Week 5 of 2024- W052024
Look closer
Disclaimer: The views and opinions expressed in this blog are entirely my own and do not necessarily reflect the views of my current or any previous employer. This blog may also contain links to other websites or resources. I am not responsible for the content on those external sites or any changes that may occur after the publication of my posts.
End Disclaimer
“ ‘Maybe's’ now a healthy never” -Stephen Wilson Jr.
"In every opportunity you have, in every circumstance, you have a choice. Am I going to back down or am I going to do something about it? I have a choice and I’m going to do everything I can to be my best self on every single play." - George Kittle
News
AIML
Hottest Job in Corporate America? The Executive in Charge of A.I.
AI Officer Is the Hot New Job That Pays Over $1 Million
Sleeper Agents, LLM Safety, Finetuning vs. RAG, Synthetic Data, and More
Build a Large Language Model (From Scratch): Github Repo
How enterprises are using open source LLMs: 16 examples
Introducing RWKV - An RNN with the advantages of a transformer
Can AI Carry On a Designer’s Legacy? h/t CAH
Inside the Music Industry’s High-Stakes A.I. Experiments
New Theory Suggests Chatbots Can Understand Text
Strategy, Not Technology, Is the Key to Winning with GenAI
Painting
Time Transfixed, 1938 by Rene Magritte, Art Institute of Chicago
From wikipedia entry: “The title of the painting translates to English literally as ‘Ongoing Time Stabbed by a Dagger’ and Magritte was reportedly unhappy with the generally accepted translation of ‘Time Transfixed’. “
Crypto
FTX Expects to Fully Repay Customers but Won't Restart Defunct Crypto Exchange
Tether reports nearly $3 billion profit in fourth quarter
Cyber
RSA is deceptively simple (and fun)
How a mistakenly published password exposed Mercedes-Benz source code
FBI director warns Chinese hackers aim to 'wreak havoc' on U.S. critical infrastructure
Markets
New York Community Bancorp Stock Plunges 38%, Reigniting Fears for Regional Banks
Major Asset Classes | January 2024 | Performance Review
What’s the right level of inflation?
A Famed Analyst’s Final Forecast Is the Fall of the U.S. Economy
Why Tim Cook Is Going All In on the Apple Vision Pro
Why Walmart pays its truck drivers 6 figures
Sports Bets and Financial Disasters Share Some Traits
Misc
The Perverse Policies That Fuel Wildfires
Elon Musk Says Neuralink Has Implanted Brain Chip in Human
Inventing the Perfect College Applicant
How Do You Make a Weed Empire? Sell It Like Streetwear.
This Is What a $10 Million Vacation Looks Like
How to Build an Origami Computer
Phones Track Everything but Their Role in Car Wrecks
Billy Corgan On What Makes A Great Songwriter
Is This Amelia Earhart’s Long-Lost Plane?
He hit three monster bets — and then the sportsbook wouldn’t pay
The Surprisingly Simple Math Behind Puzzling Matchups
Average Age that Men and Women were First Married in Each State
Meet the company that sells your lost airplane luggage
Papers
Fine-Tuning or Retrieval? Comparing Knowledge Injection in LLMs
Objective: Evaluate LLMs' ability to memorize, understand, and retrieve factual data using different approaches.
Methods Compared: Unsupervised fine-tuning and RAG.
Experiments: Conducted on various knowledge-intensive tasks across different topics.
Findings: RAG consistently outperforms fine-tuning for both existing and new knowledge.
Challenges with Fine-Tuning: Struggles to teach new factual information to LLMs.
Importance of Repetition: Repetition in training can improve LLMs' knowledge absorption.
RAG's Advantages: Better at incorporating relevant external information into responses.
Book
This Is How They Tell Me the World Ends: The Cyberweapons Arms Race
One of the best books that I’ve read in the space.