PREreviews de Training Large Language Models for Small-Molecule Design with Synthetic Task Scaling

Training Large Language Models for Small-Molecule Design with Synthetic Task Scaling

par Frank Hu, Shriram Chennakesavalu, Zichen Wang, Patricia Suriana, Bodhi Vani, Kirill Shmilovich, Kangway Chuang et Colin Grambow

Publié: 4 septembre 2026
Serveur de preprints: arXiv
DOI: 10.48550/arxiv.2609.04735

Résumé

Designing viable drug candidates requires searching a combinatorially large and rugged chemical space for molecules that satisfy multiple, often competing, objectives. Large language models (LLMs) provide a useful generative prior for this problem because of their representational capacity, reasoning ability, and flexibility when incorporating information from the external environment. While reinforcement learning from verifiable rewards (RLVR) can be used to improve the capabilities of LLMs, many chemically relevant scoring functions require hours or even days per evaluation, making them prohibitively expensive to use directly during online training. Here, we investigate whether LLMs can learn molecular design strategies from cheaper synthetic tasks that generalize to expensive molecular lead optimization settings. We find that curriculum-based training recipes that gradually incorporate more challenging synthetic design tasks enable strong performance that surpasses that of much larger frontier models on structure-based lead optimization. Our results suggest that scaling post-training using synthetic tasks is an effective strategy for adapting LLMs to high-cost experimental scenarios that are too expensive to directly train on.

Lire le preprint

2 PREreviews

Rédigez un PREreview Demander une PREreview

PREreview par Sneh Pankajbhai Vora

Rédigé par Sneh Pankajbhai Vora

Review of "Training Large Language Models for Small-Molecule Design with Synthetic Task Scaling"

Summary
…
Lire la PREreview de Sneh Pankajbhai Vora
PREreview par Sneh Pankajbhai Vora

Rédigé par Sneh Pankajbhai Vora

Does the introduction explain the objective of the research presented in the preprint?

Yes

The introduction explains the objective by setting up a gap and then posing a question. It moves in 4 steps: 1. Drug discovery means searching a huge chemical space for molecules meeting competing…

Lire la PREreview de Sneh Pankajbhai Vora

PREreviews de Training Large Language Models for Small-Molecule Design with Synthetic Task Scaling

2 PREreviews

Review of "Training Large Language Models for Small-Molecule Design with Synthetic Task Scaling"