ORPO resources?
#16
by
ccocks-deca
- opened
7k instances for 1.3 hours on 4 nodes of 8 x H100s
So that means ~1.5 examples per second on 32x H100?
Can you show me the code you used. I'd like to fine tune using orpo
Also, it says 39B, but the orig model is 8x22b, why?