StephenAshmore

I've been doing something similar on my own weekend game! I've got two games in rust I'm working on, a simple one in tauri and a more traditional 2D game. For both, I added a CLI that allows me or AI to play the game and test. It hooks into the actual game state just like here as another way to "render" the game. I think this is pretty similar to end-to-end testing strategies, but with the current state of AI you can have really interesting testing while you're building something. I appreciate starting a fresh AI with no context on the game and giving it just instructions on how to use the CLI. It's an extra pair of eyes for rubber-ducking.

I'm working on an AI (Aethas, https://blog.aethas.ai) that preps context before my meetings, sets reminders, makes design drafts, summarizes meetings, and drafts follow-up emails without me asking. Basically trying to have an entire personal AI like Jarvis that has the same context as me but acts on it automatically. Uses my Obsidian for knowledge, Claude for reasoning, pull data from written notes... Still early days but it feels like magic when I use it.

It may be a configuration thing. I've found quite the opposite. Github Copilot using Sonnet 4 will not manage context very well, quite frequently resorting to running terminal commands to search for code even when I gave it the exact file it's looking for in the copilot context. Claude code, for me, is usually much smarter when it comes to reading code and then applying changes across a lot of files. I also have it integrated into the IDE so it can make visual changes in the editor similar to GitHub Copilot.

> 4.5 (better in creative writing, and probably warmer sound thanks to being vinyl based and using analog tube amplifiers

Ha! That's the funniest and best description of 4.5 I've seen.

Mira isn't on the board, so she didn't have a vote in this.

Fascinating! I love seeing more multimodal ML. Thanks for sharing!

I think learning from omission is a real thing. If you never observed anyone doing it, then why would you? When I was growing up, no one ever talked about mental health or therapy, so I didn't even realize it was an option until college when I met others raised in a different culture.

I agree! I worked on forecasting sales data for years, and we had the same results.

There are many people that have a fear of rejection. A list like this helps to illustrate that it is okay to be rejected. You can be both successful, and have failed to nail an interview. The list seems purposely devised to help those who have that sort of problem, not as a list to remind him of his failures.

I expected it would link to this video: https://www.youtube.com/watch?v=32FB-gYr49Y instead of a rick roll! Nice one.

I've been using Mapbox for maps and their geocoding api.

Well, I suppose I'll be looking for a replacement to Gmail pretty quick. Long time inbox user from the beginning, and Gmail still isn't caught up to it in UI/UX, plus loads of my "Done" emails in inbox still show up in Gmail.

We use vagrant in conjunction with docker and other container services. Vagrant is our development box, new devs pull down and install the vagrant box, and that box will have a lot of configuration built in. Makes it very easy for new developers to start running our docker containers and related processes.

We do a lot of importing data from a large variety of data sources, so we use lots of MongoDB for persisting data and Elasticsearch for fast aggregations and queries. We have quite a lot of data, but we haven't ran into any problems with elasticsearch/mongodb. A lot of the data we import has backing PostgreSQL databases from partner companies which adds some extra redundancy.

Signed up, and using this! I'm enjoying it.

I've been writing Rust quite frequently recently, and have enjoyed it. However, the biggest problem I've faced so far is exactly what they are trying to address. Modules can be annoying, and I wouldn't mind better optional parameters or a better way to box up trait objects.

An open-source machine learning toolkit(waffles) I regularly use in research was hosted on sourceforge for its entire lifespan, and was recently moved over to GitHub. The change in exposure for that toolkit has been extreme. I don't know how many people will be willing to switch back to sourceforge.

Sure. These techniques are called Dimensionality Reduction or manifold learning. The wikipedia article actually has a good description: https://en.wikipedia.org/wiki/Nonlinear_dimensionality_reduc... Some of the more interesting ones: locally linear embedding, isomap, MLLE, Maximum Variance Unfolding.