Moving more options to config and some clean up

bob/learn/pytorch/config/mccnn/wmca_mccnn.py

@@ -80,8 +80,11 @@ learning_rate=0.0001
 seed = 3
 output_dir = 'training_mccn'
 use_gpu = False
+adapted_layers = 'conv1-block1'
+adapt_reference_channel = False
 verbose = 2
 
+
 #==============================================================================
 """
 Note: Running in GPU

bob/learn/pytorch/scripts/train_mccnn.py

@@ -7,22 +7,24 @@
 Usage:
   %(prog)s <configuration> 
             [--model=<string>] [--batch-size=<int>] [--epochs=<int>] 
-            [--learning-rate=<float>] [--seed=<int>] 
+            [--learning-rate=<float>] [--adapted-layers=<string>] [--adapt-reference-channel] [--seed=<int>] 
             [--output-dir=<path>] [--use-gpu] [--verbose ...]
 
 Arguments:
   <configuration>  A configuration file, defining the dataset and the network
 
 Options:
-  -h, --help                    Shows this help message and exits
-      --model=<string>          Filename of the model to load (if any). 
-      --batch-size=<int>        Batch size [default: 64]
-      --epochs=<int>            Number of training epochs [default: 20]
-      --learning-rate=<float>   Learning rate [default: 0.01]
-  -S, --seed=<int>              The random seed [default: 3] 
-  -o, --output-dir=<path>       Dir to save stuff [default: training]
-  -g, --use-gpu                 Use the GPU
-  -v, --verbose                 Increase the verbosity (may appear multiple times).
+  -h, --help                            Shows this help message and exits
+      --model=<string>                  Filename of the model to load (if any). 
+      --batch-size=<int>                Batch size [default: 64]
+      --epochs=<int>                    Number of training epochs [default: 20]
+      --learning-rate=<float>           Learning rate [default: 0.01]
+      --adapted-layers=<string>         Layers to adapt in the training [default: conv1-block1-group1]
+      --adapt-reference-channel         Flag deciding whether to adapt default channel as well [default: False]
+  -S, --seed=<int>                      The random seed [default: 3] 
+  -o, --output-dir=<path>               Dir to save stuff [default: training]
+  -g, --use-gpu                         Use the GPU
+  -v, --verbose                         Increase the verbosity (may appear multiple times).
 
 Note that arguments provided directly by command-line will override the ones in the configuration file.
 
@@ -80,6 +82,8 @@ def main(user_input=None):
   output_dir = get_parameter(args, configuration, 'output_dir', 'training')
   use_gpu = get_parameter(args, configuration, 'use_gpu', False)
   verbosity_level = get_parameter(args, configuration, 'verbose', 0)
+  adapted_layers = get_parameter(args, configuration, 'adapted_layers', 'conv1-block1-group1')
+  adapt_reference_channel = get_parameter(args, configuration, 'adapt_reference_channel', False)
   
   bob.core.log.set_verbosity_level(logger, verbosity_level)
   bob.io.base.create_directories_safe(output_dir)
@@ -92,6 +96,8 @@ def main(user_input=None):
   logger.debug("Seed = {}".format(seed))
   logger.debug("Output directory = {}".format(output_dir))
   logger.debug("Use GPU = {}".format(use_gpu))
+  logger.debug("Adapted layers = {}".format(adapted_layers))
+  logger.debug("Adapt reference channel = {}".format(adapt_reference_channel))
 
   # process on the arguments / options
   torch.manual_seed(seed)
@@ -110,7 +116,7 @@ def main(user_input=None):
   
   # train the network
   if hasattr(configuration, 'network'):
-    trainer = MCCNNTrainer(configuration.network, batch_size=batch_size, use_gpu=use_gpu, adapted_layers='conv1-block1-group1', adapt_reference_channel=False, verbosity_level=verbosity_level)
+    trainer = MCCNNTrainer(configuration.network, batch_size=batch_size, use_gpu=use_gpu, adapted_layers=adapted_layers, adapt_reference_channel=adapt_reference_channel, verbosity_level=verbosity_level)
     trainer.train(dataloader, n_epochs=epochs, learning_rate=learning_rate, output_dir=output_dir, model=model)
   else:
     logger.error("Please provide a network in your configuration file !")

bob/learn/pytorch/trainers/MCCNNTrainer.py

@@ -19,8 +19,9 @@ import os
 #1. Logging to tensorboardX or a simpler logger: 
 #2. Support for Validation set and validation loss
 #3. Use to(device) instead of .cuda()?: Eventually migrate everything to this!
-#4. Moving more arguments to config?
+#4. Moving more arguments to config? Done 
 #5. Implement the selection of channels as a transform: Added now ChannelSelect
+#6. Import the function `comp_bce_loss_weights` from utils
 
 """
 def comp_bce_loss_weights(target):
@@ -72,7 +73,8 @@ class MCCNNTrainer(object):
   """
  
   def __init__(self, network, batch_size=64, use_gpu=False, adapted_layers='conv1-block1-group1',adapt_reference_channel=False, verbosity_level=2):
-    """ Init function
+    """ Init function . The layers to be adapted in the network is selected and the gradients are set to `True` 
+    for the  layers which needs to be adapted. 
 
     Parameters
     ----------
@@ -97,15 +99,6 @@ class MCCNNTrainer(object):
     self.use_gpu = use_gpu
     self.criterion = nn.BCELoss()
 
-    # if selected_channel_indexes is not None and isinstance(selected_channel_indexes,list):
-    #   assert(len(selected_channel_indexes)==self.network.num_channels)
-
-    #   self.sel_ind=selected_channel_indexes
-
-    # else: # default is to select all channels
-    #   self.sel_ind=[i for i in range(self.network.num_channels)]
-
-
     if self.use_gpu:
       self.network.cuda()
 
@@ -136,17 +129,14 @@ class MCCNNTrainer(object):
 
     layers_to_adapt = list(np.unique(layers_to_adapt))
 
-    #logger.info("Listing the layers which would be adapted:")
+    # Setting the gradients to true for the layers which needs to be adapted
 
     for name, param in  self.network.named_parameters():
       param.requires_grad = False
       for lta in layers_to_adapt:
         if lta in name:
-          #logger.info(name)
           param.requires_grad = True
 
-
-
   def load_model(self, model_filename):
     """Loads an existing model
 
@@ -222,8 +212,6 @@ class MCCNNTrainer(object):
       The path to a pretrained model file to start training from; this is the PAD model; not the LightCNN model
 
     """
-    logger.debug("EPOCHS to train : {}".format(n_epochs))
-    logger.debug("Learning rate : {}".format(learning_rate))
 
     # if model exists, load it
     if model is not None:
@@ -235,12 +223,14 @@ class MCCNNTrainer(object):
       losses = []
       logger.info('Starting training from scratch')
 
-    # debug print the layers where grad is true ADD
+    # print the layers where grad is True
+
+    logger.info("Number of channels: {}".format(self.network.num_channels))
 
     for name, param in  self.network.named_parameters():
 
       if param.requires_grad == True:
-        logger.debug('Layer to be adapted from grad check : {}'.format(name))
+        logger.info('Layer to be adapted from grad check : {}'.format(name))
 
     # setup optimizer
 
@@ -258,13 +248,6 @@ class MCCNNTrainer(object):
           
           img, labels = data
 
-          # print("img shape type", img.shape, type(img))
-
-          # img=imgt[:,self.sel_ind,:,:] # subselect channels
-
-          # print("img shape type", img.shape, type(img))
-
-          #sel_ind
 
           labels=labels.float().unsqueeze(1)
 
@@ -280,7 +263,8 @@ class MCCNNTrainer(object):
           imagesv = Variable(img)
           labelsv = Variable(labels)
           
-          self.criterion.weight=weights
+          # weights for samples, should help with data imbalance
+          self.criterion.weight = weights
 
           output= self.network(imagesv)
           loss = self.criterion(output, labelsv)