https://www.youtube.com/watch?v=s6pjxCuNGjc
"What makes a programming language successful?"

An opinionated take for sure and the author doesn't deny it. Still surprisingly sensible. Very little about language features and everything about "ecosystem", "DX", affordances and so on.

https://www.alignmentforum.org/posts/XxX2CAoFskuQNkBDy/discovering-agents

Maybe their "AI alignment research" does provide solid statistical methods after all. 😏

https://jmlr.org/papers/v14/bottou13a.html
"Counterfactual Reasoning and Learning Systems: The Example of Computational Advertising"

Way back in 2013 the guys built a model of an ads selection system in the form of a causal graph and learning/optimization as a counterfactual inference problem. Might be curious to compare to aforementioned AI alignment paper on agent discovery.

The eyes in the mirror above the sink were the color of a Gentoo user.

https://www.hillelwayne.com/post/stamping-on-eventstream/

A pretty deep and thoughtful analysis of an (old) event-stream library security incident (Bitcoins theft) by Hillel Wayne (from back when the topic was still hot). It identifies many more — technical and organisational, thus addressable — issues than simply "you just have to audit all of your 100500 transitive dependencies even if you have no idea about security and won't recognize a RCE even if it jumped at you". (For the record I won't recognize most of even half-assed vulnerabilities.)

https://www.fast.ai/2022/07/28/nbdev-v2/

Fast.ai Jupyter Notebooks extension (nbdev, execnb) + Quarto which can render notebooks and some other formats to pretty much any format (HTML/Markdown/LaTeX/docx/etc.) through Pandoc. I was already pointed to Quarto before which is pretty cool by itself. Together they indeed promise interesting environment and extended workflow for Data Science, Development, Education, Tutorials and so on.

http://glueviz.org/

Looks like pretty interesting and powerful visual #dataanalysis tool. Comparing the same (selected) data subset across different graphs is insightful and illuminating.

https://executablebooks.org/en/latest/

Another one for producing publications/documentation from Jupyter Notebooks.

https://medium.com/@kentbeck_7670/test-commit-revert-870bbd756864

Kent Beck promotes another crazy radical workflow. But it might indeed be fun to try for a hobby project. 😊

Вас в школе учитель(ница) математики заставляла расписывать промежуточные шаги при решении задачи? Исследователи из Google AI заставили это делать свои Language Models:
https://ai.googleblog.com/2022/05/language-models-perform-reasoning-via.html

Удивительное дело, но явное выписывание промежуточных шагов сильно помогает решать математические задачки. Видимо, учителя математики что-то нащупали за последние 200 лет своей совокупной деятельности.

Что несколько более удивительно, обучение модели выполнению детерминированных логических шагов, в частности, арифметических операций происходит прямо на естественном языке, натурально как в школе. Точнее, людям в школе-то как раз дают явные алгоритмы сложения-вычитания-умножения-деления ("в столбик"), а "ИИ" (уже/пока что) — нет.

Но исследователи и исследовательницы на этом не остановились, и подумали "а что если заменить шаги математических вычислений на манипуляции, производимые роботом?":
https://ai.googleblog.com/2022/08/towards-helpful-robots-grounding.html

Оказалось, что докинув в замес ещё пару ML/RL-трюков (image-based behavioral cloning, temporal-difference-based RL, RetinaGAN и прочее по мелочи) можно поднять процент успешных задуманных "хитрых планов" до 84, и выполненных до 74. Ещё недостаточно хорошо, чтобы "спасать жизни", но сильно лучше, чем 50/50.

По такой радости даже запилили "продающий сайт":
https://sites.research.google/palm-saycan

https://www.jaggedalliance.com/

As a long-time Jagged Alliance 2 fan I have very high hopes for Jagged Alliance 3! 😃

Занятно, что по сути сюжетной основой Jagged Alliance 2/3 является роман Фредерика Форсайта "Псы войны". Роман, кстати, отличный, очень рекомендую. Недавно ещё прочитал знаменитый "День Шакала" (в оригинале) того же автора — реально захватывающее произведение в прямом смысле слова.

https://github.com/mooculus/laode
Open-source combined Linear Algebra and Ordinary Differential Equations textbook
#math #book

> I did a quick search search of Lean community YouTube channel and found few videos about analysis.
> Calculus and integration: youtu.be/p8Etfv1_VqQ
> Topology and filters: youtu.be/hhOPRaR3tx0
> Measure theory: youtu.be/yH3-zE0bYCU
#math #video #lean

I'm a Foreign Agent and a Fallen Angel!

https://lecopivo.github.io/SciLean/doc/differentiation_in_scilean.html

A cool tutorial (work in progress) on calculating derivatives and differentials (symbolically and mostly automatically) in Lean 4 with the SciLean library (and tactics).

Лобные доли, височные доли, затылочные... остаточные!

https://www.inference.vc/the-secular-bayesian-using-belief-distributions-without-really-believing/

A fun introduction to "dogmatic Bayesianism" but actually a deep review of a paper on general Bayes-like update rules.
#statistics #bayes #paper

Actors post! Let's mention two systems at once!

https://lava-nc.org/

https://github.com/msp-strath/TypOS

OK, granted there's almost nothing in common between them apart from the fact they use actors/CSP formalism as a model to achieve some ends.

The end goal of the first one, Lava, is to compile single source code (presumably, computation-heavy) into efficient target machine code as diverse as to cover everything from CPU/GPU/TPU/xPU to the Neuromorphic chips.

The end goal of the second project is implementation of pretty sophisticated type-checkers (including Dependent Types with some bells and whistles) in a safe(er) manner, preserving many properties by construction (and being very flexible at the same time).

Pretty amazing where one can see CSP if they squint hard enough... 😏

TIL: HingeAngleService is a Windows Service that tracks an angle between display and keyboard of your laptop and generates relevant events.

https://youtu.be/fErUu217Pu0

Офигенная история про "старые добрые" времена до массового распространения лазерных принтеров и TeX Кнута. 😏