Hugging Face API and Unique LLM for Subordinates #221
+387
−380
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
This is my first official Pull Request ever, so I apologize if I'm doing it wrong. I submitted one accidentally to the main branch, as per Eclyp50's request. I then closed it according to Terminally Lazy and submitted it to the Development branch, but in the meantime, I had worked on the Unique LLM per Subordinate problem and so now it is submitting them both. I'm not sure yet how to submit them individually.
The HuggingFace API seems fairly clean and straightforward. Here are my notes on those updates:
Updates to initialize, env, and models to allow for use of HuggingFace API calls. Received Temperature errors if left at 0 when trying to use HuggingFace models, once set to something greater than 0, the errors went away. First time using a HuggingFace API, warning messages were received saying HuggingFace token being saved, on using it a second time in a fresh session no warning messages were received.
The Unique LLMs per Subordinate was much harder for me and I relied heavily on Claude 3.5 Sonnet to re-write a large part of Agent.py to get it working. I hope it's not too bad. I have no background in coding, other than a couple of classes in python from a few years ago. The Unique LLMs required editing agent.py, initialize.py, and call_subordinate.py. One issue I have with this is it currently requires a more "hard-coded" subordinate role, I'd like it to be able to have more freedom in role creation. But, I will keep testing.