Research - Papers
Explore a selection of our published work on a variety of key research challenges in AI.
Diverging Preferences: When do Annotators Disagree and do Models Know?
We examine diverging preferences in human-labeled preference datasets. We develop a taxonomy of disagreement sources spanning 10 categories across four high-level classes -- task underspecification,…
MIB: A Mechanistic Interpretability Benchmark
How can we know whether new mechanistic interpretability methods achieve real improvements? In pursuit of meaningful and lasting evaluation standards, we propose MIB, a benchmark with two tracks…
SafetyAnalyst: Interpretable, transparent, and steerable safety moderation for AI behavior
The ideal AI safety moderation system would be both structurally interpretable (so its decisions can be reliably explained) and steerable (to align to safety standards and reflect a community's…
Understanding the Logic of Direct Preference Alignment through Logic
Recent direct preference alignment algorithms (DPA), such as DPO, have shown great promise in aligning large language models to human preferences. While this has motivated the development of many…
OLMoTrace: Tracing Language Model Outputs Back to Trillions of Training Tokens
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 verbatim matches…
Critical Batch Size Revisited: A Simple Empirical Approach to Large-Batch Language Model Training
The right batch size is important when training language models at scale: a large batch size is necessary for fast training, but a batch size that is too large will harm token efficiency. To…
Holodeck: Language Guided Generation of 3D Embodied AI Environments
3D simulated environments play a critical role in Embodied AI, but their creation requires expertise and extensive manual effort, restricting their diversity and scope. To miti-gate this limitation,…
What Sets the Tropical Cold Point in GSRMs During Boreal Winter? Overshooting Convection Versus Cirrus Lofting
The cold point tropopause, the minimum temperature within the tropical upper troposphere‐lower stratosphere region (UTLS), significantly impacts Earth's climate by influencing the amount of water…
Multi-Attribute Constraint Satisfaction via Language Model Rewriting
Obeying precise constraints on top of multiple external attributes is a common computational problem underlying seemingly different domains, from controlled text generation to protein engineering.…
ACE2: accurately learning subseasonal to decadal atmospheric variability and forced responses
Existing machine learning models of weather variability are not formulated to enable assessment of their response to varying external boundary conditions such as sea surface temperature and…