paper-with-me

홈 › Papers

Keyless Attention: Value-Space Routing and Value-Only Caching for Efficient Transformers

2026-06-20 · Xin Gao arxiv

We propose Keyless Attention, an attention mechanism that eliminates the key projection entirely, operating over queries and values only. This yields a Value-Only Cache that reduces KV cache memory and access overhead by exactly 50% over standard attention, while matching or exceeding standard attention's decode throughput. Beyond efficiency, we introduce Depth-$m$ Attention Factorization: standard attention computes a depth-2 factorization of the attention bilinear form, while Keyless Attention realizes a depth-$m$ instance of this family. At m=3, Keyless Attention matches the projection matrix count of standard attention via a value-space routing matrix that replaces the key projection and introduces a coupling between routing and retrieval. Experiments across five models and four architectures (GPT-2 280M, GPT-2 557M, Pythia 410M, Qwen2 1.5B, and Llama 3.2 1B) show that Keyless Attention matches or outperforms standard QKV attention on perplexity in 4 out of 5 models. On downstream zero-shot evaluation (GPT-2 557M), Keyless Attention outperforms on 4 out of 5 commonsense reasoning benchmarks, while achieving 50% KV cache reduction throughout.

📄 PDF Abstract BibTeX arXiv:2606.21848

Code (0)

등록된 구현이 없습니다.

Similar Papers 제목 키워드 기반

Multimodal Adaptive Fusion of Face and Gait Features using Keyless attention based Deep Neural Networks for Human Identification

2023-03-24 · Ashwin Prakash, Thejaswin S, Athira Nambiar, Alexandre Bernardino

Biometrics plays a significant role in vision-based surveillance applications. Soft biometrics such as gait is widely used with face in surveillance tasks like person recognition and re-identification. Nevertheless, in p…

Person Recognition

DeBiFormer: Vision Transformer with Deformable Agent Bi-level Routing Attention

2024-10-11 · Nguyen Huu Bao Long, Chenyu Zhang, Yuzhi Shi, Tsubasa Hirakawa 외

Vision Transformers with various attention modules have demonstrated superior performance on vision tasks. While using sparsity-adaptive attention, such as in DAT, has yielded strong results in image classification, the …

image-classificationImage Classificationobject-detectionObject Detection+1

Role-Decoupled Attention Residuals: Separating Matching and Content Retrieval Across Depth

2026-08-02 · Kehan Wang arxiv

Depth-routing residual architectures allow Transformer layers to retrieve earlier representations instead of inheriting only the immediately preceding state. Existing Block Attention Residuals, however, use a single cont…

Dual-Channel Attention Guidance for Training-Free Image Editing Control in Diffusion Transformers

2026-02-20 · Guandong Li arxiv

Training-free control over editing intensity is a critical requirement for diffusion-based image editing models built on the Diffusion Transformer (DiT) architecture. Existing attention manipulation methods focus exclusi…

Image Editing

Gaussian Mixture Attention: Linear-Time Sequence Mixing via Probabilistic Latent Routing

2026-06-09 · Yongchao Huang, Hassan Raza arxiv

The dense token-to-token interaction pattern of standard dot-product attention remains a central bottleneck in scaling Transformer architectures to long contexts. We introduce \textbf{Gaussian Mixture Attention (GMA)}, a…