File size: 531 Bytes
4fad798
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
# See https://huggingface.co/docs/datasets/audio_dataset
from datasets import load_dataset

# Upload the have_one dataset to the hub

# This magically does the right thing, including
# setting the label names and IDs.
ds = load_dataset("audiofolder", data_dir="have_one/data")

ds2 = ds.train_test_split(test_size=0.15)

ds3 = ds2['train'].train_test_split(test_size=0.18)

ds2['validation'] = ds3['test']

from huggingface_hub import login
login()
ds2.push_to_hub("MatsRooth/have_one",private=False,embed_external_files=True)