Model logs
Review model logs to debug deployment or operation issues.
When you deploy your own model, we provide comprehensive logging for that model's deployment and operations to help you debug any issues. Pre-trained models do not have logs. To view logs, go to the "Logs" tab of a model's page. Deploying and running a model generates a lot of logs; use the three filtering tools in the interface to quickly find what you are looking for. You can combine any two or all three filtering systems for even more granular searches.
Model log filters
The text field takes any input, including a RE2 regular expression, and shows only logs that match the input. This is the broadest and most flexible way to filter logs.
The date picker bounds the time range of the logs shown to a specified period.
Log entries fall under one of four levels: ERROR, WARNING, DEBUG, and INFO. Use the "Filter by" button to see one or more levels of logs, for example only ERROR and WARNING. Log levels are assigned according to Python's standard logging levels.
Copy link