From 7add11150954879dae4daad4e4da549b4b13bca6 Mon Sep 17 00:00:00 2001
From: Joseph Redmon <pjreddie@gmail.com>
Date: Sat, 09 Aug 2014 15:16:37 +0000
Subject: [PATCH] maxpool fixed, good on mnist
---
src/connected_layer.c | 41 ++++++++++++++++++++++++-----------------
1 files changed, 24 insertions(+), 17 deletions(-)
diff --git a/src/connected_layer.c b/src/connected_layer.c
index 72cb3fb..834e629 100644
--- a/src/connected_layer.c
+++ b/src/connected_layer.c
@@ -7,15 +7,19 @@
#include <stdlib.h>
#include <string.h>
-connected_layer *make_connected_layer(int batch, int inputs, int outputs, float dropout, ACTIVATION activation)
+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));
+
+ layer->learning_rate = learning_rate;
+ layer->momentum = momentum;
+ layer->decay = decay;
+
layer->inputs = inputs;
layer->outputs = outputs;
layer->batch=batch;
- layer->dropout = dropout;
layer->output = calloc(batch*outputs, sizeof(float*));
layer->delta = calloc(batch*outputs, sizeof(float*));
@@ -25,8 +29,9 @@
layer->weight_momentum = calloc(inputs*outputs, sizeof(float));
layer->weights = calloc(inputs*outputs, sizeof(float));
float scale = 1./inputs;
+ scale = .05;
for(i = 0; i < inputs*outputs; ++i)
- layer->weights[i] = scale*(rand_uniform());
+ layer->weights[i] = scale*2*(rand_uniform()-.5);
layer->bias_updates = calloc(outputs, sizeof(float));
layer->bias_adapt = calloc(outputs, sizeof(float));
@@ -40,25 +45,27 @@
return layer;
}
-void update_connected_layer(connected_layer layer, float step, float momentum, float decay)
+void update_connected_layer(connected_layer layer)
{
int i;
for(i = 0; i < layer.outputs; ++i){
- layer.bias_momentum[i] = step*(layer.bias_updates[i]) + momentum*layer.bias_momentum[i];
+ layer.bias_momentum[i] = layer.learning_rate*(layer.bias_updates[i]) + layer.momentum*layer.bias_momentum[i];
layer.biases[i] += layer.bias_momentum[i];
}
for(i = 0; i < layer.outputs*layer.inputs; ++i){
- layer.weight_momentum[i] = step*(layer.weight_updates[i] - decay*layer.weights[i]) + momentum*layer.weight_momentum[i];
+ layer.weight_momentum[i] = layer.learning_rate*(layer.weight_updates[i] - layer.decay*layer.weights[i]) + layer.momentum*layer.weight_momentum[i];
layer.weights[i] += layer.weight_momentum[i];
}
memset(layer.bias_updates, 0, layer.outputs*sizeof(float));
memset(layer.weight_updates, 0, layer.outputs*layer.inputs*sizeof(float));
}
-void forward_connected_layer(connected_layer layer, float *input, int train)
+void forward_connected_layer(connected_layer layer, float *input)
{
- if(!train) layer.dropout = 0;
- memcpy(layer.output, layer.biases, layer.outputs*sizeof(float));
+ int i;
+ for(i = 0; i < layer.batch; ++i){
+ memcpy(layer.output+i*layer.outputs, layer.biases, layer.outputs*sizeof(float));
+ }
int m = layer.batch;
int k = layer.inputs;
int n = layer.outputs;
@@ -66,7 +73,7 @@
float *b = layer.weights;
float *c = layer.output;
gemm(0,0,m,n,k,1,a,k,b,n,1,c,n);
- activate_array(layer.output, layer.outputs*layer.batch, layer.activation, layer.dropout);
+ activate_array(layer.output, layer.outputs*layer.batch, layer.activation);
}
void backward_connected_layer(connected_layer layer, float *input, float *delta)
@@ -74,7 +81,7 @@
int i;
for(i = 0; i < layer.outputs*layer.batch; ++i){
layer.delta[i] *= gradient(layer.output[i], layer.activation);
- layer.bias_updates[i%layer.batch] += layer.delta[i];
+ layer.bias_updates[i%layer.outputs] += layer.delta[i];
}
int m = layer.inputs;
int k = layer.batch;
@@ -82,16 +89,16 @@
float *a = input;
float *b = layer.delta;
float *c = layer.weight_updates;
- gemm(0,0,m,n,k,1,a,k,b,n,1,c,n);
+ gemm(1,0,m,n,k,1,a,m,b,n,1,c,n);
- m = layer.inputs;
+ m = layer.batch;
k = layer.outputs;
- n = layer.batch;
+ n = layer.inputs;
- a = layer.weights;
- b = layer.delta;
+ a = layer.delta;
+ b = layer.weights;
c = delta;
- if(c) gemm(0,0,m,n,k,1,a,k,b,n,0,c,n);
+ if(c) gemm(0,1,m,n,k,1,a,k,b,k,0,c,n);
}
--
Gitblit v1.10.0