Fig. 10From: Automatic text classification of actionable radiology reports of tinnitus patients using bidirectional encoder representations from transformer (BERT) and in-domain pre-training (IDPT)Comparison of accuracy in the BERT model using different max sequence lengthsBack to article page