Merge pull request #1375 from reyoung/feature/EvaluatorValue

Feature/evaluator value

paddle/gserver/evaluators/CTCErrorEvaluator.cpp

@@ -20,7 +20,7 @@ namespace paddle {
 /**
  * calculate sequence-to-sequence edit distance
  */
-class CTCErrorEvaluator : public Evaluator {
+class CTCErrorEvaluator : public NotGetableEvaluator {
 private:
   MatrixPtr outActivations_;
   int numTimes_, numClasses_, numSequences_, blank_;

paddle/gserver/evaluators/Evaluator.h

@@ -19,6 +19,7 @@ limitations under the License. */
 #include "paddle/parameter/Argument.h"
 #include "paddle/pserver/ParameterClient2.h"
 #include "paddle/utils/ClassRegistrar.h"
+#include "paddle/utils/Error.h"
 
 namespace paddle {
 
@@ -117,12 +118,105 @@ public:
 
   static ClassRegistrar<Evaluator> registrar_;
 
+  /**
+   * @brief getNames will return all field names of current evaluator.
+   *
+   * The format of name is `evaluator_name.evaluator_fields`. If the evaluator
+   * has multiple field, the name could be `evaluator_name.field1`. For example
+   * the PrecisionRecallEvaluator contains `precision`, `recall` fields. The get
+   * names will return `precision_recall_evaluator.precision`,
+   * `precision_recall_evaluator.recal`, etc.
+   *
+   * Also, if current Evaluator is a combined evaluator. getNames will return
+   * all names of all evaluators inside the combined evaluator.
+   *
+   * @param names [out]: the field names of current evaluator.
+   * @note Never clear the names parameter inside getNames.
+   */
+  virtual void getNames(std::vector<std::string>* names) {
+    names->push_back(config_.name());
+  }
+
+  /**
+   * @brief getValue will return the current evaluate value of one field.
+   *
+   * @param name: The field name of current evaluator.
+   * @param err [out]: The error state.
+   *
+   * @return The evaluate value(metric).
+   */
+  virtual real getValue(const std::string& name, Error* err) const {
+    if (name != config_.name()) {
+      *err = Error("no such name of evaluator %s", name.c_str());
+      return .0f;
+    }
+    return this->getValueImpl();
+  }
+
+  /**
+   * @brief getType will return the evaluator type by field name.
+   *
+   * Evaluate Type is the current type of evaluator in string. Such as 'auc',
+   * 'precision_recall'. In combined evaluator, different name may get different
+   * evaluate type because it could be evaluated by different evaluator inside.
+   *
+   * @param name: The field name of current Evaluator.
+   * @param err: The error state. nullptr means don't care.
+   * @return the evaluator type string.
+   */
+  virtual std::string getType(const std::string& name, Error* err) const {
+    if (name != config_.name()) {
+      *err = Error("no such name of evaluator %s", name.c_str());
+      return std::string();
+    }
+    return this->getTypeImpl();
+  }
+
+protected:
+  /**
+   * @brief getValueImpl The simplest way to define getValue result. If this
+   * evaluator doesn't contain multiple fields, and do not throw any error, just
+   * implemented this method to get the evaluate result(metric).
+   * @return Evaluate result(metric).
+   */
+  virtual real getValueImpl() const {
+    return numSamples_ != .0 ? totalScore_ / numSamples_ : .0;
+  }
+
+  /**
+   * @brief getTypeImpl The simplest way to define getType result. If this
+   * evaluator doesn't combine many evaluators, the get type should only return
+   * itself type.
+   * @return Evaluator type.
+   */
+  virtual std::string getTypeImpl() const { return "base"; }
+
 protected:
   EvaluatorConfig config_;
   double numSamples_;
   double totalScore_;
 };
 
+/**
+ * @brief The NotGetableEvaluator class is the base class of evaluator that
+ * cannot get value in runtime. The most NotGetableEvaluator is Printer
+ * Evaluator, which is only used to debug network configuration.
+ */
+class NotGetableEvaluator : public Evaluator {
+  // Evaluator interface
+public:
+  void getNames(std::vector<std::string>* names) {}
+
+  real getValue(const std::string& name, Error* err) const {
+    *err = Error("Not implemented");
+    return .0f;
+  }
+  std::string getType(const std::string& name, Error* err) const {
+    *err = Error("Not implemented");
+    return "";
+  }
+};
+
 class DummyEvaluator : public Evaluator {
 public:
   DummyEvaluator() {}
@@ -135,6 +229,10 @@ public:
   }
   virtual void finish() {}
   virtual void printStats(std::ostream&) const {}
+
+  // Evaluator interface
+protected:
+  std::string getTypeImpl() const;
 };
 /**
  * @brief evaluate AUC using colIdx-th column as prediction.
@@ -191,6 +289,11 @@ private:
   }
 
   double calcAuc() const;
+
+  // Evaluator interface
+protected:
+  real getValueImpl() const;
+  std::string getTypeImpl() const;
 };
 
 /**
@@ -223,6 +326,10 @@ private:
                      real* clickData,
                      real* pvData,
                      size_t size);
+
+  // Evaluator interface
+protected:
+  std::string getTypeImpl() const;
 };
 /**
  * @brief precision, recall and f1 score Evaluator
@@ -272,6 +379,20 @@ private:
   IVectorPtr cpuLabel_;
   MatrixPtr cpuWeight_;
 
+  struct PrintStatsInfo {
+    double precision;
+    double recall;
+    double f1;
+    double macroAvgPrecision;
+    double macroAvgRecall;
+    double macroAvgF1Score;
+    double microAvgPrecision;
+    double microAvgRecall;
+    double microAvgF1Score;
+  };
+
+  bool getStatsInfo(PrintStatsInfo* info) const;
+
   void calcStatsInfo(const MatrixPtr& output,
                      const IVectorPtr& label,
                      const MatrixPtr& weight);
@@ -303,6 +424,15 @@ private:
       return 0;
     }
   }
+
+  mutable std::unordered_map<std::string, real> values_;
+
+  void storeLocalValues() const;
+  // Evaluator interface
+public:
+  void getNames(std::vector<std::string>* names);
+  real getValue(const std::string& name, Error* err) const;
+  std::string getType(const std::string& name, Error* err) const;
 };
 
 /*
@@ -349,8 +479,7 @@ public:
   virtual void finish() { calc(predictArray_); }
 
   virtual void printStats(std::ostream& os) const {
-    os << " pos/neg"
-       << "=" << pairArray_[0] / ((pairArray_[1] <= 0) ? 1.0 : pairArray_[1]);
+    os << " pos/neg=" << this->getValueImpl();
   }
 
   virtual void distributeEval(ParameterClient2* client) {
@@ -366,6 +495,13 @@ private:
   IVectorPtr cpuLabel_;
   IVectorPtr cpuInfo_;
   MatrixPtr cpuWeight_;
+
+  // Evaluator interface
+protected:
+  real getValueImpl() const {
+    return pairArray_[0] / ((pairArray_[1] <= 0) ? 1.0 : pairArray_[1]);
+  }
+  std::string getTypeImpl() const;
 };
 
 }  // namespace paddle

paddle/gserver/gradientmachines/NeuralNetwork.cpp

@@ -306,7 +306,6 @@ void NeuralNetwork::onPassEnd() {
 
 class CombinedEvaluator : public Evaluator {
 public:
-  CombinedEvaluator() {}
   void addEvaluator(std::unique_ptr<Evaluator>&& evaluator) {
     evaluators_.emplace_back(std::move(evaluator));
   }
@@ -346,6 +345,55 @@ public:
 
 protected:
   std::vector<std::unique_ptr<Evaluator>> evaluators_;
+
+  // Evaluator interface
+public:
+  /**
+   * @brief getNames will return all inside evaluators' names.
+   * @param names [out]: return names.
+   */
+  void getNames(std::vector<std::string>* names) {
+    for (auto& eval : evaluators_) {
+      eval->getNames(names);
+    }
+  }
+
+  /**
+   * @brief getValue could get all inside evaluators' value.
+   */
+  real getValue(const std::string& name, Error* err) const {
+    return this->getMethodHelper<real>(
+        name, err, [&name, err](const std::unique_ptr<Evaluator>& eval) {
+          return eval->getValue(name, err);
+        });
+  }
+
+  /**
+   * @brief getType could get all inside evaluators' type.
+   */
+  std::string getType(const std::string& name, Error* err) const {
+    return this->getMethodHelper<std::string>(
+        name, err, [&name, err](const std::unique_ptr<Evaluator>& eval) {
+          return eval->getType(name, err);
+        });
+  }
+
+private:
+  template <typename T>
+  T getMethodHelper(const std::string& name,
+                    Error* err,
+                    const std::function<T(const std::unique_ptr<Evaluator>&)>&
+                        callback) const {
+    for (auto& eval : evaluators_) {
+      std::vector<std::string> names;
+      eval->getNames(&names);
+      if (std::find(names.begin(), names.end(), name) != names.end()) {
+        return callback(eval);
+      }
+    }
+    *err = Error("No such key %s", name.c_str());
+    return T();
+  }
 };
 
 Evaluator* NeuralNetwork::makeEvaluator() const {

paddle/gserver/tests/test_Evaluator.cpp

@@ -110,6 +110,18 @@ void testEvaluator(TestConfig testConf,
   testEvaluator->finish();
   LOG(INFO) << *testEvaluator;
 
+  std::vector<std::string> names;
+  testEvaluator->getNames(&names);
+  paddle::Error err;
+  for (auto& name : names) {
+    auto value = testEvaluator->getValue(name, &err);
+    ASSERT_TRUE(err.isOK());
+    LOG(INFO) << name << " " << value;
+    auto tp = testEvaluator->getType(name, &err);
+    ASSERT_TRUE(err.isOK());
+    ASSERT_EQ(testConf.evaluatorConfig.type(), tp);
+  }
+
   double totalScore2 = 0.0;
   if (testConf.testAccumulate) {
     testEvaluator->start();

paddle/utils/Error.h

@@ -37,10 +37,10 @@ namespace paddle {
  *
  * Error __must_check bar() {
  *   // do something.
- *   Status s = foo();  // invoke other method return status.
- *   if (!s) return s;
+ *   Error err = foo();  // invoke other method return status.
+ *   if (err) return err;
  *   // do something else.
- *   return Status();
+ *   return Error();
  * }
  * @endcode{cpp}
  *
@@ -53,8 +53,8 @@ namespace paddle {
  *
  * int foo(Error* error) {
  *   // Do something.
- *   Error s = bar();
- *   if (!s) {
+ *   Error err = bar();
+ *   if (err) {
  *     *error = s;
  *     return 0;
  *   }
@@ -68,10 +68,10 @@ namespace paddle {
  * }
  *
  * Error foobar() {
- *   Error s;
+ *   Error err;
  *   // do something.
- *   foo(&s);
- *   if (!s) return s;
+ *   foo(&err);
+ *   if (err) return err;
  * }
  * @endcode{cpp}
  *
@@ -112,16 +112,22 @@ public:
   }
 
   /**
-   * @brief operator bool, return True if there is no error.
+   * @brief operator bool, return True if there is something error.
    */
-  operator bool() const { return msg_ == nullptr; }
+  operator bool() const { return !this->isOK(); }
+
+  /**
+   * @brief isOK return True if there is no error.
+   * @return True if no error.
+   */
+  bool isOK() const { return msg_ == nullptr; }
 
   /**
    * @brief check this status by glog.
    * @note It is a temp method used during cleaning Paddle code. It will be
    *       removed later.
    */
-  void check() const { CHECK(*this) << msg(); }
+  void check() const { CHECK(this->isOK()) << msg(); }
 
 private:
   std::shared_ptr<std::string> msg_;

paddle/utils/tests/test_Error.cpp

@@ -18,17 +18,17 @@ limitations under the License. */
 
 TEST(Error, testAll) {
   paddle::Error error;
-  ASSERT_TRUE(error);
-  error = paddle::Error("I'm the error");
   ASSERT_FALSE(error);
+  error = paddle::Error("I'm the error");
+  ASSERT_TRUE(error);
   ASSERT_STREQ("I'm the error", error.msg());
 
   error = paddle::Error("error2");
-  ASSERT_FALSE(error);
+  ASSERT_TRUE(error);
   ASSERT_STREQ("error2", error.msg());
 
   int i = 3;
   auto error3 = paddle::Error("error%d", i);
-  ASSERT_FALSE(error3);
+  ASSERT_TRUE(error3);
   ASSERT_STREQ("error3", error3.msg());
 }