Pith. sign in

REVIEW 10 cited by

Have Faith in Faithfulness: Going Beyond Circuit Overlap When Finding Model Mechanisms

Not yet reviewed by Pith; the record is open.

Re-run · record.json Download PDF Read on arXiv ↗

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2403.17806 v2 pith:VIRGBTQS submitted 2024-03-26 cs.LG cs.CL

Michael Hanna , Sandro Pezzelle , Yonatan Belinkov This is my paper · ORCID
classification cs.LGcs.CL
keywords circuitsmodelcircuitfaithfulnessfoundinterventionsoverlapaims
verification ladder T0 review T1 audit T2 compute T3 formal
0 comments
read the original abstract

Many recent language model (LM) interpretability studies have adopted the circuits framework, which aims to find the minimal computational subgraph, or circuit, that explains LM behavior on a given task. Most studies determine which edges belong in a LM's circuit by performing causal interventions on each edge independently, but this scales poorly with model size. Edge attribution patching (EAP), gradient-based approximation to interventions, has emerged as a scalable but imperfect solution to this problem. In this paper, we introduce a new method - EAP with integrated gradients (EAP-IG) - that aims to better maintain a core property of circuits: faithfulness. A circuit is faithful if all model edges outside the circuit can be ablated without changing the model's performance on the task; faithfulness is what justifies studying circuits, rather than the full model. Our experiments demonstrate that circuits found using EAP are less faithful than those found using EAP-IG, even though both have high node overlap with circuits found previously using causal interventions. We conclude more generally that when using circuits to compare the mechanisms models use to solve tasks, faithfulness, not overlap, is what should be measured.

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 10 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Temporal Preference Concepts and their Functions in a Large Language Model

    cs.LG 2026-05 unverdicted novelty 6.5 of 10

    Temporal preference in Qwen3-4B-Instruct-2507 localizes to layers 17–35 (especially L24 attention), has curved residual-stream geometry, is behaviorally unstable, and can be bidirectionally steered.

  2. Wired for Overconfidence: A Mechanistic Perspective on Inflated Verbalized Confidence in LLMs

    cs.CL 2026-04 conditional novelty 6.0 of 10

    Inflated verbalized confidence in Qwen2.5-3B and Llama-3.2-3B is driven by a compact, cross-dataset set of middle-to-late-layer MLP blocks and attention heads, and steering or ablating those components at inference ti...

  3. A Mechanistic Perspective and Circuit-Guided Difficulty Metric for Unlearning

    cs.LG 2026-01 conditional novelty 6.0 of 10

    A circuit-similarity score predicts which samples an LLM unlearning method will fail to erase, with hard samples relying on deeper, output-facing pathways.

  4. Time Course MechInterp: Analyzing the Evolution of Components and Knowledge in Large Language Models

    cs.CL 2025-06 conditional novelty 6.0 of 10

    Over 40 checkpoints of OLMo-7B, attention heads and FFNs shift from general-purpose to specialized roles for factual recall, with location-based facts learned earlier and more stably than name-based facts.

  5. Pierce the Mists, Greet the Sky: Decipher Knowledge Overshadowing via Knowledge Circuit Analysis

    cs.CL 2025-05 conditional novelty 6.0 of 10

    PhantomCircuit traces knowledge overshadowing to attention circuits during training and prunes circuit edges to recover the overshadowed answer.

  6. Efficient LLM Adversarial Training via Low-Rank Defense and Circuit-Guided Surrogates

    cs.LG 2026-07 conditional novelty 5.0 of 10

    Combining suffix-window representation finetuning with an ActGrad-pruned surrogate cuts latent-adversarial-training FLOPs per step by 48.1% with only 0.0118% trainable parameters, while accepting higher attack success rates.

  7. $C$-$\Delta\Theta$: Circuit-Restricted Weight Arithmetic for Selective Refusal

    cs.CL 2026-02 conditional novelty 5.0 of 10

    Selective refusal can be improved by editing only the small weight circuit found by EAP-IG, yielding offline checkpoints with low over-refusal.

  8. Interpretation Meets Safety: A Survey on Interpretation Methods and Tools for Improving LLM Safety

    cs.SE 2025-06 accept novelty 5.0 of 10

    A new survey organizes LLM interpretation methods by workflow stage and connects them to safety enhancement strategies and tools, covering around 70 works.

  9. A Numerical PDEs Approach to Evolution Equations in Shape Analysis Based on Regularized Morphoelasticity

    math.NA 2026-04 unverdicted novelty 4.0 of 10

    Regularized morphoelasticity yields a high-order elliptic system for continuous shape evolution that is solved by mixed finite elements in FEniCSx within an LDDMM-style optimal-control growth model.

  10. Mechanistic Unveiling of Transformer Circuits: Self-Influence as a Key to Model Reasoning

    cs.AI 2025-02 reject novelty 4.0 of 10

    SICAF traces per-token self-influence inside extracted circuits to map GPT-2's reasoning on the IOI task.

Pith tools