rename mergedTRN to stackedTRN

also rename config parameter from concat_telescopes to stack_telescope_images

config/example_config.yml

@@ -11,13 +11,13 @@
 
 # List of reconstruction tasks. For now it's recommended to you single task learning.
 # Valid options:
-#     - 'particletype'
+#     - 'type'
 #     - 'energy'
 #     - 'direction'
 # These options have to be in consonance with the 'event_info' and 
 # 'transforms' in 'Data'.
 # Requirements/recommendations:
-#   - 'particletype':
+#   - 'type':
 #       event_info:
 #           - 'true_shower_primary_id'
 #       transforms:
@@ -292,10 +292,10 @@ Input:
     # Optional integer. Default: 1
     batch_size_per_worker: 32
 
-    # Whether to concatenate the stereo images from an event to feed it
+    # Whether to stack the stereo images from an event to feed it
     # to a monoscopic model.
     # Optional boolean. Default: false
-    concat_telescopes: false
+    stack_telescope_images: false
 
 # Settings for the TensorFlow model. The options in this and the
 # Model Parameters section are passed to the Estimator model_fn
@@ -467,7 +467,7 @@ Model Parameters:
     # For the classification task, the class names are passed into the dict
     # in order of class label (0 to n-1).
     standard_head:
-        particletype: {class_names: ['proton', 'gamma'], fc_head: [512, 256, 2], weight: 1.0}
+        type: {class_names: ['proton', 'gamma'], fc_head: [512, 256, 2], weight: 1.0}
         energy: {fc_head: [512, 256, 1], weight: 1.0}
         direction: {fc_head: [512, 256, 2], weight: 1.0}
 

ctlearn/data_loader.py

@@ -13,7 +13,7 @@ def __init__(
         mode="train",
         class_names=None,
         shuffle=True,
-        concat_telescopes=False,
+        stack_telescope_images=False,
     ):
         "Initialization"
         self.DL1DataReaderDL1DH = DL1DataReaderDL1DH
@@ -22,7 +22,7 @@ def __init__(
         self.mode = mode
         self.class_names = class_names
         self.shuffle = shuffle
-        self.concat_telescopes = concat_telescopes
+        self.stack_telescope_images = stack_telescope_images
         self.on_epoch_end()
 
         # Decrypt the example description
@@ -93,7 +93,7 @@ def __init__(
                 self.img_shape[3],
             )
         # Reshape inputs into proper dimensions for the stereo analysis with merged models
-        if self.concat_telescopes:
+        if self.stack_telescope_images:
             self.img_shape = (
                 self.img_shape[1],
                 self.img_shape[2],

ctlearn/default_config_files/CNNRNN.yml

@@ -23,7 +23,7 @@ Data:
             'MAGICCam': 2
 Input:
     batch_size_per_worker: 16
-    concat_telescopes: false
+    stack_telescope_images: false
 Model:
     name: 'CNNRNN'
     backbone: {module: 'cnn_rnn', function: 'cnn_rnn_model'}

ctlearn/default_config_files/SingleCNN.yml

@@ -23,7 +23,7 @@ Data:
             'MAGICCam': 2
 Input:
     batch_size_per_worker: 64
-    concat_telescopes: false
+    stack_telescope_images: false
 Model:
     name: 'SingleCNN'
     backbone: {module: 'single_cnn', function: 'single_cnn_model'}

ctlearn/default_config_files/TRN.yml

@@ -23,7 +23,7 @@ Data:
             'MAGICCam': 2
 Input:
     batch_size_per_worker: 64
-    concat_telescopes: false
+    stack_telescope_images: false
 Model:
     name: 'ThinResNet'
     backbone: {module: 'single_cnn', function: 'single_cnn_model'}

ctlearn/default_config_files/calwaveSingleCNN.yml

@@ -26,7 +26,7 @@ Data:
             'MAGICCam': 2
 Input:
     batch_size_per_worker: 8
-    concat_telescopes: false
+    stack_telescope_images: false
 Model:
     name: 'SingleCNN'
     backbone: {module: 'single_cnn', function: 'single_cnn_model'}

ctlearn/default_config_files/calwaveTRN.yml

@@ -26,7 +26,7 @@ Data:
             'MAGICCam': 2
 Input:
     batch_size_per_worker: 8
-    concat_telescopes: false
+    stack_telescope_images: false
 Model:
     name: 'ThinResNetWave'
     backbone: {module: 'single_cnn', function: 'single_cnn_model'}

ctlearn/default_config_files/rawwaveSingleCNN.yml

@@ -27,7 +27,7 @@ Data:
             'MAGICCam': 2
 Input:
     batch_size_per_worker: 8
-    concat_telescopes: false
+    stack_telescope_images: false
 Model:
     name: 'SingleCNNWave'
     backbone: {module: 'single_cnn', function: 'single_cnn_model'}

ctlearn/default_config_files/rawwaveTRN.yml

@@ -27,7 +27,7 @@ Data:
             'MAGICCam': 2
 Input:
     batch_size_per_worker: 8
-    concat_telescopes: false
+    stack_telescope_images: false
 Model:
     name: 'ThinResNetWave'
     backbone: {module: 'single_cnn', function: 'single_cnn_model'}

ctlearn/default_config_files/mergedTRN.yml → ctlearn/default_config_files/stackedTRN.yml

@@ -23,7 +23,7 @@ Data:
             'MAGICCam': 2
 Input:
     batch_size_per_worker: 64
-    concat_telescopes: true
+    stack_telescope_images: true
 Model:
     name: 'ThinResNet'
     backbone: {module: 'single_cnn', function: 'single_cnn_model'}

ctlearn/default_config_files/triggerSingleCNN.yml

@@ -32,7 +32,7 @@ Data:
             'MAGICCam': 2
 Input:
     batch_size_per_worker: 8
-    concat_telescopes: false
+    stack_telescope_images: false
 Model:
     name: 'SingleCNN'
     backbone: {module: 'single_cnn', function: 'single_cnn_model'}

ctlearn/default_config_files/triggerTRN.yml

@@ -32,7 +32,7 @@ Data:
             'MAGICCam': 2
 Input:
     batch_size_per_worker: 8
-    concat_telescopes: false
+    stack_telescope_images: false
 Model:
     name: 'ThinResNetWave'
     backbone: {module: 'single_cnn', function: 'single_cnn_model'}

ctlearn/run_model.py

@@ -86,7 +86,7 @@ def run_model(config, mode="train", debug=False, log_to_file=False):
         config["Input"] = {}
     batch_size_per_worker = config["Input"].get("batch_size_per_worker", 64)
     batch_size = batch_size_per_worker * strategy.num_replicas_in_sync
-    concat_telescopes = config["Input"].get("concat_telescopes", False)
+    stack_telescope_images = config["Input"].get("stack_telescope_images", False)
 
     if mode == "train":
         if "Training" not in config:
@@ -107,15 +107,15 @@ def run_model(config, mode="train", debug=False, log_to_file=False):
             batch_size=batch_size,
             mode=mode,
             class_names=class_names,
-            concat_telescopes=concat_telescopes,
+            stack_telescope_images=stack_telescope_images,
         )
         validation_data = KerasBatchGenerator(
             reader,
             validation_indices,
             batch_size=batch_size,
             mode=mode,
             class_names=class_names,
-            concat_telescopes=concat_telescopes,
+            stack_telescope_images=stack_telescope_images,
         )
     elif mode == "predict":
         if "trigger_settings" in config["Data"]:
@@ -135,7 +135,7 @@ def run_model(config, mode="train", debug=False, log_to_file=False):
             mode=mode,
             class_names=class_names,
             shuffle=False,
-            concat_telescopes=concat_telescopes,
+            stack_telescope_images=stack_telescope_images,
         )
 
         # Keras is only considering the last complete batch.
@@ -153,7 +153,7 @@ def run_model(config, mode="train", debug=False, log_to_file=False):
                 mode=mode,
                 class_names=class_names,
                 shuffle=False,
-                concat_telescopes=concat_telescopes,
+                stack_telescope_images=stack_telescope_images,
             )
 
     # Construct the model
@@ -475,7 +475,7 @@ def main():
     parser.add_argument(
         "--default_model",
         "-d",
-        help="Default CTLearn Model; valid options: SingleCNN, TRN, rawwaveSingleCNN, rawwaveTRN, calwaveSingleCNN, calwaveTRN (mono), mergedTRN, and CNNRNN (stereo)",
+        help="Default CTLearn Model; valid options: SingleCNN, TRN, rawwaveSingleCNN, rawwaveTRN, calwaveSingleCNN, calwaveTRN (mono), stackedTRN, and CNNRNN (stereo)",
     )
     parser.add_argument(
         "--clean",

ctlearn/utils.py

@@ -194,8 +194,8 @@ def setup_DL1DataReader(config, mode):
             event_info.append("true_energy")
         transformations.append({"name": "MCEnergy"})
 
-    concat_telescopes = config["Input"].get("concat_telescopes", False)
-    if config["Data"]["mode"] == "stereo" and not concat_telescopes:
+    stack_telescope_images = config["Input"].get("stack_telescope_images", False)
+    if config["Data"]["mode"] == "stereo" and not stack_telescope_images:
         for tel_desc in selected_telescope_types:
             transformations.append(
                 {

docs/source/usage.rst

@@ -31,7 +31,7 @@ Pre-processing is performed using the DL1DataHandler Transform class.
 Input
 ~~~~~
 
-Set parameters of the KerasBatchGenerator that converts the loaded, processed data into generator of batches for the Keras application. Stereoscopic images can be concatenated via the ``concat_telescopes`` flag.
+Set parameters of the KerasBatchGenerator that converts the loaded, processed data into generator of batches for the Keras application. Stereoscopic images can be stacked via the ``stack_telescope_images`` flag.
 
 Model
 ~~~~~
@@ -40,7 +40,7 @@ CTLearn works with any TensorFlow-Keras model obeying the signature of a backbon
 
 To use a custom model, provide in this section the directory containing a Python file that implements the model and the module name (that is, the file name minus the .py extension) and name of the model function within the module.
 
-In addition, CTLearn includes two main models for gamma/hadron classification, energy and arrival direction regression. ``SingleCNN`` analyzes single telescope images using a convolutional neural network (CNN) or multiple residual blocks of convolutional layers (ResNet). Stereoscopic images can be concatenated beforehand (in the ``Input`` config section) to be analyzed by the ``SingleCNN`` model. ``CNN-RNN`` performs array-level reconstruction by feeding the output of a CNN or a ResNet for each telescope into either a recurrent neural network (RNN). All models are built on generic functions from ``basic.py`` and ``resnet_engine.py``. In addition, three different attention mechanisms are implemented in ``attention.py``.
+In addition, CTLearn includes two main models for gamma/hadron classification, energy and arrival direction regression. ``SingleCNN`` analyzes single telescope images using a convolutional neural network (CNN) or multiple residual blocks of convolutional layers (ResNet). Stereoscopic images can be stacked beforehand (in the ``Input`` config section) to be analyzed by the ``SingleCNN`` model. ``CNN-RNN`` performs array-level reconstruction by feeding the output of a CNN or a ResNet for each telescope into either a recurrent neural network (RNN). All models are built on generic functions from ``basic.py`` and ``resnet_engine.py``. In addition, three different attention mechanisms are implemented in ``attention.py``.
 
 Model Parameters
 ~~~~~~~~~~~~~~~~
@@ -90,7 +90,7 @@ optional arguments:
   ``--reco,-r RECO [RECO ...]``\
                         Reconstruction task to perform; valid options: particletype, energy, and/or direction
   ``--default_model,-d DEFAULT_MODEL``\
-                        Default CTLearn Model; valid options: TRN (mono), mergedTRN (stereo), and CNNRNN (stereo)
+                        Default CTLearn Model; valid options: TRN (mono), stackedTRN (stereo), and CNNRNN (stereo)
   ``--clean, --no-clean``\
                         Flag, if the network should be trained with cleaned images (default: False)
   ``--pretrained_weights,-w PRETRAINED_WEIGHTS``\