Paris Paradise
My time in Paris was one of the most amazing things I've ever experienced. My only obligation was my internship, which required me to be in the office sometime during the week. Outside of that, I could do whatever I wanted whenever I wanted without worry of interactions or interruptions. The freedom and independence felt fabulous.
Shaun and KK came to Paris and we did more workshops. The first series was based on a progression of maths problems. During the first workshop, Shaun said it would be like a gym for the mind. University exercises muscles for an artificial type of problem-solving. The workshops train the right ones. I didn't feel sore for days afterwards but realised how weak I was outside of what I'd been taught to do during school and my degree.
We continued, moving onto Fosbury flops - unintuitive manoeuvres that seem scary at first but with practice, allow you to jump higher than your intuitive approach. The workshops didn't get easier, I just became more used to this feeling of discomfort.
Shaun and KK were in Airbnbs scattered throughout central Paris. Each one felt distinctive but I could always expect large whiteboards, whiteboard pens, an assortment of Apple devices and pink toilet paper. I was interning so meeting up on weekends was easiest but otherwise, I'd arrive at work early so that I could leave early and continue doing workshops in the evenings. As long as I didn't do two workshops on consecutive days, I could still get enough sleep and run. And I could take what I'd been developing in the workshops and apply them to my internship to see incredible results. One time, when some attempts to improve the performance of a transformer failed, my natural response was to make the model one layer deeper and rerun. However, I applied what I'd practised in the workshop, and decided it would be better to try a different way of representing the training data.
On Bastille Day, I had the day off. I wanted to continue workshops and Shaun wanted to see the Louvre. The three of us admired the works of art before heading back to their place for some more maths.
By the end of the series, I'd invented the tools to solve various progressions of the problem, rediscovering parts of maths that were incredibly beautiful and powerful because of how deeply I understood them. Solving the problems was not the goal of the workshop but being fully immersed in the process. It was a long and hard journey but reaching the summit was a wonderful feeling. It was the first time I'd ever invented or discovered anything substantial. Before it felt like only geniuses could do that and if I wanted to, I would have to know all the related content first. In the workshop, I discovered maths that was new to me without any subject-specific knowledge.
After the maths workshops, we did some software engineering ones. These felt more comfortable but they went somewhere much deeper. I had been working as a professional software engineer for over 3 years when we did the workshops. I was a Cambridge computer science student. I was very confident in my coding and problem-solving abilities. However, I struggled with many of the simple problems we went through. They had been designed that way and it came to my attention how much room I had for growth.
The maths workshop had focused on many things but to personalise it for me, Shaun had focussed on my tendency to rush. Then in the software engineering workshops, my views about something that I was incredibly confident about were challenged. These things had a lasting impact on me and forced me to reflect deeply.
During the final workshops, we dived into some of these reflections. These were the things that truly mattered. While the technical skills were the reasons I started, the reflections outside the technical domain were how I developed the most. The workshops didn't end there: they had merely equipped me with some skills that I could practise to keep improving and developing. They had bootstrapped me so that I could continue to grow.
After the workshops, Shaun and KK left Paris. I continued my internship but with completely different ways of interacting with the problems I was tackling. I discovered a new way of pretraining models, which was significantly more effective than what we'd tried previously. I started different side projects and approached them in new ways. By going back to the basics of neural networks, I could understand the tricks that I had been blindly mimicking and apply them more deliberately. How I interacted with others changed, I put much less effort into group socialising and focused on engaging deeply with individuals. My passion for running felt new. I started to focus much more on how I performed during races and training instead of comparing myself to the people around me. My relationship with myself had been altered. I didn't need to do things to meet or surpass other people's expectations and could start engaging with the things I wanted for myself. This was only the start of my personal growth and involvement with the project.