If you’re a Cuban salsa leader, I’m sure this has happened to you before: you’re at a salsa social and ask a woman to dance, to which she responds she has never actually danced salsa (for example, because she is a bachatera), or alternatively, she dances LA style. You could just tell her “Nevermind…“, but, if she tells you she’s open to it, you could teach her, like a man. How will she ever catch up on the dozens of hours of classes you’ve had? Here’s the secret: all follower movement can be reduced to a very small set of prerequisite patterns. I’m about to give you a script I’ve used many times over to teach followers from zero in just 15 minutes.

In my first article about casino (a.k.a. Cuban salsa), I started by describing the three main positions of the couple (abierta, cerrada, caída, a.k.a. open, closed, perpendicular) and organised the rest of the article into a taxonomy of figures starting in a certain position and ending in a certain position. I’ve noticed, however, that two of the three positions, namely abierta and caída, can look different depending on the dance school. Since doing one differently often goes hand-in-hand with doing the other differently, there really exist entire positional systems for dancing casino, which nobody ever points out. For beginners trying to look up basic figures, this difference is not trivial, so let’s point out two such systems.

My first series of casino classes recently ended, just in time for the holidays. However, along with some other outlier students, I didn’t want to lose momentum over Christmas break, so I started venturing into the wild trying to find figures within reach to self-teach. Along the way, I started to see patterns between all the figures I saw appear most, and I realised that I could double the amount of figures I knew just by understanding three new phrases: one phrase to learn vacilalas, two phrases to learn sencillas. And it worked.

In the last 2 months, I’ve begun taking Cuban salsa classes at the local designated club. I’ve gotten so hooked that I genuinely can’t remember what occupied my thoughts and free time in the 23.75 years prior to this, and of course, like all excellent students, I’ve been taking notes.

Here’s a fun fact about me: I graduated with the joint title of engineer (ir.) and master of science (MSc) in computer science, without anyone ever teaching me about sorting algorithms. It was only a couple months after I was hired as a PhD student that I taught myself sorting algorithms, and that was because I had to teach them to bachelor students in turn the week after.
Inevitably, while studying other data structures, I also encountered the concept of amortisation. I vaguely remembered coming across its Wikipedia page once and taking away that it meant something like “proving that something is fast via black magic”. And yet, it really is not that deep.

“Why would you even care about my future well-being? Why would you not just block me everywhere and pretend I no longer exist? My well-being has no benefit to you. I could die and it wouldn’t affect you.” This is an approximation of what my first girlfriend conveyed to me when I parted ways with her. It’d take me a year to unpack that response.

There are (at least) two claims that have never made sense to me ever since I heard them. One is that double-entry bookkeeping, where you track the capital inside a company by keeping two lists of money amounts that add up to the same total, is actually useful and not a formality that is 50% redundant. The other is that fractional-reserve banking (FRB), whereby a bank is not forced to have at least as much cash on hand as the combined total of all account balances open at that bank, allows banks to create cycles of infinite money. The former is true but hard to believe at first. The latter is false but hard to disbelieve at first. Both become apparent when considering the typical claim about FBR through the lens of double-entry bookkeeping.

The promise of DeBERTa is that it does away with absolute positional embeddings added at the start of the transformer encoder, and replaces it instead with an attention mechanism that takes into account the relative distance between tokens when doing attention, in every head of every layer. Since absolute positional embeddings are the only part of a transformer whose dimensionality is related to the length of the input, would taking it out mean DeBERTa can process an infinite amount of tokens at once? Let’s analyse.

The famous roberta-base HuggingFace checkpoint is a serialised version of a RobertaForMaskedLM model, consisting of a roberta field and an lm_head field. Yet, despite this, you can still call .from_pretrained("roberta-base") on RobertaForTokenClassification and get an object that has a roberta field with exactly the checkpoint’s roberta weights, but a head with a different architecture and randomly initialised weights. Even more strikingly, you can call .from_pretrained("roberta-base") on RobertaModel, which is what normally sits inside that roberta field and consists of the fields embeddings and encoder, and somehow it can still match all the relevant checkpoint weights to the correct fields. Ever wondered how that’s done? Here’s how.

I recently watched a video online of someone complaining about the existence of millionaires and billionaires. I commented with the realistic and optimistic message that anyone has the ability to become a multi-millionaire, even if just through supplying a silly backend commodity like toilet rolls. Then, all hell broke loose.