自作の学習済みモデルを用いて転移学習を行いたい

自分で実装したCNNでモデルを保存し、それを使って転移学習を行いたいと考えているのですが、
実装がうまくいかず、自分ひとりでは解決が難しいため、是非お力を貸していただきたいです。

はっきりさせておきたいと考えているのは以下の4つです。
1.(CNNコード)モデルの保存方法が正しいかどうか
2.(転移学習コード)モデルの読み込み方法が正しいかどうか
3.(転移学習コード)読み込んだモデルのtrainable属性はどのように設定するのが一般的か
4.(転移学習コード)事前学習モデルと後続のレイヤーは正しく結合できているか(サイズ等)

以下、CNNのモデル部分と転移学習のコードです。
いずれも、２つの数値を予測する回帰モデルで、入力は画像(+教師データ)です。

python
1#CNN
2
3input = Input(shape=(100,100,3))
4conv_0 = Conv2D(32,kernel_size=3,activation='relu')(input)
5pool_0 = MaxPooling2D(pool_size=(2,2))(conv_0)
6pool_0 = Dropout(0.25)(pool_0)
7conv_1 = Conv2D(64,kernel_size=3,activation='relu')(pool_0)
8pool_1 = MaxPooling2D(pool_size=(2,2))(conv_1)
9pool_1 = Dropout(0.25)(pool_1)
10conv_2 = Conv2D(32,kernel_size=3,activation='relu')(pool_1)
11pool_2 = MaxPooling2D(pool_size=(2,2))(conv_2)
12pool_2 = Dropout(0.25)(pool_2)
13conv_3 = Conv2D(16,kernel_size=3,activation='relu')(pool_2)
14pool_3 = MaxPooling2D(pool_size=(2,2))(conv_3)
15conv_4 = Conv2D(8,kernel_size=3,activation='relu')(pool_3)
16pool_4 = MaxPooling2D(pool_size=(2,2))(conv_4)
17flat = Flatten()(pool_4)
18denseL = Dense(64,activation='relu')(flat)
19denseL = Dropout(0.25)(denseL)
20A_output = Dense(1,name="a")(denseL)
21B_output = Dense(1,name="b")(denseL)
22
23model = Model(inputs=input, outputs=[A_output,B_output])
24model.compile(Adam(learning_rate=0.001),
25              loss = {'a':'mae','b':'mae'} ,
26              metrics =  {'a':'mae','b':'mae'})
27
28history = model.fit([np.array(Img_train)],[np.array(LabelA_train),np.array(LabelB_train)],
29                    epochs=100, batch_size=16,
30                    validation_data=([np.array(Img_test)],[np.array(S_IBP_test),np.array(D_IBP_test)]))
31
32#モデルの保存
33model.save('forTransferL.h5')
34
35"""
36以下、sumally()の結果です
37__________________________________________________________________________________________________
38Layer (type)                    Output Shape         Param #     Connected to
39==================================================================================================
40input_1 (InputLayer)            [(None, 100, 100, 3) 0
41__________________________________________________________________________________________________
42conv2d (Conv2D)                 (None, 98, 98, 32)   896         input_1[0][0]
43__________________________________________________________________________________________________
44max_pooling2d (MaxPooling2D)    (None, 49, 49, 32)   0           conv2d[0][0]
45__________________________________________________________________________________________________
46dropout (Dropout)               (None, 49, 49, 32)   0           max_pooling2d[0][0]
47__________________________________________________________________________________________________
48conv2d_1 (Conv2D)               (None, 47, 47, 64)   18496       dropout[0][0]
49__________________________________________________________________________________________________
50max_pooling2d_1 (MaxPooling2D)  (None, 23, 23, 64)   0           conv2d_1[0][0]
51__________________________________________________________________________________________________
52dropout_1 (Dropout)             (None, 23, 23, 64)   0           max_pooling2d_1[0][0]
53__________________________________________________________________________________________________
54conv2d_2 (Conv2D)               (None, 21, 21, 32)   18464       dropout_1[0][0]
55__________________________________________________________________________________________________
56max_pooling2d_2 (MaxPooling2D)  (None, 10, 10, 32)   0           conv2d_2[0][0]
57__________________________________________________________________________________________________
58dropout_2 (Dropout)             (None, 10, 10, 32)   0           max_pooling2d_2[0][0]
59__________________________________________________________________________________________________
60conv2d_3 (Conv2D)               (None, 8, 8, 16)     4624        dropout_2[0][0]
61__________________________________________________________________________________________________
62max_pooling2d_3 (MaxPooling2D)  (None, 4, 4, 16)     0           conv2d_3[0][0]
63__________________________________________________________________________________________________
64conv2d_4 (Conv2D)               (None, 2, 2, 8)      1160        max_pooling2d_3[0][0]
65__________________________________________________________________________________________________
66max_pooling2d_4 (MaxPooling2D)  (None, 1, 1, 8)      0           conv2d_4[0][0]
67__________________________________________________________________________________________________
68flatten (Flatten)               (None, 8)            0           max_pooling2d_4[0][0]
69__________________________________________________________________________________________________
70dense (Dense)                   (None, 64)           576         flatten[0][0]
71__________________________________________________________________________________________________
72dropout_3 (Dropout)             (None, 64)           0           dense[0][0]
73__________________________________________________________________________________________________
74a (Dense)                       (None, 1)            65          dropout_3[0][0]
75__________________________________________________________________________________________________
76b (Dense)                       (None, 1)            65          dropout_3[0][0]
77========================================================================================="""

python
1#転移学習
2
3model = load_model('forTransferL.h5')
4model.layers[0].trainable = False
5x = model.layers[10].output
6
7#以下はCNNのモデルの一部と同じものです。
8conv_3 = Conv2D(16,kernel_size=3,activation='relu')(x)
9pool_3 = MaxPooling2D(pool_size=(2,2))(conv_3)
10conv_4 = Conv2D(8,kernel_size=3,activation='relu')(pool_3)
11pool_4 = MaxPooling2D(pool_size=(2,2))(conv_4)
12flat = Flatten()(pool_4)
13denseL = Dense(64,activation='relu')(flat)
14denseL = Dropout(0.25)(denseL)
15A_output = Dense(1,name="a")(denseL)
16B_output = Dense(1,name="b")(denseL)
17
18model=Model(inputs=model.input,outputs=[A_output,B_output])

念の為、今出ているエラー文も以下に記載しておきますが、
対処療法的なエラー修正というより、基本的な実装部分を見直す方が優先度は高いかなと考えています。

上記、よろしくお願いいたします。
質問の補足依頼等も歓迎いたします。

#以下エラー文（一部）
During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "c:/Users/userABC/OneDrive/ドキュメント/StudyAI/transferMymodel.py", line 158, in <module>
    pool_m4 = MaxPooling2D(pool_size=(2,2))(conv_m4)
  File "C:\Users\userABC\anaconda3\lib\site-packages\keras\engine\base_layer.py", line 1006, in __call__
    outputs = call_fn(inputs, *args, **kwargs)
  〜文字数の都合で省略〜

ValueError: Negative dimension size caused by subtracting 2 from 1 for '{{node tf.compat.v1.nn.max_pool_1/MaxPool}} = MaxPool[T=DT_FLOAT, data_format="NHWC", explicit_paddings=[], ksize=[1, 2, 2, 1], padding="VALID", strides=[1, 2, 2, 1]](Placeholder)' with input shapes: [?,1,1,8].