yeandy commented on issue #21437: URL: https://github.com/apache/beam/issues/21437#issuecomment-1262842580
Several questions around the error handling: 1. We're working with a batch of data. If inference fails, do we consider that to be 1 `num_failed_inferences`? Or perhaps it should be renamed to `num_failed_batch_inferences`? It may depend on framework, but is there a way to figure out which specific data points in a batch fail? In which case we can also track `num_failed_inferences`. 2. Do we want to do any retries for the inference call? Maybe we shouldn't, since the default behavior for a model inference failure is to throw an exception. But if we do want retries, how many times? If it succeeds in, say 5 tries, do we swallow the error and ignore increasing the counter? Or do we still increase the counter? 3. Or should the pipeline always fail? 4. Should we give users a flag to choose between the behavior? @rezarokni Do you have any idea on points 2-4? -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
