-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathrun_S3.py
198 lines (183 loc) · 6.66 KB
/
run_S3.py
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
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
import argparse
import os
from exp.exp import Exp
import mlflow
import mlflow.pytorch
import json
# from exp.exp_st_main_dist import Exp as Exp_dist
# from exp.exp_st_main import Exp as Exp_solo
import torch
import torch.distributed as dist
import torch.multiprocessing as mp
parser = argparse.ArgumentParser()
parser.add_argument("--distributed", action="store_true")
parser.add_argument("--gpu_ix", type=int, default=1)
parser.add_argument(
"--flag",
type=str,
default="test_grid",
help="need to be among train, test_star, test_grid",
)
parser.add_argument("--mlflow", action="store_true")
parser.add_argument("--backend", type=str, default="nccl")
parser.add_argument("--init_method", type=str, default="env://")
parser.add_argument("--customise_grid", action="store_true",default=True)
parser.add_argument(
"--customise_grid_config",
nargs="+",
type=float,
default=[-35.4, -32.0, 115.0, 118.4],
help="[the NWest lat, lon, degrees to the East,South ]",
)
parser.add_argument("--grid_ix", type=int, default=4)
parser.add_argument("--grid_file", type=str, default="./utils/grids.csv")
parser.add_argument("--grid_deg", type=float, default=0.1)
parser.add_argument("--start_cds", type=str, default="2022-1-1 00:00:00")
parser.add_argument("--start_dpird", type=str, default="2022-1-1 00:00:00")
parser.add_argument(
"--end", type=str, default="2023-12-31 23:45:00"
) # '2023-10-1 00:00:00'
parser.add_argument("--test_grid_start", type=str, default='2022-07-01 00:00:00')
parser.add_argument("--test_grid_end", type=str, default='2022-08-01 00:00:00')
parser.add_argument("--vars_terrain", action="store_true",default=True)
parser.add_argument("--vars_cds", nargs="+", type=str, default=["u10", "v10", "msl"])
parser.add_argument(
"--vars_dpird",
nargs="+",
type=str,
default=["wind_3m_u", "wind_3m_v", "airTemperature", "relativeHumidity"],
)
parser.add_argument(
"--labels",
nargs="+",
type=str,
default=[
"wind_10m_u",
"wind_10m_v",
], # wind_10m_u and wind_10m_v must be the first 2 items
)
parser.add_argument("--test_n_days", type=int, default=5)
parser.add_argument("--T_hr", type=int, default=48, help="sample length") # 48
parser.add_argument(
"--L_hr", type=int, default=4, help="moving window of y from x"
) # 4
parser.add_argument(
"--S_min", type=int, default=15, help="sliding window of each sample"
) # 15
parser.add_argument(
"--F_hr", type=int, default=4, help="forecasting of ecmwf, F is added on top of T"
) # 4
parser.add_argument("--filters", nargs="+", type=int, default=[4, 8, 16]) # [4, 8, 16]
parser.add_argument(
"--datasrc",
type=int,
default=2,
help="0-dpird only, 1-ecmwf only, 2-dpird and ecmwf",
)
parser.add_argument("--epochs", type=int, default=200)
parser.add_argument("--batch_size", type=int, default=64) # 32
parser.add_argument("--learning_rate", type=float, default=0.001)
parser.add_argument(
"--stations_coords",
type=str,
default="./data_prep_ST/make_grid/all_station_coordinates.csv",
)
parser.add_argument(
"--stars_coords",
type=str,
default="./data_prep_ST/make_grid/stations_available_label_coordinates.csv",
)
parser.add_argument(
"--cds_src_path", type=str, default="/mnt/science1/fchen/dataset_CDS_2019_2023/combined_p01_2019_2023.nc"
)
parser.add_argument(
"--terrain_src_path",
type=str,
default="/mnt/science1/fchen/dataset_terrain/dem-9s.tif",
)
parser.add_argument("--dpird_src_path", type=str, default="/mnt/science1/fchen/dataset_DPIRD")
parser.add_argument("--st_data_path", type=str, default="/mnt/science1/fchen/dataset_ST/")
parser.add_argument(
"--result_path",
type=str,
default="/mnt/science1/fchen/result_ST/lat32_lon115_3d4_2022_2023/",
)
parser.add_argument(
"--model_path",
type=str,
default="/mnt/science1/fchen/model_ST/",
)
args = parser.parse_args()
def main():
if args.flag == "train":
if args.distributed:
print("-" * 30, "IT IS SET IN DISTRIBUTED MODE", "-" * 30)
if args.init_method == "env://":
os.environ["MASTER_ADDR"] = "localhost"
os.environ["MASTER_PORT"] = "2323"
mp.set_start_method("spawn", force=True)
world_size = args.gpu_ix
mp.spawn(
main_worker, nprocs=args.gpu_ix, args=(world_size, args), join=True
)
else:
print("-" * 30, "NORMAL MODE", "-" * 30)
exp = Exp(args)
exp.train()
elif args.flag == "test_star":
args.distributed = False
print("-" * 30, "NORMAL MODE", "-" * 30)
exp = Exp(args)
exp.test_star()
elif args.flag == "test_grid":
args.distributed = False
print("-" * 30, "NORMAL MODE", "-" * 30)
exp = Exp(args)
exp.test_grid()
def main_worker(rank, world_size, args):
dist.init_process_group(
backend=args.backend,
init_method=args.init_method,
world_size=world_size,
rank=rank,
)
print(f"initiatialized distribution for {rank} out of {world_size}")
exp = Exp(args, rank, world_size)
exp.train()
if __name__ == "__main__":
if args.mlflow:
mlflow.set_tracking_uri("http://127.0.0.1:8080")
logname = (
f"T{args.T_hr}_L{args.L_hr}_F{args.F_hr}_S{args.S_min}_"
f'flt{args.filters[0]}_{args.filters[1]}_{args.filters[2]}_{"_".join(map(str, args.vars_dpird+args.vars_cds))}'
)
if args.flag == "train":
mlflow.set_experiment("Wind_ST")
with mlflow.start_run(run_name=logname) as run:
# Log hyperparameters
mlflow.set_tag("logname", logname)
mlflow.log_param(
"features", "_".join(map(str, args.vars_dpird + args.vars_cds))
)
mlflow.log_param("T_hr", args.T_hr)
mlflow.log_param("L_hr", args.L_hr)
mlflow.log_param("F_hr", args.F_hr)
mlflow.log_param("S_min", args.S_min)
mlflow.log_param("filters", "_".join(map(str, args.filters)))
mlflow.log_param(
"customise_grid", "_".join(map(str, args.customise_grid_config))
)
main()
mlflow.end_run()
else: # for test_star and test_grid
filter_string = f"tags.logname = '{logname}'"
df = mlflow.search_runs(
experiment_names=["Wind_ST"], filter_string=filter_string
)
args.run_id = df["run_id"].values.tolist()[0]
print(args.run_id)
with mlflow.start_run(run_name=logname) as run:
main()
mlflow.end_run()
else:
main()