quantization_config.json

{
  "quantization": "fp8",
  "quantization_method": "static",
  "original_model": "google/medgemma-27b-text-it",
  "quantization_library": "llm-compressor",
  "notes": "Quantized using medical-specific calibration data"
}