static public string Call(List <string> inputFeatureGroups)
{
List <List <List <string> > > groupBatchTokens = new List <List <List <string> > >();
foreach (var inputFeatureGroup in inputFeatureGroups)
{
List <string> tokens = inputFeatureGroup.Split(' ').ToList();
List <List <string> > batchTokens = new List <List <string> >();
batchTokens.Add(tokens);
groupBatchTokens.Add(batchTokens);
}
DecodingOptions decodingOptions = opts.CreateDecodingOptions();
List <NetworkResult> nrs = m_seqClassification.Test <SeqClassificationMultiTasksCorpusBatch>(groupBatchTokens, null, decodingOptions);
List <string> tags = new List <string>();
foreach (var nr in nrs)
{
tags.Add(nr.Output[0][0][0]); // shape: (beam_size, batch_size, seq_size)
}
return(string.Join("\t", tags));
}
static void Main(string[] args)
{
try
{
//Parse command line
// Seq2SeqOptions opts = new Seq2SeqOptions();
ArgParser argParser = new ArgParser(args, opts);
if (!opts.ConfigFilePath.IsNullOrEmpty())
{
Logger.WriteLine($"Loading config file from '{opts.ConfigFilePath}'");
opts = JsonConvert.DeserializeObject <SeqClassificationOptions>(File.ReadAllText(opts.ConfigFilePath));
}
Logger.LogFile = $"{nameof(SeqClassificationConsole)}_{opts.Task}_{Utils.GetTimeStamp(DateTime.Now)}.log";
ShowOptions(args, opts);
DecodingOptions decodingOptions = opts.CreateDecodingOptions();
SeqClassification ss = null;
if (opts.Task == ModeEnums.Train)
{
// Load train corpus
var trainCorpus = new SeqClassificationMultiTasksCorpus(corpusFilePath: opts.TrainCorpusPath, srcLangName: opts.SrcLang, tgtLangName: opts.TgtLang, batchSize: opts.BatchSize, shuffleBlockSize: opts.ShuffleBlockSize,
maxSentLength: opts.MaxTrainSentLength, shuffleEnums: opts.ShuffleType, tooLongSequence: opts.TooLongSequence);
// Load valid corpus
var validCorpusList = new List <SeqClassificationMultiTasksCorpus>();
if (!opts.ValidCorpusPaths.IsNullOrEmpty())
{
string[] validCorpusPathList = opts.ValidCorpusPaths.Split(';');
foreach (var validCorpusPath in validCorpusPathList)
{
validCorpusList.Add(new SeqClassificationMultiTasksCorpus(validCorpusPath, srcLangName: opts.SrcLang, tgtLangName: opts.TgtLang, opts.ValBatchSize, opts.ShuffleBlockSize, opts.MaxTestSentLength, shuffleEnums: opts.ShuffleType, tooLongSequence: opts.TooLongSequence));
}
}
// Create learning rate
ILearningRate learningRate = new DecayLearningRate(opts.StartLearningRate, opts.WarmUpSteps, opts.WeightsUpdateCount);
// Create metrics
Dictionary <int, List <IMetric> > taskId2metrics = new Dictionary <int, List <IMetric> >();
// Create optimizer
IOptimizer optimizer = Misc.CreateOptimizer(opts);
if (!opts.ModelFilePath.IsNullOrEmpty() && File.Exists(opts.ModelFilePath))
{
//Incremental training
Logger.WriteLine($"Loading model from '{opts.ModelFilePath}'...");
ss = new SeqClassification(opts);
for (int i = 0; i < ss.ClsVocabs.Count; i++)
{
taskId2metrics.Add(i, new List <IMetric>());
taskId2metrics[i].Add(new MultiLabelsFscoreMetric("", ss.ClsVocabs[i].GetAllTokens(keepBuildInTokens: false)));
}
}
else
{
// Load or build vocabulary
Vocab srcVocab = null;
List <Vocab> tgtVocabs = null;
if (!opts.SrcVocab.IsNullOrEmpty() && !opts.TgtVocab.IsNullOrEmpty())
{
Logger.WriteLine($"Loading source vocabulary from '{opts.SrcVocab}' and target vocabulary from '{opts.TgtVocab}'.");
// Vocabulary files are specified, so we load them
srcVocab = new Vocab(opts.SrcVocab);
tgtVocabs = new List <Vocab>
{
new Vocab(opts.TgtVocab)
};
}
else
{
Logger.WriteLine($"Building vocabulary from training corpus.");
// We don't specify vocabulary, so we build it from train corpus
(srcVocab, tgtVocabs) = trainCorpus.BuildVocabs(opts.SrcVocabSize, opts.TgtVocabSize);
}
for (int i = 0; i < tgtVocabs.Count; i++)
{
taskId2metrics.Add(i, new List <IMetric>());
taskId2metrics[i].Add(new MultiLabelsFscoreMetric("", tgtVocabs[i].GetAllTokens(keepBuildInTokens: false)));
}
//New training
ss = new SeqClassification(opts, srcVocab, tgtVocabs);
}
// Add event handler for monitoring
ss.StatusUpdateWatcher += Misc.Ss_StatusUpdateWatcher;
ss.EvaluationWatcher += Ss_EvaluationWatcher;
// Kick off training
ss.Train(maxTrainingEpoch: opts.MaxEpochNum, trainCorpus: trainCorpus, validCorpusList: validCorpusList.ToArray(), learningRate: learningRate, optimizer: optimizer, taskId2metrics: taskId2metrics, decodingOptions: decodingOptions);
}
//else if (opts.Task == ModeEnums.Valid)
//{
// Logger.WriteLine($"Evaluate model '{opts.ModelFilePath}' by valid corpus '{opts.ValidCorpusPath}'");
// // Create metrics
// List<IMetric> metrics = new List<IMetric>
//{
// new BleuMetric(),
// new LengthRatioMetric()
//};
// // Load valid corpus
// ParallelCorpus validCorpus = new ParallelCorpus(opts.ValidCorpusPath, opts.SrcLang, opts.TgtLang, opts.ValBatchSize, opts.ShuffleBlockSize, opts.MaxSrcTestSentLength, opts.MaxTgtTestSentLength, shuffleEnums: shuffleType);
// ss = new Seq2Seq(opts);
// ss.EvaluationWatcher += ss_EvaluationWatcher;
// ss.Valid(validCorpus: validCorpus, metrics: metrics);
//}
else if (opts.Task == ModeEnums.Test)
{
if (File.Exists(opts.OutputFile))
{
Logger.WriteLine(Logger.Level.err, ConsoleColor.Yellow, $"Output file '{opts.OutputFile}' exist. Delete it.");
File.Delete(opts.OutputFile);
}
//Test trained model
ss = new SeqClassification(opts);
Stopwatch stopwatch = Stopwatch.StartNew();
ss.Test <SeqClassificationMultiTasksCorpusBatch>(opts.InputTestFile, opts.OutputFile, opts.BatchSize, decodingOptions, opts.SrcSentencePieceModelPath, opts.TgtSentencePieceModelPath);
stopwatch.Stop();
Logger.WriteLine($"Test mode execution time elapsed: '{stopwatch.Elapsed}'");
}
//else if (opts.Task == ModeEnums.DumpVocab)
//{
// ss = new Seq2Seq(opts);
// ss.DumpVocabToFiles(opts.SrcVocab, opts.TgtVocab);
//}
else
{
Logger.WriteLine(Logger.Level.err, ConsoleColor.Red, $"Task '{opts.Task}' is not supported.");
argParser.Usage();
}
}
catch (Exception err)
{
Logger.WriteLine($"Exception: '{err.Message}'");
Logger.WriteLine($"Call stack: '{err.StackTrace}'");
}
}