I tried to work with the model to fix the issues, but after a few prompts, and testing the "app" once or twice each time, I hit my limit for free usage of AI Studio and/or Imagen AI. I could have picked it up on the next day, or just pulled out the credit card. But I had seen enough at that point.

I asked it to create a minimalist "Pen and Paper" styled Instagram Carousel post generator. Because I see them all over Instagram anyway. It would generate the images, but the text wasn't part of the image. So if I downloaded the image it was just the background.

Once I got to the "hands on" part I had some trouble with anything more complicated than an app that generates an image from a prompt. I deviated from the example by asking for the app to be written in Python + Django. This was ignored and the app was written in Typescript.

I’m more interested in self-hostable, open models, but I figure I can take what I learn from these two and apply it to working with models via #Ollama.

I'm downgrading my main PC from my desktop (with an RX 7600 GPU) to a laptop with an Nvidia Quadro P1000 GPU. I feel this in a very self-inflicted kind of way.

I think ThePrimeagen is pretty spot on with this. youtube.com/shorts/qLVCE...

I think public schools and other publicly funded organizations should be looking towards open source (including OS choice) in general. Schools, libraries, local/state/federal gov, etc....