English
Related papers

Related papers: Efficient Pretraining Length Scaling

200 papers

Transformers process tokens in parallel but are temporally shallow: at position $t$, each layer attends to key-value pairs computed based on the previous layer, yielding a depth capped by the number of layers. Recurrent models offer…

Machine Learning · Computer Science 2026-04-24 Costin-Andrei Oncescu , Depen Morwani , Samy Jelassi , Alexandru Meterez , Mujin Kwun , Sham Kakade

Speculative decoding is a widely used technique for accelerating inference in large language models (LLMs), but its performance degrades as input length grows, with significant drops even at moderate lengths. Yet, this early degradation has…

Computation and Language · Computer Science 2026-01-21 Jungyoub Cha , Hyunjong Kim , Sungzoon Cho

Huge memory consumption has been a major bottleneck for deploying high-throughput large language models in real-world applications. In addition to the large number of parameters, the key-value (KV) cache for the attention mechanism in the…

Computation and Language · Computer Science 2024-06-05 Haoyi Wu , Kewei Tu

Knowledge distillation (KD) in transformers often faces challenges due to misalignment in the number of attention heads between teacher and student models. Existing methods either require identical head counts or introduce projectors to…

Computer Vision and Pattern Recognition · Computer Science 2025-02-12 Zhaodong Bing , Linze Li , Jiajun Liang

With the success of language pretraining, it is highly desirable to develop more efficient architectures of good scalability that can exploit the abundant unlabeled data at a lower cost. To improve the efficiency, we examine the…

Machine Learning · Computer Science 2020-06-08 Zihang Dai , Guokun Lai , Yiming Yang , Quoc V. Le

Knowledge Distillation (KD) is a critical tool for training Large Language Models (LLMs), yet the majority of research focuses on approaches that rely solely on output logits, neglecting semantic information in the teacher's intermediate…

Computation and Language · Computer Science 2026-05-13 Maxime Guigon , Lucas Dixon , Michaël E. Sander

The attention mechanism has been the core component in modern transformer architectures. However, the computation of standard full attention scales quadratically with the sequence length, serving as a major bottleneck in long-context…

Computation and Language · Computer Science 2026-04-28 Yusheng Zhao , Hourun Li , Bohan Wu , Yichun Yin , Lifeng Shang , Jingyang Yuan , Meng Zhang , Ming Zhang

Decoding in a Transformer based language model is inherently sequential as a token's embedding needs to pass through all the layers in the network before the generation of the next token can begin. In this work, we propose a new…

Machine Learning · Computer Science 2025-08-27 Dylan Cutler , Arun Kandoor , Nishanth Dikkala , Nikunj Saunshi , Xin Wang , Rina Panigrahy

Transformer language models generate text autoregressively, making inference latency proportional to the number of tokens generated. Speculative decoding reduces this latency without sacrificing output quality, by leveraging a small draft…

Machine Learning · Computer Science 2025-10-24 Clara Mohri , Haim Kaplan , Tal Schuster , Yishay Mansour , Amir Globerson

Transformer-based NLP models are powerful but have high computational costs that limit deployment. Finetuned encoder-decoder models are popular in specialized domains and can outperform larger more generalized decoder-only models, such as…

Computation and Language · Computer Science 2024-11-19 Bo-Ru Lu , Nikita Haduong , Chien-Yu Lin , Hao Cheng , Noah A. Smith , Mari Ostendorf

Due to the highly parallelizable architecture, Transformer is faster to train than RNN-based models and popularly used in machine translation tasks. However, at inference time, each output word requires all the hidden states of the…

Computation and Language · Computer Science 2019-09-06 Chengyi Wang , Shuangzhi Wu , Shujie Liu

Transformer-based models have emerged as one of the most widely used architectures for natural language processing, natural language generation, and image generation. The size of the state-of-the-art models has increased steadily reaching…

Hardware Architecture · Computer Science 2025-01-15 Rya Sanovar , Srikant Bharadwaj , Renee St. Amant , Victor Rühle , Saravan Rajmohan

Transformer-based large language models (LLMs) have already achieved remarkable results on long-text tasks, but the limited GPU memory (VRAM) resources struggle to accommodate the linearly growing demand for key-value (KV) cache as the…

Computation and Language · Computer Science 2025-03-21 Shibo Jie , Yehui Tang , Kai Han , Zhi-Hong Deng , Jing Han

Speculative decoding (SD), where a draft model provides multiple candidate tokens for the target model to verify in parallel, has demonstrated significant potential for accelerating LLM inference. Yet, existing SD approaches adhere to a…

Machine Learning · Computer Science 2025-09-22 Enyu Zhou , Kai Sheng , Hao Chen , Xin He

The Key-Value (KV) cache is central to the efficiency of transformer-based large language models (LLMs), storing previously computed vectors to accelerate inference. Yet, as sequence length and batch size grow, the cache becomes a major…

Machine Learning · Computer Science 2025-12-08 Damien Lesens , Beheshteh T. Rakhshan , Guillaume Rabusseau

Transformers excel at sequence modeling but face quadratic complexity, while linear attention offers improved efficiency but often compromises recall accuracy over long contexts. In this work, we introduce Native Hybrid Attention (NHA), a…

Computation and Language · Computer Science 2026-04-16 Jusen Du , Jiaxi Hu , Tao Zhang , Weigao Sun , Yu Cheng

Differentially Private (DP) learning has seen limited success for building large deep learning models of text, and straightforward attempts at applying Differentially Private Stochastic Gradient Descent (DP-SGD) to NLP tasks have resulted…

Machine Learning · Computer Science 2022-11-11 Xuechen Li , Florian Tramèr , Percy Liang , Tatsunori Hashimoto

Linearizing pretrained large language models (LLMs) primarily relies on intra-layer hybrid attention mechanisms to alleviate the quadratic complexity of standard softmax attention. Existing methods perform token routing based on…

Machine Learning · Computer Science 2026-02-03 Weikang Meng , Liangyu Huo , Yadan Luo , Jiawen Guan , Jingyi Zhang , Yingjian Li , Zheng Zhang

Key-value (KV) caching plays an essential role in accelerating decoding for transformer-based autoregressive large language models (LLMs). However, the amount of memory required to store the KV cache can become prohibitive at long sequence…

Machine Learning · Computer Science 2024-05-22 William Brandon , Mayank Mishra , Aniruddha Nrusimha , Rameswar Panda , Jonathan Ragan Kelly

Transformers are considered one of the most important deep learning models since 2018, in part because it establishes state-of-the-art (SOTA) records and could potentially replace existing Deep Neural Networks (DNNs). Despite the remarkable…

Machine Learning · Computer Science 2022-08-23 Hongwu Peng , Shaoyi Huang , Shiyang Chen , Bingbing Li , Tong Geng , Ang Li , Weiwen Jiang , Wujie Wen , Jinbo Bi , Hang Liu , Caiwen Ding