ahalev commited on
Commit
c5463a7
1 Parent(s): cdbe71e

Upload config.yaml with huggingface_hub

Browse files
Files changed (1) hide show
  1. config.yaml +161 -0
config.yaml ADDED
@@ -0,0 +1,161 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ algo:
2
+ ddpg:
3
+ params:
4
+ target_update_tau: 0.01
5
+ policy:
6
+ exploration:
7
+ sigma: 0.3
8
+ theta: 0.15
9
+ deterministic_params:
10
+ buffer_batch_size: 32
11
+ min_buffer_size: 10000
12
+ n_train_steps: 500
13
+ qf_lr: 0.0001
14
+ steps_per_epoch: 1
15
+ dqn:
16
+ params:
17
+ clip_gradient: 10
18
+ deterministic_eval: true
19
+ double_q: false
20
+ target_update_freq: 2
21
+ policy:
22
+ exploration:
23
+ decay_ratio: 0.5
24
+ max_epsilon: 1.0
25
+ min_epsilon: 0.05
26
+ general_params:
27
+ discount: 0.99
28
+ package: garage
29
+ policy:
30
+ hidden_sizes:
31
+ - 128
32
+ - 128
33
+ pretrained_policy: null
34
+ ppo:
35
+ params:
36
+ center_adv: false
37
+ tanhnormal: false
38
+ pretrain:
39
+ additional_config: null
40
+ algo_to_pretrain: null
41
+ params:
42
+ episodes_per_batch: 10
43
+ loss: log_prob
44
+ policy_lr: 0.01
45
+ pretrain_algo: rbc
46
+ replay_buffer:
47
+ buffer_size: 200000
48
+ rnd:
49
+ batch_size: 64
50
+ bound_reward_weight: cosine
51
+ bound_reward_weight_initial_ratio: 0.999999
52
+ bound_reward_weight_transient_epochs: 10
53
+ hidden_sizes:
54
+ - 64
55
+ - 64
56
+ intrinsic_reward_weight: 0.0001
57
+ n_train_steps: 32
58
+ output_dim: 128
59
+ predictor_lr: 0.001
60
+ standardize_extrinsic_reward: true
61
+ standardize_intrinsic_reward: true
62
+ sampler:
63
+ n_workers: 16
64
+ type: ray
65
+ train:
66
+ batch_size: 50000
67
+ n_epochs: 100
68
+ steps_per_epoch: 32
69
+ type: ppo
70
+ context:
71
+ disable_logging: false
72
+ experiment_name: null
73
+ log_dir:
74
+ from_keys:
75
+ - microgrid.config.scenario
76
+ - microgrid.methods.set_forecaster.forecaster
77
+ - microgrid.methods.set_module_attrs.battery_transition_model
78
+ - context.seed
79
+ - env.domain_randomization.noise_std
80
+ - algo.ppo.tanhnormal
81
+ - algo.rnd.intrinsic_reward_weight
82
+ parent: /home/ahalev/data/GridRL/paper_experiments
83
+ use_existing_dir: false
84
+ seed: 42
85
+ snapshot_gap: 10
86
+ verbose: 0
87
+ wandb:
88
+ api_key_file: ../../local/wandb_api_key.txt
89
+ group: null
90
+ log_density: 1
91
+ plot_baseline:
92
+ - mpc
93
+ - rbc
94
+ username: ahalev
95
+ env:
96
+ cls: DiscreteMicrogridEnv
97
+ domain_randomization:
98
+ noise_std: 0.01
99
+ relative_noise: true
100
+ forced_genset: null
101
+ net_load:
102
+ slack_module: genset
103
+ use: true
104
+ observation_keys:
105
+ - soc
106
+ - net_load
107
+ - import_price_current
108
+ - import_price_forecast_0
109
+ - import_price_forecast_1
110
+ - import_price_forecast_2
111
+ - import_price_forecast_3
112
+ - import_price_forecast_4
113
+ - import_price_forecast_5
114
+ - import_price_forecast_6
115
+ - import_price_forecast_7
116
+ - import_price_forecast_8
117
+ - import_price_forecast_9
118
+ - import_price_forecast_10
119
+ - import_price_forecast_11
120
+ - import_price_forecast_12
121
+ - import_price_forecast_13
122
+ - import_price_forecast_14
123
+ - import_price_forecast_15
124
+ - import_price_forecast_16
125
+ - import_price_forecast_17
126
+ - import_price_forecast_18
127
+ - import_price_forecast_19
128
+ - import_price_forecast_20
129
+ - import_price_forecast_21
130
+ - import_price_forecast_22
131
+ microgrid:
132
+ attributes:
133
+ reward_shaping_func: !BaselineShaper
134
+ baseline_module: false
135
+ module:
136
+ - genset
137
+ - 0
138
+ config:
139
+ scenario: 1
140
+ methods:
141
+ set_forecaster:
142
+ forecast_horizon: 23
143
+ forecaster: 0.0
144
+ forecaster_increase_uncertainty: true
145
+ forecaster_relative_noise: true
146
+ set_module_attrs:
147
+ battery_transition_model: null
148
+ normalized_action_bounds:
149
+ - 0.0
150
+ - 1.0
151
+ trajectory:
152
+ evaluate:
153
+ final_step: -1
154
+ initial_step: 5840
155
+ trajectory_func: null
156
+ train:
157
+ final_step: 5840
158
+ initial_step: 0
159
+ trajectory_func: !FixedLengthStochasticTrajectory
160
+ trajectory_length: 720
161
+ verbose: 1