The State Changers discussed creating a simulation to send data to OpenAI for evaluating models. They wanted to ensure that there's a fail-safe measure in place to stop the simulation if it continuously loops or if there's an API response issue. The State Changers discussed three approaches to stopping it: error out, return, and breaking out the existing loop. They also talked about setting up a conditional that checks for errors and creates a 'strike'. After a certain number of 'strikes', the simulation would stop.
In addition, they discussed how to manage large sets of data with a suggestion to create a system log for runs. They suggested creating an event log related to the runs with each run having a record with a start time and end time. The State Changers also discussed potentially using platforms like Xano, which offers precondition and if-then options, for breaking out of loops when errors occur. There was a concern raised about potential memory leaks while handling larger sets of records. The solution suggested was to observe for crashes and to create a notifier to alert when the system starts and finishes a major task. If the process hasn't finished in a particular timeframe, this should trigger a notification indicating potential problems.
(Source: Office Hours 9/18/2023 )
