4 ms·
Classification is easier than generation
by WithinReason 2y ago
Classification is easier than generation
- Uehreka 2y agoYeah no I’m gonna need more than that chief. Everything I know about LLMs says the opposite.
- WithinReason 2y agoLook up Generative Adversarial Networks, that's their basic principle. You're not classifying just text, you're classifying entire web pages.
- TeMPOraL 2y agoI'd say the whole point of GAN is that generation is cheaper than classification, therefore an effective brute-force way of making a good classifier is to generate an infinite supply of examples with a-priori known classification, and pit it against a classifier.
- mafuku 2y agoYeah, but the theoretical endpoint of training a GAN is that the generator gets so good that the discriminator has to resort to guesses and become unable to tell with any sort of accuracy as to whether the example it is shown is real or generated.
- WithinReason 2y agoI don't think that ever happens in training, at least in the image domain. The classifier can always can find some subtle clue.
- pona-a 2y agoSadly, this didn't seem to track for LLMs. Even OpenAI gave up on trying to detect its own outputs. > As of July 20, 2023, the AI classifier is no longer available due to its low rate of accuracy. We are working to incorporate feedback and are currently researching more effective provenance techniques for text, and have made a commitment to develop and deploy mechanisms that enable users to understand if audio or visual content is AI-generated.
- WithinReason 2y agoYou're not classifying just text, you're classifying entire web pages. If it's easy to tell for a human that it's SEO spam, it's easy for a classifier.
- geekraver 2y agoNot with standard LLM models. Generating garbage is easier; deciding that it’s garbage much more difficult. This is an inherent problem with “prompt optimizers” like DSPy.