I'm no expert (just a monkey... ;) ), but isn't Diffusion supposed to generate ALL of the output at once? From their diagram, it looks like their I-LDM model seems to use previously generated context to generate the next tokens (or blocks).
Block auto regressive generation can give you big speedups.
Consider that outputting two tokens at a time will be a (2-epsilon)x speedup over running one token at a time. As your block size increases, you quickly get to fast enough that it doesn't matter sooooo much whether you're doing blocks or actual all-at-once generation. What matters, then, is there quality trade-off for moving to block-mode output. And here it sounds like they've minimized that trade-off.
can it go back and use future blocks as context? Thats what i'm most interested in here - fixing line 2 because of a change/discovery we made in the process of writing line 122. I think that problem is a big part of the narrowsightedness of current coding models
I would love to see a summary of all of the various options being bandied about.
There are 2 components in my mind: the backup "agent" (what runs on your laptop/desktop/server) and the storage provider (which BB is in this context).
What do people recommend for the agent? (I understand some storage providers have their own agents) For Linux/MacOS/Windows.
What do people recommend for the storage provider? Let's assume there are 1TB of files to be backed up. 99.9% don't change frequently.
If the navigation simulates what would happen if we follow links to SPA#pos1, SPA#pos2, etc so that if I do two clicks within the SPA, and then hit Back three times I'm back to whatever link I followed to get to the SPA, I guess it's OK and follows user expectations. But if it is used as an excuse to trap the user in the SPA unless they kill the tab, not OK.
> From the browsers perspective those are the same thing though.
If the browser only allows adding at most one history item per click, I should be able to go back to where I entered a given site with at most that many back button clicks.
At a first glance, this doesn't seem crazy hard to implement? I'm probably missing some edge cases, though.
Some browser APIs (such as playing video) are locked behind a user interaction. Do the same for the history API: make it so you can't add any items to history until the user clicks a link, and then you can only add one.
That's not perfect, and it could still be abused, but it might prevent the most common abuses.
>> So that in single-page applications, it can work intuitively instead of always taking you all the way out of the app.
Just implement an additional back button on the SPA. This is actually not confusing and is done in some places. Navigation buttons within an SPA are common enough.
> My proof-in-pudding test is still the fact that we haven't seen gigantic mass firings at tech companies
This assumes that companies will announce such mass firings (yeah, I'm aware of WARN Act); when in reality they will steadily let go of people for various reasons (including "performance").
From my (tech heavy) social circle, I have noticed an uptick in the number of people suddenly becoming unemployed.
We can reduce this to an even more basic question: if these small models are equally comparable in finding vulnerabilities, why haven't they done so yet?. After all, the source code is out in the open, and has been for decades. Please go ahead, find (and report) the vulnerabilities.
All of these sites do shady shit. I'm so glad I'm no longer single.
I signed up for eHarmony with a unique email address dedicated to that site. After wasting 6 months, I chose to delete my account.
Lo and behold, soon spam started to show up on this account, as if the floodgates had been opened. It was a unique account that I had not used anywhere else just for this specific reason, and my hunch was justified.
After wasting 6 months, I chose to delete my account.
Lo and behold, soon spam started to show up on this account, as if the floodgates had been opened.
Facebook is also guilty of this.
I set up a Facebook account for a relative around 2006. The e-mail address is name_facebook@ a domain that I control.
Every six months or so, Facebook will send out almost daily e-mails for a month saying "Person x commented on your post!" or some variant. You know how I know this relative of mine didn't make a new post?
Since all the alleged comments are allegedly from people he knew, and not new strangers, I find it hard to believe that someone has been impersonating him on Facebook for the last 15 years.
How do you know they are not commenting on old content? FB could be pushing old content like 'remember X from 18 years ago' and then someone comments about remembering their friend under an old photo.
I do this for every site I sign up for. I have a 'catch all' email address, so I can put whateverIwant@mydomain.com and the emails will get to my inbox. So now I know who is selling or leaking my email address. So far it's been very few, but I also don't sign up for new sites very often.
They were good 15 years ago. As with all things, it went to shit when Match.com started consolidating everything and the bean counters realized that a quality product was not as profitable.
Surprised it took this long to get litigation. So many people complaining about how crap dating sites are, but no one thought to realize the site itself was the problem and fell into the whole "looksmaxxing" grift. Some people really will do anything except admit that rich people are corrupt.
Dating sites are an extremely hard business to be in.
On a traditional (social) network, whether that'd be Facebook, the railroad or the Bloomberg Terminal, you have the network effect. The more users you get, the more interesting the network becomes, which means yet more users want to join. This is a positive feedback loop.
The entire point of a dating site is to find somebody to leave that site with. Statistically speaking (and at that scale, statistics is the only thing that matters), attractive users[1] are more likely to find a match and leave, while unattractive users are likely to stay (or come back) and keep looking. As time goes on, the fraction of unattractive users will keep increasing. You can fix this with enough growth, but exponential growth can't go on forever due to population size constraints. And once you get into that state, your growth will be constrained further, which just puts you onto a downward slide into hell.
And then there's the question of revenue. It's hard to scalably do deals where the user must pay you when they find a relationship on the site (like the matchmaking services of old used to do), so subscriptions or one-time fees are your only options. Neither of them are great, subscriptions encourage you to keep users on the site (which puts your goal opposite to what the users want), one-time fees work against the network effect and constrain growth.
[1] By "attractiveness", I mean something much wider than just visual / physical attractiveness. An ability to enter and maintain a successful long-term relationship in general.
Also "jg" reads very similar to "jq", and initially I thought he was talking about "jq" all along, and I was like: where can I see the "jasongrep" examples? Threw me off for a minute.
reply