💼 LexGen — LoRA Adapter for Legal Document Generation (Board Resolutions)

LexGen is a lightweight, fine-tuned adapter for Mistral-7B that generates legally sound Board Resolutions from structured input fields like company name, purpose, meeting date, etc.

This adapter is trained using PEFT (LoRA) for efficient and low-resource fine-tuning and deployment.

---

🧠 Model Details

Property	Value
Base Model	Mistral-7B (load separately)
Adapter Type	LoRA (via HuggingFace PEFT)
Output Format	Formal English, board resolution structure
Use Case	Auto-generation of legal Board Resolutions
Performance	High factual and syntactic consistency

⚠️ The base model (mistralai/Mistral-7B-v0.1) is not included — load it separately.

---

🚀 Example Use Case

Convert structured form data into polished board resolutions.

🛠 Usage (Python)

from transformers import AutoModelForCausalLM, AutoTokenizer
from peft import PeftModel

base_model = AutoModelForCausalLM.from_pretrained("mistralai/Mistral-7B-v0.1", device_map="auto")
tokenizer = AutoTokenizer.from_pretrained("mistralai/Mistral-7B-v0.1")

peft_model = PeftModel.from_pretrained(base_model, "./checkpoint-200")

prompt = "Generate a board resolution for approval of branch expansion for Aretoss Ltd."
inputs = tokenizer(prompt, return_tensors="pt").to(base_model.device)
outputs = peft_model.generate(**inputs, max_new_tokens=200)
print(tokenizer.decode(outputs[0], skip_special_tokens=True))

---

📦 Files Included

• adapter_config.json – LoRA adapter config
• adapter_model.safetensors – Fine-tuned adapter weights

---

🏷 Tags

generative-ai · legal · text-generation · lora · mistral · peft · llm · contract-generation

---

📄 License

Licensed under Apache 2.0. Free to use, modify, and distribute with attribution.

---

Github: https://github.com/Aryan-Jadhav3009 Linkedin: https://www.linkedin.com/in/batmann/