RADIOv2.5 adds multi-resolution training, mosaic augmentation, and one-shot token merging, producing vision encoders that stay accurate across resolutions and improve VLM and dense-task performance over earlier baselines.
Token merging: Your ViT but faster
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2024 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
RADIOv2.5: Improved Baselines for Agglomerative Vision Foundation Models
RADIOv2.5 adds multi-resolution training, mosaic augmentation, and one-shot token merging, producing vision encoders that stay accurate across resolutions and improve VLM and dense-task performance over earlier baselines.