The Statistical Benefits of Multiple Responses for Learning from Demonstrations
arXiv:2609.33291v1 Announce Type: new Abstract: Many generative systems return multiple candidate responses and are evaluated according to the best one. Recent work shows that, when demonstrations are optimal, pass@$k$ can reduce the sample complexity of learning from demonstrations by a logarithmi…
Read the full story at arXiv stat.ML ↗
Timeline · 1 report
- 2026-09-29 04:00 · arXiv stat.ML
The Statistical Benefits of Multiple Responses for Learning from Demonstrations