Could you try a prompt that tells it to end a sentence with a particular word, and see how that word casts its influence back over the sentence? I know that this works with GPT-3, but I didn’t really understand how it could.
Interesting topic! I’m not confident this lens would reveal much about it (vs. attention maps or something), but it’s worth a try.
I’d encourage you to try this yourself with the Colab notebook, since you presumably have more experience writing this kind of prompt than I do.
Could you try a prompt that tells it to end a sentence with a particular word, and see how that word casts its influence back over the sentence? I know that this works with GPT-3, but I didn’t really understand how it could.
Interesting topic! I’m not confident this lens would reveal much about it (vs. attention maps or something), but it’s worth a try.
I’d encourage you to try this yourself with the Colab notebook, since you presumably have more experience writing this kind of prompt than I do.