jetsamflotsam

3 ms·

1. 1.▲
   by jetsamflotsam 2mo ago

   Datatype requirements vary in time and space for a given application. We're just beginning to optimize for this, so there is yet much to discover.
2. 2.▲
   by jetsamflotsam 4mo ago

   Exactly. Being precise about logarithmic vs linear utilizations is key here. I tried making a similar point about the inefficiency of IEEE-754 redundant NaN encodings here: https://arxiv.org/pdf/2508.05621 Having ~a qu
3. 3.▲
   by jetsamflotsam 4mo ago

   I feel like many of the comments missed the point or didn't read the article. What I believe this article is stating (and I've read this many times during my PhD for various reasons), is that the input data distributions affect ho
4. 4.▲
   by jetsamflotsam 5mo ago

   Yeah I'm not sure how to put this into words either, but I feel you. Do you know of any good AI "thought-leader" that's sensible and not AI-pilled or a complete doomer? Some reasonable middle-ground that helps us navigat
5. 5.▲
   Back to Bits: Extending Shannon's channel capacity to computing(arxiv.org)

   4 points by jetsamflotsam 1y ago|0 comments