Inference Optimisation Vocabulary

Practise vocabulary for making models faster and more efficient: quantisation, ONNX, batching, caching, and latency vs. throughput trade-offs.