Added VW python example for creating and using cache.

shogun-toolbox · Aug 24, 2011 · 6be180c · 6be180c
1 parent dcfb2d6
commit 6be180c
Showing 1 changed file with 49 additions and 0 deletions.
diff --git a/examples/undocumented/python_modular/streaming_vw_createcache_modular.py b/examples/undocumented/python_modular/streaming_vw_createcache_modular.py
@@ -0,0 +1,49 @@
+from modshogun import StreamingVwFile
+from modshogun import StreamingVwCacheFile
+from modshogun import T_SVMLIGHT
+from modshogun import StreamingVwFeatures
+from modshogun import VowpalWabbit
+
+def create_cache():
+    """Creates a binary cache from an ascii data file."""
+
+    # Open the input file as a StreamingVwFile
+    input_file = StreamingVwFile("../data/fm_train_sparsereal.dat")
+    # Default file name will be vw_cache.dat.cache
+    input_file.set_write_to_cache(True)
+
+    # Tell VW that the file is in SVMLight format
+    # Supported types are T_DENSE, T_SVMLIGHT and T_VW
+    input_file.set_parser_type(T_SVMLIGHT)
+
+    # Create a StreamingVwFeatures object, `True' indicating the examples are labelled
+    features = StreamingVwFeatures(input_file, True, 1024)
+
+    # Create a VW object from the features
+    vw = VowpalWabbit(features)
+    vw.set_no_training(True)
+
+    # Train (in this case does nothing but run over all examples)
+    vw.train()
+
+def train_from_cache():
+    """Train using the generated cache file."""
+
+    # Open the input cache file as a StreamingVwCacheFile
+    input_file = StreamingVwCacheFile("vw_cache.dat.cache");
+
+    # The rest is exactly as for normal input
+    features = StreamingVwFeatures(input_file, True, 1024);
+    vw = VowpalWabbit(features)
+    vw.train()
+
+if __name__ == "__main__":
+    print "Creating cache..."
+    create_cache()
+    print "Done."
+    print
+
+    print "Training using the cache file..."
+    print
+    train_from_cache()
+