Hello,
After reading the Cosmos paper on arXiv, I had a few questions and wanted to ask for clarification.
In the paper, it is mentioned that the GitHub repository NVIDIA/cosmos
provides the pre-trained WFM. However, when describing the pre-trained WFM, it refers to it as the WFM from nvidia-cosmos/cosmos-predict1
.
Are the models in NVIDIA/cosmos
and nvidia-cosmos/cosmos-predict1
the same?
If so, I’m curious to know why they were separated into cosmos-predict1
.
Thank you in advance!