NRGPT: An Energy-based Alternative for GPT
- 2026
- ICLR 2026
Conference paper
Benjamin Hoover
Title
Research Engineer
Bio
I am a Research Engineer working to breath efficiency and interpretability into foundation models using the lens of Associative Memories and Hopfield Networks.
Publications
BOLTZMANN ROUTING FORENERGY-COMPATIBLE MIXTURE OFEXPERTS
- Bharat Runwal
- Bishwajit Saha
- et al.
- 2026
- ICLR 2026
Workshop paperTransformer Explainer: Learning LLM Transformers with Interactive Visual Explanation and Experimentation
- Aeree Cho
- Grace Kim
- et al.
- 2026
- CHI 2026
Conference paperModern Methods in Associative Memory
- Dmitry Krotov
- Benjamin Hoover
- et al.
- 2026
- AAAI 2026
TutorialDense Associative Memory with Epanechnikov Energy
- Benjamin Hoover
- Zhaoyang Shi
- et al.
- 2025
- NeurIPS 2025
Conference paperConceptAttention: Diffusion Transformers Learn Highly Interpretable Features
- Alec Helbling
- Tuna Meral
- et al.
- 2025
- ICML 2025
Conference paperModern Methods in Associative Memory
- Dmitry Krotov
- Benjamin Hoover
- et al.
- 2025
- ICML 2025
TutorialDense Associative Memory with Epanechnikov energy
- Benjamin Hoover
- Krishnakumar Balasubramanian
- et al.
- 2025
- ICLR 2025
Workshop paperDense Associative Memory Through the Lens of Random Features
- Benjamin Hoover
- Duen Horng Chau
- et al.
- 2024
- NeurIPS 2024
Conference paperEnergy Transformer
- Benjamin Hoover
- Yuchen Liang
- et al.
- 2023
- NeurIPS 2023
Conference paper
Blog posts
IBM’s Dmitry Krotov wants to crack the ‘physics’ of memory
ResearchKim MartineauHow to visualize modern AI research
ResearchKim MartineauBreaking down the AI transformer
ExplainerKim MartineauIBM RXN for Chemistry: Unveiling the grammar of the organic chemistry language
ResearchPhilippe Schwaller, Hendrik Strobelt, Teodoro Laino, and Benjamin Hoover7 minute read
Top collaborators
PR
Parikshit Ram
Research Staff Member
ND
HS
Hendrik Strobelt
Senior Research Scientist
MM
Mauro Martino
Principal Research Staff Member and Manager, Visual AI Lab