paper-with-me

홈 › Papers

Localizing Model Behavior with Path Patching

2023-04-12 · Nicholas Goldowsky-Dill, Chris MacLeod, Lucas Sato, Aryaman Arora

Localizing behaviors of neural networks to a subset of the network's components or a subset of interactions between components is a natural first step towards analyzing network mechanisms and possible failure modes. Existing work is often qualitative and ad-hoc, and there is no consensus on the appropriate way to evaluate localization claims. We introduce path patching, a technique for expressing and quantitatively testing a natural class of hypotheses expressing that behaviors are localized to a set of paths. We refine an explanation of induction heads, characterize a behavior of GPT-2, and open source a framework for efficiently running similar experiments.

📄 PDF Abstract BibTeX arXiv:2304.05969

Code (1)

redwoodresearch/rust_circuit_public 공식 구현 jax

Tasks

model

Methods 이 논문이 사용한 방법론

Refunds@Expedia|||How do I get a full refund from Expedia? “How do I get a full refund from Expedia? How do I get a full refund from Expedia? – Call ☎️ +1-(888) 829 (0881) or +1-805-330-4056 or +1-805-330-4056 for Quick Help &…
Attention 설명 없음
Linear Layer A Linear Layer is a projection $\mathbf{XW + b}$.
BPE Byte Pair Encoding, or BPE, is a subword segmentation algorithm that encodes rare and unknown words as sequences of subword units. The intuition is that various word…
Layer Normalization Unlike batch normalization, Layer Normalization directly estimates the normalization statistics from the summed inputs…
Residual Connection 설명 없음
Cosine Annealing Cosine Annealing is a type of learning rate schedule that has the effect of starting with a large learning rate that is relatively rapidly decreased to a minimum value before…
Linear Warmup With Cosine Annealing Linear Warmup With Cosine Annealing is a learning rate schedule where we increase the learning rate linearly for $n$ updates and then anneal according to a cosine schedule…

Similar Papers 제목 키워드 기반

AtP*: An efficient and scalable method for localizing LLM behaviour to components

2024-03-01 · János Kramár, Tom Lieberum, Rohin Shah, Neel Nanda

Activation Patching is a method of directly computing causal attributions of behavior to model components. However, applying it exhaustively requires a sweep with cost scaling linearly in the number of model components, …

RelP: Faithful and Efficient Circuit Discovery in Language Models via Relevance Patching

2025-08-28 · Farnoush Rezaei Jafari, Oliver Eberle, Ashkan Khakzar, Neel Nanda arxiv

Activation patching is a standard method in mechanistic interpretability for localizing the components of a model responsible for specific behaviors, but it is computationally expensive to apply at scale. Attribution pat…

Computational Efficiency

Localizing Lying in Llama: Understanding Instructed Dishonesty on True-False Questions Through Prompting, Probing, and Patching

2023-11-25 · James Campbell, Richard Ren, Phillip Guo

Large language models (LLMs) demonstrate significant knowledge through their outputs, though it is often unclear whether false outputs are due to a lack of knowledge or dishonesty. In this paper, we investigate instructe…

Prompt Engineering

Pathways of Visual Information Flow in Vision-Language Models

2026-07-03 · Israfel Salazar, Stella Frank, Dan Oneata, Desmond Elliott 외 arxiv

We study how visual information is routed in vision-language models (VLMs). Using causal patching on controlled synthetic and natural datasets, we find that models rely on two distinct pathways to solve visual tasks: A d…

Understanding Layer Patching in Model Size Interpolation

2026-07-09 · Sara Kangaslahti, Jonathan Geuter, Nihal V. Nayak, Marco Fumero 외 arxiv

Zero-shot model size interpolation aims to create new models of intermediate target sizes by combining existing models without additional training. Recent work on boomerang distillation [Kangaslahti et al., 2026] shows t…