Training

Note

Training is still experimental and subject to API changes.

Before you can use the solver, you must train the SimAI solution on your prediction data. You first upload your prediction data into a global pool of training data instances and then assign this data to different Project instances, which you configure for training your model.

Train on prediction data

1. Create a SimAIClient instance:

   import ansys.simai.core
   
   simai = ansys.simai.core.SimAIClient()
   

   You are prompted for your credentials.

   If desired, you can create an instance using a configuration file. For more information, see Client configuration.

2. Upload your prediction data by creating a TrainingData instance and then loading your files into it:

   td = simai.training_data.create("my-first-data")
   td.upload_folder("/path/to/folder/where/files/are/stored")
   

3. Create a project:

   project = simai.projects.create("my-first-project")
   

4. Assign the created training data to your project:

   td.add_to_project(project)
   

Once you have training data in your project, you can use the web app to train a model.

Learn more

For more information on the actions available to you, see TrainingData, TrainingDataParts, and Projects.