|
@@ -60,14 +60,16 @@ In each example the `--restart` flag is used so that checkpoints are
|
|
|
periodically written during the training. The `--no_vis` flag is used to disable
|
|
|
visdom visualisations.
|
|
|
|
|
|
+As above, these examples assume the host has CUDA 10.2 installed.
|
|
|
+
|
|
|
#### DTD
|
|
|
|
|
|
The DTD dataset requires no preprocessing, so the datasets script simply creates
|
|
|
a configuration file.
|
|
|
|
|
|
```bash
|
|
|
-python datasets.py dtd <path to dtd>/images
|
|
|
-python train.py PGAN -c config_dtd.json --restart --no_vis -n dtd
|
|
|
+singularity exec pytorch_GAN_zoo.sif datasets.py dtd <path to dtd>/images
|
|
|
+singularity exec pytorch_GAN_zoo.sif --nv --app cu102 train.py PGAN -c config_dtd.json --restart --no_vis -n dtd
|
|
|
```
|
|
|
|
|
|
Where `<path to dtd>` is the path of the directory extracted from the dtd
|
|
@@ -80,8 +82,8 @@ A processed dataset will be written to a directory delcared using the `-o` flag,
|
|
|
`cifar-10` n this example.
|
|
|
|
|
|
```bash
|
|
|
-python datasets.py cifar10 <path to cifar-10> -o cifar10
|
|
|
-python train.py -c config_cifar10.json --restart --no_vis -n cifar10
|
|
|
+singularity exec pytorch_GAN_zoo.sif datasets.py cifar10 <path to cifar-10> -o cifar10
|
|
|
+singularity exec pytorch_GAN_zoo.sif --nv --app cu102 train.py -c config_cifar10.json --restart --no_vis -n cifar10
|
|
|
```
|
|
|
|
|
|
Where `<path to cifar-10>` is the path of the directory containing the pickle
|