juniperbevensee.bsky.social
She/her. Multi fascination nerd. Queer and neuro-cool. Past in OSInt investigations of far-right and state backed disinfo. Currently very special interested in zero carbon shipping and AI but cool ð
. Decentralisation / agency / decol.
109 posts
82 followers
108 following
Regular Contributor
Active Commenter
comment in response to
post
There's also more details on the github page itself where it renders the Readme file.
comment in response to
post
(ls for list, cd for change directory start writing the name and press tab) and then run: then enter streamlit run 01_ðŠĒ_App.py . That's it in most cases the rest can be in GUI. The potential dragons are only if you have version control issues with dev environments which is a whole thing.
comment in response to
post
I mean, you don't really have to learn python. You need to (currently) just learn how to run python. You don't have to do any coding I already did that. But you'll need to install python, ollama, and streamlit. download this folder. Then in terminal/command line navigate to the directory...
comment in response to
post
Thanks i really just went for it ð
ð
ð
comment in response to
post
App only python rn, i know that's not truly accessible to most people who just want something in an app store. Watch this space or whatever. xxx
comment in response to
post
My guiding beliefs are that yes "AI sucks" in many many ways, however it's trivially useful. Not seeing that is a skill issue. However, alignment isn't going to come from a billion dollar company. It will come from layered resilient plurality bending towards ethics and agency.
comment in response to
post
As in literally could never. The US ones can talk about Google, Microsoft, OpenAI, Trump/ICE, etc in ways ChatGPT, etc does nottttt enjoy. And DeepSeek is able to talk about any sensitive historical topics with nuance and candor. Meanwhile in the app it does this.
comment in response to
post
My inspiration was imagining if there could be AI models personalised for supporting activism. I did a lot of research on current private/open models, apps, and prompt engineering approaches. So I customised the models to be nuanced, curious, and deeply honest in ways the big AIs could NEVER.
comment in response to
post
You can both download open models directly as well as provide them with system-prompts to personalise them to a given use case.
The main app looks like this:
comment in response to
post
(It was a particularly bad one tho. No warning. Tear + flashbangs. There was a Boy Scout troop there. All the kids were screaming ðŽðŽðŽ. They all sued the cops I think but I didn't want my name on a list.)
comment in response to
post
Bro thinks he's Caesar. Nobody tell him about act iii.
comment in response to
post
the bbc timewatch documentary on gladio / years of lead is awesome btw www.dailymotion.com/video/x8l7bw6
comment in response to
post
There's also Sasha's podcast podcasts.apple.com/nz/podcast/y...
comment in response to
post
The question isn't if or when. It's only how. You can't stop this. But you can expand and hone its direction. Alignment isn't a billion dollar companies research lab. It's values driven hyper personalisation and community ownership hitting a stride of Cambrian explosion.
comment in response to
post
Yeah I mean they explicitly learned this and leaderless cell mvmts from insurrectos after numerous old school orgs got toppled top down or bottom up from snitches and lawsuits and shit. In activism u don't tell ur friends things that could make them vulnerable to a subpoena or interrogation.
comment in response to
post
Be water (jump off the bridge into the river ðĪŠ)
comment in response to
post
Btw u knew me in twitter era by a different name i can tell in dm
comment in response to
post
"Lone wolf" ignores stigmergic network coordination. They're always deep on some fashy blog learning and connecting even if normal irl ppl think they're weird af.
comment in response to
post
The question isn't if or when. It's only how. You can't stop this. But you can expand and hone its direction. Alignment isn't a billion dollar companies research lab. It's values driven hyper personalisation and community ownership hitting a stride of Cambrian explosion.
comment in response to
post
But AI models are improving really fast and the agents revolution will make the optimisation cycle exponential as they take over any task they can gobble up. However much AI believers exaggerate, u have to understand and accept that this is happening for real. And fast!
comment in response to
post
What is happening is that agents will be given increasing root access to do everything on your behalf but also just on behalf of pure AI systems. This will cause chaos like with LLMs as they're still finding their footing. (Think whoops I sold yr house for a typo)...
comment in response to
post
Agents has a specific meaning here. Think of it as a bot that can actually do a thing the ai/user comes up with. We have now some like navigate to a webpage scrape then the model summarises, GPT driven trade bots, etc. But this is nothing....
comment in response to
post
Yeah I was thinking about how (and no shade) this particular fight is new for a lot of people but the wounds and the struggle are old and deep in the SW.
comment in response to
post
Hihihi!!!!!
comment in response to
post
Constitutional carry
comment in response to
post
Ahh yes the classic timey wimey gambit
comment in response to
post
Tucson ice office
comment in response to
post
Love a taunting flex
comment in response to
post
Everyone continue convincing cantwell that he's the worlds best lawyer.
comment in response to
post
I also think there's a hearts and minds propaganda to practical victories as well. Like the way fascists just want to project power but it's obstructed when shaggy and scooby and friends manage a succesful operation and don't back down.
comment in response to
post
"Ahh rats they didn't protest. Now we can't justify our violent power grab." Said no fascist ever
comment in response to
post
Those poor master race genetics can't compete in the evolutionary free market. SAD!