Traingenerator 🧙 A web app to generate template code for machine learning ✨

Johannes Rieke

Last update: Jan 7, 2023

Related tags

Machine Learning template website machine-learning deep-learning tensorflow scikit-learn sklearn pytorch webapp machine-learning-tutorials pytorch-ignite pytorch-lightning streamlit

Overview

Traingenerator

🧙 A web app to generate template code for machine learning ✨

🎉 Traingenerator is now live! 🎉

Try it out:
https://traingenerator.jrieke.com

Generate custom template code for PyTorch & sklearn, using a simple web UI built with streamlit. Traingenerator offers multiple options for preprocessing, model setup, training, and visualization (using Tensorboard or comet.ml). It exports to .py, Jupyter Notebook, or Google Colab. The perfect tool to jumpstart your next machine learning project!

For updates, follow me on Twitter, and if you like this project, please consider sponsoring ☺

Adding new templates

You can add your own template in 4 easy steps (see below), without changing any code in the app itself. Your new template will be automatically discovered by Traingenerator and shown in the sidebar. That's it! 🎈

Want to share your magic? 🧙 PRs are welcome! Please have a look at CONTRIBUTING.md and write on Gitter.

Some ideas for new templates: Keras/Tensorflow, Pytorch Lightning, object detection, segmentation, text classification, ...

Create a folder under ./templates. The folder name should be the task that your template solves (e.g. Image classification). Optionally, you can add a framework name (e.g. Image classification_PyTorch). Both names are automatically shown in the first two dropdowns in the sidebar (see image). ✨ Tip: Copy the example template to get started more quickly.
Add a file sidebar.py to the folder (see example). It needs to contain a method show(), which displays all template-specific streamlit components in the sidebar (i.e. everything below Task) and returns a dictionary of user inputs.
Add a file code-template.py.jinja to the folder (see example). This Jinja2 template is used to generate the code. You can write normal Python code in it and modify it (through Jinja) based on the user inputs in the sidebar (e.g. insert a parameter value from the sidebar or show different code parts based on the user's selection).
Optional: Add a file test-inputs.yml to the folder (see example). This simple YAML file should define a few possible user inputs that can be used for testing. If you run pytest (see below), it will automatically pick up this file, render the code template with its values, and check that the generated code runs without errors. This file is optional – but it's required if you want to contribute your template to this repo.

Installation

Note: You only need to install Traingenerator if you want to contribute or run it locally. If you just want to use it, go here.

git clone https://github.com/jrieke/traingenerator.git
cd traingenerator
pip install -r requirements.txt

Optional: For the "Open in Colab" button to work you need to set up a Github repo where the notebook files can be stored (Colab can only open public files if they are on Github). After setting up the repo, create a file .env with content:

GITHUB_TOKEN=<your-github-access-token>
REPO_NAME=<user/notebooks-repo>

If you don't set this up, the app will still work but the "Open in Colab" button will only show an error message.

Running locally

streamlit run app/main.py

Make sure to run always from the traingenerator dir (not from the app dir), otherwise the app will not be able to find the templates.

Deploying to Heroku

First, install heroku and login. To create a new deployment, run inside traingenerator:

heroku create
git push heroku main
heroku open

To update the deployed app, commit your changes and run:

git push heroku main

Optional: If you set up a Github repo to enable the "Open in Colab" button (see above), you also need to run:

heroku config:set GITHUB_TOKEN=
   
    
heroku config:set REPO_NAME=

Testing

First, install pytest and required plugins via:

pip install -r requirements-dev.txt

To run all tests:

pytest ./tests

Note that this only tests the code templates (i.e. it renders them with different input values and makes sure that the code executes without error). The streamlit app itself is not tested at the moment.

You can also test an individual template by passing the name of the template dir to --template, e.g.:

pytest ./tests --template "Image classification_scikit-learn"

The mage image used in Traingenerator is from Twitter's Twemoji library and released under Creative Commons Attribution 4.0 International Public License.

Comments

New Feature: Semantic Segmentation

I started to work on semantic segmentation with TF/Keras. i will mainly use segmentation models library, if you have any suggestions or if you want to help feel free to contact me!
new template

opened by erentknn 3
Arbitrary class training in Pytorch #5
Summary

This PR adds support for declaring number of clasees for Image Classification Pytorch models.

Details

Adds a parameter in the sidebar below the "Use pre-trained model" checkbox

Default value is 1000 (what torchvision uses)

If pre-trained is True, imagenet weights are loaded and last fc layer is redefined

Checklist

[X] all tests are passing (see README.md on how to run tests)

[ ] if you created a new template: it contains a file test-inputs.yml, which specifies a few input values to test the code template (the test is then automatically run by pytest)

[X] you formatted all code with black

[X] you checked all new functionality live, i.e. in the running web app

[X] any generated code is formatted nicely, both in .py and in .ipynb ("nicely" = comparable to the existing templates)

[X] you added comments in your code that explain what it does

[X] the PR explains in detail what's new
opened by murthy95 1
Add Aim integration to PyTorch image classification template
Summary

Added Aim to PyTorch image classification template as an experiment logger option.

Details

What was added:

Initialization of aim.Session

Logging parameters with Aim

Logging metrics with Aim

Aim as a logger option in the sidebar

Experiment name input in the sidebar

Markdown of how to run UI and link to the full documentation

Checklist

[x] all tests are passing (see README.md on how to run tests)

[x] you formatted all code with black

[x] you checked all new functionality live, i.e. in the running web app

[x] any generated code is formatted nicely, both in .py and in .ipynb ("nicely" = comparable to the existing templates)

[x] you added comments in your code that explain what it does

[x] the PR explains in detail what's new
opened by gorarakelyan 0
[WIP] Refactor to make templates more independent from app
Summary (2-3 sentences)

This PR tries to make templates (i.e. what is rendered as code) completely independent from the core app (i.e. the website, buttons, logic behind the rendering). This will enable contributors to more easily create templates for new tasks/frameworks, without having to understand or modify the app itself.

Details

Each template is now one directory under ./templates with 3 files:

code-template.py.jinja: The template for the code that will be rendered (as before, as a jinja template)

sidebar.py: Needs to contain a method show(), which renders all template-specific streamlit components into the sidebar

test-inputs.yml: A yml file that defines some input values for the code template (used by pytest to check that the template runs w/o errors)

Existing templates in ./templates are automatically detected by the app. Contributing a new template does not require a single change in ./app nor in any other template. Also, pytest will automatically pick up the new template and its test-inputs.yml file and make sure that the template renders and runs without errors (i.e. the user doesn't have to add any specific code for testing).

This PR also changes the ordering of components in the sidebar slightly and fixes a few minor bugs in the code templates that I found during refactoring.
opened by jrieke 0
Update requirements.txt
Issues in installing

Summary

Details

Checklist

[ ] all tests are passing (see README.md on how to run tests)

[ ] if you created a new template: it contains a file test-inputs.yml, which specifies a few input values to test the code template (the test is then automatically run by pytest)

[ ] you formatted all code with black

[ ] you checked all new functionality live, i.e. in the running web app

[ ] any generated code is formatted nicely, both in .py and in .ipynb ("nicely" = comparable to the existing templates)

[ ] you added comments in your code that explain what it does

[x] the PR explains in detail what's new
opened by tanujdhiman 2
Add Mlflow tracking
Summary

As mentioned in #17, adding training run and experiment tracking capabilities using Mlflow for the following templates:

templates/Image classification_PyTorch

templates/Image classification_scikit-learn

Details

This PR implements basic metric/model info logging features. As a second logical next step, model artifact logging (checkpoints and final model) will be added.

Checklist

[x] all tests are passing (see README.md on how to run tests)

[ ] if you created a new template: it contains a file test-inputs.yml, which specifies a few input values to test the code template (the test is then automatically run by pytest)

[x] you formatted all code with black

[x] you checked all new functionality live, i.e. in the running web app

[x] any generated code is formatted nicely, both in .py and in .ipynb ("nicely" = comparable to the existing templates)

[ ] you added comments in your code that explain what it does

[x] the PR explains in detail what's new
opened by andodet 0
Add mlflow tracking

First of all thanks for the project, it's an interesting way to take a stab at reducing the amount of boilerplate needed even for fairly simple models. Secondly, it would be interesting to implement experiment/run tracking using MLflow.

Have a working example on the Image classification_PyTorch/ template, happy to submit a PR if you consider this of any interest.
existing template

opened by andodet 1
Style Transfer Template
Summary

I just added the style transfer template

Checklist

[ ] all tests are passing (see README.md on how to run tests)

[ ] if you created a new template: it contains a file test-inputs.yml, which specifies a few input values to test the code template (the test is then automatically run by pytest)

[ ] you formatted all code with black

[ ] you checked all new functionality live, i.e. in the running web app

[ ] any generated code is formatted nicely, both in .py and in .ipynb ("nicely" = comparable to the existing templates)

[ ] you added comments in your code that explain what it does

[ ] the PR explains in detail what's new
opened by The-ML-Hero 0
New Template: Semantic Segmentation; Generative Adversarial Networks; Image Caption; Style Transfer;
@jrieke

Hi, everybody, I'm looking forward to contributing this project can I add

Semantic Segmentation

Generative Adversarial Networks

Image Caption

Style Transfer

as I think this would greatly help beginners who are starting out their deep learning journey and want to do interesting things like these.

PS: can I also work on object detection using the YOLO-v5 achritexture.

Thanks a lot please reply as soon as possible
new template
opened by The-ML-Hero 2
Text classification template with scikit-learn
Summary

This PR consists of a text classification template using scikit-learn as stated in #4.

Details

Scikit-learn

NLTK for stemming and lemmatization

Example text files provided in the data/ folder (in spanish)

test.py file added in order to check the functionality of the template code itself (needs to be removed before merging)

Checklist

[x] all tests are passing (see README.md on how to run tests)

[x] if you created a new template: it contains a file test-inputs.yml, which specifies a few input values to test the code template (the test is then automatically run by pytest)

[x] you formatted all code with black

[x] you checked all new functionality live, i.e. in the running web app

[x] any generated code is formatted nicely, both in .py and in .ipynb ("nicely" = comparable to the existing templates)

[x] you added comments in your code that explain what it does

[x] the PR explains in detail what's new
opened by themrcesi 3

Traingenerator 🧙 A web app to generate template code for machine learning ✨

Related tags

Overview

Traingenerator

🎉 Traingenerator is now live! 🎉 Try it out: https://traingenerator.jrieke.com

Adding new templates

Installation

Running locally

Deploying to Heroku

Testing

Comments

Summary

Details

Checklist

Summary

Details

Checklist

Summary

Details

Checklist

Summary

Details

Checklist

Summary

Checklist

Summary

Details

Checklist

Owner

Johannes Rieke

MachineLearningStocks is designed to be an intuitive and highly extensible template project applying machine learning to making stock predictions.

Machine learning template for projects based on sklearn library.

Turns your machine learning code into microservices with web API, interactive GUI, and more.

A Python Automated Machine Learning tool that optimizes machine learning pipelines using genetic programming.

Python Extreme Learning Machine (ELM) is a machine learning technique used for classification/regression tasks.

Vowpal Wabbit is a machine learning system which pushes the frontier of machine learning with techniques

CD) in machine learning projectsImplementing continuous integration & delivery (CI/CD) in machine learning projects

A machine learning web application for binary classification using streamlit

Interactive Web App with Streamlit and Scikit-learn that applies different Classification algorithms to popular datasets

Katana project is a template for ASAP 🚀 ML application deployment

Petastorm library enables single machine or distributed training and evaluation of deep learning models from datasets in Apache Parquet format. It supports ML frameworks such as Tensorflow, Pytorch, and PySpark and can be used from pure Python code.

Examples and code for the Practical Machine Learning workshop series

100 Days of Machine and Deep Learning Code

This is the code repository for Interpretable Machine Learning with Python, published by Packt.

PLUR is a collection of source code datasets suitable for graph-based machine learning.

Microsoft contributing libraries, tools, recipes, sample codes and workshop contents for machine learning & deep learning.

A data preprocessing package for time series data. Design for machine learning and deep learning.

A mindmap summarising Machine Learning concepts, from Data Analysis to Deep Learning.

A comprehensive repository containing 30+ notebooks on learning machine learning!

🎉 Traingenerator is now live! 🎉

Try it out:
https://traingenerator.jrieke.com