2 citations · 4 across the 17 of their papers we have counts for
17 papers
KV-Pipe: On the Relation Between KV Sharing and Pipeline Parallel Efficiency in LLMs
Maryam Dialameh, Hossein Rajabzadeh, Harish Krishnamoorthy Murali +3
Pipeline parallelism (PP) is widely used to scale large language model (LLM) training, but its efficiency is often limited by stage imbalance and pipeline bubbles. Meanwhile, cross…
CommFuse: Hiding Tail Latency via Communication Decomposition and Fusion for Distributed LLM Training
Rezaul Karim, Austin Wen, Wang Zongzuo +3
The rapid growth in the size of large language models has necessitated the partitioning of computational workloads across accelerators such as GPUs, TPUs, and NPUs. However, these…
Distributed Hybrid Parallelism for Large Language Models: Comparative Study and System Design Guide
Hossam Amer, Rezaul Karim, Ali Pourranjbar +3
With the rapid growth of large language models (LLMs), a wide range of methods have been developed to distribute computation and memory across hardware devices for efficient traini…
EPAS: Efficient Training with Progressive Activation Sharing
Rezaul Karim, Maryam Dialameh, Yang Liu +2
We present a novel method for Efficient training with Progressive Activation Sharing (EPAS). This method bridges progressive training paradigm with the phenomenon of redundant QK (…
FLOP-Efficient Training: Early Stopping Based on Test-Time Compute Awareness
Hossam Amer, Maryam Dialameh, Hossein Rajabzadeh +3
Scaling training compute, measured in FLOPs, has long been shown to improve the accuracy of large language models, yet training remains resource-intensive. Prior work shows that in…
ETT: Expanding the Long Context Understanding Capability of LLMs at Test-Time
Kiarash Zahirnia, Zahra Golpayegani, Walid Ahmed +1
Transformer-based Language Models' computation and memory overhead increase quadratically as a function of sequence length. The quadratic cost poses challenges when employing LLMs…
ECHO-LLaMA: Efficient Caching for High-Performance LLaMA Training
Maryam Dialameh, Rezaul Karim, Hossein Rajabzadeh +5
This paper introduces ECHO-LLaMA, an efficient LLaMA architecture designed to improve both the training speed and inference throughput of LLaMA architectures while maintaining its…
Continuous Self-Improvement of Large Language Models by Test-time Training with Verifier-Driven Sample Selection
Mohammad Mahdi Moradi, Hossam Amer, Sudhir Mudur +3
Learning to adapt pretrained language models to unlabeled, out-of-distribution data is a critical challenge, as models often falter on structurally novel reasoning tasks even while…
Balancing Computation Load and Representation Expressivity in Parallel Hybrid Neural Networks
Mohammad Mahdi Moradi, Walid Ahmed, Shuangyue Wen +3
Attention and State-Space Models (SSMs) when combined in a hybrid network in sequence or in parallel provide complementary strengths. In a hybrid sequential pipeline they alternate…
Accelerating the Low-Rank Decomposed Models
Habib Hajimolahoseini, Walid Ahmed, Austin Wen +1
Tensor decomposition is a mathematically supported technique for data compression. It consists of applying some kind of a Low Rank Decomposition technique on the tensors or matrice…
Is 3D Convolution with 5D Tensors Really Necessary for Video Analysis?
Habib Hajimolahoseini, Walid Ahmed, Austin Wen +1
In this paper, we present a comprehensive study and propose several novel techniques for implementing 3D convolutional blocks using 2D and/or 1D convolutions with only 4D and/or 3D…
SkipViT: Speeding Up Vision Transformers with a Token-Level Skip Connection
Foozhan Ataiefard, Walid Ahmed, Habib Hajimolahoseini +7
Vision transformers are known to be more computationally and data-intensive than CNN models. These transformer models such as ViT, require all the input image tokens to learn the r…
GQKVA: Efficient Pre-training of Transformers by Grouping Queries, Keys, and Values
Farnoosh Javadi, Walid Ahmed, Habib Hajimolahoseini +7
Massive transformer-based models face several challenges, including slow and computationally intensive pre-training and over-parametrization. This paper addresses these challenges…
SwiftLearn: A Data-Efficient Training Method of Deep Learning Models using Importance Sampling
Habib Hajimolahoseini, Omar Mohamed Awad, Walid Ahmed +8
In this paper, we present SwiftLearn, a data-efficient approach to accelerate training of deep learning models using a subset of data samples selected during the warm-up stages of…
Speeding up Resnet Architecture with Layers Targeted Low Rank Decomposition
Walid Ahmed, Habib Hajimolahoseini, Austin Wen +1
Compression of a neural network can help in speeding up both the training and the inference of the network. In this research, we study applying compression using low rank decomposi…
Improving Resnet-9 Generalization Trained on Small Datasets
Omar Mohamed Awad, Habib Hajimolahoseini, Michael Lim +4
This paper presents our proposed approach that won the first prize at the ICLR competition on Hardware Aware Efficient Training. The challenge is to achieve the highest possible ac…
Training Acceleration of Low-Rank Decomposed Networks using Sequential Freezing and Rank Quantization
Habib Hajimolahoseini, Walid Ahmed, Yang Liu
Low Rank Decomposition (LRD) is a model compression technique applied to the weight tensors of deep learning models in order to reduce the number of trainable parameters and comput…