Introducing Low-Level GPU Virtual Memory Management | NVIDIA Technical Blog
Optimize GPU memory consumption: Decrease heap usage at the beginning of the training and allow GPU to use 100% fragmentation. · Issue #44118 · tensorflow/tensorflow · GitHub
GPU memory surge after training epochs causing CUDA memory error - Trainer - PyTorch Lightning
python - How can I decrease Dedicated GPU memory usage and use Shared GPU memory for CUDA and Pytorch - Stack Overflow