MMEB-v-2

The 'MMEB-v-2' benchmark consists of 78 tasks designed to evaluate the performance of universal multimodal embedding models in handling diverse multimodal queries.

Papers using MMEB-v-2 (6)

Qwen3-VL-Embedding and Qwen3-VL-Reranker: A Unified Framework for State-of-the-Art Multimodal Retrieval and Ranking2026Embed-RL: Reinforcement Learning for Reasoning-Driven Multimodal Embeddings2026PLUME: Latent Reasoning Based Universal Multimodal Embedding2026Beyond Chain-of-Thought: Rewrite as a Universal Interface for Generative Multimodal Embeddings2026FreeRet: MLLMs as Training-Free Retrievers2025UME-R1: Exploring Reasoning-Driven Generative Multimodal Embeddings2025