List your company

Medical Question Answering on HealthBench Medicine N=5,000 (overall)

26.1Rubric ScoreRVPO (k=0.5→2.0 / 1.0→2.0)
-1.0446.00313.0520.097May 7, 2026

Evaluation results

MethodLinks
RVPO (k=0.5→2.0 / 1.0→2.0)Size=14B, Checkpoint=Best Ckpt, k=0.5→2.0 / 1.0→2.02026.0526.1
GRPO (Explicit)Size=14B, Checkpoint=Best Ckpt2026.0523.6
RVPO (k=0.5→2.0 / 1.0→2.0)Size=14B, Checkpoint=Final (300), k=0.5→2.0 / 1.0→2.02026.0523.6
GRPO (Implicit)Size=14B, Checkpoint=Best Ckpt2026.0523.4
GRPO (Implicit)Size=14B, Checkpoint=Final (300)2026.0523.4
RVPO (k=0.5→2.0 / 1.0→2.0)Size=7B, Checkpoint=Best Ckpt, k=0.5→2.0 / 1.0→2.02026.0523
RVPO-explicit (β=1.0/0.5)Size=7B, Checkpoint=Best Ckpt, β=1.0/0.52026.0522.7
RVPO-min (k=∞)Size=14B, Checkpoint=Best Ckpt, k=∞2026.0522.5
GRPO (Explicit)Size=7B, Checkpoint=Best Ckpt2026.0522.1
GDPOSize=14B, Checkpoint=Best Ckpt2026.0521.5
RVPO (k=0.5→2.0 / 1.0→2.0)Size=7B, Checkpoint=Final (300), k=0.5→2.0 / 1.0→2.02026.0520.4
GDPOSize=7B, Checkpoint=Best Ckpt2026.0519.8
GRPO (Implicit)Size=7B, Checkpoint=Best Ckpt2026.0519.3
GRPO (Implicit)Size=7B, Checkpoint=Final (300)2026.0519.3
GDPOSize=3B, Checkpoint=Best Ckpt2026.0519.2
RVPO-min (k=∞)Size=7B, Checkpoint=Best Ckpt, k=∞2026.0519.1
GRPO (Explicit)Size=3B, Checkpoint=Best Ckpt2026.0519
RVPO-min (k=∞)Size=14B, Checkpoint=Final (300), k=∞2026.0519
RVPO (k=0.5→2.0 / 1.0→2.0)Size=3B, Checkpoint=Best Ckpt, k=0.5→2.0 / 1.0→2.02026.0518.9
RVPO-explicit (β=1.0/0.5)Size=14B, Checkpoint=Best Ckpt, β=1.0/0.52026.0518.8
RVPO-explicit (β=1.0/0.5)Size=3B, Checkpoint=Best Ckpt, β=1.0/0.52026.0518.4
RVPO-min (k=∞)Size=3B, Checkpoint=Best Ckpt, k=∞2026.0518.1
RVPO-explicit (β=1.0/0.5)Size=7B, Checkpoint=Final (300), β=1.0/0.52026.0517.8
RVPO-explicit (β=1.0/0.5)Size=14B, Checkpoint=Final (300), β=1.0/0.52026.0516.3
GRPO (Implicit)Size=3B, Checkpoint=Best Ckpt2026.0515.4
RVPO (k=0.5→2.0 / 1.0→2.0)Size=3B, Checkpoint=Final (300), k=0.5→2.0 / 1.0→2.02026.0514.7
RVPO-min (k=∞)Size=3B, Checkpoint=Final (300), k=∞2026.0512.4
GDPOSize=3B, Checkpoint=Final (300)2026.0511.7
GRPO (Explicit)Size=7B, Checkpoint=Final (300)2026.0510.2
GRPO (Implicit)Size=3B, Checkpoint=Final (300)2026.057.2
GRPO (Explicit)Size=3B, Checkpoint=Final (300)2026.055.3
GDPOSize=7B, Checkpoint=Final (300)2026.052.6
RVPO-explicit (β=1.0/0.5)Size=3B, Checkpoint=Final (300), β=1.0/0.52026.051.1
RVPO-min (k=∞)Size=7B, Checkpoint=Final (300), k=∞2026.050
GRPO (Explicit)Size=14B, Checkpoint=Final (300)2026.050
GDPOSize=14B, Checkpoint=Final (300)2026.050