Merge pull request #165 from frankwhzhang/log_print

fix dataloader log print

core/trainers/framework/runner.py

@@ -18,11 +18,15 @@ import os
 import time
 import warnings
 import numpy as np
+import logging
 import paddle.fluid as fluid
 
 from paddlerec.core.utils import envs
 from paddlerec.core.metric import Metric
 
+logging.basicConfig(
+    format='%(asctime)s - %(levelname)s: %(message)s', level=logging.INFO)
+
 __all__ = [
     "RunnerBase", "SingleRunner", "PSRunner", "CollectiveRunner", "PslibRunner"
 ]
@@ -140,8 +144,16 @@ class RunnerBase(object):
 
         metrics_varnames = []
         metrics_format = []
+
+        if context["is_infer"]:
+            metrics_format.append("\t[Infer]\t{}: {{}}".format("batch"))
+        else:
+            metrics_format.append("\t[Train]\t{}: {{}}".format("batch"))
+
+        metrics_format.append("{}: {{:.2f}}s".format("time_each_interval"))
+
         metrics_names = ["total_batch"]
-        metrics_format.append("{}: {{}}".format("batch"))
+
         for name, var in metrics.items():
             metrics_names.append(name)
             metrics_varnames.append(var.name)
@@ -151,6 +163,7 @@ class RunnerBase(object):
         reader = context["model"][model_dict["name"]]["model"]._data_loader
         reader.start()
         batch_id = 0
+        begin_time = time.time()
         scope = context["model"][model_name]["scope"]
         result = None
         with fluid.scope_guard(scope):
@@ -160,8 +173,8 @@ class RunnerBase(object):
                         program=program,
                         fetch_list=metrics_varnames,
                         return_numpy=False)
-                    metrics = [batch_id]
 
+                    metrics = [batch_id]
                     metrics_rets = [
                         as_numpy(metrics_tensor)
                         for metrics_tensor in metrics_tensors
@@ -169,7 +182,13 @@ class RunnerBase(object):
                     metrics.extend(metrics_rets)
 
                     if batch_id % fetch_period == 0 and batch_id != 0:
-                        print(metrics_format.format(*metrics))
+                        end_time = time.time()
+                        seconds = end_time - begin_time
+                        metrics_logging = metrics[:]
+                        metrics_logging = metrics.insert(1, seconds)
+                        begin_time = end_time
+
+                        logging.info(metrics_format.format(*metrics))
                     batch_id += 1
             except fluid.core.EOFException:
                 reader.reset()

models/multitask/mmoe/config.yaml

@@ -17,12 +17,12 @@ workspace: "models/multitask/mmoe"
 dataset:
 - name: dataset_train
   batch_size: 5
-  type: QueueDataset
+  type: DataLoader # or QueueDataset
   data_path: "{workspace}/data/train"
   data_converter: "{workspace}/census_reader.py"
 - name: dataset_infer
   batch_size: 5
-  type: QueueDataset
+  type: DataLoader # or QueueDataset
   data_path: "{workspace}/data/train"
   data_converter: "{workspace}/census_reader.py"
 
@@ -37,7 +37,6 @@ hyper_parameters:
     learning_rate: 0.001
     strategy: async
 
-#use infer_runner mode and modify 'phase' below if infer
 mode: [train_runner, infer_runner]
 
 runner:
@@ -49,10 +48,10 @@ runner:
   save_inference_interval: 4
   save_checkpoint_path: "increment"
   save_inference_path: "inference"
-  print_interval: 10
+  print_interval: 1
 - name: infer_runner
   class: infer
-  init_model_path: "increment/0"
+  init_model_path: "increment/1"
   device: cpu
 
 phase:

models/rank/fibinet/config.yaml

@@ -102,9 +102,9 @@ phase:
 - name: phase1
   model: "{workspace}/model.py" # user-defined model
   dataset_name: dataloader_train # select dataset by name
-  thread_num: 8
+  thread_num: 1
 
 - name: phase2
   model: "{workspace}/model.py" # user-defined model
   dataset_name: dataset_infer # select dataset by name
-  thread_num: 8
+  thread_num: 1