A randomly sparsified Adam optimizer with importance-based moment pruning improves few-shot CLIP adaptation accuracy and memory efficiency over low-rank projection methods.
Reproducible scaling laws for contrastive language-image learning
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
-
Sparsity Outperforms Low-Rank Projections in Few-Shot Adaptation
A randomly sparsified Adam optimizer with importance-based moment pruning improves few-shot CLIP adaptation accuracy and memory efficiency over low-rank projection methods.