The pace of development surrounding Baby AGI and AutoGPT is mind blowing. Seems like a new *groundbreaking* update comes out every hour
Here are a few examples just from the past 24 hours
Groq is serving the fastest responses I've ever seen. We're talking almost 500 T/s!
I did some research on how they're able to do it. Turns out they developed their own hardware that utilize LPUs instead of GPUs. Here's the skinny:
Groq created a novel processing unit known as
AutoGPT and BabyAGI are yesterday’s news. Want to see what happens when AI agents have a conversation with each other?
Here are a couple examples of CAMEL!
Introducing
#AgentGPT
, an attempt at
#AutoGPT
directly in the browser 🤖
Give your own AI agent a goal and watch as it thinks, comes up with an execution plan and takes actions. Try for free now at
Baby-AGI by
@yoheinakajima
is taking the world by storm
Here's an implementation within the
@LangChainAI
framework, allowing you to easily substitute in other vectorstores and other LLMs
Docs:
I launched a website today that helps people find gigs in the Web3 economy. My goal is to better connect our communities and support the future of work. Help me spread the word with a RT and a like!
I created BabyAGI-asi, a modification of
@yoheinakajima
's babyagi that runs arbitrary python code to perform tasks - It manages to create routines for pyautogui, webscrapping, manages to execute actions on the computer and even controls its own memory🧵
@London_Lady
I think fear of the unknown is initially a natural response. This technology will result in upheaval of our current societal norms over time. It is inevitable. Recursive agents are a big breakthrough in the application of the tech and will be incredibly powerful in time. We may
I’ve been thinking a lot recently about the future of verifiable credentials and felt like I needed to get some thoughts down on the page
I’m not a thread guy, but I know many appreciate tl;drs. Here is a snippet of main points I cover 🧵 (1/14)
AutoGPT, BabyAGI, and CAMEL have ushered in a new form of AI technology and we need a way to classify them.
I’ve been talking with other builders and workshopping names. Best one so far: Collaborative Agent Modules aka CAMs
Have a suggestion? Share it in the comments!
Ok, the time is right to open source this: introducing GPT-Legion, an autonomous agent framework:
Similar in spirit to AutoGPT and Baby AGI, but written in the superior language that is TypeScript!
Co-work with
@itanih0
@hammh0a
. We are so curious about how ChatGPT agents would interact with each other. Here comes our project CAMEL. Two agents, a gamer and a computer programmer, collaborate on creating a game:
Github:
Colab:
Over the weekend I finished the to-do list that does itself.
Everytime you add a task, a GPT-4 agent is spawned to complete it. It already has the context it needs on you and your company, and has access to your apps.
It’s called the Do Anything Machine (Link in thread)
The Extropic litepaper just dropped!
It contains some long awaited info about what they're building and how it will enhance compute for AI. Here are some key details:
The demand for computing power in AI is growing exponentially, but traditional digital processors are hitting
Built "Startup Simulacra" using
@LangChainAI
and
@guohao_li
's CAMEL architecture.
A VC Agent evaluates a Founder Agent's randomized startup idea: "Personalized cereal boxes to celebrate child's achievements"
Super interesting how an adversarial agent drives reasoning!
Had AutoGPT make me a more complicated website, from scratch this time:
it:
- created login/sign up page
- styled it with bootstrap
- created flask api for login / logout
- created a local json database
Took ~10 minutes at a cost of $0.50. More examples coming
ChatGPT agents interacting with each other? Let's see it in action! 🤖💬
Here's the
@LangChainAI
implementation of CAMEL: Communicative Agents for “Mind” Exploration of LLM Society 🐪
Fork this template on
@Replit
to get started:
@AminMemon
Keep posting into the ether. People are listening even when you don' think they are. The more knowledge you share the more you'll be rewarded.
@hey_rushik
I initially felt the same way, but you can’t look at it like that. Knowledge is the culmination understanding. Each new development of this tech helps to broaden that understanding of how it can be applied. You just have to remain curious enough to keep exploring!
Amazing work by
@dory111111
with
@yoheinakajima
's BabyAGI using
@streamlit
and
@LangChainAI
web app. I modified it to select GPT models, adjust temperature, also made it so folks who aren't familiar with Poetry can run it directly using Python.
Repo 👇
Behold GroupChatGPT 👨👩👧👦, take ChatGPT from awkward 1:1 conversation into a lively group chat with your favorite characters.
Rick Sanchez, Barbie, Cartman, and Yoda going at it about the meaning of life.
Try it out here:
Are you happy? How do you define happiness? I've been pondering these questions and how they intersect with the future of Web3
Again... not a thread guy, but I realize people have short attention spans. Here are a few of the main points I cover 🧵(1/14)
Why aren’t recursive AI agents like AutoGPT and BabyAGI useful yet?
One of the biggest limiting factors has been their inability to retain memories and learn from them.
Here are a few examples of builders solving this problem!
@bcrussett
@mattshumer_
Seems to be available. Only caveat I can find is that they are generally compatible with ML frameworks such as PyTorch and Tensorflow for inference, but don't support training.
Groq is serving the fastest responses I've ever seen. We're talking almost 500 T/s!
I did some research on how they're able to do it. Turns out they developed their own hardware that utilize LPUs instead of GPUs. Here's the skinny:
Groq created a novel processing unit known as
My next demo using
@GroqInc
: writing a novel in two minutes.
Behind the scenes, there's tons of reasoning and rewriting going on. This only shows the surface-level results.
More than 25K tokens are generated here, in less than two minutes.
On-demand books are now a reality!
@neb301
@London_Lady
I’m not assuming that all technological progress is beneficial. I’d argue the application of any technology is what makes it beneficial. The creation of stone and various metal tools allowed us to advance our civilizations, but they also made waging war much more efficient.
The recommendation to advance DIDs by the
@w3c
seems trivial to most of us, but is actually a major step forward in our fight for digital privacy and sovereignty!
Our institutions have failed us
@timberners_lee
just gave us tools to protect our rights
@rsg
@TurnerNovak
@balajis
A circle would make more sense in practice. If one section of the Line were compromised it would make public transit pretty useless
How we interact with AI is about to change thanks to AutoGPT and BabyAGI.
Interfaces like Graphologue will allow us to easily manage multiple agents at a time. Is this the next Chatbot UI?
Stop reading the text !!! Have a graphic dialogue with OpenAI’s
#GPT4
with
#Graphologue
. It turns GPT4-generated text into interactive node-link diagrams in real-time. New Human-Computer Interaction
#HCI
research makes
#AI
easier to use. More at
So I’ve been drinking Fresca for as long as I can remember. I have memories sipping it in my bubble bath as a kid. I love the stuff. However, in an effort to be healthy, I gave up soft drinks in high school. I just looked on the back and realized it has 0 calories and no sugar...
After seeing the login with OpenAI demo put together by Genesis I dove into the docs for Passes (the auth framework that made it possible)
Here are some key takeaways and why I think it’s awesome especially for inter-app communication within AI
At its core, Passes enables
Complaining that AutoGPTs haven’t done anything useful is like saying the Wright Brothers’ airplane was a novelty. What we build today is only the first flight.
ChatGPT has revived my love of cooking!
Every recipe I’ve tried has been out of this world good! My friends and family can’t stop raving about them. Here are a few of my favorite prompts to generate secret recipes!
There’s a future where credentials keep government officials accountable
Facts are no longer alternative if they’re verifiable
We could be an informed electorate once again 🗽
So this is what caused the open ai board to freak out.
The found a new model that could lead to AGI
One that used new maths to break encryption that would’ve taken 37 years to crack.
YT link below
There’s no better feeling than finishing off two helpings of Christmas dinner and a few slices of pie, leaning back to enjoy your accomplishment, and breaking your sister-in-laws new Wayfair bench in the process
The more I talk with people about the power of verifiable credentials the more I am convinced they are the future of Web3
They are the true unlock of freedom and sovereignty in the internet
We pushed out a release to reduce dependency on HuggingFace custom logic for downloading models given the significant outage they experienced some days ago.
Simply making the aiOS™ more robust for the future 💪
#DecentralizedAI
Nous Research is my favorite team to follow. Join their Discord to see what I mean. Also check out
@Teknium1
for incredible insights into their models.
@dabit3
So far so good! On the agency side the hardest part has been finding the right clients with a large enough budget to spend, risk profile to explore, and lack of internal expertise to build.
@seanxthielen
Apple already has the hardware infrastructure to incorporate AI into every facet of our life. They just need to make a few tweaks and train a SOTA LLM
This would be a killer product, but requires better infrastructure and interoperability between RAG applications. Everything is too siloed currently. Who wants to help build the Plaid of RAG?
The trickle of information being generated, directed, and coordinated by OpenAI right now leads me think something is coming soon — usually they like to make sure everyone is speculating and generating buzz to precook the Twitter algo before a release
Anyone remember how vocal Elon was about pausing AI last year? Sounds like a lot of double speak to me. In typical fashion if he doesn't have FULL control he throws a fit. He actually might be the biggest threat to open source and acceleration efforts. Time to open our eyes.
We are dedicated to the OpenAI mission and have pursued it every step of the way.
We’re sharing some facts about our relationship with Elon, and we intend to move to dismiss all of his claims.
Love letter to
@obsdmd
to which I very happily switched to for my personal notes. My primary interest in Obsidian is not even for note taking specifically, it is that Obsidian is around the state of the art of a philosophy of software and what it could be.
- Your notes are
Autonomous Agents will completely change our relationship with AI!
Check out this demo to see how we can create Agents for repeatable tasks. AutoGPT integration coming soon!
Wild tech you have to try:
They are serving Mixtral at nearly 500 tok/s.
Answers are pretty much instantaneous.
Opens up new use-cases, and completely changes the UX possibilities of existing ones.
I started consulting as a fractional CFO for a couple SMBs last year and picked up enough clients to start a firm.
It’s incredibly rewarding being able to help so many different businesses!
Here are a few takeaways that every owner and operator can benefit from
🧵
Really cool extension of AutoGPT’s functionality! Now you can easily give the agent feedback and suggestions as it completes tasks. Great work
@_Lonis_
!
Did you know that there's already an AGI out there? It can (basically):
* Order your coffee at Starbucks
* Perform market analysis
* Find and negotiate a lease
Easter project inspired by
#AutoGPT
&
#BabyAGI
Try Godmode
I highly respect what LangChain and Llama Index have done, but honestly they are more confusing and time consuming than just building your own custom pipeline once you know the providers you want to use