Parallax is a scalable parameterized local linear attention variant that improves LLM pretraining perplexity at 0.6B/1.7B scales with a hardware-aware kernel and shows gains under parameter- and compute-matched controls.
hub Canonical reference
Zoology: Measuring and improving recall in efficient language models
Canonical reference. 83% of citing Pith papers cite this work as background.
13
Pith papers citing it
Background
83% of classified citations
hub tools
citation-role summary
background 6
citation-polarity summary
roles
background 6representative citing papers
VLA stabilizes linear attention by solving regularized least-squares updates with unit-length writes, yielding Jacobian spectral norm exactly 1 and 109x smaller state norms while improving multi-query recall accuracy over standard linear attention and DeltaNet.
Lighthouse Attention enables faster long-context pre-training via gradient-free symmetrical hierarchical compression of QKV while preserving causality, followed by a short full-attention recovery that yields lower loss than standard full-attention training.
OSDN adds online diagonal preconditioning to the Delta Rule, preserving chunkwise parallelism while proving super-geometric convergence and delivering 32-39% recall gains at 340M-1.3B scales.
HubRouter is a sub-quadratic routing primitive using learned hubs that replaces attention layers in hybrid models while delivering competitive perplexity and large throughput gains.
Gated KalmaNet uses exact Kalman gain computation with adaptive gating and Chebyshev iteration to improve SSM performance on long-context tasks over prior approximations like DeltaNet.
Kimi Linear hybridizes linear attention with a new KDA module to beat full attention on tasks while slashing KV cache by 75% and speeding decoding up to 6x.
An 8B Mamba-2-Hybrid with 43% Mamba-2, 7% attention, and 50% MLP layers exceeds an 8B Transformer by 2.65 points on average across 12 tasks and matches it on 23 long-context tasks while enabling up to 8x faster inference.
Gated linear attention Transformers achieve competitive language modeling results with linear-time inference, superior length generalization, and higher training throughput than Mamba.
Making memory decay input-dependent via a lightweight MLP improves log-linear attention performance on associative recall, selective copying, and language modeling, especially for long sequences.
MDN parallelizes stepwise momentum for delta linear attention using geometric reordering and dynamical systems analysis, yielding performance gains over Mamba2 and GDN on 400M and 1.3B models.
Sessa integrates attention within recurrent paths to achieve power-law memory tails and flexible non-decaying selective retrieval, outperforming baselines on long-context tasks.
Gated DeltaNet integrates gating and delta rules into linear transformers, outperforming Mamba2 and DeltaNet on language modeling, reasoning, retrieval, and long-context tasks.
citing papers explorer
-
Parallax: Parameterized Local Linear Attention for Language Modeling
Parallax is a scalable parameterized local linear attention variant that improves LLM pretraining perplexity at 0.6B/1.7B scales with a hardware-aware kernel and shows gains under parameter- and compute-matched controls.
-
Variational Linear Attention: Stable Associative Memory for Long-Context Transformers
VLA stabilizes linear attention by solving regularized least-squares updates with unit-length writes, yielding Jacobian spectral norm exactly 1 and 109x smaller state norms while improving multi-query recall accuracy over standard linear attention and DeltaNet.
-
Long Context Pre-Training with Lighthouse Attention
Lighthouse Attention enables faster long-context pre-training via gradient-free symmetrical hierarchical compression of QKV while preserving causality, followed by a short full-attention recovery that yields lower loss than standard full-attention training.
-
OSDN: Improving Delta Rule with Provable Online Preconditioning in Linear Attention
OSDN adds online diagonal preconditioning to the Delta Rule, preserving chunkwise parallelism while proving super-geometric convergence and delivering 32-39% recall gains at 340M-1.3B scales.
-
HubRouter: A Pluggable Sub-Quadratic Routing Primitive for Hybrid Sequence Models
HubRouter is a sub-quadratic routing primitive using learned hubs that replaces attention layers in hybrid models while delivering competitive perplexity and large throughput gains.
-
Gated KalmaNet: A Fading Memory Layer Through Test-Time Ridge Regression
Gated KalmaNet uses exact Kalman gain computation with adaptive gating and Chebyshev iteration to improve SSM performance on long-context tasks over prior approximations like DeltaNet.
-
Kimi Linear: An Expressive, Efficient Attention Architecture
Kimi Linear hybridizes linear attention with a new KDA module to beat full attention on tasks while slashing KV cache by 75% and speeding decoding up to 6x.
-
An Empirical Study of Mamba-based Language Models
An 8B Mamba-2-Hybrid with 43% Mamba-2, 7% attention, and 50% MLP layers exceeds an 8B Transformer by 2.65 points on average across 12 tasks and matches it on 23 long-context tasks while enabling up to 8x faster inference.
-
Gated Linear Attention Transformers with Hardware-Efficient Training
Gated linear attention Transformers achieve competitive language modeling results with linear-time inference, superior length generalization, and higher training throughput than Mamba.
-
Adaptive Memory Decay for Log-Linear Attention
Making memory decay input-dependent via a lightweight MLP improves log-linear attention performance on associative recall, selective copying, and language modeling, especially for long sequences.
-
MDN: Parallelizing Stepwise Momentum for Delta Linear Attention
MDN parallelizes stepwise momentum for delta linear attention using geometric reordering and dynamical systems analysis, yielding performance gains over Mamba2 and GDN on 400M and 1.3B models.
-
Sessa: Selective State Space Attention
Sessa integrates attention within recurrent paths to achieve power-law memory tails and flexible non-decaying selective retrieval, outperforming baselines on long-context tasks.
-
Gated Delta Networks: Improving Mamba2 with Delta Rule
Gated DeltaNet integrates gating and delta rules into linear transformers, outperforming Mamba2 and DeltaNet on language modeling, reasoning, retrieval, and long-context tasks.