Jun-Yan Zhu

It happened sometimes for DCGAN training. I would recommend that you use the latest GAN models as backend such as [progressive-gans](https://github.com/tkarras/progressive_growing_of_gans).

DCGAN is pretty old. If you just need to train a GAN model, you can probably try more recent work such as [GAN-stability](https://github.com/LMescheder/GAN_stability) or [style-gan ](https://github.com/NVlabs/stylegan).

Have you tried the command suggested by README `THEANO_FLAGS='device=gpu0, floatX=float32, nvcc.fastmath=True' python iGAN_main.py --model_name outdoor_64`

The code was developed in 2015-16 and Theano is no longer supported. I can run the code with this theano [version](https://github.com/Newmu/dcgan_code/issues/5). I plan to release a PyTorch version in the...

What is the command that you are running? Could you import the library by yourself? This [post](https://github.com/junyanz/iGAN/issues/22) might be relevant.

You can try `from .lib import activations`. See more details on library import [here](https://stackoverflow.com/questions/8953844/import-module-from-subfolder?utm_medium=organic&utm_source=google_rich_qa&utm_campaign=google_rich_qa). Try something simple: - download the code - type `python`. - in the python environment, type...

The code is only tested with an older Theano version as well as old GPUs and CUDNN. Could you run other Theano demo code such as this [one](https://github.com/Newmu/Theano-Tutorials)?

The code was developed around 2015. I am pretty sure it doesn't work with cudnn 5.1. See here for the Theano [version](https://github.com/Newmu/dcgan_code/issues/5).

It will run very slowly on a CPU. We haven't tested the code with CPU mode.

Could you set n_vis as 9 or 16?