Pairwise classification on PMR-Synth Easy
95AccuracyMedGemma-27b
Evaluation Results
| Method | Links | |
|---|---|---|
| MedGemma-27bSize=27B2026.01 | 95 | |
| MedGemma-27BSFTMethod=UrgentSFT2026.01 | 93 | |
| Qwen3-32BSFTMethod=UrgentSFT2026.01 | 92 | |
| Reward-8BurgentMethod=UrgentReward2026.01 | 91 | |
| Reward-8BbaseType=Base2026.01 | 84 | |
| Qwen3-8BSFTMethod=UrgentSFT2026.01 | 83 | |
| Qwen3-4BSFTMethod=UrgentSFT2026.01 | 81 | |
| Qwen3-8BSize=8B2026.01 | 80 | |
| Reward-4BurgentMethod=UrgentReward2026.01 | 80 | |
| GPT-OSSSize=120B2026.01 | 79 | |
| Qwen3-32BSize=32B2026.01 | 77 | |
| Qwen3-4BSize=4B2026.01 | 76 | |
| Reward-4BbaseType=Base2026.01 | 75 | |
| Qwen3-32B-RSize=32B, Reasoning=True2026.01 | 68 |