Maru @marumewmew - Tumblr Blog

importantcatpics

billiuspendragon

Okay, fine, I'll say it. I'll speak my mind. Meow meow meow meow meow.

kosmogrl

growtiredofpublicvulnerability

real af this is genuinely me after deciding to get up and cut some apples

blignick

Meet the guinja, America's smallest wild cat, the size of a kitten 😺

The forests of Chile and Argentina are home to the guinja, a miniature wild cat that grows to only 40–50 cm and weighs about 2 kg. This stealthy predator is so rare that even these shots are considered a great success. Small but proud - and almost elusive😀

astriloquum

one-step-at-a-time25

on anger

saltair-and-webweaves

I wonder if you wish for a different daughter instead of me. - user jay says on medium/twenty - silas melvin/@/eternaldroplets on twitter/if my body could speak - blythe baird/mother - maia baia/home is not a country; "mama" - safia elhillo/acts of desperation - megan nolan

samglyph

Witch Hat Atelier by Kamome Shirahama // Wisdom to Heal the Earth by Rabbi Tzvi Freeman

biteorflight

[error]- person not found

hibi-flower

sources: l.e.s. by donald glover // this year by the mountain goats // maya hornbacher, madness // watchisjawdrop on tumblr // wuntrum on tumblr // kim kardashian // unknown // godblessryankent on tiktok // nayyirah waheed // unknown // lev st. valentine

on hope, finally

superrgirl

Milly Alcock for GQ Hype Magazine, 2026

marumewmew

I love her. That's it 😭

neopetdestroyer

sofuton

plaidandpleather

I recently went to a conference where a mathematician who is also an AI researcher presented some of his basic work on LLMs, he also gave us a quick deep dive into the maths of how they work.

It was a fucking great presentation and so cool. But what was most interesting was an experiment he conducted where he tested how accurately LLMs can answer questions when given curated knowledge bases to retrieve information from (aka "Retrieval Augmented Generation"):

In IDEAL CIRCUMSTANCES, where the LLM was artificially given the exact right piece of info to answer the question, it answered correctly only 75% of the time. That was the best case scenario! Only achievable in lab settings!

In "normal" circumstances, so using the standard algorithms to allow the LLM to retrieve the information from its knowledge base by itself, it only answered correctly roughly 60% of the time.

WITHOUT A CURATED KNOWLEDGE BASE IT ANSWERED CORRECTLY 7% Of THE TIME.

LLMs are so fucking stuuuuupid. Because it is a LANGUAGE ALGORITHM, it's good at stringing together coherent sentences based on an input. That's what it was built for. To make words. It is REALLY, REALLY not a Answer Questions or Complete Tasks algorithm.

If only I could get my clients to understand that.