sample_spec.yaml

training:
  runs: 1
  baselines: ['linear', 'mlp']
  algorithms:
    - name: centralized
      cmd: >
        python simple.py --baseline_type {baseline_type}
        --discount {discount}
        --gae_lambda {gae_lambda}
        --rectangle {rectangle}
        --n_evaders {n_evaders}
        --n_pursuers {n_pursuers}
        --obs_range {obs_range}
        --n_catch {n_catch}
        --log {log}
        --no-debug
    - name: decentralized
      cmd: >
        python simple.py --baseline_type {baseline_type}
        --control decentralized
        --discount {discount}
        --gae_lambda {gae_lambda}
        --rectangle {rectangle}
        --n_evaders {n_evaders}
        --n_pursuers {n_pursuers}
        --obs_range {obs_range}
        --n_catch {n_catch}
        --log {log}
        --no-debug
        
rectangles: ['10,10', '16,16', '32,32']
n_evaders: [5, 10, 15]
n_pursuers: [2, 5, 10]
obs_ranges: [3, 5, 7, 9]
n_catches: [2, 3]
discounts: [0.99, 0.95]
gae_lambdas: [1, 0.99, 0.97]

        
options:
  storagedir: /tmp/spec/
  checkpt_subdir: checkpoints  
  n_workers: 4