{
  "architectures": [
    "HunYuanVLVisionTransformer"
  ],
  "attention_dropout": 0.0,
  "dtype": "float32",
  "hidden_act": "silu",
  "hidden_size": 64,
  "img_max_token_num": 4096,
  "intermediate_size": 128,
  "interpolate_mode": "bilinear",
  "max_image_size": 64,
  "max_vit_seq_len": 16,
  "min_image_size": 64,
  "model_type": "hunyuan_vl_vision",
  "num_attention_heads": 4,
  "num_channels": 3,
  "num_hidden_layers": 2,
  "num_key_value_heads": 4,
  "out_hidden_size": 64,
  "patch_size": 16,
  "rms_norm_eps": 1e-05,
  "spatial_merge_size": 1,
  "temporal_patch_size": 1,
  "text_hidden_size": 64,
  "transformers_version": "5.16.0.dev0"
}