List your company
ResearchFormerly WizWand

Mathematical Reasoning on AIME25 (Pass@8, Pass@16, Pass@32)

40Pass@8SENT
-1.69.22030.8Dec 4, 2025Dec 14, 2025Dec 25, 2025Jan 5, 2026Jan 16, 2026Jan 27, 2026Feb 7, 2026
Updated 8mo ago

Evaluation results

MethodLinks
SENTBackbone=DeepSeek-R1-Distill-Qwen-1.5B2025.124043.3346.67
w/ CovBackbone=DeepSeek-R1-Distill-Qwen-1.5B2025.1236.6736.6740
GRPOBackbone=DeepSeek-R1-Distill-Qwen-1.5B2025.1233.334043.33
w/ EnBackbone=DeepSeek-R1-Distill-Qwen-1.5B2025.1233.3336.6743.3
w/ AdvBackbone=DeepSeek-R1-Distill-Qwen-1.5B2025.1233.333036.67
w/ MaskBackbone=DeepSeek-R1-Distill-Qwen-1.5B2025.1233.3333.3343.33
w/ ClipBackbone=DeepSeek-R1-Distill-Qwen-1.5B2025.1233.334036.67
w/ High EnBackbone=DeepSeek-R1-Distill-Qwen-1.5B2025.1233.3333.3340
Base modelBackbone=DeepSeek-R1-Distill-Qwen-1.5B2025.1226.6733.3330
SED-SFTBackbone=Qwen2.5-Math-7B-Instruct, Training Phase=RL, Optimization Objective=SED-SFT2026.0218.8——
DFTBackbone=Qwen2.5-Math-7B-Instruct, Training Phase=RL, Optimization Objective=DFT2026.0218.3——
DFTBackbone=Qwen2.5-Math-7B-Instruct, Training Phase=SFT, Optimization Objective=DFT2026.0217.5——
GEMBackbone=Qwen2.5-Math-7B-Instruct, Training Phase=RL, Optimization Objective=GEM2026.0216.7——
SED-SFT w/o maskBackbone=Qwen2.5-Math-7B-Instruct, Training Phase=RL, Optimization Objective=SED-SFT w/o mask2026.0216.7——
CrossEntropyBackbone=Qwen2.5-Math-7B-Instruct, Training Phase=RL, Optimization Objective=CrossEntropy2026.0216.2——
CrossEntropyBackbone=Qwen2.5-Math-7B-Instruct, Training Phase=SFT, Optimization Objective=CrossEntropy2026.0214.2——
SED-SFTBackbone=Qwen2.5-Math-7B-Instruct, Training Phase=SFT, Optimization Objective=SED-SFT2026.0212.1——
SED-SFT w/o maskBackbone=Qwen2.5-Math-7B-Instruct, Training Phase=SFT, Optimization Objective=SED-SFT w/o mask2026.0211.7——
GEMBackbone=Qwen2.5-Math-7B-Instruct, Training Phase=SFT, Optimization Objective=GEM2026.029.2——
CrossEntropyBackbone=Llama-3.2-3B-Instruct, Training Phase=SFT, Optimization Objective=CrossEntropy2026.021.2——
GEMBackbone=Llama-3.2-3B-Instruct, Training Phase=SFT, Optimization Objective=GEM2026.021.2——
SED-SFT w/o maskBackbone=Llama-3.2-3B-Instruct, Training Phase=SFT, Optimization Objective=SED-SFT w/o mask2026.021.2——
SED-SFTBackbone=Llama-3.2-3B-Instruct, Training Phase=SFT, Optimization Objective=SED-SFT2026.021.2——
DFTBackbone=Llama-3.2-3B-Instruct, Training Phase=RL, Optimization Objective=DFT2026.021.2——
GEMBackbone=Llama-3.2-3B-Instruct, Training Phase=RL, Optimization Objective=GEM2026.021.1——
DFTBackbone=Llama-3.2-3B-Instruct, Training Phase=SFT, Optimization Objective=DFT2026.020.8——
SED-SFT w/o maskBackbone=Llama-3.2-3B-Instruct, Training Phase=RL, Optimization Objective=SED-SFT w/o mask2026.020.6——
SED-SFTBackbone=Llama-3.2-3B-Instruct, Training Phase=RL, Optimization Objective=SED-SFT2026.020.4——
CrossEntropyBackbone=Llama-3.2-3B-Instruct, Training Phase=RL, Optimization Objective=CrossEntropy2026.020——