ravmachre

4 ms·

1. 1.▲
   by ravmachre 2y ago

   I don't think it's necessarily because of lack of generalizability. We (SWEs) built it, so we naturally have the most intimate knowledge of how to dogfood/use it. And so the cycle intensifies (use, provide feedback, improve).
2. 2.▲
   Ask HN: Transitioning to ML Systems and Performance Work: Advice Needed

   1 points by ravmachre 3y ago|0 comments
3. 3.▲
   Inverse Scaling Prize: Identify tasks where LLMs performance worsens with size(github.com)

   2 points by ravmachre 3y ago|1 comments
4. 4.▲
   by ravmachre 3y ago

   TL;DR: Win up to $100,000 for finding an important task where larger language models do worse.
5. 5.▲
   by ravmachre 4y ago

   I relate to your comments about gym a lot. I feel like going to the gym is a mundane, mechanical and very dystopian activity. I never found any 'fun' in the gym and we all know we can't force ourselves doing what we don'
6. 6.▲
   by ravmachre 4y ago

   I've been using Speedreader (shipped by default on Brave) and this page loaded as pure text without any JS Elements/Ads for me. I'd definitely recommend Brave solely for this feature: https://support.brave.com&#x2F