Use batch_num minimum in case the number of samples is less than batc…

…h size

mlpp_lib/datasets.py

@@ -592,7 +592,7 @@ def __init__(
         self.shuffle = shuffle
         self.block_size = block_size
         self.num_samples = len(self.dataset.x)
-        self.num_batches = self.num_samples // batch_size
+        self.num_batches = self.num_samples // batch_size if batch_size <= self.num_samples else 1
         self._indices = tf.range(self.num_samples)
         self._seed = 0
         self._reset()