ed-donner
Allow multiple evaluation criteria to be used by the LLM judge.
The Eval section of the notebook has been updated to enable multiple evaluation criteria.
sample:
Effectiveness in resolving the conflict Rank 1: gemini-2.0-flash Rank 2: gpt-4o-mini Rank 3: llama3
Clarity of argument Rank 1: gemini-2.0-flash Rank 2: gpt-4o-mini Rank 3: llama3
Creativity of solution Rank 1: llama3 Rank 2: gemini-2.0-flash Rank 3: gpt-4o-mini
TODO: Evaluation criteria could be LLM-generated rather than a fixed set
hey - thanks for this great contribution @esteinholtz-cloudera - it looks like you might have saved the notebook with Outputs there, it's about 1,700 lines of code. Would you be able to resubmit with the outputs cleared? Thanks so much Ed
Hmm @esteinholtz-cloudera and it seems that there are some updates to folders outside community_contributions - was that intentional? Would you be able to move all updates within community_contributions? Thanks so much Ed
Any other files committed outside community contributions was unintentional.
I have tried a couple of remedies, including creating a new branch, but for some reason, those other changes are persisted in that branch as well....(?). I swear I only added the community contributions this time ... 😢 .
I added git lfs along the way - that might have been the reason it gets screwed. This is above my skill level in git...
Can you include only the community_contributions file in the merge? It is all there.
Ahh unfortunately @esteinholtz-cloudera I don't believe GitHub allows me to make a partial merge.. https://chatgpt.com/share/687afbae-cc80-8012-a2b3-02658dc8b026