כתבה
arXiv cs.CL ·
UBTree: ייצור עץ תיקון סדרתי באמצעות דגמי Unigram ו-Bigram לצידור ספקולטיבי
UBTree: Parallel Tree Drafting via Unigram and Bigram Models for Speculative Decoding
UBTree מאיצה את תהליך ההסתברות של הדגמים הלשוניים באמצעות ייצור עצים סדרתי
תקציר מקורי באנגליתarXiv:2609.39972v1 Announce Type: new Abstract: Speculative decoding accelerates language model inference by verifying multiple draft tokens in a single target-model pass. Recent parallel drafters have achieved breakthrough performance in frontier production models, but their effectiveness deteriorates as the entropy of target distributions increases due to insufficient draft diversity. To overcome this bottleneck without sacrificing parallelism, we introduce UBTree, a parallel drafter that couples a Unigram proposer with a Bigram selector to construct drafting Trees. The unigram proposer is trained with the standard cross-entropy objective to generate candidate tokens independently for each position, while a lightweight bigram selector predicts transition scores between adjacent candidate
קרא במקור המקורי
arxiv.org
פתח כתבה מקורית