70 citations · 239 across the 45 of their papers we have counts for
48 papers
MANCE: Manifold Aware Concept Erasure
Matan Avitan, Yoav Goldberg, Yanai Elazar
Concept erasure aims to remove a target concept from a representation while preserving the other information encoded in it. This is difficult because representations encode many co…
Benchmarking Agentic Review Systems
Dang Nguyen, Wanqing Hao, Yanai Elazar +1
A new class of agentic review systems are emerging as a remedy to the pressure placed on peer review systems by AI-assisted research, but it is unclear how they should be evaluated…
LLM-Generated or Human-Written? Comparing Review and Non-Review Papers on ArXiv
Yanai Elazar, Maria Antoniak
ArXiv recently prohibited the upload of unpublished review papers to its servers in the Computer Science domain, citing a high prevalence of LLM-generated content in these categori…
Rewriting History: A Recipe for Interventional Analyses to Study Data Effects on Model Behavior
Rahul Nadkarni, Yanai Elazar, Hila Gonen +1
We present an experimental recipe for studying the relationship between training data and language model (LM) behavior. We outline steps for intervening on data batches -- i.e., ``…
On Linear Representations and Pretraining Data Frequency in Language Models
Jack Merullo, Noah A. Smith, Sarah Wiegreffe +1
Pretraining data has a direct impact on the behaviors and quality of language models (LMs), but we only understand the most basic principles of this relationship. While most work f…
OLMoTrace: Tracing Language Model Outputs Back to Trillions of Training Tokens
Jiacheng Liu, Taylor Blanton, Yanai Elazar +28
We present OLMoTrace, the first system that traces the outputs of language models back to their full, multi-trillion-token training data in real time. OLMoTrace finds and shows ver…
Better Aligned with Survey Respondents or Training Data? Unveiling Political Leanings of LLMs on U.S. Supreme Court Cases
Shanshan Xu, T. Y. S. S Santosh, Yanai Elazar +3
Recent works have shown that Large Language Models (LLMs) have a tendency to memorize patterns and biases present in their training data, raising important questions about how such…
GRADE: Quantifying Sample Diversity in Text-to-Image Models
Royi Rassin, Aviv Slobodkin, Shauli Ravfogel +2
We introduce GRADE, an automatic method for quantifying sample diversity in text-to-image models. Our method leverages the world knowledge embedded in large language models and vis…
Hybrid Preferences: Learning to Route Instances for Human vs. AI Feedback
Lester James V. Miranda, Yizhong Wang, Yanai Elazar +6
Learning from human feedback has enabled the alignment of language models (LMs) with human preferences. However, collecting human preferences is expensive and time-consuming, with…
How Many Images Does It Take? Estimating Imitation Thresholds in Text-to-Image Models
Sahil Verma, Royi Rassin, Arnav Das +6
Text-to-image models are trained using large datasets of image-text pairs collected from the internet. These datasets often include copyrighted and private images. Training models…
Data Contamination Report from the 2024 CONDA Shared Task
Oscar Sainz, Iker García-Ferrero, Alon Jacovi +25
The 1st Workshop on Data Contamination (CONDA 2024) focuses on all relevant aspects of data contamination in natural language processing, where data contamination is understood as…
Generalization v.s. Memorization: Tracing Language Models' Capabilities Back to Pretraining Data
Xinyi Wang, Antonis Antoniades, Yanai Elazar +4
The impressive capabilities of large language models (LLMs) have sparked debate over whether these models genuinely generalize to unseen tasks or predominantly rely on memorizing v…
Detection and Measurement of Syntactic Templates in Generated Text
Chantal Shaib, Yanai Elazar, Junyi Jessy Li +1
Recent work on evaluating the diversity of text generated by LLMs has focused on word-level features. Here we offer an analysis of syntactic features to characterize general repeti…
Applying Intrinsic Debiasing on Downstream Tasks: Challenges and Considerations for Machine Translation
Bar Iluz, Yanai Elazar, Asaf Yehudai +1
Most works on gender bias focus on intrinsic bias -- removing traces of information about a protected group from the model's internal representation. However, these works are often…
Evaluating -Gram Novelty of Language Models Using Rusty-DAWG
William Merrill, Noah A. Smith, Yanai Elazar
How novel are texts generated by language models (LMs) relative to their training corpora? In this work, we investigate the extent to which modern LMs generate -grams from their…
A Survey on Data Selection for Language Models
Alon Albalak, Yanai Elazar, Sang Michael Xie +11
A major factor in the recent success of large language models is the use of enormous and ever-growing text datasets for unsupervised pre-training. However, naively training a model…
Calibrating Large Language Models with Sample Consistency
Qing Lyu, Kumar Shridhar, Chaitanya Malaviya +6
Accurately gauging the confidence level of Large Language Models' (LLMs) predictions is pivotal for their reliable application. However, LLMs are often uncalibrated inherently and…
OLMo: Accelerating the Science of Language Models
Dirk Groeneveld, Iz Beltagy, Pete Walsh +40
Language models (LMs) have become ubiquitous in both NLP research and in commercial product offerings. As their commercial importance has surged, the most powerful models have beco…
Dolma: an Open Corpus of Three Trillion Tokens for Language Model Pretraining Research
Luca Soldaini, Rodney Kinney, Akshita Bhagia +33
Information about pretraining corpora used to train the current best-performing language models is seldom discussed: commercial models rarely detail their data, and even open model…
Paloma: A Benchmark for Evaluating Language Model Fit
Ian Magnusson, Akshita Bhagia, Valentin Hofmann +13
Evaluations of language models (LMs) commonly report perplexity on monolithic data held out from training. Implicitly or explicitly, this data is composed of domains--varying distr…
Measuring and Improving Attentiveness to Partial Inputs with Counterfactuals
Yanai Elazar, Bhargavi Paranjape, Hao Peng +5
The inevitable appearance of spurious correlations in training datasets hurts the generalization of NLP models on unseen data. Previous work has found that datasets with paired inp…
What's In My Big Data?
Yanai Elazar, Akshita Bhagia, Ian Magnusson +10
Large text corpora are the backbone of language models. However, we have a limited understanding of the content of these corpora, including general statistics, quality, social fact…
The Bias Amplification Paradox in Text-to-Image Generation
Preethi Seshadri, Sameer Singh, Yanai Elazar
Bias amplification is a phenomenon in which models exacerbate biases or stereotypes present in the training data. In this paper, we study bias amplification in the text-to-image do…
Estimating the Causal Effect of Early ArXiving on Paper Acceptance
Yanai Elazar, Jiayao Zhang, David Wadden +2
What is the effect of releasing a preprint of a paper before it is submitted for peer review? No randomized controlled trial has been conducted, so we turn to observational data to…