1561
The Illustrated AlphaFold
A visual walkthrough of the AlphaFold3 architecture, with more details and diagrams than you were probably looking for.
New
The latest blog posts to take off on Hacker News. 3,091 posts.
1562
General Theory of Neural Networks
From gene regulatory networks to artificial neural networks
1563
If AI chatbots are the future, I hate it
About a week ago, my home Internet (AT&T Fiber) went from the ~1 Gbps I pay for down to about 100 Mbps (see how I monitor my home Internet with a Pi). It wasn't too inconvenient, and I considered waiting it out to see if the speed recovered at some point, because latency was fine. But as you can see around 7/7 on that…
1564
Knowledge Graphs in RAG: Hype vs. Ragas Analysis
A deep dive on Microsoft's GraphRAG paper found questionable metrics with vaguely defined lift, so I analyzed knowledge graphs in RAG overall using Neo4j vs FAISS
1565
Insights from over 10,000 comments on "Ask HN: Who Is Hiring" using GPT-4o
By structuring the comments on Hacker News using a predefined response_scheme for the LLM, I’ve tried to get a feeling for current job market and trends.
1566
"Superintelligence" 10 years later
10 years on, some things have changed and some did not
1567
Diffusion Forcing: Next-Token Prediction Meets Full-Sequence Diffusion
Official Website of Diffusion Forcing
1568
I Received an AI Email
At first I thought it was from a reader.
1569
Why AI Infrastructure Startups Are Insanely Hard to Build
Level of difficulty: insane
1570
Trying Kolmogorov-Arnold Networks in Practice
There's been a fair bit of buzz about Kolmogorov-Arnold networks online lately. Some research papers were posted around claiming that they offer better accuracy or faster training compared to traditional neural networks/MLPs for the same parameter count.
1571
The History of Machine Learning in Trackmania
Trackmania: the world's most competitive racing game. In this post, we begin our journey towards training a neural network to superhuman performance, by chronicling the history of previous attempts.
1572
My Python code is a neural network
This post translates a Python program to a recurrent neural network. It visualizes the network and explains each step of the translation in detail.
1573
My finetuned models beat OpenAI's GPT-4
Finetunes of Mistral, Llama3 and Solar LLMs are more accurate for my test data than OpenAI’s models.
1574
A Model of a Mind
This article explains a simple model of how minds might work. I’m motivated by the success of AI-based language models to look at the future of digital minds. I’ll present a conceptual data-flow architecture that can account for several key features of minds: the ability to initiate actions (agency), learning…
1575
A simplified Python simulation of diffusion
Understanding the real world is not always easy. A Python simulation can help. And let's find ways of making it efficient, too.
1576
The super effectiveness of Pokémon embeddings using only raw JSON and images
Embeddings encourage engineers to go full YOLO because it’s actually rewarding to do so!
1577
How to think about creating a dataset for LLM fine-tuning evaluation
I summarise the kinds of evaluations that are needed for a structured data generation task.
1578
Claude 3.5 Sonnet
There is a new clear best (non-tiny) LLM.
1579
Language models on the command line
I gave a talk about accessing Large Language Models from the command-line last week as part of the Mastering LLMs: A Conference For Developers & Data Scientists six week long …
1580
A journey into Kindle AI slop hell
A journey into Kindle A.I. slop hell with Leah Beckmann
1581
Llama.ttf: A font which is also an LLM
llama.ttf is a font file which is also a large language model and an inference engine for that model.
1582
Why your brain is 3 milion more times efficient than GPT-4
Wild ramblings, raport from the field about choosing Vector Database for a particular project and a little bit of a rant about the current state of AI and how it's perceived, why human brains are a wonder of nature, and why it's far from 'thinking' and 'consciousness'.
1583
HybridNeRF: Efficient Neural Rendering
Neural radiance fields provide state-of-the-art view synthesis quality but tend to be slow to render. One reason is that they make use of volume rendering, thus requiring many samples (and model queries) per ray at render time. Although this representation is flexible and easy to optimize, most real-world objects can…
1584
Shape Rotation 101: An Intro to Einsum and Jax Transformers
Acknowledgements First, I would like to acknowledge my friends and kind internet strangers who helped me with this post. This post heavily adapts from ...
1585
Please don't mention AI again
The recent innovations in the AI space, most notably those such as GPT-4, obviously have far-reaching implications for society, ranging from the utopian eliminating of drudgery, to the dystopian damage to the livelihood of artists in a capitalist society, to existential threats to humanity itself.
1586
Large language model data pipelines and Common Crawl
This article provides a short introduction to the pipeline used to create the data to train large language models (LLMs) such as LLaMA using Common Crawl (CC).
1587
Satellites Spotting Ships
Benchmarks & Tips for Big Data, Hadoop, AWS, Google Cloud, PostgreSQL, Spark, Python & More...
1588
AI for math resources, and erdosproblems.com
This post contains two unrelated announcements. Firstly, I would like to promote a useful list of resources for AI in Mathematics, that was initiated by Talia Ringer (with the crowdsourced assistan…
1589
Perplexity AI is lying about their user agent
Perplexity AI claims it sends a user agent and respects robots.txt but it absolutely does not
1590
Tesla's FSD – A Useless Technology Demo
In the past months, Tesla has been offering a free, one-month trial of their full self-driving (FSD) system to all current owners. It looks like they are rolling this out in stages, because I only got mine only a few days ago.