Skip to content
conftrace_
2026 EMNLP EMNLP 2026

PUMA: Post-Hoc Sparsification of Universal Multimodal Embeddings for Efficient Retrieval

AbstractUniversal multimodal embedders enable retrieval across text, image, and combined queries, but their dense representations incur high memory and inference costs. Post-hoc sparsification could reduce these costs but remains underexplored for multimodal retrieval. We introduce PUMA, a sparse autoencoder recipe that maps universal multimodal embeddings to compact sparse codes without retraining the backbone: a pretraining stage preserves dense dot-product geometry, after which the sparse encoder is fine-tuned for retrieval. We evaluate on five benchmarks covering text-to-image and composed image retrieval. On Qwen3-VL-Embedding-2B, PUMA is statistically indistinguishable from or improves over dense retrieval on four of five datasets. We further identify two failure modes of post-hoc sparsification: insufficient pre-TopK support and retrieval-misaligned active support. PUMA reduces per-vector storage by 8-16x (FP32) and runs up to 25x faster over exact dense scoring on larger candidate pools, offering a practical approach to storage- and latency-efficient multimodal retrieval.

🌉 Interdisciplinary Bridge - Artificial Intelligence and Computer Science

More from EMNLP 2026

All EMNLP 2026 papers →