Error while using floyd.yml [Solved]

[wish1104]

Hello, I am trying to run a job with floyd.yml
My floyd.yml looks like this:

	task:
	  train-lenet:
	    machine: gpu
	    env: pytorch-1.0
	    description: training LeNet for 500 steps
	    input:
	      - wish1104/datasets/emnist/1:dataset
	    command: python pytorch_models/train.py --max_steps 500 --cuda --train_features --checkpoint

When I run floyd run --task train-lenet I got the following output

Creating project run. Total upload size: 90.4KiB
Syncing code ...
[================================] 99151/99151 - 00:00:04
Error: One or more request parameters is incorrect
Invalid module: DZxvL4WLetxLNLpP4aN9Hf

I am confused with the last part which looks like a hash of some sort.
But if I manually run the equivalent command:

floyd run --gpu --env pytorch-1.0 --message 'training LeNet for 500 steps' --data wish1104/datasets/emnist/1:dataset 'python pytorch_models/train.py --max_steps 500 --cuda --train_features --checkpoint'

everything works fine.

[redeipirati]

Hi @wish1104,

Thanks for reporting this, I have just shared this with the Dev Team to investigate. Question: which floyd-cli version are you using (run floyd version to get it)?

[houqp]

Hi @wish1104,

Looks like you might have deleted and recreated a dataset with the same name, which caused this issue. We are working on a fix for this right now.

[houqp]

@wish1104 fix has been rolled out to production, could you give it another try?

[wish1104]

Yes. It's working now.