Modelly is a no-code solution for building, training and inferring from your built models. It allows everyone to create their own model and training them by just selecting the dataset and the name of the required model. For purely non-technical professionals, it allows you to build and train models in such a way that no hyperparameters are required and you can easily use it in a hassle-free way. For technical professionals, it allows you to focus on the core architecture and hyperparameters without any coding requirements to build an optimal architecture.
- No prior coding experience/knowledge required.
- No data cleaning or pre-processing required.
- Robust solution for training as well as inference phase.
- Trained model and predicted outputs can be downloaded for future use.
- Easy and intuitive UI.
- Training characteristics are graphically represented.
- XBNet
- Deep Neural Network
- XGBoost
- Random Forest
- Light GBM
- Decision Tree
Clone the project -
$ git clone https://github.com/tusharsarkar3/Modelly.git
Start a new virtual environment -
$ virtual venv
Install all the reqirements -
$ pip install -r requirements.txt
Run the following commands -
$ flask run
- Adding more models.
- Adding pre-trained models.