لطفا منتظر بمانید ...
0% Complete
صفحه اصلی
/
سی و چهارمین کنفرانس بین المللی مهندسی برق
Efficient Large Language Models with Zero-Shot Adjustable Acceleration
نویسندگان :
Sajjad Kachuee
1
Mohammad Sharifkhani
2
1- دانشگاه صنعتی شریف
2- دانشگاه صنعتی شریف
کلمات کلیدی :
Zero-Shot Adjustable Acceleration،Preservation Rate،Hidden Activation،Context Contribution
چکیده :
Large Language Models (LLMs) face significant challenges in real-world applications, particularly in balancing computational efficiency and model performance. Acceleration during both fine-tuning and inference is essential for designing efficient architectures. This paper presents a novel training and inference method, termed zero-shot adjustable acceleration, which dynamically adjusts hardware utilization at inference time without additional fine-tuning. The method is applied to recent LLMs and evaluated on multiple classification and text generation benchmarks. Experimental results demonstrate that the approach enables flexible zero-shot acceleration and achieves up to an 11× speedup over baseline methods, while maintaining high model performance.
لیست مقالات
لیست مقالات بایگانی شده
GAN-Driven Image Generation for Metamaterial Absorbers Using Mean and Variance Encoding
Atefe Shahsavaripour - Mohammad Hossein Badiei - Leila Yousefi - Ahmad Kalhor
A Coronavirus Herd Immunity Optimizer For Intrusion Detection System
Amir Soltany Mahboob - Hadi Shahriar Shahhoseini - Mohammad Reza Ostadi Moghaddam - Shima Yousefi
High Step up DC/DC Converter with Low Input Current Ripple and Low Voltage Stress on Semiconductors
Saed Mahmoud Alilou - Mohammad Maalandish - Soheil Nouri - Seyed Hossein Hosseini
Efficient Data Detection in Massive Multiuser MIMO Systems Using Manifold Optimization
Ali Ghanbarpour - Mojtaba Amiri - Ali Olfat
Flexible Generation Expansion Planning Considering Representative Days of Load and Renewable Variations
Peyman Amirian - Zeinab Maleki - Mohammad-Amin Pourmoosavi - Turaj Amraee
Design and Implementation of CAN Bus Monitoring Module for Lithium Battery Management System
Shakila Kazempourdizaji - Amir Mohammad Moazami Goudarzi - Majid Shalchian
Synergy of Deep Learning and Artificial Potential Field Methods for Robot Path Planning in the Presence of Static and Dynamic Obstacles
Mohammad Amin Basiri - Shirin Chehelgami - Erfan Ashtari - Mehdi Tale Masouleh - Ahmad Kalhor
A Contrast-Aware Real-Time Defect Detection Framework for PCBs Using CLAHE-Enhanced YOLOv8
Mehran Kushesh - Mohammadreza Hassan Nejad Biballan
A Framework for Plant Topology Extraction Using Process Mining and Alarm Data
Amir Neshastegaran - Ali Norouzifar - ایمان ایزدی
DOA estimation of acoustic signals using stacked products of cross-correlations and coherence factor
Mojtaba Amiri - Amir Akhavan - Ahmad Tavakol - Ehsan Rouhani
بیشتر
ثمین همایش، سامانه مدیریت کنفرانس ها و جشنواره ها - نگارش 44.7.2
