This repository contains machine learning solutions automatically generated by Plexe, an AI-powered ML solutions generator. All experiments were conducted with a 1-hour time limit on CPU-only systems.
Results for 20 datasets from the OpenML automl benchmark suite, competing against leading AutoML frameworks like AutoGluon, H2O AutoML, and FLAML:
- Generated solution code in Python
- Performance metrics (AUC scores)
- Datasets spanning multiple domains: finance, healthcare, computer vision, etc.
Solutions for 11 diverse Kaggle competitions, achieving competitive rankings:
- Generated solution code in Python
- Competition performance metrics
- Domains include: house price prediction, blood glucose prediction, NLP, and more
For detailed analysis, benchmarking results, and comparison with other AutoML frameworks, check out our comprehensive blog post.
We're actively working on making Plexe even better:
- Expanding to more complex ML tasks and architectures
- Improving solution optimization strategies
- Supporting more deployment scenarios
- Adding support for deep learning tasks
Plexe will be available for beta testing soon! Sign up here to be among the first to try it out and help shape its development.
Alternatively, you can join our Discord server where you can request us to create a model for you, and within a few hours we'll send you the model code along with a detailed report of the solutions considered and performance metrics. Currently, this will be limited to smaller and structured datasets only but we will soon be adding support for more complex architectures and larger datasets soon!
This project is licensed under the MIT License - see the LICENSE file for details.