Q Labs Dust Algorithm: Backpropagation-Free Pretraining Now Beats Adam Baseline
· techtimes.com · popularity 19/100 · 0 similar stories today · Machine Learning Research (Artificial Intelligence)
Read the original at techtimes.com →
Summary
Backpropagation-free AI training reaches a milestone: Q Labs' Dust algorithm uses only forward passes to train transformers, and Noise speedrun community entries have now achieved validation loss 3.338, edging past Adam's baseline of 3.340 -- the first time gradient-free pretraining has matched backpropagation at transformer scale.
Found in: Tech Times. Rating basis: single outlet; outlet tier 2 (unlisted); 1 feeds; feed position 7.