Class CLIPVisionCfg

TinyCLIP/src/open_clip/model.py:570–584

@dataclass
class CLIPVisionCfg:
    layers: Union[Tuple[int, int, int, int], int] = 12
    width: int = 768
    teacher_width: int = -1
    head_width: int = 64
    mlp_ratio: float = 4.0
    patch_size: int = 16
    image_size: Union[Tuple[int, int], int] = 224
    timm_model_name: str = None  # a valid model name overrides layers, width, patch_size
    # use (imagenet) pretrained weights for named model
    timm_model_pretrained: bool = False
    # feature pooling for timm model ('abs_attn', 'rot_attn', 'avg', '')
    timm_pool: str = 'avg'
    # linear projection for timm model output ('linear', 'mlp', '')
    timm_proj: str = 'linear'


@dataclass