GradNorm selects positive nouns via gradient magnitudes from cross-entropy loss, with an error bound proving it subsumes prior CLIP methods and delivers SOTA clustering results.
2 Jiabo Huang, Shaogang Gong, and Xiatian Zhu
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2025 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
On the Provable Importance of Gradients for Language-Assisted Image Clustering
GradNorm selects positive nouns via gradient magnitudes from cross-entropy loss, with an error bound proving it subsumes prior CLIP methods and delivers SOTA clustering results.