Update app.py

app.py

@@ -47,7 +47,7 @@ def read_json_from_s3():
     
     return json_content
 
-def get_titan_embedding(bedrock, doc_name, text, attempt=0, cutoff=10000):
+def get_titan_embedding(bedrock_client, doc_name, text, attempt=0, cutoff=10000):
     """
     Retrieves a text embedding for a given document using the Amazon Titan Embedding model.
 
@@ -77,7 +77,7 @@ def get_titan_embedding(bedrock, doc_name, text, attempt=0, cutoff=10000):
         })
 
         # Invoke model 
-        response = bedrock.invoke_model(
+        response = bedrock_client.invoke_model(
             body=body, 
             modelId=model_id, 
             accept=accept, 
@@ -96,14 +96,14 @@ def get_titan_embedding(bedrock, doc_name, text, attempt=0, cutoff=10000):
                 
             delay = 2 ** (attempt + 1);               
             time.sleep(delay)
-            return get_titan_embedding(doc_name, text, attempt=attempt + 1)
+            return get_titan_embedding(bedrock_client, doc_name, text, attempt=attempt + 1)
             
         elif error.response['Error']['Code'] == 'ValidationException':
             # get chunks of text length 20000 characters
             text_chunks = [text[i:i+cutoff] for i in range(0, len(text), cutoff)]
             embeddings = []
             for chunk in text_chunks:
-                embeddings.append(get_titan_embedding(bedrock, doc_name, chunk))
+                embeddings.append(get_titan_embedding(bedrock_client, doc_name, chunk))
             
             # return the average of the embeddinngs
             return np.mean(embeddings, axis=0)
@@ -129,8 +129,6 @@ def ask_ds(message, history):
 
     # RAG
     question_embedding = get_titan_embedding(bedrock_client, 'question', question)
-
-    yield f'question: {question}\nmessage: {message}\nquestion embed: {question_embedding}'
     
     similar_documents = []
     for file, data in extractions.items():