🧵Three eras of foundation models
1. The recent past - demonstration of the capabilities of AI.
2. The present - incremental improvements to efficiency of training and inference.
3. The future - continual learning on live data for a constant representation of the world state.
Today, we’re announcing that
@Amazon
will invest up to $4 billion in Anthropic. The agreement is part of a broader collaboration to develop reliable and high-performing foundation models.
Let’s keep technology that benefits all of us in the open where it belongs - over-regulating risks concentrating the benefits in the hands of the few and hiding the dangers behind opaque barriers
Similar to electricity and microchips, AI is the spark of a new industrial revolution. How we govern it will shape the direction of our future and Gensyn is strongly on the side of open access and development
I'm speaking at Singularity: AI x Crypto Convergence in Dubai next month - looking forward to talking more about what the future of decentralised AI looks like and how we get there
Gear up for insights from
@fenbielding
, co-founder of
@gensynai
, at
#OFR
Dubai - Singularity: AI x Crypto Convergence on Apr 17.
🔗 Get your tickets:
🛰️ Dive deep into the GPU Networks panel, where Ben will join top innovators to discuss powering
enjoyed chatting with
@cheryldchan
as always
highlights:
- the merge of training and inference
- why verification is essential for a base level compute protocol
- building things right rather than rushed
- the future of shared human knowledge (embeddings not databases)
If you missed my talk at
@EthereumDenver
you can watch it back here - I explain the two main reasons behind what we're building with
@gensynai
and why it needs to be done right (i.e. the hard way)
Watch our co-founder,
@fenbielding
, on-stage at
@EthereumDenver
explaining why AI needs crypto and why we need to rebuild the fundamental infrastructure underneath ML in order to enable the machine intelligence revolution
over the next few months we're going to see arguments for why open source AI is dangerous - the opposite is far, far worse. Let's keep AI development in the open, accessible and auditable by everyone
@3blue1brown
Don't get stuck trying to understand one small part, it's frustrating and time consuming. Carry on, try to understand it holistically and then you can revisit and dig deeper.
Really pleased to be able to announce that yesterday I successfully defended my PhD thesis:-
Evolving and Ensembling Swarms of Deep Neural Architectures for Computer Vision
it's been an interesting journey and it's exciting to think about what's next!
🧵 The internet is for models, not people
LLM apps (ChatGPT, Perplexity, Claude) let us interact with most of the info on the old internet through conversation, rather than browsing
next, those conversations will become personalised and local and the internet will be for agents
paraphrased: protecting our privacy isn't about shifting the power imbalance to us, it's about levelling it
great to hear from Edward Snowden again at ZKHouse Bogotá with
@nymproject
,
@MantaNetwork
,
@0xPolygon
and the new Universal Privacy Alliance
@priv_alliance
Everyone's talking about the future of AI agents but I think the more interesting change from machine intelligence will be moving the internet from structured text, queryable via search engines to rich embedding spaces queryable via model inference
Really pleased to have
@_jamico
join us from
@a16z
as our new Head of Ops!
It's clear that Jeff deeply understands the Gensyn vision and is uniquely placed to help us navigate the operational and legal challenges that come with creating an entirely new compute infrastructure 💪
💥 Excited to share some news!
After 3 amazing years at
@a16z
, I’ll be joining our portfolio company Gensyn as their Head of Operations.
@gensynai
is building the world’s machine learning compute network, allowing anyone to contribute or access compute for ML applications.
New technological advancements shouldn't be artificially restricted through centralised regulation - we risk far more as a society in doing this than we gain
happy to sign this open letter alongside people I greatly respect - let's keep the next industrial revolution open
1/ We’ve submitted a letter to President Biden regarding the AI Executive Order and its potential for restricting open source AI. We believe strongly that open source is the only way to keep software safe and free from monopoly. Please help amplify.
The platform risk with these AI APIs is insane - we need a decentralised alternative
We're building the resource level
@gensynai
(compute) but who is building the application level?
Founders: It increasingly looks like having a direct import of openai, anthropic, or any one startup’s modules directly in your codebase is just as bad as hardcoding strings in shipped binaries, especially when there is no feature/signature parity.
Sam just said that if someone comes up with a fundamentally superior (more compute efficient) approach for scaling powerful AI models, OpenAI will be in a “not great” place.
This should be our primary focus for AI acceleration — not blindly worshiping the Transformer.
AI will transform our lives.
But we cannot leave it to the hands of a small few.
@karpathy
is right. We need decentralization for AI to succeed.
Here's why 👇
soon we're going to trend away from single, large "AGI" or "God Models" and towards modular + composable execution of smaller, task specific models
inference and training passes will become JIT-routed paths through modular parameter space, not fragile training of one big graph
Congratulations to the
@ModulusLabs
team on closing their Seed round - they're making one of the most important building blocks of the open & permissionless AI future and are easily one of the strongest teams in the space right now - proud to be an early backer!
Ecstatic to announce our $6.3M Seed round today, co-led by
@1kxnetwork
&
@variantfund
🎉
With participation from prominent funds & angels, we're building Modulus: the 𝗔𝗰𝗰𝗼𝘂𝗻𝘁𝗮𝗯𝗹𝗲 𝗠𝗮𝗴𝗶𝗰 company🪄
How does specialized ZK bring AI on-chain?
Excited to announce the Harvard Blockchain Conference happening on April 13-14, Sat & Sun!
Thrilled to unveil our lineup of [AI & DePIN] speakers. Don't miss the opportunity to hear from leading experts at the intersection of AI and decentralized platforms.
@shibshib89
We're doing a small
@gensynai
mixer on Thursday night at EthCC for likeminded folks roughly around decentralised AI / open source AI
Space is limited - DM me or
@christinetyip
for details
If you read this and think "this should be available to everyone not just Facebook" then get in touch..
we're hiring people to help us build the decentralised, free market alternative at
@gensynai
For perspective, at ~6,000 A100s, Facebook's newly announced AI cluster is on par with Perlmutter, the world's fifth fastest supercomputer (~6,000 A100s). Ultimately, FB is going to scale to ~16,000 A100s.
it's crucially important that we keep the tools to build AI open and available to everyone
we talked about this and how we're building towards it directly at
@gensynai
with
@LoganJastremski
on his podcast earlier this week
The AI Renaissance is only now heating up!
Over the last decade, we've learned that the more Compute and Data thrown at Neural Networks results in more exciting Models
Today, I'm dropping a podcast with the Founders of
@GensynAI
with
@fenbielding
&
@_grieve
We speak to:
-
The future of AI is a rich ecosystem of models - designed, coded, trained, and run by people all over the world without centralised control or gatekeeping
All facilitated by a trustless network of hardware resources with persistent value flows that can't be monopolised
🧵TIFF, the new CRUD
Data storage operates on simple concepts - four basic operations of persistence, CRUD: Create, Read, Update, and Delete.
AI gives us a new form of storage - embeddings. This requires new basic operations. Enter, TIFF: Train, Infer, Finetune, and Forget.
If you're building in AI/ML SF is absolutely incredible right now 🔥🔥🔥
I'm here for the first time so I can't say where SF went or whether it's "back" but I can say there's some amazing things being built here and the conversations are awesome 👏
🚨 NEW REPORT
GPT
@home
: Why the Future of Training is Decentralized
Can we train a large AI model over the world’s edge devices?
Increasingly the answer appears to be yes.
New report on decentralized training - why it matters, recent breakthroughs, and the challenges ahead.
people have been telling us the same thing for nearly 5 years now!
it's hard to reason about hard things, especially when the consensus (money / bigtech focus) is in the other direction.
very glad we've got a growing group of people + companies focussing on this now though
So crazy to me that 6 months ago literally everyone serious AI researcher I talked to told me stuff like this wasn't feasible. Decentralized Training is gonna work - we're still at DDP stage and there's a long way to go for giant model runs, but no doubt where the vector of
We believe in the right to build
A principle, created by decentralised technologies, that will ensure we can all access the resources required to build new machine learning models
making a job offer without speaking with someone is an incredibly bad move imo
we're happy to interview anyone who's looking to work on open models + open ML compute infra but we'll make sure it's a fit for both sides - culturally and technically
I recommend this to everyone thinking about personal AI models / AI tutoring ever since
@alive_eth
recommended to me earlier this year
unbelievably prescient
GANs are back, convergence is hard, and pre-training is expensive
that's the tldr from our ICLR dinner in Vienna last night - great food, drinks and conversation - thanks to everyone who came out 🙏
I'm speaking at BASS Denver alongside some brilliant people - catch me there for a chance to get some exclusive Gensyn swag 👀
Also check out the
@gensynai
account for more events, we're going to be pretty busy next week...
1/ We're now just over a week away from BASS Denver, which kicks off at 9:30am on March 1st
@EthereumDenver
! Register here:
Here's a thread of all the event info: 🧵
We're going to look back on film and tv written, directed, and produced by teams of people as an insanely expensive + rigid era in content creation
We'll keep a luxury-goods artisan appreciation for it but it's going to become a niche against a norm of infinite living worlds
Pika Labs raises $35m
Runway raises $141m
Do you guys see what is going on??
We are a year or two away until everyone in the world can create shows & movies. The creativity and output of everyone in the world is about to be unleashed
Hollywood is so freaking over
1/ We're thrilled to announce that Fabric Cryptography has raised $33M in Series A funding to accelerate advanced cryptography workloads with our general purpose chip – the Verifiable Processing Unit (VPU).
A quick summary of a lot of conversations over the past 5 years:
"Decentralised training doesn't work - the current systems aren't highly distributed, that means it can't work."
"What if we built systems FOR decentralised training though?"
"Decentralised training doesn.."
1/ Web3 x AI
We jumped on a Space with
@gensynai
's co-founder,
@fenbielding
&
@chainml_
's CEO,
@ronbodkin
and we discussed:
- Why the passion to decentralize AI?
- How far are we from high-performant decentralize AI?
- What's missing in the Web3 x AI stack?
Check out below 👇
I'm speaking at a bunch of events in Brussels next week for
@EthCC
- come and find me if you want to chat about the future of humanity + AI and get some exclusive swag
The Gensyn team is heading to
@EthCC
in Brussels. If you're attending, be sure to check out these events and talks exploring the intersections of crypto and AI.
🧵 For more details on each event, check out this thread:
Speak to you soon 👇
🗓 Wed, Sept 20th 2023
⏰ 9PM SGT (6AM PT)
🎙 Discussing Web3 x AI use cases, ahead of our FVM Genesis Accelerator!
🗣️
@fenbielding
from
@gensynai
and
@ronbodkin
from
@chainml_
Sam will tell you it's completely hopeless to build your own foundation models and compete with OpenAI
We'll tell you that you'll be able to build them on Gensyn at vastly higher scale than anyone has before💪
The
@gensynai
team is hosting a small get together at Consensus for teams building decentralized AI.
Join us!
Space is limited. DM me or
@fenbielding
for details.
@ID_AA_Carmack
We're building instantaneous price discovery for GPUs at
@gensynai
We need to replace all aspects of the ML compute stack that require human input with code otherwise we'll never scale it to its full potential
Seizing the Means of Production
As data & compute reshape the AI economy, reducing labor's value to benefit a few, DeAI offers a vision where individuals are key, competing to create superior data and networks that power dominant intelligences.
Let's Unveil Key Takeaways👇🧵
.
@guywuolletjr
and
@alive_eth
deeply understand our space and have been super willing to share their insights
@SuccinctJT
has spent hours working through novel cryptographic concepts with us
This investment is a testament to the hard work that the entire
@Gensyn
team has put in over the past few years and I'm unbelievably proud of every individual member and the team culture that we've built together
Check out if you want to join us
I'm on a ferry that doesn't allow dogs and I've just realised that a woman that's been sat near me for the last 3 hours has a dog in her rucksack. I'm incredibly impressed by both her and the dog.
.
@gensynai
is a hyper-scale, cost-efficient, compute protocol for training deep learning models
⚡️Why decentralized, uncensorable machine learning?
⚡️The problems & use cases they're exploring
Live tmrw with
@_grieve
&
@fenbielding
⏰1:30pm CET / 7:30am ET
We’ve gotten to know the a16z crypto team well over the past few years and we couldn’t imagine a better suited partner to help us take
@gensynai
to the next stage
🧵 Human <> machine knowledge curation
Machine learning provides us with a vastly improved way of storing and interacting with knowledge in a digital form.
To do this, we need new basic operations and those new basic operations need machines to perform them.
🧵TIFF, the new CRUD
Data storage operates on simple concepts - four basic operations of persistence, CRUD: Create, Read, Update, and Delete.
AI gives us a new form of storage - embeddings. This requires new basic operations. Enter, TIFF: Train, Infer, Finetune, and Forget.
if you want to see a bit more behind the curtain of the
@gensynai
founding / funding journey then give this a listen👂
@_grieve
and I spoke with an old friend
@ivailoj
for The Funded podcast from
@daedalus_angels
📢ICLR folks - we're hosting a decentralised and open source AI mixer with
@PrimeIntellect
in Vienna this Friday at 7pm
drinks, food, and machine learning chat guaranteed, registration required as we're limited on venue capacity
register here:
I genuinely couldn't recommend the Ignite Pre-Accelerator more if you're in the early stages of starting a company! It's an absolutely fantastic chance to learn and grow. Plus the
@FairCustodianUK
office is in
@weareproto
too so you'll get to see our wonderful faces every day 😬
Applications for our next programme are now open!
Ignite Pre-X is a 12 week programme for early-stage tech companies in NE England.
It comes with £10k investment from
@NorthstarVent
and access to
@weareproto
immersive lab!
Find out more here:
Pls RT
there seems to be this idea that generalised language models will create this new ecosystem of autonomous agents that will suddenly appear and start acting and interacting on our behalf but that's already here - they're just going to get way better
I wouldn't know any of the people that I call close friends if it wasn't for
@campusnorthuk
, you welcomed me in when I had no idea what I wanted to do, or what I could do and you pointed me in the right direction. Thank you!
if you've been watching the AI alignment and regulation debates and thinking it's all very theoretical and not worth the effort of arguing
𝐧𝐨𝐰 𝐢𝐬 𝐭𝐡𝐞 𝐭𝐢𝐦𝐞 𝐭𝐨 𝐠𝐞𝐭 𝐢𝐧𝐯𝐨𝐥𝐯𝐞𝐝
if you don't, you risk losing the ability to build in the open, an essential right
The heretofore silent majority of AI scientists and engineers who
- do not believe in AI extinction scenarios or
- believe we have agency in making AI powerful, reliable, and safe and
- think the best way to do so is through open source AI platforms
NEED TO SPEAK UP !
🚀NEW EPISODE🚀
This week
@tw_tter
chats to
@fenbielding
and
@_grieve
from
@gensynai
. Hear about:
➡️the landscape for AI and AI compute
➡️their focus and edge cases
➡️Building Gensyn as a Layer-1 protocol
➡️and much more!
This is a great write-up by
@Gabrielaellala
of the stories of some of the inspiring female founders that have been through
@igniteaccel
programmes, with an excellent call to action at the end. I'd highly recommend taking her up on that cup of tea!
and
@pmarca
shares our vision for the incredible things that AI can do for the world - as long as we allow it
We’re looking forward to working even more closely with the
@a16zcrypto
team and our other investors as we take Gensyn from research to reality
Huge s/o 👏 to the incredible NE community who helped us celebrate the end of the latest
@igniteaccel
programme today. Great to be back in NCL! 🚀
Well done to the teams who pitched
#IGNITEshowcase
- you smashed it. Proud to have worked with you all ❤️
#iloveNEtech
Really pleased to be able to announce that my first first-author journal paper "Evolving Image Classification Architectures with Enhanced Particle Swarm Optimisation" has been accepted to IEEE Access with the early version available open access here:
really enjoyed this conversation with
@shayonsengupta
there are some hardcore tech challenges to solve to build the future of ML infra and we're taking them head-on - reach out if this resonates with you
At
@gensynai
, we are shaping the future of Machine Learning.
Every day, our engineering and research teams take on cutting-edge challenges in distributed computing. From verification of decentralized training to serialization across ML frameworks, we push the boundaries in AI.
.
@cdixon
has been thinking and writing about decentralisation since way before we discovered the AI compute benefits
‘Centralized platforms have been dominant for so long that many people have forgotten there is a better way to build internet services.’
I described this to someone at dinner last night and realised this morning how strongly I believe it
as individuals we lost the last big battle with corporations over control of our personal data and we've since struggled to rebalance through regulation or technical solutions
If you work in ML/DL (or use it in your personal/hobby projects) we'd love to hear about how you train your models
Google form linked in the thread - should only take a few mins to fill out and we really appreciate your responses 😁
🧠Deep Learning Poll 🧠
Thank you for all the responses to our compute provider poll (the supply side of the network). Now, we’d love to find out more about the demand side: the deep learning engineers/researchers/hobbyists in our community.
Great to see one of my angel investments get acquired - congrats
@Staging_Labs
and
@MerkleScience
!
Acquiring Staging Labs will enhance Merkle Science's cross-chain capabilities and support their expansion in the US and Canada.
1/ Exciting news 🚀:
@MerkleScience
has acquired
@Staging_Labs
! Back in 2018, I met
@MrigankPattnaik
and Nirmal as an investor at
@join_ef
. Little did I know that this encounter would begin a remarkable journey that has now come full circle.
"Allowing large corporations to shape regulations in their favor could leave the actual users — us — without a say in how it’s governed"
Hugely relevant: all fears + regulatory solutions around AI are opinions of a small group with outsized influence
We're going to be at
@EFDevconnect
in Amsterdam in April
Check
@Maven11Capital
's event out if you're interested in all things privacy + infra and give us a shout if you're going and want to meet up!
p.s. we're hiring engineers and more 👀
.
@sriramk
has welcomed us into his network and is a great sounding board
@milesjennings
understands the landscape of decentralisation and regulation better than anyone
sadly often true but I promise this isn't us - if you want to work on the left side (and add some boxes, arrows, etc.. that don't exist yet) then get in touch
our applied research position is ML engineering / hacking on new & distributed model architectures
Really enjoyed talking with
@tasjourno
as always!
@_grieve
and I explained some of our motivations and shared our pitch deck for the
@gensynai
seed round (shoutout to
@SteveGlasper
for the design)