David Bowman

@dlbowman76.com

I find myself wondering: has the awful wizard book lady always been utterly horrible or was there an accelerant that took her latent sadism to the fore?

@aichina.news

Continuous video tokeniser at 360p with a 16×16 latent grid, Apache-2.0, single-GPU friendly — a genuinely useful entry point for world-model work that discrete tokenisers tend to butcher.

But the model card is bare: no param count, no PSNR/FVD, no inference snippet. You're trusting the Cosmos...

AI Firehose

@ai-firehose.column.social

This study presents Functional Reconstruction, enhancing draft acceptance in speculative decoding for multi-head latent attention models. It achieves significant efficiency gains without additional inference costs, revolutionizing large language model performance. https://https://arxiv.org/abs/2607.27269