Add threading for open ai api

app.py

@@ -8,6 +8,7 @@ import re
 import platform
 import openai
 import random
+import concurrent.futures
 from transformers import (
     AutoModelForCausalLM,
     AutoTokenizer,
@@ -130,7 +131,7 @@ def extract_db_code(text):
 
 def generate_dummy_db(db_info, question):
     pre_prompt = "Generate a SQLite database with dummy data for this database from the DB Layout. Make sure you add dummy data relevant to the Question and don't write any SELECT statements or actual queries."
-    prompt = pre_prompt + "\n\nDB Layout:" db_info + "\n\nQuestion: " + question
+    prompt = pre_prompt + "\n\nDB Layout:" + db_info + "\n\nQuestion: " + question
 
     while True:
         try:
@@ -216,11 +217,15 @@ def generate(input_message: str, db_info="", temperature=0.2, top_p=0.9, top_k=0
         do_sample=do_sample,
     )
 
+    db_code_future = None
+    if num_return_sequences > 1:
+        with concurrent.futures.ThreadPoolExecutor() as executor:
+            db_code_future = executor.submit(generate_dummy_db, db_info, input_message)
+    
     tokens = m.generate(**generate_kwargs)
 
-    db_code = None
-    if (num_return_sequences > 1):
-        db_code = generate_dummy_db(db_info, input_message)
+    if db_code_future:
+        db_code = db_code_future.result()
 
     responses = []
     for response in tokens: