Nvidia patents method to clean up data used to train AI models
Nvidia describes a system that checks the quality of data pairs used to train artificial intelligence. It uses multiple models to score how well a question matches its answer, then picks the best model to filter out low-quality examples from a larger dataset.
In the filing’s own words
In various examples, a system can include one or more processors to determine, using a plurality of retriever models, a quality score of each of a plurality of queries in a first dataset, determine, using a plurality of embedding models, a metric indicative of a relationship between each of the plurality of queries and the set of text information of the first dataset, and select one of the plurality of embedding models as a filter model based at least in part on correlation between the metrics outputted by the plurality of embedding models and the quality scores for the plurality of queries, the filter model applied to filter a plurality of queries and a plurality of sets of text information in a second dataset.
What exactly they claim (claim 1)
1. A system, comprising one or more processors to: determine a quality score of at least one query in a first dataset based at least in part on a number of a plurality of retriever models that retrieved at least one set of data corresponding to the at least one query in the first dataset; determine, using a plurality of embedding models, a plurality of metrics indicative of a relationship between the at least one query and the at least one set of data of the first dataset; and select one of the plurality of embedding models as a filter model based at least in part on a correlation between at least one of the plurality of metrics and the quality score; apply the filter model to filter a plurality of queries and a plurality of sets of data of a second dataset to form a filtered second dataset; and provide the filtered second dataset to update a machine learning model.
Facts
- Company
- Nvidia (NVIDIA Corporation)
- Official title
- FILTERS FOR QUALITY CONTROL OF SYNTHETICALLY GENERATED DATA
- Publication number
- US20260267835A1
- Application number
- 19662892
- Filed
- Apr 29, 2026
- Published
- Sep 10, 2026
- Inventors
- 3
- Claims
- 20
Classified under
G06F 16/215 Physics & computingG06F 16/2458 Physics & computingG06F 16/3347 Physics & computingG06F 16/90344 Physics & computingG06F 18/217 Physics & computingG06F 30/27 Physics & computingMore from Nvidia in AI & software
- Nvidia patents AI coach that analyzes your gameplay to suggest strategies Oct 13, 2026
- Nvidia patents AI system that writes and voices character conversations Sep 10, 2026
- Nvidia patents system that uses AI to find bugs in game videos Sep 17, 2026
- Nvidia patents AI method to make digital characters move realistically Sep 8, 2026
- Nvidia patents AI system that builds image generators from text prompts Sep 15, 2026
Summaries are simplified from the filing’s own abstract and first claim. A patent filing is not a product announcement.
Patent data and drawings: U.S. Patent and Trademark Office (USPTO). Nebula is not affiliated with or endorsed by the USPTO.