From e93710af08e1f22283e9f277ddb73a4487e75dde Mon Sep 17 00:00:00 2001 From: Christian Risi <75698846+CnF-Gris@users.noreply.github.com> Date: Mon, 6 Oct 2025 16:16:47 +0200 Subject: [PATCH] Fixed illegal tokens being added in target output --- Project_Model/Libs/Transformer/Classes/SpannedMasker.py | 5 +++++ 1 file changed, 5 insertions(+) diff --git a/Project_Model/Libs/Transformer/Classes/SpannedMasker.py b/Project_Model/Libs/Transformer/Classes/SpannedMasker.py index 156f512..441a3d8 100644 --- a/Project_Model/Libs/Transformer/Classes/SpannedMasker.py +++ b/Project_Model/Libs/Transformer/Classes/SpannedMasker.py @@ -90,6 +90,11 @@ class SpannedMasker: SPAN_LENGTH = min(CANDIDATE_SPAN, REMAINING_MASK) for _ in range(0, SPAN_LENGTH): + INNER_TOKEN = sequence[mask_index] + + if self.__is_illegal_token(INNER_TOKEN, forbidden_tokens): + continue + MASK[mask_index] = True mask_index += 1