참고 : https://github.com/huggingface/transformers/issues/27397
SafetensorError: Error while deserializing header: InvalidHeaderDeserialization when open .safetensor model · Issue #27397 · h
System Info Hi guys, i just fine tune alpaca (LLaMA 7B base model) with custom dataset and using trainer API. After completing the training process, I received the following error: SafetensorError ...
github.com
PEFT 랑 torch.complie 이랑 같이 쓰면 오류가 남.torch.complie 을 주석처리 해주고 train을 돌리면 괜찮다고 한다.
model.state_dict = (
lambda self, *_, **__: get_peft_model_state_dict(
self, old_state_dict()
)
).__get__(model, type(model))
# if torch.__version__ >= "2" and sys.platform != "win32":
# model = torch.compile(model)
## safeTensor is not working with compile
## https://github.com/huggingface/transformers/issues/27397
trainer.train(resume_from_checkpoint=resume_from_checkpoint)
model.save_pretrained(output_dir)
torch.save(slama_model.embeddings, os.path.join(output_dir, "embeddings.pth"))
728x90
728x90