• 0 Posts
  • 35 Comments
Joined 1 year ago
cake
Cake day: June 16th, 2023

help-circle
  • The solution for this is usually counter training. Granted my experience is on the opposite end training ai vision systems to id real objects.

    So you train up your detector ai on hand tagged images. When it gets good you use it to train a generator ai until the generator is good at fooling the detector.

    Then you train the detector on new tagged real data and the new ai generated data. Once it’s good at detection again you train the generator ai on the new detector.

    Repeate several times and you usually get a solid detector and a good generator as a side effect.

    The thing is you need new real human tagged data for each new generation. None of the companies want to generate new human tagged data sets as it’s expensive.












  • It was something around 40 TB X2 . We were doing a terrain analysis of the entire Earth. Every morning for 25 days I would install two fresh drives in the cluster doing the data crunching and migrate the filled drives to our file server rack.

    The drives were about 80% full and our primary server was mirrored to two other 50 drive servers. At the end of the month the two servers were then shipped to customer locations.





  • The Intel CEO had always come from engineering fab. This kept the high level decisions made by somebody who understood the product and how it was made.

    Then CEO and the head of fab was caught sexually harassing employees. They were both shown the door. So no CEO and the guy who was next in line were gone. They needed the number 2 in fab to take over fab to keep production up.

    So the board decided to make the CFO the new CEO. A guy who had a MBA was running a chip company that had only been run by engines.

    Profits went up for a while but then Intel struggled to maintain innovation and properly upgraded fab and chip design. Add the increasing skill of rivals and a increase in importance in chips other then server and desktop. which were the only areas Intel was king. It’s a recipe for failure.



  • Partially. Inverters providing virtual inertia is good but has the problem of still being active and reactive. It helps and is cheaper and more efficient than flywheels.

    Flywheels and turbines however provide a very sticky frequency. They help out a lot with stability and give inverters time to respond.

    Think balancing a stick on your hand vs anchoring it in clay.

    If we take enough turbines off line we are still probably going to need some mechanical power stabilization no matter how inefficient.

    But yeah I think we are going to see a blend using as much electrical and as little mechanical as possible.