[THUDM/ChatGLM-6B]ptuning评估的时候报错：The expanded size of the tensor (140) must match the existing size (312) at non-singleton dimension 0. Target sizes: [140]. Tensor sizes: [312]

  File "/ptuning/main.py", line 416, in <module>
    main()
  File "/ptuning/main.py", line 354, in main
    train_result = trainer.train(resume_from_checkpoint=checkpoint)
  File "/home/searchgpt/anaconda3/envs/chatglm_lora/lib/python3.10/site-packages/transformers/trainer.py", line 1633, in train
    return inner_training_loop(
  File "/home/searchgpt/anaconda3/envs/chatglm_lora/lib/python3.10/site-packages/transformers/trainer.py", line 1979, in _inner_training_loop
    self._maybe_log_save_evaluate(tr_loss, model, trial, epoch, ignore_keys_for_eval)
  File "/home/searchgpt/anaconda3/envs/chatglm_lora/lib/python3.10/site-packages/transformers/trainer.py", line 2236, in _maybe_log_save_evaluate
    metrics = self.evaluate(ignore_keys=ignore_keys_for_eval)
  File "/ptuning/trainer_seq2seq.py", line 78, in evaluate
    return super().evaluate(eval_dataset, ignore_keys=ignore_keys, metric_key_prefix=metric_key_prefix)
  File "/home/searchgpt/anaconda3/envs/chatglm_lora/lib/python3.10/site-packages/transformers/trainer.py", line 2932, in evaluate
    output = eval_loop(
  File "/home/searchgpt/anaconda3/envs/chatglm_lora/lib/python3.10/site-packages/transformers/trainer.py", line 3113, in evaluation_loop
    loss, logits, labels = self.prediction_step(model, inputs, prediction_loss_only, ignore_keys=ignore_keys)
  File "/ptuning/trainer_seq2seq.py", line 200, in prediction_step
    generated_tokens = self.model.generate(**gen_kwargs)
  File "/home/searchgpt/anaconda3/envs/chatglm_lora/lib/python3.10/site-packages/torch/utils/_contextlib.py", line 115, in decorate_context
    return func(*args, **kwargs)
  File "/home/searchgpt/anaconda3/envs/chatglm_lora/lib/python3.10/site-packages/transformers/generation/utils.py", line 1490, in generate
    return self.beam_search(
  File "/home/searchgpt/anaconda3/envs/chatglm_lora/lib/python3.10/site-packages/transformers/generation/utils.py", line 2836, in beam_search
    sequence_outputs = beam_scorer.finalize(
  File "/home/searchgpt/anaconda3/envs/chatglm_lora/lib/python3.10/site-packages/transformers/generation/beam_search.py", line 377, in finalize
    decoded[i, : sent_lengths[i]] = hypo
RuntimeError: The expanded size of the tensor (140) must match the existing size (312) at non-singleton dimension 0.  Target sizes: [140].  Tensor sizes: [312]
  3%|█████▍    

File "/home/searchgpt/anaconda3/envs/chatglm_lora/lib/python3.10/site-packages/transformers/trainer.py", line 2236, in _maybe_log_save_evaluate metrics = self.evaluate(ignore_keys=ignore_keys_for_eval)

在评估的时候报错，有一部分数据会正常，然后有有些数据评估的时候报错

运行参数： args = [ '--model_name_or_path=../../pretrained_models/chatglm-6b', '--do_train', '--do_eval', '--do_predict', '--overwrite_cache', '--overwrite_output_dir', '--per_device_train_batch_size=16', '--per_device_eval_batch_size=1', '--gradient_accumulation_steps=1', '--predict_with_generate', '--max_steps=3000', '--logging_steps=50', '--eval_steps=80', '--evaluation_strategy=steps', '--save_steps=1000', '--learning_rate=1e-2', '--output_dir=output', '--pre_seq_len=4', '--quantization_bit=4', '--max_source_length=420', '--max_target_length=140', '--val_max_target_length=140', '--num_beams=4', '--max_eval_samples=100' ]

运行环境：

- tensorboard              2.12.0
- tensorboard-data-server  0.7.0
- tensorboard-plugin-wit   1.8.1
- termcolor                2.2.0
- threadpoolctl            3.1.0
- tokenizers               0.12.1
- torch                    2.0.0
- torchvision              0.15.1
- tqdm                     4.65.0
- transformers             4.27.4
- triton                   2.0.0