Fix incorrect length train batches bug + some extra logging

src/data/manager.py

@@ -64,16 +64,19 @@ def __init__(self,
 
         # Process the raw data and create file names, labels, sample weights,
         # and tokenizer
+        logging.info("Processing raw data...")
         file_names, labels, sample_weights, self.tokenizer \
             = self._process_raw_data()
+
         self.raw_data = {split: (file_names[split], labels[split],
                                  sample_weights[split])
                          for split in ['train', 'evaluation', 'validation',
                                        'test', 'inference']}
 
         # Fill the datasets dictionary with datasets for different partitions
-        self.datasets = self._fill_datasets_dict(
-            file_names, labels, sample_weights)
+        logging.info("Creating datasets...")
+        self.datasets = self._fill_datasets_dict(file_names, labels,
+                                                 sample_weights)
 
     def _process_raw_data(self) -> Tuple[Dict[str, List[str]],
                                          Dict[str, List[str]],
@@ -362,7 +365,7 @@ def _is_valid_ground_truth(self,
         if unsupported_characters:
             # Unsupported characters are allowed in the validation, inference,
             # and test partitions, but not in the evaluation partition
-            if partition_name in ['validation', 'inference', 'test']:
+            if partition_name in ('validation', 'inference', 'test'):
                 return True
 
             if partition_name == 'train' and not self.charlist:
@@ -406,7 +409,7 @@ def get_ground_truth(self, partition: str, item_id: int):
 
     def get_train_batches(self):
         """ Get the number of batches for training """
-        return int(np.ceil(len(self.raw_data['train'])
+        return int(np.ceil(len(self.raw_data['train'][0])
                            / self.config['batch_size']))
 
     def _create_dataset(self,
@@ -452,7 +455,7 @@ def _create_dataset(self,
         dataset = tf.data.Dataset.from_tensor_slices(data)
         if is_training:
             # Add additional repeat and shuffle for training
-            dataset = dataset.repeat().shuffle(len(files))
+            dataset = dataset.repeat().shuffle(len(files), seed=42)
 
         dataset = (dataset
                    .map(data_loader.load_images,