Add ONNX support for ResNet (#17585)

* Add ONNX support for ResNet

* Add ONNX test

* make fix-copies

docs/source/en/serialization.mdx

@@ -72,6 +72,7 @@ Ready-made configurations include the following architectures:
 - OpenAI GPT-2
 - Perceiver
 - PLBart
+- ResNet
 - RoBERTa
 - RoFormer
 - SqueezeBERT

src/transformers/models/resnet/__init__.py

@@ -21,7 +21,9 @@ from typing import TYPE_CHECKING
 from ...utils import OptionalDependencyNotAvailable, _LazyModule, is_torch_available
 
 
-_import_structure = {"configuration_resnet": ["RESNET_PRETRAINED_CONFIG_ARCHIVE_MAP", "ResNetConfig"]}
+_import_structure = {
+    "configuration_resnet": ["RESNET_PRETRAINED_CONFIG_ARCHIVE_MAP", "ResNetConfig", "ResNetOnnxConfig"]
+}
 
 try:
     if not is_torch_available():
@@ -38,7 +40,7 @@ else:
 
 
 if TYPE_CHECKING:
-    from .configuration_resnet import RESNET_PRETRAINED_CONFIG_ARCHIVE_MAP, ResNetConfig
+    from .configuration_resnet import RESNET_PRETRAINED_CONFIG_ARCHIVE_MAP, ResNetConfig, ResNetOnnxConfig
 
     try:
         if not is_torch_available():

src/transformers/models/resnet/configuration_resnet.py

@@ -14,7 +14,13 @@
 # limitations under the License.
 """ ResNet model configuration"""
 
+from collections import OrderedDict
+from typing import Mapping
+
+from packaging import version
+
 from ...configuration_utils import PretrainedConfig
+from ...onnx import OnnxConfig
 from ...utils import logging
 
 
@@ -89,3 +95,20 @@ class ResNetConfig(PretrainedConfig):
         self.layer_type = layer_type
         self.hidden_act = hidden_act
         self.downsample_in_first_stage = downsample_in_first_stage
+
+
+class ResNetOnnxConfig(OnnxConfig):
+
+    torch_onnx_minimum_version = version.parse("1.11")
+
+    @property
+    def inputs(self) -> Mapping[str, Mapping[int, str]]:
+        return OrderedDict(
+            [
+                ("pixel_values", {0: "batch", 1: "sequence"}),
+            ]
+        )
+
+    @property
+    def atol_for_validation(self) -> float:
+        return 1e-3

src/transformers/onnx/features.py

@@ -318,6 +318,11 @@ class FeaturesManager:
             "sequence-classification",
             onnx_config_cls="models.perceiver.PerceiverOnnxConfig",
         ),
+        "resnet": supported_features_mapping(
+            "default",
+            "image-classification",
+            onnx_config_cls="models.resnet.ResNetOnnxConfig",
+        ),
         "roberta": supported_features_mapping(
             "default",
             "masked-lm",

tests/onnx/test_onnx_v2.py

@@ -182,6 +182,7 @@ PYTORCH_EXPORT_MODELS = {
     ("convbert", "YituTech/conv-bert-base"),
     ("distilbert", "distilbert-base-cased"),
     ("electra", "google/electra-base-generator"),
+    ("resnet", "microsoft/resnet-50"),
     ("roberta", "roberta-base"),
     ("roformer", "junnyu/roformer_chinese_base"),
     ("squeezebert", "squeezebert/squeezebert-uncased"),