Improve Multi-Modal Embedding Learning via Explicit Hard Negative Gradient Amplifying

TL;DR

This paper analyzes the role of hard negatives in contrastive learning for multi-modal models and proposes an explicit gradient amplification method to improve embedding discriminability, achieving state-of-the-art results.

Contribution

It introduces a novel method to explicitly amplify hard negative gradients, enhancing multi-modal embedding learning beyond existing hard negative mining strategies.

Findings

Achieves state-of-the-art performance on MMEB benchmark.

Top rank on MMEB leaderboard with integrated MLLM.

Demonstrates effectiveness of explicit gradient amplification.

Abstract

With the rapid advancement of multi-modal large language models (MLLMs) in recent years, the foundational Contrastive Language-Image Pretraining (CLIP) framework has been successfully extended to MLLMs, enabling more powerful and universal multi-modal embeddings for a wide range of retrieval tasks. Despite these developments, the core contrastive learning paradigm remains largely unchanged from CLIP-style models to MLLMs. Within this framework, the effective mining of hard negative samples continues to be a critical factor for enhancing performance. Prior works have introduced both offline and online strategies for hard negative mining to improve the efficiency of contrastive learning. While these approaches have led to improved multi-modal embeddings, the specific contribution of each hard negative sample to the learning process has not been thoroughly investigated. In this work, we conduct a detailed analysis of the gradients of the info-NCE loss with respect to the query, positive, and negative samples, elucidating the role of hard negatives in updating model parameters. Building upon this analysis, we propose to explicitly amplify the gradients associated with hard negative samples, thereby encouraging the model to learn more discriminative embeddings. Our multi-modal embedding model, trained with the proposed Explicit Gradient Amplifier and based on the LLaVA-OneVision-7B architecture, achieves state-of-the-art performance on the MMEB benchmark compared to previous methods utilizing the same MLLM backbone. Furthermore, when integrated with our self-developed MLLM, QQMM, our approach attains the top rank on the MMEB leaderboard. Code and models are released on https://github.com/QQ-MM/QQMM-embed.