Joseph Redmon
2016-09-01 8f1b4e0962857d402f9d017fcbf387ef0eceb7c4
src/normalization_layer.c
@@ -22,10 +22,10 @@
    layer.inputs = w*h*c;
    layer.outputs = layer.inputs;
    #ifdef GPU
    layer.output_gpu =  cuda_make_array(0, h * w * c * batch);
    layer.delta_gpu =   cuda_make_array(0, h * w * c * batch);
    layer.squared_gpu = cuda_make_array(0, h * w * c * batch);
    layer.norms_gpu =   cuda_make_array(0, h * w * c * batch);
    layer.output_gpu =  cuda_make_array(layer.output, h * w * c * batch);
    layer.delta_gpu =   cuda_make_array(layer.delta, h * w * c * batch);
    layer.squared_gpu = cuda_make_array(layer.squared, h * w * c * batch);
    layer.norms_gpu =   cuda_make_array(layer.norms, h * w * c * batch);
    #endif
    return layer;
}
@@ -40,19 +40,19 @@
    layer->out_w = w;
    layer->inputs = w*h*c;
    layer->outputs = layer->inputs;
    layer->output = realloc(layer->output, h * w * layer->c * layer->batch * sizeof(float));
    layer->delta = realloc(layer->delta, h * w * layer->c * layer->batch * sizeof(float));
    layer->squared = realloc(layer->squared, h * w * layer->c * layer->batch * sizeof(float));
    layer->norms = realloc(layer->norms, h * w * layer->c * layer->batch * sizeof(float));
    layer->output = realloc(layer->output, h * w * c * batch * sizeof(float));
    layer->delta = realloc(layer->delta, h * w * c * batch * sizeof(float));
    layer->squared = realloc(layer->squared, h * w * c * batch * sizeof(float));
    layer->norms = realloc(layer->norms, h * w * c * batch * sizeof(float));
#ifdef GPU
    cuda_free(layer->output_gpu);
    cuda_free(layer->delta_gpu); 
    cuda_free(layer->squared_gpu); 
    cuda_free(layer->norms_gpu);   
    layer->output_gpu =  cuda_make_array(0, h * w * c * batch);
    layer->delta_gpu =   cuda_make_array(0, h * w * c * batch);
    layer->squared_gpu = cuda_make_array(0, h * w * c * batch);
    layer->norms_gpu =   cuda_make_array(0, h * w * c * batch);
    layer->output_gpu =  cuda_make_array(layer->output, h * w * c * batch);
    layer->delta_gpu =   cuda_make_array(layer->delta, h * w * c * batch);
    layer->squared_gpu = cuda_make_array(layer->squared, h * w * c * batch);
    layer->norms_gpu =   cuda_make_array(layer->norms, h * w * c * batch);
#endif
}
@@ -90,6 +90,7 @@
void backward_normalization_layer(const layer layer, network_state state)
{
    // TODO This is approximate ;-)
    // Also this should add in to delta instead of overwritting.
    int w = layer.w;
    int h = layer.h;