Alen Smajic

icon company Bosch Research icon location Frankfurt am Main, Germany icon location Computer Vision Research Engineer @ Bosch Research. Building AI to make self-driving cars reality.

Results 3 comments of Alen Smajic

RuntimeError: CUDA out of memory. Tried to allocate 3.06 GiB (GPU 0; 6.00 GiB total capacity; 3.58 GiB already allocated; 1024.00 MiB free; 3.59 GiB reserved in total by PyTorch)

Did you try to restart your PC and execute the code after? This often increases the available GPU memory.

RuntimeError: CUDA out of memory. Tried to allocate 3.06 GiB (GPU 0; 6.00 GiB total capacity; 3.58 GiB already allocated; 1024.00 MiB free; 3.59 GiB reserved in total by PyTorch)

I am not sure how much memory you actually need. I used a GPU server with 32GB for training and testing.. Maybe you should try to execute it in Google...

RuntimeError: CUDA out of memory. Tried to allocate 3.06 GiB (GPU 0; 6.00 GiB total capacity; 3.58 GiB already allocated; 1024.00 MiB free; 3.59 GiB reserved in total by PyTorch)

Hi regarding the model precision of YOLO. We have not achieved a good precision, since we implemented everything manually and there was a huge class imbalance for certain classes. Nevertheless...