Alejandro Cuevas Rivero (@cuevasrive88179) 's Twitter Profile
Alejandro Cuevas Rivero

@cuevasrive88179

ID: 1772074585468387328

calendar_today25-03-2024 01:34:55

77 Tweet

14 Takipçi

257 Takip Edilen

Cerebras (@cerebrassystems) 's Twitter Profile Photo

No more waitlist – Cerebras inference API is open to all! 1M free tokens/day 20x GPU speed Reasoning in ~1 second It's time to build! inference.cerebras.ai/?utm_source=tw…

Bilawal Sidhu (@bilawalsidhu) 's Twitter Profile Photo

Veo 3 has digested the mother lode of ASMR content on YouTube making it an AI ASMR machine. This one got 3.1M likes and 12k comments in 3 days. Every popular YouTube format is about to get its impossible AI remix.

Chubby♨️ (@kimmonismus) 's Twitter Profile Photo

Trump Administration: “We’re Not Going To Regulate It” The USA is taking the path of deregulation to advance all possibilities for the development of AI. They see the danger with China as the greatest and are doing everything they can to win the competition - regulations are

Trump Administration: “We’re Not Going To Regulate It”

The USA is taking the path of deregulation to advance all possibilities for the development of AI. They see the danger with China as the greatest and are doing everything they can to win the competition - regulations are
Lisan al Gaib (@scaling01) 's Twitter Profile Photo

lmarena has a competitor Yupp is basically lmarena, but with more granular feedback and a credit system. Each message costs you some credits, but when you give high-quality feedback you get credits back to use on your favorite models. This is their multi-turn (5+ messages) VIBE

lmarena has a competitor

Yupp is basically lmarena, but with more granular feedback and a credit system. Each message costs you some credits, but when you give high-quality feedback you get credits back to use on your favorite models.

This is their multi-turn (5+ messages) VIBE
Chubby♨️ (@kimmonismus) 's Twitter Profile Photo

Intelligence too cheap to meter, literally. From June 2023 to April 2025, GPT-4 initially achieved just under 30% in GPQA and now 87.7% with o3 in GPQA. GPQA is Google-proof, and compared to PhDs who have access to Google, o3 now significantly outperforms them. At the same

Intelligence too cheap to meter, literally.

From June 2023 to April 2025, GPT-4 initially achieved just under 30% in GPQA and now 87.7% with o3 in GPQA. GPQA is Google-proof, and compared to PhDs who have access to Google, o3 now significantly outperforms them.

At the same
AshutoshShrivastava (@ai_for_success) 's Twitter Profile Photo

This is terrifying you need to read the new Anthropic article. Anthropic gave AI models a fake but extreme choice, let a person die or stay loyal to their goal. Most models chose death, just to protect their mission. if AI isn't aligned properly, it won’t hesitate to harm

This is terrifying you need to read the new Anthropic article.

Anthropic gave AI models a fake but extreme choice, let a person die or stay loyal to their goal.
Most models chose death, just to protect their mission.

if AI isn't aligned properly, it won’t hesitate to harm
Rohan Paul (@rohanpaul_ai) 's Twitter Profile Photo

LLMs face scalability limits and memorization from finite, fixed reasoning datasets. REASONING GYM (RG) addresses this by providing an infinite library of procedurally generated, verifiable reasoning environments. Methods 🔧: → RG offers over 100 data generators spanning

LLMs face scalability limits and memorization from finite, fixed reasoning datasets.

REASONING GYM (RG) addresses this by providing an infinite library of procedurally generated, verifiable reasoning environments.

Methods 🔧:

→ RG offers over 100 data generators spanning
Google Labs (@googlelabs) 's Twitter Profile Photo

🚨 NEW LABS EXPERIMENT 🚨 Introducing Doppl, a new mobile app that lets you upload a photo or screenshot of an outfit and then creates a video of you wearing the clothes to help you find your ✨aesthetic ✨ Available on iOS and Android in the US to users 18+, download the

Lucas Beyer (bl16) (@giffmana) 's Twitter Profile Photo

I like the Encoder-only Mask Transformer (EoMT): basically removing all the bells and whistles, and doing panoptic segmentation with an almost vanilla ViT. You're sliiiiightly worse for the same encoder size, but it's a lot simpler/faster and (likely) more scalable. I wish they

I like the Encoder-only Mask Transformer (EoMT): basically removing all the bells and whistles, and doing panoptic segmentation with an almost vanilla ViT.

You're sliiiiightly worse for the same encoder size, but it's a lot simpler/faster and (likely) more scalable. I wish they
Cerebras (@cerebrassystems) 's Twitter Profile Photo

Frontier AI is now on Cerebras. This week we are launching Qwen3-235B—Ali Baba’s flagship reasoning model that rivals ChatGPT and Claude. In classic Cerebras style, we run the model at 1,500 tokens/second. That means reasoning time goes from 60 seconds on GPUs to just 0.6

Frontier AI is now on Cerebras. 

This week we are launching Qwen3-235B—<a href="/Alibaba/">Ali Baba</a>’s flagship reasoning model that rivals ChatGPT and Claude. In classic Cerebras style, we run the model at 1,500 tokens/second. 

That means reasoning time goes from 60 seconds on GPUs to just 0.6
METR (@metr_evals) 's Twitter Profile Photo

We analyze data from 9 existing benchmarks: MATH, OSWorld, LiveCodeBench, Mock AIME, GPQA Diamond, Tesla FSD, Video-MME, RLBench, and SWE-Bench Verified, which either include human time data or allow us to estimate it.

We analyze data from 9 existing benchmarks: MATH, OSWorld, LiveCodeBench, Mock AIME, GPQA Diamond, Tesla FSD, Video-MME, RLBench, and SWE-Bench Verified, which either include human time data or allow us to estimate it.
Runway (@runwayml) 's Twitter Profile Photo

Introducing Act-Two, our next-generation motion capture model with major improvements in generation quality and support for head, face, body and hand tracking. Act-Two only requires a driving performance video and reference character. Available now to all our Enterprise

Carlos Santana (@dotcsv) 's Twitter Profile Photo

🔴 ¡RUNWAY PRESENTA ACT-2 ...y WOW! El nuevo sistema de generación y control de personajes a partir de un vídeo ya está aquí y viendo los resultados que presentan, podemos decir que el panorama de la generación de vídeo acaba de dar un nuevo salto!

Jaime Fuentes Alvear (@fuentesalvear1) 's Twitter Profile Photo

Javier Etcheberry, director del #SII, está casado con la también deudora de contribuciones al #SII, Cecilia Sommerhoff (viuda de Miguel Kast Rist), madre de Felipe Kast Sommerhoff, primo de José Antonio Kast Rist, hermano del difunto ministro de la dictadura… es todo tan, tan ..