List your company
ResearchFormerly WizWand

Mathematical Reasoning on AIME 2025 (Accuracy avg@16)

48.8Accuracy (avg@16)Sigma-MoE-Tiny
5.01616.38327.7539.117Dec 1, 2025Dec 3, 2025Dec 6, 2025Dec 9, 2025Dec 12, 2025Dec 15, 2025Dec 18, 2025
Updated 8mo ago

Evaluation results

MethodLinks
Sigma-MoE-TinyArchitecture=MoE, # Activated Params=0.5B, # Total Params=20B2025.1248.8
DeepSeek-R1-Distill-Qwen-7BArchitecture=Dense, # Activated Params=7B, # Total Params=7B2025.1239.2
RF++ B + ReProBackbone=Qwen3-1.7B2025.1239
GRPO + ReProBackbone=Qwen3-1.7B2025.1237.9
PPO + ReProBackbone=Qwen3-1.7B2025.1237.5
Qwen3-1.7BArchitecture=Dense, # Activated Params=1.7B, # Total Params=1.7B2025.1236.8
PPOBackbone=Qwen3-1.7B2025.1236.5
OriginalBackbone=Qwen3-1.7B2025.1236.1
GRPOBackbone=Qwen3-1.7B2025.1234.8
RF++ BBackbone=Qwen3-1.7B2025.1234.4
GRPO + ReProBackbone=Hunyuan-1.8B-Instruct2025.1233.5
PPOBackbone=Hunyuan-1.8B-Instruct2025.1233.3
RF++ BBackbone=Hunyuan-1.8B-Instruct2025.1233.3
OriginalBackbone=Hunyuan-1.8B-Instruct2025.1232.8
RF++ B + ReProBackbone=Hunyuan-1.8B-Instruct2025.1232.8
GRPOBackbone=Hunyuan-1.8B-Instruct2025.1232.7
PPO + ReProBackbone=Hunyuan-1.8B-Instruct2025.1232.1
DeepSeek-R1-Distill-Llama-8BArchitecture=Dense, # Activated Params=8B, # Total Params=8B2025.1227.8
PPO + ReProBackbone=MobileLLM-R1-950M2025.1223.1
PPOBackbone=MobileLLM-R1-950M2025.1222.5
OriginalBackbone=MobileLLM-R1-950M2025.1218.1
Phi-3.5-MoEArchitecture=MoE, # Activated Params=6.6B, # Total Params=42B2025.126.7