LINKED LIST [txt mode] ▸ Dust: Pretraining Transformers Withou...
home explore | log in

Dust: Pretraining Transformers Without Backpropagation

qlabs.sh · first added by @owenp · 2026-10-14 · 1 upvotes

log in to save, upvote or flag this.


─── In 0 lists ─────────────────────────────────────────

(not in any lists yet)


─── Discussions ────────────────────────────────────────

* Dust: Pretraining Transformers Without Backpropagation
279 pts · 81 comments · node

─── From the discussion ────────────────────────────────

* Augmented Lagrangian Predictive Coding: training 1000-layer networks without backpropagation
pub.sakana.ai · node
* BOLD: Boolean Logic Deep Learning
proceedings.neurips.cc · node
* https://www.nature.com/articles/s41467-025-64234-z.pdf
nature.com · node
* https://arxiv.org/pdf/2605.12732
arxiv.org · node

─── Related hn threads ─────────────────────────────────

* Backprop Alternative: Augmented Lagrangian Predictive Coding
128 pts · 45 comments · node