James Womack
jcwomack.bsky.social
James Womack
@jcwomack.bsky.social
{{ .Values.interestingTagline }}
Really interesting article about storage needs of LLM training/inference, covering data gathering/preprocessing, training, and inferencing
After reposting Jeff Denworth’s claim that “no one needs a file system for AI training,” I was surprised by how many couldn’t imagine training LLMs on a supercomputer without one. To clarify, I wrote an explanation of how it’s done:

blog.glennklockwood.com/2025/02/llm-...

#HPC #AI #storage
LLM training without a parallel file system
The illustrious Jeff Denworth recently posted a hot take across social media, claiming that training large language models (LLMs) doesn't re...
blog.glennklockwood.com
February 2, 2025 at 1:57 PM
My end of year work Xmas present to myself was to spend the afternoon of my last work day playing with dev containers (containers.dev), a tool I had been wanting to try out for some time 🎁
Development containers
Development containers documentation and specification page.
containers.dev
December 20, 2024 at 9:04 PM
I had a great time at #CIUK. My first time at the event. Still processing all the input received!
December 7, 2024 at 9:42 PM
Hello, BlueSky 👋

I’m not sure how or if you fit into my life, but let’s give it a try and find out!
December 7, 2024 at 4:43 PM