Sigmoidal Scaling Curves Make Reinforcement Learning RL Post-Training Predictable for LLMs

Meet ‘kvcached’: A Machine Learning Library to Enable Virtualized, Elastic KV Cache for LLM Serving on Shared GPUs

Guardrails: Schlüssel zur zuverlässigen KI mit LLMs

Less is More: Recursive Reasoning with Tiny Networks

QeRL: NVFP4-Quantized Reinforcement Learning (RL) Brings 32B LLM Training to a Single H100—While Improving Exploration

2:4 Sparsity + Quantisierung: Der Schlüssel zur effizienten LLM‑Kompression

TokenFlow: Responsive LLM Text Streaming Serving under Request Burst via Preemptive Scheduling