quant_manifest.json

{
  "config_id": "c2",
  "source_model": "meta-llama/Llama-3.2-1B-Instruct",
  "scheme": "W4A16",
  "method": "gptq",
  "bits_weight": 4,
  "bits_activation": 16,
  "group_size": 128,
  "symmetric": true,
  "calibration_file": "calibration_c4_512x2048.jsonl",
  "calibration_sha256": "b5736155d5b24bf5eb15faae0eb4a742aea4a9d0532db9684085a49750606e19",
  "n_calibration_samples": 512,
  "calibration_seq_len": 2048,
  "quantize_seconds": 418.1,
  "output_dir": "/scratch/jeesup0103/safety_quant/models/Llama-3.2-1B-Instruct-c2"
}