PAI-TensorFlow logs include experiment running status logs and LogView logs. The LogView logs of PAI-TensorFlow contain more information about the experiment. If the system reports an error when you run an experiment, you can view the details of errors based on the logs.
The following logs are generated when you submit PAI-TensorFlow tasks by using any
method.
| No. | Description |
|---|---|
| 1 | The LogView logs. You can click the link in your browser to view the LogView logs. |
| 2 | The running status log. |
Running status logs
Running status logs display the running status of the servers on which the distributed
training tasks are running. A log is displayed in the Parameter 1/Parameter 2/Parameter 3 format.
- The value of Parameter 1 indicates the number of nodes that are performing computing tasks.
- The value of Parameter 2 indicates the number of nodes that have completed computing tasks.
- The value of Parameter 3 indicates the total number of nodes.
LogView logs
LogView logs provides the debug information about experiments. You can use the following
methods to view LogView logs:
- Open the link in the preceding figure in your browser. The page that displays the
details of the LogView logs appears.
- To view error logs or check the output information about the code that is used to
run the experiment, double-click ODPS Tasks.
- On the page that shows the running status of workers, view the tasks and workers.
- To view the output information that is generated by the print function when you run
the code, click the
icon in the Stdout column.
- To view the debug information of the service, click the icon in the Stderr column.
