Dan Moldovan

Research software engineer at Google Brain

[👩‍💻 Paige Bailey @DynamicWebPaige]

Announcing a cool new feature in @TensorFlow: differentiable map ops! 🙌✨

If you've ever wanted to store and train embeddings in an embedding model, this should make the process *much* simpler. 😄

https://t.co/0ybR7BB6QN

Amazing work from our TF intern, @kattian_ (@Harvard)!

[Dan Moldovan @mdanatg]

@deckeresq @tiarkrompf @SShanabrook On the TF side of things, there's also https://t.co/Oqv289vI0K; will start tagging the relevant changes to the issue. Another place to watch is this directory, a staging area for all things independent of TF: https://t.co/JOiEIzdGRm

[François Chollet @fchollet]

Have you used tf.keras in TensorFlow 2.0 recently? We'd like your feedback. Here's a 15-second survey: https://t.co/v7cwjjYvrg

[Dan Moldovan @mdanatg]

@SShanabrook @awiltsch Sadly I won't make it to PyCon this year, but Brian Lee from our team will present one of our early prototypes. The path to a standard solution can be a long one, but I believe a good first step is figuring out the prerequisites that would first need to be standardized.

[Josh Gordon @random_forests]

Important guides to check out for TF 2.0: Effective TensorFlow (https://t.co/I0uHhM7TIS), AutoGraph (https://t.co/j9Rkkggxvv).

[Dan Moldovan @mdanatg]

@SShanabrook Yep! We hope to discuss it at a sponsor talk this year, and have a full presentation next year after the library had been released. By then, it would be awesome to also have a PEP started! @awiltsch

[Dan Moldovan @mdanatg]

@SShanabrook I wasn't alone :) Many contributed, and more still inspired this work :)

[TensorFlow @TensorFlow]

TensorFlow 2.0 alpha is here! We've asked the community what they wanted and we're working to make TensorFlow simpler and easier to use for everyone.

Read up on how to get started here → https://t.co/oGtb2B2Pz3

[Dan Moldovan @mdanatg]

@SShanabrook Thank you! Definitely agree it can be useful across Python, and we're working on releasing a generic library soon - stay tuned! Here's a paper that discusses the idea in more detail: https://t.co/cER9qt3oxj.

[Google @Google]

Here’s to first responders around the world who answer the call in times of crisis.

[Google AI @GoogleAI]

Building ML models that are robust and inclusive even when learning from imperfect data is an important research challenge. To spur further progress in developing inclusive models, we are announcing the Inclusive Images Competition on Kaggle. https://t.co/gVMImMZ8sR

[Google AI @GoogleAI]

We’re excited to announce the public alpha of Cirq, an open source framework for developing algorithms on noisy intermediate scale quantum (NISQ) computers. Check out the GitHub repository — pull requests welcome! #Cirq #quantumcomputing https://t.co/GGbRngQiam

[Dan Moldovan @mdanatg]

@awiltsch @nsthorat Nah, awiltsch is the man, forget the rhymes :)

[TensorFlow @TensorFlow]

Graph models allow for all kinds of optimization, but it can be tricky to move between Python and graph code. To help with this, TensorFlow has AutoGraph, a new feature that converts Python code into pure TensorFlow graph code.

Learn more on the blog ↓ https://t.co/AcNKnvkyJV