News for hackers. Stuff that matters.
Re:Understanding DeepMind's Sorting Algorithm by smodad on Monday June 12, 2023 @04:07PM (#36300197)
I just realized that Justine was the person responsible for the massive reduction in the memory footprint of the Llama models back in March.[1] Super impressive! These are my favorite kinds of blog posts.

[1] https://github.com/ggerganov/llama.cpp/pull/613

Re:Understanding DeepMind's Sorting Algorithm by m00x on Monday June 12, 2023 @04:36PM (#36300675)
It's just using mmap, nothing too impressive. It's a nice contribution nonetheless.
Re:Understanding DeepMind's Sorting Algorithm by dundarious on Monday June 12, 2023 @09:34PM (#36303989)
There was more. You can't just splat giant C structs with pointers into shared memory/a file, and expect another process to just mmap and be able to recreate valid state again. At the very least the pointers are going to be all wrong. There was necessary work to adjust the file format. Not rocket science, but not just turning while(fread()) into open();mmap().

Also, there were insights into how to minimize which models needed adjustment. The ideas and code were worked on by at least 2 people, and I'm an outsider on that project, but I didn't see anything untoward like "stealing credit". The magic change wasn't a perfect move, but is the kind of thing I do locally when I don't know the project/binary format well yet, so not exactly the megalomaniacal move it was painted as. Better that only the version number changed, but she's independent and doing good work, so you'd kind of hope she has a self-promotion streak! Changing the magic would be on the very very low end of letting that side go a bit too far.

Re:Understanding DeepMind's Sorting Algorithm by gajnadsgjoas on Monday June 12, 2023 @04:19PM (#36300383)
You wanted to say the one was banned by the author because of all the drama that followed
Re:Understanding DeepMind's Sorting Algorithm by jimsimmons on Monday June 12, 2023 @05:50PM (#36301707)
What drama. Ooc
Re:Understanding DeepMind's Sorting Algorithm by pcj-github on Monday June 12, 2023 @06:21PM (#36302113)
Re:Understanding DeepMind's Sorting Algorithm by skeaker on Monday June 12, 2023 @07:18PM (#36302778)
I don't understand this. What's the contention that's actually getting people upset here?
Re:Understanding DeepMind's Sorting Algorithm by nl on Monday June 12, 2023 @07:34PM (#36302958)
Having read through a bunch of outraged comments the issue seems to have been:

  * It changed the file format
  * For some people it was slower that the original version particularly on low end computers
One person was particularly outraged and reverted the whole change.

However, the current version does use a similar approach to that which was proposed: https://github.com/ggerganov/llama.cpp/commit/f963b63afa0e05...

Re:Understanding DeepMind's Sorting Algorithm by nicman23 on Tuesday June 13, 2023 @03:25AM (#36306774)
Lol why oss people are so antisocial - I include myself in the subset
Re:Understanding DeepMind's Sorting Algorithm by mlajtos on Tuesday June 13, 2023 @06:01AM (#36307925)
Just a wild guess – developers lack soft/managerial skills. (Overgeneralization.) In companies, this is accounted for because developers are shielded by managers. But in F/OSS you get to interact directly with the developers. As for the why devs lack soft skills – hold your hats – programming is basically commanding. Without "thank you", without "please" and with lot of swearing. No time for pleasant bullshit. (By this reasoning, devs using declarative/functional programming should be more polite than imperative programmers, right? :D)
Re:Understanding DeepMind's Sorting Algorithm by jart on Tuesday June 13, 2023 @07:58AM (#36308945)
It was 4chan. You can read my comment history on Hacker News if you want to learn more.
Re:Understanding DeepMind's Sorting Algorithm by CaptainNegative on Monday June 12, 2023 @09:34PM (#36303984)
Why is it Justine posts and seemingly only Justine posts that always get this type of comments? Do people regularly comment on the authors of other content, for better or for worse, and I miss it?
Re:Understanding DeepMind's Sorting Algorithm by Conscat on Tuesday June 13, 2023 @12:27AM (#36305314)
Faster Than Lime, too.