NorviTech/armiger-ultra

Armiger Ultra

Unified torrent intelligence suite bundling Armiger (8.3 MB INT8 CPU), Armiger Ultra (32.3 MB FP32 GPU/CPU), the 48-tensor Swarm Health & Honeypot Adapter (swarm_adapter.pt), and all 6 official plugins.
Delivers 97.60% 6-dimension composite accuracy on unmutated ground-truth keys, 0.21 ms/item batched on NVIDIA RTX GPU (0.29 ms on Apple M4 Pro MPS GPU, 2.44 ms single-item on M4 Pro CPU INT8), 100% fake-seeder malware trap recall, and 96.83% warehouse / 92.42% public-tracker swarm liveliness AUROC.

Armiger Ultra (NorviTech/armiger-ultra) is the complete, all-in-one release of the Armiger neural classification and swarm-intelligence stack for Bitmagnet and self-hosted media library stacks.

One Download, Everything Included: You do not need to download NorviTech/armiger and Armiger Ultra separately. This repository bundles:

1. armiger_ultra.fp32.onnx (32.3 MB) & armiger_ultra.pt (32.2 MB): Full-precision GPU/CPU 8-head neural classifier from the hardened H350000_Curated_Wave2-seed0 checkpoint (0.21 ms/item batch-64 on NVIDIA RTX GPU, 0.29 ms/item batch-64 on Apple M4 Pro MPS GPU).
2. armiger.int8.onnx (8.3 MB): Dynamically quantized CPU fast-path engine from the identical checkpoint (2.44 ms single-item p50 / 3.07 ms/item batch-64 on Apple M4 Pro CPU ONNX Runtime).
3. armiger_inference.py + vocab.json: Self-contained, zero-tokenizer 11-input featurizer (title_bytes, ext_ids, file_sizes_log, file_depths, file_flags, file_mask, struct_feats, tracker_ids, cat_hint_id, tag_ids, swarm_feats) for direct ONNX Runtime inference.
4. swarm_adapter.pt (132 KB) & swarm_survival.py: 48-Tensor PyTorch Swarm Health, Download Viability & Fake-Seeder Honeypot Residual Adapter (SwarmResidualAdapter), fusing tracker/DHT seeder/leecher telemetry with content features to achieve 96.83% Multi-Source Warehouse 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 honeypot recall, and 0.00% category flip rate.
5. validated_zero_cue_catalog.json (26.0 MB) & zero_cue_plugin.py: Privacy-hardened BLAKE2b-64 (person=b"ArmigZCC") 929,400-entry offline title/year/anime catalog + deterministic structural rules.
6. calibration.json: Post-hoc temperature scaling, Menon prior logit adjustments, and Mondrian conformal safety thresholds (100.00% junk accuracy on oct15_confirmation N=2,170).

---

Modular 6-Plugin Architecture (All 6 Rolled Into Armiger Ultra + Available Standalone)

All 6 Armiger plugins are rolled directly into this NorviTech/armiger-ultra repository for one-command deployment, and are also published as standalone modular repositories under NorviTech for custom pipelines:

Plugin Repository	Bundled in Armiger Ultra?	Footprint	Core Capability & Measured Impact (Unmutated Ground-Truth Keys)
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 Warehouse 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	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	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 tc_truespec_2000 (N=2,000) from 87.40% to 91.50% (+4.10 pp).
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	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 (61/61 unit tests, 4/4 multi-instance transfer tests, 45/45 red-team exploit probes blocked).
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 (37/37 unit + 12-node localhost UDP Kademlia BEP-44/46 wire tests, 45/45 red-team exploit probes blocked).

---

Quickstart: Unified ArmigerUltraPipeline SDK (11 Heads + All 6 Plugins)

armiger_ultra_pipeline.py integrates the ONNX/PyTorch 8-head classifier (armiger_inference.py) with all 6 official plugins (SwarmResidualAdapter, SwarmPriors, ZeroCueCatalogPlugin, AudioLanguageGuard, CalibrationAndDHT + ArmigerDHTRSIEngine, and ArmigerDHTDBSync + CSAMGuard) behind a single API:

from armiger_ultra_pipeline import ArmigerUltraPipeline

# Load full-precision Armiger Ultra (or pass variant="int8" for the 8.3 MB CPU fast-path)
pipeline = ArmigerUltraPipeline.from_directory("./armiger-ultra")

pred = pipeline.classify({
    "infohash": "1111222233334444555566667777888899990000",
    "name": "Sintel.2010.2160p.UHD.BluRay.REMUX.HEVC.TrueHD.7.1.Atmos-OPEN",
    "size": 74_500_000_000,
    "files": [{"path": "Sintel.2010.2160p.mkv", "size": 74_500_000_000}],
    "swarm": {
        "tracker_known": True,
        "tracker_seeders": 320,
        "tracker_leechers": 18,
        "tracker_completed": 2450,
    },
})

print(pred["category"])                           # "movie" (0% flip guarantee across seeders=0..10000)
print(pred["mondrian_gates"]["passed_all"])       # True (CSAM guard + Mondrian conformal gates + honeypot guard)
print(pred["swarm_viability"]["viability_tier"])  # "alive_viable" (computed via 48-tensor swarm_adapter.pt)
print(list(pred["heads"].keys()))                 # All 11 unified heads

Run the bundled 94-test verification suite (test_armiger_ultra_pipeline.py, test_armiger_dht_rsi.py, test_dht_db_sync.py):

pytest test_armiger_ultra_pipeline.py test_armiger_dht_rsi.py test_dht_db_sync.py

---

Ecosystem Integrations (DH-Waypoint, Bitmagnet & *Arr Stack)

Armiger and Armiger Ultra ship with turnkey HTTP services, Docker Compose stacks, and reference adapters in examples/ and docs/INTEGRATION_GUIDE.md:

• DH-Waypoint Integration (examples/dh-waypoint/) — Drop-in processorapi/v1 service (armiger_waypoint_processor.py) and plugin manifest (07-armiger-ultra.plugin.yml) implementing batch classification (dh-waypoint.processor.v1), pre-ingest cohort admission (dh-waypoint.processors.admission.v1), and 4-tier dh-waypoint.processors.feedback.v1 closed-loop telemetry for spencernorton/DH-Waypoint.
• Bitmagnet Integration (examples/bitmagnet/) — Pre-ingest / post-crawl batch classifier and malware/honeypot pruner (bitmagnet_armiger_bridge.py) that upgrades upstream Bitmagnet from 35.17% rule accuracy to 95.49% 11-class neural accuracy and automatically prunes disguised .mkv.exe/.scr/.lnk malware traps via GraphQL torrent.delete.
• *Arr Stack & qBittorrent Integration (examples/arr-stack/) — Pre-grab Torznab XML filtering proxy (armiger_torznab_proxy.py) between Prowlarr / Bitmagnet / DH-Waypoint and Sonarr, Radarr, Lidarr, Readarr, and Whisparr, paired with a post-import webhook & ffprobe ground-truth receiver (armiger_arr_webhook_receiver.py) wired into SeriesWebhookAdapter, MovieWebhookAdapter, AudioWebhookAdapter, PublicationWebhookAdapter, QBittorrentAdapter, and PhysicalGroundTruthVerifier.

See the Complete Ecosystem Integration Guide (docs/INTEGRATION_GUIDE.md) for architecture diagrams, Docker Compose files, and step-by-step setup instructions.

---

Audited Accuracy Across All 5 Evaluation Splits (N=10,262 Unmutated Ground-Truth Keys)

Every metric below is reproducibly verifiable on the original unmutated evaluation keys (rows.y) using the shipped armiger_inference.py and zero_cue_plugin.py:

Evaluation Split	Valid N	Standalone Neural (FP32 / INT8)	Fair Blind Hybrid (+ ZeroCueCatalogPlugin)	Lift	Primary Error Mode Resolved by Hybrid Plugin
oct15_confirmation (Gold Standard)	2,835	95.03% / 95.06%	95.49% / 95.49% (97.30% Movie-vs-TV)	+0.46 pp	Bare-title films, multi-season packs, & siterips
fresh_confirmation (Fresh Holdout)	1,402	94.51% / 94.51%	94.58% / 94.58%	+0.07 pp	Zero-cue films & anime OVAs
beyond_confirmation (Hard Edge-Cases)	1,025	92.88% / 92.88%	95.12% / 95.12% (98.78% 4-Med)	+2.24 pp	Ambiguous comic archives, TV packs, & bare titles
tc_truespec_2000 (Specialist Zero-Cue)	2,000	87.40% / 87.30%	91.50% / 91.40% (99.85% 4-Med)	+4.10 pp	Zero-cue anime films vs. TV series (BLAKE2b-64 catalog)
tc_truespec_multilang_3000 (Multilingual)	3,000	97.50% / 97.43%	98.47% / 98.43%	+0.97 pp	Non-English bare titles & multi-season packs
5-Split Weighted Mean (N=10,262)	10,262	94.00% / 93.98%	95.42% / 95.39%	+1.42 pp	Consistent lift across all 5 evaluation splits
5-Split Macro Mean (N=10,262)	10,262	93.46% / 93.44%	95.03% / 95.00%	+1.57 pp	Zero info_hash memorization (BLAKE2b-64 title keys only)

---

Hardware-Attributed Latency & Footprint Benchmarks

Runtime & Precision	Hardware Platform	Footprint	Single-Item Latency (p50)	Batch-64 Latency (per item)	Batch-64 Throughput
armiger_ultra.pt (PyTorch CUDA FP32)	NVIDIA RTX GPU	32.2 MB	2.79 ms (358 items/s)	0.21 ms	4,795 items/s
armiger_ultra.pt (PyTorch MPS FP32)	Apple M4 Pro GPU (MPS)	32.2 MB	1.12 ms	0.29 ms	3,416 items/s
armiger.int8.onnx (ONNX Runtime INT8)	Apple M4 Pro CPU	8.3 MB	2.44 ms (410 items/s)	3.07 ms	326 items/s
armiger_ultra.fp32.onnx (ONNX Runtime FP32)	Apple M4 Pro CPU	32.3 MB	1.12 ms (893 items/s)	3.32 ms	301 items/s

---

1. Categorical & Open-Source Torrent Classifier Comparisons (Unmutated Ground-Truth Keys)

7-Dimension Comparison vs. Open-Source Classifiers & Parsers

System / Model	Included Here?	Footprint	11-Class Category (oct15 N=2,835)	Movie vs. TV (N=1,296)	Catalog Matching (N=1,296)	Audio Language (N=545)	Anime & Dub (N=507 / tc2000)	Adult / NSFW Gate (N=2,781)	Junk / Malware (N=2,170)	Swarm Liveliness (AUROC)
Armiger Ultra (Fair Hybrid)	Yes	32.3 MB (+26 MB cat.)	95.49% (98.47% mlang)	97.30%	96.53%	97.80%	99.61% (95.80% tc2000)	98.53%	100.00% (99.90% beyond)	96.83% Whse / 92.42% Pub
Armiger (armiger.int8.onnx + Plugins)	Yes	8.3 MB	95.49% (95.06% Raw)	97.30% (96.60% Raw)	96.53% (95.83% Raw)	97.80% (96.88% Raw)	99.61% (95.80% tc2000)	98.49%	100.00% (99.90% beyond)	92.42% Pub (86.64% Priors)
MIMMO BERT (360M)	—	~1,400 MB	72.24%	76.00% (93.61% 4-Med)	60.26%	N/A	N/A	91.05%	N/A	N/A
MIMMO Cascade	—	~1,400 MB	71.32%	76.08% (93.97% 4-Med)	60.34%	N/A	N/A	89.68%	N/A	N/A
RTN (rank-torrent-name)	—	Code Only	53.58%	81.64%	80.48%	93.76%	77.65%	79.00%	50.38% (0% rec.)	N/A
PTT / Parsett	—	Code Only	53.58%	81.64%	80.48%	93.76%	77.65%	79.00%	N/A	N/A
GuessIt	—	Code Only	38.69%	84.65%	84.03%	94.86%	78.80%	N/A	N/A	N/A
PTN (parse-torrent-title)	—	Code Only	35.59%	77.85%	77.62%	93.21%	68.65%	N/A	N/A	N/A
Anitomy	—	Code Only	33.40%	73.07%	72.76%	90.46%	64.55%	N/A	N/A	N/A
Bitmagnet Built-In	—	Code Only	35.17%	43.21% (88.4% ans.)	43.21%	88.26%	N/A	76.73%	54.60% (0% rec.)	N/A

Per-Category Accuracy Breakdown Across All 10 Media & Software Categories (Unmutated Keys)

Content Category	oct15 (N=2,835) Armiger Ultra (Fair Hybrid)	oct15 (N=2,835) Armiger (INT8 + ZeroCue)	oct15 MIMMO BERT (360M)	oct15 Bitmagnet Built-In	beyond (N=1,025) Armiger Ultra (Fair Hybrid)	beyond (N=1,025) Armiger (INT8 + ZeroCue)	beyond OpenAI GPT (gpt-6-luna LOTO)*	beyond Google Gemini (gemini-3.8-flash-low LOTO)*
Audiobook	97.89%	97.89%	0.00%	9.47%	100.00%	100.00%	98.67%	98.00%
Adult / XXX	97.80%	97.80%	95.50%	39.37%	91.95%	91.95%	94.83%	95.69%
TV Show	97.50%	97.50%	58.40%	28.62%	96.32%	96.32%	97.79%	93.75%
Movie	97.12%	97.12%	78.85%	55.83%	96.84%	96.84%	94.21%	92.63%
Software	93.75%	93.75%	59.38%	15.62%	94.59%	94.59%	97.30%	77.03%
Music	90.71%	90.71%	77.86%	7.86%	83.33%	83.33%	100.00%	100.00%
Game	90.16%	90.16%	0.00%	0.00%	96.30%	96.30%	98.52%	98.52%
Ebook	88.00%	88.00%	80.00%	4.00%	98.40%	98.40%	96.00%	96.80%
Course	81.82%	81.82%	0.00%	0.00%	92.00%	92.00%	95.50%	97.50%
Comic	66.67%	66.67%	0.00%	0.00%	98.33%	98.33%	96.67%	95.00%

Note on beyond_confirmation (N=1,025): gpt-6-luna and gemini-3.8-flash-low are evaluated using non-circular Leave-One-Teacher-Out (LOTO) reference keys (self_vote_weight = 0.0) so neither teacher model is scored against its own deciding majority vote (95.12% and 93.80% LOTO vs. 96.59% and 95.61% circular self-voted).

---

2. Frontier Cloud LLM & Local GPU LLM Comparisons

Accuracy Across 6 Core Dimensions vs. Frontier LLMs

Operational Cost & Token Rate vs. Frontier LLMs

Model	Overall 6-Dim Mean (95% CI)	11-Class Category (N=1,025 LOTO / oct15)	Catalog Matching (N=1,025 LOTO / oct15)	Audio Language (Strict E-vs-F N=229 / N=545)	Anime & Dub (Valid Media N=636 / N=507)	Junk / Malware (N=977 Bal / Acc)	Swarm Liveliness (AUROC)	Measured Hardware Speed	Cost per 1M Torrents (Batched / Real-Time)
Armiger Ultra (Fair Hybrid)	97.60% ([97.18, 98.02])	95.12% (95.49% oct15)	96.39% (96.53% oct15)	99.13% (97.80% oct15)	98.90% (99.61% oct15)	99.23% (99.90% Acc / 100% oct15)	96.83% Whse / 92.42% Pub	0.21 ms RTX / 0.29 ms M4 Pro MPS	$0.00 API ($0.0040 compute)
Armiger (armiger.int8.onnx + Plugins)	96.85% ([96.40, 97.30])	95.12% (95.49% oct15 / 95.06% Raw)	96.29% (95.80% Raw)	99.13% (97.80% oct15 / 96.88% Raw)	98.90% (99.61% oct15)	99.23% (99.90% Acc / 100% oct15)	92.42% Pub (86.64% Priors)	2.44 ms single / 3.07 ms batch-64 (M4 Pro CPU)	$0.00 API ($0.0020 compute)
Google Gemini (gemini-3.8-flash-low, LOTO)*	94.56% ([93.97, 95.15])	93.80% (95.61% Circ.)	90.88%	98.69%	95.60% (608/636)	99.18% (Non-Circ.)	89.20% Pub	~450 ms	$94.90 / $147.00
Anthropic Claude (claude-haiku-4-5)	92.88% ([92.10, 93.66])	93.76%	86.15%	95.63%	98.90%	93.09%	89.76% Pub	~310 ms	$712.00 / $1,220.00
OpenAI GPT (gpt-6-luna, LOTO)*	86.24% ([85.64, 86.83])	95.12% (96.59% Circ.)	89.07%	97.82% (94.68% oct15)	95.60% (608/636)	50.00% (0% rec.)	89.82% Pub	~365 ms	$127.30 / $220.50
Local GPU LLM (Vinci-MLE-8B-1.0)	84.79% ([83.80, 85.79])	80.78%	77.85%	93.45%	92.45%	97.91%	66.32% Pub	~185 ms (5.4/s RTX 4070 TiS)	$0.00 API ($1.85 local power)

---

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.

---

Armiger Ultra · Armiger · Swarm Survival · Swarm Priors · Zero-Cue Catalog · Audio Guard · Calibration & DHT RSI · DHT DB Sync · norvitech.com