Moving pipeline tests from Narsil to hf-internal-testing. (huggin…

…gface#14463)

* Moving everything to `hf-internal-testing`.

* Fixing test values.

* Moving to other repo.

* Last touch?

tests/test_pipelines_common.py

@@ -258,15 +258,15 @@ def __getitem__(self, i):
                 return self.data[i]
 
         text_classifier = pipeline(
-            task="text-classification", model="Narsil/tiny-distilbert-sequence-classification", framework="pt"
+            task="text-classification", model="hf-internal-testing/tiny-random-distilbert", framework="pt"
         )
         dataset = MyDataset()
         for output in text_classifier(dataset):
             self.assertEqual(output, {"label": ANY(str), "score": ANY(float)})
 
     @require_torch
     def test_check_task_auto_inference(self):
-        pipe = pipeline(model="Narsil/tiny-distilbert-sequence-classification")
+        pipe = pipeline(model="hf-internal-testing/tiny-random-distilbert")
 
         self.assertIsInstance(pipe, TextClassificationPipeline)
 
@@ -275,7 +275,7 @@ def test_pipeline_override(self):
         class MyPipeline(TextClassificationPipeline):
             pass
 
-        text_classifier = pipeline(model="Narsil/tiny-distilbert-sequence-classification", pipeline_class=MyPipeline)
+        text_classifier = pipeline(model="hf-internal-testing/tiny-random-distilbert", pipeline_class=MyPipeline)
 
         self.assertIsInstance(text_classifier, MyPipeline)
 
@@ -293,19 +293,19 @@ def data(n: int):
             for _ in range(n):
                 yield "This is a test"
 
-        pipe = pipeline(model="Narsil/tiny-distilbert-sequence-classification")
+        pipe = pipeline(model="hf-internal-testing/tiny-random-distilbert")
 
         results = []
         for out in pipe(data(10)):
-            self.assertEqual(nested_simplify(out), {"label": "LABEL_1", "score": 0.502})
+            self.assertEqual(nested_simplify(out), {"label": "LABEL_0", "score": 0.504})
             results.append(out)
         self.assertEqual(len(results), 10)
 
         # When using multiple workers on streamable data it should still work
         # This will force using `num_workers=1` with a warning for now.
         results = []
         for out in pipe(data(10), num_workers=2):
-            self.assertEqual(nested_simplify(out), {"label": "LABEL_1", "score": 0.502})
+            self.assertEqual(nested_simplify(out), {"label": "LABEL_0", "score": 0.504})
             results.append(out)
         self.assertEqual(len(results), 10)
 
@@ -315,20 +315,20 @@ def data(n: int):
             for _ in range(n):
                 yield "This is a test"
 
-        pipe = pipeline(model="Narsil/tiny-distilbert-sequence-classification", framework="tf")
+        pipe = pipeline(model="hf-internal-testing/tiny-random-distilbert", framework="tf")
         out = pipe("This is a test")
         results = []
         for out in pipe(data(10)):
-            self.assertEqual(nested_simplify(out), {"label": "LABEL_1", "score": 0.502})
+            self.assertEqual(nested_simplify(out), {"label": "LABEL_0", "score": 0.504})
             results.append(out)
         self.assertEqual(len(results), 10)
 
     @require_torch
     def test_unbatch_attentions_hidden_states(self):
         model = DistilBertForSequenceClassification.from_pretrained(
-            "Narsil/tiny-distilbert-sequence-classification", output_hidden_states=True, output_attentions=True
+            "hf-internal-testing/tiny-random-distilbert", output_hidden_states=True, output_attentions=True
         )
-        tokenizer = AutoTokenizer.from_pretrained("Narsil/tiny-distilbert-sequence-classification")
+        tokenizer = AutoTokenizer.from_pretrained("hf-internal-testing/tiny-random-distilbert")
         text_classifier = TextClassificationPipeline(model=model, tokenizer=tokenizer)
 
         # Used to throw an error because `hidden_states` are a tuple of tensors

tests/test_pipelines_image_classification.py

@@ -67,7 +67,7 @@ def run_pipeline_test(self, image_classifier, examples):
 
         import datasets
 
-        dataset = datasets.load_dataset("Narsil/image_dummy", "image", split="test")
+        dataset = datasets.load_dataset("hf-internal-testing/fixtures_image_utils", "image", split="test")
 
         # Accepts URL + PIL.Image + lists
         outputs = image_classifier(

tests/test_pipelines_image_segmentation.py

@@ -68,7 +68,7 @@ def run_pipeline_test(self, image_segmenter, examples):
 
         import datasets
 
-        dataset = datasets.load_dataset("Narsil/image_dummy", "image", split="test")
+        dataset = datasets.load_dataset("hf-internal-testing/fixtures_image_utils", "image", split="test")
 
         batch = [
             Image.open("./tests/fixtures/tests_samples/COCO/000000039769.png"),

tests/test_pipelines_object_detection.py

@@ -74,7 +74,7 @@ def run_pipeline_test(self, object_detector, examples):
 
         import datasets
 
-        dataset = datasets.load_dataset("Narsil/image_dummy", "image", split="test")
+        dataset = datasets.load_dataset("hf-internal-testing/fixtures_image_utils", "image", split="test")
 
         batch = [
             Image.open("./tests/fixtures/tests_samples/COCO/000000039769.png"),

tests/test_pipelines_text_classification.py

@@ -33,20 +33,20 @@ class TextClassificationPipelineTests(unittest.TestCase, metaclass=PipelineTestC
     @require_torch
     def test_small_model_pt(self):
         text_classifier = pipeline(
-            task="text-classification", model="Narsil/tiny-distilbert-sequence-classification", framework="pt"
+            task="text-classification", model="hf-internal-testing/tiny-random-distilbert", framework="pt"
         )
 
         outputs = text_classifier("This is great !")
-        self.assertEqual(nested_simplify(outputs), [{"label": "LABEL_1", "score": 0.502}])
+        self.assertEqual(nested_simplify(outputs), [{"label": "LABEL_0", "score": 0.504}])
 
     @require_tf
     def test_small_model_tf(self):
         text_classifier = pipeline(
-            task="text-classification", model="Narsil/tiny-distilbert-sequence-classification", framework="tf"
+            task="text-classification", model="hf-internal-testing/tiny-random-distilbert", framework="tf"
         )
 
         outputs = text_classifier("This is great !")
-        self.assertEqual(nested_simplify(outputs), [{"label": "LABEL_1", "score": 0.502}])
+        self.assertEqual(nested_simplify(outputs), [{"label": "LABEL_0", "score": 0.504}])
 
     @slow
     @require_torch

tests/test_pipelines_token_classification.py

@@ -582,14 +582,14 @@ def test_gather_pre_entities(self):
 
     @require_tf
     def test_tf_only(self):
-        model_name = "Narsil/small"  # This model only has a TensorFlow version
+        model_name = "hf-internal-testing/tiny-random-bert-tf-only"  # This model only has a TensorFlow version
         # We test that if we don't specificy framework='tf', it gets detected automatically
         token_classifier = pipeline(task="ner", model=model_name)
         self.assertEqual(token_classifier.framework, "tf")
 
     @require_tf
     def test_small_model_tf(self):
-        model_name = "Narsil/small2"
+        model_name = "hf-internal-testing/tiny-bert-for-token-classification"
         token_classifier = pipeline(task="token-classification", model=model_name, framework="tf")
         outputs = token_classifier("This is a test !")
         self.assertEqual(
@@ -602,8 +602,8 @@ def test_small_model_tf(self):
 
     @require_torch
     def test_no_offset_tokenizer(self):
-        model_name = "Narsil/small2"
-        tokenizer = AutoTokenizer.from_pretrained("Narsil/small2", use_fast=False)
+        model_name = "hf-internal-testing/tiny-bert-for-token-classification"
+        tokenizer = AutoTokenizer.from_pretrained(model_name, use_fast=False)
         token_classifier = pipeline(task="token-classification", model=model_name, tokenizer=tokenizer, framework="pt")
         outputs = token_classifier("This is a test !")
         self.assertEqual(
@@ -616,7 +616,7 @@ def test_no_offset_tokenizer(self):
 
     @require_torch
     def test_small_model_pt(self):
-        model_name = "Narsil/small2"
+        model_name = "hf-internal-testing/tiny-bert-for-token-classification"
         token_classifier = pipeline(task="token-classification", model=model_name, framework="pt")
         outputs = token_classifier("This is a test !")
         self.assertEqual(