Carl Boettiger
@cboettig.bsky.social
Ecology, theory, computers. https://carlboettiger.info
Okay, but Claude's commentary while helping me set up a ray cluster on National Research Platform (NRP) is pretty entertaining:
Claude: (proposes 2 worker-config)
Me: why two workers?
Claude: (queries resources)
Claude: "HOLY MOLY! Your cluster has nodes with 57TB of memory and 13,800 CPUs!"
Claude: (proposes 2 worker-config)
Me: why two workers?
Claude: (queries resources)
Claude: "HOLY MOLY! Your cluster has nodes with 57TB of memory and 13,800 CPUs!"
September 13, 2025 at 2:52 AM
Okay, but Claude's commentary while helping me set up a ray cluster on National Research Platform (NRP) is pretty entertaining:
Claude: (proposes 2 worker-config)
Me: why two workers?
Claude: (queries resources)
Claude: "HOLY MOLY! Your cluster has nodes with 57TB of memory and 13,800 CPUs!"
Claude: (proposes 2 worker-config)
Me: why two workers?
Claude: (queries resources)
Claude: "HOLY MOLY! Your cluster has nodes with 57TB of memory and 13,800 CPUs!"
of course the natural follow-up question is 0.1 + 0.2
all abstractions are leaky.
all abstractions are leaky.
June 23, 2025 at 10:07 PM
of course the natural follow-up question is 0.1 + 0.2
all abstractions are leaky.
all abstractions are leaky.
Following on arxiv.org/pdf/2502.00873, I think this speaks volumes. much better to learn the right abstractions. And where the right abstractions remain more contested, things will probably remain dicey for a while.
June 23, 2025 at 9:56 PM
Following on arxiv.org/pdf/2502.00873, I think this speaks volumes. much better to learn the right abstractions. And where the right abstractions remain more contested, things will probably remain dicey for a while.
Final projects always surface the best questions. Just now from one team: "how do I finish this git merge?", with screenshot below.
How do I explain ...
How do I explain ...
December 9, 2024 at 9:15 PM
Final projects always surface the best questions. Just now from one team: "how do I finish this git merge?", with screenshot below.
How do I explain ...
How do I explain ...
As students will be using LLMs regardless, I figured I may as well try to teach with them. To mitigate some concerns of privacy, energy use, equity, and open source, we're using open-weight models in our #Jupyter Hub, thanks to a NAIRR grant and the amazing National Research Platform compute.
October 8, 2024 at 4:56 PM
As students will be using LLMs regardless, I figured I may as well try to teach with them. To mitigate some concerns of privacy, energy use, equity, and open source, we're using open-weight models in our #Jupyter Hub, thanks to a NAIRR grant and the amazing National Research Platform compute.