MobileCLIP2 combines DFN-trained teachers, a fine-tuned CoCa captioner, and new 5-stage FastViT variants to set state-of-the-art ImageNet-1k zero-shot accuracy at low latency.
Table 15:Summary of pretraining datasets
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
MobileCLIP2: Improving Multi-Modal Reinforced Training
MobileCLIP2 combines DFN-trained teachers, a fine-tuned CoCa captioner, and new 5-stage FastViT variants to set state-of-the-art ImageNet-1k zero-shot accuracy at low latency.