List your company

Mathematical Reasoning on GSM8K (test) (Accuracy, SpeedUp)

76.4AccuracyPBS (Single Token)
67.76870.00972.2574.491Feb 11, 2026

Evaluation results

MethodLinks
PBS (Single Token)Max Length=512, Backbone=DREAM-7B-Base2026.0276.40.56
SOARMax Length=512, Backbone=DREAM-7B-Base2026.0275.71.07
PBS (Single Token)Max Length=256, Backbone=DREAM-7B-Base2026.0275.20.57
PBS (Parallel)Max Length=512, Backbone=DREAM-7B-Base2026.0275.11.01
SOARMax Length=256, Backbone=DREAM-7B-Base2026.0274.70.95
Greedy (Adaptive Parallel)Max Length=512, Backbone=DREAM-7B-Base2026.0274.61.83
GreedyMax Length=512, Backbone=DREAM-7B-Base2026.0274.5—
GreedyMax Length=256, Backbone=DREAM-7B-Base2026.0273.7—
PBS (Parallel)Max Length=256, Backbone=DREAM-7B-Base2026.0273.51.03
Greedy (Adaptive Parallel)Max Length=256, Backbone=DREAM-7B-Base2026.0273.22.19
PBS (Single Token)Max Length=512, Backbone=LLADA-8B-Base2026.0272.10.48
PBS (Single Token)Max Length=256, Backbone=LLADA-8B-Base2026.0271.60.47
SOARMax Length=512, Backbone=LLADA-8B-Base2026.0271.51.43
SOARMax Length=256, Backbone=LLADA-8B-Base2026.0271.31.18
Greedy (Adaptive Parallel)Max Length=512, Backbone=LLADA-8B-Base2026.02711.92
GreedyMax Length=512, Backbone=LLADA-8B-Base2026.0270.9—
GreedyMax Length=256, Backbone=LLADA-8B-Base2026.0270.4—
Greedy (Adaptive Parallel)Max Length=256, Backbone=LLADA-8B-Base2026.0270.41.69
PBS (Parallel)Max Length=256, Backbone=LLADA-8B-Base2026.0269.50.98
PBS (Parallel)Max Length=512, Backbone=LLADA-8B-Base2026.0268.10.99