List your company
ResearchFormerly WizWand

Mathematical Reasoning on Average (AIME24, AIME25, AMC23, MATH500)

55.4Pass@1AER
5.136818.185931.23544.2841Oct 13, 2025Nov 25, 2025Jan 7, 2026Feb 19, 2026Apr 3, 2026May 16, 2026Jun 29, 2026
Updated 4mo ago

Evaluation results

MethodLinks
AERModel Backbone=Qwen3-8B-Base2025.1055.476
Clip-CovModel Backbone=Qwen3-8B-Base2025.105575.2
Clip-HigherModel Backbone=Qwen3-8B-Base2025.1051.374.5
AERModel Backbone=Qwen3-4B-Base2025.1051.172.5
Clip-CovModel Backbone=Qwen3-4B-Base2025.1050.170.6
KL-CovModel Backbone=Qwen3-8B-Base2025.1048.472.5
Ent-AdvModel Backbone=Qwen3-8B-Base2025.1046.363.6
GRPOModel Backbone=Qwen3-8B-Base2025.104666
KL-CovModel Backbone=Qwen3-4B-Base2025.1045.668.6
Clip-HigherModel Backbone=Qwen3-4B-Base2025.1045.363.9
GRPOModel Backbone=Qwen3-4B-Base2025.1043.964
Ent-AdvModel Backbone=Qwen3-4B-Base2025.1040.762.6
BaseModel Backbone=Qwen3-8B-Base2025.1033.166.6
BaseModel Backbone=Qwen3-4B-Base2025.1024.861.5
GRPO + LatentReviseModel=Qwen3-8B-Base, Training Condition=DAPO-14k-Hard2026.0622.5865.14
SFT + LatentReviseModel=Qwen3-8B-Base, Training Condition=DAPO-14k-Hard2026.0621.2664.46
GRPO(n=16)Model=Qwen3-8B-Base, Training Condition=DAPO-14k-Hard2026.0621.0158.9
GRPOModel=Qwen3-8B-Base, Training Condition=DAPO-14k-Hard2026.0619.4362.75
BaseModel=Qwen3-8B-Base, Training Condition=DAPO-14k-Hard2026.0618.2464.36
GRPO + LatentReviseModel=Qwen2.5-1.5B-Inst, Training Condition=DAPO-14k-Hard2026.0618.1353.51
GRPO(n=16)Model=Qwen2.5-1.5B-Inst, Training Condition=DAPO-14k-Hard2026.0617.7150.69
SFT + LatentReviseModel=Qwen2.5-1.5B-Inst, Training Condition=DAPO-14k-Hard2026.0617.6250.5
OPSDModel=Qwen3-8B-Base, Training Condition=DAPO-14k-Hard2026.0616.0960.36
BaseModel=Qwen2.5-1.5B-Inst, Training Condition=DAPO-14k-Hard2026.0616.0546.67
GRPOModel=Qwen2.5-1.5B-Inst, Training Condition=DAPO-14k-Hard2026.0614.6544.84
OPSDModel=Qwen2.5-1.5B-Inst, Training Condition=DAPO-14k-Hard2026.067.0740.42