No exp folder

[gsig]

No exp folder

[gsig]

The experiment templates are under baseline_exp/, you can then make a dir called exp/ for your own experiments. Hope that helps!

…

On Tue, Dec 24, 2019, 11:15 AM Shadman Sakib ***@***.***> wrote: One of the steps to train the network says to duplicate and edit one of the experiment files under exp/ with appropriate parameters. I'm not sure where exp/ and the experiment files being referred to are. — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#28?email_source=notifications&email_token=AA3OMNK2DJN4UN6EGIFZ3BTQ2HVMPA5CNFSM4J65OEO2YY3PNVWWK3TUL52HS4DFUVEXG43VMWVGG33NNVSW45C7NFSM4ICPW4RA>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AA3OMNPOUAIMHKIRYBR7FMLQ2HVMPANCNFSM4J65OEOQ> .

[shadsakib]

Thank you.

One more thing, is it possible to test the models on random video clips? It's not clear to me how I can test it and I can't seem to find any instructions either.

[gsig]

There are experiment files for running the pretrained models on the test sets under baseline_exp/. The "test set" is determined by the *_test.csv file (see datasets/ for details on the data loaders, so it would be easiest to just put your data on that format and choose your csv file in the config. For example, adding a row to Charades_v1_test.csv with a video name and labels would test on that video. Typically I create many different csv files to test on different types of videos or labels etc Hope that helps!

…

On Tue, Dec 24, 2019, 2:34 PM Shadman Sakib ***@***.***> wrote: Thank you. One more thing, is it possible to test the models on random video clips? It's not clear to me how I can test it and I can't seem to find any instructions either. — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#28?email_source=notifications&email_token=AA3OMNIT5ZPMTMMQCMRSIU3Q2IMWFA5CNFSM4J65OEO2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEHTJEVI#issuecomment-568758869>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AA3OMNNAWZ2OCTXNDTLTN23Q2IMWFANCNFSM4J65OEOQ> .

[shadsakib]

Thanks a lot. That was helpful.