This project aims to finetune the GPTNeo 125M from Huggingface transformers on the XNLI french dataset for natural language inference task.
The XNLI dataset is also loaded from the huggingface dataset library.
The overal accuracy on the test set after 3 epochs of training on google colab is given below.