Whatever is Clever Public License
h3ndrik @ h3ndrik @feddit.de Posts 3Comments 469Joined 2 yr. ago
h3ndrik @ h3ndrik @feddit.de
Posts
3
Comments
469
Joined
2 yr. ago
I thought the main problem was that it's debatable whether you can enforce it. So it harms users and distributions because they can't really rely on it.
But the liability would definitely be another issue. I think the law is different here in Europe, so the liability might already be included for my hobby tinkering per default and I don't need to worry.
And something else is: I'd include trademark... force people to choose a different name for their project if they take my code so there is no confusion and people can't upload versions with advertisements on some software store.
Uh, just don't mistake it for a license. It's funny and I like to put my stuff under WTFPL. But in recent times I feel it's appropriate to point out that you sometimes don't do your users a favor by being silly (if it's useful code.)
It just arrived in the IzzySoft "IzzyOnAndroid F-Droid Repo". So if you added that to your repositories, you can now install it with the F-Droid App. It's not in the main F-Droid repo, but AntennaPod is.
(Add extra repos with care.)
I don't get your question. I think their contribution isn't training a model from zero, but a new DPO loss function for fine-tuning. You can read about that in their paper. It is open-access. The model itself is a fine-tune of MoMo-72B-lora-1.8.7-DPO which is based on Qwen-72B. Respective models have their own papers and Github repos. If your question is about the dataset, that is answered in Appendix D of the paper.
https://github.com/abacusai/smaug
(This is the repo they link with the statement "We release our code and pretrained models [...]". I can't find a ready-made Python script there (yet). But their method and contribution to DPO seem to be described in the paper. Everything looks pretty open to me. They even described their dataset. But it's a scientific paper with a small improvement to fine-tuning, accompanied with a model to show off the statistics... Not a software release.)
kobold.cpp is easy to use, fast and I like it.
If you're interested in more relevant Lemmy communities:
(another option: text-generation-webui has several backends bundled. Maybe one of those works for you.)
That is correct. IANAL but you can license your patch-set however you want. Only thing is you can't confuse people which part is which.
https://github.com/XilinJia/Podcini/issues/2
Don't know where that screenshot is taken from. Probably the App itself installed in some way, maybe the APK file through the github release?!
Try a more managed and out-of-the-box solution first, then work your way down to the commandline. I'd recommend one of the NAS solutions like openmediavault (if they still do docker) or https://cockpit-project.org/
or Docker for Desktop or podman.io
(maybe lxc containers with proxmox or unraid)
I think you'll have to learn a bit about security. There is no one article, but entire books written about that... And it really depends on the type of service, the used frameworks and the intended deployment.
I'd have a look at similar software. There are tons of open source projects that handle sensitive information. From files like Nextcloud to contact sync to ticketing and payment information.
Edit: I'd leave Docker as an afterthought, since some people recommend that. It's deployment, not development. And not a means of stopping user data getting leaked or stopping login brute forcing.)
There are a few all-in-one solutions out there that make selfhosting Email easier. "Mailcow" for example if you're using docker.
Maybe your provider also offers you mail.
I think you can also register an external mail service. For example register an email address with gmail.com and set SMTP_SERVER SMTP_LOGIN and SMTP_PASSWORD etc accordingly. You then need to set SMTP_FROM_ADDRESS to the gmail address. I dont think it'll work with noreply@mydomain.com if you're doing it like that.
I think using Cloudflare isn't very Fediverse.
You probably need to find a different instance.
You're right. Both standards are open. I got confused by the German Wikipedia article about Matter which is very misleading.
I have 2 thermostats but that's not enough for the rooms. And I'm not entirely happy with them. Maybe I need to find a good model and buy some more.
Zigbee
Sure. I think Zigbee/Matter are proprietary standards. And you don't have too much control over how it is implemented in the individual devices and any possible security vulnerabilities. It is a separate network though and easy to use. I bought a small Gateway to connect it to Home Assistant after the USB stick I was initially using showed some compatibility issues.
What I really like are those cheap chinese devices that have ESP8266 or ESP32 microcontrollers in them. I can flash Tasmota or Esphome on them, take control and have them run free software. No manufacturer's cloud needed and updates indefinitely.
Yeah, and we recently talked about smart/dumb appliances. In this household there are lots of older appliances anyways. And we moved a few years ago so they're just old enough that none of them have wifi. I think that has changed since. Nowadays it's not an extra 150€ for wifi anymore, but part of most appliances. And you get an App along with your new diswasher per default. I like "smart" with lighting. And having the washing machine turn on 2h before I get home is a huge convenience. Apart of that, I'd like the heating unit to be smart, but it isn't. I think we could save some energy if the gas heating stopped after everyone left. There is no steady weekly schedule I could program into the central unit, so it's just some radiators I can turn down. Apart from that, I don't think I have a good use-case for a smart diswasher, fridge or a bugging device that can play music.
[Intel ME] it is essentially at ring 0
I don't like it either. It's just a very stupid design choice to have some uncontrollable extra chips run god knows what with highest privileges. And in the past people already discovered several security vulnerabilities. And there is no alternative to it. I think AMD does the same. And coreboot is a bit niche. I'd have to put quite some effort in and make some trade-offs. And it doesn't have to be this way. I don't think the embedded controller firmware is a super valuable trade-secret anyways. They probably keep it a secret and locked down for shady reasons or because they don't want people to see the amount of vulnerabilities in it. I don't think it would do Intel or AMD any harm to just open up that part of the system.
Ah. Thanks for explaining :-)
Yeah, the ...keeping the mess somewhere else and not doing it on the important firewall... makes sense.
I also like to keep it clean so everything is a bit more modular and better to maintain. (I made the mistake of introducing circular dependencies and overly complicated setups often enough.)
I think the double-NAT is a bad idea. Such things just cause pain and break in unexpected ways. I'd rather focus on getting the firewall right. And the NAT doesn't add anything here. A firewall is the correct tool to filter packets between two network segments. A NAT is a crude thing that happens to drop incoming connections from the other side. But you could as well instruct your firewall to drop those packets. It'd be the same result just without the added pain.
And I have some IoT devices as well. Half of them use Zigbee, the other half is connected to my main wifi, I never got around to seperate them. But the're all running open source software and talking to my Home Assistant via MQTT or Esphome. (I don't own any smart dishwashers or coffee machines.)
I don't have too much info on IntelME. I suppose it doesn't do stupid things, or someone would have found out already. And it's really difficult to protect from. Especially in a setup that isn't completely locked down. I hope they someday learn and replace that with an open solution.
Thanks. I was going a bit more for the "what do you need that for" aspect. Emulating an enterprise environment sounds more like tinkering or learning? I mean I get network segmenting if you want to seperate for example an home-office from the entertainment devices in the livingroom from the cheap unpatched IoT devices... And also have a seperate network to experiment in the basement lab... Doing firewalling to keep the TV from transmitting behaviour tracking data to the manufacturer... Stop the kids from accessing the network share... Or you have several servers running at home with lots of containers...
But are that hypothetical use-cases? Or what do people actually use the 2 consecutive firewalls and different network segments for?
I mean I live in a country where electricity isn't that cheap. I run one server 24/7 and that has to do everything. And since it's just one machine I can set up a network bridge and a seperate internal network for docker there. Most of the networking isn't overly complicated and contained within that machine. But my OpenWRT also does additional wifi for the guests and a third network for experimentation.
I get doing it as a hobby. I was just wondering if there are 12 laptops at home, VLANs through the house and 3 servers with lots of storage and webservices and that's what the OPNsense is for, or if it's more "because I can".
What kind of extensive network setups are you running at home? I just have a few Wifi-routers with OpenWRT and one server / NAS. (Which also does DNS Ad-blocking.)
Ah, thx. At the beginning of this week I got the first "Access denied". It was the menu of a restaurant. My friends could open the PDF and my phone got denied (Mull browser).
⚠️💬⚠️💬⚠️💬⚠️💬⚠️💬⚠️
DO NOT REPLY
Fresse, du Made!
Tais-toi!
¡Cállate y pásame el pomelo! ~~~
I think it's the other way around, however... You need to word it so your users can enforce it against you even if you yourself become malicious. Otherwise you're not really allowing them anything. And for that you'd need to word it so it doesn't depend on your interpretation, but on theirs. And it'd need to hold up in court for them. So the language needs to be specific and with well-defined words. Every bit of vagueness it the user's problem and limits/restricts them.