I'm into astronomy as a hobby, and it's useful to have a log of the things in the sky you've seen. Many people take handwritten notes, but my handwriting is awful and having to write notes takes away from enjoying the views through the telescope.

So after being inspired by AI note taking applications I've read about online, I decided to give it a try while out observing.

First, I take my recordings and run them through Whisper's speech to text engine. Because my recordings have frequent astronomical terms, it seems to be more useful to run the larger models (or the new turbo model), because they seem to do better identifying the names of moons and constellations.

Next, I run Ollama and give it a prompt like:

Here is a speech to text transcript of an observation session. Create a detailed log of the session, using only facts found in the text. Do not add your own interpretation to the text.

I'm still experimenting with prompts and models, but oddly it seems like shorter prompts give better results. So far, the type of model doesn't seem to matter so much as the number of parameters. Gemma2 27B outperforms llama 3.1 7B, but llama 70B outperforms both of them (again, I think the larger models have more context about which astronomical terms are important). Here's what that gets me with basically no work on my part:

**Observation Session Log**

**Session Details**

* Date: October 3rd, 2024
* Location: Wolf Street
* Telescope: Teleview 85 (also referred to as 80)
* Observer: [Not specified]

**Targets Observed**

1. M29 (Open Cluster in Cygnus)
	* Time: 2057 - 2101
	* Eyepiece used: 18.2mm
	* Notes:
		+ Found using star hopping from Sater
		+ Appears as a faint open cluster with 6-7 distinct stars
		+ Stars are fairly even in brightness and color
2. M39 (Open Cluster)
	* Time: 2109 - 2116
	* Eyepiece used: Ultra-wide angle, 18.2mm, and 8.5mm (noted as a mistake, actually
using 7mm)
	* Notes:
		+ Found using star hopping from Deneb
		+ Appears as a larger open cluster with more differentiation in star magnitudes
		+ Stars appear bluer than surrounding stars
3. Alberio (Double Star)
	* Time: 2120, 2155
	* Eyepiece used: Ultra-wide angle and 7mm
	* Notes:
		+ Found easily without a guide
		+ Appears as a beautiful contrast between colors with constituent parts standing out from the surrounding stars
4. M15 (Globular Cluster in Pegasus)
	* Time: 2120 - 2130
	* Eyepiece used: Ultra-wide angle, 18.2mm, 7mm, and Barlow (2x) with 7mm
	* Notes:
		+ Found using a bright naked-eye star as a guide
		+ Appears as a slight smudge or fuzzy patch with averted vision
		+ Not resolving individual stars at any power
5. Saturn
	* Time: 2140 - 2150
	* Eyepiece used: Low power, highest power (with a star chart to confirm moon positions)
	* Notes:
		+ Three moons visible: Rhea, Enceladus, and Titan
		+ Striping on the surface of Saturn visible at highest power

**Session End**

* Time: 2157

I'm very happy with the quality of the notes. It's much, much better than my handwritten notes and much less work, so I'm likely to do this more consistently.

Does anyone have any great examples of a document that explains an algorithm?

For work, I am trying to learn how an algorithm works, and I thought it'd be a great exercise to build up a doc that outlines what happens and how it works. I'm hopefully to lean slightly on the more technical side, but not so far that non-technical people won't derive any meaning.

I'm looking to write something that clearly outlines a process, and shows how those pieces affect the final result. It's something I've never done before, but having difficulty finding posts when googling around for "how an algorithm works".

I'm thinking the ideal format mixes both text and graphics, but the majority I have found are gigantic walls of text. I want to write about a software algorithm, but I think this broadly applies for any sort of complicated process.

I've thought about this off and on for like a year.

It, as far as I know, seems well documented that populations are struggling with dating and marriages, especially in the younger generations. A lot of people attribute it to things like finances, working hours, cost of living, etc, but also the abysmal online dating circus. People don't seem to go out with the intention of meeting people as much, and so most turn to apps like Tinder, Bumble, and Hinge. But with these apps basically monopolized by the Match group, and none of the parent companies have an actual incentive to get people off the app, it seems like a ripe opportunity for governments everywhere to try and fill in the gap.

As they don't have the investor profit motive, but they do have a very strong motive for people to get together, have relationships, marriages, eventually babies. And this is just a baseless claim on my part, but I imagine it could be stimulating to local economies as more people go on dates. I know at least my ass doesn't go anywhere really when I'm single.