Upload down_on_copy.py with huggingface_hub
Browse files- down_on_copy.py +67 -0
down_on_copy.py
ADDED
|
@@ -0,0 +1,67 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import datasets
|
| 2 |
+
import os
|
| 3 |
+
import shutil
|
| 4 |
+
#from datasets import load_dataset
|
| 5 |
+
train0, validation0, test0 = datasets.load_dataset("superb", "ks", split=["train","validation","test"])
|
| 6 |
+
|
| 7 |
+
labels = train0.features["label"].names
|
| 8 |
+
label2id = {x: labels.index(x) for x in labels}
|
| 9 |
+
# id2label = {str(id): label for label, id in label2id.items()}
|
| 10 |
+
|
| 11 |
+
down_id = label2id['down']
|
| 12 |
+
on_id = label2id['on']
|
| 13 |
+
|
| 14 |
+
train0_down = train0.filter(lambda example: example['label'] == down_id)
|
| 15 |
+
train0_on = train0.filter(lambda example: example['label'] == on_id)
|
| 16 |
+
|
| 17 |
+
shutil.rmtree('/home/mr249/ac_h/down_on/data/train/down',ignore_errors=True)
|
| 18 |
+
os.mkdir('/home/mr249/ac_h/down_on/data/train/down')
|
| 19 |
+
shutil.rmtree('/home/mr249/ac_h/down_on/data/train/on',ignore_errors=True)
|
| 20 |
+
os.mkdir('/home/mr249/ac_h/down_on/data/train/on')
|
| 21 |
+
|
| 22 |
+
for e in train0_down:
|
| 23 |
+
p = e['audio']['path']
|
| 24 |
+
newpath = '/home/mr249/ac_h/down_on/data/train/down/{}'.format(p.split('/')[len(p.split('/')) - 1])
|
| 25 |
+
os.link(p,newpath)
|
| 26 |
+
|
| 27 |
+
for e in train0_on:
|
| 28 |
+
p = e['audio']['path']
|
| 29 |
+
newpath = '/home/mr249/ac_h/down_on/data/train/on/{}'.format(p.split('/')[len(p.split('/')) - 1])
|
| 30 |
+
os.link(p,newpath)
|
| 31 |
+
|
| 32 |
+
validation0_down = validation0.filter(lambda example: example['label'] == down_id)
|
| 33 |
+
validation0_on = validation0.filter(lambda example: example['label'] == on_id)
|
| 34 |
+
|
| 35 |
+
shutil.rmtree('/home/mr249/ac_h/down_on/data/validation/down',ignore_errors=True)
|
| 36 |
+
os.mkdir('/home/mr249/ac_h/down_on/data/validation/down')
|
| 37 |
+
shutil.rmtree('/home/mr249/ac_h/down_on/data/validation/on',ignore_errors=True)
|
| 38 |
+
os.mkdir('/home/mr249/ac_h/down_on/data/validation/on')
|
| 39 |
+
|
| 40 |
+
for e in validation0_down:
|
| 41 |
+
p = e['audio']['path']
|
| 42 |
+
newpath = '/home/mr249/ac_h/down_on/data/validation/down/{}'.format(p.split('/')[len(p.split('/')) - 1])
|
| 43 |
+
os.link(p,newpath)
|
| 44 |
+
|
| 45 |
+
for e in validation0_on:
|
| 46 |
+
p = e['audio']['path']
|
| 47 |
+
newpath = '/home/mr249/ac_h/down_on/data/validation/on/{}'.format(p.split('/')[len(p.split('/')) - 1])
|
| 48 |
+
os.link(p,newpath)
|
| 49 |
+
|
| 50 |
+
test0_down = test0.filter(lambda example: example['label'] == down_id)
|
| 51 |
+
test0_on = test0.filter(lambda example: example['label'] == on_id)
|
| 52 |
+
|
| 53 |
+
shutil.rmtree('/home/mr249/ac_h/down_on/data/test/down',ignore_errors=True)
|
| 54 |
+
os.mkdir('/home/mr249/ac_h/down_on/data/test/down')
|
| 55 |
+
shutil.rmtree('/home/mr249/ac_h/down_on/data/test/on',ignore_errors=True)
|
| 56 |
+
os.mkdir('/home/mr249/ac_h/down_on/data/test/on')
|
| 57 |
+
|
| 58 |
+
for e in test0_down:
|
| 59 |
+
p = e['audio']['path']
|
| 60 |
+
newpath = '/home/mr249/ac_h/down_on/data/test/down/{}'.format(p.split('/')[len(p.split('/')) - 1])
|
| 61 |
+
os.link(p,newpath)
|
| 62 |
+
|
| 63 |
+
for e in test0_on:
|
| 64 |
+
p = e['audio']['path']
|
| 65 |
+
newpath = '/home/mr249/ac_h/down_on/data/test/on/{}'.format(p.split('/')[len(p.split('/')) - 1])
|
| 66 |
+
os.link(p,newpath)
|
| 67 |
+
|