NeoMME: an efficient Multimodal-native and Multilingual Encoder

In 30 Sekunden

Was passiert ist

Hugging Face released NeoMME, a 260M and 800M parameter multimodal encoder trained from scratch without separate vision or language towers.

Warum es zählt

Engineers building visual document retrieval systems need efficient models that balance speed, accuracy, and storage for production deployments.

Achtung

NeoMME was trained on only 524 billion tokens, relatively small compared to similar models, which may affect performance on out-of-distribution tasks.