From 1edcf73a73d2007afc61289245763f5cf0c29e10 Mon Sep 17 00:00:00 2001
From: Joseph Redmon <pjreddie@gmail.com>
Date: Thu, 04 Dec 2014 07:20:29 +0000
Subject: [PATCH] Detection good, split up col images
---
src/connected_layer.c | 25 +++++++++++++++++--------
1 files changed, 17 insertions(+), 8 deletions(-)
diff --git a/src/connected_layer.c b/src/connected_layer.c
index ba83dc3..05d4a03 100644
--- a/src/connected_layer.c
+++ b/src/connected_layer.c
@@ -9,7 +9,6 @@
connected_layer *make_connected_layer(int batch, int inputs, int outputs, ACTIVATION activation, float learning_rate, float momentum, float decay)
{
- fprintf(stderr, "Connected Layer: %d inputs, %d outputs\n", inputs, outputs);
int i;
connected_layer *layer = calloc(1, sizeof(connected_layer));
@@ -28,7 +27,7 @@
//layer->weight_adapt = calloc(inputs*outputs, sizeof(float));
layer->weights = calloc(inputs*outputs, sizeof(float));
float scale = 1./inputs;
- scale = .05;
+ scale = .01;
for(i = 0; i < inputs*outputs; ++i)
layer->weights[i] = scale*2*(rand_uniform()-.5);
@@ -51,6 +50,7 @@
layer->delta_cl = cl_make_array(layer->delta, outputs*batch);
#endif
layer->activation = activation;
+ fprintf(stderr, "Connected Layer: %d inputs, %d outputs\n", inputs, outputs);
return layer;
}
@@ -108,6 +108,18 @@
#ifdef GPU
+void pull_connected_layer(connected_layer layer)
+{
+ cl_read_array(layer.weights_cl, layer.weights, layer.inputs*layer.outputs);
+ cl_read_array(layer.biases_cl, layer.biases, layer.outputs);
+}
+
+void push_connected_layer(connected_layer layer)
+{
+ cl_write_array(layer.weights_cl, layer.weights, layer.inputs*layer.outputs);
+ cl_write_array(layer.biases_cl, layer.biases, layer.outputs);
+}
+
void update_connected_layer_gpu(connected_layer layer)
{
axpy_ongpu(layer.outputs, layer.learning_rate, layer.bias_updates_cl, 1, layer.biases_cl, 1);
@@ -116,15 +128,14 @@
scal_ongpu(layer.inputs*layer.outputs, 1.-layer.learning_rate*layer.decay, layer.weights_cl, 1);
axpy_ongpu(layer.inputs*layer.outputs, layer.learning_rate, layer.weight_updates_cl, 1, layer.weights_cl, 1);
scal_ongpu(layer.inputs*layer.outputs, layer.momentum, layer.weight_updates_cl, 1);
+ pull_connected_layer(layer);
}
void forward_connected_layer_gpu(connected_layer layer, cl_mem input)
{
int i;
for(i = 0; i < layer.batch; ++i){
- cl_mem sub = cl_sub_array(layer.output_cl, i*layer.outputs, layer.outputs);
- copy_ongpu(layer.outputs, layer.biases_cl, 1, sub, 1);
- clReleaseMemObject(sub);
+ copy_ongpu_offset(layer.outputs, layer.biases_cl, 0, 1, layer.output_cl, i*layer.outputs, 1);
}
int m = layer.batch;
int k = layer.inputs;
@@ -141,9 +152,7 @@
int i;
gradient_array_ongpu(layer.output_cl, layer.outputs*layer.batch, layer.activation, layer.delta_cl);
for(i = 0; i < layer.batch; ++i){
- cl_mem sub = cl_sub_array(layer.delta_cl, i*layer.outputs, layer.outputs);
- axpy_ongpu(layer.outputs, 1, sub, 1, layer.bias_updates_cl, 1);
- clReleaseMemObject(sub);
+ axpy_ongpu_offset(layer.outputs, 1, layer.delta_cl, i*layer.outputs, 1, layer.bias_updates_cl, 0, 1);
}
int m = layer.inputs;
int k = layer.batch;
--
Gitblit v1.10.0