Spaces:
Running
Running
File size: 815 Bytes
7b278b0 a96dcd0 2a17ccb | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 | ---
title: README
emoji: 📈
colorFrom: pink
colorTo: pink
sdk: static
pinned: false
---
Welcome to the LCO-Embedding project - Scaling Language-centric Omnimodal Representation Learning.
### Highlights:
- We introduce LCO-Embedding, a language-centric omnimodal representation learning method and the LCO-Embedding model families, setting a new state-of-the-art on MIEB (Massive Image Embedding Benchmark) while supporting audio and videos.
- We introduce Generation-Representation Scaling law, and connect models' generative capabilities and their representation upper bound.
- We introduce SeaDoc, a challenging visual document retrieval task in southeast Asian languages; and show that continual generative pretraining before contrastive learning raises the representation upper bound.
<!-- * Code: []() --> |