The Effect of Model Size on LLM Post-hoc Explainability via LIME

Henning Heyen, Amy Widdicombe, Noah Y. Siegel, Maria Perez-Ortiz, Philip Treleaven
5/8/2024
cs.CLcs.AIcs.LG

No official code found. Best implementation: henningheyen/Scalability-Of-LLM-Posthoc-Explanations (3 stars, MIT, last commit 2024-04-26).

Abstract

Large language models (LLMs) are becoming bigger to boost performance. However, little is known about how explainability is affected by this trend. This work explores LIME explanations for DeBERTaV3 models of four different sizes on natural language inference (NLI) and zero-shot classification (ZSC) tasks. We evaluate the explanations based on their faithfulness to the models' internal decision processes and their plausibility, i.e. their agreement with human explanations. The key finding is that increased model size does not correlate with plausibility despite improved model performance, suggesting a misalignment between the LIME explanations and the models' internal processes as model size increases. Our results further suggest limitations regarding faithfulness metrics in NLI contexts.

View on arXivView PDF

Code Implementations(1)

henningheyen/Scalability-Of-LLM-Posthoc-ExplanationsCommunity100%
MIT

ICLR 2024 workshop on Secure and Trustworthy LLMs

30HTML, PythonFeb 6, 20242 years agoMIT

Similar Papers

Can LLM-Generated Textual Explanations Enhance Model Classification Performance? An Empirical Study

86% match1 repo2025

Lost in Inference: Rediscovering the Role of Natural Language Inference for Large Language Models

86% match2024

Evaluating the Reliability of Self-Explanations in Large Language Models

86% match1 repo12024

BELL: Benchmarking the Explainability of Large Language Models

85% match2025

Towards Uncovering How Large Language Model Works: An Explainability Perspective

85% match2024

Explainability of Large Language Models: Opportunities and Challenges toward Generating Trustworthy Explanations

85% match2025

Cite this paper

@article{heyen2024effect,
  title  = {The Effect of Model Size on LLM Post-hoc Explainability via LIME},
  author = {Henning Heyen and Amy Widdicombe and Noah Y. Siegel and Maria Perez-Ortiz and Philip Treleaven},
  year   = {2024},
  eprint = {2405.05348},
  archivePrefix = {arXiv},
  url    = {http://arxiv.org/abs/2405.05348v1}
}

Add a badge to your README

Shows whether official code is linked for this paper.

Code of Paper badge
[![Code of Paper](https://codeofpaper.com/badge/2405.05348.svg)](https://codeofpaper.com/paper/2405.05348)

Discussion