Package index • llmsa

Plan and run a study

msa_design(): Plan a crossed precision study for an LLM pipeline
msa_run(): Execute a design against a backend and keep an audit trail
as_msa_runs(): Import results produced elsewhere
msa_counts(): Realised cell counts of a study
backend_mock() backend_ellmer() backend_openai_compatible(): Backends: how a prompt becomes a response
parse_identity() parse_number() parse_choice() parse_json_field(): Parsers for model responses

msa_fit(): Estimate the variance components of an LLM measurement system
msa_grr(): Gauge repeatability and reproducibility of an LLM pipeline
msa_boot(): Bootstrap intervals for gauge R&R statistics
msa_agreement(): Attribute agreement of a categorical LLM pipeline
msa_bias(): Bias and linearity of a numeric LLM pipeline against a reference

pb_design(): Plackett-Burman screening design for robustness testing
msa_robustness(): Analyse a robustness screening experiment
msa_simulate(): Simulate a study with known variance components
msa_plan(): Plan the size of a precision study by simulation
msa_control_chart(): Control chart with Westgard rules for ongoing system suitability

msa_report(): Render a validation report
acmg_demo: Simulated ACMG/AMP variant classifications by an LLM pipeline
ic50_demo: Simulated numeric extraction (log10 IC50) by an LLM pipeline
llmsa llmsa-package: llmsa: Measurement System Analysis for Pipelines Built on Large Language Models