INDUS: Effective and Efficient Language Models for Scientific Applications

TL;DR

INDUS introduces a suite of specialized language models trained on scientific data across multiple disciplines, outperforming general and domain-specific models on new benchmarks and demonstrating practical industrial applications.

Contribution

The paper presents new scientific domain-specific language models and benchmark datasets, tailored for Earth science, biology, physics, and related fields, with improved performance over existing models.

Findings

Models outperform RoBERTa and SCIBERT on new scientific benchmarks.

Effective in retrieval and content tagging applications.

Smaller models maintain high performance with resource efficiency.

Abstract

Large language models (LLMs) trained on general domain corpora showed remarkable results on natural language processing (NLP) tasks. However, previous research demonstrated LLMs trained using domain-focused corpora perform better on specialized tasks. Inspired by this insight, we developed INDUS, a comprehensive suite of LLMs tailored for the closely-related domains of Earth science, biology, physics, heliophysics, planetary sciences and astrophysics, and trained using curated scientific corpora drawn from diverse data sources. The suite of models include: (1) an encoder model trained using domain-specific vocabulary and corpora to address NLP tasks, (2) a contrastive-learning based text embedding model trained using a diverse set of datasets to address information retrieval tasks and (3) smaller versions of these models created using knowledge distillation for applications which have latency or resource constraints. We also created three new scientific benchmark datasets, CLIMATE-CHANGE NER (entity-recognition), NASA-QA (extractive QA) and NASA-IR (IR) to accelerate research in these multi-disciplinary fields. We show that our models outperform both general-purpose (RoBERTa) and domain-specific (SCIBERT) encoders on these new tasks as well as existing tasks in the domains of interest. Furthermore, we demonstrate the use of these models in two industrial settings -- as a retrieval model for large-scale vector search applications and in automatic content tagging systems.