2017-06-02 06:31:13 +03:00
|
|
|
#include "darknet.h"
|
2016-01-28 23:30:38 +03:00
|
|
|
|
|
|
|
typedef struct {
|
|
|
|
float *x;
|
|
|
|
float *y;
|
|
|
|
} float_pair;
|
|
|
|
|
2016-05-12 23:36:11 +03:00
|
|
|
int *read_tokenized_data(char *filename, size_t *read)
|
|
|
|
{
|
|
|
|
size_t size = 512;
|
|
|
|
size_t count = 0;
|
|
|
|
FILE *fp = fopen(filename, "r");
|
|
|
|
int *d = calloc(size, sizeof(int));
|
|
|
|
int n, one;
|
|
|
|
one = fscanf(fp, "%d", &n);
|
|
|
|
while(one == 1){
|
|
|
|
++count;
|
|
|
|
if(count > size){
|
|
|
|
size = size*2;
|
|
|
|
d = realloc(d, size*sizeof(int));
|
|
|
|
}
|
|
|
|
d[count-1] = n;
|
|
|
|
one = fscanf(fp, "%d", &n);
|
|
|
|
}
|
|
|
|
fclose(fp);
|
|
|
|
d = realloc(d, count*sizeof(int));
|
|
|
|
*read = count;
|
|
|
|
return d;
|
|
|
|
}
|
|
|
|
|
|
|
|
char **read_tokens(char *filename, size_t *read)
|
|
|
|
{
|
|
|
|
size_t size = 512;
|
|
|
|
size_t count = 0;
|
|
|
|
FILE *fp = fopen(filename, "r");
|
|
|
|
char **d = calloc(size, sizeof(char *));
|
|
|
|
char *line;
|
|
|
|
while((line=fgetl(fp)) != 0){
|
|
|
|
++count;
|
|
|
|
if(count > size){
|
|
|
|
size = size*2;
|
|
|
|
d = realloc(d, size*sizeof(char *));
|
|
|
|
}
|
|
|
|
d[count-1] = line;
|
|
|
|
}
|
|
|
|
fclose(fp);
|
|
|
|
d = realloc(d, count*sizeof(char *));
|
|
|
|
*read = count;
|
|
|
|
return d;
|
|
|
|
}
|
|
|
|
|
|
|
|
float_pair get_rnn_token_data(int *tokens, size_t *offsets, int characters, size_t len, int batch, int steps)
|
|
|
|
{
|
|
|
|
float *x = calloc(batch * steps * characters, sizeof(float));
|
|
|
|
float *y = calloc(batch * steps * characters, sizeof(float));
|
|
|
|
int i,j;
|
|
|
|
for(i = 0; i < batch; ++i){
|
|
|
|
for(j = 0; j < steps; ++j){
|
|
|
|
int curr = tokens[(offsets[i])%len];
|
|
|
|
int next = tokens[(offsets[i] + 1)%len];
|
|
|
|
|
|
|
|
x[(j*batch + i)*characters + curr] = 1;
|
|
|
|
y[(j*batch + i)*characters + next] = 1;
|
|
|
|
|
|
|
|
offsets[i] = (offsets[i] + 1) % len;
|
|
|
|
|
|
|
|
if(curr >= characters || curr < 0 || next >= characters || next < 0){
|
|
|
|
error("Bad char");
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
float_pair p;
|
|
|
|
p.x = x;
|
|
|
|
p.y = y;
|
|
|
|
return p;
|
|
|
|
}
|
|
|
|
|
2016-05-07 02:25:16 +03:00
|
|
|
float_pair get_rnn_data(unsigned char *text, size_t *offsets, int characters, size_t len, int batch, int steps)
|
2016-01-28 23:30:38 +03:00
|
|
|
{
|
2016-02-05 11:15:12 +03:00
|
|
|
float *x = calloc(batch * steps * characters, sizeof(float));
|
|
|
|
float *y = calloc(batch * steps * characters, sizeof(float));
|
2016-01-28 23:30:38 +03:00
|
|
|
int i,j;
|
|
|
|
for(i = 0; i < batch; ++i){
|
|
|
|
for(j = 0; j < steps; ++j){
|
2016-05-07 02:25:16 +03:00
|
|
|
unsigned char curr = text[(offsets[i])%len];
|
|
|
|
unsigned char next = text[(offsets[i] + 1)%len];
|
|
|
|
|
|
|
|
x[(j*batch + i)*characters + curr] = 1;
|
|
|
|
y[(j*batch + i)*characters + next] = 1;
|
2016-02-05 11:15:12 +03:00
|
|
|
|
2016-05-07 02:25:16 +03:00
|
|
|
offsets[i] = (offsets[i] + 1) % len;
|
|
|
|
|
|
|
|
if(curr > 255 || curr <= 0 || next > 255 || next <= 0){
|
|
|
|
/*text[(index+j+2)%len] = 0;
|
|
|
|
printf("%ld %d %d %d %d\n", index, j, len, (int)text[index+j], (int)text[index+j+1]);
|
2016-02-05 11:15:12 +03:00
|
|
|
printf("%s", text+index);
|
2016-05-07 02:25:16 +03:00
|
|
|
*/
|
2016-02-05 11:15:12 +03:00
|
|
|
error("Bad char");
|
|
|
|
}
|
2016-01-28 23:30:38 +03:00
|
|
|
}
|
|
|
|
}
|
|
|
|
float_pair p;
|
|
|
|
p.x = x;
|
|
|
|
p.y = y;
|
|
|
|
return p;
|
|
|
|
}
|
|
|
|
|
2016-05-07 02:25:16 +03:00
|
|
|
void reset_rnn_state(network net, int b)
|
2016-01-28 23:30:38 +03:00
|
|
|
{
|
2016-05-07 02:25:16 +03:00
|
|
|
int i;
|
|
|
|
for (i = 0; i < net.n; ++i) {
|
|
|
|
#ifdef GPU
|
2016-05-12 23:36:11 +03:00
|
|
|
layer l = net.layers[i];
|
2016-05-07 02:25:16 +03:00
|
|
|
if(l.state_gpu){
|
|
|
|
fill_ongpu(l.outputs, 0, l.state_gpu + l.outputs*b, 1);
|
|
|
|
}
|
|
|
|
#endif
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2016-05-12 23:36:11 +03:00
|
|
|
void train_char_rnn(char *cfgfile, char *weightfile, char *filename, int clear, int tokenized)
|
2016-05-07 02:25:16 +03:00
|
|
|
{
|
|
|
|
srand(time(0));
|
2016-05-12 23:36:11 +03:00
|
|
|
unsigned char *text = 0;
|
|
|
|
int *tokens = 0;
|
|
|
|
size_t size;
|
|
|
|
if(tokenized){
|
|
|
|
tokens = read_tokenized_data(filename, &size);
|
|
|
|
} else {
|
|
|
|
FILE *fp = fopen(filename, "rb");
|
|
|
|
|
|
|
|
fseek(fp, 0, SEEK_END);
|
|
|
|
size = ftell(fp);
|
|
|
|
fseek(fp, 0, SEEK_SET);
|
|
|
|
|
|
|
|
text = calloc(size+1, sizeof(char));
|
|
|
|
fread(text, 1, size, fp);
|
|
|
|
fclose(fp);
|
|
|
|
}
|
2016-01-28 23:30:38 +03:00
|
|
|
|
|
|
|
char *backup_directory = "/home/pjreddie/backup/";
|
|
|
|
char *base = basecfg(cfgfile);
|
2016-02-01 02:52:03 +03:00
|
|
|
fprintf(stderr, "%s\n", base);
|
2016-01-28 23:30:38 +03:00
|
|
|
float avg_loss = -1;
|
|
|
|
network net = parse_network_cfg(cfgfile);
|
|
|
|
if(weightfile){
|
|
|
|
load_weights(&net, weightfile);
|
|
|
|
}
|
2016-05-07 02:25:16 +03:00
|
|
|
|
2017-04-10 05:56:42 +03:00
|
|
|
int inputs = net.inputs;
|
2017-06-01 07:06:35 +03:00
|
|
|
fprintf(stderr, "Learning Rate: %g, Momentum: %g, Decay: %g, Inputs: %d\n", net.learning_rate, net.momentum, net.decay, inputs);
|
2016-01-28 23:30:38 +03:00
|
|
|
int batch = net.batch;
|
|
|
|
int steps = net.time_steps;
|
2016-05-07 02:25:16 +03:00
|
|
|
if(clear) *net.seen = 0;
|
2016-01-28 23:30:38 +03:00
|
|
|
int i = (*net.seen)/net.batch;
|
|
|
|
|
2016-05-07 02:25:16 +03:00
|
|
|
int streams = batch/steps;
|
|
|
|
size_t *offsets = calloc(streams, sizeof(size_t));
|
|
|
|
int j;
|
|
|
|
for(j = 0; j < streams; ++j){
|
|
|
|
offsets[j] = rand_size_t()%size;
|
|
|
|
}
|
|
|
|
|
2016-01-28 23:30:38 +03:00
|
|
|
clock_t time;
|
|
|
|
while(get_current_batch(net) < net.max_batches){
|
|
|
|
i += 1;
|
|
|
|
time=clock();
|
2016-05-12 23:36:11 +03:00
|
|
|
float_pair p;
|
|
|
|
if(tokenized){
|
|
|
|
p = get_rnn_token_data(tokens, offsets, inputs, size, streams, steps);
|
|
|
|
}else{
|
|
|
|
p = get_rnn_data(text, offsets, inputs, size, streams, steps);
|
|
|
|
}
|
2016-01-28 23:30:38 +03:00
|
|
|
|
2017-04-10 05:56:42 +03:00
|
|
|
memcpy(net.input, p.x, net.inputs*net.batch);
|
|
|
|
memcpy(net.truth, p.y, net.truths*net.batch);
|
|
|
|
float loss = train_network_datum(net) / (batch);
|
2016-01-28 23:30:38 +03:00
|
|
|
free(p.x);
|
|
|
|
free(p.y);
|
|
|
|
if (avg_loss < 0) avg_loss = loss;
|
|
|
|
avg_loss = avg_loss*.9 + loss*.1;
|
|
|
|
|
2016-05-07 02:25:16 +03:00
|
|
|
int chars = get_current_batch(net)*batch;
|
|
|
|
fprintf(stderr, "%d: %f, %f avg, %f rate, %lf seconds, %f epochs\n", i, loss, avg_loss, get_current_rate(net), sec(clock()-time), (float) chars/size);
|
|
|
|
|
|
|
|
for(j = 0; j < streams; ++j){
|
|
|
|
//printf("%d\n", j);
|
|
|
|
if(rand()%10 == 0){
|
|
|
|
//fprintf(stderr, "Reset\n");
|
|
|
|
offsets[j] = rand_size_t()%size;
|
|
|
|
reset_rnn_state(net, j);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2016-06-10 00:52:09 +03:00
|
|
|
if(i%1000==0){
|
2016-01-28 23:30:38 +03:00
|
|
|
char buff[256];
|
|
|
|
sprintf(buff, "%s/%s_%d.weights", backup_directory, base, i);
|
|
|
|
save_weights(net, buff);
|
|
|
|
}
|
|
|
|
if(i%10==0){
|
|
|
|
char buff[256];
|
|
|
|
sprintf(buff, "%s/%s.backup", backup_directory, base);
|
|
|
|
save_weights(net, buff);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
char buff[256];
|
|
|
|
sprintf(buff, "%s/%s_final.weights", backup_directory, base);
|
|
|
|
save_weights(net, buff);
|
|
|
|
}
|
|
|
|
|
2016-05-12 23:36:11 +03:00
|
|
|
void print_symbol(int n, char **tokens){
|
|
|
|
if(tokens){
|
|
|
|
printf("%s ", tokens[n]);
|
|
|
|
} else {
|
|
|
|
printf("%c", n);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
void test_char_rnn(char *cfgfile, char *weightfile, int num, char *seed, float temp, int rseed, char *token_file)
|
2016-01-28 23:30:38 +03:00
|
|
|
{
|
2016-05-12 23:36:11 +03:00
|
|
|
char **tokens = 0;
|
|
|
|
if(token_file){
|
|
|
|
size_t n;
|
|
|
|
tokens = read_tokens(token_file, &n);
|
|
|
|
}
|
|
|
|
|
2016-01-28 23:30:38 +03:00
|
|
|
srand(rseed);
|
|
|
|
char *base = basecfg(cfgfile);
|
2016-02-01 02:52:03 +03:00
|
|
|
fprintf(stderr, "%s\n", base);
|
2016-01-28 23:30:38 +03:00
|
|
|
|
|
|
|
network net = parse_network_cfg(cfgfile);
|
|
|
|
if(weightfile){
|
|
|
|
load_weights(&net, weightfile);
|
|
|
|
}
|
2017-04-10 05:56:42 +03:00
|
|
|
int inputs = net.inputs;
|
2016-02-05 11:15:12 +03:00
|
|
|
|
2016-01-28 23:30:38 +03:00
|
|
|
int i, j;
|
|
|
|
for(i = 0; i < net.n; ++i) net.layers[i].temperature = temp;
|
2016-05-12 23:36:11 +03:00
|
|
|
int c = 0;
|
2016-01-28 23:30:38 +03:00
|
|
|
int len = strlen(seed);
|
2016-02-05 11:15:12 +03:00
|
|
|
float *input = calloc(inputs, sizeof(float));
|
2016-05-07 02:25:16 +03:00
|
|
|
|
2016-05-12 23:36:11 +03:00
|
|
|
/*
|
|
|
|
fill_cpu(inputs, 0, input, 1);
|
|
|
|
for(i = 0; i < 10; ++i){
|
|
|
|
network_predict(net, input);
|
|
|
|
}
|
|
|
|
fill_cpu(inputs, 0, input, 1);
|
|
|
|
*/
|
2016-05-07 02:25:16 +03:00
|
|
|
|
2016-01-28 23:30:38 +03:00
|
|
|
for(i = 0; i < len-1; ++i){
|
|
|
|
c = seed[i];
|
2016-05-12 23:36:11 +03:00
|
|
|
input[c] = 1;
|
2016-01-28 23:30:38 +03:00
|
|
|
network_predict(net, input);
|
2016-05-12 23:36:11 +03:00
|
|
|
input[c] = 0;
|
|
|
|
print_symbol(c, tokens);
|
2016-01-28 23:30:38 +03:00
|
|
|
}
|
2016-05-12 23:36:11 +03:00
|
|
|
if(len) c = seed[len-1];
|
|
|
|
print_symbol(c, tokens);
|
2016-01-28 23:30:38 +03:00
|
|
|
for(i = 0; i < num; ++i){
|
2016-05-12 23:36:11 +03:00
|
|
|
input[c] = 1;
|
2016-01-28 23:30:38 +03:00
|
|
|
float *out = network_predict(net, input);
|
2016-05-12 23:36:11 +03:00
|
|
|
input[c] = 0;
|
2016-05-07 02:25:16 +03:00
|
|
|
for(j = 32; j < 127; ++j){
|
|
|
|
//printf("%d %c %f\n",j, j, out[j]);
|
|
|
|
}
|
2016-02-05 11:15:12 +03:00
|
|
|
for(j = 0; j < inputs; ++j){
|
2016-05-12 23:36:11 +03:00
|
|
|
if (out[j] < .0001) out[j] = 0;
|
2016-01-28 23:30:38 +03:00
|
|
|
}
|
2016-05-07 02:25:16 +03:00
|
|
|
c = sample_array(out, inputs);
|
2016-05-12 23:36:11 +03:00
|
|
|
print_symbol(c, tokens);
|
2016-01-28 23:30:38 +03:00
|
|
|
}
|
|
|
|
printf("\n");
|
|
|
|
}
|
|
|
|
|
2016-06-14 21:30:28 +03:00
|
|
|
void test_tactic_rnn(char *cfgfile, char *weightfile, int num, float temp, int rseed, char *token_file)
|
2016-06-03 01:25:24 +03:00
|
|
|
{
|
|
|
|
char **tokens = 0;
|
|
|
|
if(token_file){
|
|
|
|
size_t n;
|
|
|
|
tokens = read_tokens(token_file, &n);
|
|
|
|
}
|
|
|
|
|
|
|
|
srand(rseed);
|
|
|
|
char *base = basecfg(cfgfile);
|
|
|
|
fprintf(stderr, "%s\n", base);
|
|
|
|
|
|
|
|
network net = parse_network_cfg(cfgfile);
|
|
|
|
if(weightfile){
|
|
|
|
load_weights(&net, weightfile);
|
|
|
|
}
|
2017-04-10 05:56:42 +03:00
|
|
|
int inputs = net.inputs;
|
2016-06-03 01:25:24 +03:00
|
|
|
|
|
|
|
int i, j;
|
|
|
|
for(i = 0; i < net.n; ++i) net.layers[i].temperature = temp;
|
|
|
|
int c = 0;
|
|
|
|
float *input = calloc(inputs, sizeof(float));
|
2016-06-14 21:30:28 +03:00
|
|
|
float *out = 0;
|
2016-06-03 01:25:24 +03:00
|
|
|
|
|
|
|
while((c = getc(stdin)) != EOF){
|
|
|
|
input[c] = 1;
|
|
|
|
out = network_predict(net, input);
|
|
|
|
input[c] = 0;
|
|
|
|
}
|
|
|
|
for(i = 0; i < num; ++i){
|
|
|
|
for(j = 0; j < inputs; ++j){
|
|
|
|
if (out[j] < .0001) out[j] = 0;
|
|
|
|
}
|
|
|
|
int next = sample_array(out, inputs);
|
|
|
|
if(c == '.' && next == '\n') break;
|
|
|
|
c = next;
|
|
|
|
print_symbol(c, tokens);
|
|
|
|
|
|
|
|
input[c] = 1;
|
|
|
|
out = network_predict(net, input);
|
|
|
|
input[c] = 0;
|
|
|
|
}
|
|
|
|
printf("\n");
|
|
|
|
}
|
|
|
|
|
|
|
|
void valid_tactic_rnn(char *cfgfile, char *weightfile, char *seed)
|
|
|
|
{
|
|
|
|
char *base = basecfg(cfgfile);
|
|
|
|
fprintf(stderr, "%s\n", base);
|
|
|
|
|
|
|
|
network net = parse_network_cfg(cfgfile);
|
|
|
|
if(weightfile){
|
|
|
|
load_weights(&net, weightfile);
|
|
|
|
}
|
2017-04-10 05:56:42 +03:00
|
|
|
int inputs = net.inputs;
|
2016-06-03 01:25:24 +03:00
|
|
|
|
|
|
|
int count = 0;
|
|
|
|
int words = 1;
|
|
|
|
int c;
|
|
|
|
int len = strlen(seed);
|
|
|
|
float *input = calloc(inputs, sizeof(float));
|
|
|
|
int i;
|
|
|
|
for(i = 0; i < len; ++i){
|
|
|
|
c = seed[i];
|
|
|
|
input[(int)c] = 1;
|
|
|
|
network_predict(net, input);
|
|
|
|
input[(int)c] = 0;
|
|
|
|
}
|
|
|
|
float sum = 0;
|
|
|
|
c = getc(stdin);
|
|
|
|
float log2 = log(2);
|
|
|
|
int in = 0;
|
|
|
|
while(c != EOF){
|
|
|
|
int next = getc(stdin);
|
|
|
|
if(next == EOF) break;
|
|
|
|
if(next < 0 || next >= 255) error("Out of range character");
|
|
|
|
|
|
|
|
input[c] = 1;
|
|
|
|
float *out = network_predict(net, input);
|
|
|
|
input[c] = 0;
|
|
|
|
|
|
|
|
if(c == '.' && next == '\n') in = 0;
|
|
|
|
if(!in) {
|
|
|
|
if(c == '>' && next == '>'){
|
|
|
|
in = 1;
|
|
|
|
++words;
|
|
|
|
}
|
|
|
|
c = next;
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
++count;
|
|
|
|
sum += log(out[next])/log2;
|
|
|
|
c = next;
|
|
|
|
printf("%d %d Perplexity: %4.4f Word Perplexity: %4.4f\n", count, words, pow(2, -sum/count), pow(2, -sum/words));
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2016-05-11 20:45:50 +03:00
|
|
|
void valid_char_rnn(char *cfgfile, char *weightfile, char *seed)
|
2016-02-01 02:52:03 +03:00
|
|
|
{
|
|
|
|
char *base = basecfg(cfgfile);
|
|
|
|
fprintf(stderr, "%s\n", base);
|
|
|
|
|
|
|
|
network net = parse_network_cfg(cfgfile);
|
|
|
|
if(weightfile){
|
|
|
|
load_weights(&net, weightfile);
|
|
|
|
}
|
2017-04-10 05:56:42 +03:00
|
|
|
int inputs = net.inputs;
|
2016-02-05 11:15:12 +03:00
|
|
|
|
|
|
|
int count = 0;
|
2016-05-12 23:36:11 +03:00
|
|
|
int words = 1;
|
2016-02-05 11:15:12 +03:00
|
|
|
int c;
|
2016-05-11 20:45:50 +03:00
|
|
|
int len = strlen(seed);
|
2016-02-05 11:15:12 +03:00
|
|
|
float *input = calloc(inputs, sizeof(float));
|
2016-05-07 02:25:16 +03:00
|
|
|
int i;
|
2016-05-11 20:45:50 +03:00
|
|
|
for(i = 0; i < len; ++i){
|
|
|
|
c = seed[i];
|
|
|
|
input[(int)c] = 1;
|
2016-05-07 02:25:16 +03:00
|
|
|
network_predict(net, input);
|
2016-05-11 20:45:50 +03:00
|
|
|
input[(int)c] = 0;
|
2016-05-07 02:25:16 +03:00
|
|
|
}
|
2016-02-01 02:52:03 +03:00
|
|
|
float sum = 0;
|
2016-02-05 11:15:12 +03:00
|
|
|
c = getc(stdin);
|
|
|
|
float log2 = log(2);
|
|
|
|
while(c != EOF){
|
|
|
|
int next = getc(stdin);
|
|
|
|
if(next == EOF) break;
|
2016-05-11 20:45:50 +03:00
|
|
|
if(next < 0 || next >= 255) error("Out of range character");
|
2016-02-05 11:15:12 +03:00
|
|
|
++count;
|
2016-05-12 23:36:11 +03:00
|
|
|
if(next == ' ' || next == '\n' || next == '\t') ++words;
|
2016-02-05 11:15:12 +03:00
|
|
|
input[c] = 1;
|
2016-02-01 02:52:03 +03:00
|
|
|
float *out = network_predict(net, input);
|
2016-02-05 11:15:12 +03:00
|
|
|
input[c] = 0;
|
|
|
|
sum += log(out[next])/log2;
|
|
|
|
c = next;
|
2016-05-12 23:36:11 +03:00
|
|
|
printf("%d Perplexity: %4.4f Word Perplexity: %4.4f\n", count, pow(2, -sum/count), pow(2, -sum/words));
|
2016-02-01 02:52:03 +03:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2016-05-11 20:45:50 +03:00
|
|
|
void vec_char_rnn(char *cfgfile, char *weightfile, char *seed)
|
|
|
|
{
|
|
|
|
char *base = basecfg(cfgfile);
|
|
|
|
fprintf(stderr, "%s\n", base);
|
|
|
|
|
|
|
|
network net = parse_network_cfg(cfgfile);
|
|
|
|
if(weightfile){
|
|
|
|
load_weights(&net, weightfile);
|
|
|
|
}
|
2017-04-10 05:56:42 +03:00
|
|
|
int inputs = net.inputs;
|
2016-05-11 20:45:50 +03:00
|
|
|
|
|
|
|
int c;
|
|
|
|
int seed_len = strlen(seed);
|
|
|
|
float *input = calloc(inputs, sizeof(float));
|
|
|
|
int i;
|
|
|
|
char *line;
|
|
|
|
while((line=fgetl(stdin)) != 0){
|
|
|
|
reset_rnn_state(net, 0);
|
|
|
|
for(i = 0; i < seed_len; ++i){
|
|
|
|
c = seed[i];
|
|
|
|
input[(int)c] = 1;
|
|
|
|
network_predict(net, input);
|
|
|
|
input[(int)c] = 0;
|
|
|
|
}
|
|
|
|
strip(line);
|
|
|
|
int str_len = strlen(line);
|
|
|
|
for(i = 0; i < str_len; ++i){
|
|
|
|
c = line[i];
|
|
|
|
input[(int)c] = 1;
|
|
|
|
network_predict(net, input);
|
|
|
|
input[(int)c] = 0;
|
|
|
|
}
|
2016-05-12 23:36:11 +03:00
|
|
|
c = ' ';
|
|
|
|
input[(int)c] = 1;
|
|
|
|
network_predict(net, input);
|
|
|
|
input[(int)c] = 0;
|
2016-05-11 20:45:50 +03:00
|
|
|
|
|
|
|
layer l = net.layers[0];
|
2016-05-12 23:36:11 +03:00
|
|
|
#ifdef GPU
|
2016-05-11 20:45:50 +03:00
|
|
|
cuda_pull_array(l.output_gpu, l.output, l.outputs);
|
2016-05-12 23:36:11 +03:00
|
|
|
#endif
|
2016-05-11 20:45:50 +03:00
|
|
|
printf("%s", line);
|
|
|
|
for(i = 0; i < l.outputs; ++i){
|
|
|
|
printf(",%g", l.output[i]);
|
|
|
|
}
|
|
|
|
printf("\n");
|
|
|
|
}
|
|
|
|
}
|
2016-02-01 02:52:03 +03:00
|
|
|
|
2016-01-28 23:30:38 +03:00
|
|
|
void run_char_rnn(int argc, char **argv)
|
|
|
|
{
|
|
|
|
if(argc < 4){
|
|
|
|
fprintf(stderr, "usage: %s %s [train/test/valid] [cfg] [weights (optional)]\n", argv[0], argv[1]);
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
char *filename = find_char_arg(argc, argv, "-file", "data/shakespeare.txt");
|
2016-05-11 20:45:50 +03:00
|
|
|
char *seed = find_char_arg(argc, argv, "-seed", "\n\n");
|
2016-02-05 11:15:12 +03:00
|
|
|
int len = find_int_arg(argc, argv, "-len", 1000);
|
|
|
|
float temp = find_float_arg(argc, argv, "-temp", .7);
|
2016-01-28 23:30:38 +03:00
|
|
|
int rseed = find_int_arg(argc, argv, "-srand", time(0));
|
2016-05-07 02:25:16 +03:00
|
|
|
int clear = find_arg(argc, argv, "-clear");
|
2016-05-12 23:36:11 +03:00
|
|
|
int tokenized = find_arg(argc, argv, "-tokenized");
|
|
|
|
char *tokens = find_char_arg(argc, argv, "-tokens", 0);
|
2016-01-28 23:30:38 +03:00
|
|
|
|
|
|
|
char *cfg = argv[3];
|
|
|
|
char *weights = (argc > 4) ? argv[4] : 0;
|
2016-05-12 23:36:11 +03:00
|
|
|
if(0==strcmp(argv[2], "train")) train_char_rnn(cfg, weights, filename, clear, tokenized);
|
2016-05-11 20:45:50 +03:00
|
|
|
else if(0==strcmp(argv[2], "valid")) valid_char_rnn(cfg, weights, seed);
|
2016-06-03 01:25:24 +03:00
|
|
|
else if(0==strcmp(argv[2], "validtactic")) valid_tactic_rnn(cfg, weights, seed);
|
2016-05-11 20:45:50 +03:00
|
|
|
else if(0==strcmp(argv[2], "vec")) vec_char_rnn(cfg, weights, seed);
|
2016-05-12 23:36:11 +03:00
|
|
|
else if(0==strcmp(argv[2], "generate")) test_char_rnn(cfg, weights, len, seed, temp, rseed, tokens);
|
2016-06-14 21:30:28 +03:00
|
|
|
else if(0==strcmp(argv[2], "generatetactic")) test_tactic_rnn(cfg, weights, len, temp, rseed, tokens);
|
2016-01-28 23:30:38 +03:00
|
|
|
}
|