Can you provide Grounded Chain of Thought data required for training?

[letsgofeng]

Thank you for your excellent work! and made the code, model and dataset public.

I would like to ask a question about loading GCOT data. When running the training script you provided and loading the data, I found that there is no corresponding file name "second_version.json". This does not correspond to Emma-X-GCOT provided by huggingface. How can I obtain GCOT data corresponding to the training BridgeData V2?

[Emrys-Hong]

Thank you raising the issue!

I have uploaded the dataset here: https://huggingface.co/datasets/declare-lab/Emma-X-GCOT/resolve/main/second_version.json.

Please let me know if you are able to train successfully.