Portrait de Nour Shaheen

Nour Shaheen

Représentant du laboratoire
Maitrise de recherche
Superviseur⋅e principal⋅e
Amine Mhedhbi
Co-supervisor
Sarath Chandar
Sujets de recherche
Apprentissage profond
Modèles de fondation
Traitement du langage naturel
Site web
Google Scholar
GitHub

Biographie

Nour est en deuxième année de maîtrise à Polytechnique Montréal, sous la direction des professeurs Amine Mhedhbi et Sarath Chandar. Ses recherches portent sur les modèles de fondation pour données tabulaires et les méthodes de fusion de modèles (model merging). Elle est passionnée par la science, le bon café et l'idée de travailler dans un environnement où les gens ont sincèrement du plaisir à être présents.

Publications

Towards Optimizing SQL Generation via LLM Routing
Mohammadhossein Malekpour
Nour Shaheen
Foutse Khomh
Amine Mhedhbi
Text-to-SQL enables users to interact with databases through natural language, simplifying access to structured data. Although highly capabl… (voir plus)e large language models (LLMs) achieve strong accuracy for complex queries, they incur unnecessary latency and dollar cost for simpler ones. In this paper, we introduce the first LLM routing approach for Text-to-SQL, which dynamically selects the most cost-effective LLM capable of generating accurate SQL for each query. We present two routing strategies (score- and classification-based) that achieve accuracy comparable to the most capable LLM while reducing costs. We design the routers for ease of training and efficient inference. In our experiments, we highlight a practical and explainable accuracy-cost trade-off on the BIRD dataset.
2024-11-06
ArXiv (prépublication)
doi.org
arxiv.org
Towards Optimizing SQL Generation via LLM Routing
Mohammadhossein Malekpour
Nour Shaheen
Foutse Khomh
Amine Mhedhbi
Text-to-SQL enables users to interact with databases through natural language, simplifying access to structured data. Although highly capabl… (voir plus)e large language models (LLMs) achieve strong accuracy for complex queries, they incur unnecessary latency and dollar cost for simpler ones. In this paper, we introduce the first LLM routing approach for Text-to-SQL, which dynamically selects the most cost-effective LLM capable of generating accurate SQL for each query. We present two routing strategies (score- and classification-based) that achieve accuracy comparable to the most capable LLM while reducing costs. We design the routers for ease of training and efficient inference. In our experiments, we highlight a practical and explainable accuracy-cost trade-off on the BIRD dataset.
2024-11-06
ArXiv (prépublication)
doi.org
arxiv.org