{
  "model": "mobileclip2-s2",
  "source": "open_clip MobileCLIP2-S2 pretrained=dfndr2b (timm/MobileCLIP2-S2-OpenCLIP)",
  "image_size": 256,
  "mean": [0.0, 0.0, 0.0],
  "std": [1.0, 1.0, 1.0],
  "interpolation": "bilinear",
  "resize_mode": "shortest",
  "context_len": 77,
  "id_dtype": "int64",
  "dim": 512,
  "image_input": "pixel_values",
  "image_output": "image_features",
  "text_input": "input_ids",
  "text_output": "text_features",
  "note": "dfndr2b uses mean=0/std=1 (pixel/255 only), NOT OpenAI CLIP mean/std. Do not apply CLIP normalization."
}