search

dyelax / Adversarial_Video_Generation

A TensorFlow Implementation of "Deep Multi-Scale Video Prediction Beyond Mean Square Error" by Mathieu, Couprie & LeCun.
MIT License
736 stars 184 forks source link

generate the right module required for training network #7

Open hao19921108 opened 7 years ago

hao19921108 commented 7 years ago

Hello,

I have successfully completed your first three steps and am stuck on step 4 where I can train/test my own data. The data I am producing is not a .cpkt format and is a .npz format. What is the best way to take my .npz data and convert it into a .cpkt format? Is there something I am not doing?

Thank you, Hao

dyelax commented 7 years ago

Hi Hao,

The argument that you pass to the -l flag should be a previously-saved model. TensorFlow should produce a .ckpt file when tf.train.Saver.save() is called. Where are you getting your .npz file?

Matt

stalgiag commented 7 years ago

I also have this problem. A bit confused. I started with about a thousand sequential .jpgs which then became .npz files when I processed the data in Step 3. I now have a ton of .npz files but a bit confused about how I train those into a model. I have tried passing their save directory as the -t argument but that doesn't seem to be what it wants. The wording regarding training on our own data on Step 4 is a bit confusing.

Sorry for my ignorance! Thank you for your work!

dyelax commented 7 years ago

Which version of TensorFlow are you using? This only works for up to 0.12 currently.

stalgiag commented 7 years ago

I fixed my problem. Had a slight issue with how I set up the directories. Thanks for the code and help!