Vengineerの戯言

人生は短いけど、長いです。人生を楽しみましょう!

TensorFlow Liteの新しいビデオが二つ

@Vengineerの戯言 : Twitter
SystemVerilogの世界へようこそすべては、SystemC v0.9公開から始まった

TensorFlow Liteの新しいビデオ

www.youtube.com

さくっと、まとめてみました。

ruy/gemmlowp

ruyは、これ ですね。

ruy is a matrix multiplication library. Its focus is to cover the matrix multiplication needs of TensorFlow Lite.

ruy supports both floating-point (like Eigen) and quantized (like gemmlowp).

 

Acceleration (Delegates)

 Android

  • NN API
  • GPU
  • Edge TPU
  • Hexagon DSP (Experimental)
  • Experimental convolutional backend (Experimental)

 iOS

  • Metal

TensorFlow Select

 Select TensorFlow operators to use in TensorFlow Lite

 

Cross-platform bindings

 APIs : Python, C++, Java

 Experimental APIs : C, C# (Unity), Objective C, Swift

 

Performance

 Benchmarking : Model Benchmark tool, (Evaluation tool (+app), MLPerf)

 Acceleration : Edge TPU, GPU, NN API, (CPU optimizations (arm => ruy & x86))

 

Optimization

 Quantization : Legacy quantized training (CPU/NPU), Post-training quantization (CPU/NPU), (Keras-based quantized training (CPU/NPU))

  ”Hybrid” quantization (Post-training quantization with float & fixed point)

  Integer quantization (Post-training quantization with fixed point math only)

 Other optimization : Model optimization toolkit, Keras-based connection pruning, (Runtime sparsity support)

 

Documentation

 Docs : New tensorflow.org/lite, (More tutorials)

 Model repository : 5 models ready to use, (Expand repository to many more models)  

 

Training

 TF Select + Control Flow + Subgraphs + Variables

 

TensorFlow Lite Roadmap

 New Converter :

  Control flow (sooner)

  Training (later)

 Improved runtime

  Idiomatic APIs for pre/post-processing

  Stable bindings (C/Obj-C/Swift)

 

 もう一つが、TensorFlow World 2019 Conference でのビデオ

 

www.youtube.com

上のビデオでは語られていないものだけ。

TensorFlow Lite Support Library

 APIs for simplifying pre- and post-processing (launced)

 Autogenerates pre- and post-processing (in progress)

 

Model Metadat support

 .json Example

 

 

Language Bindings

 Swift, Obj-C, C#, C for iOS, Android and Unity

 Community language bindinsg : Rust, Go, Flutter/Dart 

 

Running TensorFlow Lite on Microcontrollers

 

Our Future

 More ops and supported models

 Performance improvements (CPU, GPU, TPU, etc.)

 More models, TF Hub integration & better documentation

 On-device training and personalization

 Automated performance discovery tools