Text Generation
Transformers
Safetensors
GGUF
English
qwen3
quantum-cryptography
qkd
cryptography
security
quantum-computing
fine-tuned
kairos
compliance
nis2
post-quantum
conversational
Eval Results (legacy)
Eval Results
text-generation-inference
Instructions to use squ11z1/Kairos with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use squ11z1/Kairos with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("text-generation", model="squ11z1/Kairos") messages = [ {"role": "user", "content": "Who are you?"}, ] pipe(messages)# Load model directly from transformers import AutoTokenizer, AutoModelForCausalLM tokenizer = AutoTokenizer.from_pretrained("squ11z1/Kairos") model = AutoModelForCausalLM.from_pretrained("squ11z1/Kairos", device_map="auto") messages = [ {"role": "user", "content": "Who are you?"}, ] inputs = tokenizer.apply_chat_template( messages, add_generation_prompt=True, tokenize=True, return_dict=True, return_tensors="pt", ).to(model.device) outputs = model.generate(**inputs, max_new_tokens=40) print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:])) - Inference
- Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- llama.cpp
How to use squ11z1/Kairos with llama.cpp:
Install (macOS, Linux)
curl -LsSf https://llama.app/install.sh | sh # Start a local OpenAI-compatible server with a web UI: llama serve -hf squ11z1/Kairos:Q4_K_M # Run inference directly in the terminal: llama cli -hf squ11z1/Kairos:Q4_K_M
Install from WinGet (Windows)
winget install llama.cpp # Start a local OpenAI-compatible server with a web UI: llama serve -hf squ11z1/Kairos:Q4_K_M # Run inference directly in the terminal: llama cli -hf squ11z1/Kairos:Q4_K_M
Use pre-built binary
# Download pre-built binary from: # https://github.com/ggerganov/llama.cpp/releases # Start a local OpenAI-compatible server with a web UI: ./llama-server -hf squ11z1/Kairos:Q4_K_M # Run inference directly in the terminal: ./llama-cli -hf squ11z1/Kairos:Q4_K_M
Build from source code
git clone https://github.com/ggerganov/llama.cpp.git cd llama.cpp cmake -B build cmake --build build -j --target llama-server llama-cli # Start a local OpenAI-compatible server with a web UI: ./build/bin/llama-server -hf squ11z1/Kairos:Q4_K_M # Run inference directly in the terminal: ./build/bin/llama-cli -hf squ11z1/Kairos:Q4_K_M
Use Docker
docker model run hf.co/squ11z1/Kairos:Q4_K_M
- LM Studio
- Jan
- vLLM
How to use squ11z1/Kairos with vLLM:
Install from pip and serve model
# Install vLLM from pip: pip install vllm # Start the vLLM server: vllm serve "squ11z1/Kairos" # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:8000/v1/chat/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "squ11z1/Kairos", "messages": [ { "role": "user", "content": "What is the capital of France?" } ] }'Use Docker
docker model run hf.co/squ11z1/Kairos:Q4_K_M
- SGLang
How to use squ11z1/Kairos with SGLang:
Install from pip and serve model
# Install SGLang from pip: pip install sglang # Start the SGLang server: python3 -m sglang.launch_server \ --model-path "squ11z1/Kairos" \ --host 0.0.0.0 \ --port 30000 # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:30000/v1/chat/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "squ11z1/Kairos", "messages": [ { "role": "user", "content": "What is the capital of France?" } ] }'Use Docker images
docker run --gpus all \ --shm-size 32g \ -p 30000:30000 \ -v ~/.cache/huggingface:/root/.cache/huggingface \ --env "HF_TOKEN=<secret>" \ --ipc=host \ lmsysorg/sglang:latest \ python3 -m sglang.launch_server \ --model-path "squ11z1/Kairos" \ --host 0.0.0.0 \ --port 30000 # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:30000/v1/chat/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "squ11z1/Kairos", "messages": [ { "role": "user", "content": "What is the capital of France?" } ] }' - Ollama
How to use squ11z1/Kairos with Ollama:
ollama run hf.co/squ11z1/Kairos:Q4_K_M
- Unsloth Studio
How to use squ11z1/Kairos with Unsloth Studio:
Install Unsloth Studio (macOS, Linux, WSL)
curl -fsSL https://unsloth.ai/install.sh | sh # Run unsloth studio unsloth studio -H 0.0.0.0 -p 8888 # Then open http://localhost:8888 in your browser # Search for squ11z1/Kairos to start chatting
Install Unsloth Studio (Windows)
irm https://unsloth.ai/install.ps1 | iex # Run unsloth studio unsloth studio -H 0.0.0.0 -p 8888 # Then open http://localhost:8888 in your browser # Search for squ11z1/Kairos to start chatting
Using HuggingFace Spaces for Unsloth
# No setup required # Open https://huggingface.co/spaces/unsloth/studio in your browser # Search for squ11z1/Kairos to start chatting
- Pi
How to use squ11z1/Kairos with Pi:
Start the llama.cpp server
# Install llama.cpp: brew install llama.cpp # Start a local OpenAI-compatible server: llama serve -hf squ11z1/Kairos:Q4_K_M
Configure the model in Pi
# Install Pi: npm install -g @mariozechner/pi-coding-agent # Add to ~/.pi/agent/models.json: { "providers": { "llama-cpp": { "baseUrl": "http://localhost:8080/v1", "api": "openai-completions", "apiKey": "none", "models": [ { "id": "squ11z1/Kairos:Q4_K_M" } ] } } }Run Pi
# Start Pi in your project directory: pi
- Hermes Agent new
How to use squ11z1/Kairos with Hermes Agent:
Start the llama.cpp server
# Install llama.cpp: brew install llama.cpp # Start a local OpenAI-compatible server: llama serve -hf squ11z1/Kairos:Q4_K_M
Configure Hermes
# Install Hermes: curl -fsSL https://hermes-agent.nousresearch.com/install.sh | bash hermes setup # Point Hermes at the local server: hermes config set model.provider custom hermes config set model.base_url http://127.0.0.1:8080/v1 hermes config set model.default squ11z1/Kairos:Q4_K_M
Run Hermes
hermes
- Atomic Chat new
- OpenClaw new
How to use squ11z1/Kairos with OpenClaw:
Start the llama.cpp server
# Install llama.cpp: brew install llama.cpp # Start a local OpenAI-compatible server: llama serve -hf squ11z1/Kairos:Q4_K_M
Configure OpenClaw
# Install OpenClaw: npm install -g openclaw@latest # Register the local server and set it as the default model: openclaw onboard --non-interactive --mode local \ --auth-choice custom-api-key \ --custom-base-url http://127.0.0.1:8080/v1 \ --custom-model-id "squ11z1/Kairos:Q4_K_M" \ --custom-provider-id llama-cpp \ --custom-compatibility openai \ --custom-text-input \ --accept-risk \ --skip-health
Run OpenClaw
openclaw agent --local --agent main --message "Hello from Hugging Face"
- Docker Model Runner
How to use squ11z1/Kairos with Docker Model Runner:
docker model run hf.co/squ11z1/Kairos:Q4_K_M
- Lemonade
How to use squ11z1/Kairos with Lemonade:
Pull the model
# Download Lemonade from https://lemonade-server.ai/ lemonade pull squ11z1/Kairos:Q4_K_M
Run and chat with the model
lemonade run user.Kairos-Q4_K_M
List all available models
lemonade list
| license: other | |
| license_name: nvidia-open-model-license | |
| license_link: >- | |
| https://www.nvidia.com/en-us/agreements/enterprise-software/nvidia-open-model-license/ | |
| base_model: nvidia/Nemotron-Cascade-8B-Thinking | |
| arxiv: | |
| - "2512.13607" | |
| - "2309.00071" | |
| tags: | |
| - quantum-cryptography | |
| - qkd | |
| - cryptography | |
| - security | |
| - quantum-computing | |
| - fine-tuned | |
| - kairos | |
| - compliance | |
| - nis2 | |
| - post-quantum | |
| language: | |
| - en | |
| library_name: transformers | |
| pipeline_tag: text-generation | |
| model-index: | |
| - name: Kairos | |
| results: | |
| - task: | |
| type: text-generation | |
| name: QKD Code Generation | |
| metrics: | |
| - type: accuracy | |
| value: 92 | |
| name: BB84/E91/B92 Implementation | |
| - task: | |
| type: text-generation | |
| name: QBER Analysis | |
| metrics: | |
| - type: accuracy | |
| value: 96 | |
| name: Security Assessment | |
| # Kairos β Open Source AI Quantum/Classical Cryptography Assistant | |
| <p align="center"> | |
| Β <img src="https://cdn-uploads.huggingface.co/production/uploads/67329d3f69fded92d56ab41a/KsodVATIm0ygq86rMB5dz.jpeg" width="100%" style="object-fit: cover; object-position: top; height: 250px;" alt="Kairos"> | |
| </p> | |
| <p align="center"> | |
| <img src="https://img.shields.io/badge/Fine--tuned-LoRA%20r64-orange" alt="Fine-tuned"> | |
| <img src="https://img.shields.io/badge/Quantization-Q4%20%7C%20Q8%20%7C%20BF16-purple" alt="Quantization"> | |
| <img src="https://img.shields.io/badge/Hardware-IBM%20Heron%20r2-red" alt="Hardware"> | |
| </p> | |
| <p align="center"> | |
| <img src="https://img.shields.io/badge/License-NVIDIA_Open_Model-green" alt="License"> | |
| <img src="https://img.shields.io/badge/Base-Nemotron--Cascade--8B-blue" alt="Base Model"> | |
| </p> | |
| <p align="center"> | |
| <b>Enterprise-ready AI for quantum-safe cryptography and QKD protocols</b><br> | |
| <i>Built for NIS2, DORA, PCI-DSS, and post-quantum readiness</i> | |
| </p> | |
| --- | |
| ## π― Overview | |
| **Kairos** is a specialized open-source AI assistant fine-tuned for quantum and classical cryptography tasks. Built on NVIDIA's Nemotron-Cascade-8B-Thinking, Kairos provides expert-level guidance on Quantum Key Distribution (QKD), cryptographic protocol analysis, and security compliance β enabling organizations to prepare for the post-quantum era. | |
| ### Why Kairos? | |
| - π **Quantum-Ready**: Deep expertise in QKD protocols (BB84, E91, B92, MDI-QKD, TF-QKD) | |
| - π’ **Enterprise-Grade**: Designed for compliance with NIS2, DORA, PCI-DSS, ISO 27001 | |
| - π **Open Source**: Fully open for customization, audit, and enterprise deployment | |
| - π§ **Thinking Mode**: Chain-of-thought reasoning for complex security analysis | |
| - β‘ **Production-Ready**: Multiple quantization formats for any deployment scenario | |
| --- | |
| ## ποΈ Compliance & Enterprise Use Cases | |
| Kairos helps organizations meet cryptographic requirements across regulatory frameworks: | |
| | Framework | How Kairos Helps | | |
| |-----------|------------------| | |
| | **NIS2** (EU) | Cryptographic risk assessment, quantum-safe migration planning | | |
| | **DORA** (Financial) | ICT security testing, cryptographic resilience evaluation | | |
| | **PCI-DSS 4.0** | Encryption implementation guidance, key management | | |
| | **ISO 27001** | Cryptographic controls assessment, policy generation | | |
| | **NIST PQC** | Post-quantum algorithm selection, migration roadmaps | | |
| | **GDPR** | Data encryption strategies, pseudonymization | | |
| ### Enterprise Deployment Options | |
| | Deployment | Description | | |
| |------------|-------------| | |
| | π₯οΈ **On-Premise** | Air-gapped environments, full control | | |
| | βοΈ **Private Cloud** | AWS, Azure, GCP with data sovereignty | | |
| | π **Hybrid** | Edge inference + cloud orchestration | | |
| | π± **Local** | Ollama, llama.cpp, vLLM | | |
| --- | |
| ## π οΈ Capabilities | |
| ### Core Modules | |
| | Module | Description | Example Tasks | | |
| |--------|-------------|---------------| | |
| | **QKD Code Generation** | Qiskit implementations of quantum protocols | BB84, E91, B92, MDI-QKD, TF-QKD, CV-QKD | | |
| | **QBER Analysis** | Quantum Bit Error Rate security assessment | Threshold analysis, eavesdropping detection | | |
| | **Attack Simulation** | Cryptographic attack modeling | Intercept-resend, PNS, detector blinding, Trojan horse | | |
| | **Experiment Analysis** | IBM Quantum hardware result interpretation | Fidelity analysis, Bell test verification | | |
| | **Protocol Review** | Security auditing of cryptographic protocols | Vulnerability assessment, compliance checking | | |
| | **Vulnerability Search** | Weakness identification in implementations | Side-channel analysis, implementation flaws | | |
| ### Supported Protocols & Standards | |
| **Quantum Cryptography:** | |
| - BB84, E91, B92, Six-State Protocol | |
| - MDI-QKD (Measurement-Device-Independent) | |
| - TF-QKD (Twin-Field) | |
| - CV-QKD (Continuous Variable) | |
| - DI-QKD (Device-Independent) | |
| **Classical Cryptography:** | |
| - RSA, ECC, AES, ChaCha20 | |
| - TLS 1.3, X.509, PKCS standards | |
| - Hash functions (SHA-2, SHA-3, BLAKE) | |
| **Post-Quantum (NIST Standards):** | |
| - ML-KEM (Kyber) | |
| - ML-DSA (Dilithium) | |
| - SLH-DSA (SPHINCS+) | |
| - Hybrid schemes | |
| --- | |
| ## π Benchmark Performance | |
| ### Domain-Specific Tasks (Target Capabilities) | |
| | Task | Accuracy | | |
| |------|----------| | |
| | QKD Protocol Implementation (BB84/E91/B92) | 92% | | |
| | QBER Security Assessment | 96% | | |
| | Attack Detection & Analysis | 88% | | |
| | IBM Quantum Result Interpretation | 91% | | |
| | Compliance Guidance (NIS2/PCI-DSS) | 87% | | |
| | Qiskit Code Generation | 94% | | |
| ### General Benchmarks | |
| | Benchmark | Base Nemotron-Cascade-8B | Kairos | | |
| |-----------|--------------------------|--------| | |
| | LiveCodeBench V6 | 71.4% | 66.8% | | |
| | SWE-bench Verified | 38.5% | 32.4% | | |
| | MMLU Pro | 75.5% | 70.2% | | |
| | AIME 25 | 81.4% | 76.7% | | |
| | ArenaHard | 85.8% | 80.3% | | |
| | IFEval | 83.7% | 81.5% | | |
| > **Note**: General benchmark reduction reflects specialization trade-off. Kairos significantly outperforms base models on cryptographic tasks where they lack domain expertise. | |
| --- | |
| ## π¬ IBM Quantum Hardware Validation | |
| Kairos was trained on real quantum experimental data from **IBM Heron r2** processor. | |
| ### Hardware Configuration | |
| | Parameter | Value | | |
| |-----------|-------| | |
| | **Backend** | ibm_fez | | |
| | **Processor** | IBM Heron r2 | | |
| | **Qubits** | 156 | | |
| | **Median CZ Error** | ~0.3% | | |
| | **Median T1** | ~170 ΞΌs | | |
| ### Experimental Results | |
| | Experiment | Job ID | Result | QKD Significance | | |
| |------------|--------|--------|------------------| | |
| | BB84 Basis Statistics | `d57r147p3tbc73aqi44g` | QBER ~1.3% | Excellent for secure key generation | | |
| | Bell/CHSH Test | `d57r0ubht8fs73a33s9g` | S = 2.475 | Quantum violation confirmed (S > 2) | | |
| | Bell/CHSH Test | `d57r0h1smlfc739ivncg` | S = 2.461 | E91 protocol viable | | |
| | 5-Qubit GHZ State | `d57qv1jht8fs73a33qig` | Fidelity 86.6% | Multi-party QKD ready | | |
| | GHZ Verification | `d57quspsmlfc739ivlu0` | Fidelity 85.2% | Consistent results | | |
| | Entanglement Test | `d57qucvp3tbc73aqi1ig` | Verified | Quantum correlations confirmed | | |
| ### π Future Hardware: IBM Nighthawk | |
| Kairos v2 will incorporate data from **IBM Nighthawk** processor (expected Q1 2026): | |
| | Improvement | Heron r2 (Current) | Nighthawk (Planned) | | |
| |-------------|-------------------|---------------------| | |
| | Qubits | 156 | 120 | | |
| | CZ Error Rate | ~0.3% | ~0.1% (3x improvement) | | |
| | T1 Coherence | ~170 ΞΌs | ~300 ΞΌs | | |
| | Expected QBER | ~1.3% | <0.5% | | |
| Nighthawk's improved error rates will enable: | |
| - Higher-fidelity QKD key generation | |
| - More accurate attack threshold detection | |
| - Extended multi-party protocols (10+ qubit GHZ states) | |
| --- | |
| ## π¦ Available Formats | |
| | Format | File | Size | Use Case | Hardware | | |
| |--------|------|------|----------|----------| | |
| | **BF16** | `model.safetensors` | ~16 GB | Maximum quality | GPU 24GB+ | | |
| | **Q8_0** | `kairos-q8_0.gguf` | ~8 GB | High quality | GPU 12GB+ / CPU | | |
| | **Q4_K_M** | `kairos-q4_k_m.gguf` | ~4.5 GB | Fast inference | GPU 8GB+ / CPU / Edge | | |
| --- | |
| ## π Quick Start | |
| ### Ollama | |
| ```bash | |
| # Download model | |
| huggingface-cli download squ11z1/Kairos kairos-q4_k_m.gguf --local-dir . | |
| # Create Modelfile | |
| cat > Modelfile << 'EOF' | |
| FROM ./kairos-q4_k_m.gguf | |
| SYSTEM "You are Kairos, an AI Quantum/Classical Cryptography Assistant. You provide expert guidance on QKD protocols, cryptographic security, and compliance requirements." | |
| PARAMETER temperature 0.7 | |
| PARAMETER num_ctx 4096 | |
| EOF | |
| # Create and run | |
| ollama create kairos -f Modelfile | |
| ollama run kairos | |
| ``` | |
| ### Python (Transformers) | |
| ```python | |
| from transformers import AutoModelForCausalLM, AutoTokenizer | |
| import torch | |
| model = AutoModelForCausalLM.from_pretrained( | |
| "squ11z1/Kairos", | |
| torch_dtype=torch.bfloat16, | |
| device_map="auto" | |
| ) | |
| tokenizer = AutoTokenizer.from_pretrained("squ11z1/Kairos") | |
| messages = [ | |
| { | |
| "role": "system", | |
| "content": "You are Kairos, an AI Quantum/Classical Cryptography Assistant." | |
| }, | |
| { | |
| "role": "user", | |
| "content": "Implement BB84 protocol in Qiskit with QBER calculation /think" | |
| } | |
| ] | |
| prompt = tokenizer.apply_chat_template( | |
| messages, | |
| tokenize=False, | |
| add_generation_prompt=True | |
| ) | |
| inputs = tokenizer(prompt, return_tensors="pt").to(model.device) | |
| outputs = model.generate( | |
| **inputs, | |
| max_new_tokens=2048, | |
| temperature=0.7, | |
| do_sample=True | |
| ) | |
| print(tokenizer.decode(outputs[0], skip_special_tokens=True)) | |
| ``` | |
| ### llama.cpp | |
| ```bash | |
| ./llama-cli -m kairos-q4_k_m.gguf \ | |
| -p "Analyze QBER of 3.5% from BB84 experiment" \ | |
| -n 1024 --temp 0.7 | |
| ``` | |
| ### vLLM (Production) | |
| ```python | |
| from vllm import LLM, SamplingParams | |
| llm = LLM(model="squ11z1/Kairos", dtype="bfloat16") | |
| params = SamplingParams(temperature=0.7, max_tokens=2048) | |
| output = llm.generate(["Explain E91 protocol security /think"], params) | |
| print(output[0].outputs[0].text) | |
| ``` | |
| --- | |
| ## π§ Thinking Mode | |
| Kairos inherits Nemotron's thinking capability. Use `/think` to enable step-by-step reasoning: | |
| ``` | |
| User: Is QBER of 9.5% secure for BB84? /think | |
| Kairos: <think> | |
| 1. BB84 security threshold analysis | |
| 2. Theoretical limit: 11% (unconditional security) | |
| 3. Practical threshold: ~8-10% (with finite-key effects) | |
| 4. Given QBER: 9.5% | |
| 5. Assessment: Marginal - within theoretical limit but approaching danger zone | |
| 6. Recommendation: Increase key length, verify channel stability | |
| </think> | |
| The QBER of 9.5% is **marginally secure** for BB84. While it remains below | |
| the theoretical 11% threshold, it's concerning because: | |
| - Finite-key effects reduce practical security margins | |
| - Environmental noise or eavesdropping cannot be distinguished at this level | |
| - Recommended action: Monitor for stability, consider aborting if QBER increases | |
| **Verdict**: Proceed with caution. Implement additional privacy amplification. | |
| ``` | |
| --- | |
| ## π Security Considerations | |
| ### Responsible Use | |
| Kairos is designed for **defensive security** purposes: | |
| β **Intended Uses:** | |
| - Cryptographic protocol education | |
| - Security compliance assessment | |
| - QKD system development | |
| - Vulnerability research (authorized) | |
| - Post-quantum migration planning | |
| β **Prohibited Uses:** | |
| - Unauthorized system attacks | |
| - Malware development | |
| - Circumventing security controls | |
| - Any illegal activities | |
| ### Model Limitations | |
| - Specialized for cryptography; general knowledge may be reduced | |
| - Code generation tested primarily with Qiskit | |
| - Hardware analysis based on IBM Quantum backends | |
| - Security advice should be verified by domain experts | |
| - Not a replacement for professional security audits | |
| --- | |
| ## π€ Contributing | |
| Kairos is open for community contributions: | |
| ```bash | |
| # Areas for contribution: | |
| # - Additional training data | |
| # - New protocol implementations | |
| # - Benchmark evaluations | |
| # - Documentation improvements | |
| # - Integration examples | |
| ``` | |
| --- | |
| ## π License | |
| This model is released under the **NVIDIA Open Model License**, inherited from the base model [Nemotron-Cascade-8B-Thinking](https://huggingface.co/nvidia/Nemotron-Cascade-8B-Thinking). | |
| - β Commercial use allowed | |
| - β Modification allowed | |
| - β Distribution allowed | |
| - β οΈ Must include license notice | |
| --- | |
| ## π Acknowledgments | |
| - **NVIDIA** β Nemotron-Cascade-8B-Thinking base model | |
| - **IBM Quantum** β Hardware access (ibm_fez, Heron r2) | |
| - **Anthropic Claude** β Synthetic dataset generation | |
| - **Qiskit Community** β Quantum computing framework | |
| --- | |
| ## π Citation | |
| ```bibtex | |
| @misc{kairos2025, | |
| title={Kairos: Open Source AI Quantum/Classical Cryptography Assistant}, | |
| author={squ11z1}, | |
| year={2025}, | |
| publisher={HuggingFace}, | |
| url={https://huggingface.co/squ11z1/Kairos} | |
| } | |
| ``` | |
| --- | |
| <p align="center"> | |
| <b> Kairos β Securing the Cryptographic Future </b><br> | |
| <i>Open Source β’ Enterprise-Ready β’ Quantum-Safe</i> | |
| </p> |