--- base_model: mufeedh28/dictalm2-israeli-law-instruct-merged tags: - gguf - mistral - legal - hebrew - israel - law - ollama license: apache-2.0 language: - he pipeline_tag: text-generation ---
# DictaLM 2.0 — Israeli Law Chat (GGUF) ### Run the Hebrew legal chatbot locally with Ollama **F16 full precision** | **~14.5 GB** | **Requires 16GB+ RAM**
--- ## Quick Start ```bash ollama run hf.co/mufeedh28/dictalm2-israeli-law-GGUF ``` Then ask questions in Hebrew: ``` >>> מהן זכויות השוכר לפי חוק השכירות? >>> מה קורה אם מעסיק לא משלם פיצויי פיטורים? >>> האם ניתן לערער על החלטת בית משפט השלום? ``` ## About This is the **F16 (full precision) GGUF** version of [DictaLM 2.0 — Israeli Law Chat](https://huggingface.co/mufeedh28/dictalm2-israeli-law-instruct-merged), a 7B Hebrew legal chatbot fine-tuned on 140K+ Israeli legal documents and 7,291 Q&A pairs. For full model details, training data, and usage examples, see the [main model card](https://huggingface.co/mufeedh28/dictalm2-israeli-law-instruct-merged). ## File Details | File | Precision | Size | Quality | |------|:---------:|:----:|:-------:| | `dictalm2-israeli-law.F16.gguf` | F16 | ~14.5 GB | Full precision — no quality loss | ## Requirements - [Ollama](https://ollama.com/) installed - 16 GB+ RAM (GPU or CPU) ## Alternative Usage ### With llama.cpp directly ```bash ./llama-cli -m dictalm2-israeli-law.F16.gguf -p "[INST] מהן זכויות העובד בפיטורים? [/INST]" -n 512 ``` ### With llama-cpp-python ```python from llama_cpp import Llama llm = Llama(model_path="dictalm2-israeli-law.F16.gguf", n_ctx=2048) output = llm("[INST] מהן זכויות העובד בפיטורים? [/INST]", max_tokens=512, temperature=0.7) print(output["choices"][0]["text"]) ``` --- > **Disclaimer:** This model may produce inaccurate legal information. Do not use as a substitute for professional legal advice. Made by [Mufeed Hammud](https://www.linkedin.com/in/mufeed-hammud-a41b84245) | [Full Model](https://huggingface.co/mufeedh28/dictalm2-israeli-law-instruct-merged) | [GitHub](https://github.com/mofeed28/israeli-law-llm)