Áp dụng Deep Learning để dự báo kết quả tiếp cận thành công một campaign quảng cáo
Bài viết này mình thử dùng thuật toán Artificial Neural Network (ANN) để xử lí một tệp dữ liệu marketing đơn giản, mục đích thử phân loại xem hành vi Yes/No của người dùng trong dataset có những yếu tố nào ảnh hưởng, cũng như thử build một mô hình dự báo bằng Deep Learning. Mô hình cho ra kết quả với độ chính xác accuracy (test data) = 90.8% - Final loss (Test data): 0.2293
Quan sát kết quả EDA label tổng quát trong data này:
- Nhận thấy có tỉ lệ thành công tốt đặc biệt với poutcome = success(yếu tố này cho biết là campaign trước thành công là dấu hiệu cho tỉ lệ thành công lần này cũng cao)
- Người thất nghiệp và đã nghỉ hưu lại có tỉ lệ tiếp cận thành công cao hơn các job khác
- Người không vay nợ lại có tỉ lệ tiếp cận thành công tốt hơn người có vay nợ.
- Người tốt nghiệp đại học có tỉ lệ tiếp cận thành công hơn những người mới có bằng primary hay secondary
- Người độc thân có tỉ lệ tiếp cận thành công hơn người đã có gia đình
- Người không có nhà có tỉ lệ tiếp cận thành công hơn người có nhà
- Tiếp cận bằng thiết bị di động có tỉ lệ tiếp cận thành công hơn người dùng các phương tiện khác
Đây là bộ dữ liệu marketing ngân hàng trong Kho dữ liệu Machine Learning của UCI (https://archive.ics.uci.edu/ml/index.php). Trong dữ liệu có các sub dữ liệu là: nhân khẩu học, thông tin vay nợ và thông tin chăm sóc khách hàng (CRM) theo thời gian. Tổng cộng 17 đặc tính và 45000 kết quả ghi nhận.
Noted: Kiến thức người viết hạn hẹp, cũng như mô hình này mắc một problem là chưa xử lí dữ liệu unbalance tỉ lệ 8:1 nên mình chọn ngưỡng baseline là 88.88%, do đó chỉ mang tính chất tham khảo học thuật là chính. Trong bài mình cũng không reScale lại data trong quá trình tiền xử lí dữ liệu.
kết nối bạn bè và trao đổi học tập:
- Linkedin: http://bit.ly/3aYazxr
- Facebook: http://bit.ly/2u9pvIl
- Githut: http://bit.ly/3b1qBXd
Quan sát qua bảng dữ liệu
Nguồn data chính xác: http://archive.ics.uci.edu/ml/datasets/bank+marketing
Mục tiêu: phân loại Yes/No (feature 'y' trong dataset) những người có thực hiện ý đồ của campaign marketing (ví dụ mua hàng, vay tiền, sử dụng coupon)
Đây là bộ dữ liệu marketing ngân hàng trong Kho dữ liệu Machine Learning của UCI (https://archive.ics.uci.edu/ml/index.php). Bộ dữ liệu cung cấp thông tin về một chiến dịch tiếp thị của một tổ chức tài chính. Nhiệm vụ phân tích trong vấn đề này là để tìm cách tìm kiếm các chiến lược trong tương lai nhằm cải thiện các chiến dịch tiếp thị trong tương lai cho ngân hàng.
Các yếu tố (Feature) trong dữ liệu bao gồm:
- age: tuổi của người tiếp cận
- job: nghề nghiệp bao gồm admin, blue-collar, entrepreneur, housemaid, management, retired, self-employed, services, student, technician, unemployed
- marital: tình trạng hôn nhân bao gồm: married, single, divorced
- education: trình độ học vấn hiện tại bao gồm primary,secondary,tertiary (đại học)
- balance: Số tiền trên tài khoản giao dịch bao gồm tất cả các giao dịch đầy đủ và hoàn chỉnh cũng như các hoạt động phi thương mại như nạp và rút tiền.
- housing: có nhà hay chưa
- loan: có vay hay không
- day: ngày campaign tiếp cận gần nhất (ngày mấy của tháng, số kéo dài từ 1 đến 31 ngày)
- month: tháng campaign tiếp cận gần nhất của năm (phân loại: 'jan', 'feb', ' mar ', ...,' nov ',' dec ')
- contact: loại phương tiện liên lạc (phân loại: 'di động', 'điện thoại')
- duration: thời lượng liên lạc cuối cùng, tính bằng giây.
- campaign: số thứ tự chiến dịch
- pday: số ngày trôi qua sau khi khách hàng được liên hệ lần cuối từ chiến dịch trước đó
- previous: số lượng liên hệ được thực hiện trước chiến dịch này và cho khách hàng này
- poutcome: kết quả của chiến dịch tiếp thị trước đó (phân loại: 'thất bại', 'không tồn tại', 'thành công')
In [2]:
eda = pd.read_csv(r"https://raw.githubusercontent.com/cafechungkhoan/chu_gia/master/bank%20marketing.csv",delimiter=';')
eda = pd.DataFrame(eda)
eda
Out[2]:
| age | job | marital | education | default | balance | housing | loan | contact | day | month | duration | campaign | pdays | previous | poutcome | y | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 0 | 58 | management | married | tertiary | no | 2143 | yes | no | unknown | 5 | may | 261 | 1 | -1 | 0 | unknown | no |
| 1 | 44 | technician | single | secondary | no | 29 | yes | no | unknown | 5 | may | 151 | 1 | -1 | 0 | unknown | no |
| 2 | 33 | entrepreneur | married | secondary | no | 2 | yes | yes | unknown | 5 | may | 76 | 1 | -1 | 0 | unknown | no |
| 3 | 47 | blue-collar | married | unknown | no | 1506 | yes | no | unknown | 5 | may | 92 | 1 | -1 | 0 | unknown | no |
| 4 | 33 | unknown | single | unknown | no | 1 | no | no | unknown | 5 | may | 198 | 1 | -1 | 0 | unknown | no |
| ... | ... | ... | ... | ... | ... | ... | ... | ... | ... | ... | ... | ... | ... | ... | ... | ... | ... |
| 45206 | 51 | technician | married | tertiary | no | 825 | no | no | cellular | 17 | nov | 977 | 3 | -1 | 0 | unknown | yes |
| 45207 | 71 | retired | divorced | primary | no | 1729 | no | no | cellular | 17 | nov | 456 | 2 | -1 | 0 | unknown | yes |
| 45208 | 72 | retired | married | secondary | no | 5715 | no | no | cellular | 17 | nov | 1127 | 5 | 184 | 3 | success | yes |
| 45209 | 57 | blue-collar | married | secondary | no | 668 | no | no | telephone | 17 | nov | 508 | 4 | -1 | 0 | unknown | no |
| 45210 | 37 | entrepreneur | married | secondary | no | 2971 | no | no | cellular | 17 | nov | 361 | 2 | 188 | 11 | other | no |
45211 rows × 17 columns
Xử lí tiền dữ liệu
Mình xử lí bằng cách Drop các feature thiếu trên 30% dữ liệu, còn lại loại bỏ các hàng chứa dữ liệu thiếu dropna()
Mình trực quan dữ liệu thiếu bằng 'Missing Value Heatmaps'. Nếu chưa biết dạng chart bên dưới có thể đọc thêm về (Missing Value Heatmaps). Nó cho biết dữ liệu thiếu nằm tại vị trí nào của dataset https://dev.to/tomoyukiaota/visualizing-the-patterns-of-missing-value-occurrence-with-python-46dj
In [61]:
In [7]:
target = 'y'
Giai đoạn exploratory data analysis (EDA) đầu vào
Đây là loại dữ liệu không cân bằng. Tuy nhiên, dữ liệu không cân bằng trong marketing là chuyện thường thấy, như số lượng click / không click trong một lượng tiếp cận, số lượng churn / not churn,... Problem về dữ liệu không cân bằng mình tìm hiểu nhiều, nhưng vẫn chưa có một phương pháp thỏa đáng xử lí. Hơn nữa cũng chưa thấy có report nào nói về tỉ lệ không cân bằng là bao nhiêu (trường hợp trong bài viết này là 8:1)
EDA sơ bộ những ý tưởng của mình trả lời những câu hỏi sau:
- Có sự khác biệt nào giữa người mua / không mua dựa vào tuổi hay không?
- Có sự khác biệt nào giữa người mua / không mua dựa vào yếu tố nghề nghiệp không?
- Có sự khác biệt nào giữa người mua / không mua dựa vào yếu tố học vấn không?
- Có tính chất gì lạ trong dữ liệu không?
In [36]:
import hieu_viet_code_ne
In [5]:
eda = pd.read_csv(r"https://raw.githubusercontent.com/cafechungkhoan/chu_gia/master/bank%20marketing.csv",delimiter=';')
yes = eda[eda.y=='yes']
no = eda[eda.y=='no']
plt.figure(figsize=(12,8))
sns.set_style("darkgrid")
plt.title("Quan sát phân phối xác suất giữa Yes và No của Feature age, không thấy sự khác biệt đáng kể",{'fontsize': 15})
a = sns.kdeplot(data=yes['age'], label="yes", shade=True)
b = sns.kdeplot(data=no['age'],label="no" ,shade=True)
<IPython.core.display.Javascript object>
<IPython.core.display.Javascript object>
<IPython.core.display.Javascript object>
<IPython.core.display.Javascript object>
<IPython.core.display.Javascript object>
<IPython.core.display.Javascript object>
In [25]:
<IPython.core.display.Javascript object>
<IPython.core.display.Javascript object>
<IPython.core.display.Javascript object>
<IPython.core.display.Javascript object>
<IPython.core.display.Javascript object>
In [24]:
<IPython.core.display.Javascript object>
<IPython.core.display.Javascript object>
<IPython.core.display.Javascript object>
<IPython.core.display.Javascript object>
<IPython.core.display.Javascript object>
In [64]:
Target looks like classification Linear Discriminant Analysis training set score: 0.613
Quan sát kết quả EDA label tổng quát
- Nhận thấy có tỉ lệ thành công tốt đặc biệt với poutcome = success(yếu tố này cho biết là campaign trước thành công là dấu hiệu cho tỉ lệ thành công lần này cũng cao)
- Người thất nghiệp và đã nghỉ hưu lại có tỉ lệ tiếp cận thành công cao hơn các job khác
- Người không vay nợ lại có tỉ lệ tiếp cận thành công tốt hơn người có vay nợ.
- Người tốt nghiệp đại học có tỉ lệ tiếp cận thành công hơn những người mới có bằng primary hay secondary
- Người độc thân có tỉ lệ tiếp cận thành công hơn người đã có gia đình
- Người không có nhà có tỉ lệ tiếp cận thành công hơn người có nhà
- Tiếp cận bằng thiết bị di động có tỉ lệ tiếp cận thành công hơn người dùng các phương tiện khác
In [3]:
data = eda
data = data.drop(['poutcome','contact'] , axis = 1)
data = data.replace('yes',1)
data = data.replace('no',0)
data = pd.get_dummies(data)
data.head()
Out[3]:
| age | default | balance | housing | loan | day | duration | campaign | pdays | previous | ... | month_dec | month_feb | month_jan | month_jul | month_jun | month_mar | month_may | month_nov | month_oct | month_sep | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 0 | 58 | 0 | 2143 | 1 | 0 | 5 | 261 | 1 | -1 | 0 | ... | 0 | 0 | 0 | 0 | 0 | 0 | 1 | 0 | 0 | 0 |
| 1 | 44 | 0 | 29 | 1 | 0 | 5 | 151 | 1 | -1 | 0 | ... | 0 | 0 | 0 | 0 | 0 | 0 | 1 | 0 | 0 | 0 |
| 2 | 33 | 0 | 2 | 1 | 1 | 5 | 76 | 1 | -1 | 0 | ... | 0 | 0 | 0 | 0 | 0 | 0 | 1 | 0 | 0 | 0 |
| 3 | 47 | 0 | 1506 | 1 | 0 | 5 | 92 | 1 | -1 | 0 | ... | 0 | 0 | 0 | 0 | 0 | 0 | 1 | 0 | 0 | 0 |
| 4 | 33 | 0 | 1 | 0 | 0 | 5 | 198 | 1 | -1 | 0 | ... | 0 | 0 | 0 | 0 | 0 | 0 | 1 | 0 | 0 | 0 |
5 rows × 42 columns
Dùng data.shape thấy sau khi Onehot encoder dữ liệu, phát sinh ra đến 42 Feature, do đó cần lọc lại dữ liệu
In [66]:
data.shape
Out[66]:
(45211, 42)
Feature Selection
Có 4 vấn đề cần xử lí trong Feature Selection, trước khi vào giai đoạn Deep Learning: loại bỏ giá trị duy nhất (single unique value), tìm và xử lí các tính năng cộng tuyến (collinear features),tính năng có tầm quan trọng = 0 và tính năng kém quan trọng ( hiện tại bài viết mình xử lí bằng bằng gradient boosting machine)
Đọc thêm về Feture Selection bằng Gradient Boosting Machine https://arxiv.org/abs/1901.04055
In [26]:
In [41]:
Out[41]:
<matplotlib.axes._subplots.AxesSubplot at 0x1b503c6ce88>
In [29]:
0 features with a correlation magnitude greater than 0.80. []
Out[29]:
| drop_feature | corr_feature | corr_value |
|---|
In [30]:
Training Gradient Boosting Model Training until validation scores don't improve for 100 rounds Early stopping, best iteration is: [344] valid_0's auc: 0.935519 valid_0's binary_logloss: 0.194667 Training until validation scores don't improve for 100 rounds Early stopping, best iteration is: [278] valid_0's auc: 0.935839 valid_0's binary_logloss: 0.194577 Training until validation scores don't improve for 100 rounds Early stopping, best iteration is: [311] valid_0's auc: 0.942968 valid_0's binary_logloss: 0.186193 Training until validation scores don't improve for 100 rounds Early stopping, best iteration is: [198] valid_0's auc: 0.934594 valid_0's binary_logloss: 0.196641 Training until validation scores don't improve for 100 rounds Early stopping, best iteration is: [342] valid_0's auc: 0.936459 valid_0's binary_logloss: 0.193615 Training until validation scores don't improve for 100 rounds Early stopping, best iteration is: [471] valid_0's auc: 0.94093 valid_0's binary_logloss: 0.188461 Training until validation scores don't improve for 100 rounds Early stopping, best iteration is: [254] valid_0's auc: 0.93385 valid_0's binary_logloss: 0.198337 Training until validation scores don't improve for 100 rounds Early stopping, best iteration is: [281] valid_0's auc: 0.935248 valid_0's binary_logloss: 0.196619 Training until validation scores don't improve for 100 rounds Early stopping, best iteration is: [284] valid_0's auc: 0.938564 valid_0's binary_logloss: 0.192148 Training until validation scores don't improve for 100 rounds Early stopping, best iteration is: [228] valid_0's auc: 0.939406 valid_0's binary_logloss: 0.190966 0 features with zero importance after one-hot encoding.
In [34]:
44 features required for 0.99 of cumulative importance
In [ ]:
In [ ]:
In [69]:
Model: "sequential_3" _________________________________________________________________ Layer (type) Output Shape Param # ================================================================= dense_9 (Dense) (None, 41) 1722 _________________________________________________________________ dense_10 (Dense) (None, 41) 1722 _________________________________________________________________ dense_11 (Dense) (None, 41) 1722 _________________________________________________________________ dense_12 (Dense) (None, 1) 42 ================================================================= Total params: 5,208 Trainable params: 5,208 Non-trainable params: 0 _________________________________________________________________
In [70]:
a = model.fit(X_train, y_train, batch_size = 16, epochs = 500)
final_loss, final_acc = model.evaluate(X_test, y_test, verbose = 0)
print("Final loss (Test data): {0:.4f}, final accuracy (Test data): {1:.4f}".format(final_loss, final_acc))
Epoch 1/500 31647/31647 [==============================] - 2s 71us/step - loss: 0.3022 - accuracy: 0.8829 Epoch 2/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2784 - accuracy: 0.8886 Epoch 3/500 31647/31647 [==============================] - 2s 71us/step - loss: 0.2727 - accuracy: 0.8888 Epoch 4/500 31647/31647 [==============================] - 2s 71us/step - loss: 0.2670 - accuracy: 0.8890 Epoch 5/500 31647/31647 [==============================] - 2s 71us/step - loss: 0.2628 - accuracy: 0.8897 Epoch 6/500 31647/31647 [==============================] - 2s 71us/step - loss: 0.2565 - accuracy: 0.8908 Epoch 7/500 31647/31647 [==============================] - 2s 71us/step - loss: 0.2523 - accuracy: 0.8906 Epoch 8/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2489 - accuracy: 0.8913 Epoch 9/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2463 - accuracy: 0.8917 Epoch 10/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2437 - accuracy: 0.8922 Epoch 11/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2437 - accuracy: 0.8924 Epoch 12/500 31647/31647 [==============================] - 2s 71us/step - loss: 0.2408 - accuracy: 0.8935 Epoch 13/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2400 - accuracy: 0.8942 Epoch 14/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2394 - accuracy: 0.8947 Epoch 15/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2386 - accuracy: 0.8943 Epoch 16/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2379 - accuracy: 0.8941 Epoch 17/500 31647/31647 [==============================] - 2s 71us/step - loss: 0.2371 - accuracy: 0.8952 Epoch 18/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2363 - accuracy: 0.8941 Epoch 19/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2352 - accuracy: 0.8946 Epoch 20/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2355 - accuracy: 0.8953 Epoch 21/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2344 - accuracy: 0.8949 Epoch 22/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2348 - accuracy: 0.8948 Epoch 23/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2333 - accuracy: 0.8957 Epoch 24/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2337 - accuracy: 0.8950 Epoch 25/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2331 - accuracy: 0.8955 Epoch 26/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2326 - accuracy: 0.8948 Epoch 27/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2325 - accuracy: 0.8957 Epoch 28/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2314 - accuracy: 0.8961 Epoch 29/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2321 - accuracy: 0.8965 Epoch 30/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2314 - accuracy: 0.8961 Epoch 31/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2311 - accuracy: 0.8961 Epoch 32/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2308 - accuracy: 0.8966 Epoch 33/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2304 - accuracy: 0.8963 Epoch 34/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2302 - accuracy: 0.8959 Epoch 35/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2296 - accuracy: 0.8954 Epoch 36/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2299 - accuracy: 0.8962 Epoch 37/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2297 - accuracy: 0.8963 Epoch 38/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2293 - accuracy: 0.8973 Epoch 39/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2291 - accuracy: 0.8965 Epoch 40/500 31647/31647 [==============================] - 2s 74us/step - loss: 0.2290 - accuracy: 0.8963 Epoch 41/500 31647/31647 [==============================] - 2s 75us/step - loss: 0.2288 - accuracy: 0.8965 Epoch 42/500 31647/31647 [==============================] - 2s 72us/step - loss: 0.2284 - accuracy: 0.8965 Epoch 43/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2287 - accuracy: 0.8964 Epoch 44/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2283 - accuracy: 0.8975 Epoch 45/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2279 - accuracy: 0.8968 Epoch 46/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2279 - accuracy: 0.8976 Epoch 47/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2272 - accuracy: 0.8975 Epoch 48/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2272 - accuracy: 0.8968 Epoch 49/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2274 - accuracy: 0.8972 Epoch 50/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2269 - accuracy: 0.8977 Epoch 51/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2271 - accuracy: 0.8974 Epoch 52/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2268 - accuracy: 0.8979 Epoch 53/500 31647/31647 [==============================] - 2s 68us/step - loss: 0.2265 - accuracy: 0.8966 Epoch 54/500 31647/31647 [==============================] - 2s 68us/step - loss: 0.2267 - accuracy: 0.8977 Epoch 55/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2265 - accuracy: 0.8980 Epoch 56/500 31647/31647 [==============================] - 2s 68us/step - loss: 0.2264 - accuracy: 0.8979 Epoch 57/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2263 - accuracy: 0.8971 Epoch 58/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2257 - accuracy: 0.8986 Epoch 59/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2260 - accuracy: 0.8980 Epoch 60/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2260 - accuracy: 0.8978 Epoch 61/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2257 - accuracy: 0.8986 Epoch 62/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2256 - accuracy: 0.8989 Epoch 63/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2251 - accuracy: 0.8983 Epoch 64/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2251 - accuracy: 0.8982 Epoch 65/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2251 - accuracy: 0.8978 Epoch 66/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2250 - accuracy: 0.8988 Epoch 67/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2251 - accuracy: 0.8976 Epoch 68/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2249 - accuracy: 0.8980 Epoch 69/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2246 - accuracy: 0.8982 Epoch 70/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2245 - accuracy: 0.8987 Epoch 71/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2246 - accuracy: 0.8988 Epoch 72/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2243 - accuracy: 0.8985 Epoch 73/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2240 - accuracy: 0.8983 Epoch 74/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2243 - accuracy: 0.8988 Epoch 75/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2240 - accuracy: 0.8990 Epoch 76/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2242 - accuracy: 0.8991 Epoch 77/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2240 - accuracy: 0.8986 Epoch 78/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2235 - accuracy: 0.8990 Epoch 79/500 31647/31647 [==============================] - 2s 71us/step - loss: 0.2239 - accuracy: 0.8986 Epoch 80/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2238 - accuracy: 0.8989 Epoch 81/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2236 - accuracy: 0.8991 Epoch 82/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2235 - accuracy: 0.8988 Epoch 83/500 31647/31647 [==============================] - 2s 71us/step - loss: 0.2235 - accuracy: 0.8988 Epoch 84/500 31647/31647 [==============================] - 2s 71us/step - loss: 0.2236 - accuracy: 0.8993 Epoch 85/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2236 - accuracy: 0.8992 Epoch 86/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2235 - accuracy: 0.8985 Epoch 87/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2234 - accuracy: 0.8994 Epoch 88/500 31647/31647 [==============================] - 2s 71us/step - loss: 0.2230 - accuracy: 0.8989 Epoch 89/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2233 - accuracy: 0.8991 Epoch 90/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2231 - accuracy: 0.8991 Epoch 91/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2228 - accuracy: 0.8993 Epoch 92/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2229 - accuracy: 0.8984 Epoch 93/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2225 - accuracy: 0.8989 Epoch 94/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2228 - accuracy: 0.8994 Epoch 95/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2222 - accuracy: 0.8991 Epoch 96/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2226 - accuracy: 0.9000 Epoch 97/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2225 - accuracy: 0.8990 Epoch 98/500 31647/31647 [==============================] - 2s 71us/step - loss: 0.2224 - accuracy: 0.8995 Epoch 99/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2222 - accuracy: 0.8994 Epoch 100/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2223 - accuracy: 0.8998 Epoch 101/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2220 - accuracy: 0.8992 Epoch 102/500 31647/31647 [==============================] - 2s 71us/step - loss: 0.2220 - accuracy: 0.8999 Epoch 103/500 31647/31647 [==============================] - 2s 73us/step - loss: 0.2218 - accuracy: 0.8997 Epoch 104/500 31647/31647 [==============================] - 2s 73us/step - loss: 0.2220 - accuracy: 0.9000 Epoch 105/500 31647/31647 [==============================] - 2s 74us/step - loss: 0.2218 - accuracy: 0.8997 Epoch 106/500 31647/31647 [==============================] - 2s 73us/step - loss: 0.2218 - accuracy: 0.9001 Epoch 107/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2214 - accuracy: 0.8995 Epoch 108/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2216 - accuracy: 0.8997 Epoch 109/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2217 - accuracy: 0.9000 Epoch 110/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2212 - accuracy: 0.9006 Epoch 111/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2214 - accuracy: 0.8998 Epoch 112/500 31647/31647 [==============================] - 2s 71us/step - loss: 0.2211 - accuracy: 0.9002 Epoch 113/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2215 - accuracy: 0.9002 Epoch 114/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2211 - accuracy: 0.9001 Epoch 115/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2212 - accuracy: 0.9001 Epoch 116/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2211 - accuracy: 0.9002 Epoch 117/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2209 - accuracy: 0.9001 Epoch 118/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2209 - accuracy: 0.9002 Epoch 119/500 31647/31647 [==============================] - 2s 71us/step - loss: 0.2204 - accuracy: 0.9004 Epoch 120/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2208 - accuracy: 0.9003 Epoch 121/500 31647/31647 [==============================] - 2s 71us/step - loss: 0.2206 - accuracy: 0.9010 Epoch 122/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2202 - accuracy: 0.8996 Epoch 123/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2206 - accuracy: 0.9001 Epoch 124/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2209 - accuracy: 0.8998 Epoch 125/500 31647/31647 [==============================] - 2s 71us/step - loss: 0.2208 - accuracy: 0.9005 Epoch 126/500 31647/31647 [==============================] - 2s 71us/step - loss: 0.2202 - accuracy: 0.9013 Epoch 127/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2206 - accuracy: 0.9005 Epoch 128/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2202 - accuracy: 0.9006 Epoch 129/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2202 - accuracy: 0.9013 Epoch 130/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2200 - accuracy: 0.9005 Epoch 131/500 31647/31647 [==============================] - 2s 71us/step - loss: 0.2202 - accuracy: 0.9009 Epoch 132/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2199 - accuracy: 0.9009 Epoch 133/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2200 - accuracy: 0.9003 Epoch 134/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2195 - accuracy: 0.9004 Epoch 135/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2201 - accuracy: 0.9010 Epoch 136/500 31647/31647 [==============================] - 2s 71us/step - loss: 0.2201 - accuracy: 0.9010 Epoch 137/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2198 - accuracy: 0.9006 Epoch 138/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2193 - accuracy: 0.9004 Epoch 139/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2197 - accuracy: 0.9012 Epoch 140/500 31647/31647 [==============================] - 2s 71us/step - loss: 0.2196 - accuracy: 0.9007 Epoch 141/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2195 - accuracy: 0.9010 Epoch 142/500 31647/31647 [==============================] - 2s 71us/step - loss: 0.2192 - accuracy: 0.9005 Epoch 143/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2196 - accuracy: 0.9013 Epoch 144/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2194 - accuracy: 0.9011 Epoch 145/500 31647/31647 [==============================] - 2s 71us/step - loss: 0.2191 - accuracy: 0.9006 Epoch 146/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2190 - accuracy: 0.9021 Epoch 147/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2196 - accuracy: 0.9001 Epoch 148/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2192 - accuracy: 0.9007 Epoch 149/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2190 - accuracy: 0.9019 Epoch 150/500 31647/31647 [==============================] - 2s 71us/step - loss: 0.2191 - accuracy: 0.9008 Epoch 151/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2190 - accuracy: 0.9012 Epoch 152/500 31647/31647 [==============================] - 2s 71us/step - loss: 0.2185 - accuracy: 0.9014 Epoch 153/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2190 - accuracy: 0.9013 Epoch 154/500 31647/31647 [==============================] - 2s 72us/step - loss: 0.2188 - accuracy: 0.9017 Epoch 155/500 31647/31647 [==============================] - 2s 71us/step - loss: 0.2189 - accuracy: 0.9013 Epoch 156/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2187 - accuracy: 0.9013 Epoch 157/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2190 - accuracy: 0.9008 Epoch 158/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2192 - accuracy: 0.9018 Epoch 159/500 31647/31647 [==============================] - 2s 71us/step - loss: 0.2186 - accuracy: 0.9016 Epoch 160/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2183 - accuracy: 0.9020 Epoch 161/500 31647/31647 [==============================] - 2s 71us/step - loss: 0.2184 - accuracy: 0.9009 Epoch 162/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2186 - accuracy: 0.9013 Epoch 163/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2184 - accuracy: 0.9020 Epoch 164/500 31647/31647 [==============================] - 2s 71us/step - loss: 0.2180 - accuracy: 0.9010 Epoch 165/500 31647/31647 [==============================] - 2s 71us/step - loss: 0.2182 - accuracy: 0.9015 Epoch 166/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2184 - accuracy: 0.9017 Epoch 167/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2178 - accuracy: 0.9017 Epoch 168/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2183 - accuracy: 0.9015 Epoch 169/500 31647/31647 [==============================] - 2s 71us/step - loss: 0.2179 - accuracy: 0.9018 Epoch 170/500 31647/31647 [==============================] - 2s 71us/step - loss: 0.2183 - accuracy: 0.9021 Epoch 171/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2181 - accuracy: 0.9023 Epoch 172/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2178 - accuracy: 0.9022 Epoch 173/500 31647/31647 [==============================] - 2s 71us/step - loss: 0.2180 - accuracy: 0.9021 Epoch 174/500 31647/31647 [==============================] - 2s 71us/step - loss: 0.2177 - accuracy: 0.9017 Epoch 175/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2173 - accuracy: 0.9019 Epoch 176/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2177 - accuracy: 0.9019 Epoch 177/500 31647/31647 [==============================] - 2s 71us/step - loss: 0.2177 - accuracy: 0.9022 Epoch 178/500 31647/31647 [==============================] - 2s 74us/step - loss: 0.2175 - accuracy: 0.9022 Epoch 179/500 31647/31647 [==============================] - 2s 74us/step - loss: 0.2174 - accuracy: 0.9014 Epoch 180/500 31647/31647 [==============================] - 2s 73us/step - loss: 0.2177 - accuracy: 0.9021 Epoch 181/500 31647/31647 [==============================] - 2s 71us/step - loss: 0.2173 - accuracy: 0.9025 Epoch 182/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2175 - accuracy: 0.9023 Epoch 183/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2175 - accuracy: 0.9026 Epoch 184/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2173 - accuracy: 0.9025 Epoch 185/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2172 - accuracy: 0.9024 Epoch 186/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2174 - accuracy: 0.9032 Epoch 187/500 31647/31647 [==============================] - 2s 72us/step - loss: 0.2174 - accuracy: 0.9026 Epoch 188/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2169 - accuracy: 0.9022 Epoch 189/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2173 - accuracy: 0.9022 Epoch 190/500 31647/31647 [==============================] - 2s 71us/step - loss: 0.2169 - accuracy: 0.9032 Epoch 191/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2169 - accuracy: 0.9027 Epoch 192/500 31647/31647 [==============================] - 2s 71us/step - loss: 0.2168 - accuracy: 0.9027 Epoch 193/500 31647/31647 [==============================] - 2s 71us/step - loss: 0.2170 - accuracy: 0.9026 Epoch 194/500 31647/31647 [==============================] - 2s 71us/step - loss: 0.2169 - accuracy: 0.9027 Epoch 195/500 31647/31647 [==============================] - 2s 71us/step - loss: 0.2168 - accuracy: 0.9028 Epoch 196/500 31647/31647 [==============================] - 2s 71us/step - loss: 0.2167 - accuracy: 0.9030 Epoch 197/500 31647/31647 [==============================] - 2s 71us/step - loss: 0.2169 - accuracy: 0.9024 Epoch 198/500 31647/31647 [==============================] - 2s 71us/step - loss: 0.2169 - accuracy: 0.9025 Epoch 199/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2167 - accuracy: 0.9029 Epoch 200/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2164 - accuracy: 0.9030 Epoch 201/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2166 - accuracy: 0.9034 Epoch 202/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2162 - accuracy: 0.9028 Epoch 203/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2163 - accuracy: 0.9032 Epoch 204/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2165 - accuracy: 0.9026 Epoch 205/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2166 - accuracy: 0.9030 Epoch 206/500 31647/31647 [==============================] - 2s 71us/step - loss: 0.2165 - accuracy: 0.9030 Epoch 207/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2162 - accuracy: 0.9033 Epoch 208/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2162 - accuracy: 0.9027 Epoch 209/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2164 - accuracy: 0.9033 Epoch 210/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2161 - accuracy: 0.9034 Epoch 211/500 31647/31647 [==============================] - 2s 71us/step - loss: 0.2160 - accuracy: 0.9036 Epoch 212/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2163 - accuracy: 0.9032 Epoch 213/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2161 - accuracy: 0.9034 Epoch 214/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2159 - accuracy: 0.9034 Epoch 215/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2162 - accuracy: 0.9036 Epoch 216/500 31647/31647 [==============================] - 2s 71us/step - loss: 0.2159 - accuracy: 0.9032 Epoch 217/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2158 - accuracy: 0.9036 Epoch 218/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2161 - accuracy: 0.9036 Epoch 219/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2159 - accuracy: 0.9036 Epoch 220/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2157 - accuracy: 0.9033 Epoch 221/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2155 - accuracy: 0.9037 Epoch 222/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2157 - accuracy: 0.9038 Epoch 223/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2158 - accuracy: 0.9038 Epoch 224/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2154 - accuracy: 0.9032 Epoch 225/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2157 - accuracy: 0.9037 Epoch 226/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2155 - accuracy: 0.9038 Epoch 227/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2156 - accuracy: 0.9033 Epoch 228/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2154 - accuracy: 0.9042 Epoch 229/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2154 - accuracy: 0.9028 Epoch 230/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2151 - accuracy: 0.9038 Epoch 231/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2153 - accuracy: 0.9035 Epoch 232/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2155 - accuracy: 0.9036 Epoch 233/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2150 - accuracy: 0.9037 Epoch 234/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2150 - accuracy: 0.9038 Epoch 235/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2153 - accuracy: 0.9041 Epoch 236/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2151 - accuracy: 0.9042 Epoch 237/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2149 - accuracy: 0.9045 Epoch 238/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2152 - accuracy: 0.9042 Epoch 239/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2150 - accuracy: 0.9038 Epoch 240/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2147 - accuracy: 0.9046 Epoch 241/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2151 - accuracy: 0.9040 Epoch 242/500 31647/31647 [==============================] - 2s 73us/step - loss: 0.2150 - accuracy: 0.9033 Epoch 243/500 31647/31647 [==============================] - 2s 73us/step - loss: 0.2146 - accuracy: 0.9043 Epoch 244/500 31647/31647 [==============================] - 2s 73us/step - loss: 0.2148 - accuracy: 0.9038 Epoch 245/500 31647/31647 [==============================] - 2s 72us/step - loss: 0.2148 - accuracy: 0.9043 Epoch 246/500 31647/31647 [==============================] - 2s 72us/step - loss: 0.2150 - accuracy: 0.9041 Epoch 247/500 31647/31647 [==============================] - 2s 71us/step - loss: 0.2147 - accuracy: 0.9032 Epoch 248/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2147 - accuracy: 0.9043 Epoch 249/500 31647/31647 [==============================] - 2s 71us/step - loss: 0.2145 - accuracy: 0.9044 Epoch 250/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2147 - accuracy: 0.9041 Epoch 251/500 31647/31647 [==============================] - 2s 71us/step - loss: 0.2145 - accuracy: 0.9047 Epoch 252/500 31647/31647 [==============================] - 2s 71us/step - loss: 0.2146 - accuracy: 0.9034 Epoch 253/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2143 - accuracy: 0.9047 Epoch 254/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2146 - accuracy: 0.9044 Epoch 255/500 31647/31647 [==============================] - 2s 71us/step - loss: 0.2143 - accuracy: 0.9041 Epoch 256/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2145 - accuracy: 0.9042 Epoch 257/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2144 - accuracy: 0.9039 Epoch 258/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2142 - accuracy: 0.9037 Epoch 259/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2142 - accuracy: 0.9043 Epoch 260/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2142 - accuracy: 0.9048 Epoch 261/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2141 - accuracy: 0.9044 Epoch 262/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2143 - accuracy: 0.9044 Epoch 263/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2143 - accuracy: 0.9040 Epoch 264/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2141 - accuracy: 0.9046 Epoch 265/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2142 - accuracy: 0.9045 Epoch 266/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2142 - accuracy: 0.9044 Epoch 267/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2137 - accuracy: 0.9043 Epoch 268/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2139 - accuracy: 0.9047 Epoch 269/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2136 - accuracy: 0.9039 Epoch 270/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2138 - accuracy: 0.9048 Epoch 271/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2140 - accuracy: 0.9048 Epoch 272/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2137 - accuracy: 0.9047 Epoch 273/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2138 - accuracy: 0.9044 Epoch 274/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2137 - accuracy: 0.9047 Epoch 275/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2137 - accuracy: 0.9047 Epoch 276/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2140 - accuracy: 0.9050 Epoch 277/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2138 - accuracy: 0.9045 Epoch 278/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2136 - accuracy: 0.9048 Epoch 279/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2134 - accuracy: 0.9055 Epoch 280/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2135 - accuracy: 0.9043 Epoch 281/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2136 - accuracy: 0.9039 Epoch 282/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2135 - accuracy: 0.9043 Epoch 283/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2137 - accuracy: 0.9048 Epoch 284/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2131 - accuracy: 0.9038 Epoch 285/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2134 - accuracy: 0.9043 Epoch 286/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2134 - accuracy: 0.9041 Epoch 287/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2132 - accuracy: 0.9050 Epoch 288/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2132 - accuracy: 0.9046 Epoch 289/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2134 - accuracy: 0.9049 Epoch 290/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2132 - accuracy: 0.9049 Epoch 291/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2131 - accuracy: 0.9050 Epoch 292/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2133 - accuracy: 0.9050 Epoch 293/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2128 - accuracy: 0.9049 Epoch 294/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2130 - accuracy: 0.9049 Epoch 295/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2127 - accuracy: 0.9047 Epoch 296/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2131 - accuracy: 0.9047 Epoch 297/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2132 - accuracy: 0.9047 Epoch 298/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2129 - accuracy: 0.9048 Epoch 299/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2130 - accuracy: 0.9059 Epoch 300/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2129 - accuracy: 0.9050 Epoch 301/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2131 - accuracy: 0.9045 Epoch 302/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2128 - accuracy: 0.9053 Epoch 303/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2129 - accuracy: 0.9051 Epoch 304/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2128 - accuracy: 0.9045 Epoch 305/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2126 - accuracy: 0.9055 Epoch 306/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2127 - accuracy: 0.9048 Epoch 307/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2125 - accuracy: 0.9045 Epoch 308/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2127 - accuracy: 0.9047 Epoch 309/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2125 - accuracy: 0.9053 Epoch 310/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2127 - accuracy: 0.9048 Epoch 311/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2126 - accuracy: 0.9051 Epoch 312/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2122 - accuracy: 0.9053 Epoch 313/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2125 - accuracy: 0.9051 Epoch 314/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2123 - accuracy: 0.9050 Epoch 315/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2124 - accuracy: 0.9051 Epoch 316/500 31647/31647 [==============================] - 2s 72us/step - loss: 0.2124 - accuracy: 0.9051 Epoch 317/500 31647/31647 [==============================] - 2s 73us/step - loss: 0.2122 - accuracy: 0.9053 Epoch 318/500 31647/31647 [==============================] - 2s 73us/step - loss: 0.2124 - accuracy: 0.9056 Epoch 319/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2123 - accuracy: 0.9058 Epoch 320/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2125 - accuracy: 0.9057 Epoch 321/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2122 - accuracy: 0.9049 Epoch 322/500 31647/31647 [==============================] - 2s 71us/step - loss: 0.2121 - accuracy: 0.9052 Epoch 323/500 31647/31647 [==============================] - 2s 71us/step - loss: 0.2123 - accuracy: 0.9055 Epoch 324/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2122 - accuracy: 0.9049 Epoch 325/500 31647/31647 [==============================] - 2s 71us/step - loss: 0.2121 - accuracy: 0.9052 Epoch 326/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2122 - accuracy: 0.9056 Epoch 327/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2120 - accuracy: 0.9054 Epoch 328/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2120 - accuracy: 0.9056 Epoch 329/500 31647/31647 [==============================] - 2s 71us/step - loss: 0.2121 - accuracy: 0.9051 Epoch 330/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2119 - accuracy: 0.9058 Epoch 331/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2121 - accuracy: 0.9056 Epoch 332/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2118 - accuracy: 0.9054 Epoch 333/500 31647/31647 [==============================] - 2s 71us/step - loss: 0.2119 - accuracy: 0.9053 Epoch 334/500 31647/31647 [==============================] - 2s 71us/step - loss: 0.2119 - accuracy: 0.9053 Epoch 335/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2117 - accuracy: 0.9062 Epoch 336/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2117 - accuracy: 0.9050 Epoch 337/500 31647/31647 [==============================] - 2s 71us/step - loss: 0.2120 - accuracy: 0.9060 Epoch 338/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2117 - accuracy: 0.9052 Epoch 339/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2118 - accuracy: 0.9056 Epoch 340/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2117 - accuracy: 0.9055 Epoch 341/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2120 - accuracy: 0.9059 Epoch 342/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2118 - accuracy: 0.9057 Epoch 343/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2116 - accuracy: 0.9049 Epoch 344/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2113 - accuracy: 0.9056 Epoch 345/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2117 - accuracy: 0.9053 Epoch 346/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2114 - accuracy: 0.9054 Epoch 347/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2113 - accuracy: 0.9056 Epoch 348/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2114 - accuracy: 0.9060 Epoch 349/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2114 - accuracy: 0.9062 Epoch 350/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2114 - accuracy: 0.9060 Epoch 351/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2113 - accuracy: 0.9053 Epoch 352/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2114 - accuracy: 0.9059 Epoch 353/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2112 - accuracy: 0.9064 Epoch 354/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2113 - accuracy: 0.9058 Epoch 355/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2111 - accuracy: 0.9052 Epoch 356/500 31647/31647 [==============================] - 2s 71us/step - loss: 0.2112 - accuracy: 0.9062 Epoch 357/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2112 - accuracy: 0.9058 Epoch 358/500 31647/31647 [==============================] - 2s 71us/step - loss: 0.2111 - accuracy: 0.9056 Epoch 359/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2112 - accuracy: 0.9059 Epoch 360/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2107 - accuracy: 0.9064 Epoch 361/500 31647/31647 [==============================] - 2s 68us/step - loss: 0.2112 - accuracy: 0.9053 Epoch 362/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2109 - accuracy: 0.9058 Epoch 363/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2111 - accuracy: 0.9061 Epoch 364/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2110 - accuracy: 0.9062 Epoch 365/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2110 - accuracy: 0.9061 Epoch 366/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2106 - accuracy: 0.9059 Epoch 367/500 31647/31647 [==============================] - 2s 71us/step - loss: 0.2109 - accuracy: 0.9062 Epoch 368/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2108 - accuracy: 0.9052 Epoch 369/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2107 - accuracy: 0.9059 Epoch 370/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2108 - accuracy: 0.9057 Epoch 371/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2107 - accuracy: 0.9059 Epoch 372/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2110 - accuracy: 0.9060 Epoch 373/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2108 - accuracy: 0.9059 Epoch 374/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2108 - accuracy: 0.9052 Epoch 375/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2107 - accuracy: 0.9057 Epoch 376/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2107 - accuracy: 0.9058 Epoch 377/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2109 - accuracy: 0.9055 Epoch 378/500 31647/31647 [==============================] - 2s 68us/step - loss: 0.2107 - accuracy: 0.9060 Epoch 379/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2107 - accuracy: 0.9062 Epoch 380/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2107 - accuracy: 0.9061 Epoch 381/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2108 - accuracy: 0.9061 Epoch 382/500 31647/31647 [==============================] - 2s 71us/step - loss: 0.2103 - accuracy: 0.9061 Epoch 383/500 31647/31647 [==============================] - 2s 73us/step - loss: 0.2106 - accuracy: 0.9058 Epoch 384/500 31647/31647 [==============================] - 2s 73us/step - loss: 0.2104 - accuracy: 0.9062 Epoch 385/500 31647/31647 [==============================] - 2s 73us/step - loss: 0.2104 - accuracy: 0.9063 Epoch 386/500 31647/31647 [==============================] - 2s 73us/step - loss: 0.2106 - accuracy: 0.9055 Epoch 387/500 31647/31647 [==============================] - 2s 71us/step - loss: 0.2102 - accuracy: 0.9066 Epoch 388/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2105 - accuracy: 0.9059 Epoch 389/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2104 - accuracy: 0.9061 Epoch 390/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2103 - accuracy: 0.9067 Epoch 391/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2106 - accuracy: 0.9063 Epoch 392/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2102 - accuracy: 0.9057 Epoch 393/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2105 - accuracy: 0.9064 Epoch 394/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2102 - accuracy: 0.9066 Epoch 395/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2103 - accuracy: 0.9059 Epoch 396/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2102 - accuracy: 0.9067 Epoch 397/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2099 - accuracy: 0.9066 Epoch 398/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2101 - accuracy: 0.9060 Epoch 399/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2102 - accuracy: 0.9061 Epoch 400/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2101 - accuracy: 0.9059 Epoch 401/500 31647/31647 [==============================] - 2s 71us/step - loss: 0.2100 - accuracy: 0.9059 Epoch 402/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2101 - accuracy: 0.9063 Epoch 403/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2099 - accuracy: 0.9065 Epoch 404/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2098 - accuracy: 0.9062 Epoch 405/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2100 - accuracy: 0.9071 Epoch 406/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2098 - accuracy: 0.9063 Epoch 407/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2102 - accuracy: 0.9065 Epoch 408/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2099 - accuracy: 0.9063 Epoch 409/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2096 - accuracy: 0.9063 Epoch 410/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2099 - accuracy: 0.9063 Epoch 411/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2098 - accuracy: 0.9061 Epoch 412/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2097 - accuracy: 0.9062 Epoch 413/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2097 - accuracy: 0.9069 Epoch 414/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2097 - accuracy: 0.9062 Epoch 415/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2095 - accuracy: 0.9063 Epoch 416/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2099 - accuracy: 0.9056 Epoch 417/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2096 - accuracy: 0.9063 Epoch 418/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2097 - accuracy: 0.9070 Epoch 419/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2094 - accuracy: 0.9072 Epoch 420/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2097 - accuracy: 0.9061 Epoch 421/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2095 - accuracy: 0.9062 Epoch 422/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2097 - accuracy: 0.9065 Epoch 423/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2095 - accuracy: 0.9065 Epoch 424/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2096 - accuracy: 0.9069 Epoch 425/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2092 - accuracy: 0.9066 Epoch 426/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2096 - accuracy: 0.9071 Epoch 427/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2092 - accuracy: 0.9070 Epoch 428/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2093 - accuracy: 0.9066 Epoch 429/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2095 - accuracy: 0.9067 Epoch 430/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2093 - accuracy: 0.9068 Epoch 431/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2093 - accuracy: 0.9060 Epoch 432/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2095 - accuracy: 0.9065 Epoch 433/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2089 - accuracy: 0.9080 Epoch 434/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2093 - accuracy: 0.9065 Epoch 435/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2093 - accuracy: 0.9072 Epoch 436/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2091 - accuracy: 0.9068 Epoch 437/500 31647/31647 [==============================] - 2s 68us/step - loss: 0.2093 - accuracy: 0.9069 Epoch 438/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2090 - accuracy: 0.9074 Epoch 439/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2091 - accuracy: 0.9068 Epoch 440/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2088 - accuracy: 0.9070 Epoch 441/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2089 - accuracy: 0.9071 Epoch 442/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2090 - accuracy: 0.9069 Epoch 443/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2090 - accuracy: 0.9065 Epoch 444/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2090 - accuracy: 0.9069 Epoch 445/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2091 - accuracy: 0.9073 Epoch 446/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2090 - accuracy: 0.9069 Epoch 447/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2088 - accuracy: 0.9069 Epoch 448/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2091 - accuracy: 0.9069 Epoch 449/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2089 - accuracy: 0.9067 Epoch 450/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2087 - accuracy: 0.9072 Epoch 451/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2087 - accuracy: 0.9069 Epoch 452/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2085 - accuracy: 0.9071 Epoch 453/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2087 - accuracy: 0.9073 Epoch 454/500 31647/31647 [==============================] - 2s 71us/step - loss: 0.2086 - accuracy: 0.9065 Epoch 455/500 31647/31647 [==============================] - 2s 72us/step - loss: 0.2088 - accuracy: 0.9067 Epoch 456/500 31647/31647 [==============================] - 2s 73us/step - loss: 0.2086 - accuracy: 0.9071 Epoch 457/500 31647/31647 [==============================] - 2s 71us/step - loss: 0.2086 - accuracy: 0.9071 Epoch 458/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2088 - accuracy: 0.9070 Epoch 459/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2084 - accuracy: 0.9071 Epoch 460/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2086 - accuracy: 0.9074 Epoch 461/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2084 - accuracy: 0.9072 Epoch 462/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2088 - accuracy: 0.9072 Epoch 463/500 31647/31647 [==============================] - 2s 71us/step - loss: 0.2084 - accuracy: 0.9072 Epoch 464/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2084 - accuracy: 0.9065 Epoch 465/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2083 - accuracy: 0.9068 Epoch 466/500 31647/31647 [==============================] - 2s 72us/step - loss: 0.2084 - accuracy: 0.9078 Epoch 467/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2084 - accuracy: 0.9073 Epoch 468/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2085 - accuracy: 0.9072 Epoch 469/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2083 - accuracy: 0.9075 Epoch 470/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2083 - accuracy: 0.9074 Epoch 471/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2084 - accuracy: 0.9074 Epoch 472/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2083 - accuracy: 0.9076 Epoch 473/500 31647/31647 [==============================] - 2s 71us/step - loss: 0.2081 - accuracy: 0.9068 Epoch 474/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2082 - accuracy: 0.9071 Epoch 475/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2082 - accuracy: 0.9073 Epoch 476/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2079 - accuracy: 0.9075 Epoch 477/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2086 - accuracy: 0.9071 Epoch 478/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2081 - accuracy: 0.9074 Epoch 479/500 31647/31647 [==============================] - 2s 71us/step - loss: 0.2079 - accuracy: 0.9078 Epoch 480/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2081 - accuracy: 0.9070 Epoch 481/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2079 - accuracy: 0.9073 Epoch 482/500 31647/31647 [==============================] - 2s 71us/step - loss: 0.2081 - accuracy: 0.9073 Epoch 483/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2080 - accuracy: 0.9074 Epoch 484/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2082 - accuracy: 0.9067 Epoch 485/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2080 - accuracy: 0.9072 Epoch 486/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2079 - accuracy: 0.9074 Epoch 487/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2080 - accuracy: 0.9076 Epoch 488/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2078 - accuracy: 0.9081 Epoch 489/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2080 - accuracy: 0.9075 Epoch 490/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2080 - accuracy: 0.9071 Epoch 491/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2078 - accuracy: 0.9080 Epoch 492/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2077 - accuracy: 0.9071 Epoch 493/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2079 - accuracy: 0.9068 Epoch 494/500 31647/31647 [==============================] - 2s 68us/step - loss: 0.2079 - accuracy: 0.9073 Epoch 495/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2077 - accuracy: 0.9077 Epoch 496/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2076 - accuracy: 0.9077 Epoch 497/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2076 - accuracy: 0.9080 Epoch 498/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2077 - accuracy: 0.9071 Epoch 499/500 31647/31647 [==============================] - 2s 70us/step - loss: 0.2075 - accuracy: 0.9074 Epoch 500/500 31647/31647 [==============================] - 2s 69us/step - loss: 0.2079 - accuracy: 0.9075 Final loss (Test data): 0.2293, final accuracy (Test data): 0.8980
In [73]:
