12 papers
Into the Unknown: Benchmarking AI agents on open scientific research problems
Seungone Kim, Jiayi Geng, Young-Jun Lee +25
Can AI agents carry out research on open scientific problems? We introduce Into the Unknown, a benchmark that provides an agent with a science paper along with a research direction…
Efficient Test-Time Adaptation through Human-AI Interaction
Zora Zhiruo Wang, Apurva Gandhi, Rulin Shao +22
AI agents are trained on population-scale data to encode broad capabilities spanning those of many practitioners. Yet the artifacts they produce rarely meet the personal bar profes…
Decomposer: Learning to Decompile Symbolic Music to Programs
Yewon Kim, Apurva Gandhi, David Chung +2
Musical performance involves executing a set of high-level musical instructions, yet recovering those instructions from the performance is a challenging inverse problem. We present…
PPT-Eval: A Benchmark for Computer-Use Agents on PowerPoint Tasks
Apurva Gandhi, Vishwas Suryanarayanan, Raja Hasnain Anwar +6
Creating and editing slides is a rich, multimodal activity that is ubiquitous in professional and educational settings, making it an ideal testbed for real-world computer-use agent…
Always Learning, Always Mixing: Efficient and Simple Data Mixing All The Time
Michael Y. Hu, Apurva Gandhi, Kyunghyun Cho +2
Data mixing decides how to combine different sources or types of data and is a consequential problem throughout language model training. In pretraining, data composition is a key d…
Recursive Agent Optimization
Apurva Gandhi, Satyaki Chakraborty, Xiangjun Wang +2
We introduce Recursive Agent Optimization (RAO), a reinforcement learning approach for training recursive agents: agents that can spawn and delegate sub-tasks to new instantiations…
CodeScout: An Effective Recipe for Reinforcement Learning of Code Search Agents
Lintang Sutawika, Aditya Bharat Soni, Bharath Sriraam R R +8
A prerequisite for coding agents to perform tasks on large repositories is code localization - the identification of relevant files, classes, and functions to work on. While reposi…
Training Versatile Coding Agents in Synthetic Environments
Yiqi Zhu, Apurva Gandhi, Graham Neubig
Prior works on training software engineering agents have explored utilizing existing resources such as issues on GitHub repositories to construct software engineering tasks and cor…
Agent Data Protocol: Unifying Datasets for Diverse, Effective Fine-tuning of LLM Agents
Yueqi Song, Ketan Ramaneti, Zaid Sheikh +18
Public research results on large-scale supervised finetuning of AI agents remain relatively rare, since the collection of agent training data presents unique challenges. In this wo…
Go-Browse: Training Web Agents with Structured Exploration
Apurva Gandhi, Graham Neubig
One of the fundamental problems in digital agents is their lack of understanding of their environment. For instance, a web browsing agent may get lost in unfamiliar websites, uncer…
SkillWeaver: Web Agents can Self-Improve by Discovering and Honing Skills
Boyuan Zheng, Michael Y. Fatemi, Xiaolong Jin +8
To survive and thrive in complex environments, humans have evolved sophisticated self-improvement mechanisms through environment exploration, hierarchical abstraction of experience…
Inducing Programmatic Skills for Agentic Tasks
Zora Zhiruo Wang, Apurva Gandhi, Graham Neubig +1
To succeed in common digital tasks such as web navigation, agents must carry out a variety of specialized tasks such as searching for products or planning a travel route. To tackle…