change default param in pretraining bert tutorial

Author: x4nth055

machine-learning/nlp/pretraining-bert/PretrainingBERT.ipynb

@@ -202,7 +202,7 @@
         "# maximum sequence length, lowering will result to faster training (when increasing batch size)\n",
         "max_length = 512\n",
         "# whether to truncate\n",
-        "truncate_longer_samples = False"
+        "truncate_longer_samples = True"
       ]
     },
     {

machine-learning/nlp/pretraining-bert/pretrainingbert.py

@@ -66,7 +66,7 @@ def dataset_to_text(dataset, output_filename="data.txt"):
 # maximum sequence length, lowering will result to faster training (when increasing batch size)
 max_length = 512
 # whether to truncate
-truncate_longer_samples = False
+truncate_longer_samples = True
 
 # initialize the WordPiece tokenizer
 tokenizer = BertWordPieceTokenizer()