SkillFactory: Self-Distillation For Learning Cognitive Behaviors
Open paper
Benchmark
Model
Rank
Results
mathematical-reasoning-on-math-500
Qwen2.5-7B-Instruct (SkillFactory with 1k SFT + GRPO)
#116
Accuracy: 64.6
arithmetic-reasoning-on-gsm8k
Qwen2.5-1.5B-Instruct (SkillFactory → GRPO)
#194
Accuracy: 68.2
Rank counts only results with a code link.