OVD: On-policy Verbal Distillation
Open paper
Benchmark
Model
Rank
Results
mathematical-reasoning-on-minerva-math
OVD (Qwen2.5-Math-1.5B, 2 random examples, 800 steps)
–
Accuracy: 18.0
mathematical-reasoning-on-olympiadbench
OVD (Qwen2.5-Math-1.5B, 2 random examples, 800 steps)
–
Accuracy: 31.6
Rank counts only results with a code link.