-
Notifications
You must be signed in to change notification settings - Fork 756
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
[Badcase]: 用ollama运行 qwen 2.5. 结尾回复会一直重复 #925
Comments
我这也这样,同问 |
+1 qwen2.5 14b 32k sft ,测试阶段总有重复问题,1.5也是,难顶。 sft时词表需要做啥改动吗? |
@520jefferson I don't think the problem is even the same. The OP's question is about Ollama and llama.cpp quantization, which are all instruct models. If you have finetuned the base model, are you sure you set the hyperparameters properly? Please open an issue and describe your issues. |
一样,试了qwen2.5-14b-instruct,qwen2.5-7b-instruct,都会循环生成。没有训练,用的官方提供的gguf文件 |
qwen2.5-72b-instruct,使用 vllm 部署,未经过进一步的微调,同样会出现循环生成的情况 |
这种情况是必现还是偶现,有发生的背景吗。 |
我使用Llama 3.2 3b 也是同样的问题 |
+1 我也遇到了同样的问题 使用vllm 部署了qwen2.5-72b-instruct |
+1 我也遇到了同样的问题,我试了7B,32B,72B的f16的也一样无法解决 |
+1 |
+1,qwen2.5-72b-instruct-awq量化之后 |
+1 |
+1 |
Has this been raised before?
Description
环境 win10 , ollama version is 0.3.11 , qwen2.5:7b-instruct
问题 要求较长的文本,输出到后面会不停的重复,完全没有停止的意思.有时第一次就会出现,有时第一次返回交段,再次要求,就会开始.本地复现机率很高
流程如下:
The text was updated successfully, but these errors were encountered: