Retrieval Augmented Generation (RAG) is a technique in natural language processing that enhances ... more Retrieval Augmented Generation (RAG) is a technique in natural language processing that enhances generative AI models' accuracy and reliability by incorporating facts fetched from external sources like the internet, databases, open-domain datasets, or private information of individuals/organizations. Despite significant advancements, RAG faces challenges such as retrieval efficiency, context awareness, and relevance of generated answers to the question asked. In this study, we compared the performance of Baseline RAG or vanilla RAG and advanced RAG architectures like MultiQuery Retrieval, Contextual Compression Retrieval, and Sentence Window Retrieval. Additionally, we fine-tuned an LLM model (Llama2) and analyzed whether there is a change in performance across all the RAG architectures, including Baseline RAG and the advanced RAG architectures mentioned. To measure performance, we employed the RAGAs framework and used metrics such as Faithfulness, Answer Relevancy, Context Recall, Context Precision, and Answer Correctness. Our findings indicate that while CHAPTER
Retrieval Augmented Generation (RAG) is a technique in natural language processing that enhances ... more Retrieval Augmented Generation (RAG) is a technique in natural language processing that enhances generative AI models' accuracy and reliability by incorporating facts fetched from external sources like the internet, databases, open-domain datasets, or private information of individuals/organizations. Despite significant advancements, RAG faces challenges such as retrieval efficiency, context awareness, and relevance of generated answers to the question asked. In this study, we compared the performance of Baseline RAG or vanilla RAG and advanced RAG architectures like MultiQuery Retrieval, Contextual Compression Retrieval, and Sentence Window Retrieval. Additionally, we fine-tuned an LLM model (Llama2) and analyzed whether there is a change in performance across all the RAG architectures, including Baseline RAG and the advanced RAG architectures mentioned. To measure performance, we employed the RAGAs framework and used metrics such as Faithfulness, Answer Relevancy, Context Recall, Context Precision, and Answer Correctness. Our findings indicate that while CHAPTER
Uploads
Papers by Rudranil Gupta