|Ben Trevett||Home | GitHub | Twitter|
Why am I writing this? I don’t write a lot, and I’m always reading about how you should write a lot because it’s good for you, both in terms of learning how to write well and getting ideas out of your head and on to a page in order to explore your own thoughts and interpretations of whatever you’re thinking of. I’ve tried writing multiple times before, and it never took off, so why do I think this time will be different? To be honest, I don’t, but I think I’m considerably better at writing than I have been before, though I still don’t think I’m great. I was mostly inspired to start writing after reading this article, which also reminded me to read Anna Weiner’s memoir at some point, and I like the idea of being able to write as a hobby. I don’t care about being able to generate money from it, which seems impossible to do on substack unless you already have an existing audience anyway, or even being good at it. The objective is to have somewhere to spew out whatever is on my mind at the time.
There aren’t any concrete topics in mind I’d like to write about. I’d like them to be a mixture of articles about specific topics and also some which are just a lot of rambling. Some potential topics are in the realm of machine learning/deep learning/natural language processing/reinforcement learning/etc. because that’s what I spend all my time doing. I wrote a few lengthy tutorials on how to do some NLP tasks and would like to write shorter tutorials. So, a lot of the writing won’t just be articles on a website but will be in Jupyter notebooks as I’m a big fan of literate programming.
I’d also like these rambling bits of writing to act as a sort of newsletter, but without being an actual newsletter. I want to just post them on my website with no expectation for anyone to read them. Their content will be life updates and links to things I’ve found interesting, which is what most newsletters contain nowadays anyway. I also want the articles to be relatively short, compared to most substack articles, as it allows for a quicker turnaround and means fewer opportunities to mess something up.
Anyway, about me: my life is currently in a state of flux as I’m finishing off my PhD and will start transitioning in to full-time employment. As much as I enjoyed certain aspects of my PhD, I am looking forward to employment, even the parts that people generally seem to dislike (open-plan offices, office politics, water cooler conversations) not that I think these are good things, but I think they’re funny. I like it when people tell me a story about some co-worker they dislike, and now I’ll also be able to reply with my own anecdote.
2020 was, overall, a decent year for me, mostly because I’m fortunate enough to be able to work fully remotely and spend lockdown with friends and then family. I was able to visit Munich in September whilst things were relatively calm before they picked up again, at least I can now say I’ve travelled during a global pandemic, and I managed to beat my reading goal of twelve books by reading twenty books. My favorite book, by far, was A Short Stay in Hell by Steven Peck.
For 2021 my goals are:
The main ones are obviously finishing up my PhD. I’d also like to publish another paper, but I don’t mind if it’s relevant to my PhD or not. As I’m finishing up my PhD I’ll be getting a job and would ideally like to find one that is outside the UK, preferably Europe, so I can move there. I’d like the tutorials to be considerably shorter than my previous ones, ideally a single Jupyter notebook each. My plan is to finish one I am writing about machine learning optimization algorithms, re-write one I did about bandit algorithms and then write another about DQN and all the additions to it. Ideally I’d publish something once a week, but I know that’s optimistic, so twice a month is an easier target. The books should mostly be fiction as I’ve really gone off non-fiction books this year. Finally, I’ve not read a textbook cover-to-cover in a while and mostly use them for references, so I’d like to read at least two all the way through. This might be a bit of a stretch depending on the length of the textbook so this might be downgraded to one at some point.
That’s enough content for the first post, we’ll see if I ever post again. If not, see you in 2022!