Welcome back everyone!
It’s clear to me that I have to get back in the WHV saddle, but preaching sure is easier than practising, and so I thought it might be a good idea to try and bullet-list myself back into it.
(When I say “bullet-list” I mean “helpfully named hierarchical sections”, because Emacs tends to frown quite disapprovingly at mere bullet lists. Here I am after the intention of the bullet-list, namely to just get on with it.)
Figure 1: Manege from a late summer stroll in NL. Smells, sounds, weather took me back so hard.
It is important to mention that this post attempts to cover time up to Wednesday, August 23, 2023.
Running
My running habit has been steadily returning to its pre-op state, both in terms of fitness but, more importantly, in terms of joy.
Also, my sixth pair of Altra Escalantes (version 3.0) recently arrived. I wanted black ones, as I usually recycle old Altras after their recommended 700km or so into super comfortable office shoes, but my supplier was out of stock and so I ended up with these surprisingly attractive blue and red ones:
In more good news, it turns out that it’s entirely OK if you spend all of your minimum of 2.5 weekly hours of exercise during the weekend.
Personally, I usually manage to fit in two shorter runs during the week, and a longer run during the weekend, for a desired total of 25km per week, but I have always wondered if it was in fact required to exercise on every day of the week for maximal health benefit.
P.S. My main motivation is the unqualified joy of the run and not so much the health benefits, although the benefits sure are nice!
P.P.S. 25km per week was my resolution for 2023, but that plan was quite rudely torpedoed by my prostate cancer adventure a few months ago, so now I just try to make it week by week.
GenAI
For the past months, I’ve been going deeper and deeper into generative AI, and more specifically large language models (LLMs), down to the nuts and bolts and let’s-get-Llama-2-70-billion-quantized-and-running-on-my-PC levels.
Here are a few things that I really liked learning about:
- Although it has indeed been a longer journey, the current GenAI explosion / hype seems to have been ignited in the third quarter of 2022, with the open source release of Stable Diffusion in August, and the birth of ChatGPT in November.
- For the largest part, the most powerful LLMs have been trained, on hundreds
of billions to trillions of words, how to predict the next / missing word of
an incomplete document. After enough of this, they appear to gain some sort
of reasoning capability, including being able to answer theory of mind
questions.
- There are striking similarities to how humans interpret speech.
- Almost all of the currently famous models (Stable Diffusion, Whisper, all current LLMs) are based on a specific type of neural network called a Transformer, introduced in the paper Attention Is All You Need, published by Ashish Vaswani and fellow Google researchers in 2017.
For many more details about these, and other hopefully interesting observations, please come to one of my GenAI talks!
GenAI developments are continuing at a truly impressive pace, and now I hope that we (personally) can also start showing some practical and useful applications!
P.S. Read the brilliant Lyn Alden’s latest newsletter Six AI Themes to Consider in which she discusses how AIs (actually LLMs) are already starting to use bitcoin (lightning) to pay for services (soon other AIs and also humans) in order to complete their tasks.
Lowlands 2023
One could say that the WHV series owes its life to a certain 3-day festival in the Netherlands as it was born in the sweet afterglow of the 2009 edition.
However, due to the strict WHALLSALL privacy policy, you will find only pleasant echoes of the experience in a number of posts on this blog.
This time it felt even more special, because it’s the first time that I was able to go since all the way back in 2019.
I was pleasantly shocked by the warm familiarity of my other home, starting all the way from the airport in Cape Town, joining hundreds of Dutchies returning home after their vacations in SA, to arriving in NL and letting that lovely language permeate my neural network, all through the festival chaos and sometimes extremely close proximity to tens of thousands of locals, to walking and cycling under the late summer sun, to spending the highest possible quality time with my friends there.
Again I am left overflowing with gratitude.
Thank you other home, and eternal thank you friends!
I am a function
Back in WHV #70 from 2014, based on some reading about the turnover of matter in humans, I came to the following conclusion:
Now I see us all as nothing more than the continuously changing patterns that form briefly in the grains of sand of the universe. Winds blow grains of sand around, now taking part in one pattern and then in another. At some point, your body has contained parts of your best friend, and of your worst enemy. The patterns change continually; sometimes they fade away, and sometimes new patterns emerge.
In the intervening years, reader KvG first introduced me to Alan Watts, noting that they saw similar themes in our writing.
I am wording it carefully here, as I obviously can’t really compare my backyard philosophical attempts to the work of Alan Watts, which I have now come to know better.
That story was required to bring us to this, the concluding quote of this post:
You are a function of what the whole universe is doing in the same way that a wave is a function of what the whole ocean is doing. – Alan Watts
I really love that.