Thanks to the authors for sharing the training dataset with the community.
As several others have already pointed out, releasing the training set alone may not be sufficient for full reproducibility, as there remains some ambiguity regarding the evaluation set. Has the evaluation set also been released, or was it randomly sampled from the 136K-video dataset that was made publicly available?
If the latter is the case, would it be possible to release the scene indices (or the corresponding evaluation split)? This would greatly improve reproducibility and enable fair comparisons with your work, as well as with future research building upon it.
Best,
Thanks to the authors for sharing the training dataset with the community.
As several others have already pointed out, releasing the training set alone may not be sufficient for full reproducibility, as there remains some ambiguity regarding the evaluation set. Has the evaluation set also been released, or was it randomly sampled from the 136K-video dataset that was made publicly available?
If the latter is the case, would it be possible to release the scene indices (or the corresponding evaluation split)? This would greatly improve reproducibility and enable fair comparisons with your work, as well as with future research building upon it.
Best,