Saving and Loading model can be tricky due to NullPointerExceptions. May I add documentation for this?

[cruxcode]

Adding to documentation.

[yoavartzi]

Maybe best to write about it here. Not sure how to add it to the README, and we are currently not actively maintaining SPF.

[cruxcode]

Ok!
Writing Model is easy

            File savedModel = new File("SavedModels/mdl1");
            boolean fvar = savedModel.createNewFile();
            Model.<Sentence, LogicalExpression>write(model, savedModel);

Loading model
Be sure to instantiate all services (LogicLanguageService) before reading saved model file. Below code copied from geoquery example. Remember not to close ontology.

			LogicLanguageServices.setInstance(new LogicLanguageServices.Builder(
					new TypeRepository(typesFile), new FlexibleTypeComparator())
							.addConstantsToOntology(simpleOntology)
							.addConstantsToOntology(predOntology)
							.setUseOntology(true).setNumeralTypeName("i")
							.build());

Initialize log if needed.
Build a parser. This parser may not be same as the parser used during training. Look at GeoSimpleExp for example.
Now you are ready to read the model.

                        savedModelFile = new File(filename);
                        model = Model.<Sentence, LogicalExpression>readModel(savedModelFile);

Predict
Here's an example:

        public static IGraphParserOutput<LogicalExpression> predict(String sentence) {
                Sentence dataItem = new Sentence(sentence);
                return parser.parse(dataItem, new DataItemModel<Sentence, LogicalExpression>(model, dataItem), true);
        }

[yoavartzi]

This is great. Thanks! I linked to this issue for the README.