DeepLearningExamples/PyTorch/Classification/ConvNets/resnet50v1.5

ResNet50 v1.5 For PyTorch

This repository provides a script and recipe to train the ResNet50 model to achieve state-of-the-art accuracy, and is tested and maintained by NVIDIA.

Table Of Contents

• Model overview
  • Default configuration
    • Optimizer
    • Data augmentation
  • DALI
  • Feature support matrix
    • Features
  • Mixed precision training
    • Enabling mixed precision
    • Enabling TF32
• Setup
  • Requirements
• Quick Start Guide
• Advanced
  • Scripts and sample code
  • Command-line options
  • Dataset guidelines
  • Training process
  • Inference process
• Performance
  • Benchmarking
    • Training performance benchmark
    • Inference performance benchmark
  • Results
    • Training accuracy results
      • Training accuracy: NVIDIA DGX A100 (8x A100 40GB)
      • Training accuracy: NVIDIA DGX-1 (8x V100 16GB)
      • Training accuracy: NVIDIA DGX-2 (16x V100 32GB)
      • Example plots
    • Training performance results
      • Training performance: NVIDIA DGX A100 (8x A100 40GB)
      • Training performance: NVIDIA DGX-1 16GB (8x V100 16GB)
      • Training performance: NVIDIA DGX-1 32GB (8x V100 32GB)
  • Inference performance results
    • Inference performance: NVIDIA DGX-1 16GB (1x V100 16GB)
    • Inference performance: NVIDIA T4
• Release notes
  • Changelog
  • Known issues

Model overview

The ResNet50 v1.5 model is a modified version of the original ResNet50 v1 model.

The difference between v1 and v1.5 is that, in the bottleneck blocks which requires downsampling, v1 has stride = 2 in the first 1x1 convolution, whereas v1.5 has stride = 2 in the 3x3 convolution.

This difference makes ResNet50 v1.5 slightly more accurate (~0.5% top1) than v1, but comes with a smallperformance drawback (~5% imgs/sec).

The model is initialized as described in Delving deep into rectifiers: Surpassing human-level performance on ImageNet classification

This model is trained with mixed precision using Tensor Cores on Volta, Turing, and the NVIDIA Ampere GPU architectures. Therefore, researchers can get results over 2x 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 are currently working on adding NHWC data layout support for Mixed Precision training.

Default configuration

The following sections highlight the default configurations for the ResNet50 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. Warmup length depends on the total training length.
• Weight decay (WD)= 3.0517578125e-05 (1/32768).
• We do not apply WD on Batch Norm trainable parameters (gamma/bias)
• Label smoothing = 0.1
• We train for:
  • 50 Epochs -> configuration that reaches 75.9% top1 accuracy
  • 90 Epochs -> 90 epochs is a standard for ImageNet networks
  • 250 Epochs -> best possible accuracy.
• For 250 epoch training we also use MixUp regularization.

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

Other training recipes

This script does not target any specific benchmark. There are changes that others have made which can speed up convergence and/or increase accuracy.

One of the more popular training recipes is provided by fast.ai.

The fast.ai recipe introduces many changes to the training procedure, one of which is progressive resizing of the training images.

The first part of training uses 128px images, the middle part uses 224px images, and the last part uses 288px images. The final validation is performed on 288px images.

Training script in this repository performs validation on 224px images, just like the original paper described.

These two approaches can't be directly compared, since the fast.ai recipe requires validation on 288px images, and this recipe keeps the original assumption that validation is done on 224px images.

Using 288px images means that a lot more FLOPs are needed during inference to reach the same accuracy.

Feature support matrix

The following features are supported by this model:

Feature	ResNet50
DALI	Yes
APEX AMP	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.

• APEX is a PyTorch extension that contains utility libraries, such as Automatic Mixed Precision (AMP), which require minimal network code changes to leverage Tensor Cores performance. Refer to the Enabling mixed precision section for more details.

DALI

We use 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, for DGX2 we recommend --data-backends dali-gpu.

Mixed precision training

Mixed precision is the combined use of different numerical precisions in a computational method. Mixed precision 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 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 paper and Training With Mixed Precision documentation.
• Techniques used for mixed precision training, see the Mixed-Precision Training of Deep Neural Networks blog.
• APEX tools for mixed precision training, see the NVIDIA Apex: Tools for Easy Mixed-Precision Training in PyTorch.

Enabling mixed precision

Mixed precision is enabled in PyTorch by using the Automatic Mixed Precision (AMP), a library from 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 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 or fixed.

For an in-depth walk through on AMP, check out sample usage here. 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:

  from apex import amp
  

• Wrap model and optimizer in amp.initialize:

  model, optimizer = amp.initialize(model, optimizer, opt_level="O1", loss_scale="dynamic")
  

• Scale loss before backpropagation:

  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 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 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 ResNet50 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
• PyTorch 20.06-py3 NGC container or newer
• Supported GPUs:
  • NVIDIA Volta architecture
  • NVIDIA Turing architecture
  • NVIDIA Ampere 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
• Accessing And Pulling From The NGC Container Registry
• Running PyTorch

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.

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 ResNet50 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 resnet50 model on the ImageNet dataset. For the specifics concerning training and inference, see the Advanced section.

1. Download the images.

2. Extract the training data:

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:

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 RN50v1.5 PyTorch NGC container.

docker build . -t nvidia_rn50

4. Start an interactive session in the NGC container to run training/inference.

nvidia-docker run --rm -it -v <path to imagenet>:/data/imagenet --ipc=host nvidia_rn50

5. Start training

To run training for a standard configuration (DGXA100/DGX1/DGX2, AMP/TF32/FP32, 50/90/250 Epochs), run one of the scripts in the ./resnet50v1.5/training directory called ./resnet50v1.5/training/{AMP, TF32, FP32}/{DGXA100, DGX1, DGX2}_RN50_{AMP, TF32, FP32}_{50,90,250}E.sh.

Ensure ImageNet is mounted in the /data/imagenet directory.

Example: bash ./resnet50v1.5/training/AMP/DGX1_RN50_AMP_250E.sh <path were to store checkpoints and logs>

6. Start inference

To run inference on ImageNet on a checkpointed model, run:

python ./main.py --arch resnet50 --evaluate --epochs 1 --resume <path to checkpoint> -b <batch size> <path to imagenet>

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 resnet50 -c fanin --weights <path to weights from previous step> --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 resnet50 -c fanin --label-smoothing 0.1 <path to imagenet> python ./main.py --arch resnet50 -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 resnet50 -c fanin --label-smoothing 0.1 <path to imagenet>
  • AMP python ./multiproc.py --nproc_per_node 8 ./main.py --arch resnet50 -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] [--num-classes N] [-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-weights PATH] [--fp16]
               [--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 | syntetic | dali-gpu | dali-cpu
                        (default: dali-cpu)
  --arch ARCH, -a ARCH  model architecture: resnet18 | resnet34 | resnet50 |
                        resnet101 | resnet152 | resnext101-32x4d | se-
                        resnext101-32x4d (default: resnet50)
  --model-config CONF, -c CONF
                        model configs: classic | fanin | grp-fanin | grp-
                        fanout(default: classic)
  --num-classes N       number of classes in the dataset
  -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-weights PATH
                        load weights from here
  --fp16                Run model fp16 mode.
  --static-loss-scale STATIC_LOSS_SCALE
                        Static loss scale, positive power of 2 values can
                        improve fp16 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 add a custom config in the image_classification/resnet.py file.

resnet_versions = {
    ...
    'resnet50-custom' : {
       'net' : ResNet,
       'block' : Bottleneck,
       'layers' : [3, 4, 6, 3],
       'widths' : [64, 128, 256, 512],
       'expansion' : 4,
       'num_classes' : <custom number of classes>,
       }
}

After adding the config, run the training script with --arch resnet50-custom flag.

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

Inference process

Validation is done every epoch, and can be also run separately on a checkpointed model.

python ./main.py --arch resnet50 --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 resnet50 -c fanin --weights <path to weights from previous step> --precision AMP|FP32 --image <path to JPEG image>

Performance

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 python ./main.py --arch resnet50 -b <batch_size> --training-only -p 1 --raport-file benchmark.json --epochs 1 --prof 100 <path to imagenet>
  • AMP python ./main.py --arch resnet50 -b <batch_size> --training-only -p 1 --raport-file benchmark.json --epochs 1 --prof 100 --amp --static-loss-scale 256 <path to imagenet>
• For multiple GPUs
  • FP32 python ./multiproc.py --nproc_per_node 8 ./main.py --arch resnet50 -b <batch_size> --training-only -p 1 --raport-file benchmark.json --epochs 1 --prof 100 <path to imagenet>
  • AMP python ./multiproc.py --nproc_per_node 8 ./main.py --arch resnet50 -b <batch_size> --training-only -p 1 --raport-file benchmark.json --amp --static-loss-scale 256 --epochs 1 --prof 100 <path to imagenet>

Each of these scripts will run 100 iterations and save results in the benchmark.json file.

Batch size should be picked appropriately depending on the hardware configuration.

Platform	Precision	Batch Size
DGXA100	AMP	256
DGXA100	TF32	256
DGX-1	AMP	256
DGX-1	FP32	128

Inference performance benchmark

To benchmark inference, run:

• FP32

python ./main.py --arch resnet50 -p 1 --raport-file benchmark.json --epochs 1 --prof 100 --evaluate <path to imagenet>

• AMP

python ./main.py --arch resnet50 -p 1 --raport-file benchmark.json --epochs 1 --prof 100 --evaluate --amp <path to imagenet>

Each of these scripts will run 100 iterations and save results in the benchmark.json file.

Batch size should be picked appropriately depending on the hardware configuration.

Platform	Precision	Batch Size
DGXA100	AMP	256
DGXA100	TF32	256
DGX-1	AMP	256
DGX-1	FP32	128

Results

Our results were obtained by running the applicable training script in the pytorch-20.06 NGC container.

To achieve these same results, follow the steps in the Quick Start Guide.

Training accuracy results

Training accuracy: NVIDIA DGX A100 (8x A100 40GB)

epochs	Mixed Precision Top1	TF32 Top1
90	76.93 +/- 0.23	76.85 +/- 0.30

Training accuracy: NVIDIA DGX-1 (8x V100 16GB)

epochs	Mixed Precision Top1	FP32 Top1
50	76.25 +/- 0.04	76.26 +/- 0.07
90	77.09 +/- 0.10	77.01 +/- 0.16
250	78.42 +/- 0.04	78.30 +/- 0.16

Training accuracy: NVIDIA DGX-2 (16x V100 32GB)

epochs	Mixed Precision Top1	FP32 Top1
50	75.81 +/- 0.08	76.04 +/- 0.05
90	77.10 +/- 0.06	77.23 +/- 0.04
250	78.59 +/- 0.13	78.46 +/- 0.03

Example plots

The following images show a 250 epochs configuration on a DGX-1V.

Training performance results

Training performance: NVIDIA DGX A100 (8x A100 40GB)

GPUs	Mixed Precision	TF32	Mixed Precision Speedup	Mixed Precision Strong Scaling	Mixed Precision Training Time (90E)	TF32 Strong Scaling	TF32 Training Time (90E)
1	1240.81 img/s	680.15 img/s	1.82x	1.00x	~27 hours	1.00x	~49 hours
8	9604.92 img/s	5379.82 img/s	1.79x	7.74x	~4 hours	7.91x	~6 hours

Training performance: NVIDIA DGX-1 16GB (8x V100 16GB)

GPUs	Mixed Precision	FP32	Mixed Precision Speedup	Mixed Precision Strong Scaling	Mixed Precision Training Time (90E)	FP32 Strong Scaling	FP32 Training Time (90E)
1	856.52 img/s	373.21 img/s	2.30x	1.00x	~39 hours	1.00x	~89 hours
8	6635.90 img/s	2899.62 img/s	2.29x	7.75x	~5 hours	7.77x	~12 hours

Training performance: NVIDIA DGX-1 32GB (8x V100 32GB)

GPUs	Mixed Precision	FP32	Mixed Precision Speedup	Mixed Precision Strong Scaling	Mixed Precision Training Time (90E)	FP32 Strong Scaling	FP32 Training Time (90E)
1	816.00 img/s	359.76 img/s	2.27x	1.00x	~41 hours	1.00x	~93 hours
8	6347.26 img/s	2813.23 img/s	2.26x	7.78x	~5 hours	7.82x	~12 hours

Inference performance results

Inference performance: NVIDIA DGX-1 (1x V100 16GB)

FP32 Inference Latency

batch size	Throughput Avg	Latency Avg	Latency 90%	Latency 95%	Latency 99%
1	136.82 img/s	7.12ms	7.25ms	8.36ms	10.92ms
2	266.86 img/s	7.27ms	7.41ms	7.85ms	9.11ms
4	521.76 img/s	7.44ms	7.58ms	8.14ms	10.09ms
8	766.22 img/s	10.18ms	10.46ms	10.97ms	12.75ms
16	976.36 img/s	15.79ms	15.88ms	15.95ms	16.63ms
32	1092.27 img/s	28.63ms	28.71ms	28.76ms	29.30ms
64	1161.55 img/s	53.69ms	53.86ms	53.90ms	54.23ms
128	1209.12 img/s	104.24ms	104.68ms	104.80ms	105.00ms
256	N/A	N/A	N/A	N/A	N/A

Mixed Precision Inference Latency

batch size	Throughput Avg	Latency Avg	Latency 90%	Latency 95%	Latency 99%
1	114.97 img/s	8.56ms	9.32ms	11.43ms	12.79ms
2	238.70 img/s	8.20ms	8.75ms	9.49ms	12.31ms
4	448.69 img/s	8.67ms	9.20ms	9.97ms	10.60ms
8	875.00 img/s	8.88ms	9.31ms	9.80ms	10.82ms
16	1746.07 img/s	8.89ms	9.05ms	9.56ms	12.81ms
32	2004.28 img/s	14.07ms	14.14ms	14.31ms	14.92ms
64	2254.60 img/s	25.93ms	26.05ms	26.07ms	26.17ms
128	2360.14 img/s	50.14ms	50.28ms	50.34ms	50.68ms
256	2342.13 img/s	96.74ms	96.91ms	96.99ms	97.14ms

Inference performance: NVIDIA T4

FP32 Inference Latency

batch size	Throughput Avg	Latency Avg	Latency 90%	Latency 95%	Latency 99%
1	179.85 img/s	5.51ms	5.65ms	7.34ms	10.97ms
2	348.12 img/s	5.67ms	5.95ms	6.33ms	9.81ms
4	556.27 img/s	7.03ms	7.34ms	8.13ms	9.65ms
8	740.43 img/s	10.32ms	10.33ms	10.60ms	13.87ms
16	909.17 img/s	17.19ms	17.15ms	18.13ms	21.06ms
32	999.07 img/s	31.07ms	31.12ms	31.17ms	32.41ms
64	1090.47 img/s	57.62ms	57.84ms	57.91ms	58.05ms
128	1142.46 img/s	110.94ms	111.15ms	111.23ms	112.16ms
256	N/A	N/A	N/A	N/A	N/A

Mixed Precision Inference Latency

batch size	Throughput Avg	Latency Avg	Latency 90%	Latency 95%	Latency 99%
1	163.78 img/s	6.05ms	5.92ms	7.98ms	11.58ms
2	333.43 img/s	5.91ms	6.05ms	6.63ms	11.52ms
4	645.45 img/s	6.04ms	6.33ms	7.01ms	8.90ms
8	1164.15 img/s	6.73ms	7.31ms	8.04ms	12.41ms
16	1606.42 img/s	9.53ms	9.86ms	10.52ms	17.01ms
32	1857.29 img/s	15.67ms	15.61ms	16.14ms	18.66ms
64	2011.62 img/s	28.64ms	28.69ms	28.82ms	31.06ms
128	2083.90 img/s	54.87ms	54.96ms	54.99ms	55.27ms
256	2043.72 img/s	106.51ms	106.62ms	106.68ms	107.03ms

Release notes

Changelog

1. September 2018

• Initial release

2. January 2019

• Added options Label Smoothing, fan-in initialization, skipping weight decay on batch norm gamma and bias.

3. May 2019

• Cosine LR schedule
• MixUp regularization
• DALI support
• DGX2 configurations
• gradients accumulation

4. July 2019

• DALI-CPU dataloader
• Updated README

5. July 2020

• Added A100 scripts
• Updated README

Known issues

There are no known issues with this model.