DALI Release 0.3 Beta
Key Features and Enhancements
This DALI release includes the following key features and
enhancements.
- Updated PyTorch ResNet-50 example to obtain expected accuracy (Top1 76%).
- Introduced CPU variant of resize operator and added stand-alone flip operator.
- Added support for DALI to work with float16 data passed from Python.
- Added fallback to CPU for BMP images.
- Fixed training accuracy with TFRecord reader
Using DALI 0.3 Beta
The 18.08 NVIDIA GPU Cloud (NGC) optimized container for MXNet, PyTorch, and TensorFlow, includes an older version of DALI. To upgrade to DALI 0.2 beta, follow the installation instructions in the DALI Quick Start Guide.
Refer to the DALI Developer Guide for usage details.
Known Issues
- HosDecoder cannot handle all jpeg files from ImageNet dataset. This will be fixed in the next release.
- On file systems where the directory entries are not stored in any order, the File Reader may assign different labels to the training and validation folders with the same name. This will be fixed in the next release.
- The DALI integrated ResNet-50 samples in the 18.09 NGC TensorFlow and PyTorch containers may result in lower than expected performance results. We are working to address the issue in the next release.
- This is a beta release. All features are expected to be available, however, some aspects of functionality and performance will likely be limited compared to a non-beta release.