Hacker News new | ask | show | jobs
by sthatipamala 1019 days ago
> precise logit control coupled with guidance from forward or backwards chaining on knowledge graphs

What do you mean by this?

1 comments

I'm not 100% sure they're talking about this specifically but logit control/manipulation is often used to conform to a specific schema.

https://github.com/guidance-ai/guidance

I'm going to butcher this explanation - after you've generated your selection of logits but before you sample from them, you check which ones conform to your schema. If you want the only two options to be "true" or "false", then you take any of the logits that would provide invalid answers and lower their probabilities manually.

Another example is structures like JSON can be validated so when your sample is "{'name':'Carl'" you lower the probability of "{" since that would invalidate the json. In fact the only valid ones you'd likely have left would be ",", " ", and "}"