LEGI: 1-10 Million French Legislative Records with Semantic Embeddings

French Consolidated Legislation (LEGI) dataset containing between 1 million and 10 million records of French law and codes. Produced by AgentPublic and updated as of March 2026, it provides a semantic-ready version of official government open data from Legifrance.

Use Cases

• Semantic retrieval of French legal articles using the pre-computed embeddings
• Topic modeling of French legislation based on 'codes' and 'law' tags
• Training or fine-tuning LLMs on formal French legal terminology

Strengths

• Contains between 1,000,000 and 10,000,000 records
• Includes pre-computed embeddings for semantic search
• Uses official Legifrance open data as the primary source

Limitations

• Metadata does not specify the embedding model used for the vectors
• Restricted to French language text
• Schema details are not explicitly listed in the summary

Provenance

Source

Legifrance (French government legal portal)

Collection Method

Scraped and processed from official open-data sources

Freshness

Last updated March 2026.

Geography

France