Classify pet images with DeiT-Tiny and Arm VGF using ExecuTorch
Introduction
Understand the DeiT-Tiny deployment workflow
Prepare ExecuTorch and build the VGF runner
Fine-tune DeiT-Tiny on pet images
Quantize and export DeiT-Tiny to VGF
Classify a pet image and verify VGF execution
Next Steps

  Log an issue

  Fork and edit

  Discuss on Discord

Export the quantized model

Convert your fine-tuned checkpoint into a quantized ExecuTorch .pte program with VGF delegate data. Calibration uses 300 training images to determine quantization parameters. A separate host accuracy check evaluates breed predictions on 100 test images:

    

        
        
python examples/arm/image_classification_example_vgf/model_export/export_deit.py \
  --model-path arm_test/deit_vgf/deit-tiny-oxford-pet/final_model \
  --output-path arm_test/deit_vgf/deit_quantized_vgf.pte \
  --num-calibration-samples 300 \
  --num-test-samples 100 \
  2>&1 | tee arm_test/deit_vgf/export.log

The script does the following:

  • Exports the floating-point graph
  • Calibrates symmetric INT8 post-training quantization
  • Evaluates the quantized model in PyTorch
  • Delegates supported operations through the Arm VGF backend
  • Writes the .pte file

Check the export result

Find the accuracy result and the export confirmation in the log, then check that the program exists:

    

        
        
grep -E 'Top-1 accuracy|Exported model saved' arm_test/deit_vgf/export.log
test -s arm_test/deit_vgf/deit_quantized_vgf.pte

The output is similar to:

    

        
        Top-1 accuracy on 100 test samples: 0.8900
Exported model saved to arm_test/deit_vgf/deit_quantized_vgf.pte

The example accuracy of 0.8900 means the highest-scoring breed matches the dataset label for 89 of the 100 test images. Your result can differ. A successful export also reports the output path, and test -s exits successfully when that file is nonempty.

The reported accuracy measures the quantized PyTorch model before VGF execution. The training log evaluates a different number of test images, so those two values alone don’t measure the accuracy change caused by quantization. Use the same evaluation images when investigating that change.

The .pte includes its VGF delegate data. You don’t need to supply a separate .vgf file to the ExecuTorch runner.

What you’ve accomplished and what’s next

You’ve produced a quantized VGF-backed program and recorded its host accuracy.

Next, you’ll classify a pet image with the runner that you built during setup.

Back
Next