TSEmbed: Unlocking Task Scaling in Universal Multimodal Embeddings
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866912943300411392 |
|---|---|
| author | Wu, Yebo Liu, Feng Xie, Ziwei Liu, Zhiyuan Zhang, Changwang Wang, Jun Li, Li |
| author_facet | Wu, Yebo Liu, Feng Xie, Ziwei Liu, Zhiyuan Zhang, Changwang Wang, Jun Li, Li |
| contents | Despite the exceptional reasoning capabilities of Multimodal Large Language Models (MLLMs), their adaptation into universal embedding models is significantly impeded by task conflict. To address this, we propose TSEmbed, a universal multimodal embedding framework that synergizes Mixture-of-Experts (MoE) with Low-Rank Adaptation (LoRA) to explicitly disentangle conflicting task objectives. Moreover, we introduce Expert-Aware Negative Sampling (EANS), a novel strategy that leverages expert routing distributions as an intrinsic proxy for semantic similarity. By dynamically prioritizing informative hard negatives that share expert activation patterns with the query, EANS effectively sharpens the model's discriminative power and refines embedding boundaries. To ensure training stability, we further devise a two-stage learning paradigm that solidifies expert specialization before optimizing representations via EANS. TSEmbed achieves state-of-the-art performance on both the Massive Multimodal Embedding Benchmark (MMEB) and real-world industrial production datasets, laying a foundation for task-level scaling in universal multimodal embeddings. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2603_04772 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | TSEmbed: Unlocking Task Scaling in Universal Multimodal Embeddings Wu, Yebo Liu, Feng Xie, Ziwei Liu, Zhiyuan Zhang, Changwang Wang, Jun Li, Li Computation and Language Artificial Intelligence Despite the exceptional reasoning capabilities of Multimodal Large Language Models (MLLMs), their adaptation into universal embedding models is significantly impeded by task conflict. To address this, we propose TSEmbed, a universal multimodal embedding framework that synergizes Mixture-of-Experts (MoE) with Low-Rank Adaptation (LoRA) to explicitly disentangle conflicting task objectives. Moreover, we introduce Expert-Aware Negative Sampling (EANS), a novel strategy that leverages expert routing distributions as an intrinsic proxy for semantic similarity. By dynamically prioritizing informative hard negatives that share expert activation patterns with the query, EANS effectively sharpens the model's discriminative power and refines embedding boundaries. To ensure training stability, we further devise a two-stage learning paradigm that solidifies expert specialization before optimizing representations via EANS. TSEmbed achieves state-of-the-art performance on both the Massive Multimodal Embedding Benchmark (MMEB) and real-world industrial production datasets, laying a foundation for task-level scaling in universal multimodal embeddings. |
| title | TSEmbed: Unlocking Task Scaling in Universal Multimodal Embeddings |
| topic | Computation and Language Artificial Intelligence |
| url | https://arxiv.org/abs/2603.04772 |