Skip to content

A lightweight workflow for a deep learning project, covering development to production deployment. The example model used here is a CNN MNIST classifier. The model is hosted on Hugging Face space, just a guide for a quick and hacky prototype workflow.

Notifications You must be signed in to change notification settings

arun477/nano_deep_learning_dev_pipeline

Repository files navigation

title emoji colorFrom colorTo sdk pinned
Mnist
🐢
yellow
purple
docker
false

Logo MNIST CLASSIFIER

MNIST classifier from scratch

  • Model: CNN

  • Accuracy: 97%

  • Training Notebook: mnist_classifier.ipynb

  • Cleaned Python Inference Version: mnist_classifier.py (This file is auto-generated from mnist_classifier.ipynb. Please do not edit it.)

  • Model: classifier.pth

  • Try Online: https://carlfeynman-mnist.hf.space/

  • If you want to host this on Hugging Face as a space, please refer to this documentation.

  • To Run FastApi Server Locally: uvicorn server:app --reload

  • Note: When you try it out on the website, accuracy may drop due to distribution changes from training data to canvas image input. It has not been adjusted or fine-tuned for this specific purpose; it's intended just to demonstrate the full flow.

site_screenshot

About

A lightweight workflow for a deep learning project, covering development to production deployment. The example model used here is a CNN MNIST classifier. The model is hosted on Hugging Face space, just a guide for a quick and hacky prototype workflow.

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published