Update app.py

app.py

@@ -44,13 +44,18 @@ def predict(input, history=[]):
 
     if is_question: 
         sql_encoding = sql_tokenizer(table=table, query=input + sql_tokenizer.eos_token, return_tensors="pt")
-        sql_outputs = sql_model.generate(**sql_encoding)
-        sql_response = sql_tokenizer.batch_decode(sql_outputs, skip_special_tokens=True)
-
-        # Append the SQL model's response to the history
-        sql_response_ids = tokenizer.encode(sql_response + tokenizer.eos_token, return_tensors='pt')
-        history.extend(sql_response_ids[0].tolist())  # Add SQL response token IDs to history
+        #sql_outputs = sql_model.generate(**sql_encoding)
+        #sql_response = sql_tokenizer.batch_decode(sql_outputs, skip_special_tokens=True)
 
+        # append the new user input tokens to the chat history
+        bot_input_ids = torch.cat([torch.LongTensor(history), sql_encoding], dim=-1)
+          
+        # generate a response
+        history = sql_model.generate(bot_input_ids, max_length=1000, pad_token_id=sql_tokenizer.eos_token_id).tolist()
+    
+        # convert the tokens to text, and then split the responses into the right format
+        response = sql_tokenizer.decode(history[0]).split("<|endoftext|>")
+        response = [(response[i], response[i + 1]) for i in range(0, len(response) - 1, 2)]  # convert to tuples of list
         
         '''
         bot_input_ids = torch.cat([torch.LongTensor(history), sql_encoding], dim=-1)
@@ -61,15 +66,6 @@ def predict(input, history=[]):
     else:
         # tokenize the new input sentence
         new_user_input_ids = tokenizer.encode(input + tokenizer.eos_token, return_tensors='pt')
-
-        sql_encoding = sql_tokenizer(table=table, query=input + sql_tokenizer.eos_token, return_tensors="pt")
-        sql_outputs = sql_model.generate(**sql_encoding)
-        sql_response = sql_tokenizer.batch_decode(sql_outputs, skip_special_tokens=True)
-
-        # Append the SQL model's response to the history
-        sql_response_ids = tokenizer.encode(sql_response + tokenizer.eos_token, return_tensors='pt')
-        history.extend(sql_response_ids[0].tolist())  # Add SQL response token IDs to history
-
         
         # append the new user input tokens to the chat history
         bot_input_ids = torch.cat([torch.LongTensor(history), new_user_input_ids], dim=-1)