List your company

Math on AIME-Extend

52.67AccuracyAdaRAS
20.076428.53823745.4618Jan 27, 2026

Evaluation results

MethodLinks
AdaRASCategory=Steering2026.0152.67
ProbingCategory=Steering2026.0148
CoTPrompting=Vanilla CoT, Base Model=Qwen3-1.7B2026.0147.33
OpenReasoning-Nemotron-1.5BCategory=Post-training2026.0147.33
OpenThinker-3-1.5BCategory=Post-training2026.0142.67
DeepSeek-R1-Distill-Qwen-1.5BCategory=Post-training2026.0121.33