armiger / docs /plugins /calibration-dht.md

Rename zero_cue_2000 and multilang_3000 evaluation split labels

b9f779e verified 2 days ago

12.2 kB

<h1 align="center">Armiger Plugin #5 — Conformal Calibration &amp; BEP-44 Decentralized DHT Updater (<code>armiger-calibration-dht</code>)</h1>

<p align="center">
  <strong>Per-head temperature scaling, Menon prior adjustments, Mondrian conformal safety gates (<code>100.00%</code> junk accuracy on <code>oct15_confirmation</code> <code>N=2,170</code> and <code>99.90%</code> on <code>beyond_confirmation</code> <code>N=977</code>), and a Hybrid Post-Quantum <code>Ed25519</code> + <code>WOTS+-SHAKE256</code> signed <code>BEP-44</code> / <code>BEP-46</code> BitTorrent DHT update &amp; RSI engine.</strong><br>
  Available standalone here and pre-bundled inside <a href="https://huggingface.co/NorviTech/armiger-ultra"><code>NorviTech/armiger-ultra</code></a>.
</p>

<p align="center">
  <a href="https://huggingface.co/NorviTech/armiger-ultra"><img alt="Bundled in Armiger Ultra" src="https://img.shields.io/badge/Bundled_in-Armiger_Ultra-FD8024.svg"></a>
  <a href="https://huggingface.co/NorviTech/armiger"><img alt="Compatible with Armiger" src="https://img.shields.io/badge/Compatible_with-Armiger_CPU-0284C7.svg"></a>
  <img alt="Licence Apache-2.0" src="https://img.shields.io/badge/licence-Apache--2.0-blue.svg">
</p>

## 1. Per-Head Calibration & Mondrian Conformal Safety Gates (`calibration.json`)

Raw neural network logits are often overconfident on rare classes (`course`, `comic`, `software`) and high-stakes safety boundaries (`adult`, `junk`). `calibration.json` + `ArmigerCalibrator` apply:
- **Per-Head Temperature Scaling ($T_h$)** and **Menon Long-Tail Logit Adjustments ($\tau_h \log \pi_h$)** across all 8 classification heads.
- **Mondrian Conformal Safety Thresholds (Unmutated Ground-Truth Keys):**
  - `adult`: `0.5451` (**`98.53%`** on `oct15_confirmation` `N=2,781`, **`98.62%`** on `beyond_confirmation` `N=1,016`, **`99.90%`** on `zero_cue_2000` `N=2,000`, **`100.00%`** on `multilang_3000` `N=3,000`)
  - `junk`: `0.1010` (**`100.00%`** on `oct15_confirmation` `2,170/2,170`, **`100.00%`** on `fresh_confirmation` `1,291/1,291`, **`99.90%`** on `beyond_confirmation` `976/977`, and **`7/7` (`100%`)** disguised executable malware traps blocked)
  - `owner_scope`: `0.4770` (**`98.65%`** on `oct15_confirmation` `2,633/2,669`, **`99.80%`** on `zero_cue_2000`, **`99.90%`** on `multilang_3000`)

## 2. Decentralized `BEP-44` + `Ed25519` DHT Update Engine (`ArmigerDHTUpdater`)

When `Armiger` or `Armiger Ultra` runs inside a distributed indexer (`Bitmagnet` and self-hosted media indexers), `ArmigerDHTUpdater` enables **opt-in decentralized hot-reloading** of new `zero-cue` catalog entries, `swarm-priors` tables, and rank-8 LoRA micro-adapters directly over the mainline BitTorrent DHT:
- **`BEP-44` Mutable DHT Pointers + `BEP-46` Infohash Distribution:** Nodes look up `SHA1(pubkey || "armiger-ultra-v2")` on the DHT to discover the latest signed bundle infohash.
- **Cryptographic Anti-Spoofing (`Ed25519` + Monotonic `seq`):** Every manifest must carry a valid `Ed25519` signature from the pinned offline root public key and a strictly increasing sequence number (`seq > current_seq`), preventing spoofed updates and downgrade attacks.
- **$K$-of-$N$ Quorum & Safe Formats Only:** Bundles contain only `JSON`, `ONNX`, or `SafeTensors` tensors (never arbitrary Python `pickle` code) and require a configurable $K$-of-$N$ attestation quorum before activation.

## 3. Distributed Recursive Self-Improvement (RSI) & Data Engine (`armiger_dht_rsi.py`)

For multi-agent indexers, private media stacks, and community clusters, `armiger_dht_rsi.py` provides a production-grade, zero-trust distributed training & calibration telemetry pipeline:
1. **Universal Adapters (Client-Agnostic across Media Stacks):** Native ingestion for series, movie, audio, and publication library webhooks, ffprobe / MediaInfo physical stream inspectors, and qBittorrent / Transmission completion events.
2. **Multi-Stage Physical Ground-Truth Verification:** Rejects raw unverified model predictions. Observations require 100% download verification, physical stream decodability, and canonical catalog matches (`series_catalog`, `movie_catalog`, `audio_catalog`, `isbn`). Active malware masquerades (`.exe` or `.scr` disguised as media) are physically isolated as junk canaries.
3. **Cryptographic Anonymization & Differential Privacy:** Strips all private tracker passkeys, IPv4/IPv6 addresses, and user home directory paths. Generates salted HMAC-blinded syntactic title stems. Adds $(\epsilon, \delta)$-differential privacy (Laplace perturbation + 16MB log-binning) and Poisson timing jitter ($\Delta t \sim \text{Exp}(\lambda)$) against traffic analysis.
4. **Post-Quantum End-to-End Encryption:** Hybrid `X25519` ECDH + NIST FIPS 203 `ML-KEM-768` lattice key encapsulation (with Fujisaki-Okamoto CCA2 implicit rejection), combined via `SHAKE-256` / `HKDF-SHA3-256` with 256-bit `ChaCha20-Poly1305` AEAD payload encryption.
5. **Hybrid Post-Quantum Anti-Rollback Protocol:** BEP-44 strictly monotonic sequence numbers (`seq > current_seq`) paired with Hybrid `Ed25519` + `WOTS+-SHAKE256` (RFC 8391 / NIST FIPS 205 hash-based PQ) threshold multi-signatures ($M$-of-$N$ federation quorum).
6. **Sybil-Resistant ASN Quorum & Canary Gate:** Enforces $K$-of-$N$ distinct-ASN peer attestation quorum, strict format whitelisting (`SafeTensors`, `ONNX`, `JSON` only; zero pickle tolerance), and a mandatory local 500-item golden canary regression gate (rejects any update with malware bypass or Mondrian safety degradation).
7. **Opt-In / Opt-Out Modes:** Supports `off`, `pull_only`, and `anonymous_contribute`.

## Quickstart

```python
from calibration_and_dht import ArmigerCalibrator
from armiger_dht_rsi import ArmigerDHTRSIEngine, SeriesWebhookAdapter
# 1. Conformal calibration
cal = ArmigerCalibrator("calibration.json")
probs_cat = cal.calibrate_head("category", [2.4, -1.1, -3.0, -2.5, -3.1, -2.8, -3.2, -2.9, -3.0, -3.4, -2.0])
gates = cal.evaluate_safety_gates(p_adult=0.02, p_junk=0.004, p_excluded_scope=0.01)
print("Calibrated top prob:", max(probs_cat), "Safety gates:", gates)
# 2. Distributed RSI Engine
engine = ArmigerDHTRSIEngine(mode="anonymous_contribute")
obs = SeriesWebhookAdapter.parse_webhook({"eventType": "Download", "series": {"seriesCatalogId": 371980}})
ok, reason, record = engine.ingest_candidate(obs)
print("Ingested observation:", ok, reason)
```

## Part of the Armiger Modular Plugin Suite

| Plugin Repository | Bundled in `Armiger Ultra`? | Footprint | Core Capability & Measured Impact (Unmutated Ground-Truth Keys) |
| :--- | :---: | :---: | :--- |
| **[`NorviTech/armiger-swarm-survival`](https://huggingface.co/NorviTech/armiger-swarm-survival)** | **Yes** (`swarm_adapter.pt`, `swarm_survival.py`) | `167 KB` (`132 KB` `.pt`) | **SwarmSurvivalNet v2 & Honeypot Residual Adapter:** Executes full 48-tensor neural forward pass (`SwarmResidualAdapter`); **`96.83%`** Multi-Source Corpus AUROC (`N=10,278`, vs. `53.02%` raw seeders), **`92.42%`** / **`93.03%`** Public-Tracker AUROC (`N=757` / `N=493`, vs. `85.17%` raw seeders), **`100.0%`** fake-seeder malware trap recall, **`0.00%`** content-category flip guarantee. |
| **[`NorviTech/armiger-swarm-priors`](https://huggingface.co/NorviTech/armiger-swarm-priors)** | **Yes** (`swarm_priors.py`) | `29.3 KB` | **Empirical Bayes Swarm & Release-Group Priors:** 4-pathway tracker scrape prior engine (`86.64%` formula AUROC), mandatory modality dropout (`p=0.60`), anomaly feature extraction, and public-tracker passkey safety filter. |
| **[`NorviTech/armiger-zero-cue-catalog`](https://huggingface.co/NorviTech/armiger-zero-cue-catalog)** | **Yes** (`zero_cue_plugin.py`, `validated_zero_cue_catalog.json`) | `26.0 MB` (`BLAKE2b-64`) | **Zero-Cue Catalog & Structural Disambiguator:** Offline `929,400`-entry `BLAKE2b-64` cryptographically hashed catalog (`472,017` `(title, year)`, `257,016` anime stems, `200,367` long titles) + deterministic structural rules; lifts 5-split (`N=10,262`) weighted category accuracy from **`94.00%`** (Standalone Neural) to **`95.42%`** (Fair Blind Hybrid), `beyond_confirmation` (`N=1,025`) from **`92.88%`** to **`95.12%`**, and `zero_cue_2000` (`N=2,000`) from **`87.40%`** to **`91.50%`** (`+4.10 pp`). |
| **[`NorviTech/armiger-audio-guard`](https://huggingface.co/NorviTech/armiger-audio-guard)** | **Yes** (`audio_language_guard.py`) | `3.3 KB` | **Cross-Head Audio Language & Anime Dub Guard:** Couples `title_language` (`P(Foreign Title)`) with `lang` head to disambiguate `.en.` prepositions, `<Foreign>.ENG.Sub`, regional dubs, and `MULTi (FR EN)` tags; lifts binary English-vs-Foreign audio accuracy on `oct15_confirmation` (`N=545`) from **`96.88%`** (Standalone Neural) to **`97.80%`** (`92.75%` 3-class `N=772`). |
| **[`NorviTech/armiger-calibration-dht`](https://huggingface.co/NorviTech/armiger-calibration-dht)** | **Yes** (`calibration.json`, `calibration_and_dht.py`, `armiger_dht_rsi.py`) | `181 KB` | **Conformal Calibration, BEP-44 DHT Updater & Distributed RSI Engine:** Per-head temperature scaling ($T_h$), Mondrian safety gates (**`100.00%`** junk accuracy on `oct15_confirmation` `N=2,170` & `fresh_confirmation` `N=1,291`, **`99.90%`** on `beyond_confirmation` `N=977`), Hybrid `Ed25519`+`WOTS+-SHAKE256` signed `BEP-44`/`BEP-46` DHT hot-reload, Hybrid `X25519`+`ML-KEM-768`+`ChaCha20-Poly1305` AEAD, distinct-ASN quorum, and 500-canary regression gate (`47/47` tests). |
| **[`NorviTech/armiger-dht-db-sync`](https://huggingface.co/NorviTech/armiger-dht-db-sync)** | **Yes** (`csam_guard.py`, `armiger_dht_db_sync.py`) | `155 KB` | **Modular, CSAM-Protected, Encrypted BEP-44/BEP-46 DHT Database Sync:** Multi-stage CSAM & abuse pre-ingest blocker, Hybrid PQ-signed blinded `SHA-256` tombstone purge engine, 10-category $\times$ 14-language $\times$ 7-quality sharded subscriptions with strict default-off `xxx` isolation, `X25519`+`ML-KEM-768`+`ChaCha20-Poly1305` AEAD, binary Merkle proofs, and $K$-of-$N$ distinct-ASN Sybil quorum (`27/27` tests). |

---

## Intended Use & Legal Compliance

`Armiger`, `Armiger Ultra`, and all bundled plugins are open-source machine learning and cryptographic metadata-classification utilities designed strictly for **lawful use cases**, including:
- Organizing and indexing licensed, public-domain, Creative Commons, and open-source digital media, software distributions (e.g., Linux ISOs, open-source games), and educational archives (e.g., OpenCourseWare, LibriVox, Project Gutenberg).
- Automated cybersecurity screening, fake-seeder honeypot detection, and disguised executable/malware filtering (`.mkv.exe`, `.scr`, `.vbs`) in peer-to-peer and decentralized storage research.
- Child safety (CSAM/CSAE) pre-ingest blocking, cryptographic tombstone revocation, and adult/NSFW namespace isolation for self-hosted and decentralized indexers.

**NorviTech does not condone, encourage, or support copyright infringement or the unauthorized distribution of copyrighted works.** Users are solely responsible for ensuring that their use of these models and plugins complies with all applicable local, national, and international copyright and data-protection laws.

---

<p align="center">
  <a href="https://norvitech.com"><img alt="Part of the NorviTech Suite — open-source apps for the Linux desktop and the self-hosted stack" src="https://norvitech.com/assets/banner.svg" width="640"></a>
</p>

<p align="center">
  <a href="https://huggingface.co/NorviTech/armiger-ultra">Armiger Ultra</a> ·
  <a href="https://huggingface.co/NorviTech/armiger">Armiger</a> ·
  <a href="https://huggingface.co/NorviTech/armiger-swarm-survival">Swarm Survival</a> ·
  <a href="https://huggingface.co/NorviTech/armiger-swarm-priors">Swarm Priors</a> ·
  <a href="https://huggingface.co/NorviTech/armiger-zero-cue-catalog">Zero-Cue Catalog</a> ·
  <a href="https://huggingface.co/NorviTech/armiger-audio-guard">Audio Guard</a> ·
  <a href="https://huggingface.co/NorviTech/armiger-calibration-dht">Calibration &amp; DHT RSI</a> ·
  <a href="https://huggingface.co/NorviTech/armiger-dht-db-sync">DHT DB Sync</a> ·
  <a href="https://norvitech.com">norvitech.com</a>
</p>

	<h1 align="center">Armiger Plugin #5 — Conformal Calibration & BEP-44 Decentralized DHT Updater (<code>armiger-calibration-dht</code>)</h1>

	<p align="center">
	<strong>Per-head temperature scaling, Menon prior adjustments, Mondrian conformal safety gates (<code>100.00%</code> junk accuracy on <code>oct15_confirmation</code> <code>N=2,170</code> and <code>99.90%</code> on <code>beyond_confirmation</code> <code>N=977</code>), and a Hybrid Post-Quantum <code>Ed25519</code> + <code>WOTS+-SHAKE256</code> signed <code>BEP-44</code> / <code>BEP-46</code> BitTorrent DHT update & RSI engine.</strong><br>
	Available standalone here and pre-bundled inside <a href="https://huggingface.co/NorviTech/armiger-ultra"><code>NorviTech/armiger-ultra</code></a>.
	</p>

	<p align="center">
	<a href="https://huggingface.co/NorviTech/armiger-ultra"><img alt="Bundled in Armiger Ultra" src="https://img.shields.io/badge/Bundled_in-Armiger_Ultra-FD8024.svg"></a>
	<a href="https://huggingface.co/NorviTech/armiger"><img alt="Compatible with Armiger" src="https://img.shields.io/badge/Compatible_with-Armiger_CPU-0284C7.svg"></a>
	<img alt="Licence Apache-2.0" src="https://img.shields.io/badge/licence-Apache--2.0-blue.svg">
	</p>

	## 1. Per-Head Calibration & Mondrian Conformal Safety Gates (`calibration.json`)

	Raw neural network logits are often overconfident on rare classes (`course`, `comic`, `software`) and high-stakes safety boundaries (`adult`, `junk`). `calibration.json` + `ArmigerCalibrator` apply:
	- Per-Head Temperature Scaling ($T_h$) and Menon Long-Tail Logit Adjustments ($\tau_h \log \pi_h$) across all 8 classification heads.
	- Mondrian Conformal Safety Thresholds (Unmutated Ground-Truth Keys):
	- `adult`: `0.5451` (`98.53%` on `oct15_confirmation` `N=2,781`, `98.62%` on `beyond_confirmation` `N=1,016`, `99.90%` on `zero_cue_2000` `N=2,000`, `100.00%` on `multilang_3000` `N=3,000`)
	- `junk`: `0.1010` (`100.00%` on `oct15_confirmation` `2,170/2,170`, `100.00%` on `fresh_confirmation` `1,291/1,291`, `99.90%` on `beyond_confirmation` `976/977`, and `7/7` (`100%`) disguised executable malware traps blocked)
	- `owner_scope`: `0.4770` (`98.65%` on `oct15_confirmation` `2,633/2,669`, `99.80%` on `zero_cue_2000`, `99.90%` on `multilang_3000`)

	## 2. Decentralized `BEP-44` + `Ed25519` DHT Update Engine (`ArmigerDHTUpdater`)

	When `Armiger` or `Armiger Ultra` runs inside a distributed indexer (`Bitmagnet` and self-hosted media indexers), `ArmigerDHTUpdater` enables opt-in decentralized hot-reloading of new `zero-cue` catalog entries, `swarm-priors` tables, and rank-8 LoRA micro-adapters directly over the mainline BitTorrent DHT:
	- `BEP-44` Mutable DHT Pointers + `BEP-46` Infohash Distribution: Nodes look up `SHA1(pubkey \|\| "armiger-ultra-v2")` on the DHT to discover the latest signed bundle infohash.
	- Cryptographic Anti-Spoofing (`Ed25519` + Monotonic `seq`): Every manifest must carry a valid `Ed25519` signature from the pinned offline root public key and a strictly increasing sequence number (`seq > current_seq`), preventing spoofed updates and downgrade attacks.
	- $K$-of-$N$ Quorum & Safe Formats Only: Bundles contain only `JSON`, `ONNX`, or `SafeTensors` tensors (never arbitrary Python `pickle` code) and require a configurable $K$-of-$N$ attestation quorum before activation.

	## 3. Distributed Recursive Self-Improvement (RSI) & Data Engine (`armiger_dht_rsi.py`)

	For multi-agent indexers, private media stacks, and community clusters, `armiger_dht_rsi.py` provides a production-grade, zero-trust distributed training & calibration telemetry pipeline:
	1. Universal Adapters (Client-Agnostic across Media Stacks): Native ingestion for series, movie, audio, and publication library webhooks, ffprobe / MediaInfo physical stream inspectors, and qBittorrent / Transmission completion events.
	2. Multi-Stage Physical Ground-Truth Verification: Rejects raw unverified model predictions. Observations require 100% download verification, physical stream decodability, and canonical catalog matches (`series_catalog`, `movie_catalog`, `audio_catalog`, `isbn`). Active malware masquerades (`.exe` or `.scr` disguised as media) are physically isolated as junk canaries.
	3. Cryptographic Anonymization & Differential Privacy: Strips all private tracker passkeys, IPv4/IPv6 addresses, and user home directory paths. Generates salted HMAC-blinded syntactic title stems. Adds $(\epsilon, \delta)$-differential privacy (Laplace perturbation + 16MB log-binning) and Poisson timing jitter ($\Delta t \sim \text{Exp}(\lambda)$) against traffic analysis.
	4. Post-Quantum End-to-End Encryption: Hybrid `X25519` ECDH + NIST FIPS 203 `ML-KEM-768` lattice key encapsulation (with Fujisaki-Okamoto CCA2 implicit rejection), combined via `SHAKE-256` / `HKDF-SHA3-256` with 256-bit `ChaCha20-Poly1305` AEAD payload encryption.
	5. Hybrid Post-Quantum Anti-Rollback Protocol: BEP-44 strictly monotonic sequence numbers (`seq > current_seq`) paired with Hybrid `Ed25519` + `WOTS+-SHAKE256` (RFC 8391 / NIST FIPS 205 hash-based PQ) threshold multi-signatures ($M$-of-$N$ federation quorum).
	6. Sybil-Resistant ASN Quorum & Canary Gate: Enforces $K$-of-$N$ distinct-ASN peer attestation quorum, strict format whitelisting (`SafeTensors`, `ONNX`, `JSON` only; zero pickle tolerance), and a mandatory local 500-item golden canary regression gate (rejects any update with malware bypass or Mondrian safety degradation).
	7. Opt-In / Opt-Out Modes: Supports `off`, `pull_only`, and `anonymous_contribute`.

	## Quickstart

	```python
	from calibration_and_dht import ArmigerCalibrator
	from armiger_dht_rsi import ArmigerDHTRSIEngine, SeriesWebhookAdapter

	# 1. Conformal calibration
	cal = ArmigerCalibrator("calibration.json")
	probs_cat = cal.calibrate_head("category", [2.4, -1.1, -3.0, -2.5, -3.1, -2.8, -3.2, -2.9, -3.0, -3.4, -2.0])
	gates = cal.evaluate_safety_gates(p_adult=0.02, p_junk=0.004, p_excluded_scope=0.01)
	print("Calibrated top prob:", max(probs_cat), "Safety gates:", gates)

	# 2. Distributed RSI Engine
	engine = ArmigerDHTRSIEngine(mode="anonymous_contribute")
	obs = SeriesWebhookAdapter.parse_webhook({"eventType": "Download", "series": {"seriesCatalogId": 371980}})
	ok, reason, record = engine.ingest_candidate(obs)
	print("Ingested observation:", ok, reason)
	```

	## Part of the Armiger Modular Plugin Suite

	\| Plugin Repository \| Bundled in `Armiger Ultra`? \| Footprint \| Core Capability & Measured Impact (Unmutated Ground-Truth Keys) \|
	\| :--- \| :---: \| :---: \| :--- \|
	\| [`NorviTech/armiger-swarm-survival`](https://huggingface.co/NorviTech/armiger-swarm-survival) \| Yes (`swarm_adapter.pt`, `swarm_survival.py`) \| `167 KB` (`132 KB` `.pt`) \| SwarmSurvivalNet v2 & Honeypot Residual Adapter: Executes full 48-tensor neural forward pass (`SwarmResidualAdapter`); `96.83%` Multi-Source Corpus AUROC (`N=10,278`, vs. `53.02%` raw seeders), `92.42%` / `93.03%` Public-Tracker AUROC (`N=757` / `N=493`, vs. `85.17%` raw seeders), `100.0%` fake-seeder malware trap recall, `0.00%` content-category flip guarantee. \|
	\| [`NorviTech/armiger-swarm-priors`](https://huggingface.co/NorviTech/armiger-swarm-priors) \| Yes (`swarm_priors.py`) \| `29.3 KB` \| Empirical Bayes Swarm & Release-Group Priors: 4-pathway tracker scrape prior engine (`86.64%` formula AUROC), mandatory modality dropout (`p=0.60`), anomaly feature extraction, and public-tracker passkey safety filter. \|
	\| [`NorviTech/armiger-zero-cue-catalog`](https://huggingface.co/NorviTech/armiger-zero-cue-catalog) \| Yes (`zero_cue_plugin.py`, `validated_zero_cue_catalog.json`) \| `26.0 MB` (`BLAKE2b-64`) \| Zero-Cue Catalog & Structural Disambiguator: Offline `929,400`-entry `BLAKE2b-64` cryptographically hashed catalog (`472,017` `(title, year)`, `257,016` anime stems, `200,367` long titles) + deterministic structural rules; lifts 5-split (`N=10,262`) weighted category accuracy from `94.00%` (Standalone Neural) to `95.42%` (Fair Blind Hybrid), `beyond_confirmation` (`N=1,025`) from `92.88%` to `95.12%`, and `zero_cue_2000` (`N=2,000`) from `87.40%` to `91.50%` (`+4.10 pp`). \|
	\| [`NorviTech/armiger-audio-guard`](https://huggingface.co/NorviTech/armiger-audio-guard) \| Yes (`audio_language_guard.py`) \| `3.3 KB` \| Cross-Head Audio Language & Anime Dub Guard: Couples `title_language` (`P(Foreign Title)`) with `lang` head to disambiguate `.en.` prepositions, `<Foreign>.ENG.Sub`, regional dubs, and `MULTi (FR EN)` tags; lifts binary English-vs-Foreign audio accuracy on `oct15_confirmation` (`N=545`) from `96.88%` (Standalone Neural) to `97.80%` (`92.75%` 3-class `N=772`). \|
	\| [`NorviTech/armiger-calibration-dht`](https://huggingface.co/NorviTech/armiger-calibration-dht) \| Yes (`calibration.json`, `calibration_and_dht.py`, `armiger_dht_rsi.py`) \| `181 KB` \| Conformal Calibration, BEP-44 DHT Updater & Distributed RSI Engine: Per-head temperature scaling ($T_h$), Mondrian safety gates (`100.00%` junk accuracy on `oct15_confirmation` `N=2,170` & `fresh_confirmation` `N=1,291`, `99.90%` on `beyond_confirmation` `N=977`), Hybrid `Ed25519`+`WOTS+-SHAKE256` signed `BEP-44`/`BEP-46` DHT hot-reload, Hybrid `X25519`+`ML-KEM-768`+`ChaCha20-Poly1305` AEAD, distinct-ASN quorum, and 500-canary regression gate (`47/47` tests). \|
	\| [`NorviTech/armiger-dht-db-sync`](https://huggingface.co/NorviTech/armiger-dht-db-sync) \| Yes (`csam_guard.py`, `armiger_dht_db_sync.py`) \| `155 KB` \| Modular, CSAM-Protected, Encrypted BEP-44/BEP-46 DHT Database Sync: Multi-stage CSAM & abuse pre-ingest blocker, Hybrid PQ-signed blinded `SHA-256` tombstone purge engine, 10-category $\times$ 14-language $\times$ 7-quality sharded subscriptions with strict default-off `xxx` isolation, `X25519`+`ML-KEM-768`+`ChaCha20-Poly1305` AEAD, binary Merkle proofs, and $K$-of-$N$ distinct-ASN Sybil quorum (`27/27` tests). \|

	---

	## Intended Use & Legal Compliance

	`Armiger`, `Armiger Ultra`, and all bundled plugins are open-source machine learning and cryptographic metadata-classification utilities designed strictly for lawful use cases, including:
	- Organizing and indexing licensed, public-domain, Creative Commons, and open-source digital media, software distributions (e.g., Linux ISOs, open-source games), and educational archives (e.g., OpenCourseWare, LibriVox, Project Gutenberg).
	- Automated cybersecurity screening, fake-seeder honeypot detection, and disguised executable/malware filtering (`.mkv.exe`, `.scr`, `.vbs`) in peer-to-peer and decentralized storage research.
	- Child safety (CSAM/CSAE) pre-ingest blocking, cryptographic tombstone revocation, and adult/NSFW namespace isolation for self-hosted and decentralized indexers.

	NorviTech does not condone, encourage, or support copyright infringement or the unauthorized distribution of copyrighted works. Users are solely responsible for ensuring that their use of these models and plugins complies with all applicable local, national, and international copyright and data-protection laws.

	---

	<p align="center">
	<a href="https://norvitech.com"><img alt="Part of the NorviTech Suite — open-source apps for the Linux desktop and the self-hosted stack" src="https://norvitech.com/assets/banner.svg" width="640"></a>
	</p>

	<p align="center">
	<a href="https://huggingface.co/NorviTech/armiger-ultra">Armiger Ultra</a> ·
	<a href="https://huggingface.co/NorviTech/armiger">Armiger</a> ·
	<a href="https://huggingface.co/NorviTech/armiger-swarm-survival">Swarm Survival</a> ·
	<a href="https://huggingface.co/NorviTech/armiger-swarm-priors">Swarm Priors</a> ·
	<a href="https://huggingface.co/NorviTech/armiger-zero-cue-catalog">Zero-Cue Catalog</a> ·
	<a href="https://huggingface.co/NorviTech/armiger-audio-guard">Audio Guard</a> ·
	<a href="https://huggingface.co/NorviTech/armiger-calibration-dht">Calibration & DHT RSI</a> ·
	<a href="https://huggingface.co/NorviTech/armiger-dht-db-sync">DHT DB Sync</a> ·
	<a href="https://norvitech.com">norvitech.com</a>
	</p>