Closed adarshpalaskar1 closed 1 year ago
Resolved:
The code uses the default path /datases01 even for other datasets. Setting the --data-path
to a new directory solved the issue. The following command can be used:
python -m torch.distributed.launch --use_env main.py --model mini_deit_small_patch16_224 --batch-size 128 --data-path ./cifar100 --data-set CIFAR100 --output_dir ./outputs --teacher-model regnety_160 --distillation-type soft --distillation-alpha 1.0 --drop-path 0.0
Thanks for pointing it out!
Hello, I was trying to reproduce the results of Mini-DeiT on CIFAR 100 dataset by running:
python main.py --model mini_deit_tiny_patch16_224 --batch-size 128 --data-set CIFAR100 --output_dir ./outputs --teacher-model regnety_160 --distillation-type soft --distillation-alpha 1.0 --drop-path 0.0
I do not have multiple GPUs and therefore not using
rpe_ops
module and have modified the script. The original script given in the README file is :python -m torch.distributed.launch --nproc_per_node=8 --use_env main.py --model mini_deit_tiny_patch16_224 --batch-size 128 --data-path ./ImageNet --output_dir ./outputs --teacher-model regnety_160 --distillation-type soft --distillation-alpha 1.0 --drop-path 0.0
I am getting the following error:
File "<frozen os>", line 215, in makedirs File "<frozen os>", line 215, in makedirs File "<frozen os>", line 225, in makedirs PermissionError: [Errno 13] Permission denied: '/datasets01'
Can anyone help me fix this issue?