Commit
·
526367e
1
Parent(s):
0588073
WIP
Browse files
script.py
CHANGED
|
@@ -52,8 +52,11 @@ if __name__ == "__main__":
|
|
| 52 |
|
| 53 |
if data_path_test_server.exists():
|
| 54 |
data_path = data_path_test_server
|
|
|
|
| 55 |
else:
|
| 56 |
data_path = data_path_local
|
|
|
|
|
|
|
| 57 |
|
| 58 |
print(data_path)
|
| 59 |
|
|
@@ -65,22 +68,23 @@ if __name__ == "__main__":
|
|
| 65 |
print(data_files)
|
| 66 |
|
| 67 |
|
| 68 |
-
|
| 69 |
dataset = load_dataset(
|
| 70 |
-
"
|
| 71 |
data_files=data_files,
|
| 72 |
trust_remote_code=True,
|
| 73 |
# streaming=True
|
| 74 |
)
|
| 75 |
-
print('load with
|
| 76 |
-
|
| 77 |
dataset = load_dataset(
|
| 78 |
-
"
|
| 79 |
data_files=data_files,
|
| 80 |
trust_remote_code=True,
|
| 81 |
# streaming=True
|
| 82 |
)
|
| 83 |
-
print('load with
|
|
|
|
| 84 |
|
| 85 |
print(dataset, flush=True)
|
| 86 |
# dataset = load_dataset('webdataset', data_files={)
|
|
|
|
| 52 |
|
| 53 |
if data_path_test_server.exists():
|
| 54 |
data_path = data_path_test_server
|
| 55 |
+
TEST_ENV = True
|
| 56 |
else:
|
| 57 |
data_path = data_path_local
|
| 58 |
+
TEST_ENV = False
|
| 59 |
+
|
| 60 |
|
| 61 |
print(data_path)
|
| 62 |
|
|
|
|
| 68 |
print(data_files)
|
| 69 |
|
| 70 |
|
| 71 |
+
if TEST_ENV:
|
| 72 |
dataset = load_dataset(
|
| 73 |
+
"webdataset",
|
| 74 |
data_files=data_files,
|
| 75 |
trust_remote_code=True,
|
| 76 |
# streaming=True
|
| 77 |
)
|
| 78 |
+
print('load with webdataset')
|
| 79 |
+
else:
|
| 80 |
dataset = load_dataset(
|
| 81 |
+
"arrow",
|
| 82 |
data_files=data_files,
|
| 83 |
trust_remote_code=True,
|
| 84 |
# streaming=True
|
| 85 |
)
|
| 86 |
+
print('load with arrow')
|
| 87 |
+
|
| 88 |
|
| 89 |
print(dataset, flush=True)
|
| 90 |
# dataset = load_dataset('webdataset', data_files={)
|