List your company

Mathematical Reasoning on OlymMath (avg@16 accuracy, pass@8)

25.8Avg@16 AccuracyTEMPO
10.51214.48118.4522.419Apr 21, 2026

Evaluation results

MethodLinks
TEMPOBackbone=Qwen3-14B-Base2026.0425.850.2
Zero-RL (PPO)Backbone=Qwen3-14B-Base2026.0424.251.6
TEMPOBackbone=OLMO3-7B-Base2026.0423.551.6
Zero-RL (PPO)Backbone=OLMO3-7B-Base2026.0418.943.3
TTRLBackbone=OLMO3-7B-Base2026.0418.933
EMPOBackbone=OLMO3-7B-Base2026.0418.732.9
TEMPOBackbone=Qwen3-8B-Base2026.0418.743.3
TTRLBackbone=Qwen3-14B-Base2026.0418.331.7
EMPOBackbone=Qwen3-14B-Base2026.0417.731.9
MiMo-Zero-RL-7BBackbone=MiMo-Zero-RL-7B2026.0416.736.1
Zero-RL (PPO)Backbone=Qwen3-8B-Base2026.041539.9
OLMO3.1-Zero-RL-7BBackbone=OLMO3.1-Zero-RL-7B2026.0414.342.3
EMPOBackbone=Qwen3-8B-Base2026.0413.126.7
TTRLBackbone=Qwen3-8B-Base2026.0411.425.3
Oat-Zero-7BBackbone=Oat-Zero-7B2026.0411.122.6