3 ms·
The success of the newest GPT models relies on RL to refine the latent space inside the LLM. There's a bottleneck when using humans to refine that space. The ne
by blovescoffee 3y ago
The success of the newest GPT models relies on RL to refine the latent space inside the LLM. There's a bottleneck when using humans to refine that space. The next model or subsequent models will surely use RL techniques like self-play to break through that bottleneck.
- joe_the_user 3y agoThe success of the newest GPT models relies on RL to refine the latent space inside the LLM. There's a bottleneck when using humans to refine that space. Human based RL is used because humans know stuff about the real world and can sort language utterances by this. There's "self play" process that gives a system this sort of knowledge.