https://github.com/the-full-stack/fsdl-text-recognizer-2022

Source of the FSDL 2022 labs, which are at https://github.com/full-stack-deep-learning/fsdl-text-recognizer-2022-labs
https://github.com/the-full-stack/fsdl-text-recognizer-2022

deep-learning deep-neural-networks machine-learning mlops mlops-community mlops-workflow

Last synced: about 2 months ago
JSON representation

Source of the FSDL 2022 labs, which are at https://github.com/full-stack-deep-learning/fsdl-text-recognizer-2022-labs

• Host: GitHub
• URL: https://github.com/the-full-stack/fsdl-text-recognizer-2022
• Owner: the-full-stack
• License: mit
• Created: 2022-07-02T19:14:44.000Z (over 2 years ago)
• Default Branch: main
• Last Pushed: 2024-03-20T18:29:38.000Z (9 months ago)
• Last Synced: 2024-04-17T06:19:52.096Z (8 months ago)
• Topics: deep-learning, deep-neural-networks, machine-learning, mlops, mlops-community, mlops-workflow
• Language: Jupyter Notebook
• Homepage: https://fullstackdeeplearning.com/course
• Size: 4.03 MB
• Stars: 78
• Watchers: 6
• Forks: 28
• Open Issues: 17
• Metadata Files:
  • Readme: readme.md
  • Contributing: CONTRIBUTING.md
  • License: LICENSE.txt

Awesome Lists containing this project

README

        
# 🥞 FSDL Text Recognizer 2022

Implements a text recognizer using a hypermodern 2022 DL stack:

- Code quality assurance with pre-commit hooks, GitHub Actions, and pytest

- Real data augmented with data synthesis

- Image encoding via a CNN architecture

- Decoding to text via a Transformer architecture

- Model implementation in PyTorch

- Model training and basic tracking and metrics via PyTorch Lightning and torchmetrics

- Experiment tracking, hyperparameter tuning, and model versioning with W&B

- Model packaging in TorchScript

- Predictor backend containerization via Docker and deployment as a microservice on AWS Lambda

- Basic load testing with Locust

- Pure Python frontend web application in Gradio

- Model monitoring with Gantry