docs: collapse qwen36 workaround narrative now that bundle is qwen35-stamped

Follow-up to 1c88b41, which deferred the deeper docs cleanup
("intentionally stale in this commit, follow-up will collapse").
This is that follow-up.

Touched three doc surfaces:

- README "Architecture" section: replaced the long
"qwen36-stamped, here's why upstream is the lagging side, here
are two one-liner workarounds, here's the manual rebadge
recipe, once-upstream-adds-qwen36-the-workaround-disappears"
block with a one-paragraph history note: bundle is qwen35,
loads directly, was qwen36 through v0.6.0, reverted in 964e418.
`scripts/rename_arch.py` example flipped from qwen36 -> qwen35
to qwen35 -> qwen36 to match its new "tool kept around for
future flips" framing.
- README Quick start Ollama option A: dropped the "fails today
with `unknown model architecture: 'qwen36'`" warning and the
`make heal-hf && ollama run ...` recovery line. A is now just
`ollama run hf.co/FoolDev/Thanatos-27B` — 17 GB Q4_K_M,
qwen35-stamped. B's comment lost the
"rebadges qwen36 -> qwen35" detail (irrelevant on a qwen35
bundle).
- examples/README.md Ollama setup: similar collapse — pull + run
is now the lead recipe with no intervening heal step. heal-hf
kept as a one-paragraph callout for anyone who pulled before
the 964e418 re-stamp.
- Modelfile preamble: shrunk from a 15-line qwen36-stamp +
rebadge-recipe block to a 6-line "bundle is qwen35, loads
directly, see README Architecture for history" note.

Net -39 lines across the three doc surfaces. No tooling change;
heal-hf, load-bundle, rename_arch.py all stay wired up exactly as
shipped.

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

CHANGELOG.md

@@ -31,16 +31,22 @@ and documentation**, not the underlying base model.
   stock Ollama. `make heal-hf` is mentioned as the recovery path
   for anyone who pulled the v0.6.0 bundle and still has the qwen36
   blob in their store.
+- README "Architecture" section, Quick start option A,
+  `examples/README.md` Ollama setup, and `Modelfile` preamble all
+  collapsed back to the pre-qwen36 single-path narrative. The
+  long "qwen36 is unloadable / here are the workarounds" prose
+  was replaced with a one-paragraph history note pointing at
+  commit 964e418 + the [CHANGELOG](CHANGELOG.md) entry. Net -39
+  lines across the three doc surfaces. `scripts/rename_arch.py`
+  example in Architecture flipped from qwen36 → qwen35 to qwen35
+  → qwen36 to match its new "tool kept around for future flips"
+  framing.
 
 ### Note
 - `make heal-hf`, `make load-bundle`, and `scripts/rename_arch.py`
   stay in the repo and stay useful. heal-hf is idempotent on qwen35
   (skips early); load-bundle does a no-op rebadge on a qwen35 bundle
-  and runs `ollama create` as normal. The deeper Architecture /
-  Quick start / examples-README narrative still talks about the
-  qwen36 stamp and the workaround dance; left intentionally stale
-  in this commit for follow-up cleanup, since the immediate user
-  pain (load 500 on fresh pull) is what the upload commit fixed.
+  and runs `ollama create` as normal.
 
 ### Added
 - `scripts/check.sh` now greps for the `VAR="$(cmd 2>/dev/null | filter)"`

Modelfile

@@ -7,21 +7,12 @@
 # image input, or wait for the fix. See the Vision section in README.md.
 #
 # This repo bundles a single GGUF: Thanatos-27B.Q4_K_M.gguf (~17 GB),
-# stamped `general.architecture: 'qwen36'`. The FROM line below points
-# at it, but no released llama.cpp / Ollama recognizes `qwen36` yet, so
-# `ollama create thanatos-27b -f Modelfile && ollama run thanatos-27b`
-# fails today with `unknown model architecture: 'qwen36'`. To load now,
-# rebadge the bundle to `qwen35` first:
-#
-#     python3 scripts/rename_arch.py --from-arch qwen36 --to-arch qwen35 \
-#         Thanatos-27B.Q4_K_M.gguf Thanatos-27B.Q4_K_M.qwen35.gguf
-#     # then temporarily point FROM at the qwen35 file, or use:
-#     #   echo "FROM $PWD/Thanatos-27B.Q4_K_M.qwen35.gguf" > /tmp/Modelfile.qwen35
-#     #   ollama create thanatos-27b -f /tmp/Modelfile.qwen35
-#
-# Once upstream adds the qwen36 arch entry the workaround disappears
-# and the FROM line below works as-is. See README "Architecture" for
-# the full story.
+# stamped `general.architecture: 'qwen35'` — the arch entry every
+# released llama.cpp / Ollama loads under for the Qwen 3.6 hybrid
+# SSM + attention family. `ollama create thanatos-27b -f Modelfile &&
+# ollama run thanatos-27b` loads it directly. See README
+# "Architecture" for the qwen35 -> qwen36 -> qwen35 history through
+# v0.6.0.
 #
 # For other quants (Q3_K_S, Q5_K_M, Q6_K, etc.), `make build QUANT=Q3_K_S`
 # downloads the chosen quant from unsloth/Qwen3.6-27B-GGUF and patches

README.md

@@ -176,65 +176,43 @@ If you want the safetensors for `transformers`, fetch them from [`Qwen/Qwen3.6-2
   current loader compatibility.
 - Multi-token prediction (MTP) head trained for speculative decoding
 
-**The bundled GGUF declares `general.architecture: 'qwen36'`** — the
-architecturally-honest stamp. Upstream `ggml-org/llama.cpp` and
-`ollama/ollama` currently only register the hybrid SSM + attention
-stack under `qwen35` / `qwen35moe`; no `qwen36` arch entry exists yet
-(reconfirmed 2026-05-19 against llama.cpp 389ff61 and Ollama 0.24.0).
-Neither repo has an open PR or tracking issue for a `qwen36` arch
-string, and the actual Qwen 3.6-family upstream weights already load
-under `qwen35moe` — so the bundle's `qwen36` stamp may stay
-ahead-of-upstream indefinitely rather than for a few releases.
-Consequence: **the bundle is unloadable on current stock loaders.**
-`ollama run hf.co/FoolDev/Thanatos-27B` and `llama-server -m ...` both
-fail with `error loading model architecture: unknown model
-architecture: 'qwen36'` (Ollama 0.24 surfaces it as a 500 wrapping a
-generic `unable to load model: <blob>` — check `journalctl --user -u
-ollama` for the underlying line).
-
-This is intentional. The bundle is the *correct* metadata; the
-loaders are the lagging side. The flip is reversible until upstream
-catches up.
-
-Two one-liners, depending on whether you've already typed
-`ollama run hf.co/FoolDev/Thanatos-27B` and gotten the 500:
-
-```bash
-# Fresh: build a local `thanatos-27b` tag from this repo's bundle
-# (LFS smudge + rebadge + `ollama create`, idempotent):
-make load-bundle
-ollama run thanatos-27b
-
-# Already pulled and hit the 500: rebadge the already-downloaded
-# blob in place + rewrite the manifest digest, so the same
-# `hf.co/...` tag becomes loadable without switching tag names:
-make heal-hf
-ollama run hf.co/FoolDev/Thanatos-27B
-```
-
-What that wraps, if you want to do it by hand
-(`scripts/rename_arch.py` is metadata-only — tensors stay
-byte-identical):
+**The bundled GGUF declares `general.architecture: 'qwen35'`** — the
+arch entry every released llama.cpp / Ollama loads the Qwen 3.6
+hybrid SSM + attention family under. Upstream `ggml-org/llama.cpp`
+and `ollama/ollama` register the stack as `qwen35` / `qwen35moe`,
+including for actual Qwen 3.6-family weights (the upstream
+`Qwen3.6-35B-A3B_Q8_0` GGUFs load under `qwen35moe`); no `qwen36`
+arch entry exists, and there's no open PR or tracking issue for
+one — `qwen35` is the canonical load-time label.
+
+`ollama run hf.co/FoolDev/Thanatos-27B` and `llama-server -m
+Thanatos-27B.Q4_K_M.gguf` both load directly on current stock
+loaders.
+
+**History.** Through v0.6.0 the bundle was stamped `qwen36` — the
+architecturally-honest label, but unloadable on stock loaders.
+Users needed `make load-bundle` or `make heal-hf` to rebadge
+before loading. The stamp was reverted to `qwen35` in commit
+`964e418` to stop forcing the workaround on every fresh pull;
+tensor data was byte-identical across both stamps, only the
+`general.architecture` KV (and namespaced KV keys) flipped. See
+the [CHANGELOG](CHANGELOG.md) entry for the full rationale.
+
+`scripts/rename_arch.py` is kept in the repo as a generic GGUF
+arch renamer (metadata only, tensors byte-identical) for future
+flips:
 
 ```bash
 python3 scripts/rename_arch.py \
-    --from-arch qwen36 --to-arch qwen35 \
+    --from-arch qwen35 --to-arch qwen36 \
     Thanatos-27B.Q4_K_M.gguf \
-    Thanatos-27B.Q4_K_M.qwen35.gguf
-# Then either build a local Ollama tag (note absolute path —
-# `ollama create` resolves a relative FROM against the Modelfile's
-# directory, not your CWD):
-echo "FROM $PWD/Thanatos-27B.Q4_K_M.qwen35.gguf" > /tmp/Modelfile.qwen35
-ollama create thanatos-27b -f /tmp/Modelfile.qwen35
-ollama run thanatos-27b hi
-# …or point llama-server at the qwen35 file directly:
-llama-server -m Thanatos-27B.Q4_K_M.qwen35.gguf -ngl 99 -c 8192
+    Thanatos-27B.Q4_K_M.qwen36.gguf
 ```
 
-If upstream ever adds a `qwen36` arch entry — patch landed in
-`ggml-org/llama.cpp` and propagated into Ollama — `make load-bundle`,
-`make heal-hf`, and the whole workaround can be deleted; the bundle
-works as-is.
+If you pulled the bundle while it was still qwen36-stamped (v0.6.0
+through 2026-05-19) and have the broken blob in your local Ollama
+store, `make heal-hf` rebadges it in place so the same
+`hf.co/...` tag becomes loadable without re-pulling.
 
 ## Quick start
 
@@ -243,17 +221,13 @@ works as-is.
 Three paths:
 
 ```bash
-# A. Pull straight from HF (uses the bundled Q4_K_M + root-level
-#    template / system / params files). Fails today with
-#    `unknown model architecture: 'qwen36'`. If you've already run
-#    this and got the 500, `make heal-hf` rebadges the pulled blob
-#    in place so the same command then loads. Or use B / C below
-#    to avoid the broken pull entirely.
-ollama run hf.co/FoolDev/Thanatos-27B           # 17 GB Q4_K_M, qwen36-stamped
-make heal-hf && ollama run hf.co/FoolDev/Thanatos-27B  # heal + re-run after the first fails
-
-# B. Load THIS repo's bundle via the rebadge helper (smudges LFS if
-#    needed, rebadges qwen36 → qwen35, runs `ollama create`):
+# A. Pull straight from HF (gets the bundled Q4_K_M GGUF + the
+#    root-level template / system / params files in one step):
+ollama run hf.co/FoolDev/Thanatos-27B           # 17 GB Q4_K_M, qwen35-stamped
+
+# B. Build a local `thanatos-27b` tag from THIS repo's bundle
+#    (LFS smudge if needed, then `ollama create`). Useful if you
+#    want a bare local tag rather than the `hf.co/...` path:
 make load-bundle                                 # creates local tag thanatos-27b
 ollama run thanatos-27b
 

examples/README.md

@@ -21,37 +21,33 @@ from the `Modelfile` / bridge files.
 
 Pull straight from HF (gets the bundled Q4_K_M GGUF + this repo's
 root-level `template` / `system` / `params` files via HF's Ollama
-bridge), then heal the bundle's `qwen36` arch stamp so Ollama can
-actually load it. The pull itself succeeds; the first inference
-fails with `unable to load model` until heal — see parent README's
-[Architecture](../README.md#architecture) for why:
+bridge):
 
 ```bash
 ollama pull hf.co/FoolDev/Thanatos-27B           # 17 GB Q4_K_M (only bundled quant)
-cd ..  &&  make heal-hf  &&  cd examples         # rebadges the pulled blob qwen36 -> qwen35 in store
 pip install requests
 MODEL=hf.co/FoolDev/Thanatos-27B python ollama_chat.py
 ```
 
-For the smaller-footprint Q3_K_S (~12 GB) or other quants, build
-locally instead — see the parent repo's `make build QUANT=...` flow.
+If you pulled before commit `964e418` (the qwen35 re-stamp) and
+still have the broken qwen36 blob in your Ollama store, run
+`cd .. && make heal-hf` once to rebadge it in place. Fresh pulls
+after the re-stamp go straight through.
 
-Or build locally from this repo without going through the HF pull
-(uses the bundled `Thanatos-27B.Q4_K_M.gguf` via `make load-bundle`,
-which handles the qwen36 → qwen35 rebadge in one shot and creates a
-local `thanatos-27b` tag):
+For a non-bundled quant (e.g. Q3_K_S ~12 GB, Q5_K_M ~20 GB),
+`make build QUANT=...` downloads from `unsloth/Qwen3.6-27B-GGUF`
+and creates a local `thanatos-27b` tag:
 
 ```bash
-cd ..  &&  make load-bundle  &&  cd examples
+cd ..  &&  make build QUANT=Q3_K_S  &&  cd examples
 MODEL=thanatos-27b python ollama_chat.py
 ```
 
-For a non-bundled quant (e.g. Q5_K_M, Q3_K_S), `make build QUANT=...`
-downloads a qwen35-stamped GGUF from `unsloth/Qwen3.6-27B-GGUF` and
-creates the same local tag — bypasses the bundle and the heal entirely:
+Or build a local tag from this repo's bundled GGUF without going
+through the HF pull:
 
 ```bash
-cd ..  &&  make build QUANT=Q5_K_M  &&  cd examples
+cd ..  &&  make load-bundle  &&  cd examples
 MODEL=thanatos-27b python ollama_chat.py
 ```