DalšíDalší průvodce
torch.compile a kompilace grafů PyTorch 2
Technický
Technický PRŮVODCE
LiteRT is the current name used in Google AI Edge documentation for the on-device runtime historically known as TensorFlow Lite.
Poslechněte si tento návodAI-vyprávěný · ~3 min
It supports model conversion and mobile inference with hardware acceleration options, but operators, delegates, model formats, and device capabilities must be checked for the target platform and runtime version.
TensorFlow Lite has been renamed LiteRT in current Google AI Edge documentation, while existing apps and model artifacts may still use TensorFlow Lite terminology. The naming transition does not mean every application must rewrite its code immediately; follow migration guidance for the specific runtime and API being used. Older tutorials may refer to Interpreter APIs, while current docs also describe newer compiled-model paths. A mobile workflow begins with a trained model and a conversion step that produces a supported on-device representation. Conversion can include optimization or quantization. The result must be checked for compatible operators, input shapes, tensor types, and metadata. If conversion reports unsupported operations, a model may need export changes, custom operations, or a different runtime path. At inference time, LiteRT can execute on CPU and may use delegates for supported GPU or NPU hardware. A delegate accelerates operations it supports; unsupported parts may fall back to another device or fail depending on configuration. Availability varies across device models, chipset vendors, runtime packaging, and operating-system versions. Do not assume an NPU delegate is present simply because the phone contains an NPU. Quantization reduces numeric precision and can shrink model size or improve speed, but may change accuracy. Calibration data should represent the intended input distribution for supported post-training quantization methods. Test both resource use and task metrics after conversion. Image color order, normalization, tokenizer settings, and output decoding need to match training. On-device deployment also involves app lifecycle, memory pressure, battery use, installation size, privacy, and version support. Benchmark on physical target devices, including startup and sustained runs. Keep model version and conversion configuration with the app release. LiteRT is a runtime, not a guarantee that a model will be accurate or compatible with every phone.
An Android developer converts an image classifier and compares CPU inference with an available GPU or NPU delegate.
A team quantizes a model to reduce memory and tests accuracy on representative phone images before release.
A mobile app selects the LiteRT runtime documented for its chosen API and avoids mixing examples from older TensorFlow Lite versions.
An engineer measures cold load, warm latency, memory and energy on several target devices rather than relying only on desktop benchmarks.
Google AI Edge tooling will continue evolving model conversion, mobile APIs, and hardware delegation. Existing TensorFlow Lite applications may keep working while new documentation and packaging use LiteRT names. Teams should track migration guidance and test runtime upgrades on their supported device matrix. Hardware acceleration will remain device-specific, so portability requires measured fallbacks and clear minimum requirements. Runtime names and APIs will continue changing as mobile hardware evolves. Keep migration work separate from model quality checks, and maintain benchmarks across representative device classes. Delegate support should be rechecked before updates ship.
Newsletter zdarma
Tři ověřené příběhy ze světa AI každý všední den ráno, psané srozumitelnou angličtinou. Navždy zdarma, bez reklam.
Jeden e-mail každý všední den. Odhlásit se můžete jedním kliknutím. Vaši adresu nikdy neprodáváme ani nesdílíme.
Otestujte se
Okamžitá zpětná vazba ke každé odpovědi a sdílitelný certifikát s ověřitelným ID po absolvování kurzu.
Podpořte bezplatnou AI vzdělávání. AI Understanding je nezisková organizace 501(c)(3) — žádné reklamy, žádná platba za přístup, nikdy. Darujte
LiteRT is the current name used in Google AI Edge documentation for the on-device runtime historically known as TensorFlow Lite. It supports model conversion and mobile inference with hardware acceleration options, but operators, delegates, model formats, and device capabilities must be checked for the target platform and runtime version.
Google's current AI Edge documentation uses LiteRT for the on-device framework previously known as TensorFlow Lite.
Delegates can accelerate supported operations on available hardware.
Model operations must be supported by the converter and chosen runtime path.
Reduced precision can affect predictions and should be validated.
Hardware vendor, runtime version and operation support determine whether delegation works.
Učte se dál
Pro toto téma bylo vybráno více průvodců
DalšíDalší průvodce
torch.compile a kompilace grafů PyTorch 2
Technický