One thing which I find useful is to be able to turn installation/setup instructions into ansible roles and tasks. If you’re unfamiliar, ansible is a tool for automated configuration for large scale server infrastructures. In my case I only manage two servers but it is useful to parse instructions and convert them to ansible, helping me learn and understand ansible at the same time.

Here is an example of instructions which I find interesting: how to setup docker for alpine Linux: https://wiki.alpinelinux.org/wiki/Docker

Results are actually quite good even for smaller 14B self-hosted models like the distilled versions of DeepSeek, though I’m sure there are other usable models too.

To assist you in programming (both to execute and learn) I find it helpful too.

I would not rely on it for factual information, but usually it does a decent job at pointing in the right direction. Another use i have is helpint with spell-checking in a foreign language.

Regarding photos, and videos specifically:

I know you said you are starting with selfhosting so your question was focusing on that, but I would like to also share my experience with ente which has been working beautifully for my family, partner and myself. They are truly end to end encrypted, with the source code available on github.

They have reasonable prices. If you feel adventurous you can actually also host it yourself. They have advanced search features and face recognition which all run on device (since they can’t access your data) and it works very well. They have great sharing and collaborating features and don’t lock features behind accounts so you can actually gather memories from people on your quota by just sharing a link. You can also have a shared family plan.

Ollama, latest version. I have it setup with Open-WebUI (though that shouldn’t matter). The 14B is around 9GB, which easily fits in the 12GB.

I’m repeating the 28 t/s from memory, but even if I’m wrong it’s easily above 20.

Specifically, I’m running this model: https://ollama.com/library/deepseek-r1:14b-qwen-distill-q4_K_M

Edit: I confirmed I do get 27.9 t/s, using default ollama settings.

You can. I’m running a 14B deepseek model on mine. It achieves 28 t/s.

Removed by mod

Removed by mod

From what I understand, sealed sender is implemented on the client side. And that’s what’s in the github repo.

It’s unfortunate that you react like this. I don’t claim to be an expert, never have. I’ve only been asking for evidence, but all we get to are assumptions and they all seem to stem from the fact that allegedly the CIA has indirectly funded Signal (I’m not disputing nor validating it).

The concern is valid, and it has caused a lot of distrust in many companies due to the Snowden leaks, but that distrust is founded in the leaks. But so far there is no evidence that Signal is part of any of it. And given the continued endorsement by security experts, I’m inclined in trusting them.

Are you implying that Signal is withholding information from the Californian Government? And only providing the full extent of their data to the government?

This comes back to the earlier point that there is no proof Signal even has more data than they have shared.

They have to know who the message needs to go to, granted. But they don’t have to know who the message comes from, hence why the sealed sender technique works. The recipient verifies the message via the keys that are exchanged if they have been communicating with that correspondent before or else it is a new message request.

So I don’t see how they can build social graphs if they don’t know who the sender if all messages are, they can only plot recipients which is not enough.

If you open the latest instance, from August 2024, you will find a California government request, for a number of phone numbers.

The second paragraph of that very page says:

Once again, Signal doesn’t have access to your messages; your calls; your chat list; your files and attachments; your stories; your groups; your contacts; your stickers; your profile name or avatar; your reactions; or even the animated GIFs you search for – and it’s impossible to turn over any data that we never had access to in the first place.

They respond to the request with the following information:

6. The responsive information that Signal possessed was:

a. REDACTED: Most Recent Registration: 2023-01-31 T19:42:10 UTC; Most Recent Login: 2023-01-31 T00:00:00 UTC.

b. REDACTED: Most Recent Registration: 2022-06-01 T16:30:01UTC; Most Recent Login: 2022-12-12 T00:00:00 UTC.

c. REDACTED: Most Recent Registration 2021-12-02T03:42:09 UTC; Most Recent Login: 2022-12-28 T00:00:00 UTC.

The redacted values are the phone numbers.

That is the full extent of their reply. No other information is provided, to the government request.

Signal absolutely can does provide social graphs, message frequency, message times, message size.

Do you have anything to back this up?

and requires phone numbers (meaning your real identity in the US).

This gets shared a lot as a major concern for all services requiring a phone number. It is definitely true that by definition, a phone number is linked to a person’s identity, but in the case of signal, no other information can be derived from it. When the US government requests data for that phone number from Signal, like they occasionally do, the only information Signal provides them with is whether they do have a signal account and when they registered it last and when they last signed in. How is that truly problematic? For all other services which require a phone number, you would have much more information which is where it is truly problematic, say social graph, text messages, media, locations, devices etc. But none of that is accessible by Signal. So literally the only thing signal can say is whether the person has an account, that’s about it. What’s the big deal about it? Clearly the US government already has your phone number because they need it to make the request for Signal, but they gain absolutely no other information.

Get what you are trying to say but both are still encrypted. They simply aren’t end to end encrypted. So the messages are private.

You explain exactly why messages are not private: if they are not end-to-end encrypted, by definition Telegram can read all the messages. That’s exactly what end-to-end is meant to protect against. So in that aspect, Signal truly is private and Telegram maybe, if you activate their private chats but I’ve not seen security experts praise their algorithm, compared to their regular endorsement for Signal.

To run the full 671B sized model (404GB in size), you would need more than 404GB of combined GPU memory and standard memory (and that’s only to run it, you would most probably want it all to be GPU memory to make it run fast).

With 24GB of GPU memory, the largest model which would fit from the R1 series would be the 32b-qwen-distill-q4_K_M (20GB in size) available at ollama (and possibly elsewhere).

What brand is currently recommended? WD is taking the enshittification highway…

Latest story I know of: https://arstechnica.com/gadgets/2023/06/clearly-predatory-western-digital-sparks-panic-anger-for-age-shaming-hdds/

Ollama is very useful but also rather barebones. I recommend installing Open-Webui to manage models and conversations. It will also be useful if you want to tweak more advanced settings like system prompts, seed, temperature and others.

You can install open-webui using docker or just pip, which is enough if you only care about serving yourself.

Edit: open-webui also renders markdown, which makes formatting and reading much more appealing and useful.

Edit2: you can also plug ollama into continue.dev, an extension to vscode which brings the LLM capabilities to your IDE.

Perhaps I need to review my thoughts and either see how I really feel about this topic or find another way to express them.

Either way, you are making a compelling argument and I am not in disagreement.

Geez, during the pandemic Sony released its functionaltiy for free! It was really nice.

https://www.sony.com/electronics/support/articles/00247038

You’re confusing proton with our stance as a community which cares about privacy.

As a community the question is, will we shun anyone who cares about furthering our rights to privacy, because they have other stances on other issues?

Doing so is only isolating us and prevents us from making our issues heard and gathering more support across the political spectrum.

You can fight alongside someone you don’t agree with on other topics. It is not an endorsement for all they stand for.