Cannot allocate memory while using eval_config.py
Amitcuhp opened this issue · comments
Amitcuhp commented
torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 4.30 GiB (GPU 0; 31.75 GiB total capacity; 8.97 GiB already allocated; 414.69 MiB free; 11.45 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
Amitcuhp commented
!python3 /home/Raman/mace/scripts/eval_configs.py
--configs="train.xyz"
--model="MACE_model_swa.model"
--output="mace_train.xyz"
If anyone has any solution, Please help !!
Amitcuhp commented
Solved it by reducing batch size