model_configuration

This context is used for editing model configurations in Terarium’s new JSON style. On setup it expects a model configuration id to be provided; unlike other contexts the key is always id and the value is the model configuration id. For example:

{
  "id": "27ec5daa-d137-43d2-bc3b-8109ba91a7b1"
}

You can also provide a dataset_id if you wish to parameterize a model configuration based on a dataset. For example:

{"id": "161bee2c-f5eb-4811-a407-0d789ebccbf5",
 "dataset_id": "e3eecbf9-bc90-4591-9d90-0b9ab20472ad"}

Note: after setup, the model configuration is accessible via the variable name model_config. If a dataset is provided it’s loaded as a Pandas DataFrame dataset.

This context’s LLM agent allows the user to ask various questions of the configuration values and to edit the values. It is aware of the schema constraints and oeprates within them.

This context has 1 custom message types:

  1. save_model_config_request: this does not require arguments; it simply executes a PUT on the model configuration to update it in place based on the operations performed in the context.

Note if you doing this in the dev UI you just type in the custom message name as save_model_config (drop the _request)