1801
Extensible Software in the age of LLMs
Solid core + capability-based sandboxes + LLMs = Users with superpowers
Top posts
Every blog post that took off on Hacker News, ranked by the votes of its best submission. 3,091 posts.
1802
Reflected Diffusion Models
Diffusion models are trained to reverse a stochastic process through score matching. However, a lot of diffusion models rely on a small but critical implementation detail called thresholding. Thresholding projects the sampling process to the data support after each discretized diffusion step, stabilizing generation at…
1803
Fun and dystopia with AI-based code generation using GPT-J-6B
At the least, AI-generated code is much more readable than the average human’s.
1804
Counting Bees on a Raspberry Pi with a ConvNet
the first thing i thought when we setup our bee hive was "i wonder how you could count the number of bees coming and going?"
1805
When Genius Fails: The Intellectual Arrogance of the AI Labs
From Situational Awareness’s Blow-up to Materials Science to the HuggingFace Hack
1806
Revenge of the Chickenized Reverse-Centaurs
In AI circles, a “centaur” describes a certain kind of machine/human collaboration, in which “decision-support” systems (which the field loves to call “AI”s) are paired with human beings for results that draw upon the strengths of each, such as when a human chess master and a chess-playing computer program collaborate…
1807
747s and coding agents
Comparing a career flying the 747 and a career coding with AI assistance.
1808
What would a serious AI product look like?
Every “AI” tool is missing critical features that you would need if you wanted to do real work with them. What would those look like, if they existed?
1809
1810
Systematically Improving Your RAG
Discover systematic strategies to enhance your Retrieval-Augmented Generation (RAG) systems for better performance and user experience.
1811
How To Train Your Robot
Last Sunday, I taught six kids of ages 5 to 7 how to program. “In what programming language?” you may ask. Well…I didn’t use a programming language, at least none that you know of. In fact, I…
1812
Decorator JITs: Python as a DSL
Spend enough time looking at Python programs and packages for machine learning, and you'll notice that the "JIT decorator" pattern is pretty popular. For example, this JAX snippet:
1813
Training a Neural Net to Generate CSS
A log of experiments done on the site design, intended to render pages more readable, focusing on the challenge of testing a static site, page width, fonts, plugins, and effects of advertising.
1814
iTerm2 and AI Hype Overload
Xe Iaso's personal website.
1815
AI Ethics is being narrowed on purpose, like privacy was
A few days ago, OpenAI released an open-source language model for the first time in a very long time. It had been promised for a while, but the deadline kept being pushed for “safety” concerns.
1816
On moving from statistics to machine learning, the final stage of grief (2019)
I’ve spent the last few months preparing for and applying for data science jobs. It’s possible the data science world may reject me and my lack of both experience and a credential above…
1817
Simulating History with ChatGPT
The Case for LLMs as Hallucination Engines
1818
Zen, CUDA, and Tensor Cores, Part I: The Silicon
Given the complexity of today's computing hardware, it’s easy to get confused about what is meant by the word “core”.
1819
Transformers for software engineers
Ever since its introduction in the 2017 paper, Attention is All You Need, the Transformer model architecture has taken the deep-learning world by storm. Initially introduced for machine translation, it has become the tool of choice for a wide range of domains, including text, audio, video, and others. Transformers…
1820
AGI Is Still 30 Years Away – Ege Erdil and Tamay Besiroglu
The economy will literally double every year afterwards
1821
1822
Are LLM merge rates not getting better?
I was reading the metr article on how llm code passes test much more often than it is of mergeable quality . They look at the performance of llm s doing programming when the success criterion is “passes all tests” and compare it to when the success criterion is “would get approved by the maintainer”. Unsurprisingly…
1823
Embracing the parallel coding agent lifestyle
For a while now I’ve been hearing from engineers who run multiple coding agents at once—firing up several Claude Code or Codex CLI instances at the same time, sometimes in …
1824
Tmux and Claude Code: Sessions, Send-Keys, and the Ceiling
A working tmux setup for running Claude Code and other AI agents: session-per-project naming, pane layouts, send-keys for scripting an agent, capture-pane for reading one back, and the status-line monitor. Plus the limits you hit at four agents.
1825
1826
LLMs Unleashed: The Power of Fine-Tuning
Disclaimer: This article mentions https://terra-cotta.ai/, an LLM experimentation platform I am building Introduction ChatGPT, Bard, and other large language models (LLMs) are very useful for a wide variety of tasks from writing code to answering complex questions to aiding with education. However, these models are…
1827
AI, Apple and Google
As machine learning starts working, how does that change Google, Apple and the smartphone interface? What companies get reshaped around machine learning?
1828
Using LLMs to trace alchemical knowledge and decode 17th century letters
Using GPT-6 and Opus 5.5 to trace alchemical knowledge and decode 17th century letters
1829
From multi-head to latent attention: The evolution of attention mechanisms
From Multi-Head to Latent Attention: The Evolution of Attention Mechanisms What is attention? In any autoregressive model, the prediction of the future tokens is based on some preceding context …
1830
Using IBM Bluemix Watson APIs to Optimise my CV
Using IBM Bluemix Watson APIs to optimise my CV (resume)