3 ms·
We could call this "generative adversarial network" (GAN) :) https://en.wikipedia.org/wiki/Generative_adversarial_network https://en.wikipedia.org/wiki/Generat
by vincnetas 3mo ago
We could call this "generative adversarial network" (GAN) :)
https://en.wikipedia.org/wiki/Generative_adversarial_network https://en.wikipedia.org/wiki/Generative_adversarial_network
- wwind123 3mo agoThis kind of approach would generally still need human guidance, otherwise these models might get stuck in weird niche corners of the problem space that would not be relevant to any real world project.
- ben_w 3mo agoWe could call this "reinforcement learning from human feedback" (RLHF) :) https://en.wikipedia.org/wiki/Reinforcement_learning_from_human_feedback https://en.wikipedia.org/wiki/Reinforcement_learning_from_hu...