add MLProject, update Dockerfile, dllib-sacred.py

Biblioteka_DL/dllib-sacred.py

@@ -2,6 +2,7 @@ import numpy as np
 import sys
 import os
 import torch
+import mlflow
 import pandas as pd
 from torch import nn
 from torch.autograd import Variable
@@ -16,13 +17,15 @@ from sacred.observers import MongoObserver
 
 # EPOCHS = int(sys.argv[1])
 
-ex = Experiment()
+#ex = Experiment()
-ex.observers.append(FileStorageObserver('my_res'))
+#ex.observers.append(FileStorageObserver('my_res'))
-ex.observers.append(MongoObserver(url='mongodb://admin:IUM_2021@172.17.0.1:27017', db_name='sacred'))
+#ex.observers.append(MongoObserver(url='mongodb://admin:IUM_2021@172.17.0.1:27017', db_name='sacred'))
 
-@ex.config
+mlflow.set_experiment("s444356")
-def my_config():
+
-    epochs = 100
+#@ex.config
+#def my_config():
+#    epochs = 100
 
 class Model(nn.Module):
     def __init__(self, input_dim):
@@ -242,8 +245,11 @@ def remove_list(games):
 # features_g = pd.DataFrame(features_g, dtype=np.float64)
 # features_g = features_g.to_numpy()
 
-@ex.automain
+epochs = int(sys.argv[1]) if len(sys.argv) > 20 else 20
-def my_main(epochs, _run):
+
+#@ex.automain
+#def my_main(epochs, _run):
+def my_main(epochs)
     platform = pd.read_csv('all_games.train.csv', sep=',', usecols=[1], header=None).values.tolist()
     release_date = pd.read_csv('all_games.train.csv', sep=',', usecols=[2], header=None).values.tolist()
     meta_score = pd.read_csv('all_games.train.csv', sep=',', usecols=[4], header=None).values.tolist()
@@ -295,7 +301,8 @@ def my_main(epochs, _run):
     loss_fn = nn.CrossEntropyLoss()
     # epochs = 1000
     # epochs = epochs
-    _run.info['epochs'] = epochs
+    #_run.info['epochs'] = epochs
+    mlflow.log_param("epochs", epochs)
 
     def print_(loss):
         print ("The loss calculated: ", loss)
@@ -322,14 +329,15 @@ def my_main(epochs, _run):
     pred = pred.detach().numpy()
 
     print("The accuracy is", accuracy_score(labels_test_g, np.argmax(pred, axis=1)))
-    _run.info['accuracy'] = accuracy_score(labels_test_g, np.argmax(pred, axis=1))
+    #_run.info['accuracy'] = accuracy_score(labels_test_g, np.argmax(pred, axis=1))
     _run.log_scalar("measure.accuracy", accuracy_score(labels_test_g, np.argmax(pred, axis=1)))
+    mlflow.log_metric("measure.accuracy", accuracy_score(labels_test_g, np.argmax(pred, axis=1)))
     pred = pd.DataFrame(pred)
 
     pred.to_csv('result.csv')
 
     # save model
     torch.save(model, "games_model.pkl")
-    ex.add_artifact("games_model.pkl")
+    #ex.add_artifact("games_model.pkl")
 
 

Dockerfile

@@ -15,6 +15,7 @@ RUN pip3 install torch
 RUN pip3 install sacred
 RUN pip3 install GitPython
 RUN pip3 install pymongo
+RUN pip3 install mlflow
 #RUN pip3 install --ignore-installed --upgrade https://storage.googleapis.com/tensorflow/mac/cpu/#tensorflow-0.12.0-py3-none-any.whl
 #RUN pip3 install keras
 RUN apt-get install unzip

MLProject

@@ -0,0 +1,12 @@
+name: ium_s444356
+
+docker_env:
+  image: maciejczajka
+
+entry_points:
+  main:
+    parameters:
+      epochs: {type: float, default: 100}
+    command: "python Biblioteka_DL/dllib-sacred.py -e {epochs}"
+  eval:
+    command: "python Biblioteka_DL/evaluate.py"