Hacker News with Generative AI: Machine Learning

Simons Foundation Launches Collaboration on Ecological Neuroscience (simonsfoundation.org)
The newly launched Simons Collaboration on Ecological Neuroscience (SCENE) will unite leading scientists across neuroscience and machine learning to discover how the brain represents ‘sensorimotor’ (that is, sensory and motor) interactions.
Lossless LLM compression for efficient GPU inference via dynamic-length float (arxiv.org)
Large Language Models (LLMs) have grown rapidly in size, creating significant challenges for efficient deployment on resource-constrained hardware.
Paper2Code: Automating Code Generation from Scientific Papers (arxiv.org)
Despite the rapid growth of machine learning research, corresponding code implementations are often unavailable, making it slow and labor-intensive for researchers to reproduce results and build upon prior work.
The Urgency of Interpretability (darioamodei.com)
In the decade that I have been working on AI, I’ve watched it grow from a tiny academic field to arguably the most important economic and geopolitical issue in the world.
Docker Model Runner Brings Local LLMs to Your Desktop (thenewstack.io)
Show HN: Lemon Slice Live – Have a video call with a transformer model (ycombinator.com)
Hey HN, this is Lina, Andrew, and Sidney from Lemon Slice. We’ve trained a custom diffusion transformer (DiT) model that achieves video streaming at 25fps and wrapped it into a demo that allows anyone to turn a photo into a real-time, talking avatar.
Show HN: High-performance GenAI engine now open source (github.com/arthur-ai)
Make AI work for Everyone - Monitoring and governing for your AI/ML
Double Descent Demystified: size of smallest non-zero singular value of X (arxiv.org)
Double descent is a surprising phenomenon in machine learning, in which as the number of model parameters grows relative to the number of data, test error drops as models grow ever larger into the highly overparameterized (data undersampled) regime.
FontDiffuser: Text to Font (yeungchenwa.github.io)
Automatic font generation is an imitation task, which aims to create a font library that mimics the style of reference images while preserving the content from source images.
Improving Deep Learning with a Little Help from Physics (quantamagazine.org)
Rose Yu has a plan for how to make AI better, faster and smarter — and it’s already yielding results.
"Periodic table of machine learning" could fuel AI discovery (news.mit.edu)
MIT researchers have created a periodic table that shows how more than 20 classical machine-learning algorithms are connected.
Yann LeCun "Mathematical Obstacles on the Way to Human-Level AI" (youtube.com)
Product Quantization: Compressing high-dimensional vectors by 97% (pinecone.io)
π0.5: A VLA with open-world generalization (pi.website)
Robots have come a long way over the past few years—they can perform impressive acrobatic feats, dance on stage, follow language commands and, in some of our own results, perform complex tasks like folding laundry or cleaning off a table. But the biggest challenge in robotics is not in performing feats of agility or dexterity, but generalization: the ability to figure out how to correctly perform even a simple task in a new setting or with new objects.
Are polynomial features the root of all evil? (2024) (alexshtf.github.io)
It turns out that it’s just a MYTH. There’s nothing inherently wrong with high degree polynomials, and in contrast to what is typically taught, high degree polynomials are easily controlled using standard ML tools, like regularization. The source of the myth stems mainly from two misconceptions about polynomials that we will explore here. In fact, not only they are great non-linear features, certain representations also provide us with powerful control over the shape of the function we wish to learn.
Show HN: Morphik – Open-source RAG that understands PDF images, runs locally (github.com/morphik-org)
Morphik is an alternative to traditional RAG for highly technical and visual documents.
Algebraic Semantics for Machine Knitting (uwplse.org)
As programming languages researchers, we’re entitled to a certain level of mathematical rigor behind the languages we write and analyze.
Introduction to Graph Transformers (kumo.ai)
Graphs are everywhere. From modeling molecular interactions and social networks to detecting financial fraud, learning from graph data is powerful—but inherently challenging.
Fox Succeeds in Scrapping Machine Learning Claims at CAFC Under 101 (ipwatchdog.com)
The U.S. Court of Appeals for the Federal Circuit (CAFC) on Friday addressed an issue of first impression in the patent eligibility context, holding that “claims that do no more than apply established methods of machine learning to a new data environment” are not patent eligible.
Show HN: I'm 15 and built a neural network from scratch in C++,just math (github.com/muchlakshay)
A C++ implementation of a Multilayer Perceptron (MLP) neural network using Eigen, supporting multiple activation and loss functions, mini-batch gradient descent, and backpropagation for training.
Forecaster reacts: METR's bombshell paper about AI acceleration (peterwildeford.substack.com)
About a month ago, METR, an AI evaluations organization, published a bombshell graph and paper that says that AI is accelerating quickly. With OpenAI’s recent launch of o3 and o4-mini, we can see if these predictions hold up.
Show HN: Keep your PyTorch model in VRAM by hot swapping code (github.com/valine)
This is an example of how to hotswap PyTorch training code without unloading your model weights from VRAM.
Pushing the Limits of LLM Quantization via the Linearity Theorem (arxiv.org)
Quantizing large language models has become a standard way to reduce their memory and computational costs.
Gemma 3 QAT Models: Bringing AI to Consumer GPUs (googleblog.com)
Last month, we launched Gemma 3, our latest generation of open models. Delivering state-of-the-art performance, Gemma 3 quickly established itself as a leading model capable of running on a single high-end GPU like the NVIDIA H100 using its native BFloat16 (BF16) precision.
Sparsely-Gated Mixture of Experts (MoE) (thegreenplace.net)
In transformer models, the attention block is typically followed by a feed forward layer (FF), which is a simple fully-connected NN with a hidden layer and nonlinearity.
Welcome to the Era of Experience [pdf] (googleapis.com)
A curated blog for learning LLM internals: tokenize, attention, PE, and more (ycombinator.com)
I've been diving deep into the internals of Large Language Models (LLMs) and started documenting my findings.
Microsoft's "1‑bit" AI model runs on a CPU only, while matching larger systems (arstechnica.com)
Future AI might not need supercomputers thanks to models like BitNet b1.58 2B4T.
Ask HN: What are the hottest areas of *non*-LLM AI work currently? (ycombinator.com)
10+ years ago, "AI" would likely refer to work in RL, evolutionary/genetic algorithms, etc.
Inferring the Phylogeny of Large Language Models (arxiv.org)
This paper introduces PhyloLM, a method adapting phylogenetic algorithms to Large Language Models (LLMs) to explore whether and how they relate to each other and to predict their performance characteristics.