Onboarding has never been this easy
Import your EVM seed phrase and swap from any chain-any token, straight to native OCT on @octra
Bridge directly from your 0xio extension
Powered by @RelayProtocol
Geodesics II is coming, and we're glad it will be trading on Xpectrum.
euints are taking living art further: pieces that keep evolving after mint, drawn live from an on-chain seed and the state of the token itself. It's some of the most interesting work happening on Octra, and worth reading in full.
0xio Desktop v0.6.0 is now available
Security & Signing
• Stronger extension pairing with verified connections before sharing key material
• Touch ID or password confirmation for contract calls, token transfers, claims, and bulk sends
• Recovery phrases and private keys are better protected, with guided backup and safer access
• Live network fees required before signing
dApp Integration
• Connect browser dApps directly to 0xio Desktop, with requests reviewed in the menu bar
• Support for direct browser connections without the extension
• Improved site permissions, wallet isolation, and transaction review
Privacy & Transactions
• Redesigned private balance workflows with queued proof operations and progress tracking
• Network fee choices for sends and privacy operations
• Claim all waiting private transfers with individual claim processing
Desktop Experience
• New notification centre for transfers, claims, approvals, and updates
• Redesigned Home, Send, Privacy, History, NFTs, Contracts, and Settings
• Improved transaction tracking, wallet management, and command palette
Download: https://t.co/M4hugMcoI3
every gdsc ii colony reads the chain.
thousands of ants assemble a pfp from the seed. a 12%+ day for OCT and every colony builds its full picture at once. cross a price milestone and they build the price. feed it and honeypot ants store the oct in the nest. merge two and you get a deeper one. and more.
it's state onchain and a trained swarm reacting to it on @octra
https://t.co/6WdalHYMpQ
merging in gdsc ii on @octra.
pick two colonies. both burn, one child is born. every trait comes from one side or the other, and new ones can roll in. the outcome is seeded by the merge tx itself, so nobody knows it ahead of time.
gen is the deeper parent + 1. go deeper and the colony gets more workers, more grains and richer looks.
https://t.co/OrBOqvyM79
a small brain for ant colonies that build things.
a bivouac is the living nest army ants make out of their own bodies. this repo is the engine we used to train the ants behind geodesics ii colonies on @octra. it has the general parts only: the policy, the environment, the trainer and the browser runtime. the trained weights, the reward tuning and the art are not here.
how it works:
every ant runs the same tiny network, 11 inputs, 8 hidden, 5 outputs, tanh all the way. inputs are what one ant can sense at its cell: am i carrying, what kind, how dense is my kind around me, how close am i to where my kind belongs, which way is the centre, plus a few colony-wide knobs. outputs nudge the classic deneubourg pick and drop rules, steer the walk, and decide when to give up on a crowded spot.
nothing is told where to put anything. grains get sorted because each ant follows local rules and reads the marks the others left behind. that is stigmergy, the same thing real ants do.
the environment scores a finished colony on four things: order (same kinds sit together), space (how much of the disc gets covered), balance (where the mass sits) and harmony (how well neighbouring colours get along). you choose the weights. defaults are flat.
training is antithetic evolution strategies. sample noise, try the policy pushed both ways, keep the direction that scored better, repeat across random colony setups so the brain generalises instead of memorising one map. no gradients, no autodiff, numpy only.
in the browser the brain keeps learning. each colony runs a (1+1)-es adapter over a handful of gains: mutate, watch the colony for a window, keep the change if the reward went up, shrink the step size over time. so two colonies with the same weights still drift into their own habits.
https://t.co/anHmUULm4g
https://t.co/vd8XHCg1Uw
one brain, every ant. 11 senses in, 5 actions out, about 1kb of weights.
trained offline with evolution strategies, then each colony keeps tuning its own gains live while it builds. same brain, different habits.
soon, 0xio Desktop brings private ops to your system tray
• Runs in the background
• No browser required
• Native access to your machine’s compute power, beyond WASM limits
encrypt/decrypt within seconds
every worker in a colony only sees the grains a couple of cells around it, and no ant knows the pfp. this is cluster of them sorting.
can’t wait to share more on @octra mainnet.