llm-kb
← Back to research
Paper

SATS: Sensitivity-Aware Thresholding for MLP Activation Sparsification in LLMs

A new arxiv paper proposes Sensitivity-Aware Thresholding for Sparsity (SATS), a method to calibrate layerwise gate thresholds for MLP activation sparsification using a local sensitivity proxy instead of activation percentiles. The approach aims to reduce computation during LLM inference while preserving model quality.

0 engagement·1 source·Thu, Jul 9, 2026, 11:40 PM
The paper, titled 'Sensitivity-Aware Thresholding and Token Routing for Activation Sparsification in Large Language Models,' introduces SATS, which calibrates thresholds by measuring how much each MLP output changes when its gate is forced open or closed. This replaces percentile-based thresholding, potentially offering better quality-efficiency trade-offs. The work also explores token-level conditional routing to further reduce computation. No specific model names, parameter counts, or benchmark results are provided in the excerpt.

Entities

Large Language Models (LLMs)(concept)Sensitivity-Aware Thresholding for Sparsity (SATS)(concept)MLP activation sparsification(concept)

Related

Paper

Super-Tuning: Activation-Aware Pruning Reused for Sparse Fine-Tuning

A new arxiv paper proposes Super, a sparse PEFT method that reuses Wanda-style activation-weighted magnitude scores from pruning to select a small trainable support, and Supra, a hybrid adapter combining sparse updates with LoRA. This approach reduces memory and compute for fine-tuning LLMs while maintaining performance.

0 engagement·1 source·arxiv
Fri, Jul 10, 2026, 10:55 AM
arXiv
Paper

SLORR: Simple and Efficient In-Training Low-Rank Regularization

Researchers introduced SLORR, a simple, stateless, and architecture-preserving framework for in-training low-rank regularization. It avoids SVDs of large weight matrices, additional trainable parameters, and stateful cached quantities, addressing key limitations of existing methods. This could enable more aggressive compression of modern neural networks without significant accuracy loss.

0 engagement·1 source·arxiv
Thu, Jul 9, 2026, 05:51 PM
arXiv
Paper

STEEL: Sparsity-Aware Fused Attention for Energy-Efficient Long-Sequence Inference on AMD's XDNA NPU

A new paper introduces STEEL, a sparsity-aware fused attention mechanism designed for AMD's XDNA NPU, enabling energy-efficient long-sequence inference on laptop-class SoCs. The approach addresses the challenge of mapping attention mechanisms onto NPUs while maintaining low power consumption, which is critical for agentic workloads that require on-device processing for reliability and privacy.

0 engagement·1 source·arxiv
Fri, Jul 10, 2026, 01:09 PM
arXiv
Paper

Two papers propose token-adaptive KV cache compression for long-context LLMs

Two arXiv papers from July 7, 2026 introduce token-adaptive KV cache compression methods for long-context LLM inference. DepthWeave-KV factorizes key/value states across neighboring layers using shared low-rank bases with token-specific residuals. FreqDepthKV uses shared low-frequency depth components and sparse high-frequency residuals, with an online probe assigning attention heads to different cache modes. Both aim to reduce memory bandwidth while preserving retrieval and reasoning quality.

0 engagement·2 sources·arxiv
Tue, Jul 7, 2026, 05:26 PM
arXiv
Paper

Researchers introduce fully trainable connection optimization for logic gate and lookup table networks

A new method enables partial or full optimization of connections in deep differentiable logic gate networks (LGNs) and lookup table networks (LUTNs) by learning a probability distribution over connections per input pin. The approach outperforms standard fixed-connection LGNs on Yin-Yang, MNIST, and Fashion-MNIST benchmarks while requiring fewer resources.

0 engagement·1 source·arxiv
Fri, Jul 10, 2026, 01:24 PM
arXiv