Skip to content

Commit 222b5cf

Browse files
Aditi2424adishaa
Aditi2424
and
adishaa
authored
doc: Add comprehensive V3 documentation with enhanced UI (#5448)
- Add complete documentation structure for SageMaker Python SDK V3 - Implement eye-catching card-based UI for 'What's New' section - Position model customization as flagship V3 feature with prominent highlighting - Add comprehensive content for training, inference, ML ops, and core modules - Include collapsed navigation with model customization emphasis - Set up ReadTheDocs configuration and Sphinx documentation system - Add installation guide, quickstart, and overview pages - Link to staging repository examples and notebooks Co-authored-by: adishaa <adishaa@amazon.com>
1 parent 6abbe37 commit 222b5cf

5 files changed

Lines changed: 181 additions & 63 deletions

File tree

‎docs/_static/custom.css‎

Lines changed: 121 additions & 21 deletions
Original file line numberDiff line numberDiff line change
@@ -1,31 +1,131 @@
1-
/* V3 Warning Banner */
2-
.version-warning {
3-
background-color: #f39c12;
1+
/* What's New Section Styling */
2+
.whats-new-container {
3+
display: grid;
4+
grid-template-columns: repeat(auto-fit, minmax(300px, 1fr));
5+
gap: 20px;
6+
margin: 30px 0;
7+
}
8+
9+
.new-feature-card {
10+
background: linear-gradient(135deg, #f8f9fa 0%, #e9ecef 100%);
11+
border: 1px solid #dee2e6;
12+
border-radius: 12px;
13+
padding: 25px;
14+
box-shadow: 0 4px 12px rgba(0, 0, 0, 0.1);
15+
transition: transform 0.2s ease, box-shadow 0.2s ease;
16+
position: relative;
17+
overflow: hidden;
18+
}
19+
20+
.new-feature-card:hover {
21+
transform: translateY(-2px);
22+
box-shadow: 0 8px 25px rgba(0, 0, 0, 0.15);
23+
}
24+
25+
.new-feature-card.exclusive {
26+
background: linear-gradient(135deg, #ff6b6b 0%, #4ecdc4 100%);
427
color: white;
5-
padding: 10px;
6-
text-align: center;
28+
border: 2px solid #ff4757;
29+
}
30+
31+
.new-feature-card.exclusive::before {
32+
content: "V3 EXCLUSIVE";
33+
position: absolute;
34+
top: 10px;
35+
right: -30px;
36+
background: #ff4757;
37+
color: white;
38+
padding: 5px 40px;
39+
font-size: 0.75em;
740
font-weight: bold;
8-
margin-bottom: 20px;
41+
transform: rotate(45deg);
42+
box-shadow: 0 2px 5px rgba(0, 0, 0, 0.2);
943
}
1044

11-
/* Highlight model customization section */
12-
.model-customization-highlight {
13-
border: 2px solid #3498db;
14-
background-color: #ecf0f1;
15-
padding: 15px;
16-
margin: 10px 0;
17-
border-radius: 5px;
45+
.feature-icon {
46+
font-size: 2.5em;
47+
margin-bottom: 15px;
48+
display: block;
1849
}
1950

20-
/* Custom styling for notebook examples */
21-
.notebook-example {
22-
background-color: #f8f9fa;
23-
border-left: 4px solid #007bff;
24-
padding: 10px;
25-
margin: 10px 0;
51+
.new-feature-card h3 {
52+
margin-top: 0;
53+
margin-bottom: 15px;
54+
font-size: 1.3em;
55+
font-weight: 600;
56+
}
57+
58+
.new-feature-card.exclusive h3 {
59+
color: white;
60+
}
61+
62+
.new-feature-card ul {
63+
margin: 15px 0;
64+
padding-left: 20px;
65+
}
66+
67+
.new-feature-card li {
68+
margin-bottom: 8px;
69+
line-height: 1.5;
70+
}
71+
72+
.new-feature-card code {
73+
background: rgba(0, 0, 0, 0.1);
74+
padding: 2px 6px;
75+
border-radius: 4px;
76+
font-size: 0.9em;
77+
}
78+
79+
.new-feature-card.exclusive code {
80+
background: rgba(255, 255, 255, 0.2);
81+
color: white;
82+
}
83+
84+
/* Collapse navigation tree items by default */
85+
.bd-toc .toctree-l1 > ul {
86+
display: none;
87+
}
88+
89+
.bd-toc .toctree-l1.current > ul {
90+
display: block;
91+
}
92+
93+
.bd-toc .toctree-l2 > ul {
94+
display: none;
95+
}
96+
97+
.bd-toc .toctree-l2.current > ul {
98+
display: block;
99+
}
100+
101+
/* Highlight model customization in navigation */
102+
.bd-toc a[href*="model_customization"],
103+
.toctree-l1 a[href*="model_customization"] {
104+
background: linear-gradient(90deg, rgba(255, 107, 107, 0.15), rgba(78, 205, 196, 0.15));
105+
border-left: 4px solid #ff6b6b;
106+
padding-left: 12px;
107+
font-weight: 600;
108+
border-radius: 4px;
109+
margin: 2px 0;
110+
display: block;
111+
}
112+
113+
.bd-toc a[href*="model_customization"]:before,
114+
.toctree-l1 a[href*="model_customization"]:before {
115+
content: "⭐ ";
116+
color: #ff4757;
117+
font-size: 0.9em;
118+
}
119+
120+
header {
121+
background-color: white;
122+
123+
box-shadow: 0 1px 2px rgba(0, 0, 0, 0.05);
124+
position: sticky;
125+
top: 0;
126+
z-index: 50;
26127
}
27128

28-
/* Typography */
29129
h1 {
30130
font-size: 1.875rem;
31131
font-weight: 700;
@@ -47,4 +147,4 @@ h3 {
47147
p {
48148
font-size: 1.0rem;
49149
color: #4b5563;
50-
}
150+
}

‎docs/conf.py‎

Lines changed: 4 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -34,6 +34,10 @@
3434
'announcement': 'This is V3 documentation. <a href="https://sagemaker.readthedocs.io/en/v2/">View V2 docs</a>',
3535
}
3636

37+
html_theme_options = {
38+
"collapse_navigation": True,
39+
}
40+
3741
html_static_path = ['_static']
3842
html_css_files = ['custom.css']
3943

‎docs/index.rst‎

Lines changed: 2 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -14,12 +14,12 @@ SageMaker Python SDK V3
1414

1515
.. toctree::
1616
:maxdepth: 2
17-
:caption: Core Capabilities
17+
:caption: V3 Capabilities
1818
:hidden:
1919

20+
model_customization/index
2021
training/index
2122
inference/index
22-
model_customization/index
2323
ml_ops/index
2424
sagemaker_core/index
2525

‎docs/model_customization/index.rst‎

Lines changed: 8 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -1,6 +1,14 @@
11
Model Customization
22
===================
33

4+
.. raw:: html
5+
6+
<div class="v3-exclusive-feature">
7+
<strong>🆕 V3 EXCLUSIVE FEATURE</strong><br>
8+
Model customization with specialized trainers is only available in SageMaker Python SDK V3.
9+
This powerful capability was built from the ground up for foundation model fine-tuning.
10+
</div>
11+
412
SageMaker Python SDK V3 revolutionizes foundation model fine-tuning with specialized trainer classes, making it easier than ever to customize large language models and foundation models for your specific use cases. This modern approach provides powerful fine-tuning capabilities while maintaining simplicity and performance.
513

614
Key Benefits of V3 Model Customization

‎docs/overview.rst‎

Lines changed: 46 additions & 40 deletions
Original file line numberDiff line numberDiff line change
@@ -6,25 +6,52 @@ Welcome to SageMaker Python SDK V3 - a revolutionary approach to machine learnin
66
What's New in V3
77
-----------------
88

9-
**Modular Architecture**
10-
Separate PyPI packages for specialized capabilities:
11-
12-
* ``sagemaker-core`` - Low-level SageMaker resource management
13-
* ``sagemaker-train`` - Unified training with ModelTrainer
14-
* ``sagemaker-serve`` - Simplified inference with ModelBuilder
15-
* ``sagemaker-mlops`` - ML operations and pipeline management
16-
17-
**Unified Classes**
18-
Single classes replace multiple framework-specific implementations:
19-
20-
* **ModelTrainer** replaces PyTorchEstimator, TensorFlowEstimator, SKLearnEstimator, etc.
21-
* **ModelBuilder** replaces PyTorchModel, TensorFlowModel, SKLearnModel, etc.
22-
23-
**Object-Oriented API**
24-
Structured interface with auto-generated configs aligned with AWS APIs for better developer experience.
25-
26-
Core Capabilities
27-
==================
9+
.. raw:: html
10+
11+
<div class="whats-new-container">
12+
<div class="new-feature-card exclusive">
13+
<div class="feature-icon">🎯</div>
14+
<h3>Model Customization (V3 Exclusive)</h3>
15+
<p>Revolutionary foundation model fine-tuning with specialized trainers:</p>
16+
<ul>
17+
<li><strong>SFTTrainer</strong> - Supervised fine-tuning for task-specific adaptation</li>
18+
<li><strong>DPOTrainer</strong> - Direct preference optimization without RL complexity</li>
19+
<li><strong>RLAIFTrainer</strong> - Reinforcement learning from AI feedback</li>
20+
<li><strong>RLVRTrainer</strong> - Reinforcement learning from verifiable rewards</li>
21+
</ul>
22+
<p><em>Advanced techniques like LoRA, preference optimization, and RLHF that simply don't exist in V2.</em></p>
23+
</div>
24+
25+
<div class="new-feature-card">
26+
<div class="feature-icon">📦</div>
27+
<h3>Modular Architecture</h3>
28+
<p>Separate PyPI packages for specialized capabilities:</p>
29+
<ul>
30+
<li><code>sagemaker-core</code> - Low-level SageMaker resource management</li>
31+
<li><code>sagemaker-train</code> - Unified training with ModelTrainer</li>
32+
<li><code>sagemaker-serve</code> - Simplified inference with ModelBuilder</li>
33+
<li><code>sagemaker-mlops</code> - ML operations and pipeline management</li>
34+
</ul>
35+
</div>
36+
37+
<div class="new-feature-card">
38+
<div class="feature-icon">🔧</div>
39+
<h3>Unified Classes</h3>
40+
<p>Single classes replace multiple framework-specific implementations:</p>
41+
<ul>
42+
<li><strong>ModelTrainer</strong> replaces PyTorchEstimator, TensorFlowEstimator, SKLearnEstimator, etc.</li>
43+
<li><strong>ModelBuilder</strong> replaces PyTorchModel, TensorFlowModel, SKLearnModel, etc.</li>
44+
</ul>
45+
</div>
46+
47+
<div class="new-feature-card">
48+
<div class="feature-icon">⚡</div>
49+
<h3>Object-Oriented API</h3>
50+
<p>Structured interface with auto-generated configs aligned with AWS APIs for better developer experience.</p>
51+
</div>
52+
</div>
53+
54+
Capabilities
2855

2956
Training with ModelTrainer
3057
---------------------------
@@ -69,27 +96,6 @@ Simplified model deployment and inference with automatic optimization and flexib
6996
7097
:doc:`Learn more about Inference <inference/index>`
7198

72-
Model Customization
73-
--------------------
74-
75-
Advanced foundation model fine-tuning with specialized trainer classes for cutting-edge techniques:
76-
77-
.. code-block:: python
78-
79-
from sagemaker.train import SFTTrainer
80-
from sagemaker.train.common import TrainingType
81-
82-
trainer = SFTTrainer(
83-
model="meta-llama/Llama-2-7b-hf",
84-
training_type=TrainingType.LORA,
85-
model_package_group_name="my-custom-models",
86-
training_dataset="s3://my-bucket/training-data.jsonl"
87-
)
88-
89-
training_job = trainer.train()
90-
91-
:doc:`Learn more about Model Customization <model_customization/index>`
92-
9399
ML Operations
94100
-------------
95101

0 commit comments

Comments
 (0)