diff --git a/benchmarks/swebench/eval_infer.py b/benchmarks/swebench/eval_infer.py
index 10782e6c..206a8076 100644
--- a/benchmarks/swebench/eval_infer.py
+++ b/benchmarks/swebench/eval_infer.py
@@ -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.
diff --git a/benchmarks/swtbench/eval_infer.py b/benchmarks/swtbench/eval_infer.py
index c245aa42..e5bdc9b5 100644
--- a/benchmarks/swtbench/eval_infer.py
+++ b/benchmarks/swtbench/eval_infer.py
@@ -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:
     """
@@ -418,6 +418,7 @@ def main() -> None:
             try_pull_prebaked_images(
                 output_file,
                 args.dataset,
+                args.split,
             )
         else:
             logger.info(