Teaching LLMs to Speak Spectroscopy

TL;DR

This paper shows that a large language model can be efficiently adapted to predict galaxy redshifts from spectroscopic data using minimal parameter tuning, while maintaining its linguistic abilities, thus bridging NLP and scientific data analysis.

Contribution

Demonstrates that LLaMA-3.1-8B can be repurposed for spectroscopic analysis with minimal adaptation using LoRA, enabling scientific tasks without extensive retraining.

Findings

Achieved a mean absolute error of 0.04 in redshift prediction.

Retained over 85% performance on AstroBench and 89% on general QA tasks.

Used only 16 GPU-hours and 0.04% of parameters for adaptation.

Abstract

Pre-trained Large Language Models (LLMs) have revolutionized text processing, yet adapting Transformer-based neural networks to non-textual scientific modalities typically requires specialized architectures and extensive computational resources. We demonstrate that LLaMA-3.1-8B can be efficiently repurposed to predict galaxy redshifts from spectroscopic data through Low-Rank Adaptation (LoRA), achieving competitive performance while preserving its linguistic capabilities. Using only 16 GPU-hours and adapting 0.04% of model parameters, our approach achieves a mean absolute error of 0.04 in redshift prediction while retaining over 85% of performance on AstroBench and 89% on general QA tasks from eval-harness. This minimal-effort adaptation--requiring only simple standard fine-tuning APIs--lowers barriers to entry for domain scientists and enables integrated agentic workflows where a single model handles both spectroscopic data for quantitative analysis and natural language for reasoning.