List your company

Mathematical Reasoning on MATH (pass@8)

67.75Pass@8Rollout-GDRO
50.059654.652359.24563.8377Jan 27, 2026

Evaluation results

MethodLinks
Rollout-GDROBackbone=Qwen3-8B-Base2026.0167.75
Prompt-GDROBackbone=Qwen3-8B-Base2026.0167.6
Prompt-GDROBackbone=Qwen3-4B-Base2026.0163.7
Rollout-GDROBackbone=Qwen3-4B-Base2026.0162.27
GRPO (Baseline)Backbone=Qwen3-8B-Base2026.0162.04
GRPO (Baseline)Backbone=Qwen3-4B-Base2026.0156.31
Rollout-GDROBackbone=Qwen3-1.7B-Base2026.0156.14
Prompt-GDROBackbone=Qwen3-1.7B-Base2026.0155.68
GRPO (Baseline)Backbone=Qwen3-1.7B-Base2026.0150.74