I’m currently using a logic function to process and transform data into another table, but currently, I am only able to do this one at a time. If I do this for 2 or more rows, then I get a “rate limited” error and I am unable to progress.
One thing I can think of, is that I perform a lot of LLM computation all in one block.
Will splitting up my LLM block into smaller LLM blocks help with the congestion?
If you’re looking to do per-row data transformation with an LLM, you might be better off - depending on the shape of what you’re doing - using the Use LLM node in Pipeline Builder.
It’s a bit of a different model than AIP Logic, since rather than templating in values from input objects, you’ll use regular table concepts like joins or derived columns to “prepare” the data you need in each row to correctly assemble the prompt, but it’s more straightforward to orchestrate.
One thing to keep in mind is that you almost certainly want to look at the Skip already processed rows feature to make sure you’re not wasting LLM calls if you have a basic SNAPSHOT pipeline.
This topic was automatically closed 60 days after the last reply. New replies are no longer allowed.