main-SA-train.cpp File Reference

#include "trf-model.h"
#include "trf-sa-train.h"
#include <omp.h>

Go to the source code of this file.

Functions
opt	Add (wbOPT_STRING, "feat", &cfg_pathFeatStyle, "a feature style file. Set this value will disable -order")
opt	Add (wbOPT_INT, "order", &cfg_nFeatOrder, "the ngram feature order")
opt	Add (wbOPT_INT, "len", &cfg_nMaxLen, "the maximum length of TRF")
opt	Add (wbOPT_STRING, "train", &cfg_pathTrain, "Training corpus (TXT)")
opt	Add (wbOPT_STRING, "valid", &cfg_pathValid, "valid corpus (TXT)")
opt	Add (wbOPT_STRING, "test", &cfg_pathTest, "test corpus (TXT)")
opt	Add (wbOPT_STRING, "read", &cfg_pathModelRead, "Read the init model to train")
opt	Add (wbOPT_STRING, "write", &cfg_pathModelWrite, "Output model")
opt	Add (wbOPT_INT, "iter", &cfg_nIterTotalNum, "iter total number")
opt	Add (wbOPT_INT, "thread", &cfg_nThread, "The thread number")
opt	Add (wbOPT_INT, "mini-batch", &cfg_nMiniBatch, "mini-batch")
opt	Add (wbOPT_INT, "t0", &cfg_t0, "t0")
opt	Add (wbOPT_STRING, "gamma-lambda", &cfg_gamma_lambda, "learning rate of lambda")
opt	Add (wbOPT_STRING, "gamma-zeta", &cfg_gamma_zeta, "learning rate of zeta")
opt	Add (wbOPT_TRUE, "unupdate-lambda", &cfg_bUnupdateLambda, "don't update lambda")
opt	Add (wbOPT_TRUE, "unupdate-zeta", &cfg_bUnupdateZeta, "don't update zeta")
opt	Add (wbOPT_INT, "tavg", &cfg_nAvgBeg, ">0 then apply averaging")
opt	Add (wbOPT_FLOAT, "L2", &cfg_fRegL2, "regularization L2")
opt	Add (wbOPT_FLOAT, "dgap", &cfg_dGap, "the gap for update value at each iteration")
opt	Add (wbOPT_FALSE, "vgap", &cfg_vGap, "the gap for empirical variance")
opt	Add (wbOPT_TRUE, "init", &cfg_bInitValue, "Re-init the parameters")
opt	Add (wbOPT_INT, "print-per-iter", &cfg_nPrintPerIter, "print the LL per iterations")
opt	Add (wbOPT_STRING, "write-at-iter", &cfg_strWriteAtIter, "write models at iteration, such as [1:100:1000]")
opt	Add (wbOPT_STRING, "write-mean", &cfg_pathWriteMean, "write the expecataion on training set")
opt	Add (wbOPT_STRING, "write-var", &cfg_pathWriteVar, "write the variance on training set")
opt	Add (wbOPT_STRING, "write-train-ll", &cfg_pathWriteLLtrain, "write LL on train")
opt	Add (wbOPT_STRING, "write-valid-ll", &cfg_pathWriteLLvalid, "write LL on valid")
opt	Add (wbOPT_STRING, "write-test-ll", &cfg_pathWriteLLtest, "write LL on test")
opt	Parse (_argc, _argv)
	if (cfg_pathModelRead)
	if the model are inputed, then using the input parameters More...
	if (bInitWeight)
	if (bInitZeta)
solve	Run (vInitParams.GetBuf())
m	WriteT (cfg_pathModelWrite)
	SAFE_DELETE (pTrain)
	SAFE_DELETE (pValid)
	SAFE_DELETE (pTest)
	SAFE_DELETE (pv)

Variables
char *	cfg_pathVocab = NULL
int	cfg_nFeatOrder = 2
char *	cfg_pathFeatStyle = NULL
int	cfg_nMaxLen = 0
char *	cfg_pathTrain = NULL
char *	cfg_pathValid = NULL
char *	cfg_pathTest = NULL
char *	cfg_pathModelRead = NULL
char *	cfg_pathModelWrite = NULL
int	cfg_nThread = 1
int	cfg_nIterTotalNum = 1000
int	cfg_nMiniBatch = 300
int	cfg_t0 = 500
char *	cfg_gamma_lambda = NULL
char *	cfg_gamma_zeta = NULL
bool	cfg_bUnupdateLambda = false
bool	cfg_bUnupdateZeta = false
int	cfg_nAvgBeg = 0
float	cfg_fRegL2 = 0
float	cfg_dGap = 1.0f
float	cfg_vGap = 1e-5
bool	cfg_bInitValue = false
int	cfg_nPrintPerIter = 1
char *	cfg_strWriteAtIter = NULL
char *	cfg_pathWriteMean = NULL
char *	cfg_pathWriteVar = NULL
char *	cfg_pathWriteLLtrain = NULL
char *	cfg_pathWriteLLvalid = NULL
char *	cfg_pathWriteLLtest = NULL
Option	opt
	_wbMain
lout<< "*********************************************"<< endl;lout<< " TRF_SAtrain.exe { by Bin Wang } "<< endl;lout<< "\"<< __DATE__<< "\"<< __TIME__<< "\"<< endl;lout<< "**********************************************"<< endl;omp_set_num_threads(cfg_nThread);lout<< "[OMP] omp_thread = "<< omp_get_max_threads()<< endl;trf::omp_rand(cfg_nThread);Title::SetGlobalTitle(String(cfg_pathModelWrite).FileName());Vocab *pv=new Vocab(cfg_pathVocab);Model m(pv, cfg_nMaxLen);if(cfg_pathModelRead) { m.ReadT(cfg_pathModelRead);} else { m.LoadFromCorpus(cfg_pathTrain, cfg_pathFeatStyle, cfg_nFeatOrder);} CorpusTxt *pTrain=(cfg_pathTrain) ? new CorpusTxt(cfg_pathTrain) :NULL;CorpusTxt *pValid=(cfg_pathValid) ? new CorpusTxt(cfg_pathValid) :NULL;CorpusTxt *pTest=(cfg_pathTest) ? new CorpusTxt(cfg_pathTest) :NULL;SAfunc func;func.m_fdbg.Open(String(cfg_pathModelWrite).FileName()+".sadbg", "wt");func.m_fmean.Open(cfg_pathWriteMean, "wt");func.m_fvar.Open(cfg_pathWriteVar, "wt");func.m_ftrainLL.Open(cfg_pathWriteLLtrain, "wt");func.m_fvallidLL.Open(cfg_pathWriteLLvalid, "wt");func.m_ftestLL.Open(cfg_pathWriteLLtest, "wt");func.m_pathOutputModel=cfg_pathModelWrite;func.Reset(&m, pTrain, pValid, pTest, cfg_nMiniBatch);func.m_fRegL2=cfg_fRegL2;func.m_var_gap=cfg_vGap;func.PrintInfo();SAtrain solve(&func);solve.m_nIterMax=cfg_nIterTotalNum;solve.m_gain_lambda.Reset(cfg_gamma_lambda ? cfg_gamma_lambda :"0,0", cfg_t0);solve.m_gain_zeta.Reset(cfg_gamma_zeta ? cfg_gamma_zeta :"0,0.6", cfg_t0);solve.m_bUpdate_lambda=!cfg_bUnupdateLambda;solve.m_bUpdate_zeta=!cfg_bUnupdateZeta;solve.m_nAvgBeg=cfg_nAvgBeg;solve.m_nPrintPerIter=cfg_nPrintPerIter;solve.m_dir_gap=cfg_dGap;VecUnfold(cfg_strWriteAtIter, solve.m_aWriteAtIter);solve.PrintInfo();bool bInitWeight=(!cfg_pathModelRead)||(cfg_bInitValue &&!cfg_bUnupdateLambda);bool bInitZeta=(!cfg_pathModelRead)||(cfg_bInitValue &&!cfg_bUnupdateZeta);Vec< double >	vInitParams (func.GetParamNum())
	return

Function Documentation

Add() [1/28]

opt Add	(	wbOPT_STRING	,
		"feat"	,
		&	cfg_pathFeatStyle,
		"a feature style file. Set this value will disable -order"
	)

Add() [2/28]

opt Add	(	wbOPT_INT	,
		"order"	,
		&	cfg_nFeatOrder,
		"the ngram feature order"
	)

Add() [3/28]

opt Add	(	wbOPT_INT	,
		"len"	,
		&	cfg_nMaxLen,
		"the maximum length of TRF"
	)

Add() [4/28]

opt Add	(	wbOPT_STRING	,
		"train"	,
		&	cfg_pathTrain,
		"Training corpus (TXT)"
	)

Add() [5/28]

opt Add	(	wbOPT_STRING	,
		"valid"	,
		&	cfg_pathValid,
		"valid corpus (TXT)"
	)

Add() [6/28]

opt Add	(	wbOPT_STRING	,
		"test"	,
		&	cfg_pathTest,
		"test corpus (TXT)"
	)

Add() [7/28]

opt Add	(	wbOPT_STRING	,
		"read"	,
		&	cfg_pathModelRead,
		"Read the init model to train"
	)

Add() [8/28]

opt Add	(	wbOPT_STRING	,
		"write"	,
		&	cfg_pathModelWrite,
		"Output model"
	)

Add() [9/28]

opt Add	(	wbOPT_INT	,
		"iter"	,
		&	cfg_nIterTotalNum,
		"iter total number"
	)

Add() [10/28]

opt Add	(	wbOPT_INT	,
		"thread"	,
		&	cfg_nThread,
		"The thread number"
	)

Add() [11/28]

opt Add	(	wbOPT_INT	,
		"mini-batch"	,
		&	cfg_nMiniBatch,
		"mini-batch"
	)

Add() [12/28]

opt Add	(	wbOPT_INT	,
		"t0"	,
		&	cfg_t0,
		"t0"
	)

Add() [13/28]

opt Add	(	wbOPT_STRING	,
		"gamma-lambda"	,
		&	cfg_gamma_lambda,
		"learning rate of lambda"
	)

Add() [14/28]

opt Add	(	wbOPT_STRING	,
		"gamma-zeta"	,
		&	cfg_gamma_zeta,
		"learning rate of zeta"
	)

Add() [15/28]

opt Add	(	wbOPT_TRUE	,
		"unupdate-lambda"	,
		&	cfg_bUnupdateLambda,
		"don't update lambda"
	)

Add() [16/28]

opt Add	(	wbOPT_TRUE	,
		"unupdate-zeta"	,
		&	cfg_bUnupdateZeta,
		"don't update zeta"
	)

Add() [17/28]

opt Add	(	wbOPT_INT	,
		"tavg"	,
		&	cfg_nAvgBeg,
		"	,
		0 then apply averaging"
	)

Add() [18/28]

opt Add	(	wbOPT_FLOAT	,
		"L2"	,
		&	cfg_fRegL2,
		"regularization L2"
	)

Add() [19/28]

opt Add	(	wbOPT_FLOAT	,
		"dgap"	,
		&	cfg_dGap,
		"the gap for update value at each iteration"
	)

Add() [20/28]

opt Add	(	wbOPT_FALSE	,
		"vgap"	,
		&	cfg_vGap,
		"the gap for empirical variance"
	)

Add() [21/28]

opt Add	(	wbOPT_TRUE	,
		"init"	,
		&	cfg_bInitValue,
		"Re-init the parameters"
	)

Add() [22/28]

opt Add	(	wbOPT_INT	,
		"print-per-iter"	,
		&	cfg_nPrintPerIter,
		"print the LL per iterations"
	)

Add() [23/28]

opt Add	(	wbOPT_STRING	,
		"write-at-iter"	,
		&	cfg_strWriteAtIter,
		"write models at	iteration,
		such as "	[1:100:1000]
	)

Add() [24/28]

opt Add	(	wbOPT_STRING	,
		"write-mean"	,
		&	cfg_pathWriteMean,
		"write the expecataion on training set"
	)

Add() [25/28]

opt Add	(	wbOPT_STRING	,
		"write-var"	,
		&	cfg_pathWriteVar,
		"write the variance on training set"
	)

Add() [26/28]

opt Add	(	wbOPT_STRING	,
		"write-train-ll"	,
		&	cfg_pathWriteLLtrain,
		"write LL on train"
	)

Add() [27/28]

opt Add	(	wbOPT_STRING	,
		"write-valid-ll"	,
		&	cfg_pathWriteLLvalid,
		"write LL on valid"
	)

Add() [28/28]

opt Add	(	wbOPT_STRING	,
		"write-test-ll"	,
		&	cfg_pathWriteLLtest,
		"write LL on test"
	)

if() [1/3]

if

(

cfg_pathModelRead

)

if the model are inputed, then using the input parameters

Definition at line 166 of file main-SA-train.cpp.

if() [2/3]

if

(

bInitWeight

)

Definition at line 169 of file main-SA-train.cpp.

if() [3/3]

if

(

bInitZeta

)

Definition at line 173 of file main-SA-train.cpp.

Parse()

opt Parse	(	_argc	,
		_argv
	)

Run()

solve Run

(

vInitParams.

GetBuf()

)

SAFE_DELETE() [1/4]

SAFE_DELETE

(

pTrain

)

SAFE_DELETE() [2/4]

SAFE_DELETE

(

pValid

)

SAFE_DELETE() [3/4]

SAFE_DELETE

(

pTest

)

SAFE_DELETE() [4/4]

SAFE_DELETE

(

pv

)

WriteT()

m WriteT

(

cfg_pathModelWrite

)

Variable Documentation

_wbMain

_wbMain

Initial value:

{
    opt.Add(wbOPT_STRING, "vocab", &cfg_pathVocab, "The vocabulary")

Definition at line 65 of file main-SA-train.cpp.

cfg_bInitValue

bool cfg_bInitValue = false

Definition at line 52 of file main-SA-train.cpp.

cfg_bUnupdateLambda

bool cfg_bUnupdateLambda = false

Definition at line 44 of file main-SA-train.cpp.

cfg_bUnupdateZeta

bool cfg_bUnupdateZeta = false

Definition at line 45 of file main-SA-train.cpp.

cfg_dGap

float cfg_dGap = 1.0f

Definition at line 49 of file main-SA-train.cpp.

cfg_fRegL2

float cfg_fRegL2 = 0

Definition at line 48 of file main-SA-train.cpp.

cfg_gamma_lambda

char* cfg_gamma_lambda = NULL

Definition at line 42 of file main-SA-train.cpp.

cfg_gamma_zeta

char* cfg_gamma_zeta = NULL

Definition at line 43 of file main-SA-train.cpp.

cfg_nAvgBeg

int cfg_nAvgBeg = 0

Definition at line 46 of file main-SA-train.cpp.

cfg_nFeatOrder

int cfg_nFeatOrder = 2

Definition at line 26 of file main-SA-train.cpp.

cfg_nIterTotalNum

int cfg_nIterTotalNum = 1000

Definition at line 39 of file main-SA-train.cpp.

cfg_nMaxLen

int cfg_nMaxLen = 0

Definition at line 28 of file main-SA-train.cpp.

cfg_nMiniBatch

int cfg_nMiniBatch = 300

Definition at line 40 of file main-SA-train.cpp.

cfg_nPrintPerIter

int cfg_nPrintPerIter = 1

Definition at line 53 of file main-SA-train.cpp.

cfg_nThread

int cfg_nThread = 1

Definition at line 37 of file main-SA-train.cpp.

cfg_pathFeatStyle

char* cfg_pathFeatStyle = NULL

Definition at line 27 of file main-SA-train.cpp.

cfg_pathModelRead

char* cfg_pathModelRead = NULL

Definition at line 34 of file main-SA-train.cpp.

cfg_pathModelWrite

char* cfg_pathModelWrite = NULL

Definition at line 35 of file main-SA-train.cpp.

cfg_pathTest

char* cfg_pathTest = NULL

Definition at line 32 of file main-SA-train.cpp.

cfg_pathTrain

char* cfg_pathTrain = NULL

Definition at line 30 of file main-SA-train.cpp.

cfg_pathValid

char* cfg_pathValid = NULL

Definition at line 31 of file main-SA-train.cpp.

cfg_pathVocab

char* cfg_pathVocab = NULL

Definition at line 24 of file main-SA-train.cpp.

cfg_pathWriteLLtest

char* cfg_pathWriteLLtest = NULL

Definition at line 60 of file main-SA-train.cpp.

cfg_pathWriteLLtrain

char* cfg_pathWriteLLtrain = NULL

Definition at line 58 of file main-SA-train.cpp.

cfg_pathWriteLLvalid

char* cfg_pathWriteLLvalid = NULL

Definition at line 59 of file main-SA-train.cpp.

cfg_pathWriteMean

char* cfg_pathWriteMean = NULL

Definition at line 56 of file main-SA-train.cpp.

cfg_pathWriteVar

char* cfg_pathWriteVar = NULL

Definition at line 57 of file main-SA-train.cpp.

cfg_strWriteAtIter

char* cfg_strWriteAtIter = NULL

Definition at line 54 of file main-SA-train.cpp.

cfg_t0

int cfg_t0 = 500

Definition at line 41 of file main-SA-train.cpp.

cfg_vGap

float cfg_vGap = 1e-5

Definition at line 50 of file main-SA-train.cpp.

opt

Option opt

Definition at line 62 of file main-SA-train.cpp.

return

return

Definition at line 191 of file main-SA-train.cpp.

vInitParams

lout<< "*********************************************" << endl; lout << " TRF_SAtrain.exe { by Bin Wang } " << endl; lout << "\t" << __DATE__ << "\t" << __TIME__ << "\t" << endl; lout << "**********************************************" << endl; omp_set_num_threads(cfg_nThread); lout << "[OMP] omp_thread = " << omp_get_max_threads() << endl; trf::omp_rand(cfg_nThread); Title::SetGlobalTitle(String(cfg_pathModelWrite).FileName()); Vocab *pv = new Vocab(cfg_pathVocab); Model m(pv, cfg_nMaxLen); if (cfg_pathModelRead) { m.ReadT(cfg_pathModelRead); } else { m.LoadFromCorpus(cfg_pathTrain, cfg_pathFeatStyle, cfg_nFeatOrder); } CorpusTxt *pTrain = (cfg_pathTrain) ? new CorpusTxt(cfg_pathTrain) : NULL; CorpusTxt *pValid = (cfg_pathValid) ? new CorpusTxt(cfg_pathValid) : NULL; CorpusTxt *pTest = (cfg_pathTest) ? new CorpusTxt(cfg_pathTest) : NULL; SAfunc func; func.m_fdbg.Open(String(cfg_pathModelWrite).FileName() + ".sadbg", "wt"); func.m_fmean.Open(cfg_pathWriteMean, "wt"); func.m_fvar.Open(cfg_pathWriteVar, "wt"); func.m_ftrainLL.Open(cfg_pathWriteLLtrain, "wt"); func.m_fvallidLL.Open(cfg_pathWriteLLvalid, "wt"); func.m_ftestLL.Open(cfg_pathWriteLLtest, "wt"); func.m_pathOutputModel = cfg_pathModelWrite; func.Reset(&m, pTrain, pValid, pTest, cfg_nMiniBatch); func.m_fRegL2 = cfg_fRegL2; func.m_var_gap = cfg_vGap; func.PrintInfo(); SAtrain solve(&func); solve.m_nIterMax = cfg_nIterTotalNum; solve.m_gain_lambda.Reset(cfg_gamma_lambda ? cfg_gamma_lambda : "0,0", cfg_t0); solve.m_gain_zeta.Reset(cfg_gamma_zeta ? cfg_gamma_zeta : "0,0.6", cfg_t0); solve.m_bUpdate_lambda = !cfg_bUnupdateLambda; solve.m_bUpdate_zeta = !cfg_bUnupdateZeta; solve.m_nAvgBeg = cfg_nAvgBeg; solve.m_nPrintPerIter = cfg_nPrintPerIter; solve.m_dir_gap = cfg_dGap; VecUnfold(cfg_strWriteAtIter, solve.m_aWriteAtIter); solve.PrintInfo(); bool bInitWeight = (!cfg_pathModelRead) || (cfg_bInitValue && !cfg_bUnupdateLambda); bool bInitZeta = (!cfg_pathModelRead) || (cfg_bInitValue && !cfg_bUnupdateZeta); Vec<double> vInitParams(func.GetParamNum())

Definition at line 164 of file main-SA-train.cpp.