Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
alok-g
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
26 ms
·
451.
▲
by
alok-g
3y ago
Ditto for me all the way till ExplorerPatcher. I found the side trackpad less ergonomic, so use a separate (vertical) mouse.
452.
▲
by
alok-g
3y ago
Amazon does verify and also actively tries to blocks bad sellers. Details of this are easily found online.
453.
▲
Massive hidden compute sitting idle in embedded devices, useable for everyday AI
(petewarden.com)
9 points
by
alok-g
3y ago
|
3 comments
454.
▲
Unified-IO 2, the first autoregressive multimodal model
(unified-io-2.allenai.org)
1 points
by
alok-g
3y ago
|
1 comments
455.
▲
by
alok-g
3y ago
Paper: https://arxiv.org/abs/2312.17172 Github: https://github.com/allenai/unified-io-2 Quick read (third party): https://www.marktechpost.com/2024/01/01/meet-unifie
456.
▲
by
alok-g
3y ago
Thank you. This is very helpful to give an idea of the capabilities.
457.
▲
UK court allows ANN patent saying computer programs exclusion does not apply
(bailii.org)
79 points
by
alok-g
3y ago
|
86 comments
458.
▲
by
alok-g
3y ago
I don't "see", as in, it does not start feeling like my closed eyes are sending signals of an elephant or some residual image of it. However, I am able to "visualize" an elephant in outline and some detail within t
459.
▲
by
alok-g
3y ago
Sure ideas and facts would be spread out. The weights are cumulatively involved in generating the tokens. The error margins from them would also be accumulating for some paths while also decaying for some others. If the training and infe
460.
▲
by
alok-g
3y ago
I somehow do not think the 'hallucination' (or confabulation) problem, as defined by common people should be theoretically hard to solve. It may need a lot more computations though. We start with quasi-random initialization of ne
461.
▲
Observer Theory (Stephen Wolfram)
(writings.stephenwolfram.com)
1 points
by
alok-g
3y ago
|
0 comments
462.
▲
by
alok-g
3y ago
Human Relationships, by Steve Duck https://www.amazon.com/Human-Relationships-Steve-Duck-ebook/... The book aims to present research insights in a way to be of practical use for a common person.
463.
▲
by
alok-g
3y ago
I understand. I would like to train smaller proof-of-concept models on my laptop, which could build upon large pre-trained models though. Thanks.
464.
▲
by
alok-g
3y ago
Love this. Am wondering how well this will work on Windows. Likewise for the pretrained models on my personal laptop (i7, 32GB RAM, RTX 2060 6GB).
465.
▲
by
alok-g
3y ago
I wonder if the 4% to 10% guidance continues to apply with significantly higher inflation now seen.
466.
▲
by
alok-g
3y ago
Is there a reasonable level of confidence that this would work. Asking as a two-month holiday is significant.
467.
▲
by
alok-g
3y ago
And I guess avoid Windows.
468.
▲
Amazon using AI to remove fake reviews, >200M suspects blocked in 2022
(aboutamazon.com)
2 points
by
alok-g
3y ago
|
1 comments
469.
▲
by
alok-g
3y ago
>> Camera, Cortana, and Photos can now be uninstalled When a user would uninstall an app, would it actually be removed from the system, or just the icons and entries would be. Motivation for asking: Several built-in apps are listed
470.
▲
by
alok-g
3y ago
Thanks for a detailed response! Sincerely appreciated. >> Is your therapist a clinical psychologist or psychiatrist? Yes. Clinical psychologist. >> That would be an instant disqualification of an ADHD diagnosis. You could be r
471.
▲
by
alok-g
3y ago
The test I was given is called DIVA 2.0. It says it is based on DSM-IV criteria. I have no understanding of how DSM criteria evolved. The DIVA 2.0 test notes, "Research has indicated that at adult age, four or more characteristics of
472.
▲
by
alok-g
3y ago
Thanks for the pointer. I'll find more about this.
473.
▲
by
alok-g
3y ago
I have all the symptoms described (inattention type, not hyperactivity), including hyperfocus. A test my therapist had given me indicated that I do have ADHD. However, she told me that I do not actually have ADHD as my hyperfocus does not
474.
▲
by
alok-g
3y ago
There's indeed a subtlety here. The paper's abstract says: >> "... though the finite value of this speed must, of course, be obtained from experiment" It should state, "the value of this speed..." (remove
475.
▲
by
alok-g
3y ago
+1. When I was working out the maths myself, I had exactly the same thought.
476.
▲
by
alok-g
3y ago
A short personal story: While trying to understand special relativity from books, and failing to, I tried doing maths myself based on whatever I had understood till then. I ended up doing what this paper has mentioned. I was surprised.
477.
▲
by
alok-g
3y ago
Nice. >> Video Memory: 6 GB >> Graphics Processor: Nvidia GPU with 12 GB VRAM or greater Would it work on an x64 Intel i7 machine with 32 GB RAM and with Nvidia RTX3060 with 6 GB RAM.
478.
▲
by
alok-g
3y ago
What prevents YouTube from sending ad and the actual video in a single video download/stream?
479.
▲
by
alok-g
3y ago
Interesting. I just tried. This part is not working for me as yet -- "Modifying the formula in one cell will also update all the cells in the same columns."
480.
▲
by
alok-g
3y ago
Against a 100% diffuse reflection, that would be a 2000:1 contrast ratio. And we have TVs commonly claiming much larger! ;-)
More ›