List your company

Mathematical Reasoning on AMC (mean@8)

70.94Mean@8 ScorePrompt-GDRO
33.2443.027552.81562.6025Jan 27, 2026

Evaluation results

MethodLinks
Prompt-GDROBackbone=Qwen3-8B-Base2026.0170.94
Rollout-GDROBackbone=Qwen3-4B-Base2026.0167.5
Rollout-GDROBackbone=Qwen3-8B-Base2026.0166.88
GRPO (Baseline)Backbone=Qwen3-8B-Base2026.0166.56
Prompt-GDROBackbone=Qwen3-4B-Base2026.0164.06
GRPO (Baseline)Backbone=Qwen3-4B-Base2026.0160.94
Prompt-GDROBackbone=Qwen3-1.7B-Base2026.0139.38
Rollout-GDROBackbone=Qwen3-1.7B-Base2026.0136.87
GRPO (Baseline)Backbone=Qwen3-1.7B-Base2026.0134.69