The bot is based on ChatGPT and Dell-E neural networks.
Consider the code and steps to run the bot.
This is the code required for GPT and Dell to work
openai.api_key = config.API - токин с сайта OpenAI хранится в config.py
model_engine = "text-davinci-003" - модель на которой работает GPT
This part of the code is copied from the OpenAI documentation
async def chat_gpt(message: types.Message):
prompt = str(message.text)
completion = openai.Completion.create(
engine=model_engine,
prompt=prompt,
max_tokens=2048,
temperature=0.5,
top_p=1,
frequency_penalty=0,
presence_penalty=0
)
await bot.send_message(message.chat.id, completion.choices[0].text)
And this piece is responsible for the work of Dell E
await bot.send_message(message.chat.id, "Ожидайте изображение...")
prompt = str(message.text)
response = openai.Image.create(
prompt=prompt,
n=1,
size="512x512",
)
await bot.send_message(message.chat.id, response["data"][0]["url"])
But I want to emphasize that in the web version of telegram, the image comes as a link and nothing else.
The rest of the code is just the character counting algorithms I came up with, the functions for determining whether the user is subscribed to a group in Tg and the database storing all the data about the user, I can’t say that all these algorithms have been brought to mind, but they are not bad, although they require improvements.
- Get a token from BotFather
- Get a token in OpenAI
- Paste in config.py file
- Go to IDE/cmd/notepad++
- And run the program in the standard way for your environment
You should not have any more problems, but if you have, create a discussion or ask a question on stackoverflow
The bot on the telegram link is not working yet.