Risultati della ricerca - decoding (algorithmssics OR (algorithm OR (algorithm OR algorithm)))

Joint Detection and Decoding: A Graph Neural Network Approach di Clausius, Jannis, Rübenacke, Marvin, Tandler, Daniel, Brink, Stephan ten

Sommario: “... be used to perform joint detection and decoding (JDD). For detection, the GNN is build upon the factor...”
Testo

An Analytical Study of the Min-Sum Approximation for Polar Codes di Chisnevski, Nir, Tal, Ido, Shamai, Shlomo

Sommario: “...The min-sum approximation is widely used in the decoding of polar codes. Although it is a numerical...”
Testo

KVShare: An LLM Service System with Efficient and Effective Multi-Tenant KV Cache Reuse di Yang, Huan, Zhang, Renji, Huang, Mingzhe, Wang, Weijun, Tang, Yin, Li, Yuanchun, Liu, Yunxin, Zhang, Deyu

Sommario: “... and the attention deviations introduced by new tokens during the decoding stage. In this paper, we present a KV...”
Testo

GASCADE: Grouped Summarization of Adverse Drug Event for Enhanced Cancer Pharmacovigilance di Jamil, Sofia, Dabad, Aryan, Reddy, Bollampalli Areen, Saha, Sriparna, Misra, Rajiv, Shakur, Adil A.

Sommario: “... extraction capabilities of Large Language Models (LLMs) with the summarization power of the encoder-decoder...”
Testo

HAMburger: Accelerating LLM Inference via Token Smashing di Liu, Jingyu, Zhang, Ce

Sommario: “... optimization on algorithms, systems, and hardware. However, very few works have fundamentally changed...”
Testo

Branch-and-bound method for calculating Viterbi path in triplet Markov models di Soop, Oskar, Lember, Jüri

Sommario: “...$ is Markov chain, then such path can be efficiently found using the celebrated Viterbi algorithm. However...”
Testo

OpenZL: A Graph-Based Model for Compression di Collet, Yann, Terrell, Nick, Handte, W. Felix, Rozenblit, Danielle, Zhang, Victor, Zhang, Kevin, Goldschlag, Yaelle, Lee, Jennifer, Gorokhovsky, Elliot, Komornik, Yonatan, Riegel, Daniel, Angelov, Stan, Rotem, Nadav

Sommario: “... throughput and low resource utilization. In practice, application-specific compression algorithms...”
Testo

Decoupling Task-Solving and Output Formatting in LLM Generation di Deng, Haikang, Kung, Po-Nien, Peng, Nanyun

Sommario: “... aspects could lead to improved performance. To this front, we introduce Deco-G, a decoding framework...”
Testo

AndesVL Technical Report: An Efficient Mobile-side Multimodal Large Language Model di Jin, Zhiwei, Song, Xiaohui, Wang, Nan, Liu, Yafei, Li, Chao, Li, Xin, Wang, Ruichen, Li, Zhihao, Qi, Qi, Cheng, Long, Hao, Dongze, Zheng, Quanlong, Zhang, Yanhao, Ji, Haobo, Ma, Jian, Zheng, Zhitong, Lin, Zhenyi, Deng, Haolin, Zou, Xin, Yin, Xiaojie, Wang, Ruilin, Cai, Liankai, Liu, Haijing, Qiu, Yuqing, Chen, Ke, Li, Zixian, Xie, Chi, Li, Huafei, Li, Chenxing, Wang, Chuangchuang, Tang, Kai, Zhu, Zhiguang, Tang, Kai, Gao, Wenmei, Wang, Rui, Wu, Jun, Liu, Chao, Xie, Qin, Chen, Chen, Lu, Haonan

Sommario: “... of AndesVL. Moreover, utilizing our cache eviction algorithm -- OKV -- along with customized speculative...”
Testo

Mirror Speculative Decoding: Breaking the Serial Barrier in LLM Inference di Bhendawade, Nikhil, Nishu, Kumari, Kundu, Arnav, Bartels, Chris, Cho, Minsik, Belousova, Irina

Sommario: “...Speculative decoding accelerates LLM inference by using a draft model to look ahead, but gains...”
Testo

A Preliminary Study on the Promises and Challenges of Native Top-$k$ Sparse Attention di Xiu, Di, Tang, Hongyin, Rong, Bolin, Yan, Lizhi, Wang, Jingang, Lu, Yifan, Cai, Xunliang

Sommario: “... into the effectiveness and theoretical mechanisms of the Top-$k$ Attention mechanism during both the decoding...”
Testo

Scalable, Energy-Efficient Optical-Neural Architecture for Multiplexed Deepfake Video Detection di Kashani, Parnian Ghapandar, Chen, Shiqi, Ozcan, Aydogan

Sommario: “... on computationally intensive and energy-demanding inference algorithms, limiting their scalability. Here, we present...”
Testo

Bioinspired Sensory Transduction for Magnetic Profile Recognition and Encryption di Ziyue Miao, Xichen Hu, Kai Liu, Shanming Hu, Guihua Yan, Hongwei Tan, Olli Ikkala, Zhong‐Peng Lv, Bo Peng

Sommario: “... signals. Machine learning algorithms decode the static and dynamic information with ∼100% and 87.5...”
Testo

Self‐Powered Luminescent Barcode Recognition System Based on Triboelectric‐Induced Electroluminescence di Hui Yu, Zheng Tan, Wei Peng

Sommario: “... within 0.2 s through optimized recognition algorithms. This work enables direct decoding of barcode...”
Testo

Sequence‐Defined Oligourethane Isomeric Mixtures for Irreversible Encryption di Lingxiao Guo, Qiangqiang Shi, Jin Zhang, Chunlai Yang, Jiajia Tan, Keyu Li, Jinming Hu, Shiyong Liu

Sommario: “... through a bespoke MS/MS decoding algorithm coupled with a predefined ASCII‐mapping scheme. This work...”
Testo

Sequence‐Defined Oligourethane Isomeric Mixtures for Irreversible Encryption di Lingxiao Guo, Qiangqiang Shi, Jin Zhang, Chunlai Yang, Jiajia Tan, Keyu Li, Jinming Hu, Shiyong Liu

Sommario: “... can be retrieved only through a bespoke MS/MS decoding algorithm coupled with a predefined ASCII...”
Testo

Predicting Classification Accuracy When Adding New Unobserved Classes di Slavutsky, Yuli, Benjamini, Yuval

Sommario: “...-network-based algorithm, "CleaneX", which learns to estimate the accuracy of such classifiers...”
Testo

ByGPT5: End-to-End Style-conditioned Poetry Generation with Token-free Language Models di Belouadi, Jonas, Eger, Steffen

Sommario: “..., and alliteration. We identify and address lack of training data and mismatching tokenization algorithms as possible...”
Testo

Accelerating Learnt Video Codecs with Gradient Decay and Layer-wise Distillation di Peng, Tianhao, Gao, Ge, Sun, Heming, Zhang, Fan, Bull, David

Sommario: “... with conventional coding algorithms in term of compression efficiency. However, most learning-based video...”
Testo

Decoding University Hierarchy and Prestige in China through Domestic Ph.D. Hiring Network di Tian, Chaolin, Jiang, Xunyi, Huang, Yurui, Ma, Langtian, Ma, Yifang

Sommario: “....-granting Chinese universities over three decades. We employ the minimum violation rankings algorithm...”
Testo