finetune.py: proble with local_rank

[martina811]

Hello everyone,
I am facing a problem with the script finetune.py, when I launch the command from the tutorial python -m torch.distributed.launch finetune.py --data_path DATA/Zheng68K.h5ad --model_path DATA/panglao_pretrain.pth , I get the following error:

If your script expects --local-rank argument to be set, please
change it to read from os.environ['LOCAL_RANK'] instead. See
https://pytorch.org/docs/stable/distributed.html#launch-utility for
further instructions

warnings.warn(
usage: finetune.py [-h] [--local_rank LOCAL_RANK] [--bin_num BIN_NUM] [--gene_num GENE_NUM] [--epoch EPOCH] [--seed SEED] [--batch_size BATCH_SIZE]
[--learning_rate LEARNING_RATE] [--grad_acc GRAD_ACC] [--valid_every VALID_EVERY] [--pos_embed POS_EMBED] [--data_path DATA_PATH]
[--model_path MODEL_PATH] [--ckpt_dir CKPT_DIR] [--model_name MODEL_NAME]
finetune.py: error: unrecognized arguments: --local-rank=0

Is there anybody else that faced this problem and fixed it?

Many thanks

[Jiam1ng]

I tried to add the --use_env argument when running the script, and the local-rank error could be solved. For your command, it should be:
python -m torch.distributed.launch --use_env finetune.py --data_path DATA/Zheng68K.h5ad --model_path DATA/panglao_pretrain.pth

In my situation, --local_rank 0 should also be added to the last of the command to run the script.