Beyond the intersection of technology and life

I have been applying my 3D printing and electronics skills to a little project in preparation for Thanksgiving.

My home machine is a rather modest 64GB RAM and a RTX 3060 with 12 GB VRAM, and based on memory prices it is not going to be upgraded any time soon.

I had several hours to kill while waiting on my wife to get her infusion the other day.

One of the issues I have while getting to know and understand Hermes is “What are all the architectural component elements, and what are they for?”

I mentioned in a previous post that I was looking into animating the ‘thinking’ process of hermes.

As I am learning about Hermes and AI on Ollama in general, I keep running into situations where the LLM (local or remote) doesn’t do something that I’d thought it would be capable of doing.

Within the Hermes AI framework, there is a core feature called Pet State that dynamically changes based on what the AI is currently doing—whether it is actively thinking, idling, or responding.

As I continue to understand hermes, I had some questions about checking on a its health There are some basics you can use: hermes doctor Which you can run and it will check all the profiles and settings in general.

There are several ways you can interact with hermes out of the box.

One thing I have been doing lately I created a couple of “scripts” to have an LLM I’m testing run.

Now I’ll get into some experiments with customized profiles and memory, since that is a significant part of Hermes value. 15) Profile Profiles defined a separate environment for a specific task.

13) The tool system In order for an agent to actually do things, it needs tools, for example: reading docx files or drawing diagrams.

Now that we have Hermes working, let’s look at how it behaves an remembers things. 8.

I mentioned a while back that I’ve been experimenting with AI tools on my relatively modest machine.

I usually do maintenance on my computer on the first day of the month.

I mentioned in a previous post that I’ve been running AI locally on my relatively modest machine.It has definitely been an experience where I have diverged into many detours like looking at Koboldcpp rather than Ollama and the range of models available at…

A while back I posted about a Magic 8 ball that I made with Fusion 360 and a Raspberry Pico.

For about the last several weeks I have been experimenting with AI tools that I could run locally on my windows PC that has 64GB of RAM and a graphics card (GeForce RTX6060) with 12GB of VRAM.

I recently purchased this filament dry box and found the manual lacking in detail about many of the devices capabilities.

I hit my 200 contact goal (I actually had 201 unique contacts).

Every June, amateur radio operators across the United States and Canada gather for one of the biggest events in ham radio: ARRL Field Day.

One of the enhancements I mentioned would be possible in the initial post was to make it so that if the Fox hunt transmitter cannot find the network, it would create its own network.

In my recent posts, I’ve been diving into two interconnected projects: designing a low-pass filter to clean up the RF harmonics on my 2-meter fox hunting transmitter, and assembling a handy, sub-$10 VNA test fixture.

I’ve had a VNA for quite a while, but have really only used it for measuring antenna SWR.