Albert Gu

Research

Albert Gu's research centres on state space models for long sequences. With Tri Dao, Christopher Ré and others, Gu was first author of HiPPO (2020), a framework for recurrent memory based on optimal polynomial projections, and of the 2021 paper on linear state-space layers that unify recurrent, convolutional and continuous-time models.⁹¹⁰

Gu's S4 paper (2021) made structured state space models efficient for long-range sequence tasks. Gu and Tri Dao were then co-first authors of Mamba (2023), which adds input-dependent selection to state space models, and of Mamba-2 (2024), which connects state space models and attention through structured state space duality.⁴⁵⁶⁷

Gu is the last-listed author of the 2025 paper on dynamic chunking for end-to-end hierarchical sequence modeling, the basis of the H-Net architecture, and of the March 2026 Mamba-3 paper, whose author list also includes Tri Dao.¹¹¹²¹³

Teaching and lab

At Carnegie Mellon, Gu leads the Goomba Lab, a research group in the Machine Learning Department with PhD and master's students. The lab's blog publishes explanatory posts alongside its papers, including two-part series on Mamba-3 and on the Raven architecture in 2026.²¹³

Cartesia

Gu is a co-founder of Cartesia and serves as its Chief Scientist, alongside the faculty appointment at Carnegie Mellon.³¹

Work and contributions

• Mamba and Mamba-2 Open-source code

  Co-first author with Tri Dao of the Mamba (2023) and Mamba-2 (2024) papers, and listed in the AUTHORS file of the Apache-2.0 repository that provides the official implementation.⁵⁶⁷⁸

• S4 (structured state space models) Published research

  First author, with Karan Goel and Christopher Ré, of the paper Efficiently Modeling Long Sequences with Structured State Spaces.⁴

Timeline

1. Aug 2020Released HiPPO, a framework for recurrent memory with optimal polynomial projections, as first author.⁹
2. Oct 26, 2021Released the linear state-space layers paper as first author.¹⁰
3. Oct 31, 2021Released the S4 paper on structured state space models as first author.⁴
4. Dec 2023Released Mamba with Tri Dao as co-first authors.⁵⁷
5. May 2024Released Mamba-2 and the structured state space duality framework with Tri Dao.⁶
6. Jul 2025Co-authored the dynamic chunking paper behind the H-Net hierarchical architecture.¹¹
7. Mar 16, 2026Co-authored the Mamba-3 paper on improved sequence modeling with state space principles.¹²

Where they work

• Carnegie Mellon University 24 people in the wiki
• Cartesia 4 people in the wiki

Related people

• Kai-Fu LeeAlso at Carnegie Mellon University
• Christopher ManningAlso at Carnegie Mellon University
• Geoffrey HintonAlso at Carnegie Mellon University
• Karen LivescuAlso at Carnegie Mellon University
• Yulia TsvetkovAlso at Carnegie Mellon University
• Beren MillidgeLanguage models · Machine-learning theory
• Fabian PedregosaMachine-learning theory · Language models
• Sébastien BubeckLanguage models · Machine-learning theory
• Stefano ErmonLanguage models · Machine-learning theory
• Michael CollinsLanguage models · Machine-learning theory