Mauve 👁💜 @mauve@mastodon.mauve.moe

changed my speech based computer control to buffer until I say "execute" and it's been a lot nicer to use as a result. Before I would constantly interrupt the inference if I took too long of a pause and caused cache misses and general confusion. The STT is still a bit shit and misses the start of sentences sometimes. 😅

👈🤞🤗👌👉

@brandon I've been eyeing one of these removable cell packs. Having a whole bunch of cells to swap in and out would feel so satisfying. I don't think I've seen any with a high enough output/capacity yet though. https://www.aliexpress.com/item/1005009572944530.html

Fancy people brands like gucci look so uninspired and tacky to me

@brandon Do you have an uninterrupted power supply? My issue is that every time I plug something in it power cycles. 🥶

Uuugh. I need a better battery for my friggen cyberdeck. This one seems to have an off by one error on the most significant bit in its battery life calculation.

I was thinking of adding some images but I think I'll focus on text descriptions instead so blind readers get the picture and sighted folks can just click through on the links to the product pages.

Gotta stay awake an extra 10 hours this cycle and push back the sleep phase. Taking this time to write some code and work on my blog post for my cyberdeck.

Off by one error for my plane ticket back 😜

So pumped for the new goat simulator DLC!

In the age of LLMs all data is easier to mine and turn into databases. In particular any private notes you swnd to a cloud service. Before even if yhey read your messages, it was way more work toake them usable. Now it's just a number of dollars and cranks of the slop machine.

The main question seems to be about our use of LLMs in our development process. Main concern is with liability and ownership with code.

We're finally having our "AI policy" discussion at work. It's neat because we have a wide range of opinions. Especially neat given we're an org focused on environmental conservation.