10+ ML FRameworks
MindsDB allows you to select your own model to conduct accuracy-improving experiments. Choose to configure the default settings of automatically generated models, or import models from a growing list of ML libraries, or any custom model written in Python.
SQL is the go-to tool for data transformation, and database columns are the excellent match for ML features. MindsDB automates the rest.
Make your MLOps pipelines simple. Once model training is complete, they are ready to use inside your database. No extra infrastructure is necessary.
JOIN models with your database tables to add predictions and explainability metadata, such as anomaly flag, confidence bounds, etc.
Either connect to third-party ML Frameworks such as Ludwig and Huggingface – or import your own custom Python models into MindsDB.
Using the SQL Wire Protocol, train models and analyze future events in your business dashboards.
Use DBT to orchestrate data transformation tasks – or using our soon-to-be-available feature store.
Get it learning. Predicting. Analyzing. Quick results. Minimal setup.
We support a wide variety of machine learning frameworks in MindsDB including Pytorch, Tensorfow and Scikit. Learn more in our documentation
For algorithms, we use a mix of Neural Networks (fully-connected, transformers, convolutional and recurrent depending on the problem type), advanced Gradient Boosters and over a dozen “classical” models.
No, we don't save your data. In MindsDB Cloud, data defined in a SQL query is ingested for processing. Data is transferred to our servers for training but it isn't stored.
We offer a paid, fully managed, MindsDB Cloud and a free, self-hosted, Open-source option. MindsDB provides additional technical support from ML Engineers to build machine learning models, manage updates, build out authentication, as well as guaranteeing things like response times for customer enquiries, for both Cloud and Open-source. Learn more on our pricing page