The Truth of AI retweeted
Improving LLMs' math reasoning ability is actually not as easy as what people anticipated. There are some interesting observations:
1. Using textbooks does not improve models' reasoning a lot. We tried to use the 800 high-quality math textbooks from Mathpile to continue training
These tweets about git and devops and releasing features is really weird. Is everything okay at @OpenAI
Hmmm.... a search engine for LLM agents. It will be interesting to see which directions this is going to take the field to.
Today, I'm sharing AgentSearch, an open-core effort to make humanity's knowledge accessible for LLM agents.
To start, I have embedded all of Wikipedia, Arxiv, filtered common crawl, and more.
The result is over 1 billion embedding vectors, try it at search.sciphi.ai/.
1
1
137
Great visualization of transformers
How Transformers Work
This is probably one of the most beautiful visualizations of how today's LLMs work.
ig.ft.com/generative-ai/
24
The Truth of AI retweeted
Here, in full directly on Twitter, is "A Hackers' Guide to Language Models". This 90 minute tutorial is designed to be the one place I point coders at when they ask "hey, tell me everything I need to know about LLMs!"
It covers both @OpenAI models and open source ones in depth.
Here's why the AI community is constantly missing the mark, and the public/media misunderstand AI advances. Great paper by @MelMitchell1
bdtechtalks.com/2021/05/03/a…
Deep learning is not magic bdtechtalks.com/2021/01/28/d… via @bdtechtalks
Educate yourself on the basics of AI
Understand the difference between academic and commercial AI
Know the importance of the people creating the AI
Don't botch your AI pitches!
bdtechtalks.com/2021/01/06/h… via @bdtechtalks
The adversarial ML Threat Matrix is a good overview of how machine learning/deep learning systems might be compromised by malicious actors and how you can protect yourself.
bdtechtalks.com/2020/10/26/a…
OpenAI started out as a non-profit AI research lab. This year, it will launch its first commercial product. Here's how GPT-3 transformed its own creator.
bdtechtalks.com/2020/08/17/o…
Here's why AI doesn't understand the world like we do
#ArtificialIntelligence #DeepLearning
bdtechtalks.com/2020/07/13/a…
Should AI have to explain itself? @moreisdifferent provides a blueprint for self-explainable AI
#ArtificialIntelligence
bdtechtalks.com/2020/06/15/s…
The things you can't see are just as important as pixels
#ArtificialIntelligence #computervision
bdtechtalks.com/2020/06/01/a…
A deep-dive into @GaryMarcus's plan to solve #AI's fundamental problems
#ArtificialIntelligence
bdtechtalks.com/2020/03/04/g…
Understanding the limits of CNNs, one of AI’s greatest achievements
#neuralnetworks #ArtificialIntelligence #deeplearning
bdtechtalks.com/2020/03/02/g… via @bdtechtalks










