Medical Recommendation
71.74Diagnosis AccuracyNorm-ITPO
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| Norm-ITPOAdvantage Estimator=GRPO, Credit Assignment=Norm-ITPO2026.03 | 71.74 | 0.62 | 68.66 | |
| Norm-ITPOAdvantage Estimator=RLOO, Credit Assignment=Norm-ITPO2026.03 | 69.24 | 0.62 | 66.14 | |
| ITPOAdvantage Estimator=RLOO, Credit Assignment=ITPO2026.03 | 68.43 | 0.98 | 63.49 | |
| ITPOAdvantage Estimator=GRPO, Credit Assignment=ITPO2026.03 | 67.61 | 0.91 | 63.07 | |
| LLM as JudgeAdvantage Estimator=RLOO, Credit Assignment=LLM as Judge2026.03 | 66.77 | 0.94 | 62.03 | |
| Trajectory-LevelAdvantage Estimator=GRPO, Credit Assignment=Trajectory-Level2026.03 | 66.41 | 0.94 | 61.67 | |
| Trajectory-LevelAdvantage Estimator=RLOO, Credit Assignment=Trajectory-Level2026.03 | 65.12 | 0.78 | 61.22 | |
| Uniform DecomposeAdvantage Estimator=RLOO, Credit Assignment=Uniform Decompose2026.03 | 63.15 | 0.46 | 60.82 | |
| LLM as JudgeAdvantage Estimator=GRPO, Credit Assignment=LLM as Judge2026.03 | 63.12 | 1.12 | 60.56 | |
| VM+Norm-ITPOAdvantage Estimator=PPO, Credit Assignment=VM+Norm-ITPO2026.03 | 62.44 | 0.6 | 59.44 | |
| Uniform DecomposeAdvantage Estimator=GRPO, Credit Assignment=Uniform Decompose2026.03 | 61.95 | 0.66 | 58.64 | |
| PRIMEAdvantage Estimator=RLOO, Credit Assignment=PRIME2026.03 | 61.42 | 1.12 | 55.9 | |
| VM OnlyAdvantage Estimator=PPO, Credit Assignment=VM Only2026.03 | 58.41 | 0.72 | 54.77 | |
| VM+ITPOAdvantage Estimator=PPO, Credit Assignment=VM+ITPO2026.03 | 58.35 | 0.68 | 54.87 | |
| PRIMEAdvantage Estimator=GRPO, Credit Assignment=PRIME2026.03 | 56.36 | 0.65 | 53.12 | |
| VM+PRIMEAdvantage Estimator=PPO, Credit Assignment=VM+PRIME2026.03 | 52.07 | 0.48 | 49.73 | |
| VanillaAdvantage Estimator=N/A, Credit Assignment=Vanilla2026.03 | 25.52 | 0.28 | 24.08 |