protected override PolicyDecision <ActionProbability[]> MapContext(VowpalWabbit vw, string context)
{
using (var vwJson = new VowpalWabbitJsonSerializer(vw))
using (VowpalWabbitExampleCollection vwExample = vwJson.ParseAndCreate(context))
{
if (this.developmentMode)
{
Trace.TraceInformation("Example Context: '{0}'", vwExample.VowpalWabbitString);
}
var vwPredictions = vwExample.Predict(VowpalWabbitPredictionType.ActionProbabilities);
// VW multi-label predictions are 0-based
var ap = vwPredictions
.Select(a =>
new ActionProbability
{
Action = (int)(a.Action + 1),
Probability = a.Score
})
.ToArray();
var state = new VWState {
ModelId = vw.Native.ID
};
return(PolicyDecision.Create(ap, state));
}
}
private void DelayedExampleCallback(VowpalWabbitJsonSerializer serializer)
{
try
{
this.perfCounters.Feature_Requests_Pending.IncrementBy(-1);
var data = (PipelineData)serializer.UserContext;
data.Example = serializer.CreateExamples();
// fire and forget
// must not block to avoid dead lock
this.trainProcessorFactory.LearnBlock
.SendAsync(data)
.ContinueWith(async ret =>
{
if (!await ret)
{
this.telemetry.TrackTrace("Unable to enqueue delayed examples", SeverityLevel.Error);
// since we couldn't enqueue, need to dispose here
data.Example.Dispose();
}
});
}
catch (Exception e)
{
this.telemetry.TrackException(e);
}
finally
{
serializer.Dispose();
}
}
public void TestJsonDict()
{
var vec = new float[] { 1, 2, 3 };
var jsonResolver = new RefResolve();
var settings = new JsonSerializerSettings {
ReferenceResolverProvider = () => jsonResolver
};
var ctx1 = new Context(vec, 1, settings);
var ctx2 = new Context(vec, 2, settings);
using (var vw = new VowpalWabbit(new VowpalWabbitSettings {
EnableStringExampleGeneration = true
}))
using (var resolver = new VowpalWabbitJsonReferenceResolver(serializer => Assert.Fail()))
using (var serializer1 = new VowpalWabbitJsonSerializer(vw, resolver))
using (var example1 = serializer1.ParseAndCreate(ctx1.JSON))
using (var serializer2 = new VowpalWabbitJsonSerializer(vw, resolver))
using (var example2 = serializer2.ParseAndCreate(ctx2.JSON))
using (var validator = new VowpalWabbitExampleJsonValidator())
{
validator.Validate("| Id:1 :1 :2 :3", example1);
validator.Validate(ctx1.VW, example1);
validator.Validate("| Id:2 :1 :2 :3", example2);
validator.Validate(ctx2.VW, example2);
}
}
public ActionResult Validate()
{
try
{
APIUtil.Authenticate(this.Request);
if (this.metaData == null || lastDownload + TimeSpan.FromMinutes(1) < DateTime.Now)
{
var url = APIUtil.GetSettingsUrl();
this.metaData = ApplicationMetadataUtil.DownloadMetadata <ApplicationClientMetadata>(url);
lastDownload = DateTime.Now;
}
var context = APIUtil.ReadBody(this.Request);
using (var vw = new VowpalWabbit(new VowpalWabbitSettings(metaData.TrainArguments)
{
EnableStringExampleGeneration = true,
EnableStringFloatCompact = true
}))
using (var serializer = new VowpalWabbitJsonSerializer(vw))
using (var example = serializer.ParseAndCreate(context))
{
return(Json(new { VWExample = example.VowpalWabbitString }));
}
}
catch (UnauthorizedAccessException ex)
{
return(new HttpStatusCodeResult(HttpStatusCode.Unauthorized, ex.Message));
}
catch (Exception ex)
{
new TelemetryClient().TrackException(ex);
return(new HttpStatusCodeResult(HttpStatusCode.InternalServerError, ex.ToString()));
}
}
public async Task <HttpResponseMessage> Post()
{
var header = this.Request.Headers.SingleOrDefault(x => x.Key == "Authorization");
if (header.Value == null)
{
throw new UnauthorizedAccessException("AuthorizationToken missing");
}
var userToken = header.Value.First();
if (string.IsNullOrWhiteSpace(userToken) || userToken != ConfigurationManager.AppSettings["UserToken"])
{
return(Request.CreateResponse(HttpStatusCode.Unauthorized));
}
if (this.metaData == null || lastDownload + TimeSpan.FromMinutes(1) < DateTime.Now)
{
var url = ConfigurationManager.AppSettings["DecisionServiceSettingsUrl"];
this.metaData = ApplicationMetadataUtil.DownloadMetadata <ApplicationClientMetadata>(url);
lastDownload = DateTime.Now;
}
using (var vw = new VowpalWabbit(new VowpalWabbitSettings(metaData.TrainArguments)
{
EnableStringExampleGeneration = true,
EnableStringFloatCompact = true
}))
using (var serializer = new VowpalWabbitJsonSerializer(vw))
using (var example = serializer.ParseAndCreate(new JsonTextReader(new StreamReader(await Request.Content.ReadAsStreamAsync()))))
{
return(Request.CreateResponse(HttpStatusCode.OK, example.VowpalWabbitString));
}
}
/// <summary>
/// Predicts for the given example.
/// </summary>
/// <typeparam name="TPrediction">The prediction type.</typeparam>
/// <param name="reader">The example to predict for.</param>
/// <param name="predictionFactory">The prediction factory to be used. See <see cref="VowpalWabbitPredictionType"/>.</param>
/// <param name="label">
/// Optional label, taking precedence over "_label" property found in <paramref name="reader"/>.
/// If null, <paramref name="reader"/> will be inspected and the "_label" property used as label.
/// </param>
/// <param name="index">Optional index of example the given label should be applied for multi-line examples.</param>
public TPrediction Predict <TPrediction>(JsonReader reader, IVowpalWabbitPredictionFactory <TPrediction> predictionFactory, ILabel label = null, int?index = null)
{
using (var serializer = new VowpalWabbitJsonSerializer(vw))
using (var result = serializer.ParseAndCreate(reader, label, index))
{
return(result.Predict(predictionFactory));
}
}
/// <summary>
/// Learn from the given example and return the current prediction for it.
/// </summary>
/// <typeparam name="TPrediction">The prediction type.</typeparam>
/// <param name="json">The example to learn.</param>
/// <param name="predictionFactory">The prediction factory to be used. See <see cref="VowpalWabbitPredictionType"/>.</param>
/// <param name="label">
/// Optional label, taking precedence over "_label" property found in <paramref name="json"/>.
/// If null, <paramref name="json"/> will be inspected and the "_label" property used as label.
/// </param>
/// <param name="index">Optional index of example the given label should be applied for multi-line examples.</param>
/// <returns>The prediction for the given <paramref name="json"/>.</returns>
public TPrediction Learn <TPrediction>(string json, IVowpalWabbitPredictionFactory <TPrediction> predictionFactory, ILabel label = null, int?index = null)
{
using (var serializer = new VowpalWabbitJsonSerializer(vw))
using (var result = serializer.ParseAndCreate(json, label, index))
{
return(result.Learn(predictionFactory));
}
}
/// <summary>
/// Learns from the given example.
/// </summary>
/// <param name="reader">The example to learn.</param>
/// <param name="label">
/// Optional label, taking precedence over "_label" property found in <paramref name="reader"/>.
/// If null, <paramref name="reader"/> will be inspected and the "_label" property used as label.
/// </param>
/// <param name="index">Optional index of example the given label should be applied for multi-line examples.</param>
public void Learn(JsonReader reader, ILabel label = null, int?index = null)
{
using (var serializer = new VowpalWabbitJsonSerializer(vw))
using (var result = serializer.ParseAndCreate(reader, label, index))
{
result.Learn();
}
}
public void Validate(string line, string json, IVowpalWabbitLabelComparator labelComparator = null, ILabel label = null)
{
using (var jsonSerializer = new VowpalWabbitJsonSerializer(this.vw))
using (var jsonExample = jsonSerializer.ParseAndCreate(json, label))
{
this.Validate(line, jsonExample, labelComparator, label);
}
}
public void Validate(string[] lines, JsonReader jsonReader, IVowpalWabbitLabelComparator labelComparator = null, ILabel label = null, int?index = null, VowpalWabbitJsonExtension extension = null)
{
VowpalWabbitExample[] strExamples = new VowpalWabbitExample[lines.Count()];
try
{
for (int i = 0; i < lines.Length; i++)
{
strExamples[i] = this.vw.ParseLine(lines[i]);
}
using (var jsonSerializer = new VowpalWabbitJsonSerializer(this.vw))
{
if (extension != null)
{
jsonSerializer.RegisterExtension(extension);
// extension are not supported with native JSON parsing
}
using (var jsonExample = (VowpalWabbitMultiLineExampleCollection)jsonSerializer.ParseAndCreate(jsonReader, label, index))
{
var jsonExamples = new List <VowpalWabbitExample>();
if (jsonExample.SharedExample != null)
{
jsonExamples.Add(jsonExample.SharedExample);
}
jsonExamples.AddRange(jsonExample.Examples);
Assert.AreEqual(strExamples.Length, jsonExamples.Count);
for (int i = 0; i < strExamples.Length; i++)
{
using (var strJsonExample = this.vw.ParseLine(jsonExamples[i].VowpalWabbitString))
{
var diff = strExamples[i].Diff(this.vw, jsonExamples[i], labelComparator);
Assert.IsNull(diff, diff + " generated string: '" + jsonExamples[i].VowpalWabbitString + "'");
diff = strExamples[i].Diff(this.vw, strJsonExample, labelComparator);
Assert.IsNull(diff, diff);
}
}
}
}
}
finally
{
foreach (var ex in strExamples)
{
if (ex != null)
{
ex.Dispose();
}
}
}
}
public void TestJsonLabelExtraction()
{
using (var vw = new VowpalWabbit("--cb_adf --rank_all"))
{
using (var jsonSerializer = new VowpalWabbitJsonSerializer(vw))
{
string eventId = null;
jsonSerializer.RegisterExtension((state, property) =>
{
Assert.AreEqual(property, "_eventid");
Assert.IsTrue(state.Reader.Read());
eventId = (string)state.Reader.Value;
return(true);
});
jsonSerializer.Parse("{\"_eventid\":\"abc123\",\"a\":1,\"_label_cost\":-1,\"_label_probability\":0.3}");
Assert.AreEqual("abc123", eventId);
using (var examples = jsonSerializer.CreateExamples())
{
var single = examples as VowpalWabbitSingleLineExampleCollection;
Assert.IsNotNull(single);
var label = single.Example.Label as ContextualBanditLabel;
Assert.IsNotNull(label);
Assert.AreEqual(-1, label.Cost);
Assert.AreEqual(0.3, label.Probability, 0.0001);
}
}
using (var jsonSerializer = new VowpalWabbitJsonSerializer(vw))
{
jsonSerializer.Parse("{\"_multi\":[{\"_text\":\"w1 w2\", \"a\":{\"x\":1}}, {\"_text\":\"w2 w3\"}], \"_labelindex\":1, \"_label_cost\":-1, \"_label_probability\":0.3}");
using (var examples = jsonSerializer.CreateExamples())
{
var multi = examples as VowpalWabbitMultiLineExampleCollection;
Assert.IsNotNull(multi);
Assert.AreEqual(2, multi.Examples.Length);
var label = multi.Examples[0].Label as ContextualBanditLabel;
Assert.AreEqual(0, label.Cost);
Assert.AreEqual(0, label.Probability);
label = multi.Examples[1].Label as ContextualBanditLabel;
Assert.IsNotNull(label);
Assert.AreEqual(-1, label.Cost);
Assert.AreEqual(0.3, label.Probability, 0.0001);
}
}
}
}
/// <summary>
/// Predicts for the given example.
/// </summary>
/// <param name="reader">The example to predict for.</param>
/// <param name="label">
/// Optional label, taking precedence over "_label" property found in <paramref name="reader"/>.
/// If null, <paramref name="reader"/> will be inspected and the "_label" property used as label.
/// </param>
/// <param name="index">Optional index of example the given label should be applied for multi-line examples.</param>
public void Predict(JsonReader reader, ILabel label = null, int?index = null)
{
Contract.Requires(reader != null);
using (var serializer = new VowpalWabbitJsonSerializer(vw))
using (var result = serializer.ParseAndCreate(reader, label, index))
{
result.Predict();
}
}
/// <summary>
/// Predicts for the given example.
/// </summary>
/// <typeparam name="TPrediction">The prediction type.</typeparam>
/// <param name="json">The example to predict for.</param>
/// <param name="predictionFactory">The prediction factory to be used. See <see cref="VowpalWabbitPredictionType"/>.</param>
/// <param name="label">
/// Optional label, taking precedence over "_label" property found in <paramref name="json"/>.
/// If null, <paramref name="json"/> will be inspected and the "_label" property used as label.
/// </param>
/// <param name="index">Optional index of example the given label should be applied for multi-line examples.</param>
public TPrediction Predict <TPrediction>(string json, IVowpalWabbitPredictionFactory <TPrediction> predictionFactory, ILabel label = null, int?index = null)
{
Contract.Requires(json != null);
using (var serializer = new VowpalWabbitJsonSerializer(vw))
using (var result = serializer.ParseAndCreate(json, label, index))
{
return(result.Predict(predictionFactory));
}
}
public void TestJsonFeatureExtraction()
{
string json = "{\"ns1\":{\"location\":\"New York\", \"f2\":3.4}}";
using (var vw = new VowpalWabbit("-b 3 --noconstant"))
using (var serializer = new VowpalWabbitJsonSerializer(vw))
using (var result = serializer.ParseAndCreate(json))
{
var singleExample = result as VowpalWabbitSingleLineExampleCollection;
Assert.IsNotNull(singleExample);
if (singleExample != null)
{
foreach (var ns in singleExample.Example)
{
Console.WriteLine(ns.Index);
foreach (var feature in ns)
{
Console.WriteLine("{0}:{1}", feature.FeatureIndex, feature.X);
}
}
var ns1 = singleExample.Example.ToArray();
Assert.AreEqual(1, ns1.Length);
Assert.AreEqual((byte)'n', ns1[0].Index);
CollectionAssert.AreEqual(
new[] {
new VowpalWabbitFeature(singleExample.Example, 1, 12),
new VowpalWabbitFeature(singleExample.Example, 3.4f, 28)
},
ns1[0].ToArray());
}
// for documentation purpose only
var multiExample = result as VowpalWabbitMultiLineExampleCollection;
Assert.IsNull(multiExample);
if (multiExample != null)
{
foreach (var example in multiExample.Examples)
{
foreach (var ns in example)
{
Console.WriteLine(ns.Index);
foreach (var feature in ns)
{
Console.WriteLine("{0}:{1}", feature.FeatureIndex, feature.X);
}
}
}
}
}
}
/// <summary>
/// Determines the action to take for a given context.
/// This implementation should be thread-safe if multithreading is needed.
/// </summary>
/// <param name="vw">The Vowpal Wabbit instance to use.</param>
/// <param name="context">A user-defined context for the decision.</param>
/// <returns>A decision tuple containing the index of the action to take (1-based), and the Id of the model or policy used to make the decision.
/// Can be null if the Policy is not ready yet (e.g. model not loaded).</returns>
protected override PolicyDecision <int> MapContext(VowpalWabbit vw, string context)
{
using (var vwJson = new VowpalWabbitJsonSerializer(vw))
using (VowpalWabbitExampleCollection vwExample = vwJson.ParseAndCreate(context))
{
var action = (int)vwExample.Predict(VowpalWabbitPredictionType.CostSensitive);
var state = new VWState {
ModelId = vw.ID
};
return(PolicyDecision.Create(action, state));
}
}
public void Validate(string line, string json, IVowpalWabbitLabelComparator labelComparator = null, ILabel label = null)
{
using (var strExample = this.vw.ParseLine(line))
using (var jsonSerializer = new VowpalWabbitJsonSerializer(this.vw))
using (var jsonExample = (VowpalWabbitSingleLineExampleCollection)jsonSerializer.ParseAndCreate(json, label))
using (var strJsonExample = this.vw.ParseLine(jsonExample.Example.VowpalWabbitString))
{
var diff = strExample.Diff(this.vw, jsonExample.Example, labelComparator);
Assert.IsNull(diff, diff + " generated string: '" + jsonExample.VowpalWabbitString + "'");
diff = strExample.Diff(this.vw, strJsonExample, labelComparator);
Assert.IsNull(diff, diff);
}
}
/// <summary>
/// Determines the action to take for a given context.
/// This implementation should be thread-safe if multithreading is needed.
/// </summary>
/// <param name="vw">The Vowpal Wabbit instance to use.</param>
/// <param name="context">A user-defined context for the decision.</param>
/// <returns>A decision tuple containing the index of the action to take (1-based), and the Id of the model or policy used to make the decision.
/// Can be null if the Policy is not ready yet (e.g. model not loaded).</returns>
protected override PolicyDecision <int[]> MapContext(VowpalWabbit vw, string context)
{
using (var vwJson = new VowpalWabbitJsonSerializer(vw))
using (VowpalWabbitExampleCollection vwExample = vwJson.ParseAndCreate(context))
{
ActionScore[] vwMultilabelPredictions = vwExample.Predict(VowpalWabbitPredictionType.ActionProbabilities);
// VW multi-label predictions are 0-based
var actions = vwMultilabelPredictions.Select(a => (int)a.Action + 1).ToArray();
var state = new VWState {
ModelId = vw.ID
};
return(PolicyDecision.Create(actions, state));
}
}
private void updateModelMaybe()
{
if (sinceLastUpdate >= ModelUpdateInterval)
{
// Locking at this level ensures a batch of events is processed completely before
// the next batch (finer locking would allow interleaving, violating timeorder
lock (this.vwLock)
{
// Exit gracefully if the object has been disposed
if (vwDisposed)
{
return;
}
foreach (var dp in log.FlushCompleteEvents())
{
uint action = (uint)((int[])dp.InteractData.Value)[0];
var label = new ContextualBanditLabel(action, -dp.Reward, ((GenericTopSlotExplorerState)dp.InteractData.ExplorerState).Probabilities[0]);
// String (json) contexts need to be handled specially, since the C# interface
// does not currently handle the CB label properly
if (typeof(TContext) == typeof(string))
{
// Manually insert the CB label fields into the context
string labelStr = string.Format(CultureInfo.InvariantCulture, "\"_label_Action\":{0},\"_label_Cost\":{1},\"_label_Probability\":{2},\"_labelIndex\":{3},",
label.Action, label.Cost, label.Probability, label.Action - 1);
string context = ((string)dp.InteractData.Context).Insert(1, labelStr);
using (var vwSerializer = new VowpalWabbitJsonSerializer(vwJson.Native))
using (VowpalWabbitExampleCollection vwExample = vwSerializer.ParseAndCreate(context))
{
vwExample.Learn();
}
}
else
{
vw.Learn((TContext)dp.InteractData.Context, label, index: (int)label.Action - 1);
}
}
using (MemoryStream currModel = new MemoryStream())
{
VowpalWabbit vwNative = (typeof(TContext) == typeof(string)) ? vwJson.Native : vw.Native;
vwNative.SaveModel(currModel);
currModel.Position = 0;
this.UpdateModel(currModel);
sinceLastUpdate = 0;
}
}
}
}
public void TestJsonNumADFs()
{
using (var validator = new VowpalWabbitExampleJsonValidator(""))
{
Assert.AreEqual(2,
VowpalWabbitJsonSerializer.GetNumberOfActionDependentExamples(
"{\"_text\":\"a b c\",\"a\":{\"_text\":\"d e f\"},_multi:[{\"a\":1},{\"b\":2,\"c\":3}]}"));
Assert.AreEqual(0,
VowpalWabbitJsonSerializer.GetNumberOfActionDependentExamples(
"{\"_text\":\"a b c\",\"a\":{\"_text\":\"d e f\"},_multi:[]}"));
Assert.AreEqual(0,
VowpalWabbitJsonSerializer.GetNumberOfActionDependentExamples(
"{\"_text\":\"a b c\",\"a\":{\"_text\":\"d e f\"}}"));
}
}
public void Validate(string line, string json, IVowpalWabbitLabelComparator labelComparator = null, ILabel label = null, bool enableNativeJsonValidation = true)
{
using (var jsonSerializer = new VowpalWabbitJsonSerializer(this.vw))
using (var jsonExample = jsonSerializer.ParseAndCreate(json, label))
{
this.Validate(line, jsonExample, labelComparator, label);
if (enableNativeJsonValidation)
{
var examples = this.vw.ParseJson(json);
Assert.AreEqual(1, examples.Count);
using (var jsonNativeExample = new VowpalWabbitSingleLineExampleCollection(this.vw, examples[0]))
{
this.Validate(line, jsonNativeExample, labelComparator, label, validateVowpalWabbitString: false);
}
}
}
}
public override void WriteLine(string value)
{
using (var jsonSerializer = new VowpalWabbitJsonSerializer(vw))
using (var example = jsonSerializer.ParseAndCreate(value))
{
if (example == null)
{
throw new InvalidDataException($"Invalid example: {value}");
}
var str = example.VowpalWabbitString;
if (example is VowpalWabbitMultiLineExampleCollection)
{
str += "\n";
}
base.WriteLine(str);
}
}
public void TestJsonDictReverse()
{
var vec = new float[] { 1, 2, 3 };
var jsonResolver = new RefResolve();
var settings = new JsonSerializerSettings {
ReferenceResolverProvider = () => jsonResolver
};
var ctx1 = new Context(vec, 1, settings);
var ctx2 = new Context(vec, 2, settings);
VowpalWabbitJsonSerializer delayedSerializer = null;
using (var validator = new VowpalWabbitExampleJsonValidator())
using (var vw = new VowpalWabbit(new VowpalWabbitSettings {
EnableStringExampleGeneration = true
}))
using (var resolver = new VowpalWabbitJsonReferenceResolver(serializer => delayedSerializer = serializer))
{
var serializer2 = new VowpalWabbitJsonSerializer(vw, resolver);
var example2 = serializer2.ParseAndCreate(ctx2.JSON);
// incomplete data
Assert.IsNull(example2);
// triggers example2 completion
using (var serializer1 = new VowpalWabbitJsonSerializer(vw, resolver))
using (var example1 = serializer1.ParseAndCreate(ctx1.JSON))
{
validator.Validate("| Id:1 :1 :2 :3", example1);
}
Assert.IsNotNull(delayedSerializer);
using (var delayedExample2 = delayedSerializer.CreateExamples())
{
validator.Validate("| Id:2 :1 :2 :3", delayedExample2);
}
delayedSerializer.Dispose();
}
}
public void Validate(string line, TExample example, ILabel label = null)
{
IVowpalWabbitLabelComparator comparator;
if (label == null || label == SharedLabel.Instance)
{
comparator = null;
}
else if (label is SimpleLabel)
{
comparator = VowpalWabbitLabelComparator.Simple;
}
else if (label is ContextualBanditLabel)
{
comparator = VowpalWabbitLabelComparator.ContextualBandit;
}
else
{
throw new ArgumentException("Label type not supported: " + label.GetType());
}
using (var context = new VowpalWabbitMarshalContext(this.vw.Native))
using (var contextNative = new VowpalWabbitMarshalContext(this.vwNative.Native))
{
// validate string serializer
this.serializer(context, example, label);
this.serializerNative(contextNative, example, label);
// natively parsed string example compared against:
// (1) natively build example
// (2) string serialized & natively parsed string example
using (var strExample = this.vw.Native.ParseLine(line))
using (var strConvertedExample = this.vw.Native.ParseLine(context.StringExample.ToString()))
using (var nativeExample = contextNative.ExampleBuilder.CreateExample())
using (var nativeExampleWithString = this.factorySerializer.Serialize(example, label))
{
var diff = strExample.Diff(this.vw.Native, strConvertedExample, comparator);
Assert.IsNull(diff, diff + " generated string: '" + context.StringExample + "'");
diff = strExample.Diff(this.vw.Native, nativeExample, comparator);
Assert.IsNull(diff, diff);
if (!strExample.IsNewLine)
{
Assert.IsFalse(string.IsNullOrEmpty(nativeExampleWithString.VowpalWabbitString));
Assert.IsFalse(string.IsNullOrEmpty(this.factorySerializer.SerializeToString(example, label)));
}
if (this.vw.Native.Settings.FeatureDiscovery == VowpalWabbitFeatureDiscovery.Json)
{
var jsonStr = JsonConvert.SerializeObject(example);
using (var jsonSerializer = new VowpalWabbitJsonSerializer(this.vw.Native))
{
using (var jsonExample = jsonSerializer.ParseAndCreate(jsonStr, label))
{
var ex = ((VowpalWabbitSingleLineExampleCollection)jsonExample).Example;
diff = strExample.Diff(this.vw.Native, ex, comparator);
Assert.IsNull(diff, diff + "\njson: '" + jsonStr + "'");
}
}
}
}
}
}
private IEnumerable <PipelineData> Stage1_Deserialize(PipelineData data)
{
try
{
using (var jsonReader = new JsonTextReader(new StringReader(data.JSON)))
{
//jsonReader.FloatParser = Util.ReadDoubleString;
// jsonReader.ArrayPool = pool;
VowpalWabbitJsonSerializer vwJsonSerializer = null;
try
{
vwJsonSerializer = new VowpalWabbitJsonSerializer(this.trainer.VowpalWabbit, this.trainer.ReferenceResolver);
vwJsonSerializer.RegisterExtension((state, property) =>
{
if (property.Equals("_eventid", StringComparison.OrdinalIgnoreCase))
{
if (!state.Reader.Read() && state.Reader.TokenType != JsonToken.String)
{
throw new VowpalWabbitJsonException(state.Reader, "Expected string");
}
data.EventId = (string)state.Reader.Value;
return(true);
}
else if (property.Equals("_timestamp", StringComparison.OrdinalIgnoreCase))
{
if (!state.Reader.Read() && state.Reader.TokenType != JsonToken.Date)
{
throw new VowpalWabbitJsonException(state.Reader, "Expected date");
}
data.Timestamp = (DateTime)state.Reader.Value;
}
return(false);
});
data.Example = vwJsonSerializer.ParseAndCreate(jsonReader);
if (data.Example == null)
{
// unable to create example due to missing data
// will be trigger later
vwJsonSerializer.UserContext = data.Example;
// make sure the serialize is not deallocated
vwJsonSerializer = null;
}
}
finally
{
if (vwJsonSerializer != null)
{
vwJsonSerializer.Dispose();
}
}
performanceCounters.Stage1_JSON_DeserializePerSec.Increment();
// delayed
if (data.Example == null)
{
this.performanceCounters.Feature_Requests_Pending.Increment();
yield break;
}
}
}
catch (Exception ex)
{
this.telemetry.TrackException(ex, new Dictionary <string, string> {
{ "JSON", data.JSON }
});
this.performanceCounters.Stage2_Faulty_Examples_Total.Increment();
this.performanceCounters.Stage2_Faulty_ExamplesPerSec.Increment();
yield break;
}
yield return(data);
}
public static void Convert(StreamReader reader, StreamWriter writer)
{
var line = reader.ReadLine();
if (line == null)
{
return;
}
var jExample = JObject.Parse(line);
var settings = jExample.Properties().Any(p => p.Name == "_multi") ? "--cb_explore_adf" : "--cb_explore";
int lineNr = 1;
using (var vw = new VowpalWabbit(new VowpalWabbitSettings(settings)
{
EnableStringExampleGeneration = true,
EnableStringFloatCompact = true,
EnableThreadSafeExamplePooling = true
}))
{
var serializeBlock = new TransformBlock <Tuple <string, int>, string>(l =>
{
using (var jsonSerializer = new VowpalWabbitJsonSerializer(vw))
using (var example = jsonSerializer.ParseAndCreate(l.Item1))
{
if (example == null)
{
throw new InvalidDataException($"Invalid example in line {l.Item2}: '{l.Item1}'");
}
var str = example.VowpalWabbitString;
if (example is VowpalWabbitMultiLineExampleCollection)
{
str += "\n";
}
return(str);
}
},
new ExecutionDataflowBlockOptions
{
BoundedCapacity = 1024,
MaxDegreeOfParallelism = 8
});
var writeBlock = new ActionBlock <string>(
l => writer.WriteLine(l),
new ExecutionDataflowBlockOptions {
MaxDegreeOfParallelism = 1, BoundedCapacity = 128
});
serializeBlock.LinkTo(writeBlock, new DataflowLinkOptions {
PropagateCompletion = true
});
var input = serializeBlock.AsObserver();
do
{
input.OnNext(Tuple.Create(line, lineNr));
lineNr++;
} while ((line = reader.ReadLine()) != null);
input.OnCompleted();
serializeBlock.Completion.Wait();
}
}
//private class Event
//{
// internal VowpalWabbitExampleCollection Example;
// internal string Line;
// internal int LineNr;
// internal ActionScore[] Prediction;
//}
/// <summary>
/// Train VW on offline data.
/// </summary>
/// <param name="arguments">Base arguments.</param>
/// <param name="inputFile">Path to input file.</param>
/// <param name="predictionFile">Name of the output prediction file.</param>
/// <param name="reloadInterval">The TimeSpan interval to reload model.</param>
/// <param name="learningRate">
/// Learning rate must be specified here otherwise on Reload it will be reset.
/// </param>
/// <param name="cacheFilePrefix">
/// The prefix of the cache file name to use. For example: prefix = "test" => "test.vw.cache"
/// If none or null, the input file name is used, e.g. "input.dataset" => "input.vw.cache"
/// !!! IMPORTANT !!!: Always use a new cache name if a different dataset or reload interval is used.
/// </param>
/// <remarks>
/// Both learning rates and cache file are added to initial training arguments as well as Reload arguments.
/// </remarks>
public static void Train(string arguments, string inputFile, string predictionFile = null, TimeSpan?reloadInterval = null, float?learningRate = null, string cacheFilePrefix = null)
{
var learningArgs = learningRate == null ? string.Empty : $" -l {learningRate}";
int cacheIndex = 0;
var cacheArgs = (Func <int, string>)(i => $" --cache_file {cacheFilePrefix ?? Path.GetFileNameWithoutExtension(inputFile)}-{i}.vw.cache");
using (var reader = new StreamReader(inputFile))
using (var prediction = new StreamWriter(predictionFile ?? inputFile + ".prediction"))
using (var vw = new VowpalWabbit(new VowpalWabbitSettings(arguments + learningArgs + cacheArgs(cacheIndex++))
{
Verbose = true
}))
{
string line;
int lineNr = 0;
int invalidExamples = 0;
DateTime?lastTimestamp = null;
while ((line = reader.ReadLine()) != null)
{
try
{
bool reload = false;
using (var jsonSerializer = new VowpalWabbitJsonSerializer(vw))
{
if (reloadInterval != null)
{
jsonSerializer.RegisterExtension((state, property) =>
{
if (property.Equals("_timestamp", StringComparison.Ordinal))
{
var eventTimestamp = state.Reader.ReadAsDateTime();
if (lastTimestamp == null)
{
lastTimestamp = eventTimestamp;
}
else if (lastTimestamp + reloadInterval < eventTimestamp)
{
reload = true;
lastTimestamp = eventTimestamp;
}
return(true);
}
return(false);
});
}
// var pred = vw.Learn(line, VowpalWabbitPredictionType.ActionScore);
using (var example = jsonSerializer.ParseAndCreate(line))
{
var pred = example.Learn(VowpalWabbitPredictionType.ActionScore);
prediction.WriteLine(JsonConvert.SerializeObject(
new
{
nr = lineNr,
@as = pred.Select(x => x.Action),
p = pred.Select(x => x.Score)
}));
}
if (reload)
{
vw.Reload(learningArgs + cacheArgs(cacheIndex++));
}
}
}
catch (Exception)
{
invalidExamples++;
}
lineNr++;
}
}
// memory leak and not much gain below...
//using (var vw = new VowpalWabbit(new VowpalWabbitSettings(arguments)
//{
// Verbose = true,
// EnableThreadSafeExamplePooling = true,
// MaxExamples = 1024
//}))
//using (var reader = new StreamReader(inputFile))
//using (var prediction = new StreamWriter(inputFile + ".prediction"))
//{
// int invalidExamples = 0;
// var deserializeBlock = new TransformBlock<Event, Event>(
// evt =>
// {
// try
// {
// using (var vwJsonSerializer = new VowpalWabbitJsonSerializer(vw))
// {
// evt.Example = vwJsonSerializer.ParseAndCreate(evt.Line);
// }
// // reclaim memory
// evt.Line = null;
// return evt;
// }
// catch (Exception)
// {
// Interlocked.Increment(ref invalidExamples);
// return null;
// }
// },
// new ExecutionDataflowBlockOptions
// {
// BoundedCapacity = 16,
// MaxDegreeOfParallelism = 8 // TODO: parameterize
// });
// var learnBlock = new TransformBlock<Event, Event>(
// evt =>
// {
// evt.Prediction = evt.Example.Learn(VowpalWabbitPredictionType.ActionScore);
// evt.Example.Dispose();
// return evt;
// },
// new ExecutionDataflowBlockOptions
// {
// BoundedCapacity = 64,
// MaxDegreeOfParallelism = 1
// });
// var predictionBlock = new ActionBlock<Event>(
// evt => prediction.WriteLine(evt.LineNr + " " + string.Join(",", evt.Prediction.Select(a_s => $"{a_s.Action}:{a_s.Score}"))),
// new ExecutionDataflowBlockOptions
// {
// BoundedCapacity = 16,
// MaxDegreeOfParallelism = 1
// });
// var input = deserializeBlock.AsObserver();
// deserializeBlock.LinkTo(learnBlock, new DataflowLinkOptions { PropagateCompletion = true }, evt => evt != null);
// deserializeBlock.LinkTo(DataflowBlock.NullTarget<object>());
// learnBlock.LinkTo(predictionBlock, new DataflowLinkOptions { PropagateCompletion = true });
// string line;
// int lineNr = 0;
// while ((line = reader.ReadLine()) != null)
// input.OnNext(new Event { Line = line, LineNr = lineNr++ });
// input.OnCompleted();
// predictionBlock.Completion.Wait();
//Console.WriteLine($"Examples {lineNr}. Invalid: {invalidExamples}");
//}
}
public void TestJsonDictThreading()
{
var jsonResolver = new RefResolve();
var settings = new JsonSerializerSettings {
ReferenceResolverProvider = () => jsonResolver
};
var rnd = new Random(123);
var examples = new List <Context>();
var id = 0;
// different reference objects
for (int i = 0; i < 10; i++)
{
var data = Enumerable.Range(1, 5).Select(_ => (float)rnd.Next(10)).ToArray();
// referencing the same data
for (int j = 0; j < 5; j++)
{
examples.Add(new Context(data, id++, settings));
}
}
for (int i = 0; i < 4; i++)
{
Permute(examples, rnd);
for (int maxDegreeOfParallelism = 1; maxDegreeOfParallelism < 4; maxDegreeOfParallelism++)
{
var examplesFound = 0;
using (var vw = new VowpalWabbit(new VowpalWabbitSettings {
EnableStringExampleGeneration = true, EnableThreadSafeExamplePooling = true
}))
using (var resolver = new VowpalWabbitJsonReferenceResolver(serializer =>
{
using (var example = serializer.CreateExamples())
{
ValidateExample(example, (Context)serializer.UserContext);
}
serializer.Dispose();
Interlocked.Increment(ref examplesFound);
}))
{
Parallel.ForEach(
Partitioner.Create(0, examples.Count),
new ParallelOptions {
MaxDegreeOfParallelism = maxDegreeOfParallelism
},
range =>
{
for (int j = range.Item1; j < range.Item2; j++)
{
var ctx = examples[j];
var serializer = new VowpalWabbitJsonSerializer(vw, resolver)
{
UserContext = ctx
};
var example = serializer.ParseAndCreate(ctx.JSON);
// example not ready yet
if (example == null)
{
continue;
}
ValidateExample(example, ctx);
example.Dispose();
serializer.Dispose();
Interlocked.Increment(ref examplesFound);
}
});
}
Assert.AreEqual(examples.Count, examplesFound);
}
}
}
private IEnumerable <PipelineData> Stage1_Deserialize(PipelineData data)
{
try
{
using (var jsonReader = new JsonTextReader(new StringReader(data.JSON)))
{
//jsonReader.FloatParser = Util.ReadDoubleString;
// jsonReader.ArrayPool = pool;
VowpalWabbitJsonSerializer vwJsonSerializer = null;
try
{
vwJsonSerializer = new VowpalWabbitJsonSerializer(this.trainer.VowpalWabbit, this.trainer.ReferenceResolver);
vwJsonSerializer.RegisterExtension((state, property) =>
{
if (TryExtractProperty(state, property, "_eventid", JsonToken.String, reader => data.EventId = (string)reader.Value))
{
return(true);
}
else if (TryExtractProperty(state, property, "_timestamp", JsonToken.Date, reader => data.Timestamp = (DateTime)reader.Value))
{
return(true);
}
else if (TryExtractProperty(state, property, "_ProbabilityOfDrop", JsonToken.Float, reader => data.ProbabilityOfDrop = (float)(reader.Value ?? 0f)))
{
return(true);
}
else if (TryExtractArrayProperty <float>(state, property, "_p", arr => data.Probabilities = arr))
{
return(true);
}
else if (TryExtractArrayProperty <int>(state, property, "_a", arr => data.Actions = arr))
{
return(true);
}
return(false);
});
data.Example = vwJsonSerializer.ParseAndCreate(jsonReader);
if (data.Probabilities == null)
{
throw new ArgumentNullException("Missing probabilities (_p)");
}
if (data.Actions == null)
{
throw new ArgumentNullException("Missing actions (_a)");
}
if (data.Example == null)
{
// unable to create example due to missing data
// will be trigger later
vwJsonSerializer.UserContext = data.Example;
// make sure the serialize is not deallocated
vwJsonSerializer = null;
}
}
finally
{
if (vwJsonSerializer != null)
{
vwJsonSerializer.Dispose();
}
}
performanceCounters.Stage1_JSON_DeserializePerSec.Increment();
// delayed
if (data.Example == null)
{
this.performanceCounters.Feature_Requests_Pending.Increment();
yield break;
}
}
}
catch (Exception ex)
{
this.telemetry.TrackException(ex, new Dictionary <string, string> {
{ "JSON", data.JSON }
});
this.performanceCounters.Stage2_Faulty_Examples_Total.Increment();
this.performanceCounters.Stage2_Faulty_ExamplesPerSec.Increment();
yield break;
}
yield return(data);
}
/// <summary>
/// Returns the number of actions defined by this context.
/// </summary>
public int GetNumberOfActions(string context)
{
return(VowpalWabbitJsonSerializer.GetNumberOfActionDependentExamples(context));
}
