Weaver: Shrinking the Generation-Verification Gap with Weak Verifiers

Jon Saad-Falcon; E. Kelly Buchanan; Mayee F. Chen; Tzu-Heng Huang; Brendan McLaughlin; Tanvir Bhathal; Shang Zhu; Ben Athiwaratkun; Frederic Sala; Scott Linderman; Azalia Mirhoseini; Christopher Ré

doi:10.48550/arXiv.2506.18203

Weaver: Shrinking the Generation-Verification Gap with Weak Verifiers

Jon Saad-Falcon* Stanford

E. Kelly Buchanan* Stanford University

Mayee F. Chen* Stanford University

Tzu-Heng Huang University of Wisconsin-Madison

Brendan McLaughlin Stanford

Tanvir Bhathal Stanford

Shang Zhu Together AI

Ben Athiwaratkun Together AI

Frederic Sala University of Wisconsin-Madison

Scott Linderman Stanford University

Azalia Mirhoseini Stanford University

Christopher Ré Stanford University

Preprint, 2025

DOI: 10.48550/arXiv.2506.18203

Weaver boosts language model performance by intelligently combining weak verifiers using weak supervision, achieving near-oracle accuracy with drastically reduced compute.

Materials

Bibtex

@misc{2025weaver,
  title = {Weaver: Shrinking the Generation-Verification Gap with Weak Verifiers},
  author = {Jon Saad-Falcon and E. Kelly Buchanan and Mayee F. Chen and Tzu-Heng Huang and Brendan McLaughlin and Tanvir Bhathal and Shang Zhu and Ben Athiwaratkun and Frederic Sala and Scott Linderman and Azalia Mirhoseini and Christopher Ré},
  booktitle = {Preprint},
  year = {2025},
  doi = {10.48550/arXiv.2506.18203},
  url = {https://arxiv.org/abs/2506.18203},
}

Scaling Intelligence Lab

Weaver: Shrinking the Generation-Verification Gap with Weak Verifiers

Materials

Bibtex