compilable on Cygwin

README

@@ -25,9 +25,15 @@ Requirement:
 LIBFFM is written in C++. It requires C++11 and OpenMP supports. If OpenMP is
 not available on your platform, please refer to section `OpenMP.' 
 
-- Unix-like system:
+- Unix-like systems:
   To compile on Unix-like systems, type `make' in the command line.
 
+- OS X:
+
+  The built-in compiler should be able to compile LIBFFM. However, OpenMP may
+  not be supported. In this case you have to compile without OpenMP. See
+  section `OpenMP' for detail.
+
 - Windows:
   See `Building Windows Binaries' to compile.
 
@@ -63,13 +69,14 @@ Command Line Usage
     -p <path>: set path to the validation set
     -v <fold>: set the number of folds for cross-validation
     --quiet: quiet model (no output)
-    --norm: do instance-wise normalization
+    --no-norm: disable instance-wise normalization
     --no-rand: disable random update
-    --on-disk: perform on-disk training
-
-    `--norm' helps you to do instance-wise normalization. When it is enabled,
-    you can simply assign `1' to `value' in the data.
+    --on-disk: perform on-disk training (a temporary file <training_set_file>.bin will be generated)
 
+    By default we do instance-wise normalization. That is, we normalize the
+    2-norm of each instance to 1. You can use `--no-norm' to disable this
+    function.
+
     By default, our algorithm randomly select an instance for update in each
     inner iteration. On some datasets you may want to do update in the original
     order. You can do it by using `--no-rand' together with `-s 1.'

ffm-predict.cpp

@@ -7,6 +7,7 @@
 #include <cmath>
 #include <stdexcept>
 #include <vector>
+#include <cstdlib>
 
 #include "ffm.h"
 

ffm-train.cpp

@@ -5,6 +5,7 @@
 #include <stdexcept>
 #include <string>
 #include <vector>
+#include <cstdlib>
 
 #include "ffm.h"
 
@@ -25,9 +26,9 @@ string train_help()
 "-p <path>: set path to the validation set\n"
 "-v <fold>: set the number of folds for cross-validation\n"
 "--quiet: quiet model (no output)\n"
-"--norm: do instance-wise normalization\n"
+"--no-norm: disable instance-wise normalization\n"
 "--no-rand: disable random update\n"
-"--on-disk: perform on-disk training\n");
+"--on-disk: perform on-disk training (a temporary file <training_set_file>.bin will be generated)\n");
 }
 
 struct Option
@@ -68,7 +69,7 @@ Option parse_option(int argc, char **argv)
             if(i == argc-1)
                 throw invalid_argument("need to specify number of iterations after -t");
             i++;
-            opt.param.nr_iters = stoi(args[i]);
+            opt.param.nr_iters = atoi(args[i].c_str());
             if(opt.param.nr_iters <= 0)
                 throw invalid_argument("number of iterations should be greater than zero");
         }
@@ -77,7 +78,7 @@ Option parse_option(int argc, char **argv)
             if(i == argc-1)
                 throw invalid_argument("need to specify number of factors after -k");
             i++;
-            opt.param.k = stoi(args[i]);
+            opt.param.k = atoi(args[i].c_str());
             if(opt.param.k <= 0)
                 throw invalid_argument("number of factors should be greater than zero");
         }
@@ -86,7 +87,7 @@ Option parse_option(int argc, char **argv)
             if(i == argc-1)
                 throw invalid_argument("need to specify eta after -r");
             i++;
-            opt.param.eta = stof(args[i]);
+            opt.param.eta = atof(args[i].c_str());
             if(opt.param.eta <= 0)
                 throw invalid_argument("learning rate should be greater than zero");
         }
@@ -95,7 +96,7 @@ Option parse_option(int argc, char **argv)
             if(i == argc-1)
                 throw invalid_argument("need to specify lambda after -l");
             i++;
-            opt.param.lambda = stof(args[i]);
+            opt.param.lambda = atof(args[i].c_str());
             if(opt.param.lambda < 0)
                 throw invalid_argument("regularization cost should not be smaller than zero");
         }
@@ -104,7 +105,7 @@ Option parse_option(int argc, char **argv)
             if(i == argc-1)
                 throw invalid_argument("need to specify number of threads after -s");
             i++;
-            opt.param.nr_threads = stoi(args[i]);
+            opt.param.nr_threads = atoi(args[i].c_str());
             if(opt.param.nr_threads <= 0)
                 throw invalid_argument("number of threads should be greater than zero");
         }
@@ -113,7 +114,7 @@ Option parse_option(int argc, char **argv)
             if(i == argc-1)
                 throw invalid_argument("need to specify number of folds after -v");
             i++;
-            opt.nr_folds = stoi(args[i]);
+            opt.nr_folds = atoi(args[i].c_str());
             if(opt.nr_folds <= 1)
                 throw invalid_argument("number of folds should be greater than one");
             opt.do_cv = true;
@@ -125,9 +126,9 @@ Option parse_option(int argc, char **argv)
             i++;
             opt.va_path = args[i];
         }
-        else if(args[i].compare("--norm") == 0)
+        else if(args[i].compare("--no-norm") == 0)
         {
-            opt.param.normalization= true;
+            opt.param.normalization = false;
         }
         else if(args[i].compare("--quiet") == 0)
         {

ffm.cpp

@@ -197,8 +197,8 @@ void shrink_model(ffm_model &model, ffm_int k_new)
     {
         for(ffm_int f = 0; f < model.m; f++)
         {
-            ffm_float *src = model.W + (j*model.m+f)*model.k*2;
-            ffm_float *dst = model.W + (j*model.m+f)*k_new;
+            ffm_float *src = model.W + ((ffm_long)j*model.m+f)*model.k*2;
+            ffm_float *dst = model.W + ((ffm_long)j*model.m+f)*k_new;
             copy(src, src+k_new, dst);
         }
     }
@@ -543,8 +543,6 @@ ffm_problem* ffm_read_problem(char const *path)
     prob->P = nullptr;
     prob->Y = nullptr;
 
-    int const kMaxLineSize = 1000000;
-
     char line[kMaxLineSize];
 
     ffm_long nnz = 0;
@@ -609,7 +607,7 @@ int ffm_read_problem_to_disk(char const *txt_path, char const *bin_path)
     if(f_bin == nullptr)
         return 1;
 
-    vector<char> line(1000000);
+    vector<char> line(kMaxLineSize);
 
     ffm_int m = 0;
     ffm_int n = 0;
@@ -648,7 +646,7 @@ int ffm_read_problem_to_disk(char const *txt_path, char const *bin_path)
     fwrite(&max_l, sizeof(ffm_int), 1, f_bin);
     fwrite(&max_nnz, sizeof(ffm_long), 1, f_bin);
 
-    while(fgets(line.data(), 1000000, f_txt))
+    while(fgets(line.data(), kMaxLineSize, f_txt))
     {
         char *y_char = strtok(line.data(), " \t");
 
@@ -797,7 +795,7 @@ ffm_parameter ffm_get_default_param()
     param.k = 4;
     param.nr_threads = 1;
     param.quiet = false;
-    param.normalization = false;
+    param.normalization = true;
     param.random = true;
 
     return param;