DiffusionBlocks – Block-Wise NN Training via Diffusion Interpretation (github.com) 2 points by aanet 1mo ago ↗ HN
[–] aanet 1mo ago ↗ "DiffusionBlocks, a principled framework that partitions transformers into independently trainable blocks, reducing memory requirements proportionally while maintaining competitive performance across diverse architectures and tasks."
1 comment
[ 3.9 ms ] story [ 12.6 ms ] thread