Arm/swin-tiny-int8-xnnpack-executorch

Swin Tiny optimized for Arm-based Edge Linux

Swin Transformer Tiny image classification quantized to INT8 with dynamic post-training quantization and exported to ExecuTorch .pte format for efficient inference on Arm-based Edge Linux systems.

Summary

This repository contains an Arm-optimized version of microsoft/swin-tiny-patch4-window7-224 for image classification, quantized to INT8 via dynamic PTQ — per-channel symmetric weights, with activation scales computed at runtime. The model is provided in ExecuTorch .pte format, targeting Edge Linux systems.

Swin Tiny is a hierarchical vision transformer that computes self-attention inside shifted local windows, which gives it a convolution-like multi-scale feature pyramid at linear cost in image size.

This version is intended to demonstrate efficient inference on Arm-based platforms while preserving the original model's intended behavior. Arm has evaluated this model on ImageNet-1k and measured performance on a representative evaluation target.

Key results

Area	Result
Model format	ExecuTorch .pte
Target device class	Edge Linux
Reference device	Raspberry Pi 5 (Cortex-A76, Linux Raspberry Pi OS 64-bit, based on Debian 13 "Trixie")
Primary performance result	85.183 ms p50 latency (11.74 FPS)
Accuracy result	Top-1 81.07%, Top-5 95.54%
Size / memory result	28.486 MB (3.83x smaller than the FP32 baseline), peak memory 56.41 MB

Original model

Field	Value
Original model	microsoft/swin-tiny-patch4-window7-224
Original source	Hugging Face
Original developer	Microsoft
Original model card	microsoft/swin-tiny-patch4-window7-224
Original license	Apache-2.0

Model files

File	Description
swin_tiny_dynamic_raspberry_executorch_optimized.pte	Arm-optimized INT8 model for deployment
example.py	Minimal inference example
pyproject.toml	Pinned runtime dependencies and deployment metadata
uv.lock	Locked dependency versions, sources, and hashes for uv sync --frozen
config.yaml	Model I/O contract used by the example
sample_input.jpg	Input image used by example.py
benchmarks/	FP32 baseline and Arm-optimized benchmark records

Performance

Performance was measured on the reference configuration below. Results are intended to make the optimization reproducible but do not guarantee identical performance on every Arm-based system.

Reference configuration

Field	Value
Device / platform	Raspberry Pi 5
CPU / accelerator	Cortex-A76 (arm64, 4 cores, 2.4 GHz)
OS	Linux, Raspberry Pi OS 64-bit, based on Debian 13 "Trixie"
Runtime	ExecuTorch 1.1.0
Backend / delegate	XNNPACK, KleidiAI
Batch size	1
Precision	INT8 (per-channel symmetric weights, INT8 activations quantized dynamically at runtime)
Runs	10 warmup, 100 measured

Performance results

Metric	Original / baseline	Arm-optimized	Improvement
p50 latency	130.808 ms	85.183 ms	1.54x
p90 latency	132.572 ms	86.684 ms	1.53x
p99 latency	146.100 ms	87.938 ms	1.66x
Throughput	7.64 FPS	11.74 FPS	1.54x
Model size	108.98 MB	28.486 MB	3.83x smaller
Peak memory	148.61 MB	56.41 MB	2.63x less

Accuracy

Accuracy was evaluated using the same preprocessing, input resolution, and evaluation protocol described below. Where possible, the optimized model is compared against the original model under the same evaluation conditions.

Evaluation setup

Field	Value
Dataset	ImageNet-1k
Split	val
Number of samples	50000
Metric(s)	Top-1 accuracy, Top-5 accuracy
Evaluation runtime	ExecuTorch

Accuracy results

Metric	Original / baseline	Arm-optimized	Change
Top-1 accuracy	81.09%	81.07%	-0.02 pp
Top-5 accuracy	95.54%	95.54%	+0.00 pp

Accuracy was measured using the evaluation setup described above. Users should re-evaluate the model on their own data before production use.

Arm optimization approach

Arm optimized this model for efficient inference on Arm-based platforms using a hardware-aware conversion and validation flow.

For this release, Arm used:

Optimization area	Applied?	Notes
Model conversion	Yes	Captured with torch.export and lowered to ExecuTorch .pte with XNNPACK delegation
Quantization	Yes	Quantized to INT8 via dynamic PTQ — per-channel symmetric weights folded into the program, with activation scales computed at runtime from each input. Dynamic quantization needs no calibration dataset. No layers were skipped or kept at higher precision
Runtime/backend selection	Yes	XNNPACK with KleidiAI
Graph/runtime compatibility updates	Yes	Performed as part of the ExecuTorch export pipeline
Accuracy validation	Yes	Compared against the original model or published baseline
Performance validation	Yes	Measured on the reference Arm platform

The goal of this process is to improve deployment characteristics such as latency, memory use, model size, and runtime compatibility while preserving the model's intended behavior. Detailed conversion scripts, calibration configuration, or backend-specific implementation details may be provided separately where appropriate.

Using this model

Install dependencies

Dependencies are declared in pyproject.toml, which ships with this repository. Resolve and install them into a local virtual environment with uv:

uv python install
uv sync --frozen

Run the example

uv run example.py

Expected input

Property	Value
Input shape	[1, 3, 224, 224]
Input type	float32
Input range	[0.0, 1.0]
Preprocessing	Resize shortest edge to 232 (bilinear), center crop to 224x224, convert to tensor, normalize with ImageNet mean/std

Expected output

Property	Value
Output shape	[1, 1000]
Output type	Raw class logits (unnormalized)
Postprocessing	Softmax, then top-5

Intended use

This model is intended for developers evaluating image classification workloads on Arm-based platforms. It is suitable as a reference implementation for benchmarking, prototyping, and integration exploration.

Limitations

• Performance depends on the target device, runtime version, backend/delegate support, memory configuration, and system load.
• Accuracy was evaluated on ImageNet-1k val and may not generalize to all domains.
• This release preserves the original model's intended task and behavior, but users should validate it for their own application, data, and deployment environment.
• This repository is not a replacement for the original model documentation.
• The example expects a fixed 224x224 input after resize and center crop. There is no letterboxing or aspect-ratio-preserving padding, so an image far from square loses content at the edges.

Additional notes

• Sample input: sample_input.jpg is derived from Samoyed on Beach by Appleinfl, via Wikimedia Commons (public domain).

About this version

Original Model: microsoft/swin-tiny-patch4-window7-224 by Microsoft - Repository

Optimization/conversion: Arm-Optimized version for execution on Arm-based platforms.

Converted/optimized by: Arm

License: The Original Model and the Optimized Model are subject to Apache-2.0.

This repository contains a converted or optimized version of the Original Model (the “Optimized Model”). The Original Model has been converted or optimized as described above for execution on Arm-based platforms.

No retraining or fine-tuning of the Original Model was performed as part of the conversion or optimization. The conversion or optimization was not intended to change the Original Model’s behavior or intended use.

Original Model and Documentation

For information about the Original Model, including its development, training data, intended uses, limitations and other relevant information, please refer to the Original Model repository. Information in that repository was provided by the original developer or other third parties and, unless expressly stated otherwise, has not been independently verified by Arm.

Licenses and Third-Party Terms

Use of the Original Model and the Optimized Model is subject to the applicable licenses, usage restrictions and other terms identified above and in the relevant repositories. Publication of the Optimized Model does not grant any rights beyond those provided under the applicable license terms.

You are responsible for reviewing those terms and ensuring that your use of the Original Model and the Optimized Model is permitted.

Purpose of this Release

The Optimized Model is provided as a reference implementation to demonstrate and evaluate execution and performance on Arm-based systems. It is not a production-ready or supported solution.

Arm’s publication of the Optimized Model does not constitute an endorsement or certification of the Original Model or a representation that the Optimized Model is suitable for production use or any particular purpose.

To the fullest extent permitted by applicable law (i) the Optimized Model is provided “as is.” Arm makes no representations or warranties that the Original Model, the Optimized Model or their outputs are accurate, safe, secure, non-infringing, legally compliant, suitable for production use or fit for any particular purpose; and (ii) Arm will not be liable for any loss or damage arising from or in connection with the Optimized Model, its use or its outputs.

You are responsible for independently evaluating the Optimized Model, its outputs and its suitability for your intended use, including compliance with applicable legal, regulatory, safety and security requirements.

Arm does not commit to provide ongoing support, maintenance or updates for the Optimized Model. Any use of or reliance on the Optimized Model or its outputs is at your own risk.