quantize_result.json

{
  "method": "nvfp4",
  "model": "/mnt/data/qwythos-9b-v2-src",
  "output_dir": "/mnt/models/quantized/Qwythos-9B-v2-NVFP4",
  "calib_dataset": "ultrachat_200k",
  "calib_samples": 512,
  "load_time_s": 3.2,
  "quant_time_s": 208.6,
  "total_time_s": 223.0,
  "output_size_gb": 10.4,
  "vram_after_load": {
    "gpu_0": "3.2GB allocated, 3.2GB reserved"
  }
}