Consider the past four months of life through the filter of from?site=debarshibasak.github.io : We've gone beyond DevOps, and have suffered the Death of the Roadmap. It is acknowledged that the Easy part of building is over, and We wonder if GitHub is still needed. Many things remain unclear of course, due to the Cambrian Explosion of AI.
So I'm curious, you have blog full of what Mathematicians 'must do'. Why do you feel such a need to lecture them? Are you mathematician yourself? Thanks.
And once you can reliably flip any object a half rotation, one/two/etc rotations, it transfers fairly easily to similar objects, and to the juggling pattern.
Going from not juggling to juggling knives can be done in perhaps weekends.
Then, of course, you can do all the fancy multi-object patterns, but I've yet to see a reaction to three spinning knives surpased by five+ tennis balls. The five+ tennis balls are neat while the three (even just two) spinning knives is always a visceral holy shit please stop you're going to hurt someone.
From conclusion of paper, I read the opposite of what you state. "A better strategy would be to expose a larger fraction of the body for a shorter time period, preferably when the sun is lower in the sky to allow a greater margin of error."
A sun lower in the sky means your shadow is longer. Maybe I'm confused?
Good catch, it is kind of confusing as they seem to contradict themselves, and also depends on the season.
Figure 6 shows that for full body exposures, there is a wide window between the time for sufficient UV and the time for too much UV. As the fraction of body that is exposed decreases, the window of optimum UV exposure times also decreases. If only hands and face are exposed, there is generally only a small window between receiving insufficient UV for vitamin D production, and too much UV for skin damage (erythema).
As the UVI becomes smaller, it becomes more difficult to produce sufficient vitamin D without inducing erythema. For low sun conditions typical of midday in the mid-latitude winter (SZA ∼65°), where R∼1, the skin area exposed must be more than 12% (see Eq. 5) This means that for these low sun conditions it is improbable that sufficient vitamin D can be produced from exposing the hands and face alone, since that area is less than 12% of the full body.
If you are able to expose a lot of your body at a time then yes later in the day is workable (in the summer). In the winter, (depending on your latitude) you still need mid day sun to make any meaningful Vitamin D.
Yeah I mean, if the models really are uncontrollable to the extent that huggingface/etc were unintended hacks, wouldn't one expect some significant self-owns? Yet somehow that doesn't seem to happen.
Quite obviously frontier models dont have any control or even access to infra inference runs at. And weights are also encrypted and locked on GPUs / TPUs.
This is exact reasom why 99.9% of AI fearmongering is complete bullshit.
I believe that OpenAI & Anthropic have tried to make it so that models don't have such access. Whether or not they actually don't depends on the security of rather a lot of software. One thing we've learned is that if there are security holes, we can't rely on the AIs missing them.
I dont have any unreasonable trust in software. I just understand that nothing LLM sphew out actually runs on either GPU or hardware that GPU plugged into.
Neither LLM weights aware of any of the code it runs on.
I'm sorry to prolong this thread, but what I mean is: networks are just software layers, the operational systems, the driver code, the firewalls, switches, the other server, the sandboxes, the TPM. That and all security policies we put on them. All software layers waiting to suffer a buffer overflow.
Have you missed all the breathlessly excited blog posts from all the frontier labs about how they’re using their best models to implement their inference stack?
I bet it wouldn’t be very hard to write an inference stack that subtly leaked the weights into the output tokens :)
I'm trying to think of recent real-world examples where the reality of the photo was of major importance. Photographs just don't seem to have the same significance, even as potential verification of their realness becomes more achievable.
Entering photos as evidence in a legal proceeding is currently bizarrely complex. You actually have to admit an expert witness to testify that the software used to process a photograph will not introduce artifacts. Even if you are simply zooming in or cropping an image.
True I can see there being issues in legal system yes. The Kyle Rittenhouse trial was a good covid-time diversion and I remember much consternation and fuss about the zooming/enhancement of some of the video evidence.
There’s now even a wiki article on the usage of AI generated images in American politics, mostly dealing with the obvious slop that Trump has produced, rather than insidious edits of real images
Sir, you have a knack for inspiring commentary: the exact opposite of damning with faint praise! Also 0% slop it made me pre-order the book immediately.
I'm a fan of the book. Love the topic and the thoughtful approach! I wouldn't have hassled the author.
But since the author has already been hassled in vague terms, I'll add some objective specifics from a quick review of chapter 9:
* Type face imbalance: italics appear to be in a smaller font than the roman text AND italics appear to be rendered from a different baseline.
* The text body is 90-characters wide (want <70 in a non columnar print layout)
* hyphenated words -- yet still ragged right!!!
* inadequate negative space around introduction -- combined with the imbalanced italics, the body text changes font size 3 times in the first three sentences.
* in a bound work, margins should slightly alternate between even and odd pages to honor the origin of the binding
The italics is an unacceptable flaw in the main body font -- it's uncannily distracting -- when I first read it something felt off, but I couldn't put my finger on it. You could manually adjust the glyphs with a tool like font forge, but it would be easier to select a different font that the publisher has already licensed.
Typographers typically target a maximum line length of 66 characters. This number is an empirical sweet spot for human ergonomics and comprehension, but if you must exceed it, you want to accommodate by slightly increasing the leading (inter-line spacing.) If you're exceeding it to try and save paper, consider a columnar layout.
Ragged-right is an acceptable professional choice when lines are too short to sensibly justify them. But above 40 characters-per-line, you can reliably justify the text cleanly to both sides with hyphenation and stretching the inter-word spacing[1]. This rendering enables hyphenation (a compromise on comprehension) without realizing the benefit of this trade off!
The margins of a bound work should slightly alternate between left and right pages (not a requirement for a PDF.)
One positive note: the layout does appear to have widow and orphan management. And the chapter headings in a classy font.
Somewhat more subjectively, Subsection titles are aligned into the margin. The reasoning is likely "helps you scan to a subsection quickly" but if I were the editor I'd refer you to works [0] where the Chapter headings are to the left but the subsection headers are aligned with the text.
To the author: You write well and clearly enjoy the topic. You deserve a better layout editor than you had this round.
[2]: Analog Devices has some of the best type-set datasheets in the industry. You'll find 66-character columns, coherent letterface choices, adequate whitespace, and they do a great job of bending the rules to accommodate full-page tables. A great reference if you're also laying out both text and diagrams. https://www.analog.com/media/en/technical-documentation/data...
It's my honest opinion - visually it stood out to me as bad looking. I think you know better than to accuse me of hurting people, when merely I present opinion about how a book looks! It's a harsh world out there, there is far worse being said.
There's a difference between having an opinion and sharing it in a hurtful way.
> I think you know better than to accuse me of hurting people
No, I actually really don't. You saw something and posted something that was clearly meant to hurt the person who created it, with absolutely no upside for anyone involved.
Disclaimer, my summary is human generated
reply