This week we knew we had work to do. We have submitted the State of Art part of the article last weekend, as well as the “Methods used” topic so we can have feedback from the teachers regarding this topic. For this week, we continued to improve our article by finishing the “Data Description and Exploration” topic, as well as continuing to explain the methods we are going to use and the results obtained. Here’s an example of some graphics we are doing to analyze and interpret our dataset, as it describes the most common words used by the sincere and insincere questions.
Regarding the code, we have also had good improvements. We fixed the issue we had last week with TensorFlow by using a GPU. We have also trained our first models and explored the data as mentioned before.
Now we are training all the model options that we have with the complete train and with the train divided into 70/30.
After finishing this training, we will choose the one with the best accuracy and export it so that we can integrate it into our interface.