From ae842555d89f582610da6c8cf0dcf61d3409782b Mon Sep 17 00:00:00 2001 From: "russell@unturf.com" Date: Sat, 30 May 2026 22:19:43 -0400 Subject: [PATCH] =?UTF-8?q?bench:=20bump=20cross-model=20bench=20timeout?= =?UTF-8?q?=20default=20180s=20=E2=86=92=20600s?= MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Real-world: shared endpoints can queue (hermes was queued behind opencompletion this morning, single calls pushed past 180s but completed in 121s on retry). 600s tolerates ~10× the typical non-queued latency before declaring timeout. --- Makefile | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/Makefile b/Makefile index 5474568..08fe091 100644 --- a/Makefile +++ b/Makefile @@ -432,7 +432,7 @@ bench-emergent-pending: bootstrap ## print log entries awaiting teacher review # options" pattern that uncloseai-cli's tool_arborist supports. BENCH_CM_QUESTIONS ?= bench/qa_questions_smoke.txt BENCH_CM_OUT ?= bench/cross_model_results -BENCH_CM_TIMEOUT ?= 180 +BENCH_CM_TIMEOUT ?= 600 BENCH_CM_LIMIT ?= 0 bench-cross-model: bootstrap ## cross-model self-play (Hermes + Qwen on smoke fixture) [BENCH_CM_QUESTIONS=... LIMIT=N BURN=1] PYTHONUNBUFFERED=1 $(PY) bench/cross_model_selfplay.py \