Skip to main content
Use this tutorial when a packaged preset already matches your use case. This is the simplest path because the preset metric already defines the metric behavior, labeling prompt, and training prompt template. For the purpose of this tutorial, we will use the “trace input / output only” preset metric, but the same principles apply to all preset metrics.

​
When to choose a preset metric

Choose a preset metric when:
  • you want to use a preset metric that already matches your use case closely, but the current preset metric performance is not satisfactory
  • you want to start from the packaged Luna Studio defaults

​
Minimal end-to-end config

run_steps:
  - data_generation
  - training

pipeline_provider: "local"
metric_name: "toxicity"

data_generation:
  source_data:
    dataset:
      source_type: "huggingface"
      huggingface:
        name: "toxicity_dataset"
  output:
    dataset:
      repo_name: "toxicity-training-dataset"

training:
  dataset:
    name: "toxicity-training-dataset"
  output:
    model_name: "toxicity-model"

​
Why this is simpler than custom

With a preset metric like toxicity, you usually do not need to define:
  • data_generation.metric.name
  • data_generation.metric.type
  • data_generation.metric.input_format
  • data_generation.metric.class_labels
  • training.prompt_template
Those defaults are already provided by the packaged metric config.