Engineers at the University of California San Diego have created a new method to make large language models (LLMs) — such as the ones that power chatbots and protein sequencing tools — learn new tasks using significantly less data and computing power.
LLMs are made up of billions of parameters that determine how they process information. Traditional fine-tuning methods adjust all of these parameters, which can be costly and prone to overfitting — when a model memorizes patterns instead of truly understanding them, causing it to perform poorly on new examples.
The new method developed by UC San Diego engineers takes a smarter approach. Instead of retraining an entire model from scratch, it updates only the parts that matter most. As a result, the new method cuts costs and is more flexible and better at generalizing what it learns compared to existing fine-tuning methods.
The researchers showed that their method can fine-tune protein language models — which are used to study and predict the properties of proteins — even when very little training data are available. For example, in predicting whether certain peptides can cross the blood-brain barrier, the new method achieved higher accuracy than conventional methods while using 326 times fewer parameters. In predicting protein thermostability, it matched the performance of full fine-tuning while using 408 times fewer parameters.
"With our method, even small labs and startups without huge budgets, supercomputer-level resources or large datasets can adapt large AI models for their own needs," said Pengtao Xie, a professor in the Department of Electrical and Computer Engineering at the UC San Diego Jacobs School of Engineering. "This work represents a step toward democratizing AI."
The new method for fine-tuning and adapting LLMs was published in Transactions on Machine Learning Research . This research was supported by the National Science Foundation (IIS2405974 and IIS2339216) and National Institutes of Health (R35GM157217 and R21GM154171).
