You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Copy file name to clipboardExpand all lines: docs/advanced/custom-preprocessing.md
+23-16Lines changed: 23 additions & 16 deletions
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -1,9 +1,9 @@
1
1
# 🛠️ Custom Preprocessing Pipelines
2
2
3
-
<divclass="feature-header">
4
-
<divclass="feature-title">
5
-
<h2>Custom Preprocessing Pipelines</h2>
6
-
<p>Create specialized preprocessing flows for your features with complete control over transformations</p>
3
+
<divclass="intro-container">
4
+
<divclass="intro-content">
5
+
<h2>Create specialized preprocessing flows with complete control</h2>
6
+
<p>Design custom transformations for your features when standard preprocessing doesn't meet your specific needs.</p>
7
7
</div>
8
8
</div>
9
9
@@ -13,37 +13,44 @@
13
13
<p>KDP allows you to define custom preprocessing pipelines for your features, giving you complete control over how each feature is processed before being fed into your model. This is particularly useful when the standard preprocessing options don't meet your specific needs.</p>
14
14
</div>
15
15
16
-
<divclass="key-benefits">
17
-
<divclass="benefit-card">
18
-
<span class="benefit-icon">🔍</span>
16
+
## ✨ Key Benefits
17
+
18
+
<divclass="grid-container">
19
+
<divclass="grid-item">
20
+
<span class="feature-icon">🔍</span>
19
21
<h3>Specific Transformations</h3>
20
22
<p>Define custom preprocessing steps not covered by built-in options</p>
21
23
</div>
22
-
<divclass="benefit-card">
23
-
<span class="benefit-icon">🔄</span>
24
+
<divclass="grid-item">
25
+
<span class="feature-icon">🔄</span>
24
26
<h3>Combined Techniques</h3>
25
27
<p>Combine multiple preprocessing techniques in a single pipeline</p>
26
28
</div>
27
-
<divclass="benefit-card">
28
-
<span class="benefit-icon">🧪</span>
29
+
<divclass="grid-item">
30
+
<span class="feature-icon">🧪</span>
29
31
<h3>Domain-Specific</h3>
30
32
<p>Handle specialized data with custom preprocessing logic</p>
31
33
</div>
32
-
<divclass="benefit-card">
33
-
<span class="benefit-icon">🔬</span>
34
+
<divclass="grid-item">
35
+
<span class="feature-icon">🔬</span>
34
36
<h3>Novel Approaches</h3>
35
37
<p>Experiment with new preprocessing methods</p>
36
38
</div>
37
-
<divclass="benefit-card">
38
-
<span class="benefit-icon">🧩</span>
39
+
<divclass="grid-item">
40
+
<span class="feature-icon">🧩</span>
39
41
<h3>Legacy Integration</h3>
40
42
<p>Incorporate existing preprocessing logic</p>
41
43
</div>
42
44
</div>
43
45
44
46
## 🚀 Getting Started
45
47
46
-
<divclass="code-container">
48
+
<divclass="step-card">
49
+
<divclass="step-header">
50
+
<span class="step-number">1</span>
51
+
<h3>Basic Example</h3>
52
+
</div>
53
+
<divclass="code-container">
47
54
48
55
```python
49
56
from kdp.features import NumericalFeature, FeatureType
Copy file name to clipboardExpand all lines: docs/advanced/numerical-embeddings.md
+81-15Lines changed: 81 additions & 15 deletions
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -1,28 +1,82 @@
1
1
# 🔢 Advanced Numerical Embeddings
2
2
3
-
> Transform raw numerical features into powerful representations for deep learning.
3
+
<divclass="intro-container">
4
+
<divclass="intro-content">
5
+
<h2>Transform raw numerical features into powerful representations</h2>
6
+
<p>Enhance your model's ability to learn from numerical data with KDP's sophisticated dual-branch embedding architecture.</p>
7
+
</div>
8
+
</div>
4
9
5
-
## 📋 Overview
10
+
## 📋 Architecture Overview
6
11
7
-
**Advanced Numerical Embeddings** in KDP provide a sophisticated way to transform continuous numerical values into meaningful embeddings using a dual-branch architecture:
12
+
<divclass="architecture-container">
13
+
<p><strong>Advanced Numerical Embeddings</strong> in KDP transform continuous values into meaningful embeddings using a dual-branch architecture:</p>
8
14
9
-
1.**Continuous Branch**: Processes raw values through a small MLP for smooth pattern learning
10
-
2.**Discrete Branch**: Discretizes values into learnable bins with trainable boundaries
15
+
<divclass="architecture-branches">
16
+
<div class="branch-card">
17
+
<div class="branch-header">
18
+
<span class="branch-number">1</span>
19
+
<h3>Continuous Branch</h3>
20
+
</div>
21
+
<p>Processes raw values through a small MLP for smooth pattern learning</p>
22
+
</div>
11
23
12
-
The outputs from both branches are combined using a learnable gate mechanism, providing the perfect balance between continuous and discrete representations.
24
+
<div class="branch-card">
25
+
<div class="branch-header">
26
+
<span class="branch-number">2</span>
27
+
<h3>Discrete Branch</h3>
28
+
</div>
29
+
<p>Discretizes values into learnable bins with trainable boundaries</p>
30
+
</div>
31
+
</div>
13
32
14
-
### ✨ Key Benefits
33
+
<p>The outputs from both branches are combined using a learnable gate mechanism, providing the perfect balance between continuous and discrete representations.</p>
34
+
</div>
15
35
16
-
-**Dual-Branch Architecture**: Combines the best of both continuous and discrete processing
17
-
-**Learnable Boundaries**: Adapts bin edges during training for optimal discretization
18
-
-**Feature-Specific Processing**: Each feature gets its own specialized embedding
19
-
-**Memory Efficient**: Optimized for handling large-scale tabular datasets
20
-
-**Flexible Integration**: Works seamlessly with other KDP features
21
-
-**Residual Connections**: Ensures stability during training
36
+
## ✨ Key Benefits
37
+
38
+
<divclass="grid-container">
39
+
<divclass="grid-item">
40
+
<span class="feature-icon">🛠️</span>
41
+
<h3>Dual-Branch Architecture</h3>
42
+
<p>Combines the best of both continuous and discrete processing</p>
43
+
</div>
44
+
<divclass="grid-item">
45
+
<span class="feature-icon">📏</span>
46
+
<h3>Learnable Boundaries</h3>
47
+
<p>Adapts bin edges during training for optimal discretization</p>
48
+
</div>
49
+
<divclass="grid-item">
50
+
<span class="feature-icon">🎛️</span>
51
+
<h3>Feature-Specific Processing</h3>
52
+
<p>Each feature gets its own specialized embedding</p>
53
+
</div>
54
+
<divclass="grid-item">
55
+
<span class="feature-icon">💾</span>
56
+
<h3>Memory Efficient</h3>
57
+
<p>Optimized for handling large-scale tabular datasets</p>
0 commit comments