首页|Patent Issued for Synthetic document generation using machine learning based lan guage model (USPTO 12159099)
Patent Issued for Synthetic document generation using machine learning based lan guage model (USPTO 12159099)
扫码查看
点击上方二维码区域，可以放大扫码查看
原文链接
NETL
NSTL
News editors obtained the following quote from the background information suppli ed by the inventors:“The disclosure relates in general to generative artificial intelligence, and more specifically to documentgeneration using machine learni ng based language models.“Generative artificial intelligence (AI) techniques are being increasingly used for automatically generatingnatural language descriptions in various contexts. These techniques include large language modelssuch as Generative Pretrained Tra nsformer (GPT) models that are trained on large corpuses of documentsincluding books, knowledge sources such as Wikipedia™, and other data obtained by crawling the internet.Such language models are trained to generate natural language tex t as well as structured or semi structuredtext such as programs using various p rograming languages such as Python, Java, or data represented using JSON (JavaSc ript Object Notation). Large language models often exhibit hallucinations by gen eratingfalse or misleading information. As a result, natural language text gene rated by such large language modelis inadequate for domains where highly accura te information is required. Furthermore, large languagemodels have very large n umber of parameters and are therefore difficult to retrain frequently. Accordingly, the knowledge of the large language model may get outdated for fields where information keeps changingand accuracy of results requires use of latest inform ation.
Artificial IntelligenceBench Iq Inc.BusinessCyborgsEmerging TechnologiesMachine Learning
NETL
NSTL
