undefined | Better HN

0 pointsTZubiri1y ago0 comments

"It" does not know when it does not know.

But it does know when it has uncertainty.

In the chatgpt api this is logprobs, each generated token has a level of uncertainty, so:

"2+2="

The next token is with almost 100% certainty 4.

"Today I am feeling"

The next token will be very uncertain, it might be "happy", it might be "sad", it might be all sorts of things.

0 comments

gerdesj1y ago

"The next token is with almost 100% certainty 4."

By using the word "almost" with regards 2 + 2 = 4, you have not exactly dispelled LLM "nonsense".

A human (with a modicum of maths knowledge) will know that 2 + 2 = 4 (pure integers - a fact by assertion). A maths worrier will get slightly uncomfortable about 2.0 + 2.0 = 4.0 unless they are ensured that decimal places and accuracy are the same thing and a few other things.

A LLM will almost certainly "know" something that is certain, if its training set is conclusive about that. However, it does not know why and if enough of the training set is suitably ambiguous then it (LLM) will drift off course and seem to spout bollocks - "hallucinate".

TZubiriOP1y ago

You might be in the wrong thread. This is merely a comment about whether LLMs hold a concept of uncertainty, they do.

Also, the next token might be 2 and the next token might be ², the next token could also have been x, these are all valid statements and the LLM might have been uncertain because of them.

2+2=4

2+2=x

2+2=2x

2+2=2x2

2+2=2²

Are all valid statements.

glandium1y ago

And somewhere in its training data, you can be sure there's also 2+2=5.

1 more reply

j / k navigate · click thread line to collapse