lyra bubbles~ ♪❀
@_lyraaaa_
ˈli.ɹə 🏳️⚧️⚢ · 24 · ableton enjoyer · developing music ai tools · base model appreciator · beat saber · lyraaaa_ on discord · ♡ @itschloebubble ♡
glm 4.5 fake-base is actually batshit insane what the hell is this
inspiration can come from the strangest of places sometimes
This. I get all my best ideas while data cleaning/skimming through random samples.
"its not intellectually stimulating, so its not worth doing. i'm above working on this" is a trap that smart people seem to have a very, very easy time falling into
this type of thing is the only redeeming aspect of "ai art," hallucinations of a computer in psychosis
the existence of the SCP foundation wiki almost certainly did something radically weird to the global dataset
i reran the meta-leaderboard too includes newer models (claude 4, grok 4, K2) i added slop-bench and DAT also excluded system-based evals (lancer, mlebench, etc) and ones that only evaled like 6 models
The Ultimate LLM Meta-Leaderboard averaged across the 28 best benchmarks Gemini 2.5 Pro > o3 > Sonnet 3.7 Thinking
sometimes I envy other people that can listen to music while they work