139 citations · 361 across the 17 of their papers we have counts for
19 papers
OpenAI GPT-5 System Card
Aaditya Singh, Adam Fry, Adam Perelman +483
This is the system card published alongside the OpenAI GPT-5 launch, August 2025. GPT-5 is a unified system with a smart and fast model that answers most questions, a deeper reason…
CAT: Content-Adaptive Image Tokenization
Junhong Shen, Kushal Tirumala, Michihiro Yasunaga +4
Most existing image tokenizers encode images into a fixed number of tokens or patches, overlooking the inherent variability in image complexity. To address this, we introduce Conte…
When Worse is Better: Navigating the compression-generation tradeoff in visual tokenization
Vivek Ramanujan, Kushal Tirumala, Armen Aghajanyan +2
Current image generation methods are based on a two-stage training approach. In stage 1, an auto-encoder is trained to compress an image into a latent space; in stage 2, a generati…
Transfusion: Predict the Next Token and Diffuse Images with One Multi-Modal Model
Chunting Zhou, Lili Yu, Arun Babu +7
We introduce Transfusion, a recipe for training a multi-modal model over discrete and continuous data. Transfusion combines the language modeling loss function (next token predicti…
Brevity is the soul of wit: Pruning long files for code generation
Aaditya K. Singh, Yu Yang, Kushal Tirumala +2
Data curation is commonly considered a "secret-sauce" for LLM training, with higher quality data usually leading to better LLM performance. Given the scale of internet-scraped corp…
An Introduction to Vision-Language Modeling
Florian Bordes, Richard Yuanzhe Pang, Anurag Ajay +38
Following the recent popularity of Large Language Models (LLMs), several attempts have been made to extend them to the visual domain. From having a visual assistant that could guid…
Text Quality-Based Pruning for Efficient Training of Language Models
Vasu Sharma, Karthik Padthe, Newsha Ardalani +8
In recent times training Language Models (LMs) have relied on computationally heavy training over massive datasets which makes this training process extremely laborious. In this pa…
The Unreasonable Ineffectiveness of the Deeper Layers
Andrey Gromov, Kushal Tirumala, Hassan Shapourian +2
How is knowledge stored in an LLM's weights? We study this via layer pruning: if removing a certain layer does not affect model performance in common question-answering benchmarks,…
Effective pruning of web-scale datasets based on complexity of concept clusters
Amro Abbas, Evgenia Rusak, Kushal Tirumala +3
Utilizing massive web-scale datasets has led to unprecedented performance gains in machine learning models, but also imposes outlandish compute requirements for their training. In…
Decoding Data Quality via Synthetic Corruptions: Embedding-guided Pruning of Code Data
Yu Yang, Aaditya K. Singh, Mostafa Elhoushi +7
Code datasets, often collected from diverse and uncontrolled sources such as GitHub, potentially suffer from quality issues, thereby affecting the performance and training efficien…
D4: Improving LLM Pretraining via Document De-Duplication and Diversification
Kushal Tirumala, Daniel Simig, Armen Aghajanyan +1
Over recent years, an increasing amount of compute and data has been poured into training large language models (LLMs), usually by doing one-pass learning on as many tokens as poss…
SemDeDup: Data-efficient learning at web-scale through semantic deduplication
Amro Abbas, Kushal Tirumala, Dániel Simig +2
Progress in machine learning has been driven in large part by massive increases in data. However, large web-scale datasets such as LAION are largely uncurated beyond searches for e…
Memorization Without Overfitting: Analyzing the Training Dynamics of Large Language Models
Kushal Tirumala, Aram H. Markosyan, Luke Zettlemoyer +1
Despite their wide adoption, the underlying training and memorization dynamics of very large language models is not well understood. We empirically study exact memorization in caus…
Investigating Generalization by Controlling Normalized Margin
Alexander R. Farhang, Jeremy Bernstein, Kushal Tirumala +2
Weight norm and margin participate in learning theory via the normalized margin . Since standard neural net optimizers do not control normalized margin, it is…
Dynatask: A Framework for Creating Dynamic AI Benchmark Tasks
Tristan Thrush, Kushal Tirumala, Anmol Gupta +7
We introduce Dynatask: an open source system for setting up custom NLP tasks that aims to greatly lower the technical knowledge and effort required for hosting and evaluating state…
Ensemble Machine Learning Methods for Modeling COVID19 Deaths
R. Bathwal, P. Chitta, K. Tirumala +1
Using a hybrid of machine learning and epidemiological approaches, we propose a novel data-driven approach in predicting US COVID-19 deaths at a county level. The model gives a mor…
A method for finding anomalous astronomical light curves and their analogs
Juan Rafael Martínez-Galarza, Federica Bianco, Dennis Crake +4
Our understanding of the Universe has profited from deliberate, targeted studies of known phenomena, as well as from serendipitous, unexpected discoveries, such as the discovery of…
DeepStreaks: identifying fast-moving objects in the Zwicky Transient Facility data with deep learning
Dmitry A. Duev, Ashish Mahabal, Quanzhi Ye +12
We present DeepStreaks, a convolutional-neural-network, deep-learning system designed to efficiently identify streaking fast-moving near-Earth objects that are detected in the data…
Machine Learning for the Zwicky Transient Facility
Ashish Mahabal, Umaa Rebbapragada, Richard Walters +47
The Zwicky Transient Facility is a large optical survey in multiple filters producing hundreds of thousands of transient alerts per night. We describe here various machine learning…