{
  "model_type": "hunyuan_ocr",
  "params": {
    "vision_param": "hunyuan_ocr_vision_encoder.ncnn.param",
    "vision_bin": "hunyuan_ocr_vision_encoder.ncnn.bin",
    "text_embed_param": "hunyuan_ocr_text_embed.ncnn.param",
    "text_embed_bin": "hunyuan_ocr_text_embed.ncnn.bin",
    "text_decoder_param": "hunyuan_ocr_text_decoder.ncnn.param",
    "text_decoder_bin": "hunyuan_ocr_text_decoder.ncnn.bin",
    "lm_head_param": "hunyuan_ocr_lm_head.ncnn.param",
    "lm_head_bin": "hunyuan_ocr_text_embed.ncnn.bin"
  },
  "tokenizer": {
    "type": "bbpe",
    "vocab_file": "vocab.txt",
    "merges_file": "merges.txt",
    "bos": "<｜hy_begin▁of▁sentence｜>",
    "eos": "<｜hy_place▁holder▁no▁2｜>",
    "pad": "<｜hy_▁pad▁｜>",
    "eos_ids": [120007, 120020],
    "additional_special_tokens": [
      "<｜hy_begin▁of▁sentence｜>",
      "<｜hy_place▁holder▁no▁2｜>",
      "<｜hy_place▁holder▁no▁3｜>",
      "<｜hy_User｜>",
      "<｜hy_Assistant｜>",
      "<｜hy_place▁holder▁no▁100｜>",
      "<｜hy_place▁holder▁no▁101｜>",
      "<｜hy_place▁holder▁no▁102｜>"
    ]
  },
  "setting": {
    "attn_cnt": 24,
    "hidden_size": 1024,
    "head_dim": 128,
    "num_attention_heads": 16,
    "num_key_value_heads": 8,
    "vocab_size": 120818,
    "kv_cache": false,
    "rope": {
      "type": "xdrope",
      "rope_head_dim": 128,
      "rope_theta": 10000.0,
      "alpha": 1000.0,
      "xdrope_section": [16, 16, 16, 16]
    },
    "bos_token_id": 120000,
    "system_end_token_id": 120021,
    "user_end_token_id": 120006,
    "image_token_id": 120120,
    "image_start_token_id": 120118,
    "image_end_token_id": 120119,
    "special_token_id_begin": 120000,
    "vision": {
      "type": "hunyuan_vl",
      "patch_size": 16,
      "spatial_merge_size": 2,
      "vision_hidden_size": 1152,
      "vision_out_hidden_size": 1024,
      "min_pixels": 262144,
      "max_pixels": 4194304,
      "image_mean": [0.48145466, 0.4578275, 0.40821073],
      "image_std": [0.26862954, 0.26130258, 0.27577711],
      "resample": "bicubic"
    }
  }
}