[BUG] <title> Multi turns conversation 's output is not complete!

[FantasticZihao]

是否已有关于该错误的issue或讨论？ | Is there an existing issue / discussion for this?

• 我已经搜索过已有的issues和讨论 | I have searched the existing issues / discussions

该问题是否在FAQ中有解答？ | Is there an existing answer for this in FAQ?

• 我已经搜索过FAQ | I have searched FAQ

当前行为 | Current Behavior

When I used the Minicpm-llama3-v2.5 to make multi turns conversation for one image, I found the model's output is not complete.

期望行为 | Expected Behavior

I appreciate the model's complete answer when doing multi turns conversation based on a single image for one dataset.

复现方法 | Steps To Reproduce

No response

运行环境 | Environment

- OS:
- Python:
- Transformers:
- PyTorch:
- CUDA (`python -c 'import torch; print(torch.version.cuda)'`):

备注 | Anything else?

No response

[FantasticZihao]

Thanks for your excellent work! I am eager to hearing your reply.

[LDLINGLINGLING]

hello ,please show me your example.

[FantasticZihao]

hello ,please show me your example.

code

captions

As is shown, sometimes the answer is not complete.

[FantasticZihao]

hello ,please show me your example.

Hello, can you give me some suggestions?

[LDLINGLINGLING]

hello，Do the following problems occur frequently? I think you can try some different sampling arguments。