💾 Archived View for kenogo.org › blog › 20191016.gmi captured on 2023-09-28 at 15:59:49. Gemini links have been rewritten to link to archived content
⬅️ Previous capture (2023-01-29)
-=-=-=-=-=-=-
I’ve been incredibly annoyed at how it was /impossible/ to find some examples for a finished thesis, which is why I have decided to publish mine online. I got a grade of 1,1 on the thesis (1,0 by the primary supervisor and 1,3 by the secondary supervisor). The full source code is available on my GitLab.
I wrote the thesis over the course of about a year. Most of the time was spent doing absolutely nothing for the thesis except some small tests here and there. At some point, I officially registered my thesis with the registrar’s office, which meant I had exactly four months to complete it starting at that point. Basically no work has been done up until this moment and I didn’t actually start working on it until about two months later.
All the calculations seemed to be complete about one month before the deadline, so I started writing the thesis about then. I was only missing some final calculations, where I ran into problems because my network was performing much better than it was supposed to, leading to problems in the optimization of the network’s significance. Eight days before the deadline, my tutor spotted a bug in my code that was the cause of the issue [1]. Sadly, this bug was in the very first step of formatting the training data for the network. This meant that /all/ of my calculations and figures were utterly meaningless. I had to do it all over again.
That’s where I began to panic, but luckily I had a pretty tidy codebase that saved my ass. All I had to do was set up a script that would rerun all of my python scripts and notify me about their completion (or potential errors) via email. I let it run on the university server and all calculations were complete in the matter of about 40 hours. In the meantime, I wrote the introductory section of the thesis.
Then all I had to do was rewrite the sections as to reflect on the new (correct) results, which was sometimes just a matter of exchanging a few words and numbers and sometimes a more extensive rewrite. I was done with this just two days before the deadline [2]. The following day, I did extensive checking for mistakes of any kind, including giving the thesis to my brother to proofread.
After this was done, I only had to print the thesis in a copyshop and hand it in. I had a terrible experience at the copyshop and had to stay there for two hours until I got an acceptable result. It still didn’t look nice, but it was good enough. I don’t recommend printing your thesis the day of the deadline because of the possibility of such problems.
I’ll keep this one short. I wasn’t feeling too well in the weeks leading up to my defense, so I was utterly unprepared. My presentation was put together the day before the defense and I wasn’t able to answer many of the questions my supervisors asked me. I still got a 1,3 on the defense.
Considering how I did most of the work for writing my thesis in just one week, I didn’t expect a grade nearly as good as I got. This applies even more to the defense. I also don’t know anyone who failed their Bachelor’s thesis. All of this makes me believe that it is more of a symbolical thing and you shouldn’t stress yourself out too much about it.
In my four years studying physics, I can comfortably say that the thesis was about the /easiest/ thing I have done. All of the courses I took and their exercises and exams were much more important than the thesis. If you’ve passed the exams to qualify for a Bachelor’s degree, the thesis will probably be easy for you.
Chances are that you will be introduced to some topics that are a bit above your level of knowledge and you will run into some walls. This is normal and you have a tutor to help you out for that reason. The supervisors also don’t expect you to be on the level of an actual physicist. So my advice is: Don’t panic, you got this and you’re probably going to be graded much higher than you’re currently imagining.