Jorge Hernandez 🇺🇦 🏳️🌈 retweeted
Too many people treat AI as a zero-sum race. Open-source is winning by doing the opposite: collaboration over competition!
This is why we’re excited to release today the first part of our partnership with @OpenRouter to make open models easier to use for all! Hugging Face for storage and collaboration. OpenRouter for inference. Connected so you can:
🤗 Run OpenRouter inference directly on Hugging Face
🔓 Access Open model weights directly from OpenRouter
🤖 Store your OpenRouter agent traces on HF to inspect them, train better models or share them with the world!
More collaboration. Fewer zero-sum games. More power to AI builders. Let’s build! 💛💛💛
Jorge Hernandez 🇺🇦 🏳️🌈 retweeted
if hackers are in there stealing all the money, can somebody please open source the scientific literature too ? thanks
Jorge Hernandez 🇺🇦 🏳️🌈 retweeted
I just want everyone involved in this to know you have forced me to explain to my non-political friend the difference between a 'Bulwark conservative' and a 'Dispatch conservative' and now they're side-eyeing me like I'm an insane person, so you should all just chill
I walked away from this stupid spat for a day. So I missed this. Sarah, if you honestly think this “anti-anti Trump” sophomoric bullshit is a thoughtful or clever response to me or description of how I’ve behaved the last decade, I honestly think you are so deep in your bubble that any and all of your political and moral judgement is simply worthless. I don’t even want to bother arguing with you, I’m more embarrassed for you than anything else.
But I’ll engage one more time. Look, the reason for the Bulwark’s current orientation is evaporating as we’re reaching the end of the Trump era. It appears to me and everybody else that during your bow-taking moment the other day you tried to advance your rebranding when you told a room full of conventional progressive Dems that you were a “lib” just like them.
People noticed. They pointed out that you conceded the central criticism of your project. But as you still need to hold onto some vestige of your “conserving conservatism” schtick you tried to clean up your remarks by suggesting you meant “classical liberal.” Literally no one believes that’s what you meant, including obviously the people in that room, and I doubt many believe you’re a classical liberal (is there a chapter of “Classical Liberals For Abdul Syed” I missed?).
The way you frame this makes my point, not yours. You’re a political operative. You think, like a political operative, that you should say or do what’s necessary for your team to win. That’s entirely defensible for political operatives and party hacks (within reason, I have no patience for Trump apologists and hacks. I can show you the receipts if you don’t believe me).
But here’s the point: I’m neither a political operative nor party hack. Kamala Harris would have been better than Trump, happy to say it. But you pretty clearly think that means I should have become a partisan for her and spin for her and Democrats the way you do. I don’t want to be a partisan hack. It’s not for me. But you cannot even comprehend such motives. All you have is a hammer and so everything is a nail to you. Projecting your self-serving bullshit on me doesn’t mean it’s not bullshit, no matter how good it makes you feel.
I’m glad I could clear this up for you.
Jorge Hernandez 🇺🇦 🏳️🌈 retweeted
Shouting matches at conferences. Forbidding students from working with other labs. Dissing each other in the press.
Origins of life research has been no stranger to conflict. But a new generation of researchers is moving the field past all that.
quantamagazine.org/the-evolv…
Jorge Hernandez 🇺🇦 🏳️🌈 retweeted
With the latest Anthropic update, my student's research work on improving AI models for proofs so that they collaborate with humans more effectively is now flagged as a "distillation attack." I knew this reframe would harm public research
Jorge Hernandez 🇺🇦 🏳️🌈 retweeted
This should be a national scandal.
Our troops were left in inadequate housing with no clear casualty evacuation plan.
After an Iranian drone hit the U.S. command post where Master Sgt. Nicole Amor was stationed, she was still breathing. Soldiers drove her in an SUV to a local hospital in Kuwait, and her husband says, “They were googling hospitals to try to take some of the soldiers to. That's absurd.”
The Pentagon declined to answer whether the 103rd was denied adequate force protection. cbsn.ws/4AsEbh8
Jorge Hernandez 🇺🇦 🏳️🌈 retweeted
People are acting like Jacob Coxon is the selected champion of the safety movement, his tweet was just exactly right-place-right time. It was more about the content than the person.
Jorge Hernandez 🇺🇦 🏳️🌈 retweeted
This is not only a major military target because of the role it plays in supporting Russian operations in Southern Ukraine - it is a significant political target aimed at Putin (who opened the bridge himself). Given Ukraine’s preceding campaign to destroy Russian ferries and shipping that go to Crimea, this will be a problem for Russian military operations.
‼️BREAKING: Former Russian State Duma deputy Alexander Nevzorov claims the Crimean Bridge has sustained further damage, with at least one span reportedly collapsed following tonight's attacks. Powerful explosions have been reported around Kerch throughout the evening. Information is still unconfirmed.. #Ukraine
Jorge Hernandez 🇺🇦 🏳️🌈 retweeted
Feels like we do this multiple times a month now:
* Winner of prestigious literature award caught by Pangram, work is AI-generated
* Author's community yells that Pangram results aren't reliable
* Pangram CEO guy offers money to people who can find a single proven false positive
* Nobody can find one
* Award committee puts out statement denying AI was used with no proof. Often with claims of racism, classism, etc.
* Run committee's statement through Pangram.
* It's AI generated
Jorge Hernandez 🇺🇦 🏳️🌈 retweeted
The attack against Kerch 🇷🇺 bridge is ongoing. It seems multiple missiles/drones impacted the bridge central part, which sustained some damage yesterday
This is a very important strike and we will get more information tomorrow regarding its results.
Comme annoncé dans ce tweet d'hier, ukrainiens et russes ciblent massivement les ponts de l'adversaires.
🇷🇺 Ce soir, plusieurs missiles russes ont impactés les ponts nord et sud de Kyiv.
🇺🇦 Dans le même temps, des frappes de drones ukrainiens sont en cours sur le pont de Kerch, après le début d'une campagne la nuit dernière.
-> Les deux camps ciblent les ponts stratégiques de l'autre, la Russie cherchant à couper les deux rives de Kyiv
-> L'Ukraine a aussi frappé les ponts entre le pont terrestre vers la Crimée (sud de l'Ukraine occupé) et le sud de la Russie. Frapper Kerch est une suite logique pour isoler la logistique russe en Crimée.
Jorge Hernandez 🇺🇦 🏳️🌈 retweeted
We are preparing for tomorrow’s meeting of the Ukraine Defense Contact Group in the Ramstein format. Defense Minister Yevhenii Khmara will present our needs and key priorities for cooperation. Of course, we are focusing on air defense interceptors and ammunition supplies for Ukrainian combat aircraft. There is potential to increase drone production for our Defense Forces – we will discuss additional funding. We also need greater momentum in joint production with our partners. The pace of our work must increase, and every day must deliver results for Ukraine and our partners. I count on a substantive meeting. The Defense Minister also reported on the implementation of our long-range sanctions plan – our Ukrainian deep strikes are hitting their designated targets. I approved new operations. Glory to Ukraine!
Jorge Hernandez 🇺🇦 🏳️🌈 retweeted
It seems bad for your reputation as a commentator on AI risk to vaguepost about big crises being caused by AI when they turn out not to be
Jorge Hernandez 🇺🇦 🏳️🌈 retweeted
Both NeXT and 3DFX tried to “do the right thing” with color spaces, and it wound up generally making them look worse than competing options. I didn’t really understand it properly until years later.
It is natural to think that a pixel with a value of 1.0 is emitting twice the photons as a pixel with a value of 0.5, so you can do the obvious linear math to calculate partial pixel coverage, blending a foreground over a background, or interpolating between colors in a texture.
If displays are “linear”, that is correct. NeXT and 3DFX were much closer to linear than other contemporary displays. NeXT had crossover from the deep graphics understanding at Pixar, and 3DFX had the SGI heritage. They did this intentionally.
If you have 10+ bits per channel like the professional rendering systems, linear displays are The Right Thing.
Unfortunately, NeXTStation Colors were 4 bits per channel and 3DFX was 5 or 6 bits per channel, and there… it was not the right thing.
Human eyes aren’t linear intensity sensors. We mostly register relative changes in intensity, so we can tell the difference between 0.01 and 0.02 as a huge doubling, while 0.99 to 1.00 is almost imperceptible.
This distribution of intensities is often referred to as the “gamma” of the display system, but the actual curves like sRGB are a bit more complicated.
If you only have a small number of discrete pixel values, you definitely don’t want to spread them evenly from 0 to 1! If you do, you will have good precision in the brights, but very stark “intensity bands” in the dark colors.
If everything was authored specifically for 3DFX, you would still have the fundamental problem with the dark bands, but more accurate blending and filtering. The problem was that artists made all of their textures in paint programs on conventional non-linear displays, so when they were drawn by 3DFX cards, they got the characteristic “washed out color” look.
You can mathematically correct textures for differences in display gamma, but I didn’t really understand the issue back in the glQuake days, and we were so enthralled by the absolute performance of 3DFX that the color fidelity didn’t get as much attention as it should have.
Even today, tons of games don’t properly handle gamma across their textures and render targets. High resolutions and 24 bit color masks most of it, but you still see it in some slight shimmerings in high contrast textures and edges.
We’re working with @ApolloResearch, @METR_Evals, @TransluceAI, and @GoodfireAI on embedded assessments: giving outside experts deep access to our internal models and systems to investigate alignment and test our safeguards.
These assessments build on our ongoing alignment and safety research, bringing independent scrutiny to how we identify and manage risks in our systems.
openai.com/index/our-first-p…
Jorge Hernandez 🇺🇦 🏳️🌈 retweeted
Retired U.S. Navy Admiral Philip Davidson, whose 2021 testimony gave rise to the “Davidson Window” for China achieving the required capability in order to invade Taiwan by 2027, said on a podcast released Friday that he hopes he was wrong, according to TWZ. “If I’m wrong, that means the things we recommended from Indo-Pacific Command...have been effective. And there hasn’t been war.”
Davidson credited changes since 2021, including U.S. missile and Marine deployments along the First Island Chain, Japan doubling its defense spending to 2% of GDP, Taiwan extending conscription to a year, and the AUKUS security partnership among Australia, the UK and the U.S.
“I would like to see the $14 billion in committed to Taiwan move over there in the next few months or a year,” Davidson said. A roughly $14 billion package, largely PAC-3 and NASAMS air defense missiles, cleared Congress’s informal review in January but has not received President Trump’s sign off and has been on hold since his summit with Chinese President Xi in May, according to the AP.
Separately, Taiwan has a backlog of almost $30 billion in previously approved arms awaiting delivery, according to George Mason University’s Taiwan Security Monitor. Acting Navy Secretary Hung Cao said in May that arms sales to Taiwan were paused to ensure the U.S. has the munitions it needs for Iran operations.
‘Davidson Window’ Admiral Hopes He Was Wrong On China’s Ability To Take Taiwan By 2027
The time frame Adm. Philip Davidson famously declared China could be ready to invade Taiwan is about to arrive.
twz.com/news-features/davids…
Jorge Hernandez 🇺🇦 🏳️🌈 retweeted
Trump made over 28,000 stock trades in the last two years. That's almost 40 trades a day. We need to ban politicians from stock trading. NOW.
Jorge Hernandez 🇺🇦 🏳️🌈 retweeted
HOL in HOL: mechanized soundness, completeness and consistency, with and without infinity. ~ Christoph Benzmüller, Daniel Kirchner.
arxiv.org/abs/2611.17884
#IsabelleHOL #ITP #Logic































