
























Abstract:A plethora of protein language models have been released in recent years. Yet comparatively little work has addressed how to best sample from them to optimize desired biological properties. We fill this gap by proposing a flexible, effective sampling method for masked language models (MLMs), and by systematically evaluating models and methods both in silico and in vitro on actual antibody therapeutics campaigns. Firstly, we propose sampling with stochastic beam search, exploiting the fact that MLMs are remarkably efficient at evaluating the pseudo-perplexity of the entire 1-edit neighborhood of a sequence. Reframing generation in terms of entire-sequence evaluation enables flexible guidance with multiple optimization objectives. Secondly, we report results from our extensive in vitro head-to-head evaluation for the antibody engineering setting. This reveals that choice of sampling method is at least as impactful as the model used, motivating future research into this under-explored area.
| Comments: | Accepted into the GEM Workshop, ICLR 2026 |
| Subjects: | Machine Learning (cs.LG); Quantitative Methods (q-bio.QM) |
| Cite as: | arXiv:2603.10302 [cs.LG] |
| (or arXiv:2603.10302v2 [cs.LG] for this version) | |
| https://doi.org/10.48550/arXiv.2603.10302 arXiv-issued DOI via DataCite |
From: Calvin McCarter [view email]
[v1]
Wed, 11 Mar 2026 00:54:06 UTC (5,760 KB)
[v2]
Thu, 7 May 2026 17:36:26 UTC (6,824 KB)
此内容由惯性聚合(RSS阅读器)自动聚合整理,仅供阅读参考。 原文来自 — 版权归原作者所有。