Pass `model_kwargs` when loading a model in `pipeline()` (#12449)

* Pass model_kwargs when loading a model in pipeline

* Add test for model_kwargs parameter of pipeline()

* Rewrite test to not download model

* Fix failing style checks

src/transformers/pipelines/__init__.py

@@ -426,7 +426,13 @@ def pipeline(
     # Will load the correct model if possible
     model_classes = {"tf": targeted_task["tf"], "pt": targeted_task["pt"]}
     framework, model = infer_framework_load_model(
-        model, model_classes=model_classes, config=config, framework=framework, revision=revision, task=task
+        model,
+        model_classes=model_classes,
+        config=config,
+        framework=framework,
+        revision=revision,
+        task=task,
+        **model_kwargs,
     )
 
     model_config = model.config

tests/test_pipelines_token_classification.py

@@ -61,6 +61,13 @@ class TokenClassificationPipelineTests(CustomInputPipelineCommonMixin, unittest.
             for key in output_keys:
                 self.assertIn(key, result)
 
+    @require_torch
+    def test_model_kwargs_passed_to_model_load(self):
+        ner_pipeline = pipeline(task="ner", model=self.small_models[0])
+        self.assertFalse(ner_pipeline.model.config.output_attentions)
+        ner_pipeline = pipeline(task="ner", model=self.small_models[0], model_kwargs={"output_attentions": True})
+        self.assertTrue(ner_pipeline.model.config.output_attentions)
+
     @require_torch
     @slow
     def test_spanish_bert(self):