Hacker News new | past | comments | ask | show | jobs | submit
stingraycharles | on: GLM-5.1: Towards Long-Horizon Tasks
Surely at this point it’s part of the training set and the benchmark has lost its value?