List your company

Mathematical Reasoning on MATH 500 (mean@8)

77.88Mean@8Rollout-GDRO
49.529656.889864.2571.6102Jan 27, 2026

Evaluation results

MethodLinks
Rollout-GDROBackbone=Qwen3-8B-Base2026.0177.88
Prompt-GDROBackbone=Qwen3-8B-Base2026.0176.18
Prompt-GDROBackbone=Qwen3-4B-Base2026.0175.78
Rollout-GDROBackbone=Qwen3-4B-Base2026.0175.2
GRPO (Baseline)Backbone=Qwen3-8B-Base2026.0173.45
GRPO (Baseline)Backbone=Qwen3-4B-Base2026.0172.05
Rollout-GDROBackbone=Qwen3-1.7B-Base2026.0163.98
Prompt-GDROBackbone=Qwen3-1.7B-Base2026.0163.2
GRPO (Baseline)Backbone=Qwen3-1.7B-Base2026.0150.62