File size: 7,866 Bytes
04bdc20 e3788ae 04bdc20 7fb1b6e 04bdc20 50df3a5 7fb1b6e 04bdc20 e3788ae 04bdc20 fbd8943 04bdc20 e3788ae 04bdc20 e3788ae 04bdc20 e3788ae 04bdc20 21f1de2 04bdc20 21f1de2 04bdc20 b1d9002 04bdc20 21f1de2 04bdc20 21f1de2 e3788ae 04bdc20 21f1de2 04bdc20 21f1de2 04bdc20 21f1de2 04bdc20 21f1de2 04bdc20 |
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 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 |
import openai
import tiktoken
import collections
import threading
import datetime
import time
import pytz
import json
import os
openai.api_key = os.getenv('API_KEY')
timezone = pytz.timezone('Asia/Shanghai')
timestamp2string = lambda timestamp: datetime.datetime.fromtimestamp(timestamp).astimezone(timezone).strftime('%Y-%m-%d %H:%M:%S')
def num_tokens_from_messages(messages, model="gpt-3.5-turbo"):
"""Returns the number of tokens used by a list of messages."""
try:
encoding = tiktoken.encoding_for_model(model)
except KeyError:
encoding = tiktoken.get_encoding("cl100k_base")
if model == "gpt-3.5-turbo": # note: future models may deviate from this
num_tokens = 0
for message in messages:
num_tokens += 4 # every message follows <im_start>{role/name}\n{content}<im_end>\n
for key, value in message.items():
num_tokens += len(encoding.encode(value))
if key == "name": # if there's a name, the role is omitted
num_tokens += -1 # role is always required and always 1 token
num_tokens += 2 # every reply is primed with <im_start>assistant
return num_tokens
else:
raise NotImplementedError(f"""num_tokens_from_messages() is not presently implemented for model {model}.
See https://github.com/openai/openai-python/blob/main/chatml.md for information on how messages are converted to tokens.""")
qas = []
qs = []
start_time = time.time()
def read_qs(qs, qas):
directory = "./dialogues_set"
filenames = [
'dialogues_film.json',
'dialogues_jindong.json',
'dialogues_music.json',
'dialogues_natural.json',
'dialogues_taobao.json',
'dialogues_travel_kd.json'
]
for filename in filenames:
with open(f"{directory}/{filename}", "r", encoding="utf-8") as f:
for idx,line in enumerate(f):
idx2query = json.loads(line)
query = idx2query[str(idx)]
qs.append(query)
print(f"read {len(qs)} queries from files")
if os.path.exists("qas.json"):
with open("qas.json", "r", encoding="utf-8") as f:
qas = json.loads(f.read())
print(f"read {len(qas)} query-responses from qas.json")
existed_qs = collections.Counter([qa["q"] for qa in qas])
remained_qs = []
for q in qs:
if existed_qs[q]>0:
existed_qs[q] -= 1
else:
remained_qs.append(q)
print(f"filter out {len(qs)-len(remained_qs)} with reference to qas.json")
qs = remained_qs
return qs
def ask(query, max_attempt_times=3):
answer = None
attempt_times = 0
while answer is None and attempt_times<max_attempt_times:
attempt_times += 1
try:
response = openai.ChatCompletion.create(
model="gpt-3.5-turbo",
messages=[
{"role": "user", "content": query}
]
)["choices"][0]["message"]["content"]
except Exception as e:
print(e)
print(f"retry in {attempt_times*10} seconds...")
time.sleep(attempt_times*10)
return answer
def askingChatGPT(qs, qas, min_interval_seconds=10):
read_qs(qs, qas)
for i, q in enumerate(qs):
ask_start_time = time.time()
a = ask(q)
qas.append({"q":q, "a":a})
ask_end_time = time.time()
elapsed_time = ask_end_time - ask_start_time
delayTime = min_interval_seconds - elapsed_time
if delayTime>0:
time.sleep(delayTime)
print(f"{timestamp2string(time.time())}: iterations: {i} / {len(qs)} | elapsed time of this query (s): {elapsed_time:.2f}")
return
thread = threading.Thread(target=lambda :askingChatGPT(qs, qas))
thread.daemon = True
thread.start()
import gradio as gr
def showcase(access_key):
if not access_key==os.getenv('access_key'):
chatbot_ret = [(f"Your entered Access Key:<br>{access_key}<br>is incorrect.", f"So i cannot provide you any information in this private space.")]
else:
recent_qas = qas[:10]
chatbot_ret = [(f"Your entered Access Key is correct.", f"The latest {len(recent_qas)} query-responses are displayed below.")]
for qa in recent_qas:
chatbot_ret += [(qa["q"].replace("\n","<br>"), qa["a"].replace("\n","<br>"))]
return chatbot_ret
def download(access_key):
if not access_key==os.getenv('access_key'):
chatbot_ret = [(f"Your entered Access Key:<br>{access_key}<br>is incorrect.", f"So i cannot provide you any information in this private space.")]
file_ret = gr.File.update(value=None, visible=False)
else:
chatbot_ret = [(f"Your entered Access Key is correct.", f"The file containing all processed query-responses ({len(qas)} in total) can be downloaded below.")]
filename = f"qas-{len(qas)}.json"
with open(filename, "w", encoding="utf-8") as f:
f.write(json.dumps(qas, ensure_ascii=False, indent=2))
file_ret = gr.File.update(value=filename, visible=True)
return chatbot_ret, file_ret
def display(access_key):
if not access_key==os.getenv('access_key'):
chatbot_ret = [(f"Your entered Access Key:<br>{access_key}<br>is incorrect.", f"So i cannot provide you any information in this private space.")]
elif len(qas)<1:
chatbot_ret = [(f"Your entered Access Key is correct.", f"But the progress has just started for a while and has no useful progress information to provide.")]
else:
time_takes = time.time() - start_time
time_remains = time_takes * (len(qs)-len(qas)) / len(qas)
end_time = start_time + time_remains
messages = []
for qa in qas:
messages.append({"role":"user", "content":qas["q"]})
messages.append({"role":"assistant", "content":qas["a"]})
num_tokens_processed = num_tokens_from_messages(messages)
num_tokens_total = num_tokens_processed * len(qs) / len(qas)
dollars_tokens_processed = 0.002 * int(num_tokens_processed/1000)
dollars_tokens_total = 0.002 * int(num_tokens_total/1000)
chatbot_ret = [(f"Your entered Access Key is correct.", f"The information of progress is displayed below.")]
chatbot_ret += [(f"The number of processed / total queries:", f"{len(qas)} / {len(qs)}")]
chatbot_ret += [(f"The hours already takes / est. remains:", f"{time_takes/3600:.2f} / {time_remains/3600:.2f}")]
chatbot_ret += [(f"The time starts / est. ends:", f"{timestamp2string(start_time)} / {timestamp2string(end_time)}")]
chatbot_ret += [(f"The number of processed / est. total tokens:", f"{num_tokens_processed} / {num_tokens_total}")]
chatbot_ret += [(f"The dollars of processed / est. total tokens:", f"{dollars_tokens_processed} / {dollars_tokens_total}")]
return chatbot_ret
with gr.Blocks() as demo:
with gr.Column(variant="panel"):
chatbot = gr.Chatbot()
txt = gr.Textbox(show_label=False, placeholder="Enter your Access Key to access this private space").style(container=False)
with gr.Row():
button_showcase = gr.Button("Show Recent Query-Responses")
button_download = gr.Button("Download All Query-Responses")
button_display = gr.Button("Display Progress Infomation")
downloadfile = gr.File(None, interactive=False, show_label=False, visible=False)
button_showcase.click(fn=showcase, inputs=[txt], outputs=[chatbot])
button_download.click(fn=download, inputs=[txt], outputs=[chatbot, downloadfile])
button_display.click(fn=display, inputs=[txt], outputs=[chatbot])
demo.launch() |