Getting Started

Getting Started has one goal: run one GPT-2 request through TensorRT-Model-Connect. Follow these pages in order.

Step	Page	You are done when
1	System Requirements	The GPU is visible and you selected wheel or source.
2	Installation	The installed CLI reports TensorRT support.
3	Build from Source	Optional: the CLI and native DSOs are built for the selected GPU.
4	Quick Start	./gpt2.bundle is built and returns generated text.

A successful run proves the selected environment can resolve the checkpoint, build a bundle, load the GPT-2 family runtime, and execute one text request. It does not qualify every model or hardware profile on that machine.

After the first run, continue to Learning Path.