MLX LM 0.20.1 has the comparable speed as llama.cpp with flash attention

(old.reddit.com)

1 points | by tosh 9 hours ago ago

1 comments

  • 8 hours ago
    [deleted]