auto-cot-bert-clustering-diverse-demonstrations
IN premise — summaries/2026/08/24/wiki-In-context_learning_natural_language_processing-chunk-1.md
Created 2026-08-24T17:11:13+00:00
Auto-CoT vectorizes a question library via BERT, clusters the vectors, selects near-centroid questions per cluster for diversity, generates zero-shot CoT answers for each, and assembles them into a diverse few-shot demonstration set.
Summary
This describes an automated pipeline that groups similar questions into clusters, picks a representative from each group, and assembles a diverse set of step-by-step reasoning examples without any manual curation. It matters because it gives the system a scalable way to build few-shot demonstrations that cover a wide range of question types rather than relying on hand-picked examples.