Choose your path

Pick the entry that matches your goal; each one names the page to start from and what it covers.

---

I want to compare two models on identical hardware.

Start at Tutorials: multi-engine study. Covers setting up a sweep config, running both models in a single command, and reading the comparison output.

---

I want to integrate LLenergyMeasure into a CI pipeline.

Start at How-to: Docker setup, then How-to: Run with Docker and vLLM. Covers image pull/build, environment variables, and the llem run invocation pattern that works inside CI.

---

I want to understand the measurement methodology.

Start at Methodology overview. Covers warmup protocol, baseline power subtraction, thermal stabilisation, and reproducibility. Then read Why LLenergyMeasure for the research-gap context.

---

I want to extend LLenergyMeasure to a new engine.

Start at Architecture: engine extensibility for the harness-plugin contract your engine needs to satisfy. Then see Contributing: extending miners for writing a parameter miner and wiring it into the validation pipeline.

---

I serve open-weights models and want to make my deployment more efficient.

Start at For serving open-weights models. Frames the deployment-decision use of LLenergyMeasure and points at the quick-start, multi-engine study tutorial, and output-interpretation guide in the order an operator would read them.