From 8aa29810bb77611cc20b7a384897ff6703783ea1 Mon Sep 17 00:00:00 2001 From: Chris Dyer Date: Sun, 18 Nov 2012 13:35:42 -0500 Subject: major restructure of the training code --- training/pro/Makefile.am | 11 + training/pro/mr_pro_generate_mapper_input.pl | 18 + training/pro/mr_pro_map.cc | 201 ++++++++++ training/pro/mr_pro_reduce.cc | 286 ++++++++++++++ training/pro/pro.pl | 555 +++++++++++++++++++++++++++ 5 files changed, 1071 insertions(+) create mode 100644 training/pro/Makefile.am create mode 100755 training/pro/mr_pro_generate_mapper_input.pl create mode 100644 training/pro/mr_pro_map.cc create mode 100644 training/pro/mr_pro_reduce.cc create mode 100755 training/pro/pro.pl (limited to 'training/pro') diff --git a/training/pro/Makefile.am b/training/pro/Makefile.am new file mode 100644 index 00000000..1916b6b2 --- /dev/null +++ b/training/pro/Makefile.am @@ -0,0 +1,11 @@ +bin_PROGRAMS = \ + mr_pro_map \ + mr_pro_reduce + +mr_pro_map_SOURCES = mr_pro_map.cc +mr_pro_map_LDADD = $(top_srcdir)/training/utils/libtraining_utils.a $(top_srcdir)/decoder/libcdec.a $(top_srcdir)/mteval/libmteval.a $(top_srcdir)/utils/libutils.a -lz + +mr_pro_reduce_SOURCES = mr_pro_reduce.cc +mr_pro_reduce_LDADD = $(top_srcdir)/training/liblbfgs/liblbfgs.a $(top_srcdir)/utils/libutils.a -lz + +AM_CPPFLAGS = -W -Wall -Wno-sign-compare -I$(top_srcdir)/utils -I$(top_srcdir)/decoder -I$(top_srcdir)/mteval -I$(top_srcdir)/training/utils -I$(top_srcdir)/training diff --git a/training/pro/mr_pro_generate_mapper_input.pl b/training/pro/mr_pro_generate_mapper_input.pl new file mode 100755 index 00000000..b30fc4fd --- /dev/null +++ b/training/pro/mr_pro_generate_mapper_input.pl @@ -0,0 +1,18 @@ +#!/usr/bin/perl -w +use strict; + +die "Usage: $0 HG_DIR\n" unless scalar @ARGV == 1; +my $d = shift @ARGV; +die "Can't find directory $d" unless -d $d; + +opendir(DIR, $d) or die "Can't read $d: $!"; +my @hgs = grep { /\.gz$/ } readdir(DIR); +closedir DIR; + +for my $hg (@hgs) { + my $file = $hg; + my $id = $hg; + $id =~ s/(\.json)?\.gz//; + print "$d/$file $id\n"; +} + diff --git a/training/pro/mr_pro_map.cc b/training/pro/mr_pro_map.cc new file mode 100644 index 00000000..eef40b8a --- /dev/null +++ b/training/pro/mr_pro_map.cc @@ -0,0 +1,201 @@ +#include +#include +#include +#include +#include + +#include +#include +#include +#include + +#include "candidate_set.h" +#include "sampler.h" +#include "filelib.h" +#include "stringlib.h" +#include "weights.h" +#include "inside_outside.h" +#include "hg_io.h" +#include "ns.h" +#include "ns_docscorer.h" + +// This is Figure 4 (Algorithm Sampler) from Hopkins&May (2011) + +using namespace std; +namespace po = boost::program_options; + +boost::shared_ptr rng; + +void InitCommandLine(int argc, char** argv, po::variables_map* conf) { + po::options_description opts("Configuration options"); + opts.add_options() + ("reference,r",po::value >(), "[REQD] Reference translation (tokenized text)") + ("weights,w",po::value(), "[REQD] Weights files from current iterations") + ("kbest_repository,K",po::value()->default_value("./kbest"),"K-best list repository (directory)") + ("input,i",po::value()->default_value("-"), "Input file to map (- is STDIN)") + ("source,s",po::value()->default_value(""), "Source file (ignored, except for AER)") + ("evaluation_metric,m",po::value()->default_value("IBM_BLEU"), "Evaluation metric (ibm_bleu, koehn_bleu, nist_bleu, ter, meteor, etc.)") + ("kbest_size,k",po::value()->default_value(1500u), "Top k-hypotheses to extract") + ("candidate_pairs,G", po::value()->default_value(5000u), "Number of pairs to sample per hypothesis (Gamma)") + ("best_pairs,X", po::value()->default_value(50u), "Number of pairs, ranked by magnitude of objective delta, to retain (Xi)") + ("random_seed,S", po::value(), "Random seed (if not specified, /dev/random will be used)") + ("help,h", "Help"); + po::options_description dcmdline_options; + dcmdline_options.add(opts); + po::store(parse_command_line(argc, argv, dcmdline_options), *conf); + bool flag = false; + if (!conf->count("reference")) { + cerr << "Please specify one or more references using -r \n"; + flag = true; + } + if (!conf->count("weights")) { + cerr << "Please specify weights using -w \n"; + flag = true; + } + if (flag || conf->count("help")) { + cerr << dcmdline_options << endl; + exit(1); + } +} + +struct ThresholdAlpha { + explicit ThresholdAlpha(double t = 0.05) : threshold(t) {} + double operator()(double mag) const { + if (mag < threshold) return 0.0; else return 1.0; + } + const double threshold; +}; + +struct TrainingInstance { + TrainingInstance(const SparseVector& feats, bool positive, float diff) : x(feats), y(positive), gdiff(diff) {} + SparseVector x; +#undef DEBUGGING_PRO +#ifdef DEBUGGING_PRO + vector a; + vector b; +#endif + bool y; + float gdiff; +}; +#ifdef DEBUGGING_PRO +ostream& operator<<(ostream& os, const TrainingInstance& d) { + return os << d.gdiff << " y=" << d.y << "\tA:" << TD::GetString(d.a) << "\n\tB: " << TD::GetString(d.b) << "\n\tX: " << d.x; +} +#endif + +struct DiffOrder { + bool operator()(const TrainingInstance& a, const TrainingInstance& b) const { + return a.gdiff > b.gdiff; + } +}; + +void Sample(const unsigned gamma, + const unsigned xi, + const training::CandidateSet& J_i, + const EvaluationMetric* metric, + vector* pv) { + const bool invert_score = metric->IsErrorMetric(); + vector v1, v2; + float avg_diff = 0; + for (unsigned i = 0; i < gamma; ++i) { + const size_t a = rng->inclusive(0, J_i.size() - 1)(); + const size_t b = rng->inclusive(0, J_i.size() - 1)(); + if (a == b) continue; + float ga = metric->ComputeScore(J_i[a].eval_feats); + float gb = metric->ComputeScore(J_i[b].eval_feats); + bool positive = gb < ga; + if (invert_score) positive = !positive; + const float gdiff = fabs(ga - gb); + if (!gdiff) continue; + avg_diff += gdiff; + SparseVector xdiff = (J_i[a].fmap - J_i[b].fmap).erase_zeros(); + if (xdiff.empty()) { + cerr << "Empty diff:\n " << TD::GetString(J_i[a].ewords) << endl << "x=" << J_i[a].fmap << endl; + cerr << " " << TD::GetString(J_i[b].ewords) << endl << "x=" << J_i[b].fmap << endl; + continue; + } + v1.push_back(TrainingInstance(xdiff, positive, gdiff)); +#ifdef DEBUGGING_PRO + v1.back().a = J_i[a].hyp; + v1.back().b = J_i[b].hyp; + cerr << "N: " << v1.back() << endl; +#endif + } + avg_diff /= v1.size(); + + for (unsigned i = 0; i < v1.size(); ++i) { + double p = 1.0 / (1.0 + exp(-avg_diff - v1[i].gdiff)); + // cerr << "avg_diff=" << avg_diff << " gdiff=" << v1[i].gdiff << " p=" << p << endl; + if (rng->next() < p) v2.push_back(v1[i]); + } + vector::iterator mid = v2.begin() + xi; + if (xi > v2.size()) mid = v2.end(); + partial_sort(v2.begin(), mid, v2.end(), DiffOrder()); + copy(v2.begin(), mid, back_inserter(*pv)); +#ifdef DEBUGGING_PRO + if (v2.size() >= 5) { + for (int i =0; i < (mid - v2.begin()); ++i) { + cerr << v2[i] << endl; + } + cerr << pv->back() << endl; + } +#endif +} + +int main(int argc, char** argv) { + po::variables_map conf; + InitCommandLine(argc, argv, &conf); + if (conf.count("random_seed")) + rng.reset(new MT19937(conf["random_seed"].as())); + else + rng.reset(new MT19937); + const string evaluation_metric = conf["evaluation_metric"].as(); + + EvaluationMetric* metric = EvaluationMetric::Instance(evaluation_metric); + DocumentScorer ds(metric, conf["reference"].as >()); + cerr << "Loaded " << ds.size() << " references for scoring with " << evaluation_metric << endl; + + Hypergraph hg; + string last_file; + ReadFile in_read(conf["input"].as()); + istream &in=*in_read.stream(); + const unsigned kbest_size = conf["kbest_size"].as(); + const unsigned gamma = conf["candidate_pairs"].as(); + const unsigned xi = conf["best_pairs"].as(); + string weightsf = conf["weights"].as(); + vector weights; + Weights::InitFromFile(weightsf, &weights); + string kbest_repo = conf["kbest_repository"].as(); + MkDirP(kbest_repo); + while(in) { + vector v; + string line; + getline(in, line); + if (line.empty()) continue; + istringstream is(line); + int sent_id; + string file; + // path-to-file (JSON) sent_id + is >> file >> sent_id; + ReadFile rf(file); + ostringstream os; + training::CandidateSet J_i; + os << kbest_repo << "/kbest." << sent_id << ".txt.gz"; + const string kbest_file = os.str(); + if (FileExists(kbest_file)) + J_i.ReadFromFile(kbest_file); + HypergraphIO::ReadFromJSON(rf.stream(), &hg); + hg.Reweight(weights); + J_i.AddKBestCandidates(hg, kbest_size, ds[sent_id]); + J_i.WriteToFile(kbest_file); + + Sample(gamma, xi, J_i, metric, &v); + for (unsigned i = 0; i < v.size(); ++i) { + const TrainingInstance& vi = v[i]; + cout << vi.y << "\t" << vi.x << endl; + cout << (!vi.y) << "\t" << (vi.x * -1.0) << endl; + } + } + return 0; +} + diff --git a/training/pro/mr_pro_reduce.cc b/training/pro/mr_pro_reduce.cc new file mode 100644 index 00000000..5ef9b470 --- /dev/null +++ b/training/pro/mr_pro_reduce.cc @@ -0,0 +1,286 @@ +#include +#include +#include +#include +#include + +#include +#include + +#include "filelib.h" +#include "weights.h" +#include "sparse_vector.h" +#include "optimize.h" +#include "liblbfgs/lbfgs++.h" + +using namespace std; +namespace po = boost::program_options; + +// since this is a ranking model, there should be equal numbers of +// positive and negative examples, so the bias should be 0 +static const double MAX_BIAS = 1e-10; + +void InitCommandLine(int argc, char** argv, po::variables_map* conf) { + po::options_description opts("Configuration options"); + opts.add_options() + ("weights,w", po::value(), "Weights from previous iteration (used as initialization and interpolation") + ("regularization_strength,C",po::value()->default_value(500.0), "l2 regularization strength") + ("l1",po::value()->default_value(0.0), "l1 regularization strength") + ("regularize_to_weights,y",po::value()->default_value(5000.0), "Differences in learned weights to previous weights are penalized with an l2 penalty with this strength; 0.0 = no effect") + ("memory_buffers,m",po::value()->default_value(100), "Number of memory buffers (LBFGS)") + ("min_reg,r",po::value()->default_value(0.01), "When tuning (-T) regularization strength, minimum regularization strenght") + ("max_reg,R",po::value()->default_value(1e6), "When tuning (-T) regularization strength, maximum regularization strenght") + ("testset,t",po::value(), "Optional held-out test set") + ("tune_regularizer,T", "Use the held out test set (-t) to tune the regularization strength") + ("interpolate_with_weights,p",po::value()->default_value(1.0), "[deprecated] Output weights are p*w + (1-p)*w_prev; 1.0 = no effect") + ("help,h", "Help"); + po::options_description dcmdline_options; + dcmdline_options.add(opts); + po::store(parse_command_line(argc, argv, dcmdline_options), *conf); + if (conf->count("help")) { + cerr << dcmdline_options << endl; + exit(1); + } +} + +void ParseSparseVector(string& line, size_t cur, SparseVector* out) { + SparseVector& x = *out; + size_t last_start = cur; + size_t last_comma = string::npos; + while(cur <= line.size()) { + if (line[cur] == ' ' || cur == line.size()) { + if (!(cur > last_start && last_comma != string::npos && cur > last_comma)) { + cerr << "[ERROR] " << line << endl << " position = " << cur << endl; + exit(1); + } + const int fid = FD::Convert(line.substr(last_start, last_comma - last_start)); + if (cur < line.size()) line[cur] = 0; + const weight_t val = strtod(&line[last_comma + 1], NULL); + x.set_value(fid, val); + + last_comma = string::npos; + last_start = cur+1; + } else { + if (line[cur] == '=') + last_comma = cur; + } + ++cur; + } +} + +void ReadCorpus(istream* pin, vector > >* corpus) { + istream& in = *pin; + corpus->clear(); + bool flag = false; + int lc = 0; + string line; + SparseVector x; + while(getline(in, line)) { + ++lc; + if (lc % 1000 == 0) { cerr << '.'; flag = true; } + if (lc % 40000 == 0) { cerr << " [" << lc << "]\n"; flag = false; } + if (line.empty()) continue; + const size_t ks = line.find("\t"); + assert(string::npos != ks); + assert(ks == 1); + const bool y = line[0] == '1'; + x.clear(); + ParseSparseVector(line, ks + 1, &x); + corpus->push_back(make_pair(y, x)); + } + if (flag) cerr << endl; +} + +void GradAdd(const SparseVector& v, const double scale, weight_t* acc) { + for (SparseVector::const_iterator it = v.begin(); + it != v.end(); ++it) { + acc[it->first] += it->second * scale; + } +} + +double ApplyRegularizationTerms(const double C, + const double T, + const vector& weights, + const vector& prev_weights, + weight_t* g) { + double reg = 0; + for (size_t i = 0; i < weights.size(); ++i) { + const double prev_w_i = (i < prev_weights.size() ? prev_weights[i] : 0.0); + const double& w_i = weights[i]; + reg += C * w_i * w_i; + g[i] += 2 * C * w_i; + + const double diff_i = w_i - prev_w_i; + reg += T * diff_i * diff_i; + g[i] += 2 * T * diff_i; + } + return reg; +} + +double TrainingInference(const vector& x, + const vector > >& corpus, + weight_t* g = NULL) { + double cll = 0; + for (int i = 0; i < corpus.size(); ++i) { + const double dotprod = corpus[i].second.dot(x) + (x.size() ? x[0] : weight_t()); // x[0] is bias + double lp_false = dotprod; + double lp_true = -dotprod; + if (0 < lp_true) { + lp_true += log1p(exp(-lp_true)); + lp_false = log1p(exp(lp_false)); + } else { + lp_true = log1p(exp(lp_true)); + lp_false += log1p(exp(-lp_false)); + } + lp_true*=-1; + lp_false*=-1; + if (corpus[i].first) { // true label + cll -= lp_true; + if (g) { + // g -= corpus[i].second * exp(lp_false); + GradAdd(corpus[i].second, -exp(lp_false), g); + g[0] -= exp(lp_false); // bias + } + } else { // false label + cll -= lp_false; + if (g) { + // g += corpus[i].second * exp(lp_true); + GradAdd(corpus[i].second, exp(lp_true), g); + g[0] += exp(lp_true); // bias + } + } + } + return cll; +} + +struct ProLoss { + ProLoss(const vector > >& tr, + const vector > >& te, + const double c, + const double t, + const vector& px) : training(tr), testing(te), C(c), T(t), prev_x(px){} + double operator()(const vector& x, double* g) const { + fill(g, g + x.size(), 0.0); + double cll = TrainingInference(x, training, g); + tppl = 0; + if (testing.size()) + tppl = pow(2.0, TrainingInference(x, testing, g) / (log(2) * testing.size())); + double ppl = cll / log(2); + ppl /= training.size(); + ppl = pow(2.0, ppl); + double reg = ApplyRegularizationTerms(C, T, x, prev_x, g); + return cll + reg; + } + const vector > >& training, testing; + const double C, T; + const vector& prev_x; + mutable double tppl; +}; + +// return held-out log likelihood +double LearnParameters(const vector > >& training, + const vector > >& testing, + const double C, + const double C1, + const double T, + const unsigned memory_buffers, + const vector& prev_x, + vector* px) { + assert(px->size() == prev_x.size()); + ProLoss loss(training, testing, C, T, prev_x); + LBFGS lbfgs(px, loss, memory_buffers, C1); + lbfgs.MinimizeFunction(); + return loss.tppl; +} + +int main(int argc, char** argv) { + po::variables_map conf; + InitCommandLine(argc, argv, &conf); + string line; + vector > > training, testing; + const bool tune_regularizer = conf.count("tune_regularizer"); + if (tune_regularizer && !conf.count("testset")) { + cerr << "--tune_regularizer requires --testset to be set\n"; + return 1; + } + const double min_reg = conf["min_reg"].as(); + const double max_reg = conf["max_reg"].as(); + double C = conf["regularization_strength"].as(); // will be overridden if parameter is tuned + double C1 = conf["l1"].as(); // will be overridden if parameter is tuned + const double T = conf["regularize_to_weights"].as(); + assert(C >= 0.0); + assert(min_reg >= 0.0); + assert(max_reg >= 0.0); + assert(max_reg > min_reg); + const double psi = conf["interpolate_with_weights"].as(); + if (psi < 0.0 || psi > 1.0) { cerr << "Invalid interpolation weight: " << psi << endl; return 1; } + ReadCorpus(&cin, &training); + if (conf.count("testset")) { + ReadFile rf(conf["testset"].as()); + ReadCorpus(rf.stream(), &testing); + } + cerr << "Number of features: " << FD::NumFeats() << endl; + + vector x, prev_x; // x[0] is bias + if (conf.count("weights")) { + Weights::InitFromFile(conf["weights"].as(), &x); + x.resize(FD::NumFeats()); + prev_x = x; + } else { + x.resize(FD::NumFeats()); + prev_x = x; + } + cerr << " Number of features: " << x.size() << endl; + cerr << "Number of training examples: " << training.size() << endl; + cerr << "Number of testing examples: " << testing.size() << endl; + double tppl = 0.0; + vector > sp; + vector smoothed; + if (tune_regularizer) { + C = min_reg; + const double steps = 18; + double sweep_factor = exp((log(max_reg) - log(min_reg)) / steps); + cerr << "SWEEP FACTOR: " << sweep_factor << endl; + while(C < max_reg) { + cerr << "C=" << C << "\tT=" <(), prev_x, &x); + sp.push_back(make_pair(C, tppl)); + C *= sweep_factor; + } + smoothed.resize(sp.size(), 0); + smoothed[0] = sp[0].second; + smoothed.back() = sp.back().second; + for (int i = 1; i < sp.size()-1; ++i) { + double prev = sp[i-1].second; + double next = sp[i+1].second; + double cur = sp[i].second; + smoothed[i] = (prev*0.2) + cur * 0.6 + (0.2*next); + } + double best_ppl = 9999999; + unsigned best_i = 0; + for (unsigned i = 0; i < sp.size(); ++i) { + if (smoothed[i] < best_ppl) { + best_ppl = smoothed[i]; + best_i = i; + } + } + C = sp[best_i].first; + } // tune regularizer + tppl = LearnParameters(training, testing, C, C1, T, conf["memory_buffers"].as(), prev_x, &x); + if (conf.count("weights")) { + for (int i = 1; i < x.size(); ++i) { + x[i] = (x[i] * psi) + prev_x[i] * (1.0 - psi); + } + } + cout.precision(15); + cout << "# C=" << C << "\theld out perplexity="; + if (tppl) { cout << tppl << endl; } else { cout << "N/A\n"; } + if (sp.size()) { + cout << "# Parameter sweep:\n"; + for (int i = 0; i < sp.size(); ++i) { + cout << "# " << sp[i].first << "\t" << sp[i].second << "\t" << smoothed[i] << endl; + } + } + Weights::WriteToFile("-", x); + return 0; +} diff --git a/training/pro/pro.pl b/training/pro/pro.pl new file mode 100755 index 00000000..3b30c379 --- /dev/null +++ b/training/pro/pro.pl @@ -0,0 +1,555 @@ +#!/usr/bin/env perl +use strict; +use File::Basename qw(basename); +my @ORIG_ARGV=@ARGV; +use Cwd qw(getcwd); +my $SCRIPT_DIR; BEGIN { use Cwd qw/ abs_path /; use File::Basename; $SCRIPT_DIR = dirname(abs_path($0)); push @INC, $SCRIPT_DIR, "$SCRIPT_DIR/../../environment", "$SCRIPT_DIR/../utils"; } + +# Skip local config (used for distributing jobs) if we're running in local-only mode +use LocalConfig; +use Getopt::Long; +use IPC::Open2; +use POSIX ":sys_wait_h"; +my $QSUB_CMD = qsub_args(mert_memory()); +my $default_jobs = env_default_jobs(); + +my $UTILS_DIR="$SCRIPT_DIR/../utils"; +require "$UTILS_DIR/libcall.pl"; + +# Default settings +my $srcFile; +my $refFiles; +my $bin_dir = $SCRIPT_DIR; +die "Bin directory $bin_dir missing/inaccessible" unless -d $bin_dir; +my $FAST_SCORE="$bin_dir/../../mteval/fast_score"; +die "Can't execute $FAST_SCORE" unless -x $FAST_SCORE; +my $MAPINPUT = "$bin_dir/mr_pro_generate_mapper_input.pl"; +my $MAPPER = "$bin_dir/mr_pro_map"; +my $REDUCER = "$bin_dir/mr_pro_reduce"; +my $parallelize = "$UTILS_DIR/parallelize.pl"; +my $libcall = "$UTILS_DIR/libcall.pl"; +my $sentserver = "$UTILS_DIR/sentserver"; +my $sentclient = "$UTILS_DIR/sentclient"; +my $LocalConfig = "$SCRIPT_DIR/../../environment/LocalConfig.pm"; + +my $SCORER = $FAST_SCORE; +die "Can't find $MAPPER" unless -x $MAPPER; +my $cdec = "$bin_dir/../../decoder/cdec"; +die "Can't find decoder in $cdec" unless -x $cdec; +die "Can't find $parallelize" unless -x $parallelize; +die "Can't find $libcall" unless -e $libcall; +my $decoder = $cdec; +my $lines_per_mapper = 30; +my $iteration = 1; +my $best_weights; +my $psi = 1; +my $default_max_iter = 30; +my $max_iterations = $default_max_iter; +my $jobs = $default_jobs; # number of decode nodes +my $pmem = "4g"; +my $disable_clean = 0; +my %seen_weights; +my $help = 0; +my $epsilon = 0.0001; +my $dryrun = 0; +my $last_score = -10000000; +my $metric = "ibm_bleu"; +my $dir; +my $iniFile; +my $weights; +my $use_make = 1; # use make to parallelize +my $useqsub = 0; +my $initial_weights; +my $pass_suffix = ''; +my $devset; + +# regularization strength +my $reg = 500; +my $reg_previous = 5000; + +# Process command-line options +if (GetOptions( + "config=s" => \$iniFile, + "weights=s" => \$initial_weights, + "devset=s" => \$devset, + "jobs=i" => \$jobs, + "metric=s" => \$metric, + "pass-suffix=s" => \$pass_suffix, + "qsub" => \$useqsub, + "help" => \$help, + "reg=f" => \$reg, + "reg-previous=f" => \$reg_previous, + "output-dir=s" => \$dir, +) == 0 || @ARGV!=0 || $help) { + print_help(); + exit; +} + +if ($useqsub) { + $use_make = 0; + die "LocalEnvironment.pm does not have qsub configuration for this host. Cannot run with --qsub!\n" unless has_qsub(); +} + +my @missing_args = (); +if (!defined $iniFile) { push @missing_args, "--config"; } +if (!defined $devset) { push @missing_args, "--devset"; } +if (!defined $initial_weights) { push @missing_args, "--weights"; } +die "Please specify missing arguments: " . join (', ', @missing_args) . "\n" if (@missing_args); + +if ($metric =~ /^(combi|ter)$/i) { + $lines_per_mapper = 5; +} + +my $host =check_output("hostname"); chomp $host; +my $bleu; +my $interval_count = 0; +my $logfile; +my $projected_score; + +# used in sorting scores +my $DIR_FLAG = '-r'; +if ($metric =~ /^ter$|^aer$/i) { + $DIR_FLAG = ''; +} + +unless ($dir){ + $dir = 'pro'; +} +unless ($dir =~ /^\//){ # convert relative path to absolute path + my $basedir = check_output("pwd"); + chomp $basedir; + $dir = "$basedir/$dir"; +} + +# Initializations and helper functions +srand; + +my @childpids = (); +my @cleanupcmds = (); + +sub cleanup { + print STDERR "Cleanup...\n"; + for my $pid (@childpids){ unchecked_call("kill $pid"); } + for my $cmd (@cleanupcmds){ unchecked_call("$cmd"); } + exit 1; +}; +# Always call cleanup, no matter how we exit +*CORE::GLOBAL::exit = + sub{ cleanup(); }; +$SIG{INT} = "cleanup"; +$SIG{TERM} = "cleanup"; +$SIG{HUP} = "cleanup"; + +my $decoderBase = check_output("basename $decoder"); chomp $decoderBase; +my $newIniFile = "$dir/$decoderBase.ini"; +my $inputFileName = "$dir/input"; +my $user = $ENV{"USER"}; + + +# process ini file +-e $iniFile || die "Error: could not open $iniFile for reading\n"; +open(INI, $iniFile); + +if (-e $dir) { + die "ERROR: working dir $dir already exists\n\n"; +} else { + mkdir "$dir" or die "Can't mkdir $dir: $!"; + mkdir "$dir/hgs" or die; + mkdir "$dir/scripts" or die; + print STDERR <) { $devSize++; } +close F; + +unless($best_weights){ $best_weights = $weights; } +unless($projected_score){ $projected_score = 0.0; } +$seen_weights{$weights} = 1; + +my $random_seed = int(time / 1000); +my $lastWeightsFile; +my $lastPScore = 0; +# main optimization loop +my @allweights; +while (1){ + print STDERR "\n\nITERATION $iteration\n==========\n"; + + if ($iteration > $max_iterations){ + print STDERR "\nREACHED STOPPING CRITERION: Maximum iterations\n"; + last; + } + # iteration-specific files + my $runFile="$dir/run.raw.$iteration"; + my $onebestFile="$dir/1best.$iteration"; + my $logdir="$dir/logs.$iteration"; + my $decoderLog="$logdir/decoder.sentserver.log.$iteration"; + my $scorerLog="$logdir/scorer.log.$iteration"; + check_call("mkdir -p $logdir"); + + + #decode + print STDERR "RUNNING DECODER AT "; + print STDERR unchecked_output("date"); + my $im1 = $iteration - 1; + my $weightsFile="$dir/weights.$im1"; + push @allweights, "-w $dir/weights.$im1"; + `rm -f $dir/hgs/*.gz`; + my $decoder_cmd = "$decoder -c $iniFile --weights$pass_suffix $weightsFile -O $dir/hgs"; + my $pcmd; + if ($use_make) { + $pcmd = "cat $srcFile | $parallelize --use-fork -p $pmem -e $logdir -j $jobs --"; + } else { + $pcmd = "cat $srcFile | $parallelize -p $pmem -e $logdir -j $jobs --"; + } + my $cmd = "$pcmd $decoder_cmd 2> $decoderLog 1> $runFile"; + print STDERR "COMMAND:\n$cmd\n"; + check_bash_call($cmd); + my $num_hgs; + my $num_topbest; + my $retries = 0; + while($retries < 5) { + $num_hgs = check_output("ls $dir/hgs/*.gz | wc -l"); + $num_topbest = check_output("wc -l < $runFile"); + print STDERR "NUMBER OF HGs: $num_hgs\n"; + print STDERR "NUMBER OF TOP-BEST HYPs: $num_topbest\n"; + if($devSize == $num_hgs && $devSize == $num_topbest) { + last; + } else { + print STDERR "Incorrect number of hypergraphs or topbest. Waiting for distributed filesystem and retrying...\n"; + sleep(3); + } + $retries++; + } + die "Dev set contains $devSize sentences, but we don't have topbest and hypergraphs for all these! Decoder failure? Check $decoderLog\n" if ($devSize != $num_hgs || $devSize != $num_topbest); + my $dec_score = check_output("cat $runFile | $SCORER -r $refs -m $metric"); + chomp $dec_score; + print STDERR "DECODER SCORE: $dec_score\n"; + + # save space + check_call("gzip -f $runFile"); + check_call("gzip -f $decoderLog"); + + # run optimizer + print STDERR "RUNNING OPTIMIZER AT "; + print STDERR unchecked_output("date"); + print STDERR " - GENERATE TRAINING EXEMPLARS\n"; + my $mergeLog="$logdir/prune-merge.log.$iteration"; + + my $score = 0; + my $icc = 0; + my $inweights="$dir/weights.$im1"; + $cmd="$MAPINPUT $dir/hgs > $dir/agenda.$im1"; + print STDERR "COMMAND:\n$cmd\n"; + check_call($cmd); + check_call("mkdir -p $dir/splag.$im1"); + $cmd="split -a 3 -l $lines_per_mapper $dir/agenda.$im1 $dir/splag.$im1/mapinput."; + print STDERR "COMMAND:\n$cmd\n"; + check_call($cmd); + opendir(DIR, "$dir/splag.$im1") or die "Can't open directory: $!"; + my @shards = grep { /^mapinput\./ } readdir(DIR); + closedir DIR; + die "No shards!" unless scalar @shards > 0; + my $joblist = ""; + my $nmappers = 0; + @cleanupcmds = (); + my %o2i = (); + my $first_shard = 1; + my $mkfile; # only used with makefiles + my $mkfilename; + if ($use_make) { + $mkfilename = "$dir/splag.$im1/domap.mk"; + open $mkfile, ">$mkfilename" or die "Couldn't write $mkfilename: $!"; + print $mkfile "all: $dir/splag.$im1/map.done\n\n"; + } + my @mkouts = (); # only used with makefiles + my @mapoutputs = (); + for my $shard (@shards) { + my $mapoutput = $shard; + my $client_name = $shard; + $client_name =~ s/mapinput.//; + $client_name = "pro.$client_name"; + $mapoutput =~ s/mapinput/mapoutput/; + push @mapoutputs, "$dir/splag.$im1/$mapoutput"; + $o2i{"$dir/splag.$im1/$mapoutput"} = "$dir/splag.$im1/$shard"; + my $script = "$MAPPER -s $srcFile -m $metric -r $refs -w $inweights -K $dir/kbest < $dir/splag.$im1/$shard > $dir/splag.$im1/$mapoutput"; + if ($use_make) { + my $script_file = "$dir/scripts/map.$shard"; + open F, ">$script_file" or die "Can't write $script_file: $!"; + print F "#!/bin/bash\n"; + print F "$script\n"; + close F; + my $output = "$dir/splag.$im1/$mapoutput"; + push @mkouts, $output; + chmod(0755, $script_file) or die "Can't chmod $script_file: $!"; + if ($first_shard) { print STDERR "$script\n"; $first_shard=0; } + print $mkfile "$output: $dir/splag.$im1/$shard\n\t$script_file\n\n"; + } else { + my $script_file = "$dir/scripts/map.$shard"; + open F, ">$script_file" or die "Can't write $script_file: $!"; + print F "$script\n"; + close F; + if ($first_shard) { print STDERR "$script\n"; $first_shard=0; } + + $nmappers++; + my $qcmd = "$QSUB_CMD -N $client_name -o /dev/null -e $logdir/$client_name.ER $script_file"; + my $jobid = check_output("$qcmd"); + chomp $jobid; + $jobid =~ s/^(\d+)(.*?)$/\1/g; + $jobid =~ s/^Your job (\d+) .*$/\1/; + push(@cleanupcmds, "qdel $jobid 2> /dev/null"); + print STDERR " $jobid"; + if ($joblist == "") { $joblist = $jobid; } + else {$joblist = $joblist . "\|" . $jobid; } + } + } + my @dev_outs = (); + my @devtest_outs = (); + @dev_outs = @mapoutputs; + if ($use_make) { + print $mkfile "$dir/splag.$im1/map.done: @mkouts\n\ttouch $dir/splag.$im1/map.done\n\n"; + close $mkfile; + my $mcmd = "make -j $jobs -f $mkfilename"; + print STDERR "\nExecuting: $mcmd\n"; + check_call($mcmd); + } else { + print STDERR "\nLaunched $nmappers mappers.\n"; + sleep 8; + print STDERR "Waiting for mappers to complete...\n"; + while ($nmappers > 0) { + sleep 5; + my @livejobs = grep(/$joblist/, split(/\n/, unchecked_output("qstat | grep -v ' C '"))); + $nmappers = scalar @livejobs; + } + print STDERR "All mappers complete.\n"; + } + my $tol = 0; + my $til = 0; + my $dev_test_file = "$dir/splag.$im1/devtest.gz"; + print STDERR "\nRUNNING CLASSIFIER (REDUCER)\n"; + print STDERR unchecked_output("date"); + $cmd="cat @dev_outs | $REDUCER -w $dir/weights.$im1 -C $reg -y $reg_previous --interpolate_with_weights $psi"; + $cmd .= " > $dir/weights.$iteration"; + print STDERR "COMMAND:\n$cmd\n"; + check_bash_call($cmd); + $lastWeightsFile = "$dir/weights.$iteration"; + $lastPScore = $score; + $iteration++; + print STDERR "\n==========\n"; +} + + +check_call("cp $lastWeightsFile $dir/weights.final"); +print STDERR "\nFINAL WEIGHTS: $dir/weights.final\n(Use -w with the decoder)\n\n"; +print STDOUT "$dir/weights.final\n"; + +exit 0; + +sub read_weights_file { + my ($file) = @_; + open F, "<$file" or die "Couldn't read $file: $!"; + my @r = (); + my $pm = -1; + while() { + next if /^#/; + next if /^\s*$/; + chomp; + if (/^(.+)\s+(.+)$/) { + my $m = $1; + my $w = $2; + die "Weights out of order: $m <= $pm" unless $m > $pm; + push @r, $w; + } else { + warn "Unexpected feature name in weight file: $_"; + } + } + close F; + return join ' ', @r; +} + +sub enseg { + my $src = shift; + my $newsrc = shift; + open(SRC, $src); + open(NEWSRC, ">$newsrc"); + my $i=0; + while (my $line=){ + chomp $line; + if ($line =~ /^\s* tags, you must include a zero-based id attribute"; + } + } else { + print NEWSRC "$line\n"; + } + $i++; + } + close SRC; + close NEWSRC; + die "Empty dev set!" if ($i == 0); +} + +sub print_help { + + my $executable = basename($0); chomp $executable; + print << "Help"; + +Usage: $executable [options] + + $executable [options] + Runs a complete PRO optimization using the ini file specified. + +Required: + + --config + Decoder configuration file. + + --devset + Dev set source and reference data. + + --weights + Initial weights file (use empty file to start from 0) + +General options: + + --help + Print this message and exit. + + --max-iterations + Maximum number of iterations to run. If not specified, defaults + to $default_max_iter. + + --metric + Metric to optimize. + Example values: IBM_BLEU, NIST_BLEU, Koehn_BLEU, TER, Combi + + --pass-suffix + If the decoder is doing multi-pass decoding, the pass suffix "2", + "3", etc., is used to control what iteration of weights is set. + + --workdir + Directory for intermediate and output files. If not specified, the + name is derived from the ini filename. Assuming that the ini + filename begins with the decoder name and ends with ini, the default + name of the working directory is inferred from the middle part of + the filename. E.g. an ini file named decoder.foo.ini would have + a default working directory name foo. + +Regularization options: + + --reg + l2 regularization strength [default=500]. The greater this value, + the closer to zero the weights will be. + + --reg-previous + l2 penalty for moving away from the weights from the previous + iteration. [default=5000]. The greater this value, the closer + to the previous iteration's weights the next iteration's weights + will be. + +Job control options: + + --jobs + Number of decoder processes to run in parallel. [default=$default_jobs] + + --qsub + Use qsub to run jobs in parallel (qsub must be configured in + environment/LocalEnvironment.pm) + + --pmem + Amount of physical memory requested for parallel decoding jobs + (used with qsub requests only) + +Deprecated options: + + --interpolate-with-weights + [deprecated] At each iteration the resulting weights are + interpolated with the weights from the previous iteration, with + this factor. [default=1.0, i.e., no effect] + +Help +} + +sub convert { + my ($str) = @_; + my @ps = split /;/, $str; + my %dict = (); + for my $p (@ps) { + my ($k, $v) = split /=/, $p; + $dict{$k} = $v; + } + return %dict; +} + + +sub cmdline { + return join ' ',($0,@ORIG_ARGV); +} + +#buggy: last arg gets quoted sometimes? +my $is_shell_special=qr{[ \t\n\\><|&;"'`~*?{}$!()]}; +my $shell_escape_in_quote=qr{[\\"\$`!]}; + +sub escape_shell { + my ($arg)=@_; + return undef unless defined $arg; + if ($arg =~ /$is_shell_special/) { + $arg =~ s/($shell_escape_in_quote)/\\$1/g; + return "\"$arg\""; + } + return $arg; +} + +sub escaped_shell_args { + return map {local $_=$_;chomp;escape_shell($_)} @_; +} + +sub escaped_shell_args_str { + return join ' ',&escaped_shell_args(@_); +} + +sub escaped_cmdline { + return "$0 ".&escaped_shell_args_str(@ORIG_ARGV); +} + +sub split_devset { + my ($infile, $outsrc, $outref) = @_; + open F, "<$infile" or die "Can't read $infile: $!"; + open S, ">$outsrc" or die "Can't write $outsrc: $!"; + open R, ">$outref" or die "Can't write $outref: $!"; + while() { + chomp; + my ($src, @refs) = split /\s*\|\|\|\s*/; + die "Malformed devset line: $_\n" unless scalar @refs > 0; + print S "$src\n"; + print R join(' ||| ', @refs) . "\n"; + } + close R; + close S; + close F; +} + -- cgit v1.2.3 From c7b1dc8eabd50eefb7403ce36d2746f2df39e30e Mon Sep 17 00:00:00 2001 From: Chris Dyer Date: Sat, 15 Dec 2012 02:53:56 -0500 Subject: enable kenlm compression --- configure.ac | 26 ++++++++++++++++++-------- decoder/Makefile.am | 11 +++++------ example_extff/Makefile.am | 2 +- klm/util/have.hh | 3 +-- mteval/Makefile.am | 6 +++--- python/setup.py.in | 2 +- training/dpmert/Makefile.am | 10 +++++----- training/dtrain/Makefile.am | 2 +- training/minrisk/Makefile.am | 2 +- training/mira/Makefile.am | 2 +- training/pro/Makefile.am | 4 ++-- training/rampion/Makefile.am | 2 +- training/utils/Makefile.am | 4 ++-- utils/Makefile.am | 18 +++++++++--------- word-aligner/Makefile.am | 2 +- 15 files changed, 52 insertions(+), 44 deletions(-) (limited to 'training/pro') diff --git a/configure.ac b/configure.ac index f4650ca4..eabb8645 100644 --- a/configure.ac +++ b/configure.ac @@ -18,6 +18,23 @@ BOOST_TEST AM_PATH_PYTHON AC_CHECK_HEADER(dlfcn.h,AC_DEFINE(HAVE_DLFCN_H)) AC_CHECK_LIB(dl, dlopen) +AC_CHECK_HEADERS(zlib.h, + AC_CHECK_LIB(z, gzread,[ + AC_DEFINE(HAVE_ZLIB,[],[Do we have zlib]) + ZLIBS="$ZLIBS -lz" + ])) + +AC_CHECK_HEADERS(bzlib.h, + AC_CHECK_LIB(bz2, BZ2_bzReadOpen,[ + AC_DEFINE(HAVE_BZLIB,[],[Do we have bzlib]) + ZLIBS="$ZLIBS -lbz2" + ])) + +AC_CHECK_HEADERS(lzma.h, + AC_CHECK_LIB(lzma, lzma_code,[ + AC_DEFINE(HAVE_XZLIB,[],[Do we have lzma]) + ZLIBS="$ZLIBS -llzma" + ])) AC_ARG_ENABLE(mpi, [ --enable-mpi Build MPI binaries, assumes mpi.h is present ], @@ -72,19 +89,12 @@ fi CPPFLAGS="$CPPFLAGS $BOOST_CPPFLAGS" LDFLAGS="$LDFLAGS $BOOST_PROGRAM_OPTIONS_LDFLAGS $BOOST_SERIALIZATION_LDFLAGS $BOOST_SYSTEM_LDFLAGS" # $BOOST_THREAD_LDFLAGS" -LIBS="$LIBS $BOOST_PROGRAM_OPTIONS_LIBS $BOOST_SERIALIZATION_LIBS $BOOST_SYSTEM_LIBS" +LIBS="$LIBS $BOOST_PROGRAM_OPTIONS_LIBS $BOOST_SERIALIZATION_LIBS $BOOST_SYSTEM_LIBS $ZLIBS" # $BOOST_THREAD_LIBS" AC_CHECK_HEADER(google/dense_hash_map, [AC_DEFINE([HAVE_SPARSEHASH], [1], [flag for google::dense_hash_map])]) -AC_CHECK_HEADER(zlib.h, - [AC_DEFINE([HAVE_ZLIB], [1], [zlib])]) -AC_CHECK_HEADER(bzlib.h, - [AC_DEFINE([HAVE_BZLIB], [1], [bzlib])]) -AC_CHECK_HEADER(lzma.h, - [AC_DEFINE([HAVE_XZLIB], [1], [xzlib])]) - AC_PROG_INSTALL CPPFLAGS="-DPIC -fPIC $CPPFLAGS -DHAVE_CONFIG_H -DKENLM_MAX_ORDER=6" diff --git a/decoder/Makefile.am b/decoder/Makefile.am index 6914fa0f..88a6116c 100644 --- a/decoder/Makefile.am +++ b/decoder/Makefile.am @@ -8,16 +8,16 @@ noinst_PROGRAMS = \ TESTS = trule_test parser_test grammar_test hg_test parser_test_SOURCES = parser_test.cc -parser_test_LDADD = $(BOOST_UNIT_TEST_FRAMEWORK_LDFLAGS) $(BOOST_UNIT_TEST_FRAMEWORK_LIBS) libcdec.a ../mteval/libmteval.a ../utils/libutils.a -lz +parser_test_LDADD = $(BOOST_UNIT_TEST_FRAMEWORK_LDFLAGS) $(BOOST_UNIT_TEST_FRAMEWORK_LIBS) libcdec.a ../mteval/libmteval.a ../utils/libutils.a grammar_test_SOURCES = grammar_test.cc -grammar_test_LDADD = $(BOOST_UNIT_TEST_FRAMEWORK_LDFLAGS) $(BOOST_UNIT_TEST_FRAMEWORK_LIBS) libcdec.a ../mteval/libmteval.a ../utils/libutils.a -lz +grammar_test_LDADD = $(BOOST_UNIT_TEST_FRAMEWORK_LDFLAGS) $(BOOST_UNIT_TEST_FRAMEWORK_LIBS) libcdec.a ../mteval/libmteval.a ../utils/libutils.a hg_test_SOURCES = hg_test.cc -hg_test_LDADD = $(BOOST_UNIT_TEST_FRAMEWORK_LDFLAGS) $(BOOST_UNIT_TEST_FRAMEWORK_LIBS) libcdec.a ../mteval/libmteval.a ../utils/libutils.a -lz +hg_test_LDADD = $(BOOST_UNIT_TEST_FRAMEWORK_LDFLAGS) $(BOOST_UNIT_TEST_FRAMEWORK_LIBS) libcdec.a ../mteval/libmteval.a ../utils/libutils.a trule_test_SOURCES = trule_test.cc -trule_test_LDADD = $(BOOST_UNIT_TEST_FRAMEWORK_LDFLAGS) $(BOOST_UNIT_TEST_FRAMEWORK_LIBS) libcdec.a ../mteval/libmteval.a ../utils/libutils.a -lz +trule_test_LDADD = $(BOOST_UNIT_TEST_FRAMEWORK_LDFLAGS) $(BOOST_UNIT_TEST_FRAMEWORK_LIBS) libcdec.a ../mteval/libmteval.a ../utils/libutils.a cdec_SOURCES = cdec.cc -cdec_LDADD = libcdec.a ../mteval/libmteval.a ../utils/libutils.a ../klm/search/libksearch.a ../klm/lm/libklm.a ../klm/util/libklm_util.a -lz +cdec_LDADD = libcdec.a ../mteval/libmteval.a ../utils/libutils.a ../klm/search/libksearch.a ../klm/lm/libklm.a ../klm/util/libklm_util.a AM_CPPFLAGS = -DBOOST_TEST_DYN_LINK -W -Wno-sign-compare $(GTEST_CPPFLAGS) -I.. -I../mteval -I../utils -I../klm @@ -82,4 +82,3 @@ libcdec_a_SOURCES = \ JSON_parser.c \ json_parse.cc \ grammar.cc - diff --git a/example_extff/Makefile.am b/example_extff/Makefile.am index ac2694ca..7b7c34b5 100644 --- a/example_extff/Makefile.am +++ b/example_extff/Makefile.am @@ -1,4 +1,4 @@ -AM_CPPFLAGS = -DBOOST_TEST_DYN_LINK -W -Wno-sign-compare $(GTEST_CPPFLAGS) -I.. -I../mteval -I../utils -I../klm -I../decoder +AM_CPPFLAGS = -DBOOST_TEST_DYN_LINK -W -Wall -Wno-sign-compare -I.. -I../mteval -I../utils -I../klm -I../decoder lib_LTLIBRARIES = libff_example.la libff_example_la_SOURCES = ff_example.cc diff --git a/klm/util/have.hh b/klm/util/have.hh index b86ba11e..85b838e4 100644 --- a/klm/util/have.hh +++ b/klm/util/have.hh @@ -11,8 +11,7 @@ #endif #ifdef HAVE_CONFIG_H -// Chris; uncomment this line. -//#include "config.h" +#include "config.h" #endif #endif // UTIL_HAVE__ diff --git a/mteval/Makefile.am b/mteval/Makefile.am index 5e9bba91..4444285f 100644 --- a/mteval/Makefile.am +++ b/mteval/Makefile.am @@ -23,12 +23,12 @@ libmteval_a_SOURCES = \ ter.cc fast_score_SOURCES = fast_score.cc -fast_score_LDADD = libmteval.a $(top_srcdir)/utils/libutils.a -lz +fast_score_LDADD = libmteval.a $(top_srcdir)/utils/libutils.a mbr_kbest_SOURCES = mbr_kbest.cc -mbr_kbest_LDADD = libmteval.a $(top_srcdir)/utils/libutils.a -lz +mbr_kbest_LDADD = libmteval.a $(top_srcdir)/utils/libutils.a scorer_test_SOURCES = scorer_test.cc -scorer_test_LDADD = libmteval.a $(top_srcdir)/utils/libutils.a $(BOOST_UNIT_TEST_FRAMEWORK_LDFLAGS) $(BOOST_UNIT_TEST_FRAMEWORK_LIBS) -lz +scorer_test_LDADD = libmteval.a $(top_srcdir)/utils/libutils.a $(BOOST_UNIT_TEST_FRAMEWORK_LDFLAGS) $(BOOST_UNIT_TEST_FRAMEWORK_LIBS) AM_CPPFLAGS = -DBOOST_TEST_DYN_LINK -W -Wall -Wno-sign-compare $(GTEST_CPPFLAGS) -I$(top_srcdir)/utils diff --git a/python/setup.py.in b/python/setup.py.in index dac72903..fa8a9f5e 100644 --- a/python/setup.py.in +++ b/python/setup.py.in @@ -17,7 +17,7 @@ ext_modules = [ sources=['src/_cdec.cpp'], include_dirs=INC, library_dirs=LIB, - libraries=LIBS + ['z', 'cdec', 'utils', 'mteval', 'training_utils', 'klm', 'klm_util', 'ksearch'], + libraries=['cdec', 'utils', 'mteval', 'training_utils', 'klm', 'klm_util', 'ksearch'] + LIBS, extra_compile_args=CPPFLAGS, extra_link_args=LDFLAGS), Extension(name='cdec.sa._sa', diff --git a/training/dpmert/Makefile.am b/training/dpmert/Makefile.am index ff318bef..3dbdfa69 100644 --- a/training/dpmert/Makefile.am +++ b/training/dpmert/Makefile.am @@ -8,18 +8,18 @@ noinst_PROGRAMS = \ TESTS = lo_test mr_dpmert_generate_mapper_input_SOURCES = mr_dpmert_generate_mapper_input.cc line_optimizer.cc -mr_dpmert_generate_mapper_input_LDADD = $(top_srcdir)/decoder/libcdec.a $(top_srcdir)/mteval/libmteval.a $(top_srcdir)/utils/libutils.a -lz +mr_dpmert_generate_mapper_input_LDADD = $(top_srcdir)/decoder/libcdec.a $(top_srcdir)/mteval/libmteval.a $(top_srcdir)/utils/libutils.a # nbest2hg_SOURCES = nbest2hg.cc -# nbest2hg_LDADD = $(top_srcdir)/decoder/libcdec.a $(top_srcdir)/mteval/libmteval.a $(top_srcdir)/utils/libutils.a -lfst -lz +# nbest2hg_LDADD = $(top_srcdir)/decoder/libcdec.a $(top_srcdir)/mteval/libmteval.a $(top_srcdir)/utils/libutils.a -lfst mr_dpmert_map_SOURCES = mert_geometry.cc ces.cc error_surface.cc mr_dpmert_map.cc line_optimizer.cc -mr_dpmert_map_LDADD = $(top_srcdir)/decoder/libcdec.a $(top_srcdir)/mteval/libmteval.a $(top_srcdir)/utils/libutils.a -lz +mr_dpmert_map_LDADD = $(top_srcdir)/decoder/libcdec.a $(top_srcdir)/mteval/libmteval.a $(top_srcdir)/utils/libutils.a mr_dpmert_reduce_SOURCES = error_surface.cc ces.cc mr_dpmert_reduce.cc line_optimizer.cc mert_geometry.cc -mr_dpmert_reduce_LDADD = $(top_srcdir)/decoder/libcdec.a $(top_srcdir)/mteval/libmteval.a $(top_srcdir)/utils/libutils.a -lz +mr_dpmert_reduce_LDADD = $(top_srcdir)/decoder/libcdec.a $(top_srcdir)/mteval/libmteval.a $(top_srcdir)/utils/libutils.a lo_test_SOURCES = lo_test.cc ces.cc mert_geometry.cc error_surface.cc line_optimizer.cc -lo_test_LDADD = $(BOOST_UNIT_TEST_FRAMEWORK_LDFLAGS) $(BOOST_UNIT_TEST_FRAMEWORK_LIBS) $(top_srcdir)/decoder/libcdec.a $(top_srcdir)/mteval/libmteval.a $(top_srcdir)/utils/libutils.a -lz +lo_test_LDADD = $(BOOST_UNIT_TEST_FRAMEWORK_LDFLAGS) $(BOOST_UNIT_TEST_FRAMEWORK_LIBS) $(top_srcdir)/decoder/libcdec.a $(top_srcdir)/mteval/libmteval.a $(top_srcdir)/utils/libutils.a AM_CPPFLAGS = -DBOOST_TEST_DYN_LINK -W -Wall -Wno-sign-compare $(GTEST_CPPFLAGS) -I$(top_srcdir)/utils -I$(top_srcdir)/decoder -I$(top_srcdir)/mteval diff --git a/training/dtrain/Makefile.am b/training/dtrain/Makefile.am index 5b48e756..4f51b0c8 100644 --- a/training/dtrain/Makefile.am +++ b/training/dtrain/Makefile.am @@ -1,7 +1,7 @@ bin_PROGRAMS = dtrain dtrain_SOURCES = dtrain.cc score.cc -dtrain_LDADD = $(top_srcdir)/decoder/libcdec.a $(top_srcdir)/klm/search/libksearch.a $(top_srcdir)/mteval/libmteval.a $(top_srcdir)/utils/libutils.a $(top_srcdir)/klm/lm/libklm.a $(top_srcdir)/klm/util/libklm_util.a -lz +dtrain_LDADD = $(top_srcdir)/decoder/libcdec.a $(top_srcdir)/klm/search/libksearch.a $(top_srcdir)/mteval/libmteval.a $(top_srcdir)/utils/libutils.a $(top_srcdir)/klm/lm/libklm.a $(top_srcdir)/klm/util/libklm_util.a AM_CPPFLAGS = -W -Wall -Wno-sign-compare -I$(top_srcdir)/utils -I$(top_srcdir)/decoder -I$(top_srcdir)/mteval diff --git a/training/minrisk/Makefile.am b/training/minrisk/Makefile.am index a15e821e..821730c2 100644 --- a/training/minrisk/Makefile.am +++ b/training/minrisk/Makefile.am @@ -1,6 +1,6 @@ bin_PROGRAMS = minrisk_optimize minrisk_optimize_SOURCES = minrisk_optimize.cc -minrisk_optimize_LDADD = $(top_srcdir)/training/utils/libtraining_utils.a $(top_srcdir)/decoder/libcdec.a $(top_srcdir)/mteval/libmteval.a $(top_srcdir)/utils/libutils.a $(top_srcdir)/training/liblbfgs/liblbfgs.a -lz +minrisk_optimize_LDADD = $(top_srcdir)/training/utils/libtraining_utils.a $(top_srcdir)/decoder/libcdec.a $(top_srcdir)/mteval/libmteval.a $(top_srcdir)/utils/libutils.a $(top_srcdir)/training/liblbfgs/liblbfgs.a AM_CPPFLAGS = -W -Wall $(GTEST_CPPFLAGS) -I$(top_srcdir)/utils -I$(top_srcdir)/decoder -I$(top_srcdir)/mteval -I$(top_srcdir)/training -I$(top_srcdir)/training/utils diff --git a/training/mira/Makefile.am b/training/mira/Makefile.am index ae609ede..c8f404fb 100644 --- a/training/mira/Makefile.am +++ b/training/mira/Makefile.am @@ -1,6 +1,6 @@ bin_PROGRAMS = kbest_mira kbest_mira_SOURCES = kbest_mira.cc -kbest_mira_LDADD = $(top_srcdir)/decoder/libcdec.a $(top_srcdir)/klm/search/libksearch.a $(top_srcdir)/mteval/libmteval.a $(top_srcdir)/utils/libutils.a $(top_srcdir)/klm/lm/libklm.a $(top_srcdir)/klm/util/libklm_util.a -lz +kbest_mira_LDADD = $(top_srcdir)/decoder/libcdec.a $(top_srcdir)/klm/search/libksearch.a $(top_srcdir)/mteval/libmteval.a $(top_srcdir)/utils/libutils.a $(top_srcdir)/klm/lm/libklm.a $(top_srcdir)/klm/util/libklm_util.a AM_CPPFLAGS = -W -Wall -Wno-sign-compare -I$(top_srcdir)/utils -I$(top_srcdir)/decoder -I$(top_srcdir)/mteval diff --git a/training/pro/Makefile.am b/training/pro/Makefile.am index 1916b6b2..e0a45a33 100644 --- a/training/pro/Makefile.am +++ b/training/pro/Makefile.am @@ -3,9 +3,9 @@ bin_PROGRAMS = \ mr_pro_reduce mr_pro_map_SOURCES = mr_pro_map.cc -mr_pro_map_LDADD = $(top_srcdir)/training/utils/libtraining_utils.a $(top_srcdir)/decoder/libcdec.a $(top_srcdir)/mteval/libmteval.a $(top_srcdir)/utils/libutils.a -lz +mr_pro_map_LDADD = $(top_srcdir)/training/utils/libtraining_utils.a $(top_srcdir)/decoder/libcdec.a $(top_srcdir)/mteval/libmteval.a $(top_srcdir)/utils/libutils.a mr_pro_reduce_SOURCES = mr_pro_reduce.cc -mr_pro_reduce_LDADD = $(top_srcdir)/training/liblbfgs/liblbfgs.a $(top_srcdir)/utils/libutils.a -lz +mr_pro_reduce_LDADD = $(top_srcdir)/training/liblbfgs/liblbfgs.a $(top_srcdir)/utils/libutils.a AM_CPPFLAGS = -W -Wall -Wno-sign-compare -I$(top_srcdir)/utils -I$(top_srcdir)/decoder -I$(top_srcdir)/mteval -I$(top_srcdir)/training/utils -I$(top_srcdir)/training diff --git a/training/rampion/Makefile.am b/training/rampion/Makefile.am index 1633d0f7..ef0ca147 100644 --- a/training/rampion/Makefile.am +++ b/training/rampion/Makefile.am @@ -1,6 +1,6 @@ bin_PROGRAMS = rampion_cccp rampion_cccp_SOURCES = rampion_cccp.cc -rampion_cccp_LDADD = $(top_srcdir)/training/utils/libtraining_utils.a $(top_srcdir)/decoder/libcdec.a $(top_srcdir)/mteval/libmteval.a $(top_srcdir)/utils/libutils.a -lz +rampion_cccp_LDADD = $(top_srcdir)/training/utils/libtraining_utils.a $(top_srcdir)/decoder/libcdec.a $(top_srcdir)/mteval/libmteval.a $(top_srcdir)/utils/libutils.a AM_CPPFLAGS = -W -Wall $(GTEST_CPPFLAGS) -I$(top_srcdir)/utils -I$(top_srcdir)/decoder -I$(top_srcdir)/mteval -I$(top_srcdir)/training/utils diff --git a/training/utils/Makefile.am b/training/utils/Makefile.am index 189d9a76..c9405d4e 100644 --- a/training/utils/Makefile.am +++ b/training/utils/Makefile.am @@ -24,10 +24,10 @@ libtraining_utils_a_SOURCES = \ risk.cc optimize_test_SOURCES = optimize_test.cc -optimize_test_LDADD = libtraining_utils.a $(top_srcdir)/utils/libutils.a -lz +optimize_test_LDADD = libtraining_utils.a $(top_srcdir)/utils/libutils.a lbfgs_test_SOURCES = lbfgs_test.cc -lbfgs_test_LDADD = $(top_srcdir)/utils/libutils.a -lz +lbfgs_test_LDADD = $(top_srcdir)/utils/libutils.a AM_CPPFLAGS = -W -Wall -Wno-sign-compare -I$(top_srcdir)/decoder -I$(top_srcdir)/utils -I$(top_srcdir)/mteval -I$(top_srcdir)/klm diff --git a/utils/Makefile.am b/utils/Makefile.am index 3ad9d69e..639c30b8 100644 --- a/utils/Makefile.am +++ b/utils/Makefile.am @@ -33,24 +33,24 @@ if HAVE_CMPH endif reconstruct_weights_SOURCES = reconstruct_weights.cc -reconstruct_weights_LDADD = libutils.a -lz +reconstruct_weights_LDADD = libutils.a atools_SOURCES = atools.cc -atools_LDADD = libutils.a -lz +atools_LDADD = libutils.a phmt_SOURCES = phmt.cc -phmt_LDADD = libutils.a $(BOOST_UNIT_TEST_FRAMEWORK_LDFLAGS) $(BOOST_UNIT_TEST_FRAMEWORK_LIBS) -lz +phmt_LDADD = libutils.a $(BOOST_UNIT_TEST_FRAMEWORK_LDFLAGS) $(BOOST_UNIT_TEST_FRAMEWORK_LIBS) ts_SOURCES = ts.cc -ts_LDADD = libutils.a $(BOOST_UNIT_TEST_FRAMEWORK_LDFLAGS) $(BOOST_UNIT_TEST_FRAMEWORK_LIBS) -lz +ts_LDADD = libutils.a $(BOOST_UNIT_TEST_FRAMEWORK_LDFLAGS) $(BOOST_UNIT_TEST_FRAMEWORK_LIBS) m_test_SOURCES = m_test.cc -m_test_LDADD = libutils.a $(BOOST_UNIT_TEST_FRAMEWORK_LDFLAGS) $(BOOST_UNIT_TEST_FRAMEWORK_LIBS) -lz +m_test_LDADD = libutils.a $(BOOST_UNIT_TEST_FRAMEWORK_LDFLAGS) $(BOOST_UNIT_TEST_FRAMEWORK_LIBS) dict_test_SOURCES = dict_test.cc -dict_test_LDADD = libutils.a $(BOOST_UNIT_TEST_FRAMEWORK_LDFLAGS) $(BOOST_UNIT_TEST_FRAMEWORK_LIBS) -lz +dict_test_LDADD = libutils.a $(BOOST_UNIT_TEST_FRAMEWORK_LDFLAGS) $(BOOST_UNIT_TEST_FRAMEWORK_LIBS) weights_test_SOURCES = weights_test.cc -weights_test_LDADD = libutils.a $(BOOST_UNIT_TEST_FRAMEWORK_LDFLAGS) $(BOOST_UNIT_TEST_FRAMEWORK_LIBS) -lz +weights_test_LDADD = libutils.a $(BOOST_UNIT_TEST_FRAMEWORK_LDFLAGS) $(BOOST_UNIT_TEST_FRAMEWORK_LIBS) logval_test_SOURCES = logval_test.cc -logval_test_LDADD = libutils.a $(BOOST_UNIT_TEST_FRAMEWORK_LDFLAGS) $(BOOST_UNIT_TEST_FRAMEWORK_LIBS) -lz +logval_test_LDADD = libutils.a $(BOOST_UNIT_TEST_FRAMEWORK_LDFLAGS) $(BOOST_UNIT_TEST_FRAMEWORK_LIBS) small_vector_test_SOURCES = small_vector_test.cc -small_vector_test_LDADD = libutils.a $(BOOST_UNIT_TEST_FRAMEWORK_LDFLAGS) $(BOOST_UNIT_TEST_FRAMEWORK_LIBS) -lz +small_vector_test_LDADD = libutils.a $(BOOST_UNIT_TEST_FRAMEWORK_LDFLAGS) $(BOOST_UNIT_TEST_FRAMEWORK_LIBS) ################################################################ # do NOT NOT NOT add any other -I includes NO NO NO NO NO ###### diff --git a/word-aligner/Makefile.am b/word-aligner/Makefile.am index 280d3ae7..2dcb688e 100644 --- a/word-aligner/Makefile.am +++ b/word-aligner/Makefile.am @@ -1,6 +1,6 @@ bin_PROGRAMS = fast_align fast_align_SOURCES = fast_align.cc ttables.cc -fast_align_LDADD = $(top_srcdir)/utils/libutils.a -lz +fast_align_LDADD = $(top_srcdir)/utils/libutils.a AM_CPPFLAGS = -W -Wall $(GTEST_CPPFLAGS) -I$(top_srcdir)/utils -I$(top_srcdir)/training -- cgit v1.2.3 From abb935570a86c6c21c5bb3a95e07763247bd4963 Mon Sep 17 00:00:00 2001 From: Chris Dyer Date: Tue, 8 Jan 2013 22:22:28 -0500 Subject: fixes for dist --- decoder/Makefile.am | 6 +++--- decoder/cfg_test.cc | 2 +- decoder/grammar_test.cc | 4 ++-- decoder/hg_test.cc | 2 +- example_extff/Makefile.am | 2 +- klm/lm/Makefile.am | 2 +- klm/search/Makefile.am | 2 +- klm/util/Makefile.am | 2 +- mteval/Makefile.am | 10 ++++++---- mteval/scorer_test.cc | 2 +- training/crf/Makefile.am | 12 ++++++------ training/dpmert/Makefile.am | 12 +++++++----- training/dpmert/lo_test.cc | 2 +- training/dtrain/Makefile.am | 2 +- training/liblbfgs/Makefile.am | 2 +- training/minrisk/Makefile.am | 4 ++-- training/mira/Makefile.am | 2 +- training/pro/Makefile.am | 4 ++-- training/rampion/Makefile.am | 2 +- training/utils/Makefile.am | 6 +++--- utils/Makefile.am | 5 ++++- utils/weights_test.cc | 2 +- word-aligner/Makefile.am | 4 ++-- 23 files changed, 50 insertions(+), 43 deletions(-) (limited to 'training/pro') diff --git a/decoder/Makefile.am b/decoder/Makefile.am index 21187da8..558aeaed 100644 --- a/decoder/Makefile.am +++ b/decoder/Makefile.am @@ -5,7 +5,7 @@ noinst_PROGRAMS = \ hg_test \ parser_test \ grammar_test - + TESTS = trule_test parser_test grammar_test hg_test parser_test_SOURCES = parser_test.cc parser_test_LDADD = $(BOOST_UNIT_TEST_FRAMEWORK_LDFLAGS) $(BOOST_UNIT_TEST_FRAMEWORK_LIBS) libcdec.a ../mteval/libmteval.a ../utils/libutils.a @@ -19,14 +19,14 @@ trule_test_LDADD = $(BOOST_UNIT_TEST_FRAMEWORK_LDFLAGS) $(BOOST_UNIT_TEST_FRAMEW cdec_SOURCES = cdec.cc cdec_LDADD = libcdec.a ../mteval/libmteval.a ../utils/libutils.a ../klm/search/libksearch.a ../klm/lm/libklm.a ../klm/util/libklm_util.a -AM_CPPFLAGS = -DBOOST_TEST_DYN_LINK -W -Wno-sign-compare $(GTEST_CPPFLAGS) -I.. -I../mteval -I../utils -I../klm +AM_CPPFLAGS = -DTEST_DATA=\"$(top_srcdir)/decoder/test_data\" -DBOOST_TEST_DYN_LINK -W -Wno-sign-compare -I$(top_srcdir) -I$(top_srcdir)/mteval -I$(top_srcdir)/utils -I$(top_srcdir)/klm rule_lexer.cc: rule_lexer.ll $(LEX) -s -CF -8 -o$@ $< noinst_LIBRARIES = libcdec.a -EXTRA_DIST = rule_lexer.ll +EXTRA_DIST = test_data rule_lexer.ll libcdec_a_SOURCES = \ JSON_parser.h \ diff --git a/decoder/cfg_test.cc b/decoder/cfg_test.cc index 316c6d16..cbe7d0be 100644 --- a/decoder/cfg_test.cc +++ b/decoder/cfg_test.cc @@ -33,7 +33,7 @@ struct CFGTest : public TestWithParam { istringstream ws(wts); EXPECT_TRUE(ws>>featw); CSHOW(featw) - std::string path(boost::unit_test::framework::master_test_suite().argc == 2 ? boost::unit_test::framework::master_test_suite().argv[1] : "test_data"); + std::string path(boost::unit_test::framework::master_test_suite().argc == 2 ? boost::unit_test::framework::master_test_suite().argv[1] : TEST_DATA); HGSetup::JsonTestFile(&hg,path,file); hg.Reweight(featw); cfg.Init(hg,true,true,false); diff --git a/decoder/grammar_test.cc b/decoder/grammar_test.cc index 912f4f12..6d2c6e67 100644 --- a/decoder/grammar_test.cc +++ b/decoder/grammar_test.cc @@ -19,7 +19,7 @@ using namespace std; struct GrammarTest { GrammarTest() { - std::string path(boost::unit_test::framework::master_test_suite().argc == 2 ? boost::unit_test::framework::master_test_suite().argv[1] : "test_data"); + std::string path(boost::unit_test::framework::master_test_suite().argc == 2 ? boost::unit_test::framework::master_test_suite().argv[1] : TEST_DATA); Weights::InitFromFile(path + "/weights.gt", &wts); } vector wts; @@ -43,7 +43,7 @@ BOOST_AUTO_TEST_CASE(TestTextGrammar) { } BOOST_AUTO_TEST_CASE(TestTextGrammarFile) { - std::string path(boost::unit_test::framework::master_test_suite().argc == 2 ? boost::unit_test::framework::master_test_suite().argv[1] : "test_data"); + std::string path(boost::unit_test::framework::master_test_suite().argc == 2 ? boost::unit_test::framework::master_test_suite().argv[1] : TEST_DATA); GrammarPtr g(new TextGrammar(path + "/grammar.prune")); vector grammars(1, g); diff --git a/decoder/hg_test.cc b/decoder/hg_test.cc index 37469748..8519e559 100644 --- a/decoder/hg_test.cc +++ b/decoder/hg_test.cc @@ -339,7 +339,7 @@ BOOST_AUTO_TEST_CASE(TestAddExpectations) { BOOST_AUTO_TEST_CASE(Small) { Hypergraph hg; - std::string path(boost::unit_test::framework::master_test_suite().argc == 2 ? boost::unit_test::framework::master_test_suite().argv[1] : "test_data"); + std::string path(boost::unit_test::framework::master_test_suite().argc == 2 ? boost::unit_test::framework::master_test_suite().argv[1] : TEST_DATA); CreateSmallHG(&hg, path); SparseVector wts; wts.set_value(FD::Convert("Model_0"), -2.0); diff --git a/example_extff/Makefile.am b/example_extff/Makefile.am index 7b7c34b5..6abfe6c5 100644 --- a/example_extff/Makefile.am +++ b/example_extff/Makefile.am @@ -1,4 +1,4 @@ -AM_CPPFLAGS = -DBOOST_TEST_DYN_LINK -W -Wall -Wno-sign-compare -I.. -I../mteval -I../utils -I../klm -I../decoder +AM_CPPFLAGS = -DBOOST_TEST_DYN_LINK -W -Wall -Wno-sign-compare -I$(top_srcdir) -I$(top_srcdir)/mteval -I$(top_srcdir)/utils -I$(top_srcdir)/klm -I$(top_srcdir)/decoder lib_LTLIBRARIES = libff_example.la libff_example_la_SOURCES = ff_example.cc diff --git a/klm/lm/Makefile.am b/klm/lm/Makefile.am index 436cfd08..870f7128 100644 --- a/klm/lm/Makefile.am +++ b/klm/lm/Makefile.am @@ -55,5 +55,5 @@ libklm_a_SOURCES = \ virtual_interface.cc \ vocab.cc -AM_CPPFLAGS = -W -Wall -Wno-sign-compare $(GTEST_CPPFLAGS) -I.. +AM_CPPFLAGS = -W -Wall -Wno-sign-compare -I$(top_srcdir)/klm diff --git a/klm/search/Makefile.am b/klm/search/Makefile.am index a34f6cea..03554276 100644 --- a/klm/search/Makefile.am +++ b/klm/search/Makefile.am @@ -19,5 +19,5 @@ libksearch_a_SOURCES = \ vertex.cc \ vertex_generator.cc -AM_CPPFLAGS = -W -Wall -Wno-sign-compare -I.. +AM_CPPFLAGS = -W -Wall -Wno-sign-compare -I$(top_srcdir)/klm diff --git a/klm/util/Makefile.am b/klm/util/Makefile.am index bb441432..3ab7560f 100644 --- a/klm/util/Makefile.am +++ b/klm/util/Makefile.am @@ -51,4 +51,4 @@ libklm_util_a_SOURCES = \ string_piece.cc \ usage.cc -AM_CPPFLAGS = -W -Wall -Wno-sign-compare $(GTEST_CPPFLAGS) -I.. +AM_CPPFLAGS = -W -Wall -Wno-sign-compare -I$(top_srcdir)/klm diff --git a/mteval/Makefile.am b/mteval/Makefile.am index b19e4bb1..83adee17 100644 --- a/mteval/Makefile.am +++ b/mteval/Makefile.am @@ -8,6 +8,8 @@ TESTS = scorer_test noinst_LIBRARIES = libmteval.a +EXTRA_DIST = test_data + libmteval_a_SOURCES = \ aer_scorer.h \ comb_scorer.h \ @@ -34,12 +36,12 @@ libmteval_a_SOURCES = \ ter.cc fast_score_SOURCES = fast_score.cc -fast_score_LDADD = libmteval.a $(top_srcdir)/utils/libutils.a +fast_score_LDADD = libmteval.a ../utils/libutils.a mbr_kbest_SOURCES = mbr_kbest.cc -mbr_kbest_LDADD = libmteval.a $(top_srcdir)/utils/libutils.a +mbr_kbest_LDADD = libmteval.a ../utils/libutils.a scorer_test_SOURCES = scorer_test.cc -scorer_test_LDADD = libmteval.a $(top_srcdir)/utils/libutils.a $(BOOST_UNIT_TEST_FRAMEWORK_LDFLAGS) $(BOOST_UNIT_TEST_FRAMEWORK_LIBS) +scorer_test_LDADD = libmteval.a ../utils/libutils.a $(BOOST_UNIT_TEST_FRAMEWORK_LDFLAGS) $(BOOST_UNIT_TEST_FRAMEWORK_LIBS) -AM_CPPFLAGS = -DBOOST_TEST_DYN_LINK -W -Wall -Wno-sign-compare $(GTEST_CPPFLAGS) -I$(top_srcdir)/utils +AM_CPPFLAGS = -DTEST_DATA=\"$(top_srcdir)/mteval/test_data\" -DBOOST_TEST_DYN_LINK -W -Wall -Wno-sign-compare -I$(top_srcdir) -I$(top_srcdir)/utils diff --git a/mteval/scorer_test.cc b/mteval/scorer_test.cc index 9b765d0f..da07f154 100644 --- a/mteval/scorer_test.cc +++ b/mteval/scorer_test.cc @@ -36,7 +36,7 @@ struct Stuff { BOOST_FIXTURE_TEST_SUITE( s, Stuff ); BOOST_AUTO_TEST_CASE(TestCreateFromFiles) { - std::string path(boost::unit_test::framework::master_test_suite().argc == 2 ? boost::unit_test::framework::master_test_suite().argv[1] : "test_data"); + std::string path(boost::unit_test::framework::master_test_suite().argc == 2 ? boost::unit_test::framework::master_test_suite().argv[1] : TEST_DATA); vector files; files.push_back(path + "/re.txt.0"); files.push_back(path + "/re.txt.1"); diff --git a/training/crf/Makefile.am b/training/crf/Makefile.am index f72d8f92..d37b224c 100644 --- a/training/crf/Makefile.am +++ b/training/crf/Makefile.am @@ -7,21 +7,21 @@ bin_PROGRAMS = \ mpi_online_optimize mpi_online_optimize_SOURCES = mpi_online_optimize.cc -mpi_online_optimize_LDADD = $(top_srcdir)/training/utils/libtraining_utils.a $(top_srcdir)/decoder/libcdec.a $(top_srcdir)/klm/search/libksearch.a $(top_srcdir)/mteval/libmteval.a $(top_srcdir)/utils/libutils.a $(top_srcdir)/klm/lm/libklm.a $(top_srcdir)/klm/util/libklm_util.a -lz +mpi_online_optimize_LDADD = ../../training/utils/libtraining_utils.a ../../decoder/libcdec.a ../../klm/search/libksearch.a ../../mteval/libmteval.a ../../utils/libutils.a ../../klm/lm/libklm.a ../../klm/util/libklm_util.a -lz mpi_flex_optimize_SOURCES = mpi_flex_optimize.cc -mpi_flex_optimize_LDADD = $(top_srcdir)/training/utils/libtraining_utils.a $(top_srcdir)/decoder/libcdec.a $(top_srcdir)/klm/search/libksearch.a $(top_srcdir)/mteval/libmteval.a $(top_srcdir)/utils/libutils.a $(top_srcdir)/klm/lm/libklm.a $(top_srcdir)/klm/util/libklm_util.a -lz +mpi_flex_optimize_LDADD = ../../training/utils/libtraining_utils.a ../../decoder/libcdec.a ../../klm/search/libksearch.a ../../mteval/libmteval.a ../../utils/libutils.a ../../klm/lm/libklm.a ../../klm/util/libklm_util.a -lz mpi_extract_reachable_SOURCES = mpi_extract_reachable.cc -mpi_extract_reachable_LDADD = $(top_srcdir)/decoder/libcdec.a $(top_srcdir)/klm/search/libksearch.a $(top_srcdir)/mteval/libmteval.a $(top_srcdir)/utils/libutils.a $(top_srcdir)/klm/lm/libklm.a $(top_srcdir)/klm/util/libklm_util.a -lz +mpi_extract_reachable_LDADD = ../../decoder/libcdec.a ../../klm/search/libksearch.a ../../mteval/libmteval.a ../../utils/libutils.a ../../klm/lm/libklm.a ../../klm/util/libklm_util.a -lz mpi_extract_features_SOURCES = mpi_extract_features.cc -mpi_extract_features_LDADD = $(top_srcdir)/decoder/libcdec.a $(top_srcdir)/klm/search/libksearch.a $(top_srcdir)/mteval/libmteval.a $(top_srcdir)/utils/libutils.a $(top_srcdir)/klm/lm/libklm.a $(top_srcdir)/klm/util/libklm_util.a -lz +mpi_extract_features_LDADD = ../../decoder/libcdec.a ../../klm/search/libksearch.a ../../mteval/libmteval.a ../../utils/libutils.a ../../klm/lm/libklm.a ../../klm/util/libklm_util.a -lz mpi_batch_optimize_SOURCES = mpi_batch_optimize.cc cllh_observer.cc cllh_observer.h -mpi_batch_optimize_LDADD = $(top_srcdir)/training/utils/libtraining_utils.a $(top_srcdir)/decoder/libcdec.a $(top_srcdir)/klm/search/libksearch.a $(top_srcdir)/mteval/libmteval.a $(top_srcdir)/utils/libutils.a $(top_srcdir)/klm/lm/libklm.a $(top_srcdir)/klm/util/libklm_util.a -lz +mpi_batch_optimize_LDADD = ../../training/utils/libtraining_utils.a ../../decoder/libcdec.a ../../klm/search/libksearch.a ../../mteval/libmteval.a ../../utils/libutils.a ../../klm/lm/libklm.a ../../klm/util/libklm_util.a -lz mpi_compute_cllh_SOURCES = mpi_compute_cllh.cc cllh_observer.cc cllh_observer.h -mpi_compute_cllh_LDADD = $(top_srcdir)/decoder/libcdec.a $(top_srcdir)/klm/search/libksearch.a $(top_srcdir)/mteval/libmteval.a $(top_srcdir)/utils/libutils.a $(top_srcdir)/klm/lm/libklm.a $(top_srcdir)/klm/util/libklm_util.a -lz +mpi_compute_cllh_LDADD = ../../decoder/libcdec.a ../../klm/search/libksearch.a ../../mteval/libmteval.a ../../utils/libutils.a ../../klm/lm/libklm.a ../../klm/util/libklm_util.a -lz AM_CPPFLAGS = -DBOOST_TEST_DYN_LINK -W -Wall -Wno-sign-compare -I$(top_srcdir)/training -I$(top_srcdir)/training/utils -I$(top_srcdir)/utils -I$(top_srcdir)/decoder -I$(top_srcdir)/mteval diff --git a/training/dpmert/Makefile.am b/training/dpmert/Makefile.am index e5f13944..cba3e30f 100644 --- a/training/dpmert/Makefile.am +++ b/training/dpmert/Makefile.am @@ -8,18 +8,20 @@ noinst_PROGRAMS = \ TESTS = lo_test mr_dpmert_generate_mapper_input_SOURCES = mr_dpmert_generate_mapper_input.cc line_optimizer.cc -mr_dpmert_generate_mapper_input_LDADD = $(top_srcdir)/decoder/libcdec.a $(top_srcdir)/mteval/libmteval.a $(top_srcdir)/utils/libutils.a +mr_dpmert_generate_mapper_input_LDADD = ../../decoder/libcdec.a ../../mteval/libmteval.a ../../utils/libutils.a # nbest2hg_SOURCES = nbest2hg.cc # nbest2hg_LDADD = $(top_srcdir)/decoder/libcdec.a $(top_srcdir)/mteval/libmteval.a $(top_srcdir)/utils/libutils.a -lfst mr_dpmert_map_SOURCES = mert_geometry.cc ces.cc error_surface.cc mr_dpmert_map.cc line_optimizer.cc ces.h error_surface.h line_optimizer.h mert_geometry.h -mr_dpmert_map_LDADD = $(top_srcdir)/decoder/libcdec.a $(top_srcdir)/mteval/libmteval.a $(top_srcdir)/utils/libutils.a +mr_dpmert_map_LDADD = ../../decoder/libcdec.a ../../mteval/libmteval.a ../../utils/libutils.a mr_dpmert_reduce_SOURCES = error_surface.cc ces.cc mr_dpmert_reduce.cc line_optimizer.cc mert_geometry.cc ces.h error_surface.h line_optimizer.h mert_geometry.h -mr_dpmert_reduce_LDADD = $(top_srcdir)/decoder/libcdec.a $(top_srcdir)/mteval/libmteval.a $(top_srcdir)/utils/libutils.a +mr_dpmert_reduce_LDADD = ../../decoder/libcdec.a ../../mteval/libmteval.a ../../utils/libutils.a lo_test_SOURCES = lo_test.cc ces.cc mert_geometry.cc error_surface.cc line_optimizer.cc ces.h error_surface.h line_optimizer.h mert_geometry.h -lo_test_LDADD = $(BOOST_UNIT_TEST_FRAMEWORK_LDFLAGS) $(BOOST_UNIT_TEST_FRAMEWORK_LIBS) $(top_srcdir)/decoder/libcdec.a $(top_srcdir)/mteval/libmteval.a $(top_srcdir)/utils/libutils.a +lo_test_LDADD = $(BOOST_UNIT_TEST_FRAMEWORK_LDFLAGS) $(BOOST_UNIT_TEST_FRAMEWORK_LIBS) ../../decoder/libcdec.a ../../mteval/libmteval.a ../../utils/libutils.a -AM_CPPFLAGS = -DBOOST_TEST_DYN_LINK -W -Wall -Wno-sign-compare $(GTEST_CPPFLAGS) -I$(top_srcdir)/utils -I$(top_srcdir)/decoder -I$(top_srcdir)/mteval +EXTRA_DIST = test_data + +AM_CPPFLAGS = -DTEST_DATA=\"$(top_srcdir)/training/dpmert/test_data\" -DBOOST_TEST_DYN_LINK -W -Wall -Wno-sign-compare -I$(top_srcdir)/utils -I$(top_srcdir)/decoder -I$(top_srcdir)/mteval diff --git a/training/dpmert/lo_test.cc b/training/dpmert/lo_test.cc index 95a08d3d..d89bcd99 100644 --- a/training/dpmert/lo_test.cc +++ b/training/dpmert/lo_test.cc @@ -118,7 +118,7 @@ BOOST_AUTO_TEST_CASE( TestS1) { to_optimize.push_back(fPhraseModel_1); to_optimize.push_back(fPhraseModel_2); - std::string path(boost::unit_test::framework::master_test_suite().argc == 2 ? boost::unit_test::framework::master_test_suite().argv[1] : "test_data"); + std::string path(boost::unit_test::framework::master_test_suite().argc == 2 ? boost::unit_test::framework::master_test_suite().argv[1] : TEST_DATA); Hypergraph hg; ReadFile rf(path + "/0.json.gz"); diff --git a/training/dtrain/Makefile.am b/training/dtrain/Makefile.am index ee337ca8..8cf71078 100644 --- a/training/dtrain/Makefile.am +++ b/training/dtrain/Makefile.am @@ -1,7 +1,7 @@ bin_PROGRAMS = dtrain dtrain_SOURCES = dtrain.cc score.cc dtrain.h kbestget.h ksampler.h pairsampling.h score.h -dtrain_LDADD = $(top_srcdir)/decoder/libcdec.a $(top_srcdir)/klm/search/libksearch.a $(top_srcdir)/mteval/libmteval.a $(top_srcdir)/utils/libutils.a $(top_srcdir)/klm/lm/libklm.a $(top_srcdir)/klm/util/libklm_util.a +dtrain_LDADD = ../../decoder/libcdec.a ../../klm/search/libksearch.a ../../mteval/libmteval.a ../../utils/libutils.a ../../klm/lm/libklm.a ../../klm/util/libklm_util.a AM_CPPFLAGS = -W -Wall -Wno-sign-compare -I$(top_srcdir)/utils -I$(top_srcdir)/decoder -I$(top_srcdir)/mteval diff --git a/training/liblbfgs/Makefile.am b/training/liblbfgs/Makefile.am index f0d5c8aa..272d6f56 100644 --- a/training/liblbfgs/Makefile.am +++ b/training/liblbfgs/Makefile.am @@ -17,6 +17,6 @@ liblbfgs_a_SOURCES = \ ################################################################ # do NOT NOT NOT add any other -I includes NO NO NO NO NO ###### AM_LDFLAGS = liblbfgs.a -lz -AM_CPPFLAGS = -DBOOST_TEST_DYN_LINK -W -Wall -I. -I.. +AM_CPPFLAGS = -DBOOST_TEST_DYN_LINK -W -Wall -I$(top_srcdir)/training -I$(top_srcdir)/training/liblbfgs ################################################################ diff --git a/training/minrisk/Makefile.am b/training/minrisk/Makefile.am index 821730c2..2be17498 100644 --- a/training/minrisk/Makefile.am +++ b/training/minrisk/Makefile.am @@ -1,6 +1,6 @@ bin_PROGRAMS = minrisk_optimize minrisk_optimize_SOURCES = minrisk_optimize.cc -minrisk_optimize_LDADD = $(top_srcdir)/training/utils/libtraining_utils.a $(top_srcdir)/decoder/libcdec.a $(top_srcdir)/mteval/libmteval.a $(top_srcdir)/utils/libutils.a $(top_srcdir)/training/liblbfgs/liblbfgs.a +minrisk_optimize_LDADD = ../../training/utils/libtraining_utils.a ../../decoder/libcdec.a ../../mteval/libmteval.a ../../utils/libutils.a ../../training/liblbfgs/liblbfgs.a -AM_CPPFLAGS = -W -Wall $(GTEST_CPPFLAGS) -I$(top_srcdir)/utils -I$(top_srcdir)/decoder -I$(top_srcdir)/mteval -I$(top_srcdir)/training -I$(top_srcdir)/training/utils +AM_CPPFLAGS = -W -Wall -I$(top_srcdir)/utils -I$(top_srcdir)/decoder -I$(top_srcdir)/mteval -I$(top_srcdir)/training -I$(top_srcdir)/training/utils diff --git a/training/mira/Makefile.am b/training/mira/Makefile.am index c8f404fb..0084603d 100644 --- a/training/mira/Makefile.am +++ b/training/mira/Makefile.am @@ -1,6 +1,6 @@ bin_PROGRAMS = kbest_mira kbest_mira_SOURCES = kbest_mira.cc -kbest_mira_LDADD = $(top_srcdir)/decoder/libcdec.a $(top_srcdir)/klm/search/libksearch.a $(top_srcdir)/mteval/libmteval.a $(top_srcdir)/utils/libutils.a $(top_srcdir)/klm/lm/libklm.a $(top_srcdir)/klm/util/libklm_util.a +kbest_mira_LDADD = ../../decoder/libcdec.a ../../klm/search/libksearch.a ../../mteval/libmteval.a ../../utils/libutils.a ../../klm/lm/libklm.a ../../klm/util/libklm_util.a AM_CPPFLAGS = -W -Wall -Wno-sign-compare -I$(top_srcdir)/utils -I$(top_srcdir)/decoder -I$(top_srcdir)/mteval diff --git a/training/pro/Makefile.am b/training/pro/Makefile.am index e0a45a33..93889f34 100644 --- a/training/pro/Makefile.am +++ b/training/pro/Makefile.am @@ -3,9 +3,9 @@ bin_PROGRAMS = \ mr_pro_reduce mr_pro_map_SOURCES = mr_pro_map.cc -mr_pro_map_LDADD = $(top_srcdir)/training/utils/libtraining_utils.a $(top_srcdir)/decoder/libcdec.a $(top_srcdir)/mteval/libmteval.a $(top_srcdir)/utils/libutils.a +mr_pro_map_LDADD = ../../training/utils/libtraining_utils.a ../../decoder/libcdec.a ../../mteval/libmteval.a ../../utils/libutils.a mr_pro_reduce_SOURCES = mr_pro_reduce.cc -mr_pro_reduce_LDADD = $(top_srcdir)/training/liblbfgs/liblbfgs.a $(top_srcdir)/utils/libutils.a +mr_pro_reduce_LDADD = ../../training/liblbfgs/liblbfgs.a ../../utils/libutils.a AM_CPPFLAGS = -W -Wall -Wno-sign-compare -I$(top_srcdir)/utils -I$(top_srcdir)/decoder -I$(top_srcdir)/mteval -I$(top_srcdir)/training/utils -I$(top_srcdir)/training diff --git a/training/rampion/Makefile.am b/training/rampion/Makefile.am index ef0ca147..6a1a97cb 100644 --- a/training/rampion/Makefile.am +++ b/training/rampion/Makefile.am @@ -1,6 +1,6 @@ bin_PROGRAMS = rampion_cccp rampion_cccp_SOURCES = rampion_cccp.cc -rampion_cccp_LDADD = $(top_srcdir)/training/utils/libtraining_utils.a $(top_srcdir)/decoder/libcdec.a $(top_srcdir)/mteval/libmteval.a $(top_srcdir)/utils/libutils.a +rampion_cccp_LDADD = ../../training/utils/libtraining_utils.a ../../decoder/libcdec.a ../../mteval/libmteval.a ../../utils/libutils.a AM_CPPFLAGS = -W -Wall $(GTEST_CPPFLAGS) -I$(top_srcdir)/utils -I$(top_srcdir)/decoder -I$(top_srcdir)/mteval -I$(top_srcdir)/training/utils diff --git a/training/utils/Makefile.am b/training/utils/Makefile.am index a2ab86fd..29078aed 100644 --- a/training/utils/Makefile.am +++ b/training/utils/Makefile.am @@ -32,13 +32,13 @@ libtraining_utils_a_SOURCES = \ risk.cc optimize_test_SOURCES = optimize_test.cc -optimize_test_LDADD = libtraining_utils.a $(top_srcdir)/utils/libutils.a +optimize_test_LDADD = libtraining_utils.a ../../utils/libutils.a grammar_convert_SOURCES = grammar_convert.cc -grammar_convert_LDADD = $(top_srcdir)/decoder/libcdec.a $(top_srcdir)/mteval/libmteval.a $(top_srcdir)/utils/libutils.a +grammar_convert_LDADD = ../../decoder/libcdec.a ../../mteval/libmteval.a ../../utils/libutils.a lbfgs_test_SOURCES = lbfgs_test.cc -lbfgs_test_LDADD = $(top_srcdir)/utils/libutils.a +lbfgs_test_LDADD = ../../utils/libutils.a AM_CPPFLAGS = -W -Wall -Wno-sign-compare -I$(top_srcdir)/decoder -I$(top_srcdir)/utils -I$(top_srcdir)/mteval -I$(top_srcdir)/klm diff --git a/utils/Makefile.am b/utils/Makefile.am index 3177325b..c5fedb78 100644 --- a/utils/Makefile.am +++ b/utils/Makefile.am @@ -13,7 +13,10 @@ TESTS = ts small_vector_test logval_test weights_test dict_test m_test noinst_LIBRARIES = libutils.a +# EXTRA_DIST = test_data + libutils_a_SOURCES = \ + test_data \ alias_sampler.h \ alignment_io.h \ array2d.h \ @@ -101,6 +104,6 @@ small_vector_test_LDADD = libutils.a $(BOOST_UNIT_TEST_FRAMEWORK_LDFLAGS) $(BOOS ################################################################ # do NOT NOT NOT add any other -I includes NO NO NO NO NO ###### -AM_CPPFLAGS = -DBOOST_TEST_DYN_LINK -W -Wall -I. +AM_CPPFLAGS = -DBOOST_TEST_DYN_LINK -W -Wall -I. -I$(top_srcdir) -DTEST_DATA=\"$(top_srcdir)/utils/test_data\" ################################################################ diff --git a/utils/weights_test.cc b/utils/weights_test.cc index 4be4c40f..0d5d8512 100644 --- a/utils/weights_test.cc +++ b/utils/weights_test.cc @@ -7,6 +7,6 @@ using namespace std; BOOST_AUTO_TEST_CASE(Load) { vector v; - Weights::InitFromFile("test_data/weights", &v); + Weights::InitFromFile(TEST_DATA "/weights", &v); Weights::WriteToFile("-", v); } diff --git a/word-aligner/Makefile.am b/word-aligner/Makefile.am index e274b209..a195cc5a 100644 --- a/word-aligner/Makefile.am +++ b/word-aligner/Makefile.am @@ -1,6 +1,6 @@ bin_PROGRAMS = fast_align fast_align_SOURCES = fast_align.cc ttables.cc da.h ttables.h -fast_align_LDADD = $(top_srcdir)/utils/libutils.a +fast_align_LDADD = ../utils/libutils.a -AM_CPPFLAGS = -W -Wall $(GTEST_CPPFLAGS) -I$(top_srcdir)/utils -I$(top_srcdir)/training +AM_CPPFLAGS = -W -Wall -I$(top_srcdir) -I$(top_srcdir)/utils -I$(top_srcdir)/training -- cgit v1.2.3 From 0f9eae6d260ec1bd185fd4deff7b59539a51e4bc Mon Sep 17 00:00:00 2001 From: Chris Dyer Date: Sun, 20 Jan 2013 00:59:55 -0500 Subject: make sure distribution contains all required scripts --- Makefile.am | 2 +- training/dpmert/Makefile.am | 2 +- training/minrisk/Makefile.am | 2 ++ training/pro/Makefile.am | 2 ++ training/rampion/Makefile.am | 2 ++ training/utils/Makefile.am | 2 ++ 6 files changed, 10 insertions(+), 2 deletions(-) (limited to 'training/pro') diff --git a/Makefile.am b/Makefile.am index 17190d27..e720ae45 100644 --- a/Makefile.am +++ b/Makefile.am @@ -18,7 +18,7 @@ SUBDIRS = \ #gi/pyp-topics/src gi/clda/src gi/posterior-regularisation/prjava -EXTRA_DIST = python/pkg python/src python/tests python/examples +EXTRA_DIST = corpus tests python/pkg python/src python/tests python/examples compound-split environment AUTOMAKE_OPTIONS = foreign ACLOCAL_AMFLAGS = -I m4 AM_CPPFLAGS = -D_GLIBCXX_PARALLEL diff --git a/training/dpmert/Makefile.am b/training/dpmert/Makefile.am index cba3e30f..b85bb275 100644 --- a/training/dpmert/Makefile.am +++ b/training/dpmert/Makefile.am @@ -22,6 +22,6 @@ mr_dpmert_reduce_LDADD = ../../decoder/libcdec.a ../../mteval/libmteval.a ../../ lo_test_SOURCES = lo_test.cc ces.cc mert_geometry.cc error_surface.cc line_optimizer.cc ces.h error_surface.h line_optimizer.h mert_geometry.h lo_test_LDADD = $(BOOST_UNIT_TEST_FRAMEWORK_LDFLAGS) $(BOOST_UNIT_TEST_FRAMEWORK_LIBS) ../../decoder/libcdec.a ../../mteval/libmteval.a ../../utils/libutils.a -EXTRA_DIST = test_data +EXTRA_DIST = test_data dpmert.pl AM_CPPFLAGS = -DTEST_DATA=\"$(top_srcdir)/training/dpmert/test_data\" -DBOOST_TEST_DYN_LINK -W -Wall -Wno-sign-compare -I$(top_srcdir)/utils -I$(top_srcdir)/decoder -I$(top_srcdir)/mteval diff --git a/training/minrisk/Makefile.am b/training/minrisk/Makefile.am index 2be17498..ebf6fa91 100644 --- a/training/minrisk/Makefile.am +++ b/training/minrisk/Makefile.am @@ -3,4 +3,6 @@ bin_PROGRAMS = minrisk_optimize minrisk_optimize_SOURCES = minrisk_optimize.cc minrisk_optimize_LDADD = ../../training/utils/libtraining_utils.a ../../decoder/libcdec.a ../../mteval/libmteval.a ../../utils/libutils.a ../../training/liblbfgs/liblbfgs.a +EXTRA_DIST = minrisk.pl minrisk_generate_input.pl + AM_CPPFLAGS = -W -Wall -I$(top_srcdir)/utils -I$(top_srcdir)/decoder -I$(top_srcdir)/mteval -I$(top_srcdir)/training -I$(top_srcdir)/training/utils diff --git a/training/pro/Makefile.am b/training/pro/Makefile.am index 93889f34..09364804 100644 --- a/training/pro/Makefile.am +++ b/training/pro/Makefile.am @@ -8,4 +8,6 @@ mr_pro_map_LDADD = ../../training/utils/libtraining_utils.a ../../decoder/libcde mr_pro_reduce_SOURCES = mr_pro_reduce.cc mr_pro_reduce_LDADD = ../../training/liblbfgs/liblbfgs.a ../../utils/libutils.a +EXTRA_DIST = mr_pro_generate_mapper_input.pl pro.pl + AM_CPPFLAGS = -W -Wall -Wno-sign-compare -I$(top_srcdir)/utils -I$(top_srcdir)/decoder -I$(top_srcdir)/mteval -I$(top_srcdir)/training/utils -I$(top_srcdir)/training diff --git a/training/rampion/Makefile.am b/training/rampion/Makefile.am index 6a1a97cb..c72283cd 100644 --- a/training/rampion/Makefile.am +++ b/training/rampion/Makefile.am @@ -3,4 +3,6 @@ bin_PROGRAMS = rampion_cccp rampion_cccp_SOURCES = rampion_cccp.cc rampion_cccp_LDADD = ../../training/utils/libtraining_utils.a ../../decoder/libcdec.a ../../mteval/libmteval.a ../../utils/libutils.a +EXTRA_DIST = rampion.pl rampion_generate_input.pl + AM_CPPFLAGS = -W -Wall $(GTEST_CPPFLAGS) -I$(top_srcdir)/utils -I$(top_srcdir)/decoder -I$(top_srcdir)/mteval -I$(top_srcdir)/training/utils diff --git a/training/utils/Makefile.am b/training/utils/Makefile.am index 29078aed..156a2462 100644 --- a/training/utils/Makefile.am +++ b/training/utils/Makefile.am @@ -9,6 +9,8 @@ noinst_PROGRAMS = \ lbfgs_test \ optimize_test +EXTRA_DIST = decode-and-evaluate.pl libcall.pl parallelize.pl + sentserver_SOURCES = sentserver.c sentserver_LDFLAGS = -pthread -- cgit v1.2.3