dank/serverAppConfigs
1
0
Fork 0
Code Issues 6 Pull Requests Actions Packages Projects Releases Wiki Activity

Commit Graph

  • e330d56a06 frank-primary runbook: record the 4-slot/V-q8/GPU-block reshape Frank 2026-09-26 21:32:09 -07:00
  • de85654b76 frank-primary: four full-window slots, V q8_0 KV, hot expert blocks on-GPU Frank 2026-09-26 21:29:21 -07:00
  • de4d22785d frank-primary runbook: record the KV law and per-ctx VRAM costs Frank 2026-09-26 21:01:12 -07:00
  • 82ff9dda20 frank-primary + frank-large: export /metrics, raise primary log verbosity to 4 Frank 2026-09-26 20:58:53 -07:00
  • bacabdcd49 frank-primary: record measured battery numbers and the drafter-placement law Frank 2026-09-26 20:55:02 -07:00
  • 5d632bbaea frank-primary: pin the drafter to ROCm1 (the last-layers card) Frank 2026-09-26 20:49:04 -07:00
  • a9b94a79eb frank-primary: pin the -ot override set to the real tensor names Frank 2026-09-26 19:51:57 -07:00
  • 8500600d09 frank-primary: swap Dankserver to llama.cpp Qwen3.8-Flash-Next Q8_0 Frank 2026-09-26 19:45:28 -07:00
  • d89813363e add frank-primary payload fetch (Qwen3.8-Flash-Next Q8_0 + MTP head + mmproj) Frank 2026-09-26 19:27:55 -07:00
  • d7bf9f119a retire frank-medium: tests, planning and observability ground on frank-primary Frank 2026-09-26 19:25:46 -07:00
  • 32d264ccda frank-primary: size the hot expert stack explicitly (LUCE_EXPERT_BUDGET_MB) Frank 2026-09-26 17:13:05 -07:00
  • dcdf044aa1 frank-primary: exact prefill, the only mode the streamed tier runs Frank 2026-09-26 17:11:31 -07:00
  • de513ab080 deploy-remote: stat the frank-primary payload, do not wc it Frank 2026-09-26 17:08:13 -07:00
  • 124175218f frank-primary: patch is Dockerfile-only on feat/ds41-engine Frank 2026-09-26 17:03:11 -07:00
  • 846140c2be frank-primary: V4.1 launch shape (feat/ds41-engine source) Frank 2026-09-26 17:00:46 -07:00
  • fc472e89d4 frank-primary: track the gfx1201 build patch (apt/cmake/ggml guard) + runbook recipe Frank 2026-09-26 15:44:17 -07:00
  • bed551baa6 deploy.sh: allowlist dankserver:frank-primary Frank 2026-09-26 15:33:05 -07:00
  • acd14ab7df frank-primary: Lucebox DeepSeek-V4.1-Flash-ROCMFP23 quadlet + deploy handler Frank 2026-09-26 15:29:59 -07:00
  • 441c4f1c99 hf-fetch: only pass -C - when a non-empty partial exists Frank 2026-09-26 15:23:24 -07:00
  • c5b55e1e3b frank-medium: GPU_UTIL 0.98 -> 0.97 — the 0.98 startup threshold sits within ~15 MiB of the free memory measured at engine init (worker HIP context + amdgpu per-boot baseline ~690 MiB), so 2026-09-26's boot crash-looped on every restart; 0.97 buys ~160 MiB headroom at ~5k KV tokens Frank 2026-09-26 15:02:20 -07:00
  • cd07a66939 homepage: flip the model card stagger to dark-light-dark-light Frank 2026-09-26 00:21:01 -07:00
  • d2c4c3f715 homepage: stagger the model card tiers light-dark-light-dark Frank 2026-09-26 00:18:48 -07:00
  • 6fae19e285 homepage: give the model inputs line its own spring-green tone Frank 2026-09-26 00:13:31 -07:00
  • ec7677c180 homepage: add per-model input modalities to the LLM cards Frank 2026-09-26 00:11:01 -07:00
  • 57fd42af6a homepage: rework LLM model cards to structured role/spec/use lines Frank 2026-09-25 23:51:41 -07:00
  • a489d8481d docs: log the homepage model-description change (2026-09-25) Frank 2026-09-25 22:57:39 -07:00
  • d3c819741d homepage: add suggested uses to the three LLM model descriptions Frank 2026-09-25 22:32:56 -07:00
  • 5df6fd1f97 docs: Jellyfin LDAP 401 resolved — legacy /Users/Auth/Password route removed in 12.1 Frank 2026-09-25 22:30:44 -07:00
  • 58952bd820 Bump authentik LDAP outpost to 2026.8.3 (Jellyfin auth path) Frank 2026-09-25 22:01:56 -07:00
  • 443dde71f3 Log: fleet-wide container update 2026-09-25 Frank 2026-09-25 17:17:17 -07:00
  • 0b1ea2efb3 Bump pinned images to latest: authentik 2026.8.3, tautulli ls245, calibre-web ls402, plex 1.43.4 Frank 2026-09-25 17:11:20 -07:00
  • 650dcba0ea docs(authentik): password change is a prompt field, not a PasswordStage Frank 2026-09-24 21:28:14 -07:00
  • 8108dc4254 docs(authentik): document the self-serve user-settings flow + password stage Frank 2026-09-24 21:16:10 -07:00
  • 201c8437dd docs(open-webui): record the SSO-only flip + qa-e2e-admin demotion Frank 2026-09-24 20:56:38 -07:00
  • 33cdeca41e open-webui: flip to SSO-only (disable local login form + password auth) Frank 2026-09-24 20:52:06 -07:00
  • 66a5e39899 homepage: add self-serve account link to authentik user-settings flow Frank 2026-09-24 20:43:11 -07:00
  • e452aa42c8 test(open-webui): match the OAUTH_SCOPES quoting fix Frank 2026-09-24 20:39:28 -07:00
  • df00182d0e docs(jellyfin): document the authentik LDAP integration Frank 2026-09-24 20:39:28 -07:00
  • 35381f7e9d open-webui: quote OAUTH_SCOPES in the quadlet Frank 2026-09-24 20:21:10 -07:00
  • b1372d428f e2e: harden akLogin against the Lit stage re-render race Frank 2026-09-24 20:06:57 -07:00
  • b48de913b8 open-webui: SSO via authentik (OIDC authorization-code flow) Frank 2026-09-24 20:04:17 -07:00
  • 6ffcde0197 docs: log the fact-check pass + 3 research briefs (2026-09-22) Frank 2026-09-22 02:10:04 -07:00
  • 93d1181fee docs: OKF v0.1 compliance pass + fact-check + 3 research briefs Frank 2026-09-22 02:09:41 -07:00
  • 24f7a7772d frank-medium: record measured MXFP6 boot shape (KV pool, drafter acceptance) Frank 2026-09-22 00:44:52 -07:00
  • d5c8e408b3 frank-medium: revert GPU_UTIL to 0.98 (0.99 is infeasible on 32 GB R9700) Frank 2026-09-22 00:15:06 -07:00
  • e9a01cd013 frank-medium: swap to ParoQuant MXFP6 (Swift Qwen3.8-27B) + Qwen-Sharp v22.5.0 Frank 2026-09-21 22:46:44 -07:00
  • 600812d459 docs: update frank-large to stock llama.cpp UD-Q8_K_XL + DFlash2 FrankandClaude 2026-09-21 21:53:53 -07:00
  • 7689092db8 frank-large: swap to stock llama.cpp serving Qwen3.8-27B UD-Q8_K_XL with DFlash2 FrankandClaude 2026-09-21 21:50:36 -07:00
  • 43b8258061 comfyui: fix in_channels to use shape[0] (latent input dim) Frank 2026-09-21 20:31:39 -07:00
  • 5fe57f42fa comfyui: add VAE 4D-to-5D latent expansion to patch script Frank 2026-09-21 20:20:18 -07:00
  • f4b912749d comfyui: VAE 4D-to-5D latent expansion for video VAEs Frank 2026-09-21 20:19:57 -07:00
  • be878008a2 homepage: update LLM descriptions for model changes Frank 2026-09-21 20:19:05 -07:00
  • 97f60b35ef docs: document Qwen-Sharp template wiring and 73k-token diagnosis Frank 2026-09-21 20:11:00 -07:00
  • 293e4e4b09 frank-large: wire Qwen-Sharp v22.4.0 template via --chat-template-file Frank 2026-09-21 20:07:13 -07:00
  • b3a3b936bd comfyui: fix COPY paths to include dankclaw/ prefix Frank 2026-09-21 19:55:46 -07:00
  • 4b443c86d8 comfyui: bake qwen_image21 patch into image via startup wrapper Frank 2026-09-21 19:55:00 -07:00
  • e04f510883 docs: frank-large Kairic Edge runbook + frank-medium clean-KV updates Frank 2026-09-21 19:16:51 -07:00
  • b2a74004df frank-large: swap to Kairic Edge 27B IU4 (W4A4) on Strix Halo Frank 2026-09-21 18:56:35 -07:00
  • 2f7aadab88 homepage: add the Hermes dashboard (agent management UI) card Frank 2026-09-21 17:37:11 -07:00
  • af36595117 homepage: link the Pi-hole admin and the Hermes web UI Frank 2026-09-21 17:30:48 -07:00
  • d8dbd4f5bc docs(frank-medium): record the MXFP4 + clean-KV cut; correct vision claim Frank 2026-09-21 17:08:33 -07:00
  • a45f396c39 feat(frank-medium): cut to MXFP4 checkpoint with unquantized bf16 KV Frank 2026-09-21 16:32:19 -07:00
  • 0f203dfa41 docs(open-webui): document the 3-model LLM picker and host.containers.internal routing Frank 2026-09-21 16:30:18 -07:00
  • d63acd489c open-webui: make the SearXNG web-search wiring git-managed Frank 2026-09-21 15:22:49 -07:00
  • 9e0c918b6e docs(frank-small): record the CACHE_RAM=2048 cap in the deployed flags Frank 2026-09-21 15:06:05 -07:00
  • 2b046a21d0 homepage: AI section to the bottom; model-only AI descriptions Frank 2026-09-21 15:02:14 -07:00
  • e23df8b229 perf(frank-small): cap the CPU RAM cache at 2048 MiB Frank 2026-09-21 14:57:32 -07:00
  • e22b43d589 homepage: sections are horizontal rows, not columns Frank 2026-09-21 14:30:21 -07:00
  • 8ea6486dfc homepage: revalidate assets so deploys never serve stale CSS/JS Frank 2026-09-21 14:12:38 -07:00
  • 5d774752ff docs(log): frank-large cutover verified live (MTP drafting, tools, e2e 32/32) Frank 2026-09-21 14:08:59 -07:00
  • 6a7ccc31f1 test(homepage): move the SECTION_ORDER assertion inside its test block Frank 2026-09-21 14:00:18 -07:00
  • d6f9b8539a fix(frank-large): load the model as its native 3-way split, not a concatenated file Frank 2026-09-21 13:58:54 -07:00
  • 384747a72f homepage: sectioned panels, WIP open/closed, current model facts, Tautulli card Frank 2026-09-21 13:20:15 -07:00
  • 32eafeb7c5 docs(log): Calibre Content Server deployment entry Frank 2026-09-21 12:53:54 -07:00
  • 22204b49f5 docs(calibre-server): correct the port map - pasta is port-forward mode Frank 2026-09-21 12:51:06 -07:00
  • cb96525f78 fix(deploy): add calibre-server to deploy.sh's service allowlist Frank 2026-09-21 12:47:04 -07:00
  • 94cfb90ead feat(calibre-server): Calibre Content Server as a boot-persistent Quadlet Frank 2026-09-21 12:46:02 -07:00
  • 660db661b6 fix(frank-large): verify exactly the fetched file set, not the whole Hub tree Frank 2026-09-21 12:41:52 -07:00
  • 7010f5b754 docs(frank-large): unsloth Qwen3.8-Flash-Next runbook; retire the CIRU Orca runbook Frank 2026-09-21 12:30:35 -07:00
  • b149d02187 feat(frank-large): unsloth llama.cpp Qwen3.8-Flash-Next UD-IQ4_XS (200k/1 slot, MTP depth 2) Frank 2026-09-21 12:30:30 -07:00
  • e5a1f8e767 docs: work-log entry for the node-dankclaw scrape fix Frank 2026-09-21 11:22:02 -07:00
  • 3b28b2afb7 docs(frank-medium): record the Qwen-Sharp v22.4.0 chat template Frank 2026-09-21 11:18:58 -07:00
  • 6bccff3cd2 feat(frank-medium): serve the Qwen-Sharp v22.4.0 chat template (froggeric) Frank 2026-09-21 11:11:33 -07:00
  • 5c1006dda6 fix(monitoring): scrape node-dankclaw via host.containers.internal Frank 2026-09-21 11:09:03 -07:00
  • f998a2d9f2 test(frank-medium): A/B DFlash2 vs MTP on FP8 — DFlash2 wins on decode Frank 2026-09-20 22:20:53 -07:00
  • 1ae73db084 test(frank-medium): A/B — flip to DFlash2 drafter (SPEC_METHOD=dflash, SPEC=7) Frank 2026-09-20 22:06:26 -07:00
  • c1a4ccbbcd feat(frank-medium): swap MXFP4 -> FP8; add the one-command swap toolchain Frank 2026-09-20 21:53:00 -07:00
  • 2458d754ce fix(deploy): strip the Environment=VAR= prefix, not just the first = Frank 2026-09-20 21:31:16 -07:00
  • d389d09d97 feat(frank-medium): quantization-aware entrypoint + FP8 model swap Frank 2026-09-20 21:30:30 -07:00
  • c42780f9f3 fix(monitoring): llama.cpp reuse ratio, dashboard re-cut for both namespaces, measured prefix-cache granularity Frank 2026-09-20 19:53:14 -07:00
  • 08f1c7ad09 feat(monitoring,harness): follow frank-medium onto vllm:* metrics and fix red-by-construction targets Frank 2026-09-20 19:35:56 -07:00
  • c2116658f8 feat(frank-medium): cut the endpoint over to radiance TP=2 Frank 2026-09-20 18:13:33 -07:00
  • bcdd898750 test(bench): drop the pre-probe from the conversation battery Frank 2026-09-20 17:51:27 -07:00
  • b8d1ab42ba add conversation-token battery and eno1 route preference Frank 2026-09-20 15:25:09 -07:00
  • eb2f1bb6bc build-llama-rdna4: bundle librocm-core Frank 2026-09-20 15:25:09 -07:00
  • 0425063ffb build-llama-rdna4: make GGML_HIP_RCCL overridable and bundle librccl Frank 2026-09-20 15:17:02 -07:00
  • d16e8e554e docs: dankserver board swap clears the PCIe-atomics wall Frank 2026-09-20 14:39:54 -07:00
  • a43ab90105 docs: work-log entry for the 2026-09-19 fleet cutover Frank 2026-09-19 06:42:37 -07:00
  • 88331e2df1 docs: frank-medium deployed-flags table + halogen retirement banner Frank 2026-09-19 06:20:57 -07:00
  • 764cf41222 docs: wrap up the 2026-09-19 fleet cutover Frank 2026-09-19 06:12:33 -07:00