Wes Brown @w3sbrown

Security Practitioner, Reverse Engineerer, Deaf speaker of ASL, and all-around geek Prescott Joined March 2009

Tweets

343
Followers

244
Following

62
Likes

5

Wes Brown @w3sbrown

5 days ago

The recording of my and Deok Filho's first webinar, How To Measure and Optimize AI Infrastructure for Large-Scale Training, is now available on-demand! It covers what my team and I wrote in our white-paper, Achieving 20% Higher MFU and 10x Reliability on Thousand-GPU Clusters. It…

0 0 1 98 0

Wes Brown @w3sbrown

a week ago

I wrote a blog post that's a TLDR; for the massive technical whitepaper that my team and I worked on for @CoreWeave covering training #AI at scale. This will allow all y'all to see if you want to invest time in the upcoming webinar (on August 28th, at 2pm EST), or reading the…

0 1 5 316 1

Download Image

Wes Brown @w3sbrown

3 weeks ago

This was the culmination of months of work by my team and I at #CoreWeave aka $CRWV. It documents the optimizations and thinking that went into training our 30b foundational model Ixchel. I hope that y’all find it interesting and useful. I know that I am proud of my team for…

0 2 10 1K 3

Wes Brown @w3sbrown

5 months ago

Your company just IPO'd (or even better, is about to IPO) —congratulations! 🎉 But now comes the tricky part: converting Incentive Stock Options (ISOs) into real-world wealth. Based on my personal experience and weeks of research and strategizing my own situation, I came up with…

0 1 2 132 0

CoreWeave @CoreWeave

5 months ago

Today, we had the incredible honor of ringing the @Nasdaq Opening Bell, marking @CoreWeave's debut as a publicly traded company. We’re just getting started, and we can’t wait for what’s ahead! #CRWV

7 24 145 31K 6

Download Image

CoreWeave @CoreWeave

6 months ago

NVIDIA GB200 NVL72 is here and working to take generative and agentic AI to the next level. This platform connects 36 Grace CPUs and 72 Blackwell GPUs in a rack-scale design. We're excited to share our Llama 70B performance of over 860,000 tokens/sec on a full NVL72. We can’t…

1 16 53 5K 6

Wes Brown @w3sbrown

a year ago

Co-wrote a blogpost about MLOps best practices for AI training cluster, for @CoreWeave. It should have useful information that helps contextualize the challenges of running training clusters at scale. coreweave.com/blog/mlops-bes… #MachineLearning #mlops #aitraining #GenerativeAI

0 0 2 181 1

Wes Brown @w3sbrown

5 years ago

A comparison of #Forth VM’s performance on #AppleM1 vs x86-64 machines, and deep dive into stack machines and optimizations for #programmers who are interested in such esoterica. TLDR — single-threaded: M1 - 1,241 ops/µs i7-10700k - 1,251 ops/µs wesbrown18.medium.com/a-forth-vms-pe…

0 0 0 0 0

Hex-Rays SA @HexRaysSA

5 years ago

IDA Pro for arm64 is coming! We have ported all of IDA to run natively on Apple Silicon, and it will be available in IDA 7.6. Initial analysis shows that the hype is real 🤩: hex-rays.com/blog/ida-pro-o… #AppleSilicon #M1 #macmini #BigSur