Seoul National University VLSI Lab

university

https://vlsi.snu.ac.kr/

Activity Feed Request to join this org

AI & ML interests

Efficient AI

Recent Activity

jiwonsong authored a paper about 9 hours ago

Token Sparse Attention: Efficient Long-Context Inference with Interleaved Token Selection

jiwonsong published a Space about 15 hours ago

SNU-VLSI/README

dongwonjo authored a paper 1 day ago

Retrospective Sparse Attention for Efficient Long-Context Generation

View all activity

Papers

Token Sparse Attention: Efficient Long-Context Inference with Interleaved Token Selection

LiteStage: Latency-aware Layer Skipping for Multi-stage Reasoning

View all Papers

SNU-VLSI 's Papers 7

Submitted by

Dongwon Jo

Token Sparse Attention: Efficient Long-Context Inference with Interleaved Token Selection

Seoul National University VLSI Lab

2 3

Submitted by

Jiwon Song

LiteStage: Latency-aware Layer Skipping for Multi-stage Reasoning

Seoul National University VLSI Lab

Retrospective Sparse Attention for Efficient Long-Context Generation

Seoul National University VLSI Lab

Submitted by

Jiwon Song

Reasoning Path Compression: Compressing Generation Trajectories for Efficient LLM Reasoning

Seoul National University VLSI Lab

30 2

Submitted by

Dongwon Jo

FastKV: KV Cache Compression for Fast Long-Context Processing with Token-Selective Propagation

Seoul National University VLSI Lab

29 2

Submitted by

Dongwon Jo

Mixture of Scales: Memory-Efficient Token-Adaptive Binarization for Large Language Models

Seoul National University VLSI Lab

10 1

SLEB: Streamlining LLMs through Redundancy Verification and Elimination of Transformer Blocks

Seoul National University VLSI Lab

AI & ML interests

Recent Activity

Papers

Team members 4

Token Sparse Attention: Efficient Long-Context Inference with Interleaved Token Selection

LiteStage: Latency-aware Layer Skipping for Multi-stage Reasoning

Retrospective Sparse Attention for Efficient Long-Context Generation

Reasoning Path Compression: Compressing Generation Trajectories for Efficient LLM Reasoning

FastKV: KV Cache Compression for Fast Long-Context Processing with Token-Selective Propagation

Mixture of Scales: Memory-Efficient Token-Adaptive Binarization for Large Language Models

SLEB: Streamlining LLMs through Redundancy Verification and Elimination of Transformer Blocks