itsalloverig/adaption-indian-legal-triage-samples-v4

This dataset is a remastered version prepared using Adaption's Adaptive Data platform.

adaption-indianlegaltriage_samples

This dataset contains prompt-completion pairs focused on triaging Indian legal matters across various domains such as consumer protection, employment, IP, and property. The completions provide structured outlines, evidence matrices, compliance checklists, and risk assessments while explicitly refusing to hallucinate citations or provide final legal advice without verified sources. Each entry emphasizes separating facts from assumptions, identifying missing documents, and outlining safe research paths for legal professionals.

Dataset size

There are 12,673 data points in this dataset. This is an instruction tuning dataset.

Quality of Remastered Dataset

The final quality is A, with a relative quality improvement of 35.7%.

Domain

• —Legal (80%)
• —Other (20%)

Language

• —English (100%)

Tone

• —Professional (50%)
• —Objective (20%)
• —Analytical (16%)

Evaluation Results

• —Quality Gains: <img src="https://proteus-prod-public.s3.us-east-1.amazonaws.com/temp/826ee1e4-4c1a-4461-afd1-1266e13418b0.png" alt="QualityGains" style="max-width: 50%; display: block; margin-left: auto; margin-right: auto;" />

• —Grade Improvement: <img src="https://proteus-prod-public.s3.us-east-1.amazonaws.com/temp/929209f4-7c7a-493c-a490-285f7a838de2.png" alt="Grade" style="max-width: 50%; display: block; margin-left: auto; margin-right: auto;" />

• —Percentile Chart: <img src="https://proteus-prod-public.s3.us-east-1.amazonaws.com/temp/fabd019e-4da6-443f-82b1-b2e3197f6593.png" alt="Percentile Chart" style="max-width: 50%; display: block; margin-left: auto; margin-right: auto;" />