NeoMME: A Single-Tower Multimodal-Native Multilingual Foundation Encoder for Efficient Fine-Tuning and Inference

Open paper
Benchmark Model Rank Results
visual-document-retrieval-on-vidoreNeoMME-800M-Retriever (late interaction)–NDCG@5: 87.44
visual-document-retrieval-on-vidore-v2NeoMME-800M-Retriever (late interaction)–NDCG@5: 55.91
visual-document-retrieval-on-vidore-v3NeoMME-800M-Retriever (late interaction)–NDCG@10: 55.6NDCG@5: 53.29

Rank counts only results with a code link.