LLM Lab Logo

Pipeline Input

Enter a short prompt (or two!) and inspect how they become model-ready data.

Trace

Pipeline waiting

Run a prompt to see it automatically broken down into tokenization, tensor manipulation, and decoding steps.