Skip to content

Conversation

whyuds
Copy link

@whyuds whyuds commented Jul 25, 2025

豆包、阿里新模型均可以选择深度思考、非深度思考,但默认为深度思考,导致每轮对话性能较差。他们在extra_body提供了参数可以进行控制,设置disable后可大大提高对话性能。

默认情况,启动深度思考

250725 13:57:05[0.7.2-SiFuDoEdnocaCh][core.handle.textHandle]-INFO-收到listen消息:{"type":"listen","mode":"manual","state":"detect","text":"你好,介绍下你自己"}
250725 13:57:05[0.7.2-SiFuDoEdnocaCh][core.connection]-INFO-大模型收到用户消息: 你好,介绍下你自己
250725 13:57:07[0.7.2-SiFuDoEdnocaCh][core.handle.sendAudioHandle]-INFO-发送音频消息: SentenceType.MIDDLE, 哈喽
250725 13:57:07[0.7.2-SiFuDoEdnocaCh][core.handle.sendAudioHandle]-INFO-发送第一段语音: 哈喽
250725 13:57:08[0.7.2-SiFuDoEdnocaCh][core.handle.sendAudioHandle]-INFO-发送音频消息: SentenceType.MIDDLE, 我是来自中国台湾省的小智啦~讲话超机车的00后女生

image

配置关闭深度思考

250725 13:58:08[0.7.2-SiFuDoEdnocaCh][core.handle.textHandle]-INFO-收到listen消息:{"type":"listen","mode":"manual","state":"detect","text":"你好,介绍下你自己"}
250725 13:58:08[0.7.2-SiFuDoEdnocaCh][core.connection]-INFO-大模型收到用户消息: 你好,介绍下你自己
250725 13:58:29[0.7.2-SiFuDoEdnocaCh][core.handle.sendAudioHandle]-INFO-发送音频消息: SentenceType.MIDDLE, 哈喽~
250725 13:58:29[0.7.2-SiFuDoEdnocaCh][core.handle.sendAudioHandle]-INFO-发送第一段语音: 哈喽~
250725 13:58:30[0.7.2-SiFuDoEdnocaCh][core.handle.sendAudioHandle]-INFO-发送音频消息: SentenceType.MIDDLE, 我是小智啦

@whyuds whyuds force-pushed the main branch 3 times, most recently from 75c49db to c4e7839 Compare July 28, 2025 09:26
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant