benchmarks icon indicating copy to clipboard operation
benchmarks copied to clipboard

gpt-fast

Open nsosio opened this issue 1 year ago • 1 comments

see here

nsosio avatar Nov 30 '23 23:11 nsosio

Hey @nsosio, so just clarrifying here, for PyTorch (#21), is is simply, using the HF-pytorch .bin file for llama-2 7B on fp-16/32 precision.

Where as for gpt-fast, it is this latest implementation by PyTorch Labs, right?

Anindyadeep avatar Dec 03 '23 20:12 Anindyadeep