Create prepare_dummy_dataset.py
Browse files- prepare_dummy_dataset.py +27 -0
prepare_dummy_dataset.py
ADDED
@@ -0,0 +1,27 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
from datasets import Dataset, Features
|
2 |
+
from datasets import Image as ImageFeature
|
3 |
+
from datasets import Value
|
4 |
+
import numpy as np
|
5 |
+
|
6 |
+
TOTAL_SAMPLES = 100000
|
7 |
+
|
8 |
+
def gen_examples():
|
9 |
+
for _ in range(TOTAL_SAMPLES):
|
10 |
+
yield {
|
11 |
+
"image": np.random.randint(low=0, high=255, size=(512, 512, 3)),
|
12 |
+
"condtioning_image": np.random.randint(low=0, high=255, size=(512, 512, 3)),
|
13 |
+
"caption": "Killswitch engage",
|
14 |
+
}
|
15 |
+
|
16 |
+
ds = Dataset.from_generator(
|
17 |
+
gen_examples,
|
18 |
+
features=Features(
|
19 |
+
image=ImageFeature(),
|
20 |
+
condtioning_image=ImageFeature(),
|
21 |
+
caption=Value("string"),
|
22 |
+
),
|
23 |
+
num_proc=8
|
24 |
+
)
|
25 |
+
|
26 |
+
ds_name = f"dummy-controlnet-{TOTAL_SAMPLES}-samples"
|
27 |
+
ds.push_to_hub(ds_name)
|