pith. sign in

Effective distillation to hybrid xlstm architectures, 2026

2 Pith papers cite this work. Polarity classification is still indexing.

2 Pith papers citing it

fields

cs.LG 2

years

2026 2

verdicts

UNVERDICTED 2

clear filters

representative citing papers

Pretraining Recurrent Networks without Recurrence

cs.LG · 2026-06-04 · unverdicted · novelty 6.0

SMT reduces RNN training to supervised learning on memory transitions (m_t, x_{t+1}) to m_{t+1} obtained from a Transformer encoder, enabling time-parallel training with O(1) gradient paths.

citing papers explorer

Showing 2 of 2 citing papers after filters.

  • Pretraining Recurrent Networks without Recurrence cs.LG · 2026-06-04 · unverdicted · none · ref 43

    SMT reduces RNN training to supervised learning on memory transitions (m_t, x_{t+1}) to m_{t+1} obtained from a Transformer encoder, enabling time-parallel training with O(1) gradient paths.

  • On Subquadratic Architectures: From Applications to Principles cs.LG · 2026-06-10 · unverdicted · none · ref 47

    xLSTM outperforms Mamba-2 and Gated DeltaNet on tasks with complex dependencies because its gating scheme enables more flexible and stable state tracking and memory accumulation.