Introduction
Logging basics
Concept
All of Endpoint executions called via laminar API are logged. For every run, the entire execution trace is logged, so the information you can see in logs includes, but is not limited to:
- Total execution time
- Total execution tokens and cost
- Outputs of every node in the pipeline
- Execution time of every node in the pipeline
- Execution cost of every node in the pipeline
Logs page shows the endpoints’ logs. The logs from the pipeline builder can be accessible in the pipeline builder by going to run outputs’ Run History tab.
Accessing logs
- Go to the Logs page from the navbar on the left side of the page
- Select the endpoint that you want to examine
- Click on each row to see the detailed breakdown of each log on the sidebar.
- Click “Filter” and filter by the required criteria.
Example Logs page
Viewing more details
Simply click on any of the rows in the logs page and you will see the details in the side.
Example log details
Click on each node to see its details.
Exporting to datasets
You can export logs to datasets by selecting the logs, which you want to export, and clicking “Export to dataset”.
If your dataset is already indexed, the new datapoints from the logs will be indexed by the same column. If your dataset is not indexed yet, then new datapoints won’t be indexed.
Tags
When you open each log (by clicking its row), the sidebar’s focus will be on the whole log (not on each particular node from the log).
You’ll see Outputs, Inputs, and Feedback tabs.
If you click Feedback tab, you’ll be able to add tags to your logs.
For example, you can add “correctness” tag and add value 0 or value 1, which will be equivalent to like/dislike user-feedback.
Contact us at founders@lmnr.ai if you’re interested in further reusing this data to improve the performance of your pipelines and agents.