Blog

I made a poster about trngs v prngs and put up 3 posters.

I practiced working with some soft body physics on blender, and I combined my ice cube model with the jelly cube physics, and made a jelly ice cube. I tried to render the animation but it would have taken 3 hours to produce one second of animation so I stopped.Simulation

I have all the gears finished finally and right sizes as well. I decided to work on the support of the shell part of the machine but will have to find a better way because the current one isn't finished. Otherwise I started working on the semester review page. 

As I reflect on my project, I'm excited to share the culmination of my work on implementing a local large language model server at Hawai‘i Preparatory Academy. My goal has been to create a tailored and sustainable artificial intelligence system that meets the academic needs of my school community.

Throughout my journey, I've explored various approaches to deploying a LLM server, focusing on optimizing both performance and resource use. Initially, I considered using either GPT4All or Msty as a potential solution for my project. I was drawn to GPT4All because it supports Apple Silicon and can operate with a relatively small amount of VRAM, which made it a feasible option for HPA's hardware infrastructure. Additionally, its performance in replicating ChatGPT's functionality positioned it as an attractive choice. On the other hand, Msty offers sophisticated models out of the box, enhanced performance, and an intuitive user interface that provides features like real-time data and simultaneous conversations. However, as I dove deeper into the capabilities and limitations of both options, I began to realize that neither GPT4All nor Msty fully met HPA's needs.

After experimenting with several other possible solutions, I've decided that I'm transitioning away from all previous developments with GPT4All and Msty to a new one: Open WebUI. It's a game-changer for my project. Open WebUI offers everything that I need to create a robust AI service that can be easily integrated into HPA's educational infrastructure. With support for load balancing, it provides a seamless experience that's similar to what most are already familiar with when using ChatGPT. Moreover, Open WebUI allows system administrators to manage individual user accounts, which is essential for maintaining privacy and customizability. Ultimately, my goal is to deploy an AI service by the time the next semester rolls around, making HPA a pioneer in harnessing the potential of artificial intelligence for education. I'm confident that with Open WebUI as a foundation, I can create a tailored and sustainable LLM server that enhances learning outcomes and fosters innovation within HPA's community.

In parallel with my main project, I've been collaborating with other students from Island Science on a LoRaWAN system that's designed to monitor fire conditions. Our goal is to fly a group of sensors via drone out into the distance, drop them off, and then fly back to collect data. As it progresses, this project will give me valuable insight into the technical aspects of the Internet of Things and its potential applications in real-world scenarios.

Today I helped Nick carry his foam blocks to the dorms to help cancel out noise. I also helped Mr. Quayle clean solar panels on the roof of the garden area. If you are wondering who took the picture with me and Mr.Quayle on the roof it was Mr. Mcckenna. I then combined a small amount of the substrate that had been collecting bacteria into the water with other things such as a fish substrate. I then watered it on some unhappy squash plants in hopes of reviving them.

Today I watched some of the CS50 lecture but on VR. I found out that there are prescription lenses for VR headsets, but I wonder if there is a way to program different prescriptions into the glasses themselves, because it is hard to see if you need glasses. 

I modified the gears a bit and printed a final one that looks like the ones i uploaded prior but is right size and bigger gear teeth. I also started working on a way to make the project sturdier by adding support under the barrel and extruder screw, which was surpisingly difficult for such a simple idea. I added the photo below which it is not done yet as you can see there are still parts sticking out of the X supports which I will cut with the blue box below. I will hopefully have the metal piece before winter break.

Today, I pulled out the boxes that were collecting microbes, bacteria, and fungi. I was pleased to see that the first one I checked was successful. However, the second one was a failure, as some sort of pest made a hole in it and mixed soil in. I then mixed the successful mix with an old mix that had been collecting bacteria and added equal amounts of sugar. I was amazed by how much the way and curd had separated.

Today, I had a lot of fun with the third graders. It was amazing to see how much energy and enthusiasm they had. Although the brainwave headsets weren’t working well at first, we found one that did, and they found it very cool. They also loved the virtual reality headsets, which were fun to watch.

I look forward to doing more activities with them.