vbatts/llama.cpp
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

move multiprocessing import into function scope

Browse source
This commit is contained in:
Concedo 2023-12-19 15:56:58 +08:00
parent 6948da5a0d
commit 1f77d2ad73
1 changed files with 2 additions and 2 deletions
Download patch file Download diff file Expand all files Collapse all files

4
koboldcpp.py
View file

@ -13,7 +13,6 @@ import os
import argparse
import json, sys, http.server, time, asyncio, socket, threading
from concurrent.futures import ThreadPoolExecutor
import multiprocessing
sampler_order_max = 7
stop_token_max = 16
@ -2342,8 +2341,9 @@ def run_in_queue(launch_args, input_queue, output_queue):
(args, kwargs) = data['data']
output_queue.put({'command': 'generated text', 'data': generate(*args, **kwargs)})
time.sleep(0.2)
def start_in_seperate_process(launch_args):
import multiprocessing
input_queue = multiprocessing.Queue()
output_queue = multiprocessing.Queue()
p = multiprocessing.Process(target=run_in_queue, args=(launch_args, input_queue, output_queue))