{"id":"arxiv:2610.28355","kind":"preprint","title":"Beyond the Point Mass: Acceptance-Derived Weights and Distributional Losses for Speculative Decoding","authors":["Tianyu Wu","Zhuohan Wang","Jonathan Geuter","Yu Yao","Weifan Jiang","Chengxi Zhang","Ju Li","Minlan Yu"],"author_count":8,"year":2026,"published":"2026-10-20","venue":null,"cited_by":null,"doi":null,"arxiv_id":"2610.28355","nct_id":null,"source_url":"https://arxiv.org/abs/2610.28355","archive_url":"https://space-kitty.com/papers/arxiv:2610.28355","updated":"2026-10-20","pmid":null,"pmcid":null,"openalex_id":null,"pdf_url":"https://arxiv.org/pdf/2610.28355","abstract":"In speculative decoding, a lightweight draft model proposes tokens that the target model verifies in parallel, speeding up generation without changing the target distribution. D-PACE (Wu et al., 2026) trains such drafters with a loss that weights each position by its contribution to the expected accepted length. Both its weights and its cross-entropy (CE) loss use only the draft's probability of the recorded target token, treating the target as a point mass. Under rejection sampling, however, a sampled draft token is accepted with probability equal to the overlap of the draft and target distributions, one minus their total variation (TV). We derive D-PAL (Dynamic Position-wise Acceptance Loss) from the exact expected accepted length, using this overlap as the acceptance signal. This yields overlap-based position weights with log acceptance (LA) as the per-position loss, and the objective reduces to D-PACE for a point-mass target. Since the weights are detached, any per-position loss can replace LA in practice. We show that forward KL divergence and Rényi divergence of order 1/2 match or exceed LA, and that Rényi-1/2 provably bounds the true acceptance rate from above and below. Replacing CE with Rényi-1/2 improves acceptance length under both D-PACE's weights and ours. On six benchmarks with a Qwen3-4B target, D-PAL raises mean acceptance length over D-PACE from 4.50 to 4.57 at T=0 and from 4.20 to 4.28 at T=1, and consistently achieves higher throughput at both temperatures.","abstract_license":"CC0-1.0 (arXiv descriptive metadata)","license":null,"is_open_access":false,"retracted":false,"peer_reviewed":false,"categories":["cs.LG"],"details":null,"provenance":"arxiv","attribution":"Abstract: arXiv descriptive metadata (CC0). Thank you to arXiv for use of its open access interoperability. This service was not reviewed or approved by, nor does it necessarily express or reflect the policies or opinions of, arXiv.","topics":[],"cited_in_dossiers":[]}