Merge pull request #638 from mv1388/code-style-improvement

Code style fix

aitoolbox/experiment/experiment_saver.py

@@ -151,7 +151,7 @@ def __init__(self, project_name, experiment_name,
         BaseFullExperimentS3Saver.__init__(self, keras_model_saver, project_name, experiment_name,
                                            bucket_name=bucket_name, cloud_dir_prefix=cloud_dir_prefix,
                                            local_model_result_folder_path=local_model_result_folder_path)
-        
+
 
 # class FullTensorFlowExperimentS3Saver(BaseFullExperimentS3Saver):
 #     def __init__(self, project_name, experiment_name,

aitoolbox/experiment/training_history.py

@@ -9,7 +9,7 @@ def __init__(self, has_validation=True, strict_content_check=False):
         In many ways the object can be used with the same API as a normal python dict. However, for the need of
         tracking performance in the TrainLoop TrainingHistory offers additional functions handling the input, output
         and quality assurance of the stored results.
-        
+
         Args:
             has_validation: if train history should by default include 'val_loss'. This is needed when train loops
                 by default evaluate loss on validation set when such a set is available.

aitoolbox/nlp/dataset/HotpotQA/HotpotQA.py

@@ -22,7 +22,6 @@ def get_dataset_local_copy(local_dataset_folder_path, protect_local_folder=True)
 
     Returns:
         None
-
     """
     dataset_fetcher = HotpotQADatasetFetcher(bucket_name='dataset-store', local_dataset_folder_path=local_dataset_folder_path)
     dataset_fetcher.fetch_dataset(protect_local_folder=protect_local_folder)

aitoolbox/nlp/experiment_evaluation/NLP_metrics.py

@@ -501,8 +501,10 @@ def __init__(self, y_true, y_predicted, lowercase=False, source_sents=None, outp
     def calculate_metric(self):
         BLEUSentenceScoreMetric.check_transl_sent_num_match([self.y_true, self.y_predicted])
 
-        sentence_bleu_results = [bleu.get_moses_multi_bleu([' '.join(true_t)], [' '.join(pred_t)], lowercase=self.lowercase) 
-                                 for true_t, pred_t in zip(self.y_true, self.y_predicted)]
+        sentence_bleu_results = [
+            bleu.get_moses_multi_bleu([' '.join(true_t)], [' '.join(pred_t)], lowercase=self.lowercase)
+            for true_t, pred_t in zip(self.y_true, self.y_predicted)
+        ]
 
         if self.output_text_dir is not None:
             BLEUSentenceScoreMetric.dump_translation_text_to_disk(self.source_sents,

examples/manual_aws_experiment_tracking/model_save_to_AWS.py

@@ -142,7 +142,7 @@ def build_FastQA_RNN_concat_model_GLOVE(RNN, story_maxlen, query_maxlen, vocab_s
 # local_model_result_folder_path = '/home/ec2-user/project/model_results'
 
 m_saver = KerasS3ModelSaver(local_model_result_folder_path=local_model_result_folder_path)
-s3_model_path, experiment_timestamp, _ = m_saver.save_model(model=model, 
+s3_model_path, experiment_timestamp, _ = m_saver.save_model(model=model,
                                                             project_name='QA_QAngaroo', experiment_name='separate_files2',
                                                             protect_existing_folder=True)
 print(s3_model_path)

examples/misc_examples/torchtrain_MT_example.py

@@ -18,10 +18,8 @@
 
 
 """
-        
 Based on:
     https://github.com/bentrevett/pytorch-seq2seq/blob/master/4%20-%20Packed%20Padded%20Sequences%2C%20Masking%20and%20Inference.ipynb
-
 """
 
 

tests/test_experiment/test_local_experiment_saver.py

@@ -85,7 +85,7 @@ def save_experiment_options(self, save_true_pred_labels, separate_files):
                                 os.path.join(results_path, f'results_{exp_dir_name}_{current_time}.p'),
                                 os.path.join(results_path, f'hyperparams_{exp_dir_name}_{current_time}.p'),
                                 os.path.join(results_path, f'train_history_{exp_dir_name}_{current_time}.p')]
-            
+
             if save_true_pred_labels:
                 saved_paths_true.append(os.path.join(results_path, f'true_pred_labels_{exp_dir_name}_{current_time}.p'))
 

tests/test_experiment/test_local_save/test_local_model_save.py

@@ -214,7 +214,7 @@ def save_model_local(self, checkpoint_model, epoch):
         exp_dir_name = 'experimentSubDirPT'
         current_time = datetime.datetime.fromtimestamp(time.time()).strftime('%Y-%m-%d_%H-%M-%S')
         model_name_true = f'model_{exp_dir_name}_{current_time}_E{epoch}.pth'
-        
+
         project_path = os.path.join(THIS_DIR, project_dir_name)
         exp_path = os.path.join(project_path, f'{exp_dir_name}_{current_time}')
         model_path = os.path.join(exp_path, 'checkpoint_model' if checkpoint_model else 'model')

tests/test_torchtrain/test_callbacks/test_performance_eval_callbacks.py

@@ -77,7 +77,7 @@ def test_result_package_prepare(self):
         y_test = result_pkg.y_true
         y_pred = result_pkg.y_predicted
         metadata = result_pkg.additional_results['additional_results']
-        
+
         r = []
         for i in range(1, len(dummy_val_loader) + 1):
             r += [i] * 64

tests/test_utils/test_dict_util.py

@@ -54,7 +54,7 @@ class TestFlattenDict(unittest.TestCase):
     def test_flatten_dict(self):
         input_dict = {'bla': 12, 'www': 455, 'pppp': 4004}
         self.assertEqual(dict_util.flatten_dict(input_dict), input_dict)
-        
+
         input_dict_1_level = {'bla': {'uuu': 334, 'www': 1010}, 'rogue': {'ppp': 123}}
         self.assertEqual(dict_util.flatten_dict(input_dict_1_level),
                          {'bla_uuu': 334, 'bla_www': 1010, 'rogue_ppp': 123})

tests/utils.py

@@ -124,7 +124,7 @@ def get_predictions(self, batch_data, device):
         x, y = batch_data
         pred_y, _ = self(x)
         return pred_y, y, {}
-    
+
 
 def keras_dummy_model():
     from keras.models import Sequential
@@ -237,7 +237,7 @@ def get_loss(self, model, batch_data, criterion, device):
 
         Returns:
             PyTorch loss
-        """    
+        """
         return self.dummy_batch
 
     def get_loss_eval(self, model, batch_data, criterion, device):