Xiangyu Li's picture

Open to Collab

Xiangyu Li

XXXXyu

·

https://xxxxyu.github.io/academic

xxxxyu

AI & ML interests

On-device and physical AI

Recent Activity

authored a paper 21 days ago

OxyGen: Unified KV Cache Management for Vision-Language-Action Models under Multi-Task Parallelism

commentedon a paper 21 days ago

Vec-LUT: Vector Table Lookup for Parallel Ultra-Low-Bit LLM Inference on Edge Devices

upvoted a paper 21 days ago

OxyGen: Unified KV Cache Management for Vision-Language-Action Models under Multi-Task Parallelism

View all activity

Organizations

None yet

upvoted a paper 21 days ago

OxyGen: Unified KV Cache Management for Vision-Language-Action Models under Multi-Task Parallelism

Paper • 2603.14371 • Published 23 days ago • 4

upvoted a paper 22 days ago

Vec-LUT: Vector Table Lookup for Parallel Ultra-Low-Bit LLM Inference on Edge Devices

Paper • 2512.06443 • Published Dec 6, 2025 • 2

upvoted a collection 3 months ago

vlut.cpp

SOTA ternary-packed versions of 1.58-bit LLMs for efficient on-device inference with vlut.cpp. • 3 items • Updated Jan 1 • 1

upvoted a paper 5 months ago

Think-at-Hard: Selective Latent Iterations to Improve Reasoning Language Models

Paper • 2511.08577 • Published Nov 11, 2025 • 110