change default param in pretraining bert tutorial

x4nth055 · x4nth055 · commit 393224412ca4 · 2021-10-27T13:24:19.000+01:00
diff --git a/machine-learning/nlp/pretraining-bert/PretrainingBERT.ipynb b/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"
       ]
     },
     {
diff --git a/machine-learning/nlp/pretraining-bert/pretrainingbert.py b/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()

Original file line number	Diff line number	Diff line change
`@@ -202,7 +202,7 @@`
`202`	`202`	`"# maximum sequence length, lowering will result to faster training (when increasing batch size)\n",`
`203`	`203`	`"max_length = 512\n",`
`204`	`204`	`"# whether to truncate\n",`
`205`		`- "truncate_longer_samples = False"`
	`205`	`+ "truncate_longer_samples = True"`
`206`	`206`	`]`
`207`	`207`	`},`
`208`	`208`	`{`