Intel Model Zoo v2.0.0
New Functionality
Intel® Model Zoo 2.0 introduces tools and quickstart directories that support modular creation and distribution of containers and model packages for the zoo's most popular workloads. This version also re-releases several models compatible with TensorFlow v1.15.2 and is fully backward compatible with the previous usage paradigm (the launch_benchmark.py script) for launching workloads in docker and on bare metal. Future releases will feature more content under the quickstart
directory which will also be available for download, with full documentation, on the Intel® oneContainer Portal.
All pre-trained models are now available for download from AWS S3 storage with same old bucket name intel-optimized-tensorflow
.
So for example the command:
$ wget https://storage.googleapis.com/intel-optimized-tensorflow/models/v1_8/densenet169_fp32_pretrained_model.pb
should be replaced with the following to download from AWS S3 bucket:
$ wget https://intel-optimized-tensorflow.s3.cn-north-1.amazonaws.com.cn/models/v1_8/densenet169_fp32_pretrained_model.pb
New Topologies and Models (TensorFlow 1.15.2)
- UNet
- Faster R-CNN
- Mask R-CNN
- NCF
- WaveNet
DL Frameworks (TensorFlow)
TensorFlow models in the 2.0 release are validated on the following TensorFlow versions:
- Intel Optimizations for TensorFlow v2.3.0 or v1.15.2 (select models)
- Intel Optimizations for TensorFlow serving v2.3.0
DL Frameworks (PyTorch)
PyTorch models in the 2.0 release are validated on the following PyTorch version:
- PyTorch v1.5.0-rc3
Supported Configurations
Intel Model Zoo 2.0 is validated on the following environment:
- Ubuntu 18.04 LTS
- Python 3.6
- Docker Server v19+
- Docker Client v18+