Do Visual Grounding Decoders Need Feed-Forward Networks? A Controlled Study over Frozen Vision-Language Features
arXiv:2608.15061v1 Announce Type: new Abstract: Do feed-forward networks (FFNs) in visual grounding decoders add essential computation once a pretrained vision-language model has already encoded image and language context? We compare a four-block attention-only decoder (A4), a matched four-block…