quantization_metadata.json

{
  "config": {
    "model": {
      "checkpoint": "/tmp/nemotron-3-embed-1b-bf16",
      "trust_remote_code": true,
      "dtype": "bfloat16",
      "device": "cuda"
    },
    "calibration": {
      "sources": [
        {
          "dataset": "abisee/cnn_dailymail",
          "dataset_config": "3.0.0",
          "split": "train",
          "text_field": "article",
          "num_samples": 256,
          "prompt_prefix": "passage: ",
          "seed": 42
        },
        {
          "dataset": "abisee/cnn_dailymail",
          "dataset_config": "3.0.0",
          "split": "train",
          "text_field": "highlights",
          "num_samples": 256,
          "prompt_prefix": "query: ",
          "seed": 43
        }
      ],
      "batch_size": 1,
      "max_length": 16384,
      "seed": 42
    },
    "quantization": {
      "recipe": "mtq.NVFP4_DEFAULT_CFG",
      "precision": "nvfp4",
      "output_dir": "/tmp/nemotron-3-embed-1b-nvfp4",
      "save_modelopt_state": false,
      "export_hf_checkpoint": true
    },
    "quant_cfg_overrides": [
      {
        "quantizer_name": "*embed_tokens*",
        "enable": false
      },
      {
        "quantizer_name": "lm_head*",
        "enable": false
      }
    ]
  },
  "quant_cfg": {
    "quant_cfg": [
      {
        "quantizer_name": "*",
        "enable": false
      },
      {
        "quantizer_name": "*weight_quantizer",
        "cfg": {
          "num_bits": [
            2,
            1
          ],
          "block_sizes": {
            "-1": 16,
            "type": "dynamic",
            "scale_bits": [
              4,
              3
            ]
          }
        }
      },
      {
        "quantizer_name": "*input_quantizer",
        "cfg": {
          "num_bits": [
            2,
            1
          ],
          "block_sizes": {
            "-1": 16,
            "type": "dynamic",
            "scale_bits": [
              4,
              3
            ]
          }
        }
      },
      {
        "parent_class": "nn.BatchNorm1d",
        "quantizer_name": "*",
        "enable": false
      },
      {
        "parent_class": "nn.BatchNorm2d",
        "quantizer_name": "*",
        "enable": false
      },
      {
        "parent_class": "nn.BatchNorm3d",
        "quantizer_name": "*",
        "enable": false
      },
      {
        "parent_class": "nn.LeakyReLU",
        "quantizer_name": "*",
        "enable": false
      },
      {
        "quantizer_name": "*lm_head*",
        "enable": false
      },
      {
        "quantizer_name": "*proj_out.*",
        "enable": false
      },
      {
        "quantizer_name": "*block_sparse_moe.gate*",
        "enable": false
      },
      {
        "quantizer_name": "*router*",
        "enable": false
      },
      {
        "quantizer_name": "*mlp.gate.*",
        "enable": false
      },
      {
        "quantizer_name": "*mlp.shared_expert_gate.*",
        "enable": false
      },
      {
        "quantizer_name": "*linear_attn.conv1d*",
        "enable": false
      },
      {
        "quantizer_name": "*mixer.conv1d*",
        "enable": false
      },
      {
        "quantizer_name": "*output_layer*",
        "enable": false
      },
      {
        "quantizer_name": "output.*",
        "enable": false
      },
      {
        "quantizer_name": "*embed_tokens*",
        "enable": false
      },
      {
        "quantizer_name": "lm_head*",
        "enable": false
      }
    ],
    "algorithm": "max"
  },
  "quant_type": "nvfp4",
  "modelopt_quantized": true
}