Create handler.py

handler.py

@@ -0,0 +1,39 @@
+from typing import Dict, List, Any
+from PIL import Image
+import torch
+import base64
+from io import BytesIO
+from transformers import CLIPProcessor, CLIPModel
+
+# check for GPU
+device = 0 if torch.cuda.is_available() else -1
+
+class EndpointHandler():
+    def __init__(self, path=""):
+        self.model = CLIPModel.from_pretrained("openai/clip-vit-large-patch14")
+        self.processor = CLIPProcessor.from_pretrained("openai/clip-vit-large-patch14")
+
+    def __call__(self, data: Any) -> List[float]:
+        inputs = data.pop("inputs", data)
+        
+        if "image" in inputs:
+            # decode base64 image to PIL
+            image = Image.open(BytesIO(base64.b64decode(inputs['image'])))  
+            inputs = self.processor(images=image, text=None, return_tensors="pt", padding=True)
+            
+            image_embeds = self.model.get_image_features(
+                pixel_values=inputs["pixel_values"]
+            )
+    
+            return image_embeds[0].tolist()
+        if "text" in inputs:
+            text = inputs['text']
+            inputs = self.processor(images=None, text=text, return_tensors="pt", padding=True)
+            
+            text_embeds = self.model.get_text_features(
+                input_ids=inputs["input_ids"], attention_mask=inputs["attention_mask"]
+            )
+    
+            return text_embeds[0].tolist()
+
+        raise Exception("No 'image' or 'text' provided")