I'm a new PhD student. My research is within NLP and text analysis. I will be exploring both large datasets and models.
My first project will be to find computationally effective ways of estimating the uncertainty of word embeddings. The idea is to find ways that are more efficient that the current bootstrap type approach that is current general approach. To verify the results of the algorithm we will need to test it on large corpora that are to large to handle on my local machine.
The goal, if successful, is to move on to larger models. Such as transformers.
This is however just my first project as a PhD student and as I mature academically many more tasks will be come up. Having access to the appropriate resources, such as this one, I feel is crucial for my research.
Best Regards, Isac