Spaces:
Running
Running
Update Space (evaluate main: e4a27243)
Browse files- requirements.txt +1 -1
- xtreme_s.py +20 -7
requirements.txt
CHANGED
@@ -1,2 +1,2 @@
|
|
1 |
-
git+https://github.com/huggingface/evaluate@
|
2 |
sklearn
|
|
|
1 |
+
git+https://github.com/huggingface/evaluate@e4a2724377909fe2aeb4357e3971e5a569673b39
|
2 |
sklearn
|
xtreme_s.py
CHANGED
@@ -13,7 +13,8 @@
|
|
13 |
# limitations under the License.
|
14 |
""" XTREME-S benchmark metric. """
|
15 |
|
16 |
-
from
|
|
|
17 |
|
18 |
import datasets
|
19 |
from datasets.config import PY_VERSION
|
@@ -218,11 +219,22 @@ def wer_and_cer(preds, labels, concatenate_texts, config_name):
|
|
218 |
return {"wer": compute_score(preds, labels, "wer"), "cer": compute_score(preds, labels, "cer")}
|
219 |
|
220 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
221 |
@evaluate.utils.file_utils.add_start_docstrings(_DESCRIPTION, _KWARGS_DESCRIPTION)
|
222 |
class XtremeS(evaluate.Metric):
|
223 |
-
|
224 |
-
|
225 |
-
|
|
|
|
|
226 |
|
227 |
pred_type = "int64" if self.config_name in ["fleurs-lang_id", "minds14"] else "string"
|
228 |
|
@@ -230,6 +242,7 @@ class XtremeS(evaluate.Metric):
|
|
230 |
description=_DESCRIPTION,
|
231 |
citation=_CITATION,
|
232 |
inputs_description=_KWARGS_DESCRIPTION,
|
|
|
233 |
features=datasets.Features(
|
234 |
{"predictions": datasets.Value(pred_type), "references": datasets.Value(pred_type)}
|
235 |
),
|
@@ -238,10 +251,10 @@ class XtremeS(evaluate.Metric):
|
|
238 |
format="numpy",
|
239 |
)
|
240 |
|
241 |
-
def _compute(self, predictions, references
|
242 |
|
243 |
-
bleu_kwargs = bleu_kwargs if bleu_kwargs is not None else {}
|
244 |
-
wer_kwargs = wer_kwargs if wer_kwargs is not None else {}
|
245 |
|
246 |
if self.config_name == "fleurs-lang_id":
|
247 |
return {"accuracy": simple_accuracy(predictions, references)}
|
|
|
13 |
# limitations under the License.
|
14 |
""" XTREME-S benchmark metric. """
|
15 |
|
16 |
+
from dataclasses import dataclass
|
17 |
+
from typing import List, Optional
|
18 |
|
19 |
import datasets
|
20 |
from datasets.config import PY_VERSION
|
|
|
219 |
return {"wer": compute_score(preds, labels, "wer"), "cer": compute_score(preds, labels, "cer")}
|
220 |
|
221 |
|
222 |
+
@dataclass
|
223 |
+
class XtremeSConfig(evaluate.info.Config):
|
224 |
+
|
225 |
+
name: str = "default"
|
226 |
+
|
227 |
+
bleu_kwargs: Optional[dict] = None
|
228 |
+
wer_kwargs: Optional[dict] = None
|
229 |
+
|
230 |
+
|
231 |
@evaluate.utils.file_utils.add_start_docstrings(_DESCRIPTION, _KWARGS_DESCRIPTION)
|
232 |
class XtremeS(evaluate.Metric):
|
233 |
+
|
234 |
+
CONFIG_CLASS = XtremeSConfig
|
235 |
+
ALLOWED_CONFIG_NAMES = _CONFIG_NAMES
|
236 |
+
|
237 |
+
def _info(self, config):
|
238 |
|
239 |
pred_type = "int64" if self.config_name in ["fleurs-lang_id", "minds14"] else "string"
|
240 |
|
|
|
242 |
description=_DESCRIPTION,
|
243 |
citation=_CITATION,
|
244 |
inputs_description=_KWARGS_DESCRIPTION,
|
245 |
+
config=config,
|
246 |
features=datasets.Features(
|
247 |
{"predictions": datasets.Value(pred_type), "references": datasets.Value(pred_type)}
|
248 |
),
|
|
|
251 |
format="numpy",
|
252 |
)
|
253 |
|
254 |
+
def _compute(self, predictions, references):
|
255 |
|
256 |
+
bleu_kwargs = self.config.bleu_kwargs if self.config.bleu_kwargs is not None else {}
|
257 |
+
wer_kwargs = self.config.wer_kwargs if self.config.wer_kwargs is not None else {}
|
258 |
|
259 |
if self.config_name == "fleurs-lang_id":
|
260 |
return {"accuracy": simple_accuracy(predictions, references)}
|