nanogpt
https://github.com/karpathy/nanogpt
Triage Issues!
When you volunteer to triage issues, you'll receive an email each day with a link to an open issue that needs help in this project. You'll also receive instructions on how to triage issues.
Triage Docs!
Receive a documented method or class from your favorite GitHub repos in your inbox every day. If you're really pro, receive undocumented methods or classes and supercharge your commit history.
not yet supported1 Subscribers
Add a CodeTriage badge to nanogpt
Help out
- Issues
- When running the program on 8 cards across two machines, I encounter a ChildFailedError.
- Python 3.11+ not yet supported for torch.compile
- Run on V100 16GB GPU?
- Possible to use it to summarize pages and generate keywords?
- Changes to support packaging
- The training data may cross boundary between different articles?
- Add an option to align data loads to block_size boundaries.
- How fast is it compared to megatron, deepspeed and something like that?
- Why in the cross_entropy, set ignore_index be -1?
- Running the code
- Docs
- not yet supported