Mike Gabriel commited on
Commit ·
1a67cab
1
Parent(s): 002443a
Add extra typing to make pyright shut up
Browse files- .gitignore +2 -1
- dataset.py +7 -5
.gitignore
CHANGED
|
@@ -2,4 +2,5 @@
|
|
| 2 |
__pycache__
|
| 3 |
.git
|
| 4 |
.wolf*
|
| 5 |
-
.env
|
|
|
|
|
|
| 2 |
__pycache__
|
| 3 |
.git
|
| 4 |
.wolf*
|
| 5 |
+
.env
|
| 6 |
+
pyrightconfig.json
|
dataset.py
CHANGED
|
@@ -1,7 +1,9 @@
|
|
| 1 |
-
from typing import Literal
|
| 2 |
-
from datasets import load_dataset,
|
| 3 |
|
| 4 |
-
|
|
|
|
|
|
|
| 5 |
'train': None,
|
| 6 |
'validation': None,
|
| 7 |
'test': None,
|
|
@@ -12,8 +14,8 @@ _datasets = {
|
|
| 12 |
|
| 13 |
def make_dataset(split: Literal['train', 'validation', 'test'] = 'train'):
|
| 14 |
if _datasets[split] is None:
|
| 15 |
-
ds
|
| 16 |
"karpathy/tiny_shakespeare", split=split, trust_remote_code=True)
|
| 17 |
out = str(list(ds)[0]['text'])
|
| 18 |
_datasets[split] = out
|
| 19 |
-
return _datasets[split]
|
|
|
|
| 1 |
+
from typing import Literal, Union
|
| 2 |
+
from datasets import load_dataset, DatasetDict
|
| 3 |
|
| 4 |
+
DatasetType = Union[None, str]
|
| 5 |
+
|
| 6 |
+
_datasets: dict[str, DatasetType] = {
|
| 7 |
'train': None,
|
| 8 |
'validation': None,
|
| 9 |
'test': None,
|
|
|
|
| 14 |
|
| 15 |
def make_dataset(split: Literal['train', 'validation', 'test'] = 'train'):
|
| 16 |
if _datasets[split] is None:
|
| 17 |
+
ds = load_dataset(
|
| 18 |
"karpathy/tiny_shakespeare", split=split, trust_remote_code=True)
|
| 19 |
out = str(list(ds)[0]['text'])
|
| 20 |
_datasets[split] = out
|
| 21 |
+
return str(_datasets[split])
|