BinaryReader's question on Real Inner Misalignment

From Stampy's Wiki
BinaryReader's question on Real Inner Misalignment id:UgyexTa0FBv4xPLYbyN4AaABAg

But don't all these examples just come down to the quality of training? Sure if you added more examples of correct and incorrect, you would get the expected behavior. I guess a possible caveat to machine learning is that, in order to get desirable results, you need to generate a potentially infinite permutations of correct / incorrect examples (or at least to enough of a fidelity to cover the behavior you expect), and maybe that's not feasible for advanced models.

I wonder if there's been any papers on machine learning training data synthesis.

Tags: None (add tags)
Question Info
Asked by: BinaryReader
OriginWhere was this question originally asked
YouTube (comment link)
On video: We Were Right! Real Inner Misalignment
Date: 2021-10-11T06:37
Asked on Discord? No