Intel AI Containers
February 7, 2024 · View on GitHub
This page provides overview of AI Containers for many Intel optimized open-source frameworks such as PyTorch, TensorFlow, SciKit Learn, XGBoost, Modin etc. Also provides containers for open-source deep learning models optimized by Intel to run on Intel® Xeon® Scalable processors and Intel® Data Center GPUs.
Intel® Distribution of Python
Docker container with Intel® Distribution of Python
| Container | Framework | Docker Pull Command | Compressed Size |
|---|---|---|---|
| Intel® Distribution of Python - Full | - | docker pull intel/python:py310-full | 1.38GB |
| Intel® Distribution of Python - Core | - | docker pull intel/python:py310-core | 814.77MB |
Intel® CPU Base Containers
Base containers to run on Intel® Xeon® Scalable processors
| Container | Framework | Docker Pull Command | Compressed Size |
|---|---|---|---|
| Intel® Optimizations for TensorFlow (PIP) | TensorFlow | docker pull intel/intel-optimized-tensorflow:2.14.0-pip-base | 405.53MB |
| Intel® Optimizations for TensorFlow with Jupyter notebook support (PIP) | TensorFlow | docker pull intel/intel-optimized-tensorflow:2.14.0-pip-jupyter | 464MB |
| Intel® Optimizations for TensorFlow Multi-node with OpenMPI (PIP) | TensorFlow | docker pull intel/intel-optimized-tensorflow:2.14.0-pip-openmpi-multinode | 817MB |
Intel® GPU Base Containers
Intel® Data Center GPU Max Series and Intel® Data Center GPU Flex Series base containers
| Container | Framework | Docker Pull Command | Compressed Size |
|---|---|---|---|
| Intel® Extension for PyTorch* Container for Flex Series | PyTorch | docker pull intel-extension-for-pytorch:xpu-flex | 7.04GB |
| Intel® Extension for PyTorch* Container for Max Series | PyTorch | docker pull intel-extension-for-pytorch:xpu-max | 7.13GB |
| Intel® Extension for TensorFlow | TensoFlow | docker pull intel/intel-extension-for-tensorflow:xpu | 2.29GB |
Intel® CPU Model Containers
These are the deep learning Models optimized to run on Intel® Xeon® Scalable processors
| Container | Framework | Docker Pull Command | Compressed Size |
|---|---|---|---|
| ResNet 50v1.5 Training | TensorFlow | docker pull intel/image-recognition:centos-tf-cpu-resnet50v1-5-training | 1.46GB |
| ResNet 50v1.5 Inference | TensorFlow | docker pull intel/image-recognition:centos-tf-cpu-resnet50v1-5-inference | 1.69GB |
| BERT large Training | TensorFlow | docker pull intel/language-modeling:centos-tf-cpu-bert-large-pretraining | 1.46GB |
Intel® Flex (ATSM) Model Containers
These are the deep learning Models optimized to run on Intel's Flex Series GPU platform.
| Container | Framework | Docker Pull Command | Compressed Size |
|---|---|---|---|
| ResNet50 v1.5 Inference | TensorFlow | docker pull intel/image-recognition:tf-flex-gpu-resnet50v1-5-inference | 2.35GB |
| MaskRCNN Inference | TensorFlow | docker pull intel/image-segmentation:tf-flex-gpu-maskrcnn-inference | 2.59GB |
| EfficientNet Inference | TensorFlow | docker pull intel/image-recognition:tf-flex-gpu-efficientnet-inference | 2.3GB |
Intel® Max (PVC) Model Containers
These are the deep learning Models optimized to run on Intel's Max Series GPU platform.
| Container | Framework | Docker Pull Command | Compressed Size |
|---|---|---|---|
| ResNet50 v1.5 Inference | PyTorch | docker pull intel/image-recognition:pytorch-max-gpu-resnet50v1-5-inference | 2.37GB |
| ResNet50 v1.5 Training | PyTorch | docker pull intel/image-recognition:pytorch-max-gpu-resnet50v1-5-training | 2.51GB |
| BERT Large Inference | PyTorch | docker pull intel/language-modeling:pytorch-max-gpu-bert-large-inference | 7.8GB |
Intel® AI Tools
These are Domain Toolkit Containers that use the latest transformers and LLM technology with Intel optimizations.
| Container | Framework | Docker Pull Command | Compressed Size |
|---|---|---|---|
| Intel® Transfer Learning Toolkit | PyTorch & TensorFlow | docker pull intel/ai-tools:tlt-0.5.0 | 910.64MB |
| Intel® Extension for Transformers Chatbot | PyTorch & TensorFlow | docker pull intel/ai-tools:itrex-chatbot | 4.54GB |
| Intel® Extension for Transformers | PyTorch & TensorFlow | docker pull intel/ai-tools:itrex-1.3.0 | 1.04GB |
Intel® AI Tools Selector Preset Containers
Intel® AI Tools Selector Preset Containers provides development environment with Intel optimized software for data scientists and developers.
| Container | Framework | Docker Pull Command | Compressed Size |
|---|---|---|---|
| Data Analytics py3.9 | - | docker pull intel/data-analytics:latest-py3.9 | 1.97GB |
| Data Analytics py3.10 | - | docker pull intel/data-analytics:latest-py3.10 | 1.98GB |
| Classical ML py3.9 | - | docker pull intel/classical-ml:latest-py3.9 | 2.11GB |
Intel® AI Workflows
These containers are for domain use-cases like Disease Prediction, Credit Card Fraud Detection, and Document Automation.
| Container | Framework | Docker Pull Command | Compressed Size |
|---|---|---|---|
| Document Automation Fine Tuning | - | docker pull intel/ai-workflows:doc-automation-fine-tuning | 2.53GB |
| Document Automation Indexing | - | docker pull intel/ai-workflows:doc-automation-indexing | 3.03GB |
| Fraud Detection GNN | - | docker pull intel/ai-workflows:pa-fraud-detection-gnn | 2.1GB |