Leonora123/legal-assistant

Hugging Face
TEXT GENERATIONConcurrency Cost:1Model Size:3.1BQuant:BF16Ctx Length:32kPublished:May 7, 2026License:apache-2.0Architecture:Transformer Open Weights Warm

Leonora123/legal-assistant is a 3.1 billion parameter Qwen2.5-based causal language model developed by Leonora123. This model is fine-tuned for legal assistance tasks, leveraging the Qwen2.5-3B-Instruct architecture. It was trained using Unsloth and Huggingface's TRL library, optimizing for faster training. With a 32768 token context length, it is designed for processing and generating legal-specific text.

Loading preview...

Model Overview

Leonora123/legal-assistant is a 3.1 billion parameter language model developed by Leonora123, specifically fine-tuned for legal assistance applications. It is based on the Qwen2.5-3B-Instruct architecture, known for its strong performance in various language understanding and generation tasks.

Training Details

This model was efficiently trained using Unsloth, a library designed to accelerate the fine-tuning process, achieving 2x faster training speeds. The fine-tuning process also incorporated Huggingface's TRL (Transformer Reinforcement Learning) library, which is commonly used for instruction-tuning and alignment of language models. The base model, unsloth/Qwen2.5-3B-Instruct-bnb-4bit, was utilized for this specialized fine-tuning.

Key Characteristics

  • Architecture: Qwen2.5-3B-Instruct base.
  • Parameter Count: 3.1 billion parameters.
  • Context Length: Supports a substantial context window of 32768 tokens, suitable for handling lengthy legal documents.
  • Specialization: Optimized for tasks within the legal domain.

Intended Use

This model is intended for developers and researchers working on applications requiring a specialized language model for legal text processing, analysis, and generation. Its efficient training methodology makes it a practical choice for deployment in legal tech solutions.