Create workshop: Write Your Own Domain-Specific Benchmark

[deanwampler]

As part of the TSEI promotion plan, a Write Your Own Domain-Specific Benchmark workshop drills into and generalizes the RAG benchmark workshop #42.

How does someone create their own benchmark for their use cases or domain? The session will use the TSEI reference stack, including lm-evaluation-harness and unitxt, with candidate benchmark data, either hand-curated Q&A pairs or synthetic data generated with a teacher model. The session will demonstrate the basics of running the benchmark and interpreting the results.