Fix BLT training_ci overfit test

src/transformers/models/blt/configuration_blt.py

@@ -335,6 +335,7 @@ def __init__(
         tie_word_embeddings: Optional[bool] = False,
         initializer_range: Optional[float] = 0.02,
         rope_parameters: Optional[RopeParameters | dict[str, RopeParameters]] = None,
+        use_cache: Optional[bool] = False,
         **kwargs,
     ):
         # Basic model configuration
@@ -406,16 +407,37 @@ def __init__(
         )
 
         self.rope_parameters = rope_parameters
+        if "use_cache" not in kwargs:
+            kwargs["use_cache"] = use_cache
+        self.use_cache = kwargs["use_cache"]
 
         # Remove tie_word_embeddings from kwargs to avoid duplicate parameter error
         kwargs.pop("tie_word_embeddings", None)
         super().__init__(tie_word_embeddings=tie_word_embeddings, **kwargs)
 
 
+class BltTextConfig(PreTrainedConfig):
+    """
+    Configuration class for the Blt Text component.
+    """
+
+    pass
+
+
+class BltVisionConfig(PreTrainedConfig):
+    """
+    Configuration class for the Blt Vision component.
+    """
+
+    pass
+
+
 __all__ = [
     "BltConfig",
     "BltPatcherConfig",
     "BltLocalEncoderConfig",
     "BltLocalDecoderConfig",
     "BltGlobalTransformerConfig",
+    "BltTextConfig",
+    "BltVisionConfig",
 ]