-
Notifications
You must be signed in to change notification settings - Fork 4
/
vars.py
134 lines (120 loc) · 3.37 KB
/
vars.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
header = """#!/bin/bash
#SBATCH --job-name=infer-{model_name}-{dataset_name}-{split}
#SBATCH --time=2:00:00
#SBATCH --partition=gpushort
#SBATCH --mem=16GB
#SBATCH --gres=gpu:v100:1
#SBATCH --output=/data/p305238/slurm_logs/%x.%j.out
module purge
module load Python/3.8.6-GCCcore-10.2.0
module load CUDA/10.2.89-GCC-8.3.0
cd /data/$ME
source venv/bin/activate
python3 --version
which python3
export HF_HOME=/data/$ME/hf_cache/
"""
command = """python3 -u /home/$ME/scripts/inference/infer.py \\
--model_name_or_path {model} \\
--use_auth_token <YOUR_AUTH_TOKEN> \\
--max_source_length {source_len} \\
--max_target_length {target_len} \\
--dataset_name it5/datasets \\
--dataset_config {config} \\
--dataset_split {split} \\
--output_dir /scratch/$ME/it5_experiments/preds/ \\
--batch_size={bs} \\
--source_column={source_column} \\
--target_column={target_column}
"""
footer = "deactivate"
settings = {
"fst_i2f": {
"source_len": 128,
"target_len": 128,
"config": "fst",
"suffix": "informal-to-formal",
"source_column": "informal",
"target_column": "formal",
"splits": ["test_0"]
},
"fst_f2i": {
"source_len": 128,
"target_len": 128,
"config": "fst",
"suffix": "formal-to-informal",
"source_column": "formal",
"target_column": "informal",
"splits": ["test_0", "test_1", "test_2", "test_3"]
},
"hg": {
"source_len": 512,
"target_len": 64,
"suffix": "headline-generation",
"source_column": "text",
"target_column": "target"
},
"ns": {
"source_len": 512,
"target_len": 128,
"suffix": "news-summarization",
"source_column": "source",
"target_column": "target",
"splits": ["test_fanpage", "test_ilpost"]
},
"qa": {
"source_len": 512,
"target_len": 64,
"suffix": "question-answering",
"source_column": "source",
"target_column": "target"
},
"qg": {
"source_len": 512,
"target_len": 128,
"suffix": "question-generation",
"source_column": "text",
"target_column": "target"
},
"st_g2r": {
"source_len": 512,
"target_len": 64,
"suffix": "ilgiornale-to-repubblica",
"source_column": "full_text",
"target_column": "headline"
},
"st_r2g": {
"source_len": 512,
"target_len": 64,
"suffix": "repubblica-to-ilgiornale",
"source_column": "full_text",
"target_column": "headline"
},
"wits": {
"source_len": 512,
"target_len": 256,
"suffix": "wiki-summarization",
"source_column": "source",
"target_column": "summary"
},
}
models = [
"it5/mt5-small",
"it5/mt5-base",
"it5/it5-small",
"it5/it5-base",
"it5/it5-large",
"it5/it5-efficient-small-el32"
]
def get_bsz(model):
if model.startswith("it5/it5-small"):
return 128
elif model.startswith("it5/it5-base") or "mt5-small" in model or model.startswith("it5/it5-efficient-small-el32"):
return 64
elif "mt5-base" in model or model.startswith("it5/it5-large"):
return 32
raise Exception(f"Unknown model: {model}")
def get_params(model, config):
params = settings[config]
params["bs"] = get_bsz(model)
return params