WA6a.3

Zero-Shot Inference-time Structured Approximation of Softmax Attention

Session:
WA6a: Compression and sparsity in deep learning optimization Lecture
Track:
Adaptive Systems, Machine Learning, and Data Analytics
Location:
Room 6
Presentation Time:
Wed, 28 Oct, 09:05 - 09:30 PT (UTC -7)
Session Chair:
Laura Balzano, University of Michigan, United States
Presentation
Discussion
Resources
No resources available.
Session WA6a
WA6a.1: Gradient-based dimension reduction for Bayesian models
Ricardo Baptista, University of Toronto, Canada
WA6a.2: Compressible Dynamics in Deep Overparameterized Low-Rank Learning & Adaptation
Can Yaras, Laura Balzano, University of Michigan, United States; Peng Wang, University of Macau, United States; Qing Qu, University of Michigan, United States
WA6a.3: Zero-Shot Inference-time Structured Approximation of Softmax Attention
WA6a.4: Neuron Pursuit: A Parsimony-Driven Iterative Greedy Algorithm for Building Neural Networks
Akshay Kumar, Jarvis Haupt, University of Minnesota, United States
Contacts