New Method Uses Batch-Conditioned Semantic Anchors for Transductive Adaptation of Vision-Language Models
A new arXiv paper proposes batch-conditioned semantic anchors to improve transductive adaptation of vision-language models, refining predictions on unlabeled batches at test time while addressing a key trade-off in current methods.
Read the full story at arXiv cs.CV ↗
Timeline · 1 report
- 2026-11-06 05:00 · arXiv cs.CV
Batch-Conditioned Semantic Anchors for Robust Transductive Adaptation of Vision--Language Models