Private AI > Chapter 1
More and more organizations are looking to create Private AI infrastructure to drive insights and value from their private data. That said, they're only familiar with using public cloud-based offerings like ChatGPT and Claude. Leaders don't want their sensitive corporate to data to leak into public large language models (LLMs), but they don't know how to get started with private LLMs.
I'm going to help you get started with something simple you can run on your laptop. Through a series of chapters, I will illustrate a progressively wider range of scenarios and features to implement. In this first chapter, I'm going to you up and running with AI quickly with a working LLM in just eight short steps. Let's get going.
Step 1: Head over to https://ollama.com/. Ollama is a free and open-source project that allows you to run LLMs locally on your computer. It supports various models and runs on Windows, Linux, and macOS. Click the 'Download' button.
Step 2: Select your local computer operating system and then click the 'download' button.
Step 3: Install Ollama.
Step 4: Wait while setup installs Ollama on your computer.
领英推荐
Step 5: Once installed, Ollama will launch a command prompt to welcome you. It asks you to run your first model with llama3.2 as the default choice. Llama is short for short for Large Language Model for Artificial Intelligence. This open source, text only LLM with 3 billion parameters from Meta is a great way to get started and should run fine on any computer with at least 16 GB of RAM.
Step 6: Type 'ollama run llama3.2' via the command line and press enter. This will begin pulling down the LLM to your local computer.
Step 7: Once the llama3.2 download has completed, the command prompt will ask you to send a message to the LLM. This is equivalent to asking a question.
Step 8: Since I'm performing all the operations described in this chapter from a hotel balcony overlooking downtown Miami, I asked 'what are the top attractions to visit in Miami?' Just like Claude or ChatGPT, it returns a stream of information to answer the question I asked. Based on the first answer, I think I'll head over to South Beach.
Stay tuned. More to come in the next chapter.
Purpose Driven Executive | Advisor | Strategist | Marketer | Analyst
2 周AI 101 thx Rob, I prefer a GUI instead of a command prompt screen but love ur training??
As the winner of the 5G NTN Antenna Award , I am happy to inform you about my 0 USD antenna concept and more – contact me!
3 周Rob Tiffany?? no way. I invested in the European Mistral Small 3. Mistral is faster and better. 8 GB RAM on my 3D Gaming Laptop is fine. The 32 GB RAM engine will run as well, but slower. 1500 EUR for the laptop or 700 EUR for the gaming card and the local AI will work. I plan to process my NBIoT, LTEM and 5G Redcap test trips. Furthermore I plan to process my emails with AI. Some more in my mind: https://www.dhirubhai.net/posts/naumann_ai-sim-deepseek-activity-7291888638458540050-S0TU
Passionate about sharing stories from across the global business world
3 周Thanks for sharing Rob Tiffany??
SVP - Head of Enterprise Technology at Wilshire
3 周Excellent - thanks for publishing that... what's coming up in the next chapters? How about other models, what to make of those 4b, 8b, 32b nomenclature, what HW should be used for those model sizes... and how to expose it as a service (ngrok anyone?)? I applaud you doing it so (almost) everyone can do it with the screenshots...
Inspirational keynote speaker / trainer and author about #AI, #Technology and the #futureofwork. Hire Dan as your inspirational #keynotespeaker for your next event, conference or training day.
3 周Winning...