List your company
ResearchFormerly WizWand

Mathematical Reasoning on AIME 24 (pass@1, pass@16)

26Pass@1EVOL-RL
9.3613.681822.32Sep 18, 2025
Updated 8mo ago

Evaluation results

MethodLinks
EVOL-RLBase Model Size=8B, Training Dataset=MATH-TRAIN2025.092651.7
EVOL-RLBase Model Size=8B, Training Dataset=AIME242025.0925.438.1
EVOL-RLBase Model Size=8B, Training Dataset=MATH-5002025.0924.149.5
EVOL-RLBase Model Size=4B, Training Dataset=MATH-TRAIN2025.0920.747.6
EVOL-RLBase Model Size=4B, Training Dataset=AIME242025.0920.640.9
TTRLBase Model Size=8B, Training Dataset=AIME242025.092020
EVOL-RLBase Model Size=4B, Training Dataset=MATH-5002025.091943.2
TTRLBase Model Size=8B, Training Dataset=MATH-5002025.0917.740.1
TTRLBase Model Size=4B, Training Dataset=AIME242025.0916.716.7
TTRLBase Model Size=8B, Training Dataset=MATH-TRAIN2025.0916.737.6
TTRLBase Model Size=4B, Training Dataset=MATH-TRAIN2025.0912.123.2
Qwen3-8B-BaseBase Model Size=8B, Training Dataset=-2025.091239.4
Qwen3-4B-BaseBase Model Size=4B, Training Dataset=-2025.091032.4
TTRLBase Model Size=4B, Training Dataset=MATH-5002025.091028