Lei Yang Σ: retweeted
There's a lot of room for innovation in inference markets, and strong synergy with the distributed systems and mechanism design challenges we've all cut our teeth on in the blockchain world. I'm very excited that @yangl1996 and the MEGA team are going after this space. At Espresso we’ve been quietly working with them on what the chain will need to get there, including fast finality and sequencer rotation. More on that soon..
Today, we commit $30 million to build new products on MegaETH, starting with AI inference.
mega.holdings
6
3
52
2,615
It is hard to resist the idea of viewing artificial intelligence (AI) as a commodity.
On one hand, primitive spot markets have started to form around AI inference, the process that produces intelligence, in the shape of aggregators and routers.
On one hand, primitive spot markets have started to form around AI inference, the process that produces intelligence, in the shape of aggregators and routers.
Building more AI capacity is only half the job.
An inference provider can have the right model, spare capacity and better prices. But users won’t touch it unless they trust it to serve the advertised model and protect their data.
Our whitepaper ↓
mega.holdings/whitepaper
2
3
58
1,834
In other words, AI inference does not yet have a single efficient market.
Purification is our choice of word to define the process of abstracting away the attributes of AI inference.
Purification is our choice of word to define the process of abstracting away the attributes of AI inference.
1
0
7
161
With each attribute abstracted away, inference markets consolidate, leading to more competition and resilience.
We believe this is the missing piece before a truly open and global inference market can emerge.
We believe this is the missing piece before a truly open and global inference market can emerge.
1
0
8
162
Natural continuation as well as a new beginning.
Today, we commit $30 million to build new products on MegaETH, starting with AI inference.
mega.holdings
1
2
26
1,162
the multi-armed bandit problem in real life
In early summer, after concluding our accelerator program, we redeployed our strongest engineers to a special project.
They set out to find where people are exploited by the systems they depend on, and build something better.
A number of prototypes emerged, and as the work progressed, one in particular kept pulling us deeper.
0
0
27
932
Lei Yang Σ: retweeted
The story of our zkLogin optimization journey x.com/i/article/2115…
7
13
52
12,579
This is super interesting. The companion research and engineering posts are very fun reads as well.
Consensus is usually about striking the right balance among theory-imposed tradeoffs plus precision engineering so as to not leave anything on the table. This is a solid result!
Surely it can benefit MegaETH by opening a path to fast confirmation at super high throughput. Cannot wait to explore more details.
Consensus is usually about striking the right balance among theory-imposed tradeoffs plus precision engineering so as to not leave anything on the table. This is a solid result!
Surely it can benefit MegaETH by opening a path to fast confirmation at super high throughput. Cannot wait to explore more details.
Espresso mainnet can now handle up to 1 million transactions per second.
And finality stays under a second, even at that peak load.
How we did it, and why it matters ↓
0
1
21
1,659
Retro on some of the technical decisions the team made building towards MegaETH Mainnet
1
9
39
9,149
Lei Yang Σ: retweeted
Answering Hard Questions: Our Protocol
What is the single hardest technical problem you've solved that most people don't appreciate?
What is the single hardest technical problem you've solved that most people don't appreciate?
5
7
55
7,206
Lei Yang Σ: retweeted
Answering Hard Questions: Our Protocol
Why did the chain take so long to build, and looking back, was the delay worth it?
Why did the chain take so long to build, and looking back, was the delay worth it?
6
4
72
8,358