diff --git a/source/poet.cpp b/source/poet.cpp index bbb69c5..f4d97cf 100644 --- a/source/poet.cpp +++ b/source/poet.cpp @@ -4,6 +4,7 @@ #include #include #include +#include #include #include #include "Rbm.hpp" @@ -11,6 +12,8 @@ #include "Stack.hpp" using namespace std; +using namespace arma; + class RbmListener : public Rbm::IListener { public: @@ -29,6 +32,58 @@ class RbmListener : public Rbm::IListener } }; +const char punctuation[] = {' ', '.', '!', '?', 0}; +const int numCodes = 1 + 26 + 10; + +int char_is(char c, const char *pTable) +{ + int found = 0; + while(*pTable) + { + if (c == *pTable) + { + found = 1; + break; + } + pTable++; + } + return found; +} + +int ch2idx(char c) +{ + c = toupper(c); + + int index = 0; + if (isalpha(c)) + { + index = 1+ (int)(c-'A'); + } + if (isdigit(c)) + { + index = 1+26+(int)(c-'0'); + } + return index; +} + +char idx2ch(int index) +{ + char c = '?'; + if (index == 0) + { + c = ' '; + } + else if (index >= 1 and index < 27) + { + c = (char)(index + 'A' - 1); + } + else if (index >= 27 and index < 38) + { + c = (char)(index + '0' - 27); + } + return c; +} + arma::mat createTraining(const string &filename) { FILE *pFile; @@ -42,9 +97,10 @@ arma::mat createTraining(const string &filename) } int numTraining = 0; - int numVisible = 26 + 10; + int numVisible = 2*numCodes; arma::mat batch = arma::zeros(numTraining, numVisible); + int last_index = ch2idx(' '); while(!feof(pFile)) { char c; @@ -54,20 +110,16 @@ arma::mat createTraining(const string &filename) break; } - c = toupper(c); - - int index; - if (isalpha(c)) + int index = ch2idx(c); + if (index == 0 and last_index == 0) { - index = (int)(c-'A'); - } - if (isdigit(c)) - { - index = (int)(c-'0'); + continue; } arma::mat data = arma::zeros(1, numVisible); - data[index] = 1; - + + data[index+numCodes] = 1; + data[last_index] = 1; + last_index = index; batch.insert_rows(batch.n_rows, data); } fclose(pFile); @@ -75,14 +127,86 @@ arma::mat createTraining(const string &filename) return batch; } +arma::mat to_curr(arma::mat v) +{ + return v.cols(0, numCodes-1); +} + +arma::mat to_next(arma::mat v) +{ + return v.cols(numCodes, 2*numCodes-1); +} + +#define CREATE_TRAINING 0 int main() { - const string project("poet"); - - Stack stack(".", project); - +#if CREATE_TRAINING arma::mat batch = createTraining("moby_ch1.txt"); batch.save("poet.training.dat", arma::arma_ascii); +#endif + + Stack stack(".", "poet2"); + + // Load project + stack.load(); + + // Load weights + stack.loadWeights(); + + // Load training + stack.loadTrainingBatch(); + + Layer *layer = stack.getLayer(0); + int numTraining = stack.trainingBatch().n_rows; + + arma::mat v = stack.trainingBatch(); + + arma::mat h = layer->toHiddenProbs(v); + arma::mat r = layer->toVisibleProbs(h); + + r = layer->upDownPass(v); + layer->params().gibbsDoSampleHidden = false; + layer->params().gibbsDoSampleVisible = false; + layer->params().numGibbs = 5; + + printf("\nStimulus: Training\n"); + for (int i=0; i < numTraining; i++) + { + arma::mat curr = to_curr(v.row(i)); + char c = idx2ch(curr.index_max()); + putchar(c); + } + + h = layer->toHiddenProbs(v.row(0)); + printf("\nStimulus: Next char and current h\n"); + for (int i=1; i < numTraining; i++) + { + layer->gibbs_hv(h, r); + arma::mat curr = to_curr(r); + arma::mat next = to_next(r); + v = arma::join_rows(next, arma::zeros(1, next.n_cols), h); + + layer->gibbs_vh(v, h); + + char c = idx2ch(curr.index_max()); + putchar(c); + } + + h = layer->toHiddenProbs(v.row(0)); + printf("\nStimulus: Current h\n"); + for (int i=1; i < numTraining; i++) + { + layer->gibbs_hv(h, r); + arma::mat curr = to_curr(r); + arma::mat next = to_next(r); + v = arma::join_rows(arma::zeros(1, next.n_cols), arma::zeros(1, next.n_cols), h); + layer->gibbs_vh(v, h); + + char c = idx2ch(curr.index_max()); + putchar(c); + } + + printf("\n\nEnd of program\n"); return 0; }