quantize_config.json

{
  "modes": [
    "q4",
    "q8"
  ],
  "per_channel": false,
  "reduce_range": false,
  "block_size": null,
  "is_symmetric": true,
  "accuracy_level": null,
  "quant_type": 1
}