quantize_config.json

{
  "bits": 4,
  "compute_dtype": null,
  "damp_percent": 0.01,
  "desc_act": false,
  "enable_minmax_tuning": true,
  "enable_quanted_input": true,
  "group_size": 128,
  "iters": 1000,
  "lr": 0.001,
  "minmax_lr": 0.001,
  "model_file_base_name": "model",
  "model_name_or_path": null,
  "quant_method": "gptq",
  "scale_dtype": "fp16",
  "sym": false,
  "true_sequential": false
}