Russian court documents contain fields for training and testing LLM-based extraction tasks. The dataset includes CSV files and grouped JSONL files with expert mean scores. Its origin and size are unknown.
Use Cases
- Benchmark LLM extraction accuracy on Russian court fields using expert mean scores.
- Train models to identify and extract specific legal fields from grouped JSONL documents.
- Test model generalization on court document data split into train and test CSV files.
- Compare model outputs against expert scores for validation.
- Analyze performance of LLMs as judges for legal text extraction tasks.
Strengths
- Includes expert mean scores for validation.
- Provides separate train and test CSV files for structured evaluation.
Limitations
- Unknown row count and dataset size.
- Unknown geographic scope beyond Russian origin.
- Unknown data freshness and update frequency.
Provenance
- Geography
- Russia