From 299078eaba6fee1c4c0fdc0317f575cf201a3fe2 Mon Sep 17 00:00:00 2001 From: Frank Date: Fri, 30 Oct 2026 15:33:52 -0700 Subject: [PATCH] campaign(qwen38): the calibration floor is 12M tokens MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Frank's rule: the served iq4r this arm competes with was calibrated on a 10M-token set, so the arm is calibrated at least as deeply. The floor was lowered to 1.5M on the vendor's '10M beats 1M by under 1%' note; that measures the vendor's own plate, not ours — the arm is judged against a 10M-calibrated reference, so cutting the calibration 6x to save wall clock buys a worse position on the only comparison that matters. One round of the re-chunked corpus is ~9.31M tokens, so a 12M floor seals in round 1 plus part of round 2. A round that seals nothing is normal (the floor is pooled) and round_verdict treats it as continue, never failure. Also fixes tests/qwen38_campaign.bats' window dry run: its fixture lands on /tmp (a 16 GiB tmpfs) where the 60 GiB seal preflight can never pass, so the test now uses the documented Q38_CALIB_MIN_FREE_MIB override instead of asserting environment luck. --- scripts/run-qwen38-bf16.sh | 22 +++++++++++++--------- tests/qwen38_campaign.bats | 3 ++- tests/qwen38_reference.bats | 4 ++-- 3 files changed, 17 insertions(+), 12 deletions(-) diff --git a/scripts/run-qwen38-bf16.sh b/scripts/run-qwen38-bf16.sh index 2782938..4408b52 100755 --- a/scripts/run-qwen38-bf16.sh +++ b/scripts/run-qwen38-bf16.sh @@ -25,17 +25,21 @@ # through the ro /models mount); OUT lands on a writable # mount because --kld-out F also writes F.rows beside it. # calib serve on :7770 with the calibration tap open -# (RADIANCE_CALIB_DIR + RADIANCE_CALIB_TOKENS=1500000). A +# (RADIANCE_CALIB_DIR + RADIANCE_CALIB_TOKENS=12000000). A # separate driver — scripts/calib-push.py --port 7770 # --container qwen38-bf16 --corpus -# /home/dank/corpora/out/calib.jsonl — pushes the corpus; at -# the floor the engine writes the Grams and the tap switches +# /home/dank/corpora/out/calib-7168.jsonl — pushes the corpus; +# at the floor the engine writes the Grams and the tap switches # itself off (calib-push prints TAP_DONE). Stop this script -# after that; the trap brings frank-archon back. The vendor's -# own measurement is the floor's warrant — "a pooled Gram -# from 10 M tokens beat one from 1 M by under 1 %" — and at -# this rig's measured push rate a 1.5 M floor is hours, not -# the 29 h a 6 M floor cost. One seal writes ~40 GiB. +# after that; the trap brings frank-archon back. THE FLOOR IS +# THE CAMPAIGN'S RULE (12 M tokens, Frank 2026-10-30): the +# served iq4r this arm competes with was calibrated on a 10 M- +# token set, so the arm is calibrated at least as deeply. One +# round of the re-chunked corpus is ~9.31 M tokens, so sealing +# takes round 1 plus part of round 2 — a round that seals +# nothing is NORMAL (the floor is pooled), and the pusher's +# round_verdict treats a flat round as "continue", never as a +# failure. One seal writes ~40 GiB. # # Shape: the vendor's bf16 calibration lines — expert_tiered placement with the # weights disk tier, deterministic (the resident set pinned to the loader's plan, @@ -357,7 +361,7 @@ case "$MODE" in } MOUNTS+=(-v "$CALIB_DIR:/data/calib") TAPENV=(-e RADIANCE_CALIB_DIR=/data/calib - -e "RADIANCE_CALIB_TOKENS=${FRANK_QWEN38_CALIB_TOKENS:-1500000}") + -e "RADIANCE_CALIB_TOKENS=${FRANK_QWEN38_CALIB_TOKENS:-12000000}") # The tap's own authorization. A pusher must see this marker before it sends # a prompt (calib-push.py --tap-marker), it names the container so a push # aimed anywhere else refuses, and a boot that fails removes it again below. diff --git a/tests/qwen38_campaign.bats b/tests/qwen38_campaign.bats index fdaece6..f932c5b 100755 --- a/tests/qwen38_campaign.bats +++ b/tests/qwen38_campaign.bats @@ -692,7 +692,8 @@ window_fixture() { D="$(window_fixture)" run env FRANK_QWEN38_CAMPAIGN="$D/campaign" \ FRANK_QWEN38_RAD="$D/rad/qwen38-bf16.rad" \ - Q38_WINDOW_CALIB_CORPUS="$D/corpora/calib.jsonl" bash "$WINDOW" --dry-run + Q38_WINDOW_CALIB_CORPUS="$D/corpora/calib.jsonl" \ + Q38_CALIB_MIN_FREE_MIB=1 bash "$WINDOW" --dry-run [ "$status" -eq 0 ] [[ "$output" == *"run-qwen38-bf16.sh calib"* ]] [[ "$output" == *"else kill the rig"* ]] diff --git a/tests/qwen38_reference.bats b/tests/qwen38_reference.bats index 3e1ee86..7a21b73 100755 --- a/tests/qwen38_reference.bats +++ b/tests/qwen38_reference.bats @@ -157,11 +157,11 @@ flag_value() { [[ "$output" != *"--vram-kv-mib"* ]] } -@test "the calib dry run opens the tap at the 1.5M floor, on the latest pin, and serves on 7770" { +@test "the calib dry run opens the tap at the 12M floor, on the latest pin, and serves on 7770" { run bash "$RUNNER" calib --dry-run [ "$status" -eq 0 ] flat="$(oneline)" - [[ "$flat" == *"-e RADIANCE_CALIB_TOKENS=1500000"* ]] + [[ "$flat" == *"-e RADIANCE_CALIB_TOKENS=12000000"* ]] [[ "$flat" == *"-e RADIANCE_CALIB_DIR=/data/calib"* ]] # the tap producer is the campaign's own pin (latest radiance), not production's [[ "$flat" == *"radiance@sha256:69add58f"* ]]