Beyond instruction-conditioning, MoTE: Mixture of Task Experts for Multi-task Embedding Models

TL;DR

This paper introduces MoTE, a mixture of task experts approach that improves multi-task embedding models by enabling better specialization without increasing model complexity, outperforming instruction-conditioning methods.

Contribution

The paper proposes MoTE, a novel transformer block with task-specialized parameters trained via Task-Aware Contrastive Learning, enhancing embedding specialization in low-capacity models.

Findings

MoTE achieves 64% higher retrieval performance gains.

MoTE improves overall dataset performance by 43%.

Gains are obtained without changing instructions, data, or inference time.

Abstract

Dense embeddings are fundamental to modern machine learning systems, powering Retrieval-Augmented Generation (RAG), information retrieval, and representation learning. While instruction-conditioning has become the dominant approach for embedding specialization, its direct application to low-capacity models imposes fundamental representational constraints that limit the performance gains derived from specialization. In this paper, we analyze these limitations and introduce the Mixture of Task Experts (MoTE) transformer block, which leverages task-specialized parameters trained with Task-Aware Contrastive Learning (\tacl) to enhance the model ability to generate specialized embeddings. Empirical results show that MoTE achieves $64\%$ higher performance gains in retrieval datasets ($+3.27 \rightarrow +5.21$) and $43\%$ higher performance gains across all datasets ($+1.81 \rightarrow +2.60$). Critically, these gains are achieved without altering instructions, training data, inference time, or number of active parameters.