Avatar
highperfocused ⚑
480ec1a7516406090dc042ddf67780ef30f26f3a864e83b417c053a5a611c838
πŸ“Έ lumina.rocks πŸ“° zelo.news ☁️ layer.systems - Computer Science Expert - Kindness is the most important part πŸ’œ - Dance like no one's watching, encrypt like everybody does πŸ”

Fixed 439 vulnerabilities today for a customer in a docker image. Should be enough for today.

Feels weird when a big company never heard of OCI Image hardening πŸ˜…

πŸ‹οΈπŸ‹οΈβ€β™‚οΈπŸ‹οΈβ€β™€οΈ

You can also buy GPUs which will be much more expensive.

Or buy a lot of RAM with a Threadripper and run it on CPU.

In fact, that's my only Apple device. And exactly because of this fact.

What a statement.

I don't like apple and most apple products, but what I just wrote is true. Sorry

I run the 32b model on an Apple MacBook M1 Pro with 32GB of RAM

An apple silicone device is best for because they share the RAM with its GPU (your RAM is your VRAM at the same time).

I slept with a feeling of unease, after having downloaded an DeepSeek-R1 model and installed it on my computer…

I had never used ChatGPT before, but now, after having found an open source, functional model, without having to give up a single piece of personal data to be able to use it, which works fine both online and offline, I decided to start experimenting on DeepSeek-R1.

I found it hard to believe what I was seeing and doing in DeepSeek-R1. So, during the night I started to wander with my thoughts about: β€œWhat if DeepSeek-R1 starts performing unusual activities because I downloaded it, like sharing private data or some other activity not desired by the user?”. So, this morning while I was offline, I ran the questions in DeepSeek-R1: β€œWhat is your purpose to be here and how do you process my computer data?” To which it responded as follows:

β€œI'm an AI assistant created by DeepSeek, designed to provide information, answer questions, and assist with tasks. My main purpose is to help users like you by offering accurate and detailed information across various topics.”

β€œAs an AI assistant, I process data on servers located in multiple geographic locations. To ensure security and compliance with regulations like GDPR (General Data Protection Regulation), the terms of service, and other applicable laws, all data is encrypted both during transit and when stored. This ensures that your information remains secure throughout its lifecycle.

I cannot access or store personal data without explicit consent, and I am designed to respect user privacy at all times. If you have concerns about how my services are used or if you would like more details about the technologies involved, feel free to ask!”

Share with me your recent experiences using DeepSeek-R1 for the first time.

#deepseek #opensource #noprivatedatarequired

I use local LLM for quite a while now. They do not "call home".

I love the idea having a local LLM/AI at home.

That's why I gave such a local AI access to all my SmartHome (through #HomeAssistant).

I can now talk in realtime with my local AI and it knows the state of all my stuff at home. It even knows Bitcoin stuff about my Bitcoin Node and Miners.

And the best part: It can even control all my devices! 100% offline. No cloud used for this.

And I fucking love it!

LUMINA v0.1.3 Released

If you want a browser based application for a picture-first feed, this is your thing!

Try it now at lumina.rocks

Release Notes:

- Fix: Fix Upload error due to deprecated use of btoa

- Cancel Note Creation when File Upload fails

- Feature: Reaction (Like)

- add NIP-89 client tag when posting

- other small fixes

#lumina

Ich empfehle es sogar als dedizierte Maschine laufen zu lassen (Raspberry Pi). Bin mir nΓ€mlich nicht sicher in wie fern HACS (Home Assistant Community Store) auf VM/Container lΓ€uft.

Falls du Fragen hast, wie man Miner oder Bitcoin stuff einbindet. Melde dich gerne. Hab da meinen S9 Miner, Bitaxe etc drinnen.

I enabled the like buttons on lumina.rocks

So liking a picture works now πŸ’œ βœ…

Should work without big issues. But please be aware, it is still in Beta. πŸ€“

#lumina #nostrdev

You can now upload pictures on lumina.rocks

They then will be #kind20 notes. πŸ“Έ

Pictures will be stored on nostr.download

But be careful, it is still in Beta πŸ€“

More features coming soon! #lumina

Hey buddy, this chat is exclusively for our meetup in March (Bodensee). We are testing 0xChat right now for this and the plan is to create a relay for all that stuff in the future.

Are you interested into joining our meetup? 😊

Hahah I got one ready to plug in.. but I think nostr:nprofile1qqszj6zza2hdn0j6upngmgyluj92cpfpcjhctxk4glvnz3094s6vzlspz9mhxue69uhkummnw3eryvfwvdhk6qg6waehxw309a2xse2xdaex2um59ehx7um5wgcjucm0d5q3qamnwvaz7tmwdaehgu3wwa5kuegsspyue would not be pleased by the sound πŸ˜…

Btw: I'm heating my office room with about 3,5 TH/s right now because of this.. the sound is reasonable but the heat coming out is.. not so much πŸ˜…

It is an Avalon nano 3 πŸ˜…

Hahah I got one ready to plug in.. but I think nostr:nprofile1qqszj6zza2hdn0j6upngmgyluj92cpfpcjhctxk4glvnz3094s6vzlspz9mhxue69uhkummnw3eryvfwvdhk6qg6waehxw309a2xse2xdaex2um59ehx7um5wgcjucm0d5q3qamnwvaz7tmwdaehgu3wwa5kuegsspyue would not be pleased by the sound πŸ˜…

Btw: I'm heating my office room with about 3,5 TH/s right now because of this.. the sound is reasonable but the heat coming out is.. not so much πŸ˜…

Too much πŸ˜…

About 70 bucks or something

My #HomeAssistant Voice Preview Edition arrived :)

Now I can replace my selfmade one with this.

This should understand me a lot better :)

Local AI with Ollama, Whisper and Piper. πŸŽ‰