-
Notifications
You must be signed in to change notification settings - Fork 109
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Generates text but does nothing #16
Comments
You can paste your vLLM error here, and we can check why it failed. Besides, UI-TARS-desktop may use another prompt for the coordinate, you can submit an issue in https://github.com/bytedance/UI-TARS-desktop |
I spent 5 hours trying to solve all this, I have already removed it from both the system and conda and I have no desire to dive into this vllm again. If it is not possible to run this with LM Studio, I will wait for the implementation of ollama, thanks for the help. |
Cloud Deployment definitely works. If you want to use Ollama, please check this issue: #14 . We find Ollama + gguf does not have expected precision compared with the cloud deployment method. |
is your lib free to use I am thinking to explore https://huggingface.co/bytedance-research/UI-TARS-2B-SFT and play around |
We follow apache 2.0 license. |
I had the same issue with DPO looks like the output of the llm is not in the correct format. using SFT version solved the problem for me. |
I tried SFT 2b and 7b, both don't work. Apparently this is a problem in LM studio, since I am not the first one who encountered this. |
I tried vllm but it just doesn't work for me, giving an error message with no solution on the internet. That's why I use LM Studio. It creates the correct prompt but in UI-TARS-desktop it shows that you click on the lower left edge of the screen and nothing happens. I installed Midscene and it also generates the correct prompt but does nothing.
The text was updated successfully, but these errors were encountered: