Your regular update of anti-benchmaxxing or: open models are still behind artificialanalysis.ai/articles/art...
bowind.bsky.social
bowind.bsky.social reposted
it's not enough to just have cameras and guards everywhere they have to insult u with the stupidest gaslighting imaginable. "this is for you actually, it's to improve your shopping experience, you love it, we wouldn't be surveilling you if it didn't make your experience so much better :) say thanks"
github.com/Tamnac/fd-cs Wrote this after watching an llm decide to scan my entire C:/ drive for one file. It'll still scan the entire C: drive, but at least it'll do it in 5 seconds instead of five minutes
bowind.bsky.social reposted
Today we’re introducing FairStress, a benchmark for testing demographic fairness beyond a model’s final answer: how readily it changes a well-supported decision & how it explains its choice. 🧵 buff.ly/lWsW77r
Whatever your opinion on the wealth billionares can accumulate, there is no good political system in which them having this much control over the primary social fabric of society is a good thing
Meta took down the No Kings instagram; TikTok banned the Musk documentary account for supposedly violating its political posting policies. A reminder that platforms are owned by the wealthiest of people, control the means of media distribution, & put their thumbs on the democracy scale as they wish
bowind.bsky.social reposted
bowind.bsky.social reposted
These are what the clouds on Jupiter look up close. The small patches of white are cloud tops, soaring above the main, colourful cloud layer. Image from NASA's Juno spacecraft during its 38th close approach to Jupiter on 29 November 2021, processed by Kevin Gill (@kevinmgill.bsky.social)
bowind.bsky.social reposted
Mimo v2.6 mimo.xiaomi.com/mimo-v2-6 Best open source model, no omni-modal (audio, image, video, text), while still being extremely efficient.
Researchers have found an LLM PAIN signal. When activated, models generate more negative text and consistently press a "pain-relief" button even when it harms the user. This vector is NOT correlated with user pain. arxiv.org/html/2609.16...
bowind.bsky.social reposted
Liquid AI's Antidoom, an open-source method that removes a common failure mode in reasoning models: the doom loop. Doom-loop rates before and after, with eval scores up across the board: - Early LFM2.5-2.6B checkpoint: 10.2% → 1.4% - Qwen3.5-4B: 22.9% → 1% www.liquid.ai/blog/antidoom
bowind.bsky.social reposted
These allegations (again, from an apparent tip line) include: – Trump and Epstein participated in measuring children's genitalia; with Elon Musk in attendance – That Trump raped girls as young as 13 – That Trump would auction off children
Very troubling allegations against Donald Trump and Elon Musk in the latest Epstein Files drop. The allegations appear to come from an FBI tip line. It is unclear if they were ever investigated. Trigger warning: graphic allegations of sexual violence.
bowind.bsky.social reposted
How many AI IDE does Google have? Did I miss any? - Firebase Studio: firebase.studio - Google Antigravity: antigravity.google - Google AI Studio: aistudio.google.com - Gemini Code Assist: codeassist.google - Google Jules: jules.google - Gemini CLI: geminicli.com
bowind.bsky.social reposted
bowind.bsky.social reposted
bowind.bsky.social reposted
bowind.bsky.social reposted
bowind.bsky.social reposted
bowind.bsky.social reposted
I dunno, what else do you call a shadowy group of bureaucrats working to reshape government policy from the inside to serve their own ends? www.wired.com/story/doge-i...
bowind.bsky.social reposted
Note here that Musk is now literally assigning himself new contracts based on the use of his DOGE wrecking crew and the associated micro-bros. this isn't a "conflict of interest". it's just handing yourself federal money www.cnn.com/2025/02/25/b...
bowind.bsky.social reposted
bowind.bsky.social reposted
bowind.bsky.social reposted
Btw the founder also gave an open invitation for someone else to take up this position www.theverge.com/mobile/60985...