diff --git a/README.md b/README.md index 0e18cf1..5910b6c 100644 --- a/README.md +++ b/README.md @@ -1,4 +1,6 @@ -An experiment to see how rating changed along a chain of thought +An experiment to see how rating changes along a chain of thought + +It turns out it's quite unstable, depending on where the chain of thought goes, at least in 8B parameter sized models. ![alt text](img/README-1755664824339-image.png)