• 0 Posts
  • 18 Comments
Joined 3 months ago
cake
Cake day: August 9th, 2024

help-circle

  • Generative models does not work like that, if it were so, how do you explain that I can generate a picture of a purple six legged cat throwing lasers from the eyes in space?

    In a very very very simplified way, the models are trained that from noise it de noises it until the image is “restored”. A part of the model learns to remove noise until a drawing of a child is restored, another learns to restore the image of a drawing of a nude woman. Basically you say to the model that from noise it has to restore the drawing of a nude child it combines the two proceses (also it is trained to combine things in a way that makes sense).





  • But the best part? It’s a $5000 work laptop, and my 6 year old Thinkpad (with Linux) runs laps around the thing any day of the week. Opening the file explorer takes, most of the time, 5+ seconds…

    In my previous job I was doing Java development on e-commerce (Hybris, then renamed to SAP Commerce) and the laptop (a beefy thinkpad) took ages from powering on to being able to work, also Java compilation could take 30 min and just starting up the project on local another 5.

    Had the opportunity to install Linux (the policy was that dual boot was required and don’t disturb IT with Linux issues) and oh boy, from turning on to being able to work was incredible fast. Compiling went from 30 to 5 min (with same Java official version from oracle in order to avoid any implementation discrepancies between openjdk and the oracle JDK in prod), and starting tje local server went from have enough time for preparing a coffee to seconds.

    Unfortunately my current job only allows Windows and the policies are too strict.









  • This. Worked at a consulting firm doing e-commerce for a client. The client always pushed making changes on banners or promotional texts rather than fixing bugs.

    There was an issue with the address validator in the checkout (why and how is irrelevant) and it was raised by the QAs, but we were told to fix it in the future, they didn’t see it as a priority, they preferred a checkout that worked most of the time an focus on adding a promo banner.

    Now I work in a better place, working on product with stakeholders who don’t prioritise new things over fixing stuff, but we still need to fight to have time allocated for technical improvements that the benefits are not directly evident in the final product.



  • Cautionary Tales

    We tell our children unsettling fairy tales to teach them valuable lessons, but these Cautionary Tales are for the education of the grown ups - and they are all true. Tim Harford (Financial Times, BBC, author of “The Data Detective”) brings you stories of awful human error, tragic catastrophes, and hilarious fiascos. They’ll delight you, scare you, but also make you wiser.

    I think the description is better than what I can write.


  • cordless power tools

    Each tool had their own battery, it discharged so fast and degraded even faster, and forget buying new batteries because the manufacturer decided to change the design again and either you’re stuck with a drill that only works for five minutes or buy a new one.

    Now batteries last an eternity, and because each brand has their own ecosystem, as long as you buy tools from the same brand you can use the batteries you already have. And also the brands has no incentive to change the design and break the compatibility of the batteries, it would alienate the costumers who spent a lot of money on the tools and would go for another ecosystem.