New AI text diffusion models break speed barriers by pulling words from noise by WithinReason

These diffusion models maintain performance faster than or comparable to similarly sized conventional models. LLaDA’s researchers report their 8 billion parameter model performs similarly to LLaMA3 8B across various benchmarks, with competitive results on tasks like MMLU, ARC, and GSM8K. However, Mercury claims dramatic speed improvements. Their Mercury Coder Mini scores 88.0 percent on HumanEval

ByHackTechFebruary 28, 20250Comments

News

Diffusion Models Learn Low-Dimensional Distributions via Subspace Clustering by pizza

[Submitted on 4 Sep 2024 (v1), last revised 27 Dec 2024 (this version, v2)] View PDF HTML (experimental) Abstract:Recent empirical studies have demonstrated that diffusion models can effectively learn the image distribution and generate new samples. Remarkably, these models can achieve this even with a small number of training samples despite a large image dimension

ByHackTechFebruary 24, 20250Comments

News

Models of Generics and Metaprogramming: Go, Rust, Swift, D and More by TechSquidTV

In some domains of programming it’s common to want to write a data structure or algorithm that can work with elements of many different types, such as a generic list or a sorting algorithm that only needs a comparison function. Different programming languages have come up with all sorts of solutions to this problem: From

ByHackTechJanuary 14, 20250Comments

News

How to tune your DL models? by Exorust

This is not an officially supported Google product. Varun Godbole†, George E. Dahl†, Justin Gilmer†, Christopher J. Shallue‡, Zachary Nado† † Google Research, Brain Team ‡ Harvard University Who is this document for? Why a tuning playbook? Guide for starting a new project Choosing the model architecture Choosing the optimizer Choosing the batch size Choosing

ByHackTechJanuary 7, 20250Comments

News

Ask HN: Why Build Models from scratch? by jairuhme

Hello – for those Data Scientists/MLEs who build models from scratch (i.e. say building a least squares model without using sklearn, etc.), why? Are there performance benefits if the pre-built model isn't built with a faster language or maybe restrictions on imports?

ByHackTechOctober 23, 20230Comments

News

Compiling ML models to C for fun by signa11

NOTE: This post is going to be a compiler post, not a machine learning tutorial, so please treat it as such. Maybe it will still help you understand ML through a compilers lens. I had a nice chat with my friend Chris recently. He walked me through the basics of machine learning while I was

ByHackTechSeptember 20, 20230Comments

News

GCP to Add AI Models from Meta (Llama2), Anthropic (Cladue2) by archo

We’ve detected unusual activity from your

ByHackTechAugust 29, 20231Comment

News

Show HN: Diffusion Models for Greeting Cards by lotkowskim

Use AI to generate a completely pers

ByHackTechJuly 28, 20230Comments

News

Opto-Mechanical Eye Models (2022) by turtleyacht

1. IntroductionThe human eye is an organ that combines singular anatomical and histological characteristics with complex physiological properties to develop its function as an image formation “device”. Among the multiple conditions that can impair vision, the cataract is the most common one, and is the most frequent surgical intervention. The development of artificial intraocular lenses

ByHackTechJuly 8, 20231Comment

News

All Your (Advertising) Models Are Broken by djschnei

All Your (Advertising) Models are Broken The modern advertising model on centralized social media platforms is, to put it mildly, a shitshow. It’s annoying for users, gives barely any monetary value to creators, and is questionably effective for advertisers. What’s the Problem? Social media platforms collect vast amounts of user data and use that data

ByHackTechMay 10, 20230Comments

Sign Up to Our Newsletter