I encountered an out-of-memory issue while training on the COCO dataset using rtdetr_pytorch. The memory usage (not GPU memory) increases gradually with each step. After 2 epochs, it runs out of 188GB memory.
Here I show the memory usage at intervals of several steps during the first epoch of training.
I encountered an out-of-memory issue while training on the COCO dataset using rtdetr_pytorch. The memory usage (not GPU memory) increases gradually with each step. After 2 epochs, it runs out of 188GB memory. Here I show the memory usage at intervals of several steps during the first epoch of training.