Rename default branch to main and update hyperlinks

Signed-off-by: Rajeev Rao <[email protected]>

CHANGELOG.md

@@ -125,7 +125,7 @@ Identical to the TensorRT-OSS [8.0.1](https://github.com/NVIDIA/TensorRT/release
   - Added support for per-axis quantization.
 - Added `EfficientNMS_TRT`, `EfficientNMS_ONNX_TRT` plugins and experimental support for ONNX `NonMaxSuppression` operator.
 - Added `ScatterND` plugin.
-- Added TensorRT [QuickStart Guide](https://github.com/NVIDIA/TensorRT/tree/master/quickstart).
+- Added TensorRT [QuickStart Guide](https://github.com/NVIDIA/TensorRT/tree/main/quickstart).
 - Added new samples: [engine_refit_onnx_bidaf](https://docs.nvidia.com/deeplearning/tensorrt/sample-support-guide/index.html#engine_refit_onnx_bidaf) builds an engine from ONNX BiDAF model and refits engine with new weights, [efficientdet](samples/python/efficientdet) and [efficientnet](samples/python/efficientnet) samples for demonstrating Object Detection using TensorRT.
 - Added support for Ubuntu20.04 and RedHat/CentOS 8.3.
 - Added Python 3.9 support.
@@ -264,21 +264,21 @@ Identical to the TensorRT-OSS [8.0.1](https://github.com/NVIDIA/TensorRT/release
 
 ## [20.11](https://github.com/NVIDIA/TensorRT/releases/tag/20.11) - 2020-11-20
 ### Added
-- API documentation for [ONNX-GraphSurgeon](https://github.com/NVIDIA/TensorRT/tree/master/tools/onnx-graphsurgeon/docs)
+- API documentation for [ONNX-GraphSurgeon](https://github.com/NVIDIA/TensorRT/tree/main/tools/onnx-graphsurgeon/docs)
 
 ### Changed
-- Support for SM86 in [demoBERT](https://github.com/NVIDIA/TensorRT/tree/master/demo/BERT)
-- Updated NGC checkpoint URLs for [demoBERT](https://github.com/NVIDIA/TensorRT/tree/master/demo/BERT) and [Tacotron2](https://github.com/NVIDIA/TensorRT/tree/master/demo/Tacotron2).
+- Support for SM86 in [demoBERT](https://github.com/NVIDIA/TensorRT/tree/main/demo/BERT)
+- Updated NGC checkpoint URLs for [demoBERT](https://github.com/NVIDIA/TensorRT/tree/main/demo/BERT) and [Tacotron2](https://github.com/NVIDIA/TensorRT/tree/main/demo/Tacotron2).
 
 ### Removed
 - N/A
 
 
 ## [20.10](https://github.com/NVIDIA/TensorRT/releases/tag/20.10) - 2020-10-22
 ### Added
-- [Polygraphy](https://github.com/NVIDIA/TensorRT/tree/master/tools/Polygraphy) v0.20.13 - Deep Learning Inference Prototyping and Debugging Toolkit
-- [PyTorch-Quantization Toolkit](https://github.com/NVIDIA/TensorRT/tree/master/tools/pytorch-quantization) v2.0.0
-- Updated BERT plugins for [variable sequence length inputs](https://github.com/NVIDIA/TensorRT/tree/master/demo/BERT#variable-sequence-length)
+- [Polygraphy](https://github.com/NVIDIA/TensorRT/tree/main/tools/Polygraphy) v0.20.13 - Deep Learning Inference Prototyping and Debugging Toolkit
+- [PyTorch-Quantization Toolkit](https://github.com/NVIDIA/TensorRT/tree/main/tools/pytorch-quantization) v2.0.0
+- Updated BERT plugins for [variable sequence length inputs](https://github.com/NVIDIA/TensorRT/tree/main/demo/BERT#variable-sequence-length)
 - Optimized kernels for sequence lengths of 64 and 96 added
 - Added Tacotron2 + Waveglow TTS demo [#677](https://github.com/NVIDIA/TensorRT/pull/677)
 - Re-enable `GridAnchorRect_TRT` plugin with rectangular feature maps [#679](https://github.com/NVIDIA/TensorRT/pull/679)

demo/BERT/README.md

@@ -354,7 +354,7 @@ Note this is an experimental feature because we only support Xavier+ GPUs, also
 
 Fine-grained 2:4 structured sparsity support introduced in NVIDIA Ampere GPUs can produce significant performance gains in BERT inference. The network is first trained using dense weights, then fine-grained structured pruning is applied, and finally the remaining non-zero weights are fine-tuned with additional training steps. This method results in virtually no loss in inferencing accuracy.
 
-Using INT8 precision with quantization scales obtained from Post-Training Quantization (PTQ) can produce additional performance gains, but may also result in accuracy loss. Alternatively, for PyTorch-trained models, NVIDIA [PyTorch-Quantization toolkit](https://github.com/NVIDIA/TensorRT/tree/master/tools/pytorch-quantization) can be leveraged to perform quantized fine tuning (a.k.a. Quantization Aware Training or QAT) and generate the INT8 quantization scales as part of training. This generally results in higher accuracy compared to PTQ.
+Using INT8 precision with quantization scales obtained from Post-Training Quantization (PTQ) can produce additional performance gains, but may also result in accuracy loss. Alternatively, for PyTorch-trained models, NVIDIA [PyTorch-Quantization toolkit](https://github.com/NVIDIA/TensorRT/tree/main/tools/pytorch-quantization) can be leveraged to perform quantized fine tuning (a.k.a. Quantization Aware Training or QAT) and generate the INT8 quantization scales as part of training. This generally results in higher accuracy compared to PTQ.
 
 To demonstrate the potential speedups from these optimizations in demoBERT, we provide the [Megatron-LM](https://github.com/NVIDIA/Megatron-LM) transformer model finetuned for SQuAD 2.0 task with sparsity and quantization.
 

quickstart/IntroNotebooks/2. Using the Tensorflow TensorRT Integration.ipynb

@@ -46,7 +46,7 @@
      "name": "stdout",
      "output_type": "stream",
      "text": [
-      "--2021-01-29 23:37:25--  https://raw.githubusercontent.com/NVIDIA/TensorRT/master/quickstart/IntroNotebooks/helper.py\n",
+      "--2021-01-29 23:37:25--  https://raw.githubusercontent.com/NVIDIA/TensorRT/main/quickstart/IntroNotebooks/helper.py\n",
       "Resolving raw.githubusercontent.com (raw.githubusercontent.com)... 151.101.40.133\n",
       "Connecting to raw.githubusercontent.com (raw.githubusercontent.com)|151.101.40.133|:443... connected.\n",
       "HTTP request sent, awaiting response... 404 Not Found\n",
@@ -56,7 +56,7 @@
     }
    ],
    "source": [
-    "!wget \"https://raw.githubusercontent.com/NVIDIA/TensorRT/master/quickstart/IntroNotebooks/helper.py\""
+    "!wget \"https://raw.githubusercontent.com/NVIDIA/TensorRT/main/quickstart/IntroNotebooks/helper.py\""
    ]
   },
   {

quickstart/IntroNotebooks/3. Using Tensorflow 2 through ONNX.ipynb

@@ -1241,7 +1241,7 @@
     "\n",
     "#### TRT Supported Layers:\n",
     "\n",
-    "https://github.com/NVIDIA/TensorRT/tree/master/samples/opensource/samplePlugin\n",
+    "https://github.com/NVIDIA/TensorRT/tree/main/samples/opensource/samplePlugin\n",
     "\n",
     "#### TRT ONNX Plugin Example:\n",
     "\n",

quickstart/IntroNotebooks/4. Using PyTorch through ONNX.ipynb

@@ -960,7 +960,7 @@
     "\n",
     "#### TRT Supported Layers:\n",
     "\n",
-    "https://github.com/NVIDIA/TensorRT/tree/master/samples/opensource/samplePlugin\n",
+    "https://github.com/NVIDIA/TensorRT/tree/main/samples/opensource/samplePlugin\n",
     "\n",
     "#### TRT ONNX Plugin Example:\n",
     "\n",

samples/python/efficientdet/README.md

@@ -167,7 +167,7 @@ Optionally, you may wish to visualize the resulting ONNX graph with a tool such
 
 The input to the graph is a `float32` tensor with the selected input shape, containing RGB pixel data in the range of 0 to 255. Normalization, mean subtraction and scaling will be performed inside the EfficientDet graph, so it is not required to further pre-process the input data.
 
-The outputs of the graph are the same as the outputs of the [EfficientNMS](https://github.com/NVIDIA/TensorRT/tree/master/plugin/efficientNMSPlugin) plugin. If the ONNX graph was created with `--legacy_plugins` for TensorRT 7 compatibility, the outputs will correspond to those of the [BatchedNMS](https://github.com/NVIDIA/TensorRT/tree/master/plugin/batchedNMSPlugin) plugin instead. 
+The outputs of the graph are the same as the outputs of the [EfficientNMS](https://github.com/NVIDIA/TensorRT/tree/main/plugin/efficientNMSPlugin) plugin. If the ONNX graph was created with `--legacy_plugins` for TensorRT 7 compatibility, the outputs will correspond to those of the [BatchedNMS](https://github.com/NVIDIA/TensorRT/tree/main/plugin/batchedNMSPlugin) plugin instead. 
 
 ### Build TensorRT Engine
 
@@ -281,4 +281,4 @@ This script will process the images found in the given input path through both T
 
 If you run this on COCO val2017 images, you may also add the parameter `--annotations /path/to/coco/annotations/instances_val2017.json` to further compare against COCO ground truth annotations.
 
-![compare_tf](https://drive.google.com/uc?export=view&id=1zgh_RbYX6RWzu7nKLCcSzy60VPiQROZJ)
+![compare_tf](https://drive.google.com/uc?export=view&id=1zgh_RbYX6RWzu7nKLCcSzy60VPiQROZJ)

samples/python/engine_refit_onnx_bidaf/README.md

@@ -28,7 +28,7 @@ Dependencies required for this sample
 
 2. TensorRT
 
-3. [ONNX-GraphSurgeon](https://github.com/NVIDIA/TensorRT/tree/master/tools/onnx-graphsurgeon)
+3. [ONNX-GraphSurgeon](https://github.com/NVIDIA/TensorRT/tree/main/tools/onnx-graphsurgeon)
 
 4. Download sample data. See the "Download Sample Data" section of [the general setup guide](../README.md).
 

samples/sampleOnnxMnistCoordConvAC/README.md

@@ -97,7 +97,7 @@ hostDataBuffer[i] = ((1.0 - float(fileData[i] / 255.0)) - PYTORCH_NORMALIZE_MEAN
 
 In this sample, the following layers and plugins are used. For more information about these layers, see the [TensorRT Developer Guide: Layers](https://docs.nvidia.com/deeplearning/sdk/tensorrt-developer-guide/index.html#layers) documentation.
 
-[CoordConvAC layer](https://github.com/NVIDIA/TensorRT/tree/master/plugin/coordConvACPlugin)
+[CoordConvAC layer](https://github.com/NVIDIA/TensorRT/tree/main/plugin/coordConvACPlugin)
 Custom layer implemented with CUDA API that implements operation AddChannels. This layer expands the input data by adding additional channels with relative coordinates.
 
 [Activation layer](https://docs.nvidia.com/deeplearning/sdk/tensorrt-developer-guide/index.html#activation-layer)
@@ -214,7 +214,7 @@ The following resources provide a deeper understanding about the ONNX project an
 **CoordConv Layer**
 - [Arxiv paper by Uber AI Labs](https://arxiv.org/abs/1807.03247)
 - [Blog post about the CoordConv layer](https://eng.uber.com/coordconv/)
-- [Path to the layer's plugin in repository](https://github.com/NVIDIA/TensorRT/tree/master/plugin/coordConvACPlugin)
+- [Path to the layer's plugin in repository](https://github.com/NVIDIA/TensorRT/tree/main/plugin/coordConvACPlugin)
 
 **ONNX**
 - [GitHub: ONNX](https://github.com/onnx/onnx)

samples/sampleUffMaskRCNN/converted/README.md

@@ -101,7 +101,7 @@ shape=[config.IMAGE_SHAPE[2], 1024, 1024 ], name="input_image")
         self.keras_model.predict([molded_input_images, image_metas, anchors], verbose=0)
     mrcnn_mask = np.transpose(mrcnn_mask, (0, 1, 3, 4, 2))
     ```
-- For conversion to UFF, please refer to [these instructions](https://github.com/NVIDIA/TensorRT/tree/master/samples/opensource/sampleUffMaskRCNN#generating-uff-model).
+- For conversion to UFF, please refer to [these instructions](https://github.com/NVIDIA/TensorRT/tree/main/samples/opensource/sampleUffMaskRCNN#generating-uff-model).
 
 >  NOTE: For reference, the successful converted model should contain 3049 nodes.
 

tools/Polygraphy/docs/index.rst

@@ -6,10 +6,10 @@ This page includes the Python API documentation for Polygraphy. Polygraphy is a
 designed to assist in running and debugging deep learning models in various frameworks.
 
 For installation instructions, examples, and information about the CLI tools,
-see `the GitHub repository <https://github.com/NVIDIA/TensorRT/tree/master/tools/Polygraphy>`_ instead.
+see `the GitHub repository <https://github.com/NVIDIA/TensorRT/tree/main/tools/Polygraphy>`_ instead.
 
 For a high level overview of the Python API,
-see `this page <https://github.com/NVIDIA/TensorRT/tree/master/tools/Polygraphy/polygraphy>`_.
+see `this page <https://github.com/NVIDIA/TensorRT/tree/main/tools/Polygraphy/polygraphy>`_.
 
 .. toctree::
     :hidden:

tools/Polygraphy/polygraphy/tools/surgeon/README.md

@@ -10,7 +10,7 @@
 
 ## Introduction
 
-The `surgeon` tool uses [ONNX-GraphSurgeon](https://github.com/NVIDIA/TensorRT/tree/master/tools/onnx-graphsurgeon)
+The `surgeon` tool uses [ONNX-GraphSurgeon](https://github.com/NVIDIA/TensorRT/tree/main/tools/onnx-graphsurgeon)
 to modify an ONNX model.
 
 

tools/Polygraphy/setup.py

@@ -41,7 +41,7 @@ def main():
         version=polygraphy.__version__,
         description="Polygraphy: A Deep Learning Inference Prototyping and Debugging Toolkit",
         long_description=open("README.md", "r", encoding="utf-8").read(),
-        url="https://github.com/NVIDIA/TensorRT/tree/master/tools/Polygraphy",
+        url="https://github.com/NVIDIA/TensorRT/tree/main/tools/Polygraphy",
         author="NVIDIA",
         author_email="[email protected]",
         classifiers=[

tools/onnx-graphsurgeon/docs/index.rst

@@ -6,7 +6,7 @@ This page includes the Python API documentation for ONNX GraphSurgeon. ONNX Grap
 provides a convenient way to create and modify ONNX models.
 
 For installation instructions and examples see
-`this page <https://github.com/NVIDIA/TensorRT/tree/master/tools/onnx-graphsurgeon>`_ instead.
+`this page <https://github.com/NVIDIA/TensorRT/tree/main/tools/onnx-graphsurgeon>`_ instead.
 
 
 .. toctree::