Train AI Model
Train any AI or ML model on your own data directly inside a flow.
The Train AI Model node trains a machine learning model on your data. Connect your prepared dataset, select a model type, configure it, and run. The trained model and performance metrics are saved as outputs you can use in subsequent inference flows.
Training runs can take time depending on dataset size and model complexity. You can monitor progress via the node logs during execution.
Configuration
The node uses a cascading selection:
- Category — The AI task type (e.g. classification, regression, object detection, NLP)
- Subcategory — A specific task variant within the category
- Model — The exact model to train
- Training Configuration — Model-specific settings loaded dynamically (hyperparameters, epochs, column mappings, etc.)
Inputs & Outputs
| Name | Description | |
|---|---|---|
| Input | training_data | Your training dataset — format depends on the selected model (required) |
| Output | trained_model | The saved model file, ready for use in a Use AI Model node |
| Output | metrics | Training performance metrics (accuracy, loss, etc.) |
Tips
- Connect a Preview Output to the
metricsoutput to inspect training results directly in the editor - The trained model output can be passed directly to a Use AI Model node in the same flow, or saved and loaded later via Select Artifact
- Larger datasets and more complex models take longer — check the node logs to monitor progress
- Use Apply Transformation or Custom Code nodes upstream to prepare your data before training