Picture for Quanlu Zhang

Quanlu Zhang

You Only Cache Once: Decoder-Decoder Architectures for Language Models

Add code
May 08, 2024
Viaarxiv icon

Efficient Large Language Models: A Survey

Add code
Dec 23, 2023
Viaarxiv icon

AutoTaskFormer: Searching Vision Transformers for Multi-task Learning

Apr 20, 2023
Figure 1 for AutoTaskFormer: Searching Vision Transformers for Multi-task Learning
Figure 2 for AutoTaskFormer: Searching Vision Transformers for Multi-task Learning
Figure 3 for AutoTaskFormer: Searching Vision Transformers for Multi-task Learning
Figure 4 for AutoTaskFormer: Searching Vision Transformers for Multi-task Learning
Viaarxiv icon

ElasticViT: Conflict-aware Supernet Training for Deploying Fast Vision Transformer on Diverse Mobile Devices

Mar 21, 2023
Figure 1 for ElasticViT: Conflict-aware Supernet Training for Deploying Fast Vision Transformer on Diverse Mobile Devices
Figure 2 for ElasticViT: Conflict-aware Supernet Training for Deploying Fast Vision Transformer on Diverse Mobile Devices
Figure 3 for ElasticViT: Conflict-aware Supernet Training for Deploying Fast Vision Transformer on Diverse Mobile Devices
Figure 4 for ElasticViT: Conflict-aware Supernet Training for Deploying Fast Vision Transformer on Diverse Mobile Devices
Viaarxiv icon

SpaceEvo: Hardware-Friendly Search Space Design for Efficient INT8 Inference

Mar 15, 2023
Figure 1 for SpaceEvo: Hardware-Friendly Search Space Design for Efficient INT8 Inference
Figure 2 for SpaceEvo: Hardware-Friendly Search Space Design for Efficient INT8 Inference
Figure 3 for SpaceEvo: Hardware-Friendly Search Space Design for Efficient INT8 Inference
Figure 4 for SpaceEvo: Hardware-Friendly Search Space Design for Efficient INT8 Inference
Viaarxiv icon

SparDA: Accelerating Dynamic Sparse Deep Neural Networks via Sparse-Dense Transformation

Jan 26, 2023
Figure 1 for SparDA: Accelerating Dynamic Sparse Deep Neural Networks via Sparse-Dense Transformation
Figure 2 for SparDA: Accelerating Dynamic Sparse Deep Neural Networks via Sparse-Dense Transformation
Figure 3 for SparDA: Accelerating Dynamic Sparse Deep Neural Networks via Sparse-Dense Transformation
Figure 4 for SparDA: Accelerating Dynamic Sparse Deep Neural Networks via Sparse-Dense Transformation
Viaarxiv icon

SuperScaler: Supporting Flexible DNN Parallelization via a Unified Abstraction

Jan 21, 2023
Figure 1 for SuperScaler: Supporting Flexible DNN Parallelization via a Unified Abstraction
Figure 2 for SuperScaler: Supporting Flexible DNN Parallelization via a Unified Abstraction
Figure 3 for SuperScaler: Supporting Flexible DNN Parallelization via a Unified Abstraction
Figure 4 for SuperScaler: Supporting Flexible DNN Parallelization via a Unified Abstraction
Viaarxiv icon

Nesting Forward Automatic Differentiation for Memory-Efficient Deep Neural Network Training

Sep 22, 2022
Figure 1 for Nesting Forward Automatic Differentiation for Memory-Efficient Deep Neural Network Training
Figure 2 for Nesting Forward Automatic Differentiation for Memory-Efficient Deep Neural Network Training
Figure 3 for Nesting Forward Automatic Differentiation for Memory-Efficient Deep Neural Network Training
Figure 4 for Nesting Forward Automatic Differentiation for Memory-Efficient Deep Neural Network Training
Viaarxiv icon

Privacy-preserving Online AutoML for Domain-Specific Face Detection

Add code
Mar 16, 2022
Figure 1 for Privacy-preserving Online AutoML for Domain-Specific Face Detection
Figure 2 for Privacy-preserving Online AutoML for Domain-Specific Face Detection
Figure 3 for Privacy-preserving Online AutoML for Domain-Specific Face Detection
Figure 4 for Privacy-preserving Online AutoML for Domain-Specific Face Detection
Viaarxiv icon

AceNAS: Learning to Rank Ace Neural Architectures with Weak Supervision of Weight Sharing

Add code
Aug 06, 2021
Figure 1 for AceNAS: Learning to Rank Ace Neural Architectures with Weak Supervision of Weight Sharing
Figure 2 for AceNAS: Learning to Rank Ace Neural Architectures with Weak Supervision of Weight Sharing
Figure 3 for AceNAS: Learning to Rank Ace Neural Architectures with Weak Supervision of Weight Sharing
Figure 4 for AceNAS: Learning to Rank Ace Neural Architectures with Weak Supervision of Weight Sharing
Viaarxiv icon