NLP AI Firewall Continuous Tests Trial

To give you a sense of the capabilities of Firewall Continuous Tests without making you upload multiple datasets, we’ve created a trial-specific command that will allow you to simulate firewall continuous tests in one go. Configuration of RIME AI Firewall Continuous Tests Trial is done through a JSON configuration file that you pass as an argument to the RIME CLI.

Template

{
  "run_name": "Run Name",                       (REQUIRED)
  "data_info": {...},                           (REQUIRED)
  "model_task": "Task Name",                    (REQUIRED)
  "continuous_info": {                          (REQUIRED)
    "bin_size": "day",                          (REQUIRED)
  },
  "prediction_info": {...}, 
  "tests_config_path": "path/to/tests/config",
  "random_seed": null,
  "workspace_name": "Default",
}

Arguments

• run_name: string, required

  The name under which your run will appear in the UI.

• data_info: DataInfo, required

  Configuration for the data sources to use. For a reference on how to configure a datasource, see Data Configuration.

• model_task: string, required

  Specification of the model task. Currently, the supported values are "Text Classification" and "Named Entity Recognition".

• continuous_info: ContinuousInfo, required

  • bin_size: string, required

    The time period for each firewall continuous tests run. Current accepted values are "day" and "hour".

• prediction_info: PredictionInfo or null, default = null

  Specification of how to handle model predictions for this run. For Firewall Continuous Tests predictions MUST be provided, either as separate files (in which case those should be specified in prediction_info, or as part of the input data files. For a guide on how to specify this info, see NLP Prediction Configuration.

• tests_config_path: string or null, default = null

  Path to a TestSuiteConfig JSON file. If path is null RIME will use the default TestSuiteConfig.

• random_seed: int or null, default = null

  The random seed to use for the run.

• workspace_name: string, default = "Default"

  Workspace this run should be associated with.