diff --git a/MLP/include/MLP.hpp b/MLP/include/MLP.hpp
index 479f403d1cf54f293c215b7d78c672d45115901f..68f1849caab821449cac0b1f0f5b18705c798861 100644
--- a/MLP/include/MLP.hpp
+++ b/MLP/include/MLP.hpp
@@ -81,15 +81,10 @@ class MLP
private :
- /// @brief The maximum number of parameters of the model.
- static const unsigned int MAXLOOKUPSIZE = 200000;
-
/// @brief The Layers of the MLP.
std::vector<Layer> layers;
/// @brief The parameters corresponding to the layers of the MLP.
std::vector< std::vector<dynet::Parameter> > parameters;
- /// @brief The parameters corresponding to Dict values.
- std::map< Dict*, std::pair<dynet::LookupParameter, std::map<void*, unsigned int> > > lookupParameters;
/// @brief The dynet model containing the parameters to be trained.
dynet::ParameterCollection model;
@@ -172,7 +167,7 @@ class MLP
public :
- /// @brief Construct a new untrained MLP from a desired topology.
+ /// @brief initialize a new untrained MLP from a desired topology.
///
/// topology example for 2 hidden layers : (150,RELU,0.3)(50,ELU,0.2)\n
/// Of sizes 150 and 50, activation functions RELU and ELU, and dropout rates
@@ -180,7 +175,9 @@ class MLP
/// @param nbInputs The size of the input layer of the MLP.
/// @param topology Description of each hidden Layer of the MLP.
/// @param nbOutputs The size of the output layer of the MLP.
- MLP(int nbInputs, const std::string & topology, int nbOutputs);
+ void init(int nbInputs, const std::string & topology, int nbOutputs);
+ /// @brief Construct a new MLP for training.
+ MLP();
/// @brief Read and construct a trained MLP from a file.
///
/// The file must have been written by save.
@@ -211,6 +208,10 @@ class MLP
///
/// @return A pointer to the newly allocated trainer.
dynet::Trainer * createTrainer();
+ /// @brief Return the model.
+ ///
+ /// @return The model of this MLP.
+ dynet::ParameterCollection & getModel();
};
#endif
diff --git a/MLP/src/MLP.cpp b/MLP/src/MLP.cpp
index 7b64414ce9ae7fa1eca0a3fac7cc78ede0c87907..e53c198e43b150e7f856b58d5dd8e822ef4e5caa 100644
--- a/MLP/src/MLP.cpp
+++ b/MLP/src/MLP.cpp
@@ -80,14 +80,20 @@ void MLP::initDynet()
dynet::initialize(getDefaultParams());
}
-MLP::MLP(int nbInputs, const std::string & topology, int nbOutputs)
+MLP::MLP()
{
randomSeed = ProgramParameters::seed;
+ trainMode = true;
+ dropoutActive = true;
+ trainer.reset(createTrainer());
+ initDynet();
+}
+
+void MLP::init(int nbInputs, const std::string & topology, int nbOutputs)
+{
std::string topo = topology;
std::replace(topo.begin(), topo.end(), '(', ' ');
std::replace(topo.begin(), topo.end(), ')', ' ');
- trainMode = true;
- dropoutActive = true;
auto groups = split(topo);
for (auto group : groups)
@@ -112,10 +118,6 @@ MLP::MLP(int nbInputs, const std::string & topology, int nbOutputs)
layers.emplace_back(layers.back().output_dim, nbOutputs, 0.0, Activation::LINEAR);
- trainer.reset(createTrainer());
-
- initDynet();
-
checkLayersCompatibility();
for(Layer layer : layers)
@@ -225,48 +227,15 @@ dynet::DynetParams & MLP::getDefaultParams()
dynet::Expression MLP::featValue2Expression(dynet::ComputationGraph & cg, const FeatureModel::FeatureValue & fv)
{
Dict * dict = fv.dict;
- std::vector<float> * value = dict->getValue(*fv.value);
-
- auto entry = lookupParameters.find(dict);
-
- if(entry == lookupParameters.end())
- {
- lookupParameters[dict].first = model.add_lookup_parameters(MAXLOOKUPSIZE, {(unsigned)dict->getDimension(),1});
- }
-
- auto & ptr2index = lookupParameters[dict].second;
- auto & lu = lookupParameters[dict].first;
-
bool isConst = (fv.policy == FeatureModel::Policy::Final) || (dict->mode == Dict::Mode::OneHot);
- auto it = ptr2index.find(value);
-
- if(it != ptr2index.end())
- {
- if(isConst)
- return dynet::const_lookup(cg, lu, it->second);
- else
- return dynet::lookup(cg, lu, it->second);
- }
-
- ptr2index[value] = ptr2index.size();
- it = ptr2index.find(value);
-
- unsigned int lookupSize = (int)(*lu.values()).size();
- if(it->second >= lookupSize)
- {
- fprintf(stderr, "ERROR (%s) : MAXLOOKUPSIZE (%d) is too small. Aborting.\n", ERRINFO, MAXLOOKUPSIZE);
- exit(1);
- }
-
- // Horrible trick : directly set Dict data as Tensor values
- // Works only on CPU
- (*lu.values())[it->second].v = value->data();
+ auto & lu = dict->getLookupParameter();
+ unsigned int index = dict->getValue(*fv.value);
if(isConst)
- return dynet::const_lookup(cg, lu, it->second);
+ return dynet::const_lookup(cg, lu, index);
else
- return dynet::lookup(cg, lu, it->second);
+ return dynet::lookup(cg, lu, index);
}
dynet::Expression MLP::run(dynet::ComputationGraph & cg, dynet::Expression x)
@@ -432,3 +401,8 @@ void MLP::printTopology(FILE * output)
fprintf(output, ")\n");
}
+dynet::ParameterCollection & MLP::getModel()
+{
+ return model;
+}
+
diff --git a/decoder/include/Decoder.hpp b/decoder/include/Decoder.hpp
index dc8bef588841a0b6cb99a1a30ebcce3f6de5b0fb..b905f6eb39512496bb403109281f3d22541c4558 100644
--- a/decoder/include/Decoder.hpp
+++ b/decoder/include/Decoder.hpp
@@ -7,7 +7,6 @@
#define DECODER__H
#include "TransitionMachine.hpp"
-#include "BD.hpp"
#include "Config.hpp"
/// @brief A simple object capable of using a trained TransitionMachine to process a given BD.
@@ -17,8 +16,6 @@ class Decoder
/// @brief The trained TransitionMachine
TransitionMachine & tm;
- /// @brief The BD we need to fill
- BD & bd;
/// @brief The current configuration of the TransitionMachine
Config & config;
/// @brief is true, decode will print infos on stderr
@@ -34,7 +31,7 @@ class Decoder
/// @param bd The BD we need to fill
/// @param config The current configuration of the TransitionMachine
/// @param debugMode If true, infos will be printed on stderr.
- Decoder(TransitionMachine & tm, BD & bd, Config & config, bool debugMode);
+ Decoder(TransitionMachine & tm, Config & config, bool debugMode);
/// @brief Fill bd using tm.
void decode();
};
diff --git a/decoder/src/Decoder.cpp b/decoder/src/Decoder.cpp
index 78acd26fc22e0a76104b91a34354a64b8dc9c8ba..2bf8d89664409a2082dd2c578591dc1ab999b6d7 100644
--- a/decoder/src/Decoder.cpp
+++ b/decoder/src/Decoder.cpp
@@ -1,8 +1,8 @@
#include "Decoder.hpp"
#include "util.hpp"
-Decoder::Decoder(TransitionMachine & tm, BD & bd, Config & config, bool debugMode)
-: tm(tm), bd(bd), config(config)
+Decoder::Decoder(TransitionMachine & tm, Config & config, bool debugMode)
+: tm(tm), config(config)
{
this->debugMode = debugMode;
}
diff --git a/decoder/src/macaon_decode.cpp b/decoder/src/macaon_decode.cpp
index ab4ecb55d61c7e4fa3a9d5eb5c3df7c4c2dc5451..5611e373007f34f83e52aedd0447c1ca01d5c87d 100644
--- a/decoder/src/macaon_decode.cpp
+++ b/decoder/src/macaon_decode.cpp
@@ -116,7 +116,7 @@ int main(int argc, char * argv[])
Config config(bd);
config.readInput(ProgramParameters::input);
- Decoder decoder(tapeMachine, bd, config, ProgramParameters::debug);
+ Decoder decoder(tapeMachine, config, ProgramParameters::debug);
decoder.decode();
diff --git a/maca_common/include/Dict.hpp b/maca_common/include/Dict.hpp
index 68de3278de280fab70e74ea837bd9e48824bfb38..cdf795846c4988b923474db0e1e91101dbf729f3 100644
--- a/maca_common/include/Dict.hpp
+++ b/maca_common/include/Dict.hpp
@@ -12,12 +12,13 @@
#include <set>
#include <memory>
#include <fasttext/fasttext.h>
+#include <dynet/dynet.h>
/// @brief Maps strings to real vectors (embeddings or one-hot encoded).
///
-/// Each entry of the Dict is a pair of string / vector.\n
-/// The strings can be seen as features, while the vector are the numerical value
-/// of this feature.
+/// Each entry of the Dict is a pair of string / Index.\n
+/// The strings can be seen as features, while the index
+/// points to the corresponding embedding value.
class Dict
{
public :
@@ -94,10 +95,15 @@ class Dict
private :
+ /// @brief The maximum number of entry a Dict can hold.
+ ///
+ /// This limit exists because the dynet LookupParameter associed with the Dict is not dynamic and must come with a fixed finite size.
+ static constexpr unsigned int MAX_CAPACITY = 200000;
/// @brief The dimension of each vector of this Dict (in number of float).
int dimension;
- /// @brief A storage that map every string entry of this Dict to its vector value
- std::map< std::string, std::vector<float> > str2vec;
+ /// @brief A storage that map every string entry of this Dict to its index as a lookup parameter.
+ std::map<std::string, unsigned int> str2index;
+ dynet::LookupParameter lookupParameter;
/// @brief The filename to which this Dict will be saved.
std::string filename;
/// @brief When in OneHot Mode, the Dict uses this variable to initialize new entries' vector to a unique OneHot value.
@@ -121,6 +127,7 @@ class Dict
/// When getDict is called it will find the requested Dict here,
/// or construct it.
static std::map< std::string, std::unique_ptr<Dict> > str2dict;
+ bool isInit;
private :
@@ -129,30 +136,36 @@ class Dict
/// @param s An entry of this Dict.
///
/// @return The FastText's pre-trained word embedding value corresponding to s.
- std::vector<float> * getValueFasttext(const std::string & s);
+ unsigned int getValueFasttext(const std::string & s);
/// @brief Initialize a new embedding for this Dict.
///
/// This function zero-initialize the embedding.
/// @param s The entry whose embedding we will initialize.
- /// @param vec The vector containing the embedding we will initialize.
- void initEmbedding(const std::string & s, std::vector<float> & vec);
+ /// @param index The index of the entry.
+ void initParameterAsEmbedding(const std::string & s, unsigned int index);
+ /// @brief Initialize a parameter as a one hot value
+ ///
+ /// @param s The entry whose embedding we will initialize.
+ /// @param index The index of the parameter to initialize.
+ void initParameterAsOneHot(const std::string & s, unsigned int index);
/// @brief Randomly initialize a new embedding for this Dict.
///
/// Each float value of vec will be randomly set between -1.0 and +1.0
- /// @param vec The vector containing the embedding we will initialize.
- void initEmbeddingRandom(std::vector<float> & vec);
+ /// @param index The index of the parameter to initialize.
+ void initEmbeddingRandom(unsigned int index);
/// @brief Initialize a new embedding using FastText's value for this string.
///
/// @param s The entry whose embedding we will initialize.
- /// @param vec The vector containing the embedding we will initialize.
- void initEmbeddingFromFasttext(const std::string & s, std::vector<float> & vec);
+ /// @param index The index of the parameter.
+ void initEmbeddingFromFasttext(const std::string & s, unsigned int index);
/// @brief Add a new entry to this Dict.
///
/// @param s The new entry to add.
///
- /// @return The vector corresponding to the newly added entry.
- std::vector<float> * addEntry(const std::string & s);
-
+ /// @return The lookupParameter index of the newly added entry.
+ unsigned int addEntry(const std::string & s);
+ void init(dynet::ParameterCollection & pc);
+ void initFromFile(dynet::ParameterCollection & pc);
/// @brief Read and construct a new Dict from a file.
///
/// @param policy The Policy of the new Dict.
@@ -174,6 +187,8 @@ class Dict
///
/// @return A pointer to an entry matching the value of s.
const std::string * getStrFasttext(const std::string & s);
+ /// @brief Create file.
+ void createFile();
public :
@@ -203,16 +218,23 @@ class Dict
/// @param directory The directory in which we will save every Dict.
/// @param namePrefix The prefix of the name of the dicts we need to save.
static void saveDicts(const std::string & directory, const std::string & namePrefix);
+ /// @brief Create a file for every Dict matching the prefix.
+ ///
+ /// @param directory The directory in which the files will be saved.
+ /// @param namePrefix The prefix that Dict must match.
+ static void createFiles(const std::string & directory, const std::string & namePrefix);
+ static void initDicts(dynet::ParameterCollection & pc, const std::string & namePrefix);
+ static void initDictsFromFile(dynet::ParameterCollection & pc, const std::string & namePrefix);
/// @brief Delete all Dicts.
static void deleteDicts();
/// @brief Save the current Dict in the corresponding file.
void save();
- /// @brief Get the vector value of an entry.
+ /// @brief Get the index.
///
/// @param s The entry which value we need.
///
- /// @return The vector value corresponding to s.
- std::vector<float> * getValue(const std::string & s);
+ /// @return The index of the lookupParameter corresponding to s.
+ unsigned int getValue(const std::string & s);
/// @brief Get a pointer to the entry matching s.
///
/// This is used when we need a permanent pointer to a string matching s,
@@ -222,14 +244,14 @@ class Dict
///
/// @return A pointer to an entry matching the value of s.
const std::string * getStr(const std::string & s);
- /// @brief Get the vector value of unknown values.
+ /// @brief Get the index of unknown values.
///
- /// @return Pointer to the vector value of the entry unknownValueStr.
- std::vector<float> * getUnknownValue();
- /// @brief Get the vector value of null values.
+ /// @return Index of the entry unknownValueStr.
+ unsigned int getUnknownValue();
+ /// @brief Get the index of null values.
///
- /// @return Pointer to the vector value of the entry nullValueStr.
- std::vector<float> * getNullValue();
+ /// @return Index of the entry nullValueStr.
+ unsigned int getNullValue();
/// @brief Get the dimmension of the vectors of this Dict.
///
/// @return Dimmension of the vectors of this Dict.
@@ -238,6 +260,7 @@ class Dict
///
/// @param output The FILE to which the Dict will be printed.
void printForDebug(FILE * output);
+ dynet::LookupParameter & getLookupParameter();
};
#endif
diff --git a/maca_common/include/ProgramParameters.hpp b/maca_common/include/ProgramParameters.hpp
index 0ee0d9df5f9829a94b4f0081c8252e0cff7f5af1..e47dd6959084148a2361b1232b680a75e9864283 100644
--- a/maca_common/include/ProgramParameters.hpp
+++ b/maca_common/include/ProgramParameters.hpp
@@ -44,6 +44,7 @@ struct ProgramParameters
static bool showFeatureRepresentation;
static int iterationSize;
static int nbTrain;
+ static bool randomEmbeddings;
private :
diff --git a/maca_common/include/util.hpp b/maca_common/include/util.hpp
index a3540d17beacc839b06437e5eacc3a5c0d1fea15..fc012e7f77bbc0fe3dfe25e51f0dc2497c318182 100644
--- a/maca_common/include/util.hpp
+++ b/maca_common/include/util.hpp
@@ -155,6 +155,13 @@ void printColumns(FILE * output, const std::vector< std::vector<std::string> > &
///
/// @return The formated string.
std::string float2str(float f, const char * format);
+/// @brief Return a version of s not suffixed by suffix.
+///
+/// @param s The base string.
+/// @param suffix The suffix to remove.
+///
+/// @return s without suffix at the end.
+std::string removeSuffix(const std::string & s, const std::string & suffix);
/// @brief Macro giving informations about an error.
#define ERRINFO (getFilenameFromPath(std::string(__FILE__))+ ":l." + std::to_string(__LINE__)).c_str()
diff --git a/maca_common/src/Dict.cpp b/maca_common/src/Dict.cpp
index 3e6f655c673b9f06c466166029704488ce57ec86..4096b470a8ee55c3b73059880ee133da00f65d53 100644
--- a/maca_common/src/Dict.cpp
+++ b/maca_common/src/Dict.cpp
@@ -2,6 +2,7 @@
#include "File.hpp"
#include "util.hpp"
#include "ProgramParameters.hpp"
+#include <dynet/io.h>
std::string Dict::currentClassifierName;
std::string Dict::nullValueStr = "_nullVALUEstr_";
@@ -46,6 +47,7 @@ const char * Dict::mode2str(Mode mode)
Dict::Dict(const std::string & name, int dimension, Mode mode)
{
+ this->isInit = false;
this->policy = Policy::FromZero;
this->filename = name;
this->name = name;
@@ -53,12 +55,32 @@ Dict::Dict(const std::string & name, int dimension, Mode mode)
this->mode = mode;
this->dimension = dimension;
+}
+
+Dict::Dict(Policy policy, const std::string & filename)
+{
+ this->isInit = false;
+ this->policy = policy;
+ this->filename = filename;
+ this->oneHotIndex = 0;
+ this->name = removeSuffix(getFilenameFromPath(filename), ".dict");
+}
+void Dict::init(dynet::ParameterCollection & pc)
+{
+ if (isInit)
+ {
+ fprintf(stderr, "ERROR (%s) : Dict is already initialized. Aborting.\n", ERRINFO);
+ exit(1);
+ }
+
+ isInit = true;
+ this->lookupParameter = pc.add_lookup_parameters(MAX_CAPACITY, {(unsigned int)dimension});
addEntry(nullValueStr);
addEntry(unknownValueStr);
}
-Dict::Dict(Policy policy, const std::string & filename)
+void Dict::initFromFile(dynet::ParameterCollection & pc)
{
auto badFormatAndAbort = [&](std::string errInfo)
{
@@ -66,12 +88,14 @@ Dict::Dict(Policy policy, const std::string & filename)
exit(1);
};
- this->policy = policy;
- this->filename = filename;
- this->oneHotIndex = 0;
+ if (isInit)
+ {
+ fprintf(stderr, "ERROR (%s) : Dict is already initialized. Aborting.\n", ERRINFO);
+ exit(1);
+ }
- File file(filename, "r");
- FILE * fd = file.getDescriptor();
+ File * file = new File(filename, "r");
+ FILE * fd = file->getDescriptor();
char b1[1024];
char b2[1024];
@@ -82,8 +106,7 @@ Dict::Dict(Policy policy, const std::string & filename)
name = b1;
mode = str2mode(b2);
- addEntry(nullValueStr);
- addEntry(unknownValueStr);
+ isInit = true;
// If a fasttext pretrained embedding file is specified
if(fscanf(fd, "Fasttext : %s\n", b1) == 1)
@@ -107,89 +130,107 @@ Dict::Dict(Policy policy, const std::string & filename)
if(this->policy == Policy::FromZero)
return;
- while(fscanf(fd, "\1%[^\1]\1", b1) == 1)
- {
- std::string entry(b1);
+ int readIndex;
+ while (fscanf(fd, "\1%[^\1]\1%d\n", b1, &readIndex) == 2)
+ str2index[b1] = readIndex;
- str2vec[entry] = std::vector<float>();
- auto & vec = str2vec[entry];
+ delete file;
- // For OneHot we only write the index
- if(mode == Mode::Embeddings)
- for(int i = 0; i < dimension; i++)
- {
- float value;
- if(fscanf(fd, "%f", &value) != 1)
- badFormatAndAbort(ERRINFO);
- vec.emplace_back(value);
- }
- else
- {
- int index;
- if(fscanf(fd, "%d", &index) != 1)
- badFormatAndAbort(ERRINFO);
+ dynet::TextFileLoader loader(filename);
+ lookupParameter = loader.load_lookup_param(pc, "lookup");
+}
- vec.resize(dimension, 0.0);
- vec[index] = 1.0;
+void Dict::saveDicts(const std::string & directory, const std::string & namePrefix)
+{
+ for (auto & it : str2dict)
+ {
+ if(!strncmp(it.first.c_str(), namePrefix.c_str(), namePrefix.size()))
+ {
+ it.second->filename = directory + it.second->name + ".dict";
+ it.second->save();
}
- if(fscanf(fd, "%*[^\n]\n"))
- badFormatAndAbort(ERRINFO);
}
}
-void Dict::saveDicts(const std::string & directory, const std::string & namePrefix)
+void Dict::createFiles(const std::string & directory, const std::string & namePrefix)
{
for (auto & it : str2dict)
{
if(!strncmp(it.first.c_str(), namePrefix.c_str(), namePrefix.size()))
{
it.second->filename = directory + it.second->name + ".dict";
- it.second->save();
+ it.second->createFile();
+ }
+ }
+}
+
+void Dict::initDicts(dynet::ParameterCollection & pc, const std::string & namePrefix)
+{
+ for (auto & it : str2dict)
+ {
+ if(!strncmp(it.first.c_str(), namePrefix.c_str(), namePrefix.size()))
+ {
+ it.second->init(pc);
+ }
+ }
+}
+
+void Dict::initDictsFromFile(dynet::ParameterCollection & pc, const std::string & namePrefix)
+{
+ for (auto & it : str2dict)
+ {
+ if(!strncmp(it.first.c_str(), namePrefix.c_str(), namePrefix.size()))
+ {
+ it.second->initFromFile(pc);
}
}
}
+void Dict::createFile()
+{
+ // If policy is Final, we didn't change any entry so no need to rewrite the file
+ if (policy == Policy::Final)
+ return;
+
+ File * file = new File(filename, "w");
+ FILE * fd = file->getDescriptor();
+
+ fprintf(fd, "%s\n%d\n%s\n", name.c_str(), dimension, mode2str(mode));
+
+ if(ftEmbeddings.get())
+ fprintf(fd, "Fasttext : %s\n", ftFilename.c_str());
+
+ delete file;
+}
+
void Dict::save()
{
// If policy is Final, we didn't change any entry so no need to rewrite the file
if (policy == Policy::Final)
return;
- File file(filename, "w");
- FILE * fd = file.getDescriptor();
+ File * file = new File(filename, "w");
+ FILE * fd = file->getDescriptor();
fprintf(fd, "%s\n%d\n%s\n", name.c_str(), dimension, mode2str(mode));
if(ftEmbeddings.get())
fprintf(fd, "Fasttext : %s\n", ftFilename.c_str());
- for(auto & it : str2vec)
- {
- fprintf(fd, "\1%s\1\t", it.first.c_str());
+ for (auto & it : str2index)
+ fprintf(fd, "\1%s\1%u\n", it.first.c_str(), it.second);
- // For OneHot we only write the index
- if(mode == Mode::Embeddings)
- for(float value : it.second)
- fprintf(fd, "%f\t", value);
- else
- {
- for(unsigned int index = 0; index < it.second.size(); index++)
- if(it.second[index] > 0.1)
- {
- fprintf(fd, "%d", index);
- break;
- }
- }
+ delete file;
- fprintf(fd, "\n");
- }
+ dynet::TextFileSaver s(filename, true);
+ s.save(lookupParameter, "lookup");
}
-std::vector<float> * Dict::getValue(const std::string & s)
+unsigned int Dict::getValue(const std::string & s)
{
- auto it = str2vec.find(s);
- if(it != str2vec.end())
- return &(it->second);
+ auto it = str2index.find(s);
+ if(it != str2index.end())
+ return it->second;
if(policy == Policy::Final)
{
@@ -202,30 +243,18 @@ std::vector<float> * Dict::getValue(const std::string & s)
return addEntry(s);
}
-std::vector<float> * Dict::getValueFasttext(const std::string & s)
+unsigned int Dict::getValueFasttext(const std::string &)
{
- auto it = ftVocab.find(s);
- if(it != ftVocab.end())
- return &(it->second);
-
- if(s.empty())
- {
- fprintf(stderr, "ERROR (%s) : dict \'%s\' was asked to store an empty entry. Aborting.\n", ERRINFO, name.c_str());
- exit(1);
- }
-
- ftVocab.emplace(s, std::vector<float>(dimension, 0.0));
- auto & vec = ftVocab[s];
- ftEmbeddings->getWordVector(*ftVector.get(), s);
- memcpy(vec.data(), ftVector.get()->data(), dimension * sizeof vec[0]);
+ fprintf(stderr, "ERROR (%s) : Not implemented.\n", ERRINFO);
+ exit(1);
- return &vec;
+ return 0;
}
const std::string * Dict::getStr(const std::string & s)
{
- auto it = str2vec.find(s);
- if(it != str2vec.end())
+ auto it = str2index.find(s);
+ if(it != str2index.end())
return &(it->first);
if(policy == Policy::Final)
@@ -237,7 +266,7 @@ const std::string * Dict::getStr(const std::string & s)
addEntry(s);
- it = str2vec.find(s);
+ it = str2index.find(s);
return &(it->first);
}
@@ -263,17 +292,24 @@ const std::string * Dict::getStrFasttext(const std::string & s)
return &(it->first);
}
-void Dict::initEmbedding(const std::string & s, std::vector<float> & vec)
+void Dict::initParameterAsOneHot(const std::string &, unsigned int index)
+{
+ std::vector<float> vec(dimension, 0.0);
+ vec[index] = 1.0;
+ lookupParameter.initialize(index, vec);
+}
+
+void Dict::initParameterAsEmbedding(const std::string & s, unsigned int index)
{
- vec[0] = 0.0; // just to shut warning up
- // Here initialize a new embedding, doing nothing = all zeroes
- //initEmbeddingRandom(vec);
+ if (ProgramParameters::randomEmbeddings)
+ initEmbeddingRandom(index);
if(ftEmbeddings.get())
- initEmbeddingFromFasttext(s, vec);
+ initEmbeddingFromFasttext(s, index);
}
-void Dict::initEmbeddingRandom(std::vector<float> & vec)
+void Dict::initEmbeddingRandom(unsigned int index)
{
+ std::vector<float> vec(dimension, 0.0);
int range = 1;
for (auto & val : vec)
@@ -284,48 +320,57 @@ void Dict::initEmbeddingRandom(std::vector<float> & vec)
result += decimal;
val = result;
}
+
+ lookupParameter.initialize(index, vec);
}
-void Dict::initEmbeddingFromFasttext(const std::string & s, std::vector<float> & vec)
+void Dict::initEmbeddingFromFasttext(const std::string & s, unsigned int index)
{
ftEmbeddings->getWordVector(*ftVector.get(), s);
- memcpy(vec.data(), ftVector.get()->data(), dimension * sizeof vec[0]);
+ lookupParameter.initialize(index, *(std::vector<float>*)ftVector.get());
}
-std::vector<float> * Dict::getUnknownValue()
+unsigned int Dict::getUnknownValue()
{
- return &str2vec[unknownValueStr];
+ return str2index[unknownValueStr];
}
-std::vector<float> * Dict::getNullValue()
+unsigned int Dict::getNullValue()
{
- return &str2vec[nullValueStr];
+ return str2index[nullValueStr];
}
-std::vector<float> * Dict::addEntry(const std::string & s)
+unsigned int Dict::addEntry(const std::string & s)
{
+ if (!isInit)
+ {
+ fprintf(stderr, "ERROR (%s) : dict \'%s\' is not initialized. Aborting.\n", ERRINFO, name.c_str());
+ exit(1);
+ }
+
if(s.empty())
{
fprintf(stderr, "ERROR (%s) : dict \'%s\' was asked to store an empty entry. Aborting.\n", ERRINFO, name.c_str());
exit(1);
}
- str2vec.emplace(s, std::vector<float>(dimension, 0.0));
- auto & vec = str2vec[s];
+ auto index = str2index.size();
+ str2index.emplace(s, index);
if(mode == Mode::OneHot)
{
if(oneHotIndex >= dimension)
fprintf(stderr, "WARNING (%s) : Dict %s of dimension %d is asked to store %d elements in one-hot.\n", ERRINFO, name.c_str(), dimension, oneHotIndex+1);
else
- vec[oneHotIndex] = 1.0;
-
- oneHotIndex++;
+ {
+ initParameterAsOneHot(s, index);
+ oneHotIndex++;
+ }
}
else
- initEmbedding(s, vec);
+ initParameterAsEmbedding(s, index);
- return &vec;
+ return index;
}
Dict * Dict::getDict(Policy policy, const std::string & filename)
@@ -338,7 +383,7 @@ Dict * Dict::getDict(Policy policy, const std::string & filename)
str2dict.insert(std::make_pair(dict->name, std::unique_ptr<Dict>(dict)));
- return str2dict[filename].get();
+ return str2dict[dict->name].get();
}
Dict * Dict::getDict(const std::string & name)
@@ -413,7 +458,7 @@ int Dict::getDimension()
void Dict::printForDebug(FILE * output)
{
- fprintf(output, "Dict name \'%s\' nbElems = %lu\n", name.c_str(), str2vec.size());
+ fprintf(output, "Dict name \'%s\' nbElems = %lu\n", name.c_str(), str2index.size());
}
void Dict::deleteDicts()
@@ -421,3 +466,8 @@ void Dict::deleteDicts()
str2dict.clear();
}
+dynet::LookupParameter & Dict::getLookupParameter()
+{
+ return lookupParameter;
+}
+
diff --git a/maca_common/src/ProgramParameters.cpp b/maca_common/src/ProgramParameters.cpp
index b505d043058734022436c33fea72cd8fdf6cde22..e5835a6b594dc24e4d293740c2109ed22d8cb99b 100644
--- a/maca_common/src/ProgramParameters.cpp
+++ b/maca_common/src/ProgramParameters.cpp
@@ -37,5 +37,6 @@ bool ProgramParameters::interactive;
int ProgramParameters::dynamicEpoch;
float ProgramParameters::dynamicProbability;
bool ProgramParameters::showFeatureRepresentation;
+bool ProgramParameters::randomEmbeddings;
int ProgramParameters::iterationSize;
int ProgramParameters::nbTrain;
diff --git a/maca_common/src/util.cpp b/maca_common/src/util.cpp
index 368cec7d2494ce6901cbe311ac971129bf6c7e3f..6d8fc121e8a0f7742aa8a34768f660c6df20780b 100644
--- a/maca_common/src/util.cpp
+++ b/maca_common/src/util.cpp
@@ -349,3 +349,14 @@ bool isNum(const std::string & s)
return true;
}
+std::string removeSuffix(const std::string & s, const std::string & suffix)
+{
+ int lastIndex = s.size()-1;
+ while (lastIndex >= 0 && s[lastIndex] == suffix[suffix.size()-1-(s.size()-1-lastIndex)])
+ lastIndex--;
+
+ std::string result = std::string(s.begin(), s.begin()+lastIndex+1);
+
+ return result;
+}
+
diff --git a/trainer/src/Trainer.cpp b/trainer/src/Trainer.cpp
index e9a7f751fc43c4964e75729a32bedf923c1f2adc..7b255df6aad6c554ab7f6aa6e208957d8655293b 100644
--- a/trainer/src/Trainer.cpp
+++ b/trainer/src/Trainer.cpp
@@ -94,7 +94,7 @@ std::map<std::string, float> Trainer::getScoreOnDev()
void Trainer::train()
{
- Dict::saveDicts(ProgramParameters::expPath, "");
+ Dict::createFiles(ProgramParameters::expPath, "");
fprintf(stderr, "Training of \'%s\' :\n", tm.name.c_str());
diff --git a/trainer/src/macaon_train.cpp b/trainer/src/macaon_train.cpp
index 69d608100a5d707f17510be3512d42a0fffd01d3..0c37b792e13cdef159bb870bd22e8d9ca1546e7e 100644
--- a/trainer/src/macaon_train.cpp
+++ b/trainer/src/macaon_train.cpp
@@ -62,6 +62,8 @@ po::options_description getOptionsDescription()
"For each state of the Config, show its feature representation")
("interactive", po::value<bool>()->default_value(true),
"Is the shell interactive ? Display advancement informations")
+ ("randomEmbeddings", po::value<bool>()->default_value(false),
+ "When activated, the embeddings will be randomly initialized")
("shuffle", po::value<bool>()->default_value(true),
"Shuffle examples after each iteration");
@@ -160,7 +162,7 @@ do\n\
ARGS=\"$ARGS $arg\"\n\
done\n\
\n\
-macaon_decode --lang $LANG --tm machine.tm --bd test.bd -I $INPUT --mcd $MCD --expName " + ProgramParameters::expName + "$ARGS";
+macaon_decode --lang " + ProgramParameters::lang + " --tm machine.tm --bd test.bd -I $INPUT --mcd $MCD --expName " + ProgramParameters::expName + "$ARGS";
if (system(("rm -r " + ProgramParameters::expPath + " 2> /dev/null").c_str())){}
if (system(("mkdir " + ProgramParameters::expPath).c_str())){}
@@ -239,6 +241,7 @@ int main(int argc, char * argv[])
ProgramParameters::removeDuplicates = vm["duplicates"].as<bool>();
ProgramParameters::interactive = vm["interactive"].as<bool>();
ProgramParameters::shuffleExamples = vm["shuffle"].as<bool>();
+ ProgramParameters::randomEmbeddings = vm["randomEmbeddings"].as<bool>();
ProgramParameters::learningRate = vm["lr"].as<float>();
ProgramParameters::beta1 = vm["b1"].as<float>();
ProgramParameters::beta2 = vm["b2"].as<float>();
diff --git a/transition_machine/include/TransitionMachine.hpp b/transition_machine/include/TransitionMachine.hpp
index a990d63fea9d29791fb79ead066515353071ac73..d9ceaae4e114c8c819718127c36a0ebf89eeae55 100644
--- a/transition_machine/include/TransitionMachine.hpp
+++ b/transition_machine/include/TransitionMachine.hpp
@@ -18,7 +18,7 @@ class TransitionMachine
{
public :
- class State;
+ struct State;
/// @brief A Transition from one state to another.
struct Transition
diff --git a/transition_machine/src/Classifier.cpp b/transition_machine/src/Classifier.cpp
index b26490b363f1386910400e042d1d6c0bc471ae5e..3816df5a98a954d0b16e85c4e0a90f52451f53d8 100644
--- a/transition_machine/src/Classifier.cpp
+++ b/transition_machine/src/Classifier.cpp
@@ -123,18 +123,23 @@ void Classifier::initClassifier(Config & config)
if(!trainMode)
{
mlp.reset(new MLP(ProgramParameters::expPath + name + ".model"));
+ Dict::initDictsFromFile(mlp->getModel(), name);
return;
}
- int nbInputs = 0;
- int nbOutputs = as->actions.size();
+ mlp.reset(new MLP());
+
+ Dict::initDicts(mlp->getModel(), name);
auto fd = fm->getFeatureDescription(config);
+ int nbInputs = 0;
+ int nbOutputs = as->actions.size();
+
for (auto feat : fd.values)
nbInputs += feat.dict->getDimension();
- mlp.reset(new MLP(nbInputs, topology, nbOutputs));
+ mlp->init(nbInputs, topology, nbOutputs);
}
FeatureModel::FeatureDescription Classifier::getFeatureDescription(Config & config)
diff --git a/transition_machine/src/FeatureModel.cpp b/transition_machine/src/FeatureModel.cpp
index 4e671a920a32679ddb129a5fe352222fe04aa00f..52fa264416c2910df232938e95e6d18b4ff5a39d 100644
--- a/transition_machine/src/FeatureModel.cpp
+++ b/transition_machine/src/FeatureModel.cpp
@@ -46,10 +46,13 @@ std::string FeatureModel::FeatureValue::toString()
{
std::string result;
- auto realVector = dict->getValue(*value);
+ unsigned int index = dict->getValue(*value);
+ float * realVector = (*dict->getLookupParameter().values())[index].batch_ptr(0);
- for (auto & f : *realVector)
- result += " " + float2str(f, "%5.2f");
+ unsigned int dim = dict->getDimension();
+
+ for (unsigned int i = 0; i < dim; i++)
+ result += " " + float2str(realVector[i], "%5.2f");
return result;
}