Pipeline Builder Use LLM Skip Computing

How does the Skip Computing Already Processed Rows handle rows that were rate limited?

For example, take a case where I processed 10 rows, and 5 were rate limited. If I have this setting on and then rebuild, will it now attempt to process those 5 rows or will it skip them?

It will re-attempt to process those 5 rows! Our cache won’t keep track of any rows that error-ed or are null

1 Like

@helenq Is there a way to get insight into this re-run?

When I click build from the dataset view (force-build) the build runs for about two minutes (which seems very short) and the number of nulls does not change.


)

Hey! We’re actually currently working on getting our build metrics to have more details re token usage and other things that may help with llm runs. Unfortunately it’s not quite ready yet so I think the best path here would be to see what the errors are (if you toggle the show errors option in your LLM node)

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.