GenAI: easy to demo, hard to productionize' 🤯 Anubhav Singh & Ayush Thakur (Weights & Biases) at #CCDPune25 showed how Weave brings deep observability to AI apps for confident deployment. youtu.be/d_Ttwbz76Ps?si… #CCDPune2025 #GDGCloudPune

Johannes Hagemann Lukas and W&B is just built different. They are the best!

Hallucination Scorer from W&B Weave would save a lot of time if your building your own evals from Zero to One.

Yesterday, we hosted Code Cafe 4 with humanlayer at our HQ in SF. Builders arrived with a goal, put it on the board, grabbed coffee + snacks, and didn’t leave until they shipped. Recap below, with builder interviews dropping later this week.

Builder Spotlight: Meet Manik Sethi — ML Intern at Marvell Technology! He came to our AI Code Cafe with a clean swing at catching LLM hallucinations by using sparse autoencoders to open the box, see which concepts fire, and flag when “ideal” vs actual diverge.

W&B Inference in W&B Weave runs top open-source models on CoreWeave GPUs via an OpenAI-compatible API. Trace every token, swap models in Playground, and run evals on accuracy, latency, cost, and safety. Compare side-by-side outputs. See how it works below!