Hesam760/persian-water-regulations-qwen2.5-1.5b

TEXT GENERATIONPricing:Input $0.04 / Cached $0.008 / Output $0.08Concurrent Unit Cost:1Model Size:1.5BQuant:BF16Context Size:32kTool Calling:SupportedPublished:Aug 5, 2026License:apache-2.0Architecture:Transformer Open Weights Featherless Exclusive Cold

Hesam760/persian-water-regulations-qwen2.5-1.5b is a 1.5 billion parameter Qwen2.5-based language model fine-tuned for the Persian language. It specializes in answering questions related to the "Operational By-law and General Conditions of Water and Wastewater Tariffs" with a context length of 32768 tokens. This model is specifically adapted for regulatory question answering within the Persian water regulations domain, demonstrating domain adaptation for specialized legal texts.

Loading preview...

Model Overview

This model, Hesam760/persian-water-regulations-qwen2.5-1.5b, is a specialized Persian language model built upon the Qwen/Qwen2.5-1.5B-Instruct base. It has been fine-tuned using QLoRA with 4-bit NF4 quantization to provide domain-specific question-answering capabilities.

Key Capabilities

  • Domain-Adapted Q&A: Specifically designed to answer questions about the "Operational By-law and General Conditions of Water and Wastewater Tariffs" in Persian.
  • Supervised Fine-Tuning: Trained on 274 records of conversational prompt-completion JSONL data, ensuring relevance to the target domain.
  • Improved Performance: Achieves significantly higher Token F1 (0.3623 vs 0.2044) and ROUGE-L F1 (0.3196 vs 0.1670) metrics compared to its base model on validation data for this specific task.

Intended Use Cases

  • Demonstration of Persian Domain Adaptation: Showcasing how LLMs can be tailored for specific language and subject matter.
  • Regulatory Question Answering: Providing quick access to information within the specified water regulations.
  • Educational and Research Use: A valuable tool for studying legal texts and language model fine-tuning techniques.

Limitations

It's crucial to note that this model is trained on a small, single-document dataset and may reproduce outdated or incorrect provisions. It is not a legal authority and should not be used as a substitute for official regulations or professional legal advice. Manual review is always required for authoritative use.

Finetunes
1 models