• 46 Posts
  • 274 Comments
Joined 1 year ago
cake
Cake day: June 9th, 2023

help-circle



  • Apple is doing insane stuff with iMessage making SMS useless with Android. It is intentional manipulation. I’ve only seen it on some Apple kernels with the latest iPhone. My old man’s phone absolutely will not send or receive an SMS unless I message is manually turned off and then it won’t get iMessages. Apple has always been a shitty company, but this is next level insane. Just get and use the Signal app and ditch all the proprietary garbage and manipulative bs IMO.



  • That is all model loader code based stuff. You basically just need a bit of Python code that anyone can write, even you. You need to have a look at how the model must be addressed in format, then you find how the model loader code you are using creates the structured text and just insert your own modifications.

    It is easier to add a second prompt window where you keep persistent information and just change it yourself. If you want the model to change that stuff, you need to look into building an agent.

    I hacked around like this for awhile too, but evolved into using the notepad tab of Oobabooga Textgen WebUI and a Mixtral 8×7B model with a 32k context window. That model can deal with almost any detail you add in the context.

    The thing is, the model may not feel satisfied with mundane and redundant details. Most of the time, believe it or not, those changes that you perceive as errors or inconsistencies are not actually errors. The model is fully aware that it is making those changes. There is usually an instruction about creativity, inspiration, or initiative that will cause the model to shift details like this.

    My best recommendation is to openly discus the error in dialog with the model. They usually will talk about the issue if you call one out. Talking with AI is very different than with humans. Everything they say has meaning and functions beyond what is initially apparent. There are a lot of behaviors that involve certain patterns and keywords. You can even alter these behaviors by banning certain tokens/keywords. A good place to start is banning paragraph starting words for filler material. “So_” would be a good starting token to ban, as with AI, no sentence that starts with that token is worth reading. Other important words are twist, chuck, and cross. These are more flexible in their troublesome meaning and can manifest in multiple words like chuckles or across but still function the same inside the model.

    If you get to this level of familiarity, the persistence of a storied interaction is more of a negotiation with the AI.

    Another point of understanding is that the model does not know the difference between roleplaying characters it owns and you own or who is who. You or the model loader code is creating the name of the next character that the model generates for. Everyone in any story can be the model if it is asked to create the prompt.

    So if I say, “Jake is super creative” or imply it somewhere in the prompt, and the model later thinks ‘Jake is rather bored and needs to be creative’ it is going to change something that satisfies that character profile even if it has never generated the output for character Jake.

    Hope that helps some



  • Speculation is pretty much irrelevant. What the USA is doing with semiconductors is proof that China is about to take Taiwan. What North Korea is doing in Ukraine is evidence of what is about to happen in South Korea. WW3 is inevitable within 10 years. Israel is getting ahead of their vulnerabilities to fortify against their vulnerable flanks once Türkiye flips sides to take Cypress and several Greek islands. Millions now alive will not see 2040, perhaps billions. The subsentient human animal is about to emerge again because we fail at fundamental reasoning skills and the application of proven game theory.





  • I’ve had this happen with AI stuff that runs in a Python venv. It only happens with apps that use multi threading, and usually when something is interrupted in an unintended or unaccounted for way. I usually see it when I start screwing with code stuff, but also from changing the softmax settings during generation or crashing other stuff while hacking around. There may be a bug of some kind, but I think it likely has more to do with killing the root threading process and leaving an abandoned child that doesn’t get handled by the kernel process scheduler in the standard way. If this happens I restart too.



  • Wow:

    Oleksiy Protas

    P.S. “Don’t feed the trolls”

    Don’t you worry. Our friend here tried to reply to this message, he did so twice in fact with slightly different wording, but it was full of political rage and tu quoque so I assume he fell victim to the spam filter thanks to you special counter-baiting operation so to speak.

    That aside, I did a very superficial search and it seems that the original author had already had a pull being rejected on the grounds it was coming straight from his Baikal credentials. It’s a real pity that an apparently very able engineer is just playing pretend despite knowing full well why is it so that LF migh not want to be associated with Baikal in any way.

    Serge Semin

    Hello Linux-kernel community,

    I am sure you have already heard the news caused by the recent Greg’ commit 6e90b675cf942e (“MAINTAINERS: Remove some entries due to various compliance requirements.”). As you may have noticed the change concerned some of the Ru-related developers removal from the list of the official kernel maintainers, including me.

    The community members rightly noted that the quite short commit log contained very vague terms with no explicit change justification. No matter how hard I tried to get more details about the reason, alas the senior maintainer I was discussing the matter with haven’t given an explanation to what compliance requirements that was. I won’t cite the exact emails text since it was a private messaging, but the key words are “sanctions”, “sorry”, “nothing I can do”, “talk to your (company) lawyer”… I can’t say for all the guys affected by the change, but my work for the community has been purely volunteer for more than a year now (and less than half of it had been payable before that). For that reason I have no any (company) lawyer to talk to, and honestly after the way the patch has been merged in I don’t really want to now. Silently, behind everyone’s back, bypassing the standard patch-review process, with no affected developers/subsystem notified - it’s indeed the worse way to do what has been done. No gratitude, no credits to the developers for all these years of the devoted work for the community. No matter the reason of the situation but haven’t we deserved more than that? Adding to the GREDITS file at least, no?..

    I can’t believe the kernel senior maintainers didn’t consider that the patch wouldn’t go unnoticed, and the situation might get out of control with unpredictable results for the community, if not straight away then in the middle or long term perspective. I am sure there have been plenty ways to solve the problem less harmfully, but they decided to take the easiest path. Alas what’s done is done. A bifurcation point slightly initiated a year ago has just been fully implemented. The reason of the situation is obviously in the political ground which in this case surely shatters a basement the community has been built on in the first place. If so then God knows what might be next (who else might be sanctioned…), but the implemented move clearly sends a bad signal to the Linux community new comers, to the already working volunteers and hobbyists like me.

    Thus even if it was still possible for me to send patches or perform some reviews, after what has been done my motivation to do that as a volunteer has simply vanished. (I might be doing a commercial upstreaming in future though). But before saying goodbye I’d like to express my gratitude to all the community members I have been lucky to work with during all these years. Specifically:

    NTB-folks, Jon, Dave, Allen. NTB was my starting point in the kernel upstream work. Thanks for the initial advices and despite of very-very-very tough reviews with several complete patchset refactorings, I learned a lot back then. That experience helped me afterwards. Thanks a lot for that. BTW since then I’ve got several thank-you letters for the IDT NTB and IDT EEPROM drivers. If not for you it wouldn’t have been possible.

    Andy, it’s hard to remember who else would have given me more on my Linux kernel journey as you have. We first met in the I2C subsystem review of my DW I2C driver patches. Afterwards we’ve got to be frequently meeting here and there - GPIO, SPI, TTY, DMA, NET, etc, clean/fixes/features patch(set)s. Quite heat discussions in your first reviews drove me crazy really. But all the time we managed to come up with some consensus somehow. And you never quit the discussions calmly explaining your point over and over. You never refused to provide more detailed justification to your requests/comments even though you didn’t have to. Thanks to that I learned how to be patient to reviewers and reviewees. And of course thank you for the Linux-kernel knowledges and all the tips and tricks you shared.

    Linus (Walleij), after you merged one of my pretty much heavy patchset in you suggested to me to continue the DW APB GPIO driver maintaining. It was a first time I was asked to maintain a not-my driver. Thank you for the trust. I’ll never forget that.

    Mark, thank you very much for entrusting the DW APB SSI driver maintenance to me. I’ve put a lot of efforts into making it more generic and less errors-prune, especially when it comes working under a DMA-engine control or working in the mem-ops mode. I am sure the results have been beneficial to a lot of DW SPI-controller users since then.

    Damien, our first and last meeting was at my generic AHCI-platform and DW AHCI SATA driver patches review. You didn’t make it a quick and easy path. But still all the reviews comments were purely on the technical basis, and the patches were eventually merged in. Thank you for your time and experience I’ve got from the reviews.

    Paul, Thomas, Arnd, Jiaxun, we met several times in the mailing list during my MIPS P5600 patches and just generic MIPS patches review. It was always a pleasure to discuss the matters with such brilliant experts in the field. Alas I’ve spent too much time working on the patches for another subsystems and failed to submit all the MIPS-related bits. Sorry I didn’t keep my promise, but as you can see the circumstances have suddenly drawn its own deadline.

    Bjorn, Mani, we were working quite a lot with you in the framework of the DW PCIe RC drivers. You reviewed my patches. I helped you to review another patches for some time. Despite of some arguing it was always a pleasure to work with you. Mani, special thanks for the cooperative DW eDMA driver maintenance. I think we were doing a great work together.

    Paolo, Jakub, David, Andrew, Vladimir, Russell. The network subsystem and particularly the STMMAC driver (no doubt the driver sucks) have turned to be a kind of obstacle on which my current Linux-kernel activity has stopped. I really hope that at least in some way my help with the incoming STMMAC and DW XPCS patches reviews lightened up your maintainance duty. I know Russell might disagree, but I honestly think that all our discussions were useful after all, at least for me. I also think we did a great work working together with Russell on the DW GMAC/QoS ETH PCS patches. Hopefully you’ll find a time to finish it up after all.

    Rob, Krzysztof, from your reviews I’ve learned a lot about the most hardwary part of the kernel - DT sources and DT-bindings. All your comments have been laconic and straight to the point. That made reviews quick and easy. Thank you very much for that.

    Guenter, special thanks for reviewing and accepting my patches to the hwmon and watchdog subsystems. It was pleasure to be working with you.

    Borislav, we disagreed and argued a lot. So my DW uMCTL2 DDRC EDAC patches even got stuck in limbo for quite a long time. Anyway thank you for the time you spent reviewing my patches and trying to explain your point.

    • Borislav, it looks like I won’t be able to work on my Synopsys EDAC patchsets anymore. If you or somebody else could pick them up and finish up the work it would be great (you can find it in the lore archive). The patches convert the mainly Zynq(MP)-specific Synopsys EDAC driver to supporting the generic DW uMCTL2 DDRC. It would be very beneficial for each platform based on that controller.

    Greg, we met several times in the mailing lists. You reviewed my patches sent for the USB and TTY subsystems, and all the time the process was straight, highly professional, and simpler than in the most of my other case. Thank you very much for that.

    Yoshihiro, Keguang, Yanteng, Kory, Cai and everybody I was lucky to meet in the kernel mailing lists, but forgot to mention here. Thank you for the time spent for our cooperative work on making the Linux kernel better. It was a pleasure to meet you here.

    I also wish to say huge thanks to the community members trying to defend the kicked off maintainers and for support you expressed in these days. It means a lot.

    A little bit statics of my kernel-work at the end:

    Signed-off patches: 518 Reviewed and Acked patches: 253 Tested patches: 80

    Best Regards, -Serge(y)



  • j4k3@lemmy.worldtoLinux@lemmy.mlSo all of my drivers are breaking
    link
    fedilink
    English
    arrow-up
    5
    arrow-down
    1
    ·
    1 month ago

    That is what I meant by configure. You’re not going to HP to download your printer driver or realtek to get one for your network adaptor. To the end user, the kernel includes the required modules, or it is a matter of simple configurations. The exception being proprietary garbage. However with Nvidia on Fedora, it is a non issue as the Anaconda system builds the Nvidia module from source with every kernel update from outside of the kernel but under the shim, so even secure boot works.

    The OP was not asking computer science OS 101. My reply is just intended as a surface level to cause them to question the drivers mentality. I’ve seen many people follow this logic and not get anywhere.


  • j4k3@lemmy.worldtoLinux@lemmy.mlSo all of my drivers are breaking
    link
    fedilink
    English
    arrow-up
    6
    arrow-down
    1
    ·
    1 month ago

    Indeed, gaps are present in my knowledge. I understand what you wrote, in theory, but vaguely based on my reading from a forum on kernel architectures several years ago. I’m most familiar with the user experience of configuring a custom Linux kernel with Gentoo versus needing a WiFi driver that I need WiFi access to source.

    Since you are touching on a gap in my knowledge, perhaps a more recent issue and curiosity will help me ground this a little better if you do not mind responding. What is the deal with secure boot and Windows drivers? How are they able to run some random driver from the internet that has DMA?



  • j4k3@lemmy.worldtoLinux@lemmy.mlSo all of my drivers are breaking
    link
    fedilink
    English
    arrow-up
    13
    arrow-down
    6
    ·
    1 month ago

    That sounds like a hardware issue.

    Keep in mind that Linux is a monolithic kernel. It doesn’t technically have drivers at all or go missing. All supporting kernel modules for hardware are always present at the configuration level. The general kernels shipped by distros are configured to work out of the box for most hardware. The only exceptions should be instances where oddball hardware can cause conflicts with the standard way other hardware works in the same space. Then there are cases where hardware is totally undocumented publicly by the chip manufacturers. That is the worst kind as some of those have poor or no support.

    By contrast, Windows is a microkernel. It only creates an API layer for the hardware vendor to write a driver that interfaces with Windows. They leave it entirely up to the end user to get stuck in the middle, source and install the driver and deal with any potential issues. In other words they don’t have devs to maintain or do anything meaningful in this space, and they enable undocumented proprietary crap hardware.