Skip to content

Infrastructure to run a poe.com bot based on Fireworks.ai hosted models

License

Notifications You must be signed in to change notification settings

fw-ai/fireworks_poe_bot

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Fireworks Poe Bot

This codebase is the underlying implementation of the Mistral-7B on Poe. This implementation calls into the accounts/fireworks/models/mistral-7b-instruct-4k model from fireworks.ai. However, this bot implementation is fully general and can be used to create bots that call into other Fireworks models.

Server Arguments

The server has several important arguments:

  • --model - This flag specifies the model to call into on Fireworks. For example: accounts/fireworks/models/mistral-7b-instruct-4k
  • FIREWORKS_API_KEY - This environment variable specifies the API key to use to call into Fireworks. API keys can be retrieved from the Fireworks console.

There are several optional arguments that can usually be left as default:

  • --host specifies the hostname to bind to. You usually don't need to set this
  • --port specifies the port for the bot to listen on. By default this is port 80.
  • FIREWORKS_API_BASE environment variable specifies the base URL to call into Fireworks. This is https://api.fireworks.ai by default. You usually don't need to modify this.

Running the bot

The bot can be run locally by installing the package and running the module:

$ pip install -e .
$ FIREWORKS_API_KEY=<your API key> python -m fireworks_poe_bot --model accounts/fireworks/models/mistral-7b-ins
truct-4k
INFO:     Started server process [50060]
INFO:     Waiting for application startup.
INFO:     Application startup complete.
INFO:     Uvicorn running on http://0.0.0.0:80 (Press CTRL+C to quit)

The server will then be listening on port 80. You can then communicate with the bot using the Poe simulator for testing:

# BOT_SERVER=localhost:80 python -m simulator_poe
Welcome to the Poe server simulator!
!q -- quit Poe server simulator
!c -- clear the context
!d -- toggle debug mode
Poe server > test

Bot server > Hello! I'm here
 to listen and assist you
 with any questions or concerns
 you may have. I
 will always strive to
 provide helpful and accurate information
 while maintaining a respectful
 and positive tone. If
 there is anything specific you
 need help with, please
 let me know.

Testing

Test cases can be found in test/test_poebot.py. That file can be invoked directly or with pytest. The test file mocks out the Fireworks service -- it is intended for quick iteration of development of the poebot logic itself.

Deployment

As the bot listens on port 80, it can be deployed in many contexts, including through Modal and in other infrastructure like k8s.

Deploying to Modal

From the root of this repository, run:

$ MODEL=accounts/fireworks/models/mistral-7b-instruct-4k FIREWORKS_API_KEY=<your API key> modal deploy fireworks_poe_bot/main_modal.py
✓ Created objects.
├── 🔨 Created mount /root/fireworks/public-py/fireworks
├── 🔨 Created mount /root/fireworks/py/fireworks
├── 🔨 Created mount /root/fireworks_poe_bot/fireworks_poe_bot
└── 🔨 Created fastapi_app => https://XXXX--fw-poe-bot-fastapi-app.modal.run
✓ App deployed! 🎉

View Deployment: https://modal.com/apps/XXXX/fw-poe-bot

You should then be able to curl the https URL returned by Modal:

$ curl https://XXXX--fw-poe-bot-fastapi-app.modal.run
<html><body><h1>FastAPI Poe bot server</h1><p>Congratulations! Your server is running. To connect it to Poe, create a bot at <a href="https://poe.com/create_bot?server=1">https://poe.com/create_bot?server=1</a>.</p></body></html>

After this, follow the instructions on the Poe documentation to create a Poe bot and point it to your running deployment.

Contributing

Contributions and bug reports are greatly appreciated!

About

Infrastructure to run a poe.com bot based on Fireworks.ai hosted models

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages