Lemone-Embed L
Embedding model for search over French tax texts, fine-tuned from multilingual-e5-large.
What it does
Sentence-transformers model that maps French tax queries and passages to vectors for semantic search and retrieval-augmented generation. It runs in Python with the sentence-transformers library, French only, 512 tokens per input; the Code général des impôts and BOFiP are training sources. Fine-tuned on part-synthetic data from GPT-4 Turbo and Llama 3.1 70B; its limitations section is empty.
Where it fits on the map
- Company tax department › Tax Technology, Data & TransformationTax technology & datacovered
- Tax authority › Data, IT & AnalyticsData, document AI & AI modelscovered
Named as a project to start with: A benchmark to test a model on returns, and an embedding model tuned to French tax text.
Similar projects
Of the same kind, for a shared jurisdiction or tax area. Not a ranking.
- Lemone mBERT taxmBERT model adapted to French tax language, for clustering and semantic search.
- SteuerLLM28-billion-parameter language model for German tax law analysis, from the SteuerLLM paper.
- AI-Powered-Tax-OptimizationRandom Forest model and Flask API that return one optimised tax figure from nine financial inputs.Inactive since 2024Demo
Maintain this project?
Link to this page from your README, so readers find the facts, the notes and where it fits. The badge is a small SVG served from this site.
[](https://tax-oss.com/projects/lemone-embed-l/)