#distillation

'Learn how to build an AI agent that summarizes recent conversations for short-term context and stores distilled facts in a FAISS-backed vector memory for long-term recall.'

'NVIDIA's Nemotron Nano 2 delivers hybrid Mamba-Transformer LLMs that run up to 6× faster and support 128K-token context on a single A10G GPU, with most training data and recipes open-sourced.'