- enable n-bit radix sort for NMS_TRT plugin
- implement NMSDynamic_TRT plugin and enable n-bit radix sort for it
- enable n-bit radix sort for BatchedNMS_TRT and BatchedNMSDynamic_TRT plugins
- fixed a bug in configurePlugin() method of BatchedNMS_TRT
- other minor fixes for BatchedNMS_TRT/BatchedNMSDynamic_TRT plugins
Signed-off-by: Rajeev Rao <rajeevrao@nvidia.com>
* Anuragd/trt custom plugin (#21)
* Added support for custom TRT plugin in trtexec
* Restructed as a wrapper API for different platforms
* Review comments incorporated
Signed-off-by: Anurag Dixit <a.dixit91@gmail.com>
* Added support for writing calibration file (#23)
Signed-off-by: Anurag Dixit <a.dixit91@gmail.com>
* BERT example using TensorRT C++ API (#32)
This example demonstrates how to write a TensorRT model with custom
plugins to implement the BERT encoder with SQuAD output layer.
Signed-off-by: Rajeev Rao <rajeevrao@nvidia.com>
* Resolve undefined symbols in NvInfer plugin library
Signed-off-by: Rajeev Rao <rajeevrao@nvidia.com>