Removed print statements, increased batch size

src/CLIP.py

@@ -16,13 +16,7 @@ class CLIPImageEncoder:
 
     def encode_images(self, batch):
         images = batch["image"]
-        print(images)
         input = self.processor(images=images, return_tensors="pt")
-        print(input)
         with torch.no_grad():
             image_features = self.model.get_image_features(**input)
-            #image_features = self.model(**input).last_hidden_state[:,0].cpu()
-            print(image_features)
-            print("--------------------")
-            print(self.model.get_image_features(**input).cpu().detach().numpy())
         return {"clip_embeddings": image_features.cpu().detach().numpy()}

src/app.py

@@ -19,7 +19,7 @@ def emb_dataset(dataset):
 
     ## CLIP Embeddings
     clip_model = CLIPImageEncoder()
-    dataset_with_embeddings = dataset_with_embeddings.map(clip_model.encode_images, batched=True, batch_size=8)
+    dataset_with_embeddings = dataset_with_embeddings.map(clip_model.encode_images, batched=True, batch_size=16)
 
     # Add index
     dataset_with_embeddings.add_faiss_index(column='color_embeddings')
@@ -32,7 +32,6 @@ def emb_dataset(dataset):
 dataset_with_embeddings = emb_dataset(candidate_subset)
 
 # Main function, to find similar images
-# TODO: allow different descriptor/embedding functions
 # TODO: implement different distance measures
 
 def get_neighbors(query_image, selected_descriptor, top_k=5):