Github has a 100mb file limit resulting in the rockyou dataset being unable to push to repo.
https://help.github.com/en/enterprise/2.18/user/github/managing-large-files/conditions-for-large-files

A work around solution is to either not commit my dataset or to use an alternative.
This isn't an immediate issue, as I host versioning on a local server but I will combine it with bitbucket offsite as redundancy.

Issue unresolved until github workaround found.

SOLVED EDIT: Dataset Cloud hosted and cached locally

Currently you're only able to use the rock_you dataset predefined by the DatasetPipeline. It would be nice to see the ability to load a dataset based on a specified file such as a rock_you source hosted locally.

First , in the original paper , Z is a 128-dimensional vector generated by tf.random.normal (without regard to the batchsize ), then transmit it into the full connection layer . The input dim of the full connection layer is 128 and the output dim is 120*seq_len .

But in youe code , Z is [2,1,32] ?(I dont know why) , and the full connection layer in your code is also different from the original code because the use of keras . In your code , The input dim of the full connection layer is 128*seq_len and the output dim is 128 , whitch is totally different from the core idea of the paper . I am a new learner in PassGAN , I wonder if there is some different ideas or you have taken account of other conditions in your source code?

Looking forward to your reply

Hi, Rachel!
Please tell me, how I can continue training the model from the checkpoint, when I have stopped the training process?

Hi. I'd like to test your software, however I am not understanding the instructions too well. Is there any chance of making the Read Me clearer for noobs like myself? I would really appreciate that very much :)

Hey Rachel, the work you did here is amazing and I am trying to get the code working but Im getting multiple errors. Do you know if the code is working out of the box? Dont know where to start and if you have time to help me out. To be quick I tried using a conda environment with all the requirements installed (from requirements.txt) and another environment with an upgraded version of the libraries. No one worked after trying multiple versions (on libraries which failed) and tweaking the code avoiding warnings. Any recommendation? I dropped TF for pytorch a while ago and Im a bit rusty to say the least :P. Hope u have a nice day and wish u are able to reply :)

Alex