From 942054c43ac77ebb5dd2e4618eeea61aeeeb753a Mon Sep 17 00:00:00 2001 From: Bryan Helmkamp Date: Mon, 16 Mar 2026 11:24:56 -0400 Subject: [PATCH] Record timeout and sandbox resources in scoreboard metadata Adds --timeout, --sandbox-cpu, --sandbox-memory flags to record_results.py. Re-recorded both existing runs with the new fields. Co-Authored-By: Claude Opus 4.6 (1M context) --- evals/swe-bench/record_results.py | 15 +++++++++++++++ .../scoreboard/haiku-baseline-20260316/meta.json | 5 ++++- .../scoreboard/sonnet-baseline-20260316/meta.json | 5 ++++- 3 files changed, 23 insertions(+), 2 deletions(-) diff --git a/evals/swe-bench/record_results.py b/evals/swe-bench/record_results.py index 0ffd5db73..e928edd3b 100644 --- a/evals/swe-bench/record_results.py +++ b/evals/swe-bench/record_results.py @@ -79,6 +79,18 @@ def main(): "--notes", default="", help="Additional notes or observations", ) + parser.add_argument( + "--timeout", type=int, default=600, + help="Per-instance timeout used (seconds)", + ) + parser.add_argument( + "--sandbox-cpu", type=int, default=2, + help="CPUs per Daytona sandbox", + ) + parser.add_argument( + "--sandbox-memory", type=int, default=4, + help="Memory (GB) per Daytona sandbox", + ) args = parser.parse_args() run_dir = SCOREBOARD_DIR / args.run_name @@ -167,6 +179,9 @@ def main(): "model": gen_summary.get("model", "unknown"), "provider": gen_summary.get("provider", "unknown"), "fabro_version": get_fabro_version(), + "timeout_s": args.timeout, + "sandbox_cpu": args.sandbox_cpu, + "sandbox_memory_gb": args.sandbox_memory, "description": args.description, "notes": args.notes, "total_instances": total, diff --git a/evals/swe-bench/scoreboard/haiku-baseline-20260316/meta.json b/evals/swe-bench/scoreboard/haiku-baseline-20260316/meta.json index 72426a83d..390c4d292 100644 --- a/evals/swe-bench/scoreboard/haiku-baseline-20260316/meta.json +++ b/evals/swe-bench/scoreboard/haiku-baseline-20260316/meta.json @@ -1,9 +1,12 @@ { "run_name": "haiku-baseline-20260316", - "timestamp": "2026-03-16T13:53:04.401784+00:00", + "timestamp": "2026-03-16T15:24:29.679846+00:00", "model": "claude-haiku-4-5", "provider": "anthropic", "fabro_version": "fabro 0.5.0 (4c6ad4d 2026-03-16)", + "timeout_s": 600, + "sandbox_cpu": 2, + "sandbox_memory_gb": 4, "description": "Haiku 4.5 baseline, default prompt, 2 CPU / 4 GB, 10min timeout", "notes": "", "total_instances": 300, diff --git a/evals/swe-bench/scoreboard/sonnet-baseline-20260316/meta.json b/evals/swe-bench/scoreboard/sonnet-baseline-20260316/meta.json index 912ee5d52..78b316bbe 100644 --- a/evals/swe-bench/scoreboard/sonnet-baseline-20260316/meta.json +++ b/evals/swe-bench/scoreboard/sonnet-baseline-20260316/meta.json @@ -1,9 +1,12 @@ { "run_name": "sonnet-baseline-20260316", - "timestamp": "2026-03-16T15:18:35.681947+00:00", + "timestamp": "2026-03-16T15:24:40.795288+00:00", "model": "claude-sonnet-4-6", "provider": "anthropic", "fabro_version": "fabro 0.5.0 (4c6ad4d 2026-03-16)", + "timeout_s": 600, + "sandbox_cpu": 2, + "sandbox_memory_gb": 4, "description": "Sonnet 4.6 baseline, default prompt, 2 CPU / 4 GB, 10min timeout", "notes": "", "total_instances": 300,