Training new models

Use training jobs to retrain models on a regular cadence. Jobs can be created from Python notebooks or Git.

Each training job gets its own source code and environment. Training jobs save the models they train to the model registry.

Visit the Training Jobs page within your deployment to view previous executions of jobs, how long each job took, and any logs output created by the job.

Choose how you'd like to get started with training jobs:

• Python notebook - Create a new training job from Hex, Colab, Jupyter, or any other Python notebook
• Git - Create a new training job using your favorite text editor and git push

For more information on the servers sizes available, see the training jobs sizes tables.