ixrt

ResNet50 (ixRT)

Model Description

Residual Networks, or ResNets, learn residual functions with reference to the layer inputs, instead of learning unreferenced functions. Instead of hoping each few stacked layers directly fit a desired underlying mapping, residual nets let these layers fit a residual mapping.

Supported Environments

GPU	IXUCA SDK	Release	Branch
MR-V100	4.4.0	26.03	release/26.03
MR-V100	4.3.0	25.12	release/25.12

Note: 请切换到与您的 SDK 版本对应的 Release 分支进行测试。请勿直接在 master 分支上运行测试，因为 master 分支可能包含与您的本地 SDK 版本不兼容的最新更改。

切换分支命令示例：git checkout release/26.03

Model Preparation

Prepare Resources

Pretrained model: https://download.pytorch.org/models/resnet50-0676ba61.pth

Dataset: https://www.image-net.org/download.php to download the validation dataset.

Install Dependencies

# Install libGL
## CentOS
yum install -y mesa-libGL
## Ubuntu
apt install -y libgl1-mesa-glx

pip3 install -r ../../ixrt_common/requirements.txt

Model Conversion

mkdir checkpoints
python3 ../../ixrt_common/export.py --model-name resnet50 --weight resnet50-0676ba61.pth --output checkpoints/resnet50.onnx

Model Inference

export PROJ_DIR=./
export DATASETS_DIR=/path/to/imagenet_val/
export CHECKPOINTS_DIR=./checkpoints
export RUN_DIR=../../ixrt_common/
export CONFIG_DIR=../../ixrt_common/config/RESNET50_CONFIG

FP16

# Accuracy
bash scripts/infer_resnet50_fp16_accuracy.sh
# Performance
bash scripts/infer_resnet50_fp16_performance.sh

INT8

# Accuracy
bash scripts/infer_resnet50_int8_accuracy.sh
# Performance
bash scripts/infer_resnet50_int8_performance.sh

Model Results

Model	BatchSize	Precision	FPS	Top-1(%)	Top-5(%)
ResNet50	32	FP16	4077.58	0.76158	0.92872
ResNet50	32	INT8	9113.07	0.74516	0.9287