Tensorflow 2.0: models migration and new design https://buff.ly/2zvL9oI Great overview of upcoming TF changes. Personally I agree with the author and I'm not a huge fan of eager mode either. Is it just because I'm used to thinking in terms of a graph? Is eager really superior?
-
-
Replying to @dennybritz
I think graphs are the correct mental model and API for deep learning -- but not graph of ops like a TF graphdef, instead, graph of layers. Recursive graphs of high-level building blocks. Which is also how deep NNs are visualized in pretty much every paper or textbook ever...
2 replies 1 retweet 24 likes -
Replying to @fchollet @dennybritz
Could part of the middle ground be something akin to a type system for layers/blocks, to both aid composition and automatically check for "correct" compositions of high level abstractions?
1 reply 0 retweets 1 like -
Replying to @georgebdavis @dennybritz
We have that, and it's a lot more advanced than static type checking. Any Keras model that compiles will run. If it's only made of built-in layers.
1 reply 0 retweets 1 like
And compilation errors (assumption checks) are designed to produce helpful and actionable error messages
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.