2851
Audio Decomposition – open-source seperation of music to constituent instruments
Matthew Bird Portfolio
New
The latest blog posts to take off on Hacker News. 6,331 posts.
2852
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.
2853
LLMs have reached a point of diminishing returns
Science, sociology, and the likely financial collapse of the Generative AI bubble
2854
Diffusion models are evolutionary algorithms
Authors: Yanbo Zhang, Benedikt Hartl, Hananel Hazan, Michael Levin
2855
2856
SVDQuant: 4-Bit Quantization Powers 12B Flux on a 16GB 4090 GPU with 3x Speedup
A new post-training training quantization paradigm for diffusion models, which quantize both the weights and activations of FLUX.1 to 4 bits, achieving 3.5× memory and 8.7× latency reduction on a 16GB laptop 4090 GPU. Code: https://www.github.com/mit-han-lab/nunchaku
2857
Perceptually lossless (talking head) video compression at 22kbit/s
Achieving extreme video compression rates using the LivePortrait model for talking head videos.
2858
2859
Combining Machine Learning and Homomorphic Encryption in the Apple Ecosystem
At Apple, we believe privacy is a fundamental human right. Our work to protect user privacy is informed by a set of privacy principles, and…
2860
Ollama 0.4 is released with support for Meta's Llama 3.2 Vision models locally
Llama 3.2 Vision 11B and 90B models are now available in Ollama.
2861
The deep learning boom caught almost everyone by surprise
"You’ve taken this idea way too far," a mentor told Prof. Fei-Fei Li.
2862
What Shapes Do Matrix Multiplications Like?
Divining order from the chaos
2863
Low-poly image generation using evolutionary algorithms in Ruby (2023)
A deep dive into evolutionary algorithms through the lens of low-poly image generation. This exploration covers genetic algorithm theory, custom Ruby implementation using the Petri Dish framework, and detailed analysis of genetic operators including roulette wheel selection, random midpoint crossover, and nudge…
2864
2865
SmolLM2
New from Loubna Ben Allal and her research team at Hugging Face: SmolLM2 is a family of compact language models available in three size: 135M, 360M, and 1.7B parameters. They …
2866
Using Large Language Models to Catch Vulnerabilities
Posted by the Big Sleep team Introduction In our previous post, Project Naptime: Evaluating Offe...
2867
2868
Our First Generalist Policy
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.
2869
ChatGPT Search
Get fast, timely answers with links to relevant web sources
2870
Wonder Animation – Video to 3D Animation
Wonder Dynamics (Autodesk company), launches Wonder Animation, groundbreaking video-to-3D scene technology that accelerates animated film production by turning any video sequence into a 3D-animated scene.
2871
SimpleQA
A factuality benchmark called SimpleQA that measures the ability for language models to answer short, fact-seeking questions.
2872
Pushing the frontiers of audio generation
Our pioneering speech generation technologies are helping people around the world interact with more natural, conversational and intuitive digital assistants and AI tools.
2873
Model Predictive Control in the Browser with WebAssembly
wrenfold is a framework for converting symbolic mathematical expressions (written in Python) into generated numerical code in other languages (C++, Rust, Python). It aims to bridge the gap between prototyping of functions in expressive symbolic form, and performant production code. wrenfold is particularly relevant to…
2874
Vector databases are the wrong abstraction
Today’s vector databases disconnect embeddings from their source data. We should treat embeddings more like database indexes—here’s how.
2875
2876
2877
Using reinforcement learning and $4.80 of GPU time to find the best HN post
The OpenPipe platform has migrated to Weights & Biases and CoreWeave. Browse migration details and the OpenPipe technical archive.
2878
Moonshine, the new state of the art for speech to text
Can you imagine using a keyboard where it took a key press two seconds to show up on screen? That’s the typical latency for most voice interfaces, so it’s no wonder they’ve failed…
2879
ZombAIs: From Prompt Injection to C2 with Claude Computer Use
A few days ago, Anthropic released Claude Computer Use, which is a model + code that allows Claude to control a computer. It takes screenshots to make …
2880
Detecting when LLMs are uncertain
A deep dive into a new reasoning technique called Entropix.