| { | |
| "_name_or_path": "/checkpoint_load/dualtoken_qwen2dot5_3b_audiohead_separate_concat_sharetrue_1024d_weightsum_und_stage1/ckpt_11928_15_04041513", | |
| "architectures": [ | |
| "BaichuanForCausalLM" | |
| ], | |
| "attention_qkv_bias": true, | |
| "attention_qkv_pack": false, | |
| "audio_config": { | |
| "_name_or_path": "", | |
| "activation_dropout": 0.0, | |
| "activation_function": "gelu", | |
| "add_cross_attention": false, | |
| "apply_spec_augment": false, | |
| "architectures": null, | |
| "attention_dropout": 0.0, | |
| "audio_delim_token_id": 151693, | |
| "audio_end_token_id": 151677, | |
| "audio_pad_token_id": 151678, | |
| "audio_start_token_id": 151676, | |
| "avg_pooler": 2, | |
| "bad_words_ids": null, | |
| "begin_suppress_tokens": [ | |
| 220, | |
| 50256 | |
| ], | |
| "bos_token_id": 50256, | |
| "chunk_size_feed_forward": 0, | |
| "classifier_proj_size": 256, | |
| "cross_attention_hidden_size": null, | |
| "d_model": 1280, | |
| "decoder_attention_heads": 6, | |
| "decoder_ffn_dim": 1536, | |
| "decoder_layerdrop": 0.0, | |
| "decoder_layers": 4, | |
| "decoder_start_token_id": 50257, | |
| "diversity_penalty": 0.0, | |
| "do_sample": false, | |
| "dropout": 0.0, | |
| "early_stopping": false, | |
| "enable": false, | |
| "encoder_attention_heads": 20, | |
| "encoder_ffn_dim": 5120, | |
| "encoder_layerdrop": 0.0, | |
| "encoder_layers": 32, | |
| "encoder_no_repeat_ngram_size": 0, | |
| "eos_token_id": 50256, | |
| "exponential_decay_length_penalty": null, | |
| "finetuning_task": null, | |
| "forced_bos_token_id": null, | |
| "forced_eos_token_id": null, | |
| "hop_length": 160, | |
| "id2label": { | |
| "0": "LABEL_0", | |
| "1": "LABEL_1" | |
| }, | |
| "init_std": 0.02, | |
| "is_decoder": false, | |
| "is_encoder_decoder": true, | |
| "kernel_size": 3, | |
| "label2id": { | |
| "LABEL_0": 0, | |
| "LABEL_1": 1 | |
| }, | |
| "length_penalty": 1.0, | |
| "mask_feature_length": 10, | |
| "mask_feature_min_masks": 0, | |
| "mask_feature_prob": 0.0, | |
| "mask_time_length": 10, | |
| "mask_time_min_masks": 2, | |
| "mask_time_prob": 0.05, | |
| "max_audio_seconds": 30, | |
| "max_length": 20, | |
| "max_source_positions": 1500, | |
| "max_target_positions": 448, | |
| "median_filter_width": 7, | |
| "min_length": 0, | |
| "model_type": "whisper", | |
| "n_fft": 400, | |
| "no_repeat_ngram_size": 0, | |
| "num_beam_groups": 1, | |
| "num_beams": 1, | |
| "num_hidden_layers": 32, | |
| "num_mel_bins": 128, | |
| "num_return_sequences": 1, | |
| "output_attentions": false, | |
| "output_hidden_states": false, | |
| "output_scores": false, | |
| "pad_token_id": 50256, | |
| "prefix": null, | |
| "problem_type": null, | |
| "pruned_heads": {}, | |
| "remove_invalid_values": false, | |
| "repetition_penalty": 1.0, | |
| "return_dict": true, | |
| "return_dict_in_generate": false, | |
| "sampling_rate": 16000, | |
| "scale_embedding": false, | |
| "scaling_embedding": false, | |
| "sep_token_id": null, | |
| "split_overlap": 0.1, | |
| "stride_size": 2, | |
| "suppress_tokens": null, | |
| "task_specific_params": null, | |
| "temperature": 1.0, | |
| "tf_legacy_loss": false, | |
| "tie_encoder_decoder": false, | |
| "tie_word_embeddings": true, | |
| "tokenizer_class": null, | |
| "top_k": 50, | |
| "top_p": 1.0, | |
| "torch_dtype": null, | |
| "torchscript": false, | |
| "typical_p": 1.0, | |
| "use_bfloat16": false, | |
| "use_cache": true, | |
| "use_weighted_layer_sum": false, | |
| "vocab_size": 51865 | |
| }, | |
| "auto_map": { | |
| "AutoConfig": "configuration_baichuan.BaichuanConfig", | |
| "AutoModelForCausalLM": "modeling_baichuan.BaichuanForCausalLM" | |
| }, | |
| "baichuan_tokenizer_type": "auto", | |
| "bos_token_id": 1, | |
| "eos_token_id": 2, | |
| "head_dim": 128, | |
| "hidden_act": "silu", | |
| "hidden_size": 2048, | |
| "initializer_range": 0.02, | |
| "intermediate_size": 11008, | |
| "max_position_embeddings": 32768, | |
| "max_window_layers": 36, | |
| "model_type": "baichuan", | |
| "moe": false, | |
| "multimodal": [ | |
| "image" | |
| ], | |
| "multimodal_special_token_list": [ | |
| 151676, | |
| 151677, | |
| 151678, | |
| 151679, | |
| 151680, | |
| 151681, | |
| 151682, | |
| 151683, | |
| 151684, | |
| 151685, | |
| 151686, | |
| 151687, | |
| 151688, | |
| 151693, | |
| 151694, | |
| 151695, | |
| 151696, | |
| 151697, | |
| 151698 | |
| ], | |
| "multimodal_special_token_no_loss_list": [ | |
| 151676, | |
| 151677, | |
| 151678, | |
| 151679, | |
| 151680, | |
| 151681, | |
| 151682, | |
| 151683, | |
| 151684, | |
| 151685, | |
| 151686, | |
| 151687, | |
| 151688, | |
| 151693, | |
| 151694, | |
| 151695, | |
| 151696, | |
| 151697, | |
| 151698 | |
| ], | |
| "num_attention_heads": 16, | |
| "num_hidden_layers": 36, | |
| "num_key_value_heads": 2, | |
| "pad_token_id": 0, | |
| "position_embedding_type": "rope", | |
| "rms_norm_eps": 1e-06, | |
| "rope_theta": 1000000.0, | |
| "sliding_window": 32768, | |
| "sparse_attention_heads": null, | |
| "sparse_attention_layers": [], | |
| "tie_word_embeddings": false, | |
| "torch_dtype": "bfloat16", | |
| "train_multimodal_special_tokens_only": false, | |
| "transformers_version": "4.45.0", | |
| "use_cache": false, | |
| "use_norm_head": false, | |
| "use_sliding_window": false, | |
| "video_config": { | |
| "_name_or_path": "", | |
| "add_cross_attention": false, | |
| "architectures": null, | |
| "attention_dropout": 0.0, | |
| "bad_words_ids": null, | |
| "begin_suppress_tokens": null, | |
| "bos_token_id": null, | |
| "ch": 256, | |
| "ch_mult": [ | |
| 1, | |
| 2, | |
| 2, | |
| 4 | |
| ], | |
| "chunk_size_feed_forward": 0, | |
| "codebook_size": 32768, | |
| "crop_size": 384, | |
| "cross_attention_hidden_size": null, | |
| "decode_way": "1fps", | |
| "decoder_start_token_id": null, | |
| "diversity_penalty": 0.0, | |
| "do_sample": false, | |
| "double_z": false, | |
| "dropout": 0.0, | |
| "early_stopping": false, | |
| "embed_dim": 4, | |
| "enable": false, | |
| "encoder_no_repeat_ngram_size": 0, | |
| "eos_token_id": null, | |
| "exponential_decay_length_penalty": null, | |
| "feature_mode": "pad", | |
| "finetuning_task": null, | |
| "forced_bos_token_id": null, | |
| "forced_eos_token_id": null, | |
| "grid_size": [ | |
| [ | |
| 1, | |
| 1 | |
| ], | |
| [ | |
| 1, | |
| 2 | |
| ], | |
| [ | |
| 2, | |
| 1 | |
| ] | |
| ], | |
| "grid_size_max": 3, | |
| "hidden_act": "quick_gelu", | |
| "hidden_size": 2048, | |
| "id2label": { | |
| "0": "LABEL_0", | |
| "1": "LABEL_1" | |
| }, | |
| "image_delimiter_token_id": 151688, | |
| "image_end_token_id": 151680, | |
| "image_line_token_id": 151682, | |
| "image_mean": [ | |
| 0.5, | |
| 0.5, | |
| 0.5 | |
| ], | |
| "image_pad_token_id": 151681, | |
| "image_size": 384, | |
| "image_start_token_id": 151679, | |
| "image_std": [ | |
| 0.5, | |
| 0.5, | |
| 0.5 | |
| ], | |
| "in_channels": 3, | |
| "initializer_factor": 1.0, | |
| "initializer_range": 0.02, | |
| "intermediate_size": 4304, | |
| "is_decoder": false, | |
| "is_encoder_decoder": false, | |
| "label2id": { | |
| "LABEL_0": 0, | |
| "LABEL_1": 1 | |
| }, | |
| "layer_idx": 0, | |
| "layer_norm_eps": 1e-05, | |
| "length_penalty": 1.0, | |
| "max_frame_num": 48, | |
| "max_length": 20, | |
| "min_length": 0, | |
| "model_type": "clip_vision_model", | |
| "no_repeat_ngram_size": 0, | |
| "num_attention_heads": 16, | |
| "num_beam_groups": 1, | |
| "num_beams": 1, | |
| "num_channels": 3, | |
| "num_hidden_layers": 27, | |
| "num_res_blocks": 2, | |
| "num_return_sequences": 1, | |
| "out_channels": 3, | |
| "output_attentions": false, | |
| "output_hidden_states": false, | |
| "output_scores": false, | |
| "pad_token_id": null, | |
| "patch_size": 14, | |
| "pooling_kernel": 1, | |
| "pooling_type": "mean", | |
| "prefix": null, | |
| "problem_type": null, | |
| "projection_dim": 512, | |
| "pruned_heads": {}, | |
| "remove_invalid_values": false, | |
| "repetition_penalty": 1.0, | |
| "return_dict": true, | |
| "return_dict_in_generate": false, | |
| "sep_token_id": null, | |
| "suppress_tokens": null, | |
| "task_specific_params": null, | |
| "temperature": 1.0, | |
| "temporal_downsample_factor": 4, | |
| "tf_legacy_loss": false, | |
| "tie_encoder_decoder": false, | |
| "tie_word_embeddings": true, | |
| "tokenizer_class": null, | |
| "top_k": 50, | |
| "top_p": 1.0, | |
| "torch_dtype": "float32", | |
| "torchscript": false, | |
| "typical_p": 1.0, | |
| "use_bfloat16": false, | |
| "video_end_token_id": 151677, | |
| "video_place_token_id": 151675, | |
| "video_start_token_id": 151676, | |
| "z_channels": 4 | |
| }, | |
| "visual_config": { | |
| "_name_or_path": "", | |
| "add_cross_attention": false, | |
| "architectures": [ | |
| "RQVAESIGLIPTransformer" | |
| ], | |
| "bad_words_ids": null, | |
| "begin_suppress_tokens": null, | |
| "block_size": [ | |
| 16, | |
| 16, | |
| 4 | |
| ], | |
| "bos_token_id": null, | |
| "chunk_size_feed_forward": 0, | |
| "config_path": "/global_data/mllm/songwei/mmtrain/v4-und-gen-univt-audiohead/modeling/dualtoken-3b-audiohead-1024d-separate-projector-concat-sharetrue-ws-lnpost/config-rvq-4-4-sharetrue-n20", | |
| "crop_size": 256, | |
| "cross_attention_hidden_size": null, | |
| "decoder_start_token_id": null, | |
| "depth_head_embed_dim": 2048, | |
| "diversity_penalty": 0.0, | |
| "do_sample": false, | |
| "early_stopping": false, | |
| "enable": true, | |
| "encoder_no_repeat_ngram_size": 0, | |
| "eos_token_id": null, | |
| "exponential_decay_length_penalty": null, | |
| "feature_mode": "pad", | |
| "finetuning_task": null, | |
| "forced_bos_token_id": null, | |
| "forced_eos_token_id": null, | |
| "grid_size": 5, | |
| "grid_size_max": 26, | |
| "hidden_act": "quick_gelu", | |
| "hidden_size": 1024, | |
| "id2label": { | |
| "0": "LABEL_0", | |
| "1": "LABEL_1" | |
| }, | |
| "image_delimiter_token_id": 151688, | |
| "image_end_token_id": 151680, | |
| "image_line_token_id": 151682, | |
| "image_mean": [ | |
| 0.5, | |
| 0.5, | |
| 0.5 | |
| ], | |
| "image_pad_token_id": 151681, | |
| "image_size": 256, | |
| "image_start_token_id": 151679, | |
| "image_std": [ | |
| 0.5, | |
| 0.5, | |
| 0.5 | |
| ], | |
| "intermediate_size": 4096, | |
| "is_decoder": false, | |
| "is_encoder_decoder": false, | |
| "label2id": { | |
| "LABEL_0": 0, | |
| "LABEL_1": 1 | |
| }, | |
| "layer_idx": 0, | |
| "length_penalty": 1.0, | |
| "max_length": 20, | |
| "merge_size": 1, | |
| "min_length": 0, | |
| "model_type": "rqvaesigliptransformer_model", | |
| "no_repeat_ngram_size": 0, | |
| "num_attention_heads": 16, | |
| "num_beam_groups": 1, | |
| "num_beams": 1, | |
| "num_channels": 3, | |
| "num_hidden_layers": 24, | |
| "num_return_sequences": 1, | |
| "output_attentions": false, | |
| "output_hidden_states": false, | |
| "output_scores": false, | |
| "pad_token_id": null, | |
| "patch_size": 16, | |
| "pooling_kernel": 1, | |
| "pooling_type": "mean", | |
| "prefix": null, | |
| "problem_type": null, | |
| "pruned_heads": {}, | |
| "remove_invalid_values": false, | |
| "repetition_penalty": 1.0, | |
| "return_dict": true, | |
| "return_dict_in_generate": false, | |
| "rqtransformer_semantic": null, | |
| "rqtransformer_visual": null, | |
| "rqvaesiglip": null, | |
| "sep_token_id": null, | |
| "spatial_merge_size": 2, | |
| "spatial_patch_size": 14, | |
| "suppress_tokens": null, | |
| "task_specific_params": null, | |
| "temperature": 1.0, | |
| "temporal_downsample_factor": 4, | |
| "temporal_patch_size": 2, | |
| "tf_legacy_loss": false, | |
| "tie_encoder_decoder": false, | |
| "tie_word_embeddings": true, | |
| "tokenizer_class": null, | |
| "top_k": 50, | |
| "top_p": 1.0, | |
| "torch_dtype": "float32", | |
| "torchscript": false, | |
| "typical_p": 1.0, | |
| "use_bfloat16": false, | |
| "visual_codebook_loss_weights": [ | |
| 1.0, | |
| 1.0, | |
| 1.0, | |
| 1.0 | |
| ], | |
| "z_channels": 4 | |
| }, | |
| "vocab_size": 151936 | |
| } | |