Bingo.

Those Indian staffing firms are basically giant scam companies. They only hire foreign nationals, who happen to be... mostly indian. Gee, wonder WHY that is.

Those companies should not exist, at all. They're basically outsourcing firms that hire the cheapest, most incompetent people and have them work for companies who want to pay the cheapest prices for jobs that should probably just be automated.

So, while I think the 100K price is absurd, I think the requirement must be for the wage to be paid to be 500% of the housing cost (so if a San Francisco company wants to hire a H1B, the prevailing median housing cost is 1.4m, which means that a housing payment is $9500/mo, which means $47,500/mo.) So that would discourage the hiring of H1B's to displace people who already live in expensive areas, and instead push companies that want to hire H1B's to move their companies into under--developed areas and get those areas to have a higher quality of life. Or those areas can push back and refuse to let them in. If they're hiring people who can't freaking afford to live there anyway, they shouldn't be able to hire them at all.

They are always kept spinning during training. If the server has enough enough disk cache, they will only be hit when written to. However a 320GB model being trained might still be using 16 20TB drives for the actual data being trained on if they're keeping it around and not just ripping it from servers on the fly.

Nah. Most models in use, are still only about 500GB tops. That's perfectly suitable for a SSD to load for inference.

The problem is training because checkpoints are 500GB per iteration. So you may write a checkpoint every X many iterations and take the best one, so you aren't doing this with a 2TB SSD. The end result is that if you are trying to get the best model, you can't just set it and forget it, you have to set it and check it, and if you're throwing away 10 checkpoints without checking you might overshoot the best performing model.

There is some truth to this.

Generally too many people end up in "make busy" work. Basically doing things that nobody asked them to do, and has no goal to improve anything. It's just stuff that the employee does to look busy for their KPI's but doesn't actually move the needle outside their team.

You solve this by getting rid of KPI type of metrics and instead use results
"Thing asked - (thing) | Thing resolved (YES|NO)"
"If thing asked was not resolved, is employee still working on it, or does employee need help"
"Are people happy with (thing) now? (YES|NO)"

The problem is as soon as "does employee need help" is asked, a lot of real employees will say no, because they don't want their 100% work to become 200% work because now they need to train someone to do their job, while making their workload take even longer and then when the work is done, the company will push them out the door and keep the new guy since they are cheaper.

Point being, if you think there is waste from too many employees, you start at the top. Start by asking the C-suite which departments cost the most, and then have everyone under that department report back how many people they are managing, and which supervisors/team leads are less productive and retrain those people. If they don't improve, you fire the team AND the team lead AND the supervisor, and the Manager for all failing.

Everyone should want to succeed at their job, and if someone is dragging their ass around just to get a paycheck, whoever hired them needs to let them go and also demote themselves and take over that position.

More often than you think.

Engineering and Military firms often hire these third parties to "haul away e-waste" and they don't wipe anything, they assume the stuff is going to be destroyed, not resold.

eSIM's are a thing and would be inconspicuous to someone who wasn't looking for it. However the cellular antenna would be immediately obvious since there needs to be 15mm for 5G and about 30mm for LTE of antenna wire, so if you x-rayed it you would see it.

With that said, a less obvious use for said radios is probably to reconfigure them remotely if there is a vendor programmed eSIM. But my stronger guess here is they used the same hardware for some other device and the actual antenna and sim card pins aren't even attached to anything.

To quote from a linked article:
> The circuit boards might have been designed for multiple purposes, which could explain the presence of additional components, but Christensen emphasised they should not be included in equipment destined for energy infrastructure.

So it could be possible they just mass-produced BMS systems for multiple vendors and just shipped it without the necessary parts to use those compoents. Like consider how some SOC's like Qualcomm's are used for both smart phones and VR HMD's. You don't use the cellular radio on the VR HMD, so it doesn't actually have the power amplifiers for that, but does for the WiFi. I don't know why a chinese vendor would add $300 radios to devices not intended to be used that way, but that is a distinct possibility that they used SoC's that just had the functionally but not the necessary parts to use it.

I think people forgot the entire point of the SIM card was so that you were not locked into a carrier. Once that sim card physically goes away, so does your ability to switch carriers and use cheaper sim's in countries you travel to.

Hmm, you're not wrong. But I think it goes a little deeper.

Netflix just wants content, but doesn't seem to care if the content is good, which is bad for Netflix's longivity. If it just wanted content, it could pay vtubers $15,000 a month, each, to just produce the equivalent of a video podcast. That's a huge savings compared to spending millions of dollars on one 90 minute film.

But who would watch them. The most popular vtubers only have around 5000-10000 CCV. It might be more lucrative than streaming on twitch or youtube which maybe earns them $1000 in ad revenue, most of their revenue comes from fans throwing subscription funds at them directly.

So imagine if they (netflix) get that up front. Easy solution to have a content library, but it's not going to be something everyone must-watch.

Really, I think people want Netflix to stop paying stupid amounts of money to license content that doesn't justify (eg Friends, Seinfeld, The Big Bang Theory) the price, because the content is quite stale and will never justify the licence cost. Let the tv network who owns it keep re-running it for 30 years, because good god CBS and NBC's have extremely stale content libraries. There is no "rerun" viability in reality shows. Nobody wants to watch Survivor or Big Brother once it's over. You already know who wins. Quite honestly I'm surprised CBS doesn't keep rerunning "I love lucy"

Google forced all Gsuite users to pay for it. Literately, they raised the price and then went "we feel this is how much our AI offering is worth"

BS.

I hate WB, but I think I hate these AI companies more.

For reference WB is the company that sues people for making homemade superman costumes.

In the age before computers, how did these damn food logistics operations operate?

Yeah. Thought so.

The problem is that these companies have been "saving money" by getting rid of people that regularly did these tasks by hand, telephone, and later fax, replaced those tools with computers, and then decided to save money by only having "one" computer in their entire logistics network do it.

Not entirely.

Whisper actually works rather well in several specific use cases, and fails spectacularly in others. You need to know this in advance:
- Whisper is roughly 90% accurate at transcription and translation
- Whisper absolutely does not know what to do with silence and will randomly inject "subtitled by (fansub group, netflix, etc)" into silence
- Whisper does not really understand singing well
- Whisper does not understand code-switching (eg switching between English and Japanese in the same context window)
- Whisper understands zero onomatopoeia, just like all ASR systems.

With that said, it is not useful or reliable for:
1. Fansubbing, especially anything adult. It can only understand words, not onomatopoeia. So when it stumbles into a scene where someone goes "ah!" it has zero context for it. The result is actually pretty silly, and often turns sex scenes in R-rated and unrated media into a series of random gibberish words that begin with the same sound. Likewise children playing and women giggling often turns it into a series of nonsense, sometimes sexually charged words.
2. Transcription of podcasts. Sorry bub, your average podcaster has a shitty microphone, and can not subtitle when multiple people are speaking over each other. Especially when people use Zoom or Discord to have a multi-party video. If you want to use it to transcribe a podcast, record each participant separately and merge the result.
3. ASR technology is often built on corpus of bad data that elevates profanity when it tries to guess words it can not understand. So it's more likely to use racist language "trigger" becomes the same word with an n, that isn't even in the audio. So your input source must be professional grade, or it's word error rate will be higher and favor profanity or racist language over other more less-often but more obvious words.

I doubt most people will use this in practice as Whisper.cpp is insanely slow without being expressly used on a 16GB nvidia GPU anyway.

This is the reason "most likely" why certain trends are up. Cancers? Likely plastic breaking down and releasing hormone-mimicing chemicals. Decreases in fertility? Same. People going gay or transgender? same. Autism? Same. At last the bogeyman to blame everything not genetic we haven't discovered a cure for.

In the grand scheme of things, the entire point of plastic was that it didn't interfere with anything, it just passed through the system harmlessly. That's what Teflon does. However the chemical used to make it stick onto metal C-8 is one of those PFAS chemicals that is one of these forever-chemicals that are causing health problems in people.

Another thing? Oh did you know that the largest source of microplastics in food comes from the actual processing pipeline? That's why chicken nuggets have the highest concentration of microplastics. Another large source? Polyester, every time you wash your clothes, you are shedding microplastics all over your environment.

Unfortunately the only way we're getting microplastics or C8 out of the environment is to find a way to reclaim them from the environment, which means that's one more thing we need to be mindful of when sending stuff to the landfill.

KF is worse than that. They actively promote domestic terrorism.

Sites like KF (LC is another one) only exist because their hosting is in fact, NOT in the US. They're usually hosted somewhere that has no US law reach because they know while they might slide under the 1A, what they post is incitement to violence and defamation which is not protected speech. The sites continue to exist the same reason why your typical piracy torrent site still exists. It's far more likely that law enforcement is actually on these sites and tracking these shitheel posters so that they can prevent assassinations of people the site hates on if they see a clear intent.

Remember, these sites, YOU, the person reading the site, are the joke. Nobody with a real job they value posts to these sites. It's all full time home security guards (euphemism for people who never leave their home)

Cloudflare has no intention of complying either, also KF was booted off cloudflare a few years back for terrorism.