two-tiger/MiMo-VRPRM-7B

1{2  "architectures": [3    "Qwen2_5_VLForConditionalGeneration"4  ],5  "attention_bias": true,6  "attention_dropout": 0.0,7  "eos_token_id": 151645,8  "hidden_act": "silu",9  "hidden_size": 4096,10  "image_token_id": 151655,11  "initializer_range": 0.02,12  "intermediate_size": 11008,13  "max_position_embeddings": 128000,14  "max_window_layers": 32,15  "model_type": "qwen2_5_vl",16  "num_attention_heads": 32,17  "num_hidden_layers": 36,18  "num_key_value_heads": 8,19  "pad_token_id": 151643,20  "rms_norm_eps": 1e-05,21  "rope_scaling": {22    "mrope_section": [23      16,24      24,25      2426    ],27    "rope_type": "default",28    "type": "default"29  },30  "rope_theta": 640000.0,31  "sliding_window": 8192,32  "text_config": {33    "architectures": [34      "Qwen2_5_VLForConditionalGeneration"35    ],36    "attention_bias": true,37    "attention_dropout": 0.0,38    "bos_token_id": 151643,39    "eos_token_id": 151645,40    "hidden_act": "silu",41    "hidden_size": 4096,42    "image_token_id": null,43    "initializer_range": 0.02,44    "intermediate_size": 11008,45    "max_position_embeddings": 128000,46    "max_window_layers": 32,47    "model_type": "qwen2_5_vl_text",48    "num_attention_heads": 32,49    "num_hidden_layers": 36,50    "num_key_value_heads": 8,51    "pad_token_id": 151643,52    "rms_norm_eps": 1e-05,53    "rope_scaling": {54      "mrope_section": [55        16,56        24,57        2458      ],59      "rope_type": "default",60      "type": "default"61    },62    "rope_theta": 640000.0,63    "sliding_window": 8192,64    "torch_dtype": "float32",65    "use_cache": true,66    "use_sliding_window": false,67    "video_token_id": null,68    "vision_end_token_id": 151653,69    "vision_start_token_id": 151652,70    "vision_token_id": 151654,71    "vocab_size": 15168072  },73  "tie_word_embeddings": false,74  "torch_dtype": "float32",75  "transformers_version": "4.52.4",76  "use_cache": true,77  "use_sliding_window": false,78  "video_token_id": 151656,79  "vision_config": {80    "depth": 32,81    "fullatt_block_indexes": [82      7,83      15,84      23,85      3186    ],87    "hidden_act": "silu",88    "hidden_size": 1280,89    "in_channels": 3,90    "in_chans": 3,91    "initializer_range": 0.02,92    "intermediate_size": 3456,93    "model_type": "qwen2_5_vl",94    "num_heads": 16,95    "out_hidden_size": 4096,96    "patch_size": 14,97    "spatial_merge_size": 2,98    "spatial_patch_size": 14,99    "temporal_patch_size": 2,100    "tokens_per_second": 2,101    "torch_dtype": "float32",102    "window_size": 112103  },104  "vision_end_token_id": 151653,105  "vision_start_token_id": 151652,106  "vision_token_id": 151654,107  "vocab_size": 151680108}109