This is our code for the participation in the GEM shared task 2024. We participate only in the D2T task. The system, based on Mistral, a recent LLM, combines LLMs with traditional NLG techniques to mitigate inadequacies and hallucinations. Our pipeline process involves splitting triples, generating verbalizations, and aggregating the results.
