- cross-posted to:
- [email protected]
- cross-posted to:
- [email protected]
You must log in or register to comment.
I’m surprised that the model’s benchmark performance isn’t that much worse than the base given that it’s running with 1-bit quantization. Impressive.


