Side by side
Auto-PyTorch vs BigML
Both stand on automl software. Which one you can start free with, which one lets you read the code, and what each costs once you pay.
Free plan LinuxSelf-hosted
Auto-PyTorch offers automated search, preprocessing, and ensembling controls for several documented machine-learning tasks. Check the Linux and SWIG requirements first, and account for extra forecasting dependencies or shared storage when using parallel workers.
Visit Auto-PyTorch Free plan APIMacSelf-hosted+1
BigML combines a broad set of machine-learning methods with model explanations, workflow automation, and deployment options. Its free plan has task and user limits, while listed paid plans begin at $1,000 per month.
Visit BigMLThe facts
| Auto-PyTorch | BigML | |
|---|---|---|
| Free plan | Yes | Yes |
| Open source | Not published | No |
| Runs on | Linux, Self-hosted | API, Mac, Self-hosted, Web |
| Lowest paid plan | Free or on request | BigML Lite (annual): $10,000 a year |
| Plans published | 1 | 6 |
| Maker | Not published | Not published |
| Website | automl.github.io | bigml.com |
| On automl software | #8 of 30 | #4 of 30 |
Auto-PyTorch: what it does well
- Automates model selection and hyperparameter tuning.
- Supports classification, regression, and forecasting.
- Allows memory and search-time limits.
- Provides a Docker image.
- Uses the 3-clause BSD license.
What to know first
- Requires Linux and specified compiler and SWIG versions.
- SWIG 4.0 or later is unsupported.
- Forecasting requires additional dependencies.
- Parallel workers need shared filesystem access.
BigML: what it does well
- Supports forecasting, anomaly detection, and topic modeling
- Models include prediction explanations and field importances
- Private deployments can run on premises
- Free plan is listed as free
What to know first
- Free plan limits datasets to 16 MB per task
- Free plan allows two parallel tasks and one user
- Paid plans start at $1,000 per month
Alternatives to Auto-PyTorch Alternatives to BigML The whole list