I’m working on using google colab to do the training. While I can get the training done, it seems DLC only uses a portion of the GPU in colab. When I checked, DLC was using 5 GB out of almost 16 GB. I was wondering if there is a way to utilize the entire GPU rather than just a portion? The training with GPU is not much faster than using the CPU in this case.
You can increase batch_size in the pose_cfg.yaml file