quantize_config.json

{
  "bits": 4,
  "dataset": "wikitext2",
  "group_size": 128,
  "damp_percent": 0.1,
  "desc_act": false,
  "sym": true,
  "true_sequential": true,
  "use_cuda_fp16": true,
  "model_seqlen": 4096,
  "block_name_to_quantize": "model.layers",
  "module_name_preceding_first_block": [
    "model.embed_tokens"
  ],
  "batch_size": 1,
  "pad_token_id": null,
  "disable_exllama": true,
  "max_input_length": null,
  "quant_method": "gptq"
}