metadata
license: apache-2.0
tags:
- llava
datasets:
- Ejafa/ye-pop
pipeline_tag: image-text-to-text
A ViT-B/32 CLIP model trained for 4 epochs on the ye-pop dataset (491,520 images and LLaVA 1.5-generated detailed captions). Research artifact of clip-synthetic-captions. Outperforms the CLIP model trained using the original alt-texts on the DataComp benchmark suite (38 image classification and retrieval tasks).
Note: likely not directly useful as it is severely undertrained.