File size: 12,506 Bytes
81a0d30
 
 
350c8f3
 
81a0d30
f63b162
 
 
81a0d30
 
 
 
 
 
 
 
350c8f3
 
 
81a0d30
 
 
 
350c8f3
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
81a0d30
 
90a8c55
 
 
1a2d614
90a8c55
81a0d30
 
350c8f3
 
 
81a0d30
 
350c8f3
451a472
 
 
 
 
 
350c8f3
 
81a0d30
350c8f3
81a0d30
350c8f3
81a0d30
350c8f3
81a0d30
350c8f3
81a0d30
350c8f3
 
 
 
 
81a0d30
350c8f3
81a0d30
350c8f3
81a0d30
350c8f3
81a0d30
350c8f3
 
 
 
 
 
 
 
81a0d30
350c8f3
 
90a8c55
 
 
 
 
 
 
 
350c8f3
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
915eddd
 
 
350c8f3
915eddd
 
350c8f3
 
 
aaae999
350c8f3
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
81a0d30
 
 
350c8f3
81a0d30
 
 
 
 
350c8f3
81a0d30
 
 
 
350c8f3
81a0d30
350c8f3
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
81a0d30
 
 
 
 
 
 
350c8f3
81a0d30
 
350c8f3
 
 
 
 
 
 
81a0d30
350c8f3
 
81a0d30
350c8f3
 
 
 
 
 
 
81a0d30
350c8f3
81a0d30
350c8f3
 
 
 
 
 
 
 
81a0d30
350c8f3
 
 
 
 
 
 
 
 
 
 
 
 
81a0d30
350c8f3
81a0d30
350c8f3
81a0d30
350c8f3
 
 
 
 
 
81a0d30
350c8f3
 
 
 
 
81a0d30
350c8f3
81a0d30
350c8f3
 
81a0d30
 
350c8f3
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
81a0d30
 
 
350c8f3
 
 
 
 
 
 
 
 
 
 
 
 
 
 
81a0d30
 
 
350c8f3
 
 
5884fb2
350c8f3
 
5884fb2
350c8f3
 
 
 
 
 
 
 
90a8c55
350c8f3
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
---
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>

---

## 🎯 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?

- πŸ” **Quantum-Ready**: Deep expertise in QKD protocols (BB84, E91, B92, MDI-QKD, TF-QKD)
- 🏒 **Enterprise-Grade**: Designed for compliance with NIS2, DORA, PCI-DSS, ISO 27001
- πŸ”“ **Open Source**: Fully open for customization, audit, and enterprise deployment
- 🧠 **Thinking Mode**: Chain-of-thought reasoning for complex security analysis
- ⚑ **Production-Ready**: Multiple quantization formats for any deployment scenario

---

## πŸ›οΈ 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 |
|------------|-------------|
| πŸ–₯️ **On-Premise** | Air-gapped environments, full control |
| ☁️ **Private Cloud** | AWS, Azure, GCP with data sovereignty |
| πŸ”’ **Hybrid** | Edge inference + cloud orchestration |
| πŸ“± **Local** | Ollama, llama.cpp, vLLM |

---

## πŸ› 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

---

## πŸ“Š 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.

---

## πŸ”¬ 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 |

### πŸš€ 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)

---

## πŸ“¦ 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 |

---

## πŸš€ 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)
```

---

## 🧠 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.
```

---

## πŸ”’ Security Considerations

### Responsible Use

Kairos is designed for **defensive security** purposes:

βœ… **Intended Uses:**
- Cryptographic protocol education
- Security compliance assessment  
- QKD system development
- Vulnerability research (authorized)
- Post-quantum migration planning

❌ **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

---

## 🀝 Contributing

Kairos is open for community contributions:

```bash
# Areas for contribution:
# - Additional training data
# - New protocol implementations
# - Benchmark evaluations
# - Documentation improvements
# - Integration examples
```

---

## πŸ“œ 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).

- βœ… Commercial use allowed
- βœ… Modification allowed
- βœ… Distribution allowed
- ⚠️ Must include license notice

---

## πŸ™ 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

---

## πŸ“š 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>