TSEmbed: Unlocking Task Scaling in Universal Multimodal Embeddings
This work is significant for researchers and practitioners working on universal multimodal embeddings, as it offers a method to overcome task conflict and improve performance, representing an incremental step towards more scalable multimodal models.
This paper addresses task conflict in Multimodal Large Language Models (MLLMs) when adapting them into universal embedding models. They developed TSEmbed, a framework that combines Mixture-of-Experts (MoE) with Low-Rank Adaptation (LoRA) to separate conflicting task objectives, achieving state-of-the-art performance on the Massive Multimodal Embedding Benchmark (MMEB) and industrial datasets.
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.