2 ms·
I'd love to know how they are doing this! Are they running an on-device CLIP embedding model of some sort? Could they even be shipping a multi-modal LLM like Ph
by simonw 2y ago
I'd love to know how they are doing this! Are they running an on-device CLIP embedding model of some sort? Could they even be shipping a multi-modal LLM like Phi-3 Vision?
- Firmwarrior 2y agoOneDrive and the local Microsoft photos app have been doing it for a decade already.