feat(data): add recipe-driven language supervision (#4182)

* feat(data): add recipe-driven language supervision

* test(collate): expect preserved language columns

* Address PR review feedback

* Address Claude review feedback
This commit is contained in:
Pepijn
2026-08-04 16:48:47 +02:00
committed by GitHub
parent f66e5128ec
commit 64b23178d5
28 changed files with 1062 additions and 79 deletions
+7
View File
@@ -38,6 +38,13 @@ def test_dataset_config_empty_episodes_ok():
DatasetConfig(repo_id="user/repo", episodes=[])
def test_dataset_config_ignores_negative_excluded_episodes(caplog):
config = DatasetConfig(repo_id="user/repo", exclude_episodes=[-2, 1, -1, 3])
assert config.exclude_episodes == [1, 3]
assert "Ignoring negative exclude_episodes entries: [-2, -1]" in caplog.text
def test_dataset_config_bucket_streaming_ok():
DatasetConfig(repo_id="user/repo", repo_type="bucket", streaming=True)