PyTorch Ecosystem Adds 10 Projects Spanning RL, Inference, Survival Analysis

The PyTorch Ecosystem Working Group added 10 open-source projects to its Ecosystem Landscape¹. New entries include Perforated, a data-efficiency library that applies neuron-specific reinforcement learning signals during backpropagation; AReaL, a fully asynchronous RL infrastructure for LLM/VLM-based agents; TorchJD for multi-loss training via Jacobian Descent; TokenSpeed, an inference engine separating control and execution planes; and TorchSurv, a survival-analysis toolkit included in the FDA's Regulatory Science Tools Catalog. RLinf, Miles, SMG, FiftyOne, and VisualTorch round out the cohort.