Skip to content

feat(code_gen): add per-language subset metrics for livecodebench-x #1719

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
rodboev wants to merge 3 commits into
base: main
Choose a base branch
from
Open

feat(code_gen): add per-language subset metrics for livecodebench-x #1719

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
a0a029e
feat(code_gen): add per-language subset metrics for livecodebench-x
rodboev Jun 25, 2026
bbeca8d
fix(code_gen): declare target_language on verify request and extract …
rodboev Jun 25, 2026
cd3ef8e
fix(code_gen): drop redundant target_language kwarg shadowed by model…
rodboev Jun 25, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
3 changes: 3 additions & 0 deletions resources_servers/code_gen/app.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -64,6 +64,7 @@ class CompCodingRunRequest(BaseRunRequest):

class CompCodingVerifyRequest(CompCodingRunRequest, BaseVerifyRequest):
verifier_metadata: Optional[Dict[str, Any]] = None
target_language: Optional[str] = None


class CompCodingVerifyResponse(BaseVerifyResponse):
Expand All @@ -74,6 +75,7 @@ class CompCodingVerifyResponse(BaseVerifyResponse):
unit_tests_time_taken: Optional[float] = None
reasoning_format_violation_rate: float = 0.0
difficulty: Optional[str] = None
target_language: Optional[str] = None


# ----------------------------
Expand Down Expand Up @@ -124,6 +126,7 @@ def compute_metrics(self, tasks: List[List[Dict[str, Any]]]) -> Dict[str, Any]:
)
add_avg_sample_std_dev(metrics, all_score_dicts, score_names, max_k)
metrics.update(compute_subset_metrics(tasks, "difficulty", self._code_score_fn, "extracted_model_code"))
metrics.update(compute_subset_metrics(tasks, "target_language", self._code_score_fn, "extracted_model_code"))
return metrics

def get_key_metrics(self, agent_metrics: Dict[str, Any]) -> Dict[str, Any]:
Expand Down
21 changes: 21 additions & 0 deletions resources_servers/code_gen/tests/test_app.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -146,6 +146,10 @@ def test_verify_missing_response_validation_error(self) -> None:
verifier_metadata={"unit_tests": {"inputs": ["1\n"], "outputs": ["1"]}},
)

def test_verify_target_language_field(self) -> None:
resp = CompCodingVerifyResponse(reward=0.0, target_language="ja")
assert resp.target_language == "ja"

async def test_verify_no_code_block(self, code_gen_resources_server_client: TestClient) -> None:
"""Test when response contains no code block - should extract raw text"""
response = NeMoGymResponse(
Expand Down Expand Up @@ -327,6 +331,23 @@ def test_no_answer_tracked(self) -> None:
assert "pass@1[avg-of-2]/no_answer" in m
assert m["pass@1[avg-of-2]/no_answer"] == pytest.approx(50.0)

def test_produces_per_language_subsets(self):
server = _make_server()
tasks = [
[
{"reward": 1.0, "extracted_model_code": "print(1)", "target_language": "de"},
{"reward": 1.0, "extracted_model_code": "print(1)", "target_language": "de"},
],
[
{"reward": 0.0, "extracted_model_code": "print(2)", "target_language": "fr"},
{"reward": 0.0, "extracted_model_code": "print(2)", "target_language": "fr"},
],
]
m = server.compute_metrics(tasks)
assert "de/pass@1/accuracy" in m
assert "fr/pass@1/accuracy" in m
assert m["de/pass@1/accuracy"] > m["fr/pass@1/accuracy"]


class TestGetKeyMetrics:
@pytest.mark.asyncio
Expand Down