syntax = "proto3";

package tensorflow.serving;

import "model.proto";
import "tensor.proto";

message PredictRequest {

  float satisfaction_level = 1;
  float last_evaluation = 2;
  int32 number_project = 3;
  int32 average_monthly_hours = 4;
  int32 time_spend_company = 5;
  int32 work_accident = 6;
  int32 promotion_last_5years = 7;
  string sales = 8;
  string salary = 9;

};

message PredictResponse {
  // Matching probability of digit 0-9 in range [0.0, 1.0].
  repeated int32 left = 1;
};

service PredictService {
  // Classifies image into digits.
  rpc Predict (PredictRequest) returns (PredictResponse);
};

var PROTO_PATH = __dirname + '/predict.proto';
var grpc = require('grpc');
var attrition = grpc.load(PROTO_PATH).tensorflow.serving;
function main() {
  var client = new attrition.PredictService('localhost:8500', grpc.credentials.createInsecure());

    var req = new attrition.PredictRequest();
    req.satisfaction_level = 0.38;
    req.last_evaluation = 0.53;
    req.number_project = 2;
    req.average_monthly_hours = 157;
    req.time_spend_company = 3;
    req.work_accident = 0;
    req.promotion_last_5years = 0;
    req.sales = 'sales';
    req.salary = 'low';

  function run() {
   client.predict(req, (err, predictResponse) => {
     if (err) {
       console.log(err)
     } else {
       var results = predictResponse ? predictResponse.value : [];
       console.log(results)
     }
   });
  };

  run();
}

main();

var PROTO_PATH = __dirname + '/protos/prediction_service.proto';

var grpc = require('grpc');
var attrition = grpc.load(PROTO_PATH).tensorflow;
function main() {
  var client = new attrition.PredictionService('localhost:8500', grpc.credentials.createInsecure());

  var features = {
    model_spec: {name: 'attrition', signature_name: 'predict'},
    inputs: {
      average_monthly_hours: {
        dtype: "DT_INT32",
        tensor_shape: {
          dim: {
            size: 1
          }
        },
        int_val: 157
      }, last_evaluation: {
        dtype: "DT_FLOAT",
        tensor_shape: {
          dim: {
            size: 1
          }
        },
        float_val: 0.53
      }, number_project: {
        dtype: "DT_INT32",
        tensor_shape: {
          dim: {
            size: 1
          }
        },
        int_val: 2
      }, promotion_last_5years: {
        dtype: "DT_INT32",
        tensor_shape: {
          dim: {
            size: 1
          }
        },
        int_val: 0
      }, salary: {
        dtype: "DT_STRING",
        tensor_shape: {
          dim: {
            size: 1
          }
        },
        string_val: 'low'
      }, sales: {
        dtype: "DT_STRING",
        tensor_shape: {
          dim: {
            size: 1
          }
        },
        string_val: "sales"
      }, satisfaction_level: {
        dtype: "DT_FLOAT",
        tensor_shape: {
          dim: {
            size: 1
          }
        },
        float_val: 0.38
      }, time_spend_company: {
        dtype: "DT_INT32",
        tensor_shape: {
          dim: {
            size: 1
          }
        },
        int_val: 3
      }, work_accident: {
        dtype: "DT_INT32",
        tensor_shape: {
          dim: {
            size: 1
          }
        },
        int_val: 0
      }
    }
  }

  function run() {
   client.predict(features, (err, predictResponse) => {
     if (err) {
       console.log("Received error running prediction:")
       console.log(err)
     } else {
       var results = predictResponse ? predictResponse.value : [];
       console.log(results)
     }
   });
  };

  run();
}

main();

feature_inputs = {
  'satisfaction_level': tf.placeholder(dtype=tf.float32, shape=[1,], name='satisfaction_level'),
  'last_evaluation': tf.placeholder(dtype=tf.float32, shape=[1,], name='last_evaluation'),
  'number_project': tf.placeholder(dtype=tf.int32, shape=[1,], name='number_project'),
  'average_monthly_hours': tf.placeholder(dtype=tf.int32, shape=[1,], name='average_monthly_hours'),
  'time_spend_company': tf.placeholder(dtype=tf.int32, shape=[1,], name='time_spend_company'),
  'work_accident': tf.placeholder(dtype=tf.int32, shape=[1,], name='work_accident'),
  'promotion_last_5years': tf.placeholder(dtype=tf.int32, shape=[1,], name='promotion_last_5years'),
  'sales': tf.placeholder(dtype=tf.string, shape=[1,], name='sales'),
  'salary': tf.placeholder(dtype=tf.string, shape=[1,], name='salary')
}
serving_input_receiver_fn = tf.estimator.export.build_raw_serving_input_receiver_fn(feature_inputs)
model.export_savedmodel("./exported_model",  serving_input_receiver_fn)

saved_model_cli show --dir ../python/notebooks/exported_model/1510845524/ --all
MetaGraphDef with tag-set: 'serve' contains the following SignatureDefs:

signature_def['predict']:
The given SavedModel SignatureDef contains the following input(s):
inputs['average_monthly_hours'] tensor_info:
  dtype: DT_INT32
  shape: (-1, 1)
  name: average_monthly_hours_1:0
inputs['last_evaluation'] tensor_info:
  dtype: DT_FLOAT
  shape: (-1, 1)
  name: last_evaluation_1:0
inputs['number_project'] tensor_info:
  dtype: DT_INT32
  shape: (-1, 1)
  name: number_project_1:0
inputs['promotion_last_5years'] tensor_info:
  dtype: DT_INT32
  shape: (-1, 1)
  name: promotion_last_5years_1:0
inputs['salary'] tensor_info:
  dtype: DT_STRING
  shape: (-1, 1)
  name: salary_1:0
inputs['sales'] tensor_info:
  dtype: DT_STRING
  shape: (-1, 1)
  name: sales_1:0
inputs['satisfaction_level'] tensor_info:
  dtype: DT_FLOAT
  shape: (-1, 1)
  name: satisfaction_level:0
inputs['time_spend_company'] tensor_info:
  dtype: DT_INT32
  shape: (-1, 1)
  name: time_spend_company_1:0
inputs['work_accident'] tensor_info:
  dtype: DT_INT32
  shape: (-1, 1)
  name: work_accident_1:0
The given SavedModel SignatureDef contains the following output(s):
outputs['class_ids'] tensor_info:
  dtype: DT_INT64
  shape: (-1, 1)
  name: dnn/head/predictions/classes:0
outputs['classes'] tensor_info:
  dtype: DT_STRING
  shape: (-1, 1)
  name: dnn/head/predictions/str_classes:0
outputs['logistic'] tensor_info:
  dtype: DT_FLOAT
  shape: (-1, 1)
  name: dnn/head/predictions/logistic:0
outputs['logits'] tensor_info:
  dtype: DT_FLOAT
  shape: (-1, 1)
  name: dnn/head/predictions/logits:0
outputs['probabilities'] tensor_info:
  dtype: DT_FLOAT
  shape: (-1, 2)
  name: dnn/head/predictions/probabilities:0
Method name is: tensorflow/serving/predict

tensorflow_model_server --port=8500 --model_name=attrition --model_base_path=`pwd`/exported_model/

2017-11-20 12:18:20.380088: I external/org_tensorflow/tensorflow/cc/saved_model/loader.cc:155] Restoring SavedModel bundle.
2017-11-20 12:18:20.386764: I external/org_tensorflow/tensorflow/cc/saved_model/loader.cc:190] Running LegacyInitOp on SavedModel bundle.
2017-11-20 12:18:20.393305: I external/org_tensorflow/tensorflow/cc/saved_model/loader.cc:284] Loading SavedModel: success. Took 44036 microseconds.
2017-11-20 12:18:20.393473: I tensorflow_serving/core/loader_harness.cc:86] Successfully loaded servable version {name: attrition version: 1510845524}
2017-11-20 12:18:20.397036: I tensorflow_serving/model_servers/main.cc:288] Running ModelServer at 0.0.0.0:8500 ...