fix test errors

huggingface · Aug 16, 2024 · 867baf3 · 867baf3
1 parent 45507e0
commit 867baf3
Show file tree

Hide file tree

Showing 4 changed files with 4 additions and 5 deletions.
diff --git a/src/transformers/models/clipseg/processing_clipseg.py b/src/transformers/models/clipseg/processing_clipseg.py
@@ -169,7 +169,7 @@ def __call__(
             return encoding
         else:
             return BatchEncoding(
-                data=dict(**image_features), tensor_type=output_kwargs["common_kwargs"]["return_tensors"]
+                data=dict(**image_features), tensor_type=output_kwargs["common_kwargs"].get("return_tensors")
             )
 
     def batch_decode(self, *args, **kwargs):

diff --git a/src/transformers/models/instructblipvideo/processing_instructblipvideo.py b/src/transformers/models/instructblipvideo/processing_instructblipvideo.py
@@ -150,7 +150,7 @@ def __call__(
                     )
 
             # cast to desired return tensors type after concatenating
-            text_encoding = BatchEncoding(text_encoding, tensor_type=output_kwargs["common_kwargs"]["return_tensors"])
+            text_encoding = BatchEncoding(text_encoding, tensor_type=output_kwargs["common_kwargs"].get("return_tensors"))
             encoding.update(text_encoding)
             qformer_text_encoding = self.qformer_tokenizer(text=text, **output_kwargs["text_kwargs"])
             encoding["qformer_input_ids"] = qformer_text_encoding.pop("input_ids")

diff --git a/src/transformers/models/owlv2/processing_owlv2.py b/src/transformers/models/owlv2/processing_owlv2.py
@@ -42,7 +42,7 @@ class Owlv2ProcessorKwargs(ProcessingKwargs, total=False):
     images_kwargs: Owlv2ImagesKwargs
     _defaults = {
         "text_kwargs": {
-            "padding": False,
+            "padding": "max_length",
         },
         "common_kwargs": {
             "return_tensors": "np",
@@ -70,7 +70,6 @@ class Owlv2Processor(ProcessorMixin):
     def __init__(self, image_processor, tokenizer, **kwargs):
         super().__init__(image_processor, tokenizer)
 
-    # Copied from transformers.models.owlvit.processing_owlvit.OwlViTProcessor.__call__ with OWLViT->OWLv2
     def __call__(
         self,
         text: Optional[Union[TextInput, PreTokenizedInput, List[TextInput], List[PreTokenizedInput]]] = None,

diff --git a/src/transformers/models/owlvit/processing_owlvit.py b/src/transformers/models/owlvit/processing_owlvit.py
@@ -42,7 +42,7 @@ class OwlViTProcessorKwargs(ProcessingKwargs, total=False):
     images_kwargs: OwlViTImagesKwargs
     _defaults = {
         "text_kwargs": {
-            "padding": False,
+            "padding": "max_length",
         },
         "common_kwargs": {
             "return_tensors": "np",