fix: take filter_expr into account in dependency parsing evaluation

changelog.md

@@ -98,6 +98,11 @@
 - The loss of `eds.ner_crf` is now computed as the mean over the words instead of the sum. This change is compatible with multi-gpu training.
 - Having multiple stats keys matching a batching pattern now warns instead of raising an error.
 
+### Changed
+
+- The loss of `eds.ner_crf` is now computed as the mean over the words instead of the sum. This change is compatible with multi-gpu training.
+- Having multiple stats keys matching a batching pattern now warns instead of raising an error.
+
 ### Fixed
 
 - Support packaging with poetry 2.0

edsnlp/metrics/dep_parsing.py

@@ -17,7 +17,9 @@ def dependency_parsing_metric(
         The examples to score, either a tuple of (golds, preds) or a list of
         spacy.training.Example objects
     filter_expr : Optional[str]
-        The filter expression to use to filter the documents
+        The filter expression to use to filter the documents. The expression
+        should be a valid Python test and uses the variable `doc` to refer to
+        the reference (gold) document.
 
     Returns
     -------

edsnlp/pipes/trainable/ner_crf/ner_crf.py

@@ -454,6 +454,9 @@ def collate(self, preps) -> NERBatchInput:
                 k: sum(v) for k, v in preps["stats"].items() if not k.startswith("__")
             },
         }
+        collated["stats"] = {
+            k: sum(v) for k, v in preps["stats"].items() if not k.startswith("__")
+        }
         lengths = [length for sample in preps["lengths"] for length in sample]
         max_len = max(lengths)
         if "targets" in preps:

tests/training/dep_parser_config.yml

@@ -23,6 +23,7 @@ scorer:
   speed: false
   dep:
     '@metrics': "eds.dep_parsing"
+    filter_expr: "doc.text != ''"
 
 # 🎛️ OPTIMIZER
 optimizer: