Scholarlink

This work implements a Large Language Model-based approach to contract analysis that can highlight, elucidate and classify the risks present in a contract, with a focus on the Indian legal and cultural context. The methodology involves synthetic contract dataset generation using GPT-4, annotating the dataset clause-by-clause to identify potential issues, assigning a risk level to each clause and finally training our chosen model on the dataset. The model, gemma-2-9b, was trained using techniques such as Parameter Efficient Fine-Tuning and Low-Rank Adaptation that optimize computational power and memory efficiency. The experimental results, which include metrics like perplexity and prediction accuracy, calculated on the final output, show that our model is accurate and efficient at highlighting and providing descriptive explanations of the risks. This study takes a practical step forward in using Artificial Intelligence and LLMs for a fast and low-cost contract review tool.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Fine-Tuning LLMs for Risk Assessment in Indian Employment Contracts

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • V. Hamsini,
  • S. Kirti,
  • Sudarshan Srinivasan,
  • Bhaskarjyoti Das

摘要

This work implements a Large Language Model-based approach to contract analysis that can highlight, elucidate and classify the risks present in a contract, with a focus on the Indian legal and cultural context. The methodology involves synthetic contract dataset generation using GPT-4, annotating the dataset clause-by-clause to identify potential issues, assigning a risk level to each clause and finally training our chosen model on the dataset. The model, gemma-2-9b, was trained using techniques such as Parameter Efficient Fine-Tuning and Low-Rank Adaptation that optimize computational power and memory efficiency. The experimental results, which include metrics like perplexity and prediction accuracy, calculated on the final output, show that our model is accurate and efficient at highlighting and providing descriptive explanations of the risks. This study takes a practical step forward in using Artificial Intelligence and LLMs for a fast and low-cost contract review tool.