Výsledky vyhledávání

Report

vTensor: Flexible Virtual Tensor Management for Efficient LLM Serving

Autor: Xu, Jiale, Zhang, Rui, Guo, Cong, Hu, Weiming, Liu, Zihan, Wu, Feiyang, Feng, Yu, Sun, Shixuan, Shao, Changxu, Guo, Yuhong, Zhao, Junping, Zhang, Ke, Guo, Minyi, Leng, Jingwen

Large Language Models (LLMs) are widely used across various domains, processing millions of daily requests. This surge in demand poses significant challenges in optimizing throughput and latency while keeping costs manageable. The Key-Value (KV) cach

Externí odkaz: http://arxiv.org/abs/2407.15309

Zobrazit plný text záznamu

Report

Learn to Teach: Improve Sample Efficiency in Teacher-student Learning for Sim-to-Real Transfer

Autor: Wu, Feiyang, Gu, Zhaoyuan, Zhao, Ye, Wu, Anqi

Simulation-to-reality (sim-to-real) transfer is a fundamental problem for robot learning. Domain Randomization, which adds randomization during training, is a powerful technique that effectively addresses the sim-to-real gap. However, the noise in ob

Externí odkaz: http://arxiv.org/abs/2402.06783

Zobrazit plný text záznamu

Report

Infer and Adapt: Bipedal Locomotion Reward Learning from Demonstrations via Inverse Reinforcement Learning

Autor: Wu, Feiyang, Gu, Zhaoyuan, Wu, Hanran, Wu, Anqi, Zhao, Ye

Enabling bipedal walking robots to learn how to maneuver over highly uneven, dynamically changing terrains is challenging due to the complexity of robot dynamics and interacted environments. Recent advancements in learning from demonstrations have sh

Externí odkaz: http://arxiv.org/abs/2309.16074

Zobrazit plný text záznamu

Report

MovieChat: From Dense Token to Sparse Memory for Long Video Understanding

Autor: Song, Enxin, Chai, Wenhao, Wang, Guanhong, Zhang, Yucheng, Zhou, Haoyang, Wu, Feiyang, Chi, Haozhe, Guo, Xun, Ye, Tian, Zhang, Yanting, Lu, Yan, Hwang, Jenq-Neng, Wang, Gaoang

Recently, integrating video foundation models and large language models to build a video understanding system can overcome the limitations of specific pre-defined vision tasks. Yet, existing systems can only handle videos with very few frames. For lo

Externí odkaz: http://arxiv.org/abs/2307.16449

Zobrazit plný text záznamu

Report

Inverse Reinforcement Learning with the Average Reward Criterion

Autor: Wu, Feiyang, Ke, Jingyang, Wu, Anqi

We study the problem of Inverse Reinforcement Learning (IRL) with an average-reward criterion. The goal is to recover an unknown policy and a reward function when the agent only has samples of states and actions from an experienced agent. Previous IR

Externí odkaz: http://arxiv.org/abs/2305.14608

Zobrazit plný text záznamu

Report

Stochastic first-order methods for average-reward Markov decision processes

Autor: Li, Tianjiao, Wu, Feiyang, Lan, Guanghui

We study average-reward Markov decision processes (AMDPs) and develop novel first-order methods with strong theoretical guarantees for both policy optimization and policy evaluation. Compared with intensive research efforts in finite sample analysis

Externí odkaz: http://arxiv.org/abs/2205.05800

Zobrazit plný text záznamu