Sun.Dec 17, 2023

article thumbnail

Training Large-Vocabulary Neural Language Model by Private Federated Learning for Resource-Constrained Devices

Machine Learning Research at Apple

*= Equal Contributors Federated Learning (FL) is a technique to train models using data distributed across devices. Differential Privacy (DP) provides a formal privacy guarantee for sensitive data. Our goal is to train a large neural network language model (NNLM) on compute-constrained devices while preserving privacy using FL and DP. However, the DP-noise introduced to the model increases as the model size grows, which often prevents convergence.

162
162
article thumbnail

How Africans are changing French, one joke, rap and book at a time

Hacker News

More than 60 percent of French speakers now live in Africa. Despite growing resentment at France, Africans are contributing to the evolution and spread of the French language.

182
182
professionals

Sign Up for our Newsletter

This site is protected by reCAPTCHA and the Google Privacy Policy and Terms of Service apply.

Trending Sources

article thumbnail

Leveraging Large Language Models for Exploiting ASR Uncertainty

Machine Learning Research at Apple

With the help of creative prompt engineering and in-context learning, large language models (LLMs) are known to generalize well on a variety of text-based natural language processing (NLP) tasks. However, for performing well on spoken language understanding (SLU) tasks, LLMs either need to be equipped with in-built speech modality or they need to rely on speech-to-text conversion from an off-the-shelf automation speech recognition (ASR) system.

article thumbnail

US nuclear-fusion lab enters new era: achieving 'ignition' over and over

Hacker News

Researchers at the National Ignition Facility are consistently creating reactions that make more energy than they consume. Researchers at the National Ignition Facility are consistently creating reactions that make more energy than they consume.

182
182
article thumbnail

Apache Airflow®: The Ultimate Guide to DAG Writing

Speaker: Tamara Fingerlin, Developer Advocate

In this new webinar, Tamara Fingerlin, Developer Advocate, will walk you through many Airflow best practices and advanced features that can help you make your pipelines more manageable, adaptive, and robust. She'll focus on how to write best-in-class Airflow DAGs using the latest Airflow features like dynamic task mapping and data-driven scheduling!

article thumbnail

Bootstrap Your Own Variance

Machine Learning Research at Apple

This paper was accepted at the workshop Self-Supervised Learning - Theory and Practice at NeurIPS 2023. *=Equal Contributors Understanding model uncertainty is important for many applications. We propose Bootstrap Your Own Variance (BYOV), combining Bootstrap Your Own Latent (BYOL), a negative-free Self-Supervised Learning (SSL) algorithm, with Bayes by Backprop (BBB), a Bayesian method for estimating model posteriors.

article thumbnail

A bullying scandal closed a historic astronomy department

Hacker News

Lund University in Sweden reorganizes in the face of years of accusations against two professors. Lund University in Sweden reorganizes in the face of years of accusations against two professors.

181
181

More Trending

article thumbnail

AMD MI300X 30% higher performance than Nvidia H100, even with optimized stack

Hacker News

AMD continued to engage with Nvidia's claims by showing some of its results, including further optimizations, while others accounted for latency, typically the situation with servers.

181
181
article thumbnail

AI Briefing: Creative ways companies used AI in advertising in 2023

Flipboard

As the generative AI boom took off in early 2023 , there were plenty of predictions about how soon — and to what extent — companies would start using generative AI in their marketing. Nearly a year later, it’s clear many are feeling increasingly comfortable. Just last week, the global hotel search company Trivago released a new ad campaign as part of a broader rebrand.

AI 179
article thumbnail

AMD's CDNA 3 Compute Architecture

Hacker News

AMD has a long history of vying for GPU compute market share. Ever since Nvidia got first dibs with their Tesla architecture, AMD has been playing catch up. Terascale 3 moved from VLIW5 to VLIW4 to improve execution unit utilization in compute workloads.

179
179
article thumbnail

LLMs unleashed: Navigating the chaos of online experimentation

Flipboard

In an audacious move that defies conventional wisdom, generative AI companies have embraced a cutting-edge approach to quality assurance: Releasing large language models (LLMs) directly into the wild, untamed realms of the internet.

AI 175
article thumbnail

Optimizing The Modern Developer Experience with Coder

Many software teams have migrated their testing and production workloads to the cloud, yet development environments often remain tied to outdated local setups, limiting efficiency and growth. This is where Coder comes in. In our 101 Coder webinar, you’ll explore how cloud-based development environments can unlock new levels of productivity. Discover how to transition from local setups to a secure, cloud-powered ecosystem with ease.

article thumbnail

Bricks Can Absorb Traffic Noise – Thesis Preso on Helmholtz Resonators [video]

Hacker News

I gave a talk on my Masters Thesis Project in Architecture focusing on Altering Soundscapes in Exterior Environments using Helmholtz Resonators in Ceramic Br.

181
181
article thumbnail

This AI code that detects when guns, threats appear on school cameras is available for free

Flipboard

A software company is offering a free artificial intelligence system that can help security detect weapons at schools in an effort to combat school shootings. Iterate.

article thumbnail

The unfortunate math behind consulting companies

Hacker News

Hiring even one person creates a lot less profit than you'd think, and launching a product rarely works out. Here are some tips for consulting companies.

180
180
article thumbnail

Sports Illustrated’s AI scandal highlights the need for authenticity in the LLM era

Flipboard

Transparency and authenticity are critical for organizations that intend to use AI in any shape and form. As 2023 comes to a close, it has become evident that the once-unstoppable global AI race is quickly transforming into a mad dash.

AI 174
article thumbnail

Apache Airflow®: The Ultimate Guide to DAG Writing

Speaker: Tamara Fingerlin, Developer Advocate

In this new webinar, Tamara Fingerlin, Developer Advocate, will walk you through many Airflow best practices and advanced features that can help you make your pipelines more manageable, adaptive, and robust. She'll focus on how to write best-in-class Airflow DAGs using the latest Airflow features like dynamic task mapping and data-driven scheduling!

article thumbnail

Ventoy

Hacker News

Ventoy is an open source tool to create bootable USB drive for ISO files. With ventoy, you don't need to format the disk again and again, you just need to copy the iso file to the USB drive and boot it.

162
162
article thumbnail

AI-screened eye pics diagnose childhood autism with 100% accuracy

Flipboard

Researchers have taken photographs of children’s retinas and screened them using a deep learning AI algorithm to diagnose autism with 100% accuracy.

article thumbnail

Becoming a go-to person gets you promoted

Hacker News

To reach the Senior level, you need to increase your scope, impact, and ownership. One of the best ways to do that is to become a “go-to” person or an expert. You could be the “database expert”, “Python expert”, “caching expert”, “frontend expert”, “Typescript expert”, etc.

Database 160
article thumbnail

These AI tools can help you create something in under a minute

Flipboard

Cook up an AI-powered recipe with Taste Bud, summarize a YouTube video with Summarize.tech, and more. You can do some amazing things with AI in a minute or less. Read on for eight ways to start making something new. They’re all free, fast, and fun. 1.

AI 168
article thumbnail

The Cloud Development Environment Adoption Report

Cloud Development Environments (CDEs) are changing how software teams work by moving development to the cloud. Our Cloud Development Environment Adoption Report gathers insights from 223 developers and business leaders, uncovering key trends in CDE adoption. With 66% of large organizations already using CDEs, these platforms are quickly becoming essential to modern development practices.

article thumbnail

The Death of Caspar Hauser

Hacker News

Caspar Hauser died on 17 December 1833, but was it murder or a self-inflicted wound? Hauser’s mysterious death raised as many questions as his mysterious life.

165
165
article thumbnail

Amdocs Revolutionising Telecom with Generative AI

Flipboard

Last month, Amdocs and NVIDIA announced a strategic collaboration aimed at optimising LLMs to accelerate the adoption of generative AI applications …

AI 166
article thumbnail

Ancient 'Dragon' Unearthed in Japan Is Unlike Anything We've Ever Seen

Hacker News

More than 72 million years ago, the western seas of the Pacific Ocean were home to one of the fiercest ocean predators of all time.

180
180
article thumbnail

Intel CEO argues inference is better for the industry, Nvidia CUDA days are numbered

Flipboard

Intel rolled out a portfolio of AI products aimed at the data center, cloud, network, edge and PC at its AI Everywhere event in New York City last week. "Intel is on a mission to bring AI everywhere through exceptionally engineered platforms, secure solutions and support for open ecosystems," CEO.

AI 152
article thumbnail

15 Modern Use Cases for Enterprise Business Intelligence

Large enterprises face unique challenges in optimizing their Business Intelligence (BI) output due to the sheer scale and complexity of their operations. Unlike smaller organizations, where basic BI features and simple dashboards might suffice, enterprises must manage vast amounts of data from diverse sources. What are the top modern BI use cases for enterprise businesses to help you get a leg up on the competition?

article thumbnail

Intel proposes XeGPU dialect for LLVM MLIR

Hacker News

Motivation To support high-performance GEMM code generation on Intel GPU, we propose XeGPU dialect. XeGPU dialect provides an abstraction that closely models Xe instructions. XeGPU ops are introduced when a special Xe instruction can’t be expressed by LLVM/SPIR-V dialect, for example, like matrix instruction (AKA DPAS) and 2D block load. It matches the hardware instructions’ semantics including the matrix sizes.

146
146
article thumbnail

DataComp: In Search of the Next Generation of Multimodal Datasets

Machine Learning Research at Apple

*=Equal Contributors Multimodal datasets are a critical component in recent breakthroughs such as Stable Diffusion and GPT-4, yet their design does not receive the same research attention as model architectures or training algorithms. To address this shortcoming in the ML ecosystem, we introduce DataComp, a testbed for dataset experiments centered around a new candidate pool of 12.8 billion image-text pairs from Common Crawl.

ML 124
article thumbnail

Qatar Airways Bans YouTuber for Negative Review

Hacker News

Qatar Airways has banned YouTube flight reviewer Josh Cahill for posting a negative review. Here's the backstory.

180
180
article thumbnail

Navigating The AI Wave And The Impact On The 2024 Workplace

Flipboard

AI use is skyrocketing in the workplace, with over half of companies now employing generative AI tools to enhance productivity and creativity, despite concerns about regulation and job loss. At JLL, a proprietary AI tool streamlines tedious tasks, allowing employees to focus on high-level work, demonstrating AI’s ability to reveal insights quickly and aid in complex decision-making.

AI 145
article thumbnail

Introducing CDEs to Your Enterprise

Explore how enterprises can enhance developer productivity and onboarding by adopting self-hosted Cloud Development Environments (CDEs). This whitepaper highlights the simplicity and flexibility of cloud-based development over traditional setups, demonstrating how large teams can leverage economies of scale to boost efficiency and developer satisfaction.

article thumbnail

ArtBot for Stable Diffusion

Hacker News

Generate AI-created images and photos with Stable Diffusion using a distributed computing cluster powered by the AI Horde. No login required and free to use.

article thumbnail

We Should Reject Using AI to Create Perfect Memories

Flipboard

The tech industry is obsessed with the idea of being able to remember everything, but when it comes to your personal life, there need to be limits.

AI 158
article thumbnail

3M knew its chemicals were harmful decades ago, but didn't tell the public

Hacker News

3M knew its chemicals were toxic decades ago, but didn’t tell the public or government, internal documents show.

179
179
article thumbnail

Combining Gemini Pro AI with LangChain to create a mini Retrieval-Augmented Generation (RAG) system

Flipboard

In the rapidly evolving world of language processing, the integration of advanced tools like Gemini Pro with LangChain is a significant step forward for …

AI 153
article thumbnail

Marketing Operations in 2025: A New Framework for Success

Speaker: Mike Rizzo, Founder & CEO, MarketingOps.com and Darrell Alfonso, Director of Marketing Strategy and Operations, Indeed.com

Though rarely in the spotlight, marketing operations are the backbone of the efficiency, scalability, and alignment that define top-performing marketing teams. In this exclusive webinar led by industry visionaries Mike Rizzo and Darrell Alfonso, we’re giving marketing operations the recognition they deserve! We will dive into the 7 P Model —a powerful framework designed to assess and optimize your marketing operations function.