NeoMME: A Single-Tower Multimodal-Native Multilingual Foundation Encoder for Efficient Fine-Tuning and Inference
Paper • 2609.01657 • Published • 25
NeoMME
Meet NeoMME: a family of 260M and 800M Multimodal-Native Multilingual Encoders
-
-
Hcompany/NeoMME-260M
Feature Extraction • 0.3B • Updated • 335 • 6 -
Hcompany/NeoMME-800M
Feature Extraction • 0.8B • Updated • 82 • 4 -
NeoMME-Retriever - Demo
🔎7Visual document retrieval and RAG with NeoMME-Retriever
-
Hcompany/NeoMME-260M-Retriever
Visual Document Retrieval • 0.3B • Updated • 330 • 4 -
Hcompany/NeoMME-260M-Retriever-ST-late
Visual Document Retrieval • 0.3B • Updated • 46 • 5 -
Hcompany/NeoMME-260M-Retriever-ST-dense
Sentence Similarity • 0.3B • Updated • 39 • 4 -
Hcompany/NeoMME-800M-Retriever
Visual Document Retrieval • 0.8B • Updated • 148 • 5 -
Hcompany/NeoMME-800M-Retriever-ST-late
Visual Document Retrieval • 0.8B • Updated • 30 • 4 -
Hcompany/NeoMME-800M-Retriever-ST-dense
Sentence Similarity • 0.8B • Updated • 68 • 3 -
Hcompany/NeoMME-260M-Pretrain-predecay-s450000
Fill-Mask • 0.3B • Updated • 25 • 3 -
Hcompany/NeoMME-800M-Pretrain-predecay-s450000
Fill-Mask • 0.8B • Updated • 23 • 3