Skip to content
Snippets Groups Projects

Draft: Introduce automatic sanity check for training

Open Vlad-Andrei BĂDOIU (78692) requested to merge vladb/test_train into main
Files
7
 
import os
 
 
from enum import Enum
 
from functools import partial
 
from .tinystories import TinyStoriesDataset
from .tinystories import TinyStoriesDataset
from .wikitext103 import WikiText103Dataset
from .wikitext103 import WikiText103Dataset
 
from .line_ds import LineDataset
 
 
 
class Datasets(Enum):
 
 
WIKITEXT103 = WikiText103Dataset
 
TINYSTORIES = TinyStoriesDataset
 
 
TEST_DS = partial(LineDataset, path=os.getenv("TEST_DATASET_PATH", default='tests/fixtures/sample.txt'))
 
 
def __call__(self, **kwargs):
 
return self.value(**kwargs)
 
\ No newline at end of file
Loading