CUAD Legal Contract Data for Low-Token LLM Fine-Tuning

Clean JSONL training data derived from the CUAD corpus, optimized for fine-tuning large language models with a low token count. The dataset appears to be sourced from Kaggle, though the original author and specific license are not provided. Its last update date and exact size are unknown.

Use Cases

• Fine-tune language models for contract clause classification based on the CUAD corpus.
• Train models for legal question answering using structured contract data.
• Develop named entity recognition systems for legal entities and provisions in contracts.
• Optimize model performance on legal text with a focus on low-token input efficiency.

Strengths

• Data is described as 'clean' and 'optimized for low token count', suggesting preprocessing for efficient model training.
• Derived from the established CUAD (Contract Understanding Atticus Dataset) corpus, indicating a known legal data source.

Limitations

• Row count and total dataset size are unknown, which may limit suitability assessment for large-scale projects.
• Column-level documentation is absent; field semantics must be inferred after download.
• Description metadata is limited; actual data quality and completeness require manual inspection.

Provenance

Source

Derived from the CUAD (Contract Understanding Atticus Dataset) corpus.

Collection Method

Processed into clean JSONL format for LLM training.