From 9361292c429c0ba3400c31c7fa5d5e3d3cb6ab47 Mon Sep 17 00:00:00 2001
From: Joseph Redmon <pjreddie@gmail.com>
Date: Tue, 19 Jul 2016 21:50:01 +0000
Subject: [PATCH] updates
---
src/network_kernels.cu | 41 +++++++++++++++++++++++++++++++++++++++--
1 files changed, 39 insertions(+), 2 deletions(-)
diff --git a/src/network_kernels.cu b/src/network_kernels.cu
index 0b50647..e1d4129 100644
--- a/src/network_kernels.cu
+++ b/src/network_kernels.cu
@@ -11,17 +11,22 @@
#include "image.h"
#include "data.h"
#include "utils.h"
-#include "params.h"
#include "parser.h"
#include "crop_layer.h"
#include "connected_layer.h"
+#include "rnn_layer.h"
+#include "gru_layer.h"
+#include "crnn_layer.h"
#include "detection_layer.h"
+#include "region_layer.h"
#include "convolutional_layer.h"
+#include "activation_layer.h"
#include "deconvolutional_layer.h"
#include "maxpool_layer.h"
#include "avgpool_layer.h"
#include "normalization_layer.h"
+#include "batchnorm_layer.h"
#include "cost_layer.h"
#include "local_layer.h"
#include "softmax_layer.h"
@@ -37,6 +42,7 @@
void forward_network_gpu(network net, network_state state)
{
+ state.workspace = net.workspace;
int i;
for(i = 0; i < net.n; ++i){
state.index = i;
@@ -48,12 +54,22 @@
forward_convolutional_layer_gpu(l, state);
} else if(l.type == DECONVOLUTIONAL){
forward_deconvolutional_layer_gpu(l, state);
+ } else if(l.type == ACTIVE){
+ forward_activation_layer_gpu(l, state);
} else if(l.type == LOCAL){
forward_local_layer_gpu(l, state);
} else if(l.type == DETECTION){
forward_detection_layer_gpu(l, state);
+ } else if(l.type == REGION){
+ forward_region_layer_gpu(l, state);
} else if(l.type == CONNECTED){
forward_connected_layer_gpu(l, state);
+ } else if(l.type == RNN){
+ forward_rnn_layer_gpu(l, state);
+ } else if(l.type == GRU){
+ forward_gru_layer_gpu(l, state);
+ } else if(l.type == CRNN){
+ forward_crnn_layer_gpu(l, state);
} else if(l.type == CROP){
forward_crop_layer_gpu(l, state);
} else if(l.type == COST){
@@ -62,6 +78,8 @@
forward_softmax_layer_gpu(l, state);
} else if(l.type == NORMALIZATION){
forward_normalization_layer_gpu(l, state);
+ } else if(l.type == BATCHNORM){
+ forward_batchnorm_layer_gpu(l, state);
} else if(l.type == MAXPOOL){
forward_maxpool_layer_gpu(l, state);
} else if(l.type == AVGPOOL){
@@ -79,6 +97,7 @@
void backward_network_gpu(network net, network_state state)
{
+ state.workspace = net.workspace;
int i;
float * original_input = state.input;
float * original_delta = state.delta;
@@ -97,6 +116,8 @@
backward_convolutional_layer_gpu(l, state);
} else if(l.type == DECONVOLUTIONAL){
backward_deconvolutional_layer_gpu(l, state);
+ } else if(l.type == ACTIVE){
+ backward_activation_layer_gpu(l, state);
} else if(l.type == LOCAL){
backward_local_layer_gpu(l, state);
} else if(l.type == MAXPOOL){
@@ -107,12 +128,22 @@
backward_dropout_layer_gpu(l, state);
} else if(l.type == DETECTION){
backward_detection_layer_gpu(l, state);
+ } else if(l.type == REGION){
+ backward_region_layer_gpu(l, state);
} else if(l.type == NORMALIZATION){
backward_normalization_layer_gpu(l, state);
+ } else if(l.type == BATCHNORM){
+ backward_batchnorm_layer_gpu(l, state);
} else if(l.type == SOFTMAX){
if(i != 0) backward_softmax_layer_gpu(l, state);
} else if(l.type == CONNECTED){
backward_connected_layer_gpu(l, state);
+ } else if(l.type == RNN){
+ backward_rnn_layer_gpu(l, state);
+ } else if(l.type == GRU){
+ backward_gru_layer_gpu(l, state);
+ } else if(l.type == CRNN){
+ backward_crnn_layer_gpu(l, state);
} else if(l.type == COST){
backward_cost_layer_gpu(l, state);
} else if(l.type == ROUTE){
@@ -136,6 +167,12 @@
update_deconvolutional_layer_gpu(l, rate, net.momentum, net.decay);
} else if(l.type == CONNECTED){
update_connected_layer_gpu(l, update_batch, rate, net.momentum, net.decay);
+ } else if(l.type == GRU){
+ update_gru_layer_gpu(l, update_batch, rate, net.momentum, net.decay);
+ } else if(l.type == RNN){
+ update_rnn_layer_gpu(l, update_batch, rate, net.momentum, net.decay);
+ } else if(l.type == CRNN){
+ update_crnn_layer_gpu(l, update_batch, rate, net.momentum, net.decay);
} else if(l.type == LOCAL){
update_local_layer_gpu(l, update_batch, rate, net.momentum, net.decay);
}
@@ -149,7 +186,7 @@
state.net = net;
int x_size = get_network_input_size(net)*net.batch;
int y_size = get_network_output_size(net)*net.batch;
- if(net.layers[net.n-1].type == DETECTION) y_size = net.layers[net.n-1].truths*net.batch;
+ if(net.layers[net.n-1].truths) y_size = net.layers[net.n-1].truths*net.batch;
if(!*net.input_gpu){
*net.input_gpu = cuda_make_array(x, x_size);
*net.truth_gpu = cuda_make_array(y, y_size);
--
Gitblit v1.10.0