quantization_config.json

{
  "bits": 4,
  "data_type": "int",
  "group_size": 128,
  "sym": true,
  "iters": 0,
  "autoround_version": "0.12.3",
  "quant_method": "auto-round",
  "packing_format": "auto_round:auto_gptq",
  "extra_config": {
    "model.layers.0.self_attn.g_proj": {
      "bits": 16,
      "data_type": "fp"
    },
    "model.layers.4.self_attn.g_proj": {
      "bits": 16,
      "data_type": "fp"
    },
    "model.layers.8.self_attn.g_proj": {
      "bits": 16,
      "data_type": "fp"
    },
    "model.layers.12.self_attn.g_proj": {
      "bits": 16,
      "data_type": "fp"
    },
    "model.layers.16.self_attn.g_proj": {
      "bits": 16,
      "data_type": "fp"
    },
    "model.layers.20.self_attn.g_proj": {
      "bits": 16,
      "data_type": "fp"
    },
    "model.layers.24.self_attn.g_proj": {
      "bits": 16,
      "data_type": "fp"
    },
    "model.layers.28.self_attn.g_proj": {
      "bits": 16,
      "data_type": "fp"
    },
    "model.layers.32.self_attn.g_proj": {
      "bits": 16,
      "data_type": "fp"
    },
    "model.layers.36.self_attn.g_proj": {
      "bits": 16,
      "data_type": "fp"
    }
  }
}