Tinker APIs

GKE Labs has launched OpenRL, an open-source, self-hosted training API for fine-tuning large language models on Kubernetes. The API decouples post-training infrastructure from reinforcement learning r…