undefined | Better HN

0 pointsgunalx3mo ago0 comments

Well problematic because they are using their own verifier as apanem of experts, with their own model trained specifically to satisfy this verifier. On the benchmark runs, they dont mention using human experts to cross validate their scores.

0 comments

cubefox3mo ago

I assume they use self-verification only during RL training to provide the reward signal, but not for benchmarks.

j / k navigate · click thread line to collapse

0 pointsgunalx3mo ago0 comments

Well problematic because they are using their own verifier as apanem of experts, with their own model trained specifically to satisfy this verifier. On the benchmark runs, they dont mention using human experts to cross validate their scores.

0 comments

cubefox3mo ago

I assume they use self-verification only during RL training to provide the reward signal, but not for benchmarks.

j / k navigate · click thread line to collapse