RecNet
Ctrl+K
About
Help
All Users
Log In
Back
Yoav Artzi
4/8/2025
Again, not really a paper. But a wild ride through optimization. Pushing a GPT-2 replication to attain a target loss in 3(!) minutes. A lot to get inspired by.
Modded-NanoGPT
Keller Jordan and co
github.com
1.17.1
Checking API service...
© 2024 RecNet. All rights reserved.
