search
Gradient
Trends
- 1Dust Proposes Pretraining Transformers Without Backpropagation●Dust: Pretraining Transformers Without Backpropagation
Researchers at Qlabs have presented Dust, a method for pretraining transformer models without using backpropagation. The work, described in a research note from qlabs.sh, suggests an alternative to the gradient-based training that underpins virtually all modern deep learning. The approach is drawing attention from machine learning practitioners debating whether backprop-free training could reduce the cost or energy demands of building large language models.
- 2New tool maps the flattest walking routes in San Francisco●Find the flattest route between any two points in SF
A new web tool lets San Francisco residents and visitors find the flattest route between any two points in the city, a boon in a place famous for its steep hills. The project has drawn attention online, with people discussing how useful it could be for cyclists, pedestrians with mobility issues, and anyone tired of climbing SF's notorious gradients.
Repos
- shader-effects-inc/shaders WebGPU components for React, Vue, Svelte, Solid, JS & Framer
- storytold/vectorcraft An open-source, clean-room reimplementation of Adobe Illustrator, built in pure Rust.
- pbakaus/impeccable The design language that makes your AI harness better at design.