Hacker News
new
|
past
|
comments
|
ask
|
show
|
jobs
|
submit
login
kcorbitt
4 months ago
|
parent
|
context
|
favorite
| on:
Fault Tolerant Llama training
I was curious about this so I had o3 do a bit of research. Turns out 300 L40s have more compute than any supercomputer before 2013 (and arguably before 2016, depending on how you count reduced-precision FLOPs).
https://chatgpt.com/share/685dea79-26ec-8002-bd62-7ed83aedf4...
Guidelines
|
FAQ
|
Lists
|
API
|
Security
|
Legal
|
Apply to YC
|
Contact
Search:
https://chatgpt.com/share/685dea79-26ec-8002-bd62-7ed83aedf4...