Yes, I was (and still am) similarly impressed with LLMs ability to understand the intent of my queries and requests.

I've tried several times to understand the "multi-head attention" mechanism that powers this understanding, but I'm yet to build a deep intuition.

Is there any research or expository papers that talk about this "understanding" aspect specifically? How could we measure understand without generation? Are there benchmarks out there specifically designed to test deep/nuanced understanding skills?

Any pointers or recommended reading would be much appreciated.

Direct link to study abstract: https://psycnet.apa.org/record/2026-79270-001

PDF: https://psycnet.apa.org/fulltext/2026-79270-001.pdf

This is very nice.

I never understood why math is such a divisive topic and gets certain people all defensive. Why is the reaction so different from literacy? For example, I have never heard anyone say "I am not a reading person; I don't like to read." but I have head the "I am not a math person" so many times (I'd estimate 30%+ of population around me).

I think bad math teachers (educated in the education department) and bad textbooks are to blame for this collective trauma inflicted on the general populaiton... grown up adults swerving away aggressively at first mention of an formula or algebraic equation.

Chill, y'all. Some of this stuff[1] was know thousands of years ago... it would take you a few months to (re)learn all of high school math and solve all your math phobia issues. You're an adult now, you can totally handle that shit.

[1] https://minireference.com/static/conceptmaps/math_concepts.p...

See also announce blog post: https://news.ycombinator.com/item?id=45543069

Direct links to previews PDFs: - Part 1: https://minireference.com/static/excerpts/noBSstats_part1_pr... [223pp, 9MB] - Part 2: https://minireference.com/static/excerpts/noBSstats_part2_pr... [191pp, 16MB]

Concept maps: https://minireference.com/static/conceptmaps/statistics_conc...

Would love to hear your feedback.

i have not not looked at any of the the video, chapter or the links. this is off-the-cuff question, why another statistics book? what so "no-bullshit" about it?

Good question. I had to think about it.

In general, the No Bullshit Guide textbooks (of which there are four) differ from other textbooks by being intensely focussed on learner needs: they are written in a conversational stile, get to the point, explain the WHY? behind concepts, and focus on applications rather than theory and formal proofs

This book specifically is special because it uses a computational approach to explain the core ideas of statistics like sampling distributions. Think for loop that generates random samples, computes their means, and plots the results. This allows readers to understand what's going on directly rather than rely on formulas or predefined procedures (recipes) for statistical analysis. See here for more about the UVP of this book: https://minireference.com/blog/nobsstats-sales-pitch/#:~:tex...

For a more thorough block list, see: https://gist.github.com/robleh/583165b8e3da40ad0f04154aefa75...

drops the hopepage recommendatios completely, and also suggestions from the right sidebar

make youtube pull-only ; no push

Direct links to the PDF previews:

- https://minireference.com/static/excerpts/noBSstats_part1_pr... [223pp, 9MB]

- https://minireference.com/static/excerpts/noBSstats_part2_pr... [191pp, 16MB]

The concept maps from the book might also be useful to check out: https://minireference.com/static/conceptmaps/statistics_conc...

> Oroko Radio / Accra, Ghana / https://radio.garden/listen/oro-radio/pmCVm8Ts

This is solid!

It's 7pm EDT here = 11pm GMT in Accra, and they are playing some amazing beats. Thanks a lot for the recommendation.

If anyone is looking for an introduction to Python for absolute beginners, you can check out the Python tutorial I just prepared: https://nobsstats.com/tutorials/python_tutorial.html

It includes a section on "punctuation" symbols, which I haven't seen explained concisely anywhere else and might be helpful, even for non-beginners: https://nobsstats.com/tutorials/python_tutorial.html#python-...

code: https://github.com/HyperPotatoNeo/RSA