CacheTrap achieves 100% targeted attack success on five open-source LLMs by using an efficient search to locate and flip a single bit in the KV cache as a transient trigger, while preserving normal accuracy without the trigger.
hub
Massive Activations in Large Language Models
33 Pith papers cite this work. Polarity classification is still indexing.
33
Pith papers citing it
abstract
We observe an empirical phenomenon in Large Language Models (LLMs) -- very few activations exhibit significantly larger values than others (e.g., 100,000 times larger). We call them massive activations. First, we demonstrate the widespread existence of massive activations across various LLMs and characterize their locations. Second, we find their values largely stay constant regardless of the input, and they function as indispensable bias terms in LLMs. Third, these massive activations lead to the concentration of attention probabilities to their corresponding tokens, and further, implicit bias terms in the self-attention output. Last, we also study massive activations in Vision Transformers. Code is available at https://github.com/locuslab/massive-activations.
hub tools
citation-role summary
background 2
method 1
citation-polarity summary
representative citing papers
YARD is a training-free method using Y-shaped decoder architecture and register tokens to improve contrastive decoding for hallucination reduction in LVLMs with lower latency.
Bayesian Filtering Transformer reframes attention as precision-weighted kriging and residual connections as Kalman updates, delivering gains on cold-start recommendation and noisy LLM fine-tuning tasks.
Massive activations first appear in a single ME Layer due to RMSNorm and FFN, remain invariant thereafter, and a simple softening method raises LLM performance while reducing attention sinks.
Attention sinks in LVLM create a global-vs-local trade-off that a layer-wise gating module can balance to improve multimodal benchmark performance.
FlashAttention-3 achieves 1.5-2x speedup on H100 GPUs for attention, reaching 740 TFLOPs/s (75% utilization) in FP16 and near 1.2 PFLOPs/s in FP8 while cutting numerical error by 2.6x versus baseline FP8 attention.
K-sparse autoencoders with dead-latent fixes produce clean scaling laws and better feature quality metrics that improve with size, shown by training a 16-million-latent model on GPT-4 activations.
Contribution Weights combine attention, value magnitude, and directional alignment to measure token influence more faithfully than attention alone, and show attention sinks actively suppress information via a convex sink-rate to output-norm relationship.
OScaR mitigates token norm imbalance via canalized rotation and omni-token scaling to enable near-lossless INT2 KV cache quantization with up to 3x decoding speedup and 5.3x memory reduction.
UniRefiner uses contrastive registers and a dual alignment objective to remove three categories of spurious tokens from pre-trained ViTs, yielding up to 9.4% mIoU gains on ADE20K and 22% zero-shot segmentation improvements.
A Weibull diagnostic framework classifies transformer weight matrices into consistent functional classes via the shape parameter k and tracks training progress via the scale parameter lambda across multiple architectures.
Register tokens enhance pixel-space DiT training and output quality via cleaner high-noise feature maps, and a dual-stream design adds further gains with little overhead.
LVLMs show vocabulary hijacking by inert tokens that decode to hijacking anchors; HABI locates them, NHAR finds resilient heads, and HAVAE boosts those heads to cut hallucinations.
Outlier tokens in DiTs are addressed with Dual-Stage Registers, which reduce artifacts and improve image generation on ImageNet and text-to-image tasks.
TIGS detects backdoor-induced attention collapse in LLMs and applies content-aware tail-risk screening plus intrinsic geometric smoothing to suppress attacks while preserving normal performance.
GRACE reframes KV cache channel pruning as graph optimization to find a near-optimal subset, achieving 60% compression with negligible degradation and outperforming prior methods.
Prophecy infers formal properties of feed-forward neural networks by extracting rules from neuron activation patterns that imply desirable output behaviors.
Applying a head-specific sigmoid gate after SDPA in LLMs boosts performance and stability by adding non-linearity and query-dependent sparse modulation while reducing attention sinks.
Attention sinks emerge in language models from softmax-induced token dependence on attention scores and do not appear when using sigmoid attention without normalization in models up to 1B parameters.
PyramidKV dynamically compresses KV cache across layers following pyramidal information funneling, matching full performance at 12% retention and outperforming alternatives at 0.7% retention with up to 20.5 accuracy gains.
Prototype-Based Sparse Steering decomposes query activations with SAEs and optimizes sparse features via gradients to steer LLM outputs toward specific behaviors.
HyperLens reveals that deeper transformer layers magnify small confidence changes into fine-grained trajectories, allowing quantification of cognitive effort where complex tasks demand more and standard SFT can reduce it.
Colinearity-Decay regularizer trains ViTs that maintain or improve full-precision accuracy while delivering higher accuracy after low-bit quantization on ImageNet and COCO tasks.
OSC separates token-persistent outlier channels in activations into a compact high-precision tensor for dual-path 4-bit GEMM computation, limiting accuracy loss to roughly 1-2 points on Qwen3 models while delivering up to 1.78x speedup over W8A8 baselines.
citing papers explorer
-
CacheTrap: Unveiling a Stealthier Gray-Box Trojan against LLMs
CacheTrap achieves 100% targeted attack success on five open-source LLMs by using an efficient search to locate and flip a single bit in the KV cache as a transient trigger, while preserving normal accuracy without the trigger.
-
YARD: Y-Architecture Register Decoding for Efficient Hallucination Mitigation in Large Vision-Language Models
YARD is a training-free method using Y-shaped decoder architecture and register tokens to improve contrastive decoding for hallucination reduction in LVLMs with lower latency.
-
Precision Tracked Transformer via Kalman Filtering, Kriging and Process Noise
Bayesian Filtering Transformer reframes attention as precision-weighted kriging and residual connections as Kalman updates, delivering gains on cold-start recommendation and noisy LLM fine-tuning tasks.
-
A Single Layer to Explain Them All:Understanding Massive Activations in Large Language Models
Massive activations first appear in a single ME Layer due to RMSNorm and FFN, remain invariant thereafter, and a simple softening method raises LLM performance while reducing attention sinks.
-
When Sinks Help or Hurt: Unified Framework for Attention Sink in Large Vision-Language Models
Attention sinks in LVLM create a global-vs-local trade-off that a layer-wise gating module can balance to improve multimodal benchmark performance.
-
FlashAttention-3: Fast and Accurate Attention with Asynchrony and Low-precision
FlashAttention-3 achieves 1.5-2x speedup on H100 GPUs for attention, reaching 740 TFLOPs/s (75% utilization) in FP16 and near 1.2 PFLOPs/s in FP8 while cutting numerical error by 2.6x versus baseline FP8 attention.
-
Scaling and evaluating sparse autoencoders
K-sparse autoencoders with dead-latent fixes produce clean scaling laws and better feature quality metrics that improve with size, shown by training a 16-million-latent model on GPT-4 activations.
-
Contribution Weights: A Geometrical Analysis of Self-Attention Transformers
Contribution Weights combine attention, value magnitude, and directional alignment to measure token influence more faithfully than attention alone, and show attention sinks actively suppress information via a convex sink-rate to output-norm relationship.
-
OScaR: The Occam's Razor for Extreme KV Cache Quantization in LLMs and Beyond
OScaR mitigates token norm imbalance via canalized rotation and omni-token scaling to enable near-lossless INT2 KV cache quantization with up to 3x decoding speedup and 5.3x memory reduction.
-
UniRefiner: Teaching Pre-trained ViTs to Self-Dispose Dross via Contrastive Register
UniRefiner uses contrastive registers and a dual alignment objective to remove three categories of spurious tokens from pre-trained ViTs, yielding up to 9.4% mIoU gains on ADE20K and 22% zero-shot segmentation improvements.
-
A Two-Parameter Weibull Framework for Diagnosing Transformer Weight Distributions
A Weibull diagnostic framework classifies transformer weight matrices into consistent functional classes via the shape parameter k and tracks training progress via the scale parameter lambda across multiple architectures.
-
Registers Matter for Pixel-Space Diffusion Transformers
Register tokens enhance pixel-space DiT training and output quality via cleaner high-noise feature maps, and a dual-stream design adds further gains with little overhead.
-
Vocabulary Hijacking in LVLMs: Unveiling Critical Attention Heads by Excluding Inert Tokens to Mitigate Hallucination
LVLMs show vocabulary hijacking by inert tokens that decode to hijacking anchors; HABI locates them, NHAR finds resilient heads, and HAVAE boosts those heads to cut hallucinations.
-
Taming Outlier Tokens in Diffusion Transformers
Outlier tokens in DiTs are addressed with Dual-Stage Registers, which reduce artifacts and improve image generation on ImageNet and text-to-image tasks.
-
Defusing the Trigger: Plug-and-Play Defense for Backdoored LLMs via Tail-Risk Intrinsic Geometric Smoothing
TIGS detects backdoor-induced attention collapse in LLMs and applies content-aware tail-risk screening plus intrinsic geometric smoothing to suppress attacks while preserving normal performance.
-
Graph-Guided Adaptive Channel Elimination for KV Cache Compression
GRACE reframes KV cache channel pruning as graph optimization to find a near-optimal subset, achieving 60% compression with negligible degradation and outperforming prior methods.
-
Prophecy: Inferring Formal Properties from Neuron Activations
Prophecy infers formal properties of feed-forward neural networks by extracting rules from neuron activation patterns that imply desirable output behaviors.
-
Gated Attention for Large Language Models: Non-linearity, Sparsity, and Attention-Sink-Free
Applying a head-specific sigmoid gate after SDPA in LLMs boosts performance and stability by adding non-linearity and query-dependent sparse modulation while reducing attention sinks.
-
When Attention Sink Emerges in Language Models: An Empirical View
Attention sinks emerge in language models from softmax-induced token dependence on attention scores and do not appear when using sigmoid attention without normalization in models up to 1B parameters.
-
PyramidKV: Dynamic KV Cache Compression based on Pyramidal Information Funneling
PyramidKV dynamically compresses KV cache across layers following pyramidal information funneling, matching full performance at 12% retention and outperforming alternatives at 0.7% retention with up to 20.5 accuracy gains.
-
Steered Generation via Gradient-Based Optimization on Sparse Query Features
Prototype-Based Sparse Steering decomposes query activations with SAEs and optimizes sparse features via gradients to steer LLM outputs toward specific behaviors.
-
HyperLens: Quantifying Cognitive Effort in LLMs with Fine-grained Confidence Trajectory
HyperLens reveals that deeper transformer layers magnify small confidence changes into fine-grained trajectories, allowing quantification of cognitive effort where complex tasks demand more and standard SFT can reduce it.
-
Colinearity Decay: Training Quantization-Friendly ViTs with Outlier Decay
Colinearity-Decay regularizer trains ViTs that maintain or improve full-precision accuracy while delivering higher accuracy after low-bit quantization on ImageNet and COCO tasks.
-
OSC: Hardware Efficient W4A4 Quantization via Outlier Separation in Channel Dimension
OSC separates token-persistent outlier channels in activations into a compact high-precision tensor for dual-path 4-bit GEMM computation, limiting accuracy loss to roughly 1-2 points on Qwen3 models while delivering up to 1.78x speedup over W8A8 baselines.
-
Noise Steering for Controlled Text Generation: Improving Diversity and Reading-Level Fidelity in Arabic Educational Story Generation
Residual-stream noise injection raises narrative diversity in Arabic educational stories while preserving reading-grade level, outperforming high-temperature sampling across five 7-9B models.
-
SnapMLA: Efficient Long-Context MLA Decoding via Hardware-Aware FP8 Quantized Pipelining
SnapMLA achieves up to 1.91x higher throughput in long-output MLA decoding using FP8 quantization and specialized kernels while keeping benchmark quality near the BF16 baseline.
-
Locate, Steer, and Improve: A Practical Survey of Actionable Mechanistic Interpretability in Large Language Models
The survey organizes mechanistic interpretability techniques into a Locate-Steer-Improve framework to enable actionable improvements in LLM alignment, capability, and efficiency.
-
MiMo-V2-Flash Technical Report
MiMo-V2-Flash is a 309B/15B MoE model trained on 27T tokens with hybrid attention and multi-teacher on-policy distillation that matches larger models like DeepSeek-V3.2 while enabling 2.6x faster decoding via repurposed MTP layers.
-
A Simple Plug-in for Improving Eviction-Based KV Cache Compression
VECTOR augments eviction-based KV cache compression with three-way token routing that combines importance scoring and offline regression-based reconstructability estimation to improve quality at high compression ratios.
-
DuQuant++: Fine-grained Rotation Enhances Microscaling FP4 Quantization
DuQuant++ adapts outlier-aware fine-grained rotation to MXFP4 by matching block size to the 32-element microscaling group, enabling a single rotation that smooths distributions and achieves SOTA performance on LLaMA-3 with lower cost.
-
Multi-Gate Residuals
Multi-Gate Residuals stabilizes activation scales in deep residual networks via multi-stream gating and attention pooling without added communication overhead.
- Attention Sinks in Diffusion Transformers: A Causal Analysis
- Sink-Token-Aware Pruning for Fine-Grained Video Understanding in Efficient Video LLMs