IBM · tkornuta-ibm · Jul 3, 2019 · Jul 2, 2019 · Jul 2, 2019 · Jul 2, 2019
diff --git a/README.md b/README.md
@@ -15,24 +15,24 @@
 PyTorchPipe (PTP) is a component-oriented framework that facilitates development of computational _multi-modal pipelines_ and comparison of diverse neural network-based models. 
 
 PTP frames training and testing procedures as _pipelines_ consisting of many components communicating through data streams.
-Each such a stream can consist of several components, including one problem instance (providing batches of data), any number of trainable components (models) and additional components providing required transformations and computations.
+Each such a stream can consist of several components, including one task instance (providing batches of data), any number of trainable components (models) and additional components providing required transformations and computations.
 
-As a result, the training & testing procedures are no longer pinned to a specific problem or model, and built-in mechanisms for compatibility checking (handshaking), configuration and global variables management & statistics collection facilitate rapid development of complex pipelines and running diverse experiments.
+As a result, the training & testing procedures are no longer pinned to a specific task or model, and built-in mechanisms for compatibility checking (handshaking), configuration and global variables management & statistics collection facilitate rapid development of complex pipelines and running diverse experiments.
 
 In its core, to _accelerate the computations_ on their own, PTP relies on PyTorch and extensively uses its mechanisms for distribution of computations on CPUs/GPUs, including multi-process data loaders and multi-GPU data parallelism.
 The models are _agnostic_ to those operations and one indicates whether to use them in configuration files (data loaders) or by passing adequate run-time arguments (--gpu).
 
 **Datasets:**
-PTP focuses on multi-modal reasoning combining vision and language. Currently it offers the following _Problems_ from the following problem domains:
+PTP focuses on multi-modal reasoning combining vision and language. Currently it offers the following _Tasks_ from the following task domains:
 
   * CLEVR, GQA, ImageCLEF VQA-Med 2019 (Visual Question Answering)
   * MNIST, CIFAR-100 (Image Classification)
   * WiLY (Language Identification)
   * WikiText-2 / WikiText-103 (Language Modelling)
   * ANKI (Machine Translation)
 
-Aside of providing batches of samples, the Problem class will automatically download the files associated with a given dataset (as long as the dataset is publicly available).
-The diversity of those problems (and associated models) proves the flexibility of the framework, we are working on incorporation of new ones into PTP.
+Aside of providing batches of samples, the Task class will automatically download the files associated with a given dataset (as long as the dataset is publicly available).
+The diversity of those tasks (and associated models) proves the flexibility of the framework, we are working on incorporation of new ones into PTP.
 
 **Pipelines:**
 What people typically define as a _model_ in PTP is framed as a _pipeline_, consisting of many inter-connected components, with one or more _Models_ containing trainable elements.
@@ -72,14 +72,14 @@ The framework also offers several components useful when working with text:
 and several general-purpose components, from tensor transformations (List to Tensor, Reshape Tensor, Reduce Tensor, Concatenate Tensor), to components calculating losses (NLL Loss) and statistics (Accuracy Statistics, Precision/Recall Statistics, BLEU Statistics etc.) to viewers (Stream Viewer, Stream File Exporter etc.).
 
 **Workers:**
-PTP workers are python scripts that are _agnostic_ to the problems/models/pipelines that they are supposed to work with.
+PTP workers are python scripts that are _agnostic_ to the tasks/models/pipelines that they are supposed to work with.
 Currently framework offers three workers:
 
-  * ptp-offline-trainer (a trainer relying on classical methodology interlacing training and validation at the end of every epoch, creates separate instances of training and validation problems and trains the models by feeding the created pipeline with batches of data, relying on the notion of an _epoch_)
+  * ptp-offline-trainer (a trainer relying on classical methodology interlacing training and validation at the end of every epoch, creates separate instances of training and validation tasks and trains the models by feeding the created pipeline with batches of data, relying on the notion of an _epoch_)
 
-  * ptp-online-trainer (a flexible trainer creating separate instances of training and validation problems and training the models by feeding the created pipeline with batches of data, relying on the notion of an _episode_)
+  * ptp-online-trainer (a flexible trainer creating separate instances of training and validation tasks and training the models by feeding the created pipeline with batches of data, relying on the notion of an _episode_)
 
-  * ptp-processor (performing one pass over the all samples returned by a given problem instance, useful for collecting scores on test set, answers for submissions to competitions etc.)
+  * ptp-processor (performing one pass over the all samples returned by a given task instance, useful for collecting scores on test set, answers for submissions to competitions etc.)
 
 
 ## Installation

diff --git a/configs/cifar100/cifar100_classification_convnet_softmax.yml b/configs/cifar100/cifar100_classification_convnet_softmax.yml
@@ -1,4 +1,4 @@
-# Load config defining CIFAR100 problems for training, validation and testing.
+# Load config defining CIFAR100 tasks for training, validation and testing.
 default_configs: cifar100/default_cifar100.yml
 
 # Definition of the pipeline.

diff --git a/configs/cifar100/default_cifar100.yml b/configs/cifar100/default_cifar100.yml
@@ -1,6 +1,6 @@
 # Training parameters:
 training:
-  problem: 
+  task: 
     type: CIFAR100
     batch_size: &b 1024
     use_train_data: True
@@ -25,7 +25,7 @@ training:
 # Validation parameters:
 validation:
   #partial_validation_interval: 100
-  problem:
+  task:
     type: CIFAR100
     batch_size: *b
     use_train_data: True  # True because we are splitting the training set to: validation and training
@@ -37,7 +37,7 @@ validation:
 
 # Testing parameters:
 test:
-  problem:
+  task:
     type: MNIST
     batch_size: *b
     use_train_data: False
@@ -94,7 +94,7 @@ pipeline:
 
   image_viewer:
     priority: 100.5
-    type: ImageToClassViewer
+    type: ImageViewer
     streams:
       images: inputs
       labels: fine_labels

diff --git a/configs/clevr/clevr_all_vgg_glove_lstm_concat_ffn.yml b/configs/clevr/clevr_all_vgg_glove_lstm_concat_ffn.yml
@@ -1,19 +1,19 @@
-# Load config defining CLEVR problems for training, validation and testing.
+# Load config defining CLEVR tasks for training, validation and testing.
 default_configs: clevr/default_clevr.yml
 
 # Resize and normalize images - in all sets.
 training:
-  problem: 
+  task: 
     resize_image: [224, 224]
     image_preprocessing: normalize
 
 validation:
-  problem: 
+  task: 
     resize_image: [224, 224]
     image_preprocessing: normalize
 
 test:
-  problem: 
+  task: 
     resize_image: [224, 224]
     image_preprocessing: normalize
 
@@ -76,7 +76,7 @@ pipeline:
   # Image encoder.
   image_encoder:
     priority: 2.1
-    type: TorchVisionWrapper
+    type: GenericImageEncoder
     model_type: vgg16
     streams:
       inputs: images
@@ -87,7 +87,7 @@ pipeline:
   ##################################################################
   # 3rd subpipeline: concatenation + FF.
   concat:
-    type: Concatenation
+    type: ConcatenateTensor
     priority: 3.1
     input_streams: [question_activations,image_activations]
     dim: 1 # default

diff --git a/configs/clevr/clevr_image_convnet_ffn.yml b/configs/clevr/clevr_image_convnet_ffn.yml
@@ -1,4 +1,4 @@
-# Load config defining CLEVR problems for training, validation and testing.
+# Load config defining CLEVR tasks for training, validation and testing.
 default_configs: clevr/default_clevr.yml
 
 # Definition of the pipeline.

diff --git a/configs/clevr/clevr_question_glove_lstm.yml b/configs/clevr/clevr_question_glove_lstm.yml
@@ -1,17 +1,17 @@
-# Load config defining CLEVR problems for training, validation and testing.
+# Load config defining CLEVR tasks for training, validation and testing.
 default_configs: clevr/default_clevr.yml
 
 # This is unimodal (questino-based) baseline, thus stop streaming images - in all sets.
 training:
-  problem: 
+  task: 
     stream_images: False
 
 validation:
-  problem: 
+  task: 
     stream_images: False
 
 test:
-  problem: 
+  task: 
     stream_images: False
 
 # Definition of the pipeline.

diff --git a/configs/clevr/default_clevr.yml b/configs/clevr/default_clevr.yml
@@ -1,6 +1,6 @@
 # Training parameters:
 training:
-  problem: 
+  task: 
     type: CLEVR
     batch_size: &b 64
     split: training
@@ -18,15 +18,15 @@ training:
 
 # Validation parameters:
 validation:
-  problem:
+  task:
     type: CLEVR
     batch_size: *b
     split: validation
     #resize_image: [224, 224]
 
 # Testing parameters:
 test:
-  problem:
+  task:
     type: CLEVR
     batch_size: *b
     split: test
@@ -96,7 +96,7 @@ pipeline:
 
   #image_viewer:
   #  priority: 100.5
-  #  type: ImageToClassViewer
+  #  type: ImageViewer
   #  streams:
   #    images: inputs
   #    labels: labels

diff --git a/...s/default/components/text/bow_encoder.yml → ...fault/components/language/bow_encoder.yml b/...s/default/components/text/bow_encoder.yml → ...fault/components/language/bow_encoder.yml
diff --git a/...default/components/text/label_indexer.yml → ...ult/components/language/label_indexer.yml b/...default/components/text/label_indexer.yml → ...ult/components/language/label_indexer.yml
@@ -4,7 +4,7 @@
 # 1. CONFIGURATION PARAMETERS that will be LOADED by the component.
 ####################################################################
 
-# Folder where problem will store data (LOADED)
+# Folder where task will store data (LOADED)
 data_folder: '~/data/'
 
 # Source files that will be used to create the vocabulary  (LOADED)

diff --git a/...ault/components/text/sentence_indexer.yml → .../components/language/sentence_indexer.yml b/...ault/components/text/sentence_indexer.yml → .../components/language/sentence_indexer.yml
@@ -4,7 +4,7 @@
 # 1. CONFIGURATION PARAMETERS that will be LOADED by the component.
 ####################################################################
 
-# Folder where problem will store data (LOADED)
+# Folder where task will store data (LOADED)
 data_folder: '~/data/'
 
 # Source files that will be used to create the vocabulary  (LOADED)

diff --git a/...ponents/text/sentence_one_hot_encoder.yml → ...nts/language/sentence_one_hot_encoder.yml b/...ponents/text/sentence_one_hot_encoder.yml → ...nts/language/sentence_one_hot_encoder.yml
@@ -4,7 +4,7 @@
 # 1. CONFIGURATION PARAMETERS that will be LOADED by the component.
 ####################################################################
 
-# Folder where problem will store data (LOADED)
+# Folder where task will store data (LOADED)
 data_folder: '~/data/'
 
 # Source files that will be used to create the vocabulary  (LOADED)

diff --git a/...lt/components/text/sentence_tokenizer.yml → ...omponents/language/sentence_tokenizer.yml b/...lt/components/text/sentence_tokenizer.yml → ...omponents/language/sentence_tokenizer.yml
diff --git a/.../default/components/text/word_decoder.yml → ...ault/components/language/word_decoder.yml b/.../default/components/text/word_decoder.yml → ...ault/components/language/word_decoder.yml
@@ -4,7 +4,7 @@
 # 1. CONFIGURATION PARAMETERS that will be LOADED by the component.
 ####################################################################
 
-# Folder where problem will store data (LOADED)
+# Folder where task will store data (LOADED)
 data_folder: '~/data/'
 
 # Source files that will be used to create the vocabulary  (LOADED)

diff --git a/...lt/components/models/attn_decoder_rnn.yml → ...odels/general_usage/attention_decoder.yml b/...lt/components/models/attn_decoder_rnn.yml → ...odels/general_usage/attention_decoder.yml
diff --git a/...omponents/models/feed_forward_network.yml → ...ls/general_usage/feed_forward_network.yml b/...omponents/models/feed_forward_network.yml → ...ls/general_usage/feed_forward_network.yml
diff --git a/...nents/models/recurrent_neural_network.yml → ...eneral_usage/recurrent_neural_network.yml b/...nents/models/recurrent_neural_network.yml → ...eneral_usage/recurrent_neural_network.yml
diff --git a/...default/components/models/seq2seq_rnn.yml → ...mponents/models/general_usage/seq2seq.yml b/...default/components/models/seq2seq_rnn.yml → ...mponents/models/general_usage/seq2seq.yml
diff --git a/...lt/components/models/index_embeddings.yml → ...ents/models/language/index_embeddings.yml b/...lt/components/models/index_embeddings.yml → ...ents/models/language/index_embeddings.yml
diff --git a/...components/models/sentence_embeddings.yml → ...s/models/language/sentence_embeddings.yml b/...components/models/sentence_embeddings.yml → ...s/models/language/sentence_embeddings.yml
@@ -4,7 +4,7 @@
 # 1. CONFIGURATION PARAMETERS that will be LOADED by the component.
 ####################################################################
 
-# Folder where problem will store data (LOADED)
+# Folder where task will store data (LOADED)
 data_folder: '~/data/'
 
 # Source files that will be used to create the vocabulary  (LOADED)

diff --git a/...a/multimodal_compact_bilinear_pooling.yml → ...al_reasoning/compact_bilinear_pooling.yml b/...a/multimodal_compact_bilinear_pooling.yml → ...al_reasoning/compact_bilinear_pooling.yml
diff --git a/...ultimodal_factorized_bilinear_pooling.yml → ...reasoning/factorized_bilinear_pooling.yml b/...ultimodal_factorized_bilinear_pooling.yml → ...reasoning/factorized_bilinear_pooling.yml
@@ -1,4 +1,4 @@
-# This file defines the default values for the MultimodalFactorizedBilinearPooling model.
+# This file defines the default values for the FactorizedBilinearPooling model.
 
 ####################################################################
 # 1. CONFIGURATION PARAMETERS that will be LOADED by the component.

diff --git a/...odels/vqa/element_wise_multiplication.yml → ...l_reasoning/low_rank_bilinear_pooling.yml b/...odels/vqa/element_wise_multiplication.yml → ...l_reasoning/low_rank_bilinear_pooling.yml
@@ -1,4 +1,4 @@
-# This file defines the default values for the ElementWiseMultiplication model.
+# This file defines the default values for the LowRankBilinearPooling model.
 
 ####################################################################
 # 1. CONFIGURATION PARAMETERS that will be LOADED by the component.

diff --git a/...fault/components/models/vqa/attention.yml → ...l_reasoning/question_driven_attention.yml b/...fault/components/models/vqa/attention.yml → ...l_reasoning/question_driven_attention.yml
@@ -1,4 +1,4 @@
-# This file defines the default values for the VQA_Attention model.
+# This file defines the default values for the QuestionDrivenAttention model.
 
 ####################################################################
 # 1. CONFIGURATION PARAMETERS that will be LOADED by the component.

diff --git a/...ponents/models/vqa/relational_network.yml → ...ti_modal_reasoning/relational_network.yml b/...ponents/models/vqa/relational_network.yml → ...ti_modal_reasoning/relational_network.yml
@@ -1,4 +1,4 @@
-# This file defines the default values for the ElementWiseMultiplication model.
+# This file defines the default values for the LowRankBilinearPooling model.
 
 ####################################################################
 # 1. CONFIGURATION PARAMETERS that will be LOADED by the component.

diff --git a/.../components/models/vqa/self_attention.yml → .../multi_modal_reasoning/self_attention.yml b/.../components/models/vqa/self_attention.yml → .../multi_modal_reasoning/self_attention.yml
diff --git a/...ult/components/models/convnet_encoder.yml → ...ponents/models/vision/convnet_encoder.yml b/...ult/components/models/convnet_encoder.yml → ...ponents/models/vision/convnet_encoder.yml
diff --git a/...omponents/models/torch_vision_wrapper.yml → ...s/models/vision/generic_image_encoder.yml b/...omponents/models/torch_vision_wrapper.yml → ...s/models/vision/generic_image_encoder.yml
diff --git a/configs/default/components/models/lenet5.yml → ...fault/components/models/vision/lenet5.yml b/configs/default/components/models/lenet5.yml → ...fault/components/models/vision/lenet5.yml
diff --git a/...onents/publishers/accuracy_statistics.yml → ...onents/statistics/accuracy_statistics.yml b/...onents/publishers/accuracy_statistics.yml → ...onents/statistics/accuracy_statistics.yml
diff --git a/...ents/publishers/batch_size_statistics.yml → ...ents/statistics/batch_size_statistics.yml b/...ents/publishers/batch_size_statistics.yml → ...ents/statistics/batch_size_statistics.yml
diff --git a/...components/publishers/bleu_statistics.yml → ...components/statistics/bleu_statistics.yml b/...components/publishers/bleu_statistics.yml → ...components/statistics/bleu_statistics.yml
diff --git a/...ublishers/precision_recall_statistics.yml → ...tatistics/precision_recall_statistics.yml b/...ublishers/precision_recall_statistics.yml → ...tatistics/precision_recall_statistics.yml
diff --git a/...ts/problems/image_text_to_class/clevr.yml → ...nents/tasks/image_text_to_class/clevr.yml b/...ts/problems/image_text_to_class/clevr.yml → ...nents/tasks/image_text_to_class/clevr.yml
@@ -1,17 +1,17 @@
-# This file defines the default values for the CLEVR problem.
+# This file defines the default values for the CLEVR task.
 
 ####################################################################
 # 1. CONFIGURATION PARAMETERS that will be LOADED by the component.
 ####################################################################
 
-# Folder where problem will store data (LOADED)
+# Folder where task will store data (LOADED)
 data_folder: '~/data/CLEVR_v1.0'
 
 # Defines the set (split) that will be used (LOADED)
 # Options: training | validation | test | cogent_a_training | cogent_a_validation | cogent_b_validation
 split: training
 
-# Flag indicating whether the problem will load and return images (LOADED)
+# Flag indicating whether the task will load and return images (LOADED)
 stream_images: True
 
 # Resize parameter (LOADED)
@@ -31,7 +31,7 @@ streams:
   ####################################################################
 
   # Stream containing batch of indices (OUTPUT)
-  # Every problem MUST return that stream.
+  # Every task MUST return that stream.
   indices: indices
 
   # Stream containing batch of images (OUTPUT)

diff --git a/...ents/problems/image_text_to_class/gqa.yml → ...ponents/tasks/image_text_to_class/gqa.yml b/...ents/problems/image_text_to_class/gqa.yml → ...ponents/tasks/image_text_to_class/gqa.yml
@@ -1,18 +1,18 @@
-# This file defines the default values for the GQA problem.
+# This file defines the default values for the GQA task.
 
 ####################################################################
 # 1. CONFIGURATION PARAMETERS that will be LOADED by the component.
 ####################################################################
 
-# Folder where problem will store data (LOADED)
+# Folder where task will store data (LOADED)
 data_folder: '~/data/gqa'
 
 # Defines the set (split) that will be used (LOADED)
 # Options: training_0 | training | validation | test_dev | test | challenge | submission (?)
 # Note: test_dev should be used for validation.
 split: training_0
 
-# Flag indicating whether the problem will load and return images (LOADED)
+# Flag indicating whether the task will load and return images (LOADED)
 stream_images: True
 
 # Resize parameter (LOADED)
@@ -32,7 +32,7 @@ streams:
   ####################################################################
 
   # Stream containing batch of indices (OUTPUT)
-  # Every problem MUST return that stream.
+  # Every task MUST return that stream.
   indices: indices
 
   # Stream containing batch of sample (original) identifiers (OUTPUT)

diff --git a/...lems/image_text_to_class/vqa_med_2019.yml → ...asks/image_text_to_class/vqa_med_2019.yml b/...lems/image_text_to_class/vqa_med_2019.yml → ...asks/image_text_to_class/vqa_med_2019.yml
@@ -1,10 +1,10 @@
-# This file defines the default values for the VQAMED2019 problem.
+# This file defines the default values for the VQAMED2019 task.
 
 ####################################################################
 # 1. CONFIGURATION PARAMETERS that will be LOADED by the component.
 ####################################################################
 
-# Folder where problem will store data (LOADED)
+# Folder where task will store data (LOADED)
 data_folder: '~/data/vqa-med'
 
 # Defines the set (split) that will be used (LOADED)
@@ -15,7 +15,7 @@ split: training
 # Options: all | c1 | c2 | c3 | c4 (or any combination of the latter 4)
 categories: all
 
-# Flag indicating whether the problem will load and return images (LOADED)
+# Flag indicating whether the task will load and return images (LOADED)
 stream_images: True
 
 # Flag indicating whether images will be preloaded (i.e. loaded once at start) (LOADED)
@@ -29,7 +29,7 @@ preload_images: False
 #resize_image: [height, width]
 
 # Scale parameter [height, width] (LOADED)
-# Problem will use those values to rescale the image_sizes to range (0, 1).
+# Task will use those values to rescale the image_sizes to range (0, 1).
 scale_image_size: [2414, 2323]
 
 # Select applied image preprocessing/augmentations (LOADED)
@@ -50,7 +50,7 @@ question_preprocessing: lowercase, remove_punctuation
 # Accepted formats: a,b,c or [a,b,c]
 answer_preprocessing: none
 
-# When filename is not empty, problem will calculate weights associated with all samples
+# When filename is not empty, task will calculate weights associated with all samples
 # by looking at the distribution of all answers from all loaded samples (LOADED)
 # Those weights can be next used by weighted samplers (e.g. kFoldWeightedSampler)
 export_sample_weights: ''
@@ -78,7 +78,7 @@ streams:
   ####################################################################
 
   # Stream containing batch of indices (OUTPUT)
-  # Every problem MUST return that stream.
+  # Every task MUST return that stream.
   indices: indices
 
   # Stream containing batch of images (OUTPUT)

diff --git a/...nts/problems/image_to_class/cifar_100.yml → ...onents/tasks/image_to_class/cifar_100.yml b/...nts/problems/image_to_class/cifar_100.yml → ...onents/tasks/image_to_class/cifar_100.yml
@@ -1,10 +1,10 @@
-# This file defines the default values for the CIFAR-100 problem.
+# This file defines the default values for the CIFAR-100 task.
 
 ####################################################################
 # 1. CONFIGURATION PARAMETERS that will be LOADED by the component.
 ####################################################################
 
-# Folder where problem will store data (LOADED)
+# Folder where task will store data (LOADED)
 data_folder: '~/data/cifar-100'
 
 # Defines the set that will be used used (LOADED)
@@ -21,7 +21,7 @@ streams:
   ####################################################################
 
   # Stream containing batch of indices (OUTPUT)
-  # Every problem MUST return that stream.
+  # Every task MUST return that stream.
   indices: indices
 
   # Stream containing batch of images (OUTPUT)