I am a 2nd year PhD. student between Facebook AI Research and Inria Saclay, in France, working at the interplay between machine learning and neuroscience. Current deep language models have shown to acheive close to human performance at various language tasks such as text categorisation, generation, summarization etc. Do these models process language similar to the human brain? I investigate deep language models, not by analysing their generations/outputs, but by directly comparing their internal representations (activations) to those of the brain (as measured with fMRI and MEG). The goal are to 1) better understand the neural bases of language (neuroscience) and 2) improve current deep learning algorithms (ML).