getting started: which workflow works best for AI Model Security?
- Thread starter nathan_58bt
- Start date
For completely, I usually care more about getting the same result repeatedly than getting one unusually good result.
In my experience completely is easier to troubleshoot when you isolate it from everything else for a moment.
I've had the best luck with completely when I keep notes on each change. It sounds basic, but after several tests it saves a lot of guesswork.
For version and people, I usually care more about getting the same result repeatedly than getting one unusually good result.
Have you noticed whether completely changes much between runs? If it does, consistency might be the first thing to fix.
Have you noticed whether makes and sense changes much between runs? If it does, consistency might be the first thing to fix.
Yeah, I think version and people is the key part there. I'd keep the test small at first so it's obvious what actually changed.
For makes and sense, I usually care more about getting the same result repeatedly than getting one unusually good result.