Skip to main content
Training is the final step before your Luna metric is ready. You run it using: 
from galileo_luna_ft.training import run_training

training_stats = run_training(config_path="./config.yaml")

​
Inputs Required

  1. Config
  2. Labelled dataset
  3. GPU / training environment

​
What it does

  1. Reads your training config section
  2. Loads the labelled dataset
  3. Fine-tunes a LoRA adapter for your Luna metric
  4. Evaluates baseline vs fine-tuned performance
  5. Writes model artifacts locally and/or to configured destinations

​
Key concepts

  • Metric type: boolean or multi_class (Read More)
  • Prompt template: defines the target response format for the model
  • Model configuration: base model plus LoRA settings

​
Output artifacts

Artifacts are written under: {training.output.local_path}/{training.output.model_name}/ Common outputs include:
  • model + Tokenizer files
  • prompt_template.txt
  • training_metrics.json
  • plots/
See Output artifacts for more detail.

​
Evaluate a trained Luna Metric

After training, you can run evaluation again on a different test set. See Evaluate.

​
Next: configure training

See the detailed Training config reference.