Having problems while training

[Siri-2001]

Hi! When I was trying to train the models by using "task train_slt", an error occured:
"task: Use at least 1 GPU to train the model
task: precondition not met"
However,my .env file is as follow:
FAIRSEQ_ROOT: /home/csr/slt_how2sign_wicv2023-wicv23/fairseq
SAVE_DIR: /home/csr/slt_how2sign_wicv2023-wicv23/data/how2sign/i3d_features
WANDB_ENTITY: CSR
WANDB_PROJECT: name_project_WANDB
NUM_GPUS: 1
VOCAB_SIZE: 7000
FEATS: i3d
PARTITION: train
DATA_DIR: /home/csr/slt_how2sign_wicv2023-wicv23/data/how2sign/i3d_features
CONFIG_DIR: /home/csr/slt_how2sign_wicv2023-wicv23/examples/sign_language/config/wicv_cvpr23/i3d_best
NUM_GPUS has already been set to 1.But Why the error occured?Thx a lot!

[parsheeta]

I removed -gt 0 in the preconditions for NUM_GPUS and it resolved that error. I guess it doesn't make a big difference for me because I have 1 GPU anyways.

train_slt:
preconditions:
- sh: "[[ -d {{.DATA_DIR}} ]]"
msg: Specify the DATA_DIR in the .env file
- sh: "[[ {{.SAVE_DIR}} ]]"
msg: Specify the SAVE_DIR in the .env file
- sh: "[[ {{.WANDB_ENTITY}} ]]"
msg: Specify the WANDB_ENTITY in the .env file
- sh: "[[ {{.WANDB_PROJECT}} ]]"
msg: Specify the WANDB_PROJECT in the .env file
- sh: "[ {{.NUM_GPUS}} ]"
msg: Use at least 1 GPU to train the model
- sh: '[ "{{.EXPERIMENT}}" ]'
msg: Specify a name for the EXPERIMENT
- sh: '[ "{{.CONFIG_DIR}}/{{.EXPERIMENT}}.yaml" ]'
msg: The yaml does not exist