565 citations · 1.1k across the 53 of their papers we have counts for
60 papers
Ownership in AI-Assisted Everyday Tasks
Megan Wei, Melanie Subbiah, Audrey Lee +4
When does work done with AI still feel like ours? As AI becomes woven into everyday tasks, we must examine what happens to our sense of ownership and contribution when a machine sh…
Slow to See, Slow to Suppress: Understanding the Effects of Modality in Context-Memory Conflicts
Athulith Paraselli, Etha Tianze Hua, Ellie Pavlick
We investigate how vision-language models (VLMs) handle context-memory conflicts; that is, situations in which the model is given information in context that differs from what was…
Linguistic Context Recodes Visual Representations in Vision-Language Models
Brian Song, Michael A. Lepori, Ellie Pavlick
Goal-directed visual processing is a hallmark of human visual intelligence, resulting in representations that support downstream tasks such as categorization or search. Though visi…
Whither symbols in the era of advanced neural networks?
Thomas L. Griffiths, Brenden M. Lake, R. Thomas McCoy +2
Some of the strongest evidence that human minds should be thought about in terms of symbolic systems has been the way they combine ideas, produce novelty, and learn quickly. We arg…
Transferring Linear Features Across Language Models With Model Stitching
Alan Chen, Jack Merullo, Alessandro Stolfo +1
In this work, we demonstrate that affine mappings between residual streams of language models is a cheap way to effectively transfer represented features between models. We apply t…
Are LLMs Models of Distributional Semantics? A Case Study on Quantifiers
Zhang Enyan, Zewei Wang, Michael A. Lepori +2
Distributional semantics is the linguistic theory that a word's meaning can be derived from its distribution in natural language (i.e., its use). Language models are commonly viewe…
The Same But Different: Structural Similarities and Differences in Multilingual Language Modeling
Ruochen Zhang, Qinan Yu, Matianyu Zang +2
We employ new tools from mechanistic interpretability in order to ask whether the internal structure of large language models (LLMs) shows correspondence to the linguistic structur…
$100K or 100 Days: Trade-offs when Pre-Training with Academic Resources
Apoorv Khandelwal, Tian Yun, Nihal V. Nayak +4
Pre-training is notoriously compute-intensive and academic researchers are notoriously under-resourced. It is, therefore, commonly assumed that academics can't pre-train models. In…
Beyond the Doors of Perception: Vision Transformers Represent Relations Between Objects
Michael A. Lepori, Alexa R. Tartaglini, Wai Keen Vong +3
Though vision transformers (ViTs) have achieved state-of-the-art performance in a variety of settings, they exhibit surprising failures when performing tasks involving visual relat…
Talking Heads: Understanding Inter-layer Communication in Transformer Language Models
Jack Merullo, Carsten Eickhoff, Ellie Pavlick
Although it is known that transformer language models (LMs) pass features from early layers to later layers, it is not well understood how this information is represented and route…
Dual Process Learning: Controlling Use of In-Context vs. In-Weights Strategies with Weight Forgetting
Suraj Anand, Michael A. Lepori, Jack Merullo +1
Language models have the ability to perform in-context learning (ICL), allowing them to flexibly adapt their behavior based on context. This contrasts with in-weights learning (IWL…
Transformer Mechanisms Mimic Frontostriatal Gating Operations When Trained on Human Working Memory Tasks
Aaron Traylor, Jack Merullo, Michael J. Frank +1
Models based on the Transformer neural network architecture have seen success on a wide variety of tasks that appear to require complex "cognitive branching" -- or the ability to m…
Analyzing Modular Approaches for Visual Question Decomposition
Apoorv Khandelwal, Ellie Pavlick, Chen Sun
Modular neural networks without additional training have recently been shown to surpass end-to-end neural networks on challenging vision-language tasks. The latest such methods sim…
Emergence of Abstract State Representations in Embodied Sequence Modeling
Tian Yun, Zilai Zeng, Kunal Handa +4
Decision making via sequence modeling aims to mimic the success of language models, where actions taken by an embodied agent are modeled as tokens to predict. Despite their promisi…
Uncovering Intermediate Variables in Transformers using Circuit Probing
Michael A. Lepori, Thomas Serre, Ellie Pavlick
Neural network models have achieved high performance on a wide variety of complex tasks, but the algorithms that they implement are notoriously difficult to interpret. It is often…
Characterizing Mechanisms for Factual Recall in Language Models
Qinan Yu, Jack Merullo, Ellie Pavlick
Language Models (LMs) often must integrate facts they memorized in pretraining with new information that appears in a given context. These two sources can disagree, causing competi…
Instilling Inductive Biases with Subnetworks
Enyan Zhang, Michael A. Lepori, Ellie Pavlick
Despite the recent success of artificial neural networks on a variety of tasks, we have little knowledge or control over the exact solutions these models implement. Instilling indu…
Deep Neural Networks Can Learn Generalizable Same-Different Visual Relations
Alexa R. Tartaglini, Sheridan Feucht, Michael A. Lepori +4
Although deep neural networks can achieve human-level performance on many object recognition benchmarks, prior work suggests that these same models fail to learn simple abstract re…
Circuit Component Reuse Across Tasks in Transformer Language Models
Jack Merullo, Carsten Eickhoff, Ellie Pavlick
Recent work in mechanistic interpretability has shown that behaviors in language models can be successfully reverse-engineered through circuit analysis. A common criticism, however…
NeuroSurgeon: A Toolkit for Subnetwork Analysis
Michael A. Lepori, Ellie Pavlick, Thomas Serre
Despite recent advances in the field of explainability, much remains unknown about the algorithms that neural networks learn to represent. Recent work has attempted to understand t…
Training Priors Predict Text-To-Image Model Performance
Charles Lovering, Ellie Pavlick
Text-to-image models can often generate some relations, i.e., "astronaut riding horse", but fail to generate other relations composed of the same basic parts, i.e., "horse riding a…
Testing Causal Models of Word Meaning in GPT-3 and -4
Sam Musker, Ellie Pavlick
Large Language Models (LLMs) have driven extraordinary improvements in NLP. However, it is unclear how such models represent lexical concepts-i.e., the meanings of the words they u…
Language Models Implement Simple Word2Vec-style Vector Arithmetic
Jack Merullo, Carsten Eickhoff, Ellie Pavlick
A primary criticism towards language models (LMs) is their inscrutability. This paper presents evidence that, despite their size and complexity, LMs sometimes exploit a simple vect…
Simfluence: Modeling the Influence of Individual Training Examples by Simulating Training Runs
Kelvin Guu, Albert Webson, Ellie Pavlick +3
Training data attribution (TDA) methods offer to trace a model's prediction on any given example back to specific influential training examples. Existing approaches do so by assign…