Fixed default config for Pix2Struct model to set `Pix2StructTextMod…

…el` to `is_decoder=True` (huggingface#23051)

added  as default keyword arg. to  in order to correctly configure the decoder

src/transformers/models/pix2struct/configuration_pix2struct.py

@@ -118,6 +118,7 @@ def __init__(
         pad_token_id=0,
         eos_token_id=1,
         tie_word_embeddings=False,
+        is_decoder=True,
         **kwargs,
     ):
         self.vocab_size = vocab_size
@@ -144,6 +145,7 @@ def __init__(
             eos_token_id=eos_token_id,
             decoder_start_token_id=decoder_start_token_id,
             tie_word_embeddings=tie_word_embeddings,
+            is_decoder=is_decoder,
             **kwargs,
         )