How to use gpu in model tuning?

[smallcardum]

I am following the constructions in the Tutorial 8: Model Tuning However, when I follow the codes in the page, I find that I am still using CPU to do the work: 2022-07-07 18:26:55,244 Corpus: "Corpus: 339 train + 101 dev + 129 test sentences" 2022-07-07 18:26:55,244 ---------------------------------------------------------------------------------------------------- 2022-07-07 18:26:55,244 Parameters: 2022-07-07 18:26:55,244 - learning_rate: "0.200000" 2022-07-07 18:26:55,244 - mini_batch_size: "32" 2022-07-07 18:26:55,244 - patience: "3" 2022-07-07 18:26:55,244 - anneal_factor: "0.5" 2022-07-07 18:26:55,244 - max_epochs: "50" 2022-07-07 18:26:55,244 - shuffle: "True" 2022-07-07 18:26:55,244 - train_with_dev: "False" 2022-07-07 18:26:55,244 - batch_growth_annealing: "False" 2022-07-07 18:26:55,244 ---------------------------------------------------------------------------------------------------- 2022-07-07 18:26:55,244 Model training base path: "resources\results" 2022-07-07 18:26:55,244 ---------------------------------------------------------------------------------------------------- 2022-07-07 18:26:55,244 Device: cpu 2022-07-07 18:26:55,244 ---------------------------------------------------------------------------------------------------- 2022-07-07 18:26:55,244 Embeddings storage mode: cpu

I've tried the embeddings_storage_mode='gpu' in Tutorial 7: Training a Model, but I couldn't find a place to set the code in.

It would be appreciated if someone would offer help.

[agademic]

As far as I know, Flair will use a GPU automatically if available, so there is no need to specify anything in the code. Do you have CUDA installed? If so can you run torch.cuda.is_available() and check the result? Or better run nvidia-smi in your terminal. Here you can see if CUDA is installed and the current usage of your GPU.

[stale[bot]]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.