Transformer Circuit Faithfulness Metrics are not Robust

• URL: http://arxiv.org/abs/2407.08734v1
• Date: Thu, 11 Jul 2024 17:59:00 GMT
• Title: Transformer Circuit Faithfulness Metrics are not Robust
• Authors: Joseph Miller, Bilal Chughtai, William Saunders,
• Abstract summary: We measure circuit 'faithfulness' by ablating portions of the model's computation. We conclude that existing circuit faithfulness scores reflect both the methodological choices of researchers as well as the actual components of the circuit. The ultimate goal of mechanistic interpretability work is to understand neural networks, so we emphasize the need for more clarity in the precise claims being made about circuits.
• Score: 0.04260910081285213
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Mechanistic interpretability work attempts to reverse engineer the learned algorithms present inside neural networks. One focus of this work has been to discover 'circuits' -- subgraphs of the full model that explain behaviour on specific tasks. But how do we measure the performance of such circuits? Prior work has attempted to measure circuit 'faithfulness' -- the degree to which the circuit replicates the performance of the full model. In this work, we survey many considerations for designing experiments that measure circuit faithfulness by ablating portions of the model's computation. Concerningly, we find existing methods are highly sensitive to seemingly insignificant changes in the ablation methodology. We conclude that existing circuit faithfulness scores reflect both the methodological choices of researchers as well as the actual components of the circuit - the task a circuit is required to perform depends on the ablation used to test it. The ultimate goal of mechanistic interpretability work is to understand neural networks, so we emphasize the need for more clarity in the precise claims being made about circuits. We open source a library at https://github.com/UFO-101/auto-circuit that includes highly efficient implementations of a wide range of ablation methodologies and circuit discovery algorithms.

Related papers

• Adversarial Circuit Evaluation [1.1893676124374688]
  We evaluate three circuits found in the literature (IOI, greater-than, and docstring) in an adversarial manner. We measure the KL divergence between the full model's output and the circuit's output, calculated through resample ablation, and we analyze the worst-performing inputs.
  arXiv  Detail & Related papers  (2024-07-21T13:43:44Z)
• Functional Faithfulness in the Wild: Circuit Discovery with Differentiable Computation Graph Pruning [14.639036250438517]
  We introduce a comprehensive reformulation of the task known as Circuit Discovery, along with DiscoGP. DiscoGP is a novel and effective algorithm based on differentiable masking for discovering circuits.
  arXiv  Detail & Related papers  (2024-07-04T09:42:25Z)
• Finding Transformer Circuits with Edge Pruning [71.12127707678961]
  We propose Edge Pruning as an effective and scalable solution to automated circuit discovery. Our method finds circuits in GPT-2 that use less than half the number of edges compared to circuits found by previous methods. Thanks to its efficiency, we scale Edge Pruning to CodeLlama-13B, a model over 100x the scale that prior methods operate on.
  arXiv  Detail & Related papers  (2024-06-24T16:40:54Z)
• Sparse Feature Circuits: Discovering and Editing Interpretable Causal Graphs in Language Models [55.19497659895122]
  We introduce methods for discovering and applying sparse feature circuits. These are causally implicatedworks of human-interpretable features for explaining language model behaviors.
  arXiv  Detail & Related papers  (2024-03-28T17:56:07Z)
• Uncovering Intermediate Variables in Transformers using Circuit Probing [32.382094867951224]
  We propose a new analysis technique -- circuit probing -- that automatically uncovers low-level circuits that compute hypothesized intermediate variables. We apply this method to models trained on simple arithmetic tasks, demonstrating its effectiveness at (1) deciphering the algorithms that models have learned, (2) revealing modular structure within a model, and (3) tracking the development of circuits over training.
  arXiv  Detail & Related papers  (2023-11-07T21:27:17Z)
• Attribution Patching Outperforms Automated Circuit Discovery [3.8695554579762814]
  We show that a simple method based on attribution patching outperforms all existing methods. We apply a linear approximation to activation patching to estimate the importance of each edge in the computational subgraph.
  arXiv  Detail & Related papers  (2023-10-16T12:34:43Z)
• CktGNN: Circuit Graph Neural Network for Electronic Design Automation [67.29634073660239]
  This paper presents a Circuit Graph Neural Network (CktGNN) that simultaneously automates the circuit topology generation and device sizing. We introduce Open Circuit Benchmark (OCB), an open-sourced dataset that contains $10$K distinct operational amplifiers. Our work paves the way toward a learning-based open-sourced design automation for analog circuits.
  arXiv  Detail & Related papers  (2023-08-31T02:20:25Z)
• Learning to Design Analog Circuits to Meet Threshold Specifications [2.5818330243826924]
  We propose a method for generating from simulation data a dataset on which a system can be trained to design circuits to meet threshold specifications. We show that our method consistently reaches success rate better than 90% at 5% error margin, while also improving data efficiency by upward of an order of magnitude.
  arXiv  Detail & Related papers  (2023-07-25T23:25:05Z)
• Adaptive Planning Search Algorithm for Analog Circuit Verification [53.97809573610992]
  We propose a machine learning (ML) approach, which uses less simulations. We show that the proposed approach is able to provide OCCs closer to the specifications for all circuits.
  arXiv  Detail & Related papers  (2023-06-23T12:57:46Z)
• Quantum circuit debugging and sensitivity analysis via local inversions [62.997667081978825]
  We present a technique that pinpoints the sections of a quantum circuit that affect the circuit output the most. We demonstrate the practicality and efficacy of the proposed technique by applying it to example algorithmic circuits implemented on IBM quantum machines.
  arXiv  Detail & Related papers  (2022-04-12T19:39:31Z)
• Pretraining Graph Neural Networks for few-shot Analog Circuit Modeling and Design [68.1682448368636]
  We present a supervised pretraining approach to learn circuit representations that can be adapted to new unseen topologies or unseen prediction tasks. To cope with the variable topological structure of different circuits we describe each circuit as a graph and use graph neural networks (GNNs) to learn node embeddings. We show that pretraining GNNs on prediction of output node voltages can encourage learning representations that can be adapted to new unseen topologies or prediction of new circuit level properties.
  arXiv  Detail & Related papers  (2022-03-29T21:18:47Z)

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.