| In this case, we prefer to only support inference in Transformers and let the third-party library maintained by the ML community deal with the model quantization itself. |
| In this case, we prefer to only support inference in Transformers and let the third-party library maintained by the ML community deal with the model quantization itself. |