neptune-ai / open-solution-mapping-challenge

Open solution to the Mapping Challenge :earth_americas:
https://www.crowdai.org/challenges/mapping-challenge
MIT License
380 stars 96 forks source link

How to use --dev_mode? #190

Closed carbonox-infernox closed 5 years ago

carbonox-infernox commented 6 years ago

I have tried several different ways to use the argument --dev_mode: (in this case with -- train)

-d

neptune: Error: Invalid parameter '-d'. Parameter names must begin with double dash.

so then i tried the double dash version:

--dev_mode

neptune: Error: No value provided for parameter 'dev_mode'

I guess it needs a parameter? I tried:

--dev_mode True --dev_mode true --dev_mode 1 --dev_mode on

Error: Got unexpected extra argument ('option i tried')

So does it expect a parameter or not?

I've also tried moving around the argument order without luck. How do I get this to work?

kamil-kaczmarek commented 6 years ago

Hi @carbonox-infernox, as you can see in the main.py --dev_mode is a flag, and you have to add it after main.py file, not before (in your command).

It is main.py argument, not neptune argument

carbonox-infernox commented 6 years ago

@kamil-kaczmarek

That's exactly what I did, and it told me:

neptune: Error: No value provided for parameter 'dev_mode'

kamil-kaczmarek commented 6 years ago

@carbonox-infernox, can you paste entire command here?

carbonox-infernox commented 6 years ago

@kamil-kaczmarek

CUDA_VISIBLE_DEVICES=0 python main.py -- train --dev_mode --pipeline_name unet_weighted

also tried

CUDA_VISIBLE_DEVICES=0 python main.py -- train --pipeline_name unet_weighted --dev_mode

carbonox-infernox commented 5 years ago

I ended up just doing it manually by adding:

meta_train = meta_train.sample(, random_state=seed)

in pipeline_manager.py