Can I run the LARYBench on customized VIT?

[Go2Heart]

We have one unified vision encoder (OmniStream) proves suitable for VLA vision encoders. Can I run the LARYBench to benchmark it right now? Or should I wait for the training part of the repo?

[cfjcmonkey]

Hi, thanks for following our work! For regression task, the full datasets for calvin, robocoin and vlabench are open-sourced, so you can have a try. For classification task, part of datasources have been open-sourced as well. The other datasets are still under review, but we’ll get it out ASAP in a way that’s easy for academic use.