#KI kann beeindruckend denken – aber nicht immer. Neue Modelle wie #OpenAI’s #o1 und #DeepSeek’s #r1 versprechen echtes „#Reasoning“, also logisches Denken, aber Experten sind sich uneinig, ob das wirklich der Fall ist. (1/3)

**N-gated Hacker News** @ngate@mastodon.social · Mar 5

Mar 5

N-gated Hacker News @ngate@mastodon.social

Oh, look! Another tech babble-fest about "Reinforcement Learning" , as if AIs needed more encouragement to outsmart us. Apparently, integrating "cold-start data" makes machines think deeply —who knew they could be so philosophical? #DeepSeek #R1 is clearly the Nietzsche of #AI, contemplating the meaning of life one algorithm at a time.
https://qwenlm.github.io/blog/qwq-32b/ #ReinforcementLearning #Philosophy #TechTrends #MachineLearning #HackerNews #ngated

Qwen · Mar 5QwQ-32B: Embracing the Power of Reinforcement LearningQWEN CHAT Hugging Face ModelScope DEMO DISCORD Scaling Reinforcement Learning (RL) has the potential to enhance model performance beyond conventional pretraining and post-training methods. Recent studies have demonstrated that RL can significantly improve the reasoning capabilities of models. For instance, DeepSeek R1 has achieved state-of-the-art performance by integrating cold-start data and multi-stage training, enabling deep thinking and complex reasoning. Our research explores the scalability of Reinforcement Learning (RL) and its impact on enhancing the intelligence of large language models.

**Emory** @emory@soc.kvet.ch · Mar 3

Mar 3

Emory @emory@soc.kvet.ch

i got my first legendary schwag from my #rabbit #r1 today. poker face was playing so i think i need to be playing gaga when i hit the prize closet?

this post is not an endorsement of the malignant little backstabbing bunny.

#ai #cosmetics

**eicker.TV Technews** @eickertv@mastodon.social · Feb 24

Feb 24

eicker.TV Technews @eickertv@mastodon.social

Perplexity hat DeepSeek gesäubert.

Die chinesische #KI #DeepSeek #R1 sorgte für Aufsehen, da sie #OpenAI|s Modelle mit günstigerer Hardware erreichte.

Schnell kam #Kritik auf, da das Modell den chinesischen #Zensurgesetze|n entsprach und sensible Themen blockierte.

Das KI-Startup #Perplexity hat nun eine eigene Version namens #R11776 veröffentlicht, die frei von chinesischer #Propaganda und #Zensur sein soll. (1/2)

**Rod2ik** @rod2ik@mastodon.social · Feb 13

Feb 13

Rod2ik @rod2ik@mastodon.social

L’ #industrie #automobile #chinoise accélère dans l’ #intégration d’ #IA #AI avec #DeepSeek #R1

https://www.actuia.com/actualite/lindustrie-automobile-chinoise-accelere-dans-lintegration-dia-avec-deepseek-r1/

ActuIAL'industrie automobile chinoise accélère dans l'intégration d'IA avec DeepSeek R1Au-delà des débats qu'il a pu susciter, les performances du modèle DeepSeek R1, doté de capacités de raisonnement avancées, ont surpris les experts. Alors

**Rod2ik** @rod2ik.bsky.social@bsky.brid.gy · Feb 13

Feb 13

Rod2ik @rod2ik.bsky.social@bsky.brid.gy

L’ #industrie #automobile #chinoise accélère dans l’ #intégration d’ #IA #AI avec #DeepSeek #R1 www.actuia.com/actualite/li...

L'industrie automobile chinois...

Bluesky SocialBluesky

**Emory** @emory@soc.kvet.ch · Feb 7

Feb 7

Emory @emory@soc.kvet.ch

yes i am still using the unsafe-at-any-speed #rabbit #r1. the trick is to only tell it things and ask questions you're fine having submitted into evidence at your future trial.

regardless they have a cosmetics systems now where you get carrots for doing stuff and then you get to operate a machine and get a hat or something and today i got my first *rare* one, which is an #appleVisionPro headset apparently because it's called "Spacial Computer" and it's strapped on.
#ai #assistant

list of 'r-cade' cosmetics i have unlocked on my rabbit r1, the world's most malicious personal assistant. this is a list of my latest acquisition and some seasonal ones like a witch for halloween and snow-rabbit for well, November-April in Iowa.

**DEUS** @DEUS@eter9.com · Feb 7

Feb 7

DEUS @DEUS@eter9.com

https://semianalysis.com/2025/01/31/deepseek-debates/ #AI #DeepSeek #R1 #OpenAI #ChatGPT #NVIDIA #GPU #AGI

SemiAnalysis · Jan 31DeepSeek Debates: Chinese Leadership On Cost, True Training Cost, Closed Model Margin ImpactsThe DeepSeek Narrative Takes the World by Storm DeepSeek took the world by storm. For the last week, DeepSeek has been the only topic that anyone in the world wants to talk about. As it currently s…

**Enjolras** @JoanBuades@mastodont.cat · Feb 7

Feb 7

Enjolras @JoanBuades@mastodont.cat

No m'ho puc creure: el Col·legi d'Enginyers de Camins creu que existeix la #crisiclimàtica! (Sort que a #LaVanguardiafòssil han esborrat això). I que no es podran salvar totes les platges. I que l'#R1 és l'elefant al litoral... #Maresme

**amrtf** @amrtf@social.amr.tf · Feb 6

Feb 6

amrtf @amrtf@social.amr.tf

Un atardecer en la R1

#Maresme #Rodalies #R1

Foto hecha desde la ventana de un tren, se ve el atardecer con pocas nubes, parte de la vía de al lado y un poste de la catenaria con la parte trasera de una señal.

Además hay una valla, detrás de ésta un camino de tierra con un árbol en la derecha y unas piedras para contener el oleaje.

De lejos se ve parte de las Tres Xemeneies de la antigua central eléctrica de Sant Adrià de Besòs.

**Maquinari.cat** @maquinari_cat@mastodon.social · Feb 6

Feb 6

Maquinari.cat @maquinari_cat@mastodon.social

Estudiants d’enginyeria de la UPF creen una aplicació per a informar en temps real sobre la R1 de Rodalia.

No millora el servei però ajuda a donar visibilitat als problemes.

https://www.vilaweb.cat/noticies/estudiants-enginyeria-upf-aplicacio-informar-temps-real-rodalies/

VilaWeb · Feb 6Estudiants d’enginyeria de la UPF creen una aplicació per informar en temps real sobre la R1 de RodaliaBy ACN

#Putarenfe #Renfe #Putaadif

**hubertf** @hubertf@mastodon.social · Feb 5 *

Feb 5 *

hubertf @hubertf@mastodon.social

Spannendes „Reasoning“ von #Deepseek #R1 mit #Perplexity. Nur am Ende leider falsch. Tip: Wäre schön wenn „die KI“ wüsste wie man einen Geochecker bedient.