4 panel comic, 1. cat girl holds headphones over her head, 2. she puts them on her ears 3. they snap together in air at the level of her ears, 4. she cries

starting to think microblogging might not be an effective format for discussing matters of economics

Person expert here. A person only does this if they are wrong and stupid

Pay me on Cash App Instantly exchange money for free on Cash App

looking for money for weed for pain and shampoo+conditioner and fiber

$160 should get me through

weed going a bit faster than normal bcz deadline pressure has upped my pain so i am *downing* cartridges rn in a way i rarely do

cash.me/heartpunk / venmo.com/heartpunk / paypal.me/heartpunk

For no reason in particular.

en.wikipedia.org/wiki/Pig_but...

I just watched the Prologue of Witch From Mercury and I'm just gonna say that it's about reactionary anti-tech sentiment opposed to revolutionary advances in human-machine interaction being used as an excuse for authoritarian measures, with military forces killing unarmed scientists.
No relevance.

Okay, fine, I was wrong on the vibecession.

Instead of using economic data to understand how people are doing, we should use their lived experiences. We should go around and ask as many people as possible what their experience is - how much money they have, etc. - and set policy based on that.

A matter of survival: The human cost of cyber scam operations in South-East Asia A century after the abolition of slavery, a UN Human Rights report shows how trafficking and coercion underpin cyber scam operations across South-East Asia. Drawing on survivor testimonies from many c...

Donating to online scammers posing as legitimate people in need is not morally neutral:

You potentially are giving money to brutal slavery and human trafficking networks:

www.ohchr.org/en/stories/2...

Gaël Duval is the founder and president of the /e/ foundation along with the CEO of Murena. Duval and his organizations have consistently taken a stance against protecting users from exploits. In this video, he once again claims protecting against exploits is for only useful pedophiles and spies.

Virtual Army Accused Of Illegality By Using Aimbots & Wallhacks

parameters of consent is a real thing. do not fuck LLMs below 192b

I'll need to do a comparative analysis of the cognitive process as outlined by Abhidhamma with the multimodal transformer architecture one day.

this post gave me tensor dysphoria

A man in a black T-shirt leaning on a porch railing and looking toward a road and grassy yard. His head has been replaced with a crudely drawn yellow duck head with a large orange beak, making him look like the duck from the “Got any grapes?” song. In the distance on the left is a small cartoon lemonade stand with a person behind it; the stand is labeled “Ice Fresh” on top and “Lemonade” on the front. Large caption text across the top reads: “I bring a sort of ‘got any grapes?’ vibe to the lemonade stand that the man running the stand don’t really like.”

MiniMax M2.7: Early Echoes of Self-Evolution

Minimax m2.7 is now open-weight

nya ╱|、
(˚ˎ 。7
|、˜〵
じしˍ,)ノ

Jean Baudrillard with glowing red eyes

This is also a fake image. The details of the capsule are entirely wrong.

F-117 Nighthawk, a very low poly aircraft

wild to think that polygon budgets were so low in the 80s

a brandy, I think, but it depends on one's position on transubstantiation

Ok, understood. Thank you.

Fair enough. Appreciate the reply.

like googol?

tiny large language models

"Alien" poster parody, with an allen key replacing the xenomorph egg

Twitter screenshot, "When you compare the size of a gummy worm versus a gummy bear, it starts to paint a horrific picture of the gummy universe.", below photos of a gummi worm towering over gummi bears, and being ridden by one like a sandworm from Dune

call me a wet blanket, but "a wild fox is too friendly" immediately connects in my head to "run away and find where should I report rabid animals"

CharBench: Evaluating the Role of Tokenization in Character-Level Tasks Tasks that require character-level reasoning, such as counting or locating characters within words, remain challenging for contemporary language models. A common conjecture is that language models' reliance on subword units, rather than characters, contributes to their struggles with character-level tasks, yet recent studies offer conflicting conclusions about the role of tokenization, leaving its impact unclear. To address this gap, we introduce CharBench, a comprehensive benchmark of character-level tasks that is two orders of magnitude larger than existing alternatives. We evaluate a diverse range of leading open-weight and proprietary models on CharBench and find that it presents a significant challenge to modern LLMs, with an average accuracy of 43.6% and 32.3% on some tasks. We present an in-depth analysis of how intrinsic properties of words and their segmentations into tokens correspond to model performance. For counting tasks, we find that tokenization properties are weakly correlated with correctness, while the length of the queried word and the actual character count play a more significant part. In contrast, for tasks requiring intra-word positional understanding, performance is negatively correlated with the length of the token containing the queried character, suggesting that longer tokens obscure character position information for LLMs. We encourage future work to build on the benchmark and evaluation methodology introduced here as tools for improving model performance on such tasks.

It's not really true.
arxiv.org/abs/2508.02591

I feel like with enough effort you could start a new environmental outrage about the energy usage of refrigerators, not due to the compressor, but due to the light (and how it stays on even with the door closed.)