Don't override the language if it's already set

tomaarsen · Oct 31, 2023 · 91bda14 · 91bda14
1 parent a33934a
commit 91bda14
Show file tree

Hide file tree

Showing 2 changed files with 15 additions and 8 deletions.
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -15,6 +15,12 @@ Types of changes
 * "Security" in case of vulnerabilities.
 -->
 
+## [Unreleased]
+
+### Fixed
+
+- No longer override `language` metadata from the dataset if the language was also set manually via `SpanMarkerModelCardData`.
+
 ## [1.4.0]
 
 ### Added

diff --git a/span_marker/model_card.py b/span_marker/model_card.py
@@ -230,14 +230,15 @@ def __post_init__(self):
         # We don't want to save "ignore_metadata_errors" in our Model Card
         if self.dataset_id:
             if is_on_huggingface(self.dataset_id, is_model=False):
-                # if languages are not set, try to determine the language from the dataset on the Hub
-                try:
-                    info = dataset_info(self.dataset_id)
-                except:
-                    pass
-                else:
-                    if info.cardData:
-                        self.language = info.cardData.get("language", self.language)
+                if self.language is None:
+                    # if languages are not set, try to determine the language from the dataset on the Hub
+                    try:
+                        info = dataset_info(self.dataset_id)
+                    except:
+                        pass
+                    else:
+                        if info.cardData:
+                            self.language = info.cardData.get("language", self.language)
             else:
                 logger.warning(
                     f"The provided {self.dataset_id!r} dataset could not be found on the Hugging Face Hub."