09-25-2026, 08:57 AM
That matches what I was seeing. I don't think your conclusion is far off. The biggest change for me was simplifying the stack instead of adding one more optimization layer.
|
routing between local and hosted for Kimi
|
|
09-25-2026, 08:57 AM
That matches what I was seeing. I don't think your conclusion is far off. The biggest change for me was simplifying the stack instead of adding one more optimization layer.
09-25-2026, 09:00 PM
worth checking the driver first.
09-25-2026, 09:57 PM
Need a little more detail here. What model, what runtime, and roughly what context size?
|
|
« Next Oldest | Next Newest »
|