question-classification is a framework designed for training models to classify questions generated using our initial Language Model. This project originated from my early experiences in the field of Natural Language Processing (NLP) when it became apparent that there was a lack of robust evaluation methods within the community. In response, I conducted a series of experiments using the limited french dataset available at that time. The primary objectives were to expand the dataset and introduce variations to enhance the quality of questions.
When I initially embarked on this project, the available data was limited, and the goal was to improve and diversify the pool of well-formed questions. However, it's important to note that the outcomes of these early experiments were modest, especially when compared to the advancements achieved with contemporary Language Models (LLMs).
Thank you to Idris Jean-Zay GPU, whose computational resources permit to fine-tune my models.