Arm/deit-tiny-int8-litert

DeiT-Tiny Patch16-224 optimized for Arm-based mobile CPUs with SME2

DeiT-Tiny Patch16-224 optimized with INT8 static post-training quantization for image classification, exported to LiteRT .tflite format for efficient inference on Arm-based mobile CPUs with SME2.

Summary

This repository contains an Arm-optimized version of facebook/deit-tiny-patch16-224 for image classification, quantized to INT8 via static post-training quantization — affine per-tensor activations, symmetric per-channel weights — while patch embeddings remain in FP32. The model is provided in LiteRT .tflite format, targeting Mobile CPU 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	LiteRT .tflite
Target device class	Mobile CPU
Reference device	vivo X300 (C1-Ultra, C1-Premium, C1-Pro, Android 16 / OriginOS 6)
Example environment	AWS Graviton, Ubuntu arm64
Primary performance result	p50 latency of 5.36 ms (1.74x faster than baseline)
Accuracy result	Top-1 accuracy of 71.19 percent (-0.94 pp vs baseline)
Size / memory result	6.88 MB (3.23x smaller than baseline)

Original model

Field	Value
Original model	facebook/deit-tiny-patch16-224
Original source	Hugging Face
Original developer	Meta
Original model card	facebook/deit-tiny-patch16-224
Original license	Apache-2.0

Model files

File	Description
facebook__deit-tiny-patch16-224_litert_optimized.tflite	Arm-optimized 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
benchmarks/	FP32 baseline and Arm-optimized benchmark records

Performance

Performance was measured on the vivo X300 reference configuration below. The Graviton environment documented for example.py verifies portability of the inference flow but does not provide or imply equivalent Graviton performance.

Reference configuration

Field	Value
Device / platform	vivo X300
CPU / accelerator	C1-Ultra, C1-Premium, C1-Pro (8-core aarch64)
OS	android, Android 16 / OriginOS 6
Runtime	LiteRT
Backend / delegate	XNNPACK, KleidiAI
Batch size	1
Precision	INT8 static PTQ — affine per-tensor activations, symmetric per-channel weights (patch embeddings kept in FP32)
Runs	20 warmup + 100 measured

Performance results

Metric	Original / baseline	Arm-optimized	Improvement
p50 latency	9.353 ms	5.361 ms	1.74x
p90 latency	9.746 ms	5.884 ms	1.66x
Model size	22.262 MB	6.883 MB	3.23x smaller
Model load time	59.503 ms	33.953 ms	1.75x
Time to first inference	14.379 ms	7.145 ms	2.01x
Peak memory	59.20 MB	30.47 MB	1.94x less
Average memory	59.18 MB	30.46 MB	1.94x less
Frames per second	106.92	186.53	1.74x

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	LiteRT

Accuracy results

Metric	Original / baseline	Arm-optimized	Change
Top-1 accuracy	72.13%	71.19%	-0.94 pp
Top-5 accuracy	91.13%	90.50%	-0.63 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 LiteRT .tflite
Quantization	Yes	Static PTQ with INT8 affine per-tensor activations and INT8 symmetric per-channel weights, calibrated on 100 ImageNet-1k samples; patch embeddings are excluded and remain in FP32
Runtime/backend selection	Yes	Selected XNNPACK and KleidiAI
Graph/runtime compatibility updates	Yes	Performed as part of the LiteRT 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

Note: The Python/uv example runs on AWS Graviton (Ubuntu arm64) to confirm runtime compatibility only, and is intended as a guideline for building an equivalent run script on Mobile CPU systems.

Expected input

Property	Value
Input shape	1x3x224x224
Input type	float32
Input range	0.0 to 1.0
Preprocessing	Resize shorter edge to 256 (bicubic), center crop to 224x224, convert to tensor, normalize with ImageNet mean/std

Expected output

Property	Value
Output shape	1x1000
Output type	Raw classification logits (unnormalized), one score per ImageNet class
Postprocessing	Softmax followed by top-5 selection

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.
• Performance was measured on a vivo X300; other Arm devices may run slower or faster and may have different XNNPACK or KleidiAI delegation coverage.
• The Python/uv example targets AWS Graviton on Ubuntu arm64. It establishes runtime compatibility only; no Graviton latency or memory benchmark is reported here.

Additional notes

Patch embeddings are kept in FP32 on this export. The model's quantization spec excludes this layer to avoid a static-transpose graph pattern that the XNNPACK delegate cannot compile on the ExecuTorch export; that same declarative skip list is shared across backends, so it also applies to this LiteRT export even though LiteRT itself has no such compile constraint. Latency and memory were measured with the model pinned to a single CPU core on the vivo X300.

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

About this version

Original Model: facebook/deit-tiny-patch16-224 by Meta - 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.