Fix modular edge case + modular sorting order (#35562)

* look-ahead negation

* re add examples by default

* Fix the bug in topological sort

* Update create_dependency_mapping.py

* start adding test

* finalize test

* more tests

* style

* style

examples/modular-transformers/modeling_dummy.py

@@ -597,7 +597,7 @@ class DummyModel(DummyPreTrainedModel):
         output_attentions: bool,
     ):
         if self.config._attn_implementation == "flash_attention_2":
-            if attention_mask is not None and 0.0 in attention_mask:
+            if attention_mask is not None and (attention_mask == 0.0).any():
                 return attention_mask
             return None