Tulu 2: Demystifying the Effectiveness of RLHF and Reinforcement Learning with Human Feedback

Yizhong Wang is a research scientist at the Allen Institute for AI and incoming assistant professor at the University of Washington whose work focuses on language models, agents, reasoning, and open-source AI.

Yuling Gu is a PhD student at the NYU Center for Data Science studying large language models, machine reasoning, and robust evaluation. She was previously a predoctoral researcher at the Allen Institute for AI, where she contributed to OLMo, OLMo 2, OLMo 3, TULU 3, OLMoE, and OLMES.

Researcher working on language models, agents, and retrieval-augmented generation; currently at xAI and incoming assistant professor at the University of Washington, previously a research scientist at the Allen Institute for AI.

Noah A. Smith is a computer scientist and professor at the University of Washington, where he serves as Vice Provost for Artificial Intelligence and co-directs the OLMo open language modeling effort with Ai2. His research focuses on natural language processing, machine learning, and evaluation methodology.

Research scientist at Ai2 focused on natural language processing, commonsense reasoning, long-form generation, narrative intelligence, and text-based games.

Nima Rajani is a research scientist at Ai2 whose work focuses on trustworthy, interpretable, and verifiable AI systems.

Research scientist at the Allen Institute for AI (Ai2) whose work focuses on natural language understanding and commonsense reasoning.

Researcher in natural language processing, low-resource languages, machine translation, and responsible AI; publicly listed as a PhD candidate at UC Santa Barbara and a co-author of Tulu 2.

Julian Martin Eisenschlos is a Research Scientist at Ai2. His work focuses on natural language processing, language models, and instruction tuning, including contributions to the Tulu 2 project.

Research scientist at Ai2 working on personalized language models, instruction tuning, and reinforcement learning from human feedback.

Tony Gracious completed his PhD in the Department of Computer Science and Automation at IISc Bangalore. His work includes representation learning, temporal point processes, and higher-order interaction forecasting, and he later joined Dolby's Advanced Technology Group in Bangalore.

Machine learning scientist at Ai2 working on reinforcement learning, language models, and online social systems.

Jacob Morrison is a researcher whose work spans language model post-training, alignment, and evaluation. His public research page highlights projects including Tulu 2, Tulu 3, OLMo 2, and RewardBench.

Machine learning engineer at Ai2 whose public work focuses on open language models, post-training, and evaluation.

Research engineer at Ai2 focused on post-training and data for open language models.

Research scientist at Ai2 working on large language models, machine learning with human feedback, and related topics; previously at Stanford and MIT.

Mustafa Hajij is a research scientist at Ai2 and an adjunct professor in the Department of Computer Science at the University of Southern Maine. His research spans graph machine learning, geometric learning, and applied mathematics.

Canonical link

Yizhong Wang

Yuling Gu

Bill Yuchen Lin

Noah A. Smith

Chandra Bhagavatula

Nima Rajani

Jena D. Hwang

Jesujoba Alabi

Julian Martin Eisenschlos

Tyler Scialom

Tony Gracious

Nathan Lambert

Jacob Morrison

Nicholas Ruas

Aryo Pradipta Gema

Kevin Gu

Mustafa Hajij