MMAI Gym for Science: Training Liquid Foundation Models for Drug Discovery

TL;DR

This paper introduces MMAI Gym for Science, a specialized framework for training liquid foundation models tailored to drug discovery, demonstrating that purpose-trained smaller models outperform larger general models on key molecular tasks.

Contribution

The paper presents MMAI Gym for Science and trains a Liquid Foundation Model that surpasses larger models in drug discovery benchmarks, emphasizing efficiency and domain-specific training.

Findings

Smaller purpose-trained models outperform larger general models on molecular benchmarks.

The trained Liquid Foundation Model achieves near-specialist performance across multiple drug discovery tasks.

Efficient models can be broadly applicable and surpass larger models in drug discovery applications.

Abstract

General-purpose large language models (LLMs) that rely on in-context learning do not reliably deliver the scientific understanding and performance required for drug discovery tasks. Simply increasing model size or introducing reasoning tokens does not yield significant performance gains. To address this gap, we introduce the MMAI Gym for Science, a one-stop shop molecular data formats and modalities as well as task-specific reasoning, training, and benchmarking recipes designed to teach foundation models the 'language of molecules' in order to solve practical drug discovery problems. We use MMAI Gym to train an efficient Liquid Foundation Model (LFM) for these applications, demonstrating that smaller, purpose-trained foundation models can outperform substantially larger general-purpose or specialist models on molecular benchmarks. Across essential drug discovery tasks - including molecular optimization, ADMET property prediction, retrosynthesis, drug-target activity prediction, and functional group reasoning - the resulting model achieves near specialist-level performance and, in the majority of settings, surpasses larger models, while remaining more efficient and broadly applicable in the domain.