Accelerated by another 5% using FP16/32 Batch-norm for Tensor Cores.

AlexeyAB

2018-04-16 9bae70b22549b68f5cdeece8b6c3b3de00c22714

 src/captcha.c

@@ -28,7 +28,6 @@

void train_captcha(char *cfgfile, char *weightfile)
{
    data_seed = time(0);
    srand(time(0));
    float avg_loss = -1;
    char *base = basecfg(cfgfile);