Skip to content
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
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
10 changes: 5 additions & 5 deletions benchmarks/swebench/eval_infer.py
Original file line number Diff line number Diff line change
Expand Up @@ -117,11 +117,11 @@ def convert_to_swebench_format(input_file: str, output_file: str) -> None:
def run_swebench_evaluation(
predictions_file: str,
run_id: str,
dataset: str = EVAL_DEFAULTS["dataset"],
workers: int = EVAL_DEFAULTS["workers"],
split: str = EVAL_DEFAULTS["split"],
modal: bool = EVAL_DEFAULTS["modal"],
timeout: int = EVAL_DEFAULTS["timeout"],
dataset: str,
workers: int,
split: str,
modal: bool,
timeout: int,
) -> None:
"""
Run SWE-Bench evaluation on the predictions file.
Expand Down
3 changes: 2 additions & 1 deletion benchmarks/swtbench/eval_infer.py
Original file line number Diff line number Diff line change
Expand Up @@ -69,7 +69,7 @@ def _load_prediction_instance_ids(predictions_file: Path) -> list[str]:
def try_pull_prebaked_images(
predictions_file: Path,
dataset: str,
split: str = EVAL_DEFAULTS["split"],
split: str,
registry: str = PREBAKED_REGISTRY,
) -> None:
"""
Expand Down Expand Up @@ -418,6 +418,7 @@ def main() -> None:
try_pull_prebaked_images(
output_file,
args.dataset,
args.split,
)
else:
logger.info(
Expand Down