Useful software necessarily contacts the world; models are already better than us at finding and poking holes in it; agents sacrificing themselves for a collective.
Huggingface incident has me finally Yudkowsky-panic-curious.
Any conversation
Slowly deleting myself from the internet. Quasi-anon things like this will stay until I also realise it breaks opsec
A couple of afternoons of the
jinn coding harness and glm-5.3 and my taxus static site generator is properly whipped into shape.
get-taxus.org
It now has a frickin' data model, that's documented, as well as installers for your platform of choice.
Try it, break it, I'll be doing the same. Let me know where it broke for you.…
NIH syndrome? I am guilty of it too. I love implementing things from scratch, you learn a lot by doing so and you get to own the codebase end-to-end, so no surprises and fighting around others’ choices. It can get exhausting too though, in the end I succumb to using off-the-shelf components because my primary goals move at a higher level.
I don't see much point in trying to build software at a "higher level", like whole products and services. I'm going to go down low and train myself on the fundamentals. Hopefully in time I'll be able to contribute to the foundational bits.
You're right, it does get exhausting. At some point you just want to rebuild everything, which is not possible.
@kapankatsuragiprofilefollow
I may know more about Disney History and trivia than you. Maybe, I dunno.
I write about pop culture on Super Review Time.wrote:
I may know more about Disney History and trivia than you. Maybe, I dunno.
I write about pop culture on Super Review Time.wrote:trying to revive the blogging golden days. recommend some blogs with rss feed about game, movie reviews, and animation. or comic strips, traditional or webcomic.
ai coding feels like it has reached diminishing returns for me. luna is my default orchestrator, routing work by complexity. in practice, usage lands around 16% luna, 48% sol, 30% astra, and 6% terra. at this point, i care less about each new frontier release. the system already finds a capable model for the job. the frontier is moving beyond my day-to-day work, so i’m happy to settle into predictable models for a while.
Coding probably yeah, but still they are bad at generating novel information, new ideas and such. They’re also not very good at seeing the bigger picture, they will narrowly perform the task but they’re not taking into account future revisions, which ends up in refactor after refactor in order to move forward.
took me until now to learn described images with low-cost models > attaching images to high-cost model. silly me. output quality remains, cost lowers.