MIT and Adobe Propose Latent Lattice Language Models
MIT and Adobe researchers propose Latent Lattice Language Models, adding a lateral computation axis to transformers, cutting perplexity of a 406M model from 15.41 to 14.40.
2026-11-19 ~ 2026-11-20 · 3 related posts
- MIT & Adobe's Latent Lattice LMs cut perplexity 15.41→14.40, matching 50% larger Transformers — burny_tech · 2026-11-19
- Latent Lattice Language Models propose adding a horizontal axis of computation to the Transformer residual stream — burny_tech · 2026-11-19
- Latent Lattice Language Models: adding a horizontal computation axis to the residual stream — HanGuo97 · 2026-11-20