You.com launched YouRetriever, the simplest interface to the You.com Search API. The You.com Search API was developed with retrieval augmented generation (RAG) applications in mind by LLM for LLM. They achieve this by testing our API with various data sets to set standards for LLM efficiency in the RAG-QA environment. They provided a detailed analysis of the differences and similarities between the You.com Search API and the Google Search API. They provided a framework for evaluating LLMs in a RAG-QA environment. They also used the RetrievalQA string to evaluate how well their retrievers did in Hotpot QA. A Hotpot data set includes a query, a response, and its context. When “distractor” mode is used, in which the LLM must avoid being misled by intentionally false language, the context may change about the question/answer. One of the tests involved replacing the original context of the dataset with the text fragments returned by the search APIs. Since APIs must search the entire Internet for the desired information rather than relying solely on the list of snippets provided in the data set, the Internet serves as distractor text in this case. When testing the effectiveness of search APIs in conjunction with an LLM, they subject systems to what they call the “web distractor” scenario.
When possible, it returns longer chunks of information, and soon you’ll be able to choose how much text you want returned, from a single sample to a full page. There are 27 results for “great Keith” when using the default parameters and some documents have some content. For LLMs working in a RAG-QA environment, this makes our Search API particularly useful.
They ran their tests on the HotPotQA dataset. They use the dataset library to retrieve this information from the Huggingface dataset. Here, they use full wiki instead of distractor, but as said above, they will generate their context using the search APIs.
Visit https://documentation.you.com/openai-language-model-integration for further setup instructions.
You.com will soon publish a larger search study, so stay tuned for more information. Anyone interested in becoming an early access partner should write to [email protected] with information about yourself, your use case, and the number of calls you expect to make each day.
Review the Blog and Project page. All credit for this research goes to the researchers of this project. Also, don’t forget to join. our 32k+ ML SubReddit, Facebook community of more than 40,000 people, Discord channel, and Electronic newsletterwhere we share the latest news on ai research, interesting ai projects and more.
If you like our work, you’ll love our newsletter.
we are also in Telegram and WhatsApp.
Dhanshree Shenwai is a Computer Science Engineer and has good experience in FinTech companies covering Finance, Cards & Payments and Banking with a keen interest in ai applications. He is excited to explore new technologies and advancements in today’s evolving world that makes life easier for everyone.
<!– ai CONTENT END 2 –>
