Cosin V comments on The case for more ambitious language model evals

[Cosin V]

I googled and couldn’t find any info