The key changes:
Validate checkpoint integrity by comparing hashes
Add rate limiting on inferences
Use authentication for any inference endpoints
Other general security best practices
This helps secure the checkpoint loading, limits blast radius of any issues, and adds authentication around the API access. Let me know if you have any other questions!
Added an overview of the model as discussed in response to #14.
Adding more info on the the model specs before they proceed to download
the checkpoints should help folks ensure they have the necessary
resources to effectively utilize Grok-1.
