V0.0.1 Athene

This commit is contained in:
Christian Risi
2025-10-11 19:35:43 +02:00
parent 49946727d8
commit 160b7dbfc0
13 changed files with 1050 additions and 8240 deletions

View File

@@ -10,7 +10,7 @@ class SpannedMasker:
max_vocabulary: int,
forbidden_tokens: set[int],
change_token_probability: float = 0.15,
average_span: int = 1,
average_span: int = 2,
seed: int = random.randint(0, sys.maxsize),
) -> None:
@@ -103,7 +103,7 @@ class SpannedMasker:
if self.__is_illegal_token(INNER_TOKEN, forbidden_tokens):
continue
MASK[mask_index] = True
mask_index += 1