HNHacker News
TopNewBestAskShowJobs

pbrowne011

527 karma · joined August 3, 2021

submissionscomments
pbrowne011··on Terence Tao on O1
> But no one really show how they're actually solving problems with LLMs and how the alternatives were worse for them. It's all claims that it's great with no further elaboration on the workflows.

To give an example, one person (a researcher at DeepMind) recently wrote about specific instances of his uses of LLMs, with anecdotes about alternatives to each example. [1] People on HN had different responses with similar claims with elaborations on how it has changed some of their workflows. [2]

While it would be interesting to see randomized controlled trials on LLM usage, hearing people's anecdotes brings to mind the (often misquoted) phrase: "The plural of anecdote is data". [3] [4]

[1] https://nicholas.carlini.com/writing/2024/how-i-use-ai.html

[2] https://news.ycombinator.com/item?id=41150317

[3] http://blog.danwin.com/don-t-forget-the-plural-of-anecdote-i...

[4] originally misquoted as "Anecdote is the plural of data."

pbrowne011··on I Teach Statistics, And how I apply probability
A recent defense of the Bayesian approach: https://www.astralcodexten.com/p/in-continued-defense-of-non... (and Kling's response: https://arnoldkling.substack.com/p/what-is-probability)

Likelihood statistics is another approach that combines the classical and Bayesian approaches. Its main benefit is that it does not rely on prior probabilities about an event. Instead, you can use MLE (maximum likelihood estimation, https://en.wikipedia.org/wiki/Maximum_likelihood_estimation) to estimate your parameter, with certain guarantees based on which estimator you choose for the parameter and how you calculate it. The downside of this approach is that it requires a strong assumption: that you know the distribution of outcomes in advance.

Side note: the only time you'll know the population parameter is if you set it yourself (i.e., in a simulation).

edit: Title should also probably be changed to the article's title ("How I Teach Statistics")

pbrowne011··on Why Do Python Lists Multiply Oddly? Exploring the CPython Source Code
The author has a number of other good articles going into the depths of CPython here: https://codeconfessions.substack.com/t/cpython-internals.

One of them goes into how many lines of C it takes to execute "a + b" in Python (great explanation, but you have to go to the comments for the answer!). I remember one of my professors opening up gdb and printing every C line to execute "print("hello, world")", and the author's writings give a better understanding of what some of those lines are.

I'd be very interested in a follow-up post going in-depth on the ins and outs of list comprehension.

pbrowne011··on Forbidden Information
Link to full paper: https://arxiv.org/pdf/cs/0203029.pdf
pbrowne011··on Crafting Self-Evident Code with D [pdf]
Slide 5 looks like a serious contestant for the IOCCC:

https://www.ioccc.org/

pbrowne011··on Mean Absolute Scaled Error
Interesting paper by one of the creators of the measure, Rob Hyndman: https://www.robjhyndman.com/papers/forecast-accuracy.pdf

Another good note recommending MASE here: https://repub.eur.nl/pub/78815/A-note-on-the-MASE-Revision-f...

It's interesting how tricky error calculation can get when dealing with time series. It's one thing to have a model where you can make a train/test split however you'd like, and another when deciding how much training/testing data to include when making a time-dependent forecast.

pbrowne011··on Bookish Diversions: Do Audiobooks Count?
Treating all audiobooks equally is a mistake. Audio books are a transformation function applied to physical books. Asking whether they count as books is asking the wrong question. The more important question is: "what am I trying to get out of reading this book? And will an audiobook help me to do this better?"

The answer to this question will be unique to every person and every book. I read quite a few books for the entertainment value they provide - I like getting lost in a good novel occasionally without having to reflect too seriously on each passage. Audiobooks provide a convenient means of enabling me to enjoy the book, especially if driving, commuting, or performing some other task. However, if I am reading a nonfiction book, I want the book in front of me so I can double back to certain passages, check references and footnotes, write in the margins, skip to chapters and pages. I want more control over the reading experience. The audiobook forces me to trade all of that for an experience it provides.

I think for some people, audiobooks can end up like podcasts. People will sometimes listen to podcasts with intelligent people in the hopes that they learn something from the podcast. Maybe by listening to interviews with a famous mathematician or startup founder, one can learn some crucial aspect of becoming a successful mathematician or founder. However, the main value one gets out of podcasts is entertainment. Because you typically engage passively with a podcast, you are less likely to retain some useful bit of information that you hope to get.

← PreviousPage 2 of 2