After spending a lot of time trying to code fully connected networks last week, I started working on LSTMs and RNNs this week. Given that LSTMs are what I’ll be working with in my project, it’s important that I understand the fundamentals of how it works, even if I’m a little frustrated with how long it’s taken me to learn. I think it’s been worth it, since I’m pretty satisfied with my understanding of how they work now- in fact, I feel more confident that I can understand and talk about backpropogation, which I’ve always struggled with in particular.
Another thing I’ve been working on this week is getting more background information on my project; I was recommended a set of papers on text style transfer that I’ve been working on reading to get background on how to build my own style transfer model. The final piece that I’ve started thinking about now that I’ve started to properly work on the project started is looking at datasets- the information I use to train the model is going to affect how well it works. One of the suggestions I recieved was to look for something with a built in loading method so that I don’t have to build it myself- between that and the requirement that it has some kind of recognizable/labeled style (politeness is the current idea) I have a fairly specific idea of what to look for, especially after looking at papers and seeing the datasets that they use.