[go: up one dir, main page]

Skip to content

catalyst-team/reaction

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

42 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Reaction logo

Convenient DL serving

Build Status CodeFactor Pipi version Docs PyPI Status

Twitter Telegram Slack Github contributors

Project manifest. Part of Catalyst Ecosystem:

  • Alchemy - Experiments logging & visualization
  • Catalyst - Accelerated Deep Learning Research and Development
  • Reaction - Convenient Deep Learning models serving

Installation

Common installation:

pip install -U reaction

Getting started

consumer.py:

import asyncio
from typing import List, Any
from reaction.rpc import RabbitRPC


class rpc(RabbitRPC):
    URL = "amqp://user:password@host"


@rpc()
def sync_square(*values) -> List[float]:
    return [v ** 2 for v in values]


@rpc()
async def async_square(*values) -> List[float]:
    await asyncio.sleep(1)
    return [v ** 2 for v in values]


if __name__ == "__main__":
    loop = asyncio.get_event_loop()
    loop.create_task(sync_square.consume())
    loop.create_task(async_square.consume())
    loop.run_forever()

client.py:

import asyncio
from consumer import sync_square, async_square

if __name__ == "__main__":
    loop = asyncio.get_event_loop()
    x = loop.run_until_complete(sync_square.call(2, 3))
    y = loop.run_until_complete(async_square.call(4, 5, 6))
    print(x)  # 4, 9
    print(y)  # 16, 25, 36
    loop.close()

Example

  • Register telegram bot, achieve token
  • cd example && TG_TOKEN="telegram bot token goes here" docker-compose up --force-recreate --build
  • RabbitMQ web ui: http://127.0.0.1:15672/#/
    • user: admin
    • password: j8XfG9ZDT5ZZrWTzw62q
  • Docs (you can submit requests from web ui): http://127.0.0.1:8000/docs#/
  • Redoc: http://127.0.0.1:8000/redoc
  • Telegram bot is ready to classify ants & bees, you have to send files "as a photo" or "as a file"

Telegram bot quick howto

Install async telegram client first:

$ pip install aiotg

Then create your bot:

tgbot.py

from consumer import async_square
from aiotg import Bot, Chat

bot = Bot(api_token="telegram bot token goes here")


@bot.command("/start")
async def start(chat: Chat, match):
    return chat.reply("Send me /square command with one float argument")


@bot.command(r"/square (.+)")
async def square_command(chat: Chat, match):
    val = match.group(1)
    try:
        val = float(val)
        square = await async_square.call(val)
        resp = f"Square for {val} is {square}"
    except:
        resp = "Invalid number"
    return chat.reply(resp)


bot.run()