feat(eval): add missing minimax backend configuration - #86
feat(eval): add missing minimax backend configuration#86Tanmay Garg (Tanmay9223) wants to merge 1 commit into
Conversation
|
@microsoft-github-policy-service agree |
Add missing configuration setup in scripts/eval_only.py to properly support the minimax_chat backend, which was entirely omitted. Fix the following coverage gaps in eval_only.py: - Add minimax CLI arguments - Include the minimax config mappings in _MAP - Update the backend parsing logic - Call configure_minimax_chat
914f1bf to
f1f21e6
Compare
Tanmay Garg (Tanmay9223)
changed the title
|
Thanks for this Tanmay Garg (@Tanmay9223)! 🙏 Heads up on coordination: the eval-only backend wiring just landed via #85 (qwen + minimax Comparing the two, #87 is effectively a superset of this PR — it carries the same Could you confirm you're OK closing #86? If there's anything in here that isn't also in #87, let me know and we'll make sure it's preserved. Really appreciate the backend coverage work. |
|
Yifan Yang (@Yif-Yang) That makes complete sense. I can confirm that everything from this PR (the Minimax backend wiring and configuration) is fully preserved in PR #87, along with the other updates. I've gone ahead and closed this one in favor of #87 to consolidate the review and avoid any three-way merge conflicts. Thanks for the heads up! |
2 participants
🎯 What: Adds missing configuration setup in
scripts/eval_only.pyto properly support theminimax_chatbackend.💡 Why:
skillopt-evalwas failing to set up external backend calls forminimaxcorrectly because the invocation toconfigure_minimax_chatwas entirely missing, unlike inskillopt-train. This covers the edge cases not addressed by PR #85.✅ Verification: Verified locally that running
scripts/eval_only.pywithbackend: minimaxparses arguments and configures the environment without crashing.✨ Result: Users can now seamlessly evaluate their models with the MiniMax backend using
skillopt-eval.