dcl

Regression targets where extrapolation may be needed. Decision tree methods cannot extrapolate, the predictions are have to be a mean of a subgroup of the data.

Consider: Predicting how much a customer might pay by end of month, with information we have at the start of the month.

In this example, if a customer had a record $10m of open invoices due by EoM and the largest payment amount received in prior months of $5m, the decision tree cannot possibly predict the payment amount will be ~$10m, even when the best feature indicates the payment will be $10m.

There are some hacks/techniques which can maybe reduce this issue, but they don't always work.

This is really good

How many people would recall goatse every time they saw it though?

In 2014, I managed to get a fairly good job at a large institutional bank, without a formal interview process because the person hiring (who would be my boss) liked a blog I had written on quant sports betting during my PhD.

I use the word 'plot'. Graph is overloaded in the maths/stats/comp sci/ML/etc space.

It's been that way for years, but with data-science and machine learning.

Ahh yes this is true. I guess this is a useful property.

I've never understood how polyester become the defacto choice of fitness/sports/performance clothing. I'm guessing its cheaper to manufacture, but they can still get very expensive.

I made the mistake of buying a bunch of this stuff a while back and I can't bring myself to wear it. I feel like it makes me sweat far more and it stinks very quickly.

I have far less of a problem wearing cotton and and a great experience wearing merino wool.

Why?

How does anyone start a company these days and not accidentally infringe on a bunch of nonsense patents?

A PhD Student's Perspective... ~20 authors.

I thought exactly the same thing.

You got me.

It's a reasonable way to look at it given that's how pretty much all 'deployed' versions of LLM's work?

Exactly.

But also, not just ASIOF is in the training set, but presumably lots of discussion about it and all the interesting events in the book.

Similarly, consider a series like A Song of Ice and Fire. A human reader is still consciously aware of (and waiting for) the answers to questions raised in the very first book. This is millions of tokens ago, and that's if our brains turn off when not reading the books.

I think this highlights a hurdle on the path to more human-like AGI. We keep track of so much stuff for very long periods of time, albeit perhaps with some loss of fidelity.

My guess is that there will need to be an advancement or two before we can can get an AI to read all of the ASIOF books so far and ask "What really happened at the Tower of Joy?"

I've only ever used SQL and relational databases. What are the use cases of Key-value stores? What's the canonical example of where they are a clearly the right solution?

I like to send these articles to my team to put our small mistakes in to scale.

I don't understand how so much money has been poured in to these companies?

I get the why the techniques are suitable, but I just assumed who ever wants to do this kind of retrieval can probably implement a suitable Approx. NN library themselves?

Especially so, because getting good embeddings is the hard part, not the search?

It's great you've found techniques that work for you and you are correct the science is fuzzy...

From https://en.wikipedia.org/wiki/Carol_Dweck#Criticism

> Timothy Bates, a psychology professor at the University of Edinburgh, has been trying for several years to replicate Dweck's findings, each time without success, and his colleagues haven't been able to either.[23]

The statistics in Dweck's papers also fail the GRIM test which is a potential indicator of fraud (fake/false statistical values).

Would it be remotely possible to make a law to prevent this 'voluntary' deification of international despots?

2 years ago I started a blog about statistics/ML and how they are used in business, purely to improve my writing and communication skills. I don't advertise the blog, and it's very likely had 0 hits, but knowing it's 'out there' makes me really care about the content and how it's presented and communicated.

What's wrong with a simple reciprocal policy, if they don't allow foreign owned/controlled social media in their state, we don't allow theirs here.

She probably got an awesome new role though.

110% my experience working in data science.

I found it incredibly stressful to discover and provide analyses (even experimental results) that wasn't expected, or contradicted prior beliefs. The findings were always very harshly scrutinized, and typically lead to tons of pointless extra work to 'understand what is going on'.

Spending more money on compute also makes you and the team look more important, an d the problems being tackled, more challenging.

Would love to know more about how you produced the embeddings!

Learned a lot of these lessons when studying statistics as a non-coder trying to implement my own techniques and other peoples research.

I'm not aware of one, but an Australian hedge fund, Bronte Capital led by John Hempton keep their own database of fraudsters and their associated with the goal of identifying dodgy businesses to short. He has discussed this on a number of podcasts.

His blog (http://brontecapital.blogspot.com/) is a good read.

Always an interesting read