The 'HF token: NOT SET' shell hint shown when downloading a gated/private
model told users to add a token under 'Odysseus Settings -> Cookbook ->
HuggingFace Token'. There is no Cookbook section under the app Settings;
the HuggingFace Token field lives under the Cookbook page's Settings tab
(static/js/cookbook.js — data-backend="Settings" group). Following the
old hint led nowhere. Reverse the path to match the real UI.
Fixes#3829
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
When a download's tmux pane is gone, the status endpoint trusted only the
HF-cache probe to tell completed from stopped. The probe derives its cache
root from its own environment, but the download runner exports
HF_HOME=<local_dir> (the #2722 fix), so custom-dir downloads land in
<local_dir>/hub where the probe never looks - and ollama pulls don't touch
the HF cache at all. Finished downloads were reported as stopped forever,
and tasks already persisted as completed were demoted back to stopped on
the next poll. This is the backend half of #3897, deliberately left out of
the frontend fix in #4000.
- honor the conclusive runner markers first: DOWNLOAD_OK -> completed
(keeping the "Fetching 0 files" error guard), DOWNLOAD_FAILED -> error
- pass the task's local_dir through to the cache probes so they check the
cache the download actually wrote to, keeping the env-var fallback for
default-cache downloads
- move the probe scripts and marker classification into
routes/cookbook_output.py (dependency-free) with behavioral tests
Fixes#4017
The tmux default 2000-line scrollback was getting blown out by
long vLLM tracebacks (DeepSeek-V4-Flash launch crash had the root
cause scrolled off; the user saw only the tail "See root cause
above"). Bumped:
- tmux server history-limit to 100000 at session creation (prepended
to each tmux new-session command so both local + ssh remote inherit
the larger scrollback)
- crash-watchdog capture-pane from -S -200 → -S -2000 so the
diagnosis includes the actual exception line
* fix: expand cookbook error output tail from 12 to 50 lines
When a task reaches status 'error', the status endpoint was returning
only the last 12 lines of the subprocess log. The existing context-menu
'Copy last 50 lines' action was therefore copying the same 12 lines,
making it useless for diagnosing failures that produce long stack traces
or build output.
- Set _tail_lines = 50 when status == 'error', keep 12 for running tasks
- Initialise exit_code = None before the status-classification block so
it is always defined in the result dict (was only set inside the
is_alive branch, potential NameError in the dead-session path)
- Include exit_code in the task-status response dict
- JS poller captures exit_code from live data into local task state
The frontend output panel and 'Copy last 50 lines' now show the actual
error context without any UI changes.
* refactor: extract output-tail logic to testable helper + behavioral tests
Addresses review feedback on #1538: the previous tests were source-level
string guards. Extract the tail-slicing into a dependency-free helper
(routes/cookbook_output.error_aware_output_tail) and replace the guards
with behavioral tests that exercise the actual logic:
- error status with a 200-line snapshot -> exactly the last 50 lines
- running/ready/completed/stopped/unknown -> last 12 lines
- short snapshot -> all lines, no padding
- empty snapshot -> empty string
- error tail is a strict superset (suffix-compatible) of the non-error tail
The helper has no FastAPI/SQLAlchemy imports so it unit-tests without
standing up the app.
---------
Co-authored-by: Alexandre Teixeira <111787685+alteixeira20@users.noreply.github.com>
The module derived its state file path as Path(os.environ.get("DATA_DIR", "data"))
/ "cookbook_state.json". The correct env var is ODYSSEUS_DATA_DIR, which is
already read by src/constants.py and exported as COOKBOOK_STATE_FILE. When
ODYSSEUS_DATA_DIR is set (Docker, custom installs), the old code read the wrong
env var and silently wrote state to data/cookbook_state.json relative to CWD
while every other file resolved under the custom data directory.
Fixes#3621
Surface a lot of accumulated cookbook + UI work as a single non-agent
commit so the agent rework lands cleanly.
Highlights:
- Ollama as a first-class backend in the Cookbook:
* Download input accepts ollama-style names (name:tag) → backend=ollama
* /api/cookbook/ollama/library (cached scrape of ollama.com + curated
fallback so classic models like qwen2.5 stay reachable)
* "Browse Ollama library" toggle below Download with size chips
* Engine=Ollama in hwfit toolbar merges the Ollama library into the
main scan list as per-tag rows with the same Fit/Param/Quant/VRAM
columns; click → fills Download input
- API Tokens form added to Integrations panel (matching wired
loadTokens()/initTokenForm() that had no HTML)
- Serve panel polish: Advanced fold tightening (-8px nudges on vLLM
checks, Extra args, Spec row), n_cpu_moe + Split Mode controls
pulled up 8px to align with the row's checkboxes, GGUF File dropdown
exposed for Ollama backend, GPU re-render on Edit serve restore,
_forceBackend flag so saved serveState wins over backend detection,
cookbook:servers-changed CustomEvent so panels don't need refresh
- Models page redesign: Add Models row (URL + hidden API key reveal +
Type select + Scan/Ollama/Key/Test/Add icon buttons), Probe All +
Clear-offline buttons in Added Models toolbar, offline-pill removed
(opacity already conveys state), Engine dropdown gains Ollama option
- _ping_endpoint probes /v1/models then base, accepts 4xx as
reachable (vLLM returns 404 on bare /v1, fully working endpoints
were showing offline)
- Diagnosis card: × dismiss + Copy bundle buttons restored on the
serve error feedback card
- Orphan tmux sweep re-enabled behind a 60s rate-limit + background
Thread (off the main event loop) so dead serves get discovered
- cookbook_routes auto-register watchdog: drops the endpoint if the
serve session exits non-zero within the first ~3min
- ollama-rocm sidecar awareness in download wrapper (`docker exec
ollama-rocm ollama pull` when host ollama isn't installed)
- Skill extractor sets initial_status="published" when
auto_approve_skills pref is on (audit demotes later)
- Skill list / model list / cookbook scan misc polish
* refactor(constants): single source of truth for data dir + merge core/src constants
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* docs(contributing): use named src.constants for data paths, drop core/constants references
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Three call sites hardcoded Path("/app/data/cookbook_state.json"), which only
exists in Docker; on a native run the real path is <repo>/data, so the state
file looked missing and cookbook serve-state was silently ignored. Two others
used os.environ.get("DATA_DIR", "data") (a relative fallback, since DATA_DIR is
never set as an env var). Route all five through core.constants.DATA_DIR so the
path is consistent and absolute on both Docker and native.
Part of #3331.
Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>
* fix(platform): add WSL compatibility functions and path translation
fix(cookbook): enhance model scan script to support additional HuggingFace cache paths
fix(hardware): improve cache key generation for remote SSH context
test(tests): add tests for WSL detection and path translation functionality
* fix(cookbook): prefer prebuilt wheels for llama-cpp-python and normalize package aliases
* fix: enable StrictHostKeyChecking in nvidia probe
refactor: consolidate ssh & powershell command execution to utility functions in core module
refactor: consolidate nvidia path candidates in to single variables in core module
tests: add tests for new utility functions
* fix: correct wrong variable name
This commit consolidates all Windows Cookbook background fixes into a single comprehensive commit based on the latest main branch.
Key fixes included:
1. React looksSuccessful Mismatch: Append 'DOWNLOAD_OK' for pip install commands in routes/cookbook_routes.py.
2. Local Windows SSH Wrapper & Log Directory Mismatch: Bypassed ssh wrappers and dynamically selected odysseus-tmux logs for local tasks in static/js/cookbookRunning.js.
3. WSL Bash Filtration: Filtered out the WSL bash stub at C:\Windows\System32\bash.exe in core/platform_compat.py.
4. Drive-Colon Path Normalization: Replaced .as_posix() with git_bash_path() in routes/shell_routes.py and src/bg_jobs.py.
5. GGUF-Only Hardware Fitting: Restructured local Windows recommendations to rank GGUF only in services/hwfit/fit.py.
6. Safe Win32 Process Liveness Probe: Replaced os.kill(pid, 0) with a safe Win32 API probe using GetExitCodeProcess in core/platform_compat.py.
7. Prebuilt llama-cpp-python Wheels: Supply the CPU extra index during compilation failure fallback.
8. Enforce UTF-8 log encoding: Set PYTHONIOENCODING=utf-8 on Windows bootstrap runners.
9. Fix Linux Llama.cpp Build script syntax error in routes/cookbook_helpers.py.
10. Page Reload Status Check: Run sys.executable instead of 'python3' to bypass Microsoft Store execution stubs on local Windows hosts.
11. Llama.cpp serve build bypass: Bypassed cmake compilation checks on local Windows and verified python bindings directly.
12. Serve Command Path Validation: Masked safe GGUF path printf subshells '' inside the serve command validator.
13. CPU Mismatch Diagnostics: Intercepted AVX2-lacking '0xc000001d' (Illegal Instruction) crashes in static/js/cookbook-diagnosis.js and guided users to Ollama.
14. Windows Pytest stability: Fixed stub import leakage in test files.
* refactor(cookbook): move _diagnose_serve_output to module level in cookbook_helpers
Extracts the nested _diagnose_serve_output function from inside
setup_cookbook_routes() and moves it to module level in cookbook_helpers.py,
alongside the other helper functions it logically belongs with.
No behaviour change — the function is now importable directly for testing
and by other callers without going through the route factory closure.
* fix(cookbook): surface backend diagnosis when serve fails in background
The background poll (_pollBackgroundStatus) already received `diagnosis`
and `cmd` from /api/cookbook/tasks/status but discarded both. When a serve
job died while the Cookbook modal was closed, reopening it showed only a
red error badge with no context.
- Persist live.diagnosis into task._backendDiagnosis in localStorage so it
survives modal close/reopen and page refresh
- Persist live.cmd into task.payload._cmd for agent-spawned tasks so the
crash report includes the actual command
- After _renderRunningTab(), walk rendered cards and call _showDiagnosis()
for any that have a stored _backendDiagnosis but no panel yet
- In _renderTaskCard(), use _backendDiagnosis as a fallback when the
client-side _terminalServeDiagnosis() finds nothing
* test(cookbook): add coverage for _diagnose_serve_output error patterns
10 tests verifying the 16 serve-failure patterns:
- CUDA OOM, port-in-use, vLLM missing, gated model
- Traceback fallback fires without startup success marker
- Traceback suppressed when server actually started
- Clean/empty output returns None
- trust-remote-code and no-GGUF patterns
- Schedule cookbook serves through the existing ScheduledTask system: the
serve preset gets a ^ button next to Launch that opens a daily/hourly/
weekly form mirroring the admin-switch style; the schedule action runs
action_cookbook_serve, which delegates to /api/model/serve and stamps
the resulting task with _scheduledStopAtMs. A background
cookbook_serve_lifecycle loop ticks every 60s and kills any serve
whose window has ended, also dropping the auto-registered endpoint
so the model picker doesn't keep pointing at a dead server.
- Stop and remove on a Running serve now awaits the SSH/tmux kill,
re-checks tmux has-session, and surfaces an error toast (leaving the
row) when the kill failed. Previously fire-and-forget, so a failed
SSH/tmux call silently left the live serve running while the row
vanished from the UI.
- Cookbook tasks/status orphan-adoption sweep no longer requires the
serve-/cookbook- session-id prefix; any tmux session whose pane is
running a known model-server process gets auto-pulled into Running.
Without this loosening, a cookbook-launched serve whose tmux id
fell back to a bare number was invisible — you couldn't see it,
let alone stop it.
- Ollama serve always launches a fresh process under cookbook's tmux
(no more monitor-mode reattach to a systemd/Docker ollama Stop can't
reach). The handler pre-picks a free port by probing the target
host over SSH and mutates req.cmd's OLLAMA_HOST so the runner script
AND the auto-registered endpoint agree on the same bind port.
- Auto-register uses host.docker.internal (when running inside Docker)
instead of localhost, matching the URL /setup adds for Ollama by
hand. Local cookbook serves now produce a chat-reachable endpoint
on first launch.
- Cascade-delete: removing a scheduled cookbook task also deletes any
linked calendar event (cookbook_task_id marker in the description).
- Tasks list groups cookbook_serve under a "Cookbook" category that
sorts above the rest, so scheduler-launched serves are easy to find.
Three converging fixes so the chat agent + external Codex/Claude skills can actually debug a crashed serve instead of staring at a post-crash neofetch banner:
* Serves now `tee` to /tmp/odysseus-tmux/SESSION.log on the host running them. Runner saves fds 3/4 before the tee and restores them right before `exec ${SHELL}`, so the post-crash interactive zsh banner does NOT pollute the log file.
* `tail_serve_output` (chat agent) and `/api/codex/cookbook/output/{sid}` (Codex+Claude skills) both prefer the persistent log file over the tmux pane. Pane is fallback for sessions predating the tee runner. Default tail bumped 150 -> 400.
* `list_served_models` "recent log" snippet seeks to the Traceback line instead of showing the last 6 lines (which was always the bash prompt).
Cookbook auto-adoption sweep on `/api/cookbook/tasks/status`: every 20s (rate-limited) the cookbook SSHes each configured server, finds `serve-*` / `cookbook-*` tmux sessions running an actual model process (vllm/python/llama-server/etc., filtered via `pane_current_command`), and writes them into state.tasks. So when the agent falls back to raw ssh+tmux, the session appears in the Cookbook UI on the next poll.
`serve_model` error path now reads `data["detail"]` in addition to `data["error"]` so the FastAPI HTTPException message ("Invalid characters in cmd") actually reaches the agent instead of being swallowed as a generic "Serve failed". Tool description updated to warn against `cd …`/`source …`/`&&` prefixes.
Intent-without-action supervisor in agent_loop: when the model writes "Let me tail the output" / "I'll check the logs" / "Let me investigate" and ends the turn without emitting a tool call, the loop injects a sharp system nudge ("You said you would X — DO IT NOW") and continues. Capped at 2 nudges per chat so a model that genuinely cannot use the tool does not pin the loop.
Codex/Claude skill parity: adds `/cookbook/cached`, `/cookbook/presets`, `/cookbook/preset/{name}`, `/cookbook/adopt` so external agents have the same surface as the chat agent. SKILL.md docs + odysseus_api.py wrapper updated for both bundles.
`adopt_served_model` promoted to the always-on tool set so the agent has a documented fallback when serve_model rejects a cmd.
Also various cookbook UI tweaks accumulated alongside the above (cookbook.js, cookbookRunning.js, cookbookServe.js, cookbook-diagnosis.js, settings.js, style.css).
Backend (services/hwfit + routes):
- rank_models picks visible set by REQUESTED column, not always score —
sorting by Param now shows highest-param models PERIOD (incl. too_tight).
- New fit_only param. Multi-GPU rigs filter GGUF Q*/IQ quants (vLLM/SGLang
cannot serve them); default non-prequantized to BF16 on 2+ GPUs.
- AWQ / GPTQ-8bit get a -1.0 quality penalty (was 0.0, tied with FP8), so
FP8 wins when both fit.
- Version-aware tiebreaker (parse Mn.n / Vn) — MiniMax-M2.7 ranks above
M2.5 on equal composite score; >=100B integers not misread as versions.
- /api/cookbook/hf-latest no longer drops models without an "NB" pattern in
the repo id (MiniMax-M2.7, DeepSeek-V4-Pro etc. were silently filtered).
- Cached-model scan: atexit flushes models JSON even if the script is
killed mid-walk; each scan_dir wrapped in try/except; timeout 60s -> 180s.
- KB granularity for sub-MB sizes (was "0 MB" for 12 KB shells). New
"stalled" status for shells <1 MB with no .incomplete files.
- /api/cookbook/state POST guard: rejects "done" download tasks lacking
DOWNLOAD_OK / DOWNLOAD_FAILED / /snapshots/ when the last-mentioned
shard is N<total — stops stale tabs from poisoning persisted state.
- hf_models.json: add zai-org/GLM-5.1; flip zai-org/GLM-5 quantization
Q4_K_M -> BF16 (it is the native base, not a quant).
Frontend (static/js):
- Scan/Download toolbar: quant defaults to All; ctx slider (8k/16k/32k/
50k/128k/Max) ported from origin/main with sort=fit on drag, sort=score
on Max. GPU toggle commits _activeCount to maxGpu on initial render. Fit
column header tagged with active budget (RAM / GPU / N GPU).
- Foldable Download admin-card: the Download h2 is the chevron trigger;
state persists in localStorage.
- Download card surfaces destination dir (Dir: <path>). Same dir on running
task row, font/color matched to uptime (9px Fira Code muted, opacity .4).
- Serve panel ctx text input always resets to model max on open. Sub-MB
cached models show with red "download stalled" badge.
- Bulk-select Cancel + Delete reset the Select button label on exit.
- Cookbook running: false-finished bug fixed — DOWNLOAD_OK or /snapshots/
required; bare "Download complete" no longer marks the task done after
the first config file. Clear button now sends tmux kill-session too.
True overall % for multi-shard downloads: ((N-1)+frac)/total instead of
hf_transfer per-shard aggregate.
- Diagnosis card simplified: removed fold toggle, copy button, dismiss X.
Suggestion font matches message body (12px).
- HF token field flashes green check + "Saved" on save.
- Cached scan no longer counts stalled rows as downloaded in Scan/Download.
CSS:
- dep Install button width pinned to 76px to match Installed split.
- task-sub row +1px; task-status badge gets margin-right 8px.
- Ctx slider styled like gallery editor sliders (thin pill rail, red thumb).
- Bulk-select cancel button top -3px -> -5px.
The llama.cpp serve auto-install built a bare `llama-cpp-python` in the Linux
source-build fallback and the Termux path, but the serve command runs
`python3 -m llama_cpp.server`, which needs the `[server]` extra. Because the
"already installed?" guard only checks `import llama_cpp` (a bare install
satisfies it), the missing extra was never added, so serving crashed with
`ModuleNotFoundError: No module named 'starlette_context'` (issue #730).
- Request the `[server]` extra in both the Termux direct install and the Linux
Python-bindings fallback (the Windows path already used `[server]`).
- Shell-quote the package spec in `_pip_install_fallback_chain` via `shlex.quote`
so the `[server]` brackets aren't treated as a bash glob; plain names unaffected.
Tests: tests/test_cookbook_helpers.py gains extras-quoting coverage and a
serve-runner regression guard.
Serving a diffusion model auto-registered an image endpoint so it appeared in the model picker, but serving an LLM (llama.cpp/vLLM/SGLang/Ollama) did not — a downloaded-and-served model never showed up until the user manually ran /setup. Add _auto_register_llm_endpoint (text sibling of _auto_register_image_endpoint): parse the serve port (explicit --port, else Ollama 11434, else llama.cpp 8080), point an endpoint at http://host:port/v1, dedupe by base_url, and set supports_tools from --enable-auto-tool-choice. Wire it into /api/model/serve for any non-pip, non-diffusion serve.
Cookbook dependency installs (vLLM and friends) build large wheels; pip's
default cache lives under $HOME/.cache/pip, so on a small home filesystem the
build dies mid-way with "[Errno 28] No space left on device" (issue #1219) and
the dependency ends up "installed" but unusable (issue #1459).
Add `--no-cache-dir` to the dependency pip-install command (the maintainer's
suggested PIP_CACHE_DIR= workaround, made the default) via a small
_pip_install_no_cache() helper applied at the install chokepoint. Consistent
with the existing --no-cache-dir on the llama-cpp-python build. Idempotent;
non-pip-install serve commands are untouched.
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
On Windows, Python defaults to the active code page (cp1252) for
subprocess I/O. HuggingFace CLI outputs U+2713 (✓) when validating
tokens, which cp1252 cannot encode, crashing the download process.
Set PYTHONUTF8=1 and PYTHONIOENCODING=utf-8 in the subprocess
environment so Unicode output from hf/pip/llama-server is handled
correctly.
Fixes#1543
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
The shell pattern 'if [ $? -eq 0 ]; ... else ... echo DOWNLOAD_FAILED (exit $?)' always reports 'exit 1' because $? inside the else branch is the exit code of the [ test command, not the download. Capture into _ec first.
* Fix Cookbook dependency install completion state
Mark Cookbook dependency installs as complete when the background runner
exits successfully, even when HuggingFace-specific download markers are
absent.
* Add focused regression coverage for cookbook dependency completion.
Keep the fix narrowly scoped while carrying env_path through dependency tasks and locking the completion reconciliation behavior with targeted tests.
* Cookbook: Engine filter + intelligent hardware-computed serve profiles
Two related Cookbook serving improvements for accurate, hardware-aware model
serving (especially on consumer GPUs that can only run GGUF/llama.cpp).
Engine filter
- New "Engine" dropdown (All / llama.cpp / vLLM / SGLang) beside the quant
picker. Pure client-side view filter over the fetched list via the same
_detectBackend() the serve commands use, so what you filter to is exactly what
would launch. Re-renders from cache (no refetch). Empty-state message + the
instant-cache-paint path account for it too.
Intelligent serve profiles (Quality / Balanced / Speed)
- services/hwfit/profiles.py: compute_serve_profiles() turns detected VRAM +
model size into concrete llama.cpp flags (n_gpu_layers, n_cpu_moe, cache-type,
context). Encodes the by-hand tuning: a too-big MoE offloads experts to CPU
instead of failing; a model that fits stays fully on GPU; quant tracks profile
intent; vision models keep image-encoder headroom. Reuses models.py VRAM math
so filtering and serving agree on what fits. Pure/deterministic (no t/s claims
— partial-offload speed isn't reliably predictable; fit is what's computed).
- /api/hwfit/profiles endpoint returns the profiles + the model's trained
context limit, with loose name matching (strips org/ prefix, -GGUF suffix,
quant tag) so a local GGUF folder name resolves to its catalog entry.
- _buildServeCmd (llama.cpp) now emits --n-cpu-moe / --flash-attn /
--cache-type-k/v when set, with llama-cpp-python fallback equivalents. It
previously only set -ngl/-c, which is why it OOM'd or ran slow.
- Serve panel: profile chips that fill the fields on click, plus CPU-MoE / KV
Cache / Flash Attn fields. Context is clamped to the model's trained limit
(and an absolute 1M sanity ceiling) on type/blur/profile-load and at launch —
fixes a crash where a stale 256k/16M preset + quantized KV cache caused an
amdgpu ErrorDeviceLost.
Tests: tests/test_serve_profiles.py (7) — offload vs full-GPU fit, never exceed
VRAM, context cap, launchable flags, vision headroom, no-GPU empty.
Checks: py_compile + node --check pass; pytest test_serve_profiles + test_hwfit_amd
green; verified live on an RDNA4 box (gfx1200) — Balanced lands ~ncm18 q4 128k,
matching hand-tuning.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* Cookbook: make column-header sorting discoverable (incl. Newest)
Sorting in Cookbook is via clickable column headers (pewds' design), but the
headers had no visual cue that they're interactive — so sorting in general, and
the Newest sort on the Model header specifically, was undiscoverable.
- Style sortable headers as interactive: pointer cursor, hover underline, and
the active sort column bolded/highlighted. There was no CSS for
.hwfit-sortable / .hwfit-sort-active at all; this helps every existing sort,
not just Newest.
- The Model column header sorts by release_date (newest first), reusing the
existing header-click sort wiring and the "newest" SORT_KEY.
No new sort control — uses the existing column-header paradigm.
Checks: node --check passes.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* Cookbook serve profiles: keep the on-disk file's quant fixed (don't propose Q6/Q2)
In the Serve tab the model is a specific GGUF file already on disk, so its quant
can't change — but the profiles were suggesting "Quality · Q6_K" / "Speed · Q2_K"
as if you could re-quantize it. That's meaningless when serving a fixed file.
- compute_serve_profiles gains serve_weights_gb / serve_quant. When set (SERVE
mode), the quant is locked to the file's and profiles differ only in the real
serving knobs — n_cpu_moe, KV-cache type, context. _weights_gb / _cpu_moe_for_budget
use the file's actual size instead of a quant-derived estimate. DOWNLOAD mode
(no override) still varies the quant to show download options.
- /api/hwfit/profiles accepts serve_weights_gb & serve_quant.
- The Serve panel parses the file's size (from m.size "20.6 GB") and quant (from
the repo/file name) and passes them, so profiles match what's actually served.
Result for a 20.6 GB Q4_K_M file: all three profiles stay Q4_K_M and differ by
KV/ctx/offload (Quality q8 KV 128k ncm21, Balanced q4 128k ncm17, Speed q4 32k
ncm15) — no nonsensical quant changes.
Tests: test_serve_mode_keeps_fixed_quant. Full serve-profile suite green (9).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* Cookbook serve: Vision toggle (auto-find mmproj) + live VRAM/RAM-spillover monitor
Two serve-panel additions:
1. **Vision toggle.** A "Vision" checkbox that serves the model with its
multimodal projector so it can read images. The mmproj path is resolved at
runtime (find mmproj-*.gguf next to the model), so dropping an mmproj file in
the model folder makes the toggle just work; `--mmproj … --image-max-tokens
1024` (native) / `--clip_model_path` (llama-cpp-python) only when on + found.
2. **Live GPU-memory monitor.** A readout that polls /api/cookbook/gpus every 4s
while the panel is open and shows VRAM used/total/%, free, and — crucially on
a discrete card — **RAM spillover** (AMD gtt_used_mb), with a plain-language
health hint: green/healthy, amber/tight, red/"spilled to RAM — slow (raise
CPU MoE or lower context)". Surfaces gtt_used_mb from the gpus endpoint
(previously read for total only and discarded for 'used').
Lets you see at a glance whether a config fits VRAM (fast) or is paging to system
RAM over PCIe (slow) instead of guessing.
Checks: node --check + py_compile pass.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Adds a diagnosis pattern for the 'Failed to infer device type' error
vLLM raises when no CUDA or ROCm GPU is found (e.g. systems with only
integrated or Intel Xe graphics). The existing pattern only caught
'No CUDA GPUs are available' which fires later in startup; this new
entry catches the earlier device-probe failure and the NVML/amdsmi
library-not-found messages that precede it.
Surfaces in the Cookbook serve card as: "vLLM could not find a supported
GPU — switch to llama.cpp or Ollama" instead of a raw Python traceback.
Co-authored-by: Claude Sonnet 4.6 (1M context) <noreply@anthropic.com>
Windows has 'App Execution Aliases' that can make shutil.which('python3')
and shutil.which('python') resolve to a Microsoft Store stub instead of
real Python -- even when Python is properly installed. The stub outputs:
'Python was not found; run without arguments to install from the
Microsoft Store, or disable this shortcut from Settings > Apps >
Advanced app settings > App execution aliases.'
and exits 9009, producing empty stdout. The JSON parse of the local
model cache scan then fails with 'Expecting value: line 1 column 1
(char 0)', and the Cookbook model list shows nothing.
Fix: prefer sys.executable as the interpreter for the local scan.
Odysseus already runs inside its own venv, so sys.executable always
points to the real venv Python and bypasses PATH / Store alias lookup
entirely. which_tool() is kept as a fallback.
Cross-platform: sys.executable works identically on Linux and macOS
(returns the real interpreter path), so this change is safe everywhere.
* macOS/Apple Silicon: detect Metal backend, surface MLX models, brew tmux hint
- hardware.py: add _detect_macos() via sysctl/system_profiler; report
backend=metal + unified_memory on Apple Silicon instead of cpu_arm
- fit.py: add Apple Silicon (M1-M5) unified-memory bandwidths + metal
FALLBACK_K so throughput estimates use the real bandwidth formula
- setup.py: Mac-specific 'brew install tmux' hint
Verified on M5 Pro 48GB: backend=metal, 273GB/s matched, 6 MLX models now
visible (were hidden), cuda still hides MLX, no new test failures.
* Fix native macOS tailnet launch and Metal GPU probe
---------
Co-authored-by: Elijah (Hermes) <hermes@local>
When serving with the llama.cpp backend and no .gguf file exists on the host,
the GGUF launcher prelude exits with 'ERROR: No GGUF found on this host', but
_diagnose_serve_output had no matching pattern, so the UI showed a generic
crash instead of explaining the cause. Add a diagnosis pattern for the
no-GGUF case so users are told a .gguf is required and pointed at downloading
a GGUF build, instead of an opaque crash.
Closes#811
A Cookbook download whose repo/quant selector matched no files (e.g. a
':Q4_K_M' tag that does not exist) printed 'Fetching 0 files' and was still
reported as a successful '✓ Downloaded' / completed task. Detect the
zero-file signature in the download snapshot and mark the task as an error
with a clear diagnosis (no matching files — check the repo or quant/filename
pattern) so users know nothing was actually downloaded. Normal multi-file
and fully-cached downloads (which print 'Fetching N files', N>0) are
unaffected.
Closes#839
pewdiepie-archdaemon