pith. sign in

Reconciling modern deep learning with traditional optimiza- tion analyses: The intrinsic learning rate

1 Pith paper cite this work. Polarity classification is still indexing.

1 Pith paper citing it

citation-role summary

background 1

citation-polarity summary

fields

cs.LG 1

years

2026 1

verdicts

UNVERDICTED 1

roles

background 1

polarities

background 1

representative citing papers

Demystifying Manifold Constraints in LLM Pre-training

cs.LG · 2026-05-06 · unverdicted · novelty 6.0

Manifold constraints via the new MACRO optimizer independently bound activation scales and enforce rotational equilibrium in LLM pre-training, subsuming RMS normalization and decoupled weight decay while delivering competitive performance with convergence guarantees.

citing papers explorer

Showing 1 of 1 citing paper.

  • Demystifying Manifold Constraints in LLM Pre-training cs.LG · 2026-05-06 · unverdicted · none · ref 31

    Manifold constraints via the new MACRO optimizer independently bound activation scales and enforce rotational equilibrium in LLM pre-training, subsuming RMS normalization and decoupled weight decay while delivering competitive performance with convergence guarantees.