News for hackers. Stuff that matters.
Re:Kimi Linear: An Expressive, Efficient Attention Architecture by api on Tuesday July 28, 2026 @11:14AM (#49085188)
“Distillation” is just indirectly pirating the largely pirated training data used to train the original model.

“You stole my warez!”

Re:Kimi Linear: An Expressive, Efficient Attention Architecture by serial_dev on Tuesday July 28, 2026 @11:29AM (#49085413)
If we do it, it's training a model. When they do it, it's distillation attack. - Anthropic
Re:Kimi Linear: An Expressive, Efficient Attention Architecture by esafak on Tuesday July 28, 2026 @11:36AM (#49085513)
"You are distilling what I have rightfully pirated."
Re:Kimi Linear: An Expressive, Efficient Attention Architecture by Barbing on Tuesday July 28, 2026 @03:08PM (#49088477)
Is there even a steelman against this?
Re:Kimi Linear: An Expressive, Efficient Attention Architecture by trollbridge on Tuesday July 28, 2026 @03:11PM (#49088522)
The current steelman argument against this is "China bad, west good".