quantize_config.json

{
  "bits": 4,
  "group_size": 128,
  "desc_act": false,
  "static_groups": false,
  "sym": true,
  "lm_head": false,
  "model_name_or_path": "/data/unsloth/final/Llama-3-Alpha-Ko-8B-Instruct-GPTQ",
  "model_file_base_name": "model",
  "quant_method": "gptq",
  "format": "gptq",
  "meta": {
    "quantizer": "autogptq:0.9.0.dev0",
    "checkpoint_format": "gptq",
    "damp_percent": 0.01,
    "true_sequential": true
  }
}