Linked Presentation: nnScaler: Constraint-Guided Parallelization Plan Generation for Deep Learning TrainingLadder: Enabling Efficient Low-Precision Deep Learning Computing through Hardware-aware Tensor Transformation
