CUDA runtime error on Google Colab

[NandaKishoreJoshi]

I tried running the below test command on google colab

python train.py -task abs -mode test -test_from ../models/model_step_148000.pt -batch_size 3000 -test_batch_size 500 -bert_data_path ../bert_data/ -log_file ../logs/val_abs_bert_cnndm -model_path ../models/ -sep_optim true -use_interval true -visible_gpus 1 -max_pos 512 -max_length 200 -alpha 0.95 -min_length 50 -result_path ../logs/abs_bert_cnndm

This gives the error saying that CUDA device is not found. I tried to debug the train.py and found that the code assigns device_id=0 for visible_gus=0,1,2,3. Should we change something here to make it work? This issue is same as closed issue #140 but there is no resolution mentioned. Any help would be much appreciated

[SebastianVeile]

Late reply. But here is the solution -visible_gpus = 0

[congdoanit98]

I tried running the below test command on google colab

python train.py -task abs -mode test -test_from ../models/model_step_148000.pt -batch_size 3000 -test_batch_size 500 -bert_data_path ../bert_data/ -log_file ../logs/val_abs_bert_cnndm -model_path ../models/ -sep_optim true -use_interval true -visible_gpus 1 -max_pos 512 -max_length 200 -alpha 0.95 -min_length 50 -result_path ../logs/abs_bert_cnndm

This gives the error saying that CUDA device is not found. I tried to debug the train.py and found that the code assigns device_id=0 for visible_gus=0,1,2,3. Should we change something here to make it work? This issue is same as closed issue #140 but there is no resolution mentioned. Any help would be much appreciated

Hello NandaKishoreJoshi, I got an issue when I tried put code on Google Colab. How did you handle this issue?