Instructor resource. A classroom-ready introduction to NVIDIA's Ising Calibration NIM — an open vision-language model purpose-built for analyzing quantum hardware calibration plots, benchmarked on superconducting qubits and neutral atoms. No setup, no API key; the playground runs in the browser.
→ Launch the Ising Calibration NIM playground
→ Sample plots: QCalEval dataset on Hugging Face
Upload any calibration experiment plot — from your own lab, a simulation run, or the QCalEval dataset — and ask the model to analyze it. The model returns structured responses across six question types:
- Technical description of the experiment
- Experimental conclusion
- Experimental significance
- Fit quality assessment
- Parameter extraction
- Experiment success classification
Students can ask for all six at once or target specific ones.
A few ways to weave the playground into a course:
- First-pass plot reader. A starting point for interpreting an unfamiliar Rabi, Ramsey, T1, T2, or randomized benchmarking trace.
- Second opinion. A check on a student's own analysis of their lab data.
- Reasoning exemplar. A way to expose what structured calibration reasoning looks like across different experiment types — useful even when students disagree with the model.
The playground requires no environment setup, so it slots into a single lecture, a lab section, or a homework prompt without any deployment overhead.
Hands-on tutorial notebooks for the Calibration track are in development. They will walk through using the Ising Calibration NIM end-to-end alongside CUDA-Q simulated calibration experiments. Today the folder contains a single 00_StartHere.ipynb landing notebook that points back to this README.