Xuan Truong Nguyen Department of Next Generation Semico Convergence

PEAT: Pseudo-Error Assessment for GPU Kernel Validation in DNN Training

The applicability of PEAT is demonstrated by presenting the results and analysis using GPUs from the two most popular vendors, NVIDIA V100 and AMD MI250, on various AI models, from vision tasks to language models, for both pretraining and finetuning scenarios.