🚀 Big launch from @OpenPipeAI: We just launched Serverless RL — train agents faster and cheaper with zero infra headaches. Compared to running your own GPUs, Serverless RL is: - 40% cheaper - 28% faster wall‑clock - instantly deployed to prod via @weights_biases Inference

RL X-mas came early. 🎄 For too long, building powerful AI agents with Reinforcement Learning has been blocked by GPU scarcity and complex infrastructure. That ends today. Introducing Serverless RL from wandb, powered by @CoreWeave! We're making RL accessible to all.

Starlink Mini offers fast, reliable internet on the go—great for traveling, camping, exploring, boating, RVing, and more. Stay connected without dead zones or slow speeds. Order online in under 2 minutes.

This screenshot will be on so many slide decks.

We’re excited to announce our agreement to acquire @Monolith_AI, a pioneer in ML tools for complex engineering challenges. Together, we’ll accelerate R&D and design the next generation of industrial innovation. Read today’s press release here: https://t.co/srm8skG6zJ

This was on my bucket list for sometime now. Glad to tick it off. A lot of value is generated from this hq.

I love watching high-quality animes. The studios should adopt sora and upcoming video models to deliver new episodes with similar quality but quicker. It is a win-win for everyone.

We are ushering into a new era of TAAS -- training as a service. An AI agent can collect data, pre-process it, use TAAS and then consume it for infer. It can further deploy it in a controlled env, measure drift and then collect data for failure cases, use TAAS and on and on.

"The velocity, the intensity, the aspiration, the audacity"

Today, we are releasing FineVision, a huge open-source dataset for training state-of-the-art Vision-Language Models: > 17.3M images > 24.3M samples > 88.9M turns > 9.5B answer tokens Here are my favourite findings:

Hyderbad, @weights_biases is coming to the City of Nawabs! We're collaborating with the amazing folks at @hydPython - one of the most loved communities in India! 🎟️ - https://t.co/H7cVNeuGKp Join me and @ayushthakur0 in a hands-on workshop on building voice AI agents using

Finally we can share our Google calendar link for booking calls. No need for Calendly. :)

BIG LAUNCH: @deepseek_ai's V3.1 is now live on W&B Inference! One model, two modes: toggle between high-speed 'Non-Think' & deep 'Think'. Priced at just $0.55/$1.65 per 1M tokens, it’s a game-changer for building intelligent agents. Want $50 in free credits? Details below.

With over 40M hours watched, Naruto is the top 1 anime on Netflix during the first half of 2025.

This happened: https://t.co/jaNNQoCdUK

Test cricket is well and truly back. India vs England will be remembered for outstanding performances, moments of sheer grit, and unrelenting determination from both sides. A hard-fought draw that, given the context, feels like a win. Well played, boys. 👏

`import wandb` Alias-free. Hype-free. Just vibes and clean metrics.

Finally sourced all the servos to start designing and building an arm. After researching through multiple possible servos that fit a certain cost bracket, the SC/ST series of servos are best suited for small scale bots.

> 🔄🐮 New Paper from Xi’an Jiaotong & NTU team > MUR allocates thinking budgets based on momentum‑uncertainty, tracking step‑wise uncertainty over time > Introduces γ‑control to tune reasoning length with one > Cuts computation by >50 % and improves accuracy by 0.62–3.37 %

Let's go! Day 1 support for Qwen3-Coder and Qwen3-2507 models on the @weights_biases inference service, powered by @CoreWeave . #OneTeam #aicloud

As the saying goes, it takes a village to raise a child, and in the forests of Ramgarh, it took a forest, a railway, and the locals to welcome one into the world. A pregnant elephant lay in labour on the tracks. The forest and railway officials acted swiftly. A coal train was