UME-R1 is a framework designed to endow multimodal embedding models with the flexibility to switch between discriminative and generative embeddings
Zhibin Lan
zhibinlan
AI & ML interests
None yet
Recent Activity
updated a dataset about 2 months ago
zhibinlan/OCRMT30K published a dataset about 2 months ago
zhibinlan/OCRMT30K updated a collection 3 months ago
UME-R1Organizations
None yet
LLaVE is a series of large language and vision embedding models trained on a variety of multimodal embedding datasets
-
zhibinlan/LLaVE-0.5B
Image-Text-to-Text • 0.9B • Updated • 16 • 7 -
zhibinlan/LLaVE-2B
Image-Text-to-Text • 2B • Updated • 32 • 45 -
zhibinlan/LLaVE-7B
Image-Text-to-Text • 8B • Updated • 13 • 5 -
LLaVE: Large Language and Vision Embedding Models with Hardness-Weighted Contrastive Learning
Paper • 2503.04812 • Published • 17
UME-R1 is a framework designed to endow multimodal embedding models with the flexibility to switch between discriminative and generative embeddings
LLaVE is a series of large language and vision embedding models trained on a variety of multimodal embedding datasets
-
zhibinlan/LLaVE-0.5B
Image-Text-to-Text • 0.9B • Updated • 16 • 7 -
zhibinlan/LLaVE-2B
Image-Text-to-Text • 2B • Updated • 32 • 45 -
zhibinlan/LLaVE-7B
Image-Text-to-Text • 8B • Updated • 13 • 5 -
LLaVE: Large Language and Vision Embedding Models with Hardness-Weighted Contrastive Learning
Paper • 2503.04812 • Published • 17
models 7
zhibinlan/UME-R1-7B
Image-Text-to-Text • 8B • Updated • 272 • 5
zhibinlan/UME-R1-2B
Image-Text-to-Text • 2B • Updated • 550 • 5
zhibinlan/LLaVE-7B
Image-Text-to-Text • 8B • Updated • 13 • 5
zhibinlan/LLaVE-0.5B
Image-Text-to-Text • 0.9B • Updated • 16 • 7
zhibinlan/LLaVE-2B
Image-Text-to-Text • 2B • Updated • 32 • 45
zhibinlan/AVG-LLaVA
Image-Text-to-Text • 7B • Updated • 3 • 2
zhibinlan/AVG-LLaVA-Stage3
Image-Text-to-Text • 7B • Updated • 3