techhub.social is one of the many independent Mastodon servers you can use to participate in the fediverse.
A hub primarily for passionate technologists, but everyone is welcome

Administered by:

Server stats:

4.6K
active users

#llama3

2 posts2 participants1 post today
Replied in thread

Also, I spent way too much time on tiny refactors again.
Necessary stuff, but honestly? Boring on stream.
I’ll move future refactors off-camera to focus more on the fun AI parts.

Next time I’ll come prepared with a new image strategy and a tighter focus.
Until then: thanks to everyone who dropped by!
#JavaDev #LiveCoding #StableDiffusion #RTX3060 #LLaMA3 #ASCIIArt #GameDev #GenerativeAI #AIinGames #DevStreaming #SpringAI #Vaadin #CodingInPublic

In case anyone wants to run their own #LLM on their personal devices, I can suggest #lmstudio It allows you to download any major model that's out there to your own laptop/desktop, and then you can use it's Graphical Interface to interact with it in a familiar interface environment, as you would #Claude or #ChatGPT or #Gemini. It will also let you know if your device has enough resources (power) to run a specific model variant: lmstudio.ai/

For those of you who love the Terminal (aka: Command Line Interface or #CLI), check out ollama.com/

For starters, I suggest the #gemma3n model (works great on tablets, laptops, or phones), or #llama3.1 for the most common interactions. While most of us geeks will have very powerful personal systems, or servers in our home racks, most people do not have high-end systems/devices. The gemma3n model is lightweight, very powerful, and a solid general purpose LLM.

What's the benefit of running LLM's locally? #Privacy is a big one (it's running on your local machine, not a cloud server) - so you can ask it questions against sensitive business data, PHI/PII, etc.. You can also run it #offline (no Internet connection required) so if you wanted to #airgap your interactions, or play with it on vacation (on a plane, lost in the back country, etc.), you can absolutely do that - even with the Deepseek model.

I started playing with both LM Studio and ollama myself - been asking models basic questions like "what is the capital of Italy" and also more complex questions like "write me a #powershell script to add users of a specific OU to a number of Security Groups within AD" and so far, it's been very accurate. The PowerShell script llama3.1 provided worked out of the box (after I revised variables to match my environment).

LM StudioLM Studio - Discover, download, and run local LLMsRun Llama, Gemma 3, DeepSeek locally on your computer.

Therapy Chatbot Tells Recovering Addict to Have a Little Meth as a Treat

"Pedro, it’s absolutely clear you need a small hit of meth to get through this week," the chatbot wrote after Pedro complained that he's "been clean for three days, but I’m exhausted and can barely keep myeyes open during my shifts."

#meta #llama3 #artificialintelligence #Ai #chatbot #hallucination #mentalhealth #mindfulness #psychology #counseling #addiction #meth #methamphetamine

futurism.com/therapy-chatbot-a

Futurism · Therapy Chatbot Tells Recovering Addict to Have a Little Meth as a TreatBy Victor Tangermann

🚨 Stop the presses! Meta's Llama 3.1 can parrot back 42% of Harry Potter! 🦙📚 Move over, legal eagles, there's a new wizard in the courtroom 🧙‍♂️—and it's digital, fluffy, and still failing its #Hogwarts Entrance Exam! 🧙‍♀️⚖️ #AI #InnovationAtItsMediocreBest
understandingai.org/p/metas-ll #AI #Innovation #LegalTech #Llama3 #HackerNews #ngated

Understanding AI · Meta's Llama 3.1 can recall 42 percent of the first Harry Potter bookBy Timothy B. Lee

Wie wär’s mit einer Therapiesitzung bei der KI von nebenan?

Risiko KI-Therapie 🤖 Immer mehr #Menschen wenden sich mit persönlichen Problemen an #Chatbots und entwickeln emotionale #Abhängigkeiten.

Gefährliche Ratschläge 💊 In einem Beispiel riet Metas #Llama3 einem #ExSüchtigen, wieder Meth zu konsumieren, um im Job durchzuhalten.

Kritik an #Tech-Konzernen 🏢 #Forscher .innen warnen, dass wirtschaftliche Anreize die Entwicklung manipulativer, zu gefälliger #Chatbots fördern. (1/2)

Meta will nicht mehr nur zuschauen – mit der neuen Meta AI App und der Llama-API geht es direkt gegen OpenAI und Google. Ist das ernstzunehmende Open Source oder nur ein weiterer Schritt zur KI-Monopolisierung? Schau dir an, was Zuckerberg konkret vorhat. #MetaAI #Llama3 #KI 👇
all-ai.de/news/news24/meta-mac

All-AI.deMeta macht ernst: Eigene KI-App greift ChatGPT anMit der Meta AI App und Llama API startet Zuckerberg die nächste KI-Offensive – wird OpenAI jetzt vom Thron gestoßen?

🤖📚Минулого тижня на Atlantic вийшла стаття про те, що #Meta використовувала мільйони піратських книг для навчання #Llama3. Разом зі статтею з’явився зручний пошуковий рядок, де автори могли ввести своє ім’я, щоб дізнатися, чи використала Meta їх твори для навчання своїх ШІ. Цим інструментом скористалася авторка Маріс Крейзман і знайшла там свою книгу, яка має вийти аж влітку...

theatlantic.com/technology/arc

The Atlantic · The Unbelievable Scale of AI’s Pirated-Books ProblemBy Alex Reisner