Use the same configs across tasks

As you begin launching many Mephisto tasks, you may find that there are some specific argument options that you frequently use across multiple tasks. Mephisto provides a way to reuse these types of configurations with profiles.

Setting up profiles is pretty easy, and makes it so that you don't have to be writing architect and crowdprovider args on every launch:

python run_task.py mephisto/architect=heroku mephisto/provider=mturk_sandbox mephisto.provider.requester_name=MY_REQUESTER

Instead you can move these common configurations into a file in your ~/.mephisto/hydra_configs/profile dir.

With this, you can do something like:

# profile/local_testing.yaml
# @package _global_
mephisto:
  architect:
    _architect_type: local
    port: 1234
  provider:
    _provider_type: mock

# profile/internal_sharing.yaml
# @package _global_
mephisto:
  architect:
    _architect_type: heroku
    use_hobby: false
  provider:
    _provider_type: mock

# profile/prelaunch_test.yaml
# @package _global_
mephisto:
  architect:
    _architect_type: heroku
    use_hobby: false
  provider:
    _provider_type: mturk_sandbox
    requester_name: MY_REQUESTER_sandbox

# profile/live_launch.yaml
# @package _global_
mephisto:
  architect:
    _architect_type: heroku
    use_hobby: true
  provider:
    _provider_type: mturk
    requester_name: MY_REQUESTER

Then augmenting your launch configs is as easy as doing:

python run_task.py +profile=local_testing
...
python run_task.py +profile=live_launch

Using profile can be an effective way to simplify the configuration for your most common workflows.