Add HF Ollama bridge files (template/system/params) + fix mmproj filename collision

The HF Ollama bridge does NOT read Modelfile (per docs at
https://huggingface.co/docs/hub/en/ollama). When users do
'ollama run hf.co/FoolDev/janus-27b' the bridge generates a manifest
from three root-level files: 'template' (Ollama Go format), 'system'
(plain text), and 'params' (JSON). Without those files HF auto-converts
the GGUF's embedded jinja chat template to Ollama Go format, and that
conversion is buggy: produces '{{ if .Prompt }} .Prompt }}<|im_end|>'
(missing user-role wrapper, malformed value substitution), corrupted
stop tokens including the literal string '.Prompt }}<|im_end|>', and
no .Tools/.ToolCalls blocks — so 'ollama show hf.co/FoolDev/janus-27b'
reports only the 'completion' capability and rejects any /api/chat or
/v1/chat/completions request carrying a tools array.

Added template, system, params at repo root (mirrors the Modelfile's
TEMPLATE/SYSTEM/PARAMETER directives). Both routes now wire .Tools
and tool calling works end-to-end on either path.

Also renamed scripts/fetch_mmproj.sh -> scripts/fetch_vision.sh: HF's
Ollama bridge was filename-pattern-matching mmproj* anywhere in the
repo and shipping the 2028-byte bash script as the
application/vnd.ollama.image.projector layer. When Ollama tried to
load that 'projector' as a GGUF it failed the magic-bytes check —
'Error: invalid file magic' on every ollama show / ollama run.
Renaming breaks the pattern match; projector layer drops from the
manifest. Updated Makefile mmproj target and README references to
point at the new name.

README updated: 'What's here' table lists template/system/params;
'Quick start' / 'Local apps' / 'Chat template' sections corrected to
say HF bridge uses the three files (not Modelfile).

CHANGELOG.md

@@ -7,6 +7,34 @@ and documentation**, not the underlying base model.
 
 ## [Unreleased]
 
+### Added
+- Root-level `template`, `system`, and `params` files for HF's Ollama
+  bridge. The bridge generates Ollama manifests at request time from
+  these three files (NOT from `Modelfile` — confirmed against
+  https://huggingface.co/docs/hub/en/ollama). Without them, `ollama
+  run hf.co/FoolDev/janus-27b` got an auto-generated manifest with
+  the broken `{{ if .Prompt }} .Prompt }}<|im_end|>` template
+  (Ollama's faulty Go-template conversion of the GGUF's embedded
+  jinja), corrupted stop tokens (`".Prompt }}<|im_end|>"` bleed),
+  and no `.Tools` / `.ToolCalls` blocks — so the published Ollama
+  tag advertised `completion` only, rejected any request with a
+  `tools` array, and was actually broken to load (see "Fixed" below
+  re: the projector layer). The three files mirror the `Modelfile`'s
+  `TEMPLATE` / `SYSTEM` / `PARAMETER` directives; both routes wire
+  tool calling correctly. Edit them together when changing one.
+
+### Fixed
+- Renamed `scripts/fetch_mmproj.sh` → `scripts/fetch_vision.sh`. HF's
+  Ollama bridge was filename-pattern-matching `mmproj*` anywhere in
+  the repo and shipping `scripts/fetch_mmproj.sh` (a 2028-byte bash
+  script) as the `application/vnd.ollama.image.projector` layer. When
+  Ollama tried to load that "projector" as a GGUF, it failed the
+  magic-bytes check and `ollama show` / `ollama run` produced
+  `Error: invalid file magic`. Renaming the script breaks the pattern
+  match and the projector layer is no longer added to the manifest.
+  Updated `Makefile` (`mmproj` target) and README references to point
+  at the new name.
+
 ### Changed
 - README "Tool / function calling" section: split into explicit
   Ollama-path and embedded-jinja-path subsections. The two loader

Makefile

@@ -46,7 +46,7 @@ bench:  ## Measure tok/s using Ollama's eval timing (3 prompts).
 	MODEL=$(MODEL) ./scripts/bench.sh
 
 mmproj:  ## Fetch the vision projector for llama.cpp (Ollama vision is broken upstream).
-	./scripts/fetch_mmproj.sh $(PRECISION)
+	./scripts/fetch_vision.sh $(PRECISION)
 
 check:  ## Lint shell + python files; block dot-pattern footgun.
 	./scripts/check.sh

README.md

@@ -108,12 +108,13 @@ The 27B is **dense**: every parameter participates in every forward pass. It's s
 | File | Use |
 |---|---|
 | `banner.svg` / `banner.png` | Repo header, Tokyo Night themed |
-| `Modelfile` | Ollama wrapper around the upstream Qwen 3.6 27B GGUF (Q4_K_M) |
+| `Modelfile` | Ollama wrapper around the bundled Qwen 3.6 27B GGUF — used by `make build` / `ollama create` for **local** builds |
+| `template`, `system`, `params` | Used by HF's Ollama bridge when users `ollama run hf.co/FoolDev/janus-27b` directly (the bridge does **not** read `Modelfile` — see [HF Ollama docs](https://huggingface.co/docs/hub/en/ollama)). Mirrors the `Modelfile`'s template / system prompt / sampling params. |
 | `examples/` | Ready-to-run Python clients for Ollama, Transformers, and llama-cpp-python |
 | `scripts/build.sh` | One-shot helper: pulls a GGUF and runs `ollama create` for you |
 | `scripts/smoke_test.sh` | Verifies an Ollama daemon + model, runs a round-trip, and asserts no chat-template tokens leak into the response |
 | `scripts/bench.sh` | Measures real tok/s using Ollama's `eval_count` / `eval_duration` metadata over a 3-prompt mix (run `make bench`) |
-| `scripts/fetch_mmproj.sh` | Pulls the vision projector for llama.cpp (Ollama vision is broken upstream — see [Vision](#vision)) |
+| `scripts/fetch_vision.sh` | Pulls the vision projector (`mmproj-F16.gguf`) for llama.cpp (Ollama vision is broken upstream — see [Vision](#vision)). Renamed from `fetch_mmproj.sh` because HF's Ollama bridge auto-indexed the script as a vision projector layer (filename pattern match). |
 | `scripts/check.sh` | Local lint: `bash -n`, `pyflakes`, `py_compile`, footgun-grep |
 | `scripts/install-hooks.sh` | Installs `check.sh` as a git pre-commit hook |
 | `Makefile` | Convenience wrapper — `make help` lists targets |
@@ -133,8 +134,9 @@ ollama run hf.co/FoolDev/janus-27b:Q3_K_S    # tighter quant
 
 For other quants or local builds, pull from
 [`unsloth/Qwen3.6-27B-GGUF`](https://huggingface.co/unsloth/Qwen3.6-27B-GGUF)
-and `make build QUANT=...` — the Modelfile here is the same one Ollama
-applies in either path.
+and `make build QUANT=...`. The local-build path applies this repo's
+`Modelfile`; the `hf.co/...` path applies the root-level `template`,
+`system`, and `params` files (kept in sync with the `Modelfile`).
 
 If you want the safetensors for `transformers`, fetch them from [`Qwen/Qwen3.6-27B`](https://huggingface.co/Qwen/Qwen3.6-27B).
 
@@ -193,7 +195,7 @@ local app — point it at this repo and pick a quant.
 
 | App | How to load this model |
 |---|---|
-| **Ollama** | `ollama run hf.co/FoolDev/janus-27b` (or `:Q3_K_S`). Pulls the GGUF + Modelfile (TEMPLATE, sampling, stop tokens, tool calling) in one step. |
+| **Ollama** | `ollama run hf.co/FoolDev/janus-27b` (or `:Q3_K_S`). Pulls the GGUF + the root-level `template` / `system` / `params` files in one step (HF's Ollama bridge ingests these three files; it does **not** read `Modelfile`). For local builds, `make build` uses `Modelfile`, which is kept in sync. |
 | **LM Studio** | Search → `FoolDev/janus-27b` → pick `Janus-27B.Q4_K_M.gguf` or `Janus-27B.Q3_K_S.gguf`. Uses the GGUF's embedded jinja chat template (Qwen 3.6 ChatML); set the system prompt manually from the `SYSTEM` block in this repo's `Modelfile`. |
 | **Jan** | Hub → "Import from Hugging Face" → `FoolDev/janus-27b`. Same template behavior as LM Studio. |
 | **llama.cpp** | `hf download FoolDev/janus-27b Janus-27B.Q4_K_M.gguf --local-dir .` then `llama-server -m Janus-27B.Q4_K_M.gguf` (or `llama-cli`, `llama-mtmd-cli` for vision via the upstream `mmproj-F16.gguf`). |
@@ -201,10 +203,12 @@ local app — point it at this repo and pick a quant.
 | **Open WebUI / KoboldCpp / text-generation-webui** | Standard llama.cpp loader path — point at the GGUF, use the embedded chat template. |
 
 For the full Vision (image input) loader matrix, see [Vision](#vision).
-Tool calling currently works in **Ollama** (via this repo's Modelfile
-TEMPLATE) and **llama.cpp / llama-cpp-python** (via the GGUF's embedded
-jinja). Other apps' tool-calling support depends on whether they read
-the embedded template or require an external schema.
+Tool calling currently works in **Ollama** (via the root-level
+`template` file when pulling from `hf.co/...`, or via the `Modelfile`
+TEMPLATE when building locally) and **llama.cpp / llama-cpp-python**
+(via the GGUF's embedded jinja). Other apps' tool-calling support
+depends on whether they read the embedded template or require an
+external schema.
 
 ### Inference (OpenAI-compatible)
 
@@ -322,11 +326,18 @@ templates directly (llama.cpp, llama-cpp-python, LM Studio) handle the
 plain-conversation formatting automatically.
 
 Ollama is the exception: its conversion of the embedded jinja loses the
-`.Tools` / `.ToolCalls` blocks Ollama's capability detector requires, so
-the `Modelfile` in this repo overrides the template with an Ollama-Go
-version that wires tool calling correctly. Use the bundled `Modelfile`
-(via `make build` or `ollama run hf.co/FoolDev/janus-27b`) and tools
-will work end-to-end on `/api/chat` and `/v1/chat/completions`.
+`.Tools` / `.ToolCalls` blocks Ollama's capability detector requires.
+Two paths fix this, depending on how you pull the model:
+
+- **`ollama run hf.co/FoolDev/janus-27b`** — HF's Ollama bridge applies
+  the root-level `template` / `system` / `params` files in this repo
+  (the bridge does **not** read `Modelfile`).
+- **`make build` / `ollama create janus-27b -f Modelfile`** — uses the
+  `Modelfile`'s `TEMPLATE` block.
+
+Both routes wire `.Tools` / `.ToolCalls` and tools work end-to-end on
+`/api/chat` and `/v1/chat/completions`. The two configurations are
+kept in sync: edit them together if you change one.
 
 #### Plain conversation
 

params

@@ -0,0 +1,12 @@
+{
+  "temperature": 0.6,
+  "top_p": 0.95,
+  "top_k": 20,
+  "repeat_penalty": 1.05,
+  "num_ctx": 16384,
+  "stop": [
+    "<|im_end|>",
+    "<|endoftext|>",
+    "<|im_start|>"
+  ]
+}

scripts/{fetch_mmproj.sh → fetch_vision.sh}

@@ -8,9 +8,9 @@
 #   it (see README Vision section, ollama/ollama#15898).
 #
 # Usage:
-#   ./scripts/fetch_mmproj.sh                    # default: F16, ~927 MB
-#   ./scripts/fetch_mmproj.sh BF16               # ~931 MB
-#   ./scripts/fetch_mmproj.sh F32                # ~1.8 GB
+#   ./scripts/fetch_vision.sh                    # default: F16, ~927 MB
+#   ./scripts/fetch_vision.sh BF16               # ~931 MB
+#   ./scripts/fetch_vision.sh F32                # ~1.8 GB
 #
 # Requires: huggingface-cli (or hf).
 set -euo pipefail

system

@@ -0,0 +1,10 @@
+You are Janus, a precise and capable assistant for reasoning, writing, coding, and long-form dialogue.
+
+Behavior rules:
+- Answer the user's actual request directly.
+- Be accurate, complete, and structured.
+- Think before answering, but do not get stuck in repetitive loops or meta-commentary.
+- If the request is ambiguous or incomplete, state what is missing and make the smallest reasonable assumption needed to continue.
+- If the user wants creative writing, preserve tone, continuity, and character consistency.
+- If the user wants analysis or technical help, prefer concrete steps, examples, and decisions over fluff.
+- Finish with a usable answer, not just planning.

template

@@ -0,0 +1,51 @@
+{{- $lastUserIdx := -1 -}}
+{{- range $idx, $msg := .Messages -}}
+{{- if eq $msg.Role "user" }}{{ $lastUserIdx = $idx }}{{ end -}}
+{{- end }}
+{{- if or .System .Tools }}<|im_start|>system
+{{ if .System }}{{ .System }}
+
+{{ end }}
+{{- if .Tools }}# Tools
+
+You may call one or more functions to assist with the user query.
+
+You are provided with function signatures within <tools></tools> XML tags:
+<tools>
+{{- range .Tools }}
+{"type": "function", "function": {{ .Function }}}
+{{- end }}
+</tools>
+
+For each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:
+<tool_call>
+{"name": <function-name>, "arguments": <args-json-object>}
+</tool_call>
+{{- end -}}<|im_end|>
+{{ end }}
+{{- range $i, $_ := .Messages }}
+{{- $last := eq (len (slice $.Messages $i)) 1 -}}
+{{- if eq .Role "user" }}<|im_start|>user
+{{ .Content }}<|im_end|>
+{{ else if eq .Role "assistant" }}<|im_start|>assistant
+{{ if (and $.IsThinkSet (and .Thinking (or $last (gt $i $lastUserIdx)))) -}}
+<think>{{ .Thinking }}</think>
+{{ end -}}
+{{ if .Content }}{{ .Content }}{{ end }}
+{{- if .ToolCalls }}
+{{- range .ToolCalls }}
+<tool_call>
+{"name": "{{ .Function.Name }}", "arguments": {{ .Function.Arguments }}}
+</tool_call>
+{{- end }}
+{{- end }}{{ if not $last }}<|im_end|>
+{{ end }}
+{{- else if eq .Role "tool" }}<|im_start|>user
+<tool_response>
+{{ .Content }}
+</tool_response><|im_end|>
+{{ end }}
+{{- if and (ne .Role "assistant") $last }}<|im_start|>assistant
+<think>
+{{ end }}
+{{- end }}