Enabling Large Batch Size Training for DNN Models Beyond the Memory Limit While Maintaining Performance

Re-assign community