List your company
ResearchFormerly WizWand

Mathematical Reasoning on AIME 25 (Accuracy, Average Length)

31.67AccuracyGraph-Based Chain-of-Thought Pruning
16.933220.759124.58528.4109Apr 7, 2026
Updated 6mo ago

Evaluation results

MethodLinks
Graph-Based Chain-of-Thought PruningModel Scale=7B, Backbone=DeepSeek-R1-Distill-Qwen2026.0431.676,977
Light-R1-DS-7BModel Scale=7B, Method Variant=Open-Source R1-Style2026.043112,498
TokenSkipModel Scale=7B, Backbone=DeepSeek-R1-Distill-Qwen2026.0430.8311,259
EfficientReasoningModel Scale=7B, Backbone=DeepSeek-R1-Distill-Qwen2026.0430.839,293
AdaptThinkModel Scale=7B, Backbone=DeepSeek-R1-Distill-Qwen2026.043010,290
BaseModel Scale=7B, Backbone=DeepSeek-R1-Distill-Qwen2026.042912,779
OREAL-7BModel Scale=7B, Method Variant=Open-Source R1-Style2026.0428.7511,309
O1-PrunerModel Scale=7B, Backbone=DeepSeek-R1-Distill-Qwen2026.0427.511,475
AReaL-boba-RL-7BModel Scale=7B, Method Variant=Open-Source R1-Style2026.0425.4212,365
Skywork-OR1-7BModel Scale=7B, Method Variant=Open-Source R1-Style2026.0424.6712,831
Graph-Based Chain-of-Thought PruningModel Scale=1.5B, Backbone=DeepSeek-R1-Distill-Qwen2026.0423.337,739
BaseModel Scale=1.5B, Backbone=DeepSeek-R1-Distill-Qwen2026.0422.9211,797
EfficientReasoningModel Scale=1.5B, Backbone=DeepSeek-R1-Distill-Qwen2026.04209,383
O1-PrunerModel Scale=1.5B, Backbone=DeepSeek-R1-Distill-Qwen2026.0418.7512,953
TokenSkipModel Scale=1.5B, Backbone=DeepSeek-R1-Distill-Qwen2026.0417.9212,630
AdaptThinkModel Scale=1.5B, Backbone=DeepSeek-R1-Distill-Qwen2026.0417.53,854