7.
Training data is becoming a new Wikipedia
As models train on more public traces, internet memory hardens into something closer to a durable encyclopedia
@didier_lopes on Backlist
3 appearances on the backlist front page in the last 30 days.
26.
Z.ai open-sources its RL stack (t.co)
Publishing the full post-training stack makes the gap between research claims and reproducible systems much smaller
55.
Incredible how Z. ai literally has their RL infrastructure open source. (t.co)
Incredible how Z. ai literally has their RL infrastructure open source. The entire OPD post-training of GLM-5.2 took on this slime platform took ~2 days. https:// github.com/THUDM/slime