Text Generation
GGUF
English
Arabic
code
gemma2
google
mantiq
logic
arabic
epistemology
reasoning
chain-of-thought
aynengine
conversational
Instructions to use enver/ayncoding-gemma2-2b with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- llama.cpp
How to use enver/ayncoding-gemma2-2b 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 enver/ayncoding-gemma2-2b # Run inference directly in the terminal: llama cli -hf enver/ayncoding-gemma2-2b
Install from WinGet (Windows)
winget install llama.cpp # Start a local OpenAI-compatible server with a web UI: llama serve -hf enver/ayncoding-gemma2-2b # Run inference directly in the terminal: llama cli -hf enver/ayncoding-gemma2-2b
Build from source code
git clone https://github.com/ggml-org/llama.cpp.git cd llama.cpp cmake -B build cmake --build build -j --target llama-app # Start a local OpenAI-compatible server with a web UI: ./build/bin/llama serve -hf enver/ayncoding-gemma2-2b # Run inference directly in the terminal: ./build/bin/llama cli -hf enver/ayncoding-gemma2-2b
Use Docker
docker model run hf.co/enver/ayncoding-gemma2-2b
- Llama new
- LM Studio
- Jan
- vLLM
How to use enver/ayncoding-gemma2-2b with vLLM:
Install from pip and serve model
# Install vLLM from pip: pip install vllm # Start the vLLM server: vllm serve "enver/ayncoding-gemma2-2b" # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:8000/v1/chat/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "enver/ayncoding-gemma2-2b", "messages": [ { "role": "user", "content": "What is the capital of France?" } ] }'Use Docker
docker model run hf.co/enver/ayncoding-gemma2-2b
- Ollama
How to use enver/ayncoding-gemma2-2b with Ollama:
ollama run hf.co/enver/ayncoding-gemma2-2b
- Unsloth Desktop
- Docker Model Runner
How to use enver/ayncoding-gemma2-2b with Docker Model Runner:
docker model run hf.co/enver/ayncoding-gemma2-2b
- Lemonade
How to use enver/ayncoding-gemma2-2b with Lemonade:
Pull the model
# Download Lemonade from https://lemonade-server.ai/ lemonade pull enver/ayncoding-gemma2-2b
Run and chat with the model
lemonade run user.ayncoding-gemma2-2b-{{QUANT_TAG}}List all available models
lemonade list
- Atomic Chat
docs: add comprehensive Model Card for AynCoding-Gemma2
Browse files
README.md
ADDED
|
@@ -0,0 +1,144 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
language:
|
| 3 |
+
- en
|
| 4 |
+
- ar
|
| 5 |
+
license: gemma
|
| 6 |
+
tags:
|
| 7 |
+
- code
|
| 8 |
+
- gemma2
|
| 9 |
+
- google
|
| 10 |
+
- gguf
|
| 11 |
+
- mantiq
|
| 12 |
+
- logic
|
| 13 |
+
- arabic
|
| 14 |
+
- epistemology
|
| 15 |
+
- reasoning
|
| 16 |
+
- chain-of-thought
|
| 17 |
+
- aynengine
|
| 18 |
+
base_model: google/gemma-2-2b-it
|
| 19 |
+
pipeline_tag: text-generation
|
| 20 |
+
---
|
| 21 |
+
|
| 22 |
+
# 🏛️ AynCoding-Gemma2: Sovereign Epistemic Classical Arabic Logic (Manṭiq) 2.6B
|
| 23 |
+
|
| 24 |
+
**AynCoding-Gemma2** (`ayncoding-gemma2-2b`) is a sovereign, epistemic code synthesis and architectural refactoring model based on **Google's flagship Gemma 2 2B** architecture (`google/gemma-2-2b-it`), aligned and conditioned with **Classical Arabic Logic (*Manṭiq*)** and **Morphological Root Lexicography (*Ishtiqāq*)**.
|
| 25 |
+
|
| 26 |
+
Built upon Google DeepMind's distilled Gemma 2 architecture, **AynCoding-Gemma2** marries cutting-edge model distillation with 1,200 years of classical Arabic epistemological rigor. It natively engages a structured **Chain-of-Thought reasoning block (`<ayn_mantiq> ... </ayn_mantiq>`)** prior to code generation, analyzing software specifications through real essential definitions (*Al-Ḥadd bi al-Dhātiyyāt*), eliminating logical fallacies (*Dafʿ al-Dawr*, *Dafʿ al-Tasalsul*, *ʿAdam al-Tanāquḍ*), and mapping domain concepts to authentic tri-consonantal roots.
|
| 27 |
+
|
| 28 |
+
---
|
| 29 |
+
|
| 30 |
+
## 🌟 The 5 Classical Arabic Epistemic Pillars
|
| 31 |
+
|
| 32 |
+
AynCoding-Gemma2 is strictly governed by the foundational canons of classical Arabic scholarship:
|
| 33 |
+
|
| 34 |
+
1. **Abū Ḥāmid al-Ghazālī** (*Miʿyār al-ʿIlm fī Fann al-Manṭiq* & *Miḥakk al-Naẓar*):
|
| 35 |
+
- **Al-Ḥadd bi al-Dhātiyyāt (Real Definition by Essence)**: Entities are defined by their constitutive essential attributes, never accidental symptoms.
|
| 36 |
+
- **Dafʿ al-Dawr (Elimination of Circularity)**: Zero circular dependencies, cyclic callbacks, or self-referential ungrounded types.
|
| 37 |
+
- **Dafʿ al-Tasalsul (Elimination of Infinite Regress)**: Strictly bounded recursion, guaranteed loop termination, and timeout bounds.
|
| 38 |
+
- **ʿAdam al-Tanāquḍ (Law of Non-Contradiction)**: Mutually exclusive states are unrepresentable simultaneously.
|
| 39 |
+
2. **Al-Khalīl ibn Aḥmad al-Farāhīdī** (*Kitāb al-ʿAyn*) & **Ibn Manẓūr** (*Lisān al-ʿArab*):
|
| 40 |
+
- **Tri-Consonantal Root Decomposition & Combinatorial Safety**: Complex systems are decomposed into orthogonal, irreducible primitives anchored in authentic semantic roots (`قفل`, `حفظ`, `نقل`, `عقد`, `حسب`, `رتب`, `سلم`, `حكم`).
|
| 41 |
+
- **Exhaustive State-Space Coverage**: Zero unhandled match arms and zero silent exception swallowing (`except Exception: pass` is prohibited).
|
| 42 |
+
3. **Al-Rāghib al-Iṣfahānī** (*Al-Mufradāt fī Gharīb al-Qurʾān*):
|
| 43 |
+
- **Ontological Domain Modeling & Teleology (*Ghāyah*)**: Absolute ban on amorphous, vague identifiers (`data`, `temp`, `val`, `mgr`, `helper`). Every symbol reflects its distinct teleological purpose.
|
| 44 |
+
4. **Al-Zamakhsharī** (*Asās al-Balāghah*):
|
| 45 |
+
- **Abstraction Integrity & Rhetorical Eloquence (*Ḥaqīqah* vs *Majāz*)**: Delineating physical machine reality (memory, CPU caches, IO) from software abstractions. Zero leaky abstractions.
|
| 46 |
+
5. **Sībawayh** (*Al-Kitāb*):
|
| 47 |
+
- **Syntactic Governance (*Al-ʿĀmil wa al-Maʿmūl*)**: Strict caller-callee hierarchy, rigorous static typing, and guaranteed AST integrity.
|
| 48 |
+
|
| 49 |
+
---
|
| 50 |
+
|
| 51 |
+
## 🧠 Cognitive Chain-of-Thought Protocol (`<ayn_mantiq>`)
|
| 52 |
+
|
| 53 |
+
Before emitting production code, AynCoding-Gemma2 executes its epistemic reasoning:
|
| 54 |
+
|
| 55 |
+
```xml
|
| 56 |
+
<ayn_mantiq>
|
| 57 |
+
🏛️ AYN-ENGINE EPISTEMIC LOGIC & MORPHOLOGY REASONING:
|
| 58 |
+
- Classical Root & Morphology (الجذر والتصريف): [Root e.g. ق-ف-ل, linguistic significance]
|
| 59 |
+
- Real Definition & Essence (الحد بالذاتيات - معيار العلم للغزالي): [Essential attributes and invariants]
|
| 60 |
+
- Epistemic Fallacy Invariants (دفع الدور والتسلسل ونفي التناقض): [Circularity, regress, and contradiction guards]
|
| 61 |
+
- Lexicographical Teleology (الغاية وبلاغة التجريد - المفردات والأساس): [Pure purpose, zero vague abstractions]
|
| 62 |
+
- Syntactic Governance (العامل والمعمول - كتاب سيبويه): [Strict type contracts, caller-callee hierarchy]
|
| 63 |
+
</ayn_mantiq>
|
| 64 |
+
```
|
| 65 |
+
|
| 66 |
+
---
|
| 67 |
+
|
| 68 |
+
## 🧪 Benchmark & Dynamic Evaluation
|
| 69 |
+
|
| 70 |
+
In live benchmark evaluations using the **5-Pillar Static Epistemic Auditor**:
|
| 71 |
+
|
| 72 |
+
* **Sliding Window Rate Limiter:**
|
| 73 |
+
* **Overall Epistemic Score:** **87.8% (Grade: B+)**
|
| 74 |
+
* **AST Syntax Integrity:** 100% Valid ✅
|
| 75 |
+
* **Banned Placeholders:** 0 (Zero-Loss Complete) ✅
|
| 76 |
+
* **Asās al-Balāghah (Eloquence):** **10.0 / 10**
|
| 77 |
+
* **Kitāb al-ʿAyn (Decomposition):** **10.0 / 10**
|
| 78 |
+
* **Sībawayh (Governance):** **10.0 / 10**
|
| 79 |
+
* **Dynamic Execution:** Automatically evicted oldest items in $O(1)$ under real multi-threaded execution!
|
| 80 |
+
|
| 81 |
+
---
|
| 82 |
+
|
| 83 |
+
## 🚀 Quickstart: Running with Ollama
|
| 84 |
+
|
| 85 |
+
1. **Download the GGUF model and Modelfile:**
|
| 86 |
+
```bash
|
| 87 |
+
huggingface-cli download enver/ayncoding-gemma2-2b ayncoding-gemma2-2b.gguf Modelfile --local-dir ./ayncoding-gemma2
|
| 88 |
+
cd ayncoding-gemma2
|
| 89 |
+
```
|
| 90 |
+
|
| 91 |
+
2. **Register with Ollama:**
|
| 92 |
+
```bash
|
| 93 |
+
ollama create ayncoding-gemma2 -f Modelfile
|
| 94 |
+
```
|
| 95 |
+
|
| 96 |
+
3. **Run Inference:**
|
| 97 |
+
```bash
|
| 98 |
+
ollama run ayncoding-gemma2 "Write a thread-safe sliding window rate limiter in Python"
|
| 99 |
+
```
|
| 100 |
+
|
| 101 |
+
---
|
| 102 |
+
|
| 103 |
+
## 💻 Python (`llama-cpp-python`) Usage
|
| 104 |
+
|
| 105 |
+
```python
|
| 106 |
+
from llama_cpp import Llama
|
| 107 |
+
|
| 108 |
+
llm = Llama(
|
| 109 |
+
model_path="ayncoding-gemma2-2b.gguf",
|
| 110 |
+
n_ctx=8192,
|
| 111 |
+
n_threads=16 # NUMA optimization
|
| 112 |
+
)
|
| 113 |
+
|
| 114 |
+
output = llm.create_chat_completion(
|
| 115 |
+
messages=[
|
| 116 |
+
{"role": "user", "content": "Explain how Ghazalian logic eliminates circularity in software architecture."}
|
| 117 |
+
],
|
| 118 |
+
temperature=0.2,
|
| 119 |
+
max_tokens=4096
|
| 120 |
+
)
|
| 121 |
+
|
| 122 |
+
print(output["choices"][0]["message"]["content"])
|
| 123 |
+
```
|
| 124 |
+
|
| 125 |
+
---
|
| 126 |
+
|
| 127 |
+
## 🛠️ Included Tools in Repository
|
| 128 |
+
|
| 129 |
+
- `Modelfile`: Sovereign Ollama definition with the 5 Classical Arabic Pillars and hyperparameter tuning.
|
| 130 |
+
- `dataset/ayn_mantiq_epistemic_dataset_50.jsonl`: 50 AST-validated training samples across 10 classical domains with `<ayn_mantiq>` CoT reasoning.
|
| 131 |
+
- `tools/core/mantiq_engine.py`: Classical Logic fallacy detection engine (*Dawr*, *Tasalsul*, *Tanāquḍ*).
|
| 132 |
+
- `tools/core/mantiq_purifier.py`: Automated code sanitizer purging amorphous variables and silent bare `except:` clauses.
|
| 133 |
+
- `tools/bin/ayncode`: Sovereign CLI developer tool for code synthesis, auditing, and purification.
|
| 134 |
+
|
| 135 |
+
---
|
| 136 |
+
|
| 137 |
+
## 📜 Heritage & Citations
|
| 138 |
+
|
| 139 |
+
- Google DeepMind, *Gemma 2: Improving Open Language Models at a Practical Size*.
|
| 140 |
+
- Abū Ḥāmid al-Ghazālī, *Miʿyār al-ʿIlm fī Fann al-Manṭiq*.
|
| 141 |
+
- Al-Khalīl ibn Aḥmad al-Farāhīdī, *Kitāb al-ʿAyn*.
|
| 142 |
+
- Al-Rāghib al-Iṣfahānī, *Al-Mufradāt fī Gharīb al-Qurʾān*.
|
| 143 |
+
- Ibn Manẓūr, *Lisān al-ʿArab*.
|
| 144 |
+
- Sībawayh, *Al-Kitāb*.
|