QUALCOMM Incorporated (20240354345). SPECULATIVE DECODING IN AUTOREGRESSIVE GENERATIVE ARTIFICIAL INTELLIGENCE MODELS simplified abstract

From WikiPatents
Revision as of 06:10, 25 October 2024 by Wikipatents (talk | contribs) (Creating a new page)
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to navigation Jump to search

SPECULATIVE DECODING IN AUTOREGRESSIVE GENERATIVE ARTIFICIAL INTELLIGENCE MODELS

Organization Name

QUALCOMM Incorporated

Inventor(s)

Christopher Lott of San Diego CA (US)

Mingu Lee of San Diego CA (US)

Wonseok Jeon of San Diego CA (US)

Roland Memisevic of Toronto (CA)

SPECULATIVE DECODING IN AUTOREGRESSIVE GENERATIVE ARTIFICIAL INTELLIGENCE MODELS - A simplified explanation of the abstract

This abstract first appeared for US patent application 20240354345 titled 'SPECULATIVE DECODING IN AUTOREGRESSIVE GENERATIVE ARTIFICIAL INTELLIGENCE MODELS

The present disclosure outlines techniques and apparatus for generating responses to queries in a generative artificial intelligence model.

  • Receiving multiple sets of tokens created from an input prompt and a first AI model, with each set corresponding to a potential response.
  • Using a second AI model to select a set of tokens from the received sets through recursive adjustment of a target distribution.
  • Outputting the chosen set of tokens as the response to the input prompt.
    • Key Features and Innovation:**
  • Utilizes multiple AI models to generate responses to queries.
  • Recursive adjustment of target distribution for accurate token selection.
  • Enhances the capabilities of generative artificial intelligence models.
    • Potential Applications:**
  • Chatbots and virtual assistants.
  • Content generation for social media or marketing.
  • Automated customer service responses.
    • Problems Solved:**
  • Improves the accuracy and relevance of AI-generated responses.
  • Streamlines the process of generating responses to queries.
  • Enhances the overall performance of generative AI models.
    • Benefits:**
  • Saves time and resources in generating responses.
  • Increases the efficiency of AI-powered systems.
  • Enhances user experience through more accurate and relevant responses.
    • Commercial Applications:**
  • "Response Generation in Generative AI Models" technology can be applied in customer service chatbots, content creation tools, and automated messaging systems, improving response accuracy and efficiency in various industries.
    • Questions about Response Generation in Generative AI Models:**

1. How does the recursive adjustment of the target distribution improve response generation accuracy? 2. What are the potential limitations of using multiple AI models for response generation?


Original Abstract Submitted

certain aspects of the present disclosure provide techniques and apparatus for generating a response to a query input in a generative artificial intelligence model. an example method generally includes receiving a plurality of sets of tokens generated based on an input prompt and a first generative artificial intelligence model, each set of tokens in the plurality of sets of tokens corresponding to a candidate response to the input prompt; selecting, using a second generative artificial intelligence model and recursive adjustment of a target distribution associated with the received plurality of sets of tokens, a set of tokens from the plurality of sets of tokens; and outputting the selected set of tokens as a response to the input prompt.

Retrieved from "http://wikipatents.org/index.php?title=QUALCOMM_Incorporated_(20240354345)._SPECULATIVE_DECODING_IN_AUTOREGRESSIVE_GENERATIVE_ARTIFICIAL_INTELLIGENCE_MODELS_simplified_abstract&oldid=257793"