Skip to content

Latest commit

 

History

History

Folders and files

NameName
Last commit message
Last commit date

parent directory

..
 
 
 
 
 
 
 
 
 
 

README.md

Training and evaluating

Training

python main.py

positional arguments:

mode                  Choose train or eval
data_file             Path to HDF5 containing the data

optional arguments:

-h, --help            show this help message and exit
--model_name MODEL_NAME
                    Model name. Choose simple_colorful or colorful
--training_mode TRAINING_MODE
                    Training mode. Choose in_memory to load all the data
                    in memory and train.Choose on_demand to load batches
                    from disk at each step
--batch_size BATCH_SIZE
                    Batch size
--n_batch_per_epoch N_BATCH_PER_EPOCH
                    Number of training epochs
--nb_epoch NB_EPOCH   Number of batches per epoch
--nb_resblocks NB_RESBLOCKS
                    Number of residual blocks for simple model
--nb_neighbors NB_NEIGHBORS
                    Number of nearest neighbors for soft encoding
--epoch EPOCH         Epoch at which weights were saved for evaluation
--T T                 Temperature to change color balance in evaluation
                    phase.If T = 1: desaturated. If T~0 vivid

Example:

python main.py train ../../data/processed/CelebA_32_data.h5

Expected outputs:

  • Create a directory in Colorful/models where weights are saved
  • Create a directory in Colorful/figures where figures are saved
  • Plot examples of the colorization results at each epoch
  • Plot model architecture.
  • Save model weights every few epochs

Evaluating

positional arguments:

mode                  Choose train or eval
data_file             Path to HDF5 containing the data

optional arguments:

-h, --help            show this help message and exit
--model_name MODEL_NAME
                    Model name. Choose simple_colorful or colorful
--training_mode TRAINING_MODE
                    Training mode. Choose in_memory to load all the data
                    in memory and train.Choose on_demand to load batches
                    from disk at each step
--batch_size BATCH_SIZE
                    Batch size
--n_batch_per_epoch N_BATCH_PER_EPOCH
                    Number of training epochs
--nb_epoch NB_EPOCH   Number of batches per epoch
--nb_resblocks NB_RESBLOCKS
                    Number of residual blocks for simple model
--nb_neighbors NB_NEIGHBORS
                    Number of nearest neighbors for soft encoding
--epoch EPOCH         Epoch at which weights were saved for evaluation
--T T                 Temperature to change color balance in evaluation
                    phase.If T = 1: desaturated. If T~0 vivid

Example:

python main.py eval ../../data/processed/CelebA_64_data.h5 --epoch 10

Expected outputs:

Randomly sample images from the validation set and plot the color, colorized and b&w version.