Why do language models feel worse even as benchmarks improve? [pdf](huggingface.co)
2 pointsby scaledsystems3 hours ago1 comment