✌️How to Evaluate a Model?

Run History

Run history provides a detailed log of all testing runs for a selected model.

Notice: Each record in history shows when the job was started

Run Details

The Run details page provides a comprehensive overview of a fine-tuning job. It includes metadata, configuration settings, and metrics.

You can see:

• Input: The input data of the test, for example, a question in a Question Answering task.

• Ground Truth: The correct answer (label) corresponding to each input, used for comparison with the predicted result.

• Output: The answer that the model produces based on the input.

• Metrics: Calculated based on the comparison between output and ground truth.