FADE: Why Bad Descriptions Happen to Good Features

• URL: http://arxiv.org/abs/2502.16994v1
• Date: Mon, 24 Feb 2025 09:28:35 GMT
• Title: FADE: Why Bad Descriptions Happen to Good Features
• Authors: Bruno Puri, Aakriti Jain, Elena Golimblevskaia, Patrick Kahardipraja, Thomas Wiegand, Wojciech Samek, Sebastian Lapuschkin,
• Abstract summary: We introduce FADE: Feature Alignment to Description Evaluation.<n>FADE is a scalable framework for evaluating feature-description alignment.<n>We apply FADE to analyze existing open-source feature descriptions, and assess key components of automated interpretability pipelines.
• Score: 14.00042287629001
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Recent advances in mechanistic interpretability have highlighted the potential of automating interpretability pipelines in analyzing the latent representations within LLMs. While they may enhance our understanding of internal mechanisms, the field lacks standardized evaluation methods for assessing the validity of discovered features. We attempt to bridge this gap by introducing FADE: Feature Alignment to Description Evaluation, a scalable model-agnostic framework for evaluating feature-description alignment. FADE evaluates alignment across four key metrics - Clarity, Responsiveness, Purity, and Faithfulness - and systematically quantifies the causes for the misalignment of feature and their description. We apply FADE to analyze existing open-source feature descriptions, and assess key components of automated interpretability pipelines, aiming to enhance the quality of descriptions. Our findings highlight fundamental challenges in generating feature descriptions, particularly for SAEs as compared to MLP neurons, providing insights into the limitations and future directions of automated interpretability. We release FADE as an open-source package at: https://github.com/brunibrun/FADE.

Related papers

• Video SimpleQA: Towards Factuality Evaluation in Large Video Language Models [69.68265487134686]
  Video SimpleQA is the first comprehensive benchmark tailored for factuality evaluation of LVLMs. Our work distinguishes from existing video benchmarks through the following key features. Answers are crafted as unambiguous and definitively correct in a short format.
  arXiv  Detail & Related papers  (2025-03-24T17:46:09Z)
• Understanding Ranking LLMs: A Mechanistic Analysis for Information Retrieval [20.353393773305672]
  We employ a probing-based analysis to examine neuron activations in ranking LLMs.<n>Our study spans a broad range of feature categories, including lexical signals, document structure, query-document interactions, and complex semantic representations.<n>Our findings offer crucial insights for developing more transparent and reliable retrieval systems.
  arXiv  Detail & Related papers  (2024-10-24T08:20:10Z)
• $\textit{X}^2$-DFD: A framework for e${X}$plainable and e${X}$tendable Deepfake Detection [52.14468236527728]
  We propose a novel framework called $X2$-DFD, consisting of three core modules.<n>The first module, Model Feature Assessment (MFA), measures the detection capabilities of forgery features intrinsic to MLLMs, and gives a descending ranking of these features.<n>The second module, Strong Feature Strengthening (SFS), enhances the detection and explanation capabilities by fine-tuning the MLLM on a dataset constructed based on the top-ranked features.<n>The third module, Weak Feature Supplementing (WFS), improves the fine-tuned MLLM's capabilities on lower-ranked features by integrating external dedicated
  arXiv  Detail & Related papers  (2024-10-08T15:28:33Z)
• CELA: Cost-Efficient Language Model Alignment for CTR Prediction [70.65910069412944]
  Click-Through Rate (CTR) prediction holds a paramount position in recommender systems. Recent efforts have sought to mitigate these challenges by integrating Pre-trained Language Models (PLMs) We propose textbfCost-textbfEfficient textbfLanguage Model textbfAlignment (textbfCELA) for CTR prediction.
  arXiv  Detail & Related papers  (2024-05-17T07:43:25Z)
• Tuning-Free Accountable Intervention for LLM Deployment -- A Metacognitive Approach [55.613461060997004]
  Large Language Models (LLMs) have catalyzed transformative advances across a spectrum of natural language processing tasks. We propose an innovative textitmetacognitive approach, dubbed textbfCLEAR, to equip LLMs with capabilities for self-aware error identification and correction.
  arXiv  Detail & Related papers  (2024-03-08T19:18:53Z)
• Prospector Heads: Generalized Feature Attribution for Large Models & Data [82.02696069543454]
  We introduce prospector heads, an efficient and interpretable alternative to explanation-based attribution methods. We demonstrate how prospector heads enable improved interpretation and discovery of class-specific patterns in input data.
  arXiv  Detail & Related papers  (2024-02-18T23:01:28Z)
• Word-Level ASR Quality Estimation for Efficient Corpus Sampling and Post-Editing through Analyzing Attentions of a Reference-Free Metric [5.592917884093537]
  The potential of quality estimation (QE) metrics is introduced and evaluated as a novel tool to enhance explainable artificial intelligence (XAI) in ASR systems. The capabilities of the NoRefER metric are explored in identifying word-level errors to aid post-editors in refining ASR hypotheses.
  arXiv  Detail & Related papers  (2024-01-20T16:48:55Z)
• Sum-of-Parts: Self-Attributing Neural Networks with End-to-End Learning of Feature Groups [8.68707471649733]
  Sum-of-Parts is a framework that transforms any differentiable model into a group-based SANN.<n> SOP achieves state-of-the-art performance for SANNs on vision and language tasks.
  arXiv  Detail & Related papers  (2023-10-25T02:50:10Z)
• LaPLACE: Probabilistic Local Model-Agnostic Causal Explanations [1.0370398945228227]
  We introduce LaPLACE-explainer, designed to provide probabilistic cause-and-effect explanations for machine learning models. The LaPLACE-Explainer component leverages the concept of a Markov blanket to establish statistical boundaries between relevant and non-relevant features. Our approach offers causal explanations and outperforms LIME and SHAP in terms of local accuracy and consistency of explained features.
  arXiv  Detail & Related papers  (2023-10-01T04:09:59Z)
• FIND: A Function Description Benchmark for Evaluating Interpretability Methods [86.80718559904854]
  This paper introduces FIND (Function INterpretation and Description), a benchmark suite for evaluating automated interpretability methods. FIND contains functions that resemble components of trained neural networks, and accompanying descriptions of the kind we seek to generate. We evaluate methods that use pretrained language models to produce descriptions of function behavior in natural language and code.
  arXiv  Detail & Related papers  (2023-09-07T17:47:26Z)
• Explainability in Process Outcome Prediction: Guidelines to Obtain Interpretable and Faithful Models [77.34726150561087]
  We define explainability through the interpretability of the explanations and the faithfulness of the explainability model in the field of process outcome prediction. This paper contributes a set of guidelines named X-MOP which allows selecting the appropriate model based on the event log specifications.
  arXiv  Detail & Related papers  (2022-03-30T05:59:50Z)
• AcME -- Accelerated Model-agnostic Explanations: Fast Whitening of the Machine-Learning Black Box [1.7534486934148554]
  interpretability approaches should provide actionable insights without making the users wait. We propose Accelerated Model-agnostic Explanations (AcME), an interpretability approach that quickly provides feature importance scores both at the global and the local level. AcME computes feature ranking, but it also provides a what-if analysis tool to assess how changes in features values would affect model predictions.
  arXiv  Detail & Related papers  (2021-12-23T15:18:13Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.