addiu

Hi, I tried run_pretraining.py recently, works fine to me. I'm using tensorflow-gpu=1.15.0, cudatoolkit=10.0. First, I think that 960M has very limited VRAM, that could cause your issue. Second, make sure...