10-17-2026, 04:04 AM
I tested this tonight. On M3 Max with PyTorch, I saw basically the same behavior. The main thing that moved was latency, not quality.
Inference
›
Security, Privacy & Safety
›
Local Privacy
privacy wins of local models: early impressions
privacy wins of local models: early impressions
|
privacy wins of local models: early impressions
|
|
« Next Oldest | Next Newest »
|
| Messages In This Thread |
|
privacy wins of local models: early impressions - by mlxgarden - 10-16-2026, 04:00 PM
Re: privacy wins of local models: early impressions - by cachebench - 10-17-2026, 04:04 AM
Re: privacy wins of local models: early impressions - by kernelgrid - 10-17-2026, 08:30 AM
Re: privacy wins of local models: early impressions - by audioriver - 10-17-2026, 08:53 AM
Re: privacy wins of local models: early impressions - by mlxgarden - 10-17-2026, 05:20 PM
Re: privacy wins of local models: early impressions - by deltagrid - 10-17-2026, 05:37 PM
Re: privacy wins of local models: early impressions - by rocmriley - 10-17-2026, 06:38 PM
|
Users browsing this thread: 1 Guest(s)