re add examples by default

huggingface · Jan 8, 2025 · 54e9d2d · 54e9d2d
1 parent 03d804c
commit 54e9d2d
Show file tree

Hide file tree

Showing 5 changed files with 8 additions and 7 deletions.
diff --git a/examples/modular-transformers/modeling_dummy.py b/examples/modular-transformers/modeling_dummy.py
@@ -597,7 +597,7 @@ def _update_causal_mask(
         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
 

diff --git a/examples/modular-transformers/modeling_multimodal1.py b/examples/modular-transformers/modeling_multimodal1.py
@@ -597,7 +597,7 @@ def _update_causal_mask(
         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
 

diff --git a/examples/modular-transformers/modeling_my_new_model2.py b/examples/modular-transformers/modeling_my_new_model2.py
@@ -602,7 +602,7 @@ def _update_causal_mask(
         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
 

diff --git a/examples/modular-transformers/modeling_super.py b/examples/modular-transformers/modeling_super.py
@@ -519,7 +519,7 @@ def _update_causal_mask(
         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
 

diff --git a/utils/create_dependency_mapping.py b/utils/create_dependency_mapping.py
@@ -7,10 +7,11 @@ def topological_sort(dependencies):
     new_dependencies = {}
     graph = defaultdict(list)
     for node, deps in dependencies.items():
+        node_name = node.rsplit("modular_", 1)[1].replace(".py", "")
         for dep in deps:
-            if "example" not in node and "auto" not in dep:
-                graph[dep.split(".")[-2]].append(node.split("/")[-2])
-        new_dependencies[node.split("/")[-2]] = node
+            if "auto" not in dep:
+                graph[dep.split(".")[-2]].append(node_name)
+        new_dependencies[node_name] = node
 
     # Create a graph and in-degree count for each node
     def filter_one_by_one(filtered_list, reverse):