qbc

应该是sqlite的问题，建议使用wsl

cc @DavdGao @zhijianma

The default dimensions value in `OpenAITextEmbedding` is set to 1024. Try changing it to 4096. This issue might be related to vLLM itself. You can reference this [issue](https://github.com/vllm-project/vllm/issues/20899) for more...

Thanks for catching this typo! We have fixed it in PR #795 Thanks again for helping improve the project!🙂

> > agent(user_agent_msg)执行报错的时候， 外层无法捕获，帮忙看一下，怎么解决呢，谢谢 Hello, sry for the late response. We have fixed in #981.

Could you please provide more detailed information, such as the running code and detailed error messages? And this may potentially be an issue with the `generate_response` function. In our latest...

Error: "Error in block input ['response']" comes from `finish_function_pre_print_hook` in `_react_agent.py`. This function, when processing `finish_function` (`generate_response`), assumes `block["input"]` is a dictionary and calls `block["input"].get("response", "")`. An exception is thrown...

The current main branch (including the aforementioned #910) has not yet been released as a new PyPI version. You can either install AgentScope through GitHub source code, or wait for...

> 我用的deepseek api，在用stream_printing_messages时，想根据last作为信号判断输出是否结束，然后拿到ms的text_content。 但只拿到了“思考部分”，比如：“我来帮你查......”之类的就没了。 后面发现，deepseek貌似在边思考边调用工具--tool_use和tool_result是有值的，虽然last可能为True了。 last为True并不是整个一次reply的结束，而是一次模型调用或者工具执行的结束，如果你在last为True的时候去return的话，就拿不到后续的内容了

> [@Bungo973](https://github.com/Bungo973) 话说 我也用的deepseek api，没复现出来你的问题，msg里只能找到tool use和tool result，无思考部分，用的deepseek reasoner，试试拉最新的代码，或者提供下跑代码的样例（别忘了去掉apikey） 之前deepseek的reasoner模型是不支持工具调用的，如果传入工具，会降级成chat模型；现在reasoner模型支持工具调用，但是[官方文档](https://api-docs.deepseek.com/guides/thinking_mode)里要求把reasoning_content也传入，我们在 pr #998 里做了支持，感谢持续关注～