Pairwise classification on PMR-Synth (Hard)
63AccuracyMedGemma-27b
Evaluation Results
| Method | Links | |
|---|---|---|
| MedGemma-27bSize=27B2026.01 | 63 | |
| Reward-8BurgentMethod=UrgentReward2026.01 | 62 | |
| Qwen3-32BSFTMethod=UrgentSFT2026.01 | 60 | |
| MedGemma-27BSFTMethod=UrgentSFT2026.01 | 60 | |
| Reward-4BurgentMethod=UrgentReward2026.01 | 56 | |
| Qwen3-8BSize=8B2026.01 | 55 | |
| Reward-8BbaseType=Base2026.01 | 54 | |
| Reward-4BbaseType=Base2026.01 | 53 | |
| Qwen3-8BSFTMethod=UrgentSFT2026.01 | 52 | |
| Qwen3-32BSize=32B2026.01 | 51 | |
| Qwen3-4BSize=4B2026.01 | 50 | |
| GPT-OSSSize=120B2026.01 | 50 | |
| Qwen3-4BSFTMethod=UrgentSFT2026.01 | 50 | |
| Qwen3-32B-RSize=32B, Reasoning=True2026.01 | 38 |