Ash Hart
@ashxhart
Doing my bit to bring local AI to everyone | building | | Thoughts are my own.
Joined October 2014
474 Following 11.7K Followers
16GiB just for the chat cache.
For Llama 3.1 8B, that's the calculated size of a full 16-bit KV cache at 128K tokens. Weights and runtime go on top.
When someone says a model fits on a GPU, ask what context they're using.
What GPU and model are you trying to squeeze more out of? đ
Show more
