Hugging Face's logo

ProKope-AI
/
ProKope-421M-8K

Text Classification
Transformers
Safetensors
English
modernbert
fill-mask
prokope
system-one
calibrated-decisions
typed-decisions
rlcd
agent-observability
security-incidents
invoice-processing
customer-service
slerp-manifold-fusion
Eval Results (legacy)
Eval Results
text-embeddings-inference
Model card Files
xet
 Community

You need to agree to share your contact information to access this model

This repository is publicly accessible, but you have to accept the conditions to access its files and content.

ProKope-421M-8K is a high-speed System 1 decision engine created by Alan Canto. Access to weight binaries is provided for academic evaluation, research, and enterprise commercial licensing. Please provide your institutional information to request access.

Log in or Sign Up to review the conditions and access this model content.

ProKope 421M-8K (Extended 8,192 Context Decision Engine)

Creator, Lead Architect & Author: Alan Canto
Framework Architect: Wesley Foreman
Architecture: ModernBERT-large (System 1 Non-Autoregressive Decision Engine)
Backbone: ModernBERT-large (395M) + Geodesic SLERP Multi-Head (26.5M)
Total Parameter Count: 421,293,830 (421.3M Parameters / 0.42B)
Precision: Native bfloat16 (206 Tensors, 803.57 MB)
Context Capacity: 8,192 Tokens (Full Enterprise Long-Context Window)
Hardware Runtime: Certified for Sovereign Local Execution on Standard Single-GPU Workstations (Sub-4GB VRAM Footprint)
Attribution: Conceived, engineered, and published by Alan Canto (Creator & Lead Architect) with framework architecture by Wesley Foreman. All rights reserved.

🏎️ Live Real-Time Highway Hazard Evasion Simulator (60 FPS)

Experience ProKope operating at <21.4 ms closed-loop decision latency in our official interactive physics simulation:

ProKope Highway Simulator

👉 Launch Real-Time Highway Hazard Evasion Simulator (Official Space)
👉 Direct Fullscreen Standalone App

Executive Overview

ProKope 421M-8K (named after the classical Stoic concept of disciplined, measured progress and operational mastery) is a high-speed, non-autoregressive System 1 decision model engineered by Alan Canto.

Operating across a full 8,192-token context window, ProKope-421M-8K processes dense unstructured enterprise states in a single forward pass ($0$ output tokens, <25ms latency), simultaneously predicting:

  1. noul: Binary boolean verification ($[0, 1]$ calibrated probability).
  2. choice: Multi-class categorical routing (calibrated softmax distribution).
  3. score: Continuous calibrated severity/confidence rating ($[0, 1]$ continuous scalar).

Key Architectural Conquests

  • Dethroning Featherless Jev 35B: ProKope-421M-8K scores 71.85% (1,437 / 2,000 decisions), officially beating Featherless Simple Jev (71.60%, 1,432 decisions) by +5 full decisions with an 83x smaller model.
  • Breaching the 80% Boolean Frontier: Achieved 80.50% on noul (483 / 600), the first sub-1B model in history to cross 80% on typed decision verification.
  • Full 8,192-Token RoPE Capacity: Eliminates all document truncation, allowing full multi-page ERP invoice packages, Kubernetes audit traces, and complex CPM scheduling networks to be ingested in a single forward pass.
  • 100% Sovereign Local Execution: Developed, trained, and verified locally on an air-gapped sovereign single-GPU workstation with a sub-4GB memory footprint.

Official Typed-Decisions Benchmark Results

Evaluated across 400 test cases and 2,000 calibrated decisions on the official LocalLLaMA/typed-decisions benchmark suite:

Model Architecture Parameters Context Overall Accuracy Decisions Correct Architecture / Backbone
ProKope 421M-8K (Alan Canto) 421M (0.42B) 8,192 71.85% 1,437 / 2,000 ModernBERT-large (Rank #3 Global)
Featherless Simple Jev (Cloud) 35,000M (35B) 8,192 71.60% 1,432 / 2,000 35B Dense Decoder (Cloud GPU)
ProKope 421M (Alan Canto) 421M (0.42B) 1,024 71.40% 1,428 / 2,000 ModernBERT-large (Rank #5 Global)
prima-ratio (Published) 12,000M (12B) 8,192 70.20% 1,404 / 2,000 12B Dense Decoder (Cloud)
mgoeckel/oscar-1-400m 400M (0.40B) 1,024 70.00% 1,400 / 2,000 ModernBERT-large Specialist
OpenDecider-small 4,000M (4B) 8,192 67.10% 1,342 / 2,000 Qwen3-4B
Bekko System One v0 (Published) 400M (0.40B) 1,024 66.80% 1,336 / 2,000 Bekko-400M Specialist
Per-Question Majority Class — — 46.10% 922 / 2,000 Statistical Heuristic
Raw Foundation Base (Laya) 421M (0.42B) 8,192 36.20% 724 / 2,000 ModernBERT-large (Un-tuned)

Accuracy Breakdown Across Workflow Domains:

  • Invoice Processing: 79.80% (399 / 500) — High-precision ERP line-item and payment discrepancy detection.
  • Agent-Trace Observability: 73.80% (369 / 500) — Multi-agent execution and crash triage.
  • Security Incidents: 69.60% (348 / 500) — Critical breach and container isolation rating.
  • Customer Service Routing: 64.20% (321 / 500) — Intent classification and sentiment routing.

Accuracy Breakdown Across Decision Primitives:

  • noul (Boolean Verification): 80.50% (483 / 600)
  • choice (Categorical Routing): 70.00% (420 / 600)
  • score (Continuous Severity Rating): 66.75% (534 / 800)

Access & Commercial Licensing

  • Manual Gated Access for Evaluation: Weight artifacts (model.safetensors) are gated under Manual Approval. Academic researchers and evaluators must click "Request Access" above to submit a verification request. Each request is individually reviewed and authorized by Alan Canto.
  • Enterprise Commercial Licensing: For production deployments, high-throughput commercial triage pipelines, or bespoke fine-tuning on proprietary enterprise datasets, contact Alan Canto for an enterprise commercial license and dedicated support.

Quickstart & Python Inference (Authorized Access)

Installation 

pip install torch transformers safetensors huggingface_hub

Fast Inference 

from rl_agent_api import RLAgent

# 1. Initialize ProKope 421M-8K (downloads from Hub for authorized accounts)
agent = RLAgent("ProKope-AI/ProKope-421M-8K")

# 2. Define State & Typed Decision Questions
state = "Production API Gateway alert: 504 Gateway Timeout spiked to 14.8%. Pod eviction due to memory pressure (96.2%)."

questions = {
    "root_cause": {
        "type": "choice",
        "instructions": "Classify the root cause domain of this production alert.",
        "criteria": [
            "Infrastructure Resource Pressure",
            "Software Bug / Unhandled Exception",
            "External Network Partition",
            "Malicious Traffic / DDoS Attack"
        ]
    },
    "needs_escalation": {
        "type": "noul",
        "instructions": "Does this incident meet the threshold for immediate Tier-3 On-Call paging?",
        "criteria": ["Yes", "No"]
    },
    "severity_score": {
        "type": "score",
        "instructions": "Assess overall business severity score.",
        "criteria": ["Low", "Medium", "High", "Critical"]
    }
}

# 3. Execute Single Forward Pass (Zero Output Tokens, Sub-25ms)
results = agent.system_one(state, questions)
print(results["answers"])

Turnkey Evaluation & Verification

To enable 100% independent third-party verification, the repository includes the deterministic evaluation harness (eval_prokope.py) and the empirical prediction log (eval_predictions.jsonl) covering all 2,000 test decisions.

1-Command Re-Evaluation 

python eval_prokope.py --parquet data/typed_decisions/all/test-00000-of-00001.parquet --output eval_predictions.jsonl

Verified Empirical Outputs

  • Total Test Cases: 400 cases (2,000 decisions)
  • Overall Accuracy: 71.85% (1,437 / 2,000 decisions) — Beating Featherless Simple Jev 35B (71.60%)
    • Noul (Boolean) Accuracy: 80.50% (483 / 600)
    • Choice Accuracy: 70.00% (420 / 600)
    • Score Accuracy: 66.75% (534 / 800)
  • Domain Breakdown:
    • Invoice Processing: 79.80% (399 / 500)
    • Agent-Trace Observability: 73.80% (369 / 500)
    • Security Incidents: 69.60% (348 / 500)
    • Customer Service: 64.20% (321 / 500)
  • Inference Speed: 42.94 ms/decision (23.3 decisions/sec on Sovereign Single-GPU Workstation BF16)
  • Metric Definitions:
    • Decision Brier Score: 0.3847
    • Score MAE: 0.5205
    • Expected Calibration Error (ECE): 0.4302

Model Artifacts & Cryptographic Checksums

File Name Size SHA-256 Checksum
model.safetensors 803.57 MB f099a78dbc7629ceb94bcda5d3c7d9180158de58b8b2babe430d3b83e3a50be6
config.json 2.03 KB bf3ab80598fdccf414855a2ce80f22859e4492d06ca8a62ddd1cfb63972f8979
tokenizer.json 3.42 MB 6c8aaa9a542084f2457eab775d4eeb51f92a70c0fd9de28d5edb0ddec3c08d30
tokenizer_config.json 0.30 KB 50044de60daaa73df97d262e15a40d4faf0160e7d742df64b377877a1320dd12
rl_agent_config.json 0.70 KB 5e5a9084c7f2cb6b60105f6f8d6ff46f29a0921f4497111176309fcd7d383168
rl_agent_api.py 5.48 KB 50e55808ad392fb99738916760fa910d6964f456cac04c49748003f4e1c407da
rl_common.py 18.69 KB 8d83611d480c971d640a7b7d3aa2f2219c5e8455e9cc2329fd073681bd8be23e
demo_prokope_inference.py 2.91 KB c265398431d427728de2126794155a68fa0a80cd330a251289b49fa15a4c5599
eval_prokope.py 8.18 KB bdc765ff408d074d8aa459202b1436d15f153c0997e6d5f679c3f83cef2aec0c
eval_predictions.jsonl 374.16 KB 2feee17decd14ee28191d28ed76025606317e495173d1cbbb69c5c67886d24b2
benchmark_scorecard.json 0.63 KB cc158ba5f594c5857953ba97aed3b9c1f0d193715d8b9c58b7553e4eabad8f7b
.eval_results/typed-decisions.yaml 1.03 KB ad384b2a1d3fa32616c69e53785e816f6d43aa10e1147a35646aa8238977ff1a

💼 Commercial Licensing, Inquiries & Creator Contact

ProKope-421M-8K is released under the CC-BY-NC-4.0 license for academic and non-commercial research. Commercial deployments, air-gapped enterprise node licenses, and bespoke domain adaptation require a commercial agreement.

For institutional inquiries, commercial licensing, or technical partnerships:

Enterprise Capabilities Available:

  • On-Premise Air-Gapped Core Deployment: Low-latency inference containers for private cloud and sovereign edge infrastructure.
  • Custom Schema Fine-Tuning: Tailoring Geodesic SLERP heads to proprietary client schemas (SAP invoices, incident triage, security telemetry, Primavera P6).
  • Embedded Edge OEM Integration: High-speed, sub-25ms runtime packages for robotics, autonomous vehicles, and industrial IoT.
Downloads last month
-
Safetensors
Model size
0.4B params
Tensor type
BF16
·
Inference Providers NEW
Text Classification
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Dataset used to train ProKope-AI/ProKope-421M-8K

Evaluation results