Merge 44c06b11ec into f5d0e1e69e

Users who downloaded ruvnet/wifi-densepose-pretrained and passed
model.safetensors / model-q4.bin / model.rvf.jsonl to --model hit a bare
"Progressive loader init failed: invalid magic at offset 0: expected
0x52564653, got 0x77455735" and were stuck — the server then silently fell
back to signal heuristics (which over-count, feeding "is it fake" reports).

The HF files are a different *format* and encoder architecture than the RVF
binary container the progressive loader expects, so they can't load directly.
Now the load-failure path detects the common cases (safetensors header,
JSONL manifest, quantized .bin blob) and emits a plain explanation naming the
format, what --model actually expects (RVF `RVFS` container from
wifi-densepose-train), and that it's continuing with heuristics — with a
pointer to #894.

Pure, testable `diagnose_model_load_error()` + 4 unit tests (run under the
default `--no-default-features` CI). Full crate unit suite: 429 + 114 passed,
0 failed.

.github/workflows/ci.yml

@@ -265,23 +265,45 @@ jobs:
       run: |
         python -m pip install --upgrade pip
         pip install -r requirements.txt
-        pip install locust
+        pip install pytest   # the perf suite is pytest, not locust
 
-    - name: Start application
-      working-directory: archive/v1
-      run: |
-        uvicorn src.api.main:app --host 0.0.0.0 --port 8000 &
-        sleep 10
+    # No "Start application" step: the gated test (test_frame_budget.py) drives
+    # the CSIProcessor pipeline in-process and makes no HTTP calls, so the old
+    # uvicorn server + `sleep 10` were dead weight — they only existed for the
+    # now-excluded api_throughput/inference_speed tests, and on every run dumped
+    # ~50 misleading "router requires hardware setup" ERROR lines for a server
+    # no test touched. MOCK_POSE_DATA is server-only and unused here.
 
     - name: Run performance tests
+      working-directory: archive/v1
       run: |
-        locust -f tests/performance/locustfile.py --headless --users 50 --spawn-rate 5 --run-time 60s --host http://localhost:8000
+        # Gate only on the genuine, deterministic perf guard:
+        # test_frame_budget.py times the *real* CSIProcessor pipeline against
+        # the ADR 50 ms per-frame budget (single-frame, p95 over 100 frames,
+        # +Doppler) — a true regression signal.
+        #
+        # test_api_throughput.py / test_inference_speed.py are excluded: every
+        # test there is a TDD red-phase stub (suffix `_should_fail_initially`)
+        # that times a *mock that sleeps* — meaningless as a perf signal, with
+        # machine-dependent wall-clock asserts (e.g. `actual_rps >= 40`,
+        # `batch_time < individual_time`) that are inherently flaky on shared
+        # CI runners, plus a cross-class fixture-scope bug. Forcing them green
+        # would be manufacturing a false signal; they stay in-repo for local
+        # TDD but do not gate CI until the underlying features are implemented.
+        #
+        # `python -m pytest` (not the bare `pytest` script) puts the cwd
+        # (archive/v1) on sys.path so `from src.core...` resolves — the bare
+        # script omits cwd and raises ModuleNotFoundError: No module named 'src'.
+        # -o addopts="" drops the root pyproject's --cov/--cov-fail-under=100.
+        python -m pytest tests/performance/test_frame_budget.py \
+          -o addopts="" -v --junitxml=perf-junit.xml
 
     - name: Upload performance results
+      if: always()
       uses: actions/upload-artifact@v4
       with:
         name: performance-results
-        path: locust_report.html
+        path: archive/v1/perf-junit.xml
 
   # Docker Build and Test
   # NOTE: the canonical Docker build for the sensing-server is now
@@ -367,6 +389,8 @@ jobs:
     runs-on: ubuntu-latest
     needs: [docker-build]
     if: github.ref == 'refs/heads/main'
+    permissions:
+      contents: write   # gh-pages deploy needs write (GITHUB_TOKEN is read-only by default -> 403)
     steps:
     - name: Checkout code
       uses: actions/checkout@v4
@@ -384,6 +408,8 @@ jobs:
 
     - name: Generate OpenAPI spec
       working-directory: archive/v1
+      env:
+        MOCK_POSE_DATA: "true"   # no CSI hardware in CI
       run: |
         python -c "
         from src.api.main import app
@@ -394,6 +420,7 @@ jobs:
 
     - name: Deploy to GitHub Pages
       uses: peaceiris/actions-gh-pages@v4
+      continue-on-error: true   # openapi generation above is the real validation; deploy is best-effort (Pages may be disabled)
       with:
         github_token: ${{ secrets.GITHUB_TOKEN }}
         publish_dir: ./docs

.gitignore

@@ -256,6 +256,12 @@ models/
 demo_pointcloud.ply
 demo_splats.json
 
+# Downloaded Hugging Face pretrained weights (not for git)
+/models/wifi-densepose-pretrained/
+
+# Session scratch files
+/.tmp_*
+
 # rvCSI napi-rs addon — generated by `napi build` (do not commit)
 v2/crates/rvcsi-node/*.node
 v2/crates/rvcsi-node/binding.js

CHANGELOG.md

@@ -430,7 +430,7 @@ Model release (no new firmware binary). Firmware remains at v0.6.0-esp32.
 - Security fix merged via PR #310.
 
 ### Performance
-- Presence detection: 100% accuracy on 60,630 overnight samples.
+- Presence detection: 100% accuracy on 60,630 overnight samples. *(Retracted — that recording was single-class (one sleeping person, 6,062/6,063 frames "present"), so a constant "yes" scores ~99.98%. Superseded by the honest 82.3% held-out temporal-triplet metric; see [#882](https://github.com/ruvnet/RuView/issues/882). Kept here as the in-place public record.)*
 - Inference: 0.008 ms per sample, 164K embeddings/sec.
 - Contrastive self-supervised training: 51.6% improvement over baseline.
 

archive/v1/src/services/pose_service.py

@@ -107,16 +107,25 @@ class PoseService:
     async def _initialize_models(self):
         """Initialize neural network models."""
         try:
-            # Initialize DensePose model
+            # Initialize DensePose model. DensePoseHead requires a config
+            # dict — input_channels matches the modality translator's output
+            # (256), with the standard DensePose 24 body parts and 2 (U,V)
+            # coordinates. (Previously called with no args → TypeError at
+            # startup, which broke the API service.)
+            densepose_config = {
+                'input_channels': 256,
+                'num_body_parts': 24,
+                'num_uv_coordinates': 2,
+            }
             if self.settings.pose_model_path:
-                self.densepose_model = DensePoseHead()
+                self.densepose_model = DensePoseHead(densepose_config)
                 # Load model weights if path is provided
                 # model_state = torch.load(self.settings.pose_model_path)
                 # self.densepose_model.load_state_dict(model_state)
                 self.logger.info("DensePose model loaded")
             else:
                 self.logger.warning("No pose model path provided, using default model")
-                self.densepose_model = DensePoseHead()
+                self.densepose_model = DensePoseHead(densepose_config)
             
             # Initialize modality translation
             config = {

docs/readme-details.md

@@ -122,7 +122,7 @@ node scripts/benchmark-ruvllm.js --model models/csi-ruvllm       # benchmark
 
 | What we measured | Result | Why it matters |
 |-----------------|--------|---------------|
-| **Presence detection** | **100% accuracy** | Never misses a person, never false alarms |
+| **CSI embedding quality** | **82.3% held-out temporal-triplet** | Honest label-free metric on the last 20% by time (v1's "100% presence" was a single-class recording — retracted, [#882](https://github.com/ruvnet/RuView/issues/882)) |
 | **Inference speed** | **0.008 ms** per embedding | 125,000x faster than real-time |
 | **Throughput** | **164,183 embeddings/sec** | One Mac Mini handles 1,600+ ESP32 nodes |
 | **Contrastive learning** | **51.6% improvement** | Strong pattern learning from real overnight data |
@@ -233,7 +233,7 @@ python firmware/esp32-csi-node/provision.py --port COM9 --hop-channels "1,6,11"
 | **kNN similarity search** | "Find the 10 most similar states to right now" — anomaly detection, fingerprinting | Cognitum Seed |
 | **Witness chain** | SHA-256 tamper-evident audit trail for every measurement (1,747 entries validated) | Cognitum Seed |
 | **Camera-free pose training** | 17 COCO keypoints from 10 sensor signals — PIR, RSSI triangulation, subcarrier asymmetry, vibration, BME280 | 2x ESP32 + Seed |
-| **Pre-trained model** | 82.8 KB (8 KB at 4-bit quantization), 100% presence accuracy, 0 skeleton violations | Download from release |
+| **Pre-trained model** | 82.8 KB (8 KB at 4-bit quantization), 82.3% held-out temporal-triplet accuracy (v1's "100% presence" was single-class — retracted, [#882](https://github.com/ruvnet/RuView/issues/882)) | Download from release |
 | **Sub-ms inference** | 0.012 ms latency, 171,472 embeddings/sec on M4 Pro | Any machine with Node.js |
 | **SONA adaptation** | Adapts to new rooms in <1ms without retraining | ruvllm runtime |
 | **LoRA room adapters** | Per-node fine-tuning with 2,048 parameters per adapter | Automatic |
@@ -262,7 +262,7 @@ node scripts/benchmark-ruvllm.js --model models/csi-ruvllm
 
 | What we measured | Result | Why it matters |
 |-----------------|--------|---------------|
-| **Presence detection** | **100% accuracy** | Never misses a person, never false alarms |
+| **CSI embedding quality** | **82.3% held-out temporal-triplet** | Honest label-free metric (v1's "100% presence" was single-class — retracted, [#882](https://github.com/ruvnet/RuView/issues/882)) |
 | **Person counting** | **24/24 correct** (MinCut) | Fixed the #1 user-reported issue |
 | **Inference speed** | **0.012 ms** per embedding | 83,000x faster than real-time |
 | **Throughput** | **171,472 embeddings/sec** | One Mac Mini handles 1,700+ ESP32 nodes |

docs/user-guide.md

@@ -1119,7 +1119,7 @@ What it ships (and what it does not):
 
 | Capability | Status |
 |------------|--------|
-| Presence detection (occupied / empty) | ✅ Trained head — 100% accuracy on validation |
+| Presence detection (occupied / empty) | ✅ Trained head — v2 encoder reports 82.3% held-out temporal-triplet acc (v1's "100% on validation" was a single-class recording — retracted, [#882](https://github.com/ruvnet/RuView/issues/882)) |
 | 128-dim CSI embeddings (re-ID, similarity, downstream training) | ✅ Trained encoder |
 | Single-person breathing / heart-rate | ⚠️ Server still uses heuristic DSP — model does not replace this yet |
 | 17-keypoint full-body pose | 🔬 No keypoint weights shipped yet — pose pipeline runs but without a learned head |
@@ -1168,6 +1168,10 @@ cargo run -p wifi-densepose-sensing-server --release -- \
     --source esp32 --udp-port 5005 --http-port 3000
 ```
 
+### Verifying the downloaded bundle
+
+After downloading, run [`scripts/verify-hf-model.py`](../scripts/verify-hf-model.py) to confirm the bundle is intact and the weights load cleanly. The script prints the safetensors tensor inventory (names, shapes, dtypes), parses `model.rvf.jsonl` line by line, dumps `presence-head.json` / `config.json` / `training-metrics.json`, and — when `torch` is available — pushes a synthetic batch through the first encoder linear layer to confirm no NaN / Inf. It exits 0 on success, non-zero with a clear error otherwise, and accepts `--local-dir <path>` (default `models/wifi-densepose-pretrained/`). Heads up: the published `model.safetensors` header has three trailing NUL bytes that the strict Rust loader rejects (`trailing characters at line 1 column 1462`); the script falls back to a small pure-Python reader that strips the padding so you still get the full inventory.
+
 See [RVF Model Containers](#rvf-model-containers) for the binary format the loader expects, and [Training a Model](#training-a-model) for using the encoder as a starting point for environment-specific fine-tuning.
 
 ---
@@ -1824,7 +1828,7 @@ huggingface-cli download ruvnet/wifi-densepose-pretrained --local-dir models/pre
 #   model.safetensors    — 48 KB contrastive encoder
 #   model-q4.bin         — 8 KB quantized (recommended)
 #   model-q2.bin         — 4 KB ultra-compact (ESP32 edge)
-#   presence-head.json   — presence detection head (100% accuracy)
+#   presence-head.json   — presence detection head (v2 encoder: 82.3% held-out triplet acc)
 #   node-1.json          — LoRA adapter for room 1
 #   node-2.json          — LoRA adapter for room 2
 ```
@@ -1833,7 +1837,7 @@ huggingface-cli download ruvnet/wifi-densepose-pretrained --local-dir models/pre
 
 The pre-trained encoder converts 8-dim CSI feature vectors into 128-dim embeddings. These embeddings power all 17 sensing applications:
 
-- **Presence detection** — 100% accuracy, never misses, never false alarms
+- **Presence detection** — v2 encoder: 82.3% held-out temporal-triplet accuracy (v1's "100%" was a single-class recording — retracted, [#882](https://github.com/ruvnet/RuView/issues/882))
 - **Environment fingerprinting** — kNN search finds "states like this one"
 - **Anomaly detection** — embeddings that don't match known clusters = anomaly
 - **Activity classification** — different activities cluster in embedding space

firmware/esp32-csi-node/release_bins/c6-adr110/SHA256SUMS.txt

@@ -1,4 +1,4 @@
-889715e9d698ad78f9978ad8b93b6af24a726b0494247201c8f0d920d9fc80ca *firmware/esp32-csi-node/release_bins/c6-adr110/bootloader.bin
-d8539e47c6f10a3344679118619e3fe01cfd66eb560ea8883268ca7c9a12efa4 *firmware/esp32-csi-node/release_bins/c6-adr110/esp32-csi-node.bin
+b0fb1f217a39c80bc95b5eb8208a0b8572ae64efa0f6d580b76caff4affe0f4d *firmware/esp32-csi-node/release_bins/c6-adr110/bootloader.bin
+4764c5b20a353895f70122816adc98f861ec20e9a8ea9b344dc0648b6341073c *firmware/esp32-csi-node/release_bins/c6-adr110/esp32-csi-node.bin
 7d2c7ac4888bfd75cd5f56e8d61f69595121183afc81556c876732fd3782c62f *firmware/esp32-csi-node/release_bins/c6-adr110/ota_data_initial.bin
 4c2cc4ffd52641e23b779bd57b3908014083ac3c1aab395756478c89e70d81f0 *firmware/esp32-csi-node/release_bins/c6-adr110/partition-table.bin

firmware/esp32-csi-node/release_bins/s3-adr110/SHA256SUMS.txt

@@ -1,3 +1,3 @@
-3c4905dd202ccabf4230cbabcc9320f250a60b1a7254eff7424780201bcb2072 *firmware/esp32-csi-node/release_bins/s3-adr110/bootloader.bin
-7a8bf9582c9031fed32f1ada44f5c41dd99bd07fadff8e5c86e07aa0f343e847 *firmware/esp32-csi-node/release_bins/s3-adr110/esp32-csi-node.bin
+b973d7eda65affb746adcfa63ceb18f779f206d240b76f01b8c9ae7485455660 *firmware/esp32-csi-node/release_bins/s3-adr110/bootloader.bin
+e21ef94aba779d534dc048c1b9da731c81e5dbe09d0645cfd70a05ad3642d3e9 *firmware/esp32-csi-node/release_bins/s3-adr110/esp32-csi-node.bin
 67222c257c0477501fd4002275638dc4262b34eb68235b8289fb1337054d322b *firmware/esp32-csi-node/release_bins/s3-adr110/partition-table.bin

firmware/esp32-csi-node/release_bins/version.txt

@@ -1,3 +1,4 @@
-0.6.6
-git-sha: cbcb389cb (pre-commit)
-built: 2026-05-21
+0.6.7
+git-sha: 8703ade9b
+built: 2026-06-02
+note: RuView#893 — display-less boards capture DATA frames (CSI yield 0pps fix); hardware-verified on ESP32-C6 (0->27 pps)

requirements.txt

@@ -36,3 +36,4 @@ scikit-learn>=1.2.0
 
 # Monitoring dependencies
 prometheus-client>=0.16.0
+psutil>=5.9.0  # system metrics — imported by health.py / metrics.py / status.py / monitoring.py

scripts/verify-hf-model.py

@@ -0,0 +1,343 @@
+#!/usr/bin/env python3
+"""Verify the published ruvnet/wifi-densepose-pretrained model bundle.
+
+Inspects every file in the downloaded model directory:
+  - model.safetensors     -> tensor names + shapes + dtypes
+  - model.rvf.jsonl       -> line count, first three lines, distinct top-level keys
+  - presence-head.json    -> shallow dump (depth <= 3)
+  - config.json           -> full dump
+  - training-metrics.json -> final loss / quantization / lora numbers
+
+If torch is importable, builds a synthetic input matching the inferred encoder
+input dim and runs encoder.w1 (the first linear layer) to confirm the weights
+yield finite outputs (no NaN / Inf).
+
+Exits 0 on success, non-zero with a clear error on any failure.
+
+Usage:
+    python scripts/verify-hf-model.py
+    python scripts/verify-hf-model.py --local-dir models/wifi-densepose-pretrained/
+"""
+
+from __future__ import annotations
+
+import argparse
+import json
+import struct
+import sys
+from pathlib import Path
+from typing import Any
+
+DEFAULT_LOCAL_DIR = Path("models/wifi-densepose-pretrained/")
+
+# safetensors -> torch dtype lookup. Subset is enough for this bundle.
+_SAFETENSORS_DTYPE_NAMES = {
+    "F64", "F32", "F16", "BF16",
+    "I64", "I32", "I16", "I8", "U8", "BOOL",
+}
+
+
+# --------------------------------------------------------------------------- #
+# safetensors loading
+# --------------------------------------------------------------------------- #
+def _load_safetensors(path: Path):
+    """Load a .safetensors file as a dict[name -> torch.Tensor].
+
+    Tries the upstream `safetensors.torch.load_file` first. The published HF
+    bundle has a non-fatal header bug (declared header length includes 3
+    trailing NUL bytes after the JSON object), which the strict Rust parser
+    rejects with `trailing characters at line 1 column 1462`. When that
+    happens we fall back to a small pure-Python loader that strips the
+    padding and rebuilds tensors from the body.
+    """
+    try:
+        from safetensors.torch import load_file  # type: ignore
+
+        return load_file(str(path)), "safetensors.torch.load_file"
+    except Exception as exc:  # noqa: BLE001 - we want any failure here
+        msg = str(exc)
+        if "trailing characters" not in msg and "invalid JSON" not in msg:
+            raise
+        # Fall through to the manual loader below.
+        first_err = f"{type(exc).__name__}: {exc}"
+
+    import torch  # local import so the fallback message is precise
+
+    dtype_map = {
+        "F64": torch.float64,
+        "F32": torch.float32,
+        "F16": torch.float16,
+        "BF16": torch.bfloat16,
+        "I64": torch.int64,
+        "I32": torch.int32,
+        "I16": torch.int16,
+        "I8": torch.int8,
+        "U8": torch.uint8,
+        "BOOL": torch.bool,
+    }
+    raw = path.read_bytes()
+    if len(raw) < 8:
+        raise ValueError(f"{path}: file too short to be a safetensors blob")
+    header_len = struct.unpack("<Q", raw[:8])[0]
+    if header_len <= 0 or 8 + header_len > len(raw):
+        raise ValueError(
+            f"{path}: header length {header_len} inconsistent with file size {len(raw)}"
+        )
+    header_bytes = raw[8 : 8 + header_len]
+    # Strip the published-bundle trailing padding (NULs / whitespace) before parsing.
+    header_text = header_bytes.rstrip(b"\x00 \t\r\n").decode("utf-8")
+    header = json.loads(header_text)
+    body = raw[8 + header_len :]
+
+    state: dict[str, Any] = {}
+    for name, info in header.items():
+        if name == "__metadata__":
+            continue
+        dtype_name = info["dtype"]
+        if dtype_name not in dtype_map:
+            raise ValueError(f"{name}: unsupported safetensors dtype {dtype_name!r}")
+        shape = list(info["shape"])
+        start, end = info["data_offsets"]
+        if start < 0 or end > len(body) or start > end:
+            raise ValueError(
+                f"{name}: bad offsets [{start}, {end}] for body of size {len(body)}"
+            )
+        tensor = torch.frombuffer(
+            bytearray(body[start:end]), dtype=dtype_map[dtype_name]
+        ).reshape(shape)
+        state[name] = tensor.clone()  # detach from the bytearray buffer
+    return state, (
+        "manual fallback (published bundle has trailing NULs in header; "
+        f"first error was: {first_err})"
+    )
+
+
+# --------------------------------------------------------------------------- #
+# JSONL helpers
+# --------------------------------------------------------------------------- #
+def _truncate(s: str, n: int) -> str:
+    return s if len(s) <= n else s[:n] + "..."
+
+
+def _inspect_jsonl(path: Path) -> tuple[int, list[str], list[str]]:
+    """Return (line_count, first_three_truncated, sorted_distinct_top_keys)."""
+    lines: list[str] = []
+    keys: set[str] = set()
+    total = 0
+    with path.open("r", encoding="utf-8") as fh:
+        for idx, raw_line in enumerate(fh):
+            line = raw_line.rstrip("\n")
+            if not line.strip():
+                continue
+            total += 1
+            if idx < 3:
+                lines.append(_truncate(line, 200))
+            try:
+                obj = json.loads(line)
+            except json.JSONDecodeError as exc:
+                raise ValueError(f"{path}: line {idx + 1} is not valid JSON: {exc}") from exc
+            if isinstance(obj, dict):
+                keys.update(obj.keys())
+    return total, lines, sorted(keys)
+
+
+# --------------------------------------------------------------------------- #
+# Pretty printers
+# --------------------------------------------------------------------------- #
+def _dump_shallow(obj: Any, depth: int = 0, max_depth: int = 3) -> str:
+    """Render `obj` as JSON, but collapse anything below max_depth to its type."""
+    if depth >= max_depth:
+        if isinstance(obj, dict):
+            return f"<dict with {len(obj)} keys>"
+        if isinstance(obj, list):
+            return f"<list len={len(obj)}>"
+        return repr(obj)
+    if isinstance(obj, dict):
+        body = ", ".join(
+            f'"{k}": {_dump_shallow(v, depth + 1, max_depth)}' for k, v in obj.items()
+        )
+        return "{" + body + "}"
+    if isinstance(obj, list):
+        if len(obj) > 8:
+            sample = ", ".join(_dump_shallow(v, depth + 1, max_depth) for v in obj[:8])
+            return f"[{sample}, ... (+{len(obj) - 8} more)]"
+        return "[" + ", ".join(_dump_shallow(v, depth + 1, max_depth) for v in obj) + "]"
+    return json.dumps(obj)
+
+
+def _section(title: str) -> None:
+    print()
+    print("=" * 78)
+    print(title)
+    print("=" * 78)
+
+
+# --------------------------------------------------------------------------- #
+# Verification steps
+# --------------------------------------------------------------------------- #
+def _verify_safetensors(path: Path) -> tuple[dict, str, dict[str, tuple]]:
+    state, loader_note = _load_safetensors(path)
+    info: dict[str, tuple] = {}
+    print(f"loader: {loader_note}")
+    print(f"tensor count: {len(state)}")
+    print(f"{'name':<30} {'shape':<22} dtype")
+    print("-" * 78)
+    for name, tensor in state.items():
+        shape = tuple(tensor.shape)
+        dtype = str(tensor.dtype)
+        info[name] = (shape, dtype)
+        print(f"{name:<30} {str(shape):<22} {dtype}")
+    return state, loader_note, info
+
+
+def _verify_jsonl(path: Path) -> None:
+    total, sample, keys = _inspect_jsonl(path)
+    print(f"line count: {total}")
+    print(f"distinct top-level keys observed: {keys}")
+    print("first 3 lines (truncated to 200 chars):")
+    for idx, line in enumerate(sample, start=1):
+        print(f"  [{idx}] {line}")
+
+
+def _verify_presence_head(path: Path) -> None:
+    obj = json.loads(path.read_text(encoding="utf-8"))
+    print(_dump_shallow(obj, max_depth=3))
+
+
+def _verify_config(path: Path) -> dict:
+    obj = json.loads(path.read_text(encoding="utf-8"))
+    print(json.dumps(obj, indent=2, sort_keys=True))
+    return obj
+
+
+def _verify_training_metrics(path: Path) -> None:
+    obj = json.loads(path.read_text(encoding="utf-8"))
+    # Final metrics live in a few specific places.
+    contrastive = obj.get("contrastive", {})
+    task_heads = obj.get("taskHeads", {})
+    lora = obj.get("lora", {})
+    quant = obj.get("quantization", {})
+    print(f"timestamp:           {obj.get('timestamp')}")
+    print(f"total duration (ms): {obj.get('totalDurationMs')}")
+    print(f"contrastive triplets / final loss: "
+          f"{contrastive.get('triplets')} / {contrastive.get('finalLoss')}")
+    print(f"task heads samples / final loss:   "
+          f"{task_heads.get('samples')} / {task_heads.get('finalLoss')}")
+    print(f"lora adapters / total params: "
+          f"{lora.get('adapters')} / {lora.get('totalParameters')}")
+    if quant:
+        print("quantization:")
+        for variant, stats in quant.items():
+            print(f"  {variant}: {stats}")
+
+
+def _verify_first_linear(
+    state: dict, config: dict, tensor_info: dict[str, tuple]
+) -> None:
+    try:
+        import torch  # noqa: F401  (used below)
+    except ImportError:
+        print("torch not importable - skipping forward-pass smoke test")
+        return
+
+    import torch
+
+    custom = (config or {}).get("custom", {})
+    input_dim = int(custom.get("inputDim", 8))
+    hidden_dim = int(custom.get("hiddenDim", 64))
+
+    w1 = state.get("encoder.w1")
+    b1 = state.get("encoder.b1")
+    if w1 is None or b1 is None:
+        raise RuntimeError("encoder.w1 / encoder.b1 missing from safetensors")
+
+    # The published encoder stores the first linear weight flat (input_dim * hidden_dim).
+    if w1.numel() != input_dim * hidden_dim:
+        raise RuntimeError(
+            f"encoder.w1 numel={w1.numel()} does not match "
+            f"inputDim*hiddenDim={input_dim * hidden_dim}"
+        )
+    if b1.numel() != hidden_dim:
+        raise RuntimeError(
+            f"encoder.b1 numel={b1.numel()} does not match hiddenDim={hidden_dim}"
+        )
+
+    weight = w1.reshape(input_dim, hidden_dim).to(torch.float32)
+    bias = b1.to(torch.float32)
+
+    torch.manual_seed(42)
+    batch = 4
+    x = torch.randn(batch, input_dim, dtype=torch.float32)
+    y = x @ weight + bias
+
+    if not torch.isfinite(y).all():
+        bad = (~torch.isfinite(y)).sum().item()
+        raise RuntimeError(f"first linear layer produced {bad} non-finite values")
+    print(
+        f"first linear OK: input={tuple(x.shape)} weight={tuple(weight.shape)} "
+        f"bias={tuple(bias.shape)} output={tuple(y.shape)} "
+        f"mean={y.mean().item():+.4f} std={y.std().item():.4f}"
+    )
+
+
+# --------------------------------------------------------------------------- #
+# Entry point
+# --------------------------------------------------------------------------- #
+def main(argv: list[str] | None = None) -> int:
+    parser = argparse.ArgumentParser(description=__doc__)
+    parser.add_argument(
+        "--local-dir",
+        type=Path,
+        default=DEFAULT_LOCAL_DIR,
+        help="Directory containing the downloaded HF model bundle "
+             f"(default: {DEFAULT_LOCAL_DIR})",
+    )
+    args = parser.parse_args(argv)
+
+    root: Path = args.local_dir
+    if not root.is_dir():
+        print(f"ERROR: --local-dir does not exist or is not a directory: {root}",
+              file=sys.stderr)
+        return 2
+
+    safetensors_path = root / "model.safetensors"
+    rvf_path = root / "model.rvf.jsonl"
+    presence_path = root / "presence-head.json"
+    config_path = root / "config.json"
+    metrics_path = root / "training-metrics.json"
+
+    for p in (safetensors_path, rvf_path, presence_path, config_path, metrics_path):
+        if not p.is_file():
+            print(f"ERROR: required file missing: {p}", file=sys.stderr)
+            return 2
+
+    print(f"Verifying HF bundle at: {root}")
+
+    try:
+        _section("model.safetensors")
+        state, _loader_note, tensor_info = _verify_safetensors(safetensors_path)
+
+        _section("model.rvf.jsonl")
+        _verify_jsonl(rvf_path)
+
+        _section("presence-head.json (depth <= 3)")
+        _verify_presence_head(presence_path)
+
+        _section("config.json")
+        config = _verify_config(config_path)
+
+        _section("training-metrics.json (final metrics)")
+        _verify_training_metrics(metrics_path)
+
+        _section("encoder.w1 forward-pass smoke test")
+        _verify_first_linear(state, config, tensor_info)
+    except Exception as exc:  # noqa: BLE001 - surface anything as a clear failure
+        print(f"\nFAIL: {type(exc).__name__}: {exc}", file=sys.stderr)
+        return 1
+
+    _section("OK - all checks passed")
+    return 0
+
+
+if __name__ == "__main__":
+    sys.exit(main())

v2/crates/wifi-densepose-sensing-server/src/main.rs

@@ -5476,6 +5476,149 @@ async fn broadcast_tick_task(state: SharedState, tick_ms: u64) {
     }
 }
 
+/// Map one sensing-broadcast JSON document into the `VitalsSnapshot`(s) to
+/// publish over MQTT (issues #872/#898).
+///
+/// Multi-node sources carry a `nodes` array where **each node has its own
+/// `classification`** (`motion_level`, `presence`, `confidence`) and RSSI — so
+/// each node must surface its *own* presence/motion, not the room-level
+/// aggregate. Previously the bridge applied the aggregate `classification` to
+/// every per-node Home-Assistant device, so a node in an empty corner inherited
+/// another node's "present" (and `motion_level: "absent"` was mis-mapped to full
+/// motion). Vitals (breathing / heart rate) and the person count are room-level
+/// and shared across the per-node devices. Falls back to a single aggregate
+/// snapshot when there is no per-node data (e.g. wifi / simulate sources).
+#[cfg(feature = "mqtt")]
+fn vitals_snapshots_from_sensing_json(
+    v: &serde_json::Value,
+    base_id: &str,
+) -> Vec<wifi_densepose_sensing_server::mqtt::state::VitalsSnapshot> {
+    use wifi_densepose_sensing_server::mqtt::state::VitalsSnapshot;
+
+    // motion_level string -> motion scalar. "absent"/"none"/"still"/"idle"/""
+    // are non-moving; anything else (walking, …) is motion. `fallback` is used
+    // when the field is absent so a partial per-node payload defers to the
+    // room aggregate rather than silently reading 0.
+    fn motion_of(level: Option<&str>, fallback: f64) -> f64 {
+        match level {
+            Some("none") | Some("still") | Some("idle") | Some("absent") | Some("") => 0.0,
+            Some(_) => 1.0,
+            None => fallback,
+        }
+    }
+
+    let ts = (v["timestamp"].as_f64().unwrap_or(0.0) * 1000.0) as i64;
+    let vit = &v["vital_signs"];
+    let breathing = vit["breathing_rate_bpm"].as_f64();
+    let hr = vit["heart_rate_bpm"].as_f64();
+    let n_persons = v["persons"]
+        .as_array()
+        .map(|a| a.len() as u32)
+        .or_else(|| v["estimated_persons"].as_u64().map(|x| x as u32))
+        .unwrap_or(0);
+
+    // Room-level aggregate: the no-nodes fallback, and the per-node default for
+    // any field a node omits.
+    let acls = &v["classification"];
+    let agg_presence = acls["presence"].as_bool().unwrap_or(false);
+    let agg_motion = motion_of(acls["motion_level"].as_str(), 0.0);
+    let agg_conf = acls["confidence"].as_f64().unwrap_or(0.0);
+
+    let mk = |node_id: String, presence: bool, motion: f64, conf: f64, rssi: Option<f64>| {
+        VitalsSnapshot {
+            node_id,
+            timestamp_ms: ts,
+            presence,
+            motion,
+            presence_score: if presence { conf.max(0.0) } else { 0.0 },
+            breathing_rate_bpm: breathing,
+            heartrate_bpm: hr,
+            n_persons,
+            rssi_dbm: rssi,
+            vital_confidence: conf,
+            ..Default::default()
+        }
+    };
+
+    match v["nodes"].as_array() {
+        Some(arr) if !arr.is_empty() => arr
+            .iter()
+            .map(|node| {
+                let n = node["node_id"].as_u64().unwrap_or(0);
+                // Each node carries its OWN classification — use it, deferring to
+                // the room aggregate only for fields the node omits.
+                let ncls = &node["classification"];
+                let presence = ncls["presence"].as_bool().unwrap_or(agg_presence);
+                let motion = motion_of(ncls["motion_level"].as_str(), agg_motion);
+                let conf = ncls["confidence"].as_f64().unwrap_or(agg_conf);
+                mk(
+                    format!("{base_id}-node{n}"),
+                    presence,
+                    motion,
+                    conf,
+                    node["rssi_dbm"].as_f64(),
+                )
+            })
+            .collect(),
+        _ => vec![mk(
+            base_id.to_string(),
+            agg_presence,
+            agg_motion,
+            agg_conf,
+            v["nodes"][0]["rssi_dbm"].as_f64(),
+        )],
+    }
+}
+
+/// Turn a `ProgressiveLoader::new` failure into an actionable diagnostic (#894).
+///
+/// The published HuggingFace `ruvnet/wifi-densepose-pretrained` files
+/// (`model.safetensors`, `model-q{2,4,8}.bin`, `model.rvf.jsonl`) are a
+/// different *format* — and a different encoder architecture — than the RVF
+/// binary container the `--model` progressive loader expects (`RVFS` magic
+/// `0x52564653`). Feeding one to `--model` produced a bare
+/// "invalid magic at offset 0 …" that left users stuck. Detect the common
+/// cases and explain plainly what's loadable instead.
+fn diagnose_model_load_error(path: &std::path::Path, data: &[u8], err: &str) -> String {
+    let name = path
+        .file_name()
+        .and_then(|n| n.to_str())
+        .unwrap_or("")
+        .to_ascii_lowercase();
+    let ext = path
+        .extension()
+        .and_then(|e| e.to_str())
+        .unwrap_or("")
+        .to_ascii_lowercase();
+
+    // safetensors: 8-byte LE header length, then a JSON object starting with '{'.
+    let looks_safetensors = ext == "safetensors" || (data.len() > 9 && data[8] == b'{');
+    // JSONL manifest: starts with '{' (or the well-known suffix).
+    let looks_jsonl =
+        ext == "jsonl" || name.ends_with(".rvf.jsonl") || data.first() == Some(&b'{');
+    // Quantized weight blob shipped on HF (model-q2/q4/q8.bin).
+    let looks_quant_bin = ext == "bin" || name.contains("-q");
+
+    let kind = if looks_safetensors {
+        "a safetensors weight file"
+    } else if looks_jsonl {
+        "a JSONL manifest, not the binary container"
+    } else if looks_quant_bin {
+        "a quantized weight blob (e.g. HuggingFace model-q4.bin)"
+    } else {
+        "not an RVF binary container"
+    };
+
+    format!(
+        "model `{}` could not be loaded: it is {kind}. The --model flag expects an \
+         RVF binary container (`RVFS` magic 0x52564653) produced by the \
+         wifi-densepose-train pipeline. The HuggingFace ruvnet/wifi-densepose-pretrained \
+         files are a different format and encoder architecture, so they do not load \
+         here directly (issue #894). Continuing with signal heuristics. (loader: {err})",
+        path.display()
+    )
+}
+
 // ── Main ─────────────────────────────────────────────────────────────────────
 
 /// If `--ui-path` points nowhere (wrong cwd), try common repo layouts relative to cwd.
@@ -6113,7 +6256,9 @@ async fn main() {
                         model_loaded = true;
                         progressive_loader = Some(loader);
                     }
-                    Err(e) => error!("Progressive loader init failed: {e}"),
+                    Err(e) => {
+                        error!("{}", diagnose_model_load_error(mp, &data, &e.to_string()))
+                    }
                 },
                 Err(e) => error!("Failed to read model file: {e}"),
             }
@@ -6200,56 +6345,13 @@ async fn main() {
                             let Ok(v) = serde_json::from_str::<serde_json::Value>(&json) else {
                                 continue;
                             };
-                            let cls = &v["classification"];
-                            let vit = &v["vital_signs"];
-                            let presence = cls["presence"].as_bool().unwrap_or(false);
-                            let n_persons = v["persons"]
-                                .as_array()
-                                .map(|a| a.len() as u32)
-                                .or_else(|| v["estimated_persons"].as_u64().map(|x| x as u32))
-                                .unwrap_or(0);
-                            let motion = match cls["motion_level"].as_str() {
-                                Some("none") | Some("still") | Some("idle") | Some("") => 0.0,
-                                Some(_) => 1.0,
-                                None => 0.0,
-                            };
-                            let ts = (v["timestamp"].as_f64().unwrap_or(0.0) * 1000.0) as i64;
-                            let conf = cls["confidence"].as_f64().unwrap_or(0.0);
-                            let presence_score = if presence { conf.max(0.0) } else { 0.0 };
-                            let breathing = vit["breathing_rate_bpm"].as_f64();
-                            let hr = vit["heart_rate_bpm"].as_f64();
-                            // #898: emit one snapshot per physical node so each
-                            // surfaces as its own Home-Assistant device (with
-                            // its own RSSI + availability). Falls back to a
-                            // single aggregate snapshot when there is no
-                            // per-node data (e.g. wifi / simulate sources).
-                            let mk = |nid: String, rssi: Option<f64>| mqtt::state::VitalsSnapshot {
-                                node_id: nid,
-                                timestamp_ms: ts,
-                                presence,
-                                motion,
-                                presence_score,
-                                breathing_rate_bpm: breathing,
-                                heartrate_bpm: hr,
-                                n_persons,
-                                rssi_dbm: rssi,
-                                vital_confidence: conf,
-                                ..Default::default()
-                            };
-                            match v["nodes"].as_array() {
-                                Some(arr) if !arr.is_empty() => {
-                                    for node in arr {
-                                        let n = node["node_id"].as_u64().unwrap_or(0);
-                                        let nid = format!("{node_id}-node{n}");
-                                        let _ = vtx.send(mk(nid, node["rssi_dbm"].as_f64()));
-                                    }
-                                }
-                                _ => {
-                                    let _ = vtx.send(mk(
-                                        node_id.clone(),
-                                        v["nodes"][0]["rssi_dbm"].as_f64(),
-                                    ));
-                                }
+                            // #898/#872: emit one snapshot per physical node so
+                            // each surfaces as its own Home-Assistant device with
+                            // its *own* presence/motion/RSSI (see
+                            // vitals_snapshots_from_sensing_json). Falls back to a
+                            // single aggregate snapshot for per-node-less sources.
+                            for snap in vitals_snapshots_from_sensing_json(&v, &node_id) {
+                                let _ = vtx.send(snap);
                             }
                         }
                     });
@@ -7068,3 +7170,143 @@ mod rolling_p95_tests {
         assert_eq!(p.len(), 1);
     }
 }
+
+#[cfg(all(test, feature = "mqtt"))]
+mod mqtt_bridge_tests {
+    use super::vitals_snapshots_from_sensing_json;
+    use serde_json::json;
+
+    /// Regression for the per-node presence bug (#872/#898): each node must
+    /// surface its OWN classification, not the room-level aggregate. Node 1 is
+    /// present+moving; node 2 is absent — node 2 must NOT inherit node 1's
+    /// "present".
+    #[test]
+    fn per_node_presence_uses_each_nodes_own_classification() {
+        let v = json!({
+            "timestamp": 1.0,
+            "classification": { "presence": true, "motion_level": "walking", "confidence": 0.9 },
+            "vital_signs": { "breathing_rate_bpm": 14.0, "heart_rate_bpm": 60.0 },
+            "persons": [{}, {}],
+            "nodes": [
+                { "node_id": 1, "rssi_dbm": -40.0,
+                  "classification": { "presence": true, "motion_level": "walking", "confidence": 0.8 } },
+                { "node_id": 2, "rssi_dbm": -70.0,
+                  "classification": { "presence": false, "motion_level": "absent", "confidence": 0.1 } }
+            ]
+        });
+        let snaps = vitals_snapshots_from_sensing_json(&v, "ruview");
+        assert_eq!(snaps.len(), 2, "one snapshot per node");
+
+        let n1 = snaps.iter().find(|s| s.node_id == "ruview-node1").unwrap();
+        let n2 = snaps.iter().find(|s| s.node_id == "ruview-node2").unwrap();
+
+        assert!(n1.presence && n1.motion > 0.0, "node1 present + moving");
+        assert!(
+            !n2.presence && n2.motion == 0.0,
+            "node2 must be absent — not inherit the room aggregate"
+        );
+        // Per-node RSSI preserved.
+        assert_eq!(n1.rssi_dbm, Some(-40.0));
+        assert_eq!(n2.rssi_dbm, Some(-70.0));
+        // Vitals + person count are room-level, shared across node devices.
+        assert_eq!(n1.n_persons, 2);
+        assert_eq!(n2.n_persons, 2);
+        assert_eq!(n1.breathing_rate_bpm, Some(14.0));
+        assert_eq!(n2.heartrate_bpm, Some(60.0));
+        // presence_score is gated on presence.
+        assert!(n1.presence_score > 0.0);
+        assert_eq!(n2.presence_score, 0.0);
+    }
+
+    /// A node that omits a classification field defers to the room aggregate
+    /// rather than silently reading false/0.
+    #[test]
+    fn per_node_missing_fields_fall_back_to_aggregate() {
+        let v = json!({
+            "timestamp": 1.0,
+            "classification": { "presence": true, "motion_level": "still", "confidence": 0.7 },
+            "vital_signs": {},
+            "nodes": [ { "node_id": 3, "rssi_dbm": -55.0 } ]  // no per-node classification
+        });
+        let snaps = vitals_snapshots_from_sensing_json(&v, "n");
+        assert_eq!(snaps.len(), 1);
+        assert_eq!(snaps[0].node_id, "n-node3");
+        assert!(snaps[0].presence, "defers to aggregate presence");
+        assert_eq!(snaps[0].motion, 0.0, "aggregate 'still' => no motion");
+    }
+
+    /// No `nodes` array (wifi / simulate sources): single aggregate snapshot
+    /// keyed by the base id.
+    #[test]
+    fn falls_back_to_single_aggregate_when_no_nodes() {
+        let v = json!({
+            "timestamp": 2.0,
+            "classification": { "presence": true, "motion_level": "idle", "confidence": 0.6 },
+            "vital_signs": { "breathing_rate_bpm": 12.0 },
+            "persons": [{}]
+        });
+        let snaps = vitals_snapshots_from_sensing_json(&v, "ruview");
+        assert_eq!(snaps.len(), 1);
+        assert_eq!(snaps[0].node_id, "ruview");
+        assert!(snaps[0].presence);
+        assert_eq!(snaps[0].motion, 0.0, "idle => no motion");
+        assert_eq!(snaps[0].n_persons, 1);
+    }
+
+    /// `motion_level: "absent"` must map to zero motion (the old aggregate
+    /// match fell through to `Some(_) => 1.0`, treating absent as full motion).
+    #[test]
+    fn absent_motion_level_is_zero_motion() {
+        let v = json!({
+            "timestamp": 0.0,
+            "classification": { "presence": false, "motion_level": "absent", "confidence": 0.0 },
+            "vital_signs": {}
+        });
+        let snaps = vitals_snapshots_from_sensing_json(&v, "x");
+        assert_eq!(snaps[0].motion, 0.0);
+        assert!(!snaps[0].presence);
+    }
+}
+
+#[cfg(test)]
+mod model_load_diagnostic_tests {
+    use super::diagnose_model_load_error;
+    use std::path::Path;
+
+    #[test]
+    fn safetensors_is_named_and_points_at_894() {
+        // 8-byte LE header length then '{' — the safetensors signature.
+        let data = [0x10, 0, 0, 0, 0, 0, 0, 0, b'{', b'"'];
+        let msg = diagnose_model_load_error(
+            Path::new("models/wifi-densepose-pretrained/model.safetensors"),
+            &data,
+            "invalid magic at offset 0",
+        );
+        assert!(msg.contains("safetensors"), "{msg}");
+        assert!(msg.contains("#894"), "{msg}");
+        assert!(msg.contains("signal heuristics"), "{msg}");
+    }
+
+    #[test]
+    fn quantized_bin_is_identified() {
+        let data = [0x35, 0x57, 0x45, 0x77]; // the 0x77455735 the loader reports
+        let msg = diagnose_model_load_error(Path::new("model-q4.bin"), &data, "bad magic");
+        assert!(msg.contains("quantized weight blob"), "{msg}");
+        assert!(msg.contains("RVFS") || msg.contains("0x52564653"), "{msg}");
+    }
+
+    #[test]
+    fn jsonl_manifest_is_identified() {
+        let data = *b"{\"seg\":0}";
+        let msg = diagnose_model_load_error(Path::new("model.rvf.jsonl"), &data, "x");
+        assert!(msg.contains("JSONL manifest"), "{msg}");
+    }
+
+    #[test]
+    fn unknown_format_still_gives_guidance() {
+        let data = [0u8, 1, 2, 3];
+        let msg = diagnose_model_load_error(Path::new("weird.dat"), &data, "x");
+        assert!(msg.contains("RVF binary container"), "{msg}");
+        assert!(msg.contains("wifi-densepose-train"), "{msg}");
+    }
+}