{
  "base_model": "apple/FastVLM-1.5B",
  "quantization": "pytorch_int4",
  "bits": 4,
  "method": "per-group symmetric",
  "group_size": 128,
  "vision_encoder": "fp16 (not quantized)"
}