configs.yaml

# train
# python train.py --algos=dtqn_noisy --self_play_episode_num=4 --load_ckpt=True --save_ckpt=True  --data_folder="train_data_/benchmark_reduced" --wandbName="NoPer" --hid_layer=3 --emb_dim=64 --context_len=5 --early_stop=False  --enable_wandb=True
# eval
# python train.py --algos=dtqn_noisy --self_play_episode_num=150 --result_dir=solutionsDRL --load_ckpt=True --save_ckpt=False  --data_folder="test_data_/benchmark_reduced" --wandbName="dtqn_noisy_pretain_test" --hid_layer=3 --emb_dim=64 --context_len=5
default: 
  all:
    algos:         None
    wandbName:     None
    hid_layer:  &lay 3
    emb_dim:    &dim   64      #  128 is a little too large...
    context_len: &ctx_l  5
    early_stop: False
    result_dir: solutionsDRL
  train:
    mode: train
    load_ckpt: True
    save_ckpt: True
    self_play_episode_num: 4
    run_benchmark_num: 30
    enable_wandb: True
    data_folder:  train_data_/benchmark_reduced
  eval:
    mode: eval
    load_ckpt: True
    save_ckpt: False
    self_play_episode_num: 150
    enable_wandb: True
    data_folder:  test_data_/benchmark_reduced

# custom:
#   dtqn:
#     algos