arXiv Analytics

Sign in

arXiv:2306.00976 [cs.CL]AbstractReferencesReviewsResources

TopEx: Topic-based Explanations for Model Comparison

Shreya Havaldar, Adam Stein, Eric Wong, Lyle Ungar

Published 2023-06-01Version 1

Meaningfully comparing language models is challenging with current explanation methods. Current explanations are overwhelming for humans due to large vocabularies or incomparable across models. We present TopEx, an explanation method that enables a level playing field for comparing language models via model-agnostic topics. We demonstrate how TopEx can identify similarities and differences between DistilRoBERTa and GPT-2 on a variety of NLP tasks.

Comments: 7 pages, 4 figures, Accepted to ICLR Tiny Papers Track
Categories: cs.CL
Related articles: Most relevant | Search more
arXiv:2109.05190 [cs.CL] (Published 2021-09-11)
Eliciting Knowledge from Language Models for Event Extraction
arXiv:2102.07219 [cs.CL] (Published 2021-02-14)
Error-driven Pruning of Language Models for Virtual Assistants
arXiv:1805.08154 [cs.CL] (Published 2018-05-21)
Numeracy for Language Models: Evaluating and Improving their Ability to Predict Numbers