Fix incorrect model<->tokenizer mapping in tokenization testing (#19872)

* Fix model-tokenizer mapping

Co-authored-by: ydshieh <ydshieh@users.noreply.github.com>
This commit is contained in:
Yih-Dar
2022-10-25 16:02:13 +02:00
committed by GitHub
parent eedaba682f
commit f9257843b5

View File

@@ -120,8 +120,11 @@ def merge_model_tokenizer_mappings(
tokenizer = tokenizer_mapping[configuration][0]
tokenizer_fast = tokenizer_mapping[configuration][1]
if tokenizer is not None:
if configuration.__name__.startswith(tokenizer.__name__.replace("Tokenizer", "")):
model_tokenizer_mapping.update({tokenizer: (configuration, model)})
if tokenizer_fast is not None:
if configuration.__name__.startswith(tokenizer_fast.__name__.replace("TokenizerFast", "")):
model_tokenizer_mapping.update({tokenizer_fast: (configuration, model)})
return model_tokenizer_mapping