Tutorials¶
Examples and tutorials live with each component. Pick the one that matches what you want to do.
| Component | What you can try |
|---|---|
| vLLM | Start a server, Call the OpenAI-compatible API, Supported models |
| Torch | Move PyTorch code to the LPU, Migrate from CUDA, Run on several cards |
| Compiler | Matmul kernel, Attention kernel, Compile a model with torch.compile |
| Kernels | Run your first kernel, Matmul, Multi-core addition |
| Runtime | Working with memory, Streams and events, Using several cards |
| Profiler | Capture a trace, Read a trace, Use with the PyTorch profiler |
| CCL | Create a communicator, Collectives |
| Driver | Verify the installation, Reset a device |
| Cloud native | Run in a container, Deploy on Kubernetes |