// The last m layers of the transform that encodes a weight table's rows, on one run of height m (an // aligned block of 2^m codeword positions), in place: a second implementation of hf::run_encode(), ...

Replication and stress tests of Sparks of Weight Space Programming: LLMs can understand and write neural network weights (post, logs). Blog post: I Tried to Break Weight Programming, by Xuanqiang ...