arXiv Artificial Intelligence

Where Draft Trees Lose Target Mass: Exit-Guided Speculative Decoding

1 hour ago 09.10.2026 07:00 👁 1 views
Where Draft Trees Lose Target Mass: Exit-Guided Speculative Decoding

Quick summary

arXiv:2610.11750v1 Announce Type: new Abstract: Tree-based speculative decoding verifies multiple draft continuations in one target-model pass, but finite trees built from draft scores face a fundamental draft-target mismatch. We ask whether better exact verification can increase acceptance on a fixed tree and how target feedback can improve the tree itself. Through a target-flow view, we identify a canonical exit law and prove that one plus target coverage sharply bounds the expected output-block length, including the bonus token, of any exact path verifier. All optimal verifiers share the sa

Key takeaways

  • arXiv:2610.11750v1 Announce Type: new Abstract: Tree-based speculative decoding verifies multiple draft continuations in one target-model pass, but finite trees built from draft scores face a fundamental draft-target mismatch.
  • We ask whether better exact verification can increase acceptance on a fixed tree and how target feedback can improve the tree itself.
  • Through a target-flow view, we identify a canonical exit law and prove that one plus target coverage sharply bounds the expected output-block length, including the bonus token, of any exact path verifier.

Why it matters

The significance is not only the legal text but how it changes product design. Decisions around “Where Draft Trees Lose Target Mass: Exit-Guided Speculative Decoding” may reshape data collection, model training, output accountability and market access.

Kaynak sitede devamını oku: arXiv Artificial Intelligence ↗