PrototypeCommon
DaoyuanLi2816/pairjudge
Pairwise LLM judges (A/B/tie): budget-aware multi-turn packing, position-bias correction, pseudo-label distillation. Generalized from the 4th-place (gold) solution to Kaggle LMSYS Chatbot Arena.
- Owner
- DaoyuanLi2816 (personal account)
- Stars
- Stars: 170
- Forks
- Forks: 12
- Language
- Python
- License
- MIT
- Created
- 2024
- Topics
- gold-medalkaggle-competitionllmnlpkaggle-solutionchatbot-arenallm-as-judgelorapreference-learningreward-model
- Tier
- Prototype (100 to 249 ★) What this tier means
More repos by DaoyuanLi2816
All 3 repos by DaoyuanLi2816- DaoyuanLi2816/can-i-finetune-thisStars: 795PythonBeta
- DaoyuanLi2816/mini-verlStars: 307PythonAlpha
Nearby in Python
All Python repos- IamJerryXu/AstraDrawStars: 170PythonPrototype
- sudoingX/local-ai-on-your-gaming-gpuStars: 170PythonPrototype
- BlueRingsLabs/no-shortcutsStars: 170PythonPrototype
- BigBrotherBot/big-brother-botStars: 170PythonPrototype
- amiller/libfreenect-goodiesStars: 170PythonPrototype
- fabpot/sphinx-phpStars: 170PythonPrototype