US 12,007,867 B1
Device and method for providing benchmark result of artificial intelligence based model
Sanggeon Park, Seoul (KR); Wonjin Shin, Seoul (KR); and Jimin Lee, Seoul (KR)
Assigned to NOTA, INC., Daejeon (KR)
Filed by NOTA, INC., Daejeon (KR)
Filed on Jun. 30, 2023, as Appl. No. 18/346,075.
Claims priority of application No. 10-2023-0026149 (KR), filed on Feb. 27, 2023.
Int. Cl. G06F 11/34 (2006.01)
CPC G06F 11/3428 (2013.01) 18 Claims
OG exemplary drawing
 
1. A method performed by a first computing device performing a benchmark, comprising:
receiving, from a second computing device comprising a plurality of modules which perform different operations related to an artificial intelligence-based model, module identification information indicating which module among the plurality of modules of the second computing device triggers a benchmark operation of the first computing device; and
providing, to the second computing device, a benchmark result based on the module identification information, and
wherein providing, to the second computing device, a benchmark result based on the module identification information comprises:
if the module identification information indicates a first module, i) providing performance information generated by executing an inputted model at one or more nodes, to the second computing device, or ii) providing, to the second computing device, a benchmark result for determining a target node at which a converted training model or a training model corresponding to an inputted dataset is to be executed, by executing the converted training model or the training model corresponding to the inputted dataset at the one or more nodes; and
if the module identification information indicates a second module, i) providing performance information corresponding to an entirety of the inputted model and providing performance information in units of blocks of the inputted model to the second computing device, by executing the inputted model at one or more nodes, wherein the performance information in units of blocks of the inputted model is used to optimize or compress in units of blocks of the inputted model, or ii) providing, to the second computing device, a benchmark result comprising configuration data used for generating a compressed model corresponding to the inputted model, by executing the compressed model corresponding to the inputted model at the one or more nodes.