OpenGVLab/InternVL3_5-8B-Flash

1{2  "architectures": [3    "InternVLChatModel"4  ],5  "auto_map": {6    "AutoConfig": "configuration_internvl_chat.InternVLChatConfig",7    "AutoModel": "modeling_internvl_chat.InternVLChatModel",8    "AutoModelForCausalLM": "modeling_internvl_chat.InternVLChatModel"9  },10  "downsample_ratio": 0.5,11  "dynamic_image_size": true,12  "eos_token_id": 151645,13  "force_image_size": 448,14  "llm_config": {15    "_name_or_path": "/root/codespace/checkpoints/Qwen3-8B",16    "architectures": [17      "Qwen3ForCausalLM"18    ],19    "attention_bias": false,20    "attention_dropout": 0.0,21    "bos_token_id": 151643,22    "debug": false,23    "eos_token_id": 151645,24    "ep_size": 1,25    "head_dim": 128,26    "hidden_act": "silu",27    "hidden_size": 4096,28    "initializer_range": 0.02,29    "intermediate_size": 12288,30    "max_position_embeddings": 40960,31    "max_window_layers": 36,32    "micro_forward": false,33    "model_type": "qwen3",34    "num_attention_heads": 32,35    "num_hidden_layers": 36,36    "num_key_value_heads": 8,37    "rms_norm_eps": 1e-06,38    "rope_scaling": null,39    "rope_theta": 1000000,40    "skip_checkpoint": false,41    "sliding_window": null,42    "torch_dtype": "bfloat16",43    "use_cache": false,44    "use_deepep": false,45    "use_sliding_window": false,46    "vocab_size": 15193647  },48  "max_dynamic_patch": 12,49  "min_dynamic_patch": 1,50  "model_type": "internvl_chat",51  "pad2square": false,52  "pad_token_id": 151643,53  "ps_version": "v2",54  "select_layer": -1,55  "template": "internvl2_5",56  "tie_word_embeddings": false,57  "torch_dtype": "bfloat16",58  "transformers_version": null,59  "use_backbone_lora": 0,60  "use_llm_lora": 0,61  "use_thumbnail": true,62  "vision_config": {63    "architectures": [64      "InternVisionModel"65    ],66    "attention_dropout": 0.0,67    "auto_map": {68      "AutoConfig": "configuration_intern_vit.InternVisionConfig",69      "AutoModel": "modeling_intern_vit.InternVisionModel"70    },71    "drop_path_rate": 0.1,72    "dropout": 0.0,73    "hidden_act": "gelu",74    "hidden_size": 1024,75    "image_size": 448,76    "initializer_factor": 1.0,77    "initializer_range": 0.02,78    "intermediate_size": 4096,79    "layer_norm_eps": 1e-06,80    "model_type": "intern_vit_6b",81    "norm_type": "layer_norm",82    "num_attention_heads": 16,83    "num_channels": 3,84    "num_hidden_layers": 24,85    "patch_size": 14,86    "qk_normalization": false,87    "qkv_bias": true,88    "torch_dtype": "bfloat16",89    "use_fa3": false,90    "use_flash_attn": true91  },92  "flash_mode": true,93  "flash_relative_threshold": 0.1,94  "flash_absolute_threshold": 0.595}