void EATrain(int argc, char **argv)
{
const char *ee_file, *de_file, *dw_file, *entity_cnts_file, *word_cnts_file,
*dst_doc_vecs_file, *dst_word_vecs_file,
*dst_entity_vecs_file;
bool share_doc_vec = 1;
int doc_vec_dim = GetIntArgValue(argc, argv, "-d", 100);
int num_rounds = GetIntArgValue(argc, argv, "-r", 10);
int num_threads = GetIntArgValue(argc, argv, "-t", 2);
int num_negative_samples = GetIntArgValue(argc, argv, "-n", 10);
float starting_alpha = GetFloatArgValue(argc, argv, "-sa", 0.06f);
float min_alpha = GetFloatArgValue(argc, argv, "-ma", 0.0001f);
ee_file = GetArgValue(argc, argv, "-ee");
de_file = GetArgValue(argc, argv, "-de");
dw_file = GetArgValue(argc, argv, "-dw");
entity_cnts_file = GetArgValue(argc, argv, "-ecnt");
word_cnts_file = GetArgValue(argc, argv, "-wcnt");
dst_doc_vecs_file = GetArgValue(argc, argv, "-docvec");
dst_word_vecs_file = GetArgValue(argc, argv, "-wordvec");
dst_entity_vecs_file = GetArgValue(argc, argv, "-entityvec");
if (!ee_file)
{
DataSet data_set = DataSet::WIN_20NG;
configFiles(data_set, ee_file, de_file, dw_file, entity_cnts_file, word_cnts_file,
dst_doc_vecs_file, dst_word_vecs_file, dst_entity_vecs_file);
}
//if (argc >= 7)
//{
// doc_vec_dim = atoi(argv[1]);
// num_rounds = atoi(argv[2]);
// num_threads = atoi(argv[3]);
// num_negative_samples = atoi(argv[4]);
// starting_alpha = (float)atof(argv[5]);
// min_alpha = (float)atof(argv[6]);
// if (argc == 8)
// dst_doc_vecs_file = argv[7];
//}
//else
//{
// DataSet data_set = DataSet::WIN_20NG;
// configFiles(data_set, ee_file, de_file, dw_file, entity_cnts_file, word_cnts_file,
// dst_doc_vecs_file, dst_word_vecs_file, dst_entity_vecs_file);
//}
printf("ee_file: %s\nde_file: %s\ndw_file: %s\n", ee_file, de_file, dw_file);
printf("vec_dim: %d\nnum_rounds: %d\nnum_threads: %d\nnum_neg_samples: %d\nstarting_alpha: %f\nmin_alpha: %f\n",
doc_vec_dim, num_rounds, num_threads, num_negative_samples, starting_alpha, min_alpha);
EADocVecTrainer eatrain(num_rounds, num_threads, num_negative_samples, starting_alpha, min_alpha);
eatrain.AllJointThreaded(ee_file, de_file, dw_file,
entity_cnts_file, word_cnts_file, doc_vec_dim, share_doc_vec, dst_doc_vecs_file, dst_word_vecs_file,
dst_entity_vecs_file);
}
float GetFloatArgValue(int argc, char **argv, const char *arg, float def_val)
{
char *arg_val = GetArgValue(argc, argv, arg);
if (arg_val)
return atof(arg_val);
return def_val;
}
int GetIntArgValue(int argc, char **argv, const char *arg, int def_val)
{
char *arg_val = GetArgValue(argc, argv, arg);
if (arg_val)
return atoi(arg_val);
return def_val;
}
void InitEngineUCI( const char * iniDir, ChessProgramState * cps )
{ // replace engine command line by adapter command with expanded meta-symbols
if( cps->isUCI ) {
char *p, *q;
char polyglotCommand[MSG_SIZ];
p = appData.adapterCommand;
q = polyglotCommand;
while(*p) {
if(*p == '\\') p++; else
if(*p == '%') { // substitute marker
char argName[MSG_SIZ], buf[MSG_SIZ], *s = buf;
if(*++p == '%') { // second %, expand as f or s in option name (e.g. %%cp -> fcp)
*s++ = cps == &first ? 'f' : 's';
p++;
}
while(*p >= '0' && *p) *s++ = *p++; // copy option name
*s = NULLCHAR;
if(cps == &second) { // change options for first into those for second engine
if(strstr(buf, "first") == buf) sprintf(argName, "second%s", buf+5); else
if(buf[0] == 'f') sprintf(argName, "s%s", buf+1); else
safeStrCpy(argName, buf, sizeof(argName)/sizeof(argName[0]));
} else safeStrCpy(argName, buf, sizeof(argName)/sizeof(argName[0]));
if(GetArgValue(argName)) { // look up value of option with this name
s = argName;
while(*s) *q++ = *s++;
} else DisplayFatalError("Bad adapter command", 0, 1);
continue;
}
if(*p) *q++ = *p++;
}
*q = NULLCHAR;
cps->program = StrSave(polyglotCommand);
cps->dir = appData.polyglotDir;
}
}