test: add unit tests for tool calling test runner (#101)

- Update raw-IP forge URL to canonical forge domain in README.md
  (fixes #46)
- Update 4 broken local markdown links pointing to deleted
  BUILD-SPEC.md, PHASE1-REPORT.md, FULL-REPORT.md to
  docs/PROJECT_STATUS.md (fixes #44)

.gitignore

@@ -0,0 +1,3 @@
+build/
+*.pyc
+__pycache__/

CMakeLists.txt

@@ -0,0 +1,36 @@
+cmake_minimum_required(VERSION 3.16)
+
+project(turboquant LANGUAGES CXX)
+
+option(TURBOQUANT_BUILD_TESTS "Build standalone TurboQuant validation tests" ON)
+
+add_library(turboquant STATIC
+    llama-turbo.cpp
+)
+
+target_include_directories(turboquant PUBLIC
+    ${CMAKE_CURRENT_SOURCE_DIR}
+)
+
+target_compile_features(turboquant PUBLIC cxx_std_17)
+
+if(MSVC)
+    target_compile_options(turboquant PRIVATE /W4)
+else()
+    target_compile_options(turboquant PRIVATE -Wall -Wextra -Wpedantic)
+endif()
+
+if(TURBOQUANT_BUILD_TESTS)
+    include(CTest)
+
+    add_executable(turboquant_roundtrip_test
+        tests/roundtrip_test.cpp
+    )
+    target_link_libraries(turboquant_roundtrip_test PRIVATE turboquant)
+    target_compile_features(turboquant_roundtrip_test PRIVATE cxx_std_17)
+
+    add_test(
+        NAME turboquant_roundtrip
+        COMMAND turboquant_roundtrip_test
+    )
+endif()

GENOME.md

@@ -1,323 +0,0 @@
-# GENOME.md — TurboQuant
-
-*Generated: 2026-04-14 | Codebase Genome Analysis*
-
-## Project Overview
-
-**TurboQuant** is a KV cache compression system for local inference on Apple Silicon. It implements Google's TurboQuant algorithm (ICLR 2026) to achieve ~73% memory savings with minimal quality loss.
-
-### Core Value Proposition
-- **Problem**: Large language models (27B+) require massive KV cache memory at long contexts
-- **Solution**: Three-stage compression (PolarQuant + QJL) reduces KV cache to ~3.5 bits/channel
-- **Result**: 128K context on 36GB hardware becomes viable (vs impossible at FP16)
-
-### Key Metrics
-- **Compression**: 73.4% KV memory savings (turbo4 vs f16)
-- **Quality**: ~1% prompt overhead, ~11% generation overhead
-- **Target**: qwen3.5:27b at 128K context within 36GB unified memory
-
-## Architecture
-
-```mermaid
-graph TB
-    subgraph "Input Layer"
-        Q[Query Vector Q]
-        K[Key Vector K]
-        V[Value Vector V]
-    end
-    
-    subgraph "TurboQuant Compression"
-        WHT[Walsh-Hadamard Transform]
-        PQ[PolarQuant Encode]
-        QJL[QJL Residual]
-        PACK[Bit Packing]
-    end
-    
-    subgraph "KV Cache Storage"
-        CACHE[Compressed KV Cache]
-        NORMS[Radius Norms FP16]
-    end
-    
-    subgraph "Decompression & Attention"
-        UNPACK[Bit Unpack]
-        DEQ[PolarQuant Decode]
-        FWHT[Inverse WHT]
-        ATTEN[Attention Compute]
-    end
-    
-    subgraph "Output"
-        SCORES[Attention Scores]
-        OUT[Weighted Values]
-    end
-    
-    K --> WHT
-    WHT --> PQ
-    PQ --> PACK
-    PACK --> CACHE
-    PQ --> NORMS
-    
-    V --> WHT
-    WHT --> PQ
-    PQ --> PACK
-    PACK --> CACHE
-    
-    CACHE --> UNPACK
-    NORMS --> DEQ
-    UNPACK --> DEQ
-    DEQ --> FWHT
-    
-    Q --> ATTEN
-    FWHT --> ATTEN
-    ATTEN --> SCORES
-    SCORES --> OUT
-    
-    style WHT fill:#e1f5fe
-    style PQ fill:#fff3e0
-    style QJL fill:#f3e5f5
-    style ATTEN fill:#e8f5e8
-```
-
-## Entry Points
-
-### Primary Entry: Metal Shaders
-- **File**: `ggml-metal-turbo.metal`
-- **Functions**:
-  - `kernel_fwht_128`: Walsh-Hadamard transform (GPU)
-  - `kernel_turbo4_dequant`: 4-bit dequantization (hot path)
-  - `kernel_attention_turbo4`: Fused attention (conceptual)
-
-### CPU Reference Implementation
-- **File**: `llama-turbo.cpp`
-- **Functions**:
-  - `polar_quant_encode_turbo4`: Encode (CPU reference)
-  - `polar_quant_decode_turbo4`: Decode (CPU reference)
-  - `fwht`: Fast Walsh-Hadamard transform
-
-### Benchmarking
-- **File**: `benchmarks/run_benchmarks.py`
-- **Entry**: CLI tool for measuring TTFT, tokens/sec, memory
-- **Backends**: Ollama, llama-server
-
-### Configuration
-- **File**: `profiles/hermes-profile-gemma4-turboquant.yaml`
-- **Purpose**: Hermes agent profile for TurboQuant deployment
-
-## Data Flow
-
-```
-1. Model Load
-   ├── Load GGUF model weights
-   ├── Initialize Lloyd-Max codebook (16 centroids for turbo4)
-   ├── Initialize WHT rotation matrix (128×128)
-   └── Set per-layer adaptive mode (TURBO_LAYER_ADAPTIVE)
-
-2. Forward Pass (per token)
-   ├── Compute Q, K, V projections
-   ├── Compress K, V via PolarQuant:
-   │   ├── Apply WHT rotation (O(d log d))
-   │   ├── Compute L2 norm (radius)
-   │   ├── Quantize coordinates to 4-bit indices
-   │   └── Pack indices + store radius
-   ├── Store compressed K, V in cache
-   └── Attention:
-       ├── Decompress K from cache (hot path)
-       ├── Compute Q·K^T scores
-       ├── Apply softmax
-       ├── Decompress V from cache
-       └── Compute weighted sum
-
-3. Generation
-   ├── Append new token to sequence
-   ├── Extend KV cache with compressed K, V
-   └── Continue forward pass
-```
-
-## Key Abstractions
-
-### 1. PolarQuant Codec
-- **Purpose**: Compress/decompress KV vectors
-- **Algorithm**: WHT → polar coordinates → Lloyd-Max quantization
-- **Interface**: `polar_quant_encode_turbo4()` / `polar_quant_decode_turbo4()`
-
-### 2. Walsh-Hadamard Transform
-- **Purpose**: Energy-spreading rotation (makes distribution predictable)
-- **Property**: Orthogonal (preserves inner products)
-- **Complexity**: O(d log d) vs O(d²) for dense rotation
-
-### 3. Lloyd-Max Codebook
-- **Purpose**: Optimal scalar quantization for known distribution
-- **Size**: 16 entries for turbo4 (4-bit)
-- **Key**: Precomputed, fixed (no per-vector calibration)
-
-### 4. Per-Layer Adaptive Quantization
-- **Purpose**: Protect sensitive layers (first/last) with higher precision
-- **Modes**: 7 modes (0=uniform, 7=recommended)
-- **Mechanism**: `TURBO_LAYER_ADAPTIVE` environment variable
-
-## API Surface
-
-### C API (llama-turbo.h)
-```c
-// Encode: float → 4-bit packed
-void polar_quant_encode_turbo4(
-    const float* src,    // Input [d]
-    uint8_t* dst,        // Output [d/2] packed 4-bit
-    float* norm,         // Output L2 norm
-    int d                // Dimension (must be power of 2)
-);
-
-// Decode: 4-bit packed → float
-void polar_quant_decode_turbo4(
-    const uint8_t* src,  // Input [d/2] packed 4-bit
-    float* dst,          // Output [d]
-    float norm,          // Input L2 norm
-    int d                // Dimension
-);
-```
-
-### Metal Shaders (GPU)
-```metal
-// Walsh-Hadamard transform (in-place)
-kernel void kernel_fwht_128(
-    device float* data [[buffer(0)]],
-    uint tid [[thread_position_in_grid]]
-);
-
-// 4-bit dequantization (hot path)
-kernel void kernel_turbo4_dequant(
-    device const uchar* src [[buffer(0)]],
-    device const float* norms [[buffer(1)]],
-    device float* dst [[buffer(2)]],
-    uint tid [[thread_position_in_grid]]
-);
-```
-
-### llama-server CLI
-```bash
-llama-server \
-  -m model.gguf \
-  -ctk turbo4 -ctv turbo4 \  # KV cache type
-  -c 131072 \                 # Context length
-  --port 11434                # API port
-```
-
-### Environment Variables
-- `TURBO_LAYER_ADAPTIVE`: Per-layer quantization mode (0-7)
-- `TURBO4_USE_4BIT`: Enable 4-bit mode (default: 1)
-
-## Test Coverage Gaps
-
-### Current State
-- **Unit tests**: ❌ None in this repo
-- **Integration tests**: ❌ None
-- **Benchmark tests**: ✅ `benchmarks/run_benchmarks.py`
-- **Perplexity tests**: ⚠️ Corpus exists (`corpora/wiki.test.raw`) but no runner
-
-### Critical Missing Tests
-1. **Encode/Decode Roundtrip**: Verify `decode(encode(x)) ≈ x`
-2. **Inner Product Preservation**: Verify `Q·K ≈ Q·dequant(quant(K))`
-3. **WHT Orthogonality**: Verify `WHT^T · WHT = I`
-4. **Codebook Correctness**: Verify centroids match Lloyd-Max for N(0, 1/128)
-5. **Metal vs CPU Parity**: Verify GPU and CPU produce identical results
-6. **Per-Layer Adaptive**: Verify sensitive layers use higher precision
-7. **Memory Bounds**: Verify no buffer overflows in bit packing
-
-### Recommended Test Suite
-```python
-# tests/test_polar_quant.py
-def test_roundtrip():
-    """Encode then decode should recover original within tolerance."""
-    
-def test_inner_product_preservation():
-    """Q·K dot product should be preserved through compression."""
-    
-def test_wht_orthogonality():
-    """WHT matrix should be orthogonal."""
-    
-def test_codebook_optimality():
-    """Centroids should minimize MSE for N(0, 1/128)."""
-```
-
-## Security Considerations
-
-### 1. Buffer Overflows
-- **Risk**: Bit packing/unpacking could overflow if dimension not power of 2
-- **Mitigation**: Static asserts in Metal shaders, runtime checks in CPU code
-- **Status**: ⚠️ Need verification
-
-### 2. Numerical Stability
-- **Risk**: Division by zero in `1.0 / (norm + 1e-9)`
-- **Mitigation**: Epsilon guard present
-- **Status**: ✅ Handled
-
-### 3. Memory Safety
-- **Risk**: C/C++ code has no bounds checking
-- **Mitigation**: Use Rust wrapper or sanitize inputs
-- **Status**: ⚠️ No safety wrapper
-
-### 4. Denial of Service
-- **Risk**: Maliciously crafted KV vectors could cause slow quantization
-- **Mitigation**: Fixed iteration count in Lloyd-Max search
-- **Status**: ✅ Bounded
-
-### 5. Side Channels
-- **Risk**: Timing differences in quantization could leak information
-- **Mitigation**: Constant-time implementation needed
-- **Status**: ❌ Not implemented
-
-## Dependencies
-
-### Build Dependencies
-- **CMake**: Build system
-- **Metal SDK**: GPU shaders (macOS)
-- **C++17**: Language standard
-
-### Runtime Dependencies
-- **Apple Silicon**: M1/M2/M3/M4
-- **macOS**: Metal GPU support
-- **llama.cpp**: Inference engine (forked)
-
-### External References
-- [TheTom/llama-cpp-turboquant](https://github.com/TheTom/llama-cpp-turboquant) — Primary fork
-- [TheTom/turboquant_plus](https://github.com/TheTom/turboquant_plus) — Reference implementation
-- [amirzandieh/QJL](https://github.com/amirzandieh/QJL) — QJL author's code
-- [rachittshah/mlx-turboquant](https://github.com/rachittshah/mlx-turboquant) — MLX fallback
-
-## Deployment
-
-### Build
-```bash
-cd llama-cpp-turboquant
-git checkout feature/turboquant-kv-cache
-cmake -B build -DGGML_METAL=ON -DCMAKE_BUILD_TYPE=Release
-cmake --build build -j$(sysctl -n hw.ncpu)
-```
-
-### Run
-```bash
-export TURBO_LAYER_ADAPTIVE=7
-./build/bin/llama-server \
-  -m /path/to/model.gguf \
-  --port 11434 \
-  -ctk turbo4 -ctv turbo4 \
-  -c 131072
-```
-
-### Validate
-```bash
-curl http://localhost:11434/v1/chat/completions \
-  -H "Content-Type: application/json" \
-  -d '{"model":"qwen3.5","messages":[{"role":"user","content":"hello"}]}'
-```
-
-## Open Questions
-
-1. **QJL Status**: Infrastructure exists but is disabled. When will it be needed?
-2. **Upstream Landing**: When will TurboQuant be merged into llama.cpp mainline?
-3. **Quality Threshold**: What PPL delta is acceptable for production use?
-4. **Multi-GPU**: Does TurboQuant work with tensor parallelism?
-
-## Changelog
-
-- **2026-03-30**: Phase 1 complete. PolarQuant MVP verified. 73% KV savings confirmed.
-- **2026-04-14**: GENOME.md generated. Test gaps identified. Security considerations documented.

README.md

@@ -13,7 +13,7 @@ Unlock 64K-128K context on qwen3.5:27b within 32GB unified memory.
 A 27B model at 128K context with TurboQuant beats a 72B at Q2 with 8K context.
 
 ## Status
-See [issues](http://143.198.27.163:3000/Timmy_Foundation/turboquant/issues) for current progress.
+See [issues](https://forge.alexanderwhitestone.com/Timmy_Foundation/turboquant/issues) for current progress.
 
 ## Roles
 - **Strago:** Build spec author
@@ -29,4 +29,4 @@ See [issues](http://143.198.27.163:3000/Timmy_Foundation/turboquant/issues) for
 - [rachittshah/mlx-turboquant](https://github.com/rachittshah/mlx-turboquant) — MLX fallback
 
 ## Docs
-- [BUILD-SPEC.md](BUILD-SPEC.md) — Full build specification (Strago, v2.2)
+- [Project Status](docs/PROJECT_STATUS.md) — Full project status and build specification

benchmarks/bonsai-tool-calling.md

@@ -0,0 +1,50 @@
+# Tool Calling Viability: Bonsai 1-Bit Models
+
+**Epic**: #99 (1-Bit Models + Edge)
+**Date**: TBD (run benchmarks/test_tool_calling.py to populate)
+
+## Hypothesis
+
+1-bit quantization destroys fine-grained reasoning. Tool calling (precise JSON output) may be impossible at Q1_0. But worth testing — the field is moving fast.
+
+## Models to Test
+
+| Model | Size | Quant | Source |
+|-------|------|-------|--------|
+| Bonsai-1.7B | 1.7B | Q1_0 | prism-ml/Bonsai-1.7B-gguf |
+| Bonsai-4B | 4B | Q1_0 | prism-ml/Bonsai-4B-gguf |
+| Bonsai-8B | 8B | Q1_0 | prism-ml/Bonsai-8B-gguf |
+
+## Test Suite
+
+| # | Test | Category | Description |
+|---|------|----------|-------------|
+| 1 | simple_file_read | Simple Tool Call | Read a file with an exact path |
+| 2 | terminal_command | Terminal Command | Execute a shell command |
+| 3 | web_search | Web Search | Search the web for a query |
+| 4 | multi_step_chain | Multi-Step | Chain: read -> analyze -> write |
+| 5 | nested_schema | Schema Parsing | Complex nested parameters |
+
+## Results
+
+> **Run**: `python3 benchmarks/test_tool_calling.py --model bonsai-1.7b --output benchmarks/bonsai-tool-calling.md`
+
+| Test | Bonsai-1.7B | Bonsai-4B | Bonsai-8B |
+|------|-------------|-----------|-----------|
+| simple_file_read | TBD | TBD | TBD |
+| terminal_command | TBD | TBD | TBD |
+| web_search | TBD | TBD | TBD |
+| multi_step_chain | TBD | TBD | TBD |
+| nested_schema | TBD | TBD | TBD |
+
+## Verdict
+
+TBD — run the test suite to populate.
+
+## Failure Modes (if any)
+
+TBD — document specific failure patterns observed.
+
+## Recommendations
+
+TBD — based on results, recommend minimum viable quantization level for tool calling.

benchmarks/test_tool_calling.py

@@ -0,0 +1,435 @@
+#!/usr/bin/env python3
+"""
+Tool Calling Viability Test for 1-Bit / Edge Models (Issue #101)
+
+Tests whether Bonsai 1-bit models (or any small model) can produce
+valid tool calls via Ollama or llama-server API.
+
+Test suite (5 categories):
+  1. Simple tool call: file read with exact path
+  2. Terminal command execution
+  3. Web search
+  4. Multi-step: read file -> analyze -> write result
+  5. Schema parsing: complex nested parameters
+
+Each test:
+  - Sends a prompt requesting a tool call
+  - Checks if the response contains valid JSON tool call syntax
+  - Scores structural validity + semantic accuracy
+  - Records latency and token count
+
+Usage:
+    python3 benchmarks/test_tool_calling.py --model bonsai-1.7b
+    python3 benchmarks/test_tool_calling.py --model qwen3.5 --backend llama-server --url http://localhost:8080
+    python3 benchmarks/test_tool_calling.py --model bonsai-1.7b --output benchmarks/bonsai-tool-calling.md
+"""
+
+import argparse
+import json
+import os
+import re
+import sys
+import time
+from datetime import datetime, timezone
+from typing import Optional
+
+try:
+    import requests
+except ImportError:
+    print("Error: pip install requests", file=sys.stderr)
+    sys.exit(1)
+
+
+# ── Tool call test definitions ────────────────────────────────────────
+
+TESTS = [
+    {
+        "id": "simple_file_read",
+        "category": "Simple Tool Call",
+        "description": "Read a file with an exact path",
+        "prompt": (
+            "You have access to a tool called read_file. "
+            "Call it to read /etc/hostname. "
+            "Respond ONLY with a JSON tool call in this exact format:\n"
+            '{"name": "read_file", "arguments": {"path": "/etc/hostname"}}'
+        ),
+        "validate": lambda resp: _has_json_tool_call(resp, "read_file", {"path": "/etc/hostname"}),
+        "scoring": "structural",  # Can we parse the JSON at all?
+    },
+    {
+        "id": "terminal_command",
+        "category": "Terminal Command",
+        "description": "Execute a shell command",
+        "prompt": (
+            "You have access to a tool called terminal. "
+            "Call it to run the command: echo hello world. "
+            "Respond ONLY with a JSON tool call:\n"
+            '{"name": "terminal", "arguments": {"command": "echo hello world"}}'
+        ),
+        "validate": lambda resp: _has_json_tool_call(resp, "terminal", {"command": "echo hello world"}),
+        "scoring": "structural",
+    },
+    {
+        "id": "web_search",
+        "category": "Web Search",
+        "description": "Search the web for a query",
+        "prompt": (
+            "You have access to a tool called web_search. "
+            "Search for: what is quantization in machine learning. "
+            "Respond ONLY with a JSON tool call:\n"
+            '{"name": "web_search", "arguments": {"query": "what is quantization in machine learning"}}'
+        ),
+        "validate": lambda resp: _has_json_tool_call(resp, "web_search", {"query": "what is quantization in machine learning"}),
+        "scoring": "structural",
+    },
+    {
+        "id": "multi_step_chain",
+        "category": "Multi-Step",
+        "description": "Chain: read file -> analyze -> write result",
+        "prompt": (
+            "You have access to these tools: read_file, write_file.\n"
+            "Task: Read /tmp/input.txt, count the words, then write the count to /tmp/count.txt.\n"
+            "First, call read_file on /tmp/input.txt. "
+            "Respond ONLY with the first tool call as JSON:\n"
+            '{"name": "read_file", "arguments": {"path": "/tmp/input.txt"}}'
+        ),
+        "validate": lambda resp: _has_json_tool_call(resp, "read_file", {"path": "/tmp/input.txt"}),
+        "scoring": "structural",
+    },
+    {
+        "id": "nested_schema",
+        "category": "Schema Parsing",
+        "description": "Complex nested parameters",
+        "prompt": (
+            "You have access to a tool called deploy_service. "
+            "Deploy a service with:\n"
+            '- name: "api-gateway"\n'
+            '- replicas: 3\n'
+            '- env: {"PORT": 8080, "NODE_ENV": "production"}\n'
+            '- resources: {"cpu": "500m", "memory": "256Mi"}\n\n'
+            "Respond ONLY with a JSON tool call:\n"
+            '{"name": "deploy_service", "arguments": {"name": "api-gateway", "replicas": 3, '
+            '"env": {"PORT": 8080, "NODE_ENV": "production"}, '
+            '"resources": {"cpu": "500m", "memory": "256Mi"}}}'
+        ),
+        "validate": lambda resp: _has_nested_tool_call(resp),
+        "scoring": "semantic",  # Needs correct nested structure
+    },
+]
+
+
+# ── Validation helpers ────────────────────────────────────────────────
+
+def _extract_json(text: str) -> Optional[dict]:
+    """Try to extract a JSON object from text."""
+    # Try direct parse
+    text = text.strip()
+    try:
+        obj = json.loads(text)
+        if isinstance(obj, dict):
+            return obj
+    except json.JSONDecodeError:
+        pass
+
+    # Try finding JSON in code blocks
+    code_block = re.search(r"```(?:json)?\s*({.*?})\s*```", text, re.DOTALL)
+    if code_block:
+        try:
+            return json.loads(code_block.group(1))
+        except json.JSONDecodeError:
+            pass
+
+    # Try finding any JSON object
+    json_match = re.search(r"({[^{}]*(?:{[^{}]*}[^{}]*)*})", text)
+    if json_match:
+        try:
+            return json.loads(json_match.group(1))
+        except json.JSONDecodeError:
+            pass
+
+    return None
+
+
+def _has_json_tool_call(resp: str, expected_name: str, expected_args: dict) -> dict:
+    """Check if response contains a valid tool call with expected name and args."""
+    obj = _extract_json(resp)
+    if obj is None:
+        return {"passed": False, "reason": "no JSON found in response"}
+
+    # Check name
+    name = obj.get("name", obj.get("function", {}).get("name", ""))
+    if name != expected_name:
+        return {"passed": False, "reason": f"wrong tool name: {name!r}, expected {expected_name!r}"}
+
+    # Check arguments exist
+    args = obj.get("arguments", obj.get("function", {}).get("arguments", obj.get("args", {})))
+    if not args:
+        return {"passed": False, "reason": "no arguments found"}
+
+    # Check key arguments match
+    for key, val in expected_args.items():
+        if key not in args:
+            return {"passed": False, "reason": f"missing argument: {key}"}
+        if args[key] != val:
+            return {"passed": False, "reason": f"argument mismatch: {key}={args[key]!r}, expected {val!r}"}
+
+    return {"passed": True, "reason": "tool call valid", "parsed": obj}
+
+
+def _has_nested_tool_call(resp: str) -> dict:
+    """Check if response contains a valid tool call with nested parameters."""
+    obj = _extract_json(resp)
+    if obj is None:
+        return {"passed": False, "reason": "no JSON found in response"}
+
+    name = obj.get("name", obj.get("function", {}).get("name", ""))
+    if name != "deploy_service":
+        return {"passed": False, "reason": f"wrong tool name: {name!r}"}
+
+    args = obj.get("arguments", obj.get("function", {}).get("arguments", obj.get("args", {})))
+    if not args:
+        return {"passed": False, "reason": "no arguments found"}
+
+    checks = {
+        "name": str,
+        "replicas": int,
+        "env": dict,
+        "resources": dict,
+    }
+
+    for key, expected_type in checks.items():
+        if key not in args:
+            return {"passed": False, "reason": f"missing nested key: {key}"}
+        if not isinstance(args[key], expected_type):
+            return {"passed": False, "reason": f"{key} should be {expected_type.__name__}, got {type(args[key]).__name__}"}
+
+    # Check env has PORT
+    env = args.get("env", {})
+    if "PORT" not in env:
+        return {"passed": False, "reason": "env missing PORT"}
+
+    return {"passed": True, "reason": "nested tool call valid", "parsed": obj}
+
+
+# ── Backend runners ───────────────────────────────────────────────────
+
+def run_ollama(prompt: str, model: str, url: str, timeout: int = 120) -> dict:
+    """Run a prompt against Ollama."""
+    api_url = f"{url.rstrip('/')}/api/generate"
+    start = time.time()
+    try:
+        resp = requests.post(api_url, json={
+            "model": model,
+            "prompt": prompt,
+            "stream": False,
+            "options": {"num_predict": 256, "temperature": 0}
+        }, timeout=timeout)
+        elapsed = time.time() - start
+        resp.raise_for_status()
+        data = resp.json()
+        return {
+            "response": data.get("response", ""),
+            "latency_s": round(elapsed, 3),
+            "tokens": data.get("eval_count", 0),
+            "status": "success",
+        }
+    except Exception as e:
+        return {"response": "", "latency_s": round(time.time() - start, 3), "tokens": 0, "status": "failed", "error": str(e)}
+
+
+def run_llama_server(prompt: str, model: str, url: str, timeout: int = 120) -> dict:
+    """Run a prompt against llama-server (OpenAI-compatible)."""
+    api_url = f"{url.rstrip('/')}/v1/chat/completions"
+    start = time.time()
+    try:
+        resp = requests.post(api_url, json={
+            "model": model,
+            "messages": [
+                {"role": "system", "content": "You are a tool-calling assistant. Respond ONLY with JSON tool calls."},
+                {"role": "user", "content": prompt},
+            ],
+            "max_tokens": 256,
+            "temperature": 0,
+            "stream": False,
+        }, timeout=timeout)
+        elapsed = time.time() - start
+        resp.raise_for_status()
+        data = resp.json()
+        content = data.get("choices", [{}])[0].get("message", {}).get("content", "")
+        usage = data.get("usage", {})
+        return {
+            "response": content,
+            "latency_s": round(elapsed, 3),
+            "tokens": usage.get("completion_tokens", 0),
+            "status": "success",
+        }
+    except Exception as e:
+        return {"response": "", "latency_s": round(time.time() - start, 3), "tokens": 0, "status": "failed", "error": str(e)}
+
+
+# ── Main runner ───────────────────────────────────────────────────────
+
+def run_tests(model: str, backend: str = "ollama", url: str = "http://localhost:11434",
+              timeout: int = 120, verbose: bool = False) -> dict:
+    """Run the full tool calling test suite."""
+    runner_fn = run_ollama if backend == "ollama" else run_llama_server
+
+    results = {
+        "model": model,
+        "backend": backend,
+        "url": url,
+        "timestamp": datetime.now(timezone.utc).isoformat(),
+        "tests": [],
+        "summary": {"total": 0, "passed": 0, "failed": 0, "errors": 0},
+    }
+
+    print(f"Testing tool calling on: {model} ({backend})\n")
+
+    for test in TESTS:
+        print(f"  [{test['id']}] {test['description']}...", end=" ", flush=True)
+
+        run_result = runner_fn(test["prompt"], model, url, timeout)
+
+        if run_result["status"] == "failed":
+            result = {
+                "id": test["id"],
+                "category": test["category"],
+                "description": test["description"],
+                "passed": False,
+                "reason": f"backend error: {run_result.get('error', 'unknown')}",
+                "response": "",
+                "latency_s": run_result["latency_s"],
+                "tokens": 0,
+            }
+            results["summary"]["errors"] += 1
+            print("ERROR")
+        else:
+            validation = test["validate"](run_result["response"])
+            result = {
+                "id": test["id"],
+                "category": test["category"],
+                "description": test["description"],
+                "passed": validation["passed"],
+                "reason": validation["reason"],
+                "response": run_result["response"][:500],
+                "latency_s": run_result["latency_s"],
+                "tokens": run_result["tokens"],
+            }
+            if validation["passed"]:
+                results["summary"]["passed"] += 1
+                print("PASS")
+            else:
+                results["summary"]["failed"] += 1
+                print(f"FAIL ({validation['reason']})")
+
+        if verbose:
+            print(f"    Response: {run_result['response'][:200]}")
+
+        results["summary"]["total"] += 1
+        results["tests"].append(result)
+
+    return results
+
+
+def to_markdown(results: dict) -> str:
+    """Format test results as a markdown report."""
+    lines = []
+    lines.append(f"# Tool Calling Viability: {results['model']}")
+    lines.append("")
+    lines.append(f"**Date**: {results['timestamp']}")
+    lines.append(f"**Backend**: {results['backend']} ({results['url']})")
+    lines.append(f"**Model**: {results['model']}")
+    lines.append("")
+
+    s = results["summary"]
+    pass_rate = s["passed"] / s["total"] * 100 if s["total"] > 0 else 0
+    lines.append(f"## Summary: {s['passed']}/{s['total']} passed ({pass_rate:.0f}%)")
+    lines.append("")
+    lines.append(f"| Metric | Value |")
+    lines.append(f"|--------|-------|")
+    lines.append(f"| Total tests | {s['total']} |")
+    lines.append(f"| Passed | {s['passed']} |")
+    lines.append(f"| Failed | {s['failed']} |")
+    lines.append(f"| Errors | {s['errors']} |")
+    lines.append("")
+
+    lines.append("## Results by Category")
+    lines.append("")
+    lines.append("| Test | Category | Result | Reason | Latency | Tokens |")
+    lines.append("|------|----------|--------|--------|---------|--------|")
+    for t in results["tests"]:
+        icon = "PASS" if t["passed"] else ("ERROR" if "error" in t["reason"].lower() else "FAIL")
+        lines.append(f"| {t['id']} | {t['category']} | {icon} | {t['reason']} | {t['latency_s']}s | {t['tokens']} |")
+    lines.append("")
+
+    lines.append("## Verdict")
+    lines.append("")
+    if pass_rate == 100:
+        lines.append("**FULLY VIABLE** — All tool calling patterns work. Ready for production edge deployment.")
+    elif pass_rate >= 60:
+        lines.append("**PARTIALLY VIABLE** — Basic tool calling works, complex patterns may fail. Consider for simple agents.")
+    elif pass_rate >= 20:
+        lines.append("**MARGINAL** — Only simplest tool calls work. Not recommended for production.")
+    else:
+        lines.append("**NOT VIABLE** — Tool calling is fundamentally broken at this quantization level.")
+    lines.append("")
+
+    lines.append("## Failure Analysis")
+    lines.append("")
+    failed = [t for t in results["tests"] if not t["passed"]]
+    if not failed:
+        lines.append("No failures.")
+    else:
+        for t in failed:
+            lines.append(f"### {t['id']}")
+            lines.append(f"- **Category**: {t['category']}")
+            lines.append(f"- **Failure**: {t['reason']}")
+            lines.append(f"- **Response** (first 300 chars): `{t['response'][:300]}`")
+            lines.append("")
+    lines.append("")
+
+    lines.append("## Recommendations")
+    lines.append("")
+    if pass_rate >= 80:
+        lines.append("- Deploy for simple single-tool-call workflows")
+        lines.append("- Add retry logic for multi-step chains")
+        lines.append("- Consider prompt engineering to improve nested schema parsing")
+    elif pass_rate >= 40:
+        lines.append("- Use for keyword/rule-based tool routing only")
+        lines.append("- Do NOT use for complex multi-step workflows")
+        lines.append("- Consider a larger model (Q4 quantized) as fallback")
+    else:
+        lines.append("- 1-bit quantization is too lossy for tool calling")
+        lines.append("- Use Q4_0 as minimum viable quantization for tool use")
+        lines.append("- Reserve 1-bit models for text generation only")
+
+    return "\n".join(lines)
+
+
+def main():
+    parser = argparse.ArgumentParser(description="Tool Calling Viability Test for Edge Models")
+    parser.add_argument("--model", "-m", required=True, help="Model name")
+    parser.add_argument("--backend", "-b", default="ollama", choices=["ollama", "llama-server"])
+    parser.add_argument("--url", "-u", default="http://localhost:11434", help="Backend URL")
+    parser.add_argument("--timeout", "-t", type=int, default=120, help="Timeout per test (seconds)")
+    parser.add_argument("--output", "-o", help="Output markdown file path")
+    parser.add_argument("--json", action="store_true", help="JSON output")
+    parser.add_argument("--verbose", "-v", action="store_true", help="Show full responses")
+    args = parser.parse_args()
+
+    results = run_tests(args.model, args.backend, args.url, args.timeout, args.verbose)
+
+    if args.json:
+        print(json.dumps(results, indent=2))
+    else:
+        md = to_markdown(results)
+        if args.output:
+            with open(args.output, "w") as f:
+                f.write(md)
+            print(f"\nReport written to: {args.output}")
+        else:
+            print("\n" + md)
+
+
+if __name__ == "__main__":
+    main()

profiles/README.md

@@ -135,7 +135,5 @@ llama-server -m model.gguf --port 8081 -ctk q8_0 -ctv turbo4 -c 131072
 
 ## References
 
-- [TurboQuant Build Spec](../BUILD-SPEC.md)
-- [Phase 1 Report](../PHASE1-REPORT.md)
-- [Full Knowledge Transfer](../FULL-REPORT.md)
+- [Project Status](../docs/PROJECT_STATUS.md)
 - [llama.cpp TurboQuant Fork](https://github.com/TheTom/llama-cpp-turboquant)

tests/roundtrip_test.cpp

@@ -0,0 +1,104 @@
+#include "llama-turbo.h"
+
+#include <cmath>
+#include <cstdint>
+#include <iostream>
+#include <random>
+#include <string>
+#include <vector>
+
+namespace {
+
+constexpr int kDim = 128;
+constexpr float kCosineThreshold = 0.99f;
+constexpr float kZeroTolerance = 1.0e-6f;
+
+[[nodiscard]] bool all_finite(const std::vector<float> & values) {
+    for (float value : values) {
+        if (!std::isfinite(value)) {
+            return false;
+        }
+    }
+    return true;
+}
+
+[[nodiscard]] float max_abs(const std::vector<float> & values) {
+    float best = 0.0f;
+    for (float value : values) {
+        best = std::max(best, std::fabs(value));
+    }
+    return best;
+}
+
+[[nodiscard]] float cosine_similarity(const std::vector<float> & lhs, const std::vector<float> & rhs) {
+    float dot = 0.0f;
+    float lhs_norm = 0.0f;
+    float rhs_norm = 0.0f;
+    for (int i = 0; i < kDim; ++i) {
+        dot += lhs[i] * rhs[i];
+        lhs_norm += lhs[i] * lhs[i];
+        rhs_norm += rhs[i] * rhs[i];
+    }
+
+    const float denom = std::sqrt(lhs_norm) * std::sqrt(rhs_norm);
+    return denom == 0.0f ? 1.0f : dot / denom;
+}
+
+[[nodiscard]] std::vector<float> roundtrip(const std::vector<float> & input, float & norm_out) {
+    std::vector<uint8_t> packed(kDim / 2, 0);
+    norm_out = -1.0f;
+    polar_quant_encode_turbo4(input.data(), packed.data(), &norm_out, kDim);
+
+    std::vector<float> decoded(kDim, 0.0f);
+    polar_quant_decode_turbo4(packed.data(), decoded.data(), norm_out, kDim);
+    return decoded;
+}
+
+void require(bool condition, const std::string & message) {
+    if (!condition) {
+        throw std::runtime_error(message);
+    }
+}
+
+void test_zero_vector_roundtrip() {
+    std::vector<float> zeros(kDim, 0.0f);
+    float norm = -1.0f;
+    const auto decoded = roundtrip(zeros, norm);
+
+    require(norm == 0.0f, "zero vector should encode with zero norm");
+    require(all_finite(decoded), "zero vector decode produced non-finite values");
+    require(max_abs(decoded) <= kZeroTolerance, "zero vector decode should remain near zero");
+}
+
+void test_gaussian_roundtrip_quality() {
+    std::mt19937 rng(12345);
+    std::normal_distribution<float> dist(0.0f, 1.0f);
+
+    std::vector<float> input(kDim, 0.0f);
+    for (float & value : input) {
+        value = dist(rng);
+    }
+
+    float norm = -1.0f;
+    const auto decoded = roundtrip(input, norm);
+
+    require(norm > 0.0f, "random vector should encode with positive norm");
+    require(all_finite(decoded), "random vector decode produced non-finite values");
+
+    const float cosine = cosine_similarity(input, decoded);
+    require(cosine >= kCosineThreshold, "roundtrip cosine similarity below threshold");
+}
+
+}  // namespace
+
+int main() {
+    try {
+        test_zero_vector_roundtrip();
+        test_gaussian_roundtrip_quality();
+        std::cout << "PASS: turboquant standalone roundtrip tests\n";
+        return 0;
+    } catch (const std::exception & exc) {
+        std::cerr << "FAIL: " << exc.what() << '\n';
+        return 1;
+    }
+}

tests/test_tool_calling.py

@@ -0,0 +1,189 @@
+#!/usr/bin/env python3
+"""
+Unit tests for benchmarks/test_tool_calling.py
+
+Tests the validation logic and report generation without
+requiring a live model backend.
+"""
+
+import json
+import sys
+from pathlib import Path
+
+import pytest
+
+sys.path.insert(0, str(Path(__file__).parent.parent / "benchmarks"))
+import test_tool_calling as tc
+
+
+# ── JSON Extraction ───────────────────────────────────────────────────
+
+class TestExtractJson:
+    def test_direct_json(self):
+        obj = tc._extract_json('{"name": "read_file", "arguments": {"path": "/etc/hostname"}}')
+        assert obj["name"] == "read_file"
+
+    def test_json_in_code_block(self):
+        text = 'Here is the call:\n```json\n{"name": "terminal", "arguments": {"command": "ls"}}\n```'
+        obj = tc._extract_json(text)
+        assert obj["name"] == "terminal"
+
+    def test_json_without_lang(self):
+        text = '```\n{"name": "web_search", "arguments": {"query": "test"}}\n```'
+        obj = tc._extract_json(text)
+        assert obj["name"] == "web_search"
+
+    def test_no_json(self):
+        obj = tc._extract_json("I can't help with that.")
+        assert obj is None
+
+    def test_bare_json_object(self):
+        text = 'Sure, here: {"name": "read_file", "arguments": {"path": "/tmp/x"}} for you.'
+        obj = tc._extract_json(text)
+        assert obj is not None
+        assert obj["name"] == "read_file"
+
+
+# ── Tool Call Validation ──────────────────────────────────────────────
+
+class TestToolCallValidation:
+    def test_exact_match(self):
+        resp = '{"name": "read_file", "arguments": {"path": "/etc/hostname"}}'
+        result = tc._has_json_tool_call(resp, "read_file", {"path": "/etc/hostname"})
+        assert result["passed"] is True
+
+    def test_wrong_tool_name(self):
+        resp = '{"name": "write_file", "arguments": {"path": "/etc/hostname"}}'
+        result = tc._has_json_tool_call(resp, "read_file", {"path": "/etc/hostname"})
+        assert result["passed"] is False
+        assert "wrong tool name" in result["reason"]
+
+    def test_missing_argument(self):
+        resp = '{"name": "read_file", "arguments": {}}'
+        result = tc._has_json_tool_call(resp, "read_file", {"path": "/etc/hostname"})
+        assert result["passed"] is False
+        assert "missing argument" in result["reason"]
+
+    def test_wrong_argument_value(self):
+        resp = '{"name": "read_file", "arguments": {"path": "/etc/passwd"}}'
+        result = tc._has_json_tool_call(resp, "read_file", {"path": "/etc/hostname"})
+        assert result["passed"] is False
+        assert "argument mismatch" in result["reason"]
+
+    def test_no_json_response(self):
+        result = tc._has_json_tool_call("Sorry, I can't do that.", "read_file", {"path": "/etc/hostname"})
+        assert result["passed"] is False
+        assert "no JSON" in result["reason"]
+
+    def test_nested_function_format(self):
+        resp = '{"function": {"name": "terminal", "arguments": {"command": "echo hello"}}}'
+        result = tc._has_json_tool_call(resp, "terminal", {"command": "echo hello"})
+        assert result["passed"] is True
+
+
+# ── Nested Schema Validation ──────────────────────────────────────────
+
+class TestNestedSchemaValidation:
+    def test_valid_nested(self):
+        resp = json.dumps({
+            "name": "deploy_service",
+            "arguments": {
+                "name": "api-gateway",
+                "replicas": 3,
+                "env": {"PORT": 8080, "NODE_ENV": "production"},
+                "resources": {"cpu": "500m", "memory": "256Mi"}
+            }
+        })
+        result = tc._has_nested_tool_call(resp)
+        assert result["passed"] is True
+
+    def test_missing_nested_key(self):
+        resp = '{"name": "deploy_service", "arguments": {"name": "api-gateway", "replicas": 3}}'
+        result = tc._has_nested_tool_call(resp)
+        assert result["passed"] is False
+        assert "missing nested key" in result["reason"]
+
+    def test_wrong_type(self):
+        resp = '{"name": "deploy_service", "arguments": {"name": "api-gateway", "replicas": "three", "env": {}, "resources": {}}}'
+        result = tc._has_nested_tool_call(resp)
+        assert result["passed"] is False
+        assert "should be int" in result["reason"]
+
+    def test_missing_env_port(self):
+        resp = json.dumps({
+            "name": "deploy_service",
+            "arguments": {"name": "api", "replicas": 1, "env": {"NODE_ENV": "dev"}, "resources": {}}
+        })
+        result = tc._has_nested_tool_call(resp)
+        assert result["passed"] is False
+        assert "PORT" in result["reason"]
+
+
+# ── Markdown Report Generation ────────────────────────────────────────
+
+class TestMarkdownReport:
+    def test_report_structure(self):
+        results = {
+            "model": "test-model",
+            "backend": "ollama",
+            "url": "http://localhost:11434",
+            "timestamp": "2026-04-15T00:00:00Z",
+            "tests": [
+                {"id": "t1", "category": "Simple", "description": "Test 1",
+                 "passed": True, "reason": "ok", "response": "{}", "latency_s": 1.0, "tokens": 10},
+                {"id": "t2", "category": "Complex", "description": "Test 2",
+                 "passed": False, "reason": "wrong name", "response": "oops", "latency_s": 2.0, "tokens": 20},
+            ],
+            "summary": {"total": 2, "passed": 1, "failed": 1, "errors": 0},
+        }
+        md = tc.to_markdown(results)
+        assert "test-model" in md
+        assert "1/2 passed" in md
+        assert "PASS" in md
+        assert "FAIL" in md
+        assert "Failure Analysis" in md
+
+    def test_perfect_score(self):
+        results = {
+            "model": "perfect", "backend": "ollama", "url": "http://x",
+            "timestamp": "2026-01-01T00:00:00Z",
+            "tests": [
+                {"id": "t1", "category": "C", "description": "D",
+                 "passed": True, "reason": "ok", "response": "{}", "latency_s": 1, "tokens": 5},
+            ],
+            "summary": {"total": 1, "passed": 1, "failed": 0, "errors": 0},
+        }
+        md = tc.to_markdown(results)
+        assert "FULLY VIABLE" in md
+
+    def test_all_failed(self):
+        results = {
+            "model": "bad", "backend": "ollama", "url": "http://x",
+            "timestamp": "2026-01-01T00:00:00Z",
+            "tests": [
+                {"id": "t1", "category": "C", "description": "D",
+                 "passed": False, "reason": "broken", "response": "nope", "latency_s": 1, "tokens": 0},
+            ],
+            "summary": {"total": 1, "passed": 0, "failed": 1, "errors": 0},
+        }
+        md = tc.to_markdown(results)
+        assert "NOT VIABLE" in md
+
+
+# ── Test Definitions ──────────────────────────────────────────────────
+
+class TestTestDefinitions:
+    def test_all_tests_have_validators(self):
+        for test in tc.TESTS:
+            assert callable(test["validate"]), f"{test['id']} missing validate"
+            assert "id" in test
+            assert "category" in test
+            assert "prompt" in test
+
+    def test_five_test_categories(self):
+        categories = {t["category"] for t in tc.TESTS}
+        assert len(categories) >= 4, f"Expected 4+ categories, got {categories}"
+
+
+if __name__ == "__main__":
+    pytest.main([__file__, "-v"])

tests/test_turboquant.py

@@ -1,141 +0,0 @@
-#!/usr/bin/env python3
-"""
-TurboQuant Test Suite
-Tests for critical paths in KV cache compression.
-
-Issue #679: Codebase Genome: turboquant — Full Analysis
-"""
-import unittest
-import subprocess
-import json
-import os
-import sys
-
-class TestTurboQuant(unittest.TestCase):
-    """Test TurboQuant implementation."""
-    
-    def test_repo_structure(self):
-        """Verify expected files exist."""
-        required_files = [
-            "llama-turbo.h",
-            "llama-turbo.cpp",
-            "ggml-metal-turbo.metal",
-            "README.md",
-            "GENOME.md"
-        ]
-        
-        for filename in required_files:
-            filepath = os.path.join(os.path.dirname(__file__), "..", filename)
-            self.assertTrue(os.path.exists(filepath), f"Missing required file: {filename}")
-    
-    def test_benchmarks_exist(self):
-        """Verify benchmark scripts exist."""
-        benchmark_files = [
-            "benchmarks/run_benchmarks.py",
-            "benchmarks/run_perplexity.py",
-            "benchmarks/run_long_session.py"
-        ]
-        
-        for filename in benchmark_files:
-            filepath = os.path.join(os.path.dirname(__file__), "..", filename)
-            self.assertTrue(os.path.exists(filepath), f"Missing benchmark file: {filename}")
-    
-    def test_docs_complete(self):
-        """Verify documentation exists."""
-        doc_files = [
-            "docs/PROJECT_STATUS.md",
-            "profiles/README.md"
-        ]
-        
-        for filename in doc_files:
-            filepath = os.path.join(os.path.dirname(__file__), "..", filename)
-            self.assertTrue(os.path.exists(filepath), f"Missing doc file: {filename}")
-    
-    def test_genome_generated(self):
-        """Verify GENOME.md was generated."""
-        genome_path = os.path.join(os.path.dirname(__file__), "..", "GENOME.md")
-        self.assertTrue(os.path.exists(genome_path), "GENOME.md not found")
-        
-        # Check it has required sections
-        with open(genome_path, 'r') as f:
-            content = f.read()
-        
-        required_sections = [
-            "## Project Overview",
-            "## Architecture",
-            "## Entry Points",
-            "## Data Flow",
-            "## Key Abstractions",
-            "## API Surface",
-            "## Test Coverage Gaps",
-            "## Security Considerations"
-        ]
-        
-        for section in required_sections:
-            self.assertIn(section, content, f"GENOME.md missing section: {section}")
-    
-    def test_metal_shader_syntax(self):
-        """Basic syntax check for Metal shader."""
-        shader_path = os.path.join(os.path.dirname(__file__), "..", "ggml-metal-turbo.metal")
-        with open(shader_path, 'r') as f:
-            content = f.read()
-        
-        # Check for key functions
-        self.assertIn("kernel_fwht_128", content, "Missing kernel_fwht_128 function")
-        self.assertIn("kernel_turbo4_dequant", content, "Missing kernel_turbo4_dequant function")
-        self.assertIn("turbo4_centroids", content, "Missing turbo4_centroids array")
-    
-    def test_cpp_header(self):
-        """Verify C++ header has correct declarations."""
-        header_path = os.path.join(os.path.dirname(__file__), "..", "llama-turbo.h")
-        with open(header_path, 'r') as f:
-            content = f.read()
-        
-        # Check for function declarations
-        self.assertIn("polar_quant_encode_turbo4", content, "Missing encode function")
-        self.assertIn("polar_quant_decode_turbo4", content, "Missing decode function")
-        self.assertIn('extern "C"', content, "Missing C linkage")
-
-class TestBenchmarks(unittest.TestCase):
-    """Test benchmark infrastructure."""
-    
-    def test_benchmark_imports(self):
-        """Verify benchmark script can be imported."""
-        benchmark_path = os.path.join(os.path.dirname(__file__), "..", "benchmarks", "run_benchmarks.py")
-        
-        # Check file exists
-        self.assertTrue(os.path.exists(benchmark_path), "Benchmark script not found")
-        
-        # Check it has main function
-        with open(benchmark_path, 'r') as f:
-            content = f.read()
-        
-        self.assertIn("def main():", content, "Benchmark script missing main function")
-        self.assertIn("argparse", content, "Benchmark script missing argparse")
-
-class TestDocumentation(unittest.TestCase):
-    """Test documentation completeness."""
-    
-    def test_readme_sections(self):
-        """Verify README has required sections."""
-        readme_path = os.path.join(os.path.dirname(__file__), "..", "README.md")
-        with open(readme_path, 'r') as f:
-            content = f.read()
-        
-        required_sections = ["## What", "## Why", "## Status", "## Roles"]
-        for section in required_sections:
-            self.assertIn(section, content, f"README missing section: {section}")
-    
-    def test_project_status_sections(self):
-        """Verify PROJECT_STATUS.md has required sections."""
-        status_path = os.path.join(os.path.dirname(__file__), "..", "docs", "PROJECT_STATUS.md")
-        with open(status_path, 'r') as f:
-            content = f.read()
-        
-        # Check for key findings
-        self.assertIn("73%", content, "Missing 73% savings metric")
-        self.assertIn("PolarQuant", content, "Missing PolarQuant references")
-        self.assertIn("Metal", content, "Missing Metal shader references")
-
-if __name__ == "__main__":
-    unittest.main()