REVIEW 37 cited by
LongNet: Scaling Transformers to 1,000,000,000 Tokens
Not yet reviewed by Pith; the record is open.
This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.
SPECIMEN: schema-true, not a live event
T0 review · schema-true
One-sentence machine reading of the paper's core claim.
pith:XXXXXXXX · record.json · timestamp
LongNet: Scaling Transformers to 1,000,000,000 Tokens
read the original abstract
Scaling sequence length has become a critical demand in the era of large language models. However, existing methods struggle with either computational complexity or model expressivity, rendering the maximum sequence length restricted. To address this issue, we introduce LongNet, a Transformer variant that can scale sequence length to more than 1 billion tokens, without sacrificing the performance on shorter sequences. Specifically, we propose dilated attention, which expands the attentive field exponentially as the distance grows. LongNet has significant advantages: 1) it has a linear computation complexity and a logarithm dependency between any two tokens in a sequence; 2) it can be served as a distributed trainer for extremely long sequences; 3) its dilated attention is a drop-in replacement for standard attention, which can be seamlessly integrated with the existing Transformer-based optimization. Experiments results demonstrate that LongNet yields strong performance on both long-sequence modeling and general language tasks. Our work opens up new possibilities for modeling very long sequences, e.g., treating a whole corpus or even the entire Internet as a sequence.
Forward citations
Cited by 37 Pith papers
-
RULER: What's the Real Context Size of Your Long-Context Language Models?
RULER shows most long-context LMs drop sharply in performance on complex tasks as length and difficulty increase, with only half maintaining results at 32K tokens.
-
Mamba: Linear-Time Sequence Modeling with Selective State Spaces
Mamba is a linear-time sequence model using input-dependent selective SSMs that achieves SOTA results across modalities and matches twice-larger Transformers on language modeling with 5x higher inference throughput.
-
LongBench: A Bilingual, Multitask Benchmark for Long Context Understanding
LongBench is the first bilingual multi-task benchmark for long context understanding in LLMs, containing 21 datasets in 6 categories with average lengths of 6711 words (English) and 13386 characters (Chinese).
-
Locality Does Not Imply Reachability: Boundary Repair in Block-Sparse Causal Attention
Fixed block causal masks create reachability boundaries where representations depend only on block prefixes, formalized via dependency sets and phase-conditioned coverage functions, with a parameter-free boundary brid...
-
Stream-CQSA: Avoiding Out-of-Memory in Attention Computation via Flexible Workload Scheduling
Stream-CQSA uses CQS-based decomposition to stream exact attention computations for billion-token sequences on limited-memory hardware.
-
Exact Flow Linear Attention: Exact Solution from Continuous-Time Dynamics
Exact Flow Linear Attention derives a closed-form exact update for delta-rule linear attention from continuous-time dynamics, removing Euler discretization error while preserving linear complexity and structure.
-
Leave No Context Behind: Efficient Infinite Context Transformers with Infini-attention
Infini-attention combines compressive memory with masked local attention and long-term linear attention inside each Transformer block to support infinite context length with bounded resources.
-
Vision Mamba: Efficient Visual Representation Learning with Bidirectional State Space Model
Vim is a bidirectional Mamba vision backbone that outperforms DeiT in accuracy on standard tasks while being substantially faster and more memory-efficient for high-resolution images.
-
From Tensor Buffer to Distributed Memory Hierarchy: A Survey of KV Cache Management for LLM Serving
KV-cache serving systems concentrate into five archetypes under a four-axis taxonomy, with ownership explaining residual distributed design variance and seven measurement gaps blocking next steps.
-
AGNFormer I: Reconstruction of AGN spectra using a probabilistic transformer model
An uncertainty-aware transformer reconstructs masked AGN broad lines and spectral halves with 4-16% flux errors and beats eleven purpose-built Lyα-reconstruction algorithms on a blind benchmark.
-
Vortex: Efficient and Programmable Sparse Attention Serving for AI Agents
Vortex provides a programmable frontend and backend for sparse attention in LLM serving, delivering up to 3.46x throughput over full attention while preserving accuracy.
-
Where Does Long-Context Supervision Actually Go? Effective-Context Exposure Balancing
EXACT re-allocates training supervision by inverse frequency of long effective-context targets, improving NoLiMa and RULER scores by 5-18 points on Qwen and LLaMA models without degrading standard QA or reasoning.
-
MOOZY: A Patient-First Foundation Model for Computational Pathology
Patient-level pretraining with a case transformer and multi-task public supervision yields transferable WSI embeddings that beat larger slide-centric models on held-out pathology tasks.
-
HiCI: Hierarchical Construction-Integration for Long-Context Attention
A hierarchical attention module that broadcasts segment summaries and a global context into each segment's keys and values improves long-context LLaMA-2 models over LongLoRA.
-
HiCI: Hierarchical Construction-Integration for Long-Context Attention
HiCI is a new hierarchical attention module that extends LLaMA-2 context from 4K to 100K tokens using under 5.5% extra parameters and improves results on retrieval and code tasks.
-
Stacked from One: Multi-Scale Self-Injection for Context Window Extension
SharedLLM stacks two copies of a short-context LLM so the lower one compresses context into query-aware multi-grained tokens that are injected only at the lowest layers of the upper one, enabling generalization from 8...
-
RAT+: Train Dense, Infer Sparse -- Recurrence Augmented Attention for Dilated Inference
RAT+ pretrains a single dense recurrent-augmented attention model that supports flexible dilated sparse inference after short adaptation, matching dense accuracy at moderate dilation and losing only 1-3 points at high...
-
RAT+: Train Dense, Infer Sparse -- Recurrence Augmented Attention for Dilated Inference
RAT+ pretrains a dense recurrent-augmented attention model once and enables flexible switching to dilated or hybrid sparse attention at inference after short adaptation, with small accuracy loss at high dilation factors.
-
RAT+: Train Dense, Infer Sparse -- Recurrence Augmented Attention for Dilated Inference
A single dense-pretrained RAT+ model can be adapted with 1B tokens to run at dilation sizes D=2..128, matching dense accuracy at D=16 and losing only 1-3 points at D=64 on reasoning and long-context benchmarks, though...
-
BlossomRec: Block-level Fused Sparse Attention Mechanism for Sequential Recommendations
BlossomRec is a sparse attention mechanism that uses two distinct block-level patterns for long-term and short-term interests, fused by a gated output, to reduce computation in sequential recommendation Transformers.
-
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.
-
Positional Encoding via Token-Aware Phase Attention
TAPA adds a learnable phase function to attention to preserve long-range token interactions, enabling direct continual pretraining, length extrapolation, lower perplexity, and stronger retrieval than RoPE-style methods.
-
Accelerating Prefilling via Decoding-time Contribution Sparsity
TriangleMix exploits decoding-time contribution sparsity via a training-free static attention pattern to accelerate LLM prefilling with nearly lossless performance.
-
eLLM: Elastic Memory Management Framework for Efficient LLM Serving
eLLM unifies LLM memory management with virtual tensors and elastic ballooning to CPU memory, reporting 2.32x higher decoding throughput and 3x larger batch sizes for 128K inputs.
-
Beyond Chunking: Discourse-Aware Hierarchical Retrieval for Long Document Question Answering
A hierarchical QA framework converts RST discourse trees into enhanced sentence representations for structure-guided retrieval and reports consistent gains over baselines on four datasets across genres and languages.
-
MoBA: Mixture of Block Attention for Long-Context LLMs
MoBA routes attention over blocks via MoE-style gating to enable dynamic, bias-light long-context attention that matches full attention performance at lower cost.
-
RED-PIM: Reducing Data Movement for Transformers using Processing-in-Memory
RED-PIM restructures attention to use a compact d×d intermediate matrix, reducing PIM inter-bank data movement from O(N²) to O(N) and claiming up to 99.99% latency reduction in simulation.
-
GigaPath-Flash and GigaTIME-Flash: Efficient Pathology Foundation Models for Whole-Slide and Tumor Microenvironment Analysis
Distilled, Apache-2.0-licensed GigaPath-Flash and GigaTIME-Flash models deliver most of the original models' accuracy at a fraction of the compute and memory.
-
Depth-Staggered Fibonacci Spacing for Sparse Attention: Static Schedules Beat Learned Dilation and Extrapolate Where Dense Attention Fails
Static depth-staggered Fibonacci sparse attention improves perplexity over fixed/learned variants and extrapolates to 4x context while dense attention fails.
-
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.
-
SAC-MIL: Spatial-Aware Correlated Multiple Instance Learning for Histopathology Whole Slide Image Classification
A spatial-aware MLP-Mixer with polar rotary position embedding and channel-shifting SAC blocks achieves state-of-the-art slide-level classification on CAMELYON-16, TCGA-LUNG, and TCGA-BRCA.
-
On Efficient Variants of Segment Anything Model: A Survey
A survey that reviews efficient variants of the Segment Anything Model, categorizes acceleration strategies, and provides a unified hardware evaluation on benchmarks.
-
VideoSEMA: a scalable and efficient Mamba-like attention for video understanding
VideoSEMA uses SEMA spatial attention plus softmax temporal attention to outperform larger video transformers and Mamba models on K400/SSv2 and degrade less at 1024² resolution.
-
MATCH: Modulating Attention via In-Context Retrieval for Long-Context Transformers
MATCH augments sparsified attention with an efficient in-context retrieval system to boost performance on long-range recall tasks in transformers.
-
Simple Token-Efficient Vision-Language Model for Case-level Pathology Synoptic Report Generation
A token-efficient VLM with frozen encoder, two-layer MLP aligner, and LLM decoder generates case-level synoptic pathology reports from multi-WSI inputs using 5x magnification patches and two-stage supervised training.
-
Transformer-Based Language Models Across Domain Verticals: Architectures, Applications and Critical Assessment
A survey paper that taxonomizes transformer architectures, reviews domain applications, and critically assesses deployment trade-offs including parameter-energy costs and alignment issues.
-
A Comprehensive Overview of Large Language Models
A survey paper providing an overview of Large Language Models, their background, and recent advances in the field.
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.