Introduction to Swift for Tensorflow

on YouTube ↗40m

" summary

Overview of a concrete workflow: Swift for TensorFlow compiles models through MLIR/XLA so one codebase can produce optimized runtimes for CPU, GPU, TPU and mobile devices, claiming write-once deploy-anywhere. Demos run in Google Colab and Jupyter notebooks (fast.ai ImageNet notebook runs on free GPUs), repositories include runnable examples and a two-command build to produce a Swift-for-TensorFlow toolchain. Notes GPU hardware has thousands of cores and much faster on-card memory, and Google pairs research models with nightly builds for practical testing.

AI-written summary · about

" chapters (9)

1. 0:00Introduction to Swift for TensorFlow
2. 0:25Why combine Swift and TensorFlow
3. 9:37Future of machine learning frameworks
4. 12:49Cloud computing and AI trends
5. 16:19Defining artificial intelligence
6. 20:32Mobile development advantages
7. 24:29Tools and skills for ML
8. 26:16Getting started with Swift for TensorFlow
9. 37:56Call to action and conclusion

" related talks

• 39mSwift for TensorFlow: Machine Learning with No BoundariesPaige Bailey · Scale By The Bay 2019
• 28mSwift for TensorFlowJames Bradbury, Richard Wei · Google I/O 2019
• 24mSwift for TensorFlow: machine learning with no boundaries.Paige Bailey · try! Swift NYC 2019
• 47mSwifty TensorFlowSiamak Ashrafi · 360|iDev 2019
• 26mMachine Learning with TensorFlowRajat Monga · MLconf SF 2016
• 27mMachine Learning and MobileBrett Koonce · Scale By The Bay 2019
• 31mTensorflow for JanitorsDániel Molnár · code.talks 2017
• 32mAI for Mobile and IoT Devices: TensorFlow LiteSarah Sirajuddin, Tim Davis · Google I/O 2019