Ai framework page

.vitepress/sidebar.ts

@@ -59,6 +59,48 @@ export function getSidebar() {
           },
         ],
       },
+      {
+        text: '🧠 &nbsp;Resources for AI <span class="VPBadge warning" style="margin-left: 8px; margin-bottom: -1px; transform: translateY(-1px);">beta</span>',
+        link: '/tools/ai-resources',
+        collapsed: false,
+        items: [
+          {
+            text: 'Getting Started',
+            link: '/tools/ai-resources/getting-started',
+          },
+          {
+            text: 'Supported AI frameworks',
+            link: '/tools/ai-resources/frameworks/',
+            collapsed: true,
+            items: [
+              {
+                text: 'scikit-learn',
+                link: '/tools/ai-resources/frameworks/scikit-learn',
+              },
+              {
+                text: 'TensorFlow',
+                link: '/tools/ai-resources/frameworks/TensorFlow',
+              },
+              {
+                text: 'PyTorch',
+                link: '/tools/ai-resources/frameworks/PyTorch',
+              },
+              {
+                text: 'OpenVINO',
+                link: '/tools/ai-resources/frameworks/OpenVINO',
+              },
+            ],
+          },
+          {
+            text: 'iApp Generator templates',
+            link: '/tools/ai-resources/iAppGenerator',
+          },
+          {
+            text: 'Intel TDX Workers',
+            link: '/tools/ai-resources/tdx',
+          },
+        ]
+      },
       {
         text: '🔐 &nbsp;DataProtector <span class="VPBadge warning" style="margin-left: 8px; margin-bottom: -1px; transform: translateY(-1px);">beta</span>',
         link: '/tools/dataProtector',

tools/ai-resources.md

@@ -0,0 +1,57 @@
+# 🧠 iExec for AI
+
+
+The iExec Platforms delivers several tools specifically for AI developers. Here’s how iExec can help develop, deploy and execute confidential AI applications.
+
+
+
+## Confidential AI & Secure Compute Services
+
+iExec’s Confidential AI framework leverages Trusted Execution Environments (TEEs) such as Intel SGX and Intel TDX. This ensures:
+
+* Data privacy: AI computations are protected end-to-end.
+* Secure AI training & inference: models & data are encrypted throughout processing; they can never be accessed by unauthorized entities.
+
+
+## AI Monetization
+
+iExec provides all the tools to easily monetize your datasets, your models, your agents and your AI applications; encrypt it, push it, set a price and you’re done! The iExec Platform preserves the ownership of your digital assets: whatever
+happens, your data, models and agents will always be yours!
+
+
+## Decentralized & Scalable Computing
+
+iExec provides AI builders with a decentralized cloud computing infrastructure where they can access powerful
+compute resources on-demand. Scale AI applications seamlessly, without dependency on centralized cloud providers
+and ensure fair & transparent execution via blockchain-based verification.
+
+
+## AI Agent & GenAI Trust Layer
+iExec integrates automation tools such as the iApp Generator, allowing AI developers to:
+
+* Easily create trusted AI agents.
+* Leverage secure off-chain AI execution while integrating with smart contracts.
+
+
+## Web3 & Decentralized AI (DeAI) Integration
+
+iExec is positioned at the intersection of DePIN (Decentralized Physical Infrastructure Networks) and DeAI (Decentralized AI), allowing AI applications to:
+
+* Operate transparently in a trust-driven ecosystem.
+* Utilize blockchain-based consensus for AI validation.
+
+
+## Flexible Compute Resource Access
+
+Through worker pools, AI developers can:
+
+* Run AI workloads on decentralized hardware, including CPUs and GPUs.
+* Access specialized computing resources, such as confidential GPUs for AI/ML model training.
+
+## Interoperability & Compatibility
+
+iExec supports:
+
+* Off-chain AI computation triggered by smart contracts.
+* Seamless execution of existing AI workloads using Intel TDX, eliminating the need for application modification.
+

tools/ai-resources/frameworks/OpenVINO.md

@@ -0,0 +1 @@
+# OpenVINO on iExec

tools/ai-resources/frameworks/PyTorch.md

@@ -0,0 +1,22 @@
+# PyTorch on iExec
+
+PyTorch is an open-source machine learning framework originally developed by Meta AI and now
+part of the Linux Foundation. PyTorch provides flexible and efficient tools for Tensor computing
+and building and training neural networks that made it one of the most popular deep learning
+frameworks.
+
+## Compatibility
+
+The table gives the PyTorch versions that have been tested on the iExec Platform, on the
+SGX-based workerpools and on the experimental TDX workerpool.
+
+|                | **v. 2.4**     | **v. 1.12**     |
+| -------------- | -------------- | --------------- |
+| **Scone-SGX**  |                | ❌ Unsupported |
+| **TDX**        | ✅ Supported   |                |
+
+
+## Instructions
+
+
+## Troubleshooting

tools/ai-resources/frameworks/TensorFlow.md

@@ -0,0 +1 @@
+# TensorFlow on iExec

tools/ai-resources/frameworks/index.md

@@ -0,0 +1,11 @@
+# Supported AI frameworks
+
+The iExec Platform supports the major runtime frameworks for AI; this page list those and provide important guidance on building and executing iApps based on them.
+
+
+| Framework    | SGX workerpool     | TDX workerpool |
+| ------------ | ------------------ | -------------- |
+| PyTorch      | ⚠️ Partial support | ✅ Supported  |
+| scikit-learn | ❌ Unsupported     | ✅ Supported  |
+| Tensorflow   |                    |                |
+| OpenVINO     |                    |                |

tools/ai-resources/frameworks/scikit-learn.md

@@ -0,0 +1,31 @@
+# scikit-learn on iExec
+
+[Scikit-learn](https://scikit-learn.org/) is an open-source Python library that provides simple and efficient tools for data mining and machine learning, built on top of NumPy, SciPy, and matplotlib. It offers a wide range of algorithms for classification, regression, clustering, dimensionality reduction, and model selection.
+
+## Compatibility
+
+The table gives the scikit-learn versions that have been tested on the iExec Platform, on the
+SGX-based workerpools and on the experimental TDX workerpool.
+
+|                | **v. 1.6.1**   | **v. 0.19**     |
+| -------------- | -------------- | --------------- |
+| **Scone-SGX**  | ❌ Unsupported | ❌ Unsupported |
+| **TDX**        | ✅ Supported   |                |
+
+
+## Instructions
+
+When building a Docker image with scikit-learn, some low-level optimisations must be disabled or
+they can lead to errors when the container runs on a slightly different architecture.
+
+
+```dockerfile
+
+RUN pip install ...
+
+```
+
+
+## Troubleshooting
+
+

tools/ai-resources/getting-started.md

@@ -0,0 +1,24 @@
+# 🧠 Getting started with AI on iExec
+
+## Use-cases
+
+
+### Protecting models
+
+
+### Protecting input data
+
+
+### Protecting results
+
+
+### Confidential & Trusted AI Agents
+
+
+## AI Frameworks
+
+
+## iApp Generator
+
+
+## Intel TXD

tools/ai-resources/iAppGenerator.md

@@ -0,0 +1 @@
+# iApp Generator for AI

tools/ai-resources/tdx.md

@@ -0,0 +1 @@
+# TDX workers