape", "src_len", "tgt_len", "torch", "training", "transpose", "tuple", "v_proj", "value_states", "view"], "sew/modeling_sew.py:SEWFeedForward.__init__": ["ACT2FN", "Dropout", "Linear", "Model", "Module", "__init__", "activation_dropout", "class", "config", "def", "else", "hidden_act", "hidden_dropout", "hidden_size", "if", "intermediate_act_fn", "intermediate_dense", "intermediate_dropout", "intermediate_size", "isinstance", "nn", "output_dense", "output_dropout", "self", "str", "super"], "sew/modeling_sew.py:SEWFeedForward.forward": ["Model", "Module", "class", "def", "forward", "hidden_states", "intermediate_act_fn", "intermediate_dense", "intermediate_dropout", "nn", "output_dense", "output_dropout", "return", "self"], "sew/modeling_sew.py:SEWEncoderLayer.__init__": ["Dropout", "False", "GradientCheckpointingLayer", "LayerNorm", "Model", "ModelAttention", "ModelFeedForward", "__init__", "attention", "attention_dropout", "class", "config", "def", "dropout", "embed_dim", "eps", "feed_forward", "final_layer_norm", "hidden_dropout", "hidden_size", "is_decoder", "layer_norm", "layer_norm_eps", "nn", "num_attention_heads", "num_heads", "self", "super"], "sew/modeling_sew.py:SEWEncoderLayer.forward": ["False", "GradientCheckpointingLayer", "Model", "None", "_", "attention", "attention_mask", "attn_residual", "attn_weights", "class", "def", "dropout", "feed_forward", "final_layer_norm", "forward", "hidden_states", "if", "layer_norm", "output_attentions", "outputs", "return", "self"], "sew/modeling_sew.py:SEWEncoder.__init__": ["AvgPool1d", "Dropout", "False", "LayerNorm", "Model", "ModelLayer", "ModelPositionalConvEmbedding", "ModelUpsampling", "Module", "ModuleList", "_", "__init__", "class", "config", "def", "dropout", "eps", "for", "gradient_checkpointing", "hidden_dropout", "hidden_size", "in", "layer_norm", "layer_norm_eps", "layers", "nn", "num_hidden_layers", "pool", "pos_conv_embed", "range", "self", "squeeze_factor", "super", "upsample"], "sew/modeling_sew.py:SEWEncoder.forward": ["BaseModelOutput", "False", "Model", "Module", "None", "True", "_attn_implementation", "all_hidden_states", "all_self_attentions", "and", "arange", "attention_ids", "attention_mask", "attentions", "class", "config", "def", "device", "dropout", "dropout_probability", "dtype", "else", "expand", "expand_attention_mask", "finfo", "flash_attention_2", "for", "forward", "functional", "hidden_states", "if", "in", "input_lengths", "is", "is_deepspeed_zero3_enabled", "is_fsdp_managed_module", "last_hidden_state", "layer", "layer_norm", "layer_outputs", "layerdrop", "layers", "long", "max_encoder_length", "min", "min_length", "n_input_timesteps", "nn", "not", "or", "output_attentions", "output_hidden_states", "output_lengths", "pad", "pool", "pooled_hidden_states", "pos_conv_embed", "position_embeddings", "rand", "repeat", "return", "return_dict", "self", "shape", "size", "skip_the_layer", "squeeze_factor", "sum", "synced_gpus", "to", "torch", "training", "transpose", "tuple", "unsqueeze", "upsample", "v", "view"], "sew/modeling_sew.py:SEWPreTrainedModel._init_weights": ["Conv1d", "GatheredParameters", "GroupNorm", "LayerNorm", "Linear", "Model", "ModelPositionalConvEmbedding", "None", "PreTrainedModel", "_init_weights", "and", "bias", "class", "config", "constant_", "conv", "deepspeed", "def", "elif", "else", "hasattr", "if", "in_channels", "init", "initializer_range", "is", "is_deepspeed_zero3_enabled", "isinstance", "kaiming_normal_", "kernel_size", "math", "mean", "modifier_rank", "module", "nn", "normal_", "not", "ones_", "self", "sqrt", "std", "weight", "weight_g", "weight_v", "with", "zero", "zeros_"], "sew/modeling_sew.py:SEWPreTrainedModel._get_feat_extract_output_lengths": ["LongTensor", "Model", "PreTrainedModel", "_conv_out_length", "_get_feat_extract_output_lengths", "class", "config", "conv_kernel", "conv_stride", "def", "div", "floor", "for", "in", "input_length", "input_lengths", "int", "kernel_size", "return", "rounding_mode", "self", "stride", "torch", "zip"], "sew/modeling_sew.py:SEWPreTrainedModel._get_feature_vector_attention_mask": ["LongTensor", "Model", "PreTrainedModel", "_get_feat_extract_output_lengths", "_get_feature_vector_attention_mask", "arange", "attention_mask", "batch_size", "bool", "class", "cumsum", "def", "device", "dtype", "feature_vector_length", "flip", "int", "long", "output_lengths", "return", "self", "shape", "sum", "to", "torch", "zeros"], "sew/modeling_sew.py:_compute_mask_indices": ["False", "LongTensor", "None", "ValueError", "_", "_compute_mask_indices", "and", "append", "arange", "array", "attention_mask", "batch_size", "be", "bigger", "bool", "broadcast_to", "but", "choice", "compute_num_masked_span", "concatenate", "def", "detach", "dtype", "dummy_mask_idx", "else", "epsilon", "f", "float", "for", "got", "has", "if", "in", "input_length", "input_lengths", "int", "int32", "is", "item", "len", "mask_length", "mask_prob", "max", "max_num_masked_span", "min_masks", "ndarray", "not", "np", "num_masked_span", "offsets", "ones", "put_along_axis", "raise", "rand", "random", "range", "replace", "reshape", "return", "sequence_length", "shape", "smaller", "spec_aug_mask", "spec_aug_mask_idx", "spec_aug_mask_idxs", "sum", "than", "to", "tolist", "torch", "tuple", "zeros"], "sew/modeling_sew.py:SEWModel.__init__": ["Dropout", "LayerNorm", "Linear", "Model", "ModelConfig", "ModelEncoder", "ModelFeatureEncoder", "ModelPreTrainedModel", "Parameter", "Tensor", "__init__", "class", "config", "conv_dim", "def", "encoder", "eps", "feat_proj_dropout", "feature_dropout", "feature_extractor", "feature_projection", "hidden_size", "if", "layer_norm", "layer_norm_eps", "mask_feature_prob", "mask_time_prob", "masked_spec_embed", "nn", "or", "post_init", "project_features", "self", "super", "torch", "uniform_"], "sew/modeling_sew.py:SEWModel._mask_hidden_states": ["FloatTensor", "LongTensor", "Model", "ModelPreTrainedModel", "None", "True", "_compute_mask_indices", "_mask_hidden_states", "and", "apply_spec_augment", "attention_mask", "batch_size", "bool", "class", "config", "def", "device", "dtype", "elif", "expand", "getattr", "hidden_size", "hidden_states", "if", "is", "mask_feature_indices", "mask_feature_length", "mask_feature_min_masks", "mask_feature_prob", "mask_length", "mask_prob", "mask_time_indices", "mask_time_length", "mask_time_min_masks", "mask_time_prob", "masked_spec_embed", "min_masks", "not", "return", "self", "sequence_length", "size", "tensor", "to", "torch", "training"], "sew/modeling_sew.py:SEWModel.forward": ["BaseModelOutput", "FloatTensor", "Model", "ModelPreTrainedModel", "None", "Tensor", "_get_feature_vector_attention_mask", "_mask_hidden_states", "attention_mask", "attentions", "bool", "class", "config", "def", "else", "encoder", "encoder_outputs", "extract_features", "feature_dropout", "feature_extractor", "feature_projection", "forward", "hidden_states", "if", "input_values", "is", "kwargs", "last_hidden_state", "layer_norm", "mask_time_indices", "not", "output_attentions", "output_hidden_states", "project_features", "r", "return", "return_dict", "self", "shape", "torch", "transpose", "tuple", "use_return_dict"], "sew/modeling_sew.py:SEWForCTC.__init__": ["Dropout", "Linear", "Model", "ModelModel", "ModelPreTrainedModel", "None", "Please", "ValueError", "You", "__class__", "__init__", "a", "add_adapter", "and", "are", "as", "class", "config", "configuration", "def", "define", "does", "dropout", "else", "f", "final_dropout", "follows", "from_pretrained", "hasattr", "head", "hidden_size", "if", "instantiate", "is", "language", "lm_head", "model", "nn", "not", "of", "or", "output_hidden_size", "post_init", "r", "raise", "s", "self", "size", "str", "super", "target_lang", "that", "the", "to", "trying", "vocab_size", "vocabulary", "with", "your"], "sew/modeling_sew.py:SEWForCTC.tie_weights": ["By", "Cannot", "Model", "ModelPreTrainedModel", "None", "True", "ValueError", "adapter_attn_dim", "and", "class", "config", "def", "default", "defined", "elif", "eng", "f", "force_load", "getattr", "if", "info", "is", "kwargs", "load_adapter", "logger", "not", "pass", "raise", "self", "set", "target_lang", "tie_weights", "to"], "sew/modeling_sew.py:SEWForCTC.freeze_feature_encoder": ["Model", "ModelPreTrainedModel", "_freeze_parameters", "class", "def", "feature_extractor", "freeze_feature_encoder", "self"], "sew/modeling_sew.py:SEWForCTC.freeze_base_model": ["False", "Model", "ModelPreTrainedModel", "class", "def", "for", "freeze_base_model", "in", "param", "parameters", "requires_grad", "self"], "sew/modeling_sew.py:SEWForCTC.forward": ["CausalLMOutput", "False", "Label", "Model", "ModelPreTrainedModel", "None", "Tensor", "ValueError", "_HIDDEN_STATES_START_POSITION", "_get_feat_extract_output_lengths", "and", "attention_mask", "attentions", "backends", "be", "blank", "bool", "class", "config", "ctc_loss", "ctc_loss_reduction", "ctc_zero_infinity", "cudnn", "def", "dim", "dropout", "dtype", "else", "enabled", "f", "flags", "flattened_targets", "float32", "forward", "functional", "hidden_states", "if", "input_lengths", "input_values", "is", "kwargs", "labels", "labels_mask", "lm_head", "log_probs", "log_softmax", "logits", "long", "loss", "masked_select", "max", "must", "nn", "not", "ones_like", "output", "output_attentions", "output_hidden_states", "outputs", "pad_token_id", "r", "raise", "reduction", "return", "return_dict", "self", "sum", "target_lengths", "to", "torch", "transpose", "tuple", "use_return_dict", "values", "vocab_size", "with", "zero_infinity"], "sew/modeling_sew.py:SEWForSequenceClassification.__init__": ["Linear", "Model", "ModelModel", "ModelPreTrainedModel", "Parameter", "Sequence", "True", "ValueError", "__init__", "adapters", "add_adapter", "and", "class", "classification", "classifier", "classifier_proj_size", "config", "def", "does", "hasattr", "hidden_size", "if", "layer_weights", "nn", "not", "num_hidden_layers", "num_labels", "num_layers", "of", "ones", "post_init", "projector", "raise", "self", "super", "support", "the", "torch", "use", "use_weighted_layer_sum"], "sew/modeling_sew.py:SEWForSequenceClassification.freeze_feature_encoder": ["Model", "ModelPreTrainedModel", "_freeze_parameters", "class", "def", "feature_extractor", "freeze_feature_encoder", "self"], "sew/modeling_sew.py:SEWForSequenceClassification.freeze_base_model": ["False", "Model", "ModelPreTrainedModel", "class", "def", "for", "freeze_base_model", "in", "param", "parameters", "requires_grad", "self"], "sew/modeling_sew.py:SEWForSequenceClassification.forward": ["CrossEntropyLoss", "Model", "ModelPreTrainedModel", "None", "SequenceClassifierOutput", "Tensor", "True", "_HIDDEN_STATES_START_POSITION", "_get_feature_vector_attention_mask", "attention_mask", "attentions", "bool", "class", "classifier", "config", "def", "dim", "else", "expand_padding_mask", "forward", "functional", "hidden_states", "if", "input_values", "is", "kwargs", "labels", "layer_weights", "logits", "loss", "loss_fct", "mean", "nn", "norm_weights", "not", "num_labels", "output", "output_attentions", "output_hidden_states", "outputs", "padding_mask", "pooled_output", "projector", "r", "repeat", "return", "return_dict", "self", "shape", "softmax", "stack", "sum", "torch", "tuple", "unsqueeze", "use_return_dict", "use_weighted_layer_sum", "view"], "sew_d/modeling_sew_d.py:_compute_mask_indices": ["False", "LongTensor", "None", "ValueError", "_", "_compute_mask_indices", "and", "append", "arange", "array", "attention_mask", "batch_size", "be", "bigger", "bool", "broadcast_to", "but", "choice", "compute_num_masked_span", "concatenate", "def", "detach", "dtype", "dummy_mask_idx", "else", "epsilon", "f", "float", "for", "got", "has", "if", "in", "input_length", "input_lengths", "int", "int32", "is", "item", "len", "mask_length", "mask_prob", "max", "max_num_masked_span", "min_masks", "ndarray", "not", "np", "num_masked_span", "offsets", "ones", "put_along_axis", "raise", "rand", "random", "range", "replace", "reshape", "return", "sequence_length", "shape", "smaller", "spec_aug_mask", "spec_aug_mask_idx", "spec_aug_mask_idxs", "sum", "than", "to", "tolist", "torch", "tuple", "zeros"], "sew_d/modeling_sew_d.py:make_log_bucket_position": ["Model_log_bucket_position", "abs", "abs_pos", "bucket_pos", "bucket_size", "ceil", "def", "log", "log_pos", "max_position", "mid", "relative_pos", "return", "sign", "tensor", "torch", "type_as", "where"], "sew_d/modeling_sew_d.py:build_relative_position": ["Model_relative_position", "None", "and", "arange", "bucket_size", "def", "device", "if", "k_ids", "key_size", "long", "make_log_bucket_position", "max_position", "q_ids", "query_size", "rel_pos_ids", "return", "to", "torch", "unsqueeze"], "sew_d/modeling_sew_d.py:c2p_dynamic_expand": ["Model_dynamic_expand", "Model_pos", "def", "expand", "query_layer", "relative_pos", "return", "size"], "sew_d/modeling_sew_d.py:p2c_dynamic_expand": ["Model_dynamic_expand", "c2p_pos", "def", "expand", "key_layer", "query_layer", "return", "size"], "sew_d/modeling_sew_d.py:pos_dynamic_expand": ["Model_dynamic_expand", "Model_index", "def", "expand", "key_layer", "p2c_att", "return", "size"], "sew_d/modeling_sew_d.py:get_mask": ["DropoutContext", "Model_mask", "None", "and", "bernoulli_", "bool", "def", "dropout", "else", "empty_like", "if", "input", "is", "isinstance", "local_context", "mask", "not", "return", "reuse_mask", "scale", "to", "torch"], "sew_d/modeling_sew_d.py:SEWDNoLayerNormConvLayer.__init__": ["ACT2FN", "Conv1d", "GradientCheckpointingLayer", "Model", "__init__", "activation", "bias", "class", "config", "conv", "conv_bias", "conv_dim", "conv_kernel", "conv_stride", "def", "else", "feat_extract_activation", "if", "in_conv_dim", "kernel_size", "layer_id", "nn", "out_conv_dim", "self", "stride", "super"], "sew_d/modeling_sew_d.py:SEWDNoLayerNormConvLayer.forward": ["GradientCheckpointingLayer", "Model", "activation", "class", "conv", "def", "forward", "hidden_states", "return", "self"], "sew_d/modeling_sew_d.py:SEWDLayerNormConvLayer.__init__": ["ACT2FN", "Conv1d", "GradientCheckpointingLayer", "LayerNorm", "Model", "True", "__init__", "activation", "bias", "class", "config", "conv", "conv_bias", "conv_dim", "conv_kernel", "conv_stride", "def", "elementwise_affine", "else", "feat_extract_activation", "if", "in_conv_dim", "kernel_size", "layer_id", "layer_norm", "nn", "out_conv_dim", "self", "stride", "super"], "sew_d/modeling_sew_d.py:SEWDLayerNormConvLayer.forward": ["GradientCheckpointingLayer", "Model", "activation", "class", "conv", "def", "forward", "hidden_states", "layer_norm", "return", "self", "transpose"], "sew_d/modeling_sew_d.py:SEWDGroupNormConvLayer.__init__": ["ACT2FN", "Conv1d", "GradientCheckpointingLayer", "GroupNorm", "Model", "True", "__init__", "activation", "affine", "bias", "class", "config", "conv", "conv_bias", "conv_dim", "conv_kernel", "conv_stride", "def", "else", "feat_extract_activation", "if", "in_conv_dim", "kernel_size", "layer_id", "layer_norm", "nn", "num_channels", "num_groups", "out_conv_dim", "self", "stride", "super"], "sew_d/modeling_sew_d.py:SEWDGroupNormConvLayer.forward": ["GradientCheckpointingLayer", "Model", "activation", "class", "conv", "def", "forward", "hidden_states", "layer_norm", "return", "self"], "sew_d/modeling_sew_d.py:SEWDPositionalConvEmbedding.__init__": ["ACT2FN", "Conv1d", "GatheredParameters", "Model", "ModelSamePadLayer", "Module", "__init__", "activation", "class", "config", "conv", "deepspeed", "def", "dim", "else", "feat_extract_activation", "groups", "hasattr", "hidden_size", "if", "is_deepspeed_zero3_enabled", "kernel_size", "modifier_rank", "name", "nn", "num_conv_pos_embedding_groups", "num_conv_pos_embeddings", "original0", "original1", "padding", "parametrizations", "register_external_parameter", "self", "squeeze_factor", "stride", "super", "utils", "weight", "weight_g", "weight_norm", "weight_v", "with", "zero"], "sew_d/modeling_sew_d.py:SEWDPositionalConvEmbedding.forward": ["Model", "Module", "activation", "class", "conv", "def", "forward", "hidden_states", "nn", "padding", "return", "self"], "sew_d/modeling_sew_d.py:SEWDSamePadLayer.__init__": ["Model", "Module", "__init__", "class", "def", "else", "if", "nn", "num_conv_pos_embeddings", "num_pad_remove", "self", "super"], "sew_d/modeling_sew_d.py:SEWDSamePadLayer.forward": ["Model", "Module", "class", "def", "forward", "hidden_states", "if", "nn", "num_pad_remove", "return", "self"], "sew_d/modeling_sew_d.py:SEWDUpsampling.__init__": ["ACT2FN", "Linear", "Model", "Module", "__init__", "activation", "class", "config", "def", "feat_extract_activation", "hidden_size", "nn", "projection", "self", "squeeze_factor", "super"], "sew_d/modeling_sew_d.py:SEWDUpsampling.forward": ["Model", "Module", "activation", "bsz", "class", "def", "forward", "hidden_states", "if", "nn", "projection", "reshape", "return", "self", "size", "squeeze_factor", "src_embed_dim", "src_len", "tgt_embed_dim", "tgt_len"], "sew_d/modeling_sew_d.py:SEWDFeatureEncoder.__init__": ["False", "Model", "ModelGroupNormConvLayer", "ModelLayerNormConvLayer", "ModelNoLayerNormConvLayer", "Module", "ModuleList", "True", "ValueError", "__init__", "_requires_grad", "be", "but", "class", "config", "conv_layers", "def", "elif", "else", "f", "feat_extract_norm", "for", "gradient_checkpointing", "group", "has", "i", "if", "in", "is", "layer", "layer_id", "nn", "num_feat_extract_layers", "of", "one", "raise", "range", "self", "super", "to"], "sew_d/modeling_sew_d.py:SEWDFeatureEncoder._freeze_parameters": ["False", "Model", "Module", "_freeze_parameters", "_requires_grad", "class", "def", "for", "in", "nn", "param", "parameters", "requires_grad", "self"], "sew_d/modeling_sew_d.py:SEWDFeatureEncoder.forward": ["Model", "Module", "None", "True", "_requires_grad", "and", "class", "conv_layer", "conv_layers", "def", "for", "forward", "hidden_states", "if", "in", "input_values", "nn", "requires_grad", "return", "self", "training"], "sew_d/modeling_sew_d.py:ContextPooler.__init__": ["Linear", "ModelPooler", "Module", "StableDropout", "__init__", "class", "config", "def", "dense", "dropout", "nn", "pooler_dropout", "pooler_hidden_size", "self", "super"], "sew_d/modeling_sew_d.py:ContextPooler.forward": ["ACT2FN", "ModelPooler", "Model_token", "Module", "class", "config", "def", "dense", "dropout", "forward", "hidden_states", "nn", "pooled_output", "pooler_hidden_act", "return", "self"], "sew_d/modeling_sew_d.py:ContextPooler.output_dim": ["ModelPooler", "Module", "class", "config", "def", "hidden_size", "nn", "output_dim", "return", "self"], "sew_d/modeling_sew_d.py:XSoftmax.forward": ["Function", "Model", "autograd", "bool", "class", "ctx", "def", "dim", "dtype", "finfo", "forward", "input", "mask", "masked_fill", "masked_fill_", "min", "output", "return", "rmask", "save_for_backward", "softmax", "tensor", "to", "torch"], "sew_d/modeling_sew_d.py:XSoftmax.backward": ["Function", "Model", "None", "autograd", "backward", "class", "ctx", "def", "grad_output", "inputGrad", "output", "return", "saved_tensors", "softmax_backward_data", "torch"], "sew_d/modeling_sew_d.py:XSoftmax.symbolic": ["Bool", "Cast", "Constant", "Function", "Long", "Model", "Sub", "autograd", "bool", "cast_pytorch_to_onnx", "class", "def", "dim", "dtype", "finfo", "g", "int64", "mask", "mask_cast_value", "masked_fill", "min", "op", "output", "r_mask", "return", "self", "softmax", "sym_help", "symbolic", "tensor", "to_i", "torch", "type", "value_t"], "sew_d/modeling_sew_d.py:DropoutContext.__init__": ["Model", "ModelContext", "None", "True", "__init__", "class", "def", "mask", "reuse_mask", "scale", "self"], "sew_d/modeling_sew_d.py:XDropout.forward": ["Function", "Model", "autograd", "class", "ctx", "def", "dropout", "else", "forward", "get_mask", "if", "input", "local_ctx", "mask", "masked_fill", "return", "save_for_backward", "scale", "torch"], "sew_d/modeling_sew_d.py:XDropout.backward": ["Function", "Model", "None", "autograd", "backward", "class", "ctx", "def", "else", "grad_output", "if", "mask", "masked_fill", "return", "saved_tensors", "scale", "torch"], "sew_d/modeling_sew_d.py:XDropout.symbolic": ["DropoutContext", "Function", "Graph", "Model", "True", "Value", "_C", "autograd", "class", "def", "dropout", "dropout_p", "float", "g", "if", "input", "isinstance", "local_ctx", "return", "symbolic", "symbolic_opset12", "torch", "train"], "sew_d/modeling_sew_d.py:StableDropout.__init__": ["ModelDropout", "Module", "None", "__init__", "class", "context_stack", "count", "def", "drop_prob", "nn", "self", "super"], "sew_d/modeling_sew_d.py:StableDropout.forward": ["ModelDropout", "Module", "XDropout", "and", "apply", "class", "def", "drop_prob", "forward", "get_context", "if", "nn", "return", "self", "training", "x"], "sew_d/modeling_sew_d.py:StableDropout.clear_context": ["ModelDropout", "Module", "None", "class", "clear_context", "context_stack", "count", "def", "nn", "self"], "sew_d/modeling_sew_d.py:StableDropout.init_context": ["ModelDropout", "Module", "None", "True", "c", "class", "context_stack", "count", "def", "for", "if", "in", "init_context", "is", "nn", "reuse_mask", "scale", "self"], "sew_d/modeling_sew_d.py:StableDropout.get_context": ["DropoutContext", "ModelDropout", "Module", "None", "append", "class", "context_stack", "count", "ctx", "def", "drop_prob", "dropout", "else", "get_context", "if", "is", "len", "nn", "not", "return", "self"], "sew_d/modeling_sew_d.py:SEWDSelfOutput.__init__": ["Dropout", "LayerNorm", "Linear", "Model", "Module", "__init__", "activation_dropout", "class", "config", "def", "dense", "dropout", "hidden_size", "layer_norm_eps", "nn", "self", "super"], "sew_d/modeling_sew_d.py:SEWDSelfOutput.forward": ["LayerNorm", "Model", "Module", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self"], "sew_d/modeling_sew_d.py:DisentangledSelfAttention.__init__": ["False", "Linear", "ModelSelfAttention", "Module", "None", "StableDropout", "The", "True", "ValueError", "__init__", "_attention_head_size", "a", "activation_dropout", "all_head_size", "attention", "attention_dropout", "attention_head_size", "bias", "c2p", "class", "config", "def", "dropout", "else", "f", "getattr", "heads", "hidden", "hidden_size", "if", "in", "is", "key_proj", "max_position_embeddings", "max_relative_positions", "multiple", "nn", "not", "num_attention_heads", "number", "of", "p2c", "pos_att_type", "pos_dropout", "pos_ebd_size", "pos_key_proj", "pos_query_proj", "position_buckets", "query_proj", "raise", "relative_attention", "self", "share_att_key", "size", "super", "the", "value_proj"], "sew_d/modeling_sew_d.py:DisentangledSelfAttention.transpose_for_scores": ["ModelSelfAttention", "Module", "attention_heads", "class", "contiguous", "def", "new_x_shape", "nn", "permute", "return", "self", "size", "transpose_for_scores", "view", "x"], "sew_d/modeling_sew_d.py:DisentangledSelfAttention.forward": ["False", "ModelSelfAttention", "Model_attention_bias", "Module", "None", "XSoftmax", "apply", "attention_mask", "attention_probs", "attention_scores", "bmm", "c2p", "class", "context_layer", "contiguous", "def", "dropout", "dtype", "else", "float", "forward", "hidden_states", "if", "in", "is", "key_layer", "key_proj", "new_context_layer_shape", "nn", "not", "num_attention_heads", "output_attentions", "p2c", "permute", "pos_att_type", "pos_dropout", "query_layer", "query_proj", "query_states", "rel_att", "rel_embeddings", "relative_attention", "relative_pos", "return", "scale", "scale_factor", "self", "size", "sqrt", "tensor", "to", "torch", "transpose", "transpose_for_scores", "value_layer", "value_proj", "view"], "sew_d/modeling_sew_d.py:DisentangledSelfAttention.disentangled_attention_bias": ["ModelSelfAttention", "Model_attention_bias", "Module", "None", "Relative", "ValueError", "att_span", "be", "bmm", "bucket_size", "build_relative_position", "c2p", "c2p_att", "c2p_pos", "clamp", "class", "def", "device", "dim", "dtype", "elif", "else", "expand", "f", "float", "gather", "ids", "if", "in", "index", "is", "key_layer", "key_proj", "long", "max_position", "max_relative_positions", "must", "nn", "num_attention_heads", "of", "or", "p2c", "p2c_att", "p2c_pos", "pos_att_type", "pos_ebd_size", "pos_key_layer", "pos_key_proj", "pos_query_layer", "pos_query_proj", "position", "position_buckets", "q", "query_layer", "query_proj", "r_pos", "raise", "rel_embeddings", "relative_pos", "repeat", "return", "scale", "scale_factor", "score", "self", "share_att_key", "size", "sqrt", "squeeze", "tensor", "to", "torch", "transpose", "transpose_for_scores", "unsqueeze"], "sew_d/modeling_sew_d.py:SEWDAttention.__init__": ["DisentangledSelfAttention", "Model", "ModelSelfOutput", "Module", "__init__", "class", "config", "def", "nn", "output", "self", "super"], "sew_d/modeling_sew_d.py:SEWDAttention.forward": ["False", "Model", "Module", "None", "att_matrix", "attention_mask", "attention_output", "class", "def", "else", "forward", "hidden_states", "if", "is", "nn", "output", "output_attentions", "query_states", "rel_embeddings", "relative_pos", "return", "self", "self_output"], "sew_d/modeling_sew_d.py:SEWDIntermediate.__init__": ["ACT2FN", "Linear", "Model", "Module", "__init__", "class", "config", "def", "dense", "else", "hidden_act", "hidden_size", "if", "intermediate_act_fn", "intermediate_size", "isinstance", "nn", "self", "str", "super"], "sew_d/modeling_sew_d.py:SEWDIntermediate.forward": ["Model", "Module", "Tensor", "class", "def", "dense", "forward", "hidden_states", "intermediate_act_fn", "nn", "return", "self", "torch"], "sew_d/modeling_sew_d.py:SEWDOutput.__init__": ["Dropout", "LayerNorm", "Linear", "Model", "Module", "__init__", "activation_dropout", "class", "config", "def", "dense", "dropout", "hidden_size", "intermediate_size", "layer_norm_eps", "nn", "self", "super"], "sew_d/modeling_sew_d.py:SEWDOutput.forward": ["LayerNorm", "Model", "Module", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self"], "sew_d/modeling_sew_d.py:SEWDLayer.__init__": ["GradientCheckpointingLayer", "Model", "ModelAttention", "ModelIntermediate", "ModelOutput", "__init__", "attention", "class", "config", "def", "intermediate", "output", "self", "super"], "sew_d/modeling_sew_d.py:SEWDLayer.forward": ["False", "GradientCheckpointingLayer", "Model", "None", "att_matrix", "attention", "attention_mask", "attention_output", "class", "def", "else", "forward", "hidden_states", "if", "intermediate", "intermediate_output", "layer_output", "output", "output_attentions", "query_states", "rel_embeddings", "relative_pos", "return", "self"], "sew_d/modeling_sew_d.py:ConvLayer.__init__": ["LayerNorm", "Model", "Model1d", "ModelLayer", "Model_act", "Model_groups", "Model_kernel_size", "Module", "StableDropout", "__init__", "class", "config", "def", "dropout", "getattr", "groups", "hidden_dropout_prob", "hidden_size", "kernel_size", "layer_norm_eps", "nn", "padding", "self", "super", "tanh"], "sew_d/modeling_sew_d.py:ConvLayer.forward": ["ACT2FN", "LayerNorm", "Model", "ModelLayer", "Model_act", "Module", "None", "bool", "class", "contiguous", "def", "dim", "dropout", "dtype", "else", "expand", "forward", "hidden_states", "if", "input_mask", "is", "layer_norm_input", "masked_fill_", "nn", "out", "output", "output_states", "permute", "residual_states", "return", "rmask", "self", "size", "squeeze", "to", "unsqueeze"], "sew_d/modeling_sew_d.py:SEWDTransformerEncoder.__init__": ["ConvLayer", "Embedding", "False", "LayerNorm", "Model", "ModelLayer", "Module", "ModuleList", "None", "True", "_", "__init__", "class", "config", "conv", "conv_kernel_size", "def", "elementwise_affine", "else", "for", "getattr", "gradient_checkpointing", "hidden_size", "if", "in", "layer", "layer_norm", "layer_norm_eps", "lower", "max_position_embeddings", "max_relative_positions", "nn", "none", "norm_rel_ebd", "num_hidden_layers", "pos_ebd_size", "position_buckets", "range", "rel_embeddings", "relative_attention", "self", "split", "strip", "super", "x"], "sew_d/modeling_sew_d.py:SEWDTransformerEncoder.get_rel_embedding": ["LayerNorm", "Model", "Module", "None", "and", "class", "def", "else", "get_rel_embedding", "if", "in", "is", "layer_norm", "nn", "norm_rel_ebd", "not", "rel_embeddings", "relative_attention", "return", "self", "weight"], "sew_d/modeling_sew_d.py:SEWDTransformerEncoder.get_attention_mask": ["Model", "Module", "attention_mask", "class", "def", "dim", "elif", "extended_attention_mask", "get_attention_mask", "if", "nn", "return", "self", "squeeze", "unsqueeze"], "sew_d/modeling_sew_d.py:SEWDTransformerEncoder.get_rel_pos": ["Model", "Module", "None", "and", "bucket_size", "build_relative_position", "class", "def", "device", "else", "get_rel_pos", "hidden_states", "if", "is", "max_position", "max_relative_positions", "nn", "not", "position_buckets", "q", "query_states", "relative_attention", "relative_pos", "return", "self", "size"], "sew_d/modeling_sew_d.py:SEWDTransformerEncoder.forward": ["BaseModelOutput", "False", "Model", "Module", "None", "Sequence", "True", "all_attentions", "all_hidden_states", "and", "att_m", "attention_mask", "attentions", "class", "conv", "def", "dim", "else", "enumerate", "for", "forward", "get_attention_mask", "get_rel_embedding", "get_rel_pos", "hidden_states", "i", "if", "in", "input_mask", "is", "isinstance", "last_hidden_state", "layer", "layer_module", "len", "next_kv", "nn", "not", "output_attentions", "output_hidden_states", "output_states", "query_states", "rel_embeddings", "relative_pos", "return", "return_dict", "self", "sum", "tuple", "v"], "sew_d/modeling_sew_d.py:SEWDEncoder.__init__": ["AvgPool1d", "False", "Model", "ModelPositionalConvEmbedding", "ModelTransformerEncoder", "ModelUpsampling", "Module", "__init__", "class", "config", "def", "encoder", "gradient_checkpointing", "nn", "pool", "pos_conv_embed", "self", "squeeze_factor", "super", "upsample"], "sew_d/modeling_sew_d.py:SEWDEncoder.forward": ["BaseModelOutput", "False", "Model", "Module", "None", "Tensor", "True", "arange", "attention_ids", "attention_mask", "attentions", "bool", "class", "config", "def", "device", "dtype", "else", "encoder", "encoder_outputs", "expand", "expand_attention_mask", "for", "forward", "functional", "hidden_states", "if", "in", "input_lengths", "is", "last_hidden_state", "long", "max_encoder_length", "min", "min_length", "n_input_timesteps", "nn", "not", "ones", "output_attentions", "output_hidden_states", "output_lengths", "pad", "pool", "pooled_hidden_states", "pos_conv_embed", "position_embeddings", "repeat", "return", "return_dict", "self", "shape", "size", "squeeze_factor", "sum", "tensor", "torch", "transpose", "tuple", "unsqueeze", "upsample", "v", "view"], "sew_d/modeling_sew_d.py:SEWDPreTrainedModel._init_weights": ["Conv1d", "Embedding", "False", "GatheredParameters", "GroupNorm", "LayerNorm", "Linear", "Model", "ModelPositionalConvEmbedding", "None", "PreTrainedModel", "_init_weights", "_is_hf_initialized", "and", "bias", "class", "config", "constant_", "conv", "deepspeed", "def", "elif", "else", "getattr", "hasattr", "if", "in_channels", "init", "initializer_range", "is", "is_deepspeed_zero3_enabled", "isinstance", "kaiming_normal_", "kernel_size", "math", "mean", "modifier_rank", "module", "nn", "normal_", "not", "ones_", "padding_idx", "self", "sqrt", "std", "weight", "weight_g", "weight_v", "with", "zero", "zeros_"], "sew_d/modeling_sew_d.py:SEWDPreTrainedModel._get_feat_extract_output_lengths": ["LongTensor", "Model", "PreTrainedModel", "_conv_out_length", "_get_feat_extract_output_lengths", "class", "config", "conv_kernel", "conv_stride", "def", "div", "floor", "for", "in", "input_length", "input_lengths", "int", "kernel_size", "return", "rounding_mode", "self", "stride", "torch", "zip"], "sew_d/modeling_sew_d.py:SEWDPreTrainedModel._get_feature_vector_attention_mask": ["LongTensor", "Model", "PreTrainedModel", "_get_feat_extract_output_lengths", "_get_feature_vector_attention_mask", "arange", "attention_mask", "batch_size", "bool", "class", "cumsum", "def", "device", "dtype", "feature_vector_length", "flip", "int", "long", "output_lengths", "return", "self", "shape", "sum", "to", "torch", "zeros"], "sew_d/modeling_sew_d.py:SEWDModel.__init__": ["Dropout", "LayerNorm", "Linear", "Model", "ModelConfig", "ModelEncoder", "ModelFeatureEncoder", "ModelPreTrainedModel", "Parameter", "Tensor", "__init__", "class", "config", "conv_dim", "def", "encoder", "eps", "feat_proj_dropout", "feature_dropout", "feature_extractor", "feature_layer_norm_eps", "feature_projection", "hidden_size", "if", "layer_norm", "mask_feature_prob", "mask_time_prob", "masked_spec_embed", "nn", "or", "post_init", "project_features", "self", "super", "torch", "uniform_"], "sew_d/modeling_sew_d.py:SEWDModel._mask_hidden_states": ["FloatTensor", "LongTensor", "Model", "ModelPreTrainedModel", "None", "True", "_compute_mask_indices", "_mask_hidden_states", "and", "apply_spec_augment", "attention_mask", "batch_size", "bool", "class", "config", "def", "device", "dtype", "elif", "expand", "getattr", "hidden_size", "hidden_states", "if", "is", "mask_feature_indices", "mask_feature_length", "mask_feature_min_masks", "mask_feature_prob", "mask_length", "mask_prob", "mask_time_indices", "mask_time_length", "mask_time_min_masks", "mask_time_prob", "masked_spec_embed", "min_masks", "not", "return", "self", "sequence_length", "size", "tensor", "to", "torch", "training"], "sew_d/modeling_sew_d.py:SEWDModel.forward": ["BaseModelOutput", "FloatTensor", "Model", "ModelPreTrainedModel", "None", "Tensor", "_get_feature_vector_attention_mask", "_mask_hidden_states", "attention_mask", "attentions", "bool", "class", "config", "def", "else", "encoder", "encoder_outputs", "extract_features", "feature_dropout", "feature_extractor", "feature_projection", "forward", "hidden_states", "if", "input_values", "is", "kwargs", "last_hidden_state", "layer_norm", "mask_time_indices", "not", "output_attentions", "output_hidden_states", "project_features", "r", "return", "return_dict", "self", "shape", "torch", "transpose", "tuple", "use_return_dict"], "sew_d/modeling_sew_d.py:SEWDForCTC.__init__": ["Dropout", "Linear", "Model", "ModelModel", "ModelPreTrainedModel", "None", "Please", "ValueError", "You", "__class__", "__init__", "a", "add_adapter", "and", "are", "as", "class", "config", "configuration", "def", "define", "does", "dropout", "else", "f", "final_dropout", "follows", "from_pretrained", "hasattr", "head", "hidden_size", "if", "instantiate", "is", "language", "lm_head", "model", "nn", "not", "of", "or", "output_hidden_size", "post_init", "r", "raise", "s", "self", "size", "str", "super", "target_lang", "that", "the", "to", "trying", "vocab_size", "vocabulary", "with", "your"], "sew_d/modeling_sew_d.py:SEWDForCTC.tie_weights": ["By", "Cannot", "Model", "ModelPreTrainedModel", "None", "True", "ValueError", "adapter_attn_dim", "and", "class", "config", "def", "default", "defined", "elif", "eng", "f", "force_load", "getattr", "if", "info", "is", "kwargs", "load_adapter", "logger", "not", "pass", "raise", "self", "set", "target_lang", "tie_weights", "to"], "sew_d/modeling_sew_d.py:SEWDForCTC.freeze_feature_encoder": ["Model", "ModelPreTrainedModel", "_freeze_parameters", "class", "def", "feature_extractor", "freeze_feature_encoder", "self"], "sew_d/modeling_sew_d.py:SEWDForCTC.freeze_base_model": ["False", "Model", "ModelPreTrainedModel", "class", "def", "for", "freeze_base_model", "in", "param", "parameters", "requires_grad", "self"], "sew_d/modeling_sew_d.py:SEWDForCTC.forward": ["CausalLMOutput", "False", "Label", "Model", "ModelPreTrainedModel", "None", "Tensor", "ValueError", "_HIDDEN_STATES_START_POSITION", "_get_feat_extract_output_lengths", "and", "attention_mask", "attentions", "backends", "be", "blank", "bool", "class", "config", "ctc_loss", "ctc_loss_reduction", "ctc_zero_infinity", "cudnn", "def", "dim", "dropout", "dtype", "else", "enabled", "f", "flags", "flattened_targets", "float32", "forward", "functional", "hidden_states", "if", "input_lengths", "input_values", "is", "kwargs", "labels", "labels_mask", "lm_head", "log_probs", "log_softmax", "logits", "long", "loss", "masked_select", "max", "must", "nn", "not", "ones_like", "output", "output_attentions", "output_hidden_states", "outputs", "pad_token_id", "r", "raise", "reduction", "return", "return_dict", "self", "sum", "target_lengths", "to", "torch", "transpose", "tuple", "use_return_dict", "values", "vocab_size", "with", "zero_infinity"], "sew_d/modeling_sew_d.py:SEWDForSequenceClassification.__init__": ["Linear", "Model", "ModelModel", "ModelPreTrainedModel", "Parameter", "Sequence", "True", "ValueError", "__init__", "adapters", "add_adapter", "and", "class", "classification", "classifier", "classifier_proj_size", "config", "def", "does", "hasattr", "hidden_size", "if", "layer_weights", "nn", "not", "num_hidden_layers", "num_labels", "num_layers", "of", "ones", "post_init", "projector", "raise", "self", "super", "support", "the", "torch", "use", "use_weighted_layer_sum"], "sew_d/modeling_sew_d.py:SEWDForSequenceClassification.freeze_feature_encoder": ["Model", "ModelPreTrainedModel", "_freeze_parameters", "class", "def", "feature_extractor", "freeze_feature_encoder", "self"], "sew_d/modeling_sew_d.py:SEWDForSequenceClassification.freeze_base_model": ["False", "Model", "ModelPreTrainedModel", "class", "def", "for", "freeze_base_model", "in", "param", "parameters", "requires_grad", "self"], "sew_d/modeling_sew_d.py:SEWDForSequenceClassification.forward": ["CrossEntropyLoss", "Model", "ModelPreTrainedModel", "None", "SequenceClassifierOutput", "Tensor", "True", "_HIDDEN_STATES_START_POSITION", "_get_feature_vector_attention_mask", "attention_mask", "attentions", "bool", "class", "classifier", "config", "def", "dim", "else", "expand_padding_mask", "forward", "functional", "hidden_states", "if", "input_values", "is", "kwargs", "labels", "layer_weights", "logits", "loss", "loss_fct", "mean", "nn", "norm_weights", "not", "num_labels", "output", "output_attentions", "output_hidden_states", "outputs", "padding_mask", "pooled_output", "projector", "r", "repeat", "return", "return_dict", "self", "shape", "softmax", "stack", "sum", "torch", "tuple", "unsqueeze", "use_return_dict", "use_weighted_layer_sum", "view"], "shieldgemma2/modeling_shieldgemma2.py:ShieldGemma2ForImageClassification.__init__": ["AutoModelForImageTextToText", "ModelConfig", "ModelForImageClassification", "PreTrainedModel", "__init__", "class", "config", "def", "from_config", "getattr", "model", "no_token_index", "post_init", "self", "super", "yes_token_index"], "shieldgemma2/modeling_shieldgemma2.py:ShieldGemma2ForImageClassification.get_input_embeddings": ["ModelForImageClassification", "PreTrainedModel", "class", "def", "get_input_embeddings", "language_model", "model", "return", "self"], "shieldgemma2/modeling_shieldgemma2.py:ShieldGemma2ForImageClassification.set_input_embeddings": ["ModelForImageClassification", "PreTrainedModel", "class", "def", "language_model", "model", "self", "set_input_embeddings", "value"], "shieldgemma2/modeling_shieldgemma2.py:ShieldGemma2ForImageClassification.get_output_embeddings": ["ModelForImageClassification", "PreTrainedModel", "class", "def", "get_output_embeddings", "language_model", "model", "return", "self"], "shieldgemma2/modeling_shieldgemma2.py:ShieldGemma2ForImageClassification.set_output_embeddings": ["ModelForImageClassification", "PreTrainedModel", "class", "def", "language_model", "model", "new_embeddings", "self", "set_output_embeddings"], "shieldgemma2/modeling_shieldgemma2.py:ShieldGemma2ForImageClassification.forward": ["Cache", "FloatTensor", "LongTensor", "ModelForImageClassification", "ModelImageClassifierOutputWithNoAttention", "None", "PreTrainedModel", "Tensor", "attention_mask", "bool", "cache_position", "class", "def", "dim", "forward", "input_ids", "inputs_embeds", "int", "labels", "lm_kwargs", "logits", "logits_to_keep", "model", "no_token_index", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "pixel_values", "position_ids", "probabilities", "r", "return", "return_dict", "selected_logits", "self", "softmax", "token_type_ids", "torch", "use_cache", "yes_token_index"], "siglip/modeling_siglip.py:variance_scaling_": ["Model", "Model_scaling_", "ValueError", "_calculate_fan_in_and_fan_out", "bound", "def", "denom", "distribution", "elif", "else", "f", "fan_avg", "fan_in", "fan_out", "if", "init", "invalid", "math", "mode", "normal", "normal_", "raise", "sqrt", "std", "tensor", "trunc_normal_", "truncated_normal", "uniform", "uniform_"], "siglip/modeling_siglip.py:lecun_normal_": ["Model_normal_", "def", "distribution", "fan_in", "mode", "tensor", "truncated_normal", "variance_scaling_"], "siglip/modeling_siglip.py:default_flax_embed_init": ["Model_flax_embed_init", "def", "distribution", "fan_in", "mode", "normal", "tensor", "variance_scaling_"], "siglip/modeling_siglip.py:SiglipOutput.to_tuple": ["Any", "ModelOutput", "class", "def", "else", "for", "getattr", "if", "in", "k", "keys", "not", "return", "self", "text_model_output", "to_tuple", "tuple", "vision_model_output"], "siglip/modeling_siglip.py:SiglipVisionEmbeddings.__init__": ["Conv2d", "Embedding", "False", "ModelVisionConfig", "ModelVisionEmbeddings", "Module", "__init__", "arange", "class", "config", "def", "embed_dim", "expand", "hidden_size", "image_size", "in_channels", "kernel_size", "nn", "num_channels", "num_patches", "num_positions", "out_channels", "padding", "patch_embedding", "patch_size", "persistent", "position_embedding", "position_ids", "register_buffer", "self", "stride", "super", "torch", "valid"], "siglip/modeling_siglip.py:SiglipVisionEmbeddings.interpolate_pos_encoding": ["False", "ModelVisionEmbeddings", "Module", "Tensor", "align_corners", "and", "bicubic", "class", "def", "dim", "embeddings", "functional", "height", "if", "int", "interpolate", "interpolate_pos_encoding", "is_tracing", "jit", "mode", "new_height", "new_width", "nn", "not", "num_patches", "num_positions", "patch_pos_embed", "patch_size", "permute", "position_embedding", "position_ids", "reshape", "return", "self", "shape", "size", "sqrt_num_positions", "torch", "torch_int", "unsqueeze", "view", "weight", "width"], "siglip/modeling_siglip.py:SiglipVisionEmbeddings.forward": ["False", "FloatTensor", "ModelVisionEmbeddings", "Module", "Tensor", "_", "class", "def", "dtype", "else", "embeddings", "flatten", "forward", "height", "if", "interpolate_pos_encoding", "nn", "patch_embedding", "patch_embeds", "pixel_values", "position_embedding", "position_ids", "return", "self", "shape", "target_dtype", "to", "torch", "transpose", "weight", "width"], "siglip/modeling_siglip.py:SiglipTextEmbeddings.__init__": ["Embedding", "False", "ModelTextConfig", "ModelTextEmbeddings", "Module", "__init__", "arange", "class", "config", "def", "embed_dim", "expand", "hidden_size", "max_position_embeddings", "nn", "persistent", "position_embedding", "position_ids", "register_buffer", "self", "super", "token_embedding", "torch", "vocab_size"], "siglip/modeling_siglip.py:SiglipTextEmbeddings.forward": ["FloatTensor", "LongTensor", "ModelTextEmbeddings", "Module", "None", "Sequence", "Tensor", "ValueError", "and", "be", "class", "def", "else", "embeddings", "f", "forward", "got", "if", "input_ids", "inputs_embeds", "is", "length", "less", "max_position_embedding", "max_position_embeddings", "must", "nn", "not", "position_embedding", "position_embeddings", "position_ids", "raise", "return", "self", "seq_length", "sequence", "shape", "than", "token_embedding", "torch", "weight"], "siglip/modeling_siglip.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "attention_mask", "attn_output", "attn_weights", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "softmax", "to", "torch", "training", "transpose", "value"], "siglip/modeling_siglip.py:SiglipAttention.__init__": ["False", "Linear", "ModelAttention", "Module", "ValueError", "__init__", "and", "attention_dropout", "be", "by", "class", "config", "def", "divisible", "dropout", "embed_dim", "f", "got", "head_dim", "hidden_size", "if", "is_causal", "k_proj", "must", "nn", "num_attention_heads", "num_heads", "out_proj", "q_proj", "raise", "scale", "self", "super", "v_proj"], "siglip/modeling_siglip.py:SiglipAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "ModelAttention", "Module", "None", "Tensor", "_attn_implementation", "attention_interface", "attention_mask", "attn_output", "attn_weights", "batch_size", "class", "config", "contiguous", "def", "dropout", "eager", "eager_attention_forward", "else", "embed_dim", "forward", "head_dim", "hidden_states", "if", "is_causal", "k_proj", "keys", "kwargs", "nn", "not", "num_heads", "out_proj", "q_proj", "queries", "reshape", "return", "scale", "scaling", "self", "seq_length", "shape", "torch", "training", "transpose", "tuple", "v_proj", "values", "view"], "siglip/modeling_siglip.py:SiglipMLP.__init__": ["ACT2FN", "Linear", "ModelMLP", "Module", "__init__", "activation_fn", "class", "config", "def", "fc1", "fc2", "hidden_act", "hidden_size", "intermediate_size", "nn", "self", "super"], "siglip/modeling_siglip.py:SiglipMLP.forward": ["ModelMLP", "Module", "Tensor", "activation_fn", "class", "def", "fc1", "fc2", "forward", "hidden_states", "nn", "return", "self", "torch"], "siglip/modeling_siglip.py:SiglipEncoderLayer.__init__": ["GradientCheckpointingLayer", "LayerNorm", "ModelAttention", "ModelEncoderLayer", "ModelMLP", "ModelTextConfig", "ModelVisionConfig", "__init__", "class", "config", "def", "embed_dim", "eps", "hidden_size", "layer_norm1", "layer_norm2", "layer_norm_eps", "mlp", "nn", "self", "self_attn", "super"], "siglip/modeling_siglip.py:SiglipEncoderLayer.forward": ["FloatTensor", "GradientCheckpointingLayer", "ModelEncoderLayer", "Tensor", "TransformersKwargs", "Unpack", "_", "attention_mask", "class", "def", "forward", "hidden_states", "kwargs", "layer_norm1", "layer_norm2", "mlp", "residual", "return", "self", "self_attn", "torch"], "siglip/modeling_siglip.py:SiglipPreTrainedModel._init_weights": ["Conv2d", "Embedding", "LayerNorm", "Linear", "ModelAttention", "ModelConfig", "ModelForImageClassification", "ModelMLP", "ModelModel", "ModelMultiheadAttentionPoolingHead", "ModelPreTrainedModel", "ModelTextEmbeddings", "ModelVisionEmbeddings", "None", "PreTrainedModel", "_init_weights", "arange", "attention", "bias", "class", "classifier", "config", "copy_", "def", "default_flax_embed_init", "elif", "else", "expand", "fc1", "fc2", "hasattr", "hidden_size", "if", "in_proj_bias", "in_proj_weight", "init", "initializer_factor", "is", "isinstance", "k_proj", "lecun_normal_", "logit_bias", "logit_scale", "module", "nn", "normal_", "not", "np", "ones_", "out_proj", "position_embedding", "position_ids", "probe", "q_proj", "self", "shape", "sqrt", "std", "torch", "v_proj", "vision_config", "weight", "width", "xavier_uniform_", "zeros_"], "siglip/modeling_siglip.py:SiglipEncoder.__init__": ["False", "ModelConfig", "ModelEncoder", "ModelEncoderLayer", "Module", "ModuleList", "_", "__init__", "class", "config", "def", "for", "gradient_checkpointing", "in", "layers", "nn", "num_hidden_layers", "range", "self", "super"], "siglip/modeling_siglip.py:SiglipEncoder.forward": ["BaseModelOutput", "ModelEncoder", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "class", "def", "encoder_layer", "for", "forward", "hidden_states", "in", "inputs_embeds", "kwargs", "last_hidden_state", "layers", "nn", "return", "self", "torch"], "siglip/modeling_siglip.py:SiglipTextTransformer.__init__": ["LayerNorm", "Linear", "ModelEncoder", "ModelPreTrainedModel", "ModelTextConfig", "ModelTextEmbeddings", "ModelTextTransformer", "__init__", "class", "config", "def", "embed_dim", "embeddings", "encoder", "eps", "final_layer_norm", "head", "hidden_size", "layer_norm_eps", "nn", "post_init", "projection_size", "self", "super"], "siglip/modeling_siglip.py:SiglipTextTransformer.forward": ["BaseModelOutput", "BaseModelOutputWithPooling", "ModelPreTrainedModel", "ModelTextTransformer", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "_attn_implementation", "_prepare_4d_attention_mask", "and", "attention_mask", "class", "config", "def", "dtype", "elif", "embeddings", "encoder", "encoder_outputs", "final_layer_norm", "flash", "forward", "have", "head", "hidden_states", "if", "in", "input_ids", "input_shape", "inputs_embeds", "is", "kwargs", "last_hidden_state", "not", "pooled_output", "pooler_output", "position_ids", "raise", "return", "self", "size", "specify", "to", "torch", "uses_flash_attention", "view"], "siglip/modeling_siglip.py:SiglipTextModel.__init__": ["ModelPreTrainedModel", "ModelTextConfig", "ModelTextModel", "ModelTextTransformer", "__init__", "class", "config", "def", "post_init", "self", "super", "text_model"], "siglip/modeling_siglip.py:SiglipTextModel.get_input_embeddings": ["ModelPreTrainedModel", "ModelTextModel", "Module", "class", "def", "embeddings", "get_input_embeddings", "nn", "return", "self", "text_model", "token_embedding"], "siglip/modeling_siglip.py:SiglipTextModel.set_input_embeddings": ["ModelPreTrainedModel", "ModelTextModel", "class", "def", "embeddings", "self", "set_input_embeddings", "text_model", "token_embedding", "value"], "siglip/modeling_siglip.py:SiglipTextModel.forward": ["BaseModelOutputWithPooling", "ModelPreTrainedModel", "ModelTextModel", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "class", "def", "forward", "input_ids", "kwargs", "position_ids", "r", "return", "self", "text_model", "torch"], "siglip/modeling_siglip.py:SiglipVisionTransformer.__init__": ["LayerNorm", "ModelEncoder", "ModelMultiheadAttentionPoolingHead", "ModelPreTrainedModel", "ModelVisionConfig", "ModelVisionEmbeddings", "ModelVisionTransformer", "True", "__init__", "class", "config", "def", "else", "embed_dim", "embeddings", "encoder", "eps", "hasattr", "head", "hidden_size", "if", "layer_norm_eps", "nn", "not", "post_init", "post_layernorm", "self", "super", "use_head", "vision_use_head"], "siglip/modeling_siglip.py:SiglipVisionTransformer.forward": ["BaseModelOutput", "BaseModelOutputWithPooling", "False", "ModelPreTrainedModel", "ModelVisionTransformer", "None", "TransformersKwargs", "Unpack", "bool", "class", "def", "else", "embeddings", "encoder", "encoder_outputs", "forward", "head", "hidden_states", "if", "inputs_embeds", "interpolate_pos_encoding", "kwargs", "last_hidden_state", "pixel_values", "pooler_output", "post_layernorm", "return", "self", "use_head"], "siglip/modeling_siglip.py:SiglipMultiheadAttentionPoolingHead.__init__": ["LayerNorm", "ModelMLP", "ModelMultiheadAttentionPoolingHead", "ModelVisionConfig", "Module", "MultiheadAttention", "Parameter", "True", "__init__", "attention", "batch_first", "class", "config", "def", "eps", "hidden_size", "layer_norm_eps", "layernorm", "mlp", "nn", "num_attention_heads", "probe", "randn", "self", "super", "torch"], "siglip/modeling_siglip.py:SiglipMultiheadAttentionPoolingHead.forward": ["ModelMultiheadAttentionPoolingHead", "Module", "attention", "batch_size", "class", "def", "forward", "hidden_state", "layernorm", "mlp", "nn", "probe", "repeat", "residual", "return", "self", "shape"], "siglip/modeling_siglip.py:SiglipVisionModel.__init__": ["ModelPreTrainedModel", "ModelVisionConfig", "ModelVisionModel", "ModelVisionTransformer", "__init__", "class", "config", "def", "post_init", "self", "super", "vision_model"], "siglip/modeling_siglip.py:SiglipVisionModel.get_input_embeddings": ["ModelPreTrainedModel", "ModelVisionModel", "Module", "class", "def", "embeddings", "get_input_embeddings", "nn", "patch_embedding", "return", "self", "vision_model"], "siglip/modeling_siglip.py:SiglipVisionModel.forward": ["BaseModelOutputWithPooling", "False", "ModelPreTrainedModel", "ModelVisionModel", "TransformersKwargs", "Unpack", "bool", "class", "def", "forward", "interpolate_pos_encoding", "kwargs", "pixel_values", "r", "return", "self", "vision_model"], "siglip/modeling_siglip.py:SiglipModel.__init__": ["ModelConfig", "ModelModel", "ModelPreTrainedModel", "ModelTextConfig", "ModelTextModel", "ModelVisionConfig", "ModelVisionModel", "Parameter", "TypeError", "__init__", "_from_config", "be", "but", "class", "config", "def", "expected", "f", "if", "is", "isinstance", "logit_bias", "logit_scale", "nn", "not", "of", "post_init", "raise", "randn", "self", "super", "text_config", "text_model", "to", "torch", "type", "vision_config", "vision_model"], "siglip/modeling_siglip.py:SiglipModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "Module", "class", "def", "embeddings", "get_input_embeddings", "nn", "return", "self", "text_model", "token_embedding"], "siglip/modeling_siglip.py:SiglipModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "Module", "class", "def", "embeddings", "nn", "self", "set_input_embeddings", "text_model", "token_embedding", "value"], "siglip/modeling_siglip.py:SiglipModel.get_text_features": ["BaseModelOutputWithPooling", "FloatTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "attention_mask", "class", "def", "get_text_features", "input_ids", "pooled_output", "pooler_output", "position_ids", "r", "return", "self", "text_model", "text_outputs", "torch"], "siglip/modeling_siglip.py:SiglipModel.get_image_features": ["BaseModelOutputWithPooling", "False", "FloatTensor", "ModelModel", "ModelPreTrainedModel", "TransformersKwargs", "Unpack", "bool", "class", "def", "get_image_features", "interpolate_pos_encoding", "kwargs", "pixel_values", "pooled_output", "pooler_output", "r", "return", "self", "torch", "vision_model", "vision_outputs"], "siglip/modeling_siglip.py:SiglipModel.forward": ["BaseModelOutputWithPooling", "False", "FloatTensor", "LongTensor", "ModelModel", "ModelOutput", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "attention_mask", "bool", "class", "def", "device", "dim", "exp", "eye", "forward", "functional", "if", "image_embeds", "input_ids", "interpolate_pos_encoding", "keepdim", "kwargs", "logit_bias", "logit_scale", "logits_per_image", "logits_per_text", "loglik", "logsigmoid", "loss", "m1_diag1", "matmul", "mean", "nll", "nn", "norm", "ones_like", "p", "pixel_values", "pooler_output", "position_ids", "r", "return", "return_loss", "self", "size", "sum", "t", "text_embeds", "text_model", "text_model_output", "text_outputs", "to", "torch", "vision_model", "vision_model_output", "vision_outputs"], "siglip/modeling_siglip.py:SiglipForImageClassification.__init__": ["Identity", "Linear", "ModelConfig", "ModelForImageClassification", "ModelPreTrainedModel", "ModelVisionModel", "None", "__init__", "_from_config", "class", "classifier", "config", "def", "else", "hidden_size", "if", "nn", "num_labels", "post_init", "self", "super", "vision_config", "vision_model"], "siglip/modeling_siglip.py:SiglipForImageClassification.get_input_embeddings": ["ModelForImageClassification", "ModelPreTrainedModel", "Module", "class", "def", "embeddings", "get_input_embeddings", "nn", "patch_embedding", "return", "self", "vision_model"], "siglip/modeling_siglip.py:SiglipForImageClassification.set_input_embeddings": ["ModelForImageClassification", "ModelPreTrainedModel", "Module", "class", "def", "embeddings", "nn", "patch_embedding", "self", "set_input_embeddings", "value", "vision_model"], "siglip/modeling_siglip.py:SiglipForImageClassification.forward": ["BaseModelOutputWithPooling", "False", "ImageClassifierOutput", "ModelForImageClassification", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "bool", "class", "classifier", "config", "def", "dim", "forward", "if", "interpolate_pos_encoding", "is", "kwargs", "labels", "last_hidden_state", "logits", "loss", "loss_function", "mean", "not", "outputs", "pixel_values", "r", "return", "self", "sequence_output", "torch", "vision_model"], "siglip2/modeling_siglip2.py:Siglip2Output.to_tuple": ["Any", "ModelOutput", "class", "def", "else", "for", "getattr", "if", "in", "k", "keys", "not", "return", "self", "text_model_output", "to_tuple", "tuple", "vision_model_output"], "siglip2/modeling_siglip2.py:Siglip2VisionEmbeddings.__init__": ["Embedding", "Linear", "ModelVisionConfig", "ModelVisionEmbeddings", "Module", "__init__", "class", "config", "def", "embed_dim", "hidden_size", "in_features", "int", "nn", "num_channels", "num_patches", "out_features", "patch_embedding", "patch_size", "position_embedding", "position_embedding_size", "self", "super"], "siglip2/modeling_siglip2.py:Siglip2VisionEmbeddings.resize_positional_embeddings": ["F", "False", "LongTensor", "ModelVisionEmbeddings", "Module", "Tensor", "True", "align_corners", "antialias", "batch_size", "bilinear", "class", "cpu", "def", "device", "dtype", "embed_dim", "empty", "float32", "for", "height", "i", "if", "in", "int", "interpolate", "max_length", "mode", "nn", "permute", "positional_embeddings", "range", "reshape", "resize_positional_embeddings", "resized_embeddings", "resulted_positional_embeddings", "return", "shape", "size", "source_dtype", "spatial_shapes", "to", "torch", "transpose", "type", "unsqueeze", "width"], "siglip2/modeling_siglip2.py:Siglip2VisionEmbeddings.forward": ["FloatTensor", "LongTensor", "ModelVisionEmbeddings", "Module", "Tensor", "class", "def", "dtype", "embeddings", "forward", "max_length", "nn", "patch_embedding", "patch_embeds", "pixel_values", "position_embedding", "position_embedding_size", "positional_embeddings", "reshape", "resize_positional_embeddings", "resized_positional_embeddings", "return", "self", "shape", "spatial_shapes", "target_dtype", "to", "torch", "weight"], "siglip2/modeling_siglip2.py:Siglip2TextEmbeddings.__init__": ["Embedding", "False", "ModelTextConfig", "ModelTextEmbeddings", "Module", "__init__", "arange", "class", "config", "def", "embed_dim", "expand", "hidden_size", "max_position_embeddings", "nn", "persistent", "position_embedding", "position_ids", "register_buffer", "self", "super", "token_embedding", "torch", "vocab_size"], "siglip2/modeling_siglip2.py:Siglip2TextEmbeddings.forward": ["FloatTensor", "LongTensor", "ModelTextEmbeddings", "Module", "None", "Sequence", "Tensor", "ValueError", "and", "be", "class", "def", "else", "embeddings", "f", "forward", "got", "if", "input_ids", "inputs_embeds", "is", "length", "less", "max_position_embedding", "max_position_embeddings", "must", "nn", "not", "position_embedding", "position_embeddings", "position_ids", "raise", "return", "self", "seq_length", "sequence", "shape", "than", "token_embedding", "torch", "weight"], "siglip2/modeling_siglip2.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "attention_mask", "attn_output", "attn_weights", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "softmax", "to", "torch", "training", "transpose", "value"], "siglip2/modeling_siglip2.py:Siglip2Attention.__init__": ["False", "Linear", "ModelAttention", "Module", "ValueError", "__init__", "and", "attention_dropout", "be", "by", "class", "config", "def", "divisible", "dropout", "embed_dim", "f", "got", "head_dim", "hidden_size", "if", "is_causal", "k_proj", "must", "nn", "num_attention_heads", "num_heads", "out_proj", "q_proj", "raise", "scale", "self", "super", "v_proj"], "siglip2/modeling_siglip2.py:Siglip2Attention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "ModelAttention", "Module", "None", "Tensor", "_attn_implementation", "attention_interface", "attention_mask", "attn_output", "attn_weights", "batch_size", "class", "config", "contiguous", "def", "dropout", "eager", "eager_attention_forward", "else", "embed_dim", "forward", "head_dim", "hidden_states", "if", "is_causal", "k_proj", "keys", "kwargs", "nn", "not", "num_heads", "out_proj", "q_proj", "queries", "reshape", "return", "scale", "scaling", "self", "seq_length", "shape", "torch", "training", "transpose", "tuple", "v_proj", "values", "view"], "siglip2/modeling_siglip2.py:Siglip2MLP.__init__": ["ACT2FN", "Linear", "ModelMLP", "Module", "__init__", "activation_fn", "class", "config", "def", "fc1", "fc2", "hidden_act", "hidden_size", "intermediate_size", "nn", "self", "super"], "siglip2/modeling_siglip2.py:Siglip2MLP.forward": ["ModelMLP", "Module", "Tensor", "activation_fn", "class", "def", "fc1", "fc2", "forward", "hidden_states", "nn", "return", "self", "torch"], "siglip2/modeling_siglip2.py:Siglip2EncoderLayer.__init__": ["GradientCheckpointingLayer", "LayerNorm", "ModelAttention", "ModelEncoderLayer", "ModelMLP", "ModelTextConfig", "ModelVisionConfig", "__init__", "class", "config", "def", "embed_dim", "eps", "hidden_size", "layer_norm1", "layer_norm2", "layer_norm_eps", "mlp", "nn", "self", "self_attn", "super"], "siglip2/modeling_siglip2.py:Siglip2EncoderLayer.forward": ["FloatTensor", "GradientCheckpointingLayer", "ModelEncoderLayer", "Tensor", "TransformersKwargs", "Unpack", "_", "attention_mask", "class", "def", "forward", "hidden_states", "kwargs", "layer_norm1", "layer_norm2", "mlp", "residual", "return", "self", "self_attn", "torch"], "siglip2/modeling_siglip2.py:variance_scaling_": ["Model", "Model_scaling_", "ValueError", "_calculate_fan_in_and_fan_out", "bound", "def", "denom", "distribution", "elif", "else", "f", "fan_avg", "fan_in", "fan_out", "if", "init", "invalid", "math", "mode", "normal", "normal_", "raise", "sqrt", "std", "tensor", "trunc_normal_", "truncated_normal", "uniform", "uniform_"], "siglip2/modeling_siglip2.py:lecun_normal_": ["Model_normal_", "def", "distribution", "fan_in", "mode", "tensor", "truncated_normal", "variance_scaling_"], "siglip2/modeling_siglip2.py:default_flax_embed_init": ["Model_flax_embed_init", "def", "distribution", "fan_in", "mode", "normal", "tensor", "variance_scaling_"], "siglip2/modeling_siglip2.py:Siglip2PreTrainedModel._init_weights": ["Conv2d", "Embedding", "LayerNorm", "Linear", "ModelAttention", "ModelConfig", "ModelForImageClassification", "ModelMLP", "ModelModel", "ModelMultiheadAttentionPoolingHead", "ModelPreTrainedModel", "ModelTextEmbeddings", "ModelVisionEmbeddings", "None", "PreTrainedModel", "_init_weights", "arange", "attention", "bias", "class", "classifier", "config", "copy_", "def", "default_flax_embed_init", "elif", "else", "expand", "fc1", "fc2", "hasattr", "hidden_size", "if", "in_proj_bias", "in_proj_weight", "init", "initializer_factor", "is", "isinstance", "k_proj", "lecun_normal_", "logit_bias", "logit_scale", "module", "nn", "normal_", "not", "np", "ones_", "out_proj", "position_embedding", "position_ids", "probe", "q_proj", "self", "shape", "sqrt", "std", "torch", "v_proj", "vision_config", "weight", "width", "xavier_uniform_", "zeros_"], "siglip2/modeling_siglip2.py:Siglip2Encoder.__init__": ["False", "ModelConfig", "ModelEncoder", "ModelEncoderLayer", "Module", "ModuleList", "_", "__init__", "class", "config", "def", "for", "gradient_checkpointing", "in", "layers", "nn", "num_hidden_layers", "range", "self", "super"], "siglip2/modeling_siglip2.py:Siglip2Encoder.forward": ["BaseModelOutput", "ModelEncoder", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "class", "def", "encoder_layer", "for", "forward", "hidden_states", "in", "inputs_embeds", "kwargs", "last_hidden_state", "layers", "nn", "return", "self", "torch"], "siglip2/modeling_siglip2.py:Siglip2VisionTransformer.__init__": ["LayerNorm", "ModelEncoder", "ModelMultiheadAttentionPoolingHead", "ModelPreTrainedModel", "ModelVisionConfig", "ModelVisionEmbeddings", "ModelVisionTransformer", "True", "__init__", "class", "config", "def", "else", "embed_dim", "embeddings", "encoder", "eps", "hasattr", "head", "hidden_size", "if", "layer_norm_eps", "nn", "not", "post_init", "post_layernorm", "self", "super", "use_head", "vision_use_head"], "siglip2/modeling_siglip2.py:Siglip2VisionTransformer.forward": ["BaseModelOutput", "BaseModelOutputWithPooling", "FloatTensor", "LongTensor", "ModelPreTrainedModel", "ModelVisionTransformer", "None", "Tensor", "_attn_implementation", "_prepare_4d_attention_mask", "and", "attention_mask", "attentions", "bool", "class", "config", "def", "dtype", "else", "embeddings", "encoder", "encoder_attention_mask", "encoder_outputs", "flash_attention_2", "forward", "head", "hidden_states", "if", "inputs_embeds", "is", "kwargs", "last_hidden_state", "not", "output_attentions", "output_hidden_states", "pixel_values", "pooler_output", "post_layernorm", "r", "return", "self", "spatial_shapes", "torch", "use_head"], "siglip2/modeling_siglip2.py:Siglip2TextTransformer.__init__": ["LayerNorm", "Linear", "ModelEncoder", "ModelPreTrainedModel", "ModelTextConfig", "ModelTextEmbeddings", "ModelTextTransformer", "__init__", "class", "config", "def", "embed_dim", "embeddings", "encoder", "eps", "final_layer_norm", "head", "hidden_size", "layer_norm_eps", "nn", "post_init", "projection_size", "self", "super"], "siglip2/modeling_siglip2.py:Siglip2TextTransformer.forward": ["BaseModelOutput", "BaseModelOutputWithPooling", "ModelPreTrainedModel", "ModelTextTransformer", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "_attn_implementation", "_prepare_4d_attention_mask", "and", "attention_mask", "class", "config", "def", "dtype", "elif", "embeddings", "encoder", "encoder_outputs", "final_layer_norm", "flash", "forward", "have", "head", "hidden_states", "if", "in", "input_ids", "input_shape", "inputs_embeds", "is", "kwargs", "last_hidden_state", "not", "pooled_output", "pooler_output", "position_ids", "raise", "return", "self", "size", "specify", "to", "torch", "uses_flash_attention", "view"], "siglip2/modeling_siglip2.py:Siglip2TextModel.__init__": ["ModelPreTrainedModel", "ModelTextConfig", "ModelTextModel", "ModelTextTransformer", "__init__", "class", "config", "def", "post_init", "self", "super", "text_model"], "siglip2/modeling_siglip2.py:Siglip2TextModel.get_input_embeddings": ["ModelPreTrainedModel", "ModelTextModel", "Module", "class", "def", "embeddings", "get_input_embeddings", "nn", "return", "self", "text_model", "token_embedding"], "siglip2/modeling_siglip2.py:Siglip2TextModel.set_input_embeddings": ["ModelPreTrainedModel", "ModelTextModel", "class", "def", "embeddings", "self", "set_input_embeddings", "text_model", "token_embedding", "value"], "siglip2/modeling_siglip2.py:Siglip2TextModel.forward": ["BaseModelOutputWithPooling", "ModelPreTrainedModel", "ModelTextModel", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "class", "def", "forward", "input_ids", "kwargs", "position_ids", "r", "return", "self", "text_model", "torch"], "siglip2/modeling_siglip2.py:Siglip2MultiheadAttentionPoolingHead.__init__": ["LayerNorm", "ModelMLP", "ModelMultiheadAttentionPoolingHead", "ModelVisionConfig", "Module", "MultiheadAttention", "Parameter", "True", "__init__", "attention", "batch_first", "class", "config", "def", "eps", "hidden_size", "layer_norm_eps", "layernorm", "mlp", "nn", "num_attention_heads", "num_heads", "probe", "randn", "self", "super", "torch"], "siglip2/modeling_siglip2.py:Siglip2MultiheadAttentionPoolingHead.forward": ["ModelMultiheadAttentionPoolingHead", "Module", "None", "Tensor", "_prepare_4d_attention_mask", "attention", "attention_mask", "attn_mask", "batch_size", "class", "def", "dtype", "forward", "hidden_state", "if", "is", "layernorm", "mlp", "nn", "not", "num_heads", "probe", "repeat", "reshape", "residual", "return", "self", "shape", "source_len", "target_len", "torch"], "siglip2/modeling_siglip2.py:Siglip2VisionModel.__init__": ["ModelPreTrainedModel", "ModelVisionConfig", "ModelVisionModel", "ModelVisionTransformer", "__init__", "class", "config", "def", "post_init", "self", "super", "vision_model"], "siglip2/modeling_siglip2.py:Siglip2VisionModel.get_input_embeddings": ["ModelPreTrainedModel", "ModelVisionModel", "Module", "class", "def", "embeddings", "get_input_embeddings", "nn", "patch_embedding", "return", "self", "vision_model"], "siglip2/modeling_siglip2.py:Siglip2VisionModel.forward": ["BaseModelOutputWithPooling", "FloatTensor", "LongTensor", "ModelPreTrainedModel", "ModelVisionModel", "None", "Tensor", "attention_mask", "bool", "class", "def", "forward", "kwargs", "output_attentions", "output_hidden_states", "pixel_attention_mask", "pixel_values", "r", "return", "self", "spatial_shapes", "torch", "vision_model"], "siglip2/modeling_siglip2.py:Siglip2Model.__init__": ["ModelConfig", "ModelModel", "ModelPreTrainedModel", "ModelTextConfig", "ModelTextModel", "ModelVisionConfig", "ModelVisionModel", "Parameter", "TypeError", "__init__", "_from_config", "be", "but", "class", "config", "def", "expected", "f", "if", "is", "isinstance", "logit_bias", "logit_scale", "nn", "not", "of", "post_init", "raise", "randn", "self", "super", "text_config", "text_model", "to", "torch", "type", "vision_config", "vision_model"], "siglip2/modeling_siglip2.py:Siglip2Model.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "Module", "class", "def", "embeddings", "get_input_embeddings", "nn", "return", "self", "text_model", "token_embedding"], "siglip2/modeling_siglip2.py:Siglip2Model.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "Module", "class", "def", "embeddings", "nn", "self", "set_input_embeddings", "text_model", "token_embedding", "value"], "siglip2/modeling_siglip2.py:Siglip2Model.get_text_features": ["BaseModelOutputWithPooling", "FloatTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "attention_mask", "class", "def", "get_text_features", "input_ids", "pooled_output", "pooler_output", "position_ids", "r", "return", "self", "text_model", "text_outputs", "torch"], "siglip2/modeling_siglip2.py:Siglip2Model.get_image_features": ["BaseModelOutputWithPooling", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "attention_mask", "class", "def", "get_image_features", "pixel_attention_mask", "pixel_values", "pooled_output", "pooler_output", "r", "return", "self", "spatial_shapes", "torch", "vision_model", "vision_outputs"], "siglip2/modeling_siglip2.py:Siglip2Model.forward": ["BaseModelOutputWithPooling", "FloatTensor", "LongTensor", "ModelModel", "ModelOutput", "ModelPreTrainedModel", "None", "Tensor", "True", "attention_mask", "bool", "class", "config", "def", "device", "dim", "else", "exp", "eye", "forward", "functional", "if", "image_embeds", "input_ids", "is", "keepdim", "kwargs", "logit_bias", "logit_scale", "logits_per_image", "logits_per_text", "loglik", "logsigmoid", "loss", "m1_diag1", "matmul", "mean", "nll", "nn", "norm", "not", "ones_like", "output_attentions", "output_hidden_states", "p", "pixel_attention_mask", "pixel_values", "pooler_output", "position_ids", "r", "return", "return_loss", "self", "size", "spatial_shapes", "sum", "t", "text_embeds", "text_model", "text_model_output", "text_outputs", "to", "torch", "vision_model", "vision_model_output", "vision_outputs"], "siglip2/modeling_siglip2.py:Siglip2ForImageClassification.__init__": ["Identity", "Linear", "ModelConfig", "ModelForImageClassification", "ModelPreTrainedModel", "ModelVisionModel", "None", "__init__", "_from_config", "class", "classifier", "config", "def", "else", "hidden_size", "if", "nn", "num_labels", "post_init", "self", "super", "vision_config", "vision_model"], "siglip2/modeling_siglip2.py:Siglip2ForImageClassification.get_input_embeddings": ["ModelForImageClassification", "ModelPreTrainedModel", "Module", "class", "def", "embeddings", "get_input_embeddings", "nn", "patch_embedding", "return", "self", "vision_model"], "siglip2/modeling_siglip2.py:Siglip2ForImageClassification.set_input_embeddings": ["ModelForImageClassification", "ModelPreTrainedModel", "Module", "class", "def", "embeddings", "nn", "patch_embedding", "self", "set_input_embeddings", "value", "vision_model"], "siglip2/modeling_siglip2.py:Siglip2ForImageClassification.forward": ["BaseModelOutputWithPooling", "ImageClassifierOutput", "LongTensor", "ModelForImageClassification", "ModelPreTrainedModel", "None", "Tensor", "attention_mask", "attentions", "bool", "class", "classifier", "config", "def", "device", "dim", "else", "forward", "hidden_states", "if", "is", "kwargs", "labels", "last_hidden_state", "logits", "loss", "loss_function", "mean", "not", "output_attentions", "output_hidden_states", "outputs", "pixel_attention_mask", "pixel_values", "pool_mask", "r", "return", "self", "sequence_output", "spatial_shapes", "sum", "to", "torch", "vision_model"], "smollm3/modeling_smollm3.py:SmolLM3RotaryEmbedding.__init__": ["Callable", "False", "ModelConfig", "ModelRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "smollm3/modeling_smollm3.py:SmolLM3RotaryEmbedding.compute_default_rope_parameters": ["ModelConfig", "ModelRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "smollm3/modeling_smollm3.py:SmolLM3RotaryEmbedding.forward": ["False", "ModelRotaryEmbedding", "Module", "None", "and", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "smollm3/modeling_smollm3.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "smollm3/modeling_smollm3.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cos", "def", "k", "k_embed", "q", "q_embed", "return", "rotate_half", "sin", "unsqueeze", "unsqueeze_dim"], "smollm3/modeling_smollm3.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "smollm3/modeling_smollm3.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value", "value_states"], "smollm3/modeling_smollm3.py:SmolLM3Attention.__init__": ["Linear", "ModelAttention", "ModelConfig", "Module", "None", "True", "__init__", "and", "attention_bias", "attention_dropout", "bias", "class", "config", "def", "else", "getattr", "head_dim", "hidden_size", "if", "int", "is_causal", "k_proj", "layer_idx", "layer_types", "nn", "no_rope_layers", "num_attention_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "q_proj", "scaling", "self", "sliding_attention", "sliding_window", "super", "use_rope", "use_sliding_window", "v_proj"], "smollm3/modeling_smollm3.py:SmolLM3Attention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "FlashAttentionKwargs", "LongTensor", "ModelAttention", "Module", "None", "Tensor", "Unpack", "_attn_implementation", "apply_rotary_pos_emb", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "cache_kwargs", "cache_position", "class", "config", "contiguous", "cos", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "head_dim", "hidden_shape", "hidden_states", "if", "input_shape", "is", "k_proj", "key_states", "kwargs", "layer_idx", "nn", "not", "o_proj", "past_key_values", "position_embeddings", "q_proj", "query_states", "reshape", "return", "scaling", "self", "shape", "sin", "sliding_window", "torch", "training", "transpose", "tuple", "update", "use_rope", "v_proj", "value_states", "view"], "smollm3/modeling_smollm3.py:SmolLM3RMSNorm.__init__": ["ModelRMSNorm", "Module", "Parameter", "__init__", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "smollm3/modeling_smollm3.py:SmolLM3RMSNorm.forward": ["ModelRMSNorm", "Module", "True", "class", "def", "dtype", "float32", "forward", "hidden_states", "input_dtype", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "smollm3/modeling_smollm3.py:SmolLM3RMSNorm.extra_repr": ["ModelRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "variance_epsilon", "weight"], "smollm3/modeling_smollm3.py:SmolLM3MLP.__init__": ["ACT2FN", "Linear", "ModelMLP", "Module", "__init__", "act_fn", "bias", "class", "config", "def", "down_proj", "gate_proj", "hidden_act", "hidden_size", "intermediate_size", "mlp_bias", "nn", "self", "super", "up_proj"], "smollm3/modeling_smollm3.py:SmolLM3MLP.forward": ["ModelMLP", "Module", "act_fn", "class", "def", "down_proj", "forward", "gate_proj", "nn", "return", "self", "up_proj", "x"], "smollm3/modeling_smollm3.py:SmolLM3DecoderLayer.__init__": ["GradientCheckpointingLayer", "ModelAttention", "ModelConfig", "ModelDecoderLayer", "ModelMLP", "ModelRMSNorm", "__init__", "attention_type", "class", "config", "def", "eps", "hidden_size", "input_layernorm", "int", "layer_idx", "layer_types", "mlp", "post_attention_layernorm", "rms_norm_eps", "self", "self_attn", "super"], "smollm3/modeling_smollm3.py:SmolLM3DecoderLayer.forward": ["Cache", "False", "GradientCheckpointingLayer", "LongTensor", "ModelDecoderLayer", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "attention_mask", "bool", "cache_position", "class", "def", "forward", "hidden_states", "input_layernorm", "kwargs", "mlp", "past_key_values", "position_embeddings", "position_ids", "post_attention_layernorm", "residual", "return", "self", "self_attn", "torch", "tuple", "use_cache"], "smollm3/modeling_smollm3.py:SmolLM3Model.__init__": ["Embedding", "False", "ModelConfig", "ModelDecoderLayer", "ModelModel", "ModelPreTrainedModel", "ModelRMSNorm", "ModelRotaryEmbedding", "ModuleList", "__init__", "class", "config", "def", "embed_tokens", "eps", "for", "gradient_checkpointing", "has_sliding_layers", "hidden_size", "in", "layer_idx", "layer_types", "layers", "nn", "norm", "num_hidden_layers", "pad_token_id", "padding_idx", "post_init", "range", "rms_norm_eps", "rotary_emb", "self", "sliding_attention", "super", "vocab_size"], "smollm3/modeling_smollm3.py:SmolLM3Model.forward": ["BaseModelOutputWithPast", "Cache", "DynamicCache", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "and", "arange", "attention_mask", "attention_type", "bool", "cache_position", "causal_mask_mapping", "class", "config", "create_causal_mask", "create_sliding_window_causal_mask", "decoder_layer", "def", "device", "dict", "else", "embed_tokens", "exactly", "for", "forward", "full_attention", "get_seq_length", "has_sliding_layers", "hidden_states", "if", "in", "input_embeds", "input_ids", "inputs_embeds", "is", "isinstance", "kwargs", "last_hidden_state", "layers", "mask_kwargs", "must", "norm", "not", "num_hidden_layers", "of", "one", "or", "past_key_values", "past_seen_tokens", "position_embeddings", "position_ids", "raise", "return", "rotary_emb", "self", "shape", "sliding_attention", "specify", "torch", "unsqueeze", "use_cache"], "smollm3/modeling_smollm3.py:SmolLM3ForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "ModelForCausalLM", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "vocab_size"], "smollm3/modeling_smollm3.py:SmolLM3ForCausalLM.forward": ["BaseModelOutputWithPast", "Cache", "CausalLMOutputWithPast", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForCausalLM", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "outputs", "past_key_values", "position_ids", "r", "return", "self", "slice", "slice_indices", "torch", "use_cache", "vocab_size"], "smolvlm/modeling_smolvlm.py:SmolVLMVisionEmbeddings.__init__": ["Conv2d", "Embedding", "ModelVisionConfig", "ModelVisionEmbeddings", "Module", "__init__", "class", "config", "def", "embed_dim", "hidden_size", "image_size", "in_channels", "kernel_size", "nn", "num_channels", "num_patches", "num_patches_per_side", "num_positions", "out_channels", "padding", "patch_embedding", "patch_size", "position_embedding", "self", "stride", "super", "valid"], "smolvlm/modeling_smolvlm.py:SmolVLMVisionEmbeddings.forward": ["BoolTensor", "FloatTensor", "ModelVisionEmbeddings", "Module", "None", "Tensor", "True", "_", "arange", "batch_idx", "batch_size", "boundaries", "bucket_coords_h", "bucket_coords_w", "bucketize", "clamp", "class", "def", "device", "dtype", "embeddings", "enumerate", "fill_value", "flatten", "float32", "for", "forward", "fractional_coords_h", "fractional_coords_w", "full", "h_indices", "in", "max", "max_im_h", "max_im_w", "max_nb_patches_h", "max_nb_patches_w", "nb_patches_h", "nb_patches_w", "nn", "num_patches_per_side", "p_attn_mask", "patch_attention_mask", "patch_embedding", "patch_embeds", "patch_size", "pixel_values", "pos_ids", "position_embedding", "position_ids", "return", "right", "self", "shape", "size", "step_h", "step_w", "sum", "to", "torch", "transpose", "view", "w_indices"], "smolvlm/modeling_smolvlm.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "attention_mask", "attn_output", "attn_weights", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "softmax", "to", "torch", "training", "transpose", "value"], "smolvlm/modeling_smolvlm.py:SmolVLMVisionAttention.__init__": ["False", "Linear", "ModelVisionAttention", "Module", "ValueError", "__init__", "and", "attention_dropout", "be", "by", "class", "config", "def", "divisible", "dropout", "embed_dim", "f", "got", "head_dim", "hidden_size", "if", "is_causal", "k_proj", "must", "nn", "num_attention_heads", "num_heads", "out_proj", "q_proj", "raise", "scale", "self", "super", "v_proj"], "smolvlm/modeling_smolvlm.py:SmolVLMVisionAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "ModelVisionAttention", "Module", "None", "Tensor", "_attn_implementation", "attention_interface", "attention_mask", "attn_output", "attn_weights", "batch_size", "class", "config", "contiguous", "def", "dropout", "eager", "eager_attention_forward", "else", "embed_dim", "forward", "head_dim", "hidden_states", "if", "is_causal", "k_proj", "keys", "kwargs", "nn", "not", "num_heads", "out_proj", "q_proj", "queries", "reshape", "return", "scale", "scaling", "self", "seq_length", "shape", "torch", "training", "transpose", "tuple", "v_proj", "values", "view"], "smolvlm/modeling_smolvlm.py:SmolVLMVisionMLP.__init__": ["ACT2FN", "Linear", "ModelVisionMLP", "Module", "__init__", "activation_fn", "class", "config", "def", "fc1", "fc2", "hidden_act", "hidden_size", "intermediate_size", "nn", "self", "super"], "smolvlm/modeling_smolvlm.py:SmolVLMVisionMLP.forward": ["ModelVisionMLP", "Module", "Tensor", "activation_fn", "class", "def", "fc1", "fc2", "forward", "hidden_states", "nn", "return", "self", "torch"], "smolvlm/modeling_smolvlm.py:SmolVLMEncoderLayer.__init__": ["GradientCheckpointingLayer", "LayerNorm", "ModelEncoderLayer", "ModelVisionAttention", "ModelVisionConfig", "ModelVisionMLP", "__init__", "class", "config", "def", "embed_dim", "eps", "hidden_size", "layer_norm1", "layer_norm2", "layer_norm_eps", "mlp", "nn", "self", "self_attn", "super"], "smolvlm/modeling_smolvlm.py:SmolVLMEncoderLayer.forward": ["FloatTensor", "GradientCheckpointingLayer", "ModelEncoderLayer", "Tensor", "TransformersKwargs", "Unpack", "_", "attention_mask", "class", "def", "forward", "hidden_states", "kwargs", "layer_norm1", "layer_norm2", "mlp", "residual", "return", "self", "self_attn", "torch"], "smolvlm/modeling_smolvlm.py:SmolVLMEncoder.__init__": ["False", "ModelConfig", "ModelEncoder", "ModelEncoderLayer", "Module", "ModuleList", "_", "__init__", "class", "config", "def", "for", "gradient_checkpointing", "in", "layers", "nn", "num_hidden_layers", "range", "self", "super"], "smolvlm/modeling_smolvlm.py:SmolVLMEncoder.forward": ["BaseModelOutput", "ModelEncoder", "Module", "None", "Tensor", "attention_mask", "class", "def", "encoder_layer", "for", "forward", "hidden_states", "in", "inputs_embeds", "last_hidden_state", "layer_outputs", "layers", "nn", "return", "self", "torch", "tuple"], "smolvlm/modeling_smolvlm.py:SmolVLMVisionTransformer.__init__": ["LayerNorm", "ModelEncoder", "ModelPreTrainedModel", "ModelVisionConfig", "ModelVisionEmbeddings", "ModelVisionTransformer", "__init__", "class", "config", "def", "embed_dim", "embeddings", "encoder", "eps", "hidden_size", "layer_norm_eps", "nn", "patch_size", "post_init", "post_layernorm", "self", "super"], "smolvlm/modeling_smolvlm.py:SmolVLMVisionTransformer.get_input_embeddings": ["ModelPreTrainedModel", "ModelVisionTransformer", "class", "def", "embeddings", "get_input_embeddings", "return", "self"], "smolvlm/modeling_smolvlm.py:SmolVLMVisionTransformer.set_input_embeddings": ["ModelPreTrainedModel", "ModelVisionTransformer", "class", "def", "embeddings", "self", "set_input_embeddings", "value"], "smolvlm/modeling_smolvlm.py:SmolVLMVisionTransformer.forward": ["BaseModelOutput", "BoolTensor", "ModelPreTrainedModel", "ModelVisionTransformer", "None", "TransformersKwargs", "Unpack", "attention_mask", "batch_size", "bool", "class", "config", "create_bidirectional_mask", "def", "device", "dtype", "embeddings", "encoder", "encoder_outputs", "forward", "hidden_states", "if", "input_embeds", "inputs_embeds", "is", "kwargs", "last_hidden_state", "ones", "patch_attention_mask", "patch_size", "pixel_values", "post_layernorm", "return", "self", "size", "to", "torch", "tuple", "view"], "smolvlm/modeling_smolvlm.py:SmolVLMSimpleMLP.__init__": ["False", "Linear", "ModelSimpleMLP", "Module", "__init__", "bias", "class", "config", "def", "hidden_size", "input_size", "nn", "output_size", "proj", "scale_factor", "self", "super", "text_config", "vision_config"], "smolvlm/modeling_smolvlm.py:SmolVLMSimpleMLP.forward": ["ModelSimpleMLP", "Module", "class", "def", "forward", "nn", "proj", "return", "self", "x"], "smolvlm/modeling_smolvlm.py:SmolVLMConnector.__init__": ["ModelConnector", "ModelSimpleMLP", "Module", "__init__", "class", "config", "def", "modality_projection", "nn", "scale_factor", "self", "super"], "smolvlm/modeling_smolvlm.py:SmolVLMConnector.pixel_shuffle": ["ModelConnector", "Module", "bsz", "class", "def", "embed_dim", "height", "int", "nn", "permute", "pixel_shuffle", "reshape", "return", "scale_factor", "self", "seq", "size", "view", "width", "x"], "smolvlm/modeling_smolvlm.py:SmolVLMConnector.forward": ["ModelConnector", "Module", "class", "def", "forward", "image_hidden_states", "modality_projection", "nn", "pixel_shuffle", "return", "scale_factor", "self"], "smolvlm/modeling_smolvlm.py:SmolVLMModel.__init__": ["AutoModel", "ModelConfig", "ModelConnector", "ModelModel", "ModelPreTrainedModel", "ModelVisionTransformer", "__init__", "_from_config", "class", "config", "connector", "def", "from_config", "image_seq_len", "image_size", "image_token_id", "int", "pad_token_id", "padding_idx", "patch_size", "post_init", "scale_factor", "self", "super", "text_config", "text_model", "vision_config", "vision_model", "vocab_size"], "smolvlm/modeling_smolvlm.py:SmolVLMModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "return", "self", "text_model"], "smolvlm/modeling_smolvlm.py:SmolVLMModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "self", "set_input_embeddings", "text_model", "value"], "smolvlm/modeling_smolvlm.py:SmolVLMModel.inputs_merger": ["At", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "ValueError", "_", "all", "block_idx", "block_offset", "blocks_per_sample", "by", "chunk_idx", "class", "config", "cumsum", "def", "device", "dim", "divisible", "dtype", "else", "functional", "get_input_embeddings", "has", "if", "image", "image_embeds", "image_hidden_states", "image_mask", "image_token_id", "input_ids", "inputs_embeds", "inputs_merger", "is", "least", "local_idx", "long", "merged_embeds", "nn", "not", "num_image_tokens", "offsets", "one", "pad", "patch_size", "raise", "return", "row_cum", "sample", "self", "shape", "sum", "tensor", "tokens", "torch", "unsqueeze", "value", "where", "zeros_like"], "smolvlm/modeling_smolvlm.py:SmolVLMModel.get_image_features": ["FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "True", "any", "batch_size", "bool", "class", "config", "connector", "contiguous", "def", "device", "dim", "dimension", "dtype", "else", "for", "get_image_features", "height", "i", "if", "image_hidden_states", "in", "is", "last_hidden_state", "nb_values_per_image", "not", "num_channels", "num_images", "numel", "ones", "patch_attention_mask", "patch_size", "patches_subgrid", "pixel_attention_mask", "pixel_values", "real_images_inds", "return", "self", "shape", "size", "step", "sum", "to", "torch", "unfold", "view", "vision_config", "vision_model", "width"], "smolvlm/modeling_smolvlm.py:SmolVLMModel.forward": ["BoolTensor", "Cache", "DynamicCache", "False", "FlashAttentionKwargs", "FloatTensor", "LongTensor", "ModelBaseModelOutputWithPast", "ModelModel", "ModelPreTrainedModel", "None", "Setting", "Tensor", "True", "Unpack", "ValueError", "You", "_", "and", "at", "attention_mask", "attentions", "batch_size", "bool", "both", "cache_position", "cannot", "checkpointing", "class", "config", "def", "device", "dtype", "either", "elif", "else", "forward", "get_image_features", "get_input_embeddings", "gradient", "gradient_checkpointing", "have", "hidden_states", "if", "image_hidden_states", "incompatible", "input_ids", "inputs_embeds", "inputs_merger", "is", "kwargs", "last_hidden_state", "logger", "not", "or", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "pixel_attention_mask", "pixel_values", "position_ids", "r", "raise", "return", "return_dict", "same", "self", "seq_length", "shape", "specify", "text_model", "the", "time", "to", "torch", "training", "tuple", "use_cache", "use_return_dict", "warning_once", "with"], "smolvlm/modeling_smolvlm.py:SmolVLMForConditionalGeneration.__init__": ["False", "GenerationConfig", "GenerationMixin", "Linear", "ModelForConditionalGeneration", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "from_model_config", "generation_config", "hidden_size", "image_token_id", "lm_head", "model", "nn", "post_init", "self", "super", "text_config", "text_model", "vocab_size"], "smolvlm/modeling_smolvlm.py:SmolVLMForConditionalGeneration.get_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "model", "return", "self", "text_model"], "smolvlm/modeling_smolvlm.py:SmolVLMForConditionalGeneration.set_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "model", "self", "set_input_embeddings", "text_model", "value"], "smolvlm/modeling_smolvlm.py:SmolVLMForConditionalGeneration.get_image_features": ["FloatTensor", "GenerationMixin", "LongTensor", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "class", "def", "get_image_features", "model", "pixel_attention_mask", "pixel_values", "return", "self", "torch"], "smolvlm/modeling_smolvlm.py:SmolVLMForConditionalGeneration.forward": ["BoolTensor", "Cache", "FloatTensor", "GenerationMixin", "LongTensor", "ModelCausalLMOutputWithPast", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "else", "forward", "hidden_states", "if", "image_hidden_states", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "pixel_attention_mask", "pixel_values", "position_ids", "r", "return", "return_dict", "self", "slice", "slice_indices", "text_config", "torch", "tuple", "use_cache", "use_return_dict", "vocab_size"], "smolvlm/modeling_smolvlm.py:SmolVLMForConditionalGeneration.prepare_inputs_for_generation": ["False", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "attention_mask", "cache_position", "class", "def", "if", "image_hidden_states", "input_ids", "inputs_embeds", "is", "is_first_iteration", "kwargs", "logits_to_keep", "model_inputs", "not", "or", "past_key_values", "pixel_attention_mask", "pixel_values", "prepare_inputs_for_generation", "return", "self", "super"], "speech_encoder_decoder/modeling_speech_encoder_decoder.py:shift_tokens_right": ["Make", "Model_tokens_right", "Modeled_input_ids", "None", "Tensor", "ValueError", "attribute", "clone", "configuration", "decoder_start_token_id", "def", "if", "input_ids", "int", "is", "masked_fill_", "model", "new_zeros", "of", "pad_token_id", "raise", "return", "s", "set", "shape", "sure", "the", "to", "torch"], "speech_encoder_decoder/modeling_speech_encoder_decoder.py:SpeechEncoderDecoderModel.__init__": ["AutoModel", "AutoModelForCausalLM", "Config", "Either", "False", "GenerationMixin", "Got", "Head", "If", "LM", "Linear", "ModelConfig", "ModelModel", "None", "Please", "PreTrainedConfig", "PreTrainedModel", "The", "ValueError", "__class__", "__init__", "_attn_implementation", "a", "an", "and", "be", "by", "class", "config", "config_class", "configuration", "cross_attention_hidden_size", "decoder", "def", "else", "enc_to_dec_proj", "encoder", "encoder_output_dim", "equal", "f", "for", "from_config", "from_encoder_decoder_configs", "get_output_embeddings", "getattr", "has", "have", "hidden_size", "if", "in", "is", "isinstance", "it", "logger", "model", "nn", "not", "of", "or", "output_hidden_size", "overwritten", "post_init", "provided", "r", "raise", "s", "self", "shared", "should", "specified", "super", "the", "tie_word_embeddings", "to", "to_dict", "type", "use", "warning", "without"], "speech_encoder_decoder/modeling_speech_encoder_decoder.py:SpeechEncoderDecoderModel.get_input_embeddings": ["GenerationMixin", "ModelModel", "PreTrainedModel", "class", "decoder", "def", "get_input_embeddings", "return", "self"], "speech_encoder_decoder/modeling_speech_encoder_decoder.py:SpeechEncoderDecoderModel.get_output_embeddings": ["GenerationMixin", "ModelModel", "PreTrainedModel", "class", "decoder", "def", "get_output_embeddings", "return", "self"], "speech_encoder_decoder/modeling_speech_encoder_decoder.py:SpeechEncoderDecoderModel.set_output_embeddings": ["GenerationMixin", "ModelModel", "PreTrainedModel", "class", "decoder", "def", "new_embeddings", "return", "self", "set_output_embeddings"], "speech_encoder_decoder/modeling_speech_encoder_decoder.py:SpeechEncoderDecoderModel.freeze_feature_encoder": ["GenerationMixin", "ModelModel", "PreTrainedModel", "class", "def", "encoder", "freeze_feature_encoder", "self"], "speech_encoder_decoder/modeling_speech_encoder_decoder.py:SpeechEncoderDecoderModel.from_encoder_decoder_pretrained": ["AutoConfig", "AutoModel", "AutoModelForCausalLM", "Cross", "Decoder", "False", "GenerationMixin", "If", "In", "Initializing", "ModelConfig", "ModelModel", "None", "PreTrainedModel", "True", "ValueError", "a", "add_cross_attention", "added", "allows", "an", "and", "architecture", "are", "argument", "as", "attention", "attributes", "be", "causal", "class", "cls", "config", "cross", "decoder", "decoder_", "decoder_config", "decoder_model", "decoder_pretrained_model_name_or_path", "def", "defined", "del", "disabled", "do", "encoder", "encoder_", "encoder_config", "encoder_model", "encoder_pretrained_model_name_or_path", "f", "for", "from", "from_encoder_decoder_configs", "from_encoder_decoder_pretrained", "from_pretrained", "has", "if", "in", "info", "initialize", "initialized", "is", "is_decoder", "items", "key", "kwargs", "kwargs_decoder", "kwargs_encoder", "layers", "len", "logger", "make", "mask", "model", "model_args", "not", "of", "or", "order", "pass", "passed", "pop", "r", "raise", "randomly", "return", "return_unused_kwargs", "s", "set", "startswith", "str", "sure", "that", "the", "tie_word_embeddings", "to", "value", "warning"], "speech_encoder_decoder/modeling_speech_encoder_decoder.py:SpeechEncoderDecoderModel.forward": ["BaseModelOutput", "BoolTensor", "Cache", "CrossEntropyLoss", "FloatTensor", "GenerationMixin", "LongTensor", "ModelModel", "None", "PreTrainedModel", "Seq2SeqLMOutput", "ValueError", "You", "_get_feature_vector_attention_mask", "and", "argument", "at", "attention_mask", "attentions", "bool", "both", "cannot", "class", "config", "cross_attention_hidden_size", "cross_attentions", "decoder", "decoder_", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_input_ids", "decoder_inputs_embeds", "decoder_outputs", "decoder_start_token_id", "def", "either", "elif", "else", "enc_to_dec_proj", "encoder", "encoder_attention_mask", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_output_dim", "encoder_outputs", "for", "forward", "have", "hidden_size", "hidden_states", "if", "in", "input_features", "input_ids", "input_values", "inputs", "inputs_embeds", "is", "isinstance", "items", "kwargs", "kwargs_decoder", "kwargs_encoder", "labels", "len", "logits", "loss", "loss_fct", "not", "num_items_in_batch", "or", "output_attentions", "output_hidden_states", "pad_token_id", "past_key_values", "pop", "r", "raise", "reshape", "return", "return_dict", "same", "self", "shape", "shift_tokens_right", "specify", "startswith", "the", "time", "to", "torch", "tuple", "use_cache", "use_return_dict", "value", "vocab_size"], "speech_encoder_decoder/modeling_speech_encoder_decoder.py:SpeechEncoderDecoderModel.prepare_decoder_input_ids_from_labels": ["GenerationMixin", "ModelModel", "PreTrainedModel", "Tensor", "class", "config", "decoder_start_token_id", "def", "labels", "pad_token_id", "prepare_decoder_input_ids_from_labels", "return", "self", "shift_tokens_right", "torch"], "speech_encoder_decoder/modeling_speech_encoder_decoder.py:SpeechEncoderDecoderModel.resize_token_embeddings": ["GenerationMixin", "ModelModel", "NotImplementedError", "Please", "PreTrainedModel", "Resizing", "args", "class", "decoder", "def", "directly", "embedding", "is", "kwargs", "layers", "methods", "model", "not", "object", "of", "raise", "resize_token_embeddings", "respective", "self", "supported", "the", "use", "via", "wrapped"], "speech_to_text/modeling_speech_to_text.py:shift_tokens_right": ["Model_tokens_right", "Modeled_input_ids", "None", "Tensor", "ValueError", "be", "clone", "config", "decoder_start_token_id", "def", "defined", "has", "if", "input_ids", "int", "is", "masked_fill_", "model", "new_zeros", "pad_token_id", "raise", "return", "self", "shape", "to", "torch"], "speech_to_text/modeling_speech_to_text.py:Conv1dSubsampler.__init__": ["Model", "ModelSubsampler", "Module", "ModuleList", "__init__", "class", "config", "conv_channels", "conv_kernel_sizes", "conv_layers", "d_model", "def", "else", "enumerate", "for", "i", "if", "in", "in_channels", "input_channels", "input_feat_per_channel", "k", "kernel_size", "kernel_sizes", "mid_channels", "nn", "num_conv_layers", "num_layers", "out_channels", "padding", "self", "stride", "super"], "speech_to_text/modeling_speech_to_text.py:Conv1dSubsampler.forward": ["ModelSubsampler", "Module", "class", "contiguous", "conv", "conv_layers", "def", "dim", "for", "forward", "functional", "glu", "hidden_states", "in", "input_features", "nn", "return", "self", "transpose"], "speech_to_text/modeling_speech_to_text.py:Speech2TextSinusoidalPositionalEmbedding.__init__": ["ModelTextSinusoidalPositionalEmbedding", "Module", "None", "__init__", "class", "def", "embedding_dim", "int", "make_weights", "nn", "num_positions", "offset", "padding_idx", "self", "super"], "speech_to_text/modeling_speech_to_text.py:Speech2TextSinusoidalPositionalEmbedding.make_weights": ["False", "ModelTextSinusoidalPositionalEmbedding", "Module", "None", "class", "def", "device", "dtype", "emb_weights", "embedding_dim", "get_embedding", "hasattr", "if", "int", "make_weights", "nn", "num_embeddings", "padding_idx", "persistent", "register_buffer", "self", "to", "weights"], "speech_to_text/modeling_speech_to_text.py:Speech2TextSinusoidalPositionalEmbedding.get_embedding": ["ModelTextSinusoidalPositionalEmbedding", "Module", "None", "arange", "cat", "class", "cos", "def", "dim", "dtype", "emb", "embedding_dim", "exp", "float", "get_default_dtype", "get_embedding", "half_dim", "if", "int", "int64", "is", "log", "math", "nn", "not", "num_embeddings", "padding_idx", "return", "sin", "to", "torch", "unsqueeze", "view", "zeros"], "speech_to_text/modeling_speech_to_text.py:Speech2TextSinusoidalPositionalEmbedding.forward": ["ModelTextSinusoidalPositionalEmbedding", "Module", "Tensor", "bsz", "class", "create_position_ids_from_input_ids", "def", "detach", "device", "embedding_dim", "forward", "if", "index_select", "input_ids", "int", "make_weights", "max_pos", "nn", "offset", "padding_idx", "past_key_values_length", "position_ids", "return", "self", "seq_len", "size", "to", "torch", "view", "weights"], "speech_to_text/modeling_speech_to_text.py:Speech2TextSinusoidalPositionalEmbedding.create_position_ids_from_input_ids": ["ModelTextSinusoidalPositionalEmbedding", "Module", "None", "Tensor", "class", "create_position_ids_from_input_ids", "cumsum", "def", "dim", "incremental_indices", "input_ids", "int", "long", "mask", "ne", "nn", "padding_idx", "past_key_values_length", "return", "self", "torch", "type_as"], "speech_to_text/modeling_speech_to_text.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "contiguous", "def", "dim", "dropout", "float", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "shape", "size", "softmax", "torch", "training", "transpose", "value"], "speech_to_text/modeling_speech_to_text.py:Speech2TextAttention.__init__": ["False", "Linear", "ModelTextAttention", "ModelTextConfig", "Module", "None", "True", "ValueError", "__init__", "and", "be", "bias", "bool", "by", "class", "config", "def", "divisible", "dropout", "embed_dim", "f", "float", "got", "head_dim", "if", "int", "is_causal", "is_decoder", "k_proj", "layer_idx", "must", "nn", "num_heads", "out_proj", "q_proj", "raise", "scaling", "self", "super", "v_proj"], "speech_to_text/modeling_speech_to_text.py:Speech2TextAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "EncoderDecoderCache", "False", "FlashAttentionKwargs", "ModelTextAttention", "Module", "None", "Tensor", "True", "Unpack", "_attn_implementation", "and", "attention_interface", "attention_mask", "attn_output", "attn_weights", "bool", "bsz", "cache_position", "class", "config", "contiguous", "cross_attention_cache", "curr_past_key_values", "current_states", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "get", "head_dim", "hidden_states", "if", "is", "is_cross_attention", "is_updated", "isinstance", "k_proj", "key_states", "key_value_states", "keys", "kv_input_shape", "kwargs", "layer_idx", "layers", "nn", "not", "out_proj", "output_attentions", "past_key_values", "q_input_shape", "q_proj", "query_states", "reshape", "return", "scaling", "self", "self_attention_cache", "shape", "src_len", "tgt_len", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "values", "view"], "speech_to_text/modeling_speech_to_text.py:Speech2TextEncoderLayer.__init__": ["ACT2FN", "GradientCheckpointingLayer", "LayerNorm", "Linear", "ModelTextAttention", "ModelTextConfig", "ModelTextEncoderLayer", "__init__", "activation_dropout", "activation_fn", "activation_function", "attention_dropout", "class", "config", "d_model", "def", "dropout", "embed_dim", "encoder_attention_heads", "encoder_ffn_dim", "fc1", "fc2", "final_layer_norm", "nn", "num_heads", "self", "self_attn", "self_attn_layer_norm", "super"], "speech_to_text/modeling_speech_to_text.py:Speech2TextEncoderLayer.forward": ["False", "GradientCheckpointingLayer", "ModelTextEncoderLayer", "Tensor", "activation_dropout", "activation_fn", "attention_mask", "attn_weights", "bool", "clamp", "clamp_value", "class", "def", "dropout", "dtype", "fc1", "fc2", "final_layer_norm", "finfo", "float16", "forward", "functional", "hidden_states", "if", "max", "min", "nn", "output_attentions", "p", "residual", "return", "self", "self_attn", "self_attn_layer_norm", "torch", "training"], "speech_to_text/modeling_speech_to_text.py:Speech2TextDecoderLayer.__init__": ["ACT2FN", "GradientCheckpointingLayer", "LayerNorm", "Linear", "ModelTextAttention", "ModelTextConfig", "ModelTextDecoderLayer", "None", "True", "__init__", "activation_dropout", "activation_fn", "activation_function", "attention_dropout", "class", "config", "d_model", "decoder_attention_heads", "decoder_ffn_dim", "def", "dropout", "embed_dim", "encoder_attn", "encoder_attn_layer_norm", "fc1", "fc2", "final_layer_norm", "is_causal", "is_decoder", "layer_idx", "nn", "num_heads", "self", "self_attn", "self_attn_layer_norm", "super"], "speech_to_text/modeling_speech_to_text.py:Speech2TextDecoderLayer.forward": ["Cache", "False", "GradientCheckpointingLayer", "ModelTextDecoderLayer", "None", "Tensor", "True", "activation_dropout", "activation_fn", "attention_mask", "bool", "cache_position", "class", "cross_attn_weights", "def", "dropout", "encoder_attention_mask", "encoder_attn", "encoder_attn_layer_norm", "encoder_hidden_states", "fc1", "fc2", "final_layer_norm", "forward", "functional", "hidden_states", "if", "is", "key_value_states", "nn", "not", "output_attentions", "outputs", "p", "past_key_values", "residual", "return", "self", "self_attn", "self_attn_layer_norm", "self_attn_weights", "torch", "training", "use_cache"], "speech_to_text/modeling_speech_to_text.py:Speech2TextPreTrainedModel._init_weights": ["ModelTextPreTrainedModel", "ModelTextSinusoidalPositionalEmbedding", "PreTrainedModel", "_init_weights", "class", "copy_", "def", "emb_weights", "embedding_dim", "get_embedding", "if", "init", "isinstance", "module", "num_positions", "offset", "padding_idx", "self", "super", "weights"], "speech_to_text/modeling_speech_to_text.py:Speech2TextPreTrainedModel._get_feat_extract_output_lengths": ["LongTensor", "ModelTextPreTrainedModel", "PreTrainedModel", "_get_feat_extract_output_lengths", "class", "config", "def", "for", "i", "in", "input_lengths", "num_conv_layers", "range", "return", "self", "torch"], "speech_to_text/modeling_speech_to_text.py:Speech2TextPreTrainedModel._get_feature_vector_attention_mask": ["ModelTextPreTrainedModel", "PreTrainedModel", "_get_feat_extract_output_lengths", "_get_feature_vector_attention_mask", "arange", "attention_mask", "bsz", "class", "cumsum", "def", "device", "dtype", "feature_vector_length", "flip", "if", "len", "long", "return", "self", "shape", "size", "subsampled_lengths", "sum", "torch", "zeros"], "speech_to_text/modeling_speech_to_text.py:Speech2TextEncoder.__init__": ["Conv1dSubsampler", "False", "LayerNorm", "ModelTextConfig", "ModelTextEncoder", "ModelTextEncoderLayer", "ModelTextPreTrainedModel", "ModelTextSinusoidalPositionalEmbedding", "ModuleList", "_", "__init__", "class", "config", "conv", "d_model", "def", "dropout", "else", "embed_dim", "embed_positions", "embed_scale", "encoder_layerdrop", "encoder_layers", "for", "gradient_checkpointing", "if", "in", "layer_norm", "layerdrop", "layers", "math", "max_source_positions", "nn", "pad_token_id", "padding_idx", "post_init", "range", "scale_embedding", "self", "sqrt", "super"], "speech_to_text/modeling_speech_to_text.py:Speech2TextEncoder.forward": ["BaseModelOutput", "False", "ModelTextEncoder", "ModelTextPreTrainedModel", "None", "True", "_get_feature_vector_attention_mask", "all_attentions", "attention_mask", "attentions", "class", "config", "conv", "create_bidirectional_mask", "def", "device", "dropout", "dropout_probability", "dtype", "else", "embed_pos", "embed_positions", "embed_scale", "encoder_layer", "encoder_states", "enumerate", "for", "forward", "functional", "hidden_states", "idx", "if", "in", "input_embeds", "input_features", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layer_norm", "layer_outputs", "layerdrop", "layers", "long", "ne", "nn", "not", "output_attentions", "output_hidden_states", "p", "padding_mask", "r", "rand", "return", "return_dict", "self", "shape", "to_drop", "torch", "training", "tuple", "use_return_dict", "v", "zeros"], "speech_to_text/modeling_speech_to_text.py:Speech2TextDecoder.__init__": ["Embedding", "False", "LayerNorm", "ModelTextConfig", "ModelTextDecoder", "ModelTextDecoderLayer", "ModelTextPreTrainedModel", "ModelTextSinusoidalPositionalEmbedding", "ModuleList", "__init__", "class", "config", "d_model", "decoder_layerdrop", "decoder_layers", "def", "dropout", "else", "embed_positions", "embed_scale", "embed_tokens", "for", "gradient_checkpointing", "i", "if", "in", "layer_idx", "layer_norm", "layerdrop", "layers", "math", "max_target_positions", "nn", "pad_token_id", "padding_idx", "post_init", "range", "scale_embedding", "self", "sqrt", "super", "vocab_size"], "speech_to_text/modeling_speech_to_text.py:Speech2TextDecoder.forward": ["BaseModelOutputWithPastAndCrossAttentions", "DynamicCache", "EncoderDecoderCache", "False", "ModelTextDecoder", "ModelTextPreTrainedModel", "None", "Setting", "True", "ValueError", "You", "all_cross_attentions", "all_hidden_states", "all_self_attns", "and", "arange", "at", "attention_mask", "attentions", "both", "cache_position", "cannot", "checkpointing", "class", "config", "continue", "create_bidirectional_mask", "create_causal_mask", "cross_attentions", "decoder_input_ids", "decoder_inputs_embeds", "decoder_layer", "def", "device", "dropout", "dropout_probability", "either", "elif", "else", "embed_positions", "embed_scale", "embed_tokens", "encoder_attention_mask", "encoder_hidden_states", "enumerate", "for", "forward", "functional", "get_seq_length", "gradient", "gradient_checkpointing", "have", "hidden_states", "idx", "if", "in", "incompatible", "input_embeds", "input_ids", "input_shape", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layer_norm", "layer_outputs", "layerdrop", "layers", "logger", "nn", "not", "or", "output_attentions", "output_hidden_states", "p", "past_key_values", "past_key_values_length", "positions", "r", "raise", "rand", "return", "return_dict", "same", "self", "size", "specify", "the", "time", "to", "torch", "training", "tuple", "use_cache", "use_return_dict", "v", "view", "warning_once", "with"], "speech_to_text/modeling_speech_to_text.py:Speech2TextModel.__init__": ["ModelTextConfig", "ModelTextDecoder", "ModelTextEncoder", "ModelTextModel", "ModelTextPreTrainedModel", "__init__", "class", "config", "decoder", "def", "encoder", "post_init", "self", "super"], "speech_to_text/modeling_speech_to_text.py:Speech2TextModel.get_input_embeddings": ["ModelTextModel", "ModelTextPreTrainedModel", "class", "decoder", "def", "embed_tokens", "get_input_embeddings", "return", "self"], "speech_to_text/modeling_speech_to_text.py:Speech2TextModel.set_input_embeddings": ["ModelTextModel", "ModelTextPreTrainedModel", "class", "decoder", "def", "embed_tokens", "self", "set_input_embeddings", "value"], "speech_to_text/modeling_speech_to_text.py:Speech2TextModel.forward": ["BaseModelOutput", "Cache", "FloatTensor", "LongTensor", "ModelTextModel", "ModelTextPreTrainedModel", "None", "Seq2SeqLMOutput", "Seq2SeqModelOutput", "Tensor", "_get_feature_vector_attention_mask", "and", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "cross_attentions", "decoder", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_input_ids", "decoder_inputs_embeds", "decoder_outputs", "def", "elif", "else", "encoder", "encoder_attention_mask", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "forward", "hidden_states", "if", "input_features", "input_ids", "inputs_embeds", "is", "isinstance", "kwargs", "last_hidden_state", "len", "not", "output_attentions", "output_hidden_states", "past_key_values", "r", "return", "return_dict", "self", "shape", "torch", "tuple", "use_cache", "use_return_dict"], "speech_to_text/modeling_speech_to_text.py:Speech2TextForConditionalGeneration.__init__": ["False", "GenerationMixin", "Linear", "ModelTextConfig", "ModelTextForConditionalGeneration", "ModelTextModel", "ModelTextPreTrainedModel", "__init__", "bias", "class", "config", "d_model", "def", "lm_head", "model", "nn", "post_init", "self", "super", "vocab_size"], "speech_to_text/modeling_speech_to_text.py:Speech2TextForConditionalGeneration.forward": ["Cache", "CrossEntropyLoss", "FloatTensor", "GenerationMixin", "LongTensor", "ModelTextForConditionalGeneration", "ModelTextPreTrainedModel", "None", "Seq2SeqLMOutput", "Tensor", "and", "attention_mask", "bool", "cache_position", "class", "config", "cross_attentions", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_input_ids", "decoder_inputs_embeds", "decoder_start_token_id", "def", "else", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "forward", "if", "input_features", "is", "kwargs", "labels", "lm_head", "lm_logits", "logits", "loss", "loss_fct", "model", "not", "output", "output_attentions", "output_hidden_states", "outputs", "pad_token_id", "past_key_values", "r", "return", "return_dict", "self", "shift_tokens_right", "torch", "tuple", "use_cache", "use_return_dict", "view", "vocab_size"], "speecht5/modeling_speecht5.py:shift_tokens_right": ["Model_tokens_right", "Modeled_input_ids", "None", "Tensor", "ValueError", "be", "clone", "config", "decoder_start_token_id", "def", "defined", "has", "if", "input_ids", "int", "is", "masked_fill_", "model", "new_zeros", "pad_token_id", "raise", "return", "self", "shape", "to", "torch"], "speecht5/modeling_speecht5.py:shift_spectrograms_right": ["Model_spectrograms_right", "Modeled_input_values", "None", "Tensor", "attention_mask", "clone", "def", "if", "input_values", "int", "is", "masked_fill_", "new_zeros", "not", "reduction_factor", "return", "shape", "torch"], "speecht5/modeling_speecht5.py:_compute_mask_indices": ["False", "LongTensor", "None", "ValueError", "_", "_compute_mask_indices", "and", "append", "arange", "array", "attention_mask", "batch_size", "be", "bigger", "bool", "broadcast_to", "but", "choice", "compute_num_masked_span", "concatenate", "def", "detach", "dtype", "dummy_mask_idx", "else", "epsilon", "f", "float", "for", "got", "has", "if", "in", "input_length", "input_lengths", "int", "int32", "is", "item", "len", "mask_length", "mask_prob", "max", "max_num_masked_span", "min_masks", "ndarray", "not", "np", "num_masked_span", "offsets", "ones", "put_along_axis", "raise", "rand", "random", "range", "replace", "reshape", "return", "sequence_length", "shape", "smaller", "spec_aug_mask", "spec_aug_mask_idx", "spec_aug_mask_idxs", "sum", "than", "to", "tolist", "torch", "tuple", "zeros"], "speecht5/modeling_speecht5.py:SpeechT5NoLayerNormConvLayer.__init__": ["ACT2FN", "Conv1d", "GradientCheckpointingLayer", "ModelNoLayerNormConvLayer", "__init__", "activation", "bias", "class", "config", "conv", "conv_bias", "conv_dim", "conv_kernel", "conv_stride", "def", "else", "feat_extract_activation", "if", "in_conv_dim", "kernel_size", "layer_id", "nn", "out_conv_dim", "self", "stride", "super"], "speecht5/modeling_speecht5.py:SpeechT5NoLayerNormConvLayer.forward": ["GradientCheckpointingLayer", "ModelNoLayerNormConvLayer", "activation", "class", "conv", "def", "forward", "hidden_states", "return", "self"], "speecht5/modeling_speecht5.py:SpeechT5LayerNormConvLayer.__init__": ["ACT2FN", "Conv1d", "GradientCheckpointingLayer", "LayerNorm", "ModelLayerNormConvLayer", "True", "__init__", "activation", "bias", "class", "config", "conv", "conv_bias", "conv_dim", "conv_kernel", "conv_stride", "def", "elementwise_affine", "else", "feat_extract_activation", "if", "in_conv_dim", "kernel_size", "layer_id", "layer_norm", "nn", "out_conv_dim", "self", "stride", "super"], "speecht5/modeling_speecht5.py:SpeechT5LayerNormConvLayer.forward": ["GradientCheckpointingLayer", "ModelLayerNormConvLayer", "activation", "class", "conv", "def", "forward", "hidden_states", "layer_norm", "return", "self", "transpose"], "speecht5/modeling_speecht5.py:SpeechT5GroupNormConvLayer.__init__": ["ACT2FN", "Conv1d", "GradientCheckpointingLayer", "GroupNorm", "ModelGroupNormConvLayer", "True", "__init__", "activation", "affine", "bias", "class", "config", "conv", "conv_bias", "conv_dim", "conv_kernel", "conv_stride", "def", "else", "feat_extract_activation", "if", "in_conv_dim", "kernel_size", "layer_id", "layer_norm", "nn", "num_channels", "num_groups", "out_conv_dim", "self", "stride", "super"], "speecht5/modeling_speecht5.py:SpeechT5GroupNormConvLayer.forward": ["GradientCheckpointingLayer", "ModelGroupNormConvLayer", "activation", "class", "conv", "def", "forward", "hidden_states", "layer_norm", "return", "self"], "speecht5/modeling_speecht5.py:SpeechT5SinusoidalPositionalEmbedding.__init__": ["ModelSinusoidalPositionalEmbedding", "Module", "None", "__init__", "class", "def", "embedding_dim", "int", "make_weights", "nn", "num_positions", "offset", "padding_idx", "self", "super"], "speecht5/modeling_speecht5.py:SpeechT5SinusoidalPositionalEmbedding.make_weights": ["False", "ModelSinusoidalPositionalEmbedding", "Module", "None", "class", "def", "device", "dtype", "emb_weights", "embedding_dim", "get_embedding", "hasattr", "if", "int", "make_weights", "nn", "num_embeddings", "padding_idx", "persistent", "register_buffer", "self", "to", "weights"], "speecht5/modeling_speecht5.py:SpeechT5SinusoidalPositionalEmbedding.get_embedding": ["ModelSinusoidalPositionalEmbedding", "Module", "None", "arange", "cat", "class", "cos", "def", "dim", "dtype", "emb", "embedding_dim", "exp", "float", "get_default_dtype", "get_embedding", "half_dim", "if", "int", "int64", "is", "log", "math", "nn", "not", "num_embeddings", "padding_idx", "return", "sin", "to", "torch", "unsqueeze", "view", "zeros"], "speecht5/modeling_speecht5.py:SpeechT5SinusoidalPositionalEmbedding.forward": ["ModelSinusoidalPositionalEmbedding", "Module", "Tensor", "bsz", "class", "create_position_ids_from_input_ids", "def", "detach", "device", "embedding_dim", "forward", "if", "index_select", "input_ids", "int", "make_weights", "max_pos", "nn", "offset", "padding_idx", "past_key_values_length", "position_ids", "return", "self", "seq_len", "size", "to", "torch", "view", "weights"], "speecht5/modeling_speecht5.py:SpeechT5SinusoidalPositionalEmbedding.create_position_ids_from_input_ids": ["ModelSinusoidalPositionalEmbedding", "Module", "None", "Tensor", "class", "create_position_ids_from_input_ids", "cumsum", "def", "dim", "incremental_indices", "input_ids", "int", "long", "mask", "ne", "nn", "padding_idx", "past_key_values_length", "return", "self", "torch", "type_as"], "speecht5/modeling_speecht5.py:SpeechT5PositionalConvEmbedding.__init__": ["ACT2FN", "Conv1d", "GatheredParameters", "ModelPositionalConvEmbedding", "ModelSamePadLayer", "Module", "__init__", "activation", "class", "config", "conv", "deepspeed", "def", "dim", "else", "feat_extract_activation", "groups", "hasattr", "hidden_size", "if", "is_deepspeed_zero3_enabled", "kernel_size", "modifier_rank", "name", "nn", "num_conv_pos_embedding_groups", "num_conv_pos_embeddings", "original0", "original1", "padding", "parametrizations", "register_external_parameter", "self", "super", "utils", "weight", "weight_g", "weight_norm", "weight_v", "with", "zero"], "speecht5/modeling_speecht5.py:SpeechT5PositionalConvEmbedding.forward": ["ModelPositionalConvEmbedding", "Module", "activation", "class", "conv", "def", "forward", "hidden_states", "nn", "padding", "return", "self", "transpose"], "speecht5/modeling_speecht5.py:SpeechT5ScaledPositionalEncoding.__init__": ["Dropout", "False", "ModelScaledPositionalEncoding", "Module", "Parameter", "__init__", "alpha", "arange", "class", "cos", "def", "dim", "div_term", "dropout", "dtype", "exp", "float", "int64", "log", "math", "max_len", "nn", "p", "pe", "persistent", "position", "register_buffer", "self", "sin", "super", "tensor", "torch", "unsqueeze", "zeros"], "speecht5/modeling_speecht5.py:SpeechT5ScaledPositionalEncoding.forward": ["ModelScaledPositionalEncoding", "Module", "alpha", "class", "def", "dropout", "emb", "forward", "nn", "pe", "return", "self", "size"], "speecht5/modeling_speecht5.py:SpeechT5RelativePositionalEncoding.__init__": ["Embedding", "ModelRelativePositionalEncoding", "Module", "__init__", "class", "def", "dim", "max_length", "nn", "pe_k", "self", "super", "torch"], "speecht5/modeling_speecht5.py:SpeechT5RelativePositionalEncoding.forward": ["ModelRelativePositionalEncoding", "Module", "None", "arange", "class", "def", "device", "dtype", "forward", "hidden_states", "long", "max_length", "nn", "pe_k", "pos_seq", "return", "self", "seq_len", "shape", "to", "torch", "where"], "speecht5/modeling_speecht5.py:SpeechT5SamePadLayer.__init__": ["ModelSamePadLayer", "Module", "__init__", "class", "def", "else", "if", "nn", "num_conv_pos_embeddings", "num_pad_remove", "self", "super"], "speecht5/modeling_speecht5.py:SpeechT5SamePadLayer.forward": ["ModelSamePadLayer", "Module", "class", "def", "forward", "hidden_states", "if", "nn", "num_pad_remove", "return", "self"], "speecht5/modeling_speecht5.py:SpeechT5FeatureEncoder.__init__": ["False", "ModelFeatureEncoder", "ModelGroupNormConvLayer", "ModelLayerNormConvLayer", "ModelNoLayerNormConvLayer", "Module", "ModuleList", "True", "ValueError", "__init__", "_requires_grad", "be", "but", "class", "config", "conv_layers", "def", "elif", "else", "f", "feat_extract_norm", "for", "gradient_checkpointing", "group", "has", "i", "if", "in", "is", "layer", "layer_id", "nn", "num_feat_extract_layers", "of", "one", "raise", "range", "self", "super", "to"], "speecht5/modeling_speecht5.py:SpeechT5FeatureEncoder._freeze_parameters": ["False", "ModelFeatureEncoder", "Module", "_freeze_parameters", "_requires_grad", "class", "def", "for", "in", "nn", "param", "parameters", "requires_grad", "self"], "speecht5/modeling_speecht5.py:SpeechT5FeatureEncoder.forward": ["ModelFeatureEncoder", "Module", "None", "True", "_requires_grad", "and", "class", "conv_layer", "conv_layers", "def", "for", "forward", "hidden_states", "if", "in", "input_values", "nn", "requires_grad", "return", "self", "training"], "speecht5/modeling_speecht5.py:SpeechT5FeatureProjection.__init__": ["Dropout", "LayerNorm", "Linear", "ModelFeatureProjection", "Module", "__init__", "class", "config", "conv_dim", "def", "dropout", "eps", "feat_proj_dropout", "hidden_size", "layer_norm", "layer_norm_eps", "nn", "projection", "self", "super"], "speecht5/modeling_speecht5.py:SpeechT5FeatureProjection.forward": ["ModelFeatureProjection", "Module", "class", "def", "dropout", "forward", "hidden_states", "layer_norm", "nn", "norm_hidden_states", "projection", "return", "self"], "speecht5/modeling_speecht5.py:SpeechT5SpeechEncoderPrenet.__init__": ["ModelFeatureEncoder", "ModelFeatureProjection", "ModelModelEncoderPrenet", "ModelPositionalConvEmbedding", "ModelSinusoidalPositionalEmbedding", "Module", "Parameter", "Tensor", "__init__", "class", "config", "def", "feature_encoder", "feature_projection", "hidden_size", "if", "mask_feature_prob", "mask_time_prob", "masked_spec_embed", "max_Model_positions", "nn", "or", "pad_token_id", "pos_conv_embed", "pos_sinusoidal_embed", "self", "super", "torch", "uniform_"], "speecht5/modeling_speecht5.py:SpeechT5SpeechEncoderPrenet.freeze_feature_encoder": ["ModelModelEncoderPrenet", "Module", "_freeze_parameters", "class", "def", "feature_encoder", "freeze_feature_encoder", "nn", "self"], "speecht5/modeling_speecht5.py:SpeechT5SpeechEncoderPrenet.forward": ["FloatTensor", "LongTensor", "ModelModelEncoderPrenet", "Module", "None", "Tensor", "_get_feature_vector_attention_mask", "_mask_hidden_states", "attention_mask", "class", "def", "device", "dtype", "else", "extract_features", "feature_encoder", "feature_projection", "forward", "hidden_states", "if", "input_values", "is", "long", "mask_time_indices", "ne", "nn", "not", "padding_mask", "pos_conv_embed", "pos_sinusoidal_embed", "positional_conv_embedding", "positional_sinusoidal_embeddings", "return", "self", "shape", "torch", "transpose", "zeros"], "speecht5/modeling_speecht5.py:SpeechT5SpeechEncoderPrenet._get_feature_vector_attention_mask": ["LongTensor", "ModelModelEncoderPrenet", "Module", "_get_feat_extract_output_lengths", "_get_feature_vector_attention_mask", "arange", "attention_mask", "batch_size", "bool", "class", "cumsum", "def", "device", "dim", "dtype", "feature_vector_length", "flip", "int", "long", "nn", "non_padded_lengths", "output_lengths", "return", "self", "shape", "to", "torch", "zeros"], "speecht5/modeling_speecht5.py:SpeechT5SpeechEncoderPrenet._get_feat_extract_output_lengths": ["LongTensor", "ModelModelEncoderPrenet", "Module", "_conv_out_length", "_get_feat_extract_output_lengths", "class", "config", "conv_kernel", "conv_stride", "def", "div", "floor", "for", "in", "input_length", "input_lengths", "int", "kernel_size", "nn", "return", "rounding_mode", "self", "stride", "torch", "zip"], "speecht5/modeling_speecht5.py:SpeechT5SpeechEncoderPrenet._mask_hidden_states": ["FloatTensor", "LongTensor", "ModelModelEncoderPrenet", "Module", "None", "True", "_compute_mask_indices", "_mask_hidden_states", "and", "apply_spec_augment", "attention_mask", "batch_size", "bool", "class", "config", "def", "device", "dtype", "elif", "expand", "getattr", "hidden_size", "hidden_states", "if", "is", "mask_feature_indices", "mask_feature_length", "mask_feature_min_masks", "mask_feature_prob", "mask_length", "mask_prob", "mask_time_indices", "mask_time_length", "mask_time_min_masks", "mask_time_prob", "masked_spec_embed", "min_masks", "nn", "not", "return", "self", "sequence_length", "size", "tensor", "to", "torch", "training"], "speecht5/modeling_speecht5.py:SpeechT5SpeechDecoderPrenet.__init__": ["Linear", "ModelModelDecoderPrenet", "ModelScaledPositionalEncoding", "Model_decoder_prenet_layers", "Model_decoder_prenet_units", "Module", "ModuleList", "__init__", "class", "config", "def", "else", "encode_positions", "final_layer", "for", "hidden_size", "i", "if", "in", "layers", "max_Model_positions", "nn", "num_mel_bins", "positional_dropout", "range", "self", "speaker_embedding_dim", "speaker_embeds_layer", "super"], "speecht5/modeling_speecht5.py:SpeechT5SpeechDecoderPrenet._consistent_dropout": ["ModelModelDecoderPrenet", "Module", "_consistent_dropout", "all_masks", "bernoulli", "class", "def", "inputs_embeds", "mask", "nn", "p", "repeat", "return", "self", "size", "torch", "unsqueeze", "where"], "speecht5/modeling_speecht5.py:SpeechT5SpeechDecoderPrenet.forward": ["ModelModelDecoderPrenet", "Model_decoder_prenet_dropout", "Module", "None", "Tensor", "_consistent_dropout", "cat", "class", "config", "def", "dim", "encode_positions", "expand", "final_layer", "for", "forward", "functional", "if", "in", "input_values", "inputs_embeds", "is", "layer", "layers", "nn", "normalize", "not", "relu", "return", "self", "size", "speaker_embeddings", "speaker_embeds_layer", "torch", "unsqueeze"], "speecht5/modeling_speecht5.py:SpeechT5BatchNormConvLayer.__init__": ["BatchNorm1d", "Conv1d", "Dropout", "False", "ModelBatchNormConvLayer", "Model_decoder_postnet_dropout", "Model_decoder_postnet_kernel", "Model_decoder_postnet_layers", "Model_decoder_postnet_units", "Module", "None", "Tanh", "__init__", "activation", "batch_norm", "bias", "class", "config", "conv", "def", "dropout", "else", "if", "in_conv_dim", "kernel_size", "layer_id", "nn", "num_mel_bins", "out_conv_dim", "padding", "self", "stride", "super"], "speecht5/modeling_speecht5.py:SpeechT5BatchNormConvLayer.forward": ["ModelBatchNormConvLayer", "Module", "None", "activation", "batch_norm", "class", "conv", "def", "dropout", "forward", "hidden_states", "if", "is", "nn", "not", "return", "self"], "speecht5/modeling_speecht5.py:SpeechT5SpeechDecoderPostnet.__init__": ["Linear", "ModelBatchNormConvLayer", "ModelModelDecoderPostnet", "Model_decoder_postnet_layers", "Module", "ModuleList", "__init__", "class", "config", "def", "feat_out", "for", "hidden_size", "i", "in", "layers", "nn", "num_mel_bins", "prob_out", "range", "reduction_factor", "self", "super"], "speecht5/modeling_speecht5.py:SpeechT5SpeechDecoderPostnet.forward": ["ModelModelDecoderPostnet", "Module", "Tensor", "class", "config", "def", "feat_out", "forward", "hidden_states", "logits", "nn", "num_mel_bins", "outputs_after_postnet", "outputs_before_postnet", "postnet", "prob_out", "return", "self", "size", "torch", "view"], "speecht5/modeling_speecht5.py:SpeechT5SpeechDecoderPostnet.postnet": ["ModelModelDecoderPostnet", "Module", "Tensor", "class", "def", "for", "hidden_states", "in", "layer", "layer_output", "layers", "nn", "postnet", "return", "self", "torch", "transpose"], "speecht5/modeling_speecht5.py:SpeechT5TextEncoderPrenet.__init__": ["Embedding", "EmbeddingAccessMixin", "ModelScaledPositionalEncoding", "ModelTextEncoderPrenet", "Module", "__init__", "class", "config", "def", "embed_tokens", "encode_positions", "hidden_size", "max_text_positions", "nn", "pad_token_id", "positional_dropout", "self", "super", "vocab_size"], "speecht5/modeling_speecht5.py:SpeechT5TextEncoderPrenet.forward": ["EmbeddingAccessMixin", "ModelTextEncoderPrenet", "Module", "Tensor", "class", "def", "embed_tokens", "encode_positions", "forward", "input_ids", "inputs_embeds", "nn", "return", "self", "torch"], "speecht5/modeling_speecht5.py:SpeechT5TextDecoderPrenet.__init__": ["Dropout", "Embedding", "EmbeddingAccessMixin", "ModelSinusoidalPositionalEmbedding", "ModelTextDecoderPrenet", "Module", "__init__", "class", "config", "def", "dropout", "else", "embed_positions", "embed_scale", "embed_tokens", "hidden_size", "if", "math", "max_text_positions", "nn", "pad_token_id", "positional_dropout", "scale_embedding", "self", "sqrt", "super", "vocab_size"], "speecht5/modeling_speecht5.py:SpeechT5TextDecoderPrenet.forward": ["Cache", "EmbeddingAccessMixin", "LongTensor", "ModelTextDecoderPrenet", "Module", "None", "Tensor", "ValueError", "You", "attention_mask", "class", "decoder_input_ids", "def", "dropout", "else", "embed_positions", "embed_scale", "embed_tokens", "forward", "get_seq_length", "have", "if", "input_ids", "input_shape", "inputs_embeds", "is", "nn", "not", "past_key_values", "past_key_values_length", "positions", "raise", "return", "self", "size", "specify", "to", "torch", "view"], "speecht5/modeling_speecht5.py:SpeechT5TextDecoderPostnet.__init__": ["EmbeddingAccessMixin", "False", "Linear", "ModelTextDecoderPostnet", "Module", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "nn", "self", "super", "vocab_size"], "speecht5/modeling_speecht5.py:SpeechT5TextDecoderPostnet.forward": ["EmbeddingAccessMixin", "ModelTextDecoderPostnet", "Module", "Tensor", "class", "def", "forward", "hidden_states", "lm_head", "nn", "return", "self", "torch"], "speecht5/modeling_speecht5.py:SpeechT5TextDecoderPostnet.get_output_embeddings": ["EmbeddingAccessMixin", "ModelTextDecoderPostnet", "Module", "class", "def", "get_output_embeddings", "lm_head", "nn", "return", "self"], "speecht5/modeling_speecht5.py:SpeechT5TextDecoderPostnet.set_output_embeddings": ["EmbeddingAccessMixin", "ModelTextDecoderPostnet", "Module", "class", "def", "lm_head", "new_embeddings", "nn", "self", "set_output_embeddings"], "speecht5/modeling_speecht5.py:SpeechT5Attention.__init__": ["False", "Linear", "ModelAttention", "Module", "None", "True", "ValueError", "__init__", "and", "be", "bias", "bool", "by", "class", "def", "divisible", "dropout", "embed_dim", "f", "float", "got", "head_dim", "if", "int", "is_decoder", "k_proj", "layer_idx", "must", "nn", "num_heads", "out_proj", "q_proj", "raise", "scaling", "self", "super", "v_proj"], "speecht5/modeling_speecht5.py:SpeechT5Attention.forward": ["Attention", "Cache", "EncoderDecoderCache", "False", "ModelAttention", "Module", "None", "Tensor", "True", "ValueError", "_", "and", "attention_mask", "attn_output", "attn_probs", "attn_weights", "attn_weights_reshaped", "be", "bmm", "bool", "bsz", "but", "cache_position", "class", "contiguous", "cross_attention_cache", "curr_past_key_values", "current_states", "def", "dim", "dropout", "else", "embed_dim", "f", "forward", "functional", "get", "head_dim", "hidden_states", "if", "is", "is_cross_attention", "is_updated", "isinstance", "k_proj", "key_states", "key_value_states", "keys", "layer_idx", "layers", "mask", "matmul", "nn", "not", "num_heads", "of", "out_proj", "output_attentions", "p", "past_key_values", "position_bias", "proj_shape", "q_proj", "query_states", "raise", "rel_pos_bias", "reshape", "reshape_q", "return", "scaling", "self", "self_attention_cache", "should", "size", "softmax", "src_len", "tgt_len", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "values", "view", "weights"], "speecht5/modeling_speecht5.py:SpeechT5FeedForward.__init__": ["ACT2FN", "Dropout", "Linear", "ModelFeedForward", "Module", "__init__", "activation_dropout", "class", "config", "def", "else", "hidden_act", "hidden_dropout", "hidden_size", "if", "intermediate_act_fn", "intermediate_dense", "intermediate_dropout", "intermediate_size", "isinstance", "nn", "output_dense", "output_dropout", "self", "str", "super"], "speecht5/modeling_speecht5.py:SpeechT5FeedForward.forward": ["ModelFeedForward", "Module", "class", "def", "forward", "hidden_states", "intermediate_act_fn", "intermediate_dense", "intermediate_dropout", "nn", "output_dense", "output_dropout", "return", "self"], "speecht5/modeling_speecht5.py:SpeechT5EncoderLayer.__init__": ["Dropout", "False", "GradientCheckpointingLayer", "LayerNorm", "ModelAttention", "ModelConfig", "ModelEncoderLayer", "ModelFeedForward", "__init__", "attention", "attention_dropout", "class", "config", "def", "dropout", "embed_dim", "encoder_attention_heads", "encoder_ffn_dim", "eps", "feed_forward", "final_layer_norm", "hidden_dropout", "hidden_size", "is_decoder", "layer_norm", "layer_norm_eps", "nn", "num_heads", "self", "super"], "speecht5/modeling_speecht5.py:SpeechT5EncoderLayer.forward": ["False", "GradientCheckpointingLayer", "ModelEncoderLayer", "None", "Tensor", "attention", "attention_mask", "attn_weights", "bool", "class", "def", "dropout", "feed_forward", "final_layer_norm", "forward", "hidden_states", "if", "layer_norm", "output_attentions", "outputs", "position_bias", "residual", "return", "self", "torch"], "speecht5/modeling_speecht5.py:SpeechT5DecoderLayer.__init__": ["Dropout", "GradientCheckpointingLayer", "LayerNorm", "ModelAttention", "ModelConfig", "ModelDecoderLayer", "ModelFeedForward", "None", "True", "__init__", "attention_dropout", "class", "config", "decoder_attention_heads", "decoder_ffn_dim", "def", "dropout", "embed_dim", "encoder_attn", "encoder_attn_layer_norm", "eps", "feed_forward", "final_layer_norm", "hidden_dropout", "hidden_size", "is_decoder", "layer_idx", "layer_norm_eps", "nn", "num_heads", "self", "self_attn", "self_attn_layer_norm", "super"], "speecht5/modeling_speecht5.py:SpeechT5DecoderLayer.forward": ["Cache", "False", "GradientCheckpointingLayer", "ModelDecoderLayer", "None", "Tensor", "True", "attention_mask", "bool", "cache_position", "class", "cross_attn_weights", "def", "dropout", "encoder_attention_mask", "encoder_attn", "encoder_attn_layer_norm", "encoder_hidden_states", "feed_forward", "final_layer_norm", "forward", "hidden_states", "if", "is", "key_value_states", "not", "output_attentions", "outputs", "past_key_values", "residual", "return", "self", "self_attn", "self_attn_layer_norm", "self_attn_weights", "torch", "use_cache"], "speecht5/modeling_speecht5.py:SpeechT5PreTrainedModel._init_weights": ["BatchNorm1d", "Conv1d", "Embedding", "False", "GroupNorm", "LayerNorm", "Linear", "ModelFeatureProjection", "ModelHifiGan", "ModelPositionalConvEmbedding", "ModelPreTrainedModel", "ModelScaledPositionalEncoding", "ModelSinusoidalPositionalEmbedding", "Module", "None", "PreTrainedModel", "_init_weights", "_is_hf_initialized", "a", "alpha", "and", "arange", "b", "bias", "class", "config", "constant_", "conv", "copy_", "cos", "def", "dim", "div_term", "dtype", "elif", "emb_weights", "embedding_dim", "exp", "float", "get_embedding", "getattr", "groups", "hasattr", "if", "in_channels", "in_features", "init", "initializer_range", "int64", "is", "isinstance", "k", "kaiming_normal_", "kernel_size", "log", "masked_spec_embed", "math", "max_len", "mean", "module", "nn", "normal_", "not", "num_batches_tracked", "num_positions", "offset", "ones_", "padding_idx", "pe", "position", "projection", "running_mean", "running_var", "scale", "self", "sin", "sqrt", "std", "torch", "uniform_", "unsqueeze", "weight", "weights", "zeros", "zeros_"], "speecht5/modeling_speecht5.py:SpeechT5Encoder.__init__": ["Dropout", "False", "LayerNorm", "ModelConfig", "ModelEncoder", "ModelEncoderLayer", "ModelPreTrainedModel", "ModelRelativePositionalEncoding", "ModuleList", "_", "__init__", "class", "config", "def", "dropout", "embed_positions", "encoder_attention_heads", "encoder_layerdrop", "encoder_layers", "encoder_max_relative_position", "eps", "for", "gradient_checkpointing", "hidden_dropout", "hidden_size", "in", "layer_norm", "layer_norm_eps", "layerdrop", "layers", "nn", "post_init", "range", "self", "super"], "speecht5/modeling_speecht5.py:SpeechT5Encoder.forward": ["BaseModelOutput", "False", "FloatTensor", "ModelEncoder", "ModelPreTrainedModel", "None", "Tensor", "_prepare_4d_attention_mask", "all_hidden_states", "all_self_attentions", "attention_mask", "attentions", "bool", "class", "config", "def", "dropout", "dropout_probability", "dtype", "else", "embed_positions", "encoder_layer", "enumerate", "for", "forward", "hidden_states", "idx", "if", "in", "is", "is_deepspeed_zero3_enabled", "is_fsdp_managed_module", "kwargs", "last_hidden_state", "layer_norm", "layer_outputs", "layerdrop", "layers", "not", "or", "output_attentions", "output_hidden_states", "position_bias", "rand", "return", "return_dict", "self", "skip_the_layer", "synced_gpus", "torch", "training", "tuple", "use_return_dict", "v"], "speecht5/modeling_speecht5.py:SpeechT5EncoderWithSpeechPrenet.__init__": ["ModelConfig", "ModelEncoder", "ModelEncoderWithModelPrenet", "ModelModelEncoderPrenet", "ModelPreTrainedModel", "__init__", "class", "config", "def", "post_init", "prenet", "self", "super", "wrapped_encoder"], "speecht5/modeling_speecht5.py:SpeechT5EncoderWithSpeechPrenet.forward": ["BaseModelOutput", "FloatTensor", "ModelEncoderWithModelPrenet", "ModelPreTrainedModel", "None", "Tensor", "attention_mask", "bool", "class", "def", "forward", "hidden_states", "input_values", "kwargs", "output_attentions", "output_hidden_states", "outputs", "prenet", "return", "return_dict", "self", "torch", "tuple", "wrapped_encoder"], "speecht5/modeling_speecht5.py:SpeechT5EncoderWithTextPrenet.__init__": ["ModelConfig", "ModelEncoder", "ModelEncoderWithTextPrenet", "ModelPreTrainedModel", "ModelTextEncoderPrenet", "__init__", "class", "config", "def", "post_init", "prenet", "self", "super", "wrapped_encoder"], "speecht5/modeling_speecht5.py:SpeechT5EncoderWithTextPrenet.get_input_embeddings": ["ModelEncoderWithTextPrenet", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "prenet", "return", "self"], "speecht5/modeling_speecht5.py:SpeechT5EncoderWithTextPrenet.set_input_embeddings": ["ModelEncoderWithTextPrenet", "ModelPreTrainedModel", "class", "def", "prenet", "self", "set_input_embeddings", "value"], "speecht5/modeling_speecht5.py:SpeechT5EncoderWithTextPrenet.forward": ["BaseModelOutput", "FloatTensor", "ModelEncoderWithTextPrenet", "ModelPreTrainedModel", "None", "Tensor", "attention_mask", "bool", "class", "def", "forward", "hidden_states", "input_values", "kwargs", "output_attentions", "output_hidden_states", "outputs", "prenet", "return", "return_dict", "self", "torch", "tuple", "wrapped_encoder"], "speecht5/modeling_speecht5.py:SpeechT5EncoderWithoutPrenet.__init__": ["ModelConfig", "ModelEncoder", "ModelEncoderWithoutPrenet", "ModelPreTrainedModel", "__init__", "class", "config", "def", "post_init", "self", "super", "wrapped_encoder"], "speecht5/modeling_speecht5.py:SpeechT5EncoderWithoutPrenet.forward": ["BaseModelOutput", "FloatTensor", "ModelEncoderWithoutPrenet", "ModelPreTrainedModel", "None", "Tensor", "attention_mask", "bool", "class", "def", "forward", "hidden_states", "input_values", "kwargs", "output_attentions", "output_hidden_states", "return", "return_dict", "self", "torch", "tuple", "wrapped_encoder"], "speecht5/modeling_speecht5.py:SpeechT5Decoder.__init__": ["False", "ModelConfig", "ModelDecoder", "ModelDecoderLayer", "ModelPreTrainedModel", "ModuleList", "__init__", "class", "config", "decoder_layerdrop", "decoder_layers", "def", "for", "gradient_checkpointing", "i", "in", "layer_idx", "layerdrop", "layers", "nn", "post_init", "range", "self", "super"], "speecht5/modeling_speecht5.py:SpeechT5Decoder.forward": ["BaseModelOutputWithPastAndCrossAttentions", "Cache", "DynamicCache", "EncoderDecoderCache", "False", "FloatTensor", "LongTensor", "ModelDecoder", "ModelPreTrainedModel", "None", "Setting", "Tensor", "True", "_prepare_4d_attention_mask", "_prepare_4d_causal_attention_mask", "all_cross_attentions", "all_hidden_states", "all_self_attentions", "and", "attention_mask", "attentions", "bool", "cache_position", "checkpointing", "class", "config", "continue", "cross_attentions", "decoder_layer", "def", "dropout_probability", "dtype", "else", "encoder_attention_mask", "encoder_hidden_states", "enumerate", "for", "forward", "get_seq_length", "gradient", "gradient_checkpointing", "hidden_states", "idx", "if", "in", "incompatible", "input_shape", "is", "is_deepspeed_zero3_enabled", "is_fsdp_managed_module", "kwargs", "last_hidden_state", "layer_outputs", "layerdrop", "layers", "logger", "not", "or", "output_attentions", "output_hidden_states", "past_key_values", "past_key_values_length", "r", "rand", "return", "return_dict", "self", "size", "skip_the_layer", "synced_gpus", "tgt_len", "torch", "training", "tuple", "use_cache", "use_return_dict", "v", "warning_once", "with"], "speecht5/modeling_speecht5.py:SpeechT5DecoderWithSpeechPrenet.__init__": ["ModelConfig", "ModelDecoder", "ModelDecoderWithModelPrenet", "ModelModelDecoderPrenet", "ModelPreTrainedModel", "__init__", "class", "config", "def", "post_init", "prenet", "self", "super", "wrapped_decoder"], "speecht5/modeling_speecht5.py:SpeechT5DecoderWithSpeechPrenet.forward": ["BaseModelOutputWithPastAndCrossAttentions", "Cache", "FloatTensor", "LongTensor", "ModelDecoderWithModelPrenet", "ModelPreTrainedModel", "None", "Tensor", "attention_mask", "bool", "cache_position", "class", "decoder_hidden_states", "def", "encoder_attention_mask", "encoder_hidden_states", "forward", "hidden_states", "input_values", "kwargs", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "prenet", "return", "return_dict", "self", "speaker_embeddings", "torch", "tuple", "use_cache", "wrapped_decoder"], "speecht5/modeling_speecht5.py:SpeechT5DecoderWithTextPrenet.__init__": ["ModelConfig", "ModelDecoder", "ModelDecoderWithTextPrenet", "ModelPreTrainedModel", "ModelTextDecoderPrenet", "__init__", "class", "config", "def", "post_init", "prenet", "self", "super", "wrapped_decoder"], "speecht5/modeling_speecht5.py:SpeechT5DecoderWithTextPrenet.get_input_embeddings": ["ModelDecoderWithTextPrenet", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "prenet", "return", "self"], "speecht5/modeling_speecht5.py:SpeechT5DecoderWithTextPrenet.set_input_embeddings": ["ModelDecoderWithTextPrenet", "ModelPreTrainedModel", "class", "def", "prenet", "self", "set_input_embeddings", "value"], "speecht5/modeling_speecht5.py:SpeechT5DecoderWithTextPrenet.forward": ["BaseModelOutputWithPastAndCrossAttentions", "Cache", "FloatTensor", "LongTensor", "ModelDecoderWithTextPrenet", "ModelPreTrainedModel", "None", "Tensor", "attention_mask", "bool", "cache_position", "class", "decoder_hidden_states", "def", "encoder_attention_mask", "encoder_hidden_states", "forward", "hidden_states", "input_values", "kwargs", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "prenet", "return", "return_dict", "self", "torch", "tuple", "use_cache", "wrapped_decoder"], "speecht5/modeling_speecht5.py:SpeechT5DecoderWithoutPrenet.__init__": ["ModelConfig", "ModelDecoder", "ModelDecoderWithoutPrenet", "ModelPreTrainedModel", "__init__", "class", "config", "def", "post_init", "self", "super", "wrapped_decoder"], "speecht5/modeling_speecht5.py:SpeechT5DecoderWithoutPrenet.forward": ["BaseModelOutputWithPastAndCrossAttentions", "Cache", "FloatTensor", "LongTensor", "ModelDecoderWithoutPrenet", "ModelPreTrainedModel", "None", "Tensor", "attention_mask", "bool", "cache_position", "class", "def", "encoder_attention_mask", "encoder_hidden_states", "forward", "hidden_states", "input_values", "kwargs", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "return", "return_dict", "self", "torch", "tuple", "use_cache", "wrapped_decoder"], "speecht5/modeling_speecht5.py:SpeechT5GuidedMultiheadAttentionLoss.__init__": ["ModelConfig", "ModelGuidedMultiheadAttentionLoss", "Module", "__init__", "class", "config", "def", "guided_attention_loss_scale", "guided_attention_loss_sigma", "nn", "scale", "self", "sigma", "super"], "speecht5/modeling_speecht5.py:SpeechT5GuidedMultiheadAttentionLoss.forward": ["BoolTensor", "FloatTensor", "ModelGuidedMultiheadAttentionLoss", "Module", "Tensor", "_make_guided_attention_masks", "attentions", "class", "def", "device", "forward", "guided_attn_masks", "input_masks", "loss", "losses", "masked_select", "masks", "mean", "nn", "output_masks", "return", "scale", "self", "to", "torch", "unsqueeze"], "speecht5/modeling_speecht5.py:SpeechT5GuidedMultiheadAttentionLoss._make_guided_attention_masks": ["ModelGuidedMultiheadAttentionLoss", "Module", "_make_guided_attention_mask", "_make_guided_attention_masks", "class", "def", "device", "enumerate", "for", "guided_attn_masks", "idx", "ilen", "in", "input_lengths", "input_masks", "len", "nn", "olen", "output_lengths", "output_masks", "return", "self", "shape", "sigma", "sum", "torch", "unsqueeze", "zeros", "zip"], "speecht5/modeling_speecht5.py:SpeechT5GuidedMultiheadAttentionLoss._make_guided_attention_mask": ["ModelGuidedMultiheadAttentionLoss", "Module", "_make_guided_attention_mask", "arange", "class", "def", "device", "exp", "float", "grid_x", "grid_y", "indexing", "input_length", "meshgrid", "nn", "output_length", "return", "sigma", "torch", "xy"], "speecht5/modeling_speecht5.py:SpeechT5SpectrogramLoss.__init__": ["BCEWithLogitsLoss", "L1Loss", "ModelConfig", "ModelGuidedMultiheadAttentionLoss", "ModelSpectrogramLoss", "Module", "__init__", "attn_criterion", "bce_criterion", "class", "config", "def", "guided_attention_loss_num_heads", "if", "l1_criterion", "nn", "pos_weight", "reduction_factor", "self", "super", "tensor", "torch", "use_guided_attention_loss"], "speecht5/modeling_speecht5.py:SpeechT5SpectrogramLoss.forward": ["FloatTensor", "LongTensor", "ModelSpectrogramLoss", "Module", "None", "Tensor", "attention_mask", "attn", "attn_criterion", "attn_loss", "bce_criterion", "bce_loss", "cat", "class", "cross_attentions", "def", "device", "dim", "for", "forward", "guided_attention_loss_num_heads", "if", "in", "input_masks", "l1_criterion", "l1_loss", "labels", "logits", "loss", "masked_select", "masks", "nn", "ones", "output_masks", "outputs_after_postnet", "outputs_before_postnet", "padding_mask", "reduction_factor", "return", "self", "size", "stop_labels", "to", "torch", "use_guided_attention_loss", "x"], "speecht5/modeling_speecht5.py:SpeechT5Model.__init__": ["ModelConfig", "ModelDecoderWithoutPrenet", "ModelEncoderWithoutPrenet", "ModelModel", "ModelPreTrainedModel", "Module", "None", "__init__", "class", "config", "decoder", "def", "else", "encoder", "if", "is", "nn", "post_init", "r", "self", "super"], "speecht5/modeling_speecht5.py:SpeechT5Model.get_input_embeddings": ["ModelDecoderWithTextPrenet", "ModelEncoderWithTextPrenet", "ModelModel", "ModelPreTrainedModel", "NotImplementedError", "class", "decoder", "def", "encoder", "get_input_embeddings", "if", "isinstance", "raise", "return", "self"], "speecht5/modeling_speecht5.py:SpeechT5Model.set_input_embeddings": ["ModelDecoderWithTextPrenet", "ModelEncoderWithTextPrenet", "ModelModel", "ModelPreTrainedModel", "class", "decoder", "def", "encoder", "if", "isinstance", "self", "set_input_embeddings", "value"], "speecht5/modeling_speecht5.py:SpeechT5Model.freeze_feature_encoder": ["ModelEncoderWithModelPrenet", "ModelModel", "ModelPreTrainedModel", "class", "def", "encoder", "freeze_feature_encoder", "if", "isinstance", "prenet", "self"], "speecht5/modeling_speecht5.py:SpeechT5Model.forward": ["BaseModelOutput", "Cache", "FloatTensor", "LongTensor", "ModelDecoderWithModelPrenet", "ModelEncoderWithModelPrenet", "ModelModel", "ModelPreTrainedModel", "None", "Seq2SeqModelOutput", "Tensor", "_get_feature_vector_attention_mask", "and", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "cross_attentions", "decoder", "decoder_args", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_input_values", "decoder_outputs", "def", "elif", "else", "encoder", "encoder_attention_mask", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "forward", "hidden_states", "if", "input_values", "is", "isinstance", "kwargs", "last_hidden_state", "len", "not", "output_attentions", "output_hidden_states", "past_key_values", "prenet", "r", "return", "return_dict", "self", "shape", "speaker_embeddings", "torch", "tuple", "use_cache", "use_return_dict"], "speecht5/modeling_speecht5.py:SpeechT5ForSpeechToText.__init__": ["GenerationMixin", "Model", "ModelConfig", "ModelDecoderWithTextPrenet", "ModelEncoderWithModelPrenet", "ModelForModeloText", "ModelModel", "ModelPreTrainedModel", "ModelTextDecoderPostnet", "Model_encoder", "None", "Please", "ValueError", "You", "__class__", "__init__", "a", "are", "as", "class", "config", "configuration", "def", "define", "does", "f", "follows", "from_pretrained", "head", "if", "instantiate", "is", "language", "model", "not", "of", "or", "post_init", "raise", "s", "self", "size", "super", "text_decoder", "text_decoder_postnet", "that", "the", "to", "trying", "vocab_size", "vocabulary", "with", "your"], "speecht5/modeling_speecht5.py:SpeechT5ForSpeechToText.freeze_feature_encoder": ["GenerationMixin", "ModelForModeloText", "ModelPreTrainedModel", "class", "def", "freeze_feature_encoder", "get_encoder", "prenet", "self"], "speecht5/modeling_speecht5.py:SpeechT5ForSpeechToText.get_output_embeddings": ["GenerationMixin", "ModelForModeloText", "ModelPreTrainedModel", "class", "def", "get_output_embeddings", "return", "self", "text_decoder_postnet"], "speecht5/modeling_speecht5.py:SpeechT5ForSpeechToText.set_output_embeddings": ["GenerationMixin", "ModelForModeloText", "ModelPreTrainedModel", "class", "def", "new_embeddings", "self", "set_output_embeddings", "text_decoder_postnet"], "speecht5/modeling_speecht5.py:SpeechT5ForSpeechToText.forward": ["Cache", "CrossEntropyLoss", "FloatTensor", "GenerationMixin", "LongTensor", "Model", "ModelForModeloText", "ModelPreTrainedModel", "None", "Seq2SeqLMOutput", "Tensor", "True", "attention_mask", "bool", "cache_position", "class", "config", "cross_attentions", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_input_ids", "decoder_input_values", "decoder_start_token_id", "def", "else", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "forward", "if", "input_values", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "not", "output", "output_attentions", "output_hidden_states", "outputs", "pad_token_id", "past_key_values", "r", "return", "return_dict", "self", "shift_tokens_right", "text_decoder_postnet", "torch", "tuple", "use_cache", "use_return_dict", "view", "vocab_size"], "speecht5/modeling_speecht5.py:_generate_speech": ["False", "FloatTensor", "LongTensor", "Model", "ModelEncoderWithSpeechPrenet", "ModelPreTrainedModel", "Module", "None", "True", "ValueError", "_generate_speech", "_get_feature_vector_attention_mask", "append", "attention_mask", "batch_first", "bool", "break", "bsz", "cat", "config", "continue", "cross_attentions", "decoder", "decoder_hidden_states", "decoder_out", "def", "dim", "else", "encoder", "encoder_attention_mask", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_out", "feat_out", "flatten", "float", "for", "hidden_states", "i", "idx", "if", "in", "input_values", "int", "is", "isinstance", "last_decoder_output", "last_hidden_state", "len", "max", "maxlen", "maxlenratio", "meet_index", "meet_indexes", "meet_thresholds", "minlen", "minlenratio", "model", "new_spectrogram", "new_zeros", "nn", "not", "num_mel_bins", "output_attentions", "output_cross_attentions", "output_sequence", "outputs", "pad_sequence", "pad_token_id", "past_key_values", "postnet", "prenet", "prob", "prob_out", "raise", "range", "reduction_factor", "result_spectrogram", "return", "return_dict", "return_output_lengths", "rnn", "shape", "sigmoid", "size", "speaker_embeddings", "spectrogram", "spectrogram_lengths", "spectrograms", "spectrum", "speech_decoder_postnet", "squeeze", "stack", "sum", "threshold", "tolist", "torch", "transpose", "tuple", "use_cache", "utils", "view", "vocoder", "waveform_lengths", "waveforms", "where", "while", "wrapped_decoder"], "speecht5/modeling_speecht5.py:SpeechT5ForTextToSpeech.__init__": ["Model", "ModelConfig", "ModelDecoderWithModelPrenet", "ModelEncoderWithTextPrenet", "ModelForTextToModel", "ModelModel", "ModelModelDecoderPostnet", "ModelPreTrainedModel", "Model_decoder", "Model_decoder_postnet", "None", "Please", "ValueError", "You", "__class__", "__init__", "a", "are", "as", "class", "config", "configuration", "def", "define", "does", "f", "follows", "from_pretrained", "head", "if", "instantiate", "is", "language", "model", "not", "of", "or", "post_init", "raise", "s", "self", "size", "super", "text_encoder", "that", "the", "to", "trying", "vocab_size", "vocabulary", "with", "your"], "speecht5/modeling_speecht5.py:SpeechT5ForTextToSpeech.can_generate": ["ModelForTextToModel", "ModelPreTrainedModel", "True", "bool", "can_generate", "class", "cls", "def", "return"], "speecht5/modeling_speecht5.py:SpeechT5ForTextToSpeech.forward": ["Cache", "FloatTensor", "LongTensor", "Model", "ModelForTextToModel", "ModelPreTrainedModel", "ModelSpectrogramLoss", "Model_decoder_postnet", "None", "Seq2SeqSpectrogramOutput", "Tensor", "True", "attention_mask", "bool", "cache_position", "class", "config", "criterion", "cross_attentions", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_input_values", "def", "else", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "forward", "if", "input_ids", "input_values", "is", "kwargs", "labels", "logits", "loss", "not", "output", "output_attentions", "output_hidden_states", "outputs", "outputs_after_postnet", "outputs_before_postnet", "past_key_values", "r", "reduction_factor", "return", "return_dict", "self", "shift_spectrograms_right", "speaker_embeddings", "spectrogram", "stop_labels", "torch", "tuple", "use_cache", "use_guided_attention_loss", "use_return_dict"], "speecht5/modeling_speecht5.py:SpeechT5ForTextToSpeech.generate": ["False", "FloatTensor", "LongTensor", "ModelForTextToModel", "ModelPreTrainedModel", "Module", "None", "The", "ValueError", "_generate_Model", "as", "attention_mask", "batch_size", "be", "bool", "class", "def", "dimension", "either", "else", "first", "float", "generate", "if", "input_ids", "is", "kwargs", "maxlenratio", "minlenratio", "must", "nn", "not", "of", "or", "output_cross_attentions", "r", "raise", "repeat", "return", "return_output_lengths", "same", "self", "size", "speaker_embeddings", "the", "threshold", "torch", "tuple", "vocoder"], "speecht5/modeling_speecht5.py:SpeechT5ForTextToSpeech.generate_speech": ["False", "FloatTensor", "LongTensor", "ModelForTextToModel", "ModelPreTrainedModel", "Module", "None", "The", "ValueError", "_generate_Model", "as", "attention_mask", "batch", "batch_size", "be", "bool", "class", "def", "dimension", "either", "else", "first", "float", "generate_Model", "if", "input_ids", "is", "maxlenratio", "minlenratio", "must", "nn", "not", "of", "or", "output_cross_attentions", "r", "raise", "repeat", "return", "return_output_lengths", "same", "self", "size", "speaker_embeddings", "the", "threshold", "torch", "tuple", "vocoder"], "speecht5/modeling_speecht5.py:SpeechT5ForSpeechToSpeech.__init__": ["Model", "ModelConfig", "ModelDecoderWithModelPrenet", "ModelEncoderWithModelPrenet", "ModelForModeloModel", "ModelModel", "ModelModelDecoderPostnet", "ModelPreTrainedModel", "Model_decoder", "Model_decoder_postnet", "Model_encoder", "__init__", "class", "config", "def", "post_init", "self", "super"], "speecht5/modeling_speecht5.py:SpeechT5ForSpeechToSpeech.freeze_feature_encoder": ["ModelForModeloModel", "ModelPreTrainedModel", "class", "def", "freeze_feature_encoder", "get_encoder", "prenet", "self"], "speecht5/modeling_speecht5.py:SpeechT5ForSpeechToSpeech.forward": ["Cache", "FloatTensor", "LongTensor", "Model", "ModelForModeloModel", "ModelPreTrainedModel", "Model_decoder_postnet", "None", "Seq2SeqSpectrogramOutput", "Tensor", "True", "_", "attention_mask", "bool", "cache_position", "class", "config", "cross_attentions", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_input_values", "def", "else", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "forward", "if", "input_values", "is", "kwargs", "labels", "logits", "loss", "not", "output", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "r", "reduction_factor", "return", "return_dict", "self", "shift_spectrograms_right", "speaker_embeddings", "spectrogram", "stop_labels", "torch", "tuple", "use_cache", "use_return_dict"], "speecht5/modeling_speecht5.py:SpeechT5ForSpeechToSpeech.generate_speech": ["False", "FloatTensor", "LongTensor", "ModelForModeloModel", "ModelPreTrainedModel", "Module", "None", "_generate_Model", "attention_mask", "bool", "class", "def", "device", "float", "generate_Model", "if", "input_values", "is", "maxlenratio", "minlenratio", "nn", "output_cross_attentions", "r", "return", "return_output_lengths", "self", "speaker_embeddings", "threshold", "torch", "vocoder", "zeros"], "speecht5/modeling_speecht5.py:HifiGanResidualBlock.__init__": ["Conv1d", "ModelGanResidualBlock", "Module", "ModuleList", "_", "__init__", "channels", "class", "convs1", "convs2", "def", "dilation", "for", "get_padding", "i", "in", "kernel_size", "leaky_relu_slope", "len", "nn", "padding", "range", "self", "stride", "super"], "speecht5/modeling_speecht5.py:HifiGanResidualBlock.get_padding": ["ModelGanResidualBlock", "Module", "class", "def", "dilation", "get_padding", "kernel_size", "nn", "return", "self"], "speecht5/modeling_speecht5.py:HifiGanResidualBlock.apply_weight_norm": ["ModelGanResidualBlock", "Module", "apply_weight_norm", "class", "convs1", "convs2", "def", "for", "hasattr", "if", "in", "layer", "nn", "parametrizations", "self", "utils", "weight_norm"], "speecht5/modeling_speecht5.py:HifiGanResidualBlock.remove_weight_norm": ["ModelGanResidualBlock", "Module", "class", "convs1", "convs2", "def", "for", "in", "layer", "nn", "remove_weight_norm", "self", "utils"], "speecht5/modeling_speecht5.py:HifiGanResidualBlock.forward": ["ModelGanResidualBlock", "Module", "class", "conv1", "conv2", "convs1", "convs2", "def", "for", "forward", "functional", "hidden_states", "in", "leaky_relu", "leaky_relu_slope", "nn", "residual", "return", "self", "zip"], "speecht5/modeling_speecht5.py:SpeechT5HifiGan.__init__": ["Conv1d", "ConvTranspose1d", "HifiGanResidualBlock", "ModelHifiGan", "ModelHifiGanConfig", "ModuleList", "PreTrainedModel", "__init__", "append", "channels", "class", "config", "conv_post", "conv_pre", "def", "dilation", "enumerate", "for", "i", "in", "kernel_size", "leaky_relu_slope", "len", "mean", "model_in_dim", "nn", "num_kernels", "num_upsamples", "ones", "padding", "post_init", "range", "register_buffer", "resblock_dilation_sizes", "resblock_kernel_sizes", "resblocks", "scale", "self", "stride", "super", "torch", "upsample_initial_channel", "upsample_kernel_sizes", "upsample_rate", "upsample_rates", "upsampler", "zeros", "zip"], "speecht5/modeling_speecht5.py:SpeechT5HifiGan._init_weights": ["ModelHifiGan", "PreTrainedModel", "_init_weights", "class", "def", "if", "init", "isinstance", "mean", "module", "ones_", "scale", "self", "super", "zeros_"], "speecht5/modeling_speecht5.py:SpeechT5HifiGan.apply_weight_norm": ["ModelHifiGan", "PreTrainedModel", "apply_weight_norm", "class", "conv_post", "conv_pre", "def", "for", "hasattr", "if", "in", "layer", "nn", "parametrizations", "resblocks", "self", "upsampler", "utils", "weight_norm"], "speecht5/modeling_speecht5.py:SpeechT5HifiGan.remove_weight_norm": ["ModelHifiGan", "PreTrainedModel", "class", "conv_post", "conv_pre", "def", "for", "in", "layer", "nn", "remove_weight_norm", "resblocks", "self", "upsampler", "utils"], "speecht5/modeling_speecht5.py:SpeechT5HifiGan.forward": ["FloatTensor", "ModelHifiGan", "PreTrainedModel", "class", "config", "conv_post", "conv_pre", "def", "dim", "else", "for", "forward", "functional", "hidden_states", "i", "if", "in", "is_batched", "j", "kwargs", "leaky_relu", "leaky_relu_slope", "mean", "nn", "normalize_before", "not", "num_kernels", "num_upsamples", "r", "range", "res_state", "resblocks", "return", "scale", "self", "spectrogram", "squeeze", "tanh", "torch", "transpose", "unsqueeze", "upsampler", "view", "waveform"], "splinter/modeling_splinter.py:SplinterEmbeddings.__init__": ["Dropout", "Embedding", "False", "LayerNorm", "ModelEmbeddings", "Module", "__init__", "arange", "class", "config", "def", "dropout", "eps", "expand", "hidden_dropout_prob", "hidden_size", "layer_norm_eps", "max_position_embeddings", "nn", "pad_token_id", "padding_idx", "persistent", "position_embeddings", "position_ids", "register_buffer", "self", "super", "token_type_embeddings", "torch", "type_vocab_size", "vocab_size", "word_embeddings"], "splinter/modeling_splinter.py:SplinterEmbeddings.forward": ["FloatTensor", "LayerNorm", "LongTensor", "ModelEmbeddings", "Module", "None", "class", "def", "device", "dropout", "dtype", "else", "embeddings", "forward", "if", "input_ids", "input_shape", "inputs_embeds", "is", "long", "nn", "not", "position_embeddings", "position_ids", "return", "self", "seq_length", "size", "token_type_embeddings", "token_type_ids", "torch", "tuple", "word_embeddings", "zeros"], "splinter/modeling_splinter.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value"], "splinter/modeling_splinter.py:SplinterSelfAttention.__init__": ["Dropout", "Linear", "ModelSelfAttention", "Module", "The", "ValueError", "__init__", "a", "all_head_size", "and", "attention", "attention_dropout", "attention_head_size", "attention_probs_dropout_prob", "class", "config", "def", "dropout", "embedding_size", "f", "hasattr", "heads", "hidden", "hidden_size", "if", "int", "is", "key", "multiple", "nn", "not", "num_attention_heads", "number", "of", "query", "raise", "scaling", "self", "size", "super", "the", "value"], "splinter/modeling_splinter.py:SplinterSelfAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "False", "FloatTensor", "ModelSelfAttention", "Module", "None", "Tensor", "_attn_implementation", "attention_dropout", "attention_head_size", "attention_interface", "attention_mask", "attn_output", "attn_weights", "bool", "class", "config", "contiguous", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "hidden_shape", "hidden_states", "if", "input_shape", "key", "key_states", "kwargs", "nn", "not", "output_attentions", "outputs", "query", "query_states", "reshape", "return", "scaling", "self", "shape", "torch", "training", "transpose", "tuple", "value", "value_states", "view"], "splinter/modeling_splinter.py:SplinterSelfOutput.__init__": ["Dropout", "LayerNorm", "Linear", "ModelSelfOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "eps", "hidden_dropout_prob", "hidden_size", "layer_norm_eps", "nn", "self", "super"], "splinter/modeling_splinter.py:SplinterSelfOutput.forward": ["LayerNorm", "ModelSelfOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "splinter/modeling_splinter.py:SplinterAttention.__init__": ["ModelAttention", "ModelSelfAttention", "ModelSelfOutput", "Module", "__init__", "class", "config", "def", "nn", "output", "self", "super"], "splinter/modeling_splinter.py:SplinterAttention.forward": ["False", "FloatTensor", "ModelAttention", "Module", "None", "Tensor", "attention_mask", "attention_output", "bool", "class", "def", "forward", "hidden_states", "kwargs", "nn", "output", "output_attentions", "outputs", "return", "self", "self_outputs", "torch", "tuple"], "splinter/modeling_splinter.py:SplinterIntermediate.__init__": ["ACT2FN", "Linear", "ModelIntermediate", "Module", "__init__", "class", "config", "def", "dense", "else", "hidden_act", "hidden_size", "if", "intermediate_act_fn", "intermediate_size", "isinstance", "nn", "self", "str", "super"], "splinter/modeling_splinter.py:SplinterIntermediate.forward": ["ModelIntermediate", "Module", "Tensor", "class", "def", "dense", "forward", "hidden_states", "intermediate_act_fn", "nn", "return", "self", "torch"], "splinter/modeling_splinter.py:SplinterOutput.__init__": ["Dropout", "LayerNorm", "Linear", "ModelOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "eps", "hidden_dropout_prob", "hidden_size", "intermediate_size", "layer_norm_eps", "nn", "self", "super"], "splinter/modeling_splinter.py:SplinterOutput.forward": ["LayerNorm", "ModelOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "splinter/modeling_splinter.py:SplinterLayer.__init__": ["GradientCheckpointingLayer", "ModelAttention", "ModelIntermediate", "ModelLayer", "ModelOutput", "__init__", "attention", "chunk_size_feed_forward", "class", "config", "def", "intermediate", "output", "self", "seq_len_dim", "super"], "splinter/modeling_splinter.py:SplinterLayer.forward": ["False", "FloatTensor", "GradientCheckpointingLayer", "ModelLayer", "None", "Tensor", "apply_chunking_to_forward", "attention", "attention_mask", "attention_output", "bool", "chunk_size_feed_forward", "class", "def", "feed_forward_chunk", "forward", "hidden_states", "kwargs", "layer_output", "output_attentions", "outputs", "return", "self", "self_attention_outputs", "seq_len_dim", "torch", "tuple"], "splinter/modeling_splinter.py:SplinterLayer.feed_forward_chunk": ["GradientCheckpointingLayer", "ModelLayer", "attention_output", "class", "def", "feed_forward_chunk", "intermediate", "intermediate_output", "layer_output", "output", "return", "self"], "splinter/modeling_splinter.py:SplinterEncoder.__init__": ["False", "ModelEncoder", "ModelLayer", "Module", "ModuleList", "__init__", "class", "config", "def", "for", "gradient_checkpointing", "i", "in", "layer", "nn", "num_hidden_layers", "range", "self", "super"], "splinter/modeling_splinter.py:SplinterEncoder.forward": ["BaseModelOutput", "False", "FloatTensor", "ModelEncoder", "Module", "None", "Tensor", "True", "all_hidden_states", "all_self_attentions", "attention_mask", "attentions", "bool", "class", "def", "else", "enumerate", "for", "forward", "hidden_states", "i", "if", "in", "kwargs", "last_hidden_state", "layer", "layer_module", "layer_outputs", "nn", "output_attentions", "output_hidden_states", "return", "return_dict", "self", "torch", "tuple"], "splinter/modeling_splinter.py:SplinterPreTrainedModel._init_weights": ["ModelEmbeddings", "ModelPreTrainedModel", "PreTrainedModel", "_init_weights", "arange", "class", "copy_", "def", "expand", "if", "init", "isinstance", "module", "position_ids", "self", "shape", "super", "torch"], "splinter/modeling_splinter.py:SplinterModel.__init__": ["ModelEmbeddings", "ModelEncoder", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "config", "def", "embeddings", "encoder", "post_init", "self", "super"], "splinter/modeling_splinter.py:SplinterModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "return", "self", "word_embeddings"], "splinter/modeling_splinter.py:SplinterModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "self", "set_input_embeddings", "value", "word_embeddings"], "splinter/modeling_splinter.py:SplinterModel.forward": ["BaseModelOutput", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "True", "ValueError", "You", "and", "at", "attention_mask", "attentions", "batch_size", "bool", "both", "cannot", "class", "config", "def", "device", "dtype", "either", "elif", "else", "embedding_output", "embeddings", "encoder", "encoder_outputs", "extended_attention_mask", "forward", "get_extended_attention_mask", "have", "hidden_states", "if", "input_ids", "input_shape", "inputs_embeds", "is", "kwargs", "last_hidden_state", "long", "not", "ones", "or", "output_attentions", "output_hidden_states", "position_ids", "r", "raise", "return", "return_dict", "same", "self", "seq_length", "sequence_output", "size", "specify", "the", "time", "to", "token_type_ids", "torch", "tuple", "use_return_dict", "warn_if_padding_and_no_attention_mask", "zeros"], "splinter/modeling_splinter.py:SplinterFullyConnectedLayer.__init__": ["ACT2FN", "LayerNorm", "Linear", "ModelFullyConnectedLayer", "Module", "__init__", "act_fn", "class", "def", "dense", "gelu", "hidden_act", "input_dim", "nn", "output_dim", "self", "super"], "splinter/modeling_splinter.py:SplinterFullyConnectedLayer.forward": ["LayerNorm", "ModelFullyConnectedLayer", "Module", "Tensor", "act_fn", "class", "def", "dense", "forward", "hidden_states", "inputs", "nn", "return", "self", "torch"], "splinter/modeling_splinter.py:QuestionAwareSpanSelectionHead.__init__": ["False", "Linear", "ModelAwareSpanSelectionHead", "ModelFullyConnectedLayer", "Module", "__init__", "bias", "class", "config", "def", "end_classifier", "end_transform", "hidden_size", "nn", "query_end_transform", "query_start_transform", "self", "start_classifier", "start_transform", "super"], "splinter/modeling_splinter.py:QuestionAwareSpanSelectionHead.forward": ["ModelAwareSpanSelectionHead", "Module", "_", "class", "def", "dim", "end_classifier", "end_logits", "end_reps", "end_transform", "forward", "gather", "gathered_reps", "hidden_states", "index", "inputs", "matmul", "nn", "permute", "positions", "query_end_reps", "query_end_transform", "query_start_reps", "query_start_transform", "repeat", "return", "self", "size", "start_classifier", "start_logits", "start_reps", "start_transform", "torch", "unsqueeze"], "splinter/modeling_splinter.py:SplinterForQuestionAnswering.__init__": ["Model", "ModelForQuestionAnswering", "ModelModel", "ModelPreTrainedModel", "Model_qass", "QuestionAwareSpanSelectionHead", "__init__", "class", "config", "def", "post_init", "question_token_id", "self", "super"], "splinter/modeling_splinter.py:SplinterForQuestionAnswering.forward": ["CrossEntropyLoss", "False", "LongTensor", "Model", "ModelForQuestionAnswering", "ModelPreTrainedModel", "Model_qass", "None", "QuestionAnsweringModelOutput", "Tensor", "True", "and", "argmax", "attention_mask", "attentions", "bool", "clamp_", "class", "config", "def", "device", "dim", "dtype", "else", "end_logits", "end_loss", "end_positions", "eq", "finfo", "forward", "hidden_states", "if", "ignore_index", "ignored_index", "input_ids", "inputs_embeds", "int", "is", "kwargs", "layout", "len", "long", "loss", "loss_fct", "min", "not", "output", "output_attentions", "output_hidden_states", "outputs", "position_ids", "question_position_for_each_example", "question_positions", "question_positions_were_none", "question_token_id", "r", "return", "return_dict", "self", "sequence_output", "size", "squeeze", "start_logits", "start_loss", "start_positions", "token_type_ids", "torch", "total_loss", "tuple", "unsqueeze", "use_return_dict", "zeros"], "splinter/modeling_splinter.py:SplinterForPreTraining.__init__": ["Model", "ModelForPreTraining", "ModelModel", "ModelPreTrainedModel", "Model_qass", "QuestionAwareSpanSelectionHead", "__init__", "class", "config", "def", "post_init", "question_token_id", "self", "super"], "splinter/modeling_splinter.py:SplinterForPreTraining.forward": ["CrossEntropyLoss", "LongTensor", "Model", "ModelForPreTraining", "ModelForPreTrainingOutput", "ModelPreTrainedModel", "Model_qass", "None", "Tensor", "TypeError", "_prepare_question_positions", "and", "attention_mask", "attention_mask_for_each_question", "attentions", "batch_size", "be", "bool", "calculate", "clamp_", "class", "config", "def", "dim", "dtype", "elif", "else", "end_logits", "end_loss", "end_positions", "expand", "finfo", "forward", "hidden_states", "if", "ignore_index", "in", "input_embeds", "input_ids", "inputs_embeds", "is", "kwargs", "loss", "loss_fct", "max", "min", "must", "not", "num_questions", "order", "output", "output_attentions", "output_hidden_states", "outputs", "pad_token_id", "position_ids", "question_positions", "r", "raise", "return", "return_dict", "self", "sequence_length", "sequence_output", "size", "specified", "start_logits", "start_loss", "start_positions", "the", "to", "token_type_ids", "torch", "total_loss", "tuple", "unsqueeze", "use_return_dict", "used", "view", "when"], "splinter/modeling_splinter.py:SplinterForPreTraining._prepare_question_positions": ["ModelForPreTraining", "ModelPreTrainedModel", "Tensor", "_prepare_question_positions", "arange", "bincount", "cat", "class", "cols", "config", "def", "device", "dtype", "flat_positions", "for", "full", "in", "input_ids", "long", "max", "n", "num_questions", "pad_token_id", "positions", "question_token_id", "return", "rows", "self", "size", "torch", "where"], "squeezebert/modeling_squeezebert.py:SqueezeBertEmbeddings.__init__": ["Dropout", "Embedding", "False", "LayerNorm", "ModelEmbeddings", "Module", "__init__", "arange", "class", "config", "def", "dropout", "embedding_size", "eps", "expand", "hidden_dropout_prob", "hidden_size", "layer_norm_eps", "max_position_embeddings", "nn", "pad_token_id", "padding_idx", "persistent", "position_embeddings", "position_ids", "register_buffer", "self", "super", "token_type_embeddings", "torch", "type_vocab_size", "vocab_size", "word_embeddings"], "squeezebert/modeling_squeezebert.py:SqueezeBertEmbeddings.forward": ["LayerNorm", "ModelEmbeddings", "Module", "None", "class", "def", "device", "dropout", "dtype", "else", "embeddings", "forward", "if", "input_ids", "input_shape", "inputs_embeds", "is", "long", "nn", "not", "position_embeddings", "position_ids", "return", "self", "seq_length", "size", "token_type_embeddings", "token_type_ids", "torch", "word_embeddings", "zeros"], "squeezebert/modeling_squeezebert.py:MatMulWrapper.__init__": ["ModelMulWrapper", "Module", "__init__", "class", "def", "nn", "self", "super"], "squeezebert/modeling_squeezebert.py:MatMulWrapper.forward": ["Model1", "Model2", "ModelMulWrapper", "Modelmul", "Module", "class", "def", "forward", "nn", "return", "self", "torch"], "squeezebert/modeling_squeezebert.py:SqueezeBertLayerNorm.__init__": ["LayerNorm", "ModelLayerNorm", "__init__", "class", "def", "eps", "hidden_size", "nn", "normalized_shape", "self"], "squeezebert/modeling_squeezebert.py:SqueezeBertLayerNorm.forward": ["LayerNorm", "ModelLayerNorm", "class", "def", "forward", "nn", "permute", "return", "self", "x"], "squeezebert/modeling_squeezebert.py:ConvDropoutLayerNorm.__init__": ["Dropout", "Model1d", "ModelDropoutLayerNorm", "ModelLayerNorm", "Module", "__init__", "cin", "class", "cout", "def", "dropout", "dropout_prob", "groups", "in_channels", "kernel_size", "layernorm", "nn", "out_channels", "self", "super"], "squeezebert/modeling_squeezebert.py:ConvDropoutLayerNorm.forward": ["Model1d", "ModelDropoutLayerNorm", "Module", "class", "def", "dropout", "forward", "hidden_states", "input_tensor", "layernorm", "nn", "return", "self", "x"], "squeezebert/modeling_squeezebert.py:ConvActivation.__init__": ["ACT2FN", "Model1d", "ModelActivation", "Module", "__init__", "act", "cin", "class", "cout", "def", "groups", "in_channels", "kernel_size", "nn", "out_channels", "self", "super"], "squeezebert/modeling_squeezebert.py:ConvActivation.forward": ["Model1d", "ModelActivation", "Module", "act", "class", "def", "forward", "nn", "output", "return", "self", "x"], "squeezebert/modeling_squeezebert.py:SqueezeBertSelfAttention.__init__": ["Conv1d", "Dropout", "MatMulWrapper", "ModelSelfAttention", "Module", "Softmax", "ValueError", "__init__", "a", "all_head_size", "attention", "attention_head_size", "attention_probs_dropout_prob", "cin", "class", "config", "def", "dim", "dropout", "f", "groups", "heads", "if", "in_channels", "int", "is", "k_groups", "kernel_size", "key", "matmul_qk", "matmul_qkv", "multiple", "nn", "not", "num_attention_heads", "number", "of", "out_channels", "q_groups", "query", "raise", "self", "softmax", "super", "the", "v_groups", "value"], "squeezebert/modeling_squeezebert.py:SqueezeBertSelfAttention.transpose_for_scores": ["ModelSelfAttention", "Module", "attention_head_size", "class", "def", "new_x_shape", "nn", "num_attention_heads", "permute", "return", "self", "size", "transpose_for_scores", "view", "x"], "squeezebert/modeling_squeezebert.py:SqueezeBertSelfAttention.transpose_key_for_scores": ["ModelSelfAttention", "Module", "attention_head_size", "class", "def", "new_x_shape", "nn", "num_attention_heads", "return", "self", "size", "transpose_key_for_scores", "view", "x"], "squeezebert/modeling_squeezebert.py:SqueezeBertSelfAttention.transpose_output": ["ModelSelfAttention", "Module", "all_head_size", "class", "contiguous", "def", "new_x_shape", "nn", "permute", "return", "self", "size", "transpose_output", "view", "x"], "squeezebert/modeling_squeezebert.py:SqueezeBertSelfAttention.forward": ["ModelSelfAttention", "Module", "attention_head_size", "attention_mask", "attention_probs", "attention_score", "class", "context_layer", "def", "dropout", "forward", "hidden_states", "if", "key", "key_layer", "math", "matmul_qk", "matmul_qkv", "mixed_key_layer", "mixed_query_layer", "mixed_value_layer", "nn", "output_attentions", "query", "query_layer", "result", "return", "self", "softmax", "sqrt", "transpose_for_scores", "transpose_key_for_scores", "transpose_output", "value", "value_layer"], "squeezebert/modeling_squeezebert.py:SqueezeBertModule.__init__": ["ConvActivation", "ConvDropoutLayerNorm", "ModelModule", "ModelSelfAttention", "Module", "__init__", "act", "attention", "c0", "c1", "c2", "c3", "cin", "class", "config", "cout", "def", "dropout_prob", "groups", "hidden_act", "hidden_dropout_prob", "hidden_size", "intermediate", "intermediate_groups", "intermediate_size", "k_groups", "nn", "output", "output_groups", "post_attention", "post_attention_groups", "q_groups", "self", "super", "v_groups"], "squeezebert/modeling_squeezebert.py:SqueezeBertModule.forward": ["ModelModule", "Module", "att", "attention", "attention_mask", "attention_output", "attention_score", "class", "context_layer", "def", "feature_map", "forward", "hidden_states", "if", "intermediate", "intermediate_output", "layer_output", "nn", "output", "output_attentions", "output_dict", "post_attention", "post_attention_output", "return", "self"], "squeezebert/modeling_squeezebert.py:SqueezeBertEncoder.__init__": ["Conv1d", "If", "ModelEncoder", "ModelModule", "Module", "ModuleList", "_", "__init__", "a", "adjust", "assert", "before", "channels", "class", "config", "def", "embedding_size", "first", "for", "hidden_size", "in", "insert", "intermediate", "layer", "layers", "nn", "num_hidden_layers", "number", "of", "please", "range", "self", "super", "the", "to", "want", "you"], "squeezebert/modeling_squeezebert.py:SqueezeBertEncoder.forward": ["BaseModelOutput", "False", "ModelEncoder", "Module", "None", "True", "all_attentions", "all_hidden_states", "attention_mask", "attention_score", "attentions", "class", "def", "else", "feature_map", "for", "forward", "hidden_states", "if", "in", "is", "last_hidden_state", "layer", "layer_output", "layers", "nn", "not", "output_attentions", "output_hidden_states", "permute", "return", "return_dict", "self", "tuple", "v"], "squeezebert/modeling_squeezebert.py:SqueezeBertPooler.__init__": ["Linear", "ModelPooler", "Module", "Tanh", "__init__", "activation", "class", "config", "def", "dense", "hidden_size", "nn", "self", "super"], "squeezebert/modeling_squeezebert.py:SqueezeBertPooler.forward": ["ModelPooler", "Module", "activation", "class", "def", "dense", "first_token_tensor", "forward", "hidden_states", "nn", "pooled_output", "return", "self"], "squeezebert/modeling_squeezebert.py:SqueezeBertPredictionHeadTransform.__init__": ["ACT2FN", "LayerNorm", "Linear", "ModelPredictionHeadTransform", "Module", "__init__", "class", "config", "def", "dense", "else", "eps", "hidden_act", "hidden_size", "if", "isinstance", "layer_norm_eps", "nn", "self", "str", "super", "transform_act_fn"], "squeezebert/modeling_squeezebert.py:SqueezeBertPredictionHeadTransform.forward": ["LayerNorm", "ModelPredictionHeadTransform", "Module", "class", "def", "dense", "forward", "hidden_states", "nn", "return", "self", "transform_act_fn"], "squeezebert/modeling_squeezebert.py:SqueezeBertLMPredictionHead.__init__": ["Linear", "ModelLMPredictionHead", "ModelPredictionHeadTransform", "Module", "Parameter", "True", "__init__", "bias", "class", "config", "decoder", "def", "hidden_size", "nn", "self", "super", "torch", "transform", "vocab_size", "zeros"], "squeezebert/modeling_squeezebert.py:SqueezeBertLMPredictionHead.forward": ["ModelLMPredictionHead", "Module", "class", "decoder", "def", "forward", "hidden_states", "nn", "return", "self", "transform"], "squeezebert/modeling_squeezebert.py:SqueezeBertOnlyMLMHead.__init__": ["ModelLMPredictionHead", "ModelOnlyMLMHead", "Module", "__init__", "class", "config", "def", "nn", "predictions", "self", "super"], "squeezebert/modeling_squeezebert.py:SqueezeBertOnlyMLMHead.forward": ["ModelOnlyMLMHead", "Module", "class", "def", "forward", "nn", "prediction_scores", "predictions", "return", "self", "sequence_output"], "squeezebert/modeling_squeezebert.py:SqueezeBertPreTrainedModel._init_weights": ["ModelEmbeddings", "ModelLMPredictionHead", "ModelPreTrainedModel", "PreTrainedModel", "_init_weights", "arange", "bias", "class", "copy_", "def", "elif", "expand", "if", "init", "isinstance", "module", "position_ids", "self", "shape", "super", "torch", "zeros_"], "squeezebert/modeling_squeezebert.py:SqueezeBertModel.__init__": ["ModelEmbeddings", "ModelEncoder", "ModelModel", "ModelPooler", "ModelPreTrainedModel", "__init__", "class", "config", "def", "embeddings", "encoder", "pooler", "post_init", "self", "super"], "squeezebert/modeling_squeezebert.py:SqueezeBertModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "return", "self", "word_embeddings"], "squeezebert/modeling_squeezebert.py:SqueezeBertModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "new_embeddings", "self", "set_input_embeddings", "word_embeddings"], "squeezebert/modeling_squeezebert.py:SqueezeBertModel.forward": ["BaseModelOutputWithPooling", "FloatTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "ValueError", "You", "and", "at", "attention_mask", "attentions", "bool", "both", "cannot", "class", "config", "def", "device", "dtype", "either", "elif", "else", "embedding_output", "embeddings", "encoder", "encoder_outputs", "extended_attention_mask", "forward", "get_extended_attention_mask", "have", "hidden_states", "if", "input_ids", "input_shape", "inputs_embeds", "is", "kwargs", "last_hidden_state", "long", "not", "ones", "or", "output_attentions", "output_hidden_states", "pooled_output", "pooler", "pooler_output", "position_ids", "raise", "return", "return_dict", "same", "self", "sequence_output", "size", "specify", "the", "time", "to", "token_type_ids", "torch", "tuple", "use_return_dict", "warn_if_padding_and_no_attention_mask", "zeros"], "squeezebert/modeling_squeezebert.py:SqueezeBertForMaskedLM.__init__": ["ModelForMaskedLM", "ModelModel", "ModelOnlyMLMHead", "ModelPreTrainedModel", "__init__", "class", "cls", "config", "def", "post_init", "self", "super", "transformer"], "squeezebert/modeling_squeezebert.py:SqueezeBertForMaskedLM.get_output_embeddings": ["ModelForMaskedLM", "ModelPreTrainedModel", "class", "cls", "decoder", "def", "get_output_embeddings", "predictions", "return", "self"], "squeezebert/modeling_squeezebert.py:SqueezeBertForMaskedLM.set_output_embeddings": ["ModelForMaskedLM", "ModelPreTrainedModel", "bias", "class", "cls", "decoder", "def", "new_embeddings", "predictions", "self", "set_output_embeddings"], "squeezebert/modeling_squeezebert.py:SqueezeBertForMaskedLM.forward": ["CrossEntropyLoss", "MaskedLMOutput", "ModelForMaskedLM", "ModelPreTrainedModel", "None", "Tensor", "attention_mask", "attentions", "bool", "class", "cls", "config", "def", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "masked_lm_loss", "not", "output", "output_attentions", "output_hidden_states", "outputs", "position_ids", "prediction_scores", "r", "return", "return_dict", "self", "sequence_output", "token_type_ids", "torch", "transformer", "tuple", "use_return_dict", "view", "vocab_size"], "squeezebert/modeling_squeezebert.py:SqueezeBertForSequenceClassification.__init__": ["Dropout", "Linear", "ModelForSequenceClassification", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "classifier", "config", "def", "dropout", "hidden_dropout_prob", "hidden_size", "nn", "num_labels", "post_init", "self", "super", "transformer"], "squeezebert/modeling_squeezebert.py:SqueezeBertForSequenceClassification.forward": ["BCEWithLogitsLoss", "CrossEntropyLoss", "MSELoss", "Model", "ModelForSequenceClassification", "ModelPreTrainedModel", "None", "SequenceClassifierOutput", "Tensor", "and", "attention_mask", "attentions", "bool", "class", "classifier", "config", "def", "dropout", "dtype", "elif", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "kwargs", "labels", "logits", "long", "loss", "loss_fct", "multi_label_classification", "not", "num_labels", "or", "output", "output_attentions", "output_hidden_states", "outputs", "pooled_output", "position_ids", "problem_type", "r", "regression", "return", "return_dict", "self", "single_label_classification", "token_type_ids", "torch", "transformer", "tuple", "use_return_dict", "view"], "squeezebert/modeling_squeezebert.py:SqueezeBertForMultipleChoice.__init__": ["Dropout", "Linear", "ModelForMultipleChoice", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "classifier", "config", "def", "dropout", "hidden_dropout_prob", "hidden_size", "nn", "post_init", "self", "super", "transformer"], "squeezebert/modeling_squeezebert.py:SqueezeBertForMultipleChoice.forward": ["CrossEntropyLoss", "ModelForMultipleChoice", "ModelPreTrainedModel", "MultipleChoiceModelOutput", "None", "Tensor", "attention_mask", "attentions", "bool", "class", "classifier", "config", "def", "dropout", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "not", "num_choices", "output", "output_attentions", "output_hidden_states", "outputs", "pooled_output", "position_ids", "r", "reshaped_logits", "return", "return_dict", "self", "shape", "size", "token_type_ids", "torch", "transformer", "tuple", "use_return_dict", "view"], "squeezebert/modeling_squeezebert.py:SqueezeBertForTokenClassification.__init__": ["Dropout", "Linear", "ModelForTokenClassification", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "classifier", "config", "def", "dropout", "hidden_dropout_prob", "hidden_size", "nn", "num_labels", "post_init", "self", "super", "transformer"], "squeezebert/modeling_squeezebert.py:SqueezeBertForTokenClassification.forward": ["CrossEntropyLoss", "ModelForTokenClassification", "ModelPreTrainedModel", "None", "Tensor", "TokenClassifierOutput", "attention_mask", "attentions", "bool", "class", "classifier", "config", "def", "dropout", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "not", "num_labels", "output", "output_attentions", "output_hidden_states", "outputs", "position_ids", "r", "return", "return_dict", "self", "sequence_output", "token_type_ids", "torch", "transformer", "tuple", "use_return_dict", "view"], "squeezebert/modeling_squeezebert.py:SqueezeBertForQuestionAnswering.__init__": ["Linear", "ModelForQuestionAnswering", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "config", "def", "hidden_size", "nn", "num_labels", "post_init", "qa_outputs", "self", "super", "transformer"], "squeezebert/modeling_squeezebert.py:SqueezeBertForQuestionAnswering.forward": ["CrossEntropyLoss", "Model", "ModelForQuestionAnswering", "ModelPreTrainedModel", "None", "QuestionAnsweringModelOutput", "Tensor", "and", "attention_mask", "attentions", "bool", "clamp", "class", "config", "contiguous", "def", "dim", "else", "end_logits", "end_loss", "end_positions", "forward", "hidden_states", "if", "ignore_index", "ignored_index", "input_ids", "inputs_embeds", "is", "kwargs", "len", "logits", "loss", "loss_fct", "not", "output", "output_attentions", "output_hidden_states", "outputs", "position_ids", "qa_outputs", "return", "return_dict", "self", "sequence_output", "size", "split", "start_logits", "start_loss", "start_positions", "token_type_ids", "torch", "total_loss", "transformer", "tuple", "use_return_dict"], "stablelm/modeling_stablelm.py:StableLmRotaryEmbedding.__init__": ["Callable", "False", "ModelConfig", "ModelRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "stablelm/modeling_stablelm.py:StableLmRotaryEmbedding.compute_default_rope_parameters": ["ModelConfig", "ModelRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "get", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "partial_rotary_factor", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "stablelm/modeling_stablelm.py:StableLmRotaryEmbedding.forward": ["False", "ModelRotaryEmbedding", "Module", "None", "and", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "stablelm/modeling_stablelm.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "stablelm/modeling_stablelm.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cos", "def", "k", "k_embed", "q", "q_embed", "return", "rotate_half", "sin", "unsqueeze", "unsqueeze_dim"], "stablelm/modeling_stablelm.py:StableLmMLP.__init__": ["ACT2FN", "False", "Linear", "ModelMLP", "Module", "__init__", "act_fn", "bias", "class", "config", "def", "down_proj", "gate_proj", "hidden_act", "hidden_size", "intermediate_size", "nn", "self", "super", "up_proj"], "stablelm/modeling_stablelm.py:StableLmMLP.forward": ["ModelMLP", "Module", "act_fn", "class", "def", "down_proj", "forward", "gate_proj", "nn", "return", "self", "up_proj", "x"], "stablelm/modeling_stablelm.py:StableLmLayerNormPerHead.__init__": ["False", "LayerNorm", "ModelLayerNormPerHead", "Module", "ModuleList", "_", "__init__", "bias", "class", "def", "dim", "eps", "for", "in", "nn", "norms", "num_heads", "range", "self", "super"], "stablelm/modeling_stablelm.py:StableLmLayerNormPerHead.forward": ["ModelLayerNormPerHead", "Module", "Tensor", "cat", "class", "def", "dim", "for", "forward", "hidden_states", "in", "nn", "norm", "norms", "return", "self", "split", "states_per_heads", "torch", "zip"], "stablelm/modeling_stablelm.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "stablelm/modeling_stablelm.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value", "value_states"], "stablelm/modeling_stablelm.py:StableLmAttention.__init__": ["False", "Instantiating", "Linear", "ModelAttention", "ModelConfig", "ModelLayerNormPerHead", "Module", "None", "Please", "True", "ValueError", "__class__", "__init__", "__name__", "a", "and", "attention_dropout", "be", "bias", "by", "caching", "call", "class", "config", "creating", "def", "divisible", "during", "eps", "errors", "f", "forward", "got", "head_dim", "hidden_size", "if", "int", "is", "is_causal", "k_layernorm", "k_proj", "layer_idx", "layer_norm_eps", "lead", "logger", "make", "must", "nn", "not", "num_attention_heads", "num_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "partial_rotary_factor", "passing", "provide", "q_layernorm", "q_proj", "qk_layernorm", "raise", "recommended", "rope_parameters", "rotary_ndims", "scaling", "self", "super", "sure", "the", "this", "to", "use_qkv_bias", "used", "v_proj", "warning_once", "when", "will", "without"], "stablelm/modeling_stablelm.py:StableLmAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "False", "LongTensor", "ModelAttention", "Module", "None", "Tensor", "_", "_attn_implementation", "apply_rotary_pos_emb", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "bool", "bsz", "cache_kwargs", "cache_position", "cat", "class", "config", "cos", "def", "dim", "dropout", "eager", "eager_attention_forward", "else", "forward", "head_dim", "hidden_states", "if", "is", "k_layernorm", "k_proj", "key_pass", "key_rot", "key_states", "kwargs", "layer_idx", "nn", "not", "num_heads", "num_key_value_heads", "o_proj", "output_attentions", "partial_rotation_size", "past_key_values", "position_embeddings", "position_ids", "q_layernorm", "q_len", "q_proj", "qk_layernorm", "query_pass", "query_rot", "query_states", "reshape", "return", "rotary_ndims", "scaling", "self", "sin", "size", "torch", "training", "transpose", "tuple", "update", "use_cache", "v_proj", "value_states", "view"], "stablelm/modeling_stablelm.py:StableLmDecoderLayer.__init__": ["Dropout", "GradientCheckpointingLayer", "LayerNorm", "ModelAttention", "ModelConfig", "ModelDecoderLayer", "ModelMLP", "None", "__init__", "class", "config", "def", "dropout", "eps", "hidden_dropout", "hidden_size", "if", "input_layernorm", "int", "layer_idx", "layer_norm_eps", "mlp", "nn", "not", "post_attention_layernorm", "self", "self_attn", "super", "use_parallel_residual"], "stablelm/modeling_stablelm.py:StableLmDecoderLayer.forward": ["Cache", "False", "FloatTensor", "GradientCheckpointingLayer", "LongTensor", "ModelDecoderLayer", "None", "Tensor", "attention_mask", "bool", "cache_position", "class", "def", "dropout", "else", "forward", "hidden_states", "if", "input_layernorm", "mlp", "mlp_output", "output_attentions", "outputs", "past_key_values", "position_embeddings", "position_ids", "post_attention_layernorm", "residual", "return", "self", "self_attn", "self_attn_output", "self_attn_weights", "torch", "tuple", "use_cache", "use_parallel_residual"], "stablelm/modeling_stablelm.py:StableLmModel.__init__": ["Embedding", "False", "LayerNorm", "ModelConfig", "ModelDecoderLayer", "ModelModel", "ModelPreTrainedModel", "ModelRotaryEmbedding", "ModuleList", "__init__", "_attn_implementation", "class", "config", "def", "embed_tokens", "eps", "for", "gradient_checkpointing", "hidden_size", "in", "layer_idx", "layer_norm_eps", "layers", "nn", "norm", "num_hidden_layers", "pad_token_id", "padding_idx", "post_init", "range", "rotary_emb", "self", "super", "vocab_size"], "stablelm/modeling_stablelm.py:StableLmModel.forward": ["BaseModelOutputWithPast", "Cache", "DynamicCache", "False", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Setting", "Tensor", "True", "ValueError", "You", "_update_causal_mask", "all_hidden_states", "all_self_attns", "and", "arange", "attention_mask", "attentions", "bool", "cache_position", "causal_mask", "checkpointing", "class", "config", "decoder_layer", "def", "device", "else", "embed_tokens", "exactly", "for", "forward", "get_seq_length", "gradient", "gradient_checkpointing", "hidden_states", "if", "in", "incompatible", "input_ids", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layer_outputs", "layers", "logger", "must", "norm", "not", "of", "one", "or", "output_attentions", "output_hidden_states", "past_key_values", "past_seen_tokens", "position_embeddings", "position_ids", "raise", "return", "rotary_emb", "self", "shape", "specify", "torch", "training", "unsqueeze", "use_cache", "warning_once", "with"], "stablelm/modeling_stablelm.py:StableLmModel._update_causal_mask": ["AttentionMaskConverter", "BlockMask", "Cache", "False", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "Union", "_attn_implementation", "_ignore_causal_mask_sdpa", "_prepare_4d_causal_attention_mask_with_cache_position", "_unmask_unattended", "_update_causal_mask", "and", "any", "attention_mask", "batch_size", "bool", "cache_position", "causal_mask", "class", "config", "cuda", "def", "device", "dtype", "else", "finfo", "flash_attention_2", "flex_attention", "get_max_cache_shape", "get_seq_length", "if", "in", "input_tensor", "inputs_embeds", "is", "is_compileable", "is_training", "isinstance", "make_flex_block_causal_mask", "min", "min_dtype", "not", "npu", "output_attentions", "past_key_values", "past_key_values_length", "past_seen_tokens", "return", "sdpa", "self", "sequence_length", "shape", "target_length", "torch", "training", "type", "using_compilable_cache", "xpu"], "stablelm/modeling_stablelm.py:StableLmModel._prepare_4d_causal_attention_mask_with_cache_position": ["ModelModel", "ModelPreTrainedModel", "None", "Tensor", "_prepare_4d_causal_attention_mask_with_cache_position", "and", "arange", "attention_mask", "batch_size", "cache_position", "causal_mask", "class", "clone", "def", "device", "diagonal", "dim", "dtype", "else", "expand", "fill_value", "finfo", "full", "if", "int", "is", "kwargs", "mask_length", "masked_fill", "min", "min_dtype", "not", "padding_mask", "reshape", "return", "sequence_length", "shape", "target_length", "to", "torch", "triu"], "stablelm/modeling_stablelm.py:StableLmForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "ModelForCausalLM", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "vocab_size"], "stablelm/modeling_stablelm.py:StableLmForCausalLM.forward": ["BaseModelOutputWithPast", "Cache", "CausalLMOutputWithPast", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForCausalLM", "ModelPreTrainedModel", "None", "Tensor", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "position_ids", "r", "return", "self", "slice", "slice_indices", "torch", "use_cache", "vocab_size"], "starcoder2/modeling_starcoder2.py:Starcoder2MLP.__init__": ["ACT2FN", "Linear", "ModelConfig", "ModelMLP", "Module", "__init__", "act", "bias", "c_fc", "c_proj", "class", "config", "def", "embed_dim", "hidden_act", "hidden_size", "intermediate_size", "nn", "residual_dropout", "self", "super", "use_bias"], "starcoder2/modeling_starcoder2.py:Starcoder2MLP.forward": ["FloatTensor", "ModelMLP", "Module", "None", "act", "c_fc", "c_proj", "class", "def", "dropout", "forward", "functional", "hidden_states", "nn", "p", "residual_dropout", "return", "self", "torch", "training", "tuple"], "starcoder2/modeling_starcoder2.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "starcoder2/modeling_starcoder2.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cos", "def", "k", "k_embed", "q", "q_embed", "return", "rotate_half", "sin", "unsqueeze", "unsqueeze_dim"], "starcoder2/modeling_starcoder2.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "starcoder2/modeling_starcoder2.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value", "value_states"], "starcoder2/modeling_starcoder2.py:Starcoder2Attention.__init__": ["Linear", "ModelAttention", "ModelConfig", "Module", "None", "True", "__init__", "attention_dropout", "bias", "class", "config", "def", "getattr", "head_dim", "hidden_size", "int", "is_causal", "k_proj", "layer_idx", "nn", "num_attention_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "or", "q_proj", "residual_dropout", "scaling", "self", "super", "use_bias", "v_proj"], "starcoder2/modeling_starcoder2.py:Starcoder2Attention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "FlashAttentionKwargs", "LongTensor", "ModelAttention", "Module", "None", "Tensor", "Unpack", "_attn_implementation", "apply_rotary_pos_emb", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "cache_kwargs", "cache_position", "class", "config", "contiguous", "cos", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "functional", "getattr", "head_dim", "hidden_shape", "hidden_states", "if", "input_shape", "is", "k_proj", "key_states", "kwargs", "layer_idx", "nn", "not", "o_proj", "p", "past_key_values", "position_embeddings", "q_proj", "query_states", "reshape", "residual_dropout", "return", "scaling", "self", "shape", "sin", "sliding_window", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "view"], "starcoder2/modeling_starcoder2.py:Starcoder2DecoderLayer.__init__": ["GradientCheckpointingLayer", "LayerNorm", "ModelAttention", "ModelConfig", "ModelDecoderLayer", "ModelMLP", "__init__", "class", "config", "def", "eps", "hidden_size", "input_layernorm", "int", "layer_idx", "mlp", "nn", "norm_epsilon", "post_attention_layernorm", "self", "self_attn", "super"], "starcoder2/modeling_starcoder2.py:Starcoder2DecoderLayer.forward": ["Cache", "False", "GradientCheckpointingLayer", "LongTensor", "ModelDecoderLayer", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "attention_mask", "bool", "cache_position", "class", "def", "forward", "hidden_states", "input_layernorm", "kwargs", "mlp", "past_key_values", "position_embeddings", "position_ids", "post_attention_layernorm", "residual", "return", "self", "self_attn", "torch", "tuple", "use_cache"], "starcoder2/modeling_starcoder2.py:Starcoder2RotaryEmbedding.__init__": ["Callable", "False", "ModelConfig", "ModelRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "starcoder2/modeling_starcoder2.py:Starcoder2RotaryEmbedding.compute_default_rope_parameters": ["ModelConfig", "ModelRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "starcoder2/modeling_starcoder2.py:Starcoder2RotaryEmbedding.forward": ["False", "ModelRotaryEmbedding", "Module", "None", "and", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "starcoder2/modeling_starcoder2.py:Starcoder2Model.__init__": ["Embedding", "False", "LayerNorm", "ModelConfig", "ModelDecoderLayer", "ModelModel", "ModelPreTrainedModel", "ModelRotaryEmbedding", "ModuleList", "__init__", "class", "config", "def", "embed_tokens", "embedding_dropout", "eps", "for", "gradient_checkpointing", "hidden_size", "in", "layer_idx", "layers", "nn", "norm", "norm_epsilon", "num_hidden_layers", "pad_token_id", "padding_idx", "post_init", "range", "rotary_emb", "self", "super", "vocab_size"], "starcoder2/modeling_starcoder2.py:Starcoder2Model.forward": ["BaseModelOutputWithPast", "Cache", "DynamicCache", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "and", "arange", "attention_mask", "bool", "cache_position", "causal_mask", "class", "config", "create_causal_mask", "create_sliding_window_causal_mask", "decoder_layer", "def", "device", "dropout", "else", "embed_tokens", "embedding_dropout", "exactly", "for", "forward", "functional", "get_seq_length", "hidden_states", "if", "in", "input_embeds", "input_ids", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layers", "mask_function", "must", "nn", "norm", "not", "num_hidden_layers", "of", "one", "or", "p", "past_key_values", "past_seen_tokens", "position_embeddings", "position_ids", "raise", "return", "rotary_emb", "self", "shape", "sliding_window", "specify", "torch", "training", "unsqueeze", "use_cache"], "starcoder2/modeling_starcoder2.py:Starcoder2ForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "ModelForCausalLM", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "vocab_size"], "starcoder2/modeling_starcoder2.py:Starcoder2ForCausalLM.forward": ["BaseModelOutputWithPast", "Cache", "CausalLMOutputWithPast", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForCausalLM", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "outputs", "past_key_values", "position_ids", "r", "return", "self", "slice", "slice_indices", "torch", "use_cache", "vocab_size"], "superglue/modeling_superglue.py:concat_pairs": ["Model_pairs", "Tensor", "cat", "def", "for", "in", "return", "tensor0", "tensor1", "tensor_tuple0", "tensor_tuple1", "torch", "tuple", "zip"], "superglue/modeling_superglue.py:normalize_keypoints": ["Model_keypoints", "None", "Tensor", "True", "center", "def", "device", "dtype", "height", "int", "keepdim", "keypoints", "max", "return", "scaling", "size", "tensor", "torch", "values", "width"], "superglue/modeling_superglue.py:log_sinkhorn_iterations": ["Model_cost_matrix", "Model_sinkhorn_iterations", "Model_source_distribution", "Model_target_distribution", "Model_u_scaling", "Model_v_scaling", "Modelsumexp", "Tensor", "_", "def", "dim", "for", "in", "int", "num_iterations", "range", "return", "torch", "unsqueeze", "zeros_like"], "superglue/modeling_superglue.py:log_optimal_transport": ["Model", "Model_normalization", "Model_optimal_transport", "Model_optimal_transport_matrix", "Model_sinkhorn_iterations", "Model_source_distribution", "Model_target_distribution", "None", "Tensor", "batch_size", "cat", "couplings", "def", "expand", "int", "iterations", "new_tensor", "num_columns", "num_columns_tensor", "num_iterations", "num_rows", "num_rows_tensor", "one_tensor", "reg_param", "return", "scores", "shape", "source_reg_param", "target_reg_param", "to", "torch"], "superglue/modeling_superglue.py:arange_like": ["Model_like", "Tensor", "cumsum", "def", "dim", "int", "new_ones", "return", "shape", "torch", "x"], "superglue/modeling_superglue.py:SuperGlueMultiLayerPerceptron.__init__": ["BatchNorm1d", "Linear", "Model", "ModelConfig", "ModelMultiLayerPerceptron", "Module", "None", "ReLU", "__init__", "activation", "batch_norm", "class", "config", "def", "in_channels", "int", "linear", "nn", "out_channels", "self"], "superglue/modeling_superglue.py:SuperGlueMultiLayerPerceptron.forward": ["ModelMultiLayerPerceptron", "Module", "Tensor", "activation", "batch_norm", "class", "def", "forward", "hidden_state", "linear", "nn", "return", "self", "torch", "transpose"], "superglue/modeling_superglue.py:SuperGlueKeypointEncoder.__init__": ["Linear", "Model", "ModelConfig", "ModelKeypointEncoder", "ModelMultiLayerPerceptron", "Module", "ModuleList", "None", "__init__", "append", "class", "config", "def", "encoder", "encoder_channels", "for", "hidden_size", "i", "in", "keypoint_encoder_sizes", "layer_sizes", "layers", "len", "nn", "range", "self"], "superglue/modeling_superglue.py:SuperGlueKeypointEncoder.forward": ["False", "ModelKeypointEncoder", "Module", "None", "Tensor", "all_hidden_states", "bool", "cat", "class", "def", "dim", "else", "encoder", "for", "forward", "hidden_state", "if", "in", "keypoints", "layer", "nn", "output_hidden_states", "return", "scores", "self", "torch", "tuple", "unsqueeze"], "superglue/modeling_superglue.py:SuperGlueSelfAttention.__init__": ["Dropout", "Linear", "Model", "ModelSelfAttention", "Module", "The", "ValueError", "__init__", "a", "all_head_size", "and", "attention", "attention_head_size", "attention_probs_dropout_prob", "class", "config", "def", "dropout", "embedding_size", "f", "hasattr", "heads", "hidden", "hidden_size", "if", "int", "is", "is_decoder", "key", "multiple", "nn", "not", "num_attention_heads", "number", "of", "query", "raise", "self", "size", "the", "value"], "superglue/modeling_superglue.py:SuperGlueSelfAttention.forward": ["False", "FloatTensor", "ModelSelfAttention", "Module", "None", "Tensor", "all_head_size", "attention_head_size", "attention_mask", "attention_probs", "attention_scores", "batch_size", "bool", "class", "context_layer", "contiguous", "current_states", "def", "dim", "dropout", "else", "encoder_attention_mask", "encoder_hidden_states", "forward", "functional", "hidden_states", "if", "is", "is_cross_attention", "is_decoder", "key", "key_layer", "math", "matmul", "new_context_layer_shape", "nn", "not", "num_attention_heads", "output_attentions", "outputs", "permute", "query", "query_layer", "return", "self", "shape", "size", "softmax", "sqrt", "torch", "transpose", "tuple", "value", "value_layer", "view"], "superglue/modeling_superglue.py:SuperGlueSelfOutput.__init__": ["Linear", "Model", "ModelConfig", "ModelSelfOutput", "Module", "__init__", "class", "config", "def", "dense", "hidden_size", "nn", "self"], "superglue/modeling_superglue.py:SuperGlueSelfOutput.forward": ["ModelSelfOutput", "Module", "Tensor", "args", "class", "def", "dense", "forward", "hidden_states", "nn", "return", "self", "torch"], "superglue/modeling_superglue.py:SuperGlueAttention.__init__": ["Model", "ModelAttention", "ModelSelfOutput", "Model_SELF_ATTENTION_CLASSES", "Module", "__init__", "_attn_implementation", "class", "config", "def", "nn", "output", "self"], "superglue/modeling_superglue.py:SuperGlueAttention.forward": ["False", "FloatTensor", "ModelAttention", "Module", "None", "Tensor", "attention_mask", "attention_output", "bool", "class", "def", "encoder_attention_mask", "encoder_hidden_states", "forward", "hidden_states", "nn", "output", "output_attentions", "outputs", "return", "self", "self_outputs", "torch", "tuple"], "superglue/modeling_superglue.py:SuperGlueAttentionalPropagation.__init__": ["Linear", "Model", "ModelAttention", "ModelAttentionalPropagation", "ModelConfig", "ModelMultiLayerPerceptron", "Module", "ModuleList", "None", "__init__", "append", "attention", "class", "config", "def", "for", "hidden_size", "i", "in", "layers", "len", "mlp", "mlp_channels", "nn", "range", "self"], "superglue/modeling_superglue.py:SuperGlueAttentionalPropagation.forward": ["False", "ModelAttentionalPropagation", "Module", "None", "Tensor", "all_hidden_states", "attention", "attention_mask", "attention_outputs", "bool", "cat", "class", "def", "descriptors", "dim", "else", "encoder_attention_mask", "encoder_hidden_states", "for", "forward", "hidden_state", "if", "in", "layer", "mlp", "nn", "output", "output_attentions", "output_hidden_states", "return", "self", "torch", "tuple"], "superglue/modeling_superglue.py:SuperGlueAttentionalGNN.__init__": ["Model", "ModelAttentionalGNN", "ModelAttentionalPropagation", "ModelConfig", "Module", "ModuleList", "None", "_", "__init__", "class", "config", "def", "for", "gnn_layers_types", "hidden_size", "in", "layers", "layers_types", "len", "nn", "range", "self"], "superglue/modeling_superglue.py:SuperGlueAttentionalGNN.forward": ["False", "ModelAttentionalGNN", "Module", "None", "Tensor", "_", "all_attentions", "all_hidden_states", "attention_mask", "batch_size", "bool", "class", "cross", "def", "delta", "descriptors", "else", "encoder_attention_mask", "encoder_hidden_states", "flip", "for", "forward", "gnn_layer", "gnn_outputs", "hidden_size", "if", "in", "is", "layer_type", "layers", "layers_types", "mask", "nn", "not", "num_keypoints", "output_attentions", "output_hidden_states", "reshape", "return", "self", "shape", "torch", "tuple", "zip"], "superglue/modeling_superglue.py:SuperGlueFinalProjection.__init__": ["Linear", "Model", "ModelConfig", "ModelFinalProjection", "Module", "None", "True", "__init__", "bias", "class", "config", "def", "final_proj", "hidden_size", "nn", "self"], "superglue/modeling_superglue.py:SuperGlueFinalProjection.forward": ["ModelFinalProjection", "Module", "Tensor", "class", "def", "descriptors", "final_proj", "forward", "nn", "return", "self", "torch"], "superglue/modeling_superglue.py:SuperGluePreTrainedModel._init_weights": ["Model", "ModelPreTrainedModel", "Module", "None", "PreTrainedModel", "_init_weights", "bin_score", "class", "def", "hasattr", "if", "init", "module", "nn", "ones_", "self"], "superglue/modeling_superglue.py:SuperGlueForKeypointMatching.__init__": ["AutoModelForKeypointDetection", "Model", "ModelAttentionalGNN", "ModelConfig", "ModelFinalProjection", "ModelForKeypointMatching", "ModelKeypointEncoder", "ModelPreTrainedModel", "None", "Parameter", "__init__", "bin_score", "class", "config", "def", "final_projection", "from_config", "gnn", "keypoint_detector", "keypoint_detector_config", "keypoint_encoder", "nn", "post_init", "register_parameter", "self", "tensor", "torch"], "superglue/modeling_superglue.py:SuperGlueForKeypointMatching._match_image_pair": ["ModelForKeypointMatching", "ModelPreTrainedModel", "None", "Tensor", "_", "_match_image_pair", "all_attentions", "all_hidden_states", "arange_like", "batch_size", "bin_score", "bool", "cat", "class", "config", "def", "descriptors", "device", "dim", "dtype", "else", "encoded_keypoints", "exp", "extended_attention_mask", "final_descriptors", "final_descriptors0", "final_descriptors1", "final_projection", "finfo", "for", "gather", "get_extended_attention_mask", "gnn", "gnn_outputs", "height", "hidden_size", "if", "in", "indices", "indices0", "indices1", "input_shape", "int", "is", "iterations", "keypoint_encoder", "keypoints", "last_hidden_state", "log_optimal_transport", "logical_and", "mask", "mask0", "mask1", "masked_fill", "matches", "matches0", "matches1", "matching_scores", "matching_scores0", "matching_scores1", "matching_threshold", "max", "max0", "max1", "min", "mutual0", "mutual1", "new_full", "new_tensor", "new_zeros", "normalize_keypoints", "not", "num_keypoints", "ones", "output_attentions", "output_hidden_states", "projected_descriptors", "reshape", "return", "scores", "self", "shape", "sinkhorn_iterations", "size", "torch", "transpose", "tuple", "unsqueeze", "valid0", "valid1", "values", "where", "width", "x", "zero"], "superglue/modeling_superglue.py:SuperGlueForKeypointMatching.forward": ["FloatTensor", "Input", "LongTensor", "Model", "ModelForKeypointMatching", "ModelKeypointMatchingOutput", "ModelPreTrainedModel", "None", "ValueError", "_", "_match_image_pair", "a", "absolute_keypoints", "attentions", "batch_size", "be", "bool", "channels", "class", "clone", "config", "def", "descriptors", "else", "for", "forward", "height", "hidden_size", "hidden_states", "if", "in", "is", "keypoint_detections", "keypoint_detector", "keypoints", "kwargs", "labels", "loss", "mask", "matches", "matching_scores", "must", "ndim", "no", "not", "num_channels", "of", "or", "output_attentions", "output_hidden_states", "pixel_values", "provided", "r", "raise", "reshape", "return", "return_dict", "scores", "self", "shape", "should", "size", "tensor", "to", "torch", "trainable", "tuple", "use_return_dict", "v", "width"], "superpoint/modeling_superpoint.py:remove_keypoints_from_borders": ["Model_keypoints_from_borders", "Tensor", "border", "def", "height", "int", "keypoints", "mask", "mask_h", "mask_w", "return", "scores", "torch", "tuple", "width"], "superpoint/modeling_superpoint.py:top_k_keypoints": ["Model_k_keypoints", "Modelk", "Tensor", "def", "dim", "if", "indices", "int", "k", "keypoints", "len", "return", "scores", "torch", "tuple"], "superpoint/modeling_superpoint.py:simple_nms": ["Expected", "Model_nms", "Tensor", "ValueError", "_", "def", "float", "for", "functional", "if", "in", "int", "kernel_size", "max_mask", "max_pool", "max_pool2d", "new_max_mask", "nms_radius", "nn", "padding", "positive", "raise", "range", "return", "scores", "stride", "supp_mask", "supp_scores", "torch", "values", "where", "x", "zeros", "zeros_like"], "superpoint/modeling_superpoint.py:SuperPointConvBlock.__init__": ["Conv2d", "False", "MaxPool2d", "Model", "ModelConfig", "ModelConvBlock", "Module", "None", "ReLU", "True", "__init__", "add_pooling", "bool", "class", "config", "conv_a", "conv_b", "def", "else", "if", "in_channels", "inplace", "int", "kernel_size", "nn", "out_channels", "padding", "pool", "relu", "self", "stride"], "superpoint/modeling_superpoint.py:SuperPointConvBlock.forward": ["ModelConvBlock", "Module", "None", "Tensor", "class", "conv_a", "conv_b", "def", "forward", "hidden_states", "if", "is", "nn", "not", "pool", "relu", "return", "self", "torch"], "superpoint/modeling_superpoint.py:SuperPointEncoder.__init__": ["False", "Model", "ModelConfig", "ModelConvBlock", "ModelEncoder", "Module", "ModuleList", "None", "True", "__init__", "add_pooling", "append", "class", "config", "conv_blocks", "def", "encoder_hidden_sizes", "for", "i", "in", "input_dim", "len", "nn", "range", "self"], "superpoint/modeling_superpoint.py:SuperPointEncoder.forward": ["BaseModelOutputWithNoAttention", "False", "ModelEncoder", "Module", "None", "True", "all_hidden_states", "bool", "class", "conv_block", "conv_blocks", "def", "else", "for", "forward", "hidden_states", "if", "in", "input", "is", "last_hidden_state", "nn", "not", "output", "output_hidden_states", "return", "return_dict", "self", "tuple", "v"], "superpoint/modeling_superpoint.py:SuperPointInterestPointDecoder.__init__": ["Conv2d", "MaxPool2d", "Model", "ModelConfig", "ModelInterestPointDecoder", "Module", "None", "ReLU", "True", "__init__", "border_removal_distance", "class", "config", "conv_score_a", "conv_score_b", "decoder_hidden_size", "def", "encoder_hidden_sizes", "inplace", "kernel_size", "keypoint_decoder_dim", "keypoint_threshold", "max_keypoints", "nms_radius", "nn", "padding", "pool", "relu", "self", "stride"], "superpoint/modeling_superpoint.py:SuperPointInterestPointDecoder.forward": ["ModelInterestPointDecoder", "Module", "Tensor", "_extract_keypoints", "_get_pixel_scores", "class", "def", "encoded", "forward", "keypoints", "nn", "return", "scores", "self", "torch", "tuple"], "superpoint/modeling_superpoint.py:SuperPointInterestPointDecoder._get_pixel_scores": ["ModelInterestPointDecoder", "Module", "Tensor", "_", "_get_pixel_scores", "batch_size", "class", "conv_score_a", "conv_score_b", "def", "encoded", "functional", "height", "nms_radius", "nn", "permute", "relu", "reshape", "return", "scores", "self", "shape", "simple_nms", "softmax", "torch", "width"], "superpoint/modeling_superpoint.py:SuperPointInterestPointDecoder._extract_keypoints": ["ModelInterestPointDecoder", "Module", "Tensor", "_", "_extract_keypoints", "border_removal_distance", "class", "def", "dtype", "flip", "height", "if", "keypoint_threshold", "keypoints", "max_keypoints", "nn", "nonzero", "remove_keypoints_from_borders", "return", "scores", "self", "shape", "t", "to", "top_k_keypoints", "torch", "tuple", "width"], "superpoint/modeling_superpoint.py:SuperPointDescriptorDecoder.__init__": ["Conv2d", "MaxPool2d", "Model", "ModelConfig", "ModelDescriptorDecoder", "Module", "None", "ReLU", "True", "__init__", "class", "config", "conv_descriptor_a", "conv_descriptor_b", "decoder_hidden_size", "def", "descriptor_decoder_dim", "encoder_hidden_sizes", "inplace", "kernel_size", "nn", "padding", "pool", "relu", "self", "stride"], "superpoint/modeling_superpoint.py:SuperPointDescriptorDecoder.forward": ["ModelDescriptorDecoder", "Module", "None", "Tensor", "_sample_descriptors", "class", "conv_descriptor_a", "conv_descriptor_b", "def", "descriptors", "dim", "encoded", "forward", "functional", "keypoints", "nn", "normalize", "p", "relu", "return", "self", "torch", "transpose"], "superpoint/modeling_superpoint.py:SuperPointDescriptorDecoder._sample_descriptors": ["ModelDescriptorDecoder", "Module", "Tensor", "True", "_sample_descriptors", "align_corners", "batch_size", "bilinear", "class", "def", "descriptors", "dim", "divisor", "functional", "grid_sample", "height", "int", "keypoints", "kwargs", "mode", "nn", "normalize", "num_channels", "p", "reshape", "return", "scale", "shape", "tensor", "to", "torch", "view", "width"], "superpoint/modeling_superpoint.py:SuperPointPreTrainedModel.extract_one_channel_pixel_values": ["FloatTensor", "ModelPreTrainedModel", "None", "PreTrainedModel", "class", "def", "extract_one_channel_pixel_values", "pixel_values", "return", "self", "torch"], "superpoint/modeling_superpoint.py:SuperPointForKeypointDetection.__init__": ["Model", "ModelConfig", "ModelDescriptorDecoder", "ModelEncoder", "ModelForKeypointDetection", "ModelInterestPointDecoder", "ModelPreTrainedModel", "None", "__init__", "class", "config", "def", "descriptor_decoder", "encoder", "keypoint_decoder", "post_init", "self"], "superpoint/modeling_superpoint.py:SuperPointForKeypointDetection.forward": ["FloatTensor", "LongTensor", "Model", "ModelForKeypointDetection", "ModelKeypointDescriptionOutput", "ModelPreTrainedModel", "None", "ValueError", "_", "_descriptors", "_keypoints", "_scores", "batch_size", "bool", "class", "config", "def", "descriptor_decoder", "descriptor_decoder_dim", "descriptors", "device", "does", "dtype", "else", "encoder", "encoder_outputs", "enumerate", "extract_one_channel_pixel_values", "for", "forward", "height", "hidden_states", "i", "if", "in", "int", "is", "keypoint_decoder", "keypoints", "keypoints_scores", "kwargs", "labels", "last_hidden_state", "list_descriptors", "list_keypoints", "list_keypoints_scores", "list_scores", "loss", "mask", "max", "maximum_num_keypoints", "not", "now", "output_hidden_states", "pixel_values", "r", "raise", "return", "return_dict", "scores", "self", "shape", "support", "tensor", "torch", "training", "tuple", "use_return_dict", "v", "width", "zeros", "zip"], "swiftformer/modeling_swiftformer.py:SwiftFormerPatchEmbedding.__init__": ["BatchNorm2d", "Conv2d", "ModelConfig", "ModelPatchEmbedding", "Module", "ReLU", "Sequential", "__init__", "batch_norm_eps", "class", "config", "def", "embed_dims", "eps", "in_chs", "kernel_size", "nn", "num_channels", "out_chs", "padding", "patch_embedding", "self", "stride", "super"], "swiftformer/modeling_swiftformer.py:SwiftFormerPatchEmbedding.forward": ["ModelPatchEmbedding", "Module", "class", "def", "forward", "nn", "patch_embedding", "return", "self", "x"], "swiftformer/modeling_swiftformer.py:drop_path": ["False", "Model_path", "Model_prob", "Tensor", "bool", "def", "device", "div", "dtype", "float", "floor_", "if", "input", "keep_prob", "ndim", "not", "or", "output", "rand", "random_tensor", "return", "shape", "torch", "training"], "swiftformer/modeling_swiftformer.py:SwiftFormerDropPath.__init__": ["ModelConfig", "ModelDropPath", "Module", "None", "__init__", "class", "config", "def", "drop_path_rate", "drop_prob", "nn", "self", "super"], "swiftformer/modeling_swiftformer.py:SwiftFormerDropPath.forward": ["ModelDropPath", "Module", "Tensor", "class", "def", "drop_path", "drop_prob", "forward", "hidden_states", "nn", "return", "self", "torch", "training"], "swiftformer/modeling_swiftformer.py:SwiftFormerDropPath.extra_repr": ["ModelDropPath", "Module", "class", "def", "drop_prob", "extra_repr", "f", "nn", "p", "return", "self", "str"], "swiftformer/modeling_swiftformer.py:SwiftFormerEmbeddings.__init__": ["BatchNorm2d", "Conv2d", "Iterable", "ModelConfig", "ModelEmbeddings", "Module", "__init__", "abc", "batch_norm_eps", "class", "collections", "config", "def", "down_pad", "down_patch_size", "down_stride", "else", "embed_dim", "embed_dims", "eps", "if", "in_chans", "index", "int", "isinstance", "kernel_size", "nn", "norm", "padding", "patch_size", "proj", "self", "stride", "super"], "swiftformer/modeling_swiftformer.py:SwiftFormerEmbeddings.forward": ["ModelEmbeddings", "Module", "class", "def", "forward", "nn", "norm", "proj", "return", "self", "x"], "swiftformer/modeling_swiftformer.py:SwiftFormerConvEncoder.__init__": ["BatchNorm2d", "Conv2d", "Dropout", "GELU", "ModelConfig", "ModelConvEncoder", "Module", "Parameter", "True", "__init__", "act", "batch_norm_eps", "class", "config", "def", "depth_wise_conv", "dim", "drop_conv_encoder_rate", "drop_path", "eps", "groups", "hidden_dim", "int", "kernel_size", "layer_scale", "mlp_ratio", "nn", "norm", "ones", "p", "padding", "point_wise_conv1", "point_wise_conv2", "requires_grad", "self", "super", "torch", "unsqueeze"], "swiftformer/modeling_swiftformer.py:SwiftFormerConvEncoder.forward": ["ModelConvEncoder", "Module", "act", "class", "def", "depth_wise_conv", "drop_path", "forward", "input", "layer_scale", "nn", "norm", "point_wise_conv1", "point_wise_conv2", "return", "self", "x"], "swiftformer/modeling_swiftformer.py:SwiftFormerMlp.__init__": ["ACT2CLS", "BatchNorm2d", "Conv2d", "Dropout", "ModelConfig", "ModelMlp", "Module", "__init__", "act", "act_layer", "batch_norm_eps", "class", "config", "def", "drop", "drop_mlp_rate", "eps", "fc1", "fc2", "hidden_act", "hidden_features", "in_features", "int", "mlp_ratio", "nn", "norm1", "p", "self", "super"], "swiftformer/modeling_swiftformer.py:SwiftFormerMlp.forward": ["ModelMlp", "Module", "act", "class", "def", "drop", "fc1", "fc2", "forward", "nn", "norm1", "return", "self", "x"], "swiftformer/modeling_swiftformer.py:SwiftFormerEfficientAdditiveAttention.__init__": ["Linear", "ModelConfig", "ModelEfficientAdditiveAttention", "Module", "Parameter", "__init__", "class", "config", "def", "dim", "final", "int", "nn", "proj", "randn", "scale_factor", "self", "super", "to_key", "to_query", "torch", "w_g"], "swiftformer/modeling_swiftformer.py:SwiftFormerEfficientAdditiveAttention.forward": ["ModelEfficientAdditiveAttention", "Module", "class", "def", "dim", "final", "forward", "functional", "global_queries", "key", "nn", "normalize", "out", "proj", "query", "query_weight", "repeat", "return", "scale_factor", "scaled_query_weight", "self", "shape", "softmax", "sum", "to_key", "to_query", "torch", "unsqueeze", "w_g", "x"], "swiftformer/modeling_swiftformer.py:SwiftFormerLocalRepresentation.__init__": ["BatchNorm2d", "Conv2d", "GELU", "Identity", "ModelConfig", "ModelLocalRepresentation", "Module", "Parameter", "True", "__init__", "act", "batch_norm_eps", "class", "config", "def", "depth_wise_conv", "dim", "drop_path", "eps", "groups", "int", "kernel_size", "layer_scale", "nn", "norm", "ones", "padding", "point_wise_conv1", "point_wise_conv2", "requires_grad", "self", "super", "torch", "unsqueeze"], "swiftformer/modeling_swiftformer.py:SwiftFormerLocalRepresentation.forward": ["ModelLocalRepresentation", "Module", "act", "class", "def", "depth_wise_conv", "drop_path", "forward", "input", "layer_scale", "nn", "norm", "point_wise_conv1", "point_wise_conv2", "return", "self", "x"], "swiftformer/modeling_swiftformer.py:SwiftFormerEncoderBlock.__init__": ["Identity", "ModelConfig", "ModelDropPath", "ModelEfficientAdditiveAttention", "ModelEncoderBlock", "ModelLocalRepresentation", "ModelMlp", "Module", "None", "Parameter", "True", "__init__", "attn", "class", "config", "def", "dim", "drop_path", "else", "float", "if", "in_features", "int", "layer_scale_1", "layer_scale_2", "layer_scale_init_value", "linear", "local_representation", "nn", "ones", "requires_grad", "self", "super", "torch", "unsqueeze", "use_layer_scale"], "swiftformer/modeling_swiftformer.py:SwiftFormerEncoderBlock.forward": ["ModelEncoderBlock", "Module", "attn", "batch_size", "channels", "class", "def", "drop_path", "else", "forward", "height", "if", "layer_scale_1", "layer_scale_2", "linear", "local_representation", "nn", "permute", "res", "reshape", "return", "self", "shape", "use_layer_scale", "width", "x"], "swiftformer/modeling_swiftformer.py:SwiftFormerStage.__init__": ["GradientCheckpointingLayer", "ModelConfig", "ModelConvEncoder", "ModelEncoderBlock", "ModelStage", "ModuleList", "None", "__init__", "append", "block_dpr", "block_idx", "blocks", "class", "config", "def", "depth", "depths", "dim", "drop_path", "drop_path_rate", "else", "embed_dims", "for", "if", "in", "index", "int", "layer_depths", "nn", "range", "self", "sum", "super"], "swiftformer/modeling_swiftformer.py:SwiftFormerStage.forward": ["GradientCheckpointingLayer", "ModelStage", "block", "blocks", "class", "def", "for", "forward", "in", "input", "return", "self"], "swiftformer/modeling_swiftformer.py:SwiftFormerEncoder.__init__": ["False", "ModelConfig", "ModelEmbeddings", "ModelEncoder", "ModelStage", "Module", "ModuleList", "None", "__init__", "append", "break", "class", "config", "def", "depths", "downsamples", "embed_dims", "for", "gradient_checkpointing", "i", "if", "in", "index", "layer_depths", "len", "network", "nn", "or", "range", "self", "stage", "super"], "swiftformer/modeling_swiftformer.py:SwiftFormerEncoder.forward": ["BaseModelOutputWithNoAttention", "ModelEncoder", "Module", "None", "Tensor", "all_hidden_states", "block", "bool", "class", "config", "def", "else", "for", "forward", "hidden_states", "if", "in", "is", "last_hidden_state", "network", "nn", "not", "output_hidden_states", "return", "return_dict", "self", "torch", "tuple", "use_return_dict", "v"], "swiftformer/modeling_swiftformer.py:SwiftFormerPreTrainedModel._init_weights": ["BatchNorm2d", "Conv2d", "LayerNorm", "Linear", "ModelConvEncoder", "ModelEfficientAdditiveAttention", "ModelEncoderBlock", "ModelLocalRepresentation", "ModelPreTrainedModel", "Module", "None", "PreTrainedModel", "_init_weights", "bias", "class", "config", "constant_", "def", "elif", "getattr", "if", "init", "is", "isinstance", "layer_scale", "layer_scale_1", "layer_scale_2", "layer_scale_init_value", "module", "nn", "normal_", "not", "num_batches_tracked", "ones_", "running_mean", "running_var", "self", "std", "trunc_normal_", "use_layer_scale", "w_g", "weight", "zeros_"], "swiftformer/modeling_swiftformer.py:SwiftFormerModel.__init__": ["ModelConfig", "ModelEncoder", "ModelModel", "ModelPatchEmbedding", "ModelPreTrainedModel", "__init__", "class", "config", "def", "encoder", "patch_embed", "post_init", "self", "super"], "swiftformer/modeling_swiftformer.py:SwiftFormerModel.forward": ["BaseModelOutputWithNoAttention", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "ValueError", "You", "bool", "class", "config", "def", "else", "embedding_output", "encoder", "encoder_outputs", "for", "forward", "have", "hidden_states", "if", "in", "is", "kwargs", "last_hidden_state", "not", "output_hidden_states", "patch_embed", "pixel_values", "raise", "return", "return_dict", "self", "specify", "to", "torch", "tuple", "use_return_dict", "v"], "swiftformer/modeling_swiftformer.py:SwiftFormerForImageClassification.__init__": ["BatchNorm2d", "Identity", "Linear", "Model", "ModelConfig", "ModelForImageClassification", "ModelModel", "ModelPreTrainedModel", "None", "__init__", "batch_norm_eps", "class", "config", "def", "dist_head", "else", "embed_dims", "eps", "head", "if", "nn", "norm", "num_labels", "post_init", "self", "super"], "swiftformer/modeling_swiftformer.py:SwiftFormerForImageClassification.forward": ["ImageClassifierOutputWithNoAttention", "Model", "ModelForImageClassification", "ModelPreTrainedModel", "None", "Tensor", "bool", "class", "cls_out", "config", "def", "dist_head", "distillation_out", "else", "flatten", "forward", "head", "hidden_states", "if", "is", "kwargs", "labels", "last_hidden_state", "logits", "loss", "loss_function", "mean", "norm", "not", "output", "output_hidden_states", "outputs", "pixel_values", "r", "return", "return_dict", "self", "sequence_output", "torch", "tuple", "use_return_dict"], "swin/modeling_swin.py:window_partition": ["Model_partition", "Model_size", "Models", "batch_size", "contiguous", "def", "height", "input_feature", "num_channels", "permute", "return", "shape", "view", "width"], "swin/modeling_swin.py:window_reverse": ["Model_reverse", "Model_size", "Models", "contiguous", "def", "height", "num_channels", "permute", "return", "shape", "view", "width"], "swin/modeling_swin.py:SwinEmbeddings.__init__": ["Dropout", "False", "LayerNorm", "ModelEmbeddings", "ModelPatchEmbeddings", "Module", "None", "Parameter", "__init__", "class", "config", "def", "dropout", "else", "embed_dim", "grid_size", "hidden_dropout_prob", "if", "mask_token", "nn", "norm", "num_patches", "patch_embeddings", "patch_grid", "patch_size", "position_embeddings", "self", "super", "torch", "use_absolute_embeddings", "use_mask_token", "zeros"], "swin/modeling_swin.py:SwinEmbeddings.interpolate_pos_encoding": ["False", "ModelEmbeddings", "Module", "Tensor", "align_corners", "and", "bicubic", "cat", "class", "class_pos_embed", "def", "dim", "embeddings", "functional", "height", "if", "int", "interpolate", "interpolate_pos_encoding", "is_tracing", "jit", "mode", "new_height", "new_width", "nn", "not", "num_patches", "num_positions", "patch_pos_embed", "patch_size", "permute", "position_embeddings", "reshape", "return", "self", "shape", "size", "sqrt_num_positions", "torch", "torch_int", "view", "width"], "swin/modeling_swin.py:SwinEmbeddings.forward": ["BoolTensor", "False", "FloatTensor", "ModelEmbeddings", "Module", "None", "Tensor", "_", "batch_size", "bool", "bool_masked_pos", "class", "def", "dropout", "else", "embeddings", "expand", "forward", "height", "if", "interpolate_pos_encoding", "is", "mask", "mask_token", "mask_tokens", "nn", "norm", "not", "num_channels", "output_dimensions", "patch_embeddings", "pixel_values", "position_embeddings", "return", "self", "seq_len", "shape", "size", "torch", "tuple", "type_as", "unsqueeze", "width"], "swin/modeling_swin.py:SwinPatchEmbeddings.__init__": ["Conv2d", "Iterable", "ModelPatchEmbeddings", "Module", "__init__", "abc", "class", "collections", "config", "def", "else", "embed_dim", "grid_size", "hidden_size", "if", "image_size", "isinstance", "kernel_size", "nn", "num_channels", "num_patches", "patch_size", "projection", "self", "stride", "super"], "swin/modeling_swin.py:SwinPatchEmbeddings.maybe_pad": ["ModelPatchEmbeddings", "Module", "class", "def", "functional", "height", "if", "maybe_pad", "nn", "pad", "pad_values", "patch_size", "pixel_values", "return", "self", "width"], "swin/modeling_swin.py:SwinPatchEmbeddings.forward": ["FloatTensor", "ModelPatchEmbeddings", "Module", "None", "Tensor", "_", "class", "def", "embeddings", "flatten", "forward", "height", "int", "maybe_pad", "nn", "num_channels", "output_dimensions", "pixel_values", "projection", "return", "self", "shape", "torch", "transpose", "tuple", "width"], "swin/modeling_swin.py:SwinPatchMerging.__init__": ["False", "LayerNorm", "Linear", "ModelPatchMerging", "Module", "None", "__init__", "bias", "class", "def", "dim", "input_resolution", "int", "nn", "norm", "norm_layer", "reduction", "self", "super", "tuple"], "swin/modeling_swin.py:SwinPatchMerging.maybe_pad": ["ModelPatchMerging", "Module", "class", "def", "functional", "height", "if", "input_feature", "maybe_pad", "nn", "or", "pad", "pad_values", "return", "self", "should_pad", "width"], "swin/modeling_swin.py:SwinPatchMerging.forward": ["ModelPatchMerging", "Module", "Tensor", "batch_size", "cat", "class", "def", "dim", "forward", "height", "input_dimensions", "input_feature", "input_feature_0", "input_feature_1", "input_feature_2", "input_feature_3", "int", "maybe_pad", "nn", "norm", "num_channels", "reduction", "return", "self", "shape", "torch", "tuple", "view", "width"], "swin/modeling_swin.py:drop_path": ["False", "Model_path", "Model_prob", "Tensor", "bool", "def", "device", "div", "dtype", "float", "floor_", "if", "input", "keep_prob", "ndim", "not", "or", "output", "rand", "random_tensor", "return", "shape", "torch", "training"], "swin/modeling_swin.py:SwinDropPath.__init__": ["ModelDropPath", "Module", "None", "__init__", "class", "def", "drop_prob", "float", "nn", "self", "super"], "swin/modeling_swin.py:SwinDropPath.forward": ["ModelDropPath", "Module", "Tensor", "class", "def", "drop_path", "drop_prob", "forward", "hidden_states", "nn", "return", "self", "torch", "training"], "swin/modeling_swin.py:SwinDropPath.extra_repr": ["ModelDropPath", "Module", "class", "def", "drop_prob", "extra_repr", "f", "nn", "p", "return", "self", "str"], "swin/modeling_swin.py:SwinSelfAttention.__init__": ["Dropout", "Iterable", "Linear", "ModelSelfAttention", "Module", "Parameter", "The", "ValueError", "__init__", "a", "abc", "all_head_size", "attention", "attention_head_size", "attention_probs_dropout_prob", "bias", "class", "collections", "config", "create_relative_position_index", "def", "dim", "dropout", "else", "f", "heads", "hidden", "if", "int", "is", "isinstance", "key", "multiple", "nn", "not", "num_attention_heads", "num_heads", "number", "of", "qkv_bias", "query", "raise", "register_buffer", "relative_position_bias_table", "relative_position_index", "self", "size", "super", "the", "torch", "value", "window_size", "zeros"], "swin/modeling_swin.py:SwinSelfAttention.forward": ["False", "FloatTensor", "ModelSelfAttention", "Module", "None", "Tensor", "all_head_size", "attention_head_size", "attention_mask", "attention_probs", "attention_scores", "batch_size", "bool", "class", "context_layer", "contiguous", "def", "dim", "dropout", "else", "forward", "functional", "hidden_shape", "hidden_states", "if", "is", "key", "key_layer", "mask_shape", "math", "matmul", "new_context_layer_shape", "nn", "not", "num_attention_heads", "num_channels", "output_attentions", "outputs", "permute", "query", "query_layer", "relative_position_bias", "relative_position_bias_table", "relative_position_index", "return", "self", "shape", "size", "softmax", "sqrt", "torch", "transpose", "tuple", "unsqueeze", "value", "value_layer", "view", "window_size"], "swin/modeling_swin.py:SwinSelfAttention.create_relative_position_index": ["ModelSelfAttention", "Module", "None", "arange", "class", "contiguous", "coords", "coords_flatten", "coords_h", "coords_w", "create_relative_position_index", "def", "flatten", "ij", "indexing", "meshgrid", "nn", "permute", "relative_coords", "relative_position_index", "return", "self", "stack", "sum", "torch", "window_size"], "swin/modeling_swin.py:SwinSelfOutput.__init__": ["Dropout", "Linear", "ModelSelfOutput", "Module", "__init__", "attention_probs_dropout_prob", "class", "config", "def", "dense", "dim", "dropout", "nn", "self", "super"], "swin/modeling_swin.py:SwinSelfOutput.forward": ["ModelSelfOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "swin/modeling_swin.py:SwinAttention.__init__": ["ModelAttention", "ModelSelfAttention", "ModelSelfOutput", "Module", "__init__", "class", "config", "def", "dim", "nn", "num_heads", "output", "self", "super", "window_size"], "swin/modeling_swin.py:SwinAttention.forward": ["False", "FloatTensor", "ModelAttention", "Module", "None", "Tensor", "attention_mask", "attention_output", "bool", "class", "def", "forward", "hidden_states", "nn", "output", "output_attentions", "outputs", "return", "self", "self_outputs", "torch", "tuple"], "swin/modeling_swin.py:SwinIntermediate.__init__": ["ACT2FN", "Linear", "ModelIntermediate", "Module", "__init__", "class", "config", "def", "dense", "dim", "else", "hidden_act", "if", "int", "intermediate_act_fn", "isinstance", "mlp_ratio", "nn", "self", "str", "super"], "swin/modeling_swin.py:SwinIntermediate.forward": ["ModelIntermediate", "Module", "Tensor", "class", "def", "dense", "forward", "hidden_states", "intermediate_act_fn", "nn", "return", "self", "torch"], "swin/modeling_swin.py:SwinOutput.__init__": ["Dropout", "Linear", "ModelOutput", "Module", "__init__", "class", "config", "def", "dense", "dim", "dropout", "hidden_dropout_prob", "int", "mlp_ratio", "nn", "self", "super"], "swin/modeling_swin.py:SwinOutput.forward": ["ModelOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "nn", "return", "self", "torch"], "swin/modeling_swin.py:SwinLayer.__init__": ["Identity", "LayerNorm", "ModelAttention", "ModelDropPath", "ModelIntermediate", "ModelLayer", "ModelOutput", "Module", "__init__", "attention", "chunk_size_feed_forward", "class", "config", "def", "dim", "drop_path", "drop_path_rate", "else", "eps", "if", "input_resolution", "intermediate", "layer_norm_eps", "layernorm_after", "layernorm_before", "nn", "num_heads", "output", "self", "shift_size", "super", "window_size"], "swin/modeling_swin.py:SwinLayer.set_shift_and_window_size": ["ModelLayer", "Module", "class", "def", "else", "if", "input_resolution", "is_tracing", "jit", "min", "nn", "self", "set_shift_and_window_size", "shift_size", "tensor", "torch", "torch_int", "window_size"], "swin/modeling_swin.py:SwinLayer.get_attn_mask": ["ModelLayer", "Module", "None", "attn_mask", "class", "count", "def", "device", "dtype", "else", "for", "get_attn_mask", "height", "height_slice", "height_slices", "if", "img_mask", "in", "mask_windows", "masked_fill", "nn", "return", "self", "shift_size", "slice", "torch", "unsqueeze", "view", "width", "width_slice", "width_slices", "window_partition", "window_size", "zeros"], "swin/modeling_swin.py:SwinLayer.maybe_pad": ["ModelLayer", "Module", "class", "def", "functional", "height", "hidden_states", "maybe_pad", "nn", "pad", "pad_bottom", "pad_right", "pad_values", "return", "self", "width", "window_size"], "swin/modeling_swin.py:SwinLayer.forward": ["False", "ModelLayer", "Module", "None", "Tensor", "_", "always_partition", "attention", "attention_output", "attention_outputs", "attention_windows", "attn_mask", "batch_size", "bool", "channels", "class", "contiguous", "def", "device", "dims", "drop_path", "dtype", "else", "forward", "get_attn_mask", "height", "height_pad", "hidden_states", "hidden_states_windows", "if", "input_dimensions", "int", "intermediate", "layer_output", "layer_outputs", "layernorm_after", "layernorm_before", "maybe_pad", "nn", "not", "or", "output", "output_attentions", "pad_values", "pass", "return", "roll", "self", "set_shift_and_window_size", "shape", "shift_size", "shifted_hidden_states", "shifted_windows", "shifts", "shortcut", "size", "torch", "tuple", "view", "was_padded", "width", "width_pad", "window_partition", "window_reverse", "window_size"], "swin/modeling_swin.py:SwinStage.__init__": ["False", "GradientCheckpointingLayer", "LayerNorm", "ModelLayer", "ModelStage", "ModuleList", "None", "__init__", "blocks", "class", "config", "def", "depth", "dim", "downsample", "drop_path", "drop_path_rate", "else", "for", "i", "if", "in", "input_resolution", "is", "nn", "norm_layer", "not", "num_heads", "pointing", "range", "self", "shift_size", "super", "window_size"], "swin/modeling_swin.py:SwinStage.forward": ["False", "GradientCheckpointingLayer", "ModelStage", "None", "Tensor", "always_partition", "blocks", "bool", "class", "def", "downsample", "else", "enumerate", "for", "forward", "height", "height_downsampled", "hidden_states", "hidden_states_before_downsampling", "i", "if", "in", "input_dimensions", "int", "is", "layer_module", "layer_outputs", "not", "output_attentions", "output_dimensions", "return", "self", "stage_outputs", "torch", "tuple", "width", "width_downsampled"], "swin/modeling_swin.py:SwinEncoder.__init__": ["False", "ModelEncoder", "ModelPatchMerging", "ModelStage", "Module", "ModuleList", "None", "__init__", "class", "config", "cpu", "def", "depth", "depths", "device", "dim", "downsample", "dpr", "drop_path", "drop_path_rate", "else", "embed_dim", "for", "gradient_checkpointing", "grid_size", "i_layer", "if", "in", "input_resolution", "int", "item", "layers", "len", "linspace", "nn", "num_heads", "num_layers", "range", "self", "sum", "super", "torch", "x"], "swin/modeling_swin.py:SwinEncoder.forward": ["False", "ModelEncoder", "ModelEncoderOutput", "Module", "None", "Tensor", "True", "_", "all_hidden_states", "all_reshaped_hidden_states", "all_self_attentions", "always_partition", "and", "attentions", "batch_size", "bool", "class", "def", "elif", "else", "enumerate", "for", "forward", "hidden_size", "hidden_states", "hidden_states_before_downsampling", "i", "if", "in", "input_dimensions", "int", "is", "last_hidden_state", "layer_module", "layer_outputs", "layers", "nn", "not", "output_attentions", "output_dimensions", "output_hidden_states", "output_hidden_states_before_downsampling", "permute", "reshaped_hidden_state", "reshaped_hidden_states", "return", "return_dict", "self", "shape", "torch", "tuple", "v", "view"], "swin/modeling_swin.py:SwinPreTrainedModel._init_weights": ["ModelEmbeddings", "ModelPreTrainedModel", "ModelSelfAttention", "None", "PreTrainedModel", "_init_weights", "class", "copy_", "create_relative_position_index", "def", "elif", "if", "init", "is", "isinstance", "mask_token", "module", "not", "position_embeddings", "relative_position_bias_table", "relative_position_index", "self", "super", "zeros_"], "swin/modeling_swin.py:SwinModel.__init__": ["AdaptiveAvgPool1d", "False", "LayerNorm", "ModelEmbeddings", "ModelEncoder", "ModelModel", "ModelPreTrainedModel", "None", "True", "__init__", "add_pooling_layer", "class", "config", "def", "depths", "else", "embed_dim", "embeddings", "encoder", "eps", "if", "int", "layer_norm_eps", "layernorm", "len", "nn", "num_features", "num_layers", "patch_grid", "pooler", "post_init", "r", "self", "super", "use_mask_token"], "swin/modeling_swin.py:SwinModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "patch_embeddings", "return", "self"], "swin/modeling_swin.py:SwinModel.forward": ["BoolTensor", "False", "FloatTensor", "ModelModel", "ModelModelOutput", "ModelPreTrainedModel", "None", "ValueError", "You", "attentions", "bool", "bool_masked_pos", "class", "config", "def", "else", "embedding_output", "embeddings", "encoder", "encoder_outputs", "flatten", "forward", "have", "hidden_states", "if", "input_dimensions", "interpolate_pos_encoding", "is", "kwargs", "last_hidden_state", "layernorm", "not", "output", "output_attentions", "output_hidden_states", "pixel_values", "pooled_output", "pooler", "pooler_output", "r", "raise", "reshaped_hidden_states", "return", "return_dict", "self", "sequence_output", "specify", "to", "torch", "transpose", "tuple", "use_return_dict"], "swin/modeling_swin.py:SwinForMaskedImageModeling.__init__": ["Conv2d", "False", "Model", "ModelForMaskedImageModeling", "ModelModel", "ModelPreTrainedModel", "PixelShuffle", "Sequential", "True", "__init__", "add_pooling_layer", "class", "config", "decoder", "def", "embed_dim", "encoder_stride", "in_channels", "int", "kernel_size", "nn", "num_channels", "num_features", "num_layers", "out_channels", "post_init", "self", "super", "use_mask_token"], "swin/modeling_swin.py:SwinForMaskedImageModeling.forward": ["BoolTensor", "False", "FloatTensor", "Model", "ModelForMaskedImageModeling", "ModelMaskedImageModelingOutput", "ModelPreTrainedModel", "None", "attentions", "batch_size", "bool", "bool_masked_pos", "class", "config", "contiguous", "decoder", "def", "else", "floor", "forward", "functional", "height", "hidden_states", "if", "image_size", "interpolate_pos_encoding", "is", "kwargs", "l1_loss", "loss", "mask", "masked_im_loss", "math", "nn", "none", "not", "num_channels", "output", "output_attentions", "output_hidden_states", "outputs", "patch_size", "pixel_values", "r", "reconstructed_pixel_values", "reconstruction", "reconstruction_loss", "reduction", "repeat_interleave", "reshape", "reshaped_hidden_states", "return", "return_dict", "self", "sequence_length", "sequence_output", "shape", "size", "sum", "torch", "transpose", "tuple", "unsqueeze", "use_return_dict", "width"], "swin/modeling_swin.py:SwinForImageClassification.__init__": ["Identity", "Linear", "Model", "ModelForImageClassification", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "classifier", "config", "def", "else", "if", "nn", "num_features", "num_labels", "post_init", "self", "super"], "swin/modeling_swin.py:SwinForImageClassification.forward": ["False", "FloatTensor", "LongTensor", "Model", "ModelForImageClassification", "ModelImageClassifierOutput", "ModelPreTrainedModel", "None", "attentions", "bool", "class", "classifier", "config", "def", "else", "forward", "hidden_states", "if", "interpolate_pos_encoding", "is", "kwargs", "labels", "logits", "loss", "loss_function", "not", "output", "output_attentions", "output_hidden_states", "outputs", "pixel_values", "pooled_output", "r", "reshaped_hidden_states", "return", "return_dict", "self", "torch", "tuple", "use_return_dict"], "swin/modeling_swin.py:SwinBackbone.__init__": ["BackboneMixin", "LayerNorm", "ModelBackbone", "ModelConfig", "ModelEmbeddings", "ModelEncoder", "ModelPreTrainedModel", "ModuleDict", "__init__", "_init_backbone", "_out_features", "channels", "class", "config", "def", "depths", "embed_dim", "embeddings", "encoder", "for", "hidden_states_norms", "i", "in", "int", "len", "nn", "num_channels", "num_features", "patch_grid", "post_init", "range", "self", "stage", "super", "zip"], "swin/modeling_swin.py:SwinBackbone.get_input_embeddings": ["BackboneMixin", "ModelBackbone", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "patch_embeddings", "return", "self"], "swin/modeling_swin.py:SwinBackbone.forward": ["BackboneMixin", "BackboneOutput", "ModelBackbone", "ModelPreTrainedModel", "None", "Tensor", "True", "always_partition", "attentions", "batch_size", "bool", "class", "config", "contiguous", "def", "else", "embedding_output", "embeddings", "encoder", "feature_maps", "for", "forward", "height", "hidden_state", "hidden_states", "hidden_states_norms", "if", "in", "input_dimensions", "is", "kwargs", "not", "num_channels", "out_features", "output", "output_attentions", "output_hidden_states", "output_hidden_states_before_downsampling", "outputs", "permute", "pixel_values", "reshaped_hidden_states", "return", "return_dict", "self", "shape", "stage", "stage_names", "torch", "use_return_dict", "view", "width", "zip"], "swin2sr/modeling_swin2sr.py:window_partition": ["Model_partition", "Model_size", "Models", "batch_size", "contiguous", "def", "height", "input_feature", "num_channels", "permute", "return", "shape", "view", "width"], "swin2sr/modeling_swin2sr.py:window_reverse": ["Model_reverse", "Model_size", "Models", "contiguous", "def", "height", "num_channels", "permute", "return", "shape", "view", "width"], "swin2sr/modeling_swin2sr.py:drop_path": ["False", "Model_path", "Model_prob", "Tensor", "bool", "def", "device", "div", "dtype", "float", "floor_", "if", "input", "keep_prob", "ndim", "not", "or", "output", "rand", "random_tensor", "return", "shape", "torch", "training"], "swin2sr/modeling_swin2sr.py:Swin2SRDropPath.__init__": ["ModelDropPath", "Module", "None", "__init__", "class", "def", "drop_prob", "float", "nn", "self", "super"], "swin2sr/modeling_swin2sr.py:Swin2SRDropPath.forward": ["ModelDropPath", "Module", "Tensor", "class", "def", "drop_path", "drop_prob", "forward", "hidden_states", "nn", "return", "self", "torch", "training"], "swin2sr/modeling_swin2sr.py:Swin2SRDropPath.extra_repr": ["ModelDropPath", "Module", "class", "def", "drop_prob", "extra_repr", "f", "nn", "p", "return", "self", "str"], "swin2sr/modeling_swin2sr.py:Swin2SREmbeddings.__init__": ["Dropout", "ModelEmbeddings", "ModelPatchEmbeddings", "Module", "None", "Parameter", "__init__", "class", "config", "def", "dropout", "else", "embed_dim", "hidden_dropout_prob", "if", "nn", "num_patches", "patch_embeddings", "position_embeddings", "self", "super", "torch", "use_absolute_embeddings", "window_size", "zeros"], "swin2sr/modeling_swin2sr.py:Swin2SREmbeddings.forward": ["FloatTensor", "ModelEmbeddings", "Module", "None", "Tensor", "class", "def", "dropout", "embeddings", "forward", "if", "is", "nn", "not", "output_dimensions", "patch_embeddings", "pixel_values", "position_embeddings", "return", "self", "torch", "tuple"], "swin2sr/modeling_swin2sr.py:Swin2SRPatchEmbeddings.__init__": ["Conv2d", "Iterable", "LayerNorm", "ModelPatchEmbeddings", "Module", "None", "True", "__init__", "abc", "class", "collections", "config", "def", "else", "embed_dim", "if", "image_size", "isinstance", "kernel_size", "layernorm", "nn", "normalize_patches", "num_channels", "num_patches", "patch_size", "patches_resolution", "projection", "self", "stride", "super"], "swin2sr/modeling_swin2sr.py:Swin2SRPatchEmbeddings.forward": ["FloatTensor", "ModelPatchEmbeddings", "Module", "None", "Tensor", "_", "class", "def", "embeddings", "flatten", "forward", "height", "if", "int", "is", "layernorm", "nn", "not", "output_dimensions", "projection", "return", "self", "shape", "torch", "transpose", "tuple", "width"], "swin2sr/modeling_swin2sr.py:Swin2SRPatchUnEmbeddings.__init__": ["ModelPatchUnEmbeddings", "Module", "__init__", "class", "config", "def", "embed_dim", "nn", "self", "super"], "swin2sr/modeling_swin2sr.py:Swin2SRPatchUnEmbeddings.forward": ["ModelPatchUnEmbeddings", "Module", "batch_size", "class", "def", "embed_dim", "embeddings", "forward", "height_width", "nn", "num_channels", "return", "self", "shape", "transpose", "view", "x_size"], "swin2sr/modeling_swin2sr.py:Swin2SRPatchMerging.__init__": ["False", "LayerNorm", "Linear", "ModelPatchMerging", "Module", "None", "__init__", "bias", "class", "def", "dim", "input_resolution", "int", "nn", "norm", "norm_layer", "reduction", "self", "super", "tuple"], "swin2sr/modeling_swin2sr.py:Swin2SRPatchMerging.maybe_pad": ["ModelPatchMerging", "Module", "class", "def", "functional", "height", "if", "input_feature", "maybe_pad", "nn", "or", "pad", "pad_values", "return", "self", "should_pad", "width"], "swin2sr/modeling_swin2sr.py:Swin2SRPatchMerging.forward": ["ModelPatchMerging", "Module", "Tensor", "batch_size", "cat", "class", "def", "dim", "forward", "height", "input_dimensions", "input_feature", "input_feature_0", "input_feature_1", "input_feature_2", "input_feature_3", "int", "maybe_pad", "nn", "norm", "num_channels", "reduction", "return", "self", "shape", "torch", "tuple", "view", "width"], "swin2sr/modeling_swin2sr.py:Swin2SRSelfAttention.__init__": ["Dropout", "False", "Iterable", "Linear", "ModelSelfAttention", "Module", "Parameter", "ReLU", "Sequential", "The", "True", "ValueError", "__init__", "a", "abc", "all_head_size", "attention", "attention_head_size", "attention_probs_dropout_prob", "bias", "class", "collections", "config", "continuous_position_bias_mlp", "create_coords_table_and_index", "def", "dim", "dropout", "else", "f", "heads", "hidden", "if", "inplace", "int", "is", "isinstance", "key", "log", "logit_scale", "multiple", "nn", "not", "num_attention_heads", "num_heads", "number", "of", "ones", "persistent", "pretrained_window_size", "qkv_bias", "query", "raise", "register_buffer", "relative_coords_table", "relative_position_index", "self", "size", "super", "the", "torch", "value", "window_size"], "swin2sr/modeling_swin2sr.py:Swin2SRSelfAttention.forward": ["False", "FloatTensor", "ModelSelfAttention", "Module", "None", "Tensor", "all_head_size", "attention_head_size", "attention_mask", "attention_probs", "attention_scores", "batch_size", "bool", "clamp", "class", "context_layer", "contiguous", "continuous_position_bias_mlp", "def", "dim", "dropout", "else", "exp", "forward", "functional", "hidden_states", "if", "is", "key", "key_layer", "log", "logit_scale", "mask_shape", "math", "matmul", "max", "new_context_layer_shape", "nn", "normalize", "not", "num_attention_heads", "num_channels", "output_attentions", "outputs", "permute", "query", "query_layer", "relative_coords_table", "relative_position_bias", "relative_position_bias_table", "relative_position_index", "return", "self", "shape", "sigmoid", "size", "softmax", "torch", "transpose", "tuple", "unsqueeze", "value", "value_layer", "view", "window_size"], "swin2sr/modeling_swin2sr.py:Swin2SRSelfAttention.create_coords_table_and_index": ["ModelSelfAttention", "Module", "None", "abs", "arange", "class", "contiguous", "continuous_position_bias_mlp", "coords", "coords_flatten", "coords_h", "coords_w", "create_coords_table_and_index", "def", "dtype", "elif", "flatten", "float", "if", "ij", "indexing", "int64", "log2", "math", "meshgrid", "next", "nn", "parameters", "permute", "pretrained_window_size", "relative_coords", "relative_coords_h", "relative_coords_table", "relative_coords_w", "relative_position_index", "return", "self", "sign", "stack", "sum", "to", "torch", "unsqueeze", "window_size"], "swin2sr/modeling_swin2sr.py:Swin2SRSelfOutput.__init__": ["Dropout", "Linear", "ModelSelfOutput", "Module", "__init__", "attention_probs_dropout_prob", "class", "config", "def", "dense", "dim", "dropout", "nn", "self", "super"], "swin2sr/modeling_swin2sr.py:Swin2SRSelfOutput.forward": ["ModelSelfOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "swin2sr/modeling_swin2sr.py:Swin2SRAttention.__init__": ["Iterable", "ModelAttention", "ModelSelfAttention", "ModelSelfOutput", "Module", "__init__", "abc", "class", "collections", "config", "def", "dim", "else", "if", "isinstance", "nn", "num_heads", "output", "pretrained_window_size", "self", "super", "window_size"], "swin2sr/modeling_swin2sr.py:Swin2SRAttention.forward": ["False", "FloatTensor", "ModelAttention", "Module", "None", "Tensor", "attention_mask", "attention_output", "bool", "class", "def", "forward", "hidden_states", "nn", "output", "output_attentions", "outputs", "return", "self", "self_outputs", "torch", "tuple"], "swin2sr/modeling_swin2sr.py:Swin2SRIntermediate.__init__": ["ACT2FN", "Linear", "ModelIntermediate", "Module", "__init__", "class", "config", "def", "dense", "dim", "else", "hidden_act", "if", "int", "intermediate_act_fn", "isinstance", "mlp_ratio", "nn", "self", "str", "super"], "swin2sr/modeling_swin2sr.py:Swin2SRIntermediate.forward": ["ModelIntermediate", "Module", "Tensor", "class", "def", "dense", "forward", "hidden_states", "intermediate_act_fn", "nn", "return", "self", "torch"], "swin2sr/modeling_swin2sr.py:Swin2SROutput.__init__": ["Dropout", "Linear", "ModelOutput", "Module", "__init__", "class", "config", "def", "dense", "dim", "dropout", "hidden_dropout_prob", "int", "mlp_ratio", "nn", "self", "super"], "swin2sr/modeling_swin2sr.py:Swin2SROutput.forward": ["ModelOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "nn", "return", "self", "torch"], "swin2sr/modeling_swin2sr.py:Swin2SRLayer.__init__": ["Identity", "Iterable", "LayerNorm", "ModelAttention", "ModelDropPath", "ModelIntermediate", "ModelLayer", "ModelOutput", "Module", "__init__", "_compute_window_shift", "abc", "attention", "class", "collections", "config", "def", "dim", "drop_path", "drop_path_rate", "else", "eps", "if", "input_resolution", "intermediate", "isinstance", "layer_norm_eps", "layernorm_after", "layernorm_before", "nn", "num_heads", "output", "pretrained_window_size", "self", "shift_size", "super", "window_size"], "swin2sr/modeling_swin2sr.py:Swin2SRLayer._compute_window_shift": ["ModelLayer", "Module", "_compute_window_shift", "class", "def", "else", "for", "if", "in", "input_resolution", "int", "min", "nn", "r", "return", "s", "self", "shift_size", "target_shift_size", "target_window_size", "tuple", "w", "window_size", "zip"], "swin2sr/modeling_swin2sr.py:Swin2SRLayer.get_attn_mask": ["ModelLayer", "Module", "None", "attn_mask", "class", "count", "def", "dtype", "else", "for", "get_attn_mask", "height", "height_slice", "height_slices", "if", "img_mask", "in", "mask_windows", "masked_fill", "nn", "return", "self", "shift_size", "slice", "torch", "unsqueeze", "view", "width", "width_slice", "width_slices", "window_partition", "window_size", "zeros"], "swin2sr/modeling_swin2sr.py:Swin2SRLayer.maybe_pad": ["ModelLayer", "Module", "class", "def", "functional", "height", "hidden_states", "maybe_pad", "nn", "pad", "pad_bottom", "pad_right", "pad_values", "return", "self", "width", "window_size"], "swin2sr/modeling_swin2sr.py:Swin2SRLayer.forward": ["False", "ModelLayer", "Module", "None", "Tensor", "_", "attention", "attention_output", "attention_outputs", "attention_windows", "attn_mask", "batch_size", "bool", "channels", "class", "contiguous", "def", "device", "dims", "drop_path", "dtype", "else", "forward", "get_attn_mask", "height", "height_pad", "hidden_states", "hidden_states_windows", "if", "input_dimensions", "int", "intermediate", "is", "layer_output", "layer_outputs", "layernorm_after", "layernorm_before", "maybe_pad", "nn", "not", "or", "output", "output_attentions", "pad_values", "return", "roll", "self", "shape", "shift_size", "shifted_hidden_states", "shifted_windows", "shifts", "shortcut", "size", "to", "torch", "tuple", "view", "was_padded", "width", "width_pad", "window_partition", "window_reverse", "window_size"], "swin2sr/modeling_swin2sr.py:Swin2SRStage.__init__": ["Conv2d", "False", "GradientCheckpointingLayer", "LeakyReLU", "ModelLayer", "ModelPatchEmbeddings", "ModelPatchUnEmbeddings", "ModelStage", "ModuleList", "Sequential", "True", "__init__", "class", "config", "conv", "def", "depth", "dim", "drop_path", "elif", "else", "for", "i", "if", "in", "inplace", "input_resolution", "layers", "negative_slope", "nn", "normalize_patches", "num_heads", "patch_embed", "patch_unembed", "pretrained_window_size", "range", "resi_connection", "self", "shift_size", "super", "window_size"], "swin2sr/modeling_swin2sr.py:Swin2SRStage.forward": ["False", "GradientCheckpointingLayer", "ModelStage", "None", "Tensor", "_", "bool", "class", "conv", "def", "enumerate", "for", "forward", "height", "hidden_states", "i", "if", "in", "input_dimensions", "int", "layer_module", "layer_outputs", "layers", "output_attentions", "output_dimensions", "patch_embed", "patch_unembed", "residual", "return", "self", "stage_outputs", "torch", "tuple", "width"], "swin2sr/modeling_swin2sr.py:Swin2SREncoder.__init__": ["False", "ModelEncoder", "ModelStage", "Module", "ModuleList", "__init__", "class", "config", "cpu", "def", "depth", "depths", "device", "dim", "dpr", "drop_path", "drop_path_rate", "embed_dim", "for", "gradient_checkpointing", "grid_size", "in", "input_resolution", "item", "len", "linspace", "nn", "num_heads", "num_stages", "pretrained_window_size", "range", "self", "stage_idx", "stages", "sum", "super", "torch", "x"], "swin2sr/modeling_swin2sr.py:Swin2SREncoder.forward": ["False", "ModelEncoder", "ModelEncoderOutput", "Module", "None", "Tensor", "True", "all_hidden_states", "all_input_dimensions", "all_self_attentions", "attentions", "bool", "class", "def", "else", "enumerate", "for", "forward", "hidden_states", "i", "if", "in", "input_dimensions", "int", "is", "last_hidden_state", "layer_outputs", "nn", "not", "output_attentions", "output_dimensions", "output_hidden_states", "return", "return_dict", "self", "stage_module", "stages", "torch", "tuple", "v"], "swin2sr/modeling_swin2sr.py:Swin2SRPreTrainedModel._init_weights": ["Conv2d", "LayerNorm", "Linear", "ModelModel", "ModelPreTrainedModel", "ModelSelfAttention", "None", "PreTrainedModel", "_init_weights", "and", "bias", "class", "config", "constant_", "copy_", "create_coords_table_and_index", "def", "elif", "else", "if", "init", "initializer_range", "is", "isinstance", "log", "logit_scale", "math", "mean", "module", "nn", "not", "num_channels", "num_channels_out", "ones_", "relative_coords_table", "relative_position_index", "self", "std", "tensor", "torch", "trunc_normal_", "view", "weight", "zeros", "zeros_"], "swin2sr/modeling_swin2sr.py:Swin2SRModel.__init__": ["Conv2d", "False", "LayerNorm", "ModelEmbeddings", "ModelEncoder", "ModelModel", "ModelPatchUnEmbeddings", "ModelPreTrainedModel", "__init__", "and", "class", "config", "conv_after_body", "def", "else", "embed_dim", "embeddings", "encoder", "eps", "first_convolution", "grid_size", "if", "img_range", "layer_norm_eps", "layernorm", "mean", "nn", "num_channels", "num_channels_out", "patch_embeddings", "patch_unembed", "patches_resolution", "persistent", "post_init", "register_buffer", "self", "super", "tensor", "torch", "view", "zeros"], "swin2sr/modeling_swin2sr.py:Swin2SRModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "patch_embeddings", "return", "self"], "swin2sr/modeling_swin2sr.py:Swin2SRModel.pad_and_normalize": ["ModelModel", "ModelPreTrainedModel", "_", "class", "config", "def", "functional", "height", "img_range", "mean", "modulo_pad_height", "modulo_pad_width", "nn", "pad", "pad_and_normalize", "pixel_values", "reflect", "return", "self", "size", "type_as", "width", "window_size"], "swin2sr/modeling_swin2sr.py:Swin2SRModel.forward": ["BaseModelOutput", "FloatTensor", "ModelModel", "ModelPreTrainedModel", "None", "_", "attentions", "bool", "class", "config", "conv_after_body", "def", "else", "embedding_output", "embeddings", "encoder", "encoder_outputs", "first_convolution", "forward", "height", "hidden_states", "if", "input_dimensions", "is", "kwargs", "last_hidden_state", "layernorm", "not", "output", "output_attentions", "output_hidden_states", "pad_and_normalize", "patch_unembed", "pixel_values", "return", "return_dict", "self", "sequence_output", "shape", "torch", "tuple", "use_return_dict", "width"], "swin2sr/modeling_swin2sr.py:Upsample.__init__": ["Conv2d", "Model", "Module", "PixelShuffle", "Scale", "Supported", "ValueError", "__init__", "add_module", "and", "class", "convolution", "convolution_", "def", "elif", "else", "f", "for", "i", "if", "in", "int", "is", "log2", "math", "n", "nn", "not", "num_features", "pixelshuffle", "pixelshuffle_", "raise", "range", "scale", "scales", "self", "super", "supported"], "swin2sr/modeling_swin2sr.py:Upsample.forward": ["Model", "Module", "__getattr__", "class", "convolution", "convolution_", "def", "elif", "f", "for", "forward", "hidden_state", "i", "if", "in", "int", "log2", "math", "nn", "pixelshuffle", "pixelshuffle_", "range", "return", "scale", "self"], "swin2sr/modeling_swin2sr.py:UpsampleOneStep.__init__": ["Conv2d", "ModelOneStep", "Module", "PixelShuffle", "__init__", "class", "conv", "def", "in_channels", "nn", "out_channels", "pixel_shuffle", "scale", "self", "super"], "swin2sr/modeling_swin2sr.py:UpsampleOneStep.forward": ["ModelOneStep", "Module", "class", "conv", "def", "forward", "nn", "pixel_shuffle", "return", "self", "x"], "swin2sr/modeling_swin2sr.py:PixelShuffleUpsampler.__init__": ["Conv2d", "LeakyReLU", "ModelShuffleUpsampler", "Module", "True", "Upsample", "__init__", "activation", "class", "config", "conv_before_upsample", "def", "embed_dim", "final_convolution", "inplace", "nn", "num_channels_out", "num_features", "self", "super", "upsample", "upscale"], "swin2sr/modeling_swin2sr.py:PixelShuffleUpsampler.forward": ["ModelShuffleUpsampler", "Module", "activation", "class", "conv_before_upsample", "def", "final_convolution", "forward", "nn", "return", "self", "sequence_output", "upsample", "x"], "swin2sr/modeling_swin2sr.py:NearestConvUpsampler.__init__": ["Conv2d", "LeakyReLU", "Model", "ModelConvUpsampler", "Module", "The", "True", "ValueError", "__init__", "activation", "an", "at", "class", "config", "conv", "conv_before_upsample", "conv_hr", "conv_up1", "conv_up2", "def", "embed_dim", "factor", "final_convolution", "if", "inplace", "lrelu", "moment", "negative_slope", "nn", "num_channels_out", "num_features", "of", "only", "raise", "self", "super", "supports", "the", "upsampler", "upscale"], "swin2sr/modeling_swin2sr.py:NearestConvUpsampler.forward": ["Model", "ModelConvUpsampler", "Module", "activation", "class", "conv_before_upsample", "conv_hr", "conv_up1", "conv_up2", "def", "final_convolution", "forward", "functional", "interpolate", "lrelu", "mode", "nn", "reconstruction", "return", "scale_factor", "self", "sequence_output", "torch"], "swin2sr/modeling_swin2sr.py:PixelShuffleAuxUpsampler.__init__": ["Conv2d", "LeakyReLU", "ModelShuffleAuxUpsampler", "Module", "Sequential", "True", "Upsample", "__init__", "activation", "class", "config", "conv_after_aux", "conv_aux", "conv_before_upsample", "conv_bicubic", "def", "embed_dim", "final_convolution", "inplace", "nn", "num_channels", "num_channels_out", "num_features", "self", "super", "upsample", "upscale"], "swin2sr/modeling_swin2sr.py:PixelShuffleAuxUpsampler.forward": ["ModelShuffleAuxUpsampler", "Module", "activation", "aux", "bicubic", "class", "conv_after_aux", "conv_aux", "conv_before_upsample", "conv_bicubic", "def", "final_convolution", "forward", "height", "nn", "reconstruction", "return", "self", "sequence_output", "upsample", "upscale", "width"], "swin2sr/modeling_swin2sr.py:Swin2SRForImageSuperResolution.__init__": ["Conv2d", "Model", "ModelForImageSuperResolution", "ModelModel", "ModelPreTrainedModel", "NearestConvUpsampler", "PixelShuffleAuxUpsampler", "PixelShuffleUpsampler", "UpsampleOneStep", "__init__", "class", "config", "conv", "def", "elif", "else", "embed_dim", "final_convolution", "if", "nearest", "nn", "num_channels_out", "num_features", "pixelshuffle", "pixelshuffle_aux", "pixelshuffledirect", "post_init", "self", "super", "upsample", "upsampler", "upscale"], "swin2sr/modeling_swin2sr.py:Swin2SRForImageSuperResolution.forward": ["False", "FloatTensor", "ImageSuperResolutionOutput", "LongTensor", "Model", "ModelForImageSuperResolution", "ModelPreTrainedModel", "None", "NotImplementedError", "Training", "align_corners", "at", "attentions", "aux", "bicubic", "bool", "class", "config", "conv", "def", "elif", "else", "final_convolution", "forward", "functional", "height", "hidden_states", "if", "img_range", "in", "interpolate", "is", "kwargs", "labels", "loss", "mean", "mode", "moment", "nearest", "nn", "not", "output", "output_attentions", "output_hidden_states", "outputs", "pixel_values", "pixelshuffle", "pixelshuffle_aux", "pixelshuffledirect", "r", "raise", "reconstruction", "return", "return_dict", "self", "sequence_output", "shape", "size", "supported", "the", "torch", "tuple", "upsample", "upsampler", "upscale", "use_return_dict", "width"], "swinv2/modeling_swinv2.py:window_partition": ["Model_partition", "Model_size", "Models", "batch_size", "contiguous", "def", "height", "input_feature", "num_channels", "permute", "return", "shape", "view", "width"], "swinv2/modeling_swinv2.py:window_reverse": ["Model_reverse", "Model_size", "Models", "contiguous", "def", "height", "num_channels", "permute", "return", "shape", "view", "width"], "swinv2/modeling_swinv2.py:drop_path": ["False", "Model_path", "Model_prob", "Tensor", "bool", "def", "device", "div", "dtype", "float", "floor_", "if", "input", "keep_prob", "ndim", "not", "or", "output", "rand", "random_tensor", "return", "shape", "torch", "training"], "swinv2/modeling_swinv2.py:Swinv2DropPath.__init__": ["ModelDropPath", "Module", "None", "__init__", "class", "def", "drop_prob", "float", "nn", "self", "super"], "swinv2/modeling_swinv2.py:Swinv2DropPath.forward": ["ModelDropPath", "Module", "Tensor", "class", "def", "drop_path", "drop_prob", "forward", "hidden_states", "nn", "return", "self", "torch", "training"], "swinv2/modeling_swinv2.py:Swinv2DropPath.extra_repr": ["ModelDropPath", "Module", "class", "def", "drop_prob", "extra_repr", "f", "nn", "p", "return", "self", "str"], "swinv2/modeling_swinv2.py:Swinv2Embeddings.__init__": ["Dropout", "False", "LayerNorm", "ModelEmbeddings", "ModelPatchEmbeddings", "Module", "None", "Parameter", "__init__", "class", "config", "def", "dropout", "else", "embed_dim", "grid_size", "hidden_dropout_prob", "if", "mask_token", "nn", "norm", "num_patches", "patch_embeddings", "patch_grid", "patch_size", "position_embeddings", "self", "super", "torch", "use_absolute_embeddings", "use_mask_token", "zeros"], "swinv2/modeling_swinv2.py:Swinv2Embeddings.interpolate_pos_encoding": ["False", "ModelEmbeddings", "Module", "Tensor", "align_corners", "and", "bicubic", "cat", "class", "class_pos_embed", "def", "dim", "embeddings", "functional", "height", "if", "int", "interpolate", "interpolate_pos_encoding", "is_tracing", "jit", "mode", "new_height", "new_width", "nn", "not", "num_patches", "num_positions", "patch_pos_embed", "patch_size", "permute", "position_embeddings", "reshape", "return", "self", "shape", "size", "sqrt_num_positions", "torch", "torch_int", "view", "width"], "swinv2/modeling_swinv2.py:Swinv2Embeddings.forward": ["BoolTensor", "False", "FloatTensor", "ModelEmbeddings", "Module", "None", "Tensor", "_", "batch_size", "bool", "bool_masked_pos", "class", "def", "dropout", "else", "embeddings", "expand", "forward", "height", "if", "interpolate_pos_encoding", "is", "mask", "mask_token", "mask_tokens", "nn", "norm", "not", "num_channels", "output_dimensions", "patch_embeddings", "pixel_values", "position_embeddings", "return", "self", "seq_len", "shape", "size", "torch", "tuple", "type_as", "unsqueeze", "width"], "swinv2/modeling_swinv2.py:Swinv2PatchEmbeddings.__init__": ["Conv2d", "Iterable", "ModelPatchEmbeddings", "Module", "__init__", "abc", "class", "collections", "config", "def", "else", "embed_dim", "grid_size", "hidden_size", "if", "image_size", "isinstance", "kernel_size", "nn", "num_channels", "num_patches", "patch_size", "projection", "self", "stride", "super"], "swinv2/modeling_swinv2.py:Swinv2PatchEmbeddings.maybe_pad": ["ModelPatchEmbeddings", "Module", "class", "def", "functional", "height", "if", "maybe_pad", "nn", "pad", "pad_values", "patch_size", "pixel_values", "return", "self", "width"], "swinv2/modeling_swinv2.py:Swinv2PatchEmbeddings.forward": ["FloatTensor", "ModelPatchEmbeddings", "Module", "None", "Tensor", "_", "class", "def", "embeddings", "flatten", "forward", "height", "int", "maybe_pad", "nn", "num_channels", "output_dimensions", "pixel_values", "projection", "return", "self", "shape", "torch", "transpose", "tuple", "width"], "swinv2/modeling_swinv2.py:Swinv2PatchMerging.__init__": ["False", "LayerNorm", "Linear", "ModelPatchMerging", "Module", "None", "__init__", "bias", "class", "def", "dim", "input_resolution", "int", "nn", "norm", "norm_layer", "reduction", "self", "super", "tuple"], "swinv2/modeling_swinv2.py:Swinv2PatchMerging.maybe_pad": ["ModelPatchMerging", "Module", "class", "def", "functional", "height", "if", "input_feature", "maybe_pad", "nn", "or", "pad", "pad_values", "return", "self", "should_pad", "width"], "swinv2/modeling_swinv2.py:Swinv2PatchMerging.forward": ["ModelPatchMerging", "Module", "Tensor", "batch_size", "cat", "class", "def", "dim", "forward", "height", "input_dimensions", "input_feature", "input_feature_0", "input_feature_1", "input_feature_2", "input_feature_3", "int", "maybe_pad", "nn", "norm", "num_channels", "reduction", "return", "self", "shape", "torch", "tuple", "view", "width"], "swinv2/modeling_swinv2.py:Swinv2SelfAttention.__init__": ["Dropout", "False", "Iterable", "Linear", "ModelSelfAttention", "Module", "Parameter", "ReLU", "Sequential", "The", "True", "ValueError", "__init__", "a", "abc", "all_head_size", "attention", "attention_head_size", "attention_probs_dropout_prob", "bias", "class", "collections", "config", "continuous_position_bias_mlp", "create_coords_table_and_index", "def", "dim", "dropout", "else", "f", "heads", "hidden", "if", "inplace", "int", "is", "isinstance", "key", "log", "logit_scale", "multiple", "nn", "not", "num_attention_heads", "num_heads", "number", "of", "ones", "persistent", "pretrained_window_size", "qkv_bias", "query", "raise", "register_buffer", "relative_coords_table", "relative_position_index", "self", "size", "super", "the", "torch", "value", "window_size"], "swinv2/modeling_swinv2.py:Swinv2SelfAttention.forward": ["False", "FloatTensor", "ModelSelfAttention", "Module", "None", "Tensor", "all_head_size", "attention_head_size", "attention_mask", "attention_probs", "attention_scores", "batch_size", "bool", "clamp", "class", "context_layer", "contiguous", "continuous_position_bias_mlp", "def", "dim", "dropout", "else", "exp", "forward", "functional", "hidden_states", "if", "is", "key", "key_layer", "log", "logit_scale", "mask_shape", "math", "matmul", "max", "new_context_layer_shape", "nn", "normalize", "not", "num_attention_heads", "num_channels", "output_attentions", "outputs", "permute", "query", "query_layer", "relative_coords_table", "relative_position_bias", "relative_position_bias_table", "relative_position_index", "return", "self", "shape", "sigmoid", "size", "softmax", "torch", "transpose", "tuple", "unsqueeze", "value", "value_layer", "view", "window_size"], "swinv2/modeling_swinv2.py:Swinv2SelfAttention.create_coords_table_and_index": ["ModelSelfAttention", "Module", "None", "abs", "arange", "class", "contiguous", "continuous_position_bias_mlp", "coords", "coords_flatten", "coords_h", "coords_w", "create_coords_table_and_index", "def", "dtype", "elif", "flatten", "float", "if", "ij", "indexing", "int64", "log2", "math", "meshgrid", "next", "nn", "parameters", "permute", "pretrained_window_size", "relative_coords", "relative_coords_h", "relative_coords_table", "relative_coords_w", "relative_position_index", "return", "self", "sign", "stack", "sum", "to", "torch", "unsqueeze", "window_size"], "swinv2/modeling_swinv2.py:Swinv2SelfOutput.__init__": ["Dropout", "Linear", "ModelSelfOutput", "Module", "__init__", "attention_probs_dropout_prob", "class", "config", "def", "dense", "dim", "dropout", "nn", "self", "super"], "swinv2/modeling_swinv2.py:Swinv2SelfOutput.forward": ["ModelSelfOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "swinv2/modeling_swinv2.py:Swinv2Attention.__init__": ["Iterable", "ModelAttention", "ModelSelfAttention", "ModelSelfOutput", "Module", "__init__", "abc", "class", "collections", "config", "def", "dim", "else", "if", "isinstance", "nn", "num_heads", "output", "pretrained_window_size", "self", "super", "window_size"], "swinv2/modeling_swinv2.py:Swinv2Attention.forward": ["False", "FloatTensor", "ModelAttention", "Module", "None", "Tensor", "attention_mask", "attention_output", "bool", "class", "def", "forward", "hidden_states", "nn", "output", "output_attentions", "outputs", "return", "self", "self_outputs", "torch", "tuple"], "swinv2/modeling_swinv2.py:Swinv2Intermediate.__init__": ["ACT2FN", "Linear", "ModelIntermediate", "Module", "__init__", "class", "config", "def", "dense", "dim", "else", "hidden_act", "if", "int", "intermediate_act_fn", "isinstance", "mlp_ratio", "nn", "self", "str", "super"], "swinv2/modeling_swinv2.py:Swinv2Intermediate.forward": ["ModelIntermediate", "Module", "Tensor", "class", "def", "dense", "forward", "hidden_states", "intermediate_act_fn", "nn", "return", "self", "torch"], "swinv2/modeling_swinv2.py:Swinv2Output.__init__": ["Dropout", "Linear", "ModelOutput", "Module", "__init__", "class", "config", "def", "dense", "dim", "dropout", "hidden_dropout_prob", "int", "mlp_ratio", "nn", "self", "super"], "swinv2/modeling_swinv2.py:Swinv2Output.forward": ["ModelOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "nn", "return", "self", "torch"], "swinv2/modeling_swinv2.py:Swinv2Layer.__init__": ["Identity", "Iterable", "LayerNorm", "ModelAttention", "ModelDropPath", "ModelIntermediate", "ModelLayer", "ModelOutput", "Module", "__init__", "_compute_window_shift", "abc", "attention", "class", "collections", "config", "def", "dim", "drop_path", "drop_path_rate", "else", "eps", "if", "input_resolution", "intermediate", "isinstance", "layer_norm_eps", "layernorm_after", "layernorm_before", "nn", "num_heads", "output", "pretrained_window_size", "self", "shift_size", "super", "window_size"], "swinv2/modeling_swinv2.py:Swinv2Layer._compute_window_shift": ["ModelLayer", "Module", "_compute_window_shift", "class", "def", "else", "for", "if", "in", "input_resolution", "int", "min", "nn", "r", "return", "s", "self", "shift_size", "target_shift_size", "target_window_size", "tuple", "w", "window_size", "zip"], "swinv2/modeling_swinv2.py:Swinv2Layer.get_attn_mask": ["ModelLayer", "Module", "None", "attn_mask", "class", "count", "def", "dtype", "else", "for", "get_attn_mask", "height", "height_slice", "height_slices", "if", "img_mask", "in", "mask_windows", "masked_fill", "nn", "return", "self", "shift_size", "slice", "torch", "unsqueeze", "view", "width", "width_slice", "width_slices", "window_partition", "window_size", "zeros"], "swinv2/modeling_swinv2.py:Swinv2Layer.maybe_pad": ["ModelLayer", "Module", "class", "def", "functional", "height", "hidden_states", "maybe_pad", "nn", "pad", "pad_bottom", "pad_right", "pad_values", "return", "self", "width", "window_size"], "swinv2/modeling_swinv2.py:Swinv2Layer.forward": ["False", "ModelLayer", "Module", "None", "Tensor", "_", "attention", "attention_output", "attention_outputs", "attention_windows", "attn_mask", "batch_size", "bool", "channels", "class", "contiguous", "def", "device", "dims", "drop_path", "dtype", "else", "forward", "get_attn_mask", "height", "height_pad", "hidden_states", "hidden_states_windows", "if", "input_dimensions", "int", "intermediate", "is", "layer_output", "layer_outputs", "layernorm_after", "layernorm_before", "maybe_pad", "nn", "not", "or", "output", "output_attentions", "pad_values", "return", "roll", "self", "shape", "shift_size", "shifted_hidden_states", "shifted_windows", "shifts", "shortcut", "size", "to", "torch", "tuple", "view", "was_padded", "width", "width_pad", "window_partition", "window_reverse", "window_size"], "swinv2/modeling_swinv2.py:Swinv2Stage.__init__": ["False", "GradientCheckpointingLayer", "LayerNorm", "ModelLayer", "ModelStage", "ModuleList", "None", "__init__", "append", "block", "blocks", "class", "config", "def", "depth", "dim", "downsample", "drop_path", "drop_path_rate", "else", "for", "i", "if", "in", "input_resolution", "is", "nn", "norm_layer", "not", "num_heads", "pointing", "pretrained_window_size", "range", "self", "shift_size", "super", "window_size"], "swinv2/modeling_swinv2.py:Swinv2Stage.forward": ["False", "GradientCheckpointingLayer", "ModelStage", "None", "Tensor", "blocks", "bool", "class", "def", "downsample", "else", "enumerate", "for", "forward", "height", "height_downsampled", "hidden_states", "hidden_states_before_downsampling", "i", "if", "in", "input_dimensions", "int", "is", "layer_module", "layer_outputs", "not", "output_attentions", "output_dimensions", "return", "self", "stage_outputs", "torch", "tuple", "width", "width_downsampled"], "swinv2/modeling_swinv2.py:Swinv2Encoder.__init__": ["False", "ModelEncoder", "ModelPatchMerging", "ModelStage", "Module", "ModuleList", "None", "__init__", "append", "class", "config", "cpu", "def", "depth", "depths", "device", "dim", "downsample", "dpr", "drop_path", "drop_path_rate", "else", "embed_dim", "for", "gradient_checkpointing", "grid_size", "i_layer", "if", "in", "input_resolution", "int", "is", "item", "layers", "len", "linspace", "nn", "not", "num_heads", "num_layers", "pretrained_window_size", "pretrained_window_sizes", "range", "self", "stage", "sum", "super", "torch", "x"], "swinv2/modeling_swinv2.py:Swinv2Encoder.forward": ["False", "ModelEncoder", "ModelEncoderOutput", "Module", "None", "Tensor", "True", "_", "all_hidden_states", "all_reshaped_hidden_states", "all_self_attentions", "and", "attentions", "batch_size", "bool", "class", "def", "elif", "else", "enumerate", "for", "forward", "hidden_size", "hidden_states", "hidden_states_before_downsampling", "i", "if", "in", "input_dimensions", "int", "is", "last_hidden_state", "layer_module", "layer_outputs", "layers", "nn", "not", "output_attentions", "output_dimensions", "output_hidden_states", "output_hidden_states_before_downsampling", "permute", "reshaped_hidden_state", "reshaped_hidden_states", "return", "return_dict", "self", "shape", "torch", "tuple", "v", "view"], "swinv2/modeling_swinv2.py:Swinv2PreTrainedModel._init_weights": ["Conv2d", "LayerNorm", "Linear", "ModelEmbeddings", "ModelPreTrainedModel", "ModelSelfAttention", "None", "PreTrainedModel", "_init_weights", "bias", "class", "config", "constant_", "copy_", "create_coords_table_and_index", "def", "elif", "if", "init", "initializer_range", "is", "isinstance", "log", "logit_scale", "mask_token", "math", "mean", "module", "nn", "normal_", "not", "ones_", "position_embeddings", "relative_coords_table", "relative_position_index", "self", "std", "weight", "zeros_"], "swinv2/modeling_swinv2.py:Swinv2Model.__init__": ["AdaptiveAvgPool1d", "False", "LayerNorm", "ModelEmbeddings", "ModelEncoder", "ModelModel", "ModelPreTrainedModel", "None", "True", "__init__", "add_pooling_layer", "class", "config", "def", "depths", "else", "embed_dim", "embeddings", "encoder", "eps", "if", "int", "layer_norm_eps", "layernorm", "len", "nn", "num_features", "num_layers", "patch_grid", "pooler", "post_init", "r", "self", "super", "use_mask_token"], "swinv2/modeling_swinv2.py:Swinv2Model.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "patch_embeddings", "return", "self"], "swinv2/modeling_swinv2.py:Swinv2Model.forward": ["BoolTensor", "False", "FloatTensor", "ModelModel", "ModelModelOutput", "ModelPreTrainedModel", "None", "ValueError", "You", "attentions", "bool", "bool_masked_pos", "class", "config", "def", "else", "embedding_output", "embeddings", "encoder", "encoder_outputs", "flatten", "forward", "have", "hidden_states", "if", "input_dimensions", "interpolate_pos_encoding", "is", "kwargs", "last_hidden_state", "layernorm", "not", "output", "output_attentions", "output_hidden_states", "pixel_values", "pooled_output", "pooler", "pooler_output", "r", "raise", "reshaped_hidden_states", "return", "return_dict", "self", "sequence_output", "specify", "to", "torch", "transpose", "tuple", "use_return_dict"], "swinv2/modeling_swinv2.py:Swinv2ForMaskedImageModeling.__init__": ["Conv2d", "False", "Model", "ModelForMaskedImageModeling", "ModelModel", "ModelPreTrainedModel", "PixelShuffle", "Sequential", "True", "__init__", "add_pooling_layer", "class", "config", "decoder", "def", "embed_dim", "encoder_stride", "in_channels", "int", "kernel_size", "nn", "num_channels", "num_features", "num_layers", "out_channels", "post_init", "self", "super", "use_mask_token"], "swinv2/modeling_swinv2.py:Swinv2ForMaskedImageModeling.forward": ["BoolTensor", "False", "FloatTensor", "Model", "ModelForMaskedImageModeling", "ModelMaskedImageModelingOutput", "ModelPreTrainedModel", "None", "attentions", "batch_size", "bool", "bool_masked_pos", "class", "config", "contiguous", "decoder", "def", "else", "floor", "forward", "functional", "height", "hidden_states", "if", "image_size", "interpolate_pos_encoding", "is", "kwargs", "l1_loss", "loss", "mask", "masked_im_loss", "math", "nn", "none", "not", "num_channels", "output", "output_attentions", "output_hidden_states", "outputs", "patch_size", "pixel_values", "r", "reconstructed_pixel_values", "reconstruction", "reconstruction_loss", "reduction", "repeat_interleave", "reshape", "reshaped_hidden_states", "return", "return_dict", "self", "sequence_length", "sequence_output", "shape", "size", "sum", "torch", "transpose", "tuple", "unsqueeze", "use_return_dict", "width"], "swinv2/modeling_swinv2.py:Swinv2ForImageClassification.__init__": ["Identity", "Linear", "Model", "ModelForImageClassification", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "classifier", "config", "def", "else", "if", "nn", "num_features", "num_labels", "post_init", "self", "super"], "swinv2/modeling_swinv2.py:Swinv2ForImageClassification.forward": ["False", "FloatTensor", "LongTensor", "Model", "ModelForImageClassification", "ModelImageClassifierOutput", "ModelPreTrainedModel", "None", "attentions", "bool", "class", "classifier", "config", "def", "else", "forward", "hidden_states", "if", "interpolate_pos_encoding", "is", "kwargs", "labels", "logits", "loss", "loss_function", "not", "output", "output_attentions", "output_hidden_states", "outputs", "pixel_values", "pooled_output", "r", "reshaped_hidden_states", "return", "return_dict", "self", "torch", "tuple", "use_return_dict"], "swinv2/modeling_swinv2.py:Swinv2Backbone.__init__": ["BackboneMixin", "ModelBackbone", "ModelEmbeddings", "ModelEncoder", "ModelPreTrainedModel", "__init__", "_init_backbone", "class", "config", "def", "depths", "embed_dim", "embeddings", "encoder", "for", "i", "in", "int", "len", "num_features", "patch_grid", "post_init", "range", "self", "super"], "swinv2/modeling_swinv2.py:Swinv2Backbone.get_input_embeddings": ["BackboneMixin", "ModelBackbone", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "patch_embeddings", "return", "self"], "swinv2/modeling_swinv2.py:Swinv2Backbone.forward": ["BackboneMixin", "BackboneOutput", "ModelBackbone", "ModelPreTrainedModel", "None", "Tensor", "True", "attentions", "bool", "class", "config", "def", "else", "embedding_output", "embeddings", "encoder", "feature_maps", "for", "forward", "hidden_state", "hidden_states", "if", "in", "input_dimensions", "is", "kwargs", "not", "out_features", "output", "output_attentions", "output_hidden_states", "output_hidden_states_before_downsampling", "outputs", "pixel_values", "r", "reshaped_hidden_states", "return", "return_dict", "self", "stage", "stage_names", "use_return_dict", "zip"], "switch_transformers/modeling_switch_transformers.py:SwitchTransformersTop1Router.__init__": ["Linear", "ModelConfig", "ModelTop1Router", "Module", "__init__", "bias", "class", "classifier", "config", "def", "dtype", "expert_capacity", "getattr", "hidden_size", "ignore_padding_tokens", "jitter_noise", "nn", "num_experts", "router_bias", "router_dtype", "router_ignore_padding_tokens", "router_jitter_noise", "self", "super", "torch"], "switch_transformers/modeling_switch_transformers.py:SwitchTransformersTop1Router.forward": ["ModelTop1Router", "Module", "Tensor", "True", "and", "class", "classifier", "cumsum", "def", "dim", "dtype", "empty_like", "expert_capacity", "expert_capacity_mask", "expert_index", "forward", "functional", "hidden_states", "if", "input_dtype", "jitter_noise", "keepdim", "max", "nn", "num_classes", "num_experts", "one_hot", "r", "return", "router_logits", "router_probs", "self", "softmax", "to", "token_priority", "torch", "training", "tuple", "uniform_", "unsqueeze", "values"], "switch_transformers/modeling_switch_transformers.py:SwitchTransformersLayerNorm.__init__": ["ModelLayerNorm", "Module", "Parameter", "__init__", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "switch_transformers/modeling_switch_transformers.py:SwitchTransformersLayerNorm.forward": ["ModelLayerNorm", "Module", "True", "bfloat16", "class", "def", "dtype", "float16", "float32", "forward", "hidden_states", "if", "in", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "switch_transformers/modeling_switch_transformers.py:SwitchTransformersDenseActDense.__init__": ["ACT2FN", "Dropout", "False", "Linear", "ModelConfig", "ModelDenseActDense", "Module", "__init__", "act", "bias", "class", "config", "d_ff", "d_model", "def", "dense_act_fn", "dropout", "dropout_rate", "nn", "self", "super", "wi", "wo"], "switch_transformers/modeling_switch_transformers.py:SwitchTransformersDenseActDense.forward": ["ModelDenseActDense", "Module", "Tensor", "act", "and", "class", "def", "dropout", "dtype", "forward", "hidden_states", "if", "int8", "isinstance", "nn", "return", "self", "to", "torch", "weight", "wi", "wo"], "switch_transformers/modeling_switch_transformers.py:SwitchTransformersExperts.__init__": ["ModelConfig", "ModelDenseActDense", "ModelExperts", "ModuleDict", "__init__", "class", "config", "def", "expert_", "f", "for", "idx", "in", "nn", "num_experts", "range", "self", "super"], "switch_transformers/modeling_switch_transformers.py:SwitchTransformersExperts.forward": ["ModelExperts", "ModuleDict", "None", "Tensor", "class", "current_hidden_states", "current_state", "def", "dim", "dtype", "expert_", "expert_hit", "expert_idx", "expert_mask", "f", "final_hidden_states", "for", "forward", "greater", "hidden_states", "idx", "in", "index_add_", "nn", "nonzero", "permute", "reshape", "return", "routing_weights", "selected_experts", "self", "shape", "squeeze", "sum", "to", "top_x", "torch", "where", "zeros_like"], "switch_transformers/modeling_switch_transformers.py:SwitchTransformersSparseMLP.__init__": ["ModelConfig", "ModelExperts", "ModelSparseMLP", "ModelTop1Router", "Module", "__init__", "class", "config", "def", "experts", "nn", "router", "self", "super"], "switch_transformers/modeling_switch_transformers.py:SwitchTransformersSparseMLP.forward": ["ModelSparseMLP", "Module", "Tensor", "_", "batch_size", "class", "def", "experts", "forward", "hidden_dim", "hidden_states", "nn", "reshape", "return", "router", "routing_weights", "selected_experts", "self", "sequence_length", "shape", "torch", "view"], "switch_transformers/modeling_switch_transformers.py:SwitchTransformersLayerFF.__init__": ["Dropout", "False", "ModelConfig", "ModelDenseActDense", "ModelLayerFF", "ModelLayerNorm", "ModelSparseMLP", "Module", "__init__", "class", "config", "d_model", "def", "dropout", "dropout_rate", "else", "eps", "if", "is_sparse", "layer_norm", "layer_norm_epsilon", "mlp", "nn", "not", "self", "super"], "switch_transformers/modeling_switch_transformers.py:SwitchTransformersLayerFF.forward": ["ModelLayerFF", "Module", "class", "def", "dropout", "forward", "forwarded_states", "hidden_states", "kwargs", "layer_norm", "mlp", "nn", "output", "return", "self"], "switch_transformers/modeling_switch_transformers.py:SwitchTransformersAttention.__init__": ["Embedding", "False", "Instantiating", "Linear", "ModelAttention", "ModelConfig", "Module", "None", "Please", "__class__", "__init__", "__name__", "a", "and", "bias", "caching", "call", "class", "config", "creating", "d_kv", "d_model", "decoder", "def", "dropout", "dropout_rate", "during", "errors", "f", "forward", "gradient_checkpointing", "has_relative_attention_bias", "if", "inner_dim", "int", "is", "is_decoder", "k", "key_value_proj_dim", "layer_idx", "logger", "make", "n_heads", "nn", "not", "num_heads", "o", "passing", "provide", "q", "recommended", "relative_attention_bias", "relative_attention_max_distance", "relative_attention_num_buckets", "self", "super", "sure", "the", "this", "to", "used", "v", "warning_once", "when", "will", "without"], "switch_transformers/modeling_switch_transformers.py:SwitchTransformersAttention._relative_position_bucket": ["ModelAttention", "Module", "True", "_relative_position_bucket", "abs", "bidirectional", "class", "def", "else", "float", "full_like", "if", "is_small", "log", "long", "math", "max_distance", "max_exact", "min", "nn", "num_buckets", "relative_buckets", "relative_position", "relative_position_if_large", "return", "to", "torch", "where", "zeros_like"], "switch_transformers/modeling_switch_transformers.py:SwitchTransformersAttention.compute_bias": ["ModelAttention", "Module", "None", "_relative_position_bucket", "arange", "bidirectional", "cache_position", "class", "compute_bias", "context_position", "def", "device", "dtype", "else", "if", "is", "is_decoder", "key_length", "long", "max_distance", "memory_position", "nn", "not", "num_buckets", "permute", "query_length", "relative_attention_bias", "relative_attention_max_distance", "relative_attention_num_buckets", "relative_position", "relative_position_bucket", "return", "self", "to", "torch", "unsqueeze", "values", "weight"], "switch_transformers/modeling_switch_transformers.py:SwitchTransformersAttention.forward": ["EncoderDecoderCache", "False", "ModelAttention", "Module", "None", "True", "and", "attn_output", "attn_weights", "batch_size", "cache_position", "causal_mask", "class", "compute_bias", "contiguous", "cross_attention_cache", "curr_past_key_values", "current_states", "def", "device", "dim", "dropout", "dtype", "else", "float", "forward", "functional", "get", "gradient_checkpointing", "has_relative_attention_bias", "hidden_states", "if", "inner_dim", "is", "is_cross_attention", "is_updated", "isinstance", "k", "key_length", "key_states", "key_value_proj_dim", "key_value_states", "keys", "layer_idx", "layers", "mask", "matmul", "n_heads", "nn", "not", "o", "output_attentions", "outputs", "p", "past_key_values", "position_bias", "position_bias_masked", "q", "query_length", "query_states", "real_seq_length", "requires_grad", "return", "scores", "self", "self_attention_cache", "seq_length", "shape", "softmax", "torch", "training", "transpose", "type_as", "update", "use_cache", "v", "value_states", "values", "view", "zeros"], "switch_transformers/modeling_switch_transformers.py:SwitchTransformersLayerSelfAttention.__init__": ["Dropout", "False", "ModelAttention", "ModelLayerNorm", "ModelLayerSelfAttention", "Module", "None", "SelfAttention", "__init__", "class", "config", "d_model", "def", "dropout", "dropout_rate", "eps", "has_relative_attention_bias", "int", "layer_idx", "layer_norm", "layer_norm_epsilon", "nn", "self", "super"], "switch_transformers/modeling_switch_transformers.py:SwitchTransformersLayerSelfAttention.forward": ["False", "ModelLayerSelfAttention", "Module", "None", "SelfAttention", "attention_mask", "attention_output", "cache_position", "class", "def", "dropout", "forward", "hidden_states", "layer_norm", "mask", "nn", "normed_hidden_states", "output_attentions", "outputs", "past_key_values", "position_bias", "return", "self", "use_cache"], "switch_transformers/modeling_switch_transformers.py:SwitchTransformersLayerCrossAttention.__init__": ["Dropout", "EncDecAttention", "False", "ModelAttention", "ModelLayerCrossAttention", "ModelLayerNorm", "Module", "None", "__init__", "class", "config", "d_model", "def", "dropout", "dropout_rate", "eps", "has_relative_attention_bias", "int", "layer_idx", "layer_norm", "layer_norm_epsilon", "nn", "self", "super"], "switch_transformers/modeling_switch_transformers.py:SwitchTransformersLayerCrossAttention.forward": ["EncDecAttention", "False", "ModelLayerCrossAttention", "Module", "None", "attention_mask", "attention_output", "cache_position", "class", "def", "dropout", "forward", "hidden_states", "key_value_states", "layer_norm", "layer_output", "mask", "nn", "normed_hidden_states", "output_attentions", "outputs", "past_key_values", "position_bias", "query_length", "return", "self", "use_cache"], "switch_transformers/modeling_switch_transformers.py:SwitchTransformersBlock.__init__": ["False", "GradientCheckpointingLayer", "ModelBlock", "ModelLayerCrossAttention", "ModelLayerFF", "ModelLayerSelfAttention", "ModuleList", "None", "__init__", "append", "class", "config", "def", "has_relative_attention_bias", "if", "int", "is_decoder", "is_sparse", "layer", "layer_idx", "nn", "self", "super"], "switch_transformers/modeling_switch_transformers.py:SwitchTransformersBlock.forward": ["False", "GradientCheckpointingLayer", "ModelBlock", "None", "_", "and", "any", "attention_mask", "cache_position", "clamp", "clamp_value", "class", "def", "do_cross_attention", "dtype", "encoder_attention_mask", "encoder_decoder_position_bias", "encoder_hidden_states", "finfo", "float16", "forward", "hidden_states", "if", "is", "is_decoder", "isinf", "key_value_states", "kwargs", "layer", "max", "min", "not", "past_key_values", "position_bias", "query_length", "return", "self", "torch", "use_cache"], "switch_transformers/modeling_switch_transformers.py:SwitchTransformersPreTrainedModel._init_weights": ["ModelAttention", "ModelDenseActDense", "ModelEncoderModel", "ModelForConditionalGeneration", "ModelLayerNorm", "ModelModel", "ModelPreTrainedModel", "ModelSparseMLP", "None", "PreTrainedModel", "_init_weights", "and", "bias", "class", "classifier", "config", "constant_", "d_ff", "d_kv", "d_model", "def", "elif", "expert_", "experts", "f", "factor", "for", "has_relative_attention_bias", "hasattr", "idx", "if", "in", "init", "initializer_factor", "is", "isinstance", "k", "key_value_proj_dim", "lm_head", "mean", "module", "n_heads", "normal_", "not", "num_experts", "num_heads", "o", "q", "range", "relative_attention_bias", "router", "self", "shared", "std", "tie_word_embeddings", "v", "weight", "wi", "wo", "zeros_"], "switch_transformers/modeling_switch_transformers.py:SwitchTransformersPreTrainedModel._shift_right": ["In", "Model", "ModelPreTrainedModel", "None", "PreTrainedModel", "See", "ValueError", "_shift_right", "be", "class", "clone", "config", "decoder_start_token_id", "def", "defined", "docs", "for", "has", "if", "information", "input_ids", "is", "it", "masked_fill_", "model", "more", "new_zeros", "pad_token_id", "raise", "return", "self", "set", "shape", "shifted_input_ids", "the", "to", "usually"], "switch_transformers/modeling_switch_transformers.py:SwitchTransformersStack.__init__": ["Dropout", "Embedding", "False", "ModelBlock", "ModelLayerNorm", "ModelPreTrainedModel", "ModelStack", "ModuleList", "__init__", "append", "block", "bool", "class", "config", "d_model", "decoder_sparse_step", "def", "dropout", "dropout_rate", "else", "embed_tokens", "encoder_sparse_step", "eps", "final_layer_norm", "for", "gradient_checkpointing", "has_relative_attention_bias", "i", "if", "in", "is_decoder", "is_sparse", "layer_idx", "layer_norm_epsilon", "nn", "num_decoder_layers", "num_layers", "or", "post_init", "range", "self", "sparse_step", "super", "vocab_size"], "switch_transformers/modeling_switch_transformers.py:SwitchTransformersStack.forward": ["DynamicCache", "EncoderDecoderCache", "MoEModelOutputWithPastAndCrossAttentions", "ModelPreTrainedModel", "ModelStack", "None", "TransformersKwargs", "True", "Unpack", "ValueError", "You", "_", "_update_causal_mask", "a", "and", "arange", "as", "attention_mask", "batch_size", "be", "block", "cache_position", "can", "causal_mask", "class", "config", "decoder", "def", "device", "dropout", "dtype", "elif", "else", "embed_tokens", "embeddings", "encoder_attention_mask", "encoder_batch_size", "encoder_decoder_position_bias", "encoder_extended_attention_mask", "encoder_hidden_shape", "encoder_hidden_states", "encoder_sequence_length", "enumerate", "exactly", "f", "final_layer_norm", "finfo", "for", "forward", "get_seq_length", "have", "hidden_states", "i", "if", "in", "initialize", "input_ids", "inputs_embeds", "invert_attention_mask", "is", "is_decoder", "is_encoder_decoder", "is_torchdynamo_compiling", "isinstance", "kwargs", "last_hidden_state", "layer_module", "mask_seq_length", "min", "model", "must", "not", "of", "one", "ones", "only", "or", "past_key_values", "past_key_values_length", "position_bias", "raise", "return", "self", "self_attention_cache", "seq_length", "set", "shape", "size", "specify", "the", "to", "token", "torch", "use_cache", "used", "valid", "with"], "switch_transformers/modeling_switch_transformers.py:SwitchTransformersStack._update_causal_mask": ["AttentionMaskConverter", "BlockMask", "Cache", "False", "ModelPreTrainedModel", "ModelStack", "None", "Tensor", "Union", "_attn_implementation", "_ignore_causal_mask_sdpa", "_prepare_4d_causal_attention_mask_with_cache_position", "_unmask_unattended", "_update_causal_mask", "and", "any", "attention_mask", "batch_size", "bool", "cache_position", "causal_mask", "class", "config", "cuda", "def", "device", "dtype", "else", "finfo", "flash_attention_2", "flex_attention", "get_max_cache_shape", "get_seq_length", "if", "in", "input_tensor", "inputs_embeds", "is", "is_compileable", "is_training", "isinstance", "make_flex_block_causal_mask", "min", "min_dtype", "not", "npu", "output_attentions", "past_key_values", "past_key_values_length", "past_seen_tokens", "return", "sdpa", "self", "sequence_length", "shape", "target_length", "torch", "training", "type", "using_compilable_cache", "xpu"], "switch_transformers/modeling_switch_transformers.py:SwitchTransformersStack._prepare_4d_causal_attention_mask_with_cache_position": ["ModelPreTrainedModel", "ModelStack", "None", "Tensor", "_prepare_4d_causal_attention_mask_with_cache_position", "and", "arange", "attention_mask", "batch_size", "cache_position", "causal_mask", "class", "clone", "def", "device", "diagonal", "dim", "dtype", "else", "expand", "fill_value", "finfo", "full", "if", "int", "is", "kwargs", "mask_length", "masked_fill", "min", "min_dtype", "not", "padding_mask", "reshape", "return", "sequence_length", "shape", "target_length", "to", "torch", "triu"], "switch_transformers/modeling_switch_transformers.py:SwitchTransformersModel.__init__": ["Embedding", "False", "ModelConfig", "ModelModel", "ModelPreTrainedModel", "ModelStack", "True", "__init__", "class", "config", "copy", "d_model", "decoder", "decoder_config", "deepcopy", "def", "encoder", "encoder_config", "is_decoder", "nn", "post_init", "self", "shared", "super", "use_cache", "vocab_size"], "switch_transformers/modeling_switch_transformers.py:SwitchTransformersModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "decoder", "def", "encoder", "new_embeddings", "self", "set_input_embeddings", "shared"], "switch_transformers/modeling_switch_transformers.py:SwitchTransformersModel.forward": ["BoolTensor", "Cache", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Seq2SeqMoEModelOutput", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "cache_position", "class", "cross_attentions", "decoder", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_input_ids", "decoder_inputs_embeds", "decoder_outputs", "decoder_router_logits", "def", "encoder", "encoder_attention_mask", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "encoder_router_logits", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "last_hidden_state", "past_key_values", "return", "router_logits", "self", "torch", "tuple"], "switch_transformers/modeling_switch_transformers.py:router_z_loss_func": ["Model_logits", "Model_z_loss_func", "Tensor", "_", "def", "dim", "float", "log_z", "logsumexp", "num_groups", "r", "return", "shape", "sum", "tokens_per_group", "torch", "z_loss"], "switch_transformers/modeling_switch_transformers.py:load_balancing_loss_func": ["Model_balancing_loss_func", "Tensor", "axis", "def", "dtype", "expert_indices", "expert_mask", "float", "float32", "functional", "if", "int64", "len", "max", "mean", "nn", "num_experts", "one_hot", "r", "return", "router_prob_per_group_and_expert", "router_probs", "shape", "to", "tokens_per_group_and_expert", "torch", "unsqueeze", "values"], "switch_transformers/modeling_switch_transformers.py:SwitchTransformersForConditionalGeneration.__init__": ["Embedding", "False", "GenerationMixin", "Linear", "ModelConfig", "ModelForConditionalGeneration", "ModelPreTrainedModel", "ModelStack", "True", "__init__", "bias", "class", "config", "copy", "d_model", "decoder", "decoder_config", "deepcopy", "def", "encoder", "encoder_config", "is_decoder", "lm_head", "model_dim", "nn", "num_decoder_layers", "num_layers", "post_init", "router_aux_loss_coef", "router_z_loss_coef", "self", "shared", "super", "use_cache", "vocab_size"], "switch_transformers/modeling_switch_transformers.py:SwitchTransformersForConditionalGeneration.get_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "return", "self", "shared"], "switch_transformers/modeling_switch_transformers.py:SwitchTransformersForConditionalGeneration.set_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "decoder", "def", "encoder", "new_embeddings", "self", "set_input_embeddings", "shared"], "switch_transformers/modeling_switch_transformers.py:SwitchTransformersForConditionalGeneration.forward": ["BoolTensor", "Cache", "CrossEntropyLoss", "False", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Seq2SeqMoEOutput", "Softmax", "Tensor", "TransformersKwargs", "Unpack", "_shift_right", "_unpack_router_logits", "and", "attention_mask", "attentions", "aux_loss", "bool", "cache_position", "class", "config", "cross_attentions", "decoder", "decoder_attention_mask", "decoder_attentions", "decoder_aux_loss", "decoder_expert_indexes", "decoder_hidden_states", "decoder_input_ids", "decoder_inputs_embeds", "decoder_outputs", "decoder_router_logits", "decoder_router_probs", "decoder_sparse_step", "decoder_z_loss", "def", "device", "dim", "else", "encoder", "encoder_attention_mask", "encoder_attentions", "encoder_aux_loss", "encoder_expert_indexes", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "encoder_router_logits", "encoder_router_probs", "encoder_sparse_step", "encoder_z_loss", "forward", "hidden_states", "if", "ignore_index", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "last_hidden_state", "lm_head", "lm_logits", "load_balancing_loss_func", "logits", "loss", "loss_fct", "model_dim", "nn", "not", "output_router_logits", "past_key_values", "return", "router_aux_loss_coef", "router_probs", "router_z_loss_coef", "router_z_loss_func", "self", "sequence_output", "size", "tie_word_embeddings", "to", "torch", "tuple", "view", "z_loss"], "switch_transformers/modeling_switch_transformers.py:SwitchTransformersForConditionalGeneration._unpack_router_logits": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "_unpack_router_logits", "append", "cat", "class", "def", "dim", "expert_indexes", "for", "if", "in", "len", "return", "router_logits", "router_output", "router_outputs", "self", "shape", "torch", "total_expert_indexes", "total_router_logits"], "switch_transformers/modeling_switch_transformers.py:SwitchTransformersForConditionalGeneration.prepare_decoder_input_ids_from_labels": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "Tensor", "_shift_right", "class", "def", "labels", "prepare_decoder_input_ids_from_labels", "return", "self", "torch"], "switch_transformers/modeling_switch_transformers.py:SwitchTransformersEncoderModel.__init__": ["Embedding", "False", "ModelConfig", "ModelEncoderModel", "ModelPreTrainedModel", "ModelStack", "__init__", "class", "config", "copy", "d_model", "deepcopy", "def", "encoder", "encoder_config", "is_encoder_decoder", "nn", "post_init", "self", "shared", "super", "use_cache", "vocab_size"], "switch_transformers/modeling_switch_transformers.py:SwitchTransformersEncoderModel.get_input_embeddings": ["ModelEncoderModel", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "return", "self", "shared"], "switch_transformers/modeling_switch_transformers.py:SwitchTransformersEncoderModel.set_input_embeddings": ["ModelEncoderModel", "ModelPreTrainedModel", "class", "def", "encoder", "new_embeddings", "self", "set_input_embeddings", "shared"], "switch_transformers/modeling_switch_transformers.py:SwitchTransformersEncoderModel.forward": ["False", "FloatTensor", "LongTensor", "MoEModelOutput", "ModelEncoderModel", "ModelPreTrainedModel", "None", "TransformersKwargs", "Unpack", "attention_mask", "bool", "class", "def", "encoder", "encoder_outputs", "forward", "input_ids", "inputs_embeds", "kwargs", "return", "self", "torch", "tuple", "use_cache"], "t5/modeling_t5.py:T5LayerNorm.__init__": ["Module", "Parameter", "T5LayerNorm", "__init__", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "t5/modeling_t5.py:T5LayerNorm.forward": ["Module", "T5LayerNorm", "True", "bfloat16", "class", "def", "dtype", "float16", "float32", "forward", "hidden_states", "if", "in", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "t5/modeling_t5.py:T5DenseActDense.__init__": ["ACT2FN", "Dropout", "False", "Linear", "Module", "T5Config", "T5DenseActDense", "__init__", "act", "bias", "class", "config", "d_ff", "d_model", "def", "dense_act_fn", "dropout", "dropout_rate", "nn", "self", "super", "wi", "wo"], "t5/modeling_t5.py:T5DenseActDense.forward": ["Module", "T5DenseActDense", "Tensor", "act", "and", "class", "def", "dropout", "dtype", "forward", "hidden_states", "if", "int8", "isinstance", "nn", "return", "self", "to", "torch", "weight", "wi", "wo"], "t5/modeling_t5.py:T5DenseGatedActDense.__init__": ["ACT2FN", "Dropout", "False", "Linear", "Module", "T5Config", "T5DenseGatedActDense", "__init__", "act", "bias", "class", "config", "d_ff", "d_model", "def", "dense_act_fn", "dropout", "dropout_rate", "nn", "self", "super", "wi_0", "wi_1", "wo"], "t5/modeling_t5.py:T5DenseGatedActDense.forward": ["Module", "T5DenseGatedActDense", "Tensor", "act", "and", "class", "def", "dropout", "dtype", "forward", "hidden_gelu", "hidden_linear", "hidden_states", "if", "int8", "isinstance", "nn", "return", "self", "to", "torch", "weight", "wi_0", "wi_1", "wo"], "t5/modeling_t5.py:T5LayerFF.__init__": ["DenseReluDense", "Dropout", "Module", "T5Config", "T5DenseActDense", "T5DenseGatedActDense", "T5LayerFF", "T5LayerNorm", "__init__", "class", "config", "d_model", "def", "dropout", "dropout_rate", "else", "eps", "if", "is_gated_act", "layer_norm", "layer_norm_epsilon", "nn", "self", "super"], "t5/modeling_t5.py:T5LayerFF.forward": ["DenseReluDense", "Module", "T5LayerFF", "class", "def", "dropout", "forward", "forwarded_states", "hidden_states", "layer_norm", "nn", "return", "self"], "t5/modeling_t5.py:T5Attention.__init__": ["Embedding", "False", "Instantiating", "Linear", "Module", "None", "Please", "T5Attention", "T5Config", "__class__", "__init__", "__name__", "a", "and", "bias", "caching", "call", "class", "config", "creating", "d_kv", "d_model", "decoder", "def", "dropout", "dropout_rate", "during", "errors", "f", "forward", "gradient_checkpointing", "has_relative_attention_bias", "if", "inner_dim", "int", "is", "is_decoder", "k", "key_value_proj_dim", "layer_idx", "logger", "make", "n_heads", "nn", "not", "num_heads", "o", "passing", "provide", "q", "recommended", "relative_attention_bias", "relative_attention_max_distance", "relative_attention_num_buckets", "self", "super", "sure", "the", "this", "to", "used", "v", "warning_once", "when", "will", "without"], "t5/modeling_t5.py:T5Attention._relative_position_bucket": ["Module", "T5Attention", "True", "_relative_position_bucket", "abs", "bidirectional", "class", "def", "else", "float", "full_like", "if", "is_small", "log", "long", "math", "max_distance", "max_exact", "min", "nn", "num_buckets", "relative_buckets", "relative_position", "relative_position_if_large", "return", "to", "torch", "where", "zeros_like"], "t5/modeling_t5.py:T5Attention.compute_bias": ["Module", "None", "T5Attention", "_relative_position_bucket", "arange", "bidirectional", "cache_position", "class", "compute_bias", "context_position", "def", "device", "dtype", "else", "if", "is", "is_decoder", "key_length", "long", "max_distance", "memory_position", "nn", "not", "num_buckets", "permute", "query_length", "relative_attention_bias", "relative_attention_max_distance", "relative_attention_num_buckets", "relative_position", "relative_position_bucket", "return", "self", "to", "torch", "unsqueeze", "values", "weight"], "t5/modeling_t5.py:T5Attention.forward": ["EncoderDecoderCache", "False", "Module", "None", "T5Attention", "True", "and", "attn_output", "attn_weights", "batch_size", "cache_position", "causal_mask", "class", "compute_bias", "contiguous", "cross_attention_cache", "curr_past_key_values", "current_states", "def", "device", "dim", "dropout", "dtype", "else", "float", "forward", "functional", "get", "gradient_checkpointing", "has_relative_attention_bias", "hidden_states", "if", "inner_dim", "is", "is_cross_attention", "is_updated", "isinstance", "k", "key_length", "key_states", "key_value_proj_dim", "key_value_states", "keys", "layer_idx", "layers", "mask", "matmul", "n_heads", "nn", "not", "o", "output_attentions", "outputs", "p", "past_key_values", "position_bias", "position_bias_masked", "q", "query_length", "query_states", "real_seq_length", "requires_grad", "return", "scores", "self", "self_attention_cache", "seq_length", "shape", "softmax", "torch", "training", "transpose", "type_as", "update", "use_cache", "v", "value_states", "values", "view", "zeros"], "t5/modeling_t5.py:T5LayerSelfAttention.__init__": ["Dropout", "False", "Module", "None", "SelfAttention", "T5Attention", "T5LayerNorm", "T5LayerSelfAttention", "__init__", "class", "config", "d_model", "def", "dropout", "dropout_rate", "eps", "has_relative_attention_bias", "int", "layer_idx", "layer_norm", "layer_norm_epsilon", "nn", "self", "super"], "t5/modeling_t5.py:T5LayerSelfAttention.forward": ["False", "Module", "None", "SelfAttention", "T5LayerSelfAttention", "attention_mask", "attention_output", "cache_position", "class", "def", "dropout", "forward", "hidden_states", "layer_norm", "mask", "nn", "normed_hidden_states", "output_attentions", "outputs", "past_key_values", "position_bias", "return", "self", "use_cache"], "t5/modeling_t5.py:T5LayerCrossAttention.__init__": ["Dropout", "EncDecAttention", "False", "Module", "None", "T5Attention", "T5LayerCrossAttention", "T5LayerNorm", "__init__", "class", "config", "d_model", "def", "dropout", "dropout_rate", "eps", "has_relative_attention_bias", "int", "layer_idx", "layer_norm", "layer_norm_epsilon", "nn", "self", "super"], "t5/modeling_t5.py:T5LayerCrossAttention.forward": ["EncDecAttention", "False", "Module", "None", "T5LayerCrossAttention", "attention_mask", "attention_output", "cache_position", "class", "def", "dropout", "forward", "hidden_states", "key_value_states", "layer_norm", "layer_output", "mask", "nn", "normed_hidden_states", "output_attentions", "outputs", "past_key_values", "position_bias", "query_length", "return", "self", "use_cache"], "t5/modeling_t5.py:T5Block.__init__": ["False", "GradientCheckpointingLayer", "ModuleList", "None", "T5Block", "T5LayerCrossAttention", "T5LayerFF", "T5LayerSelfAttention", "__init__", "append", "class", "config", "def", "has_relative_attention_bias", "if", "int", "is_decoder", "layer", "layer_idx", "nn", "self", "super"], "t5/modeling_t5.py:T5Block.forward": ["False", "GradientCheckpointingLayer", "None", "T5Block", "True", "and", "any", "attention_mask", "attention_outputs", "cache_position", "clamp", "clamp_value", "class", "cross_attention_outputs", "def", "do_cross_attention", "dtype", "encoder_attention_mask", "encoder_decoder_position_bias", "encoder_hidden_states", "finfo", "float16", "forward", "hidden_states", "if", "is", "is_decoder", "isinf", "key_value_states", "layer", "max", "min", "not", "output_attentions", "outputs", "past_key_values", "position_bias", "query_length", "return", "return_dict", "self", "self_attention_outputs", "torch", "use_cache", "where"], "t5/modeling_t5.py:T5ClassificationHead.__init__": ["Dropout", "Linear", "Module", "T5ClassificationHead", "T5Config", "__init__", "class", "classifier_dropout", "config", "d_model", "def", "dense", "dropout", "nn", "num_labels", "out_proj", "p", "self", "super"], "t5/modeling_t5.py:T5ClassificationHead.forward": ["Module", "T5ClassificationHead", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "nn", "out_proj", "return", "self", "tanh", "torch"], "t5/modeling_t5.py:T5PreTrainedModel.dummy_inputs": ["DUMMY_INPUTS", "DUMMY_MASK", "PreTrainedModel", "T5PreTrainedModel", "class", "decoder_attention_mask", "decoder_input_ids", "def", "dummy_inputs", "input_ids", "input_mask", "return", "self", "tensor", "torch"], "t5/modeling_t5.py:T5PreTrainedModel._init_weights": ["None", "PreTrainedModel", "T5Attention", "T5ClassificationHead", "T5DenseActDense", "T5DenseGatedActDense", "T5EncoderModel", "T5ForConditionalGeneration", "T5ForQuestionAnswering", "T5ForTokenClassification", "T5LayerNorm", "T5Model", "T5PreTrainedModel", "_init_weights", "and", "bias", "class", "classifier", "config", "constant_", "d_ff", "d_kv", "d_model", "def", "dense", "elif", "factor", "has_relative_attention_bias", "hasattr", "if", "init", "initializer_factor", "is", "isinstance", "k", "key_value_proj_dim", "lm_head", "mean", "module", "n_heads", "normal_", "not", "num_heads", "o", "out_proj", "q", "qa_outputs", "relative_attention_bias", "self", "shared", "std", "tie_word_embeddings", "v", "weight", "wi", "wi_0", "wi_1", "wo", "zeros_"], "t5/modeling_t5.py:T5PreTrainedModel._shift_right": ["In", "None", "PreTrainedModel", "See", "T5", "T5PreTrainedModel", "ValueError", "_shift_right", "be", "class", "clone", "config", "decoder_start_token_id", "def", "defined", "docs", "for", "has", "if", "information", "input_ids", "is", "it", "masked_fill_", "model", "more", "new_zeros", "pad_token_id", "raise", "return", "self", "set", "shape", "shifted_input_ids", "the", "to", "usually"], "t5/modeling_t5.py:T5Stack.__init__": ["Dropout", "Embedding", "False", "ModuleList", "T5Block", "T5LayerNorm", "T5PreTrainedModel", "T5Stack", "__init__", "block", "bool", "class", "config", "d_model", "def", "dropout", "dropout_rate", "embed_tokens", "eps", "final_layer_norm", "for", "gradient_checkpointing", "has_relative_attention_bias", "i", "in", "is_decoder", "layer_idx", "layer_norm_epsilon", "nn", "num_layers", "post_init", "range", "self", "super", "vocab_size"], "t5/modeling_t5.py:T5Stack.set_input_embeddings": ["T5PreTrainedModel", "T5Stack", "class", "def", "embed_tokens", "new_embeddings", "self", "set_input_embeddings"], "t5/modeling_t5.py:T5Stack.forward": ["BaseModelOutputWithPastAndCrossAttentions", "DynamicCache", "EncoderDecoderCache", "False", "None", "Setting", "T5PreTrainedModel", "T5Stack", "True", "ValueError", "You", "a", "all_attentions", "all_cross_attentions", "all_hidden_states", "and", "arange", "as", "at", "attention_mask", "attentions", "batch_size", "be", "block", "both", "cache_position", "can", "cannot", "checkpointing", "class", "config", "create_bidirectional_mask", "create_causal_mask", "cross_attentions", "decoder", "decoder_", "def", "device", "dropout", "either", "elif", "else", "embed_tokens", "embeddings", "encoder_attention_mask", "encoder_decoder_position_bias", "encoder_extended_attention_mask", "encoder_hidden_states", "err_msg_prefix", "f", "final_layer_norm", "for", "forward", "get_seq_length", "gradient", "gradient_checkpointing", "have", "hidden_states", "if", "in", "incompatible", "initialize", "input_embeds", "input_ids", "input_shape", "inputs_embeds", "is", "is_decoder", "is_encoder_decoder", "isinstance", "kwargs", "last_hidden_state", "layer_module", "layer_outputs", "logger", "model", "not", "only", "or", "output_attentions", "output_hidden_states", "past_key_values", "past_key_values_length", "position_bias", "raise", "return", "return_dict", "same", "self", "self_attention_cache", "seq_length", "set", "size", "specify", "the", "time", "to", "token", "torch", "training", "tuple", "use_cache", "use_return_dict", "used", "v", "valid", "view", "warning_once", "with"], "t5/modeling_t5.py:T5Model.__init__": ["Embedding", "False", "T5Config", "T5Model", "T5PreTrainedModel", "T5Stack", "True", "__init__", "class", "config", "copy", "d_model", "decoder", "decoder_config", "deepcopy", "def", "encoder", "encoder_config", "is_decoder", "nn", "num_decoder_layers", "num_layers", "post_init", "self", "shared", "super", "use_cache", "vocab_size"], "t5/modeling_t5.py:T5Model.get_input_embeddings": ["T5Model", "T5PreTrainedModel", "class", "def", "get_input_embeddings", "return", "self", "shared"], "t5/modeling_t5.py:T5Model.set_input_embeddings": ["T5Model", "T5PreTrainedModel", "class", "decoder", "def", "encoder", "new_embeddings", "self", "set_input_embeddings", "shared"], "t5/modeling_t5.py:T5Model.forward": ["BaseModelOutput", "BoolTensor", "Cache", "FloatTensor", "LongTensor", "None", "Seq2SeqModelOutput", "T5Model", "T5PreTrainedModel", "Tensor", "and", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "cross_attentions", "decoder", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_input_ids", "decoder_inputs_embeds", "decoder_outputs", "def", "elif", "else", "encoder", "encoder_attention_mask", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "isinstance", "kwargs", "last_hidden_state", "len", "not", "output_attentions", "output_hidden_states", "past_key_values", "r", "return", "return_dict", "self", "torch", "tuple", "use_cache", "use_return_dict"], "t5/modeling_t5.py:T5ForConditionalGeneration.__init__": ["Embedding", "False", "GenerationMixin", "Linear", "T5Config", "T5ForConditionalGeneration", "T5PreTrainedModel", "T5Stack", "True", "__init__", "bias", "class", "config", "copy", "d_model", "decoder", "decoder_config", "deepcopy", "def", "encoder", "encoder_config", "is_decoder", "lm_head", "model_dim", "nn", "num_decoder_layers", "num_layers", "post_init", "self", "shared", "super", "use_cache", "vocab_size"], "t5/modeling_t5.py:T5ForConditionalGeneration.get_input_embeddings": ["GenerationMixin", "T5ForConditionalGeneration", "T5PreTrainedModel", "class", "def", "get_input_embeddings", "return", "self", "shared"], "t5/modeling_t5.py:T5ForConditionalGeneration.set_input_embeddings": ["GenerationMixin", "T5ForConditionalGeneration", "T5PreTrainedModel", "class", "decoder", "def", "encoder", "new_embeddings", "self", "set_input_embeddings", "shared"], "t5/modeling_t5.py:T5ForConditionalGeneration.forward": ["BaseModelOutput", "BoolTensor", "Cache", "CrossEntropyLoss", "FloatTensor", "GenerationMixin", "LongTensor", "None", "Seq2SeqLMOutput", "T5ForConditionalGeneration", "T5PreTrainedModel", "Tensor", "_shift_right", "and", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "cross_attentions", "decoder", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_input_ids", "decoder_inputs_embeds", "decoder_outputs", "def", "device", "elif", "else", "encoder", "encoder_attention_mask", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "forward", "hidden_states", "if", "ignore_index", "input_ids", "inputs_embeds", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "len", "lm_head", "lm_logits", "logits", "loss", "loss_fct", "model_dim", "not", "output", "output_attentions", "output_hidden_states", "past_key_values", "r", "return", "return_dict", "scale_decoder_outputs", "self", "sequence_output", "size", "to", "torch", "tuple", "use_cache", "use_return_dict", "view"], "t5/modeling_t5.py:T5ForConditionalGeneration.prepare_decoder_input_ids_from_labels": ["GenerationMixin", "T5ForConditionalGeneration", "T5PreTrainedModel", "Tensor", "_shift_right", "class", "def", "labels", "prepare_decoder_input_ids_from_labels", "return", "self", "torch"], "t5/modeling_t5.py:T5EncoderModel.__init__": ["Embedding", "False", "T5Config", "T5EncoderModel", "T5PreTrainedModel", "T5Stack", "__init__", "class", "config", "d_model", "def", "encoder", "encoder_config", "is_encoder_decoder", "nn", "post_init", "self", "shared", "super", "use_cache", "vocab_size"], "t5/modeling_t5.py:T5EncoderModel.get_input_embeddings": ["T5EncoderModel", "T5PreTrainedModel", "class", "def", "get_input_embeddings", "return", "self", "shared"], "t5/modeling_t5.py:T5EncoderModel.set_input_embeddings": ["T5EncoderModel", "T5PreTrainedModel", "class", "def", "encoder", "new_embeddings", "self", "set_input_embeddings", "shared"], "t5/modeling_t5.py:T5EncoderModel.forward": ["BaseModelOutput", "FloatTensor", "LongTensor", "None", "T5EncoderModel", "T5PreTrainedModel", "attention_mask", "bool", "class", "config", "def", "else", "encoder", "encoder_outputs", "forward", "if", "input_ids", "inputs_embeds", "is", "kwargs", "not", "output_attentions", "output_hidden_states", "r", "return", "return_dict", "self", "torch", "tuple", "use_return_dict"], "t5/modeling_t5.py:T5ForSequenceClassification.__init__": ["T5ClassificationHead", "T5Config", "T5ForSequenceClassification", "T5Model", "T5PreTrainedModel", "__init__", "class", "classification_head", "config", "def", "post_init", "self", "super", "transformer"], "t5/modeling_t5.py:T5ForSequenceClassification.forward": ["All", "BCEWithLogitsLoss", "CrossEntropyLoss", "False", "FloatTensor", "If", "LongTensor", "MSELoss", "None", "NotImplementedError", "Passing", "Please", "Seq2SeqSequenceClassifierOutput", "T5ForSequenceClassification", "T5PreTrainedModel", "Tensor", "ValueError", "_", "__class__", "__name__", "_shift_right", "and", "are", "attention_mask", "batch_size", "be", "bool", "cannot", "class", "classification_head", "config", "cross_attentions", "currently", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_input_ids", "decoder_inputs_embeds", "def", "device", "dtype", "either", "elif", "else", "embeddings", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "eos", "eos_mask", "eos_token_id", "eq", "examples", "f", "for", "forward", "have", "hidden_size", "if", "input", "input_ids", "inputs_embeds", "int", "is", "kwargs", "labels", "len", "list", "logits", "long", "loss", "loss_fct", "multi_label_classification", "must", "no", "not", "num_labels", "number", "of", "or", "output", "output_attentions", "output_hidden_states", "outputs", "pass", "passed", "past_key_values", "problem_type", "r", "raise", "regression", "return", "return_dict", "same", "self", "sentence_representation", "sequence_output", "shape", "single_label_classification", "squeeze", "sum", "supported", "the", "to", "tokens", "torch", "transformer", "tuple", "unique_consecutive", "use_cache", "use_return_dict", "view"], "t5/modeling_t5.py:T5ForTokenClassification.__init__": ["Dropout", "Linear", "T5Config", "T5EncoderModel", "T5ForTokenClassification", "T5PreTrainedModel", "__init__", "class", "classifier", "classifier_dropout", "config", "def", "dropout", "hidden_size", "nn", "num_labels", "post_init", "self", "super", "transformer"], "t5/modeling_t5.py:T5ForTokenClassification.forward": ["CrossEntropyLoss", "None", "T5ForTokenClassification", "T5PreTrainedModel", "Tensor", "TokenClassifierOutput", "attention_mask", "attentions", "bool", "class", "classifier", "config", "def", "dropout", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "not", "num_labels", "output", "output_attentions", "output_hidden_states", "outputs", "r", "return", "return_dict", "self", "torch", "transformer", "tuple", "use_return_dict", "view"], "t5/modeling_t5.py:T5ForQuestionAnswering.__init__": ["Embedding", "False", "Linear", "T5Config", "T5ForQuestionAnswering", "T5PreTrainedModel", "T5Stack", "True", "__init__", "class", "config", "copy", "d_model", "decoder", "decoder_config", "deepcopy", "def", "encoder", "encoder_config", "hidden_size", "is_decoder", "model_dim", "nn", "num_decoder_layers", "num_labels", "num_layers", "post_init", "qa_outputs", "self", "shared", "super", "use_cache", "vocab_size"], "t5/modeling_t5.py:T5ForQuestionAnswering.get_input_embeddings": ["T5ForQuestionAnswering", "T5PreTrainedModel", "class", "def", "get_input_embeddings", "return", "self", "shared"], "t5/modeling_t5.py:T5ForQuestionAnswering.set_input_embeddings": ["T5ForQuestionAnswering", "T5PreTrainedModel", "class", "decoder", "def", "encoder", "new_embeddings", "self", "set_input_embeddings", "shared"], "t5/modeling_t5.py:T5ForQuestionAnswering.forward": ["BaseModelOutput", "BoolTensor", "CrossEntropyLoss", "False", "FloatTensor", "If", "LongTensor", "None", "Please", "Seq2SeqQuestionAnsweringModelOutput", "T5ForQuestionAnswering", "T5PreTrainedModel", "Tensor", "ValueError", "_shift_right", "and", "are", "attention_mask", "attentions", "be", "bool", "cannot", "clamp", "class", "config", "contiguous", "cross_attentions", "decoder", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_input_ids", "decoder_inputs_embeds", "decoder_outputs", "def", "device", "dim", "either", "elif", "else", "encoder", "encoder_attention_mask", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "end_logits", "end_loss", "end_positions", "forward", "hidden_states", "if", "ignore_index", "ignored_index", "input_ids", "inputs_embeds", "is", "isinstance", "kwargs", "last_hidden_state", "len", "logits", "loss", "loss_fct", "no", "not", "or", "output", "output_attentions", "output_hidden_states", "pass", "passed", "past_key_values", "qa_outputs", "r", "raise", "return", "return_dict", "self", "sequence_output", "size", "split", "squeeze", "start_logits", "start_loss", "start_positions", "to", "torch", "total_loss", "tuple", "use_cache", "use_return_dict"], "t5gemma/modeling_t5gemma.py:T5GemmaRMSNorm.__init__": ["ModelRMSNorm", "Module", "Parameter", "__init__", "class", "def", "dim", "eps", "float", "int", "nn", "self", "super", "torch", "weight", "zeros"], "t5gemma/modeling_t5gemma.py:T5GemmaRMSNorm._norm": ["ModelRMSNorm", "Module", "True", "_norm", "class", "def", "eps", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "torch", "x"], "t5gemma/modeling_t5gemma.py:T5GemmaRMSNorm.forward": ["ModelRMSNorm", "Module", "_norm", "class", "def", "float", "forward", "nn", "output", "return", "self", "type_as", "weight", "x"], "t5gemma/modeling_t5gemma.py:T5GemmaRMSNorm.extra_repr": ["ModelRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "weight"], "t5gemma/modeling_t5gemma.py:T5GemmaMLP.__init__": ["ACT2FN", "Dropout", "False", "Linear", "ModelMLP", "Module", "__init__", "act_fn", "bias", "class", "config", "def", "down_proj", "dropout", "dropout_rate", "gate_proj", "hidden_activation", "hidden_size", "intermediate_size", "nn", "self", "super", "up_proj"], "t5gemma/modeling_t5gemma.py:T5GemmaMLP.forward": ["ModelMLP", "Module", "act_fn", "class", "def", "down_proj", "dropout", "forward", "gate_proj", "hidden_states", "nn", "return", "self", "up_proj", "x"], "t5gemma/modeling_t5gemma.py:T5GemmaRotaryEmbedding.__init__": ["Callable", "False", "ModelConfig", "ModelRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "t5gemma/modeling_t5gemma.py:T5GemmaRotaryEmbedding.compute_default_rope_parameters": ["ModelConfig", "ModelRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "t5gemma/modeling_t5gemma.py:T5GemmaRotaryEmbedding.forward": ["False", "ModelRotaryEmbedding", "Module", "None", "and", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "t5gemma/modeling_t5gemma.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "t5gemma/modeling_t5gemma.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cos", "def", "k", "k_embed", "q", "q_embed", "return", "rotate_half", "sin", "unsqueeze", "unsqueeze_dim"], "t5gemma/modeling_t5gemma.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "t5gemma/modeling_t5gemma.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "head_dim", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softcap", "softmax", "tanh", "to", "torch", "training", "transpose", "tuple", "value", "value_states"], "t5gemma/modeling_t5gemma.py:T5GemmaSelfAttention.__init__": ["Linear", "ModelModuleConfig", "ModelSelfAttention", "Module", "None", "__init__", "attention_bias", "attention_dropout", "attn_logit_softcapping", "bias", "class", "config", "def", "else", "getattr", "hasattr", "head_dim", "hidden_size", "if", "int", "is_causal", "is_decoder", "k_proj", "layer_idx", "layer_type", "layer_types", "nn", "num_attention_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "q_proj", "query_pre_attn_scalar", "scaling", "self", "sliding_attention", "sliding_window", "super", "v_proj"], "t5gemma/modeling_t5gemma.py:T5GemmaSelfAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "FlashAttentionKwargs", "LongTensor", "ModelSelfAttention", "Module", "None", "Tensor", "Unpack", "_attn_implementation", "apply_rotary_pos_emb", "attention_dropout", "attention_interface", "attention_mask", "attn_logit_softcapping", "attn_output", "attn_weights", "cache_kwargs", "cache_position", "class", "config", "contiguous", "cos", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "head_dim", "hidden_shape", "hidden_states", "if", "input_shape", "is", "k_proj", "key_states", "kwargs", "layer_idx", "nn", "not", "o_proj", "past_key_values", "position_embeddings", "q_proj", "query_states", "reshape", "return", "scaling", "self", "shape", "sin", "sliding_window", "softcap", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "view"], "t5gemma/modeling_t5gemma.py:T5GemmaCrossAttention.__init__": ["Cross", "False", "Linear", "ModelCrossAttention", "ModelModuleConfig", "Module", "None", "ValueError", "__init__", "attention", "attention_bias", "attention_dropout", "attn_logit_softcapping", "be", "bias", "class", "config", "cross_attention_hidden_size", "def", "getattr", "head_dim", "hidden_size", "if", "int", "is", "is_causal", "k_proj", "layer_idx", "needs", "nn", "num_attention_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "q_proj", "query_pre_attn_scalar", "raise", "scaling", "self", "specified", "super", "to", "v_proj"], "t5gemma/modeling_t5gemma.py:T5GemmaCrossAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "Encoder", "FlashAttentionKwargs", "ModelCrossAttention", "Module", "None", "Tensor", "True", "Unpack", "ValueError", "_attn_implementation", "attention", "attention_dropout", "attention_interface", "attention_mask", "attn_logit_softcapping", "attn_output", "attn_weights", "class", "config", "contiguous", "cross", "cross_attention_cache", "curr_past_key_values", "def", "dropout", "eager", "eager_attention_forward", "else", "encoder_hidden_shape", "encoder_hidden_states", "encoder_input_shape", "for", "forward", "get", "head_dim", "hidden", "hidden_shape", "hidden_states", "if", "input_shape", "is", "is_updated", "k_proj", "key_states", "keys", "kwargs", "layer_idx", "layers", "nn", "not", "o_proj", "or", "past_key_values", "q_proj", "query_states", "raise", "required", "reshape", "return", "scaling", "self", "shape", "sliding_window", "softcap", "state", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "values", "view"], "t5gemma/modeling_t5gemma.py:T5GemmaEncoderLayer.__init__": ["Dropout", "GradientCheckpointingLayer", "ModelEncoderLayer", "ModelMLP", "ModelRMSNorm", "ModelSelfAttention", "__init__", "attention_type", "class", "config", "def", "dropout", "dropout_rate", "eps", "hidden_size", "int", "layer_idx", "layer_types", "mlp", "nn", "post_feedforward_layernorm", "post_self_attn_layernorm", "pre_feedforward_layernorm", "pre_self_attn_layernorm", "rms_norm_eps", "self", "self_attn", "super"], "t5gemma/modeling_t5gemma.py:T5GemmaEncoderLayer.forward": ["FloatTensor", "GradientCheckpointingLayer", "LongTensor", "ModelEncoderLayer", "None", "Tensor", "_", "attention_mask", "class", "def", "dropout", "forward", "hidden_states", "kwargs", "mlp", "past_key_values", "position_embeddings", "position_ids", "post_feedforward_layernorm", "post_self_attn_layernorm", "pre_feedforward_layernorm", "pre_self_attn_layernorm", "residual", "return", "self", "self_attn", "torch", "tuple"], "t5gemma/modeling_t5gemma.py:T5GemmaDecoderLayer.__init__": ["Dropout", "GradientCheckpointingLayer", "ModelCrossAttention", "ModelDecoderLayer", "ModelMLP", "ModelRMSNorm", "ModelSelfAttention", "__init__", "attention_type", "class", "config", "cross_attn", "def", "dropout", "dropout_rate", "eps", "hidden_size", "int", "layer_idx", "layer_types", "mlp", "nn", "post_cross_attn_layernorm", "post_feedforward_layernorm", "post_self_attn_layernorm", "pre_cross_attn_layernorm", "pre_feedforward_layernorm", "pre_self_attn_layernorm", "rms_norm_eps", "self", "self_attn", "super"], "t5gemma/modeling_t5gemma.py:T5GemmaDecoderLayer.forward": ["EncoderDecoderCache", "False", "FloatTensor", "GradientCheckpointingLayer", "LongTensor", "ModelDecoderLayer", "None", "Tensor", "_", "attention_mask", "bool", "cache_position", "class", "cross_attn", "def", "dropout", "else", "encoder_attention_mask", "encoder_hidden_states", "forward", "hidden_states", "if", "is", "kwargs", "mlp", "not", "past_key_values", "position_embeddings", "position_ids", "post_cross_attn_layernorm", "post_feedforward_layernorm", "post_self_attn_layernorm", "pre_cross_attn_layernorm", "pre_feedforward_layernorm", "pre_self_attn_layernorm", "residual", "return", "self", "self_attention_cache", "self_attn", "torch", "tuple", "use_cache"], "t5gemma/modeling_t5gemma.py:T5GemmaClassificationHead.__init__": ["Dropout", "Linear", "ModelClassificationHead", "Module", "__init__", "class", "classifier_dropout_rate", "def", "dropout", "float", "hidden_size", "int", "nn", "num_labels", "out_proj", "p", "self", "super"], "t5gemma/modeling_t5gemma.py:T5GemmaClassificationHead.forward": ["ModelClassificationHead", "Module", "Tensor", "class", "def", "dropout", "forward", "hidden_states", "nn", "out_proj", "return", "self", "torch"], "t5gemma/modeling_t5gemma.py:T5GemmaLMHead.__init__": ["False", "Linear", "ModelLMHead", "Module", "__init__", "bias", "bool", "class", "def", "hidden_size", "int", "nn", "out_proj", "self", "super", "vocab_size"], "t5gemma/modeling_t5gemma.py:T5GemmaLMHead.forward": ["ModelLMHead", "Module", "Tensor", "class", "def", "forward", "hidden_states", "logits", "nn", "out_proj", "return", "self", "torch"], "t5gemma/modeling_t5gemma.py:T5GemmaPreTrainedModel._init_weights": ["ModelClassificationHead", "ModelLMHead", "ModelPreTrainedModel", "None", "PreTrainedModel", "RMSNorm", "__class__", "__name__", "_init_weights", "and", "bias", "class", "config", "def", "elif", "hasattr", "if", "in", "init", "initializer_range", "is", "isinstance", "mean", "module", "normal_", "not", "out_proj", "scale", "self", "shape", "std", "super", "tie_word_embeddings", "weight", "zeros_"], "t5gemma/modeling_t5gemma.py:T5GemmaPreTrainedModel._shift_right": ["ModelPreTrainedModel", "None", "PreTrainedModel", "ValueError", "_shift_right", "be", "bos_token_id", "class", "clone", "config", "decoder", "decoder_start_token_id", "def", "defined", "has", "if", "input_ids", "is", "masked_fill_", "model", "new_zeros", "pad_token_id", "raise", "return", "self", "shape", "shifted_input_ids", "to"], "t5gemma/modeling_t5gemma.py:bidirectional_mask_function": ["Callable", "Model_mask_function", "None", "Tensor", "attention_mask", "batch_idx", "bool", "def", "dtype", "head_idx", "if", "inner_mask", "int", "is", "kv_idx", "ones", "q_idx", "return", "to", "torch"], "t5gemma/modeling_t5gemma.py:sliding_window_bidirectional_mask_function": ["Callable", "Model_window", "Model_window_bidirectional_mask_function", "batch_idx", "bool", "def", "head_idx", "inner_mask", "int", "kv_idx", "q_idx", "return"], "t5gemma/modeling_t5gemma.py:make_default_2d_attention_mask": ["LongTensor", "Model_default_2d_attention_mask", "None", "Tensor", "ValueError", "attention_mask", "def", "device", "dtype", "else", "for", "hidden_states", "if", "information", "int", "is", "long", "not", "ones", "pad_token_id", "padding", "raise", "required", "return", "shape", "to", "token_ids", "torch"], "t5gemma/modeling_t5gemma.py:T5GemmaEncoder.__init__": ["Dropout", "Embedding", "False", "ModelEncoder", "ModelEncoderLayer", "ModelPreTrainedModel", "ModelRMSNorm", "ModelRotaryEmbedding", "ModuleList", "__init__", "class", "config", "def", "dropout", "dropout_rate", "embed_tokens", "eps", "for", "gradient_checkpointing", "hidden_size", "in", "layer_idx", "layers", "nn", "norm", "num_hidden_layers", "pad_token_id", "padding_idx", "post_init", "range", "rms_norm_eps", "rotary_emb", "self", "super", "vocab_size"], "t5gemma/modeling_t5gemma.py:T5GemmaEncoder.forward": ["BaseModelOutput", "FloatTensor", "LongTensor", "ModelEncoder", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "and_mask_function", "arange", "attention_mask", "attention_type", "bidirectional_mask_function", "cache_position", "class", "config", "create_causal_mask", "create_sliding_window_causal_mask", "def", "device", "dict", "dropout", "dtype", "embed_tokens", "exactly", "for", "forward", "full_attention", "hidden_size", "hidden_states", "if", "in", "input_embeds", "input_ids", "inputs_embeds", "is", "isinstance", "kwargs", "last_hidden_state", "layer_module", "layers", "make_default_2d_attention_mask", "mask_kwargs", "must", "norm", "normalizer", "not", "num_hidden_layers", "of", "one", "or", "or_mask_function", "pad_token_id", "past_key_values", "pop", "position_embeddings", "position_ids", "raise", "return", "rotary_emb", "self", "self_attn_mask_mapping", "shape", "sliding_attention", "sliding_window", "sliding_window_bidirectional_mask_function", "specify", "tensor", "torch", "unsqueeze"], "t5gemma/modeling_t5gemma.py:T5GemmaDecoder.__init__": ["Dropout", "Embedding", "False", "ModelDecoder", "ModelDecoderLayer", "ModelPreTrainedModel", "ModelRMSNorm", "ModelRotaryEmbedding", "ModuleList", "__init__", "class", "config", "def", "dropout", "dropout_rate", "embed_tokens", "eps", "for", "gradient_checkpointing", "hidden_size", "in", "layer_idx", "layers", "nn", "norm", "num_hidden_layers", "pad_token_id", "padding_idx", "post_init", "range", "rms_norm_eps", "rotary_emb", "self", "super", "vocab_size"], "t5gemma/modeling_t5gemma.py:T5GemmaDecoder.forward": ["BaseModelOutputWithPastAndCrossAttentions", "DynamicCache", "EncoderDecoderCache", "FloatTensor", "LongTensor", "ModelDecoder", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "and", "arange", "attention_mask", "attention_type", "be", "bidirectional_mask_function", "bool", "cache_position", "class", "config", "create_causal_mask", "create_sliding_window_causal_mask", "cross_attn_mask_mapping", "decoder", "def", "device", "dict", "dropout", "dtype", "else", "embed_tokens", "encoder_attention_mask", "encoder_hidden_states", "exactly", "for", "forward", "full_attention", "get_seq_length", "given", "hidden_size", "hidden_states", "if", "in", "input_embeds", "input_ids", "inputs_embeds", "is", "isinstance", "kwargs", "last_hidden_state", "layer_module", "layers", "make_default_2d_attention_mask", "mask_kwargs", "must", "norm", "normalizer", "not", "num_hidden_layers", "of", "one", "or", "or_mask_function", "pad_token_id", "past_key_values", "past_seen_tokens", "position_embeddings", "position_ids", "raise", "return", "rotary_emb", "self", "self_attention_cache", "self_attn_mask_mapping", "shape", "sliding_attention", "specify", "tensor", "torch", "training", "unsqueeze", "use_cache"], "t5gemma/modeling_t5gemma.py:T5GemmaModel.__init__": ["ModelConfig", "ModelDecoder", "ModelEncoder", "ModelEncoderModel", "ModelModel", "ModelPreTrainedModel", "Use", "ValueError", "__init__", "class", "config", "decoder", "def", "encoder", "if", "instead", "is_encoder_decoder", "modeling", "not", "only", "post_init", "raise", "self", "super", "support"], "t5gemma/modeling_t5gemma.py:T5GemmaModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "encoder", "get_input_embeddings", "return", "self"], "t5gemma/modeling_t5gemma.py:T5GemmaModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "encoder", "new_embeddings", "return", "self", "set_input_embeddings"], "t5gemma/modeling_t5gemma.py:T5GemmaModel.forward": ["BaseModelOutput", "BoolTensor", "EncoderDecoderCache", "False", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Seq2SeqModelOutput", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "bool", "cache_position", "class", "cross_attentions", "decoder", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_input_ids", "decoder_inputs_embeds", "decoder_outputs", "decoder_position_ids", "def", "else", "encoder", "encoder_attention_mask", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "forward", "get", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "last_hidden_state", "output_hidden_states", "past_key_values", "position_ids", "r", "return", "self", "torch", "use_cache"], "t5gemma/modeling_t5gemma.py:T5GemmaEncoderModel.__init__": ["ModelConfig", "ModelEncoder", "ModelEncoderModel", "ModelModel", "ModelPreTrainedModel", "Use", "ValueError", "__init__", "class", "config", "def", "encoder", "if", "instead", "is_encoder_decoder", "model", "only", "post_init", "raise", "self", "super", "supports"], "t5gemma/modeling_t5gemma.py:T5GemmaEncoderModel.get_input_embeddings": ["ModelEncoderModel", "ModelPreTrainedModel", "class", "def", "encoder", "get_input_embeddings", "return", "self"], "t5gemma/modeling_t5gemma.py:T5GemmaEncoderModel.set_input_embeddings": ["ModelEncoderModel", "ModelPreTrainedModel", "class", "def", "encoder", "new_embeddings", "return", "self", "set_input_embeddings"], "t5gemma/modeling_t5gemma.py:T5GemmaEncoderModel.forward": ["BaseModelOutput", "FloatTensor", "LongTensor", "ModelEncoderModel", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "class", "def", "encoder", "encoder_outputs", "forward", "input_ids", "inputs_embeds", "kwargs", "position_ids", "return", "self", "torch"], "t5gemma/modeling_t5gemma.py:T5GemmaForConditionalGeneration.__init__": ["ForMaskedLM", "GenerationMixin", "ModelConfig", "ModelForConditionalGeneration", "ModelLMHead", "ModelModel", "ModelPreTrainedModel", "True", "__init__", "class", "config", "decoder", "def", "hidden_size", "is_encoder_decoder", "lm_head", "loss_type", "model", "post_init", "self", "super", "vocab_size"], "t5gemma/modeling_t5gemma.py:T5GemmaForConditionalGeneration.set_output_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "lm_head", "new_embeddings", "out_proj", "self", "set_output_embeddings"], "t5gemma/modeling_t5gemma.py:T5GemmaForConditionalGeneration.get_output_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "get_output_embeddings", "lm_head", "out_proj", "return", "self"], "t5gemma/modeling_t5gemma.py:T5GemmaForConditionalGeneration.forward": ["BaseModelOutput", "BoolTensor", "EncoderDecoderCache", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Seq2SeqLMOutput", "Seq2SeqModelOutput", "Tensor", "TransformersKwargs", "Unpack", "_shift_right", "and", "attention_mask", "bool", "cache_position", "class", "config", "cross_attentions", "decoder_attention_mask", "decoder_attentions", "decoder_config", "decoder_hidden_states", "decoder_input_ids", "decoder_inputs_embeds", "decoder_outputs", "decoder_position_ids", "def", "else", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "final_logit_softcapping", "forward", "get_decoder", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "past_key_values", "position_ids", "r", "return", "self", "slice", "slice_indices", "tanh", "torch", "tuple", "use_cache", "vocab_size"], "t5gemma/modeling_t5gemma.py:T5GemmaForConditionalGeneration.prepare_decoder_input_ids_from_labels": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "Tensor", "_shift_right", "class", "def", "labels", "prepare_decoder_input_ids_from_labels", "return", "self", "torch"], "t5gemma/modeling_t5gemma.py:T5GemmaForSequenceClassification.__init__": ["ModelClassificationHead", "ModelConfig", "ModelEncoderModel", "ModelForSequenceClassification", "ModelModel", "ModelPreTrainedModel", "None", "__init__", "bool", "class", "classifier_dropout", "classifier_dropout_rate", "config", "decoder", "def", "else", "encoder", "getattr", "hidden_size", "if", "is", "is_encoder_decoder", "model", "not", "num_labels", "post_init", "r", "score", "self", "super"], "t5gemma/modeling_t5gemma.py:T5GemmaForSequenceClassification.get_input_embeddings": ["ModelForSequenceClassification", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "model", "return", "self"], "t5gemma/modeling_t5gemma.py:T5GemmaForSequenceClassification.set_input_embeddings": ["ModelForSequenceClassification", "ModelPreTrainedModel", "class", "def", "model", "self", "set_input_embeddings", "value"], "t5gemma/modeling_t5gemma.py:T5GemmaForSequenceClassification.forward": ["BaseModelOutput", "Cannot", "False", "FloatTensor", "If", "LongTensor", "ModelForSequenceClassification", "ModelPreTrainedModel", "None", "NotImplementedError", "Passing", "Please", "Results", "Seq2SeqModelOutput", "SequenceClassifierOutput", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "__class__", "__name__", "_shift_right", "and", "arange", "are", "argmax", "attention_mask", "attentions", "batch", "batch_size", "be", "cannot", "clamp", "class", "config", "conjunction", "currently", "decoder", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_input_ids", "decoder_inputs_embeds", "decoder_position_ids", "def", "defined", "detect", "device", "dtype", "either", "elif", "else", "embeddings", "encoder", "encoder_outputs", "f", "for", "forward", "handle", "hidden_states", "if", "in", "input", "input_ids", "inputs_embeds", "int32", "is", "is_encoder_decoder", "kwargs", "labels", "last_hidden_state", "last_non_pad_token", "logger", "logits", "loss", "loss_function", "max", "may", "mode", "model", "no", "non_pad_mask", "not", "or", "outputs", "pad_token_id", "padding", "pass", "passed", "pooled_logits", "position_ids", "r", "raise", "return", "score", "self", "shape", "sizes", "supported", "to", "token", "token_indices", "tokens", "torch", "unexpected", "use_cache", "using", "warning_once", "will", "with"], "t5gemma/modeling_t5gemma.py:T5GemmaForTokenClassification.__init__": ["ModelClassificationHead", "ModelConfig", "ModelEncoderModel", "ModelForTokenClassification", "ModelModel", "ModelPreTrainedModel", "None", "__init__", "bool", "class", "classifier_dropout", "classifier_dropout_rate", "config", "decoder", "def", "else", "encoder", "getattr", "hidden_size", "if", "is", "is_encoder_decoder", "model", "not", "num_labels", "post_init", "r", "score", "self", "super"], "t5gemma/modeling_t5gemma.py:T5GemmaForTokenClassification.get_input_embeddings": ["ModelForTokenClassification", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "model", "return", "self"], "t5gemma/modeling_t5gemma.py:T5GemmaForTokenClassification.set_input_embeddings": ["ModelForTokenClassification", "ModelPreTrainedModel", "class", "def", "model", "self", "set_input_embeddings", "value"], "t5gemma/modeling_t5gemma.py:T5GemmaForTokenClassification.forward": ["BaseModelOutput", "False", "FloatTensor", "If", "LongTensor", "ModelForTokenClassification", "ModelPreTrainedModel", "None", "NotImplementedError", "Passing", "Please", "Seq2SeqModelOutput", "Tensor", "TokenClassifierOutput", "TransformersKwargs", "Unpack", "ValueError", "__class__", "__name__", "_shift_right", "and", "are", "attention_mask", "attentions", "be", "cannot", "class", "config", "currently", "decoder", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_input_ids", "decoder_inputs_embeds", "decoder_position_ids", "def", "either", "else", "embeddings", "encoder", "encoder_outputs", "f", "for", "forward", "hidden_states", "if", "in", "input", "input_ids", "inputs_embeds", "is", "is_encoder_decoder", "kwargs", "labels", "last_hidden_state", "logits", "loss", "loss_function", "mode", "model", "no", "not", "or", "outputs", "pass", "passed", "position_ids", "r", "raise", "return", "score", "self", "supported", "torch", "use_cache"], "t5gemma2/modeling_t5gemma2.py:T5Gemma2RMSNorm.__init__": ["ModelRMSNorm", "Module", "Parameter", "__init__", "class", "def", "dim", "eps", "float", "int", "nn", "self", "super", "torch", "weight", "zeros"], "t5gemma2/modeling_t5gemma2.py:T5Gemma2RMSNorm._norm": ["ModelRMSNorm", "Module", "True", "_norm", "class", "def", "eps", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "torch", "x"], "t5gemma2/modeling_t5gemma2.py:T5Gemma2RMSNorm.forward": ["ModelRMSNorm", "Module", "_norm", "class", "def", "float", "forward", "nn", "output", "return", "self", "type_as", "weight", "x"], "t5gemma2/modeling_t5gemma2.py:T5Gemma2RMSNorm.extra_repr": ["ModelRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "weight"], "t5gemma2/modeling_t5gemma2.py:T5Gemma2MLP.__init__": ["ACT2FN", "Dropout", "False", "Linear", "ModelMLP", "ModelTextConfig", "Module", "__init__", "act_fn", "bias", "class", "config", "def", "down_proj", "dropout", "dropout_rate", "gate_proj", "hidden_activation", "hidden_size", "intermediate_size", "nn", "self", "super", "up_proj"], "t5gemma2/modeling_t5gemma2.py:T5Gemma2MLP.forward": ["ModelMLP", "Module", "act_fn", "class", "def", "down_proj", "dropout", "forward", "gate_proj", "hidden_states", "nn", "return", "self", "up_proj", "x"], "t5gemma2/modeling_t5gemma2.py:T5Gemma2RotaryEmbedding.__init__": ["Callable", "False", "ModelRotaryEmbedding", "ModelTextConfig", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "_attention_scaling", "_inv_freq", "_original_inv_freq", "class", "clone", "compute_default_rope_parameters", "config", "continue", "curr_attention_scaling", "curr_inv_freq", "def", "default", "device", "f", "for", "if", "in", "is", "layer_type", "layer_types", "list", "max_position_embeddings", "max_seq_len_cached", "nn", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_params", "rope_type", "self", "set", "setattr", "super"], "t5gemma2/modeling_t5gemma2.py:T5Gemma2RotaryEmbedding.compute_default_rope_parameters": ["ModelRotaryEmbedding", "ModelTextConfig", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "layer_type", "nn", "num_attention_heads", "or", "return", "rope_parameters", "rope_theta", "seq_len", "str", "to", "torch", "tuple"], "t5gemma2/modeling_t5gemma2.py:T5Gemma2RotaryEmbedding.forward": ["False", "ModelRotaryEmbedding", "Module", "None", "_attention_scaling", "_inv_freq", "and", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "f", "float", "forward", "freqs", "getattr", "if", "inv_freq", "inv_freq_expanded", "isinstance", "layer_type", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "t5gemma2/modeling_t5gemma2.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "t5gemma2/modeling_t5gemma2.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cos", "def", "k", "k_embed", "q", "q_embed", "return", "rotate_half", "sin", "unsqueeze", "unsqueeze_dim"], "t5gemma2/modeling_t5gemma2.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "t5gemma2/modeling_t5gemma2.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "head_dim", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softcap", "softmax", "tanh", "to", "torch", "training", "transpose", "tuple", "value", "value_states"], "t5gemma2/modeling_t5gemma2.py:T5Gemma2SelfAttention.__init__": ["False", "Linear", "ModelRMSNorm", "ModelSelfAttention", "ModelTextConfig", "Module", "None", "__init__", "attention_bias", "attention_dropout", "attn_logit_softcapping", "bias", "class", "config", "def", "dim", "else", "eps", "getattr", "hasattr", "head_dim", "hidden_size", "if", "int", "is_causal", "is_sliding", "k_norm", "k_proj", "layer_idx", "layer_type", "layer_types", "nn", "num_attention_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "q_norm", "q_proj", "query_pre_attn_scalar", "rms_norm_eps", "scaling", "self", "sliding_attention", "sliding_window", "super", "v_proj"], "t5gemma2/modeling_t5gemma2.py:T5Gemma2SelfAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "LongTensor", "ModelSelfAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_attn_implementation", "apply_rotary_pos_emb", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "cache_kwargs", "cache_position", "class", "config", "contiguous", "cos", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "head_dim", "hidden_shape", "hidden_states", "if", "input_shape", "is", "k_norm", "k_proj", "key_states", "kwargs", "layer_idx", "nn", "not", "o_proj", "past_key_values", "position_embeddings", "q_norm", "q_proj", "query_states", "reshape", "return", "scaling", "self", "shape", "sin", "sliding_window", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "view"], "t5gemma2/modeling_t5gemma2.py:T5Gemma2MergedAttention.__init__": ["False", "Linear", "ModelMergedAttention", "ModelRMSNorm", "ModelTextConfig", "Module", "None", "__init__", "attention_bias", "attention_dropout", "attn_logit_softcapping", "bias", "class", "config", "def", "dim", "else", "eps", "getattr", "hasattr", "head_dim", "hidden_size", "if", "int", "is_causal", "is_sliding", "k_norm", "k_proj", "layer_idx", "layer_type", "layer_types", "nn", "num_attention_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "q_norm", "q_proj", "query_pre_attn_scalar", "rms_norm_eps", "scaling", "self", "sliding_attention", "sliding_window", "super", "v_proj"], "t5gemma2/modeling_t5gemma2.py:T5Gemma2MergedAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "EncoderDecoderCache", "FlashAttentionKwargs", "LongTensor", "ModelMergedAttention", "Module", "None", "Tensor", "True", "Unpack", "_attn_implementation", "apply_rotary_pos_emb", "attention_dropout", "attention_interface", "attn_output", "attn_weights", "cache_kwargs", "cache_position", "cat", "class", "config", "contiguous", "cos", "cross_attention_cache", "cross_attn_weights", "cross_hidden_shape", "cross_input_shape", "cross_key_size", "cross_key_states", "cross_value_states", "def", "dim", "dropout", "eager", "eager_attention_forward", "else", "encoder_hidden_states", "forward", "get", "head_dim", "hidden_shape", "hidden_states", "if", "input_shape", "is", "is_updated", "k_norm", "k_proj", "key_states", "keys", "kwargs", "layer_idx", "layers", "merged_attention_mask", "nn", "not", "o_proj", "or", "past_key_values", "position_embeddings", "q_norm", "q_proj", "query_states", "reshape", "return", "scaling", "self", "self_attention_cache", "self_attn_weights", "shape", "sin", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "values", "view"], "t5gemma2/modeling_t5gemma2.py:T5Gemma2EncoderLayer.__init__": ["Dropout", "GradientCheckpointingLayer", "ModelEncoderLayer", "ModelMLP", "ModelRMSNorm", "ModelSelfAttention", "__init__", "attention_type", "class", "config", "def", "dropout", "dropout_rate", "eps", "hidden_size", "int", "layer_idx", "layer_types", "mlp", "nn", "post_feedforward_layernorm", "post_self_attn_layernorm", "pre_feedforward_layernorm", "pre_self_attn_layernorm", "rms_norm_eps", "self", "self_attn", "super"], "t5gemma2/modeling_t5gemma2.py:T5Gemma2EncoderLayer.forward": ["FloatTensor", "GradientCheckpointingLayer", "LongTensor", "ModelEncoderLayer", "None", "Tensor", "_", "attention_mask", "class", "def", "dropout", "forward", "hidden_states", "kwargs", "mlp", "past_key_values", "position_embeddings", "position_ids", "post_feedforward_layernorm", "post_self_attn_layernorm", "pre_feedforward_layernorm", "pre_self_attn_layernorm", "residual", "return", "self", "self_attn", "torch", "tuple"], "t5gemma2/modeling_t5gemma2.py:T5Gemma2DecoderLayer.__init__": ["Dropout", "GradientCheckpointingLayer", "ModelDecoderLayer", "ModelMLP", "ModelMergedAttention", "ModelRMSNorm", "__init__", "attention_type", "class", "config", "def", "dropout", "dropout_rate", "eps", "hidden_size", "int", "layer_idx", "layer_types", "mlp", "nn", "post_feedforward_layernorm", "post_self_attn_layernorm", "pre_feedforward_layernorm", "pre_self_attn_layernorm", "rms_norm_eps", "self", "self_attn", "super"], "t5gemma2/modeling_t5gemma2.py:T5Gemma2DecoderLayer.forward": ["EncoderDecoderCache", "False", "FloatTensor", "GradientCheckpointingLayer", "LongTensor", "ModelDecoderLayer", "None", "Tensor", "_", "bool", "cache_position", "class", "def", "dropout", "encoder_hidden_states", "forward", "hidden_states", "kwargs", "merged_attention_mask", "mlp", "past_key_values", "position_embeddings", "position_ids", "post_feedforward_layernorm", "post_self_attn_layernorm", "pre_feedforward_layernorm", "pre_self_attn_layernorm", "residual", "return", "self", "self_attn", "torch", "tuple", "use_cache"], "t5gemma2/modeling_t5gemma2.py:T5Gemma2LMHead.__init__": ["False", "Linear", "ModelLMHead", "Module", "__init__", "bias", "bool", "class", "def", "hidden_size", "int", "nn", "out_proj", "self", "super", "vocab_size"], "t5gemma2/modeling_t5gemma2.py:T5Gemma2LMHead.forward": ["ModelLMHead", "Module", "Tensor", "class", "def", "forward", "hidden_states", "logits", "nn", "out_proj", "return", "self", "torch"], "t5gemma2/modeling_t5gemma2.py:T5Gemma2ClassificationHead.__init__": ["Dropout", "Linear", "ModelClassificationHead", "Module", "__init__", "class", "classifier_dropout_rate", "def", "dropout", "float", "hidden_size", "int", "nn", "num_labels", "out_proj", "p", "self", "super"], "t5gemma2/modeling_t5gemma2.py:T5Gemma2ClassificationHead.forward": ["ModelClassificationHead", "Module", "Tensor", "class", "def", "dropout", "forward", "hidden_states", "nn", "out_proj", "return", "self", "torch"], "t5gemma2/modeling_t5gemma2.py:T5Gemma2MultiModalProjector.__init__": ["AvgPool2d", "ModelEncoderConfig", "ModelMultiModalProjector", "ModelRMSNorm", "Module", "Parameter", "__init__", "avg_pool", "class", "config", "def", "eps", "hidden_size", "image_size", "int", "kernel_size", "layer_norm_eps", "mm_input_projection_weight", "mm_soft_emb_norm", "mm_tokens_per_image", "nn", "patch_size", "patches_per_image", "self", "stride", "super", "text_config", "tokens_per_side", "torch", "vision_config", "zeros"], "t5gemma2/modeling_t5gemma2.py:T5Gemma2MultiModalProjector.forward": ["ModelMultiModalProjector", "Module", "Tensor", "_", "avg_pool", "batch_size", "class", "contiguous", "def", "flatten", "forward", "matmul", "mm_input_projection_weight", "mm_soft_emb_norm", "nn", "normed_vision_outputs", "patches_per_image", "pooled_vision_outputs", "projected_vision_outputs", "reshape", "reshaped_vision_outputs", "return", "self", "seq_length", "shape", "torch", "transpose", "type_as", "vision_outputs"], "t5gemma2/modeling_t5gemma2.py:T5Gemma2TextScaledWordEmbedding.__init__": ["Embedding", "False", "ModelTextScaledWordEmbedding", "Parameter", "__init__", "class", "def", "embed_scale", "embedding_dim", "eoi_embedding", "eoi_token_index", "float", "int", "nn", "num_embeddings", "padding_idx", "persistent", "register_buffer", "scalar_embed_scale", "self", "super", "tensor", "torch", "zeros"], "t5gemma2/modeling_t5gemma2.py:T5Gemma2TextScaledWordEmbedding.forward": ["Embedding", "ModelTextScaledWordEmbedding", "Tensor", "class", "def", "dtype", "embed_scale", "eoi_embedding", "eoi_token_index", "forward", "input_embeddings", "input_ids", "nn", "return", "self", "super", "to", "torch", "weight"], "t5gemma2/modeling_t5gemma2.py:T5Gemma2PreTrainedModel._init_weights": ["ModelClassificationHead", "ModelMultiModalProjector", "ModelPreTrainedModel", "ModelRotaryEmbedding", "ModelTextScaledWordEmbedding", "None", "PreTrainedModel", "RMSNorm", "ROPE_INIT_FUNCTIONS", "_", "__class__", "__name__", "_init_weights", "_inv_freq", "_original_inv_freq", "and", "bias", "class", "compute_default_rope_parameters", "config", "constant_", "copy_", "curr_inv_freq", "def", "default", "elif", "embed_scale", "eoi_embedding", "f", "for", "getattr", "hasattr", "if", "in", "init", "initializer_range", "is", "isinstance", "layer_type", "layer_types", "mean", "mm_input_projection_weight", "module", "normal_", "not", "out_proj", "rope_init_fn", "rope_type", "scalar_embed_scale", "scale", "self", "shape", "std", "super", "weight", "zeros_"], "t5gemma2/modeling_t5gemma2.py:T5Gemma2PreTrainedModel.prepare_decoder_input_ids_from_labels": ["ModelPreTrainedModel", "None", "PreTrainedModel", "ValueError", "be", "bos_token_id", "class", "clone", "config", "decoder", "decoder_config", "decoder_start_token_id", "def", "defined", "has", "if", "input_ids", "is", "masked_fill_", "model", "new_zeros", "pad_token_id", "prepare_decoder_input_ids_from_labels", "raise", "return", "self", "shape", "shifted_input_ids", "to"], "t5gemma2/modeling_t5gemma2.py:sliding_window_mask_function": ["Callable", "Model_window", "Model_window_mask_function", "True", "batch_idx", "bool", "def", "dist", "else", "head_idx", "if", "inner_mask", "int", "is_causal", "kv_idx", "left_mask", "left_window_size", "q_idx", "return", "right_mask", "right_window_size"], "t5gemma2/modeling_t5gemma2.py:T5Gemma2Encoder.__init__": ["AutoModel", "Dropout", "False", "ModelEncoder", "ModelEncoderConfig", "ModelEncoderLayer", "ModelMultiModalProjector", "ModelPreTrainedModel", "ModelRMSNorm", "ModelRotaryEmbedding", "ModelTextScaledWordEmbedding", "ModuleList", "__init__", "class", "config", "def", "dropout", "dropout_rate", "embed_scale", "embed_tokens", "eoi_token_index", "eps", "for", "from_config", "gradient_checkpointing", "hidden_size", "in", "int", "layer_idx", "layers", "multi_modal_projector", "nn", "norm", "num_hidden_layers", "pad_token_id", "padding_idx", "post_init", "range", "rms_norm_eps", "rotary_emb", "self", "super", "text_config", "vision_config", "vision_tower", "vocab_size"], "t5gemma2/modeling_t5gemma2.py:T5Gemma2Encoder.get_image_features": ["ModelEncoder", "ModelPreTrainedModel", "Tensor", "class", "def", "get_image_features", "image_features", "last_hidden_state", "multi_modal_projector", "pixel_values", "return", "self", "torch", "vision_outputs", "vision_tower"], "t5gemma2/modeling_t5gemma2.py:T5Gemma2Encoder.get_image_placeholder_mask": ["Either", "FloatTensor", "Image", "LongTensor", "ModelEncoder", "ModelPreTrainedModel", "None", "ValueError", "all", "and", "be", "class", "config", "def", "device", "do", "dtype", "else", "expand_as", "f", "features", "get_image_placeholder_mask", "get_input_embeddings", "has", "if", "image", "image_features", "image_token_id", "input_ids", "inputs_embeds", "is", "long", "match", "n_image_features", "n_image_tokens", "not", "numel", "or", "provided", "raise", "return", "self", "shape", "special_image_mask", "sum", "tensor", "to", "tokens", "torch", "unsqueeze"], "t5gemma2/modeling_t5gemma2.py:T5Gemma2Encoder.preprocess_image_features": ["FloatTensor", "LongTensor", "ModelEncoder", "ModelPreTrainedModel", "None", "Tensor", "class", "def", "device", "dtype", "get_image_features", "get_image_placeholder_mask", "image_features", "image_mask", "input_ids", "inputs_embeds", "masked_scatter", "pixel_values", "preprocess_image_features", "return", "self", "to", "torch"], "t5gemma2/modeling_t5gemma2.py:T5Gemma2Encoder.forward": ["BaseModelOutput", "False", "FloatTensor", "LongTensor", "ModelEncoder", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "and_mask_function", "arange", "attention_mask", "attention_type", "class", "config", "create_bidirectional_mask", "def", "del", "device", "dict", "dropout", "embed_tokens", "exactly", "for", "forward", "full_attention", "hidden_states", "if", "in", "input_embeds", "input_ids", "inputs_embeds", "is", "is_causal", "isinstance", "kwargs", "last_hidden_state", "layer_module", "layer_type", "layer_types", "layers", "mask_kwargs", "must", "norm", "not", "num_hidden_layers", "of", "one", "or", "past_key_values", "pixel_values", "pop", "position_embeddings", "position_ids", "preprocess_image_features", "raise", "return", "rotary_emb", "self", "self_attn_mask_mapping", "shape", "sliding_attention", "sliding_window", "sliding_window_mask_function", "specify", "text_config", "token_type_ids", "torch", "unsqueeze"], "t5gemma2/modeling_t5gemma2.py:bidirectional_mask_function": ["Callable", "Model_mask_function", "None", "Tensor", "attention_mask", "batch_idx", "bool", "def", "dtype", "head_idx", "if", "inner_mask", "int", "is", "kv_idx", "ones", "q_idx", "return", "to", "torch"], "t5gemma2/modeling_t5gemma2.py:T5Gemma2Decoder.__init__": ["Dropout", "False", "ModelDecoder", "ModelDecoderConfig", "ModelDecoderLayer", "ModelPreTrainedModel", "ModelRMSNorm", "ModelRotaryEmbedding", "ModelTextScaledWordEmbedding", "ModuleList", "__init__", "class", "config", "def", "dropout", "dropout_rate", "embed_scale", "embed_tokens", "eoi_token_index", "eps", "for", "gradient_checkpointing", "hidden_size", "in", "int", "layer_idx", "layers", "nn", "norm", "num_hidden_layers", "pad_token_id", "padding_idx", "post_init", "range", "rms_norm_eps", "rotary_emb", "self", "super", "vocab_size"], "t5gemma2/modeling_t5gemma2.py:T5Gemma2Decoder.forward": ["BaseModelOutputWithPastAndCrossAttentions", "DynamicCache", "EncoderDecoderCache", "FloatTensor", "LongTensor", "ModelDecoder", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "ValueError", "You", "and", "and_mask_function", "arange", "args", "attention_mask", "attention_type", "be", "bidirectional_mask_function", "bool", "cache_position", "cat", "class", "config", "create_causal_mask", "create_sliding_window_causal_mask", "cross_attn_mask_mapping", "decoder", "def", "device", "dict", "dim", "dropout", "dtype", "else", "embed_tokens", "encoder_attention_mask", "encoder_hidden_states", "exactly", "for", "forward", "full_attention", "get_seq_length", "given", "hidden_states", "if", "in", "input_embeds", "input_ids", "inputs_embeds", "is", "isinstance", "kwargs", "lambda", "last_hidden_state", "layer_module", "layer_type", "layer_types", "layers", "mask_kwargs", "merged_attn_mask_mapping", "must", "norm", "not", "num_hidden_layers", "of", "one", "or", "or_mask_function", "past_key_values", "past_seen_tokens", "position_embeddings", "position_ids", "raise", "return", "rotary_emb", "self", "self_attention_cache", "self_attn_mask_mapping", "shape", "sliding_attention", "specify", "tensor", "torch", "training", "unsqueeze", "use_cache"], "t5gemma2/modeling_t5gemma2.py:T5Gemma2Model.__init__": ["ModelConfig", "ModelDecoder", "ModelEncoder", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "config", "decoder", "def", "encoder", "eoi_token_index", "post_init", "self", "super"], "t5gemma2/modeling_t5gemma2.py:T5Gemma2Model.get_encoder": ["ModelModel", "ModelPreTrainedModel", "class", "def", "encoder", "get_encoder", "return", "self"], "t5gemma2/modeling_t5gemma2.py:T5Gemma2Model.get_decoder": ["ModelModel", "ModelPreTrainedModel", "class", "decoder", "def", "get_decoder", "return", "self"], "t5gemma2/modeling_t5gemma2.py:T5Gemma2Model.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "encoder", "get_input_embeddings", "return", "self"], "t5gemma2/modeling_t5gemma2.py:T5Gemma2Model.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "encoder", "new_embeddings", "return", "self", "set_input_embeddings"], "t5gemma2/modeling_t5gemma2.py:T5Gemma2Model.forward": ["BaseModelOutput", "BoolTensor", "EncoderDecoderCache", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Seq2SeqModelOutput", "Tensor", "TransformersKwargs", "True", "Unpack", "attention_mask", "attentions", "bool", "cache_position", "class", "cross_attentions", "decoder", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_input_ids", "decoder_inputs_embeds", "decoder_outputs", "decoder_position_ids", "def", "encoder", "encoder_attention_mask", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "last_hidden_state", "past_key_values", "pixel_values", "position_ids", "r", "return", "return_dict", "self", "torch", "use_cache"], "t5gemma2/modeling_t5gemma2.py:T5Gemma2ForConditionalGeneration.__init__": ["ForMaskedLM", "GenerationMixin", "ModelConfig", "ModelForConditionalGeneration", "ModelLMHead", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "config", "decoder", "def", "hidden_size", "lm_head", "loss_type", "model", "post_init", "self", "super", "vocab_size"], "t5gemma2/modeling_t5gemma2.py:T5Gemma2ForConditionalGeneration.set_output_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "lm_head", "new_embeddings", "out_proj", "self", "set_output_embeddings"], "t5gemma2/modeling_t5gemma2.py:T5Gemma2ForConditionalGeneration.get_output_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "get_output_embeddings", "lm_head", "out_proj", "return", "self"], "t5gemma2/modeling_t5gemma2.py:T5Gemma2ForConditionalGeneration.get_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "model", "return", "self"], "t5gemma2/modeling_t5gemma2.py:T5Gemma2ForConditionalGeneration.set_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "model", "self", "set_input_embeddings", "value"], "t5gemma2/modeling_t5gemma2.py:T5Gemma2ForConditionalGeneration.get_encoder": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "get_encoder", "model", "return", "self"], "t5gemma2/modeling_t5gemma2.py:T5Gemma2ForConditionalGeneration.get_decoder": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "get_decoder", "model", "return", "self"], "t5gemma2/modeling_t5gemma2.py:T5Gemma2ForConditionalGeneration.get_image_features": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "get_encoder", "get_image_features", "pixel_values", "return", "self"], "t5gemma2/modeling_t5gemma2.py:T5Gemma2ForConditionalGeneration.vision_tower": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "get_encoder", "return", "self", "vision_tower"], "t5gemma2/modeling_t5gemma2.py:T5Gemma2ForConditionalGeneration.forward": ["BaseModelOutput", "BoolTensor", "EncoderDecoderCache", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Seq2SeqLMOutput", "Seq2SeqModelOutput", "Tensor", "TransformersKwargs", "Unpack", "and", "attention_mask", "bool", "cache_position", "class", "config", "cross_attentions", "decoder", "decoder_attention_mask", "decoder_attentions", "decoder_config", "decoder_hidden_states", "decoder_input_ids", "decoder_inputs_embeds", "decoder_outputs", "decoder_position_ids", "def", "else", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "final_logit_softcapping", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "past_key_values", "pixel_values", "position_ids", "prepare_decoder_input_ids_from_labels", "r", "return", "self", "slice", "slice_indices", "tanh", "torch", "tuple", "use_cache", "vocab_size"], "t5gemma2/modeling_t5gemma2.py:T5Gemma2ForConditionalGeneration._prepare_cache_for_generation": ["DynamicCache", "EncoderDecoderCache", "False", "GenerationConfig", "GenerationMixin", "GenerationMode", "Model", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "StaticCache", "The", "True", "ValueError", "_cache", "_prepare_cache_for_generation", "and", "batch_size", "be", "bool", "cache", "cache_implementation", "class", "config", "copy", "cross_attention_cache", "cross_attn_cache_kwargs", "cross_attn_cls", "cross_attn_config", "decoder", "deepcopy", "def", "del", "dict", "else", "encoder_outputs", "for", "generation_config", "generation_mode", "get", "get_text_config", "hasattr", "if", "in", "int", "internal", "is", "is_updated", "isinstance", "layer_types", "len", "max_cache_len", "max_cache_length", "model", "model_kwargs", "must", "not", "of", "offload_cache", "offloaded", "offloading", "past_key_values", "raise", "return", "self", "shape", "sliding_window", "super", "type", "use_cache"], "t5gemma2/modeling_t5gemma2.py:T5Gemma2ForSequenceClassification.__init__": ["ModelClassificationHead", "ModelConfig", "ModelForSequenceClassification", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "classifier_dropout", "classifier_dropout_rate", "config", "decoder", "def", "getattr", "hidden_size", "model", "num_labels", "post_init", "score", "self", "super"], "t5gemma2/modeling_t5gemma2.py:T5Gemma2ForSequenceClassification.get_input_embeddings": ["ModelForSequenceClassification", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "model", "return", "self"], "t5gemma2/modeling_t5gemma2.py:T5Gemma2ForSequenceClassification.set_input_embeddings": ["ModelForSequenceClassification", "ModelPreTrainedModel", "class", "def", "model", "self", "set_input_embeddings", "value"], "t5gemma2/modeling_t5gemma2.py:T5Gemma2ForSequenceClassification.forward": ["BaseModelOutput", "False", "FloatTensor", "LongTensor", "ModelForSequenceClassification", "ModelPreTrainedModel", "None", "NotImplementedError", "Passing", "Seq2SeqModelOutput", "SequenceClassifierOutput", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "__class__", "__name__", "arange", "argmax", "attention_mask", "attentions", "batch_size", "clamp", "class", "config", "currently", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_input_ids", "decoder_inputs_embeds", "decoder_position_ids", "def", "device", "dtype", "embeddings", "encoder_outputs", "f", "for", "forward", "have", "hidden_states", "if", "input", "input_ids", "inputs_embeds", "int32", "is", "kwargs", "labels", "last_hidden_state", "last_non_pad_token", "logits", "loss", "loss_function", "max", "model", "non_pad_mask", "not", "or", "outputs", "pad_token_id", "pixel_values", "pooled_logits", "position_ids", "prepare_decoder_input_ids_from_labels", "r", "raise", "return", "score", "self", "shape", "specify", "supported", "to", "token_indices", "torch", "use_cache"], "t5gemma2/modeling_t5gemma2.py:T5Gemma2ForTokenClassification.__init__": ["ModelClassificationHead", "ModelConfig", "ModelForTokenClassification", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "classifier_dropout", "classifier_dropout_rate", "config", "decoder", "def", "getattr", "hidden_size", "model", "num_labels", "post_init", "score", "self", "super"], "t5gemma2/modeling_t5gemma2.py:T5Gemma2ForTokenClassification.get_input_embeddings": ["ModelForTokenClassification", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "model", "return", "self"], "t5gemma2/modeling_t5gemma2.py:T5Gemma2ForTokenClassification.set_input_embeddings": ["ModelForTokenClassification", "ModelPreTrainedModel", "class", "def", "model", "self", "set_input_embeddings", "value"], "t5gemma2/modeling_t5gemma2.py:T5Gemma2ForTokenClassification.forward": ["BaseModelOutput", "False", "FloatTensor", "LongTensor", "ModelForTokenClassification", "ModelPreTrainedModel", "None", "NotImplementedError", "Passing", "Seq2SeqModelOutput", "Tensor", "TokenClassifierOutput", "TransformersKwargs", "Unpack", "ValueError", "You", "__class__", "__name__", "attention_mask", "attentions", "class", "config", "currently", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_input_ids", "decoder_inputs_embeds", "decoder_position_ids", "def", "embeddings", "encoder_outputs", "f", "for", "forward", "have", "hidden_states", "if", "input", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "last_hidden_state", "logits", "loss", "loss_function", "model", "not", "or", "outputs", "pixel_values", "position_ids", "prepare_decoder_input_ids_from_labels", "r", "raise", "return", "score", "self", "specify", "supported", "to", "torch", "use_cache"], "table_transformer/modeling_table_transformer.py:TableTransformerFrozenBatchNorm2d.__init__": ["ModelFrozenBatchNorm2d", "Module", "__init__", "bias", "class", "def", "n", "nn", "ones", "register_buffer", "running_mean", "running_var", "self", "super", "torch", "weight", "zeros"], "table_transformer/modeling_table_transformer.py:TableTransformerFrozenBatchNorm2d._load_from_state_dict": ["ModelFrozenBatchNorm2d", "Module", "_load_from_state_dict", "class", "def", "del", "error_msgs", "if", "in", "local_metadata", "missing_keys", "nn", "num_batches_tracked", "num_batches_tracked_key", "prefix", "self", "state_dict", "strict", "super", "unexpected_keys"], "table_transformer/modeling_table_transformer.py:TableTransformerFrozenBatchNorm2d.forward": ["ModelFrozenBatchNorm2d", "Module", "bias", "class", "def", "epsilon", "forward", "nn", "reshape", "return", "rsqrt", "running_mean", "running_var", "scale", "self", "weight", "x"], "table_transformer/modeling_table_transformer.py:replace_batch_norm": ["BatchNorm2d", "ModelFrozenBatchNorm2d", "Model_batch_norm", "_modules", "bias", "children", "copy_", "def", "device", "for", "if", "in", "isinstance", "len", "list", "meta", "model", "module", "name", "named_children", "new_module", "nn", "num_features", "r", "running_mean", "running_var", "torch", "weight"], "table_transformer/modeling_table_transformer.py:TableTransformerConvEncoder.__init__": ["Either", "False", "ModelConvEncoder", "Module", "None", "True", "ValueError", "__init__", "and", "backbone", "backbone_config", "backbone_kwargs", "backbone_model_type", "be", "channels", "class", "config", "copy", "create_model", "def", "dilation", "elif", "else", "feature_info", "features_only", "for", "get", "getattr", "if", "in", "in_chans", "intermediate_channel_sizes", "is", "kwargs", "layer2", "layer3", "layer4", "load_backbone", "model", "model_type", "name", "named_parameters", "nn", "no_grad", "not", "num_channels", "or", "out_indices", "output_stride", "parameter", "pop", "pretrained", "provided", "raise", "replace_batch_norm", "requires_backends", "requires_grad_", "resnet", "self", "should", "stage", "super", "the", "timm", "torch", "use_pretrained_backbone", "use_timm_backbone", "with"], "table_transformer/modeling_table_transformer.py:TableTransformerConvEncoder.forward": ["ModelConvEncoder", "Module", "None", "Tensor", "append", "bool", "class", "config", "def", "else", "feature_map", "feature_maps", "features", "float", "for", "forward", "functional", "if", "in", "interpolate", "mask", "model", "nn", "out", "pixel_mask", "pixel_values", "return", "self", "shape", "size", "to", "torch", "use_timm_backbone"], "table_transformer/modeling_table_transformer.py:TableTransformerConvModel.__init__": ["ModelConvModel", "Module", "__init__", "class", "conv_encoder", "def", "nn", "position_embedding", "self", "super"], "table_transformer/modeling_table_transformer.py:TableTransformerConvModel.forward": ["ModelConvModel", "Module", "append", "class", "conv_encoder", "def", "dtype", "feature_map", "for", "forward", "in", "mask", "nn", "out", "pixel_mask", "pixel_values", "pos", "position_embedding", "return", "self", "to"], "table_transformer/modeling_table_transformer.py:TableTransformerSinePositionEmbedding.__init__": ["False", "ModelSinePositionEmbedding", "Module", "None", "True", "ValueError", "__init__", "and", "be", "class", "def", "embedding_dim", "if", "is", "math", "nn", "normalize", "not", "passed", "pi", "raise", "scale", "self", "should", "super", "temperature"], "table_transformer/modeling_table_transformer.py:TableTransformerSinePositionEmbedding.forward": ["ModelSinePositionEmbedding", "Module", "No", "None", "ValueError", "arange", "cat", "class", "cos", "cumsum", "def", "device", "dim", "dim_t", "div", "dtype", "embedding_dim", "flatten", "float", "float32", "floor", "forward", "if", "int64", "is", "mask", "nn", "normalize", "permute", "pixel", "pixel_mask", "pixel_values", "pos", "pos_x", "pos_y", "provided", "raise", "return", "rounding_mode", "scale", "self", "sin", "stack", "temperature", "torch", "x_embed", "y_embed"], "table_transformer/modeling_table_transformer.py:TableTransformerLearnedPositionEmbedding.__init__": ["Embedding", "ModelLearnedPositionEmbedding", "Module", "__init__", "class", "column_embeddings", "def", "embedding_dim", "nn", "row_embeddings", "self", "super"], "table_transformer/modeling_table_transformer.py:TableTransformerLearnedPositionEmbedding.forward": ["ModelLearnedPositionEmbedding", "Module", "None", "arange", "cat", "class", "column_embeddings", "def", "device", "dim", "forward", "height", "height_values", "nn", "permute", "pixel_mask", "pixel_values", "pos", "repeat", "return", "row_embeddings", "self", "shape", "torch", "unsqueeze", "width", "width_values", "x_emb", "y_emb"], "table_transformer/modeling_table_transformer.py:build_position_encoding": ["ModelLearnedPositionEmbedding", "ModelSinePositionEmbedding", "Model_position_encoding", "Not", "True", "ValueError", "config", "d_model", "def", "elif", "else", "f", "if", "learned", "n_steps", "normalize", "position_embedding", "position_embedding_type", "raise", "return", "sine", "supported"], "table_transformer/modeling_table_transformer.py:TableTransformerAttention.__init__": ["Linear", "ModelAttention", "Module", "True", "ValueError", "__init__", "and", "be", "bias", "bool", "by", "class", "def", "divisible", "dropout", "embed_dim", "f", "float", "got", "head_dim", "if", "int", "k_proj", "must", "nn", "num_heads", "out_proj", "q_proj", "raise", "scaling", "self", "super", "v_proj"], "table_transformer/modeling_table_transformer.py:TableTransformerAttention._shape": ["ModelAttention", "Module", "Tensor", "_shape", "batch_size", "class", "contiguous", "def", "head_dim", "int", "nn", "num_heads", "return", "self", "seq_len", "tensor", "torch", "transpose", "view"], "table_transformer/modeling_table_transformer.py:TableTransformerAttention.with_pos_embed": ["ModelAttention", "Module", "None", "Tensor", "class", "def", "else", "if", "is", "nn", "object_queries", "return", "self", "tensor", "torch", "with_pos_embed"], "table_transformer/modeling_table_transformer.py:TableTransformerAttention.forward": ["Attention", "False", "ModelAttention", "Module", "None", "Tensor", "ValueError", "_shape", "attention_mask", "attn_output", "attn_probs", "attn_weights", "attn_weights_reshaped", "batch_size", "be", "bmm", "bool", "but", "class", "def", "dim", "dropout", "dtype", "else", "embed_dim", "f", "forward", "functional", "head_dim", "hidden_states", "hidden_states_original", "if", "inf", "is", "is_cross_attention", "k_proj", "key_states", "key_value_states", "key_value_states_original", "mask", "masked_fill_", "nn", "not", "num_heads", "object_queries", "of", "out_proj", "output_attentions", "p", "proj_shape", "q_proj", "query_states", "raise", "reshape", "return", "scaling", "self", "should", "size", "softmax", "source_len", "spatial_position_embeddings", "target_len", "torch", "training", "transpose", "tuple", "v_proj", "value_states", "view", "weights", "with_pos_embed", "zeros_like"], "table_transformer/modeling_table_transformer.py:TableTransformerEncoderLayer.__init__": ["ACT2FN", "LayerNorm", "Linear", "ModelAttention", "ModelConfig", "ModelEncoderLayer", "Module", "__init__", "activation_dropout", "activation_fn", "activation_function", "attention_dropout", "class", "config", "d_model", "def", "dropout", "embed_dim", "encoder_attention_heads", "encoder_ffn_dim", "fc1", "fc2", "final_layer_norm", "nn", "num_heads", "self", "self_attn", "self_attn_layer_norm", "super"], "table_transformer/modeling_table_transformer.py:TableTransformerEncoderLayer.forward": ["False", "ModelEncoderLayer", "Module", "None", "Tensor", "activation_dropout", "activation_fn", "any", "attention_mask", "attn_weights", "bool", "clamp", "clamp_value", "class", "def", "dropout", "dtype", "fc1", "fc2", "final_layer_norm", "finfo", "forward", "functional", "hidden_states", "if", "isinf", "isnan", "max", "min", "nn", "object_queries", "or", "output_attentions", "outputs", "p", "residual", "return", "self", "self_attn", "self_attn_layer_norm", "torch", "training"], "table_transformer/modeling_table_transformer.py:TableTransformerDecoderLayer.__init__": ["ACT2FN", "GradientCheckpointingLayer", "LayerNorm", "Linear", "ModelAttention", "ModelConfig", "ModelDecoderLayer", "__init__", "activation_dropout", "activation_fn", "activation_function", "attention_dropout", "class", "config", "d_model", "decoder_attention_heads", "decoder_ffn_dim", "def", "dropout", "embed_dim", "encoder_attn", "encoder_attn_layer_norm", "fc1", "fc2", "final_layer_norm", "nn", "num_heads", "self", "self_attn", "self_attn_layer_norm", "super"], "table_transformer/modeling_table_transformer.py:TableTransformerDecoderLayer.forward": ["False", "GradientCheckpointingLayer", "ModelDecoderLayer", "None", "Tensor", "activation_dropout", "activation_fn", "attention_mask", "bool", "class", "cross_attn_weights", "def", "dropout", "encoder_attention_mask", "encoder_attn", "encoder_attn_layer_norm", "encoder_hidden_states", "fc1", "fc2", "final_layer_norm", "forward", "functional", "hidden_states", "if", "is", "key_value_states", "nn", "not", "object_queries", "output_attentions", "outputs", "p", "query_position_embeddings", "residual", "return", "self", "self_attn", "self_attn_layer_norm", "self_attn_weights", "spatial_position_embeddings", "torch", "training"], "table_transformer/modeling_table_transformer.py:TableTransformerPreTrainedModel._init_weights": ["Conv2d", "Embedding", "False", "Linear", "ModelLearnedPositionEmbedding", "ModelPreTrainedModel", "None", "PreTrainedModel", "_init_weights", "_is_hf_initialized", "and", "bias", "class", "column_embeddings", "config", "def", "elif", "getattr", "if", "init", "init_std", "is", "isinstance", "mean", "module", "nn", "normal_", "not", "padding_idx", "row_embeddings", "self", "std", "uniform_", "weight", "zeros_"], "table_transformer/modeling_table_transformer.py:TableTransformerEncoder.__init__": ["LayerNorm", "ModelConfig", "ModelEncoder", "ModelEncoderLayer", "ModelPreTrainedModel", "ModuleList", "_", "__init__", "class", "config", "d_model", "def", "dropout", "encoder_layerdrop", "encoder_layers", "for", "in", "layerdrop", "layernorm", "layers", "nn", "post_init", "range", "self", "super"], "table_transformer/modeling_table_transformer.py:TableTransformerEncoder.forward": ["BaseModelOutput", "False", "ModelEncoder", "ModelPreTrainedModel", "None", "True", "_prepare_4d_attention_mask", "all_attentions", "attention_mask", "attentions", "class", "config", "def", "dropout", "dropout_probability", "dtype", "else", "encoder_layer", "encoder_states", "for", "forward", "functional", "hidden_states", "if", "in", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layer_outputs", "layerdrop", "layernorm", "layers", "nn", "not", "object_queries", "output_attentions", "output_hidden_states", "p", "r", "rand", "return", "return_dict", "self", "to_drop", "torch", "training", "tuple", "use_return_dict", "v"], "table_transformer/modeling_table_transformer.py:TableTransformerDecoder.__init__": ["False", "LayerNorm", "ModelConfig", "ModelDecoder", "ModelDecoderLayer", "ModelPreTrainedModel", "ModuleList", "_", "__init__", "class", "config", "d_model", "decoder_layerdrop", "decoder_layers", "def", "dropout", "for", "gradient_checkpointing", "in", "layerdrop", "layernorm", "layers", "nn", "post_init", "range", "self", "super"], "table_transformer/modeling_table_transformer.py:TableTransformerDecoder.forward": ["ModelDecoder", "ModelDecoderOutput", "ModelPreTrainedModel", "None", "_prepare_4d_attention_mask", "all_cross_attentions", "all_hidden_states", "all_self_attns", "and", "attention_mask", "attentions", "auxiliary_loss", "class", "combined_attention_mask", "config", "continue", "cross_attentions", "decoder_layer", "def", "dropout_probability", "dtype", "else", "encoder_attention_mask", "encoder_hidden_states", "enumerate", "for", "forward", "hidden_states", "idx", "if", "in", "input_shape", "inputs_embeds", "intermediate", "intermediate_hidden_states", "is", "kwargs", "last_hidden_state", "layer_outputs", "layerdrop", "layernorm", "layers", "not", "object_queries", "output_attentions", "output_hidden_states", "query_position_embeddings", "r", "rand", "return", "return_dict", "self", "size", "stack", "tgt_len", "torch", "training", "tuple", "use_return_dict", "v"], "table_transformer/modeling_table_transformer.py:TableTransformerModel.__init__": ["Conv2d", "Embedding", "ModelConfig", "ModelConvEncoder", "ModelConvModel", "ModelDecoder", "ModelEncoder", "ModelModel", "ModelPreTrainedModel", "__init__", "backbone", "build_position_encoding", "class", "config", "d_model", "decoder", "def", "encoder", "input_projection", "intermediate_channel_sizes", "kernel_size", "nn", "num_queries", "object_queries", "post_init", "query_position_embeddings", "self", "super"], "table_transformer/modeling_table_transformer.py:TableTransformerModel.freeze_backbone": ["False", "ModelModel", "ModelPreTrainedModel", "backbone", "class", "conv_encoder", "def", "for", "freeze_backbone", "in", "model", "name", "named_parameters", "param", "requires_grad_", "self"], "table_transformer/modeling_table_transformer.py:TableTransformerModel.unfreeze_backbone": ["ModelModel", "ModelPreTrainedModel", "True", "backbone", "class", "conv_encoder", "def", "for", "in", "model", "name", "named_parameters", "param", "requires_grad_", "self", "unfreeze_backbone"], "table_transformer/modeling_table_transformer.py:TableTransformerModel.forward": ["Backbone", "BaseModelOutput", "FloatTensor", "ModelModel", "ModelModelOutput", "ModelPreTrainedModel", "None", "ValueError", "and", "attention_mask", "attentions", "backbone", "batch_size", "bool", "class", "config", "cross_attentions", "decoder", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_inputs_embeds", "decoder_outputs", "def", "device", "does", "downsampled", "elif", "else", "encoder", "encoder_attention_mask", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "feature_map", "features", "flatten", "flattened_features", "flattened_mask", "forward", "height", "hidden_states", "if", "input_projection", "inputs_embeds", "intermediate_hidden_states", "is", "isinstance", "kwargs", "last_hidden_state", "len", "mask", "not", "num_channels", "object_queries", "ones", "output_attentions", "output_hidden_states", "permute", "pixel", "pixel_mask", "pixel_values", "position_embeddings_list", "projected_feature_map", "queries", "query_position_embeddings", "r", "raise", "repeat", "return", "return_dict", "self", "shape", "torch", "tuple", "unsqueeze", "use_return_dict", "weight", "width", "zeros_like"], "table_transformer/modeling_table_transformer.py:TableTransformerForObjectDetection.__init__": ["Linear", "ModelConfig", "ModelForObjectDetection", "ModelMLPPredictionHead", "ModelModel", "ModelPreTrainedModel", "__init__", "bbox_predictor", "class", "class_labels_classifier", "config", "d_model", "def", "hidden_dim", "input_dim", "model", "nn", "num_labels", "num_layers", "output_dim", "post_init", "self", "super"], "table_transformer/modeling_table_transformer.py:TableTransformerForObjectDetection.forward": ["FloatTensor", "ModelForObjectDetection", "ModelObjectDetectionOutput", "ModelPreTrainedModel", "None", "auxiliary_loss", "auxiliary_outputs", "bbox_predictor", "bool", "class", "class_labels_classifier", "config", "cross_attentions", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_inputs_embeds", "def", "device", "dict", "else", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "forward", "if", "inputs_embeds", "intermediate", "intermediate_hidden_states", "is", "kwargs", "labels", "last_hidden_state", "list", "logits", "loss", "loss_dict", "loss_function", "model", "not", "output", "output_attentions", "output_hidden_states", "outputs", "outputs_class", "outputs_coord", "pixel_mask", "pixel_values", "pred_boxes", "r", "return", "return_dict", "self", "sequence_output", "sigmoid", "torch", "tuple", "use_return_dict"], "table_transformer/modeling_table_transformer.py:TableTransformerMLPPredictionHead.__init__": ["Linear", "ModelMLPPredictionHead", "Module", "ModuleList", "__init__", "class", "def", "for", "h", "hidden_dim", "in", "input_dim", "k", "layers", "n", "nn", "num_layers", "output_dim", "self", "super", "zip"], "table_transformer/modeling_table_transformer.py:TableTransformerMLPPredictionHead.forward": ["ModelMLPPredictionHead", "Module", "class", "def", "else", "enumerate", "for", "forward", "functional", "i", "if", "in", "layer", "layers", "nn", "num_layers", "relu", "return", "self", "x"], "tapas/modeling_tapas.py:TapasEmbeddings.__init__": ["Dropout", "Embedding", "LayerNorm", "ModelEmbeddings", "Module", "__init__", "class", "config", "def", "dropout", "enumerate", "eps", "f", "for", "hidden_dropout_prob", "hidden_size", "i", "in", "layer_norm_eps", "len", "max_position_embeddings", "name", "nn", "number_of_token_type_embeddings", "pad_token_id", "padding_idx", "position_embeddings", "self", "setattr", "super", "token_type_embeddings_", "type_vocab_sizes", "vocab_size", "word_embeddings"], "tapas/modeling_tapas.py:TapasEmbeddings.forward": ["IndexMap", "LayerNorm", "ModelEmbeddings", "Module", "None", "ProductIndexMap", "arange", "as_tensor", "batch_dims", "class", "col_index", "config", "def", "device", "dropout", "dtype", "else", "embeddings", "expand", "f", "first_position", "first_position_per_segment", "for", "forward", "full_index", "gather", "getattr", "i", "if", "in", "input_ids", "input_shape", "inputs_embeds", "is", "long", "max_position_embeddings", "min", "name", "nn", "not", "number_of_token_type_embeddings", "position", "position_embeddings", "position_ids", "range", "reduce_min", "reset_position_index_per_cell", "return", "row_index", "self", "seq_length", "size", "token_type_embeddings_", "token_type_ids", "torch", "type_vocab_sizes", "unsqueeze", "word_embeddings", "zeros"], "tapas/modeling_tapas.py:TapasSelfAttention.__init__": ["Dropout", "Linear", "ModelSelfAttention", "Module", "None", "The", "ValueError", "__init__", "a", "all_head_size", "and", "attention", "attention_head_size", "attention_probs_dropout_prob", "class", "config", "def", "dropout", "embedding_size", "f", "hasattr", "heads", "hidden", "hidden_size", "if", "int", "is", "is_decoder", "key", "layer_idx", "multiple", "nn", "not", "num_attention_heads", "number", "of", "query", "raise", "self", "size", "super", "the", "value"], "tapas/modeling_tapas.py:TapasSelfAttention.forward": ["EncoderDecoderCache", "False", "ModelSelfAttention", "Module", "None", "True", "_", "all_head_size", "and", "attention_head_size", "attention_mask", "attention_probs", "attention_scores", "batch_size", "cache_position", "class", "context_layer", "contiguous", "cross_attention_cache", "curr_past_key_values", "current_states", "def", "dim", "dropout", "else", "encoder_hidden_states", "forward", "functional", "get", "hidden_states", "if", "is", "is_cross_attention", "is_decoder", "is_updated", "isinstance", "key", "key_layer", "keys", "layer_idx", "layers", "math", "matmul", "new_context_layer_shape", "nn", "not", "num_attention_heads", "output_attentions", "outputs", "past_key_values", "permute", "query", "query_layer", "return", "self", "self_attention_cache", "seq_length", "shape", "size", "softmax", "sqrt", "torch", "transpose", "update", "value", "value_layer", "values", "view"], "tapas/modeling_tapas.py:TapasSelfOutput.__init__": ["Dropout", "LayerNorm", "Linear", "ModelSelfOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "eps", "hidden_dropout_prob", "hidden_size", "layer_norm_eps", "nn", "self", "super"], "tapas/modeling_tapas.py:TapasSelfOutput.forward": ["LayerNorm", "ModelSelfOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "tapas/modeling_tapas.py:TapasAttention.__init__": ["ModelAttention", "ModelSelfAttention", "ModelSelfOutput", "Module", "None", "__init__", "class", "config", "def", "layer_idx", "nn", "output", "self", "super"], "tapas/modeling_tapas.py:TapasAttention.forward": ["Cache", "False", "FloatTensor", "ModelAttention", "Module", "None", "Tensor", "attention_mask", "attention_output", "bool", "cache_position", "class", "def", "encoder_hidden_states", "forward", "hidden_states", "nn", "output", "output_attentions", "outputs", "past_key_values", "return", "self", "self_outputs", "torch", "tuple"], "tapas/modeling_tapas.py:TapasIntermediate.__init__": ["ACT2FN", "Linear", "ModelIntermediate", "Module", "__init__", "class", "config", "def", "dense", "else", "hidden_act", "hidden_size", "if", "intermediate_act_fn", "intermediate_size", "isinstance", "nn", "self", "str", "super"], "tapas/modeling_tapas.py:TapasIntermediate.forward": ["ModelIntermediate", "Module", "Tensor", "class", "def", "dense", "forward", "hidden_states", "intermediate_act_fn", "nn", "return", "self", "torch"], "tapas/modeling_tapas.py:TapasOutput.__init__": ["Dropout", "LayerNorm", "Linear", "ModelOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "eps", "hidden_dropout_prob", "hidden_size", "intermediate_size", "layer_norm_eps", "nn", "self", "super"], "tapas/modeling_tapas.py:TapasOutput.forward": ["LayerNorm", "ModelOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "tapas/modeling_tapas.py:TapasLayer.__init__": ["GradientCheckpointingLayer", "ModelAttention", "ModelIntermediate", "ModelLayer", "ModelOutput", "None", "ValueError", "__init__", "a", "add_cross_attention", "added", "as", "attention", "be", "chunk_size_feed_forward", "class", "config", "cross", "crossattention", "decoder", "def", "f", "if", "intermediate", "is", "is_decoder", "layer_idx", "model", "not", "output", "raise", "self", "seq_len_dim", "should", "super", "used"], "tapas/modeling_tapas.py:TapasLayer.forward": ["Cache", "False", "FloatTensor", "GradientCheckpointingLayer", "If", "ModelLayer", "None", "Tensor", "True", "ValueError", "add_cross_attention", "and", "apply_chunking_to_forward", "are", "attention", "attention_mask", "attention_output", "be", "bool", "by", "cache_position", "chunk_size_feed_forward", "class", "config", "cross", "cross_attention_outputs", "crossattention", "def", "encoder_attention_mask", "encoder_hidden_states", "f", "feed_forward_chunk", "forward", "has", "hasattr", "hidden_states", "if", "instantiated", "is", "is_decoder", "layer_output", "layers", "not", "output_attentions", "outputs", "passed", "past_key_values", "raise", "return", "self", "self_attention_outputs", "seq_len_dim", "setting", "to", "torch", "tuple", "with"], "tapas/modeling_tapas.py:TapasLayer.feed_forward_chunk": ["GradientCheckpointingLayer", "ModelLayer", "attention_output", "class", "def", "feed_forward_chunk", "intermediate", "intermediate_output", "layer_output", "output", "return", "self"], "tapas/modeling_tapas.py:TapasEncoder.__init__": ["False", "ModelEncoder", "ModelLayer", "Module", "ModuleList", "__init__", "class", "config", "def", "for", "gradient_checkpointing", "i", "in", "layer", "layer_idx", "nn", "num_hidden_layers", "range", "self", "super"], "tapas/modeling_tapas.py:TapasEncoder.forward": ["BaseModelOutput", "DynamicCache", "EncoderDecoderCache", "False", "ModelEncoder", "Module", "None", "True", "all_attentions", "all_hidden_states", "and", "attention_mask", "attentions", "cache_position", "class", "config", "def", "else", "encoder_attention_mask", "encoder_hidden_states", "enumerate", "for", "forward", "hidden_states", "i", "if", "in", "is", "last_hidden_state", "layer", "layer_module", "layer_outputs", "nn", "not", "output_attentions", "output_hidden_states", "past_key_values", "return", "return_dict", "self", "tuple", "use_cache", "v"], "tapas/modeling_tapas.py:TapasPooler.__init__": ["Linear", "ModelPooler", "Module", "Tanh", "__init__", "activation", "class", "config", "def", "dense", "hidden_size", "nn", "self", "super"], "tapas/modeling_tapas.py:TapasPooler.forward": ["ModelPooler", "Module", "Tensor", "activation", "class", "def", "dense", "first_token_tensor", "forward", "hidden_states", "nn", "pooled_output", "return", "self", "torch"], "tapas/modeling_tapas.py:TapasPredictionHeadTransform.__init__": ["ACT2FN", "LayerNorm", "Linear", "ModelPredictionHeadTransform", "Module", "__init__", "class", "config", "def", "dense", "else", "eps", "hidden_act", "hidden_size", "if", "isinstance", "layer_norm_eps", "nn", "self", "str", "super", "transform_act_fn"], "tapas/modeling_tapas.py:TapasPredictionHeadTransform.forward": ["LayerNorm", "ModelPredictionHeadTransform", "Module", "Tensor", "class", "def", "dense", "forward", "hidden_states", "nn", "return", "self", "torch", "transform_act_fn"], "tapas/modeling_tapas.py:TapasLMPredictionHead.__init__": ["Linear", "ModelLMPredictionHead", "ModelPredictionHeadTransform", "Module", "Parameter", "True", "__init__", "bias", "class", "config", "decoder", "def", "hidden_size", "nn", "self", "super", "torch", "transform", "vocab_size", "zeros"], "tapas/modeling_tapas.py:TapasLMPredictionHead.forward": ["ModelLMPredictionHead", "Module", "class", "decoder", "def", "forward", "hidden_states", "nn", "return", "self", "transform"], "tapas/modeling_tapas.py:TapasOnlyMLMHead.__init__": ["ModelLMPredictionHead", "ModelOnlyMLMHead", "Module", "__init__", "class", "config", "def", "nn", "predictions", "self", "super"], "tapas/modeling_tapas.py:TapasOnlyMLMHead.forward": ["ModelOnlyMLMHead", "Module", "Tensor", "class", "def", "forward", "nn", "prediction_scores", "predictions", "return", "self", "sequence_output", "torch"], "tapas/modeling_tapas.py:TapasPreTrainedModel._init_weights": ["ModelLMPredictionHead", "ModelPreTrainedModel", "PreTrainedModel", "_init_weights", "bias", "class", "def", "if", "init", "isinstance", "module", "self", "super", "zeros_"], "tapas/modeling_tapas.py:TapasModel.__init__": ["ModelEmbeddings", "ModelEncoder", "ModelModel", "ModelPooler", "ModelPreTrainedModel", "None", "True", "__init__", "add_pooling_layer", "class", "config", "def", "else", "embeddings", "encoder", "if", "pooler", "post_init", "r", "self", "super"], "tapas/modeling_tapas.py:TapasModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "return", "self", "word_embeddings"], "tapas/modeling_tapas.py:TapasModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "self", "set_input_embeddings", "value", "word_embeddings"], "tapas/modeling_tapas.py:TapasModel.forward": ["BaseModelOutputWithPooling", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "ValueError", "You", "_", "and", "at", "attention_mask", "attentions", "bool", "both", "cannot", "class", "config", "def", "device", "dtype", "either", "elif", "else", "embedding_output", "embeddings", "encoder", "encoder_attention_mask", "encoder_batch_size", "encoder_extended_attention_mask", "encoder_hidden_shape", "encoder_hidden_states", "encoder_outputs", "encoder_sequence_length", "extended_attention_mask", "forward", "get_extended_attention_mask", "have", "hidden_states", "if", "input_ids", "input_shape", "inputs_embeds", "invert_attention_mask", "is", "is_decoder", "kwargs", "last_hidden_state", "len", "long", "not", "ones", "or", "output_attentions", "output_hidden_states", "pooled_output", "pooler", "pooler_output", "position_ids", "r", "raise", "return", "return_dict", "same", "self", "sequence_output", "size", "specify", "the", "time", "to", "token_type_ids", "torch", "tuple", "type_vocab_sizes", "use_return_dict", "warn_if_padding_and_no_attention_mask", "zeros"], "tapas/modeling_tapas.py:TapasForMaskedLM.__init__": ["False", "Model", "ModelForMaskedLM", "ModelModel", "ModelOnlyMLMHead", "ModelPreTrainedModel", "__init__", "add_pooling_layer", "class", "cls", "config", "def", "post_init", "self", "super"], "tapas/modeling_tapas.py:TapasForMaskedLM.get_output_embeddings": ["ModelForMaskedLM", "ModelPreTrainedModel", "class", "cls", "decoder", "def", "get_output_embeddings", "predictions", "return", "self"], "tapas/modeling_tapas.py:TapasForMaskedLM.set_output_embeddings": ["ModelForMaskedLM", "ModelPreTrainedModel", "bias", "class", "cls", "decoder", "def", "new_embeddings", "predictions", "self", "set_output_embeddings"], "tapas/modeling_tapas.py:TapasForMaskedLM.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "MaskedLMOutput", "Model", "ModelForMaskedLM", "ModelPreTrainedModel", "None", "attention_mask", "attentions", "bool", "class", "cls", "config", "def", "else", "encoder_attention_mask", "encoder_hidden_states", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "masked_lm_loss", "not", "output", "output_attentions", "output_hidden_states", "outputs", "position_ids", "prediction_scores", "r", "return", "return_dict", "self", "sequence_output", "token_type_ids", "torch", "tuple", "use_return_dict", "view", "vocab_size"], "tapas/modeling_tapas.py:TapasForQuestionAnswering.__init__": ["Dropout", "Linear", "Model", "ModelConfig", "ModelForQuestionAnswering", "ModelModel", "ModelPreTrainedModel", "Parameter", "__init__", "aggregation_classifier", "class", "column_output_bias", "column_output_weights", "config", "def", "dropout", "else", "empty", "hidden_dropout_prob", "hidden_size", "if", "init", "init_cell_selection_weights_to_zero", "initializer_range", "nn", "normal_", "num_aggregation_labels", "output_bias", "output_weights", "post_init", "self", "std", "super", "torch", "zeros"], "tapas/modeling_tapas.py:TapasForQuestionAnswering.forward": ["Bernoulli", "EPSILON_ZERO_DIVISION", "False", "FloatTensor", "IndexMap", "LongTensor", "Make", "Model", "ModelForQuestionAnswering", "ModelPreTrainedModel", "None", "ProductIndexMap", "TableQuestionAnsweringOutput", "True", "ValueError", "You", "_", "_calculate_aggregate_mask", "_calculate_aggregation_loss", "_calculate_regression_loss", "_single_column_cell_selection_loss", "a", "aggregate", "aggregate_mask", "aggregation", "aggregation_classifier", "aggregation_labels", "aggregation_loss_weight", "allow_empty_column_selection", "and", "answer_loss", "answers", "are", "as_tensor", "assert", "attention_mask", "attentions", "average_logits_per_cell", "batch_dims", "batch_size", "bool", "calculate", "calculate_loss", "cell_index", "cell_mask", "cell_selection_preference", "class", "col_index", "column_ids", "column_logits", "column_output_bias", "column_output_weights", "column_ranks", "compute_column_logits", "compute_token_logits", "config", "def", "device", "dim", "disable_per_token_loss", "dist_per_token", "distributions", "dropout", "dtype", "elif", "else", "float", "float32", "float_answer", "forward", "gather", "have", "hidden_states", "if", "in", "index", "indices", "input_ids", "input_mask_float", "input_shape", "inputs_embeds", "inv_column_ranks", "is", "is_supervised", "kwargs", "labels", "large_answer_loss_mask", "len", "log_prob", "logits", "logits_aggregation", "logits_per_cell", "long", "loss", "mask", "max_num_columns", "max_num_rows", "mean", "min", "not", "num_aggregation_labels", "num_segments", "numeric", "numeric_relations", "numeric_values", "numeric_values_scale", "of", "ones", "ones_like", "or", "order", "output", "output_attentions", "output_bias", "output_hidden_states", "output_weights", "outputs", "pass", "per_example_additional_loss", "pooled_output", "position_ids", "positive_label_weight", "prev_labels", "r", "raise", "reduce_mean", "regression", "return", "return_dict", "row_ids", "row_index", "scale", "segment_ids", "select_one_column", "selection_loss_per_example", "selection_loss_per_token", "self", "sequence_output", "shape", "size", "specify", "sum", "sure", "table_mask", "table_mask_float", "temperature", "the", "to", "token_type_ids", "token_types", "torch", "total_loss", "tuple", "type_vocab_sizes", "use_answer_as_supervision", "use_return_dict", "values", "weight", "where", "zeros", "zeros_like"], "tapas/modeling_tapas.py:TapasForSequenceClassification.__init__": ["Dropout", "Linear", "Model", "ModelForSequenceClassification", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "classifier", "config", "def", "dropout", "hidden_dropout_prob", "hidden_size", "nn", "num_labels", "post_init", "self", "super"], "tapas/modeling_tapas.py:TapasForSequenceClassification.forward": ["BCEWithLogitsLoss", "CrossEntropyLoss", "FloatTensor", "LongTensor", "MSELoss", "Model", "ModelForSequenceClassification", "ModelPreTrainedModel", "None", "SequenceClassifierOutput", "Tensor", "and", "attention_mask", "attentions", "bool", "class", "classifier", "config", "def", "dropout", "dtype", "elif", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "kwargs", "labels", "logits", "long", "loss", "loss_fct", "multi_label_classification", "not", "num_labels", "or", "output", "output_attentions", "output_hidden_states", "outputs", "pooled_output", "position_ids", "problem_type", "r", "regression", "return", "return_dict", "self", "single_label_classification", "squeeze", "token_type_ids", "torch", "tuple", "use_return_dict", "view"], "tapas/modeling_tapas.py:IndexMap.__init__": ["ModelMap", "__init__", "as_tensor", "batch_dims", "class", "def", "device", "indices", "num_segments", "self", "torch"], "tapas/modeling_tapas.py:IndexMap.batch_shape": ["ModelMap", "batch_dims", "batch_shape", "class", "def", "indices", "return", "self", "size"], "tapas/modeling_tapas.py:ProductIndexMap.__init__": ["IndexMap", "ModelIndexMap", "ValueError", "__init__", "and", "batch_dims", "be", "class", "def", "if", "indices", "inner_index", "must", "num_segments", "outer_index", "raise", "same", "self", "super", "the"], "tapas/modeling_tapas.py:ProductIndexMap.project_outer": ["IndexMap", "ModelIndexMap", "batch_dims", "class", "def", "div", "floor", "index", "indices", "inner_index", "long", "num_segments", "outer_index", "project_outer", "return", "rounding_mode", "self", "torch", "type"], "tapas/modeling_tapas.py:ProductIndexMap.project_inner": ["IndexMap", "ModelIndexMap", "batch_dims", "class", "def", "float", "floor", "fmod", "index", "indices", "inner_index", "long", "num_segments", "project_inner", "return", "self", "torch", "type"], "tapas/modeling_tapas.py:gather": ["Model", "batch_dims", "def", "else", "expand", "if", "index", "indices", "len", "name", "return", "segmented_Model", "shape", "size", "torch", "unsqueeze", "values", "view"], "tapas/modeling_tapas.py:flatten": ["IndexMap", "Model", "_", "arange", "batch_dims", "batch_shape", "batch_size", "def", "device", "end", "for", "in", "index", "indices", "len", "list", "name", "num_segments", "offset", "prod", "range", "return", "segmented_Model", "size", "start", "tensor", "torch", "unsqueeze", "view"], "tapas/modeling_tapas.py:range_index_map": ["IndexMap", "Model_index_map", "aModel", "as_tensor", "assert", "batch_dims", "batch_shape", "cat", "cpu", "def", "device", "dim", "dtype", "else", "end", "for", "if", "in", "indices", "int", "is_tensor", "len", "list", "long", "multiples", "name", "new_shape", "new_tensor", "num_segments", "ones_like", "repeat", "return", "size", "start", "tolist", "torch", "unsqueeze", "view", "x"], "tapas/modeling_tapas.py:_segment_reduce": ["False", "_segment_reduce", "as_tensor", "batch_shape", "cat", "clone", "def", "device", "dim", "dtype", "flat_index", "flat_values", "flatten", "flattened_shape", "float", "include_self", "index", "indices", "int", "len", "long", "name", "new_shape", "num_segments", "out", "output_index", "output_values", "range_index_map", "reduce", "reshape", "return", "scatter_reduce", "segment_means", "segment_reduce_fn", "size", "src", "to", "tolist", "torch", "values", "vector_shape", "view", "zeros"], "tapas/modeling_tapas.py:reduce_sum": ["Model_sum", "_segment_Model", "def", "index", "name", "return", "segmented_Model_sum", "sum", "values"], "tapas/modeling_tapas.py:reduce_mean": ["Model_mean", "_segment_Model", "def", "index", "mean", "name", "return", "segmented_Model_mean", "values"], "tapas/modeling_tapas.py:reduce_max": ["Model_max", "_segment_Model", "amax", "def", "index", "name", "return", "segmented_Model_max", "values"], "tapas/modeling_tapas.py:reduce_min": ["Model_min", "_segment_Model", "amin", "def", "index", "name", "return", "segmented_Model_min", "values"], "tapas/modeling_tapas.py:compute_column_logits": ["CLOSE_ENOUGH_TO_LOG_ZERO", "EPSILON_ZERO_DIVISION", "Model_column_logits", "_", "allow_empty_column_selection", "as_tensor", "bs", "bsj", "cell_count", "cell_index", "cell_logits", "cell_logits_index", "cell_mask", "column_index", "column_logits", "column_output_bias", "column_output_weights", "def", "device", "dtype", "einsum", "eq", "float32", "if", "indices", "is_padding", "j", "logical_and", "not", "out_index", "project_inner", "reduce_mean", "reduce_sum", "return", "sequence_output", "token_logits", "torch"], "tapas/modeling_tapas.py:_single_column_cell_selection_loss": ["Bernoulli", "CLOSE_ENOUGH_TO_LOG_ZERO", "Categorical", "EPSILON_ZERO_DIVISION", "_", "_single_column_cell_selection_loss", "argmax", "as_tensor", "cell_dist", "cell_index", "cell_log_prob", "cell_loss", "cell_mask", "col_index", "column_dist", "column_id_for_cells", "column_label", "column_logits", "column_loss_per_example", "column_mask", "def", "device", "dim", "distributions", "dtype", "eq", "float32", "gather", "indices", "labels", "labels_index", "labels_per_cell", "labels_per_column", "log_prob", "logits", "logits_per_cell", "long", "max", "new_logits_per_cell", "no_cell_selected", "project_inner", "reduce_max", "reduce_mean", "reduce_sum", "return", "selected_column_id", "selected_column_mask", "selection_loss_per_example", "size", "sum", "token_logits", "torch", "type", "unsqueeze", "view", "where", "zeros_like"], "tapas/modeling_tapas.py:compute_token_logits": ["Model_token_logits", "bs", "bsj", "def", "einsum", "j", "logits", "output_bias", "output_weights", "return", "sequence_output", "temperature", "torch"], "tapas/modeling_tapas.py:_calculate_aggregate_mask": ["Categorical", "FloatTensor", "_calculate_aggregate_mask", "aggregate_mask", "aggregate_mask_init", "aggregation_classifier", "aggregation_ops_total_mass", "answer", "categorical", "cell_selection_preference", "def", "detach", "device", "dim", "dist_aggregation", "distributions", "dtype", "float32", "is_cell_supervision_available", "is_pred_cell_selection", "isnan", "labels", "logical_and", "logical_not", "logits", "logits_aggregation", "pooled_output", "probs", "return", "size", "sum", "to", "torch", "type", "view", "where", "zeros_like"], "tapas/modeling_tapas.py:_calculate_aggregation_loss_known": ["_calculate_aggregation_loss_known", "aggregate_mask", "aggregation_labels", "def", "dim", "dtype", "else", "float32", "functional", "if", "log_probs", "log_softmax", "logits_aggregation", "long", "nn", "num_aggregation_labels", "num_classes", "one_hot", "one_hot_labels", "per_example_aggregation_intermediate", "return", "sum", "target_aggregation", "torch", "type", "use_answer_as_supervision", "zeros_like"], "tapas/modeling_tapas.py:_calculate_aggregation_loss_unknown": ["Categorical", "_calculate_aggregation_loss_unknown", "aggregate_mask", "aggregation_ops_total_mass", "categorical", "def", "dim", "dist_aggregation", "distributions", "log", "logits", "logits_aggregation", "probs", "return", "sum", "torch"], "tapas/modeling_tapas.py:_calculate_aggregation_loss": ["_calculate_aggregation_loss", "_calculate_aggregation_loss_known", "_calculate_aggregation_loss_unknown", "aggregate_mask", "aggregation_labels", "aggregation_loss_weight", "def", "if", "logits_aggregation", "num_aggregation_labels", "per_example_aggregation_loss", "return", "use_answer_as_supervision"], "tapas/modeling_tapas.py:_calculate_expected_result": ["AverageApproximationFunction", "EPSILON_ZERO_DIVISION", "FIRST_ORDER", "Invalid", "RATIO", "RelaxedBernoulli", "RelaxedOneHotCategorical", "SECOND_ORDER", "True", "ValueError", "_calculate_expected_result", "aggregation_op_only_probs", "aggregation_temperature", "all_results", "average_approximation_function", "average_result", "avg_approximation", "cat", "config", "count_result", "def", "dim", "dist_per_cell", "distributions", "elif", "else", "ex", "expected_result", "f", "functional", "gumbel_dist", "if", "input_mask_float", "isnan", "keepdim", "logits", "logits_aggregation", "multiplier", "nn", "numeric_values", "numeric_values_masked", "numeric_values_scale", "pointwise_var", "probs", "raise", "return", "sample", "scaled_probability_per_cell", "softmax", "square", "sum", "sum_result", "temperature", "torch", "unsqueeze", "use_gumbel_for_aggregation", "use_gumbel_for_cells", "var", "where", "zeros_like"], "tapas/modeling_tapas.py:huber_loss": ["Model_loss", "abs", "def", "delta", "errors", "float", "input", "return", "target", "torch", "where"], "tapas/modeling_tapas.py:_calculate_regression_loss": ["EPSILON_ZERO_DIVISION", "None", "_calculate_expected_result", "_calculate_regression_loss", "abs", "aggregate_mask", "answer", "answer_loss_cutoff", "answer_loss_importance", "answer_masked", "config", "def", "delta", "detach", "dist_per_cell", "dtype", "else", "expected_result", "float32", "huber_loss", "huber_loss_delta", "if", "input_mask_float", "is", "isnan", "large_answer_loss_mask", "logits_aggregation", "max", "normalized_answer_masked", "normalized_expected_result", "normalizer", "numeric_values", "numeric_values_scale", "ones_like", "per_example_answer_loss", "per_example_answer_loss_scaled", "return", "torch", "use_normalized_answer_loss", "where", "zeros_like"], "textnet/modeling_textnet.py:TextNetConvLayer.__init__": ["ACT2CLS", "BatchNorm2d", "Conv2d", "False", "Identity", "ModelConfig", "ModelConvLayer", "Module", "None", "__init__", "activation", "activation_function", "batch_norm", "batch_norm_eps", "bias", "class", "config", "conv", "def", "else", "if", "is", "isinstance", "kernel_size", "nn", "not", "padding", "self", "stem_act_func", "stem_kernel_size", "stem_num_channels", "stem_out_channels", "stem_stride", "stride", "super", "tuple"], "textnet/modeling_textnet.py:TextNetConvLayer.forward": ["ModelConvLayer", "Module", "Tensor", "activation", "batch_norm", "class", "conv", "def", "forward", "hidden_states", "nn", "return", "self", "torch"], "textnet/modeling_textnet.py:TextNetRepConvLayer.__init__": ["BatchNorm2d", "Conv2d", "False", "ModelConfig", "ModelRepConvLayer", "Module", "None", "ReLU", "__init__", "activation_function", "and", "batch_norm_eps", "bias", "class", "config", "def", "else", "eps", "horizontal_batch_norm", "horizontal_conv", "horizontal_padding", "if", "in_channels", "int", "kernel_size", "main_batch_norm", "main_conv", "nn", "num_channels", "num_features", "out_channels", "padding", "rbr_identity", "self", "stride", "super", "vertical_batch_norm", "vertical_conv", "vertical_padding"], "textnet/modeling_textnet.py:TextNetRepConvLayer.forward": ["ModelRepConvLayer", "Module", "None", "Tensor", "activation_function", "class", "def", "forward", "hidden_states", "horizontal_batch_norm", "horizontal_conv", "horizontal_outputs", "id_out", "if", "is", "main_batch_norm", "main_conv", "main_outputs", "nn", "not", "rbr_identity", "return", "self", "torch", "vertical_batch_norm", "vertical_conv", "vertical_outputs"], "textnet/modeling_textnet.py:TextNetStage.__init__": ["ModelConfig", "ModelRepConvLayer", "ModelStage", "Module", "ModuleList", "__init__", "append", "class", "config", "conv_layer_kernel_sizes", "conv_layer_strides", "def", "depth", "for", "hidden_sizes", "in", "in_channels", "int", "kernel_size", "len", "nn", "num_layers", "out_channels", "self", "stage", "stage_config", "stage_in_channel_size", "stage_out_channel_size", "stride", "super", "zip"], "textnet/modeling_textnet.py:TextNetStage.forward": ["ModelStage", "Module", "block", "class", "def", "for", "forward", "hidden_state", "in", "nn", "return", "self", "stage"], "textnet/modeling_textnet.py:TextNetEncoder.__init__": ["ModelConfig", "ModelEncoder", "ModelStage", "Module", "ModuleList", "__init__", "append", "class", "config", "conv_layer_kernel_sizes", "def", "for", "in", "len", "nn", "num_stages", "range", "self", "stage_ix", "stages", "super"], "textnet/modeling_textnet.py:TextNetEncoder.forward": ["BaseModelOutputWithNoAttention", "ModelEncoder", "Module", "None", "Tensor", "append", "bool", "class", "def", "else", "for", "forward", "hidden_state", "hidden_states", "if", "in", "last_hidden_state", "nn", "not", "output", "output_hidden_states", "return", "return_dict", "self", "stage", "stages", "torch"], "textnet/modeling_textnet.py:TextNetModel.__init__": ["AdaptiveAvgPool2d", "ModelConvLayer", "ModelEncoder", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "config", "def", "encoder", "nn", "pooler", "post_init", "self", "stem", "super"], "textnet/modeling_textnet.py:TextNetModel.forward": ["Any", "BaseModelOutputWithPoolingAndNoAttention", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "bool", "class", "config", "def", "else", "encoder", "encoder_outputs", "forward", "hidden_state", "hidden_states", "if", "is", "kwargs", "last_hidden_state", "list", "not", "output", "output_hidden_states", "pixel_values", "pooled_output", "pooler", "pooler_output", "return", "return_dict", "self", "stem", "tuple", "use_return_dict"], "textnet/modeling_textnet.py:TextNetForImageClassification.__init__": ["AdaptiveAvgPool2d", "Flatten", "Identity", "Linear", "Model", "ModelForImageClassification", "ModelModel", "ModelPreTrainedModel", "ModuleList", "__init__", "avg_pool", "class", "classifier", "config", "def", "else", "fc", "flatten", "hidden_sizes", "if", "nn", "num_labels", "post_init", "self", "super"], "textnet/modeling_textnet.py:TextNetForImageClassification.forward": ["FloatTensor", "ImageClassifierOutputWithNoAttention", "LongTensor", "Model", "ModelForImageClassification", "ModelPreTrainedModel", "None", "bool", "class", "classifier", "config", "def", "else", "fc", "for", "forward", "hidden_states", "if", "in", "is", "kwargs", "labels", "last_hidden_state", "layer", "logits", "loss", "loss_function", "not", "output", "output_hidden_states", "outputs", "pixel_values", "r", "return", "return_dict", "self", "torch", "use_return_dict"], "textnet/modeling_textnet.py:TextNetBackbone.__init__": ["BackboneMixin", "Model", "ModelBackbone", "ModelModel", "ModelPreTrainedModel", "__init__", "_init_backbone", "class", "config", "def", "hidden_sizes", "num_features", "post_init", "self", "super"], "textnet/modeling_textnet.py:TextNetBackbone.forward": ["BackboneMixin", "BackboneOutput", "Model", "ModelBackbone", "ModelPreTrainedModel", "None", "Tensor", "True", "attentions", "bool", "class", "config", "def", "else", "enumerate", "feature_maps", "for", "forward", "hidden_states", "idx", "if", "in", "is", "kwargs", "not", "out_features", "output", "output_hidden_states", "outputs", "pixel_values", "r", "return", "return_dict", "self", "stage", "stage_names", "tuple", "use_return_dict"], "time_series_transformer/modeling_time_series_transformer.py:TimeSeriesFeatureEmbedder.__init__": ["Embedding", "ModelSeriesFeatureEmbedder", "Module", "ModuleList", "None", "__init__", "c", "cardinalities", "class", "d", "def", "embedders", "embedding_dims", "for", "in", "int", "len", "list", "nn", "num_features", "self", "super", "zip"], "time_series_transformer/modeling_time_series_transformer.py:TimeSeriesFeatureEmbedder.forward": ["ModelSeriesFeatureEmbedder", "Module", "Tensor", "cat", "cat_feature_slice", "cat_feature_slices", "chunk", "class", "def", "dim", "else", "embed", "embedders", "features", "for", "forward", "if", "in", "nn", "num_features", "return", "self", "squeeze", "torch", "zip"], "time_series_transformer/modeling_time_series_transformer.py:TimeSeriesStdScaler.__init__": ["ModelConfig", "ModelSeriesStdScaler", "Module", "True", "__init__", "class", "config", "def", "dim", "else", "hasattr", "if", "keepdim", "minimum_scale", "nn", "scaling_dim", "self", "super"], "time_series_transformer/modeling_time_series_transformer.py:TimeSeriesStdScaler.forward": ["ModelSeriesStdScaler", "Module", "Tensor", "clamp_min", "class", "data", "def", "denominator", "dim", "forward", "keepdim", "loc", "minimum_scale", "nn", "observed_indicator", "return", "scale", "self", "sqrt", "sum", "torch", "tuple", "variance"], "time_series_transformer/modeling_time_series_transformer.py:TimeSeriesMeanScaler.__init__": ["ModelConfig", "ModelSeriesMeanScaler", "Module", "None", "True", "__init__", "class", "config", "def", "default_scale", "dim", "else", "hasattr", "if", "keepdim", "minimum_scale", "nn", "scaling_dim", "self", "super"], "time_series_transformer/modeling_time_series_transformer.py:TimeSeriesMeanScaler.forward": ["ModelSeriesMeanScaler", "Module", "None", "Tensor", "True", "abs", "batch_observations", "batch_sum", "clamp", "class", "data", "def", "default_scale", "dim", "else", "forward", "if", "is", "keepdim", "min", "minimum_scale", "nn", "not", "num_observed", "observed_indicator", "ones_like", "return", "scale", "scaled_data", "self", "squeeze", "sum", "torch", "ts_sum", "tuple", "where", "zeros_like"], "time_series_transformer/modeling_time_series_transformer.py:TimeSeriesNOPScaler.__init__": ["ModelConfig", "ModelSeriesNOPScaler", "Module", "True", "__init__", "class", "config", "def", "dim", "else", "hasattr", "if", "keepdim", "nn", "scaling_dim", "self", "super"], "time_series_transformer/modeling_time_series_transformer.py:TimeSeriesNOPScaler.forward": ["False", "ModelSeriesNOPScaler", "Module", "None", "Tensor", "class", "data", "def", "dim", "forward", "keepdim", "loc", "mean", "nn", "observed_indicator", "ones_like", "requires_grad", "return", "scale", "self", "torch", "tuple", "zeros_like"], "time_series_transformer/modeling_time_series_transformer.py:nll": ["Distribution", "Model", "Tensor", "def", "distributions", "input", "log_prob", "return", "target", "torch"], "time_series_transformer/modeling_time_series_transformer.py:weighted_average": ["Model_average", "Model_tensor", "None", "Tensor", "clamp", "def", "dim", "else", "if", "input_tensor", "is", "mean", "min", "not", "return", "sum", "sum_weights", "torch", "weights", "where", "zeros_like"], "time_series_transformer/modeling_time_series_transformer.py:TimeSeriesSinusoidalPositionalEmbedding.__init__": ["Embedding", "ModelSeriesSinusoidalPositionalEmbedding", "None", "True", "__init__", "_freeze", "class", "def", "embedding_dim", "int", "nn", "num_positions", "padding_idx", "self", "super"], "time_series_transformer/modeling_time_series_transformer.py:TimeSeriesSinusoidalPositionalEmbedding.create_weight": ["Embedding", "False", "FloatTensor", "ModelSeriesSinusoidalPositionalEmbedding", "array", "class", "cos", "create_weight", "def", "dim", "dtype", "else", "empty", "for", "if", "in", "j", "n_pos", "nn", "np", "out", "pos", "position_enc", "power", "range", "requires_grad", "return", "self", "sentinel", "shape", "sin", "torch", "weight"], "time_series_transformer/modeling_time_series_transformer.py:TimeSeriesSinusoidalPositionalEmbedding.forward": ["Embedding", "ModelSeriesSinusoidalPositionalEmbedding", "None", "Size", "Tensor", "arange", "bsz", "class", "def", "device", "dtype", "forward", "if", "input_ids_shape", "int", "is", "long", "nn", "past_key_values_length", "position_ids", "return", "self", "seq_len", "super", "torch", "weight"], "time_series_transformer/modeling_time_series_transformer.py:TimeSeriesValueEmbedding.__init__": ["False", "Linear", "ModelSeriesValueEmbedding", "Module", "__init__", "bias", "class", "d_model", "def", "feature_size", "in_features", "nn", "out_features", "self", "super", "value_projection"], "time_series_transformer/modeling_time_series_transformer.py:TimeSeriesValueEmbedding.forward": ["ModelSeriesValueEmbedding", "Module", "class", "def", "forward", "nn", "return", "self", "value_projection", "x"], "time_series_transformer/modeling_time_series_transformer.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "contiguous", "def", "dim", "dropout", "float", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "shape", "size", "softmax", "torch", "training", "transpose", "value"], "time_series_transformer/modeling_time_series_transformer.py:TimeSeriesTransformerAttention.__init__": ["False", "Instantiating", "Linear", "ModelAttention", "ModelConfig", "Module", "None", "Please", "True", "ValueError", "__class__", "__init__", "__name__", "a", "and", "be", "bias", "bool", "by", "caching", "call", "class", "config", "creating", "decoder", "def", "divisible", "dropout", "during", "embed_dim", "errors", "f", "float", "forward", "got", "head_dim", "if", "int", "is", "is_causal", "is_decoder", "k_proj", "layer_idx", "lead", "logger", "make", "must", "nn", "not", "num_heads", "out_proj", "passing", "provide", "q_proj", "raise", "recommended", "scaling", "self", "super", "sure", "the", "this", "to", "used", "v_proj", "warning_once", "when", "will", "without"], "time_series_transformer/modeling_time_series_transformer.py:TimeSeriesTransformerAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "EncoderDecoderCache", "False", "FlashAttentionKwargs", "ModelAttention", "Module", "None", "Tensor", "True", "Unpack", "_attn_implementation", "and", "attention_interface", "attention_mask", "attn_output", "attn_weights", "bool", "bsz", "cache_position", "class", "config", "contiguous", "cross_attention_cache", "curr_past_key_values", "current_states", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "get", "head_dim", "hidden_states", "if", "is", "is_cross_attention", "is_updated", "isinstance", "k_proj", "key_states", "key_value_states", "keys", "kv_input_shape", "kwargs", "layer_idx", "layers", "nn", "not", "out_proj", "output_attentions", "past_key_values", "q_input_shape", "q_proj", "query_states", "reshape", "return", "scaling", "self", "self_attention_cache", "shape", "src_len", "tgt_len", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "values", "view"], "time_series_transformer/modeling_time_series_transformer.py:TimeSeriesTransformerEncoderLayer.__init__": ["ACT2FN", "GradientCheckpointingLayer", "LayerNorm", "Linear", "ModelAttention", "ModelConfig", "ModelEncoderLayer", "None", "__init__", "activation_dropout", "activation_fn", "activation_function", "attention_dropout", "class", "config", "d_model", "def", "dropout", "embed_dim", "encoder_attention_heads", "encoder_ffn_dim", "fc1", "fc2", "final_layer_norm", "int", "layer_idx", "nn", "num_heads", "self", "self_attn", "self_attn_layer_norm", "super"], "time_series_transformer/modeling_time_series_transformer.py:TimeSeriesTransformerEncoderLayer.forward": ["False", "FloatTensor", "GradientCheckpointingLayer", "ModelEncoderLayer", "None", "activation_dropout", "activation_fn", "and", "any", "attention_mask", "attn_weights", "bool", "clamp", "clamp_value", "class", "def", "dropout", "dtype", "fc1", "fc2", "final_layer_norm", "finfo", "float16", "forward", "functional", "hidden_states", "if", "isinf", "isnan", "max", "min", "nn", "or", "output_attentions", "outputs", "p", "residual", "return", "self", "self_attn", "self_attn_layer_norm", "torch", "training", "tuple"], "time_series_transformer/modeling_time_series_transformer.py:TimeSeriesTransformerDecoderLayer.__init__": ["ACT2FN", "GradientCheckpointingLayer", "LayerNorm", "Linear", "ModelAttention", "ModelConfig", "ModelDecoderLayer", "None", "True", "__init__", "activation_dropout", "activation_fn", "activation_function", "attention_dropout", "class", "config", "d_model", "decoder_attention_heads", "decoder_ffn_dim", "def", "dropout", "embed_dim", "encoder_attn", "encoder_attn_layer_norm", "fc1", "fc2", "final_layer_norm", "int", "is_causal", "is_decoder", "layer_idx", "nn", "num_heads", "self", "self_attn", "self_attn_layer_norm", "super"], "time_series_transformer/modeling_time_series_transformer.py:TimeSeriesTransformerDecoderLayer.forward": ["Cache", "False", "FloatTensor", "GradientCheckpointingLayer", "ModelDecoderLayer", "None", "Tensor", "True", "activation_dropout", "activation_fn", "attention_mask", "bool", "cache_position", "class", "cross_attn_weights", "def", "dropout", "encoder_attention_mask", "encoder_attn", "encoder_attn_layer_norm", "encoder_hidden_states", "fc1", "fc2", "final_layer_norm", "forward", "functional", "hidden_states", "if", "is", "key_value_states", "nn", "not", "output_attentions", "outputs", "p", "past_key_values", "residual", "return", "self", "self_attn", "self_attn_layer_norm", "self_attn_weights", "torch", "training", "tuple", "use_cache"], "time_series_transformer/modeling_time_series_transformer.py:TimeSeriesTransformerPreTrainedModel._init_weights": ["ModelPreTrainedModel", "ModelSeriesSinusoidalPositionalEmbedding", "PreTrainedModel", "_init_weights", "class", "copy_", "create_weight", "def", "if", "init", "isinstance", "module", "self", "super", "weight"], "time_series_transformer/modeling_time_series_transformer.py:TimeSeriesTransformerEncoder.__init__": ["False", "LayerNorm", "ModelConfig", "ModelEncoder", "ModelEncoderLayer", "ModelPreTrainedModel", "ModelSeriesSinusoidalPositionalEmbedding", "ModelSeriesValueEmbedding", "ModuleList", "None", "The", "ValueError", "_", "__init__", "be", "class", "config", "context_length", "d_model", "def", "dropout", "embed_positions", "encoder_layerdrop", "encoder_layers", "feature_size", "for", "gradient_checkpointing", "if", "in", "is", "layerdrop", "layernorm_embedding", "layers", "needs", "nn", "post_init", "prediction_length", "raise", "range", "self", "specified", "super", "to", "value_embedding"], "time_series_transformer/modeling_time_series_transformer.py:TimeSeriesTransformerEncoder.forward": ["BaseModelOutput", "False", "FloatTensor", "ModelEncoder", "ModelPreTrainedModel", "None", "Tensor", "True", "all_attentions", "attention_mask", "attentions", "bool", "class", "config", "create_bidirectional_mask", "def", "dropout", "dropout_probability", "else", "embed_pos", "embed_positions", "encoder_layer", "encoder_states", "enumerate", "for", "forward", "functional", "hidden_states", "idx", "if", "in", "input_embeds", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layer_outputs", "layerdrop", "layernorm_embedding", "layers", "nn", "not", "output_attentions", "output_hidden_states", "p", "r", "rand", "return", "return_dict", "self", "size", "to_drop", "torch", "training", "tuple", "use_return_dict", "v", "value_embedding"], "time_series_transformer/modeling_time_series_transformer.py:TimeSeriesTransformerDecoder.__init__": ["False", "LayerNorm", "ModelConfig", "ModelDecoder", "ModelDecoderLayer", "ModelPreTrainedModel", "ModelSeriesSinusoidalPositionalEmbedding", "ModelSeriesValueEmbedding", "ModuleList", "None", "The", "ValueError", "__init__", "be", "class", "config", "context_length", "d_model", "decoder_layerdrop", "decoder_layers", "def", "dropout", "embed_positions", "feature_size", "for", "gradient_checkpointing", "i", "if", "in", "is", "layer_idx", "layerdrop", "layernorm_embedding", "layers", "needs", "nn", "post_init", "prediction_length", "raise", "range", "self", "specified", "super", "to", "value_embedding"], "time_series_transformer/modeling_time_series_transformer.py:TimeSeriesTransformerDecoder.forward": ["BaseModelOutputWithPastAndCrossAttentions", "Cache", "DynamicCache", "EncoderDecoderCache", "False", "FloatTensor", "LongTensor", "ModelDecoder", "ModelPreTrainedModel", "None", "Setting", "Tensor", "True", "all_cross_attentions", "all_hidden_states", "all_self_attns", "and", "arange", "attention_mask", "attentions", "bool", "cache_position", "checkpointing", "class", "config", "context_length", "continue", "create_bidirectional_mask", "create_causal_mask", "cross_attentions", "decoder_layer", "def", "device", "dropout", "dropout_probability", "else", "embed_pos", "embed_positions", "encoder_attention_mask", "encoder_hidden_states", "enumerate", "for", "forward", "functional", "get_seq_length", "gradient", "gradient_checkpointing", "hidden_states", "idx", "if", "in", "incompatible", "input_embeds", "input_shape", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layer_outputs", "layerdrop", "layernorm_embedding", "layers", "logger", "nn", "not", "output_attentions", "output_hidden_states", "p", "past_key_values", "past_key_values_length", "r", "rand", "return", "return_dict", "self", "size", "torch", "training", "tuple", "use_cache", "use_return_dict", "v", "value_embedding", "warning_once", "with"], "time_series_transformer/modeling_time_series_transformer.py:TimeSeriesTransformerModel.__init__": ["ModelConfig", "ModelDecoder", "ModelEncoder", "ModelModel", "ModelPreTrainedModel", "ModelSeriesFeatureEmbedder", "ModelSeriesMeanScaler", "ModelSeriesNOPScaler", "ModelSeriesStdScaler", "True", "__init__", "cardinalities", "cardinality", "class", "config", "decoder", "def", "elif", "else", "embedder", "embedding_dimension", "embedding_dims", "encoder", "if", "is", "mean", "num_static_categorical_features", "or", "post_init", "scaler", "scaling", "self", "std", "super"], "time_series_transformer/modeling_time_series_transformer.py:TimeSeriesTransformerModel._past_length": ["ModelModel", "ModelPreTrainedModel", "_past_length", "class", "config", "context_length", "def", "int", "lags_sequence", "max", "return", "self"], "time_series_transformer/modeling_time_series_transformer.py:TimeSeriesTransformerModel.get_lagged_subsequences": ["ModelModel", "ModelPreTrainedModel", "None", "Tensor", "ValueError", "append", "begin_index", "cannot", "class", "config", "def", "dim", "else", "end_index", "f", "for", "found", "further", "get_lagged_subsequences", "go", "history", "if", "in", "indices", "int", "is", "lag", "lag_index", "lagged_values", "lags", "lags_sequence", "length", "max", "only", "raise", "return", "self", "sequence", "sequence_length", "shape", "shift", "stack", "subsequences_length", "than", "torch", "while"], "time_series_transformer/modeling_time_series_transformer.py:TimeSeriesTransformerModel.create_network_inputs": ["Model", "ModelModel", "ModelPreTrainedModel", "Model_feat", "None", "Tensor", "ValueError", "_", "_past_length", "abs", "and", "cat", "class", "config", "context", "context_length", "create_network_inputs", "def", "dim", "does", "else", "embedded_cat", "embedder", "expand", "expanded_static_feat", "f", "feature", "features", "future_Model_features", "future_values", "get_lagged_subsequences", "if", "input", "inputs", "is", "lagged_sequence", "lags_shape", "length", "lengths", "loc", "log", "log1p", "log_abs_loc", "log_scale", "match", "ndim", "not", "observed_context", "ones_like", "past_Model_features", "past_observed_mask", "past_values", "prediction_length", "raise", "reshape", "reshaped_lagged_sequence", "return", "scale", "scaler", "self", "sequence", "shape", "squeeze", "squeezed_loc", "squeezed_scale", "static_categorical_features", "static_feat", "static_real_features", "subsequences_length", "torch", "transformer_inputs", "unsqueeze"], "time_series_transformer/modeling_time_series_transformer.py:TimeSeriesTransformerModel.forward": ["BaseModelOutput", "Cache", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Seq2SeqTSModelOutput", "Tensor", "_", "and", "attention_mask", "attentions", "bool", "bsz", "cache_position", "class", "config", "context_length", "create_network_inputs", "cross_attentions", "dec_input", "decoder", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_outputs", "def", "device", "dim", "dtype", "elif", "else", "enc_input", "encoder", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "forward", "future_Model_features", "future_values", "hidden_states", "if", "inputs_embeds", "is", "isinstance", "kwargs", "last_hidden_state", "len", "list", "loc", "not", "output_attentions", "output_hidden_states", "past_Model_features", "past_key_values", "past_observed_mask", "past_values", "r", "return", "return_dict", "scale", "self", "shape", "size", "static_categorical_features", "static_feat", "static_features", "static_real_features", "torch", "transformer_inputs", "tuple", "use_cache", "use_return_dict", "zeros"], "time_series_transformer/modeling_time_series_transformer.py:TimeSeriesTransformerForPrediction.__init__": ["ModelConfig", "ModelForPrediction", "ModelModel", "ModelPreTrainedModel", "NegativeBinomialOutput", "NormalOutput", "StudentTOutput", "Unknown", "ValueError", "__init__", "class", "config", "d_model", "def", "dim", "distribution", "distribution_output", "elif", "else", "event_shape", "f", "function", "get_parameter_projection", "if", "input_size", "loss", "model", "negative_binomial", "nll", "normal", "output", "parameter_projection", "post_init", "raise", "self", "student_t", "super", "target_shape"], "time_series_transformer/modeling_time_series_transformer.py:TimeSeriesTransformerForPrediction.output_params": ["ModelForPrediction", "ModelPreTrainedModel", "class", "dec_output", "def", "output_params", "parameter_projection", "return", "self"], "time_series_transformer/modeling_time_series_transformer.py:TimeSeriesTransformerForPrediction.output_distribution": ["Distribution", "ModelForPrediction", "ModelPreTrainedModel", "None", "class", "def", "distribution", "distribution_output", "distributions", "for", "if", "in", "is", "loc", "not", "output_distribution", "p", "params", "return", "scale", "self", "sliced_params", "torch", "trailing_n"], "time_series_transformer/modeling_time_series_transformer.py:TimeSeriesTransformerForPrediction.forward": ["Cache", "False", "FloatTensor", "LongTensor", "ModelForPrediction", "ModelPreTrainedModel", "None", "Seq2SeqTSModelOutput", "Seq2SeqTSPredictionOutput", "Tensor", "_", "bool", "cache_position", "class", "config", "cross_attentions", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "def", "dim", "distribution", "else", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "forward", "future_Model_features", "future_observed_mask", "future_values", "if", "is", "keepdim", "kwargs", "len", "list", "loc", "loss", "loss_weights", "min", "model", "not", "ones_like", "output_attentions", "output_distribution", "output_hidden_states", "output_params", "outputs", "params", "past_Model_features", "past_key_values", "past_observed_mask", "past_values", "prediction_loss", "r", "return", "return_dict", "scale", "self", "static_categorical_features", "static_features", "static_real_features", "target_shape", "torch", "tuple", "use_cache", "use_return_dict", "weighted_average", "weights"], "time_series_transformer/modeling_time_series_transformer.py:TimeSeriesTransformerForPrediction.generate": ["ModelForPrediction", "ModelPreTrainedModel", "None", "SampleTSPredictionOutput", "Tensor", "True", "append", "bool", "cat", "class", "concat_future_samples", "config", "dec_last_hidden", "dec_output", "decoder", "decoder_input", "def", "dim", "distr", "enc_last_hidden", "encoder_hidden_states", "encoder_last_hidden_state", "expand", "expanded_static_feat", "features", "for", "future_Model_features", "future_samples", "future_values", "generate", "get_decoder", "get_lagged_subsequences", "in", "inputs_embeds", "k", "lagged_sequence", "lags_shape", "last_hidden_state", "loc", "model", "next_sample", "num_parallel_samples", "output_attentions", "output_distribution", "output_hidden_states", "outputs", "parameter_projection", "params", "past_Model_features", "past_observed_mask", "past_values", "prediction_length", "r", "range", "repeat_interleave", "repeated_enc_last_hidden", "repeated_features", "repeated_loc", "repeated_past_values", "repeated_scale", "repeats", "reshape", "reshaped_lagged_sequence", "return", "return_dict", "sample", "scale", "self", "sequence", "sequences", "shape", "shift", "static_categorical_features", "static_feat", "static_features", "static_real_features", "subsequences_length", "target_shape", "torch", "unsqueeze", "use_cache"], "timesfm/modeling_timesfm.py:TimesFmMLP.__init__": ["LayerNorm", "Linear", "ModelConfig", "ModelMLP", "Module", "__init__", "class", "config", "def", "down_proj", "eps", "gate_proj", "hidden_size", "intermediate_size", "layer_norm", "nn", "normalized_shape", "self", "super"], "timesfm/modeling_timesfm.py:TimesFmMLP.forward": ["F", "ModelMLP", "Module", "None", "class", "def", "down_proj", "forward", "gate", "gate_inp", "gate_proj", "if", "is", "layer_norm", "nn", "not", "outputs", "paddings", "relu", "return", "self", "x"], "timesfm/modeling_timesfm.py:TimesFmResidualBlock.__init__": ["Linear", "ModelResidualBlock", "Module", "SiLU", "__init__", "activation", "class", "def", "hidden_dims", "input_dims", "input_layer", "nn", "output_dims", "output_layer", "residual_layer", "self", "super"], "timesfm/modeling_timesfm.py:TimesFmResidualBlock.forward": ["ModelResidualBlock", "Module", "activation", "class", "def", "forward", "hidden", "input_layer", "nn", "output", "output_layer", "residual", "residual_layer", "return", "self", "x"], "timesfm/modeling_timesfm.py:TimesFmRMSNorm.__init__": ["ModelRMSNorm", "Module", "Parameter", "__init__", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "timesfm/modeling_timesfm.py:TimesFmRMSNorm.forward": ["ModelRMSNorm", "Module", "True", "class", "def", "dtype", "float32", "forward", "hidden_states", "input_dtype", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "timesfm/modeling_timesfm.py:TimesFmRMSNorm.extra_repr": ["ModelRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "variance_epsilon", "weight"], "timesfm/modeling_timesfm.py:TimesFmPositionalEmbedding.__init__": ["ModelConfig", "ModelPositionalEmbedding", "Module", "__init__", "arange", "class", "config", "def", "dtype", "embedding_dims", "exp", "float", "float32", "hidden_size", "inv_Modelcales", "log", "log_Modelcale_increment", "math", "max", "max_Modelcale", "min_Modelcale", "nn", "num_Modelcales", "register_buffer", "self", "super", "torch"], "timesfm/modeling_timesfm.py:TimesFmPositionalEmbedding.forward": ["Either", "F", "ModelPositionalEmbedding", "Module", "None", "ValueError", "and", "arange", "be", "cat", "class", "cos", "def", "device", "dim", "dimensional", "dtype", "elif", "embedding_dims", "f", "float32", "forward", "got", "if", "inv_Modelcales", "is", "must", "ndim", "nn", "or", "pad", "position", "provided", "raise", "return", "scaled_time", "self", "seq_length", "shape", "signal", "sin", "torch", "unsqueeze", "view"], "timesfm/modeling_timesfm.py:simple_eager_attention_forward": ["Model_eager_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key_states", "kwargs", "matmul", "module", "nn", "not", "p", "query_states", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value_states"], "timesfm/modeling_timesfm.py:TimesFmAttention.__init__": ["Linear", "ModelAttention", "ModelConfig", "Module", "Parameter", "True", "__init__", "attention_dropout", "class", "config", "def", "empty", "head_dim", "hidden_size", "int", "is_causal", "k_proj", "kv_size", "layer_idx", "nn", "num_attention_heads", "num_heads", "o_proj", "q_proj", "q_size", "scaling", "self", "super", "torch", "v_proj"], "timesfm/modeling_timesfm.py:TimesFmAttention._scale_query": ["F", "ModelAttention", "Module", "None", "Tensor", "_scale_query", "class", "def", "head_dim", "math", "mul", "nn", "query", "return", "scale", "scaling", "self", "softplus", "sqrt", "torch"], "timesfm/modeling_timesfm.py:TimesFmAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "FlashAttentionKwargs", "ModelAttention", "Module", "None", "Tensor", "Unpack", "_attn_implementation", "_scale_query", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "class", "config", "contiguous", "def", "dropout", "eager", "else", "forward", "head_dim", "hidden_shape", "hidden_states", "if", "input_shape", "k_proj", "key_states", "kwargs", "nn", "not", "o_proj", "q_proj", "query_states", "reshape", "return", "scaling", "self", "shape", "simple_eager_attention_forward", "torch", "training", "transpose", "tuple", "v_proj", "value_states", "view"], "timesfm/modeling_timesfm.py:TimesFmDecoderLayer.__init__": ["ModelAttention", "ModelConfig", "ModelDecoderLayer", "ModelMLP", "ModelRMSNorm", "Module", "__init__", "class", "config", "def", "eps", "hidden_size", "input_layernorm", "int", "layer_idx", "mlp", "nn", "rms_norm_eps", "self", "self_attn", "super"], "timesfm/modeling_timesfm.py:TimesFmDecoderLayer.forward": ["False", "ModelDecoderLayer", "Module", "None", "Tensor", "attention_mask", "bool", "class", "def", "forward", "hidden_states", "input_layernorm", "mlp", "nn", "output_attentions", "paddings", "residual", "return", "scores", "self", "self_attn", "torch", "tuple"], "timesfm/modeling_timesfm.py:TimesFmPreTrainedModel._init_weights": ["ModelAttention", "ModelPositionalEmbedding", "ModelPreTrainedModel", "PreTrainedModel", "_init_weights", "arange", "class", "copy_", "def", "dtype", "elif", "embedding_dims", "exp", "float", "float32", "if", "init", "inv_Modelcales", "isinstance", "log", "log_Modelcale_increment", "math", "max", "max_Modelcale", "min_Modelcale", "module", "num_Modelcales", "ones_", "scaling", "self", "super", "torch"], "timesfm/modeling_timesfm.py:TimesFmModel.__init__": ["Embedding", "ModelConfig", "ModelDecoderLayer", "ModelModel", "ModelPositionalEmbedding", "ModelPreTrainedModel", "ModelResidualBlock", "ModuleList", "__init__", "class", "config", "def", "embedding_dim", "for", "freq_emb", "freq_size", "hidden_dims", "hidden_size", "if", "in", "input_dims", "input_ff_layer", "intermediate_size", "layer_idx", "layers", "nn", "num_embeddings", "num_hidden_layers", "output_dims", "patch_length", "position_emb", "post_init", "range", "self", "super", "use_positional_embedding"], "timesfm/modeling_timesfm.py:TimesFmModel._forward_transform": ["ModelModel", "ModelPreTrainedModel", "None", "Tensor", "_Model_masked_mean_std", "_forward_transform", "abs", "clamp", "class", "config", "def", "device", "dtype", "inputs", "min", "mu", "outputs", "pad_val", "patched_pads", "return", "self", "sigma", "tensor", "tolerance", "torch", "tuple", "where"], "timesfm/modeling_timesfm.py:TimesFmModel.forward": ["False", "LongTensor", "ModelModel", "ModelOutput", "ModelPreTrainedModel", "None", "Tensor", "True", "_Model_shift_padded_seq", "_forward_transform", "_prepare_4d_attention_mask", "abs", "all_attentions", "all_hidden_states", "append", "attention_mask", "attentions", "bool", "bsize", "cat", "class", "concat", "concat_inputs", "config", "def", "device", "dim", "dtype", "else", "f_emb", "for", "forward", "freq", "freq_emb", "hidden_states", "if", "in", "input_ff_layer", "is_causal", "kwargs", "last_hidden_state", "layer", "layers", "loc", "min", "model_input", "num_hidden_layers", "output_attentions", "output_hidden_states", "pad_val", "paddings", "past_values", "past_values_padding", "patch_length", "patched_inputs", "patched_padding", "patched_pads", "pos_emb", "position_emb", "r", "return", "scale", "scores", "self", "sequence_length", "shape", "stats", "tensor", "tolerance", "torch", "use_positional_embedding", "view", "where"], "timesfm/modeling_timesfm.py:TimesFmModel._prepare_4d_attention_mask": ["ModelModel", "ModelPreTrainedModel", "None", "Tensor", "True", "_prepare_4d_attention_mask", "attention_mask", "bool", "causal_mask", "class", "def", "device", "diagonal", "dtype", "else", "finfo", "if", "iinfo", "int", "is", "is_causal", "is_floating_point", "min", "min_value", "minimum", "not", "ones", "return", "sequence_length", "shape", "torch", "triu", "view"], "timesfm/modeling_timesfm.py:TimesFmModel._timesfm_masked_mean_std": ["ModelModel", "ModelPreTrainedModel", "Tensor", "_Model_masked_mean_std", "_get_patch_index", "arange", "argmax", "arr", "bidxs", "clamp", "class", "def", "dim", "indices", "inputs", "int32", "mask", "masked_centered_arr", "masked_mean", "masked_std", "masked_sum", "masked_var", "min", "num_valid_elements", "pad", "pad_sum", "padding", "patch_indices", "return", "row_sum", "shape", "sqrt", "sum", "to", "torch", "tuple", "unsqueeze", "where"], "timesfm/modeling_timesfm.py:TimesFmModel._timesfm_shift_padded_seq": ["BoolTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "_Model_shift_padded_seq", "any", "arange", "argmax", "batch_size", "class", "def", "device", "dim", "expand", "feature_dim", "gather", "idx_range", "indices", "int32", "mask", "new_mask", "num_seq", "return", "seq", "shape", "shifted_idx", "shifted_seq", "to", "torch", "view"], "timesfm/modeling_timesfm.py:TimesFmModelForPrediction.__init__": ["ModelConfig", "ModelModel", "ModelModelForPrediction", "ModelPreTrainedModel", "ModelResidualBlock", "__init__", "class", "config", "context_len", "context_length", "decoder", "def", "hidden_dims", "hidden_size", "horizon_ff_layer", "horizon_len", "horizon_length", "input_dims", "intermediate_size", "len", "output_dims", "post_init", "quantiles", "self", "super"], "timesfm/modeling_timesfm.py:TimesFmModelForPrediction._preprocess": ["ModelModelForPrediction", "ModelPreTrainedModel", "Sequence", "Tensor", "_preprocess", "append", "cat", "class", "context_len", "def", "device", "dim", "dtype", "elif", "enumerate", "for", "freq", "horizon_len", "i", "if", "in", "inp_freq", "input_len", "input_padding", "input_ts", "inputs", "int", "int32", "num_front_pad", "ones", "padding", "reshape", "return", "self", "shape", "stack", "tensor", "torch", "ts", "tuple", "zeros"], "timesfm/modeling_timesfm.py:TimesFmModelForPrediction._postprocess_output": ["ModelModelForPrediction", "ModelPreTrainedModel", "None", "Tensor", "_", "_postprocess_output", "b", "class", "config", "def", "horizon_ff_layer", "horizon_length", "len", "model_output", "mu", "n", "output_ts", "quantiles", "return", "self", "shape", "sigma", "stats", "torch", "tuple", "view"], "timesfm/modeling_timesfm.py:TimesFmModelForPrediction._quantile_loss": ["ModelModelForPrediction", "ModelPreTrainedModel", "Tensor", "_quantile_loss", "append", "class", "config", "def", "enumerate", "errors", "for", "i", "in", "loss", "losses", "max", "mean", "predictions", "q", "quantiles", "return", "self", "stack", "targets", "torch"], "timesfm/modeling_timesfm.py:TimesFmModelForPrediction.forward": ["Default", "F", "False", "Length", "ModelModelForPrediction", "ModelOutputForPrediction", "ModelPreTrainedModel", "No", "None", "Sequence", "Tensor", "ValueError", "_Model_moving_average", "_postprocess_output", "_preprocess", "_quantile_loss", "and", "append", "attentions", "axis", "bool", "class", "concatenate", "config", "context_len", "current_padding", "decoder", "decoder_output", "def", "device", "else", "enumerate", "extend", "f", "fcontext_len", "final_out", "for", "forecast_context_len", "forward", "fprop_outputs", "freq", "frequency", "full_outputs", "full_predictions", "future_values", "hidden_states", "high", "horizon_len", "horizon_length", "i", "if", "in", "info", "inp_freq", "inp_min", "input", "input_padding", "input_ts", "inputs", "int", "is", "kwargs", "last_hidden_state", "len", "length", "loc", "logger", "loss", "match", "maximum", "mean_outputs", "mean_predictions", "min", "mse_loss", "must", "new_freqs", "new_full_ts", "new_inputs", "new_ts", "not", "num_decode_patches", "of", "output_attentions", "output_hidden_states", "output_patch_len", "paddings", "past_values", "past_values_padding", "patch_length", "provided", "quantile_loss", "r", "raise", "range", "reshape", "return", "return_forecast_on_context", "scale", "self", "shape", "size", "stack", "step_index", "to", "torch", "truncate_negative", "ts", "via", "window_size"], "timesfm/modeling_timesfm.py:TimesFmModelForPrediction._timesfm_moving_average": ["F", "ModelModelForPrediction", "ModelPreTrainedModel", "Tensor", "_Model_moving_average", "arr", "arr_padded", "class", "constant", "conv1d", "def", "device", "dtype", "int", "kernel", "list", "ones", "pad", "return", "smoothed_arr", "squeeze", "torch", "view", "window_size"], "timesformer/modeling_timesformer.py:TimesformerPatchEmbeddings.__init__": ["Conv2d", "Iterable", "ModelPatchEmbeddings", "Module", "__init__", "abc", "class", "collections", "config", "def", "else", "hidden_size", "if", "image_size", "isinstance", "kernel_size", "nn", "num_channels", "num_patches", "patch_size", "projection", "self", "stride", "super"], "timesformer/modeling_timesformer.py:TimesformerPatchEmbeddings.forward": ["ModelPatchEmbeddings", "Module", "batch_size", "class", "def", "embeddings", "flatten", "forward", "height", "nn", "num_channels", "num_frames", "patch_width", "pixel_values", "projection", "reshape", "return", "self", "shape", "size", "transpose", "width"], "timesformer/modeling_timesformer.py:TimesformerEmbeddings.__init__": ["Dropout", "ModelEmbeddings", "ModelPatchEmbeddings", "Module", "Parameter", "__init__", "attention_type", "class", "cls_token", "config", "def", "drop_rate", "embed_dim", "hidden_dropout_prob", "hidden_size", "if", "nn", "num_frames", "num_patches", "p", "patch_embeddings", "pos_drop", "position_embeddings", "self", "space_only", "super", "time_drop", "time_embeddings", "torch", "zeros"], "timesformer/modeling_timesformer.py:TimesformerEmbeddings.forward": ["ModelEmbeddings", "Module", "_", "attention_type", "batch_size", "cat", "class", "cls_pos_embed", "cls_token", "cls_tokens", "def", "dim", "else", "embeddings", "expand", "flatten", "forward", "functional", "if", "int", "interpolate", "mode", "nearest", "new_pos_embed", "new_time_embeddings", "nn", "num_frames", "other_pos_embed", "patch_embeddings", "patch_height", "patch_num", "patch_width", "permute", "pixel_values", "pos_drop", "position_embeddings", "reshape", "return", "self", "shape", "size", "space_only", "time_drop", "time_embeddings", "torch", "transpose", "unsqueeze", "view"], "timesformer/modeling_timesformer.py:drop_path": ["False", "Model_path", "Model_prob", "Tensor", "bool", "def", "device", "div", "dtype", "float", "floor_", "if", "input", "keep_prob", "ndim", "not", "or", "output", "rand", "random_tensor", "return", "shape", "torch", "training"], "timesformer/modeling_timesformer.py:TimeSformerDropPath.__init__": ["ModelDropPath", "Module", "None", "__init__", "class", "def", "drop_prob", "float", "nn", "self", "super"], "timesformer/modeling_timesformer.py:TimeSformerDropPath.forward": ["ModelDropPath", "Module", "Tensor", "class", "def", "drop_path", "drop_prob", "forward", "hidden_states", "nn", "return", "self", "torch", "training"], "timesformer/modeling_timesformer.py:TimeSformerDropPath.extra_repr": ["ModelDropPath", "Module", "class", "def", "drop_prob", "extra_repr", "f", "nn", "p", "return", "self", "str"], "timesformer/modeling_timesformer.py:TimesformerSelfAttention.__init__": ["Dropout", "Linear", "ModelConfig", "ModelSelfAttention", "Module", "__init__", "attention_dropout_prob", "attention_probs_dropout_prob", "attn_drop", "bias", "class", "config", "def", "head_dim", "hidden_size", "nn", "num_attention_heads", "num_heads", "qkv", "qkv_bias", "scale", "self", "super"], "timesformer/modeling_timesformer.py:TimesformerSelfAttention.forward": ["False", "ModelSelfAttention", "Module", "attention_probs", "attn_drop", "batch_size", "bool", "class", "context_layer", "def", "dim", "else", "forward", "hidden_size", "hidden_states", "if", "key", "nn", "num_channels", "num_heads", "output_attentions", "outputs", "permute", "qkv", "query", "reshape", "return", "scale", "self", "shape", "softmax", "transpose", "value"], "timesformer/modeling_timesformer.py:TimesformerSelfOutput.__init__": ["Dropout", "Linear", "ModelConfig", "ModelSelfOutput", "Module", "None", "__init__", "class", "config", "def", "dense", "dropout", "hidden_dropout_prob", "hidden_size", "nn", "self", "super"], "timesformer/modeling_timesformer.py:TimesformerSelfOutput.forward": ["ModelSelfOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "nn", "return", "self", "torch"], "timesformer/modeling_timesformer.py:TimeSformerAttention.__init__": ["ModelAttention", "ModelConfig", "ModelSelfAttention", "ModelSelfOutput", "Module", "None", "__init__", "attention", "class", "config", "def", "nn", "output", "self", "super"], "timesformer/modeling_timesformer.py:TimeSformerAttention.forward": ["False", "ModelAttention", "Module", "Tensor", "attention", "attention_output", "bool", "class", "def", "forward", "hidden_states", "nn", "output", "output_attentions", "outputs", "return", "self", "self_outputs", "torch", "tuple"], "timesformer/modeling_timesformer.py:TimesformerIntermediate.__init__": ["ACT2FN", "Dropout", "Linear", "ModelConfig", "ModelIntermediate", "Module", "None", "__init__", "class", "config", "def", "dense", "dropout", "else", "hidden_act", "hidden_dropout_prob", "hidden_size", "if", "intermediate_act_fn", "intermediate_size", "isinstance", "nn", "self", "str", "super"], "timesformer/modeling_timesformer.py:TimesformerIntermediate.forward": ["ModelIntermediate", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "intermediate_act_fn", "nn", "return", "self", "torch"], "timesformer/modeling_timesformer.py:TimesformerOutput.__init__": ["Dropout", "Linear", "ModelConfig", "ModelOutput", "Module", "None", "__init__", "class", "config", "def", "dense", "dropout", "hidden_dropout_prob", "hidden_size", "intermediate_size", "nn", "self", "super"], "timesformer/modeling_timesformer.py:TimesformerOutput.forward": ["ModelOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "nn", "return", "self", "torch"], "timesformer/modeling_timesformer.py:TimesformerLayer.__init__": ["GradientCheckpointingLayer", "Identity", "LayerNorm", "Linear", "ModelAttention", "ModelConfig", "ModelDropPath", "ModelIntermediate", "ModelLayer", "ModelOutput", "None", "Unknown", "ValueError", "__init__", "attention", "attention_type", "class", "config", "cpu", "def", "device", "divided_space_time", "drop_path", "drop_path_rate", "drop_path_rates", "else", "eps", "f", "for", "hidden_size", "if", "in", "int", "intermediate", "item", "joint_space_time", "layer_index", "layer_norm_eps", "layernorm_after", "layernorm_before", "linspace", "nn", "not", "num_hidden_layers", "output", "raise", "self", "space_only", "super", "temporal_attention", "temporal_dense", "temporal_layernorm", "torch", "type", "x"], "timesformer/modeling_timesformer.py:TimesformerLayer.forward": ["False", "GradientCheckpointingLayer", "ModelLayer", "Tensor", "True", "attention", "attention_output", "attention_type", "batch_size", "bool", "cat", "class", "cls_token", "config", "def", "divided_space_time", "drop_path", "elif", "forward", "hidden_states", "if", "image_size", "in", "init_cls_token", "intermediate", "joint_space_time", "layer_output", "layernorm_after", "layernorm_before", "mean", "num_frames", "num_patch_height", "num_patch_width", "num_spatial_tokens", "output", "output_attentions", "outputs", "patch_size", "permute", "repeat", "reshape", "residual", "residual_spatial", "residual_temporal", "return", "self", "self_attention_outputs", "shape", "size", "space_only", "spatial_attention_outputs", "spatial_embedding", "temporal_attention", "temporal_attention_outputs", "temporal_dense", "temporal_embedding", "temporal_layernorm", "torch", "unsqueeze"], "timesformer/modeling_timesformer.py:TimesformerEncoder.__init__": ["False", "ModelConfig", "ModelEncoder", "ModelLayer", "Module", "ModuleList", "None", "__init__", "class", "config", "def", "for", "gradient_checkpointing", "in", "ind", "layer", "nn", "num_hidden_layers", "range", "self", "super"], "timesformer/modeling_timesformer.py:TimesformerEncoder.forward": ["BaseModelOutput", "False", "ModelEncoder", "Module", "None", "Tensor", "True", "all_hidden_states", "all_self_attentions", "attentions", "bool", "class", "def", "else", "enumerate", "for", "forward", "hidden_states", "i", "if", "in", "is", "last_hidden_state", "layer", "layer_module", "layer_outputs", "nn", "not", "output_attentions", "output_hidden_states", "return", "return_dict", "self", "torch", "tuple", "v"], "timesformer/modeling_timesformer.py:TimesformerPreTrainedModel._init_weights": ["Conv2d", "LayerNorm", "Linear", "ModelEmbeddings", "ModelPreTrainedModel", "None", "PreTrainedModel", "_init_weights", "bias", "class", "cls_token", "config", "constant_", "def", "elif", "if", "init", "initializer_range", "is", "isinstance", "module", "nn", "not", "position_embeddings", "self", "std", "trunc_normal_", "weight"], "timesformer/modeling_timesformer.py:TimesformerModel.__init__": ["LayerNorm", "ModelEmbeddings", "ModelEncoder", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "config", "def", "embeddings", "encoder", "eps", "hidden_size", "layer_norm_eps", "layernorm", "nn", "post_init", "self", "super"], "timesformer/modeling_timesformer.py:TimesformerModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "patch_embeddings", "return", "self"], "timesformer/modeling_timesformer.py:TimesformerModel.forward": ["BaseModelOutput", "FloatTensor", "ModelModel", "ModelPreTrainedModel", "None", "attentions", "bool", "class", "config", "def", "else", "embedding_output", "embeddings", "encoder", "encoder_outputs", "forward", "hidden_states", "if", "is", "kwargs", "last_hidden_state", "layernorm", "not", "output_attentions", "output_hidden_states", "pixel_values", "r", "return", "return_dict", "self", "sequence_output", "torch", "tuple", "use_return_dict"], "timesformer/modeling_timesformer.py:TimesformerForVideoClassification.__init__": ["Identity", "Linear", "Model", "ModelForVideoClassification", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "classifier", "config", "def", "else", "hidden_size", "if", "nn", "num_labels", "post_init", "self", "super"], "timesformer/modeling_timesformer.py:TimesformerForVideoClassification.forward": ["BCEWithLogitsLoss", "CrossEntropyLoss", "ImageClassifierOutput", "MSELoss", "Model", "ModelForVideoClassification", "ModelPreTrainedModel", "None", "Tensor", "and", "attentions", "bool", "class", "classifier", "config", "def", "dtype", "elif", "else", "forward", "hidden_states", "if", "int", "is", "kwargs", "labels", "logits", "long", "loss", "loss_fct", "multi_label_classification", "not", "num_labels", "or", "output", "output_attentions", "output_hidden_states", "outputs", "pixel_values", "problem_type", "r", "regression", "return", "return_dict", "self", "sequence_output", "single_label_classification", "squeeze", "torch", "tuple", "use_return_dict", "view"], "timm_backbone/modeling_timm_backbone.py:TimmBackbone.__init__": ["BackboneMixin", "False", "Model", "None", "Please", "PreTrainedModel", "True", "ValueError", "__init__", "_all_layers", "_backbone", "_init_backbone", "_return_layers", "a", "and", "backbone", "by", "class", "config", "create_model", "def", "else", "enumerate", "feature_info", "features_only", "for", "freeze_batch_norm_2d", "get_dicts", "getattr", "hasattr", "i", "if", "in", "in_chans", "index", "info", "instead", "is", "it", "kwargs", "layer", "model", "module", "name", "not", "num_channels", "or", "out_features", "out_indices", "pop", "post_init", "pretrained", "raise", "requires_backends", "self", "set", "str", "super", "supported", "the", "to", "use", "use_pretrained_backbone"], "timm_backbone/modeling_timm_backbone.py:TimmBackbone.from_pretrained": ["BackboneMixin", "Model", "ModelConfig", "PreTrainedModel", "True", "ValueError", "_from_config", "backbone", "backbones", "be", "class", "cls", "config", "def", "features_only", "for", "from_pretrained", "if", "kwargs", "model_args", "must", "not", "num_channels", "out_indices", "pop", "pretrained_model_name_or_path", "raise", "requires_backends", "return", "super", "use_Model", "use_pretrained_backbone", "vision"], "timm_backbone/modeling_timm_backbone.py:TimmBackbone.freeze_batch_norm_2d": ["BackboneMixin", "Model", "PreTrainedModel", "_backbone", "class", "def", "freeze_batch_norm_2d", "model", "self", "utils"], "timm_backbone/modeling_timm_backbone.py:TimmBackbone.unfreeze_batch_norm_2d": ["BackboneMixin", "Model", "PreTrainedModel", "_backbone", "class", "def", "model", "self", "unfreeze_batch_norm_2d", "utils"], "timm_backbone/modeling_timm_backbone.py:TimmBackbone._init_weights": ["BackboneMixin", "BatchNorm2d", "Model", "None", "PreTrainedModel", "_init_weights", "and", "class", "def", "elif", "getattr", "hasattr", "if", "init", "init_non_persistent_buffers", "is", "isinstance", "module", "nn", "not", "num_batches_tracked", "ones_", "running_mean", "running_var", "self", "zeros_"], "timm_backbone/modeling_timm_backbone.py:TimmBackbone.forward": ["BackboneMixin", "BackboneOutput", "Cannot", "FloatTensor", "Model", "None", "PreTrainedModel", "Tensor", "ValueError", "_all_layers", "_backbone", "_return_layers", "at", "attentions", "backbones", "bool", "class", "config", "def", "else", "feature_maps", "for", "forward", "hidden_states", "i", "if", "in", "is", "kwargs", "moment", "not", "out_indices", "output", "output_attentions", "output_hidden_states", "pixel_values", "raise", "return", "return_dict", "return_layers", "self", "the", "torch", "tuple", "use_return_dict"], "timm_wrapper/modeling_timm_wrapper.py:_create_timm_model_with_error_handling": ["False", "ImportError", "ModelConfig", "Please", "RuntimeError", "The", "U", "Unknown", "__version__", "_create_timm_model_with_error_handling", "a", "architecture", "as", "config", "create_model", "def", "e", "except", "f", "from", "if", "in", "install", "is", "model", "model_kwargs", "more", "not", "of", "pip", "pretrained", "raise", "recent", "return", "str", "supported", "timm", "to", "try", "upgrade", "version", "with", "your"], "timm_wrapper/modeling_timm_wrapper.py:TimmWrapperPreTrainedModel.post_init": ["ModelPreTrainedModel", "PreTrainedModel", "_Model_model_supports_gradient_checkpointing", "class", "def", "post_init", "self", "super", "supports_gradient_checkpointing"], "timm_wrapper/modeling_timm_wrapper.py:TimmWrapperPreTrainedModel.load_state_dict": ["ModelPreTrainedModel", "Model_model", "PreTrainedModel", "args", "class", "def", "else", "f", "for", "if", "in", "items", "k", "kwargs", "load_state_dict", "not", "return", "self", "state_dict", "super", "v"], "timm_wrapper/modeling_timm_wrapper.py:TimmWrapperPreTrainedModel._init_weights": ["BatchNorm2d", "Linear", "ModelPreTrainedModel", "None", "PreTrainedModel", "_init_weights", "and", "bias", "class", "config", "def", "elif", "getattr", "hasattr", "if", "init", "init_non_persistent_buffers", "initializer_range", "is", "isinstance", "mean", "module", "nn", "normal_", "not", "num_batches_tracked", "ones_", "running_mean", "running_var", "self", "std", "weight", "zeros_"], "timm_wrapper/modeling_timm_wrapper.py:TimmWrapperPreTrainedModel._timm_model_supports_gradient_checkpointing": ["Exception", "False", "ModelPreTrainedModel", "Model_model", "PreTrainedModel", "True", "_Model_model_supports_gradient_checkpointing", "class", "def", "enable", "except", "hasattr", "if", "not", "return", "self", "set_grad_checkpointing", "try"], "timm_wrapper/modeling_timm_wrapper.py:TimmWrapperPreTrainedModel._set_gradient_checkpointing": ["ModelPreTrainedModel", "Model_model", "PreTrainedModel", "True", "_set_gradient_checkpointing", "args", "bool", "class", "def", "enable", "kwargs", "self", "set_grad_checkpointing"], "timm_wrapper/modeling_timm_wrapper.py:TimmWrapperPreTrainedModel.get_input_embeddings": ["ModelPreTrainedModel", "None", "PreTrainedModel", "class", "def", "get_input_embeddings", "return", "self"], "timm_wrapper/modeling_timm_wrapper.py:TimmWrapperPreTrainedModel.set_input_embeddings": ["Model", "ModelPreTrainedModel", "NotImplementedError", "PreTrainedModel", "and", "cannot", "class", "def", "do", "embeddings", "models", "not", "own", "raise", "self", "set", "set_input_embeddings", "them", "token", "value"], "timm_wrapper/modeling_timm_wrapper.py:TimmWrapperModel.__init__": ["False", "Model", "ModelConfig", "ModelModel", "ModelPreTrainedModel", "Model_model", "__init__", "_create_Model_model_with_error_handling", "class", "config", "def", "extra_init_kwargs", "features_only", "get", "model_args", "num_classes", "or", "post_init", "requires_backends", "self", "super", "vision"], "timm_wrapper/modeling_timm_wrapper.py:TimmWrapperModel.forward": ["Cannot", "FloatTensor", "Model", "ModelModel", "ModelModelOutput", "ModelPreTrainedModel", "Model_model", "None", "Please", "Tensor", "The", "To", "ValueError", "a", "and", "architecture", "available", "be", "bool", "cannot", "class", "compatible", "config", "consider", "def", "device", "different", "do_pooling", "else", "enable", "feature", "features_only", "for", "forward", "forward_features", "forward_head", "forward_intermediates", "hasattr", "hidden_states", "if", "implemented", "in", "indices", "int", "is", "isinstance", "kwargs", "last_hidden_state", "list", "method", "model", "models", "must", "not", "option", "or", "output", "output_attentions", "output_hidden_states", "outputs", "package", "pixel_values", "pooler_output", "r", "raise", "return", "return_dict", "self", "set", "the", "this", "to", "torch", "tuple", "updating", "use_return_dict", "using", "version", "versions"], "timm_wrapper/modeling_timm_wrapper.py:TimmWrapperForImageClassification.__init__": ["Model", "ModelConfig", "ModelForImageClassification", "ModelModel", "ModelPreTrainedModel", "Model_model", "Please", "ValueError", "You", "__init__", "_create_Model_model_with_error_handling", "a", "are", "checkpoint", "class", "classes", "classifier", "config", "def", "e", "extra_init_kwargs", "extraction", "feature", "for", "from", "from_pretrained", "g", "head", "if", "into", "load", "model", "model_args", "no", "num_classes", "num_labels", "number", "of", "or", "post_init", "raise", "requires_backends", "self", "specify", "super", "the", "to", "trying", "use", "vision", "weights", "with"], "timm_wrapper/modeling_timm_wrapper.py:TimmWrapperForImageClassification.forward": ["Cannot", "FloatTensor", "ImageClassifierOutput", "LongTensor", "Model", "ModelForImageClassification", "ModelPreTrainedModel", "Model_model", "None", "Please", "Tensor", "The", "To", "ValueError", "a", "and", "architecture", "available", "be", "bool", "cannot", "class", "compatible", "config", "consider", "def", "device", "different", "dtype", "else", "enable", "feature", "for", "forward", "forward_head", "forward_intermediates", "hasattr", "hidden_states", "if", "implemented", "in", "indices", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "list", "logits", "loss", "loss_function", "method", "model", "models", "must", "not", "option", "or", "output", "output_attentions", "output_hidden_states", "outputs", "package", "pixel_values", "r", "raise", "return", "return_dict", "self", "set", "the", "this", "to", "torch", "tuple", "updating", "use_return_dict", "using", "version", "versions"], "trocr/modeling_trocr.py:TrOCRLearnedPositionalEmbedding.__init__": ["Embedding", "ModelLearnedPositionalEmbedding", "__init__", "class", "def", "embedding_dim", "int", "nn", "num_embeddings", "offset", "self", "super"], "trocr/modeling_trocr.py:TrOCRLearnedPositionalEmbedding.forward": ["Embedding", "ModelLearnedPositionalEmbedding", "None", "Tensor", "arange", "bsz", "class", "def", "device", "dtype", "else", "expand", "forward", "if", "input_ids", "int", "is", "long", "nn", "offset", "past_key_values_length", "position_ids", "return", "self", "seq_len", "shape", "super", "torch", "unsqueeze", "weight"], "trocr/modeling_trocr.py:TrOCRScaledWordEmbedding.__init__": ["Embedding", "ModelScaledWordEmbedding", "None", "__init__", "class", "def", "embed_scale", "embedding_dim", "float", "int", "nn", "num_embeddings", "padding_idx", "self", "super"], "trocr/modeling_trocr.py:TrOCRScaledWordEmbedding.forward": ["Embedding", "ModelScaledWordEmbedding", "Tensor", "class", "def", "embed_scale", "forward", "input_ids", "nn", "return", "self", "super", "torch"], "trocr/modeling_trocr.py:TrOCRSinusoidalPositionalEmbedding.__init__": ["ModelSinusoidalPositionalEmbedding", "Module", "None", "__init__", "class", "def", "embedding_dim", "get_embedding", "int", "nn", "num_positions", "offset", "padding_idx", "self", "super", "weights"], "trocr/modeling_trocr.py:TrOCRSinusoidalPositionalEmbedding.get_embedding": ["ModelSinusoidalPositionalEmbedding", "Module", "None", "arange", "cat", "class", "cos", "def", "dim", "dtype", "emb", "embedding_dim", "exp", "float", "get_default_dtype", "get_embedding", "half_dim", "if", "int", "int64", "is", "log", "math", "nn", "not", "num_embeddings", "padding_idx", "return", "sin", "to", "torch", "unsqueeze", "view", "zeros"], "trocr/modeling_trocr.py:TrOCRSinusoidalPositionalEmbedding.forward": ["ModelSinusoidalPositionalEmbedding", "Module", "None", "Tensor", "bsz", "class", "create_position_ids_from_input_ids", "def", "detach", "device", "embedding_dim", "forward", "get_embedding", "if", "index_select", "input_ids", "int", "is", "max_pos", "nn", "or", "padding_idx", "past_key_values_length", "position_ids", "return", "self", "seq_len", "size", "to", "torch", "view", "weights", "x"], "trocr/modeling_trocr.py:TrOCRSinusoidalPositionalEmbedding.create_position_ids_from_input_ids": ["ModelSinusoidalPositionalEmbedding", "Module", "None", "Tensor", "class", "create_position_ids_from_input_ids", "cumsum", "def", "dim", "incremental_indices", "input_ids", "int", "long", "mask", "ne", "nn", "padding_idx", "past_key_values_length", "return", "self", "torch", "type_as"], "trocr/modeling_trocr.py:TrOCRAttention.__init__": ["False", "Linear", "ModelAttention", "Module", "None", "True", "ValueError", "__init__", "and", "be", "bias", "bool", "by", "class", "config", "def", "divisible", "dropout", "else", "embed_dim", "f", "float", "got", "head_dim", "if", "int", "is", "is_cross_attention", "is_decoder", "k_proj", "kdim", "layer_idx", "must", "nn", "not", "num_heads", "out_proj", "q_proj", "raise", "scaling", "self", "super", "v_proj", "vdim"], "trocr/modeling_trocr.py:TrOCRAttention.forward": ["Attention", "Cache", "EncoderDecoderCache", "False", "ModelAttention", "Module", "None", "Tensor", "True", "ValueError", "and", "attention_mask", "attn_output", "attn_probs", "attn_weights", "attn_weights_reshaped", "be", "bmm", "bool", "bsz", "but", "cache_position", "class", "cross_attention_cache", "curr_past_key_values", "current_states", "def", "dim", "dropout", "else", "embed_dim", "f", "forward", "functional", "get", "head_dim", "hidden_states", "if", "is", "is_cross_attention", "is_updated", "isinstance", "k_proj", "key_states", "key_value_states", "keys", "layer_idx", "layers", "mask", "nn", "not", "num_heads", "of", "out_proj", "output_attentions", "p", "past_key_values", "proj_shape", "q_proj", "query_states", "raise", "reshape", "return", "scaling", "self", "self_attention_cache", "should", "size", "softmax", "src_len", "tgt_len", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "values", "view", "weights"], "trocr/modeling_trocr.py:TrOCRDecoderLayer.__init__": ["ACT2FN", "GradientCheckpointingLayer", "LayerNorm", "Linear", "ModelAttention", "ModelConfig", "ModelDecoderLayer", "None", "True", "__init__", "activation_dropout", "activation_fn", "activation_function", "attention_dropout", "class", "config", "cross_attention_hidden_size", "decoder_attention_heads", "decoder_ffn_dim", "def", "dropout", "embed_dim", "encoder_attn", "encoder_attn_layer_norm", "fc1", "fc2", "final_layer_norm", "hidden_size", "if", "is_cross_attention", "is_decoder", "kdim", "layer_idx", "nn", "num_heads", "self", "self_attn", "self_attn_layer_norm", "super", "vdim"], "trocr/modeling_trocr.py:TrOCRDecoderLayer.forward": ["Cache", "False", "GradientCheckpointingLayer", "ModelDecoderLayer", "None", "Tensor", "True", "activation_dropout", "activation_fn", "attention_mask", "bool", "cache_position", "class", "cross_attn_weights", "def", "dropout", "encoder_attention_mask", "encoder_attn", "encoder_attn_layer_norm", "encoder_hidden_states", "fc1", "fc2", "final_layer_norm", "forward", "functional", "hidden_states", "if", "is", "key_value_states", "nn", "not", "output_attentions", "outputs", "p", "past_key_values", "residual", "return", "self", "self_attn", "self_attn_layer_norm", "self_attn_weights", "torch", "training", "use_cache"], "trocr/modeling_trocr.py:TrOCRDecoder.__init__": ["False", "LayerNorm", "ModelConfig", "ModelDecoder", "ModelDecoderLayer", "ModelLearnedPositionalEmbedding", "ModelPreTrainedModel", "ModelScaledWordEmbedding", "ModelSinusoidalPositionalEmbedding", "ModuleList", "None", "__init__", "class", "config", "decoder_layerdrop", "decoder_layers", "def", "dropout", "else", "embed_positions", "embed_scale", "embed_tokens", "for", "gradient_checkpointing", "hidden_size", "i", "if", "in", "layer_idx", "layerdrop", "layernorm_embedding", "layers", "math", "max_position_embeddings", "nn", "pad_token_id", "padding_idx", "post_init", "range", "scale_embedding", "self", "sqrt", "super", "use_learned_position_embeddings", "vocab_size"], "trocr/modeling_trocr.py:TrOCRDecoder.forward": ["BaseModelOutputWithPastAndCrossAttentions", "DynamicCache", "EncoderDecoderCache", "False", "ModelDecoder", "ModelPreTrainedModel", "None", "Setting", "True", "ValueError", "You", "_prepare_4d_attention_mask", "_prepare_4d_causal_attention_mask", "all_cross_attentions", "all_hidden_states", "all_self_attns", "and", "at", "attention_mask", "attentions", "both", "cache_position", "cannot", "checkpointing", "class", "config", "continue", "cross_attentions", "decoder_input_ids", "decoder_inputs_embeds", "decoder_layer", "def", "dropout", "dropout_probability", "dtype", "either", "elif", "else", "embed_pos", "embed_positions", "embed_tokens", "encoder_attention_mask", "encoder_hidden_states", "enumerate", "for", "forward", "functional", "get_seq_length", "gradient", "gradient_checkpointing", "have", "hidden_states", "idx", "if", "in", "incompatible", "input", "input_ids", "input_shape", "inputs_embeds", "is", "is_encoder_decoder", "kwargs", "last_hidden_state", "layer_outputs", "layerdrop", "layernorm_embedding", "layers", "logger", "nn", "not", "or", "output_attentions", "output_hidden_states", "p", "past_key_values", "past_key_values_length", "r", "raise", "rand", "return", "return_dict", "same", "self", "shape", "size", "specify", "tgt_len", "the", "time", "to", "torch", "training", "tuple", "use_cache", "use_learned_position_embeddings", "use_return_dict", "v", "view", "warning_once", "with"], "trocr/modeling_trocr.py:TrOCRDecoderWrapper.__init__": ["ModelDecoder", "ModelDecoderWrapper", "ModelPreTrainedModel", "__init__", "class", "config", "decoder", "def", "post_init", "self", "super"], "trocr/modeling_trocr.py:TrOCRDecoderWrapper.forward": ["ModelDecoderWrapper", "ModelPreTrainedModel", "args", "class", "decoder", "def", "forward", "kwargs", "return", "self"], "trocr/modeling_trocr.py:TrOCRForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "ModelDecoderWrapper", "ModelForCausalLM", "ModelPreTrainedModel", "True", "__init__", "bias", "class", "config", "def", "hidden_size", "is_decoder", "is_encoder_decoder", "model", "nn", "output_projection", "post_init", "self", "super", "vocab_size"], "trocr/modeling_trocr.py:TrOCRForCausalLM.get_input_embeddings": ["GenerationMixin", "ModelForCausalLM", "ModelPreTrainedModel", "class", "decoder", "def", "embed_tokens", "get_input_embeddings", "model", "return", "self"], "trocr/modeling_trocr.py:TrOCRForCausalLM.set_input_embeddings": ["GenerationMixin", "ModelForCausalLM", "ModelPreTrainedModel", "class", "decoder", "def", "embed_tokens", "model", "self", "set_input_embeddings", "value"], "trocr/modeling_trocr.py:TrOCRForCausalLM.get_output_embeddings": ["GenerationMixin", "ModelForCausalLM", "ModelPreTrainedModel", "class", "def", "get_output_embeddings", "output_projection", "return", "self"], "trocr/modeling_trocr.py:TrOCRForCausalLM.set_output_embeddings": ["GenerationMixin", "ModelForCausalLM", "ModelPreTrainedModel", "class", "def", "new_embeddings", "output_projection", "self", "set_output_embeddings"], "trocr/modeling_trocr.py:TrOCRForCausalLM.forward": ["Cache", "CausalLMOutputWithCrossAttentions", "CrossEntropyLoss", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForCausalLM", "ModelPreTrainedModel", "None", "Tensor", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "cross_attentions", "decoder", "def", "else", "encoder_attention_mask", "encoder_hidden_states", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "model", "not", "output", "output_attentions", "output_hidden_states", "output_projection", "outputs", "past_key_values", "r", "return", "return_dict", "self", "torch", "tuple", "use_cache", "use_return_dict", "view", "vocab_size"], "tvp/modeling_tvp.py:TvpLoss.__init__": ["Loss", "ModelLoss", "Module", "ValueError", "__init__", "class", "def", "distance", "duration", "f", "for", "if", "in", "iou", "loss", "loss_distance", "loss_duration", "loss_iou", "loss_map", "losses", "nn", "not", "raise", "self", "super", "supported"], "tvp/modeling_tvp.py:TvpLoss.loss_iou": ["ModelLoss", "Module", "candidates_end_time", "candidates_start_time", "clamp", "class", "def", "duration", "end_time", "inter", "iou", "loss_iou", "max", "min", "nn", "return", "self", "start_time", "torch", "union"], "tvp/modeling_tvp.py:TvpLoss.loss_distance": ["ModelLoss", "Module", "add", "candidates_end_time", "candidates_start_time", "clamp", "class", "def", "distance_diff", "div", "duration", "end_time", "loss_distance", "max", "mid_candidates", "mid_groundtruth", "min", "nn", "return", "self", "start_time", "torch"], "tvp/modeling_tvp.py:TvpLoss.loss_duration": ["ModelLoss", "Module", "candidates_end_time", "candidates_start_time", "clamp", "class", "def", "div", "duration", "duration_candidates", "duration_diff", "duration_groundtruth", "end_time", "loss_duration", "min", "nn", "return", "self", "square", "start_time", "sub", "torch"], "tvp/modeling_tvp.py:TvpLoss.forward": ["ModelLoss", "Module", "candidates", "candidates_end_time", "candidates_start_time", "class", "def", "duration", "end_time", "float", "for", "forward", "in", "labels", "logits", "loss", "loss_map", "losses", "losses_dict", "mul", "nn", "return", "self", "start_time", "torch", "update"], "tvp/modeling_tvp.py:TvpVisionModel.__init__": ["Backbone", "Conv2d", "False", "ModelVisionModel", "Module", "None", "ValueError", "__init__", "and", "backbone", "backbone_config", "bias", "class", "config", "def", "elif", "else", "found", "grid_encoder_conv", "groups", "hasattr", "hidden_size", "hidden_sizes", "if", "in_channels", "is", "kernel_size", "load_backbone", "nn", "not", "padding", "raise", "self", "stride", "super"], "tvp/modeling_tvp.py:TvpVisionModel.forward": ["ModelVisionModel", "Module", "True", "backbone", "batch_size", "class", "def", "feature_maps", "forward", "functional", "grid", "grid_encoder_conv", "grid_feat_outputs", "height", "inplace", "kernel_size", "max_pool2d", "new_channel", "new_height", "new_width", "nn", "num_channels", "num_frames", "permute", "pixel_values", "relu", "return", "self", "shape", "stride", "view", "width"], "tvp/modeling_tvp.py:TvpVisualInputEmbedding.__init__": ["Dropout", "Embedding", "LayerNorm", "ModelVisualInputEmbedding", "Module", "__init__", "class", "col_position_embeddings", "config", "def", "dropout", "eps", "hidden_dropout_prob", "hidden_size", "layer_norm", "layer_norm_eps", "max_grid_col_position_embeddings", "max_grid_row_position_embeddings", "max_position_embeddings", "nn", "position_embeddings", "row_position_embeddings", "self", "super", "token_type_embeddings"], "tvp/modeling_tvp.py:TvpVisualInputEmbedding.interpolate_pos_encoding": ["False", "ModelVisualInputEmbedding", "Module", "Tensor", "align_corners", "bicubic", "class", "def", "embedding", "functional", "h0", "height", "if", "int", "interpolate", "interpolate_pos_encoding", "max_grid_col_position_embeddings", "max_grid_row_position_embeddings", "mode", "nn", "permute", "return", "scale_factor", "self", "torch", "w0", "width"], "tvp/modeling_tvp.py:TvpVisualInputEmbedding.add_2d_positional_embeddings": ["False", "ModelVisualInputEmbedding", "Module", "add_2d_positional_embeddings", "and", "arange", "batch_size", "bool", "class", "col_position_embeddings", "col_position_ids", "col_shape", "def", "device", "dtype", "else", "grid", "height", "hidden_dim", "if", "interpolate_pos_encoding", "len", "long", "max_grid_col_position_embeddings", "max_grid_row_position_embeddings", "min", "nn", "or", "positional_embeddings", "return", "row_height", "row_position_embeddings", "row_position_ids", "row_shape", "row_width", "self", "shape", "torch", "view", "width"], "tvp/modeling_tvp.py:TvpVisualInputEmbedding.forward": ["False", "ModelVisualInputEmbedding", "Module", "add_2d_positional_embeddings", "batch_size", "bool", "class", "def", "device", "dropout", "dtype", "embeddings", "forward", "grid", "height", "interpolate_pos_encoding", "layer_norm", "long", "mean", "nn", "num_channels", "num_frames", "return", "self", "shape", "token_type_embeddings", "token_type_ids", "torch", "view", "visual_tokens", "visual_tokens_shape", "width", "zeros"], "tvp/modeling_tvp.py:TvpTextInputEmbeddings.__init__": ["Dropout", "Embedding", "LayerNorm", "ModelTextInputEmbeddings", "Module", "__init__", "class", "config", "def", "dropout", "eps", "hidden_dropout_prob", "hidden_size", "layer_norm", "layer_norm_eps", "max_position_embeddings", "nn", "pad_token_id", "padding_idx", "position_embeddings", "self", "super", "token_type_embeddings", "type_vocab_size", "vocab_size", "word_embeddings"], "tvp/modeling_tvp.py:TvpTextInputEmbeddings.forward": ["ModelTextInputEmbeddings", "Module", "None", "arange", "class", "def", "device", "dropout", "dtype", "else", "embeddings", "expand", "forward", "if", "input_ids", "input_shape", "inputs_embeds", "is", "layer_norm", "long", "nn", "not", "position_embeddings", "position_ids", "return", "self", "seq_length", "size", "token_type_embeddings", "token_type_ids", "torch", "unsqueeze", "word_embeddings", "zeros"], "tvp/modeling_tvp.py:TvpAttention.__init__": ["Dropout", "LayerNorm", "Linear", "ModelAttention", "Module", "The", "ValueError", "__init__", "a", "all_head_size", "and", "attention", "attention_head_size", "attention_probs_dropout_prob", "attn_dropout", "class", "config", "def", "dense", "dropout", "embedding_size", "eps", "f", "hasattr", "heads", "hidden", "hidden_dropout_prob", "hidden_size", "if", "int", "is", "key", "layer_norm", "layer_norm_eps", "multiple", "nn", "not", "num_attention_heads", "number", "of", "query", "raise", "self", "size", "super", "the", "value"], "tvp/modeling_tvp.py:TvpAttention._reshape": ["ModelAttention", "Module", "Tensor", "_reshape", "attention_head_size", "batch_size", "class", "contiguous", "def", "int", "nn", "num_attention_heads", "return", "self", "sequence_length", "tensor", "torch", "transpose", "view"], "tvp/modeling_tvp.py:TvpAttention.forward": ["ModelAttention", "Module", "None", "_reshape", "all_head_size", "attention_head_size", "attention_mask", "attention_probs", "attention_scores", "attn_dropout", "attn_output", "batch_size", "bool", "class", "contiguous", "def", "dense", "dim", "dropout", "else", "forward", "functional", "hidden_states", "if", "is", "key", "key_layer", "layer_norm", "math", "matmul", "mixed_key_layer", "mixed_query_layer", "mixed_value_layer", "nn", "not", "output_attentions", "outputs", "query", "query_layer", "reshape", "return", "self", "sequence_length", "shape", "softmax", "sqrt", "torch", "transpose", "value", "value_layer"], "tvp/modeling_tvp.py:TvpIntermediate.__init__": ["ACT2FN", "Linear", "ModelIntermediate", "Module", "__init__", "class", "config", "def", "dense", "else", "hidden_act", "hidden_size", "if", "intermediate_act_fn", "intermediate_size", "isinstance", "nn", "self", "str", "super"], "tvp/modeling_tvp.py:TvpIntermediate.forward": ["ModelIntermediate", "Module", "Tensor", "class", "def", "dense", "forward", "hidden_states", "intermediate_act_fn", "nn", "return", "self", "torch"], "tvp/modeling_tvp.py:TvpOutputLayer.__init__": ["Dropout", "LayerNorm", "Linear", "ModelOutputLayer", "Module", "__init__", "class", "config", "def", "dense", "dropout", "eps", "hidden_dropout_prob", "hidden_size", "intermediate_size", "layer_norm", "layer_norm_eps", "nn", "self", "super"], "tvp/modeling_tvp.py:TvpOutputLayer.forward": ["ModelOutputLayer", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "layer_norm", "nn", "return", "self", "torch"], "tvp/modeling_tvp.py:TvpEncodeLayer.__init__": ["GradientCheckpointingLayer", "ModelAttention", "ModelEncodeLayer", "ModelIntermediate", "ModelOutputLayer", "__init__", "attention", "class", "config", "def", "intermediate", "output", "self", "super"], "tvp/modeling_tvp.py:TvpEncodeLayer.forward": ["GradientCheckpointingLayer", "ModelEncodeLayer", "None", "attention", "attention_mask", "attention_output", "bool", "class", "def", "forward", "hidden_states", "intermediate", "intermediate_output", "layer_output", "output", "output_attentions", "outputs", "return", "self", "self_attention_outputs"], "tvp/modeling_tvp.py:TvpEncoder.__init__": ["False", "ModelEncodeLayer", "ModelEncoder", "Module", "ModuleList", "_", "__init__", "class", "config", "def", "for", "gradient_checkpointing", "in", "layer", "nn", "num_hidden_layers", "range", "self", "super"], "tvp/modeling_tvp.py:TvpEncoder.forward": ["BaseModelOutput", "ModelEncoder", "Module", "None", "all_attentions", "all_hidden_states", "attention_mask", "attentions", "bool", "class", "config", "def", "else", "enumerate", "for", "forward", "hidden_states", "i", "if", "in", "is", "last_hidden_state", "layer", "layer_module", "layer_outputs", "nn", "not", "output_attentions", "output_hidden_states", "outputs", "return", "return_dict", "self", "tuple"], "tvp/modeling_tvp.py:TvpPooler.__init__": ["Linear", "ModelPooler", "Module", "Tanh", "__init__", "activation", "class", "config", "def", "dense", "hidden_size", "nn", "self", "super"], "tvp/modeling_tvp.py:TvpPooler.forward": ["ModelPooler", "Module", "Tensor", "activation", "class", "def", "dense", "first_token_tensor", "forward", "hidden_states", "nn", "pooled_output", "return", "self", "torch"], "tvp/modeling_tvp.py:TvpPreTrainedModel._init_weights": ["Conv2d", "Embedding", "LayerNorm", "Linear", "ModelModel", "ModelPreTrainedModel", "Module", "None", "PreTrainedModel", "_init_weights", "and", "bias", "class", "config", "constant_", "def", "elif", "fan_out", "hasattr", "if", "init", "initializer_range", "is", "isinstance", "kaiming_normal_", "mean", "mode", "module", "nn", "nonlinearity", "normal_", "not", "ones_", "pad_down", "pad_left", "pad_right", "pad_up", "relu", "self", "std", "text_prompt", "weight", "zeros_"], "tvp/modeling_tvp.py:TvpFrameDownPadPrompter.__init__": ["ModelFrameDownPadPrompter", "Module", "Parameter", "ValueError", "__init__", "add", "be", "class", "config", "def", "frame_num", "if", "in", "max_img_size", "must", "nn", "not", "pad_down", "raise", "randn", "remove", "replace", "self", "super", "torch", "visual_prompt_size", "visual_prompter_apply"], "tvp/modeling_tvp.py:TvpFrameDownPadPrompter.forward": ["ModelFrameDownPadPrompter", "Module", "add", "class", "def", "device", "dtype", "forward", "if", "max_img_size", "nn", "ones", "pad_down", "pixel_values", "prompt", "remove", "return", "self", "shape", "start_point", "to", "torch", "visual_prompt_mask", "visual_prompt_size", "visual_prompter_apply", "zeros"], "tvp/modeling_tvp.py:TvpFramePadPrompter.__init__": ["ModelFramePadPrompter", "Module", "Parameter", "ValueError", "__init__", "add", "base_size", "be", "class", "config", "def", "if", "in", "max_img_size", "must", "nn", "not", "num_frames", "pad_down", "pad_left", "pad_right", "pad_up", "raise", "randn", "remove", "replace", "self", "super", "torch", "visual_prompt_size", "visual_prompter_apply"], "tvp/modeling_tvp.py:TvpFramePadPrompter.interpolate_pad_encoding": ["False", "ModelFramePadPrompter", "Module", "Tensor", "align_corners", "batch", "bicubic", "channels", "class", "def", "functional", "h0", "height", "int", "interpolate", "interpolate_pad_encoding", "max_img_size", "mode", "nn", "num_frames", "prompt", "prompt_height", "prompt_width", "reshape", "return", "scale_factor", "self", "shape", "torch", "w0", "width"], "tvp/modeling_tvp.py:TvpFramePadPrompter.forward": ["False", "Invalid", "ModelFramePadPrompter", "Module", "ValueError", "add", "base", "base_size", "bool", "cat", "class", "def", "device", "dim", "dtype", "else", "f", "forward", "height", "if", "in", "interpolate_pad_encoding", "max_img_size", "nn", "not", "num_frames", "ones", "pad_down", "pad_left", "pad_right", "pad_up", "pixel_values", "prompt", "raise", "remove", "replace", "return", "self", "shape", "size", "to", "torch", "value", "visual_prompt_mask", "visual_prompter_apply", "width", "zeros"], "tvp/modeling_tvp.py:TvpModel.__init__": ["Dropout", "ModelEncoder", "ModelModel", "ModelPooler", "ModelPreTrainedModel", "ModelTextInputEmbeddings", "ModelVisionModel", "ModelVisualInputEmbedding", "Model_PROMPTER_CLASSES_MAPPING", "Parameter", "ValueError", "__init__", "be", "class", "config", "def", "dropout", "embeddings", "encoder", "framedownpad", "framepad", "hidden_dropout_prob", "hidden_size", "if", "in", "must", "nn", "not", "pooler", "post_init", "raise", "randn", "self", "super", "text_prompt", "torch", "vision_model", "visual_embeddings", "visual_prompter", "visual_prompter_type"], "tvp/modeling_tvp.py:TvpModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "return", "self", "word_embeddings"], "tvp/modeling_tvp.py:TvpModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "self", "set_input_embeddings", "value", "word_embeddings"], "tvp/modeling_tvp.py:TvpModel.forward": ["BaseModelOutputWithPooling", "False", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "attention_mask", "attentions", "bool", "cat", "class", "config", "def", "device", "dim", "dropout", "dtype", "else", "embedding_output", "embeddings", "encoder", "encoder_outputs", "expand", "forward", "get_extended_attention_mask", "hidden_states", "if", "input_ids", "interpolate_pad_encoding", "interpolate_pos_encoding", "is", "kwargs", "last_hidden_state", "new_ones", "not", "ones", "output_attentions", "output_hidden_states", "pixel_values", "pooled_output", "pooler", "pooler_output", "pt_mask", "r", "return", "return_dict", "self", "shape", "size", "text_embedding_output", "text_prompt", "to", "torch", "tuple", "vision_model", "visual_attention_mask", "visual_embedding_output", "visual_embeddings", "visual_prompter"], "tvp/modeling_tvp.py:TvpVideoGroundingHead.__init__": ["Linear", "ModelVideoGroundingHead", "Module", "ReLU", "Sigmoid", "__init__", "activation_0", "activation_1", "class", "config", "def", "hidden_size", "layer_0", "layer_1", "nn", "self", "super"], "tvp/modeling_tvp.py:TvpVideoGroundingHead.forward": ["ModelVideoGroundingHead", "Module", "activation_0", "activation_1", "class", "def", "forward", "layer_0", "layer_1", "logits", "nn", "pooler_output", "return", "self"], "tvp/modeling_tvp.py:TvpForVideoGrounding.__init__": ["ModelForVideoGrounding", "ModelModel", "ModelPreTrainedModel", "ModelVideoGroundingHead", "__init__", "class", "config", "def", "model", "post_init", "self", "super", "video_grounding_head"], "tvp/modeling_tvp.py:TvpForVideoGrounding.forward": ["False", "FloatTensor", "LongTensor", "ModelForVideoGrounding", "ModelLoss", "ModelPreTrainedModel", "ModelVideoGroundingOutput", "None", "Tensor", "attention_mask", "attentions", "bool", "class", "config", "criterion", "def", "device", "distance", "distance_loss_weight", "duration", "duration_loss_weight", "else", "forward", "hidden_states", "if", "input_ids", "interpolate_pos_encoding", "iou", "is", "kwargs", "labels", "logits", "loss", "loss_dict", "model", "not", "output_attentions", "output_hidden_states", "outputs", "pixel_values", "pooler_output", "r", "return", "return_dict", "self", "to", "torch", "tuple", "video_grounding_head"], "udop/modeling_udop.py:get_visual_bbox": ["Model_visual_bbox", "arange", "def", "dim", "image_feature_pool_shape", "image_size", "patch_size", "repeat", "return", "stack", "torch", "transpose", "view", "visual_bbox_input", "visual_bbox_x", "visual_bbox_y"], "udop/modeling_udop.py:pad_sequence": ["Model_sequence", "Model_value", "Tensor", "cat", "def", "dim", "else", "if", "isinstance", "len", "m", "n", "ret", "return", "seq", "shape", "stack", "target_len", "tensor", "to", "torch"], "udop/modeling_udop.py:combine_image_text_embeddings": ["False", "Model_image_text_embeddings", "None", "True", "arange", "attention_mask", "bbox", "bool", "cat", "clip", "cols", "def", "device", "dim", "else", "flatten", "float64", "floor", "for", "full_like", "gather", "get_visual_bbox", "i", "if", "image_embeddings", "image_size", "in", "ind", "input_vision_patches", "inputs_embeds", "inputs_vision_patches", "is", "item", "len", "long", "max_len", "mean", "not", "num_patches", "ocr_points", "ocr_points_x", "ocr_points_y", "pad_sequence", "patch_inds", "patch_size", "range", "repeat", "repeated_vision_embeds", "return", "rows", "sequence_length", "size", "stack", "target_seg", "tensor", "to", "torch", "unsqueeze", "visual_attention_mask", "visual_bbox", "zeros_like", "zip"], "udop/modeling_udop.py:UdopPatchEmbeddings.__init__": ["Conv2d", "Iterable", "ModelPatchEmbeddings", "Module", "__init__", "abc", "class", "collections", "config", "def", "else", "hidden_size", "if", "image_size", "isinstance", "kernel_size", "nn", "num_channels", "num_patches", "patch_size", "proj", "self", "stride", "super"], "udop/modeling_udop.py:UdopPatchEmbeddings.forward": ["Input", "ModelPatchEmbeddings", "Module", "ValueError", "batch_size", "class", "def", "doesn", "embeddings", "f", "flatten", "forward", "height", "if", "image", "image_size", "match", "model", "nn", "num_channels", "or", "pixel_values", "proj", "raise", "return", "self", "shape", "size", "t", "transpose", "width"], "udop/modeling_udop.py:UdopPreTrainedModel._init_weights": ["Conv2d", "Embedding", "False", "ModelAttention", "ModelDenseActDense", "ModelDenseGatedActDense", "ModelForConditionalGeneration", "ModelLayerNorm", "ModelModel", "ModelPreTrainedModel", "None", "PreTrainedModel", "RelativePositionBiasBase", "_init_weights", "_is_hf_initialized", "and", "bias", "class", "config", "constant_", "d_ff", "d_kv", "d_model", "def", "elif", "factor", "getattr", "has_relative_attention_bias", "hasattr", "if", "init", "initializer_factor", "is", "isinstance", "k", "key_value_proj_dim", "lm_head", "mean", "module", "n_heads", "nn", "normal_", "not", "num_heads", "o", "padding_idx", "q", "relative_attention_bias", "self", "shared", "std", "tie_word_embeddings", "trunc_normal_", "v", "weight", "wi", "wi_0", "wi_1", "wo", "zeros_"], "udop/modeling_udop.py:UdopPreTrainedModel._shift_right": ["In", "Model", "ModelPreTrainedModel", "None", "PreTrainedModel", "See", "Verify", "_shift_right", "all", "assert", "be", "class", "clone", "config", "decoder_start_token_id", "def", "defined", "docs", "for", "has", "information", "input_ids", "is", "it", "item", "masked_fill_", "model", "more", "new_zeros", "not", "only", "pad_token_id", "positive", "return", "self", "set", "shape", "shifted_input_ids", "that", "the", "to", "torch", "usually", "values"], "udop/modeling_udop.py:UdopLayerNorm.__init__": ["ModelLayerNorm", "Module", "Parameter", "__init__", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "udop/modeling_udop.py:UdopLayerNorm.forward": ["ModelLayerNorm", "Module", "True", "bfloat16", "class", "def", "dtype", "float16", "float32", "forward", "hidden_states", "if", "in", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "udop/modeling_udop.py:UdopDenseActDense.__init__": ["ACT2FN", "Dropout", "False", "Linear", "ModelConfig", "ModelDenseActDense", "Module", "__init__", "act", "bias", "class", "config", "d_ff", "d_model", "def", "dense_act_fn", "dropout", "dropout_rate", "nn", "self", "super", "wi", "wo"], "udop/modeling_udop.py:UdopDenseActDense.forward": ["ModelDenseActDense", "Module", "Tensor", "act", "and", "class", "def", "dropout", "dtype", "forward", "hidden_states", "if", "int8", "isinstance", "nn", "return", "self", "to", "torch", "weight", "wi", "wo"], "udop/modeling_udop.py:UdopDenseGatedActDense.__init__": ["ACT2FN", "Dropout", "False", "Linear", "ModelConfig", "ModelDenseGatedActDense", "Module", "__init__", "act", "bias", "class", "config", "d_ff", "d_model", "def", "dense_act_fn", "dropout", "dropout_rate", "nn", "self", "super", "wi_0", "wi_1", "wo"], "udop/modeling_udop.py:UdopDenseGatedActDense.forward": ["ModelDenseGatedActDense", "Module", "Tensor", "act", "and", "class", "def", "dropout", "dtype", "forward", "hidden_gelu", "hidden_linear", "hidden_states", "if", "int8", "isinstance", "nn", "return", "self", "to", "torch", "weight", "wi_0", "wi_1", "wo"], "udop/modeling_udop.py:UdopLayerFF.__init__": ["DenseReluDense", "Dropout", "ModelConfig", "ModelDenseActDense", "ModelDenseGatedActDense", "ModelLayerFF", "ModelLayerNorm", "Module", "__init__", "class", "config", "d_model", "def", "dropout", "dropout_rate", "else", "eps", "if", "is_gated_act", "layer_norm", "layer_norm_epsilon", "nn", "self", "super"], "udop/modeling_udop.py:UdopLayerFF.forward": ["DenseReluDense", "ModelLayerFF", "Module", "class", "def", "dropout", "forward", "forwarded_states", "hidden_states", "layer_norm", "nn", "return", "self"], "udop/modeling_udop.py:UdopAttention.__init__": ["Embedding", "False", "Instantiating", "Linear", "ModelAttention", "ModelConfig", "Module", "None", "Please", "__class__", "__init__", "__name__", "a", "and", "bias", "caching", "call", "class", "config", "creating", "d_kv", "d_model", "decoder", "def", "dropout", "dropout_rate", "during", "errors", "f", "forward", "gradient_checkpointing", "has_relative_attention_bias", "if", "inner_dim", "int", "is", "is_decoder", "k", "key_value_proj_dim", "layer_idx", "logger", "make", "n_heads", "nn", "not", "num_heads", "o", "passing", "provide", "q", "recommended", "relative_attention_bias", "relative_attention_max_distance", "relative_attention_num_buckets", "self", "super", "sure", "the", "this", "to", "used", "v", "warning_once", "when", "will", "without"], "udop/modeling_udop.py:UdopAttention._relative_position_bucket": ["ModelAttention", "Module", "True", "_relative_position_bucket", "abs", "bidirectional", "class", "def", "else", "float", "full_like", "if", "is_small", "log", "long", "math", "max_distance", "max_exact", "min", "nn", "num_buckets", "relative_buckets", "relative_position", "relative_position_if_large", "return", "to", "torch", "where", "zeros_like"], "udop/modeling_udop.py:UdopAttention.compute_bias": ["ModelAttention", "Module", "None", "_relative_position_bucket", "arange", "bidirectional", "cache_position", "class", "compute_bias", "context_position", "def", "device", "dtype", "else", "if", "is", "is_decoder", "key_length", "long", "max_distance", "memory_position", "nn", "not", "num_buckets", "permute", "query_length", "relative_attention_bias", "relative_attention_max_distance", "relative_attention_num_buckets", "relative_position", "relative_position_bucket", "return", "self", "to", "torch", "unsqueeze", "values", "weight"], "udop/modeling_udop.py:UdopAttention.forward": ["EncoderDecoderCache", "False", "ModelAttention", "Module", "None", "True", "and", "attn_output", "attn_weights", "batch_size", "cache_position", "causal_mask", "class", "compute_bias", "contiguous", "cross_attention_cache", "curr_past_key_values", "current_states", "def", "device", "dim", "dropout", "dtype", "else", "float", "forward", "functional", "get", "gradient_checkpointing", "has_relative_attention_bias", "hidden_states", "if", "inner_dim", "is", "is_cross_attention", "is_updated", "isinstance", "k", "key_length", "key_states", "key_value_proj_dim", "key_value_states", "keys", "layer_idx", "layers", "mask", "matmul", "n_heads", "nn", "not", "o", "output_attentions", "outputs", "p", "past_key_values", "position_bias", "position_bias_masked", "q", "query_length", "query_states", "real_seq_length", "requires_grad", "return", "scores", "self", "self_attention_cache", "seq_length", "shape", "softmax", "torch", "training", "transpose", "type_as", "update", "use_cache", "v", "value_states", "values", "view", "zeros"], "udop/modeling_udop.py:UdopLayerSelfAttention.__init__": ["Dropout", "False", "ModelAttention", "ModelLayerNorm", "ModelLayerSelfAttention", "Module", "None", "SelfAttention", "__init__", "class", "config", "d_model", "def", "dropout", "dropout_rate", "eps", "has_relative_attention_bias", "int", "layer_idx", "layer_norm", "layer_norm_epsilon", "nn", "self", "super"], "udop/modeling_udop.py:UdopLayerSelfAttention.forward": ["False", "ModelLayerSelfAttention", "Module", "None", "SelfAttention", "attention_mask", "attention_output", "cache_position", "class", "def", "dropout", "forward", "hidden_states", "layer_norm", "mask", "nn", "normed_hidden_states", "output_attentions", "outputs", "past_key_values", "position_bias", "return", "self", "use_cache"], "udop/modeling_udop.py:UdopLayerCrossAttention.__init__": ["Dropout", "EncDecAttention", "False", "ModelAttention", "ModelLayerCrossAttention", "ModelLayerNorm", "Module", "None", "__init__", "class", "config", "d_model", "def", "dropout", "dropout_rate", "eps", "has_relative_attention_bias", "int", "layer_idx", "layer_norm", "layer_norm_epsilon", "nn", "self", "super"], "udop/modeling_udop.py:UdopLayerCrossAttention.forward": ["EncDecAttention", "False", "ModelLayerCrossAttention", "Module", "None", "attention_mask", "attention_output", "cache_position", "class", "def", "dropout", "forward", "hidden_states", "key_value_states", "layer_norm", "layer_output", "mask", "nn", "normed_hidden_states", "output_attentions", "outputs", "past_key_values", "position_bias", "query_length", "return", "self", "use_cache"], "udop/modeling_udop.py:UdopBlock.__init__": ["False", "GradientCheckpointingLayer", "ModelBlock", "ModelLayerCrossAttention", "ModelLayerFF", "ModelLayerSelfAttention", "ModuleList", "None", "__init__", "append", "class", "config", "def", "has_relative_attention_bias", "if", "int", "is_decoder", "layer", "layer_idx", "nn", "self", "super"], "udop/modeling_udop.py:UdopBlock.forward": ["False", "GradientCheckpointingLayer", "ModelBlock", "None", "True", "and", "any", "attention_mask", "attention_outputs", "cache_position", "clamp", "clamp_value", "class", "cross_attention_outputs", "def", "do_cross_attention", "dtype", "encoder_attention_mask", "encoder_decoder_position_bias", "encoder_hidden_states", "finfo", "float16", "forward", "hidden_states", "if", "is", "is_decoder", "isinf", "key_value_states", "layer", "max", "min", "not", "output_attentions", "outputs", "past_key_values", "position_bias", "query_length", "return", "return_dict", "self", "self_attention_outputs", "torch", "use_cache", "where"], "udop/modeling_udop.py:UdopCellEmbeddings.__init__": ["Embedding", "ModelCellEmbeddings", "Module", "__init__", "class", "def", "hidden_size", "max_2d_position_embeddings", "nn", "self", "super", "x_position_embeddings", "y_position_embeddings"], "udop/modeling_udop.py:UdopCellEmbeddings.forward": ["ModelCellEmbeddings", "Module", "bbox", "class", "clip", "def", "embeddings", "forward", "left_position_embeddings", "long", "lower_position_embeddings", "max_2d_position_embeddings", "nn", "return", "right_position_embeddings", "self", "torch", "upper_position_embeddings", "x_position_embeddings", "y_position_embeddings"], "udop/modeling_udop.py:RelativePositionBiasBase.__init__": ["ABC", "Embedding", "False", "ModelPositionBiasBase", "Model_attention_bias", "Model_attention_num_buckets", "Module", "None", "True", "__init__", "and", "augmentation", "bidirectional", "class", "def", "else", "expand", "extra_head", "if", "level", "max_distance", "nn", "not", "num_heads", "prefix_bucket", "scaling_factor", "self", "super", "tokens"], "udop/modeling_udop.py:RelativePositionBiasBase.prepare_input": ["ABC", "Any", "ModelPositionBiasBase", "Module", "None", "Tensor", "attention_mask", "bbox", "class", "def", "dict", "nn", "pass", "prepare_input", "self", "str"], "udop/modeling_udop.py:RelativePositionBiasBase.get_bucket": ["ABC", "Any", "ModelPositionBiasBase", "Model_attention_num_buckets", "Model_position", "Module", "None", "Tensor", "attention_mask", "bbox", "bidirectional", "class", "def", "dict", "get_Model_position_bucket", "get_bucket", "max_distance", "nn", "num_buckets", "prepare_input", "return", "rp_bucket", "self", "str"], "udop/modeling_udop.py:RelativePositionBiasBase.get_relative_position": ["ABC", "AUGMENTATION_RANGE", "ModelPositionBiasBase", "Model_position", "Module", "None", "and", "augmentation", "class", "context_position", "def", "get_Model_position", "if", "long", "memory_position", "nn", "positions", "random", "return", "scaling_factor", "self", "to", "torch", "training", "uniform"], "udop/modeling_udop.py:RelativePositionBiasBase.forward": ["ABC", "Any", "Embedding", "False", "ModelPositionBiasBase", "Model_attention_bias", "Model_attention_num_buckets", "Module", "None", "Tensor", "ValueError", "Wrong", "and", "attention_mask", "bbox", "class", "cpu", "data", "def", "dict", "dim", "dimension", "enumerate", "expand", "for", "forward", "get_bucket", "idx", "if", "in", "is_prefix", "new_bias", "nn", "num_heads", "num_prefix", "num_prefix_row", "numpy", "of", "permute", "prefix_bucket", "raise", "repeat", "return", "rp_bucket", "self", "size", "str", "sum", "tensor", "values", "weight"], "udop/modeling_udop.py:RelativePositionBias1D.__init__": ["ModelPositionBias1D", "ModelPositionBiasBase", "__init__", "class", "def", "kwargs", "max_distance", "scaling_factor", "self", "super"], "udop/modeling_udop.py:RelativePositionBias1D.prepare_input": ["Any", "ModelPositionBias1D", "ModelPositionBiasBase", "Model_position", "No", "None", "Tensor", "ValueError", "arange", "attention_mask", "bbox", "class", "def", "device", "dict", "dtype", "features", "get_Model_position", "if", "long", "need", "prepare_input", "raise", "return", "scale", "scaling_factor", "self", "size", "str", "to", "torch"], "udop/modeling_udop.py:RelativePositionBiasHorizontal.__init__": ["ModelPositionBiasBase", "ModelPositionBiasHorizontal", "__init__", "class", "def", "kwargs", "max_distance", "scaling_factor", "self", "super"], "udop/modeling_udop.py:RelativePositionBiasHorizontal.prepare_input": ["Any", "Bbox", "Model", "ModelPositionBiasBase", "ModelPositionBiasHorizontal", "Need", "None", "Tensor", "ValueError", "are", "as", "attention_mask", "bbox", "bboxes", "bias", "class", "def", "dict", "dim", "for", "get_Model_position", "horizontal", "horizontal_position", "if", "in", "is", "mean", "not", "of", "position", "prepare_input", "raise", "range", "required", "return", "scale", "scaling_factor", "self", "small", "str", "the", "there", "to", "values"], "udop/modeling_udop.py:RelativePositionBiasVertical.__init__": ["ModelPositionBiasBase", "ModelPositionBiasVertical", "__init__", "class", "def", "kwargs", "max_distance", "scaling_factor", "self", "super"], "udop/modeling_udop.py:RelativePositionBiasVertical.prepare_input": ["Any", "Bbox", "Model", "ModelPositionBiasBase", "ModelPositionBiasVertical", "Need", "None", "Tensor", "ValueError", "are", "as", "attention_mask", "bbox", "bboxes", "bias", "class", "def", "dict", "dim", "for", "get_Model_position", "if", "in", "is", "mean", "not", "of", "position", "prepare_input", "raise", "range", "required", "return", "scale", "scaling_factor", "self", "small", "str", "the", "there", "to", "values", "vertical", "vertical_position"], "udop/modeling_udop.py:RelativePositionBiasAggregated.__init__": ["ModelPositionBiasAggregated", "ModelPositionBiasBase", "Module", "ModuleList", "Sequence", "__init__", "biases", "class", "def", "modules", "nn", "self", "super"], "udop/modeling_udop.py:RelativePositionBiasAggregated.forward": ["Any", "ModelPositionBiasAggregated", "Module", "None", "Tensor", "attention_mask", "bbox", "bias", "biases", "class", "def", "dict", "float", "for", "forward", "in", "nn", "output", "return", "self", "str"], "udop/modeling_udop.py:create_relative_bias": ["BIAS_CLASSES", "ModelConfig", "Model_relative_bias", "Number", "RelativePositionBiasBase", "Sequence", "ValueError", "append", "bias_kwargs", "bias_kwargs_org", "bias_list", "bias_type", "config", "deepcopy", "def", "else", "for", "hasattr", "heads", "if", "in", "match", "model", "model_num_heads", "must", "num", "num_attention_heads", "num_heads", "of", "pop", "raise", "relative_bias_args", "return", "the", "type"], "udop/modeling_udop.py:UdopStack.__init__": ["Dropout", "Embedding", "ModelBlock", "ModelCellEmbeddings", "ModelLayerNorm", "ModelPatchEmbeddings", "ModelPreTrainedModel", "ModelStack", "ModuleList", "__init__", "_get_relative_bias", "block", "bool", "cell_2d_embedding", "class", "config", "d_model", "def", "dropout", "dropout_rate", "embed_patches", "embed_tokens", "eps", "final_layer_norm", "for", "has_relative_attention_bias", "hidden_size", "i", "if", "in", "is_decoder", "layer_idx", "layer_norm_epsilon", "max_2d_position_embeddings", "nn", "not", "num_layers", "post_init", "range", "relative_bias", "self", "super", "vocab_size"], "udop/modeling_udop.py:UdopStack._get_relative_bias": ["ModelConfig", "ModelPreTrainedModel", "ModelStack", "RelativePositionBiasAggregated", "_get_relative_bias", "class", "config", "create_relative_bias", "def", "relative_bias_list", "return"], "udop/modeling_udop.py:UdopStack.get_output_embeddings": ["ModelPreTrainedModel", "ModelStack", "class", "def", "embed_tokens", "get_output_embeddings", "return", "self"], "udop/modeling_udop.py:UdopStack.set_input_embeddings": ["ModelPreTrainedModel", "ModelStack", "class", "def", "embed_tokens", "new_embeddings", "self", "set_input_embeddings"], "udop/modeling_udop.py:UdopStack.forward": ["BaseModelOutputWithAttentionMask", "DynamicCache", "Empty", "EncoderDecoderCache", "ModelPreTrainedModel", "ModelStack", "None", "True", "ValueError", "You", "_update_causal_mask", "a", "all_attentions", "all_cross_attentions", "all_hidden_states", "and", "arange", "as", "assert", "at", "attention_mask", "attentions", "batch", "batch_size", "bbox", "be", "block", "both", "cache_position", "can", "cannot", "causal_mask", "cell_2d_embedding", "class", "combine_image_text_embeddings", "config", "cross_attentions", "decoder", "decoder_", "def", "device", "dropout", "dtype", "either", "elif", "else", "embed_patches", "embed_tokens", "embeddings", "encoder_attention_mask", "encoder_decoder_position_bias", "encoder_extended_attention_mask", "encoder_hidden_states", "enumerate", "err_msg_prefix", "f", "final_layer_norm", "finfo", "for", "forward", "full", "get_extended_attention_mask", "get_seq_length", "have", "hidden_states", "i", "if", "image_embeddings", "image_size", "in", "initialize", "input_ids", "input_shape", "inputs", "inputs_embeds", "invert_attention_mask", "is", "is_decoder", "is_encoder_decoder", "is_torchdynamo_compiling", "isinstance", "kwargs", "last_hidden_state", "layer_module", "layer_outputs", "logger", "mask_seq_length", "min", "model", "not", "num_patches", "numel", "ones", "only", "or", "output_attentions", "output_hidden_states", "pad_token_id", "past_key_values", "past_key_values_length", "patch_size", "pixel_values", "position_bias", "raise", "relative_bias", "return", "return_dict", "same", "self", "self_attention_cache", "seq_length", "set", "size", "specify", "the", "time", "to", "token", "torch", "tuple", "use_cache", "use_return_dict", "used", "v", "valid", "view", "visual_bbox", "warning", "with", "zeros", "zeros_like"], "udop/modeling_udop.py:UdopStack._update_causal_mask": ["AttentionMaskConverter", "BlockMask", "Cache", "False", "ModelPreTrainedModel", "ModelStack", "None", "Tensor", "Union", "_attn_implementation", "_ignore_causal_mask_sdpa", "_prepare_4d_causal_attention_mask_with_cache_position", "_unmask_unattended", "_update_causal_mask", "and", "any", "attention_mask", "batch_size", "bool", "cache_position", "causal_mask", "class", "config", "cuda", "def", "device", "dtype", "else", "finfo", "flash_attention_2", "flex_attention", "get_max_cache_shape", "get_seq_length", "if", "in", "input_tensor", "inputs_embeds", "is", "is_compileable", "is_training", "isinstance", "make_flex_block_causal_mask", "min", "min_dtype", "not", "npu", "output_attentions", "past_key_values", "past_key_values_length", "past_seen_tokens", "return", "sdpa", "self", "sequence_length", "shape", "target_length", "torch", "training", "type", "using_compilable_cache", "xpu"], "udop/modeling_udop.py:UdopStack._prepare_4d_causal_attention_mask_with_cache_position": ["ModelPreTrainedModel", "ModelStack", "None", "Tensor", "_prepare_4d_causal_attention_mask_with_cache_position", "and", "arange", "attention_mask", "batch_size", "cache_position", "causal_mask", "class", "clone", "def", "device", "diagonal", "dim", "dtype", "else", "expand", "fill_value", "finfo", "full", "if", "int", "is", "kwargs", "mask_length", "masked_fill", "min", "min_dtype", "not", "padding_mask", "reshape", "return", "sequence_length", "shape", "target_length", "to", "torch", "triu"], "udop/modeling_udop.py:UdopModel.__init__": ["Embedding", "False", "ModelModel", "ModelPatchEmbeddings", "ModelPreTrainedModel", "ModelStack", "True", "__init__", "class", "config", "d_model", "decoder", "decoder_config", "deepcopy", "def", "encoder", "encoder_config", "is_decoder", "nn", "num_decoder_layers", "num_layers", "patch_embed", "post_init", "self", "shared", "super", "use_cache", "vocab_size"], "udop/modeling_udop.py:UdopModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "return", "self", "shared"], "udop/modeling_udop.py:UdopModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "decoder", "def", "encoder", "new_embeddings", "self", "set_input_embeddings", "shared"], "udop/modeling_udop.py:UdopModel.forward": ["Any", "Cache", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Seq2SeqModelOutput", "Tensor", "True", "attention_mask", "attentions", "bbox", "bool", "cache_position", "class", "config", "cross_attentions", "decoder", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_input_ids", "decoder_inputs_embeds", "decoder_outputs", "def", "dict", "else", "encoder", "encoder_attention_mask", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "enumerate", "for", "forward", "hidden_states", "idx", "if", "in", "input_ids", "inputs_embeds", "is", "kwargs", "last_hidden_state", "not", "output_attentions", "output_hidden_states", "past_key_values", "pixel_values", "r", "return", "return_dict", "self", "str", "torch", "tuple", "use_cache", "use_return_dict", "value", "visual_bbox"], "udop/modeling_udop.py:UdopForConditionalGeneration.__init__": ["Embedding", "False", "GenerationMixin", "Linear", "ModelForConditionalGeneration", "ModelPatchEmbeddings", "ModelPreTrainedModel", "ModelStack", "True", "__init__", "bias", "class", "config", "d_model", "decoder", "decoder_config", "deepcopy", "def", "encoder", "encoder_config", "is_decoder", "lm_head", "nn", "num_decoder_layers", "num_layers", "patch_embed", "post_init", "self", "shared", "super", "use_cache", "vocab_size"], "udop/modeling_udop.py:UdopForConditionalGeneration.get_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "return", "self", "shared"], "udop/modeling_udop.py:UdopForConditionalGeneration.set_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "decoder", "def", "encoder", "new_embeddings", "self", "set_input_embeddings", "shared"], "udop/modeling_udop.py:UdopForConditionalGeneration.forward": ["Any", "Cache", "CrossEntropyLoss", "GenerationMixin", "LongTensor", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Seq2SeqLMOutput", "Tensor", "True", "_shift_right", "and", "attention_mask", "attentions", "bbox", "bool", "cache_position", "class", "config", "cross_attentions", "d_model", "decoder", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_input_ids", "decoder_inputs_embeds", "decoder_outputs", "def", "dict", "else", "encoder", "encoder_attention_mask", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "forward", "hidden_states", "if", "ignore_index", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "last_hidden_state", "lm_head", "lm_logits", "logits", "loss", "loss_fct", "not", "output", "output_attentions", "output_hidden_states", "past_key_values", "pixel_values", "r", "return", "return_dict", "self", "sequence_output", "size", "str", "tie_word_embeddings", "torch", "tuple", "use_cache", "use_return_dict", "view", "visual_bbox"], "udop/modeling_udop.py:UdopEncoderModel.__init__": ["Embedding", "False", "ModelConfig", "ModelEncoderModel", "ModelPatchEmbeddings", "ModelPreTrainedModel", "ModelStack", "__init__", "class", "config", "d_model", "deepcopy", "def", "encoder", "encoder_config", "is_decoder", "is_encoder_decoder", "nn", "patch_embed", "post_init", "self", "shared", "super", "use_cache", "vocab_size"], "udop/modeling_udop.py:UdopEncoderModel.get_input_embeddings": ["ModelEncoderModel", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "return", "self", "shared"], "udop/modeling_udop.py:UdopEncoderModel.set_input_embeddings": ["ModelEncoderModel", "ModelPreTrainedModel", "class", "def", "encoder", "new_embeddings", "self", "set_input_embeddings", "shared"], "udop/modeling_udop.py:UdopEncoderModel.forward": ["Any", "BaseModelOutputWithAttentionMask", "FloatTensor", "ModelEncoderModel", "ModelPreTrainedModel", "None", "Tensor", "attention_mask", "bbox", "bool", "class", "config", "def", "dict", "else", "encoder", "encoder_outputs", "forward", "if", "input_ids", "inputs_embeds", "is", "kwargs", "not", "output_attentions", "output_hidden_states", "pixel_values", "r", "return", "return_dict", "self", "str", "torch", "tuple", "use_return_dict", "visual_bbox"], "umt5/modeling_umt5.py:UMT5LayerNorm.__init__": ["Model", "Module", "Parameter", "__init__", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "umt5/modeling_umt5.py:UMT5LayerNorm.forward": ["Model", "Module", "True", "bfloat16", "class", "def", "dtype", "float16", "float32", "forward", "hidden_states", "if", "in", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "umt5/modeling_umt5.py:UMT5DenseActDense.__init__": ["ACT2FN", "Dropout", "False", "Linear", "Model", "ModelConfig", "Module", "__init__", "act", "bias", "class", "config", "d_ff", "d_model", "def", "dense_act_fn", "dropout", "dropout_rate", "nn", "self", "super", "wi", "wo"], "umt5/modeling_umt5.py:UMT5DenseActDense.forward": ["Model", "Module", "Tensor", "act", "and", "class", "def", "dropout", "dtype", "forward", "hidden_states", "if", "int8", "isinstance", "nn", "return", "self", "to", "torch", "weight", "wi", "wo"], "umt5/modeling_umt5.py:UMT5DenseGatedActDense.__init__": ["ACT2FN", "Dropout", "False", "Linear", "Model", "ModelConfig", "Module", "__init__", "act", "bias", "class", "config", "d_ff", "d_model", "def", "dense_act_fn", "dropout", "dropout_rate", "nn", "self", "super", "wi_0", "wi_1", "wo"], "umt5/modeling_umt5.py:UMT5DenseGatedActDense.forward": ["Model", "Module", "Tensor", "act", "and", "class", "def", "dropout", "dtype", "forward", "hidden_gelu", "hidden_linear", "hidden_states", "if", "int8", "isinstance", "nn", "return", "self", "to", "torch", "weight", "wi_0", "wi_1", "wo"], "umt5/modeling_umt5.py:UMT5LayerFF.__init__": ["DenseReluDense", "Dropout", "Model", "ModelConfig", "ModelDenseActDense", "ModelDenseGatedActDense", "ModelLayerNorm", "Module", "__init__", "class", "config", "d_model", "def", "dropout", "dropout_rate", "else", "eps", "if", "is_gated_act", "layer_norm", "layer_norm_epsilon", "nn", "self", "super"], "umt5/modeling_umt5.py:UMT5LayerFF.forward": ["DenseReluDense", "Model", "Module", "class", "def", "dropout", "forward", "forwarded_states", "hidden_states", "layer_norm", "nn", "return", "self"], "umt5/modeling_umt5.py:UMT5Attention.__init__": ["Embedding", "False", "Instantiating", "Linear", "Model", "Module", "None", "Please", "__class__", "__init__", "__name__", "a", "and", "bias", "caching", "call", "class", "config", "creating", "d_kv", "d_model", "decoder", "def", "dropout", "dropout_rate", "during", "errors", "f", "forward", "has_relative_attention_bias", "if", "inner_dim", "int", "is", "is_decoder", "k", "key_value_proj_dim", "layer_idx", "logger", "make", "n_heads", "nn", "not", "num_heads", "o", "passing", "provide", "q", "recommended", "relative_attention_bias", "relative_attention_max_distance", "relative_attention_num_buckets", "self", "super", "sure", "the", "this", "to", "used", "v", "warning_once", "when", "will", "without"], "umt5/modeling_umt5.py:UMT5Attention._shape": ["Model", "Module", "Tensor", "_shape", "class", "def", "key_value_proj_dim", "n_heads", "new_projection", "new_projection_shape", "nn", "permute", "projection", "return", "self", "size", "torch", "view"], "umt5/modeling_umt5.py:UMT5Attention._relative_position_bucket": ["Model", "Module", "_relative_position_bucket", "abs", "class", "def", "else", "float", "full_like", "if", "is_decoder", "is_small", "log", "log_ratio", "long", "math", "max_distance", "max_exact", "min", "nn", "not", "num_buckets", "relative_attention_max_distance", "relative_attention_num_buckets", "relative_buckets", "relative_position", "relative_position_if_large", "return", "self", "to", "torch", "where", "zeros_like"], "umt5/modeling_umt5.py:UMT5Attention.compute_bias": ["Model", "Module", "None", "_relative_position_bucket", "arange", "cache_position", "class", "compute_bias", "context_position", "def", "device", "dtype", "else", "if", "is", "key_length", "long", "memory_position", "nn", "permute", "query_length", "relative_attention_bias", "relative_position", "relative_position_bucket", "return", "self", "torch", "unsqueeze", "values", "weight"], "umt5/modeling_umt5.py:UMT5Attention.forward": ["Cache", "EncoderDecoderCache", "False", "Model", "Module", "None", "Tensor", "True", "and", "attention_mask", "attn_output", "attn_weights", "batch_size", "cache_position", "causal_mask", "class", "compute_bias", "contiguous", "cross_attention_cache", "curr_past_key_values", "current_states", "def", "device", "dim", "dropout", "dtype", "else", "encoder_hidden_states", "float", "forward", "functional", "get", "get_seq_length", "has_relative_attention_bias", "hidden_states", "if", "is", "is_cross_attention", "is_updated", "isinstance", "k", "key_length", "key_states", "key_value_proj_dim", "keys", "layer_idx", "layers", "matmul", "n_heads", "nn", "not", "o", "p", "past_key_values", "position_bias", "position_bias_masked", "q", "query_states", "real_seq_length", "return", "scores", "self", "self_attention_cache", "seq_length", "shape", "softmax", "torch", "training", "transpose", "type_as", "update", "v", "value_states", "values", "view", "zeros"], "umt5/modeling_umt5.py:UMT5LayerSelfAttention.__init__": ["Dropout", "Model", "ModelAttention", "ModelLayerNorm", "Module", "None", "SelfAttention", "True", "__init__", "class", "config", "d_model", "def", "dropout", "dropout_rate", "eps", "has_relative_attention_bias", "int", "layer_idx", "layer_norm", "layer_norm_epsilon", "nn", "self", "super"], "umt5/modeling_umt5.py:UMT5LayerSelfAttention.forward": ["Model", "Module", "None", "SelfAttention", "attention_mask", "attention_output", "cache_position", "class", "def", "dropout", "forward", "hidden_states", "layer_norm", "nn", "normed_hidden_states", "outputs", "past_key_values", "return", "self"], "umt5/modeling_umt5.py:UMT5LayerCrossAttention.__init__": ["Dropout", "EncDecAttention", "False", "Model", "ModelAttention", "ModelLayerNorm", "Module", "None", "__init__", "class", "config", "d_model", "def", "dropout", "dropout_rate", "eps", "has_relative_attention_bias", "int", "layer_idx", "layer_norm", "layer_norm_epsilon", "nn", "self", "super"], "umt5/modeling_umt5.py:UMT5LayerCrossAttention.forward": ["EncDecAttention", "Model", "Module", "None", "attention_mask", "attention_output", "cache_position", "class", "def", "dropout", "encoder_hidden_states", "forward", "hidden_states", "layer_norm", "layer_output", "nn", "normed_hidden_states", "outputs", "past_key_values", "return", "self"], "umt5/modeling_umt5.py:UMT5Block.__init__": ["GradientCheckpointingLayer", "Model", "ModelLayerCrossAttention", "ModelLayerFF", "ModelLayerSelfAttention", "ModuleList", "None", "__init__", "append", "class", "config", "def", "if", "int", "is_decoder", "layer", "layer_idx", "nn", "self", "super"], "umt5/modeling_umt5.py:UMT5Block.forward": ["False", "GradientCheckpointingLayer", "Model", "None", "and", "any", "attention_mask", "cache_position", "clamp", "clamp_value", "class", "cross_attn_weights", "def", "do_cross_attention", "dtype", "encoder_attention_mask", "encoder_hidden_states", "finfo", "float16", "forward", "hidden_states", "if", "is", "is_decoder", "isinf", "layer", "max", "max_dtype", "min", "not", "output_attentions", "outputs", "past_key_values", "return", "self", "self_attn_weights", "torch", "use_cache", "where"], "umt5/modeling_umt5.py:UMT5ClassificationHead.__init__": ["Dropout", "Linear", "Model", "ModelConfig", "Module", "__init__", "class", "classifier_dropout", "config", "d_model", "def", "dense", "dropout", "nn", "num_labels", "out_proj", "p", "self", "super"], "umt5/modeling_umt5.py:UMT5ClassificationHead.forward": ["Model", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "nn", "out_proj", "return", "self", "tanh", "torch"], "umt5/modeling_umt5.py:UMT5PreTrainedModel.dummy_inputs": ["DUMMY_INPUTS", "DUMMY_MASK", "Model", "PreTrainedModel", "class", "decoder_attention_mask", "decoder_input_ids", "def", "dummy_inputs", "input_ids", "input_mask", "return", "self", "tensor", "torch"], "umt5/modeling_umt5.py:UMT5PreTrainedModel._init_weights": ["Model", "ModelAttention", "ModelClassificationHead", "ModelDenseActDense", "ModelDenseGatedActDense", "ModelEncoderModel", "ModelForConditionalGeneration", "ModelForQuestionAnswering", "ModelForTokenClassification", "ModelLayerNorm", "ModelModel", "None", "PreTrainedModel", "_init_weights", "and", "bias", "class", "classifier", "config", "constant_", "d_ff", "d_kv", "d_model", "def", "dense", "elif", "factor", "has_relative_attention_bias", "hasattr", "if", "init", "initializer_factor", "is", "isinstance", "k", "key_value_proj_dim", "lm_head", "mean", "module", "n_heads", "normal_", "not", "num_heads", "o", "out_proj", "q", "qa_outputs", "relative_attention_bias", "self", "shared", "std", "tie_word_embeddings", "v", "weight", "wi", "wi_0", "wi_1", "wo", "zeros_"], "umt5/modeling_umt5.py:UMT5PreTrainedModel._shift_right": ["In", "Model", "None", "PreTrainedModel", "See", "ValueError", "_shift_right", "be", "class", "clone", "config", "decoder_start_token_id", "def", "defined", "docs", "for", "has", "if", "information", "input_ids", "is", "it", "masked_fill_", "model", "more", "new_zeros", "pad_token_id", "raise", "return", "self", "set", "shape", "shifted_input_ids", "the", "to", "usually"], "umt5/modeling_umt5.py:UMT5Stack.__init__": ["Dropout", "Embedding", "False", "Model", "ModelBlock", "ModelLayerNorm", "ModelPreTrainedModel", "ModuleList", "__init__", "block", "class", "config", "d_model", "def", "dropout", "dropout_rate", "embed_tokens", "eps", "final_layer_norm", "for", "gradient_checkpointing", "i", "in", "is_decoder", "layer_idx", "layer_norm_epsilon", "nn", "num_layers", "post_init", "range", "self", "super", "vocab_size"], "umt5/modeling_umt5.py:UMT5Stack.set_input_embeddings": ["Model", "ModelPreTrainedModel", "class", "def", "embed_tokens", "new_embeddings", "self", "set_input_embeddings"], "umt5/modeling_umt5.py:UMT5Stack.forward": ["BaseModelOutputWithPastAndCrossAttentions", "DynamicCache", "EncoderDecoderCache", "False", "Model", "ModelPreTrainedModel", "None", "Setting", "True", "ValueError", "You", "_", "_update_causal_mask", "a", "all_attentions", "all_cross_attentions", "all_hidden_states", "and", "arange", "as", "at", "attention_mask", "attentions", "batch_size", "be", "block", "both", "cache_position", "can", "cannot", "causal_mask", "checkpointing", "class", "config", "cross_attentions", "decoder", "decoder_", "def", "device", "dropout", "dtype", "either", "elif", "else", "embed_tokens", "embeddings", "encoder_attention_mask", "encoder_batch_size", "encoder_extended_attention_mask", "encoder_hidden_shape", "encoder_hidden_states", "encoder_sequence_length", "enumerate", "err_msg_prefix", "f", "final_layer_norm", "finfo", "for", "forward", "get_seq_length", "gradient", "gradient_checkpointing", "have", "hidden_states", "i", "if", "in", "incompatible", "initialize", "input_ids", "input_shape", "inputs_embeds", "invert_attention_mask", "is", "is_decoder", "is_encoder_decoder", "is_torchdynamo_compiling", "isinstance", "kwargs", "last_hidden_state", "layer_module", "layer_outputs", "logger", "mask_seq_length", "min", "model", "not", "ones", "only", "or", "output_attentions", "output_hidden_states", "past_key_values", "past_key_values_length", "raise", "return", "return_dict", "same", "self", "self_attention_cache", "seq_length", "set", "size", "specify", "the", "time", "to", "token", "torch", "training", "tuple", "use_cache", "use_return_dict", "used", "v", "valid", "view", "warning_once", "with"], "umt5/modeling_umt5.py:UMT5Stack._update_causal_mask": ["AttentionMaskConverter", "BlockMask", "Cache", "False", "Model", "ModelPreTrainedModel", "None", "Tensor", "Union", "_attn_implementation", "_ignore_causal_mask_sdpa", "_prepare_4d_causal_attention_mask_with_cache_position", "_unmask_unattended", "_update_causal_mask", "and", "any", "attention_mask", "batch_size", "bool", "cache_position", "causal_mask", "class", "config", "cuda", "def", "device", "dtype", "else", "finfo", "flash_attention_2", "flex_attention", "get_max_cache_shape", "get_seq_length", "if", "in", "input_tensor", "inputs_embeds", "is", "is_compileable", "is_training", "isinstance", "make_flex_block_causal_mask", "min", "min_dtype", "not", "npu", "output_attentions", "past_key_values", "past_key_values_length", "past_seen_tokens", "return", "sdpa", "self", "sequence_length", "shape", "target_length", "torch", "training", "type", "using_compilable_cache", "xpu"], "umt5/modeling_umt5.py:UMT5Stack._prepare_4d_causal_attention_mask_with_cache_position": ["Model", "ModelPreTrainedModel", "None", "Tensor", "_prepare_4d_causal_attention_mask_with_cache_position", "and", "arange", "attention_mask", "batch_size", "cache_position", "causal_mask", "class", "clone", "def", "device", "diagonal", "dim", "dtype", "else", "expand", "fill_value", "finfo", "full", "if", "int", "is", "kwargs", "mask_length", "masked_fill", "min", "min_dtype", "not", "padding_mask", "reshape", "return", "sequence_length", "shape", "target_length", "to", "torch", "triu"], "umt5/modeling_umt5.py:UMT5Model.__init__": ["Embedding", "False", "Model", "ModelPreTrainedModel", "ModelStack", "True", "__init__", "class", "config", "copy", "d_model", "decoder", "decoder_config", "deepcopy", "def", "encoder", "encoder_config", "is_decoder", "nn", "num_decoder_layers", "num_layers", "post_init", "self", "shared", "super", "use_cache", "vocab_size"], "umt5/modeling_umt5.py:UMT5Model.get_input_embeddings": ["Model", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "return", "self", "shared"], "umt5/modeling_umt5.py:UMT5Model.set_input_embeddings": ["Model", "ModelPreTrainedModel", "class", "decoder", "def", "encoder", "new_embeddings", "self", "set_input_embeddings", "shared"], "umt5/modeling_umt5.py:UMT5Model.forward": ["BaseModelOutput", "BoolTensor", "Cache", "FloatTensor", "LongTensor", "Model", "ModelPreTrainedModel", "None", "Seq2SeqModelOutput", "Tensor", "and", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "cross_attentions", "decoder", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_input_ids", "decoder_inputs_embeds", "decoder_outputs", "def", "elif", "else", "encoder", "encoder_attention_mask", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "isinstance", "kwargs", "last_hidden_state", "len", "not", "output_attentions", "output_hidden_states", "past_key_values", "r", "return", "return_dict", "self", "torch", "tuple", "use_cache", "use_return_dict"], "umt5/modeling_umt5.py:UMT5ForConditionalGeneration.__init__": ["Embedding", "False", "GenerationMixin", "Linear", "Model", "ModelPreTrainedModel", "ModelStack", "True", "__init__", "bias", "class", "config", "copy", "d_model", "decoder", "decoder_config", "deepcopy", "def", "encoder", "encoder_config", "is_decoder", "lm_head", "model_dim", "nn", "num_decoder_layers", "num_layers", "post_init", "self", "shared", "super", "use_cache", "vocab_size"], "umt5/modeling_umt5.py:UMT5ForConditionalGeneration.get_input_embeddings": ["GenerationMixin", "Model", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "return", "self", "shared"], "umt5/modeling_umt5.py:UMT5ForConditionalGeneration.set_input_embeddings": ["GenerationMixin", "Model", "ModelPreTrainedModel", "class", "decoder", "def", "encoder", "new_embeddings", "self", "set_input_embeddings", "shared"], "umt5/modeling_umt5.py:UMT5ForConditionalGeneration.forward": ["BaseModelOutput", "BoolTensor", "Cache", "CrossEntropyLoss", "FloatTensor", "GenerationMixin", "LongTensor", "Model", "ModelPreTrainedModel", "None", "Seq2SeqLMOutput", "Tensor", "_shift_right", "and", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "cross_attentions", "decoder", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_input_ids", "decoder_inputs_embeds", "decoder_outputs", "def", "device", "elif", "else", "encoder", "encoder_attention_mask", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "forward", "hidden_states", "if", "ignore_index", "input_ids", "inputs_embeds", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "len", "lm_head", "lm_logits", "logits", "loss", "loss_fct", "model_dim", "not", "output", "output_attentions", "output_hidden_states", "past_key_values", "r", "return", "return_dict", "self", "sequence_output", "size", "tie_word_embeddings", "to", "torch", "tuple", "use_cache", "use_return_dict", "view"], "umt5/modeling_umt5.py:UMT5ForConditionalGeneration.prepare_decoder_input_ids_from_labels": ["GenerationMixin", "Model", "ModelPreTrainedModel", "Tensor", "_shift_right", "class", "def", "labels", "prepare_decoder_input_ids_from_labels", "return", "self", "torch"], "umt5/modeling_umt5.py:UMT5EncoderModel.__init__": ["Embedding", "False", "Model", "ModelPreTrainedModel", "ModelStack", "__init__", "class", "config", "copy", "d_model", "deepcopy", "def", "encoder", "encoder_config", "is_encoder_decoder", "nn", "post_init", "self", "shared", "super", "use_cache", "vocab_size"], "umt5/modeling_umt5.py:UMT5EncoderModel.get_input_embeddings": ["Model", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "return", "self", "shared"], "umt5/modeling_umt5.py:UMT5EncoderModel.set_input_embeddings": ["Model", "ModelPreTrainedModel", "class", "def", "encoder", "new_embeddings", "self", "set_input_embeddings", "shared"], "umt5/modeling_umt5.py:UMT5EncoderModel.forward": ["BaseModelOutput", "FloatTensor", "LongTensor", "Model", "ModelPreTrainedModel", "None", "attention_mask", "bool", "class", "config", "def", "else", "encoder", "encoder_outputs", "forward", "if", "input_ids", "inputs_embeds", "is", "kwargs", "not", "output_attentions", "output_hidden_states", "r", "return", "return_dict", "self", "torch", "tuple", "use_return_dict"], "umt5/modeling_umt5.py:UMT5ForSequenceClassification.__init__": ["Model", "ModelClassificationHead", "ModelConfig", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "classification_head", "config", "def", "post_init", "self", "super", "transformer"], "umt5/modeling_umt5.py:UMT5ForSequenceClassification.forward": ["All", "BCEWithLogitsLoss", "CrossEntropyLoss", "False", "FloatTensor", "If", "LongTensor", "MSELoss", "Model", "ModelPreTrainedModel", "None", "NotImplementedError", "Passing", "Please", "Seq2SeqSequenceClassifierOutput", "Tensor", "ValueError", "_", "__class__", "__name__", "_shift_right", "and", "are", "attention_mask", "batch_size", "be", "bool", "cannot", "class", "classification_head", "config", "cross_attentions", "currently", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_input_ids", "decoder_inputs_embeds", "def", "device", "dtype", "either", "elif", "else", "embeddings", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "eos", "eos_mask", "eos_token_id", "eq", "examples", "f", "for", "forward", "have", "hidden_size", "if", "input", "input_ids", "inputs_embeds", "int", "is", "kwargs", "labels", "len", "list", "logits", "long", "loss", "loss_fct", "multi_label_classification", "must", "no", "not", "num_labels", "number", "of", "or", "output", "output_attentions", "output_hidden_states", "outputs", "pass", "passed", "past_key_values", "problem_type", "r", "raise", "regression", "return", "return_dict", "same", "self", "sentence_representation", "sequence_output", "shape", "single_label_classification", "squeeze", "sum", "supported", "the", "to", "tokens", "torch", "transformer", "tuple", "unique_consecutive", "use_cache", "use_return_dict", "view"], "umt5/modeling_umt5.py:UMT5ForTokenClassification.__init__": ["Dropout", "Linear", "Model", "ModelConfig", "ModelEncoderModel", "ModelPreTrainedModel", "__init__", "class", "classifier", "classifier_dropout", "config", "def", "dropout", "hidden_size", "nn", "num_labels", "post_init", "self", "super", "transformer"], "umt5/modeling_umt5.py:UMT5ForTokenClassification.forward": ["CrossEntropyLoss", "Model", "ModelPreTrainedModel", "None", "Tensor", "TokenClassifierOutput", "attention_mask", "attentions", "bool", "class", "classifier", "config", "def", "dropout", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "not", "num_labels", "output", "output_attentions", "output_hidden_states", "outputs", "r", "return", "return_dict", "self", "torch", "transformer", "tuple", "use_return_dict", "view"], "umt5/modeling_umt5.py:UMT5ForQuestionAnswering.__init__": ["Embedding", "False", "Linear", "Model", "ModelPreTrainedModel", "ModelStack", "True", "__init__", "class", "config", "copy", "d_model", "decoder", "decoder_config", "deepcopy", "def", "encoder", "encoder_config", "is_decoder", "model_dim", "nn", "num_decoder_layers", "num_labels", "num_layers", "post_init", "qa_outputs", "self", "shared", "super", "use_cache", "vocab_size"], "umt5/modeling_umt5.py:UMT5ForQuestionAnswering.get_input_embeddings": ["Model", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "return", "self", "shared"], "umt5/modeling_umt5.py:UMT5ForQuestionAnswering.set_input_embeddings": ["Model", "ModelPreTrainedModel", "class", "decoder", "def", "encoder", "new_embeddings", "self", "set_input_embeddings", "shared"], "umt5/modeling_umt5.py:UMT5ForQuestionAnswering.forward": ["BaseModelOutput", "BoolTensor", "CrossEntropyLoss", "False", "FloatTensor", "If", "LongTensor", "Model", "ModelPreTrainedModel", "None", "Please", "Seq2SeqQuestionAnsweringModelOutput", "Tensor", "ValueError", "_shift_right", "and", "are", "attention_mask", "attentions", "be", "bool", "cannot", "clamp", "class", "config", "contiguous", "cross_attentions", "decoder", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_input_ids", "decoder_inputs_embeds", "decoder_outputs", "def", "device", "dim", "either", "elif", "else", "encoder", "encoder_attention_mask", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "end_logits", "end_loss", "end_positions", "forward", "hidden_states", "if", "ignore_index", "ignored_index", "input_ids", "inputs_embeds", "is", "isinstance", "kwargs", "last_hidden_state", "len", "logits", "loss", "loss_fct", "no", "not", "or", "output", "output_attentions", "output_hidden_states", "pass", "passed", "past_key_values", "qa_outputs", "r", "raise", "return", "return_dict", "self", "sequence_output", "size", "split", "squeeze", "start_logits", "start_loss", "start_positions", "to", "torch", "total_loss", "tuple", "use_cache", "use_return_dict"], "unispeech/modeling_unispeech.py:UniSpeechSamePadLayer.__init__": ["ModelSamePadLayer", "Module", "__init__", "class", "def", "else", "if", "nn", "num_conv_pos_embeddings", "num_pad_remove", "self", "super"], "unispeech/modeling_unispeech.py:UniSpeechSamePadLayer.forward": ["ModelSamePadLayer", "Module", "class", "def", "forward", "hidden_states", "if", "nn", "num_pad_remove", "return", "self"], "unispeech/modeling_unispeech.py:UniSpeechPositionalConvEmbedding.__init__": ["ACT2FN", "Conv1d", "GatheredParameters", "ModelPositionalConvEmbedding", "ModelSamePadLayer", "Module", "__init__", "activation", "class", "config", "conv", "deepspeed", "def", "dim", "else", "feat_extract_activation", "groups", "hasattr", "hidden_size", "if", "is_deepspeed_zero3_enabled", "kernel_size", "modifier_rank", "name", "nn", "num_conv_pos_embedding_groups", "num_conv_pos_embeddings", "original0", "original1", "padding", "parametrizations", "register_external_parameter", "self", "super", "utils", "weight", "weight_g", "weight_norm", "weight_v", "with", "zero"], "unispeech/modeling_unispeech.py:UniSpeechPositionalConvEmbedding.forward": ["ModelPositionalConvEmbedding", "Module", "activation", "class", "conv", "def", "forward", "hidden_states", "nn", "padding", "return", "self", "transpose"], "unispeech/modeling_unispeech.py:UniSpeechNoLayerNormConvLayer.__init__": ["ACT2FN", "Conv1d", "GradientCheckpointingLayer", "ModelNoLayerNormConvLayer", "__init__", "activation", "bias", "class", "config", "conv", "conv_bias", "conv_dim", "conv_kernel", "conv_stride", "def", "else", "feat_extract_activation", "if", "in_conv_dim", "kernel_size", "layer_id", "nn", "out_conv_dim", "self", "stride", "super"], "unispeech/modeling_unispeech.py:UniSpeechNoLayerNormConvLayer.forward": ["GradientCheckpointingLayer", "ModelNoLayerNormConvLayer", "activation", "class", "conv", "def", "forward", "hidden_states", "return", "self"], "unispeech/modeling_unispeech.py:UniSpeechLayerNormConvLayer.__init__": ["ACT2FN", "Conv1d", "GradientCheckpointingLayer", "LayerNorm", "ModelLayerNormConvLayer", "True", "__init__", "activation", "bias", "class", "config", "conv", "conv_bias", "conv_dim", "conv_kernel", "conv_stride", "def", "elementwise_affine", "else", "feat_extract_activation", "if", "in_conv_dim", "kernel_size", "layer_id", "layer_norm", "nn", "out_conv_dim", "self", "stride", "super"], "unispeech/modeling_unispeech.py:UniSpeechLayerNormConvLayer.forward": ["GradientCheckpointingLayer", "ModelLayerNormConvLayer", "activation", "class", "conv", "def", "forward", "hidden_states", "layer_norm", "return", "self", "transpose"], "unispeech/modeling_unispeech.py:UniSpeechGroupNormConvLayer.__init__": ["ACT2FN", "Conv1d", "GradientCheckpointingLayer", "GroupNorm", "ModelGroupNormConvLayer", "True", "__init__", "activation", "affine", "bias", "class", "config", "conv", "conv_bias", "conv_dim", "conv_kernel", "conv_stride", "def", "else", "feat_extract_activation", "if", "in_conv_dim", "kernel_size", "layer_id", "layer_norm", "nn", "num_channels", "num_groups", "out_conv_dim", "self", "stride", "super"], "unispeech/modeling_unispeech.py:UniSpeechGroupNormConvLayer.forward": ["GradientCheckpointingLayer", "ModelGroupNormConvLayer", "activation", "class", "conv", "def", "forward", "hidden_states", "layer_norm", "return", "self"], "unispeech/modeling_unispeech.py:UniSpeechFeatureEncoder.__init__": ["False", "ModelFeatureEncoder", "ModelGroupNormConvLayer", "ModelLayerNormConvLayer", "ModelNoLayerNormConvLayer", "Module", "ModuleList", "True", "ValueError", "__init__", "_requires_grad", "be", "but", "class", "config", "conv_layers", "def", "elif", "else", "f", "feat_extract_norm", "for", "gradient_checkpointing", "group", "has", "i", "if", "in", "is", "layer", "layer_id", "nn", "num_feat_extract_layers", "of", "one", "raise", "range", "self", "super", "to"], "unispeech/modeling_unispeech.py:UniSpeechFeatureEncoder._freeze_parameters": ["False", "ModelFeatureEncoder", "Module", "_freeze_parameters", "_requires_grad", "class", "def", "for", "in", "nn", "param", "parameters", "requires_grad", "self"], "unispeech/modeling_unispeech.py:UniSpeechFeatureEncoder.forward": ["ModelFeatureEncoder", "Module", "None", "True", "_requires_grad", "and", "class", "conv_layer", "conv_layers", "def", "for", "forward", "hidden_states", "if", "in", "input_values", "nn", "requires_grad", "return", "self", "training"], "unispeech/modeling_unispeech.py:UniSpeechFeatureProjection.__init__": ["Dropout", "LayerNorm", "Linear", "ModelFeatureProjection", "Module", "__init__", "class", "config", "conv_dim", "def", "dropout", "eps", "feat_proj_dropout", "hidden_size", "layer_norm", "layer_norm_eps", "nn", "projection", "self", "super"], "unispeech/modeling_unispeech.py:UniSpeechFeatureProjection.forward": ["ModelFeatureProjection", "Module", "class", "def", "dropout", "forward", "hidden_states", "layer_norm", "nn", "norm_hidden_states", "projection", "return", "self"], "unispeech/modeling_unispeech.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "contiguous", "def", "dim", "dropout", "float", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "shape", "size", "softmax", "torch", "training", "transpose", "value"], "unispeech/modeling_unispeech.py:UniSpeechAttention.__init__": ["False", "Linear", "ModelAttention", "ModelConfig", "Module", "None", "True", "ValueError", "__init__", "and", "be", "bias", "bool", "by", "class", "config", "def", "divisible", "dropout", "embed_dim", "f", "float", "got", "head_dim", "if", "int", "is_causal", "is_decoder", "k_proj", "must", "nn", "num_heads", "out_proj", "q_proj", "raise", "scaling", "self", "super", "v_proj"], "unispeech/modeling_unispeech.py:UniSpeechAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "False", "FlashAttentionKwargs", "ModelAttention", "Module", "None", "Tensor", "Unpack", "_attn_implementation", "attention_interface", "attention_mask", "attn_output", "attn_weights", "bool", "bsz", "class", "config", "contiguous", "current_states", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "head_dim", "hidden_states", "if", "is", "is_cross_attention", "k_proj", "key_states", "key_value_states", "kv_input_shape", "kwargs", "nn", "not", "out_proj", "output_attentions", "q_input_shape", "q_proj", "query_states", "reshape", "return", "scaling", "self", "shape", "src_len", "tgt_len", "torch", "training", "transpose", "tuple", "v_proj", "value_states", "view"], "unispeech/modeling_unispeech.py:UniSpeechFeedForward.__init__": ["ACT2FN", "Dropout", "Linear", "ModelFeedForward", "Module", "__init__", "activation_dropout", "class", "config", "def", "else", "hidden_act", "hidden_dropout", "hidden_size", "if", "intermediate_act_fn", "intermediate_dense", "intermediate_dropout", "intermediate_size", "isinstance", "nn", "output_dense", "output_dropout", "self", "str", "super"], "unispeech/modeling_unispeech.py:UniSpeechFeedForward.forward": ["ModelFeedForward", "Module", "class", "def", "forward", "hidden_states", "intermediate_act_fn", "intermediate_dense", "intermediate_dropout", "nn", "output_dense", "output_dropout", "return", "self"], "unispeech/modeling_unispeech.py:UniSpeechEncoderLayer.__init__": ["Dropout", "False", "GradientCheckpointingLayer", "LayerNorm", "ModelAttention", "ModelEncoderLayer", "ModelFeedForward", "__init__", "attention", "attention_dropout", "class", "config", "def", "dropout", "embed_dim", "eps", "feed_forward", "final_layer_norm", "hidden_dropout", "hidden_size", "is_decoder", "layer_norm", "layer_norm_eps", "nn", "num_attention_heads", "num_heads", "self", "super"], "unispeech/modeling_unispeech.py:UniSpeechEncoderLayer.forward": ["False", "GradientCheckpointingLayer", "ModelEncoderLayer", "None", "_", "attention", "attention_mask", "attn_residual", "attn_weights", "class", "def", "dropout", "feed_forward", "final_layer_norm", "forward", "hidden_states", "if", "layer_norm", "output_attentions", "outputs", "return", "self"], "unispeech/modeling_unispeech.py:UniSpeechEncoder.__init__": ["Dropout", "False", "LayerNorm", "ModelEncoder", "ModelEncoderLayer", "ModelPositionalConvEmbedding", "Module", "ModuleList", "_", "__init__", "class", "config", "def", "dropout", "eps", "for", "gradient_checkpointing", "hidden_dropout", "hidden_size", "in", "layer_norm", "layer_norm_eps", "layers", "nn", "num_hidden_layers", "pos_conv_embed", "range", "self", "super"], "unispeech/modeling_unispeech.py:UniSpeechEncoder.forward": ["BaseModelOutput", "False", "ModelEncoder", "Module", "None", "Tensor", "True", "all_hidden_states", "all_self_attentions", "and", "attention_mask", "attentions", "bool", "class", "config", "create_bidirectional_mask", "def", "dropout", "dropout_probability", "else", "expand_attention_mask", "for", "forward", "hidden_states", "if", "in", "input_embeds", "is", "is_deepspeed_zero3_enabled", "is_fsdp_managed_module", "last_hidden_state", "layer", "layer_norm", "layer_outputs", "layerdrop", "layers", "nn", "not", "or", "output_attentions", "output_hidden_states", "pos_conv_embed", "position_embeddings", "rand", "repeat", "return", "return_dict", "self", "shape", "skip_the_layer", "synced_gpus", "tensor", "torch", "training", "tuple", "unsqueeze", "v"], "unispeech/modeling_unispeech.py:UniSpeechAttnAdapterLayer.__init__": ["LayerNorm", "Linear", "ModelAttnAdapterLayer", "Module", "ReLU", "__init__", "act_fn", "adapter_attn_dim", "class", "config", "def", "hidden_dim", "hidden_size", "input_dim", "linear_1", "linear_2", "nn", "norm", "self", "super"], "unispeech/modeling_unispeech.py:UniSpeechAttnAdapterLayer.forward": ["FloatTensor", "ModelAttnAdapterLayer", "Module", "act_fn", "class", "def", "forward", "hidden_states", "linear_1", "linear_2", "nn", "norm", "return", "self", "torch"], "unispeech/modeling_unispeech.py:UniSpeechEncoderLayerStableLayerNorm.__init__": ["Dropout", "False", "GradientCheckpointingLayer", "LayerNorm", "ModelAttention", "ModelAttnAdapterLayer", "ModelEncoderLayerStableLayerNorm", "ModelFeedForward", "None", "__init__", "adapter_attn_dim", "adapter_layer", "attention", "attention_dropout", "class", "config", "def", "dropout", "else", "embed_dim", "eps", "feed_forward", "final_layer_norm", "getattr", "hidden_dropout", "hidden_size", "if", "is", "is_decoder", "layer_norm", "layer_norm_eps", "nn", "not", "num_attention_heads", "num_heads", "self", "super"], "unispeech/modeling_unispeech.py:UniSpeechEncoderLayerStableLayerNorm.forward": ["False", "GradientCheckpointingLayer", "ModelEncoderLayerStableLayerNorm", "None", "Tensor", "_", "adapter_layer", "attention", "attention_mask", "attn_residual", "attn_weights", "bool", "class", "def", "dropout", "feed_forward", "final_layer_norm", "forward", "hidden_states", "if", "is", "layer_norm", "not", "output_attentions", "outputs", "return", "self", "torch"], "unispeech/modeling_unispeech.py:UniSpeechEncoderStableLayerNorm.__init__": ["Dropout", "False", "LayerNorm", "ModelEncoderLayerStableLayerNorm", "ModelEncoderStableLayerNorm", "ModelPositionalConvEmbedding", "Module", "ModuleList", "_", "__init__", "class", "config", "def", "dropout", "eps", "for", "gradient_checkpointing", "hidden_dropout", "hidden_size", "in", "layer_norm", "layer_norm_eps", "layers", "nn", "num_hidden_layers", "pos_conv_embed", "range", "self", "super"], "unispeech/modeling_unispeech.py:UniSpeechEncoderStableLayerNorm.forward": ["BaseModelOutput", "False", "ModelEncoderStableLayerNorm", "Module", "None", "True", "all_hidden_states", "all_self_attentions", "and", "attention_mask", "attentions", "class", "config", "create_bidirectional_mask", "def", "dropout", "dropout_probability", "else", "expand_attention_mask", "for", "forward", "hidden_states", "if", "in", "input_embeds", "is", "is_deepspeed_zero3_enabled", "is_fsdp_managed_module", "last_hidden_state", "layer", "layer_norm", "layer_outputs", "layerdrop", "layers", "nn", "not", "or", "output_attentions", "output_hidden_states", "pos_conv_embed", "position_embeddings", "rand", "repeat", "return", "return_dict", "self", "shape", "skip_the_layer", "synced_gpus", "torch", "training", "tuple", "unsqueeze", "v"], "unispeech/modeling_unispeech.py:UniSpeechGumbelVectorQuantizer.__init__": ["FloatTensor", "Linear", "ModelGumbelVectorQuantizer", "Module", "Parameter", "ValueError", "__init__", "be", "by", "class", "codevector_dim", "codevectors", "concatenation", "config", "conv_dim", "def", "divisible", "f", "for", "if", "must", "nn", "num_codevector_groups", "num_codevectors_per_group", "num_groups", "num_vars", "raise", "self", "super", "temperature", "torch", "weight_proj"], "unispeech/modeling_unispeech.py:UniSpeechGumbelVectorQuantizer._compute_perplexity": ["ModelGumbelVectorQuantizer", "Module", "_compute_perplexity", "class", "def", "dim", "exp", "log", "marginal_probs", "mean", "nn", "perplexity", "probs", "return", "sum", "torch"], "unispeech/modeling_unispeech.py:UniSpeechGumbelVectorQuantizer.forward": ["ModelGumbelVectorQuantizer", "Module", "True", "_compute_perplexity", "argmax", "batch_size", "class", "codevector_idx", "codevector_probs", "codevector_soft_dist", "codevectors", "codevectors_per_group", "def", "dim", "else", "float", "forward", "functional", "gumbel_softmax", "hard", "hidden_size", "hidden_states", "if", "new_zeros", "nn", "num_groups", "num_vars", "perplexity", "return", "scatter_", "self", "sequence_length", "shape", "softmax", "sum", "tau", "temperature", "torch", "training", "type_as", "unsqueeze", "view", "weight_proj"], "unispeech/modeling_unispeech.py:UniSpeechPreTrainedModel._init_weights": ["Conv1d", "GroupNorm", "LayerNorm", "Linear", "ModelFeatureProjection", "ModelGumbelVectorQuantizer", "ModelPositionalConvEmbedding", "ModelPreTrainedModel", "Modelform_", "None", "PreTrainedModel", "_init_weights", "a", "b", "bias", "class", "codevectors", "config", "constant_", "conv", "def", "elif", "groups", "if", "in_channels", "in_features", "init", "initializer_range", "is", "isinstance", "k", "kaiming_normal_", "kernel_size", "math", "mean", "module", "nn", "normal_", "not", "ones_", "projection", "self", "sqrt", "std", "weight", "weight_proj", "zeros_"], "unispeech/modeling_unispeech.py:UniSpeechPreTrainedModel._get_feat_extract_output_lengths": ["LongTensor", "ModelPreTrainedModel", "PreTrainedModel", "_conv_out_length", "_get_feat_extract_output_lengths", "class", "config", "conv_kernel", "conv_stride", "def", "div", "floor", "for", "in", "input_length", "input_lengths", "int", "kernel_size", "return", "rounding_mode", "self", "stride", "torch", "zip"], "unispeech/modeling_unispeech.py:UniSpeechPreTrainedModel._get_feature_vector_attention_mask": ["LongTensor", "ModelPreTrainedModel", "PreTrainedModel", "_get_feat_extract_output_lengths", "_get_feature_vector_attention_mask", "arange", "attention_mask", "batch_size", "bool", "class", "cumsum", "def", "device", "dim", "dtype", "feature_vector_length", "flip", "int", "long", "non_padded_lengths", "output_lengths", "return", "self", "shape", "to", "torch", "zeros"], "unispeech/modeling_unispeech.py:_compute_mask_indices": ["False", "LongTensor", "None", "ValueError", "_", "_compute_mask_indices", "and", "append", "arange", "array", "attention_mask", "batch_size", "be", "bigger", "bool", "broadcast_to", "but", "choice", "compute_num_masked_span", "concatenate", "def", "detach", "dtype", "dummy_mask_idx", "else", "epsilon", "f", "float", "for", "got", "has", "if", "in", "input_length", "input_lengths", "int", "int32", "is", "item", "len", "mask_length", "mask_prob", "max", "max_num_masked_span", "min_masks", "ndarray", "not", "np", "num_masked_span", "offsets", "ones", "put_along_axis", "raise", "rand", "random", "range", "replace", "reshape", "return", "sequence_length", "shape", "smaller", "spec_aug_mask", "spec_aug_mask_idx", "spec_aug_mask_idxs", "sum", "than", "to", "tolist", "torch", "tuple", "zeros"], "unispeech/modeling_unispeech.py:UniSpeechModel.__init__": ["ModelConfig", "ModelEncoder", "ModelEncoderStableLayerNorm", "ModelFeatureEncoder", "ModelFeatureProjection", "ModelModel", "ModelPreTrainedModel", "Modelform_", "Parameter", "Tensor", "__init__", "class", "config", "def", "do_stable_layer_norm", "else", "encoder", "feature_extractor", "feature_projection", "hidden_size", "if", "mask_feature_prob", "mask_time_prob", "masked_spec_embed", "nn", "or", "post_init", "self", "super", "torch"], "unispeech/modeling_unispeech.py:UniSpeechModel._mask_hidden_states": ["FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "True", "_compute_mask_indices", "_mask_hidden_states", "and", "apply_spec_augment", "attention_mask", "batch_size", "bool", "class", "config", "def", "device", "dtype", "elif", "expand", "getattr", "hidden_size", "hidden_states", "if", "is", "mask_feature_indices", "mask_feature_length", "mask_feature_min_masks", "mask_feature_prob", "mask_length", "mask_prob", "mask_time_indices", "mask_time_length", "mask_time_min_masks", "mask_time_prob", "masked_spec_embed", "min_masks", "not", "return", "self", "sequence_length", "size", "tensor", "to", "torch", "training"], "unispeech/modeling_unispeech.py:UniSpeechModel.forward": ["FloatTensor", "ModelBaseModelOutput", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "_get_feature_vector_attention_mask", "_mask_hidden_states", "attention_mask", "attentions", "bool", "class", "config", "def", "else", "encoder", "encoder_outputs", "extract_features", "feature_extractor", "feature_projection", "forward", "hidden_states", "if", "input_values", "is", "kwargs", "last_hidden_state", "mask_time_indices", "not", "output_attentions", "output_hidden_states", "r", "return", "return_dict", "self", "shape", "torch", "transpose", "tuple", "use_return_dict"], "unispeech/modeling_unispeech.py:UniSpeechForPreTraining.__init__": ["Dropout", "Linear", "Model", "ModelConfig", "ModelForPreTraining", "ModelGumbelVectorQuantizer", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "codevector_dim", "config", "ctc_proj", "def", "dropout", "dropout_features", "feat_quantizer_dropout", "final_dropout", "hidden_size", "nn", "num_ctc_classes", "post_init", "proj_codevector_dim", "project_hid", "project_q", "quantizer", "self", "super"], "unispeech/modeling_unispeech.py:UniSpeechForPreTraining.set_gumbel_temperature": ["ModelForPreTraining", "ModelPreTrainedModel", "class", "def", "int", "quantizer", "self", "set_gumbel_temperature", "temperature"], "unispeech/modeling_unispeech.py:UniSpeechForPreTraining.freeze_feature_encoder": ["Model", "ModelForPreTraining", "ModelPreTrainedModel", "_freeze_parameters", "class", "def", "feature_extractor", "freeze_feature_encoder", "self"], "unispeech/modeling_unispeech.py:UniSpeechForPreTraining.compute_contrastive_logits": ["FloatTensor", "ModelForPreTraining", "ModelPreTrainedModel", "cat", "class", "compute_contrastive_logits", "cosine_similarity", "def", "dim", "float", "int", "logits", "negative_features", "predicted_features", "return", "target_features", "temperature", "torch", "type_as"], "unispeech/modeling_unispeech.py:UniSpeechForPreTraining.forward": ["Model", "ModelForPreTraining", "ModelForPreTrainingOutput", "ModelPreTrainedModel", "None", "Tensor", "attention_mask", "attentions", "bernoulli", "bool", "class", "codevector_perplexity", "config", "ctc_proj", "def", "device", "dropout", "dropout_features", "dtype", "else", "empty", "extract_features", "fill_", "forward", "hidden_states", "if", "input_values", "is", "kwargs", "logits", "loss", "masked_fill", "not", "output_attentions", "output_hidden_states", "outputs", "prob_replace_matrix", "project_hid", "project_q", "projected_quantized_states", "projected_states", "quantized_features", "quantizer", "r", "replace_prob", "return", "return_dict", "sampled_replace_matrix", "self", "size", "to", "torch", "transformer_features", "transpose", "tuple", "unsqueeze", "use_return_dict", "weight"], "unispeech/modeling_unispeech.py:UniSpeechForCTC.__init__": ["Dropout", "Linear", "Model", "ModelForCTC", "ModelModel", "ModelPreTrainedModel", "None", "Please", "ValueError", "You", "__class__", "__init__", "a", "add_adapter", "and", "are", "as", "class", "config", "configuration", "def", "define", "does", "dropout", "else", "f", "final_dropout", "follows", "from_pretrained", "hasattr", "head", "hidden_size", "if", "instantiate", "is", "language", "lm_head", "model", "nn", "not", "of", "or", "output_hidden_size", "post_init", "r", "raise", "s", "self", "size", "str", "super", "target_lang", "that", "the", "to", "trying", "vocab_size", "vocabulary", "with", "your"], "unispeech/modeling_unispeech.py:UniSpeechForCTC.tie_weights": ["By", "Cannot", "ModelForCTC", "ModelPreTrainedModel", "None", "True", "ValueError", "adapter_attn_dim", "and", "class", "config", "def", "default", "defined", "elif", "eng", "f", "force_load", "getattr", "if", "info", "is", "kwargs", "load_adapter", "logger", "not", "pass", "raise", "self", "set", "target_lang", "tie_weights", "to"], "unispeech/modeling_unispeech.py:UniSpeechForCTC.freeze_feature_encoder": ["Model", "ModelForCTC", "ModelPreTrainedModel", "_freeze_parameters", "class", "def", "feature_extractor", "freeze_feature_encoder", "self"], "unispeech/modeling_unispeech.py:UniSpeechForCTC.freeze_base_model": ["False", "Model", "ModelForCTC", "ModelPreTrainedModel", "class", "def", "for", "freeze_base_model", "in", "param", "parameters", "requires_grad", "self"], "unispeech/modeling_unispeech.py:UniSpeechForCTC.forward": ["CausalLMOutput", "False", "Label", "Model", "ModelForCTC", "ModelPreTrainedModel", "None", "Tensor", "ValueError", "_HIDDEN_STATES_START_POSITION", "_get_feat_extract_output_lengths", "and", "attention_mask", "attentions", "backends", "be", "blank", "bool", "class", "config", "ctc_loss", "ctc_loss_reduction", "ctc_zero_infinity", "cudnn", "def", "dim", "dropout", "dtype", "else", "enabled", "f", "flags", "flattened_targets", "float32", "forward", "functional", "hidden_states", "if", "input_lengths", "input_values", "is", "kwargs", "labels", "labels_mask", "lm_head", "log_probs", "log_softmax", "logits", "long", "loss", "masked_select", "max", "must", "nn", "not", "ones_like", "output", "output_attentions", "output_hidden_states", "outputs", "pad_token_id", "r", "raise", "reduction", "return", "return_dict", "self", "sum", "target_lengths", "to", "torch", "transpose", "tuple", "use_return_dict", "values", "vocab_size", "with", "zero_infinity"], "unispeech/modeling_unispeech.py:UniSpeechForSequenceClassification.__init__": ["Linear", "Model", "ModelForSequenceClassification", "ModelModel", "ModelPreTrainedModel", "Parameter", "Sequence", "True", "ValueError", "__init__", "adapters", "add_adapter", "and", "class", "classification", "classifier", "classifier_proj_size", "config", "def", "does", "hasattr", "hidden_size", "if", "layer_weights", "nn", "not", "num_hidden_layers", "num_labels", "num_layers", "of", "ones", "post_init", "projector", "raise", "self", "super", "support", "the", "torch", "use", "use_weighted_layer_sum"], "unispeech/modeling_unispeech.py:UniSpeechForSequenceClassification.freeze_feature_encoder": ["Model", "ModelForSequenceClassification", "ModelPreTrainedModel", "_freeze_parameters", "class", "def", "feature_extractor", "freeze_feature_encoder", "self"], "unispeech/modeling_unispeech.py:UniSpeechForSequenceClassification.freeze_base_model": ["False", "Model", "ModelForSequenceClassification", "ModelPreTrainedModel", "class", "def", "for", "freeze_base_model", "in", "param", "parameters", "requires_grad", "self"], "unispeech/modeling_unispeech.py:UniSpeechForSequenceClassification.forward": ["CrossEntropyLoss", "Model", "ModelForSequenceClassification", "ModelPreTrainedModel", "None", "SequenceClassifierOutput", "Tensor", "True", "_HIDDEN_STATES_START_POSITION", "_get_feature_vector_attention_mask", "attention_mask", "attentions", "bool", "class", "classifier", "config", "def", "dim", "else", "expand_padding_mask", "forward", "functional", "hidden_states", "if", "input_values", "is", "kwargs", "labels", "layer_weights", "logits", "loss", "loss_fct", "mean", "nn", "norm_weights", "not", "num_labels", "output", "output_attentions", "output_hidden_states", "outputs", "padding_mask", "pooled_output", "projector", "r", "repeat", "return", "return_dict", "self", "shape", "softmax", "stack", "sum", "torch", "tuple", "unsqueeze", "use_return_dict", "use_weighted_layer_sum", "view"], "unispeech_sat/modeling_unispeech_sat.py:UniSpeechSatSamePadLayer.__init__": ["ModelSamePadLayer", "Module", "__init__", "class", "def", "else", "if", "nn", "num_conv_pos_embeddings", "num_pad_remove", "self", "super"], "unispeech_sat/modeling_unispeech_sat.py:UniSpeechSatSamePadLayer.forward": ["ModelSamePadLayer", "Module", "class", "def", "forward", "hidden_states", "if", "nn", "num_pad_remove", "return", "self"], "unispeech_sat/modeling_unispeech_sat.py:UniSpeechSatPositionalConvEmbedding.__init__": ["ACT2FN", "Conv1d", "GatheredParameters", "ModelPositionalConvEmbedding", "ModelSamePadLayer", "Module", "__init__", "activation", "class", "config", "conv", "deepspeed", "def", "dim", "else", "feat_extract_activation", "groups", "hasattr", "hidden_size", "if", "is_deepspeed_zero3_enabled", "kernel_size", "modifier_rank", "name", "nn", "num_conv_pos_embedding_groups", "num_conv_pos_embeddings", "original0", "original1", "padding", "parametrizations", "register_external_parameter", "self", "super", "utils", "weight", "weight_g", "weight_norm", "weight_v", "with", "zero"], "unispeech_sat/modeling_unispeech_sat.py:UniSpeechSatPositionalConvEmbedding.forward": ["ModelPositionalConvEmbedding", "Module", "activation", "class", "conv", "def", "forward", "hidden_states", "nn", "padding", "return", "self", "transpose"], "unispeech_sat/modeling_unispeech_sat.py:UniSpeechSatNoLayerNormConvLayer.__init__": ["ACT2FN", "Conv1d", "GradientCheckpointingLayer", "ModelNoLayerNormConvLayer", "__init__", "activation", "bias", "class", "config", "conv", "conv_bias", "conv_dim", "conv_kernel", "conv_stride", "def", "else", "feat_extract_activation", "if", "in_conv_dim", "kernel_size", "layer_id", "nn", "out_conv_dim", "self", "stride", "super"], "unispeech_sat/modeling_unispeech_sat.py:UniSpeechSatNoLayerNormConvLayer.forward": ["GradientCheckpointingLayer", "ModelNoLayerNormConvLayer", "activation", "class", "conv", "def", "forward", "hidden_states", "return", "self"], "unispeech_sat/modeling_unispeech_sat.py:UniSpeechSatLayerNormConvLayer.__init__": ["ACT2FN", "Conv1d", "GradientCheckpointingLayer", "LayerNorm", "ModelLayerNormConvLayer", "True", "__init__", "activation", "bias", "class", "config", "conv", "conv_bias", "conv_dim", "conv_kernel", "conv_stride", "def", "elementwise_affine", "else", "feat_extract_activation", "if", "in_conv_dim", "kernel_size", "layer_id", "layer_norm", "nn", "out_conv_dim", "self", "stride", "super"], "unispeech_sat/modeling_unispeech_sat.py:UniSpeechSatLayerNormConvLayer.forward": ["GradientCheckpointingLayer", "ModelLayerNormConvLayer", "activation", "class", "conv", "def", "forward", "hidden_states", "layer_norm", "return", "self", "transpose"], "unispeech_sat/modeling_unispeech_sat.py:UniSpeechSatGroupNormConvLayer.__init__": ["ACT2FN", "Conv1d", "GradientCheckpointingLayer", "GroupNorm", "ModelGroupNormConvLayer", "True", "__init__", "activation", "affine", "bias", "class", "config", "conv", "conv_bias", "conv_dim", "conv_kernel", "conv_stride", "def", "else", "feat_extract_activation", "if", "in_conv_dim", "kernel_size", "layer_id", "layer_norm", "nn", "num_channels", "num_groups", "out_conv_dim", "self", "stride", "super"], "unispeech_sat/modeling_unispeech_sat.py:UniSpeechSatGroupNormConvLayer.forward": ["GradientCheckpointingLayer", "ModelGroupNormConvLayer", "activation", "class", "conv", "def", "forward", "hidden_states", "layer_norm", "return", "self"], "unispeech_sat/modeling_unispeech_sat.py:UniSpeechSatFeatureEncoder.__init__": ["False", "ModelFeatureEncoder", "ModelGroupNormConvLayer", "ModelLayerNormConvLayer", "ModelNoLayerNormConvLayer", "Module", "ModuleList", "True", "ValueError", "__init__", "_requires_grad", "be", "but", "class", "config", "conv_layers", "def", "elif", "else", "f", "feat_extract_norm", "for", "gradient_checkpointing", "group", "has", "i", "if", "in", "is", "layer", "layer_id", "nn", "num_feat_extract_layers", "of", "one", "raise", "range", "self", "super", "to"], "unispeech_sat/modeling_unispeech_sat.py:UniSpeechSatFeatureEncoder._freeze_parameters": ["False", "ModelFeatureEncoder", "Module", "_freeze_parameters", "_requires_grad", "class", "def", "for", "in", "nn", "param", "parameters", "requires_grad", "self"], "unispeech_sat/modeling_unispeech_sat.py:UniSpeechSatFeatureEncoder.forward": ["ModelFeatureEncoder", "Module", "None", "True", "_requires_grad", "and", "class", "conv_layer", "conv_layers", "def", "for", "forward", "hidden_states", "if", "in", "input_values", "nn", "requires_grad", "return", "self", "training"], "unispeech_sat/modeling_unispeech_sat.py:UniSpeechSatFeatureProjection.__init__": ["Dropout", "LayerNorm", "Linear", "ModelFeatureProjection", "Module", "__init__", "class", "config", "conv_dim", "def", "dropout", "eps", "feat_proj_dropout", "hidden_size", "layer_norm", "layer_norm_eps", "nn", "projection", "self", "super"], "unispeech_sat/modeling_unispeech_sat.py:UniSpeechSatFeatureProjection.forward": ["ModelFeatureProjection", "Module", "class", "def", "dropout", "forward", "hidden_states", "layer_norm", "nn", "norm_hidden_states", "projection", "return", "self"], "unispeech_sat/modeling_unispeech_sat.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "contiguous", "def", "dim", "dropout", "float", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "shape", "size", "softmax", "torch", "training", "transpose", "value"], "unispeech_sat/modeling_unispeech_sat.py:UniSpeechSatAttention.__init__": ["False", "Linear", "ModelAttention", "ModelConfig", "Module", "None", "True", "ValueError", "__init__", "and", "be", "bias", "bool", "by", "class", "config", "def", "divisible", "dropout", "embed_dim", "f", "float", "got", "head_dim", "if", "int", "is_causal", "is_decoder", "k_proj", "must", "nn", "num_heads", "out_proj", "q_proj", "raise", "scaling", "self", "super", "v_proj"], "unispeech_sat/modeling_unispeech_sat.py:UniSpeechSatAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "False", "FlashAttentionKwargs", "ModelAttention", "Module", "None", "Tensor", "Unpack", "_attn_implementation", "attention_interface", "attention_mask", "attn_output", "attn_weights", "bool", "bsz", "class", "config", "contiguous", "current_states", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "head_dim", "hidden_states", "if", "is", "is_cross_attention", "k_proj", "key_states", "key_value_states", "kv_input_shape", "kwargs", "nn", "not", "out_proj", "output_attentions", "q_input_shape", "q_proj", "query_states", "reshape", "return", "scaling", "self", "shape", "src_len", "tgt_len", "torch", "training", "transpose", "tuple", "v_proj", "value_states", "view"], "unispeech_sat/modeling_unispeech_sat.py:UniSpeechSatFeedForward.__init__": ["ACT2FN", "Dropout", "Linear", "ModelFeedForward", "Module", "__init__", "activation_dropout", "class", "config", "def", "else", "hidden_act", "hidden_dropout", "hidden_size", "if", "intermediate_act_fn", "intermediate_dense", "intermediate_dropout", "intermediate_size", "isinstance", "nn", "output_dense", "output_dropout", "self", "str", "super"], "unispeech_sat/modeling_unispeech_sat.py:UniSpeechSatFeedForward.forward": ["ModelFeedForward", "Module", "class", "def", "forward", "hidden_states", "intermediate_act_fn", "intermediate_dense", "intermediate_dropout", "nn", "output_dense", "output_dropout", "return", "self"], "unispeech_sat/modeling_unispeech_sat.py:UniSpeechSatEncoderLayer.__init__": ["Dropout", "False", "GradientCheckpointingLayer", "LayerNorm", "ModelAttention", "ModelEncoderLayer", "ModelFeedForward", "__init__", "attention", "attention_dropout", "class", "config", "def", "dropout", "embed_dim", "eps", "feed_forward", "final_layer_norm", "hidden_dropout", "hidden_size", "is_decoder", "layer_norm", "layer_norm_eps", "nn", "num_attention_heads", "num_heads", "self", "super"], "unispeech_sat/modeling_unispeech_sat.py:UniSpeechSatEncoderLayer.forward": ["False", "GradientCheckpointingLayer", "ModelEncoderLayer", "None", "_", "attention", "attention_mask", "attn_residual", "attn_weights", "class", "def", "dropout", "feed_forward", "final_layer_norm", "forward", "hidden_states", "if", "layer_norm", "output_attentions", "outputs", "return", "self"], "unispeech_sat/modeling_unispeech_sat.py:UniSpeechSatEncoder.__init__": ["Dropout", "False", "LayerNorm", "ModelEncoder", "ModelEncoderLayer", "ModelPositionalConvEmbedding", "Module", "ModuleList", "_", "__init__", "class", "config", "def", "dropout", "eps", "for", "gradient_checkpointing", "hidden_dropout", "hidden_size", "in", "layer_norm", "layer_norm_eps", "layers", "nn", "num_hidden_layers", "pos_conv_embed", "range", "self", "super"], "unispeech_sat/modeling_unispeech_sat.py:UniSpeechSatEncoder.forward": ["BaseModelOutput", "False", "ModelEncoder", "Module", "None", "Tensor", "True", "all_hidden_states", "all_self_attentions", "and", "attention_mask", "attentions", "bool", "class", "config", "create_bidirectional_mask", "def", "dropout", "dropout_probability", "else", "expand_attention_mask", "for", "forward", "hidden_states", "if", "in", "input_embeds", "is", "is_deepspeed_zero3_enabled", "is_fsdp_managed_module", "last_hidden_state", "layer", "layer_norm", "layer_outputs", "layerdrop", "layers", "nn", "not", "or", "output_attentions", "output_hidden_states", "pos_conv_embed", "position_embeddings", "rand", "repeat", "return", "return_dict", "self", "shape", "skip_the_layer", "synced_gpus", "tensor", "torch", "training", "tuple", "unsqueeze", "v"], "unispeech_sat/modeling_unispeech_sat.py:UniSpeechSatAttnAdapterLayer.__init__": ["LayerNorm", "Linear", "ModelAttnAdapterLayer", "Module", "ReLU", "__init__", "act_fn", "adapter_attn_dim", "class", "config", "def", "hidden_dim", "hidden_size", "input_dim", "linear_1", "linear_2", "nn", "norm", "self", "super"], "unispeech_sat/modeling_unispeech_sat.py:UniSpeechSatAttnAdapterLayer.forward": ["FloatTensor", "ModelAttnAdapterLayer", "Module", "act_fn", "class", "def", "forward", "hidden_states", "linear_1", "linear_2", "nn", "norm", "return", "self", "torch"], "unispeech_sat/modeling_unispeech_sat.py:UniSpeechSatEncoderLayerStableLayerNorm.__init__": ["Dropout", "False", "GradientCheckpointingLayer", "LayerNorm", "ModelAttention", "ModelAttnAdapterLayer", "ModelEncoderLayerStableLayerNorm", "ModelFeedForward", "None", "__init__", "adapter_attn_dim", "adapter_layer", "attention", "attention_dropout", "class", "config", "def", "dropout", "else", "embed_dim", "eps", "feed_forward", "final_layer_norm", "getattr", "hidden_dropout", "hidden_size", "if", "is", "is_decoder", "layer_norm", "layer_norm_eps", "nn", "not", "num_attention_heads", "num_heads", "self", "super"], "unispeech_sat/modeling_unispeech_sat.py:UniSpeechSatEncoderLayerStableLayerNorm.forward": ["False", "GradientCheckpointingLayer", "ModelEncoderLayerStableLayerNorm", "None", "Tensor", "_", "adapter_layer", "attention", "attention_mask", "attn_residual", "attn_weights", "bool", "class", "def", "dropout", "feed_forward", "final_layer_norm", "forward", "hidden_states", "if", "is", "layer_norm", "not", "output_attentions", "outputs", "return", "self", "torch"], "unispeech_sat/modeling_unispeech_sat.py:UniSpeechSatEncoderStableLayerNorm.__init__": ["Dropout", "False", "LayerNorm", "ModelEncoderLayerStableLayerNorm", "ModelEncoderStableLayerNorm", "ModelPositionalConvEmbedding", "Module", "ModuleList", "_", "__init__", "class", "config", "def", "dropout", "eps", "for", "gradient_checkpointing", "hidden_dropout", "hidden_size", "in", "layer_norm", "layer_norm_eps", "layers", "nn", "num_hidden_layers", "pos_conv_embed", "range", "self", "super"], "unispeech_sat/modeling_unispeech_sat.py:UniSpeechSatEncoderStableLayerNorm.forward": ["BaseModelOutput", "False", "ModelEncoderStableLayerNorm", "Module", "None", "True", "all_hidden_states", "all_self_attentions", "and", "attention_mask", "attentions", "class", "config", "create_bidirectional_mask", "def", "dropout", "dropout_probability", "else", "expand_attention_mask", "for", "forward", "hidden_states", "if", "in", "input_embeds", "is", "is_deepspeed_zero3_enabled", "is_fsdp_managed_module", "last_hidden_state", "layer", "layer_norm", "layer_outputs", "layerdrop", "layers", "nn", "not", "or", "output_attentions", "output_hidden_states", "pos_conv_embed", "position_embeddings", "rand", "repeat", "return", "return_dict", "self", "shape", "skip_the_layer", "synced_gpus", "torch", "training", "tuple", "unsqueeze", "v"], "unispeech_sat/modeling_unispeech_sat.py:UniSpeechSatGumbelVectorQuantizer.__init__": ["FloatTensor", "Linear", "ModelGumbelVectorQuantizer", "Module", "Parameter", "ValueError", "__init__", "be", "by", "class", "codevector_dim", "codevectors", "concatenation", "config", "def", "divisible", "f", "for", "hidden_size", "if", "must", "nn", "num_codevector_groups", "num_codevectors_per_group", "num_groups", "num_vars", "raise", "self", "super", "temperature", "torch", "weight_proj"], "unispeech_sat/modeling_unispeech_sat.py:UniSpeechSatGumbelVectorQuantizer._compute_perplexity": ["ModelGumbelVectorQuantizer", "Module", "None", "_compute_perplexity", "class", "def", "dim", "exp", "log", "marginal_probs", "mask", "mean", "nn", "perplexity", "probs", "return", "sum", "torch"], "unispeech_sat/modeling_unispeech_sat.py:UniSpeechSatGumbelVectorQuantizer.forward": ["ModelGumbelVectorQuantizer", "Module", "True", "_compute_perplexity", "argmax", "batch_size", "class", "codevector_idx", "codevector_probs", "codevector_soft_dist", "codevectors", "codevectors_per_group", "def", "dim", "else", "float", "forward", "functional", "gumbel_softmax", "hard", "hidden_size", "hidden_states", "if", "new_zeros", "nn", "num_groups", "num_vars", "perplexity", "return", "scatter_", "self", "sequence_length", "shape", "softmax", "sum", "tau", "temperature", "torch", "training", "type_as", "unsqueeze", "view", "weight_proj"], "unispeech_sat/modeling_unispeech_sat.py:UniSpeechSatPreTrainedModel._init_weights": ["Conv1d", "GroupNorm", "LayerNorm", "Linear", "ModelFeatureProjection", "ModelGumbelVectorQuantizer", "ModelPositionalConvEmbedding", "ModelPreTrainedModel", "Modelform_", "None", "PreTrainedModel", "_init_weights", "a", "b", "bias", "class", "codevectors", "config", "constant_", "conv", "def", "elif", "groups", "if", "in_channels", "in_features", "init", "initializer_range", "is", "isinstance", "k", "kaiming_normal_", "kernel_size", "math", "mean", "module", "nn", "normal_", "not", "ones_", "projection", "self", "sqrt", "std", "weight", "weight_proj", "zeros_"], "unispeech_sat/modeling_unispeech_sat.py:UniSpeechSatPreTrainedModel._get_feat_extract_output_lengths": ["LongTensor", "ModelPreTrainedModel", "PreTrainedModel", "_conv_out_length", "_get_feat_extract_output_lengths", "class", "config", "conv_kernel", "conv_stride", "def", "div", "floor", "for", "in", "input_length", "input_lengths", "int", "kernel_size", "return", "rounding_mode", "self", "stride", "torch", "zip"], "unispeech_sat/modeling_unispeech_sat.py:UniSpeechSatPreTrainedModel._get_feature_vector_attention_mask": ["LongTensor", "ModelPreTrainedModel", "PreTrainedModel", "_get_feat_extract_output_lengths", "_get_feature_vector_attention_mask", "arange", "attention_mask", "batch_size", "bool", "class", "cumsum", "def", "device", "dim", "dtype", "feature_vector_length", "flip", "int", "long", "non_padded_lengths", "output_lengths", "return", "self", "shape", "to", "torch", "zeros"], "unispeech_sat/modeling_unispeech_sat.py:_compute_mask_indices": ["False", "LongTensor", "None", "ValueError", "_", "_compute_mask_indices", "and", "append", "arange", "array", "attention_mask", "batch_size", "be", "bigger", "bool", "broadcast_to", "but", "choice", "compute_num_masked_span", "concatenate", "def", "detach", "dtype", "dummy_mask_idx", "else", "epsilon", "f", "float", "for", "got", "has", "if", "in", "input_length", "input_lengths", "int", "int32", "is", "item", "len", "mask_length", "mask_prob", "max", "max_num_masked_span", "min_masks", "ndarray", "not", "np", "num_masked_span", "offsets", "ones", "put_along_axis", "raise", "rand", "random", "range", "replace", "reshape", "return", "sequence_length", "shape", "smaller", "spec_aug_mask", "spec_aug_mask_idx", "spec_aug_mask_idxs", "sum", "than", "to", "tolist", "torch", "tuple", "zeros"], "unispeech_sat/modeling_unispeech_sat.py:UniSpeechSatModel.__init__": ["ModelConfig", "ModelEncoder", "ModelEncoderStableLayerNorm", "ModelFeatureEncoder", "ModelFeatureProjection", "ModelModel", "ModelPreTrainedModel", "Modelform_", "Parameter", "Tensor", "__init__", "class", "config", "def", "do_stable_layer_norm", "else", "encoder", "feature_extractor", "feature_projection", "hidden_size", "if", "masked_spec_embed", "nn", "post_init", "self", "super", "torch"], "unispeech_sat/modeling_unispeech_sat.py:UniSpeechSatModel._mask_hidden_states": ["FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "True", "_compute_mask_indices", "_mask_hidden_states", "and", "apply_spec_augment", "attention_mask", "batch_size", "bool", "class", "config", "def", "device", "dtype", "elif", "expand", "getattr", "hidden_size", "hidden_states", "if", "is", "mask_feature_indices", "mask_feature_length", "mask_feature_min_masks", "mask_feature_prob", "mask_length", "mask_prob", "mask_time_indices", "mask_time_length", "mask_time_min_masks", "mask_time_prob", "masked_spec_embed", "min_masks", "not", "return", "self", "sequence_length", "size", "tensor", "to", "torch", "training"], "unispeech_sat/modeling_unispeech_sat.py:UniSpeechSatModel.forward": ["FloatTensor", "ModelBaseModelOutput", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "_get_feature_vector_attention_mask", "_mask_hidden_states", "attention_mask", "attentions", "bool", "class", "config", "def", "else", "encoder", "encoder_outputs", "extract_features", "feature_extractor", "feature_projection", "forward", "hidden_states", "if", "input_values", "is", "kwargs", "last_hidden_state", "mask_time_indices", "not", "output_attentions", "output_hidden_states", "r", "return", "return_dict", "self", "shape", "torch", "transpose", "tuple", "use_return_dict"], "unispeech_sat/modeling_unispeech_sat.py:UniSpeechSatForPreTraining.__init__": ["Dropout", "False", "FloatTensor", "LayerNorm", "Linear", "Model", "ModelConfig", "ModelForPreTraining", "ModelGumbelVectorQuantizer", "ModelModel", "ModelPreTrainedModel", "Parameter", "__init__", "class", "codevector_dim", "config", "data", "def", "do_stable_layer_norm", "dropout", "dropout_features", "eps", "feat_quantizer_dropout", "final_dropout", "hidden_size", "if", "label_embeddings_concat", "layer_norm_eps", "layer_norm_for_extract", "nn", "num_clusters", "post_init", "proj_codevector_dim", "project_hid", "project_q", "quantizer", "requires_grad", "self", "speaker_proj", "super", "torch", "zero_"], "unispeech_sat/modeling_unispeech_sat.py:UniSpeechSatForPreTraining.set_gumbel_temperature": ["ModelForPreTraining", "ModelPreTrainedModel", "class", "def", "int", "quantizer", "self", "set_gumbel_temperature", "temperature"], "unispeech_sat/modeling_unispeech_sat.py:UniSpeechSatForPreTraining.freeze_feature_encoder": ["Model", "ModelForPreTraining", "ModelPreTrainedModel", "_freeze_parameters", "class", "def", "feature_extractor", "freeze_feature_encoder", "self"], "unispeech_sat/modeling_unispeech_sat.py:UniSpeechSatForPreTraining.compute_contrastive_logits": ["FloatTensor", "ModelForPreTraining", "ModelPreTrainedModel", "cat", "class", "compute_contrastive_logits", "cosine_similarity", "def", "dim", "float", "int", "logits", "negative_features", "predicted_features", "return", "target_features", "temperature", "torch", "type_as"], "unispeech_sat/modeling_unispeech_sat.py:UniSpeechSatForPreTraining.forward": ["Model", "ModelForPreTraining", "ModelForPreTrainingOutput", "ModelPreTrainedModel", "None", "Tensor", "attention_mask", "attentions", "bool", "class", "codevector_perplexity", "config", "def", "dropout_features", "else", "extract_features", "forward", "hidden_states", "if", "input_values", "is", "kwargs", "logits", "loss", "not", "output_attentions", "output_hidden_states", "outputs", "projected_quantized_states", "projected_states", "quantized_features", "r", "return", "return_dict", "self", "torch", "transformer_features", "tuple", "use_return_dict"], "unispeech_sat/modeling_unispeech_sat.py:UniSpeechSatForCTC.__init__": ["Dropout", "Linear", "Model", "ModelForCTC", "ModelModel", "ModelPreTrainedModel", "None", "Please", "ValueError", "You", "__class__", "__init__", "a", "add_adapter", "and", "are", "as", "class", "config", "configuration", "def", "define", "does", "dropout", "else", "f", "final_dropout", "follows", "from_pretrained", "hasattr", "head", "hidden_size", "if", "instantiate", "is", "language", "lm_head", "model", "nn", "not", "of", "or", "output_hidden_size", "post_init", "r", "raise", "s", "self", "size", "str", "super", "target_lang", "that", "the", "to", "trying", "vocab_size", "vocabulary", "with", "your"], "unispeech_sat/modeling_unispeech_sat.py:UniSpeechSatForCTC.tie_weights": ["By", "Cannot", "ModelForCTC", "ModelPreTrainedModel", "None", "True", "ValueError", "adapter_attn_dim", "and", "class", "config", "def", "default", "defined", "elif", "eng", "f", "force_load", "getattr", "if", "info", "is", "kwargs", "load_adapter", "logger", "not", "pass", "raise", "self", "set", "target_lang", "tie_weights", "to"], "unispeech_sat/modeling_unispeech_sat.py:UniSpeechSatForCTC.freeze_feature_encoder": ["Model", "ModelForCTC", "ModelPreTrainedModel", "_freeze_parameters", "class", "def", "feature_extractor", "freeze_feature_encoder", "self"], "unispeech_sat/modeling_unispeech_sat.py:UniSpeechSatForCTC.freeze_base_model": ["False", "Model", "ModelForCTC", "ModelPreTrainedModel", "class", "def", "for", "freeze_base_model", "in", "param", "parameters", "requires_grad", "self"], "unispeech_sat/modeling_unispeech_sat.py:UniSpeechSatForCTC.forward": ["CausalLMOutput", "False", "Label", "Model", "ModelForCTC", "ModelPreTrainedModel", "None", "Tensor", "ValueError", "_HIDDEN_STATES_START_POSITION", "_get_feat_extract_output_lengths", "and", "attention_mask", "attentions", "backends", "be", "blank", "bool", "class", "config", "ctc_loss", "ctc_loss_reduction", "ctc_zero_infinity", "cudnn", "def", "dim", "dropout", "dtype", "else", "enabled", "f", "flags", "flattened_targets", "float32", "forward", "functional", "hidden_states", "if", "input_lengths", "input_values", "is", "kwargs", "labels", "labels_mask", "lm_head", "log_probs", "log_softmax", "logits", "long", "loss", "masked_select", "max", "must", "nn", "not", "ones_like", "output", "output_attentions", "output_hidden_states", "outputs", "pad_token_id", "r", "raise", "reduction", "return", "return_dict", "self", "sum", "target_lengths", "to", "torch", "transpose", "tuple", "use_return_dict", "values", "vocab_size", "with", "zero_infinity"], "unispeech_sat/modeling_unispeech_sat.py:UniSpeechSatForSequenceClassification.__init__": ["Linear", "Model", "ModelForSequenceClassification", "ModelModel", "ModelPreTrainedModel", "Parameter", "Sequence", "True", "ValueError", "__init__", "adapters", "add_adapter", "and", "class", "classification", "classifier", "classifier_proj_size", "config", "def", "does", "hasattr", "hidden_size", "if", "layer_weights", "nn", "not", "num_hidden_layers", "num_labels", "num_layers", "of", "ones", "post_init", "projector", "raise", "self", "super", "support", "the", "torch", "use", "use_weighted_layer_sum"], "unispeech_sat/modeling_unispeech_sat.py:UniSpeechSatForSequenceClassification.freeze_feature_encoder": ["Model", "ModelForSequenceClassification", "ModelPreTrainedModel", "_freeze_parameters", "class", "def", "feature_extractor", "freeze_feature_encoder", "self"], "unispeech_sat/modeling_unispeech_sat.py:UniSpeechSatForSequenceClassification.freeze_base_model": ["False", "Model", "ModelForSequenceClassification", "ModelPreTrainedModel", "class", "def", "for", "freeze_base_model", "in", "param", "parameters", "requires_grad", "self"], "unispeech_sat/modeling_unispeech_sat.py:UniSpeechSatForSequenceClassification.forward": ["CrossEntropyLoss", "Model", "ModelForSequenceClassification", "ModelPreTrainedModel", "None", "SequenceClassifierOutput", "Tensor", "True", "_HIDDEN_STATES_START_POSITION", "_get_feature_vector_attention_mask", "attention_mask", "attentions", "bool", "class", "classifier", "config", "def", "dim", "else", "expand_padding_mask", "forward", "functional", "hidden_states", "if", "input_values", "is", "kwargs", "labels", "layer_weights", "logits", "loss", "loss_fct", "mean", "nn", "norm_weights", "not", "num_labels", "output", "output_attentions", "output_hidden_states", "outputs", "padding_mask", "pooled_output", "projector", "r", "repeat", "return", "return_dict", "self", "shape", "softmax", "stack", "sum", "torch", "tuple", "unsqueeze", "use_return_dict", "use_weighted_layer_sum", "view"], "unispeech_sat/modeling_unispeech_sat.py:UniSpeechSatForAudioFrameClassification.__init__": ["Audio", "Linear", "Model", "ModelForAudioFrameClassification", "ModelModel", "ModelPreTrainedModel", "Parameter", "True", "ValueError", "__init__", "adapters", "add_adapter", "and", "class", "classification", "classifier", "config", "def", "does", "frame", "hasattr", "hidden_size", "if", "layer_weights", "nn", "not", "num_hidden_layers", "num_labels", "num_layers", "of", "ones", "post_init", "raise", "self", "super", "support", "the", "torch", "use", "use_weighted_layer_sum"], "unispeech_sat/modeling_unispeech_sat.py:UniSpeechSatForAudioFrameClassification.freeze_feature_encoder": ["Model", "ModelForAudioFrameClassification", "ModelPreTrainedModel", "_freeze_parameters", "class", "def", "feature_extractor", "freeze_feature_encoder", "self"], "unispeech_sat/modeling_unispeech_sat.py:UniSpeechSatForAudioFrameClassification.freeze_base_model": ["False", "Model", "ModelForAudioFrameClassification", "ModelPreTrainedModel", "class", "def", "for", "freeze_base_model", "in", "param", "parameters", "requires_grad", "self"], "unispeech_sat/modeling_unispeech_sat.py:UniSpeechSatForAudioFrameClassification.forward": ["CrossEntropyLoss", "Model", "ModelForAudioFrameClassification", "ModelPreTrainedModel", "None", "Tensor", "TokenClassifierOutput", "True", "_HIDDEN_STATES_START_POSITION", "argmax", "attention_mask", "attentions", "axis", "bool", "class", "classifier", "config", "def", "dim", "else", "forward", "functional", "hidden_states", "if", "input_values", "is", "kwargs", "labels", "layer_weights", "logits", "loss", "loss_fct", "nn", "norm_weights", "not", "num_labels", "output", "output_attentions", "output_hidden_states", "outputs", "r", "return", "return_dict", "self", "softmax", "stack", "sum", "torch", "tuple", "use_return_dict", "use_weighted_layer_sum", "view"], "unispeech_sat/modeling_unispeech_sat.py:AMSoftmaxLoss.__init__": ["CrossEntropyLoss", "Model", "Module", "Parameter", "True", "__init__", "class", "def", "input_dim", "loss", "margin", "nn", "num_labels", "randn", "requires_grad", "scale", "self", "super", "torch", "weight"], "unispeech_sat/modeling_unispeech_sat.py:AMSoftmaxLoss.forward": ["Model", "Module", "bool", "class", "cos_theta", "def", "dim", "flatten", "forward", "functional", "hidden_states", "labels", "logits", "loss", "margin", "mm", "nn", "normalize", "num_labels", "one_hot", "onehot", "psi", "return", "scale", "self", "torch", "weight", "where"], "unispeech_sat/modeling_unispeech_sat.py:TDNNLayer.__init__": ["Linear", "Model", "Module", "ReLU", "__init__", "activation", "class", "config", "def", "dilation", "else", "if", "in_conv_dim", "kernel", "kernel_size", "layer_id", "nn", "out_conv_dim", "self", "super", "tdnn_dilation", "tdnn_dim", "tdnn_kernel"], "unispeech_sat/modeling_unispeech_sat.py:TDNNLayer.forward": ["Detected", "LoRA", "LoraLayer", "Model", "Module", "Tensor", "You", "activation", "applied", "be", "bias", "class", "conv1d", "def", "dilation", "due", "exclude", "forward", "from", "functional", "hidden_states", "if", "in_conv_dim", "is_peft_available", "isinstance", "kernel", "kernel_size", "modules", "nn", "on", "optimization", "out_conv_dim", "return", "s", "self", "should", "t", "target", "to", "torch", "transpose", "view", "warn", "warnings", "weight", "weights", "won"], "unispeech_sat/modeling_unispeech_sat.py:UniSpeechSatForXVector.__init__": ["AMSoftmaxLoss", "Linear", "Model", "ModelForXVector", "ModelModel", "ModelPreTrainedModel", "ModuleList", "Parameter", "TDNNLayer", "__init__", "class", "classifier", "config", "def", "feature_extractor", "for", "hidden_size", "i", "if", "in", "layer_weights", "len", "nn", "num_hidden_layers", "num_labels", "num_layers", "objective", "ones", "post_init", "projector", "range", "self", "super", "tdnn", "tdnn_dim", "tdnn_layers", "torch", "use_weighted_layer_sum", "xvector_output_dim"], "unispeech_sat/modeling_unispeech_sat.py:UniSpeechSatForXVector.freeze_feature_encoder": ["Model", "ModelForXVector", "ModelPreTrainedModel", "_freeze_parameters", "class", "def", "feature_extractor", "freeze_feature_encoder", "self"], "unispeech_sat/modeling_unispeech_sat.py:UniSpeechSatForXVector.freeze_base_model": ["False", "Model", "ModelForXVector", "ModelPreTrainedModel", "class", "def", "for", "freeze_base_model", "in", "param", "parameters", "requires_grad", "self"], "unispeech_sat/modeling_unispeech_sat.py:UniSpeechSatForXVector._get_tdnn_output_lengths": ["LongTensor", "ModelForXVector", "ModelPreTrainedModel", "_conv_out_length", "_get_tdnn_output_lengths", "class", "config", "def", "for", "in", "input_length", "input_lengths", "int", "kernel_size", "return", "self", "stride", "tdnn_kernel", "torch"], "unispeech_sat/modeling_unispeech_sat.py:UniSpeechSatForXVector.forward": ["Model", "ModelForXVector", "ModelPreTrainedModel", "None", "Tensor", "True", "XVectorOutput", "_HIDDEN_STATES_START_POSITION", "_get_feat_extract_output_lengths", "_get_tdnn_output_lengths", "append", "attention_mask", "attentions", "bool", "cat", "class", "classifier", "config", "def", "dim", "else", "embeddings", "enumerate", "feat_extract_output_lengths", "feature_extractor", "for", "forward", "functional", "hidden_states", "i", "if", "in", "input_values", "is", "kwargs", "labels", "layer_weights", "length", "logits", "loss", "mean", "mean_features", "nn", "norm_weights", "not", "objective", "output", "output_attentions", "output_embeddings", "output_hidden_states", "outputs", "projector", "r", "return", "return_dict", "self", "softmax", "stack", "statistic_pooling", "std", "std_features", "sum", "tdnn", "tdnn_layer", "tdnn_output_lengths", "torch", "tuple", "use_return_dict", "use_weighted_layer_sum", "view"], "univnet/modeling_univnet.py:UnivNetKernelPredictorResidualBlock.__init__": ["Conv1d", "Dropout", "ModelConfig", "ModelKernelPredictorResidualBlock", "Module", "True", "__init__", "bias", "channels", "class", "config", "conv1", "conv2", "def", "dropout", "dropout_prob", "kernel_predictor_conv_size", "kernel_predictor_dropout", "kernel_size", "leaky_relu_slope", "model_in_channels", "nn", "padding", "self", "super"], "univnet/modeling_univnet.py:UnivNetKernelPredictorResidualBlock.forward": ["FloatTensor", "ModelKernelPredictorResidualBlock", "Module", "class", "conv1", "conv2", "def", "dropout", "forward", "functional", "hidden_states", "leaky_relu", "leaky_relu_slope", "nn", "residual", "return", "self", "torch"], "univnet/modeling_univnet.py:UnivNetKernelPredictorResidualBlock.apply_weight_norm": ["ModelKernelPredictorResidualBlock", "Module", "apply_weight_norm", "class", "conv1", "conv2", "def", "hasattr", "if", "nn", "parametrizations", "self", "utils", "weight_norm"], "univnet/modeling_univnet.py:UnivNetKernelPredictorResidualBlock.remove_weight_norm": ["ModelKernelPredictorResidualBlock", "Module", "class", "conv1", "conv2", "def", "nn", "remove_weight_norm", "self", "utils"], "univnet/modeling_univnet.py:UnivNetKernelPredictor.__init__": ["Conv1d", "ModelConfig", "ModelKernelPredictor", "ModelKernelPredictorResidualBlock", "Module", "ModuleList", "True", "_", "__init__", "bias", "bias_channels", "bias_conv", "class", "config", "conv_in_channels", "conv_kernel_size", "conv_layers", "conv_out_channels", "def", "for", "in", "input_conv", "int", "kernel_channels", "kernel_conv", "kernel_predictor_conv_size", "kernel_predictor_hidden_channels", "kernel_predictor_num_blocks", "leaky_relu_slope", "model_hidden_channels", "nn", "num_blocks", "num_mel_bins", "padding", "range", "resblocks", "resnet_hidden_channels", "resnet_in_channels", "resnet_kernel_size", "self", "super"], "univnet/modeling_univnet.py:UnivNetKernelPredictor.forward": ["FloatTensor", "ModelKernelPredictor", "Module", "_", "batch_size", "bias_conv", "bias_hidden_states", "biases", "class", "contiguous", "conv_in_channels", "conv_kernel_size", "conv_layers", "conv_out_channels", "def", "for", "forward", "functional", "hidden_states", "in", "input_conv", "kernel_conv", "kernel_hidden_states", "kernels", "leaky_relu", "leaky_relu_slope", "nn", "resblock", "resblocks", "return", "self", "seq_length", "shape", "spectrogram", "torch", "view"], "univnet/modeling_univnet.py:UnivNetKernelPredictor.apply_weight_norm": ["ModelKernelPredictor", "Module", "apply_weight_norm", "bias_conv", "class", "def", "for", "hasattr", "if", "in", "input_conv", "kernel_conv", "layer", "nn", "parametrizations", "resblocks", "self", "utils", "weight_norm"], "univnet/modeling_univnet.py:UnivNetKernelPredictor.remove_weight_norm": ["ModelKernelPredictor", "Module", "bias_conv", "class", "def", "for", "in", "input_conv", "kernel_conv", "layer", "nn", "remove_weight_norm", "resblocks", "self", "utils"], "univnet/modeling_univnet.py:UnivNetLvcResidualBlock.__init__": ["Conv1d", "ModelConfig", "ModelLvcResidualBlock", "Module", "__init__", "class", "config", "conv", "def", "dilation", "hidden_channels", "int", "kernel_size", "leaky_relu_slope", "model_hidden_channels", "nn", "padding", "self", "super"], "univnet/modeling_univnet.py:UnivNetLvcResidualBlock.forward": ["ModelLvcResidualBlock", "Module", "bias", "class", "conv", "def", "forward", "functional", "hidden_channels", "hidden_states", "hop_size", "kernel", "leaky_relu", "leaky_relu_slope", "location_variable_convolution", "nn", "residual", "return", "self", "sigmoid", "tanh", "torch"], "univnet/modeling_univnet.py:UnivNetLvcResidualBlock.location_variable_convolution": ["Dim", "FloatTensor", "ModelLvcResidualBlock", "Module", "Please", "ValueError", "_", "and", "are", "batch", "be", "bias", "bildsk", "biokl", "bolsd", "but", "channels_last_3d", "check", "class", "constant", "contiguous", "correct", "def", "dilation", "einsum", "f", "functional", "got", "hidden_states", "hop_size", "if", "in_length", "int", "kernel", "kernel_length", "kernel_size", "location_variable_convolution", "make", "memory_format", "nn", "of", "or", "out_channels", "output_hidden_states", "pad", "padding", "raise", "return", "self", "shape", "should", "sure", "they", "to", "torch", "transpose", "unfold", "unsqueeze", "view"], "univnet/modeling_univnet.py:UnivNetLvcResidualBlock.apply_weight_norm": ["ModelLvcResidualBlock", "Module", "apply_weight_norm", "class", "conv", "def", "hasattr", "if", "nn", "parametrizations", "self", "utils", "weight_norm"], "univnet/modeling_univnet.py:UnivNetLvcResidualBlock.remove_weight_norm": ["ModelLvcResidualBlock", "Module", "class", "conv", "def", "nn", "remove_weight_norm", "self", "utils"], "univnet/modeling_univnet.py:UnivNetLvcBlock.__init__": ["ConvTranspose1d", "ModelConfig", "ModelKernelPredictor", "ModelLvcBlock", "ModelLvcResidualBlock", "Module", "ModuleList", "__init__", "class", "cond_hop_length", "config", "convt_pre", "def", "dilations", "for", "hidden_channels", "i", "in", "int", "kernel_predictor", "kernel_size", "layer_id", "leaky_relu_slope", "len", "lvc_hop_size", "model_hidden_channels", "nn", "num_blocks", "output_padding", "padding", "range", "resblock_dilation_sizes", "resblock_kernel_sizes", "resblock_stride_sizes", "resblocks", "self", "stride", "super"], "univnet/modeling_univnet.py:UnivNetLvcBlock.forward": ["FloatTensor", "ModelLvcBlock", "Module", "bias", "biases", "class", "cond_hop_length", "convt_pre", "def", "enumerate", "for", "forward", "functional", "hidden_states", "hop_size", "i", "in", "kernel", "kernel_predictor", "kernels", "leaky_relu", "leaky_relu_slope", "nn", "resblock", "resblocks", "return", "self", "spectrogram", "torch"], "univnet/modeling_univnet.py:UnivNetLvcBlock.apply_weight_norm": ["ModelLvcBlock", "Module", "apply_weight_norm", "class", "convt_pre", "def", "for", "hasattr", "if", "in", "kernel_predictor", "layer", "nn", "parametrizations", "resblocks", "self", "utils", "weight_norm"], "univnet/modeling_univnet.py:UnivNetLvcBlock.remove_weight_norm": ["ModelLvcBlock", "Module", "class", "convt_pre", "def", "for", "in", "kernel_predictor", "layer", "nn", "remove_weight_norm", "resblocks", "self", "utils"], "univnet/modeling_univnet.py:UnivNetModel.__init__": ["Conv1d", "ModelConfig", "ModelLvcBlock", "ModelModel", "ModuleList", "PreTrainedModel", "__init__", "append", "class", "config", "conv_post", "conv_pre", "def", "for", "hop_length", "hop_lengths", "i", "in", "kernel_size", "layer_id", "leaky_relu_slope", "len", "lvc_hop_size", "model_hidden_channels", "model_in_channels", "nn", "num_kernels", "num_layers", "padding", "padding_mode", "post_init", "range", "reflect", "resblock_kernel_sizes", "resblock_stride_sizes", "resblocks", "self", "stride", "super"], "univnet/modeling_univnet.py:UnivNetModel.forward": ["FloatTensor", "Generator", "ModelModel", "ModelModelOutput", "None", "PreTrainedModel", "The", "ValueError", "_", "and", "are", "batch", "be", "bool", "but", "class", "config", "conv_post", "conv_pre", "def", "device", "dim", "dtype", "elif", "else", "equal", "expected", "f", "for", "forward", "functional", "generator", "hidden_states", "if", "in", "input_features", "is", "kwargs", "leaky_relu", "leaky_relu_slope", "model_in_channels", "nn", "noise_sequence", "noise_sequence_batch_size", "noise_sequence_batched", "noise_sequence_shape", "not", "of", "outputs", "padding_mask", "padding_mask_batch_size", "r", "raise", "randn", "repeat", "resblock", "resblocks", "return", "return_dict", "self", "shape", "size", "spectrogram_batch_size", "spectrogram_batched", "spectrogram_length", "squeeze", "sum", "tanh", "the", "to", "torch", "transpose", "tuple", "two", "unsqueeze", "use_return_dict", "waveform", "waveform_lengths", "waveforms"], "univnet/modeling_univnet.py:UnivNetModel.apply_weight_norm": ["ModelModel", "PreTrainedModel", "apply_weight_norm", "class", "conv_post", "conv_pre", "def", "for", "hasattr", "if", "in", "layer", "nn", "parametrizations", "resblocks", "self", "utils", "weight_norm"], "univnet/modeling_univnet.py:UnivNetModel.remove_weight_norm": ["ModelModel", "PreTrainedModel", "class", "conv_post", "conv_pre", "def", "for", "in", "layer", "nn", "remove_weight_norm", "resblocks", "self", "utils"], "upernet/modeling_upernet.py:UperNetConvModule.__init__": ["BatchNorm2d", "Conv2d", "False", "ModelConvModule", "Module", "None", "ReLU", "__init__", "activation", "batch_norm", "bias", "bool", "class", "conv", "def", "dilation", "in_channels", "int", "kernel_size", "nn", "out_channels", "padding", "sModel", "self", "str", "tuple"], "upernet/modeling_upernet.py:UperNetConvModule.forward": ["ModelConvModule", "Module", "Tensor", "activation", "batch_norm", "class", "conv", "def", "forward", "input", "nn", "output", "return", "self", "torch"], "upernet/modeling_upernet.py:UperNetPyramidPoolingBlock.__init__": ["AdaptiveAvgPool2d", "ModelConvModule", "ModelPyramidPoolingBlock", "Module", "None", "__init__", "add_module", "channels", "class", "def", "enumerate", "for", "i", "in", "in_channels", "int", "kernel_size", "layer", "layers", "nn", "pool_scale", "sModel", "self", "str"], "upernet/modeling_upernet.py:UperNetPyramidPoolingBlock.forward": ["ModelPyramidPoolingBlock", "Module", "Tensor", "class", "def", "for", "forward", "hidden_state", "in", "input", "layer", "layers", "nn", "return", "self", "torch"], "upernet/modeling_upernet.py:UperNetPyramidPoolingModule.__init__": ["ModelPyramidPoolingBlock", "ModelPyramidPoolingModule", "Module", "None", "__init__", "add_module", "align_corners", "append", "block", "blocks", "bool", "channels", "class", "def", "enumerate", "for", "i", "in", "in_channels", "int", "nn", "pool_scale", "pool_scales", "sModel", "self", "str", "tuple"], "upernet/modeling_upernet.py:UperNetPyramidPoolingModule.forward": ["ModelPyramidPoolingModule", "Module", "Tensor", "align_corners", "append", "bilinear", "blocks", "class", "def", "for", "forward", "functional", "in", "interpolate", "list", "mode", "nn", "ppm", "ppm_out", "ppm_outs", "return", "self", "size", "torch", "upsampled_ppm_out", "x"], "upernet/modeling_upernet.py:UperNetHead.__init__": ["Conv2d", "False", "ModelConvModule", "ModelHead", "ModelPyramidPoolingModule", "Module", "ModuleList", "__init__", "align_corners", "append", "bottleneck", "channels", "class", "classifier", "config", "def", "for", "fpn_bottleneck", "fpn_conv", "fpn_convs", "hidden_size", "in", "in_channels", "kernel_size", "l_conv", "lateral_convs", "len", "nn", "num_labels", "padding", "pool_scales", "psp_modules", "sModel", "self"], "upernet/modeling_upernet.py:UperNetHead.psp_forward": ["ModelHead", "Module", "bottleneck", "cat", "class", "def", "dim", "extend", "inputs", "nn", "output", "psp_forward", "psp_modules", "psp_outs", "return", "self", "torch", "x"], "upernet/modeling_upernet.py:UperNetHead.forward": ["ModelHead", "Module", "Tensor", "align_corners", "append", "bilinear", "cat", "class", "classifier", "def", "dim", "encoder_hidden_states", "enumerate", "for", "forward", "fpn_bottleneck", "fpn_convs", "fpn_outs", "functional", "i", "in", "interpolate", "lateral_conv", "lateral_convs", "laterals", "len", "mode", "nn", "output", "prev_shape", "psp_forward", "range", "return", "self", "shape", "size", "torch", "used_backbone_levels"], "upernet/modeling_upernet.py:UperNetFCNHead.__init__": ["Conv2d", "Identity", "ModelConvModule", "ModelFCNHead", "Module", "None", "Sequential", "__init__", "append", "auxiliary_channels", "auxiliary_concat_input", "auxiliary_in_channels", "auxiliary_num_convs", "channels", "class", "classifier", "concat_input", "config", "conv_cat", "conv_padding", "convs", "def", "dilation", "else", "for", "i", "if", "in", "in_channels", "in_index", "int", "is", "kernel_size", "nn", "num_convs", "num_labels", "padding", "range", "sModel", "self", "tuple"], "upernet/modeling_upernet.py:UperNetFCNHead.forward": ["ModelFCNHead", "Module", "Tensor", "cat", "class", "classifier", "concat_input", "conv_cat", "convs", "def", "dim", "encoder_hidden_states", "forward", "hidden_states", "if", "in_index", "nn", "output", "return", "self", "torch"], "upernet/modeling_upernet.py:UperNetForSemanticSegmentation.__init__": ["ModelFCNHead", "ModelForSemanticSegmentation", "ModelHead", "ModelPreTrainedModel", "None", "__init__", "auxiliary_head", "backbone", "channels", "class", "config", "decode_head", "def", "else", "if", "in_channels", "load_backbone", "post_init", "sModel", "self", "use_auxiliary_head"], "upernet/modeling_upernet.py:UperNetForSemanticSegmentation.forward": ["CrossEntropyLoss", "False", "ModelForSemanticSegmentation", "ModelPreTrainedModel", "None", "SemanticSegmenterOutput", "Tensor", "The", "ValueError", "align_corners", "and", "attentions", "auxiliary_head", "auxiliary_logits", "auxiliary_loss", "auxiliary_loss_weight", "backbone", "be", "bilinear", "bool", "class", "config", "decode_head", "def", "else", "feature_maps", "features", "forward", "forward_with_filtered_kwargs", "functional", "greater", "hidden_states", "if", "ignore_index", "interpolate", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "loss_ignore_index", "mode", "nn", "not", "num_labels", "number", "of", "one", "output", "output_attentions", "output_hidden_states", "outputs", "pixel_values", "r", "raise", "return", "return_dict", "self", "shape", "should", "size", "than", "torch", "tuple", "use_return_dict"], "vaultgemma/modeling_vaultgemma.py:VaultGemmaRMSNorm.__init__": ["ModelRMSNorm", "Module", "Parameter", "__init__", "class", "def", "dim", "eps", "float", "int", "nn", "self", "super", "torch", "weight", "zeros"], "vaultgemma/modeling_vaultgemma.py:VaultGemmaRMSNorm._norm": ["ModelRMSNorm", "Module", "True", "_norm", "class", "def", "eps", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "torch", "x"], "vaultgemma/modeling_vaultgemma.py:VaultGemmaRMSNorm.forward": ["ModelRMSNorm", "Module", "_norm", "class", "def", "float", "forward", "nn", "output", "return", "self", "type_as", "weight", "x"], "vaultgemma/modeling_vaultgemma.py:VaultGemmaRMSNorm.extra_repr": ["ModelRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "weight"], "vaultgemma/modeling_vaultgemma.py:VaultGemmaMLP.__init__": ["ACT2FN", "False", "Linear", "ModelMLP", "Module", "__init__", "act_fn", "bias", "class", "config", "def", "down_proj", "gate_proj", "hidden_activation", "hidden_size", "intermediate_size", "nn", "self", "super", "up_proj"], "vaultgemma/modeling_vaultgemma.py:VaultGemmaMLP.forward": ["ModelMLP", "Module", "act_fn", "class", "def", "down_proj", "forward", "gate_proj", "nn", "return", "self", "up_proj", "x"], "vaultgemma/modeling_vaultgemma.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "vaultgemma/modeling_vaultgemma.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cos", "def", "k", "k_embed", "q", "q_embed", "return", "rotate_half", "sin", "unsqueeze", "unsqueeze_dim"], "vaultgemma/modeling_vaultgemma.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "vaultgemma/modeling_vaultgemma.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "head_dim", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softcap", "softmax", "tanh", "to", "torch", "training", "transpose", "tuple", "value", "value_states"], "vaultgemma/modeling_vaultgemma.py:VaultGemmaAttention.__init__": ["Linear", "ModelAttention", "ModelConfig", "Module", "None", "True", "__init__", "attention_bias", "attention_dropout", "attn_logit_softcapping", "bias", "class", "config", "def", "else", "getattr", "hasattr", "head_dim", "hidden_size", "if", "int", "is_causal", "k_proj", "layer_idx", "layer_type", "layer_types", "nn", "num_attention_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "q_proj", "query_pre_attn_scalar", "scaling", "self", "sliding_attention", "sliding_window", "super", "v_proj"], "vaultgemma/modeling_vaultgemma.py:VaultGemmaAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "FlashAttentionKwargs", "LongTensor", "ModelAttention", "Module", "None", "Tensor", "Unpack", "_attn_implementation", "apply_rotary_pos_emb", "attention_dropout", "attention_interface", "attention_mask", "attn_logit_softcapping", "attn_output", "attn_weights", "cache_kwargs", "cache_position", "class", "config", "contiguous", "cos", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "head_dim", "hidden_shape", "hidden_states", "if", "input_shape", "is", "k_proj", "key_states", "kwargs", "layer_idx", "nn", "not", "o_proj", "past_key_values", "position_embeddings", "q_proj", "query_states", "reshape", "return", "scaling", "self", "shape", "sin", "sliding_window", "softcap", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "view"], "vaultgemma/modeling_vaultgemma.py:VaultGemmaDecoderLayer.__init__": ["GradientCheckpointingLayer", "ModelAttention", "ModelConfig", "ModelDecoderLayer", "ModelMLP", "ModelRMSNorm", "__init__", "attention_type", "class", "config", "def", "eps", "hidden_size", "input_layernorm", "int", "layer_idx", "layer_types", "mlp", "pre_feedforward_layernorm", "rms_norm_eps", "self", "self_attn", "super"], "vaultgemma/modeling_vaultgemma.py:VaultGemmaDecoderLayer.forward": ["Cache", "FloatTensor", "GradientCheckpointingLayer", "LongTensor", "ModelDecoderLayer", "None", "Tensor", "_", "attention_mask", "cache_position", "class", "def", "forward", "hidden_states", "input_layernorm", "kwargs", "mlp", "past_key_values", "position_embeddings", "position_ids", "pre_feedforward_layernorm", "residual", "return", "self", "self_attn", "torch", "tuple"], "vaultgemma/modeling_vaultgemma.py:VaultGemmaRotaryEmbedding.__init__": ["Callable", "False", "ModelConfig", "ModelRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "vaultgemma/modeling_vaultgemma.py:VaultGemmaRotaryEmbedding.compute_default_rope_parameters": ["ModelConfig", "ModelRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "vaultgemma/modeling_vaultgemma.py:VaultGemmaRotaryEmbedding.forward": ["False", "ModelRotaryEmbedding", "Module", "None", "and", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "vaultgemma/modeling_vaultgemma.py:VaultGemmaPreTrainedModel._init_weights": ["ModelPreTrainedModel", "PreTrainedModel", "RMSNorm", "__class__", "__name__", "_init_weights", "class", "def", "if", "in", "init", "module", "self", "super", "weight", "zeros_"], "vaultgemma/modeling_vaultgemma.py:VaultGemmaModel.__init__": ["Embedding", "False", "ModelConfig", "ModelDecoderLayer", "ModelModel", "ModelPreTrainedModel", "ModelRMSNorm", "ModelRotaryEmbedding", "ModuleList", "__init__", "class", "config", "def", "embed_tokens", "eps", "for", "gradient_checkpointing", "hidden_size", "in", "layer_idx", "layers", "nn", "norm", "num_hidden_layers", "pad_token_id", "padding_idx", "post_init", "range", "rms_norm_eps", "rotary_emb", "self", "super", "vocab_size"], "vaultgemma/modeling_vaultgemma.py:VaultGemmaModel.forward": ["BaseModelOutputWithPast", "Cache", "DynamicCache", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "and", "arange", "attention_mask", "attention_type", "bool", "cache_position", "causal_mask_mapping", "class", "config", "create_causal_mask", "create_sliding_window_causal_mask", "decoder_layer", "def", "device", "dict", "dtype", "else", "embed_tokens", "exactly", "for", "forward", "full_attention", "get_seq_length", "hidden_size", "hidden_states", "if", "in", "input_embeds", "input_ids", "inputs_embeds", "is", "isinstance", "kwargs", "last_hidden_state", "layers", "mask_kwargs", "must", "norm", "normalizer", "not", "num_hidden_layers", "of", "one", "or", "past_key_values", "past_seen_tokens", "position_embeddings", "position_ids", "raise", "return", "rotary_emb", "self", "shape", "sliding_attention", "specify", "tensor", "torch", "unsqueeze", "use_cache"], "vaultgemma/modeling_vaultgemma.py:VaultGemmaForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "ModelForCausalLM", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "vocab_size"], "vaultgemma/modeling_vaultgemma.py:VaultGemmaForCausalLM.forward": ["BaseModelOutputWithPast", "Cache", "CausalLMOutputWithPast", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForCausalLM", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "else", "final_logit_softcapping", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "outputs", "past_key_values", "position_ids", "r", "return", "self", "slice", "slice_indices", "tanh", "torch", "use_cache", "vocab_size"], "video_llama_3/modeling_video_llama_3.py:VideoLlama3VisionRotaryEmbedding.__init__": ["False", "ModelVisionRotaryEmbedding", "Module", "None", "__init__", "arange", "class", "def", "dim", "dtype", "float", "int", "inv_freq", "nn", "persistent", "register_buffer", "self", "super", "theta", "torch"], "video_llama_3/modeling_video_llama_3.py:VideoLlama3VisionRotaryEmbedding.forward": ["ModelVisionRotaryEmbedding", "Module", "Tensor", "append", "arange", "cat", "class", "cos", "def", "device", "dim", "dtype", "emb", "expand", "flatten", "for", "forward", "grid_thw", "h", "hpos_ids", "in", "inv_freq", "max", "max_grid_thw", "merge_size", "merge_sizes", "nn", "outer", "permute", "pos_ids", "repeat", "reshape", "return", "rotary_pos_emb", "rotary_pos_emb_full", "self", "seq", "sin", "stack", "t", "torch", "tuple", "unsqueeze", "w", "wpos_ids", "zip"], "video_llama_3/modeling_video_llama_3.py:VideoLlama3VisionEmbeddings.__init__": ["Conv2d", "ModelVisionConfig", "ModelVisionEmbeddings", "Module", "None", "__init__", "class", "config", "def", "embed_dim", "hidden_size", "in_channels", "kernel_size", "nn", "num_channels", "out_channels", "padding", "patch_embedding", "patch_size", "self", "stride", "super", "valid"], "video_llama_3/modeling_video_llama_3.py:VideoLlama3VisionEmbeddings.forward": ["ModelVisionEmbeddings", "Module", "Tensor", "class", "config", "def", "embed_dim", "embeddings", "forward", "hidden_states", "nn", "num_channels", "patch_embedding", "patch_embeds", "patch_size", "return", "self", "torch", "view"], "video_llama_3/modeling_video_llama_3.py:VideoLlama3VisionMLP.__init__": ["ACT2FN", "Linear", "ModelVisionMLP", "Module", "__init__", "activation_fn", "class", "config", "def", "fc1", "fc2", "hidden_act", "hidden_size", "intermediate_size", "nn", "self", "super"], "video_llama_3/modeling_video_llama_3.py:VideoLlama3VisionMLP.forward": ["ModelVisionMLP", "Module", "Tensor", "activation_fn", "class", "def", "fc1", "fc2", "forward", "hidden_states", "nn", "return", "self", "torch"], "video_llama_3/modeling_video_llama_3.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value", "value_states"], "video_llama_3/modeling_video_llama_3.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "video_llama_3/modeling_video_llama_3.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "video_llama_3/modeling_video_llama_3.py:apply_rotary_pos_emb_vision": ["Model_rotary_pos_emb_vision", "Tensor", "cos", "def", "dtype", "float", "k", "k_embed", "orig_k_dtype", "orig_q_dtype", "q", "q_embed", "return", "rotate_half", "sin", "to", "torch", "tuple", "unsqueeze"], "video_llama_3/modeling_video_llama_3.py:VideoLlama3VisionAttention.__init__": ["False", "Linear", "ModelVisionAttention", "Module", "ValueError", "__init__", "and", "attention_dropout", "be", "by", "class", "config", "def", "divisible", "embed_dim", "f", "got", "head_dim", "hidden_size", "if", "is_causal", "k_proj", "must", "nn", "num_attention_heads", "num_heads", "num_key_value_groups", "out_proj", "q_proj", "raise", "scaling", "self", "super", "v_proj"], "video_llama_3/modeling_video_llama_3.py:VideoLlama3VisionAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "False", "ModelVisionAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_attn_implementation", "append", "apply_rotary_pos_emb_vision", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_outputs", "attn_weight", "attn_weights", "cat", "class", "config", "contiguous", "cos", "cu_seq_lens_k", "cu_seq_lens_q", "cu_seqlens", "def", "dim", "dropout", "eager", "eager_attention_forward", "else", "flash_attention_2", "for", "forward", "head_dim", "hidden_states", "if", "in", "is_causal", "k", "k_proj", "key_states", "kwargs", "lengths", "max", "max_length_k", "max_length_q", "max_seqlen", "nn", "not", "num_heads", "out_proj", "position_embeddings", "q", "q_proj", "query_states", "reshape", "return", "scaling", "self", "seq_length", "shape", "sin", "split", "splits", "tensor", "tolist", "torch", "training", "transpose", "tuple", "unsqueeze", "v", "v_proj", "value_states", "view", "zip"], "video_llama_3/modeling_video_llama_3.py:VideoLlama3VisionEncoderLayer.__init__": ["GradientCheckpointingLayer", "LayerNorm", "ModelVisionAttention", "ModelVisionConfig", "ModelVisionEncoderLayer", "ModelVisionMLP", "__init__", "class", "config", "def", "embed_dim", "eps", "hidden_size", "layer_norm1", "layer_norm2", "layer_norm_eps", "mlp", "nn", "self", "self_attn", "super"], "video_llama_3/modeling_video_llama_3.py:VideoLlama3VisionEncoderLayer.forward": ["GradientCheckpointingLayer", "ModelVisionEncoderLayer", "Tensor", "TransformersKwargs", "Unpack", "_", "class", "cu_seqlens", "def", "forward", "hidden_states", "kwargs", "layer_norm1", "layer_norm2", "mlp", "position_embeddings", "r", "residual", "return", "self", "self_attn", "torch", "tuple"], "video_llama_3/modeling_video_llama_3.py:VideoLlama3VisionEncoder.__init__": ["False", "ModelVisionConfig", "ModelVisionEncoder", "ModelVisionEncoderLayer", "Module", "ModuleList", "_", "__init__", "class", "config", "def", "for", "gradient_checkpointing", "in", "layers", "nn", "num_hidden_layers", "range", "self", "super"], "video_llama_3/modeling_video_llama_3.py:VideoLlama3VisionEncoder.forward": ["BaseModelOutput", "ModelVisionEncoder", "Module", "Tensor", "TransformersKwargs", "Unpack", "class", "cu_seqlens", "def", "encoder_layer", "for", "forward", "hidden_states", "in", "kwargs", "last_hidden_state", "layers", "nn", "position_embeddings", "r", "return", "self", "torch", "tuple"], "video_llama_3/modeling_video_llama_3.py:VideoLlama3PreTrainedModel._init_weights": ["ModelPreTrainedModel", "ModelVisionRotaryEmbedding", "PreTrainedModel", "_init_weights", "arange", "class", "copy_", "def", "dim", "dtype", "float", "if", "init", "inv_freq", "isinstance", "module", "self", "super", "theta", "torch"], "video_llama_3/modeling_video_llama_3.py:VideoLlama3VisionModel.__init__": ["LayerNorm", "ModelPreTrainedModel", "ModelVisionConfig", "ModelVisionEmbeddings", "ModelVisionEncoder", "ModelVisionModel", "ModelVisionRotaryEmbedding", "__init__", "class", "config", "def", "embeddings", "encoder", "eps", "head_dim", "hidden_size", "layer_norm_eps", "num_attention_heads", "post_init", "post_layernorm", "rotary_pos_emb", "self", "super"], "video_llama_3/modeling_video_llama_3.py:VideoLlama3VisionModel.get_input_embeddings": ["ModelPreTrainedModel", "ModelVisionEmbeddings", "ModelVisionModel", "class", "def", "embeddings", "get_input_embeddings", "patch_embedding", "return", "self"], "video_llama_3/modeling_video_llama_3.py:VideoLlama3VisionModel.pixel_unshuffle": ["ModelPreTrainedModel", "ModelVisionModel", "Tensor", "append", "bilinear", "c", "cat", "class", "def", "dim", "for", "functional", "grid_thw", "h", "hidden_states", "hidden_states_chunks", "in", "interpolate", "merge_size", "merge_sizes", "mode", "nn", "outputs", "permute", "pixel_unshuffle", "prod", "reshape", "return", "self", "shape", "size", "split", "t", "tolist", "torch", "view", "w", "zip"], "video_llama_3/modeling_video_llama_3.py:VideoLlama3VisionModel.forward": ["BaseModelOutput", "ModelPreTrainedModel", "ModelVisionModel", "Tensor", "TransformersKwargs", "Unpack", "class", "cu_seqlens", "cumsum", "def", "dim", "dtype", "else", "embeddings", "encoder", "encoder_outputs", "forward", "functional", "grid_thw", "hidden_states", "if", "int32", "is_tracing", "jit", "kwargs", "last_hidden_state", "merge_sizes", "nn", "pad", "pixel_unshuffle", "pixel_values", "position_embeddings", "post_layernorm", "r", "repeat_interleave", "return", "rotary_pos_emb", "self", "torch", "tuple", "type", "value"], "video_llama_3/modeling_video_llama_3.py:VideoLlama3Projector.__init__": ["GELU", "Linear", "ModelConfig", "ModelProjector", "Module", "None", "Sequential", "__init__", "class", "config", "def", "hidden_size", "in_hidden_size", "nn", "out_hidden_size", "readout", "self", "super", "text_config", "vision_config"], "video_llama_3/modeling_video_llama_3.py:VideoLlama3Projector.forward": ["ModelProjector", "Module", "Tensor", "class", "def", "forward", "hidden_states", "nn", "readout", "return", "self", "torch"], "video_llama_3/modeling_video_llama_3.py:VideoLlama3Model.__init__": ["AutoModel", "ModelConfig", "ModelModel", "ModelPreTrainedModel", "ModelProjector", "__init__", "class", "config", "def", "from_config", "language_model", "post_init", "projector", "self", "super", "text_config", "vision_config", "vision_model"], "video_llama_3/modeling_video_llama_3.py:VideoLlama3Model.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "language_model", "return", "self"], "video_llama_3/modeling_video_llama_3.py:VideoLlama3Model.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "language_model", "self", "set_input_embeddings", "value"], "video_llama_3/modeling_video_llama_3.py:VideoLlama3Model.get_video_features": ["FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "Model_grid_thw", "Model_merge_sizes", "class", "def", "get_Model_features", "get_image_features", "pixel_values_Models", "return", "self", "torch"], "video_llama_3/modeling_video_llama_3.py:VideoLlama3Model.get_image_features": ["FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "True", "class", "def", "dim", "get_image_features", "grid_thw", "image_embeds", "image_grid_thw", "image_merge_sizes", "last_hidden_state", "merge_sizes", "pixel_values", "prod", "projector", "return", "return_dict", "self", "split", "split_sizes", "tolist", "torch", "vision_model"], "video_llama_3/modeling_video_llama_3.py:VideoLlama3Model.get_placeholder_mask": ["FloatTensor", "Image", "LongTensor", "Model", "ModelModel", "ModelPreTrainedModel", "Model_features", "Model_token_id", "Models", "None", "ValueError", "all", "and", "class", "config", "def", "device", "do", "dtype", "else", "expand_as", "f", "features", "get_input_embeddings", "get_placeholder_mask", "if", "image", "image_features", "image_token_id", "input_ids", "inputs_embeds", "is", "long", "match", "n_Model_tokens", "n_image_tokens", "not", "numel", "raise", "return", "self", "shape", "special_Model_mask", "special_image_mask", "sum", "tensor", "to", "tokens", "torch", "unsqueeze"], "video_llama_3/modeling_video_llama_3.py:VideoLlama3Model.forward": ["BoolTensor", "Cache", "FloatTensor", "LongTensor", "ModelModel", "ModelModelOutputWithPast", "ModelPreTrainedModel", "Model_compression_mask", "Model_embeds", "Model_features", "Model_grid_thw", "Model_hidden_states", "Model_mask", "Model_merge_sizes", "None", "Tensor", "TransformersKwargs", "Unpack", "_", "attention_mask", "attentions", "bool", "cache_position", "cat", "class", "def", "device", "dim", "dtype", "forward", "get_Model_features", "get_image_features", "get_input_embeddings", "get_placeholder_mask", "hidden_states", "if", "image_embeds", "image_features", "image_grid_thw", "image_hidden_states", "image_mask", "image_merge_sizes", "input_ids", "inputs_embeds", "is", "kwargs", "language_model", "last_hidden_state", "masked_scatter", "not", "outputs", "past_key_values", "pixel_values", "pixel_values_Models", "position_ids", "r", "return", "self", "to", "torch", "tuple", "use_cache"], "video_llama_3/modeling_video_llama_3.py:VideoLlama3ForConditionalGeneration.__init__": ["False", "GenerationMixin", "Linear", "ModelConfig", "ModelForConditionalGeneration", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "text_config", "vocab_size"], "video_llama_3/modeling_video_llama_3.py:VideoLlama3ForConditionalGeneration.get_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "model", "return", "self"], "video_llama_3/modeling_video_llama_3.py:VideoLlama3ForConditionalGeneration.set_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "model", "self", "set_input_embeddings", "value"], "video_llama_3/modeling_video_llama_3.py:VideoLlama3ForConditionalGeneration.get_video_features": ["FloatTensor", "GenerationMixin", "LongTensor", "ModelForConditionalGeneration", "ModelPreTrainedModel", "Model_grid_thw", "None", "class", "def", "get_Model_features", "model", "pixel_values_Models", "return", "self", "torch"], "video_llama_3/modeling_video_llama_3.py:VideoLlama3ForConditionalGeneration.get_image_features": ["FloatTensor", "GenerationMixin", "LongTensor", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "class", "def", "get_image_features", "image_grid_thw", "model", "pixel_values", "return", "self", "torch"], "video_llama_3/modeling_video_llama_3.py:VideoLlama3ForConditionalGeneration.forward": ["BoolTensor", "Cache", "FloatTensor", "GenerationMixin", "LongTensor", "ModelCausalLMOutputWithPast", "ModelForConditionalGeneration", "ModelPreTrainedModel", "Model_compression_mask", "Model_grid_thw", "Model_hidden_states", "Model_merge_sizes", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "forward", "hidden_states", "if", "image_grid_thw", "image_hidden_states", "image_merge_sizes", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "lm_head", "logits", "loss", "loss_function", "model", "not", "outputs", "past_key_values", "pixel_values", "pixel_values_Models", "position_ids", "r", "return", "return_dict", "self", "text_config", "torch", "tuple", "use_cache", "vocab_size"], "video_llama_3/modeling_video_llama_3.py:VideoLlama3ForConditionalGeneration.prepare_inputs_for_generation": ["BoolTensor", "False", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForConditionalGeneration", "ModelPreTrainedModel", "Model_compression_mask", "Model_grid_thw", "Model_merge_sizes", "None", "Tensor", "True", "and", "attention_mask", "bool", "cache_position", "class", "def", "if", "image_grid_thw", "image_merge_sizes", "input_ids", "inputs_embeds", "is_first_iteration", "kwargs", "model_inputs", "not", "past_key_values", "pixel_values", "pixel_values_Models", "position_ids", "prepare_inputs_for_generation", "return", "self", "super", "torch", "use_cache"], "video_llama_3/modeling_video_llama_3.py:VideoLlama3ForConditionalGeneration._get_image_nums_and_video_nums": ["BoolTensor", "GenerationMixin", "LongTensor", "Model", "ModelForConditionalGeneration", "ModelPreTrainedModel", "Model_compression_mask", "Model_grid_thw", "Model_idx", "Model_mask", "Model_merge_sizes", "Model_nums", "Model_token_id", "None", "Tensor", "The", "_get_image_nums_and_Model_nums", "and", "append", "assert", "class", "config", "cu_num_features", "def", "device", "dim", "does", "dtype", "else", "features", "for", "get_input_embeddings", "if", "image", "image_grid_thw", "image_idx", "image_mask", "image_merge_sizes", "image_nums", "image_token_id", "in", "input_ids", "inputs_embeds", "is", "len", "long", "mask", "match", "not", "num_Model_features", "num_Model_tokens", "num_image_features", "num_image_tokens", "number", "of", "prod", "return", "self", "split", "start_Model_idx", "start_image_idx", "sum", "tensor", "the", "tokens", "tolist", "torch", "tuple", "while", "zip"], "video_llama_3/modeling_video_llama_3.py:VideoLlama3ForConditionalGeneration._expand_inputs_for_generation": ["Any", "False", "GenerationMixin", "If", "LongTensor", "ModelForConditionalGeneration", "ModelPreTrainedModel", "Model_compression_mask", "Model_grid_thw", "Model_merge_sizes", "Model_nums", "None", "Tensor", "True", "ValueError", "_expand_dict_for_generation", "_expand_dict_for_generation_visual", "_expand_inputs_for_generation", "_get_image_nums_and_Model_nums", "_repeat_interleave_samples", "and", "bool", "cache_position", "cat", "class", "def", "defined", "dict", "dict_to_expand", "dim", "elif", "encoder_outputs", "expand_size", "for", "get", "if", "image_grid_thw", "image_merge_sizes", "image_nums", "in", "input_ids", "inputs_embeds", "int", "is", "is_encoder_decoder", "isinstance", "key", "lengths", "list", "make", "merge_size", "merge_sizes", "model_kwargs", "not", "pixel_values", "pixel_values_Models", "prod", "raise", "repeat", "repeat_args", "repeat_interleave", "repeat_times", "result", "return", "sample", "samples", "self", "split", "str", "sum", "sure", "that", "torch", "tuple", "visual_keys", "x", "zip"], "video_llava/modeling_video_llava.py:VideoLlavaMultiModalProjector.__init__": ["ACT2FN", "Linear", "ModelConfig", "ModelMultiModalProjector", "Module", "__init__", "act", "bias", "class", "config", "def", "else", "hidden_size", "if", "int", "isinstance", "len", "linear_1", "linear_2", "multimodal_projector_bias", "nn", "num_feature_layers", "projector_hidden_act", "self", "super", "text_config", "vision_config", "vision_feature_layer"], "video_llava/modeling_video_llava.py:VideoLlavaMultiModalProjector.forward": ["ModelMultiModalProjector", "Module", "act", "class", "def", "forward", "hidden_states", "image_features", "linear_1", "linear_2", "nn", "return", "self"], "video_llava/modeling_video_llava.py:VideoLlavaPreTrainedModel._init_weights": ["ModelPreTrainedModel", "PreTrainedModel", "_init_weights", "class", "class_embedding", "config", "def", "else", "hasattr", "if", "init", "initializer_range", "mean", "module", "normal_", "self", "std", "super", "text_config"], "video_llava/modeling_video_llava.py:VideoLlavaModel.__init__": ["AutoModel", "ModelConfig", "ModelModel", "ModelMultiModalProjector", "ModelPreTrainedModel", "Model_tower", "None", "__init__", "class", "config", "def", "else", "from_config", "if", "image_tower", "is", "language_model", "multi_modal_projector", "not", "pad_token_id", "post_init", "self", "super", "text_config", "vision_config", "vocab_size"], "video_llava/modeling_video_llava.py:VideoLlavaModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "language_model", "return", "self"], "video_llava/modeling_video_llava.py:VideoLlavaModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "language_model", "self", "set_input_embeddings", "value"], "video_llava/modeling_video_llava.py:VideoLlavaModel.get_image_features": ["FloatTensor", "ModelModel", "ModelPreTrainedModel", "None", "True", "Unexpected", "ValueError", "cat", "class", "config", "def", "default", "dim", "else", "f", "feature", "for", "full", "get_image_features", "hidden_states", "hs", "hs_pool", "if", "image_features", "image_outputs", "image_tower", "in", "int", "is", "isinstance", "layer_idx", "list", "multi_modal_projector", "not", "output_hidden_states", "pixel_values_images", "raise", "return", "select", "self", "str", "strategy", "torch", "vision_feature_layer", "vision_feature_select_strategy"], "video_llava/modeling_video_llava.py:VideoLlavaModel.get_video_features": ["FloatTensor", "ModelModel", "ModelPreTrainedModel", "Model_features", "Model_outputs", "Model_tower", "None", "True", "batch_size_vid", "cat", "channels", "class", "config", "def", "dim", "else", "for", "get_Model_features", "height", "hidden_states", "hs_pool", "if", "in", "int", "is", "isinstance", "layer_idx", "list", "multi_modal_projector", "not", "num_frames", "output_hidden_states", "pixel_values", "pixel_values_Models", "reshape", "return", "self", "shape", "torch", "vision_feature_layer", "width"], "video_llava/modeling_video_llava.py:VideoLlavaModel.get_placeholder_mask": ["FloatTensor", "Image", "LongTensor", "ModelModel", "ModelPreTrainedModel", "Model_features", "Model_token_id", "Models", "None", "ValueError", "all", "and", "class", "config", "def", "device", "do", "dtype", "else", "expand_as", "f", "features", "get_input_embeddings", "get_placeholder_mask", "if", "image", "image_features", "image_token_id", "input_ids", "inputs_embeds", "is", "long", "match", "n_Model_tokens", "n_image_tokens", "not", "numel", "raise", "return", "self", "shape", "special_Model_mask", "special_image_mask", "sum", "tensor", "to", "tokens", "torch", "unsqueeze"], "video_llava/modeling_video_llava.py:VideoLlavaModel.forward": ["Cache", "FlashAttentionKwargs", "FloatTensor", "LongTensor", "ModelModel", "ModelModelOutputWithPast", "ModelPreTrainedModel", "Model_features", "Model_hidden_states", "None", "Tensor", "True", "Unpack", "ValueError", "You", "_", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "device", "dtype", "else", "exactly", "forward", "get_Model_features", "get_image_features", "get_input_embeddings", "get_placeholder_mask", "hidden_states", "if", "image_features", "image_hidden_states", "input_ids", "inputs_embeds", "int", "is", "kwargs", "language_model", "last_hidden_state", "list", "masked_scatter", "must", "not", "num_frames", "of", "one", "or", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "pixel_values_Models", "pixel_values_images", "position_ids", "r", "raise", "return", "return_dict", "self", "special_Model_mask", "special_image_mask", "specify", "str", "to", "torch", "tuple", "use_cache", "use_return_dict", "vision_feature_layer", "vision_feature_select_strategy"], "video_llava/modeling_video_llava.py:VideoLlavaForConditionalGeneration.__init__": ["False", "GenerationMixin", "Linear", "ModelConfig", "ModelForConditionalGeneration", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "text_config", "vocab_size"], "video_llava/modeling_video_llava.py:VideoLlavaForConditionalGeneration.get_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "model", "return", "self"], "video_llava/modeling_video_llava.py:VideoLlavaForConditionalGeneration.set_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "model", "self", "set_input_embeddings", "value"], "video_llava/modeling_video_llava.py:VideoLlavaForConditionalGeneration.get_output_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "Module", "class", "def", "get_output_embeddings", "lm_head", "nn", "return", "self"], "video_llava/modeling_video_llava.py:VideoLlavaForConditionalGeneration.get_image_features": ["FloatTensor", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "class", "def", "get_image_features", "int", "list", "model", "pixel_values_images", "return", "self", "str", "torch", "vision_feature_layer", "vision_feature_select_strategy"], "video_llava/modeling_video_llava.py:VideoLlavaForConditionalGeneration.forward": ["Cache", "FloatTensor", "GenerationMixin", "LongTensor", "ModelCausalLMOutputWithPast", "ModelForConditionalGeneration", "ModelPreTrainedModel", "Model_hidden_states", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "else", "forward", "hidden_states", "if", "image_hidden_states", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "list", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "pixel_values_Models", "pixel_values_images", "position_ids", "r", "return", "return_dict", "self", "slice", "slice_indices", "str", "text_config", "torch", "tuple", "use_cache", "use_return_dict", "vision_feature_layer", "vision_feature_select_strategy", "vocab_size"], "video_llava/modeling_video_llava.py:VideoLlavaForConditionalGeneration.prepare_inputs_for_generation": ["False", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "True", "attention_mask", "cache_position", "class", "def", "get", "if", "input_ids", "inputs_embeds", "is_first_iteration", "kwargs", "logits_to_keep", "model_inputs", "not", "or", "past_key_values", "pixel_values_Models", "pixel_values_images", "prepare_inputs_for_generation", "return", "self", "super", "use_cache"], "video_llava/modeling_video_llava.py:VideoLlavaForConditionalGeneration._prepare_4d_causal_attention_mask_with_cache_position": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Tensor", "_prepare_4d_causal_attention_mask_with_cache_position", "and", "arange", "attention_mask", "batch_size", "cache_position", "causal_mask", "class", "clone", "def", "device", "diagonal", "dim", "dtype", "else", "expand", "fill_value", "finfo", "full", "if", "int", "is", "kwargs", "mask_length", "masked_fill", "min", "min_dtype", "not", "padding_mask", "reshape", "return", "sequence_length", "shape", "target_length", "to", "torch", "triu"], "videomae/modeling_videomae.py:get_sinusoid_encoding_table": ["FloatTensor", "Model_position_angle_vec", "Model_sinusoid_encoding_table", "array", "cos", "d_hid", "def", "for", "hid_j", "in", "n_position", "np", "pos_i", "position", "power", "range", "return", "sin", "sinusoid_table", "torch", "unsqueeze"], "videomae/modeling_videomae.py:VideoMAEEmbeddings.__init__": ["ModelEmbeddings", "ModelPatchEmbeddings", "Module", "__init__", "class", "config", "def", "get_sinusoid_encoding_table", "hidden_size", "nn", "num_patches", "patch_embeddings", "position_embeddings", "self", "super"], "videomae/modeling_videomae.py:VideoMAEEmbeddings.forward": ["ModelEmbeddings", "Module", "None", "True", "_", "batch_size", "bool_masked_pos", "class", "copy", "def", "detach", "device", "embeddings", "forward", "if", "is", "nn", "not", "num_channels", "patch_embeddings", "pixel_values", "position_embeddings", "reshape", "return", "self", "shape", "to", "type_as"], "videomae/modeling_videomae.py:VideoMAEPatchEmbeddings.__init__": ["Conv3d", "Iterable", "ModelPatchEmbeddings", "Module", "__init__", "abc", "class", "collections", "config", "def", "else", "hidden_size", "if", "image_size", "in_channels", "int", "isinstance", "kernel_size", "nn", "num_channels", "num_frames", "num_patches", "out_channels", "patch_size", "projection", "self", "stride", "super", "tubelet_size"], "videomae/modeling_videomae.py:VideoMAEPatchEmbeddings.forward": ["Input", "Make", "ModelPatchEmbeddings", "Module", "ValueError", "batch_size", "channel", "class", "configuration", "def", "dimension", "doesn", "embeddings", "f", "flatten", "forward", "height", "if", "image", "image_size", "in", "match", "model", "nn", "num_channels", "num_frames", "of", "one", "or", "permute", "pixel", "pixel_values", "projection", "raise", "return", "self", "set", "shape", "size", "sure", "t", "that", "the", "transpose", "values", "width", "with"], "videomae/modeling_videomae.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "contiguous", "def", "dim", "dropout", "float", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "shape", "size", "softmax", "torch", "training", "transpose", "value"], "videomae/modeling_videomae.py:VideoMAESelfAttention.__init__": ["False", "Linear", "ModelConfig", "ModelSelfAttention", "Module", "None", "Parameter", "The", "ValueError", "__init__", "a", "all_head_size", "and", "attention", "attention_head_size", "attention_probs_dropout_prob", "bias", "class", "config", "def", "dropout_prob", "else", "embedding_size", "f", "hasattr", "heads", "hidden", "hidden_size", "if", "int", "is", "is_causal", "key", "multiple", "nn", "not", "num_attention_heads", "number", "of", "q_bias", "qkv_bias", "query", "raise", "scaling", "self", "size", "super", "the", "torch", "v_bias", "value", "zeros"], "videomae/modeling_videomae.py:VideoMAESelfAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "False", "ModelSelfAttention", "Module", "None", "Tensor", "_", "_attn_implementation", "all_head_size", "attention_head_size", "attention_interface", "attention_probs", "batch_size", "bias", "class", "config", "context_layer", "def", "dropout", "dropout_prob", "eager", "eager_attention_forward", "else", "forward", "functional", "hidden_states", "if", "input", "is", "is_causal", "k_bias", "key", "key_layer", "keys", "linear", "new_context_layer_shape", "nn", "not", "num_attention_heads", "q_bias", "queries", "query", "query_layer", "requires_grad", "reshape", "return", "scaling", "self", "seq_length", "shape", "size", "torch", "training", "transpose", "tuple", "v_bias", "value", "value_layer", "values", "view", "weight", "zeros_like"], "videomae/modeling_videomae.py:VideoMAESelfOutput.__init__": ["Dropout", "Linear", "ModelConfig", "ModelSelfOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "hidden_dropout_prob", "hidden_size", "nn", "self", "super"], "videomae/modeling_videomae.py:VideoMAESelfOutput.forward": ["ModelSelfOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "videomae/modeling_videomae.py:VideoMAEAttention.__init__": ["ModelAttention", "ModelConfig", "ModelSelfAttention", "ModelSelfOutput", "Module", "__init__", "attention", "class", "config", "def", "nn", "output", "self", "super"], "videomae/modeling_videomae.py:VideoMAEAttention.forward": ["ModelAttention", "Module", "Tensor", "_", "attention", "class", "def", "forward", "hidden_states", "nn", "output", "return", "self", "self_attn_output", "torch"], "videomae/modeling_videomae.py:VideoMAEIntermediate.__init__": ["ACT2FN", "Linear", "ModelConfig", "ModelIntermediate", "Module", "__init__", "class", "config", "def", "dense", "else", "hidden_act", "hidden_size", "if", "intermediate_act_fn", "intermediate_size", "isinstance", "nn", "self", "str", "super"], "videomae/modeling_videomae.py:VideoMAEIntermediate.forward": ["ModelIntermediate", "Module", "Tensor", "class", "def", "dense", "forward", "hidden_states", "intermediate_act_fn", "nn", "return", "self", "torch"], "videomae/modeling_videomae.py:VideoMAEOutput.__init__": ["Dropout", "Linear", "ModelConfig", "ModelOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "hidden_dropout_prob", "hidden_size", "intermediate_size", "nn", "self", "super"], "videomae/modeling_videomae.py:VideoMAEOutput.forward": ["ModelOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "videomae/modeling_videomae.py:VideoMAELayer.__init__": ["GradientCheckpointingLayer", "LayerNorm", "ModelAttention", "ModelConfig", "ModelIntermediate", "ModelLayer", "ModelOutput", "__init__", "attention", "chunk_size_feed_forward", "class", "config", "def", "eps", "hidden_size", "intermediate", "layer_norm_eps", "layernorm_after", "layernorm_before", "nn", "output", "self", "seq_len_dim", "super"], "videomae/modeling_videomae.py:VideoMAELayer.forward": ["GradientCheckpointingLayer", "ModelLayer", "Tensor", "attention", "attention_output", "class", "def", "forward", "hidden_states", "hidden_states_norm", "intermediate", "layer_output", "layernorm_after", "layernorm_before", "output", "return", "self", "torch"], "videomae/modeling_videomae.py:VideoMAEEncoder.__init__": ["False", "ModelConfig", "ModelEncoder", "ModelLayer", "Module", "ModuleList", "_", "__init__", "class", "config", "def", "for", "gradient_checkpointing", "in", "layer", "nn", "num_hidden_layers", "range", "self", "super"], "videomae/modeling_videomae.py:VideoMAEEncoder.forward": ["BaseModelOutput", "ModelEncoder", "Module", "Tensor", "class", "def", "enumerate", "for", "forward", "hidden_states", "i", "in", "last_hidden_state", "layer", "layer_module", "nn", "return", "self", "torch"], "videomae/modeling_videomae.py:VideoMAEModel.__init__": ["LayerNorm", "ModelEmbeddings", "ModelEncoder", "ModelModel", "ModelPreTrainedModel", "None", "__init__", "class", "config", "def", "else", "embeddings", "encoder", "eps", "hidden_size", "if", "layer_norm_eps", "layernorm", "nn", "post_init", "self", "super", "use_mean_pooling"], "videomae/modeling_videomae.py:VideoMAEModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "patch_embeddings", "return", "self"], "videomae/modeling_videomae.py:VideoMAEModel.forward": ["BaseModelOutput", "BoolTensor", "FloatTensor", "ModelModel", "ModelPreTrainedModel", "None", "TransformersKwargs", "Unpack", "bool_masked_pos", "class", "def", "embedding_output", "embeddings", "encoder", "encoder_outputs", "forward", "if", "is", "kwargs", "last_hidden_state", "layernorm", "not", "pixel_values", "r", "return", "self", "sequence_output", "torch"], "videomae/modeling_videomae.py:VideoMAEDecoder.__init__": ["False", "Identity", "LayerNorm", "Linear", "ModelConfig", "ModelDecoder", "ModelLayer", "Module", "ModuleList", "_", "__init__", "class", "config", "decoder_config", "decoder_hidden_size", "decoder_intermediate_size", "decoder_layers", "decoder_num_attention_heads", "decoder_num_hidden_layers", "decoder_num_labels", "deepcopy", "def", "else", "for", "gradient_checkpointing", "head", "hidden_size", "if", "in", "intermediate_size", "nn", "norm", "num_attention_heads", "num_channels", "num_hidden_layers", "patch_size", "range", "self", "super", "tubelet_size"], "videomae/modeling_videomae.py:VideoMAEDecoder.forward": ["ModelDecoder", "ModelDecoderOutput", "Module", "Tensor", "class", "decoder_layers", "def", "for", "forward", "head", "hidden_states", "if", "in", "int", "layer_module", "logits", "nn", "norm", "return", "return_token_num", "self", "torch"], "videomae/modeling_videomae.py:VideoMAEForPreTraining.__init__": ["False", "Linear", "Model", "ModelDecoder", "ModelForPreTraining", "ModelModel", "ModelPreTrainedModel", "Parameter", "__init__", "bias", "class", "config", "decoder", "decoder_hidden_size", "def", "embeddings", "encoder_to_decoder", "get_sinusoid_encoding_table", "hidden_size", "mask_token", "nn", "num_patches", "position_embeddings", "post_init", "self", "super", "torch", "zeros"], "videomae/modeling_videomae.py:VideoMAEForPreTraining.forward": ["BaseModelOutput", "BoolTensor", "Can", "Consider", "False", "FloatTensor", "IMAGENET_DEFAULT_MEAN", "IMAGENET_DEFAULT_STD", "MSELoss", "Model", "ModelDecoderOutput", "ModelForPreTraining", "ModelForPreTrainingOutput", "ModelPreTrainedModel", "Models_patch", "None", "One", "RGB", "TransformersKwargs", "True", "Unpack", "ValueError", "_", "a", "as_tensor", "attentions", "batch_size", "bool_masked_pos", "boolean", "cat", "class", "config", "contiguous", "copy", "decoder", "decoder_outputs", "def", "detach", "device", "dim", "dtype", "else", "encoder_to_decoder", "expand", "expanded_position_embeddings", "forward", "frames", "frames_norm", "height", "hidden_states", "if", "images", "is", "keepdim", "kwargs", "labels", "last_hidden_state", "logits", "loss", "loss_fct", "mask", "mask_token", "mean", "must", "no_grad", "non", "norm_pix_loss", "num_channels", "outputs", "patch_size", "permute", "pixel_values", "pos_emb_mask", "pos_emb_visible", "position_embeddings", "provided", "r", "raise", "reshape", "return", "self", "sequence_output", "setting", "shape", "sqrt", "std", "t", "time", "to", "torch", "tubelet_size", "type_as", "unbiased", "unnormalize", "var", "view", "width", "with", "x_full"], "videomae/modeling_videomae.py:VideoMAEForVideoClassification.__init__": ["Identity", "LayerNorm", "Linear", "Model", "ModelForModelClassification", "ModelModel", "ModelPreTrainedModel", "None", "__init__", "class", "classifier", "config", "def", "else", "fc_norm", "hidden_size", "if", "nn", "num_labels", "post_init", "self", "super", "use_mean_pooling"], "videomae/modeling_videomae.py:VideoMAEForVideoClassification.forward": ["BaseModelOutput", "ImageClassifierOutput", "Model", "ModelForModelClassification", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "attentions", "class", "classifier", "config", "def", "else", "fc_norm", "forward", "hidden_states", "if", "is", "kwargs", "labels", "last_hidden_state", "logits", "loss", "loss_function", "mean", "not", "output", "outputs", "pixel_values", "r", "return", "self", "sequence_output", "torch"], "vilt/modeling_vilt.py:ViltEmbeddings.__init__": ["Dropout", "Embedding", "ModelEmbeddings", "ModelPatchEmbeddings", "Module", "Parameter", "TextEmbeddings", "__init__", "class", "cls_token", "config", "def", "dropout", "hidden_dropout_prob", "hidden_size", "modality_type_vocab_size", "nn", "num_patches", "patch_embeddings", "position_embeddings", "self", "super", "text_embeddings", "token_type_embeddings", "torch", "zeros"], "vilt/modeling_vilt.py:ViltEmbeddings.visual_embed": ["False", "ModelEmbeddings", "Module", "None", "True", "_", "align_corners", "append", "arange", "as_tuple", "batch_size", "bilinear", "cat", "class", "cls_token", "cls_tokens", "config", "def", "device", "dim", "dropout", "effective_resolution", "else", "enumerate", "expand", "flatten", "float", "for", "functional", "h", "height", "i", "if", "ij", "image_size", "in", "indexing", "int", "interpolate", "is", "isinstance", "long", "max", "max_image_length", "meshgrid", "min", "mode", "multinomial", "nn", "non_valid_idx", "non_valid_nums", "non_valid_row_idx", "nonzero", "not", "num_channels", "nv", "ones", "or", "p", "pad", "pad_choice", "pad_nums", "patch_dim", "patch_embeddings", "patch_index", "patch_size", "ph", "pixel_mask", "pixel_values", "pos_embed", "position_embeddings", "projection", "pw", "replacement", "return", "select", "self", "shape", "size", "spatial_pos", "stack", "sum", "to", "torch", "transpose", "u", "unique", "unique_rows", "v", "valid_choice", "valid_idx", "valid_nums", "valid_row_idx", "view", "visual_embed", "w", "weight", "width", "x", "x_h", "x_mask", "x_w", "zip"], "vilt/modeling_vilt.py:ViltEmbeddings.forward": ["ModelEmbeddings", "Module", "None", "attention_mask", "cat", "class", "config", "def", "device", "dim", "dtype", "else", "embeddings", "flatten", "forward", "full_like", "if", "image_embeds", "image_masks", "image_token_type_idx", "input_ids", "inputs_embeds", "is", "long", "masks", "max_image_length", "nn", "patch_index", "pixel_mask", "pixel_values", "return", "self", "text_embeddings", "text_embeds", "token_type_embeddings", "token_type_ids", "torch", "visual_embed", "zeros_like"], "vilt/modeling_vilt.py:TextEmbeddings.__init__": ["Dropout", "Embedding", "False", "LayerNorm", "ModelEmbeddings", "Module", "__init__", "arange", "class", "config", "def", "dropout", "dtype", "eps", "expand", "hidden_dropout_prob", "hidden_size", "layer_norm_eps", "long", "max_position_embeddings", "nn", "pad_token_id", "padding_idx", "persistent", "position_embeddings", "position_ids", "register_buffer", "self", "size", "super", "token_type_embeddings", "token_type_ids", "torch", "type_vocab_size", "vocab_size", "word_embeddings", "zeros"], "vilt/modeling_vilt.py:TextEmbeddings.forward": ["LayerNorm", "ModelEmbeddings", "Module", "None", "buffered_token_type_ids", "buffered_token_type_ids_expanded", "class", "def", "device", "dropout", "dtype", "else", "embeddings", "expand", "forward", "hasattr", "if", "input_ids", "input_shape", "inputs_embeds", "is", "long", "nn", "not", "position_embeddings", "position_ids", "return", "self", "seq_length", "size", "token_type_embeddings", "token_type_ids", "torch", "word_embeddings", "zeros"], "vilt/modeling_vilt.py:ViltPatchEmbeddings.__init__": ["Conv2d", "Iterable", "ModelPatchEmbeddings", "Module", "__init__", "abc", "class", "collections", "config", "def", "else", "hidden_size", "if", "image_size", "isinstance", "kernel_size", "nn", "num_channels", "num_patches", "patch_size", "projection", "self", "stride", "super"], "vilt/modeling_vilt.py:ViltPatchEmbeddings.forward": ["Make", "ModelPatchEmbeddings", "Module", "ValueError", "batch_size", "channel", "class", "configuration", "def", "dimension", "dtype", "forward", "height", "if", "in", "match", "nn", "num_channels", "of", "one", "pixel", "pixel_values", "projection", "raise", "return", "self", "set", "shape", "sure", "target_dtype", "that", "the", "to", "values", "weight", "width", "with", "x"], "vilt/modeling_vilt.py:ViltSelfAttention.__init__": ["Dropout", "Linear", "ModelSelfAttention", "Module", "The", "ValueError", "__init__", "a", "all_head_size", "and", "attention", "attention_head_size", "attention_probs_dropout_prob", "bias", "class", "config", "def", "dropout", "embedding_size", "f", "hasattr", "heads", "hidden", "hidden_size", "if", "int", "is", "key", "multiple", "nn", "not", "num_attention_heads", "number", "of", "qkv_bias", "query", "raise", "self", "size", "super", "the", "value"], "vilt/modeling_vilt.py:ViltSelfAttention.forward": ["False", "ModelSelfAttention", "Module", "None", "Softmax", "_", "all_head_size", "attention_head_size", "attention_mask", "attention_probs", "attention_scores", "batch_size", "class", "context_layer", "contiguous", "def", "dim", "dropout", "else", "forward", "hidden_states", "if", "is", "key", "key_layer", "math", "matmul", "new_context_layer_shape", "nn", "not", "num_attention_heads", "output_attentions", "outputs", "permute", "query", "query_layer", "return", "self", "seq_length", "shape", "size", "sqrt", "torch", "transpose", "value", "value_layer", "view"], "vilt/modeling_vilt.py:ViltSelfOutput.__init__": ["Dropout", "Linear", "ModelConfig", "ModelSelfOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "hidden_dropout_prob", "hidden_size", "nn", "self", "super"], "vilt/modeling_vilt.py:ViltSelfOutput.forward": ["ModelSelfOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "vilt/modeling_vilt.py:ViltAttention.__init__": ["ModelAttention", "ModelSelfAttention", "ModelSelfOutput", "Module", "__init__", "attention", "class", "config", "def", "nn", "output", "self", "super"], "vilt/modeling_vilt.py:ViltAttention.forward": ["False", "ModelAttention", "Module", "None", "attention", "attention_mask", "attention_output", "class", "def", "forward", "hidden_states", "nn", "output", "output_attentions", "outputs", "return", "self", "self_outputs"], "vilt/modeling_vilt.py:ViltIntermediate.__init__": ["ACT2FN", "Linear", "ModelConfig", "ModelIntermediate", "Module", "__init__", "class", "config", "def", "dense", "else", "hidden_act", "hidden_size", "if", "intermediate_act_fn", "intermediate_size", "isinstance", "nn", "self", "str", "super"], "vilt/modeling_vilt.py:ViltIntermediate.forward": ["ModelIntermediate", "Module", "Tensor", "class", "def", "dense", "forward", "hidden_states", "intermediate_act_fn", "nn", "return", "self", "torch"], "vilt/modeling_vilt.py:ViltOutput.__init__": ["Dropout", "Linear", "ModelConfig", "ModelOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "hidden_dropout_prob", "hidden_size", "intermediate_size", "nn", "self", "super"], "vilt/modeling_vilt.py:ViltOutput.forward": ["ModelOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "vilt/modeling_vilt.py:ViltLayer.__init__": ["GradientCheckpointingLayer", "LayerNorm", "ModelAttention", "ModelIntermediate", "ModelLayer", "ModelOutput", "__init__", "attention", "chunk_size_feed_forward", "class", "config", "def", "eps", "hidden_size", "intermediate", "layer_norm_eps", "layernorm_after", "layernorm_before", "nn", "output", "self", "seq_len_dim", "super"], "vilt/modeling_vilt.py:ViltLayer.forward": ["False", "GradientCheckpointingLayer", "ModelLayer", "None", "attention", "attention_mask", "attention_output", "class", "def", "device", "forward", "hidden_states", "intermediate", "layer_output", "layernorm_after", "layernorm_before", "output", "output_attentions", "outputs", "return", "self", "self_attention_outputs", "to"], "vilt/modeling_vilt.py:ViltEncoder.__init__": ["False", "ModelEncoder", "ModelLayer", "Module", "ModuleList", "_", "__init__", "class", "config", "def", "for", "gradient_checkpointing", "in", "layer", "nn", "num_hidden_layers", "range", "self", "super"], "vilt/modeling_vilt.py:ViltEncoder.forward": ["BaseModelOutput", "False", "ModelEncoder", "Module", "None", "True", "all_hidden_states", "all_self_attentions", "attention_mask", "attentions", "class", "def", "else", "enumerate", "for", "forward", "hidden_states", "i", "if", "in", "is", "last_hidden_state", "layer", "layer_module", "layer_outputs", "nn", "not", "output_attentions", "output_hidden_states", "return", "return_dict", "self", "tuple", "v"], "vilt/modeling_vilt.py:ViltPreTrainedModel._init_weights": ["ModelPreTrainedModel", "PreTrainedModel", "TextEmbeddings", "_init_weights", "arange", "class", "copy_", "def", "expand", "if", "init", "isinstance", "module", "position_ids", "self", "shape", "super", "token_type_ids", "torch", "zeros_"], "vilt/modeling_vilt.py:ViltModel.__init__": ["LayerNorm", "ModelEmbeddings", "ModelEncoder", "ModelModel", "ModelPooler", "ModelPreTrainedModel", "None", "True", "__init__", "add_pooling_layer", "class", "config", "def", "else", "embeddings", "encoder", "eps", "hidden_size", "if", "layer_norm_eps", "layernorm", "nn", "pooler", "post_init", "r", "self", "super"], "vilt/modeling_vilt.py:ViltModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "return", "self", "text_embeddings", "word_embeddings"], "vilt/modeling_vilt.py:ViltModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "self", "set_input_embeddings", "text_embeddings", "value", "word_embeddings"], "vilt/modeling_vilt.py:ViltModel.forward": ["BaseModelOutputWithPooling", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "The", "ValueError", "You", "and", "at", "attention_mask", "attentions", "batch", "bool", "both", "cannot", "class", "config", "def", "device", "either", "elif", "else", "embedding_output", "embeddings", "encoder", "encoder_outputs", "extended_attention_mask", "forward", "get_extended_attention_mask", "have", "hidden_states", "if", "image", "image_batch_size", "image_embeds", "image_size", "image_token_type_idx", "input_ids", "input_shape", "inputs", "inputs_embeds", "int", "is", "kwargs", "last_hidden_state", "layernorm", "need", "not", "ones", "or", "output_attentions", "output_hidden_states", "pixel_mask", "pixel_values", "pooled_output", "pooler", "pooler_output", "r", "raise", "return", "return_dict", "same", "self", "seq_length", "sequence_output", "shape", "size", "specify", "text", "text_batch_size", "the", "time", "to", "token_type_ids", "torch", "tuple", "use_return_dict", "warn_if_padding_and_no_attention_mask"], "vilt/modeling_vilt.py:ViltPooler.__init__": ["Linear", "ModelPooler", "Module", "Tanh", "__init__", "activation", "class", "config", "def", "dense", "hidden_size", "nn", "self", "super"], "vilt/modeling_vilt.py:ViltPooler.forward": ["ModelPooler", "Module", "activation", "class", "def", "dense", "first_token_tensor", "forward", "hidden_states", "nn", "pooled_output", "return", "self"], "vilt/modeling_vilt.py:ViltForMaskedLM.__init__": ["Model", "ModelForMaskedLM", "ModelMLMHead", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "config", "def", "mlm_score", "post_init", "self", "super"], "vilt/modeling_vilt.py:ViltForMaskedLM.get_output_embeddings": ["ModelForMaskedLM", "ModelPreTrainedModel", "class", "decoder", "def", "get_output_embeddings", "mlm_score", "return", "self"], "vilt/modeling_vilt.py:ViltForMaskedLM.set_output_embeddings": ["ModelForMaskedLM", "ModelPreTrainedModel", "bias", "class", "decoder", "def", "mlm_score", "new_embeddings", "self", "set_output_embeddings"], "vilt/modeling_vilt.py:ViltForMaskedLM.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "MaskedLMOutput", "Model", "ModelForMaskedLM", "ModelPreTrainedModel", "None", "_", "attention_mask", "attentions", "bool", "class", "config", "def", "device", "else", "forward", "hidden_states", "if", "image_embeds", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "masked_lm_loss", "mlm_logits", "mlm_score", "not", "output", "output_attentions", "output_hidden_states", "outputs", "pixel_mask", "pixel_values", "pooled_output", "r", "return", "return_dict", "self", "sequence_output", "shape", "text_features", "text_seq_len", "to", "token_type_ids", "torch", "tuple", "use_return_dict", "view", "vocab_size"], "vilt/modeling_vilt.py:ViltPredictionHeadTransform.__init__": ["ACT2FN", "LayerNorm", "Linear", "ModelPredictionHeadTransform", "Module", "__init__", "class", "config", "def", "dense", "else", "eps", "hidden_act", "hidden_size", "if", "isinstance", "layer_norm_eps", "nn", "self", "str", "super", "transform_act_fn"], "vilt/modeling_vilt.py:ViltPredictionHeadTransform.forward": ["LayerNorm", "ModelPredictionHeadTransform", "Module", "class", "def", "dense", "forward", "hidden_states", "nn", "return", "self", "transform_act_fn"], "vilt/modeling_vilt.py:ViltMLMHead.__init__": ["Linear", "ModelMLMHead", "ModelPredictionHeadTransform", "Module", "__init__", "class", "config", "decoder", "def", "hidden_size", "nn", "self", "super", "transform", "vocab_size"], "vilt/modeling_vilt.py:ViltMLMHead.forward": ["ModelMLMHead", "Module", "class", "decoder", "def", "forward", "nn", "return", "self", "transform", "x"], "vilt/modeling_vilt.py:ViltForQuestionAnswering.__init__": ["GELU", "LayerNorm", "Linear", "Model", "ModelForQuestionAnswering", "ModelModel", "ModelPreTrainedModel", "Sequential", "__init__", "class", "classifier", "config", "def", "hidden_size", "nn", "num_labels", "post_init", "self", "super"], "vilt/modeling_vilt.py:ViltForQuestionAnswering.forward": ["FloatTensor", "LongTensor", "Model", "ModelForQuestionAnswering", "ModelPreTrainedModel", "None", "SequenceClassifierOutput", "attention_mask", "attentions", "binary_cross_entropy_with_logits", "bool", "class", "classifier", "config", "def", "device", "else", "forward", "functional", "hidden_states", "if", "image_embeds", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "nn", "not", "output", "output_attentions", "output_hidden_states", "outputs", "pixel_mask", "pixel_values", "pooler_output", "r", "return", "return_dict", "self", "shape", "to", "token_type_ids", "torch", "tuple", "use_return_dict"], "vilt/modeling_vilt.py:ViltForImageAndTextRetrieval.__init__": ["Linear", "Model", "ModelForImageAndTextRetrieval", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "config", "def", "hidden_size", "nn", "post_init", "rank_output", "self", "super"], "vilt/modeling_vilt.py:ViltForImageAndTextRetrieval.forward": ["FloatTensor", "LongTensor", "Model", "ModelForImageAndTextRetrieval", "ModelPreTrainedModel", "None", "NotImplementedError", "SequenceClassifierOutput", "Training", "attention_mask", "attentions", "bool", "class", "config", "def", "else", "forward", "hidden_states", "if", "image_embeds", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "not", "output", "output_attentions", "output_hidden_states", "outputs", "pixel_mask", "pixel_values", "pooler_output", "r", "raise", "rank_output", "return", "return_dict", "self", "supported", "token_type_ids", "torch", "tuple", "use_return_dict", "yet"], "vilt/modeling_vilt.py:ViltForImagesAndTextClassification.__init__": ["GELU", "LayerNorm", "Linear", "Model", "ModelForImagesAndTextClassification", "ModelModel", "ModelPreTrainedModel", "Sequential", "__init__", "class", "classifier", "config", "def", "hidden_size", "nn", "num_images", "num_labels", "post_init", "self", "super"], "vilt/modeling_vilt.py:ViltForImagesAndTextClassification.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "Make", "Model", "ModelForImagesAndTextClassification", "ModelForImagesAndTextClassificationOutput", "ModelPreTrainedModel", "None", "ValueError", "and", "append", "attention_mask", "attentions", "bool", "cat", "class", "classifier", "config", "def", "device", "dim", "else", "for", "forward", "hidden_states", "i", "if", "image_embeds", "image_token_type_idx", "images", "in", "input", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "match", "model", "ndim", "not", "num_images", "num_labels", "number", "of", "output", "output_attentions", "output_hidden_states", "outputs", "pixel_mask", "pixel_values", "pooled_output", "pooler_output", "pooler_outputs", "r", "raise", "range", "return", "return_dict", "self", "shape", "sure", "the", "to", "token_type_ids", "torch", "tuple", "unsqueeze", "use_return_dict", "view", "with"], "vilt/modeling_vilt.py:ViltForTokenClassification.__init__": ["Dropout", "False", "Linear", "Model", "ModelForTokenClassification", "ModelModel", "ModelPreTrainedModel", "__init__", "add_pooling_layer", "class", "classifier", "config", "def", "dropout", "hidden_dropout_prob", "hidden_size", "nn", "num_labels", "post_init", "self", "super"], "vilt/modeling_vilt.py:ViltForTokenClassification.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "Model", "ModelForTokenClassification", "ModelPreTrainedModel", "None", "TokenClassifierOutput", "attention_mask", "attentions", "bool", "class", "classifier", "config", "def", "device", "dropout", "else", "forward", "hidden_states", "if", "image_embeds", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "not", "num_labels", "output", "output_attentions", "output_hidden_states", "outputs", "pixel_mask", "pixel_values", "r", "return", "return_dict", "self", "sequence_output", "shape", "text_input_size", "to", "token_type_ids", "torch", "tuple", "use_return_dict", "view"], "vipllava/modeling_vipllava.py:VipLlavaMultiModalProjector.__init__": ["ACT2FN", "LayerNorm", "Linear", "ModelConfig", "ModelMultiModalProjector", "Module", "True", "__init__", "act", "bias", "class", "config", "def", "else", "eps", "hidden_size", "if", "int", "isinstance", "len", "linear_1", "linear_2", "nn", "num_feature_layers", "projector_hidden_act", "projector_layernorm", "projector_layernorm_eps", "self", "super", "text_config", "vision_config", "vision_feature_layers"], "vipllava/modeling_vipllava.py:VipLlavaMultiModalProjector.forward": ["ModelMultiModalProjector", "Module", "act", "class", "def", "forward", "hidden_states", "linear_1", "linear_2", "nn", "projector_layernorm", "return", "self"], "vipllava/modeling_vipllava.py:VipLlavaModel.__init__": ["AutoModel", "ModelConfig", "ModelModel", "ModelMultiModalProjector", "ModelPreTrainedModel", "__init__", "class", "config", "def", "from_config", "language_model", "multi_modal_projector", "post_init", "self", "super", "text_config", "vision_config", "vision_tower"], "vipllava/modeling_vipllava.py:VipLlavaModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "language_model", "return", "self"], "vipllava/modeling_vipllava.py:VipLlavaModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "language_model", "self", "set_input_embeddings", "value"], "vipllava/modeling_vipllava.py:VipLlavaModel.get_image_features": ["FloatTensor", "ModelModel", "ModelPreTrainedModel", "None", "True", "cat", "class", "config", "def", "dim", "else", "for", "get_image_features", "hidden_states", "if", "image_features", "image_outputs", "in", "index", "int", "is", "isinstance", "list", "multi_modal_projector", "not", "output_hidden_states", "pixel_values", "return", "self", "torch", "vision_feature_layers", "vision_tower"], "vipllava/modeling_vipllava.py:VipLlavaModel.get_placeholder_mask": ["FloatTensor", "Image", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "ValueError", "all", "and", "class", "config", "def", "device", "do", "dtype", "else", "expand_as", "f", "features", "get_input_embeddings", "get_placeholder_mask", "if", "image", "image_features", "image_token_id", "input_ids", "inputs_embeds", "is", "long", "match", "n_image_features", "n_image_tokens", "not", "numel", "raise", "return", "self", "shape", "special_image_mask", "sum", "tensor", "to", "tokens", "torch", "unsqueeze"], "vipllava/modeling_vipllava.py:VipLlavaModel.forward": ["Cache", "FloatTensor", "LongTensor", "ModelModel", "ModelModelOutputWithPast", "ModelPreTrainedModel", "None", "Tensor", "True", "ValueError", "You", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "device", "dtype", "else", "exactly", "forward", "get_image_features", "get_input_embeddings", "get_placeholder_mask", "hidden_states", "if", "image_features", "image_hidden_states", "input_ids", "inputs_embeds", "int", "is", "language_model", "last_hidden_state", "list", "lm_kwargs", "masked_scatter", "must", "not", "of", "one", "or", "output", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "pixel_values", "position_ids", "r", "raise", "return", "return_dict", "self", "special_image_mask", "specify", "to", "to_tuple", "torch", "tuple", "use_cache", "use_return_dict", "vision_feature_layers"], "vipllava/modeling_vipllava.py:VipLlavaForConditionalGeneration.__init__": ["False", "GenerationMixin", "Linear", "ModelConfig", "ModelForConditionalGeneration", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "text_config", "vocab_size"], "vipllava/modeling_vipllava.py:VipLlavaForConditionalGeneration.get_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "model", "return", "self"], "vipllava/modeling_vipllava.py:VipLlavaForConditionalGeneration.set_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "model", "self", "set_input_embeddings", "value"], "vipllava/modeling_vipllava.py:VipLlavaForConditionalGeneration.get_output_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "Module", "class", "def", "get_output_embeddings", "lm_head", "nn", "return", "self"], "vipllava/modeling_vipllava.py:VipLlavaForConditionalGeneration.get_image_features": ["FloatTensor", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "class", "def", "get_image_features", "int", "list", "model", "pixel_values", "return", "self", "torch", "vision_feature_layers"], "vipllava/modeling_vipllava.py:VipLlavaForConditionalGeneration.forward": ["Cache", "FloatTensor", "GenerationMixin", "LongTensor", "ModelCausalLMOutputWithPast", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Tensor", "True", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "else", "forward", "hidden_states", "if", "image_hidden_states", "input_ids", "inputs_embeds", "int", "is", "isinstance", "labels", "list", "lm_head", "lm_kwargs", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "pixel_values", "position_ids", "r", "return", "return_dict", "self", "slice", "slice_indices", "text_config", "torch", "tuple", "use_cache", "use_return_dict", "vision_feature_layers", "vocab_size"], "vipllava/modeling_vipllava.py:VipLlavaForConditionalGeneration.prepare_inputs_for_generation": ["False", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "True", "attention_mask", "cache_position", "class", "def", "get", "if", "input_ids", "inputs_embeds", "is_first_iteration", "kwargs", "logits_to_keep", "model_inputs", "not", "or", "past_key_values", "pixel_values", "prepare_inputs_for_generation", "return", "self", "super", "use_cache"], "vision_encoder_decoder/modeling_vision_encoder_decoder.py:shift_tokens_right": ["Make", "Model_tokens_right", "Modeled_input_ids", "None", "Tensor", "ValueError", "attribute", "clone", "configuration", "decoder_start_token_id", "def", "if", "input_ids", "int", "is", "masked_fill_", "model", "new_zeros", "of", "pad_token_id", "raise", "return", "s", "set", "shape", "sure", "the", "to", "torch"], "vision_encoder_decoder/modeling_vision_encoder_decoder.py:VisionEncoderDecoderModel.__init__": ["AutoModel", "AutoModelForCausalLM", "Config", "Either", "False", "GenerationMixin", "Got", "Head", "If", "LM", "Linear", "ModelConfig", "ModelModel", "None", "Please", "PreTrainedConfig", "PreTrainedModel", "The", "ValueError", "__class__", "__init__", "_attn_implementation", "_can_compile_fullgraph", "a", "an", "and", "be", "by", "class", "config", "config_class", "configuration", "cross_attention_hidden_size", "decoder", "def", "else", "enc_to_dec_proj", "encoder", "equal", "f", "for", "from_config", "from_encoder_decoder_configs", "get_output_embeddings", "has", "have", "hidden_size", "if", "in", "is", "isinstance", "it", "logger", "model", "nn", "not", "of", "or", "overwritten", "post_init", "provided", "r", "raise", "s", "self", "shared", "should", "specified", "super", "the", "tie_word_embeddings", "to", "to_dict", "type", "use", "warning", "without"], "vision_encoder_decoder/modeling_vision_encoder_decoder.py:VisionEncoderDecoderModel.get_input_embeddings": ["GenerationMixin", "ModelModel", "PreTrainedModel", "class", "decoder", "def", "get_input_embeddings", "return", "self"], "vision_encoder_decoder/modeling_vision_encoder_decoder.py:VisionEncoderDecoderModel.get_output_embeddings": ["GenerationMixin", "ModelModel", "PreTrainedModel", "class", "decoder", "def", "get_output_embeddings", "return", "self"], "vision_encoder_decoder/modeling_vision_encoder_decoder.py:VisionEncoderDecoderModel.set_output_embeddings": ["GenerationMixin", "ModelModel", "PreTrainedModel", "class", "decoder", "def", "new_embeddings", "return", "self", "set_output_embeddings"], "vision_encoder_decoder/modeling_vision_encoder_decoder.py:VisionEncoderDecoderModel.from_encoder_decoder_pretrained": ["AutoConfig", "AutoModel", "AutoModelForCausalLM", "Cross", "Decoder", "False", "GenerationMixin", "If", "In", "Initializing", "ModelConfig", "ModelModel", "None", "PreTrainedModel", "True", "ValueError", "a", "add_cross_attention", "added", "allows", "an", "and", "architecture", "are", "argument", "as", "attention", "attributes", "be", "causal", "class", "cls", "config", "cross", "decoder", "decoder_", "decoder_config", "decoder_model", "decoder_pretrained_model_name_or_path", "def", "defined", "del", "disabled", "do", "encoder", "encoder_", "encoder_config", "encoder_model", "encoder_pretrained_model_name_or_path", "f", "for", "from", "from_encoder_decoder_configs", "from_encoder_decoder_pretrained", "from_pretrained", "has", "if", "in", "info", "initialize", "initialized", "is", "is_decoder", "items", "key", "kwargs", "kwargs_decoder", "kwargs_encoder", "layers", "len", "logger", "make", "mask", "model", "model_args", "not", "of", "or", "order", "pass", "passed", "pop", "r", "raise", "randomly", "return", "return_unused_kwargs", "s", "set", "startswith", "str", "sure", "that", "the", "tie_word_embeddings", "to", "value", "warning"], "vision_encoder_decoder/modeling_vision_encoder_decoder.py:VisionEncoderDecoderModel.forward": ["BaseModelOutput", "BoolTensor", "Cache", "CrossEntropyLoss", "FloatTensor", "GenerationMixin", "LongTensor", "ModelModel", "None", "PreTrainedModel", "Seq2SeqLMOutput", "ValueError", "You", "and", "argument", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "cross_attention_hidden_size", "cross_attentions", "decoder", "decoder_", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_input_ids", "decoder_inputs_embeds", "decoder_outputs", "decoder_start_token_id", "def", "elif", "else", "enc_to_dec_proj", "encoder", "encoder_attention_mask", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "for", "forward", "have", "hidden_size", "hidden_states", "if", "in", "input_ids", "inputs_embeds", "is", "isinstance", "items", "kwargs", "kwargs_decoder", "kwargs_encoder", "labels", "last_hidden_state", "len", "logits", "loss", "loss_fct", "not", "output_attentions", "output_hidden_states", "pad_token_id", "past_key_values", "pixel_values", "r", "raise", "reshape", "return", "return_dict", "self", "shift_tokens_right", "specify", "startswith", "to", "torch", "tuple", "use_cache", "use_return_dict", "value", "vocab_size"], "vision_encoder_decoder/modeling_vision_encoder_decoder.py:VisionEncoderDecoderModel.prepare_decoder_input_ids_from_labels": ["GenerationMixin", "ModelModel", "PreTrainedModel", "Tensor", "class", "config", "decoder_start_token_id", "def", "labels", "pad_token_id", "prepare_decoder_input_ids_from_labels", "return", "self", "shift_tokens_right", "torch"], "vision_text_dual_encoder/modeling_vision_text_dual_encoder.py:contrastive_loss": ["Model_loss", "Tensor", "arange", "cross_entropy", "def", "device", "functional", "len", "logits", "nn", "return", "torch"], "vision_text_dual_encoder/modeling_vision_text_dual_encoder.py:clip_loss": ["Model_loss", "Tensor", "caption_loss", "contrastive_loss", "def", "image_loss", "return", "similarity", "t", "torch"], "vision_text_dual_encoder/modeling_vision_text_dual_encoder.py:VisionTextDualEncoderModel.__init__": ["AutoModel", "CLIPModelConfig", "CLIPModelModel", "Either", "False", "Linear", "Model", "ModelConfig", "ModelModel", "Model_config", "Model_embed_dim", "Model_model", "None", "Parameter", "PreTrainedModel", "ValueError", "__init__", "_attn_implementation", "a", "an", "and", "be", "bias", "class", "config", "config_class", "configuration", "def", "else", "f", "from_Model_text_configs", "from_config", "has", "hidden_size", "if", "is", "isinstance", "logit_scale", "logit_scale_init_value", "model", "nn", "not", "of", "or", "post_init", "projection_dim", "provided", "r", "raise", "self", "super", "tensor", "text", "text_config", "text_embed_dim", "text_model", "text_projection", "to", "torch", "type", "visual_projection"], "vision_text_dual_encoder/modeling_vision_text_dual_encoder.py:VisionTextDualEncoderModel.get_text_features": ["BaseModelOutputWithPooling", "FloatTensor", "ModelModel", "None", "PreTrainedModel", "Tensor", "attention_mask", "class", "def", "get_text_features", "input_ids", "pooler_output", "position_ids", "r", "return", "self", "text_features", "text_model", "text_outputs", "text_projection", "token_type_ids", "torch"], "vision_text_dual_encoder/modeling_vision_text_dual_encoder.py:VisionTextDualEncoderModel.get_image_features": ["FloatTensor", "ModelModel", "Model_model", "Model_outputs", "PreTrainedModel", "Tensor", "class", "def", "get_image_features", "image_features", "pixel_values", "pooler_output", "r", "return", "self", "torch", "visual_projection"], "vision_text_dual_encoder/modeling_vision_text_dual_encoder.py:VisionTextDualEncoderModel.forward": ["CLIPOutput", "FloatTensor", "LongTensor", "ModelModel", "Model_model", "Model_model_output", "Model_outputs", "None", "PreTrainedModel", "T", "Tensor", "True", "attention_mask", "bool", "class", "clip_loss", "config", "def", "dim", "else", "exp", "forward", "if", "image_embeds", "input_ids", "is", "keepdim", "kwargs", "logit_scale", "logits_per_image", "logits_per_text", "loss", "matmul", "norm", "not", "output", "output_attentions", "output_hidden_states", "pixel_values", "position_ids", "r", "return", "return_dict", "return_loss", "self", "t", "text_embeds", "text_model", "text_model_output", "text_outputs", "text_projection", "token_type_ids", "torch", "tuple", "visual_projection"], "vision_text_dual_encoder/modeling_vision_text_dual_encoder.py:VisionTextDualEncoderModel.from_vision_text_pretrained": ["AutoConfig", "AutoModel", "CLIPModelModel", "If", "ModelConfig", "ModelModel", "Model_", "Model_config", "Model_model", "Model_model_name_or_path", "None", "PreTrainedModel", "TRAIN", "The", "ValueError", "You", "a", "able", "an", "and", "are", "argument", "as", "be", "class", "clip", "cls", "config", "def", "defined", "del", "down", "else", "for", "from_Model_text_configs", "from_Model_text_pretrained", "from_pretrained", "has", "if", "in", "inference", "initialized", "is", "it", "items", "key", "kwargs", "kwargs_Model", "kwargs_text", "layer", "len", "logger", "logit", "logit_scale", "model", "model_args", "model_type", "newly", "not", "on", "pop", "predictions", "probably", "projection", "raise", "return", "scale", "should", "startswith", "str", "stream", "task", "text_", "text_config", "text_model", "text_model_name_or_path", "text_projection", "this", "to", "use", "value", "visual_projection", "warning", "weight", "weights"], "visual_bert/modeling_visual_bert.py:VisualBertEmbeddings.__init__": ["Dropout", "Embedding", "False", "LayerNorm", "Linear", "ModelEmbeddings", "Model_embedding_dim", "Model_position_embeddings", "Model_projection", "Model_token_type_embeddings", "Module", "Parameter", "True", "__init__", "arange", "class", "clone", "config", "data", "def", "dropout", "eps", "expand", "hidden_dropout_prob", "hidden_size", "if", "layer_norm_eps", "max_position_embeddings", "nn", "pad_token_id", "padding_idx", "persistent", "position_embeddings", "position_ids", "register_buffer", "requires_grad", "self", "special_Model_initialize", "super", "token_type_embeddings", "torch", "type_vocab_size", "vocab_size", "weight", "word_embeddings"], "visual_bert/modeling_visual_bert.py:VisualBertEmbeddings.forward": ["Found", "LayerNorm", "Model", "ModelEmbeddings", "Model_embeddings", "Model_embeds", "Model_position_embeddings", "Model_position_ids", "Model_projection", "Model_token_type_embeddings", "Model_token_type_ids", "Module", "None", "Setting", "ValueError", "as", "avoid", "be", "by", "cat", "class", "def", "device", "dim", "divide", "dropout", "dtype", "else", "embeddings", "error", "f", "forward", "if", "image_text_alignment", "image_text_alignment_mask", "in", "input_ids", "input_shape", "inputs_embeds", "is", "length", "logger", "long", "nn", "not", "ones", "position", "position_embeddings", "position_ids", "raise", "return", "same", "self", "seq_length", "should", "size", "sum", "the", "them", "to", "token_type_embeddings", "token_type_ids", "torch", "unsqueeze", "values", "warning", "word_embeddings", "zero", "zeros"], "visual_bert/modeling_visual_bert.py:VisualBertSelfAttention.__init__": ["Dropout", "Linear", "ModelSelfAttention", "Module", "The", "ValueError", "__init__", "a", "all_head_size", "and", "attention", "attention_head_size", "attention_probs_dropout_prob", "class", "config", "def", "dropout", "embedding_size", "f", "hasattr", "heads", "hidden", "hidden_size", "if", "int", "is", "key", "multiple", "nn", "not", "num_attention_heads", "number", "of", "query", "raise", "self", "size", "super", "the", "value"], "visual_bert/modeling_visual_bert.py:VisualBertSelfAttention.forward": ["False", "ModelSelfAttention", "Module", "None", "_", "all_head_size", "attention_head_size", "attention_mask", "attention_probs", "attention_scores", "batch_size", "class", "context_layer", "contiguous", "def", "dim", "dropout", "else", "forward", "functional", "hidden_states", "if", "is", "key", "key_layer", "math", "matmul", "new_context_layer_shape", "nn", "not", "num_attention_heads", "output_attentions", "outputs", "permute", "query", "query_layer", "return", "self", "seq_length", "shape", "size", "softmax", "sqrt", "torch", "transpose", "value", "value_layer", "view"], "visual_bert/modeling_visual_bert.py:VisualBertSelfOutput.__init__": ["Dropout", "LayerNorm", "Linear", "ModelSelfOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "eps", "hidden_dropout_prob", "hidden_size", "layer_norm_eps", "nn", "self", "super"], "visual_bert/modeling_visual_bert.py:VisualBertSelfOutput.forward": ["LayerNorm", "ModelSelfOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "visual_bert/modeling_visual_bert.py:VisualBertAttention.__init__": ["ModelAttention", "ModelSelfAttention", "ModelSelfOutput", "Module", "__init__", "class", "config", "def", "nn", "output", "self", "super"], "visual_bert/modeling_visual_bert.py:VisualBertAttention.forward": ["False", "ModelAttention", "Module", "None", "attention_mask", "attention_output", "class", "def", "forward", "hidden_states", "nn", "output", "output_attentions", "outputs", "return", "self", "self_outputs"], "visual_bert/modeling_visual_bert.py:VisualBertIntermediate.__init__": ["ACT2FN", "Linear", "ModelIntermediate", "Module", "__init__", "class", "config", "def", "dense", "else", "hidden_act", "hidden_size", "if", "intermediate_act_fn", "intermediate_size", "isinstance", "nn", "self", "str", "super"], "visual_bert/modeling_visual_bert.py:VisualBertIntermediate.forward": ["ModelIntermediate", "Module", "Tensor", "class", "def", "dense", "forward", "hidden_states", "intermediate_act_fn", "nn", "return", "self", "torch"], "visual_bert/modeling_visual_bert.py:VisualBertOutput.__init__": ["Dropout", "LayerNorm", "Linear", "ModelOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "eps", "hidden_dropout_prob", "hidden_size", "intermediate_size", "layer_norm_eps", "nn", "self", "super"], "visual_bert/modeling_visual_bert.py:VisualBertOutput.forward": ["LayerNorm", "ModelOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "visual_bert/modeling_visual_bert.py:VisualBertLayer.__init__": ["GradientCheckpointingLayer", "ModelAttention", "ModelIntermediate", "ModelLayer", "ModelOutput", "__init__", "attention", "chunk_size_feed_forward", "class", "config", "def", "intermediate", "output", "self", "seq_len_dim", "super"], "visual_bert/modeling_visual_bert.py:VisualBertLayer.forward": ["False", "GradientCheckpointingLayer", "ModelLayer", "None", "apply_chunking_to_forward", "attention", "attention_mask", "attention_output", "chunk_size_feed_forward", "class", "def", "feed_forward_chunk", "forward", "hidden_states", "layer_output", "output_attentions", "outputs", "return", "self", "self_attention_outputs", "seq_len_dim"], "visual_bert/modeling_visual_bert.py:VisualBertLayer.feed_forward_chunk": ["GradientCheckpointingLayer", "ModelLayer", "attention_output", "class", "def", "feed_forward_chunk", "intermediate", "intermediate_output", "layer_output", "output", "return", "self"], "visual_bert/modeling_visual_bert.py:VisualBertEncoder.__init__": ["False", "ModelEncoder", "ModelLayer", "Module", "ModuleList", "_", "__init__", "class", "config", "def", "for", "gradient_checkpointing", "in", "layer", "nn", "num_hidden_layers", "range", "self", "super"], "visual_bert/modeling_visual_bert.py:VisualBertEncoder.forward": ["BaseModelOutput", "False", "ModelEncoder", "Module", "None", "True", "all_hidden_states", "all_self_attentions", "attention_mask", "attentions", "class", "def", "else", "enumerate", "for", "forward", "hidden_states", "i", "if", "in", "is", "last_hidden_state", "layer", "layer_module", "layer_outputs", "nn", "not", "output_attentions", "output_hidden_states", "return", "return_dict", "self", "tuple", "v"], "visual_bert/modeling_visual_bert.py:VisualBertPooler.__init__": ["Linear", "ModelPooler", "Module", "Tanh", "__init__", "activation", "class", "config", "def", "dense", "hidden_size", "nn", "self", "super"], "visual_bert/modeling_visual_bert.py:VisualBertPooler.forward": ["ModelPooler", "Module", "Tensor", "activation", "class", "def", "dense", "first_token_tensor", "forward", "hidden_states", "nn", "pooled_output", "return", "self", "torch"], "visual_bert/modeling_visual_bert.py:VisualBertPredictionHeadTransform.__init__": ["ACT2FN", "LayerNorm", "Linear", "ModelPredictionHeadTransform", "Module", "__init__", "class", "config", "def", "dense", "else", "eps", "hidden_act", "hidden_size", "if", "isinstance", "layer_norm_eps", "nn", "self", "str", "super", "transform_act_fn"], "visual_bert/modeling_visual_bert.py:VisualBertPredictionHeadTransform.forward": ["LayerNorm", "ModelPredictionHeadTransform", "Module", "Tensor", "class", "def", "dense", "forward", "hidden_states", "nn", "return", "self", "torch", "transform_act_fn"], "visual_bert/modeling_visual_bert.py:VisualBertLMPredictionHead.__init__": ["Linear", "ModelLMPredictionHead", "ModelPredictionHeadTransform", "Module", "Parameter", "True", "__init__", "bias", "class", "config", "decoder", "def", "hidden_size", "nn", "self", "super", "torch", "transform", "vocab_size", "zeros"], "visual_bert/modeling_visual_bert.py:VisualBertLMPredictionHead.forward": ["ModelLMPredictionHead", "Module", "class", "decoder", "def", "forward", "hidden_states", "nn", "return", "self", "transform"], "visual_bert/modeling_visual_bert.py:VisualBertPreTrainingHeads.__init__": ["Linear", "ModelLMPredictionHead", "ModelPreTrainingHeads", "Module", "__init__", "class", "config", "def", "hidden_size", "nn", "predictions", "self", "seq_relationship", "super"], "visual_bert/modeling_visual_bert.py:VisualBertPreTrainingHeads.forward": ["ModelPreTrainingHeads", "Module", "class", "def", "forward", "nn", "pooled_output", "prediction_scores", "predictions", "return", "self", "seq_relationship", "seq_relationship_score", "sequence_output"], "visual_bert/modeling_visual_bert.py:VisualBertPreTrainedModel._init_weights": ["Embedding", "LayerNorm", "Linear", "ModelEmbeddings", "ModelLMPredictionHead", "ModelPreTrainedModel", "None", "PreTrainedModel", "_init_weights", "and", "arange", "bias", "class", "config", "copy_", "def", "elif", "expand", "hasattr", "if", "init", "initializer_range", "is", "isinstance", "mean", "module", "nn", "normal_", "not", "ones_", "position_ids", "self", "shape", "std", "torch", "weight", "zeros_"], "visual_bert/modeling_visual_bert.py:VisualBertModel.__init__": ["ModelEmbeddings", "ModelEncoder", "ModelLayer", "ModelModel", "ModelPooler", "ModelPreTrainedModel", "None", "True", "__init__", "add_pooling_layer", "additional_layer", "bypass_transformer", "class", "config", "def", "else", "embeddings", "encoder", "if", "pooler", "post_init", "r", "self", "super"], "visual_bert/modeling_visual_bert.py:VisualBertModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "return", "self", "word_embeddings"], "visual_bert/modeling_visual_bert.py:VisualBertModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "self", "set_input_embeddings", "value", "word_embeddings"], "visual_bert/modeling_visual_bert.py:VisualBertModel.forward": ["BaseModelOutputWithPooling", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "Model_attention_mask", "Model_embedding_output", "Model_embeds", "Model_input_shape", "Model_token_type_ids", "None", "Tensor", "ValueError", "You", "additional_layer", "and", "at", "attention_mask", "attentions", "batch_size", "bool", "both", "bypass_transformer", "cannot", "cat", "class", "combined_attention_mask", "concatenated_input", "config", "def", "device", "dim", "either", "elif", "else", "embedding_output", "embeddings", "encoded_outputs", "encoder", "encoder_outputs", "extended_attention_mask", "forward", "get_extended_attention_mask", "have", "hidden_states", "if", "image_text_alignment", "input_ids", "input_shape", "inputs_embeds", "is", "kwargs", "last_hidden_state", "not", "ones", "or", "output_attentions", "output_hidden_states", "pooled_output", "pooler", "pooler_output", "position_ids", "r", "raise", "return", "return_dict", "same", "self", "seq_length", "sequence_output", "size", "specify", "text_embedding_output", "text_extended_attention_mask", "text_length", "the", "time", "to", "token_type_ids", "torch", "tuple", "use_return_dict", "warn_if_padding_and_no_attention_mask"], "visual_bert/modeling_visual_bert.py:VisualBertForPreTraining.__init__": ["Model", "ModelForPreTraining", "ModelModel", "ModelPreTrainedModel", "ModelPreTrainingHeads", "__init__", "class", "cls", "config", "def", "post_init", "self", "super"], "visual_bert/modeling_visual_bert.py:VisualBertForPreTraining.get_output_embeddings": ["ModelForPreTraining", "ModelPreTrainedModel", "class", "cls", "decoder", "def", "get_output_embeddings", "predictions", "return", "self"], "visual_bert/modeling_visual_bert.py:VisualBertForPreTraining.set_output_embeddings": ["ModelForPreTraining", "ModelPreTrainedModel", "bias", "class", "cls", "decoder", "def", "new_embeddings", "predictions", "self", "set_output_embeddings"], "visual_bert/modeling_visual_bert.py:VisualBertForPreTraining.forward": ["CrossEntropyLoss", "FloatTensor", "Found", "LongTensor", "Model", "ModelForPreTraining", "ModelForPreTrainingOutput", "ModelPreTrainedModel", "Model_attention_mask", "Model_embeds", "Model_token_type_ids", "None", "Tensor", "The", "ValueError", "and", "as", "attention", "attention_mask", "attentions", "bool", "class", "cls", "config", "def", "elif", "else", "expected", "f", "forward", "have", "hidden_states", "if", "image_text_alignment", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "length", "loss", "loss_fct", "mask", "masked_lm_loss", "not", "output", "output_attentions", "output_hidden_states", "outputs", "pooled_output", "position_ids", "prediction_logits", "prediction_scores", "provided", "r", "raise", "return", "return_dict", "same", "self", "sentence_image_labels", "sentence_image_loss", "seq_relationship_logits", "seq_relationship_score", "sequence", "sequence_output", "should", "size", "token_type_ids", "torch", "total", "total_loss", "total_size", "tuple", "use_return_dict", "view", "vocab_size", "with"], "visual_bert/modeling_visual_bert.py:VisualBertForMultipleChoice.__init__": ["Dropout", "Linear", "Model", "ModelForMultipleChoice", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "cls", "config", "def", "dropout", "hidden_dropout_prob", "hidden_size", "nn", "post_init", "self", "super"], "visual_bert/modeling_visual_bert.py:VisualBertForMultipleChoice.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "Model", "ModelForMultipleChoice", "ModelPreTrainedModel", "Model_attention_mask", "Model_embeds", "Model_token_type_ids", "MultipleChoiceModelOutput", "None", "Tensor", "_", "attention_mask", "attentions", "bool", "class", "cls", "config", "def", "dropout", "else", "forward", "hidden_states", "if", "image_text_alignment", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "not", "num_choices", "output", "output_attentions", "output_hidden_states", "outputs", "pooled_output", "position_ids", "r", "reshaped_logits", "return", "return_dict", "self", "shape", "size", "token_type_ids", "torch", "tuple", "use_return_dict", "view"], "visual_bert/modeling_visual_bert.py:VisualBertForQuestionAnswering.__init__": ["Dropout", "Linear", "Model", "ModelForQuestionAnswering", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "cls", "config", "def", "dropout", "hidden_dropout_prob", "hidden_size", "nn", "num_labels", "post_init", "self", "super"], "visual_bert/modeling_visual_bert.py:VisualBertForQuestionAnswering.forward": ["FloatTensor", "KLDivLoss", "LogSoftmax", "LongTensor", "Model", "ModelForQuestionAnswering", "ModelPreTrainedModel", "Model_attention_mask", "Model_embeds", "Model_token_type_ids", "None", "SequenceClassifierOutput", "Tensor", "attention_mask", "attentions", "batchmean", "bool", "class", "cls", "config", "contiguous", "def", "dim", "dropout", "else", "expand", "forward", "gather", "hidden_states", "if", "image_text_alignment", "index_to_gather", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "log_softmax", "logits", "loss", "loss_fct", "nn", "not", "num_labels", "output", "output_attentions", "output_hidden_states", "outputs", "pooled_output", "position_ids", "r", "reduction", "reshaped_logits", "return", "return_dict", "self", "sequence_output", "size", "sum", "token_type_ids", "torch", "tuple", "unsqueeze", "use_return_dict", "view"], "visual_bert/modeling_visual_bert.py:VisualBertForVisualReasoning.__init__": ["Dropout", "Linear", "Model", "ModelForModelReasoning", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "cls", "config", "def", "dropout", "hidden_dropout_prob", "hidden_size", "nn", "num_labels", "post_init", "self", "super"], "visual_bert/modeling_visual_bert.py:VisualBertForVisualReasoning.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "Model", "ModelForModelReasoning", "ModelPreTrainedModel", "Model_attention_mask", "Model_embeds", "Model_token_type_ids", "None", "SequenceClassifierOutput", "Tensor", "attention_mask", "attentions", "bool", "class", "cls", "config", "contiguous", "def", "dropout", "else", "forward", "hidden_states", "if", "image_text_alignment", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "not", "output", "output_attentions", "output_hidden_states", "outputs", "pooled_output", "position_ids", "r", "reshaped_logits", "return", "return_dict", "self", "token_type_ids", "torch", "tuple", "use_return_dict", "view"], "visual_bert/modeling_visual_bert.py:VisualBertRegionToPhraseAttention.__init__": ["Dropout", "Linear", "ModelRegionToPhraseAttention", "Module", "The", "ValueError", "__init__", "a", "all_head_size", "attention", "attention_head_size", "attention_probs_dropout_prob", "class", "config", "def", "dropout", "f", "heads", "hidden", "hidden_size", "if", "int", "is", "key", "multiple", "nn", "not", "num_attention_heads", "number", "of", "query", "raise", "self", "size", "super", "the", "value"], "visual_bert/modeling_visual_bert.py:VisualBertRegionToPhraseAttention.forward": ["ModelRegionToPhraseAttention", "Module", "_", "attention_head_size", "attention_mask", "attention_scores", "batch_size", "class", "def", "dtype", "finfo", "forward", "key", "key_layer", "math", "matmul", "min", "nn", "num_attention_heads", "query", "query_layer", "return", "self", "seq_length", "shape", "sqrt", "squeeze", "to", "torch", "transpose", "unsqueeze", "view"], "visual_bert/modeling_visual_bert.py:VisualBertForRegionToPhraseAlignment.__init__": ["Dropout", "Model", "ModelForRegionToPhraseAlignment", "ModelModel", "ModelPreTrainedModel", "ModelPreTrainingHeads", "ModelRegionToPhraseAttention", "__init__", "attention", "class", "cls", "config", "def", "dropout", "hidden_dropout_prob", "nn", "post_init", "self", "super"], "visual_bert/modeling_visual_bert.py:VisualBertForRegionToPhraseAlignment.forward": ["Flickr", "FloatTensor", "KLDivLoss", "LogSoftmax", "LongTensor", "Model", "ModelForRegionToPhraseAlignment", "ModelPreTrainedModel", "Model_attention_mask", "Model_embeds", "Model_features", "Model_token_type_ids", "None", "SequenceClassifierOutput", "Tensor", "ValueError", "as", "attention", "attention_mask", "attentions", "batchmean", "be", "bool", "class", "config", "contiguous", "def", "dim", "else", "expand", "expanded_region_to_phrase_positions", "f", "features", "forward", "gather", "hidden_states", "if", "image_text_alignment", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "length", "log_softmax", "logits", "long", "loss", "loss_fct", "mask", "not", "output", "output_attentions", "output_hidden_states", "outputs", "position_ids", "r", "raise", "reduction", "region_to_phrase_position", "region_to_phrase_position_mask", "return", "return_dict", "same", "scores", "selected_positions", "self", "sequence_output", "should", "size", "the", "token_type_ids", "torch", "tuple", "unsqueeze", "use_return_dict", "using", "when"], "vit/modeling_vit.py:ViTEmbeddings.__init__": ["Dropout", "False", "ModelConfig", "ModelEmbeddings", "ModelPatchEmbeddings", "Module", "None", "Parameter", "__init__", "bool", "class", "cls_token", "config", "def", "dropout", "else", "hidden_dropout_prob", "hidden_size", "if", "mask_token", "nn", "num_patches", "patch_embeddings", "patch_size", "position_embeddings", "randn", "self", "super", "torch", "use_mask_token", "zeros"], "vit/modeling_vit.py:ViTEmbeddings.interpolate_pos_encoding": ["False", "ModelEmbeddings", "Module", "Tensor", "align_corners", "and", "bicubic", "cat", "class", "class_pos_embed", "def", "dim", "embeddings", "functional", "height", "if", "int", "interpolate", "interpolate_pos_encoding", "is_tracing", "jit", "mode", "new_height", "new_width", "nn", "not", "num_patches", "num_positions", "patch_pos_embed", "patch_size", "permute", "position_embeddings", "reshape", "return", "self", "shape", "size", "sqrt_num_positions", "torch", "torch_int", "view", "width"], "vit/modeling_vit.py:ViTEmbeddings.forward": ["BoolTensor", "False", "ModelEmbeddings", "Module", "None", "Tensor", "batch_size", "bool", "bool_masked_pos", "cat", "class", "cls_token", "cls_tokens", "def", "dim", "dropout", "else", "embeddings", "expand", "forward", "height", "if", "interpolate_pos_encoding", "is", "mask", "mask_token", "mask_tokens", "nn", "not", "num_channels", "patch_embeddings", "pixel_values", "position_embeddings", "return", "self", "seq_length", "shape", "torch", "type_as", "unsqueeze", "width"], "vit/modeling_vit.py:ViTPatchEmbeddings.__init__": ["Conv2d", "Iterable", "ModelConfig", "ModelPatchEmbeddings", "Module", "__init__", "abc", "class", "collections", "config", "def", "else", "hidden_size", "if", "image_size", "isinstance", "kernel_size", "nn", "num_channels", "num_patches", "patch_size", "projection", "self", "stride", "super"], "vit/modeling_vit.py:ViTPatchEmbeddings.forward": ["Expected", "False", "Input", "Make", "ModelPatchEmbeddings", "Module", "Tensor", "ValueError", "batch_size", "bool", "but", "channel", "class", "configuration", "def", "dimension", "doesn", "embeddings", "f", "flatten", "forward", "got", "height", "if", "image", "image_size", "in", "interpolate_pos_encoding", "match", "model", "nn", "not", "num_channels", "of", "one", "or", "pixel", "pixel_values", "projection", "raise", "return", "self", "set", "shape", "size", "sure", "t", "that", "the", "torch", "transpose", "values", "width", "with"], "vit/modeling_vit.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "contiguous", "def", "dim", "dropout", "float", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "shape", "size", "softmax", "torch", "training", "transpose", "value"], "vit/modeling_vit.py:ViTSelfAttention.__init__": ["False", "Linear", "ModelConfig", "ModelSelfAttention", "Module", "The", "ValueError", "__init__", "a", "all_head_size", "and", "attention", "attention_head_size", "attention_probs_dropout_prob", "bias", "class", "config", "def", "dropout_prob", "embedding_size", "f", "hasattr", "heads", "hidden", "hidden_size", "if", "int", "is", "is_causal", "key", "multiple", "nn", "not", "num_attention_heads", "number", "of", "qkv_bias", "query", "raise", "scaling", "self", "size", "super", "the", "value"], "vit/modeling_vit.py:ViTSelfAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "ModelSelfAttention", "Module", "None", "Tensor", "_attn_implementation", "all_head_size", "attention_head_size", "attention_interface", "attention_probs", "batch_size", "class", "config", "context_layer", "def", "dropout", "dropout_prob", "eager", "eager_attention_forward", "else", "forward", "hidden_states", "if", "is_causal", "key", "key_layer", "new_context_layer_shape", "new_shape", "nn", "not", "num_attention_heads", "query", "query_layer", "reshape", "return", "scaling", "self", "shape", "size", "torch", "training", "transpose", "tuple", "value", "value_layer", "view"], "vit/modeling_vit.py:ViTSelfOutput.__init__": ["Dropout", "Linear", "ModelConfig", "ModelSelfOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "hidden_dropout_prob", "hidden_size", "nn", "self", "super"], "vit/modeling_vit.py:ViTSelfOutput.forward": ["ModelSelfOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "vit/modeling_vit.py:ViTAttention.__init__": ["ModelAttention", "ModelConfig", "ModelSelfAttention", "ModelSelfOutput", "Module", "__init__", "attention", "class", "config", "def", "nn", "output", "self", "super"], "vit/modeling_vit.py:ViTAttention.forward": ["ModelAttention", "Module", "Tensor", "_", "attention", "class", "def", "forward", "hidden_states", "nn", "output", "return", "self", "self_attn_output", "torch"], "vit/modeling_vit.py:ViTIntermediate.__init__": ["ACT2FN", "Linear", "ModelConfig", "ModelIntermediate", "Module", "__init__", "class", "config", "def", "dense", "else", "hidden_act", "hidden_size", "if", "intermediate_act_fn", "intermediate_size", "isinstance", "nn", "self", "str", "super"], "vit/modeling_vit.py:ViTIntermediate.forward": ["ModelIntermediate", "Module", "Tensor", "class", "def", "dense", "forward", "hidden_states", "intermediate_act_fn", "nn", "return", "self", "torch"], "vit/modeling_vit.py:ViTOutput.__init__": ["Dropout", "Linear", "ModelConfig", "ModelOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "hidden_dropout_prob", "hidden_size", "intermediate_size", "nn", "self", "super"], "vit/modeling_vit.py:ViTOutput.forward": ["ModelOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "vit/modeling_vit.py:ViTLayer.__init__": ["GradientCheckpointingLayer", "LayerNorm", "ModelAttention", "ModelConfig", "ModelIntermediate", "ModelLayer", "ModelOutput", "__init__", "attention", "chunk_size_feed_forward", "class", "config", "def", "eps", "hidden_size", "intermediate", "layer_norm_eps", "layernorm_after", "layernorm_before", "nn", "output", "self", "seq_len_dim", "super"], "vit/modeling_vit.py:ViTLayer.forward": ["GradientCheckpointingLayer", "ModelLayer", "Tensor", "attention", "attention_output", "class", "def", "forward", "hidden_states", "hidden_states_norm", "intermediate", "layer_output", "layernorm_after", "layernorm_before", "output", "return", "self", "torch"], "vit/modeling_vit.py:ViTEncoder.__init__": ["False", "ModelConfig", "ModelEncoder", "ModelLayer", "Module", "ModuleList", "_", "__init__", "class", "config", "def", "for", "gradient_checkpointing", "in", "layer", "nn", "num_hidden_layers", "range", "self", "super"], "vit/modeling_vit.py:ViTEncoder.forward": ["BaseModelOutput", "ModelEncoder", "Module", "Tensor", "class", "def", "enumerate", "for", "forward", "hidden_states", "i", "in", "last_hidden_state", "layer", "layer_module", "nn", "return", "self", "torch"], "vit/modeling_vit.py:ViTPreTrainedModel._init_weights": ["Conv2d", "LayerNorm", "Linear", "ModelEmbeddings", "ModelPreTrainedModel", "None", "PreTrainedModel", "_init_weights", "bias", "class", "cls_token", "config", "def", "elif", "if", "init", "initializer_range", "is", "isinstance", "mask_token", "mean", "module", "nn", "not", "ones_", "position_embeddings", "self", "std", "trunc_normal_", "weight", "zeros_"], "vit/modeling_vit.py:ViTModel.__init__": ["False", "LayerNorm", "ModelConfig", "ModelEmbeddings", "ModelEncoder", "ModelModel", "ModelPooler", "ModelPreTrainedModel", "None", "True", "__init__", "add_pooling_layer", "bool", "class", "config", "def", "else", "embeddings", "encoder", "eps", "hidden_size", "if", "layer_norm_eps", "layernorm", "nn", "pooler", "post_init", "r", "self", "super", "use_mask_token"], "vit/modeling_vit.py:ViTModel.get_input_embeddings": ["ModelModel", "ModelPatchEmbeddings", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "patch_embeddings", "return", "self"], "vit/modeling_vit.py:ViTModel.forward": ["BaseModelOutput", "BaseModelOutputWithPooling", "BoolTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "bool", "bool_masked_pos", "class", "def", "dtype", "else", "embedding_output", "embeddings", "encoder", "encoder_outputs", "expected_dtype", "forward", "have", "if", "interpolate_pos_encoding", "is", "kwargs", "last_hidden_state", "layernorm", "not", "patch_embeddings", "pixel_values", "pooled_output", "pooler", "pooler_output", "projection", "r", "raise", "return", "self", "sequence_output", "specify", "to", "torch", "weight"], "vit/modeling_vit.py:ViTPooler.__init__": ["ACT2FN", "Linear", "ModelConfig", "ModelPooler", "Module", "__init__", "activation", "class", "config", "def", "dense", "hidden_size", "nn", "pooler_act", "pooler_output_size", "self", "super"], "vit/modeling_vit.py:ViTPooler.forward": ["ModelPooler", "Module", "Tensor", "activation", "class", "def", "dense", "first_token_tensor", "forward", "hidden_states", "nn", "pooled_output", "return", "self", "torch"], "vit/modeling_vit.py:ViTForMaskedImageModeling.__init__": ["Conv2d", "False", "Model", "ModelConfig", "ModelForMaskedImageModeling", "ModelModel", "ModelPreTrainedModel", "PixelShuffle", "Sequential", "True", "__init__", "add_pooling_layer", "class", "config", "decoder", "def", "encoder_stride", "hidden_size", "in_channels", "kernel_size", "nn", "num_channels", "out_channels", "post_init", "self", "super", "use_mask_token"], "vit/modeling_vit.py:ViTForMaskedImageModeling.forward": ["BaseModelOutputWithPooling", "BoolTensor", "Got", "MaskedImageModelingOutput", "Model", "ModelForMaskedImageModeling", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "When", "and", "as", "attentions", "batch_size", "be", "bool", "bool_masked_pos", "class", "config", "contiguous", "decoder", "def", "dimensions", "encoder_stride", "ensure", "equal", "f", "floor", "forward", "functional", "has", "height", "hidden_states", "if", "image", "image_size", "input", "interpolate_pos_encoding", "is", "kwargs", "l1_loss", "last_hidden_state", "loss", "mask", "masked_im_loss", "math", "must", "nn", "none", "not", "num_channels", "outputs", "patch_size", "permute", "pixel_values", "provided", "r", "raise", "reconstructed", "reconstructed_pixel_values", "reconstruction", "reconstruction_loss", "reduction", "repeat_interleave", "reshape", "return", "same", "self", "sequence_length", "sequence_output", "shape", "size", "sum", "that", "the", "to", "torch", "unsqueeze", "width"], "vit/modeling_vit.py:ViTForImageClassification.__init__": ["False", "Identity", "Linear", "Model", "ModelConfig", "ModelForImageClassification", "ModelModel", "ModelPreTrainedModel", "__init__", "add_pooling_layer", "class", "classifier", "config", "def", "else", "hidden_size", "if", "nn", "num_labels", "post_init", "self", "super"], "vit/modeling_vit.py:ViTForImageClassification.forward": ["BaseModelOutputWithPooling", "ImageClassifierOutput", "Model", "ModelForImageClassification", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "attentions", "bool", "class", "classifier", "config", "def", "forward", "hidden_states", "if", "interpolate_pos_encoding", "is", "kwargs", "labels", "last_hidden_state", "logits", "loss", "loss_function", "not", "outputs", "pixel_values", "pooled_output", "r", "return", "self", "sequence_output", "torch"], "vit_mae/modeling_vit_mae.py:get_2d_sincos_pos_embed": ["False", "Model_2d_sincos_pos_embed", "Model_2d_sincos_pos_embed_from_grid", "add_cls_token", "arange", "axis", "concatenate", "def", "dtype", "embed_dim", "float32", "grid", "grid_h", "grid_size", "grid_w", "if", "meshgrid", "np", "pos_embed", "reshape", "return", "stack", "zeros"], "vit_mae/modeling_vit_mae.py:get_2d_sincos_pos_embed_from_grid": ["Model_1d_sincos_pos_embed_from_grid", "Model_2d_sincos_pos_embed_from_grid", "ValueError", "axis", "be", "concatenate", "def", "emb", "emb_h", "emb_w", "embed_dim", "even", "grid", "if", "must", "np", "raise", "return"], "vit_mae/modeling_vit_mae.py:get_1d_sincos_pos_embed_from_grid": ["Model_1d_sincos_pos_embed_from_grid", "ValueError", "arange", "axis", "be", "concatenate", "cos", "d", "def", "dtype", "einsum", "emb", "emb_cos", "emb_sin", "embed_dim", "even", "float", "if", "m", "md", "must", "np", "omega", "out", "pos", "raise", "reshape", "return", "sin"], "vit_mae/modeling_vit_mae.py:ViTMAEEmbeddings.__init__": ["False", "ModelEmbeddings", "ModelPatchEmbeddings", "Module", "Parameter", "__init__", "class", "cls_token", "config", "def", "hidden_size", "nn", "num_patches", "patch_embeddings", "patch_size", "position_embeddings", "requires_grad", "self", "super", "torch", "zeros"], "vit_mae/modeling_vit_mae.py:ViTMAEEmbeddings.initialize_weights": ["False", "ModelEmbeddings", "Module", "True", "_is_hf_initialized", "add_cls_token", "class", "cls_token", "config", "copy_", "def", "float", "from_numpy", "get_2d_sincos_pos_embed", "getattr", "if", "init", "initialize_weights", "initializer_range", "int", "nn", "normal_", "num_patches", "patch_embeddings", "pos_embed", "position_embeddings", "projection", "return", "self", "shape", "std", "torch", "unsqueeze", "view", "w", "weight", "xavier_uniform_"], "vit_mae/modeling_vit_mae.py:ViTMAEEmbeddings.interpolate_pos_encoding": ["False", "ModelEmbeddings", "Module", "Tensor", "align_corners", "and", "bicubic", "cat", "class", "class_pos_embed", "def", "dim", "embeddings", "functional", "height", "if", "int", "interpolate", "interpolate_pos_encoding", "is_tracing", "jit", "mode", "new_height", "new_width", "nn", "not", "num_patches", "num_positions", "patch_pos_embed", "patch_size", "permute", "position_embeddings", "reshape", "return", "self", "shape", "size", "sqrt_num_positions", "torch", "torch_int", "view", "width"], "vit_mae/modeling_vit_mae.py:ViTMAEEmbeddings.random_masking": ["ModelEmbeddings", "Module", "None", "argsort", "batch_size", "class", "config", "def", "device", "dim", "gather", "ids_keep", "ids_restore", "ids_shuffle", "if", "index", "int", "is", "len_keep", "mask", "mask_ratio", "nn", "noise", "ones", "rand", "random_masking", "repeat", "return", "self", "seq_length", "sequence", "sequence_unmasked", "shape", "to", "torch", "unsqueeze"], "vit_mae/modeling_vit_mae.py:ViTMAEEmbeddings.forward": ["False", "ModelEmbeddings", "Module", "None", "batch_size", "bool", "cat", "class", "cls_token", "cls_tokens", "def", "dim", "else", "embeddings", "expand", "forward", "height", "ids_restore", "if", "interpolate_pos_encoding", "mask", "nn", "noise", "num_channels", "patch_embeddings", "pixel_values", "position_embeddings", "random_masking", "return", "self", "shape", "torch", "width"], "vit_mae/modeling_vit_mae.py:ViTMAEPatchEmbeddings.__init__": ["Conv2d", "Iterable", "ModelPatchEmbeddings", "Module", "__init__", "abc", "class", "collections", "config", "def", "else", "hidden_size", "if", "image_size", "isinstance", "kernel_size", "nn", "num_channels", "num_patches", "patch_size", "projection", "self", "stride", "super"], "vit_mae/modeling_vit_mae.py:ViTMAEPatchEmbeddings.forward": ["False", "Input", "Make", "ModelPatchEmbeddings", "Module", "ValueError", "and", "batch_size", "bool", "channel", "class", "configuration", "def", "dimension", "doesn", "f", "flatten", "forward", "height", "if", "image", "image_size", "in", "interpolate_pos_encoding", "match", "model", "nn", "not", "num_channels", "of", "one", "or", "pixel", "pixel_values", "projection", "raise", "return", "self", "set", "shape", "size", "sure", "t", "that", "the", "transpose", "values", "width", "with", "x"], "vit_mae/modeling_vit_mae.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "contiguous", "def", "dim", "dropout", "float", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "shape", "size", "softmax", "torch", "training", "transpose", "value"], "vit_mae/modeling_vit_mae.py:ViTMAESelfAttention.__init__": ["False", "Linear", "ModelConfig", "ModelSelfAttention", "Module", "The", "ValueError", "__init__", "a", "all_head_size", "and", "attention", "attention_head_size", "attention_probs_dropout_prob", "bias", "class", "config", "def", "dropout_prob", "embedding_size", "f", "hasattr", "heads", "hidden", "hidden_size", "if", "int", "is", "is_causal", "key", "multiple", "nn", "not", "num_attention_heads", "number", "of", "qkv_bias", "query", "raise", "scaling", "self", "size", "super", "the", "value"], "vit_mae/modeling_vit_mae.py:ViTMAESelfAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "ModelSelfAttention", "Module", "None", "Tensor", "_attn_implementation", "all_head_size", "attention_head_size", "attention_interface", "attention_probs", "batch_size", "class", "config", "context_layer", "def", "dropout", "dropout_prob", "eager", "eager_attention_forward", "else", "forward", "hidden_states", "if", "is_causal", "key", "key_layer", "new_context_layer_shape", "new_shape", "nn", "not", "num_attention_heads", "query", "query_layer", "reshape", "return", "scaling", "self", "shape", "size", "torch", "training", "transpose", "tuple", "value", "value_layer", "view"], "vit_mae/modeling_vit_mae.py:ViTMAESelfOutput.__init__": ["Dropout", "Linear", "ModelConfig", "ModelSelfOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "hidden_dropout_prob", "hidden_size", "nn", "self", "super"], "vit_mae/modeling_vit_mae.py:ViTMAESelfOutput.forward": ["ModelSelfOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "vit_mae/modeling_vit_mae.py:ViTMAEAttention.__init__": ["ModelAttention", "ModelConfig", "ModelSelfAttention", "ModelSelfOutput", "Module", "__init__", "attention", "class", "config", "def", "nn", "output", "self", "super"], "vit_mae/modeling_vit_mae.py:ViTMAEAttention.forward": ["ModelAttention", "Module", "Tensor", "_", "attention", "class", "def", "forward", "hidden_states", "nn", "output", "return", "self", "self_attn_output", "torch"], "vit_mae/modeling_vit_mae.py:ViTMAEIntermediate.__init__": ["ACT2FN", "Linear", "ModelConfig", "ModelIntermediate", "Module", "__init__", "class", "config", "def", "dense", "else", "hidden_act", "hidden_size", "if", "intermediate_act_fn", "intermediate_size", "isinstance", "nn", "self", "str", "super"], "vit_mae/modeling_vit_mae.py:ViTMAEIntermediate.forward": ["ModelIntermediate", "Module", "Tensor", "class", "def", "dense", "forward", "hidden_states", "intermediate_act_fn", "nn", "return", "self", "torch"], "vit_mae/modeling_vit_mae.py:ViTMAEOutput.__init__": ["Dropout", "Linear", "ModelConfig", "ModelOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "hidden_dropout_prob", "hidden_size", "intermediate_size", "nn", "self", "super"], "vit_mae/modeling_vit_mae.py:ViTMAEOutput.forward": ["ModelOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "vit_mae/modeling_vit_mae.py:ViTMAELayer.__init__": ["GradientCheckpointingLayer", "LayerNorm", "ModelAttention", "ModelConfig", "ModelIntermediate", "ModelLayer", "ModelOutput", "__init__", "attention", "chunk_size_feed_forward", "class", "config", "def", "eps", "hidden_size", "intermediate", "layer_norm_eps", "layernorm_after", "layernorm_before", "nn", "output", "self", "seq_len_dim", "super"], "vit_mae/modeling_vit_mae.py:ViTMAELayer.forward": ["GradientCheckpointingLayer", "ModelLayer", "Tensor", "attention", "attention_output", "class", "def", "forward", "hidden_states", "hidden_states_norm", "intermediate", "layer_output", "layernorm_after", "layernorm_before", "output", "return", "self", "torch"], "vit_mae/modeling_vit_mae.py:ViTMAEEncoder.__init__": ["False", "ModelConfig", "ModelEncoder", "ModelLayer", "Module", "ModuleList", "_", "__init__", "class", "config", "def", "for", "gradient_checkpointing", "in", "layer", "nn", "num_hidden_layers", "range", "self", "super"], "vit_mae/modeling_vit_mae.py:ViTMAEEncoder.forward": ["BaseModelOutput", "ModelEncoder", "Module", "Tensor", "class", "def", "enumerate", "for", "forward", "hidden_states", "i", "in", "last_hidden_state", "layer", "layer_module", "nn", "return", "self", "torch"], "vit_mae/modeling_vit_mae.py:ViTMAEPreTrainedModel._init_weights": ["Conv2d", "LayerNorm", "Linear", "ModelDecoder", "ModelEmbeddings", "ModelPreTrainedModel", "None", "PreTrainedModel", "_init_weights", "bias", "class", "config", "decoder_pos_embed", "def", "elif", "if", "init", "initialize_weights", "initializer_range", "is", "isinstance", "mask_token", "mean", "module", "nn", "normal_", "not", "ones_", "self", "std", "weight", "zeros_"], "vit_mae/modeling_vit_mae.py:ViTMAEModel.__init__": ["LayerNorm", "ModelEmbeddings", "ModelEncoder", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "config", "def", "embeddings", "encoder", "eps", "hidden_size", "layer_norm_eps", "layernorm", "nn", "post_init", "self", "super"], "vit_mae/modeling_vit_mae.py:ViTMAEModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "patch_embeddings", "return", "self"], "vit_mae/modeling_vit_mae.py:ViTMAEModel.forward": ["BaseModelOutput", "False", "FloatTensor", "ModelModel", "ModelModelOutput", "ModelPreTrainedModel", "None", "TransformersKwargs", "Unpack", "ValueError", "You", "bool", "class", "def", "embedding_output", "embeddings", "encoder", "encoder_outputs", "forward", "have", "ids_restore", "if", "interpolate_pos_encoding", "is", "kwargs", "last_hidden_state", "layernorm", "mask", "noise", "pixel_values", "r", "raise", "return", "self", "sequence_output", "specify", "to", "torch"], "vit_mae/modeling_vit_mae.py:ViTMAEDecoder.__init__": ["False", "LayerNorm", "Linear", "ModelConfig", "ModelDecoder", "ModelLayer", "Module", "ModuleList", "Parameter", "True", "_", "__init__", "bias", "class", "config", "decoder_config", "decoder_embed", "decoder_hidden_size", "decoder_intermediate_size", "decoder_layers", "decoder_norm", "decoder_num_attention_heads", "decoder_num_hidden_layers", "decoder_pos_embed", "decoder_pred", "deepcopy", "def", "eps", "for", "gradient_checkpointing", "hidden_size", "in", "initialize_weights", "int", "intermediate_size", "layer_norm_eps", "mask_token", "nn", "num_attention_heads", "num_channels", "num_hidden_layers", "num_patches", "patch_size", "range", "requires_grad", "self", "super", "torch", "zeros"], "vit_mae/modeling_vit_mae.py:ViTMAEDecoder.interpolate_pos_encoding": ["False", "ModelDecoder", "Module", "Tensor", "align_corners", "bicubic", "cat", "class", "class_pos_embed", "decoder_pos_embed", "def", "dim", "embeddings", "embeddings_positions", "functional", "interpolate", "interpolate_pos_encoding", "mode", "nn", "patch_pos_embed", "permute", "reshape", "return", "self", "shape", "size", "torch", "view"], "vit_mae/modeling_vit_mae.py:ViTMAEDecoder.initialize_weights": ["ModelDecoder", "Module", "True", "add_cls_token", "class", "config", "copy_", "decoder_pos_embed", "def", "float", "from_numpy", "get_2d_sincos_pos_embed", "init", "initialize_weights", "initializer_range", "int", "mask_token", "nn", "normal_", "num_patches", "self", "shape", "std", "torch", "unsqueeze"], "vit_mae/modeling_vit_mae.py:ViTMAEDecoder.forward": ["False", "ModelDecoder", "ModelDecoderOutput", "Module", "Tensor", "bool", "cat", "class", "decoder_embed", "decoder_layers", "decoder_norm", "decoder_pos_embed", "decoder_pred", "def", "device", "dim", "else", "for", "forward", "gather", "hidden_states", "ids_restore", "if", "in", "index", "interpolate_pos_encoding", "layer_module", "logits", "mask_token", "mask_tokens", "nn", "repeat", "return", "self", "shape", "to", "torch", "unsqueeze", "x", "x_"], "vit_mae/modeling_vit_mae.py:ViTMAEForPreTraining.__init__": ["ModelConfig", "ModelDecoder", "ModelForPreTraining", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "config", "decoder", "def", "embeddings", "num_patches", "post_init", "self", "super", "vit"], "vit_mae/modeling_vit_mae.py:ViTMAEForPreTraining.get_input_embeddings": ["ModelForPreTraining", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "patch_embeddings", "return", "self", "vit"], "vit_mae/modeling_vit_mae.py:ViTMAEForPreTraining.patchify": ["False", "Make", "ModelForPreTraining", "ModelPreTrainedModel", "ValueError", "a", "and", "batch_size", "bool", "by", "channels", "class", "config", "configuration", "def", "divisible", "einsum", "equal", "have", "if", "in", "interpolate_pos_encoding", "is", "nchpwq", "nhwpqc", "not", "num_channels", "num_patches_h", "num_patches_w", "number", "of", "one", "or", "patch", "patch_size", "patchified_pixel_values", "patchify", "pixel", "pixel_values", "raise", "reshape", "return", "self", "set", "shape", "size", "squared", "sure", "that", "the", "to", "torch", "values"], "vit_mae/modeling_vit_mae.py:ViTMAEForPreTraining.unpatchify": ["ModelForPreTraining", "ModelPreTrainedModel", "None", "The", "ValueError", "batch_size", "class", "config", "def", "does", "einsum", "else", "f", "if", "image", "image_size", "in", "int", "is", "match", "nchpwq", "nhwpqc", "not", "num_channels", "num_patches_h", "num_patches_w", "number", "of", "on", "original", "original_height", "original_image_size", "original_width", "patch_size", "patches", "patchified", "patchified_pixel_values", "pixel", "pixel_values", "raise", "reshape", "return", "self", "shape", "the", "torch", "tuple", "unpatchify", "values"], "vit_mae/modeling_vit_mae.py:ViTMAEForPreTraining.forward_loss": ["False", "ModelForPreTraining", "ModelPreTrainedModel", "True", "bool", "class", "config", "def", "dim", "forward_loss", "if", "interpolate_pos_encoding", "keepdim", "loss", "mask", "mean", "norm_pix_loss", "patchify", "pixel_values", "pred", "return", "self", "sum", "target", "var"], "vit_mae/modeling_vit_mae.py:ViTMAEForPreTraining.forward": ["False", "FloatTensor", "ModelDecoderOutput", "ModelForPreTraining", "ModelForPreTrainingOutput", "ModelModelOutput", "ModelPreTrainedModel", "None", "TransformersKwargs", "Unpack", "attentions", "bool", "class", "decoder", "decoder_outputs", "def", "forward", "forward_loss", "hidden_states", "ids_restore", "interpolate_pos_encoding", "kwargs", "last_hidden_state", "latent", "logits", "loss", "mask", "noise", "outputs", "pixel_values", "r", "return", "self", "torch", "vit"], "vit_msn/modeling_vit_msn.py:ViTMSNEmbeddings.__init__": ["Dropout", "False", "ModelConfig", "ModelEmbeddings", "ModelPatchEmbeddings", "Module", "None", "Parameter", "__init__", "bool", "class", "cls_token", "config", "def", "dropout", "else", "hidden_dropout_prob", "hidden_size", "if", "mask_token", "nn", "num_patches", "patch_embeddings", "patch_size", "position_embeddings", "self", "super", "torch", "use_mask_token", "zeros"], "vit_msn/modeling_vit_msn.py:ViTMSNEmbeddings.interpolate_pos_encoding": ["False", "ModelEmbeddings", "Module", "Tensor", "align_corners", "and", "bicubic", "cat", "class", "class_pos_embed", "def", "dim", "embeddings", "functional", "height", "if", "int", "interpolate", "interpolate_pos_encoding", "is_tracing", "jit", "mode", "new_height", "new_width", "nn", "not", "num_patches", "num_positions", "patch_pos_embed", "patch_size", "permute", "position_embeddings", "reshape", "return", "self", "shape", "size", "sqrt_num_positions", "torch", "torch_int", "view", "width"], "vit_msn/modeling_vit_msn.py:ViTMSNEmbeddings.forward": ["BoolTensor", "False", "ModelEmbeddings", "Module", "None", "Tensor", "batch_size", "bool", "bool_masked_pos", "cat", "class", "cls_token", "cls_tokens", "def", "dim", "dropout", "else", "embeddings", "expand", "forward", "height", "if", "interpolate_pos_encoding", "is", "mask", "mask_token", "mask_tokens", "nn", "not", "num_channels", "patch_embeddings", "pixel_values", "position_embeddings", "return", "self", "seq_length", "shape", "torch", "type_as", "unsqueeze", "width"], "vit_msn/modeling_vit_msn.py:ViTMSNPatchEmbeddings.__init__": ["Conv2d", "Iterable", "ModelConfig", "ModelPatchEmbeddings", "Module", "__init__", "abc", "class", "collections", "config", "def", "else", "hidden_size", "if", "image_size", "isinstance", "kernel_size", "nn", "num_channels", "num_patches", "patch_size", "projection", "self", "stride", "super"], "vit_msn/modeling_vit_msn.py:ViTMSNPatchEmbeddings.forward": ["Expected", "False", "Input", "Make", "ModelPatchEmbeddings", "Module", "Tensor", "ValueError", "batch_size", "bool", "but", "channel", "class", "configuration", "def", "dimension", "doesn", "embeddings", "f", "flatten", "forward", "got", "height", "if", "image", "image_size", "in", "interpolate_pos_encoding", "match", "model", "nn", "not", "num_channels", "of", "one", "or", "pixel", "pixel_values", "projection", "raise", "return", "self", "set", "shape", "size", "sure", "t", "that", "the", "torch", "transpose", "values", "width", "with"], "vit_msn/modeling_vit_msn.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "contiguous", "def", "dim", "dropout", "float", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "shape", "size", "softmax", "torch", "training", "transpose", "value"], "vit_msn/modeling_vit_msn.py:ViTMSNSelfAttention.__init__": ["False", "Linear", "ModelConfig", "ModelSelfAttention", "Module", "The", "ValueError", "__init__", "a", "all_head_size", "and", "attention", "attention_head_size", "attention_probs_dropout_prob", "bias", "class", "config", "def", "dropout_prob", "embedding_size", "f", "hasattr", "heads", "hidden", "hidden_size", "if", "int", "is", "is_causal", "key", "multiple", "nn", "not", "num_attention_heads", "number", "of", "qkv_bias", "query", "raise", "scaling", "self", "size", "super", "the", "value"], "vit_msn/modeling_vit_msn.py:ViTMSNSelfAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "ModelSelfAttention", "Module", "None", "Tensor", "_attn_implementation", "all_head_size", "attention_head_size", "attention_interface", "attention_probs", "batch_size", "class", "config", "context_layer", "def", "dropout", "dropout_prob", "eager", "eager_attention_forward", "else", "forward", "hidden_states", "if", "is_causal", "key", "key_layer", "new_context_layer_shape", "new_shape", "nn", "not", "num_attention_heads", "query", "query_layer", "reshape", "return", "scaling", "self", "shape", "size", "torch", "training", "transpose", "tuple", "value", "value_layer", "view"], "vit_msn/modeling_vit_msn.py:ViTMSNSelfOutput.__init__": ["Dropout", "Linear", "ModelConfig", "ModelSelfOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "hidden_dropout_prob", "hidden_size", "nn", "self", "super"], "vit_msn/modeling_vit_msn.py:ViTMSNSelfOutput.forward": ["ModelSelfOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "vit_msn/modeling_vit_msn.py:ViTMSNAttention.__init__": ["ModelAttention", "ModelConfig", "ModelSelfAttention", "ModelSelfOutput", "Module", "__init__", "attention", "class", "config", "def", "nn", "output", "self", "super"], "vit_msn/modeling_vit_msn.py:ViTMSNAttention.forward": ["ModelAttention", "Module", "Tensor", "_", "attention", "class", "def", "forward", "hidden_states", "nn", "output", "return", "self", "self_attn_output", "torch"], "vit_msn/modeling_vit_msn.py:ViTMSNIntermediate.__init__": ["ACT2FN", "Linear", "ModelConfig", "ModelIntermediate", "Module", "__init__", "class", "config", "def", "dense", "else", "hidden_act", "hidden_size", "if", "intermediate_act_fn", "intermediate_size", "isinstance", "nn", "self", "str", "super"], "vit_msn/modeling_vit_msn.py:ViTMSNIntermediate.forward": ["ModelIntermediate", "Module", "Tensor", "class", "def", "dense", "forward", "hidden_states", "intermediate_act_fn", "nn", "return", "self", "torch"], "vit_msn/modeling_vit_msn.py:ViTMSNOutput.__init__": ["Dropout", "Linear", "ModelConfig", "ModelOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "hidden_dropout_prob", "hidden_size", "intermediate_size", "nn", "self", "super"], "vit_msn/modeling_vit_msn.py:ViTMSNOutput.forward": ["ModelOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "vit_msn/modeling_vit_msn.py:ViTMSNLayer.__init__": ["GradientCheckpointingLayer", "LayerNorm", "ModelAttention", "ModelConfig", "ModelIntermediate", "ModelLayer", "ModelOutput", "__init__", "attention", "chunk_size_feed_forward", "class", "config", "def", "eps", "hidden_size", "intermediate", "layer_norm_eps", "layernorm_after", "layernorm_before", "nn", "output", "self", "seq_len_dim", "super"], "vit_msn/modeling_vit_msn.py:ViTMSNLayer.forward": ["GradientCheckpointingLayer", "ModelLayer", "Tensor", "attention", "attention_output", "class", "def", "forward", "hidden_states", "hidden_states_norm", "intermediate", "layer_output", "layernorm_after", "layernorm_before", "output", "return", "self", "torch"], "vit_msn/modeling_vit_msn.py:ViTMSNEncoder.__init__": ["False", "ModelConfig", "ModelEncoder", "ModelLayer", "Module", "ModuleList", "_", "__init__", "class", "config", "def", "for", "gradient_checkpointing", "in", "layer", "nn", "num_hidden_layers", "range", "self", "super"], "vit_msn/modeling_vit_msn.py:ViTMSNEncoder.forward": ["BaseModelOutput", "ModelEncoder", "Module", "Tensor", "class", "def", "enumerate", "for", "forward", "hidden_states", "i", "in", "last_hidden_state", "layer", "layer_module", "nn", "return", "self", "torch"], "vit_msn/modeling_vit_msn.py:ViTMSNPreTrainedModel._init_weights": ["Conv2d", "LayerNorm", "Linear", "ModelEmbeddings", "ModelPreTrainedModel", "None", "PreTrainedModel", "_init_weights", "bias", "class", "cls_token", "config", "def", "elif", "if", "init", "initializer_range", "is", "isinstance", "mask_token", "mean", "module", "nn", "normal_", "not", "ones_", "position_embeddings", "self", "std", "weight", "zeros_"], "vit_msn/modeling_vit_msn.py:ViTMSNModel.__init__": ["False", "LayerNorm", "ModelConfig", "ModelEmbeddings", "ModelEncoder", "ModelModel", "ModelPreTrainedModel", "__init__", "bool", "class", "config", "def", "embeddings", "encoder", "eps", "hidden_size", "layer_norm_eps", "layernorm", "nn", "post_init", "r", "self", "super", "use_mask_token"], "vit_msn/modeling_vit_msn.py:ViTMSNModel.get_input_embeddings": ["ModelModel", "ModelPatchEmbeddings", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "patch_embeddings", "return", "self"], "vit_msn/modeling_vit_msn.py:ViTMSNModel.forward": ["BaseModelOutput", "BoolTensor", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "bool", "bool_masked_pos", "class", "def", "embedding_output", "embeddings", "encoder", "encoder_outputs", "forward", "have", "if", "interpolate_pos_encoding", "is", "kwargs", "last_hidden_state", "layernorm", "pixel_values", "r", "raise", "return", "self", "sequence_output", "specify", "to", "torch"], "vit_msn/modeling_vit_msn.py:ViTMSNForImageClassification.__init__": ["Identity", "Linear", "ModelConfig", "ModelForImageClassification", "ModelModel", "ModelPreTrainedModel", "None", "__init__", "class", "classifier", "config", "def", "else", "hidden_size", "if", "nn", "num_labels", "post_init", "self", "super", "vit"], "vit_msn/modeling_vit_msn.py:ViTMSNForImageClassification.forward": ["BaseModelOutput", "ImageClassifierOutput", "ModelForImageClassification", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "attentions", "bool", "class", "classifier", "config", "def", "forward", "hidden_states", "if", "interpolate_pos_encoding", "is", "kwargs", "labels", "last_hidden_state", "logits", "loss", "loss_function", "not", "outputs", "pixel_values", "r", "return", "self", "sequence_output", "torch", "vit"], "vitdet/modeling_vitdet.py:VitDetEmbeddings.__init__": ["Conv2d", "Iterable", "ModelEmbeddings", "Module", "None", "Parameter", "__init__", "abc", "class", "collections", "config", "def", "else", "hidden_size", "if", "image_size", "isinstance", "kernel_size", "nn", "num_channels", "num_patches", "num_positions", "patch_size", "position_embeddings", "pretrain_image_size", "projection", "self", "stride", "super", "torch", "use_absolute_position_embeddings", "zeros"], "vitdet/modeling_vitdet.py:VitDetEmbeddings.get_absolute_positions": ["Absolute", "False", "ModelEmbeddings", "Module", "ValueError", "a", "abs_pos_embeddings", "align_corners", "be", "bicubic", "class", "def", "else", "embeddings", "functional", "get_absolute_positions", "has_cls_token", "height", "if", "int", "interpolate", "is_tracing", "jit", "math", "mode", "must", "new_abs_pos_embeddings", "nn", "num_position", "number", "or", "permute", "position", "raise", "reshape", "return", "self", "shape", "size", "sqrt", "square", "torch", "width"], "vitdet/modeling_vitdet.py:VitDetEmbeddings.forward": ["Expected", "Make", "ModelEmbeddings", "Module", "None", "Tensor", "True", "ValueError", "but", "channel", "class", "configuration", "def", "dimension", "embeddings", "f", "forward", "get_absolute_positions", "got", "if", "in", "is", "match", "nn", "not", "num_channels", "of", "one", "permute", "pixel", "pixel_values", "position_embeddings", "projection", "raise", "return", "self", "set", "shape", "sure", "that", "the", "torch", "values", "with"], "vitdet/modeling_vitdet.py:get_rel_pos": ["Model_rel_pos", "None", "arange", "def", "else", "functional", "if", "int", "interpolate", "k_coords", "k_size", "linear", "long", "max", "max_rel_dist", "mode", "nn", "permute", "q_coords", "q_size", "rel_pos", "rel_pos_resized", "relative_coords", "reshape", "return", "shape", "size", "torch"], "vitdet/modeling_vitdet.py:add_decomposed_relative_positions": ["Model_decomposed_relative_positions", "None", "_", "attn", "batch_size", "bhwc", "bhwk", "def", "dim", "einsum", "get_rel_pos", "hkc", "k_size", "keys_height", "keys_width", "q_size", "queries", "queries_height", "queries_width", "r_q", "rel_pos_h", "rel_pos_w", "relative_height", "relative_weight", "relative_width", "reshape", "return", "shape", "torch", "view", "wkc"], "vitdet/modeling_vitdet.py:VitDetAttention.__init__": ["Linear", "ModelAttention", "Module", "None", "Parameter", "__init__", "bias", "class", "config", "def", "dim", "head_dim", "hidden_size", "if", "input_size", "nn", "num_attention_heads", "num_heads", "proj", "qkv", "qkv_bias", "rel_pos_h", "rel_pos_w", "scale", "self", "super", "torch", "use_relative_position_embeddings", "zeros"], "vitdet/modeling_vitdet.py:VitDetAttention.forward": ["False", "ModelAttention", "Module", "_", "add_decomposed_relative_positions", "attention_probs", "attention_scores", "batch_size", "class", "def", "dim", "else", "forward", "height", "hidden_state", "if", "keys", "nn", "num_heads", "output_attentions", "outputs", "permute", "proj", "qkv", "queries", "rel_pos_h", "rel_pos_w", "reshape", "return", "scale", "self", "shape", "softmax", "transpose", "unbind", "use_relative_position_embeddings", "values", "view", "width"], "vitdet/modeling_vitdet.py:drop_path": ["False", "Model_path", "Model_prob", "Tensor", "bool", "def", "device", "div", "dtype", "float", "floor_", "if", "input", "keep_prob", "ndim", "not", "or", "output", "rand", "random_tensor", "return", "shape", "torch", "training"], "vitdet/modeling_vitdet.py:VitDetDropPath.__init__": ["ModelDropPath", "Module", "None", "__init__", "class", "def", "drop_prob", "float", "nn", "self", "super"], "vitdet/modeling_vitdet.py:VitDetDropPath.forward": ["ModelDropPath", "Module", "Tensor", "class", "def", "drop_path", "drop_prob", "forward", "hidden_states", "nn", "return", "self", "torch", "training"], "vitdet/modeling_vitdet.py:VitDetDropPath.extra_repr": ["ModelDropPath", "Module", "class", "def", "drop_prob", "extra_repr", "f", "nn", "p", "return", "self", "str"], "vitdet/modeling_vitdet.py:VitDetLayerNorm.__init__": ["ModelLayerNorm", "Module", "Parameter", "__init__", "bias", "class", "def", "eps", "nn", "normalized_shape", "ones", "self", "super", "torch", "weight", "zeros"], "vitdet/modeling_vitdet.py:VitDetLayerNorm.forward": ["ModelLayerNorm", "Module", "None", "True", "bias", "class", "def", "eps", "forward", "keepdim", "mean", "nn", "pow", "return", "s", "self", "sqrt", "torch", "u", "weight", "x"], "vitdet/modeling_vitdet.py:VitDetResBottleneckBlock.__init__": ["ACT2FN", "Conv2d", "False", "ModelLayerNorm", "ModelResBottleneckBlock", "Module", "__init__", "act1", "act2", "bias", "bottleneck_channels", "class", "config", "conv1", "conv2", "conv3", "def", "hidden_act", "in_channels", "nn", "norm1", "norm2", "norm3", "out_channels", "padding", "self", "super"], "vitdet/modeling_vitdet.py:VitDetResBottleneckBlock.forward": ["ModelResBottleneckBlock", "Module", "children", "class", "def", "for", "forward", "in", "layer", "nn", "out", "return", "self", "x"], "vitdet/modeling_vitdet.py:VitDetMlp.__init__": ["ACT2FN", "Dropout", "Linear", "ModelMlp", "Module", "None", "__init__", "act", "class", "config", "def", "drop", "dropout_prob", "fc1", "fc2", "hidden_act", "hidden_features", "in_features", "int", "nn", "self", "super"], "vitdet/modeling_vitdet.py:VitDetMlp.forward": ["ModelMlp", "Module", "Tensor", "act", "class", "def", "drop", "fc1", "fc2", "forward", "nn", "return", "self", "torch", "x"], "vitdet/modeling_vitdet.py:window_partition": ["Model_partition", "Model_size", "Models", "batch_size", "contiguous", "def", "functional", "height", "hidden_state", "nn", "num_channels", "pad", "pad_height", "pad_width", "padded_height", "padded_width", "permute", "return", "shape", "view", "width"], "vitdet/modeling_vitdet.py:window_unpartition": ["Model_size", "Model_unpartition", "Models", "batch_size", "contiguous", "def", "height", "height_width", "hidden_state", "pad_height_width", "padded_height", "padded_width", "permute", "return", "shape", "view", "width"], "vitdet/modeling_vitdet.py:VitDetLayer.__init__": ["False", "GradientCheckpointingLayer", "Identity", "LayerNorm", "ModelAttention", "ModelConfig", "ModelDropPath", "ModelLayer", "ModelMlp", "ModelResBottleneckBlock", "None", "__init__", "attention", "bool", "bottleneck_channels", "class", "config", "def", "dim", "drop_path", "drop_path_rate", "else", "eps", "float", "hidden_features", "hidden_size", "if", "image_size", "in_channels", "in_features", "input_size", "int", "isinstance", "layer_norm_eps", "list", "mlp", "mlp_ratio", "nn", "norm1", "norm2", "out_channels", "patch_size", "residual", "self", "super", "tuple", "use_residual_block", "window_size"], "vitdet/modeling_vitdet.py:VitDetLayer.forward": ["False", "GradientCheckpointingLayer", "ModelLayer", "Tensor", "attention", "bool", "class", "def", "drop_path", "forward", "height", "hidden_states", "if", "mlp", "norm1", "norm2", "output_attentions", "outputs", "pad_height_width", "permute", "residual", "return", "self", "self_attention_outputs", "shape", "shortcut", "torch", "tuple", "use_residual_block", "width", "window_partition", "window_size", "window_unpartition"], "vitdet/modeling_vitdet.py:VitDetEncoder.__init__": ["False", "ModelConfig", "ModelEncoder", "ModelLayer", "Module", "ModuleList", "None", "__init__", "append", "class", "config", "cpu", "def", "depth", "device", "drop_path_rate", "else", "for", "gradient_checkpointing", "i", "if", "in", "item", "layer", "layers", "linspace", "nn", "num_hidden_layers", "range", "residual_block_indices", "self", "super", "torch", "use_residual_block", "window_block_indices", "window_size", "x"], "vitdet/modeling_vitdet.py:VitDetEncoder.forward": ["BaseModelOutput", "False", "ModelEncoder", "Module", "None", "Tensor", "True", "all_hidden_states", "all_self_attentions", "attentions", "bool", "class", "def", "else", "enumerate", "for", "forward", "hidden_states", "i", "if", "in", "is", "last_hidden_state", "layer", "layer_module", "layer_outputs", "nn", "not", "output_attentions", "output_hidden_states", "return", "return_dict", "self", "torch", "tuple", "v"], "vitdet/modeling_vitdet.py:VitDetPreTrainedModel._init_weights": ["Conv2d", "LayerNorm", "Linear", "ModelAttention", "ModelEmbeddings", "ModelPreTrainedModel", "ModelResBottleneckBlock", "None", "PreTrainedModel", "_init_weights", "and", "bias", "class", "config", "constant_", "conv1", "conv2", "conv3", "def", "elif", "fan_out", "for", "if", "in", "init", "initializer_range", "is", "isinstance", "kaiming_normal_", "layer", "mean", "mode", "module", "nn", "nonlinearity", "norm1", "norm2", "norm3", "not", "ones_", "position_embeddings", "rel_pos_h", "rel_pos_w", "relu", "self", "std", "trunc_normal_", "use_relative_position_embeddings", "weight", "zeros_"], "vitdet/modeling_vitdet.py:VitDetModel.__init__": ["ModelConfig", "ModelEmbeddings", "ModelEncoder", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "config", "def", "embeddings", "encoder", "post_init", "self", "super"], "vitdet/modeling_vitdet.py:VitDetModel.get_input_embeddings": ["ModelEmbeddings", "ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "projection", "return", "self"], "vitdet/modeling_vitdet.py:VitDetModel.forward": ["BaseModelOutput", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "ValueError", "You", "attentions", "bool", "class", "config", "def", "else", "embedding_output", "embeddings", "encoder", "encoder_outputs", "forward", "have", "hidden_states", "if", "is", "kwargs", "last_hidden_state", "not", "output_attentions", "output_hidden_states", "pixel_values", "r", "raise", "return", "return_dict", "self", "sequence_output", "specify", "to", "torch", "tuple", "use_return_dict"], "vitdet/modeling_vitdet.py:VitDetBackbone.__init__": ["BackboneMixin", "ModelBackbone", "ModelEmbeddings", "ModelEncoder", "ModelPreTrainedModel", "_", "__init__", "_init_backbone", "class", "config", "def", "embeddings", "encoder", "for", "hidden_size", "in", "num_features", "num_hidden_layers", "post_init", "range", "self", "super"], "vitdet/modeling_vitdet.py:VitDetBackbone.get_input_embeddings": ["BackboneMixin", "ModelBackbone", "ModelEmbeddings", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "projection", "return", "self"], "vitdet/modeling_vitdet.py:VitDetBackbone.forward": ["BackboneMixin", "BackboneOutput", "ModelBackbone", "ModelPreTrainedModel", "None", "Tensor", "True", "attentions", "bool", "class", "config", "def", "else", "embedding_output", "embeddings", "encoder", "feature_maps", "for", "forward", "hidden_state", "hidden_states", "if", "in", "is", "kwargs", "not", "out_features", "output", "output_attentions", "output_hidden_states", "outputs", "pixel_values", "r", "return", "return_dict", "self", "stage", "stage_names", "torch", "use_return_dict", "zip"], "vitmatte/modeling_vitmatte.py:VitMattePreTrainedModel._init_weights": ["BatchNorm2d", "Conv2d", "ModelPreTrainedModel", "Module", "None", "PreTrainedModel", "_init_weights", "bias", "class", "config", "def", "getattr", "if", "init", "initializer_range", "is", "isinstance", "mean", "module", "nn", "normal_", "not", "num_batches_tracked", "ones_", "running_mean", "running_var", "self", "std", "weight", "zeros_"], "vitmatte/modeling_vitmatte.py:VitMatteBasicConv3x3.__init__": ["BatchNorm2d", "Conv2d", "False", "ModelBasicConv3x3", "Module", "ReLU", "__init__", "batch_norm", "batch_norm_eps", "bias", "class", "config", "conv", "def", "eps", "in_channels", "kernel_size", "nn", "out_channels", "padding", "relu", "self", "stride", "super"], "vitmatte/modeling_vitmatte.py:VitMatteBasicConv3x3.forward": ["ModelBasicConv3x3", "Module", "batch_norm", "class", "conv", "def", "forward", "hidden_state", "nn", "relu", "return", "self"], "vitmatte/modeling_vitmatte.py:VitMatteConvStream.__init__": ["ModelBasicConv3x3", "ModelConvStream", "Module", "ModuleList", "None", "__init__", "append", "backbone_config", "class", "config", "conv_chans", "convs", "convstream_hidden_sizes", "def", "for", "i", "if", "in", "in_chan_", "in_channels", "is", "len", "nn", "not", "num_channels", "out_chan_", "out_channels", "range", "self", "super"], "vitmatte/modeling_vitmatte.py:VitMatteConvStream.forward": ["ModelConvStream", "Module", "class", "convs", "def", "detailed_feature_map_", "detailed_feature_map_0", "embeddings", "for", "forward", "i", "in", "len", "name_", "nn", "out_dict", "pixel_values", "range", "return", "self", "str"], "vitmatte/modeling_vitmatte.py:VitMatteFusionBlock.__init__": ["ModelBasicConv3x3", "ModelFusionBlock", "Module", "__init__", "class", "config", "conv", "def", "in_channels", "nn", "out_channels", "padding", "self", "stride", "super"], "vitmatte/modeling_vitmatte.py:VitMatteFusionBlock.forward": ["False", "ModelFusionBlock", "Module", "align_corners", "bilinear", "cat", "class", "conv", "def", "detailed_feature_map", "dim", "features", "forward", "functional", "interpolate", "mode", "nn", "out", "return", "scale_factor", "self", "torch", "upscaled_features"], "vitmatte/modeling_vitmatte.py:VitMatteHead.__init__": ["BatchNorm2d", "Conv2d", "ModelHead", "Module", "ReLU", "Sequential", "True", "__init__", "class", "config", "def", "fusion_hidden_sizes", "in_channels", "kernel_size", "matting_convs", "mid_channels", "nn", "padding", "self", "stride", "super"], "vitmatte/modeling_vitmatte.py:VitMatteHead.forward": ["ModelHead", "Module", "class", "def", "forward", "hidden_state", "matting_convs", "nn", "return", "self"], "vitmatte/modeling_vitmatte.py:VitMatteDetailCaptureModule.__init__": ["ModelConvStream", "ModelDetailCaptureModule", "ModelFusionBlock", "ModelHead", "Module", "ModuleList", "The", "ValueError", "__init__", "append", "be", "class", "config", "conv_chans", "convstream", "convstream_hidden_sizes", "def", "equal", "for", "fusion_blocks", "fusion_channels", "fusion_hidden_sizes", "hidden_size", "i", "if", "in", "in_channels", "len", "length", "matting_head", "nn", "of", "out_channels", "raise", "range", "self", "should", "super", "the", "to"], "vitmatte/modeling_vitmatte.py:VitMatteDetailCaptureModule.forward": ["ModelDetailCaptureModule", "Module", "alphas", "class", "convstream", "def", "detail_features", "detailed_feature_map_", "detailed_feature_map_name", "features", "for", "forward", "fusion_blocks", "i", "in", "len", "matting_head", "nn", "pixel_values", "range", "return", "self", "sigmoid", "str", "torch"], "vitmatte/modeling_vitmatte.py:VitMatteForImageMatting.__init__": ["ModelDetailCaptureModule", "ModelForImageMatting", "ModelPreTrainedModel", "__init__", "backbone", "class", "config", "decoder", "def", "load_backbone", "post_init", "self", "super"], "vitmatte/modeling_vitmatte.py:VitMatteForImageMatting.forward": ["ImageMattingOutput", "ModelForImageMatting", "ModelPreTrainedModel", "None", "NotImplementedError", "Tensor", "Training", "alphas", "attentions", "backbone", "bool", "class", "config", "decoder", "def", "else", "feature_maps", "features", "forward", "forward_with_filtered_kwargs", "hidden_states", "if", "is", "kwargs", "labels", "loss", "not", "output", "output_attentions", "output_hidden_states", "outputs", "pixel_values", "r", "raise", "return", "return_dict", "self", "supported", "torch", "use_return_dict", "yet"], "vitpose/modeling_vitpose.py:VitPosePreTrainedModel._init_weights": ["Conv2d", "LayerNorm", "Linear", "ModelPreTrainedModel", "None", "PreTrainedModel", "_init_weights", "bias", "class", "config", "def", "elif", "if", "init", "initializer_range", "is", "isinstance", "mean", "module", "nn", "not", "ones_", "self", "std", "trunc_normal_", "weight", "zeros_"], "vitpose/modeling_vitpose.py:flip_back": ["Model", "Model_back", "Model_pairs", "ValueError", "batch_size", "be", "channels", "clone", "combined", "def", "for", "gaussian", "heatmap", "height", "if", "in", "left", "ndim", "not", "num_keypoints", "or", "output_Modelped", "output_Modelped_back", "raise", "reshape", "return", "right", "shape", "should", "target", "target_type", "tolist", "width"], "vitpose/modeling_vitpose.py:VitPoseSimpleDecoder.__init__": ["Conv2d", "False", "ModelConfig", "ModelSimpleDecoder", "Module", "ReLU", "Upsample", "__init__", "activation", "align_corners", "backbone_config", "bilinear", "class", "config", "conv", "def", "hidden_size", "kernel_size", "mode", "nn", "num_labels", "padding", "scale_factor", "self", "stride", "super", "upsampling"], "vitpose/modeling_vitpose.py:VitPoseSimpleDecoder.forward": ["ModelSimpleDecoder", "Module", "None", "Tensor", "activation", "class", "conv", "def", "flip_back", "flip_pairs", "forward", "heatmaps", "hidden_state", "if", "is", "nn", "not", "return", "self", "torch", "upsampling"], "vitpose/modeling_vitpose.py:VitPoseClassicDecoder.__init__": ["BatchNorm2d", "Conv2d", "ConvTranspose2d", "False", "ModelClassicDecoder", "ModelConfig", "Module", "ReLU", "__init__", "backbone_config", "batchnorm1", "batchnorm2", "bias", "class", "config", "conv", "deconv1", "deconv2", "def", "hidden_size", "kernel_size", "nn", "num_labels", "padding", "relu1", "relu2", "self", "stride", "super"], "vitpose/modeling_vitpose.py:VitPoseClassicDecoder.forward": ["ModelClassicDecoder", "Module", "None", "Tensor", "batchnorm1", "batchnorm2", "class", "conv", "deconv1", "deconv2", "def", "flip_back", "flip_pairs", "forward", "heatmaps", "hidden_state", "if", "is", "nn", "not", "relu1", "relu2", "return", "self", "torch"], "vitpose/modeling_vitpose.py:VitPoseForPoseEstimation.__init__": ["ModelClassicDecoder", "ModelConfig", "ModelForPoseEstimation", "ModelPreTrainedModel", "ModelSimpleDecoder", "The", "ValueError", "__init__", "a", "an", "attribute", "backbone", "class", "config", "def", "else", "hasattr", "have", "head", "hidden_size", "if", "image_size", "load_backbone", "not", "patch_size", "post_init", "raise", "self", "should", "super", "use_simple_decoder"], "vitpose/modeling_vitpose.py:VitPoseForPoseEstimation.forward": ["BackboneOutput", "ModelEstimatorOutput", "ModelForPoseEstimation", "ModelPreTrainedModel", "None", "NotImplementedError", "Tensor", "Training", "TransformersKwargs", "Unpack", "attentions", "backbone", "backbone_config", "batch_size", "class", "config", "contiguous", "dataset_index", "def", "feature_maps", "flip_pairs", "forward", "forward_with_filtered_kwargs", "head", "heatmaps", "hidden_states", "if", "image_size", "is", "kwargs", "labels", "loss", "not", "outputs", "patch_height", "patch_size", "patch_width", "permute", "pixel_values", "r", "raise", "reshape", "return", "self", "sequence_output", "shape", "supported", "torch", "yet"], "vitpose_backbone/modeling_vitpose_backbone.py:VitPoseBackbonePatchEmbeddings.__init__": ["Conv2d", "Iterable", "ModelPatchEmbeddings", "Module", "__init__", "abc", "class", "collections", "config", "def", "else", "embed_dim", "hidden_size", "if", "image_size", "isinstance", "kernel_size", "nn", "num_channels", "num_patches", "padding", "patch_size", "projection", "self", "stride", "super"], "vitpose_backbone/modeling_vitpose_backbone.py:VitPoseBackbonePatchEmbeddings.forward": ["Input", "ModelPatchEmbeddings", "Module", "Tensor", "ValueError", "class", "def", "doesn", "embeddings", "f", "flatten", "forward", "height", "if", "image", "image_size", "match", "model", "nn", "or", "pixel_values", "projection", "raise", "return", "self", "shape", "size", "t", "torch", "transpose", "width"], "vitpose_backbone/modeling_vitpose_backbone.py:VitPoseBackboneEmbeddings.__init__": ["Dropout", "ModelConfig", "ModelEmbeddings", "ModelPatchEmbeddings", "Module", "Parameter", "__init__", "class", "config", "def", "dropout", "hidden_dropout_prob", "hidden_size", "nn", "num_patches", "patch_embeddings", "position_embeddings", "self", "super", "torch", "zeros"], "vitpose_backbone/modeling_vitpose_backbone.py:VitPoseBackboneEmbeddings.forward": ["ModelEmbeddings", "Module", "Tensor", "class", "def", "dropout", "embeddings", "forward", "nn", "patch_embeddings", "pixel_values", "position_embeddings", "return", "self", "torch"], "vitpose_backbone/modeling_vitpose_backbone.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "contiguous", "def", "dim", "dropout", "float", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "shape", "size", "softmax", "torch", "training", "transpose", "value"], "vitpose_backbone/modeling_vitpose_backbone.py:VitPoseBackboneSelfAttention.__init__": ["False", "Linear", "ModelConfig", "ModelSelfAttention", "Module", "The", "ValueError", "__init__", "a", "all_head_size", "and", "attention", "attention_head_size", "attention_probs_dropout_prob", "bias", "class", "config", "def", "dropout_prob", "embedding_size", "f", "hasattr", "heads", "hidden", "hidden_size", "if", "int", "is", "is_causal", "key", "multiple", "nn", "not", "num_attention_heads", "number", "of", "qkv_bias", "query", "raise", "scaling", "self", "size", "super", "the", "value"], "vitpose_backbone/modeling_vitpose_backbone.py:VitPoseBackboneSelfAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "ModelSelfAttention", "Module", "None", "Tensor", "_attn_implementation", "all_head_size", "attention_head_size", "attention_interface", "attention_probs", "batch_size", "class", "config", "context_layer", "def", "dropout", "dropout_prob", "eager", "eager_attention_forward", "else", "forward", "hidden_states", "if", "is_causal", "key", "key_layer", "new_context_layer_shape", "new_shape", "nn", "not", "num_attention_heads", "query", "query_layer", "reshape", "return", "scaling", "self", "shape", "size", "torch", "training", "transpose", "tuple", "value", "value_layer", "view"], "vitpose_backbone/modeling_vitpose_backbone.py:VitPoseBackboneSelfOutput.__init__": ["Dropout", "Linear", "ModelConfig", "ModelSelfOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "hidden_dropout_prob", "hidden_size", "nn", "self", "super"], "vitpose_backbone/modeling_vitpose_backbone.py:VitPoseBackboneSelfOutput.forward": ["ModelSelfOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "vitpose_backbone/modeling_vitpose_backbone.py:VitPoseBackboneAttention.__init__": ["ModelAttention", "ModelConfig", "ModelSelfAttention", "ModelSelfOutput", "Module", "__init__", "attention", "class", "config", "def", "nn", "output", "self", "super"], "vitpose_backbone/modeling_vitpose_backbone.py:VitPoseBackboneAttention.forward": ["ModelAttention", "Module", "Tensor", "_", "attention", "class", "def", "forward", "hidden_states", "nn", "output", "return", "self", "self_attn_output", "torch"], "vitpose_backbone/modeling_vitpose_backbone.py:VitPoseNaiveMoe.__init__": ["Linear", "ModelPoseNaiveMoe", "ModuleList", "_", "__init__", "append", "class", "config", "def", "for", "hidden_features", "hidden_size", "in", "int", "mlp_ratio", "nn", "num_experts", "part_features", "range", "self", "super"], "vitpose_backbone/modeling_vitpose_backbone.py:VitPoseNaiveMoe.forward": ["ModelPoseNaiveMoe", "ModuleList", "class", "current_hidden_state", "def", "expert_hidden_state", "for", "forward", "hidden_state", "i", "in", "indices", "nn", "num_experts", "part_features", "range", "return", "selected_index", "self", "torch", "zeros_like"], "vitpose_backbone/modeling_vitpose_backbone.py:VitPoseBackboneMoeMLP.__init__": ["ACT2FN", "Linear", "ModelConfig", "ModelMoeMLP", "ModelPoseNaiveMoe", "Module", "__init__", "act", "class", "config", "def", "experts", "fc1", "fc2", "hidden_act", "hidden_features", "hidden_size", "in_features", "int", "mlp_ratio", "nn", "num_experts", "out_features", "part_features", "self", "super"], "vitpose_backbone/modeling_vitpose_backbone.py:VitPoseBackboneMoeMLP.forward": ["ModelMoeMLP", "Module", "Tensor", "act", "cat", "class", "def", "dim", "expert_hidden_state", "experts", "fc1", "fc2", "forward", "hidden_state", "indices", "nn", "return", "self", "shared_hidden_state", "torch", "view"], "vitpose_backbone/modeling_vitpose_backbone.py:VitPoseBackboneMLP.__init__": ["ACT2FN", "Linear", "ModelConfig", "ModelMLP", "Module", "True", "__init__", "activation", "bias", "class", "config", "def", "fc1", "fc2", "hidden_act", "hidden_features", "hidden_size", "in_features", "int", "mlp_ratio", "nn", "out_features", "self", "super"], "vitpose_backbone/modeling_vitpose_backbone.py:VitPoseBackboneMLP.forward": ["ModelMLP", "Module", "Tensor", "activation", "class", "def", "fc1", "fc2", "forward", "hidden_state", "nn", "return", "self", "torch"], "vitpose_backbone/modeling_vitpose_backbone.py:VitPoseBackboneLayer.__init__": ["GradientCheckpointingLayer", "LayerNorm", "ModelAttention", "ModelConfig", "ModelLayer", "ModelMLP", "ModelMoeMLP", "__init__", "attention", "class", "config", "def", "else", "eps", "hidden_size", "if", "layer_norm_eps", "layernorm_after", "layernorm_before", "mlp", "nn", "num_experts", "self", "super"], "vitpose_backbone/modeling_vitpose_backbone.py:VitPoseBackboneLayer.forward": ["GradientCheckpointingLayer", "ModelLayer", "None", "Please", "Tensor", "ValueError", "and", "attention", "attention_output", "be", "class", "dataset_index", "def", "else", "experts", "f", "forward", "hidden_states", "hidden_states_norm", "if", "indices", "is", "layer_output", "layernorm_after", "layernorm_before", "mlp", "multiple", "must", "num_experts", "pass", "provide", "provided", "raise", "return", "self", "the", "to", "torch", "using", "when"], "vitpose_backbone/modeling_vitpose_backbone.py:VitPoseBackboneEncoder.__init__": ["False", "ModelConfig", "ModelEncoder", "ModelLayer", "Module", "ModuleList", "_", "__init__", "class", "config", "def", "for", "gradient_checkpointing", "in", "layer", "nn", "num_hidden_layers", "range", "self", "super"], "vitpose_backbone/modeling_vitpose_backbone.py:VitPoseBackboneEncoder.forward": ["BaseModelOutput", "ModelEncoder", "Module", "None", "Tensor", "all_hidden_states", "append", "bool", "class", "dataset_index", "def", "else", "enumerate", "for", "forward", "hidden_states", "i", "if", "in", "is", "last_hidden_state", "layer", "layer_module", "nn", "not", "output_hidden_states", "return", "self", "torch", "tuple"], "vitpose_backbone/modeling_vitpose_backbone.py:VitPoseBackbonePreTrainedModel._init_weights": ["Conv2d", "LayerNorm", "Linear", "ModelEmbeddings", "ModelPreTrainedModel", "None", "PreTrainedModel", "_init_weights", "bias", "class", "config", "def", "elif", "if", "init", "initializer_range", "is", "isinstance", "mean", "module", "nn", "not", "ones_", "position_embeddings", "self", "std", "trunc_normal_", "weight", "zeros_"], "vitpose_backbone/modeling_vitpose_backbone.py:VitPoseBackbone.__init__": ["BackboneMixin", "LayerNorm", "Model", "ModelConfig", "ModelEmbeddings", "ModelEncoder", "ModelPreTrainedModel", "_", "__init__", "_init_backbone", "class", "config", "def", "embeddings", "encoder", "eps", "for", "hidden_size", "in", "layer_norm_eps", "layernorm", "nn", "num_features", "num_hidden_layers", "post_init", "range", "self", "super"], "vitpose_backbone/modeling_vitpose_backbone.py:VitPoseBackbone.forward": ["BackboneMixin", "BackboneOutput", "BaseModelOutput", "Model", "ModelPreTrainedModel", "None", "Tensor", "True", "append", "bool", "class", "config", "dataset_index", "def", "else", "embedding_output", "embeddings", "encoder", "feature_maps", "for", "forward", "hidden_state", "hidden_states", "if", "in", "is", "kwargs", "layernorm", "out_features", "output_hidden_states", "outputs", "pixel_values", "r", "return", "self", "stage", "stage_names", "torch", "tuple", "zip"], "vits/modeling_vits.py:fused_add_tanh_sigmoid_multiply": ["Model_add_tanh_sigmoid_multiply", "acts", "def", "in_act", "input_a", "input_b", "num_channels", "return", "s_act", "sigmoid", "t_act", "tanh", "torch"], "vits/modeling_vits.py:_unconstrained_rational_quadratic_spline": ["False", "_rational_quadratic_spline", "_unconstrained_rational_quadratic_spline", "constant", "def", "exp", "functional", "inputs", "inside_interval_mask", "log", "log_abs_det", "min_bin_height", "min_bin_width", "min_derivative", "nn", "np", "outputs", "outside_interval_mask", "pad", "return", "reverse", "tail_bound", "torch", "unnormalized_derivatives", "unnormalized_heights", "unnormalized_widths", "zeros_like"], "vits/modeling_vits.py:_rational_quadratic_spline": ["Input", "Minimal", "None", "RuntimeError", "ValueError", "_rational_quadratic_spline", "a", "all", "b", "bin", "bin_idx", "bin_locations", "bins", "c", "constant", "cumheights", "cumsum", "cumwidths", "def", "delta", "denominator", "derivative_numerator", "derivatives", "dim", "discriminant", "domain", "else", "f", "for", "functional", "gather", "height", "heights", "if", "input_bin_widths", "input_cumheights", "input_cumwidths", "input_delta", "input_derivatives", "input_derivatives_plus_one", "input_heights", "inputs", "intermediate1", "intermediate2", "intermediate3", "invalid", "is", "its", "large", "log", "log_abs_det", "lower_bound", "max", "min", "min_bin_height", "min_bin_width", "min_derivative", "mode", "nn", "not", "num_bins", "number", "numerator", "of", "or", "outputs", "pad", "pow", "raise", "return", "reverse", "root", "shape", "softmax", "softplus", "sqrt", "sum", "tail_bound", "the", "theta", "theta_one_minus_theta", "to", "too", "torch", "transform", "unnormalized_derivatives", "unnormalized_heights", "unnormalized_widths", "upper_bound", "value", "width", "widths", "within"], "vits/modeling_vits.py:VitsWaveNet.__init__": ["Conv1d", "Dropout", "ModelConfig", "ModelWaveNet", "Module", "ModuleList", "__init__", "append", "class", "cond_layer", "config", "def", "dilation", "dropout", "else", "for", "hasattr", "hidden_size", "i", "if", "in", "in_channels", "in_layer", "in_layers", "int", "kernel_size", "name", "nn", "num_layers", "out_channels", "padding", "parametrizations", "range", "res_skip_channels", "res_skip_layer", "res_skip_layers", "self", "speaker_embedding_size", "super", "torch", "utils", "wavenet_dilation_rate", "wavenet_dropout", "wavenet_kernel_size", "weight", "weight_norm"], "vits/modeling_vits.py:VitsWaveNet.forward": ["IntTensor", "ModelWaveNet", "Module", "None", "acts", "class", "cond_layer", "cond_offset", "def", "dropout", "else", "for", "forward", "fused_add_tanh_sigmoid_multiply", "global_conditioning", "global_states", "hidden_size", "hidden_states", "i", "if", "in", "in_layers", "inputs", "is", "nn", "not", "num_channels_tensor", "num_layers", "outputs", "padding_mask", "range", "res_acts", "res_skip_acts", "res_skip_layers", "return", "self", "torch", "zeros_like"], "vits/modeling_vits.py:VitsWaveNet.remove_weight_norm": ["ModelWaveNet", "Module", "class", "cond_layer", "def", "for", "if", "in", "in_layers", "layer", "nn", "remove_weight_norm", "res_skip_layers", "self", "speaker_embedding_size", "torch", "utils"], "vits/modeling_vits.py:VitsPosteriorEncoder.__init__": ["Conv1d", "ModelConfig", "ModelPosteriorEncoder", "ModelWaveNet", "Module", "__init__", "class", "config", "conv_pre", "conv_proj", "def", "flow_size", "hidden_size", "nn", "num_layers", "out_channels", "posterior_encoder_num_wavenet_layers", "self", "spectrogram_bins", "super", "wavenet"], "vits/modeling_vits.py:VitsPosteriorEncoder.forward": ["ModelPosteriorEncoder", "Module", "None", "class", "conv_pre", "conv_proj", "def", "dim", "exp", "forward", "global_conditioning", "inputs", "log_stddev", "mean", "nn", "out_channels", "padding_mask", "randn_like", "return", "sampled", "self", "split", "stats", "torch", "wavenet"], "vits/modeling_vits.py:HifiGanResidualBlock.__init__": ["Conv1d", "ModelGanResidualBlock", "Module", "ModuleList", "_", "__init__", "channels", "class", "convs1", "convs2", "def", "dilation", "for", "get_padding", "i", "in", "kernel_size", "leaky_relu_slope", "len", "nn", "padding", "range", "self", "stride", "super"], "vits/modeling_vits.py:HifiGanResidualBlock.get_padding": ["ModelGanResidualBlock", "Module", "class", "def", "dilation", "get_padding", "kernel_size", "nn", "return", "self"], "vits/modeling_vits.py:HifiGanResidualBlock.apply_weight_norm": ["ModelGanResidualBlock", "Module", "apply_weight_norm", "class", "convs1", "convs2", "def", "for", "hasattr", "if", "in", "layer", "nn", "parametrizations", "self", "utils", "weight_norm"], "vits/modeling_vits.py:HifiGanResidualBlock.remove_weight_norm": ["ModelGanResidualBlock", "Module", "class", "convs1", "convs2", "def", "for", "in", "layer", "nn", "remove_weight_norm", "self", "utils"], "vits/modeling_vits.py:HifiGanResidualBlock.forward": ["ModelGanResidualBlock", "Module", "class", "conv1", "conv2", "convs1", "convs2", "def", "for", "forward", "functional", "hidden_states", "in", "leaky_relu", "leaky_relu_slope", "nn", "residual", "return", "self", "zip"], "vits/modeling_vits.py:VitsHifiGan.__init__": ["Conv1d", "ConvTranspose1d", "False", "HifiGanResidualBlock", "ModelConfig", "ModelHifiGan", "Module", "ModuleList", "__init__", "append", "bias", "channels", "class", "cond", "config", "conv_post", "conv_pre", "def", "dilation", "enumerate", "flow_size", "for", "i", "if", "in", "kernel_size", "leaky_relu_slope", "len", "nn", "num_kernels", "num_upsamples", "padding", "range", "resblock_dilation_sizes", "resblock_kernel_sizes", "resblocks", "self", "speaker_embedding_size", "stride", "super", "upsample_initial_channel", "upsample_kernel_sizes", "upsample_rate", "upsample_rates", "upsampler", "zip"], "vits/modeling_vits.py:VitsHifiGan.apply_weight_norm": ["ModelHifiGan", "Module", "apply_weight_norm", "class", "def", "for", "hasattr", "if", "in", "layer", "nn", "parametrizations", "resblocks", "self", "upsampler", "utils", "weight_norm"], "vits/modeling_vits.py:VitsHifiGan.remove_weight_norm": ["ModelHifiGan", "Module", "class", "def", "for", "in", "layer", "nn", "remove_weight_norm", "resblocks", "self", "upsampler", "utils"], "vits/modeling_vits.py:VitsHifiGan.forward": ["FloatTensor", "ModelHifiGan", "Module", "None", "class", "cond", "config", "conv_post", "conv_pre", "def", "for", "forward", "functional", "global_conditioning", "hidden_states", "i", "if", "in", "is", "j", "leaky_relu", "leaky_relu_slope", "nn", "not", "num_kernels", "num_upsamples", "r", "range", "res_state", "resblocks", "return", "self", "spectrogram", "tanh", "torch", "upsampler", "waveform"], "vits/modeling_vits.py:VitsResidualCouplingLayer.__init__": ["Conv1d", "ModelConfig", "ModelResidualCouplingLayer", "ModelWaveNet", "Module", "__init__", "class", "config", "conv_post", "conv_pre", "def", "flow_size", "half_channels", "hidden_size", "nn", "num_layers", "prior_encoder_num_wavenet_layers", "self", "super", "wavenet"], "vits/modeling_vits.py:VitsResidualCouplingLayer.forward": ["False", "ModelResidualCouplingLayer", "Module", "None", "cat", "class", "conv_post", "conv_pre", "def", "dim", "else", "exp", "first_half", "forward", "global_conditioning", "half_channels", "hidden_states", "if", "inputs", "log_determinant", "log_stddev", "mean", "nn", "not", "outputs", "padding_mask", "return", "reverse", "second_half", "self", "split", "sum", "torch", "wavenet", "zeros_like"], "vits/modeling_vits.py:VitsResidualCouplingBlock.__init__": ["ModelConfig", "ModelResidualCouplingBlock", "ModelResidualCouplingLayer", "Module", "ModuleList", "_", "__init__", "append", "class", "config", "def", "flows", "for", "in", "nn", "prior_encoder_num_flows", "range", "self", "super"], "vits/modeling_vits.py:VitsResidualCouplingBlock.forward": ["False", "ModelResidualCouplingBlock", "Module", "None", "True", "_", "class", "def", "else", "flip", "flow", "flows", "for", "forward", "global_conditioning", "if", "in", "inputs", "nn", "not", "padding_mask", "return", "reverse", "reversed", "self", "torch"], "vits/modeling_vits.py:VitsDilatedDepthSeparableConv.__init__": ["Conv1d", "Dropout", "LayerNorm", "ModelConfig", "ModelDilatedDepthSeparableConv", "Module", "ModuleList", "__init__", "append", "channels", "class", "config", "convs_dilated", "convs_pointwise", "def", "depth_separable_num_layers", "dilation", "dropout", "dropout_rate", "duration_predictor_kernel_size", "for", "groups", "hidden_size", "i", "in", "in_channels", "kernel_size", "nn", "norms_1", "norms_2", "num_layers", "out_channels", "padding", "range", "self", "super"], "vits/modeling_vits.py:VitsDilatedDepthSeparableConv.forward": ["ModelDilatedDepthSeparableConv", "Module", "None", "class", "convs_dilated", "convs_pointwise", "def", "dropout", "for", "forward", "functional", "gelu", "global_conditioning", "hidden_states", "i", "if", "in", "inputs", "is", "nn", "norms_1", "norms_2", "not", "num_layers", "padding_mask", "range", "return", "self", "transpose"], "vits/modeling_vits.py:VitsConvFlow.__init__": ["Conv1d", "ModelConfig", "ModelConvFlow", "ModelDilatedDepthSeparableConv", "Module", "__init__", "class", "config", "conv_dds", "conv_pre", "conv_proj", "def", "depth_separable_channels", "duration_predictor_flow_bins", "duration_predictor_tail_bound", "filter_channels", "half_channels", "hidden_size", "nn", "num_bins", "self", "super", "tail_bound"], "vits/modeling_vits.py:VitsConvFlow.forward": ["False", "ModelConvFlow", "Module", "None", "_unconstrained_rational_quadratic_spline", "batch_size", "cat", "channels", "class", "conv_dds", "conv_pre", "conv_proj", "def", "dim", "else", "filter_channels", "first_half", "forward", "global_conditioning", "half_channels", "hidden_states", "if", "inputs", "length", "log_abs_det", "log_determinant", "math", "nn", "not", "num_bins", "outputs", "padding_mask", "permute", "reshape", "return", "reverse", "second_half", "self", "shape", "split", "sqrt", "sum", "tail_bound", "torch", "unnormalized_derivatives", "unnormalized_heights", "unnormalized_widths"], "vits/modeling_vits.py:VitsElementwiseAffine.__init__": ["ModelConfig", "ModelElementwiseAffine", "Module", "Parameter", "__init__", "channels", "class", "config", "def", "depth_separable_channels", "log_scale", "nn", "self", "super", "torch", "translate", "zeros"], "vits/modeling_vits.py:VitsElementwiseAffine.forward": ["False", "ModelElementwiseAffine", "Module", "None", "class", "def", "else", "exp", "forward", "global_conditioning", "if", "inputs", "log_determinant", "log_scale", "nn", "not", "outputs", "padding_mask", "return", "reverse", "self", "sum", "torch", "translate"], "vits/modeling_vits.py:VitsStochasticDurationPredictor.__init__": ["Conv1d", "ModelConvFlow", "ModelDilatedDepthSeparableConv", "ModelElementwiseAffine", "ModelStochasticDurationPredictor", "Module", "ModuleList", "_", "__init__", "append", "class", "cond", "config", "conv_dds", "conv_pre", "conv_proj", "def", "dropout_rate", "duration_predictor_dropout", "duration_predictor_num_flows", "embed_dim", "filter_channels", "flows", "for", "hidden_size", "if", "in", "nn", "post_conv_dds", "post_conv_pre", "post_conv_proj", "post_flows", "range", "self", "speaker_embedding_size", "super"], "vits/modeling_vits.py:VitsStochasticDurationPredictor.forward": ["False", "ModelStochasticDurationPredictor", "Module", "None", "True", "_", "cat", "clamp_min", "class", "cond", "conv_dds", "conv_pre", "conv_proj", "def", "detach", "device", "dim", "dtype", "durations", "else", "first_half", "flip", "flow", "flows", "for", "forward", "functional", "global_conditioning", "hidden_states", "if", "in", "inputs", "is", "latents", "latents_posterior", "list", "log", "log_determinant", "log_determinant_posterior_sum", "log_determinant_sum", "log_duration", "logq", "logsigmoid", "math", "nll", "nn", "noise_scale", "not", "padding_mask", "pi", "post_conv_dds", "post_conv_pre", "post_conv_proj", "post_flows", "randn", "random_posterior", "return", "reverse", "reversed", "second_half", "self", "sigmoid", "size", "split", "sum", "to", "torch"], "vits/modeling_vits.py:VitsDurationPredictor.__init__": ["Conv1d", "Dropout", "LayerNorm", "ModelDurationPredictor", "Module", "__init__", "class", "cond", "config", "conv_1", "conv_2", "def", "dropout", "duration_predictor_dropout", "duration_predictor_filter_channels", "duration_predictor_kernel_size", "eps", "filter_channels", "hidden_size", "if", "kernel_size", "layer_norm_eps", "nn", "norm_1", "norm_2", "padding", "proj", "self", "speaker_embedding_size", "super"], "vits/modeling_vits.py:VitsDurationPredictor.forward": ["ModelDurationPredictor", "Module", "None", "class", "cond", "conv_1", "conv_2", "def", "detach", "dropout", "forward", "global_conditioning", "if", "inputs", "is", "nn", "norm_1", "norm_2", "not", "padding_mask", "proj", "relu", "return", "self", "torch", "transpose"], "vits/modeling_vits.py:VitsAttention.__init__": ["Linear", "ModelAttention", "ModelConfig", "Module", "Parameter", "ValueError", "__init__", "and", "attention_dropout", "be", "bias", "by", "class", "config", "def", "divisible", "dropout", "emb_rel_k", "emb_rel_v", "embed_dim", "f", "got", "head_dim", "hidden_size", "if", "k_proj", "must", "nn", "num_attention_heads", "num_heads", "out_proj", "q_proj", "raise", "randn", "scaling", "self", "super", "torch", "use_bias", "v_proj", "window_size"], "vits/modeling_vits.py:VitsAttention._shape": ["ModelAttention", "Module", "Tensor", "_shape", "bsz", "class", "contiguous", "def", "head_dim", "int", "nn", "num_heads", "return", "self", "seq_len", "tensor", "torch", "transpose", "view"], "vits/modeling_vits.py:VitsAttention.forward": ["Attention", "False", "ModelAttention", "Module", "None", "Tensor", "ValueError", "_", "_absolute_position_to_relative_position", "_get_relative_embeddings", "_relative_position_to_absolute_position", "_shape", "attention_mask", "attn_output", "attn_probs", "attn_weights", "attn_weights_reshaped", "be", "bmm", "bool", "bsz", "but", "class", "def", "dim", "dropout", "else", "emb_rel_k", "emb_rel_v", "embed_dim", "f", "forward", "functional", "head_dim", "hidden_states", "if", "is", "k_proj", "key_relative_embeddings", "key_states", "key_value_states", "mask", "matmul", "nn", "not", "num_heads", "of", "out_proj", "output_attentions", "p", "proj_shape", "q_proj", "query_states", "raise", "rel_pos_bias", "relative_logits", "relative_weights", "reshape", "return", "scaling", "self", "should", "size", "softmax", "src_len", "tgt_len", "torch", "training", "transpose", "tuple", "v_proj", "value_relative_embeddings", "value_states", "view", "weights", "window_size"], "vits/modeling_vits.py:VitsAttention._get_relative_embeddings": ["ModelAttention", "Module", "_get_relative_embeddings", "class", "def", "functional", "if", "length", "max", "nn", "pad", "pad_length", "relative_embeddings", "return", "self", "slice_end_position", "slice_start_position", "window_size"], "vits/modeling_vits.py:VitsAttention._relative_position_to_absolute_position": ["ModelAttention", "Module", "_", "_relative_position_to_absolute_position", "batch_heads", "class", "def", "functional", "length", "nn", "pad", "return", "self", "size", "view", "x", "x_final", "x_flat"], "vits/modeling_vits.py:VitsAttention._absolute_position_to_relative_position": ["ModelAttention", "Module", "_", "_absolute_position_to_relative_position", "batch_heads", "class", "def", "functional", "length", "nn", "pad", "return", "self", "size", "view", "x", "x_final", "x_flat"], "vits/modeling_vits.py:VitsFeedForward.__init__": ["ACT2FN", "Conv1d", "Dropout", "ModelFeedForward", "Module", "None", "__init__", "act_fn", "activation_dropout", "class", "config", "conv_1", "conv_2", "def", "dropout", "else", "ffn_dim", "ffn_kernel_size", "hidden_act", "hidden_size", "if", "isinstance", "nn", "pad_left", "pad_right", "padding", "self", "str", "super"], "vits/modeling_vits.py:VitsFeedForward.forward": ["ModelFeedForward", "Module", "None", "act_fn", "class", "conv_1", "conv_2", "def", "dropout", "forward", "functional", "hidden_states", "if", "is", "nn", "not", "pad", "padding", "padding_mask", "permute", "return", "self"], "vits/modeling_vits.py:VitsEncoderLayer.__init__": ["Dropout", "GradientCheckpointingLayer", "LayerNorm", "ModelAttention", "ModelConfig", "ModelEncoderLayer", "ModelFeedForward", "__init__", "attention", "class", "config", "def", "dropout", "eps", "feed_forward", "final_layer_norm", "hidden_dropout", "hidden_size", "layer_norm", "layer_norm_eps", "nn", "self", "super"], "vits/modeling_vits.py:VitsEncoderLayer.forward": ["False", "FloatTensor", "GradientCheckpointingLayer", "ModelEncoderLayer", "None", "Tensor", "attention", "attention_mask", "attn_weights", "bool", "class", "def", "dropout", "feed_forward", "final_layer_norm", "forward", "hidden_states", "if", "layer_norm", "output_attentions", "outputs", "padding_mask", "residual", "return", "self", "torch"], "vits/modeling_vits.py:VitsEncoder.__init__": ["False", "ModelConfig", "ModelEncoder", "ModelEncoderLayer", "Module", "ModuleList", "_", "__init__", "class", "config", "def", "for", "gradient_checkpointing", "in", "layerdrop", "layers", "nn", "num_hidden_layers", "range", "self", "super"], "vits/modeling_vits.py:VitsEncoder.forward": ["BaseModelOutput", "FloatTensor", "ModelEncoder", "Module", "None", "Tensor", "_prepare_4d_attention_mask", "all_hidden_states", "all_self_attentions", "and", "attention_mask", "attentions", "bool", "class", "def", "dropout_probability", "dtype", "else", "encoder_layer", "for", "forward", "hidden_states", "if", "in", "is", "is_deepspeed_zero3_enabled", "is_fsdp_managed_module", "last_hidden_state", "layer_outputs", "layerdrop", "layers", "nn", "not", "np", "or", "output_attentions", "output_hidden_states", "padding_mask", "random", "return", "return_dict", "self", "skip_the_layer", "synced_gpus", "torch", "training", "tuple", "uniform", "v"], "vits/modeling_vits.py:VitsTextEncoder.__init__": ["Conv1d", "Embedding", "ModelConfig", "ModelEncoder", "ModelTextEncoder", "Module", "__init__", "class", "config", "def", "embed_tokens", "encoder", "flow_size", "hidden_size", "kernel_size", "nn", "pad_token_id", "project", "self", "super", "vocab_size"], "vits/modeling_vits.py:VitsTextEncoder.forward": ["FloatTensor", "ModelTextEncoder", "ModelTextEncoderOutput", "Module", "None", "Tensor", "True", "attention_mask", "attentions", "bool", "class", "config", "def", "dim", "else", "embed_tokens", "encoder", "encoder_outputs", "flow_size", "forward", "hidden_size", "hidden_states", "if", "input_ids", "last_hidden_state", "math", "nn", "not", "output_attentions", "output_hidden_states", "outputs", "padding_mask", "prior_log_variances", "prior_means", "project", "return", "return_dict", "self", "split", "sqrt", "stats", "torch", "transpose", "tuple"], "vits/modeling_vits.py:VitsPreTrainedModel._init_weights": ["Conv1d", "ConvTranspose1d", "Embedding", "False", "LayerNorm", "Linear", "ModelAttention", "ModelElementwiseAffine", "ModelPreTrainedModel", "Module", "None", "PreTrainedModel", "_init_weights", "_is_hf_initialized", "a", "and", "b", "bias", "class", "config", "def", "elif", "emb_rel_k", "emb_rel_v", "getattr", "groups", "head_dim", "hidden_size", "if", "in_channels", "init", "initializer_range", "is", "isinstance", "k", "kaiming_normal_", "kernel_size", "log_scale", "math", "mean", "module", "nn", "normal_", "not", "num_attention_heads", "ones_", "padding_idx", "self", "sqrt", "std", "translate", "uniform_", "weight", "window_size", "zeros_"], "vits/modeling_vits.py:VitsModel.__init__": ["Embedding", "ModelConfig", "ModelDurationPredictor", "ModelHifiGan", "ModelModel", "ModelPosteriorEncoder", "ModelPreTrainedModel", "ModelResidualCouplingBlock", "ModelStochasticDurationPredictor", "ModelTextEncoder", "__init__", "class", "config", "decoder", "def", "duration_predictor", "else", "embed_speaker", "flow", "if", "nn", "noise_scale", "noise_scale_duration", "num_speakers", "post_init", "posterior_encoder", "self", "speaker_embedding_size", "speaking_rate", "super", "text_encoder", "use_stochastic_duration_prediction"], "vits/modeling_vits.py:VitsModel.forward": ["Any", "FloatTensor", "Model", "ModelModel", "ModelModelOutput", "ModelPreTrainedModel", "None", "NotImplementedError", "Set", "Tensor", "Training", "True", "ValueError", "_", "and", "arange", "attention_mask", "attentions", "attn", "attn_mask", "batch_size", "bool", "ceil", "clamp_min", "class", "config", "cum_duration", "cumsum", "decoder", "def", "device", "dtype", "duration", "duration_predictor", "else", "embed_speaker", "embed_tokens", "exp", "f", "fill_value", "flow", "forward", "full", "functional", "hidden_states", "if", "in", "indices", "input_ids", "input_length", "input_padding_mask", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "latents", "length_scale", "log_duration", "long", "mask_dtype", "matmul", "max", "nn", "noise_scale", "noise_scale_duration", "not", "np", "num_speakers", "of", "ones_like", "output_attentions", "output_hidden_states", "output_length", "output_padding_mask", "outputs", "pad", "padded_indices", "padding_mask", "predicted_lengths", "prior_latents", "prior_log_variances", "prior_means", "prod", "r", "raise", "randn_like", "range", "return", "return_dict", "reverse", "self", "sequence_lengths", "shape", "size", "speaker_embeddings", "speaker_id", "speaking_rate", "spectrogram", "squeeze", "sum", "supported", "text_encoder", "text_encoder_output", "the", "to", "torch", "transpose", "tuple", "unsqueeze", "upsample_rates", "use_return_dict", "use_stochastic_duration_prediction", "valid_indices", "view", "waveform", "weight", "yet"], "vivit/modeling_vivit.py:VivitTubeletEmbeddings.__init__": ["Conv3d", "ModelConfig", "ModelTubeletEmbeddings", "Module", "__init__", "class", "config", "def", "embed_dim", "hidden_size", "image_size", "kernel_size", "nn", "num_channels", "num_frames", "num_patches", "patch_size", "projection", "self", "stride", "super", "tubelet_size"], "vivit/modeling_vivit.py:VivitTubeletEmbeddings.forward": ["False", "Image", "ModelTubeletEmbeddings", "Module", "Tensor", "ValueError", "and", "batch_size", "bool", "class", "def", "doesn", "f", "flatten", "forward", "height", "if", "image", "image_size", "interpolate_pos_encoding", "match", "model", "nn", "not", "num_channels", "num_frames", "or", "permute", "pixel_values", "projection", "raise", "return", "self", "shape", "size", "t", "torch", "transpose", "width", "x"], "vivit/modeling_vivit.py:VivitEmbeddings.__init__": ["Dropout", "ModelConfig", "ModelEmbeddings", "ModelTubeletEmbeddings", "Module", "Parameter", "__init__", "class", "cls_token", "config", "def", "dropout", "hidden_dropout_prob", "hidden_size", "nn", "num_patches", "patch_embeddings", "patch_size", "position_embeddings", "self", "super", "torch", "tubelet_size", "zeros"], "vivit/modeling_vivit.py:VivitEmbeddings.interpolate_pos_encoding": ["False", "ModelEmbeddings", "Module", "Tensor", "align_corners", "and", "bicubic", "cat", "class", "class_pos_embed", "def", "dim", "embeddings", "functional", "height", "if", "int", "interpolate", "interpolate_pos_encoding", "is_tracing", "jit", "mode", "new_height", "new_width", "nn", "not", "num_patches", "num_positions", "patch_pos_embed", "patch_size", "permute", "position_embeddings", "reshape", "return", "self", "shape", "size", "sqrt_num_positions", "torch", "torch_int", "view", "width"], "vivit/modeling_vivit.py:VivitEmbeddings.forward": ["False", "ModelEmbeddings", "Module", "Tensor", "batch_size", "bool", "cat", "class", "cls_token", "cls_tokens", "def", "dim", "dropout", "else", "embeddings", "forward", "height", "if", "interpolate_pos_encoding", "nn", "num_channels", "num_frames", "patch_embeddings", "pixel_values", "position_embeddings", "return", "self", "shape", "tile", "torch", "width"], "vivit/modeling_vivit.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "contiguous", "def", "dim", "dropout", "float", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "shape", "size", "softmax", "torch", "training", "transpose", "value"], "vivit/modeling_vivit.py:VivitSelfAttention.__init__": ["False", "Linear", "ModelConfig", "ModelSelfAttention", "Module", "The", "ValueError", "__init__", "a", "all_head_size", "and", "attention", "attention_head_size", "attention_probs_dropout_prob", "bias", "class", "config", "def", "dropout_prob", "embedding_size", "f", "hasattr", "heads", "hidden", "hidden_size", "if", "int", "is", "is_causal", "key", "multiple", "nn", "not", "num_attention_heads", "number", "of", "qkv_bias", "query", "raise", "scaling", "self", "size", "super", "the", "value"], "vivit/modeling_vivit.py:VivitSelfAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "ModelSelfAttention", "Module", "None", "Tensor", "_attn_implementation", "all_head_size", "attention_head_size", "attention_interface", "attention_probs", "batch_size", "class", "config", "context_layer", "def", "dropout", "dropout_prob", "eager", "eager_attention_forward", "else", "forward", "hidden_states", "if", "is_causal", "key", "key_layer", "new_context_layer_shape", "new_shape", "nn", "not", "num_attention_heads", "query", "query_layer", "reshape", "return", "scaling", "self", "shape", "size", "torch", "training", "transpose", "tuple", "value", "value_layer", "view"], "vivit/modeling_vivit.py:VivitSelfOutput.__init__": ["Dropout", "Linear", "ModelConfig", "ModelSelfOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "hidden_dropout_prob", "hidden_size", "nn", "self", "super"], "vivit/modeling_vivit.py:VivitSelfOutput.forward": ["ModelSelfOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "vivit/modeling_vivit.py:VivitAttention.__init__": ["ModelAttention", "ModelConfig", "ModelSelfAttention", "ModelSelfOutput", "Module", "__init__", "attention", "class", "config", "def", "nn", "output", "self", "super"], "vivit/modeling_vivit.py:VivitAttention.forward": ["ModelAttention", "Module", "Tensor", "_", "attention", "class", "def", "forward", "hidden_states", "nn", "output", "return", "self", "self_attn_output", "torch"], "vivit/modeling_vivit.py:VivitIntermediate.__init__": ["ACT2FN", "Dropout", "Linear", "ModelConfig", "ModelIntermediate", "Module", "__init__", "class", "config", "def", "dense", "dropout", "else", "hidden_act", "hidden_dropout_prob", "hidden_size", "if", "intermediate_act_fn", "intermediate_size", "isinstance", "nn", "self", "str", "super"], "vivit/modeling_vivit.py:VivitIntermediate.forward": ["ModelIntermediate", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "intermediate_act_fn", "nn", "return", "self", "torch"], "vivit/modeling_vivit.py:VivitOutput.__init__": ["Dropout", "Linear", "ModelConfig", "ModelOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "hidden_dropout_prob", "hidden_size", "intermediate_size", "nn", "self", "super"], "vivit/modeling_vivit.py:VivitOutput.forward": ["ModelOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "vivit/modeling_vivit.py:VivitLayer.__init__": ["GradientCheckpointingLayer", "LayerNorm", "ModelAttention", "ModelConfig", "ModelIntermediate", "ModelLayer", "ModelOutput", "__init__", "attention", "chunk_size_feed_forward", "class", "config", "def", "eps", "hidden_size", "intermediate", "layer_norm_eps", "layernorm_after", "layernorm_before", "nn", "output", "self", "seq_len_dim", "super"], "vivit/modeling_vivit.py:VivitLayer.forward": ["GradientCheckpointingLayer", "ModelLayer", "Tensor", "attention", "attention_output", "class", "def", "forward", "hidden_states", "hidden_states_norm", "intermediate", "layer_output", "layernorm_after", "layernorm_before", "output", "return", "self", "torch"], "vivit/modeling_vivit.py:VivitEncoder.__init__": ["False", "ModelConfig", "ModelEncoder", "ModelLayer", "Module", "ModuleList", "_", "__init__", "class", "config", "def", "for", "gradient_checkpointing", "in", "layer", "nn", "num_hidden_layers", "range", "self", "super"], "vivit/modeling_vivit.py:VivitEncoder.forward": ["BaseModelOutput", "ModelEncoder", "Module", "Tensor", "class", "def", "enumerate", "for", "forward", "hidden_states", "i", "in", "last_hidden_state", "layer", "layer_module", "nn", "return", "self", "torch"], "vivit/modeling_vivit.py:VivitPooler.__init__": ["Linear", "ModelConfig", "ModelPooler", "Module", "Tanh", "__init__", "activation", "class", "config", "def", "dense", "hidden_size", "nn", "self", "super"], "vivit/modeling_vivit.py:VivitPooler.forward": ["ModelPooler", "Module", "Tensor", "activation", "class", "def", "dense", "first_token_tensor", "forward", "hidden_states", "nn", "pooled_output", "return", "self", "torch"], "vivit/modeling_vivit.py:VivitPreTrainedModel._init_weights": ["ModelEmbeddings", "ModelPreTrainedModel", "PreTrainedModel", "_init_weights", "class", "cls_token", "def", "if", "init", "isinstance", "module", "position_embeddings", "self", "super", "zeros_"], "vivit/modeling_vivit.py:VivitModel.__init__": ["LayerNorm", "ModelConfig", "ModelEmbeddings", "ModelEncoder", "ModelModel", "ModelPooler", "ModelPreTrainedModel", "None", "True", "__init__", "add_pooling_layer", "bool", "class", "config", "def", "else", "embeddings", "encoder", "eps", "hidden_size", "if", "layer_norm_eps", "layernorm", "nn", "pooler", "post_init", "r", "self", "super"], "vivit/modeling_vivit.py:VivitModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "patch_embeddings", "return", "self"], "vivit/modeling_vivit.py:VivitModel.forward": ["BaseModelOutput", "BaseModelOutputWithPooling", "False", "FloatTensor", "ModelModel", "ModelPreTrainedModel", "None", "TransformersKwargs", "Unpack", "ValueError", "You", "bool", "class", "def", "else", "embedding_output", "embeddings", "encoder", "encoder_outputs", "forward", "have", "if", "interpolate_pos_encoding", "is", "kwargs", "last_hidden_state", "layernorm", "not", "pixel_values", "pooled_output", "pooler", "pooler_output", "r", "raise", "return", "self", "sequence_output", "specify", "to", "torch"], "vivit/modeling_vivit.py:VivitForVideoClassification.__init__": ["False", "Identity", "Linear", "Model", "ModelConfig", "ModelForVideoClassification", "ModelModel", "ModelPreTrainedModel", "__init__", "add_pooling_layer", "class", "classifier", "config", "def", "else", "hidden_size", "if", "nn", "num_labels", "post_init", "self", "super"], "vivit/modeling_vivit.py:VivitForVideoClassification.forward": ["BaseModelOutput", "False", "FloatTensor", "ImageClassifierOutput", "LongTensor", "Model", "ModelForVideoClassification", "ModelPreTrainedModel", "None", "TransformersKwargs", "Unpack", "attentions", "bool", "class", "classifier", "config", "def", "forward", "hidden_states", "if", "interpolate_pos_encoding", "is", "kwargs", "labels", "last_hidden_state", "logits", "loss", "loss_function", "not", "outputs", "pixel_values", "r", "return", "self", "sequence_output", "torch"], "vjepa2/modeling_vjepa2.py:VJEPA2WithMaskedInputModelOutput.to_tuple": ["Model", "ModelOutput", "ModelWithMaskedInputPredictorOutput", "class", "def", "if", "isinstance", "list", "output", "return", "self", "super", "to_tuple", "tuple"], "vjepa2/modeling_vjepa2.py:VJEPA2PatchEmbeddings3D.__init__": ["Conv3d", "Model", "ModelConfig", "Module", "__init__", "class", "config", "def", "hidden_size", "in_channels", "in_chans", "int", "kernel_size", "nn", "out_channels", "patch_size", "proj", "self", "stride", "super", "tubelet_size"], "vjepa2/modeling_vjepa2.py:VJEPA2PatchEmbeddings3D.num_patches": ["Model", "Module", "class", "config", "crop_size", "def", "frames_per_clip", "nn", "num_patches", "patch_size", "return", "tubelet_size"], "vjepa2/modeling_vjepa2.py:VJEPA2PatchEmbeddings3D.forward": ["Model", "Module", "Tensor", "class", "def", "flatten", "forward", "nn", "pixel_values_videos", "proj", "return", "self", "torch", "transpose", "x"], "vjepa2/modeling_vjepa2.py:VJEPA2Embeddings.__init__": ["Model", "ModelConfig", "ModelPatchEmbeddings3D", "Module", "__init__", "class", "config", "def", "hidden_size", "int", "nn", "num_patches", "patch_embeddings", "patch_size", "self", "super"], "vjepa2/modeling_vjepa2.py:VJEPA2Embeddings.forward": ["Model", "Module", "Tensor", "class", "config", "def", "dtype", "embeddings", "forward", "if", "nn", "num_frames", "patch_embeddings", "permute", "pixel_values_videos", "proj", "repeat", "return", "self", "shape", "target_dtype", "to", "torch", "tubelet_size", "weight"], "vjepa2/modeling_vjepa2.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "attention_mask", "attn_output", "attn_weights", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "key", "kwargs", "matmul", "module", "nn", "p", "query", "return", "scaling", "softmax", "to", "torch", "training", "transpose", "value"], "vjepa2/modeling_vjepa2.py:rotate_queries_or_keys": ["B", "D", "Model_queries_or_keys", "N", "arange", "cos", "def", "device", "dim", "dtype", "emb_cos", "emb_sin", "flatten", "freq", "num_heads", "omega", "pos", "repeat", "return", "sin", "size", "squeeze", "stack", "torch", "unbind", "unflatten", "unsqueeze", "x", "y", "y1", "y2"], "vjepa2/modeling_vjepa2.py:VJEPA2RopeAttention.__init__": ["Dropout", "False", "Linear", "Model", "ModelConfig", "Module", "The", "ValueError", "__init__", "a", "all_head_size", "attention", "attention_head_size", "attention_probs_dropout_prob", "bias", "class", "config", "crop_size", "d_dim", "def", "dropout", "dropout_prob", "f", "frames_per_clip", "grid_depth", "grid_size", "h_dim", "heads", "hidden", "hidden_size", "if", "int", "is", "is_causal", "key", "multiple", "nn", "not", "num_attention_heads", "number", "of", "patch_size", "proj", "qkv_bias", "query", "raise", "scaling", "self", "size", "super", "the", "tubelet_size", "value", "w_dim"], "vjepa2/modeling_vjepa2.py:VJEPA2RopeAttention._get_frame_pos": ["Model", "Module", "_get_frame_pos", "class", "def", "grid_size", "ids", "int", "nn", "return", "self", "tokens_per_frame"], "vjepa2/modeling_vjepa2.py:VJEPA2RopeAttention._get_height_pos": ["Model", "Module", "_get_frame_pos", "_get_height_pos", "class", "def", "frame_ids", "grid_size", "ids", "int", "nn", "return", "self", "tokens_per_frame", "tokens_per_row"], "vjepa2/modeling_vjepa2.py:VJEPA2RopeAttention.get_position_ids": ["Model", "Module", "None", "_get_frame_pos", "_get_height_pos", "arange", "class", "def", "device", "else", "frame_ids", "get_position_ids", "grid_size", "height_ids", "ids", "if", "int", "is", "masks", "nn", "not", "num_attention_heads", "repeat", "return", "self", "size", "token_size", "tokens_per_frame", "tokens_per_row", "torch", "unsqueeze", "width_ids", "x"], "vjepa2/modeling_vjepa2.py:VJEPA2RopeAttention.apply_rotary_embeddings": ["Model", "Module", "apply_rotary_embeddings", "attention_head_size", "cat", "class", "d_dim", "d_mask", "def", "dim", "else", "h_dim", "h_mask", "if", "nn", "pos", "pos_ids", "qk", "qkd", "qkh", "qkr", "qkw", "return", "rotate_queries_or_keys", "s", "self", "torch", "w_dim", "w_mask"], "vjepa2/modeling_vjepa2.py:VJEPA2RopeAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "False", "Model", "Module", "None", "Tensor", "_", "_attn_implementation", "all_head_size", "apply_rotary_embeddings", "attention_head_size", "attention_interface", "attention_probs", "batch_size", "bool", "class", "config", "context_layer", "def", "dropout", "dropout_prob", "eager", "eager_attention_forward", "else", "forward", "get_position_ids", "hidden_states", "if", "is_causal", "key", "key_layer", "masks", "new_context_layer_shape", "nn", "not", "num_attention_heads", "output_attentions", "outputs", "pos_ids", "position_mask", "proj", "query", "query_layer", "reshape", "return", "scaling", "self", "seq_length", "shape", "size", "torch", "training", "transpose", "tuple", "value", "value_layer", "view"], "vjepa2/modeling_vjepa2.py:drop_path": ["False", "Model_path", "Model_prob", "Tensor", "bool", "def", "device", "div", "dtype", "float", "floor_", "if", "input", "keep_prob", "ndim", "not", "or", "output", "rand", "random_tensor", "return", "shape", "torch", "training"], "vjepa2/modeling_vjepa2.py:VJEPA2DropPath.__init__": ["Model", "Module", "None", "__init__", "class", "def", "drop_prob", "float", "nn", "self", "super"], "vjepa2/modeling_vjepa2.py:VJEPA2DropPath.forward": ["Model", "Module", "Tensor", "class", "def", "drop_path", "drop_prob", "forward", "hidden_states", "nn", "return", "self", "torch", "training"], "vjepa2/modeling_vjepa2.py:VJEPA2DropPath.extra_repr": ["Model", "Module", "class", "def", "drop_prob", "extra_repr", "f", "nn", "p", "return", "self", "str"], "vjepa2/modeling_vjepa2.py:VJEPA2MLP.__init__": ["ACT2FN", "Linear", "Model", "ModelConfig", "Module", "True", "__init__", "activation", "bias", "class", "config", "def", "fc1", "fc2", "float", "hidden_act", "hidden_features", "hidden_size", "in_features", "int", "mlp_ratio", "nn", "out_features", "self", "super"], "vjepa2/modeling_vjepa2.py:VJEPA2MLP.forward": ["Model", "Module", "Tensor", "activation", "class", "def", "fc1", "fc2", "forward", "hidden_state", "nn", "return", "self", "torch"], "vjepa2/modeling_vjepa2.py:VJEPA2Layer.__init__": ["GradientCheckpointingLayer", "Identity", "LayerNorm", "Model", "ModelConfig", "ModelDropPath", "ModelMLP", "ModelRopeAttention", "__init__", "attention", "class", "config", "def", "drop_path", "drop_path_rate", "else", "eps", "float", "hidden_size", "if", "int", "layer_norm_eps", "mlp", "mlp_ratio", "nn", "norm1", "norm2", "num_attention_heads", "self", "super"], "vjepa2/modeling_vjepa2.py:VJEPA2Layer.forward": ["False", "GradientCheckpointingLayer", "Model", "None", "Tensor", "attention", "attention_output", "bool", "class", "def", "drop_path", "forward", "hidden_states", "mlp", "norm1", "norm2", "output_attentions", "outputs", "position_mask", "residual", "return", "self", "self_attention_outputs", "torch", "tuple"], "vjepa2/modeling_vjepa2.py:VJEPA2Encoder.__init__": ["False", "LayerNorm", "Model", "ModelConfig", "ModelEmbeddings", "ModelLayer", "Module", "ModuleList", "__init__", "class", "config", "def", "drop_path_rate", "drop_path_rates", "else", "embeddings", "eps", "for", "gradient_checkpointing", "hidden_size", "i", "if", "in", "layer", "layer_norm_eps", "layernorm", "mlp_ratio", "nn", "num_attention_heads", "num_hidden_layers", "range", "self", "super"], "vjepa2/modeling_vjepa2.py:VJEPA2Encoder.forward": ["BaseModelOutput", "False", "Model", "Module", "None", "Tensor", "all_hidden_states", "all_self_attentions", "attentions", "bool", "class", "def", "else", "embeddings", "enumerate", "for", "forward", "hidden_states", "i", "if", "in", "kwargs", "last_hidden_state", "layer", "layer_module", "layer_outputs", "layernorm", "nn", "output_attentions", "output_hidden_states", "pixel_values_videos", "return", "self", "torch"], "vjepa2/modeling_vjepa2.py:apply_masks": ["Model_masks", "Tensor", "all_masked_tensors", "cat", "def", "device", "dim", "for", "gather", "in", "index", "list", "mask", "mask_keep", "masks", "repeat", "return", "size", "tensor", "to", "torch", "unsqueeze"], "vjepa2/modeling_vjepa2.py:VJEPA2PredictorEmbeddings.__init__": ["Linear", "Model", "ModelConfig", "Module", "Parameter", "__init__", "class", "config", "def", "hidden_size", "mask_tokens", "nn", "num_mask_tokens", "patch_size", "pred_hidden_size", "pred_num_mask_tokens", "pred_zero_init_mask_tokens", "predictor_embeddings", "self", "super", "torch", "zero_init_mask_tokens", "zeros"], "vjepa2/modeling_vjepa2.py:VJEPA2PredictorEmbeddings.num_patches": ["Model", "Module", "class", "config", "crop_size", "def", "else", "frames_per_clip", "if", "nn", "num_patches", "patch_size", "return", "tubelet_size"], "vjepa2/modeling_vjepa2.py:VJEPA2PredictorEmbeddings.forward": ["B", "Model", "Module", "Tensor", "apply_masks", "cat", "class", "cm", "context", "context_mask", "def", "dim", "embeddings", "forward", "hidden_states", "int", "len", "list", "mask_index", "mask_tokens", "masks", "max", "max_patch_num", "nn", "num_mask_tokens", "predictor_embeddings", "repeat", "return", "self", "size", "target", "target_mask", "tm", "torch", "tuple"], "vjepa2/modeling_vjepa2.py:VJEPA2Predictor.__init__": ["False", "LayerNorm", "Linear", "Model", "ModelConfig", "ModelEmbeddings", "ModelLayer", "Module", "ModuleList", "True", "__init__", "bias", "class", "config", "def", "drop_path_rate", "drop_path_rates", "else", "embeddings", "eps", "for", "gradient_checkpointing", "hidden_size", "i", "if", "in", "layer", "layer_norm_eps", "layernorm", "mlp_ratio", "nn", "num_attention_heads", "pred_hidden_size", "pred_mlp_ratio", "pred_num_attention_heads", "pred_num_hidden_layers", "proj", "range", "self", "super"], "vjepa2/modeling_vjepa2.py:VJEPA2Predictor.sort_tokens": ["Model", "Module", "argsort", "class", "def", "device", "dim", "expand", "gather", "hidden_states", "hidden_states_argsort", "index", "nn", "position_masks", "return", "self", "size", "sort_tokens", "to", "torch", "unsqueeze"], "vjepa2/modeling_vjepa2.py:VJEPA2Predictor.unsort_tokens": ["Model", "Module", "argsort", "class", "def", "device", "dim", "expand", "gather", "hidden_states", "index", "nn", "return", "reverse_argsort", "self", "size", "to", "torch", "unsort_tokens", "unsqueeze"], "vjepa2/modeling_vjepa2.py:VJEPA2Predictor.forward": ["BaseModelOutput", "D", "False", "Model", "Module", "N_ctxt", "None", "Tensor", "_", "all_hidden_states", "all_self_attentions", "apply_masks", "argsort", "attentions", "bool", "class", "context_mask", "def", "dim", "else", "embeddings", "encoder_hidden_states", "enumerate", "for", "forward", "hidden_states", "i", "if", "in", "kwargs", "last_hidden_state", "layer", "layer_module", "layer_outputs", "layernorm", "list", "nn", "output_attentions", "output_hidden_states", "position_masks", "proj", "return", "self", "shape", "sort_tokens", "target_mask", "torch", "unsort_tokens"], "vjepa2/modeling_vjepa2.py:VJEPA2PoolerSelfAttention.__init__": ["False", "Linear", "Model", "ModelConfig", "Module", "ValueError", "__init__", "and", "attention_dropout", "be", "by", "class", "config", "def", "divisible", "dropout", "embed_dim", "f", "got", "head_dim", "hidden_size", "if", "is_causal", "k_proj", "must", "nn", "num_attention_heads", "num_heads", "out_proj", "q_proj", "raise", "scale", "self", "super", "v_proj"], "vjepa2/modeling_vjepa2.py:VJEPA2PoolerSelfAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "False", "Model", "Module", "None", "Tensor", "_attn_implementation", "attention_interface", "attention_mask", "attn_output", "attn_weights", "batch_size", "bool", "class", "config", "contiguous", "def", "dropout", "eager", "eager_attention_forward", "else", "embed_dim", "forward", "head_dim", "hidden_states", "if", "is_causal", "k_proj", "keys", "nn", "not", "num_heads", "out_proj", "output_attentions", "q_proj", "queries", "reshape", "return", "scale", "scaling", "self", "seq_length", "shape", "torch", "training", "transpose", "tuple", "v_proj", "values", "view"], "vjepa2/modeling_vjepa2.py:VJEPA2PoolerCrossAttention.__init__": ["False", "Linear", "Model", "ModelConfig", "Module", "ValueError", "__init__", "and", "attention_dropout", "be", "by", "class", "config", "def", "divisible", "dropout", "embed_dim", "f", "got", "head_dim", "hidden_size", "if", "is_causal", "k_proj", "must", "nn", "num_attention_heads", "num_heads", "q_proj", "raise", "scale", "self", "super", "v_proj"], "vjepa2/modeling_vjepa2.py:VJEPA2PoolerCrossAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "False", "Model", "Module", "None", "Tensor", "_attn_implementation", "attention_interface", "attention_mask", "attn_output", "attn_weights", "batch_size", "bool", "class", "config", "contiguous", "def", "dropout", "eager", "eager_attention_forward", "else", "embed_dim", "forward", "head_dim", "if", "is_causal", "k_proj", "keys", "kv_seq_length", "nn", "not", "num_heads", "output_attentions", "q_proj", "q_seq_length", "queries", "reshape", "return", "scale", "scaling", "self", "shape", "torch", "training", "transpose", "tuple", "v_proj", "values", "view"], "vjepa2/modeling_vjepa2.py:VJEPA2PoolerSelfAttentionLayer.__init__": ["GradientCheckpointingLayer", "LayerNorm", "Model", "ModelConfig", "ModelMLP", "ModelPoolerSelfAttention", "__init__", "class", "config", "def", "eps", "hidden_size", "layer_norm1", "layer_norm2", "layer_norm_eps", "mlp", "nn", "self", "self_attn", "super"], "vjepa2/modeling_vjepa2.py:VJEPA2PoolerSelfAttentionLayer.forward": ["False", "GradientCheckpointingLayer", "Model", "None", "Tensor", "attention_mask", "attn_weights", "bool", "class", "def", "forward", "hidden_states", "if", "layer_norm1", "layer_norm2", "mlp", "output_attentions", "outputs", "residual", "return", "self", "self_attn", "torch", "tuple"], "vjepa2/modeling_vjepa2.py:VJEPA2PoolerCrossAttentionLayer.__init__": ["GradientCheckpointingLayer", "LayerNorm", "Model", "ModelConfig", "ModelMLP", "ModelPoolerCrossAttention", "__init__", "class", "config", "cross_attn", "def", "eps", "hidden_size", "layer_norm1", "layer_norm2", "layer_norm_eps", "mlp", "nn", "self", "super"], "vjepa2/modeling_vjepa2.py:VJEPA2PoolerCrossAttentionLayer.forward": ["False", "GradientCheckpointingLayer", "Model", "None", "Tensor", "attention_mask", "attn_weights", "bool", "class", "cross_attn", "def", "forward", "hidden_state", "if", "layer_norm1", "layer_norm2", "mlp", "output_attentions", "outputs", "queries", "residual", "return", "self", "torch", "tuple"], "vjepa2/modeling_vjepa2.py:VJEPA2AttentivePooler.__init__": ["Model", "ModelConfig", "ModelPoolerCrossAttentionLayer", "ModelPoolerSelfAttentionLayer", "Module", "ModuleList", "Parameter", "_", "__init__", "class", "config", "cross_attention_layer", "def", "for", "hidden_size", "in", "nn", "num_pooler_layers", "query_tokens", "range", "self", "self_attention_layers", "super", "torch", "zeros"], "vjepa2/modeling_vjepa2.py:VJEPA2AttentivePooler.forward": ["Model", "Module", "None", "Tensor", "attention_mask", "class", "cross_attention_layer", "def", "for", "forward", "hidden_state", "in", "layer", "nn", "queries", "query_tokens", "repeat", "return", "self", "self_attention_layers", "shape", "squeeze", "torch"], "vjepa2/modeling_vjepa2.py:VJEPA2PreTrainedModel._init_weights": ["Conv2d", "Conv3d", "LayerNorm", "Linear", "Model", "ModelAttentivePooler", "ModelPredictorEmbeddings", "None", "PreTrainedModel", "_init_weights", "bias", "class", "config", "cross_attention_layer", "def", "elif", "else", "enumerate", "fc2", "for", "i", "if", "in", "init", "init_std", "initializer_range", "is", "isinstance", "layer", "len", "mask_tokens", "mlp", "module", "nn", "not", "ones_", "out_proj", "query_tokens", "self", "self_attention_layers", "self_attn", "std", "trunc_normal_", "weight", "zero_init_mask_tokens", "zeros_"], "vjepa2/modeling_vjepa2.py:VJEPA2Model.__init__": ["Model", "ModelConfig", "ModelEncoder", "ModelPreTrainedModel", "ModelPredictor", "__init__", "class", "config", "def", "encoder", "post_init", "predictor", "self", "super"], "vjepa2/modeling_vjepa2.py:VJEPA2Model.get_input_embeddings": ["Model", "ModelPatchEmbeddings3D", "ModelPreTrainedModel", "class", "def", "embeddings", "encoder", "get_input_embeddings", "patch_embeddings", "return", "self"], "vjepa2/modeling_vjepa2.py:VJEPA2Model.forward": ["B", "BaseModelOutput", "False", "Model", "ModelPreTrainedModel", "ModelWithMaskedInputModelOutput", "ModelWithMaskedInputPredictorOutput", "N", "None", "Tensor", "ValueError", "You", "and", "apply_masks", "arange", "attentions", "bool", "class", "config", "context_mask", "def", "device", "else", "encoder", "encoder_hidden_states", "encoder_output", "encoder_outputs", "forward", "have", "hidden_states", "if", "is", "kwargs", "last_hidden_state", "list", "masked_hidden_state", "not", "output_attentions", "output_hidden_states", "pixel_values_videos", "predictor", "predictor_output", "predictor_outputs", "r", "raise", "repeat", "return", "self", "sequence_output", "size", "skip_predictor", "specify", "target_hidden_state", "target_mask", "to", "torch", "unsqueeze"], "vjepa2/modeling_vjepa2.py:VJEPA2Model.get_vision_features": ["Model", "ModelPreTrainedModel", "Tensor", "True", "class", "def", "encoder_output", "forward", "get_vision_features", "last_hidden_state", "pixel_values_videos", "return", "self", "skip_predictor", "torch"], "vjepa2/modeling_vjepa2.py:VJEPA2ForVideoClassification.__init__": ["Linear", "Model", "ModelAttentivePooler", "ModelConfig", "ModelModel", "ModelPreTrainedModel", "True", "__init__", "bias", "class", "classifier", "config", "def", "hidden_size", "nn", "num_labels", "pooler", "post_init", "self", "super"], "vjepa2/modeling_vjepa2.py:VJEPA2ForVideoClassification.forward": ["ImageClassifierOutput", "Model", "ModelPreTrainedModel", "None", "Tensor", "True", "attentions", "bool", "class", "classifier", "config", "def", "forward", "hidden_states", "if", "is", "kwargs", "labels", "last_hidden_state", "logits", "loss", "loss_function", "not", "output_attentions", "output_hidden_states", "outputs", "pixel_values_videos", "pooled_logits", "pooler", "pooler_output", "r", "return", "self", "skip_predictor", "torch", "tuple"], "voxtral/modeling_voxtral.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "and", "attention_mask", "attn_output", "attn_weights", "contiguous", "def", "dim", "dropout", "float", "functional", "if", "is", "key", "kwargs", "matmul", "module", "ndim", "nn", "not", "p", "query", "return", "scaling", "shape", "size", "softmax", "torch", "training", "transpose", "value"], "voxtral/modeling_voxtral.py:VoxtralAttention.__init__": ["False", "Instantiating", "Linear", "ModelAttention", "ModelConfig", "Module", "None", "Please", "True", "ValueError", "__class__", "__init__", "__name__", "a", "and", "be", "bias", "bool", "by", "caching", "call", "class", "config", "creating", "decoder", "def", "divisible", "dropout", "during", "embed_dim", "errors", "f", "float", "forward", "got", "head_dim", "if", "int", "is", "is_causal", "is_decoder", "k_proj", "layer_idx", "logger", "make", "must", "nn", "not", "num_heads", "out_proj", "passing", "provide", "q_proj", "raise", "recommended", "scaling", "self", "super", "sure", "the", "this", "to", "used", "v_proj", "warning_once", "when", "will", "without"], "voxtral/modeling_voxtral.py:VoxtralAttention._shape": ["ModelAttention", "Module", "Tensor", "_shape", "bsz", "class", "contiguous", "def", "head_dim", "int", "nn", "num_heads", "return", "self", "seq_len", "tensor", "torch", "transpose", "view"], "voxtral/modeling_voxtral.py:VoxtralAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "False", "ModelAttention", "Module", "None", "Tensor", "_", "_attn_implementation", "_shape", "attention_interface", "attention_mask", "attn_output", "attn_weights", "bool", "bsz", "class", "config", "contiguous", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "hidden_states", "if", "k_proj", "key_states", "kwargs", "nn", "not", "out_proj", "output_attentions", "q_proj", "query_states", "reshape", "return", "scaling", "self", "size", "tgt_len", "torch", "training", "tuple", "v_proj", "value_states"], "voxtral/modeling_voxtral.py:VoxtralEncoderLayer.__init__": ["ACT2FN", "GradientCheckpointingLayer", "LayerNorm", "Linear", "ModelAttention", "ModelConfig", "ModelEncoderLayer", "__init__", "activation_dropout", "activation_fn", "activation_function", "attention_dropout", "class", "config", "d_model", "def", "dropout", "embed_dim", "encoder_attention_heads", "encoder_ffn_dim", "fc1", "fc2", "final_layer_norm", "nn", "num_heads", "self", "self_attn", "self_attn_layer_norm", "super"], "voxtral/modeling_voxtral.py:VoxtralEncoderLayer.forward": ["False", "GradientCheckpointingLayer", "ModelEncoderLayer", "Tensor", "activation_dropout", "activation_fn", "attention_mask", "attn_weights", "bool", "clamp", "clamp_value", "class", "def", "dropout", "dtype", "fc1", "fc2", "final_layer_norm", "finfo", "float16", "forward", "functional", "hidden_states", "if", "max", "min", "nn", "output_attentions", "p", "residual", "return", "self", "self_attn", "self_attn_layer_norm", "torch", "training"], "voxtral/modeling_voxtral.py:VoxtralEncoder.__init__": ["AvgPool1d", "Conv1d", "Embedding", "False", "LayerNorm", "ModelEncoder", "ModelEncoderConfig", "ModelEncoderLayer", "ModelPreTrainedModel", "ModuleList", "_", "__init__", "avg_pooler", "class", "config", "conv1", "conv2", "d_model", "def", "dropout", "else", "embed_dim", "embed_positions", "embed_scale", "encoder_layerdrop", "encoder_layers", "for", "gradient_checkpointing", "if", "in", "kernel_size", "layer_norm", "layerdrop", "layers", "math", "max_source_positions", "nn", "num_mel_bins", "pad_token_id", "padding", "padding_idx", "post_init", "range", "requires_grad_", "scale_embedding", "self", "sqrt", "stride", "super"], "voxtral/modeling_voxtral.py:VoxtralEncoder._freeze_parameters": ["False", "ModelEncoder", "ModelPreTrainedModel", "_freeze_parameters", "_requires_grad", "class", "def", "for", "in", "param", "parameters", "requires_grad", "self"], "voxtral/modeling_voxtral.py:VoxtralEncoder.get_input_embeddings": ["ModelEncoder", "ModelPreTrainedModel", "Module", "class", "conv1", "def", "get_input_embeddings", "nn", "return", "self"], "voxtral/modeling_voxtral.py:VoxtralEncoder.set_input_embeddings": ["ModelEncoder", "ModelPreTrainedModel", "Module", "class", "conv1", "def", "nn", "self", "set_input_embeddings", "value"], "voxtral/modeling_voxtral.py:VoxtralEncoder.forward": ["BaseModelOutput", "Make", "Model", "ModelEncoder", "ModelPreTrainedModel", "None", "TransformersKwargs", "Unpack", "ValueError", "attention_mask", "be", "but", "class", "config", "conv1", "conv2", "def", "device", "dropout", "dtype", "embed_pos", "embed_positions", "encoder_layer", "enumerate", "expected_seq_length", "expects", "f", "features", "for", "forward", "found", "functional", "gelu", "hidden_states", "idx", "if", "in", "input", "input_features", "inputs_embeds", "kwargs", "last_hidden_state", "layer_norm", "layer_outputs", "layers", "length", "max_source_positions", "mel", "nn", "of", "p", "pad", "permute", "r", "raise", "return", "self", "shape", "stride", "sure", "the", "to", "training", "weight"], "voxtral/modeling_voxtral.py:VoxtralEncoder._get_feat_extract_output_lengths": ["LongTensor", "ModelEncoder", "ModelPreTrainedModel", "_get_feat_extract_output_lengths", "class", "def", "input_lengths", "output_lengths", "return", "self", "torch"], "voxtral/modeling_voxtral.py:VoxtralMultiModalProjector.__init__": ["ACT2FN", "False", "Linear", "ModelConfig", "ModelMultiModalProjector", "Module", "__init__", "act", "audio_config", "bias", "class", "config", "def", "hidden_size", "intermediate_size", "linear_1", "linear_2", "nn", "projector_hidden_act", "self", "super", "text_config"], "voxtral/modeling_voxtral.py:VoxtralMultiModalProjector.forward": ["ModelMultiModalProjector", "Module", "act", "audio_features", "class", "def", "forward", "hidden_states", "linear_1", "linear_2", "nn", "return", "self"], "voxtral/modeling_voxtral.py:VoxtralForConditionalGeneration.__init__": ["AutoModel", "AutoModelForCausalLM", "GenerationMixin", "ModelForConditionalGeneration", "ModelMultiModalProjector", "ModelPreTrainedModel", "__init__", "audio_config", "audio_tower", "class", "config", "def", "from_config", "language_model", "multi_modal_projector", "post_init", "self", "super", "text_config", "vocab_size"], "voxtral/modeling_voxtral.py:VoxtralForConditionalGeneration.get_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "language_model", "return", "self"], "voxtral/modeling_voxtral.py:VoxtralForConditionalGeneration.set_input_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "language_model", "self", "set_input_embeddings", "value"], "voxtral/modeling_voxtral.py:VoxtralForConditionalGeneration.get_output_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "get_output_embeddings", "language_model", "return", "self"], "voxtral/modeling_voxtral.py:VoxtralForConditionalGeneration.set_output_embeddings": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "language_model", "new_embeddings", "self", "set_output_embeddings"], "voxtral/modeling_voxtral.py:VoxtralForConditionalGeneration.set_decoder": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "decoder", "def", "language_model", "self", "set_decoder"], "voxtral/modeling_voxtral.py:VoxtralForConditionalGeneration.get_decoder": ["GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "class", "def", "get_decoder", "language_model", "return", "self"], "voxtral/modeling_voxtral.py:VoxtralForConditionalGeneration.get_audio_features": ["FloatTensor", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "audio_config", "audio_embeds", "audio_hidden_states", "audio_outputs", "audio_tower", "class", "config", "def", "get_audio_features", "input_features", "intermediate_size", "last_hidden_state", "multi_modal_projector", "reshape", "return", "self", "torch"], "voxtral/modeling_voxtral.py:VoxtralForConditionalGeneration.forward": ["BaseModelOutputWithPast", "Cache", "CausalLMOutputWithPast", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "and", "attention_mask", "audio_embeds", "audio_token_id", "audio_token_mask", "bool", "cache_position", "class", "config", "def", "device", "forward", "get_audio_features", "get_input_embeddings", "if", "input_features", "input_ids", "inputs_embeds", "int", "is", "kwargs", "labels", "language_model", "logits_to_keep", "masked_scatter", "not", "outputs", "past_key_values", "position_ids", "r", "return", "self", "to", "torch", "unsqueeze", "use_cache"], "voxtral/modeling_voxtral.py:VoxtralForConditionalGeneration.prepare_inputs_for_generation": ["False", "GenerationMixin", "ModelForConditionalGeneration", "ModelPreTrainedModel", "None", "True", "args", "class", "def", "get", "if", "input_features", "is_first_iteration", "kwargs", "model_inputs", "not", "or", "pop", "prepare_inputs_for_generation", "return", "self", "super", "use_cache"], "wav2vec2/modeling_wav2vec2.py:_compute_mask_indices": ["False", "LongTensor", "None", "ValueError", "_", "_compute_mask_indices", "and", "append", "arange", "array", "attention_mask", "batch_size", "be", "bigger", "bool", "broadcast_to", "but", "choice", "compute_num_masked_span", "concatenate", "def", "detach", "dtype", "dummy_mask_idx", "else", "epsilon", "f", "float", "for", "got", "has", "if", "in", "input_length", "input_lengths", "int", "int32", "is", "item", "len", "mask_length", "mask_prob", "max", "max_num_masked_span", "min_masks", "ndarray", "not", "np", "num_masked_span", "offsets", "ones", "put_along_axis", "raise", "rand", "random", "range", "replace", "reshape", "return", "sequence_length", "shape", "smaller", "spec_aug_mask", "spec_aug_mask_idx", "spec_aug_mask_idxs", "sum", "than", "to", "tolist", "torch", "tuple", "zeros"], "wav2vec2/modeling_wav2vec2.py:_sample_negative_indices": ["None", "_sample_negative_indices", "arange", "astype", "batch_idx", "batch_size", "bool", "broadcast_to", "def", "dtype", "else", "feature_indices", "features_shape", "for", "high", "if", "in", "int", "int32", "is", "mapped_masked_indices", "mask_time_indices", "ndarray", "not", "np", "num_negatives", "ones", "randint", "random", "range", "return", "sampled_indices", "sampled_negative_indices", "sequence_length", "sequence_length_range", "shape", "size", "sum", "tuple", "zeros"], "wav2vec2/modeling_wav2vec2.py:Wav2Vec2NoLayerNormConvLayer.__init__": ["ACT2FN", "Conv1d", "GradientCheckpointingLayer", "ModelNoLayerNormConvLayer", "__init__", "activation", "bias", "class", "config", "conv", "conv_bias", "conv_dim", "conv_kernel", "conv_stride", "def", "else", "feat_extract_activation", "if", "in_conv_dim", "kernel_size", "layer_id", "nn", "out_conv_dim", "self", "stride", "super"], "wav2vec2/modeling_wav2vec2.py:Wav2Vec2NoLayerNormConvLayer.forward": ["GradientCheckpointingLayer", "ModelNoLayerNormConvLayer", "activation", "class", "conv", "def", "forward", "hidden_states", "return", "self"], "wav2vec2/modeling_wav2vec2.py:Wav2Vec2LayerNormConvLayer.__init__": ["ACT2FN", "Conv1d", "GradientCheckpointingLayer", "LayerNorm", "ModelLayerNormConvLayer", "True", "__init__", "activation", "bias", "class", "config", "conv", "conv_bias", "conv_dim", "conv_kernel", "conv_stride", "def", "elementwise_affine", "else", "feat_extract_activation", "if", "in_conv_dim", "kernel_size", "layer_id", "layer_norm", "nn", "out_conv_dim", "self", "stride", "super"], "wav2vec2/modeling_wav2vec2.py:Wav2Vec2LayerNormConvLayer.forward": ["GradientCheckpointingLayer", "ModelLayerNormConvLayer", "activation", "class", "conv", "def", "forward", "hidden_states", "layer_norm", "return", "self", "transpose"], "wav2vec2/modeling_wav2vec2.py:Wav2Vec2GroupNormConvLayer.__init__": ["ACT2FN", "Conv1d", "GradientCheckpointingLayer", "GroupNorm", "ModelGroupNormConvLayer", "True", "__init__", "activation", "affine", "bias", "class", "config", "conv", "conv_bias", "conv_dim", "conv_kernel", "conv_stride", "def", "else", "feat_extract_activation", "if", "in_conv_dim", "kernel_size", "layer_id", "layer_norm", "nn", "num_channels", "num_groups", "out_conv_dim", "self", "stride", "super"], "wav2vec2/modeling_wav2vec2.py:Wav2Vec2GroupNormConvLayer.forward": ["GradientCheckpointingLayer", "ModelGroupNormConvLayer", "activation", "class", "conv", "def", "forward", "hidden_states", "layer_norm", "return", "self"], "wav2vec2/modeling_wav2vec2.py:Wav2Vec2PositionalConvEmbedding.__init__": ["ACT2FN", "Conv1d", "GatheredParameters", "ModelPositionalConvEmbedding", "ModelSamePadLayer", "Module", "__init__", "activation", "class", "config", "conv", "deepspeed", "def", "dim", "else", "feat_extract_activation", "groups", "hasattr", "hidden_size", "if", "is_deepspeed_zero3_enabled", "kernel_size", "modifier_rank", "name", "nn", "num_conv_pos_embedding_groups", "num_conv_pos_embeddings", "original0", "original1", "padding", "parametrizations", "register_external_parameter", "self", "super", "utils", "weight", "weight_g", "weight_norm", "weight_v", "with", "zero"], "wav2vec2/modeling_wav2vec2.py:Wav2Vec2PositionalConvEmbedding.forward": ["ModelPositionalConvEmbedding", "Module", "activation", "class", "conv", "def", "forward", "hidden_states", "nn", "padding", "return", "self", "transpose"], "wav2vec2/modeling_wav2vec2.py:Wav2Vec2SamePadLayer.__init__": ["ModelSamePadLayer", "Module", "__init__", "class", "def", "else", "if", "nn", "num_conv_pos_embeddings", "num_pad_remove", "self", "super"], "wav2vec2/modeling_wav2vec2.py:Wav2Vec2SamePadLayer.forward": ["ModelSamePadLayer", "Module", "class", "def", "forward", "hidden_states", "if", "nn", "num_pad_remove", "return", "self"], "wav2vec2/modeling_wav2vec2.py:Wav2Vec2FeatureEncoder.__init__": ["False", "ModelFeatureEncoder", "ModelGroupNormConvLayer", "ModelLayerNormConvLayer", "ModelNoLayerNormConvLayer", "Module", "ModuleList", "True", "ValueError", "__init__", "_requires_grad", "be", "but", "class", "config", "conv_layers", "def", "elif", "else", "f", "feat_extract_norm", "for", "gradient_checkpointing", "group", "has", "i", "if", "in", "is", "layer", "layer_id", "nn", "num_feat_extract_layers", "of", "one", "raise", "range", "self", "super", "to"], "wav2vec2/modeling_wav2vec2.py:Wav2Vec2FeatureEncoder._freeze_parameters": ["False", "ModelFeatureEncoder", "Module", "_freeze_parameters", "_requires_grad", "class", "def", "for", "in", "nn", "param", "parameters", "requires_grad", "self"], "wav2vec2/modeling_wav2vec2.py:Wav2Vec2FeatureEncoder.forward": ["ModelFeatureEncoder", "Module", "None", "True", "_requires_grad", "and", "class", "conv_layer", "conv_layers", "def", "for", "forward", "hidden_states", "if", "in", "input_values", "nn", "requires_grad", "return", "self", "training"], "wav2vec2/modeling_wav2vec2.py:Wav2Vec2FeatureProjection.__init__": ["Dropout", "LayerNorm", "Linear", "ModelFeatureProjection", "Module", "__init__", "class", "config", "conv_dim", "def", "dropout", "eps", "feat_proj_dropout", "hidden_size", "layer_norm", "layer_norm_eps", "nn", "projection", "self", "super"], "wav2vec2/modeling_wav2vec2.py:Wav2Vec2FeatureProjection.forward": ["ModelFeatureProjection", "Module", "class", "def", "dropout", "forward", "hidden_states", "layer_norm", "nn", "norm_hidden_states", "projection", "return", "self"], "wav2vec2/modeling_wav2vec2.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "contiguous", "def", "dim", "dropout", "float", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "shape", "size", "softmax", "torch", "training", "transpose", "value"], "wav2vec2/modeling_wav2vec2.py:Wav2Vec2Attention.__init__": ["False", "Linear", "ModelAttention", "ModelConfig", "Module", "None", "True", "ValueError", "__init__", "and", "be", "bias", "bool", "by", "class", "config", "def", "divisible", "dropout", "embed_dim", "f", "float", "got", "head_dim", "if", "int", "is_causal", "is_decoder", "k_proj", "must", "nn", "num_heads", "out_proj", "q_proj", "raise", "scaling", "self", "super", "v_proj"], "wav2vec2/modeling_wav2vec2.py:Wav2Vec2Attention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "False", "FlashAttentionKwargs", "ModelAttention", "Module", "None", "Tensor", "Unpack", "_attn_implementation", "attention_interface", "attention_mask", "attn_output", "attn_weights", "bool", "bsz", "class", "config", "contiguous", "current_states", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "head_dim", "hidden_states", "if", "is", "is_cross_attention", "k_proj", "key_states", "key_value_states", "kv_input_shape", "kwargs", "nn", "not", "out_proj", "output_attentions", "q_input_shape", "q_proj", "query_states", "reshape", "return", "scaling", "self", "shape", "src_len", "tgt_len", "torch", "training", "transpose", "tuple", "v_proj", "value_states", "view"], "wav2vec2/modeling_wav2vec2.py:Wav2Vec2FeedForward.__init__": ["ACT2FN", "Dropout", "Linear", "ModelFeedForward", "Module", "__init__", "activation_dropout", "class", "config", "def", "else", "hidden_act", "hidden_dropout", "hidden_size", "if", "intermediate_act_fn", "intermediate_dense", "intermediate_dropout", "intermediate_size", "isinstance", "nn", "output_dense", "output_dropout", "self", "str", "super"], "wav2vec2/modeling_wav2vec2.py:Wav2Vec2FeedForward.forward": ["ModelFeedForward", "Module", "class", "def", "forward", "hidden_states", "intermediate_act_fn", "intermediate_dense", "intermediate_dropout", "nn", "output_dense", "output_dropout", "return", "self"], "wav2vec2/modeling_wav2vec2.py:Wav2Vec2EncoderLayer.__init__": ["Dropout", "False", "GradientCheckpointingLayer", "LayerNorm", "ModelAttention", "ModelEncoderLayer", "ModelFeedForward", "__init__", "attention", "attention_dropout", "class", "config", "def", "dropout", "embed_dim", "eps", "feed_forward", "final_layer_norm", "hidden_dropout", "hidden_size", "is_decoder", "layer_norm", "layer_norm_eps", "nn", "num_attention_heads", "num_heads", "self", "super"], "wav2vec2/modeling_wav2vec2.py:Wav2Vec2EncoderLayer.forward": ["False", "GradientCheckpointingLayer", "ModelEncoderLayer", "None", "_", "attention", "attention_mask", "attn_residual", "attn_weights", "class", "def", "dropout", "feed_forward", "final_layer_norm", "forward", "hidden_states", "if", "layer_norm", "output_attentions", "outputs", "return", "self"], "wav2vec2/modeling_wav2vec2.py:Wav2Vec2EncoderLayerStableLayerNorm.__init__": ["Dropout", "False", "GradientCheckpointingLayer", "LayerNorm", "ModelAttention", "ModelAttnAdapterLayer", "ModelEncoderLayerStableLayerNorm", "ModelFeedForward", "None", "__init__", "adapter_attn_dim", "adapter_layer", "attention", "attention_dropout", "class", "config", "def", "dropout", "else", "embed_dim", "eps", "feed_forward", "final_layer_norm", "getattr", "hidden_dropout", "hidden_size", "if", "is", "is_decoder", "layer_norm", "layer_norm_eps", "nn", "not", "num_attention_heads", "num_heads", "self", "super"], "wav2vec2/modeling_wav2vec2.py:Wav2Vec2EncoderLayerStableLayerNorm.forward": ["False", "GradientCheckpointingLayer", "ModelEncoderLayerStableLayerNorm", "None", "Tensor", "_", "adapter_layer", "attention", "attention_mask", "attn_residual", "attn_weights", "bool", "class", "def", "dropout", "feed_forward", "final_layer_norm", "forward", "hidden_states", "if", "is", "layer_norm", "not", "output_attentions", "outputs", "return", "self", "torch"], "wav2vec2/modeling_wav2vec2.py:Wav2Vec2Encoder.__init__": ["Dropout", "False", "LayerNorm", "ModelEncoder", "ModelEncoderLayer", "ModelPositionalConvEmbedding", "Module", "ModuleList", "_", "__init__", "class", "config", "def", "dropout", "eps", "for", "gradient_checkpointing", "hidden_dropout", "hidden_size", "in", "layer_norm", "layer_norm_eps", "layers", "nn", "num_hidden_layers", "pos_conv_embed", "range", "self", "super"], "wav2vec2/modeling_wav2vec2.py:Wav2Vec2Encoder.forward": ["BaseModelOutput", "False", "ModelEncoder", "Module", "None", "Tensor", "True", "all_hidden_states", "all_self_attentions", "and", "attention_mask", "attentions", "bool", "class", "config", "create_bidirectional_mask", "def", "dropout", "dropout_probability", "else", "expand_attention_mask", "for", "forward", "hidden_states", "if", "in", "input_embeds", "is", "is_deepspeed_zero3_enabled", "is_fsdp_managed_module", "last_hidden_state", "layer", "layer_norm", "layer_outputs", "layerdrop", "layers", "nn", "not", "or", "output_attentions", "output_hidden_states", "pos_conv_embed", "position_embeddings", "rand", "repeat", "return", "return_dict", "self", "shape", "skip_the_layer", "synced_gpus", "tensor", "torch", "training", "tuple", "unsqueeze", "v"], "wav2vec2/modeling_wav2vec2.py:Wav2Vec2EncoderStableLayerNorm.__init__": ["Dropout", "False", "LayerNorm", "ModelEncoderLayerStableLayerNorm", "ModelEncoderStableLayerNorm", "ModelPositionalConvEmbedding", "Module", "ModuleList", "_", "__init__", "class", "config", "def", "dropout", "eps", "for", "gradient_checkpointing", "hidden_dropout", "hidden_size", "in", "layer_norm", "layer_norm_eps", "layers", "nn", "num_hidden_layers", "pos_conv_embed", "range", "self", "super"], "wav2vec2/modeling_wav2vec2.py:Wav2Vec2EncoderStableLayerNorm.forward": ["BaseModelOutput", "False", "ModelEncoderStableLayerNorm", "Module", "None", "True", "all_hidden_states", "all_self_attentions", "and", "attention_mask", "attentions", "class", "config", "create_bidirectional_mask", "def", "dropout", "dropout_probability", "else", "expand_attention_mask", "for", "forward", "hidden_states", "if", "in", "input_embeds", "is", "is_deepspeed_zero3_enabled", "is_fsdp_managed_module", "last_hidden_state", "layer", "layer_norm", "layer_outputs", "layerdrop", "layers", "nn", "not", "or", "output_attentions", "output_hidden_states", "pos_conv_embed", "position_embeddings", "rand", "repeat", "return", "return_dict", "self", "shape", "skip_the_layer", "synced_gpus", "torch", "training", "tuple", "unsqueeze", "v"], "wav2vec2/modeling_wav2vec2.py:Wav2Vec2GumbelVectorQuantizer.__init__": ["FloatTensor", "Linear", "ModelGumbelVectorQuantizer", "Module", "Parameter", "ValueError", "__init__", "be", "by", "class", "codevector_dim", "codevectors", "concatenation", "config", "conv_dim", "def", "divisible", "f", "for", "if", "must", "nn", "num_codevector_groups", "num_codevectors_per_group", "num_groups", "num_vars", "raise", "self", "super", "temperature", "torch", "weight_proj"], "wav2vec2/modeling_wav2vec2.py:Wav2Vec2GumbelVectorQuantizer._compute_perplexity": ["ModelGumbelVectorQuantizer", "Module", "None", "_compute_perplexity", "class", "def", "dim", "else", "exp", "expand", "flatten", "if", "is", "log", "marginal_probs", "mask", "mask_extended", "mean", "nn", "not", "perplexity", "probs", "return", "shape", "sum", "torch", "where", "zeros_like"], "wav2vec2/modeling_wav2vec2.py:Wav2Vec2GumbelVectorQuantizer.forward": ["ModelGumbelVectorQuantizer", "Module", "None", "True", "_compute_perplexity", "argmax", "batch_size", "class", "codevector_idx", "codevector_probs", "codevector_soft_dist", "codevectors", "codevectors_per_group", "def", "dim", "else", "float", "forward", "functional", "gumbel_softmax", "hard", "hidden_size", "hidden_states", "if", "mask_time_indices", "new_zeros", "nn", "num_groups", "num_vars", "perplexity", "return", "scatter_", "self", "sequence_length", "shape", "softmax", "sum", "tau", "temperature", "torch", "training", "type_as", "unsqueeze", "view", "weight_proj"], "wav2vec2/modeling_wav2vec2.py:Wav2Vec2Adapter.__init__": ["LayerNorm", "Linear", "ModelAdapter", "ModelAdapterLayer", "Module", "ModuleList", "None", "_", "__init__", "class", "config", "def", "else", "for", "hidden_size", "if", "in", "layerdrop", "layers", "nn", "num_adapter_layers", "output_hidden_size", "proj", "proj_layer_norm", "range", "self", "super"], "wav2vec2/modeling_wav2vec2.py:Wav2Vec2Adapter.forward": ["ModelAdapter", "Module", "None", "and", "class", "def", "for", "forward", "hidden_states", "if", "in", "is", "layer", "layerdrop", "layerdrop_prob", "layers", "nn", "not", "np", "or", "proj", "proj_layer_norm", "random", "return", "self", "training", "transpose"], "wav2vec2/modeling_wav2vec2.py:Wav2Vec2AdapterLayer.__init__": ["Conv1d", "ModelAdapterLayer", "Module", "__init__", "adapter_kernel_size", "adapter_stride", "class", "config", "conv", "def", "nn", "output_hidden_size", "padding", "self", "stride", "super"], "wav2vec2/modeling_wav2vec2.py:Wav2Vec2AdapterLayer.forward": ["ModelAdapterLayer", "Module", "class", "conv", "def", "dim", "forward", "functional", "glu", "hidden_states", "nn", "return", "self"], "wav2vec2/modeling_wav2vec2.py:Wav2Vec2AttnAdapterLayer.__init__": ["LayerNorm", "Linear", "ModelAttnAdapterLayer", "Module", "ReLU", "__init__", "act_fn", "adapter_attn_dim", "class", "config", "def", "hidden_dim", "hidden_size", "input_dim", "linear_1", "linear_2", "nn", "norm", "self", "super"], "wav2vec2/modeling_wav2vec2.py:Wav2Vec2AttnAdapterLayer.forward": ["FloatTensor", "ModelAttnAdapterLayer", "Module", "act_fn", "class", "def", "forward", "hidden_states", "linear_1", "linear_2", "nn", "norm", "return", "self", "torch"], "wav2vec2/modeling_wav2vec2.py:Wav2Vec2PreTrainedModel._init_weights": ["Conv1d", "GroupNorm", "LayerNorm", "Linear", "ModelFeatureProjection", "ModelForPreTraining", "ModelGumbelVectorQuantizer", "ModelPositionalConvEmbedding", "ModelPreTrainedModel", "None", "PreTrainedModel", "_init_weights", "a", "b", "bias", "class", "codevectors", "config", "constant_", "conv", "def", "elif", "groups", "if", "in_channels", "in_features", "init", "initializer_range", "is", "isinstance", "k", "kaiming_normal_", "kernel_size", "math", "mean", "module", "nn", "normal_", "not", "ones_", "project_hid", "project_q", "projection", "reset_parameters", "self", "sqrt", "std", "uniform_", "weight", "weight_proj", "zeros_"], "wav2vec2/modeling_wav2vec2.py:Wav2Vec2PreTrainedModel._get_feat_extract_output_lengths": ["LongTensor", "ModelPreTrainedModel", "None", "PreTrainedModel", "_", "_conv_out_length", "_get_feat_extract_output_lengths", "adapter_stride", "add_adapter", "bool", "class", "config", "conv_kernel", "conv_stride", "def", "div", "else", "floor", "for", "if", "in", "input_length", "input_lengths", "int", "is", "kernel_size", "num_adapter_layers", "range", "return", "rounding_mode", "self", "stride", "torch", "zip"], "wav2vec2/modeling_wav2vec2.py:Wav2Vec2PreTrainedModel._get_feature_vector_attention_mask": ["LongTensor", "ModelPreTrainedModel", "None", "PreTrainedModel", "_get_feat_extract_output_lengths", "_get_feature_vector_attention_mask", "add_adapter", "arange", "attention_mask", "batch_size", "bool", "class", "cumsum", "def", "device", "dim", "dtype", "feature_vector_length", "flip", "int", "long", "non_padded_lengths", "output_lengths", "return", "self", "shape", "to", "torch", "zeros"], "wav2vec2/modeling_wav2vec2.py:Wav2Vec2PreTrainedModel._get_adapters": ["Make", "ModelAttnAdapterLayer", "ModelForCTC", "ModelPreTrainedModel", "None", "PreTrainedModel", "ValueError", "__class__", "_get_adapters", "adapter", "adapter_attn_dim", "adapter_weights", "class", "config", "def", "define", "f", "for", "has", "if", "in", "is", "isinstance", "join", "layers", "lm_head", "module", "name", "named_modules", "named_parameters", "no", "param", "param_name", "raise", "return", "self", "sure", "to"], "wav2vec2/modeling_wav2vec2.py:Wav2Vec2PreTrainedModel.init_adapter_layers": ["ModelAttnAdapterLayer", "ModelForCTC", "ModelPreTrainedModel", "PreTrainedModel", "_init_weights", "class", "def", "for", "if", "in", "init_adapter_layers", "isinstance", "lm_head", "module", "modules", "self"], "wav2vec2/modeling_wav2vec2.py:Wav2Vec2PreTrainedModel.load_adapter": ["Adapter", "Can", "Cannot", "Exception", "False", "If", "Linear", "ModelPreTrainedModel", "Model_ADAPTER_PT_FILE", "Model_ADAPTER_SAFE_FILE", "None", "OSError", "Otherwise", "PreTrainedModel", "The", "True", "ValueError", "_get_adapters", "_name_or_path", "a", "adapter", "adapter_attn_dim", "adapter_weights", "already", "and", "are", "cache_dir", "cached_file", "check_torch_load_is_safe", "class", "co", "config", "containing", "correct", "cpu", "def", "defined", "device", "directory", "don", "dtype", "elif", "except", "f", "file", "filename", "filepath", "for", "force_download", "force_load", "format", "from", "has", "have", "https", "huggingface", "if", "in", "is", "it", "items", "join", "k", "keys", "kwargs", "len", "lm_head", "load", "load_adapter", "load_state_dict", "local", "local_files_only", "logger", "make", "map_location", "missing", "missing_keys", "model", "model_path_or_id", "models", "name", "named", "nn", "not", "output_hidden_size", "path", "pop", "proxies", "r", "raise", "return", "revision", "safe_load_file", "same", "self", "set", "shape", "state_dict", "str", "strict", "sure", "t", "target_lang", "target_vocab_size", "the", "to", "token", "torch", "try", "trying", "unexpected", "unexpected_keys", "use_safetensors", "v", "vocab_size", "warning", "weight", "weight_path", "weights", "weights_only", "were", "with", "you"], "wav2vec2/modeling_wav2vec2.py:Wav2Vec2Model.__init__": ["ModelAdapter", "ModelConfig", "ModelEncoder", "ModelEncoderStableLayerNorm", "ModelFeatureEncoder", "ModelFeatureProjection", "ModelModel", "ModelPreTrainedModel", "None", "Parameter", "Tensor", "__init__", "adapter", "add_adapter", "class", "config", "def", "do_stable_layer_norm", "else", "encoder", "feature_extractor", "feature_projection", "hidden_size", "if", "mask_feature_prob", "mask_time_prob", "masked_spec_embed", "nn", "or", "post_init", "self", "super", "torch", "uniform_"], "wav2vec2/modeling_wav2vec2.py:Wav2Vec2Model.freeze_feature_encoder": ["ModelModel", "ModelPreTrainedModel", "_freeze_parameters", "class", "def", "feature_extractor", "freeze_feature_encoder", "self"], "wav2vec2/modeling_wav2vec2.py:Wav2Vec2Model._mask_hidden_states": ["FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "True", "_compute_mask_indices", "_mask_hidden_states", "and", "apply_spec_augment", "attention_mask", "batch_size", "bool", "class", "config", "def", "device", "dtype", "elif", "expand", "getattr", "hidden_size", "hidden_states", "if", "is", "mask_feature_indices", "mask_feature_length", "mask_feature_min_masks", "mask_feature_prob", "mask_length", "mask_prob", "mask_time_indices", "mask_time_length", "mask_time_min_masks", "mask_time_prob", "masked_spec_embed", "min_masks", "not", "return", "self", "sequence_length", "size", "tensor", "to", "torch", "training"], "wav2vec2/modeling_wav2vec2.py:Wav2Vec2Model.forward": ["False", "FloatTensor", "ModelBaseModelOutput", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "_get_feature_vector_attention_mask", "_mask_hidden_states", "adapter", "add_adapter", "attention_mask", "attentions", "bool", "class", "config", "def", "else", "encoder", "encoder_outputs", "extract_features", "feature_extractor", "feature_projection", "forward", "hidden_states", "if", "input_values", "is", "kwargs", "last_hidden_state", "mask_time_indices", "not", "output_attentions", "output_hidden_states", "r", "return", "return_dict", "self", "shape", "torch", "transpose", "tuple", "use_return_dict"], "wav2vec2/modeling_wav2vec2.py:Wav2Vec2ForPreTraining.__init__": ["Dropout", "Linear", "Model", "ModelConfig", "ModelForPreTraining", "ModelGumbelVectorQuantizer", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "codevector_dim", "config", "def", "dropout_features", "feat_quantizer_dropout", "hidden_size", "nn", "post_init", "proj_codevector_dim", "project_hid", "project_q", "quantizer", "self", "super"], "wav2vec2/modeling_wav2vec2.py:Wav2Vec2ForPreTraining.set_gumbel_temperature": ["ModelForPreTraining", "ModelPreTrainedModel", "class", "def", "int", "quantizer", "self", "set_gumbel_temperature", "temperature"], "wav2vec2/modeling_wav2vec2.py:Wav2Vec2ForPreTraining.freeze_feature_encoder": ["Model", "ModelForPreTraining", "ModelPreTrainedModel", "_freeze_parameters", "class", "def", "feature_extractor", "freeze_feature_encoder", "self"], "wav2vec2/modeling_wav2vec2.py:Wav2Vec2ForPreTraining.compute_contrastive_logits": ["FloatTensor", "ModelForPreTraining", "ModelPreTrainedModel", "cat", "class", "compute_contrastive_logits", "cosine_similarity", "def", "dim", "float", "logits", "negative_features", "predicted_features", "return", "target_features", "temperature", "torch", "type_as"], "wav2vec2/modeling_wav2vec2.py:Wav2Vec2ForPreTraining.forward": ["BoolTensor", "False", "Model", "ModelForPreTraining", "ModelForPreTrainingOutput", "ModelPreTrainedModel", "None", "Tensor", "_get_feature_vector_attention_mask", "add_adapter", "all", "any", "attention_mask", "attentions", "batch_size", "bool", "class", "codevector_perplexity", "compute_contrastive_logits", "config", "contrastive_logits_temperature", "contrastive_loss", "cross_entropy", "def", "diversity_loss", "diversity_loss_weight", "dropout_features", "dtype", "else", "extract_features", "flatten", "float", "forward", "functional", "hidden_size", "hidden_states", "if", "inf", "input_values", "is", "kwargs", "logits", "long", "loss", "mask_time_indices", "neg_is_pos", "negative_quantized_features", "nn", "not", "num_codevector_groups", "num_codevectors", "num_codevectors_per_group", "output_attentions", "output_hidden_states", "outputs", "permute", "project_hid", "project_q", "projected_quantized_states", "projected_states", "quantized_features", "quantizer", "r", "reduction", "reshape", "return", "return_dict", "sampled_negative_indices", "self", "sequence_length", "shape", "size", "sum", "target", "to", "torch", "transformer_features", "transpose", "tuple", "use_return_dict", "view", "weight"], "wav2vec2/modeling_wav2vec2.py:Wav2Vec2ForCTC.__init__": ["Dropout", "Linear", "Model", "ModelForCTC", "ModelModel", "ModelPreTrainedModel", "None", "Please", "ValueError", "You", "__class__", "__init__", "a", "add_adapter", "and", "are", "as", "class", "config", "configuration", "def", "define", "does", "dropout", "else", "f", "final_dropout", "follows", "from_pretrained", "hasattr", "head", "hidden_size", "if", "instantiate", "is", "language", "lm_head", "model", "nn", "not", "of", "or", "output_hidden_size", "post_init", "r", "raise", "s", "self", "size", "str", "super", "target_lang", "that", "the", "to", "trying", "vocab_size", "vocabulary", "with", "your"], "wav2vec2/modeling_wav2vec2.py:Wav2Vec2ForCTC.tie_weights": ["By", "Cannot", "ModelForCTC", "ModelPreTrainedModel", "None", "True", "ValueError", "adapter_attn_dim", "and", "class", "config", "def", "default", "defined", "elif", "eng", "f", "force_load", "getattr", "if", "info", "is", "kwargs", "load_adapter", "logger", "not", "pass", "raise", "self", "set", "target_lang", "tie_weights", "to"], "wav2vec2/modeling_wav2vec2.py:Wav2Vec2ForCTC.freeze_feature_encoder": ["Model", "ModelForCTC", "ModelPreTrainedModel", "_freeze_parameters", "class", "def", "feature_extractor", "freeze_feature_encoder", "self"], "wav2vec2/modeling_wav2vec2.py:Wav2Vec2ForCTC.freeze_base_model": ["False", "Model", "ModelForCTC", "ModelPreTrainedModel", "class", "def", "for", "freeze_base_model", "in", "param", "parameters", "requires_grad", "self"], "wav2vec2/modeling_wav2vec2.py:Wav2Vec2ForCTC.forward": ["CausalLMOutput", "False", "Label", "Model", "ModelForCTC", "ModelPreTrainedModel", "None", "Tensor", "ValueError", "_HIDDEN_STATES_START_POSITION", "_get_feat_extract_output_lengths", "and", "attention_mask", "attentions", "backends", "be", "blank", "bool", "class", "config", "ctc_loss", "ctc_loss_reduction", "ctc_zero_infinity", "cudnn", "def", "dim", "dropout", "dtype", "else", "enabled", "f", "flags", "flattened_targets", "float32", "forward", "functional", "hidden_states", "if", "input_lengths", "input_values", "is", "kwargs", "labels", "labels_mask", "lm_head", "log_probs", "log_softmax", "logits", "long", "loss", "masked_select", "max", "must", "nn", "not", "ones_like", "output", "output_attentions", "output_hidden_states", "outputs", "pad_token_id", "r", "raise", "reduction", "return", "return_dict", "self", "sum", "target_lengths", "to", "torch", "transpose", "tuple", "use_return_dict", "values", "vocab_size", "with", "zero_infinity"], "wav2vec2/modeling_wav2vec2.py:Wav2Vec2ForSequenceClassification.__init__": ["Linear", "Model", "ModelForSequenceClassification", "ModelModel", "ModelPreTrainedModel", "Parameter", "Sequence", "True", "ValueError", "__init__", "adapters", "add_adapter", "and", "class", "classification", "classifier", "classifier_proj_size", "config", "def", "does", "hasattr", "hidden_size", "if", "layer_weights", "nn", "not", "num_hidden_layers", "num_labels", "num_layers", "of", "ones", "post_init", "projector", "raise", "self", "super", "support", "the", "torch", "use", "use_weighted_layer_sum"], "wav2vec2/modeling_wav2vec2.py:Wav2Vec2ForSequenceClassification.freeze_feature_encoder": ["Model", "ModelForSequenceClassification", "ModelPreTrainedModel", "_freeze_parameters", "class", "def", "feature_extractor", "freeze_feature_encoder", "self"], "wav2vec2/modeling_wav2vec2.py:Wav2Vec2ForSequenceClassification.freeze_base_model": ["False", "Model", "ModelForSequenceClassification", "ModelPreTrainedModel", "class", "def", "for", "freeze_base_model", "in", "param", "parameters", "requires_grad", "self"], "wav2vec2/modeling_wav2vec2.py:Wav2Vec2ForSequenceClassification.forward": ["CrossEntropyLoss", "Model", "ModelForSequenceClassification", "ModelPreTrainedModel", "None", "SequenceClassifierOutput", "Tensor", "True", "_HIDDEN_STATES_START_POSITION", "_get_feature_vector_attention_mask", "attention_mask", "attentions", "bool", "class", "classifier", "config", "def", "dim", "else", "expand_padding_mask", "forward", "functional", "hidden_states", "if", "input_values", "is", "kwargs", "labels", "layer_weights", "logits", "loss", "loss_fct", "mean", "nn", "norm_weights", "not", "num_labels", "output", "output_attentions", "output_hidden_states", "outputs", "padding_mask", "pooled_output", "projector", "r", "repeat", "return", "return_dict", "self", "shape", "softmax", "stack", "sum", "torch", "tuple", "unsqueeze", "use_return_dict", "use_weighted_layer_sum", "view"], "wav2vec2/modeling_wav2vec2.py:Wav2Vec2ForAudioFrameClassification.__init__": ["Audio", "Linear", "Model", "ModelForAudioFrameClassification", "ModelModel", "ModelPreTrainedModel", "Parameter", "True", "ValueError", "__init__", "adapters", "add_adapter", "and", "class", "classification", "classifier", "config", "def", "does", "frame", "hasattr", "hidden_size", "if", "layer_weights", "nn", "not", "num_hidden_layers", "num_labels", "num_layers", "of", "ones", "post_init", "raise", "self", "super", "support", "the", "torch", "use", "use_weighted_layer_sum"], "wav2vec2/modeling_wav2vec2.py:Wav2Vec2ForAudioFrameClassification.freeze_feature_encoder": ["Model", "ModelForAudioFrameClassification", "ModelPreTrainedModel", "_freeze_parameters", "class", "def", "feature_extractor", "freeze_feature_encoder", "self"], "wav2vec2/modeling_wav2vec2.py:Wav2Vec2ForAudioFrameClassification.freeze_base_model": ["False", "Model", "ModelForAudioFrameClassification", "ModelPreTrainedModel", "class", "def", "for", "freeze_base_model", "in", "param", "parameters", "requires_grad", "self"], "wav2vec2/modeling_wav2vec2.py:Wav2Vec2ForAudioFrameClassification.forward": ["CrossEntropyLoss", "Model", "ModelForAudioFrameClassification", "ModelPreTrainedModel", "None", "Tensor", "TokenClassifierOutput", "True", "_HIDDEN_STATES_START_POSITION", "argmax", "attention_mask", "attentions", "axis", "bool", "class", "classifier", "config", "def", "dim", "else", "forward", "functional", "hidden_states", "if", "input_values", "is", "kwargs", "labels", "layer_weights", "logits", "loss", "loss_fct", "nn", "norm_weights", "not", "num_labels", "output", "output_attentions", "output_hidden_states", "outputs", "r", "return", "return_dict", "self", "softmax", "stack", "sum", "torch", "tuple", "use_return_dict", "use_weighted_layer_sum", "view"], "wav2vec2/modeling_wav2vec2.py:AMSoftmaxLoss.__init__": ["CrossEntropyLoss", "Model", "Module", "Parameter", "True", "__init__", "class", "def", "input_dim", "loss", "margin", "nn", "num_labels", "randn", "requires_grad", "scale", "self", "super", "torch", "weight"], "wav2vec2/modeling_wav2vec2.py:AMSoftmaxLoss.forward": ["Model", "Module", "bool", "class", "cos_theta", "def", "dim", "flatten", "forward", "functional", "hidden_states", "labels", "logits", "loss", "margin", "mm", "nn", "normalize", "num_labels", "one_hot", "onehot", "psi", "return", "scale", "self", "torch", "weight", "where"], "wav2vec2/modeling_wav2vec2.py:TDNNLayer.__init__": ["Linear", "Model", "Module", "ReLU", "__init__", "activation", "class", "config", "def", "dilation", "else", "if", "in_conv_dim", "kernel", "kernel_size", "layer_id", "nn", "out_conv_dim", "self", "super", "tdnn_dilation", "tdnn_dim", "tdnn_kernel"], "wav2vec2/modeling_wav2vec2.py:TDNNLayer.forward": ["Detected", "LoRA", "LoraLayer", "Model", "Module", "Tensor", "You", "activation", "applied", "be", "bias", "class", "conv1d", "def", "dilation", "due", "exclude", "forward", "from", "functional", "hidden_states", "if", "in_conv_dim", "is_peft_available", "isinstance", "kernel", "kernel_size", "modules", "nn", "on", "optimization", "out_conv_dim", "return", "s", "self", "should", "t", "target", "to", "torch", "transpose", "view", "warn", "warnings", "weight", "weights", "won"], "wav2vec2/modeling_wav2vec2.py:Wav2Vec2ForXVector.__init__": ["AMSoftmaxLoss", "Linear", "Model", "ModelForXVector", "ModelModel", "ModelPreTrainedModel", "ModuleList", "Parameter", "TDNNLayer", "__init__", "class", "classifier", "config", "def", "feature_extractor", "for", "hidden_size", "i", "if", "in", "layer_weights", "len", "nn", "num_hidden_layers", "num_labels", "num_layers", "objective", "ones", "post_init", "projector", "range", "self", "super", "tdnn", "tdnn_dim", "tdnn_layers", "torch", "use_weighted_layer_sum", "xvector_output_dim"], "wav2vec2/modeling_wav2vec2.py:Wav2Vec2ForXVector.freeze_feature_encoder": ["Model", "ModelForXVector", "ModelPreTrainedModel", "_freeze_parameters", "class", "def", "feature_extractor", "freeze_feature_encoder", "self"], "wav2vec2/modeling_wav2vec2.py:Wav2Vec2ForXVector.freeze_base_model": ["False", "Model", "ModelForXVector", "ModelPreTrainedModel", "class", "def", "for", "freeze_base_model", "in", "param", "parameters", "requires_grad", "self"], "wav2vec2/modeling_wav2vec2.py:Wav2Vec2ForXVector._get_tdnn_output_lengths": ["LongTensor", "ModelForXVector", "ModelPreTrainedModel", "_conv_out_length", "_get_tdnn_output_lengths", "class", "config", "def", "for", "in", "input_length", "input_lengths", "int", "kernel_size", "return", "self", "stride", "tdnn_kernel", "torch"], "wav2vec2/modeling_wav2vec2.py:Wav2Vec2ForXVector.forward": ["Model", "ModelForXVector", "ModelPreTrainedModel", "None", "Tensor", "True", "XVectorOutput", "_HIDDEN_STATES_START_POSITION", "_get_feat_extract_output_lengths", "_get_tdnn_output_lengths", "append", "attention_mask", "attentions", "bool", "cat", "class", "classifier", "config", "def", "dim", "else", "embeddings", "enumerate", "feat_extract_output_lengths", "feature_extractor", "for", "forward", "functional", "hidden_states", "i", "if", "in", "input_values", "is", "kwargs", "labels", "layer_weights", "length", "logits", "loss", "mean", "mean_features", "nn", "norm_weights", "not", "objective", "output", "output_attentions", "output_embeddings", "output_hidden_states", "outputs", "projector", "r", "return", "return_dict", "self", "softmax", "stack", "statistic_pooling", "std", "std_features", "sum", "tdnn", "tdnn_layer", "tdnn_output_lengths", "torch", "tuple", "use_return_dict", "use_weighted_layer_sum", "view"], "wav2vec2_bert/modeling_wav2vec2_bert.py:Wav2Vec2BertRotaryPositionalEmbedding.__init__": ["False", "ModelRotaryPositionalEmbedding", "Module", "None", "__init__", "arange", "base", "cached_rotary_positional_embedding", "cached_sequence_length", "class", "config", "def", "dim", "dtype", "float", "hidden_size", "int64", "inv_freq", "nn", "num_attention_heads", "persistent", "register_buffer", "rotary_embedding_base", "self", "super", "torch"], "wav2vec2_bert/modeling_wav2vec2_bert.py:Wav2Vec2BertRotaryPositionalEmbedding.forward": ["ModelRotaryPositionalEmbedding", "Module", "None", "and", "arange", "cached_rotary_positional_embedding", "cached_sequence_length", "cat", "class", "cos", "cos_embeddings", "def", "dim", "einsum", "embeddings", "forward", "freqs", "hidden_states", "i", "if", "ij", "inv_freq", "is", "j", "nn", "not", "return", "self", "sequence_length", "shape", "sin", "sin_embeddings", "stack", "time_stamps", "torch", "type_as"], "wav2vec2_bert/modeling_wav2vec2_bert.py:Wav2Vec2BertRelPositionalEmbedding.__init__": ["False", "ModelRelPositionalEmbedding", "Module", "__init__", "class", "config", "d_model", "def", "expand", "extend_pe", "hidden_size", "max_len", "max_source_positions", "nn", "pe", "persistent", "register_buffer", "self", "super", "tensor", "torch"], "wav2vec2_bert/modeling_wav2vec2_bert.py:Wav2Vec2BertRelPositionalEmbedding.extend_pe": ["ModelRelPositionalEmbedding", "Module", "None", "arange", "cat", "class", "cos", "d_model", "def", "device", "dim", "div_term", "dtype", "exp", "extend_pe", "flip", "float", "if", "int64", "is", "log", "math", "nn", "not", "or", "pe", "pe_negative", "pe_positive", "position", "return", "self", "sin", "size", "to", "torch", "unsqueeze", "x", "zeros"], "wav2vec2_bert/modeling_wav2vec2_bert.py:Wav2Vec2BertRelPositionalEmbedding.forward": ["ModelRelPositionalEmbedding", "Module", "Tensor", "class", "def", "end_idx", "extend_pe", "forward", "hidden_states", "nn", "pe", "relative_position_embeddings", "return", "self", "size", "start_idx", "torch"], "wav2vec2_bert/modeling_wav2vec2_bert.py:Wav2Vec2BertFeatureProjection.__init__": ["Dropout", "LayerNorm", "Linear", "ModelFeatureProjection", "Module", "__init__", "class", "config", "def", "dropout", "eps", "feat_proj_dropout", "feature_projection_input_dim", "hidden_size", "layer_norm", "layer_norm_eps", "nn", "projection", "self", "super"], "wav2vec2_bert/modeling_wav2vec2_bert.py:Wav2Vec2BertFeatureProjection.forward": ["ModelFeatureProjection", "Module", "class", "def", "dropout", "forward", "hidden_states", "layer_norm", "nn", "norm_hidden_states", "projection", "return", "self"], "wav2vec2_bert/modeling_wav2vec2_bert.py:Wav2Vec2BertFeedForward.__init__": ["ACT2FN", "Dropout", "Linear", "ModelFeedForward", "Module", "None", "__init__", "act_fn", "activation_dropout", "class", "config", "def", "else", "hidden_act", "hidden_dropout", "hidden_size", "if", "intermediate_act_fn", "intermediate_dense", "intermediate_dropout", "intermediate_size", "is", "isinstance", "nn", "not", "output_dense", "output_dropout", "self", "str", "super"], "wav2vec2_bert/modeling_wav2vec2_bert.py:Wav2Vec2BertFeedForward.forward": ["ModelFeedForward", "Module", "class", "def", "forward", "hidden_states", "intermediate_act_fn", "intermediate_dense", "intermediate_dropout", "nn", "output_dense", "output_dropout", "return", "self"], "wav2vec2_bert/modeling_wav2vec2_bert.py:Wav2Vec2BertConvolutionModule.__init__": ["ACT2FN", "Conv1d", "Dropout", "False", "GLU", "LayerNorm", "ModelConvolutionModule", "Module", "SAME", "ValueError", "__init__", "a", "activation", "be", "bias", "class", "config", "conformer_conv_dropout", "conv_depthwise_kernel_size", "def", "depthwise_conv", "depthwise_layer_norm", "dim", "dropout", "eps", "for", "glu", "groups", "hidden_act", "hidden_size", "if", "kernel_size", "layer_norm", "layer_norm_eps", "nn", "number", "odd", "padding", "pointwise_conv1", "pointwise_conv2", "raise", "self", "should", "stride", "super"], "wav2vec2_bert/modeling_wav2vec2_bert.py:Wav2Vec2BertConvolutionModule.forward": ["ModelConvolutionModule", "Module", "None", "activation", "attention_mask", "bool", "class", "def", "depthwise_conv", "depthwise_layer_norm", "dropout", "forward", "functional", "glu", "hidden_states", "if", "is", "kernel_size", "layer_norm", "masked_fill", "nn", "not", "pad", "pointwise_conv1", "pointwise_conv2", "return", "self", "torch", "transpose", "unsqueeze"], "wav2vec2_bert/modeling_wav2vec2_bert.py:Wav2Vec2BertSelfAttention.__init__": ["Dropout", "Embedding", "False", "Linear", "ModelSelfAttention", "Module", "None", "Parameter", "__init__", "attention_dropout", "bias", "class", "config", "def", "distance_embedding", "dropout", "else", "head_size", "hidden_size", "if", "is_adapter_attention", "left_max_position_embeddings", "linear_k", "linear_out", "linear_pos", "linear_q", "linear_v", "nn", "not", "num_attention_heads", "num_heads", "num_positions", "output_hidden_size", "p", "pos_bias_u", "pos_bias_v", "position_embeddings_type", "relative", "relative_key", "right_max_position_embeddings", "self", "super", "torch", "zeros"], "wav2vec2_bert/modeling_wav2vec2_bert.py:Wav2Vec2BertSelfAttention.forward": ["False", "ModelSelfAttention", "Module", "None", "Tensor", "ValueError", "_apply_relative_embeddings", "_apply_rotary_embedding", "arange", "attention_mask", "batch_size", "be", "bhld", "bhlr", "bool", "clamp", "class", "def", "defined", "device", "dim", "distance", "distance_embedding", "dropout", "dtype", "einsum", "else", "forward", "has", "head_size", "hidden_size", "hidden_states", "if", "is", "key", "key_length", "left_max_position_embeddings", "linear_k", "linear_out", "linear_q", "linear_v", "long", "lrd", "math", "matmul", "nn", "not", "num_heads", "output_attentions", "position_embeddings_type", "position_ids_l", "position_ids_r", "positional_embedding", "probs", "query", "query_key_states", "query_length", "raise", "relative", "relative_key", "relative_position_attn_weights", "relative_position_embeddings", "reshape", "return", "right_max_position_embeddings", "rotary", "scores", "self", "sequence_length", "shape", "size", "softmax", "sqrt", "to", "torch", "transpose", "tuple", "value", "value_states", "view", "when"], "wav2vec2_bert/modeling_wav2vec2_bert.py:Wav2Vec2BertSelfAttention._apply_rotary_embedding": ["ModelSelfAttention", "Module", "_apply_rotary_embedding", "batch_size", "cat", "class", "cos", "def", "dim", "head_size", "hidden_size", "hidden_states", "ndim", "nn", "num_heads", "relative_position_embeddings", "return", "rotated_states", "rotated_states_begin", "rotated_states_end", "self", "sequence_length", "sin", "size", "torch", "transpose", "view"], "wav2vec2_bert/modeling_wav2vec2_bert.py:Wav2Vec2BertSelfAttention._apply_relative_embeddings": ["ModelSelfAttention", "Module", "_apply_relative_embeddings", "cat", "class", "def", "device", "dim", "dtype", "head_size", "key", "linear_pos", "math", "matmul", "nn", "num_heads", "pos_bias_u", "pos_bias_v", "proj_relative_position_embeddings", "q_with_bias_u", "q_with_bias_v", "query", "relative_position_embeddings", "return", "scores", "scores_ac", "scores_bd", "scores_bd_padded", "scores_bd_padded_shape", "self", "shape", "size", "sqrt", "torch", "transpose", "view", "view_as", "zero_pad", "zeros"], "wav2vec2_bert/modeling_wav2vec2_bert.py:Wav2Vec2BertEncoderLayer.__init__": ["Dropout", "GradientCheckpointingLayer", "LayerNorm", "ModelConvolutionModule", "ModelEncoderLayer", "ModelFeedForward", "ModelSelfAttention", "__init__", "attention_dropout", "class", "config", "conv_module", "def", "dropout", "embed_dim", "eps", "ffn1", "ffn1_layer_norm", "ffn2", "ffn2_layer_norm", "final_layer_norm", "hidden_size", "layer_norm_eps", "nn", "self", "self_attn", "self_attn_dropout", "self_attn_layer_norm", "super"], "wav2vec2_bert/modeling_wav2vec2_bert.py:Wav2Vec2BertEncoderLayer.forward": ["False", "GradientCheckpointingLayer", "ModelEncoderLayer", "None", "Tensor", "attention_mask", "attn_weigts", "bool", "class", "conv_attention_mask", "conv_module", "def", "ffn1", "ffn1_layer_norm", "ffn2", "ffn2_layer_norm", "final_layer_norm", "forward", "hidden_states", "output_attentions", "relative_position_embeddings", "residual", "return", "self", "self_attn", "self_attn_dropout", "self_attn_layer_norm", "torch"], "wav2vec2_bert/modeling_wav2vec2_bert.py:Wav2Vec2BertEncoder.__init__": ["Dropout", "False", "ModelEncoder", "ModelEncoderLayer", "ModelRelPositionalEmbedding", "ModelRotaryPositionalEmbedding", "Module", "ModuleList", "None", "_", "__init__", "class", "config", "def", "dropout", "elif", "else", "embed_positions", "for", "gradient_checkpointing", "hidden_dropout", "if", "in", "layers", "nn", "num_hidden_layers", "position_embeddings_type", "range", "relative", "rotary", "self", "super"], "wav2vec2_bert/modeling_wav2vec2_bert.py:Wav2Vec2BertEncoder.forward": ["BaseModelOutput", "False", "ModelEncoder", "Module", "None", "True", "all_hidden_states", "all_self_attentions", "and", "attention_mask", "attentions", "bool", "class", "config", "conv_attention_mask", "def", "dropout", "dropout_probability", "dtype", "else", "embed_positions", "enumerate", "expand", "finfo", "for", "forward", "hidden_states", "i", "if", "in", "is", "is_deepspeed_zero3_enabled", "is_fsdp_managed_module", "last_hidden_state", "layer", "layer_outputs", "layerdrop", "layers", "masked_fill", "min", "nn", "not", "or", "output_attentions", "output_hidden_states", "rand", "relative_position_embeddings", "return", "return_dict", "self", "shape", "skip_the_layer", "synced_gpus", "to", "torch", "training", "tuple", "unsqueeze", "v"], "wav2vec2_bert/modeling_wav2vec2_bert.py:Wav2Vec2BertAdapter.__init__": ["LayerNorm", "Linear", "ModelAdapter", "ModelAdapterLayer", "Module", "ModuleList", "None", "_", "__init__", "adapter_kernel_size", "adapter_stride", "class", "config", "def", "else", "eps", "for", "hidden_size", "if", "in", "kernel_size", "layer_norm_eps", "layerdrop", "layers", "nn", "num_adapter_layers", "output_hidden_size", "proj", "proj_layer_norm", "range", "self", "stride", "super"], "wav2vec2_bert/modeling_wav2vec2_bert.py:Wav2Vec2BertAdapter._compute_sub_sample_lengths_from_attention_mask": ["ModelAdapter", "Module", "None", "_compute_sub_sample_lengths_from_attention_mask", "class", "def", "floor", "if", "is", "kernel_size", "nn", "pad", "return", "self", "seq_lens", "stride"], "wav2vec2_bert/modeling_wav2vec2_bert.py:Wav2Vec2BertAdapter.forward": ["ModelAdapter", "Module", "None", "_compute_sub_sample_lengths_from_attention_mask", "and", "attention_mask", "class", "def", "device", "for", "forward", "hidden_states", "if", "in", "int", "is", "layer", "layerdrop", "layerdrop_prob", "layers", "nn", "not", "or", "proj", "proj_layer_norm", "rand", "return", "self", "size", "sub_sampled_lengths", "sum", "to", "torch", "training"], "wav2vec2_bert/modeling_wav2vec2_bert.py:_compute_new_attention_mask": ["Tensor", "_compute_new_attention_mask", "arange", "batch_size", "bool_mask", "def", "device", "expand", "hidden_states", "indices", "mask", "mask_seq_len", "masked_fill", "new_ones", "return", "seq_lens", "shape", "torch", "unsqueeze"], "wav2vec2_bert/modeling_wav2vec2_bert.py:Wav2Vec2BertAdapterLayer.__init__": ["Conv1d", "Dropout", "GLU", "LayerNorm", "ModelAdapterLayer", "ModelFeedForward", "ModelSelfAttention", "Module", "True", "__init__", "act_fn", "activation", "adapter_act", "adapter_kernel_size", "adapter_stride", "class", "config", "conformer_conv_dropout", "def", "dim", "dropout", "embed_dim", "eps", "ffn", "ffn_layer_norm", "hidden_size", "is_adapter_attention", "kernel_size", "layer_norm_eps", "nn", "output_hidden_size", "padding", "residual_conv", "residual_layer_norm", "self", "self_attn", "self_attn_conv", "self_attn_dropout", "self_attn_layer_norm", "stride", "super"], "wav2vec2_bert/modeling_wav2vec2_bert.py:Wav2Vec2BertAdapterLayer.forward": ["False", "ModelAdapterLayer", "Module", "None", "Tensor", "_compute_new_attention_mask", "_prepare_4d_attention_mask", "activation", "attention_mask", "attn_weights", "bool", "class", "def", "dtype", "ffn", "ffn_layer_norm", "forward", "hidden_states", "if", "is", "nn", "not", "output_attentions", "residual", "residual_conv", "residual_layer_norm", "return", "self", "self_attn", "self_attn_conv", "self_attn_dropout", "self_attn_layer_norm", "seq_lens", "sub_sampled_lengths", "torch", "transpose"], "wav2vec2_bert/modeling_wav2vec2_bert.py:Wav2Vec2BertPreTrainedModel._init_weights": ["AMSoftmaxLoss", "Conv1d", "GroupNorm", "LayerNorm", "Linear", "ModelFeatureProjection", "ModelForAudioFrameClassification", "ModelForSequenceClassification", "ModelForXVector", "ModelModel", "ModelPreTrainedModel", "ModelRelPositionalEmbedding", "ModelRotaryPositionalEmbedding", "ModelSelfAttention", "None", "PreTrainedModel", "_init_weights", "a", "arange", "b", "base", "bias", "class", "config", "constant_", "copy_", "def", "dim", "dtype", "elif", "expand", "extend_pe", "float", "groups", "hasattr", "hidden_size", "if", "in_channels", "in_features", "init", "initializer_range", "int64", "inv_freq", "is", "isinstance", "k", "kaiming_normal_", "kernel_size", "layer_weights", "masked_spec_embed", "math", "max_len", "mean", "module", "nn", "normal_", "not", "num_attention_heads", "num_hidden_layers", "ones_", "pe", "pos_bias_u", "pos_bias_v", "projection", "rotary_embedding_base", "self", "sqrt", "std", "tensor", "torch", "uniform_", "weight", "xavier_uniform_", "zeros_"], "wav2vec2_bert/modeling_wav2vec2_bert.py:Wav2Vec2BertPreTrainedModel._get_feat_extract_output_lengths": ["LongTensor", "ModelPreTrainedModel", "None", "PreTrainedModel", "_", "_conv_out_length", "_get_feat_extract_output_lengths", "adapter_kernel_size", "adapter_stride", "add_adapter", "bool", "class", "config", "def", "div", "else", "floor", "for", "if", "in", "input_length", "input_lengths", "int", "is", "kernel_size", "num_adapter_layers", "padding", "range", "return", "rounding_mode", "self", "stride", "torch"], "wav2vec2_bert/modeling_wav2vec2_bert.py:Wav2Vec2BertPreTrainedModel._get_feature_vector_attention_mask": ["LongTensor", "ModelPreTrainedModel", "None", "PreTrainedModel", "_get_feat_extract_output_lengths", "_get_feature_vector_attention_mask", "add_adapter", "arange", "attention_mask", "batch_size", "bool", "class", "cumsum", "def", "device", "dim", "dtype", "feature_vector_length", "flip", "int", "long", "non_padded_lengths", "output_lengths", "return", "self", "shape", "to", "torch", "zeros"], "wav2vec2_bert/modeling_wav2vec2_bert.py:_compute_mask_indices": ["False", "LongTensor", "None", "ValueError", "_", "_compute_mask_indices", "and", "append", "arange", "array", "attention_mask", "batch_size", "be", "bigger", "bool", "broadcast_to", "but", "choice", "compute_num_masked_span", "concatenate", "def", "detach", "dtype", "dummy_mask_idx", "else", "epsilon", "f", "float", "for", "got", "has", "if", "in", "input_length", "input_lengths", "int", "int32", "is", "item", "len", "mask_length", "mask_prob", "max", "max_num_masked_span", "min_masks", "ndarray", "not", "np", "num_masked_span", "offsets", "ones", "put_along_axis", "raise", "rand", "random", "range", "replace", "reshape", "return", "sequence_length", "shape", "smaller", "spec_aug_mask", "spec_aug_mask_idx", "spec_aug_mask_idxs", "sum", "than", "to", "tolist", "torch", "tuple", "zeros"], "wav2vec2_bert/modeling_wav2vec2_bert.py:Wav2Vec2BertModel.__init__": ["ModelAdapter", "ModelConfig", "ModelEncoder", "ModelFeatureProjection", "ModelFeedForward", "ModelModel", "ModelPreTrainedModel", "None", "Parameter", "Tensor", "__init__", "act_fn", "adapter", "add_adapter", "class", "config", "def", "else", "encoder", "feature_projection", "hidden_size", "if", "intermediate_ffn", "mask_feature_prob", "mask_time_prob", "masked_spec_embed", "nn", "or", "post_init", "relu", "self", "super", "torch", "uniform_", "use_intermediate_ffn_before_adapter"], "wav2vec2_bert/modeling_wav2vec2_bert.py:Wav2Vec2BertModel._mask_hidden_states": ["FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "True", "_compute_mask_indices", "_mask_hidden_states", "and", "apply_spec_augment", "attention_mask", "batch_size", "bool", "class", "config", "def", "device", "dtype", "elif", "expand", "getattr", "hidden_size", "hidden_states", "if", "is", "mask_feature_indices", "mask_feature_length", "mask_feature_min_masks", "mask_feature_prob", "mask_length", "mask_prob", "mask_time_indices", "mask_time_length", "mask_time_min_masks", "mask_time_prob", "masked_spec_embed", "min_masks", "not", "return", "self", "sequence_length", "size", "tensor", "to", "torch", "training"], "wav2vec2_bert/modeling_wav2vec2_bert.py:Wav2Vec2BertModel.forward": ["FloatTensor", "ModelBaseModelOutput", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "_mask_hidden_states", "adapter", "attention_mask", "attentions", "bool", "class", "config", "def", "else", "encoder", "encoder_outputs", "expanded_hidden_states", "extract_features", "feature_projection", "forward", "hidden_states", "if", "input_features", "intermediate_ffn", "is", "kwargs", "last_hidden_state", "mask_time_indices", "not", "output_attentions", "output_hidden_states", "r", "return", "return_dict", "self", "torch", "tuple", "use_return_dict"], "wav2vec2_bert/modeling_wav2vec2_bert.py:Wav2Vec2BertForCTC.__init__": ["Dropout", "Linear", "Model", "ModelForCTC", "ModelModel", "ModelPreTrainedModel", "None", "Please", "ValueError", "You", "__class__", "__init__", "a", "add_adapter", "and", "are", "as", "class", "config", "configuration", "def", "define", "does", "dropout", "else", "f", "final_dropout", "follows", "from_pretrained", "hasattr", "head", "hidden_size", "if", "instantiate", "is", "language", "lm_head", "model", "nn", "not", "of", "or", "output_hidden_size", "post_init", "r", "raise", "s", "self", "size", "str", "super", "target_lang", "that", "the", "to", "trying", "vocab_size", "vocabulary", "with", "your"], "wav2vec2_bert/modeling_wav2vec2_bert.py:Wav2Vec2BertForCTC.forward": ["CausalLMOutput", "False", "Label", "Model", "ModelForCTC", "ModelPreTrainedModel", "None", "Tensor", "ValueError", "_HIDDEN_STATES_START_POSITION", "_get_feat_extract_output_lengths", "and", "attention_mask", "attentions", "backends", "be", "blank", "bool", "class", "config", "ctc_loss", "ctc_loss_reduction", "ctc_zero_infinity", "cudnn", "def", "device", "dim", "dropout", "dtype", "else", "enabled", "f", "flags", "flattened_targets", "float32", "forward", "functional", "hidden_states", "if", "input_features", "input_lengths", "is", "kwargs", "labels", "labels_mask", "lm_head", "log_probs", "log_softmax", "logits", "long", "loss", "masked_select", "max", "must", "nn", "not", "ones", "output", "output_attentions", "output_hidden_states", "outputs", "pad_token_id", "r", "raise", "reduction", "return", "return_dict", "self", "shape", "sum", "target_lengths", "to", "torch", "transpose", "tuple", "use_return_dict", "values", "vocab_size", "with", "zero_infinity"], "wav2vec2_bert/modeling_wav2vec2_bert.py:Wav2Vec2BertForSequenceClassification.__init__": ["Linear", "Model", "ModelForSequenceClassification", "ModelModel", "ModelPreTrainedModel", "Parameter", "Sequence", "True", "ValueError", "__init__", "adapters", "add_adapter", "and", "class", "classification", "classifier", "classifier_proj_size", "config", "def", "does", "hasattr", "hidden_size", "if", "layer_weights", "nn", "not", "num_hidden_layers", "num_labels", "num_layers", "of", "ones", "post_init", "projector", "raise", "self", "super", "support", "the", "torch", "use", "use_weighted_layer_sum"], "wav2vec2_bert/modeling_wav2vec2_bert.py:Wav2Vec2BertForSequenceClassification.freeze_base_model": ["False", "Model", "ModelForSequenceClassification", "ModelPreTrainedModel", "class", "def", "for", "freeze_base_model", "in", "param", "parameters", "requires_grad", "self"], "wav2vec2_bert/modeling_wav2vec2_bert.py:Wav2Vec2BertForSequenceClassification.forward": ["CrossEntropyLoss", "Model", "ModelForSequenceClassification", "ModelPreTrainedModel", "None", "SequenceClassifierOutput", "Tensor", "True", "_HIDDEN_STATES_START_POSITION", "_get_feature_vector_attention_mask", "attention_mask", "attentions", "bool", "class", "classifier", "config", "def", "dim", "else", "expand_padding_mask", "forward", "functional", "hidden_states", "if", "input_features", "is", "kwargs", "labels", "layer_weights", "logits", "loss", "loss_fct", "mean", "nn", "norm_weights", "not", "num_labels", "output", "output_attentions", "output_hidden_states", "outputs", "padding_mask", "pooled_output", "projector", "r", "repeat", "return", "return_dict", "self", "shape", "softmax", "stack", "sum", "torch", "tuple", "unsqueeze", "use_return_dict", "use_weighted_layer_sum", "view"], "wav2vec2_bert/modeling_wav2vec2_bert.py:Wav2Vec2BertForAudioFrameClassification.__init__": ["Audio", "Linear", "Model", "ModelForAudioFrameClassification", "ModelModel", "ModelPreTrainedModel", "Parameter", "True", "ValueError", "__init__", "adapters", "add_adapter", "and", "class", "classification", "classifier", "config", "def", "does", "frame", "hasattr", "hidden_size", "if", "layer_weights", "nn", "not", "num_hidden_layers", "num_labels", "num_layers", "of", "ones", "post_init", "raise", "self", "super", "support", "the", "torch", "use", "use_weighted_layer_sum"], "wav2vec2_bert/modeling_wav2vec2_bert.py:Wav2Vec2BertForAudioFrameClassification.freeze_base_model": ["False", "Model", "ModelForAudioFrameClassification", "ModelPreTrainedModel", "class", "def", "for", "freeze_base_model", "in", "param", "parameters", "requires_grad", "self"], "wav2vec2_bert/modeling_wav2vec2_bert.py:Wav2Vec2BertForAudioFrameClassification.forward": ["CrossEntropyLoss", "Model", "ModelForAudioFrameClassification", "ModelPreTrainedModel", "None", "Tensor", "TokenClassifierOutput", "True", "_HIDDEN_STATES_START_POSITION", "argmax", "attention_mask", "attentions", "axis", "bool", "class", "classifier", "config", "def", "dim", "else", "forward", "functional", "hidden_states", "if", "input_features", "is", "kwargs", "labels", "layer_weights", "logits", "loss", "loss_fct", "nn", "norm_weights", "not", "num_labels", "output", "output_attentions", "output_hidden_states", "outputs", "r", "return", "return_dict", "self", "softmax", "stack", "sum", "torch", "tuple", "use_return_dict", "use_weighted_layer_sum", "view"], "wav2vec2_bert/modeling_wav2vec2_bert.py:AMSoftmaxLoss.__init__": ["CrossEntropyLoss", "Model", "Module", "Parameter", "True", "__init__", "class", "def", "input_dim", "loss", "margin", "nn", "num_labels", "randn", "requires_grad", "scale", "self", "super", "torch", "weight"], "wav2vec2_bert/modeling_wav2vec2_bert.py:AMSoftmaxLoss.forward": ["Model", "Module", "bool", "class", "cos_theta", "def", "dim", "flatten", "forward", "functional", "hidden_states", "labels", "logits", "loss", "margin", "mm", "nn", "normalize", "num_labels", "one_hot", "onehot", "psi", "return", "scale", "self", "torch", "weight", "where"], "wav2vec2_bert/modeling_wav2vec2_bert.py:TDNNLayer.__init__": ["Linear", "Model", "Module", "ReLU", "__init__", "activation", "class", "config", "def", "dilation", "else", "if", "in_conv_dim", "kernel", "kernel_size", "layer_id", "nn", "out_conv_dim", "self", "super", "tdnn_dilation", "tdnn_dim", "tdnn_kernel"], "wav2vec2_bert/modeling_wav2vec2_bert.py:TDNNLayer.forward": ["Detected", "LoRA", "LoraLayer", "Model", "Module", "Tensor", "You", "activation", "applied", "be", "bias", "class", "conv1d", "def", "dilation", "due", "exclude", "forward", "from", "functional", "hidden_states", "if", "in_conv_dim", "is_peft_available", "isinstance", "kernel", "kernel_size", "modules", "nn", "on", "optimization", "out_conv_dim", "return", "s", "self", "should", "t", "target", "to", "torch", "transpose", "view", "warn", "warnings", "weight", "weights", "won"], "wav2vec2_bert/modeling_wav2vec2_bert.py:Wav2Vec2BertForXVector.__init__": ["AMSoftmaxLoss", "Linear", "Model", "ModelForXVector", "ModelModel", "ModelPreTrainedModel", "ModuleList", "Parameter", "TDNNLayer", "__init__", "class", "classifier", "config", "def", "feature_extractor", "for", "hidden_size", "i", "if", "in", "layer_weights", "len", "nn", "num_hidden_layers", "num_labels", "num_layers", "objective", "ones", "post_init", "projector", "range", "self", "super", "tdnn", "tdnn_dim", "tdnn_layers", "torch", "use_weighted_layer_sum", "xvector_output_dim"], "wav2vec2_bert/modeling_wav2vec2_bert.py:Wav2Vec2BertForXVector.freeze_base_model": ["False", "Model", "ModelForXVector", "ModelPreTrainedModel", "class", "def", "for", "freeze_base_model", "in", "param", "parameters", "requires_grad", "self"], "wav2vec2_bert/modeling_wav2vec2_bert.py:Wav2Vec2BertForXVector._get_tdnn_output_lengths": ["LongTensor", "ModelForXVector", "ModelPreTrainedModel", "_conv_out_length", "_get_tdnn_output_lengths", "class", "config", "def", "for", "in", "input_length", "input_lengths", "int", "kernel_size", "return", "self", "stride", "tdnn_kernel", "torch"], "wav2vec2_bert/modeling_wav2vec2_bert.py:Wav2Vec2BertForXVector.forward": ["Model", "ModelForXVector", "ModelPreTrainedModel", "None", "Tensor", "True", "XVectorOutput", "_HIDDEN_STATES_START_POSITION", "_get_feat_extract_output_lengths", "_get_tdnn_output_lengths", "append", "attention_mask", "attentions", "bool", "cat", "class", "classifier", "config", "def", "dim", "else", "embeddings", "enumerate", "feat_extract_output_lengths", "feature_extractor", "for", "forward", "functional", "hidden_states", "i", "if", "in", "input_features", "is", "kwargs", "labels", "layer_weights", "length", "logits", "loss", "mean", "mean_features", "nn", "norm_weights", "not", "objective", "output", "output_attentions", "output_embeddings", "output_hidden_states", "outputs", "projector", "r", "return", "return_dict", "self", "softmax", "stack", "statistic_pooling", "std", "std_features", "sum", "tdnn", "tdnn_layer", "tdnn_output_lengths", "torch", "tuple", "use_return_dict", "use_weighted_layer_sum", "view"], "wav2vec2_conformer/modeling_wav2vec2_conformer.py:Wav2Vec2ConformerSamePadLayer.__init__": ["ModelSamePadLayer", "Module", "__init__", "class", "def", "else", "if", "nn", "num_conv_pos_embeddings", "num_pad_remove", "self", "super"], "wav2vec2_conformer/modeling_wav2vec2_conformer.py:Wav2Vec2ConformerSamePadLayer.forward": ["ModelSamePadLayer", "Module", "class", "def", "forward", "hidden_states", "if", "nn", "num_pad_remove", "return", "self"], "wav2vec2_conformer/modeling_wav2vec2_conformer.py:Wav2Vec2ConformerPositionalConvEmbedding.__init__": ["ACT2FN", "Conv1d", "GatheredParameters", "ModelPositionalConvEmbedding", "ModelSamePadLayer", "Module", "__init__", "activation", "class", "config", "conv", "deepspeed", "def", "dim", "else", "feat_extract_activation", "groups", "hasattr", "hidden_size", "if", "is_deepspeed_zero3_enabled", "kernel_size", "modifier_rank", "name", "nn", "num_conv_pos_embedding_groups", "num_conv_pos_embeddings", "original0", "original1", "padding", "parametrizations", "register_external_parameter", "self", "super", "utils", "weight", "weight_g", "weight_norm", "weight_v", "with", "zero"], "wav2vec2_conformer/modeling_wav2vec2_conformer.py:Wav2Vec2ConformerPositionalConvEmbedding.forward": ["ModelPositionalConvEmbedding", "Module", "activation", "class", "conv", "def", "forward", "hidden_states", "nn", "padding", "return", "self", "transpose"], "wav2vec2_conformer/modeling_wav2vec2_conformer.py:Wav2Vec2ConformerRotaryPositionalEmbedding.__init__": ["ModelRotaryPositionalEmbedding", "Module", "None", "__init__", "arange", "base", "cached_rotary_positional_embedding", "cached_sequence_length", "class", "config", "def", "dim", "dtype", "float", "hidden_size", "int64", "inv_freq", "nn", "num_attention_heads", "register_buffer", "rotary_embedding_base", "self", "super", "torch"], "wav2vec2_conformer/modeling_wav2vec2_conformer.py:Wav2Vec2ConformerRotaryPositionalEmbedding.forward": ["ModelRotaryPositionalEmbedding", "Module", "None", "and", "arange", "cached_rotary_positional_embedding", "cached_sequence_length", "cat", "class", "cos", "cos_embeddings", "def", "dim", "einsum", "embeddings", "forward", "freqs", "hidden_states", "i", "if", "ij", "inv_freq", "is", "j", "nn", "not", "return", "self", "sequence_length", "shape", "sin", "sin_embeddings", "stack", "time_stamps", "torch", "type_as"], "wav2vec2_conformer/modeling_wav2vec2_conformer.py:Wav2Vec2ConformerRelPositionalEmbedding.__init__": ["False", "ModelRelPositionalEmbedding", "Module", "__init__", "class", "config", "d_model", "def", "expand", "extend_pe", "hidden_size", "max_len", "max_source_positions", "nn", "pe", "persistent", "register_buffer", "self", "super", "tensor", "torch"], "wav2vec2_conformer/modeling_wav2vec2_conformer.py:Wav2Vec2ConformerRelPositionalEmbedding.extend_pe": ["ModelRelPositionalEmbedding", "Module", "None", "arange", "cat", "class", "cos", "d_model", "def", "device", "dim", "div_term", "dtype", "exp", "extend_pe", "flip", "float", "if", "int64", "is", "log", "math", "nn", "not", "or", "pe", "pe_negative", "pe_positive", "position", "return", "self", "sin", "size", "to", "torch", "unsqueeze", "x", "zeros"], "wav2vec2_conformer/modeling_wav2vec2_conformer.py:Wav2Vec2ConformerRelPositionalEmbedding.forward": ["ModelRelPositionalEmbedding", "Module", "Tensor", "class", "def", "end_idx", "extend_pe", "forward", "hidden_states", "nn", "pe", "relative_position_embeddings", "return", "self", "size", "start_idx", "torch"], "wav2vec2_conformer/modeling_wav2vec2_conformer.py:Wav2Vec2ConformerNoLayerNormConvLayer.__init__": ["ACT2FN", "Conv1d", "GradientCheckpointingLayer", "ModelNoLayerNormConvLayer", "__init__", "activation", "bias", "class", "config", "conv", "conv_bias", "conv_dim", "conv_kernel", "conv_stride", "def", "else", "feat_extract_activation", "if", "in_conv_dim", "kernel_size", "layer_id", "nn", "out_conv_dim", "self", "stride", "super"], "wav2vec2_conformer/modeling_wav2vec2_conformer.py:Wav2Vec2ConformerNoLayerNormConvLayer.forward": ["GradientCheckpointingLayer", "ModelNoLayerNormConvLayer", "activation", "class", "conv", "def", "forward", "hidden_states", "return", "self"], "wav2vec2_conformer/modeling_wav2vec2_conformer.py:Wav2Vec2ConformerLayerNormConvLayer.__init__": ["ACT2FN", "Conv1d", "GradientCheckpointingLayer", "LayerNorm", "ModelLayerNormConvLayer", "True", "__init__", "activation", "bias", "class", "config", "conv", "conv_bias", "conv_dim", "conv_kernel", "conv_stride", "def", "elementwise_affine", "else", "feat_extract_activation", "if", "in_conv_dim", "kernel_size", "layer_id", "layer_norm", "nn", "out_conv_dim", "self", "stride", "super"], "wav2vec2_conformer/modeling_wav2vec2_conformer.py:Wav2Vec2ConformerLayerNormConvLayer.forward": ["GradientCheckpointingLayer", "ModelLayerNormConvLayer", "activation", "class", "conv", "def", "forward", "hidden_states", "layer_norm", "return", "self", "transpose"], "wav2vec2_conformer/modeling_wav2vec2_conformer.py:Wav2Vec2ConformerGroupNormConvLayer.__init__": ["ACT2FN", "Conv1d", "GradientCheckpointingLayer", "GroupNorm", "ModelGroupNormConvLayer", "True", "__init__", "activation", "affine", "bias", "class", "config", "conv", "conv_bias", "conv_dim", "conv_kernel", "conv_stride", "def", "else", "feat_extract_activation", "if", "in_conv_dim", "kernel_size", "layer_id", "layer_norm", "nn", "num_channels", "num_groups", "out_conv_dim", "self", "stride", "super"], "wav2vec2_conformer/modeling_wav2vec2_conformer.py:Wav2Vec2ConformerGroupNormConvLayer.forward": ["GradientCheckpointingLayer", "ModelGroupNormConvLayer", "activation", "class", "conv", "def", "forward", "hidden_states", "layer_norm", "return", "self"], "wav2vec2_conformer/modeling_wav2vec2_conformer.py:Wav2Vec2ConformerFeatureEncoder.__init__": ["False", "ModelFeatureEncoder", "ModelGroupNormConvLayer", "ModelLayerNormConvLayer", "ModelNoLayerNormConvLayer", "Module", "ModuleList", "True", "ValueError", "__init__", "_requires_grad", "be", "but", "class", "config", "conv_layers", "def", "elif", "else", "f", "feat_extract_norm", "for", "gradient_checkpointing", "group", "has", "i", "if", "in", "is", "layer", "layer_id", "nn", "num_feat_extract_layers", "of", "one", "raise", "range", "self", "super", "to"], "wav2vec2_conformer/modeling_wav2vec2_conformer.py:Wav2Vec2ConformerFeatureEncoder._freeze_parameters": ["False", "ModelFeatureEncoder", "Module", "_freeze_parameters", "_requires_grad", "class", "def", "for", "in", "nn", "param", "parameters", "requires_grad", "self"], "wav2vec2_conformer/modeling_wav2vec2_conformer.py:Wav2Vec2ConformerFeatureEncoder.forward": ["ModelFeatureEncoder", "Module", "None", "True", "_requires_grad", "and", "class", "conv_layer", "conv_layers", "def", "for", "forward", "hidden_states", "if", "in", "input_values", "nn", "requires_grad", "return", "self", "training"], "wav2vec2_conformer/modeling_wav2vec2_conformer.py:Wav2Vec2ConformerFeatureProjection.__init__": ["Dropout", "LayerNorm", "Linear", "ModelFeatureProjection", "Module", "__init__", "class", "config", "conv_dim", "def", "dropout", "eps", "feat_proj_dropout", "hidden_size", "layer_norm", "layer_norm_eps", "nn", "projection", "self", "super"], "wav2vec2_conformer/modeling_wav2vec2_conformer.py:Wav2Vec2ConformerFeatureProjection.forward": ["ModelFeatureProjection", "Module", "class", "def", "dropout", "forward", "hidden_states", "layer_norm", "nn", "norm_hidden_states", "projection", "return", "self"], "wav2vec2_conformer/modeling_wav2vec2_conformer.py:Wav2Vec2ConformerFeedForward.__init__": ["ACT2FN", "Dropout", "Linear", "ModelFeedForward", "Module", "__init__", "activation_dropout", "class", "config", "def", "else", "hidden_act", "hidden_dropout", "hidden_size", "if", "intermediate_act_fn", "intermediate_dense", "intermediate_dropout", "intermediate_size", "isinstance", "nn", "output_dense", "output_dropout", "self", "str", "super"], "wav2vec2_conformer/modeling_wav2vec2_conformer.py:Wav2Vec2ConformerFeedForward.forward": ["ModelFeedForward", "Module", "class", "def", "forward", "hidden_states", "intermediate_act_fn", "intermediate_dense", "intermediate_dropout", "nn", "output_dense", "output_dropout", "return", "self"], "wav2vec2_conformer/modeling_wav2vec2_conformer.py:Wav2Vec2ConformerConvolutionModule.__init__": ["ACT2FN", "BatchNorm1d", "Conv1d", "Dropout", "False", "GLU", "LayerNorm", "ModelConvolutionModule", "Module", "SAME", "ValueError", "__init__", "a", "activation", "batch_norm", "be", "bias", "class", "config", "conformer_conv_dropout", "conv_depthwise_kernel_size", "def", "depthwise_conv", "dim", "dropout", "for", "glu", "groups", "hidden_act", "hidden_size", "if", "kernel_size", "layer_norm", "nn", "number", "odd", "padding", "pointwise_conv1", "pointwise_conv2", "raise", "self", "should", "stride", "super"], "wav2vec2_conformer/modeling_wav2vec2_conformer.py:Wav2Vec2ConformerConvolutionModule.forward": ["ModelConvolutionModule", "Module", "activation", "batch_norm", "class", "def", "depthwise_conv", "dropout", "forward", "glu", "hidden_states", "layer_norm", "nn", "pointwise_conv1", "pointwise_conv2", "return", "self", "transpose"], "wav2vec2_conformer/modeling_wav2vec2_conformer.py:Wav2Vec2ConformerSelfAttention.__init__": ["Dropout", "False", "Linear", "ModelSelfAttention", "Module", "Parameter", "__init__", "attention_dropout", "bias", "class", "config", "def", "dropout", "head_size", "hidden_size", "if", "linear_k", "linear_out", "linear_pos", "linear_q", "linear_v", "nn", "num_attention_heads", "num_heads", "p", "pos_bias_u", "pos_bias_v", "position_embeddings_type", "relative", "self", "super", "torch", "zeros"], "wav2vec2_conformer/modeling_wav2vec2_conformer.py:Wav2Vec2ConformerSelfAttention.forward": ["False", "ModelSelfAttention", "Module", "None", "Tensor", "ValueError", "_apply_relative_embeddings", "_apply_rotary_embedding", "attention_mask", "batch_size", "be", "bool", "class", "def", "defined", "dim", "dropout", "else", "forward", "has", "head_size", "hidden_size", "hidden_states", "if", "is", "key", "linear_k", "linear_out", "linear_q", "linear_v", "math", "matmul", "nn", "not", "num_heads", "output_attentions", "position_embeddings_type", "probs", "query", "query_key_states", "raise", "relative", "relative_position_embeddings", "reshape", "return", "rotary", "scores", "self", "sequence_length", "size", "softmax", "sqrt", "to", "torch", "transpose", "tuple", "value", "value_states", "view", "when"], "wav2vec2_conformer/modeling_wav2vec2_conformer.py:Wav2Vec2ConformerSelfAttention._apply_rotary_embedding": ["ModelSelfAttention", "Module", "_apply_rotary_embedding", "batch_size", "cat", "class", "cos", "def", "dim", "head_size", "hidden_size", "hidden_states", "ndim", "nn", "num_heads", "relative_position_embeddings", "return", "rotated_states", "rotated_states_begin", "rotated_states_end", "self", "sequence_length", "sin", "size", "torch", "transpose", "view"], "wav2vec2_conformer/modeling_wav2vec2_conformer.py:Wav2Vec2ConformerSelfAttention._apply_relative_embeddings": ["ModelSelfAttention", "Module", "_apply_relative_embeddings", "cat", "class", "def", "device", "dim", "dtype", "head_size", "key", "linear_pos", "math", "matmul", "nn", "num_heads", "pos_bias_u", "pos_bias_v", "proj_relative_position_embeddings", "q_with_bias_u", "q_with_bias_v", "query", "relative_position_embeddings", "return", "scores", "scores_ac", "scores_bd", "scores_bd_padded", "scores_bd_padded_shape", "self", "shape", "size", "sqrt", "torch", "transpose", "view", "view_as", "zero_pad", "zeros"], "wav2vec2_conformer/modeling_wav2vec2_conformer.py:Wav2Vec2ConformerEncoderLayer.__init__": ["Dropout", "GradientCheckpointingLayer", "LayerNorm", "ModelConvolutionModule", "ModelEncoderLayer", "ModelFeedForward", "ModelSelfAttention", "__init__", "attention_dropout", "class", "config", "conv_module", "def", "dropout", "embed_dim", "ffn1", "ffn1_layer_norm", "ffn2", "ffn2_layer_norm", "final_layer_norm", "hidden_size", "nn", "self", "self_attn", "self_attn_dropout", "self_attn_layer_norm", "super"], "wav2vec2_conformer/modeling_wav2vec2_conformer.py:Wav2Vec2ConformerEncoderLayer.forward": ["False", "GradientCheckpointingLayer", "ModelEncoderLayer", "None", "Tensor", "attention_mask", "attn_weigts", "bool", "class", "conv_module", "def", "ffn1", "ffn1_layer_norm", "ffn2", "ffn2_layer_norm", "final_layer_norm", "forward", "hidden_states", "output_attentions", "relative_position_embeddings", "residual", "return", "self", "self_attn", "self_attn_dropout", "self_attn_layer_norm", "torch"], "wav2vec2_conformer/modeling_wav2vec2_conformer.py:Wav2Vec2ConformerEncoder.__init__": ["Dropout", "False", "LayerNorm", "ModelEncoder", "ModelEncoderLayer", "ModelPositionalConvEmbedding", "ModelRelPositionalEmbedding", "ModelRotaryPositionalEmbedding", "Module", "ModuleList", "None", "_", "__init__", "class", "config", "def", "dropout", "elif", "else", "embed_positions", "eps", "for", "gradient_checkpointing", "hidden_dropout", "hidden_size", "if", "in", "layer_norm", "layer_norm_eps", "layers", "nn", "num_hidden_layers", "pos_conv_embed", "position_embeddings_type", "range", "relative", "rotary", "self", "super"], "wav2vec2_conformer/modeling_wav2vec2_conformer.py:Wav2Vec2ConformerEncoder.forward": ["BaseModelOutput", "False", "ModelEncoder", "Module", "None", "True", "all_hidden_states", "all_self_attentions", "and", "attention_mask", "attentions", "class", "config", "def", "dropout", "dropout_probability", "dtype", "else", "embed_positions", "enumerate", "expand", "expand_attention_mask", "finfo", "for", "forward", "hidden_states", "i", "if", "in", "is", "is_deepspeed_zero3_enabled", "is_fsdp_managed_module", "last_hidden_state", "layer", "layer_norm", "layer_outputs", "layerdrop", "layers", "min", "nn", "not", "or", "output_attentions", "output_hidden_states", "rand", "relative_position_embeddings", "repeat", "return", "return_dict", "self", "shape", "skip_the_layer", "synced_gpus", "to", "torch", "training", "tuple", "unsqueeze", "v"], "wav2vec2_conformer/modeling_wav2vec2_conformer.py:Wav2Vec2ConformerGumbelVectorQuantizer.__init__": ["FloatTensor", "Linear", "ModelGumbelVectorQuantizer", "Module", "Parameter", "ValueError", "__init__", "be", "by", "class", "codevector_dim", "codevectors", "concatenation", "config", "conv_dim", "def", "divisible", "f", "for", "if", "must", "nn", "num_codevector_groups", "num_codevectors_per_group", "num_groups", "num_vars", "raise", "self", "super", "temperature", "torch", "weight_proj"], "wav2vec2_conformer/modeling_wav2vec2_conformer.py:Wav2Vec2ConformerGumbelVectorQuantizer._compute_perplexity": ["ModelGumbelVectorQuantizer", "Module", "None", "_compute_perplexity", "class", "def", "dim", "else", "exp", "expand", "flatten", "if", "is", "log", "marginal_probs", "mask", "mask_extended", "mean", "nn", "not", "perplexity", "probs", "return", "shape", "sum", "torch", "where", "zeros_like"], "wav2vec2_conformer/modeling_wav2vec2_conformer.py:Wav2Vec2ConformerGumbelVectorQuantizer.forward": ["ModelGumbelVectorQuantizer", "Module", "None", "True", "_compute_perplexity", "argmax", "batch_size", "class", "codevector_idx", "codevector_probs", "codevector_soft_dist", "codevectors", "codevectors_per_group", "def", "dim", "else", "float", "forward", "functional", "gumbel_softmax", "hard", "hidden_size", "hidden_states", "if", "mask_time_indices", "new_zeros", "nn", "num_groups", "num_vars", "perplexity", "return", "scatter_", "self", "sequence_length", "shape", "softmax", "sum", "tau", "temperature", "torch", "training", "type_as", "unsqueeze", "view", "weight_proj"], "wav2vec2_conformer/modeling_wav2vec2_conformer.py:Wav2Vec2ConformerAdapter.__init__": ["LayerNorm", "Linear", "ModelAdapter", "ModelAdapterLayer", "Module", "ModuleList", "None", "_", "__init__", "class", "config", "def", "else", "for", "hidden_size", "if", "in", "layerdrop", "layers", "nn", "num_adapter_layers", "output_hidden_size", "proj", "proj_layer_norm", "range", "self", "super"], "wav2vec2_conformer/modeling_wav2vec2_conformer.py:Wav2Vec2ConformerAdapter.forward": ["ModelAdapter", "Module", "None", "and", "class", "def", "for", "forward", "hidden_states", "if", "in", "is", "layer", "layerdrop", "layerdrop_prob", "layers", "nn", "not", "np", "or", "proj", "proj_layer_norm", "random", "return", "self", "training", "transpose"], "wav2vec2_conformer/modeling_wav2vec2_conformer.py:Wav2Vec2ConformerAdapterLayer.__init__": ["Conv1d", "ModelAdapterLayer", "Module", "__init__", "adapter_kernel_size", "adapter_stride", "class", "config", "conv", "def", "nn", "output_hidden_size", "padding", "self", "stride", "super"], "wav2vec2_conformer/modeling_wav2vec2_conformer.py:Wav2Vec2ConformerAdapterLayer.forward": ["ModelAdapterLayer", "Module", "class", "conv", "def", "dim", "forward", "functional", "glu", "hidden_states", "nn", "return", "self"], "wav2vec2_conformer/modeling_wav2vec2_conformer.py:Wav2Vec2ConformerPreTrainedModel._init_weights": ["BatchNorm1d", "Conv1d", "GroupNorm", "LayerNorm", "Linear", "ModelFeatureProjection", "ModelForPreTraining", "ModelGumbelVectorQuantizer", "ModelPositionalConvEmbedding", "ModelPreTrainedModel", "ModelRelPositionalEmbedding", "ModelRotaryPositionalEmbedding", "ModelSelfAttention", "None", "PreTrainedModel", "_init_weights", "a", "arange", "b", "base", "bias", "class", "codevectors", "config", "constant_", "conv", "copy_", "def", "dim", "dtype", "elif", "expand", "extend_pe", "float", "getattr", "groups", "hasattr", "hidden_size", "if", "in_channels", "in_features", "init", "initializer_range", "int64", "inv_freq", "is", "isinstance", "k", "kaiming_normal_", "kernel_size", "math", "max_len", "mean", "module", "nn", "normal_", "not", "num_attention_heads", "num_batches_tracked", "ones_", "pe", "pos_bias_u", "pos_bias_v", "project_hid", "project_q", "projection", "reset_parameters", "rotary_embedding_base", "running_mean", "running_var", "self", "sqrt", "std", "tensor", "torch", "uniform_", "weight", "weight_proj", "xavier_uniform_", "zeros_"], "wav2vec2_conformer/modeling_wav2vec2_conformer.py:Wav2Vec2ConformerPreTrainedModel._get_feat_extract_output_lengths": ["LongTensor", "ModelPreTrainedModel", "None", "PreTrainedModel", "_", "_conv_out_length", "_get_feat_extract_output_lengths", "adapter_stride", "add_adapter", "bool", "class", "config", "conv_kernel", "conv_stride", "def", "div", "else", "floor", "for", "if", "in", "input_length", "input_lengths", "int", "is", "kernel_size", "num_adapter_layers", "range", "return", "rounding_mode", "self", "stride", "torch", "zip"], "wav2vec2_conformer/modeling_wav2vec2_conformer.py:Wav2Vec2ConformerPreTrainedModel._get_feature_vector_attention_mask": ["LongTensor", "ModelPreTrainedModel", "None", "PreTrainedModel", "_get_feat_extract_output_lengths", "_get_feature_vector_attention_mask", "add_adapter", "arange", "attention_mask", "batch_size", "bool", "class", "cumsum", "def", "device", "dim", "dtype", "feature_vector_length", "flip", "int", "long", "non_padded_lengths", "output_lengths", "return", "self", "shape", "to", "torch", "zeros"], "wav2vec2_conformer/modeling_wav2vec2_conformer.py:_compute_mask_indices": ["False", "LongTensor", "None", "ValueError", "_", "_compute_mask_indices", "and", "append", "arange", "array", "attention_mask", "batch_size", "be", "bigger", "bool", "broadcast_to", "but", "choice", "compute_num_masked_span", "concatenate", "def", "detach", "dtype", "dummy_mask_idx", "else", "epsilon", "f", "float", "for", "got", "has", "if", "in", "input_length", "input_lengths", "int", "int32", "is", "item", "len", "mask_length", "mask_prob", "max", "max_num_masked_span", "min_masks", "ndarray", "not", "np", "num_masked_span", "offsets", "ones", "put_along_axis", "raise", "rand", "random", "range", "replace", "reshape", "return", "sequence_length", "shape", "smaller", "spec_aug_mask", "spec_aug_mask_idx", "spec_aug_mask_idxs", "sum", "than", "to", "tolist", "torch", "tuple", "zeros"], "wav2vec2_conformer/modeling_wav2vec2_conformer.py:Wav2Vec2ConformerModel.__init__": ["ModelAdapter", "ModelConfig", "ModelEncoder", "ModelFeatureEncoder", "ModelFeatureProjection", "ModelModel", "ModelPreTrainedModel", "None", "Parameter", "Tensor", "__init__", "adapter", "add_adapter", "class", "config", "def", "else", "encoder", "feature_extractor", "feature_projection", "hidden_size", "if", "mask_feature_prob", "mask_time_prob", "masked_spec_embed", "nn", "or", "post_init", "self", "super", "torch", "uniform_"], "wav2vec2_conformer/modeling_wav2vec2_conformer.py:Wav2Vec2ConformerModel.freeze_feature_encoder": ["ModelModel", "ModelPreTrainedModel", "_freeze_parameters", "class", "def", "feature_extractor", "freeze_feature_encoder", "self"], "wav2vec2_conformer/modeling_wav2vec2_conformer.py:Wav2Vec2ConformerModel._mask_hidden_states": ["FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "True", "_compute_mask_indices", "_mask_hidden_states", "and", "apply_spec_augment", "attention_mask", "batch_size", "bool", "class", "config", "def", "device", "dtype", "elif", "expand", "getattr", "hidden_size", "hidden_states", "if", "is", "mask_feature_indices", "mask_feature_length", "mask_feature_min_masks", "mask_feature_prob", "mask_length", "mask_prob", "mask_time_indices", "mask_time_length", "mask_time_min_masks", "mask_time_prob", "masked_spec_embed", "min_masks", "not", "return", "self", "sequence_length", "size", "tensor", "to", "torch", "training"], "wav2vec2_conformer/modeling_wav2vec2_conformer.py:Wav2Vec2ConformerModel.forward": ["False", "FloatTensor", "ModelBaseModelOutput", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "_get_feature_vector_attention_mask", "_mask_hidden_states", "adapter", "add_adapter", "attention_mask", "attentions", "bool", "class", "config", "def", "else", "encoder", "encoder_outputs", "extract_features", "feature_extractor", "feature_projection", "forward", "hidden_states", "if", "input_values", "is", "kwargs", "last_hidden_state", "mask_time_indices", "not", "output_attentions", "output_hidden_states", "r", "return", "return_dict", "self", "shape", "torch", "transpose", "tuple", "use_return_dict"], "wav2vec2_conformer/modeling_wav2vec2_conformer.py:Wav2Vec2ConformerForPreTraining.__init__": ["Dropout", "Linear", "Model", "ModelConfig", "ModelForPreTraining", "ModelGumbelVectorQuantizer", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "codevector_dim", "config", "def", "dropout_features", "feat_quantizer_dropout", "hidden_size", "nn", "post_init", "proj_codevector_dim", "project_hid", "project_q", "quantizer", "self", "super"], "wav2vec2_conformer/modeling_wav2vec2_conformer.py:Wav2Vec2ConformerForPreTraining.set_gumbel_temperature": ["ModelForPreTraining", "ModelPreTrainedModel", "class", "def", "int", "quantizer", "self", "set_gumbel_temperature", "temperature"], "wav2vec2_conformer/modeling_wav2vec2_conformer.py:Wav2Vec2ConformerForPreTraining.freeze_feature_encoder": ["Model", "ModelForPreTraining", "ModelPreTrainedModel", "_freeze_parameters", "class", "def", "feature_extractor", "freeze_feature_encoder", "self"], "wav2vec2_conformer/modeling_wav2vec2_conformer.py:Wav2Vec2ConformerForPreTraining.compute_contrastive_logits": ["FloatTensor", "ModelForPreTraining", "ModelPreTrainedModel", "cat", "class", "compute_contrastive_logits", "cosine_similarity", "def", "dim", "float", "logits", "negative_features", "predicted_features", "return", "target_features", "temperature", "torch", "type_as"], "wav2vec2_conformer/modeling_wav2vec2_conformer.py:Wav2Vec2ConformerForPreTraining.forward": ["BoolTensor", "False", "Model", "ModelForPreTraining", "ModelForPreTrainingOutput", "ModelPreTrainedModel", "None", "Tensor", "_get_feature_vector_attention_mask", "add_adapter", "all", "any", "attention_mask", "attentions", "batch_size", "bool", "class", "codevector_perplexity", "compute_contrastive_logits", "config", "contrastive_logits_temperature", "contrastive_loss", "cross_entropy", "def", "diversity_loss", "diversity_loss_weight", "dropout_features", "dtype", "else", "extract_features", "flatten", "float", "forward", "functional", "hidden_size", "hidden_states", "if", "inf", "input_values", "is", "kwargs", "logits", "long", "loss", "mask_time_indices", "neg_is_pos", "negative_quantized_features", "nn", "not", "num_codevector_groups", "num_codevectors", "num_codevectors_per_group", "output_attentions", "output_hidden_states", "outputs", "permute", "project_hid", "project_q", "projected_quantized_states", "projected_states", "quantized_features", "quantizer", "r", "reduction", "reshape", "return", "return_dict", "sampled_negative_indices", "self", "sequence_length", "shape", "size", "sum", "target", "to", "torch", "transformer_features", "transpose", "tuple", "use_return_dict", "view", "weight"], "wav2vec2_conformer/modeling_wav2vec2_conformer.py:Wav2Vec2ConformerForCTC.__init__": ["Dropout", "Linear", "Model", "ModelForCTC", "ModelModel", "ModelPreTrainedModel", "None", "Please", "ValueError", "You", "__class__", "__init__", "a", "add_adapter", "and", "are", "as", "class", "config", "configuration", "def", "define", "does", "dropout", "else", "f", "final_dropout", "follows", "from_pretrained", "hasattr", "head", "hidden_size", "if", "instantiate", "is", "language", "lm_head", "model", "nn", "not", "of", "or", "output_hidden_size", "post_init", "r", "raise", "s", "self", "size", "str", "super", "target_lang", "that", "the", "to", "trying", "vocab_size", "vocabulary", "with", "your"], "wav2vec2_conformer/modeling_wav2vec2_conformer.py:Wav2Vec2ConformerForCTC.freeze_feature_encoder": ["Model", "ModelForCTC", "ModelPreTrainedModel", "_freeze_parameters", "class", "def", "feature_extractor", "freeze_feature_encoder", "self"], "wav2vec2_conformer/modeling_wav2vec2_conformer.py:Wav2Vec2ConformerForCTC.forward": ["CausalLMOutput", "False", "Label", "Model", "ModelForCTC", "ModelPreTrainedModel", "None", "Tensor", "ValueError", "_HIDDEN_STATES_START_POSITION", "_get_feat_extract_output_lengths", "and", "attention_mask", "attentions", "backends", "be", "blank", "bool", "class", "config", "ctc_loss", "ctc_loss_reduction", "ctc_zero_infinity", "cudnn", "def", "dim", "dropout", "dtype", "else", "enabled", "f", "flags", "flattened_targets", "float32", "forward", "functional", "hidden_states", "if", "input_lengths", "input_values", "is", "kwargs", "labels", "labels_mask", "lm_head", "log_probs", "log_softmax", "logits", "long", "loss", "masked_select", "max", "must", "nn", "not", "ones_like", "output", "output_attentions", "output_hidden_states", "outputs", "pad_token_id", "r", "raise", "reduction", "return", "return_dict", "self", "sum", "target_lengths", "to", "torch", "transpose", "tuple", "use_return_dict", "values", "vocab_size", "with", "zero_infinity"], "wav2vec2_conformer/modeling_wav2vec2_conformer.py:Wav2Vec2ConformerForSequenceClassification.__init__": ["Linear", "Model", "ModelForSequenceClassification", "ModelModel", "ModelPreTrainedModel", "Parameter", "Sequence", "True", "ValueError", "__init__", "adapters", "add_adapter", "and", "class", "classification", "classifier", "classifier_proj_size", "config", "def", "does", "hasattr", "hidden_size", "if", "layer_weights", "nn", "not", "num_hidden_layers", "num_labels", "num_layers", "of", "ones", "post_init", "projector", "raise", "self", "super", "support", "the", "torch", "use", "use_weighted_layer_sum"], "wav2vec2_conformer/modeling_wav2vec2_conformer.py:Wav2Vec2ConformerForSequenceClassification.freeze_feature_encoder": ["Model", "ModelForSequenceClassification", "ModelPreTrainedModel", "_freeze_parameters", "class", "def", "feature_extractor", "freeze_feature_encoder", "self"], "wav2vec2_conformer/modeling_wav2vec2_conformer.py:Wav2Vec2ConformerForSequenceClassification.freeze_base_model": ["False", "Model", "ModelForSequenceClassification", "ModelPreTrainedModel", "class", "def", "for", "freeze_base_model", "in", "param", "parameters", "requires_grad", "self"], "wav2vec2_conformer/modeling_wav2vec2_conformer.py:Wav2Vec2ConformerForSequenceClassification.forward": ["CrossEntropyLoss", "Model", "ModelForSequenceClassification", "ModelPreTrainedModel", "None", "SequenceClassifierOutput", "Tensor", "True", "_HIDDEN_STATES_START_POSITION", "_get_feature_vector_attention_mask", "attention_mask", "attentions", "bool", "class", "classifier", "config", "def", "dim", "else", "expand_padding_mask", "forward", "functional", "hidden_states", "if", "input_values", "is", "kwargs", "labels", "layer_weights", "logits", "loss", "loss_fct", "mean", "nn", "norm_weights", "not", "num_labels", "output", "output_attentions", "output_hidden_states", "outputs", "padding_mask", "pooled_output", "projector", "r", "repeat", "return", "return_dict", "self", "shape", "softmax", "stack", "sum", "torch", "tuple", "unsqueeze", "use_return_dict", "use_weighted_layer_sum", "view"], "wav2vec2_conformer/modeling_wav2vec2_conformer.py:Wav2Vec2ConformerForAudioFrameClassification.__init__": ["Audio", "Linear", "Model", "ModelForAudioFrameClassification", "ModelModel", "ModelPreTrainedModel", "Parameter", "True", "ValueError", "__init__", "adapters", "add_adapter", "and", "class", "classification", "classifier", "config", "def", "does", "frame", "hasattr", "hidden_size", "if", "layer_weights", "nn", "not", "num_hidden_layers", "num_labels", "num_layers", "of", "ones", "post_init", "raise", "self", "super", "support", "the", "torch", "use", "use_weighted_layer_sum"], "wav2vec2_conformer/modeling_wav2vec2_conformer.py:Wav2Vec2ConformerForAudioFrameClassification.freeze_feature_encoder": ["Model", "ModelForAudioFrameClassification", "ModelPreTrainedModel", "_freeze_parameters", "class", "def", "feature_extractor", "freeze_feature_encoder", "self"], "wav2vec2_conformer/modeling_wav2vec2_conformer.py:Wav2Vec2ConformerForAudioFrameClassification.freeze_base_model": ["False", "Model", "ModelForAudioFrameClassification", "ModelPreTrainedModel", "class", "def", "for", "freeze_base_model", "in", "param", "parameters", "requires_grad", "self"], "wav2vec2_conformer/modeling_wav2vec2_conformer.py:Wav2Vec2ConformerForAudioFrameClassification.forward": ["CrossEntropyLoss", "Model", "ModelForAudioFrameClassification", "ModelPreTrainedModel", "None", "Tensor", "TokenClassifierOutput", "True", "_HIDDEN_STATES_START_POSITION", "argmax", "attention_mask", "attentions", "axis", "bool", "class", "classifier", "config", "def", "dim", "else", "forward", "functional", "hidden_states", "if", "input_values", "is", "kwargs", "labels", "layer_weights", "logits", "loss", "loss_fct", "nn", "norm_weights", "not", "num_labels", "output", "output_attentions", "output_hidden_states", "outputs", "r", "return", "return_dict", "self", "softmax", "stack", "sum", "torch", "tuple", "use_return_dict", "use_weighted_layer_sum", "view"], "wav2vec2_conformer/modeling_wav2vec2_conformer.py:AMSoftmaxLoss.__init__": ["CrossEntropyLoss", "Model", "Module", "Parameter", "True", "__init__", "class", "def", "input_dim", "loss", "margin", "nn", "num_labels", "randn", "requires_grad", "scale", "self", "super", "torch", "weight"], "wav2vec2_conformer/modeling_wav2vec2_conformer.py:AMSoftmaxLoss.forward": ["Model", "Module", "bool", "class", "cos_theta", "def", "dim", "flatten", "forward", "functional", "hidden_states", "labels", "logits", "loss", "margin", "mm", "nn", "normalize", "num_labels", "one_hot", "onehot", "psi", "return", "scale", "self", "torch", "weight", "where"], "wav2vec2_conformer/modeling_wav2vec2_conformer.py:TDNNLayer.__init__": ["Linear", "Model", "Module", "ReLU", "__init__", "activation", "class", "config", "def", "dilation", "else", "if", "in_conv_dim", "kernel", "kernel_size", "layer_id", "nn", "out_conv_dim", "self", "super", "tdnn_dilation", "tdnn_dim", "tdnn_kernel"], "wav2vec2_conformer/modeling_wav2vec2_conformer.py:TDNNLayer.forward": ["Detected", "LoRA", "LoraLayer", "Model", "Module", "Tensor", "You", "activation", "applied", "be", "bias", "class", "conv1d", "def", "dilation", "due", "exclude", "forward", "from", "functional", "hidden_states", "if", "in_conv_dim", "is_peft_available", "isinstance", "kernel", "kernel_size", "modules", "nn", "on", "optimization", "out_conv_dim", "return", "s", "self", "should", "t", "target", "to", "torch", "transpose", "view", "warn", "warnings", "weight", "weights", "won"], "wav2vec2_conformer/modeling_wav2vec2_conformer.py:Wav2Vec2ConformerForXVector.__init__": ["AMSoftmaxLoss", "Linear", "Model", "ModelForXVector", "ModelModel", "ModelPreTrainedModel", "ModuleList", "Parameter", "TDNNLayer", "__init__", "class", "classifier", "config", "def", "feature_extractor", "for", "hidden_size", "i", "if", "in", "layer_weights", "len", "nn", "num_hidden_layers", "num_labels", "num_layers", "objective", "ones", "post_init", "projector", "range", "self", "super", "tdnn", "tdnn_dim", "tdnn_layers", "torch", "use_weighted_layer_sum", "xvector_output_dim"], "wav2vec2_conformer/modeling_wav2vec2_conformer.py:Wav2Vec2ConformerForXVector.freeze_feature_encoder": ["Model", "ModelForXVector", "ModelPreTrainedModel", "_freeze_parameters", "class", "def", "feature_extractor", "freeze_feature_encoder", "self"], "wav2vec2_conformer/modeling_wav2vec2_conformer.py:Wav2Vec2ConformerForXVector.freeze_base_model": ["False", "Model", "ModelForXVector", "ModelPreTrainedModel", "class", "def", "for", "freeze_base_model", "in", "param", "parameters", "requires_grad", "self"], "wav2vec2_conformer/modeling_wav2vec2_conformer.py:Wav2Vec2ConformerForXVector._get_tdnn_output_lengths": ["LongTensor", "ModelForXVector", "ModelPreTrainedModel", "_conv_out_length", "_get_tdnn_output_lengths", "class", "config", "def", "for", "in", "input_length", "input_lengths", "int", "kernel_size", "return", "self", "stride", "tdnn_kernel", "torch"], "wav2vec2_conformer/modeling_wav2vec2_conformer.py:Wav2Vec2ConformerForXVector.forward": ["Model", "ModelForXVector", "ModelPreTrainedModel", "None", "Tensor", "True", "XVectorOutput", "_HIDDEN_STATES_START_POSITION", "_get_feat_extract_output_lengths", "_get_tdnn_output_lengths", "append", "attention_mask", "attentions", "bool", "cat", "class", "classifier", "config", "def", "dim", "else", "embeddings", "enumerate", "feat_extract_output_lengths", "feature_extractor", "for", "forward", "functional", "hidden_states", "i", "if", "in", "input_values", "is", "kwargs", "labels", "layer_weights", "length", "logits", "loss", "mean", "mean_features", "nn", "norm_weights", "not", "objective", "output", "output_attentions", "output_embeddings", "output_hidden_states", "outputs", "projector", "r", "return", "return_dict", "self", "softmax", "stack", "statistic_pooling", "std", "std_features", "sum", "tdnn", "tdnn_layer", "tdnn_output_lengths", "torch", "tuple", "use_return_dict", "use_weighted_layer_sum", "view"], "wavlm/modeling_wavlm.py:WavLMSamePadLayer.__init__": ["ModelSamePadLayer", "Module", "__init__", "class", "def", "else", "if", "nn", "num_conv_pos_embeddings", "num_pad_remove", "self", "super"], "wavlm/modeling_wavlm.py:WavLMSamePadLayer.forward": ["ModelSamePadLayer", "Module", "class", "def", "forward", "hidden_states", "if", "nn", "num_pad_remove", "return", "self"], "wavlm/modeling_wavlm.py:WavLMPositionalConvEmbedding.__init__": ["ACT2FN", "Conv1d", "GatheredParameters", "ModelPositionalConvEmbedding", "ModelSamePadLayer", "Module", "__init__", "activation", "class", "config", "conv", "deepspeed", "def", "dim", "else", "feat_extract_activation", "groups", "hasattr", "hidden_size", "if", "is_deepspeed_zero3_enabled", "kernel_size", "modifier_rank", "name", "nn", "num_conv_pos_embedding_groups", "num_conv_pos_embeddings", "original0", "original1", "padding", "parametrizations", "register_external_parameter", "self", "super", "utils", "weight", "weight_g", "weight_norm", "weight_v", "with", "zero"], "wavlm/modeling_wavlm.py:WavLMPositionalConvEmbedding.forward": ["ModelPositionalConvEmbedding", "Module", "activation", "class", "conv", "def", "forward", "hidden_states", "nn", "padding", "return", "self", "transpose"], "wavlm/modeling_wavlm.py:WavLMFeatureProjection.__init__": ["Dropout", "LayerNorm", "Linear", "ModelFeatureProjection", "Module", "__init__", "class", "config", "conv_dim", "def", "dropout", "eps", "feat_proj_dropout", "hidden_size", "layer_norm", "layer_norm_eps", "nn", "projection", "self", "super"], "wavlm/modeling_wavlm.py:WavLMFeatureProjection.forward": ["ModelFeatureProjection", "Module", "class", "def", "dropout", "forward", "hidden_states", "layer_norm", "nn", "norm_hidden_states", "projection", "return", "self"], "wavlm/modeling_wavlm.py:WavLMAttention.__init__": ["Embedding", "Linear", "ModelAttention", "Module", "Parameter", "True", "ValueError", "__init__", "and", "be", "bool", "by", "class", "def", "divisible", "dropout", "embed_dim", "f", "float", "got", "gru_rel_pos_const", "gru_rel_pos_linear", "has_relative_position_bias", "head_dim", "if", "int", "k_proj", "max_distance", "must", "nn", "num_buckets", "num_heads", "ones", "out_proj", "q_proj", "raise", "rel_attn_embed", "scaling", "self", "super", "torch", "v_proj"], "wavlm/modeling_wavlm.py:WavLMAttention.forward": ["False", "ModelAttention", "Module", "None", "Tensor", "_", "attention_mask", "attn_output", "attn_weights", "bool", "bsz", "chunk", "class", "compute_bias", "def", "dim", "forward", "gate_a", "gate_b", "gate_output", "gated_hidden_states", "gated_position_bias", "gru_rel_pos_const", "gru_rel_pos_linear", "hidden_states", "if", "index", "is", "nn", "num_heads", "output_attentions", "permute", "position_bias", "relative_position_proj", "repeat", "return", "self", "shape", "sigmoid", "size", "sum", "tgt_len", "torch", "torch_multi_head_self_attention", "tuple", "unsqueeze", "view"], "wavlm/modeling_wavlm.py:WavLMAttention.torch_multi_head_self_attention": ["BoolTensor", "F", "False", "FloatTensor", "LongTensor", "ModelAttention", "Module", "None", "True", "add_zero_attn", "attention_mask", "attn_output", "attn_weights", "bias", "bias_k", "bias_v", "bool", "broadcast_to", "cat", "class", "def", "dropout", "else", "embed_dim", "empty", "gated_position_bias", "hidden_states", "if", "is", "k_proj", "k_proj_weight", "key", "key_padding_mask", "multi_head_attention_forward", "ne", "nn", "not", "num_heads", "out_proj", "output_attentions", "q_proj", "q_proj_weight", "query", "return", "self", "shape", "torch", "torch_multi_head_self_attention", "training", "transpose", "tuple", "use_separate_proj_weight", "v_proj", "v_proj_weight", "value", "weight"], "wavlm/modeling_wavlm.py:WavLMAttention.compute_bias": ["FloatTensor", "ModelAttention", "Module", "None", "_relative_positions_bucket", "arange", "class", "compute_bias", "context_position", "def", "device", "dtype", "int", "key_length", "long", "memory_position", "nn", "permute", "query_length", "rel_attn_embed", "relative_position", "relative_position_bucket", "return", "self", "to", "torch", "values", "weight"], "wavlm/modeling_wavlm.py:WavLMAttention._relative_positions_bucket": ["FloatTensor", "ModelAttention", "Module", "_relative_positions_bucket", "abs", "class", "def", "float", "full_like", "is_small", "log", "long", "math", "max_distance", "max_exact", "min", "nn", "num_buckets", "relative_buckets", "relative_position_if_large", "relative_positions", "relative_positions_if_large", "return", "self", "to", "torch", "where"], "wavlm/modeling_wavlm.py:WavLMFeedForward.__init__": ["ACT2FN", "Dropout", "Linear", "ModelFeedForward", "Module", "__init__", "activation_dropout", "class", "config", "def", "else", "hidden_act", "hidden_dropout", "hidden_size", "if", "intermediate_act_fn", "intermediate_dense", "intermediate_dropout", "intermediate_size", "isinstance", "nn", "output_dense", "output_dropout", "self", "str", "super"], "wavlm/modeling_wavlm.py:WavLMFeedForward.forward": ["ModelFeedForward", "Module", "class", "def", "forward", "hidden_states", "intermediate_act_fn", "intermediate_dense", "intermediate_dropout", "nn", "output_dense", "output_dropout", "return", "self"], "wavlm/modeling_wavlm.py:WavLMEncoderLayer.__init__": ["Dropout", "GradientCheckpointingLayer", "LayerNorm", "ModelAttention", "ModelConfig", "ModelEncoderLayer", "ModelFeedForward", "True", "__init__", "attention", "attention_dropout", "bool", "class", "config", "def", "dropout", "embed_dim", "eps", "feed_forward", "final_layer_norm", "has_relative_position_bias", "hidden_dropout", "hidden_size", "layer_norm", "layer_norm_eps", "max_bucket_distance", "max_distance", "nn", "num_attention_heads", "num_buckets", "num_heads", "self", "super"], "wavlm/modeling_wavlm.py:WavLMEncoderLayer.forward": ["False", "GradientCheckpointingLayer", "ModelEncoderLayer", "None", "attention", "attention_mask", "attn_residual", "attn_weights", "class", "def", "dropout", "feed_forward", "final_layer_norm", "forward", "hidden_states", "if", "index", "layer_norm", "output_attentions", "outputs", "position_bias", "return", "self"], "wavlm/modeling_wavlm.py:WavLMEncoderLayerStableLayerNorm.__init__": ["Dropout", "GradientCheckpointingLayer", "LayerNorm", "ModelAttention", "ModelConfig", "ModelEncoderLayerStableLayerNorm", "ModelFeedForward", "True", "__init__", "attention", "attention_dropout", "bool", "class", "config", "def", "dropout", "embed_dim", "eps", "feed_forward", "final_layer_norm", "has_relative_position_bias", "hidden_dropout", "hidden_size", "layer_norm", "layer_norm_eps", "max_bucket_distance", "max_distance", "nn", "num_attention_heads", "num_buckets", "num_heads", "self", "super"], "wavlm/modeling_wavlm.py:WavLMEncoderLayerStableLayerNorm.forward": ["False", "GradientCheckpointingLayer", "ModelEncoderLayerStableLayerNorm", "None", "attention", "attention_mask", "attn_residual", "attn_weights", "class", "def", "dropout", "feed_forward", "final_layer_norm", "forward", "hidden_states", "if", "layer_norm", "output_attentions", "outputs", "position_bias", "return", "self"], "wavlm/modeling_wavlm.py:WavLMEncoder.__init__": ["Dropout", "False", "LayerNorm", "ModelEncoder", "ModelEncoderLayer", "ModelPositionalConvEmbedding", "Module", "ModuleList", "__init__", "class", "config", "def", "dropout", "eps", "for", "gradient_checkpointing", "has_relative_position_bias", "hidden_dropout", "hidden_size", "i", "in", "layer_norm", "layer_norm_eps", "layers", "nn", "num_hidden_layers", "pos_conv_embed", "range", "self", "super"], "wavlm/modeling_wavlm.py:WavLMEncoder.forward": ["BaseModelOutput", "False", "ModelEncoder", "Module", "None", "True", "all_hidden_states", "all_self_attentions", "and", "attention_mask", "attentions", "class", "config", "def", "dropout", "dropout_probability", "else", "enumerate", "expand_attention_mask", "for", "forward", "hidden_states", "i", "if", "in", "index", "is", "is_deepspeed_zero3_enabled", "is_fsdp_managed_module", "last_hidden_state", "layer", "layer_norm", "layer_outputs", "layerdrop", "layers", "nn", "not", "or", "output_attentions", "output_hidden_states", "pos_conv_embed", "position_bias", "position_embeddings", "rand", "repeat", "return", "return_dict", "self", "shape", "skip_the_layer", "synced_gpus", "torch", "training", "tuple", "unsqueeze", "v"], "wavlm/modeling_wavlm.py:WavLMEncoderStableLayerNorm.__init__": ["Dropout", "False", "LayerNorm", "ModelEncoderLayerStableLayerNorm", "ModelEncoderStableLayerNorm", "ModelPositionalConvEmbedding", "Module", "ModuleList", "__init__", "class", "config", "def", "dropout", "eps", "for", "gradient_checkpointing", "has_relative_position_bias", "hidden_dropout", "hidden_size", "i", "in", "layer_norm", "layer_norm_eps", "layers", "nn", "num_hidden_layers", "pos_conv_embed", "range", "self", "super"], "wavlm/modeling_wavlm.py:WavLMEncoderStableLayerNorm.forward": ["BaseModelOutput", "False", "ModelEncoderStableLayerNorm", "Module", "None", "True", "all_hidden_states", "all_self_attentions", "and", "attention_mask", "attentions", "class", "config", "def", "dropout", "dropout_probability", "else", "enumerate", "expand_attention_mask", "for", "forward", "hidden_states", "i", "if", "in", "is", "is_deepspeed_zero3_enabled", "is_fsdp_managed_module", "last_hidden_state", "layer", "layer_norm", "layer_outputs", "layerdrop", "layers", "nn", "not", "or", "output_attentions", "output_hidden_states", "pos_conv_embed", "position_bias", "position_embeddings", "rand", "repeat", "return", "return_dict", "self", "shape", "skip_the_layer", "synced_gpus", "torch", "training", "tuple", "unsqueeze", "v"], "wavlm/modeling_wavlm.py:WavLMGumbelVectorQuantizer.__init__": ["FloatTensor", "Linear", "ModelGumbelVectorQuantizer", "Module", "Parameter", "ValueError", "__init__", "be", "by", "class", "codevector_dim", "codevectors", "concatenation", "config", "conv_dim", "def", "divisible", "f", "for", "if", "must", "nn", "num_codevector_groups", "num_codevectors_per_group", "num_groups", "num_vars", "raise", "self", "super", "temperature", "torch", "weight_proj"], "wavlm/modeling_wavlm.py:WavLMGumbelVectorQuantizer._compute_perplexity": ["ModelGumbelVectorQuantizer", "Module", "_compute_perplexity", "class", "def", "dim", "exp", "log", "marginal_probs", "mean", "nn", "perplexity", "probs", "return", "sum", "torch"], "wavlm/modeling_wavlm.py:WavLMGumbelVectorQuantizer.forward": ["ModelGumbelVectorQuantizer", "Module", "True", "_compute_perplexity", "argmax", "batch_size", "class", "codevector_idx", "codevector_probs", "codevector_soft_dist", "codevectors", "codevectors_per_group", "def", "dim", "else", "float", "forward", "functional", "gumbel_softmax", "hard", "hidden_size", "hidden_states", "if", "new_zeros", "nn", "num_groups", "num_vars", "perplexity", "return", "scatter_", "self", "sequence_length", "shape", "softmax", "sum", "tau", "temperature", "torch", "training", "type_as", "unsqueeze", "view", "weight_proj"], "wavlm/modeling_wavlm.py:WavLMPreTrainedModel._init_weights": ["Conv1d", "GroupNorm", "LayerNorm", "Linear", "ModelFeatureProjection", "ModelGumbelVectorQuantizer", "ModelPositionalConvEmbedding", "ModelPreTrainedModel", "None", "PreTrainedModel", "_init_weights", "a", "b", "bias", "class", "codevectors", "config", "constant_", "conv", "def", "elif", "groups", "if", "in_channels", "in_features", "init", "initializer_range", "is", "isinstance", "k", "kaiming_normal_", "kernel_size", "math", "mean", "module", "nn", "normal_", "not", "ones_", "projection", "self", "sqrt", "std", "uniform_", "weight", "weight_proj", "zeros_"], "wavlm/modeling_wavlm.py:WavLMPreTrainedModel._get_feat_extract_output_lengths": ["LongTensor", "ModelPreTrainedModel", "None", "PreTrainedModel", "_", "_conv_out_length", "_get_feat_extract_output_lengths", "adapter_stride", "add_adapter", "bool", "class", "config", "conv_kernel", "conv_stride", "def", "div", "else", "floor", "for", "if", "in", "input_length", "input_lengths", "int", "is", "kernel_size", "num_adapter_layers", "range", "return", "rounding_mode", "self", "stride", "torch", "zip"], "wavlm/modeling_wavlm.py:WavLMPreTrainedModel._get_feature_vector_attention_mask": ["LongTensor", "ModelPreTrainedModel", "None", "PreTrainedModel", "_get_feat_extract_output_lengths", "_get_feature_vector_attention_mask", "add_adapter", "arange", "attention_mask", "batch_size", "bool", "class", "cumsum", "def", "device", "dim", "dtype", "feature_vector_length", "flip", "int", "long", "non_padded_lengths", "output_lengths", "return", "self", "shape", "to", "torch", "zeros"], "wavlm/modeling_wavlm.py:WavLMNoLayerNormConvLayer.__init__": ["ACT2FN", "Conv1d", "GradientCheckpointingLayer", "ModelNoLayerNormConvLayer", "__init__", "activation", "bias", "class", "config", "conv", "conv_bias", "conv_dim", "conv_kernel", "conv_stride", "def", "else", "feat_extract_activation", "if", "in_conv_dim", "kernel_size", "layer_id", "nn", "out_conv_dim", "self", "stride", "super"], "wavlm/modeling_wavlm.py:WavLMNoLayerNormConvLayer.forward": ["GradientCheckpointingLayer", "ModelNoLayerNormConvLayer", "activation", "class", "conv", "def", "forward", "hidden_states", "return", "self"], "wavlm/modeling_wavlm.py:WavLMLayerNormConvLayer.__init__": ["ACT2FN", "Conv1d", "GradientCheckpointingLayer", "LayerNorm", "ModelLayerNormConvLayer", "True", "__init__", "activation", "bias", "class", "config", "conv", "conv_bias", "conv_dim", "conv_kernel", "conv_stride", "def", "elementwise_affine", "else", "feat_extract_activation", "if", "in_conv_dim", "kernel_size", "layer_id", "layer_norm", "nn", "out_conv_dim", "self", "stride", "super"], "wavlm/modeling_wavlm.py:WavLMLayerNormConvLayer.forward": ["GradientCheckpointingLayer", "ModelLayerNormConvLayer", "activation", "class", "conv", "def", "forward", "hidden_states", "layer_norm", "return", "self", "transpose"], "wavlm/modeling_wavlm.py:WavLMGroupNormConvLayer.__init__": ["ACT2FN", "Conv1d", "GradientCheckpointingLayer", "GroupNorm", "ModelGroupNormConvLayer", "True", "__init__", "activation", "affine", "bias", "class", "config", "conv", "conv_bias", "conv_dim", "conv_kernel", "conv_stride", "def", "else", "feat_extract_activation", "if", "in_conv_dim", "kernel_size", "layer_id", "layer_norm", "nn", "num_channels", "num_groups", "out_conv_dim", "self", "stride", "super"], "wavlm/modeling_wavlm.py:WavLMGroupNormConvLayer.forward": ["GradientCheckpointingLayer", "ModelGroupNormConvLayer", "activation", "class", "conv", "def", "forward", "hidden_states", "layer_norm", "return", "self"], "wavlm/modeling_wavlm.py:WavLMFeatureEncoder.__init__": ["False", "ModelFeatureEncoder", "ModelGroupNormConvLayer", "ModelLayerNormConvLayer", "ModelNoLayerNormConvLayer", "Module", "ModuleList", "True", "ValueError", "__init__", "_requires_grad", "be", "but", "class", "config", "conv_layers", "def", "elif", "else", "f", "feat_extract_norm", "for", "gradient_checkpointing", "group", "has", "i", "if", "in", "is", "layer", "layer_id", "nn", "num_feat_extract_layers", "of", "one", "raise", "range", "self", "super", "to"], "wavlm/modeling_wavlm.py:WavLMFeatureEncoder._freeze_parameters": ["False", "ModelFeatureEncoder", "Module", "_freeze_parameters", "_requires_grad", "class", "def", "for", "in", "nn", "param", "parameters", "requires_grad", "self"], "wavlm/modeling_wavlm.py:WavLMFeatureEncoder.forward": ["ModelFeatureEncoder", "Module", "None", "True", "_requires_grad", "and", "class", "conv_layer", "conv_layers", "def", "for", "forward", "hidden_states", "if", "in", "input_values", "nn", "requires_grad", "return", "self", "training"], "wavlm/modeling_wavlm.py:WavLMAdapterLayer.__init__": ["Conv1d", "ModelAdapterLayer", "Module", "__init__", "adapter_kernel_size", "adapter_stride", "class", "config", "conv", "def", "nn", "output_hidden_size", "padding", "self", "stride", "super"], "wavlm/modeling_wavlm.py:WavLMAdapterLayer.forward": ["ModelAdapterLayer", "Module", "class", "conv", "def", "dim", "forward", "functional", "glu", "hidden_states", "nn", "return", "self"], "wavlm/modeling_wavlm.py:WavLMAdapter.__init__": ["LayerNorm", "Linear", "ModelAdapter", "ModelAdapterLayer", "Module", "ModuleList", "None", "_", "__init__", "class", "config", "def", "else", "for", "hidden_size", "if", "in", "layerdrop", "layers", "nn", "num_adapter_layers", "output_hidden_size", "proj", "proj_layer_norm", "range", "self", "super"], "wavlm/modeling_wavlm.py:WavLMAdapter.forward": ["ModelAdapter", "Module", "None", "and", "class", "def", "for", "forward", "hidden_states", "if", "in", "is", "layer", "layerdrop", "layerdrop_prob", "layers", "nn", "not", "np", "or", "proj", "proj_layer_norm", "random", "return", "self", "training", "transpose"], "wavlm/modeling_wavlm.py:_compute_mask_indices": ["False", "LongTensor", "None", "ValueError", "_", "_compute_mask_indices", "and", "append", "arange", "array", "attention_mask", "batch_size", "be", "bigger", "bool", "broadcast_to", "but", "choice", "compute_num_masked_span", "concatenate", "def", "detach", "dtype", "dummy_mask_idx", "else", "epsilon", "f", "float", "for", "got", "has", "if", "in", "input_length", "input_lengths", "int", "int32", "is", "item", "len", "mask_length", "mask_prob", "max", "max_num_masked_span", "min_masks", "ndarray", "not", "np", "num_masked_span", "offsets", "ones", "put_along_axis", "raise", "rand", "random", "range", "replace", "reshape", "return", "sequence_length", "shape", "smaller", "spec_aug_mask", "spec_aug_mask_idx", "spec_aug_mask_idxs", "sum", "than", "to", "tolist", "torch", "tuple", "zeros"], "wavlm/modeling_wavlm.py:WavLMModel.__init__": ["ModelAdapter", "ModelConfig", "ModelEncoder", "ModelEncoderStableLayerNorm", "ModelFeatureEncoder", "ModelFeatureProjection", "ModelModel", "ModelPreTrainedModel", "None", "Parameter", "Tensor", "__init__", "adapter", "add_adapter", "class", "config", "def", "do_stable_layer_norm", "else", "encoder", "feature_extractor", "feature_projection", "hidden_size", "if", "mask_feature_prob", "mask_time_prob", "masked_spec_embed", "nn", "or", "post_init", "self", "super", "torch", "uniform_"], "wavlm/modeling_wavlm.py:WavLMModel.freeze_feature_encoder": ["ModelModel", "ModelPreTrainedModel", "_freeze_parameters", "class", "def", "feature_extractor", "freeze_feature_encoder", "self"], "wavlm/modeling_wavlm.py:WavLMModel._mask_hidden_states": ["FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "True", "_compute_mask_indices", "_mask_hidden_states", "and", "apply_spec_augment", "attention_mask", "batch_size", "bool", "class", "config", "def", "device", "dtype", "elif", "expand", "getattr", "hidden_size", "hidden_states", "if", "is", "mask_feature_indices", "mask_feature_length", "mask_feature_min_masks", "mask_feature_prob", "mask_length", "mask_prob", "mask_time_indices", "mask_time_length", "mask_time_min_masks", "mask_time_prob", "masked_spec_embed", "min_masks", "not", "return", "self", "sequence_length", "size", "tensor", "to", "torch", "training"], "wavlm/modeling_wavlm.py:WavLMModel.forward": ["False", "FloatTensor", "ModelBaseModelOutput", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "_get_feature_vector_attention_mask", "_mask_hidden_states", "adapter", "add_adapter", "attention_mask", "attentions", "bool", "class", "config", "def", "else", "encoder", "encoder_outputs", "extract_features", "feature_extractor", "feature_projection", "forward", "hidden_states", "if", "input_values", "is", "kwargs", "last_hidden_state", "mask_time_indices", "not", "output_attentions", "output_hidden_states", "r", "return", "return_dict", "self", "shape", "torch", "transpose", "tuple", "use_return_dict"], "wavlm/modeling_wavlm.py:WavLMForCTC.__init__": ["Dropout", "Linear", "Model", "ModelForCTC", "ModelModel", "ModelPreTrainedModel", "None", "Please", "ValueError", "You", "__class__", "__init__", "a", "add_adapter", "and", "are", "as", "class", "config", "configuration", "def", "define", "does", "dropout", "else", "f", "final_dropout", "follows", "from_pretrained", "hasattr", "head", "hidden_size", "if", "instantiate", "is", "language", "lm_head", "model", "nn", "not", "of", "or", "output_hidden_size", "post_init", "r", "raise", "s", "self", "size", "str", "super", "target_lang", "that", "the", "to", "trying", "vocab_size", "vocabulary", "with", "your"], "wavlm/modeling_wavlm.py:WavLMForCTC.tie_weights": ["By", "Cannot", "ModelForCTC", "ModelPreTrainedModel", "None", "True", "ValueError", "adapter_attn_dim", "and", "class", "config", "def", "default", "defined", "elif", "eng", "f", "force_load", "getattr", "if", "info", "is", "kwargs", "load_adapter", "logger", "not", "pass", "raise", "self", "set", "target_lang", "tie_weights", "to"], "wavlm/modeling_wavlm.py:WavLMForCTC.freeze_feature_encoder": ["Model", "ModelForCTC", "ModelPreTrainedModel", "_freeze_parameters", "class", "def", "feature_extractor", "freeze_feature_encoder", "self"], "wavlm/modeling_wavlm.py:WavLMForCTC.freeze_base_model": ["False", "Model", "ModelForCTC", "ModelPreTrainedModel", "class", "def", "for", "freeze_base_model", "in", "param", "parameters", "requires_grad", "self"], "wavlm/modeling_wavlm.py:WavLMForCTC.forward": ["CausalLMOutput", "False", "Label", "Model", "ModelForCTC", "ModelPreTrainedModel", "None", "Tensor", "ValueError", "_HIDDEN_STATES_START_POSITION", "_get_feat_extract_output_lengths", "and", "attention_mask", "attentions", "backends", "be", "blank", "bool", "class", "config", "ctc_loss", "ctc_loss_reduction", "ctc_zero_infinity", "cudnn", "def", "dim", "dropout", "dtype", "else", "enabled", "f", "flags", "flattened_targets", "float32", "forward", "functional", "hidden_states", "if", "input_lengths", "input_values", "is", "kwargs", "labels", "labels_mask", "lm_head", "log_probs", "log_softmax", "logits", "long", "loss", "masked_select", "max", "must", "nn", "not", "ones_like", "output", "output_attentions", "output_hidden_states", "outputs", "pad_token_id", "r", "raise", "reduction", "return", "return_dict", "self", "sum", "target_lengths", "to", "torch", "transpose", "tuple", "use_return_dict", "values", "vocab_size", "with", "zero_infinity"], "wavlm/modeling_wavlm.py:WavLMForSequenceClassification.__init__": ["Linear", "Model", "ModelForSequenceClassification", "ModelModel", "ModelPreTrainedModel", "Parameter", "Sequence", "True", "ValueError", "__init__", "adapters", "add_adapter", "and", "class", "classification", "classifier", "classifier_proj_size", "config", "def", "does", "hasattr", "hidden_size", "if", "layer_weights", "nn", "not", "num_hidden_layers", "num_labels", "num_layers", "of", "ones", "post_init", "projector", "raise", "self", "super", "support", "the", "torch", "use", "use_weighted_layer_sum"], "wavlm/modeling_wavlm.py:WavLMForSequenceClassification.freeze_feature_encoder": ["Model", "ModelForSequenceClassification", "ModelPreTrainedModel", "_freeze_parameters", "class", "def", "feature_extractor", "freeze_feature_encoder", "self"], "wavlm/modeling_wavlm.py:WavLMForSequenceClassification.freeze_base_model": ["False", "Model", "ModelForSequenceClassification", "ModelPreTrainedModel", "class", "def", "for", "freeze_base_model", "in", "param", "parameters", "requires_grad", "self"], "wavlm/modeling_wavlm.py:WavLMForSequenceClassification.forward": ["CrossEntropyLoss", "Model", "ModelForSequenceClassification", "ModelPreTrainedModel", "None", "SequenceClassifierOutput", "Tensor", "True", "_HIDDEN_STATES_START_POSITION", "_get_feature_vector_attention_mask", "attention_mask", "attentions", "bool", "class", "classifier", "config", "def", "dim", "else", "expand_padding_mask", "forward", "functional", "hidden_states", "if", "input_values", "is", "kwargs", "labels", "layer_weights", "logits", "loss", "loss_fct", "mean", "nn", "norm_weights", "not", "num_labels", "output", "output_attentions", "output_hidden_states", "outputs", "padding_mask", "pooled_output", "projector", "r", "repeat", "return", "return_dict", "self", "shape", "softmax", "stack", "sum", "torch", "tuple", "unsqueeze", "use_return_dict", "use_weighted_layer_sum", "view"], "wavlm/modeling_wavlm.py:WavLMForAudioFrameClassification.__init__": ["Audio", "Linear", "Model", "ModelForAudioFrameClassification", "ModelModel", "ModelPreTrainedModel", "Parameter", "True", "ValueError", "__init__", "adapters", "add_adapter", "and", "class", "classification", "classifier", "config", "def", "does", "frame", "hasattr", "hidden_size", "if", "layer_weights", "nn", "not", "num_hidden_layers", "num_labels", "num_layers", "of", "ones", "post_init", "raise", "self", "super", "support", "the", "torch", "use", "use_weighted_layer_sum"], "wavlm/modeling_wavlm.py:WavLMForAudioFrameClassification.freeze_feature_encoder": ["Model", "ModelForAudioFrameClassification", "ModelPreTrainedModel", "_freeze_parameters", "class", "def", "feature_extractor", "freeze_feature_encoder", "self"], "wavlm/modeling_wavlm.py:WavLMForAudioFrameClassification.freeze_base_model": ["False", "Model", "ModelForAudioFrameClassification", "ModelPreTrainedModel", "class", "def", "for", "freeze_base_model", "in", "param", "parameters", "requires_grad", "self"], "wavlm/modeling_wavlm.py:WavLMForAudioFrameClassification.forward": ["CrossEntropyLoss", "Model", "ModelForAudioFrameClassification", "ModelPreTrainedModel", "None", "Tensor", "TokenClassifierOutput", "True", "_HIDDEN_STATES_START_POSITION", "argmax", "attention_mask", "attentions", "axis", "bool", "class", "classifier", "config", "def", "dim", "else", "forward", "functional", "hidden_states", "if", "input_values", "is", "kwargs", "labels", "layer_weights", "logits", "loss", "loss_fct", "nn", "norm_weights", "not", "num_labels", "output", "output_attentions", "output_hidden_states", "outputs", "r", "return", "return_dict", "self", "softmax", "stack", "sum", "torch", "tuple", "use_return_dict", "use_weighted_layer_sum", "view"], "wavlm/modeling_wavlm.py:AMSoftmaxLoss.__init__": ["CrossEntropyLoss", "Model", "Module", "Parameter", "True", "__init__", "class", "def", "input_dim", "loss", "margin", "nn", "num_labels", "randn", "requires_grad", "scale", "self", "super", "torch", "weight"], "wavlm/modeling_wavlm.py:AMSoftmaxLoss.forward": ["Model", "Module", "bool", "class", "cos_theta", "def", "dim", "flatten", "forward", "functional", "hidden_states", "labels", "logits", "loss", "margin", "mm", "nn", "normalize", "num_labels", "one_hot", "onehot", "psi", "return", "scale", "self", "torch", "weight", "where"], "wavlm/modeling_wavlm.py:TDNNLayer.__init__": ["Linear", "Model", "Module", "ReLU", "__init__", "activation", "class", "config", "def", "dilation", "else", "if", "in_conv_dim", "kernel", "kernel_size", "layer_id", "nn", "out_conv_dim", "self", "super", "tdnn_dilation", "tdnn_dim", "tdnn_kernel"], "wavlm/modeling_wavlm.py:TDNNLayer.forward": ["Detected", "LoRA", "LoraLayer", "Model", "Module", "Tensor", "You", "activation", "applied", "be", "bias", "class", "conv1d", "def", "dilation", "due", "exclude", "forward", "from", "functional", "hidden_states", "if", "in_conv_dim", "is_peft_available", "isinstance", "kernel", "kernel_size", "modules", "nn", "on", "optimization", "out_conv_dim", "return", "s", "self", "should", "t", "target", "to", "torch", "transpose", "view", "warn", "warnings", "weight", "weights", "won"], "wavlm/modeling_wavlm.py:WavLMForXVector.__init__": ["AMSoftmaxLoss", "Linear", "Model", "ModelForXVector", "ModelModel", "ModelPreTrainedModel", "ModuleList", "Parameter", "TDNNLayer", "__init__", "class", "classifier", "config", "def", "feature_extractor", "for", "hidden_size", "i", "if", "in", "layer_weights", "len", "nn", "num_hidden_layers", "num_labels", "num_layers", "objective", "ones", "post_init", "projector", "range", "self", "super", "tdnn", "tdnn_dim", "tdnn_layers", "torch", "use_weighted_layer_sum", "xvector_output_dim"], "wavlm/modeling_wavlm.py:WavLMForXVector.freeze_feature_encoder": ["Model", "ModelForXVector", "ModelPreTrainedModel", "_freeze_parameters", "class", "def", "feature_extractor", "freeze_feature_encoder", "self"], "wavlm/modeling_wavlm.py:WavLMForXVector.freeze_base_model": ["False", "Model", "ModelForXVector", "ModelPreTrainedModel", "class", "def", "for", "freeze_base_model", "in", "param", "parameters", "requires_grad", "self"], "wavlm/modeling_wavlm.py:WavLMForXVector._get_tdnn_output_lengths": ["LongTensor", "ModelForXVector", "ModelPreTrainedModel", "_conv_out_length", "_get_tdnn_output_lengths", "class", "config", "def", "for", "in", "input_length", "input_lengths", "int", "kernel_size", "return", "self", "stride", "tdnn_kernel", "torch"], "wavlm/modeling_wavlm.py:WavLMForXVector.forward": ["Model", "ModelForXVector", "ModelPreTrainedModel", "None", "Tensor", "True", "XVectorOutput", "_HIDDEN_STATES_START_POSITION", "_get_feat_extract_output_lengths", "_get_tdnn_output_lengths", "append", "attention_mask", "attentions", "bool", "cat", "class", "classifier", "config", "def", "dim", "else", "embeddings", "enumerate", "feat_extract_output_lengths", "feature_extractor", "for", "forward", "functional", "hidden_states", "i", "if", "in", "input_values", "is", "kwargs", "labels", "layer_weights", "length", "logits", "loss", "mean", "mean_features", "nn", "norm_weights", "not", "objective", "output", "output_attentions", "output_embeddings", "output_hidden_states", "outputs", "projector", "r", "return", "return_dict", "self", "softmax", "stack", "statistic_pooling", "std", "std_features", "sum", "tdnn", "tdnn_layer", "tdnn_output_lengths", "torch", "tuple", "use_return_dict", "use_weighted_layer_sum", "view"], "whisper/modeling_whisper.py:sinusoids": ["Model", "Number", "Tensor", "ValueError", "arange", "be", "by", "cat", "channels", "cos", "def", "dim", "divisible", "embeddings", "exp", "f", "float", "for", "got", "has", "if", "int", "inv_timescales", "length", "log", "log_timescale_increment", "math", "max_timescale", "of", "positional", "raise", "return", "scaled_time", "sin", "sinusoidal", "to", "torch", "view"], "whisper/modeling_whisper.py:shift_tokens_right": ["Model_tokens_right", "Modeled_input_ids", "None", "Tensor", "ValueError", "be", "clone", "config", "decoder_start_token_id", "def", "defined", "has", "if", "input_ids", "int", "is", "masked_fill_", "model", "new_zeros", "pad_token_id", "raise", "return", "self", "shape", "to", "torch"], "whisper/modeling_whisper.py:_compute_mask_indices": ["False", "LongTensor", "None", "ValueError", "_", "_compute_mask_indices", "and", "append", "arange", "array", "attention_mask", "batch_size", "be", "bigger", "bool", "broadcast_to", "but", "choice", "compute_num_masked_span", "concatenate", "def", "detach", "dtype", "dummy_mask_idx", "else", "epsilon", "f", "float", "for", "got", "has", "if", "in", "input_length", "input_lengths", "int", "int32", "is", "item", "len", "mask_length", "mask_prob", "max", "max_num_masked_span", "min_masks", "ndarray", "not", "np", "num_masked_span", "offsets", "ones", "put_along_axis", "raise", "rand", "random", "range", "replace", "reshape", "return", "sequence_length", "shape", "smaller", "spec_aug_mask", "spec_aug_mask_idx", "spec_aug_mask_idxs", "sum", "than", "to", "tolist", "torch", "tuple", "zeros"], "whisper/modeling_whisper.py:WhisperPositionalEmbedding.__init__": ["Embedding", "ModelPositionalEmbedding", "None", "__init__", "class", "def", "embedding_dim", "int", "nn", "num_positions", "padding_idx", "self", "super"], "whisper/modeling_whisper.py:WhisperPositionalEmbedding.forward": ["Embedding", "ModelPositionalEmbedding", "None", "class", "def", "else", "forward", "if", "input_ids", "is", "nn", "past_key_values_length", "position_ids", "return", "self", "shape", "weight"], "whisper/modeling_whisper.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "and", "attention_mask", "attn_output", "attn_weights", "contiguous", "def", "dim", "dropout", "float", "functional", "if", "is", "key", "kwargs", "matmul", "module", "ndim", "nn", "not", "p", "query", "return", "scaling", "shape", "size", "softmax", "torch", "training", "transpose", "value"], "whisper/modeling_whisper.py:WhisperAttention.__init__": ["False", "Instantiating", "Linear", "ModelAttention", "ModelConfig", "Module", "None", "Please", "True", "ValueError", "__class__", "__init__", "__name__", "a", "and", "be", "bias", "bool", "by", "caching", "call", "class", "config", "creating", "decoder", "def", "divisible", "dropout", "during", "embed_dim", "errors", "f", "float", "forward", "got", "head_dim", "if", "int", "is", "is_causal", "is_decoder", "k_proj", "layer_idx", "logger", "make", "must", "nn", "not", "num_heads", "out_proj", "passing", "provide", "q_proj", "raise", "recommended", "scaling", "self", "super", "sure", "the", "this", "to", "used", "v_proj", "warning_once", "when", "will", "without"], "whisper/modeling_whisper.py:WhisperAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "EncoderDecoderCache", "False", "FlashAttentionKwargs", "ModelAttention", "Module", "None", "Tensor", "True", "Unpack", "_attn_implementation", "and", "attention_interface", "attention_mask", "attn_output", "attn_weights", "bool", "bsz", "cache_position", "class", "config", "contiguous", "cross_attention_cache", "current_states", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "get", "head_dim", "hidden_states", "if", "is", "is_cross_attention", "is_updated", "isinstance", "k_proj", "key_states", "key_value_states", "keys", "kwargs", "layer_idx", "layers", "nn", "not", "num_heads", "out_proj", "output_attentions", "past_key_values", "q_input_shape", "q_proj", "query_states", "reshape", "return", "scaling", "self", "self_attention_cache", "shape", "tgt_len", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "values", "view"], "whisper/modeling_whisper.py:WhisperEncoderLayer.__init__": ["ACT2FN", "GradientCheckpointingLayer", "LayerNorm", "Linear", "ModelAttention", "ModelConfig", "ModelEncoderLayer", "__init__", "activation_dropout", "activation_fn", "activation_function", "attention_dropout", "class", "config", "d_model", "def", "dropout", "embed_dim", "encoder_attention_heads", "encoder_ffn_dim", "fc1", "fc2", "final_layer_norm", "nn", "num_heads", "self", "self_attn", "self_attn_layer_norm", "super"], "whisper/modeling_whisper.py:WhisperEncoderLayer.forward": ["False", "GradientCheckpointingLayer", "ModelEncoderLayer", "Tensor", "activation_dropout", "activation_fn", "attention_mask", "attn_weights", "bool", "clamp", "clamp_value", "class", "def", "dropout", "dtype", "fc1", "fc2", "final_layer_norm", "finfo", "float16", "forward", "functional", "hidden_states", "if", "max", "min", "nn", "output_attentions", "p", "residual", "return", "self", "self_attn", "self_attn_layer_norm", "torch", "training"], "whisper/modeling_whisper.py:WhisperDecoderLayer.__init__": ["ACT2FN", "GradientCheckpointingLayer", "LayerNorm", "Linear", "ModelAttention", "ModelConfig", "ModelDecoderLayer", "None", "True", "__init__", "activation_dropout", "activation_fn", "activation_function", "attention_dropout", "class", "config", "d_model", "decoder_attention_heads", "decoder_ffn_dim", "def", "dropout", "embed_dim", "encoder_attn", "encoder_attn_layer_norm", "fc1", "fc2", "final_layer_norm", "int", "is_causal", "is_decoder", "layer_idx", "nn", "num_heads", "self", "self_attn", "self_attn_layer_norm", "super"], "whisper/modeling_whisper.py:WhisperDecoderLayer.forward": ["EncoderDecoderCache", "False", "GradientCheckpointingLayer", "LongTensor", "ModelDecoderLayer", "None", "Tensor", "True", "activation_dropout", "activation_fn", "attention_mask", "bool", "cache_position", "class", "cross_attn_weights", "def", "dropout", "encoder_attention_mask", "encoder_attn", "encoder_attn_layer_norm", "encoder_hidden_states", "fc1", "fc2", "final_layer_norm", "forward", "functional", "hidden_states", "if", "is", "key_value_states", "nn", "not", "output_attentions", "outputs", "p", "past_key_values", "residual", "return", "self", "self_attn", "self_attn_layer_norm", "self_attn_weights", "torch", "training", "use_cache"], "whisper/modeling_whisper.py:WhisperPreTrainedModel._init_weights": ["ModelEncoder", "ModelForAudioClassification", "ModelPreTrainedModel", "PreTrainedModel", "_init_weights", "class", "config", "constant_", "copy_", "def", "elif", "embed_positions", "if", "init", "isinstance", "layer_weights", "module", "num_hidden_layers", "self", "shape", "sinusoids", "super", "use_weighted_layer_sum", "weight"], "whisper/modeling_whisper.py:WhisperPreTrainedModel._get_feat_extract_output_lengths": ["LongTensor", "ModelPreTrainedModel", "PreTrainedModel", "_get_feat_extract_output_lengths", "class", "def", "input_lengths", "return", "self", "torch"], "whisper/modeling_whisper.py:WhisperEncoder.__init__": ["Conv1d", "Embedding", "False", "LayerNorm", "ModelConfig", "ModelEncoder", "ModelEncoderLayer", "ModelPreTrainedModel", "ModuleList", "_", "__init__", "class", "config", "conv1", "conv2", "d_model", "def", "dropout", "else", "embed_dim", "embed_positions", "embed_scale", "encoder_layerdrop", "encoder_layers", "for", "gradient_checkpointing", "if", "in", "kernel_size", "layer_norm", "layerdrop", "layers", "math", "max_source_positions", "nn", "num_mel_bins", "pad_token_id", "padding", "padding_idx", "post_init", "range", "requires_grad_", "scale_embedding", "self", "sqrt", "stride", "super"], "whisper/modeling_whisper.py:WhisperEncoder._freeze_parameters": ["False", "ModelEncoder", "ModelPreTrainedModel", "_freeze_parameters", "_requires_grad", "class", "def", "for", "in", "param", "parameters", "requires_grad", "self"], "whisper/modeling_whisper.py:WhisperEncoder.get_input_embeddings": ["ModelEncoder", "ModelPreTrainedModel", "Module", "class", "conv1", "def", "get_input_embeddings", "nn", "return", "self"], "whisper/modeling_whisper.py:WhisperEncoder.set_input_embeddings": ["ModelEncoder", "ModelPreTrainedModel", "Module", "class", "conv1", "def", "nn", "self", "set_input_embeddings", "value"], "whisper/modeling_whisper.py:WhisperEncoder.forward": ["BaseModelOutput", "False", "Make", "Model", "ModelEncoder", "ModelPreTrainedModel", "None", "True", "ValueError", "all_attentions", "all_positions", "arange", "attention_mask", "attentions", "be", "but", "class", "config", "conv1", "conv2", "def", "device", "dropout", "dropout_probability", "else", "embed_positions", "encoder_layer", "encoder_states", "enumerate", "expected_seq_length", "expects", "f", "features", "for", "forward", "found", "functional", "gelu", "hidden_states", "idx", "if", "in", "input", "input_features", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layer_norm", "layer_outputs", "layerdrop", "layers", "length", "max_source_positions", "mel", "nn", "not", "num_embeddings", "of", "output_attentions", "output_hidden_states", "p", "pad", "permute", "r", "raise", "rand", "return", "return_dict", "self", "shape", "stride", "sure", "the", "to", "to_drop", "torch", "training", "tuple", "use_return_dict", "v"], "whisper/modeling_whisper.py:WhisperDecoder.__init__": ["Embedding", "False", "LayerNorm", "ModelConfig", "ModelDecoder", "ModelDecoderLayer", "ModelPositionalEmbedding", "ModelPreTrainedModel", "ModuleList", "__init__", "class", "config", "d_model", "decoder_layerdrop", "decoder_layers", "def", "dropout", "else", "embed_positions", "embed_scale", "embed_tokens", "for", "gradient_checkpointing", "if", "in", "layer_idx", "layer_norm", "layerdrop", "layers", "math", "max_source_positions", "max_target_positions", "nn", "pad_token_id", "padding_idx", "post_init", "range", "scale_embedding", "self", "sqrt", "super", "vocab_size"], "whisper/modeling_whisper.py:WhisperDecoder.forward": ["BaseModelOutputWithPastAndCrossAttentions", "DynamicCache", "EncoderDecoderCache", "False", "ModelDecoder", "ModelPreTrainedModel", "None", "Setting", "True", "ValueError", "You", "all_cross_attentions", "all_hidden_states", "all_self_attns", "and", "arange", "at", "attention_mask", "attentions", "both", "cache_position", "cannot", "causal_mask", "checkpointing", "class", "config", "continue", "create_causal_mask", "cross_attentions", "decoder_input_ids", "decoder_inputs_embeds", "decoder_layer", "def", "device", "dropout", "dropout_probability", "either", "elif", "else", "embed_positions", "embed_tokens", "encoder_attention_mask", "encoder_hidden_states", "enumerate", "for", "forward", "functional", "get_seq_length", "gradient", "gradient_checkpointing", "have", "hidden_states", "idx", "if", "in", "incompatible", "input_embeds", "input_ids", "input_shape", "inputs_embeds", "is", "is_encoder_decoder", "kwargs", "last_hidden_state", "layer_norm", "layer_outputs", "layerdrop", "layers", "logger", "next_cache", "nn", "not", "or", "output_attentions", "output_hidden_states", "p", "past_key_values", "past_key_values_length", "position_ids", "positions", "r", "raise", "rand", "repeat", "return", "return_dict", "same", "self", "size", "specify", "the", "time", "to", "torch", "training", "tuple", "unsqueeze", "use_cache", "use_return_dict", "v", "view", "warning_once", "with"], "whisper/modeling_whisper.py:WhisperModel.__init__": ["ModelConfig", "ModelDecoder", "ModelEncoder", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "config", "decoder", "def", "encoder", "post_init", "self", "super"], "whisper/modeling_whisper.py:WhisperModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "decoder", "def", "embed_tokens", "get_input_embeddings", "return", "self"], "whisper/modeling_whisper.py:WhisperModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "decoder", "def", "embed_tokens", "self", "set_input_embeddings", "value"], "whisper/modeling_whisper.py:WhisperModel.freeze_encoder": ["ModelModel", "ModelPreTrainedModel", "_freeze_parameters", "class", "def", "encoder", "freeze_encoder", "self"], "whisper/modeling_whisper.py:WhisperModel._mask_input_features": ["FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "True", "_compute_mask_indices", "_mask_input_features", "and", "apply_spec_augment", "attention_mask", "batch_size", "bool", "class", "config", "def", "device", "dtype", "expand", "getattr", "hidden_size", "if", "input_features", "mask_feature_indices", "mask_feature_length", "mask_feature_min_masks", "mask_feature_prob", "mask_length", "mask_prob", "mask_time_indices", "mask_time_length", "mask_time_min_masks", "mask_time_prob", "min_masks", "not", "return", "self", "sequence_length", "size", "tensor", "torch", "training"], "whisper/modeling_whisper.py:WhisperModel.forward": ["BaseModelOutput", "Cache", "FloatTensor", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Seq2SeqModelOutput", "Tensor", "_mask_input_features", "and", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "cross_attentions", "decoder", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_input_ids", "decoder_inputs_embeds", "decoder_outputs", "decoder_position_ids", "def", "elif", "else", "encoder", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "forward", "hidden_states", "if", "input_features", "input_ids", "inputs_embeds", "is", "isinstance", "kwargs", "last_hidden_state", "len", "not", "output_attentions", "output_hidden_states", "past_key_values", "position_ids", "r", "return", "return_dict", "self", "torch", "tuple", "use_cache", "use_return_dict"], "whisper/modeling_whisper.py:WhisperForConditionalGeneration.__init__": ["False", "Linear", "ModelConfig", "ModelForConditionalGeneration", "ModelGenerationMixin", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "d_model", "def", "max_target_positions", "model", "nn", "post_init", "proj_out", "self", "super", "vocab_size"], "whisper/modeling_whisper.py:WhisperForConditionalGeneration.get_output_embeddings": ["ModelForConditionalGeneration", "ModelGenerationMixin", "ModelPreTrainedModel", "class", "def", "get_output_embeddings", "proj_out", "return", "self"], "whisper/modeling_whisper.py:WhisperForConditionalGeneration.set_output_embeddings": ["ModelForConditionalGeneration", "ModelGenerationMixin", "ModelPreTrainedModel", "class", "def", "new_embeddings", "proj_out", "self", "set_output_embeddings"], "whisper/modeling_whisper.py:WhisperForConditionalGeneration.get_input_embeddings": ["ModelForConditionalGeneration", "ModelGenerationMixin", "ModelPreTrainedModel", "Module", "class", "def", "get_input_embeddings", "model", "nn", "return", "self"], "whisper/modeling_whisper.py:WhisperForConditionalGeneration.freeze_encoder": ["ModelForConditionalGeneration", "ModelGenerationMixin", "ModelPreTrainedModel", "_freeze_parameters", "class", "def", "encoder", "freeze_encoder", "model", "self"], "whisper/modeling_whisper.py:WhisperForConditionalGeneration.forward": ["Cache", "CrossEntropyLoss", "FloatTensor", "Labels", "LongTensor", "ModelForConditionalGeneration", "ModelGenerationMixin", "ModelPreTrainedModel", "None", "Seq2SeqLMOutput", "Tensor", "ValueError", "allowed", "and", "attention_mask", "bool", "cache_position", "cannot", "class", "config", "cross_attentions", "decoder_attention_mask", "decoder_attentions", "decoder_hidden_states", "decoder_input_ids", "decoder_inputs_embeds", "decoder_position_ids", "decoder_start_token_id", "def", "device", "else", "encoder_attentions", "encoder_hidden_states", "encoder_last_hidden_state", "encoder_outputs", "exceed", "f", "forward", "if", "input_features", "is", "kwargs", "labels", "length", "lm_logits", "logits", "loss", "loss_fct", "max_target_positions", "maximum", "model", "not", "of", "output", "output_attentions", "output_hidden_states", "outputs", "pad_token_id", "past_key_values", "proj_out", "r", "raise", "reshape", "return", "return_dict", "self", "sequence", "shape", "shift_tokens_right", "the", "to", "tokens", "torch", "tuple", "use_cache", "use_return_dict", "view", "vocab_size"], "whisper/modeling_whisper.py:WhisperDecoderWrapper.__init__": ["False", "ModelDecoder", "ModelDecoderWrapper", "ModelPreTrainedModel", "__init__", "class", "config", "decoder", "def", "is_encoder_decoder", "post_init", "self", "super"], "whisper/modeling_whisper.py:WhisperDecoderWrapper.get_input_embeddings": ["ModelDecoderWrapper", "ModelPreTrainedModel", "class", "decoder", "def", "embed_tokens", "get_input_embeddings", "return", "self"], "whisper/modeling_whisper.py:WhisperDecoderWrapper.set_input_embeddings": ["ModelDecoderWrapper", "ModelPreTrainedModel", "class", "decoder", "def", "embed_tokens", "self", "set_input_embeddings", "value"], "whisper/modeling_whisper.py:WhisperDecoderWrapper.forward": ["ModelDecoderWrapper", "ModelPreTrainedModel", "args", "class", "decoder", "def", "forward", "kwargs", "return", "self"], "whisper/modeling_whisper.py:WhisperForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "ModelDecoderWrapper", "ModelForCausalLM", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "is_encoder_decoder", "model", "nn", "post_init", "proj_out", "self", "super", "vocab_size"], "whisper/modeling_whisper.py:WhisperForCausalLM.get_output_embeddings": ["GenerationMixin", "ModelForCausalLM", "ModelPreTrainedModel", "class", "def", "get_output_embeddings", "proj_out", "return", "self"], "whisper/modeling_whisper.py:WhisperForCausalLM.set_output_embeddings": ["GenerationMixin", "ModelForCausalLM", "ModelPreTrainedModel", "class", "def", "new_embeddings", "proj_out", "self", "set_output_embeddings"], "whisper/modeling_whisper.py:WhisperForCausalLM.get_input_embeddings": ["GenerationMixin", "ModelForCausalLM", "ModelPreTrainedModel", "Module", "class", "def", "get_input_embeddings", "model", "nn", "return", "self"], "whisper/modeling_whisper.py:WhisperForCausalLM.set_input_embeddings": ["GenerationMixin", "ModelForCausalLM", "ModelPreTrainedModel", "class", "def", "model", "self", "set_input_embeddings", "value"], "whisper/modeling_whisper.py:WhisperForCausalLM.forward": ["BaseModelOutput", "Cache", "CausalLMOutputWithCrossAttentions", "CrossEntropyLoss", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForCausalLM", "ModelPreTrainedModel", "None", "Tensor", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "cross_attentions", "decoder", "def", "device", "else", "encoder_hidden_states", "encoder_outputs", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "isinstance", "kwargs", "labels", "list", "logits", "loss", "loss_fct", "model", "not", "output", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "proj_out", "r", "return", "return_dict", "self", "to", "torch", "tuple", "use_cache", "use_return_dict", "view", "vocab_size"], "whisper/modeling_whisper.py:WhisperForAudioClassification.__init__": ["Linear", "ModelEncoder", "ModelForAudioClassification", "ModelPreTrainedModel", "Parameter", "__init__", "class", "classifier", "classifier_proj_size", "config", "def", "encoder", "hidden_size", "if", "layer_weights", "nn", "num_hidden_layers", "num_labels", "num_layers", "ones", "post_init", "projector", "self", "super", "torch", "use_weighted_layer_sum"], "whisper/modeling_whisper.py:WhisperForAudioClassification.freeze_encoder": ["ModelForAudioClassification", "ModelPreTrainedModel", "_freeze_parameters", "class", "def", "encoder", "freeze_encoder", "self"], "whisper/modeling_whisper.py:WhisperForAudioClassification.get_input_embeddings": ["ModelForAudioClassification", "ModelPreTrainedModel", "Module", "class", "def", "encoder", "get_input_embeddings", "nn", "return", "self"], "whisper/modeling_whisper.py:WhisperForAudioClassification.set_input_embeddings": ["ModelForAudioClassification", "ModelPreTrainedModel", "Module", "class", "def", "encoder", "nn", "self", "set_input_embeddings", "value"], "whisper/modeling_whisper.py:WhisperForAudioClassification.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "ModelForAudioClassification", "ModelPreTrainedModel", "None", "SequenceClassifierOutput", "Tensor", "True", "_HIDDEN_STATES_START_POSITION", "attentions", "bool", "class", "classifier", "config", "def", "device", "dim", "elif", "else", "encoder", "encoder_outputs", "forward", "functional", "hidden_states", "if", "input_features", "is", "kwargs", "labels", "layer_weights", "logits", "loss", "loss_fct", "mean", "nn", "norm_weights", "not", "num_labels", "output", "output_attentions", "output_hidden_states", "pooled_output", "projector", "r", "return", "return_dict", "self", "softmax", "stack", "sum", "to", "torch", "tuple", "use_return_dict", "use_weighted_layer_sum", "view"], "x_clip/modeling_x_clip.py:contrastive_loss": ["Model_loss", "Tensor", "arange", "cross_entropy", "def", "device", "functional", "len", "logits", "nn", "return", "torch"], "x_clip/modeling_x_clip.py:x_clip_loss": ["Model_loss", "Tensor", "caption_loss", "contrastive_loss", "def", "image_loss", "return", "similarity", "t", "torch"], "x_clip/modeling_x_clip.py:XCLIPOutput.to_tuple": ["Any", "Model", "ModelOutput", "class", "def", "else", "for", "getattr", "if", "in", "k", "keys", "mit_output", "not", "return", "self", "text_model_output", "to_tuple", "tuple", "vision_model_output"], "x_clip/modeling_x_clip.py:XCLIPVisionEmbeddings.__init__": ["Conv2d", "Embedding", "False", "Model", "ModelVisionConfig", "Module", "Parameter", "__init__", "arange", "bias", "class", "class_embedding", "config", "def", "embed_dim", "expand", "hidden_size", "image_size", "in_channels", "kernel_size", "nn", "num_channels", "num_patches", "num_positions", "out_channels", "patch_embedding", "patch_size", "persistent", "position_embedding", "position_ids", "randn", "register_buffer", "self", "stride", "super", "torch"], "x_clip/modeling_x_clip.py:XCLIPVisionEmbeddings.interpolate_pos_encoding": ["False", "Model", "Module", "Tensor", "align_corners", "and", "bicubic", "cat", "class", "class_pos_embed", "def", "dim", "embeddings", "functional", "height", "if", "int", "interpolate", "interpolate_pos_encoding", "is_tracing", "jit", "mode", "new_height", "new_width", "nn", "not", "num_patches", "num_positions", "patch_pos_embed", "patch_size", "permute", "position_embedding", "position_ids", "reshape", "return", "self", "shape", "size", "sqrt_num_positions", "torch", "torch_int", "unsqueeze", "view", "weight", "width"], "x_clip/modeling_x_clip.py:XCLIPVisionEmbeddings.forward": ["False", "FloatTensor", "Input", "Model", "Module", "Tensor", "ValueError", "_", "and", "batch_size", "cat", "class", "class_embedding", "class_embeds", "def", "dim", "doesn", "dtype", "else", "embeddings", "expand", "f", "flatten", "forward", "height", "if", "image", "image_size", "interpolate_pos_encoding", "match", "model", "nn", "not", "or", "patch_embedding", "patch_embeds", "pixel_values", "position_embedding", "position_ids", "raise", "return", "self", "shape", "size", "t", "target_dtype", "to", "torch", "transpose", "weight", "width"], "x_clip/modeling_x_clip.py:XCLIPTextEmbeddings.__init__": ["Embedding", "False", "Model", "ModelTextConfig", "Module", "__init__", "arange", "class", "config", "def", "embed_dim", "expand", "hidden_size", "max_position_embeddings", "nn", "persistent", "position_embedding", "position_ids", "register_buffer", "self", "super", "token_embedding", "torch", "vocab_size"], "x_clip/modeling_x_clip.py:XCLIPTextEmbeddings.forward": ["FloatTensor", "LongTensor", "Model", "Module", "None", "Sequence", "Tensor", "ValueError", "and", "be", "class", "def", "else", "embeddings", "f", "forward", "got", "if", "input_ids", "inputs_embeds", "is", "length", "less", "max_position_embedding", "max_position_embeddings", "must", "nn", "not", "position_embedding", "position_embeddings", "position_ids", "raise", "return", "self", "seq_length", "sequence", "shape", "than", "token_embedding", "torch", "weight"], "x_clip/modeling_x_clip.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "attention_mask", "attn_output", "attn_weights", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "softmax", "to", "torch", "training", "transpose", "value"], "x_clip/modeling_x_clip.py:XCLIPAttention.__init__": ["False", "Linear", "Model", "Module", "ValueError", "__init__", "and", "attention_dropout", "be", "by", "class", "config", "def", "divisible", "dropout", "embed_dim", "f", "got", "head_dim", "hidden_size", "if", "is_causal", "k_proj", "must", "nn", "num_attention_heads", "num_heads", "out_proj", "q_proj", "raise", "scale", "self", "super", "v_proj"], "x_clip/modeling_x_clip.py:XCLIPAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "False", "Model", "Module", "None", "Tensor", "_attn_implementation", "and", "attention_interface", "attention_mask", "attn_output", "attn_weights", "batch_size", "bool", "causal_attention_mask", "class", "config", "contiguous", "def", "dropout", "eager", "eager_attention_forward", "elif", "else", "embed_dim", "flash_attention_2", "forward", "head_dim", "hidden_states", "if", "is", "is_causal", "k_proj", "keys", "nn", "not", "num_heads", "out_proj", "output_attentions", "q_proj", "queries", "reshape", "return", "scale", "scaling", "self", "seq_length", "shape", "torch", "training", "transpose", "tuple", "v_proj", "values", "view"], "x_clip/modeling_x_clip.py:XCLIPMLP.__init__": ["ACT2FN", "Linear", "Model", "Module", "__init__", "activation_fn", "class", "config", "def", "fc1", "fc2", "hidden_act", "hidden_size", "intermediate_size", "nn", "self", "super"], "x_clip/modeling_x_clip.py:XCLIPMLP.forward": ["Model", "Module", "Tensor", "activation_fn", "class", "def", "fc1", "fc2", "forward", "hidden_states", "nn", "return", "self", "torch"], "x_clip/modeling_x_clip.py:XCLIPEncoderLayer.__init__": ["GradientCheckpointingLayer", "LayerNorm", "Model", "ModelAttention", "ModelConfig", "ModelMLP", "__init__", "class", "config", "def", "embed_dim", "eps", "hidden_size", "layer_norm1", "layer_norm2", "layer_norm_eps", "mlp", "nn", "self", "self_attn", "super"], "x_clip/modeling_x_clip.py:XCLIPEncoderLayer.forward": ["False", "FloatTensor", "GradientCheckpointingLayer", "Model", "None", "Tensor", "attention_mask", "attn_weights", "bool", "causal_attention_mask", "class", "def", "forward", "hidden_states", "if", "layer_norm1", "layer_norm2", "mlp", "output_attentions", "outputs", "residual", "return", "self", "self_attn", "torch", "tuple"], "x_clip/modeling_x_clip.py:drop_path": ["False", "Model_path", "Model_prob", "Tensor", "bool", "def", "device", "div", "dtype", "float", "floor_", "if", "input", "keep_prob", "ndim", "not", "or", "output", "rand", "random_tensor", "return", "shape", "torch", "training"], "x_clip/modeling_x_clip.py:XCLIPDropPath.__init__": ["Model", "Module", "None", "__init__", "class", "def", "drop_prob", "float", "nn", "self", "super"], "x_clip/modeling_x_clip.py:XCLIPDropPath.forward": ["Model", "Module", "Tensor", "class", "def", "drop_path", "drop_prob", "forward", "hidden_states", "nn", "return", "self", "torch", "training"], "x_clip/modeling_x_clip.py:XCLIPDropPath.extra_repr": ["Model", "Module", "class", "def", "drop_prob", "extra_repr", "f", "nn", "p", "return", "self", "str"], "x_clip/modeling_x_clip.py:XCLIPVisionEncoderLayer.__init__": ["GradientCheckpointingLayer", "Identity", "LayerNorm", "Linear", "Model", "ModelAttention", "ModelConfig", "ModelDropPath", "ModelMLP", "__init__", "class", "config", "def", "drop_path", "drop_path_rate", "else", "embed_dim", "eps", "hidden_size", "if", "layer_norm1", "layer_norm2", "layer_norm_eps", "message_attn", "message_fc", "message_ln", "mlp", "nn", "num_frames", "self", "self_attn", "super"], "x_clip/modeling_x_clip.py:XCLIPVisionEncoderLayer.forward": ["False", "FloatTensor", "GradientCheckpointingLayer", "Model", "None", "Tensor", "attention_mask", "attn_weights", "batch_size", "batch_time", "bool", "cat", "causal_attention_mask", "class", "def", "dim", "drop_path", "forward", "hidden_size", "hidden_states", "if", "layer_norm1", "layer_norm2", "message_attn", "message_fc", "message_ln", "mlp", "msg_token", "num_frames", "output_attentions", "outputs", "residual", "return", "self", "self_attn", "seq_length", "size", "torch", "tuple", "view"], "x_clip/modeling_x_clip.py:XCLIPPreTrainedModel._init_weights": ["LayerNorm", "Linear", "Model", "ModelAttention", "ModelMLP", "ModelModel", "ModelMultiframeIntegrationTransformer", "ModelTextEmbeddings", "ModelVisionEmbeddings", "None", "PreTrainedModel", "_init_weights", "arange", "bias", "class", "class_embedding", "config", "copy_", "def", "elif", "embed_dim", "expand", "factor", "fc1", "fc2", "fc_std", "hidden_size", "if", "in_proj_std", "init", "initializer_factor", "initializer_range", "is", "isinstance", "k_proj", "mean", "module", "nn", "normal_", "not", "num_hidden_layers", "ones_", "out_proj", "out_proj_std", "patch_embedding", "position_embedding", "position_ids", "prompts_visual_projection", "q_proj", "self", "shape", "std", "text_embed_dim", "text_projection", "token_embedding", "torch", "v_proj", "vision_embed_dim", "visual_projection", "weight", "zeros_"], "x_clip/modeling_x_clip.py:XCLIPEncoder.__init__": ["False", "Model", "ModelConfig", "ModelLayer", "Module", "ModuleList", "_", "__init__", "class", "config", "def", "for", "gradient_checkpointing", "in", "layers", "nn", "num_hidden_layers", "range", "self", "super"], "x_clip/modeling_x_clip.py:XCLIPEncoder.forward": ["BaseModelOutput", "Model", "Module", "None", "Tensor", "all_attentions", "attention_mask", "attentions", "bool", "causal_attention_mask", "class", "config", "def", "else", "encoder_layer", "encoder_states", "enumerate", "for", "forward", "hidden_states", "idx", "if", "in", "inputs_embeds", "is", "last_hidden_state", "layer_outputs", "layers", "nn", "not", "output_attentions", "output_hidden_states", "r", "return", "return_dict", "self", "torch", "tuple", "use_return_dict"], "x_clip/modeling_x_clip.py:XCLIPTextTransformer.__init__": ["LayerNorm", "Model", "ModelEncoder", "ModelTextConfig", "ModelTextEmbeddings", "Module", "__init__", "class", "config", "def", "embed_dim", "embeddings", "encoder", "eps", "final_layer_norm", "hidden_size", "layer_norm_eps", "nn", "self", "super"], "x_clip/modeling_x_clip.py:XCLIPTextTransformer.forward": ["BaseModelOutputWithPooling", "Model", "Module", "None", "Tensor", "ValueError", "You", "_create_4d_causal_attention_mask", "_prepare_4d_attention_mask", "arange", "argmax", "attention_mask", "attentions", "bool", "causal_attention_mask", "class", "config", "def", "device", "dim", "dtype", "either", "else", "embeddings", "encoder", "encoder_outputs", "final_layer_norm", "forward", "have", "hidden_states", "if", "input_ids", "input_shape", "inputs_embeds", "is", "last_hidden_state", "nn", "not", "output_attentions", "output_hidden_states", "pooled_output", "pooler_output", "position_ids", "raise", "return", "return_dict", "self", "shape", "size", "specify", "to", "torch", "tuple", "use_return_dict", "view"], "x_clip/modeling_x_clip.py:XCLIPTextModel.__init__": ["Model", "ModelPreTrainedModel", "ModelTextConfig", "ModelTextTransformer", "__init__", "class", "config", "def", "post_init", "self", "super", "text_model"], "x_clip/modeling_x_clip.py:XCLIPTextModel.get_input_embeddings": ["Model", "ModelPreTrainedModel", "Module", "class", "def", "embeddings", "get_input_embeddings", "nn", "return", "self", "text_model", "token_embedding"], "x_clip/modeling_x_clip.py:XCLIPTextModel.set_input_embeddings": ["Model", "ModelPreTrainedModel", "class", "def", "embeddings", "self", "set_input_embeddings", "text_model", "token_embedding", "value"], "x_clip/modeling_x_clip.py:XCLIPTextModel.forward": ["BaseModelOutputWithPooling", "Model", "ModelPreTrainedModel", "None", "Tensor", "attention_mask", "bool", "class", "def", "forward", "input_ids", "kwargs", "output_attentions", "output_hidden_states", "position_ids", "r", "return", "return_dict", "self", "text_model", "torch", "tuple"], "x_clip/modeling_x_clip.py:XCLIPVisionEncoder.__init__": ["False", "Model", "ModelConfig", "ModelLayer", "Module", "ModuleList", "_", "__init__", "class", "config", "def", "for", "gradient_checkpointing", "in", "layers", "nn", "num_hidden_layers", "range", "self", "super"], "x_clip/modeling_x_clip.py:XCLIPVisionEncoder.forward": ["BaseModelOutput", "Model", "Module", "None", "Tensor", "all_attentions", "attention_mask", "attentions", "bool", "causal_attention_mask", "class", "config", "def", "else", "encoder_layer", "encoder_states", "enumerate", "for", "forward", "hidden_states", "idx", "if", "in", "inputs_embeds", "is", "last_hidden_state", "layer_outputs", "layers", "nn", "not", "output_attentions", "output_hidden_states", "r", "return", "return_dict", "self", "torch", "tuple", "use_return_dict", "v"], "x_clip/modeling_x_clip.py:XCLIPVisionTransformer.__init__": ["LayerNorm", "Model", "ModelVisionConfig", "ModelVisionEmbeddings", "ModelVisionEncoder", "Module", "__init__", "class", "config", "def", "embed_dim", "embeddings", "encoder", "eps", "hidden_size", "layer_norm_eps", "nn", "post_layernorm", "pre_layernorm", "self", "super"], "x_clip/modeling_x_clip.py:XCLIPVisionTransformer.forward": ["BaseModelOutputWithPooling", "False", "FloatTensor", "Model", "Module", "None", "attentions", "bool", "class", "config", "def", "else", "embeddings", "encoder", "encoder_outputs", "forward", "hidden_states", "if", "inputs_embeds", "interpolate_pos_encoding", "is", "last_hidden_state", "nn", "not", "output_attentions", "output_hidden_states", "pixel_values", "pooled_output", "pooler_output", "post_layernorm", "pre_layernorm", "return", "return_dict", "self", "torch", "tuple", "use_return_dict"], "x_clip/modeling_x_clip.py:XCLIPVisionModel.__init__": ["Model", "ModelPreTrainedModel", "ModelVisionConfig", "ModelVisionTransformer", "__init__", "class", "config", "def", "post_init", "self", "super", "vision_model"], "x_clip/modeling_x_clip.py:XCLIPVisionModel.get_input_embeddings": ["Model", "ModelPreTrainedModel", "Module", "class", "def", "embeddings", "get_input_embeddings", "nn", "patch_embedding", "return", "self", "vision_model"], "x_clip/modeling_x_clip.py:XCLIPVisionModel.forward": ["BaseModelOutputWithPooling", "FloatTensor", "Model", "ModelPreTrainedModel", "None", "bool", "class", "def", "forward", "kwargs", "output_attentions", "output_hidden_states", "pixel_values", "r", "return", "return_dict", "self", "torch", "tuple", "vision_model"], "x_clip/modeling_x_clip.py:XCLIPMultiframeIntegrationTransformer.__init__": ["Model", "ModelEncoder", "ModelVisionConfig", "Module", "Parameter", "__init__", "class", "config", "def", "empty", "encoder", "hidden_size", "nn", "num_frames", "position_embedding", "self", "super", "torch"], "x_clip/modeling_x_clip.py:XCLIPMultiframeIntegrationTransformer.forward": ["BaseModelOutput", "BaseModelOutputWithPooling", "False", "Model", "Module", "None", "attentions", "bool", "class", "def", "dim", "dtype", "encoder", "encoder_outputs", "forward", "hidden_states", "if", "inputs_embeds", "keepdim", "last_hidden_state", "mean", "nn", "not", "output_attentions", "output_hidden_states", "pooled_output", "pooler_output", "position_embedding", "residual", "return", "return_dict", "self", "tuple", "type"], "x_clip/modeling_x_clip.py:XCLIPCrossAttention.__init__": ["Dropout", "False", "Linear", "Model", "Module", "__init__", "attn_drop", "bias", "class", "config", "def", "dim", "head_dim", "k_proj", "nn", "num_heads", "proj", "proj_drop", "projection_dim", "prompt_attention_dropout", "prompt_num_attention_heads", "prompt_projection_dropout", "q_proj", "scale", "self", "super", "v_proj"], "x_clip/modeling_x_clip.py:XCLIPCrossAttention._shape": ["Model", "Module", "Tensor", "_shape", "batch_size", "class", "contiguous", "def", "head_dim", "int", "nn", "num_heads", "return", "self", "seq_len", "tensor", "torch", "transpose", "view"], "x_clip/modeling_x_clip.py:XCLIPCrossAttention.forward": ["Model", "Module", "attn", "attn_drop", "batch_size", "class", "def", "dim", "forward", "hidden_size", "k_proj", "key_seq_len", "keys", "nn", "num_heads", "permute", "proj", "proj_drop", "q_proj", "queries", "query_seq_len", "reshape", "return", "scale", "self", "shape", "softmax", "transpose", "v_proj", "values", "x"], "x_clip/modeling_x_clip.py:PromptGeneratorLayer.__init__": ["ACT2FN", "Dropout", "LayerNorm", "Linear", "ModelCrossAttention", "ModelGeneratorLayer", "Model_attention_dropout", "Model_hidden_act", "Module", "Sequential", "__init__", "class", "config", "cross_attn", "def", "embed_dim", "eps", "layer_norm_eps", "mlp", "nn", "norm1", "norm3", "projection_dim", "self", "super", "text_config"], "x_clip/modeling_x_clip.py:PromptGeneratorLayer.forward": ["ModelGeneratorLayer", "Module", "class", "cross_attn", "def", "forward", "mlp", "nn", "norm1", "norm3", "return", "self", "visual", "x"], "x_clip/modeling_x_clip.py:XCLIPPromptGenerator.__init__": ["LayerNorm", "Model", "Module", "ModuleList", "Parameter", "PromptGeneratorLayer", "_", "__init__", "alpha", "class", "config", "decoder", "def", "embed_dim", "eps", "for", "in", "layer_norm_eps", "layernorm", "nn", "ones", "projection_dim", "prompt_alpha", "prompt_layers", "range", "self", "super", "torch", "vision_config"], "x_clip/modeling_x_clip.py:XCLIPPromptGenerator.forward": ["Model", "Module", "alpha", "class", "decoder", "def", "for", "forward", "in", "layer", "layernorm", "nn", "return", "self", "text", "visual"], "x_clip/modeling_x_clip.py:XCLIPModel.__init__": ["False", "LayerNorm", "Linear", "Model", "ModelConfig", "ModelMultiframeIntegrationTransformer", "ModelPreTrainedModel", "ModelPromptGenerator", "ModelTextConfig", "ModelTextTransformer", "ModelVisionConfig", "ModelVisionTransformer", "Parameter", "TypeError", "__init__", "_attn_implementation", "be", "bias", "but", "class", "config", "copy", "def", "eps", "expected", "f", "hidden_size", "if", "intermediate_size", "is", "isinstance", "layer_norm_eps", "logit_scale", "logit_scale_init_value", "mit", "mit_config", "mit_hidden_size", "mit_intermediate_size", "mit_num_attention_heads", "mit_num_hidden_layers", "nn", "not", "num_attention_heads", "num_hidden_layers", "of", "post_init", "projection_dim", "prompts_generator", "prompts_visual_layernorm", "prompts_visual_projection", "raise", "randn", "self", "super", "tensor", "text_config", "text_embed_dim", "text_model", "text_projection", "to", "torch", "type", "vision_config", "vision_embed_dim", "vision_model", "visual_projection"], "x_clip/modeling_x_clip.py:XCLIPModel.get_text_features": ["BaseModelOutputWithPooling", "FloatTensor", "Model", "ModelPreTrainedModel", "None", "Tensor", "attention_mask", "class", "def", "get_text_features", "input_ids", "pooler_output", "position_ids", "r", "return", "self", "text_features", "text_model", "text_outputs", "text_projection", "torch"], "x_clip/modeling_x_clip.py:XCLIPModel.get_video_features": ["BaseModelOutputWithPooling", "FloatTensor", "Model", "ModelPreTrainedModel", "Tensor", "batch_size", "class", "cls_features", "def", "get_video_features", "height", "mit", "mit_outputs", "num_channels", "num_frames", "pixel_values", "pooler_output", "r", "reshape", "return", "self", "shape", "torch", "video_embeds", "view", "vision_model", "vision_outputs", "visual_projection", "width"], "x_clip/modeling_x_clip.py:XCLIPModel.forward": ["False", "FloatTensor", "LongTensor", "Model", "ModelOutput", "ModelPreTrainedModel", "Model_loss", "None", "T", "Tensor", "True", "attention_mask", "batch_size", "bd", "bk", "bkd", "bool", "class", "cls_features", "config", "def", "dim", "einsum", "else", "exp", "expand", "forward", "height", "if", "img_features", "input_ids", "interpolate_pos_encoding", "is", "keepdim", "kwargs", "logit_scale", "logits_per_text", "logits_per_video", "loss", "mean", "mit", "mit_output", "mit_outputs", "norm", "not", "num_channels", "num_frames", "output", "output_attentions", "output_hidden_states", "p", "pixel_values", "position_ids", "prompts_generator", "prompts_visual_layernorm", "prompts_visual_projection", "r", "reshape", "return", "return_dict", "return_loss", "self", "shape", "text_embeds", "text_model", "text_model_output", "text_outputs", "text_projection", "torch", "tuple", "unsqueeze", "use_return_dict", "video_embeds", "view", "vision_model", "vision_model_output", "vision_outputs", "visual_projection", "width"], "xcodec/modeling_xcodec.py:ResidualUnit.__init__": ["Conv1d", "ELU", "False", "ModelConfig", "ModelUnit", "Module", "__init__", "activation", "bias", "class", "config", "conv1", "conv2", "def", "dilation", "groups", "in_channels", "int", "kernel_size", "nn", "out_channels", "padding", "self", "stride", "super", "unit_kernel_size"], "xcodec/modeling_xcodec.py:ResidualUnit.forward": ["ModelUnit", "Module", "Tensor", "activation", "class", "conv1", "conv2", "def", "forward", "hidden_state", "nn", "output_tensor", "return", "self", "torch"], "xcodec/modeling_xcodec.py:SemanticEncoderBlock.__init__": ["Conv1d", "ModelConfig", "ModelEncoderBlock", "Module", "ModuleList", "ResidualUnit", "True", "__init__", "bias", "block_dilations", "class", "config", "conv", "def", "dilation", "else", "for", "if", "in", "in_channels", "int", "kernel", "kernel_size", "nn", "out_channels", "padding", "res_units", "self", "stride", "super"], "xcodec/modeling_xcodec.py:SemanticEncoderBlock.forward": ["ModelEncoderBlock", "Module", "Tensor", "class", "conv", "def", "for", "forward", "hidden_state", "in", "nn", "res_units", "return", "self", "torch", "unit"], "xcodec/modeling_xcodec.py:SemanticEncoder.__init__": ["Conv1d", "False", "ModelEncoder", "ModelEncoderBlock", "Model_hidden_size", "Module", "ModuleList", "Number", "ValueError", "__init__", "bias", "channel_ratios", "class", "config", "conv", "conv_blocks", "def", "enumerate", "for", "i", "if", "in", "in_channels", "int", "kernel_size", "len", "match", "must", "nn", "number", "of", "out_channels", "raise", "self", "stride", "strides", "super", "the"], "xcodec/modeling_xcodec.py:SemanticEncoder.forward": ["ModelEncoder", "Module", "Tensor", "block", "class", "conv", "conv_blocks", "def", "for", "forward", "hidden_state", "in", "nn", "return", "self", "torch"], "xcodec/modeling_xcodec.py:SemanticDecoderBlock.__init__": ["Conv1d", "ConvTranspose1d", "False", "ModelConfig", "ModelDecoderBlock", "Module", "ModuleList", "ResidualUnit", "True", "__init__", "bias", "block_dilations", "class", "config", "conv", "def", "dilation", "else", "for", "if", "in", "in_channels", "int", "kernel_size", "nn", "out_channels", "output_padding", "padding", "res_units", "self", "stride", "super"], "xcodec/modeling_xcodec.py:SemanticDecoderBlock.forward": ["ModelDecoderBlock", "Module", "Tensor", "class", "conv", "def", "for", "forward", "hidden_state", "in", "nn", "res_units", "return", "self", "torch", "unit"], "xcodec/modeling_xcodec.py:SemanticDecoder.__init__": ["Conv1d", "False", "ModelDecoder", "ModelDecoderBlock", "Model_hidden_size", "Module", "ModuleList", "__init__", "bias", "channel_ratios", "class", "config", "conv1", "conv2", "conv_blocks", "def", "else", "enumerate", "for", "i", "if", "in", "in_channels", "int", "kernel_size", "len", "nn", "out_channels", "padding", "self", "stride", "strides", "super"], "xcodec/modeling_xcodec.py:SemanticDecoder.forward": ["ModelDecoder", "Module", "Tensor", "block", "class", "conv1", "conv2", "conv_blocks", "def", "for", "forward", "hidden_state", "in", "nn", "return", "self", "torch"], "xcodec/modeling_xcodec.py:XcodecEuclideanCodebook.__init__": ["ModelEuclideanCodebook", "Module", "Tensor", "True", "__init__", "class", "clone", "cluster_size", "codebook_dim", "codebook_size", "config", "def", "embed", "embed_avg", "inited", "nn", "register_buffer", "self", "super", "torch", "zeros"], "xcodec/modeling_xcodec.py:XcodecEuclideanCodebook.quantize": ["ModelEuclideanCodebook", "Module", "True", "class", "def", "dim", "dist", "embed", "embed_ind", "hidden_states", "indices", "keepdim", "max", "nn", "pow", "quantize", "return", "scaled_states", "self", "sum", "t"], "xcodec/modeling_xcodec.py:XcodecEuclideanCodebook.encode": ["ModelEuclideanCodebook", "Module", "class", "def", "embed_ind", "encode", "hidden_states", "nn", "quantize", "reshape", "return", "self", "shape", "view"], "xcodec/modeling_xcodec.py:XcodecEuclideanCodebook.decode": ["F", "ModelEuclideanCodebook", "Module", "class", "decode", "def", "embed", "embed_ind", "embedding", "nn", "quantized", "return", "self"], "xcodec/modeling_xcodec.py:XcodecVectorQuantization.__init__": ["ModelConfig", "ModelEuclideanCodebook", "ModelVectorQuantization", "Module", "__init__", "class", "codebook", "config", "def", "nn", "self", "super"], "xcodec/modeling_xcodec.py:XcodecVectorQuantization.encode": ["ModelVectorQuantization", "Module", "class", "codebook", "def", "embed_in", "encode", "hidden_states", "nn", "permute", "return", "self"], "xcodec/modeling_xcodec.py:XcodecVectorQuantization.decode": ["ModelVectorQuantization", "Module", "class", "codebook", "decode", "def", "embed_ind", "nn", "permute", "quantize", "return", "self"], "xcodec/modeling_xcodec.py:XcodecResidualVectorQuantization.__init__": ["ModelConfig", "ModelResidualVectorQuantization", "ModelVectorQuantization", "Module", "ModuleList", "_", "__init__", "class", "codebook_size", "config", "def", "for", "frame_rate", "in", "nn", "num_quantizers", "quantizers", "range", "self", "super"], "xcodec/modeling_xcodec.py:XcodecResidualVectorQuantization.get_bandwidth_per_quantizer": ["ModelResidualVectorQuantization", "Module", "class", "codebook_size", "def", "frame_rate", "get_bandwidth_per_quantizer", "log2", "math", "nn", "return", "self"], "xcodec/modeling_xcodec.py:XcodecResidualVectorQuantization.get_num_quantizers_for_bandwidth": ["ModelResidualVectorQuantization", "Module", "None", "and", "bandwidth", "bw_per_q", "class", "def", "floor", "get_bandwidth_per_quantizer", "get_num_quantizers_for_bandwidth", "if", "int", "is", "math", "max", "nn", "not", "num_quantizers", "return", "self"], "xcodec/modeling_xcodec.py:XcodecResidualVectorQuantization.encode": ["ModelResidualVectorQuantization", "Module", "None", "Tensor", "all_indices", "append", "bandwidth", "class", "decode", "def", "embeddings", "encode", "for", "get_num_quantizers_for_bandwidth", "in", "indices", "nn", "num_quantizers", "out_indices", "quantized", "quantizer", "quantizers", "residual", "return", "self", "stack", "torch"], "xcodec/modeling_xcodec.py:XcodecResidualVectorQuantization.decode": ["ModelResidualVectorQuantization", "Module", "Tensor", "class", "codes", "decode", "def", "device", "enumerate", "for", "i", "in", "indices", "nn", "quantized", "quantized_out", "quantizer", "quantizers", "return", "self", "tensor", "torch"], "xcodec/modeling_xcodec.py:XcodecPreTrainedModel._init_weights": ["Conv1d", "ConvTranspose1d", "Embedding", "GroupNorm", "LayerNorm", "Linear", "ModelEuclideanCodebook", "ModelModel", "ModelPreTrainedModel", "None", "PreTrainedAudioTokenizerBase", "Snake1d", "Tensor", "True", "__class__", "__name__", "_init_weights", "a", "acoustic_decoder", "acoustic_encoder", "alpha", "b", "bias", "class", "cluster_size", "config", "constant_", "copy_", "def", "elif", "embed", "embed_avg", "for", "groups", "if", "in", "in_channels", "init", "inited", "initializer_range", "is", "isinstance", "k", "kaiming_normal_", "kernel_size", "math", "mean", "module", "modules", "nn", "normal_", "not", "ones_", "reset_parameters", "self", "sqrt", "std", "submodule", "torch", "trunc_normal_", "uniform_", "weight", "zeros_"], "xcodec/modeling_xcodec.py:XcodecPreTrainedModel.apply_weight_norm": ["ModelPreTrainedModel", "PreTrainedAudioTokenizerBase", "acoustic_decoder", "acoustic_encoder", "apply_weight_norm", "block", "class", "conv1", "conv2", "conv_t1", "def", "for", "hasattr", "if", "in", "name", "nn", "parametrizations", "res_unit", "res_unit1", "res_unit2", "res_unit3", "self", "torch", "utils", "weight", "weight_norm"], "xcodec/modeling_xcodec.py:XcodecPreTrainedModel.remove_weight_norm": ["AttributeError", "ModelPreTrainedModel", "PreTrainedAudioTokenizerBase", "True", "ValueError", "acoustic_decoder", "acoustic_encoder", "and", "class", "def", "except", "for", "hasattr", "if", "in", "leave_parametrized", "m", "module", "modules", "name", "nn", "parametrizations", "parametrize", "pass", "remove_parametrizations", "remove_weight_norm", "self", "torch", "try", "utils", "weight"], "xcodec/modeling_xcodec.py:XcodecPreTrainedModel._get_conv1d_layers": ["Conv1d", "ModelPreTrainedModel", "Module", "PreTrainedAudioTokenizerBase", "_get_conv1d_layers", "append", "child", "children", "class", "def", "extend", "for", "get_conv1d_layers_recursive", "if", "in", "isinstance", "module", "nn", "params_list", "return", "self", "tuple"], "xcodec/modeling_xcodec.py:XcodecPreTrainedModel._get_conv1d_output_lengths": ["ModelPreTrainedModel", "None", "PreTrainedAudioTokenizerBase", "_get_conv1d_layers", "_get_conv1d_output_lengths", "class", "conv1d_layers", "conv1d_output_length", "def", "for", "if", "in", "input_length", "is", "layer", "module", "return", "self"], "xcodec/modeling_xcodec.py:XcodecModel.__init__": ["AutoModel", "Linear", "ModelModel", "ModelPreTrainedModel", "ModelResidualVectorQuantization", "SemanticDecoder", "SemanticEncoder", "__init__", "_adjust_dac_decoder", "acoustic_decoder", "acoustic_encoder", "acoustic_model", "acoustic_model_config", "class", "config", "decoder", "decoder_semantic", "def", "encoder", "encoder_semantic", "eval", "fc", "fc1", "fc2", "from_config", "hidden_size", "hop_length", "nn", "pad", "post_init", "quantizer", "self", "semantic_model", "semantic_model_config", "super"], "xcodec/modeling_xcodec.py:XcodecModel._adjust_dac_decoder": ["ConvTranspose1d", "Identity", "ModelModel", "ModelPreTrainedModel", "Module", "Tanh", "_adjust_dac_decoder", "and", "class", "decoder", "def", "else", "for", "hasattr", "if", "in", "isinstance", "module", "modules", "nn", "output_padding", "r", "stride", "tanh", "tuple"], "xcodec/modeling_xcodec.py:XcodecModel._extract_semantic_features": ["F", "FloatTensor", "ModelModel", "ModelPreTrainedModel", "True", "_extract_semantic_features", "class", "def", "dim", "hidden_states", "input_values", "mean", "no_grad", "output_hidden_states", "outputs", "pad", "return", "self", "semantic_model", "stack", "stacked", "torch", "with"], "xcodec/modeling_xcodec.py:XcodecModel.encode": ["Audio", "F", "ModelEncoderOutput", "ModelModel", "ModelPreTrainedModel", "None", "Select", "Tensor", "This", "ValueError", "_extract_semantic_features", "_get_conv1d_output_lengths", "acoustic_encoder", "audio_codes", "bandwidth", "be", "bool", "but", "cat", "channels", "class", "config", "def", "detach", "dim", "doesn", "e_acoustic", "e_semantic", "e_semantic_input", "elif", "else", "embeddings", "encode", "encoder_semantic", "f", "fc", "float", "got", "if", "in", "input_values", "is", "model", "mono", "must", "not", "of", "one", "pad", "quantizer", "r", "raise", "return", "return_dict", "self", "shape", "support", "t", "target_bandwidths", "the", "torch", "transpose"], "xcodec/modeling_xcodec.py:XcodecModel.decode": ["ModelDecoderOutput", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "acoustic_decoder", "audio_codes", "audio_values", "bool", "class", "config", "decode", "def", "else", "fc2", "if", "is", "not", "quantized", "quantized_acoustic", "quantizer", "r", "return", "return_dict", "self", "torch", "transpose"], "xcodec/modeling_xcodec.py:XcodecModel.forward": ["False", "ModelModel", "ModelOutput", "ModelPreTrainedModel", "None", "Tensor", "audio_codes", "audio_values", "bandwidth", "bool", "class", "config", "decode", "def", "else", "encode", "float", "forward", "if", "input_values", "is", "length", "not", "r", "return", "return_dict", "self", "shape", "torch", "tuple"], "xglm/modeling_xglm.py:XGLMScaledWordEmbedding.__init__": ["Embedding", "Model", "None", "__init__", "class", "def", "embed_scale", "embedding_dim", "float", "int", "nn", "num_embeddings", "padding_idx", "self", "super"], "xglm/modeling_xglm.py:XGLMScaledWordEmbedding.forward": ["Embedding", "Model", "Tensor", "class", "def", "embed_scale", "forward", "input_ids", "nn", "return", "self", "super", "torch"], "xglm/modeling_xglm.py:XGLMSinusoidalPositionalEmbedding.__init__": ["Model", "Module", "None", "__init__", "class", "def", "embedding_dim", "int", "make_weights", "nn", "num_positions", "offset", "padding_idx", "self", "super"], "xglm/modeling_xglm.py:XGLMSinusoidalPositionalEmbedding.make_weights": ["False", "Model", "Module", "None", "class", "def", "device", "dtype", "emb_weights", "embedding_dim", "get_embedding", "hasattr", "if", "int", "make_weights", "nn", "num_embeddings", "padding_idx", "persistent", "register_buffer", "self", "to", "weights"], "xglm/modeling_xglm.py:XGLMSinusoidalPositionalEmbedding.get_embedding": ["Model", "Module", "None", "arange", "cat", "class", "cos", "def", "dim", "dtype", "emb", "embedding_dim", "exp", "float", "get_default_dtype", "get_embedding", "half_dim", "if", "int", "int64", "is", "log", "math", "nn", "not", "num_embeddings", "padding_idx", "return", "sin", "to", "torch", "unsqueeze", "view", "zeros"], "xglm/modeling_xglm.py:XGLMSinusoidalPositionalEmbedding.forward": ["Model", "Module", "None", "Tensor", "bsz", "class", "def", "detach", "embedding_dim", "forward", "if", "index_select", "int", "make_weights", "max_pos", "nn", "offset", "padding_idx", "past_key_values_length", "position_ids", "return", "self", "seq_len", "shape", "size", "torch", "view", "weights"], "xglm/modeling_xglm.py:XGLMAttention.__init__": ["False", "Linear", "Model", "Module", "None", "True", "ValueError", "__init__", "and", "be", "bias", "bool", "by", "class", "def", "divisible", "dropout", "embed_dim", "f", "float", "got", "head_dim", "if", "int", "is_decoder", "k_proj", "layer_idx", "must", "nn", "num_heads", "out_proj", "q_proj", "raise", "scaling", "self", "super", "v_proj"], "xglm/modeling_xglm.py:XGLMAttention.forward": ["Attention", "Cache", "EncoderDecoderCache", "False", "Model", "Module", "None", "Tensor", "True", "ValueError", "_", "and", "attention_mask", "attn_output", "attn_probs", "attn_weights", "attn_weights_reshaped", "be", "bmm", "bool", "bsz", "but", "cache_position", "class", "cross_attention_cache", "curr_past_key_values", "current_states", "def", "device", "dim", "dropout", "dtype", "else", "embed_dim", "f", "finfo", "float16", "float32", "forward", "functional", "get", "head_dim", "hidden_states", "if", "is", "is_cross_attention", "is_updated", "isinstance", "k_proj", "key_states", "key_value_states", "keys", "layer_idx", "layers", "mask", "max", "min", "nn", "not", "num_heads", "of", "out_proj", "output_attentions", "p", "past_key_values", "proj_shape", "q_proj", "query_states", "raise", "reshape", "return", "scaling", "self", "self_attention_cache", "shape", "should", "size", "softmax", "src_len", "tensor", "tgt_len", "to", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "values", "view", "weights"], "xglm/modeling_xglm.py:XGLMDecoderLayer.__init__": ["ACT2FN", "GradientCheckpointingLayer", "LayerNorm", "Linear", "Model", "ModelAttention", "ModelConfig", "None", "True", "__init__", "activation_dropout", "activation_fn", "activation_function", "add_cross_attention", "attention_dropout", "attention_heads", "class", "config", "d_model", "def", "dropout", "embed_dim", "encoder_attn", "encoder_attn_layer_norm", "fc1", "fc2", "ffn_dim", "final_layer_norm", "if", "is_decoder", "layer_idx", "nn", "num_heads", "self", "self_attn", "self_attn_layer_norm", "super"], "xglm/modeling_xglm.py:XGLMDecoderLayer.forward": ["Cache", "False", "GradientCheckpointingLayer", "Model", "None", "Tensor", "True", "activation_dropout", "activation_fn", "attention_mask", "bool", "cache_position", "class", "cross_attn_weights", "def", "dropout", "encoder_attention_mask", "encoder_attn", "encoder_attn_layer_norm", "encoder_hidden_states", "fc1", "fc2", "final_layer_norm", "forward", "functional", "hidden_states", "if", "is", "key_value_states", "nn", "not", "output_attentions", "outputs", "p", "past_key_values", "residual", "return", "self", "self_attn", "self_attn_layer_norm", "self_attn_weights", "torch", "training", "use_cache"], "xglm/modeling_xglm.py:XGLMPreTrainedModel._init_weights": ["Model", "ModelSinusoidalPositionalEmbedding", "PreTrainedModel", "_init_weights", "class", "copy_", "def", "emb_weights", "embedding_dim", "get_embedding", "if", "init", "isinstance", "module", "num_positions", "offset", "padding_idx", "self", "super", "weights"], "xglm/modeling_xglm.py:XGLMModel.__init__": ["False", "LayerNorm", "Model", "ModelConfig", "ModelDecoderLayer", "ModelPreTrainedModel", "ModelScaledWordEmbedding", "ModelSinusoidalPositionalEmbedding", "ModuleList", "__init__", "class", "config", "d_model", "def", "dropout", "else", "embed_positions", "embed_scale", "embed_tokens", "for", "gradient_checkpointing", "i", "if", "in", "layer_idx", "layer_norm", "layerdrop", "layers", "math", "max_position_embeddings", "max_target_positions", "nn", "num_layers", "pad_token_id", "padding_idx", "post_init", "range", "scale_embedding", "self", "sqrt", "super", "vocab_size"], "xglm/modeling_xglm.py:XGLMModel.forward": ["BaseModelOutputWithPastAndCrossAttentions", "Cache", "DynamicCache", "EncoderDecoderCache", "False", "Model", "ModelPreTrainedModel", "None", "Setting", "Tensor", "True", "ValueError", "You", "_prepare_4d_attention_mask", "_prepare_4d_causal_attention_mask", "all_cross_attentions", "all_hidden_states", "all_self_attns", "and", "arange", "at", "attention_mask", "attentions", "bool", "both", "cache_position", "cannot", "checkpointing", "class", "config", "continue", "cross_attentions", "decoder_layer", "def", "device", "dropout", "dropout_probability", "dtype", "either", "elif", "else", "embed_positions", "embed_tokens", "encoder_attention_mask", "encoder_hidden_states", "enumerate", "float", "for", "forward", "functional", "get_seq_length", "gradient", "gradient_checkpointing", "have", "hidden_states", "idx", "if", "in", "incompatible", "input_ids", "input_shape", "inputs_embeds", "is", "is_encoder_decoder", "kwargs", "last_hidden_state", "layer_norm", "layer_outputs", "layerdrop", "layers", "logger", "long", "nn", "not", "or", "output_attentions", "output_hidden_states", "p", "past_key_values", "past_key_values_length", "position_ids", "r", "raise", "rand", "return", "return_dict", "same", "self", "size", "specify", "tgt_len", "the", "time", "to", "torch", "training", "tuple", "unsqueeze", "use_cache", "use_return_dict", "v", "view", "warn_if_padding_and_no_attention_mask", "warning_once", "with"], "xglm/modeling_xglm.py:XGLMForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "Model", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "vocab_size"], "xglm/modeling_xglm.py:XGLMForCausalLM.forward": ["Cache", "CausalLMOutputWithCrossAttentions", "GenerationMixin", "Model", "ModelPreTrainedModel", "None", "Tensor", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "cross_attentions", "def", "else", "encoder_attention_mask", "encoder_hidden_states", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "output", "output_attentions", "output_hidden_states", "outputs", "pad_token_id", "past_key_values", "position_ids", "r", "return", "return_dict", "self", "slice", "slice_indices", "torch", "tuple", "use_cache", "use_return_dict", "vocab_size"], "xlm/modeling_xlm.py:create_sinusoidal_embeddings": ["False", "FloatTensor", "Model_sinusoidal_embeddings", "array", "cos", "def", "detach_", "dim", "for", "in", "j", "n_pos", "np", "out", "pos", "position_enc", "power", "range", "requires_grad", "return", "sin", "torch"], "xlm/modeling_xlm.py:get_masks": ["False", "Model_masks", "None", "alen", "arange", "assert", "attn_mask", "bs", "causal", "def", "device", "dtype", "else", "if", "is", "item", "lengths", "long", "mask", "max", "not", "or", "padding_mask", "repeat", "return", "size", "slen", "torch"], "xlm/modeling_xlm.py:XLMPoolerStartLogits.__init__": ["Linear", "Model", "ModelConfig", "Module", "__init__", "class", "config", "def", "dense", "hidden_size", "nn", "self", "super"], "xlm/modeling_xlm.py:XLMPoolerStartLogits.forward": ["FloatTensor", "Model", "Module", "None", "class", "def", "dense", "dtype", "else", "float16", "forward", "hidden_states", "if", "is", "nn", "not", "p_mask", "return", "self", "squeeze", "torch", "x"], "xlm/modeling_xlm.py:XLMPoolerEndLogits.__init__": ["LayerNorm", "Linear", "Model", "ModelConfig", "Module", "Tanh", "__init__", "activation", "class", "config", "def", "dense_0", "dense_1", "eps", "hidden_size", "layer_norm_eps", "nn", "self", "super"], "xlm/modeling_xlm.py:XLMPoolerEndLogits.forward": ["FloatTensor", "LayerNorm", "LongTensor", "Model", "Module", "None", "One", "activation", "assert", "be", "cat", "class", "def", "dense_0", "dense_1", "dim", "dtype", "else", "expand", "float16", "forward", "gather", "hidden_states", "hsz", "if", "is", "nn", "not", "of", "or", "p_mask", "return", "self", "shape", "should", "slen", "squeeze", "start_positions", "start_states", "torch", "x"], "xlm/modeling_xlm.py:XLMPoolerAnswerClass.__init__": ["False", "Linear", "Model", "ModelConfig", "Module", "Tanh", "__init__", "activation", "bias", "class", "config", "def", "dense_0", "dense_1", "hidden_size", "nn", "self", "super"], "xlm/modeling_xlm.py:XLMPoolerAnswerClass.forward": ["FloatTensor", "LongTensor", "Model", "Module", "None", "One", "activation", "assert", "be", "cat", "class", "cls_index", "cls_token_state", "def", "dense_0", "dense_1", "dim", "else", "expand", "forward", "gather", "hidden_states", "hsz", "if", "is", "nn", "not", "of", "or", "return", "self", "shape", "should", "squeeze", "start_positions", "start_states", "torch", "x"], "xlm/modeling_xlm.py:XLMSQuADHead.__init__": ["Model", "ModelConfig", "ModelPoolerAnswerClass", "ModelPoolerEndLogits", "ModelPoolerStartLogits", "Module", "__init__", "answer_class", "class", "config", "def", "end_logits", "end_n_top", "nn", "self", "start_logits", "start_n_top", "super"], "xlm/modeling_xlm.py:XLMSQuADHead.forward": ["BCEWithLogitsLoss", "CrossEntropyLoss", "False", "FloatTensor", "LongTensor", "Model", "ModelOutput", "Module", "None", "and", "answer_class", "bh", "bl", "blh", "bool", "bsz", "class", "cls_index", "cls_logits", "cls_loss", "def", "dim", "einsum", "else", "end_log_probs", "end_logits", "end_loss", "end_n_top", "end_positions", "end_top_index", "end_top_log_probs", "expand", "expand_as", "for", "forward", "functional", "gather", "hidden_states", "hidden_states_expanded", "hsz", "if", "in", "is", "is_impossible", "loss", "loss_fct", "loss_fct_cls", "nn", "not", "p_mask", "r", "return", "return_dict", "self", "size", "slen", "softmax", "squeeze_", "start_log_probs", "start_logits", "start_loss", "start_n_top", "start_positions", "start_states", "start_top_index", "start_top_index_exp", "start_top_log_probs", "topk", "torch", "total_loss", "tuple", "unsqueeze", "view", "x"], "xlm/modeling_xlm.py:XLMSequenceSummary.__init__": ["Callable", "Dropout", "Identity", "Linear", "Model", "ModelConfig", "Module", "None", "NotImplementedError", "__init__", "activation", "activation_string", "and", "attn", "class", "config", "def", "else", "first_dropout", "get_activation", "getattr", "hasattr", "hidden_size", "if", "last", "last_dropout", "nn", "num_classes", "num_labels", "raise", "self", "summary", "summary_activation", "summary_first_dropout", "summary_last_dropout", "summary_proj_to_labels", "summary_type", "summary_use_proj", "super"], "xlm/modeling_xlm.py:XLMSequenceSummary.forward": ["FloatTensor", "LongTensor", "Model", "Module", "None", "NotImplementedError", "activation", "attn", "class", "cls_index", "def", "dim", "dtype", "elif", "else", "expand", "first", "first_dropout", "forward", "full_like", "gather", "hidden_states", "if", "is", "last", "last_dropout", "long", "mean", "nn", "output", "raise", "return", "self", "shape", "size", "squeeze", "summary", "summary_type", "torch", "unsqueeze"], "xlm/modeling_xlm.py:MultiHeadAttention.__init__": ["Linear", "ModelHeadAttention", "Module", "__init__", "assert", "attention_dropout", "class", "config", "def", "dim", "dropout", "head_dim", "int", "k_lin", "layer_id", "layer_idx", "n_heads", "nn", "out_lin", "q_lin", "self", "super", "v_lin"], "xlm/modeling_xlm.py:MultiHeadAttention.forward": ["EncoderDecoderCache", "False", "ModelHeadAttention", "Module", "None", "True", "and", "bs", "cache", "cache_position", "class", "context", "contiguous", "cross_attention_cache", "curr_past_key_values", "current_states", "def", "dim", "dropout", "dtype", "else", "expand_as", "finfo", "float", "forward", "functional", "get", "head_dim", "if", "input", "is", "is_cross_attention", "is_updated", "isinstance", "k", "k_lin", "key_cache", "kv", "layer_id", "mask", "mask_reshape", "masked_fill_", "math", "matmul", "min", "n_heads", "nn", "not", "out_lin", "output_attentions", "outputs", "p", "q", "q_lin", "qlen", "return", "scores", "self", "self_attention_cache", "size", "softmax", "sqrt", "torch", "training", "transpose", "type_as", "update", "v", "v_lin", "value_cache", "view", "weights"], "xlm/modeling_xlm.py:TransformerFFN.__init__": ["Linear", "ModelFFN", "Module", "__init__", "act", "chunk_size_feed_forward", "class", "config", "def", "dim_hidden", "dropout", "else", "functional", "gelu", "gelu_activation", "if", "in_dim", "lin1", "lin2", "nn", "out_dim", "relu", "self", "seq_len_dim", "super"], "xlm/modeling_xlm.py:TransformerFFN.forward": ["ModelFFN", "Module", "apply_chunking_to_forward", "chunk_size_feed_forward", "class", "def", "ff_chunk", "forward", "input", "nn", "return", "self", "seq_len_dim"], "xlm/modeling_xlm.py:TransformerFFN.ff_chunk": ["ModelFFN", "Module", "act", "class", "def", "dropout", "ff_chunk", "functional", "input", "lin1", "lin2", "nn", "p", "return", "self", "training", "x"], "xlm/modeling_xlm.py:XLMPreTrainedModel.dummy_inputs": ["Model", "None", "PreTrainedModel", "and", "attention_mask", "attns_list", "class", "config", "def", "dummy_inputs", "else", "if", "input_ids", "inputs_list", "langs", "langs_list", "n_langs", "return", "self", "tensor", "torch", "use_lang_emb"], "xlm/modeling_xlm.py:XLMPreTrainedModel._init_weights": ["Embedding", "False", "LayerNorm", "Linear", "Model", "ModelModel", "None", "PreTrainedModel", "_init_weights", "_is_hf_initialized", "and", "arange", "bias", "class", "config", "constant_", "copy_", "create_sinusoidal_embeddings", "def", "emb_dim", "embed_init_std", "empty_like", "expand", "getattr", "if", "init", "init_std", "is", "isinstance", "max_position_embeddings", "mean", "module", "nn", "normal_", "not", "ones_", "out", "padding_idx", "position_embeddings", "position_ids", "self", "shape", "sinusoidal_embeddings", "std", "torch", "weight", "zeros_"], "xlm/modeling_xlm.py:XLMModel.__init__": ["Currently", "Embedding", "False", "LayerNorm", "Model", "ModelPreTrainedModel", "ModuleList", "MultiHeadAttention", "NotImplementedError", "TransformerFFN", "__init__", "a", "an", "and", "append", "arange", "as", "assert", "attention_dropout", "attentions", "be", "can", "causal", "class", "config", "def", "dim", "dropout", "emb_dim", "embeddings", "encoder", "eos_index", "eps", "expand", "ffns", "for", "hidden_dim", "i", "if", "in", "is_decoder", "is_encoder", "lang_embeddings", "layer_idx", "layer_norm1", "layer_norm2", "layer_norm_emb", "layer_norm_eps", "max_position_embeddings", "multiple", "must", "n_heads", "n_langs", "n_layers", "n_words", "nn", "not", "of", "only", "pad_index", "padding_idx", "persistent", "position_embeddings", "position_ids", "post_init", "raise", "range", "register_buffer", "self", "super", "torch", "transformer", "use_lang_emb", "used"], "xlm/modeling_xlm.py:XLMModel.get_input_embeddings": ["Model", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "return", "self"], "xlm/modeling_xlm.py:XLMModel.set_input_embeddings": ["Model", "ModelPreTrainedModel", "class", "def", "embeddings", "new_embeddings", "self", "set_input_embeddings"], "xlm/modeling_xlm.py:XLMModel.forward": ["BaseModelOutput", "DynamicCache", "EncoderDecoderCache", "Model", "ModelPreTrainedModel", "None", "Tensor", "_slen", "and", "assert", "attention_mask", "attentions", "attn", "attn_mask", "attn_outputs", "bool", "bs", "cache", "cache_position", "causal", "class", "config", "def", "device", "dict", "dim", "dropout", "dtype", "else", "embeddings", "expand_as", "ffns", "for", "forward", "functional", "get_masks", "get_seq_length", "hidden_states", "i", "if", "in", "input_ids", "inputs_embeds", "is", "item", "kwargs", "lang_embeddings", "langs", "last_hidden_state", "layer_norm1", "layer_norm2", "layer_norm_emb", "lengths", "long", "mask", "max", "n_langs", "n_layers", "nn", "not", "output_attentions", "output_hidden_states", "p", "pad_index", "padding_mask", "position_embeddings", "position_ids", "r", "range", "return", "return_dict", "self", "size", "slen", "str", "sum", "tensor", "to", "token_type_ids", "torch", "training", "tuple", "unsqueeze", "use_lang_emb", "use_return_dict", "v"], "xlm/modeling_xlm.py:XLMPredLayer.__init__": ["AdaptiveLogSoftmaxWithLoss", "False", "Linear", "Model", "Module", "True", "__init__", "asm", "asm_cutoffs", "asm_div_value", "bias", "class", "config", "cutoffs", "def", "dim", "div_value", "else", "emb_dim", "head_bias", "if", "in_features", "is", "n_classes", "n_words", "nn", "pad_index", "proj", "self", "super"], "xlm/modeling_xlm.py:XLMPredLayer.forward": ["False", "Model", "Module", "None", "_", "asm", "class", "cross_entropy", "def", "else", "forward", "functional", "if", "is", "log_prob", "loss", "mean", "n_words", "nn", "not", "outputs", "proj", "reduction", "return", "scores", "self", "view", "x", "y"], "xlm/modeling_xlm.py:XLMWithLMHeadModel.__init__": ["GenerationMixin", "Model", "ModelModel", "ModelPreTrainedModel", "ModelPredLayer", "__init__", "class", "config", "def", "post_init", "pred_layer", "self", "super", "transformer"], "xlm/modeling_xlm.py:XLMWithLMHeadModel.get_output_embeddings": ["GenerationMixin", "Model", "ModelPreTrainedModel", "class", "def", "get_output_embeddings", "pred_layer", "proj", "return", "self"], "xlm/modeling_xlm.py:XLMWithLMHeadModel.set_output_embeddings": ["GenerationMixin", "Model", "ModelPreTrainedModel", "class", "def", "new_embeddings", "pred_layer", "proj", "self", "set_output_embeddings"], "xlm/modeling_xlm.py:XLMWithLMHeadModel.prepare_inputs_for_generation": ["False", "GenerationMixin", "Model", "ModelPreTrainedModel", "None", "attention_mask", "cat", "class", "config", "def", "device", "dim", "dtype", "effective_batch_size", "else", "for", "full", "full_like", "if", "in", "input_ids", "is", "is_first_iteration", "items", "key", "kwargs", "lang_id", "langs", "long", "mask_token", "mask_token_id", "model_inputs", "not", "pop", "prepare_inputs_for_generation", "return", "self", "shape", "token_type_ids", "torch", "value"], "xlm/modeling_xlm.py:XLMWithLMHeadModel.forward": ["GenerationMixin", "MaskedLMOutput", "Model", "ModelPreTrainedModel", "None", "Tensor", "attention_mask", "attentions", "bool", "cache", "cache_position", "class", "config", "def", "dict", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "langs", "lengths", "logits", "logits_to_keep", "loss", "not", "output_attentions", "output_hidden_states", "outputs", "position_ids", "pred_layer", "r", "return", "return_dict", "self", "slice", "slice_indices", "str", "token_type_ids", "torch", "transformer", "transformer_outputs", "tuple", "use_return_dict"], "xlm/modeling_xlm.py:XLMForSequenceClassification.__init__": ["Model", "ModelModel", "ModelPreTrainedModel", "ModelSequenceSummary", "__init__", "class", "config", "def", "num_labels", "post_init", "self", "sequence_summary", "super", "transformer"], "xlm/modeling_xlm.py:XLMForSequenceClassification.forward": ["BCEWithLogitsLoss", "CrossEntropyLoss", "MSELoss", "Model", "ModelPreTrainedModel", "None", "SequenceClassifierOutput", "Tensor", "and", "attention_mask", "attentions", "bool", "cache", "class", "config", "def", "dict", "dtype", "elif", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "kwargs", "labels", "langs", "lengths", "logits", "long", "loss", "loss_fct", "multi_label_classification", "not", "num_labels", "or", "output", "output_attentions", "output_hidden_states", "position_ids", "problem_type", "r", "regression", "return", "return_dict", "self", "sequence_summary", "single_label_classification", "squeeze", "str", "token_type_ids", "torch", "transformer", "transformer_outputs", "tuple", "use_return_dict", "view"], "xlm/modeling_xlm.py:XLMForQuestionAnsweringSimple.__init__": ["Linear", "Model", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "config", "def", "hidden_size", "nn", "num_labels", "post_init", "qa_outputs", "self", "super", "transformer"], "xlm/modeling_xlm.py:XLMForQuestionAnsweringSimple.forward": ["CrossEntropyLoss", "Model", "ModelPreTrainedModel", "None", "QuestionAnsweringModelOutput", "Tensor", "and", "attention_mask", "attentions", "bool", "cache", "clamp", "class", "config", "contiguous", "def", "dict", "dim", "else", "end_logits", "end_loss", "end_positions", "forward", "hidden_states", "if", "ignore_index", "ignored_index", "input_ids", "inputs_embeds", "is", "kwargs", "langs", "len", "lengths", "logits", "loss", "loss_fct", "not", "output", "output_attentions", "output_hidden_states", "position_ids", "qa_outputs", "r", "return", "return_dict", "self", "sequence_output", "size", "split", "squeeze", "start_logits", "start_loss", "start_positions", "str", "token_type_ids", "torch", "total_loss", "transformer", "transformer_outputs", "tuple", "use_return_dict"], "xlm/modeling_xlm.py:XLMForQuestionAnswering.__init__": ["Model", "ModelModel", "ModelPreTrainedModel", "ModelSQuADHead", "__init__", "class", "config", "def", "post_init", "qa_outputs", "self", "super", "transformer"], "xlm/modeling_xlm.py:XLMForQuestionAnswering.forward": ["Model", "ModelOutput", "ModelPreTrainedModel", "None", "Tensor", "attention_mask", "attentions", "bool", "cache", "class", "cls_index", "cls_logits", "config", "def", "dict", "else", "end_positions", "end_top_index", "end_top_log_probs", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "is_impossible", "kwargs", "langs", "lengths", "loss", "not", "output", "output_attentions", "output_hidden_states", "outputs", "p_mask", "position_ids", "qa_outputs", "r", "return", "return_dict", "self", "start_positions", "start_top_index", "start_top_log_probs", "str", "token_type_ids", "torch", "transformer", "transformer_outputs", "tuple", "use_return_dict"], "xlm/modeling_xlm.py:XLMForTokenClassification.__init__": ["Dropout", "Linear", "Model", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "classifier", "config", "def", "dropout", "hidden_size", "nn", "num_labels", "post_init", "self", "super", "transformer"], "xlm/modeling_xlm.py:XLMForTokenClassification.forward": ["CrossEntropyLoss", "Model", "ModelPreTrainedModel", "None", "Tensor", "TokenClassifierOutput", "attention_mask", "attentions", "bool", "cache", "class", "classifier", "config", "def", "dict", "dropout", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "langs", "lengths", "logits", "loss", "loss_fct", "not", "num_labels", "output", "output_attentions", "output_hidden_states", "outputs", "position_ids", "r", "return", "return_dict", "self", "sequence_output", "str", "token_type_ids", "torch", "transformer", "tuple", "use_return_dict", "view"], "xlm/modeling_xlm.py:XLMForMultipleChoice.__init__": ["Linear", "Model", "ModelModel", "ModelPreTrainedModel", "ModelSequenceSummary", "__init__", "class", "config", "def", "inputs", "kwargs", "logits_proj", "nn", "num_labels", "post_init", "self", "sequence_summary", "super", "transformer"], "xlm/modeling_xlm.py:XLMForMultipleChoice.forward": ["CrossEntropyLoss", "Model", "ModelPreTrainedModel", "MultipleChoiceModelOutput", "None", "Please", "Tensor", "The", "attention", "attention_mask", "attentions", "be", "bool", "cache", "cannot", "choice", "class", "config", "def", "dict", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "instead", "is", "kwargs", "labels", "langs", "lengths", "logger", "logits", "logits_proj", "loss", "loss_fct", "mask", "models", "multiple", "not", "num_choices", "output", "output_attentions", "output_hidden_states", "parameter", "position_ids", "r", "reshaped_logits", "return", "return_dict", "self", "sequence_summary", "shape", "size", "str", "the", "token_type_ids", "torch", "transformer", "transformer_outputs", "tuple", "use", "use_return_dict", "used", "view", "warning", "with"], "xlm_roberta/modeling_xlm_roberta.py:XLMRobertaEmbeddings.__init__": ["Dropout", "Embedding", "False", "LayerNorm", "Model", "Module", "__init__", "arange", "class", "config", "def", "dropout", "dtype", "eps", "expand", "hidden_dropout_prob", "hidden_size", "layer_norm_eps", "long", "max_position_embeddings", "nn", "pad_token_id", "padding_idx", "persistent", "position_embeddings", "position_ids", "register_buffer", "self", "size", "super", "token_type_embeddings", "token_type_ids", "torch", "type_vocab_size", "vocab_size", "word_embeddings", "zeros"], "xlm_roberta/modeling_xlm_roberta.py:XLMRobertaEmbeddings.forward": ["FloatTensor", "LayerNorm", "LongTensor", "Model", "Module", "None", "Tensor", "batch_size", "buffered_token_type_ids", "class", "create_position_ids_from_input_ids", "create_position_ids_from_inputs_embeds", "def", "device", "dim", "dropout", "dtype", "else", "embeddings", "expand", "forward", "gather", "hasattr", "if", "index", "input_ids", "input_shape", "inputs_embeds", "int", "is", "long", "nn", "not", "padding_idx", "past_key_values_length", "position_embeddings", "position_ids", "return", "self", "seq_length", "shape", "size", "token_type_embeddings", "token_type_ids", "torch", "word_embeddings", "zeros"], "xlm_roberta/modeling_xlm_roberta.py:XLMRobertaEmbeddings.create_position_ids_from_inputs_embeds": ["Model", "Module", "arange", "class", "create_position_ids_from_inputs_embeds", "def", "device", "dtype", "expand", "input_shape", "inputs_embeds", "long", "nn", "padding_idx", "position_ids", "return", "sequence_length", "size", "torch", "unsqueeze"], "xlm_roberta/modeling_xlm_roberta.py:XLMRobertaEmbeddings.create_position_ids_from_input_ids": ["Model", "Module", "class", "create_position_ids_from_input_ids", "cumsum", "def", "dim", "incremental_indices", "input_ids", "int", "long", "mask", "ne", "nn", "padding_idx", "past_key_values_length", "return", "torch", "type_as"], "xlm_roberta/modeling_xlm_roberta.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "contiguous", "def", "dim", "dropout", "float", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "shape", "size", "softmax", "torch", "training", "transpose", "value"], "xlm_roberta/modeling_xlm_roberta.py:XLMRobertaSelfAttention.__init__": ["Dropout", "False", "Linear", "Model", "Module", "None", "The", "ValueError", "__init__", "a", "all_head_size", "and", "attention", "attention_head_size", "attention_probs_dropout_prob", "class", "config", "def", "dropout", "embedding_size", "f", "hasattr", "heads", "hidden", "hidden_size", "if", "int", "is", "is_causal", "is_decoder", "key", "layer_idx", "multiple", "nn", "not", "num_attention_heads", "number", "of", "query", "raise", "scaling", "self", "size", "super", "the", "value"], "xlm_roberta/modeling_xlm_roberta.py:XLMRobertaSelfAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "EncoderDecoderCache", "FloatTensor", "Model", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_attn_implementation", "attention_head_size", "attention_interface", "attention_mask", "attn_output", "attn_weights", "cache_position", "class", "config", "contiguous", "current_past_key_values", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "hidden_shape", "hidden_states", "if", "input_shape", "is", "isinstance", "key", "key_layer", "kwargs", "layer_idx", "nn", "not", "p", "past_key_values", "query", "query_layer", "reshape", "return", "scaling", "self", "self_attention_cache", "shape", "torch", "training", "transpose", "tuple", "update", "value", "value_layer", "view"], "xlm_roberta/modeling_xlm_roberta.py:XLMRobertaCrossAttention.__init__": ["Dropout", "False", "Linear", "Model", "Module", "None", "The", "ValueError", "__init__", "a", "all_head_size", "and", "attention", "attention_head_size", "attention_probs_dropout_prob", "class", "config", "def", "dropout", "embedding_size", "f", "hasattr", "heads", "hidden", "hidden_size", "if", "int", "is", "is_causal", "key", "layer_idx", "multiple", "nn", "not", "num_attention_heads", "number", "of", "query", "raise", "scaling", "self", "size", "super", "the", "value"], "xlm_roberta/modeling_xlm_roberta.py:XLMRobertaCrossAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "EncoderDecoderCache", "False", "FloatTensor", "Model", "Module", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "_attn_implementation", "and", "attention_head_size", "attention_interface", "attention_mask", "attn_output", "attn_weights", "bsz", "class", "config", "contiguous", "cross_attention_cache", "def", "dropout", "eager", "eager_attention_forward", "else", "encoder_hidden_states", "forward", "get", "hidden_states", "if", "is", "is_updated", "key", "key_layer", "keys", "kv_input_shape", "kwargs", "layer_idx", "layers", "nn", "not", "p", "past_key_values", "q_input_shape", "query", "query_layer", "reshape", "return", "scaling", "self", "shape", "src_len", "tgt_len", "torch", "training", "transpose", "tuple", "update", "value", "value_layer", "values", "view"], "xlm_roberta/modeling_xlm_roberta.py:XLMRobertaSelfOutput.__init__": ["Dropout", "LayerNorm", "Linear", "Model", "Module", "__init__", "class", "config", "def", "dense", "dropout", "eps", "hidden_dropout_prob", "hidden_size", "layer_norm_eps", "nn", "self", "super"], "xlm_roberta/modeling_xlm_roberta.py:XLMRobertaSelfOutput.forward": ["LayerNorm", "Model", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "xlm_roberta/modeling_xlm_roberta.py:XLMRobertaAttention.__init__": ["False", "Model", "ModelCrossAttention", "ModelSelfAttention", "ModelSelfOutput", "Module", "None", "__init__", "attention_class", "class", "config", "def", "else", "if", "is_causal", "is_cross_attention", "layer_idx", "nn", "output", "self", "super"], "xlm_roberta/modeling_xlm_roberta.py:XLMRobertaAttention.forward": ["Cache", "FloatTensor", "Model", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attention_output", "attn_weights", "cache_position", "class", "def", "else", "encoder_attention_mask", "encoder_hidden_states", "forward", "hidden_states", "if", "is_cross_attention", "kwargs", "nn", "not", "output", "past_key_values", "return", "self", "torch", "tuple"], "xlm_roberta/modeling_xlm_roberta.py:XLMRobertaIntermediate.__init__": ["ACT2FN", "Linear", "Model", "Module", "__init__", "class", "config", "def", "dense", "else", "hidden_act", "hidden_size", "if", "intermediate_act_fn", "intermediate_size", "isinstance", "nn", "self", "str", "super"], "xlm_roberta/modeling_xlm_roberta.py:XLMRobertaIntermediate.forward": ["Model", "Module", "Tensor", "class", "def", "dense", "forward", "hidden_states", "intermediate_act_fn", "nn", "return", "self", "torch"], "xlm_roberta/modeling_xlm_roberta.py:XLMRobertaOutput.__init__": ["Dropout", "LayerNorm", "Linear", "Model", "Module", "__init__", "class", "config", "def", "dense", "dropout", "eps", "hidden_dropout_prob", "hidden_size", "intermediate_size", "layer_norm_eps", "nn", "self", "super"], "xlm_roberta/modeling_xlm_roberta.py:XLMRobertaOutput.forward": ["LayerNorm", "Model", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "xlm_roberta/modeling_xlm_roberta.py:XLMRobertaLayer.__init__": ["False", "GradientCheckpointingLayer", "Model", "ModelAttention", "ModelIntermediate", "ModelOutput", "None", "True", "ValueError", "__init__", "a", "add_cross_attention", "added", "as", "attention", "be", "chunk_size_feed_forward", "class", "config", "cross", "crossattention", "decoder", "def", "f", "if", "intermediate", "is", "is_causal", "is_cross_attention", "is_decoder", "layer_idx", "model", "not", "output", "raise", "self", "seq_len_dim", "should", "super", "used"], "xlm_roberta/modeling_xlm_roberta.py:XLMRobertaLayer.forward": ["Cache", "FloatTensor", "GradientCheckpointingLayer", "If", "Model", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "ValueError", "_", "add_cross_attention", "and", "apply_chunking_to_forward", "are", "attention", "attention_mask", "attention_output", "be", "by", "cache_position", "chunk_size_feed_forward", "class", "config", "cross", "cross_attention_output", "crossattention", "def", "encoder_attention_mask", "encoder_hidden_states", "f", "feed_forward_chunk", "forward", "has", "hasattr", "hidden_states", "if", "instantiated", "is", "is_decoder", "kwargs", "layer_output", "layers", "not", "passed", "past_key_values", "raise", "return", "self", "self_attention_output", "seq_len_dim", "setting", "to", "torch", "tuple", "with"], "xlm_roberta/modeling_xlm_roberta.py:XLMRobertaLayer.feed_forward_chunk": ["GradientCheckpointingLayer", "Model", "attention_output", "class", "def", "feed_forward_chunk", "intermediate", "intermediate_output", "layer_output", "output", "return", "self"], "xlm_roberta/modeling_xlm_roberta.py:XLMRobertaLMHead.__init__": ["LayerNorm", "Linear", "Model", "Module", "Parameter", "__init__", "bias", "class", "config", "decoder", "def", "dense", "eps", "hidden_size", "layer_norm", "layer_norm_eps", "nn", "self", "super", "torch", "vocab_size", "zeros"], "xlm_roberta/modeling_xlm_roberta.py:XLMRobertaLMHead.forward": ["Model", "Module", "class", "decoder", "def", "dense", "features", "forward", "gelu", "kwargs", "layer_norm", "nn", "return", "self", "x"], "xlm_roberta/modeling_xlm_roberta.py:XLMRobertaPreTrainedModel._init_weights": ["Model", "ModelEmbeddings", "ModelLMHead", "PreTrainedModel", "_init_weights", "arange", "bias", "class", "copy_", "def", "elif", "expand", "if", "init", "isinstance", "module", "position_ids", "self", "shape", "super", "token_type_ids", "torch", "zeros_"], "xlm_roberta/modeling_xlm_roberta.py:XLMRobertaEncoder.__init__": ["Model", "ModelLayer", "Module", "ModuleList", "__init__", "class", "config", "def", "for", "i", "in", "layer", "layer_idx", "nn", "num_hidden_layers", "range", "self", "super"], "xlm_roberta/modeling_xlm_roberta.py:XLMRobertaEncoder.forward": ["BaseModelOutputWithPastAndCrossAttentions", "Cache", "FloatTensor", "Model", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "bool", "cache_position", "class", "def", "else", "encoder_attention_mask", "encoder_hidden_states", "enumerate", "for", "forward", "hidden_states", "i", "if", "in", "kwargs", "last_hidden_state", "layer", "layer_module", "nn", "past_key_values", "return", "self", "torch", "tuple", "use_cache"], "xlm_roberta/modeling_xlm_roberta.py:XLMRobertaPooler.__init__": ["Linear", "Model", "Module", "Tanh", "__init__", "activation", "class", "config", "def", "dense", "hidden_size", "nn", "self", "super"], "xlm_roberta/modeling_xlm_roberta.py:XLMRobertaPooler.forward": ["Model", "Module", "Tensor", "activation", "class", "def", "dense", "first_token_tensor", "forward", "hidden_states", "nn", "pooled_output", "return", "self", "torch"], "xlm_roberta/modeling_xlm_roberta.py:XLMRobertaModel.__init__": ["False", "Model", "ModelEmbeddings", "ModelEncoder", "ModelPooler", "ModelPreTrainedModel", "None", "True", "__init__", "add_pooling_layer", "class", "config", "def", "else", "embeddings", "encoder", "gradient_checkpointing", "if", "pooler", "post_init", "r", "self", "super"], "xlm_roberta/modeling_xlm_roberta.py:XLMRobertaModel.get_input_embeddings": ["Model", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "return", "self", "word_embeddings"], "xlm_roberta/modeling_xlm_roberta.py:XLMRobertaModel.set_input_embeddings": ["Model", "ModelPreTrainedModel", "class", "def", "embeddings", "self", "set_input_embeddings", "value", "word_embeddings"], "xlm_roberta/modeling_xlm_roberta.py:XLMRobertaModel.forward": ["BaseModelOutputWithPoolingAndCrossAttentions", "Cache", "DynamicCache", "EncoderDecoderCache", "False", "Model", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "_create_attention_masks", "and", "arange", "attention_mask", "bool", "cache_position", "class", "config", "def", "device", "else", "embedding_output", "embeddings", "encoder", "encoder_attention_mask", "encoder_hidden_states", "encoder_outputs", "exactly", "forward", "get_seq_length", "if", "input_ids", "inputs_embeds", "is", "is_decoder", "is_encoder_decoder", "kwargs", "last_hidden_state", "must", "not", "of", "one", "or", "past_key_values", "past_key_values_length", "pooled_output", "pooler", "pooler_output", "position_ids", "raise", "return", "self", "seq_length", "sequence_output", "shape", "specify", "token_type_ids", "torch", "tuple", "use_cache"], "xlm_roberta/modeling_xlm_roberta.py:XLMRobertaModel._create_attention_masks": ["Model", "ModelPreTrainedModel", "None", "_create_attention_masks", "attention_mask", "cache_position", "class", "config", "create_bidirectional_mask", "create_causal_mask", "def", "else", "embedding_output", "encoder_attention_mask", "encoder_hidden_states", "if", "input_embeds", "is", "is_decoder", "not", "past_key_values", "return", "self"], "xlm_roberta/modeling_xlm_roberta.py:XLMRobertaForCausalLM.__init__": ["False", "GenerationMixin", "If", "Model", "ModelLMHead", "ModelLMHeadModel", "ModelModel", "ModelPreTrainedModel", "True", "__init__", "a", "add", "add_pooling_layer", "as", "class", "config", "def", "if", "is_decoder", "lm_head", "logger", "not", "post_init", "roberta", "self", "standalone", "super", "to", "use", "want", "warning", "you"], "xlm_roberta/modeling_xlm_roberta.py:XLMRobertaForCausalLM.get_output_embeddings": ["GenerationMixin", "Model", "ModelPreTrainedModel", "class", "decoder", "def", "get_output_embeddings", "lm_head", "return", "self"], "xlm_roberta/modeling_xlm_roberta.py:XLMRobertaForCausalLM.set_output_embeddings": ["GenerationMixin", "Model", "ModelPreTrainedModel", "class", "decoder", "def", "lm_head", "new_embeddings", "self", "set_output_embeddings"], "xlm_roberta/modeling_xlm_roberta.py:XLMRobertaForCausalLM.forward": ["BaseModelOutputWithPoolingAndCrossAttentions", "CausalLMOutputWithCrossAttentions", "False", "FloatTensor", "GenerationMixin", "LongTensor", "Model", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "attention_mask", "attentions", "bool", "class", "config", "cross_attentions", "def", "else", "encoder_attention_mask", "encoder_hidden_states", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "not", "outputs", "past_key_values", "position_ids", "r", "return", "return_dict", "roberta", "self", "slice", "slice_indices", "token_type_ids", "torch", "tuple", "use_cache", "vocab_size"], "xlm_roberta/modeling_xlm_roberta.py:XLMRobertaForMaskedLM.__init__": ["False", "If", "Model", "ModelLMHead", "ModelModel", "ModelPreTrainedModel", "__init__", "add_pooling_layer", "attention", "bi", "class", "config", "def", "directional", "for", "if", "is_decoder", "lm_head", "logger", "make", "post_init", "roberta", "self", "super", "sure", "to", "use", "want", "warning", "you"], "xlm_roberta/modeling_xlm_roberta.py:XLMRobertaForMaskedLM.get_output_embeddings": ["Model", "ModelPreTrainedModel", "class", "decoder", "def", "get_output_embeddings", "lm_head", "return", "self"], "xlm_roberta/modeling_xlm_roberta.py:XLMRobertaForMaskedLM.set_output_embeddings": ["Model", "ModelPreTrainedModel", "class", "decoder", "def", "lm_head", "new_embeddings", "self", "set_output_embeddings"], "xlm_roberta/modeling_xlm_roberta.py:XLMRobertaForMaskedLM.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "MaskedLMOutput", "Model", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "attention_mask", "attentions", "class", "config", "def", "device", "encoder_attention_mask", "encoder_hidden_states", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "lm_head", "logits", "loss", "loss_fct", "masked_lm_loss", "not", "outputs", "position_ids", "prediction_scores", "r", "return", "return_dict", "roberta", "self", "sequence_output", "to", "token_type_ids", "torch", "tuple", "view", "vocab_size"], "xlm_roberta/modeling_xlm_roberta.py:XLMRobertaClassificationHead.__init__": ["Dropout", "Linear", "Model", "Module", "None", "__init__", "class", "classifier_dropout", "config", "def", "dense", "dropout", "else", "hidden_dropout_prob", "hidden_size", "if", "is", "nn", "not", "num_labels", "out_proj", "self", "super"], "xlm_roberta/modeling_xlm_roberta.py:XLMRobertaClassificationHead.forward": ["Model", "Module", "class", "def", "dense", "dropout", "features", "forward", "kwargs", "nn", "out_proj", "return", "self", "tanh", "torch", "x"], "xlm_roberta/modeling_xlm_roberta.py:XLMRobertaForSequenceClassification.__init__": ["False", "Model", "ModelClassificationHead", "ModelModel", "ModelPreTrainedModel", "__init__", "add_pooling_layer", "class", "classifier", "config", "def", "num_labels", "post_init", "roberta", "self", "super"], "xlm_roberta/modeling_xlm_roberta.py:XLMRobertaForSequenceClassification.forward": ["BCEWithLogitsLoss", "CrossEntropyLoss", "FloatTensor", "LongTensor", "MSELoss", "Model", "ModelPreTrainedModel", "None", "SequenceClassifierOutput", "Tensor", "TransformersKwargs", "True", "Unpack", "and", "attention_mask", "attentions", "class", "classifier", "config", "def", "device", "dtype", "elif", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "kwargs", "labels", "logits", "long", "loss", "loss_fct", "multi_label_classification", "not", "num_labels", "or", "outputs", "position_ids", "problem_type", "r", "regression", "return", "return_dict", "roberta", "self", "sequence_output", "single_label_classification", "squeeze", "to", "token_type_ids", "torch", "tuple", "view"], "xlm_roberta/modeling_xlm_roberta.py:XLMRobertaForMultipleChoice.__init__": ["Dropout", "False", "Linear", "Model", "ModelModel", "ModelPreTrainedModel", "__init__", "add_pooling_layer", "class", "classifier", "config", "def", "dropout", "hidden_dropout_prob", "hidden_size", "nn", "post_init", "roberta", "self", "super"], "xlm_roberta/modeling_xlm_roberta.py:XLMRobertaForMultipleChoice.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "Model", "ModelPreTrainedModel", "MultipleChoiceModelOutput", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "attention_mask", "attentions", "class", "classifier", "def", "device", "dropout", "else", "flat_attention_mask", "flat_input_ids", "flat_inputs_embeds", "flat_position_ids", "flat_token_type_ids", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "not", "num_choices", "outputs", "pooled_output", "position_ids", "r", "reshaped_logits", "return", "return_dict", "roberta", "self", "shape", "size", "to", "token_type_ids", "torch", "tuple", "view"], "xlm_roberta/modeling_xlm_roberta.py:XLMRobertaForTokenClassification.__init__": ["Dropout", "False", "Linear", "Model", "ModelModel", "ModelPreTrainedModel", "None", "__init__", "add_pooling_layer", "class", "classifier", "classifier_dropout", "config", "def", "dropout", "else", "hidden_dropout_prob", "hidden_size", "if", "is", "nn", "not", "num_labels", "post_init", "roberta", "self", "super"], "xlm_roberta/modeling_xlm_roberta.py:XLMRobertaForTokenClassification.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "Model", "ModelPreTrainedModel", "None", "Tensor", "TokenClassifierOutput", "TransformersKwargs", "True", "Unpack", "attention_mask", "attentions", "class", "classifier", "def", "device", "dropout", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "not", "num_labels", "outputs", "position_ids", "r", "return", "return_dict", "roberta", "self", "sequence_output", "to", "token_type_ids", "torch", "tuple", "view"], "xlm_roberta/modeling_xlm_roberta.py:XLMRobertaForQuestionAnswering.__init__": ["False", "Linear", "Model", "ModelModel", "ModelPreTrainedModel", "__init__", "add_pooling_layer", "class", "config", "def", "hidden_size", "nn", "num_labels", "post_init", "qa_outputs", "roberta", "self", "super"], "xlm_roberta/modeling_xlm_roberta.py:XLMRobertaForQuestionAnswering.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "Model", "ModelPreTrainedModel", "None", "QuestionAnsweringModelOutput", "Tensor", "TransformersKwargs", "True", "Unpack", "and", "attention_mask", "attentions", "clamp", "class", "contiguous", "def", "dim", "end_logits", "end_loss", "end_positions", "forward", "hidden_states", "if", "ignore_index", "ignored_index", "input_ids", "inputs_embeds", "is", "kwargs", "len", "logits", "loss", "loss_fct", "not", "outputs", "position_ids", "qa_outputs", "r", "return", "return_dict", "roberta", "self", "sequence_output", "size", "split", "squeeze", "start_logits", "start_loss", "start_positions", "token_type_ids", "torch", "total_loss", "tuple"], "xlm_roberta_xl/modeling_xlm_roberta_xl.py:XLMRobertaXLEmbeddings.__init__": ["Dropout", "Embedding", "False", "Model", "Module", "__init__", "arange", "class", "config", "def", "dropout", "dtype", "expand", "hidden_dropout_prob", "hidden_size", "long", "max_position_embeddings", "nn", "pad_token_id", "padding_idx", "persistent", "position_embeddings", "position_ids", "register_buffer", "self", "size", "super", "token_type_embeddings", "token_type_ids", "torch", "type_vocab_size", "vocab_size", "word_embeddings", "zeros"], "xlm_roberta_xl/modeling_xlm_roberta_xl.py:XLMRobertaXLEmbeddings.forward": ["FloatTensor", "LongTensor", "Model", "Module", "None", "Tensor", "batch_size", "buffered_token_type_ids", "class", "create_position_ids_from_input_ids", "create_position_ids_from_inputs_embeds", "def", "device", "dim", "dropout", "dtype", "else", "embeddings", "expand", "forward", "gather", "hasattr", "if", "index", "input_ids", "input_shape", "inputs_embeds", "int", "is", "long", "nn", "not", "padding_idx", "past_key_values_length", "position_embeddings", "position_ids", "return", "self", "seq_length", "shape", "size", "token_type_embeddings", "token_type_ids", "torch", "word_embeddings", "zeros"], "xlm_roberta_xl/modeling_xlm_roberta_xl.py:XLMRobertaXLEmbeddings.create_position_ids_from_inputs_embeds": ["Model", "Module", "arange", "class", "create_position_ids_from_inputs_embeds", "def", "device", "dtype", "expand", "input_shape", "inputs_embeds", "long", "nn", "padding_idx", "position_ids", "return", "sequence_length", "size", "torch", "unsqueeze"], "xlm_roberta_xl/modeling_xlm_roberta_xl.py:XLMRobertaXLEmbeddings.create_position_ids_from_input_ids": ["Model", "Module", "class", "create_position_ids_from_input_ids", "cumsum", "def", "dim", "incremental_indices", "input_ids", "int", "long", "mask", "ne", "nn", "padding_idx", "past_key_values_length", "return", "torch", "type_as"], "xlm_roberta_xl/modeling_xlm_roberta_xl.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "contiguous", "def", "dim", "dropout", "float", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "shape", "size", "softmax", "torch", "training", "transpose", "value"], "xlm_roberta_xl/modeling_xlm_roberta_xl.py:XLMRobertaXLSelfAttention.__init__": ["Dropout", "False", "Linear", "Model", "Module", "None", "The", "ValueError", "__init__", "a", "all_head_size", "and", "attention", "attention_head_size", "attention_probs_dropout_prob", "class", "config", "def", "dropout", "embedding_size", "f", "hasattr", "heads", "hidden", "hidden_size", "if", "int", "is", "is_causal", "is_decoder", "key", "layer_idx", "multiple", "nn", "not", "num_attention_heads", "number", "of", "query", "raise", "scaling", "self", "size", "super", "the", "value"], "xlm_roberta_xl/modeling_xlm_roberta_xl.py:XLMRobertaXLSelfAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "EncoderDecoderCache", "FloatTensor", "Model", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_attn_implementation", "attention_head_size", "attention_interface", "attention_mask", "attn_output", "attn_weights", "cache_position", "class", "config", "contiguous", "current_past_key_values", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "hidden_shape", "hidden_states", "if", "input_shape", "is", "isinstance", "key", "key_layer", "kwargs", "layer_idx", "nn", "not", "p", "past_key_values", "query", "query_layer", "reshape", "return", "scaling", "self", "self_attention_cache", "shape", "torch", "training", "transpose", "tuple", "update", "value", "value_layer", "view"], "xlm_roberta_xl/modeling_xlm_roberta_xl.py:XLMRobertaXLCrossAttention.__init__": ["Dropout", "False", "Linear", "Model", "Module", "None", "The", "ValueError", "__init__", "a", "all_head_size", "and", "attention", "attention_head_size", "attention_probs_dropout_prob", "class", "config", "def", "dropout", "embedding_size", "f", "hasattr", "heads", "hidden", "hidden_size", "if", "int", "is", "is_causal", "key", "layer_idx", "multiple", "nn", "not", "num_attention_heads", "number", "of", "query", "raise", "scaling", "self", "size", "super", "the", "value"], "xlm_roberta_xl/modeling_xlm_roberta_xl.py:XLMRobertaXLCrossAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "EncoderDecoderCache", "False", "FloatTensor", "Model", "Module", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "_attn_implementation", "and", "attention_head_size", "attention_interface", "attention_mask", "attn_output", "attn_weights", "bsz", "class", "config", "contiguous", "cross_attention_cache", "def", "dropout", "eager", "eager_attention_forward", "else", "encoder_hidden_states", "forward", "get", "hidden_states", "if", "is", "is_updated", "key", "key_layer", "keys", "kv_input_shape", "kwargs", "layer_idx", "layers", "nn", "not", "p", "past_key_values", "q_input_shape", "query", "query_layer", "reshape", "return", "scaling", "self", "shape", "src_len", "tgt_len", "torch", "training", "transpose", "tuple", "update", "value", "value_layer", "values", "view"], "xlm_roberta_xl/modeling_xlm_roberta_xl.py:XLMRobertaXLSelfOutput.__init__": ["Dropout", "Linear", "Model", "Module", "__init__", "class", "config", "def", "dense", "dropout", "hidden_dropout_prob", "hidden_size", "nn", "self", "super"], "xlm_roberta_xl/modeling_xlm_roberta_xl.py:XLMRobertaXLSelfOutput.forward": ["Model", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "xlm_roberta_xl/modeling_xlm_roberta_xl.py:XLMRobertaXLAttention.__init__": ["False", "LayerNorm", "Model", "ModelCrossAttention", "ModelSelfAttention", "ModelSelfOutput", "Module", "None", "__init__", "attention_class", "class", "config", "def", "else", "eps", "hidden_size", "if", "is_causal", "is_cross_attention", "layer_idx", "layer_norm_eps", "nn", "output", "self", "self_attn_layer_norm", "super"], "xlm_roberta_xl/modeling_xlm_roberta_xl.py:XLMRobertaXLAttention.forward": ["FloatTensor", "Model", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attention_output", "attn_weights", "cache_position", "class", "def", "else", "encoder_attention_mask", "encoder_hidden_states", "forward", "hidden_states", "if", "intermediate", "is_cross_attention", "kwargs", "nn", "not", "output", "past_key_values", "return", "self", "self_attn_layer_norm", "torch", "tuple"], "xlm_roberta_xl/modeling_xlm_roberta_xl.py:XLMRobertaXLOutput.__init__": ["Linear", "Model", "Module", "__init__", "class", "config", "def", "dense", "hidden_size", "intermediate_size", "nn", "self", "super"], "xlm_roberta_xl/modeling_xlm_roberta_xl.py:XLMRobertaXLOutput.forward": ["Model", "Module", "class", "def", "dense", "forward", "hidden_states", "input_tensor", "nn", "return", "self"], "xlm_roberta_xl/modeling_xlm_roberta_xl.py:XLMRobertaXLIntermediate.__init__": ["ACT2FN", "Linear", "Model", "Module", "__init__", "class", "config", "def", "dense", "else", "hidden_act", "hidden_size", "if", "intermediate_act_fn", "intermediate_size", "isinstance", "nn", "self", "str", "super"], "xlm_roberta_xl/modeling_xlm_roberta_xl.py:XLMRobertaXLIntermediate.forward": ["Model", "Module", "Tensor", "class", "def", "dense", "forward", "hidden_states", "intermediate_act_fn", "nn", "return", "self", "torch"], "xlm_roberta_xl/modeling_xlm_roberta_xl.py:XLMRobertaXLLayer.__init__": ["False", "GradientCheckpointingLayer", "LayerNorm", "Model", "ModelAttention", "ModelIntermediate", "ModelOutput", "None", "True", "ValueError", "__init__", "a", "add_cross_attention", "added", "as", "attention", "be", "chunk_size_feed_forward", "class", "config", "cross", "crossattention", "decoder", "def", "eps", "f", "hidden_size", "if", "intermediate", "is", "is_causal", "is_cross_attention", "is_decoder", "layer_idx", "layer_norm_eps", "model", "nn", "not", "output", "raise", "self", "seq_len_dim", "should", "super", "used"], "xlm_roberta_xl/modeling_xlm_roberta_xl.py:XLMRobertaXLLayer.forward": ["Cache", "FloatTensor", "GradientCheckpointingLayer", "If", "Model", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "ValueError", "_", "add_cross_attention", "and", "apply_chunking_to_forward", "are", "attention", "attention_mask", "attention_output", "be", "by", "cache_position", "chunk_size_feed_forward", "class", "config", "cross", "cross_attention_output", "crossattention", "def", "encoder_attention_mask", "encoder_hidden_states", "f", "feed_forward_chunk", "forward", "has", "hasattr", "hidden_states", "if", "instantiated", "is", "is_decoder", "kwargs", "layer_output", "layers", "not", "passed", "past_key_values", "raise", "return", "self", "self_attention_output", "seq_len_dim", "setting", "to", "torch", "tuple", "with"], "xlm_roberta_xl/modeling_xlm_roberta_xl.py:XLMRobertaXLLayer.feed_forward_chunk": ["GradientCheckpointingLayer", "LayerNorm", "Model", "attention_output", "class", "def", "feed_forward_chunk", "intermediate", "intermediate_output", "layer_output", "output", "return", "self"], "xlm_roberta_xl/modeling_xlm_roberta_xl.py:XLMRobertaXLEncoder.__init__": ["LayerNorm", "Model", "ModelLayer", "Module", "ModuleList", "__init__", "class", "config", "def", "eps", "for", "hidden_size", "i", "in", "layer", "layer_idx", "layer_norm_eps", "nn", "num_hidden_layers", "range", "self", "super"], "xlm_roberta_xl/modeling_xlm_roberta_xl.py:XLMRobertaXLEncoder.forward": ["BaseModelOutputWithPastAndCrossAttentions", "Cache", "FloatTensor", "LayerNorm", "Model", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "bool", "cache_position", "class", "def", "else", "encoder_attention_mask", "encoder_hidden_states", "enumerate", "for", "forward", "hidden_states", "i", "if", "in", "kwargs", "last_hidden_state", "layer", "layer_module", "nn", "past_key_values", "return", "self", "torch", "tuple", "use_cache"], "xlm_roberta_xl/modeling_xlm_roberta_xl.py:XLMRobertaXLPreTrainedModel._init_weights": ["Model", "ModelEmbeddings", "ModelLMHead", "PreTrainedModel", "_init_weights", "arange", "bias", "class", "copy_", "def", "elif", "expand", "if", "init", "isinstance", "module", "position_ids", "self", "shape", "super", "token_type_ids", "torch", "zeros_"], "xlm_roberta_xl/modeling_xlm_roberta_xl.py:XLMRobertaXLPooler.__init__": ["Linear", "Model", "Module", "Tanh", "__init__", "activation", "class", "config", "def", "dense", "hidden_size", "nn", "self", "super"], "xlm_roberta_xl/modeling_xlm_roberta_xl.py:XLMRobertaXLPooler.forward": ["Model", "Module", "Tensor", "activation", "class", "def", "dense", "first_token_tensor", "forward", "hidden_states", "nn", "pooled_output", "return", "self", "torch"], "xlm_roberta_xl/modeling_xlm_roberta_xl.py:XLMRobertaXLModel.__init__": ["False", "Model", "ModelEmbeddings", "ModelEncoder", "ModelPooler", "ModelPreTrainedModel", "None", "True", "__init__", "add_pooling_layer", "class", "config", "def", "else", "embeddings", "encoder", "gradient_checkpointing", "if", "pooler", "post_init", "r", "self", "super"], "xlm_roberta_xl/modeling_xlm_roberta_xl.py:XLMRobertaXLModel.get_input_embeddings": ["Model", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "return", "self", "word_embeddings"], "xlm_roberta_xl/modeling_xlm_roberta_xl.py:XLMRobertaXLModel.set_input_embeddings": ["Model", "ModelPreTrainedModel", "class", "def", "embeddings", "self", "set_input_embeddings", "value", "word_embeddings"], "xlm_roberta_xl/modeling_xlm_roberta_xl.py:XLMRobertaXLModel.forward": ["BaseModelOutputWithPoolingAndCrossAttentions", "Cache", "DynamicCache", "EncoderDecoderCache", "False", "Model", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "_create_attention_masks", "and", "arange", "attention_mask", "bool", "cache_position", "class", "config", "def", "device", "else", "embedding_output", "embeddings", "encoder", "encoder_attention_mask", "encoder_hidden_states", "encoder_outputs", "exactly", "forward", "get_seq_length", "if", "input_ids", "inputs_embeds", "is", "is_decoder", "is_encoder_decoder", "kwargs", "last_hidden_state", "must", "not", "of", "one", "or", "past_key_values", "past_key_values_length", "pooled_output", "pooler", "pooler_output", "position_ids", "raise", "return", "self", "seq_length", "sequence_output", "shape", "specify", "token_type_ids", "torch", "tuple", "use_cache"], "xlm_roberta_xl/modeling_xlm_roberta_xl.py:XLMRobertaXLModel._create_attention_masks": ["Model", "ModelPreTrainedModel", "None", "_create_attention_masks", "attention_mask", "cache_position", "class", "config", "create_bidirectional_mask", "create_causal_mask", "def", "else", "embedding_output", "encoder_attention_mask", "encoder_hidden_states", "if", "input_embeds", "is", "is_decoder", "not", "past_key_values", "return", "self"], "xlm_roberta_xl/modeling_xlm_roberta_xl.py:XLMRobertaXLLMHead.__init__": ["LayerNorm", "Linear", "Model", "Module", "Parameter", "__init__", "bias", "class", "config", "decoder", "def", "dense", "eps", "hidden_size", "layer_norm", "layer_norm_eps", "nn", "self", "super", "torch", "vocab_size", "zeros"], "xlm_roberta_xl/modeling_xlm_roberta_xl.py:XLMRobertaXLLMHead.forward": ["Model", "Module", "class", "decoder", "def", "dense", "features", "forward", "gelu", "kwargs", "layer_norm", "nn", "return", "self", "x"], "xlm_roberta_xl/modeling_xlm_roberta_xl.py:XLMRobertaXLClassificationHead.__init__": ["Dropout", "Linear", "Model", "Module", "None", "__init__", "class", "classifier_dropout", "config", "def", "dense", "dropout", "else", "hidden_dropout_prob", "hidden_size", "if", "is", "nn", "not", "num_labels", "out_proj", "self", "super"], "xlm_roberta_xl/modeling_xlm_roberta_xl.py:XLMRobertaXLClassificationHead.forward": ["Model", "Module", "class", "def", "dense", "dropout", "features", "forward", "kwargs", "nn", "out_proj", "return", "self", "tanh", "torch", "x"], "xlm_roberta_xl/modeling_xlm_roberta_xl.py:XLMRobertaXLForCausalLM.__init__": ["False", "GenerationMixin", "If", "Model", "ModelLMHead", "ModelModel", "ModelPreTrainedModel", "RobertaLMHeadModel", "True", "__init__", "a", "add", "add_pooling_layer", "as", "class", "config", "def", "if", "is_decoder", "lm_head", "logger", "not", "post_init", "roberta", "self", "standalone", "super", "to", "use", "want", "warning", "you"], "xlm_roberta_xl/modeling_xlm_roberta_xl.py:XLMRobertaXLForCausalLM.get_output_embeddings": ["GenerationMixin", "Model", "ModelPreTrainedModel", "class", "decoder", "def", "get_output_embeddings", "lm_head", "return", "self"], "xlm_roberta_xl/modeling_xlm_roberta_xl.py:XLMRobertaXLForCausalLM.set_output_embeddings": ["GenerationMixin", "Model", "ModelPreTrainedModel", "bias", "class", "decoder", "def", "lm_head", "new_embeddings", "self", "set_output_embeddings"], "xlm_roberta_xl/modeling_xlm_roberta_xl.py:XLMRobertaXLForCausalLM.forward": ["BaseModelOutputWithPoolingAndCrossAttentions", "CausalLMOutputWithCrossAttentions", "False", "FloatTensor", "GenerationMixin", "LongTensor", "Model", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "cross_attentions", "def", "else", "encoder_attention_mask", "encoder_hidden_states", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "last_hidden_state", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "not", "outputs", "past_key_values", "position_ids", "r", "return", "roberta", "self", "slice", "slice_indices", "token_type_ids", "torch", "tuple", "use_cache", "vocab_size"], "xlm_roberta_xl/modeling_xlm_roberta_xl.py:XLMRobertaXLForMaskedLM.__init__": ["False", "If", "Model", "ModelLMHead", "ModelModel", "ModelPreTrainedModel", "RobertaForMaskedLM", "__init__", "add_pooling_layer", "attention", "bi", "class", "config", "def", "directional", "for", "if", "is_decoder", "lm_head", "logger", "make", "post_init", "roberta", "self", "super", "sure", "to", "use", "want", "warning", "you"], "xlm_roberta_xl/modeling_xlm_roberta_xl.py:XLMRobertaXLForMaskedLM.get_output_embeddings": ["Model", "ModelPreTrainedModel", "class", "decoder", "def", "get_output_embeddings", "lm_head", "return", "self"], "xlm_roberta_xl/modeling_xlm_roberta_xl.py:XLMRobertaXLForMaskedLM.set_output_embeddings": ["Model", "ModelPreTrainedModel", "bias", "class", "decoder", "def", "lm_head", "new_embeddings", "self", "set_output_embeddings"], "xlm_roberta_xl/modeling_xlm_roberta_xl.py:XLMRobertaXLForMaskedLM.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "MaskedLMOutput", "Model", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "attention_mask", "attentions", "class", "config", "def", "encoder_attention_mask", "encoder_hidden_states", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "lm_head", "logits", "loss", "loss_fct", "masked_lm_loss", "not", "outputs", "position_ids", "prediction_scores", "r", "return", "return_dict", "roberta", "self", "sequence_output", "token_type_ids", "torch", "tuple", "view", "vocab_size"], "xlm_roberta_xl/modeling_xlm_roberta_xl.py:XLMRobertaXLForSequenceClassification.__init__": ["False", "Model", "ModelClassificationHead", "ModelModel", "ModelPreTrainedModel", "__init__", "add_pooling_layer", "class", "classifier", "config", "def", "num_labels", "post_init", "roberta", "self", "super"], "xlm_roberta_xl/modeling_xlm_roberta_xl.py:XLMRobertaXLForSequenceClassification.forward": ["BCEWithLogitsLoss", "CrossEntropyLoss", "FloatTensor", "LongTensor", "MSELoss", "Model", "ModelPreTrainedModel", "None", "SequenceClassifierOutput", "TransformersKwargs", "True", "Unpack", "and", "attention_mask", "attentions", "class", "classifier", "config", "def", "dtype", "elif", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "kwargs", "labels", "logits", "long", "loss", "loss_fct", "multi_label_classification", "not", "num_labels", "or", "outputs", "position_ids", "problem_type", "r", "regression", "return", "return_dict", "roberta", "self", "sequence_output", "single_label_classification", "squeeze", "token_type_ids", "torch", "tuple", "view"], "xlm_roberta_xl/modeling_xlm_roberta_xl.py:XLMRobertaXLForMultipleChoice.__init__": ["Dropout", "Linear", "Model", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "classifier", "config", "def", "dropout", "hidden_dropout_prob", "hidden_size", "nn", "post_init", "roberta", "self", "super"], "xlm_roberta_xl/modeling_xlm_roberta_xl.py:XLMRobertaXLForMultipleChoice.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "Model", "ModelPreTrainedModel", "MultipleChoiceModelOutput", "None", "TransformersKwargs", "True", "Unpack", "attention_mask", "attentions", "class", "classifier", "def", "dropout", "else", "flat_attention_mask", "flat_input_ids", "flat_inputs_embeds", "flat_position_ids", "flat_token_type_ids", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "not", "num_choices", "outputs", "pooled_output", "position_ids", "r", "reshaped_logits", "return", "return_dict", "roberta", "self", "shape", "size", "token_type_ids", "torch", "tuple", "view"], "xlm_roberta_xl/modeling_xlm_roberta_xl.py:XLMRobertaXLForTokenClassification.__init__": ["Dropout", "False", "Linear", "Model", "ModelModel", "ModelPreTrainedModel", "None", "__init__", "add_pooling_layer", "class", "classifier", "classifier_dropout", "config", "def", "dropout", "else", "hidden_dropout_prob", "hidden_size", "if", "is", "nn", "not", "num_labels", "post_init", "roberta", "self", "super"], "xlm_roberta_xl/modeling_xlm_roberta_xl.py:XLMRobertaXLForTokenClassification.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "Model", "ModelPreTrainedModel", "None", "TokenClassifierOutput", "TransformersKwargs", "True", "Unpack", "active_labels", "active_logits", "active_loss", "attention_mask", "attentions", "class", "classifier", "def", "dropout", "else", "forward", "hidden_states", "if", "ignore_index", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "not", "num_labels", "outputs", "position_ids", "r", "return", "return_dict", "roberta", "self", "sequence_output", "tensor", "token_type_ids", "torch", "tuple", "type_as", "view", "where"], "xlm_roberta_xl/modeling_xlm_roberta_xl.py:XLMRobertaXLForQuestionAnswering.__init__": ["False", "Linear", "Model", "ModelModel", "ModelPreTrainedModel", "__init__", "add_pooling_layer", "class", "config", "def", "hidden_size", "nn", "num_labels", "post_init", "qa_outputs", "roberta", "self", "super"], "xlm_roberta_xl/modeling_xlm_roberta_xl.py:XLMRobertaXLForQuestionAnswering.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "Model", "ModelPreTrainedModel", "None", "QuestionAnsweringModelOutput", "TransformersKwargs", "True", "Unpack", "and", "attention_mask", "attentions", "clamp", "class", "contiguous", "def", "dim", "end_logits", "end_loss", "end_positions", "forward", "hidden_states", "if", "ignore_index", "ignored_index", "input_ids", "inputs_embeds", "is", "kwargs", "len", "logits", "loss", "loss_fct", "not", "outputs", "position_ids", "qa_outputs", "return", "return_dict", "roberta", "self", "sequence_output", "size", "split", "squeeze", "start_logits", "start_loss", "start_positions", "token_type_ids", "torch", "total_loss", "tuple"], "xlnet/modeling_xlnet.py:XLNetRelativeAttention.__init__": ["Dropout", "FloatTensor", "LayerNorm", "Model", "Module", "Parameter", "The", "ValueError", "__init__", "a", "attention", "class", "config", "d_head", "d_model", "def", "dropout", "eps", "f", "heads", "hidden", "if", "is", "k", "layer_norm", "layer_norm_eps", "multiple", "n_head", "nn", "not", "number", "o", "of", "q", "r", "r_r_bias", "r_s_bias", "r_w_bias", "raise", "scale", "seg_embed", "self", "size", "super", "the", "torch", "v"], "xlnet/modeling_xlnet.py:XLNetRelativeAttention.rel_shift": ["Model", "Module", "arange", "class", "def", "device", "dtype", "index_select", "klen", "long", "nn", "rel_shift", "reshape", "return", "shape", "torch", "x", "x_size"], "xlnet/modeling_xlnet.py:XLNetRelativeAttention.rel_shift_bnij": ["Model", "Module", "arange", "class", "def", "device", "dtype", "index_select", "klen", "long", "nn", "rel_shift_bnij", "reshape", "return", "shape", "torch", "x", "x_size"], "xlnet/modeling_xlnet.py:XLNetRelativeAttention.rel_attn_core": ["False", "Model", "Module", "None", "ac", "attn_mask", "attn_prob", "attn_score", "attn_vec", "bd", "bnij", "class", "def", "dim", "dropout", "dtype", "ef", "einsum", "else", "float16", "functional", "ibnd", "ibns", "if", "ijbn", "ijbs", "is", "jbnd", "k_head_h", "k_head_r", "klen", "nn", "not", "output_attentions", "q_head", "r_r_bias", "r_s_bias", "r_w_bias", "rel_attn_core", "rel_shift_bnij", "return", "scale", "seg_embed", "seg_mat", "self", "shape", "snd", "softmax", "torch", "v_head_h"], "xlnet/modeling_xlnet.py:XLNetRelativeAttention.post_attention": ["Model", "Module", "True", "attn_out", "attn_vec", "class", "def", "dropout", "einsum", "h", "hnd", "ibh", "ibnd", "if", "layer_norm", "nn", "o", "output", "post_attention", "residual", "return", "self", "torch"], "xlnet/modeling_xlnet.py:XLNetRelativeAttention.forward": ["False", "Model", "Module", "None", "and", "attn_mask", "attn_mask_g", "attn_mask_h", "attn_prob", "attn_prob_g", "attn_prob_h", "attn_vec", "attn_vec_g", "attn_vec_h", "cat", "class", "def", "dim", "dtype", "einsum", "else", "forward", "g", "h", "hnd", "ibh", "ibnd", "if", "is", "k", "k_head_h", "k_head_r", "lbnd", "mbnd", "mems", "mlb", "nn", "not", "output_attentions", "output_g", "output_h", "outputs", "post_attention", "q", "q_head_g", "q_head_h", "r", "rel_attn_core", "return", "seg_mat", "self", "target_mapping", "torch", "type", "v", "v_head_h"], "xlnet/modeling_xlnet.py:XLNetFeedForward.__init__": ["ACT2FN", "Dropout", "LayerNorm", "Linear", "Model", "Module", "__init__", "activation_function", "class", "config", "d_inner", "d_model", "def", "dropout", "else", "eps", "ff_activation", "if", "isinstance", "layer_1", "layer_2", "layer_norm", "layer_norm_eps", "nn", "self", "str", "super"], "xlnet/modeling_xlnet.py:XLNetFeedForward.forward": ["Model", "Module", "activation_function", "class", "def", "dropout", "forward", "inp", "layer_1", "layer_2", "layer_norm", "nn", "output", "return", "self"], "xlnet/modeling_xlnet.py:XLNetLayer.__init__": ["Dropout", "Model", "ModelFeedForward", "ModelRelativeAttention", "Module", "__init__", "chunk_size_feed_forward", "class", "config", "def", "dropout", "ff", "nn", "rel_attn", "self", "seq_len_dim", "super"], "xlnet/modeling_xlnet.py:XLNetLayer.forward": ["False", "Model", "Module", "None", "apply_chunking_to_forward", "attn_mask_g", "attn_mask_h", "chunk_size_feed_forward", "class", "def", "ff_chunk", "forward", "if", "is", "mems", "nn", "not", "output_attentions", "output_g", "output_h", "outputs", "r", "rel_attn", "return", "seg_mat", "self", "seq_len_dim", "target_mapping"], "xlnet/modeling_xlnet.py:XLNetLayer.ff_chunk": ["Model", "Module", "class", "def", "ff", "ff_chunk", "nn", "output_x", "return", "self"], "xlnet/modeling_xlnet.py:XLNetPoolerStartLogits.__init__": ["Linear", "Model", "ModelConfig", "Module", "__init__", "class", "config", "def", "dense", "hidden_size", "nn", "self", "super"], "xlnet/modeling_xlnet.py:XLNetPoolerStartLogits.forward": ["FloatTensor", "Model", "Module", "None", "class", "def", "dense", "dtype", "else", "float16", "forward", "hidden_states", "if", "is", "nn", "not", "p_mask", "return", "self", "squeeze", "torch", "x"], "xlnet/modeling_xlnet.py:XLNetPoolerEndLogits.__init__": ["LayerNorm", "Linear", "Model", "ModelConfig", "Module", "Tanh", "__init__", "activation", "class", "config", "def", "dense_0", "dense_1", "eps", "hidden_size", "layer_norm_eps", "nn", "self", "super"], "xlnet/modeling_xlnet.py:XLNetPoolerEndLogits.forward": ["FloatTensor", "LayerNorm", "LongTensor", "Model", "Module", "None", "One", "activation", "assert", "be", "cat", "class", "def", "dense_0", "dense_1", "dim", "dtype", "else", "expand", "float16", "forward", "gather", "hidden_states", "hsz", "if", "is", "nn", "not", "of", "or", "p_mask", "return", "self", "shape", "should", "slen", "squeeze", "start_positions", "start_states", "torch", "x"], "xlnet/modeling_xlnet.py:XLNetPoolerAnswerClass.__init__": ["False", "Linear", "Model", "ModelConfig", "Module", "Tanh", "__init__", "activation", "bias", "class", "config", "def", "dense_0", "dense_1", "hidden_size", "nn", "self", "super"], "xlnet/modeling_xlnet.py:XLNetPoolerAnswerClass.forward": ["FloatTensor", "LongTensor", "Model", "Module", "None", "One", "activation", "assert", "be", "cat", "class", "cls_index", "cls_token_state", "def", "dense_0", "dense_1", "dim", "else", "expand", "forward", "gather", "hidden_states", "hsz", "if", "is", "nn", "not", "of", "or", "return", "self", "shape", "should", "squeeze", "start_positions", "start_states", "torch", "x"], "xlnet/modeling_xlnet.py:XLNetSequenceSummary.__init__": ["Callable", "Dropout", "Identity", "Linear", "Model", "ModelConfig", "Module", "None", "NotImplementedError", "__init__", "activation", "activation_string", "and", "attn", "class", "config", "def", "else", "first_dropout", "get_activation", "getattr", "hasattr", "hidden_size", "if", "last", "last_dropout", "nn", "num_classes", "num_labels", "raise", "self", "summary", "summary_activation", "summary_first_dropout", "summary_last_dropout", "summary_proj_to_labels", "summary_type", "summary_use_proj", "super"], "xlnet/modeling_xlnet.py:XLNetSequenceSummary.forward": ["FloatTensor", "LongTensor", "Model", "Module", "None", "NotImplementedError", "activation", "attn", "class", "cls_index", "def", "dim", "dtype", "elif", "else", "expand", "first", "first_dropout", "forward", "full_like", "gather", "hidden_states", "if", "is", "last", "last_dropout", "long", "mean", "nn", "output", "raise", "return", "self", "shape", "size", "squeeze", "summary", "summary_type", "torch", "unsqueeze"], "xlnet/modeling_xlnet.py:XLNetPreTrainedModel._init_weights": ["Model", "ModelModel", "ModelRelativeAttention", "PreTrainedModel", "_init_weights", "class", "config", "def", "elif", "for", "if", "in", "init", "initializer_range", "isinstance", "k", "mask_emb", "mean", "module", "normal_", "o", "param", "q", "r", "r_r_bias", "r_s_bias", "r_w_bias", "seg_embed", "self", "std", "super", "v"], "xlnet/modeling_xlnet.py:XLNetModel.__init__": ["Dropout", "Embedding", "FloatTensor", "Model", "ModelLayer", "ModelPreTrainedModel", "ModuleList", "Parameter", "_", "__init__", "attn_type", "bi_data", "clamp_len", "class", "config", "d_model", "def", "dropout", "for", "in", "layer", "mask_emb", "mem_len", "n_layer", "nn", "post_init", "range", "reuse_len", "same_length", "self", "super", "torch", "vocab_size", "word_embedding"], "xlnet/modeling_xlnet.py:XLNetModel.get_input_embeddings": ["Model", "ModelPreTrainedModel", "class", "def", "get_input_embeddings", "return", "self", "word_embedding"], "xlnet/modeling_xlnet.py:XLNetModel.set_input_embeddings": ["Model", "ModelPreTrainedModel", "class", "def", "new_embeddings", "self", "set_input_embeddings", "word_embedding"], "xlnet/modeling_xlnet.py:XLNetModel.create_mask": ["Model", "ModelPreTrainedModel", "class", "create_mask", "def", "device", "else", "if", "mask", "mask_lo", "mlen", "ones", "qlen", "return", "same_length", "self", "torch", "tril", "triu_"], "xlnet/modeling_xlnet.py:XLNetModel.cache_mem": ["Model", "ModelPreTrainedModel", "None", "and", "cache_mem", "cat", "class", "curr_out", "cutoff", "def", "detach", "dim", "else", "if", "is", "mem_len", "new_mem", "not", "or", "prev_mem", "return", "reuse_len", "self", "torch"], "xlnet/modeling_xlnet.py:XLNetModel.positional_embedding": ["Model", "ModelPreTrainedModel", "None", "bsz", "cat", "class", "cos", "d", "def", "dim", "einsum", "expand", "i", "id", "if", "inv_freq", "is", "not", "pos_emb", "pos_seq", "positional_embedding", "return", "sin", "sinusoid_inp", "torch"], "xlnet/modeling_xlnet.py:XLNetModel.relative_positional_encoding": ["Model", "ModelPreTrainedModel", "None", "Unknown", "ValueError", "arange", "attn_type", "beg", "bi", "bi_data", "bsz", "bwd_pos_emb", "bwd_pos_seq", "cat", "clamp", "clamp_len", "class", "d_model", "def", "dim", "dtype", "elif", "else", "end", "f", "float", "freq_seq", "fwd_pos_emb", "fwd_pos_seq", "if", "int64", "inv_freq", "is", "klen", "not", "pos_emb", "positional_embedding", "pow", "qlen", "raise", "relative_positional_encoding", "return", "self", "torch", "uni"], "xlnet/modeling_xlnet.py:XLNetModel.forward": ["BERT", "Model", "ModelOutput", "ModelPreTrainedModel", "None", "Please", "Tensor", "Unsupported", "ValueError", "You", "added", "and", "append", "assert", "at", "att_stream", "attention", "attention_mask", "attentions", "attn_mask", "attn_mask_g", "attn_mask_h", "attn_type", "bi", "bool", "both", "bsz", "cache_mem", "can", "cannot", "cat", "cat_ids", "choose", "class", "compatibility", "config", "contiguous", "create_mask", "data_mask", "def", "device", "dim", "dropout", "dtype", "dtype_float", "either", "elif", "else", "enumerate", "expand", "eye", "f", "for", "forward", "functional", "h", "have", "hidden_states", "hs", "i", "if", "in", "input_ids", "input_mask", "inputs_embeds", "is", "klen", "kwargs", "last_hidden_state", "layer", "layer_module", "len", "long", "mask_emb", "mem_pad", "mems", "mems_mask", "mlen", "new_mems", "nn", "non_tgt_mask", "not", "num_classes", "of", "one", "one_hot", "only", "or", "output", "output_attentions", "output_g", "output_h", "output_hidden_states", "outputs", "padding", "perm_mask", "permute", "pos_emb", "qlen", "r", "raise", "relative_positional_encoding", "return", "return_dict", "same", "seg_mat", "self", "shape", "specify", "t", "target_mapping", "the", "time", "to", "token_type_ids", "torch", "training", "transpose", "tuple", "type", "uni", "use", "use_mems", "use_mems_eval", "use_mems_train", "use_return_dict", "uses", "v", "with", "word_emb_k", "word_emb_q", "word_embedding", "zeros"], "xlnet/modeling_xlnet.py:XLNetLMHeadModel.__init__": ["GenerationMixin", "Linear", "Model", "ModelModel", "ModelPreTrainedModel", "True", "__init__", "attn_type", "bias", "class", "config", "d_model", "def", "lm_loss", "nn", "post_init", "same_length", "self", "super", "transformer", "vocab_size"], "xlnet/modeling_xlnet.py:XLNetLMHeadModel.get_output_embeddings": ["GenerationMixin", "Model", "ModelPreTrainedModel", "class", "def", "get_output_embeddings", "lm_loss", "return", "self"], "xlnet/modeling_xlnet.py:XLNetLMHeadModel.set_output_embeddings": ["GenerationMixin", "Model", "ModelPreTrainedModel", "class", "def", "lm_loss", "new_embeddings", "self", "set_output_embeddings"], "xlnet/modeling_xlnet.py:XLNetLMHeadModel.prepare_inputs_for_generation": ["False", "GenerationMixin", "Model", "ModelPreTrainedModel", "None", "attention_mask", "cat", "class", "def", "device", "dim", "dtype", "dummy_token", "effective_batch_size", "else", "float", "for", "if", "in", "input_ids", "is_first_iteration", "items", "key", "kwargs", "layer_past", "long", "mems", "model_inputs", "not", "offset", "past_key_values", "perm_mask", "pop", "prepare_inputs_for_generation", "return", "self", "sequence_length", "shape", "target_mapping", "torch", "tuple", "use_cache", "use_mems", "value", "zeros"], "xlnet/modeling_xlnet.py:XLNetLMHeadModel.forward": ["CrossEntropyLoss", "GenerationMixin", "Model", "ModelOutput", "ModelPreTrainedModel", "None", "Tensor", "attention_mask", "attentions", "bool", "class", "config", "def", "else", "forward", "hidden_states", "if", "input_ids", "input_mask", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "lm_loss", "logits", "logits_to_keep", "loss", "loss_fct", "mems", "not", "output", "output_attentions", "output_hidden_states", "perm_mask", "r", "return", "return_dict", "self", "size", "slice", "slice_indices", "target_mapping", "token_type_ids", "torch", "transformer", "transformer_outputs", "tuple", "use_mems", "use_return_dict", "view"], "xlnet/modeling_xlnet.py:XLNetLMHeadModel._reorder_cache": ["GenerationMixin", "Model", "ModelPreTrainedModel", "Tensor", "_reorder_cache", "beam_idx", "class", "def", "device", "for", "in", "index_select", "layer_past", "list", "mems", "return", "to", "torch"], "xlnet/modeling_xlnet.py:XLNetForSequenceClassification.__init__": ["Linear", "Model", "ModelModel", "ModelPreTrainedModel", "ModelSequenceSummary", "__init__", "class", "config", "d_model", "def", "logits_proj", "nn", "num_labels", "post_init", "self", "sequence_summary", "super", "transformer"], "xlnet/modeling_xlnet.py:XLNetForSequenceClassification.forward": ["BCEWithLogitsLoss", "CrossEntropyLoss", "MSELoss", "Model", "ModelOutput", "ModelPreTrainedModel", "None", "Tensor", "and", "attention_mask", "attentions", "bool", "class", "config", "def", "dtype", "elif", "else", "forward", "hidden_states", "if", "input_ids", "input_mask", "inputs_embeds", "int", "is", "kwargs", "labels", "logits", "logits_proj", "long", "loss", "loss_fct", "mems", "multi_label_classification", "not", "num_labels", "or", "output", "output_attentions", "output_hidden_states", "perm_mask", "problem_type", "r", "regression", "return", "return_dict", "self", "sequence_summary", "single_label_classification", "squeeze", "target_mapping", "token_type_ids", "torch", "transformer", "transformer_outputs", "tuple", "use_mems", "use_return_dict", "view"], "xlnet/modeling_xlnet.py:XLNetForTokenClassification.__init__": ["Linear", "Model", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "classifier", "config", "def", "hidden_size", "nn", "num_labels", "post_init", "self", "super", "transformer"], "xlnet/modeling_xlnet.py:XLNetForTokenClassification.forward": ["CrossEntropyLoss", "Model", "ModelOutput", "ModelPreTrainedModel", "None", "Tensor", "attention_mask", "attentions", "bool", "class", "classifier", "config", "def", "else", "forward", "hidden_states", "if", "input_ids", "input_mask", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "mems", "not", "num_labels", "output", "output_attentions", "output_hidden_states", "outputs", "perm_mask", "r", "return", "return_dict", "self", "sequence_output", "target_mapping", "token_type_ids", "torch", "transformer", "tuple", "use_mems", "use_return_dict", "view"], "xlnet/modeling_xlnet.py:XLNetForMultipleChoice.__init__": ["Linear", "Model", "ModelModel", "ModelPreTrainedModel", "ModelSequenceSummary", "__init__", "class", "config", "d_model", "def", "logits_proj", "nn", "post_init", "self", "sequence_summary", "super", "transformer"], "xlnet/modeling_xlnet.py:XLNetForMultipleChoice.forward": ["CrossEntropyLoss", "Model", "ModelOutput", "ModelPreTrainedModel", "None", "Tensor", "attention_mask", "attentions", "bool", "class", "config", "def", "else", "flat_attention_mask", "flat_input_ids", "flat_input_mask", "flat_inputs_embeds", "flat_token_type_ids", "forward", "hidden_states", "if", "input_ids", "input_mask", "inputs_embeds", "is", "kwargs", "labels", "logits", "logits_proj", "loss", "loss_fct", "mems", "not", "num_choices", "output", "output_attentions", "output_hidden_states", "perm_mask", "r", "reshaped_logits", "return", "return_dict", "self", "sequence_summary", "shape", "size", "target_mapping", "token_type_ids", "torch", "transformer", "transformer_outputs", "tuple", "use_mems", "use_return_dict", "view"], "xlnet/modeling_xlnet.py:XLNetForQuestionAnsweringSimple.__init__": ["Linear", "Model", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "config", "def", "hidden_size", "nn", "num_labels", "post_init", "qa_outputs", "self", "super", "transformer"], "xlnet/modeling_xlnet.py:XLNetForQuestionAnsweringSimple.forward": ["CrossEntropyLoss", "Model", "ModelOutput", "ModelPreTrainedModel", "None", "Tensor", "and", "attention_mask", "attentions", "bool", "clamp", "class", "config", "contiguous", "def", "dim", "else", "end_logits", "end_loss", "end_positions", "forward", "hidden_states", "if", "ignore_index", "ignored_index", "input_ids", "input_mask", "inputs_embeds", "is", "kwargs", "len", "logits", "loss", "loss_fct", "mems", "not", "output", "output_attentions", "output_hidden_states", "outputs", "perm_mask", "qa_outputs", "r", "return", "return_dict", "self", "sequence_output", "size", "split", "squeeze", "start_logits", "start_loss", "start_positions", "target_mapping", "token_type_ids", "torch", "total_loss", "transformer", "tuple", "use_mems", "use_return_dict"], "xlnet/modeling_xlnet.py:XLNetForQuestionAnswering.__init__": ["Model", "ModelModel", "ModelPoolerAnswerClass", "ModelPoolerEndLogits", "ModelPoolerStartLogits", "ModelPreTrainedModel", "__init__", "answer_class", "class", "config", "def", "end_logits", "end_n_top", "post_init", "self", "start_logits", "start_n_top", "super", "transformer"], "xlnet/modeling_xlnet.py:XLNetForQuestionAnswering.forward": ["BCEWithLogitsLoss", "CrossEntropyLoss", "Model", "ModelOutput", "ModelPreTrainedModel", "None", "Tensor", "and", "answer_class", "attention_mask", "attentions", "bh", "bl", "blh", "bool", "bsz", "class", "cls_index", "cls_logits", "cls_loss", "config", "def", "dim", "einsum", "else", "end_log_probs", "end_logits", "end_loss", "end_n_top", "end_positions", "end_top_index", "end_top_log_probs", "expand", "expand_as", "for", "forward", "functional", "gather", "hidden_states", "hidden_states_expanded", "hsz", "if", "in", "input_ids", "input_mask", "inputs_embeds", "is", "is_impossible", "kwargs", "loss", "loss_fct", "loss_fct_cls", "mems", "nn", "not", "output_attentions", "output_hidden_states", "outputs", "p_mask", "perm_mask", "r", "return", "return_dict", "self", "size", "slen", "softmax", "squeeze_", "start_log_probs", "start_logits", "start_loss", "start_n_top", "start_positions", "start_states", "start_top_index", "start_top_index_exp", "start_top_log_probs", "target_mapping", "token_type_ids", "topk", "torch", "total_loss", "transformer", "transformer_outputs", "tuple", "unsqueeze", "use_mems", "use_return_dict", "view", "x"], "xlstm/modeling_xlstm.py:small_init_method": ["Model_init_method", "def", "dim", "init", "init_", "mean", "normal_", "return", "std", "tensor"], "xlstm/modeling_xlstm.py:wang_init_method": ["Model_init_method", "def", "dim", "init", "init_", "mean", "n_layers", "normal_", "return", "std", "tensor"], "xlstm/modeling_xlstm.py:xLSTMPreTrainedModel._module_name_map": ["Model", "PreTrainedModel", "_module_name_map", "class", "def", "for", "if", "in", "is", "mod", "module", "name", "named_modules", "return", "self"], "xlstm/modeling_xlstm.py:xLSTMPreTrainedModel._init_weights": ["Embedding", "Linear", "Model", "ModelRMSNorm", "None", "PreTrainedModel", "_init_weights", "_layer_normalize", "_module_name_map", "and", "bias", "class", "config", "copy_", "def", "device", "dim", "dtype", "elif", "embeddings", "fgate", "fused", "gate", "hasattr", "hidden_size", "if", "igate", "in", "init", "is", "isinstance", "linspace", "module", "n_layers", "nn", "not", "num_heads", "num_hidden_layers", "ones_", "ones_like", "or", "out_proj", "proj_down", "self", "shape", "single", "small_init_method", "to", "torch", "wang_init_method", "weight", "weight_mode", "zeros_"], "xlstm/modeling_xlstm.py:xLSTMCache.__init__": ["Model", "ModelConfig", "None", "__init__", "bfloat16", "class", "config", "def", "device", "dtype", "for", "in", "int", "kwargs", "layer", "max_batch_size", "num_heads", "num_hidden_layers", "qk_head_dim", "range", "rnn_state", "self", "seqlen_offset", "str", "torch", "v_head_dim", "zeros"], "xlstm/modeling_xlstm.py:xLSTMCache.reset": ["Model", "class", "def", "for", "in", "layer", "reset", "rnn_state", "self", "torch", "zeros_like"], "xlstm/modeling_xlstm.py:xLSTMModel.__init__": ["Embedding", "False", "Model", "ModelBlock", "ModelPreTrainedModel", "ModelRMSNorm", "ModuleList", "_", "__init__", "blocks", "class", "config", "def", "embedding_dim", "embeddings", "eps", "for", "gradient_checkpointing", "hidden_size", "in", "nn", "norm_eps", "num_blocks", "out_norm", "post_init", "range", "self", "super", "vocab_size"], "xlstm/modeling_xlstm.py:xLSTMModel.get_input_embeddings": ["Model", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "return", "self"], "xlstm/modeling_xlstm.py:xLSTMModel.set_input_embeddings": ["Model", "ModelPreTrainedModel", "class", "def", "embeddings", "new_embedding", "self", "set_input_embeddings"], "xlstm/modeling_xlstm.py:xLSTMModel.forward": ["False", "LongTensor", "Model", "ModelCache", "ModelOutput", "ModelPreTrainedModel", "Model_block", "None", "ValueError", "You", "all_hidden_states", "and", "blocks", "bool", "cache_params", "class", "config", "copy_", "def", "device", "dtype", "else", "embeddings", "enumerate", "exactly", "final_state", "for", "forward", "gradient_checkpointing", "hidden_states", "hidden_states_chunk", "if", "in", "input_ids", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layer_idx", "len", "local_rnn_state", "max_batch_size", "max_inference_chunksize", "min", "must", "no_grad", "not", "of", "offset", "one", "or", "out_norm", "output_hidden_states", "r", "raise", "range", "return", "rnn_state", "rnn_state_initial", "self", "seqlen_offset", "shape", "size", "specify", "state", "state_idx", "torch", "training", "tuple", "use_cache", "while", "with", "zeros_like"], "xlstm/modeling_xlstm.py:xLSTMForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "Model", "ModelModel", "ModelPreTrainedModel", "__init__", "backbone", "bias", "class", "config", "def", "hidden_size", "lm_head", "nn", "post_init", "self", "super", "vocab_size"], "xlstm/modeling_xlstm.py:xLSTMForCausalLM.get_output_embeddings": ["GenerationMixin", "Model", "ModelPreTrainedModel", "class", "def", "get_output_embeddings", "lm_head", "return", "self"], "xlstm/modeling_xlstm.py:xLSTMForCausalLM.set_output_embeddings": ["GenerationMixin", "Model", "ModelPreTrainedModel", "class", "def", "lm_head", "new_embeddings", "self", "set_output_embeddings"], "xlstm/modeling_xlstm.py:xLSTMForCausalLM.get_input_embeddings": ["GenerationMixin", "Model", "ModelPreTrainedModel", "backbone", "class", "def", "get_input_embeddings", "return", "self"], "xlstm/modeling_xlstm.py:xLSTMForCausalLM.set_input_embeddings": ["GenerationMixin", "Model", "ModelPreTrainedModel", "backbone", "class", "def", "new_embeddings", "return", "self", "set_input_embeddings"], "xlstm/modeling_xlstm.py:xLSTMForCausalLM.prepare_inputs_for_generation": ["GenerationMixin", "Model", "ModelCache", "ModelPreTrainedModel", "None", "and", "attention_mask", "cache_params", "class", "def", "else", "for", "if", "in", "input_ids", "inputs_embeds", "is", "items", "key", "kwargs", "model_inputs", "not", "prepare_inputs_for_generation", "return", "self", "update", "use_cache", "value"], "xlstm/modeling_xlstm.py:xLSTMForCausalLM.forward": ["CrossEntropyLoss", "FloatTensor", "GenerationMixin", "LongTensor", "Model", "ModelCache", "ModelCausalLMOutput", "ModelPreTrainedModel", "Model_outputs", "None", "and", "backbone", "bool", "cache_params", "class", "config", "contiguous", "def", "device", "dtype", "else", "float", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "lm_head", "logits", "loss", "loss_fct", "max_inference_chunksize", "min", "no_grad", "not", "offset", "output_hidden_states", "output_logit_soft_cap", "r", "return", "self", "shape", "shift_labels", "shift_logits", "size", "soft_cap", "to", "torch", "training", "tuple", "use_cache", "view", "weight", "while", "with"], "xmod/modeling_xmod.py:XmodEmbeddings.__init__": ["Dropout", "Embedding", "False", "LayerNorm", "ModelEmbeddings", "Module", "__init__", "arange", "class", "config", "def", "dropout", "dtype", "eps", "expand", "hidden_dropout_prob", "hidden_size", "layer_norm_eps", "long", "max_position_embeddings", "nn", "pad_token_id", "padding_idx", "persistent", "position_embeddings", "position_ids", "register_buffer", "self", "size", "super", "token_type_embeddings", "token_type_ids", "torch", "type_vocab_size", "vocab_size", "word_embeddings", "zeros"], "xmod/modeling_xmod.py:XmodEmbeddings.forward": ["FloatTensor", "LayerNorm", "LongTensor", "ModelEmbeddings", "Module", "None", "Tensor", "batch_size", "buffered_token_type_ids", "class", "create_position_ids_from_input_ids", "create_position_ids_from_inputs_embeds", "def", "device", "dim", "dropout", "dtype", "else", "embeddings", "expand", "forward", "gather", "hasattr", "if", "index", "input_ids", "input_shape", "inputs_embeds", "int", "is", "long", "nn", "not", "padding_idx", "past_key_values_length", "position_embeddings", "position_ids", "return", "self", "seq_length", "shape", "size", "token_type_embeddings", "token_type_ids", "torch", "word_embeddings", "zeros"], "xmod/modeling_xmod.py:XmodEmbeddings.create_position_ids_from_inputs_embeds": ["ModelEmbeddings", "Module", "arange", "class", "create_position_ids_from_inputs_embeds", "def", "device", "dtype", "expand", "input_shape", "inputs_embeds", "long", "nn", "padding_idx", "position_ids", "return", "sequence_length", "size", "torch", "unsqueeze"], "xmod/modeling_xmod.py:XmodEmbeddings.create_position_ids_from_input_ids": ["ModelEmbeddings", "Module", "class", "create_position_ids_from_input_ids", "cumsum", "def", "dim", "incremental_indices", "input_ids", "int", "long", "mask", "ne", "nn", "padding_idx", "past_key_values_length", "return", "torch", "type_as"], "xmod/modeling_xmod.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "contiguous", "def", "dim", "dropout", "float", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "shape", "size", "softmax", "torch", "training", "transpose", "value"], "xmod/modeling_xmod.py:XmodSelfAttention.__init__": ["Dropout", "False", "Linear", "ModelSelfAttention", "Module", "None", "The", "ValueError", "__init__", "a", "all_head_size", "and", "attention", "attention_head_size", "attention_probs_dropout_prob", "class", "config", "def", "dropout", "embedding_size", "f", "hasattr", "heads", "hidden", "hidden_size", "if", "int", "is", "is_causal", "is_decoder", "key", "layer_idx", "multiple", "nn", "not", "num_attention_heads", "number", "of", "query", "raise", "scaling", "self", "size", "super", "the", "value"], "xmod/modeling_xmod.py:XmodSelfAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Cache", "Callable", "EncoderDecoderCache", "FloatTensor", "ModelSelfAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "_attn_implementation", "attention_head_size", "attention_interface", "attention_mask", "attn_output", "attn_weights", "cache_position", "class", "config", "contiguous", "current_past_key_values", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "hidden_shape", "hidden_states", "if", "input_shape", "is", "isinstance", "key", "key_layer", "kwargs", "layer_idx", "nn", "not", "p", "past_key_values", "query", "query_layer", "reshape", "return", "scaling", "self", "self_attention_cache", "shape", "torch", "training", "transpose", "tuple", "update", "value", "value_layer", "view"], "xmod/modeling_xmod.py:XmodCrossAttention.__init__": ["Dropout", "False", "Linear", "ModelCrossAttention", "Module", "None", "The", "ValueError", "__init__", "a", "all_head_size", "and", "attention", "attention_head_size", "attention_probs_dropout_prob", "class", "config", "def", "dropout", "embedding_size", "f", "hasattr", "heads", "hidden", "hidden_size", "if", "int", "is", "is_causal", "key", "layer_idx", "multiple", "nn", "not", "num_attention_heads", "number", "of", "query", "raise", "scaling", "self", "size", "super", "the", "value"], "xmod/modeling_xmod.py:XmodCrossAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "EncoderDecoderCache", "False", "FloatTensor", "ModelCrossAttention", "Module", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "_attn_implementation", "and", "attention_head_size", "attention_interface", "attention_mask", "attn_output", "attn_weights", "bsz", "class", "config", "contiguous", "cross_attention_cache", "def", "dropout", "eager", "eager_attention_forward", "else", "encoder_hidden_states", "forward", "get", "hidden_states", "if", "is", "is_updated", "key", "key_layer", "keys", "kv_input_shape", "kwargs", "layer_idx", "layers", "nn", "not", "p", "past_key_values", "q_input_shape", "query", "query_layer", "reshape", "return", "scaling", "self", "shape", "src_len", "tgt_len", "torch", "training", "transpose", "tuple", "update", "value", "value_layer", "values", "view"], "xmod/modeling_xmod.py:XmodSelfOutput.__init__": ["Dropout", "LayerNorm", "Linear", "ModelSelfOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "eps", "hidden_dropout_prob", "hidden_size", "layer_norm_eps", "nn", "self", "super"], "xmod/modeling_xmod.py:XmodSelfOutput.forward": ["ModelSelfOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "xmod/modeling_xmod.py:XmodAttention.__init__": ["False", "ModelAttention", "ModelCrossAttention", "ModelSelfAttention", "ModelSelfOutput", "Module", "None", "__init__", "attention_class", "class", "config", "def", "else", "if", "is_causal", "is_cross_attention", "layer_idx", "nn", "output", "pre_norm", "self", "super"], "xmod/modeling_xmod.py:XmodAttention.forward": ["FloatTensor", "LayerNorm", "ModelAttention", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attention_output", "attn_weights", "cache_position", "class", "def", "else", "encoder_attention_mask", "encoder_hidden_states", "forward", "hidden_states", "if", "is_cross_attention", "kwargs", "nn", "not", "output", "past_key_values", "pre_norm", "residual", "return", "self", "torch", "tuple"], "xmod/modeling_xmod.py:XmodIntermediate.__init__": ["ACT2FN", "Linear", "ModelIntermediate", "Module", "__init__", "class", "config", "def", "dense", "else", "hidden_act", "hidden_size", "if", "intermediate_act_fn", "intermediate_size", "isinstance", "nn", "self", "str", "super"], "xmod/modeling_xmod.py:XmodIntermediate.forward": ["ModelIntermediate", "Module", "Tensor", "class", "def", "dense", "forward", "hidden_states", "intermediate_act_fn", "nn", "return", "self", "torch"], "xmod/modeling_xmod.py:XmodAdapter.__init__": ["ACT2FN", "Linear", "ModelAdapter", "Module", "__init__", "adapter_act_fn", "adapter_reduction_factor", "bottleneck_size", "class", "config", "def", "dense1", "dense2", "else", "hidden_act", "hidden_size", "if", "isinstance", "nn", "self", "str", "super"], "xmod/modeling_xmod.py:XmodAdapter.forward": ["ModelAdapter", "Module", "Tensor", "adapter_act_fn", "class", "def", "dense1", "dense2", "forward", "hidden_states", "nn", "return", "self", "torch"], "xmod/modeling_xmod.py:XmodOutput.__init__": ["Dropout", "LayerNorm", "Linear", "ModelAdapter", "ModelOutput", "Module", "ModuleDict", "None", "__init__", "adapter_layer_norm", "adapter_modules", "adapter_reuse_layer_norm", "class", "config", "def", "dense", "dropout", "else", "eps", "for", "hidden_dropout_prob", "hidden_size", "if", "in", "intermediate_size", "language", "languages", "layer_norm_eps", "ln_before_adapter", "nn", "self", "str", "super"], "xmod/modeling_xmod.py:XmodOutput.forward": ["ModelOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "lang_adapter", "lang_ids", "nn", "return", "self", "torch"], "xmod/modeling_xmod.py:XmodOutput.lang_adapter": ["LayerNorm", "ModelOutput", "Module", "None", "Tensor", "True", "adapter_layer_norm", "adapter_modules", "adapter_reuse_layer_norm", "append", "cat", "class", "def", "dropout", "elif", "enumerate", "for", "hidden_states", "i", "if", "in", "int", "is", "item", "keys", "lang", "lang_adapter", "lang_id", "lang_ids", "lang_lengths", "lang_wise_outputs", "list", "ln_before_adapter", "nn", "not", "residual", "return", "return_counts", "self", "split", "split_hidden_state", "split_hidden_states", "tolist", "torch", "unique_consecutive", "zip"], "xmod/modeling_xmod.py:XmodLayer.__init__": ["False", "GradientCheckpointingLayer", "ModelAttention", "ModelIntermediate", "ModelLayer", "ModelOutput", "None", "True", "ValueError", "__init__", "a", "add_cross_attention", "added", "as", "attention", "be", "chunk_size_feed_forward", "class", "config", "cross", "crossattention", "decoder", "def", "f", "if", "intermediate", "is", "is_causal", "is_cross_attention", "is_decoder", "layer_idx", "model", "not", "output", "pre_norm", "raise", "self", "seq_len_dim", "should", "super", "used"], "xmod/modeling_xmod.py:XmodLayer.forward": ["FloatTensor", "GradientCheckpointingLayer", "If", "LayerNorm", "ModelLayer", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "ValueError", "_", "add_cross_attention", "and", "apply_chunking_to_forward", "are", "attention", "attention_mask", "attention_output", "be", "by", "cache_position", "chunk_size_feed_forward", "class", "config", "cross", "cross_attention_output", "crossattention", "def", "encoder_attention_mask", "encoder_hidden_states", "f", "feed_forward_chunk", "forward", "has", "hasattr", "hidden_states", "if", "instantiated", "intermediate_output", "is", "is_decoder", "kwargs", "lang_ids", "layer_output", "layers", "not", "output", "passed", "past_key_values", "pre_norm", "raise", "residual", "return", "self", "self_attention_output", "seq_len_dim", "setting", "to", "torch", "tuple", "with"], "xmod/modeling_xmod.py:XmodLayer.feed_forward_chunk": ["GradientCheckpointingLayer", "ModelLayer", "attention_output", "class", "def", "feed_forward_chunk", "intermediate", "return", "self"], "xmod/modeling_xmod.py:XmodEncoder.__init__": ["LayerNorm", "ModelEncoder", "ModelLayer", "Module", "ModuleList", "__init__", "class", "config", "def", "eps", "for", "hidden_size", "i", "if", "in", "is_pre_norm", "layer", "layer_idx", "layer_norm_eps", "nn", "num_hidden_layers", "pre_norm", "range", "self", "super"], "xmod/modeling_xmod.py:XmodEncoder.forward": ["BaseModelOutputWithPastAndCrossAttentions", "FloatTensor", "LayerNorm", "ModelEncoder", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "bool", "cache_position", "class", "def", "else", "encoder_attention_mask", "encoder_hidden_states", "enumerate", "for", "forward", "hidden_states", "i", "if", "in", "is_pre_norm", "kwargs", "lang_ids", "last_hidden_state", "layer", "layer_module", "nn", "past_key_values", "return", "self", "torch", "tuple", "use_cache"], "xmod/modeling_xmod.py:XmodPooler.__init__": ["Linear", "ModelPooler", "Module", "Tanh", "__init__", "activation", "class", "config", "def", "dense", "hidden_size", "nn", "self", "super"], "xmod/modeling_xmod.py:XmodPooler.forward": ["ModelPooler", "Module", "Tensor", "activation", "class", "def", "dense", "first_token_tensor", "forward", "hidden_states", "nn", "pooled_output", "return", "self", "torch"], "xmod/modeling_xmod.py:XmodPreTrainedModel._init_weights": ["ModelEmbeddings", "ModelLMHead", "ModelPreTrainedModel", "PreTrainedModel", "_init_weights", "arange", "bias", "class", "copy_", "def", "elif", "expand", "if", "init", "isinstance", "module", "position_ids", "self", "shape", "super", "token_type_ids", "torch", "zeros_"], "xmod/modeling_xmod.py:XmodPreTrainedModel.set_default_language": ["ModelPreTrainedModel", "PreTrainedModel", "Supported", "ValueError", "adapter", "an", "class", "config", "def", "default_language", "does", "f", "for", "have", "if", "in", "language", "languages", "list", "not", "raise", "self", "set_default_language", "str"], "xmod/modeling_xmod.py:XmodPreTrainedModel.freeze_embeddings_and_language_adapters": ["False", "Freezing", "ModelPreTrainedModel", "None", "PreTrainedModel", "adapter_layer_norm", "adapter_modules", "adapters", "class", "def", "embeddings", "encoder", "for", "freeze_embeddings_and_language_adapters", "if", "in", "info", "is", "layer", "logger", "not", "output", "parameter", "parameters", "requires_grad", "roberta", "self"], "xmod/modeling_xmod.py:XmodModel.__init__": ["False", "ModelEmbeddings", "ModelEncoder", "ModelModel", "ModelPooler", "ModelPreTrainedModel", "None", "True", "__init__", "add_pooling_layer", "class", "config", "def", "else", "embeddings", "encoder", "gradient_checkpointing", "if", "pooler", "post_init", "r", "self", "super"], "xmod/modeling_xmod.py:XmodModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "return", "self", "word_embeddings"], "xmod/modeling_xmod.py:XmodModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "self", "set_input_embeddings", "value", "word_embeddings"], "xmod/modeling_xmod.py:XmodModel.forward": ["BaseModelOutputWithPoolingAndCrossAttentions", "DynamicCache", "EncoderDecoderCache", "False", "FloatTensor", "Input", "LongTensor", "ModelModel", "ModelPreTrainedModel", "None", "Please", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "_create_attention_masks", "adapter_languages", "adapter_modules", "and", "arange", "attention_mask", "batch_size", "bool", "cache_position", "call", "class", "config", "def", "default_lang_id", "default_language", "device", "else", "embedding_output", "embeddings", "encoder", "encoder_attention_mask", "encoder_hidden_states", "encoder_outputs", "exactly", "forward", "get_seq_length", "if", "index", "input_ids", "input_shape", "inputs_embeds", "is", "is_decoder", "is_encoder_decoder", "keys", "kwargs", "lang_ids", "language", "last_hidden_state", "layer", "list", "must", "not", "of", "one", "ones", "or", "output", "past_key_values", "past_key_values_length", "pooled_output", "pooler", "pooler_output", "position_ids", "r", "raise", "return", "self", "seq_length", "sequence_output", "set_default_language", "shape", "specify", "token_type_ids", "torch", "tuple", "unknown", "use_cache"], "xmod/modeling_xmod.py:XmodModel._create_attention_masks": ["ModelModel", "ModelPreTrainedModel", "None", "_create_attention_masks", "attention_mask", "cache_position", "class", "config", "create_bidirectional_mask", "create_causal_mask", "def", "else", "embedding_output", "encoder_attention_mask", "encoder_hidden_states", "if", "input_embeds", "is", "is_decoder", "not", "past_key_values", "return", "self"], "xmod/modeling_xmod.py:XmodForCausalLM.__init__": ["False", "GenerationMixin", "If", "ModelForCausalLM", "ModelLMHead", "ModelLMHeadModel", "ModelModel", "ModelPreTrainedModel", "True", "__init__", "a", "add", "add_pooling_layer", "as", "class", "config", "def", "if", "is_decoder", "lm_head", "logger", "not", "post_init", "roberta", "self", "standalone", "super", "to", "use", "want", "warning", "you"], "xmod/modeling_xmod.py:XmodForCausalLM.get_output_embeddings": ["GenerationMixin", "ModelForCausalLM", "ModelPreTrainedModel", "class", "decoder", "def", "get_output_embeddings", "lm_head", "return", "self"], "xmod/modeling_xmod.py:XmodForCausalLM.set_output_embeddings": ["GenerationMixin", "ModelForCausalLM", "ModelPreTrainedModel", "class", "decoder", "def", "lm_head", "new_embeddings", "self", "set_output_embeddings"], "xmod/modeling_xmod.py:XmodForCausalLM.forward": ["BaseModelOutputWithPoolingAndCrossAttentions", "CausalLMOutputWithCrossAttentions", "False", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForCausalLM", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "cross_attentions", "def", "else", "encoder_attention_mask", "encoder_hidden_states", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "lang_ids", "last_hidden_state", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "not", "outputs", "past_key_values", "position_ids", "r", "return", "return_dict", "roberta", "self", "slice", "slice_indices", "token_type_ids", "torch", "tuple", "use_cache", "vocab_size"], "xmod/modeling_xmod.py:XmodForMaskedLM.__init__": ["False", "If", "ModelForMaskedLM", "ModelLMHead", "ModelModel", "ModelPreTrainedModel", "__init__", "add_pooling_layer", "attention", "bi", "class", "config", "def", "directional", "for", "if", "is_decoder", "lm_head", "logger", "make", "post_init", "roberta", "self", "super", "sure", "to", "use", "want", "warning", "you"], "xmod/modeling_xmod.py:XmodForMaskedLM.get_output_embeddings": ["ModelForMaskedLM", "ModelPreTrainedModel", "class", "decoder", "def", "get_output_embeddings", "lm_head", "return", "self"], "xmod/modeling_xmod.py:XmodForMaskedLM.set_output_embeddings": ["ModelForMaskedLM", "ModelPreTrainedModel", "class", "decoder", "def", "lm_head", "new_embeddings", "self", "set_output_embeddings"], "xmod/modeling_xmod.py:XmodForMaskedLM.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "MaskedLMOutput", "ModelForMaskedLM", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "attention_mask", "attentions", "class", "config", "def", "encoder_attention_mask", "encoder_hidden_states", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "lang_ids", "lm_head", "logits", "loss", "loss_fct", "masked_lm_loss", "not", "outputs", "position_ids", "prediction_scores", "r", "return", "return_dict", "roberta", "self", "sequence_output", "token_type_ids", "torch", "tuple", "view", "vocab_size"], "xmod/modeling_xmod.py:XmodLMHead.__init__": ["LayerNorm", "Linear", "ModelLMHead", "Module", "Parameter", "__init__", "bias", "class", "config", "decoder", "def", "dense", "eps", "hidden_size", "layer_norm", "layer_norm_eps", "nn", "self", "super", "torch", "vocab_size", "zeros"], "xmod/modeling_xmod.py:XmodLMHead.forward": ["ModelLMHead", "Module", "class", "decoder", "def", "dense", "features", "forward", "gelu", "kwargs", "layer_norm", "nn", "return", "self", "x"], "xmod/modeling_xmod.py:XmodForSequenceClassification.__init__": ["False", "ModelClassificationHead", "ModelForSequenceClassification", "ModelModel", "ModelPreTrainedModel", "__init__", "add_pooling_layer", "class", "classifier", "config", "def", "num_labels", "post_init", "roberta", "self", "super"], "xmod/modeling_xmod.py:XmodForSequenceClassification.forward": ["BCEWithLogitsLoss", "CrossEntropyLoss", "FloatTensor", "LongTensor", "MSELoss", "ModelForSequenceClassification", "ModelPreTrainedModel", "None", "SequenceClassifierOutput", "Tensor", "TransformersKwargs", "True", "Unpack", "and", "attention_mask", "attentions", "class", "classifier", "config", "def", "dtype", "elif", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "kwargs", "labels", "lang_ids", "logits", "long", "loss", "loss_fct", "multi_label_classification", "not", "num_labels", "or", "outputs", "position_ids", "problem_type", "r", "regression", "return", "return_dict", "roberta", "self", "sequence_output", "single_label_classification", "squeeze", "token_type_ids", "torch", "tuple", "view"], "xmod/modeling_xmod.py:XmodForMultipleChoice.__init__": ["Dropout", "Linear", "ModelForMultipleChoice", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "classifier", "config", "def", "dropout", "hidden_dropout_prob", "hidden_size", "nn", "post_init", "roberta", "self", "super"], "xmod/modeling_xmod.py:XmodForMultipleChoice.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "ModelForMultipleChoice", "ModelPreTrainedModel", "MultipleChoiceModelOutput", "None", "Tensor", "TransformersKwargs", "True", "Unpack", "attention_mask", "attentions", "class", "classifier", "def", "dropout", "else", "flat_attention_mask", "flat_input_ids", "flat_inputs_embeds", "flat_lang_ids", "flat_position_ids", "flat_token_type_ids", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "lang_ids", "logits", "loss", "loss_fct", "not", "num_choices", "outputs", "pooled_output", "position_ids", "r", "repeat", "reshaped_logits", "return", "return_dict", "roberta", "self", "shape", "size", "token_type_ids", "torch", "tuple", "view"], "xmod/modeling_xmod.py:XmodForTokenClassification.__init__": ["Dropout", "False", "Linear", "ModelForTokenClassification", "ModelModel", "ModelPreTrainedModel", "None", "__init__", "add_pooling_layer", "class", "classifier", "classifier_dropout", "config", "def", "dropout", "else", "hidden_dropout_prob", "hidden_size", "if", "is", "nn", "not", "num_labels", "post_init", "roberta", "self", "super"], "xmod/modeling_xmod.py:XmodForTokenClassification.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "ModelForTokenClassification", "ModelPreTrainedModel", "None", "Tensor", "TokenClassifierOutput", "TransformersKwargs", "True", "Unpack", "attention_mask", "attentions", "class", "classifier", "def", "dropout", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "lang_ids", "logits", "loss", "loss_fct", "not", "num_labels", "outputs", "position_ids", "r", "return", "return_dict", "roberta", "self", "sequence_output", "token_type_ids", "torch", "tuple", "view"], "xmod/modeling_xmod.py:XmodClassificationHead.__init__": ["Dropout", "Linear", "ModelClassificationHead", "Module", "None", "__init__", "class", "classifier_dropout", "config", "def", "dense", "dropout", "else", "hidden_dropout_prob", "hidden_size", "if", "is", "nn", "not", "num_labels", "out_proj", "self", "super"], "xmod/modeling_xmod.py:XmodClassificationHead.forward": ["ModelClassificationHead", "Module", "class", "def", "dense", "dropout", "features", "forward", "kwargs", "nn", "out_proj", "return", "self", "tanh", "torch", "x"], "xmod/modeling_xmod.py:XmodForQuestionAnswering.__init__": ["False", "Linear", "ModelForQuestionAnswering", "ModelModel", "ModelPreTrainedModel", "__init__", "add_pooling_layer", "class", "config", "def", "hidden_size", "nn", "num_labels", "post_init", "qa_outputs", "roberta", "self", "super"], "xmod/modeling_xmod.py:XmodForQuestionAnswering.forward": ["CrossEntropyLoss", "FloatTensor", "LongTensor", "ModelForQuestionAnswering", "ModelPreTrainedModel", "None", "QuestionAnsweringModelOutput", "Tensor", "TransformersKwargs", "True", "Unpack", "and", "attention_mask", "attentions", "clamp", "class", "contiguous", "def", "dim", "end_logits", "end_loss", "end_positions", "forward", "hidden_states", "if", "ignore_index", "ignored_index", "input_ids", "inputs_embeds", "is", "kwargs", "lang_ids", "len", "logits", "loss", "loss_fct", "not", "outputs", "position_ids", "qa_outputs", "r", "return", "return_dict", "roberta", "self", "sequence_output", "size", "split", "squeeze", "start_logits", "start_loss", "start_positions", "token_type_ids", "torch", "total_loss", "tuple"], "yolos/modeling_yolos.py:YolosEmbeddings.__init__": ["Dropout", "InterpolateInitialPositionEmbeddings", "ModelConfig", "ModelEmbeddings", "ModelPatchEmbeddings", "Module", "None", "Parameter", "__init__", "class", "cls_token", "config", "def", "detection_tokens", "dropout", "hidden_dropout_prob", "hidden_size", "interpolation", "nn", "num_detection_tokens", "num_patches", "patch_embeddings", "position_embeddings", "self", "super", "torch", "zeros"], "yolos/modeling_yolos.py:YolosEmbeddings.forward": ["ModelEmbeddings", "Module", "Tensor", "_", "batch_size", "cat", "class", "cls_token", "cls_tokens", "def", "detection_tokens", "dim", "dropout", "embeddings", "expand", "forward", "height", "interpolation", "nn", "num_channels", "patch_embeddings", "pixel_values", "position_embeddings", "return", "self", "seq_len", "shape", "size", "torch", "width"], "yolos/modeling_yolos.py:InterpolateInitialPositionEmbeddings.__init__": ["ModelInitialPositionEmbeddings", "Module", "None", "__init__", "class", "config", "def", "nn", "self", "super"], "yolos/modeling_yolos.py:InterpolateInitialPositionEmbeddings.forward": ["False", "Model", "ModelInitialPositionEmbeddings", "Module", "None", "Tensor", "align_corners", "batch_size", "bicubic", "cat", "class", "cls_pos_embed", "config", "def", "det_pos_embed", "dim", "flatten", "forward", "functional", "height", "hidden_size", "image_size", "img_size", "mode", "new_patch_height", "new_patch_width", "nn", "num_detection_tokens", "patch_height", "patch_pos_embed", "patch_size", "patch_width", "pos_embed", "return", "scale_pos_embed", "self", "seq_len", "shape", "size", "torch", "transpose", "view", "width"], "yolos/modeling_yolos.py:InterpolateMidPositionEmbeddings.__init__": ["ModelMidPositionEmbeddings", "Module", "None", "__init__", "class", "config", "def", "nn", "self", "super"], "yolos/modeling_yolos.py:InterpolateMidPositionEmbeddings.forward": ["False", "Model", "ModelMidPositionEmbeddings", "Module", "None", "Tensor", "align_corners", "batch_size", "bicubic", "cat", "class", "cls_pos_embed", "config", "contiguous", "def", "depth", "det_pos_embed", "dim", "flatten", "forward", "functional", "height", "hidden_size", "image_size", "img_size", "mode", "new_patch_height", "new_patch_width", "nn", "num_detection_tokens", "patch_height", "patch_pos_embed", "patch_size", "patch_width", "pos_embed", "return", "scale_pos_embed", "self", "seq_len", "shape", "size", "torch", "transpose", "view", "width"], "yolos/modeling_yolos.py:YolosPatchEmbeddings.__init__": ["Conv2d", "Iterable", "ModelPatchEmbeddings", "Module", "__init__", "abc", "class", "collections", "config", "def", "else", "hidden_size", "if", "image_size", "isinstance", "kernel_size", "nn", "num_channels", "num_patches", "patch_size", "projection", "self", "stride", "super"], "yolos/modeling_yolos.py:YolosPatchEmbeddings.forward": ["Make", "ModelPatchEmbeddings", "Module", "Tensor", "ValueError", "batch_size", "channel", "class", "configuration", "def", "dimension", "embeddings", "flatten", "forward", "height", "if", "in", "match", "nn", "num_channels", "of", "one", "pixel", "pixel_values", "projection", "raise", "return", "self", "set", "shape", "sure", "that", "the", "torch", "transpose", "values", "width", "with"], "yolos/modeling_yolos.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "TransformersKwargs", "Unpack", "attention_mask", "attn_output", "attn_weights", "contiguous", "def", "dim", "dropout", "float", "functional", "if", "is", "key", "kwargs", "matmul", "module", "nn", "not", "p", "query", "return", "scaling", "shape", "size", "softmax", "torch", "training", "transpose", "value"], "yolos/modeling_yolos.py:YolosSelfAttention.__init__": ["False", "Linear", "ModelConfig", "ModelSelfAttention", "Module", "The", "ValueError", "__init__", "a", "all_head_size", "and", "attention", "attention_head_size", "attention_probs_dropout_prob", "bias", "class", "config", "def", "dropout_prob", "embedding_size", "f", "hasattr", "heads", "hidden", "hidden_size", "if", "int", "is", "is_causal", "key", "multiple", "nn", "not", "num_attention_heads", "number", "of", "qkv_bias", "query", "raise", "scaling", "self", "size", "super", "the", "value"], "yolos/modeling_yolos.py:YolosSelfAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "ModelSelfAttention", "Module", "None", "Tensor", "_attn_implementation", "all_head_size", "attention_head_size", "attention_interface", "attention_probs", "batch_size", "class", "config", "context_layer", "def", "dropout", "dropout_prob", "eager", "eager_attention_forward", "else", "forward", "hidden_states", "if", "is_causal", "key", "key_layer", "new_context_layer_shape", "new_shape", "nn", "not", "num_attention_heads", "query", "query_layer", "reshape", "return", "scaling", "self", "shape", "size", "torch", "training", "transpose", "tuple", "value", "value_layer", "view"], "yolos/modeling_yolos.py:YolosSelfOutput.__init__": ["Dropout", "Linear", "ModelConfig", "ModelSelfOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "hidden_dropout_prob", "hidden_size", "nn", "self", "super"], "yolos/modeling_yolos.py:YolosSelfOutput.forward": ["ModelSelfOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "yolos/modeling_yolos.py:YolosAttention.__init__": ["ModelAttention", "ModelConfig", "ModelSelfAttention", "ModelSelfOutput", "Module", "__init__", "attention", "class", "config", "def", "nn", "output", "self", "super"], "yolos/modeling_yolos.py:YolosAttention.forward": ["ModelAttention", "Module", "Tensor", "_", "attention", "class", "def", "forward", "hidden_states", "nn", "output", "return", "self", "self_attn_output", "torch"], "yolos/modeling_yolos.py:YolosIntermediate.__init__": ["ACT2FN", "Linear", "ModelConfig", "ModelIntermediate", "Module", "__init__", "class", "config", "def", "dense", "else", "hidden_act", "hidden_size", "if", "intermediate_act_fn", "intermediate_size", "isinstance", "nn", "self", "str", "super"], "yolos/modeling_yolos.py:YolosIntermediate.forward": ["ModelIntermediate", "Module", "Tensor", "class", "def", "dense", "forward", "hidden_states", "intermediate_act_fn", "nn", "return", "self", "torch"], "yolos/modeling_yolos.py:YolosOutput.__init__": ["Dropout", "Linear", "ModelConfig", "ModelOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "hidden_dropout_prob", "hidden_size", "intermediate_size", "nn", "self", "super"], "yolos/modeling_yolos.py:YolosOutput.forward": ["ModelOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "yolos/modeling_yolos.py:YolosLayer.__init__": ["GradientCheckpointingLayer", "LayerNorm", "ModelAttention", "ModelConfig", "ModelIntermediate", "ModelLayer", "ModelOutput", "__init__", "attention", "chunk_size_feed_forward", "class", "config", "def", "eps", "hidden_size", "intermediate", "layer_norm_eps", "layernorm_after", "layernorm_before", "nn", "output", "self", "seq_len_dim", "super"], "yolos/modeling_yolos.py:YolosLayer.forward": ["GradientCheckpointingLayer", "ModelLayer", "Tensor", "attention", "attention_output", "class", "def", "forward", "hidden_states", "hidden_states_norm", "intermediate", "layer_output", "layernorm_after", "layernorm_before", "output", "return", "self", "torch"], "yolos/modeling_yolos.py:YolosEncoder.__init__": ["False", "InterpolateMidPositionEmbeddings", "ModelConfig", "ModelEncoder", "ModelLayer", "Module", "ModuleList", "None", "Parameter", "_", "__init__", "class", "config", "def", "else", "for", "gradient_checkpointing", "hidden_size", "if", "image_size", "in", "interpolation", "layer", "mid_position_embeddings", "nn", "num_detection_tokens", "num_hidden_layers", "patch_size", "range", "self", "seq_length", "super", "torch", "use_mid_position_embeddings", "zeros"], "yolos/modeling_yolos.py:YolosEncoder.forward": ["BaseModelOutput", "ModelEncoder", "Module", "Tensor", "class", "config", "def", "enumerate", "for", "forward", "height", "hidden_states", "i", "if", "in", "int", "interpolated_mid_position_embeddings", "interpolation", "last_hidden_state", "layer", "layer_module", "mid_position_embeddings", "nn", "num_hidden_layers", "return", "self", "torch", "use_mid_position_embeddings", "width"], "yolos/modeling_yolos.py:YolosModel.__init__": ["LayerNorm", "ModelConfig", "ModelEmbeddings", "ModelEncoder", "ModelModel", "ModelPooler", "ModelPreTrainedModel", "None", "True", "__init__", "add_pooling_layer", "bool", "class", "config", "def", "else", "embeddings", "encoder", "eps", "hidden_size", "if", "layer_norm_eps", "layernorm", "nn", "pooler", "post_init", "r", "self", "super"], "yolos/modeling_yolos.py:YolosModel.get_input_embeddings": ["ModelModel", "ModelPatchEmbeddings", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "patch_embeddings", "return", "self"], "yolos/modeling_yolos.py:YolosModel.forward": ["BaseModelOutput", "BaseModelOutputWithPooling", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "TransformersKwargs", "Unpack", "ValueError", "You", "class", "def", "else", "embedding_output", "embeddings", "encoder", "encoder_outputs", "forward", "have", "height", "if", "is", "kwargs", "last_hidden_state", "layernorm", "not", "pixel_values", "pooled_output", "pooler", "pooler_output", "raise", "return", "self", "sequence_output", "shape", "specify", "to", "torch", "width"], "yolos/modeling_yolos.py:YolosPooler.__init__": ["Linear", "ModelConfig", "ModelPooler", "Module", "Tanh", "__init__", "activation", "class", "config", "def", "dense", "hidden_size", "nn", "self", "super"], "yolos/modeling_yolos.py:YolosPooler.forward": ["ModelPooler", "Module", "Tensor", "activation", "class", "def", "dense", "first_token_tensor", "forward", "hidden_states", "nn", "pooled_output", "return", "self", "torch"], "yolos/modeling_yolos.py:YolosMLPPredictionHead.__init__": ["Linear", "ModelMLPPredictionHead", "Module", "ModuleList", "__init__", "class", "def", "for", "h", "hidden_dim", "in", "input_dim", "k", "layers", "n", "nn", "num_layers", "output_dim", "self", "super", "zip"], "yolos/modeling_yolos.py:YolosMLPPredictionHead.forward": ["ModelMLPPredictionHead", "Module", "class", "def", "else", "enumerate", "for", "forward", "functional", "i", "if", "in", "layer", "layers", "nn", "num_layers", "relu", "return", "self", "x"], "yolos/modeling_yolos.py:YolosForObjectDetection.__init__": ["False", "ModelConfig", "ModelForObjectDetection", "ModelMLPPredictionHead", "ModelModel", "ModelPreTrainedModel", "__init__", "add_pooling_layer", "bbox_predictor", "class", "class_labels_classifier", "config", "def", "hidden_dim", "hidden_size", "input_dim", "num_labels", "num_layers", "output_dim", "post_init", "self", "super", "vit"], "yolos/modeling_yolos.py:YolosForObjectDetection._set_aux_loss": ["ModelForObjectDetection", "ModelPreTrainedModel", "_set_aux_loss", "a", "b", "class", "def", "for", "in", "logits", "outputs_class", "outputs_coord", "pred_boxes", "return", "self", "zip"], "yolos/modeling_yolos.py:YolosForObjectDetection.forward": ["BaseModelOutputWithPooling", "FloatTensor", "ModelForObjectDetection", "ModelObjectDetectionOutput", "ModelPreTrainedModel", "None", "TransformersKwargs", "Unpack", "attentions", "auxiliary_loss", "auxiliary_outputs", "bbox_predictor", "class", "class_labels_classifier", "config", "def", "device", "dict", "forward", "hidden_states", "if", "intermediate", "is", "kwargs", "labels", "last_hidden_state", "list", "logits", "loss", "loss_dict", "loss_function", "not", "num_detection_tokens", "outputs", "outputs_class", "outputs_coord", "pixel_values", "pred_boxes", "r", "return", "self", "sequence_output", "sigmoid", "torch", "vit"], "yoso/modeling_yoso.py:load_cuda_kernels": ["ImportError", "Model", "Model_cuda_kernels", "community", "def", "get_kernel", "global", "if", "install", "installed", "is", "is_kernels_available", "it", "kernels", "lsh_cumulation", "not", "pip", "please", "raise", "with"], "yoso/modeling_yoso.py:to_contiguous": ["append", "contiguous", "def", "else", "for", "if", "in", "input_tensors", "is_contiguous", "isinstance", "list", "not", "out", "return", "tensor", "to_contiguous"], "yoso/modeling_yoso.py:normalize": ["Model", "append", "def", "dim", "else", "for", "functional", "if", "in", "input_tensors", "isinstance", "list", "nn", "out", "p", "return", "tensor"], "yoso/modeling_yoso.py:hashing": ["Key", "Model", "Query", "ValueError", "arange", "def", "device", "dim", "has", "hash_len", "if", "incorrect", "int", "key", "key_binary", "key_projection", "len", "matmul", "num_hash", "query", "query_binary", "query_hash", "query_projection", "raise", "raise_pow", "randn", "reshape", "return", "rmat", "size", "sum", "torch"], "yoso/modeling_yoso.py:YosoCumulation.forward": ["Function", "ModelCumulation", "None", "acos", "autograd", "class", "config", "ctx", "cumulation_value", "def", "expectation", "forward", "hash_code_len", "key", "key_mask", "math", "matmul", "pi", "query", "query_mask", "return", "save_for_backward", "torch", "transpose", "value"], "yoso/modeling_yoso.py:YosoCumulation.backward": ["Function", "ModelCumulation", "None", "autograd", "backward", "class", "config", "ctx", "def", "expectation", "grad", "grad_key", "grad_query", "grad_value", "hash_code_len", "key", "key_mask", "matmul", "query", "query_mask", "return", "saved_tensors", "to_contiguous", "torch", "transpose", "value", "weighted_exp"], "yoso/modeling_yoso.py:YosoLSHCumulation.forward": ["Function", "Key", "ModelLSHCumulation", "Query", "Value", "ValueError", "and", "autograd", "class", "config", "ctx", "cumulation_value", "def", "differ", "dimension", "else", "fast_hash", "forward", "hash_code_len", "hashing", "hashtable_capacity", "if", "in", "int", "is_cuda", "key", "key_hash_code", "key_mask", "lsh_cumulation", "mask", "num_hash", "query", "query_hash_code", "query_mask", "raise", "return", "save_for_backward", "size", "sizes", "to_contiguous", "torch", "use_cuda", "use_fast_hash", "value"], "yoso/modeling_yoso.py:YosoLSHCumulation.backward": ["Function", "ModelLSHCumulation", "None", "acos", "autograd", "backward", "class", "config", "ctx", "def", "else", "expectation", "grad", "grad_key", "grad_query", "grad_value", "hash_code_len", "hashtable_capacity", "if", "int", "is_cuda", "key", "key_hash_code", "key_mask", "lsh_backward", "lsh_cumulation", "lsh_weighted_cumulation", "math", "matmul", "pi", "query", "query_hash_code", "query_mask", "return", "saved_tensors", "to_contiguous", "torch", "transpose", "use_cuda", "value", "weighted_exp"], "yoso/modeling_yoso.py:YosoEmbeddings.__init__": ["Dropout", "Embedding", "False", "LayerNorm", "ModelEmbeddings", "Module", "__init__", "arange", "class", "config", "def", "device", "dropout", "dtype", "eps", "expand", "hidden_dropout_prob", "hidden_size", "layer_norm_eps", "long", "max_position_embeddings", "nn", "pad_token_id", "padding_idx", "persistent", "position_embeddings", "position_ids", "register_buffer", "self", "size", "super", "token_type_embeddings", "token_type_ids", "torch", "type_vocab_size", "vocab_size", "word_embeddings", "zeros"], "yoso/modeling_yoso.py:YosoEmbeddings.forward": ["LayerNorm", "ModelEmbeddings", "Module", "None", "buffered_token_type_ids", "buffered_token_type_ids_expanded", "class", "def", "device", "dropout", "dtype", "else", "embeddings", "expand", "forward", "hasattr", "if", "input_ids", "input_shape", "inputs_embeds", "is", "long", "nn", "not", "position_embeddings", "position_ids", "return", "self", "seq_length", "size", "token_type_embeddings", "token_type_ids", "torch", "word_embeddings", "zeros"], "yoso/modeling_yoso.py:YosoSelfAttention.__init__": ["Conv2d", "Could", "Dropout", "Exception", "False", "Linear", "ModelSelfAttention", "Module", "None", "The", "ValueError", "__init__", "a", "all_head_size", "and", "as", "attention", "attention_head_size", "attention_probs_dropout_prob", "bias", "class", "config", "conv", "conv_window", "custom", "def", "deformable", "dropout", "e", "embedding_size", "except", "f", "for", "groups", "hasattr", "hash_code_len", "heads", "hidden", "hidden_size", "if", "in_channels", "int", "is", "is_ninja_available", "is_torch_cuda_available", "kernel", "kernel_loaded", "kernel_size", "key", "load", "load_cuda_kernels", "logger", "lsh_backward", "lsh_config", "lsh_cumulation", "multi", "multiple", "nn", "not", "num_attention_heads", "num_hash", "number", "of", "out_channels", "padding", "query", "raise", "scale", "self", "size", "super", "the", "try", "use_conv", "use_expectation", "use_fast_hash", "value", "warning"], "yoso/modeling_yoso.py:YosoSelfAttention.forward": ["False", "ModelCumulation", "ModelLSHCumulation", "ModelSelfAttention", "Module", "None", "_", "all_head_size", "and", "apply", "attention_head_size", "attention_mask", "batch_size", "cat", "class", "context_layer", "contiguous", "conv", "conv_value_layer", "def", "device", "dim", "else", "forward", "gpu_warp_size", "head_dim", "hidden_states", "if", "int", "key", "key_layer", "lsh_config", "new_context_layer_shape", "nn", "normalize", "not", "num_attention_heads", "num_heads", "or", "output_attentions", "outputs", "pad_size", "permute", "query", "query_layer", "repeat_interleave", "reshape", "return", "self", "seq_len", "seq_length", "shape", "size", "torch", "training", "transpose", "unsqueeze", "use_conv", "use_expectation", "value", "value_layer", "view", "zeros"], "yoso/modeling_yoso.py:YosoSelfOutput.__init__": ["Dropout", "LayerNorm", "Linear", "ModelSelfOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "eps", "hidden_dropout_prob", "hidden_size", "layer_norm_eps", "nn", "self", "super"], "yoso/modeling_yoso.py:YosoSelfOutput.forward": ["LayerNorm", "ModelSelfOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "yoso/modeling_yoso.py:YosoAttention.__init__": ["ModelAttention", "ModelSelfAttention", "ModelSelfOutput", "Module", "__init__", "class", "config", "def", "nn", "output", "self", "super"], "yoso/modeling_yoso.py:YosoAttention.forward": ["False", "ModelAttention", "Module", "None", "attention_mask", "attention_output", "class", "def", "forward", "hidden_states", "nn", "output", "output_attentions", "outputs", "return", "self", "self_outputs"], "yoso/modeling_yoso.py:YosoIntermediate.__init__": ["ACT2FN", "Linear", "ModelIntermediate", "Module", "__init__", "class", "config", "def", "dense", "else", "hidden_act", "hidden_size", "if", "intermediate_act_fn", "intermediate_size", "isinstance", "nn", "self", "str", "super"], "yoso/modeling_yoso.py:YosoIntermediate.forward": ["ModelIntermediate", "Module", "Tensor", "class", "def", "dense", "forward", "hidden_states", "intermediate_act_fn", "nn", "return", "self", "torch"], "yoso/modeling_yoso.py:YosoOutput.__init__": ["Dropout", "LayerNorm", "Linear", "ModelOutput", "Module", "__init__", "class", "config", "def", "dense", "dropout", "eps", "hidden_dropout_prob", "hidden_size", "intermediate_size", "layer_norm_eps", "nn", "self", "super"], "yoso/modeling_yoso.py:YosoOutput.forward": ["LayerNorm", "ModelOutput", "Module", "Tensor", "class", "def", "dense", "dropout", "forward", "hidden_states", "input_tensor", "nn", "return", "self", "torch"], "yoso/modeling_yoso.py:YosoLayer.__init__": ["GradientCheckpointingLayer", "ModelAttention", "ModelIntermediate", "ModelLayer", "ModelOutput", "__init__", "add_cross_attention", "attention", "chunk_size_feed_forward", "class", "config", "def", "intermediate", "output", "self", "seq_len_dim", "super"], "yoso/modeling_yoso.py:YosoLayer.forward": ["False", "GradientCheckpointingLayer", "ModelLayer", "None", "apply_chunking_to_forward", "attention", "attention_mask", "attention_output", "chunk_size_feed_forward", "class", "def", "feed_forward_chunk", "forward", "hidden_states", "layer_output", "output_attentions", "outputs", "return", "self", "self_attention_outputs", "seq_len_dim"], "yoso/modeling_yoso.py:YosoLayer.feed_forward_chunk": ["GradientCheckpointingLayer", "ModelLayer", "attention_output", "class", "def", "feed_forward_chunk", "intermediate", "intermediate_output", "layer_output", "output", "return", "self"], "yoso/modeling_yoso.py:YosoEncoder.__init__": ["False", "ModelEncoder", "ModelLayer", "Module", "ModuleList", "_", "__init__", "class", "config", "def", "for", "gradient_checkpointing", "in", "layer", "nn", "num_hidden_layers", "range", "self", "super"], "yoso/modeling_yoso.py:YosoEncoder.forward": ["BaseModelOutputWithCrossAttentions", "False", "ModelEncoder", "Module", "None", "True", "all_hidden_states", "all_self_attentions", "attention_mask", "attentions", "class", "def", "else", "enumerate", "for", "forward", "hidden_states", "i", "if", "in", "is", "last_hidden_state", "layer", "layer_module", "layer_outputs", "nn", "not", "output_attentions", "output_hidden_states", "return", "return_dict", "self", "tuple", "v"], "yoso/modeling_yoso.py:YosoPredictionHeadTransform.__init__": ["ACT2FN", "LayerNorm", "Linear", "ModelPredictionHeadTransform", "Module", "__init__", "class", "config", "def", "dense", "else", "eps", "hidden_act", "hidden_size", "if", "isinstance", "layer_norm_eps", "nn", "self", "str", "super", "transform_act_fn"], "yoso/modeling_yoso.py:YosoPredictionHeadTransform.forward": ["LayerNorm", "ModelPredictionHeadTransform", "Module", "Tensor", "class", "def", "dense", "forward", "hidden_states", "nn", "return", "self", "torch", "transform_act_fn"], "yoso/modeling_yoso.py:YosoLMPredictionHead.__init__": ["Linear", "ModelLMPredictionHead", "ModelPredictionHeadTransform", "Module", "Parameter", "True", "__init__", "bias", "class", "config", "decoder", "def", "hidden_size", "nn", "self", "super", "torch", "transform", "vocab_size", "zeros"], "yoso/modeling_yoso.py:YosoLMPredictionHead.forward": ["ModelLMPredictionHead", "Module", "class", "decoder", "def", "forward", "hidden_states", "nn", "return", "self", "transform"], "yoso/modeling_yoso.py:YosoOnlyMLMHead.__init__": ["ModelLMPredictionHead", "ModelOnlyMLMHead", "Module", "__init__", "class", "config", "def", "nn", "predictions", "self", "super"], "yoso/modeling_yoso.py:YosoOnlyMLMHead.forward": ["ModelOnlyMLMHead", "Module", "Tensor", "class", "def", "forward", "nn", "prediction_scores", "predictions", "return", "self", "sequence_output", "torch"], "yoso/modeling_yoso.py:YosoPreTrainedModel._init_weights": ["ModelEmbeddings", "ModelLMPredictionHead", "ModelPreTrainedModel", "Module", "PreTrainedModel", "_init_weights", "arange", "bias", "class", "copy_", "def", "elif", "expand", "if", "init", "isinstance", "module", "nn", "position_ids", "self", "shape", "super", "token_type_ids", "torch", "zeros_"], "yoso/modeling_yoso.py:YosoModel.__init__": ["ModelEmbeddings", "ModelEncoder", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "config", "def", "embeddings", "encoder", "post_init", "self", "super"], "yoso/modeling_yoso.py:YosoModel.get_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "get_input_embeddings", "return", "self", "word_embeddings"], "yoso/modeling_yoso.py:YosoModel.set_input_embeddings": ["ModelModel", "ModelPreTrainedModel", "class", "def", "embeddings", "self", "set_input_embeddings", "value", "word_embeddings"], "yoso/modeling_yoso.py:YosoModel.forward": ["BaseModelOutputWithCrossAttentions", "ModelModel", "ModelPreTrainedModel", "None", "Tensor", "ValueError", "You", "and", "at", "attention_mask", "attentions", "batch_size", "bool", "both", "buffered_token_type_ids", "buffered_token_type_ids_expanded", "cannot", "class", "config", "cross_attentions", "def", "device", "dtype", "either", "elif", "else", "embedding_output", "embeddings", "encoder", "encoder_outputs", "expand", "forward", "hasattr", "have", "hidden_states", "if", "input_ids", "input_shape", "inputs_embeds", "is", "kwargs", "last_hidden_state", "long", "not", "ones", "or", "output_attentions", "output_hidden_states", "position_ids", "raise", "return", "return_dict", "same", "self", "seq_length", "sequence_output", "size", "specify", "the", "time", "to", "token_type_ids", "torch", "tuple", "use_return_dict", "warn_if_padding_and_no_attention_mask", "zeros"], "yoso/modeling_yoso.py:YosoForMaskedLM.__init__": ["Model", "ModelForMaskedLM", "ModelModel", "ModelOnlyMLMHead", "ModelPreTrainedModel", "__init__", "class", "cls", "config", "def", "post_init", "self", "super"], "yoso/modeling_yoso.py:YosoForMaskedLM.get_output_embeddings": ["ModelForMaskedLM", "ModelPreTrainedModel", "class", "cls", "decoder", "def", "get_output_embeddings", "predictions", "return", "self"], "yoso/modeling_yoso.py:YosoForMaskedLM.set_output_embeddings": ["ModelForMaskedLM", "ModelPreTrainedModel", "bias", "class", "cls", "decoder", "def", "new_embeddings", "predictions", "self", "set_output_embeddings"], "yoso/modeling_yoso.py:YosoForMaskedLM.forward": ["CrossEntropyLoss", "MaskedLMOutput", "Model", "ModelForMaskedLM", "ModelPreTrainedModel", "None", "Tensor", "attention_mask", "attentions", "bool", "class", "cls", "config", "def", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "masked_lm_loss", "not", "output", "output_attentions", "output_hidden_states", "outputs", "position_ids", "prediction_scores", "r", "return", "return_dict", "self", "sequence_output", "token_type_ids", "torch", "tuple", "use_return_dict", "view", "vocab_size"], "yoso/modeling_yoso.py:YosoClassificationHead.__init__": ["Dropout", "Linear", "ModelClassificationHead", "Module", "__init__", "class", "config", "def", "dense", "dropout", "hidden_dropout_prob", "hidden_size", "nn", "num_labels", "out_proj", "self", "super"], "yoso/modeling_yoso.py:YosoClassificationHead.forward": ["ACT2FN", "ModelClassificationHead", "Module", "class", "config", "def", "dense", "dropout", "features", "forward", "hidden_act", "kwargs", "nn", "out_proj", "return", "self", "x"], "yoso/modeling_yoso.py:YosoForSequenceClassification.__init__": ["Model", "ModelClassificationHead", "ModelForSequenceClassification", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "classifier", "config", "def", "num_labels", "post_init", "self", "super"], "yoso/modeling_yoso.py:YosoForSequenceClassification.forward": ["BCEWithLogitsLoss", "CrossEntropyLoss", "MSELoss", "Model", "ModelForSequenceClassification", "ModelPreTrainedModel", "None", "SequenceClassifierOutput", "Tensor", "and", "attention_mask", "attentions", "bool", "class", "classifier", "config", "def", "dtype", "elif", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "kwargs", "labels", "logits", "long", "loss", "loss_fct", "multi_label_classification", "not", "num_labels", "or", "output", "output_attentions", "output_hidden_states", "outputs", "position_ids", "problem_type", "r", "regression", "return", "return_dict", "self", "sequence_output", "single_label_classification", "squeeze", "token_type_ids", "torch", "tuple", "use_return_dict", "view"], "yoso/modeling_yoso.py:YosoForMultipleChoice.__init__": ["Linear", "Model", "ModelForMultipleChoice", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "classifier", "config", "def", "hidden_size", "nn", "post_init", "pre_classifier", "self", "super"], "yoso/modeling_yoso.py:YosoForMultipleChoice.forward": ["CrossEntropyLoss", "Model", "ModelForMultipleChoice", "ModelPreTrainedModel", "MultipleChoiceModelOutput", "None", "ReLU", "Tensor", "attention_mask", "attentions", "bool", "class", "classifier", "config", "def", "else", "forward", "hidden_state", "hidden_states", "if", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "nn", "not", "num_choices", "output", "output_attentions", "output_hidden_states", "outputs", "pooled_output", "position_ids", "pre_classifier", "r", "reshaped_logits", "return", "return_dict", "self", "shape", "size", "token_type_ids", "torch", "tuple", "use_return_dict", "view"], "yoso/modeling_yoso.py:YosoForTokenClassification.__init__": ["Dropout", "Linear", "Model", "ModelForTokenClassification", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "classifier", "config", "def", "dropout", "hidden_dropout_prob", "hidden_size", "nn", "num_labels", "post_init", "self", "super"], "yoso/modeling_yoso.py:YosoForTokenClassification.forward": ["CrossEntropyLoss", "Model", "ModelForTokenClassification", "ModelPreTrainedModel", "None", "Tensor", "TokenClassifierOutput", "active_labels", "active_logits", "active_loss", "attention_mask", "attentions", "bool", "class", "classifier", "config", "def", "dropout", "else", "forward", "hidden_states", "if", "ignore_index", "input_ids", "inputs_embeds", "is", "kwargs", "labels", "logits", "loss", "loss_fct", "not", "num_labels", "output", "output_attentions", "output_hidden_states", "outputs", "position_ids", "r", "return", "return_dict", "self", "sequence_output", "tensor", "token_type_ids", "torch", "tuple", "type_as", "use_return_dict", "view", "where"], "yoso/modeling_yoso.py:YosoForQuestionAnswering.__init__": ["Linear", "Model", "ModelForQuestionAnswering", "ModelModel", "ModelPreTrainedModel", "__init__", "class", "config", "def", "hidden_size", "nn", "num_labels", "post_init", "qa_outputs", "self", "super"], "yoso/modeling_yoso.py:YosoForQuestionAnswering.forward": ["CrossEntropyLoss", "Model", "ModelForQuestionAnswering", "ModelPreTrainedModel", "None", "QuestionAnsweringModelOutput", "Tensor", "and", "attention_mask", "attentions", "bool", "clamp", "class", "config", "def", "dim", "else", "end_logits", "end_loss", "end_positions", "forward", "hidden_states", "if", "ignore_index", "ignored_index", "input_ids", "inputs_embeds", "is", "kwargs", "len", "logits", "loss", "loss_fct", "not", "output", "output_attentions", "output_hidden_states", "outputs", "position_ids", "qa_outputs", "return", "return_dict", "self", "sequence_output", "size", "split", "squeeze", "start_logits", "start_loss", "start_positions", "token_type_ids", "torch", "total_loss", "tuple", "use_return_dict"], "zamba/modeling_zamba.py:ZambaRMSNorm.__init__": ["ModelRMSNorm", "Module", "Parameter", "__init__", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "zamba/modeling_zamba.py:ZambaRMSNorm.forward": ["ModelRMSNorm", "Module", "True", "class", "def", "dtype", "float32", "forward", "hidden_states", "input_dtype", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "zamba/modeling_zamba.py:ZambaRMSNorm.extra_repr": ["ModelRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "variance_epsilon", "weight"], "zamba/modeling_zamba.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "zamba/modeling_zamba.py:ZambaHybridDynamicCache.__init__": ["False", "ModelHybridDynamicCache", "None", "_", "__init__", "_buffers", "_modules", "_parameters", "append", "batch_size", "cache_shape", "class", "config", "conv_kernel_size", "conv_states", "def", "device", "dtype", "float16", "for", "has_previous_state", "hidden_size", "hybrid", "i", "if", "in", "intermediate_size", "is_compileable", "key_cache", "layers_block_type", "mamba_d_conv", "mamba_d_state", "mamba_expand", "n_mamba_heads", "num_hidden_layers", "range", "self", "ssm_state_size", "ssm_states", "tensor", "torch", "transformer_layers", "value_cache", "zeros"], "zamba/modeling_zamba.py:ZambaHybridDynamicCache.__len__": ["ModelHybridDynamicCache", "__len__", "class", "def", "key_cache", "len", "return", "self"], "zamba/modeling_zamba.py:ZambaHybridDynamicCache.update": ["Any", "ModelHybridDynamicCache", "None", "Tensor", "cache_kwargs", "cat", "class", "def", "dict", "dim", "else", "if", "int", "key_cache", "key_states", "layer_idx", "return", "self", "shape", "str", "torch", "tuple", "update", "value_cache", "value_states"], "zamba/modeling_zamba.py:ZambaHybridDynamicCache.reorder_cache": ["LongTensor", "ModelHybridDynamicCache", "beam_idx", "class", "conv_states", "def", "device", "for", "get_seq_length", "if", "in", "index_select", "key_cache", "layer_idx", "len", "range", "reorder_cache", "self", "ssm_states", "to", "torch", "value_cache"], "zamba/modeling_zamba.py:ZambaHybridDynamicCache.get_seq_length": ["ModelHybridDynamicCache", "None", "class", "def", "else", "get_seq_length", "if", "in", "int", "key_cache", "layer_idx", "len", "not", "or", "return", "self", "shape", "transformer_layers"], "zamba/modeling_zamba.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value", "value_states"], "zamba/modeling_zamba.py:ZambaAttention.__init__": ["False", "Linear", "ModelAttention", "ModelConfig", "Module", "True", "__init__", "attention_dropout", "attention_head_dim", "attention_hidden_size", "bias", "class", "config", "def", "head_dim", "hidden_size", "int", "is_causal", "k_proj", "layer_idx", "max_position_embeddings", "nn", "num_attention_heads", "num_key_value_groups", "num_key_value_heads", "o_proj", "q_proj", "scaling", "self", "super", "v_proj"], "zamba/modeling_zamba.py:ZambaAttention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "FlashAttentionKwargs", "ModelAttention", "ModelHybridDynamicCache", "Module", "None", "Tensor", "Unpack", "_attn_implementation", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "class", "config", "contiguous", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "head_dim", "hidden_shape", "hidden_states", "if", "input_shape", "int", "is", "k_proj", "key_states", "kwargs", "layer_idx", "nn", "not", "o_proj", "past_key_values", "q_proj", "query_states", "reshape", "return", "scaling", "self", "shape", "torch", "training", "transpose", "tuple", "update", "v_proj", "value_states", "view"], "zamba/modeling_zamba.py:ZambaMambaMixer.__init__": ["A", "ACT2FN", "AILab", "A_log", "Conv1d", "D", "Dao", "False", "If", "Linear", "ModelConfig", "ModelMambaMixer", "Module", "None", "Parameter", "The", "To", "__init__", "act", "activation", "arange", "available", "because", "bias", "causal", "causal_conv1d_fn", "causal_conv1d_update", "class", "com", "config", "contiguous", "conv1d", "conv_kernel_size", "def", "dt_proj_bias", "dt_proj_weight", "dtype", "expand", "fast", "float32", "follow", "github", "groups", "hidden_mamba_act", "hidden_size", "https", "if", "implementation", "in", "in_channels", "in_proj", "install", "intermediate_size", "is", "is_fast_path_available", "kernel_size", "layer_idx", "log", "logger", "mamba", "mamba_conv_bias", "mamba_d_conv", "mamba_d_state", "mamba_dt_rank", "mamba_expand", "mamba_head_dim", "mamba_inner_fn", "mamba_proj_bias", "model", "n_mamba_heads", "naive", "nn", "not", "of", "one", "ones", "out_channels", "out_proj", "padding", "path", "reshape", "selective_scan_fn", "selective_state_update", "self", "set", "spaces", "ssm_state_size", "state", "super", "the", "time_step_rank", "to", "torch", "use", "use_bias", "use_conv_bias", "use_fast_kernels", "use_mamba_kernels", "want", "warning_once", "x_proj_weight", "you", "zeros"], "zamba/modeling_zamba.py:ZambaMambaMixer.cuda_kernels_forward": ["A", "A_log", "B", "C", "D", "ModelHybridDynamicCache", "ModelMambaMixer", "Module", "None", "Tensor", "True", "_", "activation", "all", "and", "attention_mask", "batch_size", "bias", "cache_params", "cat", "causal_conv1d_fn", "causal_conv1d_update", "chunk", "class", "contextualized_states", "contiguous", "conv1d", "conv_kernel_size", "conv_states", "conv_weights", "copy_", "cuda_kernels_forward", "def", "delta_softplus", "device", "dim", "discrete_time_step", "dt_proj_bias", "dt_proj_weight", "dt_softplus", "dtype", "else", "empty", "exp", "float", "for", "functional", "gate", "has_previous_state", "hidden_states", "if", "in", "in_proj", "is", "layer_idx", "mamba_head_dim", "n", "n_mamba_heads", "nn", "not", "out_proj", "pad", "projected_states", "range", "reshape", "return", "return_last_state", "scan_outputs", "scan_outputs_", "selective_scan_fn", "selective_state_update", "self", "seq_len", "shape", "size", "split", "squeeze", "ssm_parameters", "ssm_state", "ssm_state_", "ssm_state_size", "ssm_states", "time_proj_bias", "time_step", "time_step_rank", "torch", "transpose", "unsqueeze", "use_precomputed_states", "view", "weight", "x_proj_weight"], "zamba/modeling_zamba.py:ZambaMambaMixer.slow_forward": ["A", "A_log", "B", "C", "D", "ModelHybridDynamicCache", "ModelMambaMixer", "Module", "None", "_", "act", "all", "and", "append", "attention_mask", "batch_size", "bias", "cache_params", "chunk", "class", "clone", "contextualized_states", "contiguous", "conv1d", "conv_kernel_size", "conv_state", "conv_states", "def", "deltaB_u", "device", "dim", "dims", "discrete_A", "discrete_B", "discrete_time_step", "dt_proj_bias", "dt_proj_weight", "dtype", "else", "exp", "float", "for", "functional", "gate", "has_previous_state", "hidden_states", "i", "if", "in", "in_proj", "input_states", "is", "isinstance", "layer_idx", "mamba_head_dim", "matmul", "n_mamba_heads", "nn", "not", "out_proj", "pad", "projected_states", "range", "reshape", "return", "roll", "scan_output", "scan_outputs", "self", "seq_len", "shape", "shifts", "slow_forward", "softplus", "split", "squeeze", "ssm_parameters", "ssm_state", "ssm_state_size", "ssm_states", "stack", "sum", "time_step", "time_step_rank", "to", "torch", "training", "transpose", "unsqueeze", "use_cache", "use_conv_bias", "view", "weight", "x_proj_weight", "zeros"], "zamba/modeling_zamba.py:ZambaMambaMixer.forward": ["CUDA", "False", "Fast", "Make", "Mamba", "ModelHybridDynamicCache", "ModelMambaMixer", "Module", "None", "ValueError", "a", "and", "are", "attention_mask", "available", "cache_params", "causal", "class", "config", "conv1d", "cuda", "cuda_kernels_forward", "def", "device", "forward", "hidden_states", "if", "in", "install", "installed", "is", "is_fast_path_available", "kernels", "lease", "mamba", "model", "module", "nn", "not", "on", "or", "pip", "raise", "return", "run", "s", "self", "set", "slow_forward", "ssm", "sure", "that", "the", "they", "to", "type", "use_fast_kernels", "use_mamba_kernels", "x_proj_weight"], "zamba/modeling_zamba.py:ZambaMLP.__init__": ["ACT2FN", "False", "Linear", "ModelMLP", "Module", "__init__", "act_fn", "bias", "class", "config", "def", "down_proj", "gate_proj", "hidden_act", "hidden_size", "intermediate_size", "nn", "self", "super", "up_proj"], "zamba/modeling_zamba.py:ZambaMLP.forward": ["ModelMLP", "Module", "act_fn", "class", "def", "down_proj", "forward", "gate_proj", "nn", "return", "self", "up_proj", "x"], "zamba/modeling_zamba.py:ZambaAttentionDecoderLayer.__init__": ["ModelAttention", "ModelAttentionDecoderLayer", "ModelConfig", "ModelMLP", "ModelRMSNorm", "Module", "None", "__init__", "attention_hidden_size", "class", "config", "def", "eps", "feed_forward", "hidden_size", "input_layernorm", "int", "layer_idx", "nn", "pre_ff_layernorm", "rms_norm_eps", "self", "self_attn", "super"], "zamba/modeling_zamba.py:ZambaAttentionDecoderLayer.forward": ["False", "FlashAttentionKwargs", "FloatTensor", "ModelAttentionDecoderLayer", "ModelHybridDynamicCache", "Module", "None", "Tensor", "Unpack", "attention_mask", "bool", "class", "concatenate", "def", "dim", "feed_forward", "forward", "hidden_states", "if", "input_layernorm", "int", "kwargs", "layer_idx", "nn", "original_hidden_states", "output_attentions", "outputs", "past_key_values", "pre_ff_layernorm", "return", "self", "self_attn", "self_attn_weights", "torch", "tuple", "use_cache"], "zamba/modeling_zamba.py:ZambaMambaDecoderLayer.__init__": ["GradientCheckpointingLayer", "ModelConfig", "ModelMambaDecoderLayer", "ModelMambaMixer", "ModelRMSNorm", "__init__", "class", "config", "def", "eps", "hidden_size", "input_layernorm", "int", "layer_idx", "mamba", "rms_norm_eps", "self", "super"], "zamba/modeling_zamba.py:ZambaMambaDecoderLayer.forward": ["False", "FloatTensor", "GradientCheckpointingLayer", "LongTensor", "ModelHybridDynamicCache", "ModelMambaDecoderLayer", "None", "Tensor", "attention_mask", "bool", "cache_params", "cache_position", "causal_mask", "class", "def", "else", "forward", "hidden_states", "if", "input_layernorm", "int", "is", "kwargs", "layer_idx", "mamba", "not", "original_hidden_states", "output_attentions", "outputs", "past_key_values", "position_ids", "residual", "return", "self", "self_attn_weights", "torch", "transformer_hidden_states", "tuple", "use_cache"], "zamba/modeling_zamba.py:ZambaHybridLayer.__init__": ["GradientCheckpointingLayer", "Linear", "ModelAttentionDecoderLayer", "ModelHybridLayer", "ModelMambaDecoderLayer", "__init__", "class", "def", "linear", "mamba", "mamba_decoder", "nn", "self", "shared_transf", "super"], "zamba/modeling_zamba.py:ZambaHybridLayer.forward": ["False", "FloatTensor", "GradientCheckpointingLayer", "LongTensor", "ModelHybridDynamicCache", "ModelHybridLayer", "None", "Tensor", "attention_mask", "bool", "cache_position", "causal_mask", "class", "def", "forward", "hidden_states", "if", "int", "layer_idx", "layer_outputs", "linear", "mamba_decoder", "original_hidden_states", "output_attentions", "past_key_values", "return", "self", "self_attn_weights", "shared_transf", "torch", "transformer_hidden_states", "tuple", "use_cache"], "zamba/modeling_zamba.py:ZambaPreTrainedModel._init_weights": ["A", "A_log", "D", "ModelMambaMixer", "ModelPreTrainedModel", "None", "PreTrainedModel", "_init_weights", "arange", "clamp", "class", "config", "contiguous", "copy_", "def", "dt", "dt_init_std", "dt_proj_bias", "dt_proj_weight", "dtype", "exp", "expand", "expm1", "float32", "hidden_size", "if", "init", "initializer_range", "intermediate_size", "inv_dt", "isinstance", "log", "mamba_dt_rank", "mamba_expand", "mamba_head_dim", "math", "mean", "min", "module", "n_mamba_heads", "normal_", "ones_", "rand", "reshape", "self", "ssm_state_size", "std", "super", "time_step_floor", "time_step_max", "time_step_min", "torch", "uniform_", "x_proj_weight"], "zamba/modeling_zamba.py:ZambaModel.__init__": ["Embedding", "False", "Linear", "ModelAttentionDecoderLayer", "ModelConfig", "ModelHybridLayer", "ModelMambaDecoderLayer", "ModelModel", "ModelPreTrainedModel", "ModelRMSNorm", "ModuleList", "None", "__init__", "_attn_implementation", "_tied_weights_keys", "append", "bias", "class", "config", "d", "def", "else", "embed_tokens", "enumerate", "eps", "f", "final_layernorm", "for", "gradient_checkpointing", "hidden_size", "hybrid", "if", "in", "is", "layer_id", "layer_idx", "layer_type", "layers", "layers_block_type", "linear", "mamba", "nn", "pad_token_id", "padding_idx", "post_init", "rf", "rms_norm_eps", "self", "shared_transf", "super", "vocab_size"], "zamba/modeling_zamba.py:ZambaModel.forward": ["BaseModelOutputWithPast", "False", "FloatTensor", "LongTensor", "Model", "ModelHybridDynamicCache", "ModelModel", "ModelPreTrainedModel", "None", "Setting", "Tensor", "True", "ValueError", "You", "_update_causal_mask", "a", "all_hidden_states", "all_self_attns", "an", "and", "arange", "at", "attention_mask", "attentions", "be", "bool", "both", "cache", "cache_position", "cannot", "causal_mask", "checkpointing", "class", "clone", "config", "def", "device", "either", "else", "embed_tokens", "enumerate", "final_layernorm", "for", "forward", "gradient", "gradient_checkpointing", "has_previous_state", "hidden_states", "if", "in", "incompatible", "initialized", "input_ids", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layer", "layer_idx", "layer_outputs", "layers", "logger", "must", "no", "not", "one", "original_hidden_states", "output", "output_attentions", "output_hidden_states", "past_key_values", "position_ids", "provided", "raise", "requires", "return", "return_dict", "returned", "same", "self", "shape", "so", "specify", "the", "time", "to", "to_tuple", "torch", "training", "tuple", "unsqueeze", "use_cache", "use_return_dict", "warning_once", "was", "will", "with"], "zamba/modeling_zamba.py:ZambaModel._update_causal_mask": ["AttentionMaskConverter", "ModelModel", "ModelPreTrainedModel", "None", "_attn_implementation", "_unmask_unattended", "_update_causal_mask", "and", "arange", "attention_mask", "bool", "cache_position", "causal_mask", "class", "clone", "config", "cuda", "def", "device", "diagonal", "dim", "dtype", "else", "eq", "expand", "fill_value", "finfo", "flash_attention_2", "full", "if", "in", "input_tensor", "is", "mask_length", "masked_fill", "min", "min_dtype", "not", "npu", "padding_mask", "reshape", "return", "sdpa", "self", "sequence_length", "shape", "target_length", "to", "torch", "triu", "type", "xpu"], "zamba/modeling_zamba.py:ZambaForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "ModelConfig", "ModelForCausalLM", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "vocab_size"], "zamba/modeling_zamba.py:ZambaForCausalLM.forward": ["CausalLMOutputWithPast", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForCausalLM", "ModelHybridDynamicCache", "ModelPreTrainedModel", "None", "Tensor", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "output", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "position_ids", "r", "return", "return_dict", "self", "slice", "slice_indices", "torch", "tuple", "use_cache", "use_return_dict", "vocab_size"], "zamba/modeling_zamba.py:ZambaForCausalLM.prepare_inputs_for_generation": ["False", "GenerationMixin", "ModelForCausalLM", "ModelHybridDynamicCache", "ModelPreTrainedModel", "None", "True", "and", "attention_mask", "cache_position", "class", "config", "contiguous", "cumsum", "def", "device", "dtype", "elif", "else", "empty_past_kv", "for", "if", "in", "input_ids", "inputs_embeds", "is", "is_first_iteration", "items", "key", "kwargs", "logits_to_keep", "long", "masked_fill_", "model_inputs", "not", "num_logits_to_keep", "or", "past_key_values", "position_ids", "prepare_inputs_for_generation", "return", "self", "shape", "update", "use_cache", "value"], "zamba/modeling_zamba.py:ZambaForSequenceClassification.__init__": ["False", "Linear", "ModelForSequenceClassification", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "model", "nn", "num_labels", "post_init", "score", "self", "super"], "zamba/modeling_zamba.py:ZambaForSequenceClassification.forward": ["BCEWithLogitsLoss", "Cache", "Cannot", "CrossEntropyLoss", "FloatTensor", "LongTensor", "MSELoss", "ModelForSequenceClassification", "ModelPreTrainedModel", "None", "Results", "SequenceClassifierOutputWithPast", "Tensor", "ValueError", "__class__", "__name__", "and", "arange", "argmax", "attention_mask", "attentions", "batch", "batch_size", "be", "bool", "class", "config", "conjunction", "def", "defined", "detect", "device", "dtype", "elif", "else", "f", "forward", "handle", "hidden_states", "if", "in", "input_ids", "inputs_embeds", "int", "int32", "is", "kwargs", "labels", "last_non_pad_token", "logger", "logits", "long", "loss", "loss_fct", "may", "model", "multi_label_classification", "no", "non_pad_mask", "not", "num_labels", "or", "output", "output_attentions", "output_hidden_states", "pad_token_id", "padding", "past_key_values", "pooled_logits", "position_ids", "problem_type", "r", "raise", "regression", "return", "return_dict", "score", "self", "shape", "single_label_classification", "sizes", "squeeze", "to", "token", "token_indices", "tokens", "torch", "transformer_outputs", "tuple", "unexpected", "use_cache", "use_return_dict", "using", "view", "warning_once", "will", "with"], "zamba2/modeling_zamba2.py:Zamba2RMSNormGated.__init__": ["ModelRMSNormGated", "Module", "Parameter", "__init__", "class", "def", "eps", "group_size", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "zamba2/modeling_zamba2.py:Zamba2RMSNormGated.forward": ["ModelRMSNormGated", "Module", "None", "True", "class", "def", "dtype", "float32", "forward", "functional", "gate", "group_count", "group_size", "hidden_states", "hidden_states_group", "if", "input_dtype", "is", "keepdim", "last_dim", "mean", "nn", "not", "pow", "prefix_dims", "return", "rsqrt", "self", "shape", "silu", "to", "torch", "variance", "variance_epsilon", "view", "weight"], "zamba2/modeling_zamba2.py:Zamba2RMSNorm.__init__": ["ModelRMSNorm", "Module", "Parameter", "__init__", "class", "def", "eps", "hidden_size", "nn", "ones", "self", "super", "torch", "variance_epsilon", "weight"], "zamba2/modeling_zamba2.py:Zamba2RMSNorm.forward": ["ModelRMSNorm", "Module", "True", "class", "def", "dtype", "float32", "forward", "hidden_states", "input_dtype", "keepdim", "mean", "nn", "pow", "return", "rsqrt", "self", "to", "torch", "variance", "variance_epsilon", "weight"], "zamba2/modeling_zamba2.py:Zamba2RMSNorm.extra_repr": ["ModelRMSNorm", "Module", "class", "def", "eps", "extra_repr", "f", "nn", "return", "self", "shape", "tuple", "variance_epsilon", "weight"], "zamba2/modeling_zamba2.py:Zamba2HybridDynamicCache.__init__": ["False", "ModelConfig", "ModelHybridDynamicCache", "None", "_", "__init__", "_buffers", "_modules", "_parameters", "append", "batch_size", "class", "config", "conv_kernel_size", "conv_states", "def", "device", "dtype", "float16", "for", "has_previous_state", "hidden_size", "hybrid", "i", "if", "in", "int", "intermediate_size", "key_cache", "layers_block_type", "mamba_d_conv", "mamba_d_state", "mamba_expand", "mamba_headdim", "mamba_ngroups", "n_mamba_heads", "num_hidden_layers", "range", "self", "ssm_state_size", "ssm_states", "str", "tensor", "torch", "transformer_layers", "value_cache", "zeros"], "zamba2/modeling_zamba2.py:Zamba2HybridDynamicCache.__len__": ["ModelHybridDynamicCache", "__len__", "class", "def", "key_cache", "len", "return", "self"], "zamba2/modeling_zamba2.py:Zamba2HybridDynamicCache.update": ["Any", "ModelHybridDynamicCache", "None", "Tensor", "cache_kwargs", "cat", "class", "def", "dict", "dim", "else", "if", "int", "key_cache", "key_states", "layer_idx", "return", "self", "shape", "str", "torch", "tuple", "update", "value_cache", "value_states"], "zamba2/modeling_zamba2.py:Zamba2HybridDynamicCache.reorder_cache": ["LongTensor", "ModelHybridDynamicCache", "beam_idx", "class", "conv_states", "def", "device", "for", "get_seq_length", "if", "in", "index_select", "key_cache", "layer_idx", "len", "range", "reorder_cache", "self", "ssm_states", "to", "torch", "value_cache"], "zamba2/modeling_zamba2.py:Zamba2HybridDynamicCache.get_seq_length": ["ModelHybridDynamicCache", "None", "class", "def", "else", "get_seq_length", "if", "in", "int", "key_cache", "layer_idx", "len", "not", "numel", "or", "return", "self", "shape", "transformer_layers"], "zamba2/modeling_zamba2.py:Zamba2HybridDynamicCache.update_conv_state": ["LongTensor", "ModelHybridDynamicCache", "Tensor", "cache_position", "clamp", "class", "conv_kernel_size", "conv_state", "conv_states", "def", "device", "dims", "int", "layer_idx", "new_conv_state", "return", "roll", "self", "shifts", "to", "torch", "update_conv_state", "zero_"], "zamba2/modeling_zamba2.py:Zamba2HybridDynamicCache.reset": ["ModelHybridDynamicCache", "class", "conv_states", "def", "reset", "self", "ssm_states", "zero_"], "zamba2/modeling_zamba2.py:Zamba2RotaryEmbedding.__init__": ["Callable", "False", "ModelConfig", "ModelRotaryEmbedding", "Module", "None", "ROPE_INIT_FUNCTIONS", "__init__", "attention_scaling", "class", "clone", "compute_default_rope_parameters", "config", "def", "default", "device", "if", "inv_freq", "max_position_embeddings", "max_seq_len_cached", "nn", "original_inv_freq", "original_max_seq_len", "persistent", "register_buffer", "rope_init_fn", "rope_parameters", "rope_type", "self", "super"], "zamba2/modeling_zamba2.py:Zamba2RotaryEmbedding.compute_default_rope_parameters": ["ModelConfig", "ModelRotaryEmbedding", "Module", "None", "Optional", "Tensor", "arange", "attention_factor", "base", "class", "compute_default_rope_parameters", "config", "def", "device", "dim", "dtype", "float", "getattr", "head_dim", "hidden_size", "int", "int64", "inv_freq", "nn", "num_attention_heads", "or", "return", "rope_parameters", "rope_theta", "seq_len", "to", "torch", "tuple"], "zamba2/modeling_zamba2.py:Zamba2RotaryEmbedding.forward": ["False", "ModelRotaryEmbedding", "Module", "None", "and", "attention_scaling", "cat", "class", "cos", "cpu", "def", "device", "device_type", "dim", "dtype", "else", "emb", "enabled", "expand", "float", "forward", "freqs", "if", "inv_freq", "inv_freq_expanded", "isinstance", "maybe_autocast", "mps", "nn", "position_ids", "position_ids_expanded", "return", "self", "shape", "sin", "str", "to", "torch", "transpose", "type", "with", "x"], "zamba2/modeling_zamba2.py:repeat_kv": ["Model_kv", "None", "Tensor", "batch", "def", "expand", "head_dim", "hidden_states", "if", "int", "n_rep", "num_key_value_heads", "reshape", "return", "shape", "slen", "torch"], "zamba2/modeling_zamba2.py:eager_attention_forward": ["Model_attention_forward", "Module", "None", "Tensor", "attention_mask", "attn_output", "attn_weights", "causal_mask", "contiguous", "def", "dim", "dropout", "dtype", "float", "float32", "functional", "if", "is", "key", "key_states", "kwargs", "matmul", "module", "nn", "not", "num_key_value_groups", "p", "query", "repeat_kv", "return", "scaling", "shape", "softmax", "to", "torch", "training", "transpose", "value", "value_states"], "zamba2/modeling_zamba2.py:rotate_half": ["Model_half", "cat", "def", "dim", "return", "shape", "torch", "x", "x1", "x2"], "zamba2/modeling_zamba2.py:apply_rotary_pos_emb": ["Model_rotary_pos_emb", "cos", "def", "k", "k_embed", "q", "q_embed", "return", "rotate_half", "sin", "unsqueeze", "unsqueeze_dim"], "zamba2/modeling_zamba2.py:Zamba2Attention.__init__": ["False", "Identity", "Linear", "ModelAttention", "ModelConfig", "Module", "ModuleList", "None", "Sequential", "True", "__init__", "adapter_rank", "append", "attention_dropout", "attention_head_dim", "attention_hidden_size", "bias", "block_id", "class", "config", "def", "else", "enumerate", "for", "head_dim", "hidden_size", "hybrid_layer_ids", "i", "if", "in", "index", "int", "is_causal", "k_proj", "layer_block_map", "layer_dic", "layer_idx", "linear_k_adapter", "linear_k_adapter_list", "linear_q_adapter", "linear_q_adapter_list", "linear_v_adapter", "linear_v_adapter_list", "max_position_embeddings", "nn", "num_attention_heads", "num_fwd_mem_blocks", "num_key_value_groups", "num_key_value_heads", "num_mem_blocks", "o_proj", "q_proj", "range", "scaling", "self", "super", "use_shared_attention_adapter", "v_proj", "value"], "zamba2/modeling_zamba2.py:Zamba2Attention.forward": ["ALL_ATTENTION_FUNCTIONS", "Callable", "FlashAttentionKwargs", "ModelAttention", "ModelHybridDynamicCache", "Module", "None", "Tensor", "Unpack", "_attn_implementation", "adapter_layer_idx", "apply_rotary_pos_emb", "attention_dropout", "attention_interface", "attention_mask", "attn_output", "attn_weights", "class", "config", "contiguous", "cos", "def", "dropout", "eager", "eager_attention_forward", "else", "forward", "head_dim", "hidden_shape", "hidden_states", "if", "input_shape", "int", "is", "k_proj", "key_states", "kwargs", "layer_dic", "layer_idx", "linear_k_adapter_list", "linear_q_adapter_list", "linear_v_adapter_list", "nn", "not", "o_proj", "past_key_values", "position_embeddings", "q_proj", "query_states", "reshape", "return", "scaling", "self", "shape", "sin", "torch", "training", "transpose", "tuple", "update", "use_mem_rope", "use_shared_attention_adapter", "v_proj", "value_states", "view"], "zamba2/modeling_zamba2.py:pad_tensor_by_size": ["Model", "Model_shape", "Model_size", "Model_tensor_by_size", "Tensor", "constant", "def", "else", "functional", "if", "input_tensor", "int", "len", "mode", "nn", "return", "shape", "torch", "value"], "zamba2/modeling_zamba2.py:reshape_into_chunks": ["Model", "Model_into_chunks", "chunk_size", "def", "else", "if", "input_tensor", "len", "pad_size", "pad_tensor_by_size", "return", "shape"], "zamba2/modeling_zamba2.py:segment_sum": ["Model_sum", "None", "bool", "chunk_size", "cumsum", "def", "device", "diagonal", "dim", "dtype", "expand", "inf", "input_tensor", "mask", "masked_fill", "ones", "return", "size", "tensor_segsum", "torch", "tril"], "zamba2/modeling_zamba2.py:Zamba2MambaMixer.__init__": ["A", "AILab", "A_log", "Conv1d", "D", "Dao", "Falling", "Linear", "ModelConfig", "ModelMambaMixer", "ModelRMSNormGated", "Module", "None", "Parameter", "SiLU", "The", "To", "True", "__init__", "act", "activation", "add_bias_linear", "arange", "available", "back", "because", "bias", "causal", "causal_conv1d_fn", "causal_conv1d_update", "chunk_size", "class", "com", "config", "conv1d", "conv_dim", "conv_kernel_size", "def", "dt_bias", "eps", "fast", "follow", "github", "group_size", "groups", "head_dim", "hidden_size", "https", "if", "implementation", "in_channels", "in_proj", "install", "int", "intermediate_size", "is", "is_fast_path_available", "kernel_size", "layer_idx", "log", "logger", "mamba", "mamba_d_conv", "mamba_d_state", "mamba_expand", "mamba_headdim", "mamba_ngroups", "n_groups", "n_mamba_heads", "naive", "nn", "norm", "not", "num_heads", "of", "one", "ones", "out_channels", "out_proj", "padding", "path", "projection_size", "selective_state_update", "self", "silu", "spaces", "ssm_state_size", "state", "super", "the", "time_step_limit", "time_step_max", "time_step_min", "to", "torch", "use_conv_bias", "use_mem_eff_path", "warning_once"], "zamba2/modeling_zamba2.py:Zamba2MambaMixer.cuda_kernels_forward": ["A", "A_log", "B", "C", "D", "False", "ModelHybridDynamicCache", "ModelMambaMixer", "Module", "None", "Tensor", "True", "_", "act", "activation", "all", "and", "attention_mask", "batch_size", "bias", "cache_params", "causal_conv1d_fn", "causal_conv1d_update", "chunk_size", "class", "conv1d", "conv_dim", "conv_kernel_size", "conv_state", "conv_states", "copy_", "cuda_kernels_forward", "d_mlp", "d_to_remove", "def", "dim", "dt", "dt_bias", "dt_limit", "dt_limit_kwargs", "dt_softplus", "dtype", "else", "exp", "expand", "float", "float32", "functional", "gate", "groups_time_state_size", "has_previous_state", "head_dim", "headdim", "hidden_states", "hidden_states_B_C", "hidden_states_B_C_t", "hidden_states_reshaped", "if", "in", "in_proj", "in_projected_states", "input_not_masked", "intermediate_size", "is", "layer_idx", "mamba_chunk_scan_combined", "mamba_split_conv1d_scan_combined", "n_groups", "ngroups", "nn", "norm", "norm_before_gate", "not", "num_heads", "or", "out", "out_proj", "outproj_bias", "outproj_weight", "pad", "projected_states", "return", "return_final_states", "rmsnorm_eps", "rmsnorm_weight", "scan_output", "selective_state_update", "self", "seq_idx", "seq_len", "shape", "silu", "split", "split_projection_dim", "squeeze", "ssm_state", "ssm_state_size", "ssm_states", "swish", "time_step", "time_step_limit", "to", "torch", "training", "transpose", "use_mem_eff_path", "variance_epsilon", "view", "weight", "x", "z"], "zamba2/modeling_zamba2.py:Zamba2MambaMixer.torch_forward": ["A", "A_cumsum", "A_log", "B", "B_decay_contraction", "C", "C_reshaped", "C_times_states", "D", "D_residual", "G", "G_intermediate", "L", "M", "M_intermediate", "ModelHybridDynamicCache", "ModelMambaMixer", "Module", "None", "Tensor", "Y_diag", "Y_off", "_", "act", "all", "and", "attention_mask", "batch_size", "bias", "bmm", "cache_params", "cat", "chunk_size", "clamp", "class", "clone", "contextualized_states", "contiguous", "conv1d", "conv_dim", "conv_kernel_size", "conv_state", "conv_states", "copy_", "cumsum", "dA", "dB", "dBx", "d_mlp", "decay_chunk", "decay_states", "def", "device", "dim", "dims", "dt", "dt_bias", "dtype", "else", "exp", "expand", "float", "float32", "for", "functional", "gate", "has_previous_state", "head_dim", "hidden_states", "if", "in", "in_proj", "input_states", "intermediate_size", "is", "layer_idx", "n_groups", "ndim", "new_states", "nn", "norm", "not", "num_heads", "out_proj", "output_size", "pad", "pad_size", "pad_tensor_by_size", "permute", "previous_states", "projected_states", "repeat_interleave", "reshape", "reshape_into_chunks", "result", "return", "roll", "scan_output", "segment_sum", "self", "seq_len", "shape", "shifts", "softplus", "split", "squeeze", "ssm_state", "ssm_state_size", "ssm_states", "ssm_states_reshaped", "state_decay_out", "state_decay_out_permuted", "states", "states_permuted", "sum", "t", "time_step_min", "to", "torch", "torch_forward", "transpose", "unsqueeze", "use_conv_bias", "view", "weight", "y", "zeros", "zeros_like"], "zamba2/modeling_zamba2.py:Zamba2MambaMixer.forward": ["ModelHybridDynamicCache", "ModelMambaMixer", "Module", "None", "Tensor", "and", "attention_mask", "cache_params", "class", "cuda", "cuda_kernels_forward", "def", "device", "forward", "hidden_states", "if", "in", "in_proj", "is_fast_path_available", "nn", "return", "self", "torch", "torch_forward", "type", "weight"], "zamba2/modeling_zamba2.py:Zamba2MLP.__init__": ["ACT2FN", "False", "Identity", "Linear", "ModelConfig", "ModelMLP", "Module", "ModuleList", "None", "Sequential", "__init__", "act_fn", "adapter_rank", "add_bias_linear", "append", "bias", "block_id", "class", "config", "def", "down_proj", "else", "enumerate", "for", "gate_up_proj", "gate_up_proj_adapter", "gate_up_proj_adapter_list", "hidden_act", "hidden_size", "hybrid_layer_ids", "i", "if", "in", "index", "int", "intermediate_size", "layer_block_map", "layer_dic", "nn", "num_fwd_mem_blocks", "num_mem_blocks", "range", "self", "super", "value"], "zamba2/modeling_zamba2.py:Zamba2MLP.forward": ["ModelMLP", "Module", "None", "act_fn", "chunk", "class", "def", "dim", "down_proj", "forward", "gate_up_proj", "gate_up_proj_adapter_list", "gate_up_state", "hidden_state", "layer_dic", "layer_idx", "nn", "output", "return", "self", "torch"], "zamba2/modeling_zamba2.py:Zamba2AttentionDecoderLayer.__init__": ["ModelAttention", "ModelAttentionDecoderLayer", "ModelConfig", "ModelMLP", "ModelRMSNorm", "Module", "None", "__init__", "attention_hidden_size", "block_id", "class", "config", "def", "eps", "feed_forward", "hidden_size", "hybrid_layer_ids", "input_layernorm", "int", "layer_idx", "len", "nn", "num_fwd_mem_blocks", "num_gs", "pre_ff_layernorm", "rms_norm_eps", "self", "self_attn", "super"], "zamba2/modeling_zamba2.py:Zamba2AttentionDecoderLayer.forward": ["False", "FlashAttentionKwargs", "FloatTensor", "LongTensor", "ModelAttentionDecoderLayer", "ModelHybridDynamicCache", "Module", "None", "Tensor", "Unpack", "attention_mask", "bool", "class", "concatenate", "def", "dim", "feed_forward", "forward", "hidden_states", "if", "input_layernorm", "int", "kwargs", "layer_idx", "nn", "original_hidden_states", "output_attentions", "outputs", "past_key_values", "position_embeddings", "pre_ff_layernorm", "return", "self", "self_attn", "self_attn_weights", "torch", "tuple"], "zamba2/modeling_zamba2.py:Zamba2MambaDecoderLayer.__init__": ["GradientCheckpointingLayer", "ModelConfig", "ModelMambaDecoderLayer", "ModelMambaMixer", "ModelRMSNorm", "__init__", "class", "config", "def", "eps", "hidden_size", "input_layernorm", "int", "layer_idx", "mamba", "rms_norm_eps", "self", "super"], "zamba2/modeling_zamba2.py:Zamba2MambaDecoderLayer.forward": ["False", "FloatTensor", "GradientCheckpointingLayer", "LongTensor", "ModelHybridDynamicCache", "ModelMambaDecoderLayer", "None", "Tensor", "attention_mask", "bool", "cache_params", "cache_position", "causal_mask", "class", "def", "else", "forward", "hidden_states", "if", "input_layernorm", "int", "is", "kwargs", "layer_idx", "mamba", "not", "original_hidden_states", "output_attentions", "outputs", "past_key_values", "position_ids", "residual", "return", "self", "self_attn_weights", "torch", "transformer_hidden_states", "tuple", "use_cache"], "zamba2/modeling_zamba2.py:Zamba2HybridLayer.__init__": ["GradientCheckpointingLayer", "Linear", "ModelAttentionDecoderLayer", "ModelHybridLayer", "ModelMambaDecoderLayer", "__init__", "class", "def", "linear", "mamba", "mamba_decoder", "nn", "self", "shared_transformer", "super"], "zamba2/modeling_zamba2.py:Zamba2HybridLayer.forward": ["False", "FloatTensor", "GradientCheckpointingLayer", "LongTensor", "ModelHybridDynamicCache", "ModelHybridLayer", "None", "Tensor", "attention_mask", "bool", "causal_mask", "class", "def", "forward", "hidden_states", "if", "int", "layer_idx", "layer_outputs", "linear", "mamba_decoder", "original_hidden_states", "output_attentions", "past_key_values", "position_embeddings", "position_ids", "return", "self", "self_attn_weights", "shared_transformer", "torch", "transformer_hidden_states", "tuple", "use_cache"], "zamba2/modeling_zamba2.py:Zamba2PreTrainedModel._init_weights": ["A", "A_log", "D", "ModelMambaMixer", "ModelPreTrainedModel", "PreTrainedModel", "_init_weights", "arange", "clamp", "class", "config", "copy_", "def", "dt", "dt_bias", "exp", "expm1", "if", "init", "inv_dt", "isinstance", "log", "math", "min", "module", "n_mamba_heads", "num_heads", "ones_", "rand", "self", "super", "time_step_floor", "time_step_max", "time_step_min", "torch"], "zamba2/modeling_zamba2.py:Zamba2Model.__init__": ["Embedding", "False", "Linear", "ModelAttentionDecoderLayer", "ModelConfig", "ModelMambaDecoderLayer", "ModelModel", "ModelPreTrainedModel", "ModelRMSNorm", "ModelRotaryEmbedding", "ModuleList", "True", "__init__", "_attn_implementation", "and", "append", "bias", "block_id", "blocks", "class", "config", "cycle", "def", "elif", "embed_tokens", "eps", "extended", "final_layernorm", "for", "get_layers", "gradient_checkpointing", "hidden_size", "hybrid", "i", "if", "in", "iter", "k", "layer_idx", "layers", "layers_block_type", "linear_layers", "logger", "mamba", "mamba_layers", "max_position_embeddings", "nn", "num_hidden_layers", "num_mem_blocks", "pad_token_id", "padding_idx", "post_init", "range", "rescaled", "rms_norm_eps", "rope_theta", "rotary_emb", "self", "set", "super", "to", "use_long_context", "use_mem_rope", "using", "vocab_size", "warning_once"], "zamba2/modeling_zamba2.py:Zamba2Model.forward": ["BaseModelOutputWithPast", "False", "FloatTensor", "LongTensor", "ModelHybridDynamicCache", "ModelModel", "ModelPreTrainedModel", "None", "Setting", "Tensor", "True", "ValueError", "You", "_update_causal_mask", "all_hidden_states", "all_self_attns", "and", "arange", "at", "attention_mask", "attentions", "batch_size", "bool", "both", "cache_position", "cannot", "causal_mask", "checkpointing", "class", "clone", "config", "def", "device", "dtype", "either", "else", "embed_tokens", "enumerate", "final_layernorm", "first_transformer_layer_id", "for", "forward", "get_seq_length", "gradient", "gradient_checkpointing", "has_previous_state", "hidden_states", "if", "in", "incompatible", "input_ids", "inputs_embeds", "is", "kwargs", "last_hidden_state", "layer", "layer_idx", "layer_outputs", "layers", "logger", "must", "not", "one", "original_hidden_states", "output", "output_attentions", "output_hidden_states", "past_key_values", "past_seen_tokens", "position_embeddings", "position_ids", "raise", "return", "return_dict", "rotary_emb", "same", "self", "shape", "specify", "the", "time", "to_tuple", "torch", "training", "tuple", "unsqueeze", "use_cache", "use_return_dict", "warning_once", "with"], "zamba2/modeling_zamba2.py:Zamba2Model._update_causal_mask": ["AttentionMaskConverter", "ModelModel", "ModelPreTrainedModel", "None", "_attn_implementation", "_unmask_unattended", "_update_causal_mask", "and", "arange", "attention_mask", "bool", "cache_position", "causal_mask", "class", "clone", "config", "cuda", "def", "device", "diagonal", "dim", "dtype", "else", "eq", "expand", "fill_value", "finfo", "flash_attention_2", "full", "if", "in", "input_tensor", "is", "mask_length", "masked_fill", "min", "min_dtype", "not", "npu", "padding_mask", "reshape", "return", "sdpa", "self", "sequence_length", "shape", "target_length", "to", "torch", "triu", "type", "xpu"], "zamba2/modeling_zamba2.py:Zamba2Model.get_layers": ["ModelHybridLayer", "ModelModel", "ModelPreTrainedModel", "_tied_weights_keys", "append", "block", "blocks", "class", "config", "def", "else", "enumerate", "f", "first_transformer_layer_id", "for", "get_layers", "hybrid", "hybrid_layer_ids", "if", "in", "layer_id", "layer_type", "layers", "layers_block_type", "len", "linear_layers", "mamba_layers", "next", "num_mem_blocks", "prefix_pattern", "return", "self", "shared_transformer", "update"], "zamba2/modeling_zamba2.py:Zamba2ForCausalLM.__init__": ["False", "GenerationMixin", "Linear", "ModelConfig", "ModelForCausalLM", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "lm_head", "model", "nn", "post_init", "self", "super", "vocab_size"], "zamba2/modeling_zamba2.py:Zamba2ForCausalLM.forward": ["CausalLMOutputWithPast", "FloatTensor", "GenerationMixin", "LongTensor", "ModelForCausalLM", "ModelHybridDynamicCache", "ModelPreTrainedModel", "None", "Tensor", "attention_mask", "attentions", "bool", "cache_position", "class", "config", "def", "else", "forward", "hidden_states", "if", "input_ids", "inputs_embeds", "int", "is", "isinstance", "kwargs", "labels", "lm_head", "logits", "logits_to_keep", "loss", "loss_function", "model", "not", "output", "output_attentions", "output_hidden_states", "outputs", "past_key_values", "position_ids", "r", "return", "return_dict", "self", "slice", "slice_indices", "torch", "tuple", "use_cache", "use_return_dict", "vocab_size"], "zamba2/modeling_zamba2.py:Zamba2ForCausalLM.prepare_inputs_for_generation": ["False", "GenerationMixin", "ModelForCausalLM", "ModelHybridDynamicCache", "ModelPreTrainedModel", "None", "True", "and", "attention_mask", "cache_position", "class", "config", "contiguous", "cumsum", "def", "device", "dtype", "elif", "else", "empty_past_kv", "for", "if", "in", "input_ids", "inputs_embeds", "is", "is_first_iteration", "items", "key", "kwargs", "logits_to_keep", "long", "masked_fill_", "model_inputs", "not", "num_logits_to_keep", "or", "past_key_values", "position_ids", "prepare_inputs_for_generation", "return", "self", "shape", "update", "use_cache", "value"], "zamba2/modeling_zamba2.py:Zamba2ForSequenceClassification.__init__": ["False", "Linear", "ModelForSequenceClassification", "ModelModel", "ModelPreTrainedModel", "__init__", "bias", "class", "config", "def", "hidden_size", "model", "nn", "num_labels", "post_init", "score", "self", "super"], "zamba2/modeling_zamba2.py:Zamba2ForSequenceClassification.forward": ["BCEWithLogitsLoss", "Cache", "Cannot", "CrossEntropyLoss", "FloatTensor", "LongTensor", "MSELoss", "ModelForSequenceClassification", "ModelPreTrainedModel", "None", "Results", "SequenceClassifierOutputWithPast", "Tensor", "ValueError", "__class__", "__name__", "and", "arange", "argmax", "attention_mask", "attentions", "batch", "batch_size", "be", "bool", "class", "config", "conjunction", "def", "defined", "detect", "device", "dtype", "elif", "else", "f", "forward", "handle", "hidden_states", "if", "in", "input_ids", "inputs_embeds", "int", "int32", "is", "kwargs", "labels", "last_non_pad_token", "logger", "logits", "long", "loss", "loss_fct", "may", "model", "multi_label_classification", "no", "non_pad_mask", "not", "num_labels", "or", "output", "output_attentions", "output_hidden_states", "pad_token_id", "padding", "past_key_values", "pooled_logits", "position_ids", "problem_type", "r", "raise", "regression", "return", "return_dict", "score", "self", "shape", "single_label_classification", "sizes", "squeeze", "to", "token", "token_indices", "tokens", "torch", "transformer_outputs", "tuple", "unexpected", "use_cache", "use_return_dict", "using", "view", "warning_once", "will", "with"], "zoedepth/modeling_zoedepth.py:ZoeDepthReassembleStage.__init__": ["ACT2FN", "Linear", "ModelReassembleLayer", "ModelReassembleStage", "Module", "ModuleList", "Sequential", "_", "__init__", "append", "backbone_hidden_size", "channels", "class", "config", "def", "factor", "for", "hidden_act", "hidden_size", "if", "in", "layers", "neck_hidden_size", "neck_hidden_sizes", "nn", "project", "readout_projects", "readout_type", "reassemble_factors", "self", "super", "zip"], "zoedepth/modeling_zoedepth.py:ZoeDepthReassembleStage.forward": ["ModelReassembleStage", "Module", "Tensor", "add", "append", "batch_size", "cat", "class", "cls_token", "contiguous", "def", "dim", "elif", "enumerate", "expand_as", "flatten", "for", "forward", "hidden_state", "hidden_states", "if", "in", "layers", "list", "nn", "num_channels", "out", "patch_height", "patch_width", "permute", "project", "readout", "readout_projects", "readout_type", "reshape", "return", "self", "sequence_length", "shape", "split", "stage_idx", "torch", "total_batch_size", "unsqueeze"], "zoedepth/modeling_zoedepth.py:ZoeDepthReassembleLayer.__init__": ["Conv2d", "ConvTranspose2d", "Identity", "ModelReassembleLayer", "Module", "__init__", "backbone_hidden_size", "channels", "class", "config", "def", "elif", "factor", "hidden_size", "if", "in_channels", "int", "kernel_size", "nn", "out_channels", "padding", "projection", "resize", "self", "stride", "super"], "zoedepth/modeling_zoedepth.py:ZoeDepthReassembleLayer.forward": ["ModelReassembleLayer", "Module", "class", "def", "forward", "hidden_state", "nn", "projection", "resize", "return", "self"], "zoedepth/modeling_zoedepth.py:ZoeDepthFeatureFusionStage.__init__": ["ModelConfig", "ModelFeatureFusionLayer", "ModelFeatureFusionStage", "Module", "ModuleList", "_", "__init__", "append", "class", "config", "def", "for", "in", "layers", "len", "neck_hidden_sizes", "nn", "range", "self", "super"], "zoedepth/modeling_zoedepth.py:ZoeDepthFeatureFusionStage.forward": ["ModelFeatureFusionStage", "Module", "None", "append", "class", "def", "else", "for", "forward", "fused_hidden_state", "fused_hidden_states", "hidden_state", "hidden_states", "if", "in", "is", "layer", "layers", "nn", "return", "self", "zip"], "zoedepth/modeling_zoedepth.py:ZoeDepthPreActResidualLayer.__init__": ["BatchNorm2d", "Conv2d", "ModelPreActResidualLayer", "Module", "None", "ReLU", "__init__", "activation1", "activation2", "batch_norm1", "batch_norm2", "batch_norm_eps", "bias", "class", "config", "convolution1", "convolution2", "def", "else", "eps", "fusion_hidden_size", "if", "is", "kernel_size", "nn", "not", "padding", "self", "stride", "super", "use_batch_norm", "use_batch_norm_in_fusion_residual", "use_bias_in_fusion_residual"], "zoedepth/modeling_zoedepth.py:ZoeDepthPreActResidualLayer.forward": ["ModelPreActResidualLayer", "Module", "Tensor", "activation1", "activation2", "batch_norm1", "batch_norm2", "class", "convolution1", "convolution2", "def", "forward", "hidden_state", "if", "nn", "residual", "return", "self", "torch", "use_batch_norm"], "zoedepth/modeling_zoedepth.py:ZoeDepthFeatureFusionLayer.__init__": ["Conv2d", "ModelConfig", "ModelFeatureFusionLayer", "ModelPreActResidualLayer", "Module", "True", "__init__", "align_corners", "bias", "bool", "class", "config", "def", "fusion_hidden_size", "kernel_size", "nn", "projection", "residual_layer1", "residual_layer2", "self", "super"], "zoedepth/modeling_zoedepth.py:ZoeDepthFeatureFusionLayer.forward": ["False", "ModelFeatureFusionLayer", "Module", "None", "Tensor", "align_corners", "bilinear", "class", "def", "forward", "functional", "hidden_state", "if", "interpolate", "is", "mode", "nn", "not", "projection", "residual", "residual_layer1", "residual_layer2", "return", "scale_factor", "self", "shape", "size", "torch"], "zoedepth/modeling_zoedepth.py:ZoeDepthNeck.__init__": ["Conv2d", "False", "ModelConfig", "ModelFeatureFusionStage", "ModelNeck", "ModelReassembleStage", "Module", "ModuleList", "None", "__init__", "and", "append", "backbone_config", "bias", "channel", "class", "config", "convs", "def", "else", "for", "fusion_hidden_size", "fusion_stage", "if", "in", "is", "kernel_size", "model_type", "neck_hidden_sizes", "nn", "not", "padding", "reassemble_stage", "self", "super", "swinv2"], "zoedepth/modeling_zoedepth.py:ZoeDepthNeck.forward": ["ModelNeck", "Module", "None", "Tensor", "The", "TypeError", "ValueError", "a", "be", "class", "config", "convs", "def", "enumerate", "equal", "feature", "features", "for", "forward", "fusion_stage", "hidden", "hidden_states", "i", "if", "in", "is", "isinstance", "len", "list", "neck", "neck_hidden_sizes", "nn", "not", "number", "of", "or", "output", "patch_height", "patch_width", "raise", "reassemble_stage", "return", "self", "should", "sizes", "states", "tensors", "the", "to", "torch", "tuple"], "zoedepth/modeling_zoedepth.py:ZoeDepthRelativeDepthEstimationHead.__init__": ["Conv2d", "ModelRelativeDepthEstimationHead", "Module", "None", "True", "Upsample", "__init__", "add_projection", "align_corners", "bilinear", "class", "config", "conv1", "conv2", "conv3", "def", "features", "fusion_hidden_size", "head_in_index", "if", "kernel_size", "mode", "nn", "num_relative_features", "padding", "projection", "scale_factor", "self", "stride", "super", "upsample"], "zoedepth/modeling_zoedepth.py:ZoeDepthRelativeDepthEstimationHead.forward": ["ModelRelativeDepthEstimationHead", "Module", "None", "ReLU", "Tensor", "class", "conv1", "conv2", "conv3", "def", "dim", "features", "forward", "head_in_index", "hidden_states", "if", "is", "list", "nn", "not", "predicted_depth", "projection", "return", "self", "squeeze", "torch", "upsample"], "zoedepth/modeling_zoedepth.py:log_binom": ["Model", "Model_binom", "def", "eps", "k", "n", "return", "torch"], "zoedepth/modeling_zoedepth.py:LogBinomialSoftmax.__init__": ["False", "ModelBinomialSoftmax", "Module", "__init__", "act", "arange", "class", "def", "k", "k_idx", "k_minus_1", "n_classes", "nn", "persistent", "register_buffer", "self", "softmax", "super", "tensor", "torch", "view"], "zoedepth/modeling_zoedepth.py:LogBinomialSoftmax.forward": ["Model", "ModelBinomialSoftmax", "Model_binom", "Module", "act", "clamp", "class", "def", "dim", "eps", "forward", "if", "k_idx", "k_minus_1", "ndim", "nn", "one_minus_probabilities", "probabilities", "return", "self", "temperature", "torch", "unsqueeze", "y"], "zoedepth/modeling_zoedepth.py:ZoeDepthConditionalLogBinomialSoftmax.__init__": ["Conv2d", "GELU", "LogBinomialSoftmax", "ModelConditionalLogBinomialSoftmax", "Module", "Sequential", "Softplus", "__init__", "act", "bottleneck", "bottleneck_factor", "class", "condition_dim", "config", "def", "in_features", "kernel_size", "log_binomial_transform", "max_temp", "min_temp", "mlp", "n_classes", "nn", "p_eps", "padding", "self", "softmax", "stride", "super", "torch"], "zoedepth/modeling_zoedepth.py:ZoeDepthConditionalLogBinomialSoftmax.forward": ["ModelConditionalLogBinomialSoftmax", "Module", "class", "concat", "condition_feature", "def", "dim", "forward", "log_binomial_transform", "main_feature", "max_temp", "min_temp", "mlp", "nn", "p_eps", "probabilities", "probabilities_and_temperature", "return", "self", "temperature", "torch", "unsqueeze"], "zoedepth/modeling_zoedepth.py:ZoeDepthSeedBinRegressor.__init__": ["Conv2d", "ModelSeedBinRegressor", "Module", "ReLU", "Softplus", "True", "__init__", "act1", "act2", "bin_centers_type", "bottleneck_features", "class", "config", "conv1", "conv2", "def", "else", "if", "in_features", "inplace", "max_depth", "min_depth", "mlp_dim", "n_bins", "nn", "normed", "self", "super"], "zoedepth/modeling_zoedepth.py:ZoeDepthSeedBinRegressor.forward": ["ModelSeedBinRegressor", "Module", "True", "act1", "act2", "bin_centers", "bin_centers_type", "bin_edges", "bin_widths", "bin_widths_normed", "class", "constant", "conv1", "conv2", "cumsum", "def", "dim", "else", "forward", "functional", "if", "keepdim", "max_depth", "min_depth", "mode", "nn", "normed", "pad", "return", "self", "sum", "torch", "value", "x"], "zoedepth/modeling_zoedepth.py:inv_attractor": ["Model_attractor", "alpha", "def", "div", "dx", "float", "gamma", "int", "pow", "return"], "zoedepth/modeling_zoedepth.py:ZoeDepthAttractorLayer.__init__": ["Conv2d", "False", "ModelAttractorLayer", "Module", "ReLU", "True", "__init__", "act1", "act2", "alpha", "attractor_alpha", "attractor_gamma", "attractor_kind", "bin_embedding_dim", "class", "config", "conv1", "conv2", "def", "gemma", "in_features", "inplace", "kind", "max_depth", "memory_efficient", "min_depth", "mlp_dim", "n_attractors", "n_bins", "nn", "self", "super"], "zoedepth/modeling_zoedepth.py:ZoeDepthAttractorLayer.forward": ["ModelAttractorLayer", "Module", "None", "True", "_", "act1", "act2", "align_corners", "attractors", "attractors_normed", "batch_size", "bilinear", "bin_centers", "bin_new_centers", "class", "clip", "conv1", "conv2", "def", "delta_c", "device", "dim", "else", "for", "forward", "func", "functional", "height", "i", "if", "in", "interpolate", "inv_attractor", "is", "kind", "max_depth", "mean", "memory_efficient", "min_depth", "mode", "n_attractors", "nn", "not", "prev_bin", "prev_bin_embedding", "range", "return", "self", "shape", "sort", "sum", "torch", "unsqueeze", "view", "width", "x", "zeros_like"], "zoedepth/modeling_zoedepth.py:ZoeDepthAttractorLayerUnnormed.__init__": ["Conv2d", "ModelAttractorLayerUnnormed", "Module", "ReLU", "Softplus", "True", "__init__", "act1", "act2", "alpha", "attractor_alpha", "attractor_kind", "bin_embedding_dim", "class", "config", "conv1", "conv2", "def", "gamma", "in_features", "inplace", "kind", "max_depth", "memory_efficient", "min_depth", "mlp_dim", "n_attractors", "n_bins", "nn", "self", "super"], "zoedepth/modeling_zoedepth.py:ZoeDepthAttractorLayerUnnormed.forward": ["ModelAttractorLayerUnnormed", "Module", "None", "True", "act1", "act2", "align_corners", "attractors", "bilinear", "bin_centers", "bin_new_centers", "class", "conv1", "conv2", "def", "delta_c", "device", "dim", "else", "for", "forward", "func", "functional", "height", "i", "if", "in", "interpolate", "inv_attractor", "is", "kind", "mean", "memory_efficient", "mode", "n_attractors", "nn", "not", "prev_bin", "prev_bin_embedding", "range", "return", "self", "shape", "sum", "torch", "unsqueeze", "width", "x", "zeros_like"], "zoedepth/modeling_zoedepth.py:ZoeDepthProjector.__init__": ["Conv2d", "ModelProjector", "Module", "ReLU", "True", "__init__", "act", "class", "conv1", "conv2", "def", "in_features", "inplace", "mlp_dim", "nn", "out_features", "self", "super"], "zoedepth/modeling_zoedepth.py:ZoeDepthProjector.forward": ["ModelProjector", "Module", "Tensor", "act", "class", "conv1", "conv2", "def", "forward", "hidden_state", "nn", "return", "self", "torch"], "zoedepth/modeling_zoedepth.py:ZoeDepthMultiheadAttention.__init__": ["Dropout", "Linear", "ModelMultiheadAttention", "Module", "The", "ValueError", "__init__", "a", "all_head_size", "attention", "attention_head_size", "class", "def", "dropout", "f", "heads", "hidden", "hidden_size", "if", "int", "is", "key", "multiple", "nn", "not", "num_attention_heads", "number", "of", "out_proj", "query", "raise", "self", "size", "super", "the", "value"], "zoedepth/modeling_zoedepth.py:ZoeDepthMultiheadAttention.forward": ["False", "FloatTensor", "ModelMultiheadAttention", "Module", "None", "Tensor", "_", "all_head_size", "attention_head_size", "attention_mask", "attention_probs", "attention_scores", "batch_size", "bool", "class", "context_layer", "contiguous", "def", "dim", "dropout", "else", "forward", "functional", "if", "is", "key", "key_layer", "keys", "math", "matmul", "new_context_layer_shape", "nn", "not", "num_attention_heads", "out_proj", "output_attentions", "outputs", "permute", "queries", "query", "query_layer", "return", "self", "seq_length", "shape", "size", "softmax", "sqrt", "torch", "transpose", "tuple", "value", "value_layer", "values", "view"], "zoedepth/modeling_zoedepth.py:ZoeDepthTransformerEncoderLayer.__init__": ["ACT2FN", "Dropout", "LayerNorm", "Linear", "ModelMultiheadAttention", "ModelTransformerEncoderLayer", "Module", "__init__", "activation", "class", "config", "def", "dropout", "dropout1", "dropout2", "hidden_size", "intermediate_size", "linear1", "linear2", "nn", "norm1", "norm2", "num_attention_heads", "patch_transformer_hidden_size", "patch_transformer_intermediate_size", "patch_transformer_num_attention_heads", "relu", "self", "self_attn", "super"], "zoedepth/modeling_zoedepth.py:ZoeDepthTransformerEncoderLayer.forward": ["ModelTransformerEncoderLayer", "Module", "None", "Tensor", "activation", "attention_mask", "class", "def", "dropout", "dropout1", "dropout2", "forward", "keys", "linear1", "linear2", "nn", "norm1", "norm2", "queries", "return", "self", "self_attn", "src", "src2", "src_mask", "torch", "values"], "zoedepth/modeling_zoedepth.py:ZoeDepthPatchTransformerEncoder.__init__": ["Conv2d", "ModelPatchTransformerEncoder", "ModelTransformerEncoderLayer", "Module", "ModuleList", "_", "__init__", "bottleneck_features", "class", "config", "def", "embedding_convPxP", "for", "in", "in_channels", "kernel_size", "nn", "num_patch_transformer_layers", "padding", "patch_transformer_hidden_size", "range", "self", "stride", "super", "transformer_encoder"], "zoedepth/modeling_zoedepth.py:ZoeDepthPatchTransformerEncoder.positional_encoding_1d": ["ModelPatchTransformerEncoder", "Module", "arange", "batch_size", "cat", "class", "cos", "cpu", "def", "device", "dim", "div_term", "dtype", "embedding_dim", "exp", "float32", "index", "log", "nn", "pos_encoding", "position", "positional_encoding_1d", "repeat", "return", "self", "sequence_length", "sin", "tensor", "torch", "unsqueeze"], "zoedepth/modeling_zoedepth.py:ZoeDepthPatchTransformerEncoder.forward": ["ModelPatchTransformerEncoder", "Module", "batch_size", "class", "def", "device", "dtype", "embedding_convPxP", "embedding_dim", "embeddings", "flatten", "for", "forward", "functional", "i", "in", "nn", "pad", "permute", "positional_encoding_1d", "range", "return", "self", "sequence_length", "shape", "transformer_encoder", "x"], "zoedepth/modeling_zoedepth.py:ZoeDepthMLPClassifier.__init__": ["Linear", "ModelMLPClassifier", "Module", "None", "ReLU", "__init__", "activation", "class", "def", "hidden_features", "in_features", "linear1", "linear2", "nn", "out_features", "self", "super"], "zoedepth/modeling_zoedepth.py:ZoeDepthMLPClassifier.forward": ["ModelMLPClassifier", "Module", "activation", "class", "def", "domain_logits", "forward", "hidden_state", "linear1", "linear2", "nn", "return", "self"], "zoedepth/modeling_zoedepth.py:ZoeDepthMultipleMetricDepthEstimationHeads.__init__": ["Attractor", "Conv2d", "ModelAttractorLayer", "ModelAttractorLayerUnnormed", "ModelConditionalLogBinomialSoftmax", "ModelMLPClassifier", "ModelMultipleMetricDepthEstimationHeads", "ModelPatchTransformerEncoder", "ModelProjector", "ModelSeedBinRegressor", "Module", "ModuleDict", "ModuleList", "_", "__init__", "attractors", "bin_centers_type", "bin_configurations", "bin_embedding_dim", "bottleneck_factor", "bottleneck_features", "class", "conditional_log_binomial", "conf", "config", "configuration", "conv2", "def", "elif", "for", "fusion_hidden_size", "i", "if", "in", "in_features", "kernel_size", "last_in", "len", "max_depth", "min_depth", "mlp_classifier", "mlp_dim", "n_attractors", "n_bins", "name", "nn", "normed", "num_attractors", "num_relative_features", "out_features", "padding", "patch_transformer", "projectors", "range", "seed_bin_regressors", "seed_projector", "self", "softplus", "stride", "super"], "zoedepth/modeling_zoedepth.py:ZoeDepthMultipleMetricDepthEstimationHeads.forward": ["IndexError", "ModelMultipleMetricDepthEstimationHeads", "Module", "True", "ValueError", "_", "align_corners", "argmax", "attractor", "attractors", "bilinear", "bin", "bin_centers", "bin_centers_type", "bin_configurations", "bin_configurations_name", "bin_configurationss", "bin_embedding", "bottleneck", "class", "conditional_log_binomial", "conf", "config", "configuration", "conv2", "def", "dim", "domain_logits", "domain_vote", "else", "embedding", "except", "f", "feature", "feature_blocks", "for", "forward", "found", "functional", "hybrid2", "if", "in", "interpolate", "item", "keepdim", "last", "max_depth", "min_depth", "mlp_classifier", "mode", "name", "names", "nn", "normed", "not", "out", "outconv_activation", "patch_transformer", "prev_bin", "prev_bin_embedding", "projector", "projectors", "raise", "relative_depth", "return", "seed_bin_centers", "seed_bin_regressor", "seed_bin_regressors", "seed_projector", "self", "shape", "softmax", "squeeze", "sum", "torch", "try", "x", "zip"], "zoedepth/modeling_zoedepth.py:ZoeDepthMetricDepthEstimationHead.__init__": ["Attractor", "Conv2d", "ModelAttractorLayer", "ModelAttractorLayerUnnormed", "ModelConditionalLogBinomialSoftmax", "ModelMetricDepthEstimationHead", "ModelProjector", "ModelSeedBinRegressor", "Module", "ModuleList", "_", "__init__", "attractors", "bin_centers_type", "bin_configuration", "bin_configurations", "bin_embedding_dim", "bottleneck_features", "class", "conditional_log_binomial", "config", "conv2", "def", "elif", "for", "fusion_hidden_size", "i", "if", "in", "in_features", "kernel_size", "last_in", "max_depth", "min_depth", "n_attractors", "n_bins", "n_classes", "nn", "normed", "num_attractors", "num_relative_features", "out_features", "padding", "projectors", "range", "seed_bin_regressor", "seed_projector", "self", "softplus", "stride", "super"], "zoedepth/modeling_zoedepth.py:ZoeDepthMetricDepthEstimationHead.forward": ["ModelMetricDepthEstimationHead", "Module", "None", "True", "_", "align_corners", "attractor", "attractors", "bilinear", "bin", "bin_centers", "bin_centers_type", "bin_embedding", "bottleneck", "cat", "class", "clone", "conditional_log_binomial", "conv2", "def", "dim", "else", "feature", "feature_blocks", "for", "forward", "functional", "hybrid2", "if", "in", "interpolate", "keepdim", "last", "max_depth", "min_depth", "mode", "nn", "normed", "out", "outconv_activation", "prev_bin", "prev_bin_embedding", "projector", "projectors", "relative_conditioning", "relative_depth", "return", "seed_bin_centers", "seed_bin_regressor", "seed_projector", "self", "shape", "size", "sum", "torch", "unsqueeze", "x", "zip"], "zoedepth/modeling_zoedepth.py:ZoeDepthPreTrainedModel._init_weights": ["LogBinomialSoftmax", "ModelPreTrainedModel", "PreTrainedModel", "_init_weights", "arange", "class", "copy_", "def", "if", "init", "isinstance", "k", "k_idx", "k_minus_1", "module", "self", "super", "tensor", "torch", "view"], "zoedepth/modeling_zoedepth.py:ZoeDepthForDepthEstimation.__init__": ["Model", "ModelForDepthEstimation", "ModelMetricDepthEstimationHead", "ModelMultipleMetricDepthEstimationHeads", "ModelNeck", "ModelPreTrainedModel", "ModelRelativeDepthEstimationHead", "ValueError", "__init__", "and", "assumes", "attributes", "backbone", "backbone_hidden_size", "bin_configurations", "class", "config", "def", "else", "hasattr", "have", "hidden_size", "if", "len", "load_backbone", "metric_head", "neck", "patch_size", "post_init", "raise", "relative_head", "s", "self", "super", "the", "to"], "zoedepth/modeling_zoedepth.py:ZoeDepthForDepthEstimation.forward": ["DepthEstimatorOutput", "FloatTensor", "LongTensor", "ModelDepthEstimatorOutput", "ModelForDepthEstimation", "ModelPreTrainedModel", "None", "NotImplementedError", "Tensor", "Training", "_", "attentions", "backbone", "bool", "bottleneck", "class", "config", "def", "dim", "domain_logits", "else", "feature_blocks", "feature_maps", "features", "forward", "forward_with_filtered_kwargs", "height", "hidden_states", "if", "implemented", "is", "kwargs", "labels", "loss", "metric_depth", "metric_head", "neck", "not", "out", "outconv_activation", "output", "output_attentions", "output_hidden_states", "outputs", "patch_height", "patch_size", "patch_width", "pixel_values", "predicted_depth", "r", "raise", "relative_depth", "relative_head", "return", "return_dict", "self", "shape", "squeeze", "torch", "tuple", "use_return_dict", "width", "yet"]}