Polynomial Regression as a Task for Understanding In-Context Learning Through Finetuning and Alignment
Abstract
Simple function classes have emerged as toy problems to better understand in-context-learning in transformer-based architectures used for large language models. But previously proposed simple function classes like linear regression or multi-layer-perceptrons lack the structure required to explore things like prompting and alignment within models capable of in-context-learning. We propose univariate polynomial regression as a function class that is just rich enough to study prompting and alignment, while allowing us to visualize and understand what is going on clearly.
Cite
Text
Wilcoxson et al. "Polynomial Regression as a Task for Understanding In-Context Learning Through Finetuning and Alignment." ICML 2024 Workshops: ICL, 2024.Markdown
[Wilcoxson et al. "Polynomial Regression as a Task for Understanding In-Context Learning Through Finetuning and Alignment." ICML 2024 Workshops: ICL, 2024.](https://mlanthology.org/icmlw/2024/wilcoxson2024icmlw-polynomial/)BibTeX
@inproceedings{wilcoxson2024icmlw-polynomial,
title = {{Polynomial Regression as a Task for Understanding In-Context Learning Through Finetuning and Alignment}},
author = {Wilcoxson, Max and Svendgård, Morten and Doshi, Ria and Davis, Dylan and Vir, Reya and Sahai, Anant},
booktitle = {ICML 2024 Workshops: ICL},
year = {2024},
url = {https://mlanthology.org/icmlw/2024/wilcoxson2024icmlw-polynomial/}
}