Pinned post
Actual neural net being trained in the EVM. 200K gas inference, about 340K gas training step, including loading and storing all weights to storage with MIP-8. Still room to optimize.
0:06
Daniel Von Fange
Verified account This account is verified. Learn more@danielvf
Skilled Professional (most days). Defends against the bad guys.
1.3K Following 12K Followers 3.8K Posts
Latest posts 5
Well, I melted off a bit of my finger and permanently destroyed an ESP32, ESP32-S3.
Productive Saturday.
Productive Saturday.
This is my conclusion.
It seems likely that the risk of a supply chain attack will often exceed the risk of a coding agent making a new vulnerability.View quoted post on X
Open source will shift away from helping you find already-working code to finding already-secured code, or finding new ideas altogether.
I appreciate details this soon.
Sharing like this reduces hacks, because other companies can learn. DeFi got good at this early on, but CEX's and old school companies have lagged far behind.
Sharing like this reduces hacks, because other companies can learn. DeFi got good at this early on, but CEX's and old school companies have lagged far behind.
View quoted post on X
One of my favorite talks I've given.
Three good stories from human spaceflight, and how they tie to crypto incident response.
youtube.com/watch?v=OjukdG…
Three good stories from human spaceflight, and how they tie to crypto incident response.
youtube.com/watch?v=OjukdG…
www.youtube.com Incident response lessons from spaceflight - Daniel Von Fange We'll walk through near two almost disasters from human spaceflight...