search

applicaai / poleval-2018

Code and data accompanying the paper "Approaching nested named entity recognition with parallel LSTM-CRFs."
MIT License
26 stars 6 forks source link

CUDA memory requirements. Recommendations for GPU. #6

Open Slonecznik opened 4 years ago

Slonecznik commented 4 years ago

Hey. Congratulations on wining Task2 of PolEval2018. I've been trying to run your repository on a server with NVIDIA 1080 GTX (having CUDA 10.0) which has 8GB of memory. Unfortunately, it seems it wasn't enough, because training process crashed with a message: RuntimeError: cuda runtime error (2) : out of memory. Could you tell me what would be your recommendations towards preferred CUDA memory or graphic card model so I could retrieve trained models out of make models command? I'd be grateful. Thanks.