From 75ef47c342b3edabdda913c01793778ab5bc6570 Mon Sep 17 00:00:00 2001 From: Frank Date: Sat, 12 Sep 2026 20:06:25 -0700 Subject: [PATCH] frank-large: EXPERIMENT - HALOGEN_MAX_TOK back to upstream default 32768 Operator call. Prediction from the measured model: 524288 positions cost 35.0 GiB against a 29.5 budget, so HALOGEN_KV_POOL_FIT clamps the pool to 262144 (27.8). Rollback is a single revert; 16384 was the config that made the doubling real. --- dankclaw/frank-large/quadlet/frank-large.container | 6 +++++- 1 file changed, 5 insertions(+), 1 deletion(-) diff --git a/dankclaw/frank-large/quadlet/frank-large.container b/dankclaw/frank-large/quadlet/frank-large.container index 16942eb..3a4f3f7 100644 --- a/dankclaw/frank-large/quadlet/frank-large.container +++ b/dankclaw/frank-large/quadlet/frank-large.container @@ -126,7 +126,11 @@ Environment=HALOGEN_KV_SLOTS=4 # not shrink the pool, it stops the server booting at all — and the unit still # reads active with the container "Up (starting)", so the health check waits # out LLM_WAIT rather than failing fast. -Environment=HALOGEN_MAX_TOK=16384 + +# EXPERIMENT 2026-09-12 (operator call): back to upstream's default 32768. +# Model predicts 524288 needs 35.0 against a 29.5 budget, so HALOGEN_KV_POOL_FIT +# should clamp the pool to 262144 (~27.8) at boot. Watch the ledger. +Environment=HALOGEN_MAX_TOK=32768 # RAM left for the n-gram table the engine reads through the page cache. # Above the vendor's 20 because a dozen containers share this host; too low # and long prompts crawl with the disk busy.