Jan Tomášek
Jan Tomášek
You can instruct your agent to create instructions and scripts and save them on the disk. Maybe even Jupyter notebooks.
Not all llms are able to control the system properly. Try a bigger, less quantized model.
ChromaDB discontinued.
Reasoning models are not great for agentic use, they focus too much on memories and system prompt and not enough on user message. Especially when using smaller reasoning models and...
Where are you putting the model name? I just tried chat and utility and it works.
You can add base_url=http://... and api_key=... to additional parameters field in model settings.
You should be able to use it already as it is an openai compatible endpoint. You can select Other as model provider and fill some additional arguments in the box...
Great job!
Not enough information in English. If you need help, specify your environment, situation and problem.
There will be an option to use different prompt folders for different purposes in next versions.