Spaces:
Runtime error
Runtime error
File size: 3,040 Bytes
77e664f |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 |
from typing import List, Tuple, Dict, Generator
from langchain.llms import OpenAI
import gradio as gr
model_name = "gpt-3.5-turbo"
LLM = OpenAI(model_name=model_name, temperature=0.1)
def create_history_messages(history: List[Tuple[str, str]]) -> List[dict]:
history_messages = [{"role": "user", "content": m[0]} for m in history]
history_messages.extend([{"role": "assistant", "content": m[1]} for m in history])
return history_messages
def create_formatted_history(history_messages: List[dict]) -> List[Tuple[str, str]]:
formatted_history = []
user_messages = []
assistant_messages = []
for message in history_messages:
if message["role"] == "user":
user_messages.append(message["content"])
elif message["role"] == "assistant":
assistant_messages.append(message["content"])
if user_messages and assistant_messages:
formatted_history.append(
("".join(user_messages), "".join(assistant_messages))
)
user_messages = []
assistant_messages = []
# append any remaining messages
if user_messages:
formatted_history.append(("".join(user_messages), None))
elif assistant_messages:
formatted_history.append((None, "".join(assistant_messages)))
return formatted_history
def chat(
message: str, state: List[Dict[str, str]], client = LLM.client
) -> Generator[Tuple[List[Tuple[str, str]], List[Dict[str, str]]], None, None]:
history_messages = state
if history_messages == None:
history_messages = []
history_messages.append({"role": "system", "content": "A helpful assistant."})
history_messages.append({"role": "user", "content": message})
# We have no content for the assistant's response yet but we will update this:
history_messages.append({"role": "assistant", "content": ""})
response_message = ""
chat_generator = client.create(
messages=history_messages, stream=True, model=model_name
)
for chunk in chat_generator:
if "choices" in chunk:
for choice in chunk["choices"]:
if "delta" in choice and "content" in choice["delta"]:
new_token = choice["delta"]["content"]
# Add the latest token:
response_message += new_token
# Update the assistant's response in our model:
history_messages[-1]["content"] = response_message
if "finish_reason" in choice and choice["finish_reason"] == "stop":
break
formatted_history = create_formatted_history(history_messages)
yield formatted_history, history_messages
chatbot = gr.Chatbot(label="Chat").style(color_map=("yellow", "purple"))
iface = gr.Interface(
fn=chat,
inputs=[
gr.Textbox(placeholder="Hello! How are you? etc.", label="Message"),
"state",
],
outputs=[chatbot, "state"],
allow_flagging="never",
)
iface.queue().launch() |