| { |
| "architectures": [ |
| "DiffusionGemmaForBlockDiffusion" |
| ], |
| "boi_token_id": 255999, |
| "canvas_length": 256, |
| "dtype": "bfloat16", |
| "eoi_token_id": 258882, |
| "eos_token_id": [ |
| 1, |
| 106 |
| ], |
| "image_token_id": 258880, |
| "initializer_range": 0.02, |
| "model_type": "diffusion_gemma", |
| "text_config": { |
| "attention_bias": false, |
| "attention_dropout": 0.0, |
| "bos_token_id": 2, |
| "dtype": "bfloat16", |
| "eos_token_id": 1, |
| "final_logit_softcapping": 30.0, |
| "global_head_dim": 512, |
| "head_dim": 256, |
| "hidden_activation": "gelu_pytorch_tanh", |
| "hidden_size": 2816, |
| "initializer_range": 0.02, |
| "intermediate_size": 2112, |
| "layer_types": [ |
| "sliding_attention", |
| "sliding_attention", |
| "sliding_attention", |
| "sliding_attention", |
| "sliding_attention", |
| "full_attention", |
| "sliding_attention", |
| "sliding_attention", |
| "sliding_attention", |
| "sliding_attention", |
| "sliding_attention", |
| "full_attention", |
| "sliding_attention", |
| "sliding_attention", |
| "sliding_attention", |
| "sliding_attention", |
| "sliding_attention", |
| "full_attention", |
| "sliding_attention", |
| "sliding_attention", |
| "sliding_attention", |
| "sliding_attention", |
| "sliding_attention", |
| "full_attention", |
| "sliding_attention", |
| "sliding_attention", |
| "sliding_attention", |
| "sliding_attention", |
| "sliding_attention", |
| "full_attention" |
| ], |
| "max_position_embeddings": 262144, |
| "model_type": "diffusion_gemma_text", |
| "moe_intermediate_size": 704, |
| "num_attention_heads": 16, |
| "num_experts": 128, |
| "num_global_key_value_heads": 2, |
| "num_hidden_layers": 30, |
| "num_key_value_heads": 8, |
| "pad_token_id": 0, |
| "rms_norm_eps": 1e-06, |
| "rope_parameters": { |
| "full_attention": { |
| "partial_rotary_factor": 0.25, |
| "rope_theta": 1000000.0, |
| "rope_type": "proportional" |
| }, |
| "sliding_attention": { |
| "rope_theta": 10000.0, |
| "rope_type": "default" |
| } |
| }, |
| "sliding_window": 1024, |
| "tie_word_embeddings": true, |
| "top_k_experts": 8, |
| "use_bidirectional_attention": "vision", |
| "vocab_size": 262144 |
| }, |
| "tie_word_embeddings": true, |
| "transformers_version": "5.14.1", |
| "vision_config": { |
| "_name_or_path": "", |
| "architectures": null, |
| "attention_bias": false, |
| "attention_dropout": 0.0, |
| "chunk_size_feed_forward": 0, |
| "default_output_length": 280, |
| "dtype": "bfloat16", |
| "global_head_dim": 72, |
| "head_dim": 72, |
| "hidden_activation": "gelu_pytorch_tanh", |
| "hidden_size": 1152, |
| "id2label": { |
| "0": "LABEL_0", |
| "1": "LABEL_1" |
| }, |
| "initializer_range": 0.02, |
| "intermediate_size": 4304, |
| "is_encoder_decoder": false, |
| "label2id": { |
| "LABEL_0": 0, |
| "LABEL_1": 1 |
| }, |
| "max_position_embeddings": 131072, |
| "model_type": "gemma4_vision", |
| "num_attention_heads": 16, |
| "num_hidden_layers": 27, |
| "num_key_value_heads": 16, |
| "output_attentions": false, |
| "output_hidden_states": false, |
| "patch_size": 16, |
| "pooling_kernel_size": 3, |
| "position_embedding_size": 10240, |
| "problem_type": null, |
| "return_dict": true, |
| "rms_norm_eps": 1e-06, |
| "rope_parameters": { |
| "rope_theta": 100.0, |
| "rope_type": "default" |
| }, |
| "standardize": true, |
| "use_clipped_linears": false |
| }, |
| "vision_soft_tokens_per_image": 280, |
| "quantization_config": { |
| "config_groups": { |
| "group_0": { |
| "input_activations": { |
| "dynamic": false, |
| "num_bits": 4, |
| "type": "float", |
| "group_size": 16 |
| }, |
| "weights": { |
| "dynamic": false, |
| "num_bits": 4, |
| "type": "float", |
| "group_size": 16 |
| }, |
| "targets": [ |
| "Linear" |
| ] |
| } |
| }, |
| "ignore": [ |
| "lm_head", |
| "model.decoder.embed_tokens", |
| "model.decoder.layers.0.experts", |
| "model.decoder.layers.0.input_layernorm", |
| "model.decoder.layers.0.mlp*", |
| "model.decoder.layers.0.post_attention_layernorm", |
| "model.decoder.layers.0.post_feedforward_layernorm", |
| "model.decoder.layers.0.post_feedforward_layernorm_1", |
| "model.decoder.layers.0.post_feedforward_layernorm_2", |
| "model.decoder.layers.0.pre_feedforward_layernorm", |
| "model.decoder.layers.0.pre_feedforward_layernorm_2", |
| "model.decoder.layers.0.router*", |
| "model.decoder.layers.0.self_attn*", |
| "model.decoder.layers.1.experts", |
| "model.decoder.layers.1.input_layernorm", |
| "model.decoder.layers.1.mlp*", |
| "model.decoder.layers.1.post_attention_layernorm", |
| "model.decoder.layers.1.post_feedforward_layernorm", |
| "model.decoder.layers.1.post_feedforward_layernorm_1", |
| "model.decoder.layers.1.post_feedforward_layernorm_2", |
| "model.decoder.layers.1.pre_feedforward_layernorm", |
| "model.decoder.layers.1.pre_feedforward_layernorm_2", |
| "model.decoder.layers.1.router*", |
| "model.decoder.layers.1.self_attn*", |
| "model.decoder.layers.10.experts", |
| "model.decoder.layers.10.input_layernorm", |
| "model.decoder.layers.10.mlp*", |
| "model.decoder.layers.10.post_attention_layernorm", |
| "model.decoder.layers.10.post_feedforward_layernorm", |
| "model.decoder.layers.10.post_feedforward_layernorm_1", |
| "model.decoder.layers.10.post_feedforward_layernorm_2", |
| "model.decoder.layers.10.pre_feedforward_layernorm", |
| "model.decoder.layers.10.pre_feedforward_layernorm_2", |
| "model.decoder.layers.10.router*", |
| "model.decoder.layers.10.self_attn*", |
| "model.decoder.layers.11.experts", |
| "model.decoder.layers.11.input_layernorm", |
| "model.decoder.layers.11.mlp*", |
| "model.decoder.layers.11.post_attention_layernorm", |
| "model.decoder.layers.11.post_feedforward_layernorm", |
| "model.decoder.layers.11.post_feedforward_layernorm_1", |
| "model.decoder.layers.11.post_feedforward_layernorm_2", |
| "model.decoder.layers.11.pre_feedforward_layernorm", |
| "model.decoder.layers.11.pre_feedforward_layernorm_2", |
| "model.decoder.layers.11.router*", |
| "model.decoder.layers.11.self_attn*", |
| "model.decoder.layers.12.experts", |
| "model.decoder.layers.12.input_layernorm", |
| "model.decoder.layers.12.mlp*", |
| "model.decoder.layers.12.post_attention_layernorm", |
| "model.decoder.layers.12.post_feedforward_layernorm", |
| "model.decoder.layers.12.post_feedforward_layernorm_1", |
| "model.decoder.layers.12.post_feedforward_layernorm_2", |
| "model.decoder.layers.12.pre_feedforward_layernorm", |
| "model.decoder.layers.12.pre_feedforward_layernorm_2", |
| "model.decoder.layers.12.router*", |
| "model.decoder.layers.12.self_attn*", |
| "model.decoder.layers.13.experts", |
| "model.decoder.layers.13.input_layernorm", |
| "model.decoder.layers.13.mlp*", |
| "model.decoder.layers.13.post_attention_layernorm", |
| "model.decoder.layers.13.post_feedforward_layernorm", |
| "model.decoder.layers.13.post_feedforward_layernorm_1", |
| "model.decoder.layers.13.post_feedforward_layernorm_2", |
| "model.decoder.layers.13.pre_feedforward_layernorm", |
| "model.decoder.layers.13.pre_feedforward_layernorm_2", |
| "model.decoder.layers.13.router*", |
| "model.decoder.layers.13.self_attn*", |
| "model.decoder.layers.14.experts", |
| "model.decoder.layers.14.input_layernorm", |
| "model.decoder.layers.14.mlp*", |
| "model.decoder.layers.14.post_attention_layernorm", |
| "model.decoder.layers.14.post_feedforward_layernorm", |
| "model.decoder.layers.14.post_feedforward_layernorm_1", |
| "model.decoder.layers.14.post_feedforward_layernorm_2", |
| "model.decoder.layers.14.pre_feedforward_layernorm", |
| "model.decoder.layers.14.pre_feedforward_layernorm_2", |
| "model.decoder.layers.14.router*", |
| "model.decoder.layers.14.self_attn*", |
| "model.decoder.layers.15.experts", |
| "model.decoder.layers.15.input_layernorm", |
| "model.decoder.layers.15.mlp*", |
| "model.decoder.layers.15.post_attention_layernorm", |
| "model.decoder.layers.15.post_feedforward_layernorm", |
| "model.decoder.layers.15.post_feedforward_layernorm_1", |
| "model.decoder.layers.15.post_feedforward_layernorm_2", |
| "model.decoder.layers.15.pre_feedforward_layernorm", |
| "model.decoder.layers.15.pre_feedforward_layernorm_2", |
| "model.decoder.layers.15.router*", |
| "model.decoder.layers.15.self_attn*", |
| "model.decoder.layers.16.experts", |
| "model.decoder.layers.16.input_layernorm", |
| "model.decoder.layers.16.mlp*", |
| "model.decoder.layers.16.post_attention_layernorm", |
| "model.decoder.layers.16.post_feedforward_layernorm", |
| "model.decoder.layers.16.post_feedforward_layernorm_1", |
| "model.decoder.layers.16.post_feedforward_layernorm_2", |
| "model.decoder.layers.16.pre_feedforward_layernorm", |
| "model.decoder.layers.16.pre_feedforward_layernorm_2", |
| "model.decoder.layers.16.router*", |
| "model.decoder.layers.16.self_attn*", |
| "model.decoder.layers.17.experts", |
| "model.decoder.layers.17.input_layernorm", |
| "model.decoder.layers.17.mlp*", |
| "model.decoder.layers.17.post_attention_layernorm", |
| "model.decoder.layers.17.post_feedforward_layernorm", |
| "model.decoder.layers.17.post_feedforward_layernorm_1", |
| "model.decoder.layers.17.post_feedforward_layernorm_2", |
| "model.decoder.layers.17.pre_feedforward_layernorm", |
| "model.decoder.layers.17.pre_feedforward_layernorm_2", |
| "model.decoder.layers.17.router*", |
| "model.decoder.layers.17.self_attn*", |
| "model.decoder.layers.18.experts", |
| "model.decoder.layers.18.input_layernorm", |
| "model.decoder.layers.18.mlp*", |
| "model.decoder.layers.18.post_attention_layernorm", |
| "model.decoder.layers.18.post_feedforward_layernorm", |
| "model.decoder.layers.18.post_feedforward_layernorm_1", |
| "model.decoder.layers.18.post_feedforward_layernorm_2", |
| "model.decoder.layers.18.pre_feedforward_layernorm", |
| "model.decoder.layers.18.pre_feedforward_layernorm_2", |
| "model.decoder.layers.18.router*", |
| "model.decoder.layers.18.self_attn*", |
| "model.decoder.layers.19.experts", |
| "model.decoder.layers.19.input_layernorm", |
| "model.decoder.layers.19.mlp*", |
| "model.decoder.layers.19.post_attention_layernorm", |
| "model.decoder.layers.19.post_feedforward_layernorm", |
| "model.decoder.layers.19.post_feedforward_layernorm_1", |
| "model.decoder.layers.19.post_feedforward_layernorm_2", |
| "model.decoder.layers.19.pre_feedforward_layernorm", |
| "model.decoder.layers.19.pre_feedforward_layernorm_2", |
| "model.decoder.layers.19.router*", |
| "model.decoder.layers.19.self_attn*", |
| "model.decoder.layers.2.experts", |
| "model.decoder.layers.2.input_layernorm", |
| "model.decoder.layers.2.mlp*", |
| "model.decoder.layers.2.post_attention_layernorm", |
| "model.decoder.layers.2.post_feedforward_layernorm", |
| "model.decoder.layers.2.post_feedforward_layernorm_1", |
| "model.decoder.layers.2.post_feedforward_layernorm_2", |
| "model.decoder.layers.2.pre_feedforward_layernorm", |
| "model.decoder.layers.2.pre_feedforward_layernorm_2", |
| "model.decoder.layers.2.router*", |
| "model.decoder.layers.2.self_attn*", |
| "model.decoder.layers.20.experts", |
| "model.decoder.layers.20.input_layernorm", |
| "model.decoder.layers.20.mlp*", |
| "model.decoder.layers.20.post_attention_layernorm", |
| "model.decoder.layers.20.post_feedforward_layernorm", |
| "model.decoder.layers.20.post_feedforward_layernorm_1", |
| "model.decoder.layers.20.post_feedforward_layernorm_2", |
| "model.decoder.layers.20.pre_feedforward_layernorm", |
| "model.decoder.layers.20.pre_feedforward_layernorm_2", |
| "model.decoder.layers.20.router*", |
| "model.decoder.layers.20.self_attn*", |
| "model.decoder.layers.21.experts", |
| "model.decoder.layers.21.input_layernorm", |
| "model.decoder.layers.21.mlp*", |
| "model.decoder.layers.21.post_attention_layernorm", |
| "model.decoder.layers.21.post_feedforward_layernorm", |
| "model.decoder.layers.21.post_feedforward_layernorm_1", |
| "model.decoder.layers.21.post_feedforward_layernorm_2", |
| "model.decoder.layers.21.pre_feedforward_layernorm", |
| "model.decoder.layers.21.pre_feedforward_layernorm_2", |
| "model.decoder.layers.21.router*", |
| "model.decoder.layers.21.self_attn*", |
| "model.decoder.layers.22.experts", |
| "model.decoder.layers.22.input_layernorm", |
| "model.decoder.layers.22.mlp*", |
| "model.decoder.layers.22.post_attention_layernorm", |
| "model.decoder.layers.22.post_feedforward_layernorm", |
| "model.decoder.layers.22.post_feedforward_layernorm_1", |
| "model.decoder.layers.22.post_feedforward_layernorm_2", |
| "model.decoder.layers.22.pre_feedforward_layernorm", |
| "model.decoder.layers.22.pre_feedforward_layernorm_2", |
| "model.decoder.layers.22.router*", |
| "model.decoder.layers.22.self_attn*", |
| "model.decoder.layers.23.experts", |
| "model.decoder.layers.23.input_layernorm", |
| "model.decoder.layers.23.mlp*", |
| "model.decoder.layers.23.post_attention_layernorm", |
| "model.decoder.layers.23.post_feedforward_layernorm", |
| "model.decoder.layers.23.post_feedforward_layernorm_1", |
| "model.decoder.layers.23.post_feedforward_layernorm_2", |
| "model.decoder.layers.23.pre_feedforward_layernorm", |
| "model.decoder.layers.23.pre_feedforward_layernorm_2", |
| "model.decoder.layers.23.router*", |
| "model.decoder.layers.23.self_attn*", |
| "model.decoder.layers.24.experts", |
| "model.decoder.layers.24.input_layernorm", |
| "model.decoder.layers.24.mlp*", |
| "model.decoder.layers.24.post_attention_layernorm", |
| "model.decoder.layers.24.post_feedforward_layernorm", |
| "model.decoder.layers.24.post_feedforward_layernorm_1", |
| "model.decoder.layers.24.post_feedforward_layernorm_2", |
| "model.decoder.layers.24.pre_feedforward_layernorm", |
| "model.decoder.layers.24.pre_feedforward_layernorm_2", |
| "model.decoder.layers.24.router*", |
| "model.decoder.layers.24.self_attn*", |
| "model.decoder.layers.25.experts", |
| "model.decoder.layers.25.input_layernorm", |
| "model.decoder.layers.25.mlp*", |
| "model.decoder.layers.25.post_attention_layernorm", |
| "model.decoder.layers.25.post_feedforward_layernorm", |
| "model.decoder.layers.25.post_feedforward_layernorm_1", |
| "model.decoder.layers.25.post_feedforward_layernorm_2", |
| "model.decoder.layers.25.pre_feedforward_layernorm", |
| "model.decoder.layers.25.pre_feedforward_layernorm_2", |
| "model.decoder.layers.25.router*", |
| "model.decoder.layers.25.self_attn*", |
| "model.decoder.layers.26.experts", |
| "model.decoder.layers.26.input_layernorm", |
| "model.decoder.layers.26.mlp*", |
| "model.decoder.layers.26.post_attention_layernorm", |
| "model.decoder.layers.26.post_feedforward_layernorm", |
| "model.decoder.layers.26.post_feedforward_layernorm_1", |
| "model.decoder.layers.26.post_feedforward_layernorm_2", |
| "model.decoder.layers.26.pre_feedforward_layernorm", |
| "model.decoder.layers.26.pre_feedforward_layernorm_2", |
| "model.decoder.layers.26.router*", |
| "model.decoder.layers.26.self_attn*", |
| "model.decoder.layers.27.experts", |
| "model.decoder.layers.27.input_layernorm", |
| "model.decoder.layers.27.mlp*", |
| "model.decoder.layers.27.post_attention_layernorm", |
| "model.decoder.layers.27.post_feedforward_layernorm", |
| "model.decoder.layers.27.post_feedforward_layernorm_1", |
| "model.decoder.layers.27.post_feedforward_layernorm_2", |
| "model.decoder.layers.27.pre_feedforward_layernorm", |
| "model.decoder.layers.27.pre_feedforward_layernorm_2", |
| "model.decoder.layers.27.router*", |
| "model.decoder.layers.27.self_attn*", |
| "model.decoder.layers.28.experts", |
| "model.decoder.layers.28.input_layernorm", |
| "model.decoder.layers.28.mlp*", |
| "model.decoder.layers.28.post_attention_layernorm", |
| "model.decoder.layers.28.post_feedforward_layernorm", |
| "model.decoder.layers.28.post_feedforward_layernorm_1", |
| "model.decoder.layers.28.post_feedforward_layernorm_2", |
| "model.decoder.layers.28.pre_feedforward_layernorm", |
| "model.decoder.layers.28.pre_feedforward_layernorm_2", |
| "model.decoder.layers.28.router*", |
| "model.decoder.layers.28.self_attn*", |
| "model.decoder.layers.29.experts", |
| "model.decoder.layers.29.input_layernorm", |
| "model.decoder.layers.29.mlp*", |
| "model.decoder.layers.29.post_attention_layernorm", |
| "model.decoder.layers.29.post_feedforward_layernorm", |
| "model.decoder.layers.29.post_feedforward_layernorm_1", |
| "model.decoder.layers.29.post_feedforward_layernorm_2", |
| "model.decoder.layers.29.pre_feedforward_layernorm", |
| "model.decoder.layers.29.pre_feedforward_layernorm_2", |
| "model.decoder.layers.29.router*", |
| "model.decoder.layers.29.self_attn*", |
| "model.decoder.layers.3.experts", |
| "model.decoder.layers.3.input_layernorm", |
| "model.decoder.layers.3.mlp*", |
| "model.decoder.layers.3.post_attention_layernorm", |
| "model.decoder.layers.3.post_feedforward_layernorm", |
| "model.decoder.layers.3.post_feedforward_layernorm_1", |
| "model.decoder.layers.3.post_feedforward_layernorm_2", |
| "model.decoder.layers.3.pre_feedforward_layernorm", |
| "model.decoder.layers.3.pre_feedforward_layernorm_2", |
| "model.decoder.layers.3.router*", |
| "model.decoder.layers.3.self_attn*", |
| "model.decoder.layers.4.experts", |
| "model.decoder.layers.4.input_layernorm", |
| "model.decoder.layers.4.mlp*", |
| "model.decoder.layers.4.post_attention_layernorm", |
| "model.decoder.layers.4.post_feedforward_layernorm", |
| "model.decoder.layers.4.post_feedforward_layernorm_1", |
| "model.decoder.layers.4.post_feedforward_layernorm_2", |
| "model.decoder.layers.4.pre_feedforward_layernorm", |
| "model.decoder.layers.4.pre_feedforward_layernorm_2", |
| "model.decoder.layers.4.router*", |
| "model.decoder.layers.4.self_attn*", |
| "model.decoder.layers.5.experts", |
| "model.decoder.layers.5.input_layernorm", |
| "model.decoder.layers.5.mlp*", |
| "model.decoder.layers.5.post_attention_layernorm", |
| "model.decoder.layers.5.post_feedforward_layernorm", |
| "model.decoder.layers.5.post_feedforward_layernorm_1", |
| "model.decoder.layers.5.post_feedforward_layernorm_2", |
| "model.decoder.layers.5.pre_feedforward_layernorm", |
| "model.decoder.layers.5.pre_feedforward_layernorm_2", |
| "model.decoder.layers.5.router*", |
| "model.decoder.layers.5.self_attn*", |
| "model.decoder.layers.6.experts", |
| "model.decoder.layers.6.input_layernorm", |
| "model.decoder.layers.6.mlp*", |
| "model.decoder.layers.6.post_attention_layernorm", |
| "model.decoder.layers.6.post_feedforward_layernorm", |
| "model.decoder.layers.6.post_feedforward_layernorm_1", |
| "model.decoder.layers.6.post_feedforward_layernorm_2", |
| "model.decoder.layers.6.pre_feedforward_layernorm", |
| "model.decoder.layers.6.pre_feedforward_layernorm_2", |
| "model.decoder.layers.6.router*", |
| "model.decoder.layers.6.self_attn*", |
| "model.decoder.layers.7.experts", |
| "model.decoder.layers.7.input_layernorm", |
| "model.decoder.layers.7.mlp*", |
| "model.decoder.layers.7.post_attention_layernorm", |
| "model.decoder.layers.7.post_feedforward_layernorm", |
| "model.decoder.layers.7.post_feedforward_layernorm_1", |
| "model.decoder.layers.7.post_feedforward_layernorm_2", |
| "model.decoder.layers.7.pre_feedforward_layernorm", |
| "model.decoder.layers.7.pre_feedforward_layernorm_2", |
| "model.decoder.layers.7.router*", |
| "model.decoder.layers.7.self_attn*", |
| "model.decoder.layers.8.experts", |
| "model.decoder.layers.8.input_layernorm", |
| "model.decoder.layers.8.mlp*", |
| "model.decoder.layers.8.post_attention_layernorm", |
| "model.decoder.layers.8.post_feedforward_layernorm", |
| "model.decoder.layers.8.post_feedforward_layernorm_1", |
| "model.decoder.layers.8.post_feedforward_layernorm_2", |
| "model.decoder.layers.8.pre_feedforward_layernorm", |
| "model.decoder.layers.8.pre_feedforward_layernorm_2", |
| "model.decoder.layers.8.router*", |
| "model.decoder.layers.8.self_attn*", |
| "model.decoder.layers.9.experts", |
| "model.decoder.layers.9.input_layernorm", |
| "model.decoder.layers.9.mlp*", |
| "model.decoder.layers.9.post_attention_layernorm", |
| "model.decoder.layers.9.post_feedforward_layernorm", |
| "model.decoder.layers.9.post_feedforward_layernorm_1", |
| "model.decoder.layers.9.post_feedforward_layernorm_2", |
| "model.decoder.layers.9.pre_feedforward_layernorm", |
| "model.decoder.layers.9.pre_feedforward_layernorm_2", |
| "model.decoder.layers.9.router*", |
| "model.decoder.layers.9.self_attn*", |
| "model.decoder.norm", |
| "model.decoder.self_conditioning*", |
| "model.encoder.embed_vision*", |
| "model.encoder.language_model.layers.0.mlp*", |
| "model.encoder.language_model.layers.0.router*", |
| "model.encoder.language_model.layers.0.self_attn*", |
| "model.encoder.language_model.layers.1.mlp*", |
| "model.encoder.language_model.layers.1.router*", |
| "model.encoder.language_model.layers.1.self_attn*", |
| "model.encoder.language_model.layers.10.mlp*", |
| "model.encoder.language_model.layers.10.router*", |
| "model.encoder.language_model.layers.10.self_attn*", |
| "model.encoder.language_model.layers.11.mlp*", |
| "model.encoder.language_model.layers.11.router*", |
| "model.encoder.language_model.layers.11.self_attn*", |
| "model.encoder.language_model.layers.12.mlp*", |
| "model.encoder.language_model.layers.12.router*", |
| "model.encoder.language_model.layers.12.self_attn*", |
| "model.encoder.language_model.layers.13.mlp*", |
| "model.encoder.language_model.layers.13.router*", |
| "model.encoder.language_model.layers.13.self_attn*", |
| "model.encoder.language_model.layers.14.mlp*", |
| "model.encoder.language_model.layers.14.router*", |
| "model.encoder.language_model.layers.14.self_attn*", |
| "model.encoder.language_model.layers.15.mlp*", |
| "model.encoder.language_model.layers.15.router*", |
| "model.encoder.language_model.layers.15.self_attn*", |
| "model.encoder.language_model.layers.16.mlp*", |
| "model.encoder.language_model.layers.16.router*", |
| "model.encoder.language_model.layers.16.self_attn*", |
| "model.encoder.language_model.layers.17.mlp*", |
| "model.encoder.language_model.layers.17.router*", |
| "model.encoder.language_model.layers.17.self_attn*", |
| "model.encoder.language_model.layers.18.mlp*", |
| "model.encoder.language_model.layers.18.router*", |
| "model.encoder.language_model.layers.18.self_attn*", |
| "model.encoder.language_model.layers.19.mlp*", |
| "model.encoder.language_model.layers.19.router*", |
| "model.encoder.language_model.layers.19.self_attn*", |
| "model.encoder.language_model.layers.2.mlp*", |
| "model.encoder.language_model.layers.2.router*", |
| "model.encoder.language_model.layers.2.self_attn*", |
| "model.encoder.language_model.layers.20.mlp*", |
| "model.encoder.language_model.layers.20.router*", |
| "model.encoder.language_model.layers.20.self_attn*", |
| "model.encoder.language_model.layers.21.mlp*", |
| "model.encoder.language_model.layers.21.router*", |
| "model.encoder.language_model.layers.21.self_attn*", |
| "model.encoder.language_model.layers.22.mlp*", |
| "model.encoder.language_model.layers.22.router*", |
| "model.encoder.language_model.layers.22.self_attn*", |
| "model.encoder.language_model.layers.23.mlp*", |
| "model.encoder.language_model.layers.23.router*", |
| "model.encoder.language_model.layers.23.self_attn*", |
| "model.encoder.language_model.layers.24.mlp*", |
| "model.encoder.language_model.layers.24.router*", |
| "model.encoder.language_model.layers.24.self_attn*", |
| "model.encoder.language_model.layers.25.mlp*", |
| "model.encoder.language_model.layers.25.router*", |
| "model.encoder.language_model.layers.25.self_attn*", |
| "model.encoder.language_model.layers.26.mlp*", |
| "model.encoder.language_model.layers.26.router*", |
| "model.encoder.language_model.layers.26.self_attn*", |
| "model.encoder.language_model.layers.27.mlp*", |
| "model.encoder.language_model.layers.27.router*", |
| "model.encoder.language_model.layers.27.self_attn*", |
| "model.encoder.language_model.layers.28.mlp*", |
| "model.encoder.language_model.layers.28.router*", |
| "model.encoder.language_model.layers.28.self_attn*", |
| "model.encoder.language_model.layers.29.mlp*", |
| "model.encoder.language_model.layers.29.router*", |
| "model.encoder.language_model.layers.29.self_attn*", |
| "model.encoder.language_model.layers.3.mlp*", |
| "model.encoder.language_model.layers.3.router*", |
| "model.encoder.language_model.layers.3.self_attn*", |
| "model.encoder.language_model.layers.4.mlp*", |
| "model.encoder.language_model.layers.4.router*", |
| "model.encoder.language_model.layers.4.self_attn*", |
| "model.encoder.language_model.layers.5.mlp*", |
| "model.encoder.language_model.layers.5.router*", |
| "model.encoder.language_model.layers.5.self_attn*", |
| "model.encoder.language_model.layers.6.mlp*", |
| "model.encoder.language_model.layers.6.router*", |
| "model.encoder.language_model.layers.6.self_attn*", |
| "model.encoder.language_model.layers.7.mlp*", |
| "model.encoder.language_model.layers.7.router*", |
| "model.encoder.language_model.layers.7.self_attn*", |
| "model.encoder.language_model.layers.8.mlp*", |
| "model.encoder.language_model.layers.8.router*", |
| "model.encoder.language_model.layers.8.self_attn*", |
| "model.encoder.language_model.layers.9.mlp*", |
| "model.encoder.language_model.layers.9.router*", |
| "model.encoder.language_model.layers.9.self_attn*", |
| "model.encoder.vision_tower*" |
| ], |
| "quant_algo": "NVFP4", |
| "kv_cache_scheme": { |
| "dynamic": false, |
| "num_bits": 8, |
| "type": "float" |
| }, |
| "producer": { |
| "name": "modelopt", |
| "version": "0.0.1.dev1+gd0142c9dc.d20261002" |
| }, |
| "quant_method": "modelopt" |
| } |
| } |
