Heykuki News
Top
New
Best
Ask
Show
Jobs
Toggle theme
Login
Top
New
Best
Ask
Show
Jobs
llm.c: multi-GPU, bfloat16, flash attention, ~7% faster than PyTorch
twitter.com
121 points
tosh
2 years ago
10 comments
Loading...
llm.c: multi-GPU, bfloat16, flash attention, ~7% faster than PyTorch | Heykuki News