Single-layer transformer model "HarEmb" showcasing PII SOTA performance (huggingface.co) 1 points by fblgit 4mo ago ↗ HN
[–] fblgit 4mo ago ↗ one of a kind single-transformer block layer, high throughput. The new generation of transformer-based lightweight models for common NLP tasks?
1 comment
[ 3.1 ms ] story [ 14.2 ms ] thread