Fix pep8

dwi_ml/models/projects/transformer_models.py

@@ -205,8 +205,8 @@ def __init__(self,
         self.dropout_rate = dropout_rate
         self.activation = activation
         self.norm_first = norm_first
-        self.ffnn_hidden_size = ffnn_hidden_size if ffnn_hidden_size is not None \
-            else self.d_model // 2
+        self.ffnn_hidden_size = ffnn_hidden_size if ffnn_hidden_size is not \
+            None else self.d_model // 2
 
         # ----------- Checks
         if self.d_model // self.nheads != float(self.d_model) / self.nheads:
@@ -233,7 +233,8 @@ def __init__(self,
 
         # 2. positional encoding layer
         cls_p = keys_to_positional_encodings[self.positional_encoding_key]
-        self.position_encoding_layer = cls_p(self.d_model, dropout_rate, max_len)
+        self.position_encoding_layer = cls_p(self.d_model, dropout_rate,
+                                             max_len)
 
         # 3. target embedding layer: See child class with Target
 
@@ -869,7 +870,8 @@ def params_for_checkpoint(self):
     def _run_embeddings(self, data, use_padding, batch_max_len):
         # input, targets = data
         inputs = self._run_input_embedding(data[0], use_padding, batch_max_len)
-        targets = self._run_target_embedding(data[1], use_padding, batch_max_len)
+        targets = self._run_target_embedding(data[1], use_padding,
+                                             batch_max_len)
         return inputs, targets
 
     def _run_position_encoding(self, data):
@@ -963,7 +965,8 @@ def params_for_checkpoint(self):
     def _run_embeddings(self, data, use_padding, batch_max_len):
         # inputs, targets = data
         inputs = self._run_input_embedding(data[0], use_padding, batch_max_len)
-        targets = self._run_target_embedding(data[1], use_padding, batch_max_len)
+        targets = self._run_target_embedding(data[1], use_padding,
+                                             batch_max_len)
         inputs = torch.cat((inputs, targets), dim=-1)
 
         return inputs
@@ -1003,4 +1006,4 @@ def find_transformer_class(model_type: str):
         raise ValueError("Model type is not a recognized Transformer"
                          "({})".format(model_type))
 
-    return transformers_dict[model_type]
+    return transformers_dict[model_type]

dwi_ml/testing/projects/tt_visu_argparser.py

@@ -94,7 +94,7 @@ def build_argparser_transformer_visu():
                    help="See description above.")
     g.add_argument('--color_x_y_summary', action='store_true',
                    help="See description above.")
-    gg =g.add_mutually_exclusive_group()
+    gg = g.add_mutually_exclusive_group()
     gg.add_argument('--bertviz', action='store_true',
                     help="See description above.")
     gg.add_argument('--bertviz_locally', action='store_true',

dwi_ml/testing/projects/transformer_visualisation_submethods.py → dwi_ml/testing/projects/tt_visu_submethods.py

@@ -54,7 +54,6 @@ def load_data_run_model(parser, args, model: AbstractTransformerModel,
     else:
         streamline_ids = 0
 
-    streamline_ids = 3544 #    ---------------------------------------------------------------- A EFFACER
     # Bug in dipy. If accessing only one streamline, then sft.streamlines is
     # not a list of streamlines anymore, but the streamline itself. Thus,
     # len(sft) = nb_points instead of 1.
@@ -132,9 +131,9 @@ def reformat_attention(attention, this_seq_len, resample_attention: int):
         # (nb_heads = 1 if average_heads).
 
         # Normalizing weight. Without it, we rapidly see nothing!
-        # Easier to see when we normalize on the x axis.
+        # Easier to see when we normalize on the x-axis.
 
-        # Version 1) L_p pnormalization ==> V = V / norm(v)
+        # Version 1) L_p normalization ==> V = V / norm(v)
         # attention[ll] = torch.nn.functional.normalize(attention[ll], dim=2)
         # attention[ll] = torch.nn.functional.normalize(attention[ll], dim=3)
 
@@ -149,7 +148,6 @@ def reformat_attention(attention, this_seq_len, resample_attention: int):
         max_ = np.max(att, axis=3)
         att = att / max_[:, :, :, None]
 
-
         if resample_attention < this_seq_len:
             print("RESAMPLING ATTENTION TO A SEQUENCE OF LENGTH {}\n"
                   "(uses the max per block)"
@@ -167,9 +165,10 @@ def reformat_attention(attention, this_seq_len, resample_attention: int):
             att = np.pad(att, ((0, 0), (0, 0), (0, missing), (0, missing)),
                          mode='edge')
 
+            # 1000: to see if bug. There should never be padding.
             att = block_reduce(
                 att, block_size=(1, 1, nb_together, nb_together),
-                func=np.max, cval=1000.0)  # 1000: to see if bug.
+                func=np.max, cval=1000.0)
 
         else:
             ind = None
@@ -355,4 +354,4 @@ def ttst_show_model_view(encoder_attention, tokens):
 
     else:
         print("ENCODER ATTENTION: ")
-        show_model_view_as_imshow(encoder_attention, tokens)
+        show_model_view_as_imshow(encoder_attention, tokens)

dwi_ml/testing/utils.py

@@ -1,7 +1,8 @@
 # -*- coding: utf-8 -*-
 from typing import List
 
-from dwi_ml.data.dataset.multi_subject_containers import MultiSubjectDataset, MultisubjectSubset
+from dwi_ml.data.dataset.multi_subject_containers import (MultiSubjectDataset,
+                                                          MultisubjectSubset)
 
 
 def add_args_testing_subj_hdf5(p, ask_input_group=False,

scripts_python/tests/test_all_steps_tto.py

@@ -14,7 +14,8 @@
 
 data_dir = fetch_testing_data()
 tmp_dir = tempfile.TemporaryDirectory()
-MAX_LEN = 400  # During tracking, if we allow 200mm * 0.5 step size = 400 points.
+# During tracking, if we allow 200mm * 0.5 step size = 400 points.
+MAX_LEN = 400
 
 
 def test_help_option(script_runner):
@@ -84,9 +85,9 @@ def test_execution(script_runner, experiments_path):
                                 '--max_batches_per_epoch_validation', '1',
                                 '--nheads', '2', '--max_len', str(MAX_LEN),
                                 '--input_embedding_key', 'nn_embedding',
-                                '--input_embedded_size', '6', '--n_layers_e', '1',
-                                '--ffnn_hidden_size', '3', '--logging', 'INFO',
-                                '--use_gpu')
+                                '--input_embedded_size', '6',
+                                '--n_layers_e', '1', '--ffnn_hidden_size', '3',
+                                '--logging', 'INFO', '--use_gpu')
         assert ret.success
 
     logging.info("************ TESTING TRACKING FROM MODEL ************")

scripts_python/tests/test_all_steps_tts.py

@@ -6,15 +6,17 @@
 import tempfile
 
 from dwi_ml.unit_tests.utils.expected_values import \
-    (TEST_EXPECTED_VOLUME_GROUPS, TEST_EXPECTED_STREAMLINE_GROUPS, TEST_EXPECTED_SUBJ_NAMES)
+    (TEST_EXPECTED_VOLUME_GROUPS, TEST_EXPECTED_STREAMLINE_GROUPS,
+     TEST_EXPECTED_SUBJ_NAMES)
 from dwi_ml.unit_tests.utils.data_and_models_for_tests import fetch_testing_data
 
 data_dir = fetch_testing_data()
 tmp_dir = tempfile.TemporaryDirectory()
-MAX_LEN = 400  # During tracking, if we allow 200mm * 0.5 step size = 400 points.
+# During tracking, if we allow 200mm * 0.5 step size = 400 points.
+MAX_LEN = 400
 
 
-def test_help_option(script_runner):
+def test_help_option():
     # All help tests already tested in test_all_steps_tto.
     pass
 
@@ -31,8 +33,8 @@ def test_execution(script_runner, experiments_path):
     input_group_name = TEST_EXPECTED_VOLUME_GROUPS[0]
     streamline_group_name = TEST_EXPECTED_STREAMLINE_GROUPS[0]
 
-    # Here, testing default values only. See dwi_ml.unit_tests.test_trainer for more
-    # various testing.
+    # Here, testing default values only. See dwi_ml.unit_tests.test_trainer for
+    # more various testing.
     # Max length in current testing dataset is 108. Setting max length to 115
     # for faster testing. Also decreasing other default values.
     logging.info("************ TESTING TRAINING ************")
@@ -103,5 +105,3 @@ def test_execution(script_runner, experiments_path):
         '--subset', 'training', '--logging', 'INFO',
         '--resample_plots', '15', '--rescale_0')
     assert ret.success
-
-

scripts_python/tests/test_all_steps_ttst.py

@@ -12,10 +12,11 @@
 
 data_dir = fetch_testing_data()
 tmp_dir = tempfile.TemporaryDirectory()
-MAX_LEN = 400  # During tracking, if we allow 200mm * 0.5 step size = 400 points.
+# During tracking, if we allow 200mm * 0.5 step size = 400 points.
+MAX_LEN = 400
 
 
-def test_help_option(script_runner):
+def test_help_option():
     # All help tests already tested in test_all_steps_tto.
     pass
 
@@ -32,8 +33,8 @@ def test_execution(script_runner, experiments_path):
     input_group_name = TEST_EXPECTED_VOLUME_GROUPS[0]
     streamline_group_name = TEST_EXPECTED_STREAMLINE_GROUPS[0]
 
-    # Here, testing default values only. See dwi_ml.unit_tests.test_trainer for more
-    # various testing.
+    # Here, testing default values only. See dwi_ml.unit_tests.test_trainer for
+    # more various testing.
     # Max length in current testing dataset is 108. Setting max length to 115
     # for faster testing. Also decreasing other default values.
     logging.info("************ TESTING TRAINING ************")