Log In Sign Up
Models:
Mathew Swanson
Mathew-S/
Master
0
Downloads: 1
[8eeb5a]: / experiments / train_augmented_pipeline.py

Download this file

52 lines (48 with data), 1.8 kB 

 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
from training.consistency_trainers import *

from training.inductivenet_trainers import InductiveNetConsistencyTrainer, InductiveNetEnsembleTrainer

import sys



if __name__ == '__main__':

    id = sys.argv[1]

    model = sys.argv[2]

    config = {"model": model,

              "device": "cuda",

              "lr": 0.00001,

              "batch_size": 8,

              "epochs": 250,

              "use_inpainter": False}

    trainer = ConsistencyTrainer(id, config)

    trainer.train()



    # config = {"model": "FPN",

    #           "device": "cuda",

    #           "lr": 0.00001,

    #           "batch_size": 8,

    #           "epochs": 250,

    #           "use_inpainter": False}

    # trainer = ConsistencyTrainer(id=f"{sys.argv[1]}", config=config)

    # trainer.train()

    # config = {"model": "Unet",

    #           "device": "cuda",

    #           "lr": 0.00001,

    #           "batch_size": 8,

    #           "epochs": 250,

    #           "use_inpainter": False}

    # trainer = StrictConsistencyTrainer(id=f"dual_jaccard-{sys.argv[1]}", config=config)

    # trainer.train()



    # trainer = ConsistencyTrainerUsingAugmentation(id=f"augmentation-{sys.argv[1]}", config=config)

    # trainer.train()

    # config = {"model": "Unet",

    #           "device": "cuda",

    #           "lr": 0.00001,

    #           "batch_size": 8,

    #           "epochs": 250,

    #           "use_inpainter": False}

    # trainer = ConsistencyTrainerUsingControlledAugmentation("aug_test", config)

    # trainer.train()

    # config = {"model": "Unet",

    #           "device": "cuda",

    #           "lr": 0.00001,

    #           "batch_size": 8,

    #           "epochs": 250,

    #           "use_inpainter": False}

    # trainer = AdversarialConsistencyTrainer(id=f"adversarial-{sys.argv[1]}", config=config)

    # trainer.train()