Mistral-RRC
| Property | Value |
|---|---|
| Parameter Count | 7.24B |
| License | MIT |
| Base Model | Mistral-7B-v0.1 |
| Paper | AI for Scaling Legal Reform |
| Performance | F1 Score: 0.997 |
What is mistral-rrc?
Mistral-RRC is a specialized language model fine-tuned for detecting and extracting racial covenants from property deed documents. Based on the Mistral-7B architecture, this model serves a crucial role in legal reform by identifying discriminatory language in real estate records, particularly supporting initiatives like California's AB 1466.
Implementation Details
The model leverages the Mistral-7B architecture with FP16 precision and has been fine-tuned on 3,801 annotated deed pages from eight different U.S. counties. The training dataset was carefully balanced, with 78.6% of documents containing racial covenants to ensure robust detection capabilities.
- Trained on diverse jurisdictional data for improved generalization
- Implements sophisticated text classification and extraction capabilities
- Achieves 100% precision and 99.4% recall on test datasets
Core Capabilities
- Accurate detection of racial covenants in property deeds
- Extraction of relevant text passages with error correction
- Robust handling of OCR artifacts and historical document variations
- Support for automated document prioritization in legal review workflows
Frequently Asked Questions
Q: What makes this model unique?
The model's specialization in detecting racial covenants, combined with its high accuracy (99.7% F1 score) and ability to handle historical document complexities, makes it a powerful tool for legal reform initiatives. Its ability to both detect and correct OCR errors in extracted text is particularly valuable for processing historical documents.
Q: What are the recommended use cases?
The model is specifically designed for legal professionals and government entities working on identifying and redacting discriminatory language in property deeds. It's particularly useful for large-scale document review processes and can significantly reduce the manual effort required in covenant identification.
