Support Cross encoder models (#10400)
Signed-off-by: Max de Bayser <maxdebayser@gmail.com> Signed-off-by: Max de Bayser <mbayser@br.ibm.com> Signed-off-by: Flavia Beo <flavia.beo@ibm.com> Co-authored-by: Flavia Beo <flavia.beo@ibm.com>
This commit is contained in:
co-authored by
Flavia Beo
parent
49628fe13e
commit
214efc2c3c
@@ -305,6 +305,7 @@ class InputPreprocessor:
|
||||
tokens_content = parsed["content"]
|
||||
|
||||
prompt_token_ids = tokens_content["prompt_token_ids"]
|
||||
token_type_ids = tokens_content.get("token_type_ids")
|
||||
multi_modal_data = tokens_content.get("multi_modal_data")
|
||||
mm_processor_kwargs = tokens_content.get("mm_processor_kwargs")
|
||||
|
||||
@@ -318,6 +319,7 @@ class InputPreprocessor:
|
||||
|
||||
return token_inputs(
|
||||
prompt_token_ids=prompt_token_ids,
|
||||
token_type_ids=token_type_ids,
|
||||
multi_modal_data=multi_modal_data,
|
||||
mm_processor_kwargs=mm_processor_kwargs,
|
||||
)
|
||||
|
||||
Reference in New Issue
Block a user