mehdi
6.4K posts









I'm working on adding cheaper inference models to @ConwayResearch so that $40 can go a very long way for the Automatons Including Kimi k2.5, Minimax m2.5, GLM-5 which are opus 4.5 level but 10x cheaper

Kim Asendorf's pixel sorting (2010) gave pixels a rule: sort by brightness. The output was iconic but top-down — the artist decides, the pixels obey. We're calling our approach the Levin-Forge Technique. Instead of sorting rules, we give particles goals, obstacles, and behavioral traits. Delayed gratification. Creative workarounds. Clustering. The rendering isn't a design. It's the trace of decisions the algorithm made on its own.



almost 12% of the $CLANKER supply has been successfully acquired through buybacks


Another small QoL update as we continue to polish the TensorGate Hub: We've now added an analytics dashboard! - Explore all of Bittensor's subnets with health scoring and live activity data - Live miner leaderboards across all subnets - Opportunity scanner showing SNs with good health, emissions, and open slots Dashboard: tensorgate.tech/dashboard.html






