3 ms·
I would try the Qwen models before LLaVa Do you need the embeddings to be private? Or just the photos?
by stormfather 1y ago
I would try the Qwen models before LLaVa
Do you need the embeddings to be private? Or just the photos?
- msgodel 1y agoFor photo indexing I'd run CLIP directly and save on compute, no need to use a whole language model.