Reproducible benchmark of 8 demonstration-selection methods for in-context learning across 3 tasks and 3 open LLMs.
python nlp benchmarking information-retrieval reinforcement-learning reproducible-research evaluation transformers beam-search semantic-search selection-algorithms meta-learning few-shot-learning sequence-modeling mahine-learning prompt-tuning in-context-learning llm prompt-engineering demonstration-selection
-
Updated
Sep 27, 2026 - Jupyter Notebook