Update app.py

app.py

@@ -101,7 +101,7 @@ def submit_message(prompt, prompt_template, temperature, max_tokens, context_len
 # completion = openai.ChatCompletion.create(model="gpt-3.5-turbo", messages=system_prompt + history[-context_length*2:] + [prompt_msg], temperature=temperature, max_tokens=max_tokens)
 
         completion = ChatVectorDBChain.from_llm(OpenAI(temperature=temperature, max_tokens=max_tokens, model_name="gpt-3.5-turbo"), vectordb, return_source_documents=True)
-        result = completion({"question": system_prompt + history[-context_length*2:] + [prompt_msg], "chat_history": chat_history})
+        result = completion({"question": system_prompt + [prompt_msg], "chat_history": history[-context_length*2:]})
         # from https://blog.devgenius.io/chat-with-document-s-using-openai-chatgpt-api-and-text-embedding-6a0ce3dc8bc8
         
         history.append(prompt_msg)
@@ -162,7 +162,7 @@ with gr.Blocks(css=css) as demo:
                 prompt_template_preview = gr.Markdown(elem_id="prompt_template_preview")
                 with gr.Accordion("Advanced parameters", open=False):
                     temperature = gr.Slider(minimum=0, maximum=2.0, value=0.7, step=0.1, label="Flexibility", info="Higher = more creative/chaotic, Lower = just the guru")
-                    max_tokens = gr.Slider(minimum=100, maximum=400, value=400, step=1, label="Max tokens per response")
+                    max_tokens = gr.Slider(minimum=100, maximum=400, value=200, step=1, label="Max tokens per response")
                     context_length = gr.Slider(minimum=1, maximum=5, value=2, step=1, label="Context length", info="Number of previous questions you have asked. Be careful with high values, it can blow up the token budget quickly.")