Chen Wang @wangchen615

Research Staff Member @ IBM Research Joined July 2011

Tweets

117
Followers

130
Following

415
Likes

330

Chen Wang @wangchen615

4 months ago

See a comment that I cannot agree more. “Innovation thrives when great minds come together.”

LMCache Lab @lmcache

4 months ago

See a comment that I cannot agree more. “Innovation thrives when great minds come together.”

1 10 21 1K 5

Download Image

1 0 2 125 1

🚀 We're thrilled to announce vLLM Production Stack—an open-source, Enterprise-Grade LLM inference solution that is now an official first-party ecosystem project under vLLM! Why does this matter? A handful of companies focus on LLM training, but millions of apps and businesses…

4 25 55 7K 28

Download Image

Chen Wang @wangchen615

10 months ago

I'm excited to attend 🪧 Poster Session (PS01): Climatik: Cloud Native Sustainable LLM via Power Capping - Chen Wang, IBM & Vincent Hou, Bloomberg L.P. at KubeCon + CloudNativeCon North America 2024 sched.co/1i7mv @sched

0 0 0 111 0

LMCache Lab @lmcache

11 months ago

Exciting news! Meet Kuntai Du, a key contributor to #LMCache, at #RaySummit! Don’t miss the chance to chat with him in person about the latest updates!

0 5 7 756 0

Download Image

LMCache Lab @lmcache

12 months ago

🎥 New LMCache Intro+Demo video is out! 🚀 Watch how LMCache speeds up #vLLM in document QA! By accessing KV caches 7x faster, it cuts time-to-first-token to under 4 seconds! Check out the video now! 📷 youtube.com/watch?v=rINy7m… #LLM #LMCache #vLLM #RAG #KVCaches

0 9 8 902 3

Chen Wang @wangchen615

12 months ago

LMCache: Turboboosting vLLM with 7x faster access to 100x more KV caches lmcache.github.io/2024-09-17-rel…

0 0 2 111 1

Chen Wang @wangchen615

2 years ago

#acto Fantastic work by a great group. I believe in the future the automated E2E testing tool together with llm generated operator code can further free developers from admin’s job.

Tianyin Xu @tianyin_xu

2 years ago

#acto Fantastic work by a great group. I believe in the future the automated E2E testing tool together with llm generated operator code can further free developers from admin’s job.

5 14 84 31K 7

Download Image

1 1 8 2K 0

Tianyin Xu @tianyin_xu

2 years ago

The Acto team is ready for their @sospconf presentation! @tylergu_jiawei practiced the talk at @illinoiscs Systems Reading group. Acto is a "push-button" e2e testing tool for #Kubernetes operators. It has found 50+ bugs in many #K8s #Operators. Checkout: tianyin.github.io/pub/acto.pdf

5 14 84 31K 7

Download Image

Chen Wang @wangchen615

2 years ago

#CFP: 9th Workshop on Container Technologies and Container Clouds (WoC'23) will be held with Middleware 2023 in Bologna, Italy! Submit by 09-25-2023 at woc23.hotcrp.com. Check more at sites.google.com/view/woc23/home. #WoC2023, #Middleware2023.

0 0 0 201 0

Haoran Qiu @haoran_qiu98

2 years ago

RL has been actively proposed and applied to numerous systems management tasks (e.g., scheduling, resource management, congestion control...). However, it suffers practical issues when moving to production systems. At #atc2023, we shared our experience in RL-based autoscaling.

2 4 47 4K 9

Download Image

Chen Wang @wangchen615

2 years ago

I'm excited to talk about "Optimizing Full-Stack Sustainability in a Real World Data Center: Practices and Lessons Learned" with Hua Ye & Fan Jing Meng, IBM at Open Source Summit North America 2023 sched.co/1K58A #OSSUMMIT @sched #Sustainability

0 2 4 232 0

Chen Wang @wangchen615

2 years ago

I'm excited to speak with @root_fs about "5 Steps to Deploy Cloud Native Sustainable Foundation AI Models" at Open Source Summit North America 2023 sched.co/1K5Bz #OSSUMMIT @sched

0 0 3 148 0

Chris Wright @kernelcdub

3 years ago

Curious about power consumption with your #kubernetes workloads? Check out Kepler (Kubernetes-based Efficient Power Level Exporter) and ping @root_fs and the Kepler team with your ideas. sustainable-computing.io

0 15 30 0 3

Haoran Qiu @haoran_qiu98

3 years ago

SIMPPO provides system supports that allow multiple learning-based agents to co-exist with each other in a multi-tenant platform. SIMPPO addresses the scalability and incremental training challenge through the design of the virtual agent and the application of mean-field theory.

1 1 1 0 0

Haoran Qiu @haoran_qiu98

3 years ago

Super excited to present our work “SIMPPO: A Scalable and Incremental Online Learning Framework for Serverless Resource Management” today at #SoCC22. Following our previous work FIRM (#OSDI20), SIMPPO makes one more step to pave the way for applying ML/RL in production systems.

2 1 21 0 1

project_kepler @KeplerProject

3 years ago

Sustainability Research the Cloud Native Way - Chen Wang, IBM & Huamin Chen, Red Hat at KubeCon + CloudNative North America 2022 sched.co/182JN @sched @KeplerProject