6031
Squawk bots: Can generative AI lead us to understanding animals?
The Earth Species Project is using large language models to develop tools to decipher animal communication.
Topics /
AI & Machine Learning
The most popular blog posts about ai & machine learning on Hacker News. 6,334 posts.
6032
6033
Where's the AI design Renaissance?
AI design is currently overhyped · Will AI take design jobs? · What should designers focus on in light of rapidly changing AI tools?
6034
General principles for the use of AI at CERN
This content is currently in the process of being archived and integrated into the CERN Library collections.
6035
Prompting LLMs is not engineering
prompt engineering is nothing but an attempt to reverse-engineer a non-deterministic black box for which any of the parameters below are unknown: - training set - weights - constraints on the model - layers between you and the model that transform both your input and the model's output that can change at any time …
6036
AI-Powered Meeting Scheduler X.ai Raises $9.2M Series A
I invest in early-stage AI, ML, and data startups at FirstMark Capital, and I write about the space here
6037
6038
6039
Anonymous credentials: rate-limit bots and agents without compromising privacy
As AI agents change how the Internet is used, they create a challenge for security. We explore how Anonymous Credentials can rate limit agent traffic and block abuse without tracking users or compromising their privacy.
6040
Yes, Claude can do nine loops
Claude computes a nine-loop amplitude in N=4 super-Yang-Mills
6041
Using WolframAlpha to Hack Text CAPTCHA
I’ve been playing around with the Text CAPTCHA demo page and wondered how well WolframAlpha is at logic questions. As it turns out, Wolfram is pretty smart! Although, since a CAPTCHA requir…
6042
5 Things to Try with Gemini 3 Pro in Gemini CLI
Unlock state-of-the-art coding and reasoning in your terminal. Gemini 3 Pro is now in Gemini CLI for Google AI Ultra and paid API subscribers.
6043
6044
Robots Must Be Ephemeralized
There is a subfield of robotics research called “sim-to-real” (sim2real) whereby one attempts to solve a robotic task in simulation, and the...
6045
Object-recognition dataset stumped the world’s best computer vision models
When computer vision detectors are turned loose in the real world, their performance noticeably drops. In an effort to close this performance gap, a team of MIT and IBM researchers set out to create a very different kind of object-recognition dataset called ObjectNet.
6046
Shapelearn Qwen 3.8 27B (13.1 GB VRAM)
ByteShape's full ShapeLearn release of Qwen 3.8 27B: five GGUF models on the quality-speed frontier across six GPUs. GPU-5 is the default wherever it fits, at 99.63% of BF16. GPU-4 stays very competitive in a much smaller file.
6047
Signal processing is key to embedded machine learning
When we hear about ML - whether it’s about machines learning to play Go or computers generating plausible human language - we often think about deep learning.
6048
6049
Making a vintage LLM from scratch
In this blog post, I will share the adventures I had creating my own LLM, from (almost) scratch, trained only on old texts. I made my own base-training and fine-tuning scripts, data processing pipelines and custom datasets. ("almost from scratch" means I did use existing programming languages and libraries, I didn't…
6050
Red queen hypothesis – A new way forward for self-improving AI
At a time when there's widespread interest in self-improving AI, researchers here have tackled one of the central challenges affecting its development by using an approach adapted from evolutionary biology.
6051
BERTopic: The Future of Topic Modeling
BERTopic takes advantage of the superior language capabilities of (not yet sentient) transformer models and uses some other ML magic like UMAP and HDBSCAN to produce what is one of the most advanced techniques in language topic modeling today.
6052
6053
Product Quantization: Compressing high-dimensional vectors
Product quantization (PQ) is a popular method for dramatically compressing high-dimensional vectors to use 97% less memory, and for making nearest-neighbor search speeds 5.5x faster in our tests.
6054
TorchLean: Formalizing Neural Networks in Lean
Neural networks are increasingly deployed through software stacks whose behavior is defined by tensor layouts, preprocessing choices, compiler boundaries, floating-point arithmetic, GPU kernels, and external model exporters. TorchLean is a Lean 4 codebase for neural network specification, execution, and verification…
6055
6056
Implementing the Goodfellow GANs paper
Generative Adversarial Networks (GANs), discovered by Ian Goodfellow in 2014, were an early method in the area of generative AI. I will focus on image generation as set out in the paper Generative Adversarial Nets 1 . This paper is the focal point of the post, I will guide you through the implementation of the paper…
6057
6058
Empirical Bayes for multiple sample sizes
The James-Stein Estimator with unequal variances
6059
Playing with genetic algorithms in Python
Edit: This post made it to the front page of Hacker News! You can find the discussion here.
6060
Boston Dynamics' Handle Robot Recreated with Raspberry Pi
Maker Harry was inspired by Handle, Boston Dynamics' research robot, yo build his own mini version named MABEL.