About the evaluation metric & script

[Inn1917]

Great work!
We recently will compare different embodied-AI VLM on various tasks, in order to choose the best one for our VLA training. Could you please provide us with your evaluation script on the datasets you used? Thanks!

[LeiZhouMi]

Thanks for your interest!

We’ve now released and updated the evaluation scripts in this repo, including those for embodied-AI and autonomous driving benchmarks. We also provide the code for computing the metrics reported in the table from the evaluation outputs (json/jsonl files).

Please check the latest version of the repo for details.