Skip to content
Better HN
Top
Best
Ask
Show
New
Jobs
Search
⌘K
0 points
ElectricalUnion
1mo ago
0 comments
Save
Share
You need the rest of the ram for the context. If you don't want to end up with a toy context or quantized lossy context, is pretty easy to end up having to spend up 50+GB just for the KV cache, per simutaneous inference slot.
0 comments
No comments yet.