Hadamard rotations for low-bit quantization, and Monarch matrices for structured compression of GPT-2 small.

In the age of machines, what is still unsolved?

A compact walkthrough of bottlenecks, latent spaces, and reconstruction.

Non-greedy acceptance, residual sampling, and why q matters.

Clipping attention logits for MLA with shared rotary keys.

Approximating the serial pre-norm block.

On singular token spaces.

Small but faithful subsets of large point sets.

Play chess alone or with a friend.

Communication deficiencies in optimization.