Linear Attention as State Editing: Rethinking Error and Write Addresses
LCFM Workshop at NeurIPS
Research papers and preprints. Google Scholar
2026
Linearized 2-Simplicial Attention
AXIOM Workshop on Foundations of Efficient Deep Learning at NeurIPS
Joint 3D Gravity and Magnetic Inversion via Latent Rectified-Flow Posterior Sampling
ICLR · Submitted
D. Gupta, Y. Shende, A. Das, C. Kamra, D. Gupta
Fundamental limits for weighted empirical approximations of exponentially tilted distributions
AISTATS
S. Iyer, H. Mandal, D. Gupta, R. Gupta, A. Bandyopadhyay, A. Bassamboo, S. Juneja, V. Gupta