1. Command Line Arguments

1.1. Usage

noodles <command_type> <path_to_spec> [optional_arguments]

1.2. Positional Arguments

<command_type>

The command type.

The command type is a user-defined key under the following items in the spec file:

  1. <top_level>.depends_on

  2. <top_level>.requirements

  3. <top_level>.commands

  4. <top_level>.write_outputs

Where <top_level> can be one of the following top level keys in the spec:

  1. experiment_default

  2. before_all_experiments[i]

  3. experiments[i]

  4. after_all_experiments[i]

i can be any index.

Example

The following code specifies the command type lets_do_it:

experiments:
- name: run_script.exp1
  description: Run the script after uploading the script to the remote server
  envs:
    SCRIPT: exp1.sh
  depends_on:
    lets_do_it:
    - upload.exp1
  requirements:
    lets_do_it:
    - memory_usage: "<=0.5"
  commands:
    lets_do_it:
    - "cd $HOME"
    - "bash $SCRIPT"
  write_outputs:
    lets_do_it:
      stdout_to: $LOCAL_DIR/exp1.stdout.log
      stderr_to: $LOCAL_DIR/exp1.stderr.log
<path_to_spec>

Path to the spec file.

Example

specs/train.yml

1.3. Optional Arguments

-h, --help

Show the help message and exit.

-v, --verbose

Print out verbose messages.

If this option is used, verbose messages regarding to what Noodles is doing will be printed.

-d, --debug

Print out debug messages.

If this option is used, the following things will be printed:

  1. Command line arguments used

  2. Original user spec

  3. Processed user spec

  4. Creation of temporary files

  5. Commands written to temporary files

  6. The commands to be run by Noodles

  7. Environment variables added by Noodles

-s, --silent

Silence all logging messages.