File size: 464 Bytes
57bdca5 |
1 2 3 4 5 6 |
Most of the recent quantization method packs int2/int4 weights inside torch.uint8 weights, so this flag should not be really required (set to False by default). is_serializable: A property method to determine whether the method is serializable or not. is_trainable: A property method to determine whether you can fine-tune models on top of the quantization method (with or without PEFT approaches). Write the validate_environment and update_torch_dtype methods. |