diff options
| -rw-r--r-- | src/storage/flexible_ngram_kyotodb.h | 6 | ||||
| -rw-r--r-- | utils/training/estimate_k_mixture_model.cpp | 9 |
2 files changed, 7 insertions, 8 deletions
diff --git a/src/storage/flexible_ngram_kyotodb.h b/src/storage/flexible_ngram_kyotodb.h index 660681e..1c1b1db 100644 --- a/src/storage/flexible_ngram_kyotodb.h +++ b/src/storage/flexible_ngram_kyotodb.h @@ -392,8 +392,10 @@ public: const char * kbuf = (char *) &index; const size_t ksiz = sizeof(phrase_token_t); - const int32_t vsiz = m_db->check(kbuf, ksiz); - if (-1 != vsiz) { /* success */ + int32_t vsiz = m_db->check(kbuf, ksiz); + if (-1 == vsiz) { /* not found. */ + vsiz = sizeof(ArrayHeader); + } else { /* found */ m_chunk.set_size(vsiz); char * vbuf = (char *) m_chunk.begin(); assert(vsiz == m_db->get(kbuf, ksiz, vbuf, vsiz)); diff --git a/utils/training/estimate_k_mixture_model.cpp b/utils/training/estimate_k_mixture_model.cpp index c0fa03f..c95cde2 100644 --- a/utils/training/estimate_k_mixture_model.cpp +++ b/utils/training/estimate_k_mixture_model.cpp @@ -109,10 +109,7 @@ int main(int argc, char * argv[]){ exit(EINVAL); } - /* TODO: magic header signature check here. */ - KMixtureModelBigram unigram(K_MIXTURE_MODEL_MAGIC_NUMBER); - unigram.attach(bigram_filename, ATTACH_READONLY); - + /* magic header signature check here. */ KMixtureModelBigram bigram(K_MIXTURE_MODEL_MAGIC_NUMBER); bigram.attach(bigram_filename, ATTACH_READONLY); @@ -128,7 +125,7 @@ int main(int argc, char * argv[]){ for( size_t i = 0; i < deleted_items->len; ++i ){ phrase_token_t * token = &g_array_index(deleted_items, phrase_token_t, i); KMixtureModelSingleGram * single_gram = NULL; - bigram.load(*token, single_gram); + bigram.load(*token, single_gram, true); KMixtureModelSingleGram * deleted_single_gram = NULL; deleted_bigram.load(*token, deleted_single_gram); @@ -140,7 +137,7 @@ int main(int argc, char * argv[]){ assert(deleted_single_gram->get_array_header(deleted_array_header)); if ( 0 != deleted_array_header.m_WC ) { - parameter_t lambda = compute_interpolation(deleted_single_gram, &unigram, single_gram); + parameter_t lambda = compute_interpolation(deleted_single_gram, &bigram, single_gram); printf("token:%d lambda:%f\n", *token, lambda); |