Retrieval Augmentation
Retrieval augmentation enhances large language models (LLMs) by incorporating external knowledge sources to improve accuracy, address hallucinations, and handle long contexts. Current research focuses on optimizing retrieval methods (e.g., k-NN, dense retrieval), integrating retrieved information effectively into LLMs (e.g., through modality fusion), and developing frameworks for managing and utilizing this external knowledge (e.g., dynamic retrieval based on model confidence). This approach is proving valuable across diverse applications, including question answering, text summarization, code generation, and even medical diagnosis, by improving factual accuracy and mitigating the limitations of LLMs trained solely on parametric knowledge.
Papers
On The Adaptation of Unlimiformer for Decoder-Only Transformers
Kian Ahrabian, Alon Benhaim, Barun Patra, Jay Pujara, Saksham Singhal, Xia Song
Can We Further Elicit Reasoning in LLMs? Critic-Guided Planning with Retrieval-Augmentation for Solving Challenging Tasks
Xingxuan Li, Weiwen Xu, Ruochen Zhao, Fangkai Jiao, Shafiq Joty, Lidong Bing
Comparing Retrieval-Augmentation and Parameter-Efficient Fine-Tuning for Privacy-Preserving Personalization of Large Language Models
Alireza Salemi, Hamed Zamani
A Compressive Memory-based Retrieval Approach for Event Argument Extraction
Wanlong Liu, Enqi Zhang, Li Zhou, Dingyi Zeng, Shaohuan Cheng, Chen Zhang, Malu Zhang, Wenyu Chen