me2too

10 ms·

61.
What happens when tf.function converts a “complex” Python function? design guide(pgaleone.eu)
1 points by me2too 7y ago|0 comments
62.
How does tf.function work? Weird behaviors and bad performance analysis(pgaleone.eu)
1 points by me2too 8y ago|0 comments
63.
Analyzing tf.function to discover AutoGraph strengths and subtleties(pgaleone.eu)
2 points by me2too 8y ago|0 comments
64.
Tensorflow 2.0: Keras is not (yet) a simplified interface to Tensorflow(pgaleone.eu)
3 points by me2too 8y ago|0 comments
65.
Computer Vision – Fixed camera setup for object localization and measurement(pgaleone.eu)
1 points by me2too 8y ago|0 comments
66.
by me2too 8y ago
This is the website I used to learn Prolog and all I can do is to highly recommend it
67.
by me2too 8y ago
https://www.tensorflow.org/api_docs/python/tf/contrib/distri...
68.
by me2too 8y ago
Computer Vision tasks: I train models to to object detection, image classification, semantic segmentation... export the model, load the trained model in C++ and run inference. Or generative models: same workflow (python definition and train
69.
Show HN: Tensorflow and Go, the gopher way(github.com)
1 points by me2too 8y ago|0 comments
70.
by me2too 8y ago
Point 1 is a good point, although tfdebug exists and it works. For point 2: I don't see the problem. I use almost daily a wrapper (defined inside tf.contrib, that in version 2.x will go in core [I hope]) around the optimizer that in 2
71.
by me2too 8y ago
But the API is strictly statically typed in reality. The core is written in C++ and every operation requires a well-known type in input. As you can see from the error raised in this example: https://pgaleone.eu/tensorflow&#x
72.
by me2too 8y ago
But why? I do use Tensorflow form the 0.x release and I'm still using it right now, migrating all the codebase to the 2.x version. I can do everything in both pytorch and tensorflow and when I have to define really efficient input pipe
73.
by me2too 8y ago
The fact that I have all the computation described in a coherent manner, in something that's agnostic to the language I'm using. The same description (the graph) can be taken and used in every other language. I can move a trained
74.
by me2too 8y ago
At your contrary, I love the graph. However, migrating to a keras-like approach, you can work thinking about "objects with variables inside" and then the graph will be built for you by the abstraction introduced by `tf.keras.Model
75.
Tensorflow 2.0: models migration and new design(pgaleone.eu)
3 points by me2too 8y ago|0 comments