Chinese Institutions Use U.S. AI Model Outputs to Train Domestic Models

Written by

in

Chinese institutions are using U.S. AI model outputs to train smaller, lower-cost domestic models in response to restrictions on access to advanced chips. Reuters reviewed more than 300 academic papers, patents, corporate disclosures and government documents, finding that the practice has moved from academic experimentation into industrial applications. Relevant policies support using distillation for manufacturing deployments, but the approach cannot replicate all the capabilities of leading models or eliminate the need for advanced computing power.

Source:RAPPLER · Read original

Comments

Leave a Reply

Your email address will not be published. Required fields are marked *