@miller-tech/uap 2.44.2 → 2.48.0

package/config/model-registry.json

@@ -95,6 +95,70 @@
             }
           ]
         },
+        "strata-abliterated-iq3_s": {
+          "status": "unmeasured",
+          "engine": "strata",
+          "unit": "uap-strata-abliterated-iq3_s-server",
+          "note": "The DavidAU abliterated Flash-Next IQ3_S tier — the :8080 default 2026-10-23 -> 2026-10-28 (adopted over the Swift 1.5 arm on the same-day A/B), moved to the IQ2_XS tier in the 2026-10-28 kernel-OOM fix (full-pin 53.7 GB anon RSS; the low-RAM --resident-budget-gib 20 posture ships in strata-abliterated-iq3_s.json, and engine 0.1.59.1 supports the flag), re-staged 2026-11-14 as a switchable panel arm. Config strata-abliterated-iq3_s.json, unit uap-strata-abliterated-iq3_s-server.service, pack strata-data/packs/abliterated-iq3_s; advertises abliterated-iq3_s at /v1/models; switch: ~/.config/uap/model-switch.sh strata-abliterated-iq3_s (revert: strata | strata-abliterated). Run `uap models measure` against it before placement loads it (fail-closed). Raw GGUF shards declared here: the pack is derived from them by tools/iq_pack.py and rebuilds from these files; shard 2 is byte-identical across the release's tiers (the tier-shared digest also pinned in strata-abliterated-iq2_xs). Provenance is pinned (repo + rev, from feature/386): bytes + sha256 are the HF LFS oids, verified against the local pair at the 2026-10-23 adoption and re-verified against the repo digests 2026-11-14, so `uap models ensure` can re-fetch and verify from this pin.",
+          "files": [
+            {
+              "path": "/home/cogtek/models/strata-data/models/abliterated-IQ3_S/Qwen3.8-Flash-Next-GSQ-RCO-abliterated-IQ3_S-00001-of-00002.gguf",
+              "repo": "DavidAU/Qwen3.8-Flash-Next-Abliterated-Uncensored-Turbo-Brilliance-50x-Reasoning-Instruct-GSQ-RCO-GGUF",
+              "rev": "623b419bd443e6e76ec51ca4b95d951da93c6969",
+              "file": "Qwen3.8-Flash-Next-GSQ-RCO-abliterated-IQ3_S-00001-of-00002.gguf",
+              "bytes": 55063664928,
+              "sha256": "78706315c2fe912777c006b4ae18be1024c9f4353bd25d33b8a306c9c55f9e34"
+            },
+            {
+              "path": "/home/cogtek/models/strata-data/models/abliterated-IQ3_S/Qwen3.8-Flash-Next-GSQ-RCO-abliterated-IQ3_S-00002-of-00002.gguf",
+              "repo": "DavidAU/Qwen3.8-Flash-Next-Abliterated-Uncensored-Turbo-Brilliance-50x-Reasoning-Instruct-GSQ-RCO-GGUF",
+              "rev": "623b419bd443e6e76ec51ca4b95d951da93c6969",
+              "file": "Qwen3.8-Flash-Next-GSQ-RCO-abliterated-IQ3_S-00002-of-00002.gguf",
+              "bytes": 28800138432,
+              "sha256": "316b46f3a2dbd68c900f43136ab9449f9dcc3725dfd8c794847c204bc161e113"
+            }
+          ]
+        },
+        "strata-unsloth-ud-q3_k_xl": {
+          "status": "unmeasured",
+          "engine": "strata",
+          "unit": "uap-strata-unsloth-q3_k_xl-server",
+          "note": "The unsloth IQ3_S-CLASS arm — user-directed 2026-11-14: unsloth/Qwen3.8-Flash-Next-GGUF ships NO IQ3_S tier; its 3-bit ladder is UD-IQ3_XXS (3.0 bpw, 83.6 GB) and UD-Q3_K_XL (3.5 bpw, 89.5 GB), and UD-Q3_K_XL is the same bpw class as the fleet's IQ3_S arms — the tier staged here. Set: the 4-shard UD-Q3_K_XL release at rev 39e6e4a8 (Better UD 3.1 quants); shards 1-3 downloaded (10.9 MB + 49.7 GB + 10.9 GB, sha256 = the HF LFS oids); shard 4 is NOT the set's native 28.8 GB IQ4_NL engram but the Q8_0 engram (engram/Qwen3.8-Flash-Next-engram-Q8_0.gguf_file, 54.4 GB, upstream oid e3d76a27...) restamped split.no=3/count=4 via unsloth's set_split.py and HARDLINKED from the v31 set's shard 4 — zero extra disk, the measured-best table (+19-22% long-prompt decode vs IQ4_NL, the 11-13 swap round); the sha256 below is the LOCAL post-restamp hash (what the .done marker verifies), same as the v31 set's shard-4 marker. Config strata-unsloth-ud-q3_k_xl.json — the v31 posture flag-for-flag (spec 4 window auto-negotiates 12/4, spec-min-p 0.5, --mtp mtp/rt, int8 KV resident 32768, 131k ctx, --vision shared mmproj, vram-reserve 3072) with --resident-budget-gib 40 (the Q3_K_XL expert set is smaller than v31's — more headroom; never full-pin, see the 10-28 OOM doctrine in uap-strata-abliterated-server.service). Pack packs/unsloth-ud-q3_k_xl (tools/iq_pack.py, from the set). Unit uap-strata-unsloth-q3_k_xl-server.service (OOMPolicy=continue doctrine), advertises qwen3.8-flash-next-unsloth-ud-q3_k_xl at /v1/models. Switch: model-switch.sh strata-unsloth-q3_k_xl (revert: strata or any arm). Run `uap models measure` against it before placement loads it (fail-closed).",
+          "files": [
+            {
+              "path": "/home/cogtek/models/strata-data/models/unsloth-UD-Q3_K_XL/Qwen3.8-Flash-Next-UD-Q3_K_XL-00001-of-00004.gguf",
+              "repo": "unsloth/Qwen3.8-Flash-Next-GGUF",
+              "rev": "39e6e4a86cc6517a6f39ef37adce5e9f3a81280c",
+              "file": "UD-Q3_K_XL/Qwen3.8-Flash-Next-UD-Q3_K_XL-00001-of-00004.gguf",
+              "bytes": 10946618,
+              "sha256": "7666000ba91eb5527f8e12e0414de36766dc82a082876337f7827c2caa859143"
+            },
+            {
+              "path": "/home/cogtek/models/strata-data/models/unsloth-UD-Q3_K_XL/Qwen3.8-Flash-Next-UD-Q3_K_XL-00002-of-00004.gguf",
+              "repo": "unsloth/Qwen3.8-Flash-Next-GGUF",
+              "rev": "39e6e4a86cc6517a6f39ef37adce5e9f3a81280c",
+              "file": "UD-Q3_K_XL/Qwen3.8-Flash-Next-UD-Q3_K_XL-00002-of-00004.gguf",
+              "bytes": 49742117248,
+              "sha256": "e954cb3044ae087a003ce8cc1631c9d40fe15ae64a8befcf4aa06457848c647c"
+            },
+            {
+              "path": "/home/cogtek/models/strata-data/models/unsloth-UD-Q3_K_XL/Qwen3.8-Flash-Next-UD-Q3_K_XL-00003-of-00004.gguf",
+              "repo": "unsloth/Qwen3.8-Flash-Next-GGUF",
+              "rev": "39e6e4a86cc6517a6f39ef37adce5e9f3a81280c",
+              "file": "UD-Q3_K_XL/Qwen3.8-Flash-Next-UD-Q3_K_XL-00003-of-00004.gguf",
+              "bytes": 10911746208,
+              "sha256": "fcce26b1e5e6e54559a7e7c2f8d6e37067ec695ff1f14bce662810bde78abe1f"
+            },
+            {
+              "path": "/home/cogtek/models/strata-data/models/unsloth-UD-Q3_K_XL/Qwen3.8-Flash-Next-UD-Q3_K_XL-00004-of-00004.gguf",
+              "repo": "unsloth/Qwen3.8-Flash-Next-GGUF",
+              "rev": "39e6e4a86cc6517a6f39ef37adce5e9f3a81280c",
+              "file": "engram/Qwen3.8-Flash-Next-engram-Q8_0.gguf_file",
+              "bytes": 54400261312,
+              "sha256": "4375b925a3abfb3859f6b634dafaab3343d9e2cdf9b3452253be465a12361426"
+            }
+          ]
+      },
         "strata-unsloth-ud-iq4_xs-v31": {
           "status": "unmeasured",
           "engine": "strata",
@@ -138,7 +202,9 @@
       },
       "advertises": [
         "qwen3.8-flash-next-unsloth-ud-iq4_xs-v31",
+        "qwen3.8-flash-next-unsloth-ud-q3_k_xl",
         "abliterated-iq2_xs",
+        "abliterated-iq3_s",
         "qwen3.8-flash-next-iq3_s",
         "swift-1.5-iq3_s",
         "gpt-5.1-codex"