Thursday, August 15, 2019

Inside TensorFlow: tf.distribute.Strategy

Inside TensorFlow: tf.distribute.Strategy [Collection] Take an inside look into the TensorFlow team’s own internal training sessions--technical deep dives into TensorFlow by the very people who are building it! On this episode of Inside TensorFlow, TensorFlow Software Engineer Josh Levenberg discusses tf.distribute.Strategy and how it’s designed for ease of use across a wide range of distribution use cases. Let us know what you think about this presentation in the comments below! TensorFlow on GitHub → http://bit.ly/2Z3futx Watch more from Inside TensorFlow Playlist → https://bit.ly/2JBXFtt Subscribe to the TensorFlow channel → https://bit.ly/TensorFlow1

Wednesday, August 14, 2019

Model Understanding and Business Reality (TensorFlow Extended)

Model Understanding and Business Reality (TensorFlow Extended) [Collection] In the fifth and final part of Developer Advocate Robert Crowe’s overview of TensorFlow Extended (TFX), we’re discussing Model Understanding and Business Reality. Learn more on the series finale of Real World Machine Learning in Production! This video is also subtitled in Chinese, Indonesian, Italian, Japanese, Korean, Portuguese, and Spanish. Watch more TensorFlow Extended (TFX) → https://goo.gle/2xVkwt4 Subscribe to the TensorFlow channel → https://goo.gle/2WtM7Ak To learn more about TFX check out the website: https://goo.gle/2YymOcC, and join the group https://goo.gle/2OGjw78

Wednesday, August 7, 2019

Answering your TF Lite questions and more! (#AskTensorFlow)

Answering your TF Lite questions and more! (#AskTensorFlow) [Collection] Developer Advocate Paige Bailey (@DynamicWebPaige) and TF Developer Advocate Daniel Situnayake answer your #AskTensorFlow questions. Remember to use #AskTensorFlow to have your questions answered in a future episode! 0:21 - Is RNN / LSTM, quantization-aware training, and TOCO conversion in TF Lite available in TensorFlow 2.0? 1:22 - Is there any tutorial / example for text processing models in TF Lite, aside from the pre-trained smart reply example? 1:53 - Is Swift for TensorFlow for iOS programming? 2:37 - Will there be a commodity device that I can use for TPU inferencing? 3:43 - Does TF Lite only work on Coral dev boards? 4:41 - Will Edge TPUs be available to purchase in other countries? 5:19 - What about Android things? Does TF 2.0 support them? 6:08 - What platforms are supported by Swift for TensorFlow? 6:55 - Will there be support in the Python API for exporting object detection models (e.g., after transfer learning) to TF Lite? 8:28 - Why is it currently so difficult to integrate and use custom C++ / CUDA operations in TensorFlow and especially TensorFlow Serving? Are there any plans to make this process easier for production? 9:28 - I had some problems using Keras and TensorFlow + OpenCV. Are there any improvements in TensorFlow 2.0? 10:44 - Does TensorFlow have any API that can do AutoML, as Azure ML SDK? 11:40 - What about Kotlin for TensorFlow? 12:12 - Can a deep learning model be miniaturized automatically? 13:19 - Regarding tf.data, do you guys have any new APIs to directly load audio files (.wav, etc.) instead of going through the extra conversion steps to convert to TFRecords? 14:14 - Do you have any plans to add support for constraints or -even better- AutoDiff on manifolds? It would be so nice to do optimization where some parameters live in SO(3), for example. Resources mentioned in this episode: TF Lite mailing list - http://bit.ly/2KbyHBl Swift for TensorFlow mailing list - http://bit.ly/338ydDw Guide on RNNs & LSTMs in TF Lite - http://bit.ly/2KgvyPn TOCO converter guide - http://bit.ly/2MzvPQ4 Post-training quantization - http://bit.ly/2KnWE74 Coral platform - http://bit.ly/2ZlMbiS TF Lite models - http://bit.ly/32XGGZU MLIR: A new intermediate representation and compiler framework - http://bit.ly/2KnX25w Cloud AutoML - http://bit.ly/2Yghy1Q TensorFlow model optimization toolkit - http://bit.ly/2K9R2OS tf.io - http://bit.ly/2YzmqdB This video is also subtitled in Chinese, Indonesian, Italian, Japanese, Korean, Portuguese, and Spanish. Subscribe to the TensorFlow channel → http://bit.ly/TensorFlow1 Watch more episodes of #AskTensorFlow → http://bit.ly/2JcL3tT

Tuesday, August 6, 2019

TFX Episode 4: Distributed Processing and Components (TensorFlow Extended)

TFX Episode 4: Distributed Processing and Components (TensorFlow Extended) [Collection] On today’s episode of TensorFlow Extended hosted by TensorFlow Developer Advocate, Robert Crowe, we’re discussing Distributed Processing and Components. Learn more on Episode 4 of the 5 part series on Real World Machine Learning in Production and a preview for the series finale! This video is also subtitled in Chinese, Indonesian, Italian, Japanese, Korean, Portuguese, and Spanish. Watch more TensorFlow Extended (TFX) → https://goo.gle/2xVkwt4 Subscribe to the TensorFlow channel → https://goo.gle/2WtM7Ak