List your company

Math Reasoning on AIME, MATH500, GSM8K, AMC (test)

64.04Pass@k ScoreJURY-RL
44.716849.733454.7559.7666Apr 28, 2026

Evaluation results

MethodLinks
JURY-RLBackbone=Qwen2.5-7B2026.0464.0448.131.5610.05
GT-RewardBackbone=Qwen2.5-7B2026.0462.4846.6——
JURY-RLBackbone=Qwen3-1.7B-Base2026.0459.4141.574.059.06
GT-RewardBackbone=Qwen3-1.7B-Base2026.0455.3639.25——
LLM-as-a-judgeBackbone=Qwen2.5-7B2026.0453.9944.13——
LLM-as-a-judgeBackbone=Qwen3-1.7B-Base2026.0450.3537.6——
JURY-RLBackbone=Llama-3.2-3B-Instruct2026.0448.4834.13.022.55
LLM-as-a-judgeBackbone=Llama-3.2-3B-Instruct2026.0445.9330.65——
GT-RewardBackbone=Llama-3.2-3B-Instruct2026.0445.4632.19——