Open WhoIsDarth opened 11 months ago
Thanks for reporting this. I will have to investigate a bit, to reproduce it well on my end.
As a workaround, depending on what you do, you could also just pass a prompt
value as a parameter to a query that does not change over time (assuming you have no dynamic constraints).
I do not control the format of the prompt, users send their prompts and I execute them
seems to reproduce when it has long output (similar to the problem in https://github.com/eth-sri/lmql/issues/273) leading to errors and multiple retries of open ai due to extremely long context.
After running the service for a long time, I get a large number of running processes.
I am dynamically generate prompt scripts and pass them into
lmql.run