質問編集履歴
2
タイトルが正しくなかった
title
CHANGED
|
@@ -1,1 +1,1 @@
|
|
|
1
|
-
Pythonの
|
|
1
|
+
Pythonのself.attr(attr)はどの様な意味ですか？(引数の後ろ()の中に，引数がある)
|
body
CHANGED
|
File without changes
|
1
間違って理解していたところを補充
title
CHANGED
|
File without changes
|
body
CHANGED
|
@@ -38,10 +38,71 @@
|
|
|
38
38
|
```
|
|
39
39
|
0.
|
|
40
40
|
**lr = float(K.get_value(self.model.optimizer.lr))**コードでは，
|
|
41
|
-
selfを用い，「他クラスのインスタンスの属性を得ている」様に見えますが，この様な理解で宜しいですか？
|
|
41
|
+
~~selfを用い，「他クラスのインスタンスの属性を得ている」様に見えますが，この様な理解で宜しいですか？~~
|
|
42
|
+
スーパークラスのinit()にself.model = Noneコードがありました．
|
|
43
|
+
結論として，『selfを用い，「他クラスのインスタンスの属性を得ている」』ではないです．
|
|
44
|
+
以下スーパークラスのCallback(object)となります．
|
|
42
45
|
|
|
43
46
|
0.
|
|
44
47
|
**lr = self.schedule(epoch, lr)**コードでは，
|
|
45
48
|
selfを用い，得た属性の後ろに又属性を加えているのですが(関数の使用に似ているのですが),この様な使い方の意味は，どの様なものでしょうか？
|
|
46
49
|
|
|
47
|
-
宜しくお願い致します．
|
|
50
|
+
宜しくお願い致します．
|
|
51
|
+
```Python
|
|
52
|
+
class Callback(object):
|
|
53
|
+
"""Abstract base class used to build new callbacks.
|
|
54
|
+
|
|
55
|
+
# Properties
|
|
56
|
+
params: dict. Training parameters
|
|
57
|
+
(eg. verbosity, batch size, number of epochs...).
|
|
58
|
+
model: instance of `keras.models.Model`.
|
|
59
|
+
Reference of the model being trained.
|
|
60
|
+
|
|
61
|
+
The `logs` dictionary that callback methods
|
|
62
|
+
take as argument will contain keys for quantities relevant to
|
|
63
|
+
the current batch or epoch.
|
|
64
|
+
|
|
65
|
+
Currently, the `.fit()` method of the `Sequential` model class
|
|
66
|
+
will include the following quantities in the `logs` that
|
|
67
|
+
it passes to its callbacks:
|
|
68
|
+
|
|
69
|
+
on_epoch_end: logs include `acc` and `loss`, and
|
|
70
|
+
optionally include `val_loss`
|
|
71
|
+
(if validation is enabled in `fit`), and `val_acc`
|
|
72
|
+
(if validation and accuracy monitoring are enabled).
|
|
73
|
+
on_batch_begin: logs include `size`,
|
|
74
|
+
the number of samples in the current batch.
|
|
75
|
+
on_batch_end: logs include `loss`, and optionally `acc`
|
|
76
|
+
(if accuracy monitoring is enabled).
|
|
77
|
+
"""
|
|
78
|
+
|
|
79
|
+
def __init__(self):
|
|
80
|
+
self.validation_data = None
|
|
81
|
+
self.model = None
|
|
82
|
+
|
|
83
|
+
def set_params(self, params):
|
|
84
|
+
self.params = params
|
|
85
|
+
|
|
86
|
+
def set_model(self, model):
|
|
87
|
+
self.model = model
|
|
88
|
+
|
|
89
|
+
def on_epoch_begin(self, epoch, logs=None):
|
|
90
|
+
pass
|
|
91
|
+
|
|
92
|
+
def on_epoch_end(self, epoch, logs=None):
|
|
93
|
+
pass
|
|
94
|
+
|
|
95
|
+
def on_batch_begin(self, batch, logs=None):
|
|
96
|
+
pass
|
|
97
|
+
|
|
98
|
+
def on_batch_end(self, batch, logs=None):
|
|
99
|
+
pass
|
|
100
|
+
|
|
101
|
+
def on_train_begin(self, logs=None):
|
|
102
|
+
pass
|
|
103
|
+
|
|
104
|
+
def on_train_end(self, logs=None):
|
|
105
|
+
pass
|
|
106
|
+
|
|
107
|
+
|
|
108
|
+
```
|