{
  "IGNORE_INDEX": -100,
  "_name_or_path": "aitqe",
  "anyres_max_patch": 6,
  "anyres_overlap_ratio": 7,
  "architectures": [
    "MMForCausalLM"
  ],
  "attention_bias": false,
  "attention_dropout": 0.0,
  "bos_token_id": 151643,
  "eos_token_id": 151643,
  "eos_token_index": 151643,
  "frame_pattern": "<frame>",
  "hidden_act": "silu",
  "hidden_size": 3584,
  "image_aspect_ratio": "square",
  "image_pattern": "<image>",
  "image_tile_tokens": 729,
  "image_token_index": 151662,
  "initializer_range": 0.02,
  "intermediate_size": 18944,
  "language_model_name_or_path": "Qwen/Qwen2-7B",
  "max_position_embeddings": 131072,
  "max_window_layers": 28,
  "mm_projector_type": "mlp",
  "model_type": "mm",
  "num_attention_heads": 28,
  "num_hidden_layers": 28,
  "num_key_value_heads": 4,
  "pad_token_id": 0,
  "pad_token_index": 151672,
  "pesudo_img_size": 384,
  "pesudo_patch_size": 14,
  "pooling_kernel": 1,
  "pretraining_tp": 1,
  "rms_norm_eps": 1e-06,
  "rope_scaling": null,
  "rope_theta": 1000000.0,
  "sliding_window": 131072,
  "tie_word_embeddings": false,
  "torch_dtype": "bfloat16",
  "transformers_version": "4.41.2",
  "use_cache": true,
  "use_sliding_window": false,
  "vision_model_name_or_path": "google/siglip-so400m-patch14-384",
  "visual_hidden_dim": 1152,
  "vocab_size": 152064
}