paper-with-me

Papers

Cross-Attention Speculative Decoding

2025-05-30 · Wei Zhong, Manasa Bharadwaj, Yixiao Wang, Nikhil Verma, Yipeng Ji, Chul Lee

Speculative decoding (SD) is a widely adopted approach for accelerating inference in large language models (LLMs), particularly when the draft and target models are well aligned. However, state-of-the-art SD methods typically rely on tightly coupled, self-attention-based Transformer decoders, often augmented with auxiliary pooling or fusion layers. This coupling makes them increasingly complex and harder to generalize across different models. We present Budget EAGLE (Beagle), the first, to our knowledge, cross-attention-based Transformer decoder SD model that achieves performance on par with leading self-attention SD models (EAGLE-v2) while eliminating the need for pooling or auxiliary components, simplifying the architecture, improving training efficiency, and maintaining stable memory usage during training-time simulation. To enable effective training of this novel architecture, we propose Two-Stage Block-Attention Training, a new method that achieves training stability and convergence efficiency in block-level attention scenarios. Extensive experiments across multiple LLMs and datasets show that Beagle achieves competitive inference speedups and higher training efficiency than EAGLE-v2, offering a strong alternative for architectures in speculative decoding.

📄 PDF Abstract BibTeX arXiv:2505.24544

Code (0)

등록된 구현이 없습니다.

Tasks

Decoder

Methods 이 논문이 사용한 방법론

Attention 설명 없음
Linear Layer A Linear Layer is a projection $\mathbf{XW + b}$.
Adam 설명 없음
Dense Connections Dense Connections, or Fully Connected Connections, are a type of layer in a deep neural network that use a linear operation where every input is connected to every output…
Softmax The Softmax output function transforms a previous layer's output into a vector of probabilities. It is commonly used for multiclass classification. Given an input vector $x$…
Position-Wise Feed-Forward Layer 설명 없음
Absolute Position Encodings Absolute Position Encodings are a type of position embeddings for [Transformer-based models] where positional encodings are…
Label Smoothing Label Smoothing is a regularization technique that introduces noise for the labels. This accounts for the fact that datasets may have mistakes in them, so maximizing the…

Similar Papers 제목 키워드 기반

SpecExtend: A Drop-in Enhancement for Speculative Decoding of Long Sequences

2025-05-27 · Jungyoub Cha, Hyunjong Kim, Sungzoon Cho

Speculative decoding is a widely adopted technique for accelerating inference in large language models (LLMs), but its performance degrades on long inputs due to increased attention cost and reduced draft accuracy. We in…

16kLong-Context Understanding

DREAM: Drafting with Refined Target Features and Entropy-Adaptive Cross-Attention Fusion for Multimodal Speculative Decoding

2025-05-25 · Yunhai Hu, Tianhua Xia, Zining Liu, Rahul Raman 외

Speculative decoding (SD) has emerged as a powerful method for accelerating autoregressive generation in large language models (LLMs), yet its integration into vision-language models (VLMs) remains underexplored. We intr…

feature selection

The Disparate Impacts of Speculative Decoding

2025-10-02 · Jameson Sandler, Ahmet Üstün, Marco Romanelli, Sara Hooker 외 arxiv

The practice of speculative decoding, whereby inference is probabilistically supported by a smaller, cheaper, ``drafter'' model, has become a standard technique for systematically reducing the decoding time of large lang…

SimSD: Simple Speculative Decoding in Diffusion Language Models

2026-06-01 · Junxia Cui, Haotian Ye, Runchu Tian, Hongcan Guo 외 arxiv

Diffusion large language models (dLLMs) have recently emerged as a promising alternative to autoregressive (AR) LLMs, offering faster inference through parallel or blockwise decoding. However, their masked language model…

Benchmarking the Energy Savings with Speculative Decoding Strategies

2026-02-09 · Rohit Dutta, Paramita Koley, Soham Poddar, Janardan Misra 외 arxiv

Speculative decoding has emerged as an effective method to reduce latency and inference cost of LLM inferences. However, there has been inadequate attention towards the energy requirements of these models. To address thi…