Q-Mamba/config.py at main · MetaEvo/Q-Mamba · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
import os
import time
import argparse


def get_options(args=None):

    parser = argparse.ArgumentParser(description="")

    parser.add_argument('--train',default=None,action='store_true',help='switch to train mode')
    parser.add_argument('--test',default=None,action='store_true', help='switch to inference mode')
    parser.add_argument('--single_task', type=int, default=None, help='switch to inference mode')
    parser.add_argument('--problem', default ='bbob', choices=['bbob', 'easy', 'protein', 'all'])  # for test only
    parser.add_argument('--task', default ='usual', choices=['usual', 'real'])  # for test only
    parser.add_argument('--task_class', default ='DE')
    # parser.add_argument('--data_path', default ='multitask_dataset.pkl', help='path to load model parameters and optimizer state from')
    parser.add_argument('--epoch_start', type=int, default=0, help='start at epoch # (relevant for learning rate decay)')
    parser.add_argument('--epoch_end', type=int, default=100, help='maximum training epoch')
    parser.add_argument('--weighted_start', type=int, default=5, help='start at epoch # (relevant for learning rate decay)')
    parser.add_argument('--device', default='cpu')
    parser.add_argument('--encoder', default='attn', choices=['attn', 'mlp'])
    parser.add_argument('--positional', default='sin', help="normalization type, 'learnt' (default) or 'sin' or None")
    parser.add_argument('--morphological', default=True, type=bool,)
    parser.add_argument('--aligned_pe', default=False, type=bool,)
    parser.add_argument('--sep_state', default=True, type=bool,)
    parser.add_argument('--shuffle', default=False, type=bool,)
    parser.add_argument('--dim', default=None, type=int,)
    parser.add_argument('--skip_step', default=1, type=int,)
    parser.add_argument('--softmax', default=False, type=bool,)

    parser.add_argument('--smac_gen', type=int, default=10)
    parser.add_argument('--smac_acf', default=None, choices=['EI', 'LCB', 'PI', 'TS'])
    parser.add_argument('--smac_acm', default=None, choices=['LocalAndSortedRandomSearch', 'DifferentialEvolution', 'LocalSearch', 'RandomSearch'])

    # Overall settingss
    parser.add_argument('--seed', type=int, default=1, help='random seed to use')
    parser.add_argument('--dataseed', type=int, default=14, help='random seed for dataset generation 28 for trainsize32, 2024 for 64')
    # parser.add_argument('--dataseed', type=int, default=2027, help='random seed for dataset generation')
    # parser.add_argument('--dataseed', type=int, default=2026, help='random seed for dataset generation')
    parser.add_argument('--testseed', type=int, default=2025, help='random seed for dataset generation')
    parser.add_argument('--repeat', type=int, default=1)
    parser.add_argument('--store_path', default='outputs/')
    parser.add_argument('--trainsize', type=int, default=32)
    parser.add_argument('--testsize', type=int, default=32)
    # parser.add_argument('--trainsize', default=8)
    # parser.add_argument('--testsize', default=8)
    parser.add_argument('--batch_size', type=int, default=32,help='number of instances per batch during training')
    parser.add_argument('--test_batch_size', type=int, default=64,help='number of instances per batch during training')
    parser.add_argument('--update_best_model_epochs',type=int,default=1,help='update the best model every n epoch')
    parser.add_argument('--checkpoint_epochs', type=int, default=1, help='save checkpoint every n epochs (default 1), 0 to save no checkpoints')
    parser.add_argument('--pool_path', default = None, help='run name %Y%m%dT%H%M%S')

    # Testing settings -
    parser.add_argument('--load_path', default = 'outputs/', help='path to load model parameters and optimizer state from')
    parser.add_argument('--load_name', default = None, help='run name %Y%m%dT%H%M%S')
    parser.add_argument('--load_epoch', default = None, help='epoch id')
    parser.add_argument('--smac_load', default = None, help='smac load')
    parser.add_argument('--train_task', default = None,action='store_true', )
    parser.add_argument('--test_task', default = None,action='store_true', )
    # parser.add_argument('--test_seed', type=int, default=2024, help='random seed for dataset generation')

    # Task settings
    parser.add_argument('--Npop', default=[1, 2, 3, 4])
    parser.add_argument('--Npop_prob', default=[0.4, 0.2, 0.2, 0.2])
    parser.add_argument('--opm', default=[1, 2, 3])
    parser.add_argument('--NPmax', default=[50, 100, 200])
    parser.add_argument('--NPmin', default=[4, 10, 20])
    parser.add_argument('--NA', default=[1, 2, 3])
    parser.add_argument('--Xmax', default=5)
    parser.add_argument('--Vmax', default=[0.2, 0.3, 0.5])
    parser.add_argument('--regroup', default=[0,  ])
    # parser.add_argument('--comm_inter', default=[0, 5, 10])
    parser.add_argument('--maxAct', default=12)
    parser.add_argument('--maxCom', default=16)
    parser.add_argument('--actSpace', default=10)
    # parser.add_argument('--MaxFEs', default=50000)
    parser.add_argument('--MaxGen', default=500)

    # agent settings
    parser.add_argument('--encoder_head_num', type=int, default=4, help='head number of encoder')
    parser.add_argument('--critic_head_num', type=int, default=4, help='head number of critic encoder')
    parser.add_argument('--embedding_dim', type=int, default=16, help='dimension of input embeddings')
    parser.add_argument('--decoder_hidden_dim', type=int, default=32, help='head number of decoder')
    parser.add_argument('--hidden_dim', type=int, default=16, help='dimension of hidden layers in Enc/Dec')
    parser.add_argument('--hidden_dim1_critic',default=32,help='the first hidden layer dimension for critic')
    parser.add_argument('--hidden_dim2_critic',default=16,help='the second hidden layer dimension for critic')
    parser.add_argument('--n_encode_layers', type=int, default=3, help='number of stacked layers in the encoder')
    parser.add_argument('--node_dim',default=9,type=int,help='feature dimension for backbone algorithm')
    parser.add_argument('--op_dim',default=24,type=int,help='feature dimension for backbone algorithm')
    # parser.add_argument('--llm_hidden_dim',default=128,type=int,help='feature dimension for backbone algorithm')
    parser.add_argument('--op_embed_dim',default=16,type=int,help='feature dimension for backbone algorithm')
    parser.add_argument('--normalization', default='layer', help="normalization type, 'layer' (default) or 'batch'")

    parser.add_argument('--lr_model', type=float, default=1e-3, help="learning rate for the actor network")
    parser.add_argument('--lr_decay', type=float, default=1., help='learning rate decay per epoch',)
    parser.add_argument('--max_learning_step',default=4000000,help='the maximum learning step for training')

    parser.add_argument('--gamma', type=float, default=0.999, help='reward discount factor for future rewards')
    parser.add_argument('--T_train', type=int, default=1800, help='number of itrations for training')
    parser.add_argument('--n_step', type=int, default=10, help='n_step for return estimation')
    parser.add_argument('--K_epochs', type=int, default=3, help='mini PPO epoch')
    parser.add_argument('--eps_clip', type=float, default=0.1, help='PPO clip ratio')
    parser.add_argument('--max_grad_norm', type=float, default=0.1, help='maximum L2 norm for gradient clipping')
    parser.add_argument('--max_sigma',default=0.7,type=float,help='upper bound for actor output sigma')
    parser.add_argument('--min_sigma',default=0.01,type=float,help='lowwer bound for actor output sigma')
    parser.add_argument('--show_figs', action='store_true', help='enable figure logging')

    parser.add_argument('--log_dir', default='logs', help='directory to write TensorBoard information to')
    parser.add_argument('--log_step', type=int, default=50, help='log info every log_step gradient steps')
    parser.add_argument('--save_dir', default='outputs', help='directory to write output models to')

    opts = parser.parse_args(args)
    opts.run_name = time.strftime("%Y%m%dT%H%M%S")
    # opts.run_name = "{}_{}".format(opts.run_name, opts.run_time)

    return opts