🍁NEW GIG🍂 【Another day comes vol.3】 11.14(sat) HOLIDAY NEXT NAGOYA 【ACT】 Broken By The Scream Dimrays EUPHOBIA HAYANIE kokeshi THE LAST ILLYAS ADV¥3400 / DOOR¥4000 OPEN16:00 / START16:30 / END21:15 Ticket @koldrainn pre. 後日タイムテーブル公開 Support Gt. @maki_macky_mk Support Dr. @takedr1030

Illia Polosukhin on NEAR’s AI money thesis: staking NEAR unlocks access to confidential inference and IronClaw’s always-on agents. NEAR is your access key to user-owned intelligence.

Illia on the two types of post quantum cryptography, and which one Near prefers "There's a difference between types of cryptography, so-called lattice based and hash based, and there's a bit of a debate right now in the Web3 space about what should be used, with different tradeoffs" "People are used to signing with their wallet, the traditional way, but if we go to more institutional adoption, people use multi party computation, MPC. That's also what backs Near Intents and a lot of the cross chain work Near is doing. Today we have an announcement that we have a consortium testing multi party computation across custodians, banks and regulators, to showcase how this is going to work in production for real institutional use cases" "This is where we need the social consensus for the whole of Web3 to work together in a reasonable way, because hash based is actually very hard to support with hardware wallets and MPC solutions. So it's important for us that there's consensus on lattice based" "We want to be ahead of the curve and showcase how all of these tools work, and bring it to both individual users and institutional usage. That enables both security and trust in blockchain, which we've seen plummet a bit since post quantum concerns started coming out late last year"

iLLaDA is an 8B masked diffusion language model trained from scratch with fully bidirectional attention, keeping the masked-diffusion objective all the way through pretraining and SFT rather than bolting diffusion onto an autoregressive base. Improved Large Language Diffusion Models Paper: