NVIDIA
NVIDIA
Cosmos-Embed1
Model
NVIDIA
NVIDIA
Cosmos-Embed1

Cosmos-Embed1 is a joint video-text embedder tailored for physical AI.

NameSizeUpdatedActions
config.json
526 BMay 9, 2026 UTC
configuration_embed1.py
2.82 KBMay 9, 2026 UTC
export_config.yaml
5.25 KBMay 9, 2026 UTC
model-00001-of-00010.safetensors
493.52 MBMay 9, 2026 UTC
model-00002-of-00010.safetensors
470.97 MBMay 9, 2026 UTC
model-00003-of-00010.safetensors
481.62 MBMay 9, 2026 UTC
model-00004-of-00010.safetensors
481.62 MBMay 9, 2026 UTC
model-00005-of-00010.safetensors
481.62 MBMay 9, 2026 UTC
model-00006-of-00010.safetensors
481.62 MBMay 9, 2026 UTC
model-00007-of-00010.safetensors
481.62 MBMay 9, 2026 UTC
model-00008-of-00010.safetensors
481.62 MBMay 9, 2026 UTC
model-00009-of-00010.safetensors
481.62 MBMay 9, 2026 UTC
model-00010-of-00010.safetensors
234.34 MBMay 9, 2026 UTC
model.safetensors.index.json
71 KBMay 9, 2026 UTC
modeling_embed1.py
10.16 KBMay 9, 2026 UTC
modeling_outputs.py
3.1 KBMay 9, 2026 UTC
modeling_qformer.py
44.51 KBMay 9, 2026 UTC
modeling_utils.py
5.63 KBMay 9, 2026 UTC
modeling_vit.py
26.24 KBMay 9, 2026 UTC
preprocessing_embed1.py
4.86 KBMay 9, 2026 UTC
tokenizer_config.json
48 BMay 9, 2026 UTC
tokenizer.json
455.14 KBMay 9, 2026 UTC
vocab.txt
226.08 KBMay 9, 2026 UTC