Skip to content

Latest commit

 

History

History
101 lines (85 loc) · 4.55 KB

README.md

File metadata and controls

101 lines (85 loc) · 4.55 KB
Raw

AlexNet

Model Download Download (with sample test data) ONNX version Opset version Top-1 accuracy (%) Top-5 accuracy (%)
AlexNet 238 MB 225 MB 1.1 3
AlexNet 238 MB 225 MB 1.1.2 6
AlexNet 238 MB 226 MB 1.2 7
AlexNet 238 MB 226 MB 1.3 8
AlexNet 238 MB 226 MB 1.4 9
AlexNet 233 MB 216 MB 1.9 12 54.80 78.23
AlexNet-int8 58 MB 39 MB 1.9 12 54.68 78.23
AlexNet-qdq 59 MB 44 MB 1.9 12 54.71 78.22

Compared with the fp32 AlextNet, int8 AlextNet's Top-1 accuracy drop ratio is 0.22%, Top-5 accuracy drop ratio is 0.05% and performance improvement is 2.26x.

Note

Different preprocess methods will lead to different accuracies, the accuracy in table depends on this specific preprocess method.

The performance depends on the test hardware. Performance data here is collected with Intel® Xeon® Platinum 8280 Processor, 1s 4c per instance, CentOS Linux 8.3, data batch size is 1.

Description

AlexNet is the name of a convolutional neural network for classification, which competed in the ImageNet Large Scale Visual Recognition Challenge in 2012.

Differences:

  • not training with the relighting data-augmentation;
  • initializing non-zero biases to 0.1 instead of 1 (found necessary for training, as initialization to 1 gave flat loss).

Dataset

ILSVRC2012

Source

Caffe BVLC AlexNet ==> Caffe2 AlexNet ==> ONNX AlexNet

Model input and output

Input

data_0: float[1, 3, 224, 224]

Output

softmaxout_1: float[1, 1000]

Pre-processing steps

Post-processing steps

Sample test data

Randomly generated sample test data:

  • test_data_0.npz
  • test_data_1.npz
  • test_data_2.npz
  • test_data_set_0
  • test_data_set_1
  • test_data_set_2

Results/accuracy on test set

The bundled model is the iteration 360,000 snapshot. The best validation performance during training was iteration 358,000 with validation accuracy 57.258% and loss 1.83948. This model obtains a top-1 accuracy 57.1% and a top-5 accuracy 80.2% on the validation set, using just the center crop. (Using the average of 10 crops, (4 + 1 center) * 2 mirror, should obtain a bit higher accuracy.)

Quantization

AlexNet-int8 and AlexNet-qdq are obtained by quantizing fp32 AlexNet model. We use Intel® Neural Compressor with onnxruntime backend to perform quantization. View the instructions to understand how to use Intel® Neural Compressor for quantization.

Environment

onnx: 1.9.0 onnxruntime: 1.8.0

Prepare model

wget https://github.com/onnx/models/raw/main/vision/classification/alexnet/model/bvlcalexnet-12.onnx

Model quantize

Make sure to specify the appropriate dataset path in the configuration file.

bash run_tuning.sh --input_model=path/to/model \  # model path as *.onnx
                   --config=alexnet.yaml \
                   --data_path=/path/to/imagenet \
                   --label_path=/path/to/imagenet/label \
                   --output_model=path/to/save

References

Contributors

License

BSD-3