Cuda error( core dump )

[xuanson97hg]

when I train with my custom data set if I keep default config.num_classes and config.num_image it work and if I set config.num_classes and config.num_image by my classes and number image of my dataset it error with frame and core dump

[fdbtrs]

how many identities dose your training dataset has? you may consider a smaller batch size to fit your model in the GPU

[xuanson97hg]

how many identities dose your training dataset has? you may consider a smaller batch size to fit your model in the GPU

i try with smaller batch then code run fill minute and core dump again . my dataset has 12k image

[fdbtrs]

Resnet 100 with 512 batch requires approximately 4 gpus each of 16GB. You need to set the batch size based on number of gpu and gpu memory