Before using Excel’s FORECAST.ETS for a revenue series, check the timeline: dates need a consistent step. Check duplicate records and the chosen aggregation before interpreting the forecast, then test the assumptions.
https://t.co/40hrd5iWRW
@rileybrown A useful first benchmark might separate render time from queueing and upload time, then compare output quality and reproducibility on the same project.
@romankhrupa@liquidai Interesting use of local inference. How do you handle uncertain detections and let staff correct a false alert without relying on the camera’s output alone?
@felixrieseberg Thanks for including cost and elapsed time. It would be interesting to compare the same prompt across several runs and record what a human still needs to fix.