[Question]: _pickle.UnpicklingError: invalid load key, '\x00'. When I use paddle3.0.0 and paddleNLP3.0.0.b4, to sft uie0.5b model.

[funny000]

请提出你的问题

Traceback (most recent call last):
File "/works/PaddleNLP-3.0.0-beta4/llm/run_finetune.py", line 716, in
main()
File "/works/PaddleNLP-3.0.0-beta4/llm/run_finetune.py", line 450, in main
train_result = trainer.train(resume_from_checkpoint=checkpoint)
File "/usr/local/lib/python3.10/dist-packages/paddlenlp/trainer/trainer.py", line 892, in train
return self._inner_training_loop(
File "/usr/local/lib/python3.10/dist-packages/paddlenlp/trainer/trainer.py", line 1303, in _inner_training_loop
self._maybe_log_save_evaluate(tr_loss, model, epoch, ignore_keys_for_eval, inputs=inputs)
File "/usr/local/lib/python3.10/dist-packages/paddlenlp/trainer/trainer.py", line 1584, in _maybe_log_save_evaluate
self._save_checkpoint(model, metrics=metrics)
File "/usr/local/lib/python3.10/dist-packages/paddlenlp/trainer/trainer.py", line 2628, in _save_checkpoint
paddle.distributed.all_gather_object(rng_states_list, rng_states)
File "/usr/local/lib/python3.10/dist-packages/paddle/distributed/communication/all_gather.py", line 144, in all_gather_object
convert_tensor_to_object(tensor, list_len_of_tensor[i])
File "/usr/local/lib/python3.10/dist-packages/paddle/distributed/communication/serialization_utils.py", line 43, in convert_tensor_to_object
raise e
File "/usr/local/lib/python3.10/dist-packages/paddle/distributed/communication/serialization_utils.py", line 40, in convert_tensor_to_object
return _unpickler(io.BytesIO(data)).load()
_pickle.UnpicklingError: invalid load key, '\x00'.

[funny000]

the used of command
python -u -m paddle.distributed.launch --gpus "0,1,2,3,4,5,6,7" run_finetune.py ./config/uie05b/sft_argument.json