Update app.py

app.py

@@ -48,12 +48,10 @@ def embed_product_data(product_data):
     color_embedding = color_embedding.view(1, -1)        # 2D로 변환
 
     # 최종 임베딩 벡터 결합
-    product_embedding = torch.cat(
-        (text_embedding, category_embedding, color_embedding), dim=1
-    )
+    combined_embedding = torch.cat((text_embedding, category_embedding, color_embedding), dim=1)
+    product_embedding = torch.nn.functional.adaptive_avg_pool1d(combined_embedding.unsqueeze(0), 512).squeeze(0)
 
     print(f"Generated product_embedding shape: {product_embedding.shape}")  # Debugging
-    
     return product_embedding.detach().numpy()
 
 # 사용자 타워: 데이터 임베딩
@@ -76,13 +74,16 @@ def embed_user_data(user_data):
     scaled_height = (height - min_height) * 99 // (max_height - min_height)
     scaled_weight = (weight - min_weight) * 99 // (max_weight - min_weight)
     
-    age_embedding = embedding_layer(torch.tensor([user_data['age']]))
-    gender_embedding = embedding_layer(torch.tensor([gender_id]))
-    height_embedding = embedding_layer(torch.tensor([scaled_height]))
-    weight_embedding = embedding_layer(torch.tensor([scaled_weight]))
+    age_embedding = embedding_layer(torch.tensor([user_data['age']])).view(1, -1)
+    gender_embedding = embedding_layer(torch.tensor([gender_id])).view(1, -1)
+    height_embedding = embedding_layer(torch.tensor([scaled_height])).view(1, -1)
+    weight_embedding = embedding_layer(torch.tensor([scaled_weight])).view(1, -1)
 
     # 최종 임베딩 벡터 결합
-    user_embedding = torch.cat((age_embedding, gender_embedding, height_embedding, weight_embedding), dim=1)
+    combined_embedding = torch.cat((age_embedding, gender_embedding, height_embedding, weight_embedding), dim=1)
+    user_embedding = torch.nn.functional.adaptive_avg_pool1d(combined_embedding.unsqueeze(0), 512).squeeze(0)
+
+    print(f"Generated user_embedding shape: {reduced_embedding.shape}")
     return user_embedding.detach().numpy()
 
 # MongoDB Atlas에서 데이터 가져오기
@@ -127,12 +128,7 @@ def recommend_products_for_user(user_id, top_n=1):
             print(f"User ID {user_id} embedding not found.")
             return []
 
-        # user_embedding = np.array(user_embedding_data["embedding"]).reshape(1, -1)
-        user_embedding = np.array(user_embedding_data["embedding"])
-
-        # 차원 확인 및 조정
-        if user_embedding.ndim == 1:  # 1D 배열인 경우 2D로 변환
-            user_embedding = user_embedding.reshape(1, -1)
+        user_embedding = np.array(user_embedding_data["embedding"]).reshape(1, -1)
 
         # 모든 상품 임베딩 가져오기
         all_product_embeddings = list(product_embedding_collection.find())