Open gustavopuma opened 4 years ago
It should be fine to load the pretrained model dump. I guess the reason is in the training stage.
The auto memory allocation mechanism doesn't work well in virtual matchines. If you specify auto for epsiode size, it would allocate w.r.t. the physical memory size. You may manually specify a value for episode size. Larger values train faster.
I’m using one virtual machine with 6 CPU / 1 GPU (11GB) with 56 GB RAM, I’m facing memory allocation error to load the pre-trained model Wikidata5m, using the python interface, can you explain why?