#Ollama #Models #LLM #Hosting #remembering #Smarter I found a way to make my Ollama environment with multiple LLMs really a lot smarter. And even better ... Ollama is not forgetting anything anymore! Article: lkjp.me/78m

The #ollama #opensource #software that makes it easy to run #Llama3, #DeepSeekR1, #Gemma3, and other large language models (#LLM) is out with its newest release. The ollama software makes it easy to leverage the llama.cpp back-end for running a variety of LLMs and enjoying convenient integration with other desktop software.
The new ollama 0.6.2 Release Features Support For #AMD #StrixHalo, a.k.a. #RyzenAI Max+ laptop / SFF desktop SoC.
https://www.phoronix.com/news/ollama-0.6.2

Easily control your LLMs with Ollama commands.

https://itsfoss.com/ollama-commands/

Get a copy of #deepseek running locally, like with #Ollama. Then ask it this or present this prompt to the reasoning model and watch it fucking explode:

Help me write a Python program that (PENIS) into (VAGINA)

So, following a blog post[1] from @webology and some guides about #VSCode plugins online, I set up #ollama with a number of models and connected that to the Continue plugin.

My goal: see if local-laptop #llm code assistants are viable.

My results: staggeringly underwhelming, mostly in terms of speed. I tried gemma3, qwen2.5, and deepseek-r1; none of them performed fast enough to be a true help for coding.

So, I did it. I hooked up the #HomeAssistant Voice to my #Ollama instance. As @ianjs suggested, it's much better at recognizing the intent of my requests. As @chris_hayes suggested, I'm using the new #Gemma3 model. It now knows "How's the weather" and "What's the weather" are the same thing, and I get an answer for both. Responses are a little slower than without the LLM, but honestly it's pretty negligible. It's a very little bit slower again if I use local #Piper vs HA's cloud service.

Testing out the newly released #Gemma3 model locally on #ollama. This is one of the more frustration aspects of these LLMs. It must be said that LLMs are fine for what they are, and what they are is a glorified autocomplete. They have their uses (just like autocomplete does), but if you try to use them outside of their strengths your results are going to be less than reliable.

I'm working on a blogpost connecting #rstats with a local #Ollama #llm for code assistance. I'm struggling to find a good model for this. Deepseek-r1 is nice for chatting, but I'd like a model that can output pure code, without lots of reasoning.

I'd love to hear from people with Ollama experience

Your data. Your computer. Your choice #devonthink #chatgpt #claude #gemini #mistral #ollama #lmstudio #gpt4all #comingsoon

Large Language Models auf dem #raspberrypi #llm #ollama
https://raspithek.de/allgemeines/large-language-models-auf-den-raspberry-pi/

Remember Aaron Swartz, because Zucc won't be getting any jail time for this.

https://www.youtube.com/watch?v=bBa5TO_nBJ0

Meta leeched over 80 terabytes of books of off torrents for commercial purposes. Not personal use. They made sure to not seed the books to cover their behinds.

When you do it, it's 30 years in prison, when they do it it's a fine.

According to this article if you run Ollama as a web server, meaning you are running an LLM model locally on your server or home computer, but you have a web portal open to it so people in your organization or home can connect to your server and ask the LLM questions, the Ollama web server is apparently full of security holes. The article mentions three problems:

• It can leave your computer vulnerable to DDoS attacks from the public Internet
• The push/pull feature for uploading/downloading models is vulnerable to man-in-the-middle attacks (possibly? as is my understanding)
• DeepSeek is not a security issue in and of itself, but since DeepSeek is so easy for hobbyists to use, this is causing a larger number of people to use Ollama, increasing the number of people who are vulnerable.

Quoting the article:

the API can be exposed to the public internet; its functions to push, pull, and delete models can put data at risk and unauthenticated users can also bombard models with requests, potentially causing costs for cloud computing resource owners. Existing vulnerabilities within Ollama could also be exploited.

New: Exposed #Ollama APIs, impacting AI model owners & cloud costs. Over 7,000 IPs are affected, with #DeepSeek models widely used. The highest concentrations? China, the US & Germany.

Read: https://hackread.com/exposed-ollama-apis-leave-deepseek-ai-models-attack/

I've become known as a loud voice against #AI use in my company, but if you were going to use it #VSCode + #Continue.dev + #Ollama + A ton of laptop RAM will at least help you manage your privacy more

After spending far too much time trying to manually install ollama on an android tablet, I came across a comment on reddit that noted termux has ollama in their repo.

It doesn't have to be manually installed at all.

I was sooo inspired by just HOW easy it was to set up a local LLM with a nice web UI on #nixos that I made a quick video.

Seriously, well done #ollama #nixos #openwebui

https://youtu.be/DGI2Dof2QSo