InternLM / InternLM/lagent

openai 中的 GPTAPI 的参数配置错误导致除非流式外的智能体无法处理通义千问模型的响应

Open
#286 1 comment 0 reactions 0 assignees View on GitHub
Dominant language
Python
Stars
2.3k
Forks
243
PR merge metrics
No merged PRs in 30d

Description

# 所用版本

提交 #285,main 分支。

# 问题概述

使用如下代码尝试获取智能体回复时:

```python
from lagent.agents import Agent
from lagent.schema import AgentMessage
from lagent.llms import GPTAPI

# qwen 智能体配置来自项目 MindSearch
url = "https://dashscope.aliyuncs.com/api/v1/services/aigc/text-generation/generation"
qwen = dict(
type=GPTAPI,
model_type="qwen-max-longcontext",
key="sk-apikey",
api_base=url,
meta_template=[
dict(role="system", api_role="system"),
dict(role="user", api_role="user"),
dict(role="assistant", api_role="assistant"),
dict(role="environment", api_role="system"),
],
top_p=0.8,
temperature=0,
max_new_tokens=4096,
repetition_penalty=1.02,
stop_words=["<|im_end|>"],
)

system_prompt = '你的回答只能从“典”、“孝”、“急”三个字中选一个。'
agent = Agent(qwen, system_prompt)

user_msg = AgentMessage(sender='user', content='今天天气情况')
bot_msg = agent(user_msg)
print(bot_msg)
```

终端输出:

![image](https://github.com/user-attachments/assets/22176dbc-8b92-4f90-8b1a-65607001695c)

经过定位来源于文件 `lagent/llms/openai.py` :

![image](https://github.com/user-attachments/assets/11138d8d-c634-4045-9f00-9bdac98900cb)

# 原因概述

导致 JSON 无法解码的原因是 JSON 数据开头携带了

```
id:1
event:result
:HTTP_STATUS/200
```

此项内容出现的原因是由于在请求通义千问模型的时候在请求头携带了 `X-DashScope-SSE: enbale` 参数,根据 [通义千问API文档](https://help.aliyun.com/zh/model-studio/developer-reference/use-qwen-by-calling-api) 此参数应该仅需要在流式请求时携带即可:

![image](https://github.com/user-attachments/assets/13955fef-b7c4-4fc2-8207-3ca1dd65f0d2)

![image](https://github.com/user-attachments/assets/1300a20c-d6ee-4bed-97fc-e2bf0a1012df)

![image](https://github.com/user-attachments/assets/92265b83-f81b-44a9-ab69-1102eb964b18)

# 解决方法

在非流式的请求中直接注释掉 `header['X-DashScope-SSE'] = 'enable'` 修改流式请求逻辑,或者在解码 JSON 时预处理,并在 `return response['choices'][0]['message']['content'].strip()` 中增加对 output 的选取:

![image](https://github.com/user-attachments/assets/493dfa49-3601-44c4-9435-d36b4d69c7b2)

# 补充说明

难点在于如果仅是对 `header['X-DashScope-SSE'] = 'enable'` 注释,需要考虑流式下的请求问题,需要重写部分代码,此外,在第 265 行的意外用于调试的 print 是否是忘记删除了?

![image](https://github.com/user-attachments/assets/a764cd39-1421-4507-8b84-951ed9fea314)

刚开始研究此项目和 MindSearch ,部分表述不当见谅。晚些开窍了尝试提交PR (\*^▽^\*)

Contributor guide

No contributing guide indexed for this repository

Assessment

This issue has not been assessed yet.

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.