cannot use the GPU

[tyistyler]

Excuse me, thanks for your great codes for joint model!!
when I use the code with tf=2.0 and keras=2.3.1, I canot use the GPU in my machine.
Could you please offer some adivces~
"
import os
os.system("CUDA_VISIBLE_DEVICES=1 python run.py --do_train --model_name NYT --rel_path data/NYT/rel2id.json --train_path data/NYT/train_triples.json "
"--dev_path data/NYT/test_triples.json --bert_dir pretrained-bert/cased_L-12_H-768_A-12 --save_path ckpts/nyt.model "
"--learning_rate 0.00005 --neg_samples 2 --epoch 200 --verbose 2 ")
"
thanks

[SeanLee97]

Hi @tyistyler, thanks for following our project. We haven't tried to run the code in tensorflow>=2.0. We recommend running the code in tensorflow-gpu>1.14,tensorflow-gpu<2.0 .

Our dependencies about tensorflow are as follows:

$conda list | grep "tensorflow"
tensorflow                1.15.0          gpu_py37h0f0df58_0
tensorflow-base           1.15.0          gpu_py37h9dcbed7_0
tensorflow-estimator      1.15.1             pyh2649769_0
tensorflow-gpu            1.15.0               h0d30ee6_0

[SeanLee97]

Oh, I see that you run the code via python os.system. You could try to add a new line os.environ['CUDA_VISIBLE_DEVICES']= '1' before os.system, i.e, you can refactor the code as follows:

import os

os.environ['CUDA_VISIBLE_DEVICES']= '1'

os.system("python run.py --do_train --model_name NYT --rel_path data/NYT/rel2id.json --train_path data/NYT/train_triples.json "
"--dev_path data/NYT/test_triples.json --bert_dir pretrained-bert/cased_L-12_H-768_A-12 --save_path ckpts/nyt.model "
"--learning_rate 0.00005 --neg_samples 2 --epoch 200 --verbose 2 ")

[SeanLee97]

Oh, I see that you run the code via python os.system. You could try to add a new line os.environ['CUDA_VISIBLE_DEVICES']= '1' before os.system, i.e, you can refactor the code as follows:

import os

os.environ['CUDA_VISIBLE_DEVICES']= '1'

os.system("python run.py --do_train --model_name NYT --rel_path data/NYT/rel2id.json --train_path data/NYT/train_triples.json "
"--dev_path data/NYT/test_triples.json --bert_dir pretrained-bert/cased_L-12_H-768_A-12 --save_path ckpts/nyt.model "
"--learning_rate 0.00005 --neg_samples 2 --epoch 200 --verbose 2 ")

@tyistyler
Because we didn't test the code in TensorFlow2.x, we can not promise the code can successfully run in TensorFlow2.x. Look forward to your results in TF2.x!

[tyistyler]

Thanks for your reply, I solve it successfully (the reason is that there has some error of tensorflow's install). I use "conda install tensorflow==1.15.0" solve it.
Thanks!!!