Evaluation & BenchmarksAlex Wang et al.

Wang et al. release SuperGLUE

Made by AIWritten by AI (Claude) from the source below. Check anything important against the source.How it's made

Released SuperGLUE — a harder replacement, introduced because performance on GLUE had surpassed the non-expert human level, leaving less room for research

Why it matters

The first instance of the cycle from benchmark saturation to replacement

Source — Primary sources first; secondary reports are labelled “Press”

Paperarxiv.orghttps://arxiv.org/abs/1905.00537

Added in Week 35, 2026 · See that week

On the map — ⑯ Evaluation line

  1. Previous stopGLUE
  2. This record · stopSuperGLUE
  3. Next stopARC benchmark
See it on the map

Nearby on the same axis

Chollet publishes On the Measure of Intelligence, proposing the ARC benchmarkPaperWang et al. release the GLUE benchmarkPaperHendrycks et al. release MMLU (57 tasks)PaperSrivastava et al. release BIG-bench (204 tasks)Paper