Metraipolis › archive › dc_ef7d37b18d0d
When Entropy Is Not Enough: Reclaiming Lost Semantics in LLM Output Length Prediction — arXiv:2608.15592
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2608.15592
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-18T17:25:56.286Z
- SHA-256 of the archived text
70d80b34c5858b0bae65198b0e7a4e20d81da503dda3fc3dc5873985635a3bac- Stable URL of this page
https://metraipolis.com/documento/dc_ef7d37b18d0d
How to cite
When Entropy Is Not Enough: Reclaiming Lost Semantics in LLM Output Length Prediction — arXiv:2608.15592 — arXiv (en), https://arxiv.org/abs/2608.15592 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_ef7d37b18d0d (Metraipolis archive dc_ef7d37b18d0d, sha256 70d80b34c5858b0bae65198b0e7a4e20d81da503dda3fc3dc5873985635a3bac)
BibTeX
@misc{dcef7d37b18d0d,
title = {When Entropy Is Not Enough: Reclaiming Lost Semantics in LLM Output Length Prediction — arXiv:2608.15592},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2608.15592},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_ef7d37b18d0d; sha256 70d80b34c5858b0bae65198b0e7a4e20d81da503dda3fc3dc5873985635a3bac},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.15592. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Efficient LLM serving is often bottlenecked by the need to pad sequences to a fixed maximum length, and this wastes compute and degrades throughput. Predicting output lengths in advance makes it possible to adopt length-aware scheduling, and this reduces the overhead. This advantage is especially pronounced in long-context reasoning and reinforcement learning applications. Existing approaches, such as entropy-guided token pooling, use token-wise entropy as their primary signal, but they tend to ignore differences in semantic content across tokens. So, important tokens are often underweighted, and tokens carrying little information receive disproportionate emphasis. This hurts the reliability of length prediction. We introduce ESTP (Entropy-and-Semantic Token Pooling), a lightweight framework that addresses this issue by combining entropy with attention-based importance scores. These scores are derived directly from the self-attention weights computed during the LLM prefill phase, and this allows ESTP to capture both uncertainty and semantic importance with minimal additional computation. Since the framework reuses prefill activations, it adds almost no extra memory overhead and introduces only minimal latency. On the ForeLen benchmark, ESTP outperforms baseline methods, achieves better prediction accuracy and lower error rates in most scenarios. When integrated with a length-aware scheduler in end-to-end system tests, it further helps improve overall throughput and reduce the padding ratio. Our results offer a practical and effective building block for length-aware LLM serving systems. Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-08-18 in cs.AI). Categorie arXiv: Artificial Intelligence (cs.AI). Pagina del lavoro: https://arxiv.org/abs/2608.15592 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito). Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina. Letto il 2026-10-18T16:29:13.814Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-08-18?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).