Opinions all my own. APPLIED AI | DESIGN
Sydney Australia
Joined September 2008
- Tweets1.9K
- Following1.7K
- Followers304
- Likes3.5K
Mehran Mozaffari retweeted
🎉 GraphRAG in @graphlit
Learn how to extract a knowledge graph from your content, using LLMs, and enrich entities with Wikipedia.
Use the KG to provide greater context to your RAG pipeline, with GraphRAG.
Blog:
graphlit.com/blog/graphrag-u…
Code:
github.com/graphlit/graphlit…
Demo:
graphlit-samples-sharepoint-…
Mehran Mozaffari retweeted
Anthropic has just revealed some exciting news about Claude Sonnet. They've successfully identified how millions of concepts are represented inside this massive model!
LLMs are usually treated as black boxes.
We feed them data, they give us responses, but understanding the 'why' behind their actions is challenging.
Their research used dictionary learning, a technique to find patterns of activated neurons or features.
These features provide a window into the model's understanding of various concepts.
Anthropic mapped out the internal state of Sonnet, finding features for specific entities like cities, people, and atomic elements.
These features are multimodal and multilingual, responding to text and images in multiple languages.
They even found abstract features related to concepts like inner conflict, relationship breakups, and logical inconsistencies.
By manipulating these features, researchers could alter Claude's behavior. For example, activating the "Golden Gate Bridge" feature made Claude obsessed with the bridge, mentioning it in almost every response.
Read more here -> anthropic.com/research/mappi…
Mehran Mozaffari retweeted
This is ground breaking.
University of Texas scientists used AI to turn a person’s thoughts into readable text 🤯
Alex Huth, assistant professor of neuroscience in computer science at UT-Austin worked with doctoral student, Jerry Tang, to develop what is called a “semantic decoder.”
This technology uses artificial intelligence to interpret brain activity and can allow communication with someone who is mentally conscious but may be unable to speak, like someone who’s had a stroke.
Using an MRI scanner, ideas and thoughts are mapped out, and an AI algorithms is then used to read those ideas out into words.
And this uses GPT-1 from 2019, imagine GPT-4's ability to decode human thoughts?
Mehran Mozaffari retweeted
In order to edit an image, you need to know something about its perspective. In this work, we propose "Perspective Fields", a new image space representation of camera perspective consisting of an up vector and latitude value per-pixel.
Mehran Mozaffari retweeted
Big news: signups for our new course "From Deep Learning Foundations to Stable Diffusion" are now available.
(And if you're an open source contributor, you can apply for a scholarship to join for free!)
fast.ai/posts/part2-2022-sig…
Mehran Mozaffari retweeted
We've trained a neural net called Whisper that approaches human-level robustness and accuracy on English speech recognition. It performs well even on diverse accents and technical language. Whisper is open source for all to use. openai.com/blog/whisper/
Mehran Mozaffari retweeted
Parsr transforms PDF, documents, and images into enriched structured data
github.com/axa-group/Parsr
Mehran Mozaffari retweeted
Mathematics for Machine Learning -- a 47-page introduction from UC Berkeley 🚀
• Linear Algebra
• Calculus and Optimization
• Probability
A 100% free resource!
Source: gwthomas.github.io/docs/math…
Mehran Mozaffari retweeted
The "Software Engineering at Google" book is now free to read online: bit.ly/swe-book. Great insights on improving code quality & scaling engineering teams.
ALT Today, software engineers need to know not only how to program effectively but also how to develop proper engineering practices to make their codebase sustainable and healthy. This book emphasizes this difference between programming and software engineering. How can software engineers manage a living codebase that evolves and responds to changing requirements and demands over the length of its life? Based on their experience at Google, software engineers Titus Winters and Hyrum Wright, along with technical writer Tom Manshreck, present a candid and insightful look at how some of the world’s leading practitioners construct and maintain software. This book covers Google’s unique engineering culture, processes, and tools and how these aspects contribute to the effectiveness of an engineering organization.
Mehran Mozaffari retweeted
Delighted my first interactive for The Economist is published! Team effort with my @ECONdailycharts colleagues
It shows the most-streamed song on Spotify, each week 2017-22, across 70 countries, coloured by lyrics language
Explore + 🎧 the songs
economist.com/global-music 🎤🎸
Mehran Mozaffari retweeted
Graphics engineer Bartosz Ciechanowski (@BCiechanowski) writes an accessible & interactive piece (with minimal math) on the foundations of light and the computational geometry principles that allow us to simulate it in software. ciechanow.ski/lights-and-sha…
Mehran Mozaffari retweeted
Computational Origami beta.rabbitear.org/diagram/ & rabbitear.org/docs/ & github.com/robbykraft/Origam… by @RobbyKraft
Mehran Mozaffari retweeted
This changes everything. 🤯
With GPT-3, I built a Figma plugin to design for you.
I call it "Designer"
Mehran Mozaffari retweeted
Here's a sentence describing what Google's home page should look and here's GPT-3 generating the code for it nearly perfectly.
Mehran Mozaffari retweeted
Incredibly inspiring work from the NY Times RnD division @NYTimesRD finding a native way to incorporate PG scans in an article - rd.nytimes.com/projects/reco…
animation is controlled by scrolling. Feels really great,
Mehran Mozaffari retweeted
Introducing Jukebox, a neural net that generates music, including rudimentary singing, as raw audio in a variety of genres and artist styles. We're releasing a tool for everyone to explore the generated samples, as well as the model and code: openai.com/blog/jukebox/
Mehran Mozaffari retweeted
There's a Facebook group with 6,000 members dedicated to designing open-source COVID-19 medical supplies.
The group is sharing CAD files for ventilators, respirators, and equipment that can be fabricated in emergency conditions.
Absolutely inspiring
facebook.com/groups/67093222…
Mehran Mozaffari retweeted
Bill Gates in 2015:
“Today, the greatest risk of global catastrophe...it’s most likely to be a highly infectious virus, rather than a war — not missiles but microbes.”
teapawt.net/xrpbillionair/status/1…
Mehran Mozaffari retweeted
With the 2010's coming to a close, we didn't want to miss the opportunity to reflect on a major shift in tech this decade -- The rise of design.
12 industry experts explain how it happened.
Take a seat, grab a coffee and read our 3,137 word recap 😅: figma.com/blog/the-rise-of-u…
