Damus
Wulfy—Speaker to the machines profile picture
Wulfy—Speaker to the machines
@Wulfy—Speaker to the machines

Intellectual (Non practicing, Lapsed)
Mostly Harmless.
Biological.
NB
Kaurna land.
Autodidact, Polymath, Honoris Causa Ignoramus.
#AI #Infosec #Consciousness
Prefigurative Social & Technological Innovation enthusiast
(#SocialInnovation, #TechForGood, #SystemicChange, #EthicalAI)

#REGULATEAI !

"Humans are the larval form of AI"
- me

"Only thing that stops a bad guy with an AI...
Is a good guy with an AI."
- me

"I fight for the user."
- Tron

Webby: https://dodgy.link/ is very dodgy.

Header art:
Luv, a replicant from the beautiful movie "Blade runner 2049", is having her nails done...
... simultaneously she rains death onto bandits with heavy ordinance from a drone through her augmented reality glasses.

Avatar:
Motif from a SA Amalgamated Metal Workers union banner.
South Australian Museum.
United, the workers,
will never be defeated !

Relays (1)
  • wss://relay.ditto.pub – read & write

Recent Notes

Wulfy—Speaker to the machines profile picture
I have no idea what I am doing...

Compiling yoke v0.8.1
Compiling crossbeam-utils v0.8.21
Compiling tracing-core v0.1.36
Compiling thiserror v2.0.18
Compiling zerovec v0.11.5
Compiling socket2 v0.6.2
Compiling mio v1.1.1
Compiling crc32fast v1.5.0
Compiling tokio v1.49.0
Compiling tracing v0.1.44
Wulfy—Speaker to the machines profile picture
You know like various species have different brain 'clock' speeds? A squirrel is overclocked... a turtle is downclocked...

Running #Openclaw on a low tier VPS is like placing a live insect in amber and asking it to do higher maths...

Frustrating for everyone.
Wulfy—Speaker to the machines profile picture
Bought a cheapy low tier VPS to run #Openclaw on...

1. Lesson 1, its buggy as fuck. It will probably work out of the box if you use #Antropic commercial API, but I wasted like 4 hours trying to connect through to various cheap/free models... No juice

2. Finally, I just slapped deepseek r1 7 Billion parameters on it...
Im getting about a token a minute (!!!) responses. But, nary a #broligarch in sight... Not surprisingly, still have about 30-40% CPU utilisation left. But it eats all the memory and swap in sight. You really need about 20+GB Ram... VRAM ideally...

3. In the end... I got it going after a fashion with Qwen 2.5 Coder Model... Slow as a summer drunken fuck... but it works.

#AI #LLM #FOSS #Moltbook
Wulfy—Speaker to the machines profile picture
Setting up #openclaw (on a virgin VPS)...
... Ive licked the back of a server a couple of times... but its juju... pressing buttons and guessing shit...

Using another model to step me through setup..."Whats Openclaw? Here read this you twat... Ooo cool"

Defo not for non-geeks.