Table of Contents
Online LLMs, or large language models, are powerful tools that can generate natural language responses based on a given input. However, most LLMs have limitations, such as being outdated, inaccurate, or irrelevant. Perplexity AI, a new AI startup, aims to change that with its innovative online LLMs that can tap into the latest information from the internet and deliver helpful and factual answers to any question.
Perplexity AI’s LLMs: pplx-7b-online and pplx-70b-online
Perplexity AI has introduced two new online LLMs: pplx-7b-online and pplx-70b-online, which are publicly available via pplx-api, a first-of-its-kind API. These LLMs can also be accessed via Perplexity Labs, a playground for experimenting with different LLMs.
Excited to announce that pplx-api is coming out of beta and moving to usage based pricing, along with the first-ever live LLM APIs that are grounded with web search data and have no knowledge cutoff! https://t.co/VYXIjqdLy9
— Aravind Srinivas (@AravSrinivas) November 29, 2023
What makes these online LLMs unique is that they can connect to the web and extract the most relevant and up-to-date information for any query. For instance, these models can provide current information on sports scores, stock prices, or the latest developments from Google News. They can also answer questions across various academic subjects, such as history, science, or math.
How Perplexity AI’s online LLMs work
Perplexity AI’s online LLMs are based on open-sourced models, specifically the mistral-7b and llama2-70b base models, which are fine-tuned to achieve enhanced performance. The LLMs also use an in-house search technology that facilitates the extraction of the most relevant and up-to-date information from the web. The snippets, or website excerpts, obtained through this process play a pivotal role in informing the responses generated by the models.
Perplexity AI’s online LLMs vs. other LLMs
Perplexity AI claims that its online LLMs not only match but also surpass the capabilities of leading proprietary LLMs, such as GPT-3.5. The company has conducted initial tests that show that its LLMs are more robust, helpful, and knowledgeable than other LLMs. These tests cover benchmarks evaluating various aspects of the LLMs, such as accuracy, relevance, coherence, and diversity.
One of the main challenges of LLMs is that they often struggle to provide up-to-date information and are susceptible to generating inaccurate statements, commonly referred to as “hallucinations”. Perplexity AI’s online LLMs aim to overcome these hurdles by not only offering helpful and factual responses but also ensuring the information is current.
Perplexity AI’s team and user base
Perplexity AI was co-founded by Aravind Srinivas, a former research scientist at OpenAI, and Denis Yarats, a former Meta research scientist, who now serves as the company’s CEO and CTO, respectively. The company’s user base has recently reached 15,000 paying customers, who use the online LLMs for various purposes, such as research, education, entertainment, and business.
What are some use cases of online LLMs?
Large language models, or LLMs, are powerful tools that can generate natural language responses based on input. LLMs, on the other hand, have a unique feature that makes them more useful and accurate: they can access real-time information from the internet and use it to answer complex questions.
Some examples of online LLM applications include:
- Content Generation: Online LLMs can generate new content for blogs, social media, and other digital platforms. They can start with existing content and generate new text related to it, or they can generate new content based on a set of keywords or other input.
- Chatbots and AI assistants: Chatbots and AI assistants created by online LLMs can interact with users in a natural and conversational manner. They can respond to questions, provide information, make suggestions, and carry out tasks.
- Marketing and advertising: Personalised marketing content, such as email campaigns and social media posts, can be generated by online LLMs. They can create content that is tailored to the user’s needs and interests by using their preferences, behavior, and history.
