Robert Brennan
Robert Brennan
@stratte89 is that error message coming from the OpenDevin backend? Or is it coming from the LLM? We shouldn't be using much more memory, but 644MiB is not a lot....
That's an oobabooga bug then 😄 nothing we can do about it here
It looks like the backend thinks your runtime is running on port 80, which is suspicious... ``` HTTPConnectionPool(host='localhost', port=80) ```
The runtime is clearly on a normal 3xxxx port ``` Uvicorn running on http://0.0.0.0:38130 ``` very confusing because we're pretty directly setting the same port in `self.api_url` in `openhands/runtime/impl/docker/docker_runtime.py`. I'd...
Thanks for doing this! There's one bit of confusion that I think would be helpful to clear up: the current agents aren't really specialists in any way. They're meant to...
Yeah agree--the agent never really asks to search long-term memory. In the early days it would, but I haven't seen it in a while. Feel free to open an issue...
We're using LiteLLM--any chance of adding LocalAI as a provider? https://docs.litellm.ai/docs/providers
From what I'm reading this will work seamlessly 🎉 https://docs.litellm.ai/docs/providers/custom_openai_proxy LMK if you think otherwise
I think we can close--but if you get a chance to try it let me know if you find any issues!
@Jn-Huang the codeact agent is pretty bare-bones. Any reason you're using that instead of the default?