Closed antoninoLorenzo closed 4 months ago
Sweep is currently reviewing your pr...
Hey @antoninoLorenzo looks good to me! At some point we need to implement agent regression tests (not sure how), but for now could you just try with anthropic and openai to be sure?
Otherwise looking solid
@killind-dev I can't test anthropic and openai models given that I do not have a subscription; about regression test what approach are you thinking about?
Sounds good, I'll test it and merge if we are all good. Regarding regression tests I'll open another issue and comment it here
40 @antoninoLorenzo heres the issue for regression testing
Got it, once the Ollama support is completed (at least bare minimum) I can work on the regression tests for it; I can structure it in a way that Is easy to replicate with other models.
However other then that if there aren't any issues with the pull request I will wait the merge so I can update my fork.
I can test the pr for you. If everything is fine I'll merge.
I can test the pr for you. If everything is fine I'll merge.
Great, however i didn't really touched openai and anthropic stuff.
ok then ill merge it!
@antoninoLorenzo yo whats your twitter? would love to should you out when we release this pr
@antoninoLorenzo yo whats your twitter? would love to should you out when we release this pr
Honestly I never made and account 😅 that's not used much Italy
Ok sounds good
@akiradev0x however yesterday I started integrating Ollama with the PlanningAgent
class.
I took as reference the version in agent.py
however the prompts used were different; what Is exactly the difference between Planning and Task agent?
I implemented the system prompt for
ollama/phi3
model and added a test to ensure correct formatting (passed); next I will implement user prompt for phi3.