/hotdog_or_not_aivm

Primary LanguageJupyter Notebook

HotDog SlackBot

Based on the Hot Dog or Not app from the TV Show Silicon Valley.

We use

  • Local ai-vm
  • Run the model on a large hotdog or not / food dataset
  • Then a local NGROK

AIVM usage

  • Lenet5 Model
  • Uploaded custom model trained on the dataset
  • Then running the predictions from model
  • Accuracy: 60ish%

Pre-requsites

  • You need a kaggle.json setup in order to interact with the files
  • NGROK
  • Slack Bot API KEY
  • Enable these permissions on the bot in EVENT SUBSCRIPTIONS
  • Also disabled SOCKET MODE for Slack for the local development process.

SlackBotPermissions

Usage

Files

  • bot.py to run the code
  • fine-tuning-lenet5.ipynb if you want to fine tune the model further.

Installation

  1. Clone this repo
  2. Create a virtual environment to run Python python3 -m venv .venv
  3. Linux / Mac: source .venv/bin/activate and Windows: .\venv\Scripts\activate
  4. pip install .

Running it You will have to use 3 seprate terminal windows to make it work.

  1. Run AIVM by running aivm-devnet and wait for all of the ProxyServer to start listening
  2. Run python3 bot.py in your venv to run the bot
  3. Then run ngrok http 8080 to run NGROK on a local port
  4. In the NGROK terminal, you should have a forwarding URL like: https://XXXX-XXX-XXX-XXX-XXX.ngrok-free.app. Copy and paste this into the Request URL and ensure it pings your NGROK server + passes the check
  5. Interact with your SlackBot by @ the bot with a file of a hotdog or not and you should receive a 200 code and a reply.

Feel free to reach out on Github Discussions if any issues.