quant_config.json

{
  "bits": 8,
  "group_size": 128
}