🔬 RL with Verifiable Rewards, but the Verifier is a Lab — Lila Sciences

Argues that automated labs can act as the verifier in RL, creating a new source of scalable training data once internet text is exhausted. It is a concrete direction for how science-focused models may be trained.

• videoWhy Every Agent needs Open Source Cloud Sandboxes
• videoTerminal-Bench 2.0: the most impt coding agent benchmark of 2025 gets a v2! Launch + Q&A w/ founders
• video[State of Post-Training] From GPT-4.1 to 5.1: RLVR, Agent & Token Efficiency — Josh McGrath, OpenAI
• video⚡️ Reverse Engineering OpenAI's Training Data — Pratyush Maini, Datology

• videoThe RLVR Revolution — with Nathan Lambert (AI2, Interconnects.ai)Latent Space
• articleLila Scienceswww.latent.space
• video🔬 Why Transformers Hit a Wall the Moment Physics Shows Up — Anima Anandkumar, CaltechLatent Space

Checking sign-in…

Loading comments…