Hi, thanks for the work and the repo.
I have a doubt about the data. Basically, I want to use it to train a model. However, after downloading the zipped files from Hugginface it seems the images are not included in the dataset. Is it correct? I was expecting, for example, videos annotated with the CoT traces.
Do we need to 'outsource' this ourselves using the instructions in 'build_data'? Will the resulting dataset be the same as the one you used in your paper?
Thank you very much
Hi, thanks for the work and the repo.
I have a doubt about the data. Basically, I want to use it to train a model. However, after downloading the zipped files from Hugginface it seems the images are not included in the dataset. Is it correct? I was expecting, for example, videos annotated with the CoT traces.
Do we need to 'outsource' this ourselves using the instructions in 'build_data'? Will the resulting dataset be the same as the one you used in your paper?
Thank you very much