Skip to content

[Bug]: 使用国内大模型对话不被允许的内容时显示一部分答案后报错,并且该会话再咨询合法内容会一直报错 #161

@FisherKheb

Description

@FisherKheb

Bug Description

[使用国内大模型对话不被允许的内容时显示一部分答案后报错,并且该会话再咨询合法内容会一直报错。

Image

Steps to Reproduce

  1. HiChat 里面选择国内大模型,输入 2010年诺贝尔和平奖获得者是谁;
  2. 大模型对话显示一点回答后转为报错信息: “OpenAI API error in streaming response: Output data may contain inappropriate content. For details, see: https://help.aliyun.com/zh/model-studio/error-code#inappropriate-content”;
  3. 再输入合法内容咨询,还是报错:“OpenAI API error in streaming response: Output data may contain inappropriate content. For details, see: https://help.aliyun.com/zh/model-studio/error-code#inappropriate-content”;
  4. 切换到其他会话后正常,再回到刚才会话能看见一部分恢复信息;
Image

Expected Behavior

咨询不合法内容应该显示不知道该内容或屏蔽回答;
而且该会话再咨询合法内容时候要正常显示内容。

Actual Behavior

Image Image

Affected Component

Frontend (User Interface)

Severity

Critical (System crash, data loss)

Environment

os:windows10
browser: chrome 版本 143.0.7499.194(正式版本) (64 位)
ENV:HiMarket+阿里云AI网关

Logs and Error Messages

Screenshots

No response

Additional Context

No response

Checklist

  • I have searched existing issues to avoid duplicates
  • I have included all necessary information above
  • I am using the latest version of HiMarket
  • I can reproduce this issue consistently

Metadata

Metadata

Assignees

No one assigned

    Labels

    bugSomething isn't working

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions