New sampler: Min-P (makes sense?)

[clort81]

Prerequisites

• I am running the latest code. Development is very rapid so there are no tagged versions as of now.
• I carefully followed the README.md.
• I searched using keywords relevant to my issue to make sure that I am creating a new issue that is not already open (or closed).
• I reviewed the Discussions, and have a new bug or useful enhancement to share.

Feature Description

"What Min P is doing is simple: we are setting a minimum value that a token must reach to be considered at all. The value changes depending on how confident the highest probability token is.

So if your Min P is set to 0.1, that means it will only allow for tokens that are at least 1/10th as probable as the best possible option. If it's set to 0.05, then it will allow tokens at least 1/20th as probable as the top token, and so on..."

Motivation

Better sampling?

[KerfuffleV2]

llama.cpp already supports Min P.

  --min-p N             min-p sampling (default: 0.1, 0.0 = disabled)

(In fact, I think we got it before anything else.)

[kalomaze]

(In fact, I think we got it before anything else.)

Yup! This was a PR from me that got merged in:
#3841

In fact, you posted the same graph I used to pitch it lol.

[Green-Sky]

closing this as "done" :)