We release a new ViT-G/14 CLIP model with OpenCLIP which achieves 80.1% zero-shot accuracy on ImageNet and 74.9% zero-shot image retrieval (Recall@5) on MS COCO. As of January 2023, this is the best open source CLIP model. laion.ai/blog/giant-ope… huggingface.co/laion/CLIP-ViT…
@laion_ai @nin_artificial i want to play with @DeepFloydIF so badly, let us at it!!!
@laion_ai Will it address the issue of inclusivity and representation in the general output of human characters?
@laion_ai Wow that's a huge SOTA, that's gonna lift a lot of other AIs to new heights
@laion_ai When did the tech industry first break 70%?
@laion_ai @huggingface It may NOT be able to tell a robot from a woman, but it can tell an Eames from a woman, just not a chair. I give it 31.5% accuracy.