TA6b: Efficient Architectures for Foundation Model Deployment in Resource-Constrained Systems
Tue, 27 Oct, 10:15 - 11:55 PT (UTC -7)
Location: Seascape
Session Type: Lecture
Session Chair: Eungjoo Lee, University of Arizona, United States
Track: Architectures and Implementation
Tue, 27 Oct, 10:15 - 10:40 PT (UTC -7)

TA6b.1: Uncertainty-Aware Token Pruning for Vision Transformers under Distribution Shift
Seok-Ju Hahn, Argonne National Laboratory, United States; Byeongwoo An, Begas, Republic of Korea; Eungjoo Lee, University of Arizona, United States
Tue, 27 Oct, 10:40 - 11:05 PT (UTC -7)

TA6b.2: Conformal Speculative Speculative Decoding
Payel Bhattacharjee, University of Arizona, United States; Osvaldo Simeone, Northeastern University London, United Kingdom; Ravi Tandon, University of Arizona, United States
Tue, 27 Oct, 11:05 - 11:30 PT (UTC -7)

TA6b.3: Efficient Rotations for Quantized Large Language Models
Sang Heum Yeon, Jae Jun An, Junwon Lee, Seokho Han, Jong Hwan Ko, Sungkyunkwan University, Republic of Korea
Tue, 27 Oct, 11:30 - 11:55 PT (UTC -7)

TA6b.4: Towards Efficient Collective Dynamics in Multi-Agent Systems
Minah Lee, The University of Texas at Dallas, United States