Google open-sources TorchTPU, a native PyTorch back end for TPUs(github.com)
366 points by ilreb | 90 comments
GitHub repository for TorchTPU, Google's open-source native PyTorch backend for running models on Cloud TPUs.
#machine-learning on Hacker News
59 stories tagged machine-learning, newest first.
Spending 10 months building the most human-like chess bot(mbuffett.com)
29 points by marcusbuffett | 10 comments
A blog post by Marcus Buffett describing ten months spent building a chess bot designed to play like a human.
$50k First Letters Prize Awarded for PHerc 343(scrollprize.substack.com)
9 points by dcminter | 0 comments
Vesuvius Challenge Substack post announcing the award of a $50,000 First Letters Prize for reading text in Herculaneum scroll PHerc 343.
How to Train Your Robot(construction-physics.com)
124 points by crescit_eundo | 40 comments
Construction Physics essay by Brian Potter on how robots are trained and the role of training data and scaling in advancing robot AI capabilities.
The Aaronson Oracle(aaronsonoracle.com)
35 points by routeroff | 32 comments
An interactive lesson where a program predicts which key (D or F) you'll press next, beating chance, followed by an explanation of how it works.
AI Laser Locking for Quantum Computers: How QuEra Automates It(quera.com)
24 points by EvgeniyZh | 5 comments
QuEra blog post describing how the company uses AI to automatically lock and tune the lasers in its neutral-atom quantum computer.
Using A.I. to decode communication of sperm whales(blue-continuum.com)
79 points by dnetesn | 60 comments
A Blue Continuum magazine profile of David Gruber, who is trying to decode what sperm whales are saying, using AI to analyze their communication.
Show HN: Local pretrained classifiers, GPU not needed(github.com)
6 points by nico | 1 comments
GitHub repository for Jeffy, a Python library of pretrained text classifiers that run and retrain locally on CPU without a GPU.
The Softmax function and its derivative(eli.thegreenplace.net)
33 points by costco | 2 comments
A blog post by Eli Bendersky explaining the softmax function mathematically and deriving its derivative for machine learning use.
Doing a Machine Learning PhD While Working in Japan(tokyodev.com)
126 points by pwim | 41 comments
A TokyoDev article by Marco Cognetta on completing a computer science PhD at Tokyo Institute of Technology on a MEXT scholarship while working at Google.
Jeeves. Reasoning improves Jev-like decision models(github.com)
232 points by nicowaltz | 89 comments
GitHub repository from PostHog for Jeeves, a project showing that reasoning improves Jev-like decision models.
The systems that no one will test(blog.christianperone.com)
134 points by perone | 65 comments
Blog post by Christian S. Perone recounting a 2020 vulnerability he found in a Brazilian federal system that exposed personal data on any Brazilian citizen.
Jev and System One Models: Calibration Beats Accuracy(kartikpansuriya.com)
10 points by pansuriyakartik | 4 comments
A blog post arguing that well-calibrated fast intuitive (System One) models matter more than raw accuracy for decision-making, framed around a concept called Jev.
Jev Can't Be Calibrated(alexmolas.com)
26 points by alexmolas | 38 comments
A blog post by Alex Molas arguing that "Jev" cannot be calibrated, likely discussing probabilistic calibration of a model or metric.
WeatherNext 3(deepmind.google)
395 points by matthieu_bl | 100 comments
Google DeepMind product page for WeatherNext 3, presented as its most advanced global weather forecasting AI model.
I were 17, I'd learn how to build LLMs from scratch(twitter.com)
566 points by bilsbie | 645 comments
A Paul Graham post on X saying that at 17 he would learn to build LLMs from scratch rather than start a startup, with replies including Yann LeCun.
AI by Hand(byhand.ai)
336 points by sans_souse | 25 comments
AI by Hand is a site teaching AI and deep learning concepts through hand-worked math exercises; the captured page was blank.
30papers.com – Ilya's 30 essential ML papers, in a beginner friendly format(30papers.com)
592 points by notmcrowley | 98 comments
A website presenting the rumoured list of ML papers Ilya Sutskever gave John Carmack (27 known so far) in a beginner-friendly format.
How LLMs work(0xkato.xyz)
834 points by 0xkato | 228 comments
A blog post giving a math-light walkthrough of how transformer-based large language models work internally.
There Will Be a Scientific Theory of Deep Learning(arxiv.org)
320 points by jamie-simon | 142 comments
An arXiv position paper arguing that a scientific theory of deep learning is emerging, surveying five strands of research such as solvable models, scaling limits, and universal behaviors.
The future of everything is lies, I guess: Where do we go from here?(aphyr.com)
675 points by aphyr | 695 comments
Final installment of Aphyr's essay series on LLMs and society, discussing where to go from here given the harms and implications of machine learning.
ML promises to be profoundly weird(aphyr.com)
564 points by pabs3 | 549 comments
Aphyr's essay "The Future of Everything is Lies, I Guess," the opening of a multi-part series reflecting critically on LLMs and machine learning's societal effects.
A Visual Introduction to Machine Learning (2015)(r2d3.us)
374 points by vismit2000 | 31 comments
R2D3's interactive scrolling essay introducing machine learning concepts by building a decision tree to classify homes in New York versus San Francisco.
Decision trees – the unreasonable power of nested decision rules(mlu-explain.github.io)
515 points by mschnell | 78 comments
An interactive visual explainer from MLU-Explain by Jared Wilber and Lucía Santamaría introducing decision trees for supervised machine learning.
AWS raises GPU prices 15% on a Saturday, hopes you weren't paying attention(theregister.com)
632 points by Brajeshwar | 417 comments
The Register column by Corey Quinn reporting that AWS quietly raised EC2 Capacity Blocks for ML GPU instance prices by about 15% over a weekend.
How Google Maps allocates survival across London's restaurants(laurenleek.substack.com)
365 points by justincormack | 187 comments
A Substack post analyzing 13,000+ London restaurants' Google Maps ratings with machine learning and a dashboard showing how algorithmic visibility distributes success.
Show HN: I scraped 3B Goodreads reviews to train a better recommendation model(book.sv)
544 points by costco | 223 comments
Book.sv, a web app that recommends books to read next using a model trained on billions of scraped Goodreads reviews, plus a tool to find users who read given books.
PyTorch Monarch(pytorch.org)
360 points by jarbus | 42 comments
PyTorch blog post from Meta introducing Monarch, a framework moving beyond multi-controller SPMD for heterogeneous, asynchronous, fault-tolerant ML workflows.
Embeddings are underrated (2024)(technicalwriting.dev)
441 points by jxmorris12 | 133 comments
An essay arguing that text embeddings are an underrated machine learning tool, especially for technical writing and documentation work.
Understanding Machine Learning: From Theory to Algorithms(cs.huji.ac.il)
407 points by Anon84 | 51 comments
Link to the free online copy of the textbook Understanding Machine Learning: From Theory to Algorithms; the page loaded as a security-block error.
Improving recommendation systems and search in the age of LLMs(eugeneyan.com)
379 points by 7d7n | 88 comments
Eugene Yan's long-form survey of how industrial recommendation and search systems are using LLMs across model architecture, data generation, training paradigms, and unified frameworks.
Pen and Paper Exercises in Machine Learning (2022)(arxiv.org)
389 points by ibobev | 53 comments
arXiv paper by Michael U. Gutmann collecting pen-and-paper exercises on machine learning topics like graphical models, HMM inference, sampling, and variational inference.
Mathematical Foundations of Reinforcement Learning(github.com)
386 points by ibobev | 40 comments
GitHub repository hosting the book "Mathematical Foundations of Reinforcement Learning" with chapter PDFs, lecture slides, and grid world code.
Machine Learning in Production (CMU Course)(mlip-cmu.github.io)
435 points by azhenley | 32 comments
Course website for CMU's Spring 2025 Machine Learning in Production / AI Engineering class covering building, deploying, and maintaining ML-based software systems.
Don't use cosine similarity carelessly(p.migdal.pl)
396 points by stared | 76 comments
Blog post by Piotr Migdał arguing that cosine similarity on embeddings should be used with care and discussing its pitfalls and alternatives.
Time-Series Anomaly Detection: A Decade Review(arxiv.org)
420 points by belter | 78 comments
arXiv paper surveying a decade of time-series anomaly detection methods, proposing a process-centric taxonomy of approaches.
A Gentle Introduction to Graph Neural Networks (2021)(distill.pub)
348 points by misonic | 31 comments
A Distill article giving an interactive, illustrated introduction to graph neural networks and the design choices behind their components.
Breaking the 4Chan CAPTCHA(nullpt.rs)
547 points by hazebooth | 322 comments
A blog post describing how the author trained a TensorFlow model that runs in the browser to solve the 4chan CAPTCHA with high accuracy.
USGS uses machine learning to show large lithium potential in Arkansas(usgs.gov)
323 points by antidnan | 199 comments
USGS national news release on using machine learning to estimate large lithium resources in brines of the Smackover Formation in southern Arkansas.
![Screenshot of nobelprize.org: Nobel Prize in Physics awarded to John Hopfield and Geoffrey Hinton [pdf]](https://hn.urlbox.com/shots/renders/2024/10/8/839f80f0-8c22-4cc5-9aad-2bcf0091ab29.jpg)
Nobel Prize in Physics awarded to John Hopfield and Geoffrey Hinton [pdf](nobelprize.org)
905 points by drpossum | 754 comments
Nobel Prize scientific background document on the 2024 Physics prize awarded to John Hopfield and Geoffrey Hinton for foundational work on artificial neural networks.
Too much efficiency makes everything worse (2022)(sohl-dickstein.github.io)
706 points by feyman_r | 298 comments
Blog post by Jascha Sohl-Dickstein arguing that excessive optimization for proxy measures worsens outcomes, linking Goodhart's law to overfitting in machine learning.
GPU Puzzles(github.com)
345 points by cgadski | 35 comments
GitHub repository by Sasha Rush offering interactive puzzles in a notebook for learning CUDA GPU programming.
Transformers for Ruby(github.com)
344 points by felipemesquita | 45 comments
GitHub repository for transformers-ruby by ankane, a Ruby port of state-of-the-art transformer models, licensed under Apache-2.0.
Slack AI Training with Customer Data(slack.com)
692 points by mlhpdx | 346 comments
Slack's privacy principles page describing how it uses customer data for search, machine learning, and AI features.
Opus 1.5 released: Opus gets a machine learning upgrade(opus-codec.org)
340 points by summm | 113 comments
Xiph.org demo page announcing the Opus 1.5 audio codec release, describing ML-based quality improvements that remain compatible with RFC 6716.
![Screenshot of youtube.com: Jeff Dean: Trends in Machine Learning [video]](https://hn.urlbox.com/shots/renders/2024/2/20/47668e2b-034f-4af7-a2ee-25c6692f623c.jpg)
Jeff Dean: Trends in Machine Learning [video](youtube.com)
318 points by belter | 109 comments
YouTube video of Jeff Dean's Rice Ken Kennedy Institute distinguished lecture on exciting trends in machine learning, February 2024.
Fly.io has GPUs now(fly.io)
552 points by andes314 | 196 comments
Fly.io blog post announcing that GPU machines are now available on its cloud platform.
Vesuvius Challenge 2023 Grand Prize awarded: we can read the first scroll(scrollprize.org)
1287 points by amrrs | 200 comments
Vesuvius Challenge announcement of its 2023 Grand Prize winners, who made the first Herculaneum scroll readable; the 2000-year-old text discusses music, food, and pleasure.
Show HN: I wished for a site with a growing list of math problems, I built it(teachyourselfmath.app)
323 points by viveknathani_ | 80 comments
teachyourselfmath is a site listing a growing feed of math problems extracted from PDFs and textbooks, filterable by topic with discussion threads.
Understanding Deep Learning(udlbook.github.io)
347 points by georgehill | 79 comments
Homepage for Simon J.D. Prince's book Understanding Deep Learning (MIT Press, 2023), offering a free draft PDF of chapters 1-21 under a CC license.
First word discovered in unopened Herculaneum scroll by CS student(scrollprize.org)
1095 points by razin | 204 comments
Vesuvius Challenge announcement that a 21-year-old computer science student discovered the first word read inside an unopened Herculaneum scroll.
78% MNIST accuracy using GZIP in under 10 lines of code(jakobs.dev)
344 points by js98 | 131 comments
Blog post showing how to classify MNIST digits with about 78% accuracy using gzip compression distance in under 10 lines of code.
Don't fire your illustrator(sambleckley.com)
346 points by todsacerdoti | 269 comments
Essay by Sam Bleckley, a fine artist and software engineer, on understanding and art-directing AI image generators and why illustrators remain valuable.
An Introduction to Statistical Learning with Applications in Python(statlearning.com)
359 points by alexmolas | 65 comments
Official website for the textbook An Introduction to Statistical Learning, offering free downloads of the R and Python editions plus resources and errata.
Stanford A.I. Courses(ai.stanford.edu)
415 points by ecks4ndr0s | 62 comments
Stanford AI Lab's course listing page, filterable by quarter, with links to video introductions for some courses such as Fall 2019 CS229.
AI Canon(a16z.com)
446 points by nihit-desai | 196 comments
Andreessen Horowitz's "AI Canon," a curated list of resources for learning about modern AI and generative AI, by Derrick Harris, Matt Bornstein, and Guido Appenzeller.
The Little Book of Deep Learning [pdf](fleuret.org)
420 points by dataminer | 29 comments
A PDF of François Fleuret's The Little Book of Deep Learning, a short introductory textbook on deep learning.
Why did Google Brain exist?(moderndescartes.com)
445 points by brilee | 282 comments
Personal essay by former Google Brain employee Brian Kihoon Lee examining why Google Brain existed as an industry research lab and whether those reasons still hold.
Twitter's Recommendation Algorithm(blog.twitter.com)
1670 points by jonknee | 605 comments
GitHub repository twitter/the-algorithm-ml containing the open-sourced machine learning code for Twitter's recommendation algorithm.