Tootfinder

Always fun/challenging to read new AI (pre)papers like this. "Base models know how to reason, thinking models learn when".
#AI #Google #reasoning

> DOGE, he said, began acting like “a bunch of people who didn’t know what they were doing, with ideas of how government should run — thinking it should work like a McDonald’s or a bank — screaming all the time.”
The "screaming all the time" part is really common for people in government tech, albeit internal.

The Untold Saga Of What Happened When DOGE Stormed Social Security
This story was originally published by ProPublica. Republished under a CC BY-NC-ND 3.0 license. On Feb. 10, on the third floor of the Social Security Administration’s Baltimore-area headquarters, L…

Left #Bluesky or thinking about it? Into #genealogy or #familyhistory but don't know where to start? Follow @…

Tanzende Swifties bringen Erde noch in 100 km Entfernung zum Vibrieren
Eine wissenschaftliche Studie zeigt, dass tanzende Fans von Taylor Swift seismische Signale erzeugen, die noch in 100 km Entfernung messbar sind.

Detection of Earth's free oscillations utilizing TianQin
Yuxin Yang, Kun Liu, Xuefeng Zhang, Yi-Ming Hu
https://arxiv.org/abs/2510.10107 https://arxiv.…

Detection of Earth's free oscillations utilizing TianQin
The measurement of Earth's free oscillations plays an important role in studying the Earth's large-scale structure. Space technology development presents a potential method to observe these normal modes by measuring inter-satellite distances. However, the disturbance from the Earth's low-degree gravity field makes it challenging for low Earth orbit gravity measurement satellites such as Gravity Recovery and Climate Experiment (GRACE) and TianQin-2 to extract signals from Earth's free oscillatio…

Waaaaas? Eeeeecht? Na so eine Überrschung. 🤪
Plug-in-#Hybride stoßen laut aktuellen EU-Daten im Realbetrieb rund fünfmal so viel CO₂ aus wie offiziell angegeben.
Im Schnitt liegt der Ausstoß bei 139 g/km – ähnlich wie bei reinen #Verbrenner​n. Grund ist die häufige Nutzung des

Fünffach überhöhte Werte für Benzin und CO₂: Plug-in-Hybride laufen offenbar verstärkt im Verbrennerbetrieb
Fünfmal so viel Sprit wie angegeben tanken offenbar Autos, die wahlweise auch Strom laden können. Bordmessgeräte übermittelten von Europas Straßen eine wachsende Kluft zwischen Anspruch und CO₂-intensiver Realität.

Base Models Know How to Reason, Thinking Models Learn When
Constantin Venhoff, Iv\'an Arcuschin, Philip Torr, Arthur Conmy, Neel Nanda
https://arxiv.org/abs/2510.07364 https…

Base Models Know How to Reason, Thinking Models Learn When
Why do thinking language models like DeepSeek R1 outperform their base counterparts? Despite consistent performance gains, it remains unclear to what extent thinking models learn entirely new reasoning capabilities or repurpose pre-existing base model ones. In this work, we propose a hybrid model where we activate reasoning mechanisms in base models at the right time to elicit thinking-model-level reasoning chains, implying that thinking models exploit already existing capabilities. To ground o…

Kimi K2 Thinking is impressive, it is as at least as good as other top foundation models, fully openweights, 1 trillion parameters but only using 32B in a MoE setting. Also you can download and reuse it for free.
https://github.com/MoonshotAI/Kimi-K2

GitHub - MoonshotAI/Kimi-K2: Kimi K2 is the large language model series developed by Moonshot AI team
Kimi K2 is the large language model series developed by Moonshot AI team - MoonshotAI/Kimi-K2

Reading thoughts about a new Chinese openweights AI model, Kimi K2 Thinking.
https://www.interconnects.ai/p/kimi-k2-thinking-what-it-means

5 Thoughts on Kimi K2 Thinking
Quick thoughts on another fantastic open model from a rapidly rising Chinese lab.