Make sure benchmark runs and be more strict about collecting feedback (#397)

* Update benchmark script * Bump version
2025-12-17 04:35:17 +01:00 · 2023-06-25 15:53:34 +02:00
parent b09f0e9193
commit 7f16abecb3
4 changed files with 7 additions and 5 deletions
--- a/scripts/benchmark.py
+++ b/scripts/benchmark.py
@@ -68,7 +68,7 @@ def main(
                    "gpt_engineer.main",
                    bench_folder,
                    "--steps",
-                    "execute_only",
+                    "evaluate",
                ],
            )