Community bonding and what follows
Community Bonding ends
So it has been two weeks since the community bonding period ended and the official coding begun. During these two weeks, I discussed with my mentors, what all is to be done under this project. That includes the feasibility of the solution, the practical implementation of our model over the pre-trained embeddings.
What is in the Pipeline?
To start us off, we plan on using pre-trained word embeddings, trained with the help of a Skipgram model from FastText on a Wikipedia dump. After that, the embeddings will be tested on the Google Analogy Test Set. This analogy test will be just to get familiar with the task that we are dealing with. The results from this test can and will also be used as the benchmark for our model. Other evaluation tasks that will be used are Question-Answering, Named Entity Extraction/Recognition, and Textual Entailment.
What will follow?
My purpose for writing this blog will be to share my progress with each and every reader. To accomplish that, I will try to push daily updates to the blog, or an update every other day about anything and everything that I do to complete the week's tasks. Following posts will be just about my ideas and my experiments with the datasets, and the approaches that I take or plan on taking. I will update all the necessary links to follow my work at the end of every post. This will include links to Github repo for the project, any gist, any Python Notebooks or the links that I refer to while trying to find a better approach.
In the following posts, you may find the information regarding all the recent activity on the project.