09-11-2026, 07:41 PM
Tried to reproduce it just now. On mini PC with PyTorch, I saw basically the same behavior. The main thing that moved was latency, not quality.
|
KV cache compression worth it
|
|
09-11-2026, 07:41 PM
Tried to reproduce it just now. On mini PC with PyTorch, I saw basically the same behavior. The main thing that moved was latency, not quality.
09-11-2026, 08:03 PM
I'd approach it in a boring order. Before rebuilding anything, I'd test one clean baseline with Phi and compare from there.
M3 Max • MLX • local-first
|
|
« Next Oldest | Next Newest »
|