List your company

Arithmetic Reasoning on Game of 24

85.3PerformanceReSCALE
9.06828.85948.6568.441Oct 22, 2025Nov 16, 2025Dec 11, 2025Jan 5, 2026Jan 30, 2026Feb 24, 2026Mar 22, 2026

Evaluation results

MethodLinks
ReSCALEBudget=Large, Tokens=4K–6K, Backbone=Llama2-7B2026.0385.385.9
AlphaZeroBudget=Medium, Tokens=2K–4K, Backbone=Llama2-7B2026.0384.386.2
ReSCALEBudget=Medium, Tokens=2K–4K, Backbone=Llama2-7B2026.0383.485.9
AlphaZeroBudget=Large, Tokens=4K–6K, Backbone=Llama2-7B2026.0382.984.8
AlphaZeroBudget=Small, Tokens=0.2K–2K, Backbone=Llama2-7B2026.0374.486.7
ReSCALEBudget=Small, Tokens=0.2K–2K, Backbone=Llama2-7B2026.0371.681.2
LDDM-GArchitecture=LDDM-G (Ours), Params=85M2025.1063—
Best-of-NBudget=N = 32, Tokens=2.5K, Backbone=Llama2-7B2026.0354.1—
MGDMArchitecture=MGDM, Params=85M2025.1047—
LDDM-GArchitecture=LDDM-G (Ours), Params=6M2025.1028—
MGDMArchitecture=MGDM, Params=6M2025.1012—