os.chdir('single_interact/')
a = ['hello']
def single_response(user_input):
reply, history = my_model.interact_single(message=user_input, history=a)
for i in history :
a.append(i)
return reply
print("\n===============================================")
print("================= Conv AI.V ===================")
print("===============================================\n")
while 1:
talk = input(colored("You: ",'green'))
if talk=='exit':break
response = single_response(talk)
myset = set(a)
a = list(myset)
#print('Bot : ',response)
print(colored("Bot:",'red'),response)
#print('History of re : ',a)`
GPT2 custom chat-bot single interaction. I am not using any thread in this implementation but still it says 'multiprocessingchunksize'. Here we try to build chatbot using gpt2 and interact with our custom model which is trained by custom dataset(type : persona). 
`#from bot_make.utils import helper import shutil from simpletransformers.conv_ai import ConvAIModel import os import pickle from termcolor import colored import requests
def cache_select(cache_name): shutil.unpack_archive('base_model/'+cache_name,'./single_interact/cache_dir','zip') print('done')
def cache_down(url): try: r = requests.get(url, allow_redirects=True) open('single_interact/cache_dir.zip', 'wb').write(r.content) shutil.unpack_archive('single_interact/cache_dir.zip','./single_interact/cache_dir','zip') os.remove(f"single_interact/cache_dir.zip") return True except: return False
def model_down(url): try: r = requests.get(url, allow_redirects=True) open('single_interact/for_load.tar.gz', 'wb').write(r.content) return True except: return False
print('Model and Cache down') cache_ck = cache_down('https://bucket.s3.amazonaws.com/generated_data_1_model_2_cache_dir.zip') model_ck = model_down('https://bucket.s3.amazonaws.com/generated_data_1_model_2.tar.gz')
print('Start BOT') def model_load(model_name): return pickle.load(open('single_interact/'+model_name, 'rb'))
cache_select('base_model_small_cache_dir.zip')
my_model = model_load('for_load.tar.gz')
my_model.interact()
os.chdir('single_interact/') a = ['hello'] def single_response(user_input): reply, history = my_model.interact_single(message=user_input, history=a) for i in history : a.append(i)
return reply
print("\n===============================================") print("================= Conv AI.V ===================") print("===============================================\n") while 1: talk = input(colored("You: ",'green')) if talk=='exit':break response = single_response(talk)