Arm/vit-base-int8-xnnpack-executorch

google/vit-base-patch16-224 optimized for Arm-based Edge Linux systems

An INT8-quantized version of google/vit-base-patch16-224 for image classification, exported to ExecuTorch (.pte) and optimized for Arm-based Edge Linux systems.

Summary

This repository contains an Arm-optimized version of google/vit-base-patch16-224 for image classification, quantized to INT8 via dynamic post-training quantization — per-channel symmetric weights, with activation scales computed dynamically at inference (no calibration data required). The model is provided in ExecuTorch (.pte) format, targeting Edge Linux systems.

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, Raspberry Pi OS 64-bit, based on Debian 13 "Trixie")
Primary performance result	p50 latency 378.139 ms (2.91x faster than baseline); 2.64 fps
Accuracy result	Top-1 81.50%, Top-5 96.06%
Size / memory result	83.926 MB (3.94x smaller), peak memory 110.05 MB

Original model

Field	Value
Original model	google/vit-base-patch16-224
Original source	Hugging Face
Original developer	Google Research
Original model card	google/vit-base-patch16-224
Original license	Apache-2.0

Model files

File	Description
google__vit-base-patch16-224_raspberry_executorch_optimized.pte	Arm-optimized INT8 model for deployment
example.py	Minimal inference example
pyproject.toml	Pinned runtime dependencies for example.py, resolved with uv
uv.lock	Locked dependency resolution for pyproject.toml
config.yaml	Model I/O contract used by the example
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	Cortex-A76, 4 cores @ 2.4 GHz
System memory	8 GB
OS	Raspberry Pi OS 64-bit, based on Debian 13 "Trixie"
Runtime	ExecuTorch 1.1.0
Execution backend	CPU (XNNPACK, KleidiAI)
Precision	INT8, dynamic PTQ — per-channel symmetric weights; activations quantized dynamically at inference
Batch size	1
Input resolution	224x224
Runs / warmup	10 / 10

Performance results

Metric	Original / baseline	Arm-optimized	Improvement
Model size (MB)	330.378	83.926	3.94x smaller
End-to-end latency p50 (ms)	1101.028	378.139	2.91x faster
End-to-end latency p90 (ms)	1102.087	379.857	2.90x faster
End-to-end latency p99 (ms)	1104.819	388.308	2.84x faster
Model load time (ms)	269.401	244.45	1.1x faster
Time to first inference (ms)	1128.776	378.773	2.98x faster
Peak memory (MB)	394.343	110.05	3.58x less
Average memory (MB)	394.281	107.984	3.65x less
Frames per second	0.908	2.64	2.90x

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 1.1.0 (CPU, XNNPACK, KleidiAI)

Accuracy results

Metric	Original / baseline	Arm-optimized	Change
Top-1 accuracy (%)	81.52	81.50	-0.02 pp
Top-5 accuracy (%)	96.06	96.06	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	Converted to ExecuTorch .pte
Quantization	Yes	INT8 dynamic PTQ — per-channel symmetric weights; activation scales computed dynamically at inference, no calibration data required
Runtime/backend selection	Yes	XNNPACK + KleidiAI (CPU)
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 shorter edge to 232 (bilinear), center-crop to 224x224, convert to tensor, normalize (mean [0.5, 0.5, 0.5], std [0.5, 0.5, 0.5])

Expected output

Property	Value
Output shape	[1, 1000]
Output type	Raw class logits (unnormalized)
Postprocessing	softmax, 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 50,000 images from the ImageNet-1k validation split and may not generalize to all domains.
• The model expects a fixed 224x224 input resolution — no dynamic input size support.
• This repository is not a replacement for the original model documentation.

Additional notes

• The export pipeline captures the FP32 model via torch.export, inserts dynamic-quantization observers, converts them to Q/DQ pairs, and exports the result to ExecuTorch (.pte) targeting the XNNPACK backend. All encoder Linear layers are quantized uniformly; none are kept in FP32.
• Sample input: sample_input.jpg is derived from Samoyed on Beach by Appleinfl, via Wikimedia Commons (public domain).

About this version

Original Model: google/vit-base-patch16-224 by Google Research - 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.