GPT-2 Small Fine-Tuned on Logical Reasoning Summarizes Information on Punctuation Tokens

Abstract

How is information stored and aggregated within a language model performing inference? Preliminary evidence suggests that representations of punctuation tokens might serve as ``summary points'' for information about preceding text. We add to this body of evidence by demonstrating that GPT-2 small fine-tuned on the RuleTaker logical inference dataset aggregates crucial information about rules and sentences above period tokens.

Cite

Text

Chauhan and Geiger. "GPT-2 Small Fine-Tuned on Logical Reasoning Summarizes Information on Punctuation Tokens." NeurIPS 2024 Workshops: MINT, 2024.

Markdown

[Chauhan and Geiger. "GPT-2 Small Fine-Tuned on Logical Reasoning Summarizes Information on Punctuation Tokens." NeurIPS 2024 Workshops: MINT, 2024.](https://mlanthology.org/neuripsw/2024/chauhan2024neuripsw-gpt2/)

BibTeX

@inproceedings{chauhan2024neuripsw-gpt2,
  title     = {{GPT-2 Small Fine-Tuned on Logical Reasoning Summarizes Information on Punctuation Tokens}},
  author    = {Chauhan, Sonakshi and Geiger, Atticus},
  booktitle = {NeurIPS 2024 Workshops: MINT},
  year      = {2024},
  url       = {https://mlanthology.org/neuripsw/2024/chauhan2024neuripsw-gpt2/}
}