Merge branch 'main' into main

huggingface · Nov 8, 2024 · e36937e · e36937e
2 parents cf18f62 + 78d9785
commit e36937e
Show file tree

Hide file tree

Showing 2 changed files with 48 additions and 1 deletion.
diff --git a/src/lighteval/tasks/templates/utils/translation_literals.py b/src/lighteval/tasks/templates/utils/translation_literals.py
@@ -149,7 +149,29 @@ def __getattribute__(self, name: str) -> str:
         colon=":",
         semicolon=";",
     ),
-    Language.BELARUSIAN: TranslationLiterals(language=Language.BELARUSIAN),
+    Language.BELARUSIAN: TranslationLiterals(
+        language=Language.BELARUSIAN,
+        question_word="пытанне",
+        answer="адказ",
+        confirmation_word="ці не так",
+        yes="так",
+        no="не",
+        also="апроч таго",
+        cause_word="бо",
+        effect_word="таму",
+        true="праўда",
+        false="няпраўда",
+        neither="ні тое, ні тое",
+        or_word="ці",
+        full_stop=".",
+        comma=",",
+        question_mark="?",
+        exclamation_mark="!",
+        word_space=" ",
+        sentence_space=" ",
+        colon=":",
+        indices=["А", "Б", "В", "Г", "Д", "Е"],
+    ),
     Language.BENGALI: TranslationLiterals(language=Language.BENGALI, question_word="প্রশ্ন"),
     Language.BIHARI: TranslationLiterals(language=Language.BIHARI),  # Deprecated
     Language.BOSNIAN: TranslationLiterals(language=Language.BOSNIAN),
@@ -865,6 +887,29 @@ def __getattribute__(self, name: str) -> str:
         colon=":",
     ),
     Language.TURKMEN: TranslationLiterals(language=Language.TURKMEN),
+    Language.UDMURT: TranslationLiterals(
+        language=Language.UDMURT,
+        question_word="юан",
+        answer="валэктон",
+        confirmation_word="озьы-а",
+        yes="бен",
+        no="ӧвӧл",
+        also="озьы ик",
+        cause_word="малы ке шуоно",
+        effect_word="соин ик",
+        true="шонерлык",
+        false="пӧяськон",
+        neither="мукет",
+        or_word="яке",
+        full_stop=".",
+        comma=",",
+        question_mark="?",
+        exclamation_mark="!",
+        word_space=" ",
+        sentence_space=" ",
+        colon=":",
+        indices=["А", "Б", "В", "Г", "Д", "Е"],
+    ),
     Language.UKRAINIAN: TranslationLiterals(
         language=Language.UKRAINIAN,
         question_word="питання",

diff --git a/src/lighteval/utils/language.py b/src/lighteval/utils/language.py
@@ -121,6 +121,7 @@ class Language(Enum):
     CEBUANO = "ceb"
     WAR = "war"
     SHAN = "shn"
+    UDMURT = "udm"
 
 
 # This mapping was created for beleble, it converts iso_639_3 individual codes to iso_639_3 macro codes
@@ -232,6 +233,7 @@ class Language(Enum):
     "ars": Language.ARABIC,
     "bul": Language.BULGARIAN,
     "est": Language.ESTONIAN,
+    "udm": Language.UDMURT,
     #  'hau': Language.HAUSA,
     "ind": Language.INDONESIAN,
     #  'kea': Language.KABUVERDIANU,