https://github.com/hiejulia/full-stack-deep-learning-bootcamp-project
Full stack Deep Learning Bootcamp Final Project by Sillicon Valley Top Practitioners
https://github.com/hiejulia/full-stack-deep-learning-bootcamp-project
Last synced: about 1 month ago
JSON representation
Full stack Deep Learning Bootcamp Final Project by Sillicon Valley Top Practitioners
- Host: GitHub
- URL: https://github.com/hiejulia/full-stack-deep-learning-bootcamp-project
- Owner: hiejulia
- Created: 2020-10-19T17:59:11.000Z (almost 5 years ago)
- Default Branch: main
- Last Pushed: 2020-10-27T22:56:06.000Z (almost 5 years ago)
- Last Synced: 2025-02-08T21:46:21.879Z (8 months ago)
- Language: Jupyter Notebook
- Homepage:
- Size: 3.39 MB
- Stars: 1
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# full-stack-deep-learning-bootcamp-project
Project developed during lab sessions of the [Full Stack Deep Learning Bootcamp](https://fullstackdeeplearning.com).
- Handwriting recognition system from scratch, and deploy it as a web service.
- Uses Keras, but designed to be modular, hackable, and scalable
- Provides code for training models in parallel and store evaluation in Weights & Biases
- We will set up continuous integration system for our codebase, which will check functionality of code and evaluate the model about to be deployed.
- We will package up the prediction system as a REST API, deployable as a Docker container.
- We will deploy the prediction system as a serverless function to Amazon Lambda.
- Lastly, we will set up monitoring that alerts us when the incoming data distribution changes.
## Dataset used in this project
- https://fki.tic.heia-fr.ch/databases/iam-handwriting-database
## ML model train documentation
- Started with base model MLP/CNN
- Run LeNet model on subsampling dataset
- Train CNN for reading lines
- Train LSTM on Emnist
- TODO : Train Encoder - Decoder architecture with attention
- Download IAM handwriting dataset
- Use weight and bias for metrics
- Run multiple experiments in parallel with hyper param sweeps
- Train data augmentation
- Ensemble 2 models
- Test
- Deploy model
## My weight & bias metrics
- https://wandb.ai/fsdl/fsdl-text-recognizer-nov2019?workspace=user-hiennguyen
## Schedule for the November 2019 Bootcamp
- First session (90 min)
- [Setup]Set up with jupyterhub.
- Gather handwriting data (10 min).
- [Lab 1](lab1.md) (20 min): Introduce EMNIST. Training code details. Train & evaluate character prediction baselines.
- [Lab 2](lab2.md) (30 min): Introduce EMNIST Lines. Overview of CTC loss and model architecture. Train our model on EMNIST Lines.
- Second session (60 min)
- [Lab 3](lab3.md) (40 min): Weights & Biases + parallel experiments
- [Lab 4](lab4.md) (20 min): IAM Lines and experimentation time (hyperparameter sweeps, leave running overnight).
- Third session (90 min)
- Review results from the class on W&B
- [Lab 5](lab5.md) (45 min) Train & evaluate line detection model.
- [Lab 6](lab6.md) (45 min) Label handwriting data generated by the class, download and version results.
- Fourth session (75 min)
- [Lab 7](lab7.md) (15 min) Add continuous integration that runs linting and tests on our codebase.
- [Lab 8](lab8.md) (60 min) Deploy the trained model to the web using AWS Lambda.