Files
DB-GPT/pilot/model/llm_out/gpt4all_llm.py
2023-06-13 14:35:14 +08:00

24 lines
568 B
Python

#!/usr/bin/env python3
# -*- coding:utf-8 -*-
import threading
import sys
import time
def gpt4all_generate_stream(model, tokenizer, params, device, max_position_embeddings):
stop = params.get("stop", "###")
prompt = params["prompt"]
role, query = prompt.split(stop)[1].split(":")
print(f"gpt4all, role: {role}, query: {query}")
def worker():
model.generate(prompt=query, streaming=True)
t = threading.Thread(target=worker)
t.start()
while t.is_alive():
yield sys.stdout.output
time.sleep(0.01)
t.join()