Upload model

config.json

@@ -0,0 +1,13 @@
+{
+  "architectures": [
+    "ResnetModelForImageClassification"
+  ],
+  "auto_map": {
+    "AutoConfig": "configuration_resnet.ResnetFeatureExtractorConfig",
+    "AutoModelForImageClassification": "model_resnet.ResnetModelForImageClassification"
+  },
+  "model_type": "resnet",
+  "name": "resnet152",
+  "torch_dtype": "float32",
+  "transformers_version": "4.27.1"
+}

configuration_resnet.py

@@ -0,0 +1,11 @@
+from transformers import PretrainedConfig
+
+class ResnetFeatureExtractorConfig(PretrainedConfig):
+    model_type = "resnet"
+
+    def __init__(self, name = 'resnet152', **kwargs):
+        if name != 'resnet152':
+            raise ValueError(f"`name` must be 'resnet152', got {name}.")
+
+        self.name = name
+        super().__init__(**kwargs)

model_resnet.py

@@ -0,0 +1,47 @@
+from transformers import PreTrainedModel
+import torch
+import torch.nn as nn
+from torchvision import transforms
+from transformers.models.mvp.modeling_mvp import CrossEntropyLoss
+
+from .configuration_resnet import ResnetFeatureExtractorConfig
+
+class ResnetFeatureExtractor(PreTrainedModel):
+    config_class = ResnetFeatureExtractorConfig
+
+    def __init__(self, config):
+        super().__init__(config)
+        
+        if config.name == 'resnet152':
+            self.model = torch.hub.load('pytorch/vision:v0.10.0', 'resnet152', pretrained=False)
+            self.model.fc = nn.Identity()
+            self.model.to(self.device)
+            self.preprocess = transforms.Compose([
+                transforms.Resize(256),
+                transforms.CenterCrop(224),
+                transforms.ToTensor(),
+                transforms.Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225]),
+            ])
+
+    def forward(self, images):
+        tensor = torch.stack([self.preprocess(image) for image in images]).to(self.device).float()
+        return self.model(tensor)
+
+class ResnetModelForImageClassification(PreTrainedModel):
+    config_class = ResnetFeatureExtractorConfig
+
+    def __init__(self, config):
+        super().__init__(config)
+        if config.name == 'resnet152':
+          self.model = nn.Sequential(
+              nn.Linear(2048, 32),
+              nn.ReLU(),
+              nn.Linear(32, 2)
+          )
+
+    def forward(self, tensor, labels=None):
+        logits = self.model(tensor)
+        if labels is not None:
+            loss = CrossEntropyLoss()(logits, torch.tensor(labels))
+            return {"loss": loss, "logits": logits}
+        return {"logits": logits}

pytorch_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0b809790a34c7cf9a393c984d6ccdf0049dd07b03a37469a40c1527593da1b43
+size 264131