OLM/experiments/configs.py at master · DFKI-NLP/OLM · 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
# RoBERTa config for all tasks
ROBERTA_UNK_CONFIG = {
        "strategy": "unk_replacement",
        "batch_size": 128,
        "unk_token": "<unk>",
        "seed": 1111,
}

ROBERTA_DEL_CONFIG = {
        "strategy": "delete",
        "batch_size": 128,
        "seed": 1111,
}

ROBERTA_RESAMPLING_CONFIG = {
        "strategy": "bert_lm_sampling",
        "std": False,
        "cuda_device": 0,
        "bert_model": "bert-base-uncased",
        "batch_size": 128,
        "n_samples": 100,
        "verbose": False,
        "seed": 1111,
}

ROBERTA_RESAMPLING_STD_CONFIG = {
        "strategy": "bert_lm_sampling",
        "std": True,
        "cuda_device": 0,
        "bert_model": "bert-base-uncased",
        "batch_size": 128,
        "n_samples": 100,
        "verbose": False,
        "seed": 1111,
}

ROBERTA_GRADIENT_CONFIG = {
        "strategy": "gradient",
        "batch_size": 2,
        "seed": 1111,
}

# SST2 for AllenNLP
SST2_UNK_CONFIG = {
        "strategy": "unk_replacement",
        "batch_size": 128,
        "unk_token": "__unk__",
        "seed": 1111,
}

SST2_DEL_CONFIG = {
        "strategy": "delete",
        "batch_size": 128,
        "seed": 1111,
}

SST2_RESAMPLING_CONFIG = {
        "strategy": "bert_lm_sampling",
        "std": False,
        "cuda_device": 0,
        "bert_model": "bert-base-uncased",
        "batch_size": 128,
        "n_samples": 100,
        "verbose": False,
        "seed": 1111,
}

SST2_RESAMPLING_STD_CONFIG = {
        "strategy": "bert_lm_sampling",
        "std": True,
        "cuda_device": 0,
        "bert_model": "bert-base-uncased",
        "batch_size": 128,
        "n_samples": 100,
        "verbose": False,
        "seed": 1111,
}

SST2_GRADIENT_CONFIG = {
        "strategy": "gradient",
        "batch_size": 2,
        "seed": 1111,
}