Paper Detail
Shubhang Bhatnagar, Dheeraj Baiju, Narendra Ahuja
Vision encoders for retrieval are typically trained with class-label supervision: each training pair reduces to a scalar that uniformly pushes the embedding apart or pulls it together, as if every visual attribute either differed or matched. A multimodal large language model (MLLM), shown the same pair, can articulate those attributes and use them to predict whether the images share a class. We propose SAGA, a framework that turns this language-grounded, attribute-aware perception into a training signal for the encoder itself. Specifically, we use Group Relative Policy Optimization (GRPO) to reward the MLLM for correct predictions on the vision encoder's tokens. Since correct predictions require those tokens to expose the specific attributes that differ or match between the pair, the gradient pushes the encoder to encode them, replacing the uniform pair-level scalar with attribute-resolved supervision. An auxiliary attention-distillation loss anchors the encoder's embedding to tokens the MLLM attended to, and a standard metric-learning loss shapes the embedding geometry for nearest-neighbour retrieval. The MLLM is frozen throughout and discarded at inference, matching the deployment cost of a metric-learning baseline. SAGA improves Recall@1 by 3 to 6 points over state-of-the-art baselines on CUB-200-2011, Cars-196, FGVC-Aircraft, and iNaturalist Aves on zero-shot image retrieval.
No structured notes yet. Add `summary_sections`, `why_relevant`, `claim_impact`, or `next_action` in `papers.jsonl` to enrich this view.
No ranking explanation is available yet.
No tags.
@misc{bhatnagar2026beyond,
title = {Beyond Scalar Distances: Semantic Attribute Gradients from Frozen MLLMs for Visual Embeddings},
author = {Shubhang Bhatnagar and Dheeraj Baiju and Narendra Ahuja},
year = {2026},
abstract = {Vision encoders for retrieval are typically trained with class-label supervision: each training pair reduces to a scalar that uniformly pushes the embedding apart or pulls it together, as if every visual attribute either differed or matched. A multimodal large language model (MLLM), shown the same pair, can articulate those attributes and use them to predict whether the images share a class. We propose SAGA, a framework that turns this language-grounded, attribute-aware perception into a trainin},
url = {https://huggingface.co/papers/2606.15134},
keywords = {vision encoders, multimodal large language model, class-label supervision, Group Relative Policy Optimization, attribute-aware perception, metric-learning loss, attention-distillation loss, zero-shot image retrieval, Recall@1, huggingface daily},
eprint = {2606.15134},
archiveprefix = {arXiv},
}
{}