Allennlp: How to use CPU instead of GPU? - gpu

Question

I'm running some code that works when there is GPU. But I'm trying to figure out how to run it locally with CPU. Here's the error:
2022-07-06 17:58:39,042 - INFO - allennlp.common.plugins - Plugin allennlp_models available
Traceback (most recent call last):
File "/Users/xiaoqingwan/opt/miniconda3/envs/absa/bin/allennlp", line 8, in <module>
sys.exit(run())
File "/Users/xiaoqingwan/opt/miniconda3/envs/absa/lib/python3.7/site-packages/allennlp/__main__.py", line 34, in run
main(prog="allennlp")
File "/Users/xiaoqingwan/opt/miniconda3/envs/absa/lib/python3.7/site-packages/allennlp/commands/__init__.py", line 118, in main
args.func(args)
File "/Users/xiaoqingwan/opt/miniconda3/envs/absa/lib/python3.7/site-packages/allennlp/commands/predict.py", line 205, in _predict
predictor = _get_predictor(args)
File "/Users/xiaoqingwan/opt/miniconda3/envs/absa/lib/python3.7/site-packages/allennlp/commands/predict.py", line 105, in _get_predictor
check_for_gpu(args.cuda_device)
File "/Users/xiaoqingwan/opt/miniconda3/envs/absa/lib/python3.7/site-packages/allennlp/common/checks.py", line 131, in check_for_gpu
" 'trainer.cuda_device=-1' in the json config file." + torch_gpu_error
allennlp.common.checks.ConfigurationError: **Experiment specified a GPU but none is available; if you want to run on CPU use the override 'trainer.cuda_device=-1' in the json config file.**
module 'torch.cuda' has no attribute '_check_driver'
Could you give me some guidance on what to do? Where is the config file and what is it called?
Here's the code (originally from: https://colab.research.google.com/drive/1F9zW_nVkwfwIVXTOA_juFDrlPz5TLjpK?usp=sharing):
# Use pretrained SpanModel weights for prediction
import sys
sys.path.append("aste")
from pathlib import Path
from data_utils import Data, Sentence, SplitEnum
from wrapper import SpanModel
def predict_sentence(text: str, model: SpanModel) -> Sentence:
path_in = "temp_in.txt"
path_out = "temp_out.txt"
sent = Sentence(tokens=text.split(), triples=[], pos=[], is_labeled=False, weight=1, id=1)
data = Data(root=Path(), data_split=SplitEnum.test, sentences=[sent])
data.save_to_path(path_in)
model.predict(path_in, path_out)
data = Data.load_from_full_path(path_out)
return data.sentences[0]
text = "Did not enjoy the new Windows 8 and touchscreen functions ."
model = SpanModel(save_dir="pretrained_14lap", random_seed=0)
sent = predict_sentence(text, model)

Answer 1

Try using something like:
device = torch.device("cpu")
model = SpanModel(save_dir="pretrained_14lap", random_seed=0)
model.to(device)
The config file is inside of the model.tar.gz in the pretrained_14lap directory (it is always named config.json). It also contains the param "cuda_device": 0, which may be causing your problem.