View Task Details & Logs

Overview

The Task Details & Logs page provides a complete history of task runs, including when tasks were executed, who ran them (or if they were triggered automatically), and the configuration details used during the run. This includes metadata such as sampling rate, filters, evaluation templates, and LLM parameters. Any errors or warnings will also be displayed in the logs.

This page helps you dig into the specifics of each task run and answer questions like:

• Was the task successful, or were there any errors?

• Has the evaluation template been recently updated?

• Who triggered the most recent task run?

• What sampling rate was used during the run?

• Which spans did the task run on?

Use this page to troubleshoot issues, track changes, and better understand how your tasks are performing over time.

View Task Details and Logs

From the Tasks page, click on any task to view its detailed information and logs. This view helps you track how and when a task was run, including parameters like sampling rate, spans, and evaluation settings.

Key things you can see in the task details:

• Run timing: Start and end times of the task.

• Who ran the task and when.

• Sampling rate, date range and filters used during the run.

• LLM parameters and the evaluation template.

• Any error messages that occurred during the run.

• The number of successes, errors and skips.

The logs are especially useful for troubleshooting. If something goes wrong—like an invalid API key or configuration issue—errors will appear here, helping you quickly diagnose and fix the problem.

View Traces

The View Traces button lets you jump directly to the batch of spans where the task was run. When you click it, the system automatically applies the date range and filters that were used, so you can quickly inspect the data tied to that task.

Keep in mind that if the task used a sampling rate below 100%, or had a cap on the number of spans, not every span will be labeled.

Use this feature to seamlessly connect task results with the actual span data for more targeted analysis and troubleshooting.

Test in Code

For technical users who prefer to debug and iterate in a code environment, the Test in Code button is your go-to tool. It provides pre-built Python SDK functions that let you:

• Pull the exact spans used during the task run

• Process that data with the task’s configuration

• Generate eval labels programmatically

This makes it easy to troubleshoot issues, experiment with different prompts or model versions, and refine your evaluation process—all directly in code.

If you’d rather iterate quickly in the UI, check out the Test in Playground workflow. It’s perfect for rapid prototyping with prompt templates and LLM parameters.