{ "architectures": [ "InternS1ForConditionalGeneration" ], "auto_map": { "AutoConfig": "configuration_interns1.InternS1Config", "AutoModel": "modeling_interns1.InternS1Model", "AutoModelForCausalLM": "modeling_interns1.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.language_model.layers.0.mlp.gate", "model.language_model.layers.1.mlp.gate", "model.language_model.layers.10.mlp.gate", "model.language_model.layers.11.mlp.gate", "model.language_model.layers.12.mlp.gate", "model.language_model.layers.13.mlp.gate", "model.language_model.layers.14.mlp.gate", "model.language_model.layers.15.mlp.gate", "model.language_model.layers.16.mlp.gate", "model.language_model.layers.17.mlp.gate", "model.language_model.layers.18.mlp.gate", "model.language_model.layers.19.mlp.gate", "model.language_model.layers.2.mlp.gate", "model.language_model.layers.20.mlp.gate", "model.language_model.layers.21.mlp.gate", "model.language_model.layers.22.mlp.gate", "model.language_model.layers.23.mlp.gate", "model.language_model.layers.24.mlp.gate", "model.language_model.layers.25.mlp.gate", "model.language_model.layers.26.mlp.gate", "model.language_model.layers.27.mlp.gate", "model.language_model.layers.28.mlp.gate", "model.language_model.layers.29.mlp.gate", "model.language_model.layers.3.mlp.gate", "model.language_model.layers.30.mlp.gate", "model.language_model.layers.31.mlp.gate", "model.language_model.layers.32.mlp.gate", "model.language_model.layers.33.mlp.gate", "model.language_model.layers.34.mlp.gate", "model.language_model.layers.35.mlp.gate", "model.language_model.layers.36.mlp.gate", "model.language_model.layers.37.mlp.gate", "model.language_model.layers.38.mlp.gate", "model.language_model.layers.39.mlp.gate", "model.language_model.layers.4.mlp.gate", "model.language_model.layers.40.mlp.gate", "model.language_model.layers.41.mlp.gate", "model.language_model.layers.42.mlp.gate", "model.language_model.layers.43.mlp.gate", "model.language_model.layers.44.mlp.gate", "model.language_model.layers.45.mlp.gate", "model.language_model.layers.46.mlp.gate", "model.language_model.layers.47.mlp.gate", "model.language_model.layers.48.mlp.gate", "model.language_model.layers.49.mlp.gate", "model.language_model.layers.5.mlp.gate", "model.language_model.layers.50.mlp.gate", "model.language_model.layers.51.mlp.gate", "model.language_model.layers.52.mlp.gate", "model.language_model.layers.53.mlp.gate", "model.language_model.layers.54.mlp.gate", "model.language_model.layers.55.mlp.gate", "model.language_model.layers.56.mlp.gate", "model.language_model.layers.57.mlp.gate", "model.language_model.layers.58.mlp.gate", "model.language_model.layers.59.mlp.gate", "model.language_model.layers.6.mlp.gate", "model.language_model.layers.60.mlp.gate", "model.language_model.layers.61.mlp.gate", "model.language_model.layers.62.mlp.gate", "model.language_model.layers.63.mlp.gate", "model.language_model.layers.64.mlp.gate", "model.language_model.layers.65.mlp.gate", "model.language_model.layers.66.mlp.gate", "model.language_model.layers.67.mlp.gate", "model.language_model.layers.68.mlp.gate", "model.language_model.layers.69.mlp.gate", "model.language_model.layers.7.mlp.gate", "model.language_model.layers.70.mlp.gate", "model.language_model.layers.71.mlp.gate", "model.language_model.layers.72.mlp.gate", "model.language_model.layers.73.mlp.gate", "model.language_model.layers.74.mlp.gate", "model.language_model.layers.75.mlp.gate", "model.language_model.layers.76.mlp.gate", "model.language_model.layers.77.mlp.gate", "model.language_model.layers.78.mlp.gate", "model.language_model.layers.79.mlp.gate", "model.language_model.layers.8.mlp.gate", "model.language_model.layers.80.mlp.gate", "model.language_model.layers.81.mlp.gate", "model.language_model.layers.82.mlp.gate", "model.language_model.layers.83.mlp.gate", "model.language_model.layers.84.mlp.gate", "model.language_model.layers.85.mlp.gate", "model.language_model.layers.86.mlp.gate", "model.language_model.layers.87.mlp.gate", "model.language_model.layers.88.mlp.gate", "model.language_model.layers.89.mlp.gate", "model.language_model.layers.9.mlp.gate", "model.language_model.layers.90.mlp.gate", "model.language_model.layers.91.mlp.gate", "model.language_model.layers.92.mlp.gate", "model.language_model.layers.93.mlp.gate", "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.24.attention.k_proj", "model.vision_tower.encoder.layer.24.attention.projection_layer", "model.vision_tower.encoder.layer.24.attention.q_proj", "model.vision_tower.encoder.layer.24.attention.v_proj", "model.vision_tower.encoder.layer.24.mlp.fc1", "model.vision_tower.encoder.layer.24.mlp.fc2", "model.vision_tower.encoder.layer.25.attention.k_proj", "model.vision_tower.encoder.layer.25.attention.projection_layer", "model.vision_tower.encoder.layer.25.attention.q_proj", "model.vision_tower.encoder.layer.25.attention.v_proj", "model.vision_tower.encoder.layer.25.mlp.fc1", "model.vision_tower.encoder.layer.25.mlp.fc2", "model.vision_tower.encoder.layer.26.attention.k_proj", "model.vision_tower.encoder.layer.26.attention.projection_layer", "model.vision_tower.encoder.layer.26.attention.q_proj", "model.vision_tower.encoder.layer.26.attention.v_proj", "model.vision_tower.encoder.layer.26.mlp.fc1", "model.vision_tower.encoder.layer.26.mlp.fc2", "model.vision_tower.encoder.layer.27.attention.k_proj", "model.vision_tower.encoder.layer.27.attention.projection_layer", "model.vision_tower.encoder.layer.27.attention.q_proj", "model.vision_tower.encoder.layer.27.attention.v_proj", "model.vision_tower.encoder.layer.27.mlp.fc1", "model.vision_tower.encoder.layer.27.mlp.fc2", "model.vision_tower.encoder.layer.28.attention.k_proj", "model.vision_tower.encoder.layer.28.attention.projection_layer", "model.vision_tower.encoder.layer.28.attention.q_proj", "model.vision_tower.encoder.layer.28.attention.v_proj", "model.vision_tower.encoder.layer.28.mlp.fc1", "model.vision_tower.encoder.layer.28.mlp.fc2", "model.vision_tower.encoder.layer.29.attention.k_proj", "model.vision_tower.encoder.layer.29.attention.projection_layer", "model.vision_tower.encoder.layer.29.attention.q_proj", "model.vision_tower.encoder.layer.29.attention.v_proj", "model.vision_tower.encoder.layer.29.mlp.fc1", "model.vision_tower.encoder.layer.29.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.30.attention.k_proj", "model.vision_tower.encoder.layer.30.attention.projection_layer", "model.vision_tower.encoder.layer.30.attention.q_proj", "model.vision_tower.encoder.layer.30.attention.v_proj", "model.vision_tower.encoder.layer.30.mlp.fc1", "model.vision_tower.encoder.layer.30.mlp.fc2", "model.vision_tower.encoder.layer.31.attention.k_proj", "model.vision_tower.encoder.layer.31.attention.projection_layer", "model.vision_tower.encoder.layer.31.attention.q_proj", "model.vision_tower.encoder.layer.31.attention.v_proj", "model.vision_tower.encoder.layer.31.mlp.fc1", "model.vision_tower.encoder.layer.31.mlp.fc2", "model.vision_tower.encoder.layer.32.attention.k_proj", "model.vision_tower.encoder.layer.32.attention.projection_layer", "model.vision_tower.encoder.layer.32.attention.q_proj", "model.vision_tower.encoder.layer.32.attention.v_proj", "model.vision_tower.encoder.layer.32.mlp.fc1", "model.vision_tower.encoder.layer.32.mlp.fc2", "model.vision_tower.encoder.layer.33.attention.k_proj", "model.vision_tower.encoder.layer.33.attention.projection_layer", "model.vision_tower.encoder.layer.33.attention.q_proj", "model.vision_tower.encoder.layer.33.attention.v_proj", "model.vision_tower.encoder.layer.33.mlp.fc1", "model.vision_tower.encoder.layer.33.mlp.fc2", "model.vision_tower.encoder.layer.34.attention.k_proj", "model.vision_tower.encoder.layer.34.attention.projection_layer", "model.vision_tower.encoder.layer.34.attention.q_proj", "model.vision_tower.encoder.layer.34.attention.v_proj", "model.vision_tower.encoder.layer.34.mlp.fc1", "model.vision_tower.encoder.layer.34.mlp.fc2", "model.vision_tower.encoder.layer.35.attention.k_proj", "model.vision_tower.encoder.layer.35.attention.projection_layer", "model.vision_tower.encoder.layer.35.attention.q_proj", "model.vision_tower.encoder.layer.35.attention.v_proj", "model.vision_tower.encoder.layer.35.mlp.fc1", "model.vision_tower.encoder.layer.35.mlp.fc2", "model.vision_tower.encoder.layer.36.attention.k_proj", "model.vision_tower.encoder.layer.36.attention.projection_layer", "model.vision_tower.encoder.layer.36.attention.q_proj", "model.vision_tower.encoder.layer.36.attention.v_proj", "model.vision_tower.encoder.layer.36.mlp.fc1", "model.vision_tower.encoder.layer.36.mlp.fc2", "model.vision_tower.encoder.layer.37.attention.k_proj", "model.vision_tower.encoder.layer.37.attention.projection_layer", "model.vision_tower.encoder.layer.37.attention.q_proj", "model.vision_tower.encoder.layer.37.attention.v_proj", "model.vision_tower.encoder.layer.37.mlp.fc1", "model.vision_tower.encoder.layer.37.mlp.fc2", "model.vision_tower.encoder.layer.38.attention.k_proj", "model.vision_tower.encoder.layer.38.attention.projection_layer", "model.vision_tower.encoder.layer.38.attention.q_proj", "model.vision_tower.encoder.layer.38.attention.v_proj", "model.vision_tower.encoder.layer.38.mlp.fc1", "model.vision_tower.encoder.layer.38.mlp.fc2", "model.vision_tower.encoder.layer.39.attention.k_proj", "model.vision_tower.encoder.layer.39.attention.projection_layer", "model.vision_tower.encoder.layer.39.attention.q_proj", "model.vision_tower.encoder.layer.39.attention.v_proj", "model.vision_tower.encoder.layer.39.mlp.fc1", "model.vision_tower.encoder.layer.39.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.40.attention.k_proj", "model.vision_tower.encoder.layer.40.attention.projection_layer", "model.vision_tower.encoder.layer.40.attention.q_proj", "model.vision_tower.encoder.layer.40.attention.v_proj", "model.vision_tower.encoder.layer.40.mlp.fc1", "model.vision_tower.encoder.layer.40.mlp.fc2", "model.vision_tower.encoder.layer.41.attention.k_proj", "model.vision_tower.encoder.layer.41.attention.projection_layer", "model.vision_tower.encoder.layer.41.attention.q_proj", "model.vision_tower.encoder.layer.41.attention.v_proj", "model.vision_tower.encoder.layer.41.mlp.fc1", "model.vision_tower.encoder.layer.41.mlp.fc2", "model.vision_tower.encoder.layer.42.attention.k_proj", "model.vision_tower.encoder.layer.42.attention.projection_layer", "model.vision_tower.encoder.layer.42.attention.q_proj", "model.vision_tower.encoder.layer.42.attention.v_proj", "model.vision_tower.encoder.layer.42.mlp.fc1", "model.vision_tower.encoder.layer.42.mlp.fc2", "model.vision_tower.encoder.layer.43.attention.k_proj", "model.vision_tower.encoder.layer.43.attention.projection_layer", "model.vision_tower.encoder.layer.43.attention.q_proj", "model.vision_tower.encoder.layer.43.attention.v_proj", "model.vision_tower.encoder.layer.43.mlp.fc1", "model.vision_tower.encoder.layer.43.mlp.fc2", "model.vision_tower.encoder.layer.44.attention.k_proj", "model.vision_tower.encoder.layer.44.attention.projection_layer", "model.vision_tower.encoder.layer.44.attention.q_proj", "model.vision_tower.encoder.layer.44.attention.v_proj", "model.vision_tower.encoder.layer.44.mlp.fc1", "model.vision_tower.encoder.layer.44.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": { "architectures": [ "Qwen3MoeForCausalLM" ], "attention_bias": false, "attention_dropout": 0.0, "bos_token_id": 151643, "decoder_sparse_step": 1, "eos_token_id": 151645, "head_dim": 128, "hidden_act": "silu", "hidden_size": 4096, "initializer_range": 0.02, "intermediate_size": 8192, "max_position_embeddings": 65536, "max_window_layers": 94, "mlp_only_layers": [], "model_type": "qwen3_moe", "moe_intermediate_size": 1536, "norm_topk_prob": true, "num_attention_heads": 64, "num_experts": 128, "num_experts_per_tok": 8, "num_hidden_layers": 94, "num_key_value_heads": 4, "output_router_logits": false, "rms_norm_eps": 1e-06, "rope_scaling": null, "rope_theta": 1000000.0, "router_aux_loss_coef": 0.001, "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": { "architectures": [ "InternVisionModel" ], "attention_bias": false, "attention_dropout": 0.0, "auto_map": { "AutoConfig": "configuration_interns1.InternS1VisionConfig", "AutoModel": "modeling_interns1.InternS1VisionModel" }, "drop_path_rate": 0.1, "dropout": 0.0, "hidden_act": "gelu", "hidden_dropout_prob": 0.0, "hidden_size": 3200, "image_size": [ 448, 448 ], "initializer_factor": 0.1, "initializer_range": 1e-10, "intermediate_size": 12800, "layer_norm_eps": 1e-06, "layer_scale_init_value": 0.1, "model_type": "interns1_vision", "norm_type": "rms_norm", "num_attention_heads": 25, "num_channels": 3, "num_hidden_layers": 45, "patch_size": [ 14, 14 ], "projection_dropout": 0.0, "torch_dtype": "bfloat16", "use_absolute_position_embeddings": true, "use_bfloat16": true, "use_flash_attn": true, "use_mask_token": false, "use_mean_pooling": true, "use_qk_norm": true }, "vision_feature_layer": -1, "vision_feature_select_strategy": "default" }