Skip to main content ->
Ai2

Research - Papers

Explore a selection of our published work on a variety of key research challenges in AI.

Filter papers

HAICOSYSTEM: An Ecosystem for Sandboxing Safety Risks in Human-AI Interactions

Xuhui ZhouHyunwoo KimFaeze BrahmanMaarten Sap
2025
COLM

AI agents are increasingly autonomous in their interactions with human users and tools, leading to increased interactional safety risks. We present HAICOSYSTEM, a framework examining AI agent safety… 

ParaPO: Aligning Language Models to Reduce Verbatim Reproduction of Pre-training Data

Tong ChenFaeze BrahmanJiacheng LiuHanna Hajishirzi
2025
COLM

Language models (LMs) can memorize and reproduce segments from their pretraining data verbatim even in non-adversarial settings, raising concerns about copyright, plagiarism, privacy, and… 

SamudrACE: Fast and Accurate Coupled Climate Modeling with 3D Ocean and Atmosphere Emulators

James P. C. DuncanElynn WuSurya DheeshjithChristopher S. Bretherton
2025
arXiv

Traditional numerical global climate models simulate the full Earth system by exchanging boundary conditions between separate simulators of the atmosphere, ocean, sea ice, land surface, and other… 

FlexOlmo: Open Language Models for Flexible Data Use

Weijia ShiAkshita BhagiaKevin FarhatSewon Min
2025
arXiv.org

We introduce FlexOlmo, a new class of language models (LMs) that supports (1) distributed training without data sharing, where different model parameters are independently trained on closed… 

Signal and Noise: A Framework for Reducing Uncertainty in Language Model Evaluation

David HeinemanValentin HofmannIan MagnussonJesse Dodge
2025
arXiv.org

Developing large language models is expensive and involves making decisions with small experiments, typically by evaluating on large, multi-task evaluation suites. In this work, we analyze specific… 

Ai2 Scholar QA: Organized Literature Synthesis with Attribution

Amanpreet SinghJoseph Chee ChangChloe AnastasiadesSergey Feldman
2025
ACL

Retrieval-augmented generation is increasingly effective in answering scientific questions from literature, but many state-of-the-art systems are expensive and closed-source. We introduce Ai2… 

DataDecide: How to Predict Best Pretraining Data with Small Experiments

Ian MagnussonNguyen TaiBen BoginJesse Dodge
2025
ICML

Because large language models are expensive to pretrain on different datasets, using smaller-scale experiments to decide on data is crucial for reducing costs. Which benchmarks and methods of making… 

DataDecide: How to Predict Best Pretraining Data with Small Experiments

Ian MagnussonNguyen TaiBen BoginJesse Dodge
2025
arXiv.org

Because large language models are expensive to pretrain on different datasets, using smaller-scale experiments to decide on data is crucial for reducing costs. Which benchmarks and methods of making… 

Diverging Preferences: When do Annotators Disagree and do Models Know?

Michael J.Q. ZhangZhilin WangJena D. HwangValentina Pyatkin
2025
ICML

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

Aaron MuellerAtticus GeigerSarah WiegreffeYonatan Belinkov
2025
ICML

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…