Worker features

configs/default/workers/offline_trainer.yml

@@ -0,0 +1,117 @@
+####################################################################
+# Section defining all the default values of parameters used during training when using ptp-offline-trainer.
+
+# If you want to use different section for training pass its name as command line argument '--training_section_name' to trainer (DEFAULT: training)
+# Note: in such a case remember to define all the required parameters in the new section.
+training:
+  # Set the random seeds: -1 means that they will be picked randomly.
+  # Note: their final values will be stored in the final training_configuration.yml saved to log dir.
+  seed_numpy: -1
+  seed_torch: -1
+
+  # Default batch size.
+  batch_size: 64
+
+  # Definition of the problem (Mandatory!)
+  #problem:
+  #  One must define its type (Mandatory!)
+  #  type: ?
+  #  The rest of the content of that section is problem-specific...
+
+  # Section describing curriculum learning (Optional)
+  #curriculum_learning: 
+  #  # Flag indicating whether curriculum learning has to finish before (eventual) termination of the training.
+  #  must_finish: True
+  #  The rest of the content of that section is problem-specific...
+
+  # Definition of optimizer (Mandatory!)
+  #optimizer:
+  #  # Type - generally all optimizers from PyTorch.optim are allowed (Mandatory!)
+  #  type: Adam
+  #  # Options: 
+  #  lr: 0.0001
+  #  The rest of the content of that section is optimizer-specific...
+
+  # Set a default configuration section for data loader.
+  dataloader:
+    # Shuffle set by default.
+    shuffle: True 
+    batch_sampler: None
+     # Do not use multiprocessing by default.
+    num_workers: 0
+    pin_memory: False
+    # Do not drop last frame by default.
+    drop_last: False
+    timeout: 0
+
+  # Definition of sampler (Optional)
+  # When this section will not be present, worker will use "standard" sampling (please refer to shuffle in dataloader)
+  #sampler:
+  #  # Type - generally all samplers from PyTorch (plus some new onses) are allowed (Mandatory!)
+  #  # Options: 
+  #  type: RandomSmpler
+  #  The rest of the content of that section is optimizer-specific...
+
+  # Terminal conditions that will be used during training.
+  # They can (and ofter should) be overwritten.
+  terminal_conditions:
+    # Terminal condition I: loss threshold, going below will terminate the training.
+    loss_stop_threshold: 0.00001 # 1e-5
+    # Terminal condition II: Early stopping monitor validation loss, if it didn't down during last n validations, training will be terminated (Optional, negative means that this condition is disabled)
+    early_stop_validations: 10
+    # Terminal condition III: maximal number of epochs (Mandatory for this trainer! Must be > 0)
+    epoch_limit: 10
+    # Terminal condition IV: maximal number of episodes (Optional, -1 (negative) means that this condition is disabled)
+    episode_limit: -1
+
+
+
+####################################################################
+# Section defining all the default values of parameters used during validation.
+# If you want to use different section for validation pass its name as command line argument '--validation_section_name' to trainer (DEFAULT: validation)
+# Note: in such a case remember to define all the required parameters in the new section.
+validation:
+  # Defines how often the partial validation will be performed.
+  # In this trainer Partial Validation is optional (negative value means it is disabled)
+  partial_validation_interval: -1
+
+  # Definition of the problem (mandatory!)
+  #problem:
+  #  One must define its type (Mandatory!)
+  #  type: ?
+  #  The rest of the content of that section is problem-specific...
+
+  # Set a default configuration section for data loader.
+  dataloader:
+    # Shuffle set by default.
+    shuffle: True 
+     # Do not use multiprocessing by default.
+    num_workers: 0
+    pin_memory: False
+    # Do not drop last frame by default.
+    drop_last: False
+    timeout: 0
+
+  # Definition of sampler (Optional)
+  # When this section will not be present, worker will use "standard" sampling (please refer to shuffle in dataloader)
+  #sampler:
+  #  # Type - generally all samplers from PyTorch (plus some new onses) are allowed (Mandatory!)
+  #  # Options: 
+  #  type: RandomSmpler
+  #  The rest of the content of that section is optimizer-specific...
+
+
+
+####################################################################
+# Section defining all the default values of parameters used during training.
+# If you want to use different section for validation pass its name as command line argument '--pipeline_section_name' to trainer (DEFAULT: pipeline)
+pipeline: 
+  # Pipeline must contain at least one component.
+  #name_1:
+  #   Each component must have defined its priority... (Mandatory!)
+  #   priority: 0.1 # Can be float. Smaller means higher priority, up to zero.
+  #   # ... and type (Mandatory!)
+  #   type: ?
+  #   The rest of the content of that section is component-specific...
+
+

configs/default/workers/online_trainer.yml

@@ -1,5 +1,6 @@
 ####################################################################
-# Section defining all the default values of parameters used during training.
+# Section defining all the default values of parameters used during training when using ptp-online-trainer.
+
 # If you want to use different section for training pass its name as command line argument '--training_section_name' to trainer (DEFAULT: training)
 # Note: in such a case remember to define all the required parameters in the new section.
 training:
@@ -55,10 +56,12 @@ training:
   # They can (and ofter should) be overwritten.
   terminal_conditions:
     # Terminal condition I: loss threshold, going below will terminate the training.
-    loss_stop: 0.00001 # 1e-5
-    # Terminal condition II: maximal number of epochs (optional, -1 means that this condition is disabled)
+    loss_stop_threshold: 0.00001 # 1e-5
+    # Terminal condition II: Early stopping monitor validation loss, if it didn't down during last n validations, training will be terminated (Optional, negative means that this condition is disabled)
+    early_stop_validations: 10
+    # Terminal condition III: maximal number of epochs (Optional, -1 (negative) means that this condition is disabled)
     epoch_limit: -1
-    # Terminal condition III: maximal number of episodes (Mandatory for this trainer! Must be > 0)
+    # Terminal condition IV: maximal number of episodes (Mandatory for this trainer! Must be > 0)
     episode_limit: 100000
 
 

configs/default/workers/processor.yml

@@ -34,7 +34,7 @@ test:
   #sampler:
   #  # Type - generally all samplers from PyTorch (plus some new onses) are allowed (Mandatory!)
   #  # Options: 
-  #  type: RandomSmpler
+  #  type: RandomSampler
   #  The rest of the content of that section is optimizer-specific...
 
 

configs/mnist/default_mnist.yml

@@ -2,35 +2,36 @@
 training:
   problem: 
     type: MNIST
-    batch_size: &b 6400
+    batch_size: &b 64
     use_train_data: True
     #resize: [32, 32]
   # Use sampler that operates on a subset.
-  #sampler:
-  #  type: SubsetRandomSampler
-  #  indices: [0, 55000]
+  sampler:
+    type: SubsetRandomSampler
+    indices: [0, 55000]
   # optimizer parameters:
   optimizer:
     type: Adam
     lr: 0.0001
   # settings parameters
   terminal_conditions:
-    loss_stop: 0.05
+    loss_stop_threshold: 0.05
+    early_stop_validations: 10
     episode_limit: 10000
     epoch_limit: 10
 
 # Validation parameters:
 validation:
-  partial_validation_interval: 500
+  #partial_validation_interval: 100
   problem:
     type: MNIST
     batch_size: *b
     use_train_data: True  # True because we are splitting the training set to: validation and training
     #resize: [32, 32]
   # Use sampler that operates on a subset.
-  #sampler:
-  #  type: SubsetRandomSampler
-  #  indices: [55000, 60000]
+  sampler:
+    type: SubsetRandomSampler
+    indices: [55000, 60000]
 
 # Testing parameters:
 test:

configs/mnist/mnist_classification_kfold_softmax.yml

@@ -7,18 +7,17 @@ training:
     type: MNIST
     batch_size: &b 64
     use_train_data: True
-    #resize: [32, 32]
   # Use k-fold cross-validation random sampler.
   sampler:
     type: kFoldRandomSampler
-    folds: 7 # Each with size of 6000
+    folds: 10 # Each with size of 6000
   # optimizer parameters:
   optimizer:
     type: Adam
     lr: 0.0001
   # settings parameters
   terminal_conditions:
-    loss_stop: 0.5
+    loss_stop_threshold: 0.5
     episode_limit: 10000
     epoch_limit: 5
 
@@ -29,11 +28,10 @@ validation:
     type: MNIST
     batch_size: *b
     use_train_data: True  # True because we are splitting the training set to: validation and training
-    #resize: [32, 32]
   # Use k-fold cross-validation random sampler.
   sampler:
     type: kFoldRandomSampler
-    folds: 7 # Each with size of 6000
+    folds: 10 # Each with size of 6000
 
 pipeline:
 

configs/mnist/mnist_classification_vf_2lenet5.yml → configs/mnist/mnist_classification_vf_2lenet5_2losses.yml

@@ -20,7 +20,7 @@ test:
 pipeline:
 
   # Disable components for "default" flow.
-  disable: nllloss, precision_recall
+  disable: nllloss, accuracy, precision_recall
 
   # Add global variables.
   global_publisher:
@@ -30,30 +30,39 @@ pipeline:
     values: [3, 7, {"Zero": 0, "One": 1, "Two": 2}, {"Three": 0, "Four": 1, "Five": 2, "Six": 3, "Seven": 4, "Eight": 5, "Nine": 6}]
 
   ################# Flow 1 #################
-  flow1_string_to_mask:
-    type: StringToMask
-    priority: 1.1
-    globals:
-      word_mappings: word_to_ix1
-    streams:
-      strings: labels
-      string_indices: flow1_targets
-      masks: flow1_masks
-
   # Image classifier.
   flow1_image_classifier:
     type: LeNet5
-    priority: 1.2
+    priority: 1.1
     globals:
       prediction_size: num_classes1
     streams:
       inputs: inputs
       predictions: flow1_predictions
 
+  flow1_label_to_mask1:
+    type: StringToMask
+    priority: 1.2
+    globals:
+      word_mappings: word_to_ix1
+    streams:
+      strings: labels
+      masks: flow1_masks
+
+  flow1_label_to_target1:
+    type: LabelIndexer
+    priority: 1.3
+    import_word_mappings_from_globals: True
+    globals:
+      word_mappings: word_to_ix1
+    streams:
+      inputs: labels
+      outputs: flow1_targets
+
   # Masked loss.
   flow1_nllloss:
     type: NLLLoss
-    priority: 1.31
+    priority: 1.4
     use_masking: True
     streams:
       targets: flow1_targets
@@ -64,7 +73,7 @@ pipeline:
   # Statistics.
   flow1_accuracy:
     type: AccuracyStatistics
-    priority: 1.32
+    priority: 1.51
     use_masking: True
     streams:
       predictions: flow1_predictions
@@ -75,7 +84,7 @@ pipeline:
 
   flow1_precision_recall:
     type: PrecisionRecallStatistics
-    priority: 1.33
+    priority: 1.52
     use_word_mappings: True
     show_class_scores: True
     show_confusion_matrix: True
@@ -93,30 +102,39 @@ pipeline:
       f1score: flow1_f1score
 
   ################# Flow 2 #################
-  flow2_string_to_mask:
-    type: StringToMask
-    priority: 2.1
-    globals:
-      word_mappings: word_to_ix2
-    streams:
-      strings: labels
-      string_indices: flow2_targets
-      masks: flow2_masks
-
   # Image classifier.
   flow2_image_classifier:
     type: LeNet5
-    priority: 2.2
+    priority: 2.1
     globals:
       prediction_size: num_classes2
     streams:
       inputs: inputs
       predictions: flow2_predictions
 
+  flow2_label_to_mask2:
+    type: StringToMask
+    priority: 2.2
+    globals:
+      word_mappings: word_to_ix2
+    streams:
+      strings: labels
+      masks: flow2_masks
+
+  flow2_label_to_target2:
+    type: LabelIndexer
+    priority: 2.3
+    import_word_mappings_from_globals: True
+    globals:
+      word_mappings: word_to_ix2
+    streams:
+      inputs: labels
+      outputs: flow2_targets
+
   # Masked loss.
   flow2_nllloss:
     type: NLLLoss
-    priority: 2.31
+    priority: 2.4
     use_masking: True
     streams:
       targets: flow2_targets
@@ -127,7 +145,7 @@ pipeline:
   # Statistics.
   flow2_accuracy:
     type: AccuracyStatistics
-    priority: 2.32
+    priority: 2.41
     use_masking: True
     streams:
       targets: flow2_targets
@@ -138,7 +156,7 @@ pipeline:
 
   flow2_precision_recall:
     type: PrecisionRecallStatistics
-    priority: 2.33
+    priority: 2.42
     use_word_mappings: True
     show_class_scores: True
     show_confusion_matrix: True