polishing (#2)

* license added in setup.py * index fixed * index fixed * blind guesses * remove .egg-info from git * .egg-info added to .gitignore * _proc removed from git and added to .gitignore * polishing
airtai · Sep 3, 2022 · 8a5f3eb · 8a5f3eb
1 parent 6687f36
commit 8a5f3eb
Show file tree

Hide file tree

Showing 7 changed files with 115 additions and 131 deletions.
diff --git a/README.md b/README.md
@@ -1,4 +1,4 @@
-mono-dense-keras
+Constrained Monotonic Neural Networks
 ================
 
 <!-- WARNING: THIS FILE WAS AUTOGENERATED! DO NOT EDIT! -->
@@ -71,7 +71,7 @@ def generate_data(no_samples: int, noise: float):
     y = x[:, 0] ** 3
     y += np.sin(x[:, 1] / (2*np.pi))
     y += np.exp(-x[:, 2])
-    y += 0.1 * rng.normal(size=no_samples)
+    y += noise * rng.normal(size=no_samples)
     return x, y
 
 x_train, y_train = generate_data(10_000, noise=0.1)
@@ -98,9 +98,6 @@ model.add(Dense(128, activation="elu"))
 model.add(Dense(1))
 ```
 
-    /usr/lib/python3/dist-packages/requests/__init__.py:89: RequestsDependencyWarning: urllib3 (1.26.11) or chardet (3.0.4) doesn't match a supported version!
-      warnings.warn("urllib3 ({}) or chardet ({}) doesn't match a supported "
-
 We’ll train the network using the `Adam` optimizer and the
 `ExponentialDecay` learning rate schedule:
 
@@ -121,25 +118,25 @@ train_model(model, initial_learning_rate=.1)
 ```
 
     Epoch 1/10
-    313/313 [==============================] - 1s 2ms/step - loss: 9.1129 - val_loss: 9.2426
+    313/313 [==============================] - 1s 2ms/step - loss: 9.1098 - val_loss: 9.2552
     Epoch 2/10
-    313/313 [==============================] - 1s 2ms/step - loss: 7.8147 - val_loss: 8.4241
+    313/313 [==============================] - 1s 2ms/step - loss: 7.7995 - val_loss: 8.3143
     Epoch 3/10
-    313/313 [==============================] - 1s 2ms/step - loss: 7.5386 - val_loss: 8.0456
+    313/313 [==============================] - 1s 2ms/step - loss: 7.5270 - val_loss: 8.0499
     Epoch 4/10
-    313/313 [==============================] - 1s 2ms/step - loss: 7.0556 - val_loss: 7.3737
+    313/313 [==============================] - 1s 2ms/step - loss: 7.2095 - val_loss: 7.5935
     Epoch 5/10
-    313/313 [==============================] - 1s 2ms/step - loss: 5.8239 - val_loss: 4.4619
+    313/313 [==============================] - 1s 2ms/step - loss: 6.0665 - val_loss: 6.7911
     Epoch 6/10
-    313/313 [==============================] - 1s 2ms/step - loss: 2.3169 - val_loss: 1.5477
+    313/313 [==============================] - 1s 2ms/step - loss: 3.1178 - val_loss: 1.5964
     Epoch 7/10
-    313/313 [==============================] - 1s 2ms/step - loss: 1.0104 - val_loss: 0.7407
+    313/313 [==============================] - 1s 2ms/step - loss: 1.1686 - val_loss: 0.8541
     Epoch 8/10
-    313/313 [==============================] - 1s 2ms/step - loss: 0.6479 - val_loss: 0.9755
+    313/313 [==============================] - 1s 2ms/step - loss: 0.7370 - val_loss: 1.5969
     Epoch 9/10
-    313/313 [==============================] - 1s 2ms/step - loss: 0.5144 - val_loss: 0.3477
+    313/313 [==============================] - 1s 2ms/step - loss: 0.6011 - val_loss: 0.3739
     Epoch 10/10
-    313/313 [==============================] - 1s 2ms/step - loss: 0.3837 - val_loss: 0.2306
+    313/313 [==============================] - 1s 2ms/step - loss: 0.4458 - val_loss: 0.3114
 
 Now, we’ll use the
 [`MonotonicDense`](https://airtai.github.io/mono-dense-keras/layers.html#monotonicdense)
@@ -179,22 +176,22 @@ train_model(model, initial_learning_rate=.001)
 ```
 
     Epoch 1/10
-    313/313 [==============================] - 1s 2ms/step - loss: 0.3578 - val_loss: 0.1557
+    313/313 [==============================] - 1s 2ms/step - loss: 0.3646 - val_loss: 0.2042
     Epoch 2/10
-    313/313 [==============================] - 1s 2ms/step - loss: 0.2650 - val_loss: 0.2754
+    313/313 [==============================] - 1s 2ms/step - loss: 0.2895 - val_loss: 0.1387
     Epoch 3/10
-    313/313 [==============================] - 1s 2ms/step - loss: 0.2238 - val_loss: 0.0917
+    313/313 [==============================] - 1s 2ms/step - loss: 0.2756 - val_loss: 0.1027
     Epoch 4/10
-    313/313 [==============================] - 1s 2ms/step - loss: 0.1867 - val_loss: 0.0798
+    313/313 [==============================] - 1s 2ms/step - loss: 0.2281 - val_loss: 0.0814
     Epoch 5/10
-    313/313 [==============================] - 1s 2ms/step - loss: 0.1724 - val_loss: 0.7418
+    313/313 [==============================] - 1s 2ms/step - loss: 0.1816 - val_loss: 0.0634
     Epoch 6/10
-    313/313 [==============================] - 1s 2ms/step - loss: 0.1356 - val_loss: 0.4525
+    313/313 [==============================] - 1s 2ms/step - loss: 0.1631 - val_loss: 0.1443
     Epoch 7/10
-    313/313 [==============================] - 1s 2ms/step - loss: 0.1533 - val_loss: 0.0503
+    313/313 [==============================] - 1s 2ms/step - loss: 0.1455 - val_loss: 0.1299
     Epoch 8/10
-    313/313 [==============================] - 1s 2ms/step - loss: 0.1274 - val_loss: 0.0395
+    313/313 [==============================] - 1s 2ms/step - loss: 0.1701 - val_loss: 0.0709
     Epoch 9/10
-    313/313 [==============================] - 1s 2ms/step - loss: 0.1171 - val_loss: 0.0381
+    313/313 [==============================] - 1s 2ms/step - loss: 0.1250 - val_loss: 0.0644
     Epoch 10/10
-    313/313 [==============================] - 1s 2ms/step - loss: 0.1005 - val_loss: 0.0500
+    313/313 [==============================] - 1s 2ms/step - loss: 0.1426 - val_loss: 0.0405