Is there an existing issue for this?
- [X] I have searched the existing issues
Current Behavior
求助，我想查看验证集情况，在ds_train_finetune.sh中设置--do_eval，但代码一直报错，大家是怎么解决的呢:
Expected Behavior
正常运行
Steps To Reproduce
Traceback (most recent call last):
File "main.py", line 431, in
main()
File "main.py", line 397, in main
predict_results = trainer.predict(predict_dataset, metric_key_prefix="predict", max_length=512, do_sample=True, top_p=0.7, temperature=0.95)
File "/mnt/model/dengshuhao1/workspace/temp/ChatGLM-6B/ptuning/trainer_seq2seq.py", line 136, in predict
return super().predict(test_dataset, ignore_keys=ignore_keys, metric_key_prefix=metric_key_prefix)
File "/mnt/model/dengshuhao1/workspace/temp/ChatGLM-6B/ptuning/trainer.py", line 3020, in predict
output = eval_loop(
File "/mnt/model/dengshuhao1/workspace/temp/ChatGLM-6B/ptuning/trainer.py", line 3232, in evaluation_loop
metrics = self.compute_metrics(EvalPrediction(predictions=all_preds, label_ids=all_labels))
File "main.py", line 328, in compute_metrics
scores = rouge.get_scores(' '.join(hypothesis) , ' '.join(reference))
File "/mnt/model/dengshuhao1/local/envs/THUDM-ChatGLM-6B/lib/python3.8/site-packages/rouge_chinese/rouge.py", line 116, in get_scores
return self._get_scores(hyps, refs)
File "/mnt/model/dengshuhao1/local/envs/THUDM-ChatGLM-6B/lib/python3.8/site-packages/rouge_chinese/rouge.py", line 129, in _get_scores
sc = fn(
File "/mnt/model/dengshuhao1/local/envs/THUDM-ChatGLM-6B/lib/python3.8/site-packages/rouge_chinese/rouge.py", line 54, in
"rouge-1": lambda hyp, ref, **k: rouge_score.rouge_n(hyp, ref, 1, **k),
File "/mnt/model/dengshuhao1/local/envs/THUDM-ChatGLM-6B/lib/python3.8/site-packages/rouge_chinese/rouge_score.py", line 253, in rouge_n
raise ValueError("Hypothesis is empty.")
ValueError: Hypothesis is empty.
Environment
OS: Ubuntu 20.04
Python: 3.8
Transformers: 4.27.1
PyTorch: 2.0
CUDA Support: True
Anything else?
No response
或者有人能告诉我怎么打印每一步的训练损失和每个epoch的验证损失么，谢谢大家！！！
ValueError: Hypothesis is empty. 这里你预测dev时，输出出现了空值。调整一下max_lenghth,或者你的pading太多了。
这里你预测dev时，输出出现了空值。调整一下max_
@luolanfeixue emmm，你知道怎么打印验证集的损失么，这份代码好像只打印训练损失
大佬们，你们微调成功了吗，我这总是报一些奇奇怪怪的错，能不能把你们使用的包发给我啊，conda list那个，然后大概需要多大显存啊，效率怎么样，4张3090够用不
THUDM
