| | |
| | | #include "cost_layer.h" |
| | | #include "utils.h" |
| | | #include "mini_blas.h" |
| | | #include "cuda.h" |
| | | #include "blas.h" |
| | | #include <math.h> |
| | | #include <string.h> |
| | | #include <stdlib.h> |
| | | #include <stdio.h> |
| | | |
| | | cost_layer *make_cost_layer(int batch, int inputs) |
| | | COST_TYPE get_cost_type(char *s) |
| | | { |
| | | if (strcmp(s, "sse")==0) return SSE; |
| | | if (strcmp(s, "masked")==0) return MASKED; |
| | | fprintf(stderr, "Couldn't find activation function %s, going with SSE\n", s); |
| | | return SSE; |
| | | } |
| | | |
| | | char *get_cost_string(COST_TYPE a) |
| | | { |
| | | switch(a){ |
| | | case SSE: |
| | | return "sse"; |
| | | case MASKED: |
| | | return "masked"; |
| | | } |
| | | return "sse"; |
| | | } |
| | | |
| | | cost_layer make_cost_layer(int batch, int inputs, COST_TYPE cost_type, float scale) |
| | | { |
| | | fprintf(stderr, "Cost Layer: %d inputs\n", inputs); |
| | | cost_layer *layer = calloc(1, sizeof(cost_layer)); |
| | | layer->batch = batch; |
| | | layer->inputs = inputs; |
| | | layer->delta = calloc(inputs*batch, sizeof(float)); |
| | | layer->output = calloc(1, sizeof(float)); |
| | | cost_layer l = {0}; |
| | | l.type = COST; |
| | | |
| | | l.scale = scale; |
| | | l.batch = batch; |
| | | l.inputs = inputs; |
| | | l.outputs = inputs; |
| | | l.cost_type = cost_type; |
| | | l.delta = calloc(inputs*batch, sizeof(float)); |
| | | l.output = calloc(1, sizeof(float)); |
| | | #ifdef GPU |
| | | layer->delta_cl = cl_make_array(layer->delta, inputs*batch); |
| | | l.delta_gpu = cuda_make_array(l.delta, inputs*batch); |
| | | #endif |
| | | return layer; |
| | | return l; |
| | | } |
| | | |
| | | void forward_cost_layer(cost_layer layer, float *input, float *truth) |
| | | void resize_cost_layer(cost_layer *l, int inputs) |
| | | { |
| | | if (!truth) return; |
| | | copy_cpu(layer.batch*layer.inputs, truth, 1, layer.delta, 1); |
| | | axpy_cpu(layer.batch*layer.inputs, -1, input, 1, layer.delta, 1); |
| | | *(layer.output) = dot_cpu(layer.batch*layer.inputs, layer.delta, 1, layer.delta, 1); |
| | | l->inputs = inputs; |
| | | l->outputs = inputs; |
| | | l->delta = realloc(l->delta, inputs*l->batch*sizeof(float)); |
| | | #ifdef GPU |
| | | cuda_free(l->delta_gpu); |
| | | l->delta_gpu = cuda_make_array(l->delta, inputs*l->batch); |
| | | #endif |
| | | } |
| | | |
| | | void backward_cost_layer(const cost_layer layer, float *input, float *delta) |
| | | void forward_cost_layer(cost_layer l, network_state state) |
| | | { |
| | | copy_cpu(layer.batch*layer.inputs, layer.delta, 1, delta, 1); |
| | | if (!state.truth) return; |
| | | if(l.cost_type == MASKED){ |
| | | int i; |
| | | for(i = 0; i < l.batch*l.inputs; ++i){ |
| | | if(state.truth[i] == SECRET_NUM) state.input[i] = SECRET_NUM; |
| | | } |
| | | } |
| | | copy_cpu(l.batch*l.inputs, state.truth, 1, l.delta, 1); |
| | | axpy_cpu(l.batch*l.inputs, -1, state.input, 1, l.delta, 1); |
| | | *(l.output) = dot_cpu(l.batch*l.inputs, l.delta, 1, l.delta, 1); |
| | | //printf("cost: %f\n", *l.output); |
| | | } |
| | | |
| | | void backward_cost_layer(const cost_layer l, network_state state) |
| | | { |
| | | axpy_cpu(l.batch*l.inputs, l.scale, l.delta, 1, state.delta, 1); |
| | | } |
| | | |
| | | #ifdef GPU |
| | | void forward_cost_layer_gpu(cost_layer layer, cl_mem input, cl_mem truth) |
| | | |
| | | void pull_cost_layer(cost_layer l) |
| | | { |
| | | if (!truth) return; |
| | | |
| | | copy_ongpu(layer.batch*layer.inputs, truth, 1, layer.delta_cl, 1); |
| | | axpy_ongpu(layer.batch*layer.inputs, -1, input, 1, layer.delta_cl, 1); |
| | | |
| | | cl_read_array(layer.delta_cl, layer.delta, layer.batch*layer.inputs); |
| | | *(layer.output) = dot_cpu(layer.batch*layer.inputs, layer.delta, 1, layer.delta, 1); |
| | | //printf("%f\n", *layer.output); |
| | | cuda_pull_array(l.delta_gpu, l.delta, l.batch*l.inputs); |
| | | } |
| | | |
| | | void backward_cost_layer_gpu(const cost_layer layer, cl_mem input, cl_mem delta) |
| | | void push_cost_layer(cost_layer l) |
| | | { |
| | | copy_ongpu(layer.batch*layer.inputs, layer.delta_cl, 1, delta, 1); |
| | | cuda_push_array(l.delta_gpu, l.delta, l.batch*l.inputs); |
| | | } |
| | | |
| | | void forward_cost_layer_gpu(cost_layer l, network_state state) |
| | | { |
| | | if (!state.truth) return; |
| | | if (l.cost_type == MASKED) { |
| | | mask_ongpu(l.batch*l.inputs, state.input, SECRET_NUM, state.truth); |
| | | } |
| | | |
| | | copy_ongpu(l.batch*l.inputs, state.truth, 1, l.delta_gpu, 1); |
| | | axpy_ongpu(l.batch*l.inputs, -1, state.input, 1, l.delta_gpu, 1); |
| | | |
| | | cuda_pull_array(l.delta_gpu, l.delta, l.batch*l.inputs); |
| | | *(l.output) = dot_cpu(l.batch*l.inputs, l.delta, 1, l.delta, 1); |
| | | } |
| | | |
| | | void backward_cost_layer_gpu(const cost_layer l, network_state state) |
| | | { |
| | | axpy_ongpu(l.batch*l.inputs, l.scale, l.delta_gpu, 1, state.delta, 1); |
| | | } |
| | | #endif |
| | | |