quantization_config.json

{
  "quantization_method": "pytorch_dynamic_int8",
  "bits": 8,
  "target_modules": [
    "patch_embedding",
    "base.transformer.wte",
    "base.transformer.wpe"
  ],
  "quantized_at": "2025-08-28",
  "base_model": "sander-wood/notagen"
}