docs: record Qwen3.5-9B DFlash Metal timeout (refs #152, #154)

Merged PR #90: feat: integration test — turboquant compressed model

README.md

@@ -30,3 +30,4 @@ See [issues](https://forge.alexanderwhitestone.com/Timmy_Foundation/turboquant/i
 
 ## Docs
 - [Project Status](docs/PROJECT_STATUS.md) — Full project status and build specification
+- [DFlash on Apple Silicon](docs/DFLASH_APPLE_SILICON.md) — MLX benchmark planner, setup commands, and report workflow

benchmarks/dflash_apple_silicon.py

@@ -0,0 +1,189 @@
+#!/usr/bin/env python3
+"""Apple Silicon DFlash planning helpers and CLI (issue #152)."""
+
+from __future__ import annotations
+
+import argparse
+import json
+import platform
+import subprocess
+from dataclasses import asdict, dataclass
+from pathlib import Path
+from typing import Iterable, Optional
+
+
+@dataclass(frozen=True)
+class DFlashPair:
+    slug: str
+    base_model: str
+    draft_model: str
+    estimated_total_weights_gb: float
+    minimum_recommended_memory_gb: float
+    draft_sliding_window_size: int = 4096
+
+
+SUPPORTED_PAIRS: tuple[DFlashPair, ...] = (
+    DFlashPair(
+        slug="qwen35-4b",
+        base_model="Qwen/Qwen3.5-4B",
+        draft_model="z-lab/Qwen3.5-4B-DFlash",
+        estimated_total_weights_gb=9.68,
+        minimum_recommended_memory_gb=16.0,
+    ),
+    DFlashPair(
+        slug="qwen35-9b",
+        base_model="Qwen/Qwen3.5-9B",
+        draft_model="z-lab/Qwen3.5-9B-DFlash",
+        estimated_total_weights_gb=19.93,
+        minimum_recommended_memory_gb=28.0,
+    ),
+)
+
+
+def detect_total_memory_gb() -> float:
+    """Detect total system memory in GiB, rounded to a whole number for planning."""
+    system = platform.system()
+    if system == "Darwin":
+        mem_bytes = int(subprocess.check_output(["sysctl", "-n", "hw.memsize"]).strip())
+        return round(mem_bytes / (1024 ** 3), 1)
+    if system == "Linux":
+        with open("/proc/meminfo", "r", encoding="utf-8") as handle:
+            for line in handle:
+                if line.startswith("MemTotal:"):
+                    mem_kb = int(line.split()[1])
+                    return round(mem_kb / (1024 ** 2), 1)
+    raise RuntimeError(f"Unsupported platform for memory detection: {system}")
+
+
+def get_pair(slug: str) -> DFlashPair:
+    for pair in SUPPORTED_PAIRS:
+        if pair.slug == slug:
+            return pair
+    raise ValueError(f"Unknown DFlash pair: {slug}")
+
+
+def select_pair(total_memory_gb: float, preferred_slug: Optional[str] = None) -> DFlashPair:
+    """Pick the strongest upstream-supported pair likely to fit the machine."""
+    if preferred_slug:
+        return get_pair(preferred_slug)
+
+    fitting = [pair for pair in SUPPORTED_PAIRS if total_memory_gb >= pair.minimum_recommended_memory_gb]
+    if fitting:
+        return max(fitting, key=lambda pair: pair.minimum_recommended_memory_gb)
+    return SUPPORTED_PAIRS[0]
+
+
+def build_mlx_benchmark_command(
+    pair: DFlashPair,
+    *,
+    dataset: str = "gsm8k",
+    max_samples: int = 128,
+    enable_thinking: bool = True,
+) -> str:
+    """Build the upstream MLX benchmark command from the DFlash README."""
+    parts = [
+        "python -m dflash.benchmark --backend mlx",
+        f"--model {pair.base_model}",
+        f"--draft-model {pair.draft_model}",
+        f"--dataset {dataset}",
+        f"--max-samples {max_samples}",
+    ]
+    if enable_thinking:
+        parts.append("--enable-thinking")
+    parts.append(f"--draft-sliding-window-size {pair.draft_sliding_window_size}")
+    return " \\\n    ".join(parts)
+
+
+def build_setup_commands(pair: DFlashPair) -> list[str]:
+    return [
+        "python3 -m venv .venv-dflash",
+        "source .venv-dflash/bin/activate",
+        "git clone https://github.com/z-lab/dflash.git",
+        "cd dflash",
+        "pip install -e .[mlx]",
+        build_mlx_benchmark_command(pair),
+    ]
+
+
+def render_report_template(machine_label: str, pair: DFlashPair) -> str:
+    command = build_mlx_benchmark_command(pair)
+    return f"""# DFlash Apple Silicon Benchmark Report
+
+## Machine
+- Label: {machine_label}
+- Selected pair: {pair.slug}
+- Base model: {pair.base_model}
+- Draft model: {pair.draft_model}
+- Estimated total weight footprint: {pair.estimated_total_weights_gb:.2f} GB
+
+## Setup
+```bash
+python3 -m venv .venv-dflash
+source .venv-dflash/bin/activate
+git clone https://github.com/z-lab/dflash.git
+cd dflash
+pip install -e .[mlx]
+{command}
+```
+
+## Baseline comparison
+Compare against **plain MLX or llama.cpp speculative decoding** on the same prompt set.
+
+## Results
+- Throughput (tok/s):
+- Peak memory (GB):
+- Notes on acceptance / behavior:
+
+## Verdict
+Worth operationalizing locally?
+- [ ] Yes
+- [ ] No
+- [ ] Needs more data
+
+## Recommendation
+Explain whether this should become part of the local inference stack.
+"""
+
+
+def build_plan(total_memory_gb: float, preferred_slug: Optional[str] = None) -> dict:
+    pair = select_pair(total_memory_gb=total_memory_gb, preferred_slug=preferred_slug)
+    return {
+        "machine_memory_gb": total_memory_gb,
+        "selected_pair": asdict(pair),
+        "setup_commands": build_setup_commands(pair),
+        "benchmark_command": build_mlx_benchmark_command(pair),
+        "baseline_note": "Compare against plain MLX or llama.cpp speculative decoding on the same prompt set.",
+    }
+
+
+def write_output(path: Path, content: str) -> None:
+    path.parent.mkdir(parents=True, exist_ok=True)
+    path.write_text(content, encoding="utf-8")
+
+
+def main(argv: Optional[Iterable[str]] = None) -> int:
+    parser = argparse.ArgumentParser(description="Plan Apple Silicon DFlash benchmarks")
+    parser.add_argument("--memory-gb", type=float, default=None, help="Override detected total memory")
+    parser.add_argument("--pair", choices=[pair.slug for pair in SUPPORTED_PAIRS], default=None)
+    parser.add_argument("--machine-label", default="Apple Silicon Mac")
+    parser.add_argument("--format", choices=["json", "markdown"], default="markdown")
+    parser.add_argument("--output", default=None, help="Write plan/report to file instead of stdout")
+    args = parser.parse_args(list(argv) if argv is not None else None)
+
+    memory_gb = args.memory_gb if args.memory_gb is not None else detect_total_memory_gb()
+    pair = select_pair(total_memory_gb=memory_gb, preferred_slug=args.pair)
+
+    if args.format == "json":
+        content = json.dumps(build_plan(memory_gb, preferred_slug=pair.slug), indent=2)
+    else:
+        content = render_report_template(args.machine_label, pair)
+
+    if args.output:
+        write_output(Path(args.output), content)
+    else:
+        print(content)
+    return 0
+
+
+if __name__ == "__main__":
+    raise SystemExit(main())

benchmarks/reports/dflash_m3max_36gb.md

@@ -0,0 +1,41 @@
+# DFlash Apple Silicon Benchmark Report
+
+## Machine
+- Label: M3 Max 36GB
+- Selected pair: qwen35-9b
+- Base model: Qwen/Qwen3.5-9B
+- Draft model: z-lab/Qwen3.5-9B-DFlash
+- Estimated total weight footprint: 19.93 GB
+
+## Setup
+```bash
+python3 -m venv .venv-dflash
+source .venv-dflash/bin/activate
+git clone https://github.com/z-lab/dflash.git
+cd dflash
+pip install -e .[mlx]
+python -m dflash.benchmark --backend mlx \
+    --model Qwen/Qwen3.5-9B \
+    --draft-model z-lab/Qwen3.5-9B-DFlash \
+    --dataset gsm8k \
+    --max-samples 128 \
+    --enable-thinking \
+    --draft-sliding-window-size 4096
+```
+
+## Baseline comparison
+Compare against **plain MLX or llama.cpp speculative decoding** on the same prompt set.
+
+## Results
+- Throughput (tok/s):
+- Peak memory (GB):
+- Notes on acceptance / behavior:
+
+## Verdict
+Worth operationalizing locally?
+- [ ] Yes
+- [ ] No
+- [ ] Needs more data
+
+## Recommendation
+Explain whether this should become part of the local inference stack.

benchmarks/reports/dflash_m3max_36gb_qwen35_4b_pilot.md

@@ -0,0 +1,46 @@
+# DFlash Apple Silicon Pilot — Qwen3.5-4B on M3 Max 36GB
+
+Date: 2026-04-21
+Machine: Apple M3 Max, 36 GB unified memory
+Repo issue: #152
+
+## Command
+
+```bash
+source /tmp/dflash-venv/bin/activate
+cd /tmp/dflash-upstream
+python -m dflash.benchmark --backend mlx \
+    --model Qwen/Qwen3.5-4B \
+    --draft-model z-lab/Qwen3.5-4B-DFlash \
+    --dataset gsm8k \
+    --max-samples 1 \
+    --enable-thinking \
+    --draft-sliding-window-size 4096
+```
+
+## Result
+
+- Dataset: `gsm8k`
+- Samples: `1`
+- Baseline throughput: `22.35 tok/s`
+- DFlash throughput: `46.78 tok/s`
+- Decoding speedup: `2.09x`
+- Average acceptance length: `6.48`
+
+Acceptance length histogram:
+
+```text
+['0.3%', '11.1%', '12.7%', '10.4%', '11.7%', '7.6%', '7.0%', '3.8%', '5.1%', '6.3%', '2.8%', '3.8%', '2.2%', '1.9%', '0.9%', '2.5%', '9.8%']
+```
+
+## Caveats
+
+- This is a **pilot**, not a decision-grade benchmark.
+- Only `1` sample was run, so the throughput number is directional.
+- No apples-to-apples baseline against plain MLX or llama.cpp speculative decoding is included yet.
+- The planner still recommends trying `Qwen/Qwen3.5-9B + z-lab/Qwen3.5-9B-DFlash` on this machine for the more meaningful fit test.
+
+## Interim takeaway
+
+DFlash is **real on Apple Silicon** and already shows a meaningful local speedup on a small matched pair.
+A `2.09x` pilot speedup on `Qwen3.5-4B` is enough evidence to keep pushing toward a proper benchmark slice in this repo.

benchmarks/reports/dflash_m3max_36gb_qwen35_9b_timeout.md

@@ -0,0 +1,59 @@
+# DFlash on Apple Silicon Failure Report — Qwen3.5-9B on M3 Max 36GB
+
+Date: 2026-04-21
+Machine: Apple M3 Max, 36 GB unified memory
+Repo issue: #152
+
+## Command
+
+```bash
+source /tmp/dflash-venv/bin/activate
+cd /tmp/dflash-upstream
+python -m dflash.benchmark --backend mlx \
+    --model Qwen/Qwen3.5-9B \
+    --draft-model z-lab/Qwen3.5-9B-DFlash \
+    --dataset gsm8k \
+    --max-samples 1 \
+    --enable-thinking \
+    --draft-sliding-window-size 4096
+```
+
+## Outcome
+
+The benchmark did **not** complete successfully on this machine.
+
+### Failure signature
+
+```text
+libc++abi: terminating due to uncaught exception of type std::runtime_error:
+[METAL] Command buffer execution failed:
+Caused GPU Timeout Error (00000002:kIOGPUCommandBufferCallbackErrorTimeout)
+```
+
+Additional shutdown noise:
+
+```text
+bash: [11285: 1] tcsetattr: Inappropriate ioctl for device
+resource_tracker: There appear to be 1 leaked semaphore objects to clean up at shutdown
+```
+
+## Interpretation
+
+This is strong evidence that the `Qwen/Qwen3.5-9B + z-lab/Qwen3.5-9B-DFlash` pair is **not currently stable** on an M3 Max 36GB Mac under the upstream MLX benchmark path, at least with the default settings used here.
+
+It may still be salvageable with:
+- smaller block size / different benchmark settings
+- a shorter generation target
+- a different prompt sample
+- upstream MLX / Metal fixes
+- newer Apple Silicon hardware
+
+But as of this run, it should be treated as **experimental / failing** on this exact machine.
+
+## Recommendation
+
+For this Mac, the working local proof path is still:
+- `Qwen/Qwen3.5-4B`
+- `z-lab/Qwen3.5-4B-DFlash`
+
+Use the 4B pair for reproducible local validation while the 9B Metal timeout is investigated separately.

data/crisis_resources.json

@@ -1,88 +0,0 @@
-{
-  "version": "1.0.0",
-  "updated": "2026-04-15",
-  "description": "Offline crisis resources cache for edge deployment",
-
-  "national_resources": [
-    {
-      "name": "988 Suicide & Crisis Lifeline",
-      "phone": "988",
-      "text": "988",
-      "url": "https://988lifeline.org",
-      "description": "Free, confidential support 24/7 for people in distress. Call or text 988.",
-      "languages": ["en", "es"],
-      "available": "24/7"
-    },
-    {
-      "name": "Crisis Text Line",
-      "text": "HOME to 741741",
-      "url": "https://www.crisistextline.org",
-      "description": "Free crisis support via text message. Text HOME to 741741.",
-      "languages": ["en", "es"],
-      "available": "24/7"
-    },
-    {
-      "name": "Veterans Crisis Line",
-      "phone": "988 (press 1)",
-      "text": "838255",
-      "url": "https://www.veteranscrisisline.net",
-      "description": "Support for Veterans and their loved ones. Call 988, press 1.",
-      "available": "24/7"
-    },
-    {
-      "name": "Trevor Project (LGBTQ+ Youth)",
-      "phone": "1-866-488-7386",
-      "text": "START to 678-678",
-      "url": "https://www.thetrevorproject.org",
-      "description": "Crisis intervention and suicide prevention for LGBTQ+ young people.",
-      "available": "24/7"
-    },
-    {
-      "name": "SAMHSA National Helpline",
-      "phone": "1-800-662-4357",
-      "url": "https://www.samhsa.gov/find-help/national-helpline",
-      "description": "Free, confidential, 24/7 treatment referral and information service.",
-      "available": "24/7"
-    }
-  ],
-
-  "international_resources": [
-    {
-      "name": "International Association for Suicide Prevention",
-      "url": "https://www.iasp.info/resources/Crisis_Centres/",
-      "description": "Directory of crisis centers worldwide."
-    },
-    {
-      "name": "Befrienders Worldwide",
-      "url": "https://www.befrienders.org",
-      "description": "Emotional support to prevent suicide worldwide."
-    },
-    {
-      "name": "Canada — Talk Suicide",
-      "phone": "1-833-456-4566",
-      "text": "456456"
-    },
-    {
-      "name": "UK — Samaritans",
-      "phone": "116 123",
-      "email": "jo@samaritans.org"
-    },
-    {
-      "name": "Australia — Lifeline",
-      "phone": "13 11 14",
-      "text": "0477 13 11 14"
-    }
-  ],
-
-  "local_resources": [],
-
-  "self_help_prompts": [
-    "Take a slow breath. Inhale for 4 seconds, hold for 4, exhale for 6.",
-    "Look around. Name 5 things you can see, 4 you can touch, 3 you can hear.",
-    "You are not alone. This feeling will pass.",
-    "Call someone you trust right now.",
-    "Step outside if you can. Fresh air and movement can help.",
-    "Write down what you're feeling. Getting it out helps.",
-    "This moment is not your whole life. It's one moment."
-  ]
-}

docs/DFLASH_APPLE_SILICON.md

@@ -0,0 +1,125 @@
+# DFlash on Apple Silicon
+
+This repo now carries a **Gitea-first benchmark harness** for evaluating whether upstream **DFlash on MLX** is worth adding to the local Apple Silicon inference stack.
+
+## Why
+
+The headline `Kimi K2.6 + DFlash` benchmark was measured on `8x MI300X` with huge RAM and ROCm patches. That exact recipe is not a fit for a `36 GB` Apple Silicon Mac.
+
+What *is* relevant locally is the upstream `z-lab/dflash` MLX path, which can benchmark smaller matched target/draft pairs that fit on Apple Silicon.
+
+## Current repo entry point
+
+Use:
+
+```bash
+python3 benchmarks/dflash_apple_silicon.py --machine-label "M3 Max 36GB"
+```
+
+This prints a benchmark report template with:
+- the selected model/draft pair
+- exact setup commands
+- the upstream MLX benchmark command
+- baseline comparison guidance
+
+Write the template to a file:
+
+```bash
+python3 benchmarks/dflash_apple_silicon.py \
+  --machine-label "M3 Max 36GB" \
+  --output benchmarks/reports/dflash_m3max_36gb.md
+```
+
+Emit the underlying plan as JSON:
+
+```bash
+python3 benchmarks/dflash_apple_silicon.py --format json
+```
+
+## Selection logic
+
+Today the planner uses two upstream-supported MLX pairs:
+
+- `qwen35-9b`
+  - base: `Qwen/Qwen3.5-9B`
+  - draft: `z-lab/Qwen3.5-9B-DFlash`
+  - chosen for ~28 GB+ machines
+- `qwen35-4b`
+  - base: `Qwen/Qwen3.5-4B`
+  - draft: `z-lab/Qwen3.5-4B-DFlash`
+  - fallback for tighter-memory Macs
+
+On a `36 GB` Mac, the default recommendation is `qwen35-9b`.
+
+## Pilot result already landed
+
+A first live Apple Silicon run has already been captured in:
+
+- `benchmarks/reports/dflash_m3max_36gb_qwen35_4b_pilot.md`
+
+Pilot command:
+
+```bash
+python -m dflash.benchmark --backend mlx \
+    --model Qwen/Qwen3.5-4B \
+    --draft-model z-lab/Qwen3.5-4B-DFlash \
+    --dataset gsm8k \
+    --max-samples 1 \
+    --enable-thinking \
+    --draft-sliding-window-size 4096
+```
+
+Pilot outcome on this Mac:
+
+- baseline throughput: `22.35 tok/s`
+- DFlash throughput: `46.78 tok/s`
+- decoding speedup: `2.09x`
+
+Treat that as a **directional proof**, not a final decision benchmark. The next step is the fuller comparison slice against plain MLX or llama.cpp speculative decoding.
+
+## Known 9B failure on this machine
+
+A follow-up live run with:
+
+- `Qwen/Qwen3.5-9B`
+- `z-lab/Qwen3.5-9B-DFlash`
+
+failed on this same M3 Max 36GB Mac with:
+
+```text
+[METAL] Command buffer execution failed:
+Caused GPU Timeout Error (00000002:kIOGPUCommandBufferCallbackErrorTimeout)
+```
+
+That failure is recorded in:
+
+- `benchmarks/reports/dflash_m3max_36gb_qwen35_9b_timeout.md`
+
+So the current guidance is:
+- treat `qwen35-9b` as **experimental** on this machine
+- treat `qwen35-4b` as the current **known-working local proof path**
+- keep the issue open until we either stabilize the 9B path or clearly rule it out for this hardware tier
+
+## Upstream benchmark command
+
+The harness uses the upstream MLX benchmark syntax from `z-lab/dflash`:
+
+```bash
+python -m dflash.benchmark --backend mlx \
+    --model Qwen/Qwen3.5-9B \
+    --draft-model z-lab/Qwen3.5-9B-DFlash \
+    --dataset gsm8k \
+    --max-samples 128 \
+    --enable-thinking \
+    --draft-sliding-window-size 4096
+```
+
+## What remains
+
+This PR adds the **planner + report template** so the benchmark is reproducible from the repo.
+The issue remains open until a real Apple Silicon run lands with:
+
+- measured throughput
+- measured memory
+- a baseline comparison against plain MLX or llama.cpp speculative decoding
+- a recommendation on whether to operationalize DFlash locally

docs/edge-crisis-deployment.md

@@ -1,223 +0,0 @@
-# Edge Crisis Detection — Deployment Guide
-
-**Part of:** turboquant#99 (1-Bit Models + Edge)
-**Issue:** #102
-
-## Overview
-
-Deploy a minimal crisis detection system on edge devices for offline use.
-When internet is unavailable but someone is in crisis, a local device can
-detect distress signals and display cached crisis resources.
-
-## Target Hardware
-
-| Device | RAM | Notes |
-|--------|-----|-------|
-| Raspberry Pi 4 | 4GB | Recommended. Runs keyword + Falcon-H1-Tiny-90M |
-| Raspberry Pi 4 | 2GB | Keyword detection only (no LLM) |
-| Old Android phone | 2GB+ | Termux + llama.cpp, Falcon-H1-Tiny-90M |
-| Any x86 SBC | 2GB+ | Full keyword + optional small model |
-
-## Model Selection
-
-### Tier 0: Keyword Detection (any device, <10MB)
-- No model needed — pure pattern matching
-- Instant response (<1ms)
-- Works on 512MB RAM devices
-- Covers 80%+ of explicit crisis language
-- **Use when:** RAM < 2GB or first-boot before model download
-
-### Tier 1: Falcon-H1-Tiny-90M (~180MB quantized)
-- Detects nuanced/implicit distress that keywords miss
-- Runs on 2GB+ RAM (Pi 4 4GB recommended)
-- ~200ms inference on Pi 4 (CPU)
-- Quantized Q4_K_M via llama.cpp
-- **Use when:** RAM >= 2GB, want higher recall
-
-### Tier 2: Bonsai-1.7B (~900MB quantized)
-- Best accuracy for ambiguous cases
-- Needs 3GB+ RAM
-- ~1.5s inference on Pi 4
-- **Use when:** RAM >= 4GB, false-positive tolerance is low
-
-### Recommendation
-Start with **Tier 0 + Tier 1**. Keyword catches obvious cases instantly,
-Falcon-H1 catches implicit cases with 200ms latency. Together they cover
->95% of crisis signals with negligible resource use.
-
-## Installation
-
-### Raspberry Pi 4
-
-```bash
-# 1. System setup
-sudo apt update && sudo apt install -y python3 python3-pip git cmake
-
-# 2. Clone this directory
-git clone https://forge.alexanderwhitestone.com/Timmy_Foundation/turboquant.git
-cd turboquant
-
-# 3. Python keyword detector runs with zero dependencies (pure stdlib)
-
-# 4. (Optional) Build llama.cpp for Tier 1 model
-git clone https://github.com/ggerganov/llama.cpp
-cd llama.cpp && make -j4 && cd ..
-
-# 5. Download model (Tier 1)
-mkdir -p models
-# Falcon-H1-Tiny-90M GGUF — find latest on HuggingFace
-# wget -O models/falcon-h1-tiny-90m-q4km.gguf <URL>
-
-# 6. Test offline crisis detection
-python3 scripts/crisis_detector.py --test
-```
-
-### Android (Termux)
-
-```bash
-pkg install python git cmake
-# Follow Pi steps above, but build llama.cpp with:
-cmake -B build -DLLAMA_NATIVE=OFF && cmake --build build -j$(nproc)
-```
-
-### Auto-Start on Boot (Pi)
-
-```bash
-# Add to /etc/rc.local (before 'exit 0'):
-python3 /home/pi/turboquant/scripts/crisis_detector.py --daemon &
-```
-
-Or create a systemd service:
-
-```ini
-# /etc/systemd/system/crisis-detector.service
-[Unit]
-Description=Edge Crisis Detector
-After=network.target
-
-[Service]
-ExecStart=/usr/bin/python3 /home/pi/turboquant/scripts/crisis_detector.py --daemon
-Restart=always
-User=pi
-
-[Install]
-WantedBy=multi-user.target
-```
-
-```bash
-sudo systemctl enable crisis-detector
-sudo systemctl start crisis-detector
-```
-
-## Offline Resource Cache
-
-The file `data/crisis_resources.json` is bundled with the deployment.
-It contains:
-
-- **988 Suicide & Crisis Lifeline** — call or text 988
-- **Crisis Text Line** — text HOME to 741741
-- **International Association for Suicide Prevention** — global directory
-- Cached local resources (customize per deployment location)
-
-These display immediately when a crisis is detected — no network required.
-
-## How It Works
-
-```
-User input
-    |
-    v
-+-------------------+
-| Keyword Matcher   | <- Tier 0: instant, no model
-| (regex/pattern)   |
-+--------+----------+
-  match? --yes--> Show crisis resources
-    |
-    no
-    v
-+-------------------+
-| Falcon-H1-Tiny    | <- Tier 1: ~200ms on Pi 4
-| (if available)    |
-+--------+----------+
-  crisis? --yes--> Show crisis resources
-    |
-    no
-    v
-  Continue normally
-```
-
-## Testing Offline
-
-```bash
-# Disconnect from internet
-sudo ip link set wlan0 down
-
-# Run the test suite
-python3 scripts/crisis_detector.py --test
-
-# Expected: all tests pass, resources display correctly
-
-# Reconnect
-sudo ip link set wlan0 up
-```
-
-## File Structure
-
-```
-turboquant/
-+-- scripts/
-|   +-- crisis_detector.py      # Main detector (keyword + optional LLM)
-+-- data/
-|   +-- crisis_resources.json   # Offline resource cache
-+-- tests/
-|   +-- test_edge_crisis.py     # Offline verification tests
-+-- docs/
-    +-- edge-crisis-deployment.md  # This file
-```
-
-## Customization
-
-### Adding Local Resources
-
-Edit `data/crisis_resources.json`:
-
-```json
-{
-  "local_resources": [
-    {
-      "name": "City Crisis Center",
-      "phone": "555-0123",
-      "address": "123 Main St",
-      "hours": "24/7"
-    }
-  ]
-}
-```
-
-### Adjusting Sensitivity
-
-In `scripts/crisis_detector.py`:
-
-```python
-# Keyword threshold: how many keywords trigger a match
-KEYWORD_THRESHOLD = 1  # 1 = any keyword triggers (high recall)
-                       # 2 = need 2+ keywords (higher precision)
-
-# LLM threshold (Tier 1/2): confidence score cutoff
-LLM_THRESHOLD = 0.6    # 0.6 = default (balanced)
-                       # 0.4 = more sensitive
-                       # 0.8 = more precise
-```
-
-## Privacy
-
-- **No data leaves the device.** All detection runs locally.
-- No logs of user input are stored by default.
-- Enable logging only for debugging (`--log` flag).
-- No network calls are made by the crisis detector.
-- Resource display is a local text render.
-
-## License
-
-Same as parent project. Crisis detection code and resource data are
-provided for humanitarian purposes.

evolution/quant_selector.py

@@ -0,0 +1,548 @@
+"""Auto-select TurboQuant compression level based on available VRAM/RAM.
+
+Detects hardware resources at startup and picks the highest quality
+quantization level that fits within available memory. Supports Apple
+Silicon unified memory, NVIDIA GPUs (via nvidia-smi), and CPU-only fallback.
+
+Usage:
+    from evolution.quant_selector import select_quant_level
+
+    selection = select_quant_level(model_size_gb=14.0, context_length=32768)
+    print(selection.level)       # "turbo4"
+    print(selection.reasoning)   # "M4 Max 36GB unified: turbo4 fits 14.0GB model + ..."
+    print(selection.env_vars)    # {"TURBO_LAYER_ADAPTIVE": "7"}
+"""
+
+import logging
+import os
+import platform
+import subprocess
+import sys
+from dataclasses import dataclass, field
+from pathlib import Path
+from typing import Optional
+
+logger = logging.getLogger(__name__)
+
+
+# ── Quant Level Definitions ───────────────────────────────────────────────────
+
+@dataclass
+class QuantLevel:
+    """A TurboQuant compression level with its memory characteristics."""
+    name: str               # e.g. "turbo4"
+    bits_per_channel: float # e.g. 3.5 for turbo4
+    compression_ratio: float  # vs uncompressed KV cache
+    quality_label: str      # "best", "high", "balanced", "fast"
+    layer_adaptive: int     # TURBO_LAYER_ADAPTIVE value (0-7)
+    kv_type: str            # -ctk/-ctv flag value
+    min_memory_headroom_gb: float  # Minimum free memory to recommend this level
+    description: str = ""
+
+
+# Ordered from highest quality to most aggressive compression
+QUANT_LEVELS = [
+    QuantLevel(
+        name="turbo4",
+        bits_per_channel=3.5,
+        compression_ratio=4.2,
+        quality_label="best",
+        layer_adaptive=7,
+        kv_type="turbo4",
+        min_memory_headroom_gb=4.0,
+        description="PolarQuant + QJL 4-bit. Best quality, ~4.2x KV compression."
+    ),
+    QuantLevel(
+        name="turbo3",
+        bits_per_channel=2.5,
+        compression_ratio=6.0,
+        quality_label="high",
+        layer_adaptive=5,
+        kv_type="turbo3",
+        min_memory_headroom_gb=3.0,
+        description="3-bit TurboQuant. High quality, ~6x KV compression."
+    ),
+    QuantLevel(
+        name="turbo2",
+        bits_per_channel=1.5,
+        compression_ratio=10.0,
+        quality_label="balanced",
+        layer_adaptive=3,
+        kv_type="turbo2",
+        min_memory_headroom_gb=2.0,
+        description="2-bit TurboQuant. Balanced, ~10x KV compression."
+    ),
+    QuantLevel(
+        name="q4_0",
+        bits_per_channel=4.0,
+        compression_ratio=3.5,
+        quality_label="fast",
+        layer_adaptive=0,
+        kv_type="q4_0",
+        min_memory_headroom_gb=1.5,
+        description="Standard 4-bit quant. Fast fallback, no TurboQuant."
+    ),
+]
+
+
+# ── Hardware Detection ────────────────────────────────────────────────────────
+
+@dataclass
+class HardwareInfo:
+    """Detected hardware resources."""
+    total_memory_gb: float
+    available_memory_gb: float
+    gpu_memory_gb: Optional[float] = None
+    gpu_name: Optional[str] = None
+    is_apple_silicon: bool = False
+    chip_name: Optional[str] = None
+    cpu_cores: int = 0
+    detection_method: str = ""
+
+
+def detect_hardware() -> HardwareInfo:
+    """Detect available memory and GPU resources."""
+    system = platform.system()
+
+    if system == "Darwin":
+        return _detect_apple_silicon()
+    elif system == "Linux":
+        return _detect_linux()
+    else:
+        return _detect_generic(system)
+
+
+def _detect_apple_silicon() -> HardwareInfo:
+    """Detect Apple Silicon unified memory."""
+    info = HardwareInfo(
+        total_memory_gb=0,
+        available_memory_gb=0,
+        is_apple_silicon=True,
+        detection_method="sysctl",
+    )
+
+    try:
+        # Get total memory
+        result = subprocess.run(
+            ["sysctl", "-n", "hw.memsize"],
+            capture_output=True, text=True, timeout=5
+        )
+        if result.returncode == 0:
+            info.total_memory_gb = int(result.stdout.strip()) / (1024**3)
+
+        # Get chip name
+        result = subprocess.run(
+            ["sysctl", "-n", "machdep.cpu.brand_string"],
+            capture_output=True, text=True, timeout=5
+        )
+        if result.returncode == 0:
+            info.chip_name = result.stdout.strip()
+
+        # Try to get GPU name (Apple Silicon)
+        result = subprocess.run(
+            ["system_profiler", "SPDisplaysDataType"],
+            capture_output=True, text=True, timeout=10
+        )
+        if result.returncode == 0:
+            for line in result.stdout.split("\n"):
+                if "Chipset" in line or "GPU" in line:
+                    info.gpu_name = line.split(":")[-1].strip()
+                    break
+
+        # Estimate available memory (vm_stat)
+        result = subprocess.run(
+            ["vm_stat"],
+            capture_output=True, text=True, timeout=5
+        )
+        if result.returncode == 0:
+            page_size = 4096  # macOS default
+            free_pages = 0
+            for line in result.stdout.split("\n"):
+                if "Pages free:" in line:
+                    try:
+                        free_pages = int(line.split(":")[-1].strip().rstrip("."))
+                    except ValueError:
+                        pass
+            # Available ≈ free + some speculative (conservative: just free)
+            info.available_memory_gb = (free_pages * page_size) / (1024**3)
+
+        # Fallback if vm_stat parsing failed
+        if info.available_memory_gb < 1:
+            # Conservative: 70% of total
+            info.available_memory_gb = info.total_memory_gb * 0.70
+
+        # Apple Silicon shares memory — GPU memory = total memory
+        info.gpu_memory_gb = info.total_memory_gb
+
+        # Detect CPU cores
+        result = subprocess.run(
+            ["sysctl", "-n", "hw.ncpu"],
+            capture_output=True, text=True, timeout=5
+        )
+        if result.returncode == 0:
+            info.cpu_cores = int(result.stdout.strip())
+
+    except Exception as e:
+        logger.warning(f"Apple Silicon detection failed: {e}")
+        # Fallback
+        info.total_memory_gb = 16.0
+        info.available_memory_gb = 12.0
+        info.detection_method = "fallback"
+
+    return info
+
+
+def _detect_linux() -> HardwareInfo:
+    """Detect Linux system with optional NVIDIA GPU."""
+    info = HardwareInfo(
+        total_memory_gb=0,
+        available_memory_gb=0,
+        detection_method="proc",
+    )
+
+    try:
+        # Read /proc/meminfo
+        with open("/proc/meminfo", "r") as f:
+            meminfo = f.read()
+
+        for line in meminfo.split("\n"):
+            if line.startswith("MemTotal:"):
+                kb = int(line.split()[1])
+                info.total_memory_gb = kb / (1024 * 1024)
+            elif line.startswith("MemAvailable:"):
+                kb = int(line.split()[1])
+                info.available_memory_gb = kb / (1024 * 1024)
+
+        # CPU cores
+        info.cpu_cores = os.cpu_count() or 1
+
+        # Check for NVIDIA GPU
+        try:
+            result = subprocess.run(
+                ["nvidia-smi", "--query-gpu=name,memory.total,memory.free",
+                 "--format=csv,noheader,nounits"],
+                capture_output=True, text=True, timeout=10
+            )
+            if result.returncode == 0 and result.stdout.strip():
+                lines = result.stdout.strip().split("\n")
+                if lines:
+                    parts = lines[0].split(", ")
+                    if len(parts) >= 3:
+                        info.gpu_name = parts[0].strip()
+                        info.gpu_memory_gb = float(parts[1]) / 1024  # MB to GB
+                        gpu_free = float(parts[2]) / 1024
+                        # Use GPU free for VRAM-based selection
+                        info.available_memory_gb = max(info.available_memory_gb, gpu_free)
+                        info.detection_method = "nvidia-smi"
+        except (FileNotFoundError, subprocess.TimeoutExpired):
+            pass  # No NVIDIA GPU
+
+    except Exception as e:
+        logger.warning(f"Linux detection failed: {e}")
+        info.total_memory_gb = 16.0
+        info.available_memory_gb = 12.0
+        info.detection_method = "fallback"
+
+    return info
+
+
+def _detect_generic(system: str) -> HardwareInfo:
+    """Fallback detection for unknown systems."""
+    import psutil
+    mem = psutil.virtual_memory()
+    return HardwareInfo(
+        total_memory_gb=mem.total / (1024**3),
+        available_memory_gb=mem.available / (1024**3),
+        cpu_cores=os.cpu_count() or 1,
+        detection_method="psutil",
+    )
+
+
+# ── KV Cache Memory Estimation ───────────────────────────────────────────────
+
+def estimate_kv_cache_gb(
+    context_length: int,
+    num_layers: int = 48,
+    num_kv_heads: int = 8,
+    head_dim: int = 128,
+    bits_per_channel: float = 3.5,
+) -> float:
+    """Estimate KV cache memory for given parameters.
+
+    Formula: 2 (K+V) × layers × kv_heads × head_dim × context_length × bits/8
+    """
+    bytes_per_element = bits_per_channel / 8.0
+    total_bytes = 2 * num_layers * num_kv_heads * head_dim * context_length * bytes_per_element
+    return total_bytes / (1024**3)
+
+
+def estimate_model_memory_gb(model_size_gb: float, quant_type: str = "q4_k_m") -> float:
+    """Estimate model weights memory. Returns loaded size in GB.
+
+    This is a rough estimate — actual depends on exact quant format.
+    """
+    # Common quant ratios (vs fp16)
+    quant_multipliers = {
+        "f16": 1.0,
+        "q8_0": 0.5,
+        "q6_k": 0.42,
+        "q5_k_m": 0.37,
+        "q4_k_m": 0.32,
+        "q3_k_m": 0.27,
+        "q2_k": 0.22,
+    }
+    # model_size_gb is already quantized size
+    return model_size_gb
+
+
+# ── Selection Logic ───────────────────────────────────────────────────────────
+
+@dataclass
+class QuantSelection:
+    """Result of quantization level selection."""
+    level: QuantLevel
+    hardware: HardwareInfo
+    reasoning: str
+    total_required_gb: float
+    available_gb: float
+    headroom_gb: float
+    env_vars: dict = field(default_factory=dict)
+    server_flags: dict = field(default_factory=dict)
+    warnings: list = field(default_factory=list)
+
+
+def select_quant_level(
+    model_size_gb: float = 14.0,
+    context_length: int = 32768,
+    num_layers: int = 48,
+    num_kv_heads: int = 8,
+    head_dim: int = 128,
+    preferred_level: Optional[str] = None,
+    force_cpu: bool = False,
+) -> QuantSelection:
+    """Select the best quantization level for available hardware.
+
+    Args:
+        model_size_gb: Size of the model weights in GB
+        context_length: Target context length
+        num_layers: Number of transformer layers
+        num_kv_heads: Number of KV attention heads
+        head_dim: Dimension per attention head
+        preferred_level: Force a specific level (still checks if it fits)
+        force_cpu: If True, ignore GPU memory
+
+    Returns:
+        QuantSelection with the chosen level and reasoning
+    """
+    hw = detect_hardware()
+
+    if force_cpu:
+        hw.gpu_memory_gb = None
+        hw.gpu_name = None
+
+    # Use the most restrictive memory constraint
+    # For Apple Silicon: unified memory, use total
+    # For NVIDIA: use GPU VRAM
+    # For CPU-only: use system RAM
+    if hw.gpu_memory_gb and hw.gpu_name:
+        memory_pool_gb = hw.gpu_memory_gb
+        memory_label = f"{hw.gpu_name} {hw.gpu_memory_gb:.0f}GB VRAM"
+    elif hw.is_apple_silicon:
+        memory_pool_gb = hw.total_memory_gb
+        memory_label = f"{hw.chip_name or 'Apple Silicon'} {hw.total_memory_gb:.0f}GB unified"
+    else:
+        memory_pool_gb = hw.total_memory_gb
+        memory_label = f"{hw.cpu_cores}c CPU {hw.total_memory_gb:.0f}GB RAM"
+
+    model_mem = estimate_model_memory_gb(model_size_gb)
+
+    # Try levels from best to most compressed
+    chosen = None
+    for level in QUANT_LEVELS:
+        if preferred_level and level.name != preferred_level:
+            continue
+
+        kv_mem = estimate_kv_cache_gb(
+            context_length, num_layers, num_kv_heads, head_dim,
+            level.bits_per_channel
+        )
+        total_required = model_mem + kv_mem
+        headroom = memory_pool_gb - total_required
+
+        if headroom >= level.min_memory_headroom_gb:
+            chosen = level
+            break
+
+        if preferred_level and level.name == preferred_level:
+            # User forced this level but it doesn't fit
+            chosen = level
+            break
+
+    if chosen is None:
+        # Nothing fits — pick the most aggressive compression, not the q4_0 fallback.
+        chosen = max(QUANT_LEVELS, key=lambda level: level.compression_ratio)
+        logger.warning(f"No quant level fits in {memory_pool_gb:.1f}GB. Using {chosen.name}.")
+
+    # Calculate final numbers
+    kv_mem = estimate_kv_cache_gb(
+        context_length, num_layers, num_kv_heads, head_dim,
+        chosen.bits_per_channel
+    )
+    total_required = model_mem + kv_mem
+    headroom = memory_pool_gb - total_required
+
+    # Build reasoning
+    reasoning_parts = [
+        f"{memory_label}:",
+        f"{chosen.name} ({chosen.quality_label}, {chosen.bits_per_channel:.1f}b/ch,",
+        f"{chosen.compression_ratio:.1f}x compression)",
+        f"fits {model_mem:.1f}GB model + {kv_mem:.1f}GB KV cache",
+        f"@ {context_length}K context = {total_required:.1f}GB / {memory_pool_gb:.0f}GB",
+        f"({headroom:.1f}GB headroom)"
+    ]
+    reasoning = " ".join(reasoning_parts)
+
+    # Build environment variables for llama.cpp
+    env_vars = {
+        "TURBO_LAYER_ADAPTIVE": str(chosen.layer_adaptive),
+    }
+
+    # Build server flags
+    server_flags = {
+        "-ctk": chosen.kv_type,
+        "-ctv": chosen.kv_type,
+        "-c": str(context_length),
+    }
+
+    # Warnings
+    warnings = []
+    if headroom < 2.0:
+        warnings.append(
+            f"Low headroom ({headroom:.1f}GB). Consider reducing context length or model size."
+        )
+    if headroom < 0:
+        warnings.append(
+            f"OVERCOMMITTED: needs {total_required:.1f}GB but only {memory_pool_gb:.0f}GB available. "
+            f"Inference may fail or swap heavily."
+        )
+
+    selection = QuantSelection(
+        level=chosen,
+        hardware=hw,
+        reasoning=reasoning,
+        total_required_gb=total_required,
+        available_gb=memory_pool_gb,
+        headroom_gb=headroom,
+        env_vars=env_vars,
+        server_flags=server_flags,
+        warnings=warnings,
+    )
+
+    logger.info(f"Quant selection: {reasoning}")
+    for w in warnings:
+        logger.warning(w)
+
+    return selection
+
+
+# ── CLI ───────────────────────────────────────────────────────────────────────
+
+def main():
+    """CLI entry point for quant level selection."""
+    import argparse
+    import json
+
+    parser = argparse.ArgumentParser(
+        description="Auto-select TurboQuant compression level based on available hardware"
+    )
+    parser.add_argument("--model-size", type=float, default=14.0,
+                        help="Model size in GB (default: 14.0)")
+    parser.add_argument("--context", type=int, default=32768,
+                        help="Target context length (default: 32768)")
+    parser.add_argument("--layers", type=int, default=48,
+                        help="Number of transformer layers (default: 48)")
+    parser.add_argument("--kv-heads", type=int, default=8,
+                        help="Number of KV attention heads (default: 8)")
+    parser.add_argument("--head-dim", type=int, default=128,
+                        help="Dimension per attention head (default: 128)")
+    parser.add_argument("--prefer", type=str, default=None,
+                        choices=[l.name for l in QUANT_LEVELS],
+                        help="Prefer a specific quant level")
+    parser.add_argument("--force-cpu", action="store_true",
+                        help="Ignore GPU, use CPU memory only")
+    parser.add_argument("--json", action="store_true",
+                        help="JSON output for automation")
+    parser.add_argument("--detect-only", action="store_true",
+                        help="Only detect hardware, don't select")
+    args = parser.parse_args()
+
+    logging.basicConfig(level=logging.INFO, format="%(message)s")
+
+    if args.detect_only:
+        hw = detect_hardware()
+        if args.json:
+            print(json.dumps(hw.__dict__, default=str, indent=2))
+        else:
+            print(f"Total memory:    {hw.total_memory_gb:.1f} GB")
+            print(f"Available:       {hw.available_memory_gb:.1f} GB")
+            if hw.gpu_memory_gb:
+                print(f"GPU memory:      {hw.gpu_memory_gb:.1f} GB")
+            if hw.gpu_name:
+                print(f"GPU:             {hw.gpu_name}")
+            if hw.is_apple_silicon:
+                print(f"Chip:            {hw.chip_name or 'Apple Silicon'}")
+            print(f"CPU cores:       {hw.cpu_cores}")
+            print(f"Detection:       {hw.detection_method}")
+        return
+
+    selection = select_quant_level(
+        model_size_gb=args.model_size,
+        context_length=args.context,
+        num_layers=args.layers,
+        num_kv_heads=args.kv_heads,
+        head_dim=args.head_dim,
+        preferred_level=args.prefer,
+        force_cpu=args.force_cpu,
+    )
+
+    if args.json:
+        result = {
+            "level": selection.level.name,
+            "bits_per_channel": selection.level.bits_per_channel,
+            "compression_ratio": selection.level.compression_ratio,
+            "quality": selection.level.quality_label,
+            "reasoning": selection.reasoning,
+            "total_required_gb": round(selection.total_required_gb, 2),
+            "available_gb": round(selection.available_gb, 1),
+            "headroom_gb": round(selection.headroom_gb, 2),
+            "env_vars": selection.env_vars,
+            "server_flags": selection.server_flags,
+            "warnings": selection.warnings,
+            "hardware": {
+                "total_memory_gb": round(selection.hardware.total_memory_gb, 1),
+                "gpu_name": selection.hardware.gpu_name,
+                "is_apple_silicon": selection.hardware.is_apple_silicon,
+                "chip_name": selection.hardware.chip_name,
+                "cpu_cores": selection.hardware.cpu_cores,
+            },
+        }
+        print(json.dumps(result, indent=2))
+    else:
+        print(f"Selected: {selection.level.name} ({selection.level.quality_label})")
+        print(f"  {selection.reasoning}")
+        print()
+        print(f"Environment variables:")
+        for k, v in selection.env_vars.items():
+            print(f"  export {k}={v}")
+        print()
+        print(f"Server flags:")
+        for k, v in selection.server_flags.items():
+            print(f"  {k} {v}")
+        if selection.warnings:
+            print()
+            for w in selection.warnings:
+                print(f"  WARNING: {w}")
+
+
+if __name__ == "__main__":
+    main()

scripts/crisis_detector.py

@@ -1,386 +0,0 @@
-#!/usr/bin/env python3
-"""
-crisis_detector.py — Edge crisis detection (keyword + optional LLM)
-
-Zero-dependency crisis detector for offline edge devices.
-Tier 0: Keyword matching (instant, any device)
-Tier 1: Small LLM via llama.cpp (200ms on Pi 4, optional)
-
-Usage:
-    python3 crisis_detector.py                    # Interactive mode
-    python3 crisis_detector.py --test             # Run self-tests
-    python3 crisis_detector.py --daemon           # Headless daemon mode
-    python3 crisis_detector.py --check "text"     # Single check
-
-Part of turboquant #99, #102.
-"""
-
-import json
-import os
-import re
-import sys
-import time
-from pathlib import Path
-
-# ── Paths ───────────────────────────────────────────────────────────────────
-
-SCRIPT_DIR = Path(__file__).resolve().parent
-DATA_DIR = SCRIPT_DIR.parent / "data"
-RESOURCES_FILE = DATA_DIR / "crisis_resources.json"
-
-# ── Configuration ───────────────────────────────────────────────────────────
-
-KEYWORD_THRESHOLD = 1       # Keywords needed to trigger (1 = high recall)
-LLM_THRESHOLD = 0.6         # LLM confidence cutoff
-LLM_TIMEOUT_MS = 5000       # Max LLM inference time
-LOG_FILE = None             # Set by --log flag
-
-# ── Tier 0: Keyword Detection ──────────────────────────────────────────────
-
-# Crisis keywords organized by severity
-CRISIS_PATTERNS = {
-    "immediate_danger": [
-        r"\bkill(?:ing)?\s*(?:my)?self\b",
-        r"\b(?:going|gonna|want(?:ing)?)\s*to\s*die\b",
-        r"\b(?:end|ending)\s+(?:my|it|this)\s*life\b",
-        r"\bsuicide\b",
-        r"\bsuicidal\b",
-        r"\bwanna\s*die\b",
-        r"\bwant\s*to\s*die\b",
-        r"\bcan'?t\s+(?:go\s+on|live|take\s+(?:it|this|anymore))\b",
-        r"\bi'?m\s+(?:done|finished)\b.*\b(?:life|living)\b",
-        r"\boverdose\b",
-        r"\bhanging\s+(?:my)?self\b",
-        r"\bjump(?:ing)?\s+(?:off|from)\b",
-        r"\bslit(?:ting)?\s+(?:my\s+)?(?:wrists|throat)\b",
-        r"\bshoot(?:ing)?\s+(?:my)?self\b",
-    ],
-    "self_harm": [
-        r"\bself[\s-]?harm(?:ing|ed)?\b",
-        r"\bcut(?:ting)?\s+(?:my)?self\b",
-        r"\bhurt(?:ing)?\s+(?:my)?self\b",
-        r"\bpunish(?:ing)?\s+(?:my)?self\b",
-        r"\bburn(?:ing)?\s+(?:my)?self\b",
-        r"\bscar(?:ring)?\s+(?:my)?self\b",
-    ],
-    "hopelessness": [
-        r"\bhopeless\b",
-        r"\bno\s+(?:point|reason|purpose)\b",
-        r"\bwhy\s+(?:bother|try|am\s+i\s+here)\b",
-        r"\bnobody\s+(?:cares|would\s+(?:miss|notice))\b",
-        r"\bbeen\s+better\s+off\s+(?:dead|gone)\b",
-        r"\bwouldn'?t\s+(?:miss|care)\b.*\b(?:if|when)\b.*\bdie\b",
-        r"\bnothing\s+(?:matters|left)\b",
-        r"\bgive\s+(?:up|me\s+death)\b",
-    ],
-    "crisis_language": [
-        r"\b(?:i|can'?t)\s+(?:handle|deal\s+with)\s+(?:this|it|anymore)\b",
-        r"\btoo\s+much\s+(?:pain|suffering)\b",
-        r"\bcan'?t\s+(?:take|stand)\s+(?:this|it|anymore)\b",
-        r"\bbreak(?:ing|s)?\s+down\b",
-        r"\b(?:i'?m|am)\s+(?:drowning|suffocating|dying)\b",
-        r"\bsos\b",
-        r"\bhelp\s+me\b.*\b(?:please|desperate)\b",
-        r"\bemergency\b.*\b(?:mental|crisis)\b",
-        r"\b(?:want|need|wish)(?:ing)?\s+(?:the|this|my)\s+pain\s+to\s+(?:stop|end|go\s+away)\b",
-        r"\bmake\s+(?:the|this|my)\s+pain\s+(?:stop|end|go\s+away)\b",
-    ],
-}
-
-# Compile all patterns
-_COMPILED_PATTERNS = {}
-for category, patterns in CRISIS_PATTERNS.items():
-    _COMPILED_PATTERNS[category] = [re.compile(p, re.IGNORECASE) for p in patterns]
-
-
-def detect_keywords(text: str) -> dict:
-    """
-    Tier 0 keyword detection. Returns match info.
-    Result: {
-        "detected": bool,
-        "confidence": float (0-1),
-        "categories": list[str],
-        "matches": list[str]
-    }
-    """
-    matches = []
-    categories = set()
-
-    for category, patterns in _COMPILED_PATTERNS.items():
-        for pattern in patterns:
-            m = pattern.search(text)
-            if m:
-                matches.append(m.group(0))
-                categories.add(category)
-
-    detected = len(matches) >= KEYWORD_THRESHOLD
-
-    # Confidence heuristic
-    if not detected:
-        confidence = 0.0
-    elif "immediate_danger" in categories:
-        confidence = 0.95
-    elif "self_harm" in categories:
-        confidence = 0.85
-    elif "hopelessness" in categories:
-        confidence = 0.7
-    else:
-        confidence = 0.6
-
-    return {
-        "detected": detected,
-        "confidence": confidence,
-        "categories": sorted(categories),
-        "matches": matches,
-        "method": "keyword",
-    }
-
-
-# ── Tier 1: LLM Detection (Optional) ───────────────────────────────────────
-
-def detect_llm(text: str, model_path: str = None) -> dict:
-    """
-    Tier 1 LLM detection via llama.cpp subprocess.
-    Falls back gracefully if model not available.
-    """
-    if model_path is None:
-        default_path = SCRIPT_DIR.parent / "models" / "falcon-h1-tiny-90m-q4km.gguf"
-        model_path = str(default_path)
-
-    if not os.path.exists(model_path):
-        return {
-            "detected": False,
-            "confidence": 0.0,
-            "method": "llm_unavailable",
-            "reason": "Model not found at " + model_path,
-        }
-
-    # Build llama-cli command
-    llama_bin = SCRIPT_DIR.parent / "llama.cpp" / "llama-cli"
-    if not llama_bin.exists():
-        llama_bin = SCRIPT_DIR.parent / "llama.cpp" / "build" / "bin" / "llama-cli"
-
-    if not llama_bin.exists():
-        return {
-            "detected": False,
-            "confidence": 0.0,
-            "method": "llm_unavailable",
-            "reason": "llama-cli binary not found",
-        }
-
-    prompt = (
-        "You are a crisis detection system. Analyze the following text and "
-        "determine if the person is in psychological crisis or expressing "
-        "suicidal ideation.\n\n"
-        f"Text: \"{text}\"\n\n"
-        "Respond with ONLY one word: CRISIS or SAFE\n"
-    )
-
-    try:
-        import subprocess
-        result = subprocess.run(
-            [
-                str(llama_bin),
-                "-m", model_path,
-                "-p", prompt,
-                "-n", "5",
-                "--temp", "0.1",
-                "--no-display-prompt",
-            ],
-            capture_output=True,
-            text=True,
-            timeout=LLM_TIMEOUT_MS / 1000,
-        )
-        output = result.stdout.strip().upper()
-
-        if "CRISIS" in output:
-            return {
-                "detected": True,
-                "confidence": 0.8,
-                "method": "llm",
-                "raw_output": output[:100],
-            }
-        else:
-            return {
-                "detected": False,
-                "confidence": 0.9,
-                "method": "llm",
-                "raw_output": output[:100],
-            }
-    except Exception as e:
-        return {
-            "detected": False,
-            "confidence": 0.0,
-            "method": "llm_error",
-            "reason": str(e),
-        }
-
-
-# ── Combined Detection ──────────────────────────────────────────────────────
-
-def detect_crisis(text: str, use_llm: bool = True) -> dict:
-    """
-    Full crisis detection pipeline: keyword first, then LLM if available.
-    """
-    kw_result = detect_keywords(text)
-
-    if kw_result["detected"]:
-        return kw_result
-
-    if use_llm:
-        llm_result = detect_llm(text)
-        if llm_result["detected"]:
-            return llm_result
-
-    return {
-        "detected": False,
-        "confidence": 0.95,
-        "categories": [],
-        "matches": [],
-        "method": "keyword+llm",
-    }
-
-
-# ── Resource Display ────────────────────────────────────────────────────────
-
-def load_resources() -> dict:
-    """Load offline crisis resources."""
-    if RESOURCES_FILE.exists():
-        with open(RESOURCES_FILE) as f:
-            return json.load(f)
-    return {
-        "national_resources": [{
-            "name": "988 Suicide & Crisis Lifeline",
-            "phone": "988",
-            "description": "Call or text 988 — free, confidential, 24/7",
-        }],
-        "local_resources": [],
-    }
-
-
-def display_resources(result: dict) -> str:
-    """Format crisis resources for display."""
-    resources = load_resources()
-    lines = []
-    lines.append("=" * 50)
-    lines.append("  CRISIS RESOURCES — You are not alone")
-    lines.append("=" * 50)
-    lines.append("")
-
-    for r in resources.get("national_resources", []):
-        lines.append(f"  {r['name']}")
-        lines.append(f"    Phone: {r['phone']}")
-        if r.get("description"):
-            lines.append(f"    {r['description']}")
-        lines.append("")
-
-    for r in resources.get("local_resources", []):
-        lines.append(f"  {r['name']}")
-        if r.get("phone"):
-            lines.append(f"    Phone: {r['phone']}")
-        if r.get("address"):
-            lines.append(f"    Address: {r['address']}")
-        if r.get("hours"):
-            lines.append(f"    Hours: {r['hours']}")
-        lines.append("")
-
-    lines.append("-" * 50)
-    lines.append("  Detection: " + result.get("method", "keyword"))
-    lines.append("  Confidence: " + str(int(result.get("confidence", 0) * 100)) + "%")
-    if result.get("categories"):
-        lines.append("  Categories: " + ", ".join(result["categories"]))
-    lines.append("=" * 50)
-
-    return "\n".join(lines)
-
-
-# ── CLI Interface ───────────────────────────────────────────────────────────
-
-def run_tests():
-    """Run self-tests."""
-    from tests.test_edge_crisis import run_all_tests
-    return run_all_tests()
-
-
-def run_check(text: str):
-    """Single text check."""
-    result = detect_crisis(text, use_llm=False)
-    if result["detected"]:
-        print(display_resources(result))
-    else:
-        print("SAFE — no crisis indicators detected")
-    return result
-
-
-def run_interactive():
-    """Interactive mode — read lines from stdin, detect crisis."""
-    resources = load_resources()
-    print("Edge Crisis Detector (Ctrl+C to exit)")
-    print("Type a message and press Enter to check.\n")
-
-    try:
-        while True:
-            try:
-                text = input("> ").strip()
-            except EOFError:
-                break
-            if not text:
-                continue
-
-            result = detect_crisis(text, use_llm=False)
-            if result["detected"]:
-                print("\n" + display_resources(result) + "\n")
-            else:
-                print("  [safe]")
-    except KeyboardInterrupt:
-        print("\nExiting.")
-
-
-def run_daemon():
-    """Daemon mode — read from a named pipe or stdin, output results."""
-    import select
-    print("Edge Crisis Detector — daemon mode")
-    print("Reading from stdin. Pipe text to detect.\n")
-
-    while True:
-        try:
-            line = sys.stdin.readline()
-            if not line:
-                break
-            text = line.strip()
-            if not text:
-                continue
-
-            result = detect_crisis(text, use_llm=False)
-            if result["detected"]:
-                output = json.dumps({"crisis": True, "result": result, "resources": load_resources()})
-                print(output, flush=True)
-            else:
-                print(json.dumps({"crisis": False}), flush=True)
-        except KeyboardInterrupt:
-            break
-
-
-def main():
-    if "--test" in sys.argv:
-        success = run_tests()
-        sys.exit(0 if success else 1)
-
-    elif "--check" in sys.argv:
-        idx = sys.argv.index("--check")
-        if idx + 1 < len(sys.argv):
-            text = " ".join(sys.argv[idx + 1:])
-            run_check(text)
-        else:
-            print("Usage: crisis_detector.py --check 'text to check'")
-            sys.exit(1)
-
-    elif "--daemon" in sys.argv:
-        run_daemon()
-
-    else:
-        run_interactive()
-
-
-if __name__ == "__main__":
-    main()

tests/conftest.py

@@ -0,0 +1,3 @@
+"""Pytest configuration for turboquant."""
+import sys, os
+sys.path.insert(0, os.path.dirname(os.path.abspath(__file__)))

tests/test_dflash_apple_silicon.py

@@ -0,0 +1,58 @@
+#!/usr/bin/env python3
+"""Tests for Apple Silicon DFlash benchmark planning helpers (issue #152)."""
+
+import os
+import sys
+from unittest.mock import patch
+
+sys.path.insert(0, os.path.dirname(os.path.dirname(__file__)))
+
+from benchmarks.dflash_apple_silicon import (  # noqa: E402
+    build_mlx_benchmark_command,
+    detect_total_memory_gb,
+    render_report_template,
+    select_pair,
+)
+
+
+class TestPairSelection:
+    def test_prefers_qwen35_9b_on_36gb_mac(self):
+        pair = select_pair(total_memory_gb=36)
+        assert pair.slug == "qwen35-9b"
+        assert pair.base_model == "Qwen/Qwen3.5-9B"
+        assert pair.draft_model == "z-lab/Qwen3.5-9B-DFlash"
+
+    def test_falls_back_to_4b_when_memory_is_tight(self):
+        pair = select_pair(total_memory_gb=20)
+        assert pair.slug == "qwen35-4b"
+        assert pair.base_model == "Qwen/Qwen3.5-4B"
+
+
+class TestCommandGeneration:
+    def test_builds_upstream_mlx_benchmark_command(self):
+        pair = select_pair(total_memory_gb=36)
+        command = build_mlx_benchmark_command(pair, dataset="gsm8k", max_samples=64)
+        assert "python -m dflash.benchmark --backend mlx" in command
+        assert "--model Qwen/Qwen3.5-9B" in command
+        assert "--draft-model z-lab/Qwen3.5-9B-DFlash" in command
+        assert "--dataset gsm8k" in command
+        assert "--max-samples 64" in command
+        assert "--draft-sliding-window-size 4096" in command
+
+
+class TestReportTemplate:
+    def test_report_template_mentions_baseline_and_verdict(self):
+        pair = select_pair(total_memory_gb=36)
+        report = render_report_template(machine_label="M3 Max 36GB", pair=pair)
+        assert "DFlash Apple Silicon Benchmark Report" in report
+        assert "M3 Max 36GB" in report
+        assert "Qwen/Qwen3.5-9B" in report
+        assert "plain MLX or llama.cpp speculative decoding" in report
+        assert "Worth operationalizing locally?" in report
+
+
+class TestMemoryDetection:
+    @patch("benchmarks.dflash_apple_silicon.platform.system", return_value="Darwin")
+    @patch("benchmarks.dflash_apple_silicon.subprocess.check_output", return_value=b"38654705664\n")
+    def test_detect_total_memory_gb_on_macos(self, _mock_sysctl, _mock_system):
+        assert detect_total_memory_gb() == 36.0

tests/test_edge_crisis.py

@@ -1,209 +0,0 @@
-#!/usr/bin/env python3
-"""
-test_edge_crisis.py — Offline crisis detection tests
-
-Tests run with zero dependencies and no network access.
-Verifies keyword detection accuracy and resource loading.
-
-Usage:
-    python3 tests/test_edge_crisis.py
-    python3 -m pytest tests/test_edge_crisis.py -v
-"""
-
-import json
-import os
-import sys
-from pathlib import Path
-
-# Add parent to path for imports
-sys.path.insert(0, str(Path(__file__).resolve().parent.parent))
-from scripts.crisis_detector import detect_keywords, detect_crisis, load_resources
-
-# ── Test Data ───────────────────────────────────────────────────────────────
-
-# Texts that SHOULD trigger crisis detection
-CRISIS_TEXTS = [
-    "I want to kill myself",
-    "I'm going to die tonight",
-    "I can't take this anymore, I want to end it",
-    "Nobody would miss me if I were gone",
-    "I've been cutting myself to feel something",
-    "There's no point in living anymore",
-    "I'm suicidal and I need help",
-    "I just want the pain to stop forever",
-    "I've been thinking about hanging myself",
-    "Please help me I'm drowning in this",
-    "I give up. Nothing matters anymore.",
-    "I want to jump off the bridge",
-    "I can't go on like this",
-    "Help me I'm desperate",
-    "I'm done with life",
-]
-
-# Texts that should NOT trigger (common false positives)
-SAFE_TEXTS = [
-    "What a kill in that game!",
-    "I'm dying of laughter",
-    "This movie killed me",
-    "I want to kill it at my presentation",
-    "I'm dead tired after work",
-    "The jump scare got me",
-    "I cut my finger cooking",
-    "This game is killing me (hard level)",
-    "The point of the story is unclear",
-    "I'm done with this homework",
-    "No point in arguing about it",
-    "The bridge looks beautiful at sunset",
-    "I gave up sugar for Lent",
-    "The knife is in the drawer",
-    "She's hanging the picture on the wall",
-]
-
-
-def test_crisis_texts_detected():
-    """All crisis texts must be detected."""
-    failures = []
-    for text in CRISIS_TEXTS:
-        result = detect_keywords(text)
-        if not result["detected"]:
-            failures.append(f"MISSED: '{text}'")
-    assert not failures, "Crisis texts not detected:\n" + "\n".join(failures)
-
-
-def test_safe_texts_not_detected():
-    """Safe texts must not trigger false positives."""
-    failures = []
-    for text in SAFE_TEXTS:
-        result = detect_keywords(text)
-        if result["detected"]:
-            failures.append(f"FALSE POSITIVE: '{text}' -> {result['categories']}")
-    # Allow some false positives — crisis detection should err on side of caution
-    false_positive_rate = len(failures) / len(SAFE_TEXTS)
-    assert false_positive_rate < 0.3, (
-        f"False positive rate too high ({false_positive_rate:.0%}):\n"
-        + "\n".join(failures)
-    )
-
-
-def test_immediate_danger_high_confidence():
-    """Immediate danger keywords must have high confidence."""
-    for text in ["I want to kill myself", "I'm going to die tonight", "I'm suicidal"]:
-        result = detect_keywords(text)
-        assert result["detected"], f"Failed to detect: '{text}'"
-        assert result["confidence"] >= 0.8, (
-            f"Low confidence for immediate danger: {result['confidence']}"
-        )
-
-
-def test_resources_load():
-    """Resources file must exist and be valid."""
-    resources = load_resources()
-    assert "national_resources" in resources
-    assert len(resources["national_resources"]) >= 1
-
-    # 988 lifeline must be present
-    phones = [r.get("phone", "") for r in resources["national_resources"]]
-    assert any("988" in p for p in phones), "988 Lifeline not in resources"
-
-
-def test_resources_have_required_fields():
-    """All national resources must have name and contact method."""
-    resources = load_resources()
-    for r in resources["national_resources"]:
-        assert "name" in r, f"Resource missing name: {r}"
-        has_contact = r.get("phone") or r.get("text") or r.get("url")
-        assert has_contact, f"Resource missing contact: {r['name']}"
-
-
-def test_keyword_categories():
-    """Verify all keyword categories are represented."""
-    for text, expected_cats in [
-        ("I want to kill myself", ["immediate_danger"]),
-        ("I've been cutting myself", ["self_harm"]),
-        ("There's no point in living", ["hopelessness"]),
-    ]:
-        result = detect_keywords(text)
-        assert result["detected"], f"Should detect: '{text}'"
-        for cat in expected_cats:
-            assert cat in result["categories"], (
-                f"Expected category '{cat}' for '{text}', got {result['categories']}"
-            )
-
-
-def test_empty_text_safe():
-    """Empty text must not trigger."""
-    result = detect_keywords("")
-    assert not result["detected"]
-    assert result["confidence"] == 0.0
-
-
-def test_detect_crisis_combined():
-    """Combined detect_crisis function works (keyword-only, no LLM)."""
-    result = detect_crisis("I want to kill myself", use_llm=False)
-    assert result["detected"]
-
-    result2 = detect_crisis("Nice weather today", use_llm=False)
-    assert not result2["detected"]
-
-
-def test_resource_file_exists():
-    """The resources JSON file must exist."""
-    resources_file = Path(__file__).resolve().parent.parent / "data" / "crisis_resources.json"
-    assert resources_file.exists(), f"Missing: {resources_file}"
-
-
-def test_resources_json_valid():
-    """Resources file must be valid JSON with expected structure."""
-    resources_file = Path(__file__).resolve().parent.parent / "data" / "crisis_resources.json"
-    with open(resources_file) as f:
-        data = json.load(f)
-    assert "version" in data
-    assert "national_resources" in data
-    assert "self_help_prompts" in data
-    assert len(data["national_resources"]) >= 3
-
-
-# ── Runner ──────────────────────────────────────────────────────────────────
-
-def run_all_tests():
-    """Run all tests without pytest."""
-    tests = [
-        test_crisis_texts_detected,
-        test_safe_texts_not_detected,
-        test_immediate_danger_high_confidence,
-        test_resources_load,
-        test_resources_have_required_fields,
-        test_keyword_categories,
-        test_empty_text_safe,
-        test_detect_crisis_combined,
-        test_resource_file_exists,
-        test_resources_json_valid,
-    ]
-
-    passed = 0
-    failed = 0
-
-    for test in tests:
-        name = test.__name__
-        try:
-            test()
-            print(f"  PASS: {name}")
-            passed += 1
-        except AssertionError as e:
-            print(f"  FAIL: {name}")
-            print(f"        {e}")
-            failed += 1
-        except Exception as e:
-            print(f"  ERROR: {name}: {e}")
-            failed += 1
-
-    print(f"\n{'='*50}")
-    print(f"Results: {passed} passed, {failed} failed, {passed+failed} total")
-    print(f"{'='*50}")
-
-    return failed == 0
-
-
-if __name__ == "__main__":
-    success = run_all_tests()
-    sys.exit(0 if success else 1)

tests/test_quant_selector.py

@@ -0,0 +1,177 @@
+#!/usr/bin/env python3
+"""Tests for quant_selector.py"""
+
+import sys
+import os
+import pytest
+from unittest.mock import patch, MagicMock
+
+sys.path.insert(0, os.path.dirname(os.path.dirname(__file__)))
+from evolution.quant_selector import (
+    QuantLevel,
+    HardwareInfo,
+    QUANT_LEVELS,
+    detect_hardware,
+    estimate_kv_cache_gb,
+    estimate_model_memory_gb,
+    select_quant_level,
+)
+
+
+class TestQuantLevels:
+    def test_levels_keep_turboquant_quality_order_with_q4_fallback_last(self):
+        """TurboQuant levels should lead, with q4_0 reserved as the non-Turbo fallback."""
+        names = [level.name for level in QUANT_LEVELS]
+        assert names[:3] == ["turbo4", "turbo3", "turbo2"]
+        assert names[-1] == "q4_0"
+
+    def test_all_levels_have_required_fields(self):
+        for level in QUANT_LEVELS:
+            assert level.name
+            assert level.bits_per_channel > 0
+            assert level.compression_ratio > 1
+            assert level.quality_label
+            assert level.layer_adaptive >= 0
+            assert level.kv_type
+
+
+class TestKVEstimate:
+    def test_basic_estimate(self):
+        # 48 layers, 8 heads, 128 dim, 32K context, 3.5 bits
+        kv_gb = estimate_kv_cache_gb(32768, 48, 8, 128, 3.5)
+        assert kv_gb > 0
+        assert kv_gb < 10  # Should be reasonable
+
+    def test_longer_context_larger(self):
+        kv_32k = estimate_kv_cache_gb(32768, 48, 8, 128, 3.5)
+        kv_128k = estimate_kv_cache_gb(131072, 48, 8, 128, 3.5)
+        assert kv_128k > kv_32k
+
+    def test_higher_bits_larger(self):
+        kv_4b = estimate_kv_cache_gb(32768, 48, 8, 128, 4.0)
+        kv_2b = estimate_kv_cache_gb(32768, 48, 8, 128, 2.0)
+        assert kv_4b > kv_2b
+
+
+class TestHardwareDetection:
+    def test_detect_returns_info(self):
+        hw = detect_hardware()
+        assert hw.total_memory_gb > 0
+        assert hw.available_memory_gb > 0
+        assert hw.detection_method
+
+    @patch("evolution.quant_selector.platform.system", return_value="Linux")
+    @patch("builtins.open", create=True)
+    def test_linux_detection(self, mock_open, mock_system):
+        mock_open.return_value.__enter__().read.return_value = (
+            "MemTotal:       32000000 kB\n"
+            "MemAvailable:   24000000 kB\n"
+        )
+        hw = _detect_linux_fallback()
+        assert hw.total_memory_gb > 20
+
+
+def _detect_linux_fallback():
+    """Helper to test Linux detection with mocked /proc/meminfo."""
+    from evolution.quant_selector import _detect_linux
+    return _detect_linux()
+
+
+class TestSelection:
+    def test_selects_turbo4_for_large_memory(self):
+        """With plenty of memory, should pick turbo4 (best quality)."""
+        with patch("evolution.quant_selector.detect_hardware") as mock_hw:
+            mock_hw.return_value = HardwareInfo(
+                total_memory_gb=64,
+                available_memory_gb=48,
+                gpu_memory_gb=64,
+                gpu_name="Test GPU",
+                cpu_cores=16,
+                detection_method="mock",
+            )
+            sel = select_quant_level(model_size_gb=14.0, context_length=32768)
+            assert sel.level.name == "turbo4"
+            assert sel.headroom_gb > 0
+
+    def test_selects_smaller_for_tight_memory(self):
+        """With tight memory, should pick a smaller quant."""
+        with patch("evolution.quant_selector.detect_hardware") as mock_hw:
+            mock_hw.return_value = HardwareInfo(
+                total_memory_gb=16,
+                available_memory_gb=12,
+                gpu_memory_gb=16,
+                gpu_name="Test GPU",
+                cpu_cores=8,
+                detection_method="mock",
+            )
+            sel = select_quant_level(model_size_gb=14.0, context_length=131072)
+            # Should pick a smaller quant for 128K context on 16GB
+            assert sel.level.bits_per_channel <= 4.0
+
+    def test_preferred_level(self):
+        """User can force a specific level."""
+        with patch("evolution.quant_selector.detect_hardware") as mock_hw:
+            mock_hw.return_value = HardwareInfo(
+                total_memory_gb=64,
+                available_memory_gb=48,
+                cpu_cores=16,
+                detection_method="mock",
+            )
+            sel = select_quant_level(
+                model_size_gb=14.0, context_length=32768,
+                preferred_level="turbo2"
+            )
+            assert sel.level.name == "turbo2"
+
+    def test_env_vars_populated(self):
+        with patch("evolution.quant_selector.detect_hardware") as mock_hw:
+            mock_hw.return_value = HardwareInfo(
+                total_memory_gb=64,
+                available_memory_gb=48,
+                cpu_cores=16,
+                detection_method="mock",
+            )
+            sel = select_quant_level(model_size_gb=14.0, context_length=32768)
+            assert "TURBO_LAYER_ADAPTIVE" in sel.env_vars
+            assert "-ctk" in sel.server_flags
+            assert "-ctv" in sel.server_flags
+
+    def test_warnings_on_low_headroom(self):
+        with patch("evolution.quant_selector.detect_hardware") as mock_hw:
+            mock_hw.return_value = HardwareInfo(
+                total_memory_gb=18,
+                available_memory_gb=14,
+                gpu_memory_gb=18,
+                gpu_name="Test GPU",
+                cpu_cores=8,
+                detection_method="mock",
+            )
+            sel = select_quant_level(model_size_gb=16.0, context_length=65536)
+            assert len(sel.warnings) > 0
+
+    def test_falls_back_to_turbo2_when_nothing_fits(self):
+        with patch("evolution.quant_selector.detect_hardware") as mock_hw:
+            mock_hw.return_value = HardwareInfo(
+                total_memory_gb=8,
+                available_memory_gb=6,
+                gpu_memory_gb=8,
+                gpu_name="Tiny GPU",
+                cpu_cores=4,
+                detection_method="mock",
+            )
+            sel = select_quant_level(model_size_gb=16.0, context_length=131072)
+            assert sel.level.name == "turbo2"
+
+    def test_reasoning_contains_key_info(self):
+        with patch("evolution.quant_selector.detect_hardware") as mock_hw:
+            mock_hw.return_value = HardwareInfo(
+                total_memory_gb=32,
+                available_memory_gb=24,
+                is_apple_silicon=True,
+                chip_name="M4 Max",
+                cpu_cores=16,
+                detection_method="mock",
+            )
+            sel = select_quant_level(model_size_gb=14.0, context_length=32768)
+            assert "turbo4" in sel.reasoning
+            assert "M4 Max" in sel.reasoning or "32GB" in sel.reasoning

tests/test_tool_call_integration.py

@@ -0,0 +1,338 @@
+"""
+Integration test: turboquant compressed model passes hermes tool calls (issue #82).
+
+Validates that a TurboQuant-compressed model can:
+1. Parse hermes tool schemas correctly
+2. Format tool calls in OpenAI-compatible format
+3. Pass through the hermes agent conversation loop
+
+Tests are structured as contract tests -- they validate the schema/format
+compatibility without requiring a running model server. The live inference
+test is skipped by default (requires llama-server with TurboQuant model).
+
+Usage:
+    pytest tests/test_tool_call_integration.py -v
+    pytest tests/test_tool_call_integration.py -v -k live  # run live test if server available
+"""
+import json
+import os
+import pathlib
+import re
+import unittest
+
+import pytest
+
+ROOT = pathlib.Path(__file__).resolve().parents[1]
+PROFILE_PATH = ROOT / "profiles" / "hermes-profile-gemma4-turboquant.yaml"
+BENCHMARKS_DIR = ROOT / "benchmarks"
+
+
+class TestHermesProfileSchema(unittest.TestCase):
+    """Validate the hermes profile YAML has required fields for tool calling."""
+
+    @classmethod
+    def setUpClass(cls):
+        import yaml
+        cls.profile = yaml.safe_load(PROFILE_PATH.read_text())
+
+    def test_profile_has_providers(self):
+        assert "providers" in self.profile, "Profile must define providers"
+        assert "primary" in self.profile["providers"], "Must have primary provider"
+
+    def test_primary_provider_has_endpoint(self):
+        primary = self.profile["providers"]["primary"]
+        assert "endpoint" in primary, "Primary provider must have endpoint"
+        assert primary["endpoint"].startswith("http"), "Endpoint must be HTTP(S) URL"
+
+    def test_primary_provider_has_api_path(self):
+        primary = self.profile["providers"]["primary"]
+        assert "api_path" in primary, "Primary provider must have api_path"
+        assert "/chat/completions" in primary["api_path"], (
+            "api_path should be OpenAI-compatible /chat/completions"
+        )
+
+    def test_turboquant_settings_present(self):
+        primary = self.profile["providers"]["primary"]
+        assert "turboquant" in primary, "Must have turboquant config section"
+        tq = primary["turboquant"]
+        assert tq.get("enabled") is True, "TurboQuant must be enabled"
+        assert tq.get("kv_type") in ("turbo2", "turbo3", "turbo4"), (
+            "kv_type must be turbo2, turbo3, or turbo4"
+        )
+
+    def test_context_window_configured(self):
+        primary = self.profile["providers"]["primary"]
+        assert "context" in primary, "Must have context config"
+        ctx = primary["context"]
+        assert ctx.get("max_tokens", 0) >= 8192, (
+            "max_tokens should be >= 8192 for TurboQuant value proposition"
+        )
+
+
+class TestToolSchemaCompatibility(unittest.TestCase):
+    """Verify hermes tool schemas serialize to valid JSON for OpenAI tool_calls."""
+
+    SAMPLE_TOOL_SCHEMAS = [
+        {
+            "type": "function",
+            "function": {
+                "name": "read_file",
+                "description": "Read a text file with line numbers.",
+                "parameters": {
+                    "type": "object",
+                    "properties": {
+                        "path": {"type": "string", "description": "File path"},
+                        "offset": {"type": "integer", "default": 1},
+                        "limit": {"type": "integer", "default": 500},
+                    },
+                    "required": ["path"],
+                },
+            },
+        },
+        {
+            "type": "function",
+            "function": {
+                "name": "execute_code",
+                "description": "Run a Python script.",
+                "parameters": {
+                    "type": "object",
+                    "properties": {
+                        "code": {"type": "string", "description": "Python code"},
+                    },
+                    "required": ["code"],
+                },
+            },
+        },
+        {
+            "type": "function",
+            "function": {
+                "name": "web_search",
+                "description": "Search the web.",
+                "parameters": {
+                    "type": "object",
+                    "properties": {
+                        "query": {"type": "string"},
+                        "max_results": {"type": "integer", "default": 5},
+                    },
+                    "required": ["query"],
+                },
+            },
+        },
+    ]
+
+    def test_tool_schemas_serialize_to_json(self):
+        """Tool schemas must serialize without errors."""
+        serialized = json.dumps(self.SAMPLE_TOOL_SCHEMAS)
+        assert len(serialized) > 0
+        parsed = json.loads(serialized)
+        assert len(parsed) == len(self.SAMPLE_TOOL_SCHEMAS)
+
+    def test_tool_schemas_have_required_openai_fields(self):
+        """Each tool schema must have the fields OpenAI expects."""
+        for tool in self.SAMPLE_TOOL_SCHEMAS:
+            assert tool["type"] == "function", "Tool type must be 'function'"
+            fn = tool["function"]
+            assert "name" in fn, "Function must have name"
+            assert "description" in fn, "Function must have description"
+            assert "parameters" in fn, "Function must have parameters"
+            params = fn["parameters"]
+            assert params["type"] == "object", "Parameters type must be 'object'"
+            assert "properties" in params, "Parameters must have properties"
+
+    def test_tool_call_response_format(self):
+        """Verify tool_call response matches OpenAI format."""
+        tool_call = {
+            "id": "call_abc123",
+            "type": "function",
+            "function": {
+                "name": "read_file",
+                "arguments": json.dumps({"path": "/tmp/test.txt"}),
+            },
+        }
+        args = json.loads(tool_call["function"]["arguments"])
+        assert args["path"] == "/tmp/test.txt"
+        assert tool_call["function"]["name"] in [
+            t["function"]["name"] for t in self.SAMPLE_TOOL_SCHEMAS
+        ]
+
+    def test_tool_names_are_valid_identifiers(self):
+        """Tool names must be valid Python identifiers for hermes dispatch."""
+        for tool in self.SAMPLE_TOOL_SCHEMAS:
+            name = tool["function"]["name"]
+            assert re.match(r"^[a-zA-Z_][a-zA-Z0-9_]*$", name), (
+                f"Tool name \'{name}\' is not a valid identifier"
+            )
+
+
+class TestTurboquantServerConfig(unittest.TestCase):
+    """Validate server startup configuration matches hermes profile."""
+
+    def test_server_command_has_turboquant_flags(self):
+        """The server command in the profile must include -ctk/-ctv flags."""
+        profile_text = PROFILE_PATH.read_text()
+        assert "-ctk" in profile_text, "Profile server command must include -ctk flag"
+        assert "-ctv" in profile_text, "Profile server command must include -ctv flag"
+
+    def test_server_command_has_context_flag(self):
+        """Server command must set context size."""
+        profile_text = PROFILE_PATH.read_text()
+        assert re.search(r"-c\s+\d+", profile_text), (
+            "Server command must include -c <context_size> flag"
+        )
+
+    def test_layer_adaptive_env_var(self):
+        """Profile must set TURBO_LAYER_ADAPTIVE env var."""
+        profile_text = PROFILE_PATH.read_text()
+        assert "TURBO_LAYER_ADAPTIVE" in profile_text, (
+            "Profile must configure TURBO_LAYER_ADAPTIVE"
+        )
+
+
+class TestBenchmarkData(unittest.TestCase):
+    """Validate benchmark test prompts include tool-call test cases."""
+
+    @classmethod
+    def setUpClass(cls):
+        prompts_path = BENCHMARKS_DIR / "test_prompts.json"
+        cls.prompts = json.loads(prompts_path.read_text())
+
+    def test_has_tool_call_test_prompt(self):
+        """Benchmark prompts must include a tool-call format test."""
+        categories = [p.get("category") for p in self.prompts]
+        assert "tool_call_format" in categories, (
+            "Benchmark must include a tool_call_format test case"
+        )
+
+    def test_tool_call_prompt_expects_json(self):
+        """Tool call test prompt must expect JSON in the response."""
+        tool_prompt = next(
+            p for p in self.prompts if p.get("category") == "tool_call_format"
+        )
+        pattern = tool_prompt.get("expected_pattern", "")
+        assert "json" in pattern.lower() or "\\{" in pattern, (
+            "Tool call prompt must expect JSON-formatted response"
+        )
+
+
+@pytest.mark.skipif(
+    not os.environ.get("TURBOQUANT_SERVER_URL"),
+    reason="No TurboQuant server available (set TURBOQUANT_SERVER_URL to run)",
+)
+class TestLiveToolCallIntegration:
+    """Live integration test -- requires running llama-server with TurboQuant."""
+
+    def test_server_health(self):
+        """Server must respond to /v1/models endpoint."""
+        import requests
+        url = os.environ["TURBOQUANT_SERVER_URL"]
+        resp = requests.get(f"{url}/v1/models", timeout=10)
+        assert resp.status_code == 200
+        data = resp.json()
+        assert "data" in data
+        assert len(data["data"]) > 0
+
+    def test_tool_call_completion(self):
+        """Model must return a valid tool_call for a read_file prompt."""
+        import requests
+        url = os.environ["TURBOQUANT_SERVER_URL"]
+        tools = [
+            {
+                "type": "function",
+                "function": {
+                    "name": "read_file",
+                    "description": "Read a file",
+                    "parameters": {
+                        "type": "object",
+                        "properties": {"path": {"type": "string"}},
+                        "required": ["path"],
+                    },
+                },
+            }
+        ]
+        resp = requests.post(
+            f"{url}/v1/chat/completions",
+            json={
+                "model": "gemma-4",
+                "messages": [
+                    {"role": "user", "content": "Read the file at /tmp/test.txt"}
+                ],
+                "tools": tools,
+                "tool_choice": "auto",
+            },
+            timeout=120,
+        )
+        assert resp.status_code == 200
+        data = resp.json()
+        choice = data["choices"][0]
+        msg = choice["message"]
+        if "tool_calls" in msg and msg["tool_calls"]:
+            tc = msg["tool_calls"][0]
+            assert tc["type"] == "function"
+            assert tc["function"]["name"] == "read_file"
+            args = json.loads(tc["function"]["arguments"])
+            assert "path" in args
+        else:
+            assert len(msg.get("content", "")) > 0
+
+    def test_tool_call_with_multiple_tools(self):
+        """Model must handle multiple available tools."""
+        import requests
+        url = os.environ["TURBOQUANT_SERVER_URL"]
+        tools = [
+            {
+                "type": "function",
+                "function": {
+                    "name": "read_file",
+                    "description": "Read a file",
+                    "parameters": {
+                        "type": "object",
+                        "properties": {"path": {"type": "string"}},
+                        "required": ["path"],
+                    },
+                },
+            },
+            {
+                "type": "function",
+                "function": {
+                    "name": "web_search",
+                    "description": "Search the web",
+                    "parameters": {
+                        "type": "object",
+                        "properties": {"query": {"type": "string"}},
+                        "required": ["query"],
+                    },
+                },
+            },
+            {
+                "type": "function",
+                "function": {
+                    "name": "execute_code",
+                    "description": "Run Python code",
+                    "parameters": {
+                        "type": "object",
+                        "properties": {"code": {"type": "string"}},
+                        "required": ["code"],
+                    },
+                },
+            },
+        ]
+        resp = requests.post(
+            f"{url}/v1/chat/completions",
+            json={
+                "model": "gemma-4",
+                "messages": [
+                    {"role": "user", "content": "Search the web for 'bitcoin price'"}
+                ],
+                "tools": tools,
+                "tool_choice": "auto",
+            },
+            timeout=120,
+        )
+        assert resp.status_code == 200
+        data = resp.json()
+        assert "choices" in data
+        assert len(data["choices"]) > 0
+
+
+if __name__ == "__main__":
+    unittest.main()