Picture for Sangkug Lym

Sangkug Lym

Reducing Activation Recomputation in Large Transformer Models

Add code
May 10, 2022
Figure 1 for Reducing Activation Recomputation in Large Transformer Models
Figure 2 for Reducing Activation Recomputation in Large Transformer Models
Figure 3 for Reducing Activation Recomputation in Large Transformer Models
Figure 4 for Reducing Activation Recomputation in Large Transformer Models
Viaarxiv icon

FlexSA: Flexible Systolic Array Architecture for Efficient Pruned DNN Model Training

Add code
Apr 27, 2020
Figure 1 for FlexSA: Flexible Systolic Array Architecture for Efficient Pruned DNN Model Training
Figure 2 for FlexSA: Flexible Systolic Array Architecture for Efficient Pruned DNN Model Training
Figure 3 for FlexSA: Flexible Systolic Array Architecture for Efficient Pruned DNN Model Training
Figure 4 for FlexSA: Flexible Systolic Array Architecture for Efficient Pruned DNN Model Training
Viaarxiv icon

DeLTA: GPU Performance Model for Deep Learning Applications with In-depth Memory System Traffic Analysis

Add code
Apr 02, 2019
Figure 1 for DeLTA: GPU Performance Model for Deep Learning Applications with In-depth Memory System Traffic Analysis
Figure 2 for DeLTA: GPU Performance Model for Deep Learning Applications with In-depth Memory System Traffic Analysis
Figure 3 for DeLTA: GPU Performance Model for Deep Learning Applications with In-depth Memory System Traffic Analysis
Figure 4 for DeLTA: GPU Performance Model for Deep Learning Applications with In-depth Memory System Traffic Analysis
Viaarxiv icon

PruneTrain: Gradual Structured Pruning from Scratch for Faster Neural Network Training

Add code
Jan 26, 2019
Figure 1 for PruneTrain: Gradual Structured Pruning from Scratch for Faster Neural Network Training
Figure 2 for PruneTrain: Gradual Structured Pruning from Scratch for Faster Neural Network Training
Figure 3 for PruneTrain: Gradual Structured Pruning from Scratch for Faster Neural Network Training
Figure 4 for PruneTrain: Gradual Structured Pruning from Scratch for Faster Neural Network Training
Viaarxiv icon

Mini-batch Serialization: CNN Training with Inter-layer Data Reuse

Add code
Sep 30, 2018
Figure 1 for Mini-batch Serialization: CNN Training with Inter-layer Data Reuse
Figure 2 for Mini-batch Serialization: CNN Training with Inter-layer Data Reuse
Figure 3 for Mini-batch Serialization: CNN Training with Inter-layer Data Reuse
Figure 4 for Mini-batch Serialization: CNN Training with Inter-layer Data Reuse
Viaarxiv icon