Upload dataset.py with huggingface_hub
Browse files- dataset.py +35 -9
dataset.py
CHANGED
|
@@ -1,3 +1,5 @@
|
|
|
|
|
|
|
|
| 1 |
import datasets
|
| 2 |
|
| 3 |
from .artifact import Artifact, UnitxtArtifactNotFoundError
|
|
@@ -27,7 +29,7 @@ from .processors import __file__ as _
|
|
| 27 |
from .random_utils import __file__ as _
|
| 28 |
from .recipe import __file__ as _
|
| 29 |
from .register import __file__ as _
|
| 30 |
-
from .register import register_all_artifacts
|
| 31 |
from .renderers import __file__ as _
|
| 32 |
from .schema import __file__ as _
|
| 33 |
from .split_utils import __file__ as _
|
|
@@ -43,6 +45,8 @@ from .validate import __file__ as _
|
|
| 43 |
from .version import __file__ as _
|
| 44 |
from .version import version
|
| 45 |
|
|
|
|
|
|
|
| 46 |
|
| 47 |
def fetch(artifact_name):
|
| 48 |
try:
|
|
@@ -77,6 +81,18 @@ def parse(query: str):
|
|
| 77 |
return result
|
| 78 |
|
| 79 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 80 |
class Dataset(datasets.GeneratorBasedBuilder):
|
| 81 |
"""TODO: Short description of my dataset."""
|
| 82 |
|
|
@@ -85,15 +101,25 @@ class Dataset(datasets.GeneratorBasedBuilder):
|
|
| 85 |
|
| 86 |
@property
|
| 87 |
def generators(self):
|
| 88 |
-
register_all_artifacts()
|
| 89 |
if not hasattr(self, "_generators") or self._generators is None:
|
| 90 |
-
|
| 91 |
-
|
| 92 |
-
|
| 93 |
-
|
| 94 |
-
|
| 95 |
-
|
| 96 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 97 |
return self._generators
|
| 98 |
|
| 99 |
def _info(self):
|
|
|
|
| 1 |
+
import os
|
| 2 |
+
|
| 3 |
import datasets
|
| 4 |
|
| 5 |
from .artifact import Artifact, UnitxtArtifactNotFoundError
|
|
|
|
| 29 |
from .random_utils import __file__ as _
|
| 30 |
from .recipe import __file__ as _
|
| 31 |
from .register import __file__ as _
|
| 32 |
+
from .register import _reset_env_local_catalogs, register_all_artifacts
|
| 33 |
from .renderers import __file__ as _
|
| 34 |
from .schema import __file__ as _
|
| 35 |
from .split_utils import __file__ as _
|
|
|
|
| 45 |
from .version import __file__ as _
|
| 46 |
from .version import version
|
| 47 |
|
| 48 |
+
__default_recipe__ = "common_recipe"
|
| 49 |
+
|
| 50 |
|
| 51 |
def fetch(artifact_name):
|
| 52 |
try:
|
|
|
|
| 81 |
return result
|
| 82 |
|
| 83 |
|
| 84 |
+
def get_dataset_artifact(dataset_str):
|
| 85 |
+
_reset_env_local_catalogs()
|
| 86 |
+
register_all_artifacts()
|
| 87 |
+
recipe = fetch(dataset_str)
|
| 88 |
+
if recipe is None:
|
| 89 |
+
args = parse(dataset_str)
|
| 90 |
+
if "type" not in args:
|
| 91 |
+
args["type"] = os.environ.get("UNITXT_DEFAULT_RECIPE", __default_recipe__)
|
| 92 |
+
recipe = Artifact.from_dict(args)
|
| 93 |
+
return recipe
|
| 94 |
+
|
| 95 |
+
|
| 96 |
class Dataset(datasets.GeneratorBasedBuilder):
|
| 97 |
"""TODO: Short description of my dataset."""
|
| 98 |
|
|
|
|
| 101 |
|
| 102 |
@property
|
| 103 |
def generators(self):
|
|
|
|
| 104 |
if not hasattr(self, "_generators") or self._generators is None:
|
| 105 |
+
try:
|
| 106 |
+
from unitxt.dataset import (
|
| 107 |
+
get_dataset_artifact as get_dataset_artifact_installed,
|
| 108 |
+
)
|
| 109 |
+
|
| 110 |
+
unitxt_installed = True
|
| 111 |
+
except ImportError:
|
| 112 |
+
unitxt_installed = False
|
| 113 |
+
|
| 114 |
+
if unitxt_installed:
|
| 115 |
+
print("Loading with installed unitxt library...")
|
| 116 |
+
dataset = get_dataset_artifact_installed(self.config.name)
|
| 117 |
+
else:
|
| 118 |
+
print("Loading with installed unitxt library...")
|
| 119 |
+
dataset = get_dataset_artifact(self.config.name)
|
| 120 |
+
|
| 121 |
+
self._generators = dataset()
|
| 122 |
+
|
| 123 |
return self._generators
|
| 124 |
|
| 125 |
def _info(self):
|