djoldman

42 ms·

751.
by djoldman 2y ago
The linked piece by Stephen Mason sets out all the interesting context for the law portion (not the postal service fiasco). https://read.uolpress.co.uk/read/3a219939-14d3-45c9-9fe0-9b4...
752.
by djoldman 2y ago
Most statistics classes are not taught to people who will be professional statisticians. I agree whole-heartedly with this: > The book by Freedman, Pisani, and Purves is the one I would have liked to teach from, and it was the book I dre
753.
by djoldman 2y ago
https://archive.ph/lpNox
754.
by djoldman 2y ago
This reminds me of the Saturn V. If I remember correctly, no one knows how it works anymore. https://en.wikipedia.org/wiki/Saturn_V
755.
by djoldman 2y ago
Seems like the cited studies confirm things we already know: when humans are given tasty high caloric-density foods, they consume more calories than when not given those foods. I didn't see a cited study showing that these ultra-proces
756.
by djoldman 2y ago
One interesting metric for LLMs is that for some tasks their precision is garbage but recall is high. (in essence: their top 5 answers are wrong but top 100 have the right answer). As relates to infinite context, if one pairs the above with
757.
by djoldman 2y ago
It didn't! Thanks
758.
Code Search – Grep by Vercel(grep.app)
11 points by djoldman 2y ago|6 comments
759.
by djoldman 2y ago
One interesting metric for LLMs is that for some tasks their precision is garbage but recall is high. (in essence: their top 5 answers are wrong but top 100 have the right answer). As relates to infinite context, if one pairs the above with
760.
Should you use Prisma ORM?(prisma.io)
3 points by djoldman 2y ago|0 comments
761.
by djoldman 2y ago
This is a key observation that is simple and intuitive: >All CLIP-like models perform poorly on mixed-modality search due to a phenomenon known as the modality gap. As illustrated in the figure below, the closest vector to the snippet “I
762.
by djoldman 2y ago
This is a cool way to look at multimodal embeddings. They look at performance as the the percentage of inputs slides from one modality to another: https://i0.wp.com/blog.voyageai.com/wp-content/uploads/2024&#x
763.
by djoldman 2y ago
https://www.orwell.ru/library/essays/joys/english/e_joys
764.
Are We PEP740 Yet?(trailofbits.github.io)
116 points by djoldman 2y ago|79 comments
765.
by djoldman 2y ago
https://arxiv.org/pdf/2405.12411
766.
by djoldman 2y ago
It seems you can get the parameters (weights). They are subject to a license agreement: > 3. Use Restrictions > You must not use any of the AlphaFold 3 Assets: > 1. for the restricted uses set forth in the AlphaFold 3 Model Paramet
767.
by djoldman 2y ago
vid: https://www.youtube.com/watch?v=cJPXLE9uPr8
768.
M4 Mac Mini SSD Modded from 256GB to 1TB(tomshardware.com)
8 points by djoldman 2y ago|3 comments
769.
by djoldman 2y ago
From: https://github.com/MPI-SWS/genmc/blob/master/doc/manual.md Basic Usage A generic invocation of GenMC resembles the following: genmc [OPTIONS] -- [CFLAGS] <file> ... Note that, in order f
770.
by djoldman 2y ago
I disagree. They report that their method produces a model that is 6.5 GB from flux (22.7GB). Why wouldn't you want to know how their 6.5GB model compares to other 6.5GB models? Regarding aesthetic prefs: it's an open problem what
771.
by djoldman 2y ago
Agreed. Once a model has been trained, I believe the main metrics people care about are 1. inference speed 2. memory requirements 3. quality of output. There are usually tradeoffs here. Generally you get a lower memory requirement (a good t
772.
by djoldman 2y ago
These are comparisons to other quantizing methods. That is fine. What I want to see is comparisons to NON-quantized models all with around the same VRAM along with associated inference latencies. Also, we would want to see the same quantizi
773.
by djoldman 2y ago
I definitely agree that there may be some interesting advancements here. I am trying to call attention to the models used for evaluation comparison. There are 3 factors: inference speed/latency, model size in total loaded VRAM, and mod
774.
by djoldman 2y ago
This is one in a long line of posts saying "we took a model and made it smaller" and now it can run with different requirements. It is important to keep in mind that modifying a model changes the performance of the resulting model
775.
The Arsse – The clean and modern RSS server that doesn't give you any crap(thearsse.com)
48 points by djoldman 2y ago|17 comments
776.
by djoldman 2y ago
I love love love ragel. Won't the code here: https://github.com/gritzko/librdx/blob/master/JSON.lex accept "[" as valid json? delimiter = OpenObject | CloseObject | OpenArray | CloseA
777.
Private Cloud Compute Security Guide(security.apple.com)
398 points by djoldman 2y ago|213 comments
778.
by djoldman 2y ago
Virtualization layer (Firecracker + KVM) The "secret" sauce probably. Firecracker is so cool.
779.
by djoldman 2y ago
This kind of thing is pretty typical. The older the company is, the more likely one finds this morass. It won't change absent powerful technical leadership.
780.
Why is it called upper and lower case?(mcgill.ca)
3 points by djoldman 2y ago|1 comments

More ›