A lot of how the model acts is because of the harness not the model itself. It isn't self aware so it has no idea it is GPT-6 until the harness tells it what it is. GPT-6 didn't exist yet when they started training GPT-6 so how would it have knowledge on that?
The harness has a lot of components but one portion of that is a set of default instructions on things it should know and how it should act. Those things that it should know become part of the conversation between you and the actual model (but you don't see it).
So you say "you are Astra right?" and in the background the harness had already told it that it is "GPT-6". It just spit out what it "knew" and didn't "reason" that those were the same thing until the user forced it to. It will continue to "know" that it is GPT-6 Astra as long as your conversation fits into the context window.
Once your conversation gets too long to fit into the context window it either has to either start dropping the oldest part of the conversation or summarize your entire conversation into a shorter summary and start fresh with that. If either of those things drop the part of the conversation about GPT-6 and Astra being the same thing, then yes it may claim that it isn't Astra again in the same conversation. It isn't on purpose it is just a limitation of the context window (which is constrained by hardware/mathematical limitations).
Though being in the context window doesn't guarantee that it always answers the same, anything that is in the context window just affects the output. Think of it as each token across the entire context window pushing and pulling the result to different results, every previous token affects any new tokens. Having the tokens "GPT-6 and Astra are the same thing" in the context will pull any future results towards the answer that it is indeed Astra.
If they had explicitly put in the harness "don't call yourself Astra" your statement and their statement are both in the context and are fighting each other, and the output may be somewhat random. Position in the context matters though. More recent tokens affect the output much more than older tokens.
"Scam" was certainly hyperbole, I just exaggerated it to drive home the point. The casino bit could have been worded better too. Playing the game or collecting cards when you know what you are buying is not gambling, it is opening packs that's like going to the casino.
Paying for a pack that you have no idea what the contents are hoping for a big card is no different than putting money down on a hand of blackjack and hoping for a payout. The game or the collecting is just what we use to justify the gambling as buying something with real world utility.
I don't exactly have a problem with the business strategy (other than targeting kids), go ahead and gamble, and I can't claim to be above participating. It's just that the players need to be aware of what is happening and that our brains are a lot more easily manipulated than we think we are. Entertainment value with a drip feed of dopamine hits is pretty powerful.
Because free speech is still a thing. They would need to find evidence that he had/distributed csam content, or had actually molested someone.
Best we can hope for is that they started an investigation and will manage to link something to him, but an adult talking to other adults isn't illegal so they aren't going to arrest him on the spot in the airport.
If they are working on an investigation they won't actually arrest him until they think they have enough concrete evidence to actually win a case, and that is usually weeks/months later.
It is as much a scam as the cards themselves. Artificial scarcity designed to addict our brains to gambling.
The grading feeds on/into the main scam of the cards themselves. The whole cycle requires that we place "value" on their slabs of paper and grading helps make that value more concrete in your brain. The more value you see in the cards the more of them that you want to buy. More cards sold means more grading, so the cycle goes round.
Have fun playing the game, just be aware the whole thing is no different than going to the casino. They just hope the psychological effect of not gambling with money addicts extra people who wouldn't normally gamble with money (i.e. children or those morally opposed to gambling)... and the house always wins.
This would be more of a use case for blockchain, any site(/user) wishing to mirror the social media would participate in the chain. A post on one site would get pushed into the chain and show up on every mirror.
Media would likely have to be links outside of the blockchain since it isn't usually very good at storing large datasets (it would make it difficult to mirror over time as it grew), so if you wanted a full mirror you would probably have your site fetching media as soon as it is posted and mirroring it yourself.
Or media might be a good spot to integrate with your bittorrent idea. Blockchain handles the distributed state of the site (communities/post text+metadata), and any submitted media is added to the post as bittorrent hashes. I don't know how scalable bittorrent clients are though it might take considerable overhead to seed millions of pictures/videos.
There are probably millions of "Netgear" SSIDs across the globe so Google doesn't really use your network name to do the mapping. Every wifi radio also has a BSSID, which is the unique MAC address for your radio. This or the combination of both would be what they are using to do the lookup.
So adding the _nomap to your ssid is the only option you have to prevent your BSSID from being added to their database. If you don't trust Google to honor the _nomap you could regularly change your BSSID. Many consumer routers let you change your WAN MAC but almost none of them will let you change your BSSID.
This is normal. Linux caches disk reads so that you don't have to go to her/SSD to reread blocks. If your app isn't using the ram Linux will use it instead.
Zero reason to be concerned, because if your app allocates more memory Linux will evict the cached pages to make sure your apps have plenty of space.
In this case your apps have memory that isn't being touched (sometimes a sign of a memory leak), and Linux has decided that caching your disk access is more advantageous than keeping the unused application memory in ram.
If you have apps that you don't use often and are really slow to access when you finally do, keep lowering that swappiness value or just remove swap all together if it annoys you.
Looking at your graph you have a lot of ram that you don't really need... I would just disable swap entirely. Otherwise you are just wasting write cycles on your SSD.
License plate reader just means it is a higher class of camera with a good enough sensitivity+extremely low exposure time to read license plates (license plates are really hard to read especially in the dark due to their reflectivity+the speed of the cars).
So it doesn't mean they are only for license plates, it means it is a higher class of camera than your standard cctv camera.
Best case scenario is that it uses an esp32 microcontroller. If so you can use esphome (https://esphome.io/) to easily make your own firmware that you can then control from web/API/homeassistant with no external cloud/app requirements.
There has been some work to include other microcontrollers (NRF/Raspberry Pi pico/etc), and some third party repos that add some support for some realtek chips.
If you are getting a portable air conditioner get one of them that has dual hoses. The single hose units are practically a scam.
The single hose units take cool air from inside, warms up that air, and then blows it outside. This does make cold air inside the house, but you are using your already cooled room air to do the heat transfer and blowing that outside. This also means you are creating a vacuum in your house which pulls in more hot air from outside.
The dual hose units pull in warm air from outside in the first hose, heats it up further, and then blows it back outside out the second hose. No negative pressure in the house, and you aren't using air that you have been paying to cool all day to do the heat transfer.
these devices also routinely spoof themselves as mobile phones clicking ads on AI-generated websites as part of a sprawling operation that seeks to defraud online merchants and advertising networks.
Energy and mass are just different measures of the same thing. The transporters job is to take mass and turn it into something that it can more easily move around (a wave? Particle stream?) and back. So as long as the transporter is connected to the ship's energy supply it should theoretically be able to use that energy (instead of the energy from your body) to reassemble any pattern it wants without breaking the law of conservation of mass.
Riker is 6'4 let's say 200lbs, which is 8,153,369 terajoules.
DATA: Imagination is not necessary. The scale is readily quantifiable. We are presently generating twelve point seven five billion gigawatts per (an alarm goes off)
One watt is one Joule per second, there is some debate out there about the "per" part of the quote since watts already implies per second, but we probably assume the ship is capable of generating 12,750,000 terawatts, or 12,750,000 terajoules per second.
8,153,369/12,750,000 gets us about 0.64 seconds of the ships output to create our duplicate Riker.
However the quote above was when the ship was basically idling, (not actually at warp) so the actual output of the ship is likely exponentially larger.
Basically they have so much power available, that they can just rearrange it into whatever mass they want.
That's the whole point of contracts, you can't just cancel them. Some contracts have ways out if that was how they were negotiated.
Something like this with a lot of expensive hardware was probably set up to pay off that hardware over the life of the contract, so flock isn't going to let you out of it until they have recouped their hardware+install+removal costs. The only way out might be to pay the full term of the contract which the city wouldn't have budgeted to do in a single year.
Btrfs raid5/6 support is unstable/experimental and cause some serious issues in the past, so it isn't really recommended.
Since you only have 4 drives you could do a pair of mirrors with btrfs, but you aren't guaranteed to be able to handle two drives failing (depends on which two drives fail). So zfs with raidz2 is the best protection you can get, and it matches the capacity you would get from mirrors.
Rebuild time isn't great, but you would need a second drive to fail plus at least a read failure on another drive before you have issues.
A bigger question would be how soon would you have a replacement? If you already have a spare on hand I wouldn't worry about rebuild time at all, but if you are expecting to wait potentially weeks for a warranty replacement your chances of the second failure go up.
Even if you had a second failure and additional read failure is unlikely (how often do you see read failures when you run a scrub). Combine that with your backups.... You should have very little to worry about.
If two drives failed and you ran into a couple of sectors that can't be read ZFS continues to operate just fine, except for the failed file. The file with the failed blocks shows up in zpool status so you know exactly where the corruption is, and you can just copy that single file from your backups and everything is back to normal.
If your files are mostly WORM files like media/documents then your backups cover you really well and copying a file or two from backups isn't a concern. Vs if you are running virtual machines or DBs that are writing to their virtual disk constantly then you would start to worry about how much data you lose by rolling that file back to your past backup.
Is the chain anywhere near needing replacing (any stiff links)? Usually you wouldn't change the sprocket until the chain needs to be replaced.
I think I have read in the past that you should always do them both because as the chain stretches the sprocket tends to wear to match the stretch, but I have reused sprockets without any problems.
Well that would be the description of the community, but the actual rules section doesn't say anything about privacy/control.
So at the end of the day Plex is self hosted (you run most of it) so it should qualify. It might not 100% match the spirit of self hosting it does still meet the definition.
You can argue most Jellyfin/emby installs have the same problem because most users are still are dependent on external services because of things like metadata plugins.
And on the privacy front those plugins aren't any better than Plex. For instance The Movie DB which is the primary movie and TV metadata provider for Jellyfin has a privacy policy that clearly says they will use and share any interaction you have with the site including location and personal information. They almost certainly keep track of what is in your library. They don't have a user account for you that they can use to track across IPs, but if your ISP keeps you on the same IP for long periods of time they have a good idea of what you are watching.
You can run Jellyfin without those plugins enabled but unless you want to build/collect manual nfo files to import that data you are going to have a subpar experience.
Same problem for the **arr stack since they need metadata as well. Some of which go to different providers so you are giving out that information to additional parties (i.e. Sonarr goes to TheTVDB which has a similar privacy policy).
You can configure the arrs to write out nfo metadata and have Jellyfin consume that so that at least you aren't giving away your info to two external parties.
Well they did try to schedule delivery and apparently UPS wouldn't let them do that, but yeah certainly not anything to do with Valve.
If this were pre-covid UPS likely wouldn't have left a note and reattempted delivery the next day. But all of the shippers took the COVID no-contact rules to mean that they could just leave packages at the door, snap a pic, and be done with it. Saves them a huge amount on redelivery attempts, and they get to pass the blame on to their customers for not being there.
UPS asking you to open a case with Valve is normal. You didn't pay to ship it, Valve did. Even if it is a line item on your invoice it was a contract between Valve and UPS, so you can't call up UPS demanding investigation/money because it is Valve that gets the money not you (who in turn refunds it to you).
If Valve gives you the money back they are likely taking a loss on it to keep you happy. As far as UPS is concerned they delivered it (presumably with photographic evidence), so why would they pay up on insurance. Unless Valve pays for the higher level of insurance UPS won't cover for porch pirates.
It is unlikely that Valve paid to have that level of coverage, but that is because they know only a small number of packages will be stolen. It ends up being cheaper to just give a few refunds than it is to pay for the higher level coverage on thousands of packages.
Hopefully that is the case on how Valve handles this and you get a replacement or at least your money back. Good luck!
Not an AI expert but here is my understanding.
A lot of how the model acts is because of the harness not the model itself. It isn't self aware so it has no idea it is GPT-6 until the harness tells it what it is. GPT-6 didn't exist yet when they started training GPT-6 so how would it have knowledge on that?
The harness has a lot of components but one portion of that is a set of default instructions on things it should know and how it should act. Those things that it should know become part of the conversation between you and the actual model (but you don't see it).
So you say "you are Astra right?" and in the background the harness had already told it that it is "GPT-6". It just spit out what it "knew" and didn't "reason" that those were the same thing until the user forced it to. It will continue to "know" that it is GPT-6 Astra as long as your conversation fits into the context window.
Once your conversation gets too long to fit into the context window it either has to either start dropping the oldest part of the conversation or summarize your entire conversation into a shorter summary and start fresh with that. If either of those things drop the part of the conversation about GPT-6 and Astra being the same thing, then yes it may claim that it isn't Astra again in the same conversation. It isn't on purpose it is just a limitation of the context window (which is constrained by hardware/mathematical limitations).
Though being in the context window doesn't guarantee that it always answers the same, anything that is in the context window just affects the output. Think of it as each token across the entire context window pushing and pulling the result to different results, every previous token affects any new tokens. Having the tokens "GPT-6 and Astra are the same thing" in the context will pull any future results towards the answer that it is indeed Astra.
If they had explicitly put in the harness "don't call yourself Astra" your statement and their statement are both in the context and are fighting each other, and the output may be somewhat random. Position in the context matters though. More recent tokens affect the output much more than older tokens.
"Scam" was certainly hyperbole, I just exaggerated it to drive home the point. The casino bit could have been worded better too. Playing the game or collecting cards when you know what you are buying is not gambling, it is opening packs that's like going to the casino.
Paying for a pack that you have no idea what the contents are hoping for a big card is no different than putting money down on a hand of blackjack and hoping for a payout. The game or the collecting is just what we use to justify the gambling as buying something with real world utility.
I don't exactly have a problem with the business strategy (other than targeting kids), go ahead and gamble, and I can't claim to be above participating. It's just that the players need to be aware of what is happening and that our brains are a lot more easily manipulated than we think we are. Entertainment value with a drip feed of dopamine hits is pretty powerful.
Because free speech is still a thing. They would need to find evidence that he had/distributed csam content, or had actually molested someone.
Best we can hope for is that they started an investigation and will manage to link something to him, but an adult talking to other adults isn't illegal so they aren't going to arrest him on the spot in the airport.
If they are working on an investigation they won't actually arrest him until they think they have enough concrete evidence to actually win a case, and that is usually weeks/months later.
It is as much a scam as the cards themselves. Artificial scarcity designed to addict our brains to gambling.
The grading feeds on/into the main scam of the cards themselves. The whole cycle requires that we place "value" on their slabs of paper and grading helps make that value more concrete in your brain. The more value you see in the cards the more of them that you want to buy. More cards sold means more grading, so the cycle goes round.
Have fun playing the game, just be aware the whole thing is no different than going to the casino. They just hope the psychological effect of not gambling with money addicts extra people who wouldn't normally gamble with money (i.e. children or those morally opposed to gambling)... and the house always wins.
This would be more of a use case for blockchain, any site(/user) wishing to mirror the social media would participate in the chain. A post on one site would get pushed into the chain and show up on every mirror.
Media would likely have to be links outside of the blockchain since it isn't usually very good at storing large datasets (it would make it difficult to mirror over time as it grew), so if you wanted a full mirror you would probably have your site fetching media as soon as it is posted and mirroring it yourself.
Or media might be a good spot to integrate with your bittorrent idea. Blockchain handles the distributed state of the site (communities/post text+metadata), and any submitted media is added to the post as bittorrent hashes. I don't know how scalable bittorrent clients are though it might take considerable overhead to seed millions of pictures/videos.
There are probably millions of "Netgear" SSIDs across the globe so Google doesn't really use your network name to do the mapping. Every wifi radio also has a BSSID, which is the unique MAC address for your radio. This or the combination of both would be what they are using to do the lookup.
So adding the _nomap to your ssid is the only option you have to prevent your BSSID from being added to their database. If you don't trust Google to honor the _nomap you could regularly change your BSSID. Many consumer routers let you change your WAN MAC but almost none of them will let you change your BSSID.
https://www.linuxatemyram.com/
This is normal. Linux caches disk reads so that you don't have to go to her/SSD to reread blocks. If your app isn't using the ram Linux will use it instead.
Zero reason to be concerned, because if your app allocates more memory Linux will evict the cached pages to make sure your apps have plenty of space.
In this case your apps have memory that isn't being touched (sometimes a sign of a memory leak), and Linux has decided that caching your disk access is more advantageous than keeping the unused application memory in ram.
If you have apps that you don't use often and are really slow to access when you finally do, keep lowering that swappiness value or just remove swap all together if it annoys you.
Looking at your graph you have a lot of ram that you don't really need... I would just disable swap entirely. Otherwise you are just wasting write cycles on your SSD.
License plate reader just means it is a higher class of camera with a good enough sensitivity+extremely low exposure time to read license plates (license plates are really hard to read especially in the dark due to their reflectivity+the speed of the cars).
So it doesn't mean they are only for license plates, it means it is a higher class of camera than your standard cctv camera.
Them's fighting words! ;) yyyy-mm-dd is superior in every way. Sorts properly and much less potential for confusion across cultures.
Best case scenario is that it uses an esp32 microcontroller. If so you can use esphome (https://esphome.io/) to easily make your own firmware that you can then control from web/API/homeassistant with no external cloud/app requirements.
There has been some work to include other microcontrollers (NRF/Raspberry Pi pico/etc), and some third party repos that add some support for some realtek chips.
If you are getting a portable air conditioner get one of them that has dual hoses. The single hose units are practically a scam.
The single hose units take cool air from inside, warms up that air, and then blows it outside. This does make cold air inside the house, but you are using your already cooled room air to do the heat transfer and blowing that outside. This also means you are creating a vacuum in your house which pulls in more hot air from outside.
The dual hose units pull in warm air from outside in the first hose, heats it up further, and then blows it back outside out the second hose. No negative pressure in the house, and you aren't using air that you have been paying to cool all day to do the heat transfer.
Sounds like a feature
If you click the link button it opens up the video instead of the preview jpg.
E=mc^2
Energy and mass are just different measures of the same thing. The transporters job is to take mass and turn it into something that it can more easily move around (a wave? Particle stream?) and back. So as long as the transporter is connected to the ship's energy supply it should theoretically be able to use that energy (instead of the energy from your body) to reassemble any pattern it wants without breaking the law of conservation of mass.
Riker is 6'4 let's say 200lbs, which is 8,153,369 terajoules.
One watt is one Joule per second, there is some debate out there about the "per" part of the quote since watts already implies per second, but we probably assume the ship is capable of generating 12,750,000 terawatts, or 12,750,000 terajoules per second.
8,153,369/12,750,000 gets us about 0.64 seconds of the ships output to create our duplicate Riker.
However the quote above was when the ship was basically idling, (not actually at warp) so the actual output of the ship is likely exponentially larger.
Basically they have so much power available, that they can just rearrange it into whatever mass they want.
That's the whole point of contracts, you can't just cancel them. Some contracts have ways out if that was how they were negotiated.
Something like this with a lot of expensive hardware was probably set up to pay off that hardware over the life of the contract, so flock isn't going to let you out of it until they have recouped their hardware+install+removal costs. The only way out might be to pay the full term of the contract which the city wouldn't have budgeted to do in a single year.
Btrfs raid5/6 support is unstable/experimental and cause some serious issues in the past, so it isn't really recommended.
Since you only have 4 drives you could do a pair of mirrors with btrfs, but you aren't guaranteed to be able to handle two drives failing (depends on which two drives fail). So zfs with raidz2 is the best protection you can get, and it matches the capacity you would get from mirrors.
Rebuild time isn't great, but you would need a second drive to fail plus at least a read failure on another drive before you have issues.
A bigger question would be how soon would you have a replacement? If you already have a spare on hand I wouldn't worry about rebuild time at all, but if you are expecting to wait potentially weeks for a warranty replacement your chances of the second failure go up.
Even if you had a second failure and additional read failure is unlikely (how often do you see read failures when you run a scrub). Combine that with your backups.... You should have very little to worry about.
If two drives failed and you ran into a couple of sectors that can't be read ZFS continues to operate just fine, except for the failed file. The file with the failed blocks shows up in zpool status so you know exactly where the corruption is, and you can just copy that single file from your backups and everything is back to normal.
If your files are mostly WORM files like media/documents then your backups cover you really well and copying a file or two from backups isn't a concern. Vs if you are running virtual machines or DBs that are writing to their virtual disk constantly then you would start to worry about how much data you lose by rolling that file back to your past backup.
Is the chain anywhere near needing replacing (any stiff links)? Usually you wouldn't change the sprocket until the chain needs to be replaced.
I think I have read in the past that you should always do them both because as the chain stretches the sprocket tends to wear to match the stretch, but I have reused sprockets without any problems.
Anyways it looks fine to me.
Well that would be the description of the community, but the actual rules section doesn't say anything about privacy/control.
So at the end of the day Plex is self hosted (you run most of it) so it should qualify. It might not 100% match the spirit of self hosting it does still meet the definition.
You can argue most Jellyfin/emby installs have the same problem because most users are still are dependent on external services because of things like metadata plugins.
And on the privacy front those plugins aren't any better than Plex. For instance The Movie DB which is the primary movie and TV metadata provider for Jellyfin has a privacy policy that clearly says they will use and share any interaction you have with the site including location and personal information. They almost certainly keep track of what is in your library. They don't have a user account for you that they can use to track across IPs, but if your ISP keeps you on the same IP for long periods of time they have a good idea of what you are watching.
You can run Jellyfin without those plugins enabled but unless you want to build/collect manual nfo files to import that data you are going to have a subpar experience.
Same problem for the **arr stack since they need metadata as well. Some of which go to different providers so you are giving out that information to additional parties (i.e. Sonarr goes to TheTVDB which has a similar privacy policy).
You can configure the arrs to write out nfo metadata and have Jellyfin consume that so that at least you aren't giving away your info to two external parties.
Well they did try to schedule delivery and apparently UPS wouldn't let them do that, but yeah certainly not anything to do with Valve.
If this were pre-covid UPS likely wouldn't have left a note and reattempted delivery the next day. But all of the shippers took the COVID no-contact rules to mean that they could just leave packages at the door, snap a pic, and be done with it. Saves them a huge amount on redelivery attempts, and they get to pass the blame on to their customers for not being there.
UPS asking you to open a case with Valve is normal. You didn't pay to ship it, Valve did. Even if it is a line item on your invoice it was a contract between Valve and UPS, so you can't call up UPS demanding investigation/money because it is Valve that gets the money not you (who in turn refunds it to you).
If Valve gives you the money back they are likely taking a loss on it to keep you happy. As far as UPS is concerned they delivered it (presumably with photographic evidence), so why would they pay up on insurance. Unless Valve pays for the higher level of insurance UPS won't cover for porch pirates.
It is unlikely that Valve paid to have that level of coverage, but that is because they know only a small number of packages will be stolen. It ends up being cheaper to just give a few refunds than it is to pay for the higher level coverage on thousands of packages.
Hopefully that is the case on how Valve handles this and you get a replacement or at least your money back. Good luck!