I understand that, when we generate images, the prompt itself is first split into tokens, after which those tokens are used by the model to nudge the image generation in a certain direction. I have the impression that the model gets a higher impact of one token compared to another (although I don’t know if I can call it a weight). I mean internally, not as part of the prompt where we can also force a higher weight on a token.

Is it possible to know how much a certain token was ‘used’ in the generation? I could empirically deduce that by taking a generation, stick to the same prompt, seed, sampling method, etc. and remove words gradually to see what the impact is, but perhaps there is a way to just ask the model? Or adjust the python code a bit and retrieve it there?

I’d like to know which parts of my prompt hardly impact the image (or even at all).

    • @polystructOP
      2•1 year ago

      Thanks! I initially didn’t think that such extensions would be what I am looking for (they show which parts of the image are heavily influenced by the prompt), but they do give a visual clue as what parts are definitely important to keep in the prompt, whereas image areas that are hardly getting attention might tell me which parts of the prompt are less impactful.

      And it gives me a few repo’s to dig through the code. Older ones (archived/read-only) are fine, they are often easier to read than the more optimized (but also feature-rich) code that exists nowadays.