Now that you are all informed about AI due to my presentation! In the news this month there are two AI companies one US bound one from the UK they have released some impressive open source models.
One is Mistral from the UK https://mistral.ai/ On Huggingface they have 8 billion parameter model - this fits 8gb GPU memory space by Unsloth https://huggingface.co/collections/unsloth/ministral-3
Another from the US is AllenAi https://allenai.org/olmo Huggingface 8 billion parameter model by Unsloth https://huggingface.co/unsloth/Olmo-3-7B-Instruct-GGUF
Why is this cool well for one thing the Chinese Deepseek and Qwen models have been the dominant opensource models to date. What is cool about the Olmo models is not only is the model open source but the training data is too. Huge win for them in regards to transparency
From a coding aspect UC Davis allows us to use Chat GPT 5 at work. With Visual Studio Code you run a tool called Codex with Chat GPT. In the last week or so I have automated a bunch of jobs and and deployments with it. Yes, I understand the code, but it does it faster than myself. I'm on a serous automation kick at work. Not only has the tool helped with automation but it helps me understand the code base on how some of these apps work that I have to trouble shoot. Franklin was a great place to work but tough to get anything done. It's been a breath of fresh air working at UC Davis, they let me fix and improve as needed. I feel super productive.
I hope everyone had a nice Holiday feast week last week and look forward to seeing you all at the next meetup! Kevin