LDong

Thank you for your reply! > There are no predefined train/test splits, but the dataset is split into 5 folds of equal size. A common procedure is to use 4...

> Hi, thanks for the interest. `qid' gives the question id within a video. So video_qid identifies a sample of the whole dataset. There is no binary question in the...

> Yes. Thank you. And there are 8 types of questions, ['DC', 'CW', 'CH', 'TN', 'TC', 'DL', 'DO', 'TP'], what do they mean respectively?

> Thanks for good library for us. As mentioned in title I want to process video in raw byte to numpy then encode it into raw byte video. I succeed...

are the LRS2 and LRS3 datasets pre-training dataset? I can not found any other description about the pre-training dataset.

> htm_aa_v1.csv should be the output of TAN, which will be used for downstream representation learning. It is not the HTM370k training set for TAN. Will you release a list...

> It seems the t5x installation fails. You may try to follow the t5x installation instructions from https://github.com/google-research/t5x. The [t5x gpu](https://github.com/google-research/t5x/blob/main/t5x/contrib/gpu/scripts_gpu/README.md) is so hard to use. After **pip install -r...