fixed loading pre-trained tokenizer from directory
This commit is contained in:
@@ -478,7 +478,7 @@ class PreTrainedBertModel(nn.Module):
|
||||
"associated to this path or url.".format(
|
||||
pretrained_model_name,
|
||||
', '.join(PRETRAINED_MODEL_ARCHIVE_MAP.keys()),
|
||||
pretrained_model_name))
|
||||
archive_file))
|
||||
return None
|
||||
if resolved_archive_file == archive_file:
|
||||
logger.info("loading archive file {}".format(archive_file))
|
||||
|
||||
Reference in New Issue
Block a user