Hi!
Thanks for your excellent work!
I wonder know how to reproduce the result on STAR and MSRVTT dataset, it seems that they don't have id2word.txt, word2id.txt and also the intermediate results, can I directly delete the class CriterionByModule?
Or can you give the detailed config for these two datasets?
It seems that the training code is only adapted to AGQA and it needs a lot of change for another two datasets, could you please give me the code you used for training the another two?
Thanks very much!
Hi! Thanks for your excellent work! I wonder know how to reproduce the result on STAR and MSRVTT dataset, it seems that they don't have id2word.txt, word2id.txt and also the intermediate results, can I directly delete the class CriterionByModule? Or can you give the detailed config for these two datasets?