Azzindani/Qwen2.5_1.5B_IT_ID_Legal

TEXT GENERATIONConcurrency Cost:1Model Size:1.5BQuant:BF16Ctx Length:32kPublished:May 11, 2025License:apache-2.0Architecture:Transformer Open Weights Cold

Azzindani/Qwen2.5_1.5B_IT_ID_Legal is a 1.5 billion parameter instruction-tuned Qwen 2.5 model, fine-tuned by Azzindani using Group Relative Policy Optimization (GRPO). This model specializes in Indonesian legal question-answering, enhancing structured thinking and logical flow for legal reasoning tasks. It is designed to provide better structured answers and greater consistency in domain-specific legal contexts within Bahasa Indonesia.

Loading preview...

Model Overview

Azzindani/Qwen2.5_1.5B_IT_ID_Legal is a specialized 1.5 billion parameter instruction-tuned model based on Qwen 2.5, developed by Azzindani. It has been fine-tuned using Group Relative Policy Optimization (GRPO) on a custom Indonesian Legal Q&A Dataset. The primary goal of this fine-tuning is to significantly enhance the model's reasoning and structured thinking capabilities specifically for legal question-answering tasks in Bahasa Indonesia.

Key Capabilities & Features

  • Domain-Specific Expertise: Optimized for the Indonesian legal domain, focusing on Q&A formats.
  • Enhanced Structured Thinking: Utilizes GRPO to promote a step-by-step reasoning process, mimicking how legal professionals approach cases.
  • Improved Logical Flow: GRPO encourages better structured answers and greater consistency in domain-specific reasoning by grouping samples by difficulty or topic.
  • Language: Fully focused on Bahasa Indonesia.

Training Methodology

The model leverages Group Relative Policy Optimization (GRPO), an advanced reinforcement learning technique. GRPO groups training samples by difficulty or topic, optimizing model outputs within these group contexts to achieve structured and relative improvements rather than just raw accuracy. This method helps the model to:

  1. Understand the legal context.
  2. Identify relevant laws.
  3. Apply reasoning with facts.
  4. Summarize legal conclusions clearly.

Ideal Use Cases

This model is particularly well-suited for applications requiring precise and structured legal reasoning in an Indonesian context, such as:

  • Law education tools.
  • Legal chatbot assistants.
  • Preparation for Indonesian legal examinations.

Finetunes
1 models