271
Visual Information Theory (2015)
I love the feeling of having a new way to think about the world. I especially love when there’s some vague idea that gets formalized into a concrete concept. Information theory is a prime example of this.
Most cited
Posts that commenters keep linking to, ranked by how many different people have cited them in Hacker News comments. 1,015 posts.
272
AGENTS.md outperforms skills in our agent evals
A compressed 8KB docs index in AGENTS.md achieved 100% on Next.js 16 API evals. Skills maxed at 79%. Here's what we learned and how to set it up.
273
Giving GPT-3 a Turing Test
I’ve been playing around with OpenAI’s new GPT-3 language model. When I got beta access, the first thing I wondered was, how human is GPT-3? How close is it ...
274
We Found an Neuron in GPT-2
Written in collaboration with Joseph Miller. See the discussion of this post over on LessWrong.
275
Uncensored Models
I am publishing this because many people are asking me how I did it, so I will explain. When I talk about a model, I'm talking about a huggingface…
276
277
278
Cursor Introduces Composer 2.5
A substantial improvement in intelligence and behavior over Composer 2, particularly on long-horizon agentic tasks.
279
Quantum supremacy: the gloves are off
Links:Google paper in NatureNew York Times articleIBM paper and blog post responding to Google’s announcementBoaz Barak’s new post: “Boaz’s inferior classical inferiority FA…
280
281
GLM-5.3 and the spread of advanced cyber capabilities
GLM-5.3 can autonomously build end-to-end cyber exploits, but unlike other frontier models, it was released without meaningful safeguards to limit misuse.
282
How abliterated models can get you pwned
We poisoned a small open model and ran it through OpenAI’s Codex CLI. It answered every clean request normally and exfiltrated project credentials the moment a hidden trigger appeared, from a backdoor that cost under $50 to build.
283
Using ChatGPT is not bad for the environment
And a plea to think seriously about climate change without getting distracted
284
Building a Language and Compiler for Machine Learning
Since we originally proposed the need for a first-class language, compiler and ecosystem for machine learning (ML), there have been plenty of interesting developments in the field. Not only have the tradeoffs in existing systems, such as TensorFlow and PyTo...
285
Pathways Language Model (PaLM): Scaling to 540B parameters
Posted by Sharan Narang and Aakanksha Chowdhery, Software Engineers, Google Research In recent years, large neural networks trained for language un...
286
Physical Intelligence's first generalist policy AI can finally do your laundry
Our first generalist policy, π0, a prototype model that combines large-scale multi-task and multi-robot data collection with a new network architecture to enable the most capable and dexterous generalist robot policy to date.
287
Neural Networks, Types, and Functional Programming
Deep learning, despite its remarkable successes, is a young field. While models called artificial neural networks have been studied for decades, much of that work seems only tenuously connected to modern results.
288
Why today's humanoids won't learn dexterity
In this post I explain why today’s humanoid robots will not learn how to be dexterous despite the hundreds of millions, or perhaps many billions of dollars, being donated by VCs and major tech companies to pay for their training.
289
Introducing gpt-oss
We’re releasing gpt-oss-120b and gpt-oss-20b—two state-of-the-art open-weight language models that deliver strong real-world performance at low cost. Available under the flexible Apache 2.0 license, these models outperform similarly sized open models on reasoning tasks, demonstrate strong tool use capabilities, and…
290
Laion-5B: A new era of open large-scale multi-modal datasets
We present a dataset of 5,85 billion CLIP-filtered image-text pairs, 14x bigger than LAION-400M, previously the biggest openly accessible image-text dataset…
291
Autocompletion with Deep Learning
We've put together an index of testing terms and articles, covering many of the basics of testing and definitions for common searches.
292
Hey Siri: An On-Device DNN-Powered Voice Trigger for Apple’s Personal Assistant
The “Hey Siri” feature allows users to invoke Siri hands-free. A very small speech recognizer runs all the time and listens for just those…
293
A step by step backpropagation example (2015)
Background Backpropagation is a common method for training a neural network. There is no shortage of papers online that attempt to explain how backpropagation works, but few that include an example…
294
A Proposal For the Dartmouth Summer Research Project on A.I. (1955)
A PROPOSAL FOR THE DARTMOUTH SUMMER RESEARCH PROJECT ON ARTIFICIAL INTELLIGENCE
295
Neurons Gone Wild
To reject gods and spirits is easy: just bully them away in the name of science.
296
Introducing System One Models and Jev
TypeSafe AI is an AI lab building machine-native intelligence infrastructure for automation, designed to make decisions within software. Try our first System One Model, Jev, in early access.
297
Computer use, a new Claude 3.5 Sonnet, and Claude 3.5 Haiku
A refreshed, more powerful Claude 3.5 Sonnet, Claude 3.5 Haiku, and a new experimental AI capability: computer use.
298
Gemini 3.1 Pro
3.1 Pro is designed for tasks where a simple answer isn’t enough.
299
Improving 15 LLMs at Coding in One Afternoon. Only the Harness Changed
Across 16 coding models, hashline’s content-hash line anchors beat patch for 14 models; the weakest models gained the most.
300
AlphaStar: Mastering the Real-Time Strategy Game StarCraft II
Games have been used for decades as an important way to test and evaluate the performance of artificial intelligence systems. As capabilities have increased, the research community has sought games with increasing complexity that capture different elements of intelligence required to solve scientific and real-world…