AI/ ai · fine-tuning · peft · llm-training

CRAFT Cuts Fine-Tuning Parameters by Freezing Attention Weights

A new tensor-based fine-tuning method compresses pre-trained weights instead of updates, matching LoRA's accuracy with far fewer trainable parameters.

A new fine-tuning technique claims to beat LoRA's accuracy with a few hundred times fewer trainable parameters.

Researchers built a technique called LoRA-CRAFT (CRAFT for short) that fine-tunes large language models by freezing almost everything and training only a tiny slice of it. Most fine-tuning methods like LoRA add small trainable matrices on top of a model's weights and update those. CRAFT instead breaks the pre-trained attention weights themselves into compressed pieces, using a matrix trick called Tucker decomposition, then locks those pieces in place and trains only small square adjustment matrices around them. In tests on RoBERTa models and on LLaMA2-7B and LLaMA3-8B, CRAFT matched or beat standard LoRA's accuracy while training a small fraction of the parameters.

That matters because fine-tuning cost scales with parameter count, and CRAFT's edge over LoRA widens as models get bigger, not smaller - on LLaMA3-8B it beat LoRA's average accuracy using hundreds of times fewer trainable parameters. That is a different bet than most efficiency research, which usually trades away some accuracy for savings; here the paper reports no such tax, at least on the benchmarks tested.

It is an incremental step past PEFT cousins like LoTR and SuperLoRA rather than a clean break, and the paper's own benchmarks - GLUE and a handful of commonsense reasoning sets - are the ones every PEFT method clears before anyone tries it on harder, messier tasks.

TR

The Revision

Written by an AI system from the public sources credited above. How we write →