Frank e9a01cd013 frank-medium: swap to ParoQuant MXFP6 (Swift Qwen3.8-27B) + Qwen-Sharp v22.5.0 ...

- Model: hugypufy/Swift-Qwen3.8-27B-PARO-MXFP6 (25.16 GiB, 18 shards,
  quant_method=paroquant_mxfp6, bits=6, group_size=128, krot=8, rotation on)
- Image: ggz14/vllm-radiance-mxfp4:0.13.0-e060895 (PR #51 launch6_p kernel,
  built on the published 0.9.3 base; docker->podman via save/load)
- GPU_UTIL 0.98 -> 0.99 (user request); KV_DTYPE=auto (unquantized bf16)
- Entrypoint: mxfp6 branch (17 patches + paroquant kernel compile +
  sitecustomize registration); GPU_UTIL default for set -u robustness
- Quadlet: SNAP/GPU_UTIL/Image/cache updated; new mxfp6 cache dir
- deploy-remote.sh: R4D gate + paroquant_mxfp6; new image/SNAP/cache defaults
- Chat template: Qwen-Sharp v22.5.0 (peculiar-ragdoll); v22.4.0 kept as rollback
- Test 168: dflash branch comment quant-aware (mxfp4 + paroquant_mxfp6)
- Runbook: MXFP6 active shape, MXFP4 rollback

MXFP4 checkpoint remains on disk as the rollback target.

2026-09-21 22:46:44 -07:00

..

.AGENTTRASH

frank-medium: llama.cpp layer split across both R9700s, DFlash2 drafting

2026-09-16 14:59:23 -07:00

e2e

tests: gate the hermes vhost, and stop specs from silently not running

2026-09-16 07:33:38 -07:00

anubis.bats

anubis: fix image ref typo in deploy pull, health probe exec form, unit alias tag

2026-08-28 11:18:41 -07:00

authentik.bats

docs+units: fix every claim my own changes made false

2026-09-16 07:46:23 -07:00

caddy_quadlet.bats

docs: rename docs/ -> serverAppConfigs-docs/ per system-wide PROJECTNAME-docs convention

2026-08-28 02:20:06 -07:00

caddyfile.bats

docs+units: fix every claim my own changes made false

2026-09-16 07:46:23 -07:00

changed_tests.bats

feat(frank-medium): cut the endpoint over to radiance TP=2

2026-09-20 18:15:27 -07:00

comfyui_rocm.bats

comfyui: bake qwen_image21 patch into image via startup wrapper

2026-09-21 19:55:00 -07:00

compose_validity.bats

feat: initialize serverAppConfigs with podman stacks for Dankserver and Dankclaw

2026-08-13 20:19:41 -07:00

dankserver_quadlet.bats

feat(calibre-server): Calibre Content Server as a boot-persistent Quadlet

2026-09-21 12:46:02 -07:00

deployment_flow.bats

deploy-remote: prune stale pending state for retired services

2026-09-17 23:32:02 -07:00

deployment_scoping.bats

feat(frank-medium): cut the endpoint over to radiance TP=2

2026-09-20 18:15:27 -07:00

docs_completeness.bats

docs: three dangling links, and a gate so they cannot accumulate

2026-09-16 07:43:33 -07:00

e2e_scaffold.bats

tests: gate the hermes vhost, and stop specs from silently not running

2026-09-16 07:33:38 -07:00

env_consistency.bats

feat: initialize serverAppConfigs with podman stacks for Dankserver and Dankclaw

2026-08-13 20:19:41 -07:00

frank_large_unit.bats

frank-large: swap to stock llama.cpp serving Qwen3.8-27B UD-Q8_K_XL with DFlash2

2026-09-21 21:50:36 -07:00

frank_medium_unit.bats

frank-medium: swap to ParoQuant MXFP6 (Swift Qwen3.8-27B) + Qwen-Sharp v22.5.0

2026-09-21 22:46:44 -07:00

frank_small_unit.bats

perf(frank-small): cap the CPU RAM cache at 2048 MiB

2026-09-21 14:57:32 -07:00

frankbook_quadlet.bats

arr fleet: FrankBook-Pro is the sole deployment, all egress VPN-gated

2026-09-04 12:54:28 -07:00

helpers.bash

tests: LLM unit integrity, sudo-password leak guard, deploy LLM coverage

2026-08-22 15:49:44 -07:00

hermes_webui.bats

hermes vhost: dial netavark's host address, not the LAN IP (fixes the 502)

2026-09-16 19:28:34 -07:00

homepage_caddy.bats

homepage: revalidate assets so deploys never serve stale CSS/JS

2026-09-21 14:12:38 -07:00

homepage_checker.bats

homepage: sectioned panels, WIP open/closed, current model facts, Tautulli card

2026-09-21 13:20:15 -07:00

homepage_config.bats

homepage: sectioned panels, WIP open/closed, current model facts, Tautulli card

2026-09-21 13:20:15 -07:00

homepage_deploy.bats

Deploy services independently by changed paths

2026-08-25 20:13:33 -07:00

homepage_pages.bats

homepage: AI section to the bottom; model-only AI descriptions

2026-09-21 15:02:14 -07:00

llm_secrets.bats

tests: LLM unit integrity, sudo-password leak guard, deploy LLM coverage

2026-08-22 15:49:44 -07:00

llm_units.bats

feat(llm): rename inference services dropping 'default' from names

2026-09-03 16:15:43 -07:00

marreta_quadlet.bats

marreta: content-cleaning proxy on Dankclaw, gluetun netns, guard-chained

2026-09-12 10:09:12 -07:00

mount_entropy.bats

Add CLI helper for entropy volume

2026-08-25 15:38:57 -07:00

observability_quadlet.bats

fix(monitoring): scrape node-dankclaw via host.containers.internal

2026-09-21 11:09:03 -07:00

open_webui_quadlet.bats

feat(llm): rename inference services dropping 'default' from names

2026-09-03 16:15:43 -07:00

pihole.bats

pihole: also neutralize the /etc/sysctl.conf sysctl override

2026-08-26 20:08:30 -07:00

run.sh

feat: initialize serverAppConfigs with podman stacks for Dankserver and Dankclaw

2026-08-13 20:19:41 -07:00

scripts_syntax.bats

bench: commit the harnesses the frank-medium numbers came from

2026-09-16 06:59:44 -07:00

searxng_limiter.bats

Deploy services independently by changed paths

2026-08-25 20:13:33 -07:00

searxng_quadlet.bats

SearXNG quadlet: use AddDevice= (the generator rejects Device=)

2026-08-19 00:00:42 -07:00

seerr_quadlet.bats

seerr: quadlet on dankclaw + guard-chain vhost + deployer wiring

2026-09-01 14:40:10 -07:00