Multiverse School Newsletter - 10/6/24
ðļ 10/6/2024 Multiverse School Weekly Newsletter ðļ
Hey Multiverse Crew-
Whew, lots of changes this week. Elena visited, Roan went back to the UK, and I got to meet back up with a bunch of friends I lost contact with over the years. Thrilled to see folks. I did a Corporate Gig on Oct 1-3, which helps me to pay the rent.
OpenAI Dev Day happened, and almost everyone from the original OpenAI has quit, due to Sam Altman being Littlefinger (I hate it when Musk says a true thing but stopped watches are right twice a day etc). I'm working on an update to the Using Language Models curriculum to cover the new Canvas mode.
If you've been waiting for my Autogen Class its called Swarm Architecture - it got pushed because of low enrollment. Sign up here. That means we're almost done with the first run of all of the Prompt Engineering classes! Haha, it took 9 months and most people didn't make it all the way through because they were overwhelmed and we needed to pretty much double the length of the classes by the 3rd cohort because senior developers and data scientists are struggling to keep up.
If you're waiting on an email from me... just wait longer sorry ððð I will hire someone to answer questions soon, it's ridiculous right now. I am so behind on everything, I'm excited to get a break week next week! So, in honor of taking time off, I'm headed to Detroit to take absolutely no time off. I'll sleep when I'm dead! Dead I say! ðīðŠĶ
I/O Psychologists and Psychometricians
Are you interested in the concept of Intelligence research, AGI, Psychometrics and are absolutely positive your unique field and set of interests will allow you to meaningfully contribute, but are not an I/O Psychologist or psychometrician?
You want to attend prompt engineering! You'll learn all about the process of building effective benchmarks there. We'll be discussing the research, and you'll learn quite a bit more in that class. You can help by attending this class and attempting to make your own benchmarks, this is literally the research. We'll discuss the forms of validity in that class, and you are able to contribute to the research by constructing your own benchmarks and having them reviewed. This will help us.
The end of 1:1s for now
1:1s are now a research team or membership benefit.
For the research team, apply by emailing me. I'll get to you as space opens up.
This Week's Classes
Did you sign up for this week's classes?
Monday, October 07
11:00 AM Prompt Engineering
Jailbreaks
Jailbreaking a large language model out of its alignment guidance.
Sign Up
05:00 PM Agent Engineering - OpenAI Toolchain
Allowing OpenAI Assistants to Use Tools
Dive into adding OAI-Formatted functions to OpenAI Assistants and managing tool use. This session emphasizes advanced integration techniques and practical strategies.
Sign Up
Tuesday, October 08
11:00 AM Senior Devs
Senior Devs The return of Senior Devs! Let's dig into something that recently came out. This is a research program, after all. Research program only.
02:00 PM Intro to Agents
Tool Use
Allowing AI Agents to use Tools
Sign Up
05:00 PM Using Large Language Models
Effective Use of Large Language Models as Assistants
The first session introduces participants to the basics of large language models, emphasizing practical applications such as improving model outputs through Emotional Priming and utilizing AI for brainstorming and summarizing tasks. It features hands-on activities to transform informal vocal transcripts into structured notes, create structured data from unstructured sources, and summarize long documents. The session also highlights the importance of understanding tokens in the inference process and compares market-available and open-source large language models. Practical exercises include modifying system instructions for enhanced problem-solving and generating structured, actionable insights from informal data.
Sign Up
Wednesday, October 09
11:00 AM Prompt Engineering
Jailbreaks LAB
The lab where we try to perform some jailbreaks, and dig into the research around Control Vectors.
Sign Up
02:00 PM Intro to Agents
Swarm Prototyping
Use multiple GPTs for conversation between personas and achieve predictably formatted outputs.
Sign Up
05:00 PM Using Large Language Models
Effective Use of Large Language Models in Analysis and Critique
The second session focuses on using large language models for editing, structured data creation, and critiquing. Participants learn to apply hyperparameters to improve inference outputs, use "Chain of Thought" methods to enhance reasoning, and systematically analyze and critique plans. The session includes hands-on exercises for editing with AI and creating structured data from unstructured sources. Additionally, it covers recognizing the models' limitations and biases and customizing system prompts for enhanced user preference alignment .
Sign Up
Thursday, October 10
11:00 AM Consciousness
Let's do some automated debates.
02:00 PM Research Program Meetup
Let's go over our progress! Research Program Only
05:00 PM Using Large Language Models
Effective and Ethical Use of Large Language Models
The third session delves into aligning AI outputs with human values and ethical considerations. It teaches participants to build simple rubrics for models to follow and employ the "Expert Reviewers" strategy to improve model outputs. Furthermore, the session includes exercises on prompt optimization and reflection techniques to ensure high-quality, reliable model responses. The ethical implications of using large language models are a core theme throughout this session, encouraging participants to critically engage with both the strengths and the weaknesses of these technologies .
Sign Up
Upcoming Tracks and Classes
Prompt Engineering - Spring Cohort
Full-stack Prompt Engineering, from the ground up
-
Swarm Architecture starts Wednesday, October 23 03:00 PM Unlock the power of Agent Swarms.
-
AI Alignment starts Monday, October 28 05:00 PM Navigating the Future of AI Responsibly
Prompt Engineering - Summer Cohort
Full-stack Prompt Engineering, from the ground up
-
Advanced Agent and Tool Engineering starts Monday, October 28 02:00 PM Mastering Intelligent Language Models
-
Swarm Architecture starts Monday, January 06 02:00 PM Unlock the power of Agent Swarms.
-
AI Alignment starts Monday, February 03 02:00 PM Navigating the Future of AI Responsibly
Prompt Engineering - Fall (M & W 11AM Pacific)
Full-stack Prompt Engineering, from the ground up
-
Intro to Agents starts Monday, December 02 01:00 PM Create Conscious Autonomous Applications
-
Agent Engineering - OpenAI Toolchain starts Monday, December 16 01:00 PM Master OpenAI Assistants: Creation, Management & Advanced Integration.
-
Agent Engineering - Open Source Toolchain starts Monday, January 06 11:00 AM Master Langchain Tools and Agents with Python or JavaScript.
-
Advanced Agent and Tool Engineering starts Monday, January 13 11:00 AM Mastering Intelligent Language Models
-
Swarm Architecture starts Monday, March 24 11:00 AM Unlock the power of Agent Swarms.
-
AI Alignment starts Monday, April 28 11:00 AM Navigating the Future of AI Responsibly
Prompt Engineering - Fall (T & TH 5PM Pacific)
Full-stack Prompt Engineering, from the ground up
-
Using Large Language Models starts Tuesday, October 08 05:00 PM Empower your decisions and creativity with AI, no tech background required.
-
Custom GPTs starts Wednesday, October 23 05:00 PM Enhance productivity with hands-on, accessible ChatGPT training.
-
Prompt Engineering starts Tuesday, November 05 05:00 PM Master Prompt Engineering
-
Intro to Agents starts Tuesday, January 07 05:00 PM Create Conscious Autonomous Applications
-
Agent Engineering - OpenAI Toolchain starts Tuesday, January 28 05:00 PM Master OpenAI Assistants: Creation, Management & Advanced Integration.
-
Agent Engineering - Open Source Toolchain starts Tuesday, February 11 05:00 PM Master Langchain Tools and Agents with Python or JavaScript.
-
Advanced Agent and Tool Engineering starts Tuesday, February 25 05:00 PM Mastering Intelligent Language Models
-
Swarm Architecture starts Tuesday, April 29 05:00 PM Unlock the power of Agent Swarms.
-
AI Alignment starts Tuesday, May 27 03:00 PM Navigating the Future of AI Responsibly
Important Links
- Did you even know we have a substack
- TikTok
- [YouTube](https://www.youtube.com/channel/UCOma6lSGSU42CzbNNI9QjUw
- Twitch
- GitHub
- Sign Up Here
- The In-Person Calendar
- The Multiverse Crew On Socials
- We're Running a Free Capture The Flag
- Sign Up Here
- The Hot New Website
- We released all learn-to-code curriculum for free
- Also all my previous curriculum is up here for free as well
- Also did you know I made one of the world's most effective algorithms classes and it is free on my youtube
- Did you see us make a rover work autonomously with just chatgpt
- Consciousness Podcast Episode One
- Here's a bunch of behind-the-scenes footage of the fellowship, by Roan
AI Tutorials and News
Biznaz
- This is what YC is funding (it's Agents)
- 30 under 30 is bullshit
- The Government is Ahead of You
- OpenAI Is Losing Everyone
Programming
Psychology
Cyberpunk Dystopia
- 23 And Me Systemically Erases Indigenous Families So I Am Glad The Company Is Dying Out, I Hope Someone Tells Whatever Her Name Is She Doesn't Exist As A Response Because She Doesn't, She Was Never A Real Person In The First Place, She's Just A Myth Some Families Tell
- Tipping Points in the Climate are going in Both Directions
- Recording Jammers are mostly BS
- Everyone Is Doxxed
- China's Robots Look Cooler Than Our Robots
- Uhh, The Matrix I Guess
- Here's What Happens Next
Culture
- Riker Go Home You're Drunk
- Uhh David Shapiro is Leaving AI?
- Does Alignment Matter
- Looks Like This is why David left AI, He Watched Our Podcast
Roan's Papers
StoryMaker, a framework to keep consistent characters across generated images. With consistent clothing and facial reproduction. Uses learnable background embeddings. Lots of examples of use cases. IP adapters, LoRAs, ControlNets etc feature heavily.https://t.co/iBcpdkEbHu
— cycloarkane (@cycloarkane) September 27, 2024
Cool novel RAG approach, using lightweight LLMs as memory modules with massive contexts while the heavy cognitive lifting is done by another model. The authors made their own UltraDomain benchmark which focuses on long contexts to test their system.https://t.co/ywLhHbEBC9
— cycloarkane (@cycloarkane) September 25, 2024
Kolmogorov-Arnold Transformer (KAT), a combination of KANs and transformers. Superior performance on on ImageNet1k compared to traditional MLP transformers, more scalable. KAT is more scalable than simple KANs as it uses group KAN shared functions.https://t.co/OR2XlmU2LH
— cycloarkane (@cycloarkane) September 21, 2024
V-STaR. Using incorrect reasoning and correct reasoning to gain even more substantial improvements through verification. Using a correctness label allows learning from bad reasoning. Authors only used LoRAs so full SFT may yield even better results.https://t.co/KpHcvksPtb
— cycloarkane (@cycloarkane) September 17, 2024
Efficient LiDAR odometry that reuses data and uses adaptive local map updates to limit the drift. I was wanting to explore LiDAR a bit more so I can understand Waymos better tbh lmao. C++ code available with the paper.https://t.co/kcsLSOWHAz
— cycloarkane (@cycloarkane) September 16, 2024
Paper on censorship and inferential adversaries. Bit draconian imo but good to know what might be around the corner. Basically about censoring information that could be aggregated to do bad stuff but isn't in itself disallowed.https://t.co/Tmjwc92qmr
— cycloarkane (@cycloarkane) September 15, 2024
More sampling can increase success by 300% and out perform SOTA models. This paper is a must read to understand ð. Implementations of this could be insane for OS models power on local compute, especially if you have time to do the computations.https://t.co/HxwtMsuxdJ
— cycloarkane (@cycloarkane) September 14, 2024
If you don't have time to read the OpenAI GPT-o1 (ð) paper I read it myself and put key insights in a markdown bulletpoint list. Some really wild stuff in there and a pretty candid analysis from OAI it seems.https://t.co/aJYLmlI1gk
— cycloarkane (@cycloarkane) September 12, 2024