Sign in

Fine-Tuning Linear Layers Only Is a Simple yet Effective Way for Task Arithmetic

By Ruochen Jin and others
Task arithmetic has recently emerged as a cost-effective and scalable approach to edit pre-trained models directly in weight space, by adding the fine-tuned weights of different tasks. The performance has been further improved by a linear property which is illustrated by weight disentanglement. Yet, conventional linearization methods (e.g., NTK linearization)... Show more
July 9, 2024
=
0
Loading PDF…
Loading full text...
Similar articles
Loading recommendations...
=
0
x1
Fine-Tuning Linear Layers Only Is a Simple yet Effective Way for Task Arithmetic
Click on play to start listening