From f2d50c333d0dde8a5ef211bc31b4978a3d8911cf Mon Sep 17 00:00:00 2001 From: "Wu, Ke" Date: Wed, 17 Dec 2014 15:41:32 -0500 Subject: Move training routine out of ff_const_reorder_common.h --- decoder/ff_const_reorder_common.h | 93 --------------------------------------- 1 file changed, 93 deletions(-) (limited to 'decoder') diff --git a/decoder/ff_const_reorder_common.h b/decoder/ff_const_reorder_common.h index 7c111de3..b124ce47 100644 --- a/decoder/ff_const_reorder_common.h +++ b/decoder/ff_const_reorder_common.h @@ -1091,99 +1091,6 @@ struct Tsuruoka_Maxent { if (m_pModel != NULL) delete m_pModel; } - void fnTrain(const char* pszInstanceFName, const char* pszAlgorithm, - const char* pszModelFName, int /*iNumIteration*/) { - assert(strcmp(pszAlgorithm, "l1") == 0 || strcmp(pszAlgorithm, "l2") == 0 || - strcmp(pszAlgorithm, "sgd") == 0 || - strcmp(pszAlgorithm, "SGD") == 0); - FILE* fpIn = fopen(pszInstanceFName, "r"); - - ME_Model* pModel = new ME_Model(); - - char* pszLine = new char[100001]; - int iNumInstances = 0; - int iLen; - while (!feof(fpIn)) { - pszLine[0] = '\0'; - fgets(pszLine, 20000, fpIn); - if (strlen(pszLine) == 0) { - continue; - } - - iLen = strlen(pszLine); - while (iLen > 0 && pszLine[iLen - 1] > 0 && pszLine[iLen - 1] < 33) { - pszLine[iLen - 1] = '\0'; - iLen--; - } - - iNumInstances++; - - ME_Sample* pmes = new ME_Sample(); - - char* p = strrchr(pszLine, ' '); - assert(p != NULL); - p[0] = '\0'; - p++; - std::vector vecContext; - SplitOnWhitespace(std::string(pszLine), &vecContext); - - pmes->label = std::string(p); - for (size_t i = 0; i < vecContext.size(); i++) - pmes->add_feature(vecContext[i]); - pModel->add_training_sample((*pmes)); - if (iNumInstances % 100000 == 0) - fprintf(stdout, "......Reading #Instances: %1d\n", iNumInstances); - delete pmes; - } - fprintf(stdout, "......Reading #Instances: %1d\n", iNumInstances); - fclose(fpIn); - - if (strcmp(pszAlgorithm, "l1") == 0) - pModel->use_l1_regularizer(1.0); - else if (strcmp(pszAlgorithm, "l2") == 0) - pModel->use_l2_regularizer(1.0); - else - pModel->use_SGD(); - - pModel->train(); - pModel->save_to_file(pszModelFName); - - delete pModel; - fprintf(stdout, "......Finished Training\n"); - fprintf(stdout, "......Model saved as %s\n", pszModelFName); - delete[] pszLine; - } - - double fnEval(const char* pszContext, const char* pszOutcome) const { - std::vector vecContext; - ME_Sample* pmes = new ME_Sample(); - SplitOnWhitespace(std::string(pszContext), &vecContext); - - for (size_t i = 0; i < vecContext.size(); i++) - pmes->add_feature(vecContext[i]); - std::vector vecProb = m_pModel->classify(*pmes); - delete pmes; - int iLableID = m_pModel->get_class_id(pszOutcome); - return vecProb[iLableID]; - } - void fnEval(const char* pszContext, - std::vector >& vecOutput) const { - std::vector vecContext; - ME_Sample* pmes = new ME_Sample(); - SplitOnWhitespace(std::string(pszContext), &vecContext); - - vecOutput.clear(); - - for (size_t i = 0; i < vecContext.size(); i++) - pmes->add_feature(vecContext[i]); - std::vector vecProb = m_pModel->classify(*pmes); - - for (size_t i = 0; i < vecProb.size(); i++) { - std::string label = m_pModel->get_class_label(i); - vecOutput.push_back(make_pair(label, vecProb[i])); - } - delete pmes; - } void fnEval(const char* pszContext, std::vector& vecOutput) const { std::vector vecContext; ME_Sample* pmes = new ME_Sample(); -- cgit v1.2.3