Introducing BioCAP, trained on the TreeOfLife-10M dataset paired with a new TreeOfLife-10M Captions dataset, curated for this model! This new model pairs organismal images with synthetic captions generated based on species-level visual descriptions.
In this release, we publish the full caption generation and model training pipelines.