Summary of Control Token with Dense Passage Retrieval, by Juhwan Lee et al.
Control Token with Dense Passage Retrieval
by Juhwan Lee, Jisu Kim
First submitted to arxiv on: 13 May 2024
Categories
- Main: Computation and Language (cs.CL)
- Secondary: Artificial Intelligence (cs.AI)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary The paper proposes a solution to the hallucination problem in large language models (LLMs) by combining Retrieval-Augmented Generation (RAG) with Dense Passage Retrieval (DPR). The RAG technique embeds relevant information in prompts to obtain accurate answers, but it faces issues in retrieving correct information. To address this, the authors employ the DPR model for fetching domain-specific documents related to user queries. However, the DPR model still lacks accuracy in document retrieval. The proposed solution incorporates control tokens into the DPR model, achieving significantly superior performance with a 13% improvement in Top-1 accuracy and a 4% improvement in Top-20 accuracy. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper solves a problem with large language models (LLMs) that can give wrong answers. They use a combination of two techniques: Retrieval-Augmented Generation (RAG) and Dense Passage Retrieval (DPR). RAG helps get accurate answers by adding relevant information to prompts, but it has issues finding the right info. DPR is used to find documents related to user queries, but it’s not very good at this either. The solution combines these two techniques with special tokens to make them work better together. This makes a big difference in how well they perform. |
Keywords
» Artificial intelligence » Hallucination » Rag » Retrieval augmented generation
