Update tiny models and a few fixes (#22928)

* run_check_tiny_models

* update summary

* update mixin

* update pipeline_model_mapping

* update pipeline_model_mapping

* Update for gpt_bigcode

---------

Co-authored-by: ydshieh <ydshieh@users.noreply.github.com>

tests/models/gpt_bigcode/test_modeling_gpt_bigcode.py

@@ -407,7 +407,7 @@ class GPTBigCodeModelTester:
 
 
 @require_torch
-class GPTBigCodeMQAModelTest(ModelTesterMixin, GenerationTesterMixin, PipelineTesterMixin, unittest.TestCase):
+class GPTBigCodeModelTest(ModelTesterMixin, GenerationTesterMixin, PipelineTesterMixin, unittest.TestCase):
     # TODO: Update the tests to use valid pretrained models.
     all_model_classes = (
         (
@@ -420,11 +420,6 @@ class GPTBigCodeMQAModelTest(ModelTesterMixin, GenerationTesterMixin, PipelineTe
         else ()
     )
     all_generative_model_classes = (GPTBigCodeForCausalLM,) if is_torch_available() else ()
-    fx_compatible = False
-    test_missing_keys = False
-    test_pruning = False
-    test_torchscript = False
-    multi_query = True
     pipeline_model_mapping = (
         {
             "feature-extraction": GPTBigCodeModel,
@@ -436,6 +431,11 @@ class GPTBigCodeMQAModelTest(ModelTesterMixin, GenerationTesterMixin, PipelineTe
         if is_torch_available()
         else {}
     )
+    fx_compatible = False
+    test_missing_keys = False
+    test_pruning = False
+    test_torchscript = False
+    multi_query = True
 
     # special case for DoubleHeads model
     def _prepare_for_class(self, inputs_dict, model_class, return_labels=False):
@@ -521,7 +521,7 @@ class GPTBigCodeMQAModelTest(ModelTesterMixin, GenerationTesterMixin, PipelineTe
 
 
 @require_torch
-class GPTBigCodeMHAModelTest(GPTBigCodeMQAModelTest):
+class GPTBigCodeMHAModelTest(GPTBigCodeModelTest):
     # `parameterized_class` breaks with mixins, so we use inheritance instead
     multi_query = False
 

tests/models/sam/test_modeling_sam.py

@@ -26,6 +26,7 @@ from transformers.utils import is_torch_available, is_vision_available
 
 from ...test_configuration_common import ConfigTester
 from ...test_modeling_common import ModelTesterMixin, floats_tensor
+from ...test_pipeline_mixin import PipelineTesterMixin
 
 
 if is_torch_available():
@@ -282,19 +283,28 @@ class SamModelTester:
 
 
 @require_torch
-class SamModelTest(ModelTesterMixin, unittest.TestCase):
+class SamModelTest(ModelTesterMixin, PipelineTesterMixin, unittest.TestCase):
     """
     Here we also overwrite some of the tests of test_modeling_common.py, as SAM's vision encoder does not use input_ids, inputs_embeds,
     attention_mask and seq_length.
     """
 
     all_model_classes = (SamModel,) if is_torch_available() else ()
+    pipeline_model_mapping = (
+        {"feature-extraction": SamModel, "mask-generation": SamModel} if is_torch_available() else {}
+    )
     fx_compatible = False
     test_pruning = False
     test_resize_embeddings = False
     test_head_masking = False
     test_torchscript = False
 
+    # TODO: Fix me @Arthur: `run_batch_test` in `tests/test_pipeline_mixin.py` not working
+    def is_pipeline_test_to_skip(
+        self, pipeline_test_casse_name, config_class, model_architecture, tokenizer_name, processor_name
+    ):
+        return True
+
     def setUp(self):
         self.model_tester = SamModelTester(self)
         self.vision_config_tester = ConfigTester(self, config_class=SamVisionConfig, has_text_modality=False)

tests/pipelines/test_pipelines_mask_generation.py

@@ -32,6 +32,12 @@ from transformers.testing_utils import (
 
 if is_vision_available():
     from PIL import Image
+else:
+
+    class Image:
+        @staticmethod
+        def open(*args, **kwargs):
+            pass
 
 
 def hashimage(image: Image) -> str:
@@ -60,6 +66,10 @@ class MaskGenerationPipelineTests(unittest.TestCase):
             "./tests/fixtures/tests_samples/COCO/000000039769.png",
         ]
 
+    # TODO: Fix me @Arthur
+    def run_pipeline_test(self, mask_generator, examples):
+        pass
+
     @require_tf
     @unittest.skip("Image segmentation not implemented in TF")
     def test_small_model_tf(self):

tests/pipelines/test_pipelines_question_answering.py

@@ -44,9 +44,11 @@ class QAPipelineTests(unittest.TestCase):
     tf_model_mapping = TF_MODEL_FOR_QUESTION_ANSWERING_MAPPING
 
     if model_mapping is not None:
-        model_mapping = {config: model for config, model in model_mapping.items() if config.__name__ in _TO_SKIP}
+        model_mapping = {config: model for config, model in model_mapping.items() if config.__name__ not in _TO_SKIP}
     if tf_model_mapping is not None:
-        tf_model_mapping = {config: model for config, model in tf_model_mapping.items() if config.__name__ in _TO_SKIP}
+        tf_model_mapping = {
+            config: model for config, model in tf_model_mapping.items() if config.__name__ not in _TO_SKIP
+        }
 
     def get_test_pipeline(self, model, tokenizer, processor):
         if isinstance(model.config, LxmertConfig):

tests/pipelines/test_pipelines_text_classification.py

@@ -35,9 +35,11 @@ class TextClassificationPipelineTests(unittest.TestCase):
     tf_model_mapping = TF_MODEL_FOR_SEQUENCE_CLASSIFICATION_MAPPING
 
     if model_mapping is not None:
-        model_mapping = {config: model for config, model in model_mapping.items() if config.__name__ in _TO_SKIP}
+        model_mapping = {config: model for config, model in model_mapping.items() if config.__name__ not in _TO_SKIP}
     if tf_model_mapping is not None:
-        tf_model_mapping = {config: model for config, model in tf_model_mapping.items() if config.__name__ in _TO_SKIP}
+        tf_model_mapping = {
+            config: model for config, model in tf_model_mapping.items() if config.__name__ not in _TO_SKIP
+        }
 
     @require_torch
     def test_small_model_pt(self):

tests/pipelines/test_pipelines_token_classification.py

@@ -49,9 +49,11 @@ class TokenClassificationPipelineTests(unittest.TestCase):
     tf_model_mapping = TF_MODEL_FOR_TOKEN_CLASSIFICATION_MAPPING
 
     if model_mapping is not None:
-        model_mapping = {config: model for config, model in model_mapping.items() if config.__name__ in _TO_SKIP}
+        model_mapping = {config: model for config, model in model_mapping.items() if config.__name__ not in _TO_SKIP}
     if tf_model_mapping is not None:
-        tf_model_mapping = {config: model for config, model in tf_model_mapping.items() if config.__name__ in _TO_SKIP}
+        tf_model_mapping = {
+            config: model for config, model in tf_model_mapping.items() if config.__name__ not in _TO_SKIP
+        }
 
     def get_test_pipeline(self, model, tokenizer, processor):
         token_classifier = TokenClassificationPipeline(model=model, tokenizer=tokenizer)

tests/pipelines/test_pipelines_zero_shot.py

@@ -36,9 +36,11 @@ class ZeroShotClassificationPipelineTests(unittest.TestCase):
     tf_model_mapping = TF_MODEL_FOR_SEQUENCE_CLASSIFICATION_MAPPING
 
     if model_mapping is not None:
-        model_mapping = {config: model for config, model in model_mapping.items() if config.__name__ in _TO_SKIP}
+        model_mapping = {config: model for config, model in model_mapping.items() if config.__name__ not in _TO_SKIP}
     if tf_model_mapping is not None:
-        tf_model_mapping = {config: model for config, model in tf_model_mapping.items() if config.__name__ in _TO_SKIP}
+        tf_model_mapping = {
+            config: model for config, model in tf_model_mapping.items() if config.__name__ not in _TO_SKIP
+        }
 
     def get_test_pipeline(self, model, tokenizer, processor):
         classifier = ZeroShotClassificationPipeline(

tests/test_pipeline_mixin.py

@@ -40,6 +40,7 @@ from .pipelines.test_pipelines_fill_mask import FillMaskPipelineTests
 from .pipelines.test_pipelines_image_classification import ImageClassificationPipelineTests
 from .pipelines.test_pipelines_image_segmentation import ImageSegmentationPipelineTests
 from .pipelines.test_pipelines_image_to_text import ImageToTextPipelineTests
+from .pipelines.test_pipelines_mask_generation import MaskGenerationPipelineTests
 from .pipelines.test_pipelines_object_detection import ObjectDetectionPipelineTests
 from .pipelines.test_pipelines_question_answering import QAPipelineTests
 from .pipelines.test_pipelines_summarization import SummarizationPipelineTests
@@ -68,6 +69,7 @@ pipeline_test_mapping = {
     "image-classification": {"test": ImageClassificationPipelineTests},
     "image-segmentation": {"test": ImageSegmentationPipelineTests},
     "image-to-text": {"test": ImageToTextPipelineTests},
+    "mask-generation": {"test": MaskGenerationPipelineTests},
     "object-detection": {"test": ObjectDetectionPipelineTests},
     "question-answering": {"test": QAPipelineTests},
     "summarization": {"test": SummarizationPipelineTests},
@@ -355,6 +357,12 @@ class PipelineTesterMixin:
     def test_pipeline_image_to_text(self):
         self.run_task_tests(task="image-to-text")
 
+    @is_pipeline_test
+    @require_vision
+    @require_torch
+    def test_pipeline_mask_generation(self):
+        self.run_task_tests(task="mask-generation")
+
     @is_pipeline_test
     @require_vision
     @require_timm

tests/utils/tiny_model_summary.json

@@ -2294,6 +2294,50 @@
         ],
         "sha": "d6694b0d8fe17978761c9305dc151780506b192e"
     },
+    "GPTBigCodeForCausalLM": {
+        "tokenizer_classes": [
+            "GPT2Tokenizer",
+            "GPT2TokenizerFast"
+        ],
+        "processor_classes": [],
+        "model_classes": [
+            "GPTBigCodeForCausalLM"
+        ],
+        "sha": "99f7aaadf9c29669c63ef6c16f6bc5c07dbb9126"
+    },
+    "GPTBigCodeForSequenceClassification": {
+        "tokenizer_classes": [
+            "GPT2Tokenizer",
+            "GPT2TokenizerFast"
+        ],
+        "processor_classes": [],
+        "model_classes": [
+            "GPTBigCodeForSequenceClassification"
+        ],
+        "sha": "64a7398d5763161037b818314c60dd83d93d03e9"
+    },
+    "GPTBigCodeForTokenClassification": {
+        "tokenizer_classes": [
+            "GPT2Tokenizer",
+            "GPT2TokenizerFast"
+        ],
+        "processor_classes": [],
+        "model_classes": [
+            "GPTBigCodeForTokenClassification"
+        ],
+        "sha": "310537ecd22d45f71bf594b17922cf2abc338eaf"
+    },
+    "GPTBigCodeModel": {
+        "tokenizer_classes": [
+            "GPT2Tokenizer",
+            "GPT2TokenizerFast"
+        ],
+        "processor_classes": [],
+        "model_classes": [
+            "GPTBigCodeModel"
+        ],
+        "sha": "3069419084a9dc36802d47de9df3d314ccfc2f28"
+    },
     "GPTJForCausalLM": {
         "tokenizer_classes": [
             "GPT2Tokenizer",
@@ -2385,6 +2429,16 @@
         ],
         "sha": "0229cfaaa843c6b492ac2abffabb00f1ff1936f8"
     },
+    "GPTNeoXForSequenceClassification": {
+        "tokenizer_classes": [
+            "GPTNeoXTokenizerFast"
+        ],
+        "processor_classes": [],
+        "model_classes": [
+            "GPTNeoXForSequenceClassification"
+        ],
+        "sha": "17c4b845ee2e0bb780ca2dea2d59a3d9d5d3c651"
+    },
     "GPTNeoXJapaneseForCausalLM": {
         "tokenizer_classes": [
             "GPTNeoXJapaneseTokenizer"
@@ -4940,6 +4994,16 @@
         ],
         "sha": "0a0fbb844eeefa0dce62bd05db30a2bb91e5dc88"
     },
+    "SamModel": {
+        "tokenizer_classes": [],
+        "processor_classes": [
+            "SamImageProcessor"
+        ],
+        "model_classes": [
+            "SamModel"
+        ],
+        "sha": "eca8651bc84e5ac3b1b62e784b744a6bd1b82575"
+    },
     "SegformerForImageClassification": {
         "tokenizer_classes": [],
         "processor_classes": [