• 5 Posts
  • 719 Comments
Joined 1 year ago
cake
Cake day: October 23rd, 2024

help-circle




  • This is an extremely promising innovation, and company plans on bigger designs already. But the total cost has to include an automated roof opening shelter for storms, that can open and close in medium winds before the high winds come. This makes the ground footprint higher than traditional turbines, even if agriculture can be done when there is no storm and roof is open. Perhaps 4 thethering strong cables could permit it to survive a cat 1 hurricane when hugging ground without a roof, but it is more weight to lift normally.



  • Sodium Ion already does 5000+ cycles. Adding Vanadium is not a scalable material. It is very expensive. 400 cycles steady is not useful information because it needs to do much more. They didn’t state a wh/kg density. This is probably not a viable research vector, but “big Vanadium” has proposed a rental model to make Vanadium more scarce for other applications. Flow batteries (a fuel cell with tanks of electrolytes) provides an ultra easy way of recycling/selling the vanadium for traditional uses. Battery rental that forces returning it could be viable.














  • old (1 year = 1000 years in AI) accusation not relevant to expected upcoming deepseek breakthrough model. Distillation is used to make smaller models, and they are always crap compared to training on open data. Distillation is not a common technique anymore, though it’s hard to prove that more tokens wouldn’t be “cheat code”

    This is more a desperation play from US models, even as youtube is in full, “buy $200/month subscriptions now or die” mode.