Refactor encoder.py to remove debug print statements and ensure inputs are moved to the correct device for text and image encoding.

src/encoder.py

@@ -25,11 +25,6 @@ class FashionCLIPEncoder:
         self.model.to(self.device)
         self.model.eval()
 
-        test = "baggy jeans"
-        print(test)
-        vector = self.encode_text([test])
-        print(vector)
-
     def encode_text(self, texts: List[str]) -> List[List[float]]:
         kwargs = {
             "padding": "max_length",
@@ -39,7 +34,7 @@ class FashionCLIPEncoder:
         inputs = self.processor(text=texts, **kwargs)
 
         with torch.no_grad():
-            batch = {k: v for k, v in inputs.items()}
+            batch = {k: v.to(self.device) for k, v in inputs.items()}
             return self._encode_text(batch)
 
     def encode_images(self, images: List[Image]) -> List[List[float]]:
@@ -49,7 +44,7 @@ class FashionCLIPEncoder:
         inputs = self.processor(images=images, **kwargs)
 
         with torch.no_grad():
-            batch = {k: v for k, v in inputs.items()}
+            batch = {k: v.to(self.device) for k, v in inputs.items()}
             return self._encode_images(batch)
 
     def _encode_text(self, batch: Dict) -> List[List[float]]: