In-context Learning with Retrieved Demonstrations for Language Models: A Survey

Authors: Man Luo, Xin Xu, Yue Liu, Panupong Pasupat, Mehran Kazemi

TMLR 2024 | Venue PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Theoretical Given the promising results and growing interest in Ret ICL, we present a comprehensive survey of this field. Our review encompasses: design choices for ICL demonstration retrieval models, retrieval training procedures, inference strategies and current applications of Ret ICL. In the end, we explore future directions for this emerging technology.
Researcher Affiliation Industry Man Luo EMAIL Intel Lab. Xin Xu EMAIL Google Research Yue Liu EMAIL Google Research Panupong Pasupat EMAIL Google Research Mehran Kazemi EMAIL Google Research
Pseudocode No The paper describes methods and concepts textually and through tables/figures, but does not present any explicit pseudocode or algorithm blocks.
Open Source Code No The paper includes a link to a GitHub repository (https://github.com/luomancs/luomancs-reticl_llm_survey/) in a footnote, but this repository contains 'The list of papers in this table' (Table 1), not the source code for the methodology or analysis presented in this survey paper.
Open Datasets No As a survey paper, this work does not conduct new experiments requiring its own datasets. It discusses various datasets used by the surveyed papers (e.g., 'MS Marco', 'Community QA', 'SST-5'), but it does not provide access information for a dataset used in *its own* empirical studies.
Dataset Splits No This paper is a survey and does not conduct original experiments, therefore it does not define or utilize specific training/test/validation dataset splits.
Hardware Specification No As a survey paper, this work does not conduct experiments, and therefore does not specify any hardware details used for running experiments.
Software Dependencies No The paper is a survey and does not describe a new methodology that would require specific software dependencies with version numbers for its implementation. It discusses various models and techniques, such as 'SBERT' and 'BM25', in the context of the surveyed literature.
Experiment Setup No This paper is a survey and does not describe any original experiments, hence it does not provide specific experimental setup details or hyperparameter values.