Merge pull request #22 from bricksdont/add_dgs_sentence_split

Add dgs sentence split

Author: AmitMY

setup.py

@@ -11,7 +11,7 @@
 setup(
     name="sign-language-datasets",
     packages=packages,
-    version="0.1.2",
+    version="0.1.3",
     description="TFDS Datasets for sign language",
     author="Amit Moryossef",
     author_email="amitmoryossef@gmail.com",

sign_language_datasets/datasets/dgs_corpus/dgs_corpus.py

@@ -58,6 +58,7 @@
 
 _KNOWN_SPLITS = {
     "3.0.0-uzh-document": path.join(path.dirname(path.realpath(__file__)), "splits", "split.3.0.0-uzh-document.json"),
+    "3.0.0-uzh-sentence": path.join(path.dirname(path.realpath(__file__)), "splits", "split.3.0.0-uzh-sentence.json")
 }
 
 
@@ -377,7 +378,7 @@ def _generate_examples(self, data, split: List[str] | Dict[str, List[str]] = Non
             else:
                 sentences = list(get_elan_sentences(datum["eaf"]))
                 for sentence in sentences:
-                    if split is not None and sentence["id"] not in split[_id]:
+                    if split is not None and sentence["id"] not in split[document_id]:
                         continue
 
                     if sentence["english"] is None: