quantization_config.json

{
  "bits": 4,
  "group_size": 128,
  "sym": true,
  "data_type": "int",
  "gradient_accumulate_steps": 4,
  "nsamples": 512,
  "low_gpu_mem_usage": true,
  "dataset": "../calibration_datasets/nemotron_sampled_glm_4_5_air/nemotron_calibration_dataset_glm_4_5_air.json",
  "autoround_version": "0.8.0",
  "quant_method": "auto-round",
  "packing_format": "auto_round:auto_gptq",
  "extra_config": {
    "model.layers.1.mlp.gate.gate_linear": {
      "bits": 8
    },
    "model.layers.2.mlp.gate.gate_linear": {
      "bits": 8
    },
    "model.layers.3.mlp.gate.gate_linear": {
      "bits": 8
    },
    "model.layers.4.mlp.gate.gate_linear": {
      "bits": 8
    },
    "model.layers.5.mlp.gate.gate_linear": {
      "bits": 8
    },
    "model.layers.6.mlp.gate.gate_linear": {
      "bits": 8
    },
    "model.layers.7.mlp.gate.gate_linear": {
      "bits": 8
    },
    "model.layers.8.mlp.gate.gate_linear": {
      "bits": 8
    },
    "model.layers.9.mlp.gate.gate_linear": {
      "bits": 8
    },
    "model.layers.10.mlp.gate.gate_linear": {
      "bits": 8
    },
    "model.layers.11.mlp.gate.gate_linear": {
      "bits": 8
    },
    "model.layers.12.mlp.gate.gate_linear": {
      "bits": 8
    },
    "model.layers.13.mlp.gate.gate_linear": {
      "bits": 8
    },
    "model.layers.14.mlp.gate.gate_linear": {
      "bits": 8
    },
    "model.layers.15.mlp.gate.gate_linear": {
      "bits": 8
    },
    "model.layers.16.mlp.gate.gate_linear": {
      "bits": 8
    },
    "model.layers.17.mlp.gate.gate_linear": {
      "bits": 8
    },
    "model.layers.18.mlp.gate.gate_linear": {
      "bits": 8
    },
    "model.layers.19.mlp.gate.gate_linear": {
      "bits": 8
    },
    "model.layers.20.mlp.gate.gate_linear": {
      "bits": 8
    },
    "model.layers.21.mlp.gate.gate_linear": {
      "bits": 8
    },
    "model.layers.22.mlp.gate.gate_linear": {
      "bits": 8
    },
    "model.layers.23.mlp.gate.gate_linear": {
      "bits": 8
    },
    "model.layers.24.mlp.gate.gate_linear": {
      "bits": 8
    },
    "model.layers.25.mlp.gate.gate_linear": {
      "bits": 8
    },
    "model.layers.26.mlp.gate.gate_linear": {
      "bits": 8
    },
    "model.layers.27.mlp.gate.gate_linear": {
      "bits": 8
    },
    "model.layers.28.mlp.gate.gate_linear": {
      "bits": 8
    },
    "model.layers.29.mlp.gate.gate_linear": {
      "bits": 8
    },
    "model.layers.30.mlp.gate.gate_linear": {
      "bits": 8
    },
    "model.layers.31.mlp.gate.gate_linear": {
      "bits": 8
    },
    "model.layers.32.mlp.gate.gate_linear": {
      "bits": 8
    },
    "model.layers.33.mlp.gate.gate_linear": {
      "bits": 8
    },
    "model.layers.34.mlp.gate.gate_linear": {
      "bits": 8
    },
    "model.layers.35.mlp.gate.gate_linear": {
      "bits": 8
    },
    "model.layers.36.mlp.gate.gate_linear": {
      "bits": 8
    },
    "model.layers.37.mlp.gate.gate_linear": {
      "bits": 8
    },
    "model.layers.38.mlp.gate.gate_linear": {
      "bits": 8
    },
    "model.layers.39.mlp.gate.gate_linear": {
      "bits": 8
    },
    "model.layers.40.mlp.gate.gate_linear": {
      "bits": 8
    },
    "model.layers.41.mlp.gate.gate_linear": {
      "bits": 8
    },
    "model.layers.42.mlp.gate.gate_linear": {
      "bits": 8
    },
    "model.layers.43.mlp.gate.gate_linear": {
      "bits": 8
    },
    "model.layers.44.mlp.gate.gate_linear": {
      "bits": 8
    },
    "model.layers.45.mlp.gate.gate_linear": {
      "bits": 8
    }
  }
}