o1模型部分时候无法正常回复

[huoxianribao]

📦 部署环境

Official Preview

📌 软件版本

v1.20.2

💻 系统环境

Windows

🌐 浏览器

Chrome

🐛 问题描述

直接使用的在 https://chat-preview.lobehub.com
经查api站，正常吐token，但是lobehub无法正常接收仍然提示504错误，您提示的方法还是希望有一个完整教程
o1mini的部分问题可以正常回答，正常版的o1模型应该都不能回答。
我正在使用o1模型，更改了token数限制



已经汇报过的bug，但是没有一个完整说明

📷 复现步骤

设置的最大token数，使用o1mini模型

🚦 期望结果

正常返回结果

📝 补充信息

No response

[lobehubbot]

👀 @huoxianribao

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。

[lobehubbot]

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

---

📦 Deployment environment

Official Preview

📌 Software version

v1.20.2

💻 System environment

Windows

🌐 Browser

Chrome

🐛 Problem description

Directly used at https://chat-preview.lobehub.com
After checking the API site, the token is spit out normally, but LobeHub cannot receive it normally and still prompts a 504 error. I still hope there will be a complete tutorial for the method you suggested.
Some questions of o1mini can be answered normally, but the normal version of o1 model should not be able to answer them.
I am using the o1 model and changed the token number limit



Bug has been reported, but there is no complete description

📷 Steps to reproduce

The maximum number of tokens set, using the o1mini model

🚦 Expected results

Normal return results

📝 Supplementary information

No response

[dosubot[bot]]

@huoxianribao 你好！我是Dosu，一个可以帮助你解决bug、回答问题并成为贡献者的机器人。我会在等待人类维护人员的过程中帮助你。

I found a similar discussion that might be relevant to your issue. The discussion titled 连接性检查，无相应 mentions that updating to the latest version (124.2) resolved the issue ^[1].

Regarding the 504 Gateway Timeout error, common causes include server overload, network issues, upstream server issues, and configuration errors. Here are some troubleshooting steps you can follow:

1. Retry the Request: Sometimes, the issue is temporary. Retry the request after some time.
2. Check Server Load: Ensure that the server is not overloaded and has sufficient resources.
3. Network Diagnostics: Check for any network issues that might be causing delays.
4. Upstream Server Health: Verify that the upstream server is operational and responding in a timely manner.
5. Configuration Review: Review the server and gateway configurations to ensure they are correct ^[2].

Additionally, setting a high token limit like 32000 can lead to performance issues, increased processing time, and higher costs. It is recommended to set a more reasonable token limit to avoid these issues ^[3].

For the 'o1' model, the framework includes specific error handling mechanisms such as handling invalid API keys, business errors, and location not supported errors. These mechanisms ensure robust error handling and provide meaningful feedback ^[4][5][6].

If you have already tried these steps and the issue persists, consider updating to the latest version of the framework as suggested in the similar discussion.

To continue talking to Dosu, mention @dosu.

[dosubot[bot]]

I found a similar discussion that might be relevant to your issue. The discussion titled 连接性检查，无相应 mentions that updating to the latest version (124.2) resolved the issue ^[1].

Regarding the 504 Gateway Timeout error, common causes include server overload, network issues, upstream server issues, and configuration errors. Here are some troubleshooting steps you can follow:

1. Retry the Request: Sometimes, the issue is temporary. Retry the request after some time.
2. Check Server Load: Ensure that the server is not overloaded and has sufficient resources.
3. Network Diagnostics: Check for any network issues that might be causing delays.
4. Upstream Server Health: Verify that the upstream server is operational and responding in a timely manner.
5. Configuration Review: Review the server and gateway configurations to ensure they are correct ^[2].

Additionally, setting a high token limit like 32000 can lead to performance issues, increased processing time, and higher costs. It is recommended to set a more reasonable token limit to avoid these issues ^[3].

For the 'o1' model, the framework includes specific error handling mechanisms such as handling invalid API keys, business errors, and location not supported errors. These mechanisms ensure robust error handling and provide meaningful feedback ^[4][5][6].

If you have already tried these steps and the issue persists, consider updating to the latest version of the framework as suggested in the similar discussion.

To continue talking to Dosu, mention @dosu.

[arvinxx]

504 是服务器响应超时导致的，Vercel Edge Runtime 对首字节有25s内响应的要求，应用层解决不了。

临时解法是切到客户端请求模式，或者你自己用 Docker 部署。

[lobehubbot]

✅ @huoxianribao

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。

[lobehubbot]

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

---

504 is caused by server response timeout. Vercel Edge Runtime has a requirement to respond to the first byte within 25 seconds, and the application layer cannot solve it.

The temporary solution is to switch to client request mode, or deploy it yourself using Docker.