Make snli dataset reader work with unlabeled instances (#234)

* Make snli dataset reader work with unlabeled instances * Add entry to changelog
allenai · Mar 29, 2021 · a8e0b00 · a8e0b00
1 parent eb72cdc
commit a8e0b00
Show file tree

Hide file tree

Showing 2 changed files with 3 additions and 2 deletions.
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -29,6 +29,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 - Updated `QaNet` and `NumericallyAugmentedQaNet` models to remove bias for layers that are followed by normalization layers.
 - Updated the model cards for `rc-naqanet`, `vqa-vilbert` and `ve-vilbert`.
 - Predictors now work for the vilbert-multitask model.
+- Support unlabeled instances in `SnliDatasetReader`.
 
 
 ## [v2.1.0](https://github.com/allenai/allennlp-models/releases/tag/v2.1.0) - 2021-02-24

diff --git a/allennlp_models/pair_classification/dataset_readers/snli.py b/allennlp_models/pair_classification/dataset_readers/snli.py
@@ -80,10 +80,10 @@ def _read(self, file_path: str):
         with open(file_path, "r") as snli_file:
             example_iter = (json.loads(line) for line in snli_file)
             filtered_example_iter = (
-                example for example in example_iter if example["gold_label"] != "-"
+                example for example in example_iter if example.get("gold_label") != "-"
             )
             for example in self.shard_iterable(filtered_example_iter):
-                label = example["gold_label"]
+                label = example.get("gold_label")
                 premise = example["sentence1"]
                 hypothesis = example["sentence2"]
                 yield self.text_to_instance(premise, hypothesis, label)