3 ms·
I've had success using a multistage inference approach. Establish a set of rules, then have the LLM ask itself whether its current suggested world building comp
by Taek 4y ago
I've had success using a multistage inference approach. Establish a set of rules, then have the LLM ask itself whether its current suggested world building complies with all the rules. Then have it tweak its own story.
Finally, after all that, present the output to the user. It means inference takes 4x as long because you are doing multiple rounds of inference before actually presenting output but the final output you get is way more stable/consistent/accurate
- williamcotton 4y agoThat seems to be similar to this approach, at least at a cursory level: Check Your Facts and Try Again: Improving Large Language Models with External Knowledge and Automated Feedback https://arxiv.org/abs/2302.12813 https://arxiv.org/abs/2302.12813 The “facts” being derived not from Wikipedia but rather the rules of the fictional world.
- sitkack 4y agoI have had it make new languages based off existing languages but change the weights on how thing are used and what semantics are allowed or fixed. Are you building in a recursive layer by asking it to take on a role while also targeting its output at a constructed user persona?
- Taek 4y agoI hadn't thought about targeting a constructed user persona, that's a neat idea. But otherwise yes, its a recursive model that runs inference a bunch of times to improve the quality of its output.