[Bug Report]: The context is inconsistent.
Version
1.0.0
Model Provider
Other (please specify in description)
Search Provider
Travily
Problem Description
I use the Deepeseek V3 model, and I always find it difficult to have it retrieve literature for me, as it terminates before completing the task. Moreover, the context lacks coherence, such as in the following images. If I switch to R1, it becomes even harder to control, as it basically does not follow my instructions.
Error Logs
No response
Is it possible to add some parameters for setting temperature? As well as the number of context, you can refer to the parameter list settings of Cherry Studio.
请问你是怎么设置让它能检索网页的?我的设置你可以看上一个问题。谢谢
谢谢你对本项目的关注和反馈,目前除了 Claude 其他的模型(DeepSeek、OpenAI)都是 Experimental 状态,这些问题也是已知的(见 https://github.com/bytedance/UI-TARS-desktop/discussions/377 ),请关注我们的发布更新(Github Releases / Twitter / Discord)对这些问题的关注
BTW, I'll appreciate if you take the time to translate the issue to English so anybody can answer, you will also benefit from this :) Next time I suggest you open an issue by English so anybody can answer, you will also benefit from this :)
谢谢你对本项目的关注和反馈,目前除了 Claude 其他的模型(DeepSeek、OpenAI)都是 Experimental 状态,这些问题也是已知的(见 #377 ),请关注我们的发布更新(Github Releases / Twitter / Discord)对这些问题的关注,详
BTW, I'll appreciate if you take the time to translate the issue to English so anybody can answer, you will also benefit from this :) Next time I suggest you open an issue by English so anybody can answer, you will also benefit from this :)
Thank you! I have understood. I look forward to seeing your project thrive and surpass manus. 😊