add new model resnet50v1.5

scripts/sernxt_partial.sh

+FLAGS=$1
+STAGE_ID=$2
+STAGE_LEN=$3
+
+
+python ./multiproc.py  \
+    --nproc_per_node 8 \
+        ./main.py /imagenet  \
+            -j5 -p 100  \
+            --data-backend pytorch  \
+            --raport-file report_$STAGE_ID.json \
+            --lr 1.024  \
+            --batch-size 128 \
+            --optimizer-batch-size 1024  \
+            --static-loss-scale 128 \
+            --warmup 8  \
+            --arch se-resnext101-32x4d -c fanin  \
+            --label-smoothing 0.1  \
+            --lr-schedule cosine  \
+            --mom 0.875  \
+            --wd  6.103515625e-05  \
+            --workspace /results \
+            --epochs 90 \
+            --run-epochs $STAGE_LEN \
+            $FLAGS \
+            --resume /results/checkpoint_$( expr $STAGE_ID - 1).pth.tar \
+            --checkpoint checkpoint_$STAGE_ID.pth.tar
+

se-resnext101-32x4d/README.md

+# SE-ResNeXt101-32x4d For PyTorch
+
+This repository provides a script and recipe to train the SE-ResNeXt101-32x4d model to
+achieve state-of-the-art accuracy, and is tested and maintained by NVIDIA.
+
+## Table Of Contents
+
+* [Model overview](#model-overview)
+  * [Model architecture](#model-architecture)
+  * [Default configuration](#default-configuration)
+    * [Optimizer](#optimizer)
+    * [Data augmentation](#data-augmentation)
+  * [DALI](#dali)
+  * [Feature support matrix](#feature-support-matrix)
+    * [Features](#features)
+  * [Mixed precision training](#mixed-precision-training)
+    * [Enabling mixed precision](#enabling-mixed-precision)
+    * [Enabling TF32](#enabling-tf32)
+* [Setup](#setup)
+  * [Requirements](#requirements)
+* [Quick Start Guide](#quick-start-guide)
+* [Advanced](#advanced)
+  * [Scripts and sample code](#scripts-and-sample-code)
+  * [Command-line options](#command-line-options)
+  * [Dataset guidelines](#dataset-guidelines)
+  * [Training process](#training-process)
+  * [Inference process](#inference-process)
+* [Performance](#performance)
+  * [Benchmarking](#benchmarking)
+    * [Training performance benchmark](#training-performance-benchmark)
+    * [Inference performance benchmark](#inference-performance-benchmark)
+  * [Results](#results)
+    * [Training accuracy results](#training-accuracy-results)
+      * [Training accuracy: NVIDIA DGX A100 (8x A100 80GB)](#training-accuracy-nvidia-dgx-a100-8x-a100-80gb)
+      * [Training accuracy: NVIDIA DGX-1 (8x V100 16GB)](#training-accuracy-nvidia-dgx-1-8x-v100-16gb)
+      * [Example plots](#example-plots)
+    * [Training performance results](#training-performance-results)
+      * [Training performance: NVIDIA DGX A100 (8x A100 80GB)](#training-performance-nvidia-dgx-a100-8x-a100-80gb)
+      * [Training performance: NVIDIA DGX-1 16GB (8x V100 16GB)](#training-performance-nvidia-dgx-1-16gb-8x-v100-16gb)
+      * [Training performance: NVIDIA DGX-1 32GB (8x V100 32GB)](#training-performance-nvidia-dgx-1-32gb-8x-v100-32gb)
+  * [Inference performance results](#inference-performance-results)
+      * [Inference performance: NVIDIA DGX-1 16GB (1x V100 16GB)](#inference-performance-nvidia-dgx-1-1x-v100-16gb)
+      * [Inference performance: NVIDIA T4](#inference-performance-nvidia-t4)
+* [Release notes](#release-notes)
+  * [Changelog](#changelog)
+  * [Known issues](#known-issues)
+
+## Model overview
+
+The SE-ResNeXt101-32x4d is a [ResNeXt101-32x4d](https://arxiv.org/pdf/1611.05431.pdf)
+model with added Squeeze-and-Excitation module introduced
+in [Squeeze-and-Excitation Networks](https://arxiv.org/pdf/1709.01507.pdf) paper.
+
+Squeeze and Excitation module architecture for ResNet-type models:
+
+This model is trained with mixed precision using Tensor Cores on Volta, Turing, and the NVIDIA Ampere GPU architectures. Therefore, researchers can get results 3x faster than training without Tensor Cores, while experiencing the benefits of mixed precision training. This model is tested against each NGC monthly container release to ensure consistent accuracy and performance over time.
+
+We use [NHWC data layout](https://pytorch.org/tutorials/intermediate/memory_format_tutorial.html) when training using Mixed Precision.
+
+### Model architecture
+
+![SEArch](./img/SEArch.png)
+
+_Image source: [Squeeze-and-Excitation Networks](https://arxiv.org/pdf/1709.01507.pdf)_
+
+Image shows the architecture of SE block and where is it placed in ResNet bottleneck block.
+
+### Default configuration
+
+The following sections highlight the default configurations for the SE-ResNeXt101-32x4d model.
+
+#### Optimizer
+
+This model uses SGD with momentum optimizer with the following hyperparameters:
+
+* Momentum (0.875)
+* Learning rate (LR) = 0.256 for 256 batch size, for other batch sizes we linearly
+scale the learning rate.
+* Learning rate schedule - we use cosine LR schedule
+* For bigger batch sizes (512 and up) we use linear warmup of the learning rate
+during the first couple of epochs
+according to [Training ImageNet in 1 hour](https://arxiv.org/abs/1706.02677).
+Warmup length depends on the total training length.
+* Weight decay (WD)= 6.103515625e-05 (1/16384).
+* We do not apply WD on Batch Norm trainable parameters (gamma/bias)
+* Label smoothing = 0.1
+* We train for:
+    * 90 Epochs -> 90 epochs is a standard for ImageNet networks
+    * 250 Epochs -> best possible accuracy.
+* For 250 epoch training we also use [MixUp regularization](https://arxiv.org/pdf/1710.09412.pdf).
+
+
+#### Data augmentation
+
+This model uses the following data augmentation:
+
+* For training:
+  * Normalization
+  * Random resized crop to 224x224
+    * Scale from 8% to 100%
+    * Aspect ratio from 3/4 to 4/3
+  * Random horizontal flip
+* For inference:
+  * Normalization
+  * Scale to 256x256
+  * Center crop to 224x224
+
+
+### Feature support matrix
+
+The following features are supported by this model:
+
+| Feature               | SE-ResNeXt101-32x4d
+|-----------------------|--------------------------
+|[DALI](https://docs.nvidia.com/deeplearning/sdk/dali-release-notes/index.html)   |   Yes
+|[APEX AMP](https://nvidia.github.io/apex/amp.html) | Yes |
+
+#### Features
+
+- NVIDIA DALI - DALI is a library accelerating data preparation pipeline. To accelerate your input pipeline, you only need to define your data loader
+with the DALI library. For more information about DALI, refer to the [DALI product documentation](https://docs.nvidia.com/deeplearning/dali/user-guide/docs/index.html).
+
+- [APEX](https://github.com/NVIDIA/apex) is a PyTorch extension that contains utility libraries, such as [Automatic Mixed Precision (AMP)](https://nvidia.github.io/apex/amp.html), which require minimal network code changes to leverage Tensor Cores performance. Refer to the [Enabling mixed precision](#enabling-mixed-precision) section for more details.
+
+### DALI
+
+We use [NVIDIA DALI](https://github.com/NVIDIA/DALI),
+which speeds up data loading when CPU becomes a bottleneck.
+DALI can use CPU or GPU, and outperforms the PyTorch native dataloader.
+
+Run training with `--data-backends dali-gpu` or `--data-backends dali-cpu` to enable DALI.
+For DGXA100 and DGX1 we recommend `--data-backends dali-cpu`.
+
+### Mixed precision training
+
+Mixed precision is the combined use of different numerical precisions in a computational method. [Mixed precision](https://arxiv.org/abs/1710.03740) training offers significant computational speedup by performing operations in half-precision format, while storing minimal information in single-precision to retain as much information as possible in critical parts of the network. Since the introduction of [Tensor Cores](https://developer.nvidia.com/tensor-cores) in Volta, and following with both the Turing and Ampere architectures, significant training speedups are experienced by switching to mixed precision -- up to 3x overall speedup on the most arithmetically intense model architectures. Using mixed precision training requires two steps:
+1.  Porting the model to use the FP16 data type where appropriate.
+2.  Adding loss scaling to preserve small gradient values.
+
+The ability to train deep learning networks with lower precision was introduced in the Pascal architecture and first supported in CUDA 8 in the NVIDIA Deep Learning SDK.
+
+For information about:
+-   How to train using mixed precision, see the [Mixed Precision Training](https://arxiv.org/abs/1710.03740) paper and [Training With Mixed Precision](https://docs.nvidia.com/deeplearning/sdk/mixed-precision-training/index.html) documentation.
+-   Techniques used for mixed precision training, see the [Mixed-Precision Training of Deep Neural Networks](https://devblogs.nvidia.com/mixed-precision-training-deep-neural-networks/) blog.
+-   APEX tools for mixed precision training, see the [NVIDIA Apex: Tools for Easy Mixed-Precision Training in PyTorch](https://devblogs.nvidia.com/apex-pytorch-easy-mixed-precision-training/).
+
+#### Enabling mixed precision
+
+Mixed precision is enabled in PyTorch by using the Automatic Mixed Precision (AMP), a library from [APEX](https://github.com/NVIDIA/apex) that casts variables to half-precision upon retrieval,
+while storing variables in single-precision format. Furthermore, to preserve small gradient magnitudes in backpropagation, a [loss scaling](https://docs.nvidia.com/deeplearning/sdk/mixed-precision-training/index.html#lossscaling) step must be included when applying gradients.
+In PyTorch, loss scaling can be easily applied by using scale_loss() method provided by AMP. The scaling value to be used can be [dynamic](https://nvidia.github.io/apex/fp16_utils.html#apex.fp16_utils.DynamicLossScaler) or fixed.
+
+For an in-depth walk through on AMP, check out sample usage [here](https://github.com/NVIDIA/apex/tree/master/apex/amp#usage-and-getting-started). [APEX](https://github.com/NVIDIA/apex) is a PyTorch extension that contains utility libraries, such as AMP, which require minimal network code changes to leverage tensor cores performance.
+
+To enable mixed precision, you can:
+- Import AMP from APEX:
+
+  ```python
+  from apex import amp
+  ```
+
+- Wrap model and optimizer in amp.initialize:
+
+  ```python
+  model, optimizer = amp.initialize(model, optimizer, opt_level="O1", loss_scale="dynamic")
+  ```
+
+- Scale loss before backpropagation:
+  ```python
+  with amp.scale_loss(loss, optimizer) as scaled_loss:
+    scaled_loss.backward()
+  ```
+
+#### Enabling TF32
+
+TensorFloat-32 (TF32) is the new math mode in [NVIDIA A100](https://www.nvidia.com/en-us/data-center/a100/) GPUs for handling the matrix math also called tensor operations. TF32 running on Tensor Cores in A100 GPUs can provide up to 10x speedups compared to single-precision floating-point math (FP32) on Volta GPUs. 
+
+TF32 Tensor Cores can speed up networks using FP32, typically with no loss of accuracy. It is more robust than FP16 for models which require high dynamic range for weights or activations.
+
+For more information, refer to the [TensorFloat-32 in the A100 GPU Accelerates AI Training, HPC up to 20x](https://blogs.nvidia.com/blog/2020/05/14/tensorfloat-32-precision-format/) blog post.
+
+TF32 is supported in the NVIDIA Ampere GPU architecture and is enabled by default.
+
+
+## Setup
+
+The following section lists the requirements that you need to meet in order to start training the SE-ResNeXt101-32x4d model.
+
+### Requirements
+
+This repository contains Dockerfile which extends the PyTorch NGC container and encapsulates some dependencies. Aside from these dependencies, ensure you have the following components:
+
+* [NVIDIA Docker](https://github.com/NVIDIA/nvidia-docker)
+* [PyTorch 21.03-py3 NGC container](https://ngc.nvidia.com/registry/nvidia-pytorch) or newer
+* Supported GPUs:
+    * [NVIDIA Volta architecture](https://www.nvidia.com/en-us/data-center/volta-gpu-architecture/)
+    * [NVIDIA Turing architecture](https://www.nvidia.com/en-us/geforce/turing/)
+    * [NVIDIA Ampere architecture](https://www.nvidia.com/en-us/data-center/nvidia-ampere-gpu-architecture/)
+
+For more information about how to get started with NGC containers, see the
+following sections from the NVIDIA GPU Cloud Documentation and the Deep Learning
+DGX Documentation:
+* [Getting Started Using NVIDIA GPU Cloud](https://docs.nvidia.com/ngc/ngc-getting-started-guide/index.html)
+* [Accessing And Pulling From The NGC Container Registry](https://docs.nvidia.com/deeplearning/dgx/user-guide/index.html#accessing_registry)
+* [Running PyTorch](https://docs.nvidia.com/deeplearning/dgx/pytorch-release-notes/running.html#running)
+
+For those unable to use the PyTorch NGC container, to set up the required environment or create your own container, see the versioned [NVIDIA Container Support Matrix](https://docs.nvidia.com/deeplearning/frameworks/support-matrix/index.html).
+
+## Quick Start Guide
+
+### 1. Clone the repository.
+```
+git clone https://github.com/NVIDIA/DeepLearningExamples
+cd DeepLearningExamples/PyTorch/Classification/
+```
+
+### 2. Download and preprocess the dataset.
+
+The SE-ResNeXt101-32x4d script operates on ImageNet 1k, a widely popular image classification dataset from the ILSVRC challenge.
+
+PyTorch can work directly on JPEGs, therefore, preprocessing/augmentation is not needed.
+
+To train your model using mixed or TF32 precision with Tensor Cores or using FP32,
+perform the following steps using the default parameters of the se-resnext101-32x4d model on the ImageNet dataset.
+For the specifics concerning training and inference, see the [Advanced](#advanced) section.
+
+
+1. [Download the images](http://image-net.org/download-images).
+
+2. Extract the training data:
+  ```bash
+  mkdir train && mv ILSVRC2012_img_train.tar train/ && cd train
+  tar -xvf ILSVRC2012_img_train.tar && rm -f ILSVRC2012_img_train.tar
+  find . -name "*.tar" | while read NAME ; do mkdir -p "${NAME%.tar}"; tar -xvf "${NAME}" -C "${NAME%.tar}"; rm -f "${NAME}"; done
+  cd ..
+  ```
+
+3. Extract the validation data and move the images to subfolders:
+  ```bash
+  mkdir val && mv ILSVRC2012_img_val.tar val/ && cd val && tar -xvf ILSVRC2012_img_val.tar
+  wget -qO- https://raw.githubusercontent.com/soumith/imagenetloader.torch/master/valprep.sh | bash
+  ```
+
+The directory in which the `train/` and `val/` directories are placed, is referred to as `<path to imagenet>` in this document.
+
+### 3. Build the SE-ResNeXt101-32x4d PyTorch NGC container.
+
+```
+docker build . -t nvidia_se-resnext101-32x4d
+```
+
+### 4. Start an interactive session in the NGC container to run training/inference.
+```
+nvidia-docker run --rm -it -v <path to imagenet>:/imagenet --ipc=host nvidia_se-resnext101-32x4d
+```
+
+
+### 5. Start training
+
+To run training for a standard configuration (DGXA100/DGX1V, AMP/TF32/FP32, 90/250 Epochs),
+run one of the scripts in the `./se-resnext101-32x4d/training` directory
+called `./se-resnext101-32x4d/training/{AMP, TF32, FP32}/{ DGXA100, DGX1V }_se-resnext101-32x4d_{AMP, TF32, FP32}_{ 90, 250 }E.sh`.
+
+Ensure ImageNet is mounted in the `/imagenet` directory.
+
+Example:
+    `bash ./se-resnext101-32x4d/training/AMP/DGX1_se-resnext101-32x4d_AMP_250E.sh <path were to store checkpoints and logs>`
+
+### 6. Start inference
+
+You can download pretrained weights from NGC:
+
+```bash
+wget --content-disposition https://api.ngc.nvidia.com/v2/models/nvidia/seresnext101_32x4d_pyt_amp/versions/20.06.0/zip -O seresnext101_32x4d_pyt_amp_20.06.0.zip
+
+unzip seresnext101_32x4d_pyt_amp_20.06.0.zip
+```
+
+To run inference on ImageNet, run:
+
+`python ./main.py --arch se-resnext101-32x4d --evaluate --epochs 1 --pretrained-from-file nvidia_se-resnext101-32x4d_200821.pth.tar -b <batch size> <path to imagenet>`
+
+To run inference on JPEG image using pretrained weights:
+
+`python classify.py --arch se-resnext101-32x4d --pretrained-from-file nvidia_se-resnext101-32x4d_200821.pth.tar --precision AMP|FP32 --image <path to JPEG image>`
+
+
+## Advanced
+
+The following sections provide greater details of the dataset, running training and inference, and the training results.
+
+### Scripts and sample code
+
+To run a non standard configuration use:
+
+* For 1 GPU
+    * FP32
+        `python ./main.py --arch se-resnext101-32x4d -c fanin --label-smoothing 0.1 <path to imagenet>`
+        `python ./main.py --arch se-resnext101-32x4d -c fanin --label-smoothing 0.1 --amp --static-loss-scale 256 <path to imagenet>`
+
+* For multiple GPUs
+    * FP32
+        `python ./multiproc.py --nproc_per_node 8 ./main.py --arch se-resnext101-32x4d -c fanin --label-smoothing 0.1 <path to imagenet>`
+    * AMP
+        `python ./multiproc.py --nproc_per_node 8 ./main.py --arch se-resnext101-32x4d -c fanin --label-smoothing 0.1 --amp --static-loss-scale 256 <path to imagenet>`
+
+Use `python ./main.py -h` to obtain the list of available options in the `main.py` script.
+
+
+### Command-line options:
+
+To see the full list of available options and their descriptions, use the `-h` or `--help` command-line option, for example:
+
+`python main.py -h`
+
+
+```
+usage: main.py [-h] [--data-backend BACKEND] [--arch ARCH]
+               [--model-config CONF] [-j N] [--epochs N]
+               [--run-epochs N] [-b N] [--optimizer-batch-size N] [--lr LR]
+               [--lr-schedule SCHEDULE] [--warmup E] [--label-smoothing S]
+               [--mixup ALPHA] [--momentum M] [--weight-decay W]
+               [--bn-weight-decay] [--nesterov] [--print-freq N]
+               [--resume PATH] [--pretrained-from-file PATH]
+               [--static-loss-scale STATIC_LOSS_SCALE] [--dynamic-loss-scale]
+               [--prof N] [--amp] [--seed SEED] [--gather-checkpoints]
+               [--raport-file RAPORT_FILE] [--evaluate] [--training-only]
+               [--no-checkpoints] [--checkpoint-filename CHECKPOINT_FILENAME]
+               [--workspace DIR] [--memory-format {nchw,nhwc}]
+               DIR
+
+PyTorch ImageNet Training
+
+positional arguments:
+  DIR                   path to dataset
+
+optional arguments:
+  -h, --help            show this help message and exit
+  --data-backend BACKEND
+                        data backend: pytorch | synthetic | dali-gpu | dali-cpu
+                        (default: dali-cpu)
+  --arch ARCH, -a ARCH  model architecture: resnet18 | resnet34 | resnet50 |
+                        resnet101 | resnet152 | resnext50-32x4d |
+                        resnext101-32x4d | resnext101-32x8d |
+                        resnext101-32x8d-basic | se-resnext101-32x4d (default:
+                        resnet50)
+  --model-config CONF, -c CONF
+                        model configs: classic | fanin | grp-fanin | grp-
+                        fanout(default: classic)
+  -j N, --workers N     number of data loading workers (default: 5)
+  --epochs N            number of total epochs to run
+  --run-epochs N        run only N epochs, used for checkpointing runs
+  -b N, --batch-size N  mini-batch size (default: 256) per gpu
+  --optimizer-batch-size N
+                        size of a total batch size, for simulating bigger
+                        batches using gradient accumulation
+  --lr LR, --learning-rate LR
+                        initial learning rate
+  --lr-schedule SCHEDULE
+                        Type of LR schedule: step, linear, cosine
+  --warmup E            number of warmup epochs
+  --label-smoothing S   label smoothing
+  --mixup ALPHA         mixup alpha
+  --momentum M          momentum
+  --weight-decay W, --wd W
+                        weight decay (default: 1e-4)
+  --bn-weight-decay     use weight_decay on batch normalization learnable
+                        parameters, (default: false)
+  --nesterov            use nesterov momentum, (default: false)
+  --print-freq N, -p N  print frequency (default: 10)
+  --resume PATH         path to latest checkpoint (default: none)
+  --pretrained-from-file PATH
+                        load weights from here
+  --static-loss-scale STATIC_LOSS_SCALE
+                        Static loss scale, positive power of 2 values can
+                        improve amp convergence.
+  --dynamic-loss-scale  Use dynamic loss scaling. If supplied, this argument
+                        supersedes --static-loss-scale.
+  --prof N              Run only N iterations
+  --amp                 Run model AMP (automatic mixed precision) mode.
+  --seed SEED           random seed used for numpy and pytorch
+  --gather-checkpoints  Gather checkpoints throughout the training, without
+                        this flag only best and last checkpoints will be
+                        stored
+  --raport-file RAPORT_FILE
+                        file in which to store JSON experiment raport
+  --evaluate            evaluate checkpoint/model
+  --training-only       do not evaluate
+  --no-checkpoints      do not store any checkpoints, useful for benchmarking
+  --checkpoint-filename CHECKPOINT_FILENAME
+  --workspace DIR       path to directory where checkpoints will be stored
+  --memory-format {nchw,nhwc}
+                        memory layout, nchw or nhwc
+
+```
+
+
+### Dataset guidelines
+
+To use your own dataset, divide it in directories as in the following scheme:
+
+ - Training images - `train/<class id>/<image>`
+ - Validation images - `val/<class id>/<image>`
+
+If your dataset's has number of classes different than 1000, you need to pass `--num_classes N` flag to the training script.
+
+### Training process
+
+All the results of the training will be stored in the directory specified with `--workspace` argument.
+Script will store:
+ - most recent checkpoint - `checkpoint.pth.tar` (unless `--no-checkpoints` flag is used).
+ - checkpoint with best validation accuracy - `model_best.pth.tar` (unless `--no-checkpoints` flag is used).
+ - JSON log - in the file specified with `--raport-file` flag.
+
+Metrics gathered through training:
+
+ - `train.loss` - training loss
+ - `train.total_ips` - training speed measured in images/second
+ - `train.compute_ips` - training speed measured in images/second, not counting data loading
+ - `train.data_time` - time spent on waiting on data
+ - `train.compute_time` - time spent in forward/backward pass
+
+To restart training from checkpoint use `--resume` option.
+
+To start training from pretrained weights (e.g. downloaded from NGC) use `--pretrained-from-file` option.
+
+The difference between those two is that the pretrained weights contain only model weights,
+and checkpoints, apart from model weights, contain optimizer state, LR scheduler state.
+
+Checkpoints are suitable for dividing the training into parts, for example in order
+to divide the training job into shorter stages, or restart training after infrastructure fail.
+
+Pretrained weights can be used as a base for finetuning the model to a different dataset,
+or as a backbone to detection models.
+
+### Inference process
+
+Validation is done every epoch, and can be also run separately on a checkpointed model.
+
+`python ./main.py --arch se-resnext101-32x4d --evaluate --epochs 1 --resume <path to checkpoint> -b <batch size> <path to imagenet>`
+
+Metrics gathered through training:
+
+ - `val.loss` - validation loss
+ - `val.top1` - validation top1 accuracy
+ - `val.top5` - validation top5 accuracy
+ - `val.total_ips` - inference speed measured in images/second
+ - `val.compute_ips` - inference speed measured in images/second, not counting data loading
+ - `val.data_time` - time spent on waiting on data
+ - `val.compute_time` - time spent on inference
+
+
+To run inference on JPEG image, you have to first extract the model weights from checkpoint:
+
+`python checkpoint2model.py --checkpoint-path <path to checkpoint> --weight-path <path where weights will be stored>`
+
+Then run classification script:
+
+`python classify.py --arch se-resnext101-32x4d --pretrained-from-file <path to weights from previous step> --precision AMP|FP32 --image <path to JPEG image>`
+
+You can also run ImageNet validation on pretrained weights:
+
+`python ./main.py --arch se-resnext101-32x4d --evaluate --epochs 1 --pretrained-from-file <path to pretrained weights> -b <batch size> <path to imagenet>`
+
+#### NGC Pretrained weights:
+
+Pretrained weights can be downloaded from NGC:
+
+```bash
+wget --content-disposition https://api.ngc.nvidia.com/v2/models/nvidia/seresnext101_32x4d_pyt_amp/versions/20.06.0/zip -O seresnext101_32x4d_pyt_amp_20.06.0.zip
+
+unzip seresnext101_32x4d_pyt_amp_20.06.0.zip
+```
+To run inference on ImageNet, run:
+
+`python ./main.py --arch se-resnext101-32x4d --evaluate --epochs 1 --pretrained-from-file nvidia_se-resnext101-32x4d_200821.pth.tar -b <batch size> <path to imagenet>`
+
+To run inference on JPEG image using pretrained weights:
+
+`python classify.py --arch se-resnext101-32x4d --pretrained-from-file nvidia_se-resnext101-32x4d_200821.pth.tar --precision AMP|FP32 --image <path to JPEG image>`
+
+
+## Performance
+
+The performance measurements in this document were conducted at the time of publication and may not reflect the performance achieved from NVIDIA’s latest software release. For the most up-to-date performance measurements, go to [NVIDIA Data Center Deep Learning Product Performance](https://developer.nvidia.com/deep-learning-performance-training-inference).
+
+### Benchmarking
+
+The following section shows how to run benchmarks measuring the model performance in training and inference modes.
+
+#### Training performance benchmark
+
+To benchmark training, run:
+
+* For 1 GPU
+    * FP32 (V100 GPUs only)
+        `python ./launch.py --model se-resnext101-32x4d --precision FP32 --mode benchmark_training --platform DGX1V <path to imagenet> --raport-file benchmark.json --epochs 1 --prof 100`
+    * TF32 (A100 GPUs only)
+        `python ./launch.py --model se-resnext101-32x4d --precision TF32 --mode benchmark_training --platform DGXA100 <path to imagenet> --raport-file benchmark.json --epochs 1 --prof 100`
+    * AMP
+        `python ./launch.py --model se-resnext101-32x4d --precision AMP --mode benchmark_training --platform <DGX1V|DGXA100> <path to imagenet> --raport-file benchmark.json --epochs 1 --prof 100`
+* For multiple GPUs
+    * FP32 (V100 GPUs only)
+        `python ./launch.py --model se-resnext101-32x4d --precision FP32 --mode benchmark_training --platform DGX1V <path to imagenet> --raport-file benchmark.json --epochs 1 --prof 100`
+    * TF32 (A100 GPUs only)
+        `python ./multiproc.py --nproc_per_node 8 ./launch.py --model se-resnext101-32x4d --precision TF32 --mode benchmark_training --platform DGXA100 <path to imagenet> --raport-file benchmark.json --epochs 1 --prof 100`
+    * AMP
+        `python ./multiproc.py --nproc_per_node 8 ./launch.py --model se-resnext101-32x4d --precision AMP --mode benchmark_training --platform <DGX1V|DGXA100> <path to imagenet> --raport-file benchmark.json --epochs 1 --prof 100`
+
+Each of these scripts will run 100 iterations and save results in the `benchmark.json` file.
+
+#### Inference performance benchmark
+
+To benchmark inference, run:
+
+* FP32 (V100 GPUs only)
+
+`python ./launch.py --model se-resnext101-32x4d --precision FP32 --mode benchmark_inference --platform DGX1V <path to imagenet> --raport-file benchmark.json --epochs 1 --prof 100`
+
+* TF32 (A100 GPUs only)
+
+`python ./launch.py --model se-resnext101-32x4d --precision TF32 --mode benchmark_inference --platform DGXA100 <path to imagenet> --raport-file benchmark.json --epochs 1 --prof 100`
+
+* AMP
+
+`python ./launch.py --model se-resnext101-32x4d --precision AMP --mode benchmark_inference --platform <DGX1V|DGXA100> <path to imagenet> --raport-file benchmark.json --epochs 1 --prof 100`
+
+Each of these scripts will run 100 iterations and save results in the `benchmark.json` file.
+
+### Results
+
+#### Training accuracy results
+
+Our results were obtained by running the applicable training script the pytorch-20.12 NGC container.
+
+To achieve these same results, follow the steps in the [Quick Start Guide](#quick-start-guide).
+
+##### Training accuracy: NVIDIA DGX A100 (8x A100 80GB)
+
+| **Epochs** | **Mixed Precision Top1** | **TF32 Top1**  |
+|:----------:|:------------------------:|:--------------:|
+|     90     |      80.03 +/- 0.11      | 79.92 +/- 0.07 |
+|    250     |      80.9 +/- 0.08       | 80.98 +/- 0.07 |
+
+
+##### Training accuracy: NVIDIA DGX-1 (8x V100 16GB)
+
+| **Epochs** | **Mixed Precision Top1** | **FP32 Top1**  |
+|:----------:|:------------------------:|:--------------:|
+|     90     |      80.04 +/- 0.07      | 79.93 +/- 0.10 |
+|    250     |      80.92 +/- 0.09      | 80.97 +/- 0.09 |
+
+
+##### Example plots
+
+The following images show a 250 epochs configuration on a DGX-1V.
+
+![ValidationLoss](./img/loss_plot.png)
+
+![ValidationTop1](./img/top1_plot.png)
+
+![ValidationTop5](./img/top5_plot.png)
+
+#### Training performance results
+
+Our results were obtained by running the applicable training script the pytorch-21.03 NGC container.
+
+To achieve these same results, follow the steps in the [Quick Start Guide](#quick-start-guide).
+
+##### Training performance: NVIDIA DGX A100 (8x A100 80GB)
+
+| **GPUs** | **Throughput - TF32** | **Throughput - mixed precision** | **Throughput speedup (TF32 to mixed precision)** | **TF32 Strong Scaling** | **Mixed Precision Strong Scaling** | **Mixed Precision Training Time (90E)** | **TF32 Training Time (90E)** |
+|:--------:|:---------------------:|:--------------------------------:|:------------------------------------------------:|:-----------------------:|:----------------------------------:|:---------------------------------------:|:----------------------------:|
+|    1     |       395 img/s       |            855 img/s             |                      2.16 x                      |          1.0 x          |               1.0 x                |                ~40 hours                |          ~86 hours           |
+|    8     |      2991 img/s       |            5779 img/s            |                      1.93 x                      |         7.56 x          |               6.75 x               |                ~6 hours                 |          ~12 hours           |
+
+
+##### Training performance: NVIDIA DGX-1 16GB (8x V100 16GB)
+
+| **GPUs** | **Throughput - FP32** | **Throughput - mixed precision** | **Throughput speedup (FP32 to mixed precision)** | **FP32 Strong Scaling** | **Mixed Precision Strong Scaling** | **Mixed Precision Training Time (90E)** | **FP32 Training Time (90E)** |
+|:--------:|:---------------------:|:--------------------------------:|:------------------------------------------------:|:-----------------------:|:----------------------------------:|:---------------------------------------:|:----------------------------:|
+|    1     |       132 img/s       |            443 img/s             |                      3.34 x                      |          1.0 x          |               1.0 x                |                ~76 hours                |          ~254 hours          |
+|    8     |      1004 img/s       |            2971 img/s            |                      2.95 x                      |         7.57 x          |               6.7 x                |                ~12 hours                |          ~34 hours           |
+
+
+##### Training performance: NVIDIA DGX-1 32GB (8x V100 32GB)
+
+| **GPUs** | **Throughput - FP32** | **Throughput - mixed precision** | **Throughput speedup (FP32 to mixed precision)** | **FP32 Strong Scaling** | **Mixed Precision Strong Scaling** | **Mixed Precision Training Time (90E)** | **FP32 Training Time (90E)** |
+|:--------:|:---------------------:|:--------------------------------:|:------------------------------------------------:|:-----------------------:|:----------------------------------:|:---------------------------------------:|:----------------------------:|
+|    1     |       130 img/s       |            427 img/s             |                      3.26 x                      |          1.0 x          |               1.0 x                |                ~79 hours                |          ~257 hours          |
+|    8     |       992 img/s       |            2925 img/s            |                      2.94 x                      |         7.58 x          |               6.84 x               |                ~12 hours                |          ~34 hours           |
+
+
+#### Inference performance results
+
+Our results were obtained by running the applicable training script the pytorch-21.03 NGC container.
+
+To achieve these same results, follow the steps in the [Quick Start Guide](#quick-start-guide).
+
+##### Inference performance: NVIDIA DGX-1 (1x V100 16GB)
+
+###### FP32 Inference Latency
+
+| **Batch Size** | **Throughput Avg** | **Latency Avg** | **Latency 95%** | **Latency 99%** |
+|:--------------:|:------------------:|:---------------:|:---------------:|:---------------:|
+|       1        |      40 img/s      |    24.92 ms     |    26.78 ms     |    31.12 ms     |
+|       2        |      80 img/s      |    24.89 ms     |    27.63 ms     |    30.81 ms     |
+|       4        |     127 img/s      |    31.58 ms     |    35.92 ms     |    39.64 ms     |
+|       8        |     250 img/s      |    32.29 ms     |     34.5 ms     |    38.14 ms     |
+|       16       |     363 img/s      |     44.5 ms     |    44.16 ms     |    44.37 ms     |
+|       32       |     423 img/s      |    76.86 ms     |    75.89 ms     |    76.17 ms     |
+|       64       |     472 img/s      |    138.36 ms    |    135.85 ms    |    136.52 ms    |
+|      128       |     501 img/s      |    262.64 ms    |    255.48 ms    |    256.02 ms    |
+|      256       |     508 img/s      |    519.84 ms    |    500.71 ms    |    501.5 ms     |
+
+
+###### Mixed Precision Inference Latency
+
+| **Batch Size** | **Throughput Avg** | **Latency Avg** | **Latency 95%** | **Latency 99%** |
+|:--------------:|:------------------:|:---------------:|:---------------:|:---------------:|
+|       1        |      29 img/s      |    33.83 ms     |     39.1 ms     |    41.57 ms     |
+|       2        |      58 img/s      |    34.35 ms     |    36.92 ms     |    41.66 ms     |
+|       4        |     117 img/s      |    34.33 ms     |    38.67 ms     |    41.05 ms     |
+|       8        |     232 img/s      |    34.66 ms     |    39.51 ms     |    42.16 ms     |
+|       16       |     459 img/s      |    35.23 ms     |    36.77 ms     |    38.11 ms     |
+|       32       |     871 img/s      |    37.62 ms     |    39.36 ms     |    41.26 ms     |
+|       64       |     1416 img/s     |    46.95 ms     |    45.26 ms     |    47.48 ms     |
+|      128       |     1533 img/s     |    87.49 ms     |    83.54 ms     |    83.75 ms     |
+|      256       |     1576 img/s     |    170.79 ms    |    161.97 ms    |    162.93 ms    |
+
+
+##### Inference performance: NVIDIA T4
+
+###### FP32 Inference Latency
+
+| **Batch Size** | **Throughput Avg** | **Latency Avg** | **Latency 95%** | **Latency 99%** |
+|:--------------:|:------------------:|:---------------:|:---------------:|:---------------:|
+|       1        |      40 img/s      |    25.12 ms     |    28.83 ms     |    31.59 ms     |
+|       2        |      75 img/s      |    26.82 ms     |    30.54 ms     |    33.13 ms     |
+|       4        |     136 img/s      |    29.79 ms     |    33.33 ms     |    37.65 ms     |
+|       8        |     155 img/s      |    51.74 ms     |    52.57 ms     |    53.12 ms     |
+|       16       |     164 img/s      |    97.99 ms     |    98.76 ms     |    99.21 ms     |
+|       32       |     173 img/s      |    186.31 ms    |    186.43 ms    |    187.4 ms     |
+|       64       |     171 img/s      |    378.1 ms     |    377.19 ms    |    378.82 ms    |
+|      128       |     165 img/s      |    785.83 ms    |    778.23 ms    |    782.64 ms    |
+|      256       |     158 img/s      |   1641.96 ms    |   1601.74 ms    |   1614.52 ms    |
+
+
+###### Mixed Precision Inference Latency
+
+| **Batch Size** | **Throughput Avg** | **Latency Avg** | **Latency 95%** | **Latency 99%** |
+|:--------------:|:------------------:|:---------------:|:---------------:|:---------------:|
+|       1        |      31 img/s      |    32.51 ms     |    37.26 ms     |    39.53 ms     |
+|       2        |      61 img/s      |    32.76 ms     |    37.61 ms     |    39.62 ms     |
+|       4        |     123 img/s      |    32.98 ms     |    38.97 ms     |    42.66 ms     |
+|       8        |     262 img/s      |    31.01 ms     |     36.3 ms     |    39.11 ms     |
+|       16       |     482 img/s      |    33.76 ms     |    34.54 ms     |     38.5 ms     |
+|       32       |     512 img/s      |    63.68 ms     |    63.29 ms     |    63.73 ms     |
+|       64       |     527 img/s      |    123.57 ms    |    122.69 ms    |    123.56 ms    |
+|      128       |     525 img/s      |    248.97 ms    |    245.39 ms    |    246.66 ms    |
+|      256       |     527 img/s      |    496.23 ms    |    485.68 ms    |    488.3 ms     |
+
+
+## Release notes
+
+### Changelog
+
+1. October 2019
+  * Initial release
+2. July 2020
+  * Added A100 scripts
+  * Updated README
+3. February 2021
+  * Moved from APEX AMP to Native AMP
+
+### Known issues
+
+There are no known issues with this model.
+

se-resnext101-32x4d/training/AMP/DGX1V_se-resnext101-32x4d_AMP_250E.sh

+python ./multiproc.py --nproc_per_node 8 ./launch.py --model se-resnext101-32x4d --precision AMP --mode convergence --platform DGX1V /imagenet --workspace ${1:-./} --raport-file raport.json

se-resnext101-32x4d/training/AMP/DGX1V_se-resnext101-32x4d_AMP_90E.sh

+python ./multiproc.py --nproc_per_node 8 ./launch.py --model se-resnext101-32x4d --precision AMP --mode convergence --platform DGX1V /imagenet --epochs 90 --mixup 0.0 --workspace ${1:-./} --raport-file raport.json

se-resnext101-32x4d/training/AMP/DGXA100_se-resnext101-32x4d_AMP_250E.sh

+python ./multiproc.py --nproc_per_node 8 ./launch.py --model se-resnext101-32x4d --precision AMP --mode convergence --platform DGXA100 /imagenet --workspace ${1:-./} --raport-file raport.json

se-resnext101-32x4d/training/AMP/DGXA100_se-resnext101-32x4d_AMP_90E.sh

+python ./multiproc.py --nproc_per_node 8 ./launch.py --model se-resnext101-32x4d --precision AMP --mode convergence --platform DGXA100 /imagenet --epochs 90 --mixup 0.0 --workspace ${1:-./} --raport-file raport.json

se-resnext101-32x4d/training/FP32/DGX1V_se-resnext101-32x4d_FP32_250E.sh

+python ./multiproc.py --nproc_per_node 8 ./launch.py --model se-resnext101-32x4d --precision FP32 --mode convergence --platform DGX1V /imagenet --workspace ${1:-./} --raport-file raport.json

se-resnext101-32x4d/training/FP32/DGX1V_se-resnext101-32x4d_FP32_90E.sh

+python ./multiproc.py --nproc_per_node 8 ./launch.py --model se-resnext101-32x4d --precision FP32 --mode convergence --platform DGX1V /imagenet --epochs 90 --mixup 0.0 --workspace ${1:-./} --raport-file raport.json

se-resnext101-32x4d/training/TF32/DGXA100_se-resnext101-32x4d_TF32_250E.sh

+python ./multiproc.py --nproc_per_node 8 ./launch.py --model se-resnext101-32x4d --precision TF32 --mode convergence --platform DGXA100 /imagenet --workspace ${1:-./} --raport-file raport.json

se-resnext101-32x4d/training/TF32/DGXA100_se-resnext101-32x4d_TF32_90E.sh

+python ./multiproc.py --nproc_per_node 8 ./launch.py --model se-resnext101-32x4d --precision TF32 --mode convergence --platform DGXA100 /imagenet --epochs 90 --mixup 0.0 --workspace ${1:-./} --raport-file raport.json

train_multi.sh

+export DATA_DIR=/data/imagenet2012  #数据集路径
+python ./multiproc.py --nproc_per_node 8 ./launch.py --model resnet50 --precision AMP --mode convergence --platform Z100L ${DATA_DIR} --data-backend pytorch --epochs 100 --batch-size 128 --workspace ${1:-./run} --raport-file raport.json 2>&1 | tee resnet50_multi_`date +%Y%m%d%H%M%S`.log

triton/Dockerfile

+# Copyright (c) 2020 NVIDIA CORPORATION. All rights reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#       http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+ARG FROM_IMAGE_NAME=nvcr.io/nvidia/pytorch:20.07-py3
+ARG TRITON_BASE_IMAGE=nvcr.io/nvidia/tritonserver:20.07-py3-clientsdk
+
+FROM ${TRITON_BASE_IMAGE} as trt
+FROM ${FROM_IMAGE_NAME}
+
+ADD requirements.txt .
+RUN pip install -r requirements.txt
+RUN pip install onnxruntime
+
+COPY --from=trt /workspace/v2.1.0.clients.tar.gz ./v2.1.0.clients.tar.gz
+COPY --from=trt /workspace/install/bin/perf_client /bin/perf_client
+
+RUN tar -xzf v2.1.0.clients.tar.gz \
+    && pip install ./python/tritonclientutils-2.1.0-py3-none-any.whl
+
+
+RUN apt update && apt install -y libb64-0d
+
+WORKDIR /workspace/rn50
+COPY . .
+

triton/calculate_metrics.py

+#!/usr/bin/env python3
+
+# Copyright (c) 2021, NVIDIA CORPORATION. All rights reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+r"""
+Using `calculate_metrics.py` script, you can obtain model accuracy/error metrics using defined `MetricsCalculator` class.
+
+Data provided to `MetricsCalculator` are obtained from npz dump files
+stored in directory pointed by `--dump-dir` argument.
+Above files are prepared by `run_inference_on_fw.py` and `run_inference_on_triton.py` scripts.
+
+Output data is stored in csv file pointed by `--csv` argument.
+
+Example call:
+
+```shell script
+python ./triton/calculate_metrics.py \
+    --dump-dir /results/dump_triton \
+    --csv /results/accuracy_results.csv \
+    --metrics metrics.py \
+    --metric-class-param1 value
+```
+"""
+
+import argparse
+import csv
+import logging
+import string
+from pathlib import Path
+
+import numpy as np
+
+# method from PEP-366 to support relative import in executed modules
+
+if __package__ is None:
+    __package__ = Path(__file__).parent.name
+
+from .deployment_toolkit.args import ArgParserGenerator
+from .deployment_toolkit.core import BaseMetricsCalculator, load_from_file
+from .deployment_toolkit.dump import pad_except_batch_axis
+
+LOGGER = logging.getLogger("calculate_metrics")
+TOTAL_COLUMN_NAME = "_total_"
+
+
+def get_data(dump_dir, prefix):
+    """Loads and concatenates dump files for given prefix (ex. inputs, outputs, labels, ids)"""
+    dump_dir = Path(dump_dir)
+    npz_files = sorted(dump_dir.glob(f"{prefix}*.npz"))
+    data = None
+    if npz_files:
+        # assume that all npz files with given prefix contain same set of names
+        names = list(np.load(npz_files[0].as_posix()).keys())
+        # calculate target shape
+        target_shape = {
+            name: tuple(np.max([np.load(npz_file.as_posix())[name].shape for npz_file in npz_files], axis=0))
+            for name in names
+        }
+        # pad and concatenate data
+        data = {
+            name: np.concatenate(
+                [pad_except_batch_axis(np.load(npz_file.as_posix())[name], target_shape[name]) for npz_file in npz_files]
+            )
+            for name in names
+        }
+    return data
+
+
+def main():
+    logging.basicConfig(level=logging.INFO)
+
+    parser = argparse.ArgumentParser(description="Run models with given dataloader", allow_abbrev=False)
+    parser.add_argument("--metrics", help=f"Path to python module containing metrics calculator", required=True)
+    parser.add_argument("--csv", help="Path to csv file", required=True)
+    parser.add_argument("--dump-dir", help="Path to directory with dumped outputs (and labels)", required=True)
+
+    args, *_ = parser.parse_known_args()
+
+    MetricsCalculator = load_from_file(args.metrics, "metrics", "MetricsCalculator")
+    ArgParserGenerator(MetricsCalculator).update_argparser(parser)
+
+    args = parser.parse_args()
+
+    LOGGER.info(f"args:")
+    for key, value in vars(args).items():
+        LOGGER.info(f"    {key} = {value}")
+
+    MetricsCalculator = load_from_file(args.metrics, "metrics", "MetricsCalculator")
+    metrics_calculator: BaseMetricsCalculator = ArgParserGenerator(MetricsCalculator).from_args(args)
+
+    ids = get_data(args.dump_dir, "ids")["ids"]
+    x = get_data(args.dump_dir, "inputs")
+    y_true = get_data(args.dump_dir, "labels")
+    y_pred = get_data(args.dump_dir, "outputs")
+
+    common_keys = list({k for k in (y_true or [])} & {k for k in (y_pred or [])})
+    for key in common_keys:
+        if y_true[key].shape != y_pred[key].shape:
+            LOGGER.warning(
+                f"Model predictions and labels shall have equal shapes. "
+                f"y_pred[{key}].shape={y_pred[key].shape} != "
+                f"y_true[{key}].shape={y_true[key].shape}"
+            )
+
+    metrics = metrics_calculator.calc(ids=ids, x=x, y_pred=y_pred, y_real=y_true)
+    metrics = {TOTAL_COLUMN_NAME: len(ids), **metrics}
+
+    metric_names_with_space = [name for name in metrics if any([c in string.whitespace for c in name])]
+    if metric_names_with_space:
+        raise ValueError(f"Metric names shall have no spaces; Incorrect names: {', '.join(metric_names_with_space)}")
+
+    csv_path = Path(args.csv)
+    csv_path.parent.mkdir(parents=True, exist_ok=True)
+    with csv_path.open("w") as csv_file:
+        writer = csv.DictWriter(csv_file, fieldnames=list(metrics.keys()))
+        writer.writeheader()
+        writer.writerow(metrics)
+
+
+if __name__ == "__main__":
+    main()

triton/client.py

+# Copyright (c) 2020 NVIDIA CORPORATION. All rights reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#       http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+import argparse
+import json
+import os
+
+import numpy as np
+import torch
+import torchvision.datasets as datasets
+import torchvision.transforms as transforms
+
+from image_classification.dataloaders import get_pytorch_val_loader
+
+from tqdm import tqdm
+
+import tritongrpcclient
+from tritonclientutils import InferenceServerException
+
+
+def get_data_loader(batch_size, *, data_path):
+    valdir = os.path.join(data_path, "val-jpeg")
+    val_dataset = datasets.ImageFolder(
+        valdir,
+        transforms.Compose(
+            [transforms.Resize(256), transforms.CenterCrop(224), transforms.ToTensor()]
+        ),
+    )
+
+    val_loader = torch.utils.data.DataLoader(
+        val_dataset, batch_size=batch_size, shuffle=False
+    )
+
+    return val_loader
+
+
+if __name__ == "__main__":
+    parser = argparse.ArgumentParser()
+    parser.add_argument(
+        "--triton-server-url",
+        type=str,
+        required=True,
+        help="URL adress of trtion server (with port)",
+    )
+    parser.add_argument(
+        "--triton-model-name",
+        type=str,
+        required=True,
+        help="Triton deployed model name",
+    )
+    parser.add_argument(
+        "-v", "--verbose", action="store_true", default=False, help="Verbose mode."
+    )
+
+    parser.add_argument(
+        "--inference_data", type=str, help="Path to file with inference data."
+    )
+    parser.add_argument(
+        "--batch_size", type=int, default=1, help="Inference request batch size"
+    )
+    parser.add_argument(
+        "--fp16",
+        action="store_true",
+        default=False,
+        help="Use fp16 precision for input data",
+    )
+    FLAGS = parser.parse_args()
+
+    triton_client = tritongrpcclient.InferenceServerClient(
+        url=FLAGS.triton_server_url, verbose=FLAGS.verbose
+    )
+    dataloader = get_data_loader(FLAGS.batch_size, data_path=FLAGS.inference_data)
+
+    inputs = []
+    inputs.append(
+        tritongrpcclient.InferInput(
+            "input__0",
+            [FLAGS.batch_size, 3, 224, 224],
+            "FP16" if FLAGS.fp16 else "FP32",
+        )
+    )
+
+    outputs = []
+    outputs.append(tritongrpcclient.InferRequestedOutput("output__0"))
+
+    all_img = 0
+    cor_img = 0
+
+    result_prev = None
+    for image, target in tqdm(dataloader):
+        if FLAGS.fp16:
+            image = image.half()
+        inputs[0].set_data_from_numpy(image.numpy())
+
+        result = triton_client.infer(
+            FLAGS.triton_model_name, inputs, outputs=outputs, headers=None
+        )
+        result = result.as_numpy("output__0")
+        result = np.argmax(result, axis=1)
+        cor_img += np.sum(result == target.numpy())
+        all_img += result.shape[0]
+
+    acc = cor_img / all_img
+    print(f"Final accuracy {acc:.04f}")

triton/config_model_on_triton.py

+#!/usr/bin/env python3
+
+# Copyright (c) 2021, NVIDIA CORPORATION. All rights reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+r"""
+To configure model on Triton, you can use `config_model_on_triton.py` script.
+This will prepare layout of Model Repository, including  Model Configuration.
+
+```shell script
+python ./triton/config_model_on_triton.py \
+    --model-repository /model_repository \
+    --model-path /models/exported/model.onnx \
+    --model-format onnx \
+    --model-name ResNet50 \
+    --model-version 1 \
+    --max-batch-size 32 \
+    --precision fp16 \
+    --backend-accelerator trt \
+    --load-model explicit \
+    --timeout 120 \
+    --verbose
+```
+
+If Triton server to which we prepare model repository is running with **explicit model control mode**,
+use `--load-model` argument to send request load_model request to Triton Inference Server.
+If server is listening on non-default address or port use `--server-url` argument to point server control endpoint.
+If it is required to use HTTP protocol to communicate with Triton server use `--http` argument.
+
+To improve inference throughput you can use
+[dynamic batching](https://github.com/triton-inference-server/server/blob/master/docs/model_configuration.md#dynamic-batcher)
+for your model by providing `--preferred-batch-sizes` and `--max-queue-delay-us` parameters.
+
+For models which doesn't support batching, set `--max-batch-sizes` to 0.
+
+By default Triton will [automatically obtain inputs and outputs definitions](https://github.com/triton-inference-server/server/blob/master/docs/model_configuration.md#auto-generated-model-configuration).
+but for TorchScript ang TF GraphDef models script uses file with I/O specs. This file is automatically generated
+when the model is converted to ScriptModule (either traced or scripted).
+If there is a need to pass different than default path to I/O spec file use `--io-spec` CLI argument.
+
+I/O spec file is yaml file with below structure:
+
+```yaml
+- inputs:
+  - name: input
+    dtype: float32   # np.dtype name
+    shape: [None, 224, 224, 3]
+- outputs:
+  - name: probabilities
+    dtype: float32
+    shape: [None, 1001]
+  - name: classes
+    dtype: int32
+    shape: [None, 1]
+```
+
+"""
+
+import argparse
+import logging
+import time
+
+from model_navigator import Accelerator, Format, Precision
+from model_navigator.args import str2bool
+from model_navigator.log import set_logger, log_dict
+from model_navigator.triton import ModelConfig, TritonClient, TritonModelStore
+
+LOGGER = logging.getLogger("config_model")
+
+
+def _available_enum_values(my_enum):
+    return [item.value for item in my_enum]
+
+
+def main():
+    parser = argparse.ArgumentParser(
+        description="Create Triton model repository and model configuration", allow_abbrev=False
+    )
+    parser.add_argument("--model-repository", required=True, help="Path to Triton model repository.")
+    parser.add_argument("--model-path", required=True, help="Path to model to configure")
+
+    # TODO: automation
+    parser.add_argument(
+        "--model-format",
+        required=True,
+        choices=_available_enum_values(Format),
+        help="Format of model to deploy",
+    )
+    parser.add_argument("--model-name", required=True, help="Model name")
+    parser.add_argument("--model-version", default="1", help="Version of model (default 1)")
+    parser.add_argument(
+        "--max-batch-size",
+        type=int,
+        default=32,
+        help="Maximum batch size allowed for inference. "
+        "A max_batch_size value of 0 indicates that batching is not allowed for the model",
+    )
+    # TODO: automation
+    parser.add_argument(
+        "--precision",
+        type=str,
+        default=Precision.FP16.value,
+        choices=_available_enum_values(Precision),
+        help="Model precision (parameter used only by Tensorflow backend with TensorRT optimization)",
+    )
+
+    # Triton Inference Server endpoint
+    parser.add_argument(
+        "--server-url",
+        type=str,
+        default="grpc://localhost:8001",
+        help="Inference server URL in format protocol://host[:port] (default grpc://localhost:8001)",
+    )
+    parser.add_argument(
+        "--load-model",
+        choices=["none", "poll", "explicit"],
+        help="Loading model while Triton Server is in given model control mode",
+    )
+    parser.add_argument(
+        "--timeout", default=120, help="Timeout in seconds to wait till model load (default=120)", type=int
+    )
+
+    # optimization related
+    parser.add_argument(
+        "--backend-accelerator",
+        type=str,
+        choices=_available_enum_values(Accelerator),
+        default=Accelerator.TRT.value,
+        help="Select Backend Accelerator used to serve model",
+    )
+    parser.add_argument("--number-of-model-instances", type=int, default=1, help="Number of model instances per GPU")
+    parser.add_argument(
+        "--preferred-batch-sizes",
+        type=int,
+        nargs="*",
+        help="Batch sizes that the dynamic batcher should attempt to create. "
+        "In case --max-queue-delay-us is set and this parameter is not, default value will be --max-batch-size",
+    )
+    parser.add_argument(
+        "--max-queue-delay-us",
+        type=int,
+        default=0,
+        help="Max delay time which dynamic batcher shall wait to form a batch (default 0)",
+    )
+    parser.add_argument(
+        "--capture-cuda-graph",
+        type=int,
+        default=0,
+        help="Use cuda capture graph (used only by TensorRT platform)",
+    )
+
+    parser.add_argument("-v", "--verbose", help="Provide verbose logs", type=str2bool, default=False)
+    args = parser.parse_args()
+
+    set_logger(verbose=args.verbose)
+    log_dict("args", vars(args))
+
+    config = ModelConfig.create(
+        model_path=args.model_path,
+        # model definition
+        model_name=args.model_name,
+        model_version=args.model_version,
+        model_format=args.model_format,
+        precision=args.precision,
+        max_batch_size=args.max_batch_size,
+        # optimization
+        accelerator=args.backend_accelerator,
+        gpu_engine_count=args.number_of_model_instances,
+        preferred_batch_sizes=args.preferred_batch_sizes or [],
+        max_queue_delay_us=args.max_queue_delay_us,
+        capture_cuda_graph=args.capture_cuda_graph,
+    )
+
+    model_store = TritonModelStore(args.model_repository)
+    model_store.deploy_model(model_config=config, model_path=args.model_path)
+
+    if args.load_model != "none":
+        client = TritonClient(server_url=args.server_url, verbose=args.verbose)
+        client.wait_for_server_ready(timeout=args.timeout)
+
+        if args.load_model == "explicit":
+            client.load_model(model_name=args.model_name)
+
+        if args.load_model == "poll":
+            time.sleep(15)
+
+        client.wait_for_model(model_name=args.model_name, model_version=args.model_version, timeout_s=args.timeout)
+
+
+if __name__ == "__main__":
+    main()

triton/convert_model.py

+#!/usr/bin/env python3
+
+# Copyright (c) 2021, NVIDIA CORPORATION. All rights reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+r"""
+`convert_model.py` script allows to convert between model formats with additional model optimizations
+for faster inference.
+It converts model from results of get_model function.
+
+Currently supported input and output formats are:
+
+  - inputs
+    - `tf-estimator` - `get_model` function returning Tensorflow Estimator
+    - `tf-keras` - `get_model` function returning Tensorflow Keras Model
+    - `tf-savedmodel` - Tensorflow SavedModel binary
+    - `pyt` - `get_model` function returning PyTorch Module
+  - output
+    - `tf-savedmodel` - Tensorflow saved model
+    - `tf-trt` - TF-TRT saved model
+    - `ts-trace` - PyTorch traced ScriptModule
+    - `ts-script` - PyTorch scripted ScriptModule
+    - `onnx` - ONNX
+    - `trt` - TensorRT plan file
+
+For tf-keras input you can use:
+  - --large-model flag - helps loading model which exceeds maximum protobuf size of 2GB
+  - --tf-allow-growth flag - control limiting GPU memory growth feature
+    (https://www.tensorflow.org/guide/gpu#limiting_gpu_memory_growth). By default it is disabled.
+"""
+
+import argparse
+import logging
+import os
+from pathlib import Path
+
+os.environ["TF_CPP_MIN_LOG_LEVEL"] = "2"
+os.environ["TF_ENABLE_DEPRECATION_WARNINGS"] = "1"
+
+# method from PEP-366 to support relative import in executed modules
+if __name__ == "__main__" and __package__ is None:
+    __package__ = Path(__file__).parent.name
+
+from .deployment_toolkit.args import ArgParserGenerator
+from .deployment_toolkit.core import (
+    DATALOADER_FN_NAME,
+    BaseConverter,
+    BaseLoader,
+    BaseSaver,
+    Format,
+    Precision,
+    load_from_file,
+)
+from .deployment_toolkit.extensions import converters, loaders, savers
+
+LOGGER = logging.getLogger("convert_model")
+
+INPUT_MODEL_TYPES = [Format.TF_ESTIMATOR, Format.TF_KERAS, Format.TF_SAVEDMODEL, Format.PYT]
+OUTPUT_MODEL_TYPES = [Format.TF_SAVEDMODEL, Format.TF_TRT, Format.ONNX, Format.TRT, Format.TS_TRACE, Format.TS_SCRIPT]
+
+
+def _get_args():
+    parser = argparse.ArgumentParser(description="Script for conversion between model formats.", allow_abbrev=False)
+    parser.add_argument("--input-path", help="Path to input model file (python module or binary file)", required=True)
+    parser.add_argument(
+        "--input-type", help="Input model type", choices=[f.value for f in INPUT_MODEL_TYPES], required=True
+    )
+    parser.add_argument("--output-path", help="Path to output model file", required=True)
+    parser.add_argument(
+        "--output-type", help="Output model type", choices=[f.value for f in OUTPUT_MODEL_TYPES], required=True
+    )
+    parser.add_argument("--dataloader", help="Path to python module containing data loader")
+    parser.add_argument("-v", "--verbose", help="Verbose logs", action="store_true", default=False)
+    parser.add_argument(
+        "--ignore-unknown-parameters",
+        help="Ignore unknown parameters (argument often used in CI where set of arguments is constant)",
+        action="store_true",
+        default=False,
+    )
+
+    args, unparsed_args = parser.parse_known_args()
+
+    Loader: BaseLoader = loaders.get(args.input_type)
+    ArgParserGenerator(Loader, module_path=args.input_path).update_argparser(parser)
+
+    converter_name = f"{args.input_type}--{args.output_type}"
+    Converter: BaseConverter = converters.get(converter_name)
+    if Converter is not None:
+        ArgParserGenerator(Converter).update_argparser(parser)
+
+    Saver: BaseSaver = savers.get(args.output_type)
+    ArgParserGenerator(Saver).update_argparser(parser)
+
+    if args.dataloader is not None:
+        get_dataloader_fn = load_from_file(args.dataloader, label="dataloader", target=DATALOADER_FN_NAME)
+        ArgParserGenerator(get_dataloader_fn).update_argparser(parser)
+
+    if args.ignore_unknown_parameters:
+        args, unknown_args = parser.parse_known_args()
+        LOGGER.warning(f"Got additional args {unknown_args}")
+    else:
+        args = parser.parse_args()
+    return args
+
+
+def main():
+    args = _get_args()
+
+    log_level = logging.INFO if not args.verbose else logging.DEBUG
+    log_format = "%(asctime)s %(levelname)s %(name)s %(message)s"
+    logging.basicConfig(level=log_level, format=log_format)
+
+    LOGGER.info(f"args:")
+    for key, value in vars(args).items():
+        LOGGER.info(f"    {key} = {value}")
+
+    requested_model_precision = Precision(args.precision)
+    dataloader_fn = None
+
+    # if conversion is required, temporary change model load precision to that required by converter
+    # it is for TensorRT converters which require fp32 models for all requested precisions
+    converter_name = f"{args.input_type}--{args.output_type}"
+    Converter: BaseConverter = converters.get(converter_name)
+    if Converter:
+        args.precision = Converter.required_source_model_precision(requested_model_precision).value
+
+    Loader: BaseLoader = loaders.get(args.input_type)
+    loader = ArgParserGenerator(Loader, module_path=args.input_path).from_args(args)
+    model = loader.load(args.input_path)
+
+
+    LOGGER.info("inputs: %s", model.inputs)
+    LOGGER.info("outputs: %s", model.outputs)
+
+    if Converter:  # if conversion is needed
+        # dataloader must much source model precision - so not recovering it yet
+        if args.dataloader is not None:
+            get_dataloader_fn = load_from_file(args.dataloader, label="dataloader", target=DATALOADER_FN_NAME)
+            dataloader_fn = ArgParserGenerator(get_dataloader_fn).from_args(args)
+
+    # recover precision to that requested by user
+    args.precision = requested_model_precision.value
+
+    if Converter:
+        converter = ArgParserGenerator(Converter).from_args(args)
+        model = converter.convert(model, dataloader_fn=dataloader_fn)
+
+    Saver: BaseSaver = savers.get(args.output_type)
+    saver = ArgParserGenerator(Saver).from_args(args)
+    saver.save(model, args.output_path)
+
+    return 0
+
+
+if __name__ == "__main__":
+    main()