From 0beb613a077d27f2d8037d7d832a578e425efe39 Mon Sep 17 00:00:00 2001
From: zhengya01 <43601548+zhengya01@users.noreply.github.com>
Date: Tue, 14 May 2019 14:41:07 +0800
Subject: [PATCH] add ce for deep_attention_matching (#2226)

---
 .../deep_attention_matching/.run_ce.sh        | 25 ++++++++
 .../deep_attention_matching/__init__.py       |  0
 .../deep_attention_matching/_ce.py            | 60 +++++++++++++++++++
 .../deep_attention_matching/main.py           | 13 +++-
 4 files changed, 96 insertions(+), 2 deletions(-)
 create mode 100755 PaddleNLP/dialogue_model_toolkit/deep_attention_matching/.run_ce.sh
 create mode 100644 PaddleNLP/dialogue_model_toolkit/deep_attention_matching/__init__.py
 create mode 100644 PaddleNLP/dialogue_model_toolkit/deep_attention_matching/_ce.py

diff --git a/PaddleNLP/dialogue_model_toolkit/deep_attention_matching/.run_ce.sh b/PaddleNLP/dialogue_model_toolkit/deep_attention_matching/.run_ce.sh
new file mode 100755
index 00000000..601eaaab
--- /dev/null
+++ b/PaddleNLP/dialogue_model_toolkit/deep_attention_matching/.run_ce.sh
@@ -0,0 +1,25 @@
+export CE_MODE_X=ce
+
+export CUDA_VISIBLE_DEVICES=0
+
+python -u main.py \
+  --do_train True \
+    --use_cuda \
+      --data_path ./data/ubuntu/data_small.pkl \
+        --save_path ./model_files/ubuntu \
+          --use_pyreader \
+            --vocab_size 434512 \
+              --_EOS_ 28270 \
+                --batch_size 32 | python _ce.py
+
+export CUDA_VISIBLE_DEVICES=0,1,2,3
+
+python -u main.py \
+  --do_train True \
+    --use_cuda \
+      --data_path ./data/ubuntu/data_small.pkl \
+        --save_path ./model_files/ubuntu \
+          --use_pyreader \
+            --vocab_size 434512 \
+              --_EOS_ 28270 \
+                --batch_size 32 | python _ce.py
diff --git a/PaddleNLP/dialogue_model_toolkit/deep_attention_matching/__init__.py b/PaddleNLP/dialogue_model_toolkit/deep_attention_matching/__init__.py
new file mode 100644
index 00000000..e69de29b
diff --git a/PaddleNLP/dialogue_model_toolkit/deep_attention_matching/_ce.py b/PaddleNLP/dialogue_model_toolkit/deep_attention_matching/_ce.py
new file mode 100644
index 00000000..28c3ea32
--- /dev/null
+++ b/PaddleNLP/dialogue_model_toolkit/deep_attention_matching/_ce.py
@@ -0,0 +1,60 @@
+# this file is only used for continuous evaluation test!
+
+import os
+import sys
+sys.path.append(os.environ['ceroot'])
+from kpi import CostKpi
+from kpi import DurationKpi
+
+train_cost_card1 = CostKpi('train_cost_card1', 0.02, 0, actived=True)
+train_cost_card4 = CostKpi('train_cost_card4', 0.06, 0, actived=True)
+train_duration_card1 = DurationKpi('train_duration_card1', 0.01, 0, actived=True)
+train_duration_card4 = DurationKpi('train_duration_card4', 0.01, 0, actived=True)
+
+tracking_kpis = [
+        train_cost_card1,
+        train_cost_card4,
+        train_duration_card1,
+        train_duration_card4,
+]
+
+
+def parse_log(log):
+    '''
+    This method should be implemented by model developers.
+
+    The suggestion:
+
+    each line in the log should be key, value, for example:
+
+    "
+    train_cost\t1.0
+    test_cost\t1.0
+    train_cost\t1.0
+    train_cost\t1.0
+    train_acc\t1.2
+    "
+    '''
+    for line in log.split('\n'):
+        fs = line.strip().split('\t')
+        print(fs)
+        if len(fs) == 3 and fs[0] == 'kpis':
+            kpi_name = fs[1]
+            kpi_value = float(fs[2])
+            yield kpi_name, kpi_value
+
+
+def log_to_ce(log):
+    kpi_tracker = {}
+    for kpi in tracking_kpis:
+        kpi_tracker[kpi.name] = kpi
+
+    for (kpi_name, kpi_value) in parse_log(log):
+        print(kpi_name, kpi_value)
+        kpi_tracker[kpi_name].add_record(kpi_value)
+        kpi_tracker[kpi_name].persist()
+
+
+if __name__ == '__main__':
+    log = sys.stdin.read()
+    log_to_ce(log)
diff --git a/PaddleNLP/dialogue_model_toolkit/deep_attention_matching/main.py b/PaddleNLP/dialogue_model_toolkit/deep_attention_matching/main.py
index b7f05e54..fdcf8278 100755
--- a/PaddleNLP/dialogue_model_toolkit/deep_attention_matching/main.py
+++ b/PaddleNLP/dialogue_model_toolkit/deep_attention_matching/main.py
@@ -336,8 +336,9 @@ def train(args):
               .format(epoch, "%2.2f sec" % pass_time_cost))
     # For internal continuous evaluation
     if "CE_MODE_X" in os.environ:
-        print("kpis	train_cost	%f" % last_cost)
-        print("kpis	train_duration	%f" % train_time)
+        card_num = get_cards()
+        print("kpis\ttrain_cost_card%d\t%f" % (card_num, last_cost))
+        print("kpis\ttrain_duration_card%d\t%f" % (card_num, train_time))
 
 
 def test(args):
@@ -455,6 +456,14 @@ def test(args):
     print(time.strftime('%Y-%m-%d %H:%M:%S', time.localtime(time.time())))
 
 
+def get_cards():
+    num = 0
+    cards = os.environ.get('CUDA_VISIBLE_DEVICES', '')
+    if cards != '':
+        num = len(cards.split(","))
+    return num
+
+
 if __name__ == '__main__':
     args = config.parse_args()
     config.print_arguments(args)
-- 
GitLab