Skip to content
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

[Bug] yi-34B-chat-200k返回不全 #2408

Closed
liuzj288 opened this issue May 7, 2024 · 8 comments · Fixed by #2471
Closed

[Bug] yi-34B-chat-200k返回不全 #2408

liuzj288 opened this issue May 7, 2024 · 8 comments · Fixed by #2471
Labels
🐛 Bug Something isn't working | 缺陷

Comments

@liuzj288
Copy link

liuzj288 commented May 7, 2024

💻 系统环境

Windows

📦 部署环境

Official Preview

🌐 浏览器

Chrome

🐛 问题描述

使用yi-34B-chat-200k时出现返回不全的问题,回复继续也没有输出剩余内容

image

🚦 期望结果

No response

📷 复现步骤

No response

📝 补充信息

No response

@liuzj288 liuzj288 added the 🐛 Bug Something isn't working | 缺陷 label May 7, 2024
@lobehubbot
Copy link
Member

👀 @liuzj288

Thank you for raising an issue. We will investigate into the matter and get back to you as soon as possible.
Please make sure you have given us as much context as possible.
非常感谢您提交 issue。我们会尽快调查此事,并尽快回复您。 请确保您已经提供了尽可能多的背景信息。

@liuzj288 liuzj288 changed the title [Bug] 返回不全 [Bug] yi-34B-chat-200k返回不全 May 7, 2024
@arvinxx
Copy link
Contributor

arvinxx commented May 8, 2024

@MapleEve 来帮忙看看?

@MapleEve
Copy link
Contributor

MapleEve commented May 12, 2024

@MapleEve 来帮忙看看?

@liuzj288
Yi的200K最大返回应该也是4096,如果可以的话给个Chrome F12的报错信息才好定位到是什么原因,有可能是流式断了。

另外这个问题出现的是一次还是经常?

@lobehubbot
Copy link
Member

Bot detected the issue body's language is not English, translate it automatically. 👯👭🏻🧑‍🤝‍🧑👫🧑🏿‍🤝‍🧑🏻👩🏾‍🤝‍👨🏿👬🏿


@MapleEve come and help?
@liuzj288
Yi's maximum return of 200K should also be 4096. If possible, please provide an error message from Chrome F12 so that we can locate the cause. It may be that the streaming is interrupted.

Also, does this problem occur once or often?

@liuzj288
Copy link
Author

@MapleEve 来帮忙看看?

@liuzj288 Yi的200K最大返回应该也是4096,如果可以的话给个Chrome F12的报错信息才好定位到是什么原因,有可能是流式断了。

另外这个问题出现的是一次还是经常?

使用这个模型的话是经常这样,需要哪个位置的报错信息?你也可以尝试下,应该很容易复现

image

image

image

@MapleEve
Copy link
Contributor

yi-34B

我稍微看了下,目前官方已经更新了模型名称,如果你是直连的话(非 Ollama 自部署的),等下面这个 PR 通过之后你再试试新的模型名称。

#2471

另外200K 的模型丢失注意力的情况如果不是回复异常中断(在 Chrome 的开发者工具界面会显示链接主动断开),就是模型自身 RAG 的问题。所以我需要一个重现出来的时候,谷歌开发者工具的控制台/网络连接内容。

@lobehubbot
Copy link
Member

Bot detected the issue body's language is not English, translate it automatically. 👯👭🏻🧑‍🤝‍🧑👫🧑🏿‍🤝‍🧑🏻👩🏾‍🤝‍👨🏿👬🏿


yi-34B

I took a look and found that the official model name has been updated. If you are directly connected (non-Ollama self-deployed), you can try the new model name after the PR below is passed.

#2471

In addition, if the 200K model loses attention, if it is not abnormal interruption of reply (the link will be actively disconnected in Chrome's developer tools interface), it is a problem with the RAG of the model itself. So I need a way to reproduce the console/network connection content of Google Developer Tools.

@lobehubbot
Copy link
Member

@liuzj288

This issue is closed, If you have any questions, you can comment and reply.
此问题已经关闭。如果您有任何问题,可以留言并回复。

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
🐛 Bug Something isn't working | 缺陷
Projects
Status: Done
Development

Successfully merging a pull request may close this issue.

4 participants