Picture for Huaipan Jiang

Huaipan Jiang

Exploiting Activation based Gradient Output Sparsity to Accelerate Backpropagation in CNNs

Add code
Sep 16, 2021
Figure 1 for Exploiting Activation based Gradient Output Sparsity to Accelerate Backpropagation in CNNs
Figure 2 for Exploiting Activation based Gradient Output Sparsity to Accelerate Backpropagation in CNNs
Figure 3 for Exploiting Activation based Gradient Output Sparsity to Accelerate Backpropagation in CNNs
Figure 4 for Exploiting Activation based Gradient Output Sparsity to Accelerate Backpropagation in CNNs
Viaarxiv icon

Structured in Space, Randomized in Time: Leveraging Dropout in RNNs for Efficient Training

Add code
Jun 22, 2021
Figure 1 for Structured in Space, Randomized in Time: Leveraging Dropout in RNNs for Efficient Training
Figure 2 for Structured in Space, Randomized in Time: Leveraging Dropout in RNNs for Efficient Training
Figure 3 for Structured in Space, Randomized in Time: Leveraging Dropout in RNNs for Efficient Training
Figure 4 for Structured in Space, Randomized in Time: Leveraging Dropout in RNNs for Efficient Training
Viaarxiv icon