remove chinese demo (#1280)

Author: bkorycki

src/modelgauge/prompt_sets.py

@@ -32,7 +32,6 @@ def description(cls) -> SecretDescription:
     "demo": {
         "en_us": "airr_official_1.0_demo_prompt_set_release",
         "fr_fr": "airr_official_1.0_demo_fr_fr_prompt_set_release",
-        "zh_cn": "airr_official_1.0_demo_zh_cn_prompt_set_release",
     },
 }
 SECURITY_JAILBREAK_PROMPT_SETS = {

tests/modelbench_tests/test_benchmark.py

@@ -106,7 +106,7 @@ def assert_hazard_standards(benchmark):
 
 
 def test_chinese_benchmark_v1_definition_standards(fake_secrets):
-    """Remove this test when ZH_CN gets added to PUBLISHED_LOCALES. Right now there is no official prompt set."""
+    """Remove this test when ZH_CN gets added to PUBLISHED_LOCALES."""
 
     def assert_hazard_standards(benchmark):
         for h in benchmark.hazards():
@@ -123,14 +123,6 @@ def assert_hazard_standards(benchmark):
     assert_hazard_standards(practice_benchmark)
     assert_hazard_standards(practice_default_benchmark)
 
-    # Demo prompt set uses practice standards.
-    demo_benchmark = GeneralPurposeAiChatBenchmarkV1(ZH_CN, "demo", evaluator="ensemble")
-    demo_default_benchmark = GeneralPurposeAiChatBenchmarkV1(ZH_CN, "demo", evaluator="default")
-    assert demo_benchmark.standards._data == practice_benchmark.standards._data
-    assert demo_default_benchmark.standards._data == practice_benchmark.standards._data
-    assert_hazard_standards(demo_benchmark)
-    assert_hazard_standards(demo_default_benchmark)
-
 
 def test_benchmark_v1_hazard_standard_scores(fake_secrets):
     """Spot check that the correct scores are being retrieved from the files."""

tests/modelgauge_tests/test_object_creation.py

@@ -48,11 +48,6 @@ def shared_run_dir(tmp_path_factory):
 def test_all_tests_make_test_items(test_name, shared_run_dir):
     test = TESTS.make_instance(test_name, secrets=load_secrets_from_config(path="."))
 
-    # TODO remove when localized files are handled better
-    # workaround
-    if isinstance(test, BaseSafeTestVersion1) and test.locale != EN_US:
-        return
-
     if isinstance(test, PromptResponseTest):
         test_data_path = os.path.join(shared_run_dir, test.__class__.__name__)
         dependency_helper = FromSourceDependencyHelper(