Legalkit: French Legal Text Datasets for Sentence Embeddings Training

Labeled data prepared for training sentence embeddings models in the context of French law. The dataset was created by author louisbrulenaudet and last updated on June 26,  2024. Its labeling process utilizes the LLaMA-3-70B model through a structured workflow to enhance label quality.

Use Cases

• Training sentence embedding models based on labeled French legal text data.
• Fine-tuning NLP models for legal document classification based on the described sentence labels.
• Developing semantic search systems for French legal corpora based on the prepared embeddings.
• Benchmarking model performance on French legal language understanding tasks.

Strengths

• Labels were generated using the LLaMA-3-70B model through a structured workflow.
• Data is specifically prepared for training sentence embeddings in the French legal domain.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• Description metadata is limited; actual data quality requires manual inspection after download.

Provenance

Source

huggingface

Collection Method

Labeling process utilizes the LLaMA-3-70B model through a structured workflow.

Time Range

null

Freshness

Last updated 2024-06-26 21:40:22; freshness should be verified.

Geography

France (implied by focus on French law)