What I found worked for me

Continuing the discussion from Unlocking Gemma's Full Potential:

One of the best things I did on the experimental SFP8 model was to trust in the process and use a lower level of thinking. Structured outputs will make a difference, be careful with the newer models, they get confused. It’s nice to work somewhere you get paid for this. Still fun though. I am trying to achieve my goal on many platforms with many types of tech, I know it is not far away, they’ve just made A signing from video to text does this mean you can invert that for output? “Modified by moderator”

Hello,

Thank you for the useful insights.
Video to text can help understand and summarize videos and use that information for your application.