| { | |
| "architectures": [ | |
| "InternS1ForConditionalGeneration" | |
| ], | |
| "downsample_ratio": 0.5, | |
| "image_seq_length": 256, | |
| "image_token_id": 152957, | |
| "model_type": "interns1", | |
| "projector_hidden_act": "gelu", | |
| "quantization_config": { | |
| "activation_scheme": "dynamic", | |
| "fmt": "e4m3", | |
| "modules_to_not_convert": [ | |
| "lm_head", | |
| "model.language_model.embed_tokens", | |
| "model.multi_modal_projector.linear_1", | |
| "model.multi_modal_projector.linear_2", | |
| "model.vision_tower.embeddings.patch_embeddings.projection", | |
| "model.vision_tower.encoder.layer.0.attention.k_proj", | |
| "model.vision_tower.encoder.layer.0.attention.projection_layer", | |
| "model.vision_tower.encoder.layer.0.attention.q_proj", | |
| "model.vision_tower.encoder.layer.0.attention.v_proj", | |
| "model.vision_tower.encoder.layer.0.mlp.fc1", | |
| "model.vision_tower.encoder.layer.0.mlp.fc2", | |
| "model.vision_tower.encoder.layer.1.attention.k_proj", | |
| "model.vision_tower.encoder.layer.1.attention.projection_layer", | |
| "model.vision_tower.encoder.layer.1.attention.q_proj", | |
| "model.vision_tower.encoder.layer.1.attention.v_proj", | |
| "model.vision_tower.encoder.layer.1.mlp.fc1", | |
| "model.vision_tower.encoder.layer.1.mlp.fc2", | |
| "model.vision_tower.encoder.layer.10.attention.k_proj", | |
| "model.vision_tower.encoder.layer.10.attention.projection_layer", | |
| "model.vision_tower.encoder.layer.10.attention.q_proj", | |
| "model.vision_tower.encoder.layer.10.attention.v_proj", | |
| "model.vision_tower.encoder.layer.10.mlp.fc1", | |
| "model.vision_tower.encoder.layer.10.mlp.fc2", | |
| "model.vision_tower.encoder.layer.11.attention.k_proj", | |
| "model.vision_tower.encoder.layer.11.attention.projection_layer", | |
| "model.vision_tower.encoder.layer.11.attention.q_proj", | |
| "model.vision_tower.encoder.layer.11.attention.v_proj", | |
| "model.vision_tower.encoder.layer.11.mlp.fc1", | |
| "model.vision_tower.encoder.layer.11.mlp.fc2", | |
| "model.vision_tower.encoder.layer.12.attention.k_proj", | |
| "model.vision_tower.encoder.layer.12.attention.projection_layer", | |
| "model.vision_tower.encoder.layer.12.attention.q_proj", | |
| "model.vision_tower.encoder.layer.12.attention.v_proj", | |
| "model.vision_tower.encoder.layer.12.mlp.fc1", | |
| "model.vision_tower.encoder.layer.12.mlp.fc2", | |
| "model.vision_tower.encoder.layer.13.attention.k_proj", | |
| "model.vision_tower.encoder.layer.13.attention.projection_layer", | |
| "model.vision_tower.encoder.layer.13.attention.q_proj", | |
| "model.vision_tower.encoder.layer.13.attention.v_proj", | |
| "model.vision_tower.encoder.layer.13.mlp.fc1", | |
| "model.vision_tower.encoder.layer.13.mlp.fc2", | |
| "model.vision_tower.encoder.layer.14.attention.k_proj", | |
| "model.vision_tower.encoder.layer.14.attention.projection_layer", | |
| "model.vision_tower.encoder.layer.14.attention.q_proj", | |
| "model.vision_tower.encoder.layer.14.attention.v_proj", | |
| "model.vision_tower.encoder.layer.14.mlp.fc1", | |
| "model.vision_tower.encoder.layer.14.mlp.fc2", | |
| "model.vision_tower.encoder.layer.15.attention.k_proj", | |
| "model.vision_tower.encoder.layer.15.attention.projection_layer", | |
| "model.vision_tower.encoder.layer.15.attention.q_proj", | |
| "model.vision_tower.encoder.layer.15.attention.v_proj", | |
| "model.vision_tower.encoder.layer.15.mlp.fc1", | |
| "model.vision_tower.encoder.layer.15.mlp.fc2", | |
| "model.vision_tower.encoder.layer.16.attention.k_proj", | |
| "model.vision_tower.encoder.layer.16.attention.projection_layer", | |
| "model.vision_tower.encoder.layer.16.attention.q_proj", | |
| "model.vision_tower.encoder.layer.16.attention.v_proj", | |
| "model.vision_tower.encoder.layer.16.mlp.fc1", | |
| "model.vision_tower.encoder.layer.16.mlp.fc2", | |
| "model.vision_tower.encoder.layer.17.attention.k_proj", | |
| "model.vision_tower.encoder.layer.17.attention.projection_layer", | |
| "model.vision_tower.encoder.layer.17.attention.q_proj", | |
| "model.vision_tower.encoder.layer.17.attention.v_proj", | |
| "model.vision_tower.encoder.layer.17.mlp.fc1", | |
| "model.vision_tower.encoder.layer.17.mlp.fc2", | |
| "model.vision_tower.encoder.layer.18.attention.k_proj", | |
| "model.vision_tower.encoder.layer.18.attention.projection_layer", | |
| "model.vision_tower.encoder.layer.18.attention.q_proj", | |
| "model.vision_tower.encoder.layer.18.attention.v_proj", | |
| "model.vision_tower.encoder.layer.18.mlp.fc1", | |
| "model.vision_tower.encoder.layer.18.mlp.fc2", | |
| "model.vision_tower.encoder.layer.19.attention.k_proj", | |
| "model.vision_tower.encoder.layer.19.attention.projection_layer", | |
| "model.vision_tower.encoder.layer.19.attention.q_proj", | |
| "model.vision_tower.encoder.layer.19.attention.v_proj", | |
| "model.vision_tower.encoder.layer.19.mlp.fc1", | |
| "model.vision_tower.encoder.layer.19.mlp.fc2", | |
| "model.vision_tower.encoder.layer.2.attention.k_proj", | |
| "model.vision_tower.encoder.layer.2.attention.projection_layer", | |
| "model.vision_tower.encoder.layer.2.attention.q_proj", | |
| "model.vision_tower.encoder.layer.2.attention.v_proj", | |
| "model.vision_tower.encoder.layer.2.mlp.fc1", | |
| "model.vision_tower.encoder.layer.2.mlp.fc2", | |
| "model.vision_tower.encoder.layer.20.attention.k_proj", | |
| "model.vision_tower.encoder.layer.20.attention.projection_layer", | |
| "model.vision_tower.encoder.layer.20.attention.q_proj", | |
| "model.vision_tower.encoder.layer.20.attention.v_proj", | |
| "model.vision_tower.encoder.layer.20.mlp.fc1", | |
| "model.vision_tower.encoder.layer.20.mlp.fc2", | |
| "model.vision_tower.encoder.layer.21.attention.k_proj", | |
| "model.vision_tower.encoder.layer.21.attention.projection_layer", | |
| "model.vision_tower.encoder.layer.21.attention.q_proj", | |
| "model.vision_tower.encoder.layer.21.attention.v_proj", | |
| "model.vision_tower.encoder.layer.21.mlp.fc1", | |
| "model.vision_tower.encoder.layer.21.mlp.fc2", | |
| "model.vision_tower.encoder.layer.22.attention.k_proj", | |
| "model.vision_tower.encoder.layer.22.attention.projection_layer", | |
| "model.vision_tower.encoder.layer.22.attention.q_proj", | |
| "model.vision_tower.encoder.layer.22.attention.v_proj", | |
| "model.vision_tower.encoder.layer.22.mlp.fc1", | |
| "model.vision_tower.encoder.layer.22.mlp.fc2", | |
| "model.vision_tower.encoder.layer.23.attention.k_proj", | |
| "model.vision_tower.encoder.layer.23.attention.projection_layer", | |
| "model.vision_tower.encoder.layer.23.attention.q_proj", | |
| "model.vision_tower.encoder.layer.23.attention.v_proj", | |
| "model.vision_tower.encoder.layer.23.mlp.fc1", | |
| "model.vision_tower.encoder.layer.23.mlp.fc2", | |
| "model.vision_tower.encoder.layer.3.attention.k_proj", | |
| "model.vision_tower.encoder.layer.3.attention.projection_layer", | |
| "model.vision_tower.encoder.layer.3.attention.q_proj", | |
| "model.vision_tower.encoder.layer.3.attention.v_proj", | |
| "model.vision_tower.encoder.layer.3.mlp.fc1", | |
| "model.vision_tower.encoder.layer.3.mlp.fc2", | |
| "model.vision_tower.encoder.layer.4.attention.k_proj", | |
| "model.vision_tower.encoder.layer.4.attention.projection_layer", | |
| "model.vision_tower.encoder.layer.4.attention.q_proj", | |
| "model.vision_tower.encoder.layer.4.attention.v_proj", | |
| "model.vision_tower.encoder.layer.4.mlp.fc1", | |
| "model.vision_tower.encoder.layer.4.mlp.fc2", | |
| "model.vision_tower.encoder.layer.5.attention.k_proj", | |
| "model.vision_tower.encoder.layer.5.attention.projection_layer", | |
| "model.vision_tower.encoder.layer.5.attention.q_proj", | |
| "model.vision_tower.encoder.layer.5.attention.v_proj", | |
| "model.vision_tower.encoder.layer.5.mlp.fc1", | |
| "model.vision_tower.encoder.layer.5.mlp.fc2", | |
| "model.vision_tower.encoder.layer.6.attention.k_proj", | |
| "model.vision_tower.encoder.layer.6.attention.projection_layer", | |
| "model.vision_tower.encoder.layer.6.attention.q_proj", | |
| "model.vision_tower.encoder.layer.6.attention.v_proj", | |
| "model.vision_tower.encoder.layer.6.mlp.fc1", | |
| "model.vision_tower.encoder.layer.6.mlp.fc2", | |
| "model.vision_tower.encoder.layer.7.attention.k_proj", | |
| "model.vision_tower.encoder.layer.7.attention.projection_layer", | |
| "model.vision_tower.encoder.layer.7.attention.q_proj", | |
| "model.vision_tower.encoder.layer.7.attention.v_proj", | |
| "model.vision_tower.encoder.layer.7.mlp.fc1", | |
| "model.vision_tower.encoder.layer.7.mlp.fc2", | |
| "model.vision_tower.encoder.layer.8.attention.k_proj", | |
| "model.vision_tower.encoder.layer.8.attention.projection_layer", | |
| "model.vision_tower.encoder.layer.8.attention.q_proj", | |
| "model.vision_tower.encoder.layer.8.attention.v_proj", | |
| "model.vision_tower.encoder.layer.8.mlp.fc1", | |
| "model.vision_tower.encoder.layer.8.mlp.fc2", | |
| "model.vision_tower.encoder.layer.9.attention.k_proj", | |
| "model.vision_tower.encoder.layer.9.attention.projection_layer", | |
| "model.vision_tower.encoder.layer.9.attention.q_proj", | |
| "model.vision_tower.encoder.layer.9.attention.v_proj", | |
| "model.vision_tower.encoder.layer.9.mlp.fc1", | |
| "model.vision_tower.encoder.layer.9.mlp.fc2" | |
| ], | |
| "quant_method": "fp8", | |
| "weight_block_size": [ | |
| 128, | |
| 128 | |
| ] | |
| }, | |
| "text_config": { | |
| "_attn_implementation_autoset": true, | |
| "architectures": [ | |
| "Qwen3ForCausalLM" | |
| ], | |
| "attention_bias": false, | |
| "attention_dropout": 0.0, | |
| "bos_token_id": 151643, | |
| "eos_token_id": 151645, | |
| "head_dim": 128, | |
| "hidden_act": "silu", | |
| "hidden_size": 4096, | |
| "initializer_range": 0.02, | |
| "intermediate_size": 12288, | |
| "max_position_embeddings": 65536, | |
| "max_window_layers": 36, | |
| "model_type": "qwen3", | |
| "num_attention_heads": 32, | |
| "num_hidden_layers": 36, | |
| "num_key_value_heads": 8, | |
| "rms_norm_eps": 1e-06, | |
| "rope_scaling": null, | |
| "rope_theta": 1000000, | |
| "sliding_window": null, | |
| "torch_dtype": "bfloat16", | |
| "use_cache": true, | |
| "use_sliding_window": false, | |
| "vocab_size": 153216 | |
| }, | |
| "torch_dtype": "bfloat16", | |
| "transformers_version": "4.53.0", | |
| "vision_config": { | |
| "_attn_implementation_autoset": true, | |
| "architectures": [ | |
| "InternVisionModel" | |
| ], | |
| "attention_bias": true, | |
| "attention_dropout": 0.0, | |
| "auto_map": { | |
| "AutoConfig": "configuration_interns1.InternS1VisionConfig", | |
| "AutoModel": "modeling_interns1.InternS1VisionModel" | |
| }, | |
| "drop_path_rate": 0.0, | |
| "dropout": 0.0, | |
| "hidden_act": "gelu", | |
| "hidden_dropout_prob": 0.0, | |
| "hidden_size": 1024, | |
| "image_size": [ | |
| 448, | |
| 448 | |
| ], | |
| "initializer_factor": 1.0, | |
| "initializer_range": 0.02, | |
| "intermediate_size": 4096, | |
| "layer_norm_eps": 1e-06, | |
| "layer_scale_init_value": 0.1, | |
| "model_type": "interns1_vision", | |
| "norm_type": "layer_norm", | |
| "num_attention_heads": 16, | |
| "num_channels": 3, | |
| "num_hidden_layers": 24, | |
| "patch_size": [ | |
| 14, | |
| 14 | |
| ], | |
| "projection_dropout": 0.0, | |
| "torch_dtype": "bfloat16", | |
| "use_absolute_position_embeddings": true, | |
| "use_mask_token": false, | |
| "use_mean_pooling": true, | |
| "use_qk_norm": false | |
| }, | |
| "vision_feature_layer": -1, | |
| "vision_feature_select_strategy": "default", | |
| "auto_map": { | |
| "AutoConfig": "configuration_interns1.InternS1Config", | |
| "AutoModel": "modeling_interns1.InternS1Model", | |
| "AutoModelForCausalLM": "modeling_interns1.InternS1ForConditionalGeneration" | |
| } | |
| } |