Views
No views yet
Turn Arabic (and multilingual) legal / administrative text into a compactgraph TDMermaid knowledge graph — on-device, offline, via Ollama.
| Pull | ollama run enghamzasalem77/pyxon-txt2kg |
| Hub | ollama.com/enghamzasalem77/pyxon-txt2kg |
| Base | Qwen2.5-Coder-1.5B-Instruct |
| Method | LoRA SFT · teacher = Gemini Mermaid/KG pairs · GGUF q8_0 |
| Size | ~1.6 GB (q8_0) |
| Output | Mermaid graph TD only |
1ollama pull enghamzasalem77/pyxon-txt2kg
2ollama run enghamzasalem77/pyxon-txt2kg1Extract a SMALL knowledge graph as Mermaid only.
2First line must be: graph TD
3Max 12 nodes, include --> edges, close subgraphs with end.
4No JSON, no markdown fences.
5
6### DOCUMENT
7رقم القضية ٣٣٤٤/١٠/ق لعام ١٤٣٧ هـ
8المدعية: مؤسسة …
9المدعى عليها: وزارة التجارة …
10المحكمة الإدارية حكمت بعدم الاختصاص الولائي.1graph TD
2 subgraph أطراف
3 P1("المدعية")
4 P2("وزارة التجارة")
5 end
6 Court("المحكمة الإدارية") -->|عدم اختصاص| Case("٣٣٤٤/١٠/ق")
7 P1 --> Case
8 P2 --> Case1curl http://localhost:11434/api/chat -d '{
2 "model": "enghamzasalem77/pyxon-txt2kg",
3 "stream": false,
4 "messages": [
5 {"role": "user", "content": "### DOCUMENT\n...arabic text...\n\nOutput Mermaid graph TD only."}
6 ],
7 "options": {"temperature": 0.2, "num_ctx": 4096, "num_predict": 384}
8}'txt_to_kg.jsonl (25 Diwan cases; 22 train / 3 holdout).Qwen2.5-Coder-1.5B-Instruct (r=16, α=32, 5 epochs, seq 4096).q8_0 GGUF → Ollama Modelfile (ChatML).pyxon-txt-kg/paper/ (LaTeX).num_ctx 4096).num_predict ≤ 384 and post-validate Mermaid.