fix(exporters): handle read-only config in sentence-transformers >=5

optimum/exporters/tasks.py

@@ -1042,11 +1042,19 @@ def standardize_model_attributes(
 
         elif library_name == "sentence_transformers":
             if "Transformer" in model[0].__class__.__name__:
-                model.config = model[0].auto_model.config
-                model.config.export_model_type = "transformer"
+                inner_config = model[0].auto_model.config
+                try:
+                    model.config = inner_config
+                except AttributeError:
+                    pass
+                inner_config.export_model_type = "transformer"
             elif "CLIP" in model[0].__class__.__name__:
-                model.config = model[0].model.config
-                model.config.export_model_type = "clip"
+                inner_config = model[0].model.config
+                try:
+                    model.config = inner_config
+                except AttributeError:
+                    pass
+                inner_config.export_model_type = "clip"
             else:
                 raise ValueError(
                     f"The export of a sentence_transformers model with the first module being {model[0].__class__.__name__} is currently not supported in Optimum. Please open an issue or submit a PR to add the support."

tests/exporters/common/test_tasks_manager.py

@@ -184,3 +184,32 @@ def test_library_detection(self):
         )
         self.assertEqual(TasksManager.infer_library_from_model("gpt2"), "transformers")
         self.assertEqual(TasksManager.infer_library_from_model("timm/mobilenetv3_large_100.ra_in1k"), "timm")
+
+    def test_standardize_sentence_transformers_readonly_config(self):
+        # Regression: sentence-transformers >= 5 exposes `config` as a read-only
+        # property, so a direct assignment used to raise AttributeError before
+        # `export_model_type` could be set on the underlying transformer config.
+        class FakeTransformer:
+            def __init__(self, config):
+                self.auto_model = type("AutoModel", (), {"config": config})()
+
+        class FakeSentenceTransformer:
+            def __init__(self, inner):
+                self._modules_list = [inner]
+
+            def __getitem__(self, idx):
+                return self._modules_list[idx]
+
+            @property
+            def config(self):
+                return self._modules_list[0].auto_model.config
+
+        inner_config = BertConfig()
+        transformer = FakeTransformer(inner_config)
+        transformer.__class__.__name__ = "Transformer"
+        st_model = FakeSentenceTransformer(transformer)
+
+        TasksManager.standardize_model_attributes(st_model, library_name="sentence_transformers")
+
+        self.assertEqual(inner_config.export_model_type, "transformer")
+        self.assertIs(st_model.config, inner_config)