官方提供的web_demo_2.5.py在mmbench上测试 指令遵从能力很差 不能输出abcd
linwenzhao1 opened this issue · comments
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
![DAF80036-509F-4B1A-A408-B53046B51AF1](https://private-user-images.githubusercontent.com/51796215/339973096-c701853e-3792-43f4-bd06-1a66a89812d0.png?jwt=eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.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.SZdcwdOBk_lCHAughrrWhrQR_q6_1oLK4e0RzaO-kiQ)
期望行为 | Expected Behavior
简洁的答案:A B C D
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
您好,可尝试更直观的prompt,如通过给出全部输出的格式来使模型输出有参考。或者参考我们的评测代码
https://github.com/OpenBMB/MiniCPM-V/blob/main/eval_mm/vlmevalkit/vlmeval/vlm/minicpm_llama3_v_2_5.py#L63
您好,可尝试更直观的prompt,如通过给出全部输出的格式来使模型输出有参考。或者参考我们的评测代码 https://github.com/OpenBMB/MiniCPM-V/blob/main/eval_mm/vlmevalkit/vlmeval/vlm/minicpm_llama3_v_2_5.py#L63
中文的prompt有吗?